Astroparticle, Particle And Space Physics, Detectors And Medical Physics Applications (Proceedings of the 9th Italian Conference)

•

4.5

.&

Data

» MC simulation

4

3.5 3

2.5

S 3

2

r

1.5

=

1

0.5 11.... i....

0

0.5

1

i.

1.5

2

2.5

3

3.5

4

4.5

5

l/cos(theta) Figure 3. Average number of fired strips vs l/cos8 and MC comparison Entries 336866 Mean 43.55 RMS 26.48

5fl

100

ISO

250

Time over Threshold (DAC) Figure 4. ToTs distribution for each track layers and MC comparison

143 160

• Da/a T MC simulation

140

O lM

I

1 100 "S

s

e

so

^^pl^f

40 20 A * " .

• •

0

•

i

•

0.5

•

•

•

i

1

•

•

• •

i

•

1.5

i

•

i

i

2

•

•

•

•

i

•

2.5

•

i

•

i

3

i

i l n i i li

3.5

4

4.5

5

l/cos(theta) Figure 5. ToTs vs l/cos0 and MC comparison and MC comparison As the hit strip multiplicity, the ToTs in the track layers increases linearly with the l/cos0 (proportional to the track length in the SSDs) as expected by the MC simulation (Fig.5).

5.

Conclusions

The preliminary results from analysis of cosmic ray data taking during LAT integration are shown in this paper. The study of tracker behavior shows a good agreement between data and MC simulation predictions. Cosmic rays test with the complete LAT configuration and beam test that will be performed during 2006 will be usefully to tune the Monte Carlo. After that, the GLAST Collaboration will be able to use the Monte Carlo simulations to predict the performance of the instrument.

References 1. 2. 3. 4. 5. 6.

http://glast.gsfc.nasa.gov/ http://www-glast.stanford.edu/pubfiles/proposals/bigprop/ http://proj-gaudi.web.cern.ch/proj-gaudi/welcome.html M.N Mazziotta et al., Nuc. Instrum. And Meth. A533:322-343, 2004 M. Brigida et al. (2002), LAT internal note, LAT-TD-0105 Eduardo do Couto e Silva and Lee Steele (2005), LAT-TD-04631-02

B A C K G R O U N D ANALYSIS OF CUORICINO I N V I E W OF THE FUTURE EXPERIMENT CUORE

S. CAPELLI, 0 . CREMONESI, M. PAVAN, S. PIRRO, E. PREVITALI Dipartimento di Fisica dell'Universitd. di Milano-Bicocca e Sezione di Milano dell'INFN, Milano I-20126,Italy The study performed on the measured CUORICINO background spectrum is reported. The extrapolation to CUORE background and the R&D begun in order to reduce it to a value compatible with the wanted sensitivity is also presented.

1. Introduction The discovery of neutrino oscillations has unequivocally demonstrated that neutrinos are massive and mixed particles. Experiments looking for the Neutrinoless Double Beta Decay (0/3(Ov) ) can probe the Majorana nature of the particle and the absolute value of the Majorana mass (mee) . The value of (mee) is unknown but can be partially inferred from oscillation data. The possible ranges of values at 99% C.L. for (m ee ) are (1.1 - 4.5) meV for normal hierarchy [1], (12 - 57) meV for inverted hierarchy [1] and (10 - 55) meV for quasi degenerate spectrum [2]. The sensitivity reached by the running CUORICINO experiment ranges between 0.1 and 1.1 eV [3], allowing to explore part of the quasi degenerate mass spectrum range. Next generation p/3(Qv) experiments will require at least a sensitivity of the order of 10 meV in order to explore the inverted hierarchy mass spectrum range. The f}j3(0v) is a lepton violating process, where a nucleus (A,Z) decays into (A,Z+2) with the emission of two electrons and no neutrino. The searched signature is a peak at the Q value of the transition in the sum energy spectrum of the two electrons. The decay rate of this process is directly proportional to (m ee ) squared and depends on the nuclear matrix elements for the decaying nucleus. The chosen isotope in the CUORICINO experiment is 130 Te , whose high transition energy (2528.8 ± 1.3 keV) and high isotopic abundance (33.8%) make it good DBD candidate. The detectors are Te02 crystals

144

145

operated (at a temperature of ~10 mK) as bolometric detectors [4],

2. CUORICINO and C U O R E set-up The CUORICINO detector consists in 62 Te02 crystals, arranged in a tower like structure made by 11 planes of 4 crystals of 790 g and of 2 planes of 9 crystals of 330 g, for a total mass of ~40.9 kg. It is installed in the hall A of LNGS (3200 mwe), in a diluition refrigerator able to reach temperatures of about 10 mK. The tower is surrounded by copper shields, 1.5 cm of Roman lead shield, 20 cm of commercial lead shield and 10 cm of borated P E T shield. The whole set-up is kept in nitrogen overpressure and placed in a Faraday cage. Particular care was devoted to selection and cleaning of materials. Crystals and copper structure were subjected to a dedicated surface treatement. While being a self consistent experiment CUORICINO is also a feasibility test for the next generation experiment CUORE. CUORE [5] will be constituted by 988 Te02 790 g crystals, arranged in 19 towers made by 13 modules of 4 detector each, for a total mass of ~750 kg. The detector will be placed in a specially produced dilution refrigerator and provided with copper and Roman lead inner shields and externally shielded by 30 cm of commercial lead and 10 cm of borated PET. The detector set-up will be kept in nitrogen overpressure and placed in a Faraday cage.

3. Background sources for a D B D experiment Due to the rarity of the searched process the sensitivity of a /3/3(0v) experiment can be strongly limited by the background counting rate in the region of interest. Background understanding and reduction is therefore mandatory. The main background sources for a bolometric detector like CUORICINO can be divided in: • External sources: photons and neutrons from the Gran Sasso rock or muons and muon induced particles. • Internal sources: intrinsic radioactivity and cosmogenic activation of all the experimental materials and surface contaminations of crystals and materials facing them.

146

4. Background study No peak at 2528 keV appears in the anticoincidence background spectrum obtained with the statistics collected so far with CUORICINO (5.87 kg x year of 130 Te ). Using a Maximum Likelihood procedure to evaluate the maximum number of DBDOi/ events compatible with the measured background in the /?/?(0^) region (0.18 ± 0.02 c/keV/kg/y), a lower limit of 2.0 x 10 24 years at 90%C.L on the 130 Te half-life for the /?/?(0i/) was set. A dedicated study of the measured CUORICINO background was performed in view of the future CUORE experiment. An analysis method based on sophisticated procedures and with the support of Montecarlo (MC) simulations was developed. By analysing coincidences, counting rates in different energy regions, gamma and alpha peaks position, intensity, rates and shapes, the most probable sources contributing to the measured background are identified. MC spectra are then generated, reproducing the contribution that each of these sources would give in the detector. A dedicated code to find the linear combination of input MC spectra that gives the best agreement with the measured spectrum is then used in 3 steps: a peak region (E>4 MeV), a continuum (E>3 MeV) and 7 and /3 region. The Geant4 package has been used to write a C + + based code able to simulate the entire structure of CUORICINO and CUORE. Decay processes due to natural ( 232 Th , 238 U , 2 1 0 Pb , 4 0 K ) and cosmogenic isotopes in the bulk and on the surfaces of the various experimental parts have been simulated, as far as LNGS environmental radioactivity. 4.1. CUORICINO

background

analysis

From the analysis performed on the CUORICINO data the main sources responsible of the background in the P/3(0v) region have been identified, a decays occurring on the crystals surface have been clearly identified from their characteristic asimmetric peak at the alpha transition energy and the coincidence pattern, while a decays occurring in the crystals bulk show a gaussian peak at the Q value of the decay. The flat continuum measured between 3 and 4 MeV can be ascribed to deep (1-5 /xm depth) surface contaminations of materials facing the crystals (the biggest surface is the one of the copper structure) as well as to surface contamination of the detector themselves. The contribution to the /30(Ov) region from these sources has been evaluated to be of ~10±10% and ~50±20% respectively for the Te02 and Cu surfaces. The remaining background in the /?/?(0z/) region can be ascribed to 2 3 2 Th contaminations of experimental part distant from

147 the detector ( ~30±10%). A limit of ~10 _ 8 Bq/cm 2 has been evaluated for the surface contamination of copper and crystals in 2 3 2 Th , 238 U or 2 1 0 Pb . For the bulk contamination a limit of 1 pg/g and 0.1 pg/g for copper and TeC>2 has been set. 4.2. Extrapolation

to

CHORE

In order to reach the sensitivity required to investigate (mee) in the inverted hierarchy, it is necessary to reach in CUORE a background rate in the /3/?(0f) region of the order of 1 0 _ 2 - 1 0 ~ 3 c/keV/kg/y. The construction of the CUORE detector will require about five years; a dedicated R&D is on the way, aiming to decrease the background sources that could limit the sensitivity of the experiment. The background reachable with CUORE was evaluated on the basis of the background model developed for CUORICINO and on the basis of the radioactive contaminations measured so far for the various materials commercially at our disposal [5]. Negligible contributions are expected from thermal neutrons and environmental gammas, thanks to the use of specially designed neutron and lead shields. The evaluated contribution from high energy neutrons (both produced from rock radioactivity and from muon interactions in the rock and in the lead shields) is of ~ 4 x l 0 - 4 c/keV/kg/y. Copper and crystals bulk contamination expected rate in the flfi{Qv) region is less than 3xl0~ 3 c/keV/kg/y. The most worrisome background source is the surface contamination: for contamination values of ~10~ 8 Bq/cm 2 a rate of 2xlO~ 2 c/keV/kg/y and 5xlO~ 2 c/keV/kg/y could arise for crystals and copper surface contaminations respectively. 5. R & D for background reduction In order to fulfill the CUORE goal with respect to background reduction an intense R&D had begun with respect to surface cleaning and passivation, aiming at reaching at least reduction by factors 10 and 4 in copper and crystals surface contamination respectively. Dedicated measurements have been performed in a second and smaller cryostat, installed in the hall C of LNGS, provided with 5 cm of Copper shield and 15 cm of commercial lead. No neutron shield is present. In the first measurement (RAD1) 8 crystals of 760 g subject to surface etching and polishing with clean and measured powders have been mounted in CUORICINO like tower. The copper used for the tower structure was

148

also subject to a specially studied cleaning procedure: etching, electropolishing and passivation. The measured background, with a statistics of ~6750 h x detector, gave very encouraging results with respect to crystal surface 2 3 2 Th and 238 U contamination reduction. The rate of the a peaks due to these sources were in fact reduced by a factor of ~ 4 with respect to CUORICINO, fulfilling the first CUORE milestone. The continuum background between 3 and 4 MeV was compatible with the one measured with CUORICINO. Further testswere therefore necessaries in order to find out the source responsible for this contribution. In a second test (RAD2) with the same detector array the crystals were faced to a higher mass and surface of teflon (a 10x10x0.5 cm 3 sheet facing 4 crystals), gold wires (12.7 m facing 2 crystals) and silicon heaters (~30 cm 2 facing 2 crystals), thus increasing the sensitivity for these components contaminations with respect to CUORICINO. The measured background have proved that this source can't be responsible of the rate measured in CUORICINO and in RAD1 in the 3-4 MeV energy region. Another test (RAD3) is presently running in hall C in order to have a clear answer with respect to copper surface contamination. All the copper parts facing the detectors have been covered with 5 layers of ~12/im low contaminated polyethylene sheet, in order to stop a particles generated by decays occurring in the Cu surface. In a second run thermal neutrons will be tested by changing the shields: the 5 cm of copper will be sostituted by lead and 1 cm of CB4 and 6 cm of paraffin will be added. Preliminary results show a background reduction in the 3-4 MeV region of a factor ~ 2 . 6. Conclusion The analysis performed on CUORICINO data was fundamental in order to disentangle the main sources responsible of the background measured in the (3/3(Qv) energy region. And intense and promizing R&D has begun in order to reduce the identified contaminations in view of the next generation CUORE experiment. References 1. 2. 3. 4. 5.

A.Strumia and F.Vissani Nucl.Phys. B726, 294-316 2005 S.Pascoli and S.T.Petkov Nucl. Phys. B580, 280 2004 C.Amaboldi et al. (hep-ex/0501034) D. Twerenbold, Rep. Prog. Phys. 59 349 1996 A.Arnaboldi et al, CUORE proposal

APPLICATION OF NUCLEAR TRACK DETECTORS IN ASTROPARTICLE AND NUCLEAR PHYSICS S. CECCHINI IASF/INAF, 40129 Bologna, Italy INFN, Sez. Bologna, 40127Bologna, Italy S. BALESTRA, M. COZZI, G. GIACOMELLI, M. GIORGINI, S. MANZOOR, E. MEDINACELI, L. PATRIZII, V. POPA, G. SIRRI, M. SPURIO, V. TOGO Dipartimento di Fisica, Universita di Bologna, 40127Bologna, Italy INFN, Sez. Bologna, 40127 Bologna, Italy T. CHIARUSI

Dipartimento di Fisica, Universita la Sapienza, 00185 Roma, Italy After a brief presentation of the technical methods employed and the calibrations performed on different Nuclear Track Detectors (NTDs), the results obtained by their use in different fields of astroparticle and nuclear physics are presented. In particular the results of searches for magnetic monopoles, nuclearites and other exotica in the cosmic radiation carried out by the Bologna Group. Also an application of these detectors in an experiment for the determination of the charge spectrum of primary cosmic ray particles is described. Example of the use of the same detectors for the measurement of the total and partial charge changing fragmentation cross sections of relativistic ions in several targets are also reported.

1. Introduction The technique of Solid-state Nuclear Track Detectors (SSNTDs) was born almost 50 years ago. Since then the field has grown enormously; numerous Conferences have been dedicated to the topic and several books cover different aspects of this technology [1]. Starting from 1986 the Bologna group has devoted part of its activity to the use of plastic materials (CR39, Makrofol, Lexan) in various research fields and to the improvement of industrial production of CR39 expressly for science applications [2]. Here we briefly summarize the results obtained during the decennial experience acquired in Astroparticle and Nuclear physics experiments using this detection technique. 149

150

2. The Nuclear Track Detectors (NTDs) The working principle of plastic NTDs is qualitatively presented in Fig.l. When a particle crosses such a detector the polymeric chains along its path are affected giving origin to the so-called "latent track". In the case of a particle with an electric charge Z and velocity (3 (= v/c), the produced damage can be related to Z/p. The latent tracks become visible under an optical microscope after chemical etching, typically in aqueous solution of NaOH or KOH - for the materials considered here - when the etching velocity along the latent track (vx) is larger than the one for the bulk of the material (vB). The temperature and concentration of the etching bath are chosen according to the type of application. For a through going particle, in the initial stage of the etching two cones are formed on both sides of the detector. The size and shape of such cones depend on the energy loss of the incident particle; the measurement of the cone base area or heights allows, through an appropriate calibration, to determine the characteristics of the particle.

Figure 1. (left) The breaking of polymeric chains of plastic NTD by a crossing particle; (right) crossview of the formation of the track after etching for two different times [1].

The relation between the parameter p =vT /vB, the observable quantity, and the REL is obtained through calibrations. A typical set-up at an accelerator ion beam would consist in few sheets of detectors upstream of some target material (Cu, Pb, C, Al and so on) used for the ion fragmentation, and several NTD sheets placed downstream of the target in order to record the tracks of the nonfragmented ions as well as of the fragments generated in the target. Fig 2 shows the distribution of the etch cone base areas produced by 158A GeV In + ions and their fragments at the CERN SPS in CR39. By computing the REL values corresponding to the different relativistic ions, the NTD calibration curves (p vs. REL) can be obtained [3].

151

Different calibrations for different materials and etching conditions appropriate for the specific experiment carried out have been obtained [4]. 200

pyr-

180

7

1 1 1 1 1 1 1 1 1 1

SOFT ETCHING

160 '7

Z=49

<• ' Z=7 77 1

1 120 A S100

-

|soL so '4D '20

-

o Eill

2500

5000

7500

10000 12500 15000 Average
17500

20000

22500

Figure 2. Distribution of the base areas of the etched cones in CR39 by In49+ ions and their fragments; the base area measurements are averaged over two faces.

3. Astroparticle Physics The use of NTDs can play a key role for the detection of particles incident (or hypothetically incident) on the Earth atmosphere. In many cases the event flux is expected to be small and NTDs for their characteristic of being passive, easy to use, low cost and adaptable at the scope represent a good solution.

3.1. Primary Cosmic Rays Abundances The nuclear composition of primary CR is a rich probe for understanding their origin. Precise measurements of the nuclei abundances beyond Z > 30 in fact could allow to discriminate between two competing models [5] and [6] proposed for the composition at their sources. Especially the ratios among heaviest elements could be the best discriminator for most current theories regardless of the propagation history of these nuclei. In order to be able to measure these

152 ratios, high charge resolution and high statistics are needed since the flux of heavy nuclei is very low (< 1 part/m2h). Until now, the largest statistics above the Fe threshold came from experiments of the HEAO-C3 [7] and Ariel-6 [8] missions, which unfortunately suffered from low charge resolution. The use of NTDs onboard space vehicles to study the radiation environment is well know [9, 10]; they were also used for measurements of nuclei with Z > 60. The major advantage of NTDs is the possibility of building experiments with large geometrical factors, balloon flights of 20-40 days appear feasible from Antarctica or in north circumpolar routes. Another possibility can be offered by the ISS for exposure of large area (tens of m2) for few months. The competition with active detectors is favoured for what concerns weights, cost and power consumption.

1

Wl

" 5

10

• I 15

20

I . , I. . » ! . . . . 2S 30

.1 Z

Figure 3. Charge spectrum of the 2051 events from the measurements of 22.5% area of the CAKE detector. Data (black dots) with statistical errors are compared to MonteCarlo simulations (histogram).

The large area exposed requires automatic scanning and track recognition, as it is unthinkable to use operators at optical microscope for such large areas. For this purpose we have developed a prototype experiment CAKE that has been flown on a stratospheric balloon that reached a plafond altitude of 38-40 km for 18 hours during a trans-mediterranean flight. First results of this tests have been reported [11] (Fig. 3) and suggest that it is possible to reach the sensitivity, charge resolution and statistics needed for discriminating between different scenarios of the origin of CR with reasonable dimensions and exposure times of the detectors.

153 3.2. Exotica searches Different exotic particles have been suggested to be part of the cosmic radiation flux such as Magnetic Monopoles (MMs), Nuclearites, Q-balls [12]. All of them can be constituents of the dark matter component of the Universe, and could have very wide mass ranges (from multi-TeV to 1017 GeV and beyond). A search for GUT MMs was one of the primary goals of the MACRO experiment at LNGS (3600 m w. e. depth). Different sub-detector systems were used to have redundancy for the detection of MMs with different charges and velocities; among them 1280 m2 of NTDs were used for a total acceptance of ~ 10000 m2sr. From a 10-years exposure it has been possible to establish a 90% CL flux upper limit that which at present is still the lowest for direct searches of such particles in the 4. 10"5 < p < 0.8 [13]. Also multiply charged MMs with masses from 107 to 1013 GeV could be present in the cosmic radiation and accelerated to high velocities [14]. In this case one has to look for MMs with p > 10"2. Detector on the Earth surface could detect IMMs coming from above if they have masses > 105 GeV [15]; lower mass MMs may be searched for with detectors located at high mountain altitude, balloons and satellites.

jS*wc

Figure 4. 90% CL. flux upper limits for different charged MMs obtained with the analysis of 239 m2 and average exposure time of 3.7 years with SLIM experiment

154 The SLIM experiment [16], based on 440 m2 of NTDs exposed for 4 years at the Chacaltaya (Bolivia) High Altitude Laboratory for CR (5290 m a.s.l.) is sensitive to IMMs with masses > 104 GeV for g = gD magnetic charge [17]. NTDs have been used also to search for Dirac MMs produced at high-energy accelerators [18]. Similarly interesting are other particles such as Nuclearites andQ-balls[12].

4. Other applications of NTDs We recollect here examples of measurements we performed with NTDs connected to the fields of particle physics and of the natural environmental radiation. 4.1. Fractional charged nuclei The charge resolution of NTDs is the best feature for such kind of technique. Charge resolution as low as 0.05e in the range 5 < Z < 14 has been obtained for CR39 by tracking the ions through a stack of 12 detector sheets (Fig. 5). Using 200

S, 200 wV/,iuc(*i>ti

ISO 160

5140

£«0 feioo or & so CO

§ eo z 40 20

i

,d

k 10

11

12

13

IS

16

17

CHARGeZ

Figure 5 . Charge resolution obtained by using CR39 exposed to a S,
these measurements a search for the production of fractionally charged nuclei was made. A limit at the level of 1/104 was established [19].

155 4.2 Fragmentation cross-section measurements Experimental results on the fragmentation properties of high-energy nuclei are relevant for nuclear physics, cosmic ray physics and astrophysics. We measured the fragmentation cross-sections in high-energy nucleus-nucleus collisions using relativistic Sil4+, S16+ and Pb82+ ion beams on different targets by means of CR39 NTDs which yielded an easy charge identification [20]. 4.3 Radon Indoor Radon concentration in Bologna and in the LNGS tunnel was monitored using CR39 based detectors [19]. With this type of technique it is possible to measure radon concentrations integrating over time periods of the order of months. After exposure the tracks left by alpha particles from radon decay are observed and counted under an optical microscope with 40x objective and lOx eyepiece, Fig .6. It has been also established a calibration constant 1 track/cm2 for 1 month exposure = 0.41 Bq/m3 ~ 0.011 pCi/1

G

©

• 3

©

'•

as Figure 6. Microphotograph of the alpha particle tracks in CR39 visible at a microscope after exposure in environment where Radon was present.

5. Conclusions Plastic NTDs are easy to use detectors. They offer very good charge resolution (probably the best) and have wide applicability as we have tried to illustrate above.

156 Acknowledgments R. Giacomelli, A. Kumar, G. Mandrioli, M. Errico, E. Bottazzi, L. Degli Esposti, C. Valieri all these colleagues are duly acknowledged for their collaborations References 1.

R.I. Fleischer et al., Nuclear Ttracks in Solids, Univ. California Press, Berkeley, 1975 S. A. Durrani and R. Ilic, Measurements by Etched Track Detectors: Applications in Radiation Protection, Earth Sciences and the Environment, World Scientific, Singapore, 1997 D. Nikezic and K.N. Yu, Material Sci. Eng R 46, 51 (2004)

2.

S.P. Ahlen et al., Nucl. Tracks Radiat. Meas. 19, 641 (1991)

3.

S. Cecchini et al., Nucl. Tracks Radiat. Meas. 22, 555 (1993); II Nuovo Cim. 109A, 1119 (1996); Rad. Meas. 34, 55 (2001); G. Giacomelli et al., Nucl. Instr. & Meth., A411,41 (1998)

4.

S. Cecchini et al., hep-ex/0502034

5.

M. Casse and P. Goret, Astrophys. J. 346, 997 (1978)

6.

C.J. Cesarsky and J.P. Birbing, IAU Symposium, Bologna (1981)

7.

W.R. Binns et al, Astrophys. J., 346, 997 (1989)

8.

B. Byrnak et al., 18th ICRC (Bangalore, 1983) Conf. Papers, 2, 29

9.

J. Donnelly et al., Proc. 26th ICRC (Salt Lake City, 1999) Conf. Papers, 3,113

10.

B.A Weaver and A.J. Westphal, Proc. 27th ICRC (Hamburg, 2001) Conf. Papers, 5, 1720

11.

S. Cecchini et al., 16th ESA-PAC Symp., SP-530 (S. Gallen, 2003) 529; Proc. 29th ICRC (Pune, 2005) Conf. Papers , astro-ph/0510717

12.

D. Bakari et al., hep-ex/0004019; Astrop Phys. 15, 137 (2001); M. Ambrosio et al., Eur. Phys. J. C13,452 (2000)

13.

M. Ambrosio et al., Phys. Lett. B406, 249 (1997) and Eur. Phys. J. C 25, 511 (2002)

14.

P. Bhattacherjee et al, Phys. Rept. 327, 109 (2000) and reference therein

15. J. Derkaoui et al., Astrop. Phys. 9, 173 and 339, 1998 16.

S. Cecchini et a l , 28th ICRC (Tsukuba, 2003) Conf. Papers, 3,1657

157 17.

S. Balestra et al., 29th ICRC (Pune, 2005) Conf Papers, hep-ex/0508043

18.

M. Bertani et al., Europhys. Lett. 12, 613 (1990); K. Kinoshita et al., Phys. Rev. D46,R881(1992)

19.

S. Cecchini et al., Astrop. Phys. 1, 369 (1993); H. Dekhissi et al., Nucl. Phys. A662, 207 (2000); S. Cecchini et al., Nucl. Phys. A707, 513 (2002)

20.

G. Giacomelli et al., II Nuovo Saggiatore, 4, 5 (1990); M. Beozzo et al., Nucl. Tracks Rad. Meas. 19,297 (1991); Arpesella et al., Health Phys. 72, 629 (1997)

RECENT DAMA RESULTS

R. B E R N A B E I , P. B E L L I , F . M O N T E C C H I A A N D F . N O Z Z O L I Dipartimento di Fisica, Universita and INFN, Sezione di Roma2,

di Roma "Tor 1-00133 Rome,

Vergata" Raly

F . C A P P E L L A , A. D ' A N G E L C ^ A . I N C I C C H I T T I A N D D . P R O S P E R I Dipartimento di Fisica, and INFN, Sezione

Universita di Roma "La di Roma, 1-00185 Rome,

Sapienza" Raly

R. C E R U L L I INFN

- Laboratori

Nazionali

del Gran Sasso,

1-67010 Assergi

(Aq),

Italy

C.J. DAI, H.L. H E , H.H. K U A N G , J . M . M A a n d Z.P. Y E f IHEP,

Chinese

Academy,

P.O. Box 918/3,

Beijing

100039,

China

DAMA is an observatory for rare processes and it is operative deep underground at the Gran Sasso National Laboratory of the I.N.F.N.. The experiment developes and uses low background scintillator for rare processes investigation. In this paper, after a short presentation of the main DAMA set-ups, the DAMA/Nal apparatus (~ 100 kg highly radiopure Nal(Tl)) and its main results in the Dark Matter field will be addressed. This experiment, in particular, has effectively investigated the presence of a Dark Matter particle component in the galactic halo by exploiting the modelindependent annual modulation signature over seven annual cycles (total exposure of 107731 kg x day), obtaining a 6.3 a C.L. model-independent evidence for such a presence. In addition, some corollary model-dependent quests to investigate the nature of a candidate particle will be recalled. The new additional analysis for a pseudoscalar and for a scalar bosonic candidate (whose detection only involves electrons and photons/X-rays) will be addressed as well. Some perspectives of the second generation DAMA/LIBRA set-up (~ 250 kg highly radiopure Nal(Tl)), presently in measurement deep underground, will be mentioned.

* also: Scuola di Specializzazione in Fisica Sanitaria, Universita di Roma "Tor Vergata", 1-00133 Rome, Italy talso: University of Jing Gangshan, Jiangxi, China

158

159

1. Introduction DAMA is an observatory for rare processes based on the development and use of various kinds of radiopure scintillators. Several low background setups have been realized, in particular: i) DAMA/Nal (~ 100 kg of highly radiopure Nal(Tl)), which took data underground over seven annual cycles and was put out of operation in July 2002 1.2.3.4.5.6.7; ii) DAMA/LXe (~ 6.5 kg liquid Xenon) 8 ; iii) DAMA/R&D, which is devoted to tests on prototypes and small scale experiments 9 ; iv) the new second generation DAMA/LIBRA set-up (~ 250 kg highly radiopure Nal(Tl)) in operation since March 2003. Moreover, in the framework of devoted R&D for radiopure detectors and PMTs, sample measurements are carried out by means of the low background DAMA/Ge detector, installed deep underground since more than 10 years and, in some cases, by means of Ispra facilities. By profiting of the radiopurity of these apparata, several rare processes have been investigated obtaining also some competitive results 8 ' 9 . In the following, we will address, in particular, the results obtained by the DAMA/Nal experiment exploiting the annual modulation signature to point out with a model-independent approach the presence of a particle Dark Matter component in the galactic halo. This signature (originally suggested in 10 ) is very distincive; in fact, the expected signal must simultaneously satisfy of all the following requirements: 1) the rate must contain a component modulated according to a cosine function; 2) with one year period, T; 3) phase, to, around ~ 2nd June; 4) this modulation must only be found in a well-defined low energy range, where Dark Matter particle can induce signals; 5) it must apply to those events in which just one detector in a multi-detector set-up "fires" (single-hit events), since the Dark Matter particle multi-scattering probability is negligible; 6) the modulation amplitude in the region of maximal sensitivity is expected to be <7% for usually adopted halo distributions, but it can be larger in case of some possible scenarios such as e.g. those in ref. n > 1 2 . To mimic such a signature either spurious effects or side reactions should be able not only to account for the whole observed modulation amplitude but also to contemporaneously satisfy all the requirements; none has been found (see ref. 4 ' 5 and references therein). 2. The model-independent result of D A M A / N a l The experimental residual rate of the single-hit events in the (2-6) keV energy interval is reported in Fig. 1. The data offers a 6.3 a C.L. evidence for

160

0.1
2-6 keV

H t

i~n •*<- i n ->*HVr»«-

+

vii 4*

^VI-

%> 0,05

h/ \ \\A -0.1

1

$l\

^ ,,,,

500

1000

1500

i , ,

2000

,

11

2500 Time (day)

Figure 1. Experimental residual rate for single-hit events in the cumulative (2-6) keV energy interval as a function of the time over 7 annual cycles (total exposure 107731 kg X day). The experimental points show the errors as vertical bars and the associated time bin width as horizontal bars. The superimposed curve represents the cosinusoidal function behaviour expected for a Dark Matter particle signal with a period equal to 1 year and phase exactly at 2nd June; the modulation amplitude has been obtained by best fit. 4 - 6

the presence of an annual modulation of the measured rate of the single-hit events in the lowest energy region. In fact, fitting the experimental points with modulated cosine-like function (A- cosu(t — to)), a modulation amplitude equal to (0.0200 ± 0.0032) cpd/kg/keV, t0 = (140 ± 22) days and T = — = (1.00 ± 0.01) year, are obtained. The period and phase agree with those expected in case of an effect induced by Dark Matter particles in the galactic halo (T = 1 year and to roughly at ~ 152.5*'1 day of the year). The hypothesis of unmodulated behaviour can be excluded being the X2/d.o.f. = 71/37 on the (2-6) keV residual rate, which corresponds to a probability of 7 • 10~ 4 . Other model-independent analyses on the data have also been pursued; they give similar results and support the annual modulation effect observed at the same statistical significance. For a complete discussion see ref. 4 ' 5 . A careful investigation of all the known possible sources of systematic and side reactions has been regularly carried out and published at time of each data release and quantitative discussions can be found in ref. 3 ' 4 ' 5 . No systematic effect or side reaction able to account for the observed modulation amplitude and to satisfy all the requirements of the signature has been found. A further relevant investigation has been performed on the multiplehits events collected during the DAMA/NaI-6 and 7 running periods (see Fig. 2), thanks to an improvement in the electronics performed before

161

500 600 Time (day) Figure 2. Experimental residual rates over seven annual cycles for single-hit events (open circles) - class of events to which Dark Matter particle events belong - and over the last two annual cycles for multiple-hits events (filled triangles) - class of events to which Dark Matter particle events do not belong - in the (2-6) keV cumulative energy interval. The initial time is taken on August 7 t h . 6

those cycles. The multiple-hits event class - on the contrary of the singlehit one - does not include events induced by Dark Matter particles since the probability that a Dark Matter particle would interact in more than one detector is negligible. The fitted modulation amplitudes are: A = (0.0195 ± 0.0031) cpd/kg/keV and A = -(3.9 ± 7.9) • 10~ 4 cpd/kg/keV for single-hit and multiple-hits residual rates, respectively. Thus, evidence of annual modulation is present in the single-hit residuals (event class induced by Dark Matter particle), while it is absent in the multiple-hits residual rate (event class due only to background events). Since the same identical hardware and the same identical software procedures have been used for the two classes of events, the obtained result offers an additional strong support for the presence of Dark Matter particles in the galactic halo further excluding any side effect either from hardware or from software procedures or from background. In conclusion, the data of the seven annual cycles support at 6.3 a C.L. the presence of an annual modulation in the residual rate of the single-hit events in the lowest energy interval (2 - 6) keV, satisfying all the features expected for a Dark Matter particle component in the galactic halo. No systematic effect or side reaction able to account for the observed effect has been found. No other experiment, whose result can be directly compared with this one in a model independent way, is available so far in the field of Dark Matter investigation.

162

3. Some corollary model-dependent quests for a candidate On the basis of the obtained 6.3 u model-independent result, corollary investigations can also be pursued on the nature of the Dark Matter particle candidate. This latter investigation is instead model-dependent and - considering the large uncertainties which exist on the astrophysical, nuclear and particle physics assumptions and on the parameters needed in the calculations - has no general meaning (as it is also the case of exclusion plots and of the Dark Matter particle parameters evaluated in indirect detection experiments). Thus, it should be handled in the most general way as we have pointed out with time passing 2>4-5>6. Some wide discussion on this topics and on the real validity of any claimed model-dependent comparison in the field can be found in ref. 4 ' 5 ' 6 and in the references therein. The corollary quests for a Dark Matter candidate particle will be discussed considering some candidates belonging to the WIMP class particles (i.e. particle that interact via scattering off to target nuclei) and considering light bosonic candidates (i.e. particles that interact via the conversion of the particle mass in electromagnetic radiation). For simplicity, the results of these analysis are presented in terms of allowed volumes/regions obtained as superposition of the configurations at given C.L.. Details on the followed procedure can be found e.g. in ref.4'5,6. 3.1. Corollary

quests for WIMP

class

candidates

For this class of Dark Matter particles, it has been considered so far low (of order of few GeV) and high mass (up to hudereds of GeV) candidates interacting with ordinary matter via: i) mixed Spin-Independent & SpinDependent coupling; ii) dominant Spin-Independent coupling; iii) dominant Spin-Dependent coupling; iv) preferred Spin-Independent inelastic scattering. A detailed discussion on the allowed volumes/regions for these candidates can be found e.g. in ref. 4 ' 5 . As an example, here we have only reported in Fig. 3-left the results obtained in the general case of particle belonging to the WIMP class with mixed Spin-Independent & Spin-Dependent. The coulored regions are example of slices (of the 4-dimensional allowed volume) in the plane £asi vs £,
163

defined in the [0,7r) interval). In Fig. 3-right, instead, obtained allowed region for the case of a particle belonging to the WIMP class with dominant Spin-Independent interaction for some model frameworks is reported 4 ' 5 . Obviously, larger sensitivities than those reported in these figures would e=o

e=itf4

e = wi

e=z.43S

Figure 3. On the left: example of slices (of the 4-dimensional allowed volume) in the plane £asi v s £
be reached when including the effect of other existing uncertainties on the astrophysical, nuclear and particle Physics assumptions and related parameters; similarly, the set of the best fit values would also be enlarged as well. It is worth to note that the DAMA allowed region of Fig. 3-right is also compatible with theoretical expectations in the particular case of a neutralino candidate in MSSM with gaugino mass unification at GUT scale released and with purely Spin-Independent coupling 13 ' 14 . The corollary study of other candidates is given in ref. 4 ' 5 . 3.2. New corollary quest for light bosonic pseudoscalar scalar candidates

and

Recently, DAMA collaboration has also pursued a corollary quest for the candidate particle considering a light (~ keV mass) bosonic candidate, either with pseudoscalar or with scalar couplings, as Dark Matter component

164

in the galactic halo 6 . For these candidates, the direct detection process is based on the total conversion in Nal(Tl) crystal of the mass of the absorbed bosonic particle into electromagnetic radiation. Thus, in these processes the target nuclei recoil is negligible and is not involved in the detection process; therefore, signals from these light bosonic Dark Matter candidates are lost in experiments based on rejection procedures of the electromagnetic contribution to the counting rate. For detailed discussion on this corollary quest for a bosonic candidate, see ref. 6 . The considered Dark Matter light bosonic candidates are non-relativistic since they should be trapped in the galactic halo and they can interact via "Compton - like" effect, "Axioelectric" (photoelectric-like) effect, or Primakoff effect.

Figure 4. Case of a pseudoscalar boson candidate. Left: region allowed at 3cr C.L. in the plane gase vs ma by the DAMA/Nal annual modulation data in the considered model framework. Right: allowed region in the plane gary vs ma (crossed hatched region). All the configurations in this region can be allowed depending on the values of all the gajjIn the plot examples of the expectation for some models have been reported. The solid line corresponds to a particle with lifetime equal to the age of the Universe; at least all the ga-,-,'s below this line are of cosmological interest. 6

In all these processes the total (including the secondary processes: Xrays and Auger electrons) energy release, Erei, in the detector (providing that its detection efficiency is ~ 1 for low-energy electrons and low-energy photons) matches the total energy of the a particle, Ea ~ ma since the a velocity is of the order of 10 _ 3 c. In the case of a pseudoscalar boson, the results can be presented in terms of gaSe, coupling constant of the particle to the electrons, and m Q , the particle mass (see Fig. 4-left). This allowed region is almost independent on

165 the adopted gaUu and gajd coupling constants since the axioelectric effect largely dominates on the other detection processes 6 . The pseudoscalar boson particle can decay into two photons but its lifetime can be also of cosmological interest as shown in Fig. A-right, where the allowed region in the plane gayi vs ma is reported. z10 \~z.

-2

-7

10

-8

10

io~ -10 10

0.5

1

2

3

4 5 6 78910

m h (keV) Figure 5. Region allowed at 3a C.L. in the plane ghfjN vs m/, for a scalar light boson Dark Matter candidate coupled only to the hadronic matter in the given frameworks 6 . Several configurations correspond to particles of cosmological interest (see ref. 6 ) .

Furthermore, in the case of a scalar boson, only a coupling to hadronic matter is considered in order to account for the DAMA observations 6 and the result is an allowed multi-dimensional volume in the space denned by m/j (scalar boson mass) and by all the ghjf coupling constants to quarks. For semplicity here we report only a slice obtained considering the ghNN coupling constants to nucleons (see Fig. 5). This region does not allow a direct information about the effective coupling constant to photons, gh-n, and, therefore, about h lifetime; however, the large number of free coupling constants allows the existence of a large number of h configurations of cosmological interest that can be calculated in some model frameworks 6 . 4. The new D A M A / L I B R A set-up and beyond In 1996 DAMA proposed to realize a ton set-up 15 and a new R&D project for highly radiopure Nal(Tl) detectors was funded at that time and carried out for several years in order to realize as an intermediate step the second generation DAMA/LIBRA experiment (successor of DAMA/Nal) with an exposed mass of about 250 kg. Thus, as a consequence of the results of

166 this second generation R&D, the new experimental set-up DAM A/LIBRA (Large sodium Iodide Bulk for RAre processes), ~250 kg highly radiopure Nal(Tl) crystal scintillators (matrix of twenty-five ~ 9.70 kg Nal(Tl) crystals), was funded at end 1999 and realised. In fact, after the completion of the DAMA/Nal data taking in July 2002, the dismounting of DAMA/Nal occurred and the installation of DAMA/LIBRA started. The experimental site as well as many components of the installation itself have also been implemented (environment, shield of the photomultipliers, wiring, HighPurity Nitrogen system, cooling water of air conditioner, electronics and data acquisition system, etc.). DAMA/LIBRA is taking data since March 2003 and the first data release will, most probably, occur when an exposure larger than that of DAMA/Nal will have been collected and analysed in all the aspects. The highly radiopure DAMA/LIBRA set-up is a powerful tool for further investigation on the Dark Matter particle component in the galactic halo 5 having many intrinsic merits and a larger exposed mass, an higher overall radiopurity and improved performances and increased sensitivity with the respect to DAMA/Nal. At present a third generation R&D effort towards a possible Nal(Tl) ton set-up has been funded and related works have already been started. 5. Conclusion DAMA/Nal has pointed out at 6.3 a C.L. the presence of a Dark Matter particle in the galactic halo by investigating the model independent annual modulation signature in seven annual cycles. No systematic effect or side reaction able to account for the observed effect has been found. Corollary quests for the investigation on the nature of the Dark Matter candidates have also been pursued considering the present poor knowledge on the many astrophysical, nuclear and particle physics related aspects. Here, the new corollary quest for light bosonic candidate has shortly been summarized. At present after a devoted R&D effort, the second generation DAMA/LIBRA (a ~250 kg more radiopure Nal(Tl) set-up) has been realised and put in operation since March 2003 and a third generation R&D toward a possible ton Nal(Tl) set-up, we proposed in 1996 15 , is also in progress. References 1. R. Bernabei et al., Nuovo Cimento A 112, (1999) 545; R. Bernabei et al., Phys. Lett. B 389, (1996) 757; R. Bernabei et al., Nuovo Cimento A 112, (1999) 1541.

167 2. R. Bernabei et al., Phys. Lett. B 424, (1998) 195; R. Bernabei et al., Phys. Lett. B 450, (1999) 448; P. Belli et al., Phys. Rev. D 61, (2000) 023512; R. Bernabei et al., Phys. Lett. B 480, (2000) 23; R. Bernabei et a l , Phys. Lett. B 509, (2001) 197; R. Bernabei el al., Eur. Phys. J. C 23, (2002) 61; P. Belli et al., Phys. Rev. D 66 , (2002) 043503. 3. R. Bernabei et a l , Eur. Phys. J. C 18, (2000) 283; 4. R. Bernabei et a l , La Rivista del Nuovo Cimento 26, (2003) 1-73. 5. R. Bernabei et a l , Int. J. Mod. Phys. D 13, (2004) 2127. 6. R. Bernabei et al., ROM2F/2005/19, to appear on Int. J. Mod. Phys. A. 7. R. Bernabei et al., Phys. Lett. B 408, (1997) 439; P. Belli et al., Phys. Lett. B 460, (1999) 236; R. Bernabei et a l , Phys. Rev. Lett. 83, (1999) 4918; P. Belli et al., Phys. Rev. C 60, (1999) 065501; R. Bernabei et a l , Phys. Lett. B 515, (2001) 6; F. Cappella et al., Eur. Phys. J. direct C 14, (2002) 1; R. Bernabei et al., Eur. Phys. J. A 23, (2005) 7; R. Bernabei et al., Eur. Phys. J. A 24, (2005) 51. 8. R. Bernabei et al., Nuovo Cimento A 103, (1990) 767; Nuovo Cimento C 19, (1996) 537; Astrop. Phys. 5, (1996) 217; Phys. Lett. B 387, (1996) 222 and Phys. Lett. B 389, (1996) 783 (err.); Phys. Lett. B 436, (1998) 379; Phys. Lett. B 465, (1999) 315; Phys. Lett. B 493, (2000) 12; New Journal of Physics 2, (2000) 15.1; Phys. Rev. D 61, (2000) 117301; Eur. Phys. J. direct C 11, (2001) 1; Nucl. Instrum. Methods A 482, (2002) 728; Phys. Lett. B 527, (2002) 182; Phys. Lett. B 546, (2002) 23; in the volume Beyond the Desert 03 (Springer, 2004) 541. R. Bernabei et al., in publication on the Proceed, of the NPA Int. Conf., Debrecen (Hungary), May 2005 9. R. Bernabei et a l , Astrop. Phys. 7, (1997) 73; R. Bernabei et al., Nuovo Cimento A 110, (1997) 189; P. Belli et a l , Astrop. Phys. 10, (1999) 115; P. Belli et a l , Nucl. Phys. B 563, (1999) 97; R. Bernabei et al., Nucl. Phys. A 705, (2002) 29; P. Belli et al., Nucl. Instrum. Methods A 498, (2003) 352; R. Cerulli et al., Nucl. Instrum. Methods A 525, (2004) 535. R. Bernabei et al., ROM2F/2005/18 in publication on Nuclear Inst, and Meth. A 10. K.A. Drukier et al., Phys. Rev. D 33, (1986) 3495; K. Freese et al., Phys. Rev. D 37, (1988) 3388. 11. D. Smith and N. Weiner, Phys. Rev. D 64, (2001) 043502. 12. K. Freese et al., Phys. Rev. Lett. 92, (2004) 11301. 13. A. Bottino et al., Phys. Rev. D 67, (2003) 063519; A. Bottino et al, hepph/0304080; D. Hooper and T. Plehn, MADPH-02-1308, CERN-TH/200229, hep-ph/0212226; G. Belanger, F. Boudjema, A. Pukhov and S. RosierLees, hep-ph/0212227. 14. A. Bottino et al., Phys. Rev. D 69, (2004) 037302. 15. R. Bernabei et a l , Astrop. Phys. 4, (1995) 45; R. Bernabei, "Competitiveness of a very low radioactive ton scintillator for particle Dark Matter search", in the volume The identification of Dark Matter (World Sc. pub.,1997) 574.

D O U B L E BETA D E C A Y M E A S U R E M E N T W I T H C O B R A

J. V. DAWSON ON BEHALF OF THE COBRA COLLABORATION* Department of Physics and Astronomy, University of Sussex, Brighton. BN1 9QH UK E-mail: [email protected]

The COBRA double beta decay experiment is described. Preliminary new limits from a total exposure of 3.82 kg days are presented.

1. Introduction The COBRA concept is to use a large array of Cadmium Zinc Telluride (CZT) semiconductors, to search for neutrinoless double beta decays 1 in which:

(Z,A)^(Z + 2,A) + 2e~

(Ov00 - decay). 3

(1)

Currently, the COBRA arrays are formed using 1cm CZT crystals, each with an active mass of 6.53g. The cube size ensures that for any double beta decay event, the emitted electrons are unlikely to escape the crystal. The COBRA design is modular, allowing easy upgrade and offering the ability to discriminate background events by rejecting events where more than one crystal is triggered. With this ability, a large proportion of high energy gamma ray events can be rejected, as Compton scattering in more than one crystal is likely. Additionally, events that produce a gamma ray cascade such as thermal neutron capture can be noted. Semiconductors, in general, offer good energy resolution and are produced in a clean environment. This is vital for a double beta decay experiment like COBRA. CZT semiconductors operate at room temperature, allowing an experiment which does not suffer the complication of cooling. For an experiment housed underground in a remote laboratory this characteristic is extremely valuable. * http://cobra.physik.uni-dortmund.de

168

169 There is much industrial and scientific interest in the development of CZT. Pixellated devices are used for hard X-ray/7-ray astronomy, medical imaging applications and standard devices sold as general gamma ray detectors. Pixellated devices are of interest for future COBRA detectors, potentially these devices could be used for tracking, with the detector operating as a solid-state TPC. The best energy resolution for double beta events can be achieved if the double beta isotopes are contained within the detector. This is easily obtained with CZT without any doping or isotopic enrichment.

2. Double Beta Isotopes For a double beta decay experiment, CZT offers great potential. Of the 35 known isotopes able to undergo double beta decay, CZT contains 5 of them. Table 1 shows the percentage natural abundance in the crystals, the Q value of the decay and the decay mode. Four of the isotopes can also undergo double positron decay, hence emitting positrons instead of electrons ((3+{3+). Although, due to energy constraints only 106 Cd can emit two positrons, whilst the other isotopes decay via single electron capture of a K-shell electron (/3 + /EC) and double electron capture {EC/EC) modes. In addition, the atoms can be left in excited states after the decay, subsequently emitting gamma rays to return the atom to the ground state. Whilst these events are less likely, the signals produced in the COBRA array are unique. Both electrons would be contained within one cystal, depositing a characteristic energy, and the emitted gamma ray(s) could be detected in a different crystal(s), again with a characteristic energy deposition. The likelihood of a background event mimicking these signature events is small, but further investigation is required to determine how unlikely this is. Decays with high Q-values are desirable in a double beta decay experiment, as the decay rate increases with increasing energy. Also, the background rate generally decreases with increasing energy and the line resolution improves. Three of the isotopes have Q-values greater than 2.5 MeV. One of these isotopes, 116 Cd, has a Q-value greater than the highest energy gamma ray emitted in the decay of natural thorium (2.614 MeV), and so this signal should be unaffected by the natural gamma ray background. 130 Te has the lowest Q-value of these 3 isotopes at 2.529 MeV but the highest natural abundance at 33.8%.

170 Table 1. Isotope

Double Beta Isotopes present in CZT

Nat. Abundance (%)

Q (keV)

0.62

1001

28.7

534

Decay Mode

Cd

7.5

2805

128Te

31.7

868

130Te

33.8

2529

0-00-00-00-00-0-

64

48.6

1096

0+/EC

™Zn u4

Cd

116

Zn

106

Cd

1.21

2771

108

Cd

0.9

231

i20Te

0.1

1722

0+0+ EC/EC 13+/EC

3. Technology The semiconductors are cut out of large boules. The material shows defects. Whilst the material quality has improved with better production techniques, further improvement is likely. The electron mobility is superior to that of the holes, such that the holes are trapped before reaching the cathode, so the signal is formed from the electrons only using the CoPlanar Grid technique (CPG) 2 . The CPG structure forms a virtual Frisch grid below the anode, and the resulting signal is produced by electrons travelling past this virtual grid to the anode. In this way there is almost no position dependence on the signal size. 4. Backgrounds The background events that could be detrimental to the COBRA experiment could be caused by: • High Energy Gamma Rays - of less concern with high Q-valued isotopes, and vetoed by multiple crystal triggers. • Muons - rate lowered by rock shielding and muon scintillator vetoes. • Alpha and Beta Particles - only events close to the crystal surfaces will be able to penetrate and be detected. • Cosmogenics - the time crystals spend on the Earth's surface should be minimised. • Neutrons - MCNP and GEANT 4 simulations resulted in shielding designed to reduce the ambient thermal neutron flux by 8 orders of magnitude. • 2v{3f3 - this is the irreducible background. To reduce contamination

171 of the neutrinoless peak region by the 2i//3/3 continuum, a good resolution is required. The contamination fraction is 3 F = —-^— me

(2)

5. Experimental Status Currently 4 CZT cubes are installed in the laboratory at Gran Sasso, with a cumulative exposure of 3.82 kg days. Preliminary limits for some of the double beta decay modes possible are shown in Table 2. These and limits on 0^/? + /3 + decay modes will be published soon. Previously published results can be found in 4 . Table 2. Isotope

90% C.L.s on Oi//3_/3~ half-lives derived from the fits.

Daughter State

TTC

Cd Cd 116 Cd 116 Cd 116 Cd 116

i3oTe

G.S. 1294 1757 2027 2112 G.S.

Energy (keV) 2805 1511 1048

778 693

Te

536

2529 1993

l28Te

G.S. G.S.

1001

130

70

Zn

868

Events Fitted

T ! / 2 limit (10 1 9 years)

14.79 ± 17.61 75.87 ± 27.7 -7.68 ± 34.04 -46.69 ± 43.8 -12.14 ± 57.98 10.90 ± 17.25 11.82 ± 20.94 31.67 ± 41.55 -37.13 ± 38.15

1.32 0.478 0.977 2.07 0.725 6.34 5.18 3.48 0.0253

6. T h e 64 Array The next stage of COBRA development is the 64 crystal array, shown in Figure 1. This will bring the COBRA mass to 418g. The 64-array will be the first COBRA detector able to search for the characteristic signatures of double beta decays to excited states. Greater background reduction is also anticipated with the ability to reject multiple site events. At present the crystals are being tested, and installation into Gran Sasso should begin in early 2006. 7. Conclusion Whilst the mass of the COBRA experiment is low, the project is in a research and development phase, competitive limits for some neutrinoless

172

Figure 1.

Artist's impression of the COBRA 64 crystal array, courtesy of D. Dobos.

double beta decay modes have been made. The 64 crystal array is being constructed, and is expected to improve the sensitivity to neutrinoless double beta decays by at least an order of magnitude. The 64-array is intended to test the COBRA principles, paving the way towards a large scale experiment with a total mass of 500 kg. In addition to increasing the mass of the experiment, isotopic enrichment of 116 Cd is possible. GEANT 4 simulations are already underway to explore the potential characteristics of a large (40 x 40 x 40) array. References 1. 2. 3. 4.

K. Zuber, Physics Letters B 519, 1 (2001). P. Luke, IEEE Trans. Nucl. Sci. 42, 4 (1995). S. Elliot, P. Vogel, Ann. Rev. Nucl. Part. Sci. , (2002). H. Kiel et al, Nuclear Physics A 723, 499 (2003).

ULTRAVIOLET DIAMOND PHOTODETECTOR

A.DE SIO Dep. of Astronomy and Space Science, University ofFirenze, Largo Enrico Fermi, Firenze 50125 Italy F. MONTI, F. MALVEZZI, E. BURATTINI Dep.of Information Science, University of Verona, Strada le Grazie 15, 37134 Italy A. MARCELLI National Laboratories ofFrascati, National Institute of Nuclear Physics, 00044Frascati, Roma, Italy E. PACE Dep. of Astronomy and Space Science, University ofFirenze, Largo Enrico Fermi, Firenze 50125 Italy Diamond is a very promising material for UV and X-ray sensing applications. Due to its wide bandgap (of 5.5 eV) diamond present low thermal noise, low dark current levels, high visible-XUV signal discrimination and high sensitivity to radiation with energy greater than the bang gap one. Furthermore diamond radiation hardness is very appealing for applications in extreme environments or as monitor of high fluency radiation beams. In order to analyse the performances of diamond based photoconductors under synchrotron UV light a series of measure at the DA
1. Introduction Diamond physical properties offers

unique advantages for

electronic

devices realization. In fact, due to wide bandgap, electronic and physical properties,

and

radiation hardness

diamond

based

devices

should

have

intrinsically high performances and long term stability even if used in extreme and harsh environments as space, waste solution or under high fluencies radiation.

Moreover

diamond

based

detectors

show

high

sensitivity

at

wavelength with energy over band gap (225nm) and, at lower energies signal rejection that can be up to 9 order of magnitude (solar or visible blindness). This

173

174

last characteristic is very appealing in all applications where the suppression of visible signal is needed. Thus diamond is one of the most appealing material for UV and soft X-ray detection [1]. In the last years the growth of single crystal samples was become possible with the improvement of the High Pressure High Temperature (HPHT) and omoepitaxial CVD growth technique. The usage of such kinds of diamond samples has improved the detector sensitivity and reliability [2]. In this work the use of single crystal diamond based photodetectors in synchrotron radiation monitoring is discussed. For such purpose some preliminary measurements session at the UV DAONE-Light synchrotron beam line have been carried on to verify single crystal diamond sensors detectivity, reliability and linearity versus the impinging radiation intensity.

2. Experimental The DAONE-Light DXR-2 beam line can provide radiation in the 800nm 2nm spectral range (corresponding to 1.55 eV to 1 keV). These limits are respectively imposed by the small acceptance angle and to maximum energy reflected at 2.2° by the gold coated mirror. Actually the spectral interval is further on limited, at higher energies, by the sapphire window at the end of the vacuum channel and by the many in air reflections before monochromator. The monochromator is a Jobin-Yvon (mod. HR460MST2-2XM) in Zernic Turner configuration f/5.3 with a focal length of 460mm.The equipped 2400 line/mm holographic grating, optimised for 250nm, allow to work in the 190-600nm spectral region with a resolution of 0.1-0.3%. The radiation intensity at each wavelength has been calibrated with a silicon photodiode. The current signal was amplified and converted into a voltage signal by a Keithley 427 Current Amplifier (with a maximum gain of 109 V/A) and then recorded by a digital multimeter. Different spectra of the beamline light source are reported in Fig 1. The two spectra in square dots are related to two different, non consecutive, measurements performed in order to verify the stability of the photons beam and the reliability of our apparatus. The measurement, in circle dots, is related to the same spectral range while the glass long wave pass filter, that we developed to suppress the second order of the monchormator in the 380nm 500nm spectral range, is applied. II the picture insert the measurements of the transmittance of the filter is reported. The diamond device was based on a commercially available High Pressure High Temperature (HPHT) lb diamond crystal, produced by Sumitomo. It was cut and mechanically polished to {100} within 4°, with final 4 x 4 mm2 size and thickness of 350 um. The sample appears yellowish, due to the 100 ppm nitrogen concentration, but it is optically transparent. A couple of 2 x 2 mm2

175 interdigitated coplanar gold contacts were deposited on one of its surfaces. Contact pitch and width are 20 um. The diamond sensor current signal was measured with a Keithley electrometer 6517A making use, for the device bias, of its intrinsic voltage source up to 1000 Volts.

150

10

-

:__.

150

200

250

300

OOOUX^ I

,

200

350

4Q0

450

500

550

,

l

600

Wavelength (nm) I

250

i

I

i

300

I

350

i

I

400

,

I

450

i

l

500

" ,

550

:

600

Wavelength (nm)

Fig 1: Spectra of the synchrotron light source (in squares). In circle dots is shown the same spectra applying a glass philter which transmittance is reported in the insert.

3. Results and discussion Using a simple photo-generation and collection model for photons fully absorbed in the material [3], the measured current, that is the sum of the dark current and the photoconductive signal (Iph, defined as the difference between the current measure under illumination and in the dark), is given by: Ime S ured=Idark+Iph= Idark +
(1)

where q is the electric charge, F0 is the number of incident photons per second, 0 < 77 < 1 is the quantum efficiency, //ris the mobility-lifetime product, E is the applied electric field and d is the electrode spacing. Defining the photoconductive gain G as the ratio between the drift collection distance L = JUTE and d, it is possible to define the External Quantum Efficiency (EQE) as the ratio EQE = I p h /qF 0 = nL/d = TiG

(2)

176

Therefore EQE can have values higher than unity if L > d. This can be achieved by either using very small electrode gaps or if the UT value is very high as is expected in high quality material. The EQE spectra are estimated by measuring the photocurrent at each wavelength.

150

200

250

300

350

400

450

500

550

600

Wavelength (nm)

Fig2: Sensitivity of the diamond device.

The response of the photodetector to UV continuous monochromatic illumination was then measured in the 200nm 500nm spectral range. The results of these measurements are reported in Fig 2. The sensor EQE is far from the unity but even in this case the detectivity at wavelength above the band gap is good. The rejection to wavelength greater than the band gap is three order of magnitude in 50 nm and is actually limited by the sensitivity of the experimentl apparatus. Therefore the linearity of the detector response as a function of the impinging flux has been measured at 200nm (adopting the second order of the monochromator at 400nm) and 21 Onm. The results obtained are shown in Fig. 3. Both response are linear in the whole flux interval. From the linear fit parameters, combined with the expression (1), is possible to calculate, in an independent way, the EQE at each wavelength. The comparison between the measured EQE values and the calculated one is reported in Tab 1. Table 1. Comparison between the EQE directly measured (EQEm) and calculated from the linear fit (EQEc). The fourth column shows the difference between the two data. EQEc

Diff.

200

EQEm 0.0129

0.0135

4.5%

210

0.00863

0.00813

6%

177

-40

< Q.

~

-60

8.0x10'° 1.0x10" 1.2x10" 1.4x10" 1.6x10" 1.8x10" 2.0x10" 2.2x10" Photon Flux (Ph/s)

Fig3: Linearity of the photoresponse as function of the impinging flux.

4. Conclusion A photoconductive device based on HPHT single crystal diamond has been tested under UV synchrotron light irradiation. Good detectivity, reproducibility and visible solar blindness have been observed. Linear response versus the impinging flux has been recorded at two different wavelengths (200nm and 210nm). Al measurements are corroborating synthetic diamond as one of the most promising material for UV synchrotron light monitoring.

References 1. E. Pace, A. Cidronali, S. Libertino, M. Uslenghi, S. Coffa, M. Focardi, G. Collodi, M. Fiorini, A. Pini, A. De Sio,, SPIE Proc. 4498, 121-130 (2001). 2. A.De Sio, A De Sio, J Achard, A. Tallaire , R S Sussmann, A T Collins, F Silva, and E Pace, App. Phys. Lett. 86, 213504 (2005). 3. S.M.Sze, Physics of semiconductor Devices, 2nd Ed. Wiley, New York, Chapt. 5,7,13 (1981)

STATUS OF THE KATRIN EXPERIMENT OTOKAR DRAGOUN* for the KATRIN Collaboration Nuclear Physics Institute, Academy of Sciences of the Czech CZ 250 68, Rez near Prague, Czech Republic

Republic,

The KATRIN experiment will search for the neutrino mass in a model independent way by the precise measurement of the tritium P-spectrum. A windowless source of gaseous tritium with 95% isotopic purity and column density of 5 1 0 " T2 molecules/cm2 will operate at temperature of 27 K stabilized to 0.1 % by a boiling liquid neon cryostat. The P-particles will be adiabatically guided from the source through two electrostatic retardation spectrometers up to the position-sensitive semiconductor detector by means of about 30 superconducting magnets producing fields of 3-6 T. The main spectrometer (diameter of 10 m, length of 23.3 m, vacuum better than 1 0 " mbar) will operate with resolution of 0.9 eV at 18.6 keV. The retarding high voltage will be monitored to ppm level by precise voltage measurement and independently by a third retardation spectrometer with radioactive sources of monoenergetic electrons. The pre-spectrometer will reduce incoming flux of P-particles from lO'V1 to 1 0 V in the main spectrometer. KATRTN sensitivity on the neutrino mass should reach 0.2 eV after 1000 days measurement, which is an order of magnitude improvement in comparison with the best present limits.

1. Introduction According to the present knowledge the neutrinos oscillate and thus are massive. However, the absolute scale of the neutrino mass eigenstates is still unknown. The most sensitive model-independent method for the neutrino mass determination is a precise measurement of p-spectrum in the endpoint region. The most stringent direct upper limits on m(ve) were achieved using a new type of an electron spectrometer, so called MAC-E-Filter (Magnetic Adiabatic Collimation combined with an Electrostatic Filter). Such instruments were independently developed and extensively utilized in Mainz (Germany) and Troitsk (Russia). These two spectrometers exceeded their predecessors in luminosity of tens percent of 4rc even at the resolution of several eV in measurements of the endpoint region of the tritium p-spectrum at 18.6 keV. The Mainz group reported recently their final result m(ve) < 2.3 eV at 95 % This work was supported by the Grant Agency of the Czech Republic under contract No. 202/03/0889 and by the IRP AV0Z1048050

178

179 confidence level [1]. The Troitsk group, after applying a phenomenological correction for yet unexplained small anomaly in their tritium P-spectrum, reported the upper limit /n(ve) < 2.05 eV [2]. 2. The requirements for a new direct neutrino-mass experiment As for the radioactive source, the tritium P-decay, 3H —>3He+ + e~ + ve, is the best possible choice: 1. The low endpoint energy, £"0 = 18.6 keV, maximizes the fraction of P-decays in the v-mass sensitive region close to E0. This fraction is proportional to (1/ E0)3 and tritium has the second lowest endpoint energy of all Punstable isotopes. 2. The tritium decay is superallowed ('/2+ —*V-i transition) and therefore the shape of its p-spectrum is independent of the nuclear structure. 3. In present experiments, it is not possible to distinguish P-decay branches leading to various excited states of the daughter 3He ion. Thus the recorded P-spectrum is the sum of partial spectra with different endpoint energies and different relative intensities. For simple molecules like T2 and TH, the excitation energies and the population probabilities can be reliably calculated [3] although these calculations could not yet be verified experimentally. The first electronic excited state of (T3He)+ is at 27 eV. This is advantageous for investigation of narrow region just under the p-spectrum endpoint. Nevertheless, there remain the rotational-vibrational excitations of the (T3He)+ ground state caused by the recoil of the 3He nucleus. These excitations with average energy of 1.6 eV and widths of 0.4 eV have to be included into the P-spectrum analysis. Of course, a dreamed source of atomic tritium would be free of such low-energy excitations. 4. Tritium can be investigated in gaseous form and the energy losses of P-particles within the gaseous source can be determined more precisely than in the solid sources. In fact, the last 12 eV region of P-spectrum is free of electrons scattered inelastically within the gaseous source due to scattering threshold for T2 molecules. 5. The tritium half-life of 12.3 years is still acceptable from the point of view of the source specific activity. The electron spectrometer of the MAC-E-Filter type is obviously the most suitable instrument for precise measurement of the endpoint region of tritium Pspectrum thanking to its excellent combination of large luminosity even at high resolution. Since the p-ray intensity falls drastically with increasing energy, the integral character of the measurement does not represent any difficulty.

180

3. The next generation v-mass experiment with sub-eV sensitivity In June 2001, researchers from Germany, Russia, USA and Czech Republic founded the KATRIN Collaboration (Karlsruhe Tritium Neutrino Experiment, [4]). Its aim is to reach the m(ve) sensitivity of 0.2 eV which is one order of magnitude better than that of the best current experiments. The experiment will be located at the Forschungszentrum Karlsruhe (FZK) since there is the only scientific laboratory equipped with a closed tritium cycle and licensed to handle the required amount of tritium. Detailed description of the components and measurement strategy can be found in the KATRIN Design Report 2004 [5]. The KATRIN set-up will consist of three main functional units: (i) the windowless gaseous tritium source with the differential pumping and cryopumping sections, (ii) system of two electrostatic retardation spectrometers, (iii) the position-sensitive electron detector system. 3.1. The windowless gaseous tritium source (WGTS) The molecular WGTS will be a standard source for long-term measurements of P-spectra since it offers the highest possible luminosity and small systematic uncertainties. Such a source represents a technological challenge since the traces of tritium decaying within the spectrometer would increase the detector background. The WGTS activity should be monitored with the highest achievable precision due to single channel mode of the P-spectram taking. A stainless-steel source tube of 10 m length and 90 mm inner diameter will be continuously filled in the middle with molecular tritium of isotopic purity >95% at pressure of 3.4T0"3 mbar. The tritium injection rate will be 1.7-1011 Bq/s and the transport time of T2 molecules through the WGTS will be of the order of 1 s. The column density will be fixed to 5T017 T2 molecules/cm2. The source tube will be cooled to 27 K to decrease the Doppler broadening. The temperature stability of ±30 mK will be achieved by means of boiling liquid neon. Longitudinal magnetic field of Bs= 3.6 T will adiabatically guide the Pparticles to both ends of the tube. A modular differential pumping system and cryo-pumping system will be installed to keep the flow of T2 and TH molecules into the pre-spectrometer smaller than 10"14 mbar 1/s. 3.2. Electrostatic spectrometers KATRIN will utilize three spectrometers of the MAC-E-Filter type: the prespectrometer and the main spectrometer in the beam line of P-particles to analyze their energy and a monitor spectrometer to check the stability of the

181

retarding high voltage by measuring monoenergetic electrons from a suitable radioactive source.

The pre-spectrometer (Fig.l) will work at a fixed retarding energy of about 300 eV below the (3-spectrum endpoint thus reducing the incoming flux of Pparticles from 1010 s"1 to 104 s"1 in the main spectrometer. In order to analyze |3particles with resolution of 0.93 eV, the main spectrometer must have diameter of 10 m and length of 23 m. The unprecedented instrument size together with the required pressure of <10' n mbar in the volume of 1400 m3 represents a challenge since XHV vessels of this size have never been built before. The pre-spectrometer as well as the main spectrometer will be pumped with a combination of turbomolecular pumps (TMP) and non-evaporable getter (NEG) pumps. In order to reach working pressure of 10"u mbar pumping speed of about 5T0 5 1 s"1 is necessary requiring about 3 to 6 km of NEG strips. The effective pumping speed of the TMP has to be of the order of 10000 Is"1. There will be a heating/cooling system that will allow reaching temperature of 350°C needed to activate the NEG pumps.

182 In the previous MAC-E-Filters, the vacuum vessels were grounded and the retarding high voltage was supplied to a system of cylindrical electrodes inside the spectrometer. The KATRTN will utilize a novel approach in which the accurate HV will be put on the wire system of inner electrodes while the vacuum tank put on less precise HV will serve as a guard electrode. The electrodes will be put on slightly more negative potential than the tank itself in order to electrically screen low energy electrons released from the spectrometer wall by cosmic muons and the ambient or intrinsic radioactivity. The inner electrodes should have very small geometrical coverage to prevent the electrodes themselves to become a significant source of secondary electrons. The Mainz MAC-E-filter verified recently the idea of background suppression by means of "massless" inner wire electrodes. Together with the cylindrical-mirror analyzer at Rez [6], it is currently utilized to develop radioactive calibration and monitoring sources. In future, it will serve as the KATRTN high voltage monitor independent of direct voltage measurements. 3.3. Position-sensitive electron detector The KATRTN focal plane detector will be position sensitive with about 400 pixels. Using monoenergetic electrons it will allow mapping the inhomogeneities of the electric retarding potential in the analyzing plane and performing necessary off-line correction of measured p-spectra. The detector requirements are summarized in Tab. 1. The intrinsic detector background < 1 mHz will be achieved by low level of radioactivity in construction materials (especially ceramics), a passive shielding by pure copper and low-activity lead and by active veto against contribution from cosmic rays. Table 1. Requirements for the KATRIN position-sensitive electron detector Property Performance goal Electron energies E 5-100 keV Detection threshold 5 keV Sensitive area 78 cm2 Insensitive areas <5% Detection efficiency e > 0.9 < 600 eV Energy resolution A£DCI at 18.6 keV Position resolution (5x, 5y) 0.3 mm < 5x, 5y < 3.5 mm Time resolution At < 0.5 us Detector thickness d = 200 -300 urn Dead layer entrance window X < 50 nm (Si) Dark currents < 2 nA/cm2 Operation in magnetic fields up to 6 T Operating temperature 100-250 K Use of low-level radioactive tested materials Use of outgassing approved materials

183

A post-accelerating the P-particles up to 25 keV after they have passed through the main spectrometer is considered since it would rise the signal above much of the low-energy backgrounds thus improving the signal to noise ratio by a factor of 5-7. A "dart board" geometry with segmentations along the radial direction and in polar direction is the preferred pixel configuration. Segmented PIN-diodes in a monolithic array having uniform entrance window with minimum insensitive area seem to meet the requirements. 4. Monitoring and calibration of the KATRIN energy scale Precise knowledge of the actual retarded energy EKt for each recorded event is one of crucial conditions for a reliable single-channel measurement of the tritium p-spectrum. A numerical study on the energy scale imperfections [7] showed e.g. that an unrecognized shift of EKi by 0.05 eV between partial Pspectra taken at different times would result in the systematic error of the fitted neutrino mass as large as 0.04 eV. Two methods will be utilized to monitor Elet: 1. The direct measurement of the retarding voltage up to 35 kV using a highprecision voltage divider. With support of the PTB in Braunschweig [8] we aim to develop a HV divider with long-term stability and precision on 1 ppm level. 2. The measurement of energetically well-defined and sharp conversion electron and photoelectron lines of energy not far from the tritium endpoint energy by means of the monitoring spectrometer connected to the same retarding high-voltage as the main spectrometer. The first suitable candidate is the K-conversion line of the 32 keV transition in 83 Kr (energy of 17.8 keV, natural width of 2.7 eV). The 83mKr can be applied in gaseous form for an energy calibration as well as for investigation of potential distribution within the WGTS. Recently, we have determined the binding energy of the K-shell electrons in gaseous krypton to be 14327.26(4) eV and the nuclear transition energy in 83Kr to be 32151.7(5) eV. The sub-monolayer calibration sources of quenched-condensed 83mKr films were employed successfully in the Mainz neutrino experiment [1]. We try to improve their reproducibility by annealing a graphite substrate with laser radiation and applying pre-plated films of neon at temperatures below 10 K. Film thicknesses are measured by laser ellipsometry. Since the 83mRr half-life is only of 1.8 h, we try to prepare for monitoring purposes vacuum evaporated sources of 83Rb/83mKr with more suitable half-life of 86 d. We are also developing an 241Am/Co calibration and monitoring source where the photoelectrons ejected by 26 keV y-rays from the K-shell of metallic cobalt have kinetic energy very close (~ 60 eV) and above the tritium end-point.

184 The natural width of the atomic K-shell in cobalt amounts to favorable 1.3 eV. A feasibility tests were recently completed with the electron spectrometers at Rez and Mainz. 5. Expected uncertainties and sensitivity of the KATRIN experiment A great attention is paid to investigation of systematic uncertainties [5].The physical quantity examined is m\ve). No systematic effect contributing more than Amv2 = 0.0075 eV2 was specified. A quadratic adding of the systematic uncertainties quantified so far yields an estimate of osysttot ~ 0.01 eV2. Since a complete analysis of all systematic effects cannot be performed in this early stage of the experiment we anticipate osysttot < 0.017 eV2 for the m\ve). This corresponds to 5 individual systematic uncertainties of maximal value of Amv2 = 0.0075 eV2 added quadratically. For the analysis interval (Eo-30eV, is0+5eV), background rate of 10 mHz and the net exposure time of 1000 days, the statistical uncertainty amounts to o5tat = 0.018 eV2. Thus otot =0.025 eV2 and m(ve) <0.2 eV at 90 % C.L. with no finite neutrino mass being observed. The true neutrino mass of 0.35 eV (0.30 eV) should be seen as the 5o (3a) effect. The manufacture of the largest components, the WGTS and the main spectrometer vessel started in 2004, the construction of new buildings at the site of the Forschungszentrum Karlsruhe began recently. With more than 100 collaborators from 14 institutions the KATRIN Collaboration endeavors to start tritium measurements in 2008. Aimed m(ve) sensitivity of 0.2 eV should be reached within 5 calendar years. References 1.

2. 3. 4. 5. 6. 7. 8.

Ch. Kraus, B. Bornschein, L. Bornschein, J. Bonn, B. Flatt, A. Kovalik, B. Ostrick, E. E. Often, J. P. Shall, Th. Thummler and Ch. Weinheimer, Eur. Phys. J. C40, 447 (2005) V. M. Lobashev, Proc. 17. Int. Conf. Nuclear Physics in Astrophysics, Debrecen/Hungary, 2002, Nucl. Phys. A719,153c (2003) A. Saenz, S. Jonsell and P. Froelich, Phys. Rev. Lett. 84, 242 (2000) http: //www-ik. fzk.de/katrin/ http://www-ik.fzk.de/katrin/publications/index.html O. Dragoun, M. Rysavy, N. Dragounova, V. Brabec and M. Fiser, Nucl. Instr. Meth. A365, 385 (1995) J. Kaspar, M. Rysavy, A. Spalek and O. Dragoun, Nucl. Instr. Meth. A527, 423 (2004) R. Marx, IEEE Trans. Instr.Meas. 50, 426 (2001)

THE RPC SYSTEM FOR THE OPERA SPECTROMETERS

S. DUSINI* INFN Sezione di Padova, via Marzolo 8, 35131 Padova, Italy

OPERA is one of the two detectors foreseen in the CNGS (CERN Neutrino to Gran Sasso) project, devoted to the detection of i/^ <-» vT oscillations in the parameter region suggested by SuperKamiokande. The tracking system inside the iron yoke of the muon spectrometer makes use of Resistive Plate Chambers (RPC) with bakelite electrodes in a large scale application. We present here full design of the project. Four R P C planes were instrumented and the first tests were performed confirming a good behavior of the installed RPCs in terms of intrinsic noise and operating currents and efficiency. In the paper we present also some results on R P C aging studies.

1. Introduction OPERA is part of the CNGS (Cern Neutrino to Gran Sasso) project and is an experiment dedicated to the observation of the v^ <-> vT oscillation through direct observation of vT charged current interaction. The detector designa is based on a massive target of lead/nuclear emulsions sandwiches complemented by electronic detector (scintillator bars) that allow the location of the event and drive the scanning of the emulsions. The instrumented target is followed by magnetic spectrometer which measures charge and momentum of penetrating tracks. The target section and the relative spectrometer constitutes a super-module. OPERA is made of two super-module for a total target mass of 1.8 Kton. Each spectrometer of OPERA consist of a dipolar magnet made of two iron walls of rectangular cross section (8.75 x 8.2 m 2 ), interleaved by pairs of high resolution trackers. Each wall is made of 12 iron plates 5 cm thick separated by 2 cm of air gap where RPC are installed. The iron is magnetized by a current of about 1200 A circulating in the top and bottom copper coils. The magnetic flux density in the tracking region is 1.55 T with vertical field lines of opposite "On behalf of the RPC-OPERA group (Bologna, LNGS, LNF, Napoli, Padova, Zagreb). For a detailed description of the detector together with its physics potential see [1].

a

185

186 directions in the two magnet walls. The high resolution trackers consist of vertical drift tube planes with an intrinsic resolution of 0.5 mm in the bending direction. The two tracker planes housed between the two magnet walls provide an angular measurement of the track with a 100 cm lever arm. The lever arm for the external trackers is 50 cm. This design leads to a momentum resolution better than 30% in the relevant kinematical domain. Left-right ambiguities are resolved by the two-dimensional measurements of the spectrometer RPCs and by two additional RPC planes with inclined strips (±43.6° with respect to the vertical axis) located just after the two most upper stream drift tubes stations. 2. The O P E R A R P C system The RPC technology used for OPERA is the same as that developed for LHC experiments, BaBar and Argo. Given the very low rate capability required, high resistivity bakelite electrodes are employed (p > 5 x 10 n ft cm at T = 20°). The RPCs planes are composed of 21 RPC of 2.9 x 1.1 m 2 arranged in 7 rows and 3 columns. In order to fit the large bolts that hold together the magnet, 18 RPCs have grooves of semi-circular shape with radius R = 4.3 cm along one of the two long side of the RPC. With this particular shape the RPCs plane acceptance is 97%. The double coordinate read-out is performed by means of 2.6 cm pitch vertical strips, which measure the coordinate in the bending plane, and 3.5 cm pitch horizontal strips. To guarantee the strip adherence to the RPCs, polyester fiber are used to fill the residual of the 2 cm gap space between the iron slab. The RPCs are operated with a gas mixture of argon, tetrafluoreothane and isobutane in volume ratio of 76/20/4 with the addition of 0.5% of SF6Each RPC row is flushed separately at 5 refills per day, using an open flow gas system. Copper and stainless steel pipes are used to avoid excessive humidification of the gas mixture. To control the temperature of the RPC 10 thermistor are placed on each RPC plane between the external side of the read-out strips and the filling material. 3. O P E R A R P C read-out electronics The RPC are operated in streamer mode which provide large amplitude signals (~ 100 mV). The signals from the strips are collected by means of ~ 15 meter long twisted pairs cables and carried to the top of the magnet where the read-out electronics2 is located. The input stage of the RPC Front-End Board (FEB) make use LVDS receivers as comparators. Those

187 devices, because of their differential inputs and high common mode rejection ratios allow a good matching of the signals transmitted through twisted cables. Figure (1) left shows a single LVDS-discriminator channel with the resistive network that provide the threshold bias and the capacitor to reject the DC components. The variable threshold bias is obtained with a DAC with ±300 mV voltage spam. Figure (1) right show the RPC efficiency at the working point as a function of the LVDS threshold voltage.

IVa

R«i=lMonni Rd=ik Ci.ea, C9.C4=100(lF

ID

OS 0.7

:

Ms

t

.

L

r

:

m

5

-

•

Pos. signal

-

O

Ncg. signal

, 3

0.6 0.5 r

S

0.4 7

g\ (-*-aVb

•

0.3 •

0.2

,

,

.

l

100

.

I . 200

, ,

Threshold (mV)

Figure 1. (Left) Input stage of the R P C Front End Board. (Right) R P C efficiency versus LVDS-discriminator threshold voltage for positive and negative signals.

4. Underground test After the installation of the spectrometers the last 4 plane of the first spectrometer were tested (~ 20% of all installed RPCs) in order to check the RPC performance in underground conditions. The measurements were performed with no magnetic field and using self-triggering FEBs of MACRO experiment. Muon tracks were selected requiring the coincidence of signals in three planes (for additional forth plane efficiency study as a function of HV). In the sample collected in about 62 hours, 1307 single-muon, 24 di-muons, 3 four-muons and 1 six-muons events have been found after visual scanning. The number of single muon correspond to the rate of 21.0 ± 0.6 fj,/h. By requiring a very clean track topology (only one strip cluster per view) 973 muons and a rate of 15.6±0.5 fi/h has been measured, in good agreement with the Monte Carlo b prediction of 15.2 ± 0 . 3 n / h . Figure (2) show the angular distribution of the single muon events. The shape of the distributions reflect the detector acceptance and the non-uniformity b T h e underground muon flux simulation is based on an empirical formula for vertical muon intensity modified to account for the Gran Sasso rock structure 3 .

188

of the rock distribution around the laboratory. The average counting rate of the installed RPCs at the operating voltage of 5.6 kV is 17 Hz/m 2 . e
:•?*%

: + *t»

e>o

K- ^ v ^ * * e (deg.)

50

0

-50

H s

(0 j

k

*

, 1 , f*

Weg.)

Figure 2. Angular distributions of the observed underground muons. Angles correspond to the slope of the muon track (a) in the vertical and (b)(c) horizontal projections. 9 = 0 correspond to the zenith direction.

5. Results on R P C aging studies As the RPCs inside the magnet will not be accessible during the entire OPERA lifetime, a big emphasis has been placed on the long term operation and on quality control tests. The good quality of installed RPCs is ensured by a full chain of quality control tests 4 performed in dedicated facilities at the Gran Sasso external laboratory. The quality of the RPCs is determined on the basis of mechanical test (gluing of spacers and gas tightness), electrical test (dark and working currents, resistivity) and response to cosmic ray (RPC efficiency) and noise distribution inside the chamber (noise map). This last test turns out to be the most selective one with a rejection fraction of 14%. The test is performed by locating on the detector the hits acquired in self-triggering mode. The counting in pixel of 3.5 x 3.5 cm 2 are converted into local rate by scaling it to the global RPC counting rate. An RPC is rejected if the maximum of the local rate at the operating voltage exceed 5 Hz/pixel. We call this a hot-spot. Six rejected RPCs have been operated at cosmic ray fluxes for 240 days. All RPC show an increase of the hot-spot rate with the temperature which rise during the summer from 20°C to 27°C. When the temperature return to 20°C the hot-spot rate of the high resistivity RPCs (> 1012flcm) return almost to the initial value. Contrary for the RPC with the lowest resistivity ((0.36 ± 0.01) • 1012ficm) the hot-spot rate increase by a factor 2, while the average pixel rate outside the hot-spot remain stable. This is a clear indication of a local damage of the RPC surface. The damaging of the RPC electrodes has been observed

189 on an RPC (RPCl in [5]) which was operated at the cosmic ray fluxes for 460 days. The RPC showed 10% loss of efficiency in an area corresponding to the noisiest pixels (~ 350 Hz/pixel). The RPC has been opened and the electrode surface visually checked. The surface, in correspondence of the noisiest pixels had change color turning to yellow. In that zone the anode was covered by small, 0.5mm of diameter, white spots. In correspondence on the cathode were visible small whiskers. Figure (3) left show a magnification of the white spots with an optical microscope. The same structure, see figure (3) right, has been observed on the anode of a new RPC rejected by the quality control test in correspondence of the pixel with the highest rate. This show that the noise map test is able to identify defect of the RPC electrode that can, with time, degrade the RPC performance.

LJ Figure 3. Magnification, with an optical microscope, of (left) a white spots on the anode of R P C l of [5] and of (right) the anode electrode, in correspondence of an hot-spot, of a new R P C rejected by the quality control test.

6. Conclusion The OPERA spectrometers make use of RPC in a large scale application. The R P C installation ended in May 2005 and preliminary test on 4 installed RPC planes (~ 20% of all installed RPC) confirm the good behavior of the chambers in terms of intrinsic noise and operating currents and efficiency. The first OPERA spectrometer will be commissioned in spring 2006. References 1. M. Guler et al., CERN/SPSC 2000-028. M. Guler et al, CERN/SPSC 2001-025. 2. M. Ambrosio et al, Nucl. Instrum. Meth. A 533 (2004) 173. 3. M. Ambrosio et al., Phys. Rev. D52 (1995) 3793. 4. A. Bergnoli et al., Nucl. Instrum. Meth. A 533 (2004) 203. 5. G . Barichello et al. Nucl. I n s t r u m . M e t h . A 5 3 3 (2004) 42.

ELECTROMAGNETIC SHOWER RECONSTRUCTION IN EMULSION CLOUD CHAMBER

L. S. E S P O S I T O * Laboratori Nazionali del Gran Sasso, S.S. 17 BIS km. 18.910 67010 Assergi LAquila

Atmospheric neutrino d a t a from the MACRO, Soudan II and Super-Kamiokande experiments are consistent with the hypothesis of i/M —• vT oscillations. The OPERA experiment aims to prove definitively this hypothesis with the direct observation of vT neutrinos in the v^ beam produced at CERN (CNGS). The apparatus, in construction at the Gran Sasso Underground Laboratory, is equipped with electronic detectors and a sensitive target. The target is highly segmented in units, bricks, composed of alternate nuclear emulsion plates and lead sheets. An algorithm to reconstruct electromagnetic showers in a brick was developed. The algorithm was optimized using experimental data from 1, 3 and 6 GeV electron exposures and cross-checked with detailed Monte Carlo simulations. Finally, a neural network was used as electron/pion separator.

1. I n t r o d u c t i o n Neutrino physics is one of the most challenging topics in particle physics. There is convincing evidence for neutrino oscillations, hence neutrino masses and mixing, in particular the combined analysis of atmospheric neutrino experiments favors the hypothesis of a purely v^ —• vT oscillation. The final proof for neutrino oscillations should be the detection of a vT in a terrestrial artificial (almost) pure v^ beam (appearance experiment). The OPERA experiment 1 , in preparation at the Gran Sasso Underground Laboratory, aims at the direct observation of u^ —» vT oscillations in the CNGS (CERN Neutrinos to Gran Sasso) neutrino beam produced at CERN. Since the ve contamination in the CNGS beam is low, OPERA will also be able to study the sub-dominant v^ —» ve oscillation channel. OPERA is a large scale hybrid apparatus divided in two supermodules, each equipped with electronic detectors, an iron spectrometer and a *On behalf of the OPERA Collaboration.

190

191 highly segmented 0.9 kton target section made of "Emulsion Cloud Chamber" units. Each emulsion-lead unit, called brick, is composed of alternate nuclear emulsion films and lead sheets. The use of the ECC satisfies the need of a large mass, required in long-baseline neutrino experiments, and a high precision tracking capability to detect r decays. The target trackers and the spectrometers provide the trigger and a "prediction" of the individual brick where the neutrino interaction happened. The candidate brick is then extracted and disassembled. The emulsion films are photographically processed and then scanned with automatic microscopes for an event-byevent topological classification (event location, search for decays) and further measurements (particle identification and momentum measurement). Very fast automatic scanning systems are needed to cope with the analysis of the large number of emulsion plates associated to neutrino interactions 2 . Tau leptons can decay into a hadron, a muon or an electron. The electron channel has a characteristic signature due to the electromagnetic shower produced by the electron. It is important to have a good particle identification, and, at the same time, to reduce as much as possible the mis-identification, e.g. a low energy pion can undergo a charge exchange process n~p —> w°n and produce an electromagnetic shower. Good electron-pion separation can be achieved by studying the different behavior of these particles in passing through an OPERA brick. At the energies of our interest (1 4- 10 GeV), an electron generates an electromagnetic shower, while a charged pion loses energy essentially through ionization. An experimental study of electron identification using ECC was performed with an electron-enriched ir~ beam 3 . This analysis was based on multiple Coulomb scattering. An experimental study has been performed at DESY where few bricks were exposed to a pure electron beam. A dedicated shower reconstruction algorithm and an electron-pion separator based on a Neural Network have been implemented 4 . To complement this approach, a multiple Coulomb scattering analysis with a pure pion beam has been performed5.

2. Shower reconstruction It is worth to better define the terminology used in the following. An emulsion film is one of the two 43 /im thick emulsion layers put on one side of the plastic base. The emulsion plate is the whole of the plastic base

192 and the two emulsion films. A micro-track is a chain of aligned clusters in one emulsion film. A base-track is built by connecting the two micro-tracks that cross the plastic base. A volume-track is constructed from two or more base-tracks, each one in a different emulsion plate. To reconstruct a shower, all base-tracks inside a cone opened around the primary base-track can be collected. These crude strategy provides a high efficiency in collecting the base-tracks belonging to the shower but can not avoid casual matches with background base-tracks. The bricks analyzed in this work integrated many cosmic ray tracks before and after the exposure, so it was necessary to develop a dedicated algorithm that starts the propagation of the shower from the primary basetrack inside the bricks. Another issue concerns the overlapping between adjacent showers. Since the density of the electrons impinging on the brick is ~100 electrons/cm 2 , the average distance of two primary electron is ~ 1 mm, that is of same order of shower lateral spread. In this "critical" experimental situation, the cone is used as a fiducial volume. All the base-tracks that are outside the cone are not considered during the propagation. Moreover, also base-tracks inside the cone are eliminated if they belong to a volume-track that starts outside the cone itself, i.e. they are considered as "passing-through" and interpreted as cosmic rays. To be efficient in the shower reconstruction, the angular and position tolerances must be increased with respect to a standard propagation of a mip. The Moliere angle and displacement for 100 MeV particles are of the order of 100 mrad and 100 fim, respectively. It is expected that, with this "large" spatial tolerances, background contamination will be reduced at the level of ~0.2/base-tracks/cm 2 by simple computation based on phase space reduction.

3. E x p e r i m e n t a l test Some OPERA bricks were exposed to a pure electron beam at the DESY electron synchrotron. The experimental set-up is made of a pair of scintillation counters used as trigger system, a Multiwire Proportional Chamber to measure beam profile and a lead glass that provides an energy measurement of the electrons. The bricks used in this work were exposed to different energies (1,3 and 6 GeV) with a density of 50-rl00 particle/cm 2 .

193 To reconstruct base-tracks, we used FEDRA 6 , an object-oriented tool based on C + + language developed in ROOT framework. 4. Monte Carlo simulation The Monte Carlo simulation takes into account the brick geometry and materials. All the generated primary and secondary tracks are traced through a brick considering all the relevant physical processes. Secondaries are propagated down to 1 MeV kinetic energy. Two different Monte Carlo codes were used to cross-check their reliability: FLUKA 7 and GEANT3 8 . The results are compatible. For each emulsion film, Monte Carlo provides a pair of hits, one at the entrance of the emulsion, the other at the exit. These pair of hits are then connected to form a micro-track. Micro-tracks are then properly treated by a dedicated software9. A Gaussian smearing was applied both to the micro-track angle and base intercepts. Micro-track reconstruction efficiency was simulated using the mean value of the base-track efficiency as measured with the microscope in other reference plates. The number of clusters for each micro-track was sampled from a Gaussian distribution, whose parameters were obtained from experimental data a . No local effects (e.g. distortions) were simulated. They are strongly dependent on the emulsion development phase and its simulation is a very hard task. In our case the emulsion thickness is small, so the distortions should play a minor role. No smearing of the micro-track z position was applied. The emulsion-base surface, as measured by the automatic system, changes within few microns inside the scanned area. For the simulated data, an "ideal" planar surface was considered. Once smearing and efficiency corrections are applied to the micro-tracks, base-tracks are formed. Each top micro-track is linked with the corresponding bottom one using the tolerances estimated from real data. The same area was measured in all the plates of a brick. This defines the scanned volume. To simulate the scanned volume in the Monte Carlo data, the region that includes all the primary electrons in the first plate of the simulated brick was projected into the downstream plates. All base-tracks that intercept the downstream plates outside this region were discarded. a I n the real data the distribution of the number of clusters of micro-tracks follows a poissonian distribution: however, a poissonian distribution with a mean value ~12-Hl3 is well approximated by a gaussian distribution.

194 From now on, both experimental and simulated data follow the same analysis flow, with the exception of the alignment procedure performed only for real data. The shower reconstruction algorithm has been developed using Monte Carlo simulation and an iterative comparison with experimental data. A reconstructed shower of a 6 GeV electron is shown in Fig. 1.

i Beam direction

Figure 1. A reconstructed shower generated by a 6 GeV electron. The entrance point of the primary electron is the top left angle of the figure according to the beam direction.

5. Results An electron-pion separator based on a Neural Network was implemented and applied to experimental electron data. Monte Carlo simulations were used in order to correctly take into account background tracks in the signal region. The results for electron data and Monte Carlo simulations are given in Table 1. where ee-+e is the efficiency to correctly identify an electron, %_+„• is the probability of electron mis-identification and SNC is the fraction of "not classified" particles. A particle is "not classified" if it has only 1 or 2 base-tracks besides the primary: for these particle the available information is too small.

195 Table 1. Electron identification efficiency for experimental data. Monte Carlo E (GeV) 1 3 6

£e—*e

*/e-7T

0.88 0.91 0.96

0.05 0.05 0.04

Data

SNC

0.08 0.04 0.01

£e—+e

Ve-nr

0.86 0.92 0.95

0.07 0.06 0.04

SNC

0.08 0.02 0.01

6. C o n c l u s i o n s a n d o u t l o o k s A shower reconstruction algorithm was developed and the Neural Network separator implemented. T h e y were applied to t h e electrons d a t a of 1, 3 and 6 GeV bricks exposed at DESY. Electron identification efficiencies are 86%, 92% and 95%, respectively, in good agreement with Monte Carlo results. In t h e future the realization of a hybrid algorithm based on the shower reconstruction and Multiple Coulomb scattering analysis will provide better performances.

Acknowledgments I acknowledge t h e cooperation of the members of the O P E R A Collaboration for the realizations of the experimental test and t h e d a t a sharing and I t h a n k my colleagues of University of Bologna for discussions and suggestions, in particular M a x Sioli and Michela Cozzi. References 1. OPERA Collaboration, OPERA: An appearance experiment to search for uM <-> vT oscillations in the CNGS beam, Experimental proposal, CERN-SPSC-2000-028 (2000); http://operaweb.web.cern.ch/operaweb/documents/index.shtml. 2. G. Rosa et al., Nucl. Instr. Meth. A394, 357 (1997). N. D'Ambrosio [OPERA Collaboration], Nucl. Instrum. Meth. A525, 193 (2004). 3. K. Kodama et cd., Rev. Sci. Instrum. 74, 53 (2003). 4. L. S. Esposito, Study of electron identification in the Emulsion Cloud Chambers of the OPERA experiment, Ph.D. Thesis, University of Bologna (2005). 5. M. Cozzi, Study of pion identification in the Emulsion Cloud Chambers of the OPERA experiment, Ph.D. Thesis, University of Bologna (2005). 6. http://ntslab01.na.infn.it/fedra. 7. A. Fasso, A. Ferrari and P. Sala, SLAC-REPRINT-2000-116. 8. GEANT 3.21, CERN Program Library Long Writeup W5013. 9. http://people.na.infn.it/~marotta/ORFEO.

A LOW B A C K G R O U N D M I C R O M E G A S D E T E C T O R FOR T H E CAST E X P E R I M E N T

P. ABBON, S. ANDRIAMONJE, S. AUNE, D. BESIN, S. CAZAUX, P. CONTREPOIS, N. DUPORTAIL, E. FERRER RIBAS, M. GROS, I. G. IRASTORZA, A. GIGANON, I. GIOMATARIS, M. RIALLOT AND G. ZAFFANELA DAPNIA, Centre d'etudes de Saclay, 91191 Gif sur Yvette CEDEX, Prance G. FANOURAKIS, T. GERALIS, K. KOUSOURIS AND K. ZACHARIADOU Institute of Nuclear Physics, NCSR Demokritos, Aghia Paraskevi 15310, Athens, Greece T. DAFNI* Institut fur Kernphysik,

TU-Darmstadt, Schlossgartenstr. Darmstadt, Germany

9, D-64289

T. DECKER, R. HILL, M. PIVOVAROFF AND R. SOUFLI Lawrence Livermore National Laboratory, 7000 East Avenue, CA94550, USA

Livermore

J. MORALES Instituto de Fisica Nuclear y Altas Energias, Universidad de Zaragoza, Zaragoza 50009, Spain

A low background Micromegas detector has been operating on the CAST experiment at CERN for the search of solar axions during the first phase of the experiment (2002-2004). The detector operated efficiently and achieved a very low level of background rejection (5 x 1 0 - 5 counts keV _ 1 cm _ 2 s - 1 ) thanks to its good spatial and energy resolution as well as the low radioactivity materials used in the construction of the detector. For the second phase of the experiment (2005-2007), the detector will be upgraded by adding a shielding and including focusing optics. These improvements should allow for a background rejection better than two orders of magnitude.

"now at DAPNIA, Centre d'etudes de Saclay, 91191 Gif sur Yvette CEDEX, France

196

197 1. Introduction The CAST (Cern Axion Solar Telescope) collaboration is using a decommissioned LHC dipole magnet to convert solar axions into detectable x-ray photons. Axions are light pseudoscalar particles that arise in the context of the Peccei-Quinn1 solution to the strong CP problem and can be Dark Matter candidates 2 . Stars could produce axions via the Primakoff conversion of the plasma photons. The CAST experiment aims to track the Sun in order to detect solar axions. The detection principle is based on the coupling of an incoming axion to a virtual photon provided by the transverse field of an intense dipole magnet, being transformed into a real, detectable photon that carries the energy and the momentum of the original axion. The axion to photon conversion probability is proportional to the square of the transverse field of the magnet and to the active length of the magnet. Using an LHC magnet (9 T and 9.26 m long) improves the sensitivity by a factor 100 compared to previous experiments. A more detailed description of the principle of the experiment can be found in Zioutas et al.3. For the first phase of the experiment, 2002-2004, three types of detectors have been developed to detect the x-rays originated by the conversion of the axions inside the vacuum of the magnet: a time projection chamber (TPC), a CCD and a Micromegas detector. The CCD detector has been working in conjunction with a mirror system to focus x-rays coming out of the magnet bores improving its signal to background ratio. The analysis of the data collected during 2003 shows no excess of signal over background and has resulted in the most restrictive experimental limit on the coupling constant of axions to photons 4 . The CAST second phase, 2005-2007, will allow us to scan axion masses greater than 0.02 eV/c 2 . This is made possible by filling the magnet bore with a buffer gas (4He or 3 He) allowing the photon to acquire an effective mass. During 2005 a He gas system has been designed and constructed, allowing systematic changes of the He pressure. Cold polypropylene windows have been installed in the magnet bore to minimize the thermal coupling between the cold bore and the outside of the magnet. 2. Performance of the Micromegas detector during phase I 2.1.

Description

The Micromegas detector is a double gap chamber. It consists of a conversion gap separated from an amplification gap by a gauze-light electroformed conducting micromesh. A full description of the detection principle can be found in Giomataris et al.5. For CAST, the conversion gap is 20 (30) mm

198

thick and the amplification gap used was of 50 (100) /txm in the 2003 (2004) data taking. The gas mixture is 95% Argon and 5% Isobutane. The field applied to the amplification gap is about 40 times higher than the conversion field. The charge is collected on a two dimensional readout by means of X and Y strips of 350 /xm pitch. The mechanical pieces of the detector are of Plexiglas because of its low natural radioactivity to limit as much as possible the inherent background of the detector. The Micromegas detector, which operates at atmospheric pressure, is interfaced with the LHC magnet by two thin vacuum windows. These two windows, made of polypropylene, act as a buffer for the pressure gradient and at the same time they need to maximize the x-ray transmission in the energy range of interest for the axion search. The integrated x-ray efficiency of the detector in the solar axion energy spectrum (1-8 keV) is of 73% taking into account the loss in transmission due to the thin windows as well as the conversion efficiency of the gas.

2.2.

Results

The detector has run successfully during the first phase of the experiment. The Micromegas detector records tracking data at sunrise, and during the rest of the day background data is taken. The detector is calibrated daily using a 55 Fe source. The energy resolution at 6 keV is 20% (FWHM). Tracking and background data have been analysed. Signal events (photons with a mean energy of 2-8 keV) have a well defined signature giving a typical cluster in the read out strips and a typical pulse in the micromesh. Background events, coming from cosmic rays and natural radioactivity, give out a bigger cluster in the strips, and the pulse shape in the micromesh is very different favouring an efficient rejection based on the micromesh pulse shape and on the cluster topology. Figure 1 shows the energy spectra for background events for 2003 and 2004 data. In these plots, a flat background is observed with a superimposed peak at 8 keV. This peak has been identified as the copper fluorescence peak coming from the detector materials (micromesh and readout strips). Thanks to the upgrading of the detector in 2004 and the elimination of residual cross-talk between strips, the background rejection was improved of a factor 3 with respect to 2003, achieving a level of 5 x 1 0 - 5 counts k e V - 1 c m - 2 s - 1 . The combined analysis of the three CAST detectors for the 2003 data has resulted in a limit on the coupling constant of photon to axions 4 . This limit, 1.16 x l O - 1 0 G e V - 1 for ma < 0.02 eV, improves

199 [WJ

;

-4-

!00

: ; : ; -*: '•*• ; Figure 1.

i

,

] -+

10

|MV]

Background spectra for the 2003 and 2004 data.

the existing experimental limit by a factor 6. The 2004 combined analysis should allow us to extract a limit close to the theoretical bound coming from astrophysical considerations.

3. The new Micromegas line for phase II During the transformation of the magnet for phase II, the Micromegas group has been designing an upgraded Micromegas detector surrounded by a shielding and coupled to an x-ray optic, as shown in Figure 2. This new line should be installed at the CAST experiment during spring 2006. These upgrades will improve significantly the performance of the detector. First, the x-ray optic, a concentrator with a 1.3 m focal length and 47 mm diameter, will allow us to increase the signal to noise ratio by a factor ~ 100 by focusing the photon flux in a 2 mm spot. This concentrator will consist of 14 nested polycarbonate shells, each 125 mm long and coated with iridium. The optic will transmit ~ 36% of the 0.5-10 keV flux emerging from the magnet bore. Second, the shielding, composed of copper, lead, cadmium, nitrogen and polyethylene, is expected to reduce the background by a factor of 4. Third, by changing the gas of the chamber from Argon to Xenon, the photon conversion probability can be improved by at least ~ 10%. The new detector will be running with the same electronics and acquisition developed for phase I. First tests of a prototype of the detector have started. The complete line, with the integrated optics, is expected to be operational for characterisation tests and calibration at the PANTER x-ray test facility in Munich beginning of 2006. Recent developments on the integration of Micromegas with pixel CMOS readout 6 , may lead to an ultimate upgrade of the detector for the 2007 data taking to profit of the spatial and energy resolution of pixel sensors.

200

Figure 2. The new Micromegas line consisting of a new detector with integrated x-rays optics and shielding.

4. Conclusions A low background Micromegas detector has been operating at the CAST experiment achieving a remarkable background rejection (5 x 1 0 - 5 counts k e V _ 1 c m - 2 s _ 1 ) . For the CAST second phase, an upgraded Micromegas detector with integrated x-ray optics and shielding will be installed at the experiment during spring 2006. These improvements should lead to a reduction of the background level by at least two orders of magnitude. Acknowledgments This work has been performed within the CAST collaboration. our colleagues for their support. Work at LLNL was performed auspices of the U.S. Department of Energy by the University of Lawrence Livermore National Laboratory under Contract No. ENG-48.

We thank under the California W-7405-

References 1. R. D. Peccei and H. R. Quinn, Phys. Rev. Lett. 38, 1440 (1977). 2. P. Sikivie, Int. J. Mod. Phys. D3 S, 1 (1994) and Phys. Rev. Lett. 51, 1415 (1983). 3. K. Zioutas et al., Nucl. Instrul. Meth. A 425, 480 (1999). 4. CAST Coll., "First results from the CERN Axion Solar Telescope (CAST)", Phys. Rev Lett. 94,121301 (2005), hep-ex/0411033. 5. Y. Giomataris, P. Rebourgeard, J. P. Robert and G. Charpak, Nucl. Instrum. Meth. A 376, 29 (1996). 6. P. Colas et al., Nucl .Instrum .Meth. A 535, 506 (2004).

IMAGING OPTICAL ADAPTERS FOR MULTI-ANODE PHOTO MULTIPLIERS DETECTORS * LISA GAMBICORTI + Dep. Of Astronomy and Space Science, Florence University, L.go E. Fermi 2, 50125 Florence, Italy PIERO MAZZINGHI CNR-INOA, L.go E. Fermi 6, 50125 Florence,

Italy

EMANUELE PACE Dep. Of Astronomy and Space Science, Florence University, L.go E. Fermi 2, 50125 Florence, Italy

In several experiments of Nuclear and Neutrino Physics and Cosmic-rays detection, Multi-Anode Photo-Multipliers (MAPMTs) detectors from Hamamatsu corp. are used. Aim of this work is to describe the ancillary optical system, projected to improve the photons collection efficiency of PMT detectors focusing all photons inside the sensitive area. This system optimized with ray-tracing simulations allows to maintain imaging and filtering capability, improving the signal-to-noise ratio by limiting the wave band to the region of interest. The filter glass employing and the multi-layer coating optimization allows to avoid background photons from nearby spectral regions and to minimize the wavelength shifting when the incident angle increases. The transmissivity spectrum has been realized with an high transparency in wavelength range of interest and with a sharp cut-off outside. First prototypes have been realized.

* This work is supported by INFN (Florence section), CNR-INOA and Dep. Of Astronomy and Space Science of Florence University. t Lisa Gambicorti, PhD Student, Dep. of Astronomy and Space Science of Florence University, L.go E. Fermi 2, 50125 Florence, Italy. Tel & Fax +39-055-2307750. E-mail: lisag(aarcetri.astro.it

201

202

1. Introduction The detectors developments involve several and wide areas of applied physics. The employment in experiment in which a high photons collection efficiency is required determines the motion to a research and a develop devoted to an efficiency improvement. Several up to date experiments, in particular in Nuclear and Neutrino physics fields, High Energy and Cosmic-rays detection, primary optical systems with large collection area are used, detecting the photon signal generated by the interaction of such high energy particles crossing a medium, that works as a scintillator. To collect information on the wide angle and to reconstruct the particles trajectories, large pixelated focal planes are required. So, such kind large area focal surface, have to be completely covered with detectors to collect the larger amount of photons [1,2]. Advanced typologies of Multi-Anode Photo Multipliers (MAPMTs) detectors, developed by Hamamatsu Corp., represent an appealing solution to cover large collection area. In particular these detectors will be used for EUSO (Extreme Universe Space Observatory) experiment, a telescope that will observe the Earth's atmosphere with a 60° of FoV (Field of View) from 400 km altitude on board of the ISS (International Space Station). The large volume of atmosphere observed will be necessary to detect a high number of EAS (Extensive Air Shower) events at energies above 1020eV, allowing to obtain a good statistic. A UV filter is required to select the wavelength of collected photons and to improve the signalto-noise ratio in range 300+400 nm. Therefore a possible item to improve collection efficiency is to study an ancillary system to constrain the photons impinging on total area to be focalized only on sensitive area [3,4]. Aim of this work is to describe the project and the optimization of the adapters system designed for three series of detector, which have a sensitive-total area ratio ranging from 45% to 89%. In particular the efficiency reach for the R8900 PMT series, with a sensitive-total area ratio of 79% will be shown. Part of this work deals with the study of materials to obtain an adapter with high transmissivity in wavelength range of interest. 2. The optical adapters A specific ancillary optical system is studied to improve collection efficiency for each R7600/R8900/R7400 PMTs series. An optical adapter lens with hemispherical shape has been projected according to sensitive-total area MAPMTs ratio, allows maintaining the imaging capability that, in several experiments, is a requirement necessary to obtain a well defined image on detector. The adapter materials should have high transmittance in wavelength

203

range of use. The employment in EUSO of the R7600/R8900 MAPMT series is limited by their sensitivity between 300nm and 650 nm, peaked at 420nm, a filter solution must be take in consideration to avoid the 400nm 650nm unwilled signal [5]. Therefore the BG3 Schott glass filter has been selected as EUSO baseline filter[6]. This filter presents a good transmittance in the 300-500nm spectral range. The adapters performances analysis have to be done considering the optical properties of used materials. Main parameters are the bulk transmittance and the refractive index of materials, depending on the wavelength. In particular, the internal transmittance as a function of thickness allows calculating the absorption coefficient (extinction) of material for a given wavelength, as shows the Lambert-Bouguer absorption law [7]. Data on optical and physical properties of materials and filters have been used in ray- tracing simulations. In particular considering the R8900 MAPMT, baseline detectors for EUSO experiment, the projected adapter is an hemispherical lens with a squared section, the same lateral side of each MAPMT series considered to cover the full area of detector. Lateral sides of lens forming a reflecting empty truncated pyramid with four mirrors, in order to collect all the photons coming out of the lens and to fit the lens size with the sensitive area of the detector side. The lens dome is made of PMMA {Poly-Methyl-Metha-Acyilate) with a curved BG3 glass filter glued on top side respectively. To further on reduce the pass band of the BG3 a multi layer coating has been studied and optimized in wavelength range of interest and with a sharp cut-off outside. In case of space experiments, as EUSO, the use of PMMA allows reducing the adapters' weight [8]. 3. The optical performances To evaluate the performance of the filter lens adapter, a set of simulations have been performed with an extended uniform source, emitting at atmospheric fluorescence wavelengths of EUSO. The adapters performances have been obtained, calculating two parameters that together determine the collection system efficiency: the geometric^riG) and the radiometric efficiency*(r|R). Fig. 1 for example shows as look like the top view of R8900 PMT coupled to the optical adapter after the ray-tracing simulation of 105 rays at 357nm. The resulting geometric efficiency increase from 79% to 99% because only a small fraction of rays falls in dead area of detector. Even if the absorption coefficients 'Geometric efficiency r|G: Ratio between rays collected on sensitive area with respect to the number of rays startedfromsource. "Radiometric efficiency nR: Ratio of photons on sensitive area with respect to photons startedfromsource, considering the absorption coefficient of adapter materials.

204

of materials are taken in account the radiometric efficiency reaches the 95%. A simplified prototypes in Fused silica have been fabricated in order to test the optical performances directly.

<4

•

27.0 mm Figure 1. Top view of R8900 PMT coupled to the optical adapter after the rat-tracing simulation of 10s rays at 357nm.

In general applications where are not weight limits, an hemispherical lens adapter optimized for the R8900 MAPMT has a BG3 glass filter, glued on back side of squared hemispherical lens dome of Fused Silica [9,10], A similar study have been realized also for Flat Panel. 4. Conclusion In Nuclear and Neutrino Physics, High Energy and Cosmic-rays detection, the employment of the R7600/R8900/H8500 series MAPMT detectors is limited by their small sensitive-total area ratio. In this paper an optical adapters system is studied to improve the collection efficiency including filtering capabilities, implementing a glass filter to improve the signal-to-noise ratio and reducing the MAPMT wavelengths pass band to the UV range. An optical adapter system,

capable to improve the collection efficiency of each PMTs series detector, has been projected. A filtering adapter solution for the 8900 MAPMT series has been proposed and studied to match also the detector wavelength pass band to the EUSO experiment requests. The simulations show that, using these optical adapters, the detector efficiency improves from 79% to 95%.

205

Due to its general characteristics, the hemispherical lens optical parameters can be modified with different filter solutions, studied and optimized depending by the spectral range of work. References 1. R. Pestotnik et al., Lens-based collection system for a proximity focusing RICH, Nucl. Inst. andMeth A 504 (2003) 237-239 2. A. Petrolini, A comparative study of light collector systems for use with multi-anode photo-multipliers, Nucl. Inst. andMeth A 497 (2003) 314-330 3. V. Bratina, P.Mazzinghi, L.Gambicorti, Overview of possible optical adapter for EUSO, Proc. Spie Vol. 5164, UV/EUV and Visible Space Instrumentation for Astronomy II; Oswald H. Siegmund; Ed.(2003) 100-112 4. P. Mazzinghi, V. Bratina, Concentration of radiation on EUSO focal plane detectors: theoretical limitations and practical implementation, EUSO Tech. Rep., EUSO-Florence collaboration, (2000) 5. Multi-anode Photo-Multiplier tube R7600-M16/M64 series, R8900M25/M36 series, R7400-M256 series, Hamamatsu Photonics K. K., Electron Tube Center, 314-5, Shimokanzo, Toyooka-village, Iwata-gun, Shizuokaken, 438-0193, Japan; http://www.hamamatsu.com 6. Schott Glass BG3 filter glass and Fused Silica datasheet, Schott Glass software code "FILTER'99"http://www.schott.com/ 7. Bouguer-Lambert law. Middleton, W. E. K., 1961: Pierre Bouguer's Optical Treatise on the Gradation of Light, Translation Pfeiffer, H. G., and H. A. Liebhafsky, 1951: J. Chem. Educ, 28, 123-125. Malinin, D. R., and J. H. Yoe, 1961: J. Chem. Educ, 38, 129-131. 8. Gambicorti L., P. Mazzinghi, E. Pace: 2003, UV filters and optical adapters for the EUSO focal surface detectors, 9th Pisa Meeting on Advanced Detectors Frontier Detectors for Frontier Physics, La Biodola, Isola d'Elba (Italy) May 25-31. 9. L. Gambicorti, V. Bratina, A. Gherardi, G. Corti, R. Ciaranfi, P. Mazzinghi, E. Pace, Improving the collection efficiency of EUSO focal surface detectors, Frascati Physics Series Vol. XXXVII (2004) pp. 477- 484. 10. L.Gambicorti, P. Mazzinghi, E. Pace, EUSO Optical adapters: Comparison between UV filtering optical adapters optimized for EUSO focal surface, EUSO- FS Tech. Rep. 012, 20 July 2004 11. L. Gambicorti, P. Mazzinghi, E. Pace, Optical adapters for MAPMT detectors, Nucl. Inst. andMeth A, (Submitted)

A 2D STOCHATSIC M O N T E C A R L O FOR T H E SOLAR M O D U L A T I O N OF GCR: A P R O C E D U R E TO FIT I N T E R P L A N E T A R Y P A R A M E T E R S C O M P A R I N G TO T H E E X P E R I M E N T A L DATA

P. B O B I K 1 , M . J . B O S C H I N I 2 r M . G E R V A S I 2 - 3 , D . G R A N D I 2 A N D P.G. RANCOITA2 INFN

Institute of Experimental Physics, Kosice, Slovak Republic Istituto Nazionale di Fisica Nucleare sezione di Milano, Milano and University of Milano Bicocca, Italy E-mail: [email protected]

Italy

We realized a 2 dimensional (radius and latitude) stochastic MonteCarlo for modelling the Galactic Cosmic Ray (GCR) propagation in the Heliosphere. The model solves numerically the transport equation of particles in the heliosphere, including major processes affecting the heliospheric particle transport: diffusion, convection, adiabatic energy losses and drift of particles. We estimated the cosmic rays spectrum at 1AU using this model formalism: our method allows the study of modulation behaviour in dependence on energy of interstellar particles and heliosphere parameters as Solar wind speed and the tilt angle a of the neutral sheet. We compared our simulations with the Primary CR proton spectra measured by AMS and we realized a fine tuning of our parameters in order to match real data. This work, done for the A > 0 period, can be done also for A < 0, where the best value found for the diffusion coefficient is different.

1. Introduction Models of heliospheric propagation of galactic cosmic rays (GCR) need to take into account all the features of the solar cavity in order to reproduce the observed flux, we can mention for the example the structure of heliospheric magnetic field [1], after Ulysses measurements. We used a Parker field model for the heliosphere [2], that also became time dependent due to the measurements connected to drift effects [3]. First we have produced a ID heliospheric model using a stochastic simulation approach [4,5], useful for testing the method and successful in reproducing the main feature of the *also CILBA, Segrate Milano

206

207

solar cavity. Then we have implemented a two dimensional (radius and latitude) drift model of GCR propagation in the heliosphere [6], depending on the charge sign of particles due to curvature, gradient and current sheet drifts, and time dependent too due to the variation of the measured values of the solar wind velocity in the ecliptic plane (V) and the tilt angle (a). 2. Monte Carlo 2D model with drift effects A 2D stochastic simulation is based on the equation for GCR transport in the heliosphere (without drift terms) [7,8] Coordinates variation in a small time step At for a test particle results:

AT = l d ( r 5 r r ) + VAt + Rg^2K^At H

d

AHi=

— dfi

X_

dr

2*9

At+ RAI 2(1-^)^-At

(1)

Where r is radial distance, \i = cos0 where 0 is the heliolatitude, Ar is the radial variation, A/j, = AcosO is the latitudinal variation of the particle, V is solar wind velocity, Rg is the Gaussian distributed random number with unit variance, At is the time step of calculation. Radial diffusion coefficient is Krr = Kucos2ip + K_isin2ip, where ip is the angle between radial and magnetic field directions [7]. The latitudinal coefficient is Kgg = K±. Parallel and perpendicular coefficients are

K\\ =

K0pKp(9t)^ Kx = (tfj.)o#||

(2)

where KQ — 2 x 10 2 2 cm 2 s _ 1 [9] /? is the particle velocity, Kpffi) take into accounts the dependence on rigidity (in GV), (K±)o is the ratio among parallel and perpendicular diffusion coefficient, BQ ~ 5nT is the value of heliospheric magnetic field at the Earth orbit, and B is the Parker field [10]. In our model the solar wind speed V(9) is a function of the heliolatitude 0 [11]: V = V0(l + sin0), for 0° < 0 < 60° and V = 750 km s~x, for 60° < 0 < 90°. VQ is the velocity of solar wind in ecliptic plane. Drift effects are included through analytical effective drift velocities [12]. The average drift velocity is Vd = V x ( f § ) . 5ft is the CR particle's rigidity. The Parker model allows analytical solution for drift velocities and better

208

evaluation of diffusion tensor, in this spiral field we added to the previous formulas 1 gradient, curvature and drift along the neutral sheet to calculate the position of a test particle during a time step At: Ard = Ar + {vg + v%s Afid = cos arccos(A/x) + arctan

fv$At\

(3)

\ r / Arj is the radial variation with drift effect, Afid is the latitudinal variation of the particle due to drift, vg is the velocity of gradient drift, v™ is the velocity of neutral sheet drift and v$ is the velocity of curvature drift. We use Burger's model [13] as Local Interstellar Spectrum of protons (LIS). Our model has been optimized by fine tuning the parameters Kp($l) and (-ft'x)o of the diffusion tensor, performing long time consuming simulations and comparing results with experimental data (flux spectrum at 1AU). 3. Results 3.1. Comparison

with

data

For A > 0 we selected June 1998 as the best period for testing the model results, because AMS-01 [16] experiment up to now has measured one of the most precise proton spectra, and we obtain as best value (Kj_)0 = 0.025, the same as reported in [14], but the rigidity dependence we have found is different: best spectra is obtained for Kp($l) ~ 9ft0,78 = 5R2-9, where q is the slope of the LIS at high energy. Simulation based on the same set of parameters do not fit the data for A < 0. For negative solar field polarity we obtained, comparing with IMP8 as best values the following set of parameters: (-Kjjo = 0.05 and Kp($t) ~ 3?. For both periods (A > 0 but especially A < 0) we also used Creme96 [15] model for comparison. For positive periods we have used solar wind speed in the range Vo = 100 — 1000km 5 _ 1 and tilt angle in the range a = 10 — 50°. Same set of values has been investigated for A < 0, and for both periods we evaluated the effect of different drift components (gradient, curvature and neutral sheet) turning off each one of them separately and finally all three together. 3.2. Dependence

on a and Vsw

We used the tilt angle a as main parameter for the level of the solar activity: the higher the value of a the lower the expected GCR flux, for both solar

209 V „ 400 km/s - A < 0

Va, = 400 km/s - A>0

!

block donates

o - 15*

block squo'es

a-SC

\

, Sp*CU»m ot 1 AU

Figure 1.

Different values of tilt angle a, A > 0 left and A < 0 right.

field polarity. Besides for the same value of a a higher flux of protons is expected for A > 0. In Fig. 1 we present results of our model for several value of a for A > 0 (left panel) and for A < 0 (right panel). Best values are consistent with measured ones, a ~ 25° for A > 0 and a ~ 15° for A < 0 (from Wilcox Solar Laboratory [17]).

a = 30° - A>0

a = 30° - A < 0 <#dlUP8 fc'u* US s w c t i u m

Speclrum Ot I AU

Figure 2.

Speclfum o l I Ay

Different values of solar wind velocity V, A > 0 left and A < 0 right.

We also evalueted the effect of ecliptic solar wind velocity VQ in both polarities. A higher value of V means higher solar activity and In Fig. 2 we present results of our model for several value of V for A > 0 (left panel)

210

and for A < 0 (right panel). Also for solar wind results are consistent with measured ones: V ~ 350 - 400 km/s for A > 0 and V ~ 450 - 500 km/s for A < 0 (from OMNIWEB [18])

3.3. Dependence

on the drift

effect

To study the effect of the drift velocity terms we considered a period when a = 30° and V = 400 km/s both for A > 0 and A < 0. After computing the flux including all the terms we have excluded the drift from the transport equation. Results are shown in Figure 3 for A > 0 and A < 0. The flux computed with no drift is similar for positive and negative periods, but the drift effect is more evident in positive periods. Minor differences are due to the different parameters used for the two periods.

v* 400 km/s —rt

1

I

I ••

\ / Lp2 U r i o igl s Di i . sq u ire i §t f

Vw 400 km/s - A < 0

= 30° A>0

|\

P

1

Clrun

-i

. •• '^ .

id .Irion les o squa

:

Spectrum at 1 AU

Figure 3.

\

•

t 5

I ^

( Ct'U"

\\

K

N

Specl'um ot 1 AU

Drift effect: turning off single components, A > 0 left and A < 0 right.

We can distinguish three terms involving drift velocities in the heliosphere: gradient, curvature and neutral sheet drifts. We have studied as each one of the drift terms affects the modulated spectrum. As you can see drift effect seems to be dominated by the neutral sheet one (for both solar polarities but stronger for A > 0), we may suppose that drifts play a more important role than convection. We can also see the several drift terms have different effect both in terms of intensity and shape, and that their global effect is not obtained by a simple superposition (neutral sheet drift alone is stronger than all three together).

211 4.

Conclusions

We built a 2D stochastic model of particles propagation across t h e heliosphere. Our model takes into account drift effects and show quantitatively good agreement with measured values. P r o t o n spectra predicted by model are decreasing with increasing tilt angles and flux ratio during positive and negative solar periods is in agreement with measurements. It is possible t o reach a better agreement with measured values by fine t u n i n g of parameters, solar wind velocity and tilt angle, as we did for A > 0 comparing results with AMS-01 data, and for A < 0, using I M P 8 d a t a , obtaining good results especially for A > 0. Agreements with d a t a is b e t t e r for periods with A > 0 t h a n for A < 0, t o investigate this we need accurate d a t a for this phase of the solar activity, t h a t will be available with AMS-02 measurements for C R flux starting at t h e end of 2008 and lasting for 3 t o 5 years.

References 1. L.A. Fisk, J. Geophys. Res. 101, 15547-15553 (1996). 2. R. A. Burger and M. Hitge, American Geophysical Union SH71A-04, Fall Meeting 2002. 3. G. Wibberenz, S. E. S. Ferreira, M. S. Potgieter, H. V. Cane, Space Science Reviews 97, 373 (2001). 4. M. Gervasi et al., Nuclear Phys. B (proc. suppl.) 78, 26 (1999). 5. M. Gervasi et a l , Proceedings of the 26th ICRC SH 3.1.18, 69 (1999). 6. P. Bobik et al., Proceedings of ICSC 2003ESA SP-533, (2003). 7. M. S. Potgieter, J. A. Le Roux, L. F. Burlaga, F. B. McDonald, Astrophysical Journal, Part 1 (ISSN 0004-637X) 403, no. 2, 760 (1993). 8. L. A. Fisk, J. Geophys. Res. 81, 4646 (1976). 9. M. S. Potgieter, J. A. Le Roux, Astrophysical Journal 423, 817 (1994). 10. R.A. Burger and M.S. Potgieter, The Astrophys. J. 339, 501 (1989). 11. J. P. L. Reinecke, C. D. Steenberg, H. Moraal, F. B. McDonald, Advances in Space Research 19, Issue 6, 901 (1997). 12. M. Hatting and R.A. Burger, Adv. Space. Res. 16, No. 9, 213 (1995). 13. R. A. Burger, M. S. Potgieter, B. Heber, J. Geophys. Res. 105, Issue A12, 27447 (2000). 14. J. Giacalone, J. R. Jokipii, Astrophysical Journal 520, Issue 1, 204 (1999). 15. A. J. Tylka et a l , IEEE Trans. Nuclear Sci. 44, No. 6, 2150 (1997). 16. J. Alcaraz et.al., Physics Letters B 490, 27 (2000). 17. http://quake.stanford.edu/ wso/wso.html 18. http://omniweb.gsfc.nasa.gov/form/dxl.html

SPATIAL D I S T R I B U T I O N OF E N E R G E T I C H E A V Y IONS N E A R THE EARTH

M. HAREYAMA, N. HASEBE, K. SAKURAI, S. KODAIRA, N. KAJIWARA, M. TAKANO, T. DOKE, M. FUJII Advanced Research Institute for Science and Engineering, Waseda University, 3-4-1 Okubo, Shinjuku-ku, Tokyo 169-8555, Japan T. GOKA, H. KOSHIISHI, H. MATSUMOTO Institute of Space Technology and Aeronautics, Japan Aerospace Exploration Agency, 2-1-1 Sengen, Tsukuba, Ibaraki 305-8505, Japan T. YANAGIMACHI Department,

of Physics, Rikkyo University., 3-4-14 Nishiikebukuro, Tokyo 171-8501, Japan

Toshimaku,

The observations have been made of energetic heavy ions in various places from 1 to 220 Re by means of the satellites as MDS-1 and GEOTAIL. The elemental abundances from C to Fe were observed by MDS-1 at 1 - 10 Re, GEOTAIL at 10 220 Re in the magnetosphere and beyond. Comparing these abundances with that of galactic cosmic rays (GCRs) measured by ACE/CRIS at 230 Re at LI point beyond the magnetosphere, we have found some differences between the GCR composition and our results. Referring to the results obtained thus far, we show the relative abundances and the spatial distribution obtained by these satellites. We also discuss the relation between the heavy ion composition in GCRs and the particle composition within the magnetosphere.

1. Introduction The observations of galactic cosmic rays (GCRs) and energetic particles in geomagnetic field have been done precisely by various satellites at various positions. However, it is not clear whether there are some differences of the abundances between particles in the magnetosphere and GCRs. This report examines the particle modulation effect in magnetosphere by comparing the relative abundances among three satellite data, Heavy Ion Telescope Array onboard MDS-1 called TSUBASA satellite in the near earth, GEOTAIL in distant geomagnetic tail and CRIS onboard ACE be-

212

213

charge histogram oblakd by HIT/MDS-1

£ 150

2 .-.

il - t , I , t . 0

2

4

6

8

10 12 14 16 18 20 22 24 26 28 charge Z

Figure 1. Charge histogram observed by HIT onboard MDS-1.

Figure 2. Charge Histogram observed by HEP/HI onboard GEOTAIL.

yond the magnetosphere. Then we consider the relation between the GCR composition and heavy ion composition within the magnetosphere. 2. Observations 2.1. HIT onboard

MDS-1

The Mission Demonstration Test Satellite-1 (MDS-1) was in operation from February, 2002 to September, 2003 just after the solar maximum period. It was in the geostationary transfer orbit with a perigee of 500 km and an apogee of 36000 km at the inclination of 28.5° which corresponds to i

a 10 -

2

i

r-1" I

I

i

i

i "i' i

§

i

1

i •^ |

I

i

i

i

f• l

l

• CRIS/ACE (GCRs) o HEP/GEOTAIL A HIT/MDS-1


a 03

t3

S3 3

*

•O

I ° i t

« *

1

N

O

Ne Mg element

Si

Fe

Figure 3. The relative abundances of heavy ions at 100 MeV/n obtained by HIT/MDSl(triangle), HEP/GEOTAIL (circle) and CRIS/ACE(square). These d a t a are normalized at iron.

214 L w 1 ~ 10. Its orbital period was of 10 hours and 35 min and its spin rate was of 5 spins/min. The Heavy Ion Telescope (HIT) onboard the MDS1 was designed to measure energetic ions in space by 2 layers of position sensitive Si detectors and 16 PIN typed Si detectors 1 . The HIT measured heavy ions from He to iron with energy between 20 and 180 MeV/n. This instruments separate heavy ions with 0.06 charge unit for C and O. The charge histogram obtained by HIT is shown in Fig.l. 2.2. HEP onboard

GEOTAIL

The GEOTAIL satellite was launched in July, 1992. It observed energetic particles between 10 Re to 220 Re in a wide range of the geomagnetic tail region. In the first two years, the apogee was from 80 Re to 220 Re in the distant tail in magnetosphere. Later, the apogee was 30 Re to 50 Re in the near-Earth tail region. The inclination has been set at -7 degree. In this work, we use the data set observed in the first three years during the decreasing phase in the solar activity. The instrument called HEP/HI (High Energy Particle) measured heavy ions between boron and iron with 50 ~ 200 MeV/n 2 . The charge histogram obtained by HEP is shown in Fig.2. 2.3. ORIS onboard

ACE

The ACE satellite has been observing energetic particles at 230 Re of Llpoint beyond the magnetosphere. This work used the data set obtained by CRIS instrument between February 2002 and September 2003 as overlapping the HIT data. The CRIS observed from Li to Zn with energy range between 40 and 500 MeV/n. The data set analyzed in this work were taken from CRIS level-2 data of "The ACE Science Center (ASC)". Hereafter, we treat these data as GCRs. 3. Results and Discussion The relative abundance of heavy ions between C and Fe obtained by three satellites is shown in Fig.3. The abundances are normalized at Fe = 1. The GEOTAIL data show comparable to the values from GCRs(CRIS) data as a whole, though there is a small deference at Ne and Si. The GEOTAIL observed heavy ions mostly at the distant tail region of magnetosphere far from the earth. Geomagnetic field in the distant tail is much weaker in the strength than that in the neighborhood of earth's surface

215

e[km]

800

1

" \

f :

1

•00

03

U)

1 H N

\

X

\

X

\

2-

Ar

|

j

x\l

h VL P- > 4

' \

^ \ -.

600

\ v

400

1 i-

(b)

^ \ \

1

1a

x

,

800

"

i-|km|

^\

1

N

J" \

l\ \

i

o o,

\ >

X

Af

A

]

kw

' \V \ \ \

-

^

\

^\

\

300

o' ,o'°

to' 2

io 1 4

>o16

number density [m" ]

IO'8

1010

. 1012

.

. 10'4

,

.

,

.

,018

. to18

number density [m" ]

Figure 4. The variation of atmospheric neutrals with altitude during (a):solar minimum period and (b):solar maximum period reported by U.S. Standard Atmosphere (1976).

and it is almost comparable to interplanetary magnetic field at LI point. Therefore, there is no or just a small modulation effect of geomagnetic field in the distant tail region. The HIT data seem to be enhanced more in the lighter nuclei than iron when compared with GCRs (CRIS) and GEOTAIL a . We consider one of the possible reasons that the heavier ions collide relatively easy with residual atmosphere while traversing to the deeper magnetosphere and then destroy themselves. So, the lighter ions will be enhanced near the earth. The variation of atmospheric neutral composition with altitude is shown in Fig.4 3 . The number density of neutral He and O during the solar maximum period is much larger than that of proton at more than 500 km (maybe up to about 10000 km), whereas proton is dominant component in the region higher than 500 km during the solar minimum period. If our assumption of energetic particles traversing in the magnetosphere while interacting with residual atmosphere is acceptable, their different abundance in the solar minimum phase may be observed when compared that in the solar maximum phase with our result. Then, we must consider this interaction effect to estimate precise GCRs composition even if satellite obsera

However, it is noted that no ions heavier than carbon was observed in L < 2 region, because of cut off rigidity itself.

216 vations in the neighborhood of the earth aim to detect heavy components particularly. 4. Conclusion We showed the relative abundances of heavy ions near the solar maximum period, which were obtained by three satellite, MDS-1 observed at 1 10 Re, GEOTAIL observed at 10 - 220 Re in the magnetosphere and ACE observed 230 Re beyond the magnetosphere. Comparing these data, we have found the results shown in the following : • The relative abundance observed by HEP/GEOTAIL in the distant tail far from the earth is similar to the abundance of GCRs observed by CRIS/ACE beyond the magnetosphere. • The abundance observed by HIT/MDS-1 in the magnetosphere near the earth is different from that in the distant tail and beyond the magnetosphere as GCRs. As one of possible reasons, we need to consider the cause of the observed abundances of nuclei presented here due to the interaction with residual atmospheric neutrals. Acknowledgements We thank the ACE CRIS instrument team and the ACE Science Center for providing the ACE data. Two of authors, M.H. and S.K, were partially supported by a Grant-in-Aid for The 21st Century COE Program (Physics of Self-Organization Systems) at Waseda University from MEXT and one of author, M.H., also supported by a Grant-in-Aid of Scientific Research of JSPS (Grant No. 17740153). References 1. H. Matsumoto et al, Jpn. J. Appl. Phys. , 44, 9A 6870 (2005) 2. T. Doke et al, J. Geomagn. Geodector., 46 713 (1994) 3. U.S. Standard Atmosphere, NOAA, NASA, USAF, Washington, (1976)

A P H O T O N TAG CALIBRATION B E A M FOR T H E AGILE SATELLITE

S. HASAN, M. PREST Universita degli Studi dell 'Insubria e INFN sezione di Milano, Italy L. FOGGETTA, C. PONTONI CIFS, Consorzio Interuniversitario

per la Fisica Spaziale, Torino, Italy

A. MOZZANICA Universita degli Studi di Milano e INFN sezione di Pavia, Italy G. BARBIELLINI, M. BASSET, F. LIELLO, F. LONGO, E. VALLAZZA Universita degli Studi di Trieste e INFN sezione di Trieste, Italy G. BUONOMO, G. MAZZITELLI, L. QUINTIERI INFN, Laboratori Nazionali di Frascati, Italy P. VALENTE INFN, sezione di Romal,

Italy

F. BOFFELLI, P. CATTANEO, F. MAURI INFN, sezione di Pavia, Italy

The AGILE satellite will be launched in 2006 for the study of gamma rays in the energy range 30 MeV-50 GeV. The satellite has to be calibrated using gamma rays of known energy. The calibration facility is being developed at the Beam Test Facility (BTF) at the INFN Laboratories in Frascati. The photons are produced by bremsstrahlung of electrons with a maximum momentum of 750 MeV/c. The electrons are tagged using a dipole magnet whose internal walls are covered by microstrip silicon detectors: depending on the energy loss, they impinge on a different strip once the dipole current has been set to a given value. The correlation between the direction of the electron measured by a pair of x-y silicon chambers and the impinging position on the tagging module inside the magnet allows the tagging of the photon. The paper describes the calibration layout and tests and the results, compared with the Montecarlo simulation, in terms of production rate and energy resolution.

217

218

1. Introduction The AGILE mission x will provide a powerful Observatory for 7-ray astrophysics in the energy range 30 MeV-50 GeV, during the years 2006-2008. AGILE will be launched by the Indian launcher PSLV on a equatorial orbit at 550 km with Malindi as ground base. The scientific instrument is light (wlOO kg) and will be able to detect and monitor 7-ray sources within a large field of view (wl/4 of the whole sky). Given that its launch is foreseen for the beginning of 2006, the calibration of the scientific payload has to be performed before the end of this year. The calibration facility is being developed at the Beam Test Facility (BTF) of the INFN Laboratories in Frascati. Its working principle and the experimental setup are described in Section 2, while Section 3 presents the results of the commissioning of the beamline and the comparison with the simulation.

E> >mi i i M P i h i i

Figure 1.

Working principle of the photon tag calibration beamline.

219 2. T h e calibration facility The DAFNE Beam Test Facility (BTF) 2 is a beam transfer line for bunches of electrons and positrons in the energy range tens of MeV - 750 MeV in a wide range of multiplicities down to the single electron mode. The pulse duration is 10 ns and the pulse rate is 50 Hz. Fig. 1 shows the basic idea of the photon tag beamline. Photons are produced in the bremsstrahlung process by the impinging electron beam. The last dipole magnet in the line is used as a "tagger" of the electrons that have undergone bremsstrahlung in the silicon beam chambers positioned before the magnet itself. These electrons have lost part of their energy in the photon and thus they curve towards the inner walls of the magnet itself, which is completely covered by silicon microstrip detectors.

(a)

(b)

Figure 2. (a) One of the silicon detectors of the beam chambers, (b) One of the tagging modules located inside the bending magnet; the frontend PCB and the silicon detector are clearly visible. The maximum clearance for the modules is 4 cm.

The silicon beam chambers (Fig. 2(a)) before the magnet are the medium for the bremsstrahlung process and they have to reconstruct the direction of the primary beam. They consist in two pairs of 8.9x8.9 cm 2 380 /um thick strip detectors (Micron Semiconductor ltd) with a pitch of 228 /xm. Each tile is readout by 3 TAA1 ASICs (IDEAS, Norway).The two tiles of each pair are positioned in a x-y configuration and the distance between the two pairs is 15 mm. Given the pitch and the analog readout, the spatial resolution is w40 /im while the angular one is w 2 mrad.

220

Fig. 2(b) presents one of the tagging modules, which is a 12 cm long 300 /jm thick silicon detector with a strip pitch of 100 /im and a readout pitch of 300 (im in the central region and 600 ,um in the outer 3 mm. The tagging system consists in 12 tagging modules positioned in the 4.1 cm gap of the bending magnet. Given this value, a very compact design has been chosen: the frontend electronics (3 TAAls per module) is located above the silicon detector and is connected to a pitch adapter which in turn is bonded to the silicon strips. The 12 modules are organized in 2 boxes.

10O

150

200

250

300

350

400

450 500 Energy MeV

Figure 3. Bremsstrahlung spectra for a primary beam energy of 424 MeV. The results from the Geant 3.21 simulation (triangles) are in good agreement with the data measured by the Nal calorimeter (full line).

3. Results In the commissioning phase, the bremsstrahlung spectrum and the tagging system response have been verified using a 15 Xo Nal(Tl) calorimeter. Fig. 3 shows the bremsstrahlung spectrum measured by the calorimeter when one of the tagging modules has detected an electron. The data are compared with the simulation obtained from the code written with Geant 3.21. The mimimum threshold for the calorimeter is of the order of 50 MeV. Fig. 4(a) presents the relationship between the value of the photon energy as measured by the Nal(Tl) calorimeter and the value given by the tagging

221

0

50

100

150

2M

2S0

300

3S0

400

450

-300

-200

-100

0

100

E t p w t w t m r g y M*V

(a)

200

300

400

(Expactwt • mtaturad) • " • ' O Y

500 * V

(b)

Figure 4. (a) Relationship between the value of the photon energy as computed using the tagging modules (X axis) and the one measured by the Nal calorimeter (Y axis). (b) Difference between the expected (tagging modules) and measured (Nal)photon energy.

system. Fig. 4(b) shows the residuals of the same distribution. The system energy resolution is limited at the moment by the calorimeter one, which is of the order of 6% from 424 MeV down to 100 MeV and then increases up to 18%. The system allows to compute the photon energy both when there is a single track in the beam chambers and for multi track events (the maximum allowed multiplicity is around 5 electrons per bunch). In this case, the entrance angle of the electron in the magnet is computed with a weighted average of the track positions in the chambers.

4. Conclusions A photon tagged beam facility has been commissioned at the BTF (LNF, Frascati). Photons are produced with a bremsstrahlung spectrum from 750 MeV electrons and their energy is measured tagging the interacting electron with a silicon spectrometer inside the last bending magnet. The energy resolution is limited at the moment by the calorimeter used in the system calibration phase. The photon production rate strongly depends on beam multiplicity and is of the order of 1 Hz.

222

References 1. G. Barbiellini et al, Nucl. Instr. and Methods A 490, 146-158, 2002. 2. G. Mazzitelli et al., Nucl. Instr. and Meth. in Phys. Res. A 515, 524-542 (2003).

OBSERVATION PROGRAM OF ISOTOPE COMPOSITION IN THE ULTRA HEAVY COSMIC RAYS

N. HASEBEA, M. HAREYAMAA, S. KODAIRAA, K. SAKURAIA, N. YAMASHITAA, T. MIYACHIA, O. OKUDAIRAA, M. TAKANOA, S. TORIIA, T. DOKEA, K. OGURAB, N. YASUDAC, Y. UCHIHORIc, H. TAWARAD, S. NAKAMURAE, T. SHIBATA", T. YANAGIMACHI0 AND S. WANAJO" A) Advanced Research Institute for Science and Engineering, Waseda University, Shinjuku-ku, Tokyo 169-8555, Japan B) National Institute of Radiological Sciences, Inage, Chiba 263-8555, Japan C) College of Industrial Technology, Nihon University, Narashino, Chiba, 275-8576Japan D) High Energy Accelerator Organization, Tsukuba, Ibaraki 305-0801, Japan E) Department of Physics, Yokohama National University, Yokohama, Kanagawa 2408501, Japan F) Department of Physics and Mathematics, Aoyama Gakuin University, Sagamihara, Kanagawa 229-8558, Japan G) Department of Physics, Rikkyo University, Toshima-ku, Tokyo 171-8501, Japan H) Research Center for the Early Universe, Graduate School of Science, University of Tokyo, Bunkyo-ku, Tokyo 113-0033, Japan

The observational study of isotopic composition of ultra-heavy cosmic rays (UH-CRs) is the next important step after the ACE measurement of light isotopes, since no data has been available for both elemental and isotopic abundances in these cosmic rays. These abundances provide the clue for neutron capture nucleosynthesis, which is an essential for understanding the origin of galactic cosmic rays (GCRs). Long-duration-superpressure balloon over Antarctica capable of carrying large scientific payloads for a long period is under planning. A Large Isotope Spectrometer Array (LISA) made of solid-state track detectors (SSTD) will make the first measurement of the elemental and isotopic composition of GCRs for Z > 30 in the energy above 100 MeV/n. 1.

Introduction

The observations of nuclear components consisting of stable and radioactive isotopes with various half-lives in ultra-heavy cosmic rays (UH-CRs) give crucial constraints on a source of galactic cosmic rays (GCRs), the stellar nucleosynthesis, the chemical history of Galactic material, and offer new possibilities for the study of the acceleration and propagation mechanisms of

223

224

charged particles in space [1]. Although the elemental composition of UH-CRs has been observed by several previous experiments [2-4] and recent TIGER experiment [5], the isotopic one of UH-CRs has not been observed yet because of the flux in GCRs is very low. In this paper, we present the observation program of isotopic and elemental compositions of UH-CRs for Z > 30 with the energy above 100 MeV/n using a Large Isotope Spectrometer Array (LISA) made of solid-state track detectors (SSTD) onboard long-duration-super-pressure balloon over Antarctica. 2. Scientific significance for the observation of UH-CR isotopes The element Ne in GCR abundance is enhanced relatively as compared with the solar-system abundance from the previous observations of isotopic compositions for Z < 30 based on most recent experiment ACE/CRIS [6]. This is thought as due to the contribution from Wolf-Rayet stars which usually exist in superbubble [6,7]. The measurement of the abundances of 59Ni and 59Co gives the average delay time greater than about 105 years between nucleosynthesis and acceleration, and suggests that the supernovae (SN) do not accelerate their own ejecta [8]. Furthermore, from the consideration using the condensation temperature of each nuclide, it is pointed out that the abundances of UH-CRs with high condensation temperature are enhanced in GCRs [9]. Therefore, those nuclei which are relatively overabundant in the source composition seem to have been accelerated by multiple SN shocks in the OB-association [7,10]. Moreover, the trans-iron nuclei are produced by the process different from one for below iron-group. Although the elemental composition of UH-CRs was measured by some experiments [2-5], the measured one is modified by the elemental fractionation effect. Pure s, r, /^-process isotopes such as 92Mo, 96Mo, 100Mo will thus be useful for the origin of GCRs. In particular, the separation of pure r and /^-process isotopes for Z > 30 are important since the mass number of adjacent isotopes are separated by two atomic mass units. Moreover, the comparison of the abundance in a source of GCRs (mean life time: about 107 year) with those in primitive metal-poor star (about 1010 yr ago) or solar-system (4.5xl0 9 yr ago) will be useful for understanding of the chemical evolution of the Galaxy. Using the radioactive isotopes in UH-CRs, we will be able to determine the GCRs life time with 127I (TI / 2 =1.7X10 7 yr) or 135Cs (t 1/2 =2.1xl0 6 yr), the delay time between nucleosynthesis and acceleration with 81Kr (TI/ 2 =2.3X10 5 yr), and search for the nearby source due to SN explosion with 79Se (T 1 / 2 =6.5X10 5 yr) or 93Zr (TI/2=1.5xl06yr)-

225

Thus, the observation of isotopes for Z > 30 will enable a great progress in the studies of the origin of GCRs, the stellar nucleosynthesis, the chemical evolution and GCR acceleration and propagation mechanisms. 3. Detector concept and measurement technique The SSTDs such as CR-39 plastic detector or BP-1 glass one are very promising for the large-scale observation of these nuclei in space. This is due to the fact that it is easy to realize a low-cost large detector array. The LISA consists of many SSTD stack modules of 25 cm x 25 cm x 3 cm' and total detector area is 2 m x 2 m (8 x 8 modules). Its geometric factor and weight are ~ 24 m2 sr and ~ 400 kg, respectively. The LISA has the hodoscopes at the top and bottom of instrument, which select only trans-iron events both with the energy below 800 MeV/n at the top of instrument and with the relativistic energy and determine their trajectories by searching for their through-holes. The temperature in the instrument is kept within ± 3 K. We have verified the capability of a CR-39 plastic detector for isotope identification using heavy ion beam from NIRS-HIMAC. In the case of the SSTD, we use the relation of I values and R values to identify the mass, where L denotes the cone length of the etch-pit which is a function of dE/dx, and R is the residual range of an ion. The mass identification of the ion can be achieved using the so-called "L-R pairs" [11] by tracing the trajectory (see Fig. 1). Although the preliminary mass resolution for iron isotopes in CR-39 is 0.28 amu in rms (Fig. 2), by eliminating these errors as much as possible, the mass resolution is expected to be improved to ~ 0.20 amu [12]. The BP-1 glass detector has been used to measure the elemental composition of Pt-Pb group in GCRs as TREK experiment onboard the Russian space station Mir [4]. However, the mass resolution in BP-1 has not been investigated yet. Therefore, a good SSTD with high mass resolution for ultra-heavy isotopes is under development and we will employ the most suit-able detector for the observation of UH-CR isotopes. The handling of solid-state track detector historically has been required a large number of year x person to scan and analyze the etch-pit produced on the detector. Although LISA has very large area of 4 m2 and its handling is difficult, we developed the automated digital imaging optical microscope (HSP-1000, Fig. 3) to scan and analyze the etch-pit produced on the detector, whose image acquisition speed is 50-100 times faster than conventional microscope system [13]. This scanning speed of etch-pit image produced on the detector allows quickly to analyze the hodoscopes and detector stacks. However, the development of new technique to measure the cone length of etch-pit with high accuracy as well as the diameter of etch-pit mouth is required to separate more

226

clearly the adjacent isotopes. Therefore, we have been now developing new three-dimensional scanning system for measuring the cone length with high accuracy.

Fig. 1 Cone length (Lj) and residual Fig. 2 Mass distribution for Fig. 3. High speed auto range (R,) for a particle passing iron nuclei in CR-39. scaning microscope of through detector sheets to the HSP-1000. stopping point. 4. Expected number of particles over Antarctica Since the flux of UH-CRs is very low, the detector with a large collecting power is required to observe such rare events. Definitive measurements are almost avail-able for long duration flight using super-pressure balloon over Antarctica capable of carrying very large scientific payloads for a long extended period, whose advantages are a long duration flight of ~ a few month or more, a relatively small vari-ation of air thickenss and a stability of high altitude without ballast. This super-pressure balloon is currently under development. 10'

8 io' |

10 6

£• io5

5 I Super-pressure balloon over Antarctica in a 60-day of a 4m J ) • D

E > 1000 MeV/n at the top of atmosphere E < 1000 MeV/n at the top of atmosphere

I <° 4 1 io3 •o 102

a

£8

8

io1

Atomic number

Fig 4. Expected UH-CR events observed by balloon over Antarctica. The cut-off rigidities over Antarctica have been calculated to estimate the number of particles by assuming the energy spectrum of iron nuclei observed by

227 HEAO-3/C2 [14]. The number of particles was estimated in a 60-day of a 4 m2 detector area for two cases, the observation of isotopes with the energy below 1000 MeV/n at the top of atmosphere and the observation of nuclear charges with relativistic energy above 1000 MeV/n at the top of atmosphere, respectively. The relative abundance of UH-CRs is taken from HEAO-3/C3 [3] as data. The result of assuming expected number of particles for UH-CRs is shown in Fig. 4.

5. Summary The study of isotopic composition of UH-CRs is the next important step after the ACE measurement of light isotopes (Z < 30), since no data of isotopic abundances in the UH-CR regions nor decisive data of elemental abundance in the region are available. Isotopic abundance is less biased as for the chemical fractionation, which leads us to bring a preferred probe to study the origin. LISA made of SSTDs onboard long-duration-super-pressure balloon over Antarctica will be the first measurement of the elemental and isotopic composition of UHCRs with sufficient statistics for Z > 30 with the energy above 100 MeV/n.

Acknowledgments This work was partially supported by the Grant-in-Aid for The 21st Century COE Program (Physics of Self-Organization Systems) at Waseda University from MEXT. This work was performed as a part of the accelerator experiments of the Research Project (15P146) at NIRS-HIMAC. We would like to express our thanks to the staff of NIRS-HIMAC for their supports.

References 1. N. Hasebe and M. Kobayashi: Symposium on the Origin of Matter and Evolution of Galaxies, 2000, eds. T. Kajino et al, p.94. 2. P. H. Fowler et al, Astrophys. J., 314 (1987) 739. 3. W. R. Binns et al, Astrophys. J., 346 (1989) 997. 4. A. J. Westphal et al, Nature, 396 (1998) 50. 5. S.Geietetal.,Th 29th Int. Cosmic Ray Conf., (2005) (OG1.1). 6. W. R. Binns et al, Astrophys. J., (2005) (in press). 7. J. C. Higdon and R. E. Lingenfelter, Astrophys. J., 590 (2003) 822. 8. M. E. Wiedenbeck et al, Astrophys. J., 523 (1999) L61. 9. N. Hasebe et al., Nucl. Phys., A758 (2005) 292c. 10. R. E. Lingenfelter et al, Astrophys. J., 591 (2003) 228.

228

11. 12. 13. 14.

G. F. Siegmon et al, Nucl. Instrum. Metk, 128 (1975) 461. S. Kodaira et al, Jpn. J. Appl. Phys., 43 (2004) 6358. N. Yasuda et al, Radiat. Meas., 40 (2005) 311. J. J. Engelmann et al., Astron. Astrophys., 233 (1990) 96.

HIGH-ENERGY COSMIC RAYS INVESTIGATED BY AIR-SHOWER MEASUREMENTS WITH KASCADE, KASCADE-GRANDE, AND LOPES A. HAUNGS^4, W.D. APEI/ 4 , P. BADEA A , L. B A H R E N 5 , K. BEKK A , A. BERCUCI C , M. BERTAINA D , P.L. BIERMANN 5 , J. BLUMER A ' F , H. B O Z D O C 4 , I.M. BRANCUS G , M. BRUGGEMANN G , P. BUCHHOLZ G , S. B U I T I N K 5 , H. B U T C H E R 6 , A. CHIAVASSA D , K. DAUMILLER A , A.G. DE BRUYN B , C M . DE V O S B , F. DI P I E R R O D , P. D O L l / , R. ENGEL A , J. ENGLER A , H. F A L C K E 3 " 6 - 5 , H. GEMMEKE 7 , P.L. GHIA J , H.-J. GILS 71 , R. GLASSTETTER*", C. GRUPEN G , D. HECK A , J.R.HORANDEL F , A. H O R N E F F E R 7 ^ , T. H U E G E A ' F , K . - H . K A M P E R T K , G.W. K A N T B , H.O. KLAGES A , U. K L E I N L , Y. K O L O T A E V G ,

Y. KOOPMAN B , O. KROMER 7 , J. KUIJPERS 7 7 , S. LAFEBRE 7 7 , G. MAIER A , H.J. MATHES A , H.J. MAYER A , J. MILKE A , B. MITRICA C , C. M 0 R E L L 0 7 , M. MULLER A , G. NAVARRA 75 , S. NEHLS 71 , A. NIGL 77 , R. OBENLAND 71 , J. OEHLSCHLAGER A , S. OSTAPCHENKO 71 , S. OVER G , H.J. P E P P I N G 5 , M. PETCU* 7 , J. PETROVIC 7 7 , T. PIEROG 71 , S. PLEWNIA 71 , H. REBEL 71 , A. RISSE M , M. R O T H F , H. SCHIELER 71 , G. SCHOONDERBEEK B , O. SIMA C , M. STUMPERT F , G. T O M A c , G.C. TRINCHERO J , H. ULRICH 71 , J.VAN BUREN 71 , W.VAN CAPELLEN 5 , W. WALKOWIAK G , A. WEINDL 71 , S. WIJNHOLDS 5 , J. WOCHELE 71 , J. ZABIEROWSKI M , J.A. ZENSUS 5 , D. ZIMMERMANN G Institut fur Kernphysik, Forschungszentrum Karlsruhe, Germany B ASTRON Dwingeloo, The Netherlands c NIPNE Bucharest, Romania D Dpt di Fisica Generale dell'Universita Torino, Italy E Max-Planck-Institut fur Radioastronomie, Bonn, Germany Institut fur Experimentelle Kernphysik, Uni Karlsruhe, Germany, Fachbereich Physik, Universitat Siegen, Germany Dpt of Astrophysics, Radboud Uni Nijmegen, The Netherlands 1 IPE, Forschungszentrum Karlsruhe, Germany 1st di Fisica dello Spazio Interplanetario INAF, Torino, Italy K Fachbereich Physik, Uni Wuppertal, Germany Radioastronomisches Institut der Uni Bonn, Germany Soltan Institute for Nuclear Studies, Lodz, Poland

229

230 Recent results from the multi-detector set-up KASCADE on measurements of cosmic rays in the energy range of the so called knee (at ss 3 PeV) are presented. The multidimensional analysis of the air shower data indicates a distinct knee in the energy spectra of light primary cosmic rays and an increasing dominance of heavy ones towards higher energies. This provides, together with the results of large scale anisotropy studies, implications for discriminating astrophysical models of the origin of the knee. To improve the reconstruction quality and statistics at higher energies, where the knee of the heavy primaries is expected at around 100 PeV, KASCADE has recently been extended by a factor 10 in area to the new experiment KASCADE-Grande. LOPES is set up at the location of the KASCADE-Grande experiment and measures radio pulses from extensive air showers. LOPES is designed as digital radio interferometer using high bandwidths and fast data processing and profits from the reconstructed air shower observables of KASCADE-Grande.

1. Introduction The all-particle energy spectrum of cosmic rays shows a distinctive discontinuity at few PeV, known as the knee, where the spectral index changes from —2.7 to approximately —3.1 (Fig. 1). At that energy direct measurements are presently hardly possible due to the low flux, but indirect measurements observing extensive air showers (EAS) are performed. Astrophysical scenarios like the change of the acceleration mechanisms at the cosmic ray sources (supernova remnants, pulsars, etc.) or effects of the transport mechanisms inside the Galaxy (diffusion with escape probabilities) are conceivable for the origin of the knee as well as particle physics reasons like a new kind of hadronic interaction inside the atmosphere or during the transport through the interstellar medium. It is obvious that only detailed measurements over the whole energy range of the knee from 10 14 eV to 10 18 eV and analyses of the primary energy spectra for the different incoming particle types can validate or disprove some of these models 1 . The multi-detector system KASCADE-Grande (KArlsruhe Shower Core and Array DEtector and Grande array) 2 , approaches this challenge by measuring as much as possible independent information from each single air-shower event. Recently, the original KASCADE experiment 3 was extended in area by a factor 10 to the new experiment KASCADE-Grande. KASCADE-Grande allows now a full coverage of the energy range around the knee, including the possible second knee at energies just below 10 18 eV. With its capabilities KASCADE-Grande is also the ideal testbed for the development and calibration of new air-shower detection techniques like the measurement of EAS radio emission, which is performed in the frame of the LOPES project 4 .

231 10

E—i—i—i

i i i 11|

1—r

•

PROTON

•

RUNJOB

10"

,.

. j

d , r e C , d a t a

T • •

Akeno

.

KASCADE He

HiRes I

•

KASCADE heavy

HiRes II

O

* «r

10'

KASCADE (QGSJET 01) KASCADE H

KASCADE (SBYUZI) KASCADE H KASCADE He KASCADE heavy

AGASA

10'

10'

T=

•.«S».J;°°

in"

10

n m Auovevntam

10'

il

10"

i i I mill—i

10*

i i mill—I_L_I m i l l — i

10' 5

10' 6 Energy

i i iimr—i

10"

i i mill—i

10' 8

i i Mini—i

10' 9

i i mill—

10 20

10 2 '

(eV/particle)

Figure 1. Primary cosmic ray flux and primary energy range covered by KASCADEGrande. Results of the KASCADE data analyses are also shown (see text).

2. The KASCADE-Grande and LOPES Experiments These experiments are located at the Forschungszentrum Karlsruhe, Germany, (49.1°n, 8.4°e, llOma.s.l.) and measure extensive air showers in a primary energy range from 100 TeV to 1 EeV. Their combination provides multi-parameter measurements on a large number of observables concerning electrons, muons at 4 energy thresholds, hadrons, and the radio emission. The main detector components are the KASCADE array, the Grande array, and the LOPES antenna array. The KASCADE array measures the total electron and muon numbers (Ei,,kin > 230 MeV) of the shower separately using an array of 252 detector stations containing shielded and unshielded detectors at the same place in a grid of 200 x 200 m 2 . The excellent time resolution of these detectors allows also decent investigations of the arrival directions of the showers in searching large scale anisotropics and, if existent, cosmic ray point sources. The KASCADE array is optimized to measure EAS in the energy range of 100 TeV to 80 PeV. A muon tracking detector measures the incidence angles of muons relative to the shower arrival direction. These measurements provide a sensitivity to the longitudinal development of the showers. The hadronic core of

232 Muon Tracking Detector

.Radio Antenna (LOPES-10)

rrwrrir

T

•

1

•

1

LOPES-ld

0)

o

— • o O 7

CM I

Grande stations ' Detector Station \ Grande Station Array Cluster Electronic Station

Figure 2. Sketch of the KASCADE experiment: field array, muon tracking and centra! detector. The locations of 10 radio antennas of LOPES 10 are also displayed, as well as three stations of the Grande array.

-600

-400

-200

X[m]

W->E Direction

Figure 3. Sketch of the KASCADEGrande and LOPES 10 layout inside the KASCADE-Grande experiment with its 37 Grande stations.

the shower is measured by a 300 m 2 iron sampling calorimeter installed at the KASCADE central detector: Three other components offer additional valuable information on the penetrating muonic component at different energy thresholds. The complementary information of the showers measured by the central and the muon tracking detectors is predominantly being used for tests and improvements of the hadronic interaction models underlying the analyses 5 . The multi-detector concept of the KASCADE experiment (Fig. 2), which is operating since 1996 has been translated to higher primary energies through KASCADE-Grande 6 . The 37 stations of the Grande Array (Fig. 3) extend the cosmic ray measurements up to primary energies of 1 EeV. The Grande stations, 10 m 2 of plastic scintillator detectors each, are spaced at approximative by 130 m covering a total area of ~ 0.5 km 2 . For the calibration of the radio signal emitted by the air shower in the atmosphere an array of first 10 and meanwhile 30 dipole antennas (LOPES) is set up on site of the KASCADE-Grande experiment 7 . The antennas operate in the frequency range of 40-80 MHz. The radio data is collected when a 'large event' trigger is received from KASCADE which translates to primary energies above 10 16 eV.

233

3. K A S C A D E Results The KASCADE data analyses aims to reconstruct the energy spectra of individual mass groups taking into account not only different shower observables, but also their correlation on an event-by-event basis. The content of each cell of the two-dimensional spectrum of reconstructed electron number vs. muon number is the sum of contributions from the individual primary elements. Hence the inverse problem g{y) = / K{y,x)p(x)dx with y = (Ne,NJ?) (see Fig. 4) and x = (E,A) has to be solved. This problem results in a system of coupled Fredholm integral equations of the form

+oo

d\gN?dW

= I U / 3T& • PAOsNe , l g < ^

| lg£) • d ]gE where

—oo

the probability PA is obtained by Monte Carlo simulations on basis of two different hadronic interaction models (QGSJET01 8 , SIBYLL2.1 9 ) as options embedded in CORSIKA 10 . By applying these procedures (with the assumption of five primary mass groups) to the experimental data energy spectra are obtained as displayed in Fig. 1, where the resulting spectra for primary oxygen, silicon, and iron are summed up for a better visibility. A knee like feature is clearly visible in the all particle spectrum, which is the sum of the unfolded single mass group spectra, as well as in the spectra of primary proton and helium. This demonstrates that the elemental composition of cosmic rays is dominated by the light components below the knee and by a heavy component above the knee feature. Thus, the knee feature originates from a decreasing flux of the light primary particles 11 . Comparing the unfolding results based on the two different hadronic interaction models, the model dependence when interpreting the data is obvious. Modeling the hadronic interactions underlies assumptions from particle physics theory and extrapolations resulting in large uncertainties, which are reflected by the discrepancies of the results presented here. In Fig. 4 the predictions of the Ne and AT'r correlation for the two models are overlayed to the measured distribution in case of proton and iron primaries. It is remarkable that all four lines have a more or less parallel slope which is different from the data distribution. There, the knee is visible as kink to a flatter Ne-Nff dependence above lg Nff « 4.2. Comparing the residuals of the unfolded two dimensional distributions for the different models with the initial data set we conclude that at lower energies the SIBYLL model and at higher energies the QGSJET model are able to describe the correlation consistently, but none of the present models gives a contenting description of the whole data set. These findings are confirmed by detailed investigations of further shower observables measured by KASCADE 5 .

234

Figure 4. Two dimensional electron (Ne) vs. muon (JV£r = number of muons in 40200m core distance) number spectrum measured by the KASCADE array. T h e lines display the most probable values for proton and iron primaries obtained by CORSIKA simulations employing different hadronic interaction models.

Investigations of anisotropics in the arrival directions of the cosmic rays give additional information on the cosmic ray origin and of their propagation. Depending on the model of the origin of the knee and on the assumed structure of the galactic magnetic field one expects large-scale anisotropics on a scale of 1 0 - 4 to 1 0 - 2 in the energy region of the knee. The limits of large-scale anisotropy analyzing the KASCADE data are determined to be between 1 0 - 3 at 0.7 PeV primary energy and 1 0 - 2 at 6 PeV 12 . These limits were obtained by investigations of the Rayleigh amplitudes and phases of the first harmonics. Taking into account possible nearby sources of galactic cosmic rays like the Vela Supernova remnant the limits of KASCADE already exclude particular model predictions. The interest for looking to point sources in the KASCADE data sample arises from the possibility of unknown near-by sources, where the deflection of the charged cosmic rays would be small or by sources emitting neutral particles like high-energy gammas or neutrons. The scenario for neutrons is very interesting for KASCADE-Grande, since the neutron decay length at these energies is in the order of the distance to the Galactic center. In KASCADE case no significant excess was found 13 . 4. Status of KASCADE-Grande Fig. 5 shows, for a single shower, the lateral distribution of electrons and muons reconstructed with KASCADE and the charge particle densities

235 Event 10160542

KASCADE muon detectors

5 5.5 6 6.5 muon number lg(N)

cote distance [m]

Figure 5. Particle densities in the different detector types of KASCADE-Grande measured for a single event.

Figure 6. Reconstructed electron vs. number size spectrum after one year of d a t a taking of KASCADE-Grande.

muon

measured by the Grande stations. This example illustrates the capabilities of KASCADE-Grande and the high quality of the data. The KASCADEGrande reconstruction procedure follows iterative steps: shower core position, angle-of-incidence, and total number of charged particles are estimated from Grande Array data; the muon densities and the reconstruction of the total muon number are provided by the KASCADE muon detectors 14 . At the KASCADE experiment, the two-dimensional distribution shower size - truncated number of muons played the fundamental role in reconstruction of energy spectra of single mass groups. Hence, Figure 6 illustrates the capability of KASCADE-Grande to perform an unfolding procedure like in KASCADE. A hundred percent efficiency for KASCADE-Grande is reached for all primary particle types for energies above 2 • 1016 eV, providing still a large overlap with the KASCADE energy range. KASCADE-Grande started end of 2003 with combined measurements of all detector components. 5. LOPES Data Analysis The LOPES-10 data set is subject of various analyses using different selections: With an event sample obtained by stringent cuts the proof of principle to detect air showers in the radio frequency range was given 15 . With showers fallen inside KASCADE the basic correlations with shower parameters are shown 16 . Further interesting features are investigated with a sample of very inclined showers 17 and with a sample of events measured

236

during thunderstorms 18 . Additionally, an analysis is performed correlating the radio signals measured by LOPES-10 with EAS events reconstructed by KASCADE-Grande with distant cores included 19 . Grande is taking data in coincidence with KASCADE and LOPES and enables to reconstruct showers with primary energies up to 10 18 eV and with distances between shower core and the LOPES-10 antennas up to 700 m. A crucial element of the analysis method is the digital beamforming which allows to place a narrow antenna beam in the direction of the cosmic ray event. This is possible because the phase information of the radio waves is preserved by the digital receiver and the cosmic ray produces a coherent pulse. This method is also very effective in suppressing interference from the particle detectors which radiate incoherently. The procedure of time shifting of the radio signals in the antennas is relatively safe when based on the values provided by the reconstruction for shower core and shower axis using the data of the original KASCADE field array. Due to the high granularity of the detector stations the accuracy of core and direction reconstruction is high enough to obtain a good coherence of the radio signals. The shower reconstruction using information from the Grande array is re-

Figure 7. Radio signals in the 10 antennas after first beamforming Cupper panel) and after optimized beamforming (lower panel).

Figure 8. Correlation of primary energy reconstructed with KASCADE-Grande with the reconstructed height of the radio pulse signal,

237

quired for shower cores outside KASCADE. The Grande stations, 10 m 2 of plastic scintillator detectors each, are spaced at ~130 m and cannot assure an accuracy comparable with the original KASCADE array. So, a so-called 'optimized' beamforming is performed, which searches for a maximum coherence by varying the core and the direction around the values provided by the Grande reconstruction. Fig. 7 shows the raw radio signal of an example event before and after the optimized beamforming. The event has a mean core distance to the antennas of « 100 m, a primary energy of 5 • 10 17 eV, and a zenith angle of 40°. An noticeable increase in coherence is seen after the optimized beamforming. Fig. 8 shows the final result of this analysis, where the radio pulse amplitude per unit bandwidth (e„) is plotted versus the primary energy reconstructed by KASCADE-Grande. A clear correlation between radio pulse height and primary particle energy is visible. One has to recognize that at LOPES 10 the antennas are not absolute calibrated, i.e. ev is a relative value, only, just as well LOPES 10 measures only one (east-west) polarization direction of the signal.

6. Conclusions and Outlook From KASCADE measurements we do know that at a few times 10 15 eV the knee is due to light elements, that the knee positions depend on the kind of the incoming particle, and that cosmic rays around the knee arrive our Earth isotropically. Additionally KASCADE could show that no current hadronic interaction model which are unavoidably needed for the interpretation of air shower data, describes very well cosmic ray measurements in the energy range of the knee and above. These model uncertainties are due to the lack of accelerator data at these energies and especially for the forward direction of collisions. The extension of KASCADE to the KASCADE-Grande experiment, accessing higher primary energies, is expected to solve the question of the existence of a knee-like structure corresponding to heavy elements. KASCADE-Grande keeps the multi-detector concept for tuning different interaction models at primary energies up to 10 18 eV. To improve further the data quality a self-triggering, dead-time free FADC-based DAQ system will be implemented in order to record the full time evolution of energy deposits in the Grande stations 20 . At present, LOPES operates 30 dipole radio antennas (LOPES-30) positioned inside or nearby KASCADE 21 . The longer baseline of the set

238 u p will allow a more detailed investigation of t h e lateral extension of the air shower radio signal. Additionally these antennas will be absolute calibrated, so t h a t the obtained pulse amplitude can be directly compared with theoretical expectations for the radio field strength 2 2 .

References 1. A. Haungs, H. Rebel, M. Roth Rep. Prog. Phys. 66 1145 (2003). 2. G. Navarra et al. - KASCADE-Grande coll. Nucl. Instr. Meth. A 518 207 (2004). 3. T. Antoni et al. - KASCADE coll. Nucl. Instr. Meth. A 513 429 (2003). 4. A. Horneffer et al. - LOPES coll. Proc. SPIE 5500 129 (2004). 5. A. Haungs et al. - KASCADE coll. Proc. XIIIISVHECRI 2004, Pylos, Nucl. Phys. B (Proc. Suppl.) in press (2004), astro-ph/0412610. 6. A.F. Badea et al. - KASCADE-Grande coll. Nucl. Phys. B (Proc. Suppl.) 136 384 (2004). 7. A. Haungs et al. - LOPES coll. Proc. 22nd Texas Symposium, Stanford, CA, Dec. 2004 eConf C041213 2413 (2004). 8. N.N. Kalmykov, S.S. Ostapchenko Phys. Atom. Nucl. 56 346 (1993). 9. R. Engel et al. 26th ICRC, Salt Lake City, Utah p.415 (1999). 10. D. Heck et al. Report FZKA 6019, Forschungszentrum Karlsruhe (1998). 11. T. Antoni et al. - KASCADE coll. Astrop. Phys. 24 1 (2005). 12. T. Antoni et al. - KASCADE coll. Astrophys. J. 604 687 (2004). 13. T. Antoni et al. - KASCADE coll. Astrophys. J. 608 865 (2004). 14. R. Glasstetter et al. - KASCADE-Grande coll., Proc. of 29th ICRC, Pune, India (2005). 15. H. Falcke et al. - LOPES coll., Nature 435 313 (2005). 16. A. Horneffer et al. - LOPES coll. Proc. of 29th ICRC, Pune, India (2005). 17. J. Petrovic et al. - LOPES coll. Proc. of 29th ICRC, Pune, India (2005). 18. S. Buitink et al. - LOPES coll. Proc. of 29*h ICRC, Pune, India (2005); 19. A.F. Badea et al. LOPES coll. Proc. of 29th ICRC, Pune, India (2005); A.F. Badea et al. LOPES coll. Proc. of 29th ICRC, Pune, India (2005). 20. W. Walkowiak et al. - KASCADE-Grande coll. Proc. of IEEE Nuclear Science Symposium, Rome (2004). 21. S. Nehls et al. - LOPES collab., Proc. of 29th ICRC, Pune, India (2005). 22. T. Huege and H. Falcke, Astrop. Phys. 25 116 (2005).

SCINTILLATING OPTICAL FIBERS FOR ASTROPARTICLE PHYSICS W. R. BINNS, M. H. ISRAEL, K. RIELAGE1 Dept. of Physics and McDonnell Center for the Space Sciences, Washington University St. Louis, MO 63130, USA D. H. KAPLAN Department of Physics Southern Illinois University Edwardsville, IL 62026, USA Polystyrene-based scintillating optical fibers are manufactured at Washington University for use in astroparticle physics instruments. A hodoscope of 0.2-mm square-crosssection fibers has been operating on the Cosmic Ray Isotope Spectrometer on the ACE spacecraft at LI since 1997. A hodoscope of 1-mm fibers operated in a large-area cosmic-ray detector for a total of 50 days on stratospheric balloons over Antarctica. Ribbons of 1-mm fibers are part of the anti-coincidence shield on the Gamma-ray Large Area Space Telescope (GLAST). Laboratory tests of these fibers have demonstrated insensitivity to moderate radiation doses, well understood variation of response with temperature, and attenuation lengths in excess of a meter.

1. Production of Scintillating Fibers At Washington University we fabricate scintillating optical fibers of square cross-section for use in astroparticle physics instruments on stratospheric balloons and spacecraft [1]. The scintillating core is polystyrene (index of refraction, n, 1.62) with butyl-PBD primary dye and dimethyl-POPOP secondary dye, and the single cladding is acrylic (n=1.50). We have also fabricated double-clad fibers in which the second cladding is fluorinated acrylic (n=1.43). Various fiber dimensions have been used from 0.2 x 0.2 mm2 with 26-cm detector length to 1.0 x 1.0 mm2 with 1.2-m detector length. The dyes are dissolved in the styrene monomer and a boule is created with a 3.8 x 3.8 cm2 cross-section and 40 cm length. The cladding thickness ranges from 2% to 5% of the boule dimension, depending to which we are drawing the final fiber. With controlled heating a long thin fiber is pulled off the boule and

Present address: Los Alamos National Laboratory, Los Alamos, NM 87545, USA

239

240

wrapped on a large (1.8-m diameter) drum. The final dimensions of the fibers are controlled by controlling the speed at which the drum rotates and the speed at which the boule is lowered past the heating element. The cross-section of the fibers preserves the square shape of the original boule, and the relative dimensions of the core and clad are similarly preserved. 2. Fiber hodoscope for the CRIS instrument on the ACE spacecraft The Cosmic Ray Isotope Spectrometer (CRIS) [2] on the Advanced Composition Explorer (ACE) spacecraft measures the isotopic composition of the cosmic rays, using the Energy-loss/Total-energy technique with nuclei stopping in a stack of totally depleted lithium-drifted silicon detectors. Precise measurement of energy loss (dE/dx) requires knowledge of the angle the incident particle makes with the normal to silicon detector stack (the secO correction). Also, it requires knowledge of the location of the incident particle in the detector because the detector thicknesses are not perfectly uniform. The CRIS instrument is composed of four stacks of 10-cm-diameter Si(Li) detectors next to each other, with a 26 cm x 26 cm hodoscope above. The hodoscope is composed of three xy pairs of scintillating fibers. The fibers each have 200um square cross-section including lOum cladding all around. The cladding is coated with black ink to prevent optical coupling between fibers. The light output of the fibers is detected with an image-intensified CCD camera. CRIS achieved excellent mass resolution (a < 0.25 amu). Among the many astrophysical results that depend on the excellent resolution of CRIS are demonstration of the time delay between nucleosynthesis and acceleration of heavy cosmic rays [3] and the indication that the composition of the cosmic rays reflect acceleration in the superbubbles surrounding OB associations [4]. The ACE spacecraft was launched in August 1997 and has been in space near the LI Lagrange point on the Earth-Sun line about 1.5 x 106 km sunward of Earth for over eight years. During this time the entire CRIS instrument has been operating flawlessly. We estimate that the topmost layer of fibers has been exposed during these eight years to approximately 2 krad of total radiation dose. Figure I shows the variation with time of the mean integrated intensity of the spot of light in the topmost (XI) layer and the bottommost (Y3) layer from iron nuclei stopping in the instrument. We note approximately 9% degradation of the signal amplitude in XI over nearly eight years., and less in Y3. This degradation could be a combination of the effects of radiation damage on the scintillators themselves and degradation of the microchannel plate of the image intensifier, which has seen more total charge transported in the region imaging XI than in

241 Z= 2B

3 0 — ciELy a v g

intensity

30000

soooo

—

S8QOO 37000

seooo — ssooo 199B

SOOO

3002 ye&r-

2004

2006

Figure 1. Spot intensity (arbitrary scale) in uppermost (XI) and lowermost (Y3) layers vs time.

the region imaging Y3. However, the laboratory results described in section 5 below, suggest that this degradation is not due to radiation damage to the fibers. 3. Fiber hodoscope for the TIGER stratospheric balloon instrument The Washington University group has been building scintillating fiber detector systems for balloon-flight cosmic-ray instruments for many years. Early balloon flights with these fibers demonstrated their capability prior to their selection for the CRIS instrument. Our most recent balloon-borne instrument using scintillating fibers is the Trans-Iron Galactic Element Recorder (TIGER), an instrument with 1.2 m x 1.2 m area designed to measure the elemental composition of the rare cosmic rays with atomic number (Z) greater than 30. [5] The instrument measures Z for particles penetrating a stack of plastic scintillators and two Cherenkov detectors with index of refraction 1.5 and 1.04. The particle trajectories are determined with xy fiber layers above and below the detector stack. In this detector system the fibers have 1-mm square crosssection, including um cladding. The fibers are grouped in tabs of six or seven fibers and read out using 2.5cm photomultipliers. Each of the four hodoscope layers of 196 tabs is read out with just 28 photomultipliers, using a coding system. At one end (the course end) fourteen consecutive tabs are bundled and coupled to one photomultiplier. At the other end (the fine end) every fourteenth tab is bundled and coupled to one photomultiplier. The combination of signals in one course and one fine photomultiplier defines which of the 196 tabs was penetrated by the particle.

242

4. Fiber anticoincidence strips for the GLAST spacecraft The Large-Area Telescope of the Gamma-ray Large Area Space Telescope (GLAST) measures high-energy gamma rays pair-producing in a tower of silicon-strip detectors. A key element of this instrument is its anticoincidence detector (ACD), which is required to reject nearly all of the charged particles, which are the background for any gamma-ray experiment. The requirement for the ACD is that it must have 0.9997 efficiency and hermeticity for singly charged particles [6]. The ACD is composed of individual tiles of scintillators, but to close off the inevitable cracks between tiles, our Washington University group has manufactured strips of scintillating fibers that are laid over the cracks. Calculations by the GLAST scientists (Moiseev, private communication) show that with the fiber ribbons the required ACD efficiency is achieved, while without the ribbon an efficiency of only about 0.9965 is achieved. 5. Laboratory tests of the effects of radiation on the fibers To test the effect on our scintillating fibers of moderate doses of radiation, as might be experienced on a space-flight instrument, we exposed a set of our 0.75-mm square cross-section fibers to beams of 42 MeV protons at the Indiana University Cyclotron Facility, accumulating a dose of 4 krad. (Because radiation damage can repair in the presence of oxygen, and that oxygen would not be available in a space exposure, the fibers were kept in a pure nitrogen environment during and after the exposure.) Figure 2 shows that within the accuracy and repeatability of the test set-up (-+/- 5%) there was no difference

1200 $

;m »>

\ » ^-1 37 cm

1000

•P

800

,. .C D)

600

76.3 cm

\

TO

(!) -U 01

400

^L

~i

a.

*fc

15.3 cm

200

0

5

10

15

20

Light Output (Photoelectrons) Figure 2: Light output from electrons at three distances, before (solid line) and after (dashed line) irradiation.

243

between the performance of these fibers before and after the radiation. In each case, these fibers displayed a 1/e attenuation length of approximately 160 cm, and the pulse-height spectra of light detected at one end from electrons penetrating at various distances from that end after the irradiation was indistinguishable from that recorded prior to the irradiation. 6. Laboratory tests of thermal effects on fibers Kaplan et al [8] have measured the light output of a fiber as a function of temperature over the range -30°C to +50°C. (Figure 3) These data are consistent with a model of self-absorption by the secondary dye, which is described in reference [8].

Relative Intensity

Figure 3 Signal intensity vs wavelength at various temperatures.

Acknowledgments This work was supported in part by NASA grants NNG05WC04G and NAG512929. The authors acknowledge the important work of our CRIS collaborators at Caltech, JPL, and NASA/GSFC, of our collaborators at Caltech and NASA/GSFC on TIGER, and of Alexander Moseeiv our GLAST contact at NASA/GSFC.

244

References 1. A. J. Davis, et al., Nucl. Inst. & Meth. A276, 347 (1989). 2. E. C. Stone, et al., Space Sci. Rev. 86, 285 (1998). 3. M. E. Wiedenbeck, et al., Astrophys. J. 523, L61 (1999). 4. W. R. Binns, et al., Astrophys. J. in press (2005). 5. J. T. Link, et al., 28th Intl. Cosmic Ray Conf. (Tsukuba) 4, 1781 (2003). 6. A. Moiseev, et al., 26'h Intl. Cosmic Ray Conf. (Salt Lake) 5, 160 (1999). 7. K. Rielage, Ph.D. Thesis, Washington University (2002). 8. D. H. Kaplan, et al., Nucl. Inst. & Meth. in press (2005).

T H E T2K E X P E R I M E N T : STATUS A N D I N S T R U M E N T A T I O N OF T H E 280M N E A R D E T E C T O R

YU. G. KUDENKO* Institute for Nuclear Research of the Russian Academy of Sciences October Revolution Pr. 7A, 117312 Moscow, Russia E-mail: [email protected]

The Tokai-to-Kamioka (T2K) experiment is a second generation long baseline neutrino oscillation experiment which aims at a sensitive search for the ve appearance. The T2K near neutrino detectors at 280 m from the target will measure the neutrino beam properties prior to the oscillation and interaction cross sections. The main design features of these detectors and recent studies of new scintillator/WLS fiber detectors readout by multipixel APD's are presented.

1. Introduction One of the major achievements of the particle physics in the past decade is the discovery of the neutrino masses and mixing based on the experimental studies of the neutrino oscillation1. These results show the necessity for the extension of the Standard Model. Three generation neutrino oscillations depend on the three mixing parameters sin20i2,sin2f?23,sin2013 and the two mass-squared differences Am 2 ol = Am| x = m% — m\ and Am 2 t m = Arrijj = m\ — m\. Current oscillation parameters (3a interval) from the global three-neutrino analysis2 look as follows: A m ^ = (7.1 - 8.9) x 1 0 - 5 , Ara^, = (1.4 - 3.3) x 10" 3 , sin 2 0 12 = 0.23 - 0.37, sin2023 = 0.34 - 0.66, and sin2f?i3 < 0.051. The main road of neutrino physics will be the high precision measurements of these parameters, improve sensitivity and possibly measure #13 and CP-violation in lepton sector, distinguish between the normal and invert mass hierarchies, and to identify the neutrinos as being Dirac or Majorana. The next few years the new results are expecting from long baseline neutrino experiments at Fermilab MINOS 3 and OPERA 4 at CERN. These experiments will be able to test #13 with sensitivity a few 'Presenting the T2K Collaboration

245

246

times better than CHOOZ experiment, but at least the improvement by more than order of magnitude is needed. 2. The T2K long baseline experiment The T2K project 5 is a second generation long baseline neutrino oscillation experiment which will use a high intensity narrow band neutrino beam produced by JPARC 50 GeV (initially 40 GeV) proton beam. The first phase of the T2K experiment is aimed at two main goals: the sensitive measurement of #13 and the more accurate determination of the parameters sin2#23 and Am| 3 . The design intensity of the proton beam is 3.3 x 1014 protons/pulse at a repetition rate of about 0.3 Hz. The fast extracted beam width is 5.6 /zsec and there are 8 (or 15) bunches in a spill. The width of each bunch is 58 ns. Three electromagnetic horns will be used to focus charged pions generated in a graphite target installed inside the first horn. The focused pions decay in a 130 m decay tunnel which follows the horns. The off-axis beam configuration will allow to change the offaxis angle between 2° and 3° that provides the almost pure v^ beam with mean neutrino energies from 0.5 to 0.9 GeV, respectively, and a fraction of ve of about 0.5% at the peak energy. The sensitivity of better than (5(sin2013) = 0.01 at 90% c.l. will be achieved for 5 years of running. 3. The near neutrino detectors To achieve the physics goals, it is important to provide precise measurements of the neutrino beam properties near the target, neutrino flux, spectrum and interaction cross sections. For these purposes, the near detector complex (ND280) will be built at 280 m from the target along the line between the average pion decay point and the Super-Kamiokande. The physics requirements for ND280 can be briefly summarized as follows. The energy scale of the neutrino spectrum must be understood at 2% level, and the neutrino flux should be monitored with better than 5% accuracy. The momentum resolution of muons from the charged current quasi-elastic interactions should be less than 10%, and the threshold for the detection of the recoil protons is required to be about 200 MeV/c. The ve fraction is to be measured with an uncertainty of better than 10%. The measurement of the neutrino beam direction with precision much better than 1 mrad is expected to be provided by the on-axis detector (neutrino monitor). The off-axis ND280, shown in Fig. 1, consists of the UA1 magnet operated with magnetic field of 0.2 T, a Pi-Zero detector (POD), a tracking detector which

247

SMRD Figure 1.

Tp

C

The cutaway view of the T2K ND280 near detector.

has time projection chambers (TPC's) and fine grained scintillator detectors (FGD's), an electromagnetic calorimeter (Ecal), and a side muon range detector (SMRD). POD Detector. The POD has been designed to be similar to the K2K SciBar 6 and MINERi/A 7 detectors. It is installed in the upstream end of the magnet and optimized for measurement of the inclusive n° production by v^ on oxygen. In addition, the POD will measure the intensity and spectrum of the beam-associated electron neutrino flux. The two measurements will allow to control two major sources of backgrounds in the electron appearance search. The POD consists of 76 tracking planes composed of polystyrene triangular scintillating bars (about 2 x 104) alternating with thin (0.6 mm) lead foils. Each bar has a 3 cm base, 1.5 cm height and a central hole for a WLS fiber. The measurement of neutrino interaction with oxygen is achieved by interleaving water target planes between the thirty upstream scintillating planes. The water cells consist of semiflexible pillow bladders, nominal dimensions 3 cm x 1.8 m x 2.1 m, holding about 100 kg of water each. The schematic view of the POD target is shown in Fig. 2. The total mass of the POD will be about 12 tons with a fiducial mass of 1.7 tons of water, 3.6 tons of plastic scintillators, and 0.8 tons of lead. The expected energy resolution (without photostatistics) for events

248 sjMiw

30 mm

16 cm

Figure 2.

scintillator

30 mm

The schematic view of a POD layer.

fully contained in the active target is &E — 10% + 3.5%/\/E(GeV). The efficiency of the photon (ir°) detection and reconstruction will be about 80-90% (55%). It is expected about 1.7 x 104 neutral current single 7r° events in the water target for 10 21 protons on target (one year of running). Tracking detector. Three TPC's will measure the 3-momenta of muons produced by charged current interactions in the detector and will provide the most accurate measurements of the neutrino energy spectrum. In addition, dE/dx measurements will also determine the sign of charged particles and identify muons, pions, and electrons. The outer dimensions of the three TPC modules are assumed to be approximately 2.5 m x 2.5 m in the plane perpendicular neutrino beam direction, and about 0.9 m along the beam direction. A candidate for the gas mixture is Ar:C0 2 (90:10). At drift field of 200 V/cm, the transverse diffusion is expected to be 220 nm/y/cm and the drift velocity is 14 mm/fis. The proposed baseline design uses ether GEM's or Micromegas for the gas amplification. Fig. 3 shows the TPC module. The desired spatial resolution (300 ^m) determines a small size of about 8 mm x 8 mm for pads, leading to about 105 channels for 3 TPC modules. The FGD consisting of layers of segmented scintillator bars readout by WLS fibers allows tracking of all charged particles arising from neutrino interactions with the target nuclei. The ND280 will contain two FGD's, each

249

Figure 3.

The layout of a T P C module.

with dimensions 2 m x 2 m x 30 cm resulting in the total mass of 1.2 tons. The first FGD will be an active scintillator detector, one layer of which will consist of 200 scintillator bars, and thirty layers will be arranged in alternating vertical and horizontal layers perpendicular to the beam direction. The tracking threshold is expected to be about 4 cm, which corresponds to 350 MeV/c for a proton. The second FGD will consists of x — y layers scintillator bars alternating with 3 cm thick layers of passive water. The water layer will constructed from sheets of corrugated polypropylene, of dimension 1.0 m x 2.0 m and an outer thickness 1.0 cm. An active water scintillator FGD is considered as a detector upgrade. Ecal. The main purpose of the Ecal is to measure photon energy from the decays of 7r°'s produced by neutrino interactions in the POD, FGD's and TPC's. The Ecal surrounds the POD and tracking region and consists of 15 layers, each of which has one sheet of 3-mm Pb alloy and one layer of 1 cm thick x 5 cm wide plastic scintillator bars (with signal readout by WLS fibers). Interior this section is the preradiator section, where each of the 3 layers consists of a lead alloy sheet backed by 3 layers of scintillator bars. SMRD. Muons which escape at large angles with respect to the neutrino beam can not be measured by TPC's. However, they will intersect the iron yoke and therefore a muon's momentum can be obtained from its range by instrumenting the iron at various depths. About 40% of muons from

250

charged current quasi-elastic reactions and about 15% of muons from charge current non-quasi-elastic reactions are expected to intersect the SMRD. The UA1 iron yoke consists of 16 C-shaped elements made of sixteen 5 cm thick iron plates, with 1.7 cm air gap between the plates and is segmented in 12 azimuthal sections. The active component of the SMRD will consist of 1 cm thick scintillator slabs sandwiched between the iron plates of the magnet yokes. The readout will be provided by wavelength shifting fibers (WLS) to transport the light into photosensors. The scintillators can be coated by a chemical reflector to increase the light yield. The Monte Carlo study indicated that a muon energy resolution of less than 10% can be achieved. The SMRD is required to identify MIP's with very good efficiency. It is also desirable to have good inter-layer timing resolution to distinguish inwardfrom outward-going particles.

4. P h o t o s e n s o r s The Nd280 detector will widely use the WLS fiber readout with individual detection of the light from fibers by photosensors which have to operate in magnetic field and limited space inside the magnet. There are several candidates for photosensor that can satisfy these conditions: MicroChannel-Plate Multi-Anode PMT's, Avalanche Photodiodes, Fine-Mesh Multi-Anode PMT's and multipixel avalance photodiodes, which are considered as photosensors for FGD and SMRD and extensively studied for such an application. Detailed description of the multipixel avalanche photodiodes with metalresistor-semiconductor layer-like structure operating in the Geiger mode technique 8 can be found in 9 ' 1 0 . This detector (hereinafter SiPM) is a multi-pixel semiconductor photodiode which consists of pixels on a common substrate. Each pixel operates as an independent Geiger micro-counter with a gain of the same order as a vacuum photomultiplier. Geiger discharge is initiated by a photoelectron, thermally or by high electric field. A single pixel signal does not depend on a triggered number of carriers in a pixel. In such a way, the SiPM signal is a sum of fired pixels. Each pixel operates as a binary device, but SiPM on the whole is an analogue detector with the dynamic range limited by a finite number of pixels. The pixel size can be of 15 to 70 /im, and the total number of pixels is 100-4000 per mm 2 . Potential problem of such devices is a high dark rate. For example, at the bias voltage sufficient to provide the photon detection efficiency for green light of 10-12%, the dark rate is about 1 MHz at a threshold of 0.5

251

photoelectron (p.e.) SiPM's also have rather high temperature gradient of 3-4% per degree for the signal amplitude. SiPM's of the sensitive area diameter 1.1 mm and 556 pixels were produced by the CPTA Company (Moscow). The remarkable feature of these devices is a low crosstalk that provides fast reduction of the dark rate from about 1 MHz at the threshold of 0.5 p.e. to 50 (3-5) kHz for thresholds of 1.5 (2.5) p.e., respectively. The first tests show that such devices can be used in several ND280 systems. The result of the measurement with cosmic muons of the SMRD scintillator slab with embedded a S-shape double-clad Kuraray Y l l WLS fiber and both-end SiPM readout is shown in Fig. 4. A light yield of about 17 p.e. (sum of both ends) provides high efficiency

(b) (TD Cl-TDI 22)12 | 2.5 c 2

• • •

•

Groove depth: 3 mm

Figure 4. Measured light yield (a) and time resolution (b) of the SMRD scintillator detector (c) using cosmic muons.

(more than 99%) for a minimum ionizing particle (MIP) detection. Moreover, the good timing resolution ~ 1 ns allows suppressing the dark rate in such detector using coincidence between both SiPM's and results in a space resolution of 6-8 cm for a MIP along such scintillator slab.

252 5.

Conclusion

T h e main goal of the T2K experiment is a sensitive search for the v^ -> ue transition using off-axis muon neutrino beam. T h e energy spectrum, flavor content, and interaction rates of the neutrino b e a m will be measured by a set of detectors (TV° detector, tracker, electromagnetic calorimeter, side muon range detector) located at 280 m from the neutrino production target and contained inside the UA1 magnet. T h e current design and instrumentation of the near detectors seem to be adequate the physics goals. This work was supported in p a r t by the P r o g r a m of the Russian Academy of Sciences for Basic Research "Neutrino Physics". References 1. Y. Ashie et al., Super-Kamiokande Collaboration, Phys. Rev. D 7 1 , 112005 (2005) and references therein; E. Aliu et al., K2K Collaboration, Phys. Rev. Lett. 94, 081802 (2005); Q.R. Ahmad et al, SNO Collaboration, Phys. Rev. Lett. 89 011301 (2002); K. Eguchi et al., KamLAND Collaboration, Phys. Rev. Lett. 90, 021802 (2003); M. Appolinio et al., CHOOZ Collaboration, Phys. Lett. B466, 415 (1999). 2. M. Maltoni et al., New J. Phys. 6, 122 (2004) [hep-ph/0405172]. 3. htpp://www-numi.fnal.gov/ 4. The OPERA experiment, H. Pessard, hep-ex/0504033. 5. Y. Ytow et al.„ hep-ex/0106019. 6. K. Nitta et al., Nucl. Instr. Meth. A535, 147 (2004). 7. D. Drakoulakos et al., hep-ex/0405002. 8. A.G. Gasanov et al, Lett. J. Techn. Phys. 16, 14 (1990) (in Russian); G. Bondarenko, V. Golovin, M. Tarasov, Patent for invention in Russia No. 2142175, 1999. 9. A. Akindinov et al., Nucl. Instr. Meth. A494, 474 (2002). 10. G. Bondarenko et al, Nucl. Instr. Meth. A442, 187 (2000).

RESULTS AND STATUS OF THE PICASSO EXPERIMENT F. AUBIN, M. BARNABE-HEIDER, M. DI MARCO, P. DOANE, M.-H. GENEST, R. GORNEA, R. GUENETTE, C. LEROY*, L. LESSARD, J.- P. MARTIN, N. STARINSKY, U. WICHOSKI AND V. ZACEK Departement de physique, Universite de Montreal Montreal, Quebec, H3C 3J7, Canada E. BEHNKE, W. FEIGHERY§, I. LEVINE AND C. MUTHUSI§ Department of physics & astronomy, ^Department of Chemistry, Indiana University South Bend, South Bend, Indiana, 46634, USA K. CLARK, C.B. KRAUSS AND A.J. NOBLE Department of physics, Queens University Kingston, K7L 3NG, Canada S. KANAGALINGAM AND R. NOULTY Bubble Technology Industries Chalk River, Ontario, KOJ1J0, Canada S. POSPISIL, J. SODOMKA AND I. STEKL Institute of Experimental and Applied Physics ofCTU Prague, CZ-128 00 Prague2, Czech Republic The PICASSO (Project In Canada to Search for SuperSymetric Object) experiment is searching for cold dark matter (CDM) through the direct detection of weakly interacting massive particles (WIMPs), in particular neutralinos (x) via their spin-dependent interactions with nuclei. The experiment is taking data in the Sudbury Neutrino Observatory Laboratory at a depth of 2070 m (6000 mwe). PICASSO uses a liquid fluorocarbon, C4F10 as the active material and searches for WIMP interactions on 19F. Measured limits on the spin-dependent x-proton and x-neuh"011 CT0SS sections are presented and compared with results from other experiments. Excluded regions in the plane of the effective x-proton and x-neutron coupling strengths ap and a„ are also reported. The next phases of PICASSO are presented. They will involve detectors of larger volume and increased active mass with higher sensitivity in the spin-dependent sector.

Corresponding author: [email protected]

253

254

1. Introduction There is strong evidence that a large fraction of the matter in the Universe is non-luminous and non-baryonic. The most recent evidence is based on the measurement of the cosmic radiation background (CMB) by WMAP [1]. The amount of cold dark matter is predicted by WMAP as QCDM = 0.188-0.256. From the WMAP results one derive a cosmological matter density of £}m=0.27±0.04 that may be compared with the value of the baryon density of £}b=0.044±0.004 as indicated by nucleosynthesis and the height of the first acoustic peaks in the CMB data. The lightest neutralino, %, predicted in supersymmetric (SUSY) extensions of the Standard Model (SM) [2] emerges as the most promising candidate for dark matter. This stable spin Vi particle has no electrical charge and is a weakly interacting massive particle (WIMP) with a mass in the range from 10 GeVc"2 to 1 TeVc"2. These neutralinos are assumed to be concentrated in a halo around our galaxy [3]. In the Milky-Way galaxy, the velocity, v, of these particles is assumed to follow an isotropic Maxwellian distribution of the form /(v) = vV (v+V£)2//v ° 2 , where v0 = 230 kms"1 is the velocity dispersion of the dark matter halo and vE = 244 kms"1 is the velocity of the earth relative to the dark matter distribution (the effect of a ± 2 0 kms"1 annual variation is neglected here) [3]. The local neutralino mass density at the position of the solar system is assumed to be 0.3 GeVc' cm" and the velocity distribution is truncated at the escape velocity of matter in the Milky Way, vesc = 600 kms"1. In a direct dark matter search experiment, the detection of neutralinos is through the measurement of the energy deposited by the recoiling nucleus in the WIMP elastic scattering off nuclei in the detector. For all detector materials, this energy is expected to be less than 100 keV. The rate of interaction of neutralinos with ordinary matter is of the order of a fraction of an event per day per kilogram. Therefore, it is necessary to build detectors with large active mass and extremely low intrinsic background and, in practice, to operate the experiment in a deep underground site. 2. The experimental technique PICASSO detectors [4] consist of containers filled with droplets of diameters between 10 to 100 um of a superheated liquid at room temperature (presently a fluorinated halocarbon C4F10). These droplets are dispersed and trapped in a polymerized water based gel. Liquid-to-vapour phase transitions can be induced by nuclear recoils following the neutralino interaction with a nucleus (19F or 12C) in the droplet.

255

These detectors are threshold detectors, since an explosive droplet-tobubble transition is only triggered if the energy deposited through the ionization process in the liquid by the recoiling nucleus exceeds a temperature-dependent energy threshold and within a critical length. The resulting shock waves are detected by piezoelectric transducers mounted on the container's exterior wall. The measurements of the detector response to monoenergetic neutrons as a function of operating temperature allow the extraction of threshold temperatures, T,h, for a given neutron energy and in turn the determination of the recoil threshold energy, ERi>h,, as a function of temperature for various pressures of operation. For a practical range of temperatures of operation (20° to 47°C), ERth follows an exponential temperature dependence: ER*{T)

=EbeK(T-Tb>

(1)

where K is a constant to be determined experimentally and Et is the threshold energy at the boiling temperature Tb. From purely kinematical considerations, nuclear recoil thresholds in droplet detectors fall into the same operating temperature range for neutrons of low energy (from 10 keV up to a few MeV) and massive neutralinos (from 10 GeVc"2 up to 1 TeVc"2) at velocities which are typical for dark matter particles in our galactic halo. In this operating range the detectors are insensitive to x-rays, y-rays and irunimum ionizing particles. It is found that the efficiency to detect nuclear recoils at a given temperature is not a step function, but rises gradually from threshold to full efficiency. The probability, P(E'R, E'Ri,/,(T)), that a recoil nucleus i at an energy E'R near threshold EiRjh will generate an explosive droplet-to-bubble transition is zero for E'R < E'Rlh and will increase gradually up to 1 for ElR > E'Rth. This probability is expressed as [4]: P(El R,El

Kth(T))

= l-exp

-(l.0±0.l)(ElR-E'Rjh(T)) E'R,th(T)

(2)

Combining Eqs. (1) and (2) (in the case of 19F struck by neutrons) allows the determination of the detector efficiency as a function of the recoil energy of I9F at a given temperature. In the temperature range from 20°C to 47°C, in which the PICASSO detectors are operated, 19F recoils from 500 to 6 keV can be detected with an efficiency of 50%. With this efficiency, the neutralino induced recoils become detectable above 30°C, the expected recoil energies being smaller than 100 keV. Knowing the neutralino induced recoil energy spectrum [3] and the detector efficiency for a l9F recoil energy at a given operating temperature, the

256

observable neutralino count rate, R<,bSi as a function of temperature, neutralino mass and cross section can be found to be [4] Robs(M,,
403 ATMXJ

"SD

Pb

Px 03GeVc~zcm-3

<"*>

230kms

e{Mr,T). (3)

where AT is the atomic mass of the target atoms, px is the mass density of the neutralino of mass Mx, (?SD is the spin-dependent neutralino-nucleus cross section, is the relative average neutralino velocity and e is the neutralino detection efficiency. The neutralino interaction with nuclei of ordinary matter is of electro-weak strength and has the general form given by •4Gl

MXMA MX+MAJ

CA

(4)

where GF is the Fermi constant, Mx and MA are the mass of the neutralino and the detector nucleus, respectively [5,6]. CA is an enhancement factor, which depends on the type of the neutralino interaction. Spin independent (SI) or scalar interaction proceeds via Higgs, or sfermion exchange with CA given by

CsA'=-~[zfp+(A-Z)fn}

(5)

where fnp are the neutralino couplings to the nucleons. For equal couplings to neutrons and protons the cross section is proportional to A2 (coherent interaction). Due to their large 19F content (19F is a spin-l/2+ isotope), PICASSO detectors benefit from a very favorable spin-dependent cross section and are particularly suitable to search for spin-dependent neutralino interactions [5]. The general spin-dependent cross section for WIMP scattering on 19F has the form [5] ^SD

=4G2F/J2FCF,

(6)

where GF is the Fermi coupling constant and fiF is the neutralino-19F reduced mass. cF = ($lx)(*p(sp) + a„(s„)) (J+l)/J is a spin-dependent enhancement factor, where ap, a„ are the effective proton and neutron coupling strengths, J = 1/2 is the total nuclear spin, (sp) = 0.441 and (s„) = - 0.109 are the expectation values of the proton and neutron spins within ,9 F [7]. In order to compare our results with other experiments and predictions of supersymmetric models, the

257

experimental cross section limits on 19F, oxF, are converted into limits on the neutralino-proton (o^,) and neutralino-neutron (a^) cross sections. Following the method developed in [3], we assume that all events are either due to neutralinoproton or neutralino-neutron elastic scatterings in the nucleus, i.e. a„= 0 or ap= 0, respectively, and obtain: ( a

a

xi =

2 A

£*-.

Mi

#

(7)

.A) r

where i = proton or neutron, /J; is the x-nucleon reduced mass (the mass difference between neutron and proton is neglected) and Cp(F) and C„(F) are the proton and neutron contributions to the total enhancement factor of 19F. They are related to the ap and a„ couplings by C,(f) = (8/w)af (S1,)2 (J+l)/J . cp and C„ are the enhancement factors for scattering on individual protons and neutrons. The values for the ratios Cp(F/ Cp = 0.778 and Cn(F/ C„ = 0.0475 are taken from [5,7]. As shown in [5], the allowed values of ap and a„ for a given neutralino mass are constrained by the inequality: ( a„

24G>,

vV°V

(8)

The sign in the parentheses is that of (S p) / (S n) . The allowed values of ap and a„ fall within a band defined in our case by two straight lines with a slope given by -<Sn /Sp> = 0.247:

<-^±a

a

'

(Sp)

± "

K

-

-\|24G2^%

a-

(9)

^

For experiments with two active nuclei with different < S p) I (S „ > ratios, the combination of the two bands lead to a closed elliptical contour. 3. The search for dark matter The PICASSO experiment has taken data with three detectors of 1 litre volume representing a total active mass of 19.4 ± 1.0 g of 19F [8]. The detectors were all calibrated in the same known neutron field of an Ac(Be) source before being installed in the water purification gallery of the Sudbury Neutrino Observatory Laboratory at a depth of 2070 m (6000 mwe). The detectors were placed in an acoustically and thermally insulated container in which the temperature was regulated and controlled with a precision of 0.1 °C. The data taking proceeded

258

continuously between April 2004 and October 2004 under remote control via Internet from Montreal, the detectors being operated at an ambient pressure of 1.23 bar and at temperatures varied in a random sequence from 20° to 47°C. Each detector was treated independently by the data analysis. All the events were first filtered in order to reject non-droplet like background events, such as acoustic or electromagnetic noises, for which the signal frequency, obtained by a fast Fourier transform, is usually below the one obtained for particle induced events. The acceptance of this filter was carefully studied [4] and found to be 6(=0.85±0.05 over the temperature range from 25 to 45°C, decreasing slightly for lower temperatures. In order to include only data taken when the detectors were in a stable operating configuration, time cuts were applied, described in detail in [8]. After time cuts, the effective exposure for the ensemble of three detectors was 1.98 ± 0.19 kg day of 19F. The error on the exposure comes from the uncertainty on the determination of the active mass (8%) and the statistical error on the determination of the filter efficiency (5%). There are several sources of external and internal background. The external background consists of neutrons, the detectors being insensitive to y and P radiation in the temperature operating range. Neutrons at a deep underground laboratory, like SNO, are mainly produced by spontaneous fission of 238U and following (a,n) reactions and more rarely by cosmic-ray muon interactions (photonuclear interactions of muons, muon capture). Water cubes surrounding the acoustically and thermally insulated box containing the detectors slow down these external neutrons and bring them below detection threshold. The main background is the internal background due to the alpha particles [4] produced in the decay chains of uranium and thorium nuclei which are present as contaminants in the constituents of the detectors. Purification of the detector ingredients was done by chemical and mechanical means and is described in details in [4,9]. The purification achieved so far was assayed at the level of 10"10 gU/g. The entire detector setup is designed to be radon tight. The 222Rn emanation rate for PICASSO containers was measured to be 15 ± 9 atoms/day [4]. The fit of the data to the alpha background response can be seen in the left part of Figure 1. 4. Current limits and future developments A combined fit of the alpha background and the neutralino response to the data allowed us to rule out any positive evidence for neutralino induced nuclear recoils, translating into a 90% C.L. upper limit of 1.31 pb on cr^, and 21.5 pb on o™ for a neutralino mass of Mx = 29 GeVc"2 [8]. The left part of Fi gure 1 shows

259 how the data points compare with the combined alpha background and expected response for a neutralino of mass 50 GeVc"2 with different cross section values (2, 5 and 15 pb). One can see from this figure that the data points are compatible with a response induced by alpha background only. The right part of Figure 1 and the left part of Figure 2 show our limits, compared with the limits obtained by other experiments, on the spin-dependent neutralino-proton cross section (o^) and neutralino-neutron cross section (o^) at the 90% C.L. as a function of the neutralino mass. Our comparison includes neutralino-proton and neutralinoneutron results obtained from the application of the method developed in [5] to the neutralino-nucleus cross section obtained in spin-independent experiments [10], [11]. The right part of Figure 2 shows the excluded regions in the plane of the effective coupling strengths ap , a„ for protons and neutrons for a neutralino mass of 50 GeVc"2 as obtained from Eq. (9) as well as the excluded regions obtained by other experiments. The sensitivity of PICASSO is presently limited by alpha emitting contaminants and its small active mass. The next step of the experiment (Phase lb, right part of Figure 1) is in preparation, with 2 kg of active mass, detector modules of 4.5 litre volume (Figure 3) with an active mass of 60 g, an order of magnitude higher than the loading achieved for the 1 litre volume detectors used up-to-now. Each module is equipped with 9 piezo-electric sensors mounted on the exterior wall for event localization. These detector modules will benefit from hydraulic recompression, improved purification and fabrication techniques. Phase lb will be followed by phases II and III (right part of Figure 1) with 25 kg and 100 kg of active mass, respectively. With Phase II one expects to achieve a sensitivity of lxlO"3 pb or better (reach tip of MSSM predictions) in the spin-dependent sector. With Phase III the expected sensitivity is between lxlO"4 pb and lxlO"5 pb (reach the core of the MSSM predictions). These two phases will utilize larger 30 liter modules.

260

/ /

SOS

/

f

w"

1 oo«

.

/

I |

•&

/

A 0 03

0 02

T I

LJ—4—

rU^-fr i

001

-

7

30

:5

30 35 Tempetatuie r*C(

«

4S

50

M,(GcV/c')

Figure 1. Left: Measured count rate for one of the three detectors compared to the count rate expected from an alpha background only (red curve) and with an alpha background plus a signal due to a neutralino of 50 GeVc'2 mass having a cross section of 2 pb (green dashed line), 5 pb (blue dotted line) and 15 pb (dashed black line). Right: PICASSO limit on the spin-dependent neutralinoproton cross section (o-jp) at 90% C.L. as a function of the neutralino mass (blue squares) compared with other experiments: NAIAD (green circles), DAMA/Nal 3a-allowed modulation signal (pink shaded region), EDELWEISS (red dot line), CRESST I (grey 'x') and CDMS (black solid line for Ge and black dash-dot line for Si), see [10],[11] and references therein. The SIMPLE experiment [12] reports results similar to PICASSO.

1 10-

10-

.

H

^ ^ ^

\

s ^ -

__CRESST _

n

—"~>^C

o ° ° '

rTC

'^•"""firASso

^_x>< -

—.

NAIADN

PICASSO^1

10»

\

^

^ 1

i n '

11) IOM.KicVVl

III

-20

-10

0

K

Figure 2. Left: PICASSO limit on the spin-dependent neutralino-neutron cross section (c^) at 90% C.L. as a function of the neutralino mass (blue squares) compared with other experiments: DAMA/Nal 3o-allowed modulation signal (pink shaded region), EDELWEISS (red dot line), CRESST I (grey V ) , ZEPLIN I (red triangles) and CDMS (black solid line for Ge and black dashdot line for Si), see [10],[l l] and references therein. Right: Excluded regions (outside the bands) in the plane of the effective coupling strengths ap, a„ for protons and neutrons for a neutralino mass of 50 GeVc"2 as obtained from Eq. (9) for PICASSO, compared with some other experiments, see for example [10],[11] and references therein.

261

Figure 3: Left: a module of the new 4.5 litre detectors with an active mass of 60g. The container is entirely fabricated from acrylic. The stainless steel cage serves as reinforcement during recompression. Each module is equipped with 9 piezo-electric sensors for event localization. Right: "artist's view" of the arrangement of 32 modules in two layers for the 2 kg detector ensemble of Phase lb. In reality 4 detectors each are installed in a temperature and pressure control unit.

References 1. C.L. Bennett et al., Astrophys. J. Suppl. 148, 1-27 (2003); D.N. Spergel et al., Astrophys. J. Suppl. 148, 175-194 (2003). 2. G. Jungman, M. Kamionkowski and K. Griest, Phys. Rept. 267, 195-373 (1996). 3. J.D. Lewin and P.F. Smith, Astro. Phys. 6, 87-112 (1996). 4. M. Barnabe-Heider et al., PICASSO Collaboration, physics/0508098, to appear in NIM A. 5. D. R. Tovey et al, Phys. Lett. B 488, 17-26 (2000). 6. VA. Bednyakov, hep-ph/0310041, October 3, 2003. 7. A.F. Pacheco and D.D. Strottman, Phys. Rev. D 40, 2131 (1989). 8. M. Barnabe-Heider et al., PICASSO Collaboration, Phys. Lett. B 624, 186194(2005). 9. M. Di Marco, "Reduction du bruit de fond en vue de la detection de la matiere sombre avec le projet PICASSO", Ph.D. Thesis, Universite de Montreal, November 2004. 10. D.S. Akerib et al., astro-ph/0509269vl, September 10, 2005. 11. A. Benoit et al., Phys. Lett. B 616, 25-30 (2005). 12. T.A. Girard et al, Phys. Lett. B 621, 233-238 (2005).

DATA ACQUISITION SYSTEM AND TRIGGER ELECTRONICS FOR CACTUS J. LIZARAZO, P. AFONSO, M. CHERTOK, P. MARLEAU, S. MARUYAMA, J. STILLEY and S. M. TRIPATHI Physics Department, University of California One Shields Ave., Davis, CA 95616, USA CACTUS is a ground-based Air Cherenkov Telescope located at the Solar 2 facility in California, and operated by UC, Davis. It uses an array of 168 heliostats and a camera with 80 photomultiplier tubes to detect Cherenkov radiation produced by air showers. CACTUS incorporates novel techniques of time projection imaging and pattern triggering, implemented in FPGAs, thus improving upon the first generation sampling ACTs. Here we describe the telescope and its readout and triggering electronics.

1. Solar Two Field and CACTUS Secondary Optics CACTUS (Converted Atmospheric Cherenkov Telescope Using Solar2) is a ground-based air-shower detector located near Barstow, California [1,2]. The Solar Two site consists of about 1,800 heliostats (42m2 mirrors) that can be individually steered in elevation and azimuth. Motion controls are performed by remote-controlled driver electronics located at the base of each heliostat. Communications with a central computer are performed over a doublyredundant system using RS-232 lines. The cumulative tracking and pointing error for each individual heliostat is about 0.1 arc-degree. CACTUS utilizes 168 of these heliostats that fall within the field-of-view of a secondary mirror mounted 60 m above ground level on a tower. These 168 are spread over -20,000 m2, thus providing a 35% coverage with their mirror area. The spherical secondary mirror is a composite mirror made out of 13 hexagonal facets, each lm across. With an aperture of 4.4mx2.6m and a radius of curvature of 6m, it further focuses the 3m spot-size from the heliostats down to entrance apertures (~10 cm) of light-collecting parabolic Winston cones located at the focal plane of the camera. The camera consists of 80 photomultiplier tubes (PMTs) positioned according to the spot sizes of heliostats on the focal plane. The upper 35 PMTs are positioned to align with individual heliostats, while the lower 45 PMTs form a close-packed matrix which allows images from multiple heliostats to be captured in a single channel. We have exploited this feature by implementing a scheme in which we later (offline) sort out the signals from heliostats by measuring the relative delays in the times of arrival. In order to accomplish this, we have built high bandwidth electronics that can provide twopulse separation at the level of ~5ns. This is described below. 2. Readout and Trigger Electronics There are four main components in the CACTUS data acquisition system shown in Figure 1: a) front-end amplifiers and discriminators, b) trigger system, c) time to digital converters (TDCs) and readout system and d) anode current recording.

262

263 control bus data bus

TDCs

4-

ECLRVDS

translator^" PC

80 discriminator channels 80 channels

ECULVDS

Readout FPGA

translator

trigger out

Trigger FPGA

2^

': 80

PMT

amplifier t ^ parallel port interface

--| parallel port interface

USB

w:

'80 i current j monitor

Figure 1: A box-diagram of the CACTUS amplification and readout system. 2.1 Analog Front-end The front-end consists of inline (BNC-BNC) preamplifiers, mounted on the PMT bases, which drive 50Q co-axial cables to post-amplifiers located in a counting house. Both the preamps and the post-amps use monolithic broadband circuits (Mini-circuits ERA-3) [3] providing an overall voltage gain of ~30x and a bandwidth of ~1.6 GHz. The preamps are powered by a dc current carried on the same cable that is used for amplified pulses. A separate circuit senses this current and provides a voltage output that can be related to the anode current of the PMTs. The amplified signals are fed into CAMAC discriminators (LRS 4416) that provide dual outputs on ECL busses. The minimum pulse-width at the discriminator output is ~3.5 ns. Time-over-threshold measurements made above this minimum are directly proportional to pulse-heights. One set of ECL buses is connected to TDCs and the second one drives a trigger system. 2.2 Trigger System The task of the trigger system, based around a Xilinx XCV1600 FPGA, is to determine the presence of an air-shower in a given 10 ns window. The decision time must be less than 1 ixs because the storage depth in the TDCs is 2 lis. The ECL outputs from the discriminators are translated into LVDS levels before being fed into the FPGA. Figure 2 shows a schematic of the firmware loaded into the FPGA. It is divided into three main sections: a) delay lines, b) trigger logic and c) communications. ™\Majority The pipelines are used ^vLogic dynamically in real time to delay the Threshold signals from each of the 80 input Adjustable channels to compensate for flight Delay Line path differences among photons arriving from different parts of the Parallel Port field. Figure 3 shows a schematic of ' Communication a single channel pipeline. The building blocks consist of a flipflop, a multiplexer and a decoder. Figure 2: Schematic of the trigger system. These blocks are connected in a pipelined fashion such that an input signal gets introduced into the chain at a

264 point specified by the delay setting, and thence proceeds from one flip-flop to Delay DQ D9 Input Pulse

Line

4-

Delay Code •—

4 r l _ y Rip-Fiop r .

m.

-4— Multiplexer

J]

4—

4—

m.

[£]

—•—

Output

N-1

Figure 3: A schematic of the delay pipeline used in the trigger system. the next with one step advanced every rising edge of the 100 MHz global clock. Once signals arrive at the end of the pipeline, each delayed by the specified amount, an air-shower is represented by pulses synchronized in time, within ~5 ns for gamma-ray showers and -5-20 ns for cosmic showers. Light from secondary heliostats clusters in distinct patterns in time and can be included in more complex triggering schemes to further improve efficiency. The trigger logic is designed to detect timing patterns among the 80 channels. Various versions of the firmware can be downloaded into the FPGA. The simplest logic demands occupancy higher than some minimum number within a 10 ns window. Another version includes pulses from secondary heliostats and reconstructs the expected timing pattern for each channel. The "Majority Logic" module, which counts the number of channels with a hit in a 10 ns window, is pipelined in such a way that it uses only one logic cell per summing stage and is therefore capable of working at the required 100MHz clock. The summed value is truncated at six bits, thereby saving FPGA resources and making the electronics faster and more compact. This restricts the maximum allowed threshold to be 63. However, typical values used at CACTUS are in the 7-15 range. The "Threshold" module is a digital comparator which issues the final trigger when the sum value exceeds a user-specified threshold. The third part of the design is responsible for communicating with the host computer via its parallel port. A software package running on the host computer provides updates to path delays in real time, based on the tracked source's position in the sky. Delay data, in 8-bit format, are written to the data register of the parallel port and the FPGA loads them into an internal shift register that controls the delay pipelines. The coincidence level is also downloaded via the parallel port using its address register field. 2.3. Readout System Data acquisition in CACTUS consists of recording time of arrival (TOA) and time over threshold (TOT) of all pulses that arrive up to 2 us prior to the trigger time. This is accomplished by recording the rising and falling edges of discriminated signals in two different TDC channels, thus avoiding the minimum pulse-width constraint (8 ns) of the LRS 337 TDCs used in CACTUS. The TDCs register each edge independently with a 0.5 ns resolution. The TDCs are read out using a Xilinx XCV1000E FPGA as shown in Figure 4. The firmware used to program the FPGA is divided into three sections: a) TDC interface, b) 4-event buffer memory and c) data transfer, as described below.

265

ECL/LVDS

translator Jf8Q channels from discriminators (ECL)

FPGA

DPFIAM_4K

XCV1000E

PC

r

DPRAM_4K

DPRAM 4K HOST_XFER

parallel

I " J_-=±]b0rt

^~

DPRAM 4K

Figure 4: A schematic of the FPGA used in the TDC readout system. The TDC_CTL module in figure 4 is in an interface to the bank of TDCs via their front panel control bus. It implements a handshake protocol using 6 control lines: COM, WAK, REN are outputs and WST, BSY, PASS are inputs [4]. The TDCs are programmed in the "Common Stop" mode whereby they continuously acquire and store data in a 2 us buffer until a trigger is issued on the COM line. At this time the TDC stop recording and enter a readout mode. A handshake per word transfer (10 MHz) takes place between the TDCCTL module and the TDCs. The BSY line stays high during valid data transfer and the PASS line is asserted at the end. The PASS line is daisy chained to the REN line of the next TDC, thus allowing for a sequential readout of several TDCs. The readout time for typical events at CACTUS is about 2-4 ms. In order to minimize the system's dead time, the data from the TDCs are stored in buffer memory modules, DPRAM_4K, which are dual port 4K RAMs. The RAMSTORE module controls buffer sequencing and data are stored in a cyclical fashion. The buffers themselves keep track of data record length and of their own status. In case an event is longer than 4K, it utilizes contiguous buffers until the entire event is stored or all the buffers are full. The read-back of the buffers when transferring data to the host computer is also done cyclically, such that the oldest data are read out first. The RAM_SLECT module gathers the MEMORYSTATE outputs from each of the four buffers and instructs the RAMREAD module on read out sequencing. In this scheme there is no deadtime due to transfer to host computer until the trigger rate gets high enough such that the average time between triggers is equal to the average data transfer time. CACTUS operates at trigger rates of ~20 Hz with less than 10% deadtime. The HOST_XFER module implements an extended parallel port (EPP) protocol such that data can be transferred to the host computer's parallel port, regardless of the state of the TDC readout module. There are six control lines involved in an EPP transfer: nDATASTB, nADDRSTB and nWRITE are generated by the host computer and nWAIT is asserted by the FPGA, while nRESET and nlNTR are not used in this implementation. In addition, there is a bidirectional 8-bit bus that is used to transfer data and addresses. A C++

266

program, running on the host computer, continuously reads the address port for the presence of HxOC which indicates that an event is ready to be read out from the FPGA buffer memory. The host reads a few header words followed by the event record which is transferred at the maximum allowed EPP rate of ~1 MHz. 2.4 Anode Current Readout A voltage proportional to the PMT anode current is provided by the preamps and recorded using a commercial USB pod from National Instruments. The sampling rate can be varied in the 1-20 KHz range. The readout of these digitized voltages is asynchronous from the TDC readout and the data are correlated using time-stamps. The primary purpose of this data stream is for photometry which monitors the night sky background. 3. Results and Summary — All Heliostats

Figure 5 shows a distribution of Central ' Heliostats TOA from a sample of about 1,000 showers recorded by CACTUS. The lower curve is obtained by using only the primary heliostats (one per channel) while the upper curve includes pulses from secondary heliostats. The TOAs are with - 2 0 2 4 Rise Times (ns) respect to a trigger time for that Figure 5: A histogram of recorded TOA. event and display a leading edge coherence of about 1 ns. The trigger system provides a low threshold with an effective area of~10 4 m 2 at70 GeV rising to ~6xl0 4 m2 at 500 GeV. In summary, CACTUS is operational and the electronics systems have met specifications. The telescope has measured the differential flux of the Crab Nebula [5] that is in agreement with the world dataset above 300 GeV and provides new data in the regime below 100 GeV. Acknowledgements We are grateful to the University of California, especially Vice Chancellor for Research, Barry Klein, Dean Winston Ko, Senate Committee on Research, Senior Electronics Engineer, Britt Holbrook and the Office of the President. Prof Allen Zych, UCR, provided us with 50 PMTs. Mr. Judd Kilminik of Southern California Edison continues to support our activities at the Solar Two site. References [1] Tripathi, M., et al, "The Keck Solar Two Gamma-Ray Observatory", AIP Conference Proceedings, Vol. 587., 2001., p.93. [2] The CACTUS website http://ucdcms.ucdavis.edu/solar2/ [3] For a datasheet, see http://www.minicircuits.com/dg03-168.pdf [4] For a manual, see http://www.lecroy.com/lrs/dsheets/3377.htm [5] Marleau, P., et al, "Intermediate Energy Observations with CACTUS", Proceedings of CHERENKOV05, Paris, April 2005, in press.

T H E C E N T R A L PIXEL OF T H E M A G I C TELESCOPE FOR OPTICAL OBSERVATIONS *

F. LUCARELLI Dip. di Fisica, Universita degli Studi di Roma "La Sapienza" Pie. Aldo Moro 2, 00185 Rome, Italy E-mail: [email protected]

P. ANTORANZ, M. ASENSIO*, J.A. BARRIO, M. CAMARA J.L. CONTRERAS, R. DE LOS REYES, M.V. FONSECA M. LOPEZ, J.M. MIRANDA, I. OYA Dpto. Fisica Atomica, Facultad de Ciencias Fisicas, Dept. Infra., I. Sistemas Aeroespaciales y Aerop., 28040 Madrid, Spain

Universidad Universidad

Complutense Politecnica,

The MAGIC telescope has been designed for observation of the Cerenkov light generated in Extensive Air Showers. However, its 17 m. diameter and optical design makes it suitable for optical observations of varying astronomical objects. We report here on the installation of a dedicated photo-multiplier (PMT) for optical observations placed at the center of the MAGIC camera (the so-called central pixel). An electro-optical system has been developed in order to transmit through optical fiber the output signal to the counting room, where it is digitalized and stored for off-line analysis. The system was tested by observing the optical pulsation from the Crab pulsar. The results of the tests and the possibilities of the system will be presented here.

1. Introduction The MAGIC telescope l (Fig. 1) is an innovative detector aimed to detect very high-energy 7-rays from astrophysical sources using the Imaging Atmospheric Cerenkov technique. The telescope collects the very short flashes of atmospheric Cerenkov radiation (5-20 nsec in duration) emitted during the development of the Extended Atmospheric Showers (EAS) produced in the interaction of the cosmic 7-rays with the atmospheric nuclei. "This work was supported by the Spanish CICYT (project FPA2003-9543-c02-01)

267

268

Figure 1. The MAGIC telescope.

The main characteristics of the telescope are its 17m. tessellated mirror, a system of analog signal transmission based on optical fiber and signal digital sampling with 300 MHz Flash ADCs. The main detector, or camera, located at the focus of the telescope, is composed by a matrix of 576 fastresponse PMTs. Besides the main 7-ray observations, the large collection area of the MAGIC telescope can also be used to perform optical observations of slow varying astronomical objects. That can be done by measuring the variations in the DC-current output of a PMT installed at the telescope focus. This technique was already applied by other Cerenkov telescopes (HEGRA CT1 2 , Whipple 3 , Celeste 4 , HESS 5 ) which detected the optical pulsed emission from the Crab pulsar and estimated the photon content of the nebula surrounding the pulsar itself 2 . At commissioning time, the center of the MAGIC camera was deliberately left empty in order to host a dedicated PMT for optical observations. Such modified PMT, the so-called central pixel, was installed at the end of March '05 and tested successfully with the detection of the optical pulsed emission from the Crab pulsar. 2. The central pixel The PMT installed at the center of the MAGIC camera is a standard 1" MAGIC PMT, ET9116 6 , especially designed for fast pulsed-light detection. The DC branch of the pre-amplifier placed at the PMT base has been modified in order to have an integration constant r = 0.5msec. The overall tension was set to 1.08kV, corresponding to a gain of around 20k and an average current of 0.1-0.5^A under normal Night Sky Background (NSB) illumination. The transmission of the DC output signal from the PMT base to the

269

counting room for its digitalization and storing is made through optical fiber. The electro-optical transceiver (transmitter+receiver) was designed in order to transmit Crab-like signals, that is, analog pulsed signals with 10-100 Hz periods and widths of the order of milliseconds 8 . For the optical transmitter, installed inside the camera, a wide dynamic range LED was used (Honeywell HFE 4050-014 9 ) to perform the electro-optical conversion while at the receiver (placed after 170m of optical fiber, inside the counting room), a pin diode (Honeywell HFD3038-002 10 ) with an analog pre-amplifier implements the re-conversion. The LED was set to an operation point of 40mA, thus providing linear operation over a wide dynamic range (up to 200 mV). The bandwidth of the transceiver goes from 1Hz to 4kHz, thus allowing transmission of Crab-like signals and rejecting high and low frequency noise (mainly due to the night sky background). The transceiver allows to measure DC variations at the level of 0.2%. Both the central PMT and the optical transmitter are powered through independent and isolated power supplies installed on the side of the MAGIC camera. The ON-OFF switching of the central pixel is software-controlled through a Programmable Logic Controller (PLC).

3. Crab observations The whole system has been installed at the end of March '05 and tested by observing the Crab pulsar. The DC current from the central pixel, converted to voltage, was digitized at a sampling frequency rate of 2 kHz by means of a 16-bits National Instruments PCI-6034E ADC card u and stored for off-line analysis. The ADC card was set to a resolution of 0.16/xvolts. As the Crab pulses are completely embedded in noise (generated by the night sky background and the electronics), in order to observe its optical pulsation at the expected frequency, a time-correlated analisys of the data has to performed. A time stamp reflecting the sampling frequency was associated at each event and transformed to an inertial reference frame, which is assumed as the solar barycentric system. For this transformation, the TEMPO software 13 has been used. The corrected times then are folded with the radio ephemerides corresponding to the epoch of observation provided by the Jodrell Bank observatory 14 . A phaseogram was produced around the expected Crab rotational frequency for each independent test frequency defined by the Independent Fourier Spacing IFS=1/T 0 (, S . The folded intensities were then tested

270 I v-a9.7B37ialHllll

%

i S.13

rf

5.12!5

5.11

Wfii ?

^^^^0m^.

SI

=;

;

|

j

'

1

!

!

li =]

I I

i !.

i i

* U t

1

].

29.7S

31781

M.7M

*

!

.i

U.TS3

Figure 2. Upper plot: Optical light-curve of the Crab pulsar obtained with the central pixel of the MAGIC telescope. Lower plot: Zj^ statistical test. A clear peak appeared at the frequency expected for the observation epoch.

against a uniform distribution by performing a Z^ statistical test 15 . Figure 2 (lower plot) shows the Z\Q test scanning different frequencies around the expected Crab frequency. A clear peak appeared at the frequency expected for the observation epoch. Figure 2 (upper plot) shows the wellknown double-peaked lightcurve calculated at the maximum Z\0 value and after 20 min. of observation. 4. Conclusions In this work, we have reported about the installation in the center of the MAGIC camera of a photo-multiplier, the central pixel, dedicated to optical observations. The central pixel has been tested successfully by the detection of the Crab optical pulsations. The minimum time requested for a 5cr detection was lower than 1 minute. The expected detection time was 30 sec 2 . However, due to extreme weather conditions preceding the weeks immediately before the observations, the optical conditions of the telescope were not the optimal ones. The pointing error of the telescope was estimated offline in about 0.1°, while the Point Spread Function (PSF) was

271 around 0.1° ( F W H M ) . This limited the amount of collected light over t h e Central Pixel t o about 7-9% of the t o t a l light emitted by the pulsar. Recent alignments of the mirrors have reduced the P S F to less t h a n 0.05°. T h u s , we expect t o improve the detection time by at least a factor 2. T h e applications of the central pixel will be mainly focused in t h e simultaneous observations of the C r a b pulsar b o t h in the optical and in the 7 regimes, in order t o have real-time ephemerides for periodicity search in 7-rays. Before t h a t , the central pixel will be also used to test the whole timing system of the M A G I C telescope 1 2 . Optical observations of quasiperiodic flaring AGNs (Mrk 421, 501, ..) and X-ray binary systems (AE Aquarii) are also contemplated.

Acknowledgements T h e authors wish t o t h a n k the financial support given by the C I C Y T (project FPA2003-9543-C02-01) to make this work and the IAC for providing excellent working conditions in La Palma. T h e M A G I C telescope is mainly supported by B M B F (Germany), C I C Y T (Spain), I N F N and M U R S T (Italy).

References 1. Fonseca, M.V., Acta Physica Polonica B, vol. 30, 2331 (1999). 2. Ona-Wilhelmi, E., et a l , Astropart. Phys. 22, 95 (2004). 3. Srinivasan, R., et al., Proc. Towards a Major Atmospheric Cerenkov Detector V, Durban (1997) p.51. 4. De Naurois, M., et al., ApJ 566, 343 (2002). 5. Franzen, A., et al., Proc. 28th ICRC, Tsukuba (2003). 6. Ostankov, A., et al., NIM A, vol. 442, Issue 1-3, 117 (2000). 7. http://www.magic.iac.es/subsystems/centralpixel/ 8. Asensio, M., Lucarelli, F., Antoranz, P., et al., "Design, modelling and testing of the electro-optical transmitters for the central pixel of the MAGIC telescope", Proc. SPIE Vol. 5840, p.627-637 (2005) 9. content.honeywell.com/sensing/prodinfo/infrared/catalog/0331eng.pdf 10. content.honeywell.com/sensing/prodinfo/fiberoptic/application/on7eng.pdf 11. http://sine.ni.eom/nips/cds/view/p/lang/en/nid/11916 12. Lucarelli, F., et al., "The timing system of the MAGIC telescope", Proc. of the 19th ECRS, Florence (Italy), 2004 (to be appeared on IW) http://www.magic.iac. es/subsystems/timing, html. 13. Manchester, R.N., et a l , MNRAS 328, 17 (2001). 14. http://www.jb.man.ac.uk/ pulsar/crab 15. De Jager, O.C., ApJ, 436, 239 (1994).

STATUS R E P O R T OF A R D M P R O J E C T : A N E W D I R E C T D E T E C T I O N E X P E R I M E N T , B A S E D ON LIQUID A R G O N , FOR T H E SEARCH OF D A R K MATTER*

M. M E S S I N A A N D A. R U B B I A ETH Zurich, CH-8093, SWITZERLAND E-mail: marcello. messina@cern. ch

The goal of the ArDM project is to develop and operate a detector to search for direct evidence of Weakly Interacting Massive Particle (WIMP) as Dark Matter in the Universe. The experimental approach aims at detecting recoils of Argon nuclei induced by the collisions of WIMPs. Our immediate plan is to fully design and build a 1 ton prototype. This will involve a HV system, charge amplification + readout and light readout, as described in this paper. Our first milestone is a proof of principle on gamma-rays and beta electron vs nuclear recoils discrimination.

1. Introduction The understanding of the nature of Dark Matter is one of the most exciting problems of the particle physics. A reasonable hypothesis is that Dark Matter is composed of a new kind of non-baryonic, massive particle which could possess weak interactions with ordinary matter (WIMP). The detection of these WIMPs is based on the capability of measuring the recoils of target nuclei with kinetic energy in the range of 10 —100 keV. The signal is therefore quite elusive and it is a rare event given the weak coupling. Furthermore the rate is not easily predicted, since it depends on many poorly denned variables, even in the context of well defined extensions of the SM like e.g. SUSY. The ArDM project 1 aims running a high mass detector, almost 1 ton of effective mass to potentially reach a WIMP-nucleon cross-section sensitivity of a « 1 0 - 6 pb. The project is a natural evolution of the liquid Argon technology developed within the ICARUS 2 ' 3 programme, applied to the search of DM. The novelty of our design is to detect ionization and * On behalf of ArDM collaboration

272

273

scintillation signals independently such as to fully retain the features of the LAr TPC but with a threshold capable of detecting WIMP recoils.

2. Principle of detection and detector layout When a charged particle travels in the medium, scintillation light and ionization charge are produced. The amount of charge Q and light L depends on the energy release E and the applied electric field e according to the laws 4 : Q(e,E) = Q0(E)-^\n(l L(e,E) = L0(E)~Q(e)

+ ^)

(1) (2)

where C is a constant strongly dependent on the ionization power of the particle 5 ' 6 . C is large for high stopping power, e.g. like for nuclear recoils. Hence by comparing the charge and scintillation yields, one should be able to distinguish the nuclear recoils from e/7 signatures with high discrimination power. In addition the time decay of the scintillation light is given by two components, fast and slow. The amount of the slow signal depends on the stopping power of the particle, so with a pulse shape analysis one can further discriminate the nuclear recoils from the e/7 events. The detector is contained in a cylindrical vessel (Fig 1) where the liquid and vapour of the Argon are in equilibrium. The active volume is embedded in a high strength electric field (1 — 5 kV/cm). The high electric field is considered to reduce the recombination of the ionization electrons, even in the case of highly quenched slow nuclear recoils. At the bottom of the drift region a light transparent cathode is set at a potential (100-500 kV). A series of electrodes are installed and biased along the full drift path to keep the field uniform at a level of few %. The ionization electrons are drifted to the liquid-vapor interface and are extracted into the gas phase. In the vapour a LEM 7 is installed to provide the electron amplification by means of a high field generated in small (cylindrically shaped) holes; more details about LEM will be given in section 4. Lastly a series of photosensors are foreseen to be installed outside of the drift region below the cathode. The vessel wall is covered with foils of Mylar coated with a thin layer of Al to reflect the scintillation light.

274

Figure 1.

Schematic picture of the ArDM detector experiment.

3. High Voltage system The drift voltage is provided by an immersed cascade of rectifier cells (Greinacher/Crockroft-Walton circuit) which takes as input an alternate voltage of amplitude VQ and gives an output almost 2 • Vo times the number of stages. Such a solution has been chosen to overcome the task of feeding 500kV into the vessel. The immersed circuit exploits the high insulation capability of liquid Argon.

275

We envisage two possible configurations for the HV circuit, the first with a low number of stages. In this case capacitors capable to withstand up to 20 kV are needed. For this purpose we tested Mica capacitors 8 . The second solution is to use a larger number of stages, up to 250 with low voltage capacitors. A test with a circuit of 80 stages was realized, producing 120 kV in stable conditions. The test was done by using polypropylene capacitors with a maximum recommended voltage of 600 V and high voltage 2 fcVdiodes (Phillips BY505). 4. Large electron amplification device In order to address the task of multiplying the number of primary ionization electrons we started to develop a Large Electron Multiplier (LEM) 9 based on the GEM concept 10 . The former consists of perforated vetronite, with a thickness of ~ 1.5 mm with metallisation on both sides. The diameter of the cylindrical holes is 500 \xm and the distance between the centers of the 2 neighboring holes is 800 fim. An electron, following the electric drift lines will enter a hole of the LEM, where it will generate an avalanche given the high electric field. This avalanche produces a detectable signal on the LEM itself which is stripped to provide position coordinate determination. Single stage LEMs have been run in pure Argon, at room temperature, with different values of voltage applied on their faces(2.2 — 5.5 kV), at different values of Argon pressure (1—3.5 bar), and also with different values of the LEM thickness. In such conditions we got amplification factors in the range of 800 — 1200, particularly interesting is the result obtained at 3.54 bar which is the pressure at room temperature where the Argon gas has the same density as the vapour in equilibrium with liquid at T = 90°K. In this condition we got an amplification factor of 800 in stable operation. 5. Light detector and collector device For the purpose of reflecting the scintillation light we considered a thin layer of Al deposited on Mylar foil coated with a layer of MgFi to prevent the Al oxidation. The reflectivity of such mirrors at wavelength of 128 nm has been measured to be 91 ± 5 % u . As a baseline option we consider HAMAMATSU photomultipliers (PMT), R6235-01, modified to work in cold. They have their maximal sensitivity at a wavelength of ~ 420 nm and an overall gain of 3 • 105. In order to match the PMT to the scintillation light at A ~ 128nm we coat the PMT surface with a film of polystyrene doped with a wavelength-shifter,

276

Tetra Phenyl Butadiene (TPB). The polystyrene prevents the crystallization of the TPB. We also studied the features of a solid state light detector, the windowless Large Avalanche Photo Diodes (LAAPD) of Advanced Photonics. Such a device has shown a quantum efficiency of > 40 % at A ~ 128 nm12 . This result is quite promising even if the maximal gain of 103 of the APD, at room temperature, make them not very performing to detect single photons. 6. Conclusion We have presented our design of a 1 ton prototype whose goal is to demonstrate its validity for application to DM search.This requires a successful implementation and operation of a high drift field device, a LEM based charge readoutand a highly efficient Argon scintillation light detection system. A prototype is being setup at CERN. Our first milestone is a proof of principle on 7-rays and electron rejection vs nuclear recoils. Following its successful operation, we will consider a deep underground operation. A cknow

ledgments

We thank Mr R. Chandrasekaran, Dr. M. Laffranchi, G. Natterer, Ms P. Otiougova, Dr. C. Regenfus, for the precious contribution to this paper. This work was supported by ETH Zurich and the Swiss National Science Foundation. References 1. A. Rubbia, Talk given at IXth international conference on Topics in Astroparticle and Underground Physics (TAUP05), Zaragoza, (Spain) Sep. 2005. 2. A. Benetti et al, Nucl.Instrum.Meth. A327, 205 (1993). 3. F. Arneodo et al, Nucl.Instrum.Meth. A449, 147 (2000). 4. J. Thomas and D. A. Imel, Phys. Rev. A36-2, 614 (1987). 5. T. Doke et al, Nucl.Instrum.Meth. 134, 353 (1976). 6. A. Badertscher, M. Laffranchi and A. Rubbia, hep-ph/0412166. 7. P. Jeanneret et al, Nucl.Instrum.Meth. A500, 133 (2003). 8. http://www.reynoldsindustries.com. 9. P. Otiougova et al, Paper in progress. 10. F. Sauli, Nucl.Instrum.Meth. A386, 531 (1997). 11. A. Knecht, Diploma Thesis ETHZ-IPP Internal Report 2005-09. 12. R. Chandrasekaran, M. Messina and A. Rubbia, Nucl.Instrum.Meth. A546, 426 (2005).

DEVELOPMENT OF MULTIPLE PHOTON COUNTING COINCIDENCE (MPCC) TECHNIQUE FOR CHARACTERISATION OF SCINTILLATORS FOR CRYOGENIC APPLICATIONS* HANS KRAUS, VITALIIMIKHAILIK AND DAVID WAHL Department of Physics, University of Oxford, Keble Road Oxford, 0X1 3RH, United Kingdom A new method for measurements of the scintillation characteristics of materials has been developed. This method, called multiple photon counting coincidence (MPCC) technique, is based on the recording of a sequence of individual photon pulses resulting from a scintillation event. The distribution of the arrival times of these individual photon pulses provides information about the decay characteristic of the scintillation process and the number of photons recorded per scintillation event is proportional to the scintillation light yield. The ability to reject spurious events through off-line analysis is an important advantage of the MPCC method since it allows cleaning of the data set from pile-up events. It is shown that the MPCC technique is particularly well suited for the analysis of slow scintillation processes in the investigation of temperature-dependant scintillator properties. It is now used extensively by our group for the identification and optimisation of scintillating targets for cryogenic low-background rare event searches, such as Dark Matter and 0-v double beta decay experiments.

1. Introduction The search for rare events is one of the most active research areas of the astro-particle physics community. Finding neutrinoless double beta decay and the detection of weakly interacting massive particles require detectors capable of discriminating the rare signal in the presence of a dominating background caused by natural radioactivity and cosmic rays. Low-temperature detectors equipped with scintillating absorber achieve this discrimination via the simultaneous measurement of phonon and scintillation signals [1]. This ability to reject spurious background, together with other advantages of cryogenic phonon detectors, namely greatly enhanced energy resolution and low threshold, elevated these "hybrid" cryogenic phonon-scintillation detectors (CPSD) into the category of especially promising instrumentation for the next generation of experiments

* This work is supported by PPARC grant PPA/I/S/2001/00646.

277

278

searching for WIMP Dark Matter [2, 3], 0-v double beta decay [4] and radioactive decay of very long-living isotopes [5, 6]. Inorganic scintillators are key elements of CPSD and the identification and characterisation of promising scintillating materials is becoming an important objective for future progress in this field. In this paper we introduce the multiple photon counting coincidence (MPCC) technique for measuring the light yield and decay time constants of scintillators over a wide temperature range. The development of this technique is part of a research initiative addressing the identification and optimisation of scintillating materials for targets in cryogenic searches for rare events. 2. Problem A large number of research projects on characterisation of scintillators has focused on measurements close to or at room temperature. There are quite a few challenges to overcome when expanding characterisation to a wider temperature range. The characterisation of scintillation detectors has so far relied almost exclusively on high-gain photomultipliers (PMT) or avalanche photodiodes (APD). However it is technically very difficult to carry out measurements of light yield using these conventional techniques when the scintillator is optically coupled to the detector since the responsivity of PMT and APD varies significantly with its operating temperature. In studies of decay time constants as function of temperature additional problems arise from some decay time constants increasing significantly as the scintillator sample is cooled. The delayed coincidence single-photon counting (DC-SPC) technique has excellent timing resolution (ca. 0.5 ns) and it is commonly used for measuring decay characteristics of traditional fast scintillators with decay time constants in the range of nano-seconds to micro-seconds [7]. For decay time constants in the region of several micro-seconds, pulse shape analysis (PSA) can be implemented. There is however a source of error inherent to both measurement techniques. It is not practically possible to fully control the rate at which ionising particles or photons interact with the scintillator. Hence there are always events recorded in which a second (or more) scintillation event has occurred during the measurement period of the first, so-called multiple excitation events. These events contribute to the total signal, resulting in a roughly even distribution of time differences throughout the decay time spectrum, hence creating a noticeable tailing effect. Thus, when analysing such data one might easily misinterpret this background as an additional decay component with a long time constant.

279 3. Description of the Technique To solve these problems we turned our attention to the idea of recording the time structure of multiple photons, all belonging to the same individual scintillation events. In practice, a sequence of single-photon pulses produced by a PMT is recorded for a duration which is long compared to the duration of an individual scintillation event. Each pulse in the sequence corresponds to an individual photon impinging on the photocathode of the PMT. The distribution of arrival times of these pulses provides information on the decay characteristics of the scintillation process. The number of single photon pulses recorded per event is proportional to the intrinsic light yield of the scintillator. By recording a large number of scintillation events (103 — 104) one can obtain the decay time characteristics and the light yield in a single measurement. Due to the large number of photons detected in each event, sufficient accuracy can be achieved already with far fewer scintillation events than needed for the DC-SPC technique, which only uses one time difference between arriving photons per scintillation event. A helium flow cryostat equipped with optical windows to allow the detection of scintillation light is the centre piece of our experimental setup used for MPCC measurements. The scintillator sample and the 241Am a-source are placed inside the cryostat and two PMTs see the sample at right angle to each other. The signals produced by each PMT and their associated pre-amplifiers are passed to integrating amplifiers. Only one PMT is connected to the input of a transient recorder while the trigger is based on coincidence in both PMTs in order to minimise the recording of spurious events. Event selection for triggering is based on the slow integrated and amplified signals. The transient recorder has sampling intervals ranging from 5 to 80 ns and is able to record up to 16384 samples per record. All scintillation events recorded by the transient recorder are stored in a binary data file and are then subjected to data analysis using tailor-made software. The algorithm for the elimination of double events is based on the ratio of likelihoods of the decay time constant X calculated for each event and the most likely time constant Xm taken from the distribution of X -values for all records. A detailed description of the experimental setup and data analysis procedure can be found elsewhere [8]. 4. Demonstration of the method An illustration of how well the data analysis works is given in Fig. 1, showing the decay time spectrum of CaW0 4 as it was measured (a) and after elimination of spurious events (b). It is clearly visible that when multiple events are rejected, the decay spectrum shows virtually no background. The example presented here demonstrates that the procedures implemented give a much

280

20

40

60

80

100

120

Time, iiS

Figure 1. Decay time spectrum of scintillation in CaW04 at T=295 K. as it was measured (a) and after elimination of a spurious events (b).

104

I i i i i i i i . • , i i i

CD

103

f

ivN£

102 10'

1

|

•

•

•

r

1

•

•

•

.

a:

6

' rV ,3

rfHWtoiiLjj^

L

UiPfP

• vli

-in" i .1.

M

0

JJJuJSE LJUILIJ j i i i "

iuhu.. 200 300

lil J LI HUM

100

400

500

Time us :

"A'A T,

A

A

b :

A

3

A

100

1

I

3 °

a

a

^ :

1

•

n

10 °

°

;

10

°

0

0

° o

„ ^ ° o

° o o o c ^ o

^

I s

'

100 Temperature, K

Figure 2. (a) Decay time spectra of CaW0 4 measured at temperatures 345 (1), 295 (2), 170 (3), 45 (4), 30 (5), 17 (6) and 9 K (7). (b) Temperature evolution of three decay time constants u , i2 and T3 (open symbols) and light yield (filled symbols).

281

clearer decay time spectrum of the scintillation process, which in turn permits the decay time parameters to be determined with better accuracy. The MPCC technique was used in our studies of scintillation properties of several tungstate and molybdate crystals. Figs. 2a and 2b show the results of these measurements for CaW0 4 . Both, the high and low temperature decay constants and the variation of the scintillator's light yield with temperature were found to be consistent with published data [9-11]. It was shown that the MPCC technique with its associated data analysis is most beneficial for characterising "slow" scintillators and it is definitely needed for characterisation of these scintillators at low temperatures where their response becomes even slower. 5. Conclusions It is concluded that the MPCC technique can be used successfully for investigations of scintillation characteristics (decay time and light yield) over a wide temperature range. It is particularly useful for studies of slow cryogenic scintillators used in searches for rare events. An additional benefit is that it is a costeffective experimental method which can easily merge with the conventional DC-SPC technique used routinely for decay time measurements. References 1.

P. Meunier, M. Bravin, B. Bruckmayer et al, Appl. Phys. Lett, 75, 1335 (1999). 2. S. Cebrian, N. Coron, G. Dambier, et al, Astropart. Phys. 21, 23 (2004). 3. G. Angloher, C. Bucci, P. Christ et al, Astropart. Phys. 23, 325 (2005). 4. A. Alessandrello, V. Bashkirov, C. Brofferio et al, Phys. Lett. B420, 109 (1998). 5. P. de Marcillac , N. Coron, G. Dambier et al, Nature 422, 876 (2003). 6. C. Cozzini, G. Angloher, C. Bucci et al, Phys. Rev.CIO, 064606 (2004). 7. M. Moszynski, B. Bengtson, Nucl. Instr. Meth. A 142 417 (1977). 8. H.Kraus, V.B.Mikhailik and D.Wahl, Nucl. Instr. Meth. (2005) in press. 9. G. B. Beard, W. H. Kelly and M. L. J. Mallory, J. Appl. Phys. 33, 144 (1962). 10. V. B. Mikhailik, H. Kraus, G. Miller et al, J. Appl.Phys. 97, 083523 (2005). 11. Yu. G. Zdesenko, F. T. Avignone III, V. B. Brudanin et al, Nucl. Instr. Meth. A538, 657 (2005).

RESULTS F R O M CUORICINO E X P E R I M E N T A N D P R O S P E C T S FOR C U O R E

M. Pedretti 9 , R. Ardito 1 ' 2 , C. Arnaboldi 1 , D. R. Artusa 3 , F. T. Avignone III 3 , M. Balata 4 , I. Bandac 3 , M. Barucci 5 , J.W. Beeman 6 , F. Bellini 14 , C. Brofferio1, C. Bucci 4 , S. Capelli 1 , F. Capozzi 1 , L. Carbone 1 , S. Cebrian 7 , M. Clemenza1, C. Cosmelli14, O. Cremonesi 1 , R. J. Creswick 3 ,1. Dafinei14, A. de Waard 8 , M. Diemoz 14 , M. Dolinski 6 ' 11 , H. A. Farach 3 , F. Ferroni 14 , E. Fiorini 1 , C. Gargiulo 14 , E. Guardincerri 10 , A. Giuliani 9 , P. Gorla 7 , T.D. Gutierrez 6 , E. E. Haller 6 ' 11 , I. G. Irastorza 7 , E. Longo 14 , G. Maier 2 , R. Maruyama 6 ' 11 , S. Morganti 14 , S. Nisi4, C. Nones 1 , E. B. Norman 13 , A. Nucciotti 1 , E. Olivieri5, P. Ottonello 10 , M. Pallavicini 10 , V. Palmieri 12 , M. Pavan 1 , G. Pessina 1 , S. Pirro 1 , E. Previtali 1 , B. Quiter 6,11 , L. Risegari 5 , C. Rosenfeld3, S. Sangiorgio9, M. Sisti 1 , A. R. Smith 6 , L. Torres 1 , G. Ventura 5 , N. Xu 6 , and L. Zanotti 1 (1) Dipartimento di Fisica dell'Universita di Milano-Bicocca and Sezione di Milano dell'INFN, Milano 1-20126, Italy (2) Dipartimento di Ingegneria Strutturale del Politecnico di Milano, Milano 1-20133, Italy (3) Dept.of Physics and Astronomy, University of South Carolina, Columbia, South Carolina, USA 29208 (4) Laboratori Nazionali del Gran Sasso, 1-67010, Assergi (L'Aquila), Italy (5) Dipartimento di Fisica dell'Universita di Firenze e Sezione di Firenze dell'INFN, Firenze 1-50125, Italy (6) Lawrence Berkeley National Laboratory, Berkeley, California, 94720, USA (7) Laboratorio de Fisica Nuclear y Altas Energias, Universidad de Zaragoza, 50009 Zaragoza, Spain (8) Kamerling Onnes Laboratory, Leiden University, 2300 RAQ, Leiden, The Netherlands (9) Dipartimento di Fisica e Matematica dell'Universita dell'Insubria e Sezione di Milano dell'INFN, Como 1-22100, Italy (10) Dipartimento di Fisica dell'Universita di Genova e Sezione di Gen282

283

(11) (12) (13) (14)

ova dell'INFN, Genova 1-16146, Italy University of California, Berkeley, California 94720, USA Laboratori Nazionali di Legnaro, 1-35020 Legnaro ( Padova ), Italy Lawrence Livermore National Laboratory, Livermore, California, 94550, USA Dipartimento di Fisica dell'Universita di Roma and Sezione di Roma 1 dell'INFN, Roma 1-16146, Italy

Cuoricino is a taking d a t a bolometric experiment searching for neutrinoless double beta decay (OfDBD) of 1 3 0 Te. The detector consists of an array of large cubic Te02 crystal bolometers. Cuoricino works at about 10 mK in the Gran Sasso Underground Laboratory. Good energy resolutions were obtained (2.1 keV at 911 keV and 3.9 keV at 2615 keV at best). The counting rate in the region of OfDBD is 0.18±0.02 c/keV/kg/y. The limit for the Oi^DBD half lifetime is 2.0 x 10 2 4 years at the 90% of C.L. This results correspond to a limit for the effective neutrino mass between 0.2 and 1.0 eV, depending on the nuclear matrix elements used. A large international collaboration is working on CUORE project, a future experiment with a mass of 741 kg of TeG-2 crystal bolometers. The experiment aims to probe the neutrino absolute mass down to 50 meV and to understand if the inverted hierarchy holds in the neutrino mass pattern.

1. Introduction The recent positive results coming from neutrino oscillation experiments have increased the attention on neutrino physics. Indeed these experiments indicate that the neutrino flavor and the neutrino mass eigenstates are different and that neutrinos have non zero masses. These results are important in view of new theoretical models beyond Standard Model. However many questions regarding the properties of this particle are still open. In particular, which is the neutrino mass hierarchy? Up to now three mass scenarios are possible: the normal hierarchy, where mi « inj « ni3, the inverted hierarchy with m 3 « mi < m^ and the quasi-degenerate one with mi ~ m 2 ~ 1113. Moreover the neutrino oscillations experiments indicate only the differences among the neutrino mass eigenvalues but not their absolute values, so what are the masses, M», of the neutrino mass eigenstates fi? Another open question regards the neutrino nature: are the neutrino and the antineutrino the same particle? Possible answers could come from Neutrinoless Double Beta Decay (Oi^DBD) experiments. 1.1. The Double Beta

Decay

The Double Beta Decay (DBD) is a rare spontaneous nuclear transition *,2 where a nucleus (A,Z) becomes a (A,Z±2) nucleus. This type of decay is

284

not favored with respect to the single beta decay and so it is possible to observe it only for those nuclei whose single j3 decay is either energetically forbidden or suppressed by a large change of the nuclear spin-parity state. Mainly two different channels of DBD are discussed: the first one, the 2z^DBD, is described by the following reaction 2*vDBD:

(A,Z) -> (A, Z + 2) + 2e~ + 2ue,

(1)

where two neutrinos are emitted and the lepton number is conserved. This type of decay is allowed by the SM and presently it has been observed for ~ 10 nuclei 3 . The second possible DBD channel is the Neutrinoless Double Beta Decay described by the reaction Oi/DBD :

(A, Z) -^(A,Z

+ 2) + 2e~

(2)

Here the emission and the re-absorption of a virtual neutrino mediates the decay. This last channel requires that neutrino and antineutrino are the same particle or, as said, a Majorana particle. Moreover the decay is allowed only if the neutrino has a non-zero mass. In Oz^DBD, the two electrons retain all the available kinetic energy. For this reason, the expected spectrum is just a spike at the transition energy. On the contrary, the 2^DBD is a four body decay and so it is expected a continuum spectra with a maximum value around one third of the Q value. The total expected energy spectrum is shown in fig. 1 (left). The DBD is characterized by a very long lifetime, for example those for the 2 J / D B D is in the range of ~ 10 18 - 10 22 years. The Oz/DBD probability is connected to the effective neutrino mass, (m„), by Fermi's golden rule: J

1

l/2

= G°" | M 0 l f (m„) 2

(3)

where G0v is the phase space integral that can be estimated exactly and |M°"| is the decay matrix element. The effective neutrino mass is a linear combination of the neutrino mass eigenvalues: (m,)=^^Mi|C/ei|2

(. = 1,2,3)

(4)

i

with \Uei\ the neutrino mass matrix elements. The <> /& phases that appear in last equation are the intrinsic neutrino CP parities. Their presence implies that cancellations are possible. From eq (4) it is clear that O^DBD could help to solve questions on the absolute value of neutrino masses and to probe the hierarchy scheme as shown in fig. 1 (right) where the neutrino oscillation results have been considered.

285

10"* Fraction of d r a y energy

10'

10 !

10'

| i g h l c s , ncmrino mass in cV

Figure 1. (Left) Expected spectrum of the sum of the energy of the electrons for DDB2f (dashed line) and DDBOi' (solid line). In the inset the contribution of the 2v decay to the Of DDB background is underlined. (Right) Plot of the effective mass (m„) as a function of the lightest neutrino mass (on a log-log scale). The dark regions are the possible neutrino mass range. Here neutrino oscillation experiment results have been considered with negligible errors.

From eq. (3), it is also clear that the evaluation of (m„) from an experimental measure of T®J2 will require the exact knowledge of the nuclear matrix elements. This could be the main limitation when using OfDBD results for implications on neutrino physics. 1.2. Experimental

approaches

Most of the recent experiments use direct measurement approaches to study the Of DBD that allow to determine the electron energy and its distribution, and sometimes permit also event reconstruction. In order to perform a sensitive experiment searching for Of DBD it is necessary to take into account the difficulties due to the very long half time of the DBD. Indeed, even if the signature of the Of DBD is in principle very clear, to experimentally recognize a peak over the continuum background could be very hard. In order to increase the sensitivity of the experiment usually it is necessary: • to work with very large source masses, of order of one ton or larger in next generation experiments. • to use good energy resolution and high efficiency detectors; good energy resolution is mandatory to improve the signal to background in the peak search; moreover poor resolution means that the 2f DBD spectra would be a source of dangerous background for the Of DBD

286

peak. • to perform the experiment in very radiopure conditions; this implies that detectors must be shielded from the environment and its associated radioactivity. Cosmic rays contributions are suppressed by performing experiments in underground laboratories. • to have long data taking periods. 2. The bolometric technique Large mass thermal detectors have been suggested since 1984 for search on rare decays 4 . Over the last few years, low temperature detectors have provided better performance as regard energy resolution, low energy thresholds and wide material choice than conventional detectors. A complete overview on these detectors and the bolometric technique is offered by proceedings of specific low temperature detectors conferences 5 . Bolometers are low temperature detectors sensitive to single particle interactions. The basic idea of bolometers is to measure the energy lost in particle interactions as a temperature variation of the detector. The two main components of a bolometric detector are an energy absorber, where particles lose all or part of their energy and produce elementary excitations, and a sensor which collects these excitations generating an electrical signal. As a first approximation, the height of the thermal pulse is given by the ratio of the deposited energy to the heat capacity C of the bolometer, whereas the pulse time constant is equal to the ratio between C and the thermal conductance of the detector to the heat sink. Consequently, in order to achieve big and fast signals, it is important to make bolometers with small heat capacity. This is obtained by using dielectric diamagnetic materials kept at very low temperatures (~10 mK in our case). Their heat capacity in fact scales as the cube of the ratio between the operating temperature T and the Debye temperature QD (Debye law). The mechanism that limits the energy resolution in these conditions is due to thermodynamic fluctuations. It can be shown that the intrinsic energy uncertainty is in this case 6

AErms

~ (kT^C)1'2

(5)

where k is the Boltzmann constant. The Milano group started some years ago a program to develop both large mass bolometers to search for Oz^DBD of various isotopes and microcalorimeters for X-ray and ^-spectroscopy. A series of bolometric experiments have been carried out by the Milano group since 1989 in the

287

Gran Sasso Laboratory, searching for the double beta decay of 130 Te. Prom the experience of this group and with an international collaboration, the Cuoricino experiment started at the beginning of 2003. In October 2003, an enlarged collaboration has proposed a new generation large mass experiment named CUORE (Cryogenic Underground Observatory for Rare

Events) that has the search for O^DBD of 130Te with the bolometric technique as a main goal.

3. The Cuoricino experiment: detector and results Cuoricino is the biggest running O K D B D experiment. The source of O J / D B D is 130 Te that, thanks to its high natural abundance, permits to work with large active masses also without isotopic enrichment. Moreover the transition energy value of the O^DBD of 130 Te is in a favorable position of the energy spectrum with respect to the natural radioactive background, indeed it is outside the 2 3 8 [ / background and in a window of low natural radioactivity: between the full energy and the Compton edge of 208Tl gamma peak (2615 keV). As absorbers of the bolometer, Cuoricino uses Tellurium Oxide (Te02) crystals, that of course contain 130 Te. In fact it is possible to grow large single Te02 crystals with excellent properties. Furthermore, the Debye temperature of TeOi is higher than pure Te and so lower heat capacity and higher pulses are obtainable. In addition, Te dominates in mass the compound and these crystals have a high radiopurity (< lpg/g in 232Th and 23SU). As sensors, Cuoricino uses germanium Neutron Transmutation Doped (NTD) thermistors, whose resistivity at low temperature depends steeply on temperature. More details on these devices could be found in 7 . Cuoricino detector is a tower-like structure that uses 40 kg of Te02- The array is composed by 44 crystals of 5x5x5 cm3 size and a 760 g weight and 18 crystals of 3x3x6 cm3 size and a 340 g weight placed in 13 elementary modules. A picture of Cuoricino detector is shown in fig. 2 (left). Cuoricino detector has reached good energy resolution in the DBD energy region (3.9 keV at the best, and 7 keV as average at 2615 keV with the big size crystals) and is characterized by good efficiency (86%). An example of energy spectrum obtained during a calibration measurement is shown in fig. 2 (right). The current background in Oi/DBD region is 0.18±0.02 c/keV/kg/y. The present best interpretation of this continuum background, coming from

288

1500 2000 Energy [keV]

3000

Figure 2. (Left) Picture of Cuoricino detector. It contains 62 TeC>2 crystals arranged in 13 planes. (Right) Example of an energy spectrum obtained during a calibration measurement. This spectrum is the sum of all the spectra obtained by the 44 5x5x5 cm 3 detectors.

Montecarlo simulation of the whole detector, is that it is mainly due to degraded alphas coming from passive materials that face the crystal absorbers, as the copper structures holding the detectors. This shows that Cuoricino experiment is not only a powerful experiment that is giving important results on DBD but it is also a good instrument to obtain information in view of CUORE Project. Cuoricino is presently giving a limit of 2.0 x 10 24 years at the 90 % of C.L. on T®Y2, corresponding to a bound on (mv) between 0.2 and 1.0 eV. The double decay region of the energy spectrum is shown in fig. 3. Cuoricino can reach a sensitivity on neutrino mass between 0.13 and 0.31 eV in a few years 8 . Unfortunately, due to the uncertainties on nuclear matrix elements, it will not be able to exclude totally the range indicated by part of the H-M collaboration for effective neutrino mass 10 , but has the potential to verify this claim. 4. The C U O R E experiment CUORE (Cryogenic Underground Observatory for Rare Events), that has been proposed and partially funded 9 , will be placed in hall A of the] Gran Sasso Underground Laboratory. The CUORE detector will be constituted by 19 Cuoricino-like towers packed in a cylindrical structure for a total of 988 Te02 detectors and a total mass of ~741 kg and corresponding to ~200 kg of 130 Te . Like Cuoricino detector, CUORE will work at about 10 mK in a new big dilution refrigerator built with low radioactive materials. CUORE

289

2<8Q

2520 iinergy

Figure 3. Energy spectrum in the Neutrinoless Double Beta Decay region.

is planned to start in 2010. In a conservative hypothesis about background (0.01 c/keV/kg/y) and with 10 keV of energy resolution CUORE should obtain in 10 years of data taking a sensitivity on half life of O K D B D of 2.1 1026 years corresponding to a neutrino mass limit ranging between 19 and 167 meV. In a optimistic, but realistic, hypothesis regarding the background (0.001 c/keV/kg/y), with 5 keV of energy resolution and 5 years of data taking, CUORE is foreseen to have a sensitivity on half life of 0 J / D B D of 6.5 10 26 years and a neutrino mass limit of 11-57 meV. 5. C U O R E R & D and the Surface Sensitive Bolometers Big efforts on CUORE R&D are performed in order to obtain a significant reduction of radioactive background in the O J / D B D region. Tests on a second dilution refrigerator in hall C of LNGS are in progress and the CUORE collaboration is working on both passive and active background reduction techniques. Last tests on the use of very radiopure powders to lap the crystals and on the use of chemical etching to polish their surfaces showed a reduction of about a factor 4 in the peaks of 238 U and 2 3 2 Th alpha decays. This reduction could be attributed to contaminations in the crystal surface. Study of cleaning procedures for the copper surface that surrounds the detector are in progress, in particular the collaboration is working on a plasma cleaning technique that is giving positive preliminary results. The CUORE collaboration is working on new detectors to make a discrimination of background events. These devices are called Surface Sensitive Bolometers (SSB) and are able to actively discriminate events orig-

290

inated near the crystals surface n . They are composite bolometers able to identify degraded alphas and beta particles coming from outside and reaching the crystal absorber. For this purpose thin bolometers (secondary bolometers) are glued around the main large TeC>2 bolometers and used as active shields. Tests with secondary bolometers of different materials were performed. As secondary bolometers and the TeC>2 absorbers are thermally coupled, pulses with different shapes are obtained depending on the particle impact point. If particles interact into the T e 0 2 crystal absorber, then both the sensors on shield absorbers and on the TeC>2 bolometer will develop pulses with the usual shape. Otherwise, in case that the interacting point of the particle is in a secondary bolometer, the sensor on this shield will read out fast and high pulses, whereas usual shape pulses will be read out by sensors on the other bolometers. The comparison between the pulses from two sensors could be carried out by means of a scatter plot, i.e. plotting the secondary bolometer pulse amplitudes versus the corresponding main bolometer pulse amplitudes. Two clearly separated bands distinguish events generated in the shield from events generated in the main crystal. Another possibility is to distinguish the origin of events by means of pulse shape discrimination.

(a)

Pulse amplitude (Ge)[mV] Surface events *r

J

400

o

it

(b)

/ Bulk / C _

100 200 300 Pulse amplitude (TeO z ) fmV]

6 8 10 Risetime (Ge thermistor) [ras] Figure 4. (Left) Scatter plot showing the relationship between the amplitudes of the pulses acquired in coincidence from the main absorber (X-axis) and from the active layer (Y-axis). In the inset, the energy spectrum obtained by selecting surface events is shown. (Right) (a) Rise Time distribution for pulses developed by sensor on the secondary bolometer, (b) Scatter plot where the different colors evidence the contribution due to fast and slow signal.

In fig. 4 (left) and 4 (right) it is possible to observe an experimental

291

Pufe*AnvUud«OT
Figure 5. Differences between the decay time achieved with a normal CUORE-like detector (left) and a detector with SSB (right). In the second detector, events coming from outside interact with the secondary bolometer and cause in the sensor on the main Te02 crystal pulses with different decay times with respect to those to events that interact directly with the main absorber. scatter plot obtained at Insubria University with a SSB composed by a 2 x 2 x 2 cm 3 TeC>2 bolometer and by a 1.5x1.5x0.05 cm 3 Ge secondary bolometer and working at about 20 m K . Tests of this new technique are in progress at the G r a n Sasso Underground Laboratory on a CUORE-like single module. Also, even if t h e results are really preliminary, they show a large discrimination capability and powerful rejection. Moreover they indicate the possibility t o discriminate the event origin by using not only pulse amplitude and rise time of pulses on secondary bolometers, but also decay time of corresponding pulses on the main bolometer. In fig. 5 it is shown the decay time as seen from the main TeC-2 crystal of a normal detector and of a detector with SSB.

References 1. 2. 3. 4. 5. 6.

Elliott, Steven R. and Vogel, Petr, Ann. Rev. Nucl. Part. Sci. 52, 115 (2002) A. Faessler, and F. Simkovic, J. Phys. G24, 2139 (1998). Moe, M. K., Nucl. Phys. Proc. Suppl. 38, 36 (1995). E. Fiorini, and T. O. Niinikoski, Nucl. Instr. Meth. A224, 83, (1984). A.Giuliani, Proc. SPIE, 4507, 203, (2001). S.H. Moseley, J.C. Mather and D. McCammon, J. Appl. Phys. 56, 1257, (1984). 7. M.Barucci et al., J. Low Temp. Phys. 123, 303, 2001. 8. R. Ardito, and CUORE collaboration, hep-ex/0501010. 9. R. Ardito, and Cuoricino collaboration, Phys. Rev. Lett. 95, 142501, (2005) 10. Klapdor-Kleingrothaus, H. V. and Dietz, A. and Krivosheina, I. V. and Chkvorets, O., Nucl. Instrum. Meth. A522, 371 (2004). 11. L. Foggetta et al., Appl. Phys. Lett 86, 134106, (2005)

TIMING CALIBRATION OF THE NEMO OPTICAL SENSORS M. CIRCELLA, C. DE MARZO 1 , R. MEGNA, M. RUPPI 2 for the NEMO Collaboration INFN, Sezione di Bari and Dipartimento di Fisica dell 'Universita di Bari, Italy e-mail: martino. ruppi@ba. infn. it This paper describes the timing calibration system for the NEMO underwater neutrino telescope. The NEMO Project aims at the construction of a km3 detector, equipped with a large number of photomultipliers, in the Mediterranean Sea. We foresee a redundant system to perform the time calibration of our apparatus: 1) A two-step procedure for measuring the offsets in the time measurements of the NEMO optical sensors, so as to measure separately the time delay for the synchronization signals to reach the offshore electronics and the response time of the photomultipliers to calibration signals delivered from optical pulsers through an optical fibre distribution system; 2) an all-optical procedure for measuring the differences in the time offsets of the different optical modules illuminated by calibration pulses. Such a system can be extended to work for a very large apparatus, even for complex arrangements of widely spaced sensors. The NEMO prototyping activities ongoing at a test site off the coast of Sicily will allow the system described in this work to be operated and tested in situ next year.

1.

Introduction

Very large volume underwater or underice Cherenkov detectors represent nowadays the most common approach to neutrino astrophysics. These detectors consist of very large (up to the order of 1 km3), three-dimensional arrays of photomultipliers which detect the Cherenkov light emitted by high-energy muons which are created in neutrino interactions inside or around the apparatus. The capabilities of these experiments are strictly dependent on how well time measurements are performed in the detectors. In fact, the reconstruction of the physical events, extracted from the large environmental background typical of the marine abysses, is based on the possibility to properly align in time the measurements performed by a large number of widely spaced sensors. In all cases in which the signals are digitized and time-stamped offshore, two different tasks are required for aligning in time the measurements: synchronization of the electronics, i.e. the delivery of common clock signals to the whole apparatus,

1 2

Deceased. Presenting author.

292

293 and timing calibration of the sensors, i.e. the measurement of how the local time measurements performed by the sensors compare to the time measured onshore. In this paper we illustrate the timing calibration system of the NEMO neutrino telescope, focussing on the approach to the solution of this problem, the overall design of the system and the validation measurements performed in the laboratory. 2.

The approach of the Timing Calibration in NEMO

The NEMO apparatus is described in detail elsewhere [1]. The Timing Calibration in NEMO project requires an embedded system in order to track the possible drifts of the time offsets during the operations of the apparatus underwater. The task of this embedded system is essentially to measure the offsets with which the local time counters inside the optical modules are reset on reception of the reset commands broadcasted from the shore, i.e. the time delays for such commands to reach the individual optical modules. All time measurements are in fact referred to the readout of such counters. Due of the effect of the timing calibration, we propose to realize a completely redundant system: 1) a two-step procedure for measuring the offsets in the time measurements of the NEMO optical sensors; 2) an all-optical procedure for measuring the differences in the time offsets of the different optical modules. In the first system the needed measurements are performed in two separate steps: - with an 'echo' timing calibration; - with an "optical" timing calibration. The former will allow us to measure the time delay for the signal propagation from the shore to the Floor Control Module (FCM) of each floor; the latter will allow us to determine the time offsets between the FCM and each optical module connected to it. The second system is an extension of the "optical" timing calibration, as is discussed in the next section. 3.

Optical Timing Calibration in NEMO

The "optical" timing calibration system will allow to measure the time delays between each FCM and the optical modules connected to it. The scheme of the system to be implemented on each floor of the towers is shown in Figure 1. In this scheme a blue light pulse is generated from a calibrated optical source (inside each FCM) upon reception of the calibration command from the shore. This pulse is then injected into four optical fibres of known lengths in order to simultaneously illuminate the optical modules of the floor. This simple system represents an efficient way to distribute common calibration signals to groups of optical modules. In fact it is important to point out that the use of beacons which deliver light flashes into the water, as is made in other detectors, is not a convenient solution due to the peculiar arrangements of the optical modules on the NEMO tower and the large distances among them.

294 Optical Module

yr Optical fibre

[illigt^Sil

V

splitter

Optical source

FCM (FloorControl Module)__

V.

FCM electronics Interface

Figure 1. The optical calibration system of NEMO.

The pulser indicated in Figure 1 has been implemented and has undergone a long characterization activity. The schematic of this circuit, which was evolved from the circuit discussed in [2], is shown in Figure 2.

Figure 2. Schematic of the pulser circuit.

This circuit features a LED Agilent HLMP CB15, which was carefully selected to work as the source of the fast light pulses with a blue wavelength as we need. In view of the fact that the light pulses from this LED are not very well focussed, we introduced a commercial optical collimator Thorlabs F230FC-A to increase the optical power picked up by the fibre. In order to cope with attenuation introduced by the cable and connectors we have developed a remote control to manage the pulse intensity of the LED by means of the regulation of the diode voltage. This voltage is set from two levels of regulations respectively for the coarse and fine tuning of the intensity of the light pulses. So, it will be possible to tune the intensity emission of the LED from the shore. In the Figure 3 it is possible to see a photograph of the board equipped with the pulser and digital circuit controller.

Figure 3. LED pulser complete of collimator and fiber coupling.

295 In order to measure the fast and faint light pulses emitted by the pulser, fast photomultipliers Hamamatsu H6780 were used in laboratory tests. These devices offer very appealing characteristics: they are compact, fast (rise-time less than 1 ns) and reliable; in addition, they are internally equipped with a high voltage supply. The coupling of the photocathode to the fibre can be performed easily with a commercial adapter. By means of these detectors, we performed an extensive evalutation of the performance of different multimode and singlemode fibre for the transport of the fast light pulses in the wavelength range of our interest, over the required length. We finally selected the Thorlabs ASF50/125 multimode fibre, which will be used for the implementation of the system in NEMO. The pulser performance is illustrated in Table 1. Table 1. Performance of the LED pulser of the timing calibration system of NEMO. Falling time

- 2 . 5 ns

FWHM

<7ns

Standard deviation of time from trigger

<100 ps

The pulser has also been coupled to the large photomultiplier Hamamatsu R7081SEL which will be used in the activities of NEMO Phase 1. Due to the large dimensions and high sensivity of this photomultiplier, the measurements were performed in a dedicated dark room. An illustration of the signals measured in these conditions is in Figure 4.

Figure 4. Light pulse from the pulser as viewed with a photomultiplier Hamamatsu R7081SEL (blue trace: trigger signal; black trace: photomultiplier output).

4.

Conclusions

The timing calibration is a delicate operation to perform for assuring the full functionality of a large underwater neutrino telescope. In this paper we have

296

illustrated the main features of our project for a timing calibration system tailored to the needs of the NEMO apparatus. We point out however that the approach proposed can be easily adapted to work for any underwater neutrino telescope. We want also to mention a possible extension to the system discussed so far, which offers increased reliability for the whole apparatus. In fact, the pulser that we have developed can deliver very intense light flashes which can therefore serve a larger number of photomultipliers than 4 as considered so far. It is therefore possible to distribute the signals from the same calibration pulser to the optical modules of more than one floor of a NEMO tower, as illustrated in Figure 5. Moreover, with the new design of the pulser board it is now possible to accurately control the intensity of the light pulses. This solution will provide an important redundancy to the 'echo' calibration procedure mentioned in Section 2.

O-

To upper floor

E V_.-'

•> I I

e

r

E

From lower floor

Figure 5. Extension of the optical calibration system of Figure 1 in which calibration signals are distributed to the optical modules of more than one floor of the NEMO towers.

5.

Acknowledgments

This work is dedicated to the memory of Carlo De Marzo. Carlo was an enthusiastic supporter of neutrino telescope in the Mediterranean Sea. The NEMO Collaboration misses the clever scientist and a good friend. References 1 2

E. Migneco, these proceedings. J. S. Kapustinski, Nucl. Instr. Meth. A241: 612, 1985.

D E S I G N , P R O D U C T I O N , A N D FIRST RESULTS F R O M T H E I C E C U B E DIGITAL OPTICAL M O D U L E

0. TARASOVA FOR THE ICECUBE COLLABORATION* DESY Zeuthen Platanenallee 6, D-15837 Zeuthen, Germany E-mail: [email protected]

The first string of IceCube - a kilometer-scale, high energy neutrino detector - was deployed in January this year deep in the Antarctic ice. At completion, IceCube will consist of at least 70 strings each with 60 Digital Optical Modules (DOMs) and a surface array of 280 DOMs. 2004 was the first year of real production and testing of these DOMs: 400 DOMs were produced in the US, Sweden and Germany. A series of tests such as basic functionality, phototube dark noise, optical sensitivity, linearity, time resolution, and high voltage tests were performed in special Dark Freezer Laboratories at temperatures ranging from +25 °C to -55 °C. 280 DOMs were sent to the South Pole for deployment and tested there again prior to deployment. Now frozen in the ice, the DOMs are showing exceptionally promising functionality. Here DOM production and testing procedures and corresponding results are discussed and data from the first in-ice string are presented.

1. Physics Motivation The main goal of the IceCube experiment is to reach an understanding of the origin of high-energy cosmic rays as well as to test a fundamental lows of physics. Low fluxes of HE neutrinos require kilometer-scale detector volumes to provide information about the sources of these particles with reasonable statistics. IceCube is designed to search for sources of high-energy neutrinos such as Active Galactic Nuclei (AGN), Supernova Remnants (SNR), microquasars or Gamma Ray Bursts (GRB). The sensitivity of IceCube to astroparticle sources is discussed in Ref.1. Due to low ambient noise in the ice, IceCube has the opportunity to detect low-energy (MeV) neutrinos from supernovae of your galaxy. In addition to high-energy astronomy IceCube physics extends to the field of particle *http://icecube. wisc.edu

297

298

physics by searching for neutrinos from possible cold dark matter candidates, weakly interacting massive particles (WIMPs), magnetic monopoles etc. 2 ' 3

2. The IceCube Detector IceCube is a kilometer-scale, high energy neutrino observatory located at the South Pole in Antarctica (Fig. 1). The construction of the detector has started in January 2005 and at completion it will consist of two subdetectors made of more than 4200 optical modules 5 . The in-ice part includes up to 70 strings mounted at depths of 1450 to 2450 m in the ice. Strings are regularly distributed 125 m apart. Each of them carries 60 optical sensors vertically spaced by 17 m. Above the in-ice part of the detector an air shower array IceTop is located at the surface. IceTop modules are being build on top of in-ice part at the surface. IceTop consists of 160 stations (2 modules each) and covers an area of 1 km 2 . These two sub-detectors can be used either in anti-coincidence mode where IceTop provide a veto for down-going muons (main background for IceCube) or in coincidence mode synthesizing information on chemical composition of cosmic rays for energies up to 10 18 eV. It also can be used for detector cross-calibration. The detector will provide an effective detection area of 1 km 2 for upward-going muons in the TeV range. Above 100 TeV detection of down-going neutrinos (i.e. observation of the southern hemisphere) will be possible. The effective area for PeV muons is above 0.6 km 2 and increases towards the horizon.

IceTop

Snow layer

South Pole Station

IceCube

Figure 1. The IceCube detector at the South Pole.

299

3. Digital Optical Module (DOM) 3.1. DOM

Design

The Digital Optical Module (DOM) (Fig. 2) is the most important component of the IceCube detector. The main principle of the DOM is a combination of large area photo multiplier coupled with support electronics for PMT power and digitization of the PMT anode pulses. PMT signals are digitized inside the DOM and get a global time stamp with a precision of better than 5 ns. This is a key concept for IceCube as it overcomes many problems associated with analog signal transmission over long distances, up to 3.5 km in the case of the IceCube detector array. DOM technology as well as the installation procedure of IceCube are built on the wealth of experience gained with its smaller prototype - the AMANDA detector 4 .

penetrator arcl pressure t •! •

•

•

.tl i

.!]• ' s i r d

be*— PMT optical

L Figure 2.

. metal cast-

The IceCube Digital Optical Module components.

For IceCube the Hamamatsu R7081-02 photomultiplier tube (PMT) has been chosen. It has low noise (200-300 cps @ -40 °C) and high gain as well as fast PMT pulse risetimes (4 ns). PMT high voltage bias is supplied by a HV generator board that is mounted on the PCB stack around the PMT's neck. It produces voltage from 0 up to +2048 V controllable from the surface. The divider circuit is a high impedance (130 Mfi) resistive bleeder (hias — 10 fiA nom) with a toroidal transformer at the anode signal output to block the DC high voltage of the PMT. The phototube and electronic is potted into the glass pressure sphere using a clear RTV gel. Gel holds the PMT providing a good optical coupling and mechanical

300

support. The magnetic metal cage shields the terrestrial magnetic field to improve photoelectron collection efficiency. Each DOM contains 12 individually selectable LED's installed on a DOM flasher board. This is a powerful tool to obtain optical properties of the surrounding ice and geometrical calibration information. One LED is capable to emit 1010 photons per pulse. This light can be detected by modules located few hundred meters away.

3.2. The Data Acquisition

System

(DAQ)

The core of the IceCube data acquisition system (DAQ) is the DOM Mainboard. Mounted inside the DOM affixed to the neck of the PMT, it contains the electronic circuits for controll, readout, digitizing, processing, and buffering of PMT signals (Fig. 3). The PMT signals are fed to the front end electronics where they are split into two paths: one travels directly to a trigger discriminator system which can start the acquisition when the pulse height exceeds a programmable level; the other path is delayed and then is further branched into several independent waveform capture channels. Two parallel Analog Transient Waveform Digitizer (ATWD) chips capture fast waveforms with 10-bit resolution and with sampling speeds programmable from 250 MHz to 1 GHz. Each ATWD has 4 input channels. Three of them are used with different gain paths, 16x, 2x, and 0.25x resulting in a large dynamic range and giving an effective resolution of 14 bits. The fourth channel is used for monitoring and calibration purposes. A 40 MHz, 10-bit flash ADC digitizes slow waveforms from distant high-energy events which are spread out over timescales of microseconds by scattering in the ice. Digitization is controlled by the Altera Excalibur FPGA and a hardcore ARM CPU on a single die. The FPGA provides the real-time interfacing to the digitizers, various on-board electrical and optical calibration devices, flasher board, and the communications DACs and ADCs. Waveform signal processing is also done in firmware. The CPU executes application codes such as DOM self-test and self-calibration applications and data acquisition programs that buffer the waveform data from the digitizers, packages it with time stamps from the local oscillator and other local state information, and send the data packets to the surface DAQ. System software can be downloaded to the DOM at any time before and after modules are deployed in the ice.

301

Digital Optical Module Block Diagram {^9

l 1'C'iA

jb«-»-ISJ|iS^r|)>iJS-

$

-

DC-DC |

Corifi^urdlion

•«

40 M i l /

,1H

l

I

SHRAM

tfh

SDR
'*

1 lash I C omtiifj f requeticv Cif {was Toyocoitij

Monitor J'/vepT,!?-''*." & Conlrol i*.e:WKS£v'.v'--J ° ? f * * A ' ' ? PMT l>owcr 8 b , 10b, ITh

Figure 3.

3.3. DOM

•

•

; 1lasher 1 Bodid

DOM Block Diagramm.

Integration

DOM assembling is supported at 3 institutions of the IceCube collaboration: UW - Madison (USA), DESY (Germany), Stockholm University (Sweden) and each production site is provided by all necessary DOM components. The procedure of DOM integration requires special equipment and includes several steps: (1) A molded plastic collar is attached to the PMT neck to provide a base for electronics installation. (2) The HV divider is soldered onto the PMT flying leads. (3) The phototube is placed into the lower hemisphere of the glass pressure vessel surrounded by the magnetic shield and the RTV gel, properly mixed and degassed. (4) In 24 hours the gel is cured and all electronics components such as the delay board, mainboard, HV control board, and flasher board are installed. (5) The penetrator cable is soldered to the mainboard. (6) Finally the glass sphere is sealed with an internal pressure of 0.5 atm. Those DOMs which pass the final acceptance testing are attached with a harness assembly and sent to the South Pole.

302

4. Final Acceptance Testing (FAT) Before the DOMs are sent to the South Pole series of tests such as basic functionality, dark noise, optical sensitivity, PMT calibration, time resolution, and linearity are performed on all of them in the dark freezer laboratory (DFL). This is final acceptance test (FAT) lasts for about three weeks. During this time DOMs experience a temperature ranging from -55 °C to +25 °C. This proves the operativeness of the DOMs ander real in-ice conditions. Each FAT includes also a long stability test (at least 180 hours) at a temperature of -45 °C what is very close to the temperature of the antarctic ice at the depth of IceCube.

Figure 4. Dark Freezer Laboratory Setup.

DOM test stations are arranged on shelves inside the DFL. Each DOM is placed on a plastic cylinder which is foiled on the inside to reflect more light and compensate for the bias towards illuminating the center. Three light sources are used during the FAT: a 405 nm diode laser, monochromatortuned tungsten lamp, and a fast LED pulser. Light is distributed to the station via optical fibers which are packed together into a tight bundle at the input. Light from each source is focused directly onto the ~mm-sized spot of bare fiber ends (Fig. 4). A computer-controlled filter-wheel attenuator is installed just in the front of the bundle to regulate overall intensity. At the station end, the light exiting the fiber is diffused using a holographic diffuser material to distribute light uniformly over the cylinder. The optical fiber system was designed to be invariant over the wide temperature range.

303 5. South Pole Deployment During the last austral summer season the first in-ice string of IceCube was deployed. About 300 DOMs were shipped to the South Pole and tested there again prior to deployment. To drill IceCube holes an enhanced hot water drill (EHWD) was developed. Using this technology it takes slightly less than 50 hours to drill holes of 2.5km depth and 60 cm diameter in the ice. In addition 4 IceTop tanks (8 DOMs) have been installed on the surface what makes a total of 76 deployed DOMs which are functioning and producing physics data. All DOMs are operational and meet the design requirements. Obtained noise rates for in-ice modules are about 700 Hz on average. Time resolution measured with flashers is better than 2 ns. Data taking had started after the first surface tanks and the in-ice string were deployed and will continue during the detector construction. The plan for the next season is to deploy up to 10 in-ice strings and 32 IceTop stations. Continuing at a rate of 16-18 strings and 32-36 IceTop tanks per year IceCube will be completed within the next 5 years. Acknowledgments This material is based on work supported by the following agencies: U.S National Science Foundation (NSF); German Ministry of Research and Education; Swiss National Fund; Federal Office for Scientific, Technical and Cultural Affairs, Belgium; Funds for Scientific Research (FNRS and FWO), Belgium; Institute for the Promotion of Innovation by Science and Technology in Flanders, Belgium; German Research Foundation (DFG), Germany; Inoue Foundation for Science, Japan; Knut and Alice Wallenberg Foundation, Sweden; Swedish Polar Research Secretariat, Sweden; The Swedish Research Council, Sweden; Department of Energy, Office of Science, USA; University of Wisconsin Alumni Research Foundation, USA. References 1. 2. 3. 4. 5.

J. Ahres et. al., Astropart. Phys. 20, 507 (2004). F. Halzen, astro-ph/0311004. C.Spiering for AMANDA collaboration, Proc 27th ICRC (2001) 1242. http://amanda.wisc.edu J. Ahrens et. al., The IceCube Proposal to NSF (2000), J. Ahrens et. al., PDD:IceCube Conceptual Design Document (2001).

This page is intentionally left blank

Calorimetry Organizer: C. Leroy

S. Bimbot P. Bloch A. Bornheim D. Bryman I. Evangelou M. Groll E. Giilmez S. Hagopian V. Hagopian M. Ippolitov A. Krokhotin P. Meridiani M.M. Obertino

C. Oliver J. Pinfold R. Ruchti R. Salerno

Study of Electromagnetic Calorimeter Trigger-Primitive Performances for the CMS Data Selection The CMS Electromagnetic Calorimeter The CMS ECAL Laser Monitoring System Pointing Calorimeter for Measuring K°L -> K^VV Decay and Development of Extruded Scintillator CMS Preshower In-situ Absolute Calibration Dedicated Front-End Electronics for an ILC Prototype Hadronic Calorimeter with SIPM readout Using Single Photoelectron Spectra in the Calibration of the CMS-HF Calorimeter Jet Energy Scale Determination for the Run II DO Calorimeter The Hadron Calorimeter of the CMS Experiment at the Large Hadron Collider Properties of a 256-channel Prototype of the ALICE/PHOS Calorimeter Source Calibration of the CMS Quartz Fiber Forward Calorimeter The Calibration Strategy of the CMS Electromagnetic Calorimeter The Very Front-End Cards for the CMS Electromagnetic Calorimeter: Description, Calibration and Performance Uniformity of Response of the ATLAS Electromagnetic Calorimeter Series Modules Plans for the Very Forward Region of ATLAS - the LUCID Luminosity Monitor Fast, Long-Wavelength Scintillators and Waveshifters Test Beam Results of the CMS Electromagnetic Calorimeter 305

M. Schram C. Serfon A. Topkar A. Ulyanov K.-C. Voss

Electron and Pion Results from the ATLAS Foward Calorimeter 2003 Test Beam Linearity, Energy and Position Resolution of the ATLAS Electromagnetic Calorimeter Series Modules Preshower Silicon Strip Detectors for the CMS Experiment at LHC Test Beam Results of the CMS Forward Quartz Fibre Calorimeter Inter-Calibration and Offline Compensation Studies of the ATLAS Calorimeter Using Beam Tests and Monte Carlo

STUDY OF ELECTROMAGNETIC CALORIMETER TRIGGERPRIMITIVE PERFORMANCES FOR CMS DATA SELECTION STEPHANE BIMBOT On behalf of CMS ECAL Collaboration Laboratoire Leprince-Ringuet, Ecole Polytechnique, Route de Saclay 91J28PALAISEAU, France CMS will be one of the LHC detectors at CERN. The flux of data will be so high that a drastic on-line selection is needed. The Trigger Primitives (TP) are a basic element of this selection. In the electromagnetic calorimeter (ECAL), they are used to estimate the transverse energy, and to find the bunch crossing corresponding to the event. The study presented in this paper shows that the linearity and the resolution of the ECAL-TP response are satisfactory, and that the efficiency of the bunch crossing identification is perfect for particles above 1 GeV.

1. Introduction The Large Hadron Collider (LHC) is a proton-proton collider under construction at CERN (the European Laboratory for Particle Physics). It will be able to generate collisions at energies up to 14 TeV in the center of mass frame. The nominal luminosity of this collider will be 1034 cm"2 s"1. The Compact Muon Solenoid (CMS) [1] will be one of the general-purpose detectors operating at LHC. It is designed to observe the Higgs Boson, and to study physics beyond the standard model. Because of its high resolution, the electromagnetic calorimeter (ECAL) [2], made of 75,848 lead-tungstate crystals, is a key-element to identify the Higgs boson (through gamma-gamma decay, or ZZ* into four electrons decay channels). At LHC, a bunch crossing will happen every 25 ns. For the ECAL only, the data volume to store would be 80,000 Gbytes per second. How to transmit such a data flux? As most of these data will not contain any interesting information for Physics purpose, an on-line selection will be done to transmit only useful events: the Level-1 (LI) Trigger [3]. For the ECAL, the aim of the trigger is to select the events in which high transverse momentum electrons or photons are identified. For this selection, both the transverse energy and the compactness of the shower (which signs the * For each crystal, a vector is evaluated, the norm of which corresponds to the value of the energy deposited in the crystal, and the direction of which is given by the axis linking the vertex (here

307

308

presence of an electron or photon) have to be estimated on-line. This is precisely the task of the Trigger Primitive Generation (TPG). 2. Trigger Primitives 2.1. The Trigger Primitive Generation: Hardware Description The ECAL is made of a barrel and two end-caps. The following study has been done in the barrel only, where a trigger tower is defined as a group of 25 crystals. Each trigger tower is divided into five strips of five crystals (along the phi coordinate). Each strip is linked to an electronic board called "Very Front End" (VFE). The 5 VFE of each trigger tower are linked to another board called "Front-End" [4]. This board contains six ASICS involved in the Trigger Primitive (TP) calculation called "Fenix". Five of them are configured as "FenixStrip" and one as "Fenix-TCP". The light produced in the scintillator crystals is collected by two Avalanche Photo Diodes (APD). The signal is then shaped with a multi-gain amplifier, and digitized by an ADC (Analog to Digital Converter). LMMrUM PiW»

FitMrod pulac

time dwonvoluBon

.

AOMr

=7 Mm! be piopcrly configured <

Figure 1. The different steps of the signal propagation as performed by each Fenix Strip.

A Fenix-Strip takes as input the digitized samples coming from each crystal to estimate the transverse energy in each strip. As seen on Figure 1, the signal is first linearized for each crystal: the ADC samples are multiplied by a factor taking into account the gain value and the position of the crystal, making the signal amplitude proportional to the transverse energy. Next, the "Adder" sums

approximated by the zero of CMS) and the center of the crystal front face. The norm of the transverse component (with respect to the axis of the beam) of this vector is called "transverse energy". This notion is interesting because it gives, at high energy, an evaluation of the transverse momentum.

309 the values found for the five crystals. Then, a filter reconstructs the transverse energy of the strip using a weight-method (linear combination of digitized samples), and finally, a peak-finder identifies the maximum output of the amplitude filter. The time position of this maximum makes it possible to identify the bunch crossing corresponding to the event. The outputs of the five Fenix-Strips are used by the Fenix-TCP to add up the transverse energy from the five strips (ETot), and to find out the maximum transverse energy in two consecutive strips. These two outputs are used to estimate the compactness of the electromagnetic shower characterized by a single bit [3]. This bit and ETot are combined into one word of 11 bits for each tower: the TPG output, transmitted to the Level-1 trigger processor. 2.2. Validation of the Simulation using Electron-Beam Data To study the TP response, data taken in an electron beam in November 2004 were used. However, the amount of data was not sufficient for a detailed study, especially at low energy. Moreover, the condition of the data taking was different from the conditions at LHC: in test beam operation, the phase between the ADC clock and the signal pulse is random. So, for the global study, new sets of data had to be simulated. This implied first to simulate the energy deposited in each crystal (using the Monte-Carlo Simulation "Geant 4" [5] and real pedestal runs without beam to have a realistic electronic noise), and next, to have a perfect simulation of the response of the electronics. The simulation of this electronics response follows closely the hardware description given in the previous section. In order to check it, we used digitized samples from real data sets as input. The response of this electronics simulation was compared to the Trigger Primitives measured in these real data sets. This simulation reproduces perfectly (bit to bit) the real electronics response. Table 1. Comparison between Data and Simulation t^Bcam

(GeV) 15 20 30 50 120





39.93 ±0.07 55.10 + 0.08 85.55 ±0.10 147.10±0.17 369.67 ±0.63

39.91 ±0.06 55.22 ± 0.08 86.10 ±0.12 148.18 ±0.20 367.24 ±0.51

a Data 1.72 ±0.04 2.29 ±0.04 3.13 ±0.06 5.31 ±0.10 12.20 + 0.44

a Sim 1.79 ±0.04 2.26 ±0.04 3.22 + 0.06 5.46 ±0.13 12.87 ±0.39

To check the validity of the global simulation, its results were compared with the real data at the beam energies of 15, 20, 30, 50 and 120 GeV. For each energy, we used about 1000 events (data and simulation) for the comparison. Data were

310 selected such as the incident beam could be considered as point like. For each beam energy, the distribution of the energy of the TP has been fit using a gaussian function. In Table 1, the mean value and the standard deviation are shown for the data and the simulation. Clearly, data and simulation are in good agreement within 1% for the mean values and 5.5% for the standard deviations. This last difference is only observed at 120 GeV, the agreement being much better at lower energies. The overall agreement between data and simulation is satisfactory for the purpose of this study, focused on low energies. 3. Trigger Primitive Results based on Simulation For the following study, the global simulation (Monte-Carlo plus electronics simulation) has been used, with the conditions of CMS in situ (random phase between the ADC clock and the signal pulse). The study has been done for two different impact points in the trigger tower: one in the center, and one in the corner. A simulated beam of 2 mm around the center of the crystal front face has been used. Figure 3 shows the transverse energy estimated thanks to the TP by the LI Trigger algorithmt [3] versus the energy of the incident particle up to 120 GeV. The parameters in the electronics simulation were set to get an optimized resolution at low energies. These parameters imply a saturation of the TP response around 130 GeV. pTPB Linearity!

_

f

Center Crystal:

„/r-'

A D C „ ^ , » ( 7 . B 8 ± 0 . 0 4 » E a - w - <3.S8±0.79) Carnmr

Crystal:

A O C t ( K , - ( 7 . 7 O t 0 . 0 S ) E I W V . <3.17±0.79>

-

^

^

^ '

^" """

s ^

r

^

s: . 20, . . . 4,0 . . . ,8 0. . . ,80 .

oC-r O

,

,

i , , , [ , ,i 100 120 E n a r g y (<3«V)

Figure 3. Response of TP algorithm (sum of 2 trigger towers) as function of the incident energy for 2 impact points: center crystal (plain line) and corner crystal (dashed line).

A linear fit has been applied in Figure 3 for the 2 impact points. It shows that the response of the TP is linear over the energy range considered within the statistical errors (%2/NDF=4.l/8). For a given energy, the value reconstructed is slightly lower when the impact point is in the corner because of the ECAL LI ^The trigger algorithm uses the sum of the TP values in the tower hit by the particle, and in one of the adjacent towers; among the four possibilities, the tower with the maximum energy is used.

311

Trigger which associate only two trigger towers. The resolution of the TP response measured by the ratio between the width and the mean value of the TP distributions for a given energy is shown in Figure 4 as function of the energy. The measured resolution fully satisfies the CMS trigger requirements: for example, it gives a resolution of 2% at 23 GeV (the threshold of the LI trigger for single electrons). The resolution is almost independent on the impact point. [

TINS. flhMekilkHR™

| Bunch Crossing ID I

i

Resolution.: =

\ 8-8% Y | f ° - 1 7 Y

o.oa 0.06 0.04 0-02

1

Figure 4. Center crystal (plain line) and corner crystal (dashed line) TP resolution.

10

JOa Energy (GaV)

Figure 5. Bunch crossing assignment as function of the incident energy.

The TP can also be used to identify the bunch crossing in which the event occurred: the time position of the maximum of the signal collected by a trigger tower (response of the peak finder, see section 2.1) makes it possible to determine this bunch crossing. The efficiency, i.e. the probability to assign correctly the bunch crossing to the event, is shown in Figure 5 versus the incident energy of the particle. This efficiency is almost 100% above 1 GeV. This result corresponds to CMS trigger requirements (see [3] p 89). 4. Conclusion The simulation performed in this work matches the data within the statistical error. Its use to the test of TPG suggests a satisfactory linearity and resolution. It also indicates that this TPG system allows an excellent bunch crossing identification for particles with energy higher than 1 GeV. References 1. CMS Technical Proposal, CERN/LHCC 94-38 2. CMS Collaboration, The Electromagnetic Calorimeter Project, Technical Design Report, CERN/LHCC 97-33, CMS TDR 4, 1997

3. CMS Collaboration, Trigger and Data Acquisition Systems, Technical Design Report, CERN/LHCC 2000-38, CMS TDR 6.1, 2000 4. N. Pastrone, in proceedings of Workshop on Electronics for LHC and future experiments, Heidelberg 2005, CERN-LHCC-2005-038 5. Geant4 collaboration MM A 506, 2003

T H E CMS ELECTROMAGNETIC CALORIMETER

P. B L O C H PH Department,

CERN

O n behalf of t h e C M S E l e c t r o m a g n e t i c Calorimeter G r o u p The CMS electromagnetic calorimeter (ECAL) aims at measuring with high precision the photons and electrons produced in the collisions at LHC. It contains « 76000 Lead-tungstate crystals. The main challenges are to measure the electromagnetic shower energy with a precision better than 1% above 100 GeV and to operate reliably in the harsh radiation environment of LHC. We present a short overview of the calorimeter design, recent progress in its construction and results of system tests performed recently with fully assembled Barrel supermodules.

1. Introduction The prime motivation of the LHC is to discover the nature of symmetry breaking. The Higgs mechanism is the currently favoured process. In the mass interval 114-130 GeV the two-photon decay is the most promising Higgs boson decay channel. The natural width of the Higgs boson is small and the measured width of a signal will be entirely dominated by the instrumental two-photon mass resolution. High resolution electromagnetic calorimetry is therefore a central design feature of the Compact Muon Solenoid (CMS), a general purpose detector to be installed at the LHC proton-proton collider 1. The best performance in terms of energy resolution is obtained using fully active calorimeters. CMS has chosen lead tungstate (PbW04) crystals. These crystals are fast, radiation tolerant and have a short radiation length (XQ = 0.89 cm) and Moliere radius(2.1cm), allowing the design of a compact and highly granular detector. The main challenge is to build a detector which can reliably perform during more than 10 years in the high radiation levels of the LHC interactions region. The use of lead tungstate imposes further, specific requirements. The low light yield requires use of photo-detectors with intrinsic gain that can operate in a magnetic field; silicon avalanche photo-diodes (APDs) have been chosen for the barrel crystals and vacuum phototriodes (VPTs) for the endcap crystals. The change of

312

313 response with temperature due to both the variation of the scintillation efficiency and of the APD gain, puts strong requirements on the temperature stabilisation. Finally, the small, but non negligible variation of transparency of the crystals during irradiation requires a precise light monitoring system. In section 2 , we present a brief overview of the ECAL design. Section 3 is devoted to the status of construction of the various elements. In section 4 we report on the results of system tests using fully assembled sections of the detector in the laboratory or in test beams at CERN. crystals in a

preshower

EC crystals

Figure 1. The CMS Electromagnetic Calorimeter

2. C M S E C A L design The layout of the CMS ECAL 2 is shown in Fig. 1. In the Barrel, the 23 cm long crystals (25.8Xo) are grouped into 5 x 2 matrices, held in a glass fibre alveolar submodule, of which 40 or 50 are mounted into a module. Four modules are assembled together in a supermodule, which contains 1700 crystals. Eighteen supermodules form a half barrel covering the range of pseudorapidity |?7| from 0 to 1.48. Two APDs, embedded in a plastic capsule, are glued to the back of each crystal. They are connected in parallel to the read-out electronics, which is mounted on the outside of the mechanical structure. The electronic read-out chain follows a modular structure whose basic elements are matrices of 5 x 5 crystals corresponding to a trigger tower. The APDs are connected to the Very Front End (VFE)

314 boards which shape, amplify, and digitise the signals. The digitised data are then stored in the FE board and the trigger primitives (the energy sum in a tower) are generated and transmitted to the trigger electronics by a 800 Mbit/s optical link. On receipt of a level-1 accept, the data are transmitted to the off-detector electronics by a separate optical link. The Endcaps, covering the rapidity range \r]\ from 1.48 to 3, consist of 4 Dees containing 3662 crystals each. The 22cm long crystals are held in a carbon fibre alveolar, forming a 'supercrystal' containing 25 identical crystals. A VPT is glued on the back of each crystal. The supercrystals are countilevered from a thick Aluminum back plate. The readout electronics is the same as for the Barrel. To improve the 7—7r° separation, a Preshower covers most of the Endcap area. It contains 2 planes of silicon (1.9 mm pitch) strip sensors, positioned after two lead absorbers of 2 and 1 radiation length thickness respectively.

3. Construction progress Crystals: Sixty-seven percent of the Barrel crystals have been delivered. The majority of the crystals is produced at the Bogoroditsk plant (BCTP) in Russia, but recently a second source of supply has been opened with the Shanghai Institute of Ceramics (SIC) in China. The last Barrel crystal will be delivered by the end of 2006, and the last Endcap crystal one year later. Each crystal undergoes dimensional and optical checks in fully automated equipment located at CERN and in Rome. For BCTP crystals, the radiation hardness can be inferred from optical properties. The inferred radiation hardness is checked regularly by performing irradiation measurements on a sample of crystals. For SIC crystals, no such correlation has been found yet and every crystal is pre-irradiated. Photodetectors and electronics: All the APDs and 80% of the VPTs have been produced and tested. In 2002, the electronics chain has undergone a major, but very successful revision using the 0.25/um CMOS technology. All the custom ASICs have now been produced. About 75% of the Barrel VFE boards are at hand. The serial integration of electronics in Supermodules has started. Assembly: Twenty-two of the 36 Supermodules have been mechanically assembled. By the end of 2005, all the large mechanical pieces for the Endcaps will be delivered. Preshower: The production of the Preshower silicon sensors is 90% completed, the specifics ASICS have been produced and are being tested.

315

The electronics hybrid production is launched. 4. R e s u l t s of s y s t e m t e s t s In the past year, several Supermodules have been fully assembled including the final electronics and its associated cooling. One of them was thoroughly studied during Fall 2004 in the H4 electron beam at CERN. Most of the tests focused on the energy resolution and particularly on effects which could affect the constant term, which limits the resolution at high energy. T e m p e r a t u r e stability: the variation with temperature of the calorimeter response was measured to be « —4%/°C. Given the stability of the cooling water the only source of temperature variation within the ECAL is a possible variation of the power dissipated by the electronics. To investigate the sensitivity to this, the temperatures of thermistors in the APD capsules were measured with the electronics switched on and switched off. The maximum measured change in temperature AT on _ 0 fr was 0.1°C, with a mean change of 0.056°C (Fig. 2). Therefore the contribution to the constant term of the energy resolution due to thermal fluctuations will be negligible, even without temperature corrections.

"h

10

5

/ ,

/recovery

i . . . i. . . i

u

b)

Irradiation

12

c) 0 95 0.94

2

r~i r

n -0.004 -0.002

0

0.02

0.04

0.06

"h 0.C

0.93

0.1

0.12

0.14

0.95

S/S0 = <««„)" a.1.6

0.96

0.97

0.98

0.99

flT(K)

Figure 2. A T o n - 0 f r for 40 crystals equipped with a temperature sensor

Figure 3. Normalized electron vs laser response during irradiation and recovery

C r y s t a l t r a n s p a r e n c y corrections: the validation of the light injection system with its final laser and read-out electronics and the feasibility of determining the short-term transparency corrections were the goals of dedicated beam tests. The channel response to laser light without any incident beam, normalized to reference diodes, was stable to 0.1%. During the

316

tests with beam, crystals were exposed to a high intensity 120 GeV electron beam with dose rates from 0.2 to 0.4 Gy/h at the shower maximum, larger than that expected in the barrel at high luminosity. Under these conditions, the signal loss is expected to saturate at around 5%. The correlation between the electron and laser signal is shown for one crystal in Fig. 3. It allows one to correct for the signal loss with a precision < 0.3 %. Electronics performance: the electronics noise was studied using random triggers. The noise per channel ranges from 35 to 45 MeV with very little spatial correlation between channels. Finally, the energy resolution was measured using data taken with electron beams of momenta ranging from 20 to 250 GeV/c. The Supermodule was mounted on a rotating table, allowing the beam to cover almost uniformly the front face of each crystal. Fig. 4 shows the energy resolution as a function of energy for nine crystals. The energy resolution above 100 GeV is typically 0.5%, meeting the goal set for the CMS ECAL.

Erwrgy(GeV)

Figure 4.

Energy resolution as a function of energy for 9 crystals

5. S u m m a r y The construction of the CMS Lead tungstate crystal calorimeter is well advanced, with 60% of the Barrel Supermodules already mechanically assembled. The procurement of the Barrel electronics will be completed by spring 2006. All major mechanical components for the Endcap ECAL are also available. The construction schedule is driven by the crystal delivery, which should allow completion of the Barrel before LHC startup and of the Endcaps for the first physics run in 2008. System tests and test beam results with fully equipped Supermodules confirm that the CMS ECAL will

317

meet its demanding design goals. Acknowledgments The author would like to acknowledge the large dedicated community of scientists, engineers and technical staff who have worked extremely hard to ensure the success of this project. References 1. http://cmsdoc.cern.ch/docLHCC.shtml 2. The Electromagnetic Calorimeter Technical Design Report, CERN-LHCC 9733 (1997).

THE CMS ECAL LASER MONITORING SYSTEM

ADOLF BORNHEIM On behalf of the CMS ECAL

Collaboration

Lauritsen Laboratory, CALTECH, 1200 E California Blvd. Pasadena, CA 91125, USA The CMS detector at LHC will be equipped with a high precision lead tungstatc electromagnetic calorimeter. To ensure the stability of the calorimetric response at a level of a few per mille, every channel of the detector is monitored with a laser system. This system allows to correct for fluctuations in the detector response with high precision, in particular the expected radiation induced changes of the crystal transparency, which are on the level of several per cent at nominal luminosity. We report results from recent tests on a super module of the CMS ECAL equipped with final design components. The performance of the monitoring system is demonstrated and details of the monitoring strategy will be discussed.

1. Introduction One of the main physics goals of the CMS detector at LHC is the search for the Higgs boson. In the mass range around 125 GeV, the decay H—>yy is of particular interest because of its clean final state. This mode however requires a very good energy resolution for the electromagnetic calorimeter to take advantage of the expected narrow decay width of the Higgs boson in this mass range. To fulfill this requirement, the CMS experiment features a high resolution crystal calorimeter utilizing lead tungstate (PWO) crystals [1, 2]. These crystals do experience a very slight, dose rate dependent decrease of their transparency under irradiation, which recovers to a large extend in irradiation free periods. In CMS, the variation of the crystal transparency under irradiation and the corresponding decrease in light out put of the crystals will be monitored with a high precision light monitoring system. In this paper we describe the experience with the final design monitoring-system in the test beam at CERN and discuss the strategy for correcting the variation in the crystal response during LHC operation.

318

319

2. Radiation Induced Transparency Change in PWO4 Crystals The mass produced lead tungstate crystals used for the ECAL are radiation hard to high integrated dosage, but they experience a dose rate dependent transparency loss, as shown in Figure 1 (left). Although there is no damage to the scintillation mechanism, color centers are created during irradiation. These radiation induced color centers may annihilate at room temperature. During irradiations both annihilation and creation processes coexist and the color center density reaches an equilibrium at a level depending on the dose rate applied, as shown in Figure 1 (center) [3].

Figure 1 : In the left plot the transparency loss in a lead tungstate crystal is shown as a function of the wavelength. The loss is most pronounced close to the peak of the emission spectrum at around 440 nm. On the right, the transparency loss is shown as a function of the irradiation time. The transparency loss saturates after an initial decrease.

During operation of the accelerator the does rate for the crystals in ECAL will vary significantly, depending on the instantaneous luminosity, the beam quality and the location in the ECAL. A typical rate for the barrel crystals will be 15 rad per hour and up to a factor 100 higher for the crystals closest to the beam pipe. The transparency loss will recover to a large extend when the irradiation stops, typically in a fast initial recovery with a time constant of several tens of hours followed by a slower recovery on the time scale of hundreds to a few thousand hours (Figure 1, right). This is to be compared with the time cycle in the LHC operation of 12 hours. It should be noted that the exact amount of the transparency change as well as the time constants vary from crystal to crystal. 3. The CMS ECAL Laser Monitoring System The laser monitoring system must track the change in the transparency precisely enough to maintain the constant term in the ECAL resolution of 0.55%. This requires a measurement of the transparency with an accuracy of better than 0.2%. The transparency in each crystal will have to be measured approximately every 30 minutes during LHC operation to account for the fast initial change at the beginning of an irradiation cycle. To achieve this, the monitoring has to be

320

performed while the data taking of the detector is ongoing. This can be realized by taking advantage of gaps in the LHC bunch train which occur every 90 [is.

-kiii-liin In 1» J»iiill io kkkIIIII'KII'MI'.VII'DISBIMI |<_LSrjsaa—^_ i^ ^ ^ Laartrigger

Light

J

Figure 3 ,

.T h e .

layout of the C M S .

ECAL ,

moi»™ni laser monitoring source. Two laser Figure 2 : Scheme of the continuous laser monitoring S y Ste ms provide light at 440 nm and during LHC operation. Taking advantage of a 3 LIS gap in goo nm as primary and secondary the LHC bunch train laser pulses are injected at a rate of monitoring wavelengths. 100 Hz.

To allow this operational mode the laser source has to fulfill a number of design parameters. In particular the laser pulse width must be shorter than 40 ns and stable to better than 10% to match the shaping of the ECAL electronics. Furthermore must the pulse timing be stable to a few ns to allow a synchronization which the LHC clock and the detector readout. Finally, the pulse energy must be on the order of lmJ to provide sufficient light to large groups of crystals to ensure that the entire ECAL can be monitored in 30 minutes. The pulse energy also has to be stable to better than 10% to prevent possible non-linear effects in the reference system against which the signals of the individual crystals are compared. To achieve all these performance requirements a dual-stage laser system is used [4, 5]. Two such laser systems provide two different wavelengths, on at the peak of the scintillation light wavelength at 440 nm, one far away in the red wavelength range to allow for systematic cross checks. These two systems distribute the light pulses via a system of optical switches to the ECAL super modules, as shown in Figure 3.

4. Performance of the CMS ECAL Monitoring System The feasibility of the radiation damage monitoring with the desired precision has been repeatedly demonstrated in the test beam. This is done by irradiating crystals in a high energy electron beam with dose rates similar to the ones expected in CMS. During the irradiation the crystals are calibrated with the electron beam in regular time intervals, followed by laser monitoring runs. The

321

Crystals

loss in the electron beam signal response can then be correlated with the loss in the signal response to laser shots, which is shown in Figure 4 (left). C M S E C A L M 0 ' 2002 B e a m T e s t

ft 8

-

19 Crystals

6

•

^ ^ = 6.3%

4

rJ

2

;

:..,,.: .,• , 0.5 1.0

0.96 0.97 0.98 0.93 1.00 M. Normalized Laser Monitoring Response °

15

• •

•

2.0

•

2.5

3.0

Figure 4 : The left plot shows the correlation between the transparency loss measured by the laser system and the signal loss of high energy electrons under irradiation. The right plot shows the slope of this linear correlation for a group of 19 crystals. It is relatively uniform among crystals from the same producer.

The correlation can be parameterized by a linear function with slope a. As shown in Figure 4 (right), the spread of a among different crystals was measured to be 6.3% in a sample of 19 crystals. Given the average signal loss for a CMS-like radiation environment, such a small dispersion would allow to use the same correction parameter for all crystals. It should be noted that the change in the ECAL response during normal LHC operation might be relatively small because the recovery time of the crystals is much longer than one LHC cycle, as detailed in section 2. . The slope of the correlation is known to vary for crystals from different producers and is different for the barrel and the end cap of the ECAL, due to the slight differences in the crystal geometry and the light injection. Methods to determine a in-situ from physics events used for calibration are being considered. 11.004

-_ -S

'* . - ' • - • " ' . ' ' • •

" V*

73200

73400

73600

73B00

74000

74200

74400

74600

0.002 0.0035 Monitoring stability

Figure 5 : Stability of the monitoring system over Figure 6 : Stability over a period of 450 hours for a period of 450 hours on a final design system, a group of 400 channels. The typical stability is at the level of 0.)%. shown as an example on a single channel.

322

To ensure that the constant term of the resolution can be maintained it is crucial that the transparency measurement itself is sufficiently stable. The careful design of the entire monitoring system ensures that the transparency can be monitored with a precision of typically 0.1% over several hundred hours. In Figure 5 we show the stability of an individual channel of a fully equipped super module over 450 hours. Each point comprises a transparency measurement with the laser monitoring system. In Figure 6 we show the stability of a group of 500 crystals over the same time period. 5. Summary The CMS ECAL laser monitoring system has been tested at CERN for several thousand hours. All design performance parameters have been achieved. In particular it has been shown that the radiation induced transparency change in the CMS ECAL can be monitored with a precision of typically 0.1% over several hundred hours. The effect of the transparency change expected for normal LHC operation can hence be corrected and the effect reduced to a level which allows to maintain the constant term of the calorimeter resolution within the design specifications. Acknowledgments 1 would like to thank the entire ECAL collaboration for their outstanding effort, R.-Y. Zhu as well as the CALTECH team for their superb work on the laser system and the organizers of the 9th ICATPP Conference at Como. References 1. "The Electromagnetic Calorimeter Technical Design Report", CMS Collaboration , CERN/LHCC 97-33, 1997 2. "The CMS ECAL", P. Bloch, these proceedings. 3. "Radiation Induced Color Centers and Light Monitoring for Lead Tungstate Crystals", X. Qu, L. Zhang, R.-Y, Zhu, IEEE Trans. Nucl. Sci. NS-47 (2000). 4. "Performance of the Monitoring Light Source for the CMS Lead Tungstate Crystal Calorimeter ", IEEE Trans. Nucl. Sci.Vol. 52 No. 4 (2005) 1123— 1130. 5. "ECAL Monitoring Light Source at H4", D. Bailleux, A. Bornheim, L. Y. Zhang, K. J. Zhu, R.-Y. Zhu, and D. Liu, CMS IN 2003/045.

Pointing Calorimeter for Measuring K[ -» TTVV Decay and Development of Extruded Scintillator

DOUGLAS BRYMANf, JOSS IVES Department of Physics and Astronomy, University of British Columbia, Vancouver, BC, V6T1Z, Canada PIERRE AMAUDRUZ, YURI DAVYDOV, ROBERT HENDERSON, NAIMAT KHAN, CHAPMAN LIM, ANDREW MILLER, TOSHIO NUMAO , ALEKSEY SHER, DAVID WONG TRIUMF, 4004 Wesbrook Mall, Vancouver, BC V6T 2A3Canada

A sampling calorimeter based on plastic scintillator-drift chamber sandwiches was designed to measure the angles, positions, energies, and times of medium energy photons with good resolution and high efficiency. Techniques for manufacturing extruded plastic scintillators with multiple holes for wave length shifting fibers have been developed. Light output comparable to commercial scintillator and good dimensional tolerances have been achieved for 8 mm x 70 mm x 2.5 m planks which can be glued into large sheets.

1. Calorimeter for the Measurement of K°L —> n°vv The single most incisive measurement in the study of CP violation (CPV) in the context of the Standard Model (SM) would be the branching ratio B( KL —> n vv) which uniquely allows direct access to the SM CPV phase. Since the

KL—*TCVV

reaction is predicted to be extremely rare with a

branching ratio of B(K° —> TT°VV) = ( 2 . 9 ± 0 . 6 ) x l 0 ~ u , an experiment designed to observe and study it needs unprecedented sensitivity. In the KOPIO concept for this measurement, shown schematically in fig. 1, a low energy, time-structured neutral beam is used for determining the incident K momentum using time-of-flight. The detection system for observing n —> yy decay is designed to allow a fully constrained reconstruction of the decay vertex, and mass, energy, and momentum measurements in the K

center of

mass system. This is accomplished by measuring the position of interaction, Corresponding author: [email protected]

323

324

angle, and energy of each individual photon in a fine grained preradiator (PR) detector (2.7 radiation length, X0) followed by an efficient shashlyk [1] calorimeter (15 XQ ).

Figure 1. Concept of the KOPIO experiment. Photons from n —> yy decay are detected in the preradiator (PR)-calorimeter system which measures the energy, position, time, and angle of incidence.

1.1. PR Design The PR-calorimeter combination was designed to be a high resolution, high efficiency gamma ray imaging device consisting of 64 sandwiches of scintillators and cathode strip drift chambers for photons in the range of 50 to 500 MeV. Each layer had a thickness of 0.04 Xo of which about half was scintillator. The resolutions expected for the PR-calorimeter combination were

2.7% 250um, ,

= , 25 mrad, and

jE(GeV)

90/w ,

e

= • for position, energy, angle (at

ylE(GeV)

250 MeV), and time, respectively, based on prototype detector measurements and simulations. The conversion efficiency for photons in the PR was estimated to be 88%.

325

The PR design was arranged in four quadrants of modules, each of which had eight 1.5 m x 1.5 m x 7 mm drift chambers and nine, 8-mm thick sheets of plastic scintillator. The drift chambers measure the induced charge distribution on cathode strips and drift time of ionization to the anode wires to obtain the position of e e pairs. Since both particles contribute to the induced charge distributions, an accurate measure of the average position of the pair may be obtained using the cathode information. On the anode wire, only the earliest time of arrival of the primary ionization can be measured so it is expected that anode drift time will give poorer position resolution for an e+e~ pair than from the cathode strip signals. However, the resolution depends on the angle of incidence for the strip readout whereas the anode information is relatively independent of angle. Consequently, the chamber layers have alternately vertical (horizontal) and horizontal (vertical) wires (strips). The expected precision for measuring photon angles was demonstrated experimentally using 100-300 MeV tagged photons at the BNL light source.

1.2. Extruded Scintillator During the past three years we developed the technique for production of lowcost multi-hole extruded polystyrene scintillator with CELCO Industries (Surrey, BC). Readout was intended to be done with wave length shifting (WLS) fibers which are inserted into the holes without glue or optical coupling. Using holes in the extruded scintillator rather than grooves into which the WLS fibers are glued facilitates assembly and insertion of the fibers after assembly of the chamber-scintillator modules, and results in comparable light output. The polystyrene chemistry and extrusion technique were modeled after work done at Fermilab [2] which produced single hole scintillators. The extrusion planks we developed had dimensions 8 mm x 70 mm x 2.5 m with six 1.4 mm diameter holes separated by 10 mm. Photographs of planks are shown in fig. 2. To assemble a large area sheet, the plank edges are machined with a tongueand-groove shape before gluing in a jig.

326

Figure 2. Scintillator during extrusion process (left) and planks glued to form a sheet. Techniques were developed to reliably extrude multiple hole planks with uniform dimension holes, and reasonable dimensional tolerances of <0.15 mm for flatness and <0.4 mm for excursions from "straightness" over 2 m lengths. We have studied properties including hole dimensional tolerance, light output, uniformity, and attenuation. During these tests, scintillators were read out by 1 mm dia. KURARAY Yl 1(200) S fibers. The light from the fibers was collected at one end by a HAMAMATSU 3178X photo-multiplier. The other end of the WLS fiber was blackened. The tests indicated that the amount of the scintillation light produced by the extruded scintillator was comparable to that of the commercial cast scintillator BC408. The single end light yield using WLS fibers (see below) was found to be 16 photo-electrons (p.e.)/MeV, which would correspond to statistical energy 1% resolution of 0F < , =•. The measured time resolution for a single

VE(GeV) 90 ps sheet was found to be consistent with

At = - .

= with a 700 ps

ylE(GeV) constant term. The light attenuation length in the 6 hole extruded material including surface effects was found to be 150-200 mm. This is considerably shorter than in BC408 but is of minor importance in the WLS readout geometry studied. The variation of the light yield across the scintillator planks and across the joints when the scintillators are glued together with optical cement was found to be about 1.5%. In addition, it was found that the quality (surface imperfections and diameter variations) of the extruded holes had little effect on the light output.

327

Following a recent production run of 500 planks the light output dependence of various wrapping materials and coatings was studied using a 400 mm long piece of extruded scintillator with a 1.4 m WLS fiber. We observed a 60% improvement in the light output using 3M enhanced specular reflector (ESR) [3] and Avian-D [4] paint compared to wrapping with aluminized mylar. Single end readout results for these and other materials are shown in Table 1. After testing the dependence of the light yield on the Avian-D paint thickness, we found that one layer of paint (0.11(4) mm) would provide sufficient light output (88% of the maximum) and with two layers of paint (0.28(3) mm) the maximum light output was obtained.

Material Black absorber (paper) Aluminized mylar Plastic-backed aluminum foil Tyvek 3M enhanced specular reflector Avian-D paint

Light Yield (p.e./MeV) 3.63(2) 10.4(1) 10.1(1) 9.1(1) 16.4(1) 16.2(2)

Table 1. Effect of wrapping materials on the single end light yield (p.e./MeV) from the extruded scintillator read out with WLS. fibers. The tests were done with cosmic rays.

Acknowledgments This work was supported by the Natural Sciences and Engineering Research Council, the National Research Council, and the Canada Foundation for Innovation. References 1. G. S. Atoian et al. ,Nucl.Instrum.Meth. A531, 467 (2004) 2. A. Pla-Dalmau et ah, Nucl.Instrum.Meth. A466,482 (2001). 3. See http://products3.3m.com/. 4. See http://www.aviantechnologies.com/.

CMS P R E S H O W E R / i V - S m / ABSOLUTE CALIBRATION IOANNIS EVANGELOU* University ofloannina, Physics Dept., Panepistimioupoli, 45110 Ioannina, Greece

This paper describes the in-situ absolute calibration of the Preshower detector of CMS. The Preshower is based on silicon strip sensors that will be installed in the endcaps of CMS in front of the crystal Calorimeter. Energy deposited in the lead of the Preshower is estimated by the silicon sensors, allowing a re-scaling of the energy measured by the endcap crystals. Measurement of the charge deposited in the silicon strips to a few per cent accuracy is required. A precise in-situ absolute calibration of the Preshower using minimum ionizing particle signals from physics events is examined. It is shown that sufficient calibration accuracy can be obtained by using muon or pion events, and that the time required for the calibration is of the order of a few days at low luminosity

1. The Preshower at CMS The CMS Preshower detector ("ES" [1]) is based on silicon strip sensors and will be installed in the endcaps (1.653
328

329

— > z Figure 1. The Preshower detector in front of the endcap ECAL of CMS

The ES sensors measure deposited energy, normally expressed in terms of mips (minimum ionizing particles). The total deposited energy in the EE+ES system is given by E=Eg+Epresh where Eg is the energy measured by EE and Epresh= y(Ex+aEy) is the energy measured by ES. Ex,y are the deposited energies in the planes X and Y of ES respectively, a is the relative weight of the two planes and 7 is the slope of Eg[GeV] = f(Ex+aEY)[mips] plot, which effectively relates the energy deposited in the lead to that measured by the silicon sensors. Typical values of a = 0.6-0.8 and y = 0.024GeV/mip have been evaluated from beam test data and simulation. The readout electronics of ES have a large dynamic range in order to measure the energy deposited by very high energy electron-photon showers. The front-end electronics are organized as an analogue pipeline, where events are stored in a front end ASIC called PACE3 [2], followed by an on-detector analogue to digital conversion stage and a non-zero-suppressed data transmission system to the counting room. PACE3 can operate in two modes: Low Gain (LG) for normal running with a large dynamic range (1 to 400mips with a S/N ratio for single mips of ~2) and High Gain (HG) for calibration purposes (0.1 to 50 mips with a S/N ratio for single mips of -10). A measurement of single mips in HG, together with the zero point, is then complemented by the use of a charge injection circuit (inside the PACE3) to inter-calibrate the two gains, thus allowing a calibration over the full dynamic range and between channels of an individual sensor. Details of this inter-calibration procedure can be found elsewhere [3].

330

2. The absolute calibration method The strip occupancy of the ES is relatively low - a fraction of a percent at low n and low luminosity, rising to a few percent at high r| and high luminosity. This means that it is necessary to use tracks pointing to particular strips in order to obtain efficient mip distributions. The CMS simulation framework [4] was used with two approaches: a) Muon tracks from inelastic physics events reconstructed with the "GlobalMuonReconstruction" algorithm of the CMS software packages. This is a standalone method using the muon chambers (RPCs & CSCs in the endcaps) to find its own "seed" or hit in the outer region. The muon trajectory is then evaluated by following the hits all the way to the primary vertex with the "Kalman Filter" method using all muon chambers and full Tracker information. Cuts on the quality of the muon tracks reconstructed are applied at this stage. It should be noted that the fiducial coverage of the muon chambers extends to |r)|<2.4 b) b) Charged pion tracks from minimum bias or jet events. Only tracker information is used for reconstruction with the "CombinatorialTrackFinder" method. Also here, reconstruction quality cuts are applied. The CMS Tracker extends to |r||<2.5 The impact points of the reconstructed trajectories at the ES are stored for both planes (X and Y) as well as the coordinates of the strips of the ES having signals above threshold. The impact point coordinates are then compared to all found "strips with signal" coordinates for each plane of the ES and in both directions (forward and backward). A strict match is required for the first plane (|AX(X-plane)|<0.095cm - i.e. strip pitch) and the same for the other coordinate, in the second plane (|AY(Y-plane)|<0.095cm). Looser matches are required on the other coordinates (i.e. |AY(X-plane)|<3.05cm and |AX(Y-plane)|<3.05cm respectively, corresponding to the strip length). The tracks that have matched hits in both planes are recorded and their deposited energies in the ES layers are added to the mip distributions. 3. Results To evaluate the code, Monte Carlo event samples were used from the CMS production data base. In these events Gaussian noise is included in the simulated

331 signal output of the Preshower strips and threshold is applied. Initially single track events were tested of either muons or pions (with fixed p T and flat |n|<2.5). Then several fully simulated muon event samples were tested, such as: tt->/jju + X, Z/y* Z/y*->unun + X, Drell-Yann DY->uu or B°s -> JI *¥ + O , with JI *P —> MM > with pileup included corresponding to low luminosity LHC running. The tracks pointing to the ES fiducial region are first selected. Then, reconstruction quality cuts are applied (i.e. x2/ndf<1.5) and the remaining tracks are checked for matched strips with signal on both layers of the ES, as described previously. The signal distribution of the first plane of the Preshower for the collected muons of the representative it -> MM + X events sample is shown on Fig.2. A fit of a Landau distribution convoluted with a Gaussian is also shown on the same plot with a Landau "most probable value" of 86.6±0.9keV. This is in good agreement with the expected deposited energy of mips traversing the 320um silicon of the Preshower with impact angle of 10°-20° (approximately equivalent to 3.7fC). 1

cz o

35

E

30

' ' I ' ' ' ' I ' ' ' ' L a n d a u ® Gaussian fit ~ Xz 1 ndf

ES#1

Width

25

10.1 ± 0 . 8

MP

86.57 ± 0 . 9 1

Area

2247 ± 5 3 . 2

GSigma

20

154.3 / 189

18.3± 1.7

/

15 10

i

5 O

0

50

100

150

200

250 300

energy (keV) Figure 2. Preshower signal distributions for tt —> MM + X events (ES #1 layer)

The selection efficiency of the calibration method for muon tracks varies with the type of events used. In general, about 50% of the muon tracks crossing the ES layers are selected. Note that the geometrical acceptance of the Preshower for flat eta is about 30% of the total tracks, resulting in a "global efficiency" of about 15%.

332 Fully simulated pion event samples were also tested, for example jets with 2y, jets with 1 electron or minimum bias events. Looser cuts on the reconstruction quality were applied (i.e. x2/ndf<2.0). In all cases the mip distributions obtained are similar. In the case of pions the acceptance efficiency is similar to muons if high p T pions (in jets) are considered. With low p T pions (i.e. from minimum bias) the selection efficiency is significantly lower, about 5.5%, as many are not reconstructed in the Tracker. The total time required for the full Preshower calibration, taking into account the CMS high level trigger rates [5] and the LHC duty cycle, is around one month using only muon events. This can be reduced to a few days if some geometrical regularities are included, such as the forward/backward endcaps (factor 2) and using "rings" of sensors at the same pseudorapidity (factor >10). In the case of pions, the minimum bias and jet events will be a good source of mips. The time for full calibration in this case is estimated to be of the order of a few days. A combined selection of muon and pion tracks will be the best solution for fast and effective Preshower calibration. A few days of data taking at the initial run of CMS at LHC luminosity of 2xl0 33 crnV 1 . Acknowledgments This work was partially supported by CERN-PH and GSRT (Greek General Secretariat for Research and Technology). References 1. ECALEDR-04 Report Volume 2, Preshower (ES), https://edrns.cern.ch/document/l 15565/2/TAB3 2. PACE3 Digital Specifications, P.Aspell, Nov.2002, http://proikchip.web.cern.ch/proi-Kchip/preshower/doc/P ACE3DigitalSpecV3.pdf 3. Preshower in-situ absolute calibration, I.Evangelou, CMS Note in preparation 4. CMSIM, CMS simulation framework, see: http://cmsdoc.cern.ch/cmsim ORCA, Object-Oriented Reconstruction for CMS Analysis, see: http://cmsdoc.cern.ch/orca OSCAR, Object Oriented Simulation for CMS Analysis and Reconstruction, see: http://cmsdoc.ceni.ch/oscar 5. TDR, CMS The Trigger/DAQ project, Data Acquisition and High-Level Trigger, Technical Design Report, Volume 2, CERN/LHCC 02-26

D E D I C A T E D F R O N T - E N D ELECTRONICS FOR A N ILC P R O T O T Y P E H A D R O N I C CALORIMETER W I T H SIPM READOUT

M. GROLL* University of Hamburg / DESY, 22603 Hamburg, Germany E-mail: [email protected]

A prototype hadronic calorimeter for the ILC detector is under construction at DESY. It will consist of 8000 scintillator tiles with analog readout performed via SiPM directly mounted on each tile. A dedicated ASIC has been developed to match the requirements of large dynamic range, low noise, high precision and large number of readout channels needed. Particular care is given to the optimization aspects of the ASIC in the SiPM readout chain. To fully exploit the self-calibrating potentiality of this photodetector device, capable of recording single photoelectron signals, an externally selectable gain and shaping time are included in the design. The calorimeter system is described and the first calibration data are presented.

1. I n t r o d u c t i o n The requirements imposed by the high precision physics at the International Linear Collider (ILC) set high demands on calorimetry. For excellent separation of WW jets from ZZ jets, which plays a key role in Higgs boson studies, a jet-energy resolution oias/E ~ 0.3/y/E is required * 2 . This can be achieved by combining the potentials of the particle flow approach with high granularity in transverse and longitudinal directions for both electromagnetic (ECAL) and hadronic (HCAL) calorimeters. The ECAL design is based on a Si/W sampling structure with sensitive silicon pads of 1 cm2 surface. The prototype will consist of approximatly 10,000 channels 3 . A sub-group of the CALICE collaboration3, has developed the analog op*On behalf for the calice (scintillator-heal) collaboration T h e institute members of the group are: DESY, Hamburg U, ICL (London), I T E P (Moscow), LAL (Orsay), LPI (Moscow), MEPHI (Moscow), Northern Illinois U., RAL,

a

333

334

tion for the HCAL based on a sampling structure with scintillating tiles (of smallest size 3x3x0.5 cm 3 ) individually readout by Silicon PhotoMultiplier (SiPM) 4 mounted on each tile 5 . The SiPM is a pixelated avalanche photo-diode operated in limited Geiger mode. The detector surface of l x l mm 2 is divided into 1024 pixels. The analog output is obtained by adding the response of all pixels fired as independent digital counters. The SiPM are operated at 2-3 volts overvoltage. Given an internal pixel capacitance CPixei of typically 50 fF, the charge collected for one photoelectron signal is ~160 fC (or ~10 6 electrons). The SiPM offer a very fast response with a typical rise time of 3-5 ns. The dynamic range is determined by the finite number of pixels and is ~200 pC. In order to test the feasibility of the particle flow approach prototypes of both ECAL and HCAL are being built by the CALICE collaboration and will be tested in a combined test beam experiment. To meet the needs of the analog HCAL prototype it was decided to adapt the design developed for the ECAL front-end electronics 6 to read the SiPM signal. In the following the main features of the ASIC are described and its properties analyzed. 2. Very front-end electronics description and properties An 18-channel ASIC dedicated to SiPM readout has been developed starting from the design, in AMS 0.8 [im CMOS technology, already in use for the Si-W ECAL prototype readout. The integrated ASIC components allow 16 selectable preamplification gain factors between 0.67 and 10 V/pC, and 16 CR(RC) 2 shaping times between 40 and 180 ns. After shaping the signal is held at its maximum amplitude with a track and hold method and multiplexed by an 18-channel multiplexer to provide a single analog output to a 16 bit ADC. In Fig: 1 a schematic view of the ASIC chip ILC-SiPM is given. The longest shaping time has been introduced in the ECAL design to match the long delay of the beam trigger («150 ns) in a test beam environment and to minimize electronic noise. For gain calibration using LED signal, pile-up from thermal noise-induced signals can be reduced by exploiting the fast SiPM response with a shorter shaping time. An adjustable 8 bit DAC (0-5 V) in the ASIC allows individual adjustment of the SiPM bias voltage for each one of the 18 channels. UCL (London).

335 The ASIC chip properties relevant for the SiPM operation were tested with a dedicated test board. The linear range of the ASIC with the longest shaping time proves to be sufficient to read out SiPM up to its saturation. The ASIC chip output saturates between 1.7 and 1.9 V depending on the gain. A linearity of 3% in the dynamic range up to 200 pC and a negligible channel-to-channel crosstalk of 0.1-0.2% are measured with good reproducibility for all 18 channels. Fig: 2a) shows the effect of the ASIC preamplification and fast shaping on a SiPM signal. The noise introduced by the ASIC preamplifier has been found to be « 0.25 Me in this mode. Considering a SiPM gain of 10 6 = 1 Me this would give a S/N ratio of 4 for the single photoelectron signal. Fig: 2b) shows a single photoelectron peak spectrum from SiPM with ASIC readout. The DAC output voltage is found to be linear in the range 0.24 - 4.7 V, with a noise of 2 mV. The variation of the DAC voltage output for the 18 ASIC channels is below 5% (250 mV) on the full DAC range. These values are acceptable considerung that the dependence of the SiPM signal amplitude on voltage is of the order of 7%/100 mV. We conclude that the ASIC chip can be used for SiPM-like signal readout.

SiPM input

test Binput

Figure 1.

Schematic view of ASIC components.

Exploiting the channel reduction and the individual voltage steering of the ASIC chip it was possible to design the very front-end electronics for the 8000 channels of the HCAL prototype. Each of the 38 active layers is readout by 216 SiPM. On the side of each layer two base boards for the very front-end electronics are mounted. A base board hosts 6 analog boards, each equipped with one ASIC and the control and calibration electronics

336

/\ / \

0,250.20-

? T?

? 0.05. -0.05-0.10-0.15-

—

1800 1606 1400 1200

\

0.15.

0.00-

:

A

0.30-

1000

I

\

/

\ \

800 600

^-~~ "~

\ /^"^ \ j 100.0A

400 200 0

MA/U.. !

2O0,On tfme[sl

400

600

800

1000

1200 1400 ADC counb)

Figure 2. a) ASIC shaped signal for 24 ns time constant and maximum gain factor, b) Single photoelectron peak spectrum from SiPM with ASIC readout.

necessary for its operation. Bias and low voltage as well as all communication signals are distributed to all analog boards from the main base board. One HCAL layer is readout on one front-end port of a VME 64X board and processed in series with the other layers by a fast DAQ capable of 100 Hz average readout rate.

3. Calibration method For the operation of the HCAL prototype each channel is calibrated to its MIP response. The single photon detection capability of the SiPMs allows in addition the calibration of each channel to pixels per MIP. This quantity is used to apply a unique non linearity function to correct each SiPM response. In addition the stability of the pixel separation (SiPM gain) is used to monitor voltage and temperature induced fluctuations of the SiPM response. Due to the properties of our front-end electronics it is mandatory to use two different readout modes for collecting beam data and LED data. The first mode is used to determine the MIP response and the second to determine the pixel separation. Therefore an intercalibration between both modes is necessary, which is obtained by the ratio of the signals recorded in the two modes for a fixed LED light amplitude. Each channel is adjusted to a pixels/MIP value of 15 by adjusting the SiPM bias voltage. The first beam data taking period in a DESY electron testbeam was dedicated to establish this lightyield calibration method. The first MIP beam calibration data are shown in fig: 3. The determined mean value is around 13 pix per MIP with a RMS of « 2.

337 i i i i i | , , , i | , . i ,.

w | 35

I 30 25

Jr

20 15 10 5 0

•

/

i \ I

:..J]j

()

5

1

V

[

i

10

i

, .K i , , , . -

15

20 25 pix/MIP

Figure 3. Lightyield spread in the first module calibrated with electron beam. 4.

Outlook

T h e combined ECAL-HCAL prototype test in hadronic test beams will provide experience with the operation, readout, calibration and long t e r m stability of a medium size detector based on the new SiPM technology. Hadronic shower d a t a collected with unprecedented granularity will form a basis to validate and improve the simulations of shower development and prove the particle flow approach, which to large extent determines the architecture of the LC detectors.

References 1. J.-C. Brient Proc. of Snowmass 2001 C010630 (2001) E3047. 2. F. Sefkow Proc. of Calor 2004, Perugia, Italy (28 Mar.-2 Apr. 2004) Published in Perugia 2004, Calorimetry in particle physics 435-444 3. J.-C. Brient Proc. of Calor 2004, Perugia, Italy (28 Mar.-2 Apr. 2004) Published in Perugia 2004, Calorimetry in particle physics 452-467 G. Bondarenko, P. Buzhan, B. Dolgoshein, V. Golovin, E. Guschin, A. Ilyin, V. Kaplin, A. Karakash, R. Klanner, V. Pokachalov, E. Popova, K. Smirnov, Nucl. Instrum. Methods A 442 (2000) 187. V. Andreev et al. Nucl. Instrum. Methods A430 (2004) 22. S. Manen, G. Bohner, J. Lecoq, J. Fleury, C. . D. la Taille, G. Martin and c. democrite-00023563 [the FLC Collaboration Proxy] arXiv:physics/0501063.

U S I N G SINGLE PHOTOELECTRON S P E C T R A IN T H E CALIBRATION OF T H E CMS-HF CALORIMETER

E. GULMEZ, M.DELIOMEROGLU, AND K. DINDAR Bogazici University, Bebek 34342 Istanbul, E-mail: [email protected]

Turkey

U. AKGUN, A. S. AYAN, Y.ONEL, AND I.SCHMIDT University of Iowa, Iowa City 52242 IA, USA Relative gains of the PMTs installed in the CMS-HF calorimeter were measured at the University of Iowa PMT Test Station. Using the source data taken during the HF Test Beam period 2004 at CERN, PMT gains were determined from the single photoelectron spectra. Comparing these data provides an effective pedestal value and shows a very high correlation (r=0.994). PMT gains obtained in this way can be used for final calibration of the HF calorimeter.

1. Introduction Calibration of the HF Calorimeter requires the gains of the PMTs used in the calorimeter. One way of measuring the PMT gains would be to use the single photoelectron spectra. Such a set of data were taken during the test beam 2004 period with some of the HF wedges using a 5mCi 60 Co source. The following is a report of the study made on the wedge 2-13 data to compare the relative gains obtained at the University of Iowa PMT test station and the gains obtained from the single photoelectron spectra. Then, these gains are used to calibrate the calorimeter with 100 GeV electron beam. 2. University of Iowa Measurements As part of the quality control checks performed at the University of Iowa PMT test station, the relative gain of each tube was measured. The same reference tube was used throughout the measurements and the absolute gain of the reference tube was also measured accurately. 1 ' 2 Even though all the measurements were performed at 1100 V, it was shown 3 that the

338

339 X l /ndf 167.4/6 8772 ±51.0 Constant 16.71 ±0.06 Maart 7.000 ± 0.104 Sigma

| Tower 10, EM channel |

25000

I I I I I I I 111

30000

15000

I I I I I

20000

1D000

I 11 I I

\s**s.

^^V_ .

9000

)

10

20

30

40

.

.

. .

i .

90

.

.

i .

00

Figure 1. A sample fit of a single photoelectron spectrum at 1350V.

relative gains stayed flat as the HV increased. Each HF wedge is divided into three sectors; inner, middle, and outer towers. PMTs are installed into each of these sectors according to their relative gains. Tubes with lower relative gains are installed in the inner sector (closer to the beam), medium gain tubes in the middle and the higher gain tubes in the outer sector. PMTs in each sector are also selected so that they have almost the same relative gains within a few percent. With three sets of PMTs in each wedge for both type of fiber lengths, EM and HAD,we have a total of six group of PMTs providing us with six different relative gain values. 3. C E R N Measurements During the test beam period at CERN in 2004, some of the wedges were tested under various beams and also with a 60 Co source. The source was inserted into each tower one at a time and the pulse height histograms obtained through the QIE readout system were recorded. The histograms accumulated when the source was outside were used to determine the pedestals. In this study, the source runs for the HF wedge 2-13 at three different HVs, 1150, 1250, and 1350 V, were used. For each run, pulse height histograms were fitted to a gaussian to determine the peak position corresponding to the single photoelectron distribution. Then the absolute gain

340

could be obtained from the single photoelectron peak position after correcting for the pedestals. The best results were obtained from the data taken with the 1350 V (Fig.l).

25 r 20 c

.2 1 15 0L

J* CO

8. 10 " 111 a.

w

•(

5 o *-

0

50

100

150

200

Iowa Relative Gain

Figure 2. Single photoelectron peak positions or "uncorrected absolute gains in arbitrary units" (CERN) versus relative gains (Iowa). Intercept on the absolute gain axis is the effective pedestal value.

The single photoelectron peak positions were grouped according to the corresponding Iowa relative gains into six groups. When the peak positions are corrected for the pedestals, fitting the average of each group versus the Iowa relative gains to a straight line should yield a line passing through the origin. However, this does not happen when we determine the pedestals from the pedestal histograms. Instead of subtracting the pedestals obtained from the pedestal histograms, peak position versus the Iowa relative gain data were fitted to a straight line (linear correlation coefficient r=0.994) (Fig. 2). The intercept on the peak position axis would be the effective pedestal value. At 1350 V, the effective pedestal value is higher than the average pedestal value obtained from the pedestal histograms (Table 1). At lower HVs, the effective pedestal gets closer to the value obtained from the pedestal histograms. The effective pedestal value at each HV was subtracted from the single photoelectron peak positions to obtain the corrected PMT absolute gains.

341 Table 1. Effective pedestal values obtained from the single photoelectron peak positions. P M T HV

Pedestal from Histograms

Effective Pedestals

1150 V

5.0±0.3

not enough statistics.

1250 V

4.9±0.2

4.1±1.1

1350 V

4.9±0.2

7.3±1.6

4. Energy response of the calorimeter During the same test beam period, the HF wedges are exposed to several different beam types and energies. Response of the individual towers are determined by fitting the PMT pulse height distributions obtained with a 100 GeV electron beam. Since the PMTs in each of the three sectors have different gains, the PMTs show different responses. A simple approach to calibrate the calorimeter would be to correct the energy responses with the corresponding PMT gains. Figure 3 shows the individual response of the wedges corrected with the PMT gains obtained as explained above.

160 140

§ L

ergy

t as a

w

I

120 1 100 80 60

T

T

T T T

•

t U11—u^LLr_tJ] f f { •[ I n '•••Uu fi \ H \U ' ± 1 ^

I •

i

40 20 0 Tower Number

Figure 3. number

Energy response corrected with the P M T gains of towers versus the tower

342 5.

Conclusion

In this study, we show t h a t we can calibrate the H F Calorimeter and monitor the calibration by using the P M T gains obtained from the single photoelectron spectra. Then the energy response of the individual towers to a specific beam energy can be corrected with these gain values. Such a simple method results in a calibration with an accuracy of a b o u t 10%. We also show t h a t the relative gains measured at the University of Iowa P M T test station and the gains determined from the Test Beam 2004 source d a t a are highly correlated (r=0.994). However, determining the pedestal values seems to require a more careful study.

Acknowledgements This work was supported by the US Department of Energy (DE-FG0291ER-40664), Turkish Atomic Energy Commission (TAEK) and the Scientific and Technical Research Council of Turkey ( T U B I T A K ) . We also t h a n k the University of Iowa, Office of Vice-President of research and Bogazici University Scientific and Technological Research Fund (04B301) for their support.

References 1. U.Akgun, A.S.Ayan, P.Bruecken, F.Duru, E.Gulmez, A.Mestvirishvilli, M.Miller, J.Olson, Y.Onel, and I.Schmidt, Complete tests of 2000 Hamamatsu R7525HA Phototubes for the CMS-HF Forward Calorimeter,Nucl. Instr. and Meth. A, 550, 145 (2005). 2. U.Akgun, W.Anderson, A.S.Ayan, E.Gulmez, M.Miller, Y.Onel, I.Schmidt, D.Winn, Evaluation of PMTs from three different manufacturers for CMSHF Forward Calorimeter, IEEE Iran, on Nuclear Science, 51,1909, (2004). 3. E.Gulmez, U.Akgun, Y.Onel, Variation of relative gains as a function of HV for the Hamamatsu R7525HA PMTs used in the HF Forward Calorimete, CMS Internal Report CMS IN-2005/006,(2005).

J E T E N E R G Y SCALE D E T E R M I N A T I O N FOR T H E R U N II D 0 CALORIMETER

NORMAN BUCHANAN Fermilab, M.S. 352 P.O. Box 500, Batavia, IL, 60510 USA E-mail: [email protected] SHARON HAGOPIAN Dept. of Physics, Florida State University, Tallahassee, FL 32306, USA E-mail: [email protected] for the D0 Collaboration

The D0 Calorimeter system consists of three uranium/liquid-argon calorimeters and an inter-cryostat detector. The central calorimeter covers |ij| up to ~ 1, and the two forward calorimeters extend the coverage up to |»j] of ~ 4. In order to determine the particle jet energy from the measured detector jet energy, corrections must be made to account for the energy offset not associated with the hard interaction, the calorimeter response to the jet, and losses due to out-of-cone showering. These corrections will be presented as a function of detector eta and measured jet energy.

1. Introduction The determination of the relationship between the energy of a jet of particles measured in a detector and the original energy of the particles produced in the collision is crucial in order to compare measurements to theoretical models. The uncertainty in this determination is often one of the largest sources of systematic error in the measurement of physical properties such as invariant mass and production cross sections. This paper describes the procedure the D 0 experiment uses to determine the Jet Energy Scale (JES) and gives preliminary values of these corrections and corresponding uncertainties as a function of detector eta and measured jet energy.

343

344

2. The D 0 Detector The D 0 Calorimeter system consists of three uranium/liquid-argon calorimeters and the intercryostat detector. The central calorimeter (CC) covers \rj\ up to ~ 1, and the two end calorimeters (EC) extend the coverage to |r?| ~ 4. Each calorimeter contains an electromagnetic section closest to the beam line followed by fine and coarse hadronic sections. Scintillators between the CC and EC cryostats provide sampling of developing showers for \r)\ between 1.1 and 1.4. A detailed description of the calorimeter and the Run II D 0 detector can be found in the NIM paper 1 . 3. Jet Energy Scale The jets in this note are reconstructed using the D 0 Run II cone algorithm 2 , which is an iterative seed-based algorithm (including mid-points) with radius RCone = 0.7. The conversion from measured jet energy E^faa to particle level jet energy Ev^t is performed using the following correction: rPtd^

E%r-E0(nr,£) cone

(line) where Eo(Tir]C) is the offset energy, which includes multiple interactions, underlying event energy, electronic noise, uranium noise, and pile-up from previous bunch crossings, Rjet(1Zr)£) is the calorimeter response to the hadronic jet, and Rcone(R,r)C) is the fraction of particle jet energy contained within the algorithm cone. 3.1. Offset

Factor

Any energy that is not associated with the hard interaction must be accounted for. Contributions to this correction are electronics noise, pileup, uranium noise, the energy that can be attributed to the underlying event and additional minimum bias interactions. The uranium and electronics noise can be determined using zero bias events, which have the single requirement of a bunch crossing. Minimum bias events are those where the luminosity counters on each side of the detector are hit. These events are used with the information obtained using zero bias events to determine the contribution of the underlying event, and thus the offset energy distribution. Figure 1(a) shows the offset energy density, summed over 0.177 rings for one, two, and three primary vertices. The offset energy density has been

345

parameterized by the number of primary vertices, which largely removes the dependence on instantaneous luminosity. The bump in the offset energy density data comes from large weighting factors used for the inter-cryostat detector and coarse hadronic layers. 3.2. Jet

Response

The energy deposited in the calorimeter is not equal to the measured energy. This is due to the fact that the calorimeter is not completely compensating, there is dead material in front of, and within the calorimeter, there are response fluctuations between calorimeter modules, etc. The balancing of transverse energy in 7+jet events is used to obtain the response of the calorimeter to jets. For an ideal calorimeter the energy of the photon would exactly balance that of the recoiling jet. In reality, the following balance results: RemPT'y + RhadPThad

= ~$T

(2)

where Rempr-y is the electromagnetic response of the calorimeter and ^hadPThad is the hadronic response of the calorimeter. The Z mass is used to determine the electromagnetic calorimeter response and the hadronic response can then be measured in back-to-back 7+jet events. The jet response is plotted for a jet cone size of 0.7 in figure 1(b). The response values for the end-caps are scaled to account for the corresponding cryostat walls. The dominant error in the jet response comes from the differences in photon response for the forward and central calorimeters. A correction must be made for particles that scatter into, or out of, the jet cone. Particles that have trajectories taking them outside of the algorithm cone, will only have a fraction of their energy accounted for and particles that travel into the cone, under the influence of a magnetic field for example, will artificially increase the measured jet energy. In addition to the instrumental effects of out of cone scattering, physics processes can also legitimately contribute. The physics out of cone processes are determined using Monte Carlo studies. The shower corrections are given in table 1 for cone sizes of 0.5 and 0.7. Dijet and 7+jet events are used to determine the shower correction factor. The energy density is measured outward from the cone center in a radial direction (in r)- space). The offset energy density, by r?-ring, was subtracted off of the energy densities used to calculate this correction. The final correction comes from the ratio to the energy density inside the algo-

346 y Prob

5.787/12 03264 0.79M t 007627 1.1 ± 07094 0.17711 0.03461 U O I t 0.04301

f

"•

7J2V/12 0.7913 0.863710.0*489 1.11 0.04S1S 0.1696 i 0.D383S 1.477 ± 005547 -0.1233 10.1263 0.07739 ± 0.06756

•aisos i ao«66 0.02361±00486 -OJXn6S4±a006M9

0.9

= F f ^ = 0.9179 ±0.0068 : F^™ = 0.9511 ±0.0074

i

0.85 0.8

-am 15610.009652

0.75 0.7 -

X

0.65

^

/

1* s /

nR=ona = 0.7

P, + p, too (E„ 1 /E [ ; l + rv ^ / ( ^ . / E , ) p.,= 0 7 9 9 6 4 + 0 00442

pO pi p2 p3

0.6

0.6428 1 0.0S71

p,-=0 076311-0.Q0632

1.11 0.8121

p,.~. -O.0O3856 t 0.003700

0.18 i 0.02673

-I

086791003369

,

0.0134 1 0.02919

Detechx|i)|

a000112« t 0103991

|

102

10

p4

PS P6

-

,

,

,

^..[GeV]

Figure 1. (a) Offset energy as a function of detector 77 for one, two, and three primary vertices. The error bars account for statistical, 4>, and luminosity dependence uncertainties.(b) The jet response plotted as a function of detector energy for a cone size of 0.7.

rithm cone to the total summed energy. The uncertainty in the corrections is approximately 5% and primarily attributed to statistics and model dependence.

3.3.

Detector

R c o n e =0.7

R c o r i e =0.5

central ICD forward

0.99 0.96 0.94

0.92 0.89 0.85

Total Jet Energy Scale

Correction

Combining the corrections for the offset energy, jet response, and out of cone showering gives the final jet energy scale correction factors plotted in figure 2. The correction values are plotted as a function of detector 77 and transverse, uncorrected, jet energy, for a cone size of 0.7, and for events with only one primary vertex. The T42 noise suppression algorithm 3 has been applied during jet reconstruction.

347 2

Correction f . o W v . . E

•»r

.4 j -

rrr-

fc—

... •j ^ 4

i

}••]••

i-l~'"H-

i-f

••ffcafe

.8;

1

*u

C o n a d l a n factor v i , |ij |

u 4

•jr~

2

1

^~—** ™7

"\

aS

' "'

' '

t.s

! j 2.5

3

Figure 2. Data JES correction factor (left) and its absolute error (right) as a function jet uncorrected energy (top) and pseudo-rapidity (bottom). Results are shown for R=0.7 cone jet in events with 1 reconstructed primary vertex. The jet physics and detector etas are set to the same value. Up and down errors are shown separately.

4. Conclusion The jet energy scale correction has been measured for the D 0 experiment and varies in value between 1.5 at low jet transverse energies, to 1.1 at large transverse jet energies. The uncertainty in the jet energy scale is largest, approximately 18%, at low jet transverse energy and decreases to about 6% at 500 MeV uncorrected jet energy. Work is continuing to improve the D 0 jet energy scale determination. Further refinement of these corrections will lead to smaller uncertainties in measured physical quantities. 5. Acknowledgements Thanks to the D 0 Operations Group and to the D 0 Calorimeter Algorithms Group. Special thanks to the D 0 Jet Energy Scale Group for determining the jet energy scale corrections and uncertainties. The work of these and many others has led to the results presented in this paper. References 1. D0 Collaboration, "The Upgraded D 0 Detector", submitted to Nucl. Inst, and Meth. A; hep-physics/0507191; Fermilab Pub-05/341-E. 2. G.C. Blazey et al, Proceedings of "QCD and Weak Boson Physics in Run II" Workshop, ed. U. Baur, R.K. Ellis, and D. Zeppenfeld, Batavia, Illinois (2000) p. 47. See Section 3.5 for details. 3. J-R Vlimant et al, D 0 Note 4146, (2003).

THE HADRON CALORIMETER OF THE CMS EXPERIMENT AT THE LARGE HADRON COLLIDER* VASKEN HAGOPIAN Department of Physics, Florida State University Tallahassee, Florida, 32306, USA The construction of the Hadron Calorimeter is now complete. The major components of the detector have been assembled and are now in the process of installation, integration and calibration. The hadron calorimeter inside the CMS detector is made of scintillator and copper absorber covering the |nj range of 0 to 3.0. The forward calorimeter made of quartz fibers and iron absorber covers the |r|| range of 3.0 to 5.0. Segments of the hadron calorimeter have been tested in the CERN test beam with electrons, u's and n's. 1.

Introduction

The Hadron Calorimeter1 is a critical sub-system of the CMS detector. The Hadron Calorimeter will detect jets, single hadrons and n's. It is required for the discovery of the Higgs between masses of 100 to 1,000 GeV and for searches of Dark Matter. Astronomical measurements of the rotation curves of galaxies as well as the distribution of microwave background of the sky require the existence of Dark Matter. The neutralino predicted by SUSY is a Dark Matter candidate. The CMS experiment will search for Dark Matter by triggering on missing transverse energy events. SUSY searches also require the hadron calorimeter and in one year of data taking at a luminosity of 10 cm" sec" the mass limit of supersymmetric particles will be extended from existing value of 350 GeV to about 2 TeV. Extra Dimensions can also create Dark Matter candidates and again the Hadron Calorimeter is crucial for these searches. The Hadron Calorimeter will be installed in the CMS detector, calibrated and ready to take data on day one of the experiment2. The Hadron Calorimeter (HCAL) Central Barrel (HB) is 9 meters long, one meter thick and 6 meters in outer diameter. The HB has 36 wedges of 20° segments made of brass and scintillator. The two End Caps (HE) are also made of brass and scintillator, with a diameter of 0.8 to 6.0 meters and a thickness of 1.8 meters. Both HB and HE are inside the 4-tesla solenoid coil. The n-<}) segmentation of HB and HE is 0.087 x 0.087, except near n = 3.0, where the size of the segmentation is doubled. The longitudinal segmentation for HB is one unit while for HE, varies from one to *For the CMS HCAL Collaboration

348

349 three. The two forward calorimeters (HF) are constructed to measure forward jets. Since HB is only 6.5 interaction lengths thick at n=0, additional scintillators are placed inside the muon barrel system, outside of the solenoid coil, to measure the HB energy leakage. This system is called HO and both Monte Carlo studies and test beam results show that it improves the energy measurement by reducing the tails of the energy distribution substantially. To correct for the radiation damage at n above 2.0, HE has extra longitudinal segmentation to allow correction for signal loss. The light from scintillators is transported by means of plastic fibers to the Hybrid Photodetectors (HPD's) for HB, HE and HO. The signals for HF are Cherenkov radiation in quartz fibers read by conventional phototubes. The essential electronics elements are QIE's (charge integrator and encoder), HTR's (trigger and readout module) and event builder card (DCC- data concentrator card). The detector is presently in the assembly area of the CMS above ground hall, where final integration is being done. That means the front end photodetectors and electronics are being installed and connected to services, including high voltage for HPD's, low voltage, signal cables, cooling water, nitrogen purges, etc. Eighteen wedges of HB, making a half barrel of the Central Calorimeter, have been assembled and integrated and are now detecting cosmic u's.

2.0 Calibration ofHCAL The calibration of the detector is a complex process using several radioactive wire sources, nitrogen laser at 337 nm, Light Emitting Diode (LED), u's from cosmic rays and test beams as well as charge injection. The energy scale is determined in the CERN test beam using electrons and n's. Final calibration will be accomplished using p-p collision data and will not be discussed in this paper. 2.1 Radioactive wire source Co60 radioactive source implanted at the tips of a wires passes over each scintillator segment and next to several quartz fibers (HF). Even though the signal size is small, by taking millions of readings per scintillator segment, a 3% calibration is achieved. Each sub-system has permanent radioactive wire source units that can be used when the detector is closed. There are several portable units that can be used when the detector is open. This is an absolute calibration that will be used to calibrate each scintillator segment about once a year. Figure 1 shows the readout over a single tile segment. For the final calibration a stronger Co60 source will be used, which gives a signal of about 0.5 fC.

350

2.2 Nitrogen Laser for Calibration and Monitoring A nitrogen laser is a sealed 300 micro Joule unit at X.=337 nm, built by Spectra Physics, Inc. The laser system includes two neutral density filters wheels, a quartz lens focusing system, a commutator with 20 output paths and 3 pin diodes to monitor laser light, both before and after the filters. The system can be pulsed 10 times a second and will be used for both calibration and monitoring the calorimeter system. In addition the laser pulse is used to time the various components of the detector that mimic an actual event that occurs at the center of the CMS detector. The whole calorimeter can be timed before LHC start. The laser intensity is controlled by neutral density filters and the signal can be sent to various paths using a commutator. The various paths include excitation of one scintillator segment in each calorimeter tower, lighting up each pixel of every HPD and measuring radiation damage. The laser is also used for extrapolating the energy calibration from 250 GeV measured in the test beam to 3 TeV. The laser also simulates Cherenkov light in the hadron Forward Calorimeters (HF). For the laser signal to reach scintillator tiles or HPD pixels, various splitters were designed and constructed. The splitter design uses one large diameter quartz fiber butted to several smaller diameter quartz fibers. Each side is polished and attached to each other by simple pressure. For example to split a signal six ways, the large diameter fiber is 800 microns, while the smaller fibers are 250 microns. This form of splitting makes the light in each smaller fiber uniform within 10% of the average. The fiber splitters constructed are 1->18, 1->16, l->6 and l->4. All laser signal fibers are made of quartz with quartz cladding to reduce radiation damage. 2.3 Light Emitting Diode (LED) In addition of the laser lighting each pixel of every HPD as well as each photomultiplier tube in HF, fast LED's also illuminate the pixels and PMT's. For the HB, HE and HO green LED have been used that have a color spectrum that is very similar to the green light of Kuraray Y l l fibers used inside the calorimeter. For HF, blue LED's are being used. LED's have the advantage that they can be pulsed over 100 times a second with a time jitter of about one nano second. In contrast the laser has a time jitter of about 25 nsec, but all segments of the detector get the laser signal at the same time. Each LED has its own pulser and the signal cable length to each pulser varies from unit to unit.

351

700

BOO

900 CM

Figure 1. Readout of radioactive (Co60) source passing over a scintillating tile.

Figure 2. On line display of a cosmic muon through the hadron calorimeter.

2.4 Pin Diodes The laser signal and LED light are monitored by several pin diodes. There are also pin diodes inside each Calibration Module on each 20° wedge. Pin diodes have very high quantum efficiency and a gain of one and are very stable. The laser light output varies from pulse to pulse with a sigma of about 10%. To obtain calibration constants of a few percent, the laser has to be normalized. 3.0 Test Beam Two 20° HB wedges, a 40° segment of HE, and a 40° portion of HO have been permanently installed in the H2 test beam at CERN on a rotating table. The system also has 7 x 7 ECAL module and several wedges of HF. The beam included e's up to 150 GeV, u's and n's from 3GeV to 250 GeV. The resolution, shower shape and ECAL versus HCAL responses have been

352

measured at various energies and compare well with Geant4 Monte Carlo predictions. In 2006 the plan is to measure a full complement with an ECAL production module. 4.0 The Cosmic Muon Challenge A half barrel of HB has been installed with electronics, services and front end trigger. The detector is in a free trigger mode detecting cosmic muons. Figure 2 shows an example of a cosmic muon traversing the whole HCAL detector. This is a challenge, as it requires all the various components to work together. Next, the hadron calorimeter barrel will be inserted inside the 4 tesla magnet. We plan to detect cosmic muons during the magnet test, in early 2006. Acknowledgments The CMS Hadron Calorimeter has been designed, constructed and assembled by physicists and engineers from several tens of institutions from about 10 countries. It is a monumental task that has been ongoing for 10 years. The author acknowledges the hard work of all the people involved in this task. Special thanks to the management of this task. We will be ready to take data on day one. References 1. The Hadron Calorimeter Technical Design Report CERN/LHC 97-31 and P. de Barbara and V. Hagopian, Proceeding of the 8' International Conference on Astroparticle, Particle and Space Physics, Detectors and Medical Physics Applications, ICATPP-8, ed. Barone, Borchi, Leroy, Rancoita, Riboni and Ruchti; p. 186 (World Scientific, Singapore, 2004). 2. The CMS HCAL Collaboration. Design, Performance and Calibration of CMS Hadron-Barrel Calorimeter Wedges, submitted to Nucl. Instrum. and Methods (2005).

PROPERTIES OF A 256-CHANNEL PROTOTYPE OF THE ALICE/PHOS CALORIMETER* MIKHAIL IPPOLITOV, ANREY VASILIEV FOR THE ALICE/PHOS COLLABORATION Russian Research Center "Kurchatov Institute",Kurchatov sq.l, Moscow, 123182, Russia

We discuss in this paper the measured characteristics of a 256 - channel prototype of the ALICE/PHOS lead-tungstate electromagnetic calorimeter. The prototype has been tested with electron and charged pion secondary beams delivered by the CERN PS and SPS accelerators. Energy, time of flight and two-photon invariant mass resolutions are presented. Data were taken at electron and pion momenta from 0.6 GeV/c up to 150 GeV/c. An overview on the characteristics of the ALICE mass production crystals are presented.

1.

Introduction

ALICE (A Large Ion Collider Experiment) [1] is one of the four experiments under construction at the CERN Large Hadron Collider (LHC). This experiment is optimized for the study of heavy-ion collisions at a centre-of-mass energy -5.5 TeV. At these conditions the creation of a new state of matter, called quark-gluon plasma (QGP), in which quarks and gluons behave almost as a free particles, is expected. The PHOS (PHOton Spectrometer) [2], is a high resolution electromagnetic calorimeter consisting of 17920 detection channels of lead-tungstate crystals, PbW0 4 (or PWO). It will be positioned on the bottom of ALICE set-up at a distance 460 cm from the interaction point and will cover the pseudorapidity range -0.12 < r\ < 0.12 and 100° in azimuthal angle. PHOS is subdivided into five independent modules. Each module is segmented into 3584 detection channels. The detection channel consist of 2.2x2.2x18 cm3 PWO crystal, coupled to 5x5 mm2 Avalanch Photo-Diode (APD). The PHOS is dedicated to the search for electromagnetic signals from the QGP in nucleus-nucleus interactions at high energies. Approving PWO crystals as scintillating material for a large number of highenergy physics experiments has stimulated the research on this material and the optimization of its fabrication at industrial scale. The Russian Research Center

This work was supported by Russian Federal Atomic Energy Agency.

353

354

Kurchatov Institute together with North Crystals Company performed a R&D programm on development of the technology of mass production of high quality lead-tungstate detectors. In this paper we are presenting beam-tests results of the 256-channel PHOS prototype in beam momentum range 0.6 -150 GeV/c. The tests were performed at the CERN PS and SPS accelerators in years 2002-2004. Also, an overview on the characteristics of the ALICE mass production crystals are presented. 2.

Crystals

PWO was chosen for the PHOS detector due to its basic properties. This scintillating crystal has low value of radiation length, 0.9 cm, low value of Moliere radius, 2.2 cm, and it is fast. Its emission spectrum has a good correspondence to the spectral sensitivity curves of the most commonly used photo detectors. As a main crystal production facility for the ALICE PHOS project, North Crystals Company, Apatity, Murmansk region, Russia has been chosen. Crystals have been grown by Czochralski method in a resistant ovens. About 12000 ALICE/PHOS crystals were produced in years 2001-2005 and delivered for testing at the Kurchatov Institute. Light yield (LY), optical transmission (OT), emission and scintillation decay time spectra were measured. As an example, in Figure 1 the distributions of LY and OT at 420 nm of the crystals are shown. Double structure in LY distribution is due-to

H-4

1800

i

1600 1400

J

t ' i i11M

hi Nent - 11759 Mean = 10.89 RMS

s 1.565

]

1200 1000 800

600 400 200

mIt"f-i m h f-j \ mr

H H

ji

L,L, I 35 40 45 50 55 60 65 70 75 80 transmission, %

Figure l. On the left: distribution of the LY; on the right-distribution of the longitudinal optical transmission at wavelength 420 nm (maximum of emission spectrum) for all delivered crystals.

improvements in production technology in the middle of year 2003. Only crystals which fulfilled ALICE crystal's requirements (LY >8 photoelectrons/MeV, nonuniformity of LY <5%, longitudinal optical transmission (OT) >25%, 55% and 65% at the wavelengths of 360nm, 420nm and 620nm, 96% of light are collected within 300ns gate) were accepted.

355

3.

Prototype description

A single channel of the prototype consists of a crystal, an APD and a low noise charge sensitive preamplifier. HAMAMATSU S8148 APD of 5x5 mm2 sensitive area were used as readout. A row of 8 crystal detector units constitutes the basic assembly of a crystal strip unit. The detectors were stacked to form 16 x 16 array. The LY of PWO crystals is relatively low and depends on temperature. The temperature coefficient is ~-2%/°C in the temperature range between +20 °C and -30 °C [2]. The temperature dependence gives the possibility to increase the light yield at low temperatures, and -25 °C is chosen as the working temperature for the PHOS detector. The prototype was cooled down to -25°C with cooling system providing long-temperature stability within 0.1 °C. Monitoring of the detection channels was carried out by Light Emitting Diodes (LED) monitoring system [2], More detailed description of the prototype one can find in [3]. 4.

Beam-test Results

The beam tests were performed with electron, ri and 7t+ secondary beams at the CERN PS and SPS synchrotrons in the energy range 0.6-150 GeV. The beam momentum spread was less than 1%. The typical beam intensity was ~2-5xl0 3 particles/spill during calibration runs and 2xl0 4 n /spill during neutral mesons measuring runs. 4.1. Calibration, Energy Resolution and Linearity A set of thin scintillator beam counters and a gas Cherenkov counter were used to define beam particle type. The calorimeter was calibrated with 2 GeV/c and 10 GeV/C electrons during PS and SPS tests respectively. During calibration beam was defined by a thin 10 x 10 cm2 area scintillator counter. After the calibration, the beam was centered on several chosen detectors in the matrix and the energy scan was performed. In this case the beam was defined by a small lxl cm2 area detector placed in the front of the prototype. The energy resolution was determined from a Gaussian fit to the electron peak obtained by summation of the gain corrected signals in the crystals of the 3x3 matrix around a given central detector. The experimental data are well fitted by the usual function:

£.= « © A e c E

4E

(i)

E

where a=3.5%, b=13 MeV and c=l.l% are stochastic, electronics noise equivalent and constant terms, respectively. Response of the PbW0 4 matrix in the energy range of 0.6-150 GeV is linear with a precision better than 1%.

356

4.2. Time-of-FUght Resolution For the time-of-flight resolution measurements START signal was taken from external trigger detectors with intrinsic TOF resolution -150 ps, STOP signals were taken from detectors of the PHOS prototype. The time resolution as function of energy deposited by electromagnetic shower in a single channel, is shown in Figure 2. The resolution is better than 0.5 ns was observed for deposited energies higher than 1 GeV.

5

e

Energy, GeV Figure 2. Time resolution as a function of deposited energy in a single channel of the calorimeter.

43. Two - Photon Invariant Mass Resolution The experimental setup for two-photon invariant mass measurements is shown in Figure 3. Inclusive TI" + 12C -»7i°+X reaction was used. A large area VETO scintillator counter was set to reject charged particles in the detector acceptance. A n beam with 6 GeV/c momentum was used and the calorimeter was set at a distance of 203 cm from 10% of interaction length carbon target. As an example, in Figure 4, invariant mass spectra of all photon pairs from the same event are shown. Clean peak around the n" mass is observed. This peak was fitted with Gaussian curve and position of the peak at 135.6 ± 0.2 MeV/c2 with invariant mass resolution of 4.7 ± 0.2 MeV/c2 was obtained (see Figure 4).

-He-^ Target

From PS and SPS gamma2 PiO-trigger: S L x S2 x Veto

Figure 3. Experimental setup for the measurements of the two photon invariant mass resolution of the calorimeter.

357

| Invariant mass vs E1+E2 | Chi2(ndf = 13.arM7 Constant = 41.57 ±2.439

£0 5GeV<E1+E2<6GeV

Mean

= 135.S± 0.241

Sigma =170.1 ±0.211

40

30

20

10 , l.-^ll.,ft Jll.
20

40

60

Lull ml i-mlnJl i 80 100 120 140 160 180 200 220 240

Mn, MeVfc"2 Figure 4. Two - photon invariant mass distribution from 6 GeV/c 7t" + 12C ->7t°+X reaction. Line is results of the fit by Gaussian. Only events with two photons with energies between 5 GeV and 6 GeV are taken into cionsideration.

5.

Conclusions

A test of the 256-channel prototype of the PHOS detector has been performed. An energy resolution, ranging from -3.9% at 1 GeV to - 1 . 1 % at 150 GeV, has been found. Time resolution better than 0.5 ns was found for deposited energies in crystal higher than 1 GeV. 7t° mesons have been measured via its two - photon decay channel. Invariant mass resolution of 4.7 MeV/c2 was found. An overview of the ALICE/PHOS crystals properties was presented. Acknowledgments The author wishes to thank all the members of the ALICE-PHOS group for their help during test and preparation of the prototype. We also wish to express our gratitude to the CERN accelerator division for excellent performance of the PS and SPS complex. References 1. ALICE Collaboration, Technical Proposal, CERN/LHCC/95-71 (1995) 2. ALICE Collaboration, Technical Design report of the Photon Spectrometer (PHOS), CERN/LHCC 99-4, ALICE TDR 2, 5 March 1999 3. M.Ippolitov, A.Vasiliev in Proc. Of the 11 Int. Conf. on Calorimetry in Particle Physics,Perugia, Italy, 2004, World Scientific, p. 103

S O U R C E CALIBRATION OF CMS QUARTZ F I B E R FORWARD CALORIMETER

N. AKCHURIN, M. SPEZZIGA Texas Tech University, Department of Physics, Lubbock, U.S.A. V. GAVRILOV, A. KROKHOTIN, S. SEMENOV, A. ULYANOV, E. VLASSOV Institute for Theoretical and Experimental Physics, Moscow, Russia A. ERSHOV Skobeltsyn Institute for Nuclear Physics, Moscow, Russia S. PAKTINAT Institute for Studies in Theoretical Physics and Mathematics,

Tehran, Iran

We show three different methods for the calibration of the CMS forward calorimeter (HF) using a radioactive source (Co60) and compare them against the calibration obtained with high-energy electrons in the test beam. About 5% calibration precision (rms) was obtained for the five HF wedges tested.

1. Introduction The Hadron Forward (HF) calorimeter is a part of the CMS detector which will be installed at the LHC collider at CERN. HF covers pseudorapidity range 3.0 < |r?| < 5.0. It should provide the detection of jets at the very small angles and improve the missing transverse energy measurement. HF is made from steel absorber with embedded quartz fibers [1,2]. The signal is registered by Cherenkov light produced by charged particles in quarts fibers. The choice of the design is driven by the requirement to construct the detector able to operate in the extremely hard radiation conditions (up to lOOMRad/year). While known physics channels could be used for HF calibration during the running, an opportunity to calibrate HF before starting operation is very important. In this paper, we discuss the calibration procedures using the radioactive Co60 sources.

358

359 2. Detector Description Each HF is assembled from 18 identical wedges. HF wedge is constructed from steel absorber with embedded quartz fibers running parallel to the beam axis. There are two types of fibers: long (L), running the entire lengths of the absorber and short (S). L and S fibers are separately grouped into the bundles to form L and S readout cells. Through light guides bundles are connected to PMTs, located at the rear side of the wedge. Every wedge contains 31 source tubes which are inserted in place of either L or S fibers. Each cell contains either one or two source tubes. 3. Calibration Setup 5 mCi Co60 source is used for the calibration. PMT signals are digitized every 25 ns time slice by front end electronics with QIE digitizer measuring charge in non linear scale with minimal step size equal 2.6fC. Data are automatically sampled in histograms and readout with frequency 20 Hz. Fig. 1 shows typical QIE distributions for L and S readout cells in terms of linearized QIE counts. Solid (dashed) line corresponds to the case when the source is inside (outside) the wedge. Both distributions are obtained for the same time interval and hence are normalized to the same number of entries.

Figure 1. Typical QIE distribution for L (left plot) and S (right plot) cells; solid (upper) line - source is inside the absorber; dashed (lower) line -source is outside the wedge.

From Fig. 1, it can be seen that the number of noise entries from the pedestal peak is about a few orders more than signal entries. Moreover, the gammas from the source are emitted randomly so the signal from single photoelectron (SPE) collected in QIE could be shared between two consecutive time slices and give two entries in histogram. During the test beam five wedges were exposed to lOOGeV electron

360

beam. Source data are also available for these wedges. Combined analysis of the source and test beam data allow to make an absolute normalization of the calibration done with source. 4. Calibration Procedure and Results Several methods of source data analysis were tested. They differ in the way how to extract charge related to the source activity from the data. 4.1. "Mean Charge"

Method

Charge (per sec) is calculated for the signal and background histograms: r\{S/B),~.T„i

v

40MHz

32

r-^

where i - the source tube number, qj - charge value expressed in terms of linearized QIE counts and rij - the number of entries in j t h bin. The value of Qf - Qf gives charge Qi related to the source activity. Calibration coefficient
-

U(^Ci)e(GeV)Ri Qi{QIEjs)

(2)

where U(mCi) - source activity, e(GeV) is the average signal as it would be collected by large size cell (no leakage from cell borders) for 1 mCi source and Ri - geometrical scale factor. Ri show what part of the signal is collected by fibers of the ith cell relatively to the cell of infinite size. They depend on a distance from source tubes to the cell boundaries and on the source tube position relatively to readout fibers. For each calorimeter cell a single calibration coefficient Cfource should be derived. It's equal to corresponding g if the cell contains only one source tube and mean value of two g if there are two tubes. To calculate Cj, the only unknown parameter e in Eq. 2 should be determined. It can be fixed from the requirement < c?°urce/C$eam >= 1, where average is calculated over all cells of tested wedges and C\eam calibration coefficients derived with 100 GeV electron beam [3]. RMS of the < dource IC\eam > distribution defines the precision of the calibration method. It's 7.0% (12.4%) for L (S) cells. The precision of this method is limited by pedestal stability. A small fluctuation in the pedestal position may strongly affects calculated mean charge because both the signal and background histograms consist mostly from pedestal entries.

361 4.2. "Fixed QIE Bin Range"

Method

In order to reduce the influence of the pedestal stability on the calibration precision, charge can be calculated in some predefined QIE bin range. The calibration procedure follows to the method described in Sec. 4.1, but in Eq. 1 charge is calculated in the range of 10-30 bins. The precision of the method (as defined in Sec. 4.1) is 5.4% (8.2%) for L (S) cells. Due to possible sharing of signal from SPE between two consecutive time slices the signal is distributed from maximum value to zero. Charge calculated in a fixed bin range is a part of the total charge which is different for different towers. This limits precision of the method. 4.3. "Signal Extrapolation

Under the Pedestal"

Method

Fig. 2 (solid line) shows the results of subtraction of the signal and background histograms. It gives the pure source activity at the region of large QIE values, while it's not the case for small QIE values where the shape of the histogram is determined by subtraction of pedestal entries. Large negative peak in the small QIE region results from the normalization of the signal and background histograms to the same number of entries. To estimate the part of the signal under the pedestal peak, an extrapolation should be applied. In this paper, the simplest constant extrapolation is used (Fig. 2, dashed line).

1

• subtraction extrapolation

Linearized QIE Counts

Figure 2. Solid line - QIE distribution after subtraction of signal and background histograms. Dashed line - the result of the extrapolation (see text for details).

Using the extrapolated histogram a charge related to the source is calculated as: _

AQMHz n

Ej=i i

32

^

j=iP+i

(3)

362

where rij - the number of entries in j bin of histogram, P - pedestal position, ip - bin containing pedestal value P. Then the procedure described in Sec. 4.1 is applied starting from Eq. 2. This method combines the advantages of the first two methods: the signal charge under the pedestal is taken into account and the collected charge is not strongly affected by the pedestal position fluctuations. The precision of this method is 4.9% (5.9%) for L (S) cells (Fig.4.3) It is limited by uncertainty of the extrapolation to the region of small QIE values.

13 z

z

16Hr 12r 10

3.2 0.4 0.6 0.8

Entries 108 Mean 1.006 RMS 0.05861

Entries 110 Mean 0.9947 RMS 0.04908

1

1.2

1.4 1.6 1.8

'"'Source'^-'Beam

Figure 3. < Cfource/C^am

1

r-

8.2

• t . . . i • • inl li "i .i r• . i• .•

• • i

0.4 0.6 0.8

1

Jflu. . • ! . .

• ""l

• l • i

1.2 1.4 1.6

. 1

1.8

^ Source'^' Beam

>, left (right) plot is for L (S) cells.

5. Conclusion An opportunity to use source calibration for HF initial calibration is studied. The developed methods allow to achieve calibration precision about 5%, where the precision of calibration is determined relatively to calibration derived from 100 GeV electrons. This value is comparable with intrinsic lateral nonuniformity of the calorimeter response. References 1. N. Akchurin et al., Nucl. Inst, and Meth. A Vol. 399, pp. 202-226 (1997). 2. V.B.Gavrilov and M.V.Danilov, Physics of Atomic Nuclei, Vol. 67, No. 7, pp. 1390-1397 3. A.Ulyanov, "Test Beam Results of the CMS Quartz Fiber Calorimeter", talk given at "9th ICATPP Conference".

T H E CALIBRATION S T R A T E G Y OF T H E CMS ELECTROMAGNETIC CALORIMETER*

P. MERIDIANI INFN Romal Calibration is one of the main factors that set a limit on the ultimate performance of the CMS electromagnetic calorimeter (ECAL) at LHC. In situ calibration with physics events will be the main tool to reduce the constant term of the energy resolution to the design goal of 0.5%. In the following the calibration strategy will be described in detail.

1. The CMS Electromagnetic calorimeter global calibration strategy The physics reach of the CMS electromagnetic calorimeter (ECAL) 1, in particular the discovery potential for a low mass Standard Model Higgs in the two photons decay channel, depends on its excellent energy resolution. Crystal inter-calibration precision goes directly into the energy resolution constant term with very little scaling, due to the fact that most of the energy of an electromagnetic cluster is contained within a single crystal (around 80%); this means that crystals' calibration will represent a key aspect in order to reach the design performance of the electromagnetic calorimeter. Inter-calibration, however, it is not the only calibration task, since calibration should be naturally seen as composed of a global component, giving the absolute energy scale, and the inter-calibration component. Schematically the reconstructed energy might be decomposed to three factors: Eei~f = G x F X ^ C J x A,

(1)

i

where G is a global absolute scale. depending on the type of particle, clustering algorithm used: this is containment of the shower or the

The function F is a correction function its position, its momentum and on the necessary to correct either the partial energy loss due to bremsstrahlung or

* On behalf of the CMS Collaboration

363

364

conversions in the material in front of the calorimeter. The c, factors are the inter-calibration coefficients while the A, are the signal amplitudes, in ADC counts, which are summed over the clustered crystals. 2. Inter-calibration at Startup The natural spread of the channel-to-channel response variation in the calorimeter barrel is the crystal-to-crystal variation of the scintillation light yield which has an RMS of «8%. In the endcap the VPT signal yield, the product of the gain, quantum efficiency and photo-cathode area, has an RMS variation of almost 25%. The tight CMS production and assembly schedule will allow inter-calibration only of a limited number of crystals (few super-modules) with electron beams of known energies; these crystals will have an initial inter-calibration precision better than 1%. For all the other crystals there will be two other sources to compute initial value of the inter-calibration coefficient: • Laboratory measurements • Cosmic events During the assembly phase, all the detector components are characterized 2 ' 3 . The crystal light yield (LY) is measured in the laboratory with a photo-multiplier tube, exciting the crystal with a 60 Co source. An independent LY measurement can be obtained exploiting the correlation between the crystal LY and the Longitudinal Transmission at 360 nm 5 . The precision of the inter-calibration which can be reached using the average of the direct light yield measurement and the light yield predicted from the longitudinal transmission is 4.2% and has been evaluated with the 2004 ECAL test beam data; using only the direct light yield this precision is 4.6%. After the electronics integration, all the super-modules will be commissioned using cosmic events for about one week. Aligned cosmic rays can be selected by vetoing on signals above a rather low threshold in the adjacent crystal with highest signal, avoiding the need for external tracking. Inclining the super-module by ~10° (in order to have a rather uniform number of cosmic events along the super-module length) an overall precision of 2-3% is achievable for all channels. 3. In-situ calibration Since the startup inter-calibration precision will be around 2-3%, in-situ calibration techniques will be the fundamental tool to reach the 0.5% goal precision. Over the period of time in which the physics events used to

365

provide an inter-calibration are accumulated, the response must remain stable and constant to high precision. For example, the changes in crystal transparency are tracked and corrected using the laser monitoring system 6 . Several methods have been proposed to obtain inter-calibration factors. At the start-up, the 0-independence of the energy deposited in the calorimeter can be used to rapidly to inter-calibrate regions at the same pseudo-rapidity (T/). Once the tracker will be fully functional, the abundant electrons, mainly from W —> ev, can be used for local inter-calibration in region with uniform tracker material exploiting the E/p peak. A complementary method, not relying on tracker momentum measurement, is based on 7T° —> 77 and rp —» 77 mass reconstruction. The Z mass constraint in Z —> e+e~ can be used instead for region intercalibration, for example as a complement of the 77 ring inter-calibration at the start-up, or as invaluable tool to tune the electron reconstruction algorithms and to fix the energy scale. A similar use can be made of Z —> /U/U7, where the photon is coming from inner bremsstrahlung, constraining the 7 energy using the di-muon system. 3.1. <j> Symmetry Inter-calibration can be obtained by comparing the total energy deposited in each crystal with the mean of the distribution of total energies for all crystals at that pseudo-rapidity, assuming that the energy deposit should be uniform in
Using Single

Electrons

Once the tracker is fully operational and well aligned, inter-calibration of different crystals can be obtained using the momentum measurement of isolated electrons. The main difficulty in using electrons for inter-calibration is that they radiate in the tracker material in front of the ECAL, and both the energy and the momentum measurement are affected. Moreover the

366

• Precision with 11 million events A Limit on precision

c .5 O 5 2 Q. C

o

1

i

1 2

^

pjdLu Fiducial Region —

It • tu

Precision with 11 million events

* Limit on precision

4

1 3

3

9W

M •

0.2

0.4

0.6

0.8

/

2.2

2.4

2.6

2.8

3 1

Figure 1. Inter-calibration precision which can be obtained with 11 million Level-1 jet trigger events and the limit on the inter-calibration precision due to tracker material inhomogeneity as a function of 77.

average amount of bremsstrahlung varies with tracker material thickness 9 . The single crystal inter-calibration is obtained minimizing the difference between the energy and momentum measurements. The event selection is based on variables which are sensitive to the amount of bremsstrahlung emission, and consequently measure the quality of the energy and momentum reconstruction. Due to the variation of the average value of E / p peak with 77, caused by the variation of the amount of material in front of the ECAL, inter-calibration is obtained locally in region over which the average value of the E/p is rather constant. In a second step the regions are intercalibrated with each other. The calibration precision achievable is strongly dependent on the available statistics, as shown in Figure 2; the fit function represents the sum of a statistical and a constant term.

3.3. Inter-calibration

using n0 and rf

The possibility of inter-calibrating the ECAL using mass reconstruction of 7T°, and rj° -+ 77 is also being investigated. It was found that about 1.4% LI triggered events have a usable TT° in the barrel and that almost all of these are tagged by the isolated electron Level-1 trigger information. A 7r° mass peak can be well identified in all the ECAL barrel with relatively little background, the mass resolution being about 8%. If we assume a Level-1 trigger rate of 25 kHz, and that 1.4% of these triggers have usable 7r°s, then about 100 7r°s per crystal can be obtained in less than 5 hours running.

367

(%) c o

s

3

"2 .5

0.

c o

;\|\v

vL

P

.0

TB'.5 nr - p ~ \ f r r O 1 '•

i

:

|

0 10

\

H[«r

,

I!! in

I 1 1

iv

2

0.5

1 !!

! i

;

! El li! r

IN

ttt

jj| 5fb4 III

T _

T^lfe 102 10s HLT events per crystal

Figure 2. Calibration precision versus the number of events per crystal for different 77 regions. Upper curve: the last 10 crystals in the ECAL Barrel, Middle curve: 10 crystals in the middle of ECAL Barrel, Lower curve : the first 15 crystals in the ECAL Barrel. The third point along each line gives the precision for 5 f b - 1 of integrated luminosity.

3.4. Use of Z decays 3.4.1. £ - > e + e A number of different uses for Z —> e + e~ decays are envisaged, from the tuning of the electron reconstruction algorithms (for example to obtain or verify algorithmic corrections, F, in Equation 1), to the inter-calibration of ECAL regions. An iterative calibration method has been tested taking as calorimeter regions rings of crystals (at fixed rj) in the ECAL barrel: there are 170 such rings of 360 crystals. Z decays with low-radiating electron should be selected for this purpose. The calibration precision as a function of the number of events per ring is shown in Figure 3. The x-axis is the average number of events per ring, and the plotted point before the highest, at an average of 370 events per ring, corresponds to an integrated luminosity of 1.0 fb- 1 . 3.4.2. Z -> wy Radiative decays Z —• (j,fj,j can be used in a similar way to the Z —* e+e~; the photon energy is here constrained from the di-muon system and the Z mass. The available rates have been estimated using a sample of events generated using a full matrix element calculation of radiative decays. A cross-section of 65 pb over the region \r]\ < 2.5 has been obtained, selecting events with a loose cut on the di-muon mass and searching for a photon with pT >15 GeV within a radius of AR < 0.8 of either muon. Thus for

368 «0.02 B

S?

0.018

b.ooi4!?t*oS2

0.014 U.UI..

: : :

r-+-

+

um

0

1 X)

150

200

250

300

350

400

450

Htm SamOfWdgbli par riaf

Figure 3. Ring to ring calibration precision versus average number of events per ring; the fit function represents the sum of statistical and constant term. an integrated luminosity of 1 fb * an average of nearly 1 such p h o t o n per crystal will be collected. 4.

Acknowledgments

T h e a u t h o r would like to t h a n k the entire CMS E C A L e / 7 community and the organizers of 9th ICATPP conference. References 1. CMS Collaboration, "The CMS Electromagnetic Calorimeter Technical Design Report", CERN/LHCC 97-33, 1997. 2. E. Auffray et al, "Performance of ACCOS, an Automatic Crystal Quality Control System for PWO crystals of the CMS calorimeter", Nucl. Inst, and Meth. A 456 (2001) 325-341 3. S. Baccaro et al, "An automatic device for the quality control of large-scale crystal's production", Nucl. Inst, and Meth. A 459 (2001) 278-284 4. S. Baccaro et al, "Precise determination of the light yield of scintillating crystals", Nucl. Inst, and Meth. A 385 (1997) 69 5. L.M. Barone et al., "Correlation between Light Yield and Longitudianl Transmission in PbWC"4 crystals and impact on the precision of the crystal intercalibration", CMS Rapid Note 2004/005, article in preparation 6. A. Bornheim, these proceedings 7. D. Futyan, C. Seez, "Inter-calibration of ECAL crystals in <j> Using Symmetry of Energy Deposition", CMS Note 2002/031 8. D. Futyan, "Inter-calibration of the CMS Electromagnetic Calorimeter Using Jet Trigger Events", CMS NOTE-2004/007 9. CMS Collaboration, "Data Acquisition & High-Level Trigger Technical Design Report", CERN/LHCC 2002-26, 2002.

T H E VERY F R O N T - E N D C A R D S FOR T H E CMS E L E C T R O M A G N E T I C CALORIMETER: D E S C R I P T I O N , CALIBRATION A N D P E R F O R M A N C E

M. M. OBERTINO ON BEHALF OF THE CMS ECAL CALORIMETER GROUP. I.N.F.N. Via Giuria 1, 10125 Torino, Italy E-mail: [email protected]

The CMS electromagnetic calorimeter is made of about 76000 PbW04 crystals. It has been designed to measure energy and location of electromagnetic showers with high resolution and over a 16 bit dynamic range. Part of the read-out chain (the Front End Electronics) will be installed within the detector in a high radiation environment and inside a 4T magnetic field. The heart of the system is the Very Front End card (VFE), the board where the signal coming from the photodetectors is amplified and converted into digital form. After a brief description of the Front End Electronics, the results of the tests performed during the production, including the calibration procedure will be presented.

1. Introduction The CMS electromagnetic calorimeter (ECAL) [1] is a challenging detector designed to perform high precision measurements in the very demanding environment of the CERN Large Hadron Collider (LHC). Design philosophy was strongly driven by the requirement of excellent energy and angular resolution for the detection of Higgs boson through its two photon decay. The large proton-proton inelastic interaction rate (10 9 s _ 1 ), the 25 ns bunch crossing interval, the high value of the center of mass energy at LHC (14 TeV) and the calorimeter location inside a 4T magnetic field set the other constraints: low noise, wide dynamic range (50 MeV to 1.5 TeV), high processing speed of detector signals, radiation hardness and insensitivity to the magnetic field. To fulfill all these requirements it was decided to build a homogeneus calorimeter made of lead tungstate crystals. The detector is organized as one barrel, covering the pseudorapidity region \i]\ < 1.48, and two endcaps extending the acceptance to \r}\ < 3. The barrel is subdivided into

369

370

36 supermodules, each composed of 85(7j)x20(^) crystals; the endcaps are made of 14648 crystals arranged in 4 Dees. The PbWC>4 was chosen as active medium because of its excellent radiation hardness, short decay time (80% of the light collected in 25 ns) and the small value of radiation length (0.89 cm) and Moliere radius (2.2 cm), allowing for compact design and fine granularity. The disadvantages of the lead tungstate are the low light yield ( 50 ph/MeV) and the strong temperature coefficient (-2%/°C). The first effect is compensated by use of photodetectors with internal gain (APDs in the barrel and VPTs in the endcaps), the second by keeping the temperature inside the calorimeter stable within 0.1 °C. The ECAL electronics is organized in two logical layers, one positioned inside the detector (the Front End System), immediately behind the crystals, the other in the counting-room. The connection between the two parts is made by optical fibers. The building block of the Front End Electronics is the trigger tower (TT) designed to read out 25 crystals. It consists of a Mother Board (MB), 5 Very Front End Cards (VFE), a Low Voltage Regulator Board (LVB) and a Front End Card (FE). The MBs are passive devices which provide the flat surface where the VFEs and the LVB are mounted. They distribute the regulated analog and digital voltage provided by LVB to the VFEs. In the barrel they also filter and distribute the high voltage to the APDs and drive the signals to the electronic chain; the connection with the photodetectors is made by kapton cables. In the endcaps an HV board is interposed between the MB and the VPTs. The Very Front End Card, which is described in more detail below, shapes, amplifies and digitizes the photodetector signals. On top of the trigger tower, the FE board provides the sampling clock to the 5 VFEs, generates the Trigger Primitives (TT transverse energy and bunch crossing assignment) and stores the raw data while waiting for the level 1 trigger decision.

2. The Very Front End card As shown in Fig 1 each VFE card consists of 5 identical channels, each consisting of a multi gain amplifier (MGPA), a 12-bit 40MS/s 4-channels ADC (AD41240) and a buffer. The MGPA has 3 different gains: 1, ~6 and ~12. Their number and ratio are adequate for both barrel and endcaps. It consists of a low noise

371

Figure 1. VFE card.

preamplifier buffered by source followers to 3 gain channels in parallel. The total noise is ~ 8000 (gain 6 and 12) or ~28000 (gainl) e~~ for the barrel and ~3500 e~ for the endcaps. Each gain channel is connected to one of the ADCs which are inside the AD41240. A digital logic following the analog to digital conversion stage automatically selects as output the signal of the highest, not saturated gain. The multiple gain approach allows to achieve the requirement of a 90 db dynamic range without using radiation hard 16 bit ADCs. The buffer at the end of the chain adapts the LVDS signal coming from the ADC to the single ended inputs of the FE. A Detector Control Unit (DCU) measures the APD leakage current and the crystal temperature. All the chips mounted on the Very Front End Cards are manufactured in a 0.25 yum CMOS process which guarantees radiation hardness and low power consumption.

3. V F E production and test The VFE production (12880 cards for ECAL barrel and 3000 for the endcap) was started in summer 2004 and will be finished by the first quarter of 2006. Each board is characterized and calibrated before being mounted in the supermodule, following three steps: power-on test, burn-in and calibration. 3.1. The power-on

teat

The power-on is the first electrical test of the VFE cards. It is performed by the manufacturer and repeated at CERN. The test system powers the

372

VFE and measures the total power consumption, pedestal and noise. The internal MGPA test pulse is used to inject a charge into each channel and perform a fast pulse shape analysis. The correct behaviour of the DCU is also verified. Up to now 8880 cards were tested at CERN and only 79 failed.

3.2. The burn-in

test

In order to eliminate infant mortality failures, the cards which have passed the power-on are kept for 3 days at 60 °C, powered and clocked while the current is constantly monitored. So far all the cards survived this test.

3.3. The

calibration

The calibration test system measures the gains of each channel sending a charge pulse between 100 fC and 50 pC and reading the corresponding ADC response. By fitting the output signal peak versus injected charge, a slope is obtained as measurement of the gain (in ADC counts/pC). The distribution of gain ratio (G6/G1 and G12/G1) for the first 2320 VFE tested this year is reported in Figure 2a and shows a good homogeneity of the VFE performance (a/mean ~ 1%). Since more than one year is required to test all the boards the calibration system stability is a strong requirement. The dispersion of the gain ratio was measured over a period of 6 months and found less than 1%. The linearity of ADC and MGPA was measured during the chip production and found within specification (see [2] and [3] ). An indication of the linearity of the VFE card can be extracted looking at the residuals calculated with respect to the calibration line. Their average (estimated for the 2320 cards mentioned before) was found within 2 ADC count for all charges and gains (see Fig. 2b) indicating that the linearity is satisfactory over the entire range. Pedestal and RMS noise of each VFE channel are measured with the calibration system and after the installation in the supermodules. The measured noise for gain 12 was found 1.1 ADC counts corresponding to about 40 MeV, fully complying with the requirement and in good agreement with the results obtained from the test beam of the same supermodule [4]. The measured noise for gain 1 and 6 are respectively 0.54 and 0.73 ADC counts.

373

(a)

(b)

Figure 2. (a) Gain ratios G12/G1 and G6/G1 distribution for 2320 VFE. (b) Mean value of the residuals calculated respect to the calibration line versus injected charge for gain l(top), 6(middle) and 12(bottom). 4.

Conclusion

The E C A L Very Front E n d cards are under construction. Each board is tested and calibrated before the installation in the supermodule. T h e test results show t h a t the design goals have been achieved. References 1. CMS collaboration,"The Electromagnetic Calorimeter Project Technical Design report" CERN/LHCC 97-33, CMS TDR 4, 15 Dec. 1997. 2. M. Raymond et al.,"The M G P A Electromagnetic Calorimeter readout Chip fro CMS", Proceeding of the ST Workshop on Electronics for the LHC Experiment and IEEE 2004, Rome. 3. K. Kloukinas et al.,"Characterization and production testing of a quad, 12 bits, 4 0 M s / s A / D converter with authmatic digital range selection for calorimetry", Proceeding of the 11 Workshop on Electronics for the LHC and Future Experiments, Heidelberg Sept. 12-16, 2005. 4. ECAL group,"Test b e a m results for the C M S Electromagnetic Calorimeter", Proceeding of the Sr Conference on Astroparticle, Particle, Space Physics, Detector and Medical Physics Applications, Como, Oct 17-21, 2005.

UNIFORMITY OF RESPONSE OF THE ATLAS ELECTROMAGNETIC CALORIMETER SERIES MODULES C. OLIVER* Department of Theoretical Physics, Universidad Autonoma de Madrid, Madrid, 28049, Spain

Cantoblanco

During the construction of the ATLAS electromagnetic calorimeter, four barrel modules, and three end-cap modules, were scanned with electron beams of fixed energy. Final results on the uniformity of each module, and comparisons between modules are given. The contributions to the constant term expected in ATLAS at the LHC are analyzed.

1. Introduction The ATLAS electromagnetic calorimeter1 will play a central role in providing accurate measurements of energy, position and time of electrons and photons at the future LHC proton-proton collider. Its performance requirements have been defined studying the benchmark channels H-^yy, H-> 4e. A 1% resolution on a low mass Higgs requires a constant term less than 0.7% over the full detector coverage devoted to precision physics (rj< 2.5). To achieve this resolution the response non-uniformity has to be less than 0.6%. The EM calorimeter is based on a lead-liquid argon sampling technique which guarantees intrinsic linear behavior, stability of the response and radiation hardness. The accordion geometry of the electrodes and of the absorbers avoids uninstrumented areas in the azimuthal angle ^. The gap, filled with liquid argon, allows for the drifting of ionization electrons under high voltage. The calorimeter is composed of three parts, a barrel (|^| < 1.475) closed by two endcaps (1.375 < |?7| < 3.2) and segmented in depth in three compartments. The Front sampling is made of narrow strips in TJ for the precise position measurements needed in J/T? separation. The Middle sampling has a depth of 16 to 18 XQ and collects most of the shower energy. The Back sampling recovers high energy tails and helps to separate hadronic from electromagnetic particles. In addition, a thin presampler corrects for energy losses due to dead material for

M
On behalf of the ATLAS EM group.

374

375

The barrel is subdivided into two half barrels centered over the z-axis, in their turn subdivided into 16 modules. Each module is composed of 64 absorbers and 128 electrodes2. A 2kV high voltage is applied by the readout electrodes with respect to the grounded absorber. The end-cap3 is composed of two wheels with a projective crack at rj = 2.5. It is subdivided into 8 modules. In one module, 96 absorbers and electrodes are stacked in the outer wheel and 32 in the inner wheel. The gap increases with the radius and therefore decreases with r\. The detector signal is proportional to the sampling fraction and the drift velocity and inversely proportional to the liquid argon gap thickness so that an ^-independent detector response could be achieved with a continuously varying HV with the pseudorapidty. However, for technical reasons, a high voltage varying by steps was chosen, defining seven (two) high voltage sectors in the outer (inner) wheel. 2. Signal Reconstruction and Calibration The amplitude of the ionization signal after shaping and amplification is reconstructed from the linear combination of five samples, digitized at 40 MHz, located around the peak, using the optimal filtering (OF) technique4 in order to minimize the noise. The OF coefficients are calculated from the noise autocorrelation matrix, the pulse shape and its derivative. The physics pulse is obtained from the calibration one, which is corrected for the difference in the input current and injection point. Unlike the final situation in ATLAS with its fixed delay between trigger and sampling clock the beam test trigger comes asynchronous w.r.t the sampling clock. Therefore the OF weights are calculated in steps of 1 ns in order to fill the 25 ns trigger window. The energy in one cell is obtained by applying the electronic gain to the reconstructed amplitude in ADC counts, including a small non-linear behavior. The electron energy is reconstructed by summing the clusters formed in each compartment around the cell with maximum energy deposit. To compensate for energy losses in front of the calorimeter and leakage beyond it, weights are applied to the presampler and Back compartment measurements. The energy resolution and the total energy are extracted from a Gaussian fit starting 1.5
* There are two types of electrodes in each gap: one for tj < 0.8 and the other one for JJ > 0.8

376

3. Test Beam Results 3.1. Test Beam Setup In 2001 and 2002, four barrel and three end-cap series modules of the ATLAS electromagnetic calorimeter were extensively tested under electron beams. Electron/positron secondary and tertiary beams are derived from SPS protons and their energies range from 10 to 245 GeV. Scintillators for triggering and timing and 4 multiwire proportional chambers with horizontal and vertical wire planes for beam position reconstruction were present at several positions along the beam line. 3.2. Energy Corrections Due to the finite cluster size, an observable lateral leakage is produced for events when the electrons is incident far from the centre of the cluster. This dependence is fitted and corrected for with a parabola (second order polynomial) for EMEC (EMB) modules. In <j> direction, the non-containment effect is convoluted with the local nonuniformities of electric field and sampling fraction, due to the accordion geometry. This translates to periodic non-uniformities on the cluster energy along the ^-direction, where the period is the distance between two consecutive absorbers. The ^-modulation is corrected using a Fourier polynomial (for the barrel only odd coefficients are considered). Additional corrections are needed to obtain the electron energy for EMEC modules. The specific setting of the high voltage by finite sectors in the end-cap induces a linear variation of the energy response as a function of 77 in each sector. To correct for the slope of the energy response inside HV sectors, tjdependent weights are applied on each cell. The HV correction parameters are obtained from a fit to the test beam data and their values are in good agreement for all EMEC modules. The energy measured in test beam shows an unexpected non-uniformity along ^-direction for the EMEC modules. This effect is correlated with the middle cell capacitance measured along this direction. Therefore, this leads to interpret this asymmetry in the calorimeter response as a consequence of the fluctuations of the gap thickness, generated during the module stacking. The most accurate capacitance measurements for each module are used to correct this effect. Since there is no accurate measurement for ECC1 module, an ad-hoc correction has been extracted from the ^-dependence of the energy averaged over 77.

377

The transition region (77 = 0.8) between the two kinds of electrodes in a barrel module is 2 mm long and it is not instrumented, leading a loss of energy up to 15%. This effect is well understood by MC simulation and is corrected multiplying the energy reconstructed by a 77-coefficient. Finally, not related to the calorimeter itself but the test beam setup, the two feedthroughs used in the barrel test are not identical: FT-1 was a prototype and FTO a final element. It has been observed that FT-1 degraded with time and some resistive crosstalk between connectors developed, producing a higher energy dispersion in the region covered by FT-1. To correct this effect, a fraction of the measured current in a ^-adjacent cell is subtracted to the current of the excited cell. 3.3. Uniformity Results The mean reconstructed energy after all corrections for the tested end-cap and barrel modules is shown in Table 1 and Table 2, respectively. The dispersion (RMS/mean) of the distribution of the mean cluster energy reconstructed at every scan point is the non-uniformity of the module response. The non-uniformity was found to be less than 0.6% for the whole barrel and end-cap modules (results from CPPM5 and UAM6 groups), thus ensuring a global constant term below 0.7%. Table 1. Mean energy and non-uniformity of the three tested end-cap modules over the whole analysis region. For the outer (inner wheel), statistical errors on the mean energy are around 0.02 GeV (0.1 GeV) and statistical errors on non-uniformity are around 0.02 % (0.1 %).

Module <E> RMS/<E>

ECC0

Outer wheel ECC1

ECC5

188.8GeV

119.1 GeV

119.3GeV

119.1 GeV

0.52%

0.57%

0.60%

0.59%

Inner wheel ECC5

Table 2. Mean energy and non-uniformity of the three tested barrel modules over the whole analysis region. Module <E> RMS/<E>

P13 233.3 GeV 0.57%

P15 235.8 GeV

M10 232.6 GeV

0.62%

0.62%

378

4. Conclusions The uniformity of the electromagnetic end-cap and barrel calorimeter response to a electron beam has been studied on several production modules. Corrections for leakage and geometrical effects are applied on calibrated energies. For endcap modules, additional high voltage and capacitance correction, are needed. All correction are reproducible from one module to another. The results meet the challenging ATLAS requirements despite the complex geometry of the detector. References 1. ATLAS Collaboration, ATLAS Calorimeter Performance Technical Design Report, CERN/LHCC/96-40 (1996). 2. ATLAS Electromagnetic Liquid Argon Calorimeter Group, Nucl. Inst. Meth. A500, 202 (2003). 3. ATLAS Electromagnetic Liquid Argon Calorimeter Group, Nucl. Inst. Meth. A500, 178 (2003). 4. W.E. Cleland and E.G. Stern, Nucl. Inst. Meth. A338, 467-497 (1994). 5. F. Hubaut, C. Serfon, Response uniformity of the ATLAS electromagnetic endcap calorimeter, ATL-LARG-2004-015 (2004). 6. C. Oliver, J. Del Peso, Outer Wheel uniformity of the electromagnetic endcap calorimeter, ATL-LARG-PUB-2005-002 (2005).

PLANS FOR THE VERY FORWARD REGION OF ATLAS THE LUCID LUMINOSITY MONITOR JAMES PINFOLD t Physics Department, University of Alberta, Edmonton, Alberta T6G 2N5, Canada. The ATLAS collaboration is actively pursuing a program to develop the detection capability of the ATLAS detector. A prime motivation for this development is to allow the measurement of the absolute luminosity as well as, for example the study of diffractive physics and two photon physics. It is envisaged that the measurement of elastic scattering in the Coulomb region could be used to calibrate a dedicated luminosity monitor, LUCID, that will be used to measure the luminosity of ATLAS across the full range. The challenges of the design, simulation and testing of the LUCID detector are discussed below.

1. Introduction The general purpose LHC detectors, ATLAS [1] and CMS [2] have excellent central coverage in the Ecm, up to a pseudorapidity (|TJ|) ~ 5 because of the dominating interest in high-Pr physics at the LHC. However, ATLAS is now implementing additional forward detectors, which aim to cover the forward diffractive regions with tracking and/or calorimetry [3]. The CMS detector has very similar capabilities to ATLAS with the TOTEM [4]. The approved design for the upgrade to the forward system of ATLAS incorporates a relative luminosity monitoring using LUCID, a system of 400 (2 x 200) Cerenkov tubes, surrounding the beam pipe at 2=±18 m. Absolute calibration of LUCID is performed using Coulomb scattering measured in runs with special machine optics and Roman Pot detectors at z=±240 m reaching very close to the circulating beams to intercept small-angle protons. A measurement of the elastic cross section, extending down to the Coulomb scattering region (\t\ a 6 x 10"4 GeV2), is proposed for ATLAS and should provide a 2% determination of the luminosity, as well as a direct measurement of cjei, atot, the

t Representing the Luminosity and Forward Physics Working Group.

379

380

p parameter, and the nuclear slope parameter B. We also plan to use wellcalculable physics processes (e.g. W/Z production, pp->uu) for calibration. It is also envisaged that the LUCID detector would be useful in defining a rapidity gap trigger for a wider program of diffractive physics [5]. In addition it could, in principle, be useful in determining the "event-plane" when the LHC is operating in heavy-ion collider mode.

2. A Dedicated Luminosity Monitor for ATLAS - LUCID Traditionally, for hadron colliders, the luminosity has been measured using dedicated scintillation counters which measure the fraction of bunch crossings with no interactions [6]. Such detectors, typically placed at relatively large pseudorapidities, can have large acceptance for inelastic pp(p) interactions. However, the design luminosity of the LHC is so high that the fraction of crossings with no interactions will be small. Also, radiation levels are high enough to render scintillation counters unusable. At the LHC it is necessary to utilize a dedicated detector to measure the average number of interactions per bunch crossing, directly. In order to monitor with precision the average number of pp interactions we have proposed a robust, fast, radiation hard, dedicated detector that is mostly sensitive to particles from inelastic pp collisions and insensitive to backgrounds. This detector will monitor the average number of inelastic pp interactions by

Figure 1. A sketch of the proposed positioning of one of the two LUCID detectors. The other detector would be placed in the symmetrically opposite position.

measuring the number of particles, and their arrival time, in each bunch crossing. The proposed detector is called LUCID. The design of the LUCID

381

detector is largely based on that of the Cerenkov Luminosity Counter (CLC) that is currently operating at CDF [7]. 2.1. The Design of the LUCID Detector The LUCID detector consists of two modules that are located in available space between the beam pipe and the conical beam-pipe support structure. This places LUCID in the forward shielding, after the ATLAS Endcap Toroids. A schematic depiction of the proposed placing of LUCID is given in Figure 1. The beampipe inner diameter at the position of LUCID is 123mm with thickness 1.5 mm. The range of pseudorapidity covered by LUCID is |5.4 > 6.1| when the front of the detector is placed at z = ±16.98 m from the IP. An artist's impression of the LUCID detector in place is given in Figure 2. Each LUCID detector module consists of 200 thin walled (0.5mm), 1.5 m long, cylindrical, gas filled Cerenkov counters. These counters are arranged around the beampipe in five concentric rings with 40 tubes each, pointing to the centre of the interaction region. The tubes are 0.5 mm thick aluminium pipes, with a polished interior. The tube diameters are, from the inner to the outer layer of tubes: 13.28, 15.22, 17.43, 19.97, and 22.89 mm. A cross-section though LUCID is depicted in Figure 3. The light from each tube is collected by a Winston cone machined from aluminium. The ength of the Winston Cones vary from around 80 mm for the inner layer of tubes to 130 mm for the outer layer of tubes.The wall thickness of the Winston cones is envisaged to be 0.5 mm. The light from the Winston cone is collected by a bundle of clad 1mm fused silica fibres. Fused silica is chosen for its radiation hardness. Each bundle is routed through the forward shielding

Figure 2. An artists impression of the LUCID detector in place between the beampipe and the conical support tube of the beampipe. to remote photo-detectors. We will utilize 7 fibres to cover the egress of the Winston Cone for all layers. The complete detector structures are enclosed in a

382

thin, aluminium pressure vessel filled with C4F10 or isobutane. The nominal operation point is at atmospheric pressure, however the vessel will be designed to operate up to one atmosphere above atmospheric pressure. This allows us to increase the gas radiator pressure in the event that more light is required. The high radiation environment requires that the Cerenkov light from the LUCID detector is taken out to remote photo-detectors housed on the top of the nose shielding using fused silica fibres. These fibres penetrate the pressure vessel containing the detector through leak proof connectors. At this point magnetic fields are not a problem and there are two candidates for the read-out technology currently under investigation. The first is the baseline solution: multi-anode photo-multiplier tubes with quantum efficiency -20%. The second, solution is provided by avalanche photo-diodes (APDs) with quantum efficiency -80%. It is envisaged that the readout crates servicing the MaPMTs will be placed adjacent to the "nose" of the forward shielding, as shown in Figure 4. An LED system is planned to enable light to be introduced into the mouth of each Cerenkov tube, for the purpose of calibration.

Figure 3. A cross-section through LUCID at the mouth of a detector (z = ±16.98m).

We are proposing to use C4F10 (or isobutane) as a radiator as it has one of the largest indice of refraction, n=l. 00137 (n=l.00143), at atmospheric pressure for commonly available gases and good transparency for photons in the UV region where most of the Cerenkov light is emitted. The Cerenkov light cone half angle is ~3°, and the momentum threshold for light emission is 9.3 MeV/c for electrons and 2.7 GeV/c for pions. Prompt particles coming from the IP (primaries) will traverse the full length of a Cerenkov tube and generate a large amplitude signal in the photo-detector. Particles originating from secondary interactions of prompt particles in the

383

Figure 4. The positioning of the readout structures for LUCID.

detector material and beam-pipe (secondaries) are softer and will traverse the counters at larger angles, with shorter path lengths. In addition, the light from these particles will also suffer a larger number of reflections. The signals from these secondaries are therefore usually significantly smaller than those from the primaries and can be discriminated using amplitude thresholds in the electronics and in the offline data analysis. This is not feasible with scintillation counters. Also, at higher occupancies, when two primary particles traverse a single counter the resulting signal is twice that of a single particle. Given that there are no Landau fluctuations the counter's amplitude distribution should show distinct peaks for the different particle multiplicities hitting the counters. This enables us to count the actual number of primary particles hitting LUCID and helps prevent the counting rate from saturating at high luminosity, allowing a more precise measurement of the average number of interactions [8]. The LUCID detector counts tracks from minimum bias events that consist of elastic scattering events, diffractive events, double diffractive events, low-pT scatters and semi-hard QCD (2—>2) processes. The total cross-section for these processes is estimated to be ~100 mb (PYTHIA, msel=2). Secondary particles from primary interactions as well as beam-halo and beam-gas particles are expected to form the main backgrounds.

2.2. Simulating the L UCID detector A GEANT4 simulation of the main features of the LUCID detector has been written. A complete simulation awaits the details of the support structure and pressure vessel design. The GEANT4 simulation includes a simulation of the complete path of a photon in the LUCID detector, from generation in the

384

Cerenkov light produced by an incident particle, through transmission down the Cerenkov tube to collection in the Winston cone attached to the tubes and subsequent transmission down a fused silica fibre-optic cable to the PMT. A GEANT4 simulation of a single muon incident along the axis of a Cerenkov tube of LUCID reveals some important characteristics of the LUCID detector. A high momentum particle incident on a LUCID Cerenkov tube gives the maximum light output at an angle of 0° to the axis of the tube. The light is then transmitted over ~5m of fused-silica fibres. Light is lost in the fibres due to the geometrical acceptance of the fibre bundle and attenuation in the fibre. For readout fibres with a numerical aperture 0.37 we expect roughly 200 photons with wavelength between 200 nm and 800 nm to reach the photo-detector Assuming the light is detected by the photomultiplier tube with an average quantum efficiency of approximately 10%, we would expect to generate around 20 photo-electrons per track. Twice that number of photons would be detected for two tracks in the same tube. .- *

:

•

•

RetpotM AmpMiOt OvttxjMfl

MteMI

•

" '

:

1 1

•s

• •

•

w*o4*e> W-
LL^ 100

MO

300 «00 MO • * j r « f of Prxrto"*

600

«X

- - • • - '

Figure 5. The estimate of the number of photons (200 nm -> 800 nm) generated by minimum bias tracks in LUCID tube that reach the PMT (assuming the readout fibre has a NA=0.37).

The LUCID detector performance was verified using minimum bias (MB) events generated by PYTHIA-6.152 [8] using the standard ATLAS settings for MB event generation. The distribution of the number of photons in the wavelength range 200nm-^800nm that reach the PMT (assuming a readout fibre with NA = 0.37 is utilized) is shown in Figure 5. As can be seen from Figure 6. there is a linear relationship between the luminosity and the number of tracks detected in LUCID. The results shown in Figures 5. & 6. are not appreciably disturbed by the presence of the beam pipe.

385 Rcwcrc* CorrMOon

N****f C Ev*«s J** C r o i * . ^

Figure 6. The variation of number of tracks detected by LUCID (in each arm) with luminosity for MB events in ATLAS.

A initial study of background light generated in the fused-silica readout fibres have been carried out. This problem is ameliorated by the fact that the large Cerenkov angle of the light in the fused silica fibre, with moderate to low NA ( < 0.37), precludes efficient trapping of the Cerenkov light from particles traveling at small angle to the fibre axis. Another ameliorating factor is that for efficient trapping of Cerenkov light particles must be moving through the fibre at large angle, with small path lengths and thus smaller light emission. In addition, we shall be reading out each fibre This will allow us to use pattern recognition to reduce the effect of this background. Fully detailed Monte Carlo (MC) data with test volumes corresponding to fibre optic readout in various positions is currently being generated to allow more detailed estimates of the background. Beam-tests of fibre-optic cables with a view to studying backgrounds in the readout fibres are also envisaged.

2.3. Prototyping the LUCID Detector Six aluminium Cerenkov tubes with diameter roughly 2 cms have been equipped with Winston cones and mounted in a test pressure vessel capable of taking a 2 Bar overpressure. The gas vessel has been provided with fibre-optic readout cables with 42(7x6) fibres that take the light to a 64-channel Hamamatsu (H7546B-03) MaPMT Isobutane gas will be employed as the radiator. This six tube prototype will be tested using an electron beam at DESY in October 2005. The main purpose of this first beam test is to provide data for the validation of the MC simulation. To do this the variation of the light output with angle, position in tube, pressure, temperature and different fibre type will be measured.

386

2.4. The Radiation Environment Around the LUCID Detector The radiation environment at the LUCID detector is severe, at 105 Gy/yr total ionizing dose. In addition, there is a flux of 1013 (1 MeV equiv.) neutrons/ (cm2.yr) through the same region. Of particular concern are the epoxy joints that provide the gas seals for LUCID. As, the LUCID detector will be relatively inaccessible it is of great importance that all elements of the LUCID detector be tested for the equivalent of 10 years of running at these radiation levels. The radiation level in the region of the MaPMT readout is estimated to be only a few to several Gy/yr The ionizing dose expected for the primary read-out crates is of the order of 0.1 Gy/yr. The neutron flux for the same region is approximately 1 kHz/cm2, or ~1010 neutrons/(cm2.yr). The electronics and photo-detectors will be tested for radiation levels equivalent to ten years of running in this less critical radiation environment.

3. Calibrating the LUCID Detector The elastic scattering data will be recorded at an luminosity of 1027 c m ' V and thus the luminosity monitor will be calibrated at this luminosity. The luminosity for normal running of ATLAS will be in the range 1033-1034 cm"V. This implies that the calibration has to be carried over six to seven orders of magnitude. Roughly, there are three orders of magnitude from the difference in p., two orders from the difference in the number of bunches and two orders of magnitude from the lower bunch intensity. Thus, there is an advantage in using a luminosity monitor that can resolve the individual bunches i.e. with a time resolution better than 25 ns. This implies that the six to seven orders of magnitude reduces to four or five as the number of bunches factorize out in the extrapolation from low to high luminosity. In practice this means that the calibration has to be transferred from running conditions with 2xl0"4 interactions per bunch crossing to conditions with about 20-30 interactions per bunch crossing. These numbers illustrate the two main requirements for any luminosity monitor to be calibrated via special low luminosity runs: there has to be very good background rejection to avoid counting of fake interactions at low luminosity and there has to be a dynamic range of at least 20-30 to avoid saturation of the monitor at high luminosity. The LUCID concept addresses both these questions. Using Cerenkov tubes pointing to the interaction point implies possibilities of rejecting low energy secondary particles and thus being sensitive to mainly primary particles from the interaction point. Furthermore, should be possible to separate one and two particle peaks in the amplitude distribution from each tube, allowing counting of particles instead of counting only hits. If these double tracks were counted as a single hits a non-linearity in the response of roughly 5-10% would result.

387

Particle counting will be of importance to: avoid saturation at high luminosity; obtain the necessary dynamic range; and, ensure linearity of response. The cross-calibration of LUCID with the elastic scattering data will take place at a luminosity of 1027 c m ' V . Based on Figure 5, we can expect approximately 1.7 tracks to be detected per arm of the LUCID detector per pp collision, where we include single diffractive processes. At the luminosity of 1027 cm"V, the collision rate is 100 Hz for a 100 mb cross-section, which thus corresponds to 170 Hz tracks per arm. The double-arm rate is slightly less. The rate of 170 Hz per arm in LUCID is to be compared with the elastic rate in the RP of about 30 Hz. Hence, the calibration of LUCID in a high p run can count on at least the same order of numbers of events being available from LUCID. There are (at least) three other methods of luminosity determination that could be employed to give cross-checks and useful consistency criteria when used with the detectors based approaches discussed above.. For example the machine measurement of the luminosity by the Van de Meer method [9] could well be used. However, it is envisaged to provide only a 5%->10% accuracy. In principle, physics processes could give a better measurement. Two main methods are being considered by ATLAS at present [10]. The first uses a QED process: the production of a muon pair by double photon exchange[ll]. This process is experimentally clean but has a very small observable cross section because of the necessary pT trigger conditions imposed on the muons. Another process that has been proposed, and has been studied in some detail by ATLAS, is the QCD production of W/Z gauge bosons, and the measurement of their production rate times leptonic branching fraction [12]. This process measures the parton luminosity directly, and hence may serve to normalize the parton luminosity in many other production processes of interest. Alternatively, W/Z boson production is one of the theoretically best investigated QCD processes, and the parton distribution functions participating in the production process are already well and continue to improve further. Moreover, this process has a high enough rate that it could serve as a luminosity monitor as well as an absolute luminosity gauge. 4. Conclusion The LUCID approach to luminosity monitoring, based on the CLC detector at CDF provides a practical, robust and economical approach to luminosity monitoring across the full luminosity range. The LUCID detector is also relatively light (~40Kg/end) and should be supportable in the available space between the beampipe and the its conical support vessel without any costly changes to the beampipe or beam line elements. The fast timing characteristics of this detector allow a bunch by bunch luminosity determination. Also, LUCID monitors MB events with a very large cross-section (~100 mb) allowing an online luminosity measurement. However, more work needs to be done in

388

prototyping, simulation and modeling to prove that LUCID can meet the challenges of precise luminosity monitoring in the ATLAS environment. This work is well underway.

References 1. 2. 3.

4. 5.

6. 7. 8. 9. 10.

11. 12.

The ATLAS Collaboration, "Detector and Physics Performance Technical Design Report", CERN/LHCC/99-15, 1999. The CMS Collaboration, "The Compact Muon Solenoid, Technical Proposal, CERN/LHCC/94-38", 1994. The ATLAS Luminosity Lol CERN/LHCC/2004-010 (22 March 2004); ATLAS Collaboration: ATLAS-TDR-15, Vol I, 438-447, 25 May 1999; http://atlasinfo.cern.ch/Atlas/GROUPS/PHYSICS/TDR/access.html; http:// sbhepl.physics.sunysb.edu/wrijssenbeek/ATLAS-LumiForwPhys.html. TOTEM Technical Proposal, CERN/LHCC 1999 - http:// totem, web. cern. ch/ totem; CERN-LHCC/2004-002, TOTEM TDR. FP420 : "An R&D Proposal to Investigate the Feasibility of Installing Proton Tagging Detectors in the 420m Region at LHC", CERN-LHCC2005-025, LHCC-I-015 F. Abe et al., Phys. Rev. D 50, 5550, (1994); D. Cronin-Hennessy, A. Beretvas, P.F. Derwent, Nucl. Instr. & Meth. A 443, 37 (2000). D. Acosta et al., Nucl. Inst. & Meth. A461, p540, (2001); D. Acosta et al., Nucl. Inst. & Meth. A 494, 57, (2002). T. Sjostrand, Computer Physics Commun. 82, 74, (1994). K. Potter, CERN Yellow Report, CERN 71-01 (1971). Eg: The ATLAS Luminosity Lol CERN/LHCC/2004-010 (22 March 2004); J. L. Pinfold, "Determining Luminosity Using Physics Processes", ATLAS Physics Workshop, Rome, June 2005. G.Shamov and V.I. Telnov, Nucl. Instrum. Methods A 494 (2002) 51. M. Dittmar, F. Pauss & D. Zurchner, Phys. Rev. D36, 7284, (1994). 2002.

FAST, LONG-WAVELENGTH SCINTILLATORS AND WAVESHIFTERS* K. ANDERT, B. BAUMBAUGH, A. BROTHERS, A. DAVID, H. GUNTHER, J. GURROLA, D. KARMGARD, T. MADLEM, J. MARCHANT, P. MCGOUGH, M. MCKENNA, R. RUCHTIf , J. THOMPSON AND M. VIGNEAULT Department of Physics and QuarkNet Center, University of Notre Dame Notre Dame, Indiana 46556, USA L. HERNANDEZ AND C. HURLBUT Ludlum Measurements Inc. 501 Oak Street Sweetwater, Texas 79556, USA

Studies are presented of new Hue-green to red emitting scintillator and waveshifter materials for tracking and calorimetric applications for the detection of ionizing radiation. Materials include plastic scintillators, liquid scintillators, and plastic scintillating and waveshifting fibers. Program goals are to develop faster and more efficient detection media for a variety of experimental applications. 1.

Introduction

This paper focuses on the development of scintillating and waveshifting materials for calorimetric and tracking applications. Ultimate objectives are to identify and produce materials with high quantum efficiency, fast fluorescence decay and that afford tolerance to radiation exposure. For calorimetric detectors, a common detector technique is the scintillation sampling calorimeter. 1 ' 2 ' 3 Plates or "tiles" of blue-violet emitting scintillator are read out with embedded multi-clad waveshifting fiber containing the dye Y l l . Light emission is near X = 510nm and the fluorescence decay time is relatively long > 7ns. This paper provides preliminary measurements of scintillating organic plastic and liquid materials containing new fluorescent dyes that are potential replacements for conventional scintillator and waveshifter. In a high radiation environment, scintillator-based liquids may be flushed from a system * This work is supported in part by the U. S. Department of Energy STTR Program, the U. S. National Science Foundation and Department of Energy QuarkNet Program, the National Science Foundation RET Program and the University of Notre Dame. + Current address is the National Science Foundation, Arlington, VA 22230 USA.

389

390 and replaced by new material as needed. Additionally several of these dyes have been incorporated as waveshifters in multiclad fiber for the readout of tile scintillators. 2.

Scintillators and Wavelength Shifters

Bulk plastic scintillator and waveshifter samples were polymerized in either polystyrene or polyvinyltoluene. Liquid samples were prepared from aromatic, organic liquid-scintillator base. Fiber waveshifters were produced with doublecladding. 2.1. Optical Characteristics of Scintillators and Waveshifters Initially the materials were characterized optically. Emission and excitation spectra for each of the samples were measured with a Varian Cary Eclipse Fluorescence Spectrophotometer attached to a custom data acquisition system. Figure 1 displays a sample of emission spectra for three new shifter dyes, DSFl and AH328B in organic liquid base and DSF2 in a polyvinyltoluene base. The fluorescence decay times of several samples were measured with a Horiba Jobin Yvon Pulsed Diode Laser Excitation System with TBX-04 Picosecond Photo Detection Module. Samples were excited at X = 377 nm and the fluorescence decay observed at emission maximum. As can be seen in Figure 2, the photo-excited fluorescence decay time of DSFl (T = 1.3ns) is significantly faster than standard Y l l (T = 7.4ns) by a factor of over 5. The decay of AH238B is found to be slower (T = 11.5ns), but the emission wavelength for this dye is considerably longer (in the yellow/red).

2.2. Scintillation Efficiency Studies Efficiencies of the scintillator samples were characterized by exposing them to a Sr beta source. Signals were detected with a Hamamatsu R1104 photo multiplier connected to a LeCroy QVT pulse height analyzer and read into a custom computer data acquisition system. Optical filtering was used between the samples and the phototube to cut out emission from the bulk material and other primary dyes (if used). 4 Table 1 displays the measured scintillation efficiencies as a function of wavelength for several scintillation materials prepared for this study. The efficiencies are corrected (scaled) for the wavelength dependent quantum efficiency of the Rl 104 PMT. 90

391

Emission 3DZT1 (DSF1| Ex»350nm

400

A

350 300-

•

*•

|250

1 200

X.

150 100

50

J-

,

450

T*—-<—

500

550

600

Wiv*length (nm)

Emission 4D (AH238B) Ex=450nm

/ \

/\y 570

590 610 Wavalangth (nm)

\ 630

650

Figure 1. Emission spectra for several waveshifters. DSFl waveshifter organic liquid base (upper left). DSF2 waveshifter in PVT base (upper right), AH238B waveshifter in organic liquid base flower left).

Figure 2. Fluorescence decay times excited by a Diode Laser at 377 nm. Light is detected at the emission maximum for the samples. Fits to the curves indicate the following: DSFl in organic liquid (upper left, 1.3ns); Y l l in PVT (upper right, 7.4ns); AH238B in organic liquid (lower left,

11.5ns).

392 Table 1. Scintillation Efficiency of Samples Composition

Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid

3.

L1 L1 L1 L1 L1 L1 L1 L2 L2

Primary

Secondary

Dye

Dye

AH238B P1 P1 P2 P3 P3 P3 P3 P3

AH238B BBQ DSF1 DSB1 DSB2 DSF1 DSF2

Emission Wavelengt h (nm)

Efficiency Relative to BBQ

600 450 600 470 480 500 510 480 480

0.84 1.05 1.42 1 1.02 1.11 1.25 1.08 1.02

Tile-Fiber Measurements

Those secondary dyes that revealed promising scintillation performance and rapid fluorescence decay were then incorporated into double-clad fiber for waveshifter studies. The dye materials so chosen included DSFl, DSF2, DSB1 and DSB2 and these were compared to the standard dye Yl 1. All were used to read out scintillation tiles of 10cm x 10cm dimension and 6mm thickness of scintillation material (EJ200) with emission maximum in the blue-violet (X = 425nm). The detection system has been described elsewhere4 and includes a three-fold coincidence of trigger counters, including a PMT that views the scintillation tile itself and two other thin independent counters one above and one below the tile. The coincidence of these counters was used as a gate for the signal from the waveshifter fiber, located in a groove in the tile. The optical signal from the fiber was read out using a Hamamatsu H7422P-40 GaAsP PMT. The signals from the PMT were averaged over 2048 pulses and displayed and recorded on an HP54502A digitizing oscilloscope and read into a computer. Such signals were recorded for triggers that occurred in response to a 90 Sr beta source. Figure 3 displays the recorded signals from the tile/fiber combinations. As can be seen, the best performance is observed for DSB1 shifter and it is superior to Yl 1 in both overall signal as well as signal response time. DSFl and DSF2 have lower overall signal, but have very fast response. The data show that for DSFl, the full width at half maximum (FWHM) is 9.0ns and a fit to the tail with a single exponential function reveals a decay time of T = 6. Ins. The values of these quantities for Yl 1 are FWHM = 17ns and T = 12.1ns, indicating the

393 superiority of the DSFl in time response. Table 2 summarizes the results of these measurements and others. r0465 DSF1 f.02-stc01 Ru106 140mV/div 10-12-05

r0466 DSF2 f.02-stc01 Ru106 140mV/div 10-12-05

h

AiitoFVfFof; riM55:Calc CoftrnBl 1 -

C 6 T«+.'• 0.3S93 B: 1.S3+.'-0.0320 RMSE 0.463

1 /

\ \

i%^ i_ _ _ —

**J

1

Tims (ta'ni J f a f l 2 n s « c )

i

T

Time c'fcris aw 0.2 n 3 K | JOB, « . * »

r0468 DSB1 f.13-stc01 RU106 140mV/div 10-13-05

r0467 Y-11 f.17-stc01 RU106 140mV/div 10-13-05

AuloFrtFor; [<S$7.Cate C a ( o m n T c y = A'exp(-»t;6ttB A: 358 +/- T75 C 12.1 -.'-0.0B94 6 1 07 +,'- 0.0343 fiMSE 0 353

'^0F\t?'Qt'''r0ttB:C'aic7Coiu'ffii™i E A 3 32E-0G3 ->i- 123 C:6.21 4V-0C5W

•e

/ ,

•f

„,

B

i

I i I

.

Time [bins ire 0 2 nsac)

Figure 3. Averaged output pulses from various waveshifters in double-clad fiber of 940mm diameter and lm length reading out EJ200 scintillating tile of 6mm thickness and 10cm x 10cm area. Photosensor is a GaAsP PMT. DSFl (upper left); DSF2 (upper right); DSBl (lower left); Y l l in PVT (lower right). Values are summarized in Table 1 below. Table 1. Summary of Waveshifting Efficiency for Sample Materials. Waveshifter

Emission wavelength

Efficiency rel .to Yl 1

Decay Time Source Exc.

FWHM Source Exc.

Decay Time Laser Exc.

Yll DSBl DSFl DSF2

529nm 500nm 480nm 480nm

1 1.02 0.84 0.79

12.1ns 6.2ns 6.1ns 6.8ns

17ns 9.4ns 9.0ns 9.6ns

7.4ns 1.3ns

394

4.

Conclusions

From the studies underway in this program, several new scintillation and waveshifter materials have been produced and studied with fluorescence emission wavelengths in the red to green and with rapid fluorescence decay times and reasonable efficiency. These have been incorporated into multi-clad fiber and tested as waveshifters in tile-fiber geometry (DSF1, DSF2, DSB1, DSB2). Additionally, several of these have been used as secondaries in liquid scintillation cocktails (DSF1, DSF2, DSB1, DSB2, and AH238B). Initial measurements indicate that DSB1, DSF1 and DSF2 offer superior performance in fluorescence decay time to conventional Y l l waveshifter. DSB1 and DSF1 offer comparable integral fluorescence efficiency. These materials are therefore potential replacements for Y l l in any application where fast timing is required, or where improved signal to noise would be an advantage by utilizing short integration times.

Acknowledgments We would like to thank P. Kamat and the staff of the Notre Dame Radiation Laboratory for the use of the Diode laser spectrometer, E. Hahn of Fermilab for fiber end-polishing, and to the University of Notre Dame for support facilities and resources for high school students and teachers at the Notre Dame QuarkNet Center. References 1. The CDF II Technical Design Report, Fermilab-Pub-96/390-E. 2. CMS The Hadron Calorimeter Project Technical Design Report, CERN/LHCC 97-31. 3. ATLAS Tile Calorimeter Technical Design Report (TDK), CERN/LHC/9642. 4. M. Albrecht, et al, Scintillators and Wavelength Shifters for the Detection of Ionizing Radiation, Astroparticle, Particle and Space Physics, Detectors ad Medical Physics Applications, ICATPP-8, M. Barone, E. Borchi, C. Leroy, P.-G. Rancoita, P.-L. Riboni, R. Ruchti, Eds, World Scientific, 502-511 (2004).

T E S T B E A M RESULTS OF T H E CMS E L E C T R O M A G N E T I C CALORIMETER

ROBERTO SALERNO On behalf of CMS ECAL

collaboration

The CMS (Compact Moun Solenoid) experiment is a general purpose detector designed to explore the physic of proton-proton collisions at the centre-of-mass energy of 14 TeV. The CMS electromagnetic calorimeter (ECAL) will play a major role in the study of the electroweak symmetry breaking, particularly through the exploration of the Higgs boson sector. In order to test the ECAL performance, a full supermodule with the final mechanics, cooling and electronics was exposed in a test beam during the summer 2004. The most relevant results obtained at this test beam are presented together with the calibration procedure and its comparison with the precalibration obtained from the laboratory optical measurements and from cosmic rays.

1. Introduction One of the major goals of the Large Hadronic Collider is to elucidate the origin of the mass, in particular by discovering the Higgs Boson predicted by the Standard Model. For a low Higgs mass (< 120 GeV) the main channel is H —> 77 whereas for a higher mass range the channels H —> WW and H —> ZZ become important. In both cases the electromagnetic calorimeter plays a central role in the selection of these events with high efficiency and low background. In particular an excellent energy and angular resolutions are mandatory for the detection of electrons and photons. The above considerations have led the CMS collaboration to opt for an electromagnetic calorimeter built of Lead Tungstate (PbW04) crystals. The low time constant, as well as the short radiation length and Moliere radius of PbW04 allow the building of a rapid, compact and granular detector. However the relative low light yield of the crystals entails the usage of avalanche photodiodes with an intrinsic gain. Since both the gain of the APDs and the light yield of the crystals are very sensitive to the temperature fluctuations, a precise control of the temperature is required. Exhaustive detector description can be found in [1] and [2].

395

396 The energy resolution of ECAL can be parametrized as a function of the energy as
_

S

N

£ ~7I®£

'

(1)

where S is called the stochastic term which includes the shower fluctuations and the photo-statistics, N is the noise term which comprises the electronic noise and C is the constant term which includes the calibration precision, the longitudinal non-uniformities of the light collection and the fluctuations due to the temperature. 2. The testbeam setup During the summer 2004 a complete supermodule of ECAL barrel (SM10), built accordingly to the final production specifications, was tested with an electron beam at the H4 facility in the CERN North Area. The laser monitor system, the HV and LV distribution systems and the cooling system were built in the final design. At the test facility the supermodule was positioned on a moving table allowing to center each crystal on the beam. The position of the impact point of electrons on the crystal surface is measured, event by event, by a set of scintillating fibers, with a resolution of 145 /xm. SM10 had 1700 channels equipped with the Multi Gain PreAmplifier chip in the VeryFrontEnd cards.

3. Amplitude reconstruction The whole procedure for reconstructing the particle energy starting from the electromagnetic calorimeter has been investigated at the test beam. The signal from the ECAL-APD is digitized at 40 MHz and 10 samples are recorded, including 3 samples before the start of the signal to allow an estimation of the baseline. The signal amplitude A is obtained with a general technique as a weighted sum of the information of the samples (Si):

A = Y^ SiWi .

(2)

i

The weight Wi are obtained by minimizing the variance A. The additional requirement Sw, = 0 allows an event by event baseline substraction since any costant level does not contribute to A. This procedure and the resulting expression for the weights are described in detail in [3].

397 4. Noise A measurement of the electronic noise can be made by applying the amplitude reconstruction procedure to data with non signal, taken with a random trigger. The distribution of the noise measurements in a single channel is about 39 MeV while in a 3 x 3 and 5 x 5 clusters is about 120 MeV and about 200 MeV, this corresponds to 3 times and 5 times the single channel noise, demonstrating the removal of the small component of noise which is channel-to-channel correlated. 5. Calibration A precise intercalibration of the CMS ECAL is mandatory to keep the constant term in energy resolution at the level of 0.5%. The final calibration of the CMS ECAL will do obtain in situ at LHC from physical events. Nevertheless it is important to have a preliminary initial intercalibration at LHC startup. This can be obtain with a good accuracy from calibration with a beam of electrons, from laboratory measurements of the crystal light yield and from the calibration with the cosmic ray. In the testbeam the intecalibration (d) can be defined as the ratio of the response of each channel to impinging electrons of a given energy (Mj) over the response of a reference one (Mref). The position of the peak in the energy distribution measured in one single crystal is chosen as the estimator of the channel response. Detail studies have demonstrated that accuracy of the intercalibration coefficient is at the level of the 0.3% [4]. An initial intercalibration for the whole detector can be obtained from the APD and electronics gain measured in the laboratory and the light yield determined by combining the measurements of the trasparency and the response to 7 from a mCo source. The agreement with the intercalibration is at the 4% level [5]. The ECAL supermodules will be intercalibrated using cosmic rays before the installation in CMS. A statistical precision of 2 — 3% on all crystals is achievable [6]. 6. Energy Resolution The shower, for energetic electrons, is not contained into a single crystal. Therefore the energy of the electron is reconstructing as the energy deposited in a 3 x 3 and 5 x 5 clusters of crystals. The sum of the energy deposited in each the 9 or 25 crystals in the array, is carried out by using the calibration coefficients obtained with the procedure explained in Sec. 4.

398

CMS ECA^ Preliminary Resolution 3x3 «/E "0.44* 0.01%

Preliminary Resolution SxS IT/E "0.45 ± 0 . 0 1 *

Figure 1. Distribution of the energy reconstrued in 3 X 3 crystals (left) and in 5 X 5 crystals (right) at 120 GeV/c.

6.1. Resolution

for central

impact

Restricting the impact to a 4 x 4 mm2 region, over which the containment remains almost unchanged, it gives a measure of the intrinsic resolution of the calorimeter without any corrections. In figure 1 the energy resolution in the case of central impact are presented for a 3 x 3 and 5 x 5 clusters of crystals [7]. 6.2. Cluster

containment

corrections

Since the sum of 9 or of 25 crystals are vary with position of the incident particle, a correction for this effect is implemented. The position of the impinging particle is evaluated comparing the amplitudes measured in the crystals to the left and to the right of the hit. The Log(E\/(E2) method used has been extensively tested with simulated events and beam data [8]. The effect of the cluster containment on the energy resolution obtained displacing the beam to a crystal corner is in perfect agreement with the resolution for central impact. 6.3. Energy resolution

as a function

of

energy

The energy dependence of the energy resolution has been fitted by the expression 1. Data was taken at a series of beam momenta corresponding to the energies: 20, 30, 50, 80, 120, 180 and 250 GeV. The resolution as a function of energy is extracted for the case of uniform impact across 20 x 20 mm2 area of the trigger. The cluster containment corrections described in the Sec. 6.2 are applied. The figure 2 shows the energy resolution when summing the energy in a 9 and 25 crystals arrays, the errors quoted

399

1.4

- \\ : l\

CMS ECAL Preliminary Resolution for 704

1.2 1 O.B

r

0.6

^\

S=2.35V-0.1% fV=0.25OV- 3.0 GeV C'0.33-/- 0.01 % 3x3 $=2.96*/-0.1 % N=0.166*/- 3.4 GeV C'0.32*/- 0.01 %

0.4 0.2 0

. . . .

1 . . . .

200

I

250 Energy (GeV)

Figure 2. Resolution as a function of t h e energy reconstrued by summing 5 x 5 crystal (black line) compared to the sum of 3 x 3 crystals (red line).

on the fitted parameters include b o t h the statistical and the estimated systematic uncertainties. T h e resolution is better t h a n 0.5% for electrons energy greater t h a n 100 GeV. 7. C o n c l u s i o n T h e CMS electromagnetic calorimeter collaboration has achieved relevant results in the test beam 2004, all of t h e m consistent with the target values established by the collaboration to deal with the challenging physics program of C M S a t L H C . T h e C M S E C A L aims a t an excellent energy resolution for uniform impact. The comparison with the testbeam calibration have demonstrated t h a t the initial pre-calibration of t h e whole detector by laboratory and cosmic ray measurements is accurate at the 4% and 2-3% level. References 1. CMS collaboration, The electromagnetic calorimeter project. Technical Desing Report, C E R N / L H C C 97-33. 2. CMS collaboration, Data Acquisition and High-Level Trigger technical Design Report, C E R N / L H C C 2002-26. 3. R. Bruneliere and A. Zabi,Amplidute reconstruction in the CMS ECAL with the weights method", C M S N o t e in preparation 4. G. Franzoni at al, Intercalibration for the CMS ECAL at 2003 H4 testbeam, C M S IN 2 0 0 4 / 0 2 3 .

400 5. L.M. Barone at al., Correlation between Light Yield and Longitudinal Trasmission in PbWOi crystals and impact on the precision of crystal intercalibration, CMS R N 2004/005. 6. A. De Min at al., Pre-calibration of the CMS electromagnetic calorimeter with cosmic rays, C M S note in preparation. 7. R. Bruneliere, P. Jarry and A. Zabi, Energy resolution performanca of the CSM electromagnetic calorimeter, C M S note in preparation. 8. J. Descamps and P. Jarry, Test beam results on the performance of the cluster containment correction for the CMS electromagnetic calorimeter barrel, C M S note in preparation.

ELECTRON A N D P I O N RESULTS FROM T H E ATLAS FORWARD CALORIMETER 2003 T E S T B E A M

M. S C H R A M * Carleton

University 1125 Colonel By Drive, Ottawa, K1S 5B6, CANADA E-mail: mschram@physics. carleton. ca

The ATLAS Forward Calorimeter (FCal) modules have been exposed to beams of electrons, pions, and muons within the energy range of 5 GeV ^ E ^ 200 GeV at the CERN SPS, in Geneva, Switzerland. These modules are unique calorimeters with fine transverse segmentation allowing for a significant improvement in the hadronic resolution and reconstructed energy. Presented in this paper are leptonic performance results. Also included are the hadronic performance results using a transverse weighting technique.

1. ATLAS Forward Calorimeter Within the ATLAS experiment, the calorimeters will be used to reconstruct the electron, photon, jet, and missing transverse energy 1 . In particular, the ATLAS Forward Calorimeter (FCal) has been designed to provide complete forward coverage for jets in the pseudo-rapidity range of 3.2 < \rj\ < 4.9. Results of the 2003 test beam are presented in this paper. 2. The 2003 Test Beam Setup For the 2003 test beam, the full ATLAS FCal modules were placed in a cryostat in the H6 test beam area at CERN. The FCal modules consist of three longitudinal segments: one electromagnetic copper-liquid argon section and two hadronic tungsten-liquid argon sections. These modules have been exposed to beams of electrons, pions, and muons within the energy range of 5 GeV < E < 200 GeV delivered by the CERN SPS. The positioning of the modules was such that the beam impact point corresponded to T] « 3.8 in ATLAS. The 2003 beam line downstream of the last bending *on behalf of the ATLAS FCal group

401

402

magnet includes several scintillation counters and multi-wire proportional chambers followed by the cryostat containing the FCal modules, a warm tailcatcher, a concrete beam stop, and a muon scintillation counter. 3. Energy Reconstruction 3.1. Pulse

Shape

The pulse shape in the FCal modules is generated by a charged particle which ionizes the active medium (liquid argon). The initial pulse shape caused by the ionized liquid argon is triangular. After passing through the electronics, the signal peak is rounded, followed by a small depression. In the 2003 test beam, the pulse shape was sampled every 25 ns for a total of seven samples. The first sample was used to determine the pedestal (internal offset) and the signal peak was typically located in the fourth sample. To determine the current deposited in each channel the amplitude of the pulse was calculated using the optimal filtering (OF) technique 2 . 3.2. Optimal

Filtering

The pedestal-subtracted signal was processed by applying the OF method using five of the seven samples. The OF weights were determined using the physics pulse shape and the autocorrelation matrix. The physics pulse shape was derived from the physics events and the autocorrelation matrix was derived from the pedestal events. The OF weights were calculated with one nanosecond binning. 3.3.

Clustering

For electrons, a three-dimensional topological cluster algorithm 3 was used to determine which readout cell had sufficient energy to be included in the analysis. Three noise dependent parameters of the topological cluster algorithm were investigated in this study: (1) The seed threshold (crseed): clusters are seeded by cells with a signalto-noise ratio above the seed threshold. (2) The neighbor threshold (aneighbor)'- cells with an absolute value of their signal-to-noise ratio above the neighbor threshold are asked for neighboring cells. (3) The cell threshold (
403

For pions, no clustering was used, hence all readout channels were included in the analysis. 4. Test B e a m 2003 Results 4 . 1 . Energy

Resolution

The energy resolution of the FCal modules was determined by measuring the width of the reconstructed energy for particles of the same type and energy. 4.2. Electron

Results

For the electron analysis, the energy ranged from 5 GeV to 200 GeV. The energy resolution was parametrized using the following equation: a E

a y/E

b E

,. w

Here a is the sampling term, b is the noise in the electronics, and c is all remaining uncertainties of the detector. The results of the electron study are presented in Table 1. At high energy the weighted average of the three settings gives an energy resolution of (3.10±0.05)%. Table 1. Test beam 2003 ATLAS Forward Calorimeter electron energy resolution using different topological clustering settings. The cluster parameters are ®seed I & neighbor

I®cell

Cluster Parameters

Sampling [% • \/GeV]

4/3/2 4/3/1 4/2/0

4.3. Pion

Noise [% • GeV]

Constant [%]

29.594±0.741

91.84±2.47

3.07±0.07

28.066±0.825

106.95±2.31

3.12±0.07

30.407±0.917

115.33±2.68

3.13±0.08

Results

Due to the non-compensative nature of the FCal modules, two software techniques were investigated to minimize the difference between the detector response to electrons and hadrons. The flat weighting technique applies a single calibration term for each FCal module. The transverse technique 4 applies weights based on the radial distance from the cluster center. In

404

this study, the transverse technique has 16 calibration constants for each module, one for each radial centimeter spacing. For this analysis the pion energy ranged from 40 GeV to 200 GeV. The energy resolution was parametrized using the following equation:

E^VZ®0

(2)

Here the terms are the same as in Section 4.2, with the exception of the noise term which was accounted for before the fit. The results of this study are presented in Table 2 and show that the transverse weighting technique improves both the sampling and constant term. Moreover, at high energy the energy resolution is (2.74±0.34)% which is within the ATLAS design requirements. Table 2. Test beam 2003 ATLAS Forward Calorimeter pion noise subtracted energy resolution using different weighting methods. Sampling [% • VGeV]

Constant [%]

Flat Weights

96.81±8.52

5.24±1.25

Transverse Weights

90.44±0.59

2.74±0.34

In addition to the improvement in the energy resolution, the transverse weighting technique nearly achieves compensation with a e/7r-ratio ranging from 1.05 at 40 GeV to 0.98 at 200 GeV. 5. Outlook In the future, a full Monte-Carlo study will be performed to help understand properties such as the energy deposited in the upstream material and shower containment. In addition, a study on jet reconstruction using the transverse weighting technique will be performed. References 1. The ATLAS Collaboration, ATLAS Calorimeter: Performance Design Report, 1997, CERN-LHCC-96-40.

405 2. W. E. Cleland and E. G. Stern, Nucl. Instrum. Meth. A 338 467 (1994). 3. C. Cojocaru et al. "Hadronic calibration of the ATLAS liquid argon end-cap calorimeter in the pseudorapidity region 1.6 < \q\ < 1.8 in beam tests," Nucl. Instrum. Meth. A 531 (2004) 481 [arXiv:physics/0407009]. 4. A. Savine, Calorimttry in High Energy Physics 595 (2000).

LINEARITY, E N E R G Y A N D POSITION RESOLUTION OF T H E ATLAS ELECTROMAGNETIC CALORIMETER SERIES MODULES

C. SERFON* * Centre de Physique des Particules de Marseille 163 avenue de Luminy - case 902, 13288 MARSEILLE Cedex 09 FRANCE, E-mail: [email protected]

During the construction of the ATLAS Electromagnetic Calorimeter, several series modules were exposed to electron beams of energies between 10 and 245 GeV. Final results on the linearity and energy resolution are given, together with the precision obtained on position and angular measurements.

1. Introduction The ATLAS 1 experiment on the future LHC 2 (proton-proton collider) is dedicated to the search of the Higgs boson, the study of the Standard Model and the search of physics beyond the Standard Model. To reach these goals, high performances of the detector are needed. The Higgs boson decay H - • 77 (one of the most promising discovery channel for low mass Higgs) imposes some constraints on the energy and position resolution of the electromagnetic calorimeter 3 . Moreover, to extrapolate the absolute energy scale from Z° mass (used for calibration) to other energy domains, a good linearity (non-linearity less than 0.1%) is needed. 2. The ATLAS electromagnetic calorimeter The ATLAS electromagnetic calorimeter is a lead-liquid argon (LAr) sampling calorimeter with an accordeon geometry which allows a good hermiticity. It is divided in a barrel section (|^| <1.4) and in two end-caps *Now at department fur physik, LMU Munchen. Am Coulombwall 1, D-85748 Garching, Germany tQn behalf of ATLAS Liquid Argon group

406

407

(1.375< I77I < 3.2) and segmented longitudinally in three samplings (see figure 1). The sampling 1 (front) is finely segmented into the rj direction in strips which allow to perform precise position measurements (see section 3.2) and 7/71-° separation. The sampling 2 (middle) is the one where the electromagnetic shower deposit most of its energy. The sampling 3 (back) allows to recover the longitudinal leak of the most energetic electrons or photons. For \r)\ <1.8 a presampler is located in front of the calorimeter.

Towers in Sampling 3 A<|>xAr| = 0.0245)0.05

Figure 1. Cut view of a sector of the EM calorimeter. The typical granularity of the samplings is indicated.

3. Test-beam studies To study the performances of the calorimeter, 4 (resp. 3) modules of the barrel (resp. end cap) have been tested under beam. The studies performed with these data concern in particular linearity and energy resolution, angular and position resolution, and uniformity4. The two first analyses are detailed hereafter.

408

3.1. Linearity

and energy

resolution

The first kind of study made with test-beam data is the study of the linearity of the EM calorimeter 5 . This study relies on the Monte-Carlo simulation. It needs in particular a good description of the matter in front of the calorimeter, as well as the as a good knowledge of the calorimeter itself (dead material, inactive layers, e t c . ) . This simulation was tuned with the test-beam data, and all the energy distributions obtained are in good agreement between data and MonteCarlo. A calibration scheme has therefore been developped on the simulation and applied to the data leading to linear response within 0.1% in the energy range 15 < E < 180 GeV (see figure 2). 1.006 LU

1.004 -

E Hi

1.002 -

ate gearn energy uncertainty

uncorrellated beam energy uncertainty correllated ~"

0.9960.994 0.992 0.99

^ 50

100

150

200 EbeamtGeV]

Figure 2. Ratio of reconstructed electron energy to the beam energy as a function of the beam energy at the point (r/=0.687,0=0.282). The ratio is normalised to 1 at 100

GeV. The calibration is made according to equation (1) :

Erec =

a(E) + b{E)E™ + c(E)

EvisEms

+

d(E)f;samp

£

._ i=l,3

E

f°

•fcell impact(A^) • (1 + /jea*a,e((0))(l)

where a, b, c are parameters extracted from MC to take into account energy loss in front of the presampler, and between the presampler and the calorimeter, d is used to correct for variation of sampling fraction. The sampling term and the local constant term of energy resolution obtained after this procedure are respectively !0%y/GeV and 0.20% (see

409

figure 3) and reach the ATLAS requirements.

0.05 "^0.045 m

0.04 0.035-e 0.03 0.025

Sampl. Linear term [%/^(GeV)J [%] 10.0±0.1 0.20±0.04 ~

Data Data noise subtracted Noise

0.02 0.015 0.01 0.005 0

100

200 Ebean, [ G e V ]

Figure 3.

Energy resolution as a fuction of the energy.

It has to be underlined that the linearity study has been made in only one i] position in a setup different of the ATLAS one (no magnetic field, material in front of the calorimeter different). Study to transport this calibration scheme to ATLAS is being carried on.

3.2. Position

and angular

resolution

The position and angular resolution have been studied in the barrel as well as in the end cap 6 . The position measurement in the front and middle section is made using energy-weighted bary center. Because of the finite size of the cells leading to the so-called S-shape effect (bias in the reconstructed position), corrections have been applied. The position resolution in the front (resp. middle) along the r\ direction is 240 /im (resp. 540 Mm)- By combining the position of the particle reconstructed in these two samplings, the polar angle of the incoming particle can be reconstructed. The corresponding angular resolution is plotted as a function of the energy on figure 4. An average resolution in the range 50-60 mx&d/y/E(GeV) is obtained on the whole calorimeter, in agreement with the design expectations.

410

60

+

u755 •

•

^

t

•

t

t

1

+

250 E, t?45

•

40

; b

25 50 75 100125150175200225250 Energy (GeV)

20

40

60 80 100 120 140 160 Energy (GeV)

Figure 4. Variation of polar angle resolution with energy for barrel at 77=0.69, <j>=0.28 (a) and end-cap at TJ = 1.74, 0=0.18 (b). 4.

Conclusion

T h e tests beam performed on the series modules of the ATLAS E M calorimeter have shown t h a t the performances required (linearity, energy and position resolution) by the benchmark physics channels (H —)• 7 7 , H —> 4e, and W —> ev especially) could be obtained on series modules. These studies have now to be carried out in simulation with the real ATLAS design, quite different from the test-beam setup.

Acknowledgements I want to t h a n k my colleagues of the ATLAS liquid argon group for providing me results and figures.

References 1. The ATLAS Technical Proposal, CERN/LHCC/94-43 (1994). 2. The Large Hadron Collider, CERN/AC/95-05 (1995). 3. ATLAS Calorimeter Performance, Technical Design Report, CERN/LHCC/96-40 (1996). 4. C. Oliver, Uniformity of Response of the ATLAS EM Calorimeter Series Modules, these proceedings. 5. J. Colas et al., Energy Linearity and Resolution Performance of the ATLAS Electromagnetic Barrel Calorimeter in the CERN Electron Test-Beam, to be submitted to Nucl. lustrum. Meth. A. 6. J. Colas et al, Position resolution and particle identification with the ATLAS EM calorimeter, Nucl. lustrum. Meth. A 550, 96 (2005).

PRESHOWER SILICON STRIP DETECTORS FOR THE CMS EXPERIMENT AT LHC ANITA TOPKAR, PRAVEENKUMAR S, BHARTI AGARWAL & S K KATARIA Bhabha Atomic Research Centre, Trombay Mumbai, 400 080, India Y P PRABHAKAR RAO, N SHANKAR NARAYANA & NIKHIL SHARMA Bharat Electronics Limited, Jalahalli Bangalore, 560 013, India, A specific research and development program has been carried out in India to develop the technology for 32-strip silicon strip detectors for application as a preshower detector for CMS experiment at LHC, CERN. The production of strip detectors is already underway to deliver 1000 detector modules and 80% production is completed. In this paper, we summarize this research & development work carried out to develop the detector fabrication technology. The results of various tests carried out to evaluate the performance of detectors have been presented. The quality control procedures used for detector qualification during the production have been described.

1. Introduction The preshower silicon detectors will be used in the CMS experiment for the n°/y rejection. The detector has a geometry of 63mm x 63mm and it incorporates 32 P+strips with width of 1.78 mm with a pitch of 1.9 mm [1]. The fabrication technology to produce such silicon strip detectors with very good uniformity over a large area of ~40mm2, low leakage currents of the order of 10nA/cm2 per strip and high breakdown voltage of > 500V has been developed in India. The strip detectors have been characterized using IV and CV measurements to obtain the leakage currents, breakdown voltages and full depletion voltage. The performance of the detectors has been evaluated after exposure to fast neutrons and protons to ensure their radiation hardness for operation in the harsh radiation environment of LHC. A strict quality control procedure is being used for qualification of detectors during the production. In the subsequent sections, we discuss the fabrication technology development, detector characterization and present the performance of the strip detectors.

411

412

2.

Detector Design & Fabrication Process Development

In view of the expected radiation damage during ten years of operation in the LHC environment, the electrical specifications of the detectors are quite stringent i.e. low leakage currents and high breakdown voltage. Therefore, the strip detector mask has been specially designed to incorporate features such as strips with rounded corners and floating field P+ guard-rings [2]. The floating field guard rings reduce the surface electric fields and hence increase the breakdown voltages of the strips. Figure 1 shows the schematic cross section of the detector. windows In the passivation

Scribe-line

"^j

1000

Passivation

^

Figure 1 Schematic cross section of the strip detector (All dimensions are in microns)

The detector has been fabricated using a four layer mask process. The process for the detectors has been developed by Bhabha Atomic Research Centre (BARC) in a close collaboration with the foundry, Bharat Electronics Limited (BEL) [2-3]. The 32-strip silicon detectors have been fabricated using 4", retype, <111>, 3-5 kQ-cm, 300 urn thick silicon wafers. The front P + strips and the back N+ contact have been obtained by ion implantation The silicon surface of the strip detectors is passivated with a thick silicon dioxide layer. The PSG layer is used as the top passivation layer over the front metal. This layer is resistant to contaminants which could diffuse and degrade the detector performance with time. The bad performance of a strip i.e. high leakage, low breakdown voltage is known to be due to pin holes in the implant & bad oxide quality. The increase of current around the full depletion voltage is attributed to the injection from the back plane [4]. The desired specifications of high breakdown voltage and low leakage currents have been achieved by strict contamination control, incorporating implant damage at the back side for gettering, having an uniform and thick >T layer at the back for suppressing back injection. Critical process parameters such as temperature, time, ambient for processes such as oxidation,

413 implantation anneal and drive-in, etc have been optimized. Generation of defects in the bulk of the silicon and the quality of the oxide which passivates the silicon were of more concern in order to reduce leakage currents and improve breakdown voltage. The fabrication process has been optimized in several batches to tune the process parameters so as to achieve the desired specifications. 3.

Performance of Strip Detectors

The reverse IV and CV characteristics of the 32 strips of the detectors have been measured simultaneously using PC based automated measurement setups. The static characteristics of the detectors are as shown in Figure 2 and Figure 3. The characteristics show that the strip leakage currents are of the order of 5-10 nA at high voltages of the order of 500V. The CV curves are uniform for all the strips and the full depletion voltages are about 110V. The radiation hardness of the detectors have been tested by irradiating them to neutrons in a reactor (at BARC and at Dubna, Russia) or using a 24 GeV proton beam (at CERN). Figure 4 shows the strip currents measured at -18 °C and at a voltage of 300V after irradiation to a total neutron fluence of 1.36x 1014n/cm2. The detectors have been found to meet the specifications and show a stable behavior after the irradiation. 2.0E-10 T

1.5E-10

1.0E-10

5.0E-11

0.0E-KJ0 200

400

50

Bias Voltage (Volts)

Figure 2 TV Characteristics of all 32 strips of the strip detector

4.

100

150

Bias voltage (Volts)

Figure 3 CV characteristics of all 32 strips of the strip detector

Production of Strip Detectors and Quality Control

Detectors with very good uniformity and at a production yield of 40%-50% are being produced at BEL and 80% of the production has been completed. During production, the quality control is being carried out as per the qualification procedures specified by CERN - IV & CV measurements, geometrical tolerance

414

of 100 microns over detector length and width and thickness of the detector (within 300um-340um). Various parameters such as i. Leakage current of strips at full depletion voltage and at 300V, ii. Total leakage current at full depletion voltage at 300V and at 500V, iii. Breakdown voltage of strips and iv. Full depletion voltage are extracted from the static measurements and are used to qualify the detectors. All the tests are being carried out in a class 10,000 environment with controlled humidity and temperature. Figure 5 and Figure 6 show the statistical distribution of breakdown voltage and total leakage current at 300V for about 800 sensors. As can be seen, the breakdown voltage exceeds 500V for majority of the detectors. The total leakage current of the detector is less than 1.0 uA for most of the detectors though this tolerance is lO.OuA 800i

8.0E-06

600

6.0E-06

I

400

4.0E-06 200 2.0E-06

n.0E-!O0

275

1

4

Strip Nurrber

Figure 4 Leakage currents of the 32 strips after irradiation to neutrons

5.

350

425

500

7 10 13 16 19 22 25 28 31 Breakdown Voltage (V)

Figure 5 Distribution of the breakdown voltags for the detectors

Assembly of Detectors

The silicon strip detectors are assembled in the form of detector modules so as to integrate the front end electronics hybrid to the detector. The assembly of detectors involves very precise mechanical alignment of the detectors to the ceramic tile which carries the hybrid. The detector mounted in this form is finally glued to the aluminum tile which subsequently will be used to make ladders. The quality control procedure for the detector modules involves monitoring the degradation of the detector leakage current due to assembly and the mechanical damage such as chipping at the edges. Prototype modules have been made at India and the detector performance has been observed to be stable during assembly (Figure 7). Full scale production of micro modules has been initiated for delivering 1000 micro modules for the preshower detector of the CMS.

415 6.

Summary

We have developed the demanding technology for fabrication of strip detectors 250

1.0E-07 o initial D ceramic A A! tile

9

wjmmujmmimutu 0.1

2.1

4.1

6.1

8.1

Total Current at 300 V (M icroamps)

Figure 6 Distribution of the total current for the detectors

0

5

10

15

20

25

30

35

Strip Number

Figure 7 Leakage current of 32 strips of the 300V detector at various stages during assembly

with good uniformity, low leakage currents and high breakdown voltage. The detectors also meet the radiation hardness requirements for operation at LHC. The detector fabrication process has been briefly described. Subsequent to the development of technology, the full scale production of the detectors has been started in India. More than 800 detectors have been produced so far and the detector production will be completed by the end of 2005. Acknowledgement We acknowledge Dr Anna Peisert, CERN EP Division for her valuable suggestions during this development work and for evaluating the radiation hardness of detectors. References 1. Ph Bloch, Y H Chang, et al, Nucl. Instr. and Meth., A479, 265 (2002) 2. S K Kataria, Anita Topkar, et al , Proceedings of the DAE-BRNS Symposium on Intelligent Nuclear Instrumentation, 8 (2001) 3. Anita Topkar, Proceedings of the DAE-BRNS symposium on compact nuclear instruments and radiation detectors, 441 (2005) 4. Ph Bloch, A Cheremukhin, et al, IEEE Trans. Nucl. Sci. NS-49, 321 (2001)

TEST B E A M RESULTS OF THE CMS FORWARD QUARTZ FIBRE CALORIMETER

A. ULYANOV* Institute

for Theoretical and Experimental B. Cheremushkinskaya 25, 117218 Moscow, Russia E-mail: [email protected]

Physics,

The forward calorimeter extends the pseudorapidity coverage of the CMS calorimeter system from 77=3 to 77=5.2. It is composed of quartz fibres embedded into a steel absorber and measures Cherenkov light radiated by shower particles. Six production modules of the forward calorimeter have been tested and calibrated using particle beams at CERN. An outline of the calibration procedure and results on the energy resolution, response uniformity and linearity for electrons and pions are presented in this paper.

1. I n t r o d u c t i o n The forward calorimeter (HF) in the CMS experiment extends the pseudorapidity coverage of the calorimeter system from 77=3 to 77=5.2 which improves jet detection and the missing transverse energy resolution. 1 ' 2 HF calorimeter is especially important for identification of forward tagging jets associated with the Higgs boson production via vector boson fusion. The CMS forward calorimeter is based on detection of Cherenkov light radiated by shower particles in optical quartz fibres embedded into a steel absorber. The choice of quartz fibres as the active medium was dictated by the extremely hostile radiation environment in the region of high rapidities: at the nominal LHC luminosity the calorimeter will be absorbing up to 100 MRad/year. Several calorimeter prototypes were built in the past and successfully tested with high energy particle beams. 3 ' 4,5 ' 6 ' 7 In this paper are reported recent test beam results obtained with final production modules.

* On behalf of the CMS Collaboration

416

417 2. Design of the Forward Calorimeter The forward calorimeter is located at 11.15 m on either side from the CMS interaction point. It is a cylindrical structure with a hole for the beam pipe. This structure is physically composed of 18 wedges. A wedge (schematically shown in Figure 1) is made of 5 mm thick steel plates with grooves separated by 5 mm. These grooves are parallel to the beam line and are filled with 600 (im core-diameter quartz fibres (fused-silica core and polymer hardclad). There are two types of fibres which alternate in grooves: long fibres run through the full length of the absorber (165 cm), and short fibres start 22 cm deep from the front face of the calorimeter and thus are sensitive mostly to hadronic showers.

E.

£ E

I

|

473 mm

1

I

Figure 1. Transverse view of a wedge. The readout segmentation is shown by dashed lines. Open and filled circles represent long and short fibres.

The fibres are bundled to form 0.175 x 0.175 (A77 x A) towers and connected to photomultipliers (Hamamatsu R7525) by means of 40 cm long air-core lightguides with highly reflective walls. Long and short fibres are bundled and read out separately. 3. Test B e a m S e t u p Six assembled wedges of the forward calorimeter were extensively tested at the H2 beam line of the Super Proton Synchrotron at CERN. The wedges were placed one by one on a table which could move horizontally and vertically across the beam. The results presented here were obtained with the

418 wedges positioned at an angle of two degrees with respect to the beam that corresponded to rj « 4. The trigger was provided by a set of scintillator counters in front of the calorimeter with an overlapping area of 5 cm x 5 cm. A muon veto counter was installed behind an iron block downstream the calorimeter. Five multiwire proportional chambers upstream the detector were used for accurate tracking of incoming particles. 4. Results 4.1. Energy

Calibration

Hadronic showers are subject to large transverse leakage effects in a single wedge, especially close to the beam pipe. That is why electrons were used to calibrate the six wedges in question. The entire front face of each module was scanned with a 100 GeV electron beam. For all electrons hitting tower i, we calculated Aij as the average signal in the long segment of tower j . The full wedge response was then required to be equal to the beam energy: ^AijWj

= 100GeV,

where Wj is the unknown calibration coefficient for tower i. Solving these 24 linear equations (there are 24 towers per wedge) we determined the calibration coefficients for all towers. With equal PMT gains the signal produced by 100 GeV electrons in the short segment was typically 30% of the signal obsereved in the long segment. Therefore the same calibration procedure was applied to the short segment but the full response was normalized to 30 GeV. Events with the particle impact points within 2 cm range from the module edges were excluded from the calibration procedure to reduce the leakage effect. The beam to radioactive source response ratio was measured that will allow radioactive sources to be used for the calibration of the rest of the calorimeter. 4.2. Response

Linearity

and Energy

Resolution

Calorimeter response to electrons and pions measured at different energies is shown in Figure 2. The long segment response to electrons is linear within 2%. The short segment response is non-linear indicating deeper penetration of electromagnetic showers into the calorimeter at higher particle energies. The quartz fibre calorimeter is sensitive mostly to the electromagnetic component of hadronic showers. As a result, the response to pions is lower than

419

1

.

I''

••

11 11 1

•

'''" •

1

11

1 1 1 1 , 1 1 1 1 1 1

~ _

410.8 0.6

-

=

x

-

0.2 7

•

•

•

,, 1 , , ., 1 . ,

50

11

electrons (long) electrons (short) pions (long) plons (short) 1, , . , , ,

CO

-

0.6

.

. i i i | , . i i | i i i . | • i

fl

-

1

#0.8

100 150 200 250 300 350

i,

"a 1.2 — t *

:

•

i,


°Q .

0.4 [- •

iu

UJ

« c

=

0

>, CD

~ ~

:

0.4 7

*

electrons (long+short)

0.2

*

plana (lahg+short).

",,.,! ,,,.!.,,, 50

~. •

i i i i 1 • i . i 1 i i t i 1 i .

100 150 200 250 300 350

Energy (GeV)

Energy (GeV)

Figure 2. Average response to electrons and pions in long and short segments as a function of particle energy.

•

20 40 60 80 100120140160180200

Energy (GeV)

50

i

•

•

•

•

i

.

.

.

.

i

.

.

. .

i

.

100 150 200 250 300 350

Energy (GeV)

Figure 3. Energy resolution for electrons (left) and pions (right) as a function of particle energy. The upper curves are for the long segment alone and the lower curves are for the two segments added together.

the response to electrons. The 7r/e ratio increases at higher particle energies due to growing fraction of neutral pions in hadronic showers. Response difference between different particles contributes to the jet energy resolution and can be reduced by using a sum of long and short segment signals as illustrated on the right plot in Figure 2. Optimization of additional weights for long and short fibre readouts is a subject of Monte-Carlo simulations. The electron energy resolution (Figure 3) is dominated by photoelectron statistical fluctuations. The stochastic term is in a good agreement with the light yield of 0.28 pe/GeV and the 40% fractional width of a single photoelectron peak. The constant term in the electron energy resolution is

420

dominated by structural non-uniformities: fiber-to-fiber distance is comparable to the transverse size of electromagnetic showers. The electron energy resolution is slightly improved when responses from the two segments are added together, mainly because the light yield is increased. The pion energy resolution is dominated by fluctuations of the fraction of neutral pions in a hadronic shower. However, in the 30-300 GeV range the resolution can be parametrized as a sum of stochastic and constant terms as shown in Figure 3 (right plot). 4.3. Response

Uniformity

In order to check the uniformity of the calorimeter, the wedges were scanned with 100 GeV electrons and pions across their front face. Apart from response oscillations with a step corresponding to the fibre-to-fibre to distance, these scans revealed response non-uniformities of a larger dimensional scale which were related to imperfections of the detector construction. In order to quantify the effect of these non-uniformities on the energy resolution, the wedge surface was divided into 2 cm x 2 cm cells and the average response was calculated in each cell. The cell-to-cell response fluctuations were about 5% for electrons and about 3% for pions. These are small contributions to the single particle energy resolutions. The effect on the jet energy resolution is even less important because the jet size is essentially larger than 2 cm. 5. Conclusions Six (out of 36) calorimeter modules were extensively tested and calibrated with high energy particle beams. These tests revealed no particular problem in the detector construction and assembly. Performance results were in agreement with the tests of prototype calorimeters. A radioactive source will be used to transfer the energy calibration to the rest of the calorimeter. References 1. 2. 3. 4. 5. 6. 7.

CMS Collaboration, Technical Proposal, CERN/LHCC 94-39 (1994). CMS Collaboration, Heal TDR, CERN/LHCC 97-31 (1997). N.Akchurin et al., Nucl. Instr. and Meth. A379, 526 (1996). N.Akchurin et al., Nucl. Instr. and Meth. A399, 202 (1997). N.Akchurin et al., Nucl. Instr. and Meth. A400, 267 (1997). N.Akchurin et al., Nucl. Instr. and Meth. A408, 380 (1998). N.Akchurin et al., Nucl. Instr. and Meth. A409, 593 (1998).

INTER-CALIBRATION A N D OFFLINE C O M P E N S A T I O N STUDIES OF T H E ATLAS CALORIMETER U S I N G B E A M TESTS A N D M O N T E CARLO

K.-C. VOSS* University of Victoria CERN PH, 1211 Geneve 23 Switzerland E-mail: [email protected]

The inter-calibration of the different ATLAS calorimeters has been studied using beam tests at the CERN SPS accelerator and also with Monte-Carlo techniques. Particularly important in these studies has been the development of techniques allowing the offline compensation of the hadronic response of the ATLAS calorimeter system. Results on calorimeter inter-calibration, offline hadronic compensation, and comparisons of data with Geant 4 simulation, are reported from beam tests in the ATLAS barrel, end-cap and forward calorimeter regions performed from 2002-2004. Results on ongoing studies extrapolating these techniques to the full ATLAS calorimeter are also discussed.

1. Introduction 1.1. The ATLAS Calorimeters

Hadronic

and

Electromagnetic

The ATLAS calorimeters are designed to provide an accurate measurement of the energy and position of electrons, photons and jets in the pseudo rapidity range \r]\ < 5. This is achieved by several sub detectors (EMEC, EMBARREL, HEC, FCAL, TILE) [1] [2] which use different techniques for the shower detection. For all calorimeter components the response to hadrons differs significantly from that for electrons of the same energy. Therefore for hadrons a correction has to be applied to the signals obtained on the corresponding electromagnetic scale.

*on behalf of the atlas liquid argon collaboration

421

422

1.2. Electromagnetic

and hadronic

showers

Electromagnetic (EM) showers are purely based on electromagnetic processes. Due to the short distance between the individual EM processes the showers are very compact and have high energy densities. Showers from electromagnetic particles are known to a high precision and therefore can be simulated very well. A hadronic shower on the other hand consists of EM and hadronic energy. It is usually much larger than an electromagnetic shower and has large fluctuations. The hadrons in the shower can react with the nuclei of the calorimeter. Energy deposited in nuclei excitation is not detected and hence is regarded as detector invisible energy. Also particles like neutrinos which escape the detector are produced for example in pion decays. These are effects which the calibration has to account for.

2. Hadronic calibration cell weighting The different response of the calorimeter to electrons and hadrons can be corrected using so-called software compensation. The goal is to get an equal response for the electromagnetic as well as purely hadronic component of a hadronic shower. Thus fluctuations in the hadronic shower on the energy reconstructed can be minimized and the energy resolution can be substantially improved. The technique of signal weighting has been successfully used in previous experiments (see [4]). The deposited energies in individual read-out cells of a segmented calorimeter can be identified as of electromagnetic or hadronic origin on a statistical basis. The goal is to reconstruct calorimeter final state objects (clusters) first and calibrate those using a "local" normalization based on locally deposited energies in the calorimeter. For ATLAS these weighting constants have to be derived from MonteCarlo (MC) simulations, and the procedure is tested proving that beam test data are well described. The essential Monte-Carlo information in this tuning process is the knowledge of all the shower components as the visible energy and the total energy in each read-out cell. Thus the weights could be directly derived from the MC simulation and applied to beam test or Monte-Carlo data. 3. 2002 Liquid Argon End-cap Beam Test setup The beam tests were carried out in the H6 beam at the CERN SPS. This beam provides pions, electrons or muons in the energy range 6 < E <

423 200GeV. Positioned in the liquid argon (LAr) cryostat were one EMEC module, three HEC1 modules and two HEC2 rear modules. Cerenkov counters were used for the trigger up to 80 GeV. The impact position and angle of particles have been determined from four multi wire proportional chambers (MWPC) with vertical and horizontal planes. A veto wall rejected beam halo particles and muons. 3.1. Energy reconstruction

- cluster

algorithm

A cluster algorithm has been developed to reconstruct the energy of the impact particle. In each layer the topological neighbours of a given cell are considered: they have to share at least one common corner. A seed is chosen if the energy is significantly above noise: Eaeed > 4crnoise. The general cut-off at the cell level is Eceu > 2crnoise, the cluster is expanded as long as one neighbour has an energy of Eneighbour > 3<jnoise. In [5] and [6] it was already presented that it is possible to extract the cell weights as a function of the energy density using the beam test data alone. In these publications it is also shown that the Monte-Carlo simulation and the real data agree very well. In addition to this data driven method the cell weights for the local hadronic calibration have been extracted using the Monte-Carlo simulation. In the simulation of single pions the amount of EM, non-EM, visible and escaped energy per read-out cell as it is calculated by GEANT4 [3] is stored in so-called calibration hits and can be used for analyses. With this knowledge the weight per cell can be calculated e.g. as a function of the cell energy density. The application of these weights give a better energy resolution of the reconstructed pions and electrons and is compatible with the calibration scheme based purely on the beam test data. 4. First implementation of local hadronic calibration scheme A first version of the local hadronic calibration scheme was implemented in ATLAS. First the clusters in the calorimeter are classified as hadronic or electromagnetic. In case they are tagged as hadronic, cell weights are applied to each cell in the cluster in order to correct the energy scale. The weights and criteria for the classification were extracted from 400k simulated single pion events in the full ATLAS simulation. They were produced in batches of 40k for 1,3,5,10,20,50,100,200,500,1000 GeV keeping all the information of the true energy deposits in the calorimeter cells stored

424 in t h e calibration hits. Afterwards t h e cluster algorithm was carried out and different cluster moments were calculated. For the cluster classification look-up tables in 120 bins of energy and pseudo-rapidity were created. In every table t h e t r u e electromagnetic energy (EM) fraction is stored as a function of the average cell energy density < Pceii > and t h e depth of the cluster center Xcenter in the calorimeter. At t h e classification step for each reconstructed cluster the corresponding electromagnetic energy fraction is determined by extracting t h e value using the two moments < pCeii > and Xcenter and the look-up tables. If the E M fraction is below 85%, t h e cluster is classified as hadronic. T h e weights were calculated using t h e calibration hits. In the simulation the weights for each cell in each event can be calculated using the formula: T?l

—

,., . p w

„

-arhprp

II!

—

/pem i p n o n - e m v i s , r^non — eminvia ^l-Ar+Ab^^LAr+Ab. -re
• rpescaped \ +aLAT+Ab,<

"ceil ~ Menu wnere w — _M„_emt,„ m T h e weights were calculated from the simulation as function of the cluster energy and the cell energy density and are also stored in tables. All cells in hadronic tagged clusters are t h e n weighted with t h e weights which correspond t o the cluster energy and the individual cell energy density. Applying the described local hadronic calibration scheme consisting of the cluster classification and cell weighting to a pion test sample results in a significantly (about 10%) improved energy reconstruction. At the same time the the energy resolution is only slightly improved. Analysing the t r u t h information in t h e simulation reveals t h a t an improved weighting scheme should also be able t o improve the resolution. References 1. The ATLAS Collaboration, ATLAS liquid argon calorimeter: Technical design r e p o r t C E R N - L H C C - 9 6 - 4 1 , 1996. 2. The ATLAS Collaboration, ATLAS calorimeter: Performance design report, C E R N - L H C C - 9 6 - 4 0 (1997). 3. S. Agostinelli et al. GEANT4 Collab., Nucl. Instrum. M e t h . A 506 , (2003) 4. B. Andrieu et al. (HI Calorimeter Collab.),Nucl. Instrum. M e t h . A 336,1993 5. C. Cojocaru et al. [ATLAS Liquid Argon EMEC/HEC Collaboration], Hadronic calibration of the ATLAS liquid argon end-cap calorimeter in the pseudorapidity region 1.6 j \eta\ j 1.8 in beam tests, Nucl. Instrum. Meth. A 531 (2004) 481. 6. S. Menke [ATLAS Liquid Argon Collaboration], Energy-calibration of the ATLAS hadronic and electromagnetic liquid-argon endcap calorimeters, arXiv:physics/0310127.

GEANT4 and Software Applications Organizers: S. Giani and P. Binko

S. Giani M. Asai K. Amako M.J. Boschini A. Brahme M. Ellis F. Gargano P. Goncalves V.M. Grichine V. Ivantchenko

0 . Okudaira

I. Papadopoulos D. Rebuzzi A. Salzburger G. Santin J.P. Wellish

Symposium on the Applications of the GEANT4 Simulation Software GEANT4 Applications for the NASA Space Missions GEANT4 Application to Hadrontherapy in Japan AMS02 Italian Data Transfer System: Real Life Experience Application of GEANT4 in the Development of New Radiation Therapy Treatment Methods G4 Accelerator Applications A Full Monte Carlo Simulation of Silicon Strip Detectors Simulation of Radiation Monitors for Future Space Missions Electromagnetic Physics Modeling in GEANT4 Package Perspectives in Medical Applications of Monte Carlo Simulation Software for Clinical Practice in Radiotherapy Treatments GEANT4 Monte Carlo Simulation to estimate Gamma-Ray Emissions from Lunar Surface and SELENE Spacecraft CORAL, a Software System for Vendor-Neutral Access to Relational Databases GEANT4 Simulation of the ATLAS Muon Spectrometer Track Extrapolation with Intrinsic Navigation in the New ATLAS Tracking Scheme Applications of GEANT4 for the ESA Space Programme Detector Simulation in High Energy Physics 425

This page is intentionally left blank

SYMPOSIUM ON THE APPLICATIONS OF THE GEANT4 SIMULATION SOFTWARE

SIMONE GIANI PH, CERN Geneva 23, CH-1211, Switzerland The following executive summary includes an introduction to the Symposium on the GEANT4 applications, as well as the key elements of its program, and the related overall conclusions.

1. Executive Summary GEANT4 was born as an R&D project to develop object-oriented engineered simulation software. Laboratories and institutes such as CERN, KEK, SLAC, TRIUMF, ESA were amongst the main founders of the project, which was managed by the author from conception to the release of the first production version (1994-1998). A MoU then formalized the creation of an international collaboration (including also members such as INFN-Italy, IN2P3-France, JeffersonLab-Usa, Karolinska-Sweden, LPI of Russian Academy of Science, Tera foundation, W.Goethe Inst.-Germany, Budker Inst.-Russia, Pittsburg Univ.-Usa), which has continuously evolved and supported the software to date. GEANT4 simulates the propagation and interactions of elementary particles in matter and vacuum within an energy range varying from the eV (down to thermal and cold energies for neutrons) to the TeV (up to PeV for muons). The research phase granted to GEANT4 the possibility to study and develop innovative features: particle-production cuts in range, processes' final state computation in space and time per simulation step (e.g. with applications in gas detectors and optics), multiple models per physics process, tracking of particles until zero-range (with extended validity range for energy losses and multiple scattering), automatic secondaries creation below production-threshold near geometric boundaries (and handling of their thickness), integration of equation of motion in arbitrary fields, 3D voxelization algorithm for optimized navigation in geometry databases, parameterized volumes by material, solid models moving 427

428

in time, numeric precision tolerances, system of units and physics constants, user actions interface (e.g. including triple-stacks mechanism and biasing techniques). Most of these features have made the tool applicable and exploitable for a wide variety of advanced technological and scientific fields: radiation damage in space, astronauts safety, dosimetry, radiotherapy, detector simulation, design of imaging devices, accelerators, beam lines, radiation and neutron background, astrophysics. The value of scientific software is not evaluated on the metric of the number of customers, but rather on the relevance and impact of its applications. The Symposium focuses on multidisciplinary applications of GEANT4, in the domains of: • • • • •

INDUSTRY - including at General Electrics and Siemens in the field of medical instrumentation. SPACE - including for missions of the ESA space program, of NASA, and the International Space Station. MEDICINE - including radio/brachi/hadro-therapy, at Karolinska Inst., IST-Italy, LIP-Portugal, CHU Quebec-Canada, and other medical centers. TECHNOLOGY - including PET research by the international OpenGate collaboration and ion-therapy developments at KEK-CHIBA (Japan) PHYSICS - including detectors (LHC experiments, BaBar, AMS) and accelerators (ILC, CLIC, MICE, T2K).

The relative weight of the different applications indicates that the potential of the GEANT4 software is exploited for science beyond the physics domain.

GEANT4 APPLICATIONS FOR NASA SPACE MISSIONS MAKOTO ASAI Scientific Computing and Computing Services, Stanford Linear Accelerator Center, 2575 Sand Hill Road, Menlo Park, CA 94025, USA

ROBERT A. REED, ROBERT A. WELLER School of Engineering, Vanderbilt University VUStation B 351553, Nashville, TN37235, USA Geant4 is nowadays widely adopted as the simulation engine for NASA space missions. We will review three major application areas of Geant4: apparatus simulation, including pre-launch design and post-launch analysis; planetary scale simulation, including radiation spectra; micro-dosimetry simulation, including single-event effects on electronics components.

1. Apparatus Simulation 1.1. The GLAST mission GLAST1* will measure the direction, energy and arrival time of celestial gamma rays (launch foreseen for 2007). The LAT (Large Area Telescope) instrument measures gamma rays in the energy range from about 20 MeV to greater than 300 GeV (there are no other detectors in operation covering this range). In addition, the GBM instrument provides correlative observations of transient events in the energy range between 20 KeV and 20 MeV. GEANT4 was adopted since 1999 for the test beam simulations and the balloon flights2'3'. The geometry of the system was described and managed via a dedicated repository. The GEANT4 simulation was then integrated in the GAUDI framework to manage the event loop, source and hits generation, digitization and reconstruction. At present, massive Monte Carlo productions have been run. 1.2. Cassini-Huygens mission to Saturn and Titan The low-energy magnetospheric measurement system (LEMMS) has been simulated using GEANT44). All the sensitive elements of the LEMMS sensor contained in the aluminum and platinum inner and outer cylinders, as well as the 429

430

full collimator geometry, have been reproduced and simulated. Simulations of 2 MeV electron pencil beam, 50 MeV/n Oxygen pencil beam, and 2 MeV electrons circular plate distributions have been validated in comparison to calibration data taken on 22-micron foils. Channel efficiency and pass band determination have also been simulated and reproduced. 1.3. ACE/EPAM solid state detector system The NASA Advanced Composition Explorer EPAM detector is simulated with GEANT4 to study the effect of solar energetic electrons scattering5'. Single detector configurations have been simulated using a 91 KeV mono-energetic electron beam onto a 200-micron silicon detector (preceded by !4 micron dead layer). The double detector configuration has been simulated using an incident beam of 126 KeV. The study of the energy deposited in the SSD has shown that a percentage of the incident electrons will trigger channels lower than the incident energy because some electrons scatter out of the SSD. 1.4. Messenger mission to Mercury The Messenger Energetic Particle Spectrometer engineering specifications were reproduced via GEANT4 geometric modeling, including the EPS collimator, the MCP high voltage cup, the SSD mounting bracket, the PCB board, the SSD array, the large and the small pixels, and the Al flashing. In particular, the four rings of the EPS collimator have been reproduced including all holes, as well as the MCP screws and bolts. Placing the EPS model inside a spherical shell has simulated an isotropic environment: such simulation was produced generating 10 millions of 200 KeV electrons. The gamma ray spectra at the MCP from energetic electrons have thus been calculated via the GEANT4 simulation described above. GEANT4 simulations under development include the computation of background rates due to solar energetic particle events, the determination of the EPS geometrical factor, the examination of the channel pass-bands and of the SSD rates due to solar X-rays, and the study of the efficiency of the foil flashing to discriminate electrons for low energy ions6). 1.5. Combined ion and neutron spectrometer for space applications The National Space Biomedical Research Institute CINS (Combined Ion and Neutron Spectrometer) instrument is designed and developed using GEANT4 simulations7'8'9'. The first simulations for the ion spectrometer have allowed computing the energy deposition in the last silicon SSD in function of the

431 energy deposition in the thick BGO detector. Results show that the vast majority of the protons can be separated from the electrons, and that with the BGO protons up to about 300 MeV can be uniquely identified; results also show that a proton depositing 80 MeV in the BGO yields primary and penetrating depositions of 1 MeV resolution in the fourth SSD. The initial simulations for the energetic neutron detector were compared to balloon flight observations. Results show that balloon flight data match GEANT4 simulations of incident neutron of E"3 spectra. Significant simulations are under way to test the instrument concept, the rejection efficiency requirements and the expected rates. 2. Planetary Scale Simulation 2.1. Validation We recall that the near-Earth regions are usually subdivided in three zones, defined as LEO (Low Earth Orbit, at about 400 Km X 51.6 deg), MEO (Medium Earth Orbit, at about 20200 Km X 55deg), and GEO (Geo-stationary orbit, at about 36000 Km X Odeg). LEO is the typical orbit of the International Space Station, MEO is the typical orbit of GPS satellites, and GEO orbits are of course also of wide interest for space missions. Thus LEO electron and proton spectra are particularly relevant for the ISS. At NASA they are based on the AE-8MAX electron integral and differential spectra, and on the AE-8MESI proton integral and differential spectra10). AE8 and AP8 models of Earth radiation belts have known limitations11,12,13^ (limited range of observations from the '60s, simplified account of 11-years solar cycle), thus making even more crucial a proper modeling of the radiation environment via simulations. A validation of GEANT4 dose rates has been conducted at Boeing. At GEO orbits, GEANT4 simulation results for electron dose rates have been validated for different Aluminum thicknesses, ranging from 0.01 to 10.0 g/cm2. At LEO orbits, GEANT4 simulation results for 200 MeV incident protons, as well as for 1.0 MeV incident electrons, have been validated onto equivalent Aluminum thicknesses14). 2.2. Modelling of the LEO radiation environment Theoretical modelings of the inner zone using GEANT4 simulations as inputs were conducted at The Aerospace Corporation and at CalTech15\ As sources, the Cosmic Ray Albedo Neutron Decay, and the cosmic ray protons modulated by solar activity (F10.7), were taken into account16,17'18*. The Geomagnetic access to the atmosphere was modeled using vertical rigidity cutoffs plus

432

transmission function to account for off-vertical variation. Mono-energetic protons have been simulated incident isotropically at an altitude of 200 Km. The atmosphere was modeled with average of NRLMSISE-00 N. O. and Ar densities over latitude, longitude, local time, and season, with typical geomagnetic conditions: F10.7 = 150, Ap = 419). GEANT4 physics list LHEP_PRECO_HP was selected. Neutrons exiting top of simulation were tabulated. Neutron production was convolved with incident modulated/cutoff proton spectrum. Results were shown at 41 deg geomagnetic latitude (4.83 GV), obtaining a vertical neutron flux (cm2 s MeV)"1 in agreement with observations20'2^, and especially good at the crucial high energies from 100 MeV to 8000 MeV. Future improvements foreseen for the LEO modeling include the addition of neutrons from cosmic ray He to the CRAND source, and the light-ions secondaries (deuterium, tritium, alphas, He3) that have been observed with AP8 inputs: both this improvements will be based on GEANT4 simulation. 3. Micro-dosimetry Simulation 3.1. Physics case for the RADSAFE concept Any CCD image of the sun (SOHO observatory, LASCO experiment) reveals the great activity of particles that can reach electronics components. From radio and microwaves (COBE, WMAP), to x-ray (CHANDRA, XMM) and gammarays (INTEGRAL, GLAST) astronomy, to gravity waves physics (LISA), electronics devices required for imaging need to be protected from particle upsets. Three sources in the natural space environment contribute to various effects on microelectronics22), solar particles (protons and heavier ions), galactic cosmic rays, trapped particles (in the Earth's belts). There are three basic effects that occur when components are exposed to these environments: Single Event Effects (SEEs)23'24), total ionizing dose25,26), and displacement damage27). It is noted that these effects can occur when components are exposed to any radiation environment; the space environment is just one of those environments. Single Event Upset (SEU) is one element of a set of SEEs that occur when components are exposed to a radiation environment, others are single event latch ups, transients, gate ruptures, and functional interrupts. Single-Event Upset (SEU) observations recorded by the ISS show clear patterns of temporal and special dependence (this is true for all space radiation effects). SEUs are mainly due to two ionization cases: direct (incident particle generates electron-hole

433

pairs) and indirect (secondary particles add to generated electron-hole pairs). In all cases, charge collected on a sensitive node of electrical circuit causes unwanted change in the information stored in the system. Commercial Technology Computer Aided Design (TCAD) tools are unable to predict indirect ionization case. Also, classical models have given unsatisfactory performance when addressing emerging technologies28'. Hence the requirements for the development of a tool capable of taking into account orbit information, interfaced to electronic engineering design tools (detailed device and circuit response and approximation models), and equipped with full microscopic physics modeling were set. The RADSAFE system is currently being developed by researchers at Vanderbilt University and is focused on addressing the farreaching needs of the spacecraft design community. Researchers at NASA, SLAC, and other government and industrial facilities are integrally involved in the development. The next two sections give an overview of the RADSAFE concept and applications. 3.2. The RADSAFE concept The RADSAFE system is based on advanced software design and technologies. RADSAFE is a revolutionary, modular computational system for predicting the radiation response of electronic devices, circuits, and systems from basic device structure, circuit topology, and fundamental radiation characteristics. Its goals to be informed by the best available physics, implemented by robust algorithms, enabled by supercomputer technology, and calibrated by data. It allows for input of geometry and material specifications from TCAD systems for electronics devices. Given any source of radiation environment, the energy deposition and partitioning are computed. This energy partitioning information is returned to the TCAD tools so that the effects of ionization on electrical parameters can be modeled, thus leading to the calculation of device and/or circuit performance degradation. At the core of the system lies the GEANT4-based MRED component (MonteCarlo Radiative Energy Deposition tool), developed at Vanderbilt University: it includes run-time selectable physics lists features and a Python interface for providing output with a great degree of flexibility. 3.3. RADSAFE applications We have demonstrated the RADSAFE concept on a simple Silicon diode, showing excellent agreement with expect values. A Si diode was reproduced as a 5-micron cube, then MRED was used to simulate energy deposition profiles for 1 million of 12C ions with 1 GeV per nucleon. A single energetic event was

434

selected and passed to the TCAD tool. The charge collected on the anode and cathode of the diode was comparable to what is expected assuming all the generated charge is collected. The RADSAFE concept has already been applied to investigate single event upsets in SRAMS and single event transients in IR focal plane arrays29,30'3''32'33).

4. Spin-off to Medicine The requirements for medical imaging, dosimetry, and treatment planning software have strong synergy with space simulation software, both in terms of relevant physics processes and in terms of energy ranges. Applications include external beam radiotherapy, beam-line design and quality assurance, absolute dosimetry, patient dose calculation, time-dependent simulations (to account for patient breathing phase), radiation risk and protection, brachyterapy, iso-dose distributions, DICOM interface and voxelized geometry compression, PET and SPECT design. In order to provide a reference point for the rapidly growing GEANT4 medical user community in North America, and to meet regional certification needs, the G4NAMU (GEANT4 North America Medical Users Organization) has been created in May 200534). It currently involves 27 institutes throughout Unites States and Canada, including Harvard, UCSF, U. Laval, CHU Quebec, Stanford, Louisiana U., McGill, Memorial Sloan-Kettering N.Y, TRIUMF, Johns Hopkins, U. Arkansas, Jefferson Lab, U. Wisconsin, U. Alberta, WRA Medical Center, Texas A&M, U. Washington, Hopital Maisonneuve-Rosemont, U. Minnesota and others. Conclusions We gave an overview of GEANT4 applications in NASA missions, particularly for apparatus simulation, planetary scale simulation, and micro-dosimetry simulation. Related application areas include spacecrafts charging, radiation shielding for spacecrafts, astronauts' safety, and bio-medical dose calculation. Comprehensive coverage of physics models and powerful geometrical modeling capability provided by GEANT4 were proven to be suitable for mission critical simulations.

435

References 1.

2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25.

L. Latronico, for the GLAST collaboration, "Como 2003, Astroparticle, particles and space physics, detectors and medical physics applications" 69 (2003) W. Atwood et. al. "Perugia 2004, Calorimetry in particle physics" 329 (2004) T.H. Burnett et. al., IEEE Trans. Nucl. Sci .49, 1898 (2002) D.K. Haggerty and S. Livi, Adv. Space Res. 33, 2303 (2004) D.K. Haggerty and E.C. Roelof, Adv. Space Res. 32, 423 (2003) D.K. Haggerty, private communication. J.D. Kinnison, R.H. Maurer, D.R. Roth and R.C. Haight, Radiation Research 159, 154(2003) R.H. Maurer, J.D. Kinnison and D.R. Roth, accepted for publication in Radiation Protection Dosimetry. R.H. Maurer, C. Zeitlin, D.K. Haggerty, D.R. Roth and J.O. Goldson, 2005 Nuclear Science Symposium, October 24-27, 2005. J.I. Vette, NSSDC/WDC-A-R&S 91-24, 1991. E.J. Daly, J. Lemaire, D. Heynderickx, and D.J. Rodgers, IEEE Transactions on Nuclear Science 43, 403-415 (1996). S.F. Fung, Radiation Belts: Models and Standards, AGU Geophysical Monograph 97, 79-91 (1996). S.L. Huston, G.A. Kuck, and K.A. Pfitzer, Radiation Belts: Models and Standards, AGU Geophysical Monograph 97, 119-122 (1996). R. Reddell and B. Atwell, SPENVIS & Geant4 Space Users' Workshop, October 3-7, 2005. M.D. Looper, R.S. Selesnick and R.A. Mewaldt, SPENVIS & Geant4 Space Users' Workshop, October 3-7, 2005. D. Heynderickx, M. Kruglanski, V. Pierrard, J. Lemaire, M. D. Looper and J. B. Blake, IEEE Trans. Nucl. Sci., 46, 1475, (1999) G. Kanbach, C. Reppin and V. Schonfelder, J. Geophys. Res.79, 5159, (1974) I.G. Usoskin, K. Alanko, K. Mursula and G. A. Kovaltsov, Sol. Phys., 207, 389, (2002) J.M. Picone, A.E. Hedin and D.P. Drob, J. Geophys. Res., 107, 1468, doi:10.1029/2002JA009430, (2002) A.M. Preszler, S. Moon and R. S. White, J. Geophys. Res., 81,4715, (1976) R.S. Selesnick and R. A. Mewaldt, J. Geophys. Res., 101, 19,745, (1996) J.L. Barth, C.S. Dyer and E.G. Stassinopoulos, IEEE Trans. Nucl. Sci.,50, 3, 466 (2003). P.E. Dodd and L.W. Massengill, IEEE Trans. Nucl. Sci., 50, 3, 583 (2003). P.E. Dodd and L.W. Massengill, IEEE Trans. Nucl. Sci., 50, 3, 603 (2003). T.R. Oldham and F.B. McLean, IEEE Trans. Nucl. Sci., 50, 3, 483 (2003).

436

26. R.L. Pease, IEEE Trans. Nucl. Sci., 50, 3, 539 (2003). 27. J.R. Srour, C.J. Marshall and P.W. Marshall, IEEE Trans. Nucl. Sci., 50, 3 653 (2003). 28. R.A. Reed, J. Kinnison, J.C. Pickel, S. Buchner, P.W. Marshall, S. Kniffin and K.A. LaBel, IEEE Trans. Nucl. Sci., 50, 3, 622 (2003). 29. K. M. Warren, et al., accepted for publication in IEEE Trans. Nucl. Sci., 2005 30. A. S. Kobayashi, et al., accepted for publication in IEEE Trans. Nucl. Sci., 2005 31. C. Howe, et al. accepted for publication in IEEE Trans. Nucl. Sci., 2005 32. J. Pickel, et al., accepted for publication in IEEE Trans. Nucl. Sci., 2005 33. D. Ball, et al., accepted for publication in IEEE Trans. Nucl. Sci., 2005 34. J. Perl, et al., http://geant4.slac.stanford.edu/g4namu/

GEANT4 APPLICATION TO HADRONTHERAPY IN JAPAN KATSUYA AMAKOt High Energy Accelerator Research Organization, KEK Institute of Particle and Nuclear Studies, 1-1 Oho, Tsukuba-shi, Ibaraki-ken 305-0801, Japan The Geant4 simulation toolkit provides a wide variety of physics models which describe interactions of elementary particles and ions with matter in a broad range of energy. In this article, an overview of an activity in Japan to apply the toolkit for simulation of hadrontherapy is given. Also an experiment to collect ion-matter interaction data applying the emulsion technique to validate Geant4 physics models is introduced.

1. Introduction Hadrontherapy is a radiation therapy technique for tumor treatment using hadron beams from accelerator machine. Historically this technique was proposed by R. Willison [1], who recognized the merit of applying proton and ion beams to radiotherapy. For details of hadrontherapy including historical overview, see the article by P. L. Petti and A. J. Lennox [2] and also U. Amaldi and G. Kraft [3] for more recent advance. According to the information provided by Particle Therapy Cooperative Group (PTCOG) [4], currently there are 27 hadrontherapy facilities (23 for proton beams and 4 for ion beams) in operation in the world and the number of patients treated so far by proton and ion beams is more than 46,000. There is a history of more than 20 years in Japan in the practical tumor treatment using proton and ion beams; Proton Medical Research Center in Tsukuba University (PMRC) started the first treatment in 1983 using the proton booster in High Energy Accelerator Research Center (KEK). Currently there are 5 proton therapy facilities in Japan as shown in the Fig. 1. For ion beam therapy, the HIMAC synchrotron (the maximum energy is 800 MeV/u) at National Institutes of Radiological Sciences (NIRS) in Chiba started its operation in 1994 and more than 1,800 patients have been treated so far using the 12C beam. Also Hyogo Heavy Ion Medical Center (HIBMC) in Hoyogo has a synchrotron machine, which can accelerate ions up to 320 MeV/u, This work has been partially supported by CREST of JST (Japan Science and Technology Agency).

437

438

for iontherapy (Fig. 1). Currently there are only four iontherapy facilities in the world and the half of them are in Japan.

Tha Energy Research Cenlar Wakaaa Bay (Touruoa 200 MaV)

U ofTsukubaPMRC (Taukiiba 250 MaV) Hyogo Ian Baam Madlcal Centar . (N,.b,-H.r1m. 32.M.V,,,, *

Nr

_ „__ _ . u ^ ^ ^ S - L " ' , " , ? ^ (Kashhva 235 MaV) • NIRS (Chlba proton 90 MaV, lona 400MeWu) Shbuoka Cancar Cantor (Mlshlma 230 MaV) ' 0

Proton beam

yf

Ion beam

Figure 1. Hadrontherapy facilities in Japan.

The choice of the 12C beam for iontherapy treatments in Japan comes from the study of several competing factors of the characteristics of various ion beams. To obtain a large Relative Biological Effectiveness (RBE) and a small Oxygen Enhancement Ratio (OER), a heavier ion is better. Also to get a well collimated beam a heavier element is better because the effect of multiple scattering is smaller. For a small dose at the entrance channel and a tolerable fragmentation production, a lighter ion is better. An optimal choice taking into account these competing factors is the 12C beam [5]. The clinical results from HIMAC support this choice. NIRS is now promoting a project for a compact carbon accelerator for iontherapy [6]. 2. Geant4 and Hadrontherapy in Japan As described in the previous section, the research community of hadrontherapy in Japan has more than 20 years of clinical experiences. Based on these experiences medical physicists committing to hadrontherapy in Japan want to have a simulation tool which can be used to study a performance of their treatment devices, to study a new design of treatment elements, and to evaluate the validity and performance of their treatment planning systems. Meanwhile, the Geant4 [7] development team in Japan is a founder of the Geant4 International Collaboration and has more than 10 years of experience in developing the Geant4 toolkit. The developers in the team want to see their

439 product used not only in the high energy physics domain (HEP) but also in medical, radiology, space and other domains. These requirements from the two completely difference research domains triggered in year 2002 a formation of a collaboration, which consists of Japanese researchers who belong to Geant4 Japan development team, medical physicists in the hadrontherapy facilities, physicists in the space science related institutes and high energy physicists in universities/institutes. This collaboration spawned two projects; a project to create a generic Geant4 toolkit for simulation of radiotherapy (the CREST project) and a project to perform a series of experiments to collect basic data of interactions of ions with material for validation of the Geant4 physics implementations (the PI52 experiment). 3. The CREST project The goal of this project is to develop a Geant4 simulation framework which can be generally utilized for researches in the radiation therapy domain. The activities included in the project are development of G4-DICOM interface, visualization and interactivity, parallel processing under GRID environment. Also included is the validation of Geant4 physics processes. The project is founded by the Core Research for Evolutional Science and Technology (CREST) program organized by Japan Science and Technology Agency (JST). Under this project there are 11 institutes from HEP, medical and space domains in Japan; High Energy Accelerator Research Organization (KEK), Ritsumeikan Univ., Kobe Univ., Naruto Univ. of Education, Toyama National College of Maritime Technology, National Institute of Radiological Science (NIRS) , National Cancer Center-Kashiwa, Gunma Univ., Faculty of Medicine Hyogo Ion Beam Medical Center (HIBMC), Kitasato Univ., and Japan Aerospace Exploration Agency (JAXA). The project started in 2003 and will continue till the end of 2008. For the details of this project, see the presentation given at the 2005 Geant4 User Conference [8]. 4. Critical Aspect in Applying Geant4 to Hadrontherapy One of the most critical requirements when we want to apply Geant4 to hadrontherapy (and also true for other radiation therapy) is the reproducibility of the depth dose distribution in human body with the accuracy of about 1mm. This requires Geant4 to provide a high-reliability description of not only electromagnetic processes but also hadronic processes and nuclear decay processes in the relevant energy regions and particle types. To attain this level

440

of accuracy, it requires extensive elaboration of G4 physics descriptions through validations by experiment data and tuning the program codes. In this respect the most critical aspect in the project is validation of Geant4 physics description. We are currently executing extensive studies of comparing available hadrontherapy related data with Geant4 simulation. One of such studies is the comparison with the data of projectile fragments' fluence and linear energy transfer from H 2 0 with various ion beams measured at HIMAC/NIRS [9]. It is still under the study but some preliminary results demonstrate that Geant4 reproduces the relative dose distribution including the Bragg peak of the 12C beam in H 2 0 quite well, though there are some discrepancies in the fluence distributions of projectile fragmentations [10]. Another study is the validation using the dose distribution data from the treatment beam line at Hyogo Ion Beam Medical Center HIBMC (NishiHarima/Japan). The data used for the validation are observed (spread-out) Bragg peak distributions with the proton beams in the water phantom. It demonstrates that Geant4 reproduces the data with the accuracy which nearly satisfies the medical requirement [11]. 5. The P152 Experiment Albeit its importance, a systematic accumulation of ion interaction data relevant to ion-therapy has not been established yet. This is probably due to the fact that there are not many therapeutic ion accelerator machines available in the world. Another reason might be the priority of usage of an ion accelerator machine - it is clear that treatment has much higher demand and priority than collection of experimental data. However, the shortage of reliable data causes a difficulty in Geant4 physics validation. The second project, spawned from the effort to establish a collaboration between the Geant4 development team and medical physicists, is to perform a series of experiments to collect basic data of interactions of ions with material for validation of the Geant4 physics implementations. The PI52 experiment was approved as a research project using the HIMAC/NIRS machine in 2003. The purpose of the experiment is to collect data of heavy ion interactions with H, C, O, Ca, P, etc in the energy region of iontherapy (several hundreds Mev/u). Also it is to collect data of heavy ion interactions which are important in the radiation shielding of space laboratories. The number of institutes joining the experiment is twelve; High Energy Accelerator Research Organization (KEK), Nagoya Univ., Ritsumeikan Univ., Kobe Univ., Naruto Univ. of Education, Toho Univ., Aichi Univ. of Education, Univ. of Tokyo, Stanford Linear Accelerator Center (SLAC),

441 National Institute of Radiological Science (MRS), Faculty of Medicine Gunma Univ. and Japan Aerospace Exploration Agency (JAXA). The key technology employed in this experiment is the emulsion chamber. The major reason we decided to use this technology is its excellent characteristics; it has 4K acceptance, a high spatial resolution (~l(J.m), capability of recording a very short track and charge identification. The detector has no active parts, and it is also easy to handle and very compact. Thanks to the development of the automatic scanning machine for the emulsion chamber by Nagoya Univ. [12, 13], we do not need to scan emulsion films for the track reconstruction with a manually handled microscope as we did in the old days; the scanning machine automatically does the all necessary works of track and vertex findings. Fig. 2 shows the detail of the emulsion chamber (water-emulsion chamber) which was used in the experiment last year. It consists of water layers sandwiched by emulsion films.

11

!•

i

Figure 2. The upper figure shows the overall view of the water emulsion chamber. It consists of multiple layers of water sandwiched by emulsion files. The lower figure shows the enlarged view of a sandwich of water with two emulsion films.

The water-emulsion chamber consists of 87 layers of emulsion films put in a water tank in an equal spacing. A single sheet of emulsion film consists of a film-base (TAC) and two 44um emulsion layers at the front and back of TAC. The ion beam was injected from the left in the Fig. 2. A typical density of beam particles injected to the chamber is the order of 10,000 particles/cm2. The allowable maximum density is limited by the performance of the automatic track reconstruction algorithm to separate one track from other. The automatic scanning machine reconstructs all tracks in a single emulsion film with the

442

speed of ~10 hours/plate. A reconstructed track has the spatial resolution of ~lum and the angular resolution of ~5mrad. The efficiency of finding a track is -100% (tan9<0.4). For charge identification of a track we use the technique so called 'refresh method'. This technique has been newly developed and established for the PI52 experiment. It is known that a film is put under the environment of a high temperature and humidity, a latent image center in the film is destroyed. This is sometimes called 'force fading'. The refresh technique applies this fact and forces fading a track image in the film by controlling temperature and humidity so that the saturation of the grain density by highly charged particle can be resolved [14]. Using this technique it has been demonstrated the charge identification up to Z=6 is possible for particles with (3y~0.8. So far we have performed about 10 times of beam exposures using emulsion chambers. The most of them are for the study of establishing the 'refresh method', though we also have taken data of interactions of 12C with H 2 0. A preliminary result of the measurement of the total charge-changing cross section of 12C with H 2 0 is already obtained and now its systematic errors are under the study. Also available are charge multiplicity distributions of secondary particles. These results will be published soon. 6. Summary Recently the Geant4 developers in Japan established with domestic medical physicists two projects. The first one is a project to create a generic Geant4 toolkit for simulation of radiotherapy (CREST Project). The second project is to perform a series of experiments to collect basic data of interactions of ions with material (P152 Experiment). The CREST project is to develop a G4-DICOM interface, visualization and interactivity, and parallel processing under GRID environment. Also included is the validation of Geant4 physics processes. Currently the Geant4 validation using existing data measured at NIRS is on-going. The PI52 experiment at NIRS aims to collect systematically data of heavy ion interactions with H, C, O, Ca, P, etc. in the energy region of ion therapy with the HEP emulsion chamber technology. The major reason for choosing this technology is that it provides a detector with excellent characteristics; a wide acceptance, a high spatial resolution, etc. Also the refresh method newly developed provides the capability of charge identification of ions. The basic technique to apply emulsion technology to measure ion interaction has been successfully established and some preliminary data of the total charge-changing

443

cross section of C with H 2 0 has been obtained; more data including the charge multiplicity of secondary particles, angular distributions, energy distributions, etc. will come soon. Acknowledgments I would like to thank to the following persons for providing information and data to prepare the presentation; T. Aso, N. Kanematsu, T. Koi, K. Murakami, T. Sasaki, T. Toshito and K. Yusa, References 1. RR. Willsion, "Radiological Use of Fast Proton", Radiology 47, 487 (1946). 2. P. L. Petti and A. J. Lennox, "Hadron Radiotherapy", Ann. Rev. Nucl. Part. Sci, 44, 155 (1994). 3. U. Amalidi and G. Kraft, "Radiotherapy with beams of carbon ions", Rep. Prog. Phys. 68, 1861 (2005). 4. Particle Therapy Cooperative Group: url; http://ptcog.mgh.harvard.edu/. 5. HIMAC/NIRS: url; http://www.nirs.go.jp/tiryou/himac/himac_t.htm. 6. K. Noda, T. Fujisawa, T. Furukawa, et al., European Particle Accelerator Conference 9, 2631 (2004). 7. S. Agostinelli, J. Allison, K. Amako, J. Apostolakis, H. Araujo, et al., "Geant4 - a simulation toolkit", Nucl. Instru. Methods A506, 250 (2003). 8. T. Sasaki, "Status and plan for the hadron therapy simulation project in Japan", Talk in Geant4 2005, 10th User Conference and Collaboration Workshop. 9. M. Komori, N. Matsufuji, T. Kanai, A. Fukumura, M.Hirai, et al., "Comprehensive Study of the Fluence and LET Distribution of Projectile Fragmentation Produced from Heavy Ion Therapeutic Beams", Submitted to Atomic Data and Nuclear Data Tables, Oct.,2003. 10. K. Yusa, Gumma University, Japan; Private communication. 11. K. Aso, A. Kimura, S. Tanaka, H. Yoshida, N. Kanematsu et al., "Verfication of the Dose Distributions with Geant4 Simulation for Proton Therapy", IEEE TNS 52, 896 (2005). 12. S. Aoki et al., Nucl. Instrum. Methods B51, 466 (1990). 13. K. Kodama et al., Nucl. Instrum. Methods A493, 45 (2002). 14. T. Toshito, K. Kodama, K. Yusa, M. Ozaki, K. Amako, et al., "Charge identification of highly ionizing particles in desensitized nuclear emulsion using high speed read-out system", Nucl. Instrum. Methods, the paper accepted in Nov., 2005.

AMS02 ITALIAN DATA T R A N S F E R SYSTEM: REAL LIFE EXPERIENCE

M. J. B O S C H I N I CILEA Via R. Sanzio, 4, Segrate, Mi-Italy E-mail: [email protected]

D. GRANDI, R. MANGONI AND P.G. RANCOITA INFN-Milano P.zza Scienza, 3, Milano, Italy

In this work we present "real life" experiences with the AMS Italian Ground Segment Data Transfer System. Its goal is moving data for the AMS-02 experiment, which will take data on board of the International Space Station, from the Central Facility, located elsewhere, to the Italian Ground Segment, which will act as Master Copy of the whole experiment data sample. The system has been in use since 2004 to transfer MonteCarlo data from various remote sites to AMS Science Operation Center, located at CERN. Furthermore the system is in use to transfer the whole MonteCarlo data sample from AMS Science Operation Center to the Italian Ground Segment. Data Transfer proves to be fast and reliable, achieving the goal of 3 times the expected DAQ rate and fully consistent book-keeping mechanisms. Technical implementation, experiences and "lessons learned" will be described and discussed.

1. Introduction Accurate measurements of cosmic ray spectra, particularly of protons, are important because of their close connection with the foundation of the theories of elementary particles, Grand Unified Theory and Dark Matter. The Alpha Magnetic Spectrometer AMS is a space-bourne experiment dedicated to investigate and measure with high precision such spectra, with particular attention to the detection of primordial anti-matter. AMS-01 was flown on the Space Shuttle Discovery on flight STS-91 in June 1998. This was primarily a test flight that enabled the AMS team to gather data on background source and adjust and verify the spectrometer performance for the

444

445

actual data taking, known also as AMS-02. The AMS-02 experiment, described in greater detail elsewhere1, is scheduled to start data taking by the second half of year 2008 on board of the International Space Station Alpha (ISSA) and continue to operate through at least 2010. 2. Data Flow The data flow, which is described in greater detail elsewhere2, can be summarized as follows: data will be recorded on board of International Space Station (ISS) and transmitted initially to Marshall Space Flight Center (MSFC) Alabama and then to AMS ground centers: Payload Operation and Control Center and Science Operation Center. From there, data will be sent over the Internet to the Science Operation Center and Central Production Facility (CPF), located at CERN, where it will be processed and analyzed. Furthermore, a Master Copy of the whole data sample, both raw, reconstructed and simulated, will be stored at the Italian Ground Segment Data Storage (IGSDS). The IGSDS will act not only as a master copy, but also as a " distribution center", in order to allow fast access to data to all collaborators without interfering with the CPF activity. The expected Data Sample for the ISS Data taking is summarized in the following table:

Table 1.

AMS02 related data transmitted to Earth - NASA-POCC

Stream

BandWidth (Avg.)

High Rate Slow Rate NASA

Data Category

Volume (GB/day)

Volume (TB/year)

3-4 Mbit/s

Scientific

30-43

11-15

16 kbit/s

House Keeping

0.16

0.06

Auxiliary Data

0.027

0.01

Facing the long running time and the way how the data will be transmitted from the detector to the analysis and production facilities, the development of a secure and reliable way of data transferring and organization is one of ground data handling vital issues. In particular, consistency between the CPF and the IGSDS data samples and data catalogs is of prominent importance. 3. AMS02 Italian Data Transfer One major component of AMS Ground Data Handling system will be the Data Transfer (DT). Its goal will be moving data efficiently from CPF to IGSDS without interfering with CPF's activities.

446

The main goal of the Data Transfer System is to move data from CPF to IGSDS and keep track of what has been moved and if successfully or not. There are thus 2 components, which, for the sake of portability, we addressed separately and then merged together in the general architecture: • data transfer protocol (DTP): part of DT in charge of actually moving the data. • book-keeping mechanism (BK): part of D T in charge of keeping track of what has been moved and how (successfully or not). The "natural" solution to integrate the two parts is a client/server architecture: move data from CPF to IGSDS , keep track at CPF of what/how was moved, notify IGSDS about moving, IGSDS receives data, IGSDS keeps track of what/how has been received, and so on... This approach decouples the technical solutions adopted for the 2 components, making the whole system more portable and maintainable. BK has been taken care of by means of a Relational Data Base. For the whole system we adopted OpenSource solutions, which are: Linux O.S., MySQL5 for the RDBMS, Perl5 and C as programming languages*, OpenSSL3 for encrypted socket communication and bbftp 4 for the DTP. Inconsistency of the book-keeping can be divided in two components: first-level inconsistency, which occurs whenever a file does not appear in one of the two databases 5 . The AMS02 Data Transfer System provides a mechanism that checks for first-level inconsistencies and triggers a retransmission of the file. These inconsistencies are mainly due to network outages or hardware hangs at one of the sites. Any inconsistency that still remains, and which has to be fixed by human intervention, is defined as second-level inconsistency. The general requirements can be summarized as follows: low CPU load, sustain 3 times max data-rate, both in terms of bandwidth and of number of files, optimize bandwidth usage and reduce second-level inconsistency. Thorough tests, described elsewhere6, showed that the system satisfies the requirements and is suited for use in a production environment. In particular, it has been put in use since February 2004 to transfer MonteCarlo data from various remote locations to CPF . Furthermore, in 2005 it a Perl is used for the main programs, while C is used for the core of the DB communication system. b There is a DB at the sending site, keeping track of what and how has been transferred, and one at the receiving site, keeping track of what and how has arrived.

447

has been used to transfer the whole MonteCarlo data sample from CPF to IGSDS. 4. Real Life Experience As already mentioned, the system is in production since February 2004 to transfer MonteCarlo data from various remote locations to CPF. Locations, scattered across Europe, have used the AMS-02 Data Transfer System to transfer, totally, 3.3 Tbyte of data, split into 5200 files. The system's architecture implements configuration in such a way that TCP (mainly WIN.SIZE) and bbftp parameters are automatically configured for the location, and dinamically modified, according to run-time environment. This choice proved to be an essential issue in order to provide maximum effective throughput. This because the remote sites had very different networking and storage capabilities, and fixed parameters (as demonstrated during the tests 6 ) would have degraded performance of the single remote site and lower the level of consistency of the book-keeping systems. In fact, total first-level inconsistencies, obtained by summing the inconsistencies of all sites, is of 0.02%, meaning that only 10 files required human intervention. Nevertheless, this number, as will be shown later on, could be drastically reduced if remote sites provide some administrative privilege over storage and network configuration and maintenance. Altough very useful, both as a testbench of the system itself and as a tool for the AMS collaboration, this layout0 is not considered fully compliant with the original system design, which is transfering data from the CPF to the IGSDS d . The latter layout has been used in 2005 to transfer the whole 2005 MonteCarlo production, stored at the CPF , to the IGSDS . This first unicast implementation, besides the obvious goal of making a master copy of all data, had also the purpose of testing the system's integration with offline storage systems like CASTOR - CERN Advanced STORage Manager7, deployed at both locations. Main goal was measuring how well the throughput requirement of 3 times the maximum expected data rate (which, for the time being has been set to 12 Mbit/sec) e has been fulfilled. The nominal bandwidth of the CPF to IGSDS link is 100 Mbit/sec, but since it is not a dedicated link, a meac

Sometimes referred to as multicast. Sometimes referred to as unicast. e Which means that the throughput requirement is 36 Mbit/sec. d

448

surement of the available bandwidth of the link was necessary. We relied on pathloariP to perform such measurements. In fact, if the available bandwidth of a network path is defined as the maximum throughput that can be provided to a flow, without reducing the throughput of the cross traffic, one needs a non-intrusive tool 8 to measure it, and pathload is such a tool. We thus deployed this technique, after a first period of tuning, measuring pathload's estimate of the available bandwidth every three transfers. Differences between this value and the effective DT throughput indicate an inefficient use of the network link, due to bad configuration of DT's parameter, or eccessive overhead introduced by some component of the DT (main candidates beeing the book-keeping, data validation and interaction with off-line storage like CASTOR). The following plot shows the effective DT throughput and pathload's measurements for a subset of the data sample transfer. 1

1

,

x

,

"thrdughput" — i — "pathload" — x — -

?

„

\ h

+

]

i| JS l:

TM

j 111 *i

Sk

r

X

*

ff

irl 3CT

1 T

'

"

ft

_

J

y

\J k1 I k

H

*

I

: 400

420

i

1i

440

—i—i

460

' 480

500

Data set

Figure 1. Effective DT Throughput and pathload's measurements for a subset of the data sample transfer.

The average difference, calculated on all mesurements g , is of 7%, indif

Defined as total transferred data / RealElapsedTime of the whole DT + Book-keeping mechanism. g In some cases, pathload's measurement is lower than the correspondig DT value. This

449 eating an optimal bandwidth usage. Average I T effective throughput is 45 Mbit/sec. We also evaluated the throughput of saving data to the off-line storage system, which we performed using CASTOR's rfio APIs; this never proved to be a problem, being the average throughput approximately 80 Mbit/sec, thus ruling CASTOR out as a possible bottleneck. A second goal was measuring the second-level inconsistency in a fullycontrolled layout: the full data transfer had a second-level inconsistency of 0.01%, which fully satisfies the DT systems requirements. Finally, the overall Real Elapsed Time has been analyzed. The following plot shows the incremental time line of the transfered data sample, and different plateaus are evident. wuu

0

H^Wf**^ 20

T

r

1

1 40

1 1 60 80 Day of Run

1

r

1 100

1 120

1 140

Figure 2. Incremental time line of the transfered data sample.

These inactive periods are due to major hardware downtimes at CPF or IGSDS , which were still in a setup phase. The overall throughput, including these periods, is around 6 Mbit/sec, which is too low with respect to the requirement of 36 Mbit/sec, especially if we compare it with the average effective throughput of 45 Mbit/sec. The overall throughput, calculated including only 5% of downtime periods, which we believe can can be due to network glitches or some other application heavily using the link.

450

be considered a good estimate of a "natural" performance of such sites, is around 40 Mbit/sec, perfectly consistent with the requirements. 5. Conclusions We presented experiences and results of the AMS02 Italian Data Transfer System during nearly two years of full operation. In particular, we showed that the DT system is able to achieve an effective throghput that maximizes the bandwidth usage, thanks to the adaptive TCP and bbftp run-time tuning. This leads to an effective throughput of 45 Mbit/sec on a nominal 100 Mbit/sec link, which fully satisfies the initial requirements. The system also shows a very low second-level inconsistency when there is administrative control at both sites. Finally, the overall throughput shows a high dependence on the receiving and/or sending sites reliability: if we assume a "natural" downtime of 5%, an overall throughput of at least 40 Mbit/sec is expected, fully satisfying the requirements. These results clearly show that the AMS02 Italian Data Transfer System is reliable and efficient, and can thus be adopted by the AMS02 experiment. Acknowledgment s We wish to aknowledge the continuous support received by CILEA and its director, Prof. A. Cantore. References 1. AMS Collaboration Physics Reports, 366/6, 331 (2002) 2. V. Choutko, A. Klimentov Proceedings of Computing in High Energy Physics (CHEP 2001), Sep. 2001. 3. 4. 5.

http://www.openssl.org http://doc.in2p3.fr/bbftp/ http://www.mysql.com

6. M. J. Boschini et al, Proceedings of 8th ICATPP Conference (2004). 7.

http://castor.web.cern.ch/castor/

8. M. Jain, C. Dovrolis, Proceedings of Passive and Active Measurement Conference (2002). 9.

http://www.cc.gatech.edu/fac/Constantinos.Dovrolis/pathload.html

APPLICATION OF GEANT4 IN THE DEVELOPMENT OF N E W RADIATION THERAPY TREATMENT METHODS

ANDERS BRAHME, IRENA GUDOWSKA, SUSANNE LARSSON, BJORN ANDREASSEN, RICKARD HOLMBERG, ROGER SVENSSON Karolinska Institutet and Stockholm University Stockholm, S-J71 76, Sweden VLADIMIR IVANCHENKO Budker Institute for Nuclear Physics Novosibirsk, 630090, Russia ALEXANDER BAGULYA, VLADIMIR GRICHINE, NIKOLAY STARKOV Lebedev Physical Institute Moscow, 119991, Russia

There is a very fast development of new radiation treatment methods today, from advanced use of intensity modulated photon and electron beams to light ion therapy with narrow scanned beam based treatment units. Accurate radiation transport calculations are a key requisite for these developments where Geant4 is a very useful Monte Carlo code for accurate design of new treatment units. Today we cannot only image the tumor by PET-CT imaging before the treatment but also determine the tumor sensitivity to radiation and even measure in vivo the delivered absorbed dose in three dimensions in the patient. With such methods accurate Monte Carlo calculations will make radiation therapy an almost exact science where the curative doses can be calculated based on patient individual response data. In the present study results from the application of Geant4 are discussed and the comparisons between Geant4 and experimental and other Monte Carlo data are presented. 1.

Introduction

The increasing cancer incidence demands more advanced and cost effective treatment techniques in busy radiotherapy departments. Accurate, safe and fast dose delivery is required to shorten the treatment time at the same time, as the treatment outcome should be maximized. A high accuracy and flexibility in dose delivery is therefore of major importance in the clinical application. A detailed spatial and energy distribution of the narrow photon beam generated by the almost point monodirectional and monoenergetic electrons incident on the target

451

452

can be calculated for the complex therapy head configuration of a modern medical accelerator, which uses advanced field shaping techniques with multileaf collimators and electromagnetic scanned beams and purging magnets to remove primary and secondary electrons as shown in Fig. 1 [1], [2]. This information is of particular importance both for accurate treatment planning as well as for intensity and energy modulation of narrow elementary beams operating in scanning mode to deliver almost any desirable intensity distribution in the target volume [3].

Figure 1. Exploded view of the Treatment Head of an advanced scanned electron and photon beam accelerator. The scanning magnets in the bending plane and across it can scan both the photon and electron beam over the patient to deliver arbitrary intensity modulated beams.

Monte Carlo (MC) simulation of therapy beams allows accurate prediction of the radiation dose distribution delivered to a patient. An optimization of the

453

therapy beam delivered by different treatment head configurations can be preformed with this type of simulation. Additionally, the MC technique represents a powerful tool to quantify possible sources of scattered dose and radiation transport in the heterogeneous tissues of the patient. In this study the result of such MC simulations are discussed. The simulations have been done using the object-oriented Monte Carlo simulation toolkit Geant4 [4]. Geant4 is a general toolkit for simulation of particle transport through matter, originally designed for applications in high energy physics. Through the years Geant4 has been extended, redefined and new physics models have been developed for applications in different fields like particle and nuclear as well as medical, accelerator and space physics [5]-[7]. However, in some areas validation of the code for transport of particles used in radiation therapy is still required and this is the subject of the present study. The electron-photon transport through a treatment head has been studied for the therapeutic photon beam of energy 50 MV from the MM50 racetrack microtron at the Karolinska Hospital, Stockholm, as shown in the exploded view of Fig. 1. Light ion beam transport in tissue equivalent media was simulated for ions of Z = 1-6 and energy range up to 400 MeV/u. The Geant4 results are compared with experimental results from the radiation therapy facilities in Sweden, Germany and Japan and other Monte Carlo calculations. 2. Radiation Transport Calculations for scanned Electron and Photon Beam Therapy A new thin transmission target technique for dose delivery using narrow scanned photon beams has been developed [1], [2], [8], [9]. The aim of this technique is to allow very fast intensity modulated photon dose delivery especially useful for radiobiological optimized radiation therapy. High-energy, 50-70 MeV, electron beams of low emittance incident on thin low-Z targets produce very narrow intense high-energy bremsstrahlung beams. However, the electrons transmitted through the target have to be removed from the therapeutic field with a purging magnet and an electron collector.

The 50 MV racetrack microtron (MM50) installed at the Karolinska Hospital, Stockholm, delivers electron and photon beams with maximum energies in the range 2-52 MeV. Two different composite targets for bremsstrahlung production were used: one almost fully absorbing target made of 9 mm beryllium plus 6 mm tungsten (9Be6W) and one thin transmission target made of 3 mm beryllium

454

(3Be). The radial dose distribution of the 50 MV photon beam have been measured. The details of the experiment and simulation are described in Refs. [13], [14]. A special Geant4 application called GammaTherapy was designed for the simulation of the experiment, which is now a part of the Geant4 toolkit as an example of medical applications.

Bremsstrahlung thin Be target

10

ofSOMeVe-

i i i i i i i

25

50

75

100

125

150

175

200

radius(mm) Figure 2. Radial energy deposition normalized to the integral in a water phantom 1000 mm from the 3Be target experimental data (dots) and Geant4 calculations: solid histogram - the Standard subpackage, dashed line - the Low energy subpackage. It is seen that most of the bremsstrahlung photons are emitted within a few degrees from the incident electrons.

A comparison was made between different Geant4 models for electronphoton interactions both for thin and for thick target. The models used in this comparison are the Standard [11] and Low energy [12] implementations of the electromagnetic processes. The results show that the radial dose profiles inside the phantom calculated by these two models are similar. For the case of the 3 mm Be target, these predictions are in good agreement with the measurements (Fig, 2). The result is shown on a logarithmic scale to demonstrate validity of Geant4 predictions over three orders of magnitude. Below this level, the experimental uncertainty becomes significant but the results are still realistic. For

455 the case of the 9Be6W target, the agreement between simulation and the experimental data is also good [10], [14]. However, some difference exists at small radii, which was also observed in comparisons with MCNP4B [13]. The results of this work confirm that Geant4 is quite accurate and can be used to accurately simulate also very extreme photon therapy beams. Figure 2 clearly shows that it is possible to make very narrow bremsstrahlung beams with high energy electrons and thin transmission targets. By increasing the energy to about 70 MeV and treating the patient at 750 mm from the target the half width of the photon beam is decreased to about 12 mm, which is very useful for Intensity Modulated Radiation Therapy (IMRT). 2.2. Simulation of Electron Collection with the purging magnet In the thin transmission target technique only a fraction of the electron energy is converted to the narrow useful photon beam. Electrons transmitted through the target have to be bent away from the therapeutic beam by a strong purging magnet and effectively absorbed in an electron collector. The choice of collector material is essential to achieve maximum absorption of the transmitted electrons and minimal secondary bremsstrahlung contamination of the therapeutic beam. A crude design of the collector is shown in Figs. 1 and 3. It is made of a 9 cm thick block of tungsten, and it is placed immediately below the purging magnet, displaced from the central axis in y-direction since the purging magnet deflects the electron beam in this direction. The thickness of the collector in the y direction is varied to study leakage of bremsstrahlung photons. In the simulations with three different collector designs most of the electrons are stopped in the collector only leaving a few low energy electrons scattered mainly in the opposite direction to the patient [14]. Geant4 is a powerful toolkit that allows detailed study of complex beam configuration with different absorber geometry, materials and magnetic field distributions. Also the fluence and dose contribution due to scattered protons can be studied in order to minimize their contamination at the location of the patient.

456

Figure 3. Cross-section through the purging magnet and electron collector of Figure 1 showing the influence of the magnetic field on the primary electrons (left panel). In the right panel a large number of forward bremsstrahlung photons as well as back scattered photons from the electron collector are seen. The electron beam is deviated in positive y-direction hitting the collector from above, about 2 cm from the edge. The incident mean electron energy is close to 50 MeV.

2.3. In Vivo Dose Delivery verification using Photonuclear reactions inside the body The new thin transmission target technique [14] with narrow high-energy bremsstrahlung beams of high intensity in the forward direction is also suitable for 3D in vivo verification of dose delivery using PET-CT imaging of the induced /f activity [1], [2]. This approach is based on photonuclear reactions in the human tissues to convert normal 12C and 16 0 to n C and 15 0 by knocking out neutrons with the high energy photons. The generated PET emitters that are

457

ideally suited for in Vivo dose delivery verification, which is an important clinically task that never before has been possible by non invasive methods.

Figure 4. This figure illustrates the activity distribution induced by the narrow photon pencil beam in a water phantom. The PET activity distribution (/T(z)), the absorbed dose (D(z)), the fluence (<2>(z)) and the energy fluence (Viz)) in the water phantom are shown. The good agreement between b* activity absorbed dose and the energy fluence is particularly clear. The fluence profile is broader because of the background of low energy photons, which does not contribute much to the energy fluence.

Today we are using Geant4 with an extensive photonuclear cross section library to calculate the density of positron emitters in the irradiated medium or patient. Figure 4 illustrates the photonuclear PET emitter activity due to the narrow photon pencil beam in Figure 2. Interestingly, this distribution will be spread out over the whole tumor volume during therapy (cf. Fig. 5) so all regions being irradiated can be accurately traced by PET-CT imaging in relation to the normal tissue anatomy depicted by the CT part of the PET-CT camera. A broad thick target bremsstrahlung beam would not be so useful here since the mean photon energy will vary considerably over the patient and make it very difficult to interpret the true dose delivery in the patient. In Figure 5 it is seen that if the broad scanned photon beam has a uniform dose (upper right panel) both the activity distribution and energy fluence are over compensated, whereas the low energy photon fluence is strongly forward directed

458

Figure 5. Same as Fig. 4 but using scanned photon beams so that the absorbed dose is closely uniform. The intensity or energy fluence (Viz)) is then slightly over compensated and so is the PET activity.

3. Simulation of Light Ion Therapy Beams To be able to treat well also the most severe radiation resistant tumors of complex local spread, intensity modulated ion beams combined with biologically based therapy optimization methods [15] are under development and belong to the most advanced treatment techniques that are possible. The development is carried out at the Karolinska Institute and Hospital [16]. The ultimate step in the therapy development process is therefore to optimize the beam quality for the relevant tumor and normal tissue situation. To facilitate beam quality modulation and treatment optimization light ions are really needed and the Monte Carlo method is a very important tool to develop this approach. The Geant4 code is today very suitable for calculations of several important problems relating to the use of particle beam transport for new treatment facilities.

459

dE " dz 0.5

Carbon ions 195MeV/u in water

i EXP GSI Kramer et al 2000 |GEANT4 ver7.1 Bin Case. Stand EM

0.4 SHIELD-HIT veMl'l

0.3

0.2

0.1

0.0' "0

, • — ^ '

5

10

Depth/cm

15

Figure 6. Comparison of the central axis depth absorbed dose distributions in a carbon ion beam, measured in a water phantom [19], calculated by Geant4 and by the dedicated light ion code SHIELD-HIT [20]. A generally good agreement is seen between all the three data sets.

As a first step of this work the comparison between Geant4 calculations and the experimental data for different ion beams has been performed [17]. Figure 6 shows the results of simulation for carbon beam of 195 MeV/u from the carbon therapy facility at GSI, Germany [19]. It is seen that Geant4 calculations reproduce well the experimental depth-dose distribution. The calculated position of the Bragg peak is about 2 mm shallower, which indicates the necessity of implementation of more recent stopping power values such as those from ICRU 73 (2005) in the Geant4 code. On the other hand the Geant4 Binary Cascade model [18] well describes the fragmentation tail of the carbon ion beam. Conclusions The possibility to combine advanced Monte Carlo calculations with the influence of strong electromagnetic fields makes Geant4 a very flexible transport code far beyond standard MC programs. This property has been essential in the present study where the magnetic field of the purging magnet

460

severely affects the multiple scatter of the electrons in the air. A good agreement between Geant4 and experimental data was seen for all cases studied. The very narrow bremsstrahlung beam is indeed a very challenging test where experimental data are well described almost over 3 orders of magnitude. Also Geant4 shows good performance for carbon ion transport at least in the energy range up to 200 MeV/u. Acknowledgements This work was supported by VINNOVA and the EU project IOTAS (grant INTAS-2001-0323). References 1. A. Brahme, In: Radiation Therapy Physics, Ed. Smith A., Berlin Springer, 209 (1995). 2. A. Brahme, Acta Onco.l 39, 579 (2000). 3. A. Brahme, Acta Oncol. 42, 123 (2003). 4. Geant4 Collaboration (S. Agostinelli et al), Nucl. Instr. Meth. A506, 250 (2003). 5. M. Ellis, Talk at this conference. 6. G. Santin, Talk at this conference. 7. M. G. Pia, Talk at this conference. 8. A. Brahme, T. Kraepelien and R. Svensson, Acta Radiol. Oncol. 19, 305 (1980). 9. A. Brahme, Acta Oncol. 26,403 (1987). 10. I. Gudowska, V. N. Ivanchenko, S. Larsson and B. Sorcini, IEEE Nucl. Sci. Symp. Conf. Record*, 2148 (2004). 11. H. Burkhardtetal,/£££'7Vwc/. Sci. Symp. Conf. Record!, 1907 (2004). 12. S. Chauvie et al, IEEE Nucl. Sci. Symp. Conf. Record 3, 1881 (2004). 13. I. Gudowska, B. Sorcini, and R. Svensson, In: Proc. of MC 2000 Conf., Lisbon, 23-26 October, 2000, p 317. 14. S. Larsson et al, In: ICRS 10 /RPS 2004 Conference, Madeira Island, Portugal, 9-14 May 2004, accepted to Radiat. Prot. Dosim. (2005). 15. A. Brahme, Int. J. Rad. Oncol. Biol. Phys. 58, 603 (2004). 16. H. Svensson, U. Ringborg, I. Naslund and A. Brahme, Radiother. Oncol, Suppl. 2, 73 (2004).

461 17. I. Gudowska, A. Bagulya, V. Ivanchenko and N. Starkov, In: Geant4 10 User Conference and Collaborative Workshop, 3 - 1 0 Nov., Bordeaux, France, 28 (2005). 18. G. Folger, V. N. Ivanchenko and J. P. Wellisch, Eur. Phys. J. A21, 407 (2005). 19. M. Kramer et al, Phys. Med. Biol. 45, 3299 (2000). 20. I. Gudowska, N.M. Sobolevsky, P. Andreo, D. Belkic, and A. Brahme, Phys. Med. Biol. 49, 1933 (2004)

G4 ACCELERATOR APPLICATIONS MALCOLM ELLIS High Energy Physics Group,Imperial

College London, South Kensington SW7 2AZ, UK

GRAHAME BLAIR* Dept of Physics, Royal Holloway Univ of London, Egham, Surrey TW20 OEX, UK YAGMUR TORUNt Illinois Institute of Technology, 3300 South Federal St, Chicago, IL 60616, USA TOM ROBERTS t t Muonslnc,

Batavia, IL 60510, USA

This paper will review GEANT4 applications and extensions for accelerator studies, both as a general framework and for the simulation of specific accelerators. G4Beamline is a Geant4-based program intended to easily simulate many different aspects of beam-line design. G4Beamline implements accelerator components via an object-oriented ASCII input format. For instance, it allows the definition of a quadrupole magnet by its physical geometry, materials, and also magnetic field (including several models for fringe fields). The Muon Ionization Cooling Experiment (MICE) poses some challenges for simulation not normally found in a particle physics experiment. The G4MICE package is under development to allow a complete simulation of the MICE cooling channel (including the simulation of magnetic and RF fields) and instrumentation. BDSIM is a Geant4 based code that implements fast accelerator-style tracking for particles within the beam-pipe together with the full set of Geant4 physics processes for particles that leave the beam-pipe. The program has been used to simulate and optimize the design of proposed electron positron linear colliders (ILC and CLIC). The input to the code is based on the MAD format and allows definition of geometries, fields and materials. Geometry descriptions common to those used by the ILC detector groups have

This work is supported in part by the Commission of the European Communities under the 6th Framework Programme "Structuring the European Research Area", contract number RIDS011899 and by the British Council Alliance programme. * Work supported by the U.S. Department of Energy and the Minoise Board of Higher Education. " Work supported by the U.S. Department of Energy.

462

463 enabled common studies of the interaction region, specification of the required collimation depths and optimization of the extraction lines.

1. Introduction It has been common to separate the simulation and optimisation of an accelerator system from that of the detectors that use the supplied beam. Research and development towards future accelerator systems (the International Linear Collider and Neutrino Factory) has resulted in the development of Geant4-based software packages that provide tools to accurately simulate the accelerator system as well as the detectors. These software tools can be used for some of the design and optimization tasks normally performed by dedicated fast accelerator simulation packages, which typically have a minimal set of material description and physics processes. These tools can also be used to simulate the interaction and passage of the beam through the relevant detector system in the more conventional manner. The merging of these abilities to design, model and simulate a beamline and detector system in one consistent package can be particularly beneficial when there is a very strong coupling between the two. 2. G4Beamline G4beamline [1] is a general-purpose simulation program based on the Geant4 [2] toolkit that is specifically optimized for the simulation of beam lines and other systems. By implementing a large repertoire of beamline elements it relieves the user of C" programming, but permits the realistic simulation of many interesting systems. G4beamline is open source and easily extensible, and users can add their own elements when necessary, via C* programming. G4beamline uses a single ASCII input file to define all aspects of the simulation, including the geometry, beam definition, physics use-case, program control parameters, output NTuples, and visualisation. As it is inherently object oriented, and beam elements include both physical materials and electromagnetic fields, describing the system is usually a matter of selecting the objects to use, setting their parameters, and placing them into the beamline. Geant4's exampleN02 consists of ~900 lines of C ^ code; Figure 1 shows it described in 17 lines of G4beamline input. This enormous reduction in size and complexity is common, as unlike C ^ code the G4beamline input is no more complex than the system being simulated. Note this includes visualisation using most visualisation drivers supported by Geant4, plus histograms using HistoScope [4]; additional histogram packages are planned.

464

G4beamline implements "centerline coordinates" that easily permit the use of bending magnets. Figure 2 shows the MICE muon beamline in G4beamline. As it is designed for flexibility, G4beamline has been used for simulating many different systems. An example is the helical cooling channel summarised in Figure 4. The meaning of "beamline" is quite flexible, and G4beamline includes a cosmic-ray "beam" used in the simulation of muon tomography of a shipping container shown in Figure 3. # select physics use-case, define materials and beam physics QGSP material Pb A=207.19 Z=82 density=l 1.35 material Xe A=131.29 Z=54 density=0.005458 beam gaussian beamZ=-200 sigmaX=0 sigmaY=0 sigmaXp=0 sigmaYp=0

\ meanMomentum=3000 sigmaP=0 particle=proton nEvents=5 # define elements of the simulation box Target width=50 height=50 length=50 material=Pb color=l ,0,0 virtualdetector Detl width=480 height=480 length=100material=Xe virtualdetector Det2 width=1344 height=1344 length=100 material=Xe virtualdetector Det3 width=2208 height=2208 length=100 material=Xe virtualdetector Det4 width=3072 height=3072 length=100 material=Xe virtualdetector Det5 width=3936 height=3936 length=100 material=Xe # place elements into the simulation "World" volume (which expands) place Target z=0 place Detl z=800 place Det2z=l 600 place Det3 z=2400 nlace Det4 7^3200 Figure 1. ExampleN02 in G4beamline, units are mm and MeV.

.

\

• ^ 4 ,

Figure 2. The MICE Beamline, visualised using Openlnventor.

Figure 3. Cosmic-Ray exposure of a shipping container.

465 Evolution of Beam emittance ICOOL

0

10

20

*30

40

50

60

70

W""° ^T^lOO

0

LO

20

:J0

40

50

60

70

SO

90

100

Z(m)

Figure 4. Emittance reduction (cooling) along a helical cooling channel.

3. G4MICE G4MICE is the Geant4-based software suite that provides Simulation,. Reconstruction and Analysis tools for the MICE experiment [3]. The MICE experiment requires the construction, operation and measurement of the performance of an ionisation cooling channel. The simulation of MICE therefore requires a combination of accelerator components (201 MHz RF cavities, liquid hydrogen absorbers, quadrupole triplets and solenoid magnets) and instrumentation (Scintillating Fibre tracker, Time of Flight and Cherenkov detectors and a Calorimeter), shown in Figure 5. G4MICE provides an accurate model of the components of the downstream section of the muon beam line as well as the full cooling channel and instrumentation. Magnetic fields (including fringe fields) can be calculated directly from magnet currents (supplied as input parameters) or read in directly from an external file. An important aspect of particle tracking through the channel is the correct simulation of the time varying EM field associated with the 201 MHz RF cavities. The simulation of the production of X-rays and electrons by the high accelerating gradient of the RF cavities has also been implemented and was crucial to the validation of the design of the tracking detector. The process of field emission from RF cavities has been studied with a high frequency cavity and the results extrapolated to the MICE cavities. The resulting extrapolation was used to create a spectrum of particles at the RF cavities which are then propagated through the experiment by GEANT4. Figure 6 shows the shower of

466

Figure 5: Muon Ionisation Cooling Experiment (MICE). Instrumentation and cooling channel comprising eight 201 MHZ RF cavities and three Liquid Hydrogen absorbers.

Figure 6: Visualisation of the passage of one muon through the MICE experiment with the production of X-rays and electrons by the RF cavities.

X-rays and electrons produced by two RF cavities overlaid with a single muon passing through the channel. The simulation of the tracking and particle identification detectors in G4MICE is as detailed as can be reasonably achieved. An example is the scintillating fibre tracker model, which includes the active core and two layers of cladding for each fibre. Each fibre is explicitly represented and is contained inside a layer of glue and supported by a carbon fibre frame. The digitisation parameters have been taken from previous experience and updated in light of the results of test-beam experience as part of the ongoing research and development efforts for the experiment. G4MICE has been used to simulate the performance of the full experiment with a realistic set of accelerator components and a range of input beams. Ten

467

different beams (each matched to the cooling channel and of a given transverse emittance) were simulated passing through the experiment. The digitised events were reconstructed and analysed using one potential analysis technique under study. The results show the expected performance of a transverse cooling channel and are summarised in Figure 7. Cooling Measurement

Cooling Measurement Resolution

x Truth EmJttanca Difference + Reconstructed Emit-tance Dtlference

Initial Enlluiice (pi ran rail)

Generated Eialttanee (pi n

Figure 7: Left: The reconstructed (+) and simulated (x) change in emittance as a function of the input beam emittance. Cold beams are seen to be heated (positive change) and hot beams are seen to be cooled (negative change). Right: The difference between the reconstructed and simulated change in emittance is plotted as a function of the input beam emittance.

4. BDSIM BDSIM is a Geant4-based program created originally to simulate the beam delivery system (BDS) of the International Linear Collider (ILC), with early application to the CLIC [5]. However BDSIM is designed flexibly in order to track particles through any beam-line and includes particle interactions and production of secondary particles in materials. The accelerator input to the program is controlled by a file that is based on the MAD format, but also includes additional descriptions of geometries, apertures and materials [6]. The code is maintained under CVS and is available freely [7]. Inside the beam-pipe a new approach is adopted. Rather than track the particles using a locally defined magnetic field strength, the approach more commonly adopted by accelerator tracking codes is followed, using an analytic solution to the equations of motion. By using the analytic solutions directly within a G4Stepper, a significant time saving is obtained over the more usual Geant approach of solving locally for each step in a magnetic field. In addition, an

468

interface to full field maps has been implemented for the specialised magnets such as are needed in the ILC interaction region and extraction line. The accurate tracking performance was tested in detail as part of a comparison of a wider set of tracking codes [8] and showed that the BDSIM technique has comparable accuracy to other accelerator trackers. The magnitude of the step length is provided by Geant4 and depends on which processes are present. For most cases, the step length is equal to the length of the beam-line element because the particles are traveling in high vacuum. Scattered particles, or halo particles in outer positions of phase-space, can leave the beam-pipe; once outside the beam-pipe the tracking defaults to the usual Geant approach. BDSIM naturally has access to all the impressive range of processes that are included in the Geant4 toolkit. These processes include multiple scattering off beam-gas particles or in detector elements and the usual electromagnetic shower processes such as electron-positron pair creation and bremsstrahlung. The beam-gas is specified by a normal Geant4 material with a pressure that can be defined by the user. Neutron generation from photo-production can also be included using the Geant4 photo-nuclear processes. Synchrotron radiation (SR) from electrons in the magnet elements can be generated and tracked down the beam-line. An example is shown in Fig.8 (Left) where the subsequent absorption of the SR is obtained as a function of position in the CLIC BDS.

Figure 8: Left: The SR absorption in GeV/m as a function of length along the CLIC BDS showing impacts on limiting apertures such as collimators. Right: muons created in electro-magnetic showers in the (TESLA) collimation system are transported down the tunnel to the interaction point (IP). Only the tracks of those muons that actually reach the IP are shown.

Additional processes can also be included as options. One example is tracking of beam halo and its collimation in the spoilers and absorbers of the ILC. The resultant generation of muons within electromagnetic showers in the collimation

469 sections, and the tracking of these muons down the accelerator tunnel, is illustrated in Fig.8 (Right). BDSIM is particularly useful for the detailed design of accelerator components and layouts. A common interface has been set up with the ILC detector groups so that detector geometries can be included rapidly for performance studies and optimisation. An example is shown in Fig. 9 (Left) where the ILC interaction region has been modeled and is currently being studied [9] to quantify and minimise the backgrounds in the detector from reflection of SR off downstream components (Fig. 9,Right).

M

0

0.01 0,02 0.03 0.04

0.05 0.06

0.07

0.08

0.00 0.1 XlmJ

Figure 9. Left: The Interaction Region for one of the ILC detectors ("LDC") as modeled in BDSIM, the IP is beyond the left of the figure. Right: The SR impinging on the first quadrupole downstream of the IP for various collimation depths; the black outline is the air-pocket of the quadrupole.

References 1. 2. 3. 4. 5. 6. 7. 8. 9.

T. J. Roberts, http://g4beamline.muonsinc.com The Geant4 Collaboration, http://geant4.web.cem.ch/ueant4/ The MICE Collaboration, http://mice.iit.edu/ http ://fermi tools, fnal. gov/abstracts/histoscope/abstract .html G. Blair CERN-CLIC-NOTE-509 (2002). G. Blair, Proc. LCWS05, Stanford (2005). The BDSIM code can be obtained from the website: cvs.pp.rhul.ac.uk. S. Redaelli et al Proc. EPAC 2002. J. Carter, Univ. of London PhD thesis (in preparation).

A F U L L MONTE CARLO SIMULATION OF SILICON STRIP DETECTORS

M.BRIGIDA, C.FAVUZZI, P.FUSCO, F.GARGANO*. N.GIGLIETTO, F.GIORDANO, F.LOPARCO, B.MARANGELLI, M.N.MAZZIOTTA, N.MIRIZZI, S.RAINO', P.SPINELLI DipartimentoInterateneo diFisica "Michelangelo Merlin"andINFNBari, Orabona4, 70126 Bari, Italy

Via

We have developed a full Monte Carlo simulation code to evaluate the electric signals produced in silicon strip detectors (SSDs) by absorbed X-rays and by thoroughgoing charged particles. This simulation can be applied in the design stage of a SSD, to study its performance as a function of the main detector parameters, like the strip pitch and the silicon thickness. All the physical processes leading to the generation of electronhole (e-h) pairs in silicon have been taken into account. Induced current signals on the readout strips are evaluated by applying the Shockley-Ramo's theorem to the charge carriers propagating inside the detector volume. Detailed maps of the electricfieldand of the weighting fields have been implemented to describe the motion of charge carriers and the time development of the induced current signals. A simulation of the front-end electronics, including the main sources of noise, allows to convert the current pulses generated on each strip into voltage signals. This simulation code can be very useful in studying the performance of different kind of silicon detectors (pixel, strip, drift) tailored for different applications (particle identification, tracking, X-ray detection) 1.

Introduction

Silicon strip detectors (SSDs) were first introduced as high precision tracking devices for high energy physics experiments in the early 1980s. Since then, impressive technological progress has been made, and SSDs have been used both in accelerator and space experiments. SSDs allow to obtain space resolutions up to a few um and, thanks to the reduction of their costs, large detector areas can be equipped with these devices. For these reasons, SSDs will be widely adopted in the next generation detectors, that will be mounted at the colliders (LHC, RHIC) or that will be flown on satellites. During the design phase of a detector, there are some parameters that need to be optimized in order to ensure the performance required by the experiment where the detector will be operated. An accurate detector model is therefore

* Corresponding author: fabio.gargano(a),ba.infn.it Tel. +390805443172

470

471 essential for understanding the behaviour of the device. We have written and developed a full Monte Carlo simulation code that simulates all the physical processes taking place in a SSD and allows to calculate the electrical signals generated by ionizing particles. A detailed description of the simulation code can be found on Ref.l. , where the ionization cross-section using the algorithm described in Ref.l.

2. Energy loss of a charged particle in silicon A charged particle crossing a silicon layer leaves an ionization track along its trajectory. The most probable energy loss Ap is between 180 and 300 eV/um, increasing with the silicon thickness. Since the energy required to produce an e-h pair in silicon is about 3.6eV, these values correspond to about 50 e-h pairs/um and 80 e-h pairs/um respectively. A complete and accurate calculation of the differential cross section cr(E) as a function of the energy loss E in silicon is given in Ref.2. The cross-section shows three peaks at energies of about 17, 150 and 1840eV, that are associated with the M, L and K shells of silicon. The energy transfer in a single collision is mainly due to the contribution of the M shell, while the contribution from energies above the K shell is less than 1%. A Monte Carlo method is used to calculate the straggling functions that describe the distributions of energy loss for different silicon thicknesses. The energy loss E; in the ith collision is determined by extracting a random number from the differential probability distribution F(E). Individual energy losses are added up to give the total energy loss of the particle. The distance travelled between collisions is sorted according to an exponential collision probability, P(x)=exp(-xA,)M, where X is the number of ionizing collisions per unit path length. The calculation is repeated for N particles and the energy loss distribution is thus generated. The charge carriers are produced both in the primary collision and by the subsequent energy losses of the secondary e-h pairs. Usually, the energy loss in aprimary collision is larger than the energy W needed to produce an e-h pair («3.6eV). Thus, for each collision of the incident particle, several e-h pairs will be produced. The primary collision occur with the exchange of "virtual" photons between the incident charged particle and the medium. The absorption of these photons and the subsequent relaxation of the excited atoms

472

yield the photoelectrons, the Auger and Coster-Kronig "primary electrons" and the corresponding "primary holes" in the valence band. All primary electrons and holes lose part of their kinetic energy by phonon scattering and another part by impact ionizations, producing secondary e-h pairs. When photons are absorbed by an inner shell (KjLijI^), the photoelectron energy is determined by the photon energy, whereas the energies of primary electrons and primary holes depend only on the vacancies created in the valence band by the photoionization process. Photons having energies less than Egap are not absorbed. A primary electron (hole) with energy E large compared to the semiconductor band gap can interact both by ionization, i.e. creating secondary e-h pairs, and by phonon emission. The generation of secondary e-h pairs is a cascade process, that is simulated with a Monte Carlo method. All secondary electronhole particles produced by the primary particle are followed up with the same procedure until all particles have residual energies below the threshold. These particles are the total charge carriers produced by the primary particle. In Fig. 1 are shown the results of the simulation for the number of carrier produced and for the pair creation energy W.

Fig. 1 LEFT: Average number of secondary e-h pairs, np (black circles) and carriers, n,; (open circles) as a function of the photon energy. The inset shows an expanded view of the behaviour of np in the low energy region. RIGHT: Electron-hole pair creation energy, W, as function of the photon energy. The inset shows an expanded view of the low energy region.

3.

Signal simulation

Silicon detectors can be described as p-n junction diodes, operated with a reverse bias voltage that forms a sensitive region depleted of mobile charge

473

carriers, in order to allow the collection of free charge carriers generated by radiation. The motion of a carrier (an electron or a hole) produced by ionization inside a silicon detector can be described by the equation v = u(E)-E where v is the carrier velocity, E is the electric field and u is the mobility, that depends on the electric field. During their drift, electrons and holes are diffused by multiple collisions, and their distribution follows a Gaussian law. In our Monte Carlo code the diffusion effect is taken into account during the solution of the equation of motion. The current induced at the time t by a moving carrier on the kth electrode can be evaluated with the Shockley-Ramo theorem ik(0=-qo-v-Ek where q0 is the charge of the carrier, v is its velocity and Ek represents the weighting field associated to the k* electrode. The induced currents are evaluated during the motion of the carriers, and the total currents are computed by adding the contributions from all carriers. A typical detector front-end consists of a preamplifier, that provides a gain, followed by a shaper, which tailors the overall frequency response for optimum noise performance, and limits the pulse length to accommodate the signal rate. The analytical relation between input and output signals can be derived using the Laplace transform. We have considered a standard front-end in which each induced current signal ik(t) is fed into the electronic chain formed by a preamplifier followed by a shaper. In our simulation we also taken into account different sources of electronic noise: • the leakage current Ij of the semiconductor detector, that can be represented by a current noise generator in parallel with the detector, and is a shot noise; • the noise due to the bias resistor, that can be assumed as a current generator in parallel with the detector, and is a thermal noise; • the noise from the feedback resistance, that can be described by a current generator and is a thermal noise; • the electronic noise of the amplifier, that is described by a combination of a voltage source and a current source at its input.

474

In this way we can simulate the voltage output event by event. In Fig. 2 are shown the result of the electronic simulation.

Time Ins)

Timeout

Fig. 2 LEFT: Simulated current signal for the hit strip and the adjacent ones. RIGHT: simulated shaper voltage output for the hit strip with noise added.

4. Conclusion We have developed a new Monte Carlo full simulation code that can be used for designing SSDs both for high energy and space physics. Our code includes all the physical processes taking place in a SSD and allows to study its performance. It can be used for different detector geometries and front-end electronics. A complete calculation of the e-h pairs is performed to generate the charge carries produced by the absorption both of virtual, e.g. exchange photons of charged particles with the silicon, and real photons. The signal simulation includes a noise simulation, that takes into account the detector and the electronic noise contributions.

References 1. M.Brigida et al. Nucl. Instr. and Meth. A 533 (2004) 322-343 2. H. Bichsel, Rev. Mod. Phys. 60 (1988) 663.

SIMULATION OF RADIATION MONITORS FOR FUTURE SPACE MISSIONS PATRICIA GONCALVES, MARIO PIMENTA AND BERNARDO TOME LIP -Laboratorio

de Instrumentacao e Fisica Experimental Av. Elias Garcia, n°14,l° Lisboa, 1000-149, Portugal

de

f

Particulas,

A GEANT4 based simulation of a compact lightweight radiation monitor to be included in the payload of future space missions is presented. The instrument must meet severe mass and power constrains, satisfy mission safety requirements and it is also required to perform as a scientific instrument. GEANT4 is a powerful tool for developing and optimising such detector concept thanks to its capabilities for describing the complex detector geometries and simulating the passage of particles through matter. Moreover, the presence of an optical physics process category in GEANT4 is of the utmost importance in the development of a scintillation based detector concept.

1. Introduction Radiation monitors are becoming an essential component in space missions, providing crucial radiation environment information for the in-flight protection of the spacecraft and instruments onboard. In addition, the data acquired during the mission are a valuable input for space environment models; in particular data gathered in missions to other planets (e.g. particle fluxes and spectral distributions as a function of the distance to the Sun) are essential for models describing the injection and propagation of particles from the Sun. Several of the future space missions (e.g. LISA, Gaia, JWST, BepiColombo) are planned to carry radiation monitors. Given the limited resources on mass, power and accommodation onboard a spacecraft, a new generation of compact and lightweight general purpose energetic particle detectors are being explored. Monte Carlo simulation tools are essential to develop and optimize a given detector concept and to explore alternative detector concepts. The ability to predict the detector performance in realistic radiation environments is also crucial. This requires the implementation of software models in order to perform an end-to-end simulation, including particle propagation from the source to the spacecraft location, followed by a detailed simulation of the detector response. The GEANT4 toolkit is an ideal framework for simulating particle interactions with detector materials in complex geometries. A GEANT4 based ^is work was supported SFRH/BPD/11547/2002.

through

FCT

475

grants

SFRH/BPD/11500/2002

and

476

simulation of a compact lightweight radiation monitor concept for future space missions is presented. 2. Space Radiation Environment The high energy ionizing radiationt environment in the solar system consists of three main sources: the radiation belts, galactic cosmic rays and solar energetic particles. The radiation belts are composed of electrons and protons trapped in the magnetosphere of the planets. In Earth's radiation belts (Van Allen belts), electrons have energies smaller than about 6-10 MeV and protons have energies up to 250 MeV [1]. Earth's radiation belts are mostly relevant for low earth orbit (LEO) missions. The galactic cosmic rays (GCR) component consists of a continuous flux of electrons, protons and ions, arriving from beyond the solar system, which are originated mainly in the supernovae. The GCR flux is maximal at energies around 1 GeV/n, decreasing as a power law for higher energies. Lower energy particles are strongly affected by the 11-year solar cycle. Although the GCR flux is comparatively low, energetic heavy ions can locally produce significant energy depositions originating single-event upsets. Solar energetic particle events (SPE) consist of a sudden and dramatic increase in the flux of particles from the sun, including electrons, protons and heavier ions. SPE are associated with impulsive solar flares and Coronal Mass Ejections and occur with higher probability during the solar cycle maxima. Although the energies of SPE rarely exceed few hundred MeV they give rise to potential risks for space missions due to the high fluxes attained. On the other hand their study is relevant for developing and testing solar particle propagation models [2].

3. The GEANT4 toolkit GEANT4 [3,4] is a Monte Carlo radiation transport simulation toolkit, with applications in areas as high energy physics, nuclear physics, astrophysics or medical physics research. It follows an Object-Oriented design which allows for the development of flexible simulation applications.

t In the context of space radiation effects, particles are considered energetic if their energy is high enough to penetrate the outer skin of the spacecraft. This means electrons with energies above 100 keV and protons and ions with energies above 1 MeV.

477

GEANT4 includes an extensive set of electromagnetic, hadronic and optics physics processes and tracking capabilities in 3D geometries of arbitrary complexity. The electromagnetic physics category covers the energy range from 250 eV to 10 TeV (up to 1000 PeV for muons) while hadronic physics models span over 15 orders of magnitude in energy, starting from neutron thermal energies. The optical physics process category allows the simulation of scintillation, Cherenkov or transition radiation based detectors. A distinct class of particles, the optical photons, is associated to this process category. The tracking of optical photons includes refraction and reflection at medium boundaries, Rayleigh scattering and bulk absorption. The optical properties of a medium, such as refractive index, absorption length and reflectivity coefficients, can be expressed as functions of the photon's wavelength. The characteristics of the interfaces between different media can be defined using the UNIFIED optical model [5]. Full characterisation of scintillators include emission spectra, light yields, fast and slow scintillation components and associated decay time constants. 4. Simulation of a generic space radiation concept The simulation of a simple space radiation monitor concept [6] was implemented using the GEANT4 toolkit. The detector consists of a tracker made of two position sensitive silicon planes followed by a scintillating crystal surrounded by photodetectors (Fig. 1.). An aluminium foil placed on the backside of the crystal is used to tune the energy range of the detector. In the present simulation the silicon tracker planes were 0.065 mm and 0.5 mm thick, the crystal size was 3x3x3 cm3 and the Figure 1. View of the simulated aluminium absorber was 2 mm thick. The particle detector. From left to scintillation properties of the CsI(Tl) crystal right: the two silicon planes, the were used. These included a light yield of scintillating crystal surrounded by 65000 photons per MeV of deposited energy photodetectors, the aluminum and two scintillation components with decay plane followed by a charged particle sensitive detector. time constants of 0.68 us (64%) and 3.34 us (36%). The scintillation light is readout by large area silicon PIN photodiodes, with a spectral sensitivity matching the CsI(Tl) emission spectrum. Since the silicon photodiodes are sensitive both to photons and charged particles, this makes them suitable to be used also as anti-coincidence shield. Figure 1 displays the fraction of the initial particle energy deposited in the crystal for protons with energies ranging between 25 MeV and 125 MeV.

478

Protons with energy amounting to 25 MeV or 50 MeV are absorbed in the crystal. For 100 MeV protons the effect of straggling is visible. 10'

10'

10'

""r 10" .I 1

Figure 2. Total energy deposited in the crystal by incident protons expresses as a fraction of the incident particle energy.

2

3

.. «

I 5

~tL^_ 6

/

I

S

••••

>

'

Figure 3. Simulated distribution of the photodetectors' hits time spectrum.

The principle of the anti-coincidence operation mode is illustrated in Fig.2. where the timing of the simulated hits in the photodiodes, produced by a 100 MeV proton that is not stopped inside the crystal is shown. The fast signal at t=0 is due to the direct ionization produced by the proton in the photodetector, while the slower signal comes from the scintillation light emitted by the crystal. Thus, by a suitable signal processing it is possible to disentangle the direct ionisation from the scintillation. This allows the identification of events where particles entered the detector from the sides or were not fully absorbed in the crystal.

Incident Energy (MeV)

Figure 4. Energy deposited in the first (left) and second (right) silicon planes by electrons, protons and alphas as a function of the energy of the incident particle.

Particle identification is performed by measuring the energy loss in the thin silicon trackers. Figure 4 shows the simulated energy loss in the first (left) and second (right) silicon planes, as a function of the initial particle energy, for electrons, protons and alphas.

479

5. Conclusions A new generation of compact, lightweight, general purpose radiation monitors needed for future space missions (e.g. BepiColombo) is under study. A simple concept based on a scintillating crystal was presented. LIP will be responsible for implementing the required Geant4 based detector simulations for the described setup in the context of a contract with EFACEC/Portugal. References 1. 2.

3.

4. 5.

6.

E.R. Benton and E.V. Benton, "Space Radiation dosimetry in low-Earth orbit and beyond", Nucl. Instrum. Methods B184, 255 (2001). D. Maia, " Particles from the Sun", 5th international workshop on New Worlds in Astroparticle Physics, 8-10 January 2005, Universidade do Algarve - Faro, Portugal. S. Agostinelli, J. Allison, K. Amako, J. Apostolakis, H. Araujo, P. Arce et al, "GEANT4-a simulation toolkit", Nucl. Instrum. Methods, 506, 250 (2003). M. G. Pia, "The Geant4 Toolkit: simulation capabilities and application results", Nucl. Phys. B - Proc. SuppL, 125, 60 (2003). A. Levin and C. Moisan, "A more physical approach to model the surface treatment of scintillation counters and its implementation in DETECT, in Proc. IEEE 1996 NSS, 2 (1996). A. Owens et al. (ESA-SCI/A), private communication.

E L E C T R O M A G N E T I C PHYSICS MODELING I N G E A N T 4 PACKAGE

V.M. G R I C H I N E P.N. Lebedev Physical

Institute

of RAS,

Moscow

119991,

Russia

Electromagnetic physics processes of GEANT4 package are considered in terms of model approach. The approach allows one to define the priority of models in different GBANT4 regions and to test their predictions versus the region cuts. The examples of ionisation process model implementation are discussed. Recent developments for GEANT4 X-ray transition radiation package are presented. The simulation results are in agreement with experimental data.

1. Geant4 Ionisation Models 1

uses by default so called standard ionisation model. It is essentially Landau-like fluctuation model with additional Poisson-like fluctuations for soft energy transfers. The model parameters relate with the Bethe-Bloch truncated energy loss with a number of corrections. Another approach is based on photo-absorption ionisation (PAI) model allowing one to estimate the cross section of ionizing collisions. PAI-model reads: GEANT4

a

(PNx Mw dx

T

irhc Im

»-£]ta(irjji;.l>-

<»

and dPNn hdudx

aN nh/32u

. . , 2mv2 1 [u . ... , In — 1— / c 7 [u ) aw - " - ' ~ niij •=-• ' w • • 'J0 ~ " ' ' r (w)

^

where a is the fine structure constant, /? = v/c, dx — v dt, m is the electron mass, N is the atomic density, hw is the energy transfer, e is the permittivity, and
480

481 0 . 9 3 A r + 0 . 0 7 C H 4 . 1.5 c m < 2 0 0 C , 2 a t m )

^ 900 "3

g> soo

600 500 400 300 2O0 IOO O

Figure 1. Ionisation energy loss distributions of pion with the momentum 9 GeV/c in the gas mixture 93%AT -f 7%CHA with the thickness of 1.5 cm (STP). Histogram is the experiment, points are GEANT4 simulation according to the PAI model, PAI model with photons and standard ionisation model. The cut value is 1 mm.

Si, 0.032 m m

I

I

6O0 O 4k Z^ O

SOO

PAI model: IO e v e n t s P A I w i t h p h o t o n s : IO4 e v e n t s G E A N T 4 s t a n d a r d : IO4 e v e n t s experiment: positron, p = 2 Gev/c

400

3O0

2O0

IOO

O

2

4

6

8

IO

12

14 16 18 20 Energy Loss (keV)

Figure 2. Ionisation energy loss distributions of positron with the momentum 2 GeV/c in silicon with the thickness of 0.032 mm. Histogram is the experiment, points are GEANT4 simulation according to the PAI model, PAI model with photons and standard ionisation model. The cut value is 0.1 mm. charge e crossing the interface between two media against an additional electric fields induced by the charge in the vicinity of the interface. This field is defined from the boundary conditions. Using the methods developed in 3 one can derive t h e relation describing t h e mean number of X T R photons, JVjn , generated per unit photon frequency OJ and 92 (0 is the emitting angle), d?Nin/hduidO2, inside radiator

482

for the most general XTR radiator consisting of n different absorbing media with fluctuating gap thicknesses tj (j = 2 , 3 , . . . , n — 1): fc-ln-l

atuO2

d?Ni,

2 2 Re[Yt(Zi-Zi+1) +2^2^2(Zi-Zi+1)l[Hj(Zk

Mcjde2 whc

i=l

•>k+l

)]•

i = l k=2

(3) In the case of gamma distributed gap thicknesses tj the correlation factors Hj are:

Hi = r

dti

Jo

t:

-l

r(^)

exp

t

l

2Zj

l+» 2Zu

jVj

(4) where T is the Euler gamma function, tj is the mean thickness of the j - t h medium in the radiator and Vj > 0 is the parameter roughly describing the relative fluctuations of tj. In fact, the relative fluctuation of tj is, $i = V\A^- ^ n e complex formation zone of XTR in the j-th medium 3 ' 4 , Zj (j = 1,2,..., n) is defined by: -l

1 - iLj/lj'

(5)

where i2 = —1, 9 is the angle between the charged particle velocity and the direction of XTR photon, 7~ 2 = 1 — /3~ 2 , /? = v/c is the ration of the particle velocity v and the speed of light c; uij and lj are the plasma frequency and the photon absorption length in the medium, respectively. In the case of a transparent medium, lj —>• oo, the complex formation zone is reduced to the coherence length Lj of XTR, which is real number. In many high energy physics experiments the XTR is detected by the arrays of straw tube gas proportional counters, Fig. (3), inserted in radiator. The G E A N T 4 class G4StrawTubeXTRadiator describes the generation of XTR in straw tube. In the particular case of n foils of the first medium interspersed with gas gaps of the second medium, the summation of (3) results in: cPNin hdw d62

2a

_we2Re{{R(n)}^ (6) H = H1H2, nhc2 " (1 - #i)(l - H2) , ( l - f f ^ ^ l - f T * ) (RW) = (Z, - Z2)2 1-H (1 - H)2

This approach allows us to implement XTR as G E A N T 4 as standard electromagnetic process. The integration of (6) in respect to 62 can be simplified for the case of regular radiator {vi$ -> oo) with transparent in terms of

483

Straw tube crossed by relativistic charges

^ZBBtu'

••-"'"

(P^\' ^^^r i

Figure 3. The diagram of straw tube crossed by charged particles with the charge e generating XTR on the straw tube surface interfaces. The medium 1 is air, the medium 2 is the straw tube wall, and the medium 3 is the detector gas mixture.

dE/dx and X-ray TR (Li/He 300 folis): 0.9 Xe + 0.1 CH,,, 3 cm (STP) 450 histogram: PAI+XTR model 7 = 4xl0 3 , 5 x l 0 3 events • experiment: electrons, p = 2 GeV/c £• 400

_

s 2

*

5 350 300

^

250 200

Fb

150 100

n

'^W

v++r>

50 40 0

50 6(1 Energy Loss (keV)

Figure 4. The ionisation energy loss distribution produced by electrons with a momentum of 2 GeV/c in a gas mixture 0.9Xe + O.ICH4 with a thickness of 3 cm at pressure 1 atm. Closed circles are the experimental data, the histogram is simulation according to PAI and XTR models. XTR radiator: 300 • ( Li / He = 0.04/0.126) mm.

XTR generation media, and n » 1 5 ' 6 . The frequency spectrum of emitted

484

XTR photons reads: f]N-

/•~107

J2|\f-

hdudO = TTHUJ ^4^k ^.(c1 + c2f

hdu -J J — o n

k

* E ( - 'l) (

(K

fc C

(-Jrnin)

2 fc

_ ti, 2 ((»?-fa;|) "J

~ ,

„;„2

+ C 2) 2

fc=fcmin

^1,2 —

4rvT7 2

_

sin

1

"^1

*1 +*2

u(h + h)

<-^mm — ~j

7

2

(* + c2)

+

(7)

huf + i 2 ^2

The sum in (7) is defined by terms with k -^ ^ m i n corresponding to the region of 8 > 0. Therefore femj„ should be the nearest to C m j„ integer kmin > Cmin. The value of fcmaa; is defined by the maximum emitting angle 82max ~ 1 0 7 - 2 . It can be evaluated as the integer part of: n

-C

• I ^

^ m a i — ^jnzn

*

h

+ 1

. 47TC

^

10

i-

or

7

0 J

5

ul

>

"-max

_*

.

^min

~102^103^1 LyJ

•

1,J

^> 1,

however numerically, only few tens of terms contribute substantially to the sum, i.e. one can choose kmax ~ kmin + 20. The comparison between 7 G E A N T 4 simulation and experimental data is shown in Fig. (4). Some of G E A N T 4 models show satisfactory agreement between experimental data and simulation and other require more experimental details for comparisons. This investigation was partly supported by Software Development Group of Physics Division of CERN in the framework of GEANT4 Collaboration and INTAS-2001-0323 grant. References 1. 2. 3. 4. 5. 6. 7.

Collaboration, GEANT4 : A Simulation Toolkit, Nucl. Instr. and Meth., A506 250 (2003). J. Apostolakis, S. Giani, V. Grichine et al., Nucl. Instr. and Meth., A453 597 (2000). V.M. Grichine, Physics Letters, B525 225 (2002). J. Apostolakis, S. Giani, V. Grichine et al., Comput. Phys. Commun., 132 241 (2000). G.M. Garibyan, Sov. Phys. JETP 32 23 (1971). C.W. Fabian and W. Struczinski Physics Letters, B57 483 (1975). Y. Watase, Y. Suzuki, Y. Kurihara et al., Nucl. Instr. and Meth., A248 379 (1986). GEANT4

PERSPECTIVES IN MEDICAL APPLICATIONS OF MONTE CARLO SIMULATION SOFTWARE FOR CLINICAL PRACTICE IN RADIOTHERAPY TREATMENTS MATTEO BOSCHINI SyS&Net dpt.,CILEA, via. R. Sanzio 4, Segrate (MI) Italy SIMONE GIANI PHdpt., CERN, Geneva 23, CH-1211, Switzerland VLADIMIR IVANCHENKO PHdpt., CERN, Geneva 23, CH-1211, Switzerland PIER-GIORGIO RANCOITA INFNsez.

di Milano, piazza della Scienza 3 Milan.20126, Italy

We discuss the physics requirements to accurately model radiation dosimetry in the human body as performed for oncological radiotherapy treatment. Recent advancements in computing hardware and software simulation technology allow precise dose calculation in real-life imaging output, with speed suitable for clinical needs. An experimental programme, based on physics published literature, is proposed to demonstrate the actual possibility to improve the precision of radiotherapy treatment planning.

1.

Physics Background

The physics of the electromagnetic processes in presence of material, density, atomic number Z, and discontinuities is well measured and understood since years [1]. The key effects are on the energy distribution in zones of sensitive materials followed and/or preceded by higher density and higher Z materials (exactly what happens in the human body). It has been shown since over a decade that standard MonteCarlo programs (EGS [2], GEANT3 [3]) take into 485

486

account and reproduce those processes and effects. However the computing time necessary to maintain the physics quality were prohibitive for clinical usage. On the contrary, the analytical and numerical dose calculations conventionally used for radiotherapy provided adequate speed for clinical needs. The question, whether the physics effects described above were properly taken into account or not, was purely academic, because de facto the speed requirements left no alternatives. The combination of different factors [4] has progressively started to allow performing the highest precision MonteCarlo calculations at speed compatible or superior than what is required for conventional radiotherapy treatment planning. The following key factors can now bring MonteCarlo computational methods to be exploited for clinical practice (assuming that the patient positioning precision will be performed with a sufficient accuracy): 1.GEANT4 [5] expert features created during its related R&D [6] phase (such as combination of production threshold in range and tracking to 0 range, dynamic generation of secondary particles below threshold near boundaries, geometry navigation in 3D voxelized models, volumes parameterization by material, time movement/variation of volumes). 2.1ncrease of CPUs speed up to 3 GHz on the market. 3.Parallelism using commodity hardware (50GHz fit in a 50cm box). Thus it can now be appropriate to re-evaluate the physics performance (and/or limitations) of classical analytical tools, for which we propose a detailed programme of benchmarks. 2. Computing Hardware and Software Performance The experimental setup used to prove the feasibility of real-time dose evaluation, is based on a cluster of high-end computers-off-the-shelf. In particular, the system was made out of 4 dual-cpu nodes, with a cpu clock of 3GHz each, amounting to a total of 8000 SPEC CFP2000. Nodes are connected through a dedicated 1 Gbit/sec LAN. The nodes share all software applications by means of NFS, but jobs are run locally, using a sand-box mechanism. The GEANT4 toolkit was installed, together with the related medical applications software, onto the CPU farm described above. The simplest parallel architecture has been observed to maximize speed performance: the simulation software was run on a master-slave parallelism architecture, where i/o was constrained just to the jobs (and random seeds) distribution at initialization phase, and to the merging of the physics results at finalization phase. The performance of the system was found to scale proportionally to the number of processors. It has to be emphasized that the full system fits within standard

487

office furniture; moreover, current dimensional specifications allow including 50 GHz in a box 50 cm high, which allows standalone-dedicated systems to be used as commodity PC hardware. An ion-therapy dose-calculation has been chosen as first case study. Scanner output with granularity of 1 cubic millimeter has been reproduced. A 3D simulation accuracy of at least 100 micron was enforced to maximize physics quality (Figure 1), including the exact determination of the Bragg peak width and height, of all background sources contributing to shorter-range dose release, and of the higher-range radiation tail due to fragmentation. Configurations with incident Carbon, Neon and Helium ions at different energies have been simulated, in all cases with a performance equal to or better than 7 minutes using 24 GHz. An alternative case study has been preliminarily tested, involving a complete accelerator beam line for photon-based radiotherapy in a leading European medical center. Target extraction, beam collimation, and patient phantom are included. For a typical 5 cubic millimeter scan granularity, 10 million Monte-Carlo events generate accurate dose distribution in 3 dimensions. Preliminary results induce the expectation that results can be computed in about 5-10 minutes using order of 30 GHz. 700

250 300 Depth (mm) Figure 1. Energy deposition of Carbon ions in water.

488

3. Validation and Benchmarks The following set of measurements can be performed using dose calculation tools such as Varian [7], Plato [8], BrainScan [9], or other treatment planning software used in radiotherapy. 1."Virtual Calorimeter": this measurement would consist in inputting a geometry model into the fore mentioned programs without using a scanner (e.g. via XML, ASCII, CAD...); in such a case it would be extremely interesting to describe therein the SICAPO [10] calorimeter geometry (Fe-Si-Pb sampling), for which all dimensions and materials (and results) are published; thus a dosimetry calculation could be done directly using those software programs on such a calorimeter configuration; this measurement would be almost at no-cost, because it would be entirely computer-based. 2."Inertial Calorimeter": this measurement is certainly possible at a limited cost, because it would not require irradiation nor operation of detector read-out electronics; it would consist in constructing a "passive detector" built by few sandwiches of lead/silicon/iron layers (just passive material); since data are available and published for different energies and SICAPO configurations, it would be possible to manufacture the passive detector according to the geometrical and chemical specifications of the original SICAPO calorimeter; then, the full scanning imaging system should be used to produce the 3D output of SICAPO (i.e. the calorimeter should be put on the bed of the CT scanner as if it was the patient); such 3D output should finally be given in input to the classical treatment plan software programs as before; and again the results produced could be evaluated. 3."MonteCarlo runs": the same procedures as above should be repeated by executing GEANT4 simulation calculations, both in full geometry mode (case 1.) and in scanner output mode (case 2.); the maximal performance achievable and physics quality should be exploited using GEANT4 expert features and hardware parallelism; all results can finally be compared to the experimental data. 4. Conclusions and Perspectives In case results will demonstrate advantages in favor of the MonteCarlo approach, then one could move to easily show how the discontinuities (in atomic number, interaction and radiation lengths, density) in a SICAPO-like calorimeter actually correspond to the real life situation in the human body composition

489 [11]. Ultimately it can be proposed to investigate how to use CCD technology to take measurements also in corpses, whose composition in terms of bones, soft tissues, air, water, allows to cross-check the predictions. It must be emphasized that the imaging systems must maintain a central role during the computation of results, because all calculations in clinical practice are ultimately done in what is "seen and reconstructed" by the diagnostic apparatus, and not in the ideal engineering geometry/materials models. Computing hardware architecture and software optimization also play a key role, because decisions in clinical practice need to be taken within human-driven time constraints: if results will confirm our expectations, there is a perspective for providing superior dose precision calculations at speed compliant to clinical needs, today. In all cases, we consider interesting to investigate how an integrated solution, including all the aspects discussed in this document, might realistically be put into operation. Such a system could integrate in a unique framework the imaging and diagnostic devices, the GEANT4-based dosimetry for treatment planning, and the treatment hardware. References 1. C. Leroy and P. G. Rancoita, Rep. Prog, Phys. 63, 505 (2000). 2. W. R. Nelson, H. Hirayama and D. W. Rogers, SLACReport 165, (1985). 3. S. Giani, Nucl. Phys. B32, 361 (1993). 4. F.Foppiano et al., Distributed Processing, MonteCarlo and CT Interface for Medical Treatment Planning, Proc. of the 8th ICATPP, World Scientific, Singapore, 324 (2004). 5. S. Agostinelli et al., Nucl. Instr. Meth in Phys Res.. A506, 250 (2003). 6. RD44 Collaboration, CERNLHCC 98-44, (1998). 7. www.varian.com, (2005). 8. www.nucletron.com, (2005). 9. www.brainlab.com, (2005). 10. G. Barbiellini et al., Nucl. Instr. Meth. in Phys. Res. A235, 55 (1985). 11. Particle Data Group, Phys. Rev. D66, (2002).

GEANT4 MONTE CARLO SIMULATION TO ESTIMATE GAMMA-RAY EMISSIONS FROM LUNAR SURFACE AND SELENE SPACECRAFT O. OKUDAIRA, N. YAMASHITA, N. HASEBE, M. MIYACHI, M. MIYAJIMA, M. HAREYAMA, K. SAKURAI, H. YAMAMOTO, M. -N. KOBAYASHI, E. SHIBAMURA, T. HIRAMOTO Advanced Research Institute for Science and Engineering, Waseda Univ, 3-4-1 Okubo, Shinjuku, Tokyo 169-8555, JAPAN Using Geant4 Monte Carlo simulation code, background gamma rays for Gamma Ray Spectrometer onboard SELENE were estimated. Gamma ray emissions from the spacecraft body by leakage neutrons from the Moon and GCR protons were calculated as well as ones from the Moon. It is found that a BGO background reduction system works so well that the background gamma rays produced from the spacecraft body is small in the energy spectrum to be observed by the spectrometer in SELENE.

1. Introduction Determining the chemical composition of lunar surface is an essential part of the investigation of the Moon. Gamma ray lines emitted by various isotopoes allow the abundances of many elements to be determined by gamma ray spectroscopy [1,2]. A Japanese lunar orbitar, SELENE, is to be launched in 2007. A gamma ray spectrometer (GRS) is onboard SELENE to study the chemical composition of the lunar surface by the remote sensing method [3,4]. In the present spectroscopic method, it is essential to identify gamma ray lines. Gamma ray measurement in space is inevitably carried out in a severe environment with background due to natural and cosmic ray induced gammas as well as prompt gamma rays induced by secondary neutrons from the Moon. Quantitative understanding of the sources of the background gamma rays is important in order to derive the chemical composition of the lunar surface material with high precision. However, the background gamma rays are difficult to be measure directly during the mission. By the use of Geant4 Monte Carlo code, the background gamma rays from the spacecraft body were calculated and the simulation was made for on the performance of the gamma ray spectrometer.

490

491 2. Gamma Ray Spectrometer onboard SELENE The GRS complies of a cylindrical germanium detector (252 cm3, 66mm and 78mm of diameter and length, respectively) as a main detector, a BGO scintillator (1255 cm3) and a plastic scintillator (5mm in thickness). The BGO scintillation detector serves to reduce Compton background events in the Ge detector. An energy window of the GRS ranges from 100 keV to 12 MeV, with an energy resolution of 3 keV at 1.33 MeV (FWHM) [4]. Such a high energy resolution can identify a number of elements in comparison with those of Apollo and Lunar Prospector. 3. Background Gamma Rays The GRS is mounted on the spacecraft body and looks at the lunar nadir. Therefore, the measurement of lunar gamma rays is interfered by background gamma rays emitted from the spacecraft body. It is important to estimate not only gamma rays from the Moon but also background gamma rays from the spacecraft for increasing the identification capability of elements in the lunar surface. The origin of gamma rays from the spacecraft body are categorized as (1) GCR, (2) neutrons from the Moon, (3) Compton-scattered continuum in the germanium detector and its ambient materials. The germanium detector is surrounded with the BGO scintillator which works as a passive shield against gamma rays from the spacecraft body and as well as an active shield for the anticoincidence system, which rejects signals of energy deposition exceeding 100 keV. The BGO is expected to effectively reduce gamma rays from spacecraft body, Compton-scattered gamma rays and charged particles. 4 Simulation by Geant4 Using Geant4 Monte Carlo simulation codes, we studied gamma rays emitted from the Moon and the spacecraft body, as categorized in (1) and (2) in section 3. Then, by considering the passive/active shield of BGO and plastic scintillators for background suppression, the energy spectrum of the Ge detector was derived, which contains the background occurred in (3), in order to estimate the performance of the spectrometer. To simulate the processes in lunar surface, we used a chemical composition of ferroan anorthosites as a model of the lunar surface material. It consisted of 45.8 wt% Si0 2 , 28.2 wt% A1203, 17.0 wt% CaO, 4.81 wt% FeO, 3.89 wt% MgO, 0.29 wt% Na 2 0 [5]. On the other hands, GCR flux model was based on the equation:

492 1.244xl0 6 £ , (g + 1876)(.E + M + 780exp(-2.5xl0 4 g))(E + M)(E + W6 + M)

J(E,M)[protons/cm2 I si MeV] =

where E is the initial energy of GCR, M is a modulation coefficient [6,7]. The coefficient M is 950 MeV at solar maximum, 550 MeV at solar cycle averaged, 375 MeV at solar minimum. Here, we adopted the M value at 550 MeV. For simplicity, the energy distribution of the primary GCR was divided into 20 bins each 0.5 GeV. For this simulation, a simplified model of spacecraft body (1650 kg, 2.1 X 2.1 X 4.2 m) and detectors was assumed. In this model eight aluminum panels with a thickness of 11.5 mm and three propulsion fuel tanks were incorporated. One of the tanks was spherical and filled with liquid N2O4, the others were cylindrical and filled with liquid 77 26, 30 . nt I ^ » N2H3. Hydrogen in the fuel t contributes to producing _ 16 %S% 10 0 ) l f 66 background gamma rays because it a acts as an effective moderator for S^^^^ 50 neutrons. In this model the z , 140 detectors, as in Fig. 1, was placed vJ on the spacecraft body facing to the Fig.l Simplified model of GRS detectors. lunar surface.

^GeJ

m

• • m

5. Results and Discussions Neutron intensity at the orbital altitude leaked from the Moon was simulated by assuming that GCRs are isotropically incident to the lunar surface materials. A spectrum of gamma rays leaked from the spacecraft body (only lunar side) produced by the lunar 10" neutrons was •fiiel:full calculated. In • fiiel:empty 10"2 addition, a leakage gamma ray spectrum - 1 0 by GCRs directly o Is NU incident to the 1 1 0 spacecraft body was 10" also calculated. Then, both the spectrum 10"6 summed as in Fig.2. 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 Energy fceV] From the Fig.2, in the Fig.2 The comparison of gamma ray emissions from the spacecraft with case of empty and without full fuel.

k

s

493

propulsion fuel, the total number of incident background gamma rays was approximately 4.7 times more than that in the case of full fuel. In addition, peaks originated from nitrogen and hydrogen disappeared, whereas peaks from aluminum remain. The background spectrum obtained by GRS was calculated. A spectrum obtained in BGO active shield mode and one in passive shield mode are shown in Fig. 3. It is clear that gamma rays from the spacecraft are substantially suppressed by approximately one order of magnitude by the BGO active shield. l

10"6 0

1000

2000

3000

4000 5000 6000 Energy [keV]

7000

8000

9000

10000

Fig.3 Background spectra of gamma rays detected by GRS with and without anticoincidence. The case of full fuel is shown.

As the results, some peaks appear clearly. An energy spectrum expected to be measured by SELENE GRS in the lunar orbit was calculated and the result is shown in Fig.4. In this estimation, gamma rays from the Moon was re-calculated by taking into account gamma rays on the lunar surface as mentioned by Reedy [7]. BGO active shield was enabled and from tile M oon and spacecraft b.g. from spacecraft

-a cur* SwA^W,

^^^H.

10"3

"sLulJU

wwm

«w 0

1000

2000

3000

4000

5000 6000 Energy DteV]

7000

8000

9000 10000

Fig.4 Expected gamma ray spectrum by SELENE GRS with anticoincidence (upper). The lower spectrum is the same as one with anticoincidence at Fig.3. The fuel tanks are filled.

494

the fuel was full. Figure 5 shows energy spectra for 1809 keV gamma rays of 27A1 (left) and 2313 keV gamma rays of 14N (right). From this result, the counting rate of the former gamma rays from the lunar surface is much enhanced in comparison with that from the spacecraft body. We can easily obtain the flux of the gamma rays from Al-composed | ftwn th« Moon and spavawafi• • - - from thw spacecraft | material in the Moon after subtracting the A |AI 1BQ9 [ ft IN 2313 1 background gamma rays from the spacecraft. For the latter, most of them come from the spacecraft 7VA7\A J '•" \ V/\"A/ " ' \>v' s* i \ body. So far as this peak is concerned, it is hard to 27 4 detect gamma rays from Fig.5 Gamma ray spectra around A1 1809 keV ancP N 2313 keV nitrogen-composed material in the Moon amid the background.

i-

!

Vi - /

6. Conclusion We estimated the emission spectrum of gamma rays from the Moon and the spacecraft body with or without anticoincidence system. In the case of full fuel, the number of gamma rays registered by the germanium detector from the spacecraft was reduced to about 1/8 with anticoincidence measurement. The BGO background reduction system worked effectively. Actually, since the fuel left at the beginning of observation is considered to be about 20 - 30% amount before the launch at the beginning of observation, the background level of gamma rays will be considerably small. Further detailed studies of simulation using a more realistic model of the spacecraft structure are in progress. References 1. R.C.Reedy, Proc. Lunar Planet. Sci. Conf. 9, 2961(1987) 2. C.M.Pieters and P.A J.Englert, "Remote Geochemical Aanalysis: Elemental and Mineralogical Composition", Cambrigde Univ. Press, Cambridge, 1993 3. N. Hasebe et al., Lunar Planet. Sci. Conf. XXX, 117(1999). 4. M. -N. Kobayashi et al., Nucl. Instr. and Methods, Res. A548, 401(2005). 5. P.H.Warren, E. Jerde and G. W. Kallemeyn, J. Geophys. Res. 92, E303(1987) 6. G. Castagnoli and D. Lai, Radiocarbon 22, 133(1980). 7. KCReedy, J. Geophys. Res. 92, E697(1987).

CORAL, A S O F T W A R E S Y S T E M FOR V E N D O R - N E U T R A L ACCESS TO RELATIONAL D A T A B A S E S

I. PAPADOPOULOSf R. CHYTRACEK, D. DUELLMANN, AND G. GOVI CERN, IT Department Geneva 23, CH-1211, Switzerland Y. SHAPIRO CERN, PH Department ATLAS Database Group Geneva 23, CH-1211, Switzerland Z. XIE Princeton University Princeton, New Jersey 08544 USA

The COmmon Relational Abstraction Layer (CORAL)1 is a C + + software system ,developed within the context of the LCG persistency framework, which provides vendor-neutral software access to relational databases with defined semantics. The SQL-free public interfaces ensure the encapsulation of all the differences that one may find among the various RDBMS flavours in terms of SQL syntax and data types. CORAL has been developed following a component architecture where the variouw RDBMS-specific implementations of the interfaces are loaded as plugin libraries at run-time whenever required. The system addresses the needs related to the deployment of relational data by providing hooks for client-side monitoring, database service indirection and application-level connection pooling.

1. Introduction CORAL is the set of software deliverables of the "Database Access and Distribution" work package of the POOL project 2 . The development of libraries for the vendor-independent database access, collectively referred to as the Relational Abstraction Layer (RAL), started within POOL in spring 2004. At that time the main requirement that needed to be met was the

'Corresponding author

495

496 creation of an insulation layer that would allow the development of software components responsible for accessing data in RDBMS without the knowledge of the subtle differences among the various technology-specific solutions. Since then RAL was adopted by the relational components of POOL (File Catalog, Collections), the COOL (Conditions Database) project 3 , and several components of the software frameworks of the LHC experiments. In spring 2005 a formal review took place of the existing RAL API with feedback from the direct clients of RAL, within and outside POOL. During the review new emerging use cases have been considered that are related to the database deployment and distrubution issues. The outcome was the design of an improved version of RAL, which is now being developed and packaged independently of the rest of the POOL components under the new name CORAL (COmmon Relational Abstraction Layer).

2. Project scope The primary goal of CORAL is to provide functionality for accessing data in relational databases using a C + + API, free of SQL commands and types, shielding the user from the technology-specific APIs. The need for the maximal SQL insulation from the client software can be demonstrated simply showing the SQL statements for two technologies (Oracle 5 and MySQL 6 ), which are used mostly in HEP, in rather simple tasks: (1) Creation of a table • MySQL syntax: CREATE TABLE T_t (I BIGINT, X DOUBLE) • Oracle syntax: CREATE TABLE "T.t" (I NUMBER(20), X BINARYTDOUBLE) (2) Query fetching only the first rows of the result set • MySQL syntax: SELECT X FROM T.t ORDER BY I LIMIT 5 • Oracle syntax: SELECT * FROM (SELECT X FROM "T.t" ORDER BY I) WHERE ROWNUM < 6 The CORAL approach is to present an identical API for such cases, which allows the development of software components that can be used without any code modification or conditional constructs against multiple relational technologies.

497 CORAL is expected to be used by applications running on a gridenabled and distributed environment. It is therefore defining the necessary interfaces for database service indirection, multiple authentication mechanisms (certificate- and user/password pair-based), client-side connection pooling and client-side monitoring. CORAL is not a general purpose C + + connectivity library, such as ODBC, JDBC, the python or perl DBI. It is neither a system to accomodate all access patterns to relational data. Its scope is restricted to serve primarily the use cases relevant to the data handling and analysis of the LHC experiments. 3. Architectural Choices The architecture of CORAL is component-based. The overall system consists of a set of abstract interfaces and several implementation components which have no dependency on each other. Data and control flow only through the abstract interfaces in a user application. The technology insulation is achieved through the set of the abstract interfaces; a client component depends only on them. The functionality expressed by each interface is minimal but complete. The component architecture and the fact that the functionality expressed by each interface is minimal but complete, enable the parallel and independent development of the various CORAL components, as well as additional and user-provided implementations of the same interfaces. They also facilitate efficient unit and integration testing. The realization of the component architecture has been based on the plugin management and the component framework of the SEAL4 project. The latter allows the construction of context hierarchies onto which the various implementation components are hooked. In this configuration if a component requires the functionality of a particular interface, it simply queries its context tree for an available implementation. 4. Functionality of the public A P I A user may connect to a database schema by prividing a contact string which should have a format tech nology-protocol://hostna me: port-n u m ber/schema _na me The technology may take values such as oracle, rnysql, sqlite, etc. The protocol is optional and used only on embedded technologies such as SQLite 7 . It may take values such as file, http, etc.

498

The loaded implementation of the IRelationalService interface parses the connection string and searches for available plugins providing the RDBMS functionality for the particular technology. No authentication credentials appear in the connection string. These are provided by the loaded implementation of the IAuthenticationService interface to the RDBMS plugin whenever an authentication is attempted. A user may enable client-side monitoring for a stated verbosity level. In this case the RDBMS plugin searches for a loaded implementation of the IMonitoringService interface and registers information such as begin and end of user sessions, signaling of transaction boundaries and the response times of the various SQL statements that are issued. Alternatively to the format specified above, a user may choose a free format which indicates a logical name for a database service. In this case a query to an implementation of the ILookupService interface has to be issued in order to obtain the corresponding connection string that corresponds to an actual service. The best way of doing so is to use an implementation of the IConnectionService interface which coordinates the control flow among the above components and provides application-level connection pooling. After having retrieved a valid handle to a database schema the CORAL API provides the user with a large subset of the functionality that is possible through generic SQL and RDBMS-specific connectivity APIs: • Schema definition and manipulation operations, such as creating, droping and redefining tables and views, as well as defining indices, keys and constraints. It is also possible to fully describe a given schema and its elements. • Manipulating data in tables such as inserting, modifying and deleting rows. It also allows operations to be executed in bunches, where only the input data change between subsequent iterations. This minimizes the total roundtrips to the database server and increases the overall performance of the application. Operations with BLOB types are facilitated through a Blob C++ type provided by CORAL. • Issuing queries involving one or more tables or views, allowing for row ordering, application of set operations, inclusion of sub-queries and the limiting the rows in the result set. Client-side row caching can be enabled to minimize the total roundtrips to the server. The CORAL interfaces have been designed such that the RDBMSspecific optimizations or standard "best" practices in RDBMS programming are handled internally by the implementation plugins.

499

5. Implementation components Every CORAL release provides • three RDBMS-specific implementations for the interface set related to the functionality of accessing data in a relational database (Oracle, SQLite and MySQL), • two implementations of the interface set related to the retrieval of authentication credentials: one based on XML files and another on environment variables, • an implementation of the interface set responsible for peforming the necessary technology dispatching given a connection string and the available RDBMS-specific implementations found at run time, • an XML-based implementation of the interface set responsible for peforming logical to physical database service lookup operations, • a simple implementation of the interface set responsible for registering and reporting monitoring events, • an implementation of the interface set responsible for the client-side connection pooling and the overall system configuration. 6. Development and release procedures The development of CORAL is done incrementally based on individual component tags submitted by the developers. A collection of tags which is validated through a series of integration tests becomes a release candidate. Frequent internal releases are used for early validation by the experiment software integrators and for providing a reference release for incremental development for the CORAL developers. Every component provides the corresponding documentation. During the release procedure, the documentation fragments from all packages are automatically assembled to produce the CORAL User Guide and web documentation. References 1. 2. 3. 4. 5. 6. 7.

http://pool.cern.ch/coraJ/ http://pool.cern.ch/ http://lcgapp.cern.ch/project/CondDB/ http://seal.cern.ch/ http://www.oracle.com/ http://www.mysql.com/ http://www.sqlite.org/

G E A N T 4 SIMULATION OF T H E ATLAS M U O N SPECTROMETER

N. CHR. BENEKOS Max-Planck-Institut fr Physik, Fohringer Ring 6 80805 Muenchen, Germany E-mail: [email protected] D. REBUZZI INFN-Pavia Via A. Bassi, 6 27100 Pavia (Italy) E-mail: [email protected] The Large Hadron Collider (LHC) at CERN will offer an The ATLAS detector is designed to fully exploit the potential of LHC providing high accuracy measurements in terms of mass and type identification of particles emerging from the LHC proton-proton collisions. Given the complexity of the apparatus in this demanding environment, an Object-Oriented(OO) methodology is required to manage the full detector simulation. The GEANT4 simulation of the ATLAS Muon Spectrometer, implemented in a OO perspective, is an ambitious task since it deserves not only a very detailed description of the apparatus geometry, but also it covers issues connected to a fast but precise simulation and a correct implementation of the different detector responses. A report on the Muon Simulation activities during the last year is the content of the present paper.

1. Introduction The ATLAS detector 1 , currently being installed at CERN, is designed to provide precise measurements of 14 TeV proton-proton collisions at the Large Hadron Collider2, starting in 2007. The apparatus, one of the biggest and most complex ever designed, requires a detailed and flexible simulation to deal with questions related to design optimization and to issues raised by staging scenarios, enabling detailed physics studies which will lay the basis for the first physics discoveries. The reference simulation tool adopted by all the detector component applications is the GEANT4 4 , 5 package. It deals with the new physics environment, testing extensively the appara-

500

501

tus prototypes and exploiting comparison with real data from the ATLAS Combined Test-Beam (CTB). The full simulation of the ATLAS detector consists of a collection of independent modules developed by the different teams of each subdetector (Inner Tracker, Calorimeters, Muon Spectrometer and Magnet System). Dynamic loading and organization in the form of plug-in modules make the implementation extremely simple and versatile. In this note we present a review of the activities of the simulation for the Muon System during the last year, starting from an analysis of the program design for the different components of the Spectrometer 7 , describing all the applications implemented (such as Detector Description 6 and and Digitization 8 ), with focus on the the test-beam physics and on the physics validation issues.

2. The ATLAS Muon Spectrometer Simulation The Muon Spectrometer of the ATLAS experiment is a large and complex system relying on several detector technologies. The simulation of muon events require a very careful description of these detectors, some of which provide a muon trigger (RPC,TGC) while the others make precision measurements of tracks (MDT,CSC). A detailed description of the passive materials, toroidal magnet systems and shield is also needed to account for Coulomb scattering and energy losses. Input for simulations comes from event generators after passing a particle filtering stage. The Muon Hit is the response of the detector following the interaction of a particle with its sensitive part (the gas in the CSC for example). Hits are labelled by a Simulation Identifier, SimID, a 32-bit integer in which the geometry information about the hit position is folded. The SimlDs are enough to access to all the geometrical information needed later on in the ATLAS offline full chain. Muon hits are stored in HitCollections, one for each Muon technology, where they are inserted in a random way (no sorting is performed at the simulation level). There are therefore four independent hit collections in the Muon Spectrometer, one for each technology, for each event. The hit information is combined with estimates of internal noise and subjected to a parametrization of the known response of the detectors to produce the simulated digital output, the Muon Digits, which are the input to the offline reconstruction programs. The digitization of GEANT4 hits in ATLAS Muon Spectrometer 8 is performed subdetector-by-subdetector.

502

The Muon digit object carries two pieces of information: an offline compact identifier (OID) of a readout element and the simulated data coming from the same element (TDC, ADC counts on MDT tubes, charge on CSC strips, RPC time of flight, etc). The digits can be feed to the pattern recognition and track reconstruction algorithms as if they were real data. In addition of being capable of handling hits coming from a single bunch crossing, the Muon Digitization is also able to handle piled-up collisions. Hits from several bunch crossings are overlaid taking into account the global time of the hit which is defined as the GEANT4 hit time, plus the bunch crossing time with respect to the main crossing. 2.1. Muon Detector

Description

The Muon Detector Description 13 of the Muon Spectrometer, is provided by the GeoModel 6 ' 7 toolkit, a library of geometrical primitives that can be used to describe detector geometries. The main purpose of GeoModel, here used in its implementation for the Muon Spectrometer, MuonGeoModel (MGM), is to support a central store for a unique detector description information that can be accessed by many clients, the simulation, the digitization and reconstruction programs. All Muon System components and corresponding materials are described in full details using MGM mechanisms. Recentely, also the dead matter component of the Muon System (Toroids, Endcaps, Feet and Rails) has been included in MGM. The shielding description is ongoing. MGM copes witht the GEANT4 simulation via the Geo2G4 interface, which translates the MuonGeoModel description to GEANT4 by navigating the raw geometry graph. The ATLAS GeoModel project started in summer 2002. Since then the geometries of all ATLAS detector subsystems have been described using GeoModel mechanisms. 3. TestBeam Simulation Beam tests are an ideal proving ground for a detector simulation program as they provide a direct comparison with experimental data in a simplified detector setup. In 2004, a large scale system test has been performed in H8 experimental area at CERN. As extensively described in 9 , the H8 setup included twelve Monitored Drift Tube (MDT) chambers arranged to emulate one barrel tower (six barrel MDTs) and a wedge of one endcap octant (six endcap MDTs). Simulating the ATLAS Combined Test Beam permitted to test the reliability of the simulation suite and to validate the

503

physics models employed. A 3D view of the simulated testbeam setup is shown in Fig 1.

\ iracKer Mag not

Figure 1. The ATLAS combined testbeam.

Using the data collected in the summer 2004 period, the barrel sagitta reconstruction was studied as a function of the incident muon momentum 12 , thus allowing to disentangle the intrinsic resolution contribution to its resolution, from the multiple scattering one. The results obtained have been compared to the GEANT4 simulation, resulting in an important validation step for the simulation code. For instance, an initial misagreement, between data and simulation, on the sagitta resolution, allowed to fix an RPC component material, which had been otherwise wrongly set in the simulation Detector Description. Among other parameters, data and simulation residuals has been compared, to check the digitization (and simulation) procedure, finding a good agreement, as shown in Figure 2. The Testbeam simulation used the same software as the full experiment, based on the same framework and implementation as for the ATLAS DataChallenges. 4. Conclusions The computing Data Challenges phase 2, ("DC2" n ), completely performed in the GEANT4 environment, proved the robustness and the stability 10 of the whole simulation apparatus on huge amount of generated event samples (single particle, B scans and Higgs) as well as on high statistics single particle, minimum-bias events , QCD-jets and physics event productions. The Muon Spectrometer simulation has been used successfully in the context of the ATLAS Data Challenges. It has also been validated in the context of

504 | Residual Comparison

\

J "'• 23 DATA I

•

SIMULATION

Residuals (mm)

Figure 2. Residuals for the data/simulation comparison.

CTB

Muon

barrel

chambers

(BIL/BML/BOL),

the experimental program of the ATLAS Testbeam, where analysis of the chamber performances relies crucially upon the simulation and the Detector Description model. 5. Acknowledgements The authors would like to thank all the core developers for the robust, versatile and complete code provided, the Muon subdetector people for their prompt implementation of any new proposed functionality, and Dr. Armin Nairz as ex-ATLAS Muon Simulation Coordinator for his excellent work and his important role in the success of this project. Sure to be incomplete in this review the authors would like to extend all their gratitude to everyone they might have forgotten. References 1. ATLAS Collaboration (1994), ATLAS Technical Proposal for a GeneralPurposes pp Experiment at the Large Hadron Collider at CERN, CERN/LHCC/94-43, CERN, Geneve. 2. D. Boussard et al., The Large Hadron Collider Conceptual Design, CERN/AC/95-05 (1995). 3. D. Boussard et al., ATLAS Muon Spectrometer Technical Design Report, CERN/LHCC/97-22, 1997. 4. The GEANT4 Collaboration, GEANT4 - A Simulation Toolkit, Nuclear Instruments and Methods in Physics Research, NIM A 506 (2003), 250303. 5. GEANT4 webpage, http://wwwasd. web. cern. ch/wwasd/geant4/geant4 • html 6. J. Boudreau, V. Tsulaia, The GeoModel Toolkit for Detector Description, Proceedings of CHEP04 Conference.

505 7. K. Assamagan et al., The Description of the ATLAS detector, Proceedings of CHEP04 Conference. 8. D. Rebuzzi, K. A. Assamagan, A. Di Simone, N. Van Eldik, Y. Hasegawa, GEANT4 Muon Digitization in the Athena Framework, in publication. 9. J.Bensinger et al., Muon Spectrometer Program in H8, http://cerutti.home.cern.ch/cerutti/h8-program2004.pdf. 10. D. Costanzo et al., Validation of the Geant^-Based Full Simulation Program for the ATLAS Detector: An Overview of Performance and Robustness, ATLAS Internal Note, ATL-SOFT-PUB-2005-002 (2005). 11. Gonzlez de la Hoz, et al., ATLAS Data Challenge 2 : A massive Monte Carlo production on the GRID, ATLAS Internal Note, ATL-SOFT-PUB2005-001 (2005). 12. G- Avolio, E. Meoni, A. Policicchio, F. Cerutti, S. Ventura, S. Rosati, D. Rebuzzi, Barrel Sagitta Resolution versus Momentum at 2004 B\8 Test Beam data and Comparison with GEANT4 Simulation, in publication. 13. ATLAS Muon Detector Description, http://muondoc.home.cern.ch/ muondoc/Software/DetectorDescription

TRACK E X T R A P O L A T I O N W I T H I N T R I N S I C NAVIGATION IN THE N E W ATLAS T R A C K I N G SCHEME

A. S A L Z B U R G E R * Institut

fur Experimentalphysik,

Universitat

Innsbruck

The transport of track parameters and their associated errors to surfaces within the detector is a very frequent process in the event reconstruction of high energy experiments. Track predictions are necessary for various pattern recognition algorithms and the transport of track parameter errors is essential for most of the track fitting techniques that, in general, rely on the first and second momentum of the underlying probability distribution. The realization of the newly developed track extrapolation engine for the ATLAS experiment and its specific property of a dedicated connective reconstruction geometry providing fast navigation and access to material and magnetic field information is presented in this paper.

1. Introduction The transport of track parameters (i.e. a track representation on a specific surface) and their associated covariances can be divided into two realms: the purely mathematical propagation of the track and its according errors depending on the underlying track model (in the following referred to as propagation) and the application of material effect updates according to the traversed material. The full process of propagation and material effects integration will be referred to as track extrapolation. The ATLAS experiment is one of the four experiments under construction to be operated with the Large Hadron Collider (LHC) at CERN, Geneva. Its design follows a classical scheme of particle collider detectors, consisting of a central inner tracker (embedded in a 2 Tesla solenoidal magnetic field), followed by electromagnetic and hadronic calorimeters, finally by an outer muon system that is realized as a stand alone tracking device with an extra toroidal magnetic field (peak intensity of up to 4 Tesla). However, the complex geometrical and magnetic field setup of the ATLAS *[email protected], CERN/PH-ATC

506

507

detector puts stringent requirements on the reconstruction software, in particular for the extrapolation package: (a) extrapolation to surfaces of different types and arbitrary orientation (b) extrapolation within a highly non-homogeneous magnetic field (c) correct material effects integration during the extrapolation process (d) good performance in respect of cpu time consumption due to the high track density and event rates 2. Design of the Extrapolation Package The new track extrapolation engine has been fully integrated in the ATLAS C + + software framework ATHENA 1 and is embedded in the common ATLAS Event Data Model (EDM). Following the ATLAS software design principles, a rigorous separation between EDM objects and algorithms adhered to by the extrapolation software design. Each building block of the extrapolation package is implemented with an extra layer of an abstract interface carrying all public methods. a This allows flexibility and facilitates the exchange or introduction of newly developed algorithms.

Figure 1. Illustration of the involved algorithms in a sample extrapolation process. For the user only the highest level - consisting of input parameters, extrapolator engine and output parameters - is visible. The illustration shows also the three columns the extrapolation is built on: navigation, propagation, material effects integration

The extrapolation package is built on a three column structure, separating the conceptually different tasks, see Fig. 1: a T h i s enables in addition dynamical loading of libraries at runtime and therefore shortens the building and linking time and decreases the library sizes

508

(a) propagation: realized through the IPropagator interface, see Sec. 3 (b) navigation: realized through the INavigator interface, see Sec. 4 (c) material effects integration: realized through the IMaterialEffectsUpdator interface, also described in more detail in Sec. 5 A central steering tool implemented under the IExtrapolator interface invokes these three processes by calling the dedicated sub-algorithms using only detector-technology independent representations of EDM objects. This structure along with a powerful configuration scheme is the key feature to make the extrapolation engine a universal tool for ATLAS. Its full power can be used on, for example, different layouts of the full ATLAS or any test-beam setup by exchanging the underlying reconstruction geometry only. In a similar way different track models can be taken into account by exchanging the propagation without compromising the navigation capabilities. In combination with the optimal choice of track fitting and reconstruction strategy, this extrapolation engine is aimed at achieving the highest possible accuracy in track reconstruction.

3. Propagation Four different propagation algorithms are currently implemented in the new track extrapolation package, following fully analytical or step-wise analytical track models. The different propagator algorithms extend the common IPropgator interface class and are designed to perform propagations to surfaces of all given types and orientations in the ATLAS reconstruction geometry: Propagations of both track parameters and associated errors according to an underlying linear or helical track model can be performed fully analytically with the so-called StraightLinePropagator respectively the HelixPropagator. Two propagation algorithms based on a Runge-Kutta integration algorithm are available in the new ATLAS extrapolation realm. They estimate the intersection of a track with a surface in a step-wise approach. A special flavor of a Runge-Kutta based propagator is the so called STEP-Propagator that corrects for material effects continuously during the propagation, as both energy loss and multiple scattering effects are embedded in the equations of motion through the magnetic field. This propagator is designed for the propagation through dense material, such as e.g. the propagation through the ATLAS calorimeter for matching of tracks determined in the

509 inner tracker with corresponding muon system tracks or simply for combined track fitting. A more detailed description of some of the propagator types can be found in 2 .

4. Navigation between Volumes and Layers The navigation and access to material and magnetic field information is based on a completely connective reconstruction geometry that is built by dedicated geometry builders by parsing the full ATLAS detector description. The reconstruction geometry is a simplification of the ATLAS detector geometry and combines the access to material information through confined material layers and magnetic field information by building the interface to the common magnetic field service. The simplification of the full detector geometry is vital for time performance; the number of describing volumes of the full detector geometry exceeds the number of volumes in the reconstruction geometry by five orders of magnitudes.

- SCT Barrel

Pixel SCT ^-""Vtft. v Gap Volume

Beam Pipe Volume J**^

Pixel Barrel

Figure 2. Left: Schematic illustration of the extrapolation process through three different volumes by means of navigating through their boundary surfaces. Right: Realization of the volume and layer structure for the ATLAS Silicon detectors in the central region.

In the reconstruction geometry a volume is characterized by its confining boundary surfaces which in turn provide information of the attached volumes on both sides. Each boundary surfaces is uniquely oriented by its normal vector and can therefore serve as an intrinsic guide to the next volume along a track direction. Within a volume, the navigation is based on layers representing the simplified material description of the reconstruction geometry. Both boundary surfaces and layers extend the common surface

510

representation and can therefore be naturally used together with the extrapolation engine. 5. Material Effects Integration The correct treatment of material interactions of the particle while traversing the detector to the given destination surface is crucial for the quality of the track reconstruction. Two different strategies are realized for the inclusion of material effects updates during the track extrapolation: a classical layer based approach and a volume based approach which is closely related with the development of the STEP-Propagator. In a layer based material update, the material has to be modelled as thin surfaces on which the update of the track parameters is done in a point-like correction to the curvature and an according smearing of the angular representation of the track. The volume based material update requires the inclusion of material interactions into the equations of motion as briefly described in Sec. 3. It is designed for propagation through dense material. Dedicated validation and calibration algorithms are embedded in the ATLAS track extrapolation package to facilitate the control of the inert material budget within the reconstruction geometry. 6. Conclusion A new extrapolation package has been deployed in the restructured ATLAS reconstruction software. It enhances the transport of track parameters and their associated errors to arbitrarily oriented surfaces of different types and within a highly non-homogenous magnetic field. The navigation involved in this task is given by a fully connective reconstruction geometry facilitating the access to material and magnetic field information. The extrapolation package is a vital component of the new ATLAS track reconstruction software and has been successfully used to reconstruct real data from the ATLAS Combined Testbeam in 20043. References 1. http://atlas.web.cern.ch/Atlas/GROUPS/SOFTWARE/00/architecture 2. Salzburger A., "The Track Extrapolation Package in the new ATLAS Tracking realm", Proc. of CHEP2004 3. Liebig W. et. al., "Test of the ATLAS Inner Detector reconstruction software using combined test beam data", Proc. of CHEP2004

APPLICATIONS OF GEANT4 FOR THE ESA SPACE PROGRAMME G.SANTIN Rhea System SA ,Louvain-La-Neuve, Belgium Currently based at the: Space Environment and Effects Analysis Section, European Space Keplerlaan 1, Noordwijk, 2200AG, The Netherlands

Agency

P.NIEMINEN, E.DALY Space Environment and Effects Analysis Section, European Space Keplerlaan 1, Noordwijk, 2200AG, The Netherlands

Agency

Knowledge of the potential impact of the radiation environment on evolving space borne devices rely on effective analysis tools for the understanding and the prediction of the basic effects of the particle environment on new technologies. In addition to experimental measurements, detailed numerical analyses are mandatory for the evaluation of the effects to detectors and components and, in human exploration initiatives, also to humans. The GEANT4 particle transport software is being used for a broad range of space applications. Current ESA-initiated activities in this area include calculation of radiation fluxes inside the ISS and for interplanetary missions, improved hadronic models, prediction of the radiation environment on other planets in presence of local atmospheres and magnetic fields, and studies of effects of radiation at the cellular and sub-cellular level.

1. Introduction Hostile environments such as energetic particles, hot and cold plasmas, meteoroids and space debris, or the effects of the residue of the atmosphere at very low orbital altitudes can severely limit space missions. Radiation shielding analysis is a crucial process in the spacecraft and space instrument development cycle to address the issues posed by the energetic particle environment, and simulations of the particle transport play a major role in spacecraft development, from design to operational phase and to post-flight analyses. In addition, particle transport studies are used for design, calibration and analysis of scientific detectors onboard space missions.

511

512 1.1. GEANT4 and the space particle radiation environment GEANT4 is a toolkit for the simulation of particle transport in matter1 . It has been initially developed by a collaboration (CERN RD44) of more than 100 scientists from over 40 institutes in Europe, America and Japan, and is currently maintained and still upgraded by an international collaboration. While developed in the context of High Energy Physics (HEP) experiments, it is currently used also in nuclear physics, space applications, medical physics, astrophysics and radioprotection. The capability of handling particle transport in complex geometries, together with a good physics description make GEANT4 a good simulation tool for the description of the effects of radiation in space. In addition, open source approach, object orientation and high modularity ease the extendibility of the toolkit and interfacing to external modules in areas such as geometry modeling, visualization, post-processing, and analysis. 2. Selected GEANT4-based analyses for ESA missions The European Space Agency (ESA) has been a Member of the GEANT4 Collaboration since 1997 and has supported via its R&D programme the development of a number of space-specific GEANT4 tools, models and mission or instrument applications. These include the development of physics models of relevance to space applications and in the promotion of new engineering tools for the standard radiation analysis procedure for future European missions. The internal use of GEANT4 is also part of the standard support to mission design and a number of Space Science missions have benefited from GEANT4 analysis capabilities. We present here a selection of the applications of these models and tools to ESA and other missions. Some of the GEANT4-based tools developed to aid the engineering design and operational phases of ESA missions will also be outlined. 2.1. Detector response in optical systems The description of optical processes in GEANT4 has been a key advantage for the study of the AMS veto system. An analysis of the veto scintillating detectors and their optical signals has been performed by LIP Lisbon within an ESA funded project. An investigation of the influence of the scintillator surface properties has led to the development of detailed models for the simulation of

513 the internal reflection processes, which finally led to an agreement of the simulated behavior with the results of experimental ground tests. Optical simulations were also performed for the EUSO experiment by LIP Lisbon. In particular, the properties of the Fresnel lens were investigated; a detailed geometrical model of the lens profile led to the conclusion that a strong non-uniformity of the focusing power was caused by unexpected diffraction of optical photons in some areas of the lens.

2.2. Physics extensions in the low energy electromagnetic domain The Low-Energy extension of the electromagnetic processes is of interest for future space missions, with examples in the study of the elemental composition of planetary surfaces through X-ray fluorescence, in the new space borne X and gamma ray telescopes, with the possibility of polarized radiation analysis, or in the internal charge deposition analysis for the proof masses in the LISA mission. A significant example of the importance of these extensions is the case of the BepiColombo mission to Mercury: surface analysis will be performed with a spectrograph, which will detect the X-ray emission generated from the impact of solar protons on the materials on the planet. An ESA co-funded improvement in the Proton Induced X-ray Emission (PIXE) model by INFN Genova has enabled the description (with a spectacular agreement) of the X-ray fluorescence in ground experiments with protons impinging on several sample materials2' . The LISA mission aims at the detection of gravitational waves with a constellation mission composed of three spacecraft, with an arm length of 5 million kilometers and a required control of the relative position of the proof masses to the picometer level. Knowledge and control of the charging of the proof masses is of extreme importance. The internal charging of the masses by cosmic rays, solar particles, delta rays and displaced ions has been investigated by Imperial College, London3 both for LISA and its precursor mission LISA Pathfinder. For verification purposes, comparisons between simulations and experimental charging data from the NASA Gravity Probe B mission have also been made by IC. The low energy extensions of the GEANT4 electromagnetic packages have been essential to analyze these effects down to a lower energy of 250 eV. 2.3. Low energy grazing angle proton scattering The X-ray Multi Mirror mission (XMM) objectives include study of black holes and formation of galaxies through the observation of sources with three X-ray

514

telescopes. An unexpected loss of Charge Transfer Efficiency of CCD's due to Displacement Damage was observed on the ACIS instrument on board the NASA Chandra mission, which has a similar configuration, just months before the launch of XMM. The cause of the damage was traced to radiation belt low-energy protons scattering off the X-ray mirrors. Surface scattering is the underlying physics process, for which an improved description has been obtained in a collaborative effort with participation from the GEANT4 developer collaboration and ESA experts. New physics was required for treating proton scattering at very small angles: in particular, the multiple scattering model has been improved and the Firsov model (for the description of grazing angle proton scattering from the surface of the X-ray mirrors) has been developed and added to the GEANT4 physics, giving a final prediction for the proton rate onto the XMM detectors which is in very good agreement with the observed rate in space4' 5' . The GEANT4-based study has led to a new operation planning with an extra beam shutter procedure during radiation belt passages, which has successfully protected the XMM X-ray detectors from damage by low energy protons. 2.4. Physics development and validation for the exploration initiative A recent increasing focus on robotic and human missions for exploration initiatives has led to new requirements for the particle transport simulations. The description of the radiation effects during long interplanetary journeys and a long permanence on other planets with scarce natural shielding brings the attention to planetary radiation models (e.g. scaling of solar particle event fluence to other planets, magnetic field description, surface maps, etc.), to physics models for hadronic and in particular nucleus-nucleus processes (for the description of the interactions of cosmic-ray high energy heavy ions), and to the description of biological effects (with macroscopic and microscopic models). The main goal of the DESIRE project is the accurate Monte Carlo calculations of the radiation fields and doses to astronauts inside the European Columbus module of the International Space Station. An extensive physics validation has been performed in the framework of the project (co-funded by ESA) with detailed comparisons the simulation of proton hadronic processes with experimental data6' . A complete geometry model of the International Space Station has also been developed, enabling the investigation of the dependence of the dose to astronauts on the directionality of the radiation impinging on the spacecraft.

515 Physics models for nucleus-nucleus collisions are the subject of recent and ongoing development within ESA funded projects. A recent addition to the GEANT4 hadronic package includes the implementation by QinetiQ7 of the abration/ablation model and new nuclear-nuclear cross-sections8' 9' . The evaluation of risks from human exposure to space radiation is also being addressed with the ESA funded REMSIM and GEANT4-DNA projects. The former has established preliminary strategies for manned missions to Moon and Mars, while the latter aims at the description of nano-scale damage effects of energetic particles at the cellular and DNA molecule level. Extensions of some of the electromagnetic processes in water down to energies of a few eV are being developed by the GEANT4-DNA collaboration. In addition, biological effect models are being reviewed and developed and a library of human phantoms is being added to the GEANT4 distribution10 . Simulations of particle transport on Mars in particular raise some specific issues, including the scaling of solar events and GCR fluxes at 1.5 AU, a broad range of altitudes on the Martian surface, seasonal, diurnal and local variations of atmospheric pressure, dust storms (for UV and X-rays), and surface backscattering (for which knowledge of the local geology is mandatory). In addition, local magnetic fields are present in the Martian southern hemisphere. The PLANETOCOSMICS GEANT4-based tool, by Bern University, enhances and generalizes the previous MAGNETOCOSMICS and ATMOCOSMICS11' tools, which were developed for the transport of particles in the Earth magnetosphere and atmosphere. The new tool has been extended with a description of the local magnetic field on Mars and of the dipole field on Mercury. Models for the planetary atmospheres are also included. Similar ESAfunded developments, with different Martian atmospheric models, are also ongoing in the frame of the RESTEC activity with Lisbon Institute of Physics. 2.5. Usability and interfaces ESA GEANT4-related R&D plans for the next years will focus strongly on making GEANT4 readily accessible to a variety of engineering applications and WWW-based radiation effects studies through the development of easy-to-use interfaces, advanced models, and auxiliary software. Particular attention is paid to the development and improvement of tools of relevance for the radiation environment in space to replace/supplement tools currently employed in ESA projects and missions12' . This has brought to the development of ready-to-use engineering tools utilizing the full GEANT4 physics and other capabilities but allowing rapid radiation analyses.

516 The Sector Shielding Analysis Tool (SSAT) performs ray tracing from a user-defined point within a GEANT4 geometry to determine, with very little computing resources, shielding levels and shielding distribution. The tool has recently been upgraded with the calculation of doses based on externally provided dose-depth curves. The Multi-Layered Shielding Simulation Software (MULASSIS)13' is a successful example of ready-to-use analysis tool with user-friendly interface. It performs fluence, Total Ionizing Dose (TID) and NIEL dose analysis in ID geometries made of several layers of arbitrary materials. A web interface to MULASSIS has been developed in SPENVIS14' . The recently developed Geant4 Radiation Analysis for Space (GRAS) tool15' addresses some of the issues left open, such as modularity of the analysis, flexibility of the design and complex 3D geometry models. Analysis types include at present TID, NIEL, NIEL, SEE, dose equivalent and equivalent dose16 17 , path length and charge deposit. A SPENVIS web interface is under development. 3. Conclusions We presented the application of GEANT4 to space missions in the context of the ESA programme. GEANT4-based applications and tools developed with ESA support have been applied for a wide range of studies including particle transport in planetary magnetic fields, instrument design, data analysis, shielding assessment and optimization. Synergies with medical activities have been useful to address critical areas of improvement, including interfaces and physics model extensions. Fruitful collaborations among ESA, academia and external institutes has led to the creation of Geant4-based applications and tools, which are now openly offered to the community. Further coordination and cooperation in basic research and related engineering developments are highly desirable. Acknowledgments This work was supported in part by the ESA Technology Research Programme, General Studies Programme, Portuguese Task Force, and the ESA Aurora Programme.

517 References 1. S.Agostinelli et al. [GEANT4 Collaboration], "GEANT4: a simulation toolkit", Nucl .Instrum. Meth. A506, 250 (2003). GEANT4 web site: http://cern.ch/geant4 2. S.Guatelli, A.Mantero, B.Mascialino, P.Nieminen, M.G.Pia, S.Saliceti, "GEANT4 atomic relaxation", Nuclear Science Symposium Conference Record, 2004 IEEE Volume 4, 16-22 Oct. 2004, 2178-2181 Vol. 4 3. T.Sumner, H.Araujo, D.Davidge, A.Howard, C.Lee, G.Rochester, D.Shaul, P.Wass, "Description of charging/discharging processes of the LISA sensors", Class.Quant.Grav.21:S597-S602 (2004) 4. R. Nartallo, E. Daly, H. Evans, P. Nieminen, F. Lei and P. Truscott, "Lowangle scattering of protons on the XMM-Newton optics and effects on the on-board CCD detectors", IEEE Trans. Nucl.Sci. 48, 1815-1821 (2001) 5. F.Lei, R.Nartallo, P.Nieminen, E.Daly, H.Evans, P.Truscott, "Update on the use of Geant4 for the Simulation of low-energy protons scattering off X-ray Mirrors at grazing incidence angles", IEEE Trans. Nucl. Sci. 51, Issue 6, Part 2, 3408-3412 (2004) 6. T.Ersmark, PCarlson, E.Daly, C.Fuglesang, I.Gudowska, B.Lund-Jensen, R.Nartallo, P.Nieminen, M.Pearce, G.Santin, N.Sobolevsky, "Status of the DESIRE project: GEANT4 physics validation studies and first results from Columbus/ISS radiation simulations," IEEE Trans. Nucl. Sci., 51, Issue: 4, 1378-1384 (2004). DESIRE web page: http://gluon.particle.kth.se/desire/ 7. P.R.Truscott, "Nuclear-Nuclear Interaction Models in GEANT4", QINETIQ/KI/ SPA CE/ SUM040821/1.1 (2004). 8. J. W. Wilson, RK Tripathi, FA Cucinotta, JK Shinn, FF Badavi, SY Chun, JW Norbury, CJ Zeitlin, L Heilbronn, J Miller, "NUCFRG2: An evaluation of the semiempirical nuclear fragmentation database," NASA Technical Paper 3533 (1995). 9. L.W.Townsend, J.W.Wilson, R.K.Tripathi, J.W.Norbury, F.F.Badavi, and F.Khan, "HZEFRG1, An energy-dependent semiempirical nuclear fragmentation model," NASA Technical Paper 3310 (1993). 10. P. Nieminen and M.G. Pia, "II progetto GEANT4-DNA", AIRO Journal, March 2001 (in Italian) 11. L. Desorgher, E.O. Fluckiger, M.R. Moser, R. Butikofer, "GEANT4 simulation of the propagation opf cosmic rays through the Earth atmosphere", published in *Tsukuba 2003, Cosmic Ray* 4277-4280 (2003). For information about the MagnetoCosmics tool, see the URL: http://reat.space.qinetiq.com/septimess/magcos 12. G. Santin et al., "New GEANT4 Based Simulation Tools for Space Radiation Shielding and Effects Analysis", Nuclear Physics B (Proc. Suppl.) 125, 69-74 (2003)

518

13. F.Lei, P.RTruscott, C.S.Dyer, B.Quaghebeur, D.Heynderickx, P.Nieminen, H.Evans, and E.Daly "MULASSIS: A GEANT4 Based MultiLayered Shielding Simulation Tool", Proceedings of IEEE Nuclear and Space Radiation Effets (NSREC02), Phoenix (USA), 20-24 July 2002. 14. D.Heynderickx, B.Quaghebeur,, E.Speelman, E.Daly, "Space Environment Information System (SPENVIS): A WWW interface to models of the space environment and its effects", AIAA-2000-0371 (2000). SPENVIS web-site: http://www.spenvis.oma.be/spenvis/ 15. G.Santin, V.Ivanchenko, H.Evans, P.Nieminen and E.Daly "GRAS: A General Purpose 3D Modular Simulation Tool for the Space Environment Effects Analysis", accepted for publication on IEEE Trans. Nucl. Sci. (2005). 16. "1990 Recommendations of the International Commission on Radiological Protection", Annals of the ICRP, Vol. 21, No. 1-3 (1991). 17. "Relative Biological Effectiveness (RBE), Quality Factor (Q), and Radiation Weighting Factor (wR)", Annals of the ICRP Vol. 33, No. 4 (2003).

DETECTOR SIMULATION IN HIGH ENERGY PHYSICS J.P. WELLISCH Private scholar, 1279 Avenue de Jura Sergy, F-01630, France The geant4 simulation tool-kit has been developed as a simulation tool-kit for particle physics, and an ever growing list of laboratories and institutes are using geant4 as baseline for their simulation needs. In addition, there are a growing number of high energy physics experiments that use the technology provided by this program in the preparation of the detector and the interpretation of the data. We will give a brief history of detector simulation in high energy physics as it relates to geant4, and summarize the activities in the field. We will also analyze the reasons for the current successes of the geant4 tool-kit with a penchant towards hadronic physics simulation.

1. History of Geant4 At the time of writing, the effort that resulted in the current Geant4[l] releases is close to 12 years old. The first prototypes for large parts of the code were developed not inside the geant4 collaboration, but as part of a CERN R&D project, RD44[2], This project was originally conceived as a technology R&D project, with the goal to study modern software development techniques, software processes, and technologies (C++), to eventually construct the next CERN version of the GEANT[3] simulation program. As such it was the first successful attempt to replace a major CERN software package for the next generation of high energy physics experiments. At the end of the RD44 phase, a Memorandum of Understanding was written to give the program a future R&D and deployment path. CERN management decided that they were not interested to be host laboratory for this effort in future. The name GEANT was abandoned, being a well known trademark of CERN, and was replaced by the name under which the program is known today - Geant4. Already during the RD44 phase a sizable number of requirements were collected. The main source of this was high energy physics and the preparation of LHC detector simulation, but a variety of requirements were also collected from other areas like heavy ion physics, CP violation physics, cosmic ray simulation, medical applications, and space science and engineering. In order to meet such requirements, a component approach was chosen for the physics 519

520

implementations, which in turn allowed the group to build in a large degree of flexibility and functionality. 2. Geant4: Components and use-cases The basic considerations that drove the overall software design, in particular for the physics, was the need for extensibility. An open and extensible architecture was used. Implementation frameworks, in particular for the case of hadronic physics[4], were used to create abstractions of the responsibilities in the problem domain, and RD44 invited collaboration with all experts. The abstract interfaces defined in the implementation frameworks then allowed to de-couple individual efforts, and enabled independent development by physicists with deep expertise in the field of modeling hadronic interactions. 2.1. Overall architecture

GUI

Hits

Digits

Run

« systems Vis.

Event

«sfub» Event Generator

Tracking

i «systern» j GHAD I physics «stub» i Persistancy

«stub» Analysis

Transport

«systern» EM physics..

globals

i and others

«system» i

Figure 1: Package diagram showing the architecture of Geant4 version 7.0.

The package diagram of the Geant4 architecture is shown in fig.l at the example of the Geant4 7.0 release. The architecture[5] is centered around an implementation framework for physics processes, that generalizes all physics implementations in Geant4, and provides the necessary top level abstractions. It implements the interaction with the technical categories concerning event generation, events, runs, visualization, etc. This interaction is through its being

521 used by the tracking package, the heart of the Geant4 simulation infrastructure. The physics framework also provides the information on geometrical quantities and detector properties through its usage of the track package. This apparently simple architecture allows to decouple any physics implementation from the simulation infrastructure part of the system. Note that even the transportation process is funneled through the process interfaces. 2.2. Electromagnetic Physics Packages

|

—\

P

Low energy

Parametrize tlon

Utfte

«uses» «uset

MSB

p! p!

Di Xraya

ses» *usss#

Muon*

Standard

Optical

High energy

Figure 2: Package diagram of electromagnetic physics in Geant4 version 7.0.

Figure 2 shows the packages in electromagnetic physics for Geant4 version 7.0[6]. The Standard package provides the complete basic functionality needed for HEP detector simulation, while the Muons and X-rays packages provide this physics for muon interactions and X-ray generation and interactions. The Optical package provides for optical photon transport and interactions of optical photons with bulk matter and optical surfaces. The High Energy package contains extensions to the Standard package, describing processes that gain in relevance as the TeV energy scale is approached. The Parameterizations package allows for shower parameterizations, and the Low Energy package contains an incomplete set of processes alternative to those provided in Standard, with a focus on low energy particle generation. 2.3. Hadronic Physics Packages Hadronic physics in Geant4 was constructed around a set of implementation frameworks, as an aggregation hierarchy that is five levels deep. Today it constitutes more than half of the Geant4 source code.

522

Figure 3 shows the top-level package diagram. The Processes package implements the interfaces of the hadronic physics framework, the Hadronic management package, which implements the physics framework, and provides additional framework functionality for biasing, material selection, exception handling, white-boards, multithreading, etc. It delegates any further action that directly relates with physics. The Stopping package provides a set of processes for hadrons coming to a rest through energy loss. The Inclusive Cross-section package provides the cross-section data sets that are used at tracking time to determine step sizes. The final state modeling package is a subsystem that is illustrated in figure 4 and discussed later. They all make use of a class utility package that contains classes describing for example nuclear properties, but also technical codes like exception classes, whiteboards, monitoring or templatemeta-programs for type-list based subsystem configuration at compile-time.

' Hadronic man.

processes

Ijwlus__ cross sections

((subsystem* | Final state modeling

Stopping hadrons

Ilk General s : utilities 1

Figure 3: Package diagram of hadronic physics top-level architecture in Geant4 version 7.0.

Figure 4 shows the package structure of Hasfinalstate modeling subsystem. This level contains all physics implementations. In the case of hadronic physics, the strategy was to name the packages after the physical concept used to model the final state generation, hence the Binary Cascade package implements the binary cascade concept[7] the chips package implements chiial invariant phase-space decay[8] and Imaginary R-matrix implements the imaginary part of the R-matrix through resonant and non-resonant 2-particle scattering[7]. A more detailed description of these packages is beyond the scope of this paper, and can be found in the Geant4 physics manual available from the Geant4 web site. Note also that iheparton string[9] and de-excitation packages[10] are subsystems in their own right, containing themselves 12 packages in total.

523

RadloacDvs decay

Muon nuclear

imaginary R-matrlx

U-»i ! Model ( < utilities i

teotop* prod.

Mars rewrite

1

^

HUironic Datasct

^subsystems Verfncjtlon engine

Model man.

I

chips

~~High energy tfiBory

« subsystem* De excitation

High energy parametr.

Coherent elastic

Binary cascade

« subsystem* Parton string

Neutron transport

em dissoc.

Free on) pound

Berfenl cascade

Nucleon nucleon elivlk.

Figure 4: We show the package diagram of hadronic final state generation in Geant4 version 7.0.

Given the complexity of physics modeling in the hadronic subsystem, a set of pre-packaged "educated guess" simulation engines were provided[l 1]. They can be found at the physics lists web site. The use cases currently included in this are high energy physics calorimetry excluding linear collider "tracking" calorimetry, high energy physics trackers, the "average" particle physics detector, low energy dosimetry applications with neutrons, low energy nucleon penetration shielding, linear collider neutron fluxes, high energy penetration shielding, medical and military neutron applications, low energy dosimetric applications, high energy production targets (ex. pion fluxes from 400 GeV protons on Carbon or Beryllium), medium energy production targets (ex. fast particle fluxes from 15-50 GeV protons on light targets), LHC neutron fluxes, air shower applications, and low background experiments. These pre-packaged and versioned physics list were elementary pre-condition for both wide user validation of geant4 hadronic physics and its supportability.

3. Geant4: Predictive power of associated physics modeling Another key to the successes that Geant4 sees in the field of High Energy Physics was an extensive verification program. The psychological need

524

addressed was to convince people that the models in particular in hadronic physics have strong predictive power at the microscopic level. The verification effort performed by the Geant4 hadronic working group is grouped into several sections concerning thin target comparisons for final state generators, inclusive cross-section testing, verification of model components, code comparisons, complete application tests, and robustness. 3.1. Thin target comparisons for final state generators The thin target comparison suite contains at the time of writing close to 2500 individual plots. Figure 5 shows two of these as illustration of the method and the predictive power of quark gluon string model[12]. It compares predictions for negative pion production in K+ Gold interactions at 100 GeV kaon energy to experimental data[13]. We show rapidity and transverse momentum distributions. We see good agreement between the model prediction and experimental data. QGS model is the currently most suitable high energy generator for high energy physics calorimeter simulation. Whltmore et.al, Zeitschrift f. Physik C62, 1 99 (1 9 9 4 ) 1.6 1.4

r

i

F

l

i

i

! K*Au — n " x

t

P»=100GeV

1.2 1 0.8 0.6 0.4 0.2 0

&oooooo°o° ~ I

1 I

1 1 1 1 I

1 1 1 1 I

oo°"ooo 1 1 I

I I

1 1 1 1 I

1 1 1 I

I

1 1 1 1 I

1 1 1 1 I

1 1 I

r 1 1 1 1

1

•

o

o 10

K*Au — n " X p,.b = 1 0 0 GeV

0 T

o 00

1

V Y

0~ 0

0.25

0.5

1

,

.J 0.75

r it H f f .t 1..III .1.25 ,!. 1.1.1.1.5,11. .1.75 . ,1. . . p,2 [GeV2]

Figure 5. The quark gluon string model prediction for negative pion production in K+ Gold interactions at 100 GeV compared to experimental data. We show rapidity and transverse momentum distributions.

525 3.2. Cross-section comparisons As an illustration of this type of comparison, we show in figure 6 the comparison of the proton reaction cross-section prediction of the Wellisch-Axen systematics[14] with experimental cross-section values for various targets[14]. We find good agreement.

10

TO2 ^o3

io+

IO

io 2

io 3

io+

10

ici 2

io 3

io+

EHn[MeV] Figure 6: Comparison of the proton reaction cross-section prediction of the Wellisch-Axen systematics with experimental cross-section values for various targets.

526 3.3. Verification of model components This part of the verification suite is aimed at demonstrating the individual model components are correct, and not tuned away from their physical values. This includes among others verification of nuclear masses, density distributions, particle width, and total reaction cross-section predictions by cascade codes. In figure 7 we show the prediction of the A4* production cross-sections in protonproton scattering, which is part of the description of the imaginary part of the Rmatrix in binary cascade or QMD codes. The data points are taken from the CERN/HERA collections[15], and scaled with the appropriate Clebsch-Gordon coefficients. The line is the result of the Monte Carlo simulation component.

10

1

0

2

4

6

8

10

12

14

Figure 7: Comparison of the prediction of the A " production cross-sections in proton-proton scattering. This is part of the description of the imaginary part of the R-matrix in binary cascade or QMD codes. The data points (solid symbols) are taken from the CERN/HERA collections. They are scaled with the corresponding Clebsch-Gordon coefficient.

3.4. Code comparisons Code comparisons are generally considered the least effective comparison technique, and are resorted to only in cases where no data exist for a quantity of relevance to a simulation problem. Preconditions for such a comparison are that the codes compared use different modelling techniques to arrive at the prediction, and that they have not been tuned on the same experimental data for other quantities. If these conditions are fulfilled, code comparisons can be very useful. We give in figure 8 as an example the result of a code comparison

527 between the RADLIST[16] code and the Geant4 photon evaporation and internal conversion codes. For the reader to judge if anything can be concluded.

Radiation CEK CE CEL CE CE CE CEK CE CEL CEM+ CEK CE 7 7 y y

RADLIST (BNL) G«ant4 Energy (keV) Intensity (lOOdks) Energy (keV) Intensity (lOOdks) 7.301 71.00 (6.0) 7.301 70.55 (1.88) 12.899 10.00 (0.70) 13.567 7.40 (0.6) 5.95 (0.54) 13.562 13.687 0.35 (0.13) 14.315 0.85 (0.21) 14.405 0.45 (0.19) 114949 1.83 (0.14) 114.949 1.95 (0.31) 120.497 5.70 (0.53) 121.215 0.19 (0.020) 121.968 0.03 (0.005) 129.361 1.30 (0.16) 129.362 1.25 (0.25) 134.910 0.25 (0.11) 14.413 9.16 (0.15) 14.413 10.05 (0.71) 122.061 85.60 (0.17) 122.061 86.05 (2.07) 136.474 10.68 (0.08) 136.474 10.05 (0.71) 692.410 0.15 (0.01) 692.030 0.15 (0.09)

Figure 8: Code comparison between the RADLIST code and the photon evaporation and internal conversion code used in Geant4.

3.5. Complete application tests Complete application tests are introduced to second the validation efforts in the experiments. Complete application tests can only be done in close collaboration with the test-beam communities. In figure 9 and 10 we show the result of such an analysis using a modern physics simulation engine, QGSP 2.9 and PACK 4.0. The ATLAS hadronic end-cap calorimeter pion test-beam measurement[17] is simulated. The end-cap is a flat-plate liquid argon sampling calorimeter with an EST readout structure. The absorber material is Copper. It has 4 longitudinal compartments (readout layers). For details of the detector set-up we refer to the original publication of the ATLAS community[18]. We see excellent agreement for energy resolution and signal ratio of electrons to pions. We also see good agreement in the shower shape comparison for all longitudinal compartments of the calorimeter and all pion energies. It is this kind of comparison that must constitute the final acceptance test of a simulation engine for physics simulation.

528

• -*-

[s 1^17.5

u

:1 _ •

°

15 12.5 10

¥

•

¥ •

-

7.5

f

0

:, ,

i

FLUKA99, 7.1, org.

V

GCALOR, 3.21, org.

.9

T

•

,

• ft •

9

•

•

5 2.5

QGSP 2.9, Pack 4.0. g4 7.0 potchl. CLHETP

a

, ,

I

•

, |

I i i

9

f

•

•

•

, , , i . . . . i . . . , i . . .

I . I !

200 EIGeV]

s

'-5

iTl.45 1.4 1.35 1.3 1.25 1.2

: •=

"

i-

• T

• a T

•

data, org

•

QGSP 2.9 Pack 4.0, g4 7.0pl. CLHEP 1 g

S •D • •

s

a

V

T

•

V

•

FLUKA99, 7.1

T

GCALOR, 3.21

s

•

*

•

•

•

•

1.15 •

1.1

,

I

I

,

I

,

,

,

,

,

•

•

•

1

I 1 40

1 60 EIGeV]

Figure 9: Energy resolution and e/pi ratio in the ATLAS hadronic end-cap calorimeter as predicted by QGSP compared with data and other simulation engines as a function of the pion beam energy.

i;. -

J

Lny.r 1

"TO. 75

0

= 0.4 0,

5

!_•i

0.3 0.25 0.2 0.15

| _ 0.7

-

•

dot

n

oc P2.9. Puck*

J4D.65 ?

"i

0.6 0.55

8

i

•

0.5

B

'-

• • u

0.45

-

0

~3

"•

0

••

i :* i

1 m

i

*

• «

»

«

0.4 0.35 200 EIGeV]

l„ ,0.3

—

|0.08

-

iiO.07

»

jBO.06

0.2

« *

0.15 0.1 0.05

Lo,»<

~cO.09

" • "•

*

•

0.05 IK

9 0.03 0.D2 0.01

" •• - • s•

• ft

y

i

B

,

Figure 10: Mean energy fraction per longitudinal compartment in the ATLAS hadronic end-cap calorimeter as predicted by QGSP in comparison with data as a function of pion beam energy.

529

4. Geant4: Deployment in High Energy Physics The deployment of Geant4 in high energy physics was largely based on initiatives of individual working groups. The first experiment to adopt Geant4 as infrastructure for their simulation program was BaBar[19]. Geant4 was used there starting from 1999, a time where Geant4 was not really fit for the job, but by now more than 1010 high quality BaBar events have been simulated for physics analysis. In the years to follow new high energy physics experiments like AMS[20] and PANDA[21] adopted Geant4 for their simulation needs, and the international linear collider community became very active. The HARP[22] experiment was approved also with the argument that the resulting data will be useful to verify final state generators in hadronic simulation engines, and has itself adopted Geant4 for their simulation program. For LHC[23], from 1999 onwards, with the advent of "educated guess physics lists", and a set of agreements of the hadronic working group of geant4 with individual detector groups, a very solid validation program based on test-beam comparison was put together. Much later this was harnessed into LCG[24] validation meetings, and Fabiola Gianotti did a good job in keeping the effort alive. In parallel, during the year 2000, it became clear that for the purpose of LHC hadronic event simulation test-beam studies were not sufficient to address the validation of a simulation program in full. With the kind of integral energy flows and complexities involved in full detector simulation at LHC, robustness of the physics codes, but also of geometry, transportation and in particular magnetic field transport is a major issue. An agreement was found between the hadronic working group of Geant4 and the CMS [25] simulation management that CMS would focus their validation efforts mainly on the robustness aspect, while ATLAS[26] would focus mainly on the test-beam. CMS and LHCb[27] have begun to look into aspects of radiation shielding. Today three of four LHC experiments have adopted Geant4 as their primary simulation infrastructure, and have their simulation programs in routine use. The work for the physics TDR of CMS is based on Geant4 simulated data with the QGSP physics simulation engine. The number of full LHC events simulated today approaches 109. 5. Conclusions After 11 years of consistent effort, Geant4 today comes with the most complete physics modeling abilities. It is the de-facto standard detector simulation tool in High Energy Physics.

530

References 1. S. Agostinelli et al, Nucl. Instrum. Meth. A 506, 250 (2003). 2. S. Giani et al, Geant4: CERN/LHCC 98-44, 1998. 3. R. Brun, F. Bruyant, A.C. McPherson, M. Maire, P. Zanarini, CERN Data Handling Division, DD/EE/84-1, 1987. 4. J.P. Wellisch, Comput. Phys. Commun. 140, 65 (2001). 5. S. Agostinelli et al, Nucl. Instrum. Meth. A 506, 250 (2003). 6. M. Maire et al, Geant4 electromagnetic physics, Prepared for Monte Carlo 2000, Lisbon, Portugal, 23-26 Oct 2000 7. G. Folger, V.N. Ivanchenko and J.P. Wellisch, The European Physical Journal A, Vol. 21 No. 3,407 (2004). 8. P.V. Degtyarenko, M.V. Kossov and H.P. Wellisch, Eur. Phys. J. A 9 (2000) 421; Eur. Phys. J. A 9 (2000) 411; Eur. Phys. J. A 8 (2000) 217. 9. G. Folger, J.P. Wellisch, „Parton string models in Geant4", Prepared for Monte Carlo 2005, Chattanooga, TN, USA, 18.-22. Apr 2005. 10. V. Lara, J.P. Wellisch, Prepared for "Computing in high energy and nuclear physics", Padova, Italy, 7.-11. Feb 2000, pp52-55. 11. J.P. Wellisch, "Hadronic Simulation Engines for Geant4", Prepared for Monte Carlo 2005, Chattanooga, TN, USA, 18.-22. Apr 2005. 12. A.B. Kaidalov, K. A.Ter-Martirosyan, Phys. Lett. B117 247 (1982); Sov. J. Nucl. Phys. 39 (1984) 979. 13. J.J. Whitmore et al, Zeitschrift fur Physics C 62, 199 (1994). 14. H.P. Wellisch and D. Axen, Phys. Rev. C 54, 1329 (1996). 15. CERN High Energy Analysis Group records, preprint denominations CERN-HERA-YY. 16. T.W. Burrows, "The Program RADLST", Brookhaven National Laboratory Report BNL-NSC-52142, February 29, (1988). 17. ATLAS LARG unit, ATLAS Liquid Argon Calorimeter technical design report, CERN/LHCC 96-41, Dec 1996. 18. B. Dowler et al, Nucl. Instrum. Meth. A 482 (2002) 94. 19. B. Aubert et al, "The BaBar detector", Nucl. Instr. Meth. A 479, 1 (2002) 20. AMS collaboration, "AMS: Alpha magnetic spectrometer on the international space station", Nucl. Instrum. Meth. A 535, 134. 21. The Panda Collaboration, "Technical progress report for PANDA", from http://www.epl.rub.de/~panda/db/papersDB/PC19-050217_panda_tpr.pdf 22. The HARP Collaboration, CERN-SPS/2003-027, SPSC-P-325. 23. LHC Conceptual Design Report, CERN/AC/95-05. 24. F. Gianotti et al., CERN-LCGAPP-2004-10. 25. The CMS Collaboration: The Compact Muon Solenoid Technical Proposal, CERN/LHCC 9438, LHCC/P1 1994. 26. ATLAS Collaboration, Technical Design Report, CERN/LHCC/99-01. 27. LHCb collaboration, LHCb Technical Proposal, CERN/LHCC 984.

High Energy Physics Organizers: L. Price (High Energy Physics)

R. Ruchti (Accelerator and Data/Networking Developments)

E. Barberio M. Barone P. Casado A. Di Simone A. Di Ciaccio R. Gardner M. Krammer K. Lang W. Liebig P. Paolucci Fr. Pastore J. Puerta-Pelayo

R. J. J. A.

Schmidt Smyrski Ulbricht Ventura

Fast Shower Parametrisation for the Electromagnetic Cascades Technical Aspects of the Manufacture of 4 Grease Pad Assemblies for the CMS Experiment Implementation and Performance of a Tau Lepton Selection within the ATLAS Trigger System at the LHC ATLAS Detector Simulation: Status and Outlook The ATLAS Experiment at CERN: Two Years Before the Start of the Data Taking Progress in Grid Computing for Particle Physics Design, Production and Quality of the Sensors for the Silicon Strip Tracker of CMS Running MINOS ATLAS Inner Detector Results from the 2004 Combined Test Beam Data The CMS Muon System Performances of the ATLAS Level-1 Muon Trigger Processor in the Barrel Certification, Installation and Commissioning of Muon Drift Tube Chambers for the CMS Central Muon Detector The LHC Collider: Status and Outlook to Operation PANDA: a Detector for Research with Antiprotons Test of QED with the Reaction e+e~ -» 77(7) Muon Reconstruction and Identification for the Event Filter of the ATLAS Experiment 531

This page is intentionally left blank

FAST SHOWER PARAMETERISATION FOR ELECTROMAGNETIC CASCADES

J. WENG CERN & University of Karlsruhe Switzerland / Germany E-mail: Joanna. [email protected] E. BARBERIO AND A. WAUGH Univerity of Melbourne, University of Sydney Australia E-mail: [email protected], [email protected]

We developed a package for parameterizing electromagnetic showers in the context of Geant4. The algorithm is based on previous work done for the GFLASH package, but has been adapted to the new simulation context and modified to better reproduce the performance of the LHC calorimeters. The algorithm uses a probability density function (PDF) of the shower profile for electrons/positrons with an energy above 800 MeV. This PDF includes shower shape and energy fluctuations. A mixture of full simulation and fast parameterisation is also possible. This new package, included in Geant 4.7, leads to a significant gain in simulation time for LHC events, with minimal sacrifice of accuracy. This algorithm has been integrated within the ATLAS and CMS simulation framework and compared to full simulation.

1. I n t r o d u c t i o n T h e ATLAS and CMS detector simulation is based on t h e G E A N T 4 package 3 . G E A N T 4 simulates detector effects on physics events using a detailed microscopic description of the interactions between particles and m a t t e r . It is very accurate, but it can be very time consuming. In particular the simulation of electromagnetic cascades in calorimeters is expected t o account for a considerable amount of the total simulation time. Since it increases almost linearly with the energy absorbed in the detector, at LHC energies of 14 TeV, fully simulating one event, using individual particle tracking, may take several minutes. W h e n a large number of simulated events are needed for physics analyses, a full simulation approach is not viable .

533

534 In the past, within ATLAS and CMS, stand-alone fast simulations (6 and 5 ) were developed as an alternative to full simulation, however, this approach is not sufficiently accurate for many analyses; for example, studying pathological events that may look like a discovery. The running of past HEP experiments, CDF, HI, Zeus proved the importance of an intermediate level of simulation, faster than full simulation, but integrated in the same framework so that it could provide the same kind of output as full simulation on which the full reconstruction chain could be run. A set of equations, derived from the HI and Zeus parameterisation, can be used to parameterize the electromagnetic shower development in different calorimeters 1>2. These equations were implemented in the GEANT3 (GFLASH) framework. In this paper we present the implementation of an electromagnetic shower parameterisation inside the Geant4 3 framework. Shower parameterisations based on this approach have also been integrated into the simulation framework of ATLAS and CMS to speed up the simulation of the ATLAS liquid argon sampling and the CMS homogenous lead tungstate electromagnetic calorimeter. 1.1. Theoretical

background

The spatial energy distribution of electromagnetic showers is given by three probability density functions (PDF), dE(r) = Ef(t)dtf{r)drf{<)>)dct>,

(1)

describing the longitudinal, radial, and azimuthal energy distributions. Here t denotes the longitudinal shower depth in units of radiation length, r measures the radial distance from the shower axis in Moliere units, and (j) is the azimuthal angle. In , it is assumed that the energy is distributed uniformly: /() = 1/27T. The mean longitudinal profile of a shower is described by : {

E

dt

'

n t )

T(a)

[Z)

where t is the shower depth in units of radiation length and E is the shower energy. The longitudinal position of the shower maximum is T = (a —1)//3. For the radial distribution, f(r), the original GFLASH and CMS uses a twocomponent Ansatz as used in \ while ATLAS follows a description proposed in 2 . In both cases the mean and the sigma of fn(T) and In (a) are used to parameterize the longitudinal profile, as they are Gaussian variables. For an homogeneous calorimeter they parameterize as a function of the critical

535

energy, the material of the calorimeter and the direction of the incident particle, as a function of the shower energy. The inhomogeneous material of a sampling calorimeter influences the shower shapes. The shower maximum occurs earlier than in an homogeneous calorimeter with the same effective material properties. Hence, ln(T) and ln(a) parameterize also as a function of the sampling frequency and the value of e/MIP averaged over the whole shower. 2. Use cases of the new shower parameterisation 2.1. Shower parameterisation

in Geant 4

The new implementation of the GFLASH parameterisation has been integrated into Geant4 and is available from version 4.7.0 onwards. The following table shows the performance of the shower parameterisation package in a simple test setup: a block of lead tungstate, the material the CMS calorimeter consists of. The parameters needed for the parameterisation are calculated on the fly for the considered material. The speed-up is significant and reaches 100 times and more for electrons above 50 GeV.

Table 1. Speed-up of single electrons in a pure lead tungstate cube in Geant4 version 4.7.0)

Electron Energy

Time/event full Geant 4

Time/event parameterisation

Speed-up factor

1 GeV 5 GeV 10 GeV 50 GeV 100 GeV 500 GeV 1000 GeV

0.10 0.46 0.92 4.60 9.37 46.50 91.75

0.006 0.009 0.013 0.045 0.080 0.312 0.566

16 48 67 102 117 149 162

In general, a somewhat worse agreement with Geant4 is observed than with GEANT3 - especially for the radial profiles. This is one the hand due to technical differences (ex. shower start point), on the other hand the parameterisation parameters obtained from GEANT3 do not describe Geant4 shower profiles with equal accuracy. Here a retuning may be necessary. Typical longitudinal and radial profile are shown in Figure 1.

536 1 1 111

10

\f

1 i

10' 10

s

3

Iff

10*

; ! !

I,

: ! !

(D

jH-

•

S

102

:

S.

[

n

"'"''•Si^-

!

101

5

102

!

1CT

3

* ! 10

12

14

*v 1 16

f!

i

U !

18, , • 120

r[cm]

x[cm]

Figure 1. Longitudinal (left) and radial shower profile (right) for 50 GeV photons in pure lead tungstanate (PbWOi). The points correspond to the parameterisation (radial profile retuned), the histogram to the full Geant4 simulation.

2.2. Shower parameterisation

in CMS

The concept of parameterized showers has been included into the CMS full detector simulation OSCAR 4 . It is used to parameterize electrons and positrons in the barrel and endcap electromagnetic calorimeter. The radial profiles were retuned. Comparisons between the GFLASH based and the full OSCAR simulation of the energy depositions in the central crystal, the 3x3 and 5x5 crystal matrices show good agreement to the > 1% level, as illustrated in Fig. 2. The transverse and longitudinal shower profiles are also well modeled to within 1-3%. The shower parameterisations allow a significant time performance gain in the simulation, with speed increases in the range of a factor of 3-10. The gain for a simulation including the full detector geometry depends on the event type, the particle energy and the detector r\ region. For instance, a single electron or photon with an energy of 100 GeV in the ECAL barrel is simulated 10 times faster using parameterisation. For a large extra dimensions full signal event, pp —> j+G, with a single photon above 1000 GeV, the gain in speed is around a factor of 4. 2.3. Shower parameterisation

in

ATLAS

The new implementation of the shower parameterisation has been integrated into the ATLAS detector simulation. The longitudinal profiles show acceptable agreement and needed only small adjustments. The radial profile is derived by fitting the core and the tail of the radial distribution, once the longitudinal parameters are fixed. The number of spots neces-

537 103

::::!::::!::::te±::fc±:::!=!::::!::::i

1

"i

102 *a

10

*'

Jgf

=

=zzjf;jl!Sp pEE: 1

liiHiii: _ :

10

20

30

"i

40

i

.

.

1 . "3 i . , . . i , . . , i

50

60

70

Energy /GeV Figure 2. Energy depositions in a 5x5 crystal matrix for 50 GeV electrons. The histogram corresponds to the full Geant4 simulation and the markers to the shower parameterisation.

sary to correctly reproduce the radial fluctuations is very large. For the time being, an alternative function is used, that well describes the core and tail region of the radial profile. As described elsewhere, the ATLAS electromagnetic calorimeter has a 'accordion' geometry; for this reason, it was not initially clear if the procedure developed for a standard sampling calorimeter would be applicable. However, if the proper sampling frequency and e/MIP are used, the parameterisation described above reproduces the longitudinal shape of electromagnetic showers over a wide range of energies (0.5-100 GeV). The radial profile is fitted using a two parameter function which takes into account the dependence on energy, E, and shower depth, T. After first fitting for E, radial parameters are derived as a function of r . 3. Conclusion We implemented an electro-magnetic shower parameterization inside the simulation framework of Geant4. Experiment specific versions based on this implementation has been intergrated in the detector simulation of CMS and ATLAS. After tuning of the longitudinal and radial shower profiles the shower model yields good agreement with the Geant4 particle tracking and has proven to significantly speed-up the simulation in CMS and ATLAS. References 1. G.Grindhammer, M.Rudowicz and S.Peters, NIM A290(1990)469 G.Grindhammer and S.Peters,hep-ex/0001020

538 2. J. del Peso and E. Ros, NIM A306(1991)485-499 3. Geant4 see http://wwwinfo.cern.ch/asd/geant4/geant4.html and documentation therein. 4. OSCAR: CMS Simulation Package see http://cmsdoc.cern.ch/oscar 5. FAMOS: CMS Fast Simulation Package see http://cmsdoc.cern.ch/famos 6. E. Richter-Was, D. Froidevaux, L. Poggioli, ATL-PHYS-98-131

TECHNICAL ASPECTS OF THE MANUFACTURE OF 4 GREASE PAD ASSEMBLIES FOR THE CMS EXPERIMENT M. BARONE Institute of Nuclear Physics "NCSR "Demokritos "-Athens-Greece e-mail: [email protected] P. L. RIBONI ETHZ-Zurich-Switzerland;

emaihpierluigi. riboni@cern. ch C. SPITAS

Technical University of Crete, Dept of Production Engineering Greece; e-mail: [email protected]

and

Management-

V. SPITAS Technical University of Crete, Laboratory of Applied e-mail: vspitas@central. ntua.gr

Mechanics-Greece

The CMS detector is a worldwide endeavour gathering an extended number of contributors and putting together many parts coming from different countries. In this context CERN operates like a sort of 'general coordinator' and is obliged to put a special emphasis on strict criteria of quality, calculations, control certificates and full traceability of materials and operations, which are all of paramount importance to the successful integration and function of the parts. This paper discusses the technicalities related for the selection of machining parameters to optimally balance geometrical/ dimensional accuracy of the produced parts and their cost-effectiveness. These are the components of the grease pad assemblies that support and allow the precise positioning and alignment of the CERN CMS apparatus. The work, an in-kind Greek contribution, mainly consists of machining several large steel plates to various forms, including large pockets and inclined surfaces.

1. Introduction CERN's current important project is the Large Hadron Collider, LHC, which will enter in operation by the middle of 2007. LHC is a super-conducting particle accelerator of 27 km in circumference being installed in an existing underground tunnel, previously devoted to another accelerator, the LEP. One of the 4 experiments to be performed at the LHC is the CMS (Compact Muon Solenoid). 539

540

This experiment is built and funded by an international collaboration of highenergy physics institutes from thirty-six countries and by CERN. The CMS experiment consists of a massive magnet equipped with several dedicated particle detectors. The magnet system consists of a 4 Tesla solenoid Superconducting Coil, having a free bore of 6 m and a length of 13 m, enclosed in a Return Yoke comprising the Barrel Yoke and the two End cap Yokes. The Return Yoke is designed as a regular twelve-sided structure. The important dimensions of the completed detector are: a length of 21.6 m, an outer diameter of 14.8 m, and a total mass of 12'500 tonnes. On both extremities of the Yoke is located one Hadronic Forward Calorimeter (HF). It is supported by a mechanical lifting system and centered on the beam pipe. This paper covers the manufacture of mechanical components of 4 grease pads for the YE3 disc. 2. Organization of the work The CMS experiment is financed by its supporting collaboration through mutually agreed procedures. A CERN team was asked to organize and specify the production of the grease pads for the Endcap discs, the Greece representative of the CMS collaboration was in charge of the tendering, contract and follow-up of the manufacture and supply by the firm Metallo in Athens, as in-kind contribution. Consultants from the Technical University of Crete have played an important role from the tendering till to the acceptance of the goods. The supply was covering the following points: 1) The Engineering Report (ER) including la) The bill of material, lb) Definition of machining procedures, lc) Definition of control procedures, Id) The Quality Assurance Manual, 2) Base materials certificates, 3) Machining, 4) Dimensional and conformity measurements, 5) Protection against rust and conservation, 6) Packaging, 7) Transport and delivery to Point 5 Cessy, France. In this context CERN operates like a "general contractor" therefore is obliged to make controls to get quality and the full traceability of the materials and operations which are all of primary importance to the successful integration and function of the parts. 3. Grease pads' description Each pad is fitted with two movable cylindrical pots inside cavities of appropriate shape. The pots are free to slide on a mating stainless steel plate and pressurised grease contained by means of special gaskets keeps the surface friction coefficient to a minimum and prevents damage to the sliding surfaces.

541

Additionally, to compensate for a 1.23° inclination of the cavern floor at point 15, where the final detector assembly is to take place, a further set of angular plates per pad is foreseen, so that each final grease pad assembly is a superposition of no less than 7 plates of specialised shapes, as shown in Fig. 1. The total weight per pad is about 2.5 tonnes. The detailed work description is contained in specification CMS-SpF-4181 [1]. 4. Fabrication process The proper function of the grease pads can be severely impaired by geometrical deviations caused by the fabrication process, which chiefly involves material removal over large surfaces by milling. Such deviations can be the result of 1) residual stresses 2) static deflections and/ or 3) dynamical deflections. Residual stresses can in turn either la) be induced by raw material manufacture, meaning that they are present in the hot rolled stock delivered from the mill and cause bending when material is removed in an asymmetrical manner, or lb) be induced by heat and plastic work generated locally during cutting. Residual stresses can be measured by the hole-drilling method either by use of strain-gauges [2] or laser speckle interferometry [3], by reflective photoelasticity [4] or x-ray diffraction [5], Stresses of type (lb) mostly affect thin and intricate geometries, and are furthermore not normally a problem unless a poor selection of machining parameters is made, resulting in excessive heating of the workpiece. If the machining parameters are carefully monitored and the chip is checked for visual signs of excessive heating (shape, length and colour), then such residual stresses have a negligible effect. Statical deflections (2) are a direct result of the cutting force itself, and tend to deform the work-piece by compression, bending and/ or torsion. Depending on the magnitude of this force and the combined stiffness of the work-piece, fixture and machine tool, such deflections can range from quite insignificant to catastrophic. Large work-piece overhangs, thin-sectioned beams and knife-edges are all associated with low rigidity and are at a greater risk of causing significant static deflections. It is not always possible to increase the stiffness without conflicting with other significant operational and/ or manufacturing requirements, so, when deflections must be closely controlled, the cutting force must be kept in check by suitably selecting the cutting parameters. Finally, dynamical deflections (3) are produced by the same mechanism as static deflections. The cutting forces consist of a fixed and a periodically variable vector component, the former producing static deflections and the latter dynamical deflections, tool chattering and noise. Dynamical deflections in particular tend to produce defects that are in most cases excessive roughness or waviness and are therefore less critical to dimensional accuracy, though they do have a marked effect on surface integrity and tool life. Compared to static

542

deflections, however, vibration is quite easier to spot, due to the noise and characteristic tool failures that come with it.

CMSSYBXM0076

Figure 1. Sub-assembly of a grease pad manufactured by firm Metallo under contract no. CA 1305972 for CERN, identifying plates CMSSYBXM0071 and CMSSYBXM0076

Though vibration problems can be solved solely on empirical basis most of the time, a machining set-up that is prone to static deflections is also bound to suffer from excessive vibration. Therefore remedies towards decreasing static deflections, i.e. reducing the cutting forces or enhancing stiffness, will often improve the dynamical response as well, though additional measures may have to be taken. In relation to the above, and given the high performance demands placed on the grease pads, this paper presents in the form of an engineering analysis some key technical considerations that have gone into the manufacture by the Hellenic firm Metallo of 4 such assemblies under CERN contract no. CA 1305972. Table 1

Goals: flatness of large plates according to cost-effective machining form tolerances Means: maximum depths of cut & maximum low deflections feeds ...hence high cutting forces ...hence high deflections & residual stress/ strain fields ...hence accuracy of form may be compromised The analysis treats the matter of geometrical and dimensional accuracy, as is influenced by the aforementioned factors, and calculates the effect of machining

543

parameters on statical deflections and form accuracy with the aim to produce tangible manufacturing guidelines. Only the most critical processes and factors, as identified by the analysis, are isolated and analysed, whereas typical procedures and work-piece geometries are not considered. This analysis is triggered by the need to establish cost-effective manufacturing procedures and parameters without compromising the tolerance specifications, as is shown in Tab. 1 5. Analysis a) Deflections and tolerances An explanation of the modelling and calculus used in the FEA analysis can be found in [8]. 'initial selection of tool/ machining parameters for max. productivity

verification with machine power readout to check validity of estimate

calculation of material removal rate

calculation of deflections using FEA

estimation of cutting forces based on specific power

trial cut

instruct machine shop and produce first prototype

Figure 2. Analysis flow chart

The main stage in the analysis is to evaluate the effect of the calculated deflections on the geometrical integrity of the work-pieces. Machining an elastically deformed piece means that material is removed in a different way than anticipated. This can affect both geometrical and dimensional tolerances. Main problems to consider are 1) localisation of defects, 2) large cumulative deflections, 3) large dynamical cutting force components (vibration). Engineering materials can present random deviations from their assumed properties, which affect the forces involved in machining. As a rule, better quality materials are associated with smaller standard deviations, but the stochastic nature of the problem means that extra allowances must be made to ensure that no parts are compromised. Keeping maximum errors one order of

544

magnitude smaller than the official tolerances is a safe rule of thumb for prototype production. Based on the above analysis, the employed methodology is as shown in Fig. 2. 6. Results and discussion Plate CMSSYBXM0071 This is the plate that houses the grease pots. Its general geometry and way of fixturing on the machine table is depicted in Fig. 3. Due to machine limitations, more rigid fixturing alternatives [9] were not deemed practical. The long side slots were identified as potential problem areas and an analysis was called for to determine if the initial cost-effective choice of milling parameters was causing excessive deflections. In particular, two distinct cases were considered: case 1- loading at the edge of a small cantilever beam feature (localised deflection?) case 2- midpoint loading (large overall deflection?)

Figure 3. Basic work-piece orientation for plate CMSSYBXM0071

Figure 4. FEA results for plate CMSSYBXM0071

The FEA (Fig. 4) established that case 1 produced the most unfavourable conditions (almost 3 times the displacement of case 2), but at under 3% of the tolerance limit specified for plate CMSSYBXM0071, the choice of rnacliining parameters, proposed for cost effectiveness, was deemed safe. 7. Conclusion In this paper an applied methodology was presented to estimate the effect of tool and machining parameter selection on the cutting forces and resulting workpiece deflection. This methodology has been applied on the manufacture of 4 grease pad assemblies for the CMS detector and a selected example of the analysis regarding parts CMSSYBXM0071 has been given.

545 The results suggest that factors such as part fixturing/ orientation along with the tool selection and machining parameters have a significant effect on the induced deflections. Non-rigid features, knife-edges and large part overhangs are all high-risk cases that deserve special analysis, especially if cost-effective production parameters are to be chosen. If the initial choice is unsatisfactory, then this process of analysis- parameter selection can evolve as an iterative cycle until a satisfactory solution is reached. Benefits that have been observed from the application of this process are: • Less of the usual trial-and-error (without excessive safety margins) • Better-controlled quality • Workshop savings in material and man-hour resources This paper focuses purely on the deflection issues associated with machining parameter selection. Other important aspects involved in the selection, such as controlling maximum roughness, chip evacuation, tool life etc lie outside the scope of this paper and have not been mentioned. Acknowledgements Most of this work was done within the framework of the CERN order no. CA 1305972, placed on firm Metallo. Permission by CERN and Metallo to present pertinent photographs of the grease pad assembly is kindly acknowledged. Permission by AIAS Engineering to publish 3-D part models and FEA results used in the project is also acknowledged. References 1. CERN specification CMS-SpF-4181 2. ASTM Standard E837-01, ASTM (2001) 3. Diaz F. U., Kaufmann G. H., Galizzi G. E., 33, p.p. 39-48, (2005) 4. Blum A. E., 13, p.p. 96-101 (1977) 5. Hilley M. E., Ed., Residual stress measurement by x-ray diffraction, SAE J784a, SAE, Warrendale, PA, p.p. 21-24 (1971) 6. Stirnmimann J., Kirchheim A., New cutting force dynamometers for highprecision machining, Industrial Tooling Conference, Southampton (1997) 7. ASM Handbook Vol. 16 'Machining', ASM International (1995) 8. Zienkiewicz O. C , Taylor R. L., The Finite Element Method, McGraw-Hill (1989) 9. W. E. Boyes, Handbook of jig and fixture design, 2nd Edition, SME (1989) 10. SECO Main Catalogue and Technical Guide 'Milling Cutters and Inserts', SECO

IMPLEMENTATION A N D P E R F O R M A N C E OF A TAU L E P T O N SELECTION W I T H I N T H E ATLAS T R I G G E R SYSTEM AT T H E LHC

A. dos Anjos, S. Armstrong, J.T. Baines, C.P. Bee, M. Biglietti, A. Bogaerts, M. Bosman, B. Caron, P. Casado, G. Cataldi, D. Cavalli, G. Comune, P. Conde, G. Crone, D. Damazio, A. De Santo, M. Diaz Gomez, A. Di Mattia, N. Ellis, D. Emeliyanov, B. Epp, S. Falciano, H. Garitaonandia, S. George, V. Ghete, R. Goncalo, J. Haller, S. Kabana, A. Khomich, G. Kilvington, J. Kirk, N. Konstantinidis, A. Kootz, A.J.. Lankford, A. Lowe, L. Luminari, T. Maeno, J. Masik, C. Meessen, A.G. Mello, R. Moore, P. Morettini, A. Negri, N. Nikitin, A. Nisati, C. Osuna, C. Padilla, N. Panikashvili, F. Parodi, E. Pasqualucci, V. Perez-Reale, J.L. Pinfold, P. Pinto, Z. Qian, S. Resconi, S. Rosati, C. Sanchez, C. Santamarina, D.A. Scannicchio, C. Schiavi, E. Segura, J M. de Seixas, S. Sivoklokov, A. Sobreira, R. Soluk, E. Stefanidis, S. Sushkov, M. Sutton, S. Tapprogge, S. Tarem, E. Thomas, F. Touchard, G. Usai, B. Venda Pinto, A. Ventura, V. Vercesi, T. Wengler, P. Werner, S.J. Wheeler, F.J. Wickens, W. Wiedenmann, M. Wielers, H. Zobernig THE ATLAS HIGH LEVEL TRIGGER GROUP * presented by P. CASADO + IFAE, Edifici Cn, Bellaterra, E-08193, Spain

The ATLAS experiment at the Large Hadron Collider (LHC) has an interaction rate of up to 109 Hz. The trigger must efficiently select interesting events while rejecting the large amount of background. The First Level trigger will reduce this rate to around 0(75 kHz ). Subsequently, the High Level Trigger (HLT), comprising the Second Level trigger and the Event Filter, will reduce this rate by a factor of O(103). Triggering on taus is important for Higgs and SUSY searches at the LHC. In this paper tau trigger selections are presented based on a lepton trigger if the tau decays leptonically or via a dedicated tau hadron trigger if the tau disintegrates semileptonically. We present signal efficiency with the electron trigger using the data sample A —> TT —• e hadron, and rate studies obtained from the dijet sample. 'http://atlas.web.cern.ch/atlas/groups/daqtrig/hlt/authorlists/como2005.pdf tWork supported by the Institut de Fisica d'Altes Energies (IFAE) and Universitat Autonoma de Barcelona

546

547

1. Introduction The Large Hadron Collider(LHC) is expected to start data taking in 2007 at CERN (European Organisation for Nuclear Research). Proton-proton collisions will be produced at a center of mass energy of 14 TeV and design luminosity of 10 3 4 cm~ 2 s - 1 . ATLAS is a multipurpose detector which will have to cover a wide variety of aspects of high energy physics phenomenology : from discovering new physical phenomena to performing precision measurements. In order to achieve this task, the detector is composed of different tracking subdetectors, a solenoid, electromagnetic (e.m.) and hadronic calorimeters, muon subdetectors and a toroid, as well as the Trigger and Data AcQuisition systems (TDAQ) 1 . 2. The ATLAS Trigger system (1) The ATLAS Trigger system is organised in three levels. The first level (LVL1) is hardware-based and has to reduce the input rate from 40 MHz to 75 kHz in less than 2.5 /xs. The result of the LVL1 selection contains information about the type of trigger and position of the possible particle candidates that cause the event to be accepted. After a positive LVL1 decision the data are transferred from electronic pipeline memories to distributed buffers in so called Read-Out Systems (ROSs). The second (LVL2) and third (Event Filter -EF-) levels are software based systems running on linux PC farms. They are referred collectively as High Level Trigger (HLT). They use full granularity data of all detectors and can also combine information from different detectors. The LVL2 accesses a few percent of the total detector data in a Region of Interest (Rol) provided by the LVL1 result by direct network request to the ROSs. This reduces the event rate to about 2 kHz. After a positive LVL2 decision the complete event is assembled and made available to the EF, which has access to the full detector data and the latest calibration and alignment information. Events accepted by the HLT are stored on tape for further analysis. The goal is to achieve an average decision time of 10 ms and 1 s for LVL2 and EF, respectively, although the system could easily scale to accommodate larger execution times if needed. (2) The HLT selection software 2 has been designed with Object Oriented methodology and is implemented in C + + . It can run in the LVL2, EF and offline environments. This is possible because the differences between the environments are hidden by well defined interfaces and the same framework

548

(Athena 3 ) is used. The HLT selection software is subdivided into four main sub-packages: (a) The Steering organizes the processing of the HLT algorithms, (b) The Event Data Model (EDM) covers all data entities in the event, (c) The Data Manager provides the means for accessing the event data during the trigger processing, (d) The HLT Algorithms can either reconstruct the event or check hypothesis about previously calculated quantities. (3) The data access of the HLT algorithms is a fundamental part in the design of the HLT architecture. The HLT algorithm requests data in a Rol using the offline transient data store (Athena Storegate 4 ) as interface and the Region of Interest package, a software component which returns Storegate pointers to the data inside the specified detector geometrical region. These data are in form of C + + objects, convenient for the algorithm (e.g. calorimeter cells which hold energy and position), contrary to the data coming from the detector electronics which is in raw format (i.e. a binary file with the read-out channel information). 3. Tau selection The selection of taus is important at the LHC for the discovery of new particles. Taus appear in decay modes with significant branching fractions, for example in charged Higgs, in A —> TT (A being the pseudo-scalar Higgs from the Minimal Supersymmetric Standard Model). In this paper we focus on A —> TT -+ e hadron which represents 25% of the A —» TT production. The tau can decay via two modes: leptonically, with one electron or muon in the final state (we only consider here the electron decay); or semileptonically, with hadrons in the final state. In both cases the HLT selection procedure follows the Region of Interest mechanism. The data request is restricted to a region with different size for the various trigger. In the case of the tau decaying into an electron, one can use the single isolated electron trigger of ATLAS. At LVL1 calorimeter information is used with a segmentation in trigger towers of 0.1x0.1 in A77 x A. A cluster is defined with the two most energetic pairs of e.m. towers (adjacent in eta or phi) from the central 2x2 core of the 4x4 window. An isolation cut is applied in the outer region of the 4x4 region and in the hadronic calorimeter. At LVL2, the energy and position measurements obtained at LVL1 are refined; the leakage into the hadronic calorimeter is evaluated and variables related to the shower shape in the e.m. calorimeter are used to perform preliminary particle identification. If a candidate is found to be

549

Figure 1. E.m. radius (left) and isolation fraction (right) for taus decaying hadronically (black) and jets (red).

consistent with an electron, track reconstruction is performed in the Inner Detector (ID); cluster to track association is done using (77, (j>) matching criteria, achieving further rejection against fake candidates; in case the matching was successful, the ET/PT ratio between the transverse energy measured in the e.m. calorimeter and the transverse momentum of the corresponding ID track is evaluated for particle identification. If the objects under analysis fulfill the required signatures, the event and its LVL2 result are passed to the EF, where more precise clustering algorithms can be performed, and where information on the complete event is available, along with more precise calibrations and alignment constants. In the case of tau decaying into hadrons one has to use the genuine tau trigger, used in conjunction with missing ET at pr « 35 GeV, and standalone for pr « 60 GeV. At LVL1 the cluster is defined from the e.m. towers as in the electron case and in addition, the inner core of the hadronic calorimeter is used. The isolation region is defined from the outer part of the 4x4 window of the e.m. and the hadronic calorimeters. ET of the cluster can be higher than for the electron trigger while the applied isolation cut is looser. The selection of LVL2 and EF has as a guidance the offline. In this case two of the variables with highest rejection power are depicted in Fig. 1. The rejection factor obtained using the whole selection is 200 for an efficiency of 50%. At LVL2 we are using previously considered variables and new quantities imported from the offline. The list is the following: (1) Total energy (ET) calculated in an inner and outer region for the e.m. and the hadronic calorimeters. (2) The e.m. radius: energy weighted radius of the cluster calculated in the e.m. calorimeter. The distribution of this variable for the offline is shown in Fig. 1 (left). (3) The width in the energy deposition: energy weighted standard deviation in 77. (4) The isolation fraction: fraction of energy deposited in a region of

550

Figure 2. Efficiency of the electron trigger as a function of the PT of the electron (red) and as a function of the PT of the tau (black).

0.1 < R < 0.2. T h e distribution of this variable for signal and background is also shown in Fig. 1 (right). 4.

Results

T h e sample A -> TT -^ ehadron at low luminosity with electronic noise and pile u p has been used to calculate efficiencies of the isolated electron trigger as a function of p r of the electron and of pr of the r , normalizing with respect to events having an offline reconstructed electron. This is shown in Fig 2. For the electron case, there is the expected sharp rise between 20 and 30 GeV and the maximum value is « 85 %. These results confirm those obtained in the collaboration for electrons a t py = 25 GeV and extend t h e m to a broader region in pr- For the taus, the slope is much less sharp, the trigger reaches an efficiency plateau for pr > 70 GeV. In the lower pr region the electron isolated trigger on taus complements with the hadron t a u plus missing ET trigger. Finally, the HLT rates of two relevant triggers for these studies are estimated using the dijet sample with electronic noise a n d pile up a t low luminosity: for the electron isolated trigger the value is s»32 Hz and for the t a u and missing ET trigger the result is « 5 Hz. References 1. ATLAS High-Level Trigger, Data Acquisition and Controls, Technical Design Report, ATLAS TDR-016. ATLAS HLT/DAQ/ DCS Group. 2. "Architecture of the ATLAS online physics-selection software at LHC", Proceedings of the 8th ICATPP Conference on Astroparticle, Particle, Space Physics, Detectors and Medical Applications, 255-259 (2003). 3. http://atlas.web.cern.ch/Atlas/GROUPS/SOFTWARE/00/architecture 4. The StoreGate: a Data Model for the Atlas Software Architecture, Proceedings CHEP03, La Jolla, California, March 24-28, 2003.

ATLAS D E T E C T O R SIMULATION: STATUS A N D OUTLOOK

A. RIMOLDI University

of Pavia and INFN,

Italy

A. D E L L ' A C Q U A , A. DI S I M O N E , M. G A L L A S A N D A. N A I R Z CERN,

Switzerland

J. B O U D R E A U A N D V. T S U L A I A University

of Pittsburgh,

USA

D. C O S T A N Z O University

of Sheffield,

UK

The simulation program for the ATLAS experiment at CERN is currently in a full operational mode and integrated into the ATLAS's common analysis framework, ATHENA. The robustness of the application was proved during the Data Challenge 2 (DC2). The simulation software, based on GEANT4, has been interfaced within ATHENA and to GEANT4 itself using the LCG dictionaries and Python scripting. The Python interface has added the flexibility, modularity and interactivity that the simulation tool needs to manage, in a common way, the different simulation setups: full ATLAS detector, test beams and cosmic ray studies. The comparison with real data has been possible in the context of the ATLAS Combined Test Beam (2004) and ongoing cosmic ray studies.

1. The Geant4 ATLAS detector simulation Starting from 2004, the GEANT4 simulation framework has replaced the previous GEANT3-based simulation program l as the only officially supported software for the ATLAS detector simulation. It consists of a fullfeatured, object-oriented simulation suite (G4ATLAS), integrated in the common framework for the ATLAS offline software (ATHENA) 2 . The implementation of G4ATLAS has been driven by key concepts like dynamic loading and action-on-demand. The resulting software is thus highlyconfigurable and flexible, and can therefore accommodate all the require-

551

552

ments the simulation of a complex experiment such as ATLAS calls for. The user can choose at run time which kind simulation to run (whether full detector, test beam setup or cosmic simulation). However, since the underlying simulation code is always the same, consistency is ensured throughout all the different applications. 1.1. Geometry

description

The geometrical description of the ATLAS detector has been decoupled from the simulation framework. A dedicated set of classes (GeoModel) provides the necessary volume and material description. The GeoModel description is automatically translated into a GEANT4-compatible geometry at run time. On the other hand, being GeoModel completely independent from GEANT4, the same description can be used also by reconstruction applications, thus ensuring total agreement between the geometries used for simulation and reconstruction. The GeoModel description has been optimized to handle the large number of volumes (about 106) involved in the description of the ATLAS experiment: the number of volumes itself is kept to the minimum and, whenever possible, volume parameterization is used to improve the performance. 1.2. Job

control

In the ATHENA framework, jobs are configured and controlled using the python scripting language: python commands are grouped in job Options files, which contain all the information about the job to run (e.g. which algorithms to run, what input files to use, the data to be written in the output file, etc.). The user has thus all the advantages from a full featured programming language, such as flexibility and ease-of-use, and can use the existing bindings to interact directly with the underlying C + + code. GEANT4, on the other hand, has no native interface to python and the configuration of a simulation job is normally achieved using specific commands, which can be grouped into macro files. For the simulation of a complex experiment such as ATLAS, the number of macro files involved increases rapidly, leading to maintainability issues. Moreover, the macro files mechanism does not integrate with ATHENA's python job configuration, resulting in an unacceptable lack of flexibility. To solve this problem, a python layer has been developed (PyG4Atlas) which allows, using bindings to the underlying C + + G4ATLAS code, to fully configure a simulation job by means of python commands. Integration in the ATHENA frame-

553

work is thus also achieved for what concerns job configuration, while the efforts requested to the users to interact with the simulation are kept to a minimum. 2. Examples of applications 2.1. Combined

test

beam

A full slice of the ATLAS detector (inner tracker, calorimetry, muon spectrometer) has been tested in 2004 on CERN's H8 beam line. The simulation infrastructure is successfully managing all the different aspects of such a complex experimental setup. Simulation of the subdetectors must be possible in both combined and stand-alone modes, while beams with different particles, energies and profiles must be simulated; extra detectors, such as scintillators, must be properly handled, together with some extra dead material used for material studies. All the above mentioned simulation parameters are indeed fully accessible from the python interface for the advanced user. Moreover, a database is provided associating the different configurations to a run number: a user only needs to specify in the jobOptions the run number, and the simulation infrastructure will automatically setup the corresponding set of parameters. 2.2. Cosmic

simulation

G4ATLAS provides full support for cosmic simulation allowing the user to choose a specific setup, consisting of the ATLAS detector, the ATLAS cavern, the rock overburden and the surface buildings. The geometrical description of the different elements is provided by GeoModel, and primary cosmic generation is done via a specific generator (CosmicGenerator) providing single muons at gound level. A scoring layer surrounds the ATLAS detector: particles crossing it will be saved to an external file, which can then be used to start a new simulation without having to propagate particles through the rock overburden.

3. Performance Being the simulation used for massive productions on the GRID, its performance in terms of CPU time per event and memory usage at run time needs to be constantly monitored. They are indeed measured at each nightly build of the ATLAS offline software, by a set of dedicated jobs using different sin-

554

gle particle samples. More detailed tests, including full physical events, are performed at each ATHENA release.

Table 1. Average CPU time per event for single particle samples. Average disk size per event is reported as well. # simulated events

CPU time per event (kSI2K)

Event size (KB)

Single muons Pt = 200GeV, |»7| < 6.0

1000

75.74

10.05

P t = 50GeV, \rj\ < 3.0

1000

1.93

7.9

Pt = 5GeV, \r,\ < 6.0

1000

1.97

3.2

Single pions P t = 50GeV, |»/| < 3.2

1000

63.13

34.25

Pt = 5GeV, \ri\ < 6.0

1000

17.43

33.29

Single electrons P t = lOOGeV, |JJ| < 6.0

1000

604.52

Pt = 50GeV, |TJ| < 2.5

1000

82.8

14.43

Pt = 5GeV, \t)\ < 6.0

1000

34.32

56.54

Table 1 shows the timing results obtained for ATHENA release 10.5.0 using single particle events. Concerning memory usage at run time, it is measured at different steps of any job startup (ATHENA initialization, GeoModel initialization and G4ATLAS startup), as well as during the event processing. In particular, at each nightly build, plots are automatically produced showing the memory usage as a function of the number of processed events, in order to easily spot any major software problem such as memory leaks. Figure 1 shows the results of such a measurement done using single muons in ATHENA release 10.5.0. A summary of the tests done in the past years can be found in Ref. 3. Concerning physics performance, the 2004 combined test beam has been an important benchmark to evaluate the reliability of the simulation, and to compare real data with simulation. For all the subdetectors, analysis has shown very good agreement between the simulated response and the one measured at the test beam. Moreover, reconstruction of full physical events provided results at least as good as the ones obtained with the previous GEANT3-based simulation.

555

m 1-558

£• 0 total

i £556 554

552 550 i

X_l

I

I

0

1 1 l__J

50

1 I

100

1 I

I

!

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

150

200

250 300 number of events

Figure 1. Memory usage at run time as a function of the number of processed events. The plot refers to single muons with Pt = bGeV. 4.

Conclusions

T h e simulation software of t h e ATLAS detector, implemented using t h e G E A N T 4 toolkit, is presently a robust and fully featured application suite, integrated in t h e ATLAS offline software. T h e needed flexibility, configurability and ease-of-use are achieved by means of python scripts, which allow t h e users to have full control on t h e simulation job. T h e simulation suite is being successfully used for massive production since 2004, and its performance is continuously monitored using a u t o m a t e d nightly tests, as well as more detailed measurements at each software release.

References 1. ATL-SOFT-2003-013, "Strategy for the transition from Geant3 to Geant4 in ATLAS", by D. Barberis, G. Polesello, A. Rimoldi. 2. The A T H E N A framework: http://atlas.web.cern.ch/Atlas/GROUPS/SOFTWARE/00/architecture/General/ Documentation/AthenaDeveloperGuide-8.0.0-draft.pdf

3. ATL-SOFT-PUB-2005-004, "Validation of the GEANT4-Based Full Simulation Program for the ATLAS Detector" by D. Costanzo, A. Dell'Acqua, M. Gallas, A.Nairz, N. Benekos, A. Rimoldi, J. Boudreau, V. Tsulaia,

THE ATLAS EXPERIMENT AT CERN: TWO YEARS BEFORE THE START OF THE DATA TAKING ANNA DI CIACCIO On behalf of the ATLAS Collaboration University of Roma Tor Vergata and INFN Tor Vergata, via della ricerca scientiflca, 00133 Roma, Italy The ATLAS experiment at the Large Hadron Collider, under construction at CERN, is the biggest high energy-physics detector ever built on an accelerator. It will be one of the two general purpose experiments installed at the LHC. The ATLAS detector is now approaching the end of the construction and it is entering the installation and commissioning phase. The status of the different sub-detectors will be described.

1. Introduction ATLAS [1] is a general purpose detector which is designed to exploit the full discovery potential of the Large Hadron Collider (LHC) that is being built at CERN, Geneva. The LHC [2] is a proton-proton accelerator which will reach an energy in the center of mass of Vs = 14 TeV, giving the possibility to investigate a wide range of physics up to masses of several TeV. The most prominent issue is the search for the origin of the spontaneous symmetry breaking mechanism in the electroweak sector of the Standard Model. One of the possible manifestation of such a mechanism could be the existence of a SM Higgs bosons (H) or a family of Higgs particles (H*, h, H and A) when considering the Minimal Supersymmetric extension of the Standard Model (MSSM). The search for supersymmetric particles and of other final-state physics signals beyond the Standard Model (heavier W and Z bosons, extra-dimensions, quark compositeness, etc) are possible thanks to a detector that has been designed to provide as many signature as possible using electrons, gamma, muons, jets and missing trasverse energy as well as b-quark tagging. The actual plan foresees the start of the LHC operation and of the detector data taking in summer 2007.

556

557

2. The overall detector concept The ATLAS detector, shown in Fig. 1, has been designed to have a large pseudo-rapidity coverage up to |n| < 5. It has a diameter of 24 m and a length of 44 m. The total weight is 7000 tons. The basic design criteria are the following: • an efficient inner tracking system for precise lepton momentum measurements and b-quark tagging; • a very good electromagnetic calorimeter complemented by a full coverage hadronic calorimeter for jet and missing trasverse energy measurement; • a stand-alone muon system for precision momentum measurement up to the highest luminosity and with low p T muon trigger capability. Muon Detectors

Barrel Toroid

Electromagnetic Calorimeters

Inner Detector

Hadronic Calorimeters

Shielding

Figure 1. A view of the ATLAS experiment.

3. The magnetic system The magnet configuration is based on a super-conductive solenoid with a field of 2 Tesla in the inner detector cavity and three super-conducting air-core toroids: a long barrel consisting of eight independent coils arranged with an eight-fold

558

symmetry outside the barrel calorimeter and two end-cap magnets. They generate a large field volume and a strong bending power within an open structure. Multiple scattering effects are therefore minimal and a high resolution and a high acceptance muon momentum measurement is achieved with three muon stations of high precision tracking chambers. The total bending power, integrated between the first and the last muon chambers, increases from about 3 Tm at n =0 to about 8 Tm at n =2.8. The whole magnet system represents a cold mass of 700 tons and a total weight of 1400 tons. The inner solenoid coil is integrated into the vacuum vessel of the barrel LAr calorimeter cryostat, thus eliminating the material and space of independent vessel walls. Recently the assembly of the eight coils of the barrel toroid has been completed and shortly after the barrel calorimeter system, that a year ago has been transferred into the pit, has been inserted in the central position at z=0.

Figure 2. A photography of the ATLAS barrel spectrometer assembled with its eight coils by the end of September 2005. At the other end of the hall, the barrel LAr and Tile calorimeter system is visible with in front the solenoid coils and cryostat.

4. The inner tracker The ATLAS inner detector uses three sub-detectors for tracking of charged particles from r = 5 cm to r = 107 cm inside a solenoid magnet of 2 T. The

559 innermost detector is a high resolution silicon pixel detector. It provides precise 3D tracking information close to the interaction point allowing secondary vertex reconstruction and hence b identification. It is followed by the SCT, a large area tracking device based on silicon strip detectors. The TRT, based on straw tubes, provides continuous tracking and improves electron identification due to its ability to detect transition radiation. In the barrel region the high precision detector layers are arranged in concentric cylinders around the beam axis, in the end-cap the detectors are mounted on disks perpendicular to the beam axis. The pixel layers are segmented in R and z- The SCT uses small angle stereo strips to measure both coordinates. The TRT is made of straw tubes parallel to the beam direction. 4.1.1. The pixel vertex detector The ATLAS silicon pixel detector consists of more than 1700 modules for a total sensitive area of 1.7 m2 and over 80 million pixel cells organized in 3 layers in the barrel and 3 disks for each of the end-caps. The basic unit is a silicon sensor with an active area of 60.8 mm x 16.4 mm, divided in pixels of 50um x 400um size. The pixels are read out by 16 front-end chips arranged in two rows and bump bonded to the sensors providing pulse height information by a time-over-threshold technique. In summer 2005 a corrosion of some of the cooling pipe of the staves has been found. It is caused by the presence of nickel in the Aluminum pipe endings (needed to braze the fittings) and by the exposure to water without a proper drying procedure. The galvanic currents has produced corrosion in 15% of the pipes that are now leaky. The problem has produced a six months delay in the pixel detector installation that is now foreseen only by February 2007.

4.1.2. The Semi conductive tracker The Semiconductor Tracker will be located between the innermost pixel and the outer Transition Radiation Tracker at radii between 25 and 50 cm. It consists of a central barrel section (4 barrel layers) and two end-caps (9 end-cap disks on each side) allowing for charge particle tracking up to a pseudorapidity of ± 2.5. The total number of silicon microstrip detector modules that will compose the SCT is 4088 with about 6 million read-out channels.

560

The radiation environment requires the detectors and electronics to be radiation hard up to a total fluencies of 2 x 1014 1 MeV neutrons/cm2 for the anticipated 10 years of operation. In order to limit the anti-annealing of the silicon detectors, the SCT will operate at an ambient temperature of -7° C. An efficient and accurate track and vertex reconstruction requires an SCT detector efficiency of at least 95%, as well as a resolution of 16|xm in Rep and 500 pan in z direction. The four barrel layers (Figure 3) have been recently assembled together and put on the thermal enclosure, the preparation of the disks for the end-caps is proceeding according to the schedule.

.• i

••

*

• & • •

•,-,

~-

" • " . - .

J

."•''•

' . ' * . ' i

x

*

••*'

-

4

•*••**

- * g

• "£•

Figure 3. A photography of the four layers of SCT tracker assembled on the thermal enclosure.

4.1.3. The Transition radiation tracker The Transition Detector Radiation (TRT) consists of 36 layers of 4 mm diameter straw tubes with a resolution of ~200um, interspaced with a radiator to produce transition radiation from electrons. There are two thresholds for recording hits,

561

the high threshold being used to detect TR photons. The TRT provides separation between hadrons and electrons. The barrel TRT is organized in three rings, each containing 32 modules made of axially aligned straws, and 2 endcaps made each of 2 set of independent eight plane wheels with radially oriented straws. The barrel assembly has been completed by September 2005 and fully tested. Figure 4 shows a cosmic muon traversing the barrel TRT. The end-cap rings will be both completed and transported to CERN by December 2005. In the next months the silicon tracker will be inserted into the barrel TRT and they will be tested together. The SCT end-caps will be inserted into the TDR end-caps during the spring 2006.

SideC v ,~

• ' - • - , gaxt'aVeto£>f

Figure 4. An event display of a cosmic ray muon traversing the Barrel TDR. The trigger is provided by 2 scintillators bars mounted on top and on the bottom of the detector.

562

5. The electromagnetic and hadronic calorimeter The calorimeter provides precision measurements for energy and direction for electrons, photons, hadronic jets and missing transverse momentum. In addition it provides a fast trigger up to the highest luminosities. It consists of an inner barrel cylinder and two end-caps cryostats, using the radiation resistant Liquid Argon (LAr) technology for the electromagnetic calorimeter, which are surrounded over their full length by a hadronic scintillator tile calorimeter. The barrel electromagnetic LAr calorimeter consists of accordion shaped lead absorbers, separated by 2 mm (LAr gap) interleaved with read-out electrodes. At n=0 the thickness is 26 XO. The barrel e.m. calorimeter has a highly granular first sampling (integrated pre-shower detector) which is preceded by a separated coarser pre-sampler to preserve the energy and direction after the cryostat and the coil material. The measured performance of the barrel and end-caps e.m. calorimeters at various n positions from 0 up to 2.7 can be described with a 10% /VE sampling term and a constant term of 0.3%. Wt^mr

^iffw

Figure 5. A cosmic ray muon traversing the tile calorimeter.

563

The hadronic scintillator tile calorimeter is based on a sampling technique with plastic scintillator plates (tiles) embedded in the iron absorber and read-out by wavelength shifting fibers. The active calorimeter depth at n= 0 is about 9.5 Xai,s, followed by about 1.4 X^ of passive iron in the support structure and integrated with the solenoid flux return yoke. The hadronic calorimeter uses scintillator tiles in the barrel and liquid argon in the end-caps (n > 1.5); its resolution can be described with a 50% /VE sampling term and a constant term of 3%. Forward calorimeters cover the regions 3 < n < 5 with a resolution better than 100% /VE± 10%.

6. The muon chambers The muon system is based on the magnetic deflection of the muon tracks in the large superconducting air core toroid magnets, instrumented with trigger and high precision tracking chambers. The momentum resolution is ApT/pT = 2xl0~2 at 100 GeV and ApT/pr =10"x at 1000 GeV. In the barrel region tracks are measured in three cylindrical layers (stations); in the transition and end-cap regions the stations are installed vertically. Over most of the pseudorapidity range (|r|| < 2) a precision measurement of the muon tracks in the principal bending direction of the magnetic field is provided by Monitored Drift Tubes (MDTs). At larger pseudorapidity range (2.0 < |n| < 2.7) Cathode Strip Chambers (CSCs) are used to stand higher rate and background conditions. The MDTs provides a single wire resolution of 80 um when operated at high gas pressure (3 bars). The trigger chambers covers the pseudorapidity range between 2.0 < |r|| < 2.7. Resistive Plate Chambers (RPCs) are used in the barrel region at |n| < 1.05 and Thin Gap Chambers (TGCs) in the forward region (1.05< |T)| < 2.4). The trigger chambers serves several purposes: a trigger with a well defined muon p? discrimination, bunch cross identification and the second coordinate measurement in the non-bending plane with a precision of a less of 1 cm. Most of the muon chambers have been produced and they are going to be completed at CERN with electronics and service cables. For the barrel the assembly of the stations MDT-RPC (in the middle and outer layers) is progressing in the BB5 hall at CERN and for the end-caps the first TGC-MDT sector (EIL4) has been assembled. Moreover 56 barrel muon stations have already been installed inside the coils of the spectrometer. By the end of the year the installation of all the stations

564

(BMS) in the middle layer of the small sectors of the spectrometer will be completed and of a part of the outer stations (BOS). The completion of the muon barrel stations installation is foreseen by the first half of 2006. 7. Conclusions A relevant part of the ATLAS detector construction is finished. The installation of the detector and all related services underground is advancing at a good speed. Recently major milestones passed are the successfully completion of the installation of the eight toroidal coils of the barrel spectrometer and the insertion of the barrel calorimeter at z=0. Commissioning activities have started with cosmic rays at full detector level in surface and soon will start the data taking with cosmic rays in situ for the calorimeter and a part of the barrel muon chambers (sector 13). The ATLAS experiment is on track to start the data taking in 2007 for the start of the operation of the Large Hadron Collider.

References 1. 2.

ATLAS Collaboration, ATLAS Technical Proposal, CERN/LHCC/94-43, 1994 The LHC project at CERN, LHC-project-report-36 (1996).

PROGRESS IN GRID COMPUTING FOR PARTICLE PHYSICS ROBERT GARDNER Enrico Fermi and Computation Institutes, University of Chicago Chicago, Illinois 60637, USA We discuss progress in building and exploiting distributed Grid computing infrastructures with an emphasis on capabilities required for the next generation of particle physics experiments at the LHC. Associated with a persistent and reliable Grid infrastructure linking compute, storage and network resources together, experiments, organized as Virtual Organizations (VOs), are building information architectures that allow discovery, access, and manipulation of datasets and algorithms. We use the ATLAS experiment as a context for describing the scale and complexity of the processing tasks associated with large scale data analysis relevant to the LHC.

1. Introduction 1.1. Overview In this article we take a brief look at recent progress in what has now become a large, worldwide effort in building distributed computing infrastructure capable of meeting the demands of the experiments built as part of the Large Hadron Collider Project fl] at CERN [2]. The LHC experiments, which are leading drivers of distributed computing technology, are acquiring lessons using this infrastructure that will provide for a large-scale, distributed, secure and reliable computing platform for 0(1000) physicists that will require the ability to discover and use many types of computing services for their processing tasks. The efforts are also benefiting existing high energy and nuclear physics experiments (some of which have been busy converting their legacy production systems to open Grid software, protocols and standards) as well as applications from other science domains. After reviewing the essentials of the computing requirements which set the scale for the challenge at hand, we consider various approaches to joining together distributed resources, review a few of the results and general lessons learned to date, and examine approaches for moving forward. 565

566

1.2. The Pressing Needfor Grid Infrastructure in Particle Physics The LHC opens a new regime in particle physics made accessible for the first time by a combination of high luminosity and high proton-proton center-of-mass energy. The ATLAS collaboration and detector is one of four experiments [3-5] at the LHC that will make discoveries designed to elucidate the nature of matter and the interactions of the fundamental particles. The initial focus of the large, general-purpose detector will be to probe the mechanism of electroweak symmetry breaking by discovering new particles (such as the "Higgs" particle) and measuring their properties. Further studies will look for the presence of supersymmetric partners to the Standard Model groupings of quarks, leptons and gauge bosons. The datasets created during ATLAS runs will be larger than any previous particle physics effort, with multiple-Petabytes of data produced each year. These datasets will hold event collections corresponding to and event rate of 25 MHz (each raw trigger composed of about 2 megabytes of raw electronic data which is subsequently processed into tracks, jets, showers, leptons, vertices and the like). During the lifetime of the experiment the scale will grow from 10's to 100's of Petabytes (1015 bytes) per year requiring Petaflops of processing capacity and high bandwidth networks to push data to those resources. To access and process this enormous data archive, a distributed computing model has been adopted [6], which requires at least two-thirds of the compute resources be located outside the CERN laboratory. The model is intrinsically distributed, which is (largely) a departure from generations of previous experiments. The reasons for this boil down essentially to the availability of large scale computing resources outside of CERN and the rapidly advancing capacity of the networks joining them. The picture is of a hierarchy of computing centers, or "Tiers", which present at various levels data processing capabilities appropriate to the center and role of that center. 1.3. The Pressing Need for New Information Architectures Along with the very large scale of the data and compute resources comes the need to create an information system capable of providing effective management of the data and processing algorithms and tasks involved. A typical high level process flow, from theory to data analysis and back again, is depicted in Figure 1.

567

Figure 1 High level process flow in high energy particle physics, view from the perspective of a process management framework.

Viewed from a process management perspective, there is relatively little complicated in the traditional "chain" of processing operations that take raw data from the detector to the physicist's desktop where subsequent, fine grained analysis happens (calibration, examination of cuts, comparison with theoretical models, Monte Carlo simulations, fitting). However, the problem complexity grows enormously when one maps the set of processes onto a physically distributed set of resources at the scales required here. Decomposing what is involved here, we look at some typical capabilities expected by the LHC physicist and the resulting patterns that emerge which suggest and provide guidance to building an information architecture to go along with the core computing infrastructure. An individual physicist-user or working group will wish to apply algorithms to a particular selection of input data in a given execution environment. The input datasets are discovered from a global archive typically via queries to metadata catalog services, which collect and store assigned name/value pair attributes for a number of experimental conditions and event trigger requirements. These queries will involve dialogs to determine the appropriateness of the returned datasets. For example the physicist may wish to place restrictions on the dataset type (raw detector data, reconstructed event data, analysis object data, or result objects from previous analyses) or event-level attributes (describing attributes of the collision such as the presence of large transverse momentum electrons). Candidate datasets may also be inspected for their provenance information - what was the specific set of abstract and concrete transformations that generated the dataset? Which specific software releases and calibration constants were used during the reconstruction? The physicist may wish to recursively inspect the dataset at many levels - to examine individual provenance of event fragments corresponding to different

568

algorithms employed during the processing. The dataset may be described virtually, with pieces described but not yet materialized or once materialized and now deleted. A full provenance chain describing intermediate processing steps and dataset dependencies may need to be examined for appropriateness to a given analysis. Decisions on whether to proceed may be governed by the availability of the dataset, or whether significant re-processing may be required to deliver it to the algorithms. Dataset properties relating to splitting may be examined to determine if a distributed computation is possible. Transformation metadata may be used to provide estimates of resource requirements, information which the physicist may also weigh in deciding whether to proceed with a given computation. The dataset and application configurations are submitted to a workload management service, which itself will depend on a number of other services providing, for example, dataset catalog services, a basic security infrastructure, information services containing attributes of distributed compute and storage elements, authorization and accounting services, job monitoring, data provenance services, etc. Work requests can be minimal, low-time latency executions, which satisfy requirements for interactive analysis, or long-lived processes requiring large scale resources and which may take weeks or longer to satisfy. The physicist-user monitors overall task progress and job processes which in turn may interact with dataset and replica catalog services during processing. The results are gathered together and passed back to the owner of the request who may then perform additional processing steps. A framework for describing the state of process threads, faults, and recovery procedures is needed. These are being developed within the individual VO communities by teams of software developers. 2. Grid Architectures, Principles and Deployments There are now many successful Grid projects worldwide that encompass a variety of architectures, deployment approaches, and targeted application domains. For example, European efforts include the LHC Computing Grid Project (LCG) [7] project, the European Data Grid (EDG) and its follow-on EGEE (Enabling Grids for E-Science) [8], and DataTAG [9], which focused on transatlantic grid testbeds and high performance networks. NorduGrid [10] links computational centers in Scandinavia to deliver production services for highenergy physics applications. The U.S.-based Grid3 project [11], which began operations in Fall 2003 and completed by Spring 2005, extended these and other

569 efforts by relying on immediate application requirements to drive architectural principles and decisions. 2.1. Physical and Service Architectures, and Principles Projects develop a distributed architecture for linking resources taking into account several factors, including recognized emerging standards, such as being developed in the Global Grid Forum, lessons from previous projects, constraints on site policies of the resource providers, and naturally, the realistic use cases from VOs. We briefly touch on the physical, service, and principles guiding decisions in implementation and technology choices. 2.1.1. Typical Site Architecture Some abstractions for compute and storage resources have entered into common use as they describe a canonical computing center, as show in Figure 2. A Compute Element (or CE) describes a typical Linux farm in which execution nodes (which can be high-end servers or white box commodity) are connected via Gigabit Ethernet switches, sometimes completely internally with no visibility or connection to the public internet, and a small number of "gatekeeper" nodes which provide a hosting environment for the Grid services which communicate to external, higher level Grid services. Optionally within a site's administrative domain there can be a Storage Element (or SE) which provides a disk-based (tertiary, backend tape-based systems are provided by larger facilities) storage for caching of datasets. These usually consist of a set of network attached servers or commodity disks configured as RAID volumes. Both CE and SE elements are made known to the external environment through an information service, which may a local database or a publishing service to a higher level server which aggregates information from several such servers, providing an information index service.

570

CE: Compute Element

£E: Storage Element

Figure 2 A common abstraction for computing and storage resources as viewed from the point of view of various, higher level services which may be Grid specific (such as a discovery service) or VO-specific, such as user authentication and authorization database.

The CE is defined in terms of its platform (operating system and processor type), local batch queuing system (such as PBS or the Condor scheduler), the environment a job expects to find once it begins execution, including pointers to various pre-installed utilities, common software installation areas, and characteristics of the local storage on each execution node. The Unix account structure is setup by the facility administrator to map to various supported VOs according to roles, policies, and agreements between the resource provider and the accessing VO. There are a number of systems that have been developed for this purpose, including the Virtual Organization Management System (VOMS) [12], the Grid User Management System (GUMS) [13], only to name a couple. 2.1.2. Typical Software and Service Architectures Very quickly, one arrives at a layered view of the software and service infrastructure that must be in place in order to provide users access to the underlying resources. This is show schematically in Figure 3.

571 Applications, Portals and VO Services Wor&oed and Laeof KepftcQ

[

Monitoring and Accounting

Satekeepers ami Web Service Interfaces

Farm CE Site A

Farm CE Site 8

External Storage SE

External Storage SE

Network Router

Edge Server

Network Switch

Edge Server

Figure 3 Simple middleware architecture diagram showing services that span sites (hosting CE and SE elements) in a Grid.

Usually what is called for is a simple architecture that links execution and storage sites and provides services for virtual organization management, workload and replica management, monitoring and accounting services, all provisioned through middleware. There are leading providers of middleware that the LHC experiments are using. The EGEE project, which is delivering technology to the LCG project, has a large scale software engineering effort to produce a set of packages, collectively known as "gLite". The main components of gLite span a set of service groups including security services (providing a service to establish the authenticity of users, as well as auditing and authorization), information and monitoring services that provide remote services used by either the VO and Grid operations centers, data services which host replica and metadata catalogs, and job management services which provide package installation, job provenance, policy enforcement and workload management. The EGEE gLite software utilizes at its core the Virtual Data Toolkit (VDT) [14]. The VDT is also used in the U.S.-based Open Science Grid project. The VDT provides services from the Globus Toolkit [15], Condor [16], GriPhyN, and PPDG, as well as components from other providers. VDT allows grid facility administrators to configure their sites easily with simple and welldefined interfaces to existing facility configurations, information service providers, and storage elements. Additional services such as Replica Location Service (RLS) [17], Storage Resource Manager (SRM) [18], and dCache [19] are deployed at sites according to requirements from individual VOs.

572

2.1.3. Policy and Interoperability Policy issues affecting use of the resources come into play through sharing agreements between VOs and priorities and optimizations made within a VO. Sites comprised of CE's and SE's are typically associated with a VO, and some sites may be providing resources to the wider Grid environment on behalf of several VOs (such as a large national computing facility or laboratory). Thus a policy and accounting infrastructure is required to establish fair share policies and bartering among VOs. The reality today is that there are multiple, core Grid infrastructures. This has come about from a number of factors, including a natural process of trial and error by one project, and the need to take independent decisions and approaches by another. But by and large the underlying protocols are the same or identical, and it helps that between OSG and LCG/EGEE a common core layer VDT is shared between the two. However, several differences exist in specifics of how a site's architecture is organized, which services are expected to be running, and what their interfaces are. Thus, efforts at Grid interoperability are focused on providing compatibility between the infrastructures, abstracting these differences away from the end-user physicist. There are several approaches here, ranging from publishing the essential details of the resource (as expressed in an agreed to, common schema) to a resource broker that the VO's workload management system (or end-user client tools) sends jobs. 2.2. Existing Deployments and Projects 2.2.1. The LCG/EGEE Grid The LCG project is organized by CERN to provide a computing environment for the four LHC experiments. The project receives deliverables for middleware technology, infrastructure, and operational support from the EU-funded EGEE project which is providing a multi-science core foundation for Europe. The aforementioned re-engineered middleware set of packages, gLite, is provided by EGEE. The deployed infrastructure consists, as of July 2005, of 140 sites dispersed in 34 countries, aggregating more than 15,000 processors. The experiences gained while supporting data challenges by LHCb and ATLAS include ramping to a capacity of >3500 concurrent jobs, with ATLAS averaging 8,000 jobs per day at peak of operations. Over 1 million SI2K-years of CPU time were consumed, resulting in 400 TB of data produced, transported, and archived in its storage facilities.

573

Of major importance were lessons provided by the exercises which included dealing with site-level stability (and the varying degree of systems expertise between the contributing sites), and a job efficiency of > 90% under controlled conditions. 2.2.2. The Open Science Grid The Open Science Grid is a persistent, shared, multi-virtual organization (VO) Grid infrastructure capable of providing production level services for large-scale computational and data-intensive science applications. The OSG Consortium receives contributions from several projects, including the GriPhyN virtual data research project [20], the Particle Physics Data Grid, PPDG [21], and the International Virtual Data Grid Laboratory, iVDGL [22], as well as the U.S. ATLAS and U.S. CMS Software and Computing Projects of the LHC. The OSG extends the capabilities of the previous project Grid3 but still follows a simple two-tier approach, in which each resource (compute, storage, application, site, user) is logically associated with a VO. At each site, a core set of grid middleware services with VO-specific configuration and additions are installed, with registration to a VO-level set of services such as index servers and grid certificate databases. Where appropriate, VO-level services were combined into top-layer services at the OSG Grid Operations Center, which provided monitoring applications, display clients, and verification tasks and an aggregate view of the collective OSG resource and performance. Approximately 30 sites comprised of national laboratories and university-based centers contributed approximately 14,000 processors to the environment, providing resources to more than 35 VOs. 3. Performance and Lessons from ATLAS We briefly summarize here some results from the past two years of data challenge exercises taken by the ATLAS experiment. ATLAS used LCG, NorduGrid, and Grid3, and designed a production system capable of reaching all three. A major exercise was launched in June 2004 and consisted of phases covering large scale production of physics datasets (order 10M events) with a chain of production steps including Pythia event generation, GEANT4-based simulation, digitization and pileup. Almost 45 million CPU days (in SI2K-CPU) were used to execute over 100K jobs that contained nearly 8 million events and about 30 TB of data. The three Grids contributed with almost an equal amount of work each. The lessons drawn consisted of the following:

574

1. Protect grid services that are vulnerable in a multi-VO environment. 2. The most likely single point of failure is data staging. 3. An infrastructure to establish policy is needed to distinguish user roles within a VO. 4. Delay binding of jobs to resources until the last possible moment in order to optimize utilization. 5. Robustness and reliability are required to keep operating costs low. 4. Approaches Moving Forward There are two significant courses being undertaken by Grid infrastructure providers and VO application framework developers. The first major key is to develop an infrastructure that will provision and protect servers on the "edge" of the compute resource, essentially providing a container for a VO manager to drop in a service. This approach facilitates the complex workflows and methods that may be specific to the application and not general enough to be a common middleware service. An example is a distributed data management system, in which a VO has specific requirements on the functionality of reliable file transfers, local space management, and catalog structure. A second major thrust is to deploy a fully services oriented architecture using, for example, the web services resource framework (WSRF) that intends to hide much of the complexity and details of using the resource behind a standard interface, simplifying access and discovery of datasets and other services made available by resource providers. Acknowledgments This work was supported in part by a grant of the National Science Foundation, NSF ITR-0122557. The author gratefully acknowledges useful discussions with J. Shank, R. Jones, D. Barberis, L. Robertson, R. Pordes, I. Legrand, M. Lamana, E. Laure and M. Mambelli. References 1. 2. 3.

The Large Hadron Collider Project at CERN; Available from: http;//lhcnew-homepage.web.cern.ch/lhc-new-homepage/. CERN, the European Laboratory for Particle Physics; Available from: http://cern.ch/public/. The ALICE Experiment at CERN; Available from: http://alice.web.cem.ch/Alice/AliceNew/.

575

4. 5. 6.

7. 8. 9. 10. 11. 12. 13. 14. 15.

16.

17.

18.

19. 20. 21. 22.

CMS: Compact Muon Solenoid at CERN; Available from: http://cmsinfo.eern.ch/Welcome.html/. The LHCb Experiment at CERN; Available from: http://lhcb.web.cern.ch/lhcb/. Report of the Steering Group of the LHC Computing Review (2001); Available from: http://lhc-computing-review-publie.web.cern.ch/lhccomputing-review-public/Public/Report final.PDF. The LHC Computing Grid Project (LCG); Available from: http ://lcg. web. cem.ch/LCG/. EGEE: Enabling Grids for E-Science in Europe; Available from: http://public.eu-egee.org/. Research and Technological Development for a Data TransAtlantic Grid; Available from: http://datatag.web.cem.ch/datatag/proiect.html. NorduGrid: Nordic Testbedfor Wide Area Computing and Data Handling; Available from: http://www.nordugrid.org/. The Grid2003 Project, The Grid2003 Production Grid: Principles and Practice, in High Performance Distributed Computing (2004). EU DataGrid Java Security Working Group, VOMS Architecture vl.l (2003); Available from: http://grid-auth.infn.it/docs/VOMS-vl 1 .pdf. G. Carcassi, Grid User Management System; Available from: http://prid.racf.bnJ.gov/GUMS/index.html. The Virtual Data Toolkit (VDT); Available from: http://www.lscgroup.phvs.uwm.edu/vdt/. I. Foster, Kesselman, C, Globus: A Metacomputing Infrastructure Toolkit, International Journal of Supercomputer Applications 11 (2), 115-129 (1998). M.J. Litzkow, M. Livny and M.W. Mutka, Condor - A Hunter of Idle Workstations, 8th International Conference on Distributed Computing Systems 104-111(1988). A. Chervenak, E. Deelman, et al., Giggle: A Framework for Constructing Scalable Replica Location Services, in SC'02: High Performance Networking and Computing (2002). B. Sim, A. Shoshani, J. Gu, ed. Storage Resource Managers: Essential Components for the Grid. Resource Management for Grid Computing (2003). The dCache Project; Available from: http://www.dcache.org/. P. Avery, I. Foster, Towards Petascale Virtual Data Grids (GriPhyN Project); Available from: http://www.griphvn.org/. Particle Physics Data Grid; Available from: http://www.ppd g.org/. P. Avery , I. Foster, R. Gardner, H. Newman, A. Szalay, An International Virtual-Data Grid Laboratory for Data Intensive Science, Technical Report: GriPhyN-2001-2 (2001. http://www.griphvn.orgA.

D E S I G N , P R O D U C T I O N A N D QUALITY OF T H E SENSORS FOR T H E SILICON STRIP T R A C K E R OF CMS

M. K R A M M E R Institute

for High Energy

Physics of the Austrian Academy of Nikolsdorfergasse 18, Vienna, Austria E-mail: Manfred. KrammerQhephy. oeaw. ac. at On behalf of the CMS Silicon Tracker Collaboration

Sciences,

The construction of the Silicon Strip Tracker of CMS requires about 25000 large area silicon sensors. By the time of the conference the delivery of these sensors was almost completed. The paper reviews the design of the sensors including the considerations to achieve the required radiation hardness. A comprehensive quality assurance program was developed by CMS to monitor the quality of the sensors produced by two companies. After explaining the basic elements of the quality assurance program an overview of the results is shown. During the production several problems were identified and eventually solved in cooperation with the manufacturing companies.

1. The CMS Experiment At CERN, the European Laboratory for Particle Physics in Switzerland, the large proton-proton collider LHC is under construction. The LHC will produce collisions at a centre of mass energy of 14 TeV. The startup of the LHC collider is foreseen for summer 2007. To exploit this machine the CMS experiment, a multipurpose experiment, was designed and is under assembly at CERN. The acronym CMS stands for 'Compact Muon Solenoid' and indicates one of the key features of this experiment, the precise measurement of muons. The CMS detector will consist of several shells of different detector elements. Particles created in the high energy collisions in the very center of the detector will first traverse the 'Inner Tracker' a system of silicon sensors designed to detect charged particles. Outside of the Inner Tracker is the electromagnetic calorimeter, a detector consisting of about 80000 scintillating crystals, whose purpose is to absorb electrons and photons and to measure their energy. Further outside is a shell of hadronic calorimeters

576

577 measuring the energy of hadrons, e.g. pions, protons, etc. Just outside the hadronic calorimeter is a superconductive solenoid magnet which will create a magnetic field of 4 Tesla. All charged particles traversing this field are deflected with a bending radius which depends upon their charge and momentum. The outside shell of CMS is formed by the muon chamber system, a gas detector system designed to measure the tracks of through going muons. Combined with the muon tracks measured in the Inner Tracker the momentum of the muons can be determined with high precision. The complete CMS detector will be 21.6 m long, with a diameter of 15 m and with an overall weight of 12500 tons.

2. The Layout of the CMS Inner Tracker The CMS Inner Tracker : is divided into substructures: The pixel detector very close to the interaction point, which is not further explained in this paper, and the strip tracker consisting of the inner barrel detector (TIB), the inner discs (TID), the outer barrel (TOB) and the two end cap detector systems (TEC). The overall length of the Inner Tracker will be 5.4 m with a diameter of 2.4 m. The total weight will be about 3 tons. The temperature inside the tracker will be adjusted such that the maximum temperature of the silicon sensors will not exceed —10°C. The total electrical power dissipation is expected to be about 45 kW. Figure 1 shows a cut through a quarter of the Inner Tracker. The solid lines in this sketch represent the silicon sensor modules. The modules of the TIB are mounted on four and the TOB modules on six concentric shells. The supporting structures for the TID and TEC are discs, two times three discs for the TID and two times nine discs for the TEC. The basic construction element of the Silicon Strip Tracker is a module (see figure 2). The supporting frame of a module is made of carbon fiber or graphite. Glued onto the frame is a Kapton layer to isolate the frame from the silicon and to provide the electrical connection to the silicon backplane. A ceramic multilayer hybrid holds the readout chips and the auxiliary chips. A glass pitch adapter is mounted between the hybrid and the first silicon sensor to match the different pitches of the chips input pads and the sensor strips. Wire bond connections between the individual channels of the readout chips and the pitch adapter, between the pitch adapter and the first sensor and between the two sensors make the electrical connections. The modules of the TIB, the TID and the four inner rings of TEC will consist of only one silicon sensor, whereas the modules of the TOB and

578

0

200

400

600

800

1000

1200

1400

1600

1800

2000

2200

2400

2600

2800 2[mm3

Figure 1.

The CMS Inner Tracker Layout.

the three outer rings of TEC will hold two sensors. All barrel modules are rectangular. The disc modules have a wedge shape in order to form rings. Figure 2 shows a final module of the second ring of the TEC. The first two layers in TIB and TOB, the first two rings in TID and rings 1, 2, 5 in TEC are instrumented with double-sided modules. These are made of two independent single-sided modules, mounted back to back and rotated by 100 mrad with respect to each other. Table 1 lists some numbers illustrating the overall dimension of this project. Table 1. Some key numbers from the Silicon Strip Tracker construction Area of active silicon Number of silicon sensors Different sensor designs Number of modules Number of strips Number of electronics channels Number of readout chips Number of wire bonds

« 200 m 2 24,244 15 15,232 RJ 9,600,000 ss 9,600,000 « 75,000 ss 25,000,000

3. The Design of the Silicon Sensors To build such a large silicon system, the influence of the design of the silicon sensors on the mass production was studied first. In close collaboration with industry the basic parameters of the sensors were defined to reach maximum

579

Figure 2. A ring 2 module for the TEC mounted on a transport plate

efficiency for the mass production. The result was a rather simple design compared to previous projects in high energy physics 2 . In this respect it was decided to use only single sided sensors. A second criterion driving the design was to optimize the use of the wafer area as the silicon sensor prize is determined by the number of wafers rather than the actual sensor surface. The 15 different sensor types were designed to make maximum use of the area available on 6 inch wafers as most of the manufacturers use 6 inch fabrication lines for the sensor production. Concerning the silicon material standard high resitivity n-type silicon wafers are used. The potential advantage of oxygenated material with respect to radiation hardness was neglected in favor of the safety of using well understood material. The proton-proton collisions inside the LHC detectors will create an enormous radiation background which will damage every detector material. The innermost layers of the silicon strip detector will receive an estimated fluence of 1.6xl0 14 neutrons c m - 2 (1 MeV neutron equivalent) during 10 years of LHC operation. The fluence is gradually lower for sensors mounted further away from the interaction point. Silicon is inherently radiation hard compared to other detector technologies. However the silicon sensors will be nevertheless modified and damaged. The two most important macroscopic

580

effects of the irradiation are the increase of the leakage current which is linear with fluence, and the conversion of the n-type bulk into p-type with ever increasing carrier concentration. A consequence of the second effect is the need for increasing bias voltage after the type inversion which eventually leads to detector breakdown. For the longtime operation of the sensors in the Inner Tracker, it is therefore essential that the sensors can sustain very high bias voltages up to 400 - 500 V. An appropriate choice of the resistivity of the original silicon wafers is important to influence the evolution of the depletion voltage. CMS has decided to use different material for the inner part of the tracker compared to the outer part: the inner four barrel layers (TIB), the inner discs (TID) and the four innermost rings of the forward discs (TEC) are fabricated using silicon of relatively low resistivity (1.25 - 3.25 kfi cm) with a thickness of 320 ^m. The outer six barrel layers (TOB) and the three outermost rings of the end-cap discs (TEC) will be made of silicon with high resistivity (3.5 - 7.5 kfl cm), using wafers of 500 fim thickness. The low resitivity sensors for the inner part of the detector will require a high bias voltage at startup (up to 330 V), however the type inversion will occur at a higher fluence and hence later in time. At the end of the foreseen lifetime of 10 years, the sensors which are exposed to the highest fluence will.require a bias voltage which is not much higher than the one at startup. For the sensors in the outer part the choice of high resistivity material allows to use thicker wafers with still manageable high depletion voltage. Another choice related to the radiation hardness was to use silicon wafers with a lattice orientation of (100). CMS studies have shown that this material has advantages compared to material with a lattice orientation of (111). The interstrip capacitance remained nearly unchanged after irradiation for sensors produced on (100) material, but is increased significantly if (111) material was used 3 . In order to cover the barrel layers and the end-cap discs with silicon sensors 15 different designs are needed: four different rectangular designs for the barrel and 11 different wedge shaped designs for the inner discs and the endcaps. Table 2 and 3 show an overview of the different sensor types, their dimensions, the strip pitches and the number of sensors needed for the particular type. The principal design parameters are identical for all designs. Figure 3 shows a corner of a wedge shaped sensor. The sensors have p + strips on n-type bulk material. The strip pitch varies from 80 /um up to 205 fxm. The ratio strip width to strip pitch is 0.25 for all types. The strips are

581 AC coupled and the coupling isolation is formed by two thin layers of Si02 and Si3N4. The thickness of the aluminum readout lines is required to be at minimum 1.2 /im. An array of polysilicon resistors (1.5 ± 0.5 MQ) is implemented to connect the individual strips to a common p + bias ring. To enable the sensors to be operated at a bias voltage of up to 500 V the aluminum readout strips are wider than the p + strip implants (4 /um to 8 fim metal overhang). For each strip two aluminum pads are foreseen for bonding. The DC pad is in direct contact with the p + strip implant for testing. The active sensor area is surrounded by the p + bias ring and one floating p + guard ring. Metal field plates on top of the guard ring extend also beyond the guard ring p + implant. An n + implant is realized along the sensor edges. Outside of the sensor area a set of standardized test structures, identical for all 15 wafer types, is implemented which is used for the CMS quality control. Two companies have produced silicon sensors for the CMS tracker: Hamamatsu Photonics K.K. (Hamamatsu-City, Japan) is producing the large part of sensors (> 25000 sensors, including spares). The second company ST Microelectronics (Catania, Italy) has produced about 2700 sensors for the outer part (TOB and rings 5 to 7 of TEC).

t *

:-• -

*

* : \ /

1

'i'i- •' ' * *

••&

.-„<

•r"

--.

'

~' •*

'

| gj h * ,' t • >• IIIWPI ww 'Wl' SaS «*•* % a-*... iy

,

^yw?

,

<

*

I

s^

Figure 3. Picture of the corner region of a CMS sensor

582 Table 2. Inner Barrel 320 fim thick sensors and Outer Barrel 500 /wn thick sensors, geometrical dimensions and multiplicities Type IB1 IB2 OBI OB2

Length [mm] 63.3 63.3 96.4 96.4

Height [mm] 119.0 119.0 94.4 94.4

Pitch [/jm] 80 120 122 183

Strips 768 512 768 512

multipl. 1536 1188 3360 7056

Table 3. Geometrical dimensions and multiplicities for 320 jim thick (W1-W4) and 500 ftm thick (W5a-W7b) wedge sensors for TID and TEC: W l has two different versions for TID and TEC, whereas the TID shares identical W2 and W3 sensors with the TEC. Type W l TEC W l TID W2 W3 W4 W5a W5b W6a W6b W7a W7b

Length 1 [mm] 64.6 63.6 112.2 64.9 59.7 98.9 112.5 86.1 97.5 74.0 82.9

Length 2 [mm] 87.9 93.8 112.2 83.0 73.2 112.3 122.8 97.4 107.5 82.9 90.8

Height [mm] 87.2 112.9 90.2 112.7 117.2 84.0 66.0 99.0 87.8 109.8 90.8

Pitch [/im] 81-112 80.5-119 113-143 123-158 113-139 126-142 143-156 163-185 185-205 140-156 156-172

Strips 768 768 768 512 512 768 768 512 512 512 512

multipl. 288 288 864 880 1008 1440 1440 1008 1008 1440 1440

4. Sensor Quality Assurance To ensure the quality of more than 25000 sensors the CMS sensor group has developed a comprehensive procedure to qualify the sensors 4 . This quality assurance program contains four parts: • Tests performed at the companies prior to shipment. The contract lists a set of measurements and acceptance criteria. These measurements are for example tests of the leakage current, the determination of the depletion voltage and a check for pinholes and broken or shorted aluminum lines. Only sensors passing these tests are delivered to CMS. • Quality Control on sensors. This is an exhaustive test of the sensors. The total sensor leakage current and the total sensor capacitance is scanned as a function of the bias voltage and, at a bias voltage of 400 V, several parameters for each strip are measured in a strip scan.

583 • Process Control on test structures. Using the specially designed set of test structures more than 10 parameters are measured. These parameters allow to monitor the production process and provide data which cannot be measured on the sensors themselves. • Irradiation Tests. Using a neutron and a proton irradiation facility test structures and sensors are checked for the required radiation hardness. The goal of the CMS quality assurance procedure is to deliver 99% good sensors to the module assembly centers where the sensors, the readout hybrids etc. are glued to mechanical frames. A failure of a sensor at this stage will cause significant additional costs due to the lost material and working time. Furthermore the tests have to make sure that the sensors will be sufficiently radiation hard for the expected 10 years of operation in CMS. In the following the Quality Control, the Process Control and the Irradiation Tests are explained in more details. 4.1. CMS Quality

Control

The Quality Control (QTC) is done in five CMS centers: Karlsruhe (D), Perugia (I), Pisa (I), Rochester (USA) and Vienna (A). It consists of an optical inspection and an electrical characterization. During the optical inspection the sensor is checked by eye for obvious damages and the edges of the sensor are scanned using a microscope to check for defects introduced during cutting. The electrical characterization is done using fully automatic systems which measure the total leakage current and the detector capacitance as a function of the bias voltage up to 550 V and subsequently measure on each strip the values of the single strip leakage current, the polysilicon resistor, the coupling capacitance and the leakage current of the dielectric. This strip scan is done with the sensor kept at a bias voltage of 400 V. This measurement takes several hours and is therefore only done on a sample of sensors. 4.2. CMS Process

Control 5

The Process Control (PQC) is done in three CMS centers: Florence (I), Strasbourg (F) and Vienna (A). A set of standard test structures, identical for all 15 sensor designs and for both producers, is used to perform ten measurements in an automated way: C(V) on a MOS structure, C(V) on

584

a diode, interstrip resistance, I(V) on a mini-sensor, interstrip capacitance, I(V) on a gate controlled diode, resistance of the polysilicon resistors, resistance of the aluminum lines and resistance of the p + implants, coupling capacitances, and the breakdown voltage of the decoupling capacitor. The results of these measurements are compared to the specifications and are used to detect possible changes in the production process of the companies. 4.3. Irradiation

Tests

The irradiation tests for a small number of sensors and test structures are done by two laboratories: a neutron irradiation facility in Lovain-la-Neuve (B) and a proton irradiation facility in Karlsruhe (D). The irradiations are performed up to the fluence expected in CMS for the particular sensor type at a temperature of —10°C . After irradiation and a short period at room temperature to allow for beneficial annealing the structures are remeasured at —10°C. The tests before and after irradiation consist of the measurement of the leakage current, the determination of the depletion voltage and the measurement of several single strip parameters: the values of the strip capacitance, the interstrip resistance, the polysilicon resistance and the coupling capacitance. 5. Status of Sensor Production and Quality Summary The status of the production and sensor delivery at the end of September 2005 is the following. In total 28600 sensors have been ordered from HPK and STM. This number includes spare sensors, but also the sensors used for the various stages of prototyping. From these ordered sensors 90% are already delivered. Until end of September 2005 in total 8673 sensors and 4144 test structures have been tested by the CMS groups. Remember that the test procedure foresees only sample tests after a 100% test at the companies. From the delivered and already tested sensors we have qualified 25555 sensors for the use in CMS. These sensors are subsequently shipped to the assembly centers and a good fraction is already mounted into modules. The final shipment of sensors is expected to arrive at CERN in November 2005. The results shown in this paper can therefore be considered as almost final. The first parameter measured during the QTC procedure, and probably the one which provides the most important information on the quality of the sensors, is the total leakage current. Figure 4 shows the total leakage currents measured at a bias voltage of 450 V. The figure shows the results

585

measured by the QTC centers on samples of HPK sensors (left plot) and STM sensors (right plot). Only accepted sensors within the specification of maximum 10 fiA are plotted. The average value measured is about 580 nA for HPK sensors and about 1870 nA for STM sensors.

Total Leakage Current at 450V for accepted HPK Sensors Entries: 1252 »- H Mean: 576.4

Total Leakage Current at 450V for accepted STM Sensors Entries: 2203 Mean: 1865-48

, @450V|nA|

Figure 4.

Leakage currents at 450 V for HPK and STM sensors

The second very important parameter for the CMS sensors is the depletion voltage. Figure 5 shows the depletion voltage measured by the QTC centers on the same sample of sensors (HPK sensors left plot, STM sensors right plot). The specifications allow a depletion voltage range of roughly 100 V to 300 V, slightly different for the inner layer and the outer layer sensors. Both companies use up essentially the full range.

Full Depletion Voltage for accepted HPK Sensors Entries: 1252 Mean: 174.7

FuH Depletion Voltage |V]

Figure 5.

Full Depletion Voltage for accepted STM Sensors Entries 2203 Mean: 172.90

Full Depletion Voftage [V|

Depletion voltage for HPK and STM sensors

Figure 6 shows per sensor the distribution of the number of strips outside

586

the specification. The plots in this figure show the combined results from both the companies data and the QTC data measured on samples. A strip is outside the specifications if one of the following defects is present: high single strip leakage current, polyresistor value outside the specification, aluminum short or open, coupling capacitance outside specification or short (pinhole). In the case of the HPK sensors on average one strip is bad in ten sensors while for the STM sensors each sensor has on average 1.9 bad strips. The specification allows a maximum of 1% bad strips on a sensor (the sensors have either 512 or 768 strips). The sensors accepted from both companies fulfill this requirement. In the case of HPK the percentage of bad strips on a sensor is 0.02% and for STM sensors this value is 0.3%.

Number of Bad StriDs for accepted STM Sensors Entries: 2681 Mean: 1.91

Number of Bad Strips for all accepted HPK Sensors Entries: 17778 Mean: 0.1

o W

s I z

Number of Bad Strips

Figure 6.

Number of bad strips per sensor

The measurements done for the so-called Process Control provided additional information on the quality of the sensors in many details. Most of the measurements explained in section 4.2. confirm a very stable production within the specifications and with time. Examples can be seen in figure 7 for the measured poly resistor values (STM sensors) and in figure 8 for the measured breakdown voltage of the coupling capacitance (HPK sensors). The specification of the breakdown voltage requires a minimum value of 100 V. This measurement is a destructive measurement and is therefore a good example for the usefulness of the test structures as such a measurement cannot be performed on the real sensors. Both figures show the parameter as a function of time. These parameters as most others are very stable. We have however also detected some non-conformities while checking

587

01/02

Figure 7.

07/02

01/03

07/03

01/04

07/04

12/04 07/05 12/05 Oat* [month/y>ar]

Measured poly resistor values as a function of the production period (STM).

IV_dl«l*ctrlc vs. Data of Measurement for HPK

"190

|

mmm

>'180

• Urn •!&»

170 160 150 140 130 120 01/02

Figure 8. (HPK).

07/02

01/03

07/03

01/04

07/04

12/04 07/05 12/05 Data [month/yaar]

Measured dielectric breakdown voltage as a function of the production period

the measurements of the Process Control. The following figures shows two examples. Figure 9 shows the calculated specific resistivity of the thin sensors from HPK deduced from the measurement of the depletion voltage of a test diode. The specification for this parameter follows the radiation resistance requirements as explained in section 3. For the sensors of the inner layers the specific resistivity should not exceed 3.25 kf2 cm. For a production period in 2003 HPK has used a wafer material with too high specific resistivity. As a consequence part of these sensors had to be rejected

588

while some sensors of these lots could be used in layers further outside the interaction region. |

Resistivity vs. Tims for thin HPK asnsora

|

o

r

•

*

) » • ; & , :

- • • . • • •

•

:

lft'i. K: „

. _,

OT/02

Figure 9.

i

07/02

i

,

i

01/03

i

i

07/03

i

01/04

i

07/0*

i,,

i

i

i

i

i

i

12/04 07/05 Data [month/ysar]

Measured specific resitivity as a function of the production period (HPK)

Figure 10 shows the flatband voltage measured on a MOS test device for STM wafers. The flatband voltage is an indicator for the oxide quality. In an irradiation experiment we have determined the tolerable maximum value for the flatband voltage of this test stucture to be 10 V. For some batches this value was exceeded by far. This defect is particularly dangerous as the non-irradiated sensors behave perfectly. After irradiation a large increase of the interstrip capacitance would have occurred causing the sensors to fail after a few years of LHC operation. Sensors from these batches were clearly rejected. 6. Conclusions The delivery of the sensors for the Silicon Strip Tracker of CMS is almost completed. The sensors and the associated test structures were tested and qualified by CMS. The quality assurance procedure developed by CMS is of an unprecedented scale and did involve eight institutes. In total 16 electrical parameters have been measured on a total of 13000 structures as of end of September 2005. This procedure has worked very efficiently and we have identified several problems which were eventually solved together with the companies. The quality of the sensors accepted for CMS is excellent and fulfills all the

589 Flatband Voltaga vs. Data of M«a«urem«nt for STM |

01/02

07/02

01/03

07/03

01/04

07/01

12/04 07/05 12/05 Data [month/ytar]

Figure 10. Measured flatband voltage as a function of the production period (STM) requirements for the long time operation of the Strip Tracker inside the CMS experiment. Only a very small p a r t of the results could be shown in this paper. As a final indicator for the quality of the sensors accepted for CMS the percentage of good strips can b e quoted to b e 99.94%. Acknowledgments T h e results presented in this paper were collected by a large number of competent colleagues from the CMS sensor working group over a period of several years. I would like t o acknowledge their effort and t h a n k t h e m for this work. References CMS Tracker TDR, CERN/LHCC 98-6 CMS TDR 5 (1998) and CMS TDR Addendum, LHCC 2000-016 (2000) Semiconductor Tracking Detectors in e+e- Vertex Detectors, M. Krammer, Nucl. Instr. and Methods A 379 (1996) 384. Investigation of design parameters for radiation hard silicon microstrip detectors, S. Braibant et al., Nucl. Instrum. and Meth. A 485 (2002) 343. The Silicon Sensors for the Compact Muon Solenoid Tracker - Design and Qualification Procedure, J.-L. Agram et al., Nucl. Instrum. and Meth. A 517 (2004) 77. Process control strategy of the silicon sensors production for the CMS tracker, T. Bergauer et al., Nucl. Instrum. and Meth. A 494 (2002) 218.

R U N N I N G MINOS

KAROL LANG* The University of Texas at Austin, Department of Physics, 1 University Station CI 600, Austin, TX 78712-0264, USA E-mail: [email protected]

MINOS is a long baseline experiment at Fermilab to study neutrino oscillations. The experiment started its operations in February 2005 and is now recording beam neutrino interactions in its two magnetized toroidal tracking calorimeters. The distance between the two detectors and the energy of the neutrino beam produced by protons from the Main Injector cover a range of the "atmospheric" A m 2 . The MINOS far detector has been active since 2003 and registers interactions of cosmic ray muons and neutrinos. We briefly describe experimental details, our experience running MINOS, give a glimpse of beam neutrino data, and present preliminary results with atmospheric neutrinos.

1. Introduction The long baseline MINOS experiment at Fermilab was proposed 10 years ago l'2 when the world of particle physics was puzzled by "the solar neutrino deficit" and "the atmospheric neutrino anomaly". The planning, the design, and the construction of MINOS turned out to be challenging largely due to the complexity of the civil construction of the neutrino beam line. The experiment, comprising two detectors and a beam line, was fully commissioned in early 2005 and is now running. During the construction period of MINOS several experiments improved our understanding of the physics of the atmospheric and solar neutrinos. Results from Super-Kamiokande, 3 ' 4 SNO, 5 MACRO, 6 and Soudan-2 7 have now firmly established the existence of neutrino oscillations. Recently, the first long baseline accelerator and reactor experiments, K2K 8 and KamLAND, 9 provided further evidence for neutrino mixing and a non-zero mass *On behalf of the MINOS Collaboration.

590

591

Spectrum ratios | ME7,4x1O20p-O.t c 1.2 0.8

0.6

0.4 ^ r

t+^ f~

+tl#ti Strnafeted dota

0.2 J +40 P-

4

8 8 10 Neutrino «n«rgy (Gev)

x 10

Allowed regions

0.4 0.35

L 7Axi(fp O.t

0.3 0.25

L

C "^'r^ZZzz

: 90RC.U 0.2 - .—- 99J5C.L 0.1S |_ * Input parameters : 5uptsr-K, 90% C.L 0.1 : i • • 1 i i i 1 , , . 0.6 0.9 0.7 0,8

1

Figure 1. Monte Carlo studies of the MINOS sensitivity for an integrated flux of 7.4 x 10 2 0 protons on target. Points with error bars in the plot on the left show the ratio of the expected oscillated measured energy spectrum in the far detector for a hypothetical A m 2 = 0.0025 eV 2 and sin 2 20 = 1.0 to the no-oscillation prediction. The plot on the right shows the resulting regions of sensitivity of the oscillation parameters. The effects of the hypothesis of neutrino decay and decoherence are also shown in the figure on the left.

of neutrinos. These fundamentally new phenomena extend the horizons of physics. Only further and more precise studies will help to better understand them. MINOS, a next generation accelerator long baseline neutrino experiment is expected to make strides in these explorations. 2. The Physics Program of MINOS MINOS is conducting a two-detector long baseline study of neutrino oscillations. To cover an atmospheric range of Am 2 and sin22# neutrino mixing, MINOS employs two magnetized toroidal tracking calorimeters 735 km apart and a neutrino beam which uses the Main Injector at Fermilab. The main physics goals of MINOS are: (i) Decisive observation of the v^ -» vT transition and verification of its dominating role in the atmospheric Am 2 region. (ii) Measurements of Am 2 and sin22# mixing parameters with about 10% level of accuracy. (iii) Measurement (or improvement of existing limits) of the v^ ->• ve , the Up -> Sterile transitions, or test more exotic hypotheses for neutrino disappearance, such as neutrino decay, extra dimensions, or decoherence, all for the atmospheric Am 2 . (iv) Test the CPT symmetry in the atmospheric neutrino interactions through independent detection of the sign of the muon charge in the charged current events.

592

Figure 1 illustrates the main MINOS technique to study neutrino oscillations and the sensitivity for the integrated number of protons on target of 7.4 x 1020 (which possibly corresponds to about 3-year running time). The mixing parameters will be determined by measuring the disappearance of the charged current events, as projected from the near detector to the far detector. 3. The experimental setup 3.1. The NuMI beam line The NuMI beam line, schematically depicted in Figure 2, is about 1 km long and declines toward the Soudan Mine in Minnesota at a 58 mrad angle. It uses 120 GeV protons extracted from the Main Injector. Presently, the 8GeV Booster provides up to 6 proton batches into the Main Injector. Most of the time the first batch is used for the production of anti-protons for Tevatron, so a typical 5-batch proton intensity in NuMI is about 2.2 x 10 13 protons per pulse (ppp) (a 6-batch intensity of 2.7 x 10 13 ppp has recently been achieved). The primary beam from the Main Injector is delivered Absorber

Hadron Monitor

Muon Monitors

im

rtm » i 210«

Figure 2. A schematic not-to-scale view of the NuMI neutrino beam line with the main focusing and beam monitoring elements.

in a 8.67 fis single-turn extraction every 2-3 s onto a graphite target of 6.4 x 15 x 940 mm 3 segmented longitudinally into 48 fins. The target is water cooled and designed to withstand up to 4 x l 0 1 3 ppp. Early problems in the cooling system of the target have been successfully resolved and the beam line is operating well. Particles produced in the target are focused by two magnetic "horns", which generate strong focusing toroidal magnetic field when pulsed with about 200 kA peak current. The sign and momentum selected secondaries enter then the 675m-long and 2m-diameter decay pipe. The two-horn sys-

593

tern yields a wide-band neutrino beam. The central energy of the beam is determined by the relative position of the two horns and the target, which select a particular focusing momentum band for pions and kaons produced in the target. The target can be moved remotely in a matter of minutes, while changing the position of the second horn requires several weeks of effort and has not yet been attempted. Presently, the horns are separated by about 10 m and the downstream end of the target is inserted into the upstream horn. By shifting the target upstream different neutrino spectra are achieved. As a commissioning step as well as a part of the beam

Figure 3. Left: Preliminary charged current energy spectra in the near detector registered for the main three beam configurations. Right: Intensity spectra from three beam muon monitors taken for the "LE-10" beam setting described in the text.

systematics studies program, MINOS has exercised several target and horncurrent settings: the "LE" setting was with the upstream target-end 35 cm in front of the first horn and horns pulsed with 200 kA, the "LE-10" setting had the target 10 cm more upstream and the horns pulsed with 185 kA, the "PME" setting had the target 90 cm further upstream and the horns pulsed with 200 kA, and the "PHE" setting had the target 150 cm yet further upstream and the horns pulsed with 200 kA. Also, for calibration and testing purposes MINOS took data with the "LE-10" configuration and the horns pulsed with 170, 185 or 200 kA. The resulting preliminary energy spectra of charged current events in the near detector for the main three configurations are shown in Figure 3. At the end of the decay volume the beam line is instrumented with ionization chambers to monitor the secondary beam. 10 The hadron monitoring station is located just upstream of the beam absorber. The three muon monitoring stations are located in three excavated enclosures down-

594

stream of the beam absorber with 0 m, 12 m, and 30 m rock thickness to the absorber. The pointing accuracy of the muon stations allows to align the neutrino beam direction to about 25 /iradians in one spill, as illustrated in Figure 3. The physics reach of MINOS directly depends on the integrated neutrino flux so improvements in the proton intensity are paramount to the success of the experiment. The currently operating beam is expected to deliver about 1 x 1020 protons on target in 2005. Higher beam intensities are needed and future scenarios for increasing intensity provide for a gradual increase to 6 x 10 20 protons on target per year.

3.2. The near and far

detectors

The 1 kton, "near" detector operates 1,040 m downstream of the target, or 300 m downstream of the end of the decay pipe, and about 100 m underground. The near detector determines the neutrino beam characteristics such as the energy spectrum, the rate of neutral and charged current events, and the flavor composition of the beam. While the entire detector can be used for beam monitoring only the central region within a 25 cm radius around the beam axis will be used for the neutrino oscillation analysis. The larger mass, 5.4 kton, "far" detector is placed in the Soudan Mine 735 km away from the target. The role of this detector is to measure the rate and energy of the v^ neutral current and charged current interactions and identify statistically a fraction of i/e or anti-neutrinos. Both detectors have an identical segmentation, but different front-end readout optimized for cost, a low event-rate in the far detector and multiple events per spill in the near detector. The far detector consists of 484 scintillator planes assembled in two super-modules, each an 8 m-diameter octagonal toroid. The near detector is a 3.8 m by 4.8 m octagonal toroid. Each 2.54 cm-thick steel plane supports scintillator strips which are 1cmthick and 4.1 cm-wide and grouped into 20- or 28-strip wide light-tight modules. The scintillator strips were extruded using polystyrene doped with 1% of PPO and 0.03% of POPOP. Each strip is co-extruded with a reflective outer layer, except along a 1.4 mm-wide and 2.0 mm-deep groove for embedding a light-collecting 1.2 mm-diameter wavelength-shifting fiber. In the far detector, fibers from strips spaced apart by about 1 m in a scintillator plane are grouped into sets of 8 and at each end are coupled to one pixel of a 16-anode photomultiplier (PMT), n so that each PMT reads out 128 fibers. In the near detector fibers are read out individually from

595 one end by 64-anode PMTs. 12 In the far detector the front-end electronics is based on the VA chip from IDE Inc., and in the near detector it is based on the modified QIE chip designed at Fermilab. The data acquisition reads out effectively all the hits in the detectors during a NuMI spill. The far detector has been in full operation since August 2003 and since that time has been used as an observatory for the atmospheric neutrinos. The near detector was fully commissioned at the end of 2004 and is now collecting hundreds of thousands of beam neutrino interactions. 3.3. The calibration

detector

One of the technical challenges before MINOS is to determine the energy scale between the two distant detectors located deep underground. It will be necessary to determine the relative (inter-detector) energy scale to within 2% and the absolute scale to 5% at each location. An important component of understanding the detector response to beam and cosmic ray muons, hadrons, and electrons was a special MINOS calibration detector (CalDet) which was exposed to a test beam at CERN in 2001-2003. Measurements by CalDet provide an energy and topology response function for MINOS and allow the transfer of the energy scale. CalDet events can also be used to tune the Monte Carlo simulations. CalDet was a non-magnetized mini-MINOS detector. The l x l m 2 transverse size and 60 planes contained hadronic and electromagnetic showers with energy of several GeV. The readout fibers were lengthened to approximate a large size of the far and near detectors. CalDet has verified the MINOS calibration scheme and the expected detector response. Figure 4 shows the hadronic and electromagnetic preliminary energy resolutions measured by CalDet and expected to characterize both larger de-

-J.3

mi

-8

_ '•

0,5

-

o ?

fl

01S • ~

04

0.2

-

•v.

•

•.! •o*c*v II0G.V • UG.V T70O«v • > "-*• f40O«V *«0G«V |
_

A :WL

-J

0 1 -•

*

Available E (G«V)

M

*

•

B

»• ••'•

0 05

" •

•

r

^

W P

Figure 4. Preliminary results from the MINOS calibration detector (CalDet). T h e measured hadronic and electromagnetic resolutions, shown in the left plot, are 5 6 % / \ / E and 2 3 % / \ / E , respectively. The plot on the right shows the CalDet response for electrons of energies 0.6 to 9.0 GeV.

596

4. B e a m neutrinos NuMI beam is the most intense high energy neutrino beam ever constructed. Even at the low-energy setting there are multiple interactions per spill in the near detector. The front-end electronics readout, the event topology, and the timing of recorded scintillator signals allow to reconstruct most of events without biases. Figure 5 shows the scatter plot of the transverse coordinates of vertices of charged current interactions recorded in the target section of the near detector and illustrates the statistical strength of the MINOS near detector. The figure also shows a preliminary distribution of time of events in the far detector relative to the spill time. These charged current events were sampled from the full data set following a blind analysis procedure. As expected, the events clearly cluster within a ~10 /is spill window.

•3

-2

MINOS PRtllMiNAHV

Figure 5. Preliminary results from the near and far detector. The left plot shows the scatter plot of the reconstructed transverse coordinates of vertices of charged current interactions. The density of the events clearly shows the cross-sectional shape of the detector, the central coil hole and two different sizes of scintillator planes. The events were recorded for the three main beam energy settings shown in Figure 3. The right plot shows the timing distribution relative to the spill. The reconstructed and identified charged current cluster within the central 10 /is reflecting the duration of the spill. The identified cosmic ray muon events are shown by a dashed histogram.

5. Atmospheric neutrinos in the far detector The far detector is located 713 m, or 2,070 meters water-equivalent, underground and can naturally serve as a cosmic ray observatory. This is the first large underground detector which is magnetized and thus capable of measuring momenta and the sign of the cosmic ray muons, permitting to distinguish between neutrino and anti-neutrino interactions.

597 MINOS detectors are optimized for interactions of beam neutrinos originating from Fermilab which exhibit a mostly horizontal flow of energy. The detector lacks hermeticity for interactions of particles entering from the side, from above or beneath it. A cosmic ray veto shield surrounds the top portion of the detector to mitigate this shortcoming. 5.1. Upward

muons

A separation of downward-going from upward-going muons can be achieved by requiring that muons traverse a minimum of 20 planes or 2 m, have wellreconstructed trajectories within the fiducial volume of the detector, and their scintillator signals have good time records in the crossed planes. Figure 6 shows preliminary distribution of a signed ratio 1//3 of the speed of light to the velocity of well-measured muons. Downward muons cluster around l/{3 « + 1 , while upward muons, resulting from neutrino interactions in the rock beneath the detector, have 1//3 « — 1. 30

• Data ~~ MC - No OsciSiatSons --• 9in"j20)=1

CM

J2 c ©

liho MINOS Preliminary

-0.8 -0.6 -0.4 -0.2 cosQ Figure 6. Preliminary results for upward muons. The plot on the left shows the distribution of the inverse speed of light, 1//?, of cosmic muons reconstructed in the far detector for 464 days of running (160 days with the reversed magnetic field). 91 events near 1//3 = — 1 are identified as upward-going. The plot on the right shows the distributions of events as a function of the cosine of their angle on incidence with respect to the zenith. The histograms show the expected distributions for no-oscillations and for Am§ 3 = 0.0024eV 2 and sin 2 20 2 3 = 1-0.

5.2. Contained

vertex

events

Charged current atmospheric neutrino interactions with a vertex contained within the fiducial volume of the far detector are required to have a well

598

reconstructed track and no associated hits in the cosmic ray veto. The track direction is determined from timing in the scintillator planes. In addition, the events are imposed charge-weighted topological cuts determined from Monte Carlo studies. Using data recorded in the 18 month period from August 2003 to February 2005, which correspond to a live-time of 418 days or an exposure of fiducial 4.54 kiloton-years, 107 events were identified. Among these, 69 are fully contained and 38 are partially contained (i.e., have muon which exits the detector). For no neutrino oscillations 127 ± 13 are expected, while 96 ± 10 are expected for A m | 3 = 0.0024eV2 and sin 2 20 23 = 1-0. Through timing, the events can be further classified to upward or downward-going. Finally, 54 events with cleanly reconstructed tracks have the sign of the muon well measured which allows to attribute the interactions to neutrinos or anti-neutrinos. Figure 7 shows the charge-separated up-down distributions of contained events for muon neutrinos and muon anti-neutrinos. This is a low statistics measurement compatible with previous experiments with atmospheric neutrinos, but for the first time showing separate zenith angle distributions for neutrinos and anti-neutrinos. A full analysis and results are described in an article submitted for publication. 13

0 Data M C n o osculations - - - MC (ifiii 2 = 4 m ! m 0.0024 tV 2 / MC(Jim1=leVJ,AmJ=a0024*V1) • MC ba ckground

==+-

•OS

05 cos 9 „

Figure 7. The charge-separated up-down distributions of contained events cleanly identified in the far detector as muon neutrinos and muon anti-neutrinos.

6. Outlook At the time of this writing MINOS has integrated about 6.5 x 1019 protons on target or about 2 million events in the near detector. The events in the near detector are being used for tuning the Monte Carlo modeling of both the neutrino interactions and the detector response. A significant effort is

599 currently directed toward the optimization and improvements of the event reconstruction and the energy calibration of b o t h detectors. For the longer term, the precision with which MINOS will measure neutrino oscillation parameters strongly depends on the neutrino flux, thus the collaboration is involved in the R&D program to improve the proton b e a m intensity. For the far detector, MINOS has adopted the blind analysis protocol which at present precludes statistical assessment of the events. However, the collaboration plans for an internal "unblinding" at the end of 2005 to reassess the o p t i m u m of the beam setting. First MINOS publication with results on b e a m neutrino oscillations are expected in 2006. This work was supported in p a r t by grant DE-FG03-93ER40757 of the US Department Energy. References 1. The MINOS Collaboration includes research groups from Argonne National Laboratory, University of Athens, Benedictine University, Brookhaven National Laboratory, California Institute of Technology, University of Cambridge, College de France, Fermi National Accelerator Laboratory, Harvard University, Illinois Institute of Technology, Indiana University, ITEP Moscow, Lebedev Institute, Lawrence Livermore National Laboratory, University College London, University of Minnesota, University of Minnesota-Duluth, University of Oxford, University of Pittsburgh, IHEP Protvino, Rutherford Appleton Laboratory, University of Campinas, University of Sao Paolo, University of South Carolina Stanford University, University of Sussex, Texas A&M University, University of Texas at Austin, Tufts University, Western Washington University, College of William and Mary, and Universisty of Wisconsin. 2. Fermilab Proposal P875, February, 1995; The MINOS Detectors, Technical Design Report, Fermilab, NuMI-L-337, October, 1998. 3. Y. Ashie et al., Phys. Rev. D 7 1 , 112005 (2005). 4. S. Fukuda et al., Phys. Lett. B539, 179 (2002). 5. Q. R. Ahmad et al, Phys. Rev. Lett. 89, 011301 (2002). Q. R. Ahmad et al., Phys. Rev. Lett. 89, 011302 (2002). 6. M. Ambrosio et al., Eur. Phys. J. C36, 323 (2004). 7. M. C. Sanchez et al., Phys. Rev. D68, 113004 (2003). W. W. M. Allison et al., Phys. Rev. D72, 052005 (2005). 8. E. Aliu et al, Phys. Rev. Lett. 94, 081802 (2005). 9. T. Araki et al, Phys. Rev. Lett. 94, 081801 (2005). 10. D. Indurthy et al., it AIP Conf. Proc. 732, 332 (2004). 11. K. Lang et al., Nucl. Instrum. Meth. A545, 852 (2005). 12. N. Tagg et al., Nucl. Instrum. Meth. A539, 668 (2005). 13. P. Adamson et ah, First Observations of Separated Atmospheric Muon Neutrino and Muon Anti-neutrino Events in the MINOS Detector, to be submitted to Phys. Rev. D.

ATLAS I N N E R D E T E C T O R RESULTS F R O M T H E 2004 C O M B I N E D T E S T B E A M DATA

W. LIEBIG and R. PETTI* CERN, 1211 Geneve 23, Switzerland E-mail: Wolfgang. [email protected], [email protected]

The ATLAS Inner Detector has taken part in the 2004 Combined Test Beam programme and has made efficient use of this unique opportunity to exercise the hardware and software integration before commissioning the full ATLAS detector. Data from runs with e, it, /j. and 7 beams have been used to determine the detector performance (noise, occupancy, stability etc.), alignment corrections and reconstruction performance of the Inner Detector and its contribution to combined particle reconstruction with the calorimeters and muon spectrometer. Preliminary results are presented and compared to test beam simulation, allowing feed-back and improvements to the ATLAS simulation long before the start of LHC.

1. Introduction The ATLAS experiment 2 is a general-purpose detector for the Large Hadron Collider at CERN and consists of an inner tracking detector, calorimeters and muon spectrometer. As final part of a 10-year long detector development and testing programme, the ATLAS collaboration has operated modules from all detector systems in a Combined Test Beam. It has been arranged to resemble as close as possible a slice of ATLAS as it will be seen by particles originating from the interaction point of the LHC beams at polar angles of 25° and 90°. This paper describes the Inner Detector (ID) part of the Combined Test Beam and first results obtained from the data. The ID group has gained valuable experience from integrating the hardware and software in this small-scale version of the ATLAS experiment, from studying the performance of the sub-systems and of the entire ID and from the combined particle reconstruction using also the calorimeters and muon chambers. The comparison of the real data to a dedicated test beam simulation allows to tune the detector simulation and digitisation. * On behalf of the Inner Detector Combined Testbeam working group and of the ATLAS Collaboration. See ref. 1 and 2 for a detailed list of authors.

600

601

Figure 1. ATLAS Combined Test Beam setup as described by the Geant4 simulation.

1.1. Setup and

Data-Taking

The test beam setup consisted of six pixel detector modules arranged in three tracking layers, eight silicon strip detector modules (semiconductor tracker, SCT) in four layers and 6 transition-radiation tracker modules (TRT). The pixel and SCT modules were placed in a magnetic field of 1.4 T while the TRT - unlike the full ATLAS - was outside the field due to space constraints. Charged tracks were bent in the direction which is covered by the precision co-ordinate measurements (50 fxm pixel pitch and 70/jm average strip pitch). The second co-ordinate is determined by measurements from both pixel and SCT (using a 40 mrad stereo angle between strips from back-to-back modules). A track then passes though an average of 40 TRT straw tubes measuring its distance to a central wire from the drift-time information. The TRT thus provides only measurements in the direction of magnetic deflection. The visualisation from the detector simulation in Fig. 1 shows how the ID is then followed by the calorimeters and muon chambers. The ID has taken data in full combined mode in Oct-Nov 2004. It has collected data from runs with /x ± , TT*, mixed e±/n±, p and 7 beams at energies ranging from 1 to 350 GeV. For most runs data were taken with and without magnetic field. A total of 22 million events have been validated for the analysis of detector performance.

602

1.2. Event Reconstruction

and

Simulation

The software chain to reconstruct test beam events is in most cases identical to what will be used for real ATLAS data. Nevertheless many components had to be newly developed mainly due to the need to deal with real data and due to the commissioning of an entirely new, modular reconstruction chain in the ATLAS software framework. The offline software commences by decoding the raw data stream followed by the clustering and drift circle creation, the pattern recognition and tracking and finally the output to data formats suitable for analysis. Geant4 has been used to simulate the test beam, obtaining the geometry information from a detector description common to reconstruction and simulation. The output hits from the simulation are transformed to raw measurements by the digitisation code which simulates the charge drift and diffusion as well as signal read-out.

2. Calibration and Alignment The reconstruction of real data requires the software to deal with alignment and calibration corrections, and dead/noisy channel information. Once determined for a denned interval of validity, they are stored in the conditions database and provided to the detector description and reconstruction code. Dead/noisy channels have been monitored for all three detector technologies and are masked in the reconstruction. The pixel and SCT module positions are aligned in the two directions orthogonal to the beam using an iterative procedure which minimises the residuals of straight line tracks on a small set of selected reference runs without magnetic field. The residuals obtained after alignment are displayed in Fig. 2 (left) for all six pixel modules combined. The resulting sigma is 16 fim, which is slightly smaller than the simulation with a sigma of 17.4 /mx. For the SCT, residuals with a sigma of 11.2/un in data and 12.2 fan in simulation have been found. For the TRT, the calibration and alignment are performed in a common procedure which determines the drift velocity in the tubes, the drift time offset and the wire position from the extrapolation of tracks reconstructed in silicon detectors. While the intrinsic precision of the overall ID alignment is close to 1 (j,m, studies on the stability of the alignment as a function of run indicate residual variations of ~ lOfim for the silicon detectors, and larger for the TRT and for the runs with field on. The alignment strategies foreseen for ATLAS are now being exercised on the test beam data so that they can be validated providing at the same time a better understanding of alignment uncertainties at the test beam.

603

Figure 2. Residuals obtained on data with magnetic field off are shown for the pixel detector in the r coordinate (left) and for the TRT (right).

3. Particle Reconstruction 3.1. Momentum

Measurement

The momentum has been measured for different beam energies and tracking strategies as well as for tracks with and without TRT. The results from using only pixel and SCT which are fully contained in the magnetic field reproduce the beam energy very accurately. Including the TRT improves the momentum precision due to the larger lever arm, but also introduces some biases on the reconstructed momentum. The detailed description of the field and the corresponding silicon alignment are being finalised. 3.2. Material

Description

and Multiple

Scattering

The multiple scattering corrections in the reconstruction have been probed by parameterising the residuals between track segments from different detector parts as a function of momentum. A rather good agreement between data and simulation has been found for pion beams as can be seen from Fig. 3. Also the increased multiple scattering due to additional material layers, which were inserted during part of the running to emulate the ID services at low polar angles, is well described by simulation. 3.3. Particle

Identification

with the

TRT

The transition radiation from electron tracks in the TRT is detected by a signal discrimination operating at two different levels. A large measured fraction of high-level to low-level hits on a track can therefore be used to distinguish e from IT. Such a rejection curve is displayed in Fig. 3 for the lowenergy data at 2 GeV showing a much improved agreement with simulation compared to previous stand-alone test beam data after adjusting the highlevel hit threshold in the simulation.

604

DATA PION 1

ti'af

E

MS

,,

I 2GeV \

I.WI2 IJS6±0:06294

0

jj ' 1 3 K 04 f, 8M

> -

i

•— Testb&m

3

• • • MO/KP Qario

OKI

™

(

,X

~ ~ -

_

E 0.

"

df

~

1

1ft ! (L2

t-...j

f

;

.>

;

•

i

i

i

i

i

i llil TH

i l l 111 (....1.1.1.1)1 I 1 1 , t . . l . M . . I

5.S5

0,1

075

OJ

, 1 ( 1 ... 1

H»5 M 095 ELECtfiONEFBaEMCY

Figure 3. Residuals between track segments from different sub-detectors as a function of pion beam energy (left). Pion rejection curve for a 2 GeV e/ir beam (right).

3.4. Combined

Tracking

Tracks from the ID have been matched with hits in the calorimeters. Full correlation between track and shower coordinates is visible after geometric corrections were applied. This technique is employed in the reconstruction of photon conversions inside the ID and for calibrating the e — ir separation. High-energy muon beams form another part of the Combined Test Beam data. They were injected at low settings for the bending fields, such that they could be tracked through the whole apparatus over a length of about 40 m. These muon tracks have been reconstructed successfully by making use of the new software chain. 4. Conclusions ATLAS has learnt a lot from the hardware and software integration work done for the Combined Test Beam and from the detailed analysis of detector performance. In particular, we benefit from the chance to study the combined performance well before the full ATLAS detector takes data. The test beam also serves as a benchmark to test and develop new algorithms. Acknowledgments We would like to thank the members of the ATLAS collaboration and in particular all those involved with the combined test beam. References 1. G. Gorfine, Tracking Performance of the ATLAS Pixel Detector in the 2004 Combined Test Beam, proceedings of Pixel 2005, Bonn. Submitted to NIM A. 2. The ATLAS Collab. ATLAS Detector and Physics Performance: Technical Design Report, CERN/LHCC/00-14. May 1999.

THE CMS MUON SYSTEM PIERLU1GIPAOLUCCI* Istituto Nazionale di Fisica Nucleare, Sezione di Napoli, Complesso Universitario di Monte Sant 'Angelo Napoli, 80126, Italy The Compact Muon Solenoid Muon system, designed to trigger and identify muons, is based on three different technologies; drift tube (DT) in the barrel region, cathode strip chamber (CSC) in the endcap and resistive plate chambers (RPC) in both the barrel and endcap. The three subsystems, the local muon trigger and the global muon trigger will be described here giving some results obtained during the quality control tests. The chambers installation is now underway; the strategy adopted for the commissioning of the muon system will be described in this paper. 1.

Introduction The Compact Muon Solenoid [1] is one of the four experiments of the Large

Hadron Collider [2] at CERN. CMS is a complex apparatus consisting of several subdetectors in order to identify particles and measure their energies and momentum.

Figure 1. The Compact Muon Solenoid detector. * On behalf of CMS muon collaboration

605

606

A principle feature of CMS is its strong magnetic field of 4 Tesla, which is produced by a superconducting coil. The coil volume houses the tracker and the electromagnetic and hadrons calorimeters. An iron return yoke surrounds the coil and houses the muon chambers as is shown in figure 1. Since the very high collision rate and the low cross section of the "new physics" events a very important part of the experiment is the trigger and data acquisition system [3] [4]. The experiment is currently being assembled and tested in a surface hall SX5 and it will be lowered into the cavern during the 2006. 2. The Muon Spectrometer At LHC, electrons and muons will play a fundamental role in the studies of any physics sector; from the Higgs search, where the H —* 1111 are the so called "golden channels", to the "new physics" and from the electroweak precision measurements to the b and t physics. For these reasons the CMS muon system [5] has been designed and built to have excellent trigger, identification and reconstruction performances. The LHC physics signatures and their separation from the large expected background, the muon trigger rates and the trigger limits dictated by the DAQ and trigger systems, impose several requirements to be fulfilled by the muon system: • Muon trigger: excellent trigger performances on single and multi-muons events and an unambiguous identification of the bunch crossing is obtained by combining fast dedicated trigger detectors, Resistive Plate Chambers, with detectors having precise spatial resolution, Drift Tube and Cathode Strip Chambers. • Redundancy, in both trigger and reconstruction, is obtained using three technologies, combined in order to have two independent muon systems in the whole angular region. • CMS has been designed with a magnetic field of 4 Tesla and the possibility to measure muons twice, in the tracker and in the muon spectrometer in order to perform very good results in the muon momentum and charge measurements in the whole t| region and from few GeV up to a TeV. • Muon identification: to achieve a very high efficiency (> 95%) up to r) = 2.4 are required at least 16 interaction lengths of material and a very sophisticated set of algorithms able to use the three sub-detectors data. • Robustness: The detectors must be capable to work with a strong magnetic field and high radiation and interaction background [5].

607

The muon spectrometer consists of 250 DTs, 540 CSCs and 912 RPCs chambers to be built and installed for the 2006. It cover an angular region delimited by 0 < | r) | < 2.4 and is divided in two major regions: barrel and endcap (figure 2). The barrel region (0 < | n | < 1.2) is composed by 5 wheels, each divided in 12 sectors with 4 iron gaps. The muon stations consist of one DT chamber and two RPC chambers joint together and are placed in the gaps of the iron return yoke plates. In this region the magnetic field is low and almost uniform and the expected muon rate is about 1 Hz/cm2 compared with a neutron induced background of 1-10 Hz/cm2. The endcap region (0.9 < | n | < 2.4) has a strong and not uniform magnetic field (up to 3.5 T) with a y and neutron induced background rate of about 1 KHz/cm2. The two endcaps, made of 3 iron disks and 4 layers divided in 2 or 3 stations (ME1/1, ME1/2, ME 1/3, ME2/1, ME2/2, ME3/1, ME3/2, ME4/1 and ME4/2) of CSC and RPC chambers. Drift T u b e s

n=J.s

n=i-°

n=i-2

„ u._«r..._ 0

200

400

600

800

1000

1200 Z(em)

Figure 2. The CMS muon spectrometer (r\ view) is based on three different technologies; Drift Tube (DT), Cathode Strip Chamber (CSC) and Resistive Plate Chamber (RPC).

3. The CMS Muon Trigger system For the nominal LHC design luminosity of 1034 cm~V', an average of 17 inelastic events occurs every 25 ns. This input rate of 109 interactions per second must be reduced by a factor of at least 107 to 100 Hz, the maximum rate that can be achieved by the DAQ. The CMS trigger system [3] is based on two levels, at first level (LI) all data are stored for 3.2 fxs, after which no more than 100 KHz (50 KHz for the

608

first year) of the stored events are forwarded to the second level, called High Level Trigger. The LI is an hardware trigger based on custom electronics while the HLT is a software trigger based on commercial processors. DT hits

local trigger track segments

regional trigger Barrel Track Finder i 4 muon candidates (pt. i). O: quality)

"I

CSC hits

local trigger track segments to. 6$. n. 5n)

regional trigger Endcap Track Finder £4 muon candidates (p,. n-«, quality)

RPC hits

PAttern Comparator Trigger <4 barrel* <4 endcap muon candidates (Pi, n. e>. quality)

Global Muon Trigger £4 muons (p,. n. o. quality)

Figure 3. The Muon trigger schema from the chamber trigger to the Global Muon Trigger.

The LI muon trigger [6] uses all three muon detectors DT, CSC and RPC, each with its own trigger logic as shown in figure 3. DT and CSC first process the local information of every chamber generating local triggers then muons from different chambers are collected by the Track Finder which combines them to form a muon track with an assigned transverse momentum value. Up to four best (highest p T and quality) muon candidates from each system are selected and sent to the Global Muon Trigger. In the case of RPC hits from all the stations are collected (PACT) [7] and if they are aligned along a possible muon track, a p T value is assigned and the information is sent to the Muon Sorter. It select the four highest p T muon from the barrel and four from the endcaps and sends them to the Global Muon Trigger (GMT). Finally transverse momentum thresholds are applied by the GMT for all the trigger conditions. 4. Drift Tube chamber The barrel muon system consists of 4 concentric shells of drift tube chambers (MBl, MB2, MB3 and MB4). Each chamber, except MB4, is made of 3 independent units, called superlayer (SL), and a thick honeycomb plate glued together (fig.4). Each superlayer is composed by 4 layers of drift tubes, with all wires parallel. The wires of odd layers inside a SL are staggered by half-cell width with respect to the even layers. The two external SLs measure the muon trajectory in the CMS bending plane (r,<j>) while the third SL measure the z

609

coordinate. The drift tube is 43 x 13 mm2 and is composed by three electrodes: a wire, two I-beams (cathodes) and two strips.

Figure 4. The drift tube chamber is made by 3 superlayers and a thick honeycomb plate glued together.

The time needed by the primary electrons to reach the anode is measured and the spatial position of the track is calculated from the time measurement and the known time-to-distance relationship. To resolve the left-right ambiguity of a single cell it is necessary to use the staggered layers and reconstruct the track position in at least three of the four layers. A spatial resolution of 100 um in the (r,<(>) coordinate and of 150 (am in the (r,z) coordinate has been measured in the test beam [8]. 4.1. DT local trigger system The DT front-end electronics, called Bunch and Track Identifier (BTI), forms track segment from coincidences of at least three aligned hits in four layer of one DT superlayer, using the mean-timer method, obtaining a resolution of 1.4 mm and 60 mrad. The positions and directions of the segments are sent to the Track Correlator (TRACO), which attempts to combine segments from the two SL of the same chamber and produce <)> coordinate with a resolution of 10 mrad. The Trigger Server collects the two best <j) segments and the 6 segments and selects, at every bunch crossing, two segments with highest quality and p-r. The Track Finder identify the best tracks per sector (30 sectors in (> | and 12 in r\) assigning the (p, r|, p T and quality. 5. Resistive Plate Chamber RPCs has been chosen in both the barrel and endcap as dedicated trigger detectors. Because of their fast response and good time resolution (a < 1.5 ns) they guarantee a precise bunch crossing assignment of the muon tracks.

610

The barrel muon stations are equipped with two RPC layers for the innermost stations (RB1 and RB2) and one layer for the outer stations (RB3 and RB4) for a total number of 6 layers per sectors. The endcap are instrumented with one RPC layer per station for a total of 4 layers. -HV: Fiy-np

B W - UJ>

-HV

-KV eiscswwfe * (grapfcitt)

Figure 5. The schema of a barrel RPC made by two double-gaps with a strip plane in the middle.

An RPC gap is made by two parallel bakelite plates (1-2 1010 flxm) placed at a distance of 2 mm and filled with a gas mixture of 96% C2H2F4, 3.5% i-C4H10 and 0.5% of SF6. High voltage is applied to the outer graphite coated surface of the bakelite plates in order to have an electric field inside the gas gap able to generate a charge avalanche along the track of a ionizing particle. The avalanche induces a signal on the aluminium strips placed outside the gap and isolated from the graphite. An RPC chamber (fig.5) consists of two double-gaps (forward and backward), made by 2 gaps with a read-out strips plane in the middle. Every chamber is equipped with 10, 12 or 18 front-end electronic boards [9] each connected to 16 readout strips. 5.1. Production, test and results RPC barrel chambers are assembled at the HiTec and General Tecnica factories and in the Bari and Sofia laboratories. Then are characterized with a cosmic rays at the station sites of Bari, Pavia and Sofia. The endcap chambers are assembled and tested at CERN. Tests done during the production phase are: gas and cooling tightness, dark current versus high voltage and strip noise rate at the working point (9.5 KV). The chamber is accepted if the dark current distribution at 9.5 KV has a mean value lower than 5 uA and a standard deviation below 1 uA and if the noise rate is less than 5 Hz/cm2. During the cosmic tests the chambers are characterized by

611

the efficiency, dark current and noise rate curves as function of HV working point.

Figure 6. The endcap chamber efficiency distribution at 9.5 KV for a subset of chambers.

The average efficiency measured for the RPC chambers is about 97% that become the 99% using the tracking capabilities of the cosmic tower (fig 6.) [10][11]. The accepted chambers are sent at ISR where the chamber stability is studied for a month, after that RPCs are ready to be joint at the DTs 6. Cathode Strip Chamber The endcaps are instrumented with CSC chambers [12] providing precise time and position measurements. Each chamber is trapezoidal in shape and consists of six gas gaps, each equipped with a plane of radial cathode strips (3.15-16 mm) and a plane of anode wires (spacing 2.5-3.16 mm, diameter 30-50 ixm) running almost perpendicular to the strips (fig.7). w ^ w w Cathode •

Wires

"*-

" Cathode

ii.

— —

Segmented J cathode

Wires

'"""• "-*•' Grounded cathode

Figure 7. The schema of Cathode Strip Chamber with wires and cathode segmented in strips.

The gas ionization and subsequent electron avalanche produces a charge on the anode wire and an image charge on a group of cathode strips. Thus each

612 chamber measures up to six space coordinates (r, <> | , z). The wire signal gives a fast information with a spatial resolution of about 1 mm while a precise spatial measurement (a,A 100-240 urn) is achieved using the center of gravity of the charge distribution induced on the cathode strips [13]. All CSCs, except those in the station ME 1/3 are overlapped in § to avoid gaps in the muon acceptance. 6.1. CSC local trigger system The CSC local trigger uses both the cathode and anode data to found the local charged tracks (LCT). The two trigger views, based on different electronics boards and different algorithms, are able to find up to two local tracks per chamber during any bunch crossing. The two projection are then combined in a 3-D LCT by a timing coincidence. The data of each endcap sector (60°) are collected by the Track Finder that identify the best 3 muons assigning to them <> |, T], PT and a quality bit. 7. Muon system installation and commissioning The installation and commissioning of the muon system is going on since the 2004 when the first CSC chambers have been installed. At present every muon sub-systems is working on the installation and commissioning of the system following its own test procedures. CSC: every installed chamber is commissioned with a gas leak test, an HV and electronics test made with a portable DAQ in which cosmics (fig. 8) are taken in a self-triggering mode. After that a long term electronic test is done keeping the low voltage on for a long period. Up to now about the 96% of the CSC chambers have been installed and 50% of those have been commissioned.

Figure 8. A cosmic ray observed with CSC chambers during the commissioning.

613

DT-RPC: after having been coupled with the RPC at the ISR, the chambers are sent at SX5 where are installed in the iron. At that point every chamber is commissioned testing the gas and the cooling connections, the high voltage and the electronics integrity, looking at dead channels and measuring the noise rate. More tests are under development in both the DT and RPC groups. 20 sectors of the wheels +1 and +2 have been already installed, corresponding to about the 30% of the whole system. About 50% of the endcap RPC chambers (reduced system) have been produced and they will be installed from October 2005. The commissioning will follow the same barrel procedures [14]. 8. Expected muon system performances A full detector simulation has given the following results for the transverse momentum resolution of muon tracks (fig.9): • Muon system stand-alone ^ 8-15% ApT/pT at 10 GeV V 16-53% ApT/pT at 1000 GeV • Muon system with tracker matching S 0.8-1.5% ApT/pT at 10 GeV S 5-13% ApT/pT at 1000 GeV For low p T the resolution measured in the muon system is limited by multiple scattering in the iron yoke. For high p T the resolution is limited by the muon chambers resolution, even when the tracker is taken into account. 10';;'

1

"'

'

" " " •

Drifl Tubes stand alone

i Sbjcmreconstructionefltaloitcy (W~*H..p\y)>20 3BV '

" "

-•'"!

• 10 | •

•

;

•• *i • »

r

+ tracker i

: . barrel:rj" • 0.5 10

,„.*,.,

to

J . * . ^ *

io

~1T 0.0-

J CSC siand alone 1

. . . . . ^

to

•

.

-

• *

+tracfcer

c ! 04

.

03

ji 1

. 1 . ,

io

p, (QeV)

forward: ti= 2.0 ; ! "a 10

' ~i 10

J

0.? 0."

% «

B.4 U

M

1

14! M

1.»

l.»

2

1

Figure 9. The transverse momentum resolution of the muon tracks (with and without tracker matching) (left) and the muon reconstruction efficiency versus r| (right).

614 The efficiency of the reconstructed muon tracks is almost flat in r\ up to | r] | ~ 2.1 as is shown in figure 9, for W —* uv events [15]. The Global Muon Trigger combines 3 independent triggers having a very high efficiency over the full r) angular coverage. The efficiency of the LI muon trigger to identify a single muon as a function of the generated r| is shown in figure 9.

0

0.5

1.5

1

Barrel Track Finder

.1.

*T* DT

2

Endcap Track Finder .

1 s .„ (m its)'

CSC

RPC

Figure 10. Efficiency of the level-1 muon trigger to identify single muon tracks as function of n.

The DT, CSC and RPC trigger systems show different efficiencies in r) and §, and different behavior in p T due to cracks, construction properties and different trigger algorithms. The GMT can make use of these differences resulting not only in a higher efficiency overall but also a smoothing effect on the less efficient regions as is shown in figure 10. I < 1.04

I

1

l'°'lr

Endcap: 1.04,<|rH< 2.4

* ^

I

S>10 *

•^w " - *$£"

£

E

io't F ,1 10 •r

— Generated — - DT • RPC - * - GM1 opt GMT and CM t or

ioi

i

L -»-GMTopt 10 * CM I <ml CM i nr

,r

imion p T (GeV/c)

Figure 11. Muon trigger rate in barrel and endcap region as function of p-r.

615

In figure 11 is shown the single muon trigger rates of the regional triggers and of the GMT as a function of the p T threshold applied (at a luminosity of 1034 c r n V ) . A single muon rate of 8 KHz is obtained with a p T threshold of 25 GeV (efficiency 90%). The dimuon trigger rate is 2.8 KHz with a p T threshold of 8.5 GeV. 9. Conclusion The construction of the three muon detectors is close to the end (summer 2006) and about 96% of the CSC and 30% of the DT-RPC chambers have been already installed at SX5. The commissioning of the system is going on showing a very good results. In the first part of the 2006 a very important test, called cosmic challenge, will be done using a fraction of the whole CMS detectors (2 or 3 sectors for the muon system) in order to verify the estimated global performances of the muon system described in this paper. References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.

12. 13. 14. 15.

CMS collaboration, Technical Design Report, CERN/LHCC 94-38 (1994). The LHC project at CERN, LHC-project-report-36 (1996). CMS collaboration, The Trigger and Data Acquisition project, Volume I CERN/LHCC 2000-038 (2000). CMS collaboration, The Trigger and Data Acquisition project, Volume II CERN/LHCC 02-026 (2002). CMS collaboration, The Muon project, CERN/LHCC 97-32 (1997). H. Sakulin and A. Taurok, "The LI Global Muon Trigger for the CMS experiment", CMS-CR-2003-04 (2003). M. Andlinger et al., "Pattern Comparator Trigger for the muon system of the CMS experiment", CERN-PPE-94-22. C. Albajar et al., Nucl. Instr. Methods A 525 (2004) 465. M. Abbrescia, et al., "New developments on front-end electronics for the CMS Resistive Plate Chamber" Nucl. Instr. Methods A 456 (2000) 143-14. M. Abbrescia, et al., Nucl. Instr. Methods A 508 (2003) 137-141. M. Abbrescia, et al., "The cosmics rays quality tests procedures for the CMS barrel resistive plate chambers", Nucl. Instr. Methods A 533 (2004) 208-213. G. Charpac, et al., Nucl. Instr. Methods 167(1979) 455. D. Acosta , et al., Nucl. Instr. Methods A 494 (2002) 504-508. A. Sharma, CMS RPC Endcap Status Report, CERN-OPEN-2004-041. I. Belotelov, N. Neumeister, "Performance of the CMS offline reconstruction software", CMS AN 2005/010 (2005;.

P E R F O R M A N C E S OF T H E ATLAS LEVEL-1 M U O N T R I G G E R PROCESSOR IN T H E B A R R E L

V. BOCCI, G. CHIODI.G. CIAPETTI, D. DE PEDIS, A. DI GIROLAMO, A. DI MATTIA, E. GENNARI.C. LUCI, A. NISATI, E. PASQUALUCCI, FR. PASTORE, E. PETROLO, F. SPILA, R. VARI, S. VENEZIANO, L. ZANELLO Un. of Rome "La Sapienza" and I.N.F.N. Rome, Italy G. AIELLI, R. CARDARELLI, A. DI CIACCIO, A. DI SIMONE, L. DI STANTE, A. SALAMON, R. SANTONICO Un. of Rome "Tor Vergata" and I.N.F.N. Rome 2, Italy A. ALOISIO, M. ALVIGGI, V. CANALE, G. CARLINO, F. CONVENTI, R. DE ASMUNDIS, M. DELLA PIETRA, D. DELLA VOLPE, P. IENGO, V. IZZO, A. MIGLIACCIO, S. PATRICELLI, G. SEKHNIAIDZE Un. of Napoli "Federico II" and I.N.F.N. Napoli, Italy E. BRAMBILLA, G. CATALDI, E. GORINI, F. GRANCAGNOLO, R. PERRINO, M. PRIMAVERA, S. SPAGNOLO Un. of Lecce and I.N.F.N. Lecce, Italy V. APRODU, D. BARTOS, S. BUDA, S. CONSTANTIN, M. DOGARU, C. MAGUREANU, M. PECTU, L. PRODAN, A. RUSU, C. UROSEVITEANU NIPNE-HH Bucarest,

Romania

The ATLAS level-1 muon trigger will select events with high transverse momentum and tag them to the correct machine bunch-crossing number with high efficiency. Three stations of dedicated fast detectors provide a coarse PT measurement, with tracking capability on bending and non-bending projections. In the Barrel region, hits from doublets of Resistive Plate Chambers are processed by custom ASIC, the Coincidence Matrices, which performs almost all the functionalities required by the trigger algorithm and the readout. In this paper we present the performance of the level-1 trigger system studied on a cosmic test stand at CERN, concerning studies on expected trigger rates and efficiencies.

616

617

1. Barrel Level-1 Muon Trigger overview TGC2 —

Figure 1. The ATLAS Level-1 Muon Barrel Trigger schema.

The ATLAS muon trigger system 1 provides the correct association of the muon candidates with the corresponding bunch-crossing and classifies tracks based on their transverse momentum. Rejection of cavern background is assured by fine time resolution and required independent coincidence on bending (rj) and not bending ((p) views in dedicated systems . In the Barrel, three layers of Resistive Plate Chambers (RPC) are used, tightly integrated with the MDT precision chambers in the muon spectrometer. Two of them, located inside the air-core toroids, provide the Low-px trigger selection (px >6 GeV/c) required by B-physics studies, while an outer station allows to select muons above 20 GeV/c. A detector layer is composed by two RPC gaps, each read out by two layers of orthogonal pick-up strips. A schema of the trigger system is sketched in Fig.l. The trigger logic identifies candidate muon tracks coming from the interaction vertex within a px range. Coincidence of hits is required within geometrical roads, calculated on the basis of the expected curvature of the track, assuring a 90% trigger efficiency at the imposed nominal px threshold. The geometrical roads are projected from the central layer, called pivot, on to the coincidence layers, pointing to the interaction vertex. The High-px trigger is made more robust by requiring the confirmation of a Low-px trigger. The baseline trigger configuration requires a 3/4 majority coincidence for the Low-p T system and a 1/2 for the High-px • Different criteria, allowed by the programmability of all the trigger parameters, can be imposed in order to control the trigger rate under different conditions

618 of luminosity, background and detector performance. The logic for the Level-1 trigger relies on custom ASICs, called Coincidence Matrix (CMA) 2 , which apply time and space coincidences on hits from the pivot and the coincidence layers and provide the trigger result and the readout of the RPC strips with a time resolution of 1/8 of a 25 ns bunch crossing period. The granularity is given by the region of interest (Rol), A<j> x ATJ = 0.1x0.1. The trigger and readout information from two (j) and two 77 CMA's are combined in a PAD, the on-detector trigger electronics processing unit. The algorithm of a tower is performed by connecting a Low-px and a High-px PAD. The latest collects all the trigger and readout data and sends them, via a single optical link, to the off-detector electronic system using a time multiplexing approach, which gives highest priority and fixed latency to trigger data. The Receiver and Sector Logic (RX/SL) board 3 combines data from up to eight PAD boards, generating the trigger signal of one of the 64 trigger sectors of ATLAS. When a level-1 accept signal is generated, the readout data are made accessible, through dedicated readout buffers, to the higher level triggers. Stringent requirements must be fulfilled by the CMA trigger processor ASIC. All the trigger part of the electronics is organized in pipelines, running at the bunch crossing frequency, and the readout buffers contained in the chip allow for a level-1 latency of maximum 2.5 [is. The CMA processor generates the output trigger patterns with a latency of a few bunch crossing periods. After digitization, RPC signals are aligned in time with programmable delay lines and shaped, adding dead-time if necessary, in order to perform the coincidence. The time alignment of signals from different layers and stations within the same bunch crossing window is required, compensating delays due to time of flight of muons, signal propagation along the RPC strips and the cabling. The CMA chip allows a time calibration with 16 channels granularity and the readout windows is optimized to perform bunch crossing identification with high efficiency and preserve the readout information.

2. Studies of performances of a trigger tower The whole trigger slice has been extensively tested with 25 ns beams to ensure the integrity of readout and trigger chains 4 . The CMA chip underwent a second and final redesign to cope with extended input delay lines of up to about 200 ns 5 . We used some pre-production samples to perform additional tests with real RPC signals at the CERN BB5 site with cosmic

619 Table 1. Summary of the trigger tower acceptance measurements (Low-px , 3/4 majority) with different R P C working conditions. The expected value from the toy MC is 47% at full detector efficiency. R P C HV (V) R P C F E threshold (V) Ll-trigger acceptance ±0.5 (%)

9400

9.500

9600

9700

9600

9600

1.1

1.1

1.1

1.1

0.9

1.0

42.7

44.6

45.4

46.0

42.7

44.2

rays a . A tower with up to three RPC stations has been equipped with the trigger electronics, in order to set up one trigger tower on each half-chamber side. We used Barrel Medium Small (BMS) chambers, corresponding to the inner chambers of a small sector of ATLAS, and reproduced the complete level-1 trigger slice of ATLAS. A reference trigger signal has been provided by two additional RPC chambers, situated at the top and bottom of the tower. To set up the system, we programmed readout windows with 8 BC's length and used the raw hits time distributions from each RPC layer to extract the calibration constants needed for time alignment. These data were very useful to validate calibration procedures. For validating the trigger logic and the coincidence procedure, a simple toy Monte Carlo (MC) has been generated reproducing the geometry of the tower and the cosmic rays angular distribution at the ground (oc cos 2 0). This model identifies a muon tracks with 100% efficiency, digitizing chamber hits with a cluster size equal to 1. Imposing the Low-px3/4 coincidence in extended full geometrical roads, a 47% expected trigger acceptance is obtained. We measured the same quantity, convoluted with the RPC efficiencies, for detectors at different high voltage values and different front-end thresholds15. Results listed in Tab. 1 are coherent with the expected value and show the good robustness of the trigger system. Using dedicated counters on the trigger processors, we checked the consistency of cosmics trigger rates, when running in self-trigger mode. With the Monte Carlo we calculated a rate of 290 Hz for a single tower, from 2004PDG estimations of muon fluxes at sea level passing through a horizontal detector (1 hit cm _ 2 min _ 1 ). When the RPC detectors efficiency is introduced, this value is corrected for the expected 3/4 coincidence probability. We gave a raw estimation of the detectors efficiency, measuring the probaa

T h i s location is used for validating RPC-MDT chambers integration before being installed in the ATLAS site. b Front-end logic is inverted, so that the highest low-voltage discriminator threshold is the loosest.

620

"350 E 300 250 200 150 Tower trigger rates (Low p 3/4 majority) 100

b

• data —-_.„.} expected from cosmic muon flux

50

, . , . i , . , , i , , . . i , , , , i

9.4

9.45

9.5

9.55

9.6

9.65

9.7 HV(kV)

Figure 2. The measured tower trigger rate compared with the values estimated from the Monte Carlo rate, convoluted with detectors efficiency.

bility of a hit on a single layer (on each view) in events where all the other layers have at least one. This is an overestimation of the real efficiency, as it includes the random hits contribution. We introduced a systematic error of 4%, obtained by subsequent measurements with precise MDT tracking for single detector operating point. The comparison between measured and expected values is shown in Fig. 2. Errors on MC estimation take into account the raw parametrization on the muon flux (10%), the MC model (5%) and the estimation of efficiency. Errors on data are given by trigger rate variations during data taking (~4 Hz), which remained symmetric and stable on both chamber sides. 3. Conclusions The reported studies allowed the development of tools to control the stability of the Muon Barrel trigger system, monitored with real RPC data. The timing alignment precedures which have been optimized will be the starting point for the commissioning of the system. References 1. 2. 3. 4. 5.

ATLAS First-Level Trigger Technical Design Report, CERN/LHCC/98-24 V.Bocci et al. , IEEE Trans, on Nucl. Sci. 50, n. 4 (2003) . A.Salamon et al., Procs. 7th Wks on Electr. for LHC Exp., Stockholm (2001) G.Aielli et al., IEEE Trans.Nucl.Sci. 51 n.4 (2004) 1581 R.Vari et al., Procs. 2005 Nucl. Sci. Symp. and Med. Imag. Conf., Puertorico

CERTIFICATION, INSTALLATION AND COMMISSIONING OF MUON DRIFT TUBE CHAMBERS FOR THE CMS CENTRAL MUON DETECTOR JESUS PUERTA-PELAYO CERN-CMM Geneva, Switzerland The Drift Tube chambers for the muon tracking in the CMS barrel are assembled in four different European institutes (Aachen, Madrid, Padova and Turin), and sent to the ISR area (CERN) for storage, dressing and final certification prior to their installation in the iron yoke of CMS. This installation phase started in June 2004. Currently (October 2005) the installation in the surface hall has been completed for the first two iron wheels (20% of the full detector). Once installed, the detectors go through a commissioning procedure to ensure its reliability. This contribution reviews the quality control, tests and certification procedures at CERN, the installation mechanism and the current status of the commissioning.

1. The CMS central muon detector 1.1. Drift Tube Chambers The CMS muon spectrometer1 is composed of three different detectors: two of them are tracking and trigger devices for the endcap (Cathode Strip Chambers) and barrel (Drift Tube chambers) regions, while the third one, the Resistive Plate Chambers (RPCs), are used for triggering purposes in both regions. The Drift Tube chambers (DTs) are gaseous detectors composed of two or three units called superlayers, each one of them containing four layers of staggered drift cells allowing a three-dimensional reconstruction of the muon tracks. A honeycomb panel is interleaved between superlayers in order to ensure mechanical stability. The DTs have self-trigger capability, bunch crossing identification and can resolve tracks with better than 200 microns accuracy per cell. The readout and local trigger electronics are housed by rectangular crates inserted in the honeycomb panel, the so-called minicrates. In total CMS will be equipped with 250 DT chambers.

621

622

1.2. The CMS DT collaboration Five different European institutes are responsible for the DT chambers construction, installation and maintenance, including mechanics and associated electronics. These are RWTH-Aachen, CIEMAT-Madrid, INFN-Padova, INFNBologna and INFN-Turin, although components production is shared as well between several other institutions in Russia, China, etc. Chamber assembly takes place in Aachen, Madrid, Legnaro and Turin. The performance of the detector was extensively tested 2 ' 3 before the construction phase started in 2001. These tests under muon beams as well as cosmic rays data tests at the sites confirmed the good behavior of the detector under the conditions of magnetic field, radiation and muon rates expected in CMS. 2. Chamber certification at the ISR area Once the chambers are constructed and tested at the sites4,5, they are sent to the experimental area ISR at CERN. There, the chambers are received, stored, fully equipped, tested and certified for installation in their final positions in the CMS iron yoke. Although chambers are mechanically finished at the construction sites, and the internal elements are certified, still a lot of work is required to complete them before installation. The external cabling (HV, LV distribution, RPC link, gas pressure measurements, signal routing...), piping (gas, cooling), insertion mechanisms, protections, alignment forks, grounding straps etc. are added to the chambers at the ISR area. This is the so called "dressing" process. Additionally, the minicrates and the RPCs are coupled together with the DT at the ISR just before installation.

623

Figure 1: View on lhe I.SK area ;il ( I KN \l the righl kind sidr m ilic pkiuii; 1)1 sinrage area can be seen. Left side shows CSC and RPC testing areas.

In order to validate the chambers and their external components for installation after they have been dressed, several procedures are carried out on them. These are the quality control steps followed to certify DT chambers: 1. 2.

3.

4.

Optical alignment: The relative positions between superlayers are measured with a photogrammetric system. Gas tests: The gas tightness of the chamber and the gas quality in its interior (in terms of air contamination) are checked. The gas purity is important in order to achieve high detection efficiency. HV behavior: The chambers are tested by supplying high voltage to all their electrostatic elements up to nominal values. The currents induced in them are constantly stored for a period one month minimum. The momentary increase of currents, discharges, sparks, etc. are detected as hint of problems inside the chamber. Cosmic rays tests: This is the main test block performed on the chambers. The whole functionality of the detector is tested by taking data with cosmic muons thanks to a test stand. The analysis of these data allows the recognition of problems, such as disconnected elements, noisy channels, inefficient cells, etc. In addition, the front end system of the chamber is tested, checking the integrity of the preamplifiers inside of the gaseous volume of the chamber. Features as test pulses, masking capability of single channels and groups, etc. are put under test.

624

5. Tests on external elements: Once the chamber is cosmic certified, it has to be equipped with the final cabling and electronics, which require their own certification tests. 3. Installation and commissioning The installation of barrel muon packages (DT+RPC) is done on the surface hall of the CMS experimental area (SX5). There, the chambers are inserted in the pockets of the return iron yoke of the magnet. Once chambers are installed and pipelined under gas, high voltage is supplied to them for at least one week in order to test HV stability after insertion. Single chamber commissioning takes place at this stage. It consists of several tests, including numerate performance, data taking with cosmic rays using several trigger configurations, etc. The purpose of these tests is to ensure the chamber functionality before the final cabling of the wheel, since the cable routing can difficult the access to certain chamber components.

4. Future prospects At the present (October 2005) most of the mechanical assembly of DT chambers is finished. The full production is expected to be completed at the beginning of next year. Progressively manpower and resources will be progressively moved from construction to installation and commissioning issues. The ISR area houses currently 126 chambers (out of 170 still to be installed). Chambers for the central wheel are under test, being prepared for further installation rounds. The activities in this area will continue toward the end of installation, preparing also a set of spare chambers for eventual replacements. The installation will resume at the beginning of next year, inserting chambers in the central wheel. This will be a rather challenging task since half of the chambers will have to be inserted under the magnet vacuum tank. All chambers, except for two sectors used as grips for the lowering crane, will be installed on surface before the end of 2006. The full installation will be completed underground in January 2007. The progress of the commissioning activities is satisfactory. A whole wheel has been successfully completed, whereas the second one is in progress. The

625

commissioned wheel is being cabled with the final cabling layout. Next step will be to commission a foil sector (4 chambers simultaneously). The purpose of this test will be to synchronize the data acquisition system, check the integrity of cabling and electronics and validate the regional trigger. Another important milestone for the detector will be the Magnet Test, when the superconducting solenoid will be first tested and raised to nominal field value on the surface hall. During this period the chambers will be functional and data from cosmic muons will be taken. This will allow an integrity test of the experiment as a whole. The muon system will be the main trigger signal for CMS during the test. With the current schedule we foresee the completion and ready-to-operate declaration of the barrel muon detector for the middle of 2007.

Figure 2: View on two of the five wheels of the iron return yoke for the CMS magnet. Almost all muon barrel packages can be seen already installed in their pockets (Picture taken in September 05).

References 1. CMS Collaboration, "The Muon Project TDR", CERN/LHCC 97-32, (1997), see also P. Paolucci's contribution, these proceedings. 2. M. Aguilar-Benitez et al, NIM A 416 (1998), 243. 3. M. Aguilar-Benitez et al, NIM A 480 (2002), 658. 4. M. C. Fouz, J. Puerta-Pelayo, CMS Note 2004/012. 5. E. Conti, et al, CMS Note 2003/017.

THE LHC COLLIDER: STATUS AND OUTLOOK TO OPERATION RUDIGER SCHMIDT CERN, Switzerland,

Geneva

For the LHC to provide particle physics with proton-proton collisions at the centre of mass energy of 14 TeV with a luminosity of 1034 cm"V, the machine will operate with high-field dipole magnets using NbTi superconductors cooled to below the lambda point of helium. In order to reach design performance, the LHC requires both, the use of existing technologies pushed to the limits as well as the application of novel technologies. The construction follows a decade of intensive R&D and technical validation of major collider sub-systems. This paper will focus on the required LHC performance, and on the implications on the used technologies. The consequences of the unprecedented quantity of energy stored in both magnets and beams will be discussed. A brief outlook to operation and its consequences for machine protection will be given.

1. Introduction The motivation to construct the Large Hadron Collider (LHC) at CERN comes from fundamental questions in particle physics. To get into the TeV scale, a proton collider with two beams of 7 TeV/c momentum is being constructed in the 27 km long LEP tunnel. A magnetic field of 8.33 Tesla is necessary to achieve the required deflection of 7 TeV/c protons that can only be generated with superconducting magnets. The machine is also designed for collisions of heavy ions (for example lead) at very high centre of mass energies. The LHC accelerator has been under preparation since the beginning of the eighties, with the first design of the machine parameters and the lattice, and the start of research and development programs for superconducting dipole magnets. The CERN Council approved the LHC between 1994 and 1996. The LHC accelerator is being constructed by CERN in collaboration with laboratories from both member and non-member states and beam operation shall start in 2007. Particle physics requires an unprecedented luminosity for the LHC in the o r d e r o f 10

34

? i

cm" s" . M o r e details o n the L H C are g i v e n in [1] a n d [2].

626

627 2. Basic LHC Accelerator Physics Already from early LEP days it has been foreseen to install the future LHC in the existing LEP tunnel. The tunnel determines the bending radius. An increase of the energy of a charged particle is with electrical fields, while magnetic fields are used to deflect the particles. The energy gain in an electrical field is proportional to the voltage. For the acceleration to 7 TeV/c a voltage of 7 TV is required. Radio-frequency (RF) acceleration is used in the LHC as for all high energy accelerators. In an RF cavity a time-varying electrical field accelerates charged particles that must enter the cavities at the correct phase. For continuous operation with superconducting RF cavities, the electrical field is in the order of 10 MV/m. In circular machines, particles pass many times through the cavity and are accelerated during each passage over many turns, for the LHC 7

in about 10 turns. Dipole magnets keep the particles during the acceleration on an (approximately) circular path. Deflection of charged particles in a magnetic field is determined by the Lorentz force. Assuming that a particle moves on a circle with radius p in a homogenous magnetic field perpendicular to the velocity, the Lorentz force must be kept equal to the centrifugal force. The magnetic field for the momentum p can be calculated: B = p / e-p. The strength of the magnetic field is therefore determined by the LHC radius and by the particle energy. At injection for 450 GeV/c the magnetic field is about 0.54 T. During acceleration, the field in the dipole magnets is increased to 8.33 T for the maximum momentum of 7 TeV/c. Such high field can only be achieved with superconducting magnets. NbTi is the optimum conductor material for constructing a large number of superconducting magnet coils. To achieve a field of 8.33 Tesla with NbTi magnets requires cooling to 1.9 K with superfluid Helium. 3. How to get to high luminosity? When two bunches cross in the centre of a physics detector, only a tiny fraction of the particles collide to produce the wanted events, since the cross-section to produce a particle such as the Higgs Boson is very small. The rate of particles N/At created in the collisions is given by the product of cross-section and luminosity: N/At = La. To exploit the LHC energy range, a luminosity in the 34

2 l

order of 10 cm" s" is required, which exceeds the luminosity of today's proton colliders by more than one order of magnitude. The luminosity L increases with the number of protons N in each bunch, with the number of bunches nb, with the

628

revolution frequency f, and decreases with the beam size CTX and a y at the collision points: N 2 fnb 4-Ti-
Figure 1. Energy stored in LHC beams and magnets compared with other accelerators

insertions with experiments, the beams collide at an angle, in order to separate the beams as efficiently as possible. In the arcs and in the insertions without experiments, the beams are guided in two separate vacuum chambers. The most relevant parameters for the LHC are summarised in Table 1. The energy of 10 GJ stored in the magnets is very high. To achieve the required

629 luminosity, beams with high currents need to be accelerated to 7 TeV. The energy stored in each beam is 362 MJ. For comparison, only 500 kJ are sufficient to heat and melt 1 kg of copper. Figure 1 compares the energy stored in LHC beam and magnets with the beam energy for other machines. A small fraction of the beam energy released into equipment could lead to damage, and a tiny fraction could quench a superconducting magnet. Table 1. Most relevant LHC Parameters Energy at collision Energy at injection Circumference Dipole field at 7 TeV Luminosity Protons per bunch Number of bunches / beam Nominal bunch spacing Beam size at IP / 7 TeV Total crossing angle Typical beam size in arcs (rms) / 7 TeV Arc magnet coil inner diameter Distance between beams (arc) Stored energy per beam / 7 TeV Stored energy in magnets / 7 TeV

7 0.45 26658 8.33 1034 1.1510" 2808 25 15.9 300 300 56 194 362 10

TeV TeV m T -2 -1

cm s ns um urad p.m mm mm MJ GJ

4. LHC Overview Beams for the LHC machine are prepared and accelerated by existing particle sources and pre-accelerators [3], The injector complex includes many accelerators at CERN (see Figure 2): Linacs, Booster, LEIR (Low Energy Ion Ring) as an ion accumulator, CPS and the SPS. The beams will be injected from the SPS into the LHC at 450 GeV/c, accelerated to 7 TeV/c in about 25 min, and then collide for many hours. The pre-accelerators are operational and the modifications required to achieve the LHC beam parameters are essentially finished. During autumn 2004 beam has been extracted from the SPS and transported through TI8 to a beam dump at the end of this 2.5 km long LHC transfer line. The LHC has an eight-fold symmetry with eight arc sections, and eight straight sections with experiments and systems for machine operation (Figure 3). Two counter-rotating proton beams will circulate in separate beam pipes installed in the twin-aperture magnets and cross over at four points. There will be four experimental insertions, for the main experiments ATLAS, CMS, ALICE and LHCb and one small experiment, TOTEM. The injection elements are

630

installed in the insertions for ALICE and LHCb. The other insertions are dedicated to machine operation, two for beam cleaning, one for the beam dumping system, and one for RF and beam instrumentation.

Figure 2. CERN accelerator complex

The particle transport through the arcs requires dipole magnets for deflection, quadrupole magnets for focusing, and magnets with a magnetic field of higher order for beam stability. The major systems installed are the high field superconducting magnets operating at 1.9 K and the large cryogenics system including the supply and recovery of helium with a 26 km long cryogenic distribution line [4]. Four vacuum systems are required: one for each beam, one insulation vacuum system for the magnets, and one insulation vacuum for the cryogenic distribution line. Along the arc, several thousand crates with radiation tolerant electronic are required, for quench protection, power converters and instrumentation. At the end of the long arc cryostats and at the end of other cryostats, current feed-throughs are installed in distribution feed-boxes (DFB). Feeding current from ambient temperature into the magnets operating at 1.9 K includes the industrial use of High Temperature Superconducting material [5]. For the regular arc, 1232 main dipole and 392 main quadrupoles are required [6]. About 150 additional main superconducting dipoles and quadrupoles will be installed in the long straight sections. Further, several thousand smaller superconducting magnets are required for beam steering, chromaticity compensation and correction of multipole field errors. About 140 normal conducting magnets will be installed in the LHC ring, mainly in the

631 cleaning insertions. More than 600 normal conducting magnets are required for the SPS-LHC transfer lines, the majority is already installed. The status of the equipment production is updated on dashboards of the LHC homepage [7]. Niobium-titanium cable production for the dipole magnets is close to be finished. More than 70 % of all dipole magnets have been delivered to CERN. For other magnets, the advancement in the production is similar. The installation of the magnets started, more than 100 dipole magnets have been installed in the tunnel.

Figure 3. LHC layout

5. Operation and machine protection Operation of the LHC will be strongly constrained due to the risks from the large stored energy in the magnet system and in the beams. Safe operation of the magnet and powering systems, cryogenics and vacuum is already required for the commissioning of the LHC hardware systems starting in 2005. Beam operation is expected to start in 2007. 5.1. LHC cycle The beams are accelerated in the SPS from 26 GeV/c to 450 GeV/c, and then transferred to the LHC. During the injection phase, 12 batches per beam (one batch has either 216 or 288 bunches) from the SPS are injected into the LHC (see Figure 4). Injection of the two beams will take about 10 min. Then the field of the LHC dipole magnets is ramped within 25 min to 8.33 T corresponding to a momentum of 7 TeV/c. Normally, the beams will collide for several hours

632

(physics fill). At the end of the fill, the beams will be dumped and the magnets will ramp down to prepare for the next injection.

coast

beam dump

energy ramp

12000

7TeV

10000 8000 6000 4000

injection phase 12 batches from the SPS (every 20 sec' one batch 216 / 288 bunches

2000 460 GeV -4000

-2000

0

2000

4000

time from start of injection (s)

Figure 4. LHC cycle

The luminosity lifetime is estimated to be approximately 10 h (after 10 hours only 1/3 of initial luminosity). At that time, the beam current somewhat reduced - but not much. The energy stored in each beam is still about 200-300 MJ. The beams are extracted into beam dump blocks being the only components that can stand a loss of the full beam without damage. At 7 TeV, fast beam loss with an intensity of about 5% of one single "nominal bunch" could damage superconducting coils. 5.2. Beam losses into material Beam impact in material produces particle cascades. The temperature increases with the energy deposition that depends mainly on the material and on the number and energy of the particles. Already small beam losses into superconducting magnets could lead to a quench. Quench recovery to reestablish the conditions for beam operation will take several hours. For large beam impact the material could be damaged, leading to long and costly repairs. For example, the exchange of one superconducting magnet would take about 5 weeks. In order to verify the damage calculations, an experiment has been performed with the 450 GeV SPS extracted proton beam with intensities between 210 1 2 and 8-1012 (0.1 % of the beam energy at LHC for 7 TeV). The beam was deflected into a 30 cm long target consisting of a stack of metal plates.

633

The expected beam induced damage level from simulation programs was compared with the results from an experiment [8]. The result for a copper plate is shown in Figure 5: the beam with maximum intensity clearly damaged the plate. Damage is also visible for an intensity of 6-1012, for lower intensity no damage was observed. The good agreement between simulations and experimental results suggest that calculations of the damage level are realistic.

Figure 5. Damage of a copper plate from the SPS proton beam extracted towards a target, with different beam intensities

5.3. Beam dumping system The only element in the LHC that can absorb the full energy of the 7 TeV beam without being damaged is the beam dump block. At the end of a physics fill or in case of failure the beams are extracted into this block: Fast kicker magnets deflect the beam by an angle of 260 urad. The extracted beam is further deflected by septum magnets towards the beam dump block. The average beam size at 7 TeV is about a = 0.2 mm. The beam dump block has a graphite core and a concrete shielding. Even this block could not stand such a small beam. The beam size increases in the 700 m long extraction line. In addition, the beam is blown up by two additional sets of kicker magnets deflecting the beam in horizontal and vertical direction [9], For nominal beam parameters, the maximum temperature in the beam dump block is expected to be in the order of 700 °C.

634

5.4. Beam cleaning To avoid quenching a magnet, beam losses must be kept below the quench level. At injection energy, the magnet would quench after an instantaneous loss of about 5-109 protons, at top energy of only about 5 -106 protons. Particles must be captured by collimators in the cleaning insertions to limit particle losses and the associated heat load into the superconducting magnets. Below a lifetime of about 12 min, the heat load on the collimators would become unacceptable and the beams must be dumped. Collimators in the beam cleaning system are blocks of solid materials that should capture all particles with large amplitudes, to prevent particles losses around the accelerator, in particular in the superconducting magnets. The LHC will be the first accelerator requiring collimators to define the mechanical aperture throughout the entire cycle. For efficient beam cleaning, the collimators are adjusted to a position of 5-9 a from the beam axis. For operating at 7 TeV with nominal beam parameters, more than 99.98 % of the protons in the beam halo should be captured in the cleaning insertions to minimise protons impacting on the cold magnets [10]. For the first year(s) of operation, the parameters are relaxed due to operation below nominal intensity. In case of equipment failures, collimators will be the first devices to intercept the beam and must absorb part of the energy until the beams are extracted. For the LHC start-up, all collimators will have graphite jaws. Graphite collimators are very robust, and in case of beam impact not easily damaged. 6. Outlook to operation The experiments ATLAS and CMS expect proton collisions at the highest energy (7 TeV) with the nominal luminosity 10 cm'V 1 . They will also take data when the LHC is operating to provide ion collisions [11]. LHCb is conceived for B physics with a luminosity of ~5T032 cm'V 1 . ALICE is an experiment to observe ion collisions at various energies. The nominal luminosity is 1027 cm'V 1 . ALICE will also use proton collisions, with a luminosity of 1030 cm'V 1 . TOTEM will use proton collisions at various energies. If the Higgs particle as predicted by the Standard Model decays into four leptons and its mass is in the range of some hundreds GeV, it can be discovered with an integrated luminosity ~4 fb"1. Some super-symmetric particles can be discovered at more modest luminosities with an integrated luminosity of ~1 fb"1. Potential for b-physics will be there soon after the startup. Operating the LHC for 300 hours with an average luminosity of 1034 cm'V 1 will result in an integrated luminosity of 1 fb'1.

635

The experiments will make use of any beam for detector commissioning. In order to minimize event pileup they desire to operate early on with many bunches and go to 25 ns bunch distance as soon as possible. LHC operation for physics will start with a pilot physics run with few bunches. There will be no crossing angle, the bunch current will be reduced, and operation is with 43 bunches. The performance will be pushed, filling more bunches, reducing the beam size in the interaction points, and increasing the bunch current. The next phase is with bunches 75 ns apart. The parameters such as squeeze and crossing angel are still relaxed. The intensity will be slowly pushed. The third phase is with bunches 25 ns apart. This is the limit for the phase I collimator system, to increase the bunch intensity requires the second phase collimators. Beam scrubbing is required. The last phase is the operation with 25 ns bunch distance, at nominal performance. 7. Conclusions Construction of the Large Hadron Collider is well advanced. Installation of the LHC components into the former LEP tunnel is ongoing. Services like electricity, ventilation etc. have been installed. Installation of the cryogenic distribution line is under way. Magnet installation has started. Before the first protons will be injected into the LHC, many of the technical systems will have been commissioned: cryogenics, vacuum, powering, quench protection and interlocks. Commissioning of individual systems has started. A period of more than one year is foreseen for commissioning the powering of the superconducting magnets up to nominal current. It is planned to finish machine installation and hardware commissioning and to start operation with beam in 2007. After a meeting of the external Machine Advisory Committee in 2002 the chairman concluded: "The LHC is a global project with the world-wide highenergy physics community devoted to its progress and results. As a project, it is much more complex and diversified than the SPS or LEP or any other large accelerator project constructed to date". The Machine Advisory Committee, chaired by Prof. M. Tigner, June 2005, concluded for the machine protection systems: "We recognize that the planned schedule is very aggressive, given the complexity and potential for damage involved in the initial phases of operation. It will be important to understand the performance of the machine protection system, the collimation system and the orbit feedback system as well as cycle repeatability and adequate beta-beat

636

control before proceeding to run with significant stored beam energy. Pressure to take shortcuts must be resisted." 8. Acknowledgements The material presented in this paper reflects a tremendous amount of excellent work by many colleagues, inside and outside CERN. In addition, the help of those who helped writing this report is gratefully acknowledged. I also appreciate my colleagues for providing me with a number of figures and for their comments. References 1.

2. 3. 4. 5. 6.

7. 8. 9. 10. 11.

R. Schmidt, The LHC accelerator and its Challenges, in: LHC Phenomenology, Institute of Physics Publishing, edited by M.Kramer and J.P.Soler, (2004). LHC Design Report, Vol. 1, The LHC Main Ring, CERN-2004-003-V1, Geneva, CERN (2004). LHC Design Report, Vol.2, The LHC Injector Chain CERN-2004-003-V3, Geneva, CERN (2004). Ph.Lebrun, Large Cryogenic Helium Refrigeration System for the LHC, CERN-LHC-Project-Report-629, Geneva, CERN (2003). A.Ballarino, HTS Current Leads for the LHC Magnet Powering System, CERN-LHC-Project-Report-608, Geneva, CERN (2002). L.Rossi, Experience with LHC Magnets from Prototyping to Large-Scale Industrial Production and Integration, EPAC 2004, Lucerne, Switzerland, July 2004. LHC Status Dashboards: http://lhc-new-homepage.web.cem.ch/lhc-new-homq5age/DashBoard/index.asp V.Kain et al., Material Damage Test with 450 GeV LHC-Type Beam, PAC 2005, Knoxville, Tennessee (2005). R.Filippini et al., Reliability Issues of the LHC Beam Dumping System, EPAC 2004, Lucerne, Switzerland (2004). R.Assmann et al., LHC Collimation: Design and Results from Prototyping and Beam Tests, PAC 2005, Knoxville, Tennessee (2005). D.Macina, Desires and constraints of experiments during early LHC operation, 2nd LHC Project Workshop, Chamonix 17-21 January 2005 (2005).

PANDA: A DETECTOR FOR RESEARCH WITH ANTIPROTONS JERZY SMYRSKI Institute of Physics, Jagellonian University ul. Reymonta 4, 30-059 Cracow, Poland PANDA is a 4% internal detector planned for experiments at the antiproton storage ring HESR at the future international accelerator facility FAIR in GSI-Darmstadt. The detector is designed to take advantage of the extraordinary physics potential, e.g. in the field of charmonium spectroscopy or of exotic mesons, which will be available utilizing high intensity, phase space cooled antiproton beams with momentum in the range of 1.5 to 15 GeV/c. As a target, frozen H2 pellets or cluster jet for the pbar-p reactions and a thin wire for the pbar-A reactions will be applied. For a full angular coverage and a wide range of energies the detector is subdivided into a target spectrometer based on a solenoid surrounding the interaction region and a forward spectrometer using a dipole magnet for momentum analysis of the forward-going particles. In this contribution an overview of the design is presented.

1. Introduction PANDA (Antiproton Annihilations at Darmstadt) is a general purpose detector which was proposed for studies of reactions induced by antiproton beams on hydrogen as well as on nuclear targets at the future accelerator facility at GSIDarmstadt. This new facility called FAIR (Facility for Antiproton and Ion Research) will offer high intensity beams of heavy ions and antiprotons for a wide range of studies in the field of nuclear and particle physics, as well as atomic physics and plasma physics. PANDA will be a detector system at the internal target of the High Energy Storage Ring (HESR) at FAIR. HESR can store up to 10" antiprotons and accelerate them in the momentum range 1.515 GeV/c. Application of electron cooling will make it possible to perform experiments with antiproton beam of unprecedented quality and, in particular, to conduct high precision measurements in the field of: • Charmonium spectroscopy: precision measurement of mass, width and decay branches of all charmonium states. • Establishment of the QCD-predicted qluonic excitations (charmed hybrids, glueballs) in the charmonium mass range (3-5 GeV/c 2 ). • Search for modifications of properties of mesons with open and hiddencharm in nuclear medium. 637

638

• Precision y-ray spectroscopy of single and double hypernuclei. For the envisaged experimental program a nearly full coverage of the solid angle together with good particle identification and high energy and angular resolutions for charged particles and photons are mandatory.

Fig. 1 Schematic drawing of the PANDA detector seen from the top. The general layout of PANDA is based on two spectrometers and is schematically shown in Fig. 1. The target spectrometer surrounds the interaction region and has a 2 T superconducting solenoid as momentum analyzer. The most forward scattering angles - ±10° horizontally and ±5° vertically - are covered by the forward spectrometer using a 1 m gap dipole magnet for momentum analysis of the forward-going particles. The target and the basic detector components of PANDA including tracking system, particle identification detectors and calorimeters are described in subsequent chapters. 2. Target To achieve the design luminosity of 2-1032cm"Y1 with a number of 10 n antiprotons and a HESR circumference of 574 m an effective target thickness of 3.8-1015 hydrogen atoms per cm2 is required. To reach this density a pellet target is a very promising option. It consists of a stream of frozen hydrogen microdroplets, called pellets, traversing the antiproton beam perpendicularly. The design will be based on the pellet target built for the WASA detector [1]. At the

639 interaction point, typical parameters comprise a pellet rate of ~10 kHz, a pellet speed of ~60 m/s and a pellet size of 25-35 um. The second option considered for the hydrogen target is a cluster jet target which sends an ultra-dense hydrogen jet through the beam. It provides a homogeneous stream of gas which is easy to handle from the accelerator point of view. Currently, the R&D work is concentrated on improving the densities such that the design luminosity can be reached. A cluster jet target or a pellet target can also by used for reactions with nuclei. However, higher luminosities and a better definition of the interaction point will be provided by using wire or strip targets. 3. Tracking For measuring displaced vertices of charmed mesons a micro vertex detector will be used. It will consist of five barrel shaped layers surrounding the interaction point and four disk-shaped parts in forward direction. The three innermost barrel layers are composed of pixel detectors to achieve best resolution and to be able to easily detect decay vertices displaced from the interaction point. The outer layers are composed of microstrip detectors which are easier to handle. The baseline technology chosen for the pixel detectors are hybrid active pixel sensors used by several LHC experiments (ALICE, ATLAS, CMS). The electronics still has to be modified due to planned continuous readout. For tracking of charged particles in the field of the PANDA solenoid, either a straw tube tracker (STT) or a time projection chamber (TPC) will be used. The straw tube tracker consists of self supporting straw tubes arranged in 11 cylindrical double layers occupying a radial distance of 15 cm from the beam line up to 42 cm. The track position along the beam direction will be determined either by application of the stereo layers or by using the charge division technique. The expected momentum resolution dp/p is about 1%. The TPC is a very attractive alternative to the STT due to the smaller material budget and an additional option of particle identification via dE/dx. However, the TPC is the technically more challenging solution since due to the high rate of proton-antiproton annihilations of 2T0 7 s"1 expected at PANDA it has to operate continuously, i.e. the technique of gating can not be applied. Apart of the obvious problem of ions feeding back into the drift volume, continuous tracking calls for about an order of magnitude higher granularity in all three spatial coordinates than the previous TPCs in order to disentangle hits belonging to different tracks. These problems may by overcome by application of the Gas Electron Multiplier (GEM) [2] instead of typically used MWPCs for measurement of the track projection on the TPC endplate.

640

Particles emitted at angles below 22° which are not covered fully by the Straw Tube Tracker will be traced with two planar drift chambers placed 1.4 m and 2.0 m downstream of the target. The chambers have to stand a high counting rate of particles peaked at the most forward angles due to the relativistic boost as well as due to the small angle pbar-p elastic scattering. With the envisaged luminosity the expected particle flux in the first chamber in the vicinity of the 5 cm diameter beam pipe is about 2-104cm"2s"1. Besides, the chambers have to work in the 2 T magnetic field produced by the PANDA solenoid. In order to fulfill these requirements quadratic drift cells with an active area of 1 cm2 were chosen. Each chamber will contain four pairs of detection planes with wires mounted on frames of octagonal shape. We also explore a possibility of using frames having a form of closed semi-circles. The detection planes would consist of pairs of such frames put together along the common diameter. The frames could be mounted and dismounted without necessity of removing the beam-pipe. Drift chambers with 1 cm2 cells are also foreseen for tracing particle trajectories in the Forward Spectrometer. 4. Charged Particle Identification Particle identification (PID) for charged hadrons (p, n1, K'1') will be accomplished using a combination of Cherenkov type detectors, covering higher momenta, and of time-of-flight counters foreseen for lower momenta. In the barrel section of the target spectrometer, the particle identification will be based on the Detection of Internally Reflected Cherenkov (DIRC) light as realized in the BaBar experiment [3]. DIRC detector uses radiator bars with a rectangular cross section made of artificially fused quartz. Since the refraction index of the radiator is large (n = 1.47), some of the Cherenkov photons are internally reflected, regardless of the incidence angle of the tracks and propagate along the length of the bar. To avoid having to instrument both bar ends with photon detectors, a mirror is placed at one end, perpendicular to the bar axis. The photons leave the bar at the opposite end. They are then proximity focused by expanding through a region filled with purified water on an array of 7000 PMTs. The Cherenkov angle is determined through the Cherenkov ring registered with the PMTs. To reduce the number of PMTs one considers replacement of the two dimensional photon detector by a one dimensional readout where one dimension is replaced by the timing information of the detector. A time reference will come from a scintillator barrel which will be placed inside the DIRC barrel. The scintilator barrel will be also used for PID at low particle momenta on the basis of the time of flight information.

641 Another Cherenkov detector will be located in the forward end cap of the target spectrometer. In the relevant momentum range the detector can be either a disc DIRC detector with quartz as radiator or a proximity imaging C6Fi4/CsI RICH. In the forward spectrometer, the charged particle identification will be based on the time of flight information on a 7 m path between the target and a wall made of plastic scintillators. The option of an additional RICH detector with an aerogel radiator, extending the PID towards higher momenta, is under investigation. The identification of muons will be done using a hodoscope made of plastic scintillator strips placed behind the iron yoke of the target spectrometer. The momentum of the muons will be determined by the target spectrometer tracking. The scintillator strips will be read out on both ends with photomultipliers fed to ADCs and TDCs. This information can be used to determine the position for matching with the target spectrometer tracking and the time of flight, to suppress random coincidences. 5. Calorimeters Expected high count rate and a geometrically compact design of the target spectrometer require a fast scintillator material with a short radiation length and Moliere radius for the construction of the electromagnetic calorimeter. To meet these requirements, as a detector material PbW0 4 is foreseen, since it is fast and has a reasonable resolution. Nevertheless, the use of BGO is also considered due to higher light yield albeit slower signals. The electromagnetic calorimeter in the target spectrometer is placed outside the DIRC. It consists of a barrel part with 11360 crystals, and two end caps comprising additional 7680 modules. The crystals will have a form of tapered prisms with a length of approximately 20 X0, a front size of 2.2 x 2.2 cm2 and they will be mounted with an inner radius of 50 cm. The calorimeter will be cooled down to -25° C in order to increase the light yield of PbW0 4 . The readout of crystals will be accomplished by large area photo diodes with optimized light collection, a tolerable nuclear counter effect, and a fast timing behavior. As a calorimeter for the forward region, a system composed of two parts is considered. The first part will be a Shashlyk-type calorimeter, suited for detection of photons and electrons with high resolution and efficiency. The second calorimeter part is devoted to the energy measurement of neutral hadrons, in particular neutrons and antineutrons, and serves as a fast trigger and an active muon filter.

642

6. Trigger and Data Acquisition PANDA has a rich physics program with many different topics based on varying physical selection criteria. Due to a very high expected raw data rate up to 200 GB/s a very efficient and flexible triggering system is required. A conventional approach based on a hierarchical trigger and the hard-wire detector connectivity severely constrains both complexity and the flexibility of the possible trigger schemes. Therefore a concept of a continuously sampling data acquisition in which event selection takes place in programmable processing units is planned for PANDA. In this approach all detector channels are self triggering entities. They automatically detect signals and pre-process them to extract and transmit only the physically relevant information. The data related to a particle hit substantially reduced in the pre-processing step are marked by a precise timestamp and buffered for further processing. The trigger selection finally occurs in so called compute nodes which access the buffers via a high bandwidth network fabric. The new concept provides a high degree of flexibility in the choice of trigger algorithms. It makes trigger conditions available which definitely are outside the capabilities of the standard approach, an obvious example being displaced vertex triggering. Acknowledgments This work is part of the EU Integrated Infrastructure Initiative Hadron Physics Project (JRA1, JRA4). It is also part of the EU Design Study - "DIRACsecondary-beams" (PANDA3, PANDA4). References 1. 2. 3.

C. Ekstrom et al., Phys. Scripta T99, 169 (2002). F. Sauli, Nucl. Inst. Meth. A386, 531 (1997). R. Aleksan et al., Nucl. Inst. Meth. A397, 261 (1997).

T E S T OF QED W I T H T H E R E A C T I O N e + e ~ -> 7 7 ( 7 )

U. BURCH, A. RUBBIA, J. ULBRICHT, * Swiss Institute of Technology, ETH Zurich, CH-8093 Zurich,

Switzerland

A.S. SAKHAROV, Theory Division, Physics Department, CERN, 1211 Geneva 23, Switzerland and Swiss Institute of Technology, ETH Zurich, CH-8093 Zurich, Switzerland C H . LIN, National Space Organization, HsinChu, TAIWAN

300

J. ZHAO, Laboratory of research into the fundamental laws of the Universe, CEA - Saclay, F-91191 Gif sur Yvette Cedex, France I. DYMNIKOVA Institute of Mathematics and Informatics, UWM in Olsztyn, PL-10-561 Poland

Olsztyn,

We search for a non point-like behavior of fundamental particles, which could be caused by hypothesis of an exited electron. In particular we focus on the measurements of the differential cross sections for QED process e + e - —• 77(7) made at center-of-mass energies from 51.8 GeV to 209 GeV by LEP and TRISTAN. The global fit we performed indicates about 5<7 deviation from the standard QED expectations when the mass of the exited electron approaches me* = 308±56 GeV.

1. I n t r o d u c t i o n In the case of electromagnetic interaction the process e + e _ -» 7 7 ( 7 ) is ideal t o test the QED because it is not interfered by the Z° decay. This reaction proceeds via the exchange of a virtual electron in t h e t - and u channels, while the s - channel is forbidden due to the angular m o m e n t u m 'Corresponding author, E-mail: [email protected]

643

644

conservation. The differential cross-section is well known from QED, so that any deviation from this expectation hints at beyond standard model processes contributing to the photonic final states. We use a comparison of the measured photon angular distribution with the QED expectation to test the QED cut-off parameters 1 . The obtained limit on the QED cut-off parameters put constraints on the size of the interaction area, which can be connected to the size of selfgravitating particlelike structure with de Sitter vacuum core 2 . The other application of our results is related to the study of low energy effects of large extra dimensions 3 in the string theory framework 4 . 2. Modification of QED photon distribution A possible modification of QED cross section can be described in terms of Drell's parametrization which is actually applicable to any modification of standard model at short distance. We use here the model of excited electron 1 where the heavy electron of mass me* could couple to an electron and a photon via magnetic interaction with an effective Lagrangian ^excited = ~

Ipe'^iiV^eF^"

(1)

In this equation A is the coupling constant, F^" the electromagnetic field tensor, ipe. and ipe are the wave function of the heavy electron and the electron respectively. The model has A and me» as parameters. It is possible to write down the corresponding differential cross-section by adding a deviation term <Snew to the QED differential cross-section including radiative effects up to 0(a3). The modified equation is {dtr/d&l) = (dff/dn) 0 ( a a)(l + ( W ) If the center-of-mass energy y/s satisfies the condition sjm\, <^new would read as <5new = ±s 2 /2(l/A 4 ± )(l - cos2 0 )

(2) «

1, then

(3)

In this Drell's approximation, the parameters A± are the QED cut-off parameters with A+ = m 2 , / A . In the case of arbitrary
(4)

645

B -o

ScaleIo91.2GeV 0.1

0.2

03

0.4

0.5

0.6

0.7

0.8

0.9

1

0

0.1

02

03

0.4

|COS(0)|

0.3

0.4

03

|cos(e)i

0.5

0.6

0.7

0.8

0.9

0.6

0.7

0.8

0.9

|COS(0)|

0.3

0.4

0.5 |C0S(e)|

Figure 1. The measurements of the differential cross section from VENUS, TOPAZ, ALEPH, DELPHI, OPAL, L3 and an overlay of all available data together. For comparison with theory QED calculations of the differential cross section up to radiative effects up to O(o3) is displayed in a solid line. All values are normalized to 91.2 GeV.

1

646

ALEPH

Delphi

» *

.K

L3

J* OPAL

TRISTAN LEP1 LEP2 ALL

I 1 1 1 • I t 1 1 T9\ 1 1 1 1 I 1 1 i 1 1

-

2

-

1

0

1

LX10- 9

2

A+

Figure 2.

The best fit values of T r - [ G e V - 4 ] using different subsets of the available

data.

The third order QED differential cross section up to 0(az) is calculated by numerically generating a high number of Mont Carlo (MC) e + e~ -> 77(7) events. 3. Overall fit We performed the global analysis of measured differential cross sections for the process e + e~ —> 77(7). The data has been taken in the energy range from v/i=55 GeV to Vs=207 GeV with the VENUS 5 , TOPAZ 6 , ALEPH 7 , DELPHI 8 , L3 9 and OPAL 10 detectors during the period from 1989 to 2003. The data are summarized in Fig. 1. We applied the x 2 test to the whole available data set using j t as the fit parameter. We also analyzed different subsets of data belonging to different experiments and years Fig. 2. Systematic errors could arise from the luminosity evaluation, from the

647

selection efficiency, background evaluations, the choice to use the Born level or a3 theoretical QED cross section as reference cross section, the choice of the fit procedure, the choice of the fit parameter and the choice of the scattering angle |cos©| in particular in comparison between data and theoretical calculation. The maximum estimated error for the value of the fit from the luminosity, selection efficiency and background evaluations is approximately <5A/A = 0.01. The choice of the theoretical QED cross section was studied with 1882 e + e~ —> 77(7) events from the L3 detector. 4. Discussion The hypothesis used in equation (1) and (2) assumes that an exited electron will increase the total QED- a3 cross section and change the angular distribution. Opposite to the hypothesis the best fit value of (1/A 4 ) b e s t = -(1.11 ± 0.20)10- 10 [GeV]- 4 is negative with significance about 5 a. The later indicates that our data set prefers to decrease the cross section of e + e~ —> 77(7) with respective to that predicted by pure QED. Using the best value of A4 one can calculate the mass of an exited electron, which amounts me* = 308 ± 56 GeV. To add is that any exceptional claim requires exceptional evidence, so for the future refinement of the analysis we must scrutinize the signal of a deviation from QED predictions very carefull against systematics inherent to all experiments the data of which have been used in the overall fit. References 1. A. Litke, Harvard Univ., Ph.D Thesis, unpublished (1970). 2. Dymnikova, I., Ulbricht, J., and Zhao, J., hep-ph/0109138. 3. Arkani-Hamed, N., Dimopoulos, S., and Dvali, G., Phys. Lett. B429, 263 (1998); Antoniadis, I., Arkani-Hamed, N., Dimopoulos, S., and Dvali, G., Phys. Lett. 436, 257 (1998). 4. Cullen, S., Perelstein, M., and Peskin, M. E., Phys. Rev. D62, 055012 (2000). 5. The VENUS Collaboration K. Abe et at, Z.Phys.C45 175 ( 1989 ) 6. The TOPAZ Collaboration K. Shimozawa et al., Phys.Lett.B284 144 ( 1992 ) 7. The ALEPH Collaboration D. Decamp et al. , Phys.Rept.216 253 ( 1992 ) 8. The DELPHI Collaboration P. Abreu et al. Phys.Lett.B327 386 ( 1994 ). The DELPHI Collaboration P. Abreu et al.. Phys.Lett.B433 429 ( 1998 ). The DELPHI Collaboration P. Abreu et al.. Phys.Lett.B491 67 ( 2000 ) 9. The L3 Collaboration P. Achard et al. , Phys.Lett.B531 28 ( 2002 ) 10. The OPAL Collaboration M.Z. Akrawy et al, Phys.Lett.B275 531 ( 1991 ). The OPAL Collaboration G. Abbiendi et al, Eur.Phys.J.C26 331 ( 2003 )

MUON RECONSTRUCTION AND IDENTIFICATION FOR THE EVENT FILTER OF THE ATLAS EXPERIMENT

A. V E N T U R A * on behalf of t h e A T L A S T D A Q - H L T g r o u p t The ATLAS Trigger requires high efficiency and selectivity in order to keep the full physics potential of the experiment and to reject uninteresting processes from the 40 MHz event production rate of the LHC. These goals are achieved with a trigger composed of three sequential levels of increasing accuracy that have to reduce the output event rate down to ~100 Hz . This work focuses on muon reconstruction and identification for the third level (Event Filter), for which specific algorithms from the off-line environment have been adapted to work in the trigger framework. Two different strategies for accessing data (wrapped and seeded modes) are described and their reconstruction potential is then shown in terms of efficiencies, resolutions and fake muon rejection power.

f

1

Contact author: Andrea.Ventura@!e.infn.it — INFN Sezione di Lecce, via per Arnesano, 73100 Lecce, Italy. a b c d e / a

A. dos Anjos , S. Armstrong , J. T. M. Bainea , C. P. Bee , M. Bellomo , M. Biglietti , J. A. Bogaerts h f i h 3 k l a M. Bosman , G. Carlino , B. Caron , P. Casado , G. Cataldi , D. Cavalli , G. Comune , P. Conde f fn b n o p a F. Conventi , G. Crone , D. Damazio , A. De Santo , M. Diaz Gomez , A. Di Mattia , N. Ellis c q p h. n q n D. Emeliyanov , B. Epp , S. Falciano , H. Garitaonandia , S. George , V. Ghete , S. Goncalo E. Gorini , J. Haller , S. Kabana , A. Khomich , G. Kilvington , N. Kirk , N. Konstantinidis t u n p b v d tit A. Kootz , A. J. Lankford , A. Lowe , L. Luminari , T. Maeno , J. Masik , C. Meessen , A, G. Mello R. Moore , P. Morettini , A. Negri , N. Nikitin" , A. Nisati P , C. Osuna , C. Padilla 9 , N. Panikashvili* X p r i g j d F. Parodi , E. Pasqualucci , V. Perez Reale , J. L. Pinfold , P. Pinto , M. Primavera , Z. Qian k g h g e x h S. Resconi , S. Rosati , C. Sanchez , C. Santamarina , D. A. Scannicchio , C. Schiavi , E. Segura y

IL>

J. M . d e Seixaa m

, S. Sivoklokov a.a

M. Sutton

, S. Tapprogge

A. Ventura

, V. Vercesi

M. Wielers

, G. Zobernig

Laboratory

(BNL),

, S. Tarem

, T. Wengler a

.

Upton, New York, USA,

Pavia, Italy, f Dipartimento

Geneva, Switzerland, of Alberta,

Edmonton,

Fisica deU'Universitk and Astronomy,

" Institut

c

, E. Thomas

d'Aix-Marseille

College London, London,

fiir Experimentalphysik

Wuppertal,

s

u

Republic, Prague, Czech Republic, deU'Universitk

n

UK, p

fiir Informatik

Haifa, Tsrae/,

Pisa, e I.N.F.N., Pisa, Italy,

CC

Dipartimento

, B. Venda Pinto

d

Department

Dipartimento Universitat,

, W. Wiedenmann

USA, " Brookhaven

Centre de Physique

"Federico II" e I.N.F.N., de Barcelona,

Barcelona,

rn

t

Spain,

di Pav'm

University

Dipartimento

Department

di

of Physics

of Physics, Royal Holloway, University of London, Egh&m, UK, di Fisica deU'Universitk Innsbruck,

Mannheim, v

di Roma 'La Sapienza' e I.N.F.N.,

Rome,

Austria, l a b o r a t o r y for High Energy Physics, Uni-

Institute

Germany,

Fachbereich Physik,

Bergische

of Physics, Academy of Sciences of the Czech

Rio de Janeiro, Brazil,

x

Dipartimento

of Nuclear Physics, Moscow State University, Moscow, Russia,

Institut fiir Physik, Universitat Mainz, Mainz, Germany, - Universidade de Lisboa, Faculdade de Ciencias, Lisbon,

648

National

des Partic-

Napoli, Italy, 9 CERN,

degli Studi di Lecce e I.N.F.N., Lecce, Italy,

Universidade Federal do Rio de Janeiro, COPPE-EE,

a a

CFNUL

UK,

h

, S. Sushkov cc

di Fisica Nucleare e Tfeorica deU'Universitk

Universidad Autdnoma

V, [/niversitat Mannheim,

di Genova e I.N.F.N., Genoa, Italy, V Institute

of Physics, lechnion,

e

University of California at Irvine, Irvine, USA, w

, G. Usai

, F. J. Wickens

University of Michigan, Ann Arbor, Michigan, USA,

der Leopald-Franzens

Lehrstuhl

Germany,

m

, E . Stefanidis bb

Chilton, Didcot,

degli Studi di Napoli

Energies (IFAE),

di Milano e I.N.F.N., Milan, Italy,

versity of Bern, Switzerland, Universitat

2, France,

Laboratory,

di Fisica deU'Universitk

° Section de Physique, University de Geneve, Switzerland, Italy, $ Institut

, F. Touchard

, S. J. Wheeler '

Rutherford Appleton

de Ffeica d'Altes

j

, S. Spagnolo d

of Physics, University of Wisconsin, Madison, Wisconsin,

di Fisica deU'Universitk

Canada, 3 Dipartimento

University

i

, R. Soluk r

, P. Werner

Department

ules de Marseille, IN2P3-CNRS-Universite' e I.N.F.N.,

g

, A. Sobreira z _

z

di Fisica Department

** Dipartimento di Fisica deU'Universitk Portugal.

di

649

1. Introduction The ATLAS experiment (A Toroidal LHC Apparatus) 1 is a multi-purpose experiment to run at the LHC (Large Hadron Collider), the new accelerator facility under construction at CERN, the European Laboratory for Particle Physics in Geneva, Switzerland. At its design luminosity (10 34 c m _ 2 s _ 1 ) , the LHC will provide 23 inelastic proton-proton collisions for each bunch crossing at a center of mass energy of 14 TeV. The ATLAS detector is composed of concentric shells of specialized sub-detectors arranged in a cylindrical symmetry around the beam axis: an inner tracking detector inside a solenoidal magnetic field of about 2 T, a calorimetric system for energy measurements and a large air-core muon spectrometer that extends for 42 m in length and 22 m in diameter. 2. The ATLAS Trigger DAQ System The ATLAS Trigger and Data Acquisition (TDAQ) system has to face the unprecedented rate of 109 interactions per second and to reduce this to a final event rate of the order of 100 Hz (as imposed by the limited storage data flow), being able to select rare physics events while rejecting the huge amount of background expected at the LHC. To achieve this goal, the ATLAS TDAQ system is structured in three levels (Figure 1): each level has to refine the hypotheses formed at the previous one.

Figure 1.

Block diagram of the three-level ATLAS Trigger/DAQ system.

The first trigger level (LVL1) 2 is implemented with electronic modules directly connected to the calorimeters and muon detectors. It has to reduce the 40 MHz bunch crossing rate to 75 kHz (upgradable to 100 kHz) within

650

a ~2 /zs latency time. At this stage Regions of Interest (Rols) are defined, i.e. parts of the apparatus where relevant physics signatures are detected. The amount of data are then transmitted to the High Level Triggers (LVL2 and Event Filter) 3 , that run on commercial computer farms and are fully software-based. The LVL2 accesses to the data from the LVL1 Rols and processes them with fast algorithms optimized for working in a latency of 10 ms and for reducing the event rate to ~2 kHz. The Event Filter starts from the LVL2 classification to perform a more detailed event reconstruction, including alignment and calibration data. It is expected to perform the event selection in 1 s with an output of the order of 100 Hz. At the end of the selection process, events are finally saved on mass storage.

3. Muon Identification In order to retain events with muons in the final state that can give evidence for important physics processes, the Event Filter has been designed with two complementary software packages entirely developed in C + + for off-line reconstruction: MOORE (Muon Object-Oriented REconstruction) and Muld (Muon Identification), that use information from the Muon Spectrometer to provide excellent muon reconstruction and identification over a wide range of transverse momentum (px from few GeV/c to ~ 1 TeV/c). Monitored Drift Tubes (MDT) and Cathode Strip Chambers (CSC) allow the ATLAS tracking system to reach a high-precision measurements, while Resistive Plate Counters (RPC) and Thin Gas Chambers (TGC) are used to provide the LVL1 trigger. The MOORE package 4 starts searching relevant activities in the Muon Spectrometer volume, subsequently running pattern recognition and track fitting. Owing to the toroidal magnetic field, charged particles are bent in the r-z plane and not in the (p view: for this reason MOORE firstly looks for straight segments from the <j> trigger hits, and then refines track reconstruction by considering the precision hits in the r-z view. Muld 5 runs in two steps. At the beginning, it refits the tracks reconstructed by MOORE to obtain their parameters at the production vertex (Muld StandAlone); propagation through the magnetic field, multiple scattering and energy loss in the calorimeters are properly taken into account. In a second step (Muld Combined), tracks in the Muon Spectrometer are matched together with those found in the Inner Detector by the iPatRec package 6 performing a combined fit, and when matches with %2 probability greater than 1 0 - 3 are found, these are finally kept as identified muons.

651 3.1. MOORE

and Muld in the triggering

process

To use all the required software components in the trigger environment and to avoid dependencies on the off-line environment, the algorithms for the Muon Event Filter have been isolated in the TrigMoore package 7 and implemented in such a way to run both in a "wrapped" mode (e.g. executing MOORE/Muld as in the off-line) and in a "seeded" mode (namely performing reconstruction only in given Rols defined at earlier trigger stages). 4. Reconstruction performance and background rejection

p resolution

Detailed studies on MOORE and Muld have been performed on single muon Monte Carlo samples of different pr's. Fig. 2 shows the l/pr resolutions and the efficiencies obtained with the off-line versions of the algorithms. Muld Combined provides the best px resolution over the full PT range since 1

iPatRec

0.14

t-

; si •

0

MOORE

•

MuldSUmlMone

A

Muld Combined

i

•i

i

i

i

*

t

i

K 0.8

u

0.12

0.6 0.1 0.08 0.06 m 0.04 0.02

D O

I

*. '

,

I

8!

1 . ,1

•

J

1 I

s

0.4

0.2

r

iPalRec

0

MOORE

1

Muld StandAlone

1

Muld Combined

1 Muonp T (GeV/c)

'"

'"

Muon p T (GeV/cj

Figure 2. Transverse momentum resolution (left) and reconstruction efficiency (right) for the algorithms MOORE, Muld StandAlone, iPatRec and Muld Combined.

it exploits both Inner Detector and Muon Spectrometer measurements. The main sources of muon rate in the LVL1 trigger are in-flight decays of charged K and n. The HLT aims to reject such fake muons while keeping high efficiency on prompt muons (mainly from b and c decays 8 ) . This can be achieved through Muld Combined, asking for back-extrapolated Muld StandAlone tracks with small impact parameters having good matching with the corresponding Inner Detector tracks. In Fig. 3 (left) the recostruction efficiency is shown for prompt muons and for decay muons in the low pr region: differential muon trigger rates computed at pr > 6 GeV/c are observed to be ~2.5 times larger from prompt muons than from K/n in-flight decays. Another source of noise in the spectrometer is the uncorrelated cavern background that will be present in the ATLAS experimental area, essentially due to particles (mostly neutrons) produced in the interaction

Figure 3. Muld Combined efficiency versus PT for prompt p, and for fj, from K/7T decays (left). Muld StandAIone efficiency for prompt muons at different transverse momenta without and with addition of different amounts of background (right).

of primary hadrons from p-p collisions with the detector/collider materials, and then interact with matter producing neutral and charged secondaries, that diffuse like a gas through the apparatus and the cavern. Such background has been simulated and extensively studied with TrigMoore, also in terms of execution times 7 ; the plot on the right of Fig. 3 shows the Muld StandAIone efficiency on simulated single muon samples of different px, both without and with background (the "nominal" background intensity has been increased by factors x2, x5, xlO). 5. Conclusions In this work the implementation and the performance of the ATLAS muon off-line reconstruction packages MOORE and Muld in the trigger environment are discussed. The results described here demonstrate that they can be successfully used as Event Filter algorithms. References 1. ATLAS collab., ATLAS Detector and Physics Performance Technical Design Report, CERN-LHCC 99-14 and 99-15, 1999. 2. ATLAS collab., ATLAS Level-1 Trigger: Technical Design Report, CERNLHCC-98-014, ATLAS-TDR-12, June 1998. 3. ATLAS collab., ATLAS High-Level Triggers, DAQ and DCS Technical Proposal, CERN-LHCC-2000-17, March 2000. 4. J. Shank et al, Track Reconstruction in the ATLAS Muon Spectrometer with MOORE, ATL-COM-MUON-2003-012, ATL-COM-SOFT-2003-007. 5. Th. Lagouri et al., A Muon Identification and Combined Reconstruction Procedure for the ATLAS Detector, IEEE Trans.Nucl.Sci. Vol.51 #6 (2004), 3030. 6. R. Clifft and A. Poppleton, IPATREC: Inner Detector Pattern-Recognition and Track Fitting, ATL-SOFT-94-009. 7. D. Adams et al., Moore as Event Filter in the ATLAS High Level Trigger, ATL-DAQ-2003-012, 2003. 8. ATLAS Muon collab., ATLAS Muon Spectrometer Technical Design Report, CERN/LHCC 97-22, May 1997.

Medical Applications and Instrumentations Organizers: S. Pospisil (Medical Application Instrumentation)

V. Sossi (Radiotherapy and Medical Imaging)

G. Bartesaghi Th. Berghofer M.G. Bisogni

A. Bocci S. Braccini A.K. Converse D.B. Crosetto V.J. Cunningham G. Gambarini F. Groppi

G. Iurlaro M.L. Magliozzi

F. Marchetto C.J. Moore

Spectral and Spatial Distribution of the Scattered Radiation of an Electron Radiotherapy Beam A Liquid Ionization Chamber as Monitor in Radiotherapy The Mammographic Head Demonstrator Developed in the Framework of the "IMF Project: First Imaging Tests Results Optimized Infrared Detectors for Infrared Synchrotron Radiation Microspectroscopy and Biomedical Imaging MATRIX: an Innovative Pixel Ionization Chamber for On-line Beam Monitoring in Hadrontherapy Development of a Dual Tracer PET Method for Imaging Dopaminergic Neuromodulation Rethinking Positron Emission Technology for Early Cancer Detection Positron Emission Tomography - Tracer Kinetic Modelling in Drug Development 3D-Reconstruction of Absorbed Dose Obtained from Gel-Dosimeter-Layers Accurate Determination of Radionuclide Purity and Half-Life of Reactor Produced Lu-177g for Metabolic Radioimmunotherapy Spatial Linearity Improvement for Discrete Scintillation Imagers High Resolution, High Sensitivity Detectors for Molecular Imaging of Small Animals and Tumor Detection Strip Ionization Chamber as Beam Monitor in the Proton Therapy Eye Treatment Low Dose, Low Energy 3D Image Guidance During Radiotherapy 653

S. Morzenti

R. Novario A. Santagata

H. Szymanowski L. Tlustos D. Vavrik Z. Vykydal

Alpha Cyclotron Production Studies of the Alpha Emitter 2 1 1 A t / 2 U s P o for High-LET Metabolic Radiotherapy Treatment Planning with IVIS Imaging and Monte Carlo Simulation Monte Carlo Simulations of a Human Phantom Radio-Pharmacokinetic Response on a Small Field of View Scintigraphic Device Applications of the Monte Carlo Code GEANT to Particle Beam Therapy Charge Sharing in Pixel Detectors for Spectroscopic Imaging Direct Thickness Calibration: Way to Radiographic Study of Soft Tissues A portable Pixel Detector Operating as an Active Nuclear Emulsion and its Application for X-Ray and Neutron Tomography

SPECTRAL AND SPATIAL DISTRIBUTION OF THE SCATTERED RADIATION OF AN ELECTRON RADIOTHERAPY BEAM

G. BARTESAGHI, V. CONTI, V. MASCAGNA, M. PREST Universita dell'lnsubria

eINFN, Sezione di Milano

A. MOZZANICA Universita di Milano e INFN, Sezione di Pavia G. FRIGERIO, A. MONTI Ospedale Sant'Anna, Servizio di Fisica Sanitaria,

Como

E. VALLAZZA INFN, sezione di Trieste

The characterization of the radiation scattered by the collimating system of a linear accelerator for medical applications is becoming a must given the increase in the survival of the radiotreated patients. This paper describes both the spectral and spatial study of the scattered radiation of an electron beam in the energy range 6-20 MeV, produced by a Varian CLINAC 1800 accelerator at the Radiotherapy Unit of the Sant'Anna Hospital in Como. The detectors used are plastic and fiber scintillators characterized by a timing response of « 25 nsec in order to cope with the high fluxes. The results have been compared with a Montecarlo simulation based on Geant 3.21 and are presented both in terms of spectral and dose distributions.

1. Introduction The quality of a radiotherapic treatment depends strongly on the capability to measure the dose released in the treated volume 1 and the one absorbed by the surrounding volumes, which is mainly due to the scattered radiation produced by the primary beam interaction with the accelerator collimating system 2 . The increase of the survival rate, and thus the increase in life expectancy, of the radio-treated patients 3 requires the detailed knowledge of the dose due to the scattered radiation in order to estimate the probability of developing radio-induced tumours. The scattered radiation is produced by the beam interactions with the collimating systems of the accelerator head and with the screening material (for example, with

655

656

(a)

(b)

Figure 1. (a) Picture of the CLINAC 1800 setup, (b) The plastic scintillator detector and its photomultiplier.

PMT bias (V)

(a)

lime In)

(b)

Figure 2. a) Calibration curve; the exponent is compatible with the dinode number (10) multiplied by the usual 0.7 factor, (b) Beam bunch duration with respect to the klystron signal.

the cerrobend inserts, where cerrobend is an eutectic alloy of lead and bismuth; in particular the inserts are used to shape the beam whose dimensions usually range between 6x6 and 20x20 cm 2 ). The spatial and spectral characterization of the scattered radiation in a linear accel-

657

erator for medical use, when in electron mode, is very complex. The dosimeters that are commonly used are the ionization chamber (with a typical volume of at least 0.1 cm3) for absolute dosimetry measurements, and the thermoluminescence dosimeters (TLDs) and silicon diodes for relative dosimetry. These dosimeters are integration devices characterized by low efficiency in the interesting energy range; they don't allow the measurement of the spectrum and of the spatial and time distribution of the scattered radiation. On the other side, the use of detectors sensitive to single particles is limited by the high fluxes. The measurements described in this paper have been performed at the Radiotherapy Unit of the Sant'Anna Hospital (Como, Italy) using a Varian CLINAC 1800 in electron mode exploiting the full energy range. The measurements have been performed using plastic and fiber scintillators with a very good time resolution, with different energies and as a function of the distance from the beam axis. The direct beam was completely blocked by a cerrobend layer 1 cm thick. In the following, the results (Section 3) obtained with the plastic scintillator (Section 2) and with a 6 MeV electron beam are described and compared with the simulation of the accelerator setup. 2. The experimental setup Fig. 1(a) shows the CLINAC 1800 setup. The scattered radiation maps have been measured along the treatment couch and under the accelerator head, once the electron beam has been stopped with a cerrobend layer. The detector (fig. 1(b)) is a 3 x 2 x 1 cm3 (6.5 gr) plastic scintillator made of polystirene with 1% PTP and 0.01% bis-MSB a . It is readout by a head-on P30CW5 photomultiplier (Electron Tubes 4 ) with an integrated power supply. The photomultiplier bias has been kept much lower than the nominal value (450 V instead of 1100 V) to stand the high flux from the accelerator, even dealing only with scattered radiation. The data acquisition chain is triggered by the accelerator klystron signal and the charge deposited in the scintillator is integrated by a charge integrating ADC (CAEN V792). The spectral distributions have been measured with the scintillator in auto-trigger mode. A dedicated calibration procedure has been developed to find the conversion factor between the ADC units and the energy deposit in MeV (and thus, given the detector mass, the absolute dose value). The detector has been calibrated from 1300 V to 950 V using a 60 Co source and fitting the Compton edge. For the lower "thanks to the courtesy of Prof. G. Giannini, University of Trieste

658 6MeV

I

si

E


^ '

-WI 3 ^™^* 2-5 i ^ — '

/ / -

^

12 7 K

4 ^K 2 \ 0 ^ -2

j

"

0

10

20

3D

40

50

70

60

dlslance romthe Isoct nter(c m

(a)

^ ^ ' ^ ^ -15 *™ -10 1 5 11

^

^ (jstnl

2D

(b)

Figure 3. (a) ID scan with a 6 MeV electron beam and a 6x6 cm2 field with the cerrobend in place, (b) 2D map of the scattered radiation below the accelerator head.

I

A

•

TLD

A

/N 1f

-1

• " " .

i

distance frome the Isocenter (cm)

Figure 4. Comparison between the dose values measured by the scintillator and the TLDs using a 6 MeV electron beam.

bias voltages, data have been acquired at the accelerator from 450 V to 1000 V and then rescaled using the common values. Fig. 2(a) shows the resulting curve with superimposed the fit with the standard gain function while fig. 2(b) presents

659

0 (a)

0.2

0.4

0.6

0.8 1 1.2 Energia (MeV)

1.4

1.6

1.8

(b)

Figure 5. (a) Example of the simulation of the CLINAC 1800; 1000 electron events are generated, (b) Comparison data-simulation of the scattered radiation spectral distribution at 75 cm from the isocenter along the couch; the dark central part is made of electrons, while the light tracks are photons.

the particle beam bunch duration with respect to the trigger signal. The beam rate with a 6 MeV electron beam is 300 Hz (150 Hz for all the other energies) while the number of filled bunches varies from 15% at 80 MU b to 67% at 400 MU. 3. Results Fig. 3(a) shows the ID scattered radiation map along the couch with a 6 MeV electron beam and a 6x6 cm2 field. The dose values have been computed considering the scintillator mass. The presence of a valley at the isocenter is due to the cerrobend stopping the primary beam. Fig. 3(b) shows the 2D map under the accelerator head. The dose values have to be compared with the typical 20-70 Gy of a radio-treatment. Fig. 4 shows the comparison of the scintillator dose values with the ones measured with the TLDs (TLD100, Harshaw Chemical Company). The trend is clearly the b

MU = Monitor Unit; 1 MU is 1 cGy in water at the isocenter, that is 100 cm from the source spot

2

660

same while there is still a discrepancy (a factor 5 below in the highest dose region) in the absolute dose values that is under investigation. The setup has been simulated with Geant 3.21 (fig. 5(a)) and the spectral distribution of the scattered radiation has been compared with the data acquired with a large scintillator (10x3x 1 cm3) as shown in fig. 5(b). 4. Conclusions This article describes only a few of the measurements that have been performed with the CLINAC 1800. There is still a discrepancy between the absolute dose values computed with the scintillator and with the TLDs, which is probably due partly to the scintillator calibration procedure and partly to the fact that the TLDs have been calibrated with very low doses, instead of the primary beam typical ones. Once the discrepancy has been understood, two full prototypes will be built, a matrix of plastic scintillators and a bundle of scintillating fibers. These detectors, which have the clear advantage of low cost and tissue-equivalence, will be able to monitor the scattered radiation real-time during the patient treatment. Tests are on-going to use them also as dosimeters for the primary beam. References 1. D. W. O. Rogers, Phys. in Can. 51, 4, 178-181 (1995). 2. L. J. van Battum et al, Phys. Med. Biol. 48, 2493-2507 (2003). 3. U. Amaldi, G. Tosi, The TERA project and the centre for oncological hadrontherapy, U.Amaldi and M. Silari editors, 2nd Edition - April 1995 4. Photodetector package P30CW5 data sheet, Electron Tubes Inc., United Kingdom, http://www.electrontubes.co.uk

A LIQUID IONIZATION C H A M B E R AS M O N I T O R IN RADIOTHERAPY

TH. BERGHOFER, J. ENGLER, J. M. MILKE AND J. R. HORANDEL University and Forschungszentrum Karlsruhe, Institut fur Kernphysik, PO Box 3640, 76021 Karlsruhe, Germany G. H. HARTMANN Deutsches Krebsforschungszentrum (DKFZ), Im Neuenheimer Feld 280, 69120 Heidelberg, Germany First measurements with a prototype liquid ionization chamber are described to be applied as an online-monitor for intensity modulated radiotherapy. The detector consists of 480 individual electronic channels which allow parallel read-out of radiation induced currents at frequencies exceeding 10 Hz. Dose gradients in the direction of leaf movement of a multileaf collimator have been measured and a reconstruction method for individual leaf positions has been developed. The achieved reconstruction accuracy will be described.

1. Introduction Modern conformational and intensity modulated radiotherapy (IMRT) with photons and high gradient fields require an increased effort in dosimetry to guarantee safety and reproducibility of the treatment. In most cases, the radiation field is shaped by means of a multileaf collimator (MLC) to adapt the delivered dose to the shape of the tumor while sparing the healthy tissue. The aim of our research is to develop a detector to monitor the radiation during the treatment at the exit of the MLC before it is applied to the patient. The chamber design is adapted to be used gantry mounted together with a miniature MLC a and a linear accelerator b at the DKFZ. To verify the correct shape of the beam, the detector has to monitor the leaf positions of the MLC and therefore the corresponding dose distribution. As the chamber is placed in the radiation field, it has to be thin to absorb a a

ModuLeaf, MRC Systems GmbH, Heidelberg, Germany. PRIMUS, 6/15 MV, Siemens OCS, Concord, CA, USA.

b

661

662

Valve

Is Figure 1.

conductive pixel

ionization liquid

"

exit window

Schematic cross-section of the detector.

minimum of radiation. On the other hand, one has to sample a certain amount of the radiation to get a reasonable noise-free picture. Air- or gasfilled chambers would need quite a large extension in direction of the beam to meet this condition. Thus, we have chosen to develop a liquid-filled ionization chamber which allows, due to the higher density of the liquid, a thin design of the detector while still delivering high current signals for typical clinical dose rates of several Gy m i n - 1 . In order to monitor dynamic leaf positions with a leaf speed of approximately 2 cm s _ 1 in the case of dynamic IMRT, the repetition rate of the read-out has to be in the order of several 10 Hz. Consequently, the charge yield and the charge carrier mobility of the irradiated liquid have to be quite high to achieve sufficiently high signals in the read-out interval. The liquid used in our test chamber, tetramethylpentane (TMP), meets the above-mentioned conditions x. It is purified to a few ppb of oxygen equivalent using a vacuum distillation facility on site 2 . The liquid commonly used in ionization chambers for medical applications is isooctane. For purified TMP the current signal can be increased by a factor of 4 with respect of isooctane, analysis grade 3 . 2. Material and Methods The chamber consists of three printed circuit boards which are kept at constant distance by two stainless steel rings, see Fig. 1. The distance between the electrodes and therefore the thickness of the rings has been chosen to be 5 mm in order to maximize the signal to noise ratio. The chamber is filled with the ionization liquid via two stainless steel pipes with valves at their ends welded in the lateral surface of the stainless steel rings. The boards are a multilayer construction made of a special glass reinforced ceramic laminate which has turned out as inert regarding the ionization liquid. The outer boards serve as beam entry- and exit-window and high voltage cathode likewise. For this purpose, they exhibit a circular copper cladding with a diameter of 136 mm on their inside which has been

663

Leaf

4

la

\

>- 0.6 -

* 1.5 2 2.5 3

3.5 4 4.5 5

dose rate [Gy min-1]

-60

-40

reconstructed leaf positions

-20

0

20

L 40

60

x-position at isocenter [mm]

Figure 2. Left: Dose rate dependence of the ionization current at an electric field strength of 2 kV c m - 1 . The dashed line is a power-law fit to the data. Right: Measured beam profile (filled points) for a field size of 60x40 mm in t h e isocentric plane. The dashed curve is a profile fit according to Eq. 1. The solid lines indicate the reconstructed leaf positions at 50 % of the relative dose maximum.

chemically gold-plated. The anode-board in the middle is realized as a five-layer construction with 24 x 20 interconnected square conductive pixels on each side. The pixel rows are aligned with the collimator leaves and the pixel size of 3.3x3.3 mm 2 with a spacing of 0.2 mm is adapted to the leaf width of 1.75 mm. The actual detector design allows the control of two adjacent leaf pairs by one row of pixels. The pixels are surrounded by a guard electrode the same size as the high voltage electrode to form an adequate homogeneous electric field at the edges. Both the pixels and the field shaping electrode are on virtual ground potential. The collected charge of each pixel is traced to eight SCSI connectors at the outside of the chamber using a circuit track system in the middle layer of the anode board. The data acquisition has been described elsewhere 3 .

3. Results In the following we present test measurements to determine the detector characteristics at a photon energy of 6 MeV. 3.1. Signal

response

The detector response has been measured for different dose rates at a fixed electric field strength of 2 kV c m - 1 . The dose rate has been varied between 1.1 and 4.4 Gy m i n - 1 by changing the distance between the detector and

664

Mean: -0.243 mm RMS: 0.134mm

deviation from nominal position [mm] -20

0

20

40

60

nominal leaf-position [mm] Figure 3. Reconstructed leaf positions in a range from -43 mm to + 4 3 mm at isocentre versus the nominal position. The solid line indicates a linear fit to the data points. The inlay shows a frequency distribution of the deviation from the nominal leaf position.

the radiation source. The measurements have been performed in a waterequivalent plastic phantom to incorporate the dose buildup. The corresponding dose rate has been measured with a diamond detector mounted below the exit window of the detector. In Figure 2 (left) the dose-rate dependence of the ionization current is plotted. A deviation from strict proportionality of the ionization current due to space charge effects in the liquid is noticeable. Due to the recombination of charge, the relationship between the measured current / and the dose rate D is a power-law of the form / = N • DA 4 with a constant of proportionality N. This non-linear behaviour affects the imaging properties of the detector because for small currents the corresponding dose-rate is overestimated, but the effect can be corrected using the inverse function. 3.2. Reconstruction

of leaf

positions

Knowing the relationship between the measured ionization current and the dose rate, the signal of each pixel can be converted to the corresponding dose rate value. The measured beam profiles are normalized to the open field values and the dose rate distribution caused by leakage of the collimator, i.e. the dose rate measured when all collimator leaves are closed, is subtracted as a general offset. The positions of the collimator leaves in the isocentric plane

665

are determined from the relative dose values caused by partial irradiation of the pixel. With the assumption of a gaussian detector kernel, the beam profiles can be fitted using a general function of the form 5 D(x) =

\\eri

y/n(L - 2(x - x0)) 20V,

+ erf

VTT(L + 2(x -

a;0))

2CT„

(1)

where D is the relative dose value, erf(x) the error-function, L the field width and
J. Engler, J. Phys. G: Nucl. Phys. 22, 1 (1996). J. Engler, Nucl. Instrum. Methods A 427, 528 (1999). K. Eberle et al, Phys. Med. Biol. 48, 3555 (2003). J. F. Fowler, Radiation dosimetry: Instrumentation vol 2, 291 (1966). F. Garcia-Vicente et al, Phys. Med. Biol. 45, 645 (2000). G. H. Hartmann and F. Fohlisch, Phys. Med. Biol. 47, N171 (2002).

THE MAMMOGRAPHIC HEAD DEMONSTRATOR DEVELOPED IN THE FRAMEWORK OF THE "IMI" PROJECT: FIRST IMAGING TESTS RESULTS MARIA GIUSEPPINA BISOGNI + Physics Department

"E. Fermi", University of Pisa, Largo B. Pisa, 56127, Italy

Pontecorvo3

In this paper we report on the performances and the first imaging test results of a digital mammographic demonstrator based on GaAs pixel detectors. The heart of this prototype is the X-ray detection unit, which is a GaAs pixel sensor read-out by the PCC/MEDIPDQ circuit. Since the active area of the sensor is 1 cm2, 18 detectors have been organized in two staggered rows of nine chips each. To cover the typical mammographic format (18 x 24 cm2) a linear scanning is performed by means of a stepper motor. The system is integrated in mammographic equipment comprehending the X-ray tube, the bias and data acquisition systems and the PC-based control system. The prototype has been developed in the framework of the integrated Mammographic Imaging (IMI) project, an industrial research activity aiming to develop innovative instrumentation for morphologic and functional imaging. The project has been supported by the Italian Ministry of Education, University and Research (MIUR) and by five Italian High Tech companies in collaboration with the universities of Ferrara, Roma "La Sapienza", Pisa and the INFN.

1. Introduction The digital mammographic demonstrator presented in this paper was developed in collaboration among Italian high-technology industries and public research institutions. The Integrated Mammographic Imaging (IMI) is in fact an industrial research project aimed to produce prototypes of new systems for the morphologic and functional imaging to be applied in the screening, diagnosis and follow-up of the breast cancer [1]. In this paper, we report on the performances and the first imaging test results of a digital mammographic demonstrator based on GaAs pixel detectors. We will provide a description of the mammographic system and the experimental procedures followed to characterize the device and to assess its working point. Finally, we will report on the experimental results obtained in the imaging tests performed with an in house-made mammographic phantom.

On behalf of the Integrated Mammographic Imaging (IMI) proiect.http://imamint.df,unipi.it/

666

667

2. Materials and Methods 2.1. The digital mammographic demonstrator The digital mammographic demonstrator is composed by an X-ray generation unit and by the detection head, both handled by a controller PC [2]. The detection head is based on GaAs pixel detectors as sensor units read-out by the VLSI integrated circuit PPC, developed within the MEDIPIXI collaboration and operating in single photon counting mode [3, 4]. The detectors have been produced on 3 inches 200 um thick Semi Insulating GaAs wafers in collaboration with AMS (Alenia Marconi Systems, Roma, Italy). Each detector is a matrix of 64 x 64 150 um square pixels, with an inter pixel distance of 20 um, geometrically matching the electronics chip pixels. The active area of the sensor is 1.2cm2[5, 6]. We have produced in total 25 assemblies with an overall yield (number of bad pixels between 30 and 180 over 4096 per chip) between 95.5% and 99.5%. Since the typical format of a mammographic film is 18 x 24 cm2, to cover such a field with 1.2 cm2 sensor units, we adopted a mosaic geometry with staggered rows of nine assemblies each one. For this demonstrator we have implemented a version with two rows scanned across the whole exposure field by means of a stepper motor. In addition, a lead collimator, mechanically coupled to the detection head, has been positioned at the tube output to limit the irradiation area to the sensor units. The best 18 assemblies produced by AMS have been mounted onto a custom designed chipboard and read-out by an acquisition system named MMRS (Multichip Medipix Readout System), both developed by LABEN. The MMRS consists of two boards: an interface board (IB), directly connected to the chipboard, and a PCI board hosted in the controller PC. The IB implements all the logic functions of the MMRS while the PCI one is a commercial board (NI 6533) that interfaces the IB to the PC. Another board, named the BIAS Board, developed by CAEN, provides the analog and digital biases of the chips. Gilardoni provided the X ray tube by adapting the clinical mammographer SYLVIA to host the collimator and the detection head. A controller PC handles the whole image acquisition process (the chips settings, the X rays emission gate, the data taking and the stepper motor sequence) by means of a software program written in Lab Windows CVI. Figure 1 (a) shows the whole mammographic system: the X ray tube with the collimator mounted at the exit window and the box containing the detection head. Figure 1 (b) shows the detection head, in particular, the 18 assemblies mounted onto the chipboard in a mosaic configuration.

668

Figure 1 (a) The whole mammographic system: the X ray tube with the collimator mounted at the exit window and the box containing the detection head, (b) The detection head, in particular, the 18 assemblies mounted onto the chipboard in a mosaic configuration.

2.2. Set-up optimization We assessed the electrical working point of the 18 GaAs detectors mounted onto the chipboard, biasing them at 350 V and measuring the inverse current drawn by each one. The high bias voltage arises is needed to over-deplete the GaAs sensor and to achieve a uniform charge collection efficiency across the detector surface. After this test, 16 chips over 18 drawn a current less than 5 uA, which is a limit fixed to prevent the detector breakdown. Unfortunately, for the chessboard configuration of the assemblies, we can use only a region 12 cm wide of contiguous good chips. To produce images of high quality, the sensitivity of the 12 assemblies must be uniform. This can be achieved calibrating the energy discrimination threshold of the chips and setting them at the same level. The minimum threshold of the 12 chips ranges from 11 keV to 16 keV. If we exclude one chip with a very high threshold (17 keV), the average threshold that can be set-up for the remaining chips is 13 keV. This value is low enough to detect photons in the mammographic energy range ( 1 2 - 3 0 keV).

669

3. Results To test the imaging capability of our system, we acquired the radiographs of an in house made phantom simulating breast tissue with malignancies. The phantom consists of a lucite cylinder, 4 cm thick 10 cm in diameter, containing nine Al disks, 4 mm in diameter and thickness ranging from 125 down to 15 urn, embedded in as many wax cylinders. The phantom was exposed to the X ray beam (tube settings: 27 kV, 30 mAs) and the image acquired by scanning the detector across the phantom surface with a sequence of 13 exposure-motion steps. For each step, we have chosen a shutter time of 0.5 sec and a delay time of 20 s (time between two successive exposures). The heat dissipation rate from the mammographic tube dictates this long delay time. The dose per exposure (measured with a silicon dosimeter) delivered to the phantom was 4.8 mGy. The image of the phantom, shown in figure 2, has been obtained by weighting the raw image with a high statistics flat field and corrected for the noisy pixels. The quality of the image is good especially for the detection of the low contrast Al disks. In particular, the 15 urn thick Al disk (top one in the middle column of the image) shows a contrast against the wax background below 1 %. Unfortunately, at the center of the image the effect of the chip with the high threshold is clearly visible.

Figure 2 Radiograph of the phantom consisting on a lucite cylinder, 4 cm thick 10 cm in diameter, with nine Al disks, 4 mm in diameter and thickness ranging from 125 down to 15 urn, embedded in as many wax cylinders. The image has been acquired by scanning the detector across the phantom surface with a sequence of 13 exposure-motion steps.

670

4. Conclusions In this paper, we have reported on the performances of the digital mammographic demonstrator developed in the framework of the Integrated Mammographic Imaging project. The demonstrator has been fully tested and optimized to acquire radiographs of a mammographic phantom in exposure conditions typical of a clinical examination. The system is able to produce high quality radiographic images with a standard dose. Its performance is particularly high in the detection of low contrast details. With this system, we were able to detect a 15 um thick Al disk against a wax background. The contrast measured on the image is below 1 %.

Acknowledgments This project has been supported by the Italian ministry of Education, University and Research (MIUR) and by the Italian High Tech companies, AMS, CAEN, LABEN, Pol-Hi-Tech, in collaboration with the Universities and the INFN sections of Ferrara, Napoli, Roma "La Sapienza", Pisa. References 1. A. Stefanini et al., Nucl. Instr. Meth A518 (2004), 376-379. 2. S.R. Amendolia et al., Nucl. Instr. Meth A518 (2004), 382-385. 3. M.Campbell, E.H.M.Heijne, G.Meddeler, E.Pernigotti, W.Snoeys, IEEE Trans.Nucl.Sci. 45 (3), (1998) 751-753 4. http://medipix.web.cern.ch/MEDrPIX/. 5. M. Novelli et al., Nucl. Instr. Meth A509 (2003), 283. 6. M. G. Bisogni et al, IEEE 2004 Conference Records 0-7803-8701-5/04.

OPTIMIZED INFRARED DETECTORS FOR INFRARED SYNCHROTRON RADIATION MICROSPECTROSCOPY AND BIOMEDICAL IMAGING ALESSIO BOCCI*, EMANUELE M. PACE Dipartimento di Astronomia e Scienza dello Spazio, Universita di Firenze, Largo E. Fermi 2, Firenze, 50125, Italy AUGUSTO MARCELLI INFN-Laboratori Nazionali di Frascati, Via E. Fermi 40, Frascaii, 1-00044, Italy PIERANGELO MORINI, DIEGO SALI Bruker Optics Sri, Via G. Pascoli 70/3, Milano, 20133, Italy Infrared Focal Plane Arrays are bidimensional detectors allowing images of large sample area and a large number of spectra with shorter acquisition time, higher spatial resolution and higher S/N ratio than single-element detectors. Imaging spectrometers based on focal plane arrays are now used in several biomedical applications to investigate different diseases by comparing IR spectra in healthy and pathological tissues. Moreover, the availability of brilliant IR synchrotron radiation sources against bench-top IR spectrometer allows diffraction-limited spectra with a higher S/N ratio.

1. Introduction The significant development of the infrared (IR) detector technology, in particular of large IR Focal Plane Array (FPA) detectors, allows imaging spectroscopy with high spatial resolution. nIR (1-5 um) and midIR (5-20 um) FPA detectors are used for spectroscopy imaging in astronomy, space applications and currently also for biomedical applications [1]. Recently, FPA detectors have been mounted inside IR microscopes coupled to Fourier Transform Infrared (FTIR) interferometers for micro-spectroscopy applications using conventional sources [2]. They offer several advantages in microspectroscopy with respect to a single-element IR detector, such as higher spatial resolution, higher S/N ratio and lower collection time to store the same area at 1

Alessio Bocci, PhD Student, DASS, Universita di Firenze, L.go E. Fermi 2, 50125 Firenze, Italy. Tel. & Fax +39-055-2307750. E-mail: [email protected].

671

672

comparable S/N ratio [3]. Despite these advantages, commercial IR FPAs are not yet optimized for such applications and they are still limited in response and readout times. Moreover, their pixel pitch can limit the effective spatial resolution in many experiments. Furthermore, commercial FPAs have been mainly designed for classified applications and are not optimized for the most brilliant IR sources now available such as synchrotron radiation sources (IRSR). Presently, the availability of several IRSR beamlines and their use for microspectroscopy is rapidly growing because they allow the collection of spectra with the highest S/N ratio and the highest spatial resolution (down to the diffraction limit) [4]. FTIR spectrometers are designed to work with globar sources, so that synchrotron radiation sources illuminate a significantly smaller area of a 2-D FPA [5]. Therefore, new optimized FPA devices are required in order to exploit the advantages of IRSR micro-spectroscopy against bench-top spectrometers. 2. FTIR micro-spectroscopy with conventional sources Micro-spectroscopy imaging can be performed with a FTIR spectrometer coupled to an IR microscope with an FPA detector [6]. IR microscopes are based on reflective optics: two Cassegrain objectives are used to illuminate the sample at the focal point with a first Schwarzschild objective while the second one - the condenser - is used to collect the radiation from the sample and re-focus the radiation to the FPA. Imaging of the sample can be realized also with a raster scan using a single-element detector: a map of the sample can be obtained by moving the sample stage of the microscope and collecting spectra at different points [6]. Using single-element detectors in a microscope gives a small aperture and then a small illuminated sample area. Instrumentation with standard IR sources can probe areas of only about 20-50 um of diameter with a good S/N spectral ratio. This capability strongly limits the study of biological systems and in particular single cell investigations [7]. Micro-spectroscopy imaging provides a spatial resolution that is defined by the optics and by the pixel pitch of the array. Moreover, microscopes are typically used without apertures to illuminate large areas of the sample that are then measured by large area FPA detectors. Typical cooled FPA detectors for FTIR spectrometers have a 64 x 64 pixel format with a 40 um pixel pitch, so that the illuminated area is about 2.5 mm x 2.5 mm; using a 15X (36X) objective, the measured sample area is about 170 x 170 um (85 x 85 um) achieving a nominal resolution for pixel of 2.6 um (1.3 um). The advantages of an FPA over a singleelement detector are: (a) larger sample can be measured in a shorter acquisition time and with higher spatial resolution, almost near the diffraction limit, and

673

higher S/N ratio (the S/N ratio depends on the size of the element and pixels are typically significantly smaller than in a single-element detector) [5,6]. Taking into account their performances and the availability of intense IRSR sources in the last decade, FTIR micro-spectroscopy has been applied to investigate biological systems and also the first biomedical research has appeared [8]. Indeed, FTIR imaging offers unique opportunities to biologists to analyze tissue specimens and diagnosing diseases and/or monitoring progression of diseases appear promising applications [9]. midIR light is used to excite vibrational states of the molecules in a tissue; the absorption frequencies depend on the molecular chemical structure. The IR absorption spectra in biological samples may provide information on the components of the tissue and the distribution of nucleic acid, protein, lipid and carbohydrate contained in the tissue. By combining imaging and the sensitivity of the technique, we may detect small chemical variations and identify the distribution of the fundamental components of a biological sample providing important and unique details about the chemistry of pathological states which can be eventually correlated to specific diseases [10]. However, biological tissues are very heterogeneous, several different components may occur and also different cell types can be involved in these studies. Cells have also different sizes in different tissues with values ranging between 10 to 50 urn. In order to detect the pure tissue components in a single cell and to obtain accurate and reliable information about the differences between healthy and diseased tissues, a high spatial resolution coupled to a high S/N ratio has to be obtained [7]. 3. New FPA detectors for IR micro-spectroscopy Why can synchrotron radiation sources be used in imaging micro-spectroscopy achieving the highest performances? The main reasons are due to the unique characteristic of the IRSR emission: high brightness, sub-nanosecond temporal structure and absence of thermal noise [4]. Moreover, with a SR source it is possible to use the microscope in a confocal configuration, e.g., with two apertures in the microscope before and after the sample stage to obtain the maximum spatial resolution, a property that is precious for biomedical applications. The area of the specimen that can be measured is of the order of the investigated wavelength, e.g., at the diffraction limit, and spectral imaging of the sample can be obtained with a single-element detector using a mapping technique [6]. This technique achieves higher resolution and image contrast than a conventional source with an FPA detector, although longer acquisition times are required [5]. However, the use of an FPA detector may significantly improve

674

the advantages of micro-spectroscopy with synchrotron radiation if an optimized array detector is used. The first characterization of FPA performances with a synchrotron radiation has been realized at ANKA [5,11]. The microscope was a BRUKER Hyperion 3000 coupled to a 64 x 64 pixels cooled HgCdTe FPA detector. Experimental results clearly addressed the well know advantages of IRSR with respect to bench-top instrumentation. Moreover, the experimental data showed that commercial FPA detectors have a larger sensitive area compared to the natural area illuminated by the IRSR (at ANKA, ~ 30 x 30 pixels de-focusing the IRSR source without apertures and - 8 x 8 pixels with an aperture of 300 urn and 36X objective equivalent to an area of the sample of ~ 8 x 8 urn). Therefore, a frame format of 5 x 5 or 10 x 10 pixels, with a pixel size of 10 or 20 urn may better fit an IRSR source. In addition, reducing the number of pixels should likely reduce the time to collect the images. Shorter collection time makes now feasible the use of FPA detectors inside FTIR spectrometers in rapid scan mode or in experiments approaching the sub-microsecond domain [6,12,13]. The pixel pitch should be reduced down to at least 10 urn in the midlR to still improve the spatial resolution. The future technological developments follow this trend [14]. The response time of a FPA detector, now limited to a few us, should also be improved [6].

S[mV]

. 4 7 19-22 12 14

""•

. I '

27-Si \

I ,

" 1

J

.

1 !

•

$4

:

41 -,* Si

•

•

'

34

,

61

'

49 73

'

•

,

••

t

•••

'

|

••

•

^ ™ -

~

t[nsl •

;•"

' T'

'—

•

Figure 1. The IR emission of the DAONE bunches as measured by the fast single element detector. The x and y scales are 50 ns/div and 2mV/div respectively.

We are developing an ultra-fast compact infrared uncooled array detector. This new technology is based on infrared HgCdTe detectors (photoconductive or photovoltaic) working at room temperature or with Peltier coolers, optimized to detect nIR and midlR wavelengths [15,16]. The main advantage of these detectors is their capability to operate at room temperature (or at Peltier cooler temperatures) that contributes to the reduction of the overall size of the devices. The main drawback is the detectivity being typically two orders of magnitude -

675 or more - lower than that of liquid-nitrogen cooled photodetectors [15]. However, optically immersed photovoltaic devices, which are cooled by 2-stage Peltier coolers, show detectivity up to 1011 cm Hz1/2/W at 5 urn, closely approaching the BLIP limit. The great advantage of these devices is the very fast response time with a time constant of about 1 ns. Moreover, photovoltaic and photoconductive detectors can be assembled in linear or small 2D arrays [17]. A single photo-voltaic detector element operated at room temperature has been recently tested to monitor the IRSR emission of the electron bunches of the DAONE collider [18,19]. This detector, an optically immersed photovoltaic multiple junction detector, is characterized by a ns response time in the 8 to 12 urn range with optimized sensitivity at 10.6 urn [20]. Figure 1 shows the IR signal measured by the detector when a particular bunch structure was selected and stored in DAONE [19]. In this experiment the IR detector allows to separate the emission of individual bunches. In these preliminary tests the measured response time of the detector was about 3.5 ns. Figure 2 shows the detector and the experimental setup [21].

Figure 2. The IR uncooled detector (left) and the IR experimental area at DAONE (right).

4. Conclusions IR FPA technology grows rapidly in the last decade. However, only recently FTIR spectrometers have been equipped with FPA detectors for imaging. This detector allows the acquisition of a large number of spectra from large samples in a short time (tens of minutes) and with high spatial resolution, close to the diffraction limit. This is particularly important because both large areas and different types of tissues may be studied with a high spatial resolution and in a shorter time, as required in biomedical diagnostics. The use of SR sources can drive many other biological researches towards new applications because spectra can be collected with an S/N ratio and a contrast not achievable by any

676

conventional source. IR uncooled array detectors with sub-ns response time should also be developed and applied as real-time monitors being sensitive to the sub-nanosecond infrared emission of a synchrotron radiation ring or used in a particle accelerator for beam diagnostics. Acknowledgments We acknowledge J. Piotrowski for useful discussions and his valuable suggestions. A special thank is due to A. Drago, M. Cestelli Guidi, M. Piccinini and to the staff of the DA<SNE-L laboratory for their support. One of the authors (A.B.) acknowledges IMONT for partial financial support. References 1. P. Colarusso et al, Appl. Specrosc. 52, 3, 106 (A) (1998). 2. E.N. Lewis et al, Anal Chem. 67, 19, 337 (1995). 3. J.L. Koenig et al, Anal. Chem. 73, 360A (2001). 4. G.L. Carr, Rev. Sci. Instr. 72, 3, 1613 (2001). 5. D.A. Moss et al, ANKA Annual Report, 117 (2004). 6. I.W. Levin, R. Bhargava, Annu. Rev. Phys. Chem. 56, 42 (2005). 7. L.M. Miller et al, Jour. Of Biol. Phys. 29, 219 (2003). 8. A. Marcelli and P. Calvani, Nucl. Sci. Tech. 14, 93 (2003). 9. D.L. Wetzel and S.M. LeVine, Science 285, 5431, 1224 (1999). 10. R.K. Dukor, Handbook of Vibrational Spectroscopy V5, Wiley & Sons (2001). 11. D. Moss, Proc. Wirms 2005, Infr. Phys. & Tech. (2005) in press. 12. CM. Snively et al, Opt. Lett. 24, 1841 (1999). 13. L.H. Kidder et al, Opt. Lett. 22, 742 (1997). 14. P.R. Norton, Proc. SPIE 3698, 652 (1999). 15. J. Piotrowski, Opto-Electron. Rev. 12, 111 (2004). 16. J. Piotrowski and A. Rogalski, Infrared Phys. Technol 46, 115 (2004). 17. A. Piotrowski etal, Bull. Pol. Acad. Sci., Tech. Sci. 53, 139 (2005). 18. A. Drago et al, Proc. PAC2003, Portland, Oregon, (2003). 19. A. Bocci et al, LNF-05-12 (NT), (2005). 20. http://www.vigo.com.pl/dane/detektorv/pvmi.pdf. 21. M. Cestelli Guidi et al, Jour. Opt. Soc. Amer. All, 2810 (2005).

MATRIX: AN INNOVATIVE PIXEL IONIZATION CHAMBER FOR ON-LINE BEAM MONITORING IN HADRONTHERAPY S. BRACCINI + , G. PITTA' TERA Foundation, Novara,

Italy

M. DONETTI CNAO Foundation, Milan, Italy R. CIRIO, A. LA ROSA, M. A. GARELLA, S. GIORDANENGO, F. MARCHETTO, C. PERONI University andlNFN,

Turin, Italy

The control of intensity, position and shape of clinical beams are key issues in the treatment of tumours using hadron beams, especially in the case of active dose distribution systems. For this purpose an innovative pixel ionization chamber, named MATRIX, has been designed, constructed and tested. The chamber is conceived to be located very near the patient to precisely monitor the beam parameters used to verify the treatment planning specifications. MATRIX operates in air and is characterized by a 21 x 21 cm2 sensitive area subdivided in 1024 pixels of 6.5 x 6.5 mm2. To minimize the amount of material crossed by the beam, the anode is made of a 50 ^m kapton foil, with a deposit of 17 nm copper on each side. A very sensitive electronics is used for the readout, based on a dedicated chip. In this paper the construction of the chamber and the very positive results of the first beam tests are described.

1. Introduction Hadrontherapy is a technique of cancer radiation therapy based on the use of proton and carbon ion beams. Due to their characteristic ionization loss, the so called Bragg curve, and due to the fact that being charged they can be precisely directed toward a small target, hadrontherapy allows to conform the dose to the tumour volume much better than conventional X-ray radiotherapy. The healthy tissues that surround the tumour are therefore spared at best and hadrontherapy plays a key role in the cure of deep seated tumours located in proximity of organs at risk (OAR). Up to present, about 45000 patients have Corresponding author. E-mail: [email protected].

677

678

been treated in the world with hadron beams [1] and this number will significantly increase in the coming years when the many hospital based hadrontherapy centres now under construction will become operational [2]. During a treatment, the on-line control of beam intensity, position and shape is of paramount importance in order to assure the conformity of the dose given to the patient according to the treatment planning specifications. The integral of the dose is usually measured using ionization chambers and accurate measurements of the clinical fields are performed off-line before irradiating the patient. In case of a passive dose distribution system, like in conventional X-ray radiotherapy, the tumour is globally irradiated by a large beam shaped by personalized collimators. In case of an active dose distribution, a "pencil" beam of about 8 mm FWHM is magnetically controlled and driven to distribute the dose to the tumour volume. In this case the position and the intensity of the beam must be continuously monitored on-line to assure the correct irradiation for each position of the clinical beam. To accurately control the clinical beams during hadrontherapy treatments an innovative pixel ionization chamber, named MATRIX, has been designed and constructed. In this paper the construction and the first beam tests of this detector are described. 2. Design and construction The structure of the chamber is made of five successive fiberglass frames, as shown in Figure 1.

Figure 1. Schematic view of the MATRIX pixel ionization chamber.

679 The external frames are equipped with glued mylar foils which allow to maintain the planarity of the internal cathodic and anodic electrodes in case a gas is used at non atmospheric pressure. The sensitive volume of the detector corresponds to the air gap between the anode and the cathode foils. MATRIX is designed to have a gap of 5 mm but a different gas volume can be easily obtained by substituting the spacer. The anode and cathode foils are glued to the respective spacers using a non-conductive epoxy araldite. The gluing procedure required the construction of a special tool, designed to obtain a good planarity of the electrodes. The chamber is characterized by a 21 x 21 cm2 sensitive area subdivided in 1024 pixels of 6.5 x 6.5 mm2. In order to minimize the amount of material traversed by the beam, the anodic electrode is made of a thin 50 urn kapton foil, with a deposition of 17 um copper on each side. As shown in Figure 2 (left), the pixels are located in one side of the anode foil together with a guard ring and pads for soldering the read-out connector. The other side of the anode foil contains the pixel-pad connections. Metallized holes are used to electrically connect the two sides.

The fully assembled MATRIX detector is shown in Figure 2 (right). The 1024 channels are read-out by 8 electronic boards which handle 128 channels each. Each board hosts two VLSI chips, named TERA 06, specifically designed for this specific application [3]. Each chip has 64 independent channels based on a recycling integrator architecture which allows the implementation of a wide dynamic range charge-to-digital converter. Each channel is equipped with a 16bit digital counter whose output value is proportional to the integrated charge. The sensitivity can be adjusted by varying the quantum of charge, which corresponds to one digital count, in the range between 100 and 800 fC. The

680

maximum frequency of 5 MHz limits the maximum current for each channel to 4 uA. The linearity of the measurement of the charge has been measured to be better than 1% in a wide range. The tested minimum read-out time for all the 1024 channels of MATRIX is 1 ms. 3. Beam tests Laboratory tests have been performed before the first beam tests. The electronic noise has been measured to be less than one count per channel per second. The chamber has been irradiated with X-rays in order to verify the correct operation of all the 1024 pixels. For all these tests and the following beam tests the voltage of 400 V has been applied which corresponds to an electric field of 800 V/cm in the sensitive volume. The first beam tests have been performed at the cyclotron of the Joint Research Centre of the European Commission in Ispra (Italy) at the end of 2004. Using a proton beam of 17 MeV and approximately 5 nA, the chamber was irradiated under several conditions and different beam shapes were measured. These first beam tests showed that MATRIX was successfully designed and constructed, fulfilling all the design requirements. In March 2005, several tests have been performed on the clinical beams of the proton-therapy centre of the Loma Linda University Medical Center in California (USA). Figure 3 shows the results of the calibration procedure [4], based on three different measurements (reference, 90° rotation and one pixel shift). No assumption on the shape of the beam is used for the calibration. MATRIX has been installed in several treatment rooms equipped with horizontal beams or gantries and the full system could be dismounted and put in operation in about 30 minutes. This shows the system is highly reliable and user friendly.

Figure 3. Two-dimensional representation of the read-out of the MATRIX pixel chamber before (left) and after calibration (right) for a square shaped field.

681 5.00E+06 -, 4.00E+06

i

3,00E+06 2,00E*06

£**•

1,00E+06 O.OOE+00

0

2

4

6

8

10 12 14 16 18

Depth in water (cm) Figure 4. Two-dimensional map of a field shaped for a prostate cancer treatment (left). The Bragg peak measured for a 149 MeV proton beam (right).

The two-dimensional map of several fields has been measured. As an example, a field shaped for the treatment of a prostate cancer is shown in Figure 4 (left). The Bragg peak due to the ionization loss of a 149 MeV proton beam has been measured by interposing layers of water equivalent absorber between the exit window of the accelerator and the chamber, as presented in Figure 4 (right). Several tests have been performed using the read-out speed of 2 ms for the full chamber. This demonstrates that MATRIX can be successfully used for fast beam-monitoring in active dose distribution systems. The very positive results of all these beam tests show that MATRIX is an innovative detector that can be successfully used for on-line beam monitoring in hadrontherapy. Acknowledgments We gratefully acknowledge the Monzino Foundation (Milan, Italy) for the financial support of this project. The authors would like to thank U. Amaldi, President of the TERA Foundation, whose ideas are at the basis of MATRIX, G. Dellacasa and his team of the University of Western Piedmont in Alessandria (Italy) for the technical support given during the construction. We sincerely acknowledge the support for the beam tests of the research groups of the Joint Research Centre in Ispra and of the Loma Linda University Medical Center. References 1. 2. 3. 4.

Particle Therapy Co-Operative Group, http://ptcog.mgh.harvard.edu/ or http://ptcog.web.psi.ch/. See for example U. Amaldi and G. Kraft, Rep. Prog. Phys. 68 (2005) 1861. S. Amerio et al., Proceedings of the 7th ICATPP Conference, Como (Italy), 2001, M. Barone et al. Editors, World Scientific, p. 487. S. Amerio et al., Medical Physics, Vol. 31, No. 2, February 2004.

DEVELOPMENT OF A DUAL TRACER PET METHOD FOR IMAGING DOPAMINERGIC NEUROMODULATION ALEXANDER K. CONVERSE1, ONOFRE T. DEJESUS2, LEO G. FLORES2, JAMES E. HOLDEN2, ANN E. KELLEY3, JEFFREY M. MOIRANO2, ROBERT J. NICKLES2, TERRENCE R. OAKES1, ANDREW D. ROBERTS4, THOMAS J. RUTH5, NICHOLAS T. VANDEHEY2, RICHARD J. DAVIDSON3 'Waisman Center, Departments of2Medical Physics and 3Psychology, University of Wisconsin, Madison, WI53705, USA 4

Department of Physics, University of Manchester, UK TRIUMF, University of British Columbia, Vancouver, BC CANADA

5

The modulatory neurotransmitter dopamine (DA) is involved in movement and reward behaviors, and malfunctions in the dopamine system are implicated in a variety of prevalent and debilitating pathologies including Parkinson's disease, attention deficit/hyperactivity disorder, schizophrenia, and addiction. Positron emission tomography (PET) has been used to separately measure changes in DA receptor occupancy and blood flow in response to various interventions. Here we describe a dual tracer PET method to simultaneously measure both responses with the aim of comparing DA release in particular areas of the brain and associated alterations in neural activity throughout the brain. Significant correlations between reductions in DA receptor occupancy and blood flow alterations would be potential signs of dopaminergic modulation, i.e. modifications in signal processing due to increased levels of extracellular DA. Methodological development has begun with rats undergoing an amphetamine challenge while being scanned with the blood flow tracer [17F]fluoromethane and the dopamine D2 receptor tracer [18F]desmethoxyfallypride. 1. Dopaminergic Neuromodulation Dopaminergic neuromodulation is the process in which dopamine released into the extracellular space modifies neurotransmssion at downstream synapses. It plays an important role in brain function and its malfunction is implicated in a number of pathologies. 1.1. The Dopamine System Dopamine (DA) is one of several modulatory neurotransmitters in the brain along with serotonin, norepinephrine, acetylcholine, and others. Dopamine neuron cell bodies are concentrated in two midbrain regions, the substantia nigra (SN) and the ventral tegmental area (VTA). In the nigrostriatal circuit, cells in SN project to striatum, and in the mesolimbic circuit, cells in VTA project to

682

683 limbic areas including nucleus accumbens (NAcc) and prefrontal cortex (PFC). The nigrostriatal circuit is involved in movement while the mesolimbic circuit is associated with reward processing and attention. 1.2. Modulation of Neurotransmission The majority of signalling events in the brain involve point to point activation of fast exitatory or inhibitory ionotropic receptors by glutamate or GABA. These synaptic connections may be suppressed or- indirectly enhanced by the release over relatively large volumes of modulatory neurotransmitters such as dopamine, which act on slow presynaptic metabotropic receptors'. These processes are studied in animals with invasive procedures including stimulation and recording electrodes to follow electrical activity, as well as forward and reverse microdialysis to observe extracellular concentrations of neurotransmitters or to inject various agonists and antagonists. 2. Simultaneous Measures To understand dopaminergic neuromodulation at the systems level in the brain, it is necessary to observe both the release of dopamine and related alterations in neural activity. Even with precise control of the experimental conditions, there will be some variability in the physiological state of the brain between measurements, and it is therefore best to make these observations simultaneously. This is especially true when studying the neural basis of human behavior where within subject variability is expected to be larger than in anesthetized animals. Except for rare cases where invasive techniques are practical in neurological patients, an imaging method is required that permits simultaneous measures of dopamine release and neural activity. Radiotracer techniques can be used to observe dopamine release by measuring the change in binding of receptor ligands due to competition from the endogenous neurotransmitter, in this case dopamine2. There are a number of possibilities for extending these PET and SPECT methods to obtain simultaneous measures of dopamine release and the related changes in brain activity. 2.1. PET/EEG One possibility for studying dopaminergic neuromodulation is the combination of PET and electroencephalography (EEG). This technique was explored as early as 19843. Published studies involve EEG measurement duing uptake of FDG outside the scanner with subsequent PET scanning, which reflects glucose

684

metabolism during the uptake period4. It might be feasible to perform EEG within the PET scanner while observing a dopamine receptor tracer. EEG provides good temporal resolution, however it suffers from poor spatial resolution in subcortical regions. 2.2. PET/MRI Combining PET with magnetic resonance imaging (MRI) would also permit simultaneous measurement of dopamine release and neural activity as indicated by changes in the blood oxygen level dependent (BOLD) signal. A small PET ring has been successfully operated inside an MR scanner with light from the scintillators carried by fiber optics to photomultiplier tubes placed at a distance from the magnetic field of the MR5. Work is also underway to read out the scintillators with avalanche photodiodes (APD), which are insensitive to the magnetic field6. 2.3. Dual Tracer SPECT In single photon nuclear imaging techniques, tracers may be distinguished based on the energy of the emitted gamma. This was demonstrated in 1967 by simultaneously scanning patients with different combinations of tracers, for example visualizing lung with 131I albumin macroaggregate and liver with 198Au colloidal gold7. Recent work with simulations and phantoms suggests the feasibility of using dual tracer single photon emission computed tomography (SPECT) for simultaneous imaging of dopamine receptor binding and blood flow8. 3. Dual Tracer PET The present work involves a dual tracer PET method. In PET the annihilation photons are essentially monoenergetic at 511 keV regardless of the radioisotope, so there is no information directly available to the scanner that permits tracers to be distinguished from one another. However, it is possible to use the measured time course to distinguish tracers based on their physical half lives and their pharmacokinetics. 3.1. Distinguishing Half Life In 1982 Huang et al. demonstrated with phantoms that 13N (ti/2 = 10 min) and 18F (ti/2 = 110 min) could be distinguished using PET based on their half lives9. In this case it was possible to parameterize the sinograms with a sum of

685

exponentials and reconstruct the F and N images separately. However, because this technique requires the tracer spatial distribution to be constant over time it would not work in vivo except in limited cases. 3.2. Distinguishing Pharmacokinetics Because of different uptake and clearance rates for the tagged molecules, it is possible to distinguish tracers. Koeppe et al. demonstrated this concept in humans with staggered paired injections of the benzodiazepine receptor agonist ["CJflumazenil (FMZ), the acetylcholinesterase substrate N[uC]methylpiperidinyl propionate (PMP), and the vesicular monamine transporter ligand [uC]dihydrotetrabenazine (DTBZ)10. With injections staggered by as little as 10-20 minutes, it was possible to measure relevant pharmacokinetic parameters with accuracy similar to that obtained in single injection scans for FMZ.DTBZ and DTBZ:FMZ as well as FMZ:PMP and PMP.FMZ. 3.3. Modulated Delivery As a proof of principle for the simultaneous dual tracer technique described here, we performed baseline studies in rhesus monkeys11. Two monkeys were anesthetized with isoflurane, and positioned in the UW ECAT 933/04 tomograph with 6 mm FWHM resolution (CTI, Knoxville, TN). [18F]fallypride (tI/2 =110 min), a dopamine D2 receptor antagonist, was injected (5.1 and 2.1 mCi). Starring 99 - 137 min after injection, 10 second PET images were acquired while the blood flow tracer [17F]fluoromethane (t1/2 = 65 s) was administered by inhalation (0.46 mCi/s in 2 mL/s neon carrier) in a repeating boxcar pattern of 45 s on / 45 s off for twelve minutes ("modulated delivery"). Images were reconstructed, and eight 2.0 mL regions of interest (ROI) were specified. The observed 12 minute time-activity curves were split into 6 minute segments and independently fit with a three compartment lung - body - brain model of fluoromethane kinetics with whole brain perfusion fixed (Fig. 1). Repeatability of the dual tracer measures between one pair of consecutive 6 min scans is shown in Fig. 2. Averaging over both monkeys and all scans, the mean absolute variation in rCBF between consecutive 6 min scans in all 8 ROIs was 8.9% ± 7.3% (SD, n=32). The mean absolute variation in [18F]fallypride concentration in the 2 striatal ROIs between consecutive scans was 7.9% ± 7.4% (SD, n=8). Sensitivity of measured rCBF to 20% variation in each of the assumed model parameters was shown to be <0.3%, except for the fixed value of whole brain perfusion, which was effectively a scale factor.

686

•

i

i i

s

I

I

6 a

200 100

a 5

I

&3°

200

5 IT

5 3

I 1

? £

b

f 8

LSlr

S 100

a

RStr

5

150 100

Time (s)

Figure 1. (left) Representative TACs from a 6-minute scan segment (10 second resolution) during modulated delivery of [17F]fluoromethane, 117 min post-injection of [18F]fallypride. (wb) whole brain TAC. (cM) modeled [17F]fluoromethane cerebral arterial input function from fit to wb TAC. (LFr, LStr) TACs for 2 mL ROIs fit using cra(t). • observed radioactivity. — modeled total radioactivity. — modeled [18F]fallypride radioactivity. Figure 2. (right) Observed rCBF (a) and [18F]fallypride concentration (b) from two consecutive 6minute scans during modulated delivery of [17F]fluoromethane, O 117 min and D 123 min postinjection of [18F]fallypride. Whole brain perfusion fixed at 0.48 mL/min/mL.

While these results suggested that the concept is feasible, it should be noted that the methods described below differ in a number of respects: (1) Blood sampling will be used to better determine the tracer input functions; (2) a pharmacokinetic model of the neurochemical tracer will be applied; (3) separate boluses of the blood flow tracer will be given making the analysis more precise and reducing the radiation dose to the subject. 4. Methods The goal of the current work is to optimize a dual tracer PET protocol for measuring dopaminergic neuromodulation. Exploratory/developmental work is being carried out in anesthetized animals responding to a pharmacological challenge. A short lived blood flow tracer is administereed in the presence of a long lived dopamine receptor ligand. The measures of interest will be ArCBF, the change in blood flow, and ABP, the change in receptor binding. 4.1. Subjects Sprague Dawley male rats (n=8, weight = 380 ± 42g, mean ± SD) were anesthetized with isoflurane (1.5%), and 24g catheters were placed in lateral tail veins for i.v. injection of radiotracers and amphetamine. Four animals at a time were positioned in the scanner with their heads fixed by ear and tooth bars.

687

4.2. Scanner Scanning sessions were performed on the UW Concorde microPET P4 with 8 cm axial x 19 cm transaxial FOV, 2x2x2 mm3 reconstructed image resolution, and 2% sensitivity at center12. Following a 10 min transmission scan, emission data were collected for 150 minutes in 3D list mode. 4.3. Radiotracers The blood flow tracer [17F]fluoromethane (FM) for injection was prepared using an in-line gas phase method developed in our lab13. The intermediate affinity dopamine D2 receptor ligand [18F]desmethoxyfallypride (DMFP), (S)-N-[(lallyl-2-pyrrolidinyl)methyl]-5-(3 -F-18-fluoropropyl)-2-methoxy benzamide, was produced using published methods14. 4.4. Pharmacokinetic Model A dual tracer pharmacokinetic model will be employed to analyze the data (Fig. 3). We combine a two compartment ligand model with a single compartment flow tracer model, each of which is based on standard methods15. For a given ROI, blood flow and DMFP binding are assumed to respond instantaneously to a stimulus (amphetamine), i.e., k , —> k i + Ak\ (k2 = k {) and k 3 —> k 3 Ak'3. Modelled TACs are generated iteratively on a 1 s grid. K

^t

ligand

*4

flow

C'„

PR

plasma

a tissue (free)

tissue (bound)

Figure 3. Dual tracer pharmacokinetic model showing compartments for ligand and flow tracer in plasma (c'p and cfp), free and bound ligand in tissue (C'F and C'B), and flow tracer in tissue (cf).

Preliminary simulations of TACs generated using this model are shown in Fig. 4. In these simulations, an approximate [uC]raclopride infusion input function is assumed, c'p(t) = e'Ut, where X1 is the "C radioactive decay rate, ln(2)/20.4 min. Input functions for bolus injections of [17F]fluoromethane at 900, 1800, 3600, and 4500 s are simulated with cfp(t') = e"Xff x sXet, where t' is the time relative to the fluoromethane injection, Xf is the 17F radioactive decay rate, ln(2)/64.5 s, and X.e is an approximate excretion rate, ln(2)/60 s. In the

688

simulation, rate constants were taken from the literature for human striatum: k\ = 0.15 min"1, k'2 = 0.37 min 1 , k'3 = 0.51 min"1, and k'4 = 0.14 min'116.

|K 0

lifwii.il.N. 1000

2000 3000 time (»)

I,>j,.„„,\.i^ 0 l ) ri 4000

SOOO

0

i 1000

2000

i

i

3000 limo (s)

1000

J1..E, . i . . . . i . . . . i . . . . i . . . . i .. .; 6000

3000

3500

4000 4500 lime (s|

5000

Figure 4. Compartmental model simulations of a dual tracer scan, a Contributions to the observed TAC from the flow tracer, bound ligand, and free ligand, with 20% reduction in ligand binding and 20% increase in blood flow at 2700 s. b Simulation of observed TAC with no response, 20% increase in blood flow, or 20% decrease in binding, c detail of post-stimulus observed TAC (27005400s).

5. Preliminary Results First results were obtained in October 2005 with combined FM and DMFP imaging of rats during a d-amphetamine (AMPH) challenge (0.4 mg/kg i.v.). Eight rats were scanned with various combinations of FM x DMFP x AMPH. The PET images was coregistered with template ROIs based on histology17. The resulting TACs for one subject are shown in Fig. 5.

0 18

2000

4000 6000 time (s) l7

8000

Figure 5. Dual tracer scan of a rat with [ F]desmethoxyfallypride (DMFP) and [ F]fluoromefhane (FM). Coronal and sagittal views of (a) brain atlas, (b) regions of interest, and (c) PET image (2000 - 9000 s) with striatal ROI outlined, (d) Corresponding time activity curves at 60 s resolution. The bolus injection of DMFP (1.1 mCi) is seen at 2700 s. Six bolus injections of FM are seen, two prior to DMFP injection (0.3-0.5 mCi) and four after (0.4-1.4 mCi). d-amphetamine (0.4 mg/kg i.v.) was administered at 6850 s to elicit dopamine release. Cb = cerebellum, Ctx = cortex, Str = striatum, Th = thalamus.

689 6. Planned Experiments 6.1. Schedule We expect to carry out this project on the following schedule: Sep 2005 - Feb 2006: methods development in rats, Mar 2006 - Aug 2006: methods development in rhesus, Sep 2006 - Feb 2007: neuromodulation study in rhesus, Mar 2007 - Aug 2007: simulations of human studies. 6.2. Arterial Blood Sampling Arterial catheters will be placed by a surgical procedure18. Arterial blood will be sampled at 0.5 uL/s throughout and 1.3 uL/s for 2 min following injection of each [l7F]fluoromethane bolus, yielding 150 uL per 5 minute sample and total withdrawl of 3.5 mL. Blood will be analyzed on-line and in samples (Fig. 6). j

arterial line

i In-line detector

I j

samples

]

well counter

|

i

1

chromatography

1 , 1 1 d> 1 ',

<\* = < * • <*«.

Figure 6. Separation of flow tracer and neuroligand input functions using arterial sampling. The inline detector yields the whole blood TAC, c ^ t ) = c ^ t ) + c ^ t ) . Well counter measures of samples decayed by > 10 minutes yield the ligand whole blood TAC, c'wb(t), so the flow tracer input function is given by c p(t) = c ^(t) = c ^ t ) - c ^(t). Chromatography measures of plasma give the unmetabolized parent fraction of neuroligand, fparent(t), and the input function is then c'p(t) = fparent(0 x c'wbCt).

6.3. Fitting Algorithm Based on the model presented in section 4.4, parameters will be determined by least squares fitting to observed data. The parameters of interest are Akf1; i.e. the change in blood flow, and Ak'3, i.e. the change in DA receptor ligand binding. Statistical errors will be calculated by bootstrap and covariance between parameters will be determined. 6.4. Amphetamine Challenge in Rhesus Five monkeys will each undergo 5 scanning sessions using the optimum protocol determined in pilot scans. In 3 sessions, the dual tracer method will be

690

used to correlate dopamine release and rCBF alteration in response to 0, 0.2, and 0.4 mg/kg amphetamine. In 2 additional sessions, 0.4 mg/kg amphetamine challenges will be delivered to each subject while imaging with [17F]fluoromethane and [18F]DMFP alone. These sequential results will be compared to the simultaneously measured response to 0.4 mg/kg amphetamine. 7. Dreamed Experiments If this dual tracer PET method proves feasible, it might be applied in a variety of studies. For example, it would be interesting to use other pharmacological challenges such as methylphenidate. Other modulatory neurotransmitters, e.g. serotonin, may be amenable to such techniques19. The ultimate goal of this development work is to use this technique in studies of behaving humans, for instance to probe the role of dopaminergic neuromodulation in attention. Acknowledgments Funding was provided by the United States National Institutes of Health, NIH NIBIB R21 EB004482. We are grateful to Todd Bamhart for tracer development, to Dhanabalan Murali for [I8F]DMFP synthesis and to Mary Schneider and Julie Larson for collaboration on rhesus studies. References 1.

2.

3.

4.

5.

6.

Hasselmo,M.E. Neuromodulation and cortical function: modeling the physiological basis of behavior. Behavioural Brain Research 67, 1-27 (1995). Laruelle,M. & Huang.Y. Vulnerability of positron emission tomography radiotracers to endogenous competition: new insights. Quarterly Journal of Nuclear Medicine AS, 124-138 (2001). BuchsbaumM.S., Cappelletti,J.D., JohnsonJ.L., Kessler,R. & King,C. Simultaneous topographic EEG and glucose use by PET. Electroencephalography and Clinical Neurophysiology 58, 79 (1984). Oakes.T.R. et al. Functional coupling of simultaneous electrical and metabolic activity in the human brain. Hum. Brain Mapp. 21, 257-270 (2004). Slates,R.B. et al. A study of artefacts in simultaneous PET and MR imaging using a prototype MR compatible PET scanner. Physics in Medicine and Biology 44, 2015-2027 (1999). Judenhofer,M.S. et al. Development of an APD Based PET Detector for a Simultaneous PET and 7 Tesla MRI Imaging [abstract]. IEEE Nuclear Science Symposium and Medical Imaging Conference (2005).

691 7. 8.

9.

10. 11.

12. 13.

14.

15.

16.

17.

18.

19.

Ben-Porath,M., Clayton,G. & Kaplan,E. Modification of a multi-isotope color scanner for multi-purpose scanning. J. Nucl. Med. 8, 411-425 (1967). El Fakhri,G., Moore,S.C, Maksud,P., Aurengo.A. & Kijewski.M.F. Absolute activity quantitation in simultaneous 123I/99mTc brain SPECT. J. Nucl. Med. 42, 300-308 (2001). Huang,S.C, Carson,R.E., Hoffman,E J., Kuhl,D.E. & Phelps,M.E. An investigation of a double-tracer technique for positron computerized tomography. J. Nucl. Med. 23, 816-822 (1982). Koeppe,R.A. et al. Dual-[' 'C]tracer single-acquisition positron emission tomography studies. J. Cereb. Blood Flow Metab. 21, 1480-1492 (2001). Converse,A.K. et al. PET Measurement of rCBF in the presence of a neurochemical tracer. Journal of Neuroscience Methods 132, 199-208 (2004). Tai,C. et al. Performance evaluation of the microPET P4: a PET system dedicated to animal imaging. Phys. Med. Biol. 46, 1845-1862 (2001). Barnhart,T.E. et al. Production of [17F]CH3F (tI/2=65 sec), an improved PET tracer for rCBF measurement. Journal of Applied Radiation and Isotopes 62, 525-532 (2005). Mukherjee,J., Yang,Z.Y., Brown,T., Roemer.J. & Cooper,M. Evaluation of d-amphetamine effects on dopamine D-2 receptor radiotracers, F-18 fallypride and F-18 desmethoxyfallypride. J. Nucl. Med. 35, 139-140 (1994). Mintun,M.A., Raichle,M.E., Kilbourn,M.R., Wooten,G.F. & Welch,M.J. A quantitative model for the in vivo assessment of drug binding sites with positron emission tomography. Annals of Neurology 15, 217-227 (1984). Farde,L., Eriksson,L., Blomquist.G. & Halldin.C. Kinetic analysis of central [nC]raclopride binding to D2-dopamine receptors studied by PET - a comparison to the equilibrium analysis. J. Cereb. Blood Flow Metab. 9, 696-708 (1989). Rubins,D.J. et al. Development and evaluation of an automated atlas-based image analysis method for microPET studies of the rat brain. Neuroimage 20,2100-2118(2003). Guo,Z.K. & Zhou,L.Z. Dual tail catheters for infusion and sampling in rats as an efficient platform for metabolic experiments. Lab Animal 32, 45-48 (2003). Giovacchini,G. et al. Differential effects of paroxetine on raphe and cortical 5-HT1A binding: A PET study in monkeys. Neuroimage 28, 238-248 (2005).

RETHINKING POSITRON EMISSION TECHNOLOGY FOR EARLY CANCER DETECTION D. B. CROSETTO 3D-Computing, Inc.900 Hideaway Place DeSoto, TX 75115, USA E-mail: Crosetto@3d-computing. com

Current PET technology, which detects cancer indirectly by measuring the size of tumors, is contrasted with a new technology that emphasizes measuring directly the abnormal metabolic activity characteristic of the very first cancerous cells by improving the efficiency of the medical imaging device hundreds of times. Rather than maximizing spatial resolution, this approach minimizes the abnormal metabolic threshold needed for early cancer diagnosis. A redesigned detector assembly takes full advantage of innovations in electronics to implement real-time algorithms that allow the use of economical crystal detectors. The tradeoffs inherent in choosing components, such as crystals, sensors, and electronics, for achieving the largest reduction in cancer deaths is discussed and justified. Theoretical calculations and prototype data are provided in references that demonstrate that all this can be achieved at a significantly lower radiation dosage and per exam cost to the patient.

1. Introduction Positron Emission Technology is a non-invasive imaging method to obtain quantitative biochemical and molecular information of abnormal physiological processes in the body. In order to make the best use of this technology, it is important • to accurately measure and count, within a given time, the maximum number of signals (pair of photons) related to the physiological functions of interest (HEMODYNAMICS: blood flow, blood volume, perfusion; METABOLISM: glucose, oxygen, amino acid, etc.; RECEPTOR FUNCTION: transporter/receptor of benzodiazepine, receptor of oppioid, estrogen, etc.; MOLECULAR MECHANISM: DNA synthesis, etc.); • to display the information in a meaningful way to the physician. Current PET and PET/CT devices, however, capture only 1 out of 10,000 signals emitted from the tracer, and do not accurately measure their characteristics. They focus on measuring the dimensions of already formed tumors, instead of accurately measuring abnormal metabolism. This misuse of

692

693

the principle of operation of positron emission technology does not provide the information on metabolism needed by the physician. Even if current PET is modified to provide photon counts within a given time, the information is meaningless because the efficiency is too low. 2. Rethinking Positron Emission Technology for early cancer detection In order to use Positron Emission Technology effectively for early cancer detection, a paradigm shift is necessary. To illustrate this, I am going to compare objectives, specifications and implementations of current PET, "Type A," and the new technology, "Type B." Type B makes use of innovations [1], [2] and has been implemented in a device called 3D-CBS [3]. The starting point for this paradigmatic shift must be the "Social Objective." For Type B, it is to reduce cancer deaths through screening, early diagnosis and removal of the very first cancer cells. For Type A, it is to measure the shrinkage of tumors in order to justify the use of drugs, but in most cases does not save lives (as confirmed by the unchanged death rate). In some cases, it prolongs life for a few months at a high cost in suffering. The specifications for the PET Type A and Type B (3D-CBS) follow from the social objectives. These specifications are compared in Table I. Table I. Specifications for PET of Type A and Type B

Type A prioritized for profit • •

• • • •

•

Low cost of manufacturing the machine is important for maximizing profit Optimized for measuring spatial resolution (to measure the shrinkage of a tumor for drug usage purposes) Low efficiency (for the same reason as above, only measuring rumor shrinkage) High radiation dose deemed acceptable as it is only for seriously ill patients Long examination times (more than one hour to acquire data from 140 cm) High cost of the examination (due to long exams times and high radiation dose) deemed acceptable as seriously ill patients or their insurance will pay. Competition could be sustained by playing on the following parameters: higher radiation dosage to the patient, scanning a smaller section of the body, recording fewer data which allow detection of tumors at an advanced stage only Short detector is less expensive but results in long exam times and low efficiency

Type B prioritized for the patient •

•

•

• •

• •

Low cost of the examination (achieved by lowering radiation dose, and examination time) necessary for repetitive exams on ill patients and annual exams on asymptomatic patients, and they or their insurance will pay Low radiation dose and short examination times (about 4 minutes) necessary to allow screening and repetitive exams on ill patients Very high efficiency (able to count more photons, detect the very first cancerous cells and preempt metastasis) Optimized for measuring sensitivity (same reason as above, preempting metastasis) Long detector is more expensive but results in short exam times and higher efficiency because it acquires many data points simultaneously from the entire body Higher cost of the machine is acceptable if it is sufficiently more efficient Short time to recoup investment (because more patients can be examined per day)

694 Although a Type B machine has a higher initial cost, calculations show it will be advantageous in operating costs and it will take a shorter time to recoup the investment because it has the capability to examine more patients per day (see more details on page 160-165 of reference [4]). Next, considering the constraints set by the objectives and the specifications, the details of the implementation can be defined. Table II describes these for Type A and B machines in terms of design implementation. Table II. Implementation of the specification for PET of Type A and Type B

Implementation of Type A PET Reuse as much of existing technology as possible to maximize profit. Satisfy the expectations of the current market. Use nearly ideal, although costly crystals, especially if your company has a patent on that crystal. Continue to use the current "block detector" assembly to keep development costs low (although limitations in efficiency of such assembly have been determined experimentally by Andreaco [5], twelve years ago). Continue to use electronics with limited performance, with no communication between neighboring electronic channels, to keep production costs low. Use a short detector since, for the current detector design, sensitivity increases linearly with detector length. Do not try to reduce radiation for the safety of the patient (to less than 100 mrem required for annual screening) because the current design using a short detector cannot capture oblique photons at large angle making it impractical for screening anyway. Display pretty pictures (as close as possible to fine-grained photos) for marketing purposes and for showing tumor shrinkage, but not able to show minimum abnormal metabolism. Increase sensitivity only to acquire the data more quickly for making pictures so more patients can be examined.

Implementation of Type B PET 1.

The implementation must be completely different from the current PET. 2. A new, different market is envisioned. 3. Find the most cost effective combination of low cost crystals, sophisticated electronics, and detector length that meets the safety and exam cost requirements for annual screening and for a safe, repetitive exam on ill patient. 4. Eliminate the "block detector" [6], assembly to allow hundreds of times greater efficiency improvements necessary to meet the objectives. 5. Build sophisticated electronics that enable more efficient use of low cost crystals to detect and accurately measure all characteristics of as many photons as possible. 6. Increase detector length since sensitivity increases exponentially with detector length for the 3D-CBS design. 7. Do what is necessary (longer detector, capturing oblique photons at greater angle and more accurately) to lower the radiation dose to that permitted for annual screening. 8. Display the information to the physician in a way that clearly shows abnormal metabolism. It must be easy to interpret, using color codes or symbols, but also provide precise numbers. 9.

Increase sensitivity to provide metabolic (nutrient flow rate) information necessary for early cancer detection. 10. Capture accurately 1 out of 25 photons instead of 1 out of 10,000 captured by current PET. Use a detector assembly with no boundary limitation, a 3x3 algorithm for better spatial and energy resolution and a high computing power on each channel for best DOI measurement.

695

The innovative technology described in more detail in references [1], [2], [3], [4], has been developed to build a cost-effective Type B PET that finds its realization in the 3D-CBS. The key innovations are in four main areas: 1. Increasing the length of the detector (Field of View) 2. Improvement and simplification of the detector assembly 3. Innovations in the electronics which enable other innovations in: a) executing precise algorithms for photon identification b) accurately measuring the impact point and energy of oblique photons c) reducing the initial number of the electronic channels, and d) simplifying the method for identifying events in time coincidence 4. Innovation in the visualization of the information obtained The synergy of coupling several innovations, such the detector, sensors and the electronic system, that enables execution of real-time algorithms for more accurate photon identification, allows a staggering increase in efficiency. Figure 1 shows the clustering (see right in figure) that can be obtained when a simplified/improved detector assembly is coupled to the programmable electronics called 3D-Flow [1]. This allows the extraction of all characteristics (energy, "x", "y", "z" location, etc.) of the incident photon with the detector, and enables using economical crystals and capturing more photons [2]. Current PET systems • Boundary limitation at the block and module segmentation.

3D-F1QW System - NO Boundary limitation, - NO duplication of events (local maxima)

'!**.,

_ (NW+W+SW) - (NE'+E+SE) m(3tf) NW+W+SW+NE+E+SE _ (SW+S+SEl-tNW+N+NEl »M SW+S+SE+NW+N+NE

x

Figure 1 - Comparison between the signals obtained by the 2x2 PMT assembly (left) in current PET and clustering made possible with the 3D-Flow architecture (right) which allows centering a "cluster" (or trigger channel) on any electronic channel of the detector.

3. Current prospects for future Type A PET development cannot significantly reduce cancer death with early cancer detection Why continue to build Type A PET when it misuses the principle of operation of positron emission technology and does not serve the need of the patient as its first priority? The specifications of current PET show sensitivity is

696 not considered important. Articles in scientific journals also confirm this trend for the future. Table III reports parameters extracted from the IEEE, 2003 article "NEMA NU2-2001 Guided Performance Evaluation of Four Siemens ECAT PET-Scanner" 0-7803-7, 2003, IEEE. Table III The trend in development of Type A PET shows sensitivity has gotten worse over time.

Siemens/CTI PET and PET/CT Models YEAR Model Detector Material: Crystal Crystal Dimensions (mm3) Number of Crystals Detector Ring Diameter (cm) Detector Length FOV (cm) Tangential Resolution (mm) Radial Resolution (mm) Axial Resolution (mm) Sensitivity (cps/MBq)

1992 EXACT BGO 6.75X6.75X20 9216 82.4 16.2 6.06 6.38 6.25 5982

1996 HR+ BGO 4.05X4.39X30 18432 82.4 15.5 4.81 4.97 5.22 6650

2003 2000 EMERGE ACCEL LSO LSO 6.45X6.45X25 6.45X6.45X25 9216 <5000 82.4 82.4 16.2 16.2 6.25 6.63 6.14 6.63 6.52 6.49 6362 2259

The patient is definitely not the priority if sensitivity is reduced from 6650 cps/MBq referred to 1996, to 6362 in the year 2000, to be even further reduced to 2259 cps/MBq in 2003. Over the past 10 years of submitting funding proposals for Type B PET, I have constantly been asked to modify my proposal towards building a Type A PET. Instead, my innovations enable the implementation of a PET of Type B, (or 3D-CBS) with increase sensitivity and emphasize directly measuring the abnormal metabolic activity characteristic of the very first cancerous cells by improving the efficiency of the medical imaging device hundreds of times, but it has been blocked for a decade. The book in reference [4] and the slides presented at the Conference, provide more details about my innovative technology, its proof of concept in hardware, the engineering of the concept into hardware for an industrial device. Acknowledgments I thank M. Barone, P.G. Rancoita and V. Sossi for accepting this paper at the ICATTP conference and K, Spielmann and M. Bentley for editing assistance. [1] D. Crosetto, Nucl. Ins. Met. in Phys. Sec. A, vol. 436, 341 (1999) [2] D. Crosetto,. 400+ times improved PET efficiency. ISBN 0-9702897-0-7. [3] D. Crosetto, The 3D-CBS, IEEE-NSS-MIC. Conf. Record. M7-129 (2003). [4] D. Crosetto, "Come Vincere il Cancro", Ed. Clavilux: www.clavilux.it or English version "How to Win the War on Cancer" (2005). [5] J.G. Rogers at al., IEEE. Conf. Rec NSS-MIC, vol. 3, 1837 (1993). [6] M. Casey and R. Nutt, IEEE TNS, NS-33, 760 (1996).

POSITRON EMISSION TOMOGRAPHY - TRACER KINETIC MODELLING IN DRUG DEVELOPMENT VINCENT J. CUNNINGHAM and ROGER N. GUNN Translational Medicine and Genetics, GlaxoSmithKline, 891-995 Greenford Greenford, Middlesex UB6 7HE, U.K

Road,

There is a growing interest in the application of imaging techniques to drug discovery and development to enable earlier decisions to be made on the suitability of a given drug for the treatment of disease in humans. PET has the ability to image pharmacological functions in vivo in conscious humans and is ideally suited to this task. Either the drug itself can be radiolabelled and its biodistribution in tissues measured directly, or alternatively, a given pharmacological target can be labelled, enabling quantification of exogenous drug interaction. A wide range of mathematical modelling techniques maybe applied to the spatial and kinetic PET data giving rise to functional images or regional parameter values that reflect the pharmacology of the drug under study. Examples will be shown and some of the mathematical techniques behind quantification will be discussed.

1. Introduction Positron Emission Tomography is ideally suited for pharmacological studies in vivo. The pharmaceuticals themselves can be radiolabelled, and also there is an increasingly large library of radioligands with high specificity and affinity for specific receptor sites, particularly neuroreceptor sites in the brain, enabling drug-receptor interactions to be characterized. Typically the radiotracer is administered intravenously, and the time course of the regional uptake and retention of radioactivity is followed over time at a spatial resolution of a few millimeters in specific tissues. Application of kinetic models to this dynamic data enables the quantitative estimation of pharmacologically and physiologically based parameters. These include the concentration and functional status of receptor sites, together with down stream effects such as second messenger status, blood flow, and metabolic rate, amongst an increasing range of applications. The value of these techniques from the point of view of drug discovery and development is increasingly apparent, as it enables studies to be carried out in man earlier in the development process, and enables measurements to be made closer to the putative site or mechanism of action of the drug, rather than relying on indirect surrogate measures. We will first briefly describe typical applications of PET in the drug development process and then

697

698

consider a particular class of kinetic models, namely compartmental models, which can be applied to PET data to extract relevant biological parameters. 2. Applications In addition to the study of underlying mechanisms of disease, there are currently two areas where PET is particularly useful and can have a direct impact on the drug development process - namely biodistribution and drug occupancy studies. 2.1. Biodistribution studies For biodistribution studies, the drug itself is radiolabelled and its concentration in blood together with its delivery, uptake and retention in specific regions of a tissue is followed over time. This may be carried out with trace (picomolar) quantities of the radiolabelled drug, in which case it is sometimes referred to as 'microdosing', with obvious advantages for first time in man studies. Alternatively, it may be more appropriate to carry out the tracer study at pharmacologically active doses of the cold drug. For example, if the drug is a substrate for one or more of the drug efflux transporters such as P-glycoprotein, multidrug resistance proteins or organic anion transporting polypeptides, then a lack of significant tissue penetration at trace doses does not preclude entry at higher concentrations. Conversely, if uptake of the radiotracer involves saturable processes then the fractional uptake of the radiotracer in a microdosing study will be higher than that of the cold drug at pharmacological doses. The tissue time course of a tracer following intravenous administration does not, of course, reflect the bioavailability following oral dosing. The PET data does however enable an impulse response function between plasma and tissue to be established under a given dosing regimen of the cold drug. Estimation of such response functions will be considered later. The response function can then be convolved with the time course of the cold drag in plasma following oral administration to complete the picture. 2.2. Drug Occupancy Studies A question which is frequently asked in drug development concerns the relationship between a given pharmacological dose of a drag, or its concentration in plasma, and the fractional occupancy by the drug of specific receptor sites. Considerable effort has therefore gone into the development of specific radioligands targeting receptor sites of interest. The ideal properties of such radioligands is a complex subject, which, in brain for example, needs to

699 take into account how the physico-chemical properties of the tracer relate to its blood profile, its metabolism and whole body disposition, its transport across the blood brain barrier, and the partitioning of the tracer between free, nonspecifically bound, and specifically bound states in the tissue. The principal behind drug occupancy studies relies simply on determining changes in the concentration of the specifically bound radioligand in the presence of the cold drug. In general the radioligand is a different chemical species than the cold drug, usually selected on the grounds of specificity and affinity for a given site, whereas the cold drug may have multiple sites of action. We will now consider the tracer kinetic compartmental models which form the basis for estimating the concentration of specific receptor sites in brain. 3. Compartmental Models in PET 3.1. Classic Compartmental models There are three 'classic' compartmental models which have formed the basis for the analysis and interpretation of kinetic PET data. The first of these is the Kety blood flow model [1]. This is a single tissue compartment model in which it is assumed that the radiotracer exists in a single homogeneous state in the tissue. The solution to this model involves a convolution of the blood input function with a single exponential term. If, furthermore, the radiotracer is completely extracted from the capillary bed to the tissue in a single blood transit, then the concentration of the radiotracer in the venous blood is in complete equilibrium with that in the tissue and the system gives an estimate of the rate of blood flow ((ml blood).min"1.(ml tissue)"1). This model has been applied to [150]-water studies in PET to give quantitative functional images of blood flow. The second model is a two tissue compartment model describing the behaviour of the glucose analogue [18F]-fluorodeoxyglucose, used widely for the quantification of metabolic rate for glucose [2], and which has had a major impact on PET development. In the original formulation by Sokoloff et al. [3], applied to ex vivo studies with deoxyglucose, the first tissue compartment accounts for the tracer exchanging reversibly with blood and leads to a second 'trapping' compartment. The model has a further term relating the kinetics of the analogue to those of native glucose, and allows the generation of quantitative functional images of the cerebral metabolic rate for glucose (umol.min~1.(ml tissue)"1). The third model is that described by Mintun et al. [4], which describes the behaviour of radioligands in terms of three tissue compartments - free,

700

nonspecifically bound, and specifically bound, and which forms the basis for the interpretation of many of the drug development studies (Figure 1. - For radioligands, the input function to the system is usually taken as the concentration of the parent radioligand in plasma). This work introduced the concept of 'binding potential' as applied to PET tracer radioligand studies, which may be defined as the ratio of the concentration of specifically bound ligand relative to that of the free ligand in the tissue which would exist under steady state equilibrium conditions. Its importance lies in the fact that this parameter can be shown to equal the ratio of the concentration of available receptor sites to the equilibrium dissociation constant of the radioligand, and may be estimated from nonsteady state tracer studies. It is changes in binding potential which can be used to measure drug occupancy.

6 NonSpecifically Bound

Figure 1. Compartmental description of the standard radioligand model.

3.2. Reference Tissues If there are regions of brain, for example, which have a negligible concentration of specific binding sites, then the problem of estimating binding potential, and hence drug occupancy, in target regions is greatly facilitated. If it can be assumed that the partition coefficient of the free and non-specifically bound fraction in target region relative to the radioligand in plasma, is the same as that for the reference region, then the problem reduces simply to determining and comparing the total partition coefficients for each region, which, as is discussed below, is a relatively robust parameter which does not require numerical identification of the individual rate constants defining the binding potential. A further advantage of the existence of a reference region is the possibility of

701 avoiding arterial blood sampling which is necessary to define a suitable plasma input function, and expressing target kinetics directly as a function of reference kinetics. Drug occupancy can of course still be estimated in the absence of a suitable reference tissue from the total partition coefficient by examining the system over a range of cold drug concentrations and fitting the data to a competitive binding model which includes a component for the nonspecific fraction. 3.3.

Micro-andMacroparameters

It is usually the case that the individual rate constants in the above models, which describe the exchange of the radiotracers between blood/plasma and the tissue, and between the individual compartments, are not identifiable from typical dynamic PET data. It may also be the case that when individual compartments are distinguishable in a kinetic analysis, then these apparent compartments do not correspond to their ideal counterparts in the original model. This is not surprising. In the first case the signal to noise characteristics of PET limit the numerical identifiability of the system. In the second case, one might not expect the complex behaviour of a biological system to condense neatly into a finite number of components exactly matching one's preconceived ideas. However this does not diminish the usefulness of the compartmental models. Whereas the interpretation of individual rate constants in the model (microparameters) is dependent on the particular configuration and number of compartments, there are several model parameters (macroparameters) which are independent of the particular topology. For fully reversible tracer compartmental models these are the initial unidirectional clearance of the radiotracer from plasma (or blood) to tissue (e.g. Ki in Figure 1), and the equilibrium partition coefficient between total tissue tracer concentration and plasma tracer concentration, or 'Total Volume of Distribution' (VDtot, (ml plasma)/ (ml tissue)) as it is often referred to in the PET literature. As noted above, the binding Potential used to measure drug occupancy can be calculated from the VDtot in reference and target regions. For irreversible systems, there is K^ and a parameter 'The combined forward rate constant' or K[ which describes the net rate of irreversible transfer of tracer from blood to tissue, and is the key measure in metabolic studies using glucose analogues. For the practical estimation of these macroparameters, we now consider the general compartmental model for PET tracer studies.

702

4. The General Compartmental Model As has been emphasized above, compartmental models play an important role in PET tracer kinetic analysis. They give rise to systems of first order linear differential equations, parameterized by a set of rate constants which can be interpreted in terms of the underlying physiology and pharmacology. In practice however, there can be problems with the numerical identifiability of individual rate constants and the identification of the appropriate compartmental structure. These problems can be addressed in terms of a general compartmental model, and by the definition of macroparameters in this general framework. Thus, it can be shown [5] for any reversible system that, if Cj(t) and CA(t) are the time courses of the concentrations of radioactivity in tissue and plasma and n is the total number of tissue compartments, then -

CT(t)

= IRF(t)®CA(t)

(1)

Where,

^e-9-'

IRF(t) = '•-•

(2)

And,

Kx = 5 > , VDtot = £ f M

„

i=l

' •

(3)

Thus the tissue impulse response function, (IRF(t)), consists of the convolution of a sum of exponential terms with the input function (either a blood or plasma parent input function, or a reference tissue input in a more general case), and the macroparameters are simple functions of these exponential terms. 4.1. Fitting the General Compartmental Model to PET Kinetic Data The general model leads, in turn, to methods for the estimation of model order and the value of the macroparameters themselves. At the voxel level, the problem is to fit the many thousands of time activity curves which comprise a 4D dynamic PET data set. Pre-formation of a library of basis functions consisting of exponential terms convolved with the input function greatly facilitates this fitting problem. Typical PET data may consist of about 30 frames of data and in practice a sparse selection from a set of 100 basis functions, covering the expected kinetic

703

range of the tissue response, gives a suitable description of the data. Given the ill -posed nature of this basis selection problem the addition of constraints to the standard least squares metric is required for solution. One approach involves the incorporation of non-negativity constraints on the coefficients in equation (2) which has been shown to be consistent with PET compartmental systems using a plasma input function. [6, 7] A more general approach, which does not involve a non-negativity constraint, and which can therefore also be used to fit general reference tissue models, is that of Basis Pursuit de-noising. This involves the addition of a onenorm penalty term to inhibit the entry of extra basis functions into the solution and its application to PET data is described in detail in Gunn et al [8]. 5. 'Slow' and irreversible behaviour The general compartmental model framework covers not only the behaviour of reversible tracers exchanging with plasma, but also the case when there is effectively a net irreversible accumulation of the tracer in the tissue over the time course of the PET scan. In this case there are corresponding expressions for the macroparameters K1; clearance, and KI; the combined forward rate constant [5]. However it is important to emphasise that K] refers to the combined effects of flow, transport from plasma to tissue, and trapping in the tissue. This parameter is the measure of choice for quantifying the behaviour glucose analogues for example, in metabolic studies. Here it is the net rate of uptake of the radiotracer from blood or plasma which is of interest. For radioligands however, we require a binding parameter independent of flow and transport effects. In the standard model (Figure 1) this would correspond to k3 with 1Q close to or equal to zero. Thus in this case we are forced back to estimation of individual rate constants to obtain biological useful parameters. New methodology is therefore being developed [9], which aims to improve the identifiability of individual rate constants using hierarchical models which fit all tissue regions simultaneously, and which distinguish locally varying parameters, e.g. K.! and k3, from parameters which are common to all regions such as the size of the free and nonspecific compartments. 6. Summary The strengths of PET as a tool in drug discovery and development lie, on the one hand, in the wide range of radiotracers and radioligands which can be used to probe molecular binding sites. On the other hand is the fact that the technique is quantitative and that biologically based kinetic models can be used to derive

704

relevant physiological and pharmacological parameters describing drug delivery, interactions with specific sites, and downstream metabolic and physiological effects. Here we have considered the role that compartmental modeling plays in this process of quantification and how these models relate the observed kinetic behaviour to the underlying biology. References 1. S.S. Kety, Pharmacol. Rev. 3, 1 (1951). 2. M.E. Phelps, S.C. Huang, E.J. Hoffman, C. Selin, L. Sokoloff and D.E. Kuhl, Ann. Neurol. 6, 371 (1979). 3. L. Sokoloff, M. Reivich, C. Kennedy, M.H. DesRosiers, C.S. Patlak, K.D. Pettigrew, O. Sakurada and M. Shinohara, J. Neurochem. 28, 897 (1977). 4. M.A. Mintun, M.E. Raichle, MR. Kilburn, G.F. Wooten and M.J. Welch, Ann. Neurol.. 15, 217 (1984). 5. R.N. Gunn, S.R. Gunn and V.J. Cunningham, J. Cereb. Blood Flow Metab. 21, 635 (2001). 6. K. Schmidt, J. Cereb. Blood Flow Metab. 19, 560 (1999). 7. V.J. Cunningham and T.Jones, , J. Cereb. Blood Flow Metab. 13, 15 (1993). 8. R.N. Gunn, S.R. Gunn, F.E. Turkheimer, J.A.D. Aston and V.J. Cunningham, J. Cereb. Blood Flow Metab. 22, 1425 (2002). 9. V.J. Cunningham, J.C. Matthews, R.N. Gunn, E.A. Rabiner and A.D. Gee, Neurolmage, 22, suppl. 2 , T13 (2004).

3D-RECONSTRUCTION OF ABSORBED DOSE OBTAINED FROM GEL-DOSIMETER-LAYERS * G. GAMBARINlt, M. CARRARA, M. VALENTE Department of Physic of the University andlNFN, via Celoria 16 20133 Milan, Italy An experimental method for obtaining images and volume reconstruction of in-phantom absorbed dose is described. The method utilises layers of a tissue-equivalent gel matrix in which a proper chemical dosimeter has been incorporated (gel dosimeter). From the images of visible light transmittance, detected with a CCD camera before and after exposure, suitably developed software gives dose images and 3D representations.

1. Introduction The progression in radiotherapy techniques, aimed at achieving local energy release in tumours saving the surrounding healthy tissue, has caused the necessity of getting the images of in-phantom absorbed dose, especially when different radiation fields are adopted, achieving high precision and the most reliable resolution. To this end, nowadays both dosimetry methods and computational modalities are suitably developed and improved, with the purpose of achieving high accuracy and reliability. Also if computerised modelling has achieved considerable level, the obtained reliability is not always certain, in particular for exposure configurations far from the standard ones. Therefore it is convenient to perform experimental validations. Various techniques for dose imaging have been developed during last years, such as those employing radiochromic or radiographic films. A dosimetric approach, still in progress and showing great development, is gel dosimetry, which is based on tissue-equivalent gels in which a chemical dosimeter is infused. Phantoms made up of such gels act as continuous dosimeters and consequently the absorbed dose distribution can be imaged along profiles, surfaces or phantoms' volumes. Different kinds of gel dosimeters have been proposed by the scientific community, and different kinds of imaging techniques have been developed. * This work was partially supported by INFN (Italy).

705

706

The two main families of gel dosimeters are represented by the Fricke-infused and the polymer-acrylamide gels. In the first system, ionising radiation causes conversion of ferrous ion to ferric ions with oxidation yield proportional to the absorbed dose. Owing to ferric ion diffusion in the aqueous matrix, it is necessary to perform dosimeter analysis at a short time after irradiation. In polymer gels, ionising radiation produces polymerisation, and no diffusion effect takes place. Dose images, initially deduced from magnetic resonance images [1,2] are obtained by measuring visible light absorbance [3]. Fricke-gel has been employed in this work, in spite of ion diffusion, because of the best reliability of the results obtained with such a system. The apparatus for detecting light-transmittance images is transportable, so that it is possible to perform acquisition just before and after phantom exposure [4,5]. In such a way, diffusion effects do not affect images. 2. Method for dose imaging 2.1. Gel dosimeters and analysis technique The proposed method based on gel dosimeters' layers has been recently improved with the aim of attaining reliable 2D or 3D dose distributions. The here considered gel dosimeters are radiochromic Fricke gels (FrickeXylenol-Orange-Infused gels) whose standard composition is: Agarose [C|2H1405(0H)4] in the amount of 1% of the final weight, ferrous sulphate solution [1 mM FeCNH^CSO^^HzO], sulphuric acid [25 mM H 2 S0 4 ] and Xylenol-Orange [0.165 mM C3iH27N2Na5Oi3S]. The tissue equivalence of this dosimeter has resulted to be very good. In fact, the effective atomic number for a photoelectric effect is equal to 7.70 (in the muscle it is 7.71) and similar consistency was found for the other parameters. In the proposed method, gel dosimeters are in form of layers of convenient shape and thickness (1-3 mm), and the various layers can be piled up to compose a phantom. For the analysis, gel layers are placed on a plane light source near a strip of grey-level (GL) standards. Optical transmittance images are detected by means of a CCD camera provided with optical filter around

585nm. The grey-level strip allows to control, and eventually amend, the CCD linearity as well as the instability in time of the light source intensity. Optical transmittance imaging is performed just before and about 30-40 minutes after exposure. This time has resulted to be necessary for the dosimeters to reach the chemical equilibrium after irradiation.

707

The absorbed dose (D) results to be linearly correlated to the difference of optical absorbance (AA) that is to the difference of optical density (A(OD)) of the transmittance images. A(OD) is promptly obtained from the Gray Level (AGL) values evaluated before and after irradiation:

DocA4ocA(QD') = log10

^GLbefore &GLafter

j

(1)

The minimum detectable dose has resulted to be between 0.5 Gy and 1 Gy and the linearity interval is up to about 20 Gy or 30 Gy. Both sensitivity and linearity intervals depend on the dosimeter composition and can be slightly adjusted according to the radiation field of interest; the most immediate mean is to change the amount of Xylenol Orange in the dosimeter composition. A wider interval of linearity is obtained with lower dosimeter sensitivity and vice versa. 2.2. Suitably developed software A dedicated software (SW) has been developed in order to obtain an interactive rendering of dose profiles, surfaces and volumes, as well as isodose curves, starting from the transmittance gel-layers' images acquired before and after irradiation. Volumetric dose distribution is obtained semi-automatically by this SW, applying the following steps: i) registration of the non-irradiated and irradiated gel-layers by means of automatic recognition of two reference points located at the vertical side of each dosimeters' frame; ii) automatic recognition and amendment of eventual small defects in the gel structure, like little bubbles or dust grains; iii) noise removal with adequate low-pass filters; iv) determination of the superficial dose distribution by means of (1); v) determination of the 3D dose distribution by superimposition and linear interpolation of the values obtained in iv). 3. Results Six gel-layers have been closely inserted in a tissue-equivalent phantom and exposed to a 18MV Linac X-ray beam (Varian Clinac 2100c) with a sourcesurface distance of 100cm and a field size of 3x2cm2. The prescribed dose at the build-up of the field was 18Gy. Absorbed dose profiles detected by the gellayers have been evaluated and compared with those calculated with a clinical treatment planning system (Prowess 3D) and by means of Monte Carlo (MC) simulation (Penelope), in order to check the reliability of the dosimeters' readings (see figure 1). The results show a good agreement between such profiles.

708

A further confirmation of the method's reliability is obtained by plotting the entire superficial dose distribution evaluated on one of the central dosimeters' layers (see figure 2). 110.00 T X,T

100.00

T

90.00

Jsl

80.00 70.00 60.00 50.00

*- I^Miifcj

•4$

.__4 •

• TPS OMC « Fricke

•

40.00 ~a 0.00

,

,

,

,

2.00

4.00

6.00 depth (cm)

8.00

10.00

-

12.00

Figure 1. Percentage depth-dose (PDD) profiles measured with a Fricke gel-layer, calculated with Penelope code and obtained with a clinical treatment planning system (Prowess 3D).

Figure 2. Superficial dose distribution measured in the white-bordered region selected inside the Fricke-gel-layer's image shown in the figure.

709

Furthermore, 3D absorbed dose distribution has been reconstructed; in figure 3 the 95%., 80% and 40% percentage isodose surfaces are represented. This last figure, as well, demonstrates that the proposed method allows reliable 3D imaging of absorbed dose with good spatial resolution.

Figure 3. 3D absorbed dose distribution; 95% (===), 80% (===) and 40% (===) percentage isodose surfaces are represented.

References 1. J. C. Gore, Y. S. Kang, and R.J. Schulz, Phys. Med. Biol. 29, 1189 (1984). 2. R. J. Schulz, A. deGuzman, D. Nguyen, and J. C. Gore, Phys. Med. Biol. 35,1611(1990). 3. A. Appleby, and A. Leghrouz, Med. Phys. 18, 309 (1991). 4. G. Gambarini, G. Gomarasca, R. Marchesini, A. Pecci, L. Pirola and S. Tomatis, Nucl. Inst.r andMeth. A 422, 643 (1999). 5. G. Gambarini, C. Birattari M. Mariani, R. Marchesini, L. Pirola, P. Prestini, M. Sella and S.Tomatis, Nucl. Instr. and Meth. B 213, 321 (2004).

ACCURATE DETERMINATION OF RADIONUCLIDE PURITY AND HALF-LIFE OF REACTOR PRODUCED LU-177G FOR METABOLIC RADIOIMMUNOTHERAPY F. GROPPI, L. CANELLA, M.L. BONARDI, C. ZONA, S. MORZENTI LASA, Universita degli Studi di Milano e INFN, via F.lli Cervi 201 1-20090 Milano, Italy E. MENAPACE ENE'A, Division for Advanced Physics Tecnologies, via Don Fiammelli 2, 1-40128 Bologna, Italy Z.B. ALFASSI Department

of Nuclear Engineering Ben Gurion University of the Negev, 11-84105 Be 'er Sheva, Israel M. CHINOL, S. PAPI, G. TOSI

Istituto Europeo di Oncologia — IEO, via Ripamonti 1-20141 Milano, Italy

435,

The accurate determination of radionuclidic purity and half-life of the beta emitter 177gLu used for metabolic radioimmunotherapy is presented. High-resolution gamma spectrometry, as well as deconvolution of beta decay curve measured by spectrometry with liquid scintillation counting, have been adopted for quality control of different samples available on the market. A simple method was developed to distinguish between the different methods available for production of 177gLu: i.e. either direct (n,y) reactions of enriched 176Lu or indirect (n,y) activation of enriched 176Yb followed by P" decay. In the first case, the long-lived metastable level 177mLu is present in the radioactive preparation and a low specific activity radionuclide is obtained, in the latter a very high purity and high specific activity 177gLu is produced.

1. Introduction Thanks to its favorable decay characteristics (half-life of the order of few days, relative low energy of negatrons and gamma emission suitable for detection), ,77gLu (ti/2 = 6.734 d, (T emission 100%, Ep_jmax= 489.3 keV and ET = 208.4 keV)1 is starting to find several applications in nuclear medicine, especially for metabolic radiotherapy of cancer.2 The production of 177Lu is

710

711

carried out mainly in thermal nuclear reactor.3'4 The principal methods of production is either by direct neutron capture reaction Lu(n,y) Lu on enriched 176Lu target, or by the neutron capture reaction on an enriched 176Yb target followed by beta decay 176Yb(n,Y)177Yb—^—> 177g Lu. u The second method produces a high specific activity no-carrier-added (NCA) 177gLu, while the first method produces a low specific activity carrier-added (CA) mixture of both 177gLu and 177mLu. Thus, in this case the 177gLu is contamined by the longlived radionuclidic impurity 177mLu. The second method does not produced at all this impurity, since 177Yb decays only to the ground state 177gLu.' Due to the long half-life of ,77mLu it is important from point of view of radioactivity waste disposal and dose to the patient to measure the percentage of this radionuclide in order to evaluate the radioprotection measures to be taken into account. Finally, the measurement of the isomeric ratio of 177Lu allows identifying the production method used for the sample supplied to the hospital. 2. Materials and methods Some radiopharmaceutical samples labelled with 177gLu(177mLu) were investigated with two experimental methods: gamma spectrometry by HPGe detectors and beta spectrometry by liquid scintillation counting (LSCS). Samples were those used for radiotherapy by nuclear medicine division of Istituto Europeo di Oncologia (Milano). Gamma spectrometry was used to identify the radionuclides present in the sample due to their gamma lines energies and to measure the activity of the various radionuclides. LSCS was used to measure the rate of beta decay. The ratio of the activities of the two isomers was measured from the deconvolution of the decay curve. The samples were counted with a Beckman LS5000 scintillation counter (anticoincidence circuits, quench correction with Horrocks method and three counting window capability). 3. Results and discussion In Fig. 1 it is shown a typical energy spectrum of a sample containing 177Lu. As can be seen, the sample consists of both 177gLu and 177mLu. In all 177gLu samples investigated the only radionuclidic impurity identified was 177mLu, and no other gamma emitters were detected. The half-life of radionuclides in the solution was calculated from the analysis of the rate of decay of the various gamma lines in the energy spectrum acquired. In Figs. 2 and 3 are shown the decay curves of the 208 keV and the 228 keV lines of g and m levels, while the half-life is calculated from the linear fit of the experimental data decay curve.

712

Figure 1. Characteristic gamma spectrum of used for metabolic radioimmunotherapy.

l7?T

'Lu

For the correct evaluation of the half-life of 177gLu we have used the gamma emission at 112.95 keV (6.4%) and at 208.37 keV (11.0%). Each of these two lines is unfortunately in common with those of 177mLu. Nevertheless the experimental half-life found (6.724 ± 0.006 d) is in good agreement with the value taken from the literature (6.734 d) for 177gLu.1 As

the half-life of " ,m Lu is much larger than that of Lu, the agreement of our results with the literature value indicates the negligible contribution of 177mLu to these gamma lines. 177g

208 keV emission

\

228 keV amission t ] o =150.395±10.496d Discrepancy from literature1 value tm = 160.4 c

\

A = 6.24 %

•-.. \ \ "•v. Vu^r6™*0006'

^*'\

1

Discrepancy from literature value 1^=6.734 d A = 0.15%

23

29

35

41

Time (d)

Fig. 2. Decay curve of 208 keV line of 177sLu.

Fig. 3. Decay curve of 228 keV line of 177mLu.

For the half-life of 177mLu the 3 gamma lines analyzed are the most abundant and not in common with those of 177gLu: 228.48 keV (37%), 327.68 keV (18.1%) and 378.50 keV (29.77%).' However, since the measurement of the sample were done to less than one half-life of the metastable nuclide the resulted half-life obtained in this case is not accurate. Anyway for the most abundant line, the 228 keV, the obtained half-life was found to be 150.4 ± 10.5 d which differs from the literature value of 160.4 d of less than 10%. The ratio of activities of the two isomers of l77Lu was calculated from the measurement of their peaks neglecting the contribution of mLu to the lines 112.95 and 208.37 keV, for the same reason cited above. The results are given in Tab. 1. It can be seen that the radionuclidic purity is better than 99.98% at measurement time. This measurement was done one week after injection onto

713 patients and hence at the time of labeling and injection the radionuclidic purity was better than 99.99%. Although the fraction of Table 1. Activity of' 7 7 g m d m Lu. 177m Activity 17?8Lu (23/12/2004) 77 746 ± 65 Bq Lu is very small, this analysis Activity 177mLu (23/12/2004) 18.0 ± 0.8 Bq demonstrates that 177Lu was produced by direct activation of a 176 Lu enriched target since the measured fraction of 177mLu is very close to what is calculated (0.014% at the end of irradiation, see Fig. 6) from the half-lives and the thermal neutron activation cross-sections for the two isomers (2 100 b and 7 b for the ground state and the metastable state, respectively).2 LSCS was used to verify the results obtained by the gamma spectrometry. The study of the decay curve was performed in three counting windows: a whole range interval (from channel 0 up to channel 1000 corresponding to a energy range from 0 to 2 MeV), from 400 to 675 (channel) and from 100 to 700 (channel). In Fig. 4 is shown, the decay curve of a sample over the whole energy range. The linearity of the curve suggests that the time of the study was too short to identify long-lived radionuclides. The half-life of 17?8Lu was found to be 6.688 ± 0.004 d with a discrepancy of-0.68% from the value taken from the literature.1 A similar measurement for another sample was done for a •^ longer period of time and in Fig. 5 is shown a representative decay curve for the whole range. As it can be clearly seen there is more than one radionuclide in the sample. From the first 23 points it was Fig. 4. Decay curve in semilog scale for the energy interval from 0 to 2 MeV possible to obtain the half-life of 177gLu (discrepancy of -0.21% from literature),1 but from following measurements it was not possible neither distinguishing the number of components in the long-lived tail nor to get the half-life of' mLu, due to the too short time between different measurements, which is much smaller than the half-life of 177mLu. Moreover, in order to find the fraction of 177mLu, we have applied the Biller method to data for every window considered, assuming that the only constituents of the sample are the two isomers under investigation and using their decay constants taken from the literature (A.i(177gLu) = 1.19135 10'6 s"1, A,2(i77mLU) = 5.00158 10"8 s"1).1 If our assumption is correct, that there are only two t,„=6.70ie8tO.OOS45d

Discrepancy from literature' valua 1,^=6.7340 d A=0.475S1 %

714

components Fig. 6. \

17?8

Lu and

177m

Lu, the Biller plots must be linear as can be seen in

Decay curve (0-1000 channel)

y"

Biller plot (0-1000 channel) \

1,^=8.7210.06 d Discrepancy from literature' va ue 6.734 d A =0.21%

V ^

\

y' 0*--..

\ \ Fig. 5. Decay curve for the 0-2 MeV interval

/

y

nr

,m

Lu)/A( ,77 'Lu) = 0.012% ]

Fig. 6. Biller plot for the 0-2 MeV interval

4. Conclusions The only radionuclidic impurity identified in all samples was 177mLu, and the half-life of the beta emitter 177gLu is in agreement with the literature. These results were performed with two different techniques: HPGe gamma spectrometry and spectrometry by liquid scintillation counting. Another goal is the identification of production method of 177gLu, which heavily determines the specific activity of the labeled compounds. With this study we demonstrate that the 177gLu used in the radiopharmaceutical compounds was produced in a nuclear reactor by 176 Lu(n,y)177Lu nuclear reaction, as we found the presence of the long-lived impurity 177mLu. The l77mLu is a fingerprint of production with this method. References 1. 2. 3. 4. 5. 6.

R.B. Firestone, CM. Baglin, F.S. Y. Chu, Table of Isotopes, update on CDROM{\99%). P. Olivier, Nucl. Instr. Meth. Phys. Res., 527, 4 (2004). M.R.A. Pillai, S. Chakraborty, T. Das, et al., Appl. Radial hot. 59, 109 (2003). R. Mikolajczak, J.L. Parus, D. Pawlak, et al, J. Radioanal. Nucl. Chem., 257, 53 (2003). G. Pfennig, H. Klewe-Nebenius, W. Seelxnann, Karlsruher Nuklidkarte (1998). F. Groppi, H.S. Mainardi, A. Martinotti, S. Morzenti, M.L. Bonardi, J. Radioanal. Nucl. Chem., 263(2), 521 (2005).

SPATIAL LINEARITY IMPROVEMENT FOR DISCRETE SCINTILLATION IMAGERS L.MONTANI, G.IURLARO, A.SANTAGATA, N.BURGIO, R.SCAFE* Casaccia Research Centre, ENEA 00060 S.Maria di Galeria, Rome, Italy In principle, a scintillation imager made by a crystal array and a position-sensitive photomultiplier shows spatial non linearities at borders due to the undersampling of scintillation light distributions. To improve the response of such detectors a method has been carried out based on the comparison between experimental and simulated images. A computer simulation code has been developed and used to foresee the imager response. The code computes the tapered charge-weighting factors giving the optimized event centroiding in terms of crystal-pixel identification. The comparison between measured centroids and crystal axes produces the effective correction table for image linearization. Experimental results concerning a 8x8 crystal array coupled to a position sensitive photomultiplier tube (1 inch square) are presented and discussed.

1. Introduction The interest in small scintillation detectors for nuclear medicine applications has been increased in the last years [1]. Small Field Of View (FOV) scintigraphic handy devices with few millimeters spatial resolution have been developed with the aim of small lesion detection for early diagnosis. In principle these detectors, operating in single photon emission modality, show distortions at peripheral crystals due to the undersampling of Scintillation Light Distributions (SLD). This effect is particularly relevant if the device has a FOV as large as the photodetector overall area. A method for spatial linearity recovering is presented here, based on the comparison between experimental and simulated images. The latter are obtained from a simulation code capable of foreseeing the detector response [2]. The method computes optimized charge weighting factors to be used for events centroiding. An image with proper pixel identification is thus obtained which represents the first necessary step for the linearization corrections to be applied.

* Corresponding author: [email protected]; Ph: +3906-3048-3434; Fax: +3906-3048-3989

715

716 2. Equipment and Method The method has been applied to images obtained with a small FOV detector made by a scintillation array and a Position Sensitive Photomultiplier Tube (PSPMT) for light readout. The crystal array, having 25.7mm side, is composed of 8x8 CsI(Tl) pixels (3.0x3.0x5.Omm3) manufactured by Hilger Crystals (Fig.l left). The array is coupled to the PSPMT through a planar Perspex light guide. The PSPMT used for light detection is a Hamamatsu R8520-00-C12, having 25.7x25.7mm2 overall area, 22.0x22.0mm2 active area, and 6X+6Y crossed anodes [3]. An exploded view of the detector components is shown in Fig.l (right). PSPMT (R8520-00-C12)

Light guide \ Scintillation array

-is-

Fig. 1 The 8x8 crystal array (left) and an exploded picture of the detector components (right).

Charge readout is performed by single-anode electronics, each chain consisting of a low noise preamplifier and a pulse amplifier-shaper. The signals digitization is achieved by 3 computer boards, having 4 channels each [4]. Each signal is sampled throughout lOus by a 12-bit ADC at a rate of 5 Msample/s and a LabWindows C++ software performs data acquisition. Triggering is taken from the 10th dynode signal. Experimental images are obtained irradiating the detector with a "Co (Ey =122keV and 136keV, 89.0% and 11.0% abundance respectively) 1mm diameter spherical source positioned 40cm apart, coaxially with the detector. The image is then reconstructed calculating the single event centroid coordinates using the algorithm: X =

Til*

Y =

(1)

wXi, wYi being the charge-weighting factors and qxu <7K the digitized values proportional to the charges on the i* anode along X and Y directions respectively.

717

Images obtained with such a detector suffer of non-linearities related to the SLD undersampling at the detector boundaries. This effect is particularly important if the crystal-pixel involved in the scintillation is located at array's limits and, afortiori, if the crystal-array dimensions exceed the PSPMT active area, as it is in the present assembly. To improve the image linearity over the FOV an

appropriate procedure has to be used. Whatever the procedure, it can be correctly applied only if an adequate pixel identification is previously obtained, otherwise a number of events generated in a pixel would be wrongly reconstructed in areas corresponding to neighboring crystals. Our pixelidentification strategy consists, as described below, of charge weighting factors (herein termed as tapered) to be used in Eq. (1) for events centroiding. These factors are computed by a simulation code (DISIS, Discrete Scintillation Imager Simulator) developed to foresee the planar imager response corresponding to the assumed detector assembly [2]. DISIS is based on the analytical model [5] in which the SLD emerges at pixel axis and is described by: T SLD

^

=

(T>+(x-xPfHy-Ypyy

'

(2)

where T is the distance between the crystal-pixel base and the photocathode, Xp, Yp are the crystal-pixel axis coordinates and q is the parameter characterizing the SLD spread. In this case FWHM (Full With at Half Maximum) and FWTM (Full With at Tenth Maximum) are, respectively: FWHM = 2Tyj(2Vg - 1 ) ,

FWTM = 27\/(l0 V » -1) .

(3)

To perform the simulation DISIS needs the q value in the considered experimental setup. This value can be measured using different methods as described in ref. [6] or it can be evaluated by an iterative procedure, based on the comparison between experimental and simulated images. Once the q value has been found DISIS can calculate the tapered charge weighting factors giving the best image reconstruction, defined as the one which minimizes the sum of distances between each pixel centroid and the corresponding crystal axis over the array. Once the spots corresponding to all crystals have been distinguished, it is possible to apply a linearization correction, image-pixel by image-pixel. To this aim two arrays (X and Y) are used, obtained as additional quantities calculated from the difference between measured centroids coordinates and crystal axes coordinates.

718

3. Results and Discussion Fig.2 shows a comparison between simulated (a) and experimental images, reconstructed with both linear (b) and tapered (c) charge weighting factors. Note that each square image frame corresponds to the 25.7x25.7mm2 PSPMT area.

» i 11 •tt ** • * : « • • * ** * » » » « • •«

#

a*

•

• • * *• • * • m

IS t

: t tit a

b

* *

Ifc

»

*

»

« j£

».%.

*

»

•

*

a «

*n

*

•

»

*

**

« *

* *

*

.*

4

#

-P

*-

-*

j

*

«

»

*

*

*

« * :* * * » * '

» * c

*

s

d

Fig.2 Flood field irradiation: simulated image (a) and experimental image reconstructed using linear (b) and tapered (c) charge weighting factors. The last image (d) is obtained applying the linearization procedure to image (c).

In Fig.2(a) and Fig.2(b) images have been reconstructed using in Eq.(l) charge weighting factors linearly proportional to anode's positions. The 4x4 inner pixels appear well resolved in all cases: the SLDs corresponding to each crystal are well sampled and centroids are then correctly reconstructed. The border pixels suffer, on the contrary, of a compression effect related to the SLD undersampling at the detector boundaries. When the image is reconstructed using tapered weighting factors, Fig.2(c), this effect is corrected for and even the pixels at the image boundaries can be distinguished. The pixel identification obtained on the whole image area becomes sufficient for the linearization technique to be applied leading to the image shown in Fig.2d. The procedure gives quite good results in the investigated assembly of 3.0mm crystal pixel side. In the case of smaller crystal pixels, comparable with PSPMT

719

dead zone width, the method is less effective due to the poor visibility of peripheral spots. It is worth noticing that the method may also correct the internal R8520-C12 anodic asymmetry along X and Y reported by Hamamatsu [7]. Acknowledgement This work has been supported by Italian National Council for New Technologies, Energy and the Environment (ENEA) under Project P466. References 1. E.J.Hoffman, M.P.Tornai, M.Janecek, B.E.Patt, J.S.Iwanczyk, "Intraoperative probes and imaging probes", Eur J Nucl Med, Vol.26, No.8, 913-935, August (1999). 2. R.Scafe, G.Iurlaro, L.Montani, A.Santagata and N.Burgio, "DISIS - A Computer Simulation Code for Discrete Scintillation Imagers", IEEE NSSMIC Conference Record, Vol. 6, 4004-4008, 16-22 October (2004). 3. Hamamatsu Photonics K.K., Electron Tube Center, "Position Sensitive Photomultiplier Tubes R8520-00-C12 R8520U-00-C12"; TPMH1276E01 Oct. 2001 IP, Japan. 4. National Instruments Corporation, "PCI-6110E/6111E Multifunction I/O Boards for PCI Bus Computers, User Manual", (1998). 5. R.Scafe, R.Pellegrini, A.Soluri, L.Montani, A.Tati, M.N.Cinti, F.Cusanno, G.Trotta, R.Pani, F.Garibaldi, "A Study of Intrinsic Crystal-pixel Lightoutput Spread for Discrete Scintigraphic Imagers Modeling", IEEE Trans Nucl Sci, Vol.51, No.l, 80-84, February (2004). 6. R.Scafe, R.Pellegrini, A.Soluri, N.Burgio, G.Trotta, A.Tati, M.N.Cinti, G.De Vincentis, R.Pani, "Light Output Spatial Distribution of CsI(Tl) Scintillation Arrays for Gamma-rays Imaging", Nucl Instr & Meth A, vol.497, 249-254 (2003). 7. Y.Yoshizawa, Y.Hotta, Hamamatsu Photonics K.K., Electron Tube Center, private communication (2004).

HIGH RESOLUTION, HIGH SENSITIVITY DETECTORS FOR MOLECULAR IMAGING OF SMALL ANIMALS AND TUMOR DETECTION M.L. MAGLIOZZI*, E. CISBANI, S. COLILLI, F. CUSANNO, R. FRATONI, F. GARIBALDI, F. GIULIANI, M. GRICIA, S. LO MEO, M. LUCENTINI, F. SANTAVENERE, P. VENERONI Istituto Superiore di Sanita and Sezione INFN Sanita, Rome, Italy O. SCHILLACI, G. SIMONETTI University Tor Vergata, Rome, Italy S. MAJEWSKY Jefferson Lab, Newport News, USA M.N. CINTI, G. DE VINCENTIS, R. PANI, R. PELLEGRINI, F. SCOPINARO University La Sapienza, Rome, Italy Imaging techniques with radionuclides provide very sensitive measures of a wide range of specific processes underying disease in the body. Detection of very small tumors with high specificity is therefore possible but the tecnique requires both high spatial resolution and high sensitivity. We present the first simulations, performed by means of GEANT4 code, of breast tumors, imaged by different configurations of a compact discrete gamma camera, in order to optimize the performances of dedicated detectors for these tasks. Simulated planar images from 6 to 10 mm diameter tumors, placed at 5 mm from the collimator, were generated for Nal scintillator pixel sizes of 1.0x1.0 and 1.2x1.2 mm2, hexagonal hole Pb collimators with hole size of 1.5 and 1.9 mm. The generated photons have been sampled by two modelled Hamamatsu H8500 and H9500 PMT. Tumor to background uptake ratio from 1:6 to 1:12 has been considered. The preliminary results in terms of spatial resolution and SNR show a slightly better performance of the high efficiency collimator, larger crystal size and H9500 combination.

Corresponding author. E-mail address: [email protected]

720

721

1. Introduction The power of imaging techniques with radionuclides lies in their ability to provide very sensitive measures of a wide range of biologic processes in the body. These techniques allow studies to be targeted to the very specific biologic processes underlying disease, the detection of very small tumors with high specificity is becoming possible. Moreover the recent availability of genetically modified mice has generated a rapid growth of interest in radio imaging of small animals, because it enables a wide range of human diseases to be studied in animal models. Nevertheless, the technique is challenging due to the concurrent requirements in terms of high spatial resolution and high sensitivity2. Spatial resolution at millimeter scale is required for oncological nuclear medicine while many small animal imaging applications need submillimeter spatial resolution. Good sensitivity is also required, especially for small tumor detection (breast for example). PET technique, regardless of intrinsic limitations in terms of spatial resolution, suffers for high cost and complexity, while SPECT is less expensive and can be improved accordingly to the final purpose 12 . The performance of compact discrete gamma camera is strongly influenced by camera geometry, so both collimator configuration, scintillator pixel size and PMT anode size, affect breast tumor images. The preliminary simulations discussed in this paper have been designed to study how imaging is affected by camera components but also by tumor size and tumor to backgroung uptake ratio (T/B). 2. Monte Carlo simulation program and image reconstruction A GEANT4 3 program has been coded to simulate a scintimammography detector, schematically shown in Figure 1, together with a real system. The gamma rays are generated in a large 3D phantom breast of 10x10x6 cm3, including a spherical tumor, and simulates all possible physical interactions in the world volume. SIMULATED DETECTOR

BuUaatefc^

REAL DETECTOR

SsHUJSaJM Crystal Array COMPONENTS

Figure 1. Basic components of simulated gamma camera and real detector.

722

In order to simulate a clinical scintimammography scan, the number of generated 140 keV gamma rays are calculated assuming a background activity density of 80 nCi/ cm3 for breast phantom, T/B, of 6, 10, 12 and an imaging time of 10 minutes 4. Images are obtained cutting out the events with number of detected photons less than 270 in order to consider primary 140 keV gamma only (see Figure 2). x'/ndf Constant Mean;;;*;;;: Sigma

216.3/44 917.3 311:9 : 40.32

tb?.-!jfll

Imaged Photons

100

200

300

400"•••••500.M; 600 TOOBSOfl N u m b e r o f detected photons v

Figure 2. Left: distribution of the detected optical photons. The gaussian fit represents the photoelectric peak of the 140 keV emitted gamma. Right: image of 10 mm diameter tumor, T/B=10, collimator hole size of 1.5 mm, crystal pitch of 1.2 mm, PMT anode size of 3mm.

Collimator hole size, scintillating crystal size, PMT anode size, tumor diameter and T/B ratio are free parameters of the simulation. In particular we considered, all the possible combinations of: • collimator hole size of 1.5 and 1.9 mm, (High resolution, HR, and High Efficiency, HE, respectively); • crystal pixel size of 1.0x1.0 and 1.3x1.3 mm2, pixels are spaced by epoxy glue of 0.2 mm in both configuration. • PMT anode size of 6 and 3 mm (simulated model H8500 and H9500 from Hamamatsu5); • Tumor diameter of 6, 8, 10 mm; • T/B values of 6, 10, 12. 3. SNR and detector configuration In order to compare tumor images from different scans we use the Signal to Noise Ratio (SNR) operatively defined as the maximum - respect to regions of interest (ROI) surrounding the tumor - of "the background subtracted counts in the ROI" over "the ROI total counts squared". The ROIs are square boxes centered on the tumor maximum intensity pixel.

723

Figure 3 shows the SNR versus both uptake (T/B) for tumor sizes of 6 and 10 mm and tumor sizes for T/B of 6 and 10.

•

15

L._

—

i a^

3 1

:--

*g3

'- i|35S ':

|I

..~.-.-^irfc-- Z?^~

!

1

Size = 6 mm i

i

j Size = 10 mm

10 11 Tumor to Background uptake Ratio

7 8 , 9 10 11 Tumor to Background uptake Ratio

12

a .

:

•y*,^~-^...-\

• .

.

.

.

-

•

• •

•

. ' ...... < < ^ . , .1

==••

•

.7

^

.

IF :

.

• " •"

": "

.

: ::

-

^£U ^SKg^*"3* i

_.

» * & %

"::^~^

B=*

•

..... s » = ™ „ , j

IL . . '

^

;r::-^r S -::r-:i

.

•••-••

U^0Z f ^ l

- j '

:

-

• •

•

•

.

;

'

•

'

"

:

•

"

;

'

.

gS^

:::™'S'6'. i i i i

i

\.

T/B = 1 0

:t.:..:,.,d

Tumor Siz&;[:mrh]:

".""." i v . vt;".' Tumor Sizejrnm];

:

• HE-1. -H850O D HE-1. -H9500 Sli HEvl:3r-H8S00 2i HE-1.3-H9S00 *Si 'HR'-i; -H8S00 ^O:M^MH9500

Figure 3. SNR versus both uptake (T/B) for tumor sizes of 6 and 10 mm (a e b) and tumor sizes for T/B of 6 and 10 (c erf).

4. Spatial sampling In order to evaluate the optimal spatial sampling, the influence of different PMT anode sizes on the centroid reconstruction has been investigated. Figure 4 summarizes the results in terms of the FWHM of the photon shower emitted by the Nal scintillator crystal, either continuous or pixellized. The estimated Centroid Standard Deviation (STD[Xrec-Xreal]) is an index of the spatial resolution degradation induced by the sampling; smaller values correspond to smaller degradation.

724

Nal continuous A 4 mm thick * 5 mm thick f 6mm thick

0.22 0.2

FWHM (mm) 9.2 10.1 ]0.3

Nal pixellated • 1.2rampitch 8 15 mm pitch

\ \

to-1

S

'\ 1

0.12 0.1 O.OS

006

V\\ ~-\ v

0.04

— ^^*r?-

FWHM/anode size

—- .. h " - --

10"

10 15 FWHM/anode size

20

25

Figure 4. Estimated Centroid Standard Deviation (STD[Xrec-Xreal]) versus width of the photon shower emitted by the Nal scintillator crystal. The optimal anode size (minimum degradation at maximum pixel size) is ~ 0.5 * [FWHM Optical Distribution]. The left plot is a zoom of the right one in the curves knees.

Conclusions The above investigation is still underway; more configurations and different

tumor depths will be simulated and compared to real measurements. The very preliminary results show that the High Efficiency, 1.3 mm scintillator pixel combined to the H9500 PMT seems to represent the optimal configuration in terms of SNR, in most of the simulated cases. The use of H9500 PMT (when coupled to a 1.3-1.5 mm crystal pixel) is also confirmed by the spatial resolution analysis. References 1. 2. 3. 4. 5.

F. Cusanno et al., Nucl. Instr. and Meth. A527, 140 (2004). F. Garibaldi et al., IEEE Trans. Nucl. Set 52, 573 (2005). GEANT4 Home Page: http://wwwasd.web.cern.ch/wwwasd/geant4/geant4.html GJ. Gruber et al., IEEE Trans. Nucl. Sci. 46, 2119 (1999). Hamamatsu H8500 and H9500 specifications, www.hamamatsu.com.

STRIP IONIZATION CHAMBER AS BEAM MONITOR IN THE PROTON THERAPY EYE TREATMENT F. MARCHETTO, R. CIRIO, M.A. GARELLA, S. GIORDANENGO Istituto Nazionale di Fisica Nucleare (INFN), Torinojtaly A.BORIANO Istituto Nazionale di Fisica Nucleare (INFN), Torinojtaly and Scuola di Specializzazione in Fisica Sanitaria, University of Torinojtaly N. GIVEHCHI, A. LA ROSA, C. PERONI Istituto Nazionale di Fisica Nucleare (INFN), Torinojtaly and Dipartimento di Fisica Sperimentale, University of Torino, Italy M. DONETTI Istituto Nazionale di Fisica Nucleare (INFN), Torinojtaly and Fondazione CNAO, Milano.Italy F. BOURHALEB, G. PITTA' Fondazione TERA, Novarajtaly G.A.P. CIRRONE, G. CUTTONE Laboratori NazionaU del Sud, INFN, Catania Jtaly L. RAFFAELE , M.G. SABINI, L. VALASTRO Laboratori NazionaU del Sud, INFN, Catania, Italy and Dipartimento di Radiologia e Radioterapia, University of Catania, Italy

Since spring 2002, ocular pathologies have been treated in Catania at the Centro di AdroTerapia e Applicazioni Nucleari Avanzate (CATANA) within a collaboration between INFN Laboratori NazionaU del Sud (LNS), Physics Department, Ophthalmology Institute, Radiology Institute of the Catania University and CSFNSM Catania. A beam line from a 62 MeV Superconducting Cyclotron is used to treat shallow tumors. The beam is conformed to the tumor shape with a passive delivery system. A detector system has been developed in collaboration with INFN-Torino to be used as real time beam monitor. The detector, placed upstream of the patient collimator, consists of two parallel plate ionization chambers with the anode segmented in strips. Each anode is made of 0.5 mm-wide 256 strips corresponding to (12.8 x 12.8) cm2 sensitive area. With the two strip ionization chambers one can measure the relevant beam parameters during treatment to probe both asymmetry and flatness. In the test carried out at CATANA the detector has

725

726 been used under different and extreme beam conditions. Preliminary results are given for profiles and skewness, together with a comparison with reference detectors.

1. Introduction Energy loss of protons crossing the matter is characterized by the increase at the end of range (Bragg peak). In cancer therapy, this feature has been used to conform the dose distribution to the tumor while sparing the healthy tissue. At the Laboratori Nazionali del Sud-INFN (LNS), Catania, Italy, the proton beam facility, used mainly for atomic and nuclear research, has been updated to host the therapy treatment for ocular pathologies. The therapy facility has been built and operated by the Centro di Adroterapia e Applicazioni Nucleari Avanzate (CATANA)[1,2] as from 2002. The continuous flux of protons from the 62 MeV Superconducting Cyclotron is steered to the treatment room. We remark that at this energy the beam penetration is about 30 mm, which is suitable to treat most of the ocular tumors. In brief, the beam encounters the scatterer, a plastic slab shaped to flatten and spread transversally the beam. The scatterer is followed by a range shifter to spread out the Bragg peak and to conform the longitudinal beam energy loss to the tumor dimensions. A system based on two full area ionization chambers ensures the precise measurement of the delivered fluence. A collimator, personalized for each patient, shapes the beam to the tumor. The patient alignment is obtained with great accuracy with a motorized chair and a laser pointing system. A collaboration between CATANA and INFN-Torino has developed a detector (in jargon MOPI) to monitor the beam online, this is to say during the patient treatment. This paper describes in detail the detector and presents the preliminary results from several measurements performed on the treatment beam line. 2. Detector design The detector is a parallel plate ionization chamber with one of the electrodes (anode) segmented in strips. Similar ionization chambers have been developed by the same group to be used in hadron-therapy [3]. The detector is a sandwich of two similar chambers sharing the same cathode. An exploded view of MOPI is shown in Fig. 1. The chamber electrodes are glued to frames, which ensure the mechanical rigidity and provide the gas gap. The first and the last frames are two anodes with 256 strips oriented horizontally and vertically respectively. Each strip is 400um wide, with lOO^im gap between adjacent strips.

727

Fig. 1 An exploded view of the chamber.

The total sensitive area is (12.8 x 12.8) cm2. The anodes are made of 35|xm thick Kapton foil, covered with 15um thick Aluminum. The cathode is made of 25um thick Mylar foil aluminized on both sides. Protons cross a sensitive volume 6 mm thick that, for this application, is filled with air. To limit the impact of the detector on the beam, the total thickness, crossed by the beam, was kept as thin as possible. Indeed the total water equivalent thickness of the chambers is ~ 200 |xm. The electric field is set to ~ 850 V/cm by applying a high voltage of 500 V. The read out is performed via a 64-channel VLSI CMOS 0.8nm chip, Tera06, being each strip served by an individual channel [4]. An electronic board, hosting four chips, serves each chamber. The signals from the strips are carried to four connectors located on the edge of the anode frame, and finally from the connectors to the electronic board. Each channel of Tera06 is a current-to-frequency converter with an adjustable gain between 100 and 800 fA/count (charge quantum), which has been set to 200 fA/count for this project. The front end is an integrator followed by a comparator. Every time the integrator output exceeds the threshold of the comparator a signal is sent to a pulse generator which performs two functions: a) send pulses to the 16-bit counter; b) send pulses of definite charge (charge quantum) which are subtracted from the input. This mechanism stops only when the integrator output returns below the threshold. The maximum pulse frequency is 5 MHz corresponding to a maximum current of 1 (jA for a charge quantum of 200 fA. The read out can be done at any time by issuing a strobe to the counters to transfer the present count values to a register. Finally the register can be acquired at any time: this architecture makes the design dead time free. The maximum counter read out rate is 10 MHz (100 ns/channel). From this figure one can predict that the full two-plane chambers

728 can be read at a maximum rate of ~20 kHz. Some more features of the electronic include: 1) background count rate ~1 Hz corresponding to a background current of 200 fA; 2) deviation from linearity better than 0.4% for current spanning over six orders of magnitude; 3) channel to channel gain spread better than 1 %. A processor mounted close to the electronic board is programmed to read the chambers, to compute the beam parameters and to transfer the information to the Treatment Control Program. The full cycle to check the beam is done at a speed of ~2 Hz. 3. Methods The beam Quality Assurance (QA) is checked by the measurement of the beam profile before the patient treatment. A fine scan along the (x-y) coordinates is performed with a diode placed at the iso-center to measure the dose delivered as a function of the position in the plane transverse to the beam propagation. From these measurements, beam flatness, ST, and asymmetry, RT, are evaluated. The beam is accepted when both ST and RT for either x and y direction are within a given range. The check is performed with the native beam, which corresponds to the most severe beam conditions. Indeed both scatterer and modulator tend to flatten the beam. MOPI is used as beam monitor during the patient treatment: the beam is checked continuously and if the beam parameters deviate from the required conditions the beam can be stopped. A set of measurements have been performed to connect the radio-therapeutical quantities, ST and RT, at the iso-center to the beam parameters as measured upstream the collimator. In this paper, the preliminary results of these measurements are reported. 4. Results The raw strip responses are corrected to take into account the electronic and gas gain. The calibration coefficients are derived by comparing the fluence integrated along a strip to the GafChromic MD-55 film measurements integrated over the same area. The spread of the stability of the coefficients have been proved to be better than 0.5% (r.m.s.) by directly comparing set of coefficients derived at different times. To evaluate the beam flatness and asymmetry we computed the skewness, y, of the profile: at each acquisition the computation is extended only to the strips corresponding to the collimator.

729

Chomber profile

Fig. 2 Chamber profile for three beam settings

Diode profile

Fig. 3 Diode profile for three beam settings

The diameter of the collimator is of 25 mm, thus about 50 strips/plane are used for the computation. To study the relationship between y and the variables ST and RT, the beam was deviated on purpose changing the current in the last pair of dipole steering magnets. In Fig.2 we show the chamber profiles for three different beam settings. For the same settings we show in Fig. 3 the profiles as measured with the diodes. Indeed the two sets of profiles are different: the diode measures the point-to-point dose along a line, on the other hand the chamber integrates over a line, where the line integral changes as a function of the strip position with respect to the beam center.

.

0.075

045

0.025

•

•'"

•OJHS

*

-OJ»

•QJB76

Skewness V vs time (see.)

Fig 4 y as a function of time.

: Skewness vs. steering magnet current

Fig. 5 y as a function of the steering magnet current.

We remark that the variation of y with stable beam conditions is of the order of 10" : this is shown in Fig. 4 where y is shown as a function of time for a typical data acquisition. Finally we show the skewness value, averaged over a run, as a

730

function of the current in a dipole magnet used to steer the beam on the horizontal plane (black dots) and on the vertical plane (blue dots). We observe that the strong correlation y - magnet current can be used to probe beam instabilities. In other terms the y value is associated to a beam shape. It is thus necessary to define the y range, which indicates a beam shape within acceptable limits. From the correlation between y and the radio-therapeutical quantities, ST and RT, (see Fig. 6) and by inferring the ST and RT constraints we plan to extract the y limits which when broken will require to stop the beam. As an example in Fig. 6 we show a y-S T correlation curve for a scan of the current bending magnet. The QA requirements ask for a value of ST within 97 and 103%, that will project the limits on y. Studies are ongoing and it will require some more data taking and in this paper we will not present further details of the analysis.

0.1

0.07S

0-O»

0.02S

0

r

_

-0.02S

7

-O.OS

7

-0.07S

0-1

•

7

• • • * •

: 0O

05

1CO

lOS

• • no

115

1 20

»kw vs. «i

Fig. 6 Skewness vs ST for steering magnet current.

5. Conclusions A strip ionization chamber has been developed and used as online beam monitor at the CATANA beam. The chamber, made of 256 strip/chamber, is almost transparent to the beam, being the total thickness of ~0.2 mm equivalent water. The read out of the full chamber is performed at a rate of 2 Hz. Charges collected by the strips at each read out are then used to determine the beam spatial stability. Studies to derive the beam acceptance limits are underway.

731

References 1. G.A.P Cirrone et al., A 62 MeV proton beam for the treatment of ocular melanoma at Laboratori Nazionali del Sud - INFN, Transaction on nuclear Science, in press. 2. G.Cuttone et al., News on the status of the CATANA project at INFN-LNS, Particles Newsletter 28(2001)8-10. 3. R. Bonin et al., A pixel chamber to monitor the beam performances in hadron therapy, Nucl. Instrum. and Methods in Phys. Res. A 519 (2004), 674-86 and references therein. 4. G.C. Bonazzola et al., Performances of a VLSI wide dynamic range current-to-frequency converter for strip ionization chambers, Nucl. Instrum. and Methods in Phys. Res. A 405 (1998), 111-20.

LOW DOSE, LOW ENERGY 3D IMAGE GUIDANCE DURING RADIOTHERAPY C J MOORE 1 , T MARCHANT 1 , A AMER 1 , P SHARROCK 1 , P PRICE 2 , D BURTON 3 'North Western Medical Physics and 2Academic Department of Radiation Oncology Christie Hospital, Manchester M20 4BX, United Kingdom General Engineering Research Institute, Liverpool John Moores University Liverpool L3 3AF, United Kingdom Patient kilo-voltage X-ray cone beam volumetric imaging for radiotherapy was first demonstrated on an Elekta Synergy mega-voltage X-ray linear accelerator. Subsequently low dose, reduced profile reconstruction imaging was shown to be practical for 3D geometric setup registration to pre-treatment planning images without compromising registration accuracy. Reconstruction from X-ray profiles gathered between treatment beam deliveries was also introduced. The innovation of zonal cone beam imaging promises significantly reduced doses to patients and improved soft tissue contrast in the tumour target zone. These developments coincided with the first dynamic 3D monitoring of continuous body topology changes in patients, at the moment of irradiation, using a laser interferometer. They signal the arrival of low dose, low energy 3D image guidance during radiotherapy itself.

1. Introduction 1.1 X-ray Imaging and Therapy: the Historical Divide W C Rontgen mastered X-ray production in 1895 and the first X-ray imaging machine made its debut in 1896. By the 1920s X-ray radiotherapy was routine and quite separate from radiology. Since then patient setup in radiotherapy has remained manual, currently guided by skin marks and the bony detail just visible on mega-voltage X-ray port films. In the 1970s CT scanning revolutionised radiotherapy treatment planning (RTP). The 1980s saw mega-voltage electronic portal imaging devices (MEPIDs) appear and industrial researchers Feldcamp, Davis and Kress (FDK) describe an algorithm for volume cone beam CT scanning [1]. In the 1990s Antonuk described an MEPID based on an amorphous silicon transducer [2]. The inherent limitations of mega-voltage X-rays for imaging spurred research into low Z targets in Linacs [3]. However, it soon became obvious that, just as CT had revolutionized RTP, it could do the same for treatment delivery. Furthermore, every mega-voltage X-ray treatment room had a rotating gantry onto which an X-ray tube could be mounted for scanning.

732

733

Hence, by the 1990s all the components for 'cone beam tomography' (CBT) were in place for use in the radiotherapy treatment room. Jaffray et al [4] reported kilo-voltage CBT in 2002 after integrating a diagnostic X-ray tube onto the gantry of an Elekta treatment machine. An Elekta research consortium then worked on machines derived from this concept. In 2003 the first patient was CBT scanned at the Christie Hospital in Manchester using what was by then the preproduction version of the Elekta Synergy specifically designed for image guided radiotherapy (IGRT). The long delayed marriage of X-ray imaging and X-ray therapy had come more than a century after Rontgen first mastered X-ray production. 1.1. Image Guide Radiotherapy Moore et al [5] have offered a definition of image guided radiotherapy; 'IGRT aims to actively inform setup and disease targeting by imaging bone and soft tissues, showing tumour and organs at risk in 3D, as close to moment of therapeutic irradiation as practical'. Apart from there being no mention of Xrays the definition distinguishes IGRT from image assisted procedures, which make use of CT, MRI and PET generated outside of the treatment room. Several important issues are implicit: 1. The importance of the body surface as the primary 3D interface for patient setup and dosimetry. The interfaces between anatomical structures below the body surface offer additional information for setup and dosimetry. 2. The need for soft-tissue contrast detail for targeting, specifically to reveal changes to the shape of the tumour and adjacent organs at risk (OAR). This determines the suitability of the pre-treatment plan as treatment progresses. 3. The need to contain potentially large imaging doses to patients, since IGRT, like conventional radiotherapy, is often delivered daily for weeks. 4. Imaging just before, or just after, the treatment itself is not 'as close as practical' to viewing the patient during radiotherapy itself and may add significantly to the total time taken to treat the patient. At the Christie Hospital, physicists have been addressing these issues. It was shown that low-dose, reduced-profile CBT could be used for 3D geometric setup by registration to pre-treatment planning images without compromising accuracy [6]. The feasibility of FDK-CBT reconstruction from X-ray profiles gathered between treatment beam deliveries, within a single fraction, was demonstrated [7]. Innovative, zonal cone beam imaging was then shown to

734

improve soft tissue contrast where it was needed most, in the target zone, whilst substantially reducing dose and scatter, at source in the patient [8]. Finally, radiation-free, dynamic 3D monitoring of body surface changes in patients, during irradiation, using a laser interferometer, was developed [9]. The latter signaled that IGRT at the moment of therapeutic irradiation had arrived. This paper describes the status of these developments. 2. Reduced Profile, Low Dose, Cone Beam Tomography for Setup

Figure 1. RTP-CT 5mm thick axial image (left), CBT 1mm images from an optimum FDK reconstruction using 382 profiles (middle) and with 77 profiles (right). Note the preservation on anatomical interfaces necessary for cross correlation registration with the RTP-CT scan for patient setup.

2.1.

Background

A cylindrical CBT scanning volume, base diameter and length N pixels has 7iN3/4 volume elements projected onto an N2 pixel flat panel image transducer. Hence, ~7tN/4 (0.785N) image profiles are required to calculate all the volume element attenuations. For N=512 this ~402 profiles. However, where the attenuation values are less important than tissue interfaces (body contour, skeletal surfaces, cavities and muscle-fat transitions) as would be the case for patient setup imaging based upon registration to an RTP-CT scan, then far fewer profiles should suffice. This promises reduced dose to the patient and fast CBT scanning, provided the IEC gantry rotation limit of 60 seconds can be relaxed. Recall that phase sorted scanning is used to ameliorate the effects of breathing on CT reconstructions. In this technique a suboptimal number of projections are used to reconstruct each breathing phase, which for the lung site is dominated by air cavity interfaces to the tumours and healthy, soft tissues.

735

2.2 Materials and Methods Sykes et al [6] took CBT scans of a head and neck anthropomorphic phantom and reconstructed the X-ray image volume using the FDK algorithm with 382, 148 and 77 image profiles. The reconstructions were then compared to a conventional RTP-CT phantom scan, produced to a 5mm slice width in 5mm steps protocol, using the Philips Syntegra correlation-matching facility running on a Pinnacle treatment planning platform. 2.3 Results Figure 1 shows a cross-section from an RTP-CT (left) scan, a CBT volume reconstructed with 382 profiles (middle) and 77 profiles (right). Cross matching RTP-CT and CBT volumes consistently achieved sub-millimetre, sub-degree accuracy, regardless of the number of profiles used for CBT 3. In-treatment Cone Beam Tomography 3.1 Background IGRT should be as close to treatment delivery as practical, ideally during treatment itself, in order to see and measure the patient when it is most relevant and to minimize any disruption to existing work flow. Most radiotherapy is delivered in daily fractions and makes use of beams of radiation delivered to a tumour from different, fixed angles. Whilst rotating between these angles the treatment beam is off and hence there is the potential to collect kV X-ray partial projection sequences, which, when summed, can be used for CBT [7]. 3.2 Materials and Methods Angular sequences of kilo-voltage X-ray projection images of a consented bladder patient were acquired during treatment. Three distinct projection acquisition techniques were applied on consecutive treatment days: a) Conventional CBT, continuous rotation acquisition, post treatment. b) Projections for CBT acquired between four orthogonal treatment beams. c) Patient movement during scanning simulated by combining two 180° acquisitions from separate treatment fractions with known patient displacements of 2.1, 0.7 and 2.6 mm anterior-posterior, left-right and superior-inferior respectively, and 0.8° rotation along the left-right axis.

736

The three projection sequences were each reconstructed using the FDK algorithm to produce 256x256x256 volume with isotropic 1mm resolution.

3.3 Results

Figure 2. 1mm thick views of a bladder patient from post-treatment (left), in-treatment (middle) and inter-fraction shifted (right) X-ray profiles seen in a coronal plane through the CBT image volumes.

Despite the different modes of acquisition, reconstructed volumes for techniques 3.2a and 3.2b are remarkably similar (figure 2, left and centre respectively). In particular the CBT for segmented, in-treatment acquisition was free from additional artifacts and soft-tissue reproduction remains of high quality. Technique 3.2c shows double-edging of the bony features reflecting the expected differences over the two fractions (figure-2, right). This clearly has practical value in detecting patient shifts during the weeks of radiotherapy. Note that patients are by definition seriously ill, some individuals uncomfortably so by mid to late treatment. 4. Zonal CBT: Improved Contrast Detail with Reduced Dose 4.1 Background CBT employs a minimally collimated, divergent (cone) beam of X-rays to generate a rotation fluoroscopy sequence of X-ray image profiles for reconstruction. The large irradiated volume of patient tissues produces a significant scatter field that serves only to increase the dose to the patient and compromise image quality once at the plane of the imaging transducer. Postpatient air gaps and grids at the transducer plane help reduce recorded scatter, do not reduce scatter in, or dose to, the patient. 'Coning down' clearly helps, but

737

the price paid is a loss of wide field interface detail needed for patient setup. Double exposure CBT mimics an established use of mega-voltage EPIDs, where a wide field image is immediately followed by one that is small. This 'local' CBT increases patient dose in return for improved image quality close to the scanning axis. However, there is no practical reason to maintain optimal contrast detail beyond the tumour target and organs at risk (target zone), since it has already been shown (section 2) that interface detail drives patient setup by registration with the RTP-CT scan. Reducing the dose outside the target zone should maintain setup capability and improve image quality inside the target zone by reducing scatter at source, in the patient bulk [8].

Figure 3. Coronal (left) and sagittal (right) views through zonal cone beams scan of a patient being treated for bladder cancer. Equivalent slice width 5mm. Note the high contrast target zone and the preservation of extra zonal detail achieved at reduced dose.

4.2 Materials and Methods The lead diaphragms of the kilo-voltage X-ray tube used for CBT were set open and a pair of aluminium transmission diaphragms, 4 half-value layers thick, were positioned to reduce dose inferiorly and superiorly to the target zone. The gap between the diaphragms defined a 5cm long target zone along the scanning rotation axis, and unrestricted cross sectional width through the target zone, normal to the scanning axis. Dose was measured at the central axis of CTDI phantom and compared to that for conventional wide field CBT. The contrast noise ratio was measured with the transmission diaphragms in place, then removed. A bladder patient was then scanned using the zonal CBT approach.

738

4.3 Results The central dose in the CTDI phantom is reduced by 50% inside and outside the target zone. This appears to be almost entirely due to the control of bulk scatter. Image contrast noise ratio increased by 15% compared to the equivalent wide field CBT. Figure 3 shows views through the zonal CBT scan of a bladder patient, at 5mm equivalent slice width in order to clearly demonstrate the relative quality of target zone and extra-zonal image detail. The preservation of softtissue detail in the superior and inferior extra-zonal volumes is noteworthy. 5.

In-treatment Dynamic 3D Body Surface Sensing of Patient Motion

5.1

Background

For intensity modulated radiotherapy (IMRT) discrete beams (segments) of radiation are sequentially superimposed to perform dose shaping about the tumour target. In less complicated applications, such as the large number of breast treatments, IMRT compensates for the 3D body contour to remove unwanted 'hotspots'. Even then, the time taken per beam can exceed a minute, during which time the breast surface will move in an unknown manner reflecting the particular breathing mode of the patient. The same is true for lung cases where illness directly affects motion. Anatomical motion is lost in the minute or two taken to CBT-scan a patient. A single breathing cycle cannot be charted in real time. To reconstruct representations of the different phases of breathing requires the iso-phase sorting of large numbers of profiles acquired over many motion cycles, resulting in increased dose. This is not practical during therapy itself. Optical sensors offer radiation free body surface measurement and unlimited use, most significantly during irradiation itself. The most common takes the form of manually positioned markers, which offer dynamic sampling but can only approximate a moving, continuous 3D surface. Hence an alternative measurement approach, based on structured light projection, is desirable for characterizing in-treatment body surface motion. 5.2 Materials and Methods Moore et al [9] have described a dynamic surface sensing system and have demonstrated deformable correction of a conformal radiotherapy treatment plan using the body surface movement measured during treatment of rectal carcinoma. The system is able to generate a 440x440 point surface height map,

739

with better than millimetre accuracy, from each 4 millisecond video frame of interference fringes projected onto and modulated by a patient's body surface. The spatial phase encoded in the fringes allows the relative height to be computed for any point on the body. A structured light projection system developed from this prototype was subsequently installed above the gantry of the Elekta Synergy research accelerator at the Christie in order to study the breathing motion of volunteers and patients in the context of CBT-IGRT. 5.3

Results

Figure-4. Structured light fringe pattern, (top left) . Surface height map from spatial phase (top right). Discrete sequence showing deviations from the mean surface moving periodically appearing as a 'sheet' best viewed head downward (bottom). The upper and lower scale-lines are 10mm either side of the mean plane represented as a central line at Omm.

Figure 4 shows a video frame of a single, wide-field interference pattern projected on to the body surface (top left) and the rendered height map (top right) computed from the corresponding spatial phases. This particular example is taken mid-inspiration. The shallow illumination used for rendering exaggerates the perturbing effect of dark body hair, which stands somewhat

740

variably above the skin surface itself, and the potential for researching filtering optimisation prior to spatial phase computation. By computing the mean surface height map over all the video frames and then subtracting this from the height maps corresponding to the individual frames, it is possible to view dynamically the breathing cycle deviations to scale across the body surface. Figure 4 (bottom row) shows a sample sequence viewed from head to foot. The most striking feature is the simplicity of the time varying deviations and the scale of motion above and below the mean surface, which in this case ranges from approximately -8mm to +12mm in a distinctly periodic fashion. Dynamic viewing of the data clearly shows an additional, small amplitude, high frequency cycle of deviations, most visible near to expiration, revealing the effect of cardiac motion. 6.

Discussion and Conclusions

In the space of little more than a decade, radiotherapy has taken on board the Xray imaging advances of the last century. In terms of cone beam technology it has arguably taken the lead. IGRT looks certain to progress into frequent, routine use and thence to combine with other radiation free technologies, such as optical sensing, to emerge as measurement guided radiotherapy. Acknowledgments This work was funded in part by Christie Hospital charitable funds, Elekta Oncology Systems and the European Commission. The authors thank Julie Davies and her research radiographers for their expert clinical assistance.

References 1. 2.

3. 4.

L.A. Feldkamp, L.C. Davis and J.W. Kress. Jnl Opt Soc Am. 1(6), 612 (1984). L.E. Antonuk, Y. El-Mohri, W. Huang, K.W. Jee, J.H. Siewerdsen, M. Maolinbay, V.E. Scarpine, H. Sandler, J. Yorkston. Int Jnl Radiat Oncol Biol Phys. 42(2), 437 (1998). G. Pang, D.L. Beachey, P.F. O'Brien and J.A. Rowlands. Ml Jnl Radiat Oncol Biol Phys. 52(2), 532 (2002) D.A. Jaffray, J.H. Siewerdsen, J.W. Wong and A.A. Martinez AA. Int Jnl Radiat Oncol Biol Phys. 53(5), 1337 (2002).

741

5.

C J Moore, A Amer, T Marchant, J R Sykes, J Davies, J Stratford, C McCarthy, C MacBain, A Henry, P Price and P C Williams, British Jnl Radiol. (2005) (in press) 6. J.R. Sykes, A. Amer, J. Czajka and C. Moore C. Radiotherapy & Oncology (2005) (in press). 7. P. Simpson, A. Amer, J. Sykes and C.J. Moore, Radiotherapy & Oncology 73 (Suppl-1), Abstr-515, S231 (2004). 8. C.J. Moore, T. Marchant T, A. Amer, J. Sykes and A. Henry Radiotherapy & Oncology. 73 (Suppl-1), Abstr-111, S51 (2004). 9. C. Moore, F. Lilley, V. Sauret, M. Lalor and D. Burton. Intl Jnl Rad One BiolPhys. 56(1), 248 (2003).

ALPHA CYCLOTRON PRODUCTION STUDIES OF THE ALPHA EMITTER 211 AT/ 211g PO FOR HIGH-LET METABOLIC RADIOTHERAPY S. MORZENTI, M. L. BONARDI, F. GROPPI, C. ZONA, L. CANELLA LASA, Universita degli Studi di Milano andINFN-Milano, via F.lli Cervi 201 Segrate.I- 20090, Italy E. MENAPACE ENEA, Division for Advanced Physics Technologies, via Don Fiammelli 2 Bologna, 1-40128, Italy Z. B. ALFASSI Department of Nuclear Engineering, Ben Gurion University of the Negev Beer-Sheva, 11-84105, Israel K. ABBAS, U. HOLZWARTH IHCP, Institute for Health and Consumer Protection, JRC-Ispra, via E. Fermi Varese, 1-21020, Italy A series of high specific activity accelerator-produced radionuclides in no-carrier-added (NCA) form, for uses in metabolic radiotherapy and for PET, has been investigated and produced at JRC-Ispra Cyclotron Laboratory. In this study we present, in particular, the NCA 211At/21l8Po (LET = 130 eV.nm"1, ti/ 2 = 7.214 h), produced by 209Bi(a,2n) reaction, with internal spike of gamma emitter 210At (e.g. negligible amount of 210Po as radiotoxic long-lived impurity), for high-LET targeted radiotherapy and immunoradiotherapy. A selective radiochemical separation, based on liquid/liquid extraction, of At radionuclides from Bi target and Po impurities has been developed. High resolution gamma, X and alpha spectrometric techniques have been adopted for quality controls of different radiochemical fractions.

1. Introduction Metabolic radiotherapy represents an effective approach against metastatic diseases originated from different primary tumours. The labelling with radionuclides of monoclonal antibodies (or their fragments) specific for overexpressed receptors or antigens in tumour cells is a favourable strategy. Radioimmunotherapy involves injecting the patient with a specific and selective vector labelled with a radionuclide, with the aim of destroying targeted tumour cells. The P-emitters have demonstrated to be effective in solid cancers larger than 5 mm and radiosensitive lymphomas (i.e. non-Hodgkin's

742

743

lymphomas), but they do not seem suitable for micrometastases or tiny clusters of cancer cells remaining after chemoterapy or surgery. Alpha particles have very-high linear energy transfer (LET) and release their energy over a few cell diameters. Their path length is sufficient to kill tumour cells through double or multiple breaking DNA without necessity of aemitters internalization in cell nuclei, but not so long to affect the surrounding healthy tissue 1_5. In this work we present the production of the a emitter 21l At/ 2 " s Po that is suitable for high-LET metabolic radiotherapy, due to its decay properties: 1. The half-life of 7.214 h, which is reasonably sufficient long for labelling organic compounds and for medical applications; moreover this half-life allows diminishing the problems of radwasters in hospital . 2. The 211At a-branching of 41.80%, which is the highest among of the "medium-lived" At RNs. The further 58.20%, although decaying by EC, also leads to a emission through 2Ug Po. In fact, the EC decay leads to the ultra-short-lived a-emitter 211gPo, having an half-life of 516 ms and an abranching of 100%. As a consequence the "overall" a branching of the decayof 21, Atis 100%. 3. The 2 main a particles ( E a = 5869 keV and 7450 keV)6 of the 211At -> 211gPo system have an "average" range of 60 to 67 um in water (and soft animal tissues), and a nearly optimal LET*, of some 130 eV-nm"1, which is around the maximum of the Q curve for energetic ions. As a crude estimate, the 2il At -» 2l,g Po system could impart to a tissue the considerable dose of 4.1 cGy.kBq"1 (i.e. 1.5 Gy.uCi"1) to soft tissues. 2. Materials and Methods 21

'At has been obtained via 209Bi(a, 2n) reaction by oc-cyclotron irradiation, using a energies larger than the reaction threshold of 20.72 MeV. Bi targets were irradiated by using a particles beam with energy higher than 28.61 MeV, in order to produce also small amounts of the j emitter 210At, via 209Bi(oc, 3n) reaction. The presence of a small amount of 210At does facilitate the radiochemical processing of irradiated target, because 2 At/ BPo is an almost pure a emitter, with emission of X-rays of energy not optimal for high-purity germanium (HPGe) spectrometry. 210At decays on the long-lived impurity 210Po, which can settle irreversibly in the body tissue with undesired radiation damage; nonetheless the activity produced under the irradiation conditions adopted, is negligibly small compared to 2II At one, due to the favourable ratio of the halflives. To check the reliability of the procedure and to measure the production thick-target yield (TTY), irradiations of 209Bi targets at two a energies (28.8

744

MeV and 32.8 MeV) were carried out at the Scanditronix MC40 cyclotron (K=38) of JRC-Ispra of the European Commission, that can deliver proton or alpha beams, with variable energy up to 38 MeV (and deuteron beams up to 19 MeV) and a beam current up to 60 uA. Gamma and X spectra were collected at LASA-INFN with several coaxial HPGe detectors of 50 cm3 active volume, connected to Ortec MCAs (mod. 918 and 919A), with resolution of 2.3 keV (FWHM), efficiency of 15% relative to 3"x3" Nal(Tl) and a Compton to peak ratio of 30:1 at the 1 332 keV photopeak of 60Co. HPGe spectrometry also offered the possibility to obtain the typical 4 peak X-ray spectra (76.9, 79.3, 89.6 and 92.4 keV) of Po, even if the autoabsorption did not allow a quantitative determination. Alpha spectra were measured with: • a Si surface barrier detector (EG&G, Ortec, 600 mm2), with a resolution of 27 keV (FWHM). In this case, to improve the resolution, plastic collimators of 5 mm in diameter were put in between the sample and the detector; • two different liquid scintillation counting and spectrometry systems: a conventional LSC system with Horrocks number capability (Beckman, USA, mod. LS5000TD) and an higher-resolution liquid scintillation portable spectrometer with oc/p pulse shape analysis (PSA) discriminator (Hidex, Finland, mod. TRIATHLER). 3. Radiochemical Separation and quality control In order to produce NCA 2nAt/211gPo for metabolic radiotherapy, a suitable radiochemical separation of the At radioisotopes from Bi target and Po byproducts is mandatory, without voluntary addition of isomorphous carrier. The radiochemical separation adopted is a "wet" method based on liquid/liquid extraction, carried out in transparent Teflon PFA funnels. The radiochemical processing of Bi target was based on target dissolution in oxidizing acidic media, followed by liquid/liquid extraction of 211(210)At in organic solvent and back-extraction in aqueous-basic medium. The extraction of the astatine in water phase is however better since the presence of an organic solvent turns out disadvantageous during the labelling phase. The wet chemistry method developed allowed to obtain very pure 2,1 At or 21l 210 ' At fractions, according to the irradiation energy; no 2l0Po was co-extracted in the organic phase (as shown in fig. 1 and fig. 2). The 2ll(210>At present in the organic phase (fig.l) and the 210Po present in the aqueous phase (fig.2) was electroplated on Ag foils and a-counted with a Si surface barrier detector. Back-extractions was processed with different concentration of NaOH, in a range from 0.009 M to 2.256 M, to study At behavior in different basic condition. In fig. 3, in which the results are summarized, we can see that we obtained a back-extraction yield greater than 90 % in a range from 0.75 M to 2.00 M.

745

'"Po . S304J8 keV

"«Po 7450.3 keV kcounts.channcl

80

"'At 5869.5 kcV

1

. 20

J

5000

J1

5500

.

energy
Figure 2. a spectrum of the aqueous pha

III

Figure 1. a spectrum of the organic phase.

«'

9080-

T •

4

/ i

\

t

\

5- 70%

60-

£

50.

|

30,

»

20-

r

22MeV 32MeV Bftcfc-oOract

J*

10-

CoacentratJon oTNaOH (M)

Figure 3. Back - extraction fit of experimental data.

The measurements of a spectra are very selective giving highly resolved peaks, but require a time consuming procedure during the depositing stage, with a low overall recovery yield. On the contrary, the liquid scintillation technique permitted to measure in a very short time a activities with very high efficiency (about 100%) and without selective effects, as a real sample of the extracted solution is used. In this way it was possible to reduce the time of measurements, especially mandatory for short-lived radionuclides and to reduce the contamination doses to personnel, although on account of less well resolved spectrum. 4. Results and conclusions The activity measured has confirmed that the radiation conditions can be set up in order to obtain high activity of 2U At and, at the same time, a wished amount of the 210At spike (1 to 7%); in fact the choice of an incident energy of 28.8 MeV allows obtaining an 210At activity negligibly equal to approximately 0.001% of that 211 At one.

746

For the 28.8 MeV irradiation our experimental thick-target yield measured7'8, before any chemical treatment, from the y spectrum of the irradiated Bi target was 8 085 ± 176 MBq-C"1. This value agrees pretty well with the value obtained by integration of the literature data for the excitation functions of this reaction. The integration of the excitation functions from 28.8 down to 20 MeV, is 8 341 MBq-C"1 and the discrepancy of the order of 3% between our thicktarget yield and the integration value is well within the experimental error9. For the 32.8 MeV irradiation, the thick-target yields were found to be 13 018 ± 416 and 974 ± 20 for 211At and 210At respectively. The integration of the experimental excitation functions (32.8 to 20 MeV) yields 17 720 and 1 078 MBq-C"1, respectively. In this case the comparison with the integrated values has turned out in smaller agreement because of an introduced systematic error during the irradiation due to the partial melting of the target. In this case the irradiation was suitable for setting up both radiochemical separation and spectrometric measurements. The wet-chemistry method is effective for production of very high-specific activity 21l(2l0)At/211gPo, characterized by a radionuclidic purity close to 100 %. In agreement with the energy of the beam, during the separation some pg or fg of At have been separated from 0.200 - 0.250 g of Bi target with an overall radiochemical yield variable regarding the chosen concentration of NaOH. For example in the experiment with NaOH 1.88 M we obtained an overall radiochemical yield of (73.60 ± 3.69) %, and we had separated 98.17 fg of 211At (7.46 kBq, due to a AS(CF) of 76 MBq-ng"1) from 0.265 g of massive Bi target. References 1. R. H. Larsen, B.W.W., M. R. Zalutsky, Appl. Rad. hot., 47(2),135 (1996). 2. G. Vidyanathan, M.R. Zalutszy, Phys. Med. Biol., 41 (1996) 1915-1931. 3. S. Mirzadeth, Appl. Rad. hot, 49(4) (1998) 345-349. 4. Unak, T., Appl. Rad. hot., 58 (2003) 115-117. 5. R. M. Lambrecht, S. Mirzadeh, Int. J. Appl. Radiat. hot., 36(6), 443 (1985). 6. R.B. Firestone, CM. Baglin, F.S.Y. Chu, Table of Isotopes, 8-th Ed., Update on CD-ROM, John Wiley and Sons, New York, USA, 1998. 7. F. Groppi, M.L. Bonardi, C. Birattari, E. Menapace, K. Abbas, U. Holzwarth, A. Alfarano, S. Morzenti, C. Zona, Z.B. Alfassi, Appl. Rad. hot. (2005) in press. 8. C. Birattari, M. Bonardi, F. Groppi, A. Martinotti, S. Morzenti, S. Ridone, M. Acchini, K. Abbas, A. Alfarano, U. Holzwarth, E. Menapace, Quart. J. Nucl. Med. Mol. Imag, 48 (suppl.l to n. 3) (2004), in press. 9. E. Menapace, C. Birattari, M.L. Bonardi, F. Groppi, S. Morzenti, C. Zona, Proc. of Intern. Conf. on Nuclear Data for Science and Technology, Santa Fe, USA, American Institute of Physics, (2004) in press.

TREATMENT PLANNING WITH IVIS IMAGING AND MONTE CARLO SIMULATION* RAFFAELE NOVARIO* Dept of Medical Physics, University oflnsubria Varese VA, Italy RITA LORUSSO, CARLA BIANCHI Dept of medical Physics, University Hospital Varese VA, Italy FABIO TANZI Dept of medical Physics, University Hospital Varese VA, Italy MARIO VESCOVI Research Systems Agrate Brianza, MI, Italy MARCO ROVERE, CHIARA CAPPELLINI Dept of Physics, University oflnsubria Como, CO, Italy MASSIMO CACCIA Dept of Physics, University oflnsubria Como, CO, Italy LEOPOLDO CONTE Dept of Medical Physics, University oflnsubria Varese VA, Italy

' This research is supported by the European Commission under the contract G1-CT2001-00561SUCIMA.

747

748 The vessel wall is the planned target volume in intracoronary brachytherapy. The success of the treatment is based on the need of delivering doses possibly not lower than 8 and not higher than 30 Gy. An automatic procedure in order to acquire intravascular ultrasound images of the whole volume to be irradiated is pointed out; a motor driven pullback device, with velocity of the catheter of 0.5 and 1 mm/s allows to acquire the entire target volume of the vessel with a number of slices normally ranging from 400 to 1600. A semiautomatic segmentation and classification of the different structures in each slice of the vessel is proposed. The segmentation and the classification of the structures allows the calculation of their volume; this is very useful in particular for plaque volume assessment in the follow-up of the patients. A 3D analyzer tool was developed in order to visualize the walls and the lumen of the vessel. The knowledge, for each axial slice, of the source position (in the center of the catheter) and the target position (vessel walls) allows the calculation of a set of source-target distances. Given a time of irradiation, and a type of source a dose volume histogram (DVH) describing the dose distribution in the whole target can be obtained with a Monte Carlo simulation. The whole procedure takes few minutes and then is compatible with a safe treatment of the patient, giving an important indication about the quality of the radiation treatment selected.

1. Introduction Coronary artery diseases are the leading cause of morbidity and mortality in the western world. Re-establishing a stable and normal artery cross section (lumen) is the primary goal of angioplasty. Clinical studies indicate that intra-arterial irradiation reduces substantially the problem of restenosis1'2. This is particularly proven in the case of intra-stent restenosis3'4. It is estimated that the re-restenosis rate may drop from the original 50-60%) below 25-30% if radiation is delivered to the obstruction site during or after angioplasty. The intravascular radiotherapy is performed on a few centimeters long section of the vessel and it is usually accomplished either by multiple point-like radioactive sources or by a coilshaped radioactive wire tip moving inside a catheter or by wrapping with a radioactive foil the angioplasty balloon. During the radiation treatment, the patient will have significantly reduced arterial blood flow, therefore, in order to reduce risks of complications, dose rates greater than 5 Gy/min would be optimal, with a total integrated dose of 15-20 Gy. These high dose rates require P-emitter activities in the GBq range. The question of the optimal choice of the radionuclide is still open but P-emitters like 144Ce, 32P and 90Sr/90Y offer clear advantages in terms of radiation safety. Intravascular brachytherapy is characterized by steep dose gradients, thus three dimensional dose distributions have to be planned with high spatial resolution. At the moment, almost all the treatment are planned only measuring a set of diameters of the vessel lumen using angiographic imaging and thus assessing a radiation dwell time using as a reference a given dose at 2 mm from the source. An optimization of the treatment planning must take in account the geometry of the target volume (that is the vessel wall) and the position of the source. The planning must be fast enough to be done during the angioplastic procedure

749

without any additional problem for the patient. A method is proposed for intravascular brachytherapy planning, using intravascular ultrasound imaging (IVUS).

2. Material and methods Intravascular ultrasound is a technology that is used within blood vessels. In case of coronary arteries, it provides the possibility of imaging from within the lumen. In addition to its role in diagnosis and determination of vessel injuries, IVUS has evolved to interventional cardiology, particularly in the assessment of the appropriate placement of intracoronary stents and in general in Percutanueous Transluminal Coronary Angioplasty (PTCA). Intravascular ultrasound images the vessel from inside visualizing the lumen, the vessel walls, and any structure eventually present in the vessel (i.e. plaque, stent, etc.).

It can depict cross-sectional or longitudinal or three dimensional luminal images and the shape and the thickness of the various layers of the walls like the intima, media and adventitia. Although angiography is the most commonly used method for assessment of coronary artery disease, it has a number of limitations. The main limitation of angiographic imaging in case of brachytherapy treatment planning is the fact that wall structures are not imaged anyway. In any case IVUS provides additional complementary information to that provided by conventional angiography. It should be viewed as an adjunct to angiography rather than as a replacement. The IVUS (Endosonics Jomed© inc., Rancho Cordova, CA, USA) intravascular imaging system was used. It is an electronic phased array acquisition system using a circular array of 64 transducers. It is placed in a catheter 135 cm long and 3.2 F in diameter (1F=1 French=0.292 mm; this unit is commonly used in cardiology). It can travel over a guidewire 0.014" in diameter. The transducers center frequency is 20 MHz, with a bandwidth 15-25 MHz; the image display frame rate is 30 Hz and the image storage frame rate is 10 Hz. The images of the vessel cross section are stored in standard DICOM 3.0 format; image dimensions are 384x384 pixels with a depth of 8 bit (147 kb per image). Using angiography, it is possible to move the IVUS catheter from the femoral artery to the coronary to be studied, passing the whole segment to be investigated. From this point in the vessel, an automatic motor driven pull back device (Track Back™, Endosonics Jomed© Inc., Rancho Cordova, CA, USA) is used to acquire images of the vessel part to be treated; it allows velocity of 0.5 or 1 mm/s. A fast treatment planning system (TPS) was developed using the Interactive Data Language (IDL, ©) and the stored data.

750

A longitudinal view of the vessel at 4 different cut angles (0°, 45°, 90°, 135°) is calculated. On each view the operator trace 4 lines corresponding to the upper and lower vessel wall and the upper and lower lumen of the vessel (if different because presence of a plaque). At the end of this tracing, on each axial IVUS image 8 control points are so identified, describing the vessel wall (and the lumen). An automatic segmentation tool with the use of an elliptic template is used. After this vessel wall (or lumen) segmentation, a new window is used in the longitudinal view, in order to fix the position (beginning and end) of the stent and of the source. The position and the length of the source is chosen by the cardiologist and the radiotherapist. A 3D representation of all the structures is possible. This is an useful tool for the cardiologist and the radiotherapist. The information regarding the geometry of the vessel and the position of the source is transferred to the GEANT4 framework for the MC simulation. 3. Results and discussion The proposed segmentation is based on the marking of points on each slice and a fitting of the points. The error associated with this procedure can be assessed. Five patient were studied in the carotid artery both with angio-CT (with contrast agent) and IVUS, being the CT scan a gold standard for the evaluation of the geometry of the carotid. The comparison of the minimum and maximum (external wall to external wall) diameter of the carotid in a segment 5 cm long shows that the difference between CT and IVUS is always below 7%. Given a type of source (isotope, activity, length, number and type of seeds, etc.), given a dwell time and assuming that the position of the IVUS catheter is the maximum likelihood position of the source catheter during the treatment, a dose volume histogram can be calculated. The procedure gives also information about the plaque volume. The calculation of the plaque volume is another important tool in the follow-up of the patient in order to evaluate the evolution (plaque volume increasing or decreasing) of the disease and the degree of efficacy of the brachytherapy. An accurate computation of the dose is possible computing the dose released through a Monte Carlo method. The program is implemented on Geant4 libraries: the geometry resulted from the segmentation procedure is imported as a set of Geant4 Volumes and different material are implemented (e.g. water/blood, calcified plaque, steel, plastic). In the simulation free electrons, with a given energy spectrum distribution, are generated inside the virtual source. The geometry is imported directly from the visualized artery. The segmentation program produces an ASCII file with the following information: for each slice the two axis, the center and the rotation angle of the ellipses describing the lumen and the adventitia. In addition the starting point and the ending point of the stent along the axis is imported. The stent is

751 modelized in the simulation as a set of steel wires lying between the lumen and the adventitia. The final geometry is a realistic description of the radioactive source inserted through a catheter, a lumen, a calcified plaque in which the stent is situated, and the adventitia. The Geant4 application can run as stand alone program and read any file describing an arbitrary geometry (mainly for checking purpose). A simulation of 100,000 of generated beta particles computes the dose released by the ionising particles with a precision of about 10%. On a Athlon based PC, 1.7 GHz clock frequency, with 1 Gbytes of RAM this operation takes about 1 minute without any particular optimization. This leads the possibility to use the Monte Carlo simulation directly for dose computation as a module of the segmentation program. Given a type of source and a dwell time, it is so possible calculate the dose at any point in the target and to show a dose-volume-histogram (DVH). Changing the dwell time other DVH are obtained until the treatment is judged optimized (all the doses between 8 and 30 Gy).

4. Conclusions The proposed method for the fast treatment planning in intravascular brachytherapy is useful in order to assess the quality of the selected type of source and dwell time and to better describe the geometry of the vessel to be treated. It takes only few minutes being so compatible with the angiographic procedure. It improves the dosimetric optimization of the treatment as regards the only angiographic approach and gives also the possibility to quantify the plaque growth in the follow-up of the patients. The main limitations of the method are the DVHs averaging on the cardiac cycle and the fact that the correction of the dose due to the presence of stent and/or plaque is done only on a subset of point of the target. References 1. 2. 3. 4.

J. Wiedermann et al, J. Am. Coll. Cardiol. 23, 1491 (1994). R. Waksman et al, Circulation. 92, 3025 (1995). R. Waksman et al, Circulation. 101, 2165 (2000). R. Waksman et al, Circulation. 101, 1895 (2000).

MONTE CARLO SIMULATIONS OF A HUMAN PHANTOM RADIO-PHARMACOKINETIC RESPONSE ON A SMALL FIELD OF VIEW SCINTIGRAPHIC DEVICE.

BURGIO N , CIAVOLA C , SANTAGATA A. ENEA FIS-ION, CR-Casaccia, Rome Italy IURLARO G., MONTANI L , SCAFE R. ENEA Mat-NANO,

CR-Casaccia, Rome Italy

The limiting factors for the scintigraphic clinical application are related to i) biosource characteristics (pharmacokinetic of the drug distribution between organs), Detection chain (photons transport, scintillation, analog to digital signal conversion, etc.) Imaging (Signal to Noise ratio, Spatial and Energy Resolution, Linearity etc) In this work, by using Monte Carlo time resolved transport simulations on a mathematical phantom and on a small field of view scintigraphic device, the trade off between the aforementioned factors was preliminary investigated.

1.

Introduction

The functional nature of the medical Nuclear Imaging allows early disease diagnosis, therapeutic follow up, pharmacological assessment and radio-guided surgery [1-5]. A scintigraphic image is basically a map of the detected radioactivity distribution induced by the perfusion and uptake of a hot labeled biomolecule in a target organ. The detectability of a pathological condition is strongly influenced by several factors and a multidisciplinary approach is requested to properly define the problem. As observed elsewhere [6], the intrinsic resolution of a imaging device is related to the radiation transport stage. Knowing the prior, a proper model of the device's transport stages could be set up [7-8]. Detection and imaging stages could be investigated and spreads could be limited by an appropriate development cycle [9] that also include simulations and definition of the various stage in a Quantum Account Diagram [10]. In the following, the simulation of the entire system (i. e. Patient prior, scintigraphic device, nuclear image rendering of the final posterior) was set up in order to state which stages can be easily evaluate, what must be measured and finally

752

753 what is partially unknown and need further investigation.

Figure 1. Phantom cross sectionfromMCNP plotter: a) Side view of the phantom, b) Plant view. The center of the Imager field of view is aligned with the prostate center (see text). 2.

Model Assumptions and Calculations method

Following the detailed description reported in [12], the Oak Ridge mathematical Phantom was totally implemented in a MCNP [13] input deck (see fig. 1) along with the Imager components relevant for gamma photons transport (see fig. 1 and [9] for detail). In order to defines the distributions of a 99mTc (t1/2=6.01 h, Er=140 keV) labeled drug between organs, a proper pharmacokinetic compartment model was set up (see Appendix). The source strength (photon sec"1) is expressed as:

S(x, y, z, t) = N ^

P, (*, y, z, E)C, (0

(1)

the summation was on perfused organs, N is Avogadro Number and E is the y emission energy. Pj(x,y,z,E) is probability that a y photon of energy E is emitted at the point (x,y,z) of the i-th organ: y emission are uncorrelated . Q(t) is the drug concentration at time t into of the organ i. It was assumed that the drug concentration is referred to the entire volume of the correspondent phantom. The first stage output is a mapping of S trough y energy deposition at the matrix detector pixels:

H(i,j,t,E')= fa jS(x,y,z)dr

I

(2)

Where Ty is the probability matrix that a y photon (or its progeny), emitted at a

754

given point source with energy E, undergo interaction with a given crystals element (at column j , row i) of the matrix detector depositing here an amount of energy E' and V is the phantom volumes. Tys were estimated by using the so called "pulse height tally" of MCNP. In the second stage visible light photons, from the scintillation loci, were transported along the crystal. It was demonstrated that, for each single photon impinging a small size parallelepiped crystals, the Scintillation Light Distribution (SLD) is defined as Lorentzian bivariate curve of order q [14] of crystal pixel axis coordinates. SLD is the final light distribution (second stage output) for each event. In the third stage an NxN crossed strip anode sensor is considered and the charge readout is performed strip by strip. The stages 2 and 3 - were simulated by using a semi-empirical analytical code DISIS (Discrete Scintillation Imager Simulator) developed by our team [15]. The output from DISIS simulation was integrated over a finite time and energy intervals to obtain the final posterior output:

I(i,j,m)= t+2 tA>H(iJ,T,E')drdE'

(3)

that is the estimation of the expected number of event in given pixels in the time step m of amplitude At. Often, to facilitate physician interpretation, the so obtained scintigraphic images were smoothed at 128x128 pixels and reported in a suitable gray-scale proportional to the count intensity.

3.

Results and Discussions

Three different cases, belonging from the same "marnmilary-like" pharmacokinetic (see Appendix), were evaluated. In case 1, Patient has a spherical prostatic tumor (0.3 mm of radius) located exactly in the center of the

V=Imin

t~2mta" ' "

t = iamin

Figure 2. Drug distribution vs time in the different organs for the case 1. Correspondent Imager

755 response at different time (1,1.8 and 10 min from drug administration).

prostate. In case 2 Patient's prostate doesn't show any pathological evidence (no tumor). Case 3 the same tumor was misplaced of 6 mm respect to prostate center. Figure 2 reports the kinetic drug distribution in the various organ along with Imager response for case 1. The images were obtained by using instantaneous values of concentration from the compartment models whereas, in real experiments, the Imager requests a fixed time integration period to accumulated a significant count statistic. Images deals, respectively, with early perfusion time (1 minute), maximum drug accumulation in lesion (1.8 minutes) and clearance dominated regime (10 minutes). Central images (1.8 minutes) denotes a tumor intensity maximum well resolved respect to the underlying normal prostatic tissue emission. The large halo in the upper image zone is due to the progressive drug accumulation into bladder. Figure 3 reports the concentration-time histograms obtained with different time steps: the transition from 10 seconds to 5 minutes time steps clearly indicates a progressive loss of kinetic features. Conversely, images gain in feature sharpness with the increase of the counting time.

Figure 3. Drug concentration integrated over various time step (from left to right At=10,20 sec. and 5 min.) for the case 2. Correspondent scintigraphic images.

Figure 4a reports the comparison between a series of images from case 2, case 1 and case 3 at the same acquisition time (1.8 min.). Case 1 (Fig. 4a, central image) clearly shows a maximum correlated with the tumor uptake, whereas in case 2 (Fig. 4a, left image) the response is compatible with the typical uniform spherical distribution activity. Finally case 3 (Fig. 4a, right image) shows, due to the 6 mm displacement of the tumor center, a spot like maximum slightly shifted from the center of the FOV. This last result can be explained by the geometric constrains imposed by square collimator septa and crystals network. In case 1, the source center is coincident with the intersection of four collimator septa (see

756

fig. 1), and the tumor signal is realized by 4 crystals equally impinged by photons. In case 3 the tumor is displaced the 4 pixels were not equally impinged by the photons and the weakest pixels signals where hidden in the "noise" of the normal prostatic tissues emission. As consequence tumor spot was reduced to sub-crystal dimensions. Figure 4b reports the previous outcomes after a "reverse" treatment with DISIS code. The introduction of stages 2 and 3 yield in a generalized lack of information that made equivalent the posteriors of the three cases.

Figure 4. a) Comparisons (left to right) between normal tissues (case 2), centered (case 1) and misplaced tumor (case 3). b) Same images sequence after the simulation of the light guide and readout answer functions with DISIS code.

Conclusions As hypothesized in a previous work [9], the method confirms that the visible light transport and charge readout stages spread the intrinsic resolution of the system. The time dependent outcomes show that the kinetic modality requires different acquisition time than the diagnostic modality. Works are in progress to a performs a detailed quantitative analysis of the model. Future experimental work, as indicated by the present results, will be address on the improvement of light guide and electronic readout. Appendix The pharmacokinetic model applied in this paper was inspired by the general concept of linear, multi-compartmental, mammillary model [11]. The differential equations system constituted by eqs. A. 1 were symbolically solved with Laplace transforms method by using the open source software Maxima [16] for a given set of kinetic constants. The first Equation refers to the bolus injection modality that is represented as an Heaviside function. (Heart=H, Intestine=I, Liver=L, Kidney=K, Prostate=P, Bladder=B, Prostate=P, Tumor= T,

757

Dose= D0).

D0 (t) =

(Heaviside(t) - Heaviside(t- At)) / At

^(t) = at

-(K I +K I H )I(t) + KfflH(t)

—(t)=

-(K m +K H L +K H K )H(t) + KIHI(t) + KLHL(t) + K KH K(t)-D 0 (t)

^(t)=

- K ^

+ K^t)

dK — ( t ) = -(K KP +K KH )K(t) + KHKH(t) dt ~(t) = -(K M +K T P )T(t) + KPTP(t) dt ^(t)= dt

(A 1}

'

-(K P T +K r a )P(t) + K11)T(t)

JD

—(t)= dt

-K B B(t) + KPB(P(t)

References 1. R. Pani, et al., Nucl. Instr. and Meth. A 477 (2002) 509 2. F. Scopinaro, et al., Tumori 86 (2000) 329. 3. F. Scopinaro, et al., Nucl. Instr. and Meth. A 497/1 (2003) 105. 4. A. Soluri, et al., Nucl. Instr. and Meth. A 497/1 (2003) 114. 5. A. Soluri, et al., Nucl. Instr. and Meth. A 497/1 (2003) 122. 6. A. Rose, J. Opt. Soc. Am. 38 (1948). 7. H. Zaidi, Med. Phys. 26 (1999) 574. 8. H. Zaidi, IEEE Trans. Nucl. Sci. NS-47 (2000) 2722. 9. N. Burgio, et al., Nucl. Instr. and Meth. A to be published 10.1. A. Cunningham, et al., Med. Phys. 21 (3) (1994) 417. ll.Makoid Michael C , et al., "Basic pharmacokinetics [electronic book]" Omaha; Creighton University, 1996-1999. 12. Eckerman K F, Cristy M and Ryman J C 1996 The ORNL Mathematical Phantom Series (Oak Ridge, TN: Oak Ridge National Laboratory). 13. J. F. Briesmeister (Ed.), LANL Report, LA-13709-M. 14. R. Scare, et al., Nucl. Instr. and Meth. A 497 (2003) 249. 15. R. Scafe, et al., 2004 IEEE NSS-MIC Conference Record. 16. http://maxima.sourceforge.net.

APPLICATIONS OF THE MONTE CARLO CODE GEANT TO PARTICLE BEAM THERAPY H. SZYMANOWSKI, T. FUCHS, S. NILL, J.J. WILKENS, D. PFLUGFELDER, U.OELFKE Dept. of Medical Physics, German Cancer Research Center, Heidelberg, Germany Y. GLINEC, J. FAURE, V. MALKA Laboratoire d'Optique Appliquee, ENSTA-CNRS-Ecole Polytechnique, Palaiseau, France

We report on the use of the Monte Carlo simulation code GEANT for two different applications in the field of particle beam therapy. The first application relates to the planning of intensity-modulated proton therapy (IMPT) treatments. An important issue is thereby the accurate prediction of the dose while irradiating complex inhomogeneous patient geometries. We developed an improved method to account for tissue inhomogeneities in pencil beam algorithms. We show that GEANT3 can be successfully used to validate the new model before its integration in our treatment planning system. Another project concerns the investigation of the potential of high-energy particles produced by laser-plasma interactions for radiotherapy. GEANT4 simulations of the dosimetric properties of an experimental laser-accelerated electron beam were performed. They show that this technique may be very attractive for the development of new therapy beam modalities such as very-high energy (170 MeV) electrons. 1. Introduction The aim of radiation therapy is to deliver a lethal dose to the tumor volume while sparing the surrounding normal tissues and organs at risk. New radiation delivery techniques have been developed in order to increase the capability at applying complex dose patterns. One of the most advanced techniques is the intensity modulated proton therapy (IMPT) [1]. For the planning of intensity modulated proton therapy treatments, the dose calculation engine plays a crucial role not only in the assessment of the treatment plan quality but also in the determination of the optimized particle fluence. An important issue is thereby the accurate prediction of the dose while irradiating complex inhomogeneous patient

758

759

regions. In our group, we have developed an improved proton dose calculation algorithm with increased accuracy in presence of inhomogeneities. The new algorithm has been implemented into our treatment planning system in order to evaluate the effects of patient inhomogeneities on IMPT treatment planning system. In this paper we first present the role of the Monte Carlo code GEANT3 [2] for the assessment in inhomogeneous phantoms of the new proton dose algorithm and of the conventional one currently used in the clinics. We also suggest a procedure to use GEANT4 [3] as a benchmarking tool for IMPT treatment planning in order to validate the results of the dose algorithms under complex treatment conditions and patient geometries. The third application of GEANT presented in this work is related to the investigation of new particle beam modalities which could have a significant impact in radiation therapy. Besides proton therapy, our group is currently interested in the study of particle beams generated by laser-plasma interaction. In this paper we finally present the results of simulations with GEANT4 of the dosimetric properties of a very-high energy (170 MeV) electron beam generated by laser-plasma interaction. 2. Assessment of two proton dose models 2.1. Basic principles Two methods to account for tissue inhomogeneities with proton pencil beam algorithms have been assessed by using Monte Carlo simulations with the GEANT3 code. Starting point is the dose distribution for a proton pencil beam computed in water. The corresponding dose distribution for a heterogeneous medium is derived from the dose for a homogeneous water phantom via two different scaling methods. The first approach is the well known one-dimensional (ID) pathlength scaling applied along the central axis of the pencil beam. Only the energy loss due to the material traversed is considered, but neither the depth nor the scattering properties of the inhomogeneities are taken into account. The lateral spread of the pencil beam at a given depth in a heterogeneous medium is simply equal to the spread in water at the depth where protons have the same residual energy. We developed a new proton dose algorithm which is based on a twodimensional (2D) scaling of the dose distribution in water [4]. It combines a pathlength scaling with an additional lateral scaling. The factors influencing the lateral scaling are derived from Highland's and Gottschalk's formulas for the description of multiple Coulomb scattering for protons [5]. This scaling approach depends on the radiation lengths and the stopping powers of the traversed materials, and also on the depth location of the inhomogeneities.

760 2.2. Comparison to Monte Carlo

simulations

Dose distributions were computed for both scaling methods in homogeneous non-water media and for various slab geometries. The results were compared to Monte Carlo simulations performed with the GEANT3 code. A good agreement between the results of the new (2D) scaling method and the simulations was observed (less than 2.0 % ) , see for example figure 1. These results were found to improve the conventional (lD)scaling of proton pencil beams, which showed in some cases non negligible dose deviations of maximal 10.0%, e.g., observed for a 2 cm large air cavity.

b!jrfc[K1,12lcm

-

lH£~^ H^S

Mortlf Carlo

turf l ( X l 2 k ' - n

I>Xt^

-

M.;r"ile C a i l u

u"> J

Figure 1: Upper graphs: Monte Carlo simulations of a 160 MeV proton spot impinging in a water phantom containing a 2 cm thick bone slab at 10 cm depth (left) and an air slab of same dimension and position (right). Middle: Deviations between the new algorithm and Monte Carlo. Lower: Deviations between the conventional algorithm and Monte Carlo.

761 2.3. Effects ofinhomogeneities on IMPT treatment planning The new dose calculation algorithm has been implemented into the inverse treatment planning system KonRad for IMPT. In order to integrate the improved dose algorithm into a CT based treatment planning system, a calibration curve relating the lateral scaling factors to the CT Hounsfield numbers has been introduced using a stoechiometric method, additionally to the standard stopping powers to Hounsfield numbers calibration [6], see figure 5. The effects of the dose calculation method on IMPT treatment planning were investigated for patient cases with inhomogeneities such as lungs. Each IMPT plan was determined twice, using the (ID) and the (2D) pencil beam scaling methods, resulting into (ID)- and (2D)-optimized fluence matrices. The final dose distributions based on these optimized fluence matrices were computed with the (ID) and the (2D) proton pencil beam scaling methods. For the (lD)-optimized fluence matrices, severe differences were observed between the final dose distributions computed with both scaling methods, e.g. for the irradiation of a 60 cm3 target volume located in lung by five intensity modulated proton beams, the respective dose distributions showed significant deviations in the order of 10%, see figure 2. In particular in that case it can be shown that the (ID) scaling overestimates the dose within the target volume and therefore leads to a significant underdosage of the target volume.

Figure 2. IMPT dose distributions in a lung tumor using as input the proton fluence distribution optimized using the conventional dose algorithm. Left: the conventional ID scaling algorithm was used to calculate the final dose distribution. Right: the new 2D scaling algorithm was applied. The dashed line corresponds to the 90% isodose curve. Figure 3 s h o w s that using t h e ( 2 D ) s c a l i n g m e t h o d for the o p t i m i z a t i o n p r o c e s s

leads a more appropriate coverage of the target volume.

762

Figure 3: Left: Proton fluence matrix after optimization using the (ID) scaling method (upper left) leads to a poor coverage of the target volume (lower left). Right: Proton fluence matrix after optimization using the (2D) scaling method (upper right) leads to a good coverage of the target volume (lower right).

3. Development of a Monte Carlo benchmarking tool The new proton dose algorithm demonstrates a good agreement with the Monte Carlo simulations using GEANT3 in simple phantom geometries. However the significant results presented in the previous section need to be validated as well, since they are obtained under more complex CT-based patient geometries. For this purpose, we suggest to use the simulation code GEANT4, which is more flexible for the implementation of complex geometries. In this section we present a procedure which allows to develop an IMPT dose benchmarking tool based on GEANT4 Monet Carlo simulations on CT data. 3.1. Verification of Monte Carlo simulations in biological media Monte Carlo codes are a tool of choice for the validation of dose calculation for proton therapy, provided that the physics of particle transport in biological tissue is appropriately simulated. Hence a thorough validation of the performances of a Monte Carlo code using measured data or exact analytical predictions is needed prior to the use of the code as a benchmarking tool. Coulomb interactions have a predominant effect on the dose deposition pattern

763

of protons used in radiotherapy. Before applying GEANT4 as a benchmarking tool for IMPT, we suggest to verify the ability of two versions of the Monte Carlo code GEANT4 at simulating Coulomb interactions of protons in water and in biological materials. Two versions of the Monte Carlo code GEANT4 were tested (version 6.2.p02 and the current version 7.0) . The electromagnetic and hadronic physics processes likely to occur for high-energy protons were implemented for our simulations. Furthermore homogeneous phantoms filled with biological tissues were designed as simulation geometry. Because of the perspective to use GEANT4 as a dose engine based on CT data, the chemical compositions of the biological tissues were chosen so as to span the range of the conversion curves relating CT-Hounsfield numbers to the fractions of chemical components of biological materials developed by Schneider et al. [7] Two-dimensional dose distributions in water and in the biological phantoms for proton pencil beams were simulated. While performing simulations in biological media using the former version of GEANT4 (6.2.p02), we found that the proton ranges and hence the stopping powers relative to water were appropriately modelled, however about 16.0% deviations were observed between the values of the lateral scaling factors obtained from the simulations and those calculated analytically. Such discrepancies did not occur while using the current version of GEANT4 (7.0). In figure 4 comparisons between Monte Carlo and Highland formula are shown for the standard deviations of the pencil beam lateral spreads in different bone materials. There again the discrepancies remain below the spatial resolution of the scoring grid. The values of the stopping powers and of the lateral scaling factors extracted from the simulations with GEANT4.7.0 are displayed in figure 5 as functions of the Hounsfield numbers for the various biological tested. The agreement between simulated values and analytical calculations is better than 3.0%. Protons 160 M*V

Protons Ifin \* V

HUlWGfcVlMJO i

u \m6BHti i .

-

•

Hu'l*'0OGEANI4 0 I

•

Hu-ieooewMc?

1

IU 1400 GCAMT4 7 0

j

M-Wir^M w-roGEAtmewa U-WCEANT470 U-7i» *»**)*•
' 1 -

, .. • - • •

U •

,_ ,

,

Figure 4: Depth dose curves and standard deviation of the lateral spread of a 160 MeV proton pencil beam in various bone materials characterized by their Hounsfield number HU. Monte Carlo simulations using GEANT4.6.2p02 and GEANT4.7.0 are compared to analytical predications.

764

Figure 5: Relative stopping powers and lateral scaling factors as functions of the Hounsfield numbers of the biological materials. Results from the Monte Carlo code GEANT4.7.0 are compared to analytical values and the corresponding fitting functions. CT implmentation and physics

3.2. Input data The input data for GEANT4 are the patient geometry and the initial phase space corresponding to the IMPT beam. The patient geometry is described as a volume of voxels built using the CT data. Each voxel is filled with an homogeneous material characterized by its Hounsfield number. The conversion from Hounsfield number to material composition is achieved using the method suggested by Schneider et al. [7] The initial phase space consists of the definition of the initial energy, position, direction and fluence of the proton spots used for the IMPT treatment. The distribution of these information is provided by our treatment planning system KonRad. Using standard sampling methods such as the inverse cumulative probability method, the initial kinematic parameters of each new proton is generated according to the treatment planning distributions. 3.3. Dose scaring and denoising The dose is scored on a grid from the energy deposited at each step by a particle. Unfortunately for a relatively low number of histories the appearance of the simulated isodose curves is quite noisy because of the lack of statistics. The appearance of the isodose curves improves as the number of simulated particles increases. In order to improve this appearance without increasing the simulation time, we apply a denoising method based on the adaptive anisotropic diffusion filtering suggested by Miao et al.[8] The advantage of this denoising method is its speed and its ability to preserve the real high gradient regions. An example is give in figure 6.

765 Prolon Pencil Beam 10^ histories

1g'

250-

w

200-

?

E -IB. a.

1

Proton Pencil Beam (itf 1 histories) after denoisini

?E

| l »

i 100-

100-

1

SO'

-30

mr w

250

-20

-10 0 10 Lateral position (mm)

i

50-

20

3C

-30

-20

-10 0 10 Lateral position (mm)

20

30

Figure 6: Effect of the anisotropic diffusion filtering on proton isodose curves simulated by GEANT4. Left: before the denoising, right: after the denoising.

3.4. Examples Examples of dose distributions in an inhomogeneous phantom for two different IMPT delivery techniques are shown in figure 7.

Figure 7:Dose distributions simulated usingGEANT4 in a inhomogeneous CT phantom. The IMPT proton spots were applied according to two different delivery methods. Left: 3D, right: distal-edge-tracking.

Work is currently in progress in order to assess the results presented in section 2.3 using this CT-based Monte Carlo benchmarking tool. 4. Investigation of the dosimetric properties of very-high energy laser generated electron beams Laser plasma accelerators have been intensively studied during the past decade. It was shown that it is possible to accelerate electrons using a high intensity laser impinging on a fluid or solid target. The laser pulse ionizes the target and

766

creates a plasma. The electrons of the plasma itself undergo an acceleration by the laser pulse. Under some special conditions which are detailed in Faure et al [9], it was possible to generate experimentally at LOA a 170 MeV quasi monoenergetic electron beam. The feasibility for radiation therapy with low energy (6-25 MeV) laser-acccelerated electrons and their dosimetric properties were studied in [10,11]. The results imply that this kind of accelerators may provide an alternative to conventional ones. In particular for heavier and high energy particles such as protons, the compactness of the laser systems and their decreasing installation costs could also compete with larger scale particle accelerators [12]. This is particularly the case for very-high energy electrons, i.e. with energy larger than 50 MeV. Such a beam modality is not yet available for clinical use, however their potential has been shown theoretically by DesRosiers et al [13]. and by Yeboah et al. [14]. We have performed an evaluation of the dosimetric characteristics of the very-high energy electron beam obtained at LOA using Monte Carlo simulations with GEANT4. For the simulations, the high energy peak of the experimental energy spectrum (figure 8) was used, as well as the initial angular divergence of 10 mrad and the electron charge per shot. xllp Ne-Hflo'X-m! 8

-

6

•

4 -

>•

A M I'' \

^1 I iA TkreahoU

V«s—W**

too Figure 8.: Energy spectrum of the electron beam obtained at LOA by laser-plasma interaction. Electromagnetic interaction and electronuclear interactions were accounted for in the simulations. The integrated depth dose curve was simulated and shows a deep penetration depth with a maximum at 20 cm. The lateral dose profiles display a narrow Gaussian shape with FWHM of about 2 cm at 18 cm depth for a source skin distance of 100 cm. In figure 9 isodose curves for an electron bunch sweeping a 5x5 cm2 field are shown.

767

§*&»!***>&£***&*

1M***»*«*)*'*W*a***#

Figure 9: Isodose curves for an 170 MeV electron bunch generated by laser-plasma interaction and sweeping a 5x5cm2 field.

From this study it was found that each electron spot delivers a narrow radial dose profiles, so that the width of the lateral penumbra for broad fields is comparable to that of 15 MV photon beams. The dose delivered is high enough and can be decreased easily. It can be concluded that compact laser systems could offer an novel and probably affordable source for very high energy electron beams in the future. Acknowledgments This work was partially supported by Siemens PT. References 1. T.Lomax, Phys.Med. Biol. 44,185 (1999) 2. GEANT3, CERN Program Library, long writeup W5013 (1994) 3. S.Agostinelli et al., Nucl. Instrum. Methods A506, 250 (2003) 4. H.Szymanowski and U.Oelfke, Phys. Med. Biol. 47,3313 (2002) 5. B.Gottschalk et al., Nucl. Instrum Methods B74,467 (1993) 6. H.Szymanowski and U. Oelfke, Phys. Med. Biol. 48, 861 (2003) 7. W.Schneider et al., Phys. Med. Biol. 45,459 (2000) 8. B.Miao et al., Phys. Med. Biol. 48,2767 (2003) 9. J.Faure et al., Nature 431,541 (2004) 10. C.Chiu et al., Med. Phys. 31,2042 (2004). U.K. Kainz et al., Med. Phys 31,2053 (2004) 12. V.Malka et al., Med. Phys. 31,1587 (2004) 13 .C.DesRosiers et al. Phys. Med. Biol. 45,1781 (2000) 14. C.Yeboah et al. Phys. Med. Biol. 47,1285 (2002)

C H A R G E S H A R I N G I N PIXEL D E T E C T O R S FOR SPECTROSCOPIC IMAGING

L. TLUSTOS UTEF, GTU, Horskd CZ128 00 Praha 2, CZECH

3a/22 REPUBLIK

X. L L O P A R T , M. C A M P B E L L , E . H E I J N E , R. B A L L A B R I G A CERN,

1211 Geneva

23,

Switzerland

Photon counting pixel detectors with several energy bins can provide valuable additional information for imaging applications using spectral X-ray sources. This information is not accessible using integrating detectors systems. In order to profit from this advantage, the actually detected spectrum should match the photonic spectrum as closely as possible. The signal spectrum at the collection electrodes of the readout electronics deviates significantly from the photonic spectrum when planar segmented sensor geometries are used. Simulations of charge deposition and charge collection and measurements using a Medipix2 readout chip bump bonded to a 300 fj,m Si detector are presented. The effects of charge sharing and flourescence photons on image quality are discussed.

1. Introduction Single event processing X-ray detector systems open the possibility to extract additional information from images taken with spectral X-ray sources such as conventional X-ray generators. Provided a sufficient energy resolution and an appropriate number of energy bins, such detector systems could lead to improvements in K-edge imaging techniques, material identification techniques such as p/Z-reconstruction and energy weighting methods in low contrast imaging applications 1 ' 2 . Unfortunately high spatial resolution and excellent energy resolution are conflicting objectives and are difficult to achieve at the same time. 768

769

Enorgy [KeV]

Energy pwV]

(a)

Energy [keV]

(b)

(c)

Figure 1. The spectrum of the energy deposited by 40 keV photons in a a) Si, b)GaAs and c) CdTe sensor of 300 nm thickness within a pixel of 55 and 300 /im width in the center of an uniformly irradiated area.

55x55 fim2 Si GaAs CdTe photopeak events [%] @ 20keV 40keV

84.2 51.8

64.1 50.0

95.6 28.7

300x300 /xm2 Si GaAs CdTe 93.8 69.1

85.6 85.0

98.8 55.4

2. Simulations 2.1. Charge

deposition

In a detector system the first step of image formation is the conversion of incident X-ray energy into electric charge, which is further processed by the readout electronics. Simulations have been carried out in order to quantify the amount of charge detected in a detector element as a function of the pixel size and to estimate the percentage of photopeak events for different pixel geometries. All charge deposition simulations presented were performed using PENELOPE 3 . Fig. 1 shows the spectrum of the energy deposited by mono-energetic 40 keV photons within a square pixel of 55 and 300 itm size in 300 /xm thick sensor of Si, GaAs and CdTe. Whilst for Si the deformation of the detected spectrum is mainly due to Compton scattering, both GaAs and CdTe show significant contributions from fluorescent photons and the corresponding escape peaks. Table 2 gives the percentage of the detected events in the photopeak for 20 and 40 keV photons. It is interesting to note, that due to fluorescence photons even though the high stopping power of CdTe only ~29 % of the photons converted produce a signal that fall into to photopeak.

770

(a)

(b)

(c)

Figure 2. Simulation of the detected energy spectra by a photon counting detector system with a) 300 |im b) 170 fim and c) 55 \an thick Si sensor and an ENC of 150 e for mono-energetic photons of 20 and 40 keV.

2.1.1. Charge collection After conversion of the incident X-ray, the carriers drift under the influence of the applied field towards the collection electrodes. In high resistivity Si, due to its low impurity concentration, charge trapping and recombination of free carriers can be neglected. In the Medipix2 4 readout electronics the signal collected at the input electrodes is integrated with a time constant much longer than the collection time. Therefore it is sufficient to model the motion of the free carriers alone, without taking into account the exact pulse shapes. In the simulation of the collection of the generated free carries in the sensor volume the diameter of an average initial charge cloud, conversion probabilities, Ka and Kp fluorescence photons, Compton and Rayleigh scattering, velocity saturation effects in the drift time and surface losses were taken into account. Fig. 2 illustrates the effects of decreasing pixel size on the spectroscopic response. The electronic noise charge ENC of the readout electronics has been assumed to be 150 e~ rms. For monoenergetic photons of 8 keV the peak to background ratio decreases from 860 at 300 pun pixel pitch to 145 at 55 pbm pixel pitch. For 20 keV photons the respective values are 476 and 51. 3. Experiment 3.1.

Setup

For the measurements we used a hybrid photon counting detector, consisting of a Medipix24 readout chip bump-bonded to a 300 |jm thick p+n silicon sensor diode. The Medipix2 readout is a CMOS readout chip operating in

771 single photon counting mode. Its active matrix of 256x256 cells covers an area of 1.98 cm 2 . Within each 55 fxm square pixel a shaping preamplifier, a double-threshold window discriminator and a 13-bit counter is implemented. The chip can be run either with the lower threshold active of in energy window mode. The measured threshold dispersion over the matrix is <100 e~ rms, and the ENC of a pixel of 140 e~ rms. The lower detection threshold can be linearly adjusted from ~ 3 keV up to ~100 keV. Though it can also process inputs from other detector materials such as GaAs and CdTe, Si was used as sensor material since it is available from industrial suppliers with superior material homogeneity. The detector assembly was glued onto a PCB board, either using Ag-filled or unfilled mounting glue. For the measurements we used a Seifert FK-61-04xl2 X-ray tube with a W-target, a filter of 2.5 mm of Al and a tube voltage of 50 kV. 3.2.

Results

To obtain reference energies in the photon spectrum and to derive an energy calibration for the detector, additional absorbers emitting fluorescence photons were placed on top of the sensor. To determine the detected spectrum first a series of measurements was performed, scanning the lower discriminator threshold over the energy range of interest. The spectrum was then obtained from the gradient of the mean count with respect to the threshold value. Fig. 3a) shows the spectrum obtained with a detector glued to the PCB board using Ag-filled conductive glue. A 30 \j,m thick Zr foil and 2 mm of Iodine solution were placed on top of the sensor. At energies below the .£sTa-line of Zr the tail of charge shared events is clearly visible. Also the .K^-line of I is visible. In addition two other lines at the energies of the Ka-lines of Ag at 22.1 keV and Sn at 25.1 keV are present. Fig. 3b) shows the spectrum obtained with a second detector mounted on the PCB board using Ag-free glue and 30 fim Mo foil and 2 mm of Iodine solution placed on top. In this case the additional peak of Ka-\me of Ag is missing. The Sn .K^-line present in both measurements is attributed to the Sn/Pb bump bond which connects each sensor pixel with its readout channel. The Ag line stems from the conductive glue used to connect the back of the readout chip to the PCB board. 4. Conclusions Charge sharing becomes an increasingly important effect leading to a distortion of the energy resolution as pixel sizes decrease. Additional distortion of

772

K a Mo K a Sn

20

30 40 energy [keV]

(a)

20

KJ

30 energy [keV]

40

(b)

Figure 3. Spectra measured with Medipix2 detector mounted on the PCB board with a a) Agfilledand b) Ag free glue. Note the absence of the Ag XQ-line in b) image quality may be caused by fluorescence photons emitted from structures in close vicinity of the detector, such as bump bonds and support material underneath the readout electronics. In order to minimize these effects, the materials used and their dimensions have to be chosen carefully. Aiming at spectroscopic imaging devices, countermeasures have to be applied to minimize these effects. One possible way is to implement more complex signal processing schemes in the readout electronics, such as charge summing of all pixels involved in the detection of an event. The summed signal can be used to bin the energy deposited by each individual photon in a number of counters, implementing an ADC logic per pixel. In this way the spectrum of the energy deposited can be conserved even with very small pixel sizes. The distortion of image quality by fluorescence photons can be excluded by setting the energy bin accordingly and excluding the contributions from the fluorescence lines from the final image. References 1. J. Giersch, D. Niederlohner and G. Anton, Nuclear Instruments and Methods in Physics Research A 531 (2004) 68. 2. R.N. Cahn et al., Medical Physics 26 (1999) 2680. 3. J. Bar6 et al., Nucl. Instrum. Methods B 100 (1995) 31. 4. X. Llopart et al., IEEE Trans. Nucl. Sci. 49 (2001) 2279, Proc. Conf. Rec. IEEE Nuclear Science Symp. and Medical Imaging Conf. (San Diego, CA, 4-10 November 2001).

DIRECT THICKNESS CALIBRATION: WAY TO RADIOGRAPHIC STUDY OF SOFT TISSUES DANIEL VAVRIK*, TOMAS HOLY, JAN JAKUBEK, STANISLAV POSPISIL, ZDENEK VYKYDAL, JIRIDAMMER Institute of Experimental and Applied Physics of the Czech Technical University in Prague, Horska 3a/22, 128 00, Prague 2, Czech Republic, E-mail: [email protected] The diagnosis of soft tissues is commonly a highly required but complicated task due to low contrast and high distortion of soft tissue radiographs. An important source of radiograph distortion arises from beam hardening effects. When left uncorrected, distortion blurs low contrast radiograph parts. The use of a cooled digital X-ray semiconductor detector together with the proposed beam hardening correction procedure brings high dynamic range and very low noise of the acquired radiographs over the entire X ray tube spectrum. Both soft and hard parts of the object appear with abundant contrast and spatial resolution in the resulting radiographs.

1. Introduction X-ray computer radiography enables the visualization of the internal structure of biological tissues. Standard radiographic systems allow routine imaging of high contrast structures such as bone. However, soft tissues are imaged with very low contrast rendering difficult the observation of inner structures and soft tissues which can even become radiographically invisible. Essentially, soft tissues are radiographically visible with low energy X rays only. High energy X-rays are very weakly attenuated by soft tissue yielding radiographs with extremely low contrast. An effect of relevance that has to be corrected is the so called beam hardening effect which is a significant source of X ray image distortion affecting low density radiograph parts. Using a cooled digital X-ray detector together with the proposed beam hardening correction procedure brings high dynamic range and very low noise of the acquired radiographs over the entire X-ray tube spectrum. Both soft and hard parts of the object appear with sufficient contrast and spatial resolution in the resulting radiographs.

Also affiliated with Institute of Theoretical and Applied Mechanics of the Czech Academy of Sciences, Prosecka 76, Prague 9, Czech Republic

773

774

2. Beam hardening correction Typical X-ray tube spectra are polychromatic. X-ray attenuation processes in matter are energy dependent and low energy X-rays are preferentially absorbed. Consequently, after traversing an object, high energy photons remain preferentially in the beam. For any given initial X-ray spectrum and material type, the phenomenon of beam hardening represents a monotonic increase in beam hardness as a function of matter thickness traversed. The traversing of the sample with thickness variations is accompanied by beam hardness variations from point to point. Moreover, the sensitivity of individual detector pixels depends on photon energy. The beam hardening effect and the variations of pixel efficiency are found most significant for soft X-rays. Common radiographic systems are usually equipped with a filter, such as a thin aluminium plate, which reduces the beam hardening effect [1]. The filter is placed between the source and the object. Then, low energy X-ray photons are removed from the beam before they reach the object. The main disadvantage of this technique is a suppression of soft X-rays, which results in a decrease of the image signal-to-noise ratio (SNR), especially for soft tissues. Furthermore, this method gives only a reduction of the beam hardening effect [2], A linearization procedure [3, 4] is commonly used for correction of the beam hardening effect. The measured nonlinear relationship between material thickness and the logarithmic attenuation is fitted by a polynomial approximation. Every value on the polynomial is corrected towards the linear trendline which is expected in the monochromatic case. For high density materials it is necessary to use high order polynomials which are numerically unstable. The linearization procedure is usually done for the entire X-ray detector, i.e. it does not include efficiency variations among detector pixels. Distortion of roentgenograms due to beam hardening and pixel efficiency variations can be effectively eliminated simultaneously by the "Direct thickness calibration method" proposed in [5]. 2.1. Direct thickness calibration At the calibration phase we measure the response of individual pixels of the Xray detector to specific levels of "beam hardening" using different absorber thicknesses (calibrators) under conditions of stable and constant X-ray tube parameters. The measured nonlinear relationship between calibrator thickness and counts is fitted by a semi exponential calibration function for each individual pixel. Because this procedure is done for each detector pixel, both individual pixel efficiency and beam hardening are corrected simultaneously.

775 At the experimental "imaging" phase the measured intensity is directly assigned to a specific value of the calibration function for each pixel. A map of "equivalent thickness" values substituting radiograph intensities is obtained by this method. The direct thickness calibration technique improves significantly the transmission roentgenogram quality regardless of the object material and thickness variations. Introducing an "equivalent thickness" for objects composed of materials different to those of the calibrator (from the point of view of their Xray absorption), roentgenograms of rather non-homogenous objects can be improved by this method as well. One has to stress that radiographs have to be measured under the same X ray tube conditions set in the calibration phase. 3. Experimental System A new experimental system devoted to digital high resolution X-ray transmission radiography was designed and is presented in this work. The basic concept of the experimental system is the stable position both, the X-ray tube and detector during measurements as well as the operational movement of the observed object which is kept fixed to the computer controlled motorized stage. The entire experimental system is placed in a fully shielded case allowing measurements without further safety arrangement. This concept is quite common for micro computer tomography setup. The principal requirements to be fulfilled in order to achieve high resolution X-ray imaging (i.e., in the micrometric scale) are a "point" source (or collimated) X ray beam and a high performance X-ray detector. We used an X-ray tungsten microfocus tube Hamamatsu with X-ray emission spot of 5 fim and divergent cone beam ("point" source), which enabled a magnification up to micrometer scale spatial resolution. The tube output Beryllium window does not influence the X-ray spectrum significantly. As X-ray detector we used the digital Medipix-2 [6] device with a 300 /nn thick Si sensor arranged into a matrix of 256x256 square pixels, 55 /jm by 55 /im each. The detector was attached to a water cooling system ensuring temperature stabilisation within 0.1 °C accuracy. All detector support and read out system are integrated into one compact USB 1.1 controlled interface [7]. Two disc holders of calibration filters are employed in the experimental setup for calibration phase in the "Direct thickness calibration method". Each disc carries ten calibrators, providing up to one hundred combinations of calibrator thickness values. Pure Aluminium calibrators with thickness steps of

776

50 um up to 5 mm and with steps of 500 um in the thickness range between 5 and 10 mm were used in the present setup. The complete experimental system is fully controlled by a notebook via USB interface with a possibility to synchronise properly all control and monitoring commands. 4. Experimental A number of radiographic measurements of soft tissue objects have been carried out using the described experimental setup. The quality of the beam hardening correction can be illustrated by the radiography of a termite body [8] in Fig. 1. The termite body was selected as an example of a tiny object to demonstrate both, the resolution and contrast in an imaging of structure made of soft tissues (a termite has no hard exoskeleton). For this image, the exposition time was 50 sec, tube voltage 40 kV, current 100 uA and a geometrical magnification factor of 25. The temperature of the Medipix detector was stabilized at 10 °C. A set of aluminium foils was used for the "Direct thickness calibration method". Even it is possible to resolve the structure of feelers which are 70 um thick.

Figure 1. Radiograph of a termite.

Radiographs of a Gammex RMI 156 mammographic phantom were realized as well [9]. Sixteen test objects of different sizes, shapes and densities representing malignancies or small breast structures were embedded in a wax block enclosed in an acrylic box . All test objects fibrils and tumor-like masses and three groups of micro-calcifications were detected using our experimental system above described. Regarding the small variations of phantom absorptions, only one 0.7 mm Aluminium filter (calibrator) has been used for the correction of individual

777 pixels in the Medipix detection efficiency variations (i.e. with constant beam hardening). A comparison between the radiographs acquired with the mamograph GE Senographe DMR+ and our digital system is given in Fig. 2. Note that the digital radiographs obtained by the single photon counting pixel device appear with markedly better contrast and sharpness.

Figure 2. Comparison between radiographs acquired by the described system (left) and by the mamograph GE Senographe DMR + (right); a) Test object 1 - a fibre 1,56 mm diameter; b) Test object 2 - a tumor-like mass 2,00 mm diameter

5. Conclusion The experimental system for high resolution digital X ray transmission radiography has been successfully realized and tested on a number of different soft tissue objects. Excellent quality roentgenograms are obtained using the "Direct thickness calibration method" which effectively compensates for the beam hardening effect even in the case of imaging of extremely low contrast and low absorption objects. Acknowledgments This work has been carried out in framework of the Medipix-2 collaboration based at CERN and has been supported in part by Grant No. 106/04/0567 of the Grant Agency of the Czech Republic and by the Ministry of Education, Youth and Sports of the Czech Republic under Research Projects 1P04LA211, MSM210000019/M/2005 and AV0Z20710524. References 1. R. J. Jennings, A method for comparing beam hardening filter materials for diagnostic radiology.Medical Physics, Vol. 15, (1988), 588 2. E. Van de Casteele , D. Van Dyck, J. Sijbers and E. Raman, Journal of XRay Science and Technology, Vol. 12, (2004), 43 3. R. A. Brooks, G. Di Chiro, Beam hardening in X-ray reconstructive tomography. Physics in Medicine and Biology, Vol. 21 (1076), 390

778

4. P. Hammersberg and M. Mangard, Journal of X-ray Science and Technology, Vol. 8 (1998), 75 5. J. Jakubek, D. Vavrik, S. Pospisil, J. Uher, Nuclear Instruments and Methods in Physics Research Section A, Vol. 546 (2005), 113 6. Medipix colaboration: http://www.cern.ch/MEDIPIX/ 7. Z. Vykydal, J. Jakubek, T. Holy, S. Pospisil, A portable pixel detector operating as an active nuclear emulsion and its application for X-ray and neutron tomography. To appear in this Proceedings 8. R. Tykva, R. Hanus, J. Jakubek, Z. Wimmer, J. Novak, V. Vlasakova, To be published In: XVIII World Congress IMEKO, 2006 9. J. Dammer, Diploma thesis, First Faculty of Medicine, Charles University in Prague, 2005

A PORTABLE PIXEL DETECTOR OPERATING AS AN ACTIVE NUCLEAR EMULSION AND ITS APPLICATION FOR X-RAY AND NEUTRON TOMOGRAPHY* Z. VYKYDAL f , J. JAKUBEK, T. HOLY, S. POSPISIL Institute of Experimental and Applied Physics, Czech Technical University in Prague Horska 3a/22, Praha 2, CZ-12800, Czech Republic E-mail: zdenek. vvkvdal(a)utef.cvut. cz This work is devoted to the development of a USB 1.1 (Universal Serial Bus) based read out system for the Medipix2 detector to achieve maximum portability of this position sensitive detecting device. All necessary detector support is integrated into one compact system (80 x 50 x 20 mm3) including the detector bias source (up to 100 V). The read out interface can control external I2C* based devices, so in case of tomography it is easy to synchronize detector shutter with stepper motor control. An additional significant advantage of the USB interface is the support of back side pulse processing. This feature enables to determine the energy additionally to the position of a heavy charged particle hitting the sensor. Due to the small pixel dimensions it is also possible to distinguish the type of single quanta of radiation from the track created in the pixel detector as in case of an active nuclear emulsion.

1. Introduction Nowadays fully digital radiation imagers such as Medipix [1] are available and make possible real time imaging with high sensitivity and broad dynamic range. The hybrid pixel detector device Medipix2 consists of a sensor chip, with 256 x 256 square pixels of 55 um size each, which is bump bonded to a read-out chip. The read-out chip [2] contains an amplifier, two pulse height discriminators and a 13-bit counter for each pixel cell. The energy window can be adjusted by the discriminators. The counter is incremented only in case that the energy of the interacting particle deposited in a pixel of the sensor chip falls within the energy window. The hybrid structure of the detector allows using different sensor materials for specific applications and makes possible real time X-ray [3] as well as neutron [4] imaging and tomography. This work was carried out within the CERN Medipix2 Collaboration. * On temporary leave of absence at NIKHEF. The National Institute for Nuclear Physics and High Energy Physics, Kruislaan 409,1098 SJ Amsterdam, The Netherlands * I2C stands for Inter Integrated Circuit bus.

779

780

A special interface is needed to perform data acquisition from the Medipix2 chip. Several such interfaces have been already developed in the Medipix collaboration [5-7]. All these interfaces require additional support devices for their operation (supplementary PC data acquisition cards, external power supplies, etc.). Our goal was to develop a really portable, folly independent interface integrating the entire necessary detector support into one compact device without any compromises in functionality. 2. Medipix! USB interface The USB [8] is the contemporarily most widespread computer interface. The architecture of the USB offers serial communication and a power line (5 V, up to 500 mA) and it supports "Plug & Play" and "Hot swap" technologies. Consequently, a new USB device can be connected to a running computer and it is recognized automatically. 2.1. Hardware The whole interface can be divided into several main parts [9]. The block composed of FT245BM chip [10] takes care of the USB protocol management and provides bi-directional conversion between USB line and 8-bit wide CMOS bus. This CMOS bus is connected to the ADuC841 MicroConverter® [11] which controls the functionality of the entire interface. The microcontroller has integrated high speed 420 ksps ADC and DAC (12-bit precision both) which are used for Medipix2 purposes and for the back-side pulse processing feature. Several power supplies are needed for interface and Medipix2 device powering. The block of power supplies consists of 2.2 V power supply for Medipix2 chip, 3.3 V power supply for logic converters and MAX1932 chip based variable voltage source for Medipix2 detector bias (from 5 to 100 V). For the 2.2 V power supply the DC-DC step down regulator is used before the low-drop linear regulator to reduce the current drawn from the USB host controller. All the power supplies are derived from the USB 5 V power line. Power consumption of the whole Medipix2/USB device is about 250 mA during operation (also various power-saving modes are supported). All above mentioned components (mostly SMDb) are mounted on a four layer PCBC with dimensions of 75 x 46 mm2. The right side of the board is occupied by the I/O connectors. The left side of the PCB is adapted to be

b SMD stands for Surface Mount Device ' PCB stands for Printed Circuit Board

781

plugged directly to the VHDCId female connector used on all current Medipix2 chipboards (see Figure la).

Figure 1: Photographs of the developed interface: a) USB interface board connected to one of the current Medipix2 chip carrier boards, b) USB interface board in duralumin box (80 x 50 x 20 mm3).

2.2.

Software

The entire interface is driven by the ADuC841 microcontroller. Software for this microcontroller has a simple organizational structure. In the main endless loop the program waits for a command received via USB. After a command is received it is compared with the internal table of known commands. If the command is recognized the relevant operation is executed. After the command procedure is finished, or the command is not recognized, the program jumps back into the main endless loop. This organization allows to add a new command into the previous program structure. User friendly software, called "Pixelman", for controlling measurement from the host PC has been also developed in our group as described in [12]. 2.3. Back-side pulse processing This new feature enables to measure the current pulse generated when the charge produced by a heavy charged particle in the depleted region of the Medipix2 sensor is collected. Using this mechanism one can get additional information about the energy of the interacting particle [13]. Achieved energy resolution of the Medipix2/USB device for 5.5 MeV alpha particles was 44 keV in terms of FWHM. Application of this feature is limited by the minimum energy of the interacting particle (giving sufficiently high signal to noise ratio) and by the maximum count rate of about 20 kHz (given by the shaping time of preamplifier).

d

VHDCI stands for Very High Density Cable Interconnect

782

3. Measurements In this section some measurements using the Medipix2 device and the Pixelman software package [12] are presented to demonstrate its capabilities. 3.1. Analogy with nuclear emulsion The device can be used as a universal radiation monitor visualizing naturally occurring ionizing radiation, as seen in Figure 2. Photon traces, tracks and clusters produced by electrons and alpha particles are clearly recognizable similarly as in the case of a nuclear emulsion or cloud and bubble chambers.

Figure 2: Sample measurements of the natural background radiation with Medipix2 device assembled by 700 urn Si sensor (lOmin exposition time), a),b) Many electron tracks and some alpha particle clusters, c) Rare cosmic muon track sideways transversing the sensor.

3.2. Neutron micro-tomography A 300 urn Si Medipix2 sensor was adapted to a position sensitive neutron detector by coating it with LiF and used for the tomographic measurements. Results of the measurement of a LEMO connector as an example object are shown in Figure 3. 100 projections (150 seconds each) have been taken and the Filtered Back Projection (FBP) algorithm [14] was used for 3D reconstruction.

Figure 3: Neutron tomography of the LEMO connector using Medipix2 with 300 urn Si sensor coated by LiF: a) Photograph b) Neutron transmission image c) 3D reconstruction.

783

3.3. X-ray micro-tomography Details of a termite head were measured to demonstrate the applicability of the Medipix2 chip for a soft tissue X-ray tomography. 360 projections were measured with 20 s exposure time per projection. Results of the object reconstruction using FBP algorithm are depicted in Figure 4. As one can see, the tiny mandibles and antennae are clearly visible.

**

^

lOOym

' ^ P I i Jr a)

b)

c)

-

-

Figure 4: X-ray tomography of the termite head using Medipix2 with the 300 um Si sensor: a) Photograph b) X-ray transmission image c) 3D reconstruction.

4. Conclusions and future plans A really portable interface for Medipix2 device based on USB 1.1 has been manufactured and is being successfully used. Further improvements can be accomplished by the plug-in module assembly. The readout speed is 6 Mbit/s (6 fps) which is suitable for many applications (dental radiography, small animal imaging, X-ray and neutron radiography and tomography, etc.). The Medipix2 pixel detector in connection with presented interface works as a really portable "camera-like" universal radiation monitor. With an on-line visualization of the tracks of individually registered particles, one can even recognize the type of the interacting particle like in the case of nuclear emulsion. The device may have interesting educational value. As the next step we plan to speed up the interface using USB 2.0 and improve the reliability by accommodating the Medipix2 chip and USB read-out interface on the same board. Acknowledgments This work has been carried out in frame of the CERN Medipix Collaboration with the support of the Ministry of Education, Youth and Sports of the Czech Republic under Research Projects MSM210000019 / M / 2005 and 1P04LA211.

784

1. 2.

3. 4.

5.

6. 7.

8. 9. 10. 11. 12.

13.

14.

Medipix collaboration homepage at http://www.cern.ch/medipix X. Llopart, M. Campbell, R. Dinapoli, D. San Segundo Bello, E. Pernigotti: "Medipix2, a 64k pixel readout chip with 55 um square elements working in single photon counting mode", Proceedings of the IEEE Nuclear Science Symposium and Medical Imaging Conference, San Diego, November 4-10, 2001. J. Jakubek, T. Holy, S. Pospisil, D. Vavrik: 'Tomography for XRDD", Nucl. Instr. and Meth. A 532 (2004) 307-313. J. Uher, T. Holy, J. Jakubek, E. Lehmann, S. Pospisil, J. Vacik: "Performance of a pixel detector suited for slow neutrons", Nucl. Instr. and Meth. A 542 (2005) 283-287. D. San Segundo Bello, M. Van Beuzekom, P. Jansweijer, H. Verkooijen, J. Visschers: "An interface board for the control and data acquisition of the Medipix2 chip", Nucl. Instr. and Meth. A 509 (2003) 164-170. V. Fanti, R. Marzeddu, P. Randaccio: "Medipix2 parallel readout system", Nucl. Instr. and Meth. A 509 (2003) 171-175. V. Fanti, R. Marzeddu, G. Piredda, P. Randaccio: "Design and test of data acquisition systems for the Medipix2 chip based on PC standard interfaces", Nucl. Instr. and Meth. A 546 (2005) 286-290. USB 1.1 specifications at http://www.usb.org. Z. Vykydal: "Microprocessor controlled USB interface for Medipix2 detector", Diploma thesis, CTU in Prague, 2005. FT245BM - Single chip bi-directional USB to Parallel FIFO datasheet at h ttp: //www.ftdichip. com. ADuC841 - Precision 12-bit Analog Microcontroller datasheet at http://www.analog.com. T. Holy, J. Jakubek, S. Pospisil, J. Uher, D. Vavrik, Z. Vykydal: "Acquisition and Data Processing Software Package for Medipix2 Device", Book of abstracts of the 7th IWORID conference, 4-7th July 2005, Grenoble, France. To be published in Nucl. Instr. and Meth. A. Z. Vykydal, J. Jakubek, S. Pospisil: "USB interface for Medipix2 pixel device enabling energy and position sensitive detection of heavy charged particles", Book of abstracts of the 7th IWORID conference, 4-7th July 2005, Grenoble, France. To be published in Nucl. Instr. and Meth. A. A.C. Kak, M. Slaney: "Principles of Computerized Tomographic Imaging", IEEE Press, New York, 1988.

Radiation Damage Organizer: S. Baccaro

S. Assouak A. Cecilia S. Di Luise I. Di Sarcina M.E. Dinardo P. Fuochi K.K. Gan P. Giubilato I. Lazanu P. Mangoni

J.-P. Merlo A. Messineo E. Mihokova F. Nessi-Tedaldi R. Rando E. Scapparone A. Sopczak C. Zhu

Statistical Study of Radiation Hardness of CMS Silicon Sensors SIC PbWO"4 Crystals for the Electromagnetic Calorimeter of CMS Experiment MDT Chamber Ageing Test at ENEA Casaccia Neutron and Gamma Facilities Behavior of Thin Film Materials Under 7 Irradiation for Astronomical Optics Full Characterization of Non-Uniformly Irradiated Silicon Micro-Strip Sensors Beam Energy Monitor for 4-10 MeV Electron Accelerators Optical Link of the ATLAS Pixel Detector Ion Electron Emission Microscopy for SEE Studies An Analysis of the Expected Degradation of Silicon Detectors in the Future Ultra High Energy Facilities Investigation of VLSI Bipolar Transistors Irradiated with Electrons, Ions and Neutrons for Space Application Radiation-Hardness Studies of High OH~ Content Quartz Fibers Irradiated with 24 GeV Protons Development of Radiation Hard Silicon Detectors: the SMART Project Scintillator and Phosphor Materials: Latest Developments and Applications Crystals for High-Energy Physics Calorimeters in Extreme Environments Radiation Testing of GLAST LAT Tracker ASICS Magnetic Field and Radiation Tests of a Programmable Delay Line LCFI Charge Transfer Inefficiency Studies for CCD Vertex Detectors Photoluminescence and 7-Ray Irradiation of SrO-B 2 0 3 -P20 5 :Eu 2 + and SrMo0 4 :Eu 3 + Phosphors 785

This page is intentionally left blank

STATISTICAL STUDY OF RADIATION HARDNESS OF CMS SILICON SENSORS SAMIA ASSOUAK* Catholic University ofLouvain,

2 chemin du Cyclotron B-1348 Louvain la Neuve

On behalf of CMS Tracker Sensor Working Group Silicon microstrip and pixel detectors are presently the most precise electronic tracking detectors for charged particles in high energy physics. This technology has been chosen for the central trackers of CMS and ATLAS experiments at the LHC (CERN). These detectors are intended to work in a very harsh environment where the luminosity of pp collisions reaches 1034 cm'Y '.Therefore, the effects of radiation damage must be studied in detail. For that purpose, we have irradiated CMS silicon microstrip sensors with a very high flux of neutrons to achieve a fluence of 1.6 1014lMeV-equivalent neutron.cm"2. More than 500 samples were analysed. This work presents the evolution of important sensor characteristics (i.e. bulk depletion voltage, total leakage current, interstrip capacitance, bias resistance) after neutron irradiation.

1. Introduction The CMS central tracker consists of about 24200 microstrip sensors covering an area of 208 m2. At the LHC protons will collide at very high center of mass energy of 14 TeV. The expected luminosity is 1034 cm"2 s"1. These collisions will create enormous radiation backgrounds, which will damage the detectors material. The considered equivalent fluence is 1.6 1014 IMeV n/cm2 for ten years of LHC operation. The sensors selected must then obey very strict criteria to survive the harsh LHC environment. To ensure a good functionality of the tracker, the CMS tracker group developed an elaborate design and a detailed quality control procedure.

E-mail address: samia.assouaktaifvnu.ucl.ac.be (S.Assouak) Tel: +32 (0)10 47 32 34

787

788

2. Sensor design Sensors are cut from 6" wafers. From each wafer one extracts one sensor. Each sensor has 512 or 768 implanted p+ strips on n substrate. The strips are connected via an array of polysilicon resistors to a common bias ring. They are AC coupled. Two thin layers of Si02 and Si3N4 form the coupling capacity. Two thicknesses were selected: 320 fim thick sensors with a resistivity of 1.5-3.0 kQ.cm are used for the inner part of the tracker and 500 /xm thick for the outer part with a resistivity of 3.5-7.5 kQ.cm. This will allow one to delay the bulk type inversion due to irradiation. The lattice orientation of the silicon crystal is <100> to reduce the surface damage. From each wafer four half-moons remain. One halfmoon called standard test-structure is dedicated to the quality assurance tests. It contains a minisensor with the same elements and characteristics as the full-size sensor and others elements (diodes, MOS structure...) which allow the check of the production process [1].

3. Irradiation quality control procedure To check the radiation hardness of the sensors during their production, a qualification tests for radiation hardness are made before mounting them on the tracker. In Louvain-la-Neuve, 4% of test structures are irradiated at the LHC fluence and tested after their irradiation to ensure suitable performances according to the CMS collaboration criteria. The irradiation and the subsequent tests are done at a temperature of-10°C. After irradiation, detectors are annealed at a temperature of 60°C for 80 minutes to accelerate the ageing. The tests performed before and after irradiation consist of bulk measurements (total leakage current, depletion voltage) and strip parameters (bias resistors, coupling capacity, interstrip resistors and capacity, check of pinholes).

4. Irradiation and test setups Neutrons are produced from the reaction d (Be, n) B. The average energy is 20.4 MeV. The corresponding hardness factor is K = 1.95. The setup uses a cold box (-15°C) and sensors are biased to reproduce as close as possible the LHC conditions. More details can be found in references [2] and [3]. The experimental setup performs electrical measurements. It measures currents (for total leakage current and strip current), capacitances (total, coupling and interstrip capacitances) and resistors (bias and interstrip resistors). All tests are fully automatized using a probe station and a switching matrix. The

789 temperature and the humidity are controlled (T= -10CC; RH ~ 4%). The tests are monitored with a LabView software. The instruments are read via a GPIB bus [3].

5. Results 5.1. Bulk measurements One of the most important effects due to the radiation damage on the silicon semiconductor is the change in the doping concentration leading to the type inversion of the substrate within several years [4]. This results in an increase of the depletion voltage and the leakage current. In figure 1 the leakage current after irradiation is shown. It increases linearly with the fluence. The estimated values for the normalized current agree well with those in the literature [5]. We have measured the depletion voltage after irradiation and compared them to the predicted values the Hamburg model [4]. The experimental data for the 320 ^m thick sensors are lower than the predicted values. The sensors with a thickness of 500 fim showed a good agreement between the experiment and the predictions if we take into account the high error (~ 40V) on the measured values after irradiation (fig. 2).

- 6,0x10 •

500 iim S

6.0x10*-

HPK320//m ••

4,0x10'"

r *

.:.-£
2,0x10*-

0,0. 1.0

— 1,2

1,4

1,6

1,6

2.0

2,2

2,4

2,6

2,6 3,0

Fluence (10'4cm'2)

0,4

0.6

0.6

1.0

1.2

1,4

1.6

1,6

2,0 2.2

Fluence(10"cm'2)

Figure 1: Total leakage current after irradiation versus fluence.

5.2. Strip measurements We will describe only polysilicon resistors and interstrip capacitance. Surface ionization affects seriously the interstrip impedance causing a cross talk which increases noise in the signal. It is the main source of noise at the CMS working temperature (-10°C). The interstrip capacitance C„,t for CMS sensors must be less than 1.2 pF/cm strip length. In the figures we represent the histogram

790

of the values of Cj,,, after irradiation for all test-structures tested (fig.3). We observe that for the 500 /im thick minisensors, Cmt increases of about 60% from its nominal value (measured before irradiation) at full depletion representing 0.3 pF/(l 10l4cm"2) (fig. 4). Investigations showed that the sensor must be overdepleted to reach the initial value. The 320 /am detectors show similar values to the nominal ones with an increase around 15 % which is within CMS specifications. For the bias resistors, the values are 1.5 + 0.5 MQ (fig. 5). We do not notice any significant changes after irradiation. Only a small fraction showed a little increase of about 0.2-0.3 M£2 (figure 6). S . 550f 500.

320 nm

300250200-

1.61.8 2,0 2.2 2*2,6 2,B3,0 3.2 0,0 0,2 0.4 0,6 0.6 1,0 1,2

1,4 1,6

1,8 2.0 2^2

Fluence (10"cm~2)

Figure 2: Distribution of the full depletion voltage after irradiation compared to the predicted values. 500 pm After Irradiation

0.0

0.2

0.4

jiim§i

0.6

0,6

1,0

1,2

1.4

320 fim After Irradiation

1,6

1,8

2,0

•-. J i l l C

JPF)

Figure 3: Histogram of interstrip capacitance values 500 ^m 320 /jm

JH I_JL 0.0

0.5

1,0

1,5

2,0

2,5

3.C

ACM/fluence (pF/1014cm'z)

III •9 .•Mi -0,2

0,0

0,2

0,4

0.6

0,8

ACln/fluence (pF/1014cm'2)

Figure 4: Histogram of interstrip capacitance values normalized to fluence

791

-requen

700-j

5 0 0 ym After Irradiation

§ 1

^

320 //m After Irradiation

1 11

400-

•111

30020010000,0

mini ni| i 0,5

1111 .•Ills

, i illlRl*ffiffi"T"MIMii}l™ • 1,0

1,5

2,0

i

2,5

3,0

Figure 5: Bias resistors values distribution "

120-

3 2 0 jum

500 jum

,2

0.0

0,2

0,4

0,8

0,8

1,0

-0.8

J V\

-0,6 -0,4 -0.2

0.0

0,2

0,4

0,6

0.8

Figure 6: Distribution of the bias resistors variation

6. Conclusion The CMS microstrip detectors for the tracker were tested under neutrons flux equivalent to more than ten years of LHC operation. More than 500 teststructures were irradiated and tested. The depletion voltage stays within CMS specifications (<300V). The sensors can be biased at 500 V. The leakage current scales with the fluence and the deduced alpha factor is close to the value found in literature [4].

References 1. Specifications for Quality Control & Assurance of the CMS Silicon Detectors, Draft review 2.0 19/05/00 by the CMS Tracker Sensor Working Group. 2. K. Bernier PhD Thesis UCL Louvain-la-Neuve 2001, http://cds.fynu.ucl.ac.be/ 3. S. Assouak et al. Nucl. and Instr. Meths. In Physical research A 514 (2003) 156-162 4. M. Moll PhD Thesis DESY Hamburg 1999. 5. M. Moll et al. Nucl. and Instr. Meths. In Physical research A 426 (1999) 87-93

SIC P b W 0 4 CRYSTALS FOR THE ELECTROMAGNETIC CALORIMETER OF CMS EXPERIMENT ANGELICA CECILIA ENEA-Casaccia RC, Via Anguillarese 301 S. Maria di Galeria (Rome), 00060, Italy On behalf of CMS ECAL Collaboration PbW04 crystals produced in China by Shanghai Institute of Ceramics are presently considered by CMS Collaboration for their use at the construction of the Electromagnetic Calorimeter. A statistically representative group of crystals from batches recently produced in China is studied in this work for their radiation hardness characteristic.

1. Introduction After an extensive program undertaken from 1994 to 1998 and a preproduction run carried out from September 1998 to December 2000, PbW0 4 (PWO) crystals are now in mass production. More than 42000 crystals have been produced by Chzocralski method at Bogoroditsk Techno-Chemical Plant (BCTP) in Russia [1]. Recently, the optimization of raw materials purification procedures and PWO crystal growth conditions at Shanghai Institute of Ceramics (SIC) led to the production of high quality crystals to be used in the PrimEx experiment at Jefferson Laboratory [2]. Moreover, a new production technology based on Yttrium doping was developed, which turned out to be effective in reducing slow scintillation component and leading to PWO crystals with adequate radiation hardness for CMS experiment [3]. In this work we analyzed a group of 114 SIC made crystals in terms of transmission and light yield parameters. In particular, we investigated the degradation of the optical parameters under y irradiation. 2. Samples and experimental set-up Crystals studied in this work include 74 endcap (EE) (22x2.8x3cm3) and 40 barrel (EB) calorimeter crystals (23x2.2x2.4cm3) produced at SIC. Before irradiation, crystals were characterized by transmission (longitudinal and transversal) and Light Yield (LY) characteristics. The degradation of these parameters was followed at different dose rates during several irradiation cycles. 792

793

The degradation of the optical transmission was estimated through the radiation induced absorption coefficient (u) defined as: 1 ( T \ // = —ln A . d

\Tirr

(1)

y

where d is the sample length in the optical path direction and T0, T-m are the initial and the transmission after irradiation respectively. Crystals were irradiated at the Calliope 60Co plant (ENEA Casaccia, Rome) [4] in dosimetric positions corresponding to the following dose rates: 11.8 Gy/h, 34 Gy/h and 350 Gy/h. A typical irradiation cycle consisted in y radiation exposure at increasing doses until damage saturation. Irradiation was followed by room temperature (RT) recovery and finally, crystals were submitted to thermal bleaching process consisting in: (a) ramp-up from room temperature to 200 °C in not less than 3 hours; plateau of 6 hours at 200 °C and (b) finally power stop and natural cooling. Due to irradiation plant security regulation, transmission and LY measurements were performed at typically 15 min from the stop of irradiation. Transmission and LY measurements were performed with the automatic measuring system ACCOR [5]. Transmission runs consisted in measuring the optical transmission along the crystal and transversally in four points: 2, 8, 14 and 20 cm starting from the front face (FF). Dedicated transmission measurements were also performed with the home made double-ray spectrophotometer LUMEN [6]. Each LY run consisted in measuring the 60Co y spectrum in 11 points along the PWO crystals, starting at 2 cm from the FF, with 1 cm step. 3. Results and discussion The electromagnetic calorimeter of CMS needs crystals which under y irradiation exposure have low optical transmission deterioration and therefore low light yield loss to preserve the high resolution of the detector. Besides this, the nature and the concentration of defects responsible for the light yield deterioration under ionizing radiation have to be such to allow an easy qualification procedure for these crystals. In the case of BTCP PWO crystals currently used for ECAL barrel, the qualification procedure is based on the correlation between measurable optical parameters and the [i(X) at full saturation of defects. In particular, for BTCP crystals a defect saturation dose was experimentally defined, which corresponds to irradiate crystals at 350 Gy/h for lh 20 min (466.7 Gy) and is referred as Standard Full Saturation (SFS) Dose. The corresponding u limit value at 420 nm was set to 1.5 m"1 [7]. All analyzed

794

SIC crystals were irradiated at the SFS dose and representative |j. spectra are reported in Figure la for a set of 15 samples: as it can be seen the SIC n shape is similar to the radiation distribution of color centers induced by radiation in BCTP crystals, which has been shown to arise from two main color centers, around 420 nm and 520 nm [8, 9]. The only difference lies in the relative intensity of the aforementioned two bands and in the damage spread which is slightly higher for the SIC samples. In spite of this, the SIC mean value of u(420 nm) is equal to 1.24±0.5 m'1 and it is acceptable from the viewpoint of BTCP qualification procedure. 3,0 - S I C 1 1 4 - " - S I C 1 2 0 ••••* SIC122-T SIC126 -SIC1S7—<-SIC138- • S1C162 # S I C 1 7 3 - S I C 1 7 * -* SIC2S5-»~SIC270 a SIC271 - S I C 2 7 7 - a - SIC289-9— SIC 104

2,5 ^

24-

i-i

2,0 J2 16c

1

D

o O

8

I i •-!'-'-

si ,iui

400

500 600 Wavelength (nm)

700

Bjqtfwfoy^pgsa 0 0,0 0,5 1,0 1,5 2,0 2,5 3,0 n after 1 h & 20min (ip 350 Gy/h (m"1)

Figure 1. u,(X) spectra of representative SIC crystals (a) and u.(420nm) distribution of all analyzed samples after 466.7 Gy @ 350 Gy/h (b).

Further irradiation tests performed at the SFS rate for a time longer than lh 20 min evidenced that for some SIC crystals such irradiation dose is not sufficient to saturate all defects. Nevertheless, as we will explain below, we may reasonably suppose to be on the safe side from the crystal application point of view in the LHC radiation environment. In fact, the irradiation rate of 11.8 Gy/h is the maximum dose rate in LHC-like conditions. The crystal (i saturation curves at 11.8 Gy/h (Figure 2a) can be described by a double exponential growth curve from whose fit we calculated the damage saturation p. (usat) [10]. For each crystal we reported the u value related to the SFS (uSFS) vs. the (isat. Results were particularly interesting: in fact, the USFS is linearly correlated with the (xsat and the usa, is slightly smaller than that with SFS (Figure 2b). Considering that and taking into account the aforementioned comments about the dose rate in LHClike conditions we may hypothesize to be on the safe side from the crystal damage point of view in LHC radiation environment.

795

200

400 600 Dose (Gy)

800

0,0

0,5 1,0 1,5 2,0 u . (m"1) @ 420nm

2,5

Figure 2. n vs. irradiation dose @ 11.8 Gy/h for representative crystals (a) and HSFS @ 466.7 Gy vs. Usat saturation damage @ 11.8 Gy/h.

As far as LY is concerned, Figure 3 reports the LY measured on 40 preproduction crystals before and after the SFS (466.7 Gy @ 350 Gy/h). Its mean value is equal to 10.93±0.60 pe/MeV and to 7.33+1.43 pe/MeV before and after irradiation, indicating a light loss of about 33 %. 30 • before irr 2 after 466.7Gy

8 10 LY@8X0 (pe/MeV) Figure 3. LY before and after SFS dose of 466.7 Gy @ 350 Gy/h measured on 40 pre-production crystals.

Before the radiation induced characterization at the Calliope 60Co plant, SIC crystals undergo irradiation tests at a 60Co industrial irradiator available at SIC. Hence, it was extremely important to calibrate the damage measured in China on the damage measured in our institute. To do that, we have selected 9 crystals previously irradiated in China @ 30 Gy/h for 24 hours and possessing quite different radiation hardness properties. Those crystals were irradiated at Calliope facility in the dosimetric point corresponding to 34 Gy/h for 21 h llmin (720

796

Gy). The damage evaluated in China turned to be linear dependent on the damage evaluated in our laboratories with a slope of 1.14±0.02 (Figure 4).

y=1.14*x

.*--''

1.14±0.02

.%'

,-**""'

1 *'' 0,0

0,4

0,8

1,2

1,6

2,0

n(420 nm) Rome (m'1)

Figure 4. Correlation between u.(420 nm) measured at Shanghai irradiator and u.(420 nm) measured in our plant after the irradiation dose of 720 Gy.

4. Conclusions The radiation damage tests performed on 114 SIC crystals gave interesting and promising results. In spite of the missed saturation of defects observed at the SFS condition, the mean value of the measured damage (1.24±0.5 m"1) is acceptable. A correlation was found between the damage after SFS dose and the saturation damage corresponding to 11.8 Gy/h. The comparison of the u measured in China at Shanghai Institute irradiator and the u measured by us after the same dose of 720 Gy at -30 Gy/h turned out to be reliable and the established linear relationship allowed to calibrate SIC results on our tests. References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.

E. Auffray, NIMA 486, 111 (2002). R. Mao, IEEE Trans. Nucl. Science VOL: 51, NO. 4 (2004). X. Qu, et al., NIMA 480,470 (2002). S. Baccaro, ENEA Technical Report 28/FIS/2005. S. Baccaro, NIMA 459,278(2001). M. Montecchi, Rev. Sci. Instrum. 75, NO. 11 (2004). E. Auffray, CMS Technical Note 038 (1998). S. Baccaro, CMS Technical Note 067 (1998). S. Baccaro, CMS Technical Note 002 (2000). A. Annenkov, CMS Technical Note 008 (1997).

M D T C H A M B E R AGEING T E S T AT E N E A CASACCIA N E U T R O N A N D G A M M A FACILITIES

G. AVOLIO 1 , P. BRANCHINI 2 , S. DI LUISE 2 , E. GRAZIANI 2 , L. LA ROTONDA 1 , C. MAZZOTTA 1 , E. MEONI 1 , A. PASSERI 2 , F. PETRUCCI 2 , A. POLICICCHIO 1 , D. SALVATORE 1 , M. SCHIOPPA 1 , A. TONAZZO 2 1 2

Dipartimento di Fisica, Calabria University and INFN, Cosenza, Italy Dipartimento di Fisica, Roma Tre University and INFN,Roma, Italy

The Monitored Drift Tubes (MDT) are used for precision tracking in the muon spectrometer of the ATLAS experiment. They have to cope with a counting rate reaching 500 Hz/cm 2 mostly due to photon and neutron background in the experimental hall which corresponds to an accumulated charge of 0.6 C/cm per wire in 10 years of LHC operation at the nominal luminosity of 10 34 c m - 2 s _ 1 . The neutron flux is expected to be about 5 kHz/cm2, corresponding to a counting rate of 2.5 Hz/cm 2 . MDTs ageing studies have been performed at the ENEA-Casaccia Research Center (Italy) irradiating a test chamber with neutrons up to an integrated flux of 2.4-1012 neutrons/cm 2 and with photons up to an integrated charge of 4.8 C/cm. 1. The Test Chamber A set of 24 identical drift tubes, 47 cm long, were built and tested following the standard ATLAS wiring and QC procedures. The 4 layers were glued together to form a 4x6 drift tubes bundle a . The bundle was equipped with the ATLAS on chamber gas distribution system components. The Ar : COi (93 : 7) gas mixture was supplied through a bottle of certified premixed gas. The pressure and the flux of the gas mixture were adjusted through a Bronkhorst system. Gas temperature, flow and absolute pressure were monitored during the whole test period. The absolute values of T and P are known within 0,2 A* and 3 mbar respectively. a

The uncertainty on the wire position is less then 20 (im in the plane orthogonal to the wire

797

798

The tubes were read out by a front-end chip (ASD) which contains the TDC for the drift time measurement and a Wilkinson ADC for pulse charge measurement 1 . Each mezzanine contains three MDT-ASDs for a total of 24 channels per mezzanine-hedgehog combination. A Wilkinson ADC provides the measurement of the leading edge charge (the ADC gate can be set into the time interval 10 — 45 ns at the leading edge of the signal). 2. The neutron source and the photon source The employed neutron source is the TAPIRO reactor b at the ENEACasaccia Research Center 2 . The neutron energy spectrum at the position of the experimental setup ranges between 0.025 eV and 104 eV and shows a broad peak around 1 keV. The neutron flux scales linearly with the reactor thermal power. A calibration was performed during an irradiation session to derive the actual flux at the chamber surface. The values measured in three different points varies from 2.57 • 107 neutrons/cm 2 s to 1.06 • 107 neutrons/cm 2 s at 200PF with an uncertainty of about 3%. At the end of the irradiation period a maximum flux of 2.4 • 1012 neutrons/cm 2 was integrated. The gamma induced ageing test has been performed at the ENEA Casaccia Calliope Facility3 with photons from a ^Co source (about 10 1 5 S^). The two gamma from the ™Co deacy have energies of 1.173 MeV and 1.332 MeV. Two test chambers, placed at about 3 m far from the center of the source, have been exposed to an irradiation of about 1.1 • 109 photons/cm 2 s. The total average integrated charge was measured to be equal to 4.8 C/crn per wire. 3. Data taking Two different neutron irradiation programmes were performed. During the first test the reactor operated with a power varying from 20W to 200W over a two weeks period alterning irradiation with data taking runs (source switched off) using a cosmic ray trigger in order to monitor the chamber performances (a preliminary run was taken as reference). The current drawn was monitored over the whole period so as to measure the collected b

T h e reactor consists of a cylindrical core of 6.29 cm radius and 10.87 cm height and uses a metal alloy (U 98,5%-Mo 1.5%) with enriched uranium as fuel. T h e critical mass is 21.46 kg.The facility used for the test is the thermal column cave (110x110x94 cm**) where the experimental setup was located.

799

charge per tube. The chamber always worked at 3080 V with the standard gas mixture Ar : C 0 2 (93 : 7) at 3 bar. A system of two scintillator counters were used as a trigger for cosmics. The photomultipliers were protect by paraffin bricks and switched off before starting the irradiation in order to prevent photocathode and dynodes damage For the second irradiation programme one of the chambers already tested at the Calliope facility (see below) was used. Different cosmic data runs were taken while irradiating the chamber at the power of 50mWc and 100 W. A refence run was taken also. The trigger system is the same adopted for the gamma test, with Borum powder and Cadmium foils in addiction in order to increase P.M. and scintillators neutron shielding. For the ageing test with photons at the Calliope facility, two test chambers were irradiated during a three week long period. Cosmic data runs have been taken in between (source off). Three scintillator counters were used as a trigger for cosmics. The counters were completely wrapped with a 2 cm thick layer of lead. 4. Data analysis The ADC spectrum for cosmic rays shows a first narrow peak (pedestal) and a large one on the right. The position of the peak is related to the gas gain of the chamber d . The pedestal has been fitted with a gaussian and the large peak with an ad hoc function ratio of two polynomials. To estimate changes in gas gain (in particular reduction caused by ageing) the following variable has been built: Delta = (ADCpeak — ADCref)/ADCref where ADCpeak is the peak position and ADCref is the peak position relative to the reference run. Figure 1 shows the distribution of Delta as a function of the integrated charge for a central tube. The pulse height loss is not evident in any of the 24 tubes of the chamber on the 5 cosmic runs taken after different amount of collected charge per tube. The average of Delta over the whole neutron irradiation period has been computed for all tubes shown in fig. 1. The single tube efficiency was studied by reconstructing tracks in the chamber with all tubes but the one under observation. A hit may be absent (hardware inefficiency) or present with a residual from the fitted track c

Value corresponding to the nominal neutron background flux in ATLAS. C o n s i d e r i n g the gas gain dependence on the gas temperature (3.3% per degree), composition and pressure, it has assumed an overall uncertainty on gas gain variation of about 5%.

800

ADC peak ptnWon averaged over (ho whole irradiation period

1-

§r»s

\

12

I

B

|

G

f

I i

i j

i

' . . i;

L .4

_

-0.3

-0.2

>... 0

-0.1

0.1

0.2 0.3 0 overage A0C p.ak position

4

Figure 1. Distribution of Delta variable as a function of the integrated charge (left). Average Delta value over the whole neutron test (right).

larger than five times its resolution (5 a inefficiency). Fig. 2 shows both hardware and 5 a efficiency as a function of the radius for the reference and the last run of the neutron test run period. Fig. 3 (left side) shows the global 5 a efficiencies for different integrated charge for the 8 central tubes. No significative efficiency loss is evident. Fig.3 (right side) shows for the same set of tubes global 5 a efficiencies for three different running conditions: nuclear reactor off, 50mW and lOOmW neutron flux. A drop in the efficiency seems to be present in the last case. afffcl.iwlw VS rodtu* for MM 83302 •$

B-

'-

466, i $ 0

;;*<

'it

••••

v

ef fcfonelM

u

^ j*J; «»«

*•$•

•••'

/S r o * w lor tub* 22302

i

1

T

\

tube22 302

tube22 302 8

B

10

12

rodius(rnm)

0

2

+

6

10

12 „

,14

v

rodiua(mm)

Figure 2. Hardware and 5
801 200&1V05 12.1;

:

.

.

0. C/ O T

.

1.3 C/em

o

3.2 C/cm

,

4.8 C/cm

- ! : 5

f

i

p

e

•

-

Figure 3. 5
References 1. M. Posch, E. Hazen and J. Oliver. MDT-ASD: CMOS front-end for ATLAS MDT chamberst, ATL-MUON-2002-003; 7.6.2001 . 2. S. Baccaro, A. Cecilia, Facilities for Radiation Hardness Qualifications at ENEA-Casaccia (Rome-Italy), Proceedings "2nd SIRAD Workshop", Legnaro (Padova), 1-2 Aprile, 2004 Notes: ISBN 88-7337-007-1. 3. Baccaro S. et al.,Gamma and neutron irradiation facility at ENEA-Casaccia Center, Report CERN-CMS/TN, 95-192 (RADH) 1995

BEHAVIOR OF THIN FILM MATERIALS UNDER y IRRADIATION FOR ASTRONOMICAL OPTICS ILARIA DI SARCINA, ANGELA PIEGARI Advanced Technological Physics/OTT,

ENEA-Casaccia

RC (Rome, Italy)

ANGELICA CECILIA Advanced Technological Physics/ION, ENEA-Casaccia

RC (Rome, Italy)

A set of optical coating materials for astronomical and space applications were submitted to Y irradiation at the °Co radioisotope source (ENEA-Casaccia Research Center) in order to simulate the hostile radiation environment in which optical instrumentation is working. Firstly, samples were submitted to 50 Gy irradiation dose, typical of Low Earth Orbit missions. Successively, coatings were irradiated at doses between 1000-8400 Gy characterizing interplanetary mission expositions.

1. Introduction Optical components operating in space environment are exposed to fluxes of energetic particles which may deteriorate the image quality [1]. The space radiation environment is characterized by a large variety of primary particles covering an extended energy range, the main sources of such particles are: 1.

Galactic cosmic rays (particles produced outside the solar system including protons, heavy ions, electrons and positrons); 2. Particles entrapped in the Van Allen belts (electrons and protons); 3. Charged particles associated with the solar activity (electrons, protons and heavy ions) [2]. Beside the aforementioned particles, it is necessary to take into account the secondary particles (knockout protons, neutrons, a-particles, recoil nuclei and yrays) produced during the energy loss of primary particles as they pass through a spacecraft. In LEO, also albedo neutrons are sometimes mentioned, even if their component is small and of low energy [3, 4, 5]. Changes induced by radiation on the optical properties of the materials used in space instrumentation can be responsible for the failure of the optical performance of the system; consequently it is necessary to perform a preliminary investigation of the radiation resistance of such components in order to avoid unwanted malfunctioning.

802

803

Space devices are exposed to irradiation dose depending on several parameters, such as altitude, inclination and South Atlantic Anomaly (SAA). In case of Earth Orbiting missions we can distinguish two main dose ranges: a) low inclination Low Orbiting Missions (< 28°): in both northern and southern hemispheres, doses range between 1-10 Gy(Si)/year. b) high inclination Low Orbiting Missions (<20-85°): in both northern and southern hemispheres, doses range between 10-100 Gy(Si)/year. Doses involved in interplanetary missions are of the order of 104 Gy(Si)/year. In this work, we investigated the ionizing radiation damage of a set of optical coatings which can find application in space devices. In particular, our attention was focused on optical components to be used for Earth observation from the polar sun-synchronous orbit and for interplanetary missions. To simulate the radiation environment of Earth Orbiting missions, coating were irradiated at 50 Gy corresponding to the dose absorbed during 5 years of working of astronomical telescopes in the polar sun-synchronous orbit at an altitude of 700 km. To simulate the higher interplanetary radiation level, samples underwent further irradiation tests up to 8400 Gy. The optical coating behavior under irradiation was investigated performing spectrophotometric measurements before and after y-ray exposure. 2. Samples and experimental set-up Analyzed samples are described in Table I and they include an optical substrate and (single or multilayer) coatings that are typically used in optical instrumentation. All coatings, containing oxide and/or metal layers, were fabricated by PVD (Physical Vapor Deposition [6]) using materials of interest for the VIS-NIR spectral region. These materials were selected because they appear to be sensitive to y irradiation [7]. Table 1. Analyzes samples. Sample Substrate Single-layer material Optical component

Material Quartz Suprasil Aluminum oxide AI2O2 Yttrium oxide Y2O3 Protected mirror: SiCVAg

Samples were irradiated at the Calliope Co plant [8] which is located at the Research Centre ENEA-Casaccia (Rome, Italy), with a dose between 50 Gy (Low Dose Irradiation Test) and 8400 Gy (High Dose Irradiation Test). On all samples, measurements of both reflectance and transmittance were carried out by using a UV-VIS-NIR Perkin Elmer Lambda 900 Spectrophotometer. Measurements were performed in the spectral region from 250 to 800 nm.

804

3. Results and discussion Transmission measurements performed on quartz Suprasil evidence the absence of radiation induced variation on the optical properties of such substrate in the dose range between 50-8400 Gy. As a consequence of that, all the radiation induced variations observed on deposited films could be ascribed to modifications induced in the layer by ionizing radiation. As reported in our previous work [7], many coatings when exposed to the 50 Gy low irradiation dose, do not show a significant change in their optical performance. The only exceptions are represented by Al203/quartz whose transmittance slightly decreases and by Y203/quartz sample whose transmittance curve slightly shifts towards the infrared spectral region. On the contrary, relevant effects were observed at high irradiation doses between 1000-8400 Gy. In Figure 1 we reported the transmittance spectra of A1203 film before and after each irradiation dose: as it can be seen, the exposure to ionizing radiation induced a decrease of transmittance in the UV region.

(before irr.) (1000Gy) (3600Gy) (8400Gy)

95 o o c re *-»

90

c

85

Ito

2

x

Cs^

/A /

.^

^SZ'

*5>^r>

•y

i-

80

i

300

•

—

i

—

•

—

i

—

•

—

i

—

•

—

i

—

400 500 600 700 Wavelength (nm)

•

—

800

Figure 1. Transmittance spectra of AI2O3 film before and after irradiation up to 8400 Gy

In the case of Y 2 0 3 film (Figure 2a) y irradiation induces both a red-shift of the whole spectrum and a slight transmission decrease in the UV region. To better quantify the red shift, we studied the dependence of the wavelength shift on the applied dose for the three peaks characterizing the Y 2 0 3 transmission spectrum. Results concerning the peak at 550nm are reported in Figure 2b and they evidence some kind of kinetic process which seems to saturate at 4000 Gy.

805

100

25

95 20

I 90 15

85 80

5

(1000Gy) •• (3600Gy)

75

•(8400Gy)

70 300

400 500 600 700 Wavelength (nm)

(b)

104

(before irr.) (400Gy)

0 800

2000 4000 6000 Irradiation Dose (Gy)

8000

Figure 2. Transmittance spectra of Y2O3 film before and after irradiation, up to 8400 Gy (a); wavelength shift of the 550 nm peak (b).

As far as the Si0 2 protected silver mirror is concerned, in order to understand the irradiation resistance we analyzed both reflectance and transmittance variations, it turned out that the sample is strongly deteriorated by irradiation (Figure 3). 10

~ 8 0

o c n

• (before irr.) (1000Gy) (3600Gy) • (8400Gy)

6e

1 * 300

400 500 600 700 Wavelength (nm)

800

Figure 3. Transmittance spectra of Si02 protected silver mirror before and after irradiation, up to 8400 Gy.

806

4. Conclusions Most space instruments are exposed to fluxes of energetic particles that can induce performance deterioration. In this work we investigated the behaviour of substrates and thin film materials used in space applications, after exposure to y irradiation. Our studies confirmed stable behaviour of selected materials at 50 Gy low irradiation dose. On the contrary, after the high irradiation dose up to 8400 Gy the Si02/Ag mirror, Y 2 0 3 and A1203 coatings show a meaningful deterioration of the optical performances in the investigated spectral region. Further studies are in progress in order to investigate, from the structural point of view, the variations of the material properties induced by y exposure. Acknowledgments Authors are grateful to Dr. Stefania Baccaro (ENEA-Casaccia, FIS/ION) for the scientific support. References 1. M. Fernandez-Rodriguez et al, Thin Solid Films 455-456, 545 (2004). 2. E.R. Benton, E.V. Benton, NIMB 184, 255 (2001). 3. F. Spumy, Rad. Phys. and Chem. 61, 301 (2001). 4. L. Lanzerotti et al., Bell Lab TechnicalJournal (1997). 5. NASA PRACTICE NO. PD-ED-1258 (April 1996). 6. R.F.Bunshah et al., (Noyes Publications, New Jersey, 1982). 8. S. Baccaro, accepted to be published on IEEE Transactions on Nuclear Science. 7. S. Baccaro et al., ENEA Technical Report 28/FIS/2005.

FULL CHARACTERIZATION OF N O N - U N I F O R M L Y I R R A D I A T E D SILICON MICRO-STRIP SENSORS

G. CHIODINI INFN

Lecce - Italy

G. A L I M O N T I , P. D ' A N G E L O , M. E . D I N A R D O , L. M O R O N I A N D S. S A L A INFN

and Universita

degli Studi di Milano

- Italy

We describe a method we used to fully characterize micro-strip sensors, which were non-uniformly irradiated up to a fluence of ~ 1 0 1 4 lMeV equivalent neutrons per cm 2 . The method allows for a complete bidimensional mapping of the sensor characteristics over the entire active area. Information is gathered through the Q-V characteristic, measured scanning the sensor with an infra-red laser source. Q-V characteristics are then fitted to a simple analytical model, which returns local full-depletion voltages, carrier lifetimes, e t c . . . With the present method one can even obtain the profile of the absorbed fluence. The development and tuning of the present method have been done in the context of the R &; D programs for the micro-strip forward tracker of the BTeV experiment at the Tevatron.

1. Introduction One of main issues in employing Silicon detectors in HEP experiments is to certify through dedicated measurements their radiation tolerance l. This is particularly crucial for detectors, which are going to be used at hadron colliders, where the total fluence after ten years of operation can reach values well in excess of 10 14 lMeV equivalent neutrons per cm 2 . The picture is even more complicated in presence of highly non-uniform irradiation since the performance of the sensors can dramatically vary over the active area. In these cases a local characterization, point to point, of the sensors is required. To this extent we developed a simple analytical model, which we employed to fit the local Q-V characteristics measured on several points of a micro-strip sensor previously irradiated up to a fluence of ~10 1 4 200MeV protons per cm 2 . The local Q-V characteristic was measured by illuminating the sensor with infra-red laser pulses.

807

808

2. The Silicon micro-strip sensor and readout electronics The sensors we characterized are Silicon micro-strips type IB2 of the CMS experiment. They are of standard FZ p/n type, 320 /j,m thick and with fulldepletion voltage of 150 V. They have 512 strips, 116 mm long and with a 120 /im pitch. The active area is ~ 6 1 x l l 6 mm 2 . The readout electronics we employed is the TAA1 produced by Ideas with 3 /is shaping time. 3. Irradiation and global behavior We irradiated the sensors at Indiana University Cyclotron Facility (IUCF) with a proton beam of 200 MeV energy. The beam shape was measured with the wire scanner technique, it had a quasi-Gaussian shape with a a of ~19 mm. The beam was approximately centered on the edge of the sensors opposite to the readout electronics. We irradiated a sensor up to a peak fiuence of ~ 0 . 8 x l 0 1 4 lMeV equivalent neutrons per cm 2 . Sensors after the

0

100

KO

3CO

«00

Bias voltage [V]

»0

800

'

?

3

ll'

Fiuence [10

$

$

?

°J

I M e V e q . ncm*]

Figure 1. On the left: I-V characteristics at different fluences measured at room temperature. In the insert, in the upper right corner, is shown the I-V characteristic before irradiation. On the right: leakage current measured at 400 V at room temperature at different peak fluences. The straight line is the expected curve,

irradiation were always kept at a temperature lower than —12° C without applying any annealing cycle. In Fig. 1 we report the I-V characteristics measured at different fluences during the irradiation. In the same figure we compare the measured leakage current Iieak at 400 V for different fluences with the empirical formula keak = a ^ f f i where $\%£v is the integrated fiuence on the sensor and was calculated assuming a two-dimensional Gaussian shape with a peak fiuence of 0.8 x 1014 lMeV equivalent neutrons per cm2; d is the sensor thickness and a = 4 . 5 6 x l 0 - 7 A c m - 1 was taken from literature 2 . The expected behavior of the leakage current with the fiuence is confirmed by our measurements.

809 4. Laser measurements We illuminated the sensors on the strip side. In this situation, only the fraction of the laser light impinging on the sensor between adjacent strip metallizations could reach the bulk of the sensor. To reduce to a negligible level the dependence of the quantity of absorbed light on the position of its source, we illuminated a wide region, ~20 strips, by placing the sensor out of the focal plane of the lens. Given the ratio of metallization to pitch, 40 /zm/120 fxm, only 2/3 of the laser light penetrated the bulk. For each position of the laser beam on the sensor, we measured the total collected charge at different bias voltages. The total collected charge was obtained by summing up all the signals of the illuminated strips, once equalized and common-mode subtracted. In Fig. 2 we present two typical sets of measurements, which were performed in two particular regions of the sensor.

Figure 2. Q-V characteristics with superimposed the fit obtained from the model. On the left: data from the non-irradiated region. On the right: data from the highly irradiated region.

5. Collected charge model In this model we approximate our detector with a simple p/n diode with infinite plane electrodes. The validity of this approximation is only limited by the different electric field configuration near the p-implant. Interestingly enough, for long carrier lifetimes as for non-irradiated sensors, our model reproduces the data with good accuracy (cfr. Fig. 2). This means that the progression of the depleted region with the bias voltage is practically identical even in proximity of the p-implant. The main advantage of the present approach is that we can apply the simplest version of the Ramo theorem and obtain an analytical expression for the total collected charge, which is extremely advantageous for these kind of applications. The model

810

has 4 free parameters: carrier trapping-time, r , charge density generated by the laser, po, effective full-depletion voltage, V^, and V0ff, which is used to take into account electric field non-linearities presented in irradiated sensors. The actual full-depletion voltage is Vd = V'd + V0ff.

6. Results I will now discuss the progression of full-depletion voltage and carrier trapping-time along the strips with two significant examples of our measurements: laser spot centered on a lateral strip (77) and laser spot centered on a central strip (281). In Fig. 3 is reported the full-depletion voltage along the strips. For all the strips there is a big drop in the measured fulldepletion voltage in correspondence of the highest irradiated region (from 80 mm on). For the central strip the full-depletion voltage has a minimum in correspondence of ~90 mm, related to the intrinsic-Silicon-like behavior. In Fig. 4 is reported the trapping-time along the strips. For the lateral strip

BO 10

s £ 120

»

" '

* *

JO JO

Legend Strip 77

10

" Coord. Along Strips [mm]

Legend 1 — Strip 281 i

.

.

i

. ,

Coord. Along Stripe [mm]

Figure 3. Full-depletion voltage along the strips. On the left: lateral strip, 77. On the right: central strip, 281.

the trapping-time drops to ~50 ns, whereas for the central strip it drops to ~20 ns. This is coherent with a proton irradiation beam centered on the edge of the sensor. The Q-V characteristics, in correspondence of the most

irradiated region, were fitted by inverting the junction side in our model, thus exchanging the role of electrons and holes. From the trapping-time we were even able to recover the radiation fluence 3 using the following relation ^ = J^l^nV• The sensor fluence-map was then fitted with a 2DGaussian shape giving the irradiation beam profile shown in Fig. 5. The beam properties measured at IUCF are in very good agreement with our measurements.

811

%

75

60

BS ' 00 06 100 106 110 Coord. Along Strips [mm]

I

so

H HO i u n o 11 Coord. Along S t r l p i (mm)

Figure 4. Trapping-time along the strips. On the left: lateral strip, 77. On the right: central strip, 281.

oV

eb

'

ub' ' 'ioo'

lio' ' W

160 130' 2O0 710 Coord. Along Strips [I

Figure 5. Beam contour from laser measurements (color scale is in unit of 10 14 lMeV equivalent neutrons per cm 2 ). The fit parameters are: (1.01±0.18)xl0 1 4 lMeV equivalent neutrons per cm 2 , mean along strips 120±6 mm, u along strips 21.9±3.0 mm, mean transversal to strips 33.4±1.5 mm, a transversal to strips 20.0±0.8 mm. Dashed line: sensor position.

7. C o n c l u s i o n s

With a very simple method based on an analytical model for the Q-V characteristics we were able to extract all the physical information required to characterize a Silicon sensor for use in HEP experiments. The method allows for a local characterization of the sensor. The analytical model has 4 free parameters only and it is capable to extract reliable information even in case of highly radiation-damaged sensors. This method has also been proven to provide a very good estimate of the dose locally absorbed by the sensor. Thanks to this method, we can conclude that the sensors we tested are well suited for use in the forward region of hadron collider experiments. References 1. ROSE collaboration, Nucl. Inst, and Meth. A 466 (2001) 308. 2. D. Bechevet et al., Nucl. Inst, and Meth. A 479 (2002) 487. 3. G. Kramberger et al, Nucl. Inst, and Meth. A 476 (2002) 645.

BEAM ENERGY MONITOR FOR 4-10 MEV ELECTRON ACCELERATORS P. FUOCHI**, M. LAVALLE, U. CORDA ISOF-CNR, Via P. Gobetti 101 Bologna, 1-40129, Italy A. KOVACS Institute of Isotopes, CRC, P.O. Box 77 Budapest, H-1525, Hungary K. MEHTA Brimner Strasse 133-3-29 Vienna, A-1210, Austria The electron beam energy is one of the critical parameters of electron accelerators since it can affect the dose distribution inside the body or the product irradiated with the beam. An easy-to-use device has been developed for monitoring small variations in the electron beam energy during an irradiation run. It involves measurement of currents (or charges) collected by two identical aluminium plates, except for their thickness, and electrically insulated from each other, located in the beam. The ratio of these two currents (or collected charges) is quite sensitive to the beam energy; optimization of sensitivity is obtained by selecting the appropriate thickness of the front plate depending on the nominal beam energy. In this paper, we present the results of our investigation of using this energy device at energies from 4 to 10 MeV.

1. Introduction For radiation processing as well as for therapeutic use of electron beams from an accelerator, the beam energy is one of the critical parameters. A basic prerequisite for successful radiation treatment is that the variations in the beam energy be minimised since any small variation could affect the depth-dose distribution in the irradiated body or the product. A typical method to determine the electron beam energy is to measure the depthdose distribution in a reference material [1]. But since this method is time consuming, a faster method, sufficiently precise, has been developed by us for frequent checks. This

'Corresponding author: P. Fuochi, E-mail: [email protected]: fax: +39-051-6399844

812

813

method is based on the fact that any beam energy variation affects the charge-deposition distribution, just like dose distribution. The main difference is that the charge or current can be measured on-line using an electron-beam energy device. 2.

Experimental

The electron-beam energy device was constructed in the form of a thick-walled electrically grounded aluminium cage shaped like a Faraday cup (see Fig. 1) [2]. Two aluminium collector plates of 100 mm diameter and different thicknesses were inserted into the cage separated from each other by an air gap of 5 mm, and electrically insulated from the cage by ceramic posts. The thickness of thefrontplate, based on the measured charge-deposition distribution in the material of construction, was determined by the position of the peak in the differential charge-deposition distribution curve (Fig. 2). This was accomplished by adding a series of discs of 0.5 mm thick and 100 mm diameter, one disc at a time after each charge measurement Based on the results of these measurements, the optimum thickness for the front plate was established to be 12 mm (see Fig. 2) and 5 mm for Bologna and Budapest accelerators, respectively. The back plate has a thickness of 25 mm which is sufficient to stop all the electrons at the maximum operating beam energy. The two plates were connected by two coaxial cables to two current integrators, placed outside the irradiation room, using a low leakage capacitor as input storage network. The tests were carried out using electron accelerators located in two laboratories, with the operating characteristics as reported in Table 1. Table 1.Operating characteristics of the accelerators used for e-beam measurements. Nominal beam energy

Pulse width

Repetition rate

CMeVI

(us}

(Hz)

Bologna Vickers

12

0.5-5.0

50

LINAC 30-300 uA Average current

Budapest Tesla, LPR-4

4

2.6

50

LINAC 2-20 uA Average current

Beam characteristics

To be able to use the energy device to measure beam energy or detect any shift in the beam energy, a relationship was established between the response of the energy device and the most probable electron beam energy Ep. To achieve this, the beam energy was varied by changing the pulse width and/or the average beam current. The beam energy was determined by measuring the depth-dose distribution by irradiating radiochromic dosimetric film strips inserted in a 60° aluminium wedge-pair [3].

814

Measurements of the charge collected on the two plates of the energy device were done immediately before and after the irradiation of the aluminium wedge-pair. The irradiated radiochromic film strips were read by a spectrophotometer equipped with an automatic strip-feed mechanism operating at a measurement resolution of 10 points/cm. The size of the analysing beam ( 0 3 mm) determined the spatial resolution of the measured depth-dose distribution. The charge collected on the two plates of the energy device, which gives the response of the device, was measured during irradiation (on-line measurements).

e

*ry^ Figure 1. Cross-sectional view of the electron-beam energy device. Reprinted from Ref. [2] with permission of Elsevier Science.

differential collected charge

30 S O 20 ~°

1 mm Al

Figure 2. Differential charge-deposition distribution in Al for Bologna accelerator (note the peak at 12 mm). Depth-dose distribution under the same irradiation conditions is also shown, with the practical electron range Rp.

815 The practical electron beam range Rv in aluminium was determined from the measured depth-dose distribution [4]. The value of Rp (in cm) was then used to calculate Ev based on the following expression [1]: Ep (MeV) = 0.2 + 5.09 Rp

(1)

The response of the energy device (referred to as energy ratio) was calculated as follows: Charge from the front plate Energy ratio = Sum of the charges from the two plates 3.

Results and Discussion

Figure 2 shows the results of the experiments for determining the charge deposition with depth and the depth-dose distribution using an aluminium wedge for the Bologna accelerator. Similar results were obtained for the Budapest accelerator. From these data, the thickness of the front plate and Rp were determined for each accelerator. The data on energy ratio and electron beam energy for the Budapest TESLA LPR-4 accelerator and the Bologna 12 MeV Vickers accelerator are plotted in Fig. 3 with the 95 % prediction limits, which represents about ±0.3 MeV.

11 10 9

>CD

8

\

7 6 5 4

j

0.4

0.5

i

i

0.6

i

i

0.7

i

i

0.8

i

L

0.9

Energy ratio Figure 3. Correlation between the measured values of the energy ratio and the most probable beam energy Ep for the two types of accelerators. The computer generated best linear fit to these data as well as 95% prediction limits are shown.

816 These data were collected for different conditions of the accelerators, as described in Sec. 2, specifically to cover as wide a range as possible for the beam energy. All values are average of several replicate measurements. Figure 3 shows that there is a good linear correlation between these two parameters, suggesting that this relationship can be exploited for quickly assessing any variation in the beam energy by on-line measurements using the energy device. The slope of the linear fit indicates the sensitivity of the device to detect any shift in the beam energy. The largest contribution to the uncertainty in Ep arises from the determination of Rp with the aluminium wedge-pair. Because of this uncertainty, the usefulness of the device may be limited for low beam energy due to the relatively short length of the exposed radiochromic film compared to the size of the analysing beam ( 0 3 mm). However, this can be easily overcome by more precise energy measurements using improved optical analysis equipment. It should also be borne in mind that the energy device should always be calibrated at the same accelerator where it is to be used for energy checks. The advantages of using this device include: > > >

rugged construction, ease of manufacturing and low cost, simplicity of on-line measurements, precision suitable for process control application.

Acknowledgments This work was done within the Italy-Hungary (CNR-MTA) co-operation research program. We also acknowledge the valuable technical assistance provided by Eng. A. Martelli, Eng. P. Hargittai and Mr. A. Monti. References 1. 2. 3. 4.

ICRU Report No. 35, International Commission on Radiation Units and Measurements, Bethesda, MD, USA, (1984). P.G. Fuochi, M. Lavalle, A. Martelli, U. Corda, A. Kovacs, P. Hargittai, K. Mehta, Radial Phys. Chem. 67,593 (2003). K. Mehta, P.G. Fuochi, A. Kovacs, M. Lavalle, P. Hargittai, Radial Phys. Chem. 55,773 (1999). ISO/ASTM 51649, Annual Book of ASTM Standards 12.02, Philadelphia, PA, USA, 1058 (2004).

OPTICAL LINK OF THE ATLAS PIXEL DETECTOR K.K. Gan, W. Fernando, P.D. Jackson, M. Johnson, H. Kagan, A. Rahimi, R. Kass, S. Smith Department of Physics, The Ohio State University, Columbus, OH 43210, USA P. Buchholz, M. Holder, A. Roggenbuck, P. Schade, M. Ziolkwoski Fachbereich Physik, Universitaet Siegen, 57068 Siegen, Germany The on-detector optical link of the ATLAS pixel detector contains radiation-hard receiver chips to decode bi-phase marked signals received on PIN arrays and data transmitter chips to drive VCSEL arrays. The components are mounted on hybrid boards (opto-boards). We present results from the irradiation studies with 24 GeV protons up to 32 Mrad (1.2 x 1015 p/cm2) and the experience from the production.

1.

Introduction

The ATLAS pixel detector [1] consists of three barrel layers, three forward, and three backward disks which provide at least three space point measurements. The pixel sensors are read out by front-end electronics controlled by the Module Control Chip (MCC). The low voltage differential signal (LVDS) from the MCC is converted by the VCSEL Driver Chip (VDC) into a single-ended signal appropriate to drive a VCSEL (Vertical Cavity Surface Emitting Laser). The optical signal from the VCSEL is transmitted to the Readout Device (ROD) via a fiber. The 40 MHz beam-crossing clock from the ROD, bi-phase mark (BPM) encoded with the command signal to control the pixel detector, is transmitted via a fiber to a PIN diode. This BPM encoded signal is decoded using a Digital Opto-Receiver Integrated Circuit (DORIC). The clock and command signals recovered by the DORIC are in LVDS form for interfacing with the MCC. The ATLAS pixel optical link system contains 632 VDCs and 544 DORICs with each chip containing four channels. The chips are mounted on 272 chip carrier boards (opto-boards). Each opto-board contains seven active optical links. The silicon components (VDC/DORIC and PIN) of the optical link will be exposed to a maximum total fluence of 3.7 x 1014 1-MeV neq/cm2 817

818

during ten years of operation at the LHC. The corresponding fluence for the GaAs component (VCSEL) is 2 x 1015 1-MeV neq/cm2. We study the response of the optical link to a high dose of 24 GeV protons. The expected equivalent fluences at LHC are 6.3 and 3.8 x 10M p/cm2, respectively. In this paper we describe the results from the irradiations and the experience from production. 2. VDC Circuit The VDC is used to convert an LVDS input signal into a single-ended signal appropriate to drive a VCSEL in a common cathode array. The output current of the VDC should be variable between 0 and 20 mA through an external control current, with a standing (dim) current of ~1 mA to improve the switching speed of the VCSEL. The nominal operating current for a VCSEL is 10 mA. The electrical output should have rise and fall times (20-80%) between 0.5 and 2.0 ns; nominally 1.0 ns. In order to minimize the power supply noise on the opto-board, the VDC should also have constant current consumption independent of whether the VCSEL is in the bright (on) or dim (off) state. Figure 1 shows a block diagram of the VDC circuit. An LVDS receiver converts the differential input into a single-ended signal. The driver controls the current flow from the positive power supply into the anode of the VCSEL. The VDC circuit is therefore compatible with a common cathode VCSEL array. An externally controlled voltage (Viset) determines the current Let that sets the amplitude of the VCSEL current (bright minus dim current), while another externally controlled voltage (Tunepad) determines the dim current. The driver contains a dummy driver circuit which, in the VCSEL dim state, draws an identical amount of current from the positive power supply as is flowing through the VCSEL in the bright state. This enables the VDC to have constant current consumption. *mmte

V.MI

Figure 1. Block diagram of the VDC circuit.

3. DORIC Circuit The DORIC decodes BPM encoded clock and data signals received by a PIN diode. The BPM signal is derived from the 40 MHz beam-crossing clock by

819

sending transitions corresponding to clock leading edges only. In the absence of data bits (logic level 0), the resulting signal is a 20 MHz square wave. Data bits are encoded as an extra transition at beam-crossing clock trailing edges. The amplitude of the current from the PIN diode is expected to be in the range of 40-1000 uA. The 40 MHz clock recovered by the DORIC is required to have a duty cycle of (50 ± 4)% with a total timing error (jitter) of less than 1 ns. The bit error rate of the DORIC is required to be less than 10~u at end of life. Figure 2 shows a block diagram of the DORIC circuit. In order to keep the PIN bias voltage (up to 15 V) off the DORIC, we employ a single-ended preamp circuit to amplify the current produced by the PIN diode. Since singleended preamp circuits are sensitive to power supply noise, we utilize two identical preamp channels: a signal channel and a noise cancellation channel. The signal channel receives and amplifies the input signal from the anode of the PIN diode, plus any noise picked up by the circuit. The noise cancellation channel amplifies noise similar to that picked up by the signal channel. This noise is then subtracted from the signal channel in the differential gain stage. To optimise the noise subtraction, the input load of the noise cancellation channel should be matched to the input load of the signal channel (PIN capacitance) via an external dummy capacitance.

input

HBIM

C&jxfsrry

Mfl.

•girt

H*«*t

Figure 2. Block diagram of the DORIC circuit.

4. Irradiation Studies We implemented the VDC and DORIC in the standard deep submicron (0.25 um) CMOS technology. Employing enclosed layout transistors and guard rings [2], this technology was expected to be very radiation hard. We have extensively tested the chips to verify that they satisfy the design specifications [3]. The performance of the chips on opto-boards has been studied in detail. The typical PIN current thresholds for no bit errors are low, < 20 uA.

820

In the last four years, we have performed four irradiations of the optical electronics using 24 GeV protons at CERN. In June 2004, we irradiated four opto-boards containing complete optical links up to the total fluence of 1.2 x 1015 p/cm2 (32 Mrad). The boards were mounted on a shuttle so that they could be pulled back from the beam line each day for annealing of the VCSEL arrays after each dosage of ~ 5 Mrad (5 hours). We observed that the PIN current thresholds for no bit errors were all below 40 uA and remained constant through out the irradiation. The optical power was also monitored and some VCSEL channels lost some of their power after the irradiation due to insufficient time for adequate annealing as shown in Fig. 3. We received the irradiated optoboards in September 2004 and after extended annealing, the optical power of all VCSEL channels were well above 350 uW, the specification for absolute minimum power after irradiation. This confirms the radiation hardness of the fully populated opto-boards.

0

20

40

60

80

100

120

140

Time (h)

Figure 3. Optical power as a function of time (dosage) in the data channels for one of the optoboards with seven active links in the shuttle setup.

5. Opto-board Production The opto-board production is governed by a rigorous quality assurance procedure. Each board must passed 72 hours of bum-in at 50°C, followed by 18 hours of 10 thermal cycles between -25 and 50 °C, and then 8 hours of optical and electrical measurements. Our initial plan was not to test the chips before mounting on an opto-board due to the high yield during the pre-production exercise. However, we soon encountered several boards that drawn excessive currents of which we were able to trace the problem quickly to a power to ground short using a high-resolution thermal imaging camera. Consequently, we

821

tested all chips prior to the mounting on an opto-board. We are also able to recover opto-boards populated with bad chips by stacking new chips on the top. Each reworked opto-board must pass the same rigorous QA procedure but it will be classified as second class for use as a spare. A total of 21 opto-boards are recovered. The production proceeds smoothly as shown in Fig. 4 with high yield, ~ 93%. The production is now completed. 240

0 +

4

8

12

Week 16

20

24

28

32

200 -

160•o

g 120m 80 400-U #<>

H&

^

,flS>

„\1

,(*

&


^

Date

Figure 4: Total number of opto-boards produced as a function of time. The kink at ~ 21 a corresponds to the switch from producing one flavor of opto-broads to the other. Also shown is the expected yield (cross) and total number of failed boards (triangle) which may decrease with time sometimes due to successful rework.

6. Summary We have developed opto-electronics that meet all the requirements for operation in the ATLAS pixel optical link and appear to be sufficiently radiation hard for ten years of operation at the LHC. The production of the opto-boards has proceeded smoothly and is now completed. Acknowledgments This work was supported in part by the U.S. Department of Energy under contract No. DE-FG-02-91ER-40690 and by the German Federal Minister for Research and Technology (BMBF) under contract 056Si74. References 1. ATLAS Pixel Detector Technical Design Report, CERN/LHCC/98-13. 2. G. Anelli et al., IEEE Trans, on Nucl. Sci. 46, 1690 (1999). 3. K.E. Arms et al, Nucl. Instr. Meth. A554, 458 (2005).

ION ELECTRON EMISSION MICROSCOPY FOR SEE STUDIES P. GIUBILATO, D. BISELLO, A. KAMINSKI, S. MATTIAZZO, M. NIGRO, D. PANTANO, R. RANDO, M. TESSARO Physics Department,

University od Padova and INFNPadova, Padova, Italy.

via Marzolo 8, 35131,

J. WYSS SMS Department,

University of Cassino and INFN Pisa, via Buonarroti 2, 56127, Pisa, Italy.

At the INFN Legnaro Laboratories (Padova, Italy) a new instrument has been recently installed in the SIRAD beam line, a facility devoted to the study of radiation induced damage in microelectronic devices, systems and materials. This new instrument, based on an idea first proposed at SANDIA, consists of an electronic microscope system capable of recognizing with micrometric resolution the impact points of single ions impinging on the target. Linking the ion impact positions as recorded by the microscope with the single event effects (SEE) induced in a microelectronic component under test (DUT) allows one to perform a micro-map of the sensitive areas of the device under test. Compared with traditional microbeams, the standard tool to perform SEE micro mapping, an IEEM system is relatively simple and it may be installed on preexisting beam-lines with only minor requirements, an important feature that allows IEEM applications at high-energy ion accelerators that allow a wide selection of ion species (cyclotrons, tandems).

1. SEE studies on microelectronic devices. The study of the effects of radiation on semiconductor devices is an important and lively field in scientific and technological research. In particular, radiation tolerance is a fundamental issue for electronic devices and systems in many applications such as space research, telecommunications, avionics and highenergy physics. The SIRAD irradiation facility, located at the 15 MV Tandem accelerator of the INFN Laboratory in Legnaro (Italy), is actively dedicated to bulk damage and to Single Event Effect (SEE) studies in semiconductor devices and electronic systems for high energy physics and space applications [1]. An impinging ion will deposit energy in the semiconductor material (silicon) of an electronic device and generate electron-hole pairs. High electron-hole pair densities along a single ion track can influence the device functionality if the charge is generated in a high electric field region and/or it is collected at a

822

823 sensitive node of the circuit, so single ionizing ions are able to induce device failures of various types [2]. For SEE studies the SIRAD facility can deliver beams ranging from 7Li+3 (60 MeV) to 197Au+25 (300 MeV) with a present limit in magnetic rigidity of about 1.4 T-m. Global device SEE characterizations are routinely performed at SIRAD using broad beams to uniformly irradiate areas up to several cm2 of a DUT. To improve this capability, the group is developing Ion Electron Emission Microscopy (IEEM) capabilities that will allow the location of SEE sensitive points of a DUT with micrometric resolution. In the IEEM technique [3] a broad (not micro-focused) ion beam irradiates the portion of the DUT that is inside the field of view of a commercial photon electron emission microscope. The ion impact positions are reconstructed by collecting the secondary electrons emitted from the DUT surface during ion impact and focusing them onto a two-dimensional electron detector, placed at the focal plane. 2. IEEM practical implementation Practical working resolution of an actual IEEM is basically limited by the aberration introduced by the electronic microscope itself, and the necessity to maximize the electrons transmission efficiency: exactly the same as for an optical system, when maximizing light collection by widening lens aperture results in a loss of image sharpness. Our electron emission microscope can reach a resolution of 0.6um over a circular field of view (FOV) of 250um in diameter with an electron transmission efficiency of about 10-30% [4]. As a focal plane electron detector, we use an high gain (up to 107 e7e") double stack Micro Channel Plate (MCP) coupled with a fast phosphor layer. In the SIRAD IEEM (Fig. 1) the focal plane MCP-phosphor stack is annular to allow for the ions to pass axially through the microscope and impact normally the DUT surface. The intrinsic spatial resolution set by the physical size of micro-channels of the MCP (some tens of microns) suits the intrinsic resolution of the IEEM, and the phosphor response time (~100ns) is fast enough to time-resolve each individual ion impact. The photon electron microscope gathers and focuses the electrons emitted by the DUT on the MCP placed at its focal plane, where they are first multiplied and then converted to photons by the subsequent phosphor layer (~ 70 y/e" efficiency). An optical system extracts this resulting image from the vacuum chamber onto an image intensifier where it is further amplified and finally analyzed by a fast photon position sensitive detector. This opto-electronic approach, original to the SIRAD IEEM, allows the use of complex sensing devices for greater sensing performance, flexibility and ease of use. The use of a

824

variable gain image intensifier (up to 106 e7e") is necessary as the light collection efficiency of the system, due to the complex optical path, is low, in fact not better than 0.9%. A very positive aspect of using the image intensifier is that it allows one to reduce random noise by adjusting the focal plane MCP to low gain levels, the optimal working condition. Indeed the noise of the image intensifier is orders of magnitude smaller than the noise of the focal plane MCP as amplified by the image intensifier.

Figure 1. SIRAD IEEM schematic.

3. Position sensor As the output of our setup sensing chain consists of a continuous stream of short, blue light spot which position varies accordingly to the original ion impact position on DUT, a suited final sensor must be able to record the 2D position coordinates of any spot together with its time position inside the stream. The most important feature of the choice is its spatial resolution: it must suit the microscope resolution, this meaning to be capable of distinguishing at least 250 um /0.6um ~ 400 linear points on the FOV diameter. Moreover, it should also exhibit sufficient time resolution to distinguish individual ion impacts in the

825

FOV. For SEE applications, a maximum sustainable ion flux of-1000 ions/s in the FOV of the microscope (2><106 ions/cm2-s) requires a sensor with a repetition rate of at least 10 kHz. Commercially available CCD or CMOS devices do not reach such frame rates with the desired spatial resolution, not to mention the difficulty of performing any useful real time analysis from the resulting data stream that a similar device would give (~21 GBit/s). One solution we adopted [5] uses a position sensitive device (PSD), a device working accordingly to the photodiode principle that splits the charge generated by the incoming light into a resistive layer: the position of incident light spot is obtained by weighting the pulses collected by two orthogonal couples of electrodes. Even if this system has worked as development test arrangement up to now, its spatial resolution results strongly limited by the high electromagnetic noise level present during irradiation shift and is not sufficient to match our needs. A more advanced and conceptually different novel device had been first proposed [6] and then realized to solve this problem. This device uses a pair of linear NMOS chip to read the light spot position, one for each coordinate: a complex optical system first splits the image and then squeezes each copy onto one of the two sensor. One must then read out just the square root of the number of pixels of an equivalent square sensor, i.e. 1000 pixels instead of 1 million, and consequently the reading speed is decreased by the same factor. A first prototype of this new device is full working and first beam test are planned for autumn 2005. 4. First results and conclusion Figure 2 shows four images taken by our PSD system with four different ion species using a copper grid with 70um pitch as target, the ion beam coining from the upper right. The IEEM was in a non-axial configuration and the ions impinged at an angle of 70° respect to the IEEM axis with a flux in the field of view of about 500 ions/s. The spurious electron emission from the backplane of the grid confuses the images, but the resolution is clearly not better than 2jj,m, a factor 3 worse than our goal. As explained before, the PSD system cannot offer a better resolution; by using the new NMOS solution, should overtake this limitation. Nevertheless, IEEM microscope demonstrated to be able to take space and time resolved images of heavy ions impacts on a target, even if the beam onto the target is not yet orthogonal, and the desired spatial resolution have not yet been achieved. The present SIRAD beam line setup is undergoing a major upgrade, and the final setup is expected to give a practical tool to perform

826

SEE studies with energetic ion species and with a relatively cheap and unobtrusive instrument.

Figure 2. Each image is the representation of about half million events. For every event, XY position, time and other relevant parameters are stored.

References 1. J. Wyss, D. Bisello and D. Pantano, Nucl. Instr. Meth. A 462 (2001) 426. 2. A. Holmes - Siedle, L. adams, (.(Handbook of radiaion effect », Oxford Science, 1993 . 3. B. L. Doyle, G. Vizkelethy, D. S. Walsh, B. Senftinger and M. Mellon., Nucl. Instr. Meth. B 158 (1999) 6. 4. D. Bisello, A. Kaminsky, A. Magalini, M. Nigro, D. Pantano, S. Sedykh and J. Wyss, Nucl. Instr. Meth. B 181 (2001) 254. 5. D. Bisello, A. Candelori, M. Dal Maschio, P. Giubilato, A. Kaminski, M. Nigro, D. Pantano, R. Rando, S. Sedykh, M. Tessaro and J. Wyss, Nucl. Instr. Meth. B 210 (2003) 146. 6. D. Bisello, M. Dal Maschio, P. Giubilato, A. Kaminsky, M. Nigro, D. Pantano, R. Rando, M. Tessaro and J. Wyss, Nucl. Instr. Meth. B 219-220 (2004) 1000 -1004.

AN ANALYSIS OF THE EXPECTED DEGRADATION OF SILICON DETECTORS IN THE FUTURE ULTRA HIGH ENERGY FACILITIES** IONEL LAZANU University of Bucharest, Faculty of Physics, POBox Bucharest-Magurele, Romania e-mail: i lazanu&vahoo.co.uk

MG-11,

SORINA LAZANU National Institute for Materials Physics, POBox MG-7, Bucharest-Magurele,

Romania,

In this contribution we discuss how to prepare some possible detectors - only silicon option being considered, for the new era of HEP challenges because the bulk displacement damage in the detector, consequence of irradiation, produces effects at the device level that limit their long time utilisation, increasing the leakage current and the depletion potential, eventually up to breakdown, and thus affecting the lifetime of detector systems. Physical phenomena that conduce to the degradation of the detector are analysed both at the material and device levels, and some predictions of the time degradation of silicon detectors in the radiation environments expected in the LHC machine upgrade in luminosity and energy as SLHC or VLHC, or at ULHC are given. Possible effects at the detector level after high energy cosmic proton bombardment are investigated as well. Time dependences of these device parameters are studied in conditions of continuous irradiation and the technological options for detector materials are discussed, to obtain devices harder to radiation.

1. Introduction: the Future Ultra High Proton Facilities Particle physics has made striking advances in the last fifty years in describing the intimate structure of matter and the forces that determine the architecture of the universe. Major steps towards answering the open questions would come from developing global theories but invoke energy scales that are far beyond the reach of conceivable accelerators and from obtaining new experimental results at higher energies, or at least observable traces of these theories at accessible energies. Experiments relevant to this quest are not restricted to the highest &

Work in the frame of the CERN RD-50 Collaboration This work has partially been supported by the Romanian Scientific National Programmes CERES and MATNANTECH, under contracts C4-69/2004 and 219 (404) 72004 respectively.

827

828

energy frontier; they are also possible at low energy scales. A discussion of the search for new physics at the future collider facilities is given in Ref. [1]. The LHC at CERN will be operational from 2007. But the physics results expected from this facility will not be able to clarify all the unknowns of the Standard Model. So, despite the technological difficulty, significant upgrades of the accelerator in energy and luminosity are considered as SLHC and VLHC respectively. The upgrade path will be defined by the results from the initial years of LHC operation, and now only possible scenarios can be done up to 240 TeV, the final energy of the project. Ultimate Large Hadron Collider (ULHC) is imagined as accelerating protons to energies above 107 TeV in a ring with a circumference L=4><104 km around Earth and such a facility will permit to explore the "intermediate" energy scale, about 107 TeV, between the electroweek scale (1 TeV) and the scale of grand unification (around 1014 TeV). Table 1 Main parameters of future colliders Parameters

LHC

SLH (stage 1)

SLHC (stage 2)

VLHC (stage 1)

Vs fTeVl Circumference [kiri| Luminosity

14 26.7

14 26.7

28 26.7

40 233

1 25 = 117 *80 20

10 12.5

1(10)

1 18.8 « 143 «100

xio34rcm-y'i

Bunch spacing fnsl rjpp-total [mb] Opp-inel. [mb] Interactions <ET> charged partJMeV]

450

« 136 «90 100 500

VLHC (stage 2) 175 (up to 240) 233

ULHC

1-5-10 18.8 « 166 »140 up to 250

1 (10)

600

2xl0 7 40000

770 «270

1280

Other possibility to have access to ultra high energy particles, complementary to colliders is the exploitation of the of astroparticles or cosmic rays (CR). Protons are the most abundant charged particles in space and in this contri-bution aspects of the CR primary proton spectrum are discussed, in the kinetic energy range 0.2 -*- 200 GeV, in the neighbourhood of Earth, as have been measured by the Alpha Magnetic Spectrometer during space shuttle flight STS-91. The energy range between 105-108 TeV of the CR is closed by the expected energy of protons at ULHC, but the flux is irrelevant for degradation effects in detectors to be considered. In this work we discuss how to prepare some possible detectors - only silicon option being considered, for the new era of HEP challenges, because bulk displacement damage in the detector, consequence of irradiation, produces effects at the device level that limit their long time utilisation, increasing the

829 leakage current and the depletion voltage up to breakdown, and thus affects the lifetime of detector systems. The main parameters of LHC and of future colliders, relevant for the present investigation, are presented in Table 1. For ULHC, estimates of some physical quantities are based on the linear extrapolation of known results, but surprises are not excluded at ultra high energies. 2. Interactions of Particles in the Detector and the of Kinetics of Defects The nuclear interaction between the incident particle and the lattice nuclei produces bulk defects: the recoil nucleus(i) produced in the interaction are displaced from lattice positions into interstitials, their initial positions being vacant. Then, the primary knock-on nucleus, if its energy is large enough, could produce new displacements and the process continues as long as the ener-gy of the colliding nucleus is higher than the threshold for atomic displa-cements. The physical quantity characterising the process is the concentration of primary defects produced per unit of fluence of incident particles (CPD). The values of CPD corresponding to the expected <ET> energies for the spectra after primary interactions at future colliders are in the vicinity of the asymptotic high energy limit, where:

CPD-^E^^-^L^h < JLL-foK, and have approximately the values in the same domain. Here E is the kinetic energy of the incident particle, N is the atomic density of the target element, Ed is the threshold energy for displacements (function on the orientation of the target lattice), ERi - the recoil energy of the residual nucleus produced in interaction i, and cr^m\ - the cross section of the interaction between the incident particle and the nucleus of the lattice for the process or mechanism (i), responsible in defect production or the total cross-section respectively. CPD is not proportional to the modifications of material parameters after irradiation, due to the subsequent interactions of vacancies and interstitials and with other defects and impurities (only phosphorus, carbon and oxygen are considered as pre-existent) in the lattice. The formation and time evolution of complex defects, associations of primary defects or of primary defects and impurities are studied in accord with the model developed by the authors; see for example [2] and previously author's papers cited therein.

830

The primary point defects considered here are silicon self interstitials and vacancies: "classic" vacancies and fourfold coordinated defects - SiFFCD; new defect predicted by Goedecker and co-workers [3]. The characteristics of the SiFFCD defect have been determined indirectly by the authors [4]. So, Si—^—»(V + Si F F C D )+1, where the rate G is a sum of contributions from thermal generation and irradiation. The integral irradiation rate is a convolution between CPD and the flux of particles. In the irradiation rate, the information about the identity and characteristics of primary particles that produce degradation is lost. The phenomena occurring in the material after primary defect production are considered in the frame of the theory of diffusion limited reactions. The complete list of reactions considered was done in Ref. [4]. The model predicts absolute values for microscopic quantities - concentrations of defects and the characteristics of the device, without free parameters. A good accord with available experimental data was obtained [4]. 3. Predictions of the Behaviour of Detectors The experimental results and model calculation for step by step irradiation at high rates of generation of defects, up to fluences expected at new colliders, indicate a very good agreement see [5], but the degradation is not clearly correlated with oxygen concentration in silicon or with the resistivity of the starting material. Consequently, for these experimental conditions, it is not possible to make a choice between FZ and DOFZ materials. In Figs. 1 and 2, the time dependence of the leakage current and of effective carrier concentration of for FZ and DOFZ materials, in conditions of continuous irradiation at rates between those corresponding to cosmic protons up to new colliders are presented. The rates of generation of defects are about 200 pairs/cm3 for cosmic particles, about 7* 108 pairs/cm3 for LHC, and roughlylO times higher for SLHC and VLHC.

Time [years]

Time (years]

Figure 1. Time dependence of the volume density of the leakage current for FZ and DOFZ silicon.

831 For leakage current, an approximately linear dependence of time is obtained in log-log scale, FZ materials presenting a lower slope in com-parison with DOFZ ones. The produced degradation of silicon detectors scales with luminosity, and is roughly independent on particle spectra. If silicon detectors are exposed at primary protons fluxes from CR, the degradations are minor, and for these rates of generation of defects, DOFZ material is more adequate. For higher rates, both materials present inversion. In Fig. 2, the rates corresponding to the beginning of inversion are also presented.

III

SFZ

•

. . . . 510"WcmJ 1 10'VlW 5 10° Wcm1

•

. SIFZ

1 10" Wcm' 510*vlW 1 10'VI/cmJ S10'vi/cm3

7,9x10 a.oixio

...

w .! il

SiFZ

|

I

L

1

Sio'wcm 1

200 Wcm3 1000 Wcm' 3000 Wcm3

-M

•H-8K10"

s^

1 10'Wcm3

x^

\:\i SI DOFZ

1 Time [years]

. f

•f

SI DOFZ

Time [yearej

Figure 2. Time dependence of Nefffor detectors from FZ and DOFZ Si, for different rates of irradiation

4. Conclusions and Summary The present model is able to reproduce main features of available experimental data of FZ and DOFZ Si and permits to understand the mechanisms of degradation. At high rates of generation of defects, the contribution of primary defects is important and must be considered. No major differences in degradation between FZ and DOFZ technologies are obtained and from this point of view a decision is not possible. Because the processes are temperature dependent, in order to diminish these macroscopic effects, the decrease of the temperature is strongly required. References 1. F. Gianotti, ICFA, 9/10/2002. 2. I. Lazanu, S. Lazanu, Phys. Scripta 71 (2005) 31. 3. S. Goedecker et al., Phys. Rev. Lett. 88 (2002) 235501. 4. I. Lazanu, S. Lazanu, http://xxx.lanl.gov arXiv physics/0507058. 5. S. Lazanu, I. Lazanu, paper presented to PSD-7, Liverpool, Sept. 2005.

INVESTIGATION OF VLSI BIPOLAR TRANSISTORS IRRADIATED WITH ELECTRONS, IONS AND NEUTRONS FOR SPACE APPLICATION P. D'ANGELO 1 , G. FALLICA 2 , A. GALBIATI 1 , R.MANGONI 1 , R. MODICA 2 , S.PENSOTTI 1 ' 3 AND P.G. RANCOITA 1 ; INFN - Istituto Nazionale di Fisica Nucleare, sezione di Milano, Milan, Italy 2 STMicroelectronics, Catania, Italy and University of Milano-Bicocca, Italy

A systematic investigation of radiation effects on a BICMOS technology manufactured by STM has been undertaken. Bipolar transistors were irradiated by neutrons, C, Ar and Kr ions, and recently by electrons. Fast neutrons, as well as other types of particles, produce defects mainly by displacing silicon atoms from their lattice positions to interstitial locations, i.e. generating vacancy-interstitial pairs (the so-called Frenkel pairs). Although imparted doses differ largely, the experimental results indicate that the gain (|3) variation is mostly related to the non-ionizing energy-loss (NIEL) deposition for neutrons, ions and electrons. The variation of the inverse of the gain degradation, A(l/p), is found to be linearly related (as predicted by the Messenger-Spratt equation for neutron irradiations) to the concentrations of the Frenkel pairs generated independently of the kind of incoming particle. For space applications, this linear dependence on the concentration of Frenkel pairs allows to evaluate the total amount of the gain degradation of VLSI components due to the flux of charged particles during the full life of operation of any pay-load. In fact, the total amount of expected Frenkel pairs can be estimated taking into account the isotopic spectra. It has to be point out that in cosmic rays there is relevant flux of electrons and isotopes up to Ni, which are within the range of particles presently investigated.

1. Introduction The flux of Galactic Cosmic Rays (GCRs) is low compared to trapped particles but it is a hazard to spacecraft electronics, because GCRs can penetrate shielding materials [1]. For instance, after a standard aluminum shield of 100 mils (see Sect. 5.4 in [1]) fast energetic charged particles (i.e., above 22.5 MeV for protons and 90 MeV/nucl for iron ions, but only above 1.2-1.5 MeV for electrons) will be able to make radiation damages. In this case, each isotopic element of these penetrating charged particles (see for instance, fluxes and energy distributions of protons, helium, carbon and iron shown in Sect. 23 of [2]) releases a dose and loses an amount of energy by Non Ionizing Energy Loss (NIEL) processes ([3] and references therein), which are non negligible fractions of those of the proton component. As a consequence, even if a 100 mils aluminum shield is taken into account the proton component is expected to 832

833

contribute with an important, but not dominant, fraction of both the dose and the NIEL deposition. The high-energy electron component in GCR's is about 410 % of the dominant proton component [4,5] and is larger than any other isotopic component. In addition is one of the main components of trapped particles in Earth Radiation Belts. 2. Irradiations and Measurements The results obtained by irradiations of BJT's with neutrons, C, Ar and Kr ions have been recently published [6-8]. At present, additional data have been obtained with electrons. The neutron irradiation was performed at the Triga reactor RC:1 of the National Organization of Alternative Energy (ENEA) at Casaccia, Rome [6]. The obtained fluences were in the range of (1.2xl0131.2xl015) n/cm2 and are shown in Table 1. The C ions were made available at GANIL, at two different energies: 12C accelerated at 95 MeV/amu (High Energy, HE), and 13C ions at energy of 11.1 MeV/amu (Medium Energy, ME), while Ar ions at 13.6 MeV/amu and 86Kr ions at 60 MeV/amu. Electrons have been obtained with a mean kinetic energy of 9.1 MeV , at the ISOF Institute in Bologna. Both npn and pnp devices were characterized, using a modular DC source/monitor (Hewlett-Packard HP4142B), controlled by a dedicated workstation. The forward voltage applied to the emitter-base junction (VBE), was in the range 0.2-1.2 V, which allows to measure the value of forward gain for collector currents, Ic, within the range 10"6
C(HE) [ion/cm2! 5.2xl010

C(ME) [ion/cm2] 1.0x10"

Ar rion/cm2] l.OxlO10

Kr rion/cm2l 1.0x10'

e re/cm2l 2.0xl014

1.35xl014

1.0x10"

5.0x10"

5.0xl010

5.0x10'

7.0xl014

6.75xlO

H

5.1x10"

l.OxlO12

1.0x10"

l.OxlO10

2.0xl015

1.35xl0

ls

12

12

l.OxlO

5.0xl0

5.0xl012

l.OxlO13

10

5.0x10

l.OxlO13

3. Irradiations and Measurements The Fast neutrons, as well as other types of particles, produce defects mainly by displacing silicon atoms from their lattice positions to interstitial locations, i.e., generating vacancy-interstitial pair, the so-called Frenkel pairs. This energy-loss process is known as Non-Ionization Energy-Loss (NIEL). Displacements

834

produced in silicon during neutron irradiation have been widely studied ([8] and references therein). In case of incoming neutrons, the concentration of Frenkel pairs (FP), i.e. the number of Frenkel pairs per cm3, can be evaluated by means of the modified Kinchin-Pease formula [9] as FP = Edis 1(2.5 E^, where Ed, about 25 eV, is the energy required to displace a silicon atom from its lattice position to an interstitial location and Edis is the energy density, i.e. the energy per cm3, deposited through atomic displacements by neutrons. It can be computed using the neutron spectral fluence and the damage function [8]. Similarly to an incoming neutron, an incoming ion interacts and displaces silicon atoms from the lattice position. This primary displacement is followed by a small cascading process, in which secondary displacements are induced by the recoil silicon atom. The concentration of the displaced atoms has been calculated [7,8] by means of the SRIM Monte Carlo program [10]. For incoming electrons, the NIEL deposition in silicon has been estimated by interpolating the computed NIEL deposition for Ed = 21 and 30 eV given in [3,11]. The relationship between the variation of the inverse of the gain and the fast neutron fluence is provided by means of the Messenger-Spratt Equation ([12,13] and references therein):

where /? and fi are the common emitter current gain before and after the irradiation, a>r is 2K times the common emitter gain cut-off frequency, K the relevant damage constant which depends also on the base resistivity and <&„ the fast neutron fluence in n/cm2. Eq. (1) can be rewritten [8] as: NIEL A „„ A /1 \ A k « = — FP (2) where A, is independent of the damage function, which is accounted by Edis and, in turn, by FP. To a first approximation, Eq. (2) can be applied to the case of damages induced by electrons as already observed for irradiations with neutrons and C ions in [6] and for irradiation with Ar and Kr in [7,8] namely there is a NIEL scaling of the gain variation of bipolar transistors. 4. Experimental Results In previous papers [10-12], it has already been shown that for the case of small area npn transistors (with the emitter region of 5um x 5um and irradiated with C, Ar and Kr ions, and neutrons) there is a similar gain degradation when the concentration of the created Frenkel pairs (FP), interstitial-vacancy, is the same. This occurs in spite of the large difference in the absorbed dose between neutron and ion irradiations for generating an equal amount of FP concentration. Similarly, it was shown that npn large emitter area transistors (with the emitter region of 50um x 50um), vertical and lateral pnp transistors have their gain

835 degradations depending linearly on the concentration of the created Frenkel pairs. Ic=1 uA, npn 50x50 1

.

•i

•

Ar Medium E

• • • •

CHtghE C Medium E Election Kr High E Fit

•

A A

•

•

*•

• •

•

!

IE-4

,—>.•

IEI3

.....J

. — — -

1EU

1615

1E16

1E17

Frenkel Pairs [cm'3)

lc=1mA, npn 50x50 1—'•* •'"•"1—•"-*-'

.

^

1E-3

v ,

» • A • •

...—,

-

, -

T

•—'

1—

Ar Medium E C High E C Medium E Electron Kr High E Fit

••••"'I

«—»-•-

A A

A

•

• •

Frenkel Pairs [cm'3] Figure I. Linear dependence of the quantity A(l//?) on the concentration of Frenkel pairs for npn large area transistors area transistors with collector current of 1 p.A and 1 mA.

The present investigation was carried out with 9.1 MeV electrons and fluences between 2xlOM and 2x10' e/cm2 (see Table 1). The gain degradation was investigated for collector currents Ic between 1 p.A and 1 mA. A linear dependence of A(\//J) as a function of the concentration of Frenkel pairs was found. For instance in Fig. 1 (2), collected data (including the present data obtained with incoming electrons) are shown for npn large (small) area transistors with Ic = 1 uA and Ic = 1 mA. This confirms that the basic

836 mechanism of the bipolar gain variation is mainly related to silicon displacements, namely to the amount of NIEL deposition. In addition, a preliminary investigation of the time dependence of the gain variation after the irradiation indicate a that the transistors undergo a selfannealing process. IC=1LIA, npn 1

01

T • • • • •

Ai Medium E CHiflhE C Medium E Election K i Hioh E Neutron Fit

"

5x5 1

'

1

'

" ^

• * •

001

•

1E-3

1EJ

' -,, 1EI4

i

1E15

'

1E16

IEt7

Frenkel Pairs [cm'3)

lc=1mA, npn 5x5 v Argon M«diurp E » CHighE * C Medium E • Electron • Kr High E • Neutron

'.>-

— Sf IE-3 t

•'*

Frenkel pairs [cm'3]

Figure 2. Linear dependence of the quantity A(\/ft) on the concentration of Frenkel pairs for npn small area transistors area transistors with collector current of 1 p.A and 1 mA.

837

5. Conclusions The measurements confirm the fact that in bipolar transistors the inverse of gain degrades linearly with respect to the Frenkel pair concentration, for incoming electrons, neutrons, and C, Ar, and Kr ions and is mainly accounted by the NIEL deposition. For space application the known and linear (as observed from experimental data) A(l//?) dependence on the concentration of Frenkel pairs allows the possibility to evaluate the total amount of the gain degradation of VLSI components due to the flux of charged particles during the full life of operation of any pay-load. In fact, the total amount of expected Frenkel pairs can be computed taking into account the isotopic spectra, and subsequently related to the overall gain degradation. It has to be point out that in cosmic rays there is relevant flux of electrons and of massive particles up to Ni nuclei, which are within the range of particles presently investigated. References 1. J. Barth, Short Course on Applying Computer Simulation Tools to Radiation Effects Problems, IEEE Nuclear and Space Radiations Effects Conference, Snowmass Village (1977). 2. K. Hagiwara et al.., Phys. Rev. D, 6, 010001.1-010001.974 (2002). 3. S.R. Messenger et al., IEEE Trans, on Nucl. Sci., 46, 1595-1602 (1999). 4. J. Alcaraz et al., Nucl. Instr. and Meth. in Phys. Res. B, All, 215-226 (2000). 5. J. Alcaraz et al., Phys. Lett. B, 484, 10-22 (2000). 6. A. Colder et al., Nucl. Instr. and Meth. in Phys. Res. B, 179, 397^02 (2001). 7. A. Colder et al., "Effects of Ionizing Radiation on BiCMOS Components for Space Application", Proc. of the European Space Component Conference (Toulose 24-27 September 2002), 377-386 (2002). 8. D. Codegoni et al., Nucl. Instr. and Meth. in Phys. Res. B, 111, 65-76 (2004). 9. M.J. Norgett et al., Nucl. Engin. andDes., 33, 50 (1975). 10. J.F. Ziegler, J.P. Biersack and U. Littmark, The Stopping and Range of Ions in Solids, Pergamon Press, New York (1985). 11. A. Akkerman, J. Barak, M.B. Chadwick, J. Levinson, M. Murat and Y. Lifshitz, Radial Phys. Chem. 62 301-310 (2001). 12.G.C. Messenger, IEEE Trans, on Nucl. Sci, 13, 141-159 (1966). 13.G.C. Messenger, IEEE Trans, on Nucl. Sci., 39, 468 (1992).

R A D I A T I O N - H A R D N E S S S T U D I E S OF H I G H OH~ C O N T E N T QUARTZ F I B R E S I R R A D I A T E D W I T H 24 G E V PROTONS*

JEAN-PIERRE MERLO* Department of Physics and Astronomy, University of Iowa, Iowa City, IA 52242 , USA Jean-Pierre. Merlo Qeern.ch KEREM CANKOCAK Department of Physics, Mugla University, Mugla, 48000, Turkey kerem. cankocakQcern. ch

We investigated the darkening of two high OH" content quartz fibres irradiated with 24 GeV protons at the CERN PS facility IRRAD. The two fibers, 0.6 mm quartz core diameter, one with hard plastic cladding (qp), the other with quartz cladding (qq), are both supplied by the US firm Polymicro Tech. Inc. (PT). These fibres were exposed at more than 1 Grad in 3 weeks. The fibres became opaque below 380 nm, and in the range 580-650 nm. Darkening under irradiation has variation versus dose similar to what we observed with electrons.

1. Introduction For a few years the University of Iowa group, in the frame of the Forward Hadronic calorimeter (HF) of the Compact Muon Solenoid experiment (CMS) at LHC, investigated the radiation hardness of different types of fibers, in association with Turkish Particle Physics laboratories. The results of darkening measurements of quartz fibres (0.3 and 0.4 mm core diameter) irradiated with 500 MeV electrons up to 100 Mrad have been already published 1. *Work supported by the US Depart, of Energy (DE-FG02-91ER 40664) and NSF (NSFINT-98-20258) and the Scientific and Technical Research Council of Turkey, TUBITAK, (TBAG-1591). ^Presented by Erhan Gfilmez, Physics Dept. Bogazici University, Istanbul, Turkey on behalf of the authors. 838

839

To complete our knowledge of radiation damage of the HF quartz fibres (qp 0.6 mm core diameter) and to investigate the possible use of quartz fibers at SLHC we initiated a high level irradiation with 24 GeV protons at CERN. About 4.6 1016protons/cm2 corresponding to a dose of 1.2 Grad (i.e. about 12 years of HF operation in the hottest tower) have been sent onto 1.20 m of quartz fibers. The fibre radiation damage induced by protons exhibits the same well known behaviour as with electrons : high light attenuation below 380 nm and in the band 550-680 nm. Moderate attenuation in the band 400-520 nm, no attenuation above 700 nm. With a constant dose rate, the damage has an exponential variation versus time (Fig. 2b), fast in the first hours and slow after. Above 0.5 Grad the radiation damage is no more recoverable in the range 580-650 nm and below 380 nm. 2. 2.1.

Experimental set-up and data acquisition Experimental

set-up

The fibres were installed, with their light support, on a remote controlled table (Fig. 1) of the CERN facility IRRAD 2 .

24 GeV protons

Computer

V

J

3 loops of qq + qp fibers

Figure 1. Experimental set-up

Three loops of qq and qp fibers (40 cm long) are wrapped together and placed along the beam with a 2.5% slope. The total irradiated length (L) is about 1.2 m. No material was inserted in front of the fibers. The 24 GeV proton beam at IRRAD was delivered as 3-4 sub-cycles in a (19.6 ± 3.0)s long supercycle. This beam delivers 10np/cm2 /burst in a spot of 2x2 cm 2 . Some of these parameters were changed during the

840

irradiation. The number of protons delivered was recorded and the dose calibration was done using a thin Al sample. We checked regularly the proton beam stability in size and time but it was not as stable as the LIL electron beam 1. The optical setup uses Ocean Optics components. The two-channels spectrometer (SD 200) is the same as in electron measurements 1. The deuterium halogen light source (d+h) generates a wide range spectrum (200-1000 nm). The spectrometer is sensitive in the range 350-850 nm only. The attenuation measurements were performed in-situ, the irradiated fibres being linked to the spectrometer and source with 25 m long fibers. The light is injected in a Y fibre splitter and then in the qq and qp fibres. 2.2. Data

acquisition

The time interval between two successive light injections and data acquisition corresponds to one cycle (19.6 s) at the beginning of irradiation to 40 or 80 cycles (13 or 26 min) at the end, according to the damage variation versus time. The two SD200 channels were read and registered at the same time using the Ocean Optics ADCIOOO card. A PC Dell registers the spectra data as well as the time and the number of protons. The intensity of the light source (d+h) was quite low for the SD2000 and we used an integration time of 1 sec inducing a noticeable background (Fig.2a). In the range of interest (350-750 nm wavelength, dose < 1 Grad) the signal strength in the SD2000 was significant. The qp channel was less sensitive than the qq one. 3. Analysis and results As the PS cycling was changing, some runs included Cherenkov light generated by protons. These runs were discarded. Then, the data are histogrammed in 10 nm wavelength bins. Runs are arranged in groups covering a few minutes of irradiation at the beginning and hours after 100 Mrad dose. At a given dose D, the starting point, c(A, D), of the raw signal distribution 5(A), in each bin are determined and the distribution [S — c}(\, £>), is summed over all the wavelengths and then fitted to a Moyal (Landau) distribution. The maximum of this distribution d(D), characterises the spectrum. The same is done for the background to get the parameter b independent of dose and wavelength.

^

(a)

4

tj

(*

t«

U

!

^

3

+

(b)

Figure 2. Evolution of the spectra versus increasing dose (a), and of the ratio of spectra with and without irradiation at 455 nm (b) for the qq fibre.

We assume that /(A, D) = c(A, D) + d(D) — b, and then calculate the ratios/(A,.D)/J(A,0). The light attenuation in the fiber, is well represented by the function A(\,D)=a(\)[D/Ds}PW a and /? parameters for qq and qp fibres are then determined by fitting the ratios as a function of wavelength and dose. 7(A,£))/7(A,0) = ezp[-(L/4.343)a(\)(D/D.)M% Choosing a scale factor D g = 100 Mrad, a is the attenuation at 100 Mrad, in dB/m , L being expressed in m. The results are shown in Figure 3: 4. S u m m a r y The radiation damage due to protons are in agreement with our published results on the radiation damage due to electrons. This is significant because the two types of irradiation are very different: - 24 GeV protons compared to 0.5 GeV electrons with very different mode of interaction - Higher dose: 1 Grad with protons compared 0.1 Grad - Different set-up and different fiber sizes (all supplied by Polymicro) As with electrons up to 0.8 Grad we don't observed a significant difference in attenuations versus dose between qq and qp fibers (Fig. 3a). Above 0.5 Grad there is no recovery of damage below 380nm and in

Figure 3. a (a) and /3 (b) parameters for qq and qp fibres versus wavelength, where a is the attenuation in d B / m at 100 Mrad.

the range of 580-650 nm, the quartz becomes opaque. The recovery was watched over 4 months. The decrease of signal (Fig. 2a) in fibres is very fast at the beginning: 30% loss after 200 Mrad at 455 nm, i.e., after about 2 months of operation of a detector at psedo rapidity r\ = 5 (tower 12 of the CMS-HF). The complete results of this analysis up to 1.2 Grad with a recovery study in qq fiber will be published soon. Acknowledgments We thank Maurice Glaser responsible of the IRRAD facility who helped us in the installation of our set-up and calibrated the dose received as well as the PS operators and technicians providing the proton beam, the SPS-PS coordinator and the CMS test beam coordinator Emmanuel Tsesmelis. We would also like to thank Graduate College of the University of Iowa for their continuing support for this research. References 1. I.Dumanoglu et al., Nucl. Insi. and Meth. in Phys. Res. A490 (2002) 444-455. 2. M. Glaser et al., Nucl. Inst, and Meth. in Phys. Res. A426 (1999) 72-77.

DEVELOPMENT OF RADIATION HARD SILICON DETECTORS: THE SMART PROJECT A. MESSINEO1, L. BORRELLO, G. SEGNERI, D. SENTENAC INFNsez. di Pisa and Universita di Pisa, Italy D. CREANZA, M. DEPALMA, N.MANNA, V. RADICCI INFNsez. di Bari and Universita di Bari, Italy E. BORCHI, M. BRUZZI, E. FOCARDI, A. MACCHIOLO, D. MENICHELLI, M. SCARINGELLA, C. TOSI INFNsez. di Firenze and Universita di Firenze, Italy D. BISELLO, A. CANDELORI, V. KHOMENKOV INFNsez. di Padova and Universita di Padova, Italy M. PETASECCA, G. U. PIGNATEL INFNsez. di Perugina and Universita di Perugia, Italy M. BOSCARDIN, G.-F. DALLA BETTA, C. PIEMONTE, S. RONCHIN, N. ZORZI ITC-IRST Trento, Italy The research actitvity of the SMART project, a collaboration of Italian research institutes funded by the I.N.F.N., has been focused on the development of radiation hard silicon position sensitive detectors for the CERN Large Hadron Collider luminosity upgrade. Electrical characterization of pad and micro-strip devices as well as study of microscopic defects on the bulk material have been carried out on silicon 4" wafers of n- and p-type, grown with Standard Float Zone (SFz), high resistivity Magnetic Czochralski (MCz) and epitaxial (EPI) techniques. Manufactured devices have been irradiated with 24 GeV/c and 26 MeV protons up to ~ 310 15 cm'2 1 MeV neutrons eq. (n^/cm2) and with reactor neutrons up to ~810 15 n^cm 2 . Preliminary results of measurements before and after irradiations as well as material radiation hardness are shown and discussed.

1

Presenting author: e-mail address alberto.messineo(a!pi,infn,it

843

844

1. Introduction Tracker detectors based on silicon devices are now under construction for high energy physics experiments at the Large Hadron Collider (LHC, CERN). They have been designed to work in an hostile radiation environment, to withstand a fast hadron fluence up to 310 15 cm"2 at the minimum instrumented distance from the collision point (r;=4cm). The option for increasing the LHC luminosity up to 1035cm"V is already envisaged in order to extend the discovery potential of the machine (SLHC) [1]. To preserve the physics potential of SLHC the tracker detectors should maintain their performance up to the largest integrated luminosity. In such experimental conditions, at the innermost radius ri, the fast hadron fluence will reach values as high as ~1.61016cm"2 for five years of machine operation. New radiation hard material and detector design must be defined to instrument the region near to the collision point. This paper deals with the activity of a network of italian groups working in the framework of the CERN RD50 [2] collaboration. This project, named S.M.A.R.T. (Structures and Materials for Advanced Radiation hard Trackers), is funded by the INFN 5th commission to develop new radiation hard materials suitable for micro-strip detector design and production with the presently available planar technologies. Material engineering research has already proven that oxygen-enriched Float Zone (DOFZ) silicon bulk has increased radiation tolerance with respect to gamma and proton irradiations [3]. Nowadays, MCz Si [4], more enriched in oxygen (-5-910 17 [0]/cm3) than DOFZ (1-210 17 [0]/cm3), can be grown with a resistivity suitable for micro-strip detector production. The use of n-on-p devices is also appealing to increase the charge collection efficiency due to the higher electron mobility and to the expected non-appearance of bulk typeinversion at high fluences of irradiation. Furthermore an improved growth technique of thick (-100-150 \im) epitaxial Si layer on Cz substrates has recently shown [5] that this material has a noticeable potential as ultra-radiation hard material. The activity of the SMART collaboration includes: microscopic material defect study, material characterization with pad diodes, measurement on minisensors and study of their performance assembled in detector units, simulation of radiation damage on silicon. In this paper we describe the results achieved with new test structures and micro-strip detectors processed with these materials and their radiation hardness performance after heavy irradiation tests.

845

2. Materials, processing and irradiation Bare wafer materials with both n- and p-type bulk doping used in this study belong to the common RD50 procurement. MCz 4 " wafers were produced by Okmetic Ltd (Vantaa, Finland), few n-type epitaxial wafers (resistivity of 50Qcm and epitaxial layer thickness 50um) grown at ITME, (Varsaw, Poland) have been also used. SFz silicon wafers were also processed for comparison. Wafer layout includes 66 test structures and 10 micro-strip sensors of different geometries. Processing has been performed by ITC-IRST (Trento, Italy) in two successive runs. First run for n-type (RUN-I) and second one for ptype (RUN-II) materials have been processed with the same mask-set designed to be used with all the different silicon materials. For RUN-II the uniform pspray technique has been used to increase n+ implants isolation with two different implantation doses, namely 3-1012 cm"2 (low p-spray) or 510 12 cm"2 (high p-spray). Details can be found on references [6,7]. Devices have been irradiated with different particle beams. A first proton irradiation campaign was carried out at the CERN-SPS with 24 GeV/c protons with fluences up to 3-1015 neq/cm2. A second irradiation was performed at the Compact Cyclotron Forshungszentrum in Karlsruhe with 26 MeV protons with fluences up to 2-1015 neq/cm2. The 50um thick epitaxial devices have been irradiated with 24 GeV/c protons up to 5-1015 neq/cm2 and also with fast neutron in the Ljubljana nuclear reactor up to 71015neq/cm2.

3. Pre-irradiation measurements MCz and SFz diodes and mini-sensors were characterized following the procedures defined within the RD50 Collaboration. Pre-irradiation results of ntype devices (SFz,MCz) showed a uniform resistivity and current density within the same wafer and good breakdown performance for all micro-strip sensors design [8]. MCz p-type wafers have shown local depletion voltage (Vdepi) variations due to thermal donors. Moreover low breakdown voltages on microstrip sensors processed in both SFz and MCz material with a high dose of the pspray has been detected [9]. In order to quantify the contribution to the noise of a micro-strip device assembled in a tracking detector unit, the pre-irradiation characterization includes the measurement of the inter-strip capacitance. Measured values show typical behavior as a function of the bias voltage with a plateau in overdepletion condition ranging from 0.5 to 1.2 pF/cm as expected by the devices geometry. Micro-strip sensors have been assembled in detector units with the front-end electronics designed for the CMS silicon strip tracker and read-out with the standard DAQ system used for this experiment [10]. Detector

846

performance has been studied with the analysis of the collected charge spectrum by electrons from 106 Ru source, which are minimum ionizing particles. Preliminary measurement show that MCz and SFz n-type detectors have similar performances: the average value of the charge collected on a 50um pitch detector is for both materials =18.8 + 0.3 (ADC Channels), suggesting that collection efficiency is also similar and values of S/N ~19 have been measured for both detectors. 4. Post-irradiation results All devices, diodes and mini sensors, were fully characterized after irradiation: IV and CV measurements were performed at 20°C, 0°C and -3°C keeping the first guard ring connected at ground. Annealing of radiation damage was performed maintaining the devices at room temperature or heating them up to a temperature of 60°C or 80°C for known time intervals [8,9]. Microscopic investigation by Thermally Stimulated Currents (TSC) measurements has been performed on a wide temperature range, starting from 4.2K to detect the shallow levels produced by traditional dopants, Thermal Donors (TDs) and other defects which may be present in oxygen enriched Si [11]. The current related damage has been evaluated for all radiation sources and on nand p-type materials. Measurements of the leakage current density increase rate (a) [8] have shown good agreement with values expected if the NIEL hypothesis is taken into account [12]. The MCz and SFz n-type Si micro-strip detectors have shown good performances before and during the annealing treatments, with breakdown voltages well above their Vdepi value and leakage currents proportional to the received fluence as expected (see fig.l). The p-type detectors show an improved performance after irradiation than before beam exposure: in the entire fluence range the breakdown voltages of the detectors with a low p-spray dose are fully comparable with the n-type sensors and are in excess of 600V, although the detectors with a high dose p-spray recover completely only at irradiation fluences around 610' 4 1-MeV neq/cm2 [9].

Bias Voltage Figure -1 IV curves of n-type micro-strip sensors irradiated with 26 MeV protons from 610 neq/cm2 up to 2-10 neq/cm2

Figure-2 Bias voltage annealing curves for SFZ (full line) MCz (dashed line) n- and p- type Si diodes irradiated with 26 MeV protons

847

An example of depletion voltage annealing measurement is shown in fig.2 for SFZ and MCz diodes of both n- and p-type bulks irradiated with 26 MeV protons at two different fluences 6-1014 neq/cm2and 8-1014 neq/cm2. For both bulk types an improved reverse annealing behavior of MCz Si is observed with a clear saturation beyond 200 minutes at 80°C, while the Vdepi values of SFZ materials still increase. The reduced reverse annealing growth of MCz devices would simplify damage recovery in experimental operational conditions. Diodes made with n-type SFZ material were already type inverted thus the measured V dep i, as well as effective doping Neff (fig. 3), values increase in the whole fluence range [6-1013 n^cm 2 , 3-1015 neq/cm2]. On the contrary the Vdcpl values of the n-type MCz wafers show a minimum centered at 1.86-1014 neq/cm2 with a small spread due to the different initial resistivity. At the minimum the full depletion voltage is significantly different from zero as well as the effective doping concentration. This could be explained considering that no space charge sign inversion (SCSI) occurred although the annealing curves of Vdcpi values for devices irradiated further than the fluence of minimum shows a behavior similar to the one of an inverted SFZ device with the beneficial annealing that leads toward a minimum [13]. Epitaxial devices show similar results for proton irradiation and a less pronounced effect for neutron irradiation [14].

Flences (10M4 nart>-2)

Figure-3 Effective doping concentration on SFZ (dashed lines) and MCz (full lines) Si as a function of the fluence

According to our measurements of current transients at constant temperature, and to preliminary Transient Current Technique (TCT) analysis n-MCz Si does not exhibit SCSI in the range 1014-1015 neq/cm2 [15]. The occurring of a minimum of the depletion voltage without SCSI could be explained in terms of the double junction effect. On the contrary a SFZ diode irradiated with 24 GeV/c protons at a fluence of 2.04-1014 neq/cm2 exhibited a clear SCSI effect due to the charge carrier emission at the so called "I defect", demonstrating that at this fluence the SFZ under study is type-inverted at room temperature. Furthermore our TSC analysis evidences the formation of a shallow donor, which could partially compensate the radiation induced deep acceptors related

848

to divacancy. This shallow donor appears to be introduced in MCz with a generation rate much higher then in Fz in agreement with the smaller introductory rate of deep acceptors observed in MCz [8] and with the experimental evidence that at 1015 neq/cm2n-MCz Si is still not type inverted. The MCz and the SFz micro-strip detectors have comparable performances after irradiation in terms of inter-strip capacitance (Cint). Concerning the n-type detectors Cint is not varying significantly with the fluence and the postirradiation value is in the same range of the pre-irradiation one: 1.2-1.7 pF/cm for the detectors with a strip pitch of 50 um and 0.5-1 pF/cm for the 100 (im pitch. On the contrary, in p-type detectors Cint decreases with the fluence down to the n-type values at around 410 14 neq/cm2. Fig.4 shows the Cint values: SFz devices follow the classical <111> behavior [10], and MCz n-type ones reach their plateau value in advance with respect to p-type material devices for irradiations at fluences lower than 8-1014 n^/cm2 [13]. The simulation activity of radiation damage in Silicon detectors carried out within SMART achieved significant results too. The tool used to model radiation hardness is based on ISE-TCAD with discrete time and spatial solutions to transport equations. The damage modeling is performed by taking into account activation energies, cross sections for majority and minority carriers and trap concentrations of the main defects observed by TSC and deep level transient spectroscopy (DLTS). Up to now, in p-type FZ Si, a three level model has been considered, with two acceptors at 0.42eV and 0.46eV plus a donor at 0.36eV as in. Simulated results for CCE after fast hadrons fluence of 1015neq/cm2 well reproduce experimental data [16]. This simulation predicts that the charge collected on a 300um thick p-type device due to an incident mip at the highest fluences foreseen at SLHC (see fig.5) show that the expected charge from a mip will be below 5000 electrons. ............... 1

4KjH!3*'.il

7.0E-12 i 6.0E-12 •

;

jf/BfZjtigpti

: A 'c"*wv

<1

*

5.0E-12 4.0E-12

O

1

XJ*»W

3.0E-12 •

•

2.0E-12

lt

*"-*1'

....... ...... 300

400

I ' ©

'

1.0E-12

'

....... .

500

VUas(V)

Figure -4 CU: SFZ (brown), MCz p-type (red), MCz n-type (green and blue); devices irradiated at 410 14 n eq /cm 2

Figure -5 Simulation: Collected charge b; a p-type device from a m.i.p. particle as i function of the irradiation fluence

849 5. Conclusions This work has been focused on the study of radiation hardness of devices processed on new Si materials, in particular high resistivity Czochralski (n- and p-type) and epitaxial Si, by the SMART INFN project. Micro-strip devices from MCz n-type wafer assembled in a LHC-like detector configuration showed good performance in terms of CCE and S/N before irradiation. Preliminary results from TCT show that MCz n-type wafers do not show SCSI up to HO15 rieq/cm2 and that shallow donor levels have a higher concentration after irradiation in MCz respect to SFZ material. The radiation hardness evaluation carried out in this work showed good performance of the SMART devices and identifies them as very promising for processing of micro-strip detectors used to instrument the future SLHC tracker.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.

F.Gianotti et al., Eur. Phys. J. C 39, 293 (2005) RD50 collaboration: http://rd50.web.cern.ch/rd50/ G. Lindstroem et al., Nucl. Instr. Meth. A 466, 308 (2001) V.Savolainen et al., J. Crystal Growth 243, 2 (2002). G. Kramberger et al, Nucl. Instrum. Methods, A515, 665 (2003) M. Bruzzi et al. "Process and first characterisation of detectors made with radiation-hard Si materials", in press on Nucl. Instr. and Meth. A. C.Piemonte Preliminary electrical characterization of n-on-p devices fabricated at ITC-irst presented at the 5th RD50 Workshop G.Segneri et al., "Radiation hardness of high resistivity n- and p-type magnetic Czochralski silicon", proceedings of PSD07 conference A.Macchiolo et al., "Radiation hardness of high resistivity n- and p-type magnetic Czochralski silicon", proceedings of PSD07 conference CMS Tracker Technical Design Report, CERN/LHCC 98-6 M. Bruzzi et al., Nucl. Instr. and Meth. A 552, 20 (2005) M. Moll et al., Nucl. Instr. and Meth. A 426 (1999) 87. V.Radicci et al., proceedings of RD05 conference A. Candelori "Semiconductor materials and detectors for future very high luminosity colliders" presented at NSREC 2005 M. Scaringella et al.,JVwc/. Instr. and Meth. A submitted for publication M. Petasecca et al., Nucl. Instrum. Methods, A546, 291 (2005)

SCINTILLATOR AND PHOSPHOR MATERIALS: LATEST DEVELOPMENTS AND APPLICATIONS E. MIHOKOVA, M. NIKL Institute of Physics, ASCR, Cukrovarnickd 162 53 Prague 6, Czech Republic

10,

S. BACCARO, A. CECILIA ENEA FIS-ION, 000 60 S. Maria di Galeria, 162 53 Rome, Italy We review recent research in the field of phosphor and scintillator materials used for Xray detection. We describe parameters and characteristics crucial for performance of the materials in applications. Description of materials currently used or intensely studied is provided. We show recent results of radiation damage study of Y3Al5O12.Fr as a novel promising scintillator. Principal applications of scintillators and phosphors are briefly summarized.

1. Introduction Since the discovery of X-rays by Wilhelm Conrad Roentgen in 1895 there was a need to find materials efficient in converting X-rays to visible light. Simple photographic film was soon replaced by CaW0 4 powder and ZnS-based powders that are used until today. Scintillation detector consists of a scintillator (phosphor) material followed by an optional relay element and a photodetector. Wide band-gap materials are employed to convert X-rays to UV/visible photons. Entire scintillation conversion can be divided into three processes: conversion, transport and luminescence. Conversion process involves an interaction of high-energy photon with the material lattice. As a result many electrons and holes are created and thermalized in the conduction and valence bands. During the transport stage these charge carriers migrate through the lattice. Their capture at trapping levels within the material forbidden gap delays migration. Moreover, energy losses due to nonradiative recombination may occur. Final, luminescence stage involves radiative recombination of the electron and hole trapped at the luminescence center.

850

851

There is a wide variety of materials used and investigated for X-ray detection. Significant amount of literature in the field can be found (see e.g. books[l,2,3] or recent review papers [4,5]). This work is focused on one class of materials, namely crystalline wide band gap solid state materials. Principal characteristics of scintillators and phosphors, important materials and their applications are summarized. 2. Material characteristics Historically, mainly due to different demands on materials used in different applications where high energy detection is involved, scintillating materials are divided into phosphors and scintillators. Phosphors are used in integrating (steady-state) mode of detection. Scintillators are employed in X- (or y) ray photon counting regime. Nowadays some of materials are used in both detection modes. Principal characteristics considered for phosphors/scintillators are following: • Overall efficiency of X-ray-to-light conversion / light yield • X-ray stopping power • Luminescence decay time and afterglow (persistence) / scintillation response - decay time • Spectral matching between the phosphor/scintillator emission spectrum and photo-detector • Chemical stability and radiation resistance • Linearity of light response with the incident X-ray dose and intensity / linearity of light response with the incident X(y)-ray photon energy energy resolution • Spatial resolution across the screen (only phosphors) 3. Materials There are several classes of inorganic phosphors and scintillators that appear in various applications. They are namely powders, optical ceramics and single crystals. 3.1. Powders Until seventies CaW0 4 phosphor was dominantly used. Then more efficient Tb doped oxysulfides (R 2 0 2 S, R=Y, La, Gd) appeared. Various dopants were

852

studied [6], the most important appear doping by Tb3+ and Pr3+. Other rare earth doped efficient phosphors are oxyhalides LnOX (Ln=Y,La,Gd: X=C1, Br) as well as binary oxides Ln 2 0 3 (Ln=Y, Gd, Lu) dominantly doped by Eu3+. LaOBr doped by Tb3+ and Tm3+ [7] have the dominant emission line in the blue spectral region. Y2O3 is well-known red emitting phosphor. Lu 2 0 3 host offers exceptionally high density about 9.4g/cm3 which is together with LuTa04-based phosphors (9.75 g/cm3) the densest phosphor available. Sr 2 Ce0 4 is an interesting phosphor due to its not very common luminescence mechanism - charge transfer luminescence of Ce4+-based complex. Phosphors containing hafnium are not only materials offering high density, but unlike Lu compounds (due to natural abundance of 176Lu radioisotope) they have practically no background radioactivity. Recently studied SrHf03 doped by Ce3+ or Tm3+ can provide an efficient phosphor emitting in the violet-blue spectral region. Other powders to be mentioned are compositions AB 2 0 4 (A=Sr, Zn, Ca; B=Ga, In, Y) mostly doped by Mn2+, Cr3+, Eu3+, Tb3+, ternary sulfides AB2S4 (A=Ca, Sr, Ba; B=Ga, Y, Al) doped by Mn2+, Eu3+, Tb3+, or Y3A15012 (YAG) mostly doped by Ce3+, Eu3+, Tb3+. Summary of characteristics of selected materials is given in Table 1. Table 1. Summary of characteristics of selected phosphor materials [4-6,8-10]. Phosphor ZnS:Ag

Density [8/cm5] 3.9

Decay time [ns] -1000

6.1

6xl0 3 6xl0 5

CaW0 4 Gd202S:Tb

7.3

Gd202S:Pr,Ce,F LaOBr:Tb

7.3 6.3

4000

YTa04:Nb

7.5

-2000

Lu2C>3:Eu SrHf03:Ce

9.4 7.7

~10

6

6

~10 40

Efficiency [%1 17-20 5

Emis. max. Afterglow [nm] Very high 450 Very low 420

13-16

540

Very low

8-10 19-20

490 425

Very low low

11

410

low

611 390

medium

-8 2-4

Not reported

3.2. Optical ceramics These transparent or translucent materials consist of tight aggregating, randomly oriented, crystallite micro-grains. They provide bulky elements when single crystals cannot be prepared or when ceramics show superior properties. Some materials reached the highest degree of perfection for their application in the field of solid state lasers (e.g. YAG:Nd, Y 2 0 3 :Nd). However, application demands on especially grain size and interface layer thickness in the case of scintillators are higher than those of laser optical ceramics due to introduction of

853

trapping sites that can severely deteriorate a scintillator performance. Among materials reported there are (Y,Gd)202:Eu; Gd202S:Pr,Ce,F; Gd3Ga5012:Cr,Ce and Eu3+ or Tb3+-doped Lu 2 0 3 . Recently YAG:Ce optical ceramics from Baikowski, Japan Ltd. has been compared with an industrial standard quality single crystal produced by CRYTUR, Ltd., Czech Republic [11]. The ceramics shows faster decay (60-65 ns), explained by higher Ce concentration achieved (5000 ppm vs 1000 ppm) and the absence of YA) antisite defects due to lower preparation temperature. 3.3. Single crystals First single crystal scintillators used in the late forties were NaLTl and CsLTl. Later Bi4Ge3Oi2 (BGO) became a widespread scintillator used as a "standard" for comparison with new materials. Extremely rich class of materials investigated within last decades is represented by Ce-doped materials [12,13], due to the fast decay time of 5d-4f radiative transition of Ce3+ (20-60 ns) and its high quantum efficiency (close to 1) at room temperature. Search for heavy, fast and efficient materials is focused on Lu-compounds. Lu2Si05:Ce (LSO:Ce) and (LuxY!.x)2Si05:Ce were optimized, Gd2Si05:Ce (GSO.Ce) was improved. Certain effort was devoted to the development of LuA103:Ce. Its growth is difficult due to instability of perovskite phase and an easy switch to the garnet phase. Increased stability of growth was achieved for mixed (Y-Lu) A103 systems. Lu3Al50i2:Ce crystals are already available from the industrial production, their figure-of-merit is comparable to GSO:Ce. Despite attractive characteristics of Lu compounds their widespread applications might be limited by high cost of Lu 2 0 3 , demanding technology and increased background (mentioned above). Medium density halide scintillators recently attracted attention due to high light yield and very good energy resolution. Ce-doped LaCl3, LaBr3, Lul3 were reported as efficient scintillators, the last one having a light yield 76000 ph/MeV [14]. Single crystals of ternary compounds K2LaX5:Ce (X=C1, Br, I) were also studied. K2LaI5:Ce appears especially promising scintillator with density comparable to that of Csl, but much faster scintillation response. The main drawback of halide scintillators is their high hygroscopicity, comparable or even higher than that of NaLTl. During the last decade PbW0 4 single crystal was intensely studied (see review [15]) as selected scintillator for high energy physics detectors. To enable the use of this heavy, fast and relatively cheap scintillator in applications outside the field of high energy physics there were attempts to increase its light yield by multiple doping and/or annealing. Double doping (Mo, A3+), A=Y, La, later

854

accompanied by codoping by Nb5+ and other doping combinations were tested. Despite all attempts the light yield in any of the samples reported so far did not exceed 10% of BGO. Summary of characteristics of selected materials is given in Table 2. Table 1. Summary of characteristics of selected single crystal scintillators [3,5,14,16,17].

Crystal

Density [g/cm3]

CsI:Tl

4.51

NaI:H

3.67

LaBr3:Ce

5.3

LuI3:Ce

5.6 4.4

Light yield Ph/MeV 66000 41000 61000 76000

K2LaI5:Ce BaF2 (only crosslumin.) Bi4Ge30i2

4.88

1500

7.1

8600

PbWO,

8.28

300

CdW0 4 Yal03:Ce LuA103:Ce Y3Al50i2:Ce

7.9

20000 21000

5.6 8.34

55000

Dominant scint. decay time [ns| 800 230 35 50 24 0.6-0.8 300 2-3 5000 20-30

12000 24000

18 90-120

Lu3Al50i2:Ce

4.56 6.67

12500

55

Gd2Si05:Ce

6.7

8000

60

Lu2Si05:Ce

7.4

26000

30

Emis. AE/E at 662 max. [nm] keV [%] 550 410

6.6 5.6

358 472-535

2.9

420

4.5

180-220 480 410 495 360 365 550 530 420 390

4.7 7.7 9.0 30-40 6.8 4.6 -15 7.3 11 7.8 7.9

4. Radiation damage study of YAG:Pr novel scintillator Analogous to fast 5d-4f luminescence of Ce can be obtained also from Pr ion in suitable crystal hosts. Comparing to Ce the luminescence is high energy shifted and even faster. Recently a study of YAG:Pr as a potential novel scintillator material has been performed. Its principal emission band is peaking at 318 nm. We studied a set of samples with various concentrations of Pr. The radiation damage was characterized by radiation induced absorption coefficient p=(\/l)]n(To/T), where /, r 0 and Tare the length of the sample, initial transmission and transmission after irradiation. Samples were irradiated by 60Co y source. After each irradiation and measurements cycle the samples were bleached by annealing at 700°C for

855

approximately 5 hours. Results of the radiation damage study are displayed in Fig. 1. They show that with increasing concentration of Pr the radiation damage reduces. LY measurements show also a tendency of reduction with increasing Pr concentration. Therefore for potential use of this material it is necessary to find a compromise for reasonable value of both characteristics. Comparing to YAG:Ce scintillator an advantage of YAG.Pr is its much faster scintillation decay (16 ns at room temperature) and absence of slow components in the decay. Therefore it appears as potentially interesting material.

•"•

120 % 80

o undoped n 0.16% Pr • 0.33% Pr <5>0.50% Pr • 0.65% Pr

0 300 400 500 600 700 800 Wavelength (nm)

.(b) .

•

•

140

i

o undoped 0.16% Pr • 0.33% Pr * 0.5% Pr • 0.65% Pr

D

- 100

D

-

•

-

1?

m^

...2 10

• -60 <0 9

--20 .

100 Dose (Gy)

Figure 1. Radiation induced absorption coefficient og YAG:Pr samples (a) after the dose of 250 Gy, (b) dose dependence at 318 nm (emission maximum). Concentrations in legends refer to a content of Pr203 in the melt.

5. Applications The main applications using X-rays are medical imaging, general flaw detection, high resolution 2D-imaging and radio astronomy. In medical imaging the applications can be divided into static and dynamic imaging. The most frequent applications including static imaging are general and dental radiography as well as mammography. Classical way of imaging where photographic film is coupled to Gd202S:Tb phosphor screen starts gradually being replaced by the modern flat panel detectors. In dynamical imaging the main applications are fluoroscopy and Computer Tomography.

856

Acknowledgments Projects no. AV0Z10100521 and GA AV A1010305 are gratefully acknowledged.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17.

G. Blasse and B. C. Grabmaier, Luminescent materials. Springer, Berlin (1994). G. F. Knoll, Radiation detection and Measurement. John Wiley&Sons, New York (2000). P. A. Rodnyi, Physical processes in inorganic scintillators. CRC, New York (1997). M. J. Weber, J. Lumin. 100, 35 (2002). C. W. van Eijk, Phys. Med. Biol. 47, R85 (2002). J. A. Shepherd, S. M. Gruner, M. W. Tate, and M. Tecotzky, Opt. Eng. 36, 3212(1997). L. H. Brixner, Material Chem. Phys. 16,253 (1987). S. Shionoya and W. M. Yen, Phosphor Handbok. CRC Press New York (1998). W. M. Yen and M. J. Weber, Inorganic Phosphors: Compositions, Preparation and Optical Properties. CRC Press New York (2004). A. Lempicki et al., Nucl. Instr. Meth. Phys. Res. A 488, 579 (2002). M. Nikl, unpublished results. C. W. E. van Eijk, J. Andriessen, P. Dorenbos, and R. Visser , Nucl. Instr. Meth. Phys. Res. A 348, 546 (1994). C. W. E. van Eijk, Nucl. Instr. Meth. Phys. Res. A 460,1 (2001). M. D. Birowosuto et al., IEEE Trans. Nucl.Sci., 52, 1114 (2005). M. ~NM,phys. stat. sol. (a) 178, 595 (2000). J. A. Mares et al., Rad. Meas. 38,353 (2004). E. V. D. van Loef et al., Nucl. Instr. Meth. Phys. Res. A 537, 232 (2005).

CRYSTALS FOR H I G H - E N E R G Y P H Y S I C S CALORIMETERS IN E X T R E M E E N V I R O N M E N T S

F. NESSI-TEDALDI Swiss Federal Institute of Technology (ETH) CH-8093 Zurich, Switzerland E-mail: Prancesca.Nessi- Tedaldi@cern. ch Scintillating crystals are used for calorimetry in several high-energy physics experiments. For many of them, performance has to be ensured in very difficult operating conditions, like a high radiation environment and large particle fluxes, which place constraints on response and readout time. An overview is presented of the knowledge reached up to date, and of the newest achievements in the field, with particular attention given to the performance of Lead Tungstate crystals exposed to large particle fluxes.

1. Introduction This report addresses the performance of scintillating crystals used for highenergy physics calorimetry, when operation implies high radiation levels and intense particle fluxes. The effect of high levels of ionising radiation on crystals has been studied in depth and reported upon by many authors, as crystals were used e.g. in e+e~ collider experiments, and their growth parameters were optimised for best performance in such environments. They are briefly summarised herein. Hadron collider detectors today share the same concern, but add to it the need to ensure adequate performance when crystals are exposed to large particle fluxes. Such running conditions are namely expected in several experiments under construction or designed. Some new results are thus presented here, together with a fresh look at existing, older ones, to provide, as far as possible, a complete picture. 2. Performance under high ionising radiation levels Ionising radiation is known to produce absorption bands through formation of colour centres, which reduce the Light Transmission (LT) and thus the Light Output (LO), due to oxygen contamination in alkali halides like 857

858

BaF2 and Csl, and to oxygen vacancies and impurities in oxides like BGO and PbW04 *. Phosphorescence or afterglow appear sometimes 2 , which increase the noise levels in the detected light signal, possibly worsening the energy resolution (in a negligible way for PbWC"4 in LHC experiments 3 ), while the scintillation mechanism is generally not damaged. Recovery of damage at room temperature can occur depending on crystal type and growth parameters, giving rise to a dose-rate dependence of damage equilibrium levels1'4 and to a recovery speed dependent on the depth of traps. That ionising radiation only affects LT, means the damage can be monitored through light injection and corrected for, as it is done in the CMS Electromagnetic Calorimeter (ECAL) 5 .

3. Performance in large particle fluxes The way hadron fluxes affect crystals has become a crucial question while detectors making use of this calorimetry technique are being constructed. In particular, it had to be ascertained whether such fluxes cause a specific, possibly cumulative damage, and if so, what its quantitative importance is, whether it only affects LT or also the scintillation mechanism. Extensive studies have been recently performed on PbWC-4 at IHEP Protvino 6 and, for the CMS ECAL, at CERN and ETH-Ziirich7. Their main results are quoted and discussed herein. Crystal tests at Protvino were using e~ and 7r beams and 7 sources up to a few krad at 1 to 60 rad/h at one end, and a very intense mixed beam of charged hadrons, neutrons and 7 up to 3 Mr ad at 1 krad/h and 100 krad/h equivalent fluxes at the other end. In individual e~, n and 7 irradiations, the signal loss behaviour is found to be qualitatively similar between electrons and pions, and the damage appears to reach equilibrium at a dose-rate dependent level. Furthermore, no indication of damage to the scintillation mechanism from TV irradiation is found8. A concern remains however, that an additional, specific, possibly cumulative damage from hadrons cannot be excluded and could appear when a high total integrated dose is reached. This concern is partially confirmed by irradiations in the very intense, mixed beam. Under the constant flux used, the damage appears in fact to be steadily increasing with accumulated dose. This is unlike pure ionising radiation damage, which reaches equilibrium at a level depending on dose rate, not beyond what saturation of all colour centres can yield. Therefore, this constitutes an indication for a cumulative, hadron-specific damage.

859

For CMS, hadron fluences have been calculated 9 for 5 x 105 p b _ 1 (10 y running at LHC), yielding in the ECAL barrel (end caps) ~ 10 12 (~ 1014) charged hadrons/cm 2 . A hadron-specific damage could arise from the production, above a ~ 20 MeV threshold, of heavy fragments ("stars"), with up to 10 ^m range and energies up to ~ 100 MeV, causing a displacement of lattice atoms and energy losses along their path up to 50000 times the one of minimum-ionising particles. The damage caused by these processes is likely different from the one of ionising radiation, thus possibly cumulative. The primarily investigated quantity was the damage to Light Transmission measured longitudinally through the length (L) of the crystal and quantified as the induced absorption coefficient at peak-of-emission wavelength, /HJATD (420 nm) = j ; l n L^NIT, with LTINIT and LTEND the longitudinal Light Transmission at 420 nm before and after irradiation. The Perkin Elmer A900 spectrophotometer used, allows in fact to measure LT very accurately, to better than 1%. Transmission is furthermore related to LO changes, provided scintillation is not affected. Eight CMS production crystals of consistent quality were irradiated at the IRRAD1 facility of the CERN PS accelerator T7 beam line 10 in a 20 GeV/c proton flux of 10 12 p/cm 2 /h (crystals a", b, c, d, e, h) or of

(a) For proton-induced damage.

(b) For 7-induced damage. The top thin line shows LT prior to irradiation.

Figure 1. Longitudinal Light Transmission curves for crystals with various degrees of radiation damage.

860

10 13 p/cm / h (crystals E, F', G) a . To disentangle the contribution to damage from the associated ionising dose, complementary 60 Co 7-irradiations were performed at a dose rate of 1 kGy/h on seven further crystals (t, u, v, w, x, y, z) at the ENEA Casaccia Calliope plant 11 . In fact, a flux of 1012 p/cm / h has an associated ionising dose rate in PbWC-4 of 1 kGy/h. The LT data in Fig. 1(a) show a smooth worsening of LT with increasing proton fluence over the entire range of wavelengths, and a clear shift of the Transmission band-edge. In 7-irradiated crystals (Fig. 1(b), where also the emission spectrum 12 is indicated), the band-edge does not shift at all, even after the highest cumulated dose reached: just the usual absorption band appears around 420 nm. These data thus give prominence to the qualitatively different fundamental nature of proton-induced and 7-induced damage. The correlation in Fig. 2(a), between /i/jv.D(420 nm) and fluence, is consistent with a linear behaviour over two orders of magnitude, showing

EO„( 22 Na) (cm 2 )

\ (nm)

(a) Induced absorption HIND (420 nm) as

a function of cumulated proton

(b)

fluence.

HIND (420 nm)

as a function of A

Figure 2. Behaviour of irradiation damage for crystals irradiated with protons (a", b, c, d, E, F', G, h) and with 7 (v, y).

a

P r i m e (respectively ") indicates a second (or third) irradiation of the same crystal.

861

loss

that proton-induced damage in PbWC*4 is predominantly cumulative, unlike 7-induced damage, which reaches equilibrium 1,4 . Figure 2(b) shows M/N£>(420 nm) plotted versus light wavelength for the proton-irradiated crystal a" and for the two 7 - irradiated crystals v and y. The dot-dashed line shows A - 4 fitted to the data of the proton-damaged crystal a". The good agreement is an indication of Rayleigh scattering from small centres of severe damage. This is consistent with an origin of damage due to the high energy deposition of heavily ionising fragment along their path, that changes locally the crystal structure. Taking into account the difference in composition and energy spectra between 20 GeV/c protons and CMS, simulations indicate that the test results cover the CMS running condi1 0.9 0.8 p-1

0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

loss

|aIND(420 nm) (m"1) 1 0.9 0.8 H-l

0.7

f PRELIMINARY **

'-

«?*AA

0.6 0.5

* • •

0.4

• w A x O y

0.3 0.2 0.1 0

t U V

• z

f

, , 1 , , , , 1 , ,

0.5

1.5

•

•

1

2.5

,

,

,

,

1 .

,

,

,

3.5

^ I N D (420nm)(nT) Figure 3. Correlation between ^*/;v£>(420 nm) and Light Output loss for proton-induced (top) and 7-induced (bottom) damage.

862

tions up to ~ 2.6. An experimental confirmation is expected in the future from a pion-irradiation of PbWC>4, closely approximating the CMS particle spectrum and energies. The evolution of Light Output was also monitored on the same set of irradiated crystals, using cosmic muons, traversing the crystals transversely and thus leaving approximately 30 MeV of energy deposit, to excite scintillation. The correlation 13 between /i/jv£>(420 nm) and Light Output loss is shown in the top part of Fig. 3 for all proton-irradiated crystals, and at the bottom for all 7-irradiated ones. The vertical bars indicate the systematic scale uncertainty affecting the data for a' and F'. For both, proton-irradiated and 7-irradiated crystals, the measured Light Output loss correlates well with /XJJV£>(420 nm). Furthermore, within the precision of the measurements, no difference can be observed in this correlation between the two sets of crystals and thus no hadron-specific alteration of the scintillation properties can be claimed. Proton and 7 data are also compared in a study performed on BGO 1 4 . The changes in band-edge are similar to what is seen in PbW04, and long enough after irradiation, when the ionising-radiation damage contribution has recovered, one can extract a remaining proton-induced damage that behaves linearly with fluence, as visible in Fig. 4. The same exercise is not possible on Csl data from the same authors 15 because the damage caused

Proton damage of BGO crystals 1

•

1

1

• 8Sd after irradiation

11

• B

01

•

•

200

400

•

•

•

600 BOO 1000 proton fluence (krad)

•

1200

•

J40Q

Figure 4. Correlation between IIIND (420 nm) and proton fluence in BGO extracted from published d a t a (see text).

863 by ionising radiation gives a contribution which is too important to allow disentangling the proton-specific one. In conclusion, one can say that for all crystals commonly used in calorimetry, beyond the well-studied damage from ionising radiation, the understanding of additional contributions to the damage, when crystals experience a substantial hadron flux, has become important since experiments are being built having to cope with such running conditions. A hadron-specific, cumulative contribution, likely due to the intense local energy deposition from heavy fragments, has been observed in PbW04 and BGO. Over the explored flux and fluence ranges and within the accuracy of the measurements, this contribution is observed to only affect Light Transmission, and thus can be monitored through light injection. Additional studies are expected to consolidate the present understanding of hadron damage. References 1. R.Y. Zhu et al., Nucl. Instr. Meth. A413 (1998) 297-311. 2. H. Hofer, P. Lecomte, F. Nessi-Tedaldi, Nucl. Instr. Meth. A433 (1999) 630636. 3. R.Y.Zhu et al. Nucl. Instr. Meth. A376 (1996) 319. 4. H.F. Chen, K. Deiters, H. Hofer, P. Lecomte, F. Nessi-Tedaldi, Nucl. Instr. Meth. A414 (1998) 149-155. 5. L. Zhang et al., IEEE Trans. Nucl. Sci. 52 (2005) 1123-1130. 6. V Batarin et al., Nucl. Instr. Meth. A512 (2003) 488-505; V. Batarin et al., Nucl. Instr. Meth. A530 (2004) 286-292. 7. M. Huhtinen, P. Lecomte, D. Luckey, F. Nessi-Tedaldi, F. Pauss, Nucl. Instr. Meth. A545 (2005) 63-87. 8. V. Batarin et al., Nucl. Instr. Meth. A540 (2005) 131-139. 9. M. Huhtinen, P. Lecomte, D. Luckey, F. Nessi-Tedaldi, First Results on radiation damage in PbWOi crystals exposed to a 20 GeV/c proton beam, Talk presented at the 8th ICATPP conference, Como, October 2003. 10. M. Glaser et al., Nucl. Instr. Meth. A 426 (1999) 72. 11. S. Baccaro, A. Festinesi, B. Borgia, CERN CMS TN-1995/192, Geneva, 1995. 12. R.Y. Zhu, IEEE Trans. Nucl. Sci. 51 (2004) 1560-1567. 13. P. Lecomte, D. Luckey, F. Nessi-Tedaldi, F. Pauss, to be published. 14. M. Kobayashi et al., Nucl. Instr. Meth. 206 (1983) 107-117. 15. M. Kobayashi et al., Nucl. Instr. Meth. A328 (1993) 501-505.

RADIATION TESTING OF GLAST LAT TRACKER ASICS R. RANDO, D. BISELLO, A. CANDELORI, P. GIUBILATO, Istituto Nazionale di Fisica Nucleare and Dipartimento di Fisica, Universita di Padova, via Marzolo 8,1-35131, Padova, Italy A. BANGERT, M. HIRAYAMA, R. JOHNSON, H. F.-W. SADROZINSKI, M. SUGIZAKI, M. ZIEGLER, Santa Cruz Institute for Particle Physics, University of California, 1156 High Street, Santa Cruz, CA 95064, USA J. WYSS DIMSAT, Facolta di Ingegneria, Universita di Cassino, via DiBiasio 43,1-03043, Cassino (FR), Italy, and Istituto Nazionale di Fisica Nucleare, via F. Buonarotti 2,156100, Pisa, Italy GLAST is a next generation high-energy gamma-ray space observatory designed for observations of celestial gamma-ray sources in the energy band extending from 10 MeV to more than 100 GeV. This study summarizes the results obtained during the radiation testing of the ASIC chips used in the LAT Tracker. Both Single Event Effects (SEE) and Total Dose (TID) tests have been performed, as part of the Radiation Hardness Assurance (RHA) for the planned 5 year mission. Heavy ion SEE tests have been performed at the SIRAD irradiation facility at the INFN National Laboratories of Legnaro, Italy (LNL), with LET values ranging up to ~80 MeVxcm2/mg. The tolerance of the chips to ionizing radiation has been evaluated by irradiating chips with the spherical 60Co gamma source of the LNL CNR-ISOF laboratory.

1. GLAST LAT Tracker Electronics 1.1. Introduction NASA's EGRET experiment on the Compton Gamma Ray Observatory [1] revolutionized the field of gamma ray astronomy; its success and the new questions it posed demanded a follow-up mission with widely expanded capabilities. The Large Area Telescope (LAT) [2],[3] on the Gamma-ray Large Area Space Telescope (GLAST) is a next generation gamma-ray pair conversion telescope that makes use of silicon-strip detector technology [4]. Gamma rays above ~20 MeV cross a plastic Anticoincidence Detector (ACD) without triggering it and convert in one of the 16 tungsten foils within the 864

865

Tracker (TKR). The electron-positron pairs pass through the remaining silicon microstrip detectors in the TKR, leaving behind tracks that allow the reconstruction of the gamma ray direction. Below the tracker secondary particles enter a hodoscopic Csl Calorimeter (CAL) where energy is measured. 1.2. TKR Electronics The GLAST LAT detector is highly modular: it consists of 16 tower modules, each of which has an 18 layer TKR, the Csl CAL and the Tower Electronic Module (TEM) readout electronics. The whole tracker has a total silicon surface of about 83 m2, segmented into almost 885,000 channels. Each silicon layer in a TKR module is connected through a pitch adapter to a hybrid Multi-Chip Module (MCM), containing the layer readout ASICs plus some standard surface-mount electronic components to provide appropriate biasing and power. The readout is implemented using two ASICs, produced in Agilent 0.5 um CMOS technology. The GLAST LAT Front End (GTFE) ASICs are composed of an analog part (amplifier-shaper-threshold; 64 channels) plus the digital blocks for digitalization and communication [5]. Each MCM accommodates 24 GTFE ASICs, each connected to its neighbors to allow data transfer and trigger propagation. At both ends of this chain lies a GLAST LAT Readout Controller (GTRC) ASIC, a fully digital chip that interfaces the GTFEs with the data acquisition electronics in the DAQ. Taking into account the number of towers, layers and channels in a layer we obtain a total of 13,824 GTFEs and 1,152 GTRCs in the whole TKR. Each GTFE ASIC contains three 64 bit registers that enable-disable calibration pulse, data taking and trigger generation capabilities for each channel (henceforth labeled CAL, CHN and TRG respectively). Another 14 bit register (DAC) sets calibration pulse and cutoff threshold levels, while a small, two bit register (DEAF) permits the selection of leftward and rightward readout and the switching off of a malfunctioning chip. Each GTRC ASIC has two registers: REG (34 bits) has 22 bits used to store configuration parameters, 7 bits used as error and status flags and 5 bits which enable the modification of specific parts of the configuration, while SYNC (5 bits) controls synchronization with the GTFEs.

866

2. Radiation Damage Assessment 2.1. Radiation Effects Deployment in space exposes TKR ASICs to ionizing radiation and, consequently, to radiation induced damage, which can be classified into two categories: Single Event Effects (SEE), where the passage of a single ionizing particle causes an undesired process to occur within an ASIC, and Total Ionizing Dose effects (TID) due to the cumulative action of ionizing radiation during a prolonged exposure. Among all SEEs the most relevant for the LAT are Single Event Upsets (SEU) and Single Event Latchup (SEL). In a SEU a single particle releases enough charge in the proximity of a memory cell to change its status (l-»0 or 0—>1), thus corrupting data or modifying the ASIC configuration. To limit the impact of these phenomena all registers in both GTFE and GTRC ASICs are SEU-hardened [6]. In a SEL the energy released by a single particle is injected into parasitic pn-p-n structures inherent to CMOS technology, triggering a short-circuit that can lead to the destruction of the affected device. The Agilent 0.5 um CMOS process is relatively radiation hard, being based on an epitaxial structure; the thickness of the processed layer is about 13 um including the epitaxial layer. This technology has already been shown to be SEU resistant to heavy ions [7]. 2.2. The GLAST Radiation Environment GLAST will be in a low Earth orbit, at an altitude of 565 km and an inclination of 28.5 degrees. The instrumentation will be consequently shielded from a large portion of the space radiation. The orbit will, however, intersect the South Atlantic Anomaly (SAA) which will give the most important contribution to total dose. The radiation environment in which the telescope will operate is well understood [2]. The spectra of Galactic Cosmic Ray nuclei (GCR) are well represented by the CREME model; the Solar Particle (SPE) contribution is calculated, taking into account the modulation of solar activity and the corresponding atmospheric influence on the magnetic belts. Albedo from charged particles interacting with the Earth atmosphere is described by data collected during previous experiments. The expected total ionizing dose is due to trapped protons in the SAA. The accumulated TID in a 5 year mission is about 0.8 krd in the TKR outer layers. Allowing for a factor five engineering margin, we obtain about 4 krd; we

867 required testing of all ASICs up to 10 krd to allow for an overall margin of 12.5 times the expected radiation levels. All dose values reported in this paper are referred to silicon. For SEE issues we will consider GCR and SPE, taking into account the Linear Energy Transfer (LET) at the surface of silicon. For GCR we can estimate an upper energy cutoff at around 28 MeVxcm2/mg, while for SPE the limit is about 100 MeVxcm2/mg. Irradiation performed on analog-digital test structures indicated a SEE threshold of about 8 MeVxcmVmg; since this threshold value is confirmed by the data presented in this report, the threshold will be held fixed when fitting to simplify the convergence. Above this LET we estimate a worst case scenario by multiplying the maximum possible particle flux in the planned 5 years of operation by the total mission time, obtaining ~0.2 ions/cm2 due to GCR and <0.1 ions/cm2 due to SPE in 5 years. As a conservative upper limit we will then assume a fluence of 1 ion/cm2 in 5 years above the SEE LET threshold. For validation purposes we require the expected number of upsets in the whole LAT in 5 years to be smaller than about 700. As LAT configuration data will be reloaded periodically, this SEU rate is an acceptable safety factor to avoid an excessive data loss. Expected SELs, far more dangerous, must be less than 0.5 in 5 years. 2.3. Experimental Approach During the planning phase of the Radiation Hardness Assurance of TKR ASICs we first investigated several samples to understand their behavior after irradiation (SEE threshold, saturation, TID effects); the obtained results are reported in the following sections, while a much more detailed description is given in [8]. The derived test plan was subsequently applied in the screening phase (validation) for the ASIC flight lot, from which 2 GTFEs and 2 GTRCs were required to be tested for SEE effects and 7 ASICs of each type for TID. For all irradiations we used smaller area test-MCMs which can accommodate only 7 GTFEs and 2 GTRCs. Heavy ions irradiations were performed at the SIRAD beam line at the INFN National Laboratories of Legnaro (LNL), Italy [9]. A Tandem Van de Graaf 15 MV accelerator provides ion beams ranging from H to Au. A summary of the ion beams used for this study is reported in Table I. The surface LET values range from the aforementioned threshold (8 MeVxcm2/mg) up to about 80 MeVxcm2/mg, well above the value when SEE saturation is expected from previous measurements on test structures.

868 Table 1. Ion species at the SIRAD facility Ion Species 28 Si S8 Ni 79 Br 107 Ag 197 Au

LET (MeVxcm2 /mg) 8.5 28.4 38.8 54.7 81.7

Range in Si (Mm) 62 34 31 28 23

For TID we used gamma rays from a 4n 60Co source of the CNR-ISOF laboratory at LNL. Following standard ASTM and MIL procedures (see for example [10]) devices were placed within a Pb-Al box (wall thickness: 2.5 mm Al, 2 mm Pb) to eliminate low energy photon and electron components that can lead to dose enhancement on the surface of the device under test. Each ASIC was powered and clocked during irradiation; uniformity and dose rate were experimentally verified before the validation phase started and were found to be those expected by geometrical considerations (dose rate 1 rd/s, uniformity better than 10% on a test-MCM). 3. Experimental Results 3.1. Experimental results: SEU and SEL The SEU cross sections per ASIC measured during the irradiations at SIRAD were fitted with the Weibull function [11] ( &SEU(L)

=

S

1-exp

T - i L-L 0

W

W J

where L0 is the threshold LET, L>L0 the ion LET, S the saturation limiting value and W the curve width. Fit curves and the experimental data they are based on are shown in Figure 1. To obtain the SEU estimates in the worst possible scenario, we multiplied the average saturation limit S by the number of ions hitting the TKR in 5 years and by the total number of bits in the considered ASIC, thus assuming that all impinging ions have saturation LET. For GTRC we have ignored "special" bits within the REG register as a SEU occurring there will not change the behavior of

869

the electronics. We obtain an upper limit SEU rate of 0.7 SEU in the whole TKR in 5 years. 100.0

=£ §10.0

A

1.0

^OAL fit

O

'

... • •

^^^~

0.1 10

20

30

40 50 LET [MeV-cm2/mg)] Figure 1. Weibull fit of SEU cross sections.

60

CJ CAL

CT

TRG

m

<*TRG

CT

CHN

f i ' CTCHN CT

CONF

fit

°CONF

70

80

No latchups were observed even after 5><106 Au ions/cm2 on GTRC ASICs and 4*106 Au ions/cm2 on GTFEs. SEL upper limits are calculated in a fashion similar to what we did for SEU, considering that Poisson statistics implies that when no SEL was observed an upper limit of 3 can be derived at 95% confidence level. We therefore calculate an upper limit of 0.01 SEL in the whole TKR in 5 years, at 95% confidence level; see also [8]. 3.2. Verifying the SEE Experimental Results At the Radiation Effects Facility of the Cyclotron Institute, Texas A&M University [12] we investigated a possible dependence of SEL cross sections on the ion range. With respect to previous measurements at SIRAD, ion ranges were increased by a factor of 4 while surface LET was kept the same. Delivered faiences were increased by a factor of ~10 with respect of those employed at SIRAD. No significant change in the SEU cross sections was found; once again no SEL was observed.

870

The increased delivered fluence allowed us to improve the upper limits on the SEL rates: now the upper limit on the probability of suffering a SEL in the whole TKR is ~5x 10"4 in 5 years at 95% confidence level. 3.3. Total Dose Effects Power consumption was monitored to look for an increase due to ionizing radiation; no significant increase after irradiation was observed, while fluctuations remained within 5%. A series of functionality tests was performed on both GTRC and GTFE ASICs. Monitored functionalities include register access and modification and, for GTFE only, analog performances: noise, gain, threshold and calibration pulse were thoroughly examined. The entire test set was repeated before irradiation and after each 2.5 krd step up to a total of 10 krd. No significant change in DUT behavior was observed. For more details see [8].

4. Conclusions We tested LAT TKR ASICs both for SEE and for TID effects. No catastrophic damage was observed in any tested device. The number of expected SEE events in the LAT TKR is negligible for the planned 5 year mission, even considering that over 13,000 ASICs will be employed. References 1. 2. 3. 4. 5. 6. 7.

P.L. Nolan et al, IEEE Trans. Nucl. Sci., vol. 39, no. 4, Aug 1992. P.F. Michelson, Response to AO 99-OSS-03, Stanford Univ., E 2002. P.F. Michelson, Proc. SPIE, vol. 2806, October 1996. W. B. Atwood et al, Nucl. Instrum. Methods A 435, no. 1-2, Oct 1999. R.P. Johnson et al, IEEE Trans. Nucl. Sci., vol. 45, no. 3, Jun. 1998. L.R. Rockett, IEEE Trans. Nucl. Sci., vol 35, no. 6, Dec. 1998. J.V. Osborn et al, Proceedings of the 7th NASA Symposium on VLSI Design, 1998 8. R. Rando et al, IEEE Trans. Nucl. Sci., vol. 51, no. 3, Jun. 2004. 9. J. Wyss et al, Nucl. Instrum. Methods, A 462, no. 3, Aug. 2001. 10. "Standard Guide for Ionizing Radiation (Total Dose) Effects Testing of Semiconductor Devices," ASTM guide, F 1892-98 11. E.L. Petersen et al, IEEE Trans. Nucl. Sci., vol. 39, no. 6, December 1992 12. Texas A&M University Cyclotron Institute, MS #3366, College Station TX 77843-3366.

M A G N E T I C FIELD A N D RADIATION TESTS OF A P R O G R A M M A B L E DELAY LINE

A. AKINDINOV, YU. GRISHUK, S. KISELEV,D. MAL'KEVICH, A. MARTEMIYANOV, A. SMIRNITSKIY, M. RYABININ, K. VOLOSHIN, B. ZAGREEV, A. ZININE Institute for Theoretical and Experimental Physics, Moscow, Russia A. ALICI+, P. ANTONIOLI, S. ARCELLI+, M. BASILE+, G. CARA ROMEO, L. CIFARELLI+, F. CINDOLO, I. D' ANTONE D. HATZIFOTIADOU, G. LAURENTI, M.L. LUVISETTO, A. MARGOTTI, S. MENEGHINI, R. NANIA, F. NOFERINI+, G. PELLEGRINI, A. PESCI, O. PINAZZA, M. RIZZI, E. SCAPPARONEr G. SCIOLI+, G. VALENTI, M.C.S. WILLIAMS, C. ZAMPOLLI+, A. ZICHICHI+, M. ZUFFA. Istituto Nazionale di Fisica Nucleare, Sez. di Bologna, Bologna, Italy (+) also Universita degli Studi di Bologna, Bologna, Italy A. DE CARO, S. DE PASQUALE, A. DI BARTOLOMEO, M. FUSCO GIRARD, M. GUIDA, S. SELLITTO, Dipartimento

di Fisica dell'Universita and INFN, Salerno, Italy D.W. KIM

Dept. of Physics, Kangnung National University, Kangnung, South Korea S. DON ATI Istituto Nazionale di Fisica Nucleare, Sez. di Pisa, Pisa, Italy D. H. KIM World Laboratory, Lausanne,

Switzerland

•presenting and corresponding author: [email protected]

871

872 Programmable Delay Lines (PDLs) are widely used in trigger systems. We performed tests under radiation and magnetic field for the chip 3D3418-0.25, to be used in the ALICE-TOF trigger system. The tests showed that this chip can comfortably operate in a 0.6 T magnetic field and tolerate a dose larger than 446 Gy.

1. Introduction The goal of the ALICE-TOF detector 1 is the identification of the charged particles produced in Pb-Pb collisions at LHC, with momentum 0.5 GeV/c20 MeV). Although the total absorbed dose is quite moderate, single event upset (SEU) and single event latch-up (SEL) could spoil the performance of the electronic components. We selected the 3D3418-0.25 as Programmable Delay Line, made by Data Delay Device3. This 8-bit PDL is a CommercialOff-The-Shelf(COTS) device and is not specifically designed to withstand radiation. Since no data on the performance under radiation and magnetic field was available4, we planned dedicated measurements. 2. Setup description A pattern of six synchronous LVDS signals, generated by a VME I/O register, was sent to a VME board, called LTM (Local Trigger Module). Each LTM was equipped with two LVDS Quad receivers DS90LV048A followed by six PDLs 3D3418-0.25. The PDL output signals, after passing through a FPGA Altera FLEX10K100, were sent to a LVDS transceiver SN65LVDM1677, set in driver mode. A 6 m long Amphenol cable 1643099-984 brought the 6 signals to a HPTDC 5 board, equipped with a high resolution TDC. The HPTDC and the I/O register were housed in a standard 6U VME crate, while the LTM was connected to the P I connector of the VME crate through a i m long Simpex FBK0.635-68-HF cable. Such a configuration was mandatory since a VME crate could not be placed in the

873

beam line nor inside the magnet. The LTM was mounted on a iron support,

4000

5000 Time(s)

Figure 1. Relative delay of the i-th channel (i=2,6) with respect to the first one. The i-th channel is delayed by (0 ns, 0.5 ns, 0.75 ns, 1 ns, 1.5 ns, 2 ns) every 30 s before the irradiation (see text).

1 cm thick, protecting the FPGA and the other components; a dedicated hole, drilled in the iron, allowed the irradiation of the PDLs. The PDLs were set by the FPGA, implementing the VME protocol too. 3. Setup configuration The first LTM channel served as a reference time, therefore the corresponding PDL delay was not set. The other five PDL delays were changed every 30 s. At each step the delay was increased for each PDL respectively of 0.5 ns , 0.75 ns, 1 ns, 1.5 ns and 2 ns. In this way a large variety of delay configurations are checked. Fig. 1 shows the relative delay between the i-th channel (i=2,6) with respect to the first one, as a function of the running time. 4. Irradiation test The PDLs were irradiated at PSI on December 2004 using a 60 MeV proton beam, collecting a total fluence F=3.2-10 11 p/cm 2 . Since the dE/dx in silicon for protons with energy Ep=60 MeV is 8.72-10 -3 MeVcm 2 mg~ x and 1 Gy=6.25-10 6 MeVmg" 1 , the total absorbed dose by the PDLs is: 8.72 • 10- 3 -Gy = U6.bGy. (1) 6.25 • 106 Several sources of time jitter contribute to the time resolution of the system D = 3.2-10 l i

874

as

|10J

r

%

:

^10

r

Z10

r

ktA 2

s3

1

r...

-10 Figure 2.

A

BEFORE IRRADIATION

A

1

r

1 i . . .

-8

i . . .

i . . .

i jr..

-6

G variable distribution

i.

U i . . .

i

8 10 G(ns) before (a) and during (b) the irradiation

.

The I/O pattern register has a time jitter from channel to channel giving a rms of about 50 ps, while the intrinsinc time resolution of the HPTDC (INL calibration not applied 5 ), is about 60 ps. The larger contribution to the jitter comes anyway from the PDL itself. We considered the distribution of the difference G=T^p-T*'J~1-Ti, where T*'Jp is the measured time difference between the channel i and the first channel for the j-th event. T, is a value to be accounted for when, every 30 s, the delay with respect to the channel one is changed. Such variable measures the time difference between a given event with respect to the previous one, and is therefore a sensitive SEU indicator. Fig. 2 (top ) shows the distribution of G before irradiation (the five channels are summed). The G distribution is fully contained within ±2ns: considering that the LSB is 0.25 ns, we can control the five most significant bits of the PDL. A SEU in one of these five bits under control, JVc6ifs, would manifest as a difference of at least ±2 ns. Fig. 2 (bottom) shows the G distribution during the irradiation test. Since no event was detected beyond ±2 ns, we can conclude that the upper limit for the SEU cross section is: osEuim

=

2 NpD LNcbitF

= 2 - 8 7 ' 1 0 ~ 1 3 c m 2 (90%C.L.)

(2)

The full ALICE-TOF trigger system will implement 48 PDLs in each of the 72 LTM boards, giving a total of 3456 PDLs. The rate of SEU for the

875

TOF trigger is: RSEU

= Nbit/PDL

< 22.3 SEU/year

• NpDL

• OSEU/bit

• Rhad =

(3)

(90%C.L.)

This is an extremely low SEU rate that can be comfortably tolerated. The LTM board was supplied by a Low Voltage Power supply, providing a 5 V voltage. The Low Voltage and the current values, displayed in the Power Supply, were web casted and stored using the video capturing Snagit TechSmith software. We did not observe any variation in the driven current, excluding the occurence of SEL. 5. Magnetic field test The PDLs were operated in a 0.6 T magnetic field using the CERN SPS XH2B beam line dipolar magnet. Three different runs were collected, orienting the PDL three axes, parallel to the magnetic field direction. The obtained G variable distribution for the three orientation did not show any event beyond ± 2 ns. We can conclude that the PDLs performance is not affected by a 0.6 T magnetic field, independent of the field orientation. 6. Conclusions The 3D3418-0.25 Programmable Delay Lines have been tested during irradiation tests with a total dose of 446.5 Gy. We did not oberve any single event upset, nor a latch-up. Then the PDLs were operated in a 0.6 T magnetic field: dedicated run were taken orienting the PDL three axis parallel to the magnetic field. We did not observe any change in the PDL performance. Considering these results, we can conclude that this chip is well suited to delay signals in the ALICE-TOF trigger system. References 1. N. Ahmad et al.(ALICE Coll.) , Time of Flight, Technical Design Report, CERN-LHCC 2000-12, ALICE-TDR 8, Febraury 16, 2000; P. Cortese et al., (ALICE Collab.) , Time of Flight, Technical Design Report, Addendum, CERN-LHCC 2002-016, Addendum to ALICE TDR 8, April 24, 2002. 2. A. V. Akindinov et al., Nucl. Instr. and Meth. A 532(2004) 611-621. 3. See http://www.datadelay.com/datasheets/3d3418.pdf 4. Data Delay device, private communication. 5. A. V. Akindinov et al., Nucl. Instr. and Meth. A 533(2004) 178-182.

LCFI C H A R G E T R A N S F E R INEFFICIENCY STUDIES FOR CCD VERTEX D E T E C T O R S

ANDRE SOPCZAK Lancaster

University

The Linear Collider Flavour Identification (LCFI) collaboration studies CCD detectors for quark flavour identification in the framework of a future linear e+e~ collider. The flavour identification is based on precision reconstruction of charged tracks very close t o the interaction point. Therefore, this detector will be exposed to a high level of radiation and thus an important aspect of the vertex detector development are radiation hardness studies. Results of detailed simulations of the charged transport properties of a CCD prototype chip are reported and compared with initial measurements. The simulation program allows to study the effect of radiation damage after the exposure of the detector to a realistic radiation dose, which is expected in the environment of detector operation at a future LC.

1. I n t r o d u c t i o n An important requirement of a vertex detector is to remain tolerant to radiation damage for its anticipated lifetime. Two different CCDs have been considered in this study. The majority of simulations so far have been performed for the 3-phase CCD, CCD58, with serial readout. First data measurements have been performed on an unirradiated 2-phase column parallel CCD, CPC-1. CCDs suffer from both surface and bulk radiation damage, however, when considering charge transfer losses in buried channel devices only bulk traps are important. These defects create energy levels between the conduction and valance band, hence electrons may be captured by these new levels. Captured carriers are also emitted back to the conduction band, but on a different time scale. For a signal packet this may lead to a decrease in charge as it is transfered to the output and may be quantified by its Charge Transfer Inefficiency (CTI), where a charge of amplitude QQ transported across m pixels will have a reduced charge given by Qm = Qo(l - CTI) m .

(1)

The CTI value depends on many parameters, some related to the trap 876

877

characteristics such as: trap energy level, capture cross-section, and trap concentration. Operating conditions also affect the CTI as there is a strong temperature dependence on the trap capture rate and also a variation of the CTI with the readout frequency. Other factors are also relevant, for example the occupancy ratio of pixels, which influences the fraction of filled traps in the CCD transport region. Dark current effects have been measured. Previous studies are for example reported in Ref. : . 2. Simulation The simulations with ISE-TCAD (version 7.5) are performed using a 2dimensional model for a 3-phase CCD (Fig. 1). Parameters of interest are the readout frequency, up to 50 MHz, and the operating temperature between 100 K and 250 K. The charge in transfer and the trapped charge is shown in Fig. 2. From the two traps considered only the 0.17 eV trap produced a non-negligible CTI for the explored parameter ranges. The signal charge used in the simulation represents a charge deposited from an Fe 55 source. The X-ray emission (mainly 5.9 keV) generates about 1620 electrons in the CCD, which is similar to the charge generated by a MIP. The linearity of the CTI value with respect to the trap concentration was verified in the simulation. Node

0

3

10

6

20 30 Length (microns)

9

Ec — Et 0.17 eV 0.44 eV

40

C (cm" 3 ) 1 x 10 11 1 x 10 11

an (cm - 2 ) 1 x 10- 1 4 1 x lO-15

Figure 1. Left: Detector structure and potential at gates (nodes) after initialization. The signal charge is injected under node 3. Right: Energy levels E, trap concentrations C, and electron-capture cross-section crn used in simulation.

20

25 Length (microns)

30

Figure 2. Left: Signal charge density, almost at output gate. Right: Trapped charge density, from transfer of signal charge.

878

2.1. 0.44 eV Trap CTI

Contribution

We consider partially filled traps to improve the simulation by representing a continuous readout process. The results from the initially empty and partially filled traps are compared in Fig. 3. A negligible contribution to the CTI from 0.44 eV trapping for partially filled traps (due to long emission time) are obtained. Thus, the 0.44 eV traps are ignored in further studies. CTI using initially empty 0.44eV traps, 50MHz.

2

&10 0.4

250

CTI using new code for o.44eV traps, 50MHz.

300

250

Temp (K) Figure 3. eV traps.

300

Temp K

Left: CTI value for initially empty 0.44 eV traps. Right: Partially filled 0.44

2.2. 0.17 eV Trap CTI

Contribution

Figure 4 shows the CTI simulation for initially empty and partially filled traps. A clear peak structure is observed. New experimental data will cover the simulated temperature range. ISE-TCAD Simulation 50 MHz readout

Simulation S i m . single

30 0.17eV 20 -

"\^ \

\

10 0.44eV

ISO

trap levels

\

\ \ ^ 200

250

: ; '. '

Partial trap filling D.17ev traps, SOMHz Partial traps

100

1SO

200

250

T e m p (K) T e m p (K) Figure 4. Left: CTI value for initially empty 0.17 eV traps, 0.44 eV traps and their sum. Right: Partially filled 0.17 eV traps.

2.3. Frequency

Dependence

The frequency dependence is shown in Fig. 5 for initially empty and partially filled traps. For higher readout frequency there is less time to trap the charge, and thus the CTI is reduced near the CTI peak region. At high temperatures, the emission time is so short that trapped charges rejoin the passing signal.

879

CTI for Initially empty 0.17oV trap*

&

Partially filled 0.17eV traps

-

X

7MHz 25MHz SOMHz

so 40

-"

-

/

\

--

30

*a

• t 20 •

—

">

*-^Nx

10 •

150

0

200

7MH* 25MHz 50MHz

\

:

•^v

•

100

125 150

«v 175 2 0 0 225

250

T e m p (K) T e m p (K) Figure 5. Left: Frequency dependence for initially empty 0.17 eV traps. Right: Partially filled 0.17 eV traps.

3. Simple Model A simple model of charge trapping was constructed that considered the capture and emission of electrons from traps to and from the conduction band. These processes are parameterized by two timescales to form a differential rate equation. Solution of this equation with relevant boundary conditions allowed this simple model to be compared with the full ISE simulation results. Capture

Emission

Emission

'"-

6 0 -r Comparison of simulated and modelled CTI

f O

50 • — S i m p l e Mod >l - Simulatior

40 •

— — —

30

7MHz 25MHz 50MH.

20 lO -

Dunns transfer

After transfer

100

125

150

175

200

225

Temp (K)

Figure 6. Left: Traps capture electrons from the signal charge and electrons are emitted from the filled traps. Right: Comparison between simple model (solid lines) and ISE simulation (dots) for three readout frequencies.

Figure 6 shows the comparison between the full simulation and the simple model as a function of the temperature. At low temperatures the time for traps to emit captured electrons is far longer than the readout time, hence traps remain filled and no further electrons can be captured. At high temperatures the emission time is much faster than the readout frequency, so captured electrons are released back to the conduction band fast enough to rejoin their original signal packet. The CTI value is again reduced. Slower readout frequencies have higher CTI values near their peaks as each pixel has a longer occupation time for the signal charge resulting in greater net electron capture.

880

4. Initial Data Measurements For the environment of a future Linear Collider a serial readout for CCDs is no longer an option. Instead column parallel technology, using readout electronics for each column, are in development to cope with the required readout rate. CPC-1 is a prototype 2-phase CCD capable of 25 MHz readout frequency. Initial measurements have been performed on an unirradiated device in standalone mode, where four columns of the CCD were connected to external ADC amplifiers. An Fe 55 source provides the signal charge (Fig. 7). The determination of the CTI involves measuring the charge reduction as a function of the pixel number from a known initial charge. The results so far have shown small CTI values (< 10~5) for the unirradiated CCD under normal operation conditions. It is possible to induce CTI-like effects by reducing the clock voltages used to transfer charge. For 1 MHz readout the CTI value was observed to increase sharply below 1.9 V (peak-to-peak). I Channels 1,243 I

trff

f4—

Pedestal K

Fe (5.9keV) Signal

Figure 7. Measured signal distribution for an Fe 5 5 source. Noise « 60 e - . Frequency = 1 MHz.

I

I

IMF-50

-100

Integration time = 500 ms.

Gain-HeVADC

X 0

50

100

TRS

....ISft 150

200

- 3 0 °C.

2000 frames.

250

ADC

4.1. CTI - Event

Selection

The signal is induced by an Fe 55 source which provides isolated hits of about 1620 electrons in order to determine the CTI value. Hits are located using a 3x3 cluster method and selection criteria are applied: Pixel amplitude > 5(Xnoise, £®=i |clusterj| < 8a n o j s e . Events are selected within ±2a of the signal peak as shown in Fig. 8. 4.2. CTI -

Determination

The CTI is determined from isolated pixel hits. The distribution of the ADC amplitude Q against the pixel number gives CTI = — J - d ,pffel-,, where Qo is the intercept from a straight-line fit. An example for an unirradiated device and low clock voltage is given in Fig. 9. The decrease of the clock voltage reduces the transfer efficiency which provides a possibility to measure CTI values as function of the clock voltage.

881 Isolated pixel M s . ch 2.

cenl:3[:cluslef|cluster.:353fi8cer.iral<20C) _

I

Isolated pixel hits

I'M

Entries

12896

x'/ndt

60.07/32

Constant 2754? 4.7 Mean

129.810.1

Sigma

9.52:0.16

?^ViVwr*K// •'

»

100

ISO

WO : - • . ; : - . :

Figure 8.

Left: Isolated hit. Right: Extracted signal peak. ?7nS

CTI [file: c0O567.dat) I

•.:••

s

254/276

P0

1 3 6 ? : 0.3252

pi

•aOKU--0.0OI8M

<mnf

Cn.|17.558l1.334)-10

Figure 9.

4.3. Dark

Left: CTI determination. Right: CTI as function of clock voltage.

Current

Some thermally generated electrons are captured in the potential wells. The collected charge is proportional to the integration time. 10 overclocks sampled per frame are used as reference level. The gain (e~/ADC) is calibrated from an Fe 55 source (at each temperature). From a fit to Jdc = T3 exp (a — 0/T), a uniform dark current characteristics is observed across the four channels (Fig. 10). 5. S u m m a r y Radiation hardness effects have been simulated for a serial readout CCD, CCD58. A simple model has been developed and compares well with the simulation which shows the strong dependence on the operating conditions of the CCD. Measurements performed on a prototype column parallel CCD (CPC-1) display as expected no measurable CTI for the unirradiated system. CTI-like effects were induced by reducing the clock voltage to explore the CTI measurement method. Comparisons with experimental data are being carried out at a test-stand. The test-stand at Liverpool University has been shown to operate in the required temperature range. The

882 I DarkCurrant

-30

-20

|

-10 Inlegratlbn'tlme (s) X'ln&l 3.30673 oi 17,65± 1.204 ;., 73631389.5

Figure 10. Upper left: Dark current measurement method. Upper right: Dark current at different temperatures. Lower left: Dark current fit to theory expectation.

performances of the CCD58 and C P C prototypes will be compared with simulations. Experiences with the prototypes have already been gained at the Rutherford Appleton Laboratory (RAL). T h e verification of the simulation results and the tuning of the simulation is i m p o r t a n t for the studies to guide the future C C D development as vertex detector for a future Linear Collider. Acknowledgments This work is supported by the Particle Physics and Astronomy Research Council ( P P A R C ) and Lancaster University. I would like t o t h a n k Konstantin Stefanov and James Walder for comments on the manuscript.

References 1. K. Stefanov, PhD thesis, Saga University (Japan), "Radiation damage effects in CCD sensors for tracking applications in high energy physics", 2001; O. Ursache, Diploma thesis, University of Siegen (Germany), "Charge transfer efficiency simulation in CCD for application as vertex detector in the LCFI collaboration", 2003; and references therein.

PHOTOLUMINESCENCE AND y-RAY IRRADIATION OF SrOB203-P205:Eu2+ AND SrMo04:Eu3+ PHOSPHORS CHAOFENG ZHU, XIANLING CHEN, SHUANGLONG YUAN, YUNXIA YANG, GUORONG CHEN Institute of Inorganic Materials, School of Materials Science and Engineering, East China University of Science & Technology, Shanghai 200237, China S. BACCARO, A. CECILIA, M. FALCONIERI, L. SERALESSANDRI, M. BELLUSCI CR ENEA CASACCIA, Via Anguillarese 301, 00060 S.Maria di Galeria (Rome, Italy) Europium-doped SrO-B203-P20s and SrMoC>4 phosphors for application in white LEDs sources were synthesized and characterized. Photoluminescence spectra showed intense blue and red emissions after UV excitation which can be ascribed to Eu2+ and to Eu3+ respectively. The shape of the blue emission in samples was correlated to the existence of different phases in the Eu2+ doped samples as pointed out by X-ray diffraction Moreover, y- irradiation tests were performed on some samples in order to explore possible effects by X-ray diffraction, photoluminescence emission spectra and luminescence decay kinetics after short pulse excitation..

1. Introduction Nowadays, light emitting diodes (LEDs) have emerged as an important class of lighting devices. In particular white-light LEDs show high potential for replacement of conventional lighting sources like incandescent and fluorescent lamps, the advantage being their long lifetime, lower energy consumption, and environmental-friendly characteristics [4]. A white LED device has been commendably realized using YAG:Ce as a broad band yellow phosphor coated on the blue emitting InGaN-based chip [1,2,3]. Nowadays most researchers have concentrated on the excitation of phosphors by using an ultraviolet source. In our study, a blue phosphor, namely SrO-B203-P205:Eu2+ and a red phosphor, SrMo04:Eu3+ , have been synthesized by high temperature solid state reaction for white LEDs application. The excitation and emission spectra indicate that this kind of phosphors can be effectively excited by ultraviolet light and that they exhibit a satisfactory blue and red performance, nicely fitting with the widely applied UV emitting chips. It is known that y ray irradiation can induce defects in materials that might play a positive role of modifying properties of phosphor. However, there are very few reports on this topic. Therefore, in the present study, we also investigate the effect of y irradiation on these two phosphors aiming at exploring a possible way to modify their properties. 883

884

2. Experimental The composition of analyzed samples is reported in Table 1. Eu2+ doped samples (BP2, BP3 and BP10) phosphors were synthesized by high temperature solid-state reactions of mixtures of high purity SrC0 3 , Eu 2 0 3 , H 3 B0 3 and NH 4 H 2 P0 4 (the molar ratio of europium is 1% for all samples). Sample BP2 BP3 BP10 S2

Table 1. Composition on analyzed samples Composition 2SrO-O.lOB2O3-O.9OP2O5.Eu2* 2SrO-0.16B2O3-0.84P2O5:Eu2+ 2SrO-0.10B2O3-0.90P2O5:Eu2+ SrMo04:Eu3+

The mixture was ground thoroughly and mixed well in an agate mortar and preheated at 500°C in air for 2h, cooled at room temperature and grounded again. Mixtures were fired subsequently in alumina crucibles at 1100 °C for two hours under a carbon reducing atmosphere in muffle furnace in the case of BP2 and BP3 samples. For BP10 sample preparation, the firing temperature was higher and equal to 1250 °C. After firing, the samples were cooled down to room temperature in the furnace and were ground again. Then, the fired sample was washed in hot deionized water for several times in order to remove any residual boron and other impurities and dried at 100 °C in an oven. Eu3+ doped SrMo0 4 phosphor (the molar ratio of europium is 6%) was prepared by the same method with Mo0 3 , Eu 2 0 3 and SrC0 3 as starting materials and sintered at 700 °C in air for 2 hrs. The crystal structure of the sample was identified by X-ray diffraction (XRD) recorded on a Seifert PAD VI diffractometer equipped with Mo K„ radiation and a LiF monochromator on diffracted beam. Luminescence measurements of Eu2+ activated powders were performed by a Hitachi Perkin Elmer MPF-2A fluorescence spectrophotometer in the wavelength range 380-650 nm (A.EXc=365 nm). In the case of the S2 sample, considering that Eu3+ luminescence could be excited also by 532 nm wavelength, luminescence measurements were done by using a home-made microfluorimeter equipped with a frequency-doubled CW Nd:YAG laser source, which allowed to achieve a higher signal-to noise ratio and consequently much higher peak resolution with respect to conventional spectrofluorimeter. The microfluorimeter used a 63x microscope objective which was fiber-coupled to a 550 cm focal length monochromator equipped with a 500 nm blazed grating, and the luminescence was detected by a liquid-nitrogen cooled CCD. S2 sample underwent also pulsed luminescence measurements using a 12 nanosecond pulse

885

width, 10 Hz repetition rate Ti:Al 2 0 3 laser source equipped with a second harmonic generator allowing for emission of continuously tunable pulses from 360 nm to 460 nm. Typical energies of frequency-doubled pulses used for measurements were 1-3 mJ. The luminescence emission is collected by a f/1.2 Cassegrain optics and spectrally dispersed by a 18 cm focal length monochromator equipped with a 500 nm blazed grating. A long wave pass absorbing glass filter with 395 nm edge was placed in front of the photomultiplier detector to improve laser light rejection. The overall time response of this system was measured using the laser pulses and resulted to be around 200 ns. Luminescence emission waveforms were acquired by a digital oscilloscope and then transferred to a PC for data analysis. y irradiation tests were carried out at the Calliope 60Co source (ENEAFIS/ION, Italy) in a dosimetric point corresponding to the dose rate of 5000Gy/h at doses between 20k Gy-600 kGy. 3. Results & discussion The emission spectrum of BP2, BP3 and BP10 samples excited at X.exc=365 nm are shown in Figure la and in Figure lb respectively. Luminescence measurements are not absolute and to facilitate the comparison, BP2 and BP3 spectra are normalized to the luminescence intensity at 420 nm.

400

440

480 520 560 Wavelength [nm]

600

400

450 500 550 Wavelength [nm]

600

650

Figurel. Emission spectra of sample BP2 &BP3 (a) and of sample BP10 (b) (^cxc=365 nm)

As it can be seen, BP2 and BP3 spectra seem to be composed of an emission peak centered at 422 nm (FWHM=35 nm) and by a broad band centered at around 480 nm, whose relative intensity is different and which can be reasonably assigned to the Eu2+ 4f55d—>8S7/2 (4f7) transition [5]. On the contrary, the BP10 emission spectrum is lacking in the 422 nm luminescence peak. As it was observed by other authors in potassium halide matrices [6] the double peak structure of Eu2+ luminescence spectrum can be ascribed to the existence of

886

different phases in the analyzed material, where Eu + is subject to interaction with a different crystalline field.. As a matter of fact, the XRD patterns recorded on our samples evidence the existence of different phases as listed in Table 2. The main phases are related to divalent europium but some minor phases where trivalent europium enters where identified, probably due to the incomplete reduction of Eu3+ operated by the reducing atmosphere during synthesis.

Phase Sr 2 P 2 0 7 Sr 3 (P0 4 ) 2 Sr(P0 3 ) 2 SrB 2 0 4 EuP0 4 Eu(B0 2 ) 3 Eu 3 0 4

Table 2. Phases individuated by XRD patterns Eu valence BP2 BP3 Eu2+ Main Main Eu2+ Main Minor Eu2+ Main Traces Eu2+ Minor Traces Eu3+ Traces Traces 3+ Eu Traces Traces Eu2+/Eu3+ Traces Traces

BPIO Minor Absent Main Main Traces Traces Traces

The greatest contribution is related to Sr2P207 and Sr3(P04)2 phases where Eu2+ ions can enter in place of ST2*. In particular, BP2 powder is characterized by a higher content of Sr2P207 and BP3 sample contains equal concentration of Sr2P207 and Sr3(P04)2. As far as BP10 powder is concerned, its composition is lacking in the Sr3(P04)2 phase. In view of these evidences and considering the relative intensity of Eu2+ emission peaks reported in figure 1, we can hypothesize that emission at 422 nm could be related to Eu2+ placed inside Eu2P207 structure. For what concerns the broad peak at around 480 nm, probably it is formed by the convolution of Eu2+ emission placed inside the minor phases whose relative concentration respect to the prevalent phase increases in sample BP3. yirradiation tests using dose of 5.6x105 Gy on these three samples showed that no variations were observed both in the emission spectra and in XRD spectra, indicating that radiation did not induce any improvement on the optical properties of the analyzed samples. Regarding sample S2, doped with Eu3+, we show the emission spectrum after 532 nm_excitation in Figure 2a. Also in this case y irradiation does not induce any variation in the spectrum confirming the above results . From the applicative point of view the most interesting Eu3+ emission band is the one centered at around 612 nm corresponding to the 5 D 0 -» 7 F2 transition, where the greatest part of energy is emitted. The S2 decay kinetic at 612 nm turned out to be composed of a single exponential component indicating the absence of energy migration between Eu3+ ions inside the matrix. An interesting effect induced by y radiation is the modest decay time shortening

887

from to 5x10"" up to 4.8xl0"4s after the 5.6xl05 Gy irradiation dose, which can be explained by considering the formation of radiation induced defects which may favor the appearance of a non radiative recombination channel. ~ 6,0x10'-

a

t|

4,0x104-

8

i

,II \

2,0x10'-

imrnes

|

J

JV 0,0600

605

610 615 620 Wavelength (nm)

625

630

0,001 „ m e (ms)

'

0,002

Figure2. Emission spectra of sample S2 (a); decay kinetic at 612 nm before and after y ray irradiation (b).

4. Conclusions Summarizing our work, we have produced some blue and red emitting phosphors for application in UV LEDs. Luminescence measurements evidence that there is a definite relationship between crystal phase structure and emission behavior. Irradiation turned out to not modify luminescence and structural propertied of the powders with the only exception of a modest time decay shortening of the Eu3+ emission at 612 nm, probably due to radiation induced production of non-radiative defect centers. References 1. 2. 3. 4. 5. 6. 7.

C.R. Ronda, H. Nikol, J. All. Comp., 275 (1998). T.Tamura, T.Setomoto, and T.Taguchi, J. Lumin. 87,1180 (2000). P.Joung, Kim.Chang,P.Hee.Appl. Phy. Lett. 84,1647(2004). Y.Q.Li et al., Chem. Mat. 17, 3242 (2005). P.Yang, G.Yao, J.Lin. Opt. Mater.26, 327 (2004). J.E. Munoz, Phys. Rev. B, 42, 11339 (1990) J.Park, C.Han, S.Choi. Electro-Chem. Solid-State Lett.5, H l l (2002).

This page is intentionally left blank

Space Experiments Organizer: S. Volonte X. Barcons A. Basili P. Bastia P. Bobik M. Cherry

L. Foggetta M. Gervasi F. Giordano M. Hareyama D. Hudson F. Loparco V. Malvezzi A. Mozzanica S. Raino S. Russo B.F. Schutz F.R. Spada R. Sparvoli M. Uslenghi

The Extreme Universe PAMELA Data Acuisition System AGILE MCAL, the Mini-Calorimeter Primary Helium CR Inside the Magnetosphere: a Transmission Function Study Lanthanum Halide Scintillators and Optical Fiber Readout for X-Ray/Gamma-Ray Astronomy and National Security Applications The AGILE Silicon Tracker: Construction and Calibration Results Ions Abundance Close to the Earth Surface: the Role of the Magnetosphere Design of a Silicon Transition Radiation Detector (SiTRD) for Accelerators and Space Applications The Origin of Helium-3 isotope Enhancement in the Magnetosphere Observed by TSUBASA Satellite The Microscope Mission and Pre-Flight Performance Verification Performance of the Integrated Tracker Towers of the GLAST Large Area Telescope Performance of Neutron Detector and Bottom Trigger Scintillator of the Space Instrument PAMELA High Granularity Silicon Beam Monitors for Wide Range Multiplicity Beams Environmental Test Activity on the Flight Modules of the GLAST LAT Tracker The Time of Flight Detector and Trigger for the PAMELA Experiment in Space Fundamental Physics in ESA's Cosmic Vision Plan The AMS-02 Electronics System Launch in Orbit of the Space Telescope PAMELA and Ground Data Results CZT Detector Development for New Generation Hard-X/Gamma-Ray Astronomical Instruments 889

This page is intentionally left blank

THE EXTREME UNIVERSE

X. BARCONS Instituto

de Fisica de Cantabria (CSIC-UC), E-39005, Santander, Spain E-mail: barcons @ifca. unican. es

The study of matter under extreme conditions in various places of the Universe has been boosted over the last decades, thanks to t h e availability of X-ray and 7-ray observatories in space. For the future decades, ESA's Cosmic Vision 2020 plan puts forward the study of black holes and neutron stars, the assembly of baryons in the cosmic structures, and the growth of massive black holes in the centres of galaxies. I finally discuss the space missions necessary to make progress in these topics.

1. Introduction Extreme physical phenomena occur throughout the Universe. From the Earth's magnetosphere all the way through the most distant quasars, not forgetting about the very early Universe, there are places and times in the Cosmos where the physical conditions are as extreme (or even more) than in the most challenging experiments ever devised. Extreme physical conditions lead very often to energetic particles, whose energy we see released as high-energy electromagnetic radiation (X-rays and 7-rays). Observing the Unverse at the shortest available wavelengths allows us to find those places where physical conditions are truly extreme and to study the behaviour of matter under these circumstances, even at very long distances. Astronomical observations in the high-energy domain are currently enjoying what could be called as a golden age. The most powerful X-ray observatories in orbit launched in 1999, NASA's Chandra) and ESA's XMMNewton, have been recently joined in orbit by JAXA's Suzaku (formerly known as ASTRO-E2). At 7-ray energies, ESA's INTEGRAL is performing smoothly since October 2002. These and other space X-ray observatories are complemented by a number of ground-based facilities sensitive to very high 7-ray energies, which despite their limited sensitivity, are opening a new window to the Universe in the TeV photon range. 891

892 The Cosmic Vision 2015-2025 of the European Space Agency (ESA) exercise has provided an opportunity to revise the most challenging and exciting science that can be performed in the realm of the most extreme Universe in the decade after next, when the current facilities in orbit have been fully exploited. I discuss these science goals and also possible space missions to make progress on them.

2. Matter under strong gravity The Universe provides us with places where the environmental conditions can be as extreme, or more, than in any existing or foreseen laboratory. The fourth of the fundamental forces of nature, gravity, has a tiny direct effect on the interactions between elementary particles that are probed in laboratory experiments. Gravity can, however, can be dominant under cosmic conditions. Indeed, the strongest gravitational fields in nature are those around neutron stars and black holes (BH). How does matter behave under the influence of gravity in the realm of General Relativity, well beyond the post-Newtonian or any similar perturbative deviation from Newtonian gravity, can only be studied by observing the immediate vicinity of black holes and compact stars. General Relativity (GR) is indeed the best ever formulated theory of gravitation. GR predicts deviations from Newtonian gravity that have been confirmed and measured very accurately in the weak field limit. In the strong field limit, GR predicts a suite of phenomena which are no longer small perturbations of Newtonian gravity, and which reflect a strongly curved space-time: strong gravitational redshift, Lense-Thirring precession, etc. These phenomena need the gravity of a black hole or a neutron star to be revealed. Strong variations in these strong gravitational fields will produce gravity waves that will be the ultimate probes of the behaviour of space-time closest to the event horizon. Accretion of matter around a black hole or a neutron star can also probe the curved space-time within a few Schwarzschild radii. High-energy (X-ray and 7-ray) radiation from the

innermost regions of the accreting material can be used to reveal GR effects in the strong field limit, and to test this theory where its most spectacular effects are expected. One of the effects of strong gravity on the high-energy emission from accretion disks is the broadening of X-ray emission lines. The Fe K a line (at 6.4 keV for neutral Fe), which has been and will continue to be the best handle for this, likely arises from fluorescence as the accretion disk

893

is irradiated by the primary X-ray source. In the innermost parts of the accretion disk, where most of the power is produced, this line is broadened by various effects [1], which include the Doppler effect, relativistic beaming, and gravitational redshift due to the nearby presence of the black hole. The resulting line profile is skewed towards softer photon energies due to this last effect. The specific profile of the Fe line depends on many parameters (disk inclination, among others), but the importance of the low energy tail is dictated by how close the reflecting material orbits around the black hole. This, in turn, depends on the spin of the black hole, because the Innermost Stable Circular Orbit (ISCO) decreases as the spin parameter of the Black Hole a = J/Mc increases [2]. For a non-rotating (Schwarzschild) black hole, Risco = 3.Rs, where the Schwarzschild radius is Rs = 1GMj
894

does not occur in a plane, nor does it describe closed orbits. Regardless on whether the compact object rotates or not, there is a peri-astron precession. If the compact object rotates, then there is Lense-Thirring precession of the orbital plane (i.e., the plane of the orbit precesses around the spin of the compact object). The very detection of the ISCO around a BH or a NS is in itself a confirmation of General Relativity. If the mass of the compact object were known, its spin could then be deduced. Kilo-Hertz QPOs, possibly related to the ISCO, have been indeed detected in a number of Galactic Black Hole candidates and Neutron Stars, but the interpretation of the various peaks in the Fourier spectrum is not straightforward [8]. The best possible approach would be, however, to have the possibility to study orbits individually, i.e., to have a very high-throughput X-ray detector capable of reliably measuring high-accuracy fluxes every fraction of a milli-second orbit.

3. "Inside" neutron stars What happens to matter when it is squeezed to supra-nuclear densities?. Neutron stars are amongst the densest objects in the Universe, with their core density being 5 to 10 times larger than nuclear. From the phenomenological point of view, laboratory ion-collision experiments can partially approach the environmental conditions in the core of a NS, but at a different "temperature and with a different proton fraction (typically very small in NS, see [9] for a review). The Equation of State (EoS) of the core of a NS is not known. Matter at that density could be well rich in pion or kaon condensates or it could consist of a soup of unconfined quarks (see [10] for a collection of models). Even more, it could consist of "Strange" quark matter. While it is not possible to probe directly the EoS at these densities, the mass and radius of the resulting NS are strongly affected by it. X-ray observations that can help measuring M and R of neutron stars. Following with the timing analysis discussed in the previous section, the detection of the frequency of orbital motion from kilo-Hertz QPOs places useful constraints on the M-R parameter space. This comes simply from the fact that Rorb > Risco a n d that R < RiscoAn alternative method to constrain this diagram, even for isolated neutron stars, has been put forward in [11] in the case of the NS EXO 0748-676. Photospheric absorption lines in its X-ray spectrum are strongly gravitationally redshifted by the gravitational field at the NS's surface (z = 0.35

895

in this case), providing a direct measurement of M/R. More detailed information on the EoS of matter at supranuclear density could be gained with high signal to noise high spectral resolution X-ray spectroscopy. The Stark effect due to the electric field at the NS surface would broaden the photospheric absorption lines with FWHM oc M/R2, whose measurement would break the degeneracy between M and R to sufficient accuracy to single out an EoS.

4, Assembly of baryons and the creation of heavy elements The Universe began some 14 billion years ago in the so-called Big Bang. Today, ordinary, baryonic matter represents ~ 4.5% of the total content of the Universe, and almost half of it is in an unknown location. About 23% is made of Dark Matter (DM), which binds galaxies and clusters of galaxies via gravitational attraction (the only manifestation of DM so far). The bulk of the Universe is contributed by an even more exotic, extremely uniform component, Dark Energy (DE), which in fact shapes the geometry of the Universe and acts as a peculiar "repulsive force. In principle, only the 4.5% of the baryons is all we can observe via electromagnetic radiation. The DM potential wells of groups and clusters of galaxies, the largest gravitationally bound structures in the Universe, have virial temperatures which imply X-ray temperatures for the baryons trapped in them. Observing and analyzing clusters of galaxies at high redshift is the best handle to trace the assembly of baryons onto bound structures. Although gravity is what brings the baryons together onto clusters, it is also known that other processes (like heating by Supernovae or Active Galaxies) provide an "entropy floor" to these baryon assemblies. If these processes were well understood, clusters of galaxies could be used as prime cosmological tracers. The amount of dark energy could, for example, be measured by observing the gas baryon to dark matter fraction in clusters as a function of redshift [12]. The equation of state of dark energy (w = PDE/PDEC2) could also be measured if the cluster number counts as a function of redshift and/or their spatial clustering could be determined [13]. But almost 50% of the baryons do not reside in gravitationally bound structures. The Warm and Hot component (T ~ 105 — 107 K) of the Intergalactic Medium (WHIM) is likely to contain the remainder. Although a few first detections of WHIM filaments have been achieved with current instrumentation [14], there is much to learn and discover about the heating

896 of the WHIM and the baryon budget in the Universe. Finally, the chemical enrichment of the Universe is also a subject related to the extreme Universe. This is a process that we see in our immediate vicinity in Supernovae which can be studied in detail at X-ray and 7-ray energies. At the highest redshifts, the evolution of the chemical abundances, and therefore the processes that gave rise to the various elements, can be seen in the emission lines seen in groups and clusters, via high resolution X-ray spectroscopy.

5. The growth of massive black holes in galaxy centers One of the very first discoveries of X-ray Astronomy was the Cosmic X-ray Background (XRB, [15]). This energetic radiation that fills the Universe is known today to be the integrated radiation produced by accretion onto supermassive black holes (SMBH) along cosmic history. These grown SMBHs are those that we see today in the centers of virtually all galaxies and that comprise ~ 0.4% of their bulge mass [16]. According to the AGN unified models for the XRB [17], most of this accretion occurs in obscured mode, and therefore its direct detection can only be achieved in hard X-rays. This general qualitative picture of the XRB being the echo of the growth of the super-massive black holes opens a number of questions. The first one is how SMBHs (or their seeds) were born and whether they can be detected or not at the time of birth. Perhaps Gamma-ray bursts are related to this process. The second one is how they grow from their probably small initial mass to their very large masses that we see today (> 109 M Q ). Accretion is likely to be mostly responsible for this [18], but mergers and tidally captured stars can also contribute. Finally there is the question on how the birth and growth of SMBHs relates to the formation of galaxies and their stars. We have now clear clues that there is a link between both processes, but how exactly the feedback works is not yet understood.

6. How to make progress: missions for the future Most of the above scientific questions for the decade after next would require a large X-ray observatory-class mission. The effective collective area should approach 10m 2 at 1 keV (for early Universe and cluster studies), and 1 — 2 m2 at 6 keV (for Fe line studies). This mission should be equipped with suitable instrumentation: high-spectral resolution, large field of view for surveys, high count-rate capability for timing studies. ESA and JAXA are

897

studying a mission called XEUS a which would fulfill most of the above goals. Steep variations of strong magnetic fields (such as those produced by BHs) will be best detected by the gravitational wave observatory LISA under certain circumstances. Other ancillary missions would also be very beneficial for some of the above goals, including a far-infrared observatory or a focussing gamma-ray observatory. Acknowledgments Financial support was provided by the Spanish Ministerio de Educacion y Ciencia, under project ESP2003-00812. References 1. 2. 3. 4. 5. 6.

Fabian A.C., Rees M.J., Stella L., White N.E., 1989, MNRAS, 238, 729 Bardeen J., Press, W. & Teukolsky S., 1972, ApJ, 178, 347 Tanaka Y., et al., 1995, Nat, 375, 659 Wilms J., et al., 2001, MNRAS, 3218, L27 Vaughan S., Fabian A.C., 2004, MNRAS, 348, 1415 Streblyanska A., Hasinger G., Finoguenov, S., Barcons X., Mateos S., Fabian A.C., 2005, A&A, 432, 395 7. Brusa M., Gilli R., Comastri A., 2005, ApJ, in the press (astro-ph/0501542) 8. van der Klis M., 2004, in Compact stellar X-ray sources, Lewin & van der Klis (eds), Cambridge University Press 9. Lattimer J.M. & Prakash M., 2004, Science, 304, 536 10. Lattimer J.M. & Prakash M., 2001, ApJ, 550, 426 11. Cottam J., Paerels F., Mendez M., 2002, Nat, 420, 51 12. Allen S.W., Schmidt R.W., Ebeling H., Fabian A.C., van Speybroek L., 2004, MNRAS, 353, 457 13. Griffiths R., et al., 2004, Proceedings of the SPIE, 5488, 209 14. Nicastro F., et al, 2005, Nat, 433, 495 15. Giacconi R., Gursky H., Paolini F.R., Rossi B., 1962, Phys. Rev. Lett., 9, 439 16. Ferrarese L., Ford H., 2005, Space Sci. Rev., 116, 523 17. Comastri A., Setti G., Zamorani G., Hasinger G., 1995, A&A, 296, 1 18. Marconi A., Risaliti G., Gilli R., Hunt L.K., Maiolino R., Salvati M., 2004, MNRAS, 351, 169

a

http://www.rssd.esa.int/XEUS

PAMELA DATA ACQUISITION SYSTEM ALESSANDRO BASILS INFN Rome II, University of Tor Vergata, Rome, Italy The PAMELA experiment is a satellite-borne apparatus devoted to the study of antiparticle component of cosmic rays. The instrument core is a permanent magnet surrounded by several instruments with different issues. Besides the TOF (Time-OfFlight), spectrometer, calorimeter, an anticoincidence system and a neutron detector, the experiment has an on-board computer responsible of the whole acquisition and housekeeping. In this work we will show the Data Acquisition flux for PAMELA, explaining the read-out mechanism from the Front End and the data-storing trough the 2x2 GB Solid State Mass Memory and the Resurs VRL system toward the ground station.

1. Introduction PAMELA (a Payload for Antimatter Matter Exploration and Light-nuclei Astrophysics) is a satellite-borne experiment designed to explore the charged particles in cosmic rays, with particular attention to the antimatter component. The detector is a combination of different subdetectors which have different characteristics in order to identify, with high accuracy, particles within the experiment energy range (80 MeV - 190 s-GeV for antiprotons and 50 MeV 270 GeV for protons). The telescope is based on a Magnetic Spectrometer, an Anticounter System, a Time-of-Flight System, an Electromagnetic Imaging Calorimeter, a bottom scintillator called S4 and a Neutron Detector [1]. The spectrometer is made of a permanent magnet (with a quasi-constant magnetic field in its cavity of 0.4 T) and a silicon tracker. The cavity is 445 mm tall and a cross-section of 132 mm * 162 mm which gives the geometric factor of the apparatus that is 20.5 cm2xsr. The silicon tracker is made of six planes of double-layer, double-metal and AC coupled microstrip silicon detectors of 300 um thickness and 25 um strip pitch, with a total number of strips equal to 6144 per plane (total number of channel equal to 36864) [2]. The magnet is surrounded by the Anticounter System, needed to exclude particles that are out of the telescope geometric factor. It is made of 9 scintillator planes, read by phototubes [3].

For the Pamela Collaboration

898

899

The Time-of-Flight System provides the trigger to the experiment; is made by three scintillator planes and it will give information about the energy loss in the detector (dE/dx) and the absolute Z of the particles. These data, integrated with the trajectory, will give the velocity of the particle crossing and allows the rejection of albedo particles [4]. The Calorimeter is a 44 silicon planes detector, with 22 planes of tungsten in between to generate showers and measure the loss of energy. The structure is Si-x/W/Si-y (with a 380 urn thick and 2.4 mm wide strip in a 8x8 cm2 ministrip silicon sensors arranged 3x3 in each plane, for 4416 total number of channels) in order to reconstruct the trace of the particles and their nature, is estimated that the protons (antiprotons) and positron (electrons) are distinguished at 95% retaining efficiency with a rejection factor of 104 [5]. S4 and the Neutron Detector are needed to extend Pamela energy range and to capture the showers that come out of the Calorimeter. The Neutron Detector, triggered by S4, will improve Pamela's capabilities of electron-proton discrimination up to higher energies (10 u -10 13 eV). Due to its orbital characteristics (elliptical and semi-polar, with an inclination of 70.4 and an altitude varying between 350 and 600 km) and its geometric factor (20.5 cm2xsr) it is expected an average trigger rate of 12 Hz. The amount of data for each event is roughly about 6 KB. Redundancy has been used to avoid the most critical points of the experiment and to eliminate all the "single point failure". In the following paragraphs will be shown what the system is composed by and which is the data flux of any event in Pamela. 2. Event acquisition The data acquisition is based on a data reduction philosophy, getting data from every subdetector starting from a single central unit (PSCU [6]: Pamela Storage and Control Unit). As it is shown in picture n° 1, there is a chain of data collection points: 1) PSCU is the central unit which starts the data acquisition. 2) IDAQ multiplexes out all the PSCU commands to the Front-End electronics. 3) DSP boards are the first computational step in the data reduction. 4) Front-End boards are the analog to digital read-out boards. All the data are stored in a 2+2 GB solid state mass memory in the PSCU and downloaded to the host satellite where there is a 30 GB unit for data storage (VRL: Very high speed Radio Link). The radio link towards the Data Downlink

900 Station (near Moscow) is active 6 times a day and has a transmission window of 200 seconds at a speed of 153.6 Mbps. With these parameters has been designed the Pamela Mass Memory in order to guarantee the maximum flux of data without any losses. Trigger board

tracker DSP calorimeter DSP

1>S --{A/D] A/D

PSCU

:

IDAQ

:

DSP

^0 ND S4

Figure 1. Electronic chain in the data reduction of Pamela. The system has a global "busy" signal that is the only veto to the acquisition: if the IDAQ is busy (processing the event) no others triggers can be delivered. On the other hand, if the IDAQ is not busy, no commands can be sent to any subsystem. At startup the system is "busy", so that all the configuration settings can be done before starting acquiring data. The handshake protocol between PSCU and IDAQ allows setting a fixed number of "data acquisition commands queues" in the interface DMA, so that all the acquisitions have practically no cpu-time consuming. The first command of each acquisition command queue is a "release busy" command to the IDAQ, which will wait, without acknowledging the next command, the first trigger; by this command the trigger system can start a trigger pulse as soon as the first particle generates the trigger mode defined (ex. SI or S2 or S3) in the acquisition mode. All the subsystems will receive the trigger signal with different timing constraints, chosen considering the forming time of each dedicated readout electronics. As the trigger gets to the IDAQ the busy is set and a timeout of 3.5 ms starts to run on this board, in order to allow

901 compression algorithms on the DSP boards; only after this time the acknowledge to the pending command is sent back to the PSCU and all the commands start to reach every DSP board. Once all the "read event" commands has got to every single subdetector and all the data are collected in the PSCU, the command queue starts again. Monitoring the "busy" signal, the trigger board can calculate the "life-time" and "dead-time" of the experiment. The Idaq protocol towards the CPU has a parallel link with a single handshake at a speed of 2 MB/s while the serial protocol towards the Dsp boards is a 10 Mbps link. Once the packet is stored in the Mass Memory of the PSCU, there is a command from the host satellite to the Pamela system to download the data towards the data storage unit of the satellite itself. The interface unit that will handle the transmission between the PSCU and the satellite is called VRL adapter and which will guarantee a 12 MBps data link. 3. IDAQ The Interface Data Acquisition board is an FPGA [7] based board, which has the major issues of commands redirection, protocol handling and trigger-busy managing in order to keep the acquisition going, secondly it is equipped with a DSP (Analog Device ADSP2187L), a RAM memory (CY62146V 4Mb Static RAM CYPRESS) and a FLASH memory (Am29LV800B 8Mb, AMD) in order to implement a second level trigger to cut out bad data from the acquisition if the trigger rate is too high and the "rubbish" data are too much (it is estimated that more then 80% of data will not be useful). The command queue needed to implement a second level trigger is different from the one used as default and this choice is written in the configuration parameters on the CPU software. All the software parameters can be set from the ground station and the same "second level trigger routine" can be loaded via the uplink. All the communication links are LVDS based and the strobe-acknowledge protocol towards the PSCU gives to the Idaq the complete control of the data acquisition. Once the command queue is started there is no way to stop the acquisition unless a reset signal is sent to the system and the DMA is stopped. In the other side, towards the DSP boards, it is implemented a serial protocol to reduce the number of connections and an hardware timeout of 2 (is to close the communication. The data-strobe protocol fulfills the requirements of speed and power consumption, moreover it fixes a standard interface between the different boards and makes the hardware development faster. The RAM memory is used to collect data and allow any calculation wiht the DSP; there are two ram chips controlled by a dedicated FPGA (A54SX32A),

902

which is demanded to implement a CRC protection on the data stored, so that any single bit error is corrected and a double one is recognized. The same CRC encoding is implemented on the Flash chips which will contain the DSP programs and a backup of some configuration parameters (loaded from the PSCU). The DSP is set in IDMA configuration and from the command queue point of view all the addressing overlays are handled by the controller which shows DSP memory as a linear 32 KWord (overlay equal to 1, 4 and 5). In the Idaq architecture, the Main controller is master and all the peripherals (ram, flash, dsp and external links towards the dsp boards) are slaves and this "acquisition philosophy" has simplified radically both the hardware and the software organizations. 4. PSCU The Pamela Storage and Control Unit, produced by Laben, is based on an ERC32 spare V7 processor on which has been installed a real time operative system (RTEMS) and its structure is divided in five logical modules: 1) 2) 3) 4) 5)

CPU module Memory modules PIF: Pamela Interface module HKU: housekeeping unit DC/DC: power supply

The CPU module hosts the processor, a SRAM 1Mb x 32 EDAC Protected, a Boot PROM, an EEPROM 256Kx32 EDAC Protected, a MIL-Std-1553 Bus Controller/Remote Terminal Function with its associated 64K x 16 RAM buffer, a Glue Logic ASIC including PCMCIA Bus Controller, Parallel System Bus interface (SBus90) called CRIMEA and developed by Laben. The 1553 bus is the interface with the satellite and has a dedicated buffer memory to manage its commands which are the highest priority ones. The Memory modules are organized to provide 16 Gbits storage capability, divided in eleven modules, independently monitored with a current absorption checker to avoid any latch-up faults. The ASIC designed to handle the access to the memory chips (SDRAM 8 MByte per unit), called DRAMMA (DRAM Manager ASIC), has a Reed-Salomon coding process and allow a maximum throughput of 750 Mbps. The PIF unit is a dedicated interface board with three main issues: 1) Serving die communication with the Idaq through a DMA controller 2) handling the interface with the Mass Memory 3) providing the interface with the VRL adapter to download the data

903 Moreover it is implemented the interface with the PCMCIA bus to communicate with the processor. An FPGA (Actel RT54SX32S) is used to absolve its functions and a memory unit (DRAM 256 Mb x 8) is used for buffering command and data paging. Eventually the Housekeeping module has two serial links RS422 to communicate with the Kayser Housekeeping Board, which handles the connections with different subsystem (power supply control board, High Voltage control board, Tracker relays board, Idaq, S4 settings, Tracker sensors boards) to monitor the status of the system, plus a large number of ADC channel for analog monitoring (16 voltages, 16 sensors), contact closure inputs and differential bilevel inputs. This module is also equipped with 24 high level pulsed commands (26 V). The DC/DC module is needed to provide all the voltages used inside the different modules. Is based on a push-pull topology and provides both common and differential mode. It is secured with latches for overloading and for output overvoltage, while has a non latching protection for input undervoltage. The module can be monitored with an analog output for the secondary voltage, a status bit and a thermistor. 5. Conclusions The data acquisition system has been developed to fulfill requirements on the data handling and for its reliability. Moreover the housekeeping system and procedures will guarantee a full fault recovery. Pamela will be launched at the beginning of 2006 and it will last for three years, acquiring a very large amount of data and enriching the whole scientific community with better results and maybe new aspects of our universe.

References 1. M. Bongi et al., IEEE Trans. Nuc. Sc. Vol 51, n 3 (June 2004). 2. O. Adriani et al., Nuc. Instr. & Meth.. Vol 511, Issue 1-2 (2003). 3. M. Pearce et al., ICRC Tsukaba, Japan.August 2003 5. M. Boezio et al., Nuc. Instr. & Meth. Vol 487, Issue 3 (2002). 4. G. Osteria et al., Nuc. Instr. & Meth. Vol 518, Issue 1-2 (2004). 6. Laben Interface control document, rel 3 (2004) 7. M. Boscherini et al., Nuc. Instr. & Meth. Vol 513, Issue 1-3 (2003). (2001).

AGILE MCAL, the MINI-CALORIMETER PAOLO BASTIA, JENS MICHAEL POULSEN, FRANCO MONZANI, PAOLO RADAELLI, PAOLO MARCHESI, ALCATEL ALENIA SPACE ITALIA S.p.A. - LABEN", S.S. Padana Superiore 20090 Vimodrone, Milano, Italy

290,

CLAUDIO LABANTI, MARTINO MARISALDI, FABIO FUSCHINO, ANDREA BULGARELLI INAF-IASF sezione Bologna, Via P. Gobetti 101, Bologna,

Italy

AGILE is a scientific mission dedicated to gamma-ray astrophysics in space, and the mini-calorimeter MCAL is one of four detector systems on the satellite. The MCAL instrument is sensitive in the energy range: 300 keV - 100 MeV. It has two main functions: one autonomous mode for detection of impulsive cosmic events and the other as "a slave" supporting the energy measurements of the pair-conversion tracker. The AGILE Small Mission is funded by the Italian Space Agency (ASI), and the INAF-IASF section at Bologna has the scientific responsibility for MCAL. LABEN develops the MCAL instrument with its detectors and electronics. This paper gives an overview of the detectors on AGILE, and then it gives details on the design of MCAL, and finally we report on the tests at instrument level.

1. Introduction AGILE is a Small Scientific Mission dedicated to gamma-ray astrophysics in space. It is funded by the Italian Space Agency (ASI), and the INAF-IASF section at Bologna has the scientific responsibility for MCAL, the minicalorimeter instrument on AGILE. Alcatel Alenia Space Italia Spa - LABEN, has a contract for the mini-calorimeter, and several other units on the Integrated Payload, IPL. In particular LABEN designs and develops the complete MCAL instrument, including the detectors, as well as its electronics. AGILE detectors are sensitive in two energy ranges: 1 5 - 4 5 keV and 30 MeV - 30 GeV with imaging capabilities, and 300 keV - 30 MeV as monitor. First in this paper we shall give an overview of the photon detectors on AGILE, then we shall give details on the design of the Mini-Calorimeter, and finally we report on the performance tests. a

Our company WWW pages at: http://www.laben.it/index.html

904

905

Figure 1. Photo of AGILE "core" detectors comprising the Mini-calorimeter (bottom), the Tracker with preprocessor electronics, and the X-ray Imager, Super-AGILE (top). The height is 59 cm.

2. Overview of AGILE detectors Here we give a short description of the four detector systems on AGILE. They are a combination of X and gamma-ray detectors mounted on the top of a «shell» structure, and characterize the Integrated Payload (IPL). The Silicon Tracker (ST) is a gamma-ray pair converter and comprises 12 planes, each of which has two crossed layers of silicon micro-strips (total 36,864 detectors) for position directional measurement of charged particles. The upper 10 layers are interleaved with tungsten sheets for conversion of gamma rays in charged particles. The ST instrument has a good sensitivity to photons in the range 30 MeV to 30 GeV. The Mini-Calorimeter (MCAL) is a gamma-ray instrument based on 15 + 15 Csl crystal bars with photo-diode read-out and is sensitive in the range 300 keV to 100 MeV. It contributes to measure the energy of gamma rays in combination with the ST, and independently detects photons of impulsive cosmic events. MCAL is placed on a mechanical structure, below the ST (see fig.l). The Anti-coincidence System (AC) is an active shield based on plastic scintillators with photo-multiplier read-out [1]. The AC detectors completely surround the ST and the MCAL on five sides, and it is mounted on the top of the "shell" structure. The combination of these three instruments (ST, MCAL and

906

AC) operates together as a gamma-ray imaging detector (GRID mode) with a good angular resolution (0.6° at 1 GeV), and a large field-of-view (approx. 1/5 of the whole sky) [2]. Finally Super-AGILE is an imaging X-ray detector with 4 detector assemblies each associated to a "coded mask", and it is sensitive in the range from 15 to 45 keV. Super-AGILE is mounted on the top of the ST Detectors (fig. 1), and placed below the AC Shield. The total height of the IPL is 1.05 meter and its weight is 189 kg.

Figure 2. The Mini-Calorimeter Detector showing the upper 15 Crystal Bar Assemblies, and two boxes with pre-amplifiers.

3. Mini-Calorimeter Design The major units of the Mini-Calorimeter subsystem are 30 Crystal Bar Assemblies, four preamplifier boxes and one MCAL Acquisition Box with digital electronics. These units are mounted on a dedicated frame structure (see fig. 2). The Crystal Bar Assemblies form two planes with 15 + 15 crossed bars. The Acquisition Box interfaces its signal and data lines to the Payload Data Handling Unit (PDHU) [3]. The design with independent crystal bars was chosen to provide a modular structure of the detection plane, and allows for substitution of a single detection element, if necessary. Each Crystal Bar Assembly contains the proper X-ray detector, a scintillator crystal (Csl-Tl) with the dimensions 375 x 23 x 15 mm (1-w-h), which is optically coupled at its ends to two photodiodes. The crystals are treated for optimum optical properties, and thus are slightly grinded on the surfaces, then wrapped with a reflecting paper

907

(VM2000) and kapton tape. The housing of the Bar Assembly is made of carbon fiber material in order to be light. The electronics in the MCAL Acquisition Box elaborates the signals of the 60 photodiodes/pre-amplifiers on a single circuit board (415 x 415mm), which contains 5000 components. Basically there are two independent circuits: the "burst chain" and the "GRID chain": The former handles self-triggering signals in order to detect impulsive cosmic events, and in the PDHU these data are used to build several rate-meters (RMs) based on different geometrical regions, energies and various integration times. A burst search algorithm will compare the contents of the RMs with background fluctuations to detect impulsive events. The "GRID chain" collects all bar signals following an external trigger generated by the tracker electronics. The power consumption of the Acquisition Box is 3.6 Watts, and the total power consumption of the MCAL front-end electronics is 4.5 Watts. 10000 8000 26000 c S

4000 2000 U I lilrf i i i i i j l j i f c l j i f c l U.D U.O I I.* !.•• 1.D

1.0

Energy (MeV)

Figure 3. 22 Na reconstructed spectra detected by a MCAL bar. The energy resolution is 13 % FWHM at 1275 keV.

4. MCAL Test Results A large test session was carried out to characterize the MCAL "stand alone". In particular the functional tests aimed to prove that the instrument could be integrated on the payload. The detector can give information both on the energy and the position of interaction of an absorbed photon by combining the two signals from the PDs of a bar [4]. The two fundamental parameters of the Crystal Bar Assembly are the total light output (electrons measured per unit energy deposit at a fixed distance from a PD), and the light attenuation coefficient (the inverse of the distance from a PD required to reduce the light outputs by a factor 1/e, assuming a simple exponential law). The energy and the position information of a detected event can be obtained only by using dedicated algorithms based on the fundamental

908 parameters, and the accuracy in the reconstruction depends strongly on the accuracy of these parameters. To obtain the fundamental parameters the bar were illuminated with a collimated 22Na radioactive source in different positions. Light outputs and attenuation coefficient were measured on all flight bars, and found to be respectively 20 electrons/keV (average value near a photo-diode) and 0,025 0,035 cm"1. Figure 3 shows a 22Na reconstructed spectra detected by a MCAL bar when characterized. In this case the source is not collimated along the bar direction. The energy resolution is 13 % FWHM at 1275 keV. For collimated measurements, a 1.6 cm standard deviation in position reconstruction is obtained at 1275 keV. Figure 4 shows a reconstructed track of a muon, which generates events with high energy deposits on all the bars across one MCAL plane. It can be seen that the position reconstruction capabilities of the MCAL are quite good. • • •I

28 26

• • •

S24 •D

S22 u20 18

16 -30

1 . , .

-20

• • •

•

• • • . 0, . . . .10, « . . . , . . , .

-10 Position (cm)

Figure 4. Reconstructed muon track.

The MCAL has passed functional testing, and is mounted on the payload ready for testing at system level. References 1. A. Bonati et al. in "Third Rome Workshop on Gamma-ray Bursts in the Afterglow Era", ASP Conf. Series, Vol. 312, 513 (2004). 2. M. Tavani et al. in "Proc. of the 4th AGILE Science Workshop", Agile Publ. n. 02, 3 (2004) 3. A. Argan et al, IEEE NSS Rome Conf. Proc. (2004). 4. C. Labanti et al, SPIEProc, Vol. 5488, 700 (2004)

P R I M A R Y HELIUM CR I N S I D E T H E M A G N E T O S P H E R E : A TRANSMISSION FUNCTION STUDY

P. BOBIK1, M.J. BOSCHINI^M. GERVASI2'3, D. GRANDI2, K. KUDELA 2 AND P. G. RANCOITA2 Institute of Experimental Physics, Kosice, Slovak Republic Instituto Nazionale di Fisica Nucleare, Milano, Italy Physics Department - University of Milano Bicocca, Italy E-mail: [email protected]

We have applied the Transmission Function approach to evaluate the primary Helium CR spectrum inside the magnetosphere. We have evaluated the trajectories of simulated alpha particles through the Earth magnetic field by backtracing up to the magnetopause or down the atmosphere. In this way we have computed first the trasmission function for a particular position inside the magnetosphere; then we have evaluated the spectrum of primaries, propagating a energy population of GCR. The analysis has been restricted to the geographic coverage of the AMS-01 detector (between 51.6 and -51.6 degrees) and altitude ( 400 km) and divided in geomagnetic zones. Finally we have compared our results with the measured data.

1. I n t r o d u c t i o n Isotropically distributed particles outside of Earth's magnetosphere are inside magnetosphere affected by geomagnetic field. Calculation of particle trajectories in models of geomagnetic field [1],[2] are widely used for finding the distribution of these particles inside the magnetosphere. Transmission function [3] is a statistical approach based on trajectories calculations, used to evaluate particle intensity at selected points or regions in the magnetosphere. Cosmic ray spectrum at 1AU consist mainly of protons and Helium nuclei [4]. There were many measurements of CR Helium spectra inside the magnetosphere in the previous decades. Most precise were balloon measurements and AMS-01 spectrometer measurement during precursor flight of AMS project in June 1998. Precision of balloon experiments like BESS [5],[6], CAPRICE [7] and CREAM [8] is limited because of the effect CILEA, Segrate (Mi), Italy

909

910

of residual atmosphere, however, these are most precise measurements yet available except AMS-01 spectrometer measurements. 2. Method 2.1.

Backtracing

Backtracing method [9],[10] is based on the inversion of charge sign (Zq) and velocity vector (v) in the equation of motion for a particle with relativistic mass m in the magnetic field B : dv

„

,

„,

m— = Zq[v x B\

(1)

Such inversion in equation (1) means that an antiparticle with ridity Sft generated in a selected point and with initial direction has in magnetic field the same trajectory like a particle with same rigidity and direction coming to this point. In all simulations presented in this article we used a combined model of geomagnetic field B. The total magnetic field B consists of the external geomagnetic field described by the model Tsyganenko 96 [11] (see also the web site: http://nssdc.gsfc.nasa.gov/space/model/magnetos/databased/modeling.html) and the internal geomagnetic field described by the IGRF model (see web site: http://nssdc.gsfc.nasa.gov/space/model/magnetos/igrf.html) : B = B{nt + Bext. If the particle trajectory started from a selected location inside the magnetosphere crosses the magnetopause [12] at dayside of magnetosphere or a sphere with radius 25 Earth radii at nightside of magnetosphere we call it allowed trajectory. Trajectory crossing a sphere with altitude 40 km over the Earth or not crossing any of these borders after preselected number of calculations steps is forbidden trajectory. Following the Liouville theorem, if the cosmic ray flux is isotropic outside the magnetosphere, the flux in a random point inside the magnetosphere is same in all the directions allowed for primaries, while it is zero in the forbidden directions [13]. Allowed trajectories are trajectories of primary cosmic rays (PCR) inside the magnetosphere. Forbidden trajectories can contain just secondary particles created in interactions with the atmosphere. We provide such calculation for 3600 locations distributed uniformly over a complete sphere surrounding Earth at an altitude 400 km. Space Shuttle orbits ( \6iat\ < 51.6° ) cover 78.9% of all calculated locations.

911 For every location we calculate particle trajectories in 270 isotropically distributed directions within the outward hemisphere inside the 32° acceptance cone (around the local geocentric Zenith) of the AMS-01 spectrometer. In every direction we determine 310 trajectories inside 31 rigidity intervals of the AMS-01 data (from 0.37GV to 200GV). To take into account the energy dependence of the proton flux, each energy interval has been subdivided in 10 equally spaced sub-intervals. Table 1. Geomagnetic regions covered by AMS-01 measurements [14]. The regions are defined using the Corrected Geomagnetic latitude (CGM). Region (M)

2.2. Transmission

CGM latitude 6M (rad)

1

0 < |0M| <0.4

2

0.4 < \0M\ < 0 . 8

3

0.8 < | 0 M |

Function

Transmission function (TF hereafter) is a probability function of particle rigidity and position inside the magnetosphere. TF shows the probability of a particles from the magnetopause with rigidity in range 5R ±d3? to reach a selected point (or region) inside the magnetosphere. In our calculation TF shows the probability of a particle with energy from selected AMS energy sub-bin to reach chosen geomagnetic region at low orbit. We determine TF for 3 geomagnetic regions (see Table 1.), for which the AMS-01 data are available. South-Atlantic anomaly region was excluded (i.e. the region with latitude between —55° and 0° and with longitude between —80° and 20°). The value of transmission function for trajectory of particle with rigidity 3? is equal 1 for allowed trajectory and equal 0 for forbidden trajectory. TFM(3tb,iM) is the transmission function for location «M inside magnetosphere for particle with rigidity 3?& from bth rigidity bin. Value of TFM^-biiM) is ratio of number of allowed trajectories to total number of trajectories at IM location in bth rigidity bin. In our article has T F M (3?b, »M) value for trajectories with directions inside AMS-01 spectrometer acceptance cone.

912 Transmission Function for Helium

10 Rigidity [GV]

Figure 1.

100

T F M ( ^ 6 ) for all three geomagnetics regions.

For the location IM, TFjvf(9fy>, ijvf) is given by: 10

TFM{^b,iM)

=^2wb,s

Ku(®b,s)

(2)

s=l

where 3J(,iS is the mean rigidity of the sth sub-interval of width AIRb/10 for the bth rigidity bin, N^ft is the number of allowed trajectories and the total (allowed and forbidden) number of computed trajectories is N total = NaU +Nforb- T h e sub-intervals of rigidity bins have been weighted as function of their relative fluxes. For rigidities larger than 10 GV, the weights, Wb,s (where J2l^=iwb,s — 1), are derived by the flux dependence on the proton rigidity 3?, i.e <j>{E) oc ft-2-78 [14]. Below 10 GV, since the rigidity distribution becomes less steep, the sub-interval flux variation has been interpolated using three adjacent rigidity bins. Transmission function for region M is averaged from all TFM^bjijw) in region.

TFM(Mb)

__EiMTFM($tb,iM) Z)*M

(3)

where 3fy, is the particle rigidity in the bth rigidity interval of width A5R&

913

and Yl^M is the total number of locations for the same region. TFM^Stb) for all three geomegnetics regions are presentet at Figure 1. 1 st geomagnetic region :

1

r

''";+ +

I

+ + ,

T—i—r—r-]—

+

+ +

+

_ 10.000 rr

« E

> CD

\ -

+

+

++

+

+

+ t

0.001

,

,

, ,,

i

;

__j

T

E ;

+

x$ K

*

*

*

*

1

;

" Er

0

f

ij: ff

measured reg. 1 evaluated reg. 1

X

1.000 r

; " r ; . 0.010 = "

_

+ AMS-01 primary

*

*

^_ z \ ~ : :

*m

m

1 m

,

10 Rigidity [GV]

,

,

1

i

"

1L .

:

100

Figure 2. Comparison of evaluated primaries with AMS-01 He fluxes for 1st geomagnetic region.

2.3. Flux at low orbit

reconstruction

We take published differential primary AMS-01 helium spectra [15] as flux at 1AU outside the magnetosphere $1AU(3?b). For the AMS-01 observations, the predicted PCR fluxes per unit of solid angle [<&M(^&)] in geomagnetic region (M) are obtained by convolving the transmission function of geomagnetic region (M) with the estimated AMS01 flux $1AU(3fy,) [14] at 1 AU, i.e. outside the magnetosphere, as functions of the helium rigidity (%). Thus, we have $M(Mb) = $ 1 A U (%)

TFM{Ub).

(4)

The SCR fluxes per unit of solid angle $3M(?flb) can be obtained as $sM(Rb) =

$MPJV(%)

-

$M(%).

(5)

914

2nd geomagnetic region +++ 10.000 x

X CO

o

1.000

+ x xx + o° ° o$

+ AMS-01 primary x measured reg. 2o evaluated reg. 2

:

*

*

™E

_ *

5

*

I

iart

I

:

0.100

0.010

!

*

}

1

a 0.001

-i

1

i

10 Rigidity [GV]

i

i

u

\ m

m_ •_!

100

Figure 3. Comparison of evaluated primaries with AMS-01 He fluxes for 2nd geomagnetic region.

3. Computational Results AMS collaboration published the He fluxes in all three geomagnetic regions [16]. These fluxes contain all registered He particles included also secondaries. In the figures 2., 3. and 4. we compare our evaluated primary fluxes in different geomagnetic regions with AMS published fluxes. As we can see at figure 2., computed PCR spectrum for first geomagnetic region, so for region with highest cut-off rigidity is very similar to the measured AMS-01 helium spectrum. This means that in penumbra part of helium spectra is ratio of SCR to PCR small in first geomagnetic region. Figure 3. shows comparison of evaluated PCR in second geomagnetic AMS-01 region with AMS-01 measured data. Evaluated PCR are lower than the measured helium data. The difference is due to relatively bigger ratio of SCR to PCR in the penumbra part of spectra in second region. Figure 4. show same comparison in the region with lowest cut-off.

4. Conclusions A back-tracing procedure of simulated He nuclei entering the AMS-01 spectrometer has provided the fraction of allowed trajectories of primary cos-

915 3rd geomagnetic region . + +++ . x0o

:*£*$,

>

+ AMS-01 primary x measured reg. 3 o evaluated reg. 3

10.000

°>

1.000

> CD ra a.

0.100

0.010

0.001 10 Rigidity [GV]

100

Figure 4. Comparison of evaluated primaries with AMS-01 He fluxes for 3rd geomagnetic region. mic rays ( P C R s ) . Consequently, it has allowed t o determine t h e so-called Transmission Function ( T F ) able t o describe the t r a n s p o r t properties of the P C R s t o t h e space surrounding the E a r t h (at altitude of a b o u t 400km) from t h e upper limit of the geomagnetic field i.e. the magnetopause located at 1AU. T h e T F has finally allowed t o determine t h e He fluxes of t h e P C R s in t h e 3 geomagnetic regions for comparison with the AMS-01 observations. References 1. E.O. Fluckiger et al., Proc. of the 19th ICRC 5, 336 (1985). 2. D.F. Smart and M.A. Shea, Cosmic-ray penumbral effects for selected balloon launching locations USAF (1975). 3. P. Bobik, K. Kudela and I. Usoskin, Proc. of the 27th ICRC 10, 4056 (2001). 4. B. Wiebel-Sooth, P.L. Biermann, H. Meyer, Astronomy and Astrophysics 330, 389 (1998). 5. J. W. Mitchell et al., Nucl. Phys. B 134, 31 (2004). 6. S. Haino et al., Proc. 28th Int. Cosmic Ray Conf4, 1825 (2003). 7. M. Boezio et al., Astrophysical Journal 518, 457 (1999). 8. E. S. Seo et. al., Advances in Space Research 33, 1777 (2004). 9. D.F. Smart, M.A. Shea, Fluckiger, Space Sci. Rev. 9 3 , 305 (2000). 10. P. Bobik, M. Boschini, M. Gervasi, D. Grandi, E. Micelotta and P.G. Rancoita, SIF Conference Proceedings 73, 417 (2001).

916 11. 12. 13. 14. 15.

N.A. Tsyganenko and D.P. Stern, J. Geophys. Res. 101, 27187 (1996). D.G. Sibeck et al., J. Geophys. Res. 96, 5489 (1991). M.S. Vallarta, Handbuch der Physik 46, (1961). J. Alcaraz et al., Phys. Lett. B 494, 193 (2000). D.F. Smart and M.A. Shea, Handbook of Geophysics and the Space Environment 6, (1985). 16. M. Aguilar et al., Phys. Rep. 366/6, 331 (2002).

LANTHANUM HALIDE SCINTILLATORS AND OPTICAL FIBER READOUT FOR X-RAY/GAMMA-RAY ASTRONOMY AND NATIONAL SECURITY APPLICATIONS MICHAEL L. CHERRY, GARY L. CASE, and CHRISTOPHER E. WELCH Dept. of Physics and Astronomy, Louisiana State Univ., Baton Rouge, LA 70803 USA The Black Hole Finder Probe (BHFP) mission is intended to survey the local Universe for black holes. One approach to such a survey is a hard X-ray coded aperture imaging telescope operating in the 20 - 600 keV energy band. A sensitive hard X-ray/gamma ray imaging telescope is also well suited to surveillance applications searching for shielded sources of illicit nuclear materials, for example "dirty bomb" materials being smuggled into a harbor or city. The development of new inorganic scintillator materials (e.g., LaBr3 and LaCU) provides improved energy resolution and timing performance that is well suited to the requirements for these national security and astrophysics applications. LaBr3 or LaCU detector arrays coupled with waveshifting fiber optic readout represent a significant advance in the performance capabilities of scintillator-based gamma cameras and provide the potential for a feasible approach to affordable, large area, extremely sensitive detectors. We describe the Coded Aperture Survey Telescope for Energetic Radiation (CASTER), a mission concept for a BHFP, and the High Sensitivity Gamma Ray Imager (HiSGRI), a device intended for surveillance for nuclear materials, and present laboratory test results demonstrating the expected scintillator performance.

1. Introduction The Black Hole Finder Probe (BHFP)1'2, part of NASA's Beyond Einstein program, is a spacecraft mission designed to provide a sensitive wide-field survey of stellar- and galactic-scale black holes. One approach to such a survey is a hard X-ray/y ray (20 - 600 keV) coded aperture imaging mission. The large detector area, high sensitivity, and directional resolution required for such a mission can be achieved at a reasonable cost with inorganic scintillators: New scintillator materials (e.g., LaCl3 and LaBr3) provide improved light output, energy resolution, and timing; and the use of segmented scintillators coupled to waveshifting optical fibers promises to reduce the number of electronics channels and simplify the readout complexity. The Coded Aperture Survey Telescope for Energetic Radiation (CASTER)2 is a mission concept optimized to meet the BHFP science goals using an array of wide-field-of-view coded apertures with detection planes based on inorganic scintillators.

917

918

A primary scientific emphasis in the CASTER design is maximizing sensitivity. This emphasis on sensitivity also makes the instrument suitable for monitoring for illicit shipments of radioactive cargoes, for example shielded shipments of "dirty bomb" materials entering a harbor aboard a ship. Curie quantities of 24IAm, 137Cs, or 60Co (quantities characteristic of medical gauges, oil well surveying devices, and commercial irradiators) shielded by several inches of lead can be detected by a sufficiently sensitive detector similar to CASTER. The High Sensitivity Gamma Ray Imager (HiSGRI) is intended to combine the light yield of LaBr3/LaCl3 with the readout simplification provided by a waveshifting fiber/multianode photomultiplier tube (MAPMT) readout to produce an affordable high-sensitivity gamma ray imager suitable for national security surveillance applications. We briefly describe the CASTER and HiSGRI applications here and present initial results showing the feasibility of the scintillator/fiber readout approach. 2. Implementation Approach Considerable recent attention has been devoted to the development of room temperature solid state spectrometers composed of cadmium zinc telluride (CZT). CZT detectors promise excellent energy and spatial resolution, and in principle satisfy the CASTER/HiSGRI requirements. However, large area CZT arrays are expensive and complex, and it is difficult to grow CZT sufficiently thick to provide good sensitivity at 511 keV. As an alternative, cerium-doped lanthanum chloride and lanthanum bromide scintillators provide improved light yield and performance that is well suited to CASTER and HiSGRI. The task of a hard X-ray BHFP mission1'2 is to perform an all-sky census of black hole sources with a 1-year 5a sensitivity level ~ 5 x 10"13 erg cm"2 s"1 (20 100 keV) or ~ 0.02 mCrab. HiSGRI, an imaging X-ray/y ray detector similar in design to CASTER, is designed to detect a 3 Ci source shielded by 4" of Pb at 1 km in a 15 minute survey, or 3 mCi at 10 m in 10 seconds. The device will be suitable for a truck-, dock-, or ship-mounted survey of cargoes entering a port, or a search for contraband cargo in vehicles passing a stationary checkpoint. CASTER employs 16 wide-field-of-view coded aperture telescopes with detection planes utilizing a total scintillator area ~ 6 m2. HSGRI uses one telescope with total area 0.2 m2. In order to provide directional resolution on the order of arcminutes, the detector plane must provide ~ 1 mm position resolution. To provide sensitivity at 511 keV, the scintillator thickness must be 1 - 2 cm and the tungsten mask must have a thickness of 0.5 - 1 cm. The CASTER/HiSGRI detector plane will consist of a layer of scintillator read out by crossed optical

919 fibers. An array of ~ 38 x 38 cm2 x 1 cm thick LaCl3 or LaBr3 forms the basic Xray detector. In the case of LaBr3, 37% of incident 511 keV photons interact in a 1 cm thick scintillator, with 13% of the total cross section due to photoelectric absorption. The LaBr3 light output peaks near 380 run. A layer of 189 2 mm square, double clad waveshifting fibers is laid in the x-direction across the top of the LaBr3 layer (separated from the LaBr3 by a thin glass seal that provides a moisture barrier), and a second layer of fibers is laid in the y-direction across the bottom (Fig. 1). With an absorption peak near 375 nm, the fibers effectively absorb the scintillation light and re-emit a portion of it down the fiber axis with X- fibers

\

-..-.

scintillator array

\,

•''••• • • • ' s " - ' ! v " - : .

N

\

I

.*.-;

. ^ ^

*; /

, /

glass window

v ^V'l ^ ^ ^ H

• \ 1 y-fibers

f W \ /W 5"PMTs

• •

light guides

^ B

Fig. 1. CASTER/HiSGRI detector plane schematic showing LaBr3/LaCl3 scintillator array viewed by a layer of waveshifting x-fibers above and a layer of y-fibers below. The upper and lower fiber arrays are viewed by three multi-anode photomultiplier tubes on each layer in order to determine lateral position. Energy is measured in the large "energy measuring" tubes on the bottom which view the scintillation light that passes through the bottom of the y-fiber layer.

the peak of the emission spectrum near 430 nm. The light at the fiber ends is viewed by 3 64-channel MAPMTs (e.g., Hamamatsu R5900-M64). The crossed fiber layers are intended to measure x- and y-position only. Since only a small fraction of the light is trapped in the fibers, the energy is measured in nine large "energy measuring" PMTs (e.g., 130 mm Electron Tubes 9390KB) viewing the scintillator through the bottom fiber layer and a set of light pipes. 3. LaBr3 and LaCl3 Scintillator: Measured Results In order to detect X-rays efficiently in a single fiber, the light output must be maximized and the fiber absorption spectrum must be matched to the scintillator output. The key features of potential scintillator materials are listed in Table 1, along with those of other materials commonly used for hard X- and y-ray detection. The light output of LaBr3 doped with 0.5% Ce, in particular, is in excess of 60,000 photons/MeV3, among the highest values for inorganic scintillators. The proportionality of light output as a function of energy also contributes to the resolution if multiple Compton interactions occur before the energy is fully absorbed. Over the range 60 - 1275 keV, the non-proportionality

920 in light yield is ~ 6% for LaBr3 compared to ~ 20% for Nal(Tl) and CsI(Tl)4. Its combination of higher proportionality and higher light output gives LaBr3 better resolution than any other scintillator. Table 1. Summary of Detector Characteristics LaBr3 Density (g/cm3) Light Output (ph/MeV) AEZE(FWHM) @ 662 keV Peak X (nm) Fast decay (ns)

LaClj

CsI(Na)

NaI(TI)

BGO

CZT

Ge

5.29

3.86

4.51

CsI(TI) 4.51

3.67

7.13

5.78

5.33

63000

49000

39000

52000

39000

9000

N/A

N/A

<3%

3.5%

7.5%

10%

7%

>10%

<3%

0.3°/

358 -385 25

330 -352 25

420

550

415

480

N/A

N/A

630

1000

230

300

N/A

N/A

Fibers with peak absorption and emission wavelengths well matched to the CsI(Na), LaCl3, and LaBr3 emission spectra include BCF 91A with absorption and emission peaks at 420 and 494 nm respectively, BCF 99-90 (345 and 435 nm), and BCF 99-33A (375 and 430 nm), where we use the designations assigned to their fibers by St. Gobain Crystals. In all three cases, the absorption and emission spectra are sufficiently narrow and well separated that the scintillation photons can be efficiently absorbed and a portion of the re-emitted light trapped and propagated to the MAPMTs at the ends of the fibers. Fig. 2a shows the spectrum of a 137Cs source obtained with a 2.5 cm diameter x 2.5 cm thick LaBr3 detector at room temperature. The energy resolution at 662 keV is 2.7% FWHM, comparable to the quoted resolution (3%) 662 keV Peak Positional 96 Sigma=205 Resolution=15.7% (FWHM)

662 keV Peok Position=3564 Sigma=40.09 Resolutions.74% (FWHM)

1000 2000 3000 Pulse Height (Arbitrary Units)

4000

1000 2000 3000 Pulse Height (Arbitrary Units)

4000

Fig. 2a (left): LaBr3 viewed directly; Fig. 2b (right): LaBr3 viewed at the end of a 2 mm fiber.

for off-the-shelf spectroscopy grade CZT from eV Products (http://www.evproducts.com). The high LaBr3 signal level will permit lower energy thresholds than possible with other scintillators. Although not optimized for low energy response, note that the 32 keV line in the spectrum is visible as well. The spectrum in the energy-measuring tubes viewed through a layer of 2 mm St. Gobain BCF 99-33A fibers yields 662 keV resolution (FWHM) of 5.6%.

921 Fig. 2b shows the spectrum viewed at the end of a ribbon of 12 2-mm fibers, with resolution 16%. The response has been measured for sources from 32 to 662 keV with linearity observed to be better than 2% and a measured integrated yield of 1 photoelectron per 3.3 keV at the fiber ends. In the same geometry, Nal viewed directly gives a FWHM of 7.5%; viewed in the energy tube through a layer of 2 mm fibers, the resolution deteriorates to 9.7%; and viewed at the end of the fibers, the observed 662 keV resolution is 32%, corresponding to 15 keV/ photoelectron. The LaCl3 resolution is measured to be intermediate between Nal and LaBr3: 4.1%, 9.4%, and 20.3% for the same three configurations. 4. Conclusions The ability to fabricate lanthanum halides in large volumes offers improved detection efficiency at relatively high energies. Several manufacturers are currently growing both LaCl3 and LaBr3 crystals using proprietary processes. No fundamental barriers have been identified that would prevent crystal growth and detector fabrication with volumes as large as are presently possible for Nal. Early measurements are promising. We are currently planning a series of radiation damage and activation measurements, preparing to construct a full scale 38 x 38 cm HiSGRI imager, and carrying out a detailed mission study to optimize the design of a scintillator-based BHFP instrument and understand its expected capabilities and performance in detail. Acknowledgements This work has been supported by US DOE/Natl. Nuclear Security Admin, award DE-SC52-04NA25441 and NASA grant NNG04GH78G. We appreciate the assistance of C. Dothy, M. Kushner, C. Rozsa, and their colleagues at St. Gobain, and R. Binns and P. Dowkontt at Washington Univ. We also appreciate numerous useful discussions with our CASTER collaboration colleagues. References 1. J. Grindlay et al., in GAMMA 2001, S. Ritz, N. Gehrels & C.R. Shrader, ed., A1P Conf. Proc. 587, 899 (2001); and in Gamma-Ray Burst and Afterglow Astronomy 2001, G.R. Ricker & R.K. Vanderspek, ed., AIP Conf. Proc. 662, 477 (2003). 2. M. McConnell et al., SPIE Conf. Proc. 5488, 5488-56 (2004); and SPIE Conf. Proc. 5898, paper 5898-01, to be published (2005). 3. E.V.D. van Loef et al., Appl. Phys. Lett. 19, 1573 (2001); K.S. Shah et al., IEEE Trans. Nucl. Sci. 50, 2410 (2003). 4. P. Dorenbos, J.T.M. de Hass & C.W.E. van Ejik, IEEE Trans. Nucl. Sci. 42, 2190(1995).

T H E AGILE SILICON T R A C K E R : C O N S T R U C T I O N A N D CALIBRATION RESULTS

L.FOGGETTA*, C.PONTONI CIFS - Consorzio Interuniversitario

la Fisica Spaziale - Torino

M.PREST Universita degli Studi dell'Insubria e INFN sezione di Milano G.BARBIELLINI, F.LIELLO, F.LONGO Universita degli Studi di Trieste e INFN sezione di Trieste M.BASSET, E.VALLAZZA INFN sezione di Trieste

The AGILE (Light Imager for Gamma-ray Astrophysics) gamma-ray satellite will be launched on an equatorial orbit in 2006 to become the only observatory (till the GLAST launch) for gamma rays in the energy range 30 MeV-50 GeV. The scientific payload consists of a silicon tungsten tracker, a CsI(Tl) minicalorimeter, a plastic scintillator anticoincidence system and a coded mask X-ray detector (in the range 15-45 keV). The silicon tracker is the heart of the satellite. It consists of 12 x-y planes of 16 silicon strip detectors per plane for a total of PS37k channels, organized in 13 Al honeycomb-carbon fiber trays. The first 10 trays have a 245 /an tungsten layer (0.07 Xo) for the photons to convert in a e + / e _ pair. The strips are read by analog-digital, low noise, low power and self triggering ASICs, the TAAls. We will describe the present status of the silicon tracker, whose commissioning has required the equivalent of 10 man-years of tests.

1. Introduction In December 1998 AGILE (Light Imager for 7-ray Astrophysics) 1 was selected as the first small scientific mission supported by ASI. The AGILE mission will provide a powerful observatory for 7-ray astrophysics in the energy range 30 MeV-50 GeV, during the years 2006-2008. AGILE will be "Corresponding author, address: Universita degli Studi dell'Insubria, Via Valleggio 11 22100 Como, Italy. E-mail address: [email protected]

922

923

launched by the Indian launcher PSLV on an equatorial orbit at 550 km with Malindi (Kenya) as ground base. No other 7-ray mission in this energy range is planned in the same period. The instrument 2 is light (sslOO kg) and will be able to detect and monitor 7-ray sources within a large field of view ( « l / 4 of the whole sky). The instrument (Fig. 1(a)) consists of: • a Silicon-Tungsten Tracker (ST) that detects the e + / e ~ pair created in the photon converter in order to provide the trigger to the whole instrument and a complete representation of the event topology allowing the reconstruction of the incoming direction of the 7-ray • a silicon X-ray detector 3 (SuperAGILE) in the range 15-45 keV with a coded mask system • a 1.5 Xo deep CsI(Tl) Minicalorimeter4 (MCAL) that measures the energy released by the pair • an Anticoincidence system (AC) made of segmented plastic scintillators that is used to reject charged particle background.

X-ray X-ray Coded-Mask Imaging System Super-AGILE IASF Roma

Plastic Scintillator Anticoincidence IASF Mllano Converter layers 0.07 X.W

for charged particles INFN Trieste s* Csl bars Calorimeter IASF Bologna

(a)

. ,A

.<_\ " .£]

. _

Trigger & Data Handling IASF MILANO INFN Roma 2

(b)

Figure 1. (a) Photo of the scientific payload. From the top: SuperAGILE, ST and MCAL. (b) Working principle of AGILE

At present, the status of the whole apparatus is the following: • the pre-integration tests of the four sub-detectors of the scientific payload have been accomplished with success • the Power Supply Unit (PSU) has been checked with the flight detectors

924

• the Payload Data Handling Unit (PDHU) is being debugged with all the three main detectors in data-taking mode At the same time, our group is developing the photon calibration line at the Beam Test Facility (BTF) at the Laboratori Nazionali di Frascati (LNF) for the satellite calibration 5 . Fig. 1(b) shows the working principle of the AGILE instrument. In the following, section II will present a brief description of the AGILE tracker and the flight model status, while section III is devoted to the preliminary results with cosmic rays.

(a)

(b)

Figure 2. (a) ST after the assembly, (b) Detail of assembled Silicon tile.

2. The AGILE Silicon Tracker The Silicon Tracker6 is the heart of the AGILE mission. It is a compact, « 37000 channel detector with low power consumption («350juW/channel), self-triggering capability, full analog readout and low deadtime (<150/zs, around 3 orders of magnitude less than the previous gamma ray mission, EGRET). The AGILE Tracker (Fig. 2(a)) is made of 12 planes of silicon

strip detectors organized in 13 trays with 2 views in a x-y configuration of 16 silicon tiles each organized in 4 ladders of 4 detectors. The first 10 trays have a tungsten layer 245 yam thick (corresponding to 0.07 Xo) (Fig. 2(b)) shows detail of a tile (with W under the kapton foil). Each tray is made of a 14 mm core of aluminum honeycomb covered on both sides by a 0.5 mm thick carbon fibre layer obtained from four 0.125 mm plies (0-90-90-0). The active element of the AGILE tracker is a single-sided,

925

AC-coupled, 410 //m thick, 9.5x9.5 cm 2 silicon strip detector with a readout pitch of 242 jum and one floating strip with polysilicon resistors for biasing. It has been manufactured on high resistivity (>4 kfi-cm) 6" substrate by HAMAMATSU PK. Each ladder is readout by 3 TAAls (IDEAS), which are analog-digital, low noise, self triggering ASICs with full analog readout. The trigger of the satellite is given by the tracker itself with a majority condition of 3 over 4 trays; the overall syncronous trigger logic is organized in three levels. A dedicated technology for the High Density Interconnection board for the readout ASICs has been developed by ILFA (Hannover, Germany) based on a design by INFN Trieste, which allows the assembly of the ASICs inside the PCB. The data, control and power lines belonging to the HDIs of a tray are carried through a single, flexible MultiLayer Cable (MLC) that interconnects the four ladders for a multiplexed data transfer. The detector and ASIC biases, the control and readout signals and the trigger signals are generated on the 23x25 cm 2 readout boards (FTB, Frontend and Trigger Board) which are located on two of the lateral sides of the Tracker. The assembly of the flight model has required the equivalent of almost 10 manyears of tests to ensure the quality of the production and of the assembly procedures themselves. 3. ST P e r f o r m a n c e All the components of the Silicon Tracker have been tested both from the thermal, the mechanical7 and the performance point of view8. The high quality of the components and the reliability of the assembly procedures resulted in the rejection of only 10 noisy strips out of 36864. The analog readout and the presence of the floating strips ensures a spatial resolution of «35 ^m for an 15° incident beam on the ladder. The cluster is defined by the presence of at least one strip with a signal 9 times the rms noise. Each nearby strip with a signal greater than 5a is included. The following quantities can be used to describe the ST performance, keeping in mind that the dynamic range of the 12-bit ADC corresponds to 9 MIPs: • Pull, defined as the ratio of the maximum strip signal and its mean noise level. A simplified Landau fit gives a most probable value of «20 (Fig. 3(a)). • Cluster SNR (Signal to Noise Ratio). The most probable value is around 26 (Fig. 3(b)). • Noise Value after pedestal subtraction less than 10 ADC.

926

• Trigger efficiency; it has been measured close to 0.95 for each ladder with a 1/4 of a MIP threshold. The average ladder parameters in terms of noise and gain have not shown a significant variation after the ST assembly. The expected performance is represented by a PSF better than 0.2° with an incoming 7 energy greater than 10 GeV. This results in an error in source positioning from 5' to 20' depending on the source spectrum, its intensity and its off-axis position. 4. Conclusion The AGILE Silicon Tracker has been assembled and tested. The assembly of the outer shell with the other four detectors, PDHU and PSU is being completed. Once the payload is ready, it will be moved to the BTF line in Frascati, where the calibration line is completed. The AGILE launch is foreseen for middle 2006.

PI P2 .pa

1 | i

1 S40. d 1 Md 2

• - - —

PI

12.13 0.JI139 -.WHfe-01-

JJMi

nan

! :

1

.268 0. E239

«

P2

1 —

'" "

—-

1

.

„

_

-4—

\

L

.

\

n

1 \

.7886

1

,

™v*Uw

•* C l u s e r S MR

(a)

(b)

Figure 3. (a) Pull Distribution, (b) Cluster SNR. The fits are performed with a simplified Landau, whose P3 parameter is the most probable value.

References 1. M. Tavani et a/., X-ray and Gamma-ray Telescopes and Instruments for Astronomy, Proceedings of the SPIE, Vol. 4851, 1151-1162, 2003.

927 2. G. Barbiellini et al, The AGILE scientific Instrument, Proceedings of AlP Conference 587, G A M M A 2001, 774-778, 2001. 3. P. Soffitta et al., Proceedings of SPIE, 4140, 283-292, 2000. 4. E. Celesti et al, New Astronomy Reviews 48, 315-320, 2004. 5. S. Hasan et al, Proceedings of this conference 6. E. Vallazza et al, The AGILE gamma ray satellite: the construction and performance of the silicon tracker, Proceedings of IEEE-NSS-MIC 2004 7. C. Pontoni et al, The Thermal and Vibrational Tests of the AGILE Silicon Tracker, Proceedings of IEEE-NSS-MIC 2004 8. G. Barbiellini et al, Nucl. Instr. and Methods A 490, 146-158, 2002

IONS A B U N D A N C E CLOSE TO THE EARTH SURFACE: THE ROLE OF THE MAGNETOSPHERE P. B O B I I C M . J . B O S C H I N I J M . GERVASI* D. G R A N D I A N D P . G . R A N C O I T A Istituto

Nazionale di Fisica Nucleare - Sezione di Milano Piazza della Scienza 3, 20126 Milano (Italy) E-mail: [email protected]

We have compared the data of protons and helium from the AMS-01 detector in the energy range 0.1-200 GeV, and heavier ions (from Z=4 to Z=28) from the HEAO-3C2 in the energy range 0.6-35 GeV/n. Data have been collected almost in the same solar period and the same polarity (1980 for HEAO and 1998 for AMS-01). The experimental conditions are also comparable: altitude and inclination of the orbit, angular acceptance of the detectors. The effect of magnetosphere is to introduce a rigidity cut-off. The results are that the abundance ratio of nuclei/protons is larger than what expected from the spectra measured outside the magnetosphere. We discuss this effect in relation to what has been observed in experiments on board of satellites orbiting near the Earth surface.

1. Introduction Apart from particles generated during the solar flares, the measured cosmic radiation comes from outside the solar system. This component, the Galactic Cosmic Rays (GCR), is dominant at energy higher than few hundreds MeV, but is modulated by the solar activity inside the heliosphere up to few GeV. Protons are largely the most abundant component of this radiation. Nevertheless to GCR contribute also electrons, ions (all the stable nuclei), and a small fraction of antimatter of secondary origin 1 ' 2 . The problem of evaluating both the absolute and the relative abundance of the Cosmic Rays (CR) is very important in relation to the radiation damage and radiation dose in space. An accurate evaluation of these quantities is even more important in prevision of the long duration space missions * Institute of Experimental Physics, Kosice, Slovak Republic tCILEA, Segrate (Mi) t Physics Department - University of Milano Bicocca

928

929

already scheduled, like the International Space Station (ISS), a permanent orbiting station at an altitude of 400 km, or the planned manned interplanetary nights. The effect of radiation on both electronics stuff and organic tissue is depending on the absolute rate of CR, but it is related also to the relative abundance of ions 3 - 5 . 10 4

"I

~102

\ f

•-..

1

(A 10 m

'—'

— • —cosmic p -•--cosmicHe j - A — cosmic C i - » cosmic Fe

103

10°

*10*

10^ _i

10"1

10°

i

'

i

i

10 1

i

'

i

i i

10'

Kinetic energy (GeV/nucleon) Figure 1. Differential Energy spectra of Cosmic Rays. Data of p and He come from AMS-01, data of C and Fe come from HEAO-3-C2.

Current estimates are based on measurements performed on board of satellites and stratospheric balloons during the last 30-40 years. The abundance of different particles or nuclei is usually described in relation to the kinetic energy, which is the natural parameter outside the magnetosphere, in the free propagation space. The intensity of primary nulceons in the energy range from few GeV up to several TeV is approximately given by 1 ' 2 : nucle (1) cm2 s sr GeV where E is the energy per nulceon, a ~ 2.7 is the differential spectral index of the Cosmic Ray flux, and Kn is depending on the element considered. The fraction of the several components is approximately constant, as shown in Figure 1. When CR enter into the magnetosphere the Earth magnetic field provides a shield against the penetration of CR down to the Earth surface. The result is a rigidity cut-off below which CR can not penetrate. Therefore inside the magnetosphere the natural parameter to deal with is the rigidity (see Figure 2). In(E)~KnE~a

930 10 4 10 s

1 ! ' '- i - i - r i i | ...... r

r M

10 1

r

.

i

|

1

i i

i i i ii|

i

i 1

— cosmic p \ -•--cosmic He ; * cosmic C i • cosmic Fe :

'••.,, • • • • • « . "*••» * *•••« ^ «

r E.10- 1 *10"2

E

W IP4

r

*

*

*

*

*

*

*

*

•

*

.

*

*

•

"

1

*

r \

> v ,

r

>

Br

s

••#

*

l

^ 10°

10 1

-i 10 2

Rigidity (GV) Figure 2. Differential Rigidity spectra of Cosmic Rays. Data of p and He come from AMS-01, d a t a of C and Fe come from HEAO-3-C2.

2. D a t a set and detectors We have evaluated CR measurements inside the magnetosphere. We considered the data of protons and helium from the AMS-01 detector in the energy range 0.1 - 200 GeV, and heavier ions (from Z = 4 to Z = 28) from the HEAO-3-C2 in the energy range 0.6 - 35 GeV/n. Data have been collected almost in the same solar period and the same polarity: 1979-80 for HEAO; and June 1998 for AMS-01. The experimental conditions are also comparable. The altitude is ~ 380 km for AMS-01 and ~ 500 km for HEAO-3; the inclination of the orbit is 51.6 deg for AMS-01 and 43.6 deg for HEAO-3; the angular acceptance of the detector is a cone of ~ 32 deg from the axis for AMS-01 and ~ 28 deg for HEAO. The Alpha Magnetic Spectrometer (AMS)6 had a precursor flight (AMS-01) on board of the space shuttle Discovery (flight STS-91), lasting 10 days, from June 2 to 12, 1998. AMS-01 has been the first magnetic spectrometer in space with large collecting area (~ 1 m 2 ). AMS-01 has provided detailed measurements of charged particles flux outside the Earth's atmosphere, collecting ~ 107 protons in the energy range 0.1 — 200 GeV, ~ 105 electrons in the energy range 0.2 — 30 GeV and ~ 106 helium nuclei in the energy range 0.1 — 100 GeV/n. The HEAO-3 7 (High Energy Astronomical Observatory) mission performed a sky survey of gamma rays and CR. The scientific objectives of the experiment HEAO-3-C2 were the determination of the isotopic composition of the most abundant components of the cosmic-ray flux with atomic mass

931 between 7 and 56, the measurement of the flux of each element with atomic number (Z) between Z = 4 and Z = 50, and the search for super-heavy nuclei up to Z = 120. The normal operating mode was a continuous celestial scan spinning around an axis pointed towards the sun. The satellite has been launched in September 1979. The data used here, and shown in Figures 1 and 2, covers the period from October 1979 to June 1980. 3. CR fluxes inside the magnetosphere The effect of the magnetosphere, the region where the Earth magnetic field is present, on CR can be described in terms of the magnetic rigidity R: R = pc/Ze. A particle with a rigidity lower than a threshold value (usually called rigidity cut-off) can not go beyond the magnetosphere and reach the Earth surface. The value of the rigidity cut-off is position dependent, in particular it is decreasing going towards the magnetic poles.

10

• f

1

M

'» ~

0.1

E, x £L

0.01

1E-3

10

100 Rigidity (GV)

Figure 3. Rigidity spectra of protons and He inside the magnetosphere. We have considered two geomagnetic regions, M l and M 5 , as denned by AMS-01.

We have computed the probability for a CR to penetrate the magnetosphere and be detected 8 ' 9 . We have considered the AMS-01 geomagnetic regions (M) 6 . We have then computed the flux of primary CR entering each geomagnetic region, according to the measurements of AMS-01. For instance in Figure 3 we present the flux of p and He for the geomagnetic regions Ml and M5, at low and medium geomagnetic latitude respectively. For the same geomagnetic region M, the flux cut-off in both p and He measured spectra occurs at the same value of rigidity R, as shown in Figure 3.

932

This is a further confirmation that inside the magnetosphere the rigidity represents the natural parameter describing CR spectra. 4. Abundance calculations We have computed the abundance ratio 3? of Helium, Carbon, Iron, respect to protons. We have considered that at a certain position inside the magnetosphere all the particles above the rigidity cut-off are present. Therefore the ratio has been computed using the flux $ integrated above the quoted value of kinetic energy (E) or rigidity (R):

f~$He(E')dE>

XHe/P(E)=jErj:e;:/::, !~%{E>)dE>

m

_ J~*He(R')dR> ®He/p{R)

~ / " %{R)

(3)

dBf

In the same way the ratios $lC/p(E), 9?c/p(.R), &Fe/p(E), a n d $lFe/p(E) have been computed. The contribution of the modelled high energy tail is negligible (< 1%) in this calculation; only the highest energy value could be affected. The results are shown in Figure 4 and 5. In this way we compare the flux ratio in rigidity (Figure 5), corresponding to what we can measure inside the magnetosphere, with the flux ratio in kinetic energy (Figure 4), corresponding to the cosmic abundances. Table 1. Comparison of the integrated flux ratio of He/p, C / p and F e / p in kinetic Energy and Rigidity (see the text for details).

He/p

K K ( 3 GV)

SRB(1 GeV/n)

13 x 10~ 2

6 x 10-2

3

nR/nE 2.2

3

2.1 2.2

C/p

3.8 x 1 0 -

Fe/p

4.0 x 1 0 - 4

1.8 x 1 0 - 4

SRR(75 GV)

K £ ( 3 0 GeV/n)

UR/UE

17 x 1 0 - 2

1.8 x 1 0 -

5 x 10-2

3.4

C/p

5.3 x 1 0 "

3

2.4 x 1 0 - 3

2.2

Fe/p

11 x 1 0 " 4

3.5 x 1 0 " 4

3.1

He/p

The flux ratios, evaluated both in energy and rigidity (see Figure 4 and 5), of ions/protons is roughly constant, apart Fe/p ratio which looks slightly

933 —1

1 1—I I I I

0,1 -

' - • • " « — • • - • - • « -•-•-•—•—•a. 0,01 - -

,n*r^f.

• *4-**^A—A'A-*-**-*-

1E-3

.•^•^•r^rr^T^T ' ^qr~*&y^~i$r-i^~*^F~W'

1

10 kinetic Energy (GeV/nucleon)

100

Figure 4. Integrated flux ratio of He/p (•), C / p ( A ) , Fe/p ( • ) , vs kinetic Energy (s the text for details).

.. • . • : . : . j

•

:... '.. ' . . . ' . • :." j

:

'

*

• • • : . •

i

•....

0,1 — - ^ M * " * * * * - -

I 1

0,01

1E-3

MLMA*JI^~^'''^~

""""" ...~.-.s.,,

*~A~*-&MJ*-M1M&M

r.7"r""^#^w.^3t?^.

""""

:::::::::;::::;::::::^i**^ _.:::::::::::::::::::::::::::::::::::::::::::::: 1F-4

10 Rigidity (GV)

100

Figure 5. Integrated flux ratio of He/p (•), C / p (A), Fe/p (T), VS Rigidity (see the text for details).

increasing with the rigidity (or the energy). As it can be observed in Figure 4 and 5, the flux ratio is enhanced when evaluated in rigidity by a factor going from ~ 2 to ~ 3. In Table 1 the flux ratio has been computedfortwo different conditions: in the upper part 5R has been evaluated considering all the data down to R = 3 GV and E = 1 GeV/n; in the lower part only the highest energy particles have been considered (R > 75 GV and E > 30

934

GeV/n). In the first case most of the incoming flux is considered, but the results could be affected by the magnetospheric cut-off and by the solar modulation. In the second case we are neglecting a large fraction of the total flux, but the results are not affected by propagation in the heliosphere or magnetosphere. In both the conditions we find 5Rfi/5RB ~ 2 — 3. From the Figure 4 and 5 a systematic effect on ^ § e / p and K# e /„ can be observed at low energy, probably due to the propagation inside the Earth and solax cavity. Besides spectra of C and Fe can be affected at high energy by a larger uncertainty due to a lower statistics. 5. Conclusions It is important to evaluate both the absolute and the relative abundance of the Cosmic Rays in relation to the radiation damage and radiation dose in space. Besides, inside the magnetosphere we need to consider the ions abundance in terms of rigidity instead of kinetic energy. When the flux ratio is evaluated inside the magnetosphere (in rigidity) its value is at least 2 times larger than what is usually quoted as the cosmic abundance 10 . For instance the cosmic abundance 3?^ e , ~ 5 x 1 0 - 2 , but inside a geomagnetic region like Ml (for R > 10 GV) we obtain 5Rge/p ~ 0.15. This result must be considered when the effects of cosmic radiation on orbiting satellites are evaluated. This effect is easily understood. In fact when we compute the abundance ratio in terms of rigidity instead of kinetic energy we need to take into account that the charge over mass ratio is Z/A ~ 1/2 for almost all the nuclei except the H for which the ratio is Z/A ~ 1. References 1. M.S. Longair, High Energy Astrophysics, Cambridge University press, Cambridge (1992). 2. Review of Particle Physics, S. Eidelman et al, Phys. Lett. B592, 1 (2004). 3. D. Codegoni et al., Nucl. Instr. and Meth. in Phys. Res. B 217, 65 (2004). 4. A. Colder et al, Proc. of the 7th ICPPAT, Como 15-19 October 2001, World Scientific (Singapore) 780 (2002). 5. A. Colder et al., Proc. of the European Space Component Conference (Toulouse 24-27 September 2002), ESA SP-507, 377 (2002). 6. AMS collaboration - M. Aguilar et al., Phys. Rep. 366, 331 (2002). 7. J.J. Engelmann et al, Astron. & Astrophys. 233, 96 (1990). 8. P. Bobik et al., AGU Geophysical Monograph Series 155, 301 (2005). 9. P. Bobik et al., J. Geophys. Res. - Space Physics, submitted (2005). 10. ISO 15390, Space Environment - Galactic Cosmic Ray model, (2004).

DESIGN OF A SILICON TRANSITION RADIATION DETECTOR (SITRD) FOR ACCELERATORS AND SPACE APPLICATIONS M. BRIGIDA, G.A.CALIANDRO,C. FAVUZZI, P. FUSCO, F. GARGANO, N.GIGLIETTO, F.GIORDANO*, F. LOPARCO, B. MARANGELLI, M.N.MAZZIOTTA, N. MIRIZZI, S. RAINO, P. SPINELLI Universita andINFNBari G.BARBARINO, D.CAMP ANA, G.OSTEPJA, S.RUSSO, V.PALLADINO Universita and INFN Napoli G.BARBIELLINI, F.LONGO, Universita and INFN Trieste A.DE ANGELIS, B.DE LOTTO Universita and INFN Udine We are developing an unconventional Transition Radiation Detector (SiTRD), based on silicon strip detectors (SSDs) operating in a magnetic field region. The SiTRD combines the particle identification performance of a TRD with the high precision tracking capability of SSDs and can be suitable for both accelerator and cosmic ray experiments, where particle identification and momentum reconstruction are required. The main issues related to the design of a SiTRD will be presented and the experimental results obtained with reduced scale SiTRD prototypes exposed to an electron-pion beam with momenta up to 5 GeV/c will be shown.

1. Introduction Transition Radiation (TR) is emitted in the X-ray region whenever an ultrarelativistic particle crosses the boundary between two regions with different dielectric constants. The TR yield is proportional to the Lorentz factor of the particle over a wide range of momentum. Transition radiation detectors (TRDs) are currently used in many accelerator and cosmic ray experiments, to discriminate between different charged particles with the same momentum. Usually, a TRD is composed by multiple modules, each consisting of a radiator and an X-ray detector. The most commonly used X-ray detectors are gaseous counters (MWPCs, drift chambers, straw tubes). In this case, the TR X-rays are absorbed in the same detector region where the radiating particle releases its ionization energy deposit. While in a gaseous detector the signal produced by TR X-rays is * Corresponding author: francesco,giordano@,ba. infii.it: tel: +390805443168

935

936

usually of the same order of magnitude of the particle ionization energy loss, in SSDs it is ten times smaller than the ionization one. Hence, for a silicon TRD, it becomes mandatory to separate the particle from the TR photons by means of a magnetic field (Figure 1). *am*xK

SM> BKotute

Figure 1 Artist's view of a SiTRD module.

2. The SiTRD Concept The Silicon TRD (SiTRD), proposed for the first time in 2001 [1], is composed by multiple TRD modules, each consisting of a radiator and a silicon strip detector (SSD), operated in a magnetic field region. Despite of traditional TRDs, the background due to the ionization energy loss in the TR measurement is overcome. Referring to Figure 1, the detector strips are parallel to the z axis, perpendicular to the bending plane, so allowing the SSD to detect the TR X-rays and at the same time to reconstruct the track of the radiating particle in the x-y plane. A second SSD with the strips parallel to the y axis can be added behind the first SSD, to reconstruct the particle track also in the x-z plane. 2.1. The Design To optimize the design of the SiTRD in dependence on the application it is intended for, some crucial aspects have to be taken into account. One of these concerns the choice of the radiator, that contributes to determine both the detector performance and size. The radiator should be a compromise between the need of increasing the yield of TR X-rays, to achieve good particle identification capability, and the need of a reduced thickness, to minimize the quantity of material crossed by the particles and to build devices as much compact as possible. The other parameters to be optimized are those concerning the detector geometry, i.e. the magnetic field B, the module length d, the SSD strip pitch p and thickness t (Figure 1). The magnetic field value determines the bending

937

power of the spectrometer (that is roughly equal to Bd for a single SiTRD module) and consequently its momentum range. Nevertheless, the use of superconducting magnets allows to achieve magnetic fields up to few Tesla, corresponding to a bending power of a few Tm. The magnetic field also contributes, together with the other geometrical parameters, to determine the particle identification performance. The radiating particle is detected in a different region of the silicon detector far from the region where the TR X-rays is absorbed. In order to have well separate signals on the readout strips, the distance between the X-rays and the particle should be greater than the value p of the strip pitch. This separation increases with the magnetic field B and with the distance d-r between the radiator and the SSD, but decreases with increasing momentum. The upper momentum limit for the TRD is reached when the separation between the particle and the X-rays becomes of the same order of the strip pitch p. Also the SSD thickness has to be chosen in order to have a high absorption probability for the TR X-rays while limiting the detector noise. The typical TR X-ray spectra are peaked at energies around 10 keV, corresponding to absorption lengths of about 100 p.m: a detector thickness of about 500 pm ensures about 100% absorption probability for 10 keV X-rays.

Figure 2 Experimental setup for the study of the SiTRD performance.

3. Experimental study of the SiTRD performance After examining the performance of various radiators [2], we have built a reduced scale SiTRD prototype consisting of three modules. This prototype has been exposed to an electron-pion beam with momenta up to 5 GeV/c at the CERN-PS T9 beam facility and its performance has been evaluated. Figure 2 shows the layout of this test. The three module SiTRD prototype has been placed between the expansions of a dipole magnet, that could supply a magnetic field up to IT. Each module was composed by a radiator, followed by a SSD with the strips oriented along the direction perpendicular to the bending plane. The first and the third SSD were 400 um thick, while the second and the fourth SSD were 500 um thick. The particle tagging was performed using the

938

information provided by the two Cerenkov counters CI and C2 and the lead glass calorimeter. The plastic scintillators S1,S2 and S3 were used as trigger. 3.1. Experimental data The data analysis concerning the definition of the particle cluster, the X ray signal and the photon search algorithm is detailed in another paper [2]. Two different SiTRD configurations with three and four modules have been tested. The module length was 17cm and 12.5cm respectively and the magnetic field was set at IT. In Figure 3 are shown some results concerning the electron identification efficiency (£e) and the pion contamination (sK) for each set-up.

»

i

3

i

i

s

i

;

'

*

'

fc

Figure 3 Electron identification efficiency and pion contamination for both configuration tested with different radiators.

The identification efficiency depends on the X-ray search algorithm: more bent is the trajectory and more separated are the particle cluster from the X ray signal (i.e. high field or low momentum), better is the electron identification. This justifies a higher identification obtained with three modules if compared with the four modules TRD results. A same behavior is exhibited by the pion contamination: it slightly increases from 1% in the three module configuration to 5% in the four module runs due to a higher number of strips investigated by the X ray search algorithm. 3.2. Simulation results A Monte Carlo code [3] has been developed to fully simulate the SiTRD, to study the performances in different configurations of module length, magnetic field strength and strip pitch. The most significant simulated set-ups are configurations with a very precise strip pitch (25 um) and an intense magnetic field (3T) with a total length of 20cm, that can be employed in experiments at accelerators, and with a more

939 coarse strip pitch (228^m) and a lower B field (IT) 50cm long, more suitable for cosmic ray experiments in outer space. •*-^,-—i- — - : • : : . - : : ; .

-A

**-*,__ *+•••*

•

•

>••-,

I i

•

<

1 N \

* , i

•

\

\

.v, \

>f- i

f ^ -f.

X \

i \

: •

Figure 4 Electron identification efficiency as a function of momentum for accelerator (left) and space (right) configurations. A: one module SiTRD; • : 2 modules; •: 4 modules.

Figure 4 shows the electron identification efficiency versus the particle momentum for the two configurations and different SiTRD lengths, for a 5cm long polyetilene regular radiator. An almost complete identification (~100%) up to lOOGeV/c momentum can be reached by four TRD modules in the accelerator configuration, though it drops to 70% for the low B coarse strip pitch configuration. 4. Conclusions A SiTRD has been designed and amply tested. It exhibits good particle identification capabilities and precise momentum measurements. We have carried out a test with an electron/pion beam with momenta up to 5 GeV/c with a reduced scale SiTRD prototype, studying various types of radiator and exploring different detector configurations. On the basis of the data collected, a full Monte Carlo simulation code has been developed and used to optimize the SiTRD design for different high energy physics experiment purposes. References 1 M. N. Mazziotta et al., Frascati Physics Series 25, 75 (2002). 2 Brigida et al., Nuclear Instr. and Methods A 550 (2005) 157-168. 3 Brigida et al., Nuclear Instr. and Methods A 533-3 (2004) pp. 322-343

T H E ORIGIN OF HELIUM-3 ISOTOPE E N H A N C E M E N T IN T H E M A G N E T O S P H E R E OBSERVED B Y T S U B A S A SATELLITE

M. H A R E Y A M A , M. F U J I I , N. H A S E B E , N. K A J I W A R A , S. K O D A I R A , K. S A K U R A I , M. T A K A N O Advanced

Research Institute for Science 3-4-1 Okubo, Shinjuku-ku,

and Engineering, Tokyo 169-8555,

Waseda Japan

University,

T . G O K A , H. K O S H I I S H I , H. M A T S U M O T O Institute

of Space Technology and Aeronautics, Japan Aerospace Exploration Agency, 2-1-1 Sengen, Tsukuba, Ibaraki 305-8505, Japan

The energetic helium isotopes, 3 He and 4 He, were observed by Heavy Ion Telescope (HIT) onboard the TSUBASA satellite from March, 2002 to September, 2003 in the geostationary transfer orbit. These isotope data in the quiet periods were analyzed to obtain the spatial and temporal variation of fluxes of both isotopes. The enhancement and variation of the 3 H e / 4 H e ratio in low L-value were found to be large in comparison with that of galactic cosmic rays (GCR), while the ratio in high L-value is almost stable and comparable to GCR one. These results suggest that the injection and loss mechanism of heavy ions differ between inner and outer radiation belts.

1. Introduction The observations on both short-term and long-term spatial and temporal evolutions of trapped particles provide us the vital clue to understand the injection and loss processes of them in the radiation belts. Up to present, however, the observations of energetic heavy ions and their isotopes in the earth's radiation belts have been performed only by a few satellites 1,2 ' 3 , while proton, alpha particles and electrons have been done with many satellites. CRRES 1 , SAMPEX 2 and NINA 3 reported that the isotopic ratio of energetic helium ions, 3 He/ 4 He, is strikingly enhanced in inner radiation belt as compared with solar abundance 10~ 4 . In numerical simulation, Selesnick and Mewaldt suggested that 3 He isotopes originated from the interaction of protons with residual atmosphere. 4 However, the origin of 3 He has not become cleared as yet. 940

941 The Mission Demonstration Test Satellite-1 (MDS-1), called "TSUBASA" Satellite, provided a good opportunity to study various important problems as observed in radiation belts. This report shows the Ldistribution of helium fluxes in quiet periods, excluding the active periods associated with SEP events and Coronal Mass Ejections. We discuss the origin of helium isotopes in the radiation belts. 2. Observation The TSUBASA satellite was in operation from February, 2002 to September, 2003 just after solar maximum period. It was in the geostationary transfer orbit with a perigee of 500 km and an apogee of 36000 km at the inclination of 28.5° which corresponds to I a 1 ~ 10. Its orbital period was of 10 hours and 35 min and its spin rate was of 5 spins/min. The Heavy Ion Telescope (HIT) onboard the MDS-1 was designed to measure energetic ions in space by 2 layers of position sensitive Si detectors and 16 PIN typed Si detectors. The mass resolution for helium isotopes were of 0.24 amu in rms in the energy range 20~40 MeV/nucleon. The details of instruments were reported elsewhere.5 3 . R e s u l t s and Discussion The time profile of helium elemental flux integrated for all L-value is shown in Fig. 1. Many peaks associated with SEP events were seen more often in the first half term than in the last half term. Hereafter, we use the data for term I to IV as shown in Fig. 1, which correspond to quiet periods (see the thick red line), in which helium counts per day were less than 100. Figure 2 shows the variation in the L-distribution of helium isotope

a»

300

«o

Day of Year from JanJOl/2002

Figure 1. The time profile of helium flux during whole mission period, the quiet period was shown by thick red line.

fluxes, 3 He (blue square) and 4 He (red circle), with left axis and their ratio of 3 He/ 4 He (green diamond), with right axis in four quiet terms. Both fluxes of helium isotopes are enriched greatly and varied largely in L < 3, throughout whole observing period. Looking at them in detail, there is a dip or slot of 4 He flux around 3 - 4L in each of the first two terms, while such a dip is around 2L in the later two terms. The situation for 3 He in the

942 DCIY = 5 7 - I 8 4

$

s

ir

\

>

l-e-miiul

quiet period as term I '

DoY = 253-301

1

\

I

10"

iff1

$ 10 !

\ \

*

lff»

'H |'Hr+S*-*'V I 5

J 10*

Iff' if

_~_o-o-o-o-o-o-o-o-o-»-o-5 -"-'He -"-'He

•

2

4

6 L-valuc

8

DoY = 318-43» 5 , % Iff"

s

i

10

lff» 2

4

< L-valuc

8

(, L-value

8

^

"V J

10

1 *--rati«|

4

f I0 4

Iff' 3

Iff' Iff1

10'

10"

Iff5

Iff' ™^-«-<^,-o-<>--i>-o-o-c^Y

It* -°->Hc -o-'He

4

I)oY = 447-575

IO"

•

?

-n-'He ""-"He

•

| -*~mtio|

vV*^*>^v- i a: ^^ssss<^

:

X»5S^^CV

s w S

1" •-ralidj

5

s

1 10'' -Q-»He -°-"rfc

• 6 L-vulue

8

Figure 2. The L-distribution of the helium isotopes. The number in each panel corresponds to the number of term in Fig. 1.

inner region is almost the same as for 4 He, but such slots are not clearly seen. In contrast, the both fluxes are stable in outer L region larger than 4 in L. The ratio of helium isotopes in low L obtained by us varied dynamically its value among these terms as varying both isotope intensity. At L ~ 1 1.5 in first term, the ratio is a value of about 3, being much greater than solar abundance as the order of 1 0 - 4 . This result is comparable to those obtained for the past observations as CRRES 1 , SAMPEX 2 and NINA 3 . On the other hand, the ratio in L > 5 is almost constant of 0.3 ~ 0.5 that are a few times larger than those of BESS 6 and IMAX 7 , though their energy range was from a few 100 MeV/n to a few GeV/n as galactic cosmic rays (GCRs). Here, we note that these results were taken into account contamination effect of fragment particles produced by the interaction in aluminum window covering top of telescope. However, their effect for helium ratio as observed seems to be small as about one third in the amount compared with heavier nuclei such as C and O. Moreover, since the window of 2.1 mmthickness-aluminum corresponds to only 0.54 g/cm 2 , most of the particles seem to freely pass through the window without the nuclear interaction.

943

DoY from Jan. 1,2002 Figure 3.

Daily average of He isotope fluxes with L-value. (a) 4 He, (b) 3 H e

These results show that the helium isotopes trapped in outer radiation belt are identified as GCRs components. Our ratio is a few times larger than that of GCRs. However, the atmospheric thickness near the earth larger than traversing the interstellar medium may be responsible for producing 3 He from fragmentation of 4 He during the period trapped in the radiation belts. In addition, neutral 0 and He are largely dominant as compared with neutral hydrogen in upper atmosphere higher than 1000 km during the solar maximum periods. 9 Therefore, energetic protons can also produce He ions from interaction with these atmospheric target atoms. We also analyzed neon isotopes, 20 Ne and 22 Ne, in the same quiet periods. The ratio of neon isotopes with 50 - 100 MeV/n, 2 2 Ne/ 2 0 Ne, obtained by this work is ~ 0.7 in L > 4 which is similar to one of GCRs as 0.6 at several hundreds MeV/n reported by ACE/CRIS. 8 The ratio of Ne isotopes in the solar abundance is less than 0.1 and thus it is impossible for 22 Ne to be produced from both 20 Ne and atmospheric neutrals except for fragmentation of energetic particles heavier than Ne ions. On the other hand, the origin of He ions in inner radiation belt have not been cleared as yet. The He isotopes with 20 - 40 MeV/n in less than 3L is more enhanced in their flux than its outer region, especially for 3 He, despite the difficulty for them to penetrate to the lower L region. The daily average of He isotopes with L-value is shown in Fig.3. He isotopes by SEP events can not penetrate to low L from high L. Both fluxes of He isotopes

944

at L ~ 1 are stable, whereas these fluxes around L ~ 2 are unstable and the variations are synchronous between 3 He and 4 He which are enhanced after the SEPs at day 150 and lost after the SEPs at day 210. This fact indicates that trapped protons produce both He isotopes in the lowest L region to interact with atmospheric neutrals. If this is the case, we may detect the enhancement of light nuclei such as Li, Be and B in lower L region, except particles heavier than O ions. However, the variation mechanism of He fluxes and their ratio in low L still remains as an important question. 4. Conclusion This report presented the fluxes and ratios of trapped He isotopes observed by HIT/TSUBASA and discussed their possible origin. The results obtained are summarized as follows: • The fluxes of He isotopes are enhanced more in inner radiation belt as compared to the outer one. • He fluxes in L < 3 are highly variable, while they are rather stable in the outer region. • The isotope ratios of 3 He/ 4 He and 2 2 Ne/ 2 0 Ne are similar to those of GCRs in outer radiation belt, while the ratio of 3 He/ 4 He in inner one varied its value from 0.1 to 3. Based on these results, we suggest that the heavy isotopes trapped in the outer radiation belt may have been originated from GCRs, but their origin in the inner belt differs from that in the outer one. Acknowledgements This report was partially supported by a Grant-in-Aid for The 21st Century COE Program (Physics of Self-Organization Systems) at Waseda University from MEXT and a Grant-in-Aid of Scientific Research of JSPS (Grant No. 17740153). References 1. 2. 3. 4. 5. 6. 7. 8. 9.

J. Chen et al, J. Geophys. Res. 24 787, (1996). J. R. Cummings et al, J. Geophys. Res. Lett. 20 2003, (1993). A. Bakaldin et al, J. Geophys. Res. 10 1029, (2002). R. S. Selesnick and R. A. Mewaldt, J. Geophys. Res. 100 9503, (1995). H. Matsumoto et al, Jpn. J. Appl. Phys. , 44, 9A 6870 (2005) J. Z. Wang et al., Astrophys. J. 564 244, (2002). O. Reimer et al, Astrophys. J. 496 490, (1998). W.R. Binns et al, Adv. Space Res. 27 767, (2001) U.S. Standard Atmosphere, NOAA, NASA, USAF, Washington, (1976)

T H E M I C R O S C O P E MISSION A N D PRE-FLIGHT P E R F O R M A N C E VERIFICATION

D . H U D S O N ? P. T O U B O U L , M. R O D R I G U E S ONERA - DMPH, 29 avenue de la Division Leclerc, 92322 Chatillon, France E-mail: [email protected]

Recent developments in fundamental physics have renewed interest in disproving the equivalence principle. The MICROSCOPE mission will be the first test to capitalize on the advantages of space to achieve an accuracy of 1 0 - 1 5 , more than two orders of magnitude better than current ground based results. It is a joint CNES, ONERA, and Observatoire de la Cote d'Azur mission in the CNES Myriade microsatellite program. The principle of the test is to place two masses of different material on precisely the same orbit and measure any difference in the forces required to maintain the common orbit. The test is performed by a differential electrostatic accelerometer containing two concentric cylindrical test masses. This paper will present both an overview of the mission, and a description of the accelerometer development and performance verification.

1. I n t r o d u c t i o n

The equivalence principle (EP) was first formally stated by Einstein in 1907 as a requirement for his theory of general relativity. Although generally accepted as true, it has in fact never been measured beyond some 1 0 - 1 3 . Modern theories in the search for a unification of fundamental forces require a new force in addition to gravity, which has motivated a better analysis of general relativity to determine the limits of its applicability. A variety of tests have been performed or proposed, but it is the equivalence principle test that examines a fundamental basis of GR, without requiring interpretations or related theories. An improved accuracy of the test will place limits on GR and on new unification theories, which predict violations at levels up to 10~ 12 [1], and will, as well, directly detect the new force, in the case of a violation. *Work partially supported by the french Centre National d'Etudes Spatiales.

945

946

Tests of the equivalence principle date back as far as Galileo, who tested the universality of free fall in the early 17th century. Ground based testing has not been able to surpass a precision of 1 0 - 1 3 , the best results from both torsion pendulums and lunar laser ranging only approaching this [2]. However today's technology allows testing in space, where a low background noise level can be further reduced by an almost unlimited integration time. 2. Mission Overview The MICROSCOPE mission is in fact a test of universality of free fall. Two masses of different material will be maintained on exactly the same orbit to within 1 0 - 1 1 m by means of electrostatic forces in a differential accelerometer. A difference in the forces required for each mass will indicate a violation of equivalence. The satellite will be launched in early 2009 to a heliosynchronous quasipolar orbit, in the dusk-dawn plane to provide the necessary thermal stability. The orbit will be nearly circular (e < 5 x 10~3) at an altitude of 710 km. The altitude is chosen as a compromise between the strength of a violation signal, which is proportional to the local gravity, and the amount of drag that must be accommodated by the drag compensation system. The orbit eccentricity is minimized in order to reduce perturbing effects related to Earth's gravity gradient, magnetic field, and similar sources. The EP violation signal is well known, with a frequency corresponding to the sum of the orbit frequency and the frequency of any rotation of the measurement axis in the orbit plane, and a phase related to the measurement axis attitude with respect to the position in orbit. Therefore both the frequency and phase can be controlled in order to reject different perturbation sources, and verify that the signal is due to an EP violation. 3. The Microsatellite MICROSCOPE is the third scientific mission in the CNES Myriade microsatellite program. The science payload will be two differential electrostatic accelerometers, one with test masses of platinum and titanium materials for the EP test, and a second with both test masses in platinum for baseline precision measurement. The satellite will have a drag compensating attitude control system utilizing one star camera, data from the two accelerometers, and twelve FEEP thrusters. Precise attitude determination is required to limit the inertial accelerations induced by the satellite attitude fluctuations, and to calculate

947

the Earth's gravity gradient at the test masses in order to accurately correct its effects in the measurement. The angular acceleration specification of 10~ 8 rad • s~2/y/Hz is achieved thanks to accelerometer angular measurements furnished by the payload, and utilized in conjunction with the star camera data with a hybridization frequency of about 6 x 1 0 - 3 Hz. The satellite drag free system reduces background noise, while allowing the proof masses to follow a free fall trajectory. The common acceleration from the two inertial sensors of one differential accelerometer is maintained at zero by actuation of the FEEPs, with a residual acceleration of 3 x 1 0 - 1 0 m • s~2/vHz. The drag free control loop is also utilized for in-flight calibration, as it can produce specific oscillations of the satellite about its mean position to create a well known acceleration which is then measured by the inertial sensors.

4. The Science Payload The differential accelerometer of MICROSCOPE is the first cylindrical electrostatic accelerometer to function in six degrees of freedom. The cylindrical form was chosen due to the necessity of exposing the two test masses to exactly the same gravitational field simultaneously. Cylindrical masses with spherical moments of inertia were selected due to the impracticality of using actual spheres. The two inertial sensors are concentric coaxial cylinders, with a common center of mass between the two proof masses. The cylinder axis corresponds to the sensitive measurement axis. Each inertial sensor of the differential instrument consists of a proof mass surrounded by electrodes etched into two gold plated silica cylinders, as shown in Figure 1. The electrodes function in pairs to capacitively sense the proof mass position, and to apply the electrostatic forces required to maintain the mass perfectly centered. The only physical contact to the proof mass is a 5 /xm gold wire used for charge control and for applying a sinusoid voltage for the capacitive sensing. Precise alignment of the six cylinders is crucial to achieve the desired performance, with specifications of better than 5 /xm on assembly. The actual alignment of the two masses will be measured precisely during the in orbit calibration. The science instrument will contain proof masses of 0.46 kg of a platinum-rhodium alloy for the internal sensor, and 0.31 kg of titanium for the external sensor. These materials were chosen for their large difference in nucleic components, as an EP violation may be associated with different contributions from the subatomic particles. However, consideration was

948

> Vacuum System Electric) Connectn-s

External Sensor Electrode Cylinders External Mass (PtRhlOorTAoV) Internal Mass (PlRhlo) -—- Internal Sensor Electrode Cylinders

Blocking Mechanism

Figure 1. Differential accelerometer cross section. The blocking system supports the mass during launch via a constant 1600 N force on the axial stops, then retracts the stops once in orbit. The getter material of the vacuum system maintains the vacuum throughout the mission duration, as pumping is used only prior to launch.

also given to their macroscopic properties, such as thermal stability and machinability. Each inertial sensor measures the proof mass acceleration, which can be expressed as follows when the mass is servo-controlled motionless with respect to the satellite. F

rfc = ±^.+

'Ma Mi

mGfc^ _ln fmGk x , [T] - [In] OkOsat 9 (O ) + m/fc J " ' sat • \ mik

(1)

Where Tfc is the acceleration detected by inertial sensor k and Fng is the residual non-gravitational forces on the satellite. M refers to the satellite mass, while m is the mass of the proof mass, with J and G indicating inertial or gravitational mass. The final term includes the Earth's gravity gradient, [T], and the inertial accelerations, [In], due to the satellite's angular motion. The equivalence parameter, 5 = r^2- — ^ S 1 , appears in the difference of detected acceleration between the two inertial sensors: T d = Sg (Osat) + (5 [T] - [In]) 0~&2

(2)

The separation between the two masses, 0\0
949 formance of the analogue electronics, and produce changes in the sensor geometry. A temperature gradient in the sensor core is also of concern, as even radiation pressure and radiometric effects must be limited to achieve the required performance. The core materials are chosen to minimize thermal expansion, while a temperature stability has been specified at the mechanical interface of all instrument components: maximum 1 mK variation at the EP frequency for the sensor unit, and ten times that for the associated electronics. The stability will be achieved via passive methods on the satellite, with the temperature monitored for housekeeping purposes. 5. Instrument Development 5.1. Numeric

Analysis

One major basis for the instrument and mission optimization is the MICROSCOPE Error Budget. This analysis takes into consideration all the parameters affecting the instrument performance, including the sensor unit geometries (~50/mass), the electronics (~100/mass), and the environment (~30), in order to calculate the expected instrument performance. The output parameters under consideration are the bias and scale values that convert the output voltages to accelerations, and more specifically their noise levels and thermal variability, as well as the instrument sensitivity. The first function of the Error Budget is to place limits on the input pa10'

• Requirement - Inner Sensor Performance - Outer Sensor Performance

-» i o -

10 °

10"

Ereq [Hz] Figure 2.

Instrument sensitivity along cylinder (sensitive) axis.

950

rameters in order to achieve the targeted performance of 1 0 - 1 2 m-s~2/vHz. Once the parameters have been specified, and when possible measured, the actual values are input to the Error Budget so that the expected performance can be calculated, as the second function. This analysis is complete, with the performance shown in Figure 2, and it will be kept up to date as more measured parameters become available. Other numeric investigations include finite element analyses of both thermal and vibration susceptibility, and a full accelerometer simulation used both for performance verification and to design the servo-control loops. These have all been completed in designing the instrument, with the results now being verified through testing of the actual components.

5.2. Development

Models

As the flight models can not function in 1 g, the development philosophy is to analyze a series of sensor models adapted for specific purposes. The Vibration Test Model fully conforms to the Flight Model design with masses of Inermet 180, a tungsten alloy nearly as dense as the platinum alloy, and a test version of the blocking system that will support the proof masses through the vibrations of launch. Its purpose is to verify all mechanical aspects of the instrument design, and verify that the blocking system both functions and provides sufficient support. Testing involves subjecting the model to specific vibrations, then analyzing all surfaces and alignments. After the minimum launch level of 10 g vibrations, the blocking system functioned as expected: no evidence of contact between the electrode cylinders and proof masses was observed, and there was no significant transfer of material between the masses and the support stops. The Laboratory Prototype is a single inertial sensor adapted to facilitate operation in a l g environment. The proof mass is 14 g of silica, and the separation between the mass and electrodes is reduced in order to obtain more force from the same applied voltage. To operate in l g , the mass must be levitated artificially, by means of high voltages, in an unstable configuration. This is the first cylindrical accelerometer with control in six degrees of freedom, and its main objective is to demonstrate that control is possible with this electrode configuration. The artificial levitation has been achieved, with control on four of six axes. Control must be added along the cylinder axis, and for the rotation about it, but this does not present the challenges of the radial control. Once full six-axis control is achieved, all axes will be fine-tuned for maximum stability.

951 The Engineering Model conforms to the Flight Model design, but with masses in silica to permit operation in l g . It will be used for extensive performance testing on a pendulum test bench which provides a stabilized platform with a background noise below 1 0 - 8 m • s~2/VHz, but also allows controlled accelerations to be applied in the horizontal plane. The Qualification Model also follows the Flight Model design, with highdensity test masses. Therefore it will operate only in drop tower testing, before and after being subjected to thermal and vibration stresses. The testing will be done at the Zarm facility in Bremen, using a double drop capsule specially designed to reduce noise to below 10~ 6 m/s2 for the MICROSCOPE test campaigns. 6. Conclusions More accurate verification of the equivalence principle will place constraints on the range of validity of general relativity and developing unification theories. To improve test accuracy by two orders of magnitude, the MICROSCOPE mission will verify the universality of free fall with a differential electrostatic accelerometer in orbit about the Earth. The targeted test precision of 1 0 - 1 5 requires an accelerometer performance of at least 10~ 1 2 m • s~2/y/Hz, at the frequency of the potential violation signal. This places stringent specifications on all aspects of the mission, with emphasis on the perfect alignment of the two test masses. As the flight instrument is not designed to function on ground, its development includes extensive numeric analysis, component testing, and a series of functional test models adapted to operate in 1 g. If MICROSCOPE detects a violation, general relativity will have to be adapted, and the mission can be repeated with different test mass materials to further investigate the source of the violation. If no violation is detected, restrictions will be placed on unification theory development, and tests with even better precision will follow, such as the STEP [3] mission. References 1. P. Fayet, Theoretical Motivations for Equivalence Principle Tests, Adv. Space Res. Vol. 32, No. 7, pp. 1289-1296, 2003. 2. S. Baessler, B.R. Heckel, et al, Improved Test of the Equivalence Principle for Gravitational Self-Energy, Phys. Rev. Letters, 1999, Vol. 13, No. 18. 3. J. Mester, S. Buchman, et al., Gravitational Experiments in Space: Gravity Probe B and STEP, Nuclear Physics B (Proc. Suppl.) 134 (2004) 147154.

P E R F O R M A N C E OF T H E I N T E G R A T E D T R A C K E R T O W E R S OF THE GLAST LARGE A R E A TELESCOPE

M. B R I G I D A , A. C A L I A N D R O , C. FAVUZZI, P. F U S C O , F . G A R G A N O , N. G I G L I E T T O , F . G I O R D A N O , F . L O P A R C O , B . M A R A N G E L L I , M. N. M A Z Z I O T T A , N. MIRIZZI, S. R A I N O A N D P. S P I N E L L I for the GLAST Dipartimento

Interateneo

E-mail:

LAT

Collaboration

di Fisica "M. Merlin" and INFN Via E. Orabona 4, 70126 Bari, Italy [email protected]

Bari

The GLAST Large Area Telescope (LAT) is a high energy gamma ray observatory, mounted on a satellite that will be flown in 2007. The LAT tracker consists of an array of tower modules, equipped with planes of silicon strip detectors (SSDs) interleaved with tungsten converter layers. Photon detection is based on the pair conversion process; silicon strip detectors will reconstruct tracks of electrons and positrons. The instrument is actually being assembled. The first towers have been already tested and integrated at Stanford Linear Accelerator Center (SLAC). An overview of the integration stages of the main components of the tracker and a description of the pre-launch tests will be given. Experimental results on the performance of the tracker towers will be also discussed.

1. The Tracker of the GLAST Large Area Telescope The GLAST mission will study the gamma-ray sky in the energy range from IkeV to 300GeV, allowing the investigation of many fields of the gammaray astrophysics. The GLAST satellite will be instrumented with two main detectors: the Large Area Telescope (LAT), that will operate in the energy range from 20MeV to ZOOGeV, and the GLAST Burst Monitor (GBM), that will operate in the range from IkeV to 3QMeV. The LAT is a high energy pair conversion telescope. It will consist of three subsystems: a precision silicon strip detector (SSD) tracker (TKR), a Csl calorimeter (CAL), and a plastic scintillator anticoincidence detector (ACD). The LAT design is modular, an is composed by an array of 4 x 4 identical towers, supported by a low-mass grid structure. Each tower 952

953

includes a TKR module, a CAL module and a DAQ unit. The tower array is surrounded by the segmented plastic scintillators composing the ACD. Each TKR tower has a 37.2 x 37.2cm2 cross section and it is 75cm high. It is composed of layers of SSDs and thin tungsten converter foils. The e + - e~ pairs produced by photons interacting in the conversion foils are tracked by the SSDs and their energy is mesured by the calorimeter downstream. The SSDs and the converter foils are arranged in a structure consisting of a stack of 19 panels, called trays. Each tray holds a converter foil and two of SSD layers. The upper 12 trays are equipped with a 3%X0 converter foil, the middle 4 trays with a 18%X0 converter and bottom 3 trays without converter foils. The active detectors are 400//m thick single-sided SSD wafers with a cross section of 8.95 x 8.95 cm2, bonded together into 35.8cm x 8.95cm ladders. The strip pitch is of 228 /xm and there are 384 strips per ladder 1 . The TKR read-out electronics encompasses about one milion of strips and amplifier-discriminator channels. The current signal from each strip is converted into a voltage pulse by a charge-sensitive preamplifier coupled with a shaper. Finally, the shaper output signal is converted into a digital signal by a discriminator. The discrimination threshold corresponds to about 1/4 of the pulse amplitude induced by a minimum ionizing particle. For each event, the readout electronics provides a digital output that includes two kinds of information: the list of the fired strips, i.e. of the channels whose signal amplitude exceeds the threshold, and a set of Time over Threshold (ToT) values, each corresponding to a single SSD layer. The ToT represents the time interval during which, at least a signal from a strip belonging to the plane, exceeds the threshold. The information about the fired strips is used for tracking purpose. On the other hand, since the ToT is almost proportional to the energy deposited inside the SSDs, the ToT information can be used to estimate the number of particles crossing a SSD planes, In particular, in gamma-ray events, the ToT information can be used to search the position of the photon conversion vertex.

2. Integration of the LAT Tracker The construction of the LAT TKR is responsibility of the Italian INFN collaboration involved in the project. After being assembled in the INFN laboratories, the TKR towers are delivered to the Alenia test facility in Rome, where both vibration and thermal-vacuum tests are performed. During the environmental testing, functional tests are also performed on the

954 TKR towers. After the environmental tests, TKR towers are shipped to SLAC, where they are integrated on the grid structure together with the CAL modules. Actually, both the construction and testing of the TKR towers have been completed, and all towers have been delivered to SLAC for integration. Before being integrated on the grid, TKR towers are tested with cosmic rays. Table 1 shows the average efficiencies of the SSD layers composing the TKR towers delivered to SLAC. All TKR towers exhibit an average efficiency greater than 99%, with the exception of the first tower (labelled as tower "A"), that shows a slightly lower efficiency. Actually, 14 towers have already been integrated, and the remaining towers are ready to be installed. Table 1. Average efficiencies of the SSDs composing the TKR towers delivered to SLAC. T K R ID Efficiency (%)

A 98.59

B 99.59

1 99.51

2 99.62

3 99.43

4 99.62

T K R ID Efficiency (%)

7 99.68

9 99.69

10 99.67

11 99.68

13 99.74

12 99.68

5 99.68 14 99.68

6 99.61 8 99.77

3. Study of the performance of the LAT T K R towers To study the performance of the integrated LAT TKR towers, we have analyzed the cosmic ray data samples. The present analysis has been performed using the data samples obtained in the 8-tower configuration. Further data samples are being taken during the stages of the LAT integration. For or analysis we selected a data sample consisting of single muon events fully contained in a single TKR tower. The TKR trigger condition requires the coincidence among 3 consecutive SSD planes in both x and y views. Hence, the selected muon tracks have to cross at least 6 SSD layers. About 2 x 106 events survived to these cuts. 3.1. Study of the ToT in SSD

layers

Figure 1 shows the distribution of the ToTs in the SSD layers crossed by muon tracks. We have also examined the dependence of the ToT in the track layers on the geometrical parameters describing muon tracks, i.e. the zenith angle 6 and the azimuth angle . Figure 2(a) shows the average ToT as a function of l/cos#. As could be expected, the average ToT increases linearly with l/cos#. As discussed

955

SO

100

ISO

200

ISO

Time over Threshold (DAC)

Figure 1. Distribution of the ToT in the SSD layers crossed by muon tracks. A DAC unit corresponds to 200ns. in section 1, the ToT is proportional to the energy deposited in SSD layers, that is proportional to the track length. Hence, the ToT has to be depend linearly to the track length, that is proportional to 1/ cosO. The dependence of the ToT on the azimuth angle <j> is more complicated, because SSD strips are oriented along two orthogonal direction. For tracks inducing signals on groups of strips, the ToT will be proportional to the maximum pulse amplitude. A useful physical quantity that can be introduced is the projected track length, I', that is the projection of the track length I along the strip view:

V

l\/cos2 6 + sin 2 0 cos2 <> / for x-view layers 2 I y cos 9 + sin 6 sin for y-view layers

(1)

Figure 2(b) shows that the ToT increases linearly with the ratio l/V. 3.2. ToT in triggering efficiency

layers:

evaluation

of the hit

capture

After a trigger is being issued, a trigger acknowledgment signal (TACK) is generated, that takes a time of about 2/is to be sent to the readout chips. If the pulse amplitudes are still over the threshold when the chips are reached by the TACK signal, the corresponding hits are captured. On the other hand, if the pulses are below the threshold, the hits are lost and a null ToT is recorded. The study of ToT in triggering layers is useful because it allows to evaluate the hit capture efficiency. To generate a trigger, a 6-fold coincidence is required among 3 consecutive pairs of x-view and y-view layers. Hence,

956 ISO

1....!....!....!....!....!

A

so 70

y>

M

60

160

* " * 140

\ -A

Ql20

yiso &40 3»

\ / \ if"...,,:

; ^

\

^100 :

IP | 30

•*

•n; 60

20

.....

\

t

\ ! i

40

10

0

20 • i • • • • i • • • 1 1 1 1 • • i • • • • i . . . .

•••I

0.5

I

i.5

2

2.5

3

3.5

4

4.5

J

5

J.5

2

I/cos(Theta)

2.5

3

i • • • • i • •

3.5

4

W

Figure 2. Average ToT in the SSD layers crossed by muon tracks as a function of the track parameters: (a) ToT as a function of the inverse cosine of the zenith angle; (b) ToT as a function of the ratio between the track length and the projected track length along the strip view.

the probability of noisy layers being involved in the trigger is negligible. The hit capture unefficiency 1 — e is evaluated as the fraction of events with null ToT. From table 2 it is evident that the hit capture unefficiencies are always less than 10~ 3 , except for the tower "A", that has been mounted on the bay 0. Table 2. towers.

Hit capture unefficiencies of the first 8 integrated T K R

Bay ID (1 - e) x 10 4

0 32

1 8.2

4 7.9

5 8.1

8 8.0

9 9.1

12 6.4

13 5.4

4. Conclusions The construction and testing of the LAT TKR towers has been completed and their integration is currently in progress. We have performed an analysis of the cosmic ray data collected by the first 8 integrated towers, that shows that the TKR behaviour is consistent with expectations. References 1. GLAST- Scientific and Technical Plane, AO 99-OSS-03

PERFORMANCE OF NEUTRON DETECTOR AND BOTTOM TRIGGER SCINTILLATOR OF THE SPACE INSTRUMENT PAMELA

V. M A L V E Z Z I F O R T H E P A M E L A C O L L A B O R A T I O N IN FN and University of Rome "Tor Vergata", 1, Via della Ricerca Scientifica, 1-00133, Rome, Italy E-mail: Valeria. malvezzi@roma2. infn. it

PAMELA (Payload for Antimatter Matter Exploration and Light-nuclei Astrophysics) is a satellite-borne experiment designed to study charged particles in the cosmic radiation over the largest energy range ever achieved. The apparatus is composed by a magnetic spectrometer, to determine the charge of the particles, an electromagnetic calorimeter and a neutron detector (ND), which perform the particle identification; in addition, an anticoincidence system, a time of flight (TOF) system and an additional shower tail catcher scintillator (S4) located below the calorimeter complement the instrument. The system composed by the electromagnetic calorimeter, the S4 and the ND will look for events with high energy release (more than 100 GeV) in the calorimeter, and with a larger field of view than that of PAMELA (~ 900 cm2sr). In this work we will show preliminary results concerning the performances of ND and S4 detectors coming from data collected during ground tests (January-May 2005) with PAMELA in final configuration.

1. I n t r o d u c t i o n The final particles identification (i.e. positrons, electrons, antiprotons, protons, etc.) is provided by the combination of the calorimeter [1] and the neutron detector informations plus the velocity measurements from the ToF system at low momenta. For high-energy particles, both electromagnetic and hadronic cascades, produced in our apparatus, cannot be fully contained in the calorimeter. Since neutrons are produced only by hadronic interactions, the neutron yield is 10 -T- 20 times larger in an hadronic cascade than in an electromagnetic one. Therefore, adding a neutron counter to our apparatus can improve the capabilities of the calorimeter to distinguish between primary electrons and hadrons [2], [3]. The main task of the Bottom Scintillator is to detect particles passing 957

958 through and then send the signal to the trigger system for elaborating the main trigger pulse. S4 gives also information for lepton/hadron rejection. Infact, when the signal output detected by S4 corresponds to more then 10 minimum ionizing particles, this signal goes to the trigger system and its coincidence with the main trigger is the mark for the Neutron Detector to read out the data from its counters. Finally ,an additional task of the Bottom Scintillator is to play the role of trigger for detecting super high energy electrons coming from the lateral sides of the PAMELA apparatus, out of its field of view. Such electrons would produce a very high flux of neutrons in the Calorimeter and in the Neutron Detector. This task will be fulfilled by using a S4 signal exceeding the threshold of 200-300 minimum ionizing particles, due to the large number of secondaries produced by a particle with energy more then some hundreds of GeV.

2. The Neutron Detector and B o t t o m Scintillator instruments

Figure 1. Left: view of the PAMELA flight Model. Right: [top] view of the bottom scintillator S4, [bottom] view of the neutron detector.

The main difficulty related to the detection of neutrons is that they have mass but no electrical charge. Because of this they cannot directly produce ionization in a detector, and therefore cannot be directly detected. This

959

means that neutron detectors must rely upon a conversion process where an incident neutron interacts with a nucleus to produce a secondary charged particle. These charged particles are then directly detected and from them the presence of neutrons is deduced. The Neutron Detector is attached to the bottom of the PAMELA apparatus and is situated beneath the scintillation detector S4 (fig. 1). It is made of proportional counters, filled with 3He, surrounded by a polyethylene moderator enveloped in a thin cadmium layer. The 36 counters are stacked in two planes of 18 counters each, oriented along the y-axis of the instrument. The size of the ND is 600 x 550 x 150 mm3. S4 has an area of 482 x 482 mm,2 and a thickness of 10 mm; six PMTs are situated along the two opposite sides of the scintillator.

3. On ground cosmic-ray acquisition

.:i

« •

_

*

Rome data Samara data

* •

s

\

cu

> • * * M -

1

1

i

1

1

Energy (mip)

Figure 2. Efficiency of the scintillator S4 vs. the energy realeased in the last plane of the calorimiter.

During the assembling at the laboratories of the University of Rome "Tor Vergata" (Italy) and once PAMELA reached Samara (Russia) to be integrated inside the spacecraft, the system was tested with ground muons.

960

The first session of cosmic-ray acquisition started in Rome the 12th of February until 24th of March 2005; a total acquisition time of about 480 hours has been collected. In this time interval the "good" events collected by Neutron Detector and S4 were 286275 and 344071, respectively. The second session of cosmic-ray acquisition started in Samara the 12th of May 2005 until 24th of same month (acceptance test); a total acquisition time of about 140 hours has been collected. 132448 and 133884 of "good" events have been collected in this time interval by Neutron Detector and S4, respectively. From data we have analyzed PAMELA ND and S4 response. About S4, the correct use of the three calibrations available for the detector have been cheeked together with the threshold value, useful when the instrument is in trigger. To study the efficiency of the scintillator we selected the good (at least one track in TRK) events, interacting or not, releasing a positive amount of energy in the last calorimeter plane (plane 22, view X). The results are reported in figure 2. Both in Rome and in Samara data sample, the statistics for high energy is rather low, therefore it is not possible to study S4 efficiencies for energy released in Calo22x > 6 MIPs. Concerning the ND performance, in figure 3 the acquisition rate of background neutrons for the two acquisition sessions are reported. The red histograms correspond to the counting of the bottom half of the

Figure 3. Acquisition rate of background neutrons for the two acquisition session in Rome (left) and Samara (right).

ND and the blue ones to the counting of the upper half of the ND. Each bin contains neutrons counted in 5s.

961 The behaviour of neutron detector with good interacting (applying CALO standard cuts) particles, both for Rome and Samara data, seems to be consistent with test beam data collected in September 2003 at CERN facility. Some differences exist due to the low particles energy collected on ground. Neutron distribution for straight particles is consistent with expectations (no neutrons produced). 4. Conclusion Concerning S4, the threshold works in a good way when the detector is in trigger, and the efficiency increase with energy; this is important for trigger ND at high energy to discriminate positrons, protons and electrons. About ND, background performance shows right behavior of the detector with an acquisition rate consistent with licterature, and the upper and lower part of the detector behave consistently. References 1. M. Boezio et al., Nucl. Instr. fc Meth. A 487, 407 (2002). 2. A.M. Galper for PAMELA collaboration, Proceedings of ICRG, 2219 (2001). 3. 50. Y.I. Stozhkov for PAMELA collaboration, Int. J. Modern Phys. A (2005).

HIGH G R A N U L A R I T Y SILICON B E A M M O N I T O R S FOR W I D E R A N G E MULTIPLICITY B E A M S

A. MOZZANICA*, F. RISIGO Universita degli Studi di Milano and INFN Milano A. BULGHERONI, M. CACCIA, C. CAPPELLINI, M. PREST Universita degli Studi dell'Insubria and INFN Milano L. FOGGETTA CIFS, Consorzio Interuniversitario

per la Fisica Spaziale, Torino

B. BUONOMO, G. MAZZITELLI, P. VALENTE INFN, Laboratori Nazionali di Frascati E. VALLAZZA INFN Trieste

The DAFNE BeamTest Facility (BTF) at the INFN Laboratories in Frascati provides electron and positron beams in the energy range tens of MeV-750 MeV in a wide range of intensity, from 1 up to 1010 particles per pulse. The pulse rate is 50 Hz. This paper describes the implementation of two types of high granularity silicon beam monitors: two pairs of silicon strip detectors readout by single particle ASICs for the low multiplicity range (1-100 particles) and a silicon strip detector with charge integrating electronics to cover the remaining range (100-1010). Both the silicon detectors are characterized by large dimensions (up to 9.5x9.5 cm 2 ) and a granularity in the 100 /im range. The paper describes the two systems and the results obtained during several dedicated runs.

'Corresponding author, address: Universita dell'Insubria, sede di Como, via Valleggio 11 22100 Como, Italy; e-mail: [email protected]

962

963

(a)

(b)

Figure 1. (a) The silicon beam chambers inside the aluminum box. (b) The SUCIMA hybrid.

1. Introduction The DAFNE Beam Test Facility1 (BTF) at the INFN Laboratories in Frascati provides electron and positron beams in the energy range tens of MeV750 MeV in a wide range of intensity, from 1 up to 10 10 particles per pulse. The high current LINAC beam can be extracted directly to the BTF or to a 1.7 Xo W target to produce a shower. The development of a silicon on-line monitor covering all the energy and multiplicity range can be a true improvement of the Test Facility capabilities. Section 2 will describe the silicon monitors, Section 3 the multiplicity reference calorimeter, section 4 and 5 low and high multiplicity results. 2. Silicon monitors description Two different silicon based systems have been assembled and used: • The low multiplicity monitor, made of four 8.9x8.9 cm 2 228 izm pitch silicon strip detectors (Micron Semiconductor ltd) 380 /xm thick (Fig. la). Each detector is AC-coupled and readout by 3 TAA1 ASICs 2 characterized by low noise, analog readout and self triggering capabilities. The chambers are able to measure the beam profile with an expected spatial resolution of ~50 /xm in low multiplicity conditions. The 4 detectors are organized in a double x-y configuration, with a 15 mm gap between the planes. • The high multiplicity monitor consisting of a 9.5x9.5 cm 2 , 410 /xm

964 thick silicon strip detector (HAMAMATSU Photonics) developed for the AGILE satellite 3 and used in the SUCIMA4 configuration (Fig. lb). The 121 fim pitch 768 DC-coupled strips are readout by 6 128 channel charge integrating VA_SCM2 ASICs 2 , characterized by 4 possible gains (gain 1-4 from lowest to highest) and a maximum charge range of 41 pC/channel. 3. N a l calorimeter at low voltage Testing the silicon beam monitor multiplicity response needs an indipendent method for measuring the number of arriving particles. A 15 Xo Nal(Tl) calorimeter has been used with photomultiplier, shaping amplifier and peak sensing ADC readout. Since the expected multiplicity range extends up to 105 particles, the PMT bias voltages had to be reduced to very low values. Starting from a single particle calibration point at 600 V the gain ratio is computed for each bias step down to 140 V (Fig. 2a).

__ .

*

10

100

200

300

400

500 600 PMT Bias (Volt)

(a)

(b)

Figure 2. (a) Calorimeter gain vs. bias curve. At 140 V a range of 10 5 particles is reached, (b) Example of a 2d profile obained in single electron mode on chamber 2.

4. Low multiplicity beam chamber results Fig. 2b shows the beam profile measured by the beam chambers in single electron mode. The tracks are fully reconstructed and the angular resolu-

965

tion is limited to a mrad level by the low distance between tracker planes. If more than one electron per bunch is crossing the chambers, no tracking is possible due to the presence of ghosts, but a profile can still be extracted, using a charge weighted clusters (Fig. 3a). The total charge of the detector clusters is proportional to the number of incoming particles, so this value can be used for an event by event estimation of the particle crossing with a ± 1 uncertainy (Fig. 3b).

'60

J

3 SO

20

idtuajurf J\rflrf m-rflX

50

100

*t"

150

200

%M* M I M

250

300

4

350 channels

(a)

5

6 7 multiplicity

(b)

Figure 3. (a) Id profile with silicon chambers at an average multiplicity of ~10 (b) Total cluster charge in silicon vs. multiplicity as given by the Nal calorimeter.

5. High multiplicity SUCIMA hybrid results In case of high multiplicity the integrated profile of the SUCIMA detector is computed event by event, after pedestal subtraction, to obtain the total charge (in ADC units) collected by the detector. This charge is proportional to the number of particles. At the maximum gain the detector starts to be sensitive to a few hundreds of electrons (Fig. 4a) and is still linear up to 5-104 particles (Fig. 4b). This is the maximum allowed multiplicity at the BTF with the W target in place. The maximum number of particles before electronic saturation can be extimated measuring the ASIC gain ratio (Fig. 5a); being 36 the ratio between the lowest and the highest gains the maximum multiplicity is 36-5-104=1.8-106. This result refers to a 2.35 cm sigma gaussian beam and has to be scaled down for narrower beams. The

966

multiplicity response of the detector was checked to be indipendent from both the beam size and the particle energy.

tJOOO

5 fOOO 2500 2000 1500 1000 500 0

0

200 400 600 800 100012001400160018002000 Expected multiplicity

(a)

10000

20000

30000 40000 50000 Expected multiplicity

(b)

Figure 4. (a) Mean integrated signal for multiplicities up to 2000 particles, with gain 4 and a beam energy of 433 MeV (b) The integrated signal in the 2000-5000 multiplicity range, with gain 4 and electrons of 112 MeV . The uncertaines on the calorimeter data become relevant since the errors due to the bias lowering procedure sum up.

Tests have been done removing the W target so that the whole LINAC beam could reach the detector. The calorimeter has to be removed, thus no indipedent multiplicity measurement was provided. With the slits full opened the detector shows saturation at any gain (Fig. 5b), but the system is still able to return useful information on size, position and tail structures.

6. Conclusions Silicon beam chambers have proven to be a very useful tool for the BTF operation in both single electron and low multiplicity mode. Silicon detectors with charge integrating ASICs show a linear behavior in measuring the multiplicity from some hundreds to 5-104 particles per bunch, covering the whole BTF range when the W target is in place. Futhermore, the high multiplicity monitor is able to provide position, size and tail structure information in a much wider range ( up to 1 0 8 - 9 particles per bunch). A future beam monitor dedicated detector is foreseen; it will feature a smaller size,

967

-

i

i

, 150

200

250

300

350

400

450 500 channels

J :::^E«™ 250 300 350

400

450 500 channels

•s,

' ! 'r

\

- • •

300

(a)

400

500

600

700 channels

150

200

(b)

Figure 5. (a) Profile of the beam at 2 different gains, (b) Beam profiles without W target, during slits opening (up) and with full opened slits (down), with the lowest gain.

to match the beam pipe dimensions, a finer pitch, to decrease the number of particles per strip thus increasing the range and an AC coupling of the strip with the frontend to avoid pedestal contributions. References 1. G. Mazzitelli et al., Nucl. Instr. and Meth. in Phys. Res. A 515/3, 516-534, 2003. 2. www.ideas.no 3. G. Barbiellini et al, Nucl. Instr. and Meth. in Phys. Res. A 490, 146-158, 2002. 4. C. Cappellini et al., Nucl. Instr. and Meth. in Phys. Res. A 527, 46-49, 2004.

ENVIRONMENTAL TEST ACTIVITY ON THE FLIGHT MODULES OF THE GLAST LAT TRACKER M.BRIGIDA, A.CALIANDRO, C.FAVUZZI, P.FUSCO, F.GARGANO, N.GIGLIETTO, F.GIORDANO, F.LOPARCO, B.MARANGELLI, M.N.MAZZIOTTA, N. MIRIZZI, S.RAINO'. P.SPINELLI For the GLAST Dipartimento

Collaboration

Interateneo di Fisica, Universita di Bari and

INFN-Bari,

Via Orabona, 4 Bari, Italy The GLAST Large Area Telescope (LAT) is a gamma-ray telescope consisting of a silicon micro-strip detector tracker followed by a segmented Csl calorimeter and covered by a segmented scintillator anticoincidence system that will search for y-rays in the 20 MeV-300 GeV energy range. The results of the environmental tests performed on the flight modules (towers) of the Tracker are presented. The aim of the environmental tests is to verify the performance of the silicon detectors in the expected mission environment. The tower modules are subjected to dynamic tests that simulate the launch environment and thermal vacuum test that reproduce the thermal gradients expected on orbit. The tower performance is continuously monitored during the whole test sequence. The environmental test activity, the results of the tests and the silicon tracker performance are presented.

1. Introduction GLAST, the Gamma Ray Large Area Space Telescope, is a high energy gammaray astronomy mission planned for launch in September 2007. It is composed of two main instruments: the Large Area Telescope (LAT), to search for y-rays in the energy range from 20 MeV to 300 GeV, and the Gamma-ray Burst Monitor (GBM) for high variability phenomena studies at lower energies. The LAT is a gamma-ray telescope consisting of a silicon micro-strip detector tracker followed by a segmented Csl calorimeter, to reconstruct y-rays direction and energy. The tracker and the calorimeter are covered by a segmented scintillator anticoincidence system to reject charged particle background [1]. 968

969 The Large Area Telescope is based on the conversion of gamma-rays into electron-positron pairs and is arranged in a 4x4 modular array of towers, consisting of one Tracker and one Calorimeter module. Each Tracker module (that will be referred to as a "tower", from now on) corresponds to a stack of 19 carbon fiber panels (trays) supporting the silicon strip detectors and the electronics. With the exceptions of the top and bottom trays of the tower, each tray supports silicon detectors both on upper and lower faces. A more detailed description of the LAT instruments can be found at reference [1].

2. Environmental Test flow of the LAT Tracker Towers All the trays and all the towers have been subjected to an environmental tests sequence before being delivered to SLAC for the Tracker assembly to verify that all of them are acceptable for flight and can operate in space environment. At the moment these proceedings are written (October 2005), all the 16 flight towers have been assembled, tested and delivered to SLAC, where 14 of them have already been assembled in the flight grid structure. After each tower assembly, the environmental test sequence consists in the performance of vibration and thermal vacuum tests preceded and followed by functional tests to verify the electrical integrity and optimum functionality of silicon detectors and electronics after dynamical loads and thermal transitions. These tests have been performed in the Alenia-Alcatel Assembly, Integration and Test facility in Rome. The GLAST INFN-Bari group is responsible for the environmental testing of the TKR towers, completed in October 2005. 2.1. Vibration Test The goal of the vibration tests is to demonstrate that the hardware is acceptable for flight and that it will survive to the environments imposed during the launch. This is accomplished by subjecting the flight Tracker modules to tests at acceptance levels. LDS shake tables are used for testing the towers. The transfer function of the towers is evaluated by subjecting it to low level signature sweep test from 5 to 2000 Hz that are performed before and after each dynamic proof to compare the tower behaviour when it is subjected to the dynamic environment. Changes in the fundamental frequency and/or amplitude are used to identify possible structural damage that may have occurred during the dynamic test sequence. Figure 1 shows the responses of one flight TKR tower to the pre- and post-test low level signature sweeps along the Z axis: the first normal mode is at about 370 Hz.

970

The dynamic environments are simulated by sinusoidal vibration in the 5 to 50 Hz frequency range and random vibration from 20 to 2000 Hz. The response under dynamic excitation is studied both along the thrust axis (Z) first and lateral directions (X,Y) successively. Harmon* Spectrum A6 . 1

!!!!!!!!!!!!!!!! !!!!:

in! !!!i in! mm

wm-w .0°

!•!

I!!!!!!!!!

!!!!::!l!!:f!!!IM:!!l

::;:::::.:::;: ::'.: :;:: :.::•. :;::;.

?iii;

10

m l l l l j H Mi.-iii

•; !::!

I'M !l;l

I'll '!!!j

Figure 1 Pre-tcst (blue line) and post test (red line) responses to low level signature sweep.

The random vibration is performed at various levels starting from -12 dB, 6dB up to the full level with an input spectrum of 6.8 g ^ . A summary plot showing the Z axis normal modes and Q values of all the flight towers tested is shown in Figure 2. All the normal mode frequencies measured are between 350 and 400 Hz [2].

>

o

-o_ -v

-i_ 'r,

-^>

*7u "a. •*. ti

"/£, Ti

-o

.*% <^_ •%

s*.

*•*. i i

-"v

•" \, W* % % \ % \ %%* \ *••*. ^ \ % \ % ->r %. \ •9

<>

•»

tp,

•?

•»

»

*

*

•*

"%,

d

'

"ip *

•*«

'.*

">

<>

"V ^ p

'*

Figure 2 Summary plot of the first mode frequencies and Q values for all the flight TKR towers.

971

2.2. Thermal vacuum tests Thermal vacuum tests on flight hardware are performed to demonstrate that the tower modules will get over the thermal gradients expected during the mission and that the functional capability of the hardware is not degraded by thermal transients. For each tower four thermal cycles are performed in the -15°C + +45°C temperature range at a vacuum level of 10"5 Torr. A cold plate is used to drive the towers in temperature by conduction. Moreover, heaters located on an external box surrounding the towers provided a better warming up of the modules. Functional tests are performed during transients and at hot/cold plateaus to check the silicon detectors performance as a function of temperature. Moreover, at each temperature extremes the Turn Off/Turn On capability is verified. The temperature is monitored by means of thermocouples located on the tower sidewalls and on the ground support equipment and, also, by means of the thermistors located on the tower flex cables. Figure 3 shows the temperature profile as a function of time measured by the thermistors [2].

12/3-<2005 Q0 00 00

12/32005 12 00 00

13/9/2005 00.00.00

10/9/2005 12 00 00

14/9/2005 00 00.00

14/9/2005 12.00 CO

15/9/2005 00 00 00

15/3/2005 12 00 00

16/9/2005 00 00 00

Dat«

Figure 3 Temperature profile measured by the tower thermistors as a function of time.

2.3. Functional tests Electrical tests are performed before and after all the environmental proofs. Comprehensive Performance Tests (CPT) and Limited Performance Tests (LPT) include the verification of the load and read-back capability of registers and front-end ASICs, the complete monitoring of the noise, gain and noise occupancy of all the channels in a Tracker module (~55000), the check of trigger lines, cosmic rays data acquisition.

972

The main objective of the functional tests is to verify that the tower performance is in conformance with the requirements established for the detector. The measurement of the gain and noise of all the channels of a tower allows to identify the dead strips (gain less than 50 mV/fC)) and the disconnected strips (noise less than 500 ENC). The single strips noise occupancy test searches for the noisy strips: any channel with a noise occupancy larger than 10"3 is considered noisy. Functional tests are also continuously performed during thermal vacuum transients and at hot/cold dwell points (-15°C, +45°C) in order to study how the SSDs parameters change with temperature. The summary plot of the disconnected channels measured before, during and after one tower test is shown in Figure 4 .

Figure 4 Summary of the disconnected channels identified in a complete sequence of functional tests before during and after a thermal vacuum test.

3. Conclusions The environmental test sequence including dynamic loads, thermal vacuum cycles and functional tests have been successfully performed on all the 16 flight Tracker modules. The assembly of the LAT Tracker will be completed at SLAC by the end of October 2005. GLAST launch is scheduled for September 2005. References 1. 2. 3.

http://www-glast.slac.stanford.edu/pubfiles/proposals/bigprop http://www.ba.infh.it/~glast/env test.htm GLAST Bari Environmental Test Web Page http://www-glast.slac.stanford.edu/ GLAST LAT Internal Document: LATTD-00191-02

THE TIME OF FLIGHT DETECTOR AND TRIGGER FOR THE PAMELA EXPERIMENT IN SPACE S. RUSSO, G. BARBARINO, D. CAMP ANA, G. OSTERIA Dip. Scienze Fisiche Universita Federico II and INFN Sez. Via Cintia Napoli 80126, Italy

diNapoli

M. BOSCHERINI, W. MENN, M. SIMON Universitat-GH Siegen, FB Physik Siegen 57078, Deutschland

The electronics of the Time of Flight telescope and trigger of PAMELA experiment are described. The time resolution requested by the ToF system must be less than 120 ps. The contribution of the digitization electronics is negligible if the TDC resolution is < 50 ps. The peculiarity of the developed electronics arises from the need to obtain such a time resolution associated to a wide dynamic range for charge measurements, operating in satellite environment, which implies low power consumption, radiation hardness, redundancy and high reliability

1. Introduction The PAMELA [1] (a Pay load for Antimatter Matter Exploration and Lightnuclei Astrophysics) is a satellite born experiment scheduled to be lunched at the end of 2005. The primary objective of PAMELA is to measure the energy spectrum of cosmic ray protons and electrons with a special care on their antiparticle counterparts and the search for antimatter covering an energy range and with a sensitivity unreachable by previous similar experiments. Another experimental goal is to measure the antihelium to helium ratio with a sensitivity of the order of 10"7. PAMELA is built around a permanent magnet spectrometer equipped with double-sided silicon sensor tracker, which will be used to measure the sign, absolute value of charge and momentum of particles. The tracker is surrounded by a scintillator veto shield (called anticounters) that will reject particles that do not pass cleanly through the acceptance of the tracker. Below the tracker is a silicon-tungsten calorimeter, to measure the energy of electrons and allow 973

974

topological discrimination between electromagnetic and hadronic showers. A scintillator telescope system will provide the primary experimental trigger and time of flight particle identification. A scintillator mounted beneath the calorimeter will provide an additional trigger for high-energy electrons. This is followed by a neutron detector system for selections of very high-energy electrons and positrons (up to 3 TeV). 2. The PAMELA ToF The Time-of-Flight (ToF) system [2] is composed by 6 layers of fast plastic scintillators arranged in 3 groups (SI, S2 and S3) read out by 48 photomultipliers tubes (PMTs). The ToF must fulfill the following goals: • Provide a fast signal for triggering data acquisition in the whole instrument; • Measure the flight time of particles crossing its planes; once this information is integrated with the measurement of the trajectory length through the instrument their velocity p can be derived. This feature enable also the rejection of albedo particles; • Determine the absolute value of charge z of the incident particles through the multiple measurement of the specific energy loss dE/dx in the scinitllator counters. Additionally, segmentation of each detector layers in strips can provide a rough tracking of particles, thus helping the magnetic spectrometer to reconstruct their trajectory outside the magnet volume. The ToF is divided in 6 layers, arranged in 3 planes (SI, S2, S3), each plane composed of 2 layers (Sll, S12, S21, S22, S31, S32). The overall geometry of the ToF for PAMELA is summarized in tab. 1

Sll

8

Strip dim. (nun x mm) 330x51

S12

6

408 x 55

7

385

S21

2

180x75

5

375

S22

2

150x90

5

450

S31

3

150x60

7

420

S32

3

180x50

7

350

Plane

No. of strips

Thickness (mm) 7

Section area (mm2) 357

975 3. The ToF and Trigger electronics The ToF and trigger electronics is a complex system made by 9 boards. These are the 6 Front-End (FE) boards, the DSP board and the two identical trigger boards (one called "hot" and one called "cold"). The purpose of this system is to collect the signals coming from 48 PMTs of the ToF, measure their arrival time with respect to the trigger pulse and their charge, generate signal for the trigger, handle the busy logic for all subsystems and interfacing the ToF system with the general data acquisition system of the PAMELA apparatus. The ToF electronics has to comply to the following requirements: •

Guarantee a wide dynamic range in the charge measurement to allow the measurement of energy release of z > 1 particles up to Carbon; • Add a negligible contribution to the overall ToF time resolution. This implies a contribution from time digitalization, which has to be less or equal than 50 ps. In the mean time, the electronics must ensure low power consumption, a high reliability in terms of radiation hardness and temperature range. Therefore a great effort has been paid in implementation both hardware and software redundancy, that is to mean the replication of at last all critical components.

3.1. Front-End boards Each front end board receives the analog signals coming from 8 PMTs. For each channel the input is split in two branches, which are fed into the time and charge section respectively. The first section measures the arrival time of the signal with respect to the trigger pulse, and generates signal for trigger information. The other section measure the charge of the PMT signal.

3.1.1. Time section In the time section of the FE board, each anode line is coupled to a fast discriminator. Each discriminated signal is in turn split: on one side, the signal is shaped ad sent to trigger board. On the other side, it is fed to a double-ramp Time-to-Amplitude-to-Time converter (TAT). The fixed level of our discriminator threshold generates a pulse height dependency of the measured time due to the finite rise time of the pulses. Simply stated, signals which are generated at the same time, but with different amplitude, will cross the discriminator threshold at different times, the

976

difference being greater the higher the threshold level is set. To minimize this time-walk effect, one should use a threshold as low as possible; but lowering the threshold the comparator increases the probability to be crossed by noise pulse. The double-ramp time expansion works in the so called COMMON STOP mode: the arrival of the PMT pulse produces the start of the fast ramp, while the trigger signal (COMMON STOP) starts the slow ramp. The moment when the PMT pulse crosses the threshold corresponds to the START of the ramp: a low loss, low thermal drift capacitor C is charged with a high stability constant current source Ic. The charging of the capacitor continues until a STOP signal arrives, namely the trigger pulse. The time interval Tc (typically around 60-65 ns, mainly due to the trigger signal generation and propagation) between the START and STOP signals is the one of the interest, and is given by

Where V is the voltage across the capacitor. With the arrival of the STOP signal, the capacitor is slowly discharged with a much lower constant current Id. The discharge time td is given by

'

h

This time interval is measured by a Time-to-Digital Converter (TDC), which has an internal clock at 100 MHz. Therefore the time digitalization resolution related to the TDC least significant bit (LSB) is 10 ns. The ratio between the discharging and charging currents sets an expansion factor (independent of the capacitor's value) a=Ic/Id. Form the equation follows that the relationship between the expanded time and the time to measured is

Td=^Tc=aTc Hence, if the expanded time Td is measured with a resolution of 10 ns, Tc is measured with a resolution which is a time lower. In our case a=200 and the intrinsic time resolution of our electronics is 50 ps. 3.1.2. Charge section The anode current coming from PMT is collected by a charge amplifier which provides an output voltage which is proportional to the total current. This peak will quickly discharge exponentially, therefore the output of the amplifier is connected to a so-called "pulse stretcher", where a J-FET charges up rapidly a capacitor at the peak value of the input waveform. The capacitor is linearly

977 discharged with a (quite low) constant current source. Clearly, the discharge time is once again proportional to peak voltage, and therefore to the PMT total current. Likewise the time section, also in the charge section of the FE board the discharge time is measured with the TDC, whose gate is opened by trigger signal. 4. The Performance 4.1. Qualification test The performances of the ToF and Trigger electronics has been subject to extensive test. The time resolution and integral non-linearity of the flight model of the FE boards have been measured with an AGILENT 81132 pulse generator (RMS jitter of the time base = 15 ps +/- 0.001 % of the delay). The results, shown in fig. 1, fully satisfy the design specifications. NU error <± 50 ps (±1 LSB)

Figure 1. Time resolution measurement results of the PAMELA FE boards

4.2. Flight model test The performance of the ToF system and trigger are measured in the PAMELA integrated flight model. Cosmic muons are acquired using the ToF scintillators and trigger board in order respectively to detect the particles passing through the apparatus and to trigger and to manage the data acquisition. Data are taken in different working conditions relative both to the HV values as well as to the threshold values of the ToF system in order to evaluate the trigger efficiency and the time resolution as function of the working conditions. 4.2.1. Efficiency measurement The efficiency of single paddle is evaluated using a set of perfectly reconstructed tracks pointing in the paddle. The tracks selection criteria requires

978

a good reconstructed track from the tracker in agreement whit the position of the track as reconstructed from the timing information of the ToF in the planes different from the one under study. The overall measured efficiency is more than 98% 4.2.2. Time resolution measurement In order to obtain a first evaluation of the time resolution of the single paddle, the position of the track along the paddle is measured from the tracker and compared to the one obtained from the difference of the TDCs. Fig.2 shows the result for a paddle of S2.

W€ ckmw!

Figure 2. Time resolution, in unit of 50 ps, for a S2 paddle measured in flight conditions

Acknowledgments We thank to M. Boscherini and S. Ricciarini. We also thank V. Marzullo, R. Rocco, E. Vanzanella, for help in construction of ToF system and its integration in the PAMELA apparatus References 1. 2.

O. Adriani et. Al., IEEE Transaction on Nuclear Science Vol 51 n.3 854 (2004). G. Osteria et. al, Nuclear Instrumentation and Methods A535 (2004), 152,157.

F U N D A M E N T A L P H Y S I C S I N ESA'S COSMIC VISION PLAN

BERNARD F SCHUTZ* Albert Einstein Cardiff Email:

Institute, Potsdam, Germany and University, Cardiff, UK [email protected]

ESA's Cosmic Vision document sets out the most important and exciting scientific questions that European scientists want to see addressed by space missions in the time-frame 2015-2025. Cosmic Vision also marks a breakthrough for fundamental physics: for the first time, a major space agency has given full emphasis in its forward planning to missions dedicated to exploring and advancing the limits of our understanding of deep physical issues, including gravitation, unified theories, and quantum theory. In my talk I will present the conclusions of the Cosmic Vision document and discuss how it may be implemented. If we are to see experiments in space by 2015, exploring quantum measurement theory or looking for violations of deeply held beliefs about space and time, then the community needs to get ready quickly to make proposals next year.

1. Cosmic Vision: the key questions The European Space Agency (ESA) has recently completed its forward planning exercise for missions in the period 2015-2025. This plan, the successor to its Horizons 2000+, is called Cosmic Vision.1 For the first time, ESA envisions a systematic program of missions to perform high-precision experiments in Fundamental Physics. For this reason, a new constitutency of physicists — accustomed to working in university or accelerator labs — needs to be aware of the new possibilities that can be open to them. Cosmic Vision is structured around four key questions that cross the usual internal ESA discipline boundaries (astronomy, solar system science, and fundamental physics). These questions seem to be the drivers for most of the research that the European space research community wants to perform a decade from now. They are: * Chair, Fundamental Physics Advisory Group, European Space Agency

979

980

(1) What are the conditions for planet formation and the emergence of life? (2) How does the solar system work? (3) What are the fundamental physical laws of the universe? (4) How did the universe originate and what is it made of? In this article I will concentrate on how ESA might help to answer Question 3, on the funamental laws of physics. Of course, there are important implications for fundamental physics from missions designed to answer Question 4. Cosmology, the study of inflation and the current acceleration of the universe: these may well be answered only by fundamental theory. These issues will be addressed in the paper by Professor Barcons. Here I will focus on the possibilities for missions that directly perform fundamental physics experiments. Cosmic Vision is the result of a long consulation exercise. In 2004 ESA issued a call for scientists to submit theme proposals: short documents pointing at the principal questions that seem interesting and answerable by missions ten to twenty years from now. The community is well aware of ESA's current suite of missions and projects, and in particular what questions they are likely to answer. The question that ESA posed was: given the current program, what should happen afterwards? The community responded with over 150 theme proposals. There was a remarkably even division among the three traditional ESA research areas. In particular, there were as many proposals in fundamental physics as in astronomy or solar system science. The proposals were distilled by discussions within the science advisory committees of ESA, in particular in the Space Science Advisory Committee (SSAC) and its working groups in the three areas. The resulting synthesis was presented to the community at a town meeting in Paris, 15-16 September 2004. The feedback obtained there was incorporated into the plan, which was written early in 2005 and presented to the Science Programme Committee (SPC: ESA's principal science committee, which must authorise Cosmic Vision) in May 2005. With the approval of SPC, the plan was published in October 2005. 2. Key Question 3: What are the Fundamental Physical Laws of the Universe? Cosmic Vision divides this key question into three sub-themes: • Topic 3.1: Explore the limits of contemporary physics: where do we

981 look for evidence of unified theories? • Topic 3.2: The gravitational wave universe: a LISA follow-on in the 1 Hz waveband. • Topic 3.3: Matter under extreme conditions: investigating the Extreme Universe, the corners where really exotic conditions allow us to probe the deepest laws of gravity and nuclear/particle physics. These are amplified in the following sections. 2.1. Topic 3.1: Exploring physics

the limits of

contemporary

Theorists are convinced that our understanding of physics today is incomplete and even inconsistent. General relativity is a classical theory, the rest of basic physics is described in a quantum way. Quantum measurement theory, governed by the experimentally robust "Copenhagen Interpretation", does not extend in any obvious way to questions regarding the Big Bang or quantum black holes. The highly successful Standard Model of the strong and weak interactions is known to be incomplete. Unifying gravity with the other interactions presents a further challenge. Most of the work on unified theories and quantum gravity is driven by aesthetic principles and demands of mathematical consistency: there is a paucity of experimental data. There are many scenarios in which data that could help resolve these issues is obtainable in low-energy experiments, by performing "standard" experiments with extremely high precision, looking for deviations from the standard expectations. Such experiments can usefully be done in space: the stable gravity-free environment allows long-duration high-precision experiments. Here are some potential experimental techiques and technologies that could be transported into space: • Cryogenic accelerometers based on superconduction test masses and readouts (SQUIDS), pioneered by NASA's current GP-B mission, described elsewhere in this volume. • Cold-atom sources for a variety of experiments. • Bose-Einstein condensates, built from the cold atoms. • Atom trap, atomic lasers. • Ultra-stable optical lasers, microwave sources, Raman lasers. Some of these techniques have already attracted Nobel Prizes in Physics, including this year (2005. They are ready to be transferred into space,

982

where applications of equal interest are waiting. Given the large number of possible experimental techniques, and the small size of each likely experiment, Cosmic Vision suggests that an appropriate tool is a Fundamental Physics Explorer Programme. This would be a series of small satellite missions in Earth orbit, all using the same basic platform, several instances of this platform could be manufactured on an assembly line, realising considerable cost savings. It may, with careful design, be possible to launch 3 such satellites in succession within an envelope of cost that would normally characterise a medium-cost ESA mission. What would such a platform provide? Here are some preliminary ideas: • A 3-axis stabilised spacecraft with drag-free control. • Low-vibration environment without moving parts (e.g. bodymounted solar array instead of deployable units). • Sun-synchronous, low-altitude (500-700 km) circular orbit. • Limited total mass to allow for an optimised launch vehicle. • Mission lifetime typically 1 year. It will be important to refine these ideas in the very near future. I will come back to that at the end of this paper. Here are some of the exciting and challenging themes proposed by ESA's community, which could be supported by a plaform like this, and which could therefore become experiments in the Fundamental Physics Explorer series: • High-precision test of the equivalence principle (either using the design already developed for the proposed STEP satellite, 2 or employing a newer cold-atom approach). Unified theories all predict that the equivalence principle will be violated at some level, although predicting the size of the violation is harder. Finding it would be a breakthrough of enormous importance. • Test the inverse square law of gravity at small distances (also looking for evidence of a fifth force). "Braneworld" scenarios suggest possible string-theory effects at macroscopic distances. 3 A deviation of this size would again be a result of enormous importance. • Test the inverse square law at scales of 104 km, a few times larger than the Earth. This could be performed by putting a sensitive gravity gradiometer in elliptical orbit around the Earth. • Put very precise atomic clocks into orbit. This is already a field in

983

which there have been important space missions and experiments, but one that can probe the limits of conventional physics. Time, after all, is the basic metrology standard today; even the standard metre is mow defined in terms of the second and an agreed, fixed value of the speed of light. • Seek time-dependence of the fundamental constants. This would, if found, undermine the most fundamental tenets of physics, including the famous CPT theorem. • Test the isotropy of space. Like the time-dependence of constants, this would require major changes in the basis of physical theory. • Perform demanding tests of quantum measurement theory, including decoherence and entanglement. Earth experiments on entanglement 4 have lately shown how closely the Nature conforms to the expectations of the Copenhagen interpretation of quantum measurement theory, especially that the wave function "collapses" at the moment of measurement. How this passes over to the classical behaviour of macroscopic systems (decoherence) is poorly understood. But the distinction between "observer" and "observed", between quantum and classical, is hard to support in unified theories, which hope to describe the quantum evolution of the universe as a whole, and to resolve the dark energy question. Any of these experiments could change physics forever. In addition to these scientific payoffs, the emerging technology of cold atoms could eventually lead to technology spinoffs. In particular, the use of atomic interferometry to do jobs that light interferometry does today offers big increases in precision, due to the much shorter wavelength of matter waves. Future space missions in astronomy and Earth observation may use ultra-high-precision gyroscopes, pointing systems, ranging systems, and station keeping, all based on the kind of cold atom technology that has been developed for fundamental physics experiments. The paper by Prof. Schleich in this volume gives a much more detailed description of the possibilities for cold-atom experimental physics in space. 2.2. Topic 3.2: The Gravitational

Wave

Universe

The joint ESA-NASA mission LISA is planned for launch in 2013, and will open the low-frequency gravitational wave window, observing between 10~ 4 and 10~ 2 Hz. 5 Already, ground-based observatories like LIGO, VIRGO, GEO, and TAMA 6 are operating in the high-frequency band, above a few

984

Hz, with best sensitivities around 100 Hz. The fact that ground-based detectors cannot observe at lower frequencies motivated the need for LISA. But going into space has given LISA another advantage: its sensitivity is very high because it can operate with very much longer arms than are possible on the ground. While ground-based detectors will struggle to extract weak signals from instrumental noise, LISA will be confusion-limited, fighting over much of its band more against a background of gravitational waves from distant sources than against instrumental noise. Between the LISA and LIGO bands is an intermediate frequency band not so far covered by any planned detector, from about 0.1 Hz to a few Hz. This is a particularly interesting wave band because source confusion is probably not a serious problem, and it is therefore a clean window through which to look for gravitational radiation from the Big Bang. 7 At LISA frequencies and below, it is likely that astrophysical sources of gravitational waves produce random backgrounds that greatly exceed anything anticipated from the Big Bang. At ground-based frequencies the strength of the radiation expected from the Big Bang is very weak, making it hard to detect. The 1 Hz waveband seems to be the ideal place to look right back to the earliest moments of the history of the universe. Standard inflation theories make predictions of the level of stochastic radiation that one might expect, simply from the parametric amplification of quantum fluctuations at the time inflation begins. This level is rather low, and would require a big program of technology development to make it accessible to a successor to LISA. But non-standard variations on the conditions that led to inflation can produce much more radiation, enough to be detectable even by LISA itself.8 It should also be noted that gravitational waves in the LISA waveband had, when blue-shifted back to earlier times, a wavelength comparable with the horizon size when the universe was going through the electroweak phase transition. There is thus some possibility of radiation from the dynamics of this phase transition itself. This could be a spectral feature for LISA and it might even extend into the 1 Hz band. Although the astrophysical sources in this band are expected to be sparse, they are nevertheless interesting. They are typically transient, since anything able to radiate gravitational waves strongly with intrinsic timescales of 1 s will quickly evolve through the loss of energy to gravitational waves. Most sources are expected to be binaries of neutron stars and stellar-mass black holes, in various combinations, passing through this band on their way to coalescence at frequencies around 1 kHz. Other systems could be mergers of intermediate-mass black holes, which are expected to

985

have been formed in abundance in the first generation of stars, and which coalesce at around lOtM/lOOOM©) -1 Hz. A reasonable goal for a LISA follow-on might to be to detect and study every such system in the universe that passes through the 1 Hz waveband during the mission. By detecting these signals and following them for several months, such a mission will be able to measure their distances to accuracies of a few percent, and thereby pinpoint the onset of star formation as the universe evolved. In addition it would determine the populations of such binaries, measure the mass distributions of their components, and test models of the evolution of the first generation of stars. Given the strong indications that star formation and heavy element production began remarkably early, such a mission might be the only tool we have to study individual stellar systems at such early times. Cosmic Vision identifies a Gravitational Wave Explorer mission as the appropriate mission to follow LISA, near the end of the planning timeframe (2025). LISA itself is hoped to observe from 2015 for ten years, so the new mission could provide continuity when LISA turns off. The Gravitational Wave Explorer would have a design similar to that of LISA, but with shorter arms (appropriate for its higher frequency band) and nextgeneration technology: bigger mirrors, stronger lasers, improved drag-free control. Development of this technology should start soon. Mirrors and lasers can progress right away. Next-generation drag-free control will benefit from the lessons learned in LISA Pathfinder and LISA itself. As with LISA, a partnership with other agencies would also be desirable. NASA is exploring an even bigger jump in technology to what is called the "Big Bang Observer", with enough sensitivity to see a cosmological background at the level suggested by standard models of inflation. Such a system would require two or more LISA-like configurations. The challenges are significant, but there are cost savings in building identical spacecraft all at once, and perhaps also in distributing launch costs among two or more cooperating agencies.

2.3. Topic 3.3: Matter

in Extreme

Conditions

We learn the most about physics by examining places where matter is subjected to extreme and unusual conditions. Near black holes, matter at high temperature, high density, and with strong magnetic fields somehow generates the quasar phenomenon. Do quasar jets come from a combination of "normal" effects, essentially magnetohydrodynamics in some form, or

986 does it involve the direct extraction of rotational energy from the central black hole? What, indeed, does the central black hole look like in such extreme systems: is it really a Kerr hole, is it rapidly rotating? Similarly, neutron stars are objects whose detailed physics still defies a convincing description in the nearly 40 years since the discovery of pulsars. Their combination of degenerate matter, superfluidity, superconductivity, strong magnetic fields, exotic nuclear physics, and rapid rotation place them among the most fascinating physical objects in the universe. Understanding their interior physics would provide deep insight into nuclear physics and the high-density details of the Standard Model, and could in principle reveal new physics outside the Standard Model. These systems can be investigated through the X-rays and gamma rays that they emit. Cosmic Vision identifies two useful tools: a large-area X-ray telescope mission, known in Europe as XEUS and in the USA as Con-X, would have sufficient sensitivity to do detailed spectroscopy and time-series analysis on the emission. It could identify normal mode frequencies of neutron stars (a key to their interior structure), measure the mass and spin of central black holes, test the metric outside a black hole against the Kerr model by measuring a number of expected orbit-related frequencies in the emission, and unravel the mystery of quasar jet production by enabling detailed disc and jet models to be fit to the high-quality data. A later mission that did spectroscopy at even higher energies, in the gamma-ray band, would complement these studies very nicely by looking at the very central regions of quasars and accreting neutron stars. Cosmic rays currently present one of the most challenging puzzles to our understanding of matter. The flux of high-energy cosmic rays (above about 10 20 eV) is much higher than expected: such high energy protons should rapidly lose energy as they travel through, and scatter, the photons of the cosmic microwave background. If the current ground-based Pierre Auger Laboratory observations verify that the flux is anomalously large, then a more sensitive, space-based detector capable of looking at cosmic ray showers over a large portion of the Earth may be required in order to get enough statistics to understand the properties of these particles and their sources, which have not been identified.

3. N e x t steps Cosmic Vision is the plan for 2015-2025 created by the SSAC. To implement it, the SPC must allocate resources. If SPC agrees to the plan proposed by

987

SSAC, then we can expect the following timetable: • In the Spring of 2006, ESA will issue a call for proposals for the Cosmic Vision time-frame. • Proposals will be due at the end of 2006. • A number of proposals will be selected for assessment studies, leading to a further review and to the start of Phase A studies in January 2008 for two or three missions. • The number of missions approved for construction and launch will depend, of course, on the results of the Phase A studies and on the level of resources available. It is expected that approved mission(s) will be able to launch in the first three years of Cosmic Vision, 2015-2018. • Further calls will be issued later for the second and third "slices" of Cosmic Vision, 2018-2021 and 2021-2025. For a substantial part of this ambitous program to become reality, the science budget of ESA must at a minimum hold steady or, hopefully, increase gradually even in the period from 2009 onward, when missions are in preparation. One cannot be completely optimistic here: the purchasing power of ESA's annual science budget has fallen by more than 20% in the last ten years. SSAC hopes that, by presenting the members states with an exciting and rich program of missions, with a strong demand from the European scientific community, it will motivate a reversal of this depressing trend. Whether this is successful will depend not just on the ESA program, but also on whether European space scientists make their ambitions and requirements clear to their own national governments. But besides these long-range strategic issues, there is an immediate requirement: European scientists must get ready to respond to the call for mission proposals in 2006. In particular, the fundamental physics community, a large part of which has little experience of space missions, must turn their ambitious suggested themes into serious and practical proposals. To assist this new community, ESA is planning to organise a meeting in early May 2006 on the theme of the Fundamental Physics Explorer series. At this workshop, participants will be able to give ESA staff feedback about what features are desirable in the common platform that will host the experiments in this series. And they will be given suggestions by ESA staff on what is expected from a good proposal. Dates and venue for this meeting will be announced soon. The success of Cosmic Vision depends on excellent proposals from the community!

988

References 1. Cosmic Vision: Space Science for Europe 2015-2025, European Space Agency, Paris (2005). 2. J W Cornelisse, Class. Quantum Grav. 13 A59-A65 (1996). 3. L. Randall and R. Sundrum, Phys. Rev. Lett. 83 4690-4693 (1999). 4. P. Walther, et al, Phys. Rev. Lett. 95, 020403 (2005). 5. K. Danzmann, et al., LISA Pre-Phase A Report, Max- Planck-Institut fur Quantenoptik, Report MPQ 208, Garching, Germany (1996). 6. J. Hough and S. Rowan, Living Rev. Relativity 3, (2000); URL (cited on 19 October 2005): http://www.livingreviews.org/lrr-2000-3. 7. C. Ungarelli and A. Vecchio, Phys. Rev. D 6 3 064030 (2001). 8. A. Buonanno, M. Maggiore, and C. Ungarelli, Phys. Rev. D 5 5 3330-3336 (1997).

T H E AMS-02 ELECTRONICS S Y S T E M

F. R. SPADA University

of Rome "La Sapienza", P.le Aldo Moro 5, Rome, Italy E-mail: spada<8romal. infn. it

AMS-02 is a precision TeV spectrometer that will fly during 3 years on the International Space Station (ISS) to perform direct search of antimatter and indirect search of dark matter. The different subdetectors require in total about 300,000 electronic channels that deliver 7 GBits per second of data. The electronics that has to provide these capabilities and, at the same time, has to work in the adverse conditions of outer space, is currently being builded and tested by the AMS-02 collaboration.

1. Introduction The AMS-02 (Alpha Magnetic Spectrometer) will operate on the International Space Station performing spectroscopy of cosmic rays, directly searching for antimatter (i.e., antihelium nuclei) and indirectly searching for dark matter, trying to identify the products of the annihilation of candidates such as neutralinos 1 . For the AMS-02 electronics, high performance technologies used in high energy physics on Earth have been adapted to work in low Earth orbit. Offthe-shelf components have been selected and extensively tested, allowing a drastical reduction of the costs: this approach led to the implementation of high performance, high reliability space qualified electronics. Different subtedectors build up the apparatus: a transition radiation detector (TRD), a time of flight (ToF) with and anti-coincidence counter (ACC), a silicon strip tracker, a ring image cherenkov counter (RICH), and an electromagnetic calorimeter (ECal). The detector, together with the electronics crates arranged on radiators, is shown in figure 1. A unified approach has been adopted to meet the requirements imposed by the different subdetectors, by the safety rules imposed by NASA to fly on the Space Shuttle and on the ISS, and by the physics goals. 989

990

Figure 1. The AMS-02 detector with the electronics crates located on radiator panels.

2. The ISS interfaces 2.1. Electrical

interfaces

The electrical power comes from the ISS solar arrays. They provide AMS with less than 3 kW of power at 109 to 124 VDC, with very stringent electromagnetic compatibility requirements. 2.2. Data link

interfaces

Two data transmission lines are available from the ISS: low rate and high rate links. The Low Rate Data Link (LRDR) is based on the MIL-STD-1553B dual serial bus and is used in AMS-02 for commanding and monitoring. The available data rate is 10 Kbit/s with duty cycle of 55% to 90%. At most, a 120 byte command is allowed each second. In addition, critical data are transmitted continuously at 10 byte/s. The High Rate Data Link (HRDL) has a 125 MBaud rate over fiber optic cables. This stream is used for the transmission of the event data and of a copy of the monitoring data. The downlink is performed on Ku-band via Tracking and Data Relay satellites to the AMS-02 Payload Operations and Control Center (POCC) on the ground in real time. The allocated

991 downlink bandwidth for AMS-02 is 2 Mbit/s on average. Moreover, on board of the ISS, the AMS-02 Crew Operation Post (AC OP) computer records all the data on hard disks for a total capacity of 30 days of acquisition, or about four months if disks are swapped. It can also play back the recorded data if necessary, and provides a dedicated monitoring post and an alternate path to the detector.

3. The data acquisition chain 3.1. Requirements

from the

subdetectors

The TRD discriminates high-energy positrons from protons and consists of 20 layers of radiator/straw tube planes for a total of 5,248 tubes (needing one electronics channel each) in which the gas gain control is critical. The signal size is 100 fC. The Tracker measures the trajectory of charged particles in the 0.8 T magnetic field of the superconducting magnet with a resolution of 10 [im in 192 ladders with 1024 readout strips each. The signals are again of a few fC, and shielding from electromagnetic interference, power supply filtering and careful signal processing are needed. The ToF is composed by 4 scintillator planes with 144 photomultipliers in total. To measure the particle velocity, a 100 ps accuracy is required. The energy deposited is also measured to determine the charge Z. The ACCs have 16 photomultipliers. The total number of channels is 1,536. Together, ToF and ACC provide the primary trigger. The RICH identifies particles by measuring the location and number of photons collected by 680 16-pixel photomultipliers. The required dynamic range of each pixel channel (21,760 in total) is about 10,000. The ECal discriminates electrons from hadrons by shower shape and energy deposition measured in 324 4-pixel photomultipliers with a dynamic range of 60,000 on each of the 2,916 channels. ECal also provides a trigger signal for high energy photons. In total the detector requires 227,300 channels, each providing 16 bits of information for each event, with trigger rates up to 2 kHz. The resulting raw data rate is over 7 Gbit/s. To adapt the data flux to the allocated 2 Mbit/s downlink rate, during the data acquisition process the electronics must reduce the event size and filter out mistriggers.

992

3.2. Data acquisition

process

The data acquisition tree is shown in figure 2. Even though the requirements imposed by the various subdetectors differ in the details, a unified approach was adopted for their electronics, updating to the new technologies available the successful experience gained with the AMS-01 engineering flight2'3. In this unified approach, for all the subdetectors, analog signals enter a detector specific Application Specific Integrated Circuit (ASIC) and are first shaped and then held in response to a trigger that has been generated using signals from some of the detectors, then multiplexed with the ASIC. After digitization via an ADC, the output is then sent over into a Common Digital Part (CDP) and buffered into memories. With multiplexing, each CDP collects data from up to 1,000 different subdetector signals. The trigger recognizes that a charged particle has passed through the detector when a coincidence of fast signals from the ToF counters and no ACC signals happens. From the AMS-01 experience a total trigger rate from 200 to 2,000 Hz is expected, depending on the geomagnetic latitude. Thus, the electronics system is being conservatively designed to run at twice these rates. The RICH and the Tracker have two independent CDPs, while ECal, ToF and TRD have two CDPs in cold redundancy. The following data acquisition circuitry is the same for all the subdetectors: asynchronously with subsequent triggers, the data within a CDP is reduced using a subdetector specific algorithm in a digital signal processor and shipped over a custom developed 10 MByte/s serial link into the global data acquisition tree. The next node is a Command Distributor and Data Concentrator Subdetector DAQ j {

CDP: Common Digital j ^art

Global DAQ r

CDDC~\co***BilBh1rib,

DSP

DSP

[^^]^=|M|]r|AM}| j^te

GA

Sample ASK I & Hold

Mem

as

7

DataCtilkcL

s

««»

i H£?"-pJ—

DC-OC pBgh

ElBamy)

2»Vfillcr(UO)

Figure 2. Chain of the Data Acquisition in AMS-02.

993

(CDDC) circuit that receives data from up to 24 CDP. Here the data from an event are treated in order to be passed to the Main DAQ Computer, where the detector performances are monitored and a first event analysis is performed: the selected events are then buffered and sent out over the HRDL when it becomes available. 4. Construction status No access is possible to the electronics once the detector is in space, and moreover, it also has to withstand the launch stress (vibration and depressurization), the large thermal range and the vacuum. Thus, redundancy principles have been adopted, and all of the components and circuits used for the electronics have gone through a rigorous and extensive testing process to ensure that they will meet the performance requirements and operate reliably in the prohibitive environmental conditions. Tests of the single components and of the assembled boards include vibration, radiation, thermal and thermal-vacuum resistance. The serial production of most of the flight boards has been started at the ChungShan Institute of Science and Technology, Taiwan, and will be completed this year, together with that of the Main DAQ Computer. References 1. F. Barao, Nucl. Instrum. Meth. A535 134 (2004). 2. M. Aguilar et al., Phys. Kept 366 331 (2002) [Erratum-ibid. 380 97 (2003)]. 3. M. Pohl, Nucl. Phys. Proc. Suppl. 122 151 (2003).

L A U N C H I N O R B I T OF T H E SPACE TELESCOPE PAMELA A N D G R O U N D DATA RESULTS

R. S P A R V O L I F O R T H E P A M E L A C O L L A B O R A T I O N University

of Rome

"Tor

Vergata"

and IN FN structure

of "Tor

Vergata"

PAMELA is a satellite-borne experiment that aims to measure the antiproton and positron spectra in the cosmic radiation over the largest energy range ever achieved, and to search for antinuclei with unprecedented sensitivity. In addition, it will measure the light nuclear component of cosmic rays and investigate phenomena connected with Solar and Earth physics. All detectors have been successfully integrated in the apparatus that has been installed on-board the Russian ResursDK1 satellite. In the first months of 2006 PAMELA will be launched from the Baikonur cosmodrome in Kazakhstan, for a 3 year long mission. In this paper an overview of the mission and the instrument will be presented, along with results of cosmic ray muons recorded at ground during the final integration phase.

1. Experiment overview The PAMELA* apparatus is designed to study charged particles in the cosmic radiation. It will be hosted by a Russian Earth-observation satellite, the Resurs-DKl, that will be launched into space by a Soyuz rocket in 2006 from the Baikonur cosmodrome. The orbit will be elliptical and semi-polar, with an inclination of 70.4° and an altitude varying between 350 km and 600 km. The mission will last at least three years. The main scientific goal of the experiment is the precise measurement of the cosmic-ray antiproton and positron energy spectra. The satellite orbit and mechanical design of the apparatus will allow the identification of these particles in an unprecedented energy range (between 80 and 270 MeV for positrons and between 80 and 190 MeV for antiprotons) and with high statistics (~ 104 p and ~ 105 e + per year). The importance of these measurements stems from the information that they can provide about solar modulation effects (below a few GeV), cosmic-ray propagation and primary production from exotic sources such as primordial black holes, annihilation of supersymmetric particles or Kaluza-Klein particles. Almost a

a Payload for Antimatter Matter Exploration and Light-nuclei Astrophysics.

994

995 T0F(S1) |—1> a u D O D D a D

^f—I

^MTieeiMCIPEMCB

SPECTROMETER

CALORIMETER

Figure 1. A sketch of the PAMELA apparatus.

all data available so far have been obtained by balloon-borne experiments. New and extensive measurements with high statistics are strongly needed. Additionally, PAMELA will search for antimatter in the cosmic radiation (sensitivity to the He/He of ~ 10~ 8 ), it will measure the light nuclear component of cosmic rays and investigate phenomena connected with Solar and Earth physics. 2. The apparatus PAMELA The apparatus : is composed of the following sub-detectors, arranged as in Figure 1, from top to bottom: a time of flight system (ToF (S1,S2,S3)), an anticoincidence system (CARD, CAT, CAS), a magnetic spectrometer, an electromagnetic imaging calorimeter, a shower tail catcher scintillator (S4) and a neutron detector. Particles trigger the experiment via the main trigger b provided by the ToF system 2 composed of 6 layers of segmented plastic scintillators b

Additional triggers can be provided by the calorimeter and S4.

996

arranged in three planes (SI, S2, S3). The ToF system also measures the absolute value of the particles charge and flight time crossing its planes. In this way downgoing particles can be separated from upgoing ones. Furthermore, the acceptance of the apparatus can be varied by changing the configuration of the ToF layers used to form the trigger. Particles not cleanly entering the PAMELA acceptance are rejected by the anticounter system 3 . The rigidities of the particles are determined by the magnetic spectrometer 4 consisting of a permanent magnet and a silicon tracking system. Thus, positively and negatively charged particles can be identified. The final identification (i.e. positrons, electrons, antiprotons, etc.) is provide by the combination of the calorimeter (see 5 ) and neutron detector information plus the velocity measurements from the ToF system at low momenta. The detector is approximately 120 cm high, has a mass of about 450 kg and the power consumption is 360 W. A daily amount of about 10 GByte of data from the instrument are expected. The main downlink center is located at the Research Center for Earth operative monitoring "Nts-OMZ" in Moscow (Russia); an average of 8 passages per day over the station are foreseen, during which PAMELA will download all scientific data along with telemetry data about the status of the detector (temperatures, power consumptions, voltage levels, ..). An additional station, located in Khanty-Mansiisk (Siberia, Russia), is under consideration, but has not been officially established yet.

3. Ground data results Prior to the delivery to Samara (Russia), where the spacecraft is built, the PAMELA apparatus was assembled at the laboratories of the University of Rome "Tor Vergata", Italy. Here the system was tested with ground muons over a period of several months, in order to calibrate the sub-detectors and check the overall performance of the instrument. As a whole, a total of about 480 hours of ground cosmic rays have been collected. Figure 2 shows an event recorded in Rome. The event is a 1.5 GeV/negatively charged particle, with high probability of being a \x~~ considering the clean non-interacting pattern in the calorimeter. All PAMELA detectors are shown in the figure along with the signals produced by the particle in the detectors and derived information. It can be clearly seen the highly detailed information provided for each cosmic-ray event. The solid line indicates the track reconstructed by the fitting procedure of the

997

Figure 2.

A muon track in PAMELA.

tracking system. From this information the muon charge ratio measured at ground in Rome (50 m a.s.l.) has been obtained and shown in Figure 3. Muons were selected as non interacting particles in the calorimeter and charge one in the ToF scintillators. Their momenta were determined by the tracking system. In the figure PAMELA data are compared with other experimental results 6 ' 7 ' 8 . A good agreement can be seen. Once PAMELA reached Samara, it was extensively tested again before being integrated inside the spacecraft. As a result, about 140 hours of cosmic ray acquisition have been recorded. Analysis of Samara data is in progress. 4. S t a t u s of t h e mission Extensive space qualification tests of PAMELA detectors, electronics and mechanical structures have been performed. Furthermore, the detectors were tested at test beam facilities at CERN and shown to comply with their design specification. Three models of the PAMELA apparatus were developed: a mass/thermal model, a technological model for electrical and compatibil-

998 2.5

2.25 2

— o * o

Hebbeker and Timmermans (world average 2002) BESS-95 (Tsukuba, Japan) BESS-97/98/99 (Lynn Lake, Canada) CAPRICE-98 (Fort Sumner, New Mexico)

1.75 1.5 " 1.25 1

0.75 0.5

0.25

PAMELA preliminary (Rome, Italy) I

1

i

i

i

i

10

i

i

10

momentum (GeV/c) Figure 3. A t+ /^ ratio as measured by PAMELA during ground data acquisition preliminary results. ity tests with the satellite, and the flight model for actual d a t a taking in space 9 . All these models were delivered for testing to Samara, and shown to be fully compliant with t h e requirements of the Resurs-DKl spacecraft environment. Specifically, t h e flight model underwent incoming and acceptance tests in April/May 2005 and was integrated with the satellite in September 2005. Electrical and d a t a transmission tests of t h e integrated PAMELA-spacecraft system are still taking place. Subsequently, P A M E L A and the satellite will be t r a n s p o r t e d t o Baikonur for the launch into space, foreseen for the beginning of 2006 after about two months of tests in-situ. References 1. 2. 3. 4. 5. 6. 7. 8. 9.

M. Bongi et al., IEEE Trans. Nucl. Sci. 51, 854 (2004). D. Campana et a l , 28th ICRC, Tsukuba, 2141 (2003). S. Orsi et al., 35th COSPAR, Paris, to appear in Adv. Sp. Res. (2004). O. Adriani et al., Nucl. Instr. & Meth. A 511, 72 (2003). M. Boezio et al., Nucl. Instr. & Meth. A 487, 407 (2002). T. Hebbeker and C. Timmermans, Astropart. Phys. 18, 107 (2002). M. Motoki et a l , Astropart. Phys. 19, 113 (2003). M. Boezio et al., Phys. Rev. D 67, 072003 (2003). R. Sparvoli et al., 35th COSPAR, Paris, to appear in Adv. Sp. Res. (2004).

CZT DETECTOR DEVELOPMENT FOR NEW GENERATION HARD-X/GAMMA-RAY ASTRONOMICAL INSTRUMENTS MICHELA USLENGHI, GIANCARLO CONTI, SERGIO D'ANGELO, MAURO FIORINI, EGIDIO M. QUADRINI INAF - Istituto di Astrofisica Spaziale e Fisica sez.Milano

cosmica

LORENZO NATALUCCI, PIETRO UBERTINI INAF - Istituto di Astrofisica Spaziale e Fisica sez.Roma

cosmica

In the context of the definition of a future European gamma-ray mission, following the now on-orbit INTEGRAL observatory, we are carrying out a feasibility study on a Gamma Ray Wide Field Camera (5-500 KeV) for transient event detection. Recent achievements in high energy astronomy have validated the CZT detectors performances in terms of good spatial resolution, detection efficiency, energy resolution and low noise at room temperature. We started a development program aimed to explore the possibilities to improve and optimize the performance of this kind of detectors, acting at the level of both the readout system and crystal quality. Preliminary results of characterization of pixelated crystals provided by BMARAD (now Orbotech) are presented, along with their analysis and interpretation based on an analytical model of signal formation.

1. Introduction Gamma Ray Astronomy was born fifty years ago with the first evidence of relevant emissions from extra solar systems. Since the beginning, it was clear that this large amount of energy was generated by violent events related to collapsing or collapsed compact objects, but the difficulty to realize instruments with the proper sensitivity and resolution capacity, strongly delayed a clear understanding for such phenomena. During last decade, thanks to the improved performances of the new generation instruments, it was provided a new description of the sky in the hard X-ray and Gamma-ray range. To date the first priority problems in the gamma-ray field are connected to the comprehension of mechanisms leading to the cosmic accelerators and cosmic explosions. These processes, almost invisible in other wavebands, can provide a 999

1000 wealthy of information on the violent Universe answering the most exciting questions on Cosmology and Universe evolution [1],[2]. The capacity of such a radiation to deeply penetrate the matter, the most useful to discover sources behind path density of 1024 H atoms /cm2, on the other hand put hard challenges in realizing instrumentation. The intensity of these incoming signals is very low (except for unpredictable dramatic events) implying very large collecting area. Then the detector must have elevated stopping power to interact with the radiation energy, this conversely imply strong mechanical structures and huge local secondary background generation. This effect can be prevented using small detection area coupled with large area collection, but all focusing techniques developed in the past reduce their efficacy above few tens of keV. Today, a big effort has been put on developing focusing optics for the y-ray range, and Laue lens seams much promising. However, the good performance of the INTEGRAL observatory shows that a mask imager in an energy range from few kev to 1-2 MeV can still be the instrument of choice for several applications. In both cases (focused and codified mask telescopes) the detector is a very crucial element of the instrument, its performance strongly affecting the kind of astronomical measures that the instrument can carry out. In this framework, we started a development study on CZT detectors since their good performance makes them excellent candidates for both applications, if properly optimized. Our development plan includes different lines: crystal growth/processing and contact deposition (in collaboration with CNR-IMEM), modelization, laboratory characterization, readout electronics, Market Research & Cost Analysis for CZT mass production.

2. CdZnTe detectors overview Cd(Zn)Te is a very convenient material for gamma-ray detectors. In fact, it can be operated at room temperature, due to its high band gap (1.4-1.6 eV at 300 K), and it can easily stops high energy photons with a moderate thickness, due to its high density (6.2 g/cm3) and high Z. Unfortunately, due to poor hole transport in the material (compared with other semiconductors widely used in radiation detectors, like Silicon), the signal strongly depends on the photon interaction depth, causing a broadening of the left side of the gamma-ray peak ("hole tailing"). Generally, correction methods, based on waveform shape information, are used to improve the energy resolution. Alternatively, elaborate electrode structures (such as coplanar-grid CZT) are often used to compensate for hole trapping, for example using the so-

1001

called "single-polarity charge sensing", which makes the detector insensitive to the hole contribution. We started a study to address peculiar problems affecting this kind of detectors (e.g. response dependent on the interaction depth and multiple hit events) using a digital approach. This also facilitates operations like pixel to pixel equalization and background rejection. The detector electronic chain thus includes a minimal analog stage for charge pre-amplification, coupled to a flash ADC for waveform digitalization at a high time resolution sampling, and a powerful, FPGA based digital processing unit, devoted to waveform elaboration. In this framework, signal modeling is important not only for performance evaluation and interpretation of calibration results, but also to develop algorithms to be implemented in the digital processing unit.

3. Signal Model In CZT detectors, the incoming radiation produces free-moving charges (holeelectrons pairs) inside the sensitive material, then charges drift in an appropriate electric field, inducing during their motion a charge on an electrode. Then, the problem of calculating the signal shape is essentially the problem of calculating the induced charge and current. We implemented an analytical model, described in a previous paper [3]. Although simplified, the model is in good agreement with experimental data and, contrary to simulation based descriptions, shows clearly the dependency of relevant parameters on physical and geometrical characteristics of the detector, also allowing straightforward interpretation of characterization results. In the following section, some model outcomes are summarized: Charge Collection Efficiency (CCE): the shape of the electrodes play a fundamental role in the CCE: a fine segmented positive electrode makes the detector insensitive to the interaction depth, unless the interaction is very close to the positive electrode. Rise time as estimator of the interaction depth: since for a given energy the amplitude is a function of the interaction depth, a second parameter (generally the rise time Trise) is to be used as an estimator of the depth. Ideally, this second parameter should be a monotone function of the depth. This is not always true for Trise, the range of applicability being strongly dependent on the geometrical characteristics of the detector and on the shape and strength of the electric field. However, for the configuration of the detector we characterized (see next

1002 paragraph), provided that T 6 U of the preamplifier is long enough, the monotonicity is verified in the range of interest, so that x^^ can be used.

4. Experimental results A preliminary characterization of a pixelated CZT crystal (by IMARAD), 5 mm thick, has been carried out and results have been compared to the model. One side is covered with a single, grid-shaped, electrode, whereas on the other side 16x16 square electrodes (2.46 mm pitch) are deposited. The common grid is held at negative potential (respect of the pixel side). The readout electronics (consisting in a charge sensitive preamplifier Amptek A250, rise time=4.5 ns, and a high-pass filter with T~3 US) is connected at one pixel at a time. The output of the front-end electronics is then digitized at 100 Ms/s. The 8 pixels surrounding the one to be readout are held to constant potential in order to form a "guard ring". Model fitting: the analytical model has been fitted to the recorded waveforms, using as fit parameters the energy of the interacting photon and the interaction depth. The mobility lifetime product (U-T) for electrons has been evaluated independently (see below), for holes has been derived from literature. depth :

0.17 mm

depth :

1.61 mm

depth

1.b A

r**

1.0

0.5

0.0

)

rV"YW

J

05 -2.0xWT?5xWT?0x1-e5!0x10"' 0

-2.0xWT?5xWTf0xt«?0xl0"' 0

-2.0xreT?5x1flTfOxve5fox10"' 0

Figure 1. Signals generated by one of the pixels under 57Co irradiation compared with model.

He'te evaluation: under Co57 irradiation, the amplitude of the signal corresponding to the centroid of the 122 KeV photopeak has been measured at different bias voltage and then compared to the dependence of the signal amplitude on bias voltage foreseen by the model for different values of ue-Te (the dependence of the model on Uh-Th is much less important, due to the detector geometry). Best fit value is (VTe«2.75-10"3cm2/V. Hole tailing minimization: in order to correct for the depth dependence of the signal, the recorded waveforms have been classified in groups with similar rise time and for each group the spectrum has been reconstructed. The rise time binning has been chosen to have the same number of waveforms for each group, providing a proper SNR for each spectrum. In this way, the dependence of the

1003 Uncorrected spectrum

Co57(122 KeV)

- l — • — ' — ' — ( -

FWHM 12.6%ffi)

Amplitude (V)

Corrected spectrum -i—'—•—•—r~

FWHM 9.4% rffi

1

i

.

.

L_

" tn—

Amplitude (V)

Figure 2. S7Co total spectrum before and after correction by peak amplitude. photo-peak centroid on the rise time has been evaluated and then used to correct the overall spectrum. References 1. P. Ubertini et al. Unveiling the high energy obscured Universe.hunting esplosive and collapsed objects phisics.,Vioc.291h ESLAB Symposium, Noordijk, 19-21 April 2005, F.Favata, A.Gimenez eds. In press 2. J.Knoedlseder et al. Prospects in space-based gamma-ray astronomyfroc.W* ESLAB Symposium, Noordijk, 19-21 April 2005, F.Favata, A.Gimenez eds. In press 3. E.M.Quadrini, G.Conti, S.D'Angelo, M.Fiorini, M.Uslenghi, L.Natalucci, P.Ubertini - SPIE Proc. Vol.5898-25, in press

This page is intentionally left blank

Tracking Devices Organizer: D. Abbaneo

M. Andreotti L. Bagby G. Batigne F. Bellini N.Chr. Benekos S. Bianco C. Bloch M. Bomben

R. Brauer LA. Cali M. Eppard M. Prey J. Kaminski A. Khomich F. Lehner G. Leibenguth M. Lenzi F. Massa

0 . Militaru

A Barrel IFR Instrumented with Limited Streamer Tubes for BaBar Experiment A New Inner Layer Silicon Strip Detector for DO Calibration System for the Silicon Drift Detector of the ALICE Experiment Experience with the Resistive Plate Chamber in the BaBar Experiment Muon Identification and Reconstruction in the ATLAS Detector at the LHC Micrometric Position Monitoring Using Fiber Bragg Grating Sensors in Silicon Detectors Commissioning of the CMS Tracker Outer Barrel New Effects Observed in the BaBar Silicon Vertex Tracker: Interpretation and Estimate of their Impact on the Future Performance of the Detector Tests of Substructures of the CMS Silicon Strip Tracker Test, Qualification and Electronics Integration of the ALICE Silicon Pixel Detector Modules Layout and Status of the CMS Silicon Tracker The CMS-Tracker Detector Controls System Optimization of the Readout Pad Geometry for a GEM-based Time Projection Chamber Tracking Strategy and Performance for the ATLAS High Level Triggers The LHCb Silicon Tracker The HI Silicon Tracker The LHCb Muon System Two- and Three-Dimensional Reconstruction and Analysis of the Straw Tubes Tomography in the BTeV Experiment Cryogenic Operation of Edge-Sensitive Silicon Microstrip Detectors 1005

1006 H.-G. Moser J.R. Pater

W. Wallraff M. Weber G. van Nieuwenhuizen

The DEPFET Active Pixel Sensor as Vertex Detector for the ILC Final Assembly and Intergration of the ATLAS Semiconductor Tracker and Transition Radiation Tracker 3-Dimensional Position Control for the AMS-02 Tracke with Infrared Laser Beams Development of a GEM-based high Resolution TPC foi the International Linear Collider STAR Inner and Forward Tracking Upgrade

A B A R R E L IFR I N S T R U M E N T E D W I T H LIMITED S T R E A M E R T U B E S FOR B A B A R E X P E R I M E N T

MIRCO ANDREOTTI University/INFN of Ferrara Via Saragat 1, 44-100, Ferrara, Italy E-mail: [email protected] on behalf of the BaBar LST Collaboration

The new barrel Instrumented Flux Return (IFR) of BABAR detector will be reported here. Limited Streamer Tubes (LSTs) have been chosen to replace the existing RPCs as active elements of the barrel IFR. The layout of the new detector will be discussed: in particular, a cell bigger than the standard one has been used to improve efficiency and reliability. The extruded profile is coated with a resistive layer of graphite having a typical surface resistivity between 0.2 and 0.4 MOhm/square. The tubes are assembled in modules and installed in 12 active layers of each sextant of the IFR detector. R&D studies to choose the final design and Quality Control procedure adopted during the tube production will be briefly discussed. Finally the performances of installed LSTs into 2/3 of IFR after 8 months of operations will be reported.

1. The flux return geometry As originally built, the muon and KQL detection system for BABAR consisted of 19 layers of resistive plate chambers (RPCs) interleaved with the flux return iron in the barrel region and 18 layers in the forward and backward endcaps. Each detector gap contained a single layer of RPCs based on a bakelite and linseed oil design. During initial BABAR operations, the temperature in the iron increased to as much as 30°C. Since that time efficiencies have continued to decline at the rate of approximately 1.2% per month. Efforts to reverse or halt the underlying decline in efficiency have not been successful. All the chambers in the forward endcap were replaced in the summer and fall of 2002 with RPCs built with more stringent quality control (the upgrade project foresaw also the replacement of 5 active layers with 2.5 cm thick brass plates to improve the pion rejection. In December 2002 BABAR it has been decided to replace RPCs with limited streamer 1007

1008 tubes (LSTs) for the barrel upgrade. 2. The LST Concept A "standard" LST configuration [1] consists of a silver plated wire 100 ^m in diameter, located at the center of a cell of 9x9 mm 2 section. A plastic (PVC) extruded structure, or profile, contains 8 such cells, open on one side (Fig. 1). The profile is coated with a resistive layer of graphite, having a typical surface resistivity between 0.1 and 1 MOhm/square. The profiles, coated with graphite and strung with wires, are inserted in plastic tubes ("sleeves") of matching dimensions for gas containment. The signals for the measurement of one coordinate can be read directly either from the wires or from external strip planes attached on both side of the sleeve.

Figure 1. Schematic of the "standard" Limited Streamer Tube configuration.

3. R & D studies, Design and Performance More than one year of R&D studies have been done before choosing the final LST design. Our R&D program has been concentrated on several critical issues like: selection of safe gas mixture, rate capability, wire surface quality and uniformity, aging test and performance of the prototypes. Detailed studies concerning all these issues can be found at [2]. The LST tubes are somewhat fragile mechanically so careful design, handling, and operation are of paramount importance in preventing failures.

1009 The "mortality" of the LSTs depends on the cell size, on the care and attention given during construction and installation, and on the strictness of the acceptance tests. Given these constraints to build a highly efficient detector means: reduce tube mortality and/or introduce redundancy to decrease its effect on detector efficiency; arrange tubes into modules that can be extracted and replaced and to minimize the dead space; feed each tube with one or more independent HV channels and finally locate HV distribution boxes and front end electronics on the outside of the detector. Such indications led us to a 15x17 mm 2 cell design (which is more reliable and efficient) where each tube is composed by 7 or 8 cells and assembled in modules. We use wire readout for the azimuthal coordinate, (f>, and strips plans for the z coordinate (along the beam direction). In order to obtain high performances and to respect the safety requirements it has been chosen a ternary gas mixture of Ar/C 4 Hi 0 /CO 2 (3/8/89)% [3]. 4. LST production and quality control The LST production has been commissioned to an external farm, the Pol. Hi. Tech (PHT), which also assembled all the prototype used for R&D. We worked together with PHT to improve the cleanliness of the tubes and to setup a system of stringent quality (QC) control to guarantee the high performance of the product. The QC system was designed to check the quality of all mechanical components, the goodness and the resistivity of the graphite coating, the gas tightness of all assembled LSTs. Finally a set of electrical tests was applied to check the behavior and the performances of each LST under normal and critical electric (e.g. maximum allowed voltage) and physics (e.g. high particle rate simulated with a radioactive source) conditions. In order to be accepted for the installation into BaBar, each LST did have to satisfy all stringent criteria required for each test. 5. Installation into BaBar The first 2 sextants were installed during summer 2004. While the remaining four will be installed in the 2006 shutdown. Installation began with the insertion into each gap of a strip plane covering an entire gap. These strips have been attached to the iron and supported by gravity. Then the modules were brought onto a supporting structure and inserted into the gap on top of the strip layer. The full installation and commissioning of the two sector took less than 2 months and has been completed on schedule. QC tests after installation certify the proper working of the apparatus.

1010

6. Performances of the two LST-sextants BaBar experiment began RUN5 on March 2005 and after 8 months of operation LSTs are showing very good performances concerning number of dead channels, plateau curves, efficiency, muon identification and pion rejection. The main results will be described in the following.

Silent channels. Since the installation, the number of silent channels is quite constant during the time and it fluctuates around 0.2%. There are a total of 2 dead ^-channels (which could be broken wire, broken HV cable or broken signal cables) over 1552 (0.1%) and 7 dead Z-channels (which could be disconnected strips or broken signal cables) over 2284 (0.3%).

Plateau curves. Plateau curves are monitored monthly and all LSTs show a plateau width varying between 300-600 V around the value of 5600 V. In Fig.2 plateau curves of the 4 high voltage channels of a 7-cells LST are shown.

5800

6000 Voltage

Figure 2. Plateau curves (counts versus voltage) of the 4 high voltage channels of a 7-cells LST; the firsts 3 HV channels are are connected to 2 cells, while the 4th channel are connected to one cell, then it shows half counts than the others.

Efficiency. The efficiency of installed LSTs is monitored daily using fifi pairs from colliding beams and monthly from cosmics rays. The calculated efficiency results to be constant around 90%. The geometric efficiency is 92.5%. The fluctuations of the efficiency are mostly related to the fluctuations on the number of silent channels, but no loss of efficiency for each single LST is detected.

1011

Muon ID and pion rejection. The Fig.3 shows the pion rejection as a function of the muon identification efficiency for high and low momentum. Comparing the results from LSTs and RPCs at different time it appears clear that LSTs are working better than RPCs ever did. LSTs introduced an excellent improvement in pion/muon discrimination.

Muon Efficiency

Muon Efficiency

Figure 3. Pion rejection versus muon identification efficiency in IFR barrel region for RPCs during years 2000, 2004, 2005 and for LSTs during the year 2005.

Rate and current versus luminosity. There are a linear correlation between rate/current of the installed LSTs and the instantaneous luminosity. The corresponding slope shows that LSTs will work properly also at the early estimated instantaneous luminosity of 2 • 10 34 c m _ 2 s - 1 , only a firmware upgrade (already planned) of the high voltage power supply will be necessary to improve their performances. The two sextants of barrel IFR instrumented with LSTs is working properly and it is showing good performances in terms of detection efficiency and muon identification. References 1. G. Battistoni, E. Iarocci, M. M. Massai, G. Nicoletti and L. Trasatti, Operation of Limited Streamer Tubes, Nucl. Instrum. Meth. 164 (1979) 57. 2. BaBar LST Group, A barrel IFR Instrumented with Limited Streamer Tubes, (2003) 3. G.D. Alekseev, N. A. Kalinina, V. V. Karpukhin, D. M. Khazins and V. V. Kruglov, Investigation of Self quenching Streamer discharge in a wire chamber, Nucl. Instrum. Meth. 177 (1980) 385.

A NEW INNER LAYER SILICON STRIP DETECTOR FOR DO LINDA BAGBY FOR THE DO COLLABORATION Fermilab, P. O. Box 500 Batavia, Illinois 60510, USA The DO experiment at the Fermilab Tevatron is building a new inner layer detector to be installed inside the existing DO Silicon Microstrip Tracker (DOSMT). The Layer 0 detector is based on R&D performed for the Runllb silicon upgrade that was cancelled in the fall of 2003. Layer 0 will be installed inside the ~2.2 cm radius opening available in the DOSMT support structure with the detector in the collision hall. Layer 0 will reduce the radius of first sampling from 2.7 to 1.6 cm and substantially improve on the radiation hardness of the DOSMT, insuring that the silicon tracker remains viable through Tevatron Run II.

1. Introduction 1.1. Physics Motivation The physics motivations for building a new inner layer silicon strip detector, Layer 0, are to mitigate tracking losses due to detector failures, provide more robust tracking and pattern recognition for higher luminosities, and improve impact parameter resolution. Figure 1 illustrates the increased parameter resolution by a factor of approximately two with the addition of Layer 0 for low transverse momentum particles. This translates into a 20% increase in the efficiency of tagging b jets and the possible resolution of B s meson flavor oscillations. Impact Parameter Resolution 180 160 140 | 120 ^100 80 60 40 20

•

SMT2a simulation -*- DO data, 2a L0-noL1

XV

* No L1

Figure 1. Impact parameter resolution with the addition of Layer 0.

1012

1013 1.2. Mechanical Specifications Layer 0 is designed to fit inside the existing Runlla detector and will utilize much of the existing infrastructure as well as the new electronics and readout chip designed for the cancelled Runllb silicon replacement [1]. The current DO Silicon Tracking detector consists of 12 disks positioned between six 4-layer barrel structures. Figure 2 shows an end view of the current detector with the new Layer 0 superimposed in the center. •jjn+Y)

EAST(+X)

Figure 2: South End View with Layer 0 superimposed in center.

Layer 0 is designed to slide over the beryllium beam pipe and associated mounting flanges that have a diameter of 30.48 mm. The detector is also required to maintain ~1 mm separation from the beam pipe to limit capacitive coupled noise. The new detector clears the inner most layer of the current detector through a 44.04 mm diameter aperture. In addition to the tight mechanical constraints the detector is designed to maximize acceptance (98.5%) and readout segmentation. Figure 3 is an enlarged view of Layer 0 illustrating sensor positions.

Figure 3: Layer 0 end view of sensor positions.

A 6-fold geometry was chosen with an A (inner) layer of sensors positioned at a radius of 16 mm and B (outer) layer sensors at a radius of 17.6 mm. Hybrids are located outside of the active volume and are connected to the sensors with fine pitch analog cables. Z segmentation is limited to eight sensors of 70 and 120

1014 mm lengths by the radial buildup of the cable bundles. This arrangement provides better segmentation near Z=0 and equalizes the load capacitance by having lower sensor capacitance on the strings with longest analog cables. Table 1 summarizes the geometric dimensions of the Layer 0 components. Table 1. Geometric parameters of the Layer 0 detector. Layer

Radius

Z segment

Readout/ (Strip) pitch

Sensor Length

0A

16 mm

inner

71 (35.5) um

70 mm

320, 346 mm

0A

16 mm

outer

71 (35.5) urn

120 mm

167, 244 mm

0B

17.6 mm

inner

81 (40.5) um

70 mm

320, 346 mm

0B

17.6 mm

outer

81 (40.5) um

120 mm

167, 244 mm

Cable Lengths

1.3. Electronics Layer 0 SVX4-based instrumentation is designed to operate within the constraints of the currently installed SVX2 data acquisition system. Newly designed components consist of silicon sensors, pitch adapters, analog cables, hybrids, digital jumper cables, junction cards, twisted pair cables, and adapter cards. 1.3.1. Sensors, pitch adapters Layer 0 is composed of 48 silicon sensors manufactured by Hamamatsu. Inner radius positions utilize 71um pitch sensors while 81um pitch sensors are used at larger radius positions. Both sensor types have intermediate strips that are not read out. A single cable pitch is accomplished by utilizing ceramic pitch adapters mounted on the sensors that also carry decoupling capacitors. 1.3.2. Analog Cables Analog cables provide an interface between the pitch adapters and the SVX hybrids. To minimize mass in the active region, 91 um pitch kapton cables are used. The analog cables are flexible circuits manufactured by Dyconex. The cable lengths vary, with the longest at 34.6 mm, and have a capacitance of .35 pF/cm. Careful design of these cables was required to minimize the capacitance presented to the SVX4 preamplifier. Kapton mesh spacers with ~90% open area are used to separate analog cables thus rriinimizing capacitive coupling. The signal/noise is roughly equal at each Z location with adequate signal available

1015

for tracks incident at the extreme edges of the detector for reliable readout and reconstruction. 1.3.3. SVX4, hybrids, digitaljumper cables, twistedpair, adapter card The SVX4 chip is a 0.25 urn technology silicon readout chip originally developed for the Runllb upgrades [2]. These chips use a protocol similar to the currently installed SVX2 chips but operate with a single 2.5V supply rather than the 3.3-5V supplies needed for the SVX2. Each ceramic BeO hybrids hold two SVX4 chips. Grounding of the SVX4 reference to the support structure is through vias plated through the hybrid to contacts on the co-cured flex circuit on the supports. Digital signals from the hybrid are carried to the end of the support structure using a kapton flex digital jumper cable. The jumper cable is then coupled to a twisted pair cable using a junction card located on the existing silicon support structure. An adapter card with active circuitry, mounted on the wall of the DO calorimeter, interfaces Layer 0 to the existing readout. 1.3.4. Grounding and Isolation Techniques A number of Runllb studies have established that low coherent noise can be achieved by good low inductance ground connections to the support structure [3]. This is accomplished by co-curing mesh ground planes onto the carbon fiber support structure and utilizing low inductance flex circuits which carry bias and ground from the bottom to the top of the sensors. Since Layer 0 is a longitudinally continuous conductor, the potential exists for a serious ground loop encircling the DO calorimeter. The adapter card was designed to provide electrical isolation from DO. This card converts single ended SVX2 control signals, supplied by existing electronics, into differential signals needed by the SVX4. Utilizing the high impedance of the differential signals ground isolation is achieved. In addition, a separate isolated 2.5 Volt supply provides power to the SVX4 chips. The isolation requirement is greater than 10 Ohms. 2. Performance 2.1. Mechanical Measurements The completed Layer 0 detector mechanical dimensions have been compared to aperture measurements of the Runlla silicon detector taken during an access in 2004. The aperture measurements show that Layer 0, with an outer radius of

1016

22.02 mm will have a radial clearance to the existing structures by .86 mm horizontally and 1.67 mm vertically. The 1mm spacing requirement, to prevent capacitive noise coupling from the beam pipe, has also been insured. 2.2. Grounding and Isolation Two noise issues arose while testing Layer 0. The isolated low voltage power supply for the SVX4 chips was found to require a filter to reduce pick-up noise on the readout. Five turns of the power cables on a ferrite core reduced the noise to acceptable levels. A LC filter has been design and is ready for testing. The resistive temperature device (RTD) system also caused noise on the readout. The RTD system consists of a RTD soldered onto a flex circuit affixed longitudinally to Layer 0. The flex circuit is then adapted to cryogenic type wire. A braided shield around the cryogenic wire connected to the Layer 0 isolated ground eliminated the noise. 2.3. Signal to Noise Response Maximizing the signal to noise response has been addressed throughout the design of Layer 0. With a single minimum ionizing particle equivalent to ~ 30 ADC counts, readout test show that a 16:1 noise ratio has been attained with a 200V sensor bias and a ferrite core on the isolated power supply leads. 3. Conclusion A new inner layer silicon detector has been designed, built, and tested for the DO experiment. Strict mechanical specifications, electrical isolation from DO, and a high signal to noise ratio have been demonstrated. Layer 0 will be installed during the next shutdown period scheduled for early 2006. Acknowledgments The Layer 0 project was funded by the National Science Foundation and the Department of Energy. References 1. DO Collaboration, "DO Layer 0 Conceptual Design report," (2003). 2. L. Christofek, et al, "SVX4 User's Manual", DO Note 004251 (2003). 3. K. Hanagaki, Nucl. Inst. And Meth.Phys. A511, 121 (2003).

CALIBRATION S Y S T E M FOR T H E SILICON D R I F T D E T E C T O R OF T H E ALICE E X P E R I M E N T

G. BATIGNE FOR THE ALICE COLLABORATION Istituto

Nazionale

di Fisica Nucleare Sezione via Giuria 1 10125 Torino Italy E-mail: [email protected]

di

Torino

During beam tests of large area silicon drift detectors (SDD), it has been observed that the reconstructed position is affected by inhomogeneities in silicon doping. Such deviations, if not corrected, would introduce errors dominating the position resolution and would jeopardize the detector performance. A laser system has been therefore developed to measure these deviations in detail for the 260 SDDs which will be used in the ALICE experiment. This system allows to extract with precision the fluctuations in resistivity and other systematic errors in the spatial measurements and to check the consistency between laser and beam test results. A method has been devised to drastically compress the data on deviations without affecting noticeably the resolution.

1. Introduction One of the distinctive features of the ALICE experiment[l] is the tracking system, designed to handle events with thousands of tracks. The tracking in the innermost region is performed by six layers of high resolution silicon detectors with different choices of technologies [2]. The two innermost layers use silicon pixel detectors, the two next ones silicon drift detectors (SDD) and the outermost layers consist of double-sided silicon strip detectors. For the layers 3 and 4, 260 large area SDDs will be used. 2. Description of the ALICE S D D s Each SDD of ALICE has a thickness of 300/mi and a size of 7cm by 7.5cm divided into two active zones [3,4]. Along the two longest sides there are 256 anodes with a pitch of 294/um. The electrons created by ionization are first collected in the middle plane of the detector and then drift toward the anodes by the mean of an uniform electric field created by 291 cathodes (pitch of 120/jm), whose potential is decreasing with distance from 1017

1018

anodes. Hence, the mean drift time of the electrons, from the point where the ionisation is generated to the anodes, provides the position along the orthogonal axis while the centroid of the charge collected by anodes gives the position along one axis. Moreover the total collected charge represents an energy loss sample which is used for particle identification. The precision of this kind of detector is affected by two main systematic errors. The first one is the non-linearity of the voltage divider, due to local defects generating high leakage currents (hot spots) or to shorts between cathodes, which gives a non-linearity of the drift field and thus of the drift time. The second one is the fluctuation of the effective doping concentration in the silicon crystal introducing electrical parasitic fields which deviate the electrons from their straight trajectory towards the anodes. Beam tests[6] performed at CERN have demonstrated that the error on the reconstructed position of the impact point can reach values as large as 700/um. Therefore these errors on position must be corrected to achieve the 30/xm resolution required in the ALICE experiment. The beam test measurements have also shown that the non-linearity of the voltage divider and the inhomogeneities of the effective doping are stable in time. Hence, the systematic errors on position measurements, due to these defects, can be measured once and then corrected for during data analysis. Therefore each of the 260 SDDs have to be scanned in order to extract the "map" of deviations with a precision at the level of 10/zm. The "mapping" system must meet the following requirements : • the set-up must be simple : short development time, simple analysis and reliability • the scanning time has to be short (few hours) because 260 SDDs must be mapped • the systematics effects, among which the precision on the true impact position, have to be well below the detector resolution of 30/im. We have constructed from scratch a system based on a laser beam to precisely extract the deviation "map" of SDDs. 3. Laser mapping set-up The general principle of the mapping is to generate charges with a laser on precisely known locations on the detectors and to calculate the differences between the measured and the real positions. The chosen wavelength of the

1019

laser (A =1060nm) allows to create charges along the whole thickness of the detector, like particles do. The laser is moved over the detector surface by means of micrometric XYZ stages, having an accuracy of 5/zm and remotely controlled by a computer. A CCD camera is used to position the detector with a 2/im. The optical fiber carrying the laser beam is fixed close to the lens. The front end electronics[5] sends data to CARLOS/CARLOSrx boards[7] which generate the signals to control the read-out electronics and compress the data before sending them to the acquisition computer through optical fibers. The data acquisition software used is DATE[8]. Thus the read-out and the data acquisition system (DAQ) reproduces the ALICE one. The scan of the detector is performed all over its surface with a sampling of 100/xm in the anode direction and 120/im in the drift direction, the latter being imposed by the pitch of the metallized cathodes. When the motors reach the positions of measurement, the controller generates signals which trigger both the laser and the acquisition boards. With an event rate of about 100Hz, a full scan of the detector takes 2h for about 440,000 events.

4. Set-up calibration and mapping The set-up described above has been tested to check if a precision on measurement better than 30/xm can be reached. The time resolution of the system is better than Ins. This is mainly due to the fact that the 40MHz clock and the signals to trigger the laser and the DAQ are synchronized. With a typical electron speed of 8/um/ns, it implies a resolution better than 8fim. In addition, the error on the position given by the motor controller is 5/zm. Thus the intrinsic spatial resolution on the measurement is of 9.4/xm. The detector is positioned with the video system. To position the SDD on its support, several cathodes have some cross-shaped holes which are used as reference points for the scanning (cf. left photo in fig. 1). By scanning with the laser over a cross (cf. fig.l), one can extract the relative position of the laser spot with respect to the lens while a scan along the direction perpendicular to the detector plane allows to measure the incident angle of the laser beam. Therefore, the position of the laser spot on the detector is precisely known and the mapping can be performed with an accuracy of 15/im. Prom the linear fit of measured positions (anode and drift time centroids) versus the real positions, one can extract the anode pitch, as a cross-check, (294±0.05/xm) and the average electron speed (8^mi/ns at the nominal bias

1020

Figure 1. Left : photography of a positioning cross (hole in the metallization) ; Right : charge collected (in arbitrary unit) during the scanning over this cross, each pixel has a size of 5/im by 5/tm.

voltage). As said before, the electron speed varies during the drift because of the non-linearity of the voltage divider. The laser mapping system allows to precisely measure this effect which can induce errors of several hundreds of microns in our case. After the correction of this non-linearity along the drift axis, the remaining deviations correspond to inhomogeneities in the effective doping. The left plot of figure 2 shows an example of these residual errors that can reach 100/xm. The circular structures are characteristic of the production mode of the wafers. These results prove the ability of the laser mapping system to extract with precision the non-linearity of the voltage divider and the map of deviations due to effective doping fluctuations. The price to pay for this precision is the amount of data it implies. The deviation maps of the 260 SDDs would then require about 900Mb of memory which represent an excessive burden for the analysis program and this number must absolutely be reduced without loosing spatial resolution. To this aim, we have developed a method where we take advantage of the circular structure. First, one changes the system of coordinates from cartesian to cylindrical ; the new origin is of course the common center of the circles. Therefore the circles become straight lines in this frame. The second step is to perform the Fourier transform of this new map. Finally, a threshold on the Fourier coefficients is applied to select the most significant ones. During physics data analysis, the inverse Fourier transform must be performed to calculate the correction in position to apply. The right plot of the figure 2 shows the map of the inverse Fourier transform using only coefficients above threshold ; about 800 coefficients are used in this example. This method allows to reduce the amount of calibration data down to around 15Mb with a reduction factor of 60 and an error on the deviations of 10/im typically. Finally, when adding quadrati-

1021 cally the errors on the laser s p o t position a n d t h e error induced by the d a t a reduction, the total on the deviations in position is at the level of 15/im. W h e n compared t o t h e required 30/tm resolution, this error will n o t affect significantly the performance of the SDD. p o v i s l k m In Xin.u m J

Error on deviation In X in ,i m

|

Figure 2. Differences in pm between reconstructed and real impact positions in drift direction, after correction of the high voltage non-linearity. T h e white strips correspond to dead anodes or bad quality data. T h e figure on the left displays t h e measured differences, t h e figure on t h e right t h e differences after d a t a reduction and inverse Fourier transform.

5.

Conclusion

T h e m a p p i n g system for t h e calibration of the SDDs of A L I C E has been presented. I t allows t o quantify t h e deviations in position, induced by t h e non-linearity in t h e voltage divider a n d t h e inhomogeneity of resistivity, w i t h a precision a t the level of 15/zm. A p r o c e d u r e has also been introduced t o reduce drastically t h e a m o u n t of calibration d a t a by a factor around 60. In a near future, this system will b e used t o calibrate t h e final version of t h e S D D modules. References 1. ALICE web site : http://aliceinfo.cern.ch 2. Technical Design Report of ITS : http://aliceinfo.cern.ch/NewAlicePortal/en/Collaboration/ Documents/TDR/index.html 3. A. Rashevsky, et al, Nucl. Instr. and Meth. A 461 (2001) 133 4. D. Nouais, et al., Nucl. Instr. and Meth. A 501 (2003) 119 5. A. Rivetti et al., Nucl. Instr. and Meth. A 485 (2002) 188 6. E. Crescio et al., Nucl. Instr. and Meth. A 539 (2005) 250 7. http://www.bo.infn.it/ falchier/alice.html 8. http://ph-dep~aid.web.cern.ch/ph-dep-aid/

E X P E R I E N C E W I T H T H E RESISTIVE PLATE C H A M B E R IN T H E B A B A R E X P E R I M E N T

FABIO BELIINI University/INFN of Rome P.le Aldo Moro 2, 00185 Rome, Italy E-mail: [email protected] on behalf of the BABAR RPC collaboration

The BABAR detector has operated nearly 200 Resistive Plate Chambers (RPCs), constructed as part of an upgrade of the forward endcap muon detector, for the past two years. The RPCs experience widely different background and luminositydriven singles rates (0.01-10 Hz/cm 2 ) depending on position within the endcap. Some regions have integrated over 0.3 C/cm 2 . R P C efficiency measured with cosmic rays and beam is high and stable. However, a few of the highest rate RPCs have suffered efficiency losses of 5-15%. Although constructed with improved techniques many of the RPCs, which are operated in streamer mode, have shown increased dark currents and noise rates that are correlated with the direction of the gas flow and the integrated current.

1. Instrument Flux Return Overview The BaBar detector 1 , operating at the P E P I I B factory of the Stanford Linear Accelerator Center (SLAC), installed over 200 2nd generation Resistive Plate Chambers 2 (RPCs) as part of an upgrade 3 of the forward endcap muon and neutral hadron detector (IFR) in 2002. Most of the RPCs were operated nearly continuously for the two years of BaBar data taking covered in this report. The new RPCs were built from bakelite sheets at General Tecnica a . A stringent quality assurance (QA) program was developed by the IFR group to keep the inner bakelite surfaces as clean as possible and to ensure that the final linseed oil coating was thin and well cured. New molded corner pieces were designed to replace the drilling method previously used for the gas fittings. Multiple filters were added to the linseed oil filling stations and the oil was periodically analyzed for impurities. a

General Tecnica S. r. 1., 1-03030 Colli (FR), Italy

1022

1023 Chambers are made from two RPC high voltage modules interconnected to form a single gas volume and share one view of the readout strips. The modules are numbered from 1 to 6 counting from the bottom. The gas lines of two high voltage modules are connected in series to form a single gas volume of ~ 8 1. Details of the RPC construction and testing may be found in 3 . 1.1. RPCs

operating

conditions

BaBar RPCs operate in limited streamer mode, using a gas mixture of 4.5% isobutane, 60.6% argon and 34.9% Preon 134a. The gas flows in the inner (outer) layers were originally set to 22 (45) cc/min. corresponding to 4(8) volume changes per day. Concern for increased currents in the higher rate middle chambers prompted an increase in the gas flows in these RPCs to 8 volume changes per day (on day 300). During the Christmas 2003 break (day 420), all flows in the forward endcap were reversed. The streamer rates produced by backgrounds and signals from normal BaBar running varied considerably depending on the layer and position of the chambers. In the inner layers 1-12 the chamber occupancy was highest around the beam line. Signal rates (and occupancy) were proportional to the PEPII luminosity and were typically 30 to 50 kHz per module in layer 1 with peak rates above 10 Hz/cm 2 . RPC modules in the bottom of the endcap saw very low rates (little more than the cosmic ray rate) and never drew more than a few microamps. The rates in the outermost layers (15 and 16) were sensitive to backgrounds from PEPII, which enter the outside of the endcap. These backgrounds were often too high (> 12 Hz/cm 2 ) to allow normal operation. The rates in the next outermost layers (13-14) were lower (~ 4 Hz/cm 2 ) with typical PEPII backgrounds. Although the rates per module of RPCs in Layer 14 were typically higher than for RPCs in Layer 1, hits were spread over much of the chamber surface, resulting in lower peak rates per unit area. 1.2. Efficiency

Trends

Most of the endcap RPCs have stable efficiencies (Fig. l)with moderately increasing currents and noise rates. Other RPCs have significantly increased noise rates and currents coupled with significant efficiency losses4. At least part of the observed efficiency changes are probably due to increases of the bakelite resistance with time (due to the bakelite drying

1024

RPC Efficiency 1.0 "N

0.9

+-*i

A

A

A AA

A

A

*v ^

I 0-8 LU

0.7 0.6 6-2005 8715A29

200

400 Days

600

Figure 1. The average R P C efficiency of layers 1-12 was measured with beam using /ipair (black points) and with cosmic rays (open triangles). The crosses show the efficiency of the 4th R P C in layer 1 of the west door.

out) as has been previously observed by the ATLAS experiment during LHC chamber testing 5 . Initial measurements indicate that the relative humidity of the IFR input gas is ~ 0 % RH, while the exhaust gas from 2 modules in series is 20-30% RH, consistent with the removal of water from the RPCs. This observation suggests the following model. The initially dry gas entering the first module absorbs water from the bakelite as it flows through the module. The removal of water from the bakelite raises the bulk resistivity. When exposed to substantial signal rates, the higher resistance leads to a voltage drop across the bakelite, reducing the voltage across the gap thus lowering the RPC efficiency. The regions near the gas inlets are exposed to the driest gas at high flows. There is strong evidence in BaBar of high resistance regions developing in the bakelite near the gas inlets. These regions have a much reduced observed noise rate or reduced efficiency. When the gas flows were reversed and gas inlets became outlets, thereby increasing the humidity of the gas in that regions, the efficiency returned to normal. New inefficient regions appeared at the new gas inlet locations. It is likely that most of the water removed in the exhaust gas is from the first module. Preliminary measurements show that the humidity of the

1025 gas from the BaBar belt chambers (about 1/4 the size of the typical layer 1-16 chambers) is already 80% of that measured in the larger chambers. The gas entering the second (downstream) module is therefore more humid and drying effects in the second module are reduced. This may explain why the gas flow reversals also affected the efficency in the region of the high rate ring around the beam-line which is far from the gas inlet/outlets. The efficiencies of the chambers originally first in the gas flow chain increased after the reversal. Humidified (~ 30% RH) gas has been supplied from the beginning of 2005 to a sub-sample of the RPCs, a clear efficiency recovery effect showed up (Fig. 2) with no negative effect so far. Layer 14

Layer 15

Layer 16

Figure 2. The two-dimensional RPC efficiency of layer 11,15,16 before (upper plots) and after 5 months (lower plots) of ~ 30% RH gas flow.

1.3. Increased

Noise

rates and

currents

The correlation between increased dark currents, background rates, and position in the gas flow string 4 (Fig. 3) suggest that the downstream RPCs are being exposed to contaminants in the gas produced in the first RPC. Several RPC tests 6 have independently measured the presence of HF (hydrogen fluoride) in the exhaust gas presumably from the decomposition of the freon in the avalanche or streamer. High HF levels have been also correlated to increased ohmic currents 7 . Applying the same model to the BaBar

1026 1 1 Module 5

1

1

1

1

-

20 HA

10

M m

— . ','

i

J^dMifi ~

WTFfl

6-2005 A715A27

200

400 Days

600

200

400 Days

Figure 3. The average current drawn by layers 1-12 in the west door as a function of time. The gas flow was increased on day 280 and reversed on day 425 (arrow). The current with beam (black points) has contributions from streamers and an ohmic contribution (dashed line) which is estimated from the current when the voltage was below the gas gain region (4500V)

chambers would explain the preferential rise of "hot spots" in the high rate regions and in the downstream regions. Extensive HF measurements are on going. 1.4. Integrated

Charge

The 2002 RPCs in the forward endcap have integrated very different amounts of charge depending on position and background rates. By integrating the current history for each high voltage module, an integrated charge (varying from 10 to 1600 Coulombs) for each module was computed. The modules which have integrated more than 1000 Coulombs are either in the outermost active layer (14) or in the middle chambers of the inner layers. Taking into account the different occupancy, the integrated charge per unit area can be estimated to be « 0.075 C/cm 2 in Layer 14 and ~ 0.3 C/cm 2 in the inner layers 4 . The latter ones start to develop a loss of efficiency at small radii and thereby they will be operated in avalanche mode in the next run. References 1. 2. 3. 4. 5. 6. 7.

B . A u b e r t et al., Nucl. Instr. Meth. A 4 7 9 1-116 (2002). R. Santonico a n d R. Cardarelli, Nucl. Instr. Meth. A 1 8 7 377 (1981). F.Anulli et al., Nucl. Instr. Meth. A 5 3 9 , 155 (2005). F.Anulli et al, Nucl. Instr. Meth. A 5 5 2 , 276 (2005). G. Aielli et al, Nucl. Instr. Meth. A 5 3 3 86-92 (2004). R. Santonico, Nucl. Instr. Meth. A 5 3 3 1-6 (2004). G. Aielli et al., Proceedings of I E E E NSS R o m a ( 2 0 0 4 ) , t o b e published.

M U O N IDENTIFICATION A N D R E C O N S T R U C T I O N IN T H E ATLAS D E T E C T O R AT T H E LHC

N. CHR. B E N E K O S Max-Planck-Institut fir Physik, Fohringer Ring 6 80805 Muenchen, Germany E-mail: [email protected] On behalf of the ATLAS

Muon/Muid

Reconstruction

group

The muon detection system of the ATLAS detector is characterized by two high precision tracking systems, namely the Inner Detector and the Muon Spectrometer plus a thick calorimeter that ensures a safe hadron absorption filtering with high purity muons with energy above 3 GeV. In order to combine the muon tracks reconstructed in the Inner Detector and the Muon Spectrometer the Muon Identification (Muid) Object-Oriented software package has been developed. In this note the Muid reconstruction procedure is briefly described, followed by a more detailed presentation of its performance. The impact of the missing components of the ATLAS Detector on the performance of the measurements is investigated.

1. Introduction The ATLAS detector 1 , currently being installed at CERN, is designed to make precise measurements of 14 TeV proton-proton collisions at the LHC, starting in 2007. ATLAS consists of four main subdetectors: • the Inner detector(ID) for the measurement of momentum and impact parameter of charged particles. • the Calorimeter system, for measurement of particle and jet energies • the Muon Spectrometer(MS) for muon identification and momentum measurement, consisting of high precision drift tubes for tracking(MDT,CSC), and a set of two subsystems of trigger chambers: resistive plate chambers (RPC) and thin gap chambers (TCG) and • A magnet system for bending of charged particles for momentum measurements 1027

1028 The identification of muons is performed using a combination of high precision tracking detector components, including an ID, housed in a uniform solenoidal field, and a precision MS housed in toroidal fields. The Spectrometer measures curved tracks in 3 stations in the barrel toroid and straight-lines tracks before and after the endcap toroid. There is a sufficient calorimeter depth between them to ensure the absorptions of hadrons before the Spectrometer, yielding high purity muons with momenta above 3GeV.

Figure 1. Muonidentification ATHENA algorithms (left) and data objects exchanged with the Transient Event Store(right)

2. ATLAS Detector Layout The Inner Detector 2 has been in a state of continuous evolution up to the present time. Engineering developments and cost limitations have necessitated changes to the layout geometry and detector materials. Changes since the DC1 layout 3 with which the present results are compared include and presented 4 . The Muon Spectrometer 5 will not be fully deployed at the beginning of the collision period 6 . In particular, the installation of the EE(Endcap) wheel and half of the CSC stations will be postponed. This will lead to a less precise measurement of the first segment on the muon trajectory and thus to a deterioration of the muon momentum resolution at \r)\ < 2. Moreover, in the initial layout, the rapidity region between 1 and 1.3 will be

1029 characterized by both worsened momentum resolution and reconstruction efficiency as is indeed seen in the performance of the full reconstruction of a first simulation of this layout. 3. Muon Reconstruction Software MOORE 7 identifies track segments using local pattern recognition at the detector module level in each of the stations of the MS and performs a track fit, based on the package developed for the ID, iPatRec 8 . The end result of MOORE is a collection of data objects, which describes the reconstructed tracks at the entrance of the MS. MOORE has been developed within the ATLAS ATHENA 9 software environment. 4. Combined muon reconstruction in ATLAS In order to combine the muon tracks reconstructed in the MS and the ID, the Muonidentification (Muid) Object-Oriented(OO) software package has been developed. The combined reconstruction improves the track measurements, giving the best possible momentum resolution and reducing the tails in the momentum resolution distribution of the MS, accounting for the fluctuations in the energy loss in the calorimeter. Moreover it improves: the charge determination for high energy muons by means of the longer lever arm helps to discriminate muons from secondaries in the MS; to reject decay muons(from Kaons and pions) by requiring tracks to originate from the primary vertex and to resolve track finding ambiguities which occur in the MS from local showering and cavern background. 4.1.

Muonidentification-Muid

The reconstructed objects produced by MOORE are tracks whose parameters are expressed at the first measured point inside the MS. MuidStandAlone propagates the muon track to the vertex. MuidStandAlone re-express the track parameters with covariance at the closest approach to the nominal vertex (the centre of the beam intersection region). Calorimeter Coulomb scattering is taken into account by a parametrization of the width of the simulated broadening of position and angular distributions. A correction is also applied to account for Energy loss. This may be from a parametrization or a direct measurement from the CalloCells according to the isolation criteria. The procedure is a full track refit to the

1030 PT = 20GeV/c

I

p r = 100 GeV/c

-•Iipiflf-

1

fc-

yf liner Dai . £ UildSlsndalna ^ UildCoitinsd

1.5

2.5

I

••I**™"Ml|g^ i V lwwrD« ^

A-

lAjJdCbmbned

0.5

1.5

2.5

1 Figure 2. Global Efficiency as a function of \r)\ for a p y = 2 0 GeV

Figure 3. Global Efficiency as a function of |r;| for a P T = 1 0 0

3T = 20 GeV/c

f:

: 100 GeV/c OTSS

0"°™ £ ^

"-,0.05 j O. r 0,

GeV

•js 0.15

kbldSlandalens Mi1dCoirfcln«d

****> 0.5

1.5 71

Figure 4. resolution on p y as a function of |?j| for a P T = 2 0 GeV

^0.05 Q. ^ 0.

Y tineruel | UildSBnddoia A WildOomBiMd

0.5

1.5 11

2.5

Figure 5. resolution on p y as a function of |rj| for a p r = 1 0 0 GeV

MS measurement to allow for reprocessing with updated calibrations at ESD(DST) level plus extra "measurements" representing the Calorimeters. The combination between the ID and the MS track is performed by the algorithm MuidComb. Tracks are matched by forming a \ 2 with 5 degrees of freedom from the parameter differences and summed covariances. A schematic sketch of the algorithms and the objects exchanged with the Transient Event Store is shown in Figure 1. 5. Muon Reconstruction Expected Performance The specification is to reach 10 % APT/PT in the barrel region (|r/| < 1) and the endcap toroid region (1.5 < \T]\ < 2.7). At high momentum the resolution is dominated by the intrinsic precision of the muon chambers and alignment errors; at moderate momentum, the resolution is limited by the multiple scattering to APT/PT ~ 2%. Finally, at low momenta, the fluctuations in energy loss in the Calorimeters and in the MS are such that the purpose is to identify which ID tracks are muons. They are sufficiently well measured by the ID. These specifications are sufficient for the possible discovery and study of exotic physics such as

1031

new gauge bosons. The reconstruction performance has been tested with single muon samples in a range of transverse momentum from 3 GeV/c to 1 TeV/c, uniformly spread over a range of |r]| up to 2.7. The full reconstruction chain has been executed, namely: the reconstruction in the MS alone (MOORE), the extrapolation to the vertex of the track found in the MS (MuidStandalone), the reconstruction in the ID (iPatRec) and the combination of the track found in the MS and in the ID (MuidComb).

6. Conclusions The ATLAS detector has been designed to have a good muon transverse momentum(pT) resolution of few % which should be independent of px and |77| for a wide pr range. A robust muon identification and high precision measurement is crucial to full exploit the physics potential of the LHC. The muon energy of physics interest ranges in a large interval from few GeV, where the B-physics studies dominate the physics program, up to the highest values that could indicate the presence of new physics. The performance of the Reconstruction program on simulated data is in agreement with the ATLAS design specifications. References 1. ATLAS Detector and Physics Performance Technical Design Report, Volume I, II (25 May 1999), ATLAS TDR, 14,15 CERN/LHCC 99-14,15. 2. Atlas Collaboration, Inner Detector, Technical Design Report, Vol.I, CERN/LHCC/97-16(1997). 3. N. Benekos et al., Inner Detector Note, ATL-INDET-2004-002. 4. N. Benekos et al., ATL-COM-INDET-2005-008. 5. ATLAS Muon Spectrometer Technical Design Report, CERN/LHCC/'97-22, 1997. 6. Atlas completion plan, September 2002 CERN-RRB-2002-114 . 7. D. Adams et al., ATL-SOFT-2003-007. 8. R. Clifft, A. Poppleton, ATL-SOFT-94-009. 9. The ATLAS Common Framework developers guide, CERN, February 2004. http://atlas.web.cern.ch/Atlas/GROUPS/SOFTWARE/00/architecture/ General/Documentation/AthenaDeveloperGuide-8.0.0-dr aft.pdf

MICROMETRIC POSITION MONITORING USING FIBER BRAGG GRATING SENSORS IN SILICON DETECTORS E. BASILE (*), F. BELLUCCI (***), L. BENUSSI, M. BERTANI, S. BIANCO, M.A. CAPONERO (**), D. COLONNA (*), F. DI FALCO (*), F.L. FABBRI, F. FELLI(*), M. GIARDONI, A. LA MONACA, F.MASSA (*), G. MENSITIERI (***), B. ORTENZI, M. PALLOTTA, A. PAOLOZZI (*), L. PASSAMONTI, D.PIERLUIGI, C. PUCCI (*), A. RUSSO, G. SAVIANO (*)f Laboratori Nazionali diFrascati dell'INFN, v.E.Fermi 40 00044Frascati (Rome) Italy October 17,2005 We show R&D results including long term stability, resolution, radiation hardness and characterization of Fiber Bragg Grating sensors used to monitor structure deformation, repositioning, and surveying of silicon detectors in High Energy Physics.

1. Introduction FBG sensors are widely used in telecommunication as optical filters. For the first time we have used FBG sensors as optical, low-noise, high-resolution strain gauges to monitor structure deformation, repositioning, and surveying silicon (pixel and microstrips) detectors for HEP experiments at hadron machines. We show R&D results including long term stability, precision, resolution, radiation hardness and characterization. 2. FBG Sensors Fiber Bragg Grating (FBG) sensors have been used so far as telecommunication filters, and as optical strain gauges in civil and aerospace engineering [1], and, only recently, in HEP detectors [2]. The BTeV[3] detectors utilize FBG sensors to monitor online the positions of the straw tubes, pixels, and microstrips. The optical fiber is used for monitoring displacements * Permanent address: "La Sapienza" University - Rome. ** Permanent address: ENEA Frascati. *** Permanent address: "Federico II" University - Naples. ****Permanent address: INFN Sez. Roma 1. This work was supported by the Italian Istituto Nazionale di Fisica Nucleare and Ministero dell'Istruzione, dell'Universita' e della Ricerca. This work was partially funded by contract EU RII3-CT-2004-506078.

1032

1033 and strains in mechanical structures such as the straw tube-microstrip support presented here. A modulated refractive index along the FBG sensor produces Bragg reflection at a wavelength dependent on the strain in the fiber (Fig.l), permitting real-time monitoring of the support. According to these properties, an FBG sensor is going to be placed in the MOX structure between the Rohacell® foam and the CFRP shell. Sensors will be located in spots of maximal deformation, as predicted by FEA simulation. Figure 2 shows long-term behaviour of FBG sensors while monitoring micron-size displacements, compared to monitoring via microphotographic methods. 3. Long-term Stability and Radiation Hardness The optical fiber is used for monitoring displacements and strains in mechanical structures such as the presented straw tubes-microstrip support. A wavelength selective light diffraction grating (Fig.l) along the FBG sensor is placed in the fiber, and it permits an on-time monitoring of the support. Fig.2 shows long-term behaviour of FBG sensors while monitoring micron-size displacements, compared to monitoring via photographic methods. Sensors have been tested for radiation damage. Fig.3 shows spectral response up to a neutron fluence of l.6»lo" 14-MeV neutrons/cm2, corresponding to 6 months BTeV integrated dose.

4. The Omega-like Repositioning Device FBG sensors have been also applied to instrument a novel repositioning device with micrometric resolution. The Omega-like device (shown as prototype in Fig.4,5) follows the displacement of the pixel detector designed for the BTeV experiment at the Fermilab Tevatron which, at each accelerator store, has to be moved out and in of the beamline. Fig.6 shows a Finite Element Analysis of the Omega-like device. FBG sensors are located on area of largest strain in order to maximize sensitivity. Preliminary results show how a repositioning precision of about lOmicrons is reached. Work is in progress to reach the required 3 micron precision.

1034 Input signal (before diffraction)

Signal after diffraction

Optical fiber

FBG

FBG output signal

Tipical Values As = 1 u s

AA. = 1 pm

+1 A T = 1 K

A?,= i o p m

Figure 1. Principle of FBG sensors operation. A laser pulse is injected in the fiber and reflected selectively accoding to the grating pitch. Strain As changes the grating pitch thus changing the wavelength of reflected pulse. The sensor is also sensitive to temperature changes.

2

IT

r

0

20

40

ISO

80 Time [hour]

100

120

140

160

1

TV camera XXX FBG

Figure 2. FBG long-term monitoring stability results. FBG output (crosses) is validated by TV camera (bars).

1035 0.040

0.035

-

0.000

Bragg Figure 3. Radiation hardness of FBG sensors. Spectral response up to a neutron fluence of 1.6»10l: 14-MeV neutrons/cm2, corresponding to 6 months BTeV integrated dose.

Figure 4. Sketch of BTeV pixel detector and its Carbon Fiber Reinforced Plastic support frame.

1036

FBG sensors

..otfiftSWMSfc,,

1

•'." •".'• -"." •'.• •".• •"." •'.•

• - • • • - • • • - - •

Figure 5. The Omega-like repositioning devices, equipped with FBG sensors, follows the pixel support structure in and out of beam, assuring repositioning accuracy.

Figure 6. Finite Element Analysis of the Omega-like repositioning system. Sensors arc located on the areas of larger mechanical strain in order to maximize sensitivity.

1037

5. Conclusions We have used FBG sensors in HEP for the first time as precise, stable, optical devices for micrometric position monitoring of silicon pixel and strips detectors. FBG sensors provide position monitoring with micrometric resolution. Under radiation with doses typical of year-long operation at hadron colliders they show no sign of spectral response shift. We have used sensors to characterize and optimize pixel support structures in Carbon Fiber Reinforced Plastic. Finally, we have proposed a novel device to precisely reposition the pixel detector in and out of the beams at each accelerator store. Preliminary results show a lOmicron resolution, improvements are undergoing and we think we can reach the 3micron precision required by the experimental operation.

References 1. S. Berardis et al., "Fiber optic sensors for space missions" 2003 IEEE Aerospace Conference Proceeding, Big Shy Montana, March 8-15, 2003, pp. 1661-1668. 2. L. Benussi et al., "Results of Long-Term Position Monitoring by Means of Fiber Bragg Grating Sensors for the BTeV Detector", Frascati preprint LNF -03/15(IR). 3. Fermilab Experiment E-0897/E-0918, J. Butler, S. Stone co-spokespersons; see www-btev.fnal.gov.

COMMISSIONING OF T H E CMS TRACKER OUTER BARREL

CHRISTOPH BLOCH CERN Route de Meyrin, 1211 Geneve 23, Switzerland E-mail: [email protected] On behalf of the CMS Tracker Collaboration

Fully equipped final substructures of the CMS Tracker are installed in a dedicated mechanical support, the Cosmic Rack, providing a geometry suitable for tracking cosmic muons, and equipped with a dedicated trigger that allows the selection of tracks synchronous with the fast readout electronics. Data collected at room temperature and at the tracker operating temperature of -10°C can be used to test reconstruction and alignment algorithms for the tracker, as well as to perform a detailed qualification of the geometry and the functionality of the structures at different temperatures. The CMS Monte Carlo simulation has been adapted to the geometry of the cosmic rack, and the comparison with the data will provide a valuable test to improve the tracker simulation in CMS.

1. Introduction The CMS Tracker is composed by an inner part populated with silicon pixel detectors and an outer part populated with silicon strip detectors. The silicon strip system, with an active area of 210m 2 , is by far the largest tracker ever built using such technology. In the central volume of the Silicon Strip Tracker1 rectangular detectors are arranged on cylindrical shells, with the readout strips parallel to the beam axis; this region is further split into Tracker Inner Barrel, made of four layers, and Tracker Outer Barrel (TOB), made of six layers. In the forward regions wedge-shaped detectors with radial readout strips are mounted on supporting disks. In the Tracker TOB, the silicon detectors are mounted on 688 subassemblies, called rods, which also house all the interconnection, control and readout electronics. Rods of layers 1 and 2 have two detectors mounted back-to-back in each position: one of the two detectors has the readout strips tilted by 100 mrad with respect to the z axis, and the com1038

1039 bination of the coordinates reconstructed by the two detectors of the pair provides also a measurement of the z coordinate. The data from the silicon detectors of the Tracker are converted to analogue optical signals at the front-end2, and transmitted via optical fibers to the back-end, where they are converted back to electrical signals, digitized and analyzed 3 . The control signals that drive the readout electronics are also converted to optical signals and travel on optical fibers from the back-end to the front-end and back 4 .

2. The Cosmic Rack In order to operate successfully a complex device as the CMS Silicon Strip Tracker, and to fully exploit its potential, the properties of the hardware need to be characterized as deeply and precisely as possible, and the reconstruction software needs to be commissioned with physics signals. A number of issues have been identified, which need to be carefully studied to commission the detector, some of which concern the entire Tracker, while some are specific to the TOB: - the time evolution of the signals in the readout electronics need to be precisely measured and correctly simulated, as it affects the expected occupancy and the data volume, critical issues in high-luminosity running; - the electronics coupling between neighbouring channels affects the cluster size and hence the hit resolution, the occupancy and the data volume; - the speed of the front-end chips changes with temperature, therefore a possible variation of the above effects between room temperature and the Tracker operating temperature of — 10°C needs to be investigated; - the mechanical structure of the rods is composed by carbon fiber elements; aluminium inserts glued to the carbon fiber members provide efficient cooling contacts between the silicon detectors and the thin cooling pipe, made of a copper-nickel alloy; the different thermal expansion coefficients of the various components induce stress on the structure when this is cooled down to the operating temperature, possibly causing small deformations; a detailed characterization of the geometrical precision of the rods and of its possible evolution with temperature is a valuable input for track reconstruction in CMS. A dedicated setup has been designed and realized, to investigate the above issues. A maximum of 20 rods can be housed in 10 layers. The distance between the layers, the overlap and the relative angle between the two rods in each layer reproduce the average corresponding parameters of

1040 PMT

Figure 1.

Schematics of the CRack

the TOB. The distribution of the control signals and of the cooling fluid to the rods also mimics the situation of the real detector. The mechanics is designed so that it can be turned by 90° to operate the device on a test beam, or it can be complemented with two scintillator modules, above and below the module housing the rods, to track cosmic muons: hence its name of "cosmic rack". The compact design of the scintillator modules, with the light guides bent by 180°, as shown in Fig. 1, allows the device to fit into a climatic chamber, where it can be operated in nominal environmental conditions. The cosmic rack has been initially operated in two beam tests, populated with 6 rods in 6 different layers, that were important steps for the comissioning of the reconstruction and tracking software. Subsequently it has been upgraded to 12 rods, and used to commission the powering and readout systems to be used for the TOB integration, which are made of final CMS components, and to validate the grounding scheme of the TOB prior to the start of the integration. Finally it has been complemented with a dedicated trigger logic, suitable to track cosmic muons with the fast readout electronics of the Tracker, so that the program of studies outlined above can now continue using cosmic muons. 3. The Trigger System The readout electronics of the CMS Tracker works with an internal 40 MHz clock, that will run synchronous with the arrival of the particles emerging from the LHC collisions. The amplitude of the output signals drops rapidly to zero to avoid collecting hits from nearby bunch crossings. In order to be able to study the signal over noise ratio of the detectors, and to characterize the time evolution of the physics signals and of

1041

the electronics coupling of neighbouring channels, the trigger system must be able to select only tracks that are synchronous, within 2-3 ns, with the rising edge of the internal clock, to simulate correctly the LHC operation environment. In addition, the difference in time between the signals collected in the various layers has also a time jitter, caused by the spread in the incident angles of the muons, resulting in different path lengths. The trigger system must therefore select tracks with a limited spread around a preferential incident angle. The obvious choice is to select vertical tracks, for which the rate is highest. Finally, the trigger must deliver a decision within 4/is from the actual charge deposit, to cope with the length of the readout pipeline that holds 192 samples. A trigger system compliant with all the above requirements has been realised using standard NIM components. It is based on two large plastic scintillators mounted on top and on bottom of the cosmic rack, covering the active area of the rods. Each scintillator is equipped with Photo Multiplier Tubes (PMTs) on both sides (Fig.l); a lead plate is placed just above the bottom scintillator to absorb low energy particles. First the top and bottom PMTs on each side of the cosmic rack are connected to a coincidence unit, after delaying the top scintillator by 3 ns, corresponding to the time of flight of straight relativistic particles. This first coincidence reject tracks impinging on the cosmic rack with large incindence angles. Second, the signals from the two sides are sent to a further coincidence unit, that has a time window large enough to accomodate for the propagation of the signals over the entire length of the two scintillators. The purpose of this second coincidence is to reduce background from noise in the scintillators and in the PMTs. This part of the logic provides good trigger signals for vertical particles, but its timing suffers from the jitter due to the large size of the scintillators, and it is therefore not suitable for gating with the internal electronics trigger. For this purpose, the signals from the two upper scintillators are also sent to a meantimer unit, which provides a signal that corresponds to the mean arrival time on the two sides, delayed by a fix amount. This signal carries the precise information on the arrival time of the particle on the surface of the top scintillator. The signal from the meantimer is sent to a further coincidence unit together with the basic trigger signal, suitably delayed, so obtaining a signal that has a well defined delay with respect to the arrival time of the particle on the top scintillator. This final trigger signal is then sent to a gate that only accepts triggers within a window of 5 nsec of the electronics clock. The phase of the clock in the different layers is then adjusted to account for the travel

1042

time of vertical relativistic muons from the top scintillator to each layer. With the logic described the trigger rate is approximately 0.8 Hz, translating to about 1 hit per module every 25 sec. For tracking studies, where the characteristics of the signals are not critical, the final gate can be removed to gain a factor of 5 in the trigger rate.

4. Study of the grounding scheme In the TOB a number of rods that varies between 8 and 22 are served by the same cooling manifold. Cooling pipes are made of a Cu-Ni alloy, and all joints are soldered, therefore the rods belonging to the same cooling segment have the mechanical structures electrically connected together. For each rod, the power return line is connected to the manifold through a dedicated multiwire cable. So the manifold serves as local ground for all the rods of a cooling segment. Different manifolds are then connected together and to the main support structure of the Outer Barrel through a system of metallic strips and rings. In designing the cosmic rack, care has been taken to mimic the grounding scheme of a TOB cooling segment, in order to be able to test it and validate it before the actual start of the TOB assembly. However, contrary to the TOB, in the cosmic rack the cooling manifolds are made of plastic, with quick connections to the rod pipes. To recover the possibility of implementing the same grounding scheme as in the TOB, a copper bar has been added close to the rod ends, to which the multiwire cables are connected. The noise has been analyzed on a data sample collected with 12 rods arranged in 6 layers, after implementing the final grounding scheme described above. The rods were powered with final production power supplies and cables, and the readout system also used final hardware and software. The noise of one electronic channel can be considered as the combination of an intrinsic noise and a common mode noise, correlated among the channels. The common mode noise is affected by the quality of the grounding. Assuming full correlation among the strips of a readout chip, one can define the common mode level in one event as the median of the readings of all the strips in that event, and calculate the common mode noise as the RMS of the common mode level. The common mode noise can then be subtracted from the total noise, to obtain the common mode subtracted noise. The total noise for one modules is compared to the common mode subtracted noise in Fig. 2. The difference is negligible, which demonstrates the good quality of the grounding.

1043

!

100

200

300

400 500 Strp number

Figure 2. Noise (black) and common mode subtracted noise (grey) per strip. 5. S u m m a r y a n d o u t l o o k U p to 12 rods were successfully operated in the cosmic rack, using final CMS powering and readout hardware and software. T h e system has been complemented with a trigger system compliant with all requirements implied by the fast readout electronics, designed to run synchronous with t h e LHC clock. T h e analysis of the d a t a collected so far validated the grounding scheme of the T O B . Further investigations will focus on time evolution of signals, and electronics coupling between nearby channels. As soon as double-sided rods will become available, precise reconstruction of cosmic muon tracks will give the ability t o characterize t h e geometry of the rods at room t e m p e r a t u r e and at -10°C, and t o test and develop alignment algorithms for the CMS Tracker.

References 1. D. Abbaneo. Layout and performance of the cms silicon strip tracker. Nucl. Instrum. Meth., A518:331-335, 2004. 2. M. T. Brunetti et al. Design and performance of a circuit for the analog optical transmission in the cms inner tracker. Prepared for 7th Workshop on Electronics for LHC Experiments, Stockholm, Sweden, 10-14 Sep 2001. 3. J. A. Coughlan et al. The CMS Tracker Front-End Driver. Technical report, CERN, 2003. 4. F. Drouhin et al. Progress on the CMS Tracker control system. In Proceedings of the 11th Workshop on Electronics for LHC Experiments, Heidelberg, 2005.

N E W EFFECTS OBSERVED IN T H E BABAR SILICON VERTEX T R A C K E R : I N T E R P R E T A T I O N A N D ESTIMATE OF T H E I R IMPACT O N T H E F U T U R E P E R F O R M A N C E OF T H E D E T E C T O R *

V. RE 1 , D. KIRKBY2, M. BRUINSMA2, S. CURRY2 J. BERRYHILL3, S. BURKE3, D. CALLAHAN3, C. CAMPAGNARI3, B. DAHMES3, D. HALE3, P. HART3, S. KYRE3, S. LEVY3, O. LONG3, M, MAZUR3, J. RICHMAN3, J. STONER3, W. VERKERKE3, T. BECK4, A.M. EISNER4, J. KROSEBERG4, W.S. LOCKMAN4, G. NESOM4, A. SEIDEN4, P. SPRADLIN4, W. WALKOWIAK4, M. WILSON4 C. BOZZI5, G. CIBINETTO5, L. PIEMONTESE5, H. L. SNOEK6, D. BROWN7, E. CHARLES7, S. DARDIN7, F. GOOZEN7, L.T. KERTH7, A. GRITSAN7, G. LYNCH7, N. A. ROE7 C. CHEN8, C. K. LAE8, W. HULSBERGEN8, V. LILLARD8, D. ROBERTS8, A. LAZZARO9, F. PALOMBO9, L. RATTI10, P.F. MANFREDI10, E. MANDELLI10, C. ANGELINI11, G. BATIGNANI11, S. BETTARINI11, M. BONDIOLI11, F. BOSI11, F. BUCCI11, G. CALDERINI11, M. CARPINELLI11, M. CECCANTI11, F. FORTI11, M.A. GIORGI11, A. LUSIANI11, P. MAMMINI11, G. MARCHIORI11, M. MORGANTI11, F. MORSANI11, N. NERI11, E. PAOLONI11, A. PROFETI 11 , M. RAMA11, G. RIZZO11, G. SIMI11, J. WALSH11, P. ELMER12, A. PERAZZO13, P. BURCHAT14, A.J. EDWARDS14, S. MAJEWSKI14, B.A. PETERSEN14, C. ROAT14, M. BONA15, F. BIANCHI15, D. GAMBA15, P. TRAPANI15, M. BOMBEN 16 + L. BOSISIO16, C. CARTARO16, F. COSSUTTI16, G. DELLA RICCA16, S. DITTONGO16, S. GRANCAGNOLO16, L. LANCERI16, L. VITALE16, M. DATTA17, A. MIHALYI17 1

INPN-PAVIA AND UNIVERSITA DI B E R G A M O 2 U N I V E R S I T Y OP CALIFORNIA, IRVINE U N I V E R S I T Y O F CALIFORNIA, SANTA BARBARA 4 U N I V E R S I T Y O F CALIFORNIA, SANTA CRUZ 5 I N F N - F E R R A R A AND UNIVERSITA DI F E R R A R A 6 N I K H E F , NATIONAL I N S T I T U T E F O R N U C L E A R PHYSICS AND HIGH E N E R G Y PHYSICS, AMSTERDAM, T H E NEDERLANDS 7 L A W R E N C E B E R K E L E Y NATIONAL LABORATORY 8 U N I V E R S I T Y O F MARYLAND 9 I N F N - M I L A N O AND UNIVERSITA DI MILANO 10 I N F N - P A V I A AND UNIVERSITA DI PAVIA " i N F N - P I S A AND UNIVERSITA DI PISA 12 PRINCETON UNIVERSITY ' ' S T A N F O R D LINEAR A C C E L E R A T O R C E N T E R , S T A N F O R D 14 STANFORD U N I V E R S I T Y 15 I N F N - T O R I N O AND UNIVERSITA DI T O R I N O 16 I N F N - T R I E S T E AND UNIVERSITA DI T R I E S T E 17 U N I V E R S I T Y O F WISCONSIN, MADISON 3

1044

1045 The Silicon Vertex Tracker (SVT) of the BABAR experiment at PEP-II consists of five-layers of double sided AC-coupled Silicon strip detectors, read out by a fullcustom IC, capable of simultaneous acquisition, digitization and transmission of data. It represents the core of the BABAR tracking system and it is the crucial device for measuring B meson decay vertex position, to extract CP-asymmetries. After six years of operation, some unexpected affects have appeared. In particular, a shift in the pedestal for the channels of the AToM readout chips more exposed to radiation, and an anomalous increase of the bias leakage current for the modules of the outer layers, have been observed. In both cases the reason has been understood and reproduced. Results will be presented together with extrapolated performances.

1. The BABAR Silicon Vertex Tracker The BABAR experiment 1 is dedicated to a systematic study of CP asymmetries in B decays and to precision measurements of the CKM quark mixing matrix. It has thus far recorded over 200 million BB pairs produced in collisions of a 3.1 GeV positron beam and a 9.0 GeV electron beam. The beams provide a center-of-mass energy of 10.58 GeV, near the T(45) resonance, with a net boost that gives a measurable separation of the two B decays. The BABAR SVT is the sub-detector that is nearest to the e+e~ interaction point. Its primary purpose is a precise determination of track parameters of charged particles to allow the reconstruction of vertices with a resolution that is sufficient to disentangle the two B decays. They have an average separation along the beam axis (z) of 280 /xm and the achieved vertex resolution in z is 80 /urn. The SVT also enables the reconstruction of low-momentum (pr < 120 MeV) charged particles, in particular slow pions from D* decays, that do not fully traverse the main BABAR tracker. The SVT, described in greater detail elsewhere 2 , is a silicon micro-strip detector comprising over 140000 channels on 340 double-sided, AC-coupled silicon wafers arranged in five layers. The first layer is as close as 3.2 cm to the beam axis, while the last layer is 14.0 cm away. Each wafer has strips oriented perpendicular ('2-strips') and parallel ('(^-strips') to the beam axis on either side, and the readout pitch ranges from 50 ^m for the ^-strips in the first layer to 210 /xm for the z-strips in the outermost layers. The entire detector is immersed in a 1.5 T magnetic field. 2. Performance Of the 208 readout sections of the SVT, nine were inoperable from the start of data taking in 1999. Five of these failures were recovered during 'presented by M. Bomben at 9th ICATPP Conference. tcorresponding author, [email protected]

1046

an access to the detector in 2002. In addition, approximately 2% of the individual channels were unbonded or otherwise unresponsive. The 5% missing channels do not significantly affect the overal performance of the SVT, thanks to a sufficient redundancy in the detector design. The hit efficiency averages at 97%. The efficiency for reconstructing low momentum particles is 75% for momenta above 100 MeV and 90% for momentum above 200 MeV. The hit resolution is measured using high-momentum particles in two-prong events. For particles incident on the wafer under a straight angle, the hit resolution is between 10 and 15 fjm for the three innermost layers, and between 30 and 40 fan for the two outermost layers. This gives a vertex resolution in for fully reconstructed decays of better than 80 /zm in z. The resolution in the separation of the vertices of a fully reconstructed decay and an inclusively reconstructed decay, is 190 /xm, dominated by the uncertainty in the inclusively reconstructed (tag side) vertex. No loss of efficiency or decrease in resolution has been observed in the first five years of operation. 3 . U n e x p e c t e d behaviours After five years of operations, some unexpected behaviours have been observed. A shift in the pedestal of the AToM readout chips for mid-plane modules has appeared. There has been then an anomalous increase of the bias leakage current for some modules of the outer layers. We will describe briefly both of the phenomena; we will present also their interpretation and the remedy that is in place. 3.1. Pedestal

shift in AToM

chips

A shift in the value of the pedestal for the channels of the AToM chips has been observed, as it can be see in figure 1.

&Sfcfc*%i

Figure 1. Average noise per channel with zero charge injected. increase in noise for chip 3 and 4.

It can be seen the

The values plotted in figure 1 correspond to a zero charge injection performed during a so-called 'Noise calibration'.

1047 After a very detailed study of the structure of the AToM chip, we found that the effect is radiation-induced and it is due to a well-identified component of the chip. The effect of this pedestal shift was a loss of efficiency for those channels. The net effect for the affected inner layer module was around 10%. We have raised the threshold for the AToM chip to cope with the problem. The efficiency has gone back to the original value. 3.2. Increase

in bias leakage

current

An anomalous increase in the bias leakage current has been observed since spring 2004. This increase was affecting only some modules of the layer 4. In figure 2 we show an example of the increase of the bias leakage current as a function of time. It was clearly not a simple background related effect due

Figure 2.

Bias leakage current as a function of time for one half module.

to the absence of the effect in inner layers. Moreover, no one layer 5 module was affected, even if they were only a few millimeters far away from layer 4 ones. A series of analysis and tests were performed to try to understand the origin of the problem and some conclusions have been driven. First of all, the presence of the beams have an effect: a couple of hours without beams in the machine has determined a stop in the bias leakage current increase. Then the increase of a few percent of the relative humidity of the air in the detector have helped in order to halt the increase. Moreover, there was a non zero voltage drop across the space between layer 4 and 5. Varying this voltage drop we were able to stop the effect. All this hints pointed to the existence of charges, induced by the beam presence, on the passivated surface of the sensors. The charge drifts toward the sensors thanks to the electrical field present in the space region between different layers. When deposited on the surface of the sensors, the charge accumulation modifies the internal field in the silicon, inducing a beginning of breakdown effect. A simulation of the effect has been performed and the results are in figure 3. The modification of the reference voltage for layer 4 and 5 modules and the increase of relative humidity of the air in the detector volume (from

1048

Figure 3. Simulation to map the electric field inside the silicon in the junction region in presence of additional positive charge on the surface passivation. The intense electric field at the tip of the p + implant can induce a breakdown, which is responsible for the increase in the measured leakage current.

roughly 2% to about 5%) has helped in stopping the effect, limiting the drift of the charges toward the sensors and the helping the discharge of the charges on the passivated surfaces. We were eventually able to decrease the bias leakage current to the former values, 4. Estimate of impact on data quality of radiation damages The current plans for the PEP-II accelerator are shown in table 1. Table 1. one.

Current plans for the PEP-II accelerator. HER is the electron ring, LER the positron

September 2005 July 2005 July 2005 July 2005

HER current (A) 1.65 1.75 2.00 2.20

LER current (A) 2.65 3.10 3.30 3.50

Peak luminosity ( 1 0 3 3 s - 1 c m - 2 ) 9.0 12.0 17.0 20.0

With this plan for the currents an extrapolation for dose is computed. The Signal to Noise ratio has to be greater than 10. This translates into a radiation budget of about 5 MRad. Relying on the computed extrapolation, SVT will be operated smoothly at least till 2007. Conclusions The BABAR Silicon Vertex Tracker has been operated since 1999, matching all the design requirements for resolution and efficiency. After five years some unexpected effects have appeared. In particular a shift in the pedestal of the readout chips and an anomalous increase in the bias leakage current. Both the effects were fully understood and cured. References 1. B. Aubert et al. , NIM A479, (2002) 1-116. 2. V. Re et al., IEEE Trans. Nucl. Sci. 49 (2002) 3284.

TESTS OF S U B S T R U C T U R E S OF T H E CMS SILICON STRIP T R A C K E R

R. BRAUER 1. Physikalisches Institut B, RWTH Aachen Somrnerfeldstr. 14, 52074 Aachen, Germany E-mail: [email protected] On behalf of t h e C M S tracker collaboration

The silicion strip tracker of the CMS experiment is at present being integrated and will be completed by the end of 2007. With its more than 15000 single silicon modules covering an active detector area of more than 200 square meters, it will be the largest silicon detector ever built. Large substructures of the tracker have been tested in test beams at CERN in May/June and September 2004. In this report, results from these test beams are presented.

1. The CMS Silicon Strip Tracker The CMS silicon strip tracker consists of four subsystems: the Tracker Inner Barrel (TIB), the Tracker Inner Disks (TID), the Tracker Outer Barrel (TOB) and the Tracker End Caps (TEC). Fig. 1 shows one quarter of the longitudinal cross section of the tracker (more details are available in [1]). In the barrel region, rectangular silicon strip modules are mounted with the strips parallel to the beam axis, whereas in the inner disks and end caps wedge shaped modules are mounted with the strips in radial direction. Modules at radii up to 60 cm consist of one single silicon sensor with a thickness of 320 /an, while modules at larger radii consist of two daisychained sensors with a thickness of 500/xm. In the tracker both singleand double-sided modules are used, the latter consisting of two single-sided modules that are mounted back-to-back at a stereo angle of 100 mrad. Throughout the tracker, the ratio of strip width over pitch is kept constant at ^p = 0.25. The TIB consists of four cylindrical layers which are each constructed from four carbon fibre (CF) half-shells. On the inside and outside of the half-shell surfaces, the silicon modules are mounted in strings carrying three 1049

1050 0.1 0.2 0.3

4

»

«

0.4

M

0.5

M

0.6

1M0

0.7

1200

0.8

1400

0.9

1400

1

lgOO

1.1

2M0

l.Z

2200

1.3

24W

1444

1.4

1.5

28O0

Figure 1. One quarter of a cross section of the CMS silicon strip tracker in longitudinal view. The black and gray lines represent double-sided and single-sided modules respectively.

thin modules with pitches of 80 ixm (layer 1 and 2) or 120 fim (layer 3 and 4). The TID is built from three CF disks per beam direction with three rings of wedge shaped, thin silicon modules per disk, the strip pitch ranging from 81 /im to 158 /im. The basic substructure of the TOB is a rod, a rectangular CF support frame carrying 6 thick silicon modules. Sensors of layers 1-4 of the TOB have 512 strips at a pitch of 183 /xm, sensors of layers 5 and 6 have 768 strips at a smaller pitch of 122 /xm. The two endcaps of the TEC consist of nine CF disks carrying 16 CF support structures (petals) each, with front petals and back petals being mounted in alternating order on the disk sides facing the interaction point and the far end of the detector respectively. On the petals, up to 28 modules are arranged in up to seven rings with the strip length increasing radially. The strip pitch in the end caps ranges from 81 /im to 205
1051 2. Results from the Test Beams Both the TEC and TOB collaborations participated in a test beam experiment at CERN in May/June 2004. This was the first time such large substructures of the CMS tracker were operated, with both setups comprising about 50 silicon modules (approximately 1% of the respective subdetectors). The TEC setup [2] consisted of one back petal and one front petal, which together represent one autonomous unit in terms of data acquisition and detector control. The TOB group used the Cosmic Rack [3], a mechanical structure that can house up to 20 rods in precision mounting. The cosmic rack was equipped with six rods for the test beam. Both setups had a fully optical control and readout system and used almost final FEDs and the latest available version of the official CMS Data Acquisition software. 2.1. The Tracker End Cap Test

Beam

The two petals of the TEC setup were mounted back to back in an insulated aluminium box. The box was flushed with dry Nitrogen and could be cooled to a temperature of about —15°C. The modules were mostly kept at CMS operation temperatures. Only a small amount of data was recorded at room temperature. For both low and high voltages floating power supplies (not of the final design) were used and the modules were operated with a bias voltage of 300 V, which is well above the depletion voltage for all used sensors. In order to be able to test all modules in the setup, the petal box was mounted on an X-Y-table which could be moved in the plane vertical to the beam. The system showed an excellent overall performance, running stably with a low and uniform noise and a high signal-to-noise-ratio (S/N). Typical S/N distributions for a thin sensor are shown in Fig. 2(a) for peak and deconvolution mode. The mean S/N for each geometry is shown in Fig. 2(b). The data set closest to LHC running conditions is the one recorded in deconvolution mode in the cold, where the S/N stays above 19 for all geometries. This guarantees a sufficiently high S/N even after ten years of operation in the hostile environment at the LHC, where a decrease of the S/N of at most 25% is expected due to radiation damage [4]. The signal arriving at the ADCs is scaled by an a priori unknown factor depending on the gain of the optical readout components. In order to calibrate the readout chain, it is assumed that a MIP creates most likely 24000 e~ in 300 /xm of silicon. Based on the most probable value of the cluster charge distributions, each APV25 is calibrated seperately. Figure

1052 4

Deconv -19.63 Peak • 29.01

-

•

'r

i

I 1 i 4

i

-

.

*

r

i

*

i

f Peak cold

-

* «

Peak warm

• (h)

,

,

,

-

*-:

Deconv cold

• Deconv warm

1

6 7 Ring number

S/N

Figure 2. (a) Distributions of the signal-to-noise-ratio in deconvolution (black) and peak mode (gray) for a ring 4 module and (b) mean S/N for each ring. In (b), each entry represents the mean S/N of a certain geometry, calculated from t h e mean of all APV25 with sufficient beam statistics. The uncertainties correspond to the RMS of the resulting distributions.

3(a) shows the obtained equivalent noise charge (ENC), plotted with respect to the strip length. The noise depends linearly on the module capacitance which scales linearly with the strip length. The resulting linear dependence of the calibrated noise on the strip length is clearly observed. The measured noise is in reasonable agreement with the expectations [2]. The measured ENC ranges from 1019 e~ (ring 1) to 1681 e~ (ring 7) in deconvolution mode in the cold environment. The mean common mode noise in electrons amounts to (173 ± 28) e~ in peak mode and (299 ± 76) e~~ in deconvolution mode (Fig. 3(b)).

r-^2200

, •

-i £1800 1600

-I

A TEC Peak cold

' ' ' 1 '

TOB peak cold

r""""

600

.

i

.Aw*"""*"'^

.

.

.

i

120

i

.

.

1

<

•:dr

_ .1 i

.

140

.

.

~

(a . . 1 . . 1 1 1 f 160 180 200 Strip length [ m m ] i

13 I : (b)

•

i

L ---""iStr*" .

.

"J"""

1000 800

l

\

1400 7 1200

i

~

TEC Deconv cold TEC Deconv warm

-o *

•

h

^

is: :

™w, 100

-j Peak mode -

<

U

Deconvolution mode

1

200 300 400 500 600 700 800 900 Width of common mode [e" ]

Figure 3. (a) The mean ENC versus the strip length for peak and deconvolution mode and the two operating temperatures and (b) distributions of calibrated common mode noise. In (a), each entry represents the mean ENC of a certain geometry, calculated from the mean of all APV25 with sufficient beam statistics. The uncertainties correspond to the RMS of the resulting distributions. ENC obtained in the cold for T O B sensors is included for comparison.

1053 2.2. The Tracker

Outer Barrel

Test

Beam

The Cosmic Rack of the TOB was operated at room temperature in the first test beam (May 2004) and with a silicon temperature of about —15°C in the second one (September 2004). Measurements of the S/N and ENC match the numbers obtained on TEC petals. In deconvolution mode, a S/N of 25 was measured and the observed ENC was (1582 ± 74) e~ for modules with a pitch of 183 /xm [3]. With a strip length of 186 mm, the TOB modules are slightly longer than TEC ring 6 modules (strip length 183.9 mm), where an ENC of (1517 ± 47) e~ was determined. With the cosmic rack it is possible to track particles through several silicon modules with well known relative positions, allowing to measure the hit resolution. In deconvolution mode, a resolution of (65 ± 5) /xm was obtained for modules with a pitch of 183 fim. The measured hit efficiencies were above 99%. 3. Conclusions and Outlook In 2004, large subsystems of the CMS silicon strip tracker have been tested in test beam experiments. The stably running systems showed an excellent overall performance and give confidence for the integration of the individual subdetectors. In the case of the TIB and TID, integration has already started and for TOB and TEC it will start at the end of 2005. Acknowledgments This work is supported by the Graduate College 729 of the German Research Foundation (DFG). References 1. D. Abbaneo, Layout and Performance of the CMS Silicon Strip Tracker Nucl. Instr. and Methods A518, 331-335 (2004). 2. R. Brauer, K. Klein et al, Design and Test Beam Performance of Substructures of the CMS Tracker End Caps, CMS Note in preparation (2005). 3. A. Dierlamm, private communication. 4. CMS Collaboration, The Tracker Project, Technical Design Report, CERN/LHCC 98-6, CMS TDR 5 (15. April 1998) and CMS collaboration, Addendum to the CMS Tracker TDR, CERN/LHCC 200016, CMS TDR 5 Addendum 1 (21. April 2000).

TEST, QUALIFICATION A N D ELECTRONICS I N T E G R A T I O N OF T H E ALICE SILICON PIXEL D E T E C T O R MODULES

I.A.CALI 7 ' 2 , G.ANELLI 2 , F.ANTINORI 5 , A.BADALA^, A.BOCCARDI 2 , G.E.BRUNO-', M.BURNS 2 , M.CAMPBELL 2 , M.CASELLE"*, S.CERESA 2 , P.CHOCHULA 2 ' 5 , M.CINAUSERO 6 , J.CONRAD 2 , R.DIMA 5 , D.ELIA ; , D.FABRIS 5 , R.A.FINI 1 , E.FIORETTO 6 , F.FORMENTI 2 , B.GHIDINI 1 , S.KAPUSTA 5 , A.KLUGE 2 , M.KRIVDA 7 , V.LENTI^, F.LIBRIZZP*, M.LUNARDON 3 , V.MANZARI^, M.MOREL 2 , S.MORETTO 5 , F.NAVACH J , P.NILSSON 2 , F.OSMIC 2 , G.S.PAPPALARDO 4 , V.PATICCHIO-*, A.PEPATO 5 , G.PRETE 6 , A.PULVIRENTI^, P.RIEDLER 2 , F.RIGGT*, L.SANDOR 7 , R.SANTORO^, F.SCARLASSARA 3 , G.SEGATO 5 , F.SORAMEL*, G.STEFANINI 2 , C.TORCATO DE MATOS 2 , R.TURRISI 5 , L.VANNUCCI 6 , G.VIESTI 3 , T.VIRGILI 9 1 Dipartimento di Fisica e Sez. INFN di Bari, 1-70126, Bari, Italy 2 CERN - European Organization for Nuclear Research, CH-1211 Geneva 23, Switzerland 3 Dipartimento di Fisica e Sezione INFN, 1-35131 Padova, Italy 4 Dipartimento di Fisica e Sezione INFN, 1-95129 Catania, Italy Comenius University, SK-84215 Bratislava, Slovakia Laboratori Nazionali di Legnaro, 1-35020 Legnaro, Italy 7 Slovak Academy of Sciences, SK-04353, Kosice, Slovakia 8 Dipartimento di Fisica dell'Universita' e Gruppo collegato INFN di Udine, 1-33100 Udine, Italy 9 Universita' di Salerno e Sezione INFN, 1-84081 Baronissi, Italy

The ALICE Silicon Pixel Detector (SPD) consists of two cylindrical barrel layers with ~ 107 active hybrid pixel cells in total. The requirements in radiation hardness and the challenging material budget and dimensional constraints have led to specific technology developments and novel solutions. An overview of the SPD and its electronics readout system is presented. The procedures for detector module testing, qualification and integration are reported.

1054

1055

1. Introduction The ALICE Silicon Pixel Detector (SPD) consists of two cylindrical barrel layers with ~ 107 active hybrid pixel cells in total. The SPD comprises 1200 front-end ASICs, 150 \mx thick, bump bonded in groups of 5 to 200 /im thick silicon sensors to form ladders. Each ladder contains nearly 41K active cells. The basic detector module is the half-stave (HS) which consists of two ladders and one PILOT Multi Chip Module (MCM) interconnected by an aluminium/polyimide multilayer bus. The full SPD contains 120 half-staves (HS) on two half-barrel layers. Comprehensive tests and qualification procedures for the ladders and the HS have been developed. Laboratory and beam tests have been carried out on prototype elements. The following sections give an overview of the on-detector and off-detector electronics. The status of production modules test, qualification and integration is presented. 2. Overview of the ALICE S P D The SPD (see Fig. 1(a)) constitutes the two innermost layers of the ALICE Inner Tracking System (ITS) f1] at radii of 3.9 cm and 7.6 cm respectively, with a total of nearly 10M active hybrid pixel cells with dimensions of 50 /xm (r) x 425 [im (z). The SPD half-staves (see Fig. 1(b)) are mounted on 10 carbon fibre support

(a) Figure 1. jig (b)

(b)

A CAD drawing of the full SPD (a) and an image of an HS on the mounting

sectors. A HS is an assembly of two ladders glued to a multi-layer bus that carries and distributes power and signals. Each HS is equipped with a Multi Chip Module (MCM) connected to a Link Receiver card (LRx) in the Router back-end electronics. The MCM receives and distributes the 10MHz clock and control signals, generates analog reference levels, performs

1056

data multiplexing and serialization, and drives the outgoing data stream at 800Mb/s. The LRx cards provide the clock and control signals and perform zero suppression on the data streams. The I/O communication is via fibreoptic links. Each half-sector (6 HS) is read out by a Router, a VME module that is interfaced to the ALICE DAQ system. A specific feature of the the SPD is that it will provide a multiplicity signal (FastOR) contributing to the L0 (lowest latency) trigger.

3. Test of system components Comprehensive test procedures have been developed for the acceptance and qualification of components and half-staves throughout the prototyping, production and assembly sequence. The pixel ASIC [2] includes 42 internal DACs that allow adjusting all the key operating parameters, such as the global threshold as well as the individual threshold in each pixel cell. The most elementary functional test is based on electrical pulses, internally generated with programmable amplitude and timing. The detector efficiency has been studied by varying the pulse amplitude at various threshold settings (S-curves). Detector efficiency and calibration have also been studied using a 109Cd radioactive source. Measurements have been taken at various settings of the global threshold DAC. The correspondence between the charge deposited in the detector and the DAC value has been determined. Combining the results of the electrical test pulse and the source measurements, a mean threshold of ~ 1500e~ with a RMS noise of ~ 200e~ has been found. The test pulse has also been calibrated and a conversion factor of ~ 66e~ /mV has been determined. The results obtained are in agreement with those found in previous ladder tests [3]. Each Multi Chip Module (MCM) contains several ASICs [4]: the Digital Pilot (DP), the Analog Pilot (AP) and the Gigabit optical link serializer/driver (GOL). It also contains an optical transducer package connected via fibres to the the back-end electronics (LRx cards in the Router). Following electrical/optical conversion in the transducer, the incoming clock and control signals are distributed by the DP to the 10 pixel chips in the corresponding half-stave. The 10 MHz pixel chip clock is generated by the DP using the incoming 40 MHz TTC clock. The raw data of the 10 pixel chips are multiplexed and serialized in the DP, then transmitted to the GOL that drives a laser diode in the optical package at 800Mb/s using a G-link compatible protocol. The jitter recovery performed by the DP has been studied applying clocks with different jitter levels (29 ps incoming clock —» 9 ps produced clock). The optical link has

1057

been studied measuring the signal integrity of both the reconstructed clock and the control data stream. The incoming optical power has been reduced until the DP was no longer able to decode the data correctly. The optical power margin is 12dB approximately. This is adequate to compensate for losses due to connector break points and radiation effects. Long time stability tests have been carried out and eye diagrams examined on the 800 MHz G-Link communication varying the configurations parameters. The GOL configuration has been optimized in order to minimize both jitter (12 ps) and bit error rate (10~ 17 ). The SPD can provide a trigger input signal to the ALICE central trigger processor using the built-in Fast-OR functionality: in each chip, an electric pulse is fired whenever a hit is detected in a cell. The Fast-OR efficiency has been studied using a 90Sr radioactive source with a scintillator telescope. The Fast-OR efficiency, defined as the ratio of the number of events recorded to the number of scintillator triggers, has been found to be 99%. The Fast-OR response uniformity across the pixel matrix has been determined by pulsing single pixels and counting the corresponding number of Fast-OR signals generated. The response uniformity is well within the requirements.

4. Half Stave test system The half-stave functionality has been studied using specially developed electronics cards and software tools. The test system is based on a VME module with extensive FPGA programmable functionality that, together with an LRx card and dedicated software, is capable of emulating the data acquisition, trigger and detector control systems. The communication between the HS and the test system is via the same optical link as used in the final DAQ. The system makes use of boundary scan to identify missing connections or short circuits both on the multi-layer bus and on the wire bonding connections of the HS. With a sequence of successive configuration and status accesses all possible defects are identified. In a fully automated way the system scans the pixels matrices using test pulses and modifies both the internal DACs settings of the readout chip and the pulses amplitude. An online analysis tool provides, studying the data acquired, response matrices, S-curves and determines the best DACs settings. Noisy and dead pixels are also automatically identified during source tests. A full test and calibration run requires 30 minutes approximately. The system generates a configuration file that contains all the information required to operate

1058 the detector and keep track of its status. Some of these data are the test parameters, the hardware status (integrity of connections, dead or noisy pixels) and the calibration parameters found during the test procedure. The files produced are automatically stored in the construction database. The test procedure is repeated several times during the various steps of the SPD assembly in order to detect any possible malfunction and track the stability of the system. 5. B e a m test 2004 and ALICE Trigger, DAQ and DCS integration A combined beam test of prototypes of the ALICE Inner Tracking System (ITS) was performed in November 2004 in the H4 line at CERN SPS with 158 GeV/c protons. The setup included two detector modules in each of the three silicon ITS technologies: pixels, drift (SDD) and strips (SSD). This was the first test of the detectors integrated with the ALICE DAQ and trigger systems. The full SPD electronic chain was operated successfully. Fig.2(a) shows an example of the correlation plots - SPD and SDD in this case - that were obtained. The full data analysis is in progress. 6. S P D calibration in the ALICE D C S framework The SPD calibration system is being developed within the ALICE Detector Control System framework. A block diagram is shown in fig.2(b). The communication between the different component is based on T C P / I P using a DIM protocol. Software tools in C + + , PVSS and ROOT allow to simulate the ALICE DAQ and trigger system during the calibration phases. The data are retrieved via VME. A dedicated tool will analyze the detector response and identify the configuration required for the operating conditions. The prototype of this system is nearing completion and will be used in December 2005 to test the first SPD sector during final integration and commissioning. The control and readout electronics will be the final one based on the Router and the LinkRx cards. 7. S u m m a r y The SPD components prototypes have been fully validated in laboratory as well as beam tests. The production is under way; the first sector has been fully assembled. Test procedures have been developed and implemented in all the production steps. The ALICE on-detector and off-detector electronics performs to specifications. A prototype of the final system for detector

1059 Correlation SPD plane 0 - SDD plane 1 run 69 -1000

ftveilt*

t

ANALYSIS TOOL SCADACWSS) | (ROOT) Control Unit j jCommands & (Server Status' DI^ CLIENT

SDD coordinate 1 (anodes)

(a)

(b)

Figure 2. Correlation plot between the second plane of the the SPD and the first plane of the SDD (a); A block diagram of the ALICE SPD detector control system (b) calibration and control system is being developed a n d will b e operated in the integration test of t h e first sector.

References 1. ITS Technical Design Report, CERN-LHCC 99-12 (1999). 2. K. Wyllie et al., "Front-end pixel chips for tracking in ALICE and particle identification in LHCb", Proceeding of the Pixel 2002 Conference, SLAC Electronic Conference Proceedings, Carmel, USA, September 2002. 3. P. Riedler et al., "Recent test results of the ALICE silicon pixel detector", Nuclear Instruments and Methods in Physics Research A, (2005), 549, p. 65-69. 4. A. Kluge et al., "The ALICE Silicon Pixel Detector Front-end and Read-out Electronics", Proceedings of the Vertex 2004 Conference, to be published in Computer Physics Communications.

LAYOUT A N D STATUS OF T H E CMS SILICON T R A C K E R

M. E P P A R D * CERN, PH department, CH-1211 Geneva 23, Switzerland E-mail: Michael. [email protected]

With a total area of more than 200 square meters and about 15,000 silicon detectors the Tracker of the CMS experiment at the Large Hadron Collider will be the largest silicon strip detector ever built. This device is immersed in a hostile radiation environment in presence of a 4 T magnetic field and operates in a controlled subzero temperature and dry humidity atmosphere. Together with a pixel detection system the CMS silicon strip Tracker will determine the charged particle momenta and will play a determinant role in lepton reconstruction and heavy flavour quark tagging. A description of the detector and the status of the construction will be given.

1. Introduction CMS (Compact Muon Solenoid) is one of the two general purpose experiments at the Large Hadron Collider (LHC) at CERN. The apparatus is designed to detect new physics at the TeV-scale (Higgs physics, SuperSymmetry, extra dimensions). To reach this goal a robust tracking and vertex reconstruction with fine granularity and fast response time of detectors and readout electronics is necessary. At high luminosity (10 3 4 cm _ 2 s _ 1 ), on average 24 minimum bias events are produced every 25ns. The Tracker is immersed in a 4T solenoidal magnetic field for precise transverse momentum measurement that modifies the 1/r2 scaling low for the charged track density. A transverse momentum resolution of 1 — 2 % for 100 GeV/c tracks is required for precise reconstruction of heavy narrow particles. An impact parameter resolution of 10 — 20 ^m is needed for b and r tagging with displayed vertices.

* On behalf of the CMS Collaboration

1060

1061 2. Layout Since silicon sensors enable a fast read-out and stand the harsh LHC radiation environment (up to Feq = 1.6 • 10 14 niMev/cm 2 ), a silicon tracker was chosen for the CMS experiment 1 ' 2 . 2.1. Pixel

Detector

The innermost region of the CMS Tracker is occupied by a pixel detector with an outer radius of 10.2 cm. The barrel of the pixel detector has three layers with radii of 4.3 cm, 7.3 cm and 10 cm, for a total of 48 million pixels. In addition, there are two pairs of disks with additional 18 million pixels. The two disks extend from 3 — 7.5 cm in radius and are located at 34.5 cm and 46.5 cm from the centre of the barrel. The silicon modules in the four disks are tilted by 20° to increase the charge sharing between the pixels. The pixels are made of diffusively oxygenated 285 mm thick silicon with a resistivity of 3 — 5 kWcm and have the dimensions of 100 f»m x 150 /jm. An active cooling system will cool down the silicon base plate to a temperature of — 10°C for most of the time to reduce the leakage currents and the reverse annealing effects of the silicon sensors. The pixel sensors are connected to the readout chip via Indium bumps. The readout chip is realized in 0.25 [im technology. Each pixel is connected to a preamplifier and shaper with 25 ns shaping time. 2.2. Silicon Strip

Tracker

The CMS Silicon Strip Tracker (SST) covers the radial range between 20 cm and 110 cm. Figure 1 shows a schematic view of one quarter of the SST in the r - z projection and the pseudorapidity coverage. The barrel region (| z |< 120 cm) is split into an inner barrel (TIB) made of four layers, and an outer barrel (TOB) made of six layers. The TIB is shorter than the TOB, and is complemented by three inner disks (TID) made of three rings. The region 120cm < | z |< 280cm is covered by two end caps (TEC), which consist of nine disks each. Detectors of the TIB, TID and of the four innermost rings of the TEC have strip lengths of around 10 cm and pitch of around 100 (im, giving a surface of about 0.1 cm 2 per channel. These detectors are made of one sensor with 320 fim thickness. In the outer part of the Tracker (TOB and three outermost TEC rings) strip length and pitch are increased by about a factor of two in order to limit the number of channels. The increase in

1062

0 •-—- " 0 20 40

60

80 100 120 140 160 180 200 220 240 260 280 cm

Figure 1. One quarter cut of the CMS Silicon Strip Tracker. Double sided modules populate the regions with 20 cm < R < 40 cm and 60 cm < R < 75 cm (indicated with dark bars). The rest is equipped with single sided modules. The TIB has 4 layers of modules, the T O B is made of 6 layers, one TID has 3 disks and one T E C consists of 9 disks segmented in 7 to 4 rings.

strip length is realized by bonding together two sensors. To compensate for the increase of noise due to the higher inter-strip capacitance, a silicon thickness of 500 /im is chosen for these larger detectors. 2.3. Silicon

Modules

In the first integration step the silicon sensors are mounted on modules with 31 different geometries. Each module consists of a carbon fibre frame, one or two sensors and a front-end hybrid containing the read-out electronics (see Sec. 3.2). In total 15,232 modules have to be assembled with high precision. This is achieved by using automatic assembly robots ("gantry") in a clean room environment. There are six gantries in operation with a production capacity of 20 modules per gantry per day. Only 1% of the produced modules fails the specifications. After the assembly the modules are wire bonded on fully automated bonding machines with a throughput of five modules per machine per day. All bonded modules are tested for channel defects like shorts, pinholes or global faults like high currents or faulty connectivity. The modules are then mounted on larger structures with geometries depending on the location in the Tracker. 3. Production &: Status The unprecedented size and complexity of the CMS Tracker requires atten-

1063 tion to logistics, industry involvement, quality assurance (QA) and design validations 3 . 3.1. Pixel

detector

The barrel part of the pixel detector will be built at the Paul Scherrer Institut in Switzerland. In September 2005 they have received 160 out of 400 sensors. They have in hand 10 good 8 inch wafers of which approx. 2000 read-out chips can be produced. In November 2005 there will be a pre-series of pixel barrel modules. The forward pixel detector will be completely built in USA. A successful test of one panel with I of the pixels mounted on it was performed. The first batch of the final sensor delivery is expected in December 2005; and the second batch will come in January 2006. The production of the forward pixel detector will start in spring 2006 and will be delivered to CERN in summer 2007.

3.2. Hybrids

&

Modules

The front-end hybrid is a flexible Kapton substrate laminated onto a ceramic carrier. On the hybrid four or six ASIC readout chips are mounted 4 . In addition the Detector Control Unit, the PLL chip, and a multiplexer chip are mounted on the hybrid. After several problems with broken electrical lines, weak bonds and open vias 5 , new quality assurance and quality control procedures were implemented by CMS at the producers. Since then the production rate of 400 known good hybrids is sustained and the end of hybrid production is foreseen for December 2005. The assembly of silicon modules for the forward TIB and TID have been completed. The production yield of TIB/TID modules is 96% and the production is expected to end in December 2005 (including 10% of spare modules). The production of modules for the TOB has been slowed down due to a problem with the glued contact on the backplane of the module for the bias voltage. In total 2,600 TOB modules have been produced (46%), with a production yield of 99%. Following the ramping up of the delivery of TEC hybrids to the gantry centres in Europe and US, the TEC module production has gained full speed. In total 2,900 TEC modules have been produced (41%) with a yield of 93%.

1064 3.3. Tracker Inner Barrel and

Disks

The mechanics of the Tracker Inner Barrel consist of two barrels (forward and backward) with four layers each. Each half layer is made of two half cylinders. The third and fourth layer of the forward TIB have been completely assembled (see Fig. 2). The third layer is fully tested and the fourth layer is being tested in Pisa (Italy). The mechanical structure of forward layer 2 is ready and an integration exercise with dummy modules has been performed. The mechanical structure for the first layer in forward direction is ready. All mechanical structures for the backward TIB will be ready by the end of 2005. The integration of the first disk of the TID has started in September 2005.

Figure 2.

View of the forward TIB layer 4 half-shell assembly in Pisa (Italy).

3.4. Tracker Outer

Barrel

The Tracker Outer Barrel silicon modules are mounted on a ladder structure called "rod". The 688 (+ 65 spares) carbon fibre rod frames have been produced in Helsinki (Finland), where they undergo a mechanical quality control. The pins, which define the positions of the module frames, are mounted with a precision of 30 fim. The frames are sent to CERN, where the electronics are mounted, and a detailed quality control is performed including electrical connection checks and functional tests of the compo-

1065 nents. The cabled rods are sent to the two module assembly centres at the University of California, Santa Barbara, and FermiLab in Chicago (both USA), where the silicon modules are mounted on the rods. The assembled rods are shipped back to CERN, where another setup is used to verify the functionality by taking signal data at room temperature using a beta source. The final rods will be inserted into the TOB structure, which is a cylinder made of carbon fibre with a length of 2180 mm and an outer diameter of 2320 mm. The TOB cylinder has been completely assembled and measured at CERN (see Fig. 3.4). The precision of the assembly is better than 200/xm over its whole size.

Figure 3.

3.5. Tracker End

View of the Tracker Outer Barrel structure at CERN.

Caps

The TEC silicon modules are mounted on disk segments called "petals". 150 out of 288 petal frames have been produced. The precision machining of the module support inserts and the integration of the interconnect boards is done in Aachen (Germany) for all petals. The control units, analogue optical hybrids and modules are mounted in seven different petal integration

1066 centres, which are expected to produce 1.5 petals per week per centre. The fully equipped and tested petals are delivered to Aachen and CERN, where the two Tracker end caps are assembled. The two carbon fibre structures which will hold the petals with their service channels, inner shells and aluminium- inserts are available and measured. The precision of the carbon fibre structure of the forward TEC is better than 200 /xm. The insertion of petals into the TEC structures will be completed in June 2006. 3.6. Tracker

Integration

The carbon fibre support structure to hold all Tracker sub-structures was delivered to CERN in September 2004. The whole Silicon Strip Tracker will be operated at a temperature of —10° C to reduce the leakage currents and the reverse annealing effects of the silicon sensors. Therefore, an active cooling system ("thermal screen") is installed on the inner surface of the support tube. A thermal screen panel consists of a curved cold plate, a layer of insulating material and a thin heating foil on top. Thirty-two panels, divided in two rows, each one spanning half-length of the Tracker, accomplish the coverage of the detector surface. The thermal screen has been fully commissioned in 2005 with an achieved minimum air temperature inside the Tracker support tube of —20° C and a dew point of —60° C. A new clean room with a surface area of 350 m 2 is being installed at CERN for the final assembly of the CMS Silicon Strip Tracker. In December 2005 the final integration will start with the assembly of the Tracker Outer Barrel. The Tracker integration will end in summer 2006 and the Tracker will be delivered to Cessy (Prance) were the CMS experiment is located. The clean room will be used afterwards for the integration of the CMS pixel detector. The pixel detector will be installed into CMS in 2008 after the LHC pilot run. References 1. Tracker Technical Design Report, CERN, CH-1211 Geneva 23, Switzerland, CERN/LHCC 1998/006, (1998). 2. Addendum to the CMS tracker TDR, CERN, CH-1211 Geneva 23, Switzerland, CERN/LHCC 2000/016, (2000). 3. F. Hartmann, Nuclear Instruments and Methods in Physics Research, A 549 (2005). 4. M. Raymond et al., Nuclear Science Symposium Conference Record, 2000 IEEE Volume 2, (2000). 5. A. Dierlamm, Nuclear Science Symposium Conference Record, 2004 IEEE Volume 1, (2004).

T H E C M S - T R A C K E R D E T E C T O R CONTROLS S Y S T E M

MARTIN FREY Institut fur Experimentelle Kernphysik, Wolfgang-Gaede-Str. 1, 76131 Karlsruhe,

Universitdt Karlsruhe Baden-Wiirttemberg,

(TH) Germany

ON BEHALF O F T H E CMS T R A C K E R COLLABORATION The Tracker of the CMS detector will be at its completion the biggest and most complex silicon detector ever built. T h e Detector Controls System has to deal with the information of thousands of environmental sensors and has to control thousands of device channels to ensure safety, stability and operability. Diverse components of the system communicate with each other via different communication standards. A finite state machine maintains control for the user. In a test system on a 2% scale of the Tracker and during the Tracker integration the whole system will be verified before final implementation in the CMS Detector Controls System.

1. Introduction The Tracker of the CMS a detector consists of silicon pixel and silicon strip detectors covering a surface of around 206 m 2 . There will be 9648128 channels available on 75376 APV b front-end chips on 15232 modules, built of 24328 silicon sensors 1 . The power supply of the detector modules is split up in 3888 low voltage channels for the front-end electronics and, 3888 high voltage channels for the bias voltage of the sensors and 352 channels for the control power electronics2. The Tracker will have to work for at least 10 years in low humidity at -20°C in a hostile radiation environment. These specific environmental conditions are monitored in run mode of the detector, when physics data can be taken, by measuring of around 100000 values of temperatures, low voltages and leakage currents by the Detector Control Units (DCU) on the Front End Controllers which belong to the Data Acquisition System (DAQ). In addition around 3000 temperature sensors and around 1000 sensors for relative humidity are installed to provide permanently basic a b

Compact Muon Solenoid Analogue Pipeline Voltage

1067

1068 background information. The enormous dimension and complexity of the Tracker claim for special requirements on a control system to ensure safety and controllability. 2. E n v i r o n m e n t a l M o n i t o r i n g The electronics of the detector modules, the control rings and the sensors produce heat which can damage the detector. Therefore the environment of the Tracker has to be cooled to a temperature of -20°C 3 . This also decreases the current level of the irradiated sensors and prevents the reverse annealing, an effect happening at higher temperatures after a period of time, worsening the sensor properties by increasing the effects of the radiation damages 4 . A low humidity is needed to prevent condensation. In certain shutdowns higher temperatures are favored for a period of time to exploit the effect of beneficial annealing 4 . It decreases the amount of radiation damages in the beginning of the annealing. If the period gets too long the result becomes negative as the mentioned reverse annealing starts. Therefore even in times of shutdowns of the detector, the environmental conditions must be monitored, to be informed of the temperature dependent annealing of the radiation damages. Around 3000 hardwired temperature sensors and around 1000 hardwired sensors for relative humidity are installed for this purpose to always deliver basic background information. 3. T h e D e t e c t o r Safety S y s t e m The task of the Detector Safety System (DSS) is to keep the detector in a safe state and to react on technical problems or critical environmental conditions, which might have negative effects on the sensor properties or might even lead to damages on the detector. The Tracker DSS is made of an interlock system, which guarantees safety and works independently from any software acting as a basic pillar thereby. Its central elements are Programmable Logical Controllers (PLCs) 5 , which read out the hardwired temperature and humidity sensors and get additional information of the cooling temperature, the cooling flow, the radiation level and the general conditions of the cavern and CMS. The temperature readout is done by PtlOOO sensors and thermistors. The latter provide an input to the PLCs via conditioning cards, which transform the measured voltages and currents into values usable by the PLCs. The PLCs are programmed to identify potentially dangerous situa-

1069

Figure 1.

Dataflow in the CMS Tracker Detector Controls and Safety System

tions and react with immediate shutdowns of the Tracker power supplies if values exceed certain thresholds (see Figure 1). Emergency shutdowns by the PLCs can also be triggered by the CMS controls or CMS safety system.

4. The Detector Controls System The main task of the Detector Controls System (DCS) is to run and to control the detector. It also serves as a second pillar of the safety system, as it has to react on any emergency situation before the DSS takes its actions. This is realized by interpreting environmental data on a higher level to diagnose problems and eventually ramp down power supplies in a controlled way. The necessary basic environmental data is provided by the same fixed temperature and relative humidity sensors which are connected to the PLCs. The controls software of the DCS can read the information

1070

from the PLCs via the communication standards OPC a , S7 b or SOAP c (see Figure 1). In the OPC communication standard, which requires Windows operating systems, the PLCs and the control PCs are clients communicating via a server which can be installed on any PC. The native Siemens S7 driver is meant for a direct communication between the controls PC and the PLC without a server. If the communication shall be based on SOAP, the data of the PLC has to be translated in SOAP messages and a software tool on the controls PC then has to retranslate these messages so that the actual values can be handled by the controls software. Tests are going on to choose the most reliable standard. To achieve a higher granularity of information, the DCS can get additional information of about 100000 measurements of temperatures, low voltages and leakage currents taken by the DAQ. The designated communication standard between the DCS and the DAQ is based on SOAP. The control of the detector by the DCS is achieved by the usage of a finite state machine (FSM). In exactly defined states, depending on the hardware status, specific commands to the devices permit adequate transitions between the states. These commands are mainly switching commands for the voltage channels. For the essential communication between the DCS and the power supplies, the OPC standard is used. The Tracker controls system will be embedded in the DCS of the whole CMS detector and will have to make sure that general CMS states are implemented while maintaining more Tracker specific ones. In the example of switching from OFF to ON by an ON command of the CMS DCS to be ready for taking physics data, different steps have to be done which cause state changes. The system first has to check environmental conditions and the state of the cooling plant before enabling the power supply of the control rings which comprise several electronics for the DAQ. When the DAQ informs the DCS that this part is working, the DCS can read out more environmental sensors of the control rings to have a higher granularity of information for completing the analysis of the conditions. After this the low voltage channels can be activated to power the electronics of the detector modules which leads to a state change to the state ON_LV. If again the DAQ reports that everything is alright, the DCS switches on the high voltage channels which provide the voltage for the silicon sensors. During the ramp up of the voltage, the

a

O L E for Process Control Native Siemens driver c Simple Object Access Protocol b

1071 Tracker is in the s t a t e H V R A M P I N G U P a n d b y reaching t h e final value the s t a t e changes t o ON. During these steps m a n y more sensors and devices have t o be checked and parts of the DCS on lower levels have t o go through several more internal states. For the design of t h e state machine and for the production of user interfaces, a professional S C A D A a software ( P V S S b , by E T M ) has been extended by C E R N in a common LHC framework to adjust it t o t h e project's needs and to gain from overall developments. T h e software also allows the archiving of the values and the t r e a t m e n t of alarms, warnings and error messages. T h e state machine is encoded in a new version of t h e S M I C + + language, which was in use in L E P .

5. T h e Test S y s t e m " C o s m i c R a c k " T h e first prototype of the full DCS is used as an application in t h e test system "Cosmic Rack", which is an autonomous system representing the whole Tracker on a lower scale (2 % ) 6 . T h e prototype will also be needed in t h e next steps of the integration of t h e Tracker in t h e CMS detector, where a p a r t of the Tracker will have t o be operated together with parts of other CMS subdetectors in a final magnetic field configuration. T h e experience, gained in these phases, will help to improve t h e system for the final integration of t h e Tracker in the CMS detector and for refining the end diagnostic before t h e final implementation in the CMS controls system.

References 1. F. Hartmann et al., The CMS All-Silicon Tracker - Strategies to ensure a high quality and radiation hard Silicon Detector, NIM A 478, (2002) 2. S. Paoletti et al., The Powering Scheme of the CMS Silicon Strip Tracker, CERN-2004-010, CERN-LHCC-2004-030, (2004) 3. D. Oellers, Kuhlung des CMS-Spurdetektors, Diploma Thesis, RWTH Aachen, (2005) 4. M. Moll, Radiation Damage in Silicon Particle Detectors, Doctoral Thesis, DESY-THESIS-1999-040, (1999) 5. H. Berger, Automating with STEP7 in STL and SCL, Publicis MCD, (2005) 6. C. Bloch, Commissioning of the CMS Tracker Outer Barrel, ICATPP, Como, (2005)

"Supervisory Control and Data Acquisition b Prozessvisualisierungs- und Steuerungssystem c State Management Interface

OPTIMIZATION OF T H E R E A D O U T P A D G E O M E T R Y FOR A G E M - B A S E D TIME P R O J E C T I O N C H A M B E R

J. K A M I N S K I , S. K A P P L E R * B . L E D E R M A N N , T H . M U L L E R Institut

fur Experimentelle Kernphysik, Universitat Karlsruhe (TH) Postfach 3640, 76137, Karlsruhe, Germany E-mail: [email protected]. de M. T . R O N A N

Lawrence

Berkeley National Laboratory, Mail stop 50B-5239 1 Cyclotron Road, Berkeley, CA 94720, USA

In future time projection chamber (TPC) designs strong magnetic fields are foreseen to reduce transverse cluster sizes and thus improve transverse spatial resolution and double-track resolution. This implies a high granularity of the T P C readout - a feature, which is provided by micropattern T P C readouts based on the Gas Electron Multiplier (GEM). Besides numerous further advantages, GEM readouts feature narrow and direct electron collection signals and allow to take full advantage of small cluster sizes. In this paper we present the study on the pad geometry optimization for large-scale GEM-based T P C s with magnetic fields up to 5T. Measurements with six different pad geometries in a prototype-TPC are presented and the results are compared to Monte Carlo simulations.

1. Introduction Future high energy physics experiments require strong magnetic fields and excellent performance of the central tracking detectors to resolve the momentum of highly energetic particles. In the technical design report (TDR) of the TESLA-project 1 for example, the use of a time projection chamber (TPC) embedded in a 4 T magnetic field is foreseen. The readout is sug*S. Kappler is now with the Rheinisch-Westfalische Technische Hochschule, Aachen, Germany

1072

1073

gested to be based on micropattern gas amplification stages, such as Gas Electron Multipliers (GEMs), and micro pads (rectangular, 2 mm x 6 mm). These devices have many favorable features, some of which are: • suppressed ion backflow into the drift volume, • negligible E x 5-effects, • direct electron collection by pads below the last GEM. The last item mentioned leads to fast and narrow signals. Hence the cluster size depends mostly on the diffusion processes in the drift region and therefore the spatial resolution in the transverse and the drifting direction as well as the double track resolution are significantly improved. However, if the cluster size becomes so small that only one pad collects the complete signal within one pad row, then the transverse spatial resolution (t.s.r.) degrades rapidly to (pitch of pads)/\/l2- We have therefore studied the influence of the pad geometry on the t.s.r. in the limit of low diffusion.

2. Methods of Studying a Pad Geometry In this study six different readout pads were compared to each other by measuring the t.s.r. in an experimental setup and confirming the results with a Monte Carlo simulation.

2.1. Experimental

setup

For testing the various pad geometries a prototype chamber with a drift length of 25 cm and an inner diameter of 20 cm was used (detailed description in 2 ). This prototype detector has been operated in high magnetic fields and hadronic testbeams before3 and has demonstrated good performances with rectangular 1.27 mm x 12.5 mm pads. For the study presented in this paper, a new readout area had been developed to easily exchange the pad geometry, while leaving the remaining detector (especially the GEMs) untouched. The setup (s. Fig. 1) had been placed in a 1 T-dipole magnet at DESY, Hamburg, and 5.2 GeV-electrons had been used to create straight tracks in the detector. In the gas mixture Ar:CH4 (95:5) and in an electric field of 60 V/cm the transverse diffusion coefficient is 116.8 fim/y/cm and therfore only 1.7 times larger than in the aforementioned TESLA-detector. The six pad geometries, that were studied under these conditions, are shown in Fig. 2

1074

Figure 1. Photograph of the prototype detector mounted together with the front-end electronics onto the support.

2.2. Monte

Carlo Simulation

Figure 2. Schematic drawing: a) rectangular pads, b) staggered rectangular pads, c) chevron-shaped pads, d) comblike pads, e) rhombic pads, and f) '3+1'-pads

Tool

The Monte Carlo tool traces electrons generated in the drift volume to the readout pads. It takes into account a realistic model of ionization along the track path, diffusion in the drift region, gas amplification inside the GEM-holes and diffusion below each GEM. The different processes have been modelled according data given in reference4. 2.3. Data

Analysis

The data of the experimental setup and the Monte Carlo simulation were reconstructed and analyzed with the same JAVA-based software package according to the following procedure: First, the baseline was subtracted and the noise of individual pads was determined. Then a center of gravity algorithm was used to calculated the position of the charge clusters in every pad row and these clusters were merged to tracks by a combinatorical track finder. Finally, the t.s.r. is determined by the width of the residuals distribution ares of target row clusters with the reference tracks. Since the reference track is not known accurately, the uncertainty of the track parameters have to be taken into account. This is done by determining the parameters of the reference track by once including and then excluding the information of the target row cluster. The true spatial resolution is then given by the geometric mean of the two different widths of the residual distributions 5 : t.s.r. =y/<7ii7.es * ere,resSince a non-linear charge sharing is expected, if a narrow Gaussiandistributed charge is collected by two broad pads, a correction function has to be applied. An example of such a correction function can be seen in Fig. 3. Here a good agreement of Monte Carlo results, experimental data

1075 L

|

1

1

1

.

,

.

.

.

. 1

0.3 0.2 1 0.1 0

WA W?

'

'

'

*

fc

<J*'

\ xiu*^£

-1

.

.

.

i

.

.

.

.

i ' •'

i ' ' • i • ' '

i '

0.25

i

.

.

.

.

i

.

0.15

1

a.-»«-.T*r~-~

0.1

» a chevrons A A comblike * * rectangular

0.05

-0.3 .

i ' ' '

0.2

~ -1

i ' ' '

0.3

i

7

-0.1 -0.2

i • ' '

•

theo. Modell o A v Monte Carlo 7.5, U.5, 17.5 cm • * T measurement 6.8,12.2,17.9 cm

.

.

.

• o

i •

-0.5 0 0.5 1 local pad coordinate [mm]

Figure 3. Comparison of the correction function for staggered rectangular pads. The tree different functions are derived from measured values, Monte Carlo data and a theoretical model.

sf U_l

6

8

10 12 14 16 18 drift distance [cm]

Figure 4. Transverse spatial resolution in dependence on drift distance for various pad geometries: closed symbols are experimental results, open symbols Monte Carlo results.

and a theoretical model based on a numerical integration of a homogeneous charge distribution is seen. 3. results The results of the testbeam measurements are shown in Table 1, where the t.s.r. is given for effective gas gains of 4-10 3 at drift distances of 7.5 cm and 17.5 cm. Also the result for higher effective gas gains of 104 and with track inclinations of <j) = —10° are shown for a drift distance of 7.5 cm. Finally, also the probability of hitting only one pad is given. Despite the fact that this probability is highest for the classical rectangular and the staggered pad geometry, they perform best with respect to the t.s.r. This fact is also shown in Fig. 4, where the t.s.r. is shown for four different pad geometries in dependence on the drift distance. Here, the test beam results as well as the Monte Carlo simulation results are shown. Both results agree qualitatively very well, but quantitatively, the Monte Carlo simulation results are 20-50 /mi below the experimental ones originating most likely from simplification in the simulation. 4. Conclusion and Outlook Both the testbeam experiment as well as the Monte Carlo simulation agree, that the staggered rectangular pads performed best and are therefore recommendable for the use in any large scale detector. However, reasons for

1076 Table 1. Summary of experimental results. Listed are transverse spatial resolutions as —2.0°, an effective gas gain of 4 • 10 3 and a drift distance of 7.5 cm and 17.5 cm, respectively. Also transverse spatial resolutions = 10°) for drift distances of 7.5 cm with a gain = 10 4 and track inclinations of 10° are included, ncc indicates the fraction of clusters without charge sharing. Values without errors are linear interpolations between measured values. Pad geometry

<7x


0~x


(7.5 c m )

(17.5 c m )

(gain = 1 0 4 )

(
ncc

in /im

in /im

in /im

in /xm

Rectangular pads

172.3

190

158

186

0.150

2 X 6 mm2

±0.6

±2

±1

±2

±0.001

146

126

119

189

0.211

±1

±2

±2

±0.001

150

130

232

S t a g g e r e d rectangular pads 2 x 6 m m 2 Rhombic pads

179 ±2

Chevron-shaped pads

265

250

220

±12 Comblike p a d s ' 3 + 1 ' pads

315

0.141 ±0.004

231

260

286

0.078

±22

±0.001

437

0.039

±20

±0.001

±16

±9

178

191

174

445

0.074

±3

±3

±3

±3

±0.001

the worse performance of the more exotic pad geometries could be identified and improvements are tested with the Monte Carlo. Acknowledgments The authors would like to thank Norbert Meyners from DESY for helping with test beam related issues and providing the magnet, the FLC-group of DESY and especially T. Behnke and M. Ball for supporting this project in many ways, D. Karlen for modifying the electronics and T. Barvich for building the detector and parts of the equipment. References 1. G. Alexander et al., TESLA Technical Design Report, Part IV: A Detector for TESLA. DESY-01-011 (2001). 2. S. Kappler et al. IEEE Trans. Nucl. Sci. 51(4), 1524 (2004). 3. J. Kaminski et al., Nucl. Instr. Meth. A535/1-2, 201 (2004). 4. W. Blum et al., "Particle Detection with Drift Chambers", Springer, (1993). 5. R. K. Carnegie et al, Nucl. Instr. Meth. A538, 372 (2005).

T R A C K I N G S T R A T E G Y A N D P E R F O R M A N C E FOR T H E ATLAS HIGH LEVEL T R I G G E R S

A. KHOMICH* on behalf of the ATLAS TDAQ-HLT group+ Tracking has a central role in the event selection at the High Level Triggers of ATLAS. The earliest stage where tracking information can be used is the Second Level Trigger, where about 10 ms will be available for event processing. This constraint, together with the high multiplicity environment of ATLAS due to the multiple p p collisions, poses great challenges to the track reconstruction algorithms. In this review, we will describe the pattern recognition strategy for tracking in the HLT, and present results on (a) the tracking performance for different trigger signatures, such as single high-pt leptons, b-jets, and exclusive B decays; and (b) timing measurements of the complete tracking chain, including data access, unpacking, clustering, space point formation and the final pattern recognition.

1. Introduction ATLAS (A Toroidal LHC Apparatus) is one of the four detectors at Large Hadron Collider (LHC) facility at the European Organization for Nuclear Research (CERN). LHC will start operation with initial ("low") luminosity of 2 x 10 3 3 cm~ 2 s - 1 and scale up to the full design ("high") luminosity of 10 3 4 cm _ 2 s _ 1 . ATLAS is a general purpose detector which will register products of pp collisions with center of mass energy of 14 TeV. It is a com* Contact author: [email protected] - University of Mannheim, B6 26, 68131 Mannheim, Germany t A. dos Anjos, S. Armstrong, J.T.M. Baines, C.P. Bee, M. Biglietti, J.A. Bogaerts, M. Bosnian, B. Caron, P. Casado, G. Cataldi, D. Cavalli, M. Cervetto, G. Comune,P. Conde, G. Crone, D. Damazio, M. Diaz Gomez, N. Ellis, D. Emeliyanov, B. Epp, S. Falciano, H. Garitaonandia, S. George, V. Ghete, R. Goncalo, J. Haller, S. Kabana, A. Khomich, G. Kilvington, J. Kirk, N. Konstantinidis, A. Kootz, A.J. Lankford, A. Lowe, L. Luminari, T. Maeno, J. Masik, A. Di Mattia, C. Meessen, A.G. Mello, R. Moore, P. Morettini, A. Negri, N. Nikitin, A. Nisati, C. Osuna, C. Padilla, N. Panikashvili, F. Parodi, V. Perez Reale, J.L. Pinfold, P. Pinto, Z. Qian, S. Resconi, S. Rosati, C. Sanchez, C. Santamarina, A. De Santo, D.A. Scannicchio, C. Schiavi, E. Segura, J.M. de Seixas, S. Sivoklokov, A. Sobreira, R. Soluk, E. Stefanidis, S. Sushkov, M. Sutton, S. Tapprogge, S. Tarem, E. Thomas, F. Touchard, G. Usai, B. Venda Pinto, A. Ventura, V. Vercesi, T. Wengler, P. Werner, S.J. Wheeler, F.J. Wickens, W. Wiedenmann, M. Wielers, G. Zobernig

1077

1078 plex apparatus with nearly 47T geometry around interaction point. From the inside to the outside, it consists of pixel detector, silicon strip detector (SCT) and transition radiation detector (TRT). These three tracking detectors (so called "Inner Detector") are surrounded by electro-magnetic and hadronic calorimeters. Outermost subdetector is a muon spectrometer. The LHC bunch crossing rate of 40 MHz implies fast and efficient trigger system which must bring the event rate to the order of 100 Hz. ATLAS Trigger1 has a three-level architecture. The hardware-based first level (LVL1) makes decision from quick analysis of data from calorimeters and muon subdetectors. High Level Trigger (HLT) consists of the Level-2 (LVL2) and the Event Filter (EF). Both are based on software algorithms running on PC farms. To reduce the amount of data requested to a few percent of the full event size only a part of event data from so called Regions-of-Interest (regions of the detector where the LVL1 Trigger found some activity which lead to accepting an event) is analyzed at Level-2. There are about 1.6 Rols per event in average 1 . At the Event Filter reconstruction will be seeded by the LVL2 Rols and full event processing may not always be necessary. The next important feature of the HLT event selection strategy is a sequential signature validation. Processing is performed in steps of feature extraction and hypothesis testing algorithms. Events should be rejected at the earliest possible step.

2. Algorithms Reconstruction at LVL2 can exploit full-granularity data from all detectors. LVL2 is the earliest stage where the data from the tracking detectors is processed. Detector data should be converted before it can be used by tracking algorithms. The conversion process includes the ByteStream decoding, clusterization, and recording the created cluster collections into transitional data store. Data from transitional data store can be used for future processing. The average execution time for LVL2 algorithms should be about 10 ms per event on 8 GHz CPU. Several options of using detectors information and different conditions (machine and detector) especially at start-up are taken into account, hence more than one algorithm is available to accomplish a defined task. Here is a description of some LVL2 tracking algorithms.

1079 2.1.

SiTrack

SiTrack2 is using only three layers of the silicon trackers. The strategy is to build first track seeds from pairs of hits, with one hit always from the innermost pixel layer and using a Look-Up Table (LUT) connect this layer with the next one out. The track seeds which pass certain selection criteria (e.g. minimum PT) are then used to determine the z0 of the interesting interaction by histogramming their extrapolated z positions along the beam axis. Track seeds consistent with the calculated ZQ are then extended further to a third layer, using another LUT, ambiguities are removed, and the remaining triplets form the track candidates, which are then fitted (currently : analytic least squares fit). 2.2.

IDSCAN

IDSCAN 3 is composed of several sub-steps. First hits are selected in narrow (p slices, pairs of hits in each slice are extrapolated back to the beam line entering the z of the intersection in a histogram. The z-value corresponding to the peak of the histogram is taken as that of the primary vertex. Next step puts all hits into a histogram binned in r\ and
in TRT

One of the possible strategies is to combine stand-alone track finding in silicon trackers with subsequent propagation of found tracks into the TRT using concurrent "hit-to-track" association and track fitting. The Probabilistic Data Association Filter (PDAF) 5 can be used for this purpose. The PDAF is a recursive algorithm computationally similar to the Kalman filter with almost linear execution time. The PDAF consists of two main blocks: data association and track update. First the set of hit-to-track association hypotheses explaining the hit pattern observed inside the validation region are created and probabilities of these hypotheses are calculated. After that track parameters are updated by Kalman filter.

1080 Another strategy is stand-alone reconstruction in TRT. There are two algorithms TRTxK and TRTLUT which can be used for that. Detailed description of trigger algorithms can be found in 6 . 2.4. Event

Filter

The average execution time for event filter algorithms should be about 1 s per event. Therefore offline reconstruction algorithms can be adapted and used. 3. Performance Tracking performance was estimated by using realistic Monte-Carlo simulations of the expected behavior of the ATLAS detector, electronic noise and pile-up. Efficiency of tracking in silicon detectors for isolated electron with pile-up at initial luminosity (2 x 1 0 3 3 c m - 2 s _ 1 ) is about 98 % and about 96 % for pile-up at design luminosity (10 3 4 cm _ 2 s _ 1 ) and for b-jet with pile-up is about 87 % and 81 % correspondingly. Tracking efficiency is high for the entire rj region as it is shown on Figure 1.

»! efficiencv (%Ua

> 15000 MeWl T

80 60

" 1

_ ^

j ^ * "

40 20

q3 Figure 1.

-

2

-

1

0

1

2

i

Silicon tracker algorithms efficiency as a function of pseudorapidity r/.

Timing measurements of full reconstruction chain (including data preparation) was done on workstation with 3.2 GHz Intel CPU. Results of these measurements for isolated electron with pile-up at low and high luminosities are shown in Table 1. Pattern recognition time for b-jet with pile-up is 2.1 ms per Rol for initial luminosity and 3.8 ms per Rol for design luminosity. Both algorithms IDSCAN and SiTrack show similar physics and timing

1081 performance. Most critical part is d a t a preparation especially in T R T and additional work for improving it execution speed should be done. Table 1. Track reconstruction for isolated electron with pile-up. Execution times on a 3.2 GHz PC.

4.

Step

2 x 10 3 3 cm~ 2 s~' 1

1034cm-2s_1

SCT+Pixel Data Preparation T R T Data Preparation Pattern recognition (per Rol) Extension to TRT (per Rol)

2.7 6.2 0.3 0.8

7.6 ms 12 ms 0.4 ms 1.8 ms

ms ms ms ms

Conclusions

Track reconstruction at the ATLAS High Level Trigger is a challenging task. Results we have presented in this paper show t h a t this task can be performed efficiently and fast. There are different LVL2 tracking algorithms with similar physics/time performance b u t different strengths to a d a p t t o different conditions (machine and detector) especially at s t a r t - u p and to achieve flexibility in choosing the optimum reconstruction tools for specific trigger selection.

References 1. ATLAS HLT/DAQ/DCS Group, ATLAS High-Level Triggers, Data Acquisition and Controls Technical Design Report, C E R N / L H C C / 2 0 0 3 - 0 2 2 , (2003). 2. C. Schiavi, M. Cervetto, F. Parodi, N. Kostantinidis, M. Sutton, J. Baines, D. Emeliyanov and H. Drevermann, Fast Tracking for the Second Level Trigger of the ATLAS Experiment Using Silicon Detectors Data, A T L - D A Q - C O N F 2005-011, (2005). 3. J. Baines, H. Drevermann, D. Emeliyanov, N. Konstantinidis, F. Parodi, C. Schiavi and M. Sutton, Fast Tracking for the ATLAS LVL2 Trigger, ATLD A Q - C O N F - 2 0 0 5 - 0 0 1 (2005). 4. D. Emeliyanov, A Kahnan filter for track fitting in TriglDSCAN, ATL-COMDAQ-2004-012, (2004). 5. D. Emeliyanov, The Probabilistic Data Association Filter for the fast tracking in ATLAS Transition Radiation Tracker, ATL-COM-DAQ-2005-022, (2005). 6. S. Armstrong, Algorithms for the ATLAS High Level Trigger, ATL-DAQ2003-002, (2003).

THE LHCB SILICON TRACKER FRANK LEHNER* Physik Institut, Universitat Zurich, Winterthurerstr. CH-8057 Zurich, Switzerland

190

The LHCb Experiment at CERN's Large Hadron Collider LHC is a dedicated B physics experiment that is setup as a single-arm magnetic spectrometer. To fully exploit the physics potential, a good tracking performance with high efficiency in a high particle density environment close to the beam pipe is required. Silicon strip detectors with large readout pitch and long strips will be used as part of the tracking system of the LHCb detector. We will present the design and the actual production status of the LHCb Silicon Tracker.

1. Introduction The LHCb experiment [2] at the Large Hadron Collider LHC is a dedicated B physics experiment to investigate CP-violating phenomena. The detector is set up as a forward spectrometer with a 4 Tm dipole magnet and covers a polar angle of 300 mrad in the bending (horizontal) plane of the magnet. The experiment comprises a vertex detector system (Velo), a tracking system, two Ring Imaging Cherenkov counters for particle identification, a calorimeter system consisting of pre-shower detector, electromagnetic and hadronic calorimeters and a muon system. Tracking of charged particles is provided through the Velo and four detector stations: one station in front of and three behind the dipole magnet. The first tracking station (TT) in front of the magnet consists entirely of silicon microstrip detectors. Using the magnetic fringe field in front of the TT, this detector is part of the Level-1 trigger, which selects events containing particles with high pt. In case of the other three tracking stations only the inner part (IT) is employing silicon strip devices while the outer area is covered with straw tubes. This split solution allows to better cope with the high particle densities around the beam pipe, thus keeping the occupancies tolerable. Simulation studies have shown that the track momentum resolution in LHCb is dominated by multiple scattering over a wide range of momenta. This results in a spatial resolution requirement, which can be met by silicon strip detectors with wide pitch of about 200 um. Large readout pitch and long strips For a complete list of authors see [1].

1082

1083

adapted to the expected hit occupancies are used throughout the Silicon Tracker (IT+TT) in order to reduce the number of readout channels and hence the costs. The Silicon Tracker covers a total surface area of 12 m2 and is segmented in 336 (IT) and 280 (TT) readout units with about 270k channels. 2. Design of the ST detector 2.1. The TT detector The TT station situated in front of the magnet consists of four planar detection layers covering the entire acceptance between 15 mrad and 300 mrad in the horizontal plane. The four detection planes are arranged in two groups of two layers each. The first two detection layers have a strip orientation of 0° and +5° with respect to the vertical axis followed by the second group with two layers of strip orientation of -5° and 0°. All silicon sensors are kept in a common light tight, dry and thermally insulating detector housing with an ambient temperature of about 5°C provided by liquid cooling. The TT detector box is vertically split into two half stations allowing a retraction from the beam pipe. Each detection plane of the TT detector is build out of 150 cm long silicon modules equipped on both ends with readout electronics situated outside the acceptance of the experiment. The modules are mounted vertically or close to vertically on cooling balconies into the detector station frame. Each of the four detector planes consists of 14 or 16 silicon modules of full length and two halflength modules above and below the beam pipe. The basic construction unit during the assembly is a half-length module consisting of seven silicon sensors plus two or three readout hybrids located at the end of the half module. The sensors and readout hybrids are hold together by two carbon fiber rails that are glued along the sensor and hybrid edges. The seven sensors of a half module are segmented into either a 4-3 or a 4-2-1 readout grouping with maximum active strip length of 36 cm. The half modules with the finer grouping of 4-2-1 are situated around the beam pipe. For the 4-3 (4-2-1) half modules, one (two) flexible interconnect cables on polyimide basis route the silicon analogue signal from the sensors to the readout hybrids. 2.2. The IT stations Although the IT covers only few percent of the total area of the tracking stations behind the magnet, about 20% of all tracks are passing through the IT. Due to increased particle fluxes the IT has to cover a larger area in the bending plane of the magnet. The IT comprises of four independent detector boxes, which are

1084

arranged in a cross-shaped way around the beam pipe. Each detector box has four detection layers of silicon planes with strips oriented in 0°, ±5° and 0° relative to the vertical axis. The silicon planes are enclosed in a light-tight and thermally insulating housing. As in the TT station liquid cooling provides an operation temperature of 5°C in the IT detector boxes. The sensitive planes of one detector box are assembled in a modular fashion. The basic building units are silicon ladders with either 11 cm (one sensor) or 22 cm (two sensors) long and 7.8 cm wide active area. The front-end electronics is located on a hybrid at the end of the ladder. A carbon fiber composite sandwich serves as support for sensors and hybrid. The 28 ladders per detector box are mounted via small aluminum balconies to cooling rods, which provide the cooling passage for the liquid coolant. 2.3. Silicon sensors and front-end electronics The IT and TT detectors comprise of AC coupled, single-sided silicon microstrip sensors with 384 (IT) and 512 (TT) strips respectively. In total 896 (504) TT (IT) silicon sensors are necessary to build all detector modules. The TT sensors have a thickness of 500 um and are identical in their layout to the CMS OB2 sensors. They have a readout pitch of 183 um and a size of 96 x 94 mm2. The IT sensors have a size of 78 x 110 mm2 and strip pitch of 198 um. They are either 320 um (one-sensor ladders) or 410 um (two-sensor ladders) thick. The silicon thickness was adjusted to the different ladder types of IT and TT in order to maintain a signal over noise ratio of at least 12:1. The sensors are presently manufactured by Hamamatsu Photonics, Japan and about 85% (25%) of the IT (TT) sensors have been delivered so far. It is expected to have all sensors in hand by end of the year. Upon receiving, the silicon sensors are subject to a quality assurance program. The sensors are visually inspected for scratches and other blemishes. Afterwards, leakage currents and depletion voltages are determined. Moreover, on selected sensor samples an automatic strip testing is performed to detect bad channels. The preliminary results obtained from almost 500 tested sensors indicate an excellent quality of the silicon sensors. The strip of the silicon sensors are read out by integrated circuits, the Beetle chips [3] that are located on front-end hybrids at the end of the detector modules. Analogue data from the Beetle chips are then further sent via differential lines on up to 5 m long twisted pair cables to so-called service boxes located outside the detector acceptance. In the service boxes, the data are then digitized with 8-bit ADC and converted into an optical signal. The digital-optical signal is further

1085 transmitted via fibers of up to 120 m length to the counting house to an optical receiver card, which is located on the Level-1 preprocessing boards. The Beetle chip is a 128-channel ASIC device for 40 MHz sampling and multiplexed deadtimeless readout that is manufactured in a 0.25 um CMOS process and was irradiation tested up to 40 MRad. The chip features for each channel a low-noise charge-sensitive preamplifier, an active CR-RC pulse shaper with a minimum rise time (~13 ns) well below the LHC requirements and an analogue pipeline with a programmable latency of up to 160 sampling times. Upon a trigger the corresponding signals stored in the pipeline are readout within 900 ns. The two main hybrid packages that are employed by the Silicon Tracker carry either three chips for the IT detector modules or four chips for the TT detector modules. The Beetle chips are mounted together with passive electronic components on a flexible printed circuit board. Ceramic pitch adapters are necessary to match the silicon sensor pitch to the pitch of the Beetle input pads. The complete package is attached to heat spreader substrates that provide a low thermal impedance path to further cooling plates. The production of the pitch adapters and the assembly of the hybrid packages including the ultrasonic aluminium wire bonding are done by industry. The preseries consisting of 15% of the total hybrid production quantity has been delivered recently and went successfully through numerous functionality tests including a long-term burn-in. 3. Production and testing of detector modules The series production of the IT silicon ladders and TT half modules has just started. Production room facilities at CERN/Lausanne for IT and Zurich for TT are in place and assembly equipment is set up and in operation. In a pre-series phase, which lasted from March to August 2005 many details in the module assembly procedures were improved and refined. A few problems that appeared during the pre-series production could be identified and solved. For instance the HV insulation between carbon fiber support and silicon sensor backplane for IT and TT detectors was significantly improved to prevent sparking. After optimizing the production jigs for the detectors, the achieved mechanical accuracy of assembled silicon devices is now well within the specifications. Although the production is not yet proceeding at full pace, it is expected to be able to ramp up to full production speed within the next months. The anticipated rate will then be 12 ladders/week for the IT and 5 half modules/week for the TT in order to guarantee the completion of all detector devices in time. The schedule

1086

calls for the installation of the IT and TT detector boxes into the LHCb experimental hall by June/July 2006. After the assembly the IT and TT detector modules are subject to a burn-in test. For that purpose several dedicated teststands with final readout chain electronics have been set up at CERN (IT) and Zurich (TT) and are presently being commissioned. The goals of the burn-in measurement are a thorough characterization and final electrical grading of the detector modules. The two burn-in boxes for the TT half modules have a very similar but downsized layout as the final TT detector box. The TT burn-in stands are further equipped with optical fibers fed by pulsed infrared laser diodes which illuminate the silicon sensors at various points to generate charge in the silicon and to determine the operation voltage of the modules. During burn-in measurements, the detector signals, the leakage currents, the temperatures and relative humidity within the burn-in box are permanently recorded. Moreover, pulse shape scans will verify the optimal timing settings of the Beetle preamplifiers. Finally, the burn-in stands will allow a controlled thermal cycling of the detector modules and all characterization measurements are repeated at pre-defined temperature set points. 4. Summary Due to the increased particle densities close to the beam pipe a good portion of the LHCb tracking system is realized with silicon detectors. The tracking station before (TT) and the inner part of the three tracking stations (IT) after the LHCb dipole magnet are using silicon microstrip sensors with a large pitch of about 200 urn and readout strips with length of up to 36 cm. The series assembly of detector grade modules has just started after finalizing a pre-series production, which led to several assembly procedure refinements. Even if the LHCb ST detector has not the scale and dimension of the Atlas or CMS silicon trackers, its series production of all detector modules still represents a formidable production task. References 1. 2. 3.

F. Lehner et al., LHCb note 2005-077. LHCb Collaboration, Reoptimized Detector Design and Performance Technical Design Report, CERN-LHCC/2003-30 N. van Bakel, The Beetle Reference Manual, http://wwwasic.kip.uniheidelberg.de

T H E H I SILICON T R A C K E R

G. L E I B E N G U T H * Institute for particle physics ETH Zurich, Ch-8093 Zurich E-mail: Guillaume.Leibenguth@desy.

de

The HI silicon tracker is composed of a barrel (Central Silicon Tracker, CST), and two end-caps (Forward Silicon Tracker, FST and Backward Silicon Trackers, BST). A detailed description of the layout and performance of these detectors is given. Emphasis is also put on to the readout chain, and on the BST trigger part.

1. I n t r o d u c t i o n The HI detector 1 at the electron-proton collider HERA has been equipped with a Central (CST) and a Backward (BST) Silicon Tracker (the backward and forward direction are defined by the proton direction along the z-axis). During the luminosity upgrade of HERA in 2001, the CST and BST were modified to accommodate a new elliptical beam pipe and withstand higher background rates, and the Forward Silicon Tracker (FST) was added. All three detectors employ the same readout scheme, using an application specific integrated circuit (ASIC), analog pipeline chip (APC128) 2 . The BST primary goal is to detect the track of a scattered electron, to permit the distinction of charged and neutral particle, used to measure F2 and Fj,. The CST and FST, in contrary, are used to tag heavy quarks and to improve the resolution of the HI tracker system. 2. Detectors 2.1. The Central

Silicon

Tracker

The Central Silicon Tracker (CST) 3 consists of two cylindrical layers of doubled sided DC-coupled sensors. The inner (outer) layers comprise from 12 (20) ladders. Each ladder is divided into two electrical units, so called "this work is supported by the Swiss National Science Foundation.

1087

1088 half ladders, consisting of three sensors of 300 /xm thickness and of a size of 5.9 x 3.4 cm 2 . The active length of the CST is 35.6 cm for both layers. The sensors have 12 fim wide strip implants on both sides. The pside strips are parallel to the z-axis, leading to a measurement of the coordinate for a given radius. The implants of the n-side are perpendicular to the p-side, and have a pitch of 88 /xm for a measurement of the zcoordinate. 640 channels are read out for each side of each sensor (81920 readout channels in total) and connected to a preamplifier chip on the hybrid (for a description of the readout electronic, see Section 3.). Each hybrid is mounted on balconies glued to a carbon fiber endplate. This arrangement leads to a minimal amount of dead material in the sensitive region of the detector. The total thickness of the CST is about 0.40 g/cm 2 , corresponding to 1.4% Xo. From 1997 to 2000, signal over noise ratios of 19 for the p side and 7 for the n side have been measured, with an asymptotic impact parameter resolution of 57 /xm for high pt tracks. For the 2001 luminosity upgrade, the arrangement of the ladders 4 has been modified to account for the new elliptical beam pipe, and the readout hybrid were replaced with new ones equipped radiation with radiation hard chips 5 . Similar signal over noise ratio are measured with the radiation hard readout chip.

2.2. The Backward

Silicon

Tracker

The BST 6 , installed in 1995, was composed of four disks with 16 single side AC coupled detectors. Two types of modules were used: The r sensors are double metal AC coupled strip detectors with arc shaped strips, having a radius starting form 59.0 mm to 120.4 mm with a pitch of 48 /xm. Every second strip is read out. The contact to the readout pad of the hybrid at the outer edge is provided by a second metal layer. The <j> sensors, AC coupled strip detectors, have the same dimensions as the r sensors, but with strips parallel to one edge of the detector. They are limited at an inner (outer) radius of 59.0 mm (119.9 mm) and the readout pitch is 75 /xm. This design presents two main advantages. First, there is no need of a second metal layer and second, this leads to a reduced capacitive load to the preamplifier. These tf> sensors are called u o r u depending on the strip orientation. The r and 4> modules are mounted on carbon fiber reinforced plastic wheels. In 1998, four extra wheels were added, such that 128 r sensors were in operation between 1998 and 2000 together with eight prototype of the <j> ones, as shown on Fig. 1. A spatial resolution better than 20 /xm has been

1089

Figure 1. View of the Backward Silicon Tracker.

measured. To accommodate for the new beam pipe geometry, after the HERA II upgrade in 2000, one quarter of the wheels had to be removed, leading to a <j> acceptance of only 270 degrees. In addition, four new wheels with pad sensors for triggering purposes were installed (see below). After radiation damage in 2002-2003, all the r sensors had to be replaced with six wheels of u/v sensors (one of them was equipped only of u sensor), the u and v sensors being mounted back to back.

2.3. The Forward

Silicon

Tracker

Using the same wheel layout as the upgraded BST, the FST 7 , new at HERA II, covers 3/4 of the azimuthal angle. Five u/v and two r wheels are used, thus permitting a sufficient resolution to distinguish different tracks of a single event passing through the same sensor. Therefore, stand-alone tracking with the FST is possible. The complete FST has 92160 channels. Unfortunately, due to a water leak that occurred in June 2004, the FST had to be taken out of the HI detector for a year of repair. The signal over noise ratio was found to be 30 for u/v sensors and about 15 for the r sensors. The spatial resolution after alignment is about 12 /j,m.

1090 3. Readout and Signal Processing Until summer 2004, both BST and FST were read out with the APC128 chip, designed at PSI. In 2001, the CST readout chip has been upgraded to a radiation hard version APC128D 5 , in DMILL technology. The output voltage of the APC128 preamplifier is captured on one of the 32 capacitors, that form the analog pipeline for one of the 128 channels. Following a level-2 trigger accept decision, the signal gathered by 10 APC chips on two neighboring hybrid are sequentially read out at a frequency of 1.5 MHz. The full readout time is about 1.1 ms. The CST uses an optical data transmission 8 of the analog readout and the steering signals, whereas the BST and FST employ copper cables. Digitization is performed on a custom PCIbus mezzanine card containing eight 12 bit FADCs, this information being sent to a PowerPC farm, where further data sparsification is performed. The algorithm first makes a rough calculation of the baseline, pedestals are taken into account, and subtracts the baseline in average for 1280 channels. Event-by-event variation of the baseline is comparable to the single channel RMS noise. A second scan of all the channels is performed to remove the pedestals and hits are identified.

4. Trigger Since the hit information of the HI silicon tracker is only available during the event reconstruction, the BST pad detector 9 has been designed to deliver trigger signal within 2.5 /xs as input to the HI level 1 trigger. Four wheels of the AC-coupled pad sensors have been installed to this purpose, having the same geometry as the r or u/v sensors described above. Each sensor contains eight concentric rings of four pad each, having a capacitance between 15 pF (smallest pad) and 50 pF (largest pad). The PRO/A readout chip, an ASIC, was designed in collaboration with IDE AS (Oslo). It provides charge sensitive amplification with variable gain and signal discrimination with adjustable threshold. The hit search is also performed on this chip, at the bunch crossing frequency of 10.4 MHz. The BST pads are controlled and read out by a dedicated firmware running on a high density ALTERA chip. The task of this chip is, using content-addressable memories, to compare the measured hit pattern with preprogrammed masks. They contain a set of hit patterns from high momentum tracks originating from the interaction point or behind the detector (background event). An online counting of hits all over the pad detector area for every bunch crossing makes possible to control the background rate

1091 for HERA and the HI experiment. 5. Radiation Damage Standard running conditions imply relatively low radiation of the HI silicon tracker, less than 100 Gy per year, measured with radio photo luminescence glass dosimeters, mounted on different locations of the detector. However, mis-steering of the beam can cause a significant increase of the dose. In 1999, the inner layer of the CST received a dose of about 250 Gy which lead to leakage current such that the APC128 stored charge was lost during the readout time. Therefore, for the HERA luminosity upgrade, a APC128 radiation hard chip has designed for the CST. Since the FST and the BST hybrids are located at a somewhat larger radius, the old APC128 was not replaced. The startup of the HERA machine after the upgrade was very difficult, leading to very large doses measured close to the beam pipe in the BST. It seems that the origin of the dose comes from synchrotron radiation hitting a collimator close to the BST electronic. The affected modules have been replaced during the 2003 shut-down. Due to severe radiation damage in the detector front-end of both sub-detectors in 2004, and a water leak damaging the FST, the APC128 chip has been redesigned in deep sub-microns process within three months. The new FST and BST will be installed during an upcoming shutdown in November 2005. Acknowledgments I would like to thank Benno List, Peter Kostka and Ilya Tsurin for their help and support. This work has been supported by the Swiss National Fund. References 1. I. Abt et al., [HI Collaboration]. Nucl. Instr. and Meth A 386 (1997) 310-347, and ibid. 348-396. 2. R. Horisberger, D. Pitzl, Nucl. Instr. and Meth A326 (1993) 92-99. 3. D. Pitzl et al, Nucl. Instr. and Meth. A 454 (2000) 344-349. 4. B. List, Nucl. Instr. and Meth. A 501 (2001) 49-53. 5. M. Hilgers, R. Horisberger, Nucl. Instr. and Meth. A481 (2002) 556-565. 6. H. Henschel, R. Lahmann, Nucl. Instr. and Meth. A 453 (2000) 93-97. 7. M. Nozicka, Nucl. Instr. and Meth. A 501 (2003) 54-59. 8. W. Erdmann et al, Nucl. Instr and Meth. A372 (1996) 188-194. 9. I. Tsourine, Nucl. Instr. and Meth. A 501 (2003) 219-221.

T H E LHCB M U O N SYSTEM

M. L E N Z I (on behalf of the LHCb muon group) INFN and University of Florence E-mail: [email protected]

The ability to provide fast muon triggering and efficient offline muon identification is an essential feature of the LHCb experiment. The muon detector is required to have a high efficiency over a large area and an appropriate time resolution to identify the bunch crossing for level-0 triggers. The LHCb muon detector consists of five stations equipped with 1368 Multi Wire Proportional Chambers and 12 Gas Electron Multiplier chambers. The technical design of the chambers is briefly presented and the Quality Control procedures during the various construction steps are described. The method developed for gas gain uniformity measurement is also described together with the results on efficiency of detectors fully equipped with the front-end electronics, obtained from tests with cosmic rays.

1. Introduction The LHCb experiment 1, that will operate at the Large Hadron Collider (LHC) at CERN, has been designed to study CP violation in B meson decays. Muon triggering and offline muon identification are essential to reach these objectives as muons are present in the final states of many C P sensitive decays. In addition, muons provide a very efficient flavour tagging through b —>/xX semileptonic decays. The main goal of the LHCb muon detector 2 is to provide an efficient and robust level-0 muon trigger 3 through a 5-fold coincidence of hits in all stations. Therefore an efficiency greater than 99% per station is required in a gate of 20 ns and, therefore, a time resolution better than 4 ns. Due to the high rate of incident particles (up to 0.1 MHz/cm 2 for inner chambers) the muon system is required to have good rate capability and aging resistance to ensure full functionality for the 10 years lifetime of the experiment. The design of the muon system has been optimized in order to fulfill all these requirements and allows to reach a trigger efficiency of 46% for inclusive b —» /iX events inside the geometrical acceptance. 1092

1093 2. Muon system layout The muon detector consists of five muon tracking stations placed along the beam axis for a total active area of 435 m 2 . The first station, Ml, is placed in front of the calorimeters, while stations M2 to M5 are interleaved with three iron niters and placed downstream the calorimeters. The acceptance of the muon detector is about 20% for muons from inclusive b decays. Each station is subdivided into four regions (R1-R4) with dimensions and granularity shaped in order to keep the occupancy roughly constant over the detector channels. The muon detector is fully equipped with Multi Wire Proportional Chambers (MWPCs) except for Region 1 of Station 1 (1% of the area), where triple-GEM (Gas Electron Multiplier) 2 are used. A MWPC is made of four gaps (two in station Ml), each one with a plane of anode wires between two cathode planes (Fig. 1).

Figure 1. Schematic view of a four gap chamber.

The anode-cathode distance is as short as 2.5mm to have a fast charge collection. The cathode panels are composed of two copper/gold clad FR4 laminates filled with a rigid polyurethanic foam. The anode plane is composed of 30/zm diameter gold-plated tungsten wires with a pitch of 2 mm. The chambers are filled with an Ar/C0 2 /CF4 gas mixture (40%, 55%, 5%). The readout is performed through a combination of cathode and/or wire pads depending on the granularity and particle fluxes foreseen in each region. To ensure a good efficiency the four gas gaps are hard wired in pairs to form two independent double gaps before the connection to the frontend readout where the two double-gaps are logically OR-ed. This structure provides adequate redundancy and robustness to the system. The electronics is based on custom chips especially developed for the Muon System in 0.25/im CMOS radiation hard technology 4 . Short peaking time (10 ns) and low noise (ENC ~ 2000+40 e~/pF) ensure a good time

1094 resolution. Several tests have demonstrated 5 that chambers equipped with such final electronics easily achieve the required time resolution of about 4ns, with 99% efficiency within a 20ns gate.

3. C o n s t r u c t i o n a n d Quality Control The Muon detector consists of 1380 chambers, with 20 different sizes and readout types, that have to be built within two years in six production centers. To guarantee the construction of such a large number of chambers with high quality and reproducibility with limited manpower, several automatic procedures have been developed for all the production phases. Stringent mechanical constraints are imposed by the requirement of chamber response uniformity within a relatively small plateau (~ 150V around a nominal working point of 2620V, see ref. 5 ) limited from below by the 99% efficiency threshold and from above by a maximum allowed cluster size of 1.2, as imposed by the trigger algorithm. To be conservative, a maximum voltage change of ±50V is allowed, corresponding to a gain change of a factor 1.4. A full Monte Carlo simulation of the chambers was performed to evaluate the sensitivity of gas gain to the chamber imperfections 6 . The results have been confirmed on several prototypes tested on particle beams and have set limits to the allowed chamber imperfections, in particular on wire position, wire tension and gap size. In order to check that the produced chambers fulfill these constraints a series of quality tests have been devised. The gap size is one of the more sensitive parameters with respect to gas gain stability. The first consequence is a stringent requirement on panel planarity stating that at least 95% of the surface should be within 50/um and the maximum allowed deviation is lOO^m. In addition the wire fixation bars thickness (that determines the half gap size) should be within ±0.05mm from the nominal value of 2.5mm. The requirement for the wire pitch is 2±0.05 mm. The wire position is precisely determined by the pitch of the wiring machine combs; however it is important to check that no wire is out of acceptance. The wire pitch measurement is performed with two CCD cameras, viewing the two ends of the wires close to the fixation bars, moved with a step motor.A scan of the whole panel is performed taking pictures of a group of wires at each step. The mutual position of the wires is obtained by an analysis of the acquired pictures. A typical result of the pitch measurement for a whole panel with 760 wires is shown in Fig. 2; the precision obtained with this method is

1095 about 10/xm.

Figure 2. Typical result of an automated wire pitch measurement for a panel of with 760 wires.

To avoid mechanical instabilities due to electrostatic repulsion, the mechanical wire tension T must be larger than 0.3N. The upper limit on T is set by the wire elastic limit which is 1.2N. A safe condition is then 0.5N< T < 0.9N. The wire tension is controlled by a brake motor during the wiring procedure; however a test on the whole panel is needed to detect any possible failure occurring during the wiring process. The total number of wires of the muon system is about 3.2 millions: this implies that a fast, automated and reliable system is needed to check their mechanical tension. The wire mechanical tension T is related to its mechanical resonance frequency F through the equation T ~ n{2lF)2 where /x is the wire linear density and I is its length. The adopted method consists in hitting the wire with a light Mylar hammer and let it vibrate with its own fundamental frequency. The light of a few mW laser beam is reflected on the wire and then detected by a photo-diode producing a signal of the same frequency of the vibrating wire. This signal is sent to a standard PC sound card and the wire fundamental frequency is evaluated by applying a Fast Fourier Transform algorithm. This innovative method allows to measure the wire tension in a few seconds with an accuracy of 0.2%. In Fig. 3 the mean value of the wire tension is shown for several panels. To check the gas tightness, a small ( 5mbar) over-pressure is applied to the chamber under test and the pressure difference with respect to a reference hermetically closed chamber is recorded for one hour. The measured curve is then fitted to an exponential function. The maximum allowed gas leakage rate is 2mbar/hour.

1096

•l^'^/~i^**ffr^j^y?f^'&r.!!Jl'f-£*f*mJt/r*S*

||| .

Figure 3.

Mean value of wire tension as a function of panel number.

An automated training procedure is performed on all chambers before applying the operating voltage. The chamber is accepted if all the gaps are able to stand a voltage of 2850V for a few hours with a total dark current smaller than 50nA. As mentioned before, the gas gain uniformity is fundamental to ensure that the working voltage is within the voltage plateau. The test consists of an automatic scan of the wire planes with a radioactive source ( 90 Sr or 137 Cs), recording the measured current at each position. The total measured double-gap current is then compared with the average double-gap current evaluated over the whole set of produced chambers. The maximum gain variation should not deviate by more than a factor 1.4 from the average value. In Fig. 4 the normalized double gap gain is shown as a function of chamber number for chambers of Region 3 of Station M5.

MR3

g

it

*

J

5

Figure 4.

1 0 t 3

•*

-j. M l

-J

2 S 2 $ 3 9 3 $ 4 g < 1 S

Normalized double gap gain as a function of chamber number.

Finally, a sample (~ 10 -j- 20%) of the chambers that satisfy all the previous specifications, equipped with the final readout electronics, is tested with cosmic rays to determine the efficiency plateau and time resolution. Up to six chambers are piled up on a stand with two plastic scintillator planes to provide triggers. All double gaps tested up to now are well within

1097 the required time resolution and detection efficiency.

4.

Conclusions

T h e L H C b muon detector requirements are good time resolution, high efficiency, high rate capability and aging resistance. Extensive tests have shown t h a t the detector design satisfies all these requirements. Several tests are performed on the produced chambers t o verify t h a t all the specifications are satisfied. Due to the high number of chambers to be built, automatic procedures have been developed for all the production and test phases. T h e construction is currently well advanced and the detector should be ready for the first LHC beam.

References 1. LHCb Collaboration, "LHCb Technical Proposal", CERN/LHCC 98-004 (1998). LHCb Collaboration, "LHCb Reoptimized Detector, Design and Performance Technical Design Report", CERN/LHCC 2003-030 (2003). 2. LHCb Collaboration, "LHCb Muon System Technical Design Report", CERN/LHCC 20001-010 (2001). LHCb Collaboration, "Addendum to the Muon System Technical Design Report", CERN/LHCC 2003-002 (2003). LHCb Collaboration, "Second Addendum to the Muon System Technical Design Report", CERN/LHCC 2005-012 (2005). 3. LHCb Collaboration, "LHCb Trigger System Technical Proposal", CERN/LHCC 2003-031 (2003). LHCb Collaboration, "LHCb Reoptimized Detector, Design and Performance Technical Design Report", CERN/LHCC 2003-030 (2003). 4. W. Bonivento et al., Nucl. Instr. and Meth. A 4 9 1 , 233 (2002). 5. M. Anelli et al., "Test of MWPC Prototypes for Region 3 of Station 3 of the LHCb Muon System", CERN-LHCb-2005-021. 6. W. Riegler, LHCb 2000-060 7. P. Ciambrone et al., Nucl. Instr. and Meth. A545, 156 (2005).

TWO- AND THREE-DIMENSIONAL RECONSTRUCTION AND ANALYSIS OF THE STRAW TUBES TOMOGRAPHY IN THE BTEV EXPERIMENT E. BASILE (*), F. BELLUCCI (***), L. BENUSSI, M. BERTANI, S. BIANCO, M.A. CAPONERO (**), D. COLONNA (*), F. DI FALCO (*), F.L. FABBRI, F. FELLI(*), M. GIARDONI, A. LA MONACA, F.MASSA (*), G. MENSITIERI (***), B. ORTENZI, M. PALLOTTA, A. PAOLOZZI (*), L. PASSAMONTI, D.PIERLUIGI, C. PUCCI (*), A. RUSSO, G. SAVIANO (*)t LaboratoriNazionalidiFrascatidell'INFN,

v.E.Fermi 40 00044Frascati (Rome) Italy

F.CASALI, M.BETTUZZI, D. BIANCONI University of Bologna and INFN, Bologna, Italy October 17,2005 A check of the eccentricity of the aluminised kapton straw tubes used in the BTeV experiment is accomplished using X-ray tomography of the sections of tubes modules. 2 and 3-dimensional images of the single tubes and of the modules are reconstructed and analysed. Preliminary results show that a precision better than 40 urn can be reached on the measurement of the straws radii.

1. Introduction The BTeV experiment [1] uses straw tubes glued in modules and embedded in a structure mechanically untensioned[2], where straws and microstrip detectors are integrated, allowing a minimum amount of materials. A check of the eccentricity of the straws tubes and their position is accomplished using Xray tomography of the module sections.

* Permanent address: "La Sapienza" University - Rome. ** Permanent address: ENEA Frascati. *** Permanent address: "Federico II" University - Naples. ••"Permanent address: INFN Sez. Roma 1. + This work was supported by the Italian Istituto Nazionale di Fisica Nucleare and Ministero dell'Istruzione, deH'Universita' e della Ricerca. This work was partially funded by contract EU RII3-CT-2004-506078.

1098

1099

2. Experimental procedure Each straw section is scanned orthogonal to the vertical axis of the tomograph with 27 um resolution. Data are initially reduced in a numerical 8-bit matrix of 1024x1024 points, then converted to an IMAQ Image of Lab VIEW. Figure 1 shows the raw tomographic image of a straw tube section (section#10, 1024x1024 pixels, 27 urn/pixel, 256 values of greys). Not all the tubes of the module are contained in the field of view. Figure 1 also shows evident traces of glue, deposited on the external surfaces of the straw tubes, especially in the points of contact of adjacent tubes. Only the internal surfaces are well defined in the images and, in order to detect and to measure possible mechanical deformations of the tubes, this forces to study the geometry of the these surfaces. Figure 2 shows the distribution of the grey values of the previous figure. An improvement of the signal-to-noise can be obtained just setting an upper threshold to the intensity, as it is shown in Fig.3, reporting the effect of a threshold of 210 on the image of Fig. 1. As an example, an arbitrary straight line crossing two straw tubes is overlay to the same image of Fig.3. The intensity of the pixels along this line is shown in Fig.4, where the two peaks point out the positions of the crossing points (Edges) of the line with the internal surfaces of the two adjacent tubes. The transformation of the Edges coordinates from the line reference to the image reference is easily obtained from the coordinates of the line end points in the image reference. On this base, an automatic procedure is defined in order to obtain the Edges of 14 tubes of a section of the module.

, y<

-H*-.^

}

}

~v=; Fig.l Example of raw tomography image of straw tubes module (1024x1024 pixel, 27 um/pixel)

1100 Intensity Histo (Arbitrary Unity)

C

15

bL

/-.

111 1,«5 "•p \ t

:C3

Intonsrty

Fig.2 Grey intensity histogram of Fig.l.

Fig.3 Same tomography of Fig.l with a grey intensity threshold of 210. The intensity distribution along the segment is shown in Fig.4.

1101 Intensity (Arbitrary Units)

Fig.4 Intensity along the line of Fig.3

The procedure is as follows. First an image containing a Region of Interest is built, then the patterns recognising such a region are extracted from the tomography. At each Edge pointed out on the base of contrast figures, three orthogonal coordinates X,Y,Z, defined in the tomograph system (Tomo reference), are attributed, where the Z is common to all the Edges of the same section. This allows the reconstruction of the 2 and 3-dimensional images of single tubes and of the entire module. We do not expect a perfect positioning of the module on the tomograph reference plane, and even in the case of perfect positioning we would not expect a perfect parallelism between the tubes of the module. Therefore, the section Edges of each straw tube are fitted to an ellipse. The centres of the ellipses of all the sections are in turn fitted to a straight line: the axis of the straw tube. Projecting the Edges of a section on the plane orthogonal to the axis of its straw tube the contribution to an elliptical configuration due to a not perfect verticality of the tube is eliminated. 3. Results We fit the data points to an ellipse, and define as Ellipse Parameter the quantity (PFrt-PFzV2

1102 where P is a point on the ellipse, F 12 are the ellipse foci, and Pfr their distances. In order to evaluate the amount of the mechanical deformation of the straw tubes cross section from the expected circular shape, we then fit the data to a circle. Figure 5 shows the standard deviation of the histograms of the Edge radius, and the width of gaussian fit to the Ellipse Parameter of the projected edges for the 14 straw tubes analyzed. In the worst case the mechanical deformations respect to the circular cross section have a distribution with a standard deviation of about 1.5 pixel, corresponding to about 40um, largely contained in the lOOum specification in order not to spoil the electric field inside the straw tube. The precision of our technique in determining the variation of the straw cross-section from circularity can be estimated by the difference in quadrature of the two variances in Fig.5, which is about 1.2 pixels at most, corresponding to about 30um. The three-dimensional rendering of slices reconstructed is shown in Figure 6.

Straw Tube f

Fig.5 Standard deviation of the histograms of the Edge radius (crosses) and sigma of the gaussian fit to the ellipse parameters of the projected edges (dots) for 14 straw tubes.

1103

!

lb.' • "I'"' •

^HfcJ |i .

"

^

^

-°Qr***

"

v

Fig.6 3-D reconstruction in Tomo reference of the 14 -straw module.

4. Conclusions We have developed a new technique to visualize low-mass surfaces of cylindrical shapes widely used in HEP detectors, such as straw tubes. The technique uses x-ray computed tomography, implemented with an original optical recognition, pattern recognition and analysis code, Labview-based. Preliminary results show how the precision of our technique in determining deviations from circular shapes are better than 30um..

References 1. Fermilab Experiment E-0897/E-0918, J.Butler, S.Stone co-spokespersons; see www-btev.fnal. 2. E.Basile et al., A Novel Approach for an Integrated Straw tube-Microstrip Detector, accepted by Transactions on Nuclear Science (2005).

CRYOGENIC OPERATION OF EDGE-SENSITIVE SILICON MICROSTRIP DETECTORS OTILIA MILITARU(1), STEFFEN GROHMANN(2), GEORG NUSSLE(5), JAAKKO HARKONEN(5), ESA TUOVINEN(3), ZHENG LI(4), TAPIO NIINIKOSKI(4), BLANCA PEREA SOLANO(5), XAVIER ROUBY(1), KRZYSZTOF PIOTRZKOWSKI(1) (1>

Universite Catholique de Louvain, B-1348 Louvain-la-Neuve, Belgium ILK Dresden, Bertolt-Brecht-Allee 20, D-01309 Dresden, Germany (3> Helsinki Institute of Physics, 00014 Helsinki, Finland (4> Brookhaven National Laboratory, Upton, NY 11973-5000, USA (5> CERN, CH-1211 Geneva, Switzerland <2>

In the frame ofRD39

CERN

Collaboration

The CERN RD39 Collaboration is developing radiation hard cryogenic Si detectors for use in the LHC tracking systems after their future upgrades. Within this framework, we are studying the operation of silicon microstrip detector with readout electronics at low temperature. In addition, we have studied the operation efficiency of the silicon micro-strips detectors very close to the physical border of the silicon crystal, in order to reduce as much as possible the insensitive part of such a device.

1. Introduction Silicon sensors are widely used in existent high energy physics experiments and future upgrades. Given that as high as 1016 n^/cm2 fluence is required by experiments for LHC upgrades, new strategies are developing to make silicon and Si sensors more tolerant and hard at particle radiation. During the operation, the silicon sensors suffer from severe radiation damages, which lead to performance degradation. The defects accumulation in the silicon bulk cause an increase of the reverse leakage current due to the radiation induced deep level defects that act as generation-recombination centers. Another effect is the fall depletion voltage increase, causing either breakdown if operated at high biases or an incomplete non-equilibrium charge collection, if operated at partial depletion. 1104

1105 Since the process of carrier generation via the deep levels is an activated process, the resulting leakage current is extremely sensitive to temperature: it decreases exponentially with decreasing temperature. This allows the operation of silicon microstrip detectors at low temperature with greatly reduced leakage current, fast charge colection due to higher mobility and faster readout electronics with less noise. The benefit of cryogenic temperature operation of Si detectors was first highlighted in 1998 [1], when CERN RD39 put in evidence the so called Lazarus effect, which means a significant charge collection efficiency (CCE) recovery at cryogenic temperature for very heavily irradiated Si detectors (>1015 n/cm2) that have, though, minimum CCE at room temperature. Detailed modeling of Lazarus effect was presented and explained in [2]. However, this effect is limited since a decrease of CCE in time, even at cryogenic temperature, from 80% down to around 20%, is observed for heavily irradiated silicon. This effect is due to the trapped charge that can accumulate and lead to the detector medium polarization resulting in a not uniform, time-dependent electric field. In forward bias operation, CCE stays to values that are about 3 times higher than in inverse bias operation [3]. Therefore, current-injection mode operation might be an answer in keeping the CCE at high level. In cases when heavy irradiation is expected, the detector could be used with reverse bias until the resistivity is such that forward bias can be applied. This alternative requires, nevertheless, bipolar readout electronics and bias voltage supply. For certain experiments that have their tracking system as close as possible to the colliding beams (few mm), special geometry with the active area extended all the way up to the dicing edge is needed. Such sensors/devices were developed by RD39 as a direct application of the cryogenic detectors, since low temperature is a key factor in keeping the surface leakage current at few nA level. RD39 shares resources and hardware development with the current experiments at CERN. Essential results will be summarized here, that were achieved during the progress in cryogenic silicon microstrip and edge sensitive detectors study.

2. Development of cryogenic silicon microstrip detector module The key problem in the thermoelastic design of a cryogenic module arises from the thermal stress developed during cool down due to thermal dilatation mismatch of die components, which may lead to the silicon sensor breaking. A prototype of such a cryogenic module using silicon as a structural material can be seen in figure 1 [4], [5], [10]. It was built with a standard design full size FZ

1106 silicon detector (processed in HIP, Helsinki), 520um thick, with p+/n/n+/Al layout, [12]. All strips are connected to the CMS-experiment type hybrid with 4 APV25 readout chips, [9]. Under the pitch adapter a carbon fiber spacer is glued, with two embedded Cu/Ni micropipes for liquid nitrogen flow.

Figure 1. First cryogenic module built with FZ silicon microstrip detector 380 \xm thick, 512 strips.

During operation, the liquid and gaseous phase are separated in such a way that just liquid enters the cooling pipe. The gas flows back and cools the thermal shield of the cryogenic transfer line. With this system good results were obtained, showing a uniform cooling of the module over a wide mass flow range (between 40 up to 170 mg/s). The temperature varied between 145 K, for the A1203 hybrid support and 135 K for the Si sensor and it was stable for several days, being limited in time only by the liquid nitrogen supply. A dedicated test setup has been designed for module operation cooled down in direct contact to liquid nitrogen and thermal regulated by several resistors. The module was tested using the standard CMS-experiment setup for detectors testing with Front-End Driver [6], Front End Controller [7] and Trigger Sequencer Card [8] devices. For detector biased at 400 V, much over the full depletion voltage, the reconstructed pulse shape given by one single strip is presented in Figure 2, in both polarities, inverter on and inverter off. The CMOS readout electronics performance is strong temperature dependent, due to the cryoacceleration effects. At lower temperature the carrier mobility increases, entailing higher current-signal speed, which explains the observed shorter signal rising time. The APV25 readout chips functionality at low temperature was well demonstrated [11], down to 130K. Other electronic components from the module

1107

hybrid, like Phase Locker Line (PLL, [12]) that distributes the clock and trigger signals to the readout chip, at low temperature, is not in phase anymore, therefore our results are correct just down to 220 K. Further investigations will be concentrated to make the readout operational at much lower temperatures. I Ch 0 Strip 07flCalProt

' Ch 0 Strip 079GalProt

Peak Inverter Of F temperature •16"C temperature -20*C tempeiature-SVC

Peak Inverter ON temperature *18*C temperature -20'C tcmpcir.iturc -SVC

Figure 2. Detector response signal shape at different temperatures.

3. Development of edge-sensitive silicon microstrip module The edge-sensitive detectors were conceived as microstrip detectors where the dead, insensitive area, around the dicing edge, is reduced to minimum. Several single-sided 1.5 cm long baby-microstrip detectors from 320um thick wafers were chosen, with p+/n/n+/Al implant layout. Before cut and chemical treatment, the leakage current was around few nA/cm at full depletion voltage. The dicing tool used in this study was the standard laser-dicing machine and the operation was performed at BNL, USA. Two different geometries have been chosen for our purpose. In the first layout, the silicon margin is along the strip, as close as few um. — IQ

.

Strolcjhi-B (no etching) Cut border atrip 85 ,'edgeless' detectors with the cut facing ach other

' 10

I i i i I 40

60

filters for bias voltage

SO

Strip numbex pitch adapter

Figure 3. (a) single strip current measurement for an 'edgeless' detector where the cut is along strip number 85 at 10nm distance; (b) 'edgeless' module.

1108

In the second geometry the angle between the cut edge and the strip direction is of few degrees. After dicing the leakage current increased drastically by five orders of magnitude. Single-strip current measurements (Figure 3, (a)) showed that responsible for the dramatic current increase are the two strips nearest to the cut, at 120 um inter-strip distance. Both geometries were used for the 'edgeless' module, the detectors being mounted in a way that the cuts are facing each other at a distance of around 1mm (Figure 3 (b)). The twenty strips nearest to the margin were wire bonded to the readout channels. Our goal is to test this module at very low temperature, with a Picosecond Injection Laser setup in order to determine the edge strips charge collection efficiency with respect to the strips that are far from the silicon crystal cut and belong to the same detector. By placing the 'edgeless' detector between a telescope planes, the beam particles tracks reconstruction can give us an image of the effectiveness of the strips close to the diced border. A good evaluation will be made by comparing the physical distance between the detectors that are facing their cuts, with the distance determined by tracks reconstruction. References 1.

V.Palmieri & al., Nucl. Instr.& Meth., A413, 475 (1998); K.Borer & al. CERN/LHCC98-27, DRDC P53 Add.l (1998). 2. E.Verbitskaya & al., Nucllnstr.& Meths. A514, 47 (2003). 3. L.Casagrande & al., Nucllnstr.& Meths. A461, (2001). 4. G. Nussle, Cooling tests on dummy modules in the RD39 beam test cryostat, CERN/ST Division, 30* of July 2004. 5. Blanca Perea Solano, Ph.D. Thesis, Cryogenic Silicon Microstrip Detector Modules for LHC, CERN, Universitat Politecnica de Catalunya. ISBN 84-6888952-0, http://cdsweb.cern.ch/search.py?recid=802087. 6. http://www.te.rl.ac.uk/esdg/cms fed pmc/ 7. http://www.hep.ua.ac.be/cms/archive/testing/software/ 8. http://cms.pi.infh.it/roby/moduletest/doc/tscO 1 .pdf 9. http://www.te.rl.ac.uk/med/projects/High_Energy_Physics/CMS/APV25S l/pdf/User_Guide_2.2.pdf 10. T.Niinikoski & al, Nucl. Instr.& Meth., A520, (2004) 87-92. 11. K. Borer & al, RD39 Status Report, CERN/LHCC 2002-004 (2002). 12. J. Harkonen, E. Tuominen, E. Tuovinen, P. Heikkila, V. Ovchinnikov, M. Yli-Koski, L. Palmu, S. Kallijarvi, H. Nikkila, and O. Anttila, Nucl. Instr.& Meth., A 514 (2003) 173-179.

THE DEPFET ACTIVE PIXEL SENSOR AS VERTEX DETECTOR FOR THE ILC H.-G. Moser\ L. Andricek1, P. Fischer3, F. Giesen3 M. Harter3, M. Karagounis2, R. Kohrs2, H. Kriiger2, G. Lutz1, I. Peric3, L. Reuerf, R.H. Richter1, C. Sandow2, L. Struder1, J. Treis1, M. Trimpl2, N. Wermes2, S. Wolfel1 'Semiconductor Laboratory of the Max-Planck-Institut fiir Physik and Max-PlanckInstitutfiir extraterrestrische Physik, Otto-Hahn-Ring 6, D-81739 Munich, Germany 2 Physikalisches Institut der Universitdt Bonn Nussallee 12 D-53115 Bonn, Germany 3 Institut fiir Technische Informatik, B6, 26, D-68131 Mannheim, Germany

For the International Linear Collider a vertex detector with unprecedented performance is needed. The DEPFET, which integrates a MOSFET into the high resistivity detector substrate offers such performance: large signal/noise, small pixel size, thin detectors, low power consumption, high readout speed and radiation tolerance. This paper presents the concept of the DEPFET and results of a complete prototype system with dedicated control and readout electronics. Measurements of the radiation hardness will be presented and the technology to achieve thin detectors (50 urn) will be discussed.

1. Introduction A vertex detector at the international linear collider [1] needs excellent position resolution (5 urn), fast readout (50 MHz) and radiation tolerance up to 100 kRad. The detector should introduce minimal scattering material, 0.1% X0 per layer at most. The DEPFET active pixel sensor offers a detector concept which can fulfill these requirements. In the DEPFET Pixel concept [2], the first amplifying transistor is directly integrated into a high resistivity silicon substrate (Fig. 1). By sideward depletion and an additional n-implantation below the FET, a potential minimum for electrons is created underneath the transistor channel, which can be considered as an internal gate of the FET. The signal electrons created by an impinging particle are collected and stored in the internal gate, which results in a modulation of the transistor current. After readout the stored electrons are removed from the internal gate by a clear pulse. In this concept very good noise performance can be achieved. For ILC studies, prototype DEPFET pixel structures with dimensions of 22x30jum2 and matrix sizes of 64x128 pixels have 1109

1110

been devised. Dedicated steering (SWITCHER II) and readout ASICs (CUROII) have been developed [3] optimized for the timing and speed requirements at an ILC. The matrix and electronics are arranged to allow parallel readout of the 64 columns for the pixels activated in a single row (Fig. 2).

Figure 1. The DEPFET detector and amplification structure is based on a sideward depleted substrate material (a) into which a planar field effect transistor (b) is embedded (a MOS device is shown here). The electric potential is schematically drawn on the right side with the p+ implants set to ground. gate

DEPFET- matrix

Figure 2. DEPFET Matrix: The external gates and clear ("reset") contacts are controlled in rows by SWITCHERs. The drain contacts are connected in columns to the CURO readout chip. This allows reading all pixels of a row in parallel.

1111

2. Characterization of a Prototype in the Lab and Test Beams Noise and clearing studies using single pixel devices and small mini pixel matrices have been performed. The low intrinsic noise of the DEPFET could be demonstrated in spectroscopic measurements of X-rays using special electronics with long shaping times (6 usee). An ENC noise of 2.2 e" at room temperature was measured [4]. To optimize the noise of the entire system under ILC timing conditions, the clearing noise must be minimized. As the readout cycle contains the sample (signal) clear (pedestal) sequence, clearing noise is minimal if the clear process fully empties the internal gate. It can be shown that over a large range of operating parameters a complete clearing can be achieved.

2000

4000

6000

Signal (e)

8000 10000 x10

Figure 3. Signal distribution of 6 GeV electrons traversing a 450 urn thick DEPFET detector. The signal is from clusters defined by a 6 a seed. Typical cluster sizes are 4-6 pixels. The peak value corresponds to 32441 e".

An ILC DEPFET Pixel prototype module with close to ILC specifications has been characterized by lab measurements and in a 6 GeV e" test beam at DESY, Hamburg. All individual components have been shown to operate close to ILC timing requirements. In the test beam the system has been operated without on chip zero suppression and with slower speed, however. The noise of the total system with fast ILC shaping in the CUROII chip has been measured to 225 e". The signal for minimum ionizing particles in the 450um thick detector corresponds to 32500 e" (Fig. 3), resulting in a signal to noise ration of 144/1. Scaling to the proposed thickness of 50um gives S/N = 15/1. The position resolution has been measured to be lOum, however, this value is dominated by the multiple scattering error due to the low energy beam.

1112 3. Thinning Technology In order to reduce multiple scattering the material of the sensors must be minimized. Using a wafer bonding technique it is possible to thin down the active region of a DEPFET matrix to 50um. Tests with diode structures showed that the thinning did not deteriorate the performance of the device [5]. In order to achieve mechanical stiffness and stability a frame with the original thickness can be left over. Including this frame and electronics an effective thickness of 0.1% X0 can be achieved (Fig. 4). It should be stressed that the DEPFET technology allows producing large, self supporting, module size devices avoiding the need for extra mounting frames.

Figure 4. A silicon sample thinned down from 300nm to 50nm using the wafer bonding technique. A perforated frame has been left over to ensure mechanical stiffness.

4. Radiation Hardness The inner layer will be exposed to 100 KRad in 5 years of ILC operation. Like all MOS devices, the DEPFET is inherently susceptible to ionizing radiation. The predominant effect is the shift of the threshold voltage to more negative values due to the build up of positive oxide charges. Irradiations of such devices with hard X-Rays and 60Co Gamma rays up to lMRad were performed under various biasing conditions. It turns out that the threshold shifts are small (~4V) and can be compensated without compromising the performance of the device (Fig- 5). 5. Power Consumption In a DEPFET matrix only the single row which is actually read out dissipates power: about 0.5W per row of 1000 pixels. In addition 6.3mW per active row are dissipated by the switcher and 2.8W by the CURO readout chip. Taking into account the duty cycle of the ILC accelerator, 1 ms beam followed by a pause of 199ms, a five layer detector as proposed in [1] would dissipate 3.1W. Only

1113

0.5 W is dissipated in the active area of the device. This reduces considerably the cooling needs and allows for a low mass detector concept. 6. Conclusions The DEPFET active pixel detector offers large signal/noise, fast signal collection, small pixel size and low power operation. The sensors can be thinned in order to introduce minimal scattering material. They are sufficiently radiation tolerant for ILC conditions. Control and readout ASICs performing close to ILC specifications has been developed. The system has been operated successfully in a test beam. Thus the DEPFET can be considered an ideal candidate for an ILC vertex detector. PX04-2. J I 4 . 60CO. (CSr)

„

6

0 0

200

'09

800

400

>000

1200

Figure 5. Threshold shifts of DEPFET Test structures as function of the dose. "ON": gate voltage of the DEPFET was in on state, "OFF' the gate voltage was off, which represents the ILC operation mode, since the device will be on less then 0.1% of the time.

References 1. TESLA Technical Design Report, ISBN 3-935702-00-0 (March 2001). 2. J. Kemmer, G. Lutz, New semiconductor detector concepts, Nucl. Inst. & Meth. A253 365 (1987). 3. M. Trimpl et al.: A Fast Readout using Switched Current Techniques for a DEPFET Pixel Vertex Detector at TESLA. Vertex 2002 Conference, Hawaii, Nucl. Inst. & Meth. A511 257-264 (2003). 4. M. Porro et al., Spectroscopic Performances of DEPMOS Detector/Amplifier Device with Respect to Different Filtering Techniques and Operating Conditions. Submitted to IEEE TNS, Rome (November 2004). 5. L. Andricek: Processing Ultra-Thin Silicon Sensors for Future e+e- Linear Collider Experiments, IEEE Nuclear Science Symposium (NSS), Portland 19-24. Oct. 2003.

FINAL ASSEMBLY AND INTEGRATION OF THE ATLAS SEMICONDUCTOR TRACKER AND TRANSITION RADIATION TRACKER J.R.PATER The University of Manchester Manchester Ml 3 9PL, United Kingdom On behalf of the ATLAS

Collaboration

ATLAS, a general-purpose detector for the Large Hadron Collider, is currently under construction. At the heart of ATLAS is a three-component Inner Tracking Detector, which includes a silicon strip tracker and a straw-tube transition radiation tracker. A brief description of these two subdetectors will be presented, together with a report on the status of their construction and the status of and plans for their final assembly and integration, which will take place at CERN starting in August 2005.

1. The LHC and ATLAS The Large Hadron Collider, currently under construction at CERN, will collide protons on protons with a centre-of-mass energy of 14 TeV, starting in 2007. ATLAS (A Toroidal LHC Apparatus) is one of two large general-purpose detectors being built to study these proton collisions. ATLAS is a classical layered detector, with an inner tracking detector in the middle, surrounded by electromagnetic and hadronic calorimeters, with a muon tracking system immersed in a toroidal magnetic field at the outside. 2. The Status of the ATLAS Inner Tracker The ATLAS Inner Tracker [1] consists of three subdetectors. The innermost is a pixel detector which will provide three very precise tracking points per track out to a radius of 12 cm in the barrel region and down to a pseudorapidity value of 2.5 in the endcap regions. Outside the pixel tracker is the Semiconductor Tracker, or SCT, which is a silicon-strip tracker which will provide 4 precise tracking points per track out to a radius of about 52 cm in the barrel and also to a pseudorapidity value of 2.5 in the endcaps. Outside the SCT is the Transition Radiation Tracker, or TRT, which is a straw-tube tracker which will provide up

1114

1115

to 40 two-dimensional tracking coordinates, out to a radius of about 1 meter in the barrel region, and 2.7 meters axially on either side of the interaction point. The TRT will also provide electron identification capability through detection of transition radiation photons produced when an electron passes through the radiator foils embedded between the straws. All three of these subdetectors are currently under construction. The TRT and SCT will be integrated together before being lowered into the ATLAS cavern in 2006; the pixel subdetector will be added later. This paper will focus on the current status of construction of the TRT and SCT, and the plans for their assembly and integration at CERN.

2.1. The Transition Radiation Tracker The TRT is being constructed in three basic pieces: the barrel and two identical endcaps, herein referred to as "endcap A" and "endcap C" [2]. The TRT barrel consists of three concentric rings of modules, each ring containing 32 identical modules. These modules are about 1.5m long, but are electrically divided in the middle so that each end is read out separately. Each module consists of many layers of axial straws, which are 4mm-diameter tubes filled with a mixture of 70% Xenon, 27% carbon dioxide and 3% oxygen. The straws are embedded in a radiator foam; the passage of an electron through this radiator produces high-energy transition photons which trip the higher of two thresholds, providing electron identification. The barrel has a total of 52,544 straws, and hence twice that many readout channels. The carbon fibre support structure for the TRT barrel was built in Russia. The barrel modules were built in the United States, subjected to strict quality assurance testing and then sent to CERN. At CERN they were re-tested (and reworked if necessary) then inserted into the space frame. Then the front-end electronics boards were added, and also the gas and cooling manifolds. The construction of the barrel was finished earlier this year, and final testing of the whole system is nearing completion now, including measurement of the intrinsic noise, verification of the performance on a sector-by-sector basis, and studies with cosmic rays. Integration of the TRT barrel with the SCT barrel is scheduled to begin in November of this year. The two TRT endcaps are composed of "wheels", which are layers of radial straws interleaved with radiator foils. Each endcap consists of 12 A-type wheels and 8 B-type wheels (the layers of the A-type wheels being closer together than those of the B-type wheels) for a total of 122,880 straws. The wheels are being

1116 made in Russia (construction will be completed this month); once completed they are sent to CERN for integration. At CERN the wheels are stacked together in a vertical position, then rotated to the correct orientation and the electronics and services added before final testing. At this point in time, endcap C is very nearly completed; all wheels have been stacked and rotated and all the front-end electronics have been installed. Services are being mounted now, and the insertion of the SCT endcap is foreseen for early 2006. Endcap A is in an earlier stage of production as it is scheduled to be installed later than endcap C. At the moment the wheels are being stacked as they arrive from Russia and the front-end electronics are being prepared. Insertion of the SCT is foreseen for mid-2006.

2.2. The Semiconductor Tracker The Semiconductor Tracker, or SCT, is composed of double-sided silicon-strip modules [3], each of which has 768 active strips and 6 ASICs per side. The two sides are rotated by a stereo angle of 40 mRadians with respect to one another, to provide 3-dimensional tracking coordinates. There are 4088 modules in the SCT, for a total of 60 square meters of silicon and about 6 million readout channels. The modules are of four types:

1117 rectangular modules arranged on cylinders in the barrel and three types of wedge-shaped modules arranged in concentric rings in the endcaps. The SCT barrel modules, support structures and services were produced at many different places throughout the world, then shipped to the University of Oxford (UK) for assembly. After assembly and testing [4], the four complete individual barrels were shipped to CERN for integration. The first barrel arrived at CERN during the first week of January of this year; the last one arrived in August. The integration of the four barrels with the thermal enclosure began in July and was completed on schedule on the 20th of September. Figure 2 shows the insertion of the second-smallest barrel into the assembly of the two outer barrels and the outer thermal enclosure. Some mechanical work, for example the addition of the ends of the thermal enclosure and the completion of gas-tight seals around the services, remains to be done, then in November the SCT barrel will be inserted into the TRT barrel.

The SCT endcaps, like the barrel, are international projects, with the various components being produced at many different places. The assembly sites for the two endcaps are the University of Liverpool in the United Kingdom and the NIKHEF laboratory in the Netherlands, where modules are mounted onto disks and completed disks are inserted into the endcap support cylinders. When finished, the completed endcaps will be shipped to CERN, where, after

1118 testing and the addition of the thermal enclosures, they will be inserted into the TRT endcaps. At the time of this conference, endcap C is nearing completion. The most time-consuming part, namely the completion and testing of the individual disks, is finished. Three of the disks have been mounted into the support cylinder and in-cylinder testing has begun. The endcap is scheduled for delivery to CERN in December 2005. Endcap A is also well underway, with five disks finished and tested. Three of the finished disks have been mounted in the cylinder although their external services must be added before final testing can start. Delivery of endcap A to CERN is foreseen for March 2006. Once an endcap is delivered to CERN, it will be tested to verify that it suffered no damage in transit, then it will be transferred from its test box to a "cantilever stand" which supports it from one end so that its thermal enclosures can be slipped onto it. Once the inner and outer thermal enclosures are fitted, they are tested for leak-tightness and general functionality. 3. Plans for SCT-TRT Integration The procedure for integrating the SCT and TRT is basically the same for barrel and endcaps. The SCT, still on its cantilever stand, is aligned to a set of rails which have been fixed to the floor of the working area, as shown in figure 3. The TRT meanwhile has been mounted on a special trolley, also shown in figure 3; this trolley is rolled over the SCT on the rails, maintaining alignment using a frame fixed to the rails and a system of alignment wires.

Figure 3. Conceptual drawing of a TRT endcap mounted on its trolley, being rolled over an SCT endcap (on its cantilever stand) along the alignment rails.

1119 Once integrated, the combined TRT/SCT systems (barrel and each endcap) will be tested thoroughly before being lowered into the ATLAS cavern. First the readout of the SCT and TRT will be checked separately, then, running together, measurements will be made of intrinsic noise with special attention given to looking for signs of crosstalk between the systems. Long-term operational stability will be monitored and cosmic ray events will be recorded. Of course, in addition to testing the detector itself, this will provide invaluable tests of the data acquisition systems and the control and monitoring systems. These combined tests are foreseen to take place in early 2006 for the barrel and mid-2006 for the endcaps. 4. Summary and Conclusions The ATLAS Transition Radiation Tracker and Semiconductor Tracker are now nearing completion, bringing to a close many years of design and construction work by hundreds of physicists, engineers and technicians. Final integration of the TRT and SCT detectors will take place in the near future, and the finished barrel and endcaps will be lowered into the ATLAS cavern during 2006, to await the first LHC collisions scheduled for 2007.

References 1. The ATLAS Collaboration, "The ATLAS Inner Detector Technical Design Report", CERN/LHCC/97-16 and CERN/LHCC/97-17, April 1997. 2. F.Dittus., "ATLAS TRT Production Status", presentation given at the 3 rd Workshop for Advanced Transition Radiation Detectors for Accelerator and Space Applications; Ostuni, Italy, 7-10 September 2005. 3. T.Kondo et al., "Construction and Performance of the ATLAS Silicon Microstrip Barrel Modules", Nucl. Instrum. Meth. A485 (2002): 27-42. 4. P.W.Phillips, "Functional Testing of the ATLAS SCT Barrels", 7th International Conference on Large Scale Applications and Radiation Hardness of Semiconductor Detectors; Florence, Italy, 5-7 October 2005.

3-DIMENSIONAL POSITION CONTROL FOR THE AMS-02 TRACKER WITH INFRARED LASER BEAMS WOLFGANG WALLRAFF' 1. Physikalisches

Institut lb, RWTH Aachen, Sommerfeldstrasse Aachen, Germany

14

The 6 m , 8 plane Si tracker of the AMS-02 experiment can record charged particles from the cosmic radiation with high precision. The single hit resolution transverse (parallel) to the magnetic field is a y 8.5 pm (a, 30 p-m). AMS will measure with unprecedented resolution cosmic particles in a near earth orbit. The AMS-tracker is equipped with a position control system based on pulsed infrared Laser beams, that are directly seen by the Si sensors of the tracker, allowing to detect lateral displacements of less than 5 |xm. The diffraction of the Laser light at the read out structure of the Si sensors does generate clearly visible patterns. It will be demonstrated that changes in the relative separation of these diffraction fringes can be used to determine the distance between layers (depth z) of the AMS tracker with a precision of 100 pm.

1. AMS Tracking 1.1. Si Tracking in Space The AMS-01 Silicon Tracker [1] was the first application in space of the high precision silicon technology developed for position measurements in accelerator experiments [2]. The high modularity, low voltage levels (< 100 V) and gas-free operation of the device are well suited to operation in space. With the AMS-02 silicon tracker, charged particle tracks are traced at 8 space points in an approximately 1 m3 sized B-field to an accuracy of better than 10 urn. In the 0.9 T field provided by the superconducting magnet of AMS-02 protons of 30 GeV do show a sagitta of approximately 1 mm. Space based particle detection systems have to cope with a far wider range of environmental conditions than those at accelerators. This concerns notably the vibrations during the transport before deployment and the rapid periodic changes in the thermal settings due to solar radiation and cooling while in the shadow of Earth. f

This work is supported by DLR, Koln-Porz, Germany.

1120

1121

1.2. Tracker Alignment Control The alignment system, developed by RWTH-Aachen, provides optically generated signals in the 8 layers of the silicon tracker that mimic straight (infinite rigidity) tracks. It has been shown with AMS-01 [3, 4] that these artificial straight tracks allow the tracing of changes of the tracker geometry with a position (angular) accuracy of better than 5 um (2 ^irad). With AMS-01 it was found that the carbon fiber tracker support structure is stable at the 15 \im level but that excursions up to 30 \im did occur. These excursions were correlated with changes in the thermal conditions following changes in spacecraft attitude [3, 4]. For long observation periods, the overall system stability is especially important in that it limits the ultimate momentum resolution for high rigidity particles and can introduce dominating systematic errors in the pointing accuracy of the AMS tracker for converted photons, particularly from astrophysical point sources. Sub arc-minute precision pointing of weak sources is possible provided sufficiently frequent checks of the tracker geometry can be performed with laser beams and stiff cosmic tracks without going to a zero field (thereby spending superfluid He magnet coolant). a ) AlvlS-02 Tracker Afignmenl System

AMS-02 Si-tracker S 1 its.i!t beams Tor alignment control

b)

basic coittponenit,

LFCR # 1 . . . M-Ctate

i,»&€*r i-1fcrs Coupler 6

"LfSL.. ; •

-

1 * I 3 ' B teoaomoJe Ma M I

/ U , * * '' v,

LFCR #5

'-

„

n

LBBX#1.S LDDR *S

.-—-"'

-* •*

top, rale < 2OO0HI, \ amplitude <.20fln»A, w l r f l h * 5000 m

| c)

^

AMS-02 ~ - - - e r v r

Si laddei' instated 6n CFs •

';

,

. i !.. i

£ •

§ * ' ' » / ***

.

i

1

-

•9

H* -_l.

<•••

li'-ri*"'*.

(1

;1 i

1,0

f'

,-.,,.. n.

1

< • ' - *

—

•"""*

f' *

LBBXfKJ L.la CM I »

1

10 ^ " V IUM

liij,."...is'-.",.ciK^ ..T;.I

M l . - M N \l> i n I v ; < mil -I iii'.'i ^vbi

Figure 1. AMS-02 silicon tracker laser alignment system overview (a). Geometry of upward and downward going laser beams (b). Microphotograph of an Anti Reflective area (c)

1122 2. AMS Alignment Tests The system uses the same silicon sensors for both particle detection and control of the alignment. It serves to generate position control data within seconds at regular time intervals (4 to 6 times per orbit), for example, while the ISS flies into the shadow of the Earth or comes back into the sunlight.

2.1. AMS Alignment System Layout As shown in Figure 1, the AMS-02 tracker is equipped with 2 x 10 pairs of alignment control beams. The beams are narrow (diameter < 0.5 mm) and of small divergence (< 1 mrad). The beams enter the tracker volume through 2 x 5 beamport boxes (LBBX) mounted on the outer faces of the two outer tracker support plates. The photons of these beams are generated with laser diodes mounted outside of the tracker volume and are brought practically loss fiee to the LBBX via mono-mode optical fibers. The pulse length is less than the electronics integration time so that the repetition rate is limited only by the readout. The wave length of these beams, 1082 nm, has been chosen such as to penetrate all 8 Si detector layers of the tracker at once. At this wavelength only a small fraction of the generated photons is absorbed (10% in the 300 jim-thick Si sensors). The reflection at the Si surface however has to be suppressed in order to overcome the intensity limitations in recording the alignment beams due to the strong effective attenuation (factor of-10 per Si layer) caused by the high refractive index (dielectric constant) of Si. Furthermore the transparency of the Si particle detector surfaces is obstructed by the aluminized readout strips. In consequence, the tracker sensors on the alignment beams have been equipped with antireflective coatings (Si0 2 and Si3N4) optimized for the wavelength chosen (residual reflectivity ~ 1%). In addition, the readout strip metallization width was reduced to 10 ^m width in the coated areas and the other implants not metallized. Together these measures have resulted in a transparency of the alignment sensors of 50% [5] and the 8l layer of the tracker receives about 0.8% of the intensity coming out of the LBBX. With high quality DBR diodes the very high electro-optical efficiency allows for comfortably large photon fluxes even during small optical power (100 nJ/pulse) operation. This results in signals induced by the laser that can exceed those of Z = 26 nuclei in the tracker planes closest to the LBBX.

1123 The Laser sources used with AMS-02 provide coherent infrared radiation. The regularly spaced metallized readout strips -while illuminated with the coherent alignment beams- produce a diffraction pattern similar to that one produced by a grating. This diffraction pattern is most clearly observed on the 2nd plane (no. 2 or 7) in the beams (Fig. 2). The spacing of the typical intensity maxima and minima can serve to measure the distance in depth (z) between planes 1 and 2 (8 and 7). At the layers deeper inside the detector, the diffraction pattern is much more complicated. An analysis of these pattern has not been tried so far. At the alignment beam wave length of 1082 nm and a y-strip spacing of 110 urn on the Si sensors fringes are expected at equal spacing in angle (± 9.8, ± 19.6, ± 29.4 mrad). las avg (2350-2399) ped corr si L12AJ054

J00

125

150

175

200

225

s t r i p no. (pitch : 110 \m) Figure 2. Plane 2, p-side (y) AMS-02 alignment beams and diffraction patterns (insert: fit).

2.2. AMS Alignment Control Results At AMS-02 the 2 outer planes 1 and 2 (8 and 7) are 20 cm apart. Hence the crests as well as the peaks of the interference pattern will appear at approximately 2 mm spacing (dpk). The angular geometry of this interference pattern is very stable, for it is determined by the laser wavelength stability (DBR type: precise to < 0.2 nm) and the geometry changes of the wafers in the alignment beams during flight. These will be small (< 50 nm) in spite of the large temperature range foreseen for AMS-02 tracker operation, because only a small region (< 10 strips) is illuminated by the alignment beams.

1124

The wafer to wafer fluctuations in the geometry during fabrication of the Al strips contacting the implants in the Si detector (a single set of masks was used for the alignment sensors) are small also. From microphotographs this accuracy has been estimated as being better than 0.5 (im. Lateral displacements have for sure to be taken into account while evaluating the depth measurements. As shown in Figure 2, the laser beam spot covers 10 (5) strips on the p- (n) side of the sensors. Position changes are determined concurrently for both coordinates from changes of the measured centroids of the laser profiles. Alignment beams are arranged in pairs in order to distinguish between changes in beam geometry and sensor displacements. The peak positions from this figure also show that, with only 50 laser pulses, accuracies below one um perpendicular to the laser beam can be achieved. Both beams exhibit interference fringes. The spacing of the extrema of these fringes (dpk) is evaluated by fitting. Due to the smallish interference signal the accuracy of the measured spacings is approximately 4(2) \x,m in this test (AMS-02). This accuracy is however sufficient for determining changes in the distance between planes (Fig.3) at the level of 2/5 (1/10; 1/40) mm for this test (AMS-02) with 200 (800; > 10") events.

depth sensitivity (at 100 ( 200 with AMS-02 } mm plane spacing )

I

* - „ t e s t s l o p e : 9.3 * 0.5 p m / mm

1«0

m,

• » &

tl

1000

s«

1$

980

Jj $»

<>fi0

fti

*" ^ v

I

940

111

1'_

3L

1

»20

"o~~~

t"~~~~"~>"""

~T

t z054lmm)

10

12

14

ladder displacement i n depth

Figure 3. Peak to peak (crest to crest) distances in the diffraction pattern (dpk) vs. ladder displacement in depth (z054); for details see text.

1125 Laser alignment will be performed coincident with data taking. This allows any possible changes in the tracker geometry, from rapid thermal deformations to long term drift, to be identified and corrected offline. 3.

Conclusions

The AMS approach to silicon tracker alignment control using IR laser beams fulfills the requirements of a space bome experiment. It is light weight (3kg), low power (lmW), low dead time (< 1%) and provides a precision exceeding the tracker resolution (8 ^.m) with less than 100 laser shots. From the observation of read out strip generated interference patterns the absolute distance between tracker planes can be determined with an accuracy exceeding that achieved in assembly. The success of the AMS approach in Si tracker alignment control by IR laser beams has lead the team building the largest Si tracker array [6] to develop a similar system for 10 years of operation at the LHC. Acknowledgments I like to thank for the support by the AMS tracker group and its leader Prof. R. Battiston. I am very grateful for the help I got from the Geneva university tracker group and especially to Dr. D. Haas, who has run the AMS tracker DAQ for the results presented here. And last but not least "mille grazie" to the organizers of this meeting for their interesting choice of subjects. References [1] R. Battiston Nucl.Phys.Proc.Suppl. 44, 274 (1995). W.J. Burger et al., Nucl. Instr. andMeth. A 512 , 517 (2003). Ph. Azarello, AMS-02 tracker report, this meeting. [2] M. Acciarri et al., Nucl. Instr. andMeth. A 351, 300 (1994). G.F. Dalla Betta et al., Nucl. Instr. andMeth. A 431, 83 (1999). [3] J. Vandenhirtz et al., Proc. 27th International Cosmic Ray Conference (ICRC2001), D-Hamburg, session OG, 5, 2197 (2001). [4] W. Wallraff et al., Proc. 71 International Conference on Advanced Technology and Particle Physics (ICATPP-7), I-Villa Olmo Como, M.Barone et al. (Eds.), World Scientific, Singapore, ISBN 981-238-180-5, 149 (2002). [5] V. Vetterle, Silizium Alignment Sensoren fur den AMS-02 Tracker auf der ISS, 2001, Diplomarbeit, RWTH-Aachen. [6] A. Ostaptchouk et al., 9 May 2001, CMS Note 2001/053, CMS - CERN.

D E V E L O P M E N T OF A G E M - B A S E D HIGH RESOLUTION T P C FOR T H E I N T E R N A T I O N A L LINEAR COLLIDER

M. KILLENBERG, S. LOTZE, A. MUNNICH, S. ROTH, M. WEBER* III. Phys. Inst. RWTH Aachen,

Germany

J. MNICH DESY, Hamburg,

Germany

A promising candidate for the main tracker at the international linear collider (ILC) detector is a high resolution time projection chamber (TPC) with gas amplification based on micro pattern gas detectors. This paper presents the R&D work for such a TPC using gas electron multipliers x (GEM) for gas amplification. To study the spatial resolution of a GEM based TPC, a prototype with low mass fieldcage as well as a high resolution hodoscope as external reference was constructed. The development of a new and fast readout electronics for this prototype is ongoing. The studies with the prototype are assisted by the development of a simulation framework for a GEM TPC.

1. Construction of a T P C prototype The goal was to design a flexible TPC prototype with GEM readout to perform various tests and measurements. Its diameter of 26 cm allows opperation in a 5 T magnet with a bore of 28 cm located at DESY Hamburg. It has a drift length of 26 cm and a maximum drift field of 1000 V/cm which allows the study of various drift gases. Special attention was paid to building the field cage with as little material as possible. Using a composite structure (Fig. 1 left) a material budget of ~ 1 % of a radiation length has been achieved (Fig. 1 right). The design goal in the TESLA-TDR 2 of 3% of a radiation length for the full size TPC, including inner and outer fieldcage, seems achievable with a similar design. On the left of of figure 2 a photograph of the prototype with removed cathode is shown. Inside the fieldcage, the uppermost GEM of the ampli* Corresponding author: III. Phys. Inst. RWTH Aachen, D-52056 Aachen, Germany E-mail: [email protected]

1126

1127 Al Foil. 0.05 mm

Figure 2. Prototype and event display of a track.

fication structure consisting of three GEMs can be seen. A shield at the z-position of this GEM prevents field inhomogeneities at the edges of the structure. On the right hand side an event from an electron testbeam is depicted, using an event display illustrating the sensitive area in the center of the chamber in the upper view. In the bottom view the individual charge deposition on the pad plane with 1.27 x 6.985 cm 2 pads is shown.

1128 2. Development of fast readout electronics for the prototype Until now a readout designed for the ALEPH TPC has been used. It is optimized for the slow induced signals of an amplification structure using anode wires. To take full advantage of the fast electron signals of a GEM amplification structure new electronics is being developed. As a first step the Preshape 32 3 (Fig. 3 left) was chosen as a preamplifier. It is an integrated, charge sensitive amplifier/shaper with 32 channels and a nominal peaking time of 45 ns. To transfer its single ended output signals to a reasonable distance, a cable driver card (Fig. 3 right) with bipolar output has been constructed.

Figure 3. Preshape 32 preamplifier and its cable driver

This preamplifier is currently in use together with the ALEPH ADCs, and a reconstructed event can be seen on the right side of figure 2. The next step in the electronics development will be the choice of a new ADC. The minimum requirements are 10 bit resolution and a sampling rate of 40 MHz. Several prototypes are being tested to determine the best system. 3. Operation of the T P C in a hodoscope In order to perform precise studies of the TPC properties, such as the homogeneity of the drift field, a hodoscope has been built. It gives an external reference to each track measured in the TPC. Four silicon strip modules, two above and two below the TPC, are used to obtain two indepent reference points of the particle track (Fig. 4 left). Each pair of modules is

1129 arranged with a stereo angle of 5 degrees to allow a two dimensional measurement. The modules consist of two 500 fj,m thin sensors with 768 strips and a pitch of 122 //m. The hodoscope reaches a resolution of 58 /um in x and 624 /j,m in z. After calibrating the system, consisting of the TPC and the hodoscope, to account for mechanical offsets and rotations, comparisons between the two measurements can be made. The right plot of figure 4 shows the single point resolution of the TPC in x-direction, denned as the distance of a reconstructed point in the TPC to the corresponding track measured with the hodoscope. pwlklc hack

7

Figure 4. Schematic setup of hodoscope and measurement of the single point resolution of the TPC.

4. Simulation of a T P C with GEM readout The goal of this simulation framework is to obtain a better understanding of the influence of electric and magnetic fields on the performance of a GEM TPC. The simulation should be straightforward, fast, and therefore independend of big simulation packages. To allow flexibility and minimize data overhead, the simulation is divided into four modules corresponding to the actual processes in a TPC: The first module creates primary ionisation along the particle track. The second drifts the electrons through the chosen gas to the amplification structure, including the diffusion process. In a third step, the electrons

1130 are transferred through a triple GEM structure where they are amplified. This module uses information about the charge transfer through the GEMs from measurements performed by our group 4 ' 5 . The last module accounts for the electronics effects like shaping and digitisation. The first task of the analysis is to compare the results from the simulation to the measurement. Knowing the entry points and directions of the particles from the hodoscope, identical events as measured in the TPC are simulated and then reconstructed with the same software. The right plot of figure 4 shows this comparison for the single point resolution in x-direction in dependence on the drift distance z. The resolution of the simulation is slightly better due to the fact that it does not include multiple scattering and spatial separation of ^-electrons from the track. However, the agreement between measurement and simulation is good, and both curves show the same degration of the resolution due to diffusion. 5. Conclusion &: outlook A TPC prototype with a GEM readout has been constructed and is ready for measurements. Lightweight construction of a fieldcage for such a TPC has been tested. The Preshape 32 preamplifier as first part of a new readout electronics for this prototype is already being used, while new ADCs still have to be chosen. A hodoscope consisting of silicon strip modules has been constructed to act as an external reference for tracks in the TPC. The measurements with the prototype are aided by detailed simulations of the processes in a TPC. Currently measurements in an electron testbeam at DESY are carried out with the prototype and the hodoscope. We plan to measure the field inhomogenieties as well as the spatial resolution. The measurements will help to verify the simulation, which will then be used for systematic studies for a large TPC at the ILC. References 1. 2. 3. 4. 5.

F. Sauli, Nucl. Instr. and Meth. A 286 (1997) 531 TESLA Technical Design Report, DESY-01-011, ECFA 2001-209 (2001) Loukas, D. et al, CERN-DRDC-94-39; DRDC-RD-20, CERN, 1994 M. Killenberg et al., Nucl. Instr. Meth. A498 (2003) 369 M. Killenberg et al., Nucl. Instr. Meth. A530 (2004) 251-257

STAR INNER A N D F O R W A R D TRACKING U P G R A D E

G E R R I T VAN N I E U W E N H U I Z E N , F O R T H E S T A R C O L L A B O R A T I O N MIT Group National Laboratory Bldg. 555 P.O. Box 5000 Upton, NY 11973-5000 United States of America E-mail: [email protected]

Brookhaven

The STAR experiment at the Relativistic Heavy Ion Collider (RHIC) is planning a major tracking upgrade to prepare for the expected higher luminosity running of RHIC. This tracking upgrade will also enhance physics capabilities such as W physics with polarized proton beams and heavy flavor physics with heavy ion and polarized proton beams. The proposed upgrade will consist of a central barrel part and a forward disk part. Both barrel and disk parts will make extensive use of silicon strip sensors and Gas Electron Multiplier (GEM) detectors. In addition active pixel sensors will provide micro-vertexing close to the interaction point. The design of these new STAR detectors will be discussed.

1. Introduction To meet with the increasing physics demands and to cope with the higher luminosities of an upgraded Relativistic Heavy Ion Collider (RHIC), STAR is planning for a tracking upgrade. A comprehensive description of the existing detectors of the STAR experiment and of RHIC has already been published 1 . Both the heavy ion program and the polarized proton program will profit in many ways of such an upgrade. After discovering that a high density, highly interacting medium has been created in Au+Au collisions at 200 GeV 2 , the focus will now be on characterizing this medium with an extended physics program. One of the components of this program is to study heavy flavor production in detail. The proposed Heavy Flavor Tracker (HFT) is a microvertex device that will make it possible to track C and B with high accuracy. Spectra, elliptic flow and heavy flavor energy loss in the medium will make it possible to have an exciting look at the 1131

1132 first stages of the collision. Unfortunately the existing STAR Silicon Vertex Tracker (SVT) will suffer from major aging problems and mechanical incompatibilities when the HFT gets installed. Since the STAR Time Projection Chamber (TPC) doesn't have sufficient backpointing resolution to the HFT, a new Inner STAR Tracker is being proposed. The polarized proton program tries to resolve the question of the gluon contribution to the proton spin 3 . A major part is to concentrate on W production through the W —> e + v channel. To distinguish between e + and e~ accurate tracking is needed in forward rapidity. A forward silicon detector and GEM chambers will facilitate tracking close to the interation point and improve the tracking through the STAR TPC. In the following sections all the new components of the tracking upgrade will be described in detail.

2. B e a m p i p e

Figure 1. Small diameter Be beampipe with exoskeleton (left) and Heavy Flavor •with insertion structure (right).

Tracker

The new Be beampipe is shown on the left side in figure 1. It will have a diameter of 29 mm and a wall thickness of 0.5 mm. The diameter is a tradeoff between the wish to mount the first active pixel layer as close to the interaction point as possible and the expected beam envelope. To reduce scattering and secondaries the wall has been designed to be as thin as feasible while still insuring vacuum integrity. However, to prevent sagging, an exoskeleton is neccesary. The active pixel layers will be located inside of the exoskeleton.

1133

3. Heavy Flavor Tracker The right side of figure 1 shows the Heavy Flavor Tracker (HFT). On the left side the active pixel layers are visible. Also shown is the kinematic mount which insures repeatable detector insertion and positioning. The layers consist of 6 ladders at a radius of 15 mm and 18 ladders at 50 mm. The 20 cm long ladders are constructed out of 19.2 mm x 19.2 mm active pixel sensors with 640 x 640 pixels of 30 (im x 30 fim. The sensors will be thinned to 50 /jm. In total there will be about 100 million pixels. This device will have a pointing resolution of better than 10 /mi 4 . 4. Inner STAR Tracker

Figure 2.

The 3 layers of the Inner STAR

Tracker

The three layers of the Inner STAR Tracker (1ST) are shown in figure 2. The layers are located at radii of 70, 120 and 170 mm. They are designed to cover - 1 < r\ < + 1 and are able to accomodate interactions from -10 cm to +10 cm around the nominal interaction point. The active elements are silicon strip sensors with a size of 40 mm x 40 mm and a strip pitch of 60 fim. Each layer has sensors arranged in stereo pairs with a stereo angle of 5 degrees. There are 1236 sensors with about 800,000 channels. The channels will be read out with APV25-S1 chips 5 . The system will be kept at room temperature with underpressure water cooling. 5. The two G E M Tracker options To improve the tracking through the STAR Time Projection Chamber (TPC) for 1 < T) < 2 two options are being pursued. Figure 3 shows the

1134

Figure 3.

Inner GEM Tracker barrels (left) and Forward GEM Tracker

(right).

two options. The Forward GEM Tracker (FGT) would be located behind the endcap of the TPC. However, due to the large amount of material in the TPC endcap this could lead to problems with rescattering and conversions. If simulations show that this is the case then the Inner GEM Tracker, located in between the 1ST and the TPC, could be a more viable solution. Both options use the same triple GEM technology6 pioneered by the COMPASS experiment 7 . The sensors would all have an active area of about 10 cm x 10 cm and would be utilizing the APV25-S1 readout chip. 6. Forward STAR Tracker The Forward STAR Tracker, shown in figure 4, will enable high resolution tracking for 1 < rj < 2. The four annular disks will be around the beamline at 280, 310, 340 and 370 mm from the nominal interaction point. Each disks will have 21, wedge shaped, silicon strip stereo pairs, with 640 radial strips per sensor. The total of about 100,000 channels will be read out by APV25-S1 chips. The same water cooling as for the 1ST will keep the

1135

Figure 4.

The four disks of the Forward STAR

Tracker

(FST).

system at room t e m p e r a t u r e . 7. C o n c l u d i n g r e m a r k s Currently a large simulation effort is in progress to determine the final configuration of the tracking upgrade. A full proposal should be ready in 2006. Installation is expected to s t a r t in 2009. References 1. The Relativistic Heavy Ion Collider Project: RHIC and its Detectors, Nucl. Instr. and Meth. A Volume 499, Issues 2-3, 1 March 2003, Pages 235-880. 2. First Three Years of Operation of RHIC, Nuclear Physics A Volume 757, Issues 1-2 , 8 August 2005, Pages 184-283. 3. J. Ashman et al., Phys. Lett. B206, 364 (1988). 4. K. Schweda, Proceedings of the Quark Matter 2005 Conference, Budapest, Hungary, to appear in Nucl. Phys. A, arXiv: nucl-ex/0510003. 5. L.L. Jones, Fifth Workshop on Electronics for LHC Experiments. University of Wisconsin/Cern LHCC 232 - 236 USA/CERN 1999. ISBN 92-9083-147-2 ISSN 0007-8328 Fifth Workshop on Electronics for LHC Experiments. Snowmass, Colorado, September 20-24, http://www.hep.wisc.edu/LEB99/ 6. F. Sauli, Nucl.Instr. and Meth. A386 531-534, 1997. 7. B. Ketzer, Nucl. Instr. and Meth. A494, 142 2002.

This page is intentionally left blank

LIST OF PARTICIPANTS

ABBANEO AHMED ALFASSI AMAKO AMBROSI ANDREOTTI ANDREOTTI

Duccio Syed Naeem Zeev B. Katsuya Giovanni Mirco Erica

ANDRIAMONJE ANGARANO ASAI ASSOUAK

Samuel Matteo M. Makoto Samia

AZZARELLO BACCARO BAECHLER BAGBY BARCONS

Philipp Stefania Joachim Linda Xavier

BARONE

Michele

BARTESAGHI

Giacomo

BASILI

Alessandro

BASTIA

Paolo

BATIGNE BECKER BELLINI

Guillaume Julia Fabio

BENEKOS

Nektarios

BERGHOEFER

Thomas

BIANCO

Stefano

BIANCO BIMBOT

Michele Stephane

BINETRUY

Pierre

CERN Queen's University Ben Gurpin University KEK INFN - Sezione di Perugia INFN - Ferrara Universit dell'Insubria INFN Gran Sasso CEA Saclay CAEN SpA SLAC Universit Catholique de Louvain Belgium INFN - Sezione di Perugia ENEA CERN Fermilab Instituto de Fisica de Cantabria Institute of Nuclear Physics - NCSR " Demokritos" Universit dell'Insubria, sede di Como INFN and University of Roma Tor Vergata Alenia Spazio SpA Laben INFN Sezione di Torino Universitat Dortmund INFN and Univ. Di Roma La Sapienza Max-Planck-Intitute fuer Physik Univ. Karlsruhe, Institut fuer Experimentelle Kernphysik INFN - Laboratori Nazionale di Frascati INFN - Lecce Laboratoire Leprince-Ringuet APC

1137

Switzerland Canada Israel Japan Italy Italy Italy France Italy USA Belgium Italy Italy Switzerland USA Spain Greece

Italy Italy Italy Italy Germany Italy Germany Germany

Italy Italy France France

1138 BINKO

Pavel

BISOGNI BLOCH BLOCH BOBIK

Maria Giuseppina Philippe Christoph Pavol

BOCCI

Alessio

BOMBEN

Marco

BONARDI

Mauro L.

BONESINI

Maurizio

BORCHI BORNHEIM

Emilio Adol

BOSCHINI BRACCINI BRAHME BRAUER BRIGIDA BRYMAN

Matteo Saverio Anders Richard Monica Douglas

CABRERA

Juan

CALF CAMPI CAPELLI

Ivan Amos Domenico Silvia

CAPONE CAPPELLA

Antonio Fabio

CASADO CASIERI CECCHINI CECILIA CERULLI CHEFDEVILLE CHERRY

M.Pilar Augusto Stefano Angelica Riccardo Maximilien Michael

CHRIST

Tassilo

Observatoire de Geneve and SYnSpace SA INFN and Universit di Pisa CERN CERN Institute of Experimental Physics SAS INFN - Laboratori Nazionale di Frascati INFN and Trieste University INFN Milano and Universit di Milano INFN Milano and Universit di Milano Bicocca Universit di Firenze CALTECH/CMS ECAL Collaboration CILEA TERA Foundation Karolinska Institutet Physikalisches Institut B INFN - Bari University of British Columbia Center for Space Radiations INFN - Bari CERN INFN Milano and Universit di Milano Bicocca INFN - Roma 1 Univ. Di Roma La Sapienza UAB and IFAE Edslan INFN - Bologna ENEA INFN - LNGS NIKHEF Louisiana State University

Switzerland

Technische Universitaet Muenchen

Germany

Italy Switzerland Switzerland Slovak Republic Italy Italy Italy Italy

Italy USA Italy Switzerland Sweden Germany Italy Canada Belgium Italy Switzerland Italy

Italy Italy Spain Italy Italy Italy Italy The Netherlands USA

1139 CHRISTIANS

Ludwig

CIAVOLA CISBANI CONVERSE COSENTINO CROSETTO CUNNINGHAM D'ANGELO D'ESPOSITO DA VIA' DAWSON DESIO DI CIACCIO

Claudio Evaristo Alexander Dario Dario Vincent Pasquale Massimo Cinzia Jaime Antonio Anna

DI LUISE DI SARCINA DI SIMONE

Silvestro Ilaria Andrea

DICKENS DINARDO DRAGOUN DUSINI ELLIS EMEZUE

Jeremy Mauro Otokar Stefano Malcom Henry

ENGSIG EPPARD ESPOSITO EVANGELOU FABJAN FARGION FERNANDEZ

Bjorn Michael Luigi Salvatore Ioannis Christian Daniele Juan Pablo

FERRER RIBAS FINK

Esther Johannes

FOGGETTA

Luca

FOSTER FRENCH FREY

Brian Marcus Martin

Iseg Spezialelektronik GmbH ENEA INFN - Roma 1 University of Wisconsin Avocent 3D-Computing, Inc. GlaxoSmithKline INFN - Milano Avocent Brunei University University of Sussex Universit di Firenze University of Roma Tor Vergata Univ. "RomaTre" ENEA CERN PH/ATC and INFN-CNAF University of Cambridge INFN - Milano Nuclear Physics Institute INFN - Padova Imperial College London National Academy of Science Oracle Corporation CERN INFN - Bologna University of Ioannina CERN INFN - Roma 1 CIEMAT/Fisica Experimental de Particulas CEA Saclay Bonn University Institute of Physics Consorzio Interuniversitario per la Fisica Spaziale University of Oxford CCLRC Univ. Karlsruhe, Institut fuer Experimentelle Kernphysik

Germany Italy Italy USA Italy USA U.K. Italy Italy U.K. U.K. Italy Italy Italy Italy France U.K. Italy Czech Republic Italy U.K. Nigeria Denmark Switzerland Italy Greece Switzerland Italy Spain

France Germany Italy

U.K. U.K. Germany

1140 FUOCHI GADDI GAGLIARDI GALLAS TORREIRA GAMBARINI GAMBICORTI GAN

Piergiorgio Andrea Guido A. A.

ISOF-CNR CERN INFN Genova and CERN INFN-Bari

Italy Switzerland Italy Switzerland

Grazia Lisa K.K.

INFN - Milano Universit di Firenze The Ohio State University

Italy Italy USA

GARABATOS GARDNER GARGANO GERVASI

Chilo Robert Fabio Massimo

Germany USA Italy Italy

GIANI GIORDANO GIUBILATO GONCALVES

Simone Francesco Piero Patricia

GORISEK GRANDI

Andrej Davide

GRICHINE

Vladimir

GSI University of Chicago INFN - Bari INFN Milano and Universit di Milano Bicocca CERN INFN - Bari INFN - Padova LIP - Laboratorio de Instrumentacao e CERN INFN Milano and Universit di Milano Bicocca CERN / Lebedev Institute

GRIECO GROLL GROPPI

Giovanni Marius Flavia

Italy Germany Italy

GROTELUESCHEN GULMEZ HAGOPIAN HAGOPIAN HAJDAS HAREYAMA HASAN

Frank Erhan Vasken Sharon Wojtek Makoto Said

HATTENBACH HAUNGS

Jan Andreas

HEIJNE HUDSON IEVA IPPOLITOV

Erik Danya Michela Mikhail

CAEN SpA University of Hamburg INFN Milano and Universit di Milano German National Radio Bogazici University Florida State University Florida State University Paul Scherrer Institut Waseda University Universit dell'Insubria, sede di Como Physikalisches Institut B Univ. Karlsruhe, Institut fuer Experimentelle Kernphysik CERN ONERA INFN - Bari Russia Research Center "Kurchatov Inst.:

Switzerland Italy Italy Portugal Switzerland Italy

Switzerland

Germany Turkey USA USA Switzerland Japan Italy Germany Germany

Switzerland France Italy Russia

1141 ISAKSEN ISRAEL IURLARO IVANTCHENKO KAGAN KAMINSKI

Lasse Martin Giorgia Vladimir Harris Jochen

KHOMICH KRAMMER

Andrei Manfred

KRIZAN KROKHOTIN

Peter Andrey

KUDENKO

Yury

LAGOMARSINO

Stefano

LANG

Karol

LAZANU LAZARO

Ionel Delphine

LECOQ LEHNER LEIBENGUTH

Paul Frank Guillaume

LENZI LEROY LIZARAZO

Michela Claude Juan

LOPARCO LUCARELLI

Francesco Fabrizio

LUNGESCU MAGLIOZZI MALTEZOS

Andrea Maria Lucia Stavros

MALVEZZI

Valeria

MANGONI

Roberto

MANOLOPOULOS MARCELLI

Spyros Laura

CERN Washington University ENEA CERN Ohio State University Univ. Karlsruhe, Institut fuer Experimentelle Kernphysik University of Mannheim Institute for High Energy Physics University of Ljubljana Institute for Theoretical and Experimental Physics Institute for Nuclear Research INFN and Universit di Firenze University of Texas at Austin University of Bucharest Inserm U678 - SIEMENS Medical Solution CERN University of Zurich Institute for particle physics, ETH Zurich INFN - Florence Universit de Montral University of California, Davis INFN - Bari Univ. Di Roma La Sapienza Institute of Space Science INFN - Roma 1 National Technical University of Athens (NTUA) INFN and University of Roma Tor Vergata INFN Milano and Universit di Milano Bicocca CCLRC INFN and University of Roma Tor Vergata

Switzerland USA Italy Switzerland USA Germany

Germany Austria Slovenia Russia Russia Italy USA Romania France Switzerland Switzerland Germany Italy Canada USA Italy Italy Romania Italy Greece

Italy Italy

U.K. Italy

1142 MARCHETTO MARRO MASSA MATTIAZZO MERIDIANI MESSINA MESSINEO

Flavio Enzo Romeo Fabrizio Serena Paolo Marcello Alberto

MESTER MIGNECO MIHOKOVA

John Emilio Eva

MIKHAILIK MILITARU

Vitalii Otilia

MIRELA-CAMELIA MONTANI MOORE MORZENTI MOSER MOZZANICA

Zgura Livia Christopher Sabrina HansGuenther Aldo

NAPPI NESSI-TEDALDI NOVARIO

Eugenio Francesca Raffaele

OBERTINO OLDRATI OLIVER

Maria Margherita Massimo Concepcion

PAOLUCCI PAPADOPOULOS PASTORE

Pierluigi Ioannis Francesca

PATEL PATER PATERNO' PEDRETTI

Mitesh Joleen Giovanni Marisa

PIA PINFOLD POMORSKI POSPISIL

Maria Grazia James Michal Stanislav

INFN Sezione di Torino Open Systems srl INFN - Roma 1 INFN - Padova INFN - Roma 1 ETH-Z INFN and Universit di Pisa Stanford University INFN - LNS Institute of Physics, ASCR University of Oxford Institut de Physique Nuclaire, UCL Institute of Space Science ENEA Christie Hospital Universit di Milano Max-Planck-Intitute fuer Physik Universit degli Studi di Milano INFN - Bari ETH-Zurich University of Insubria, Varese INFN - Torino

Italy Italy Italy Italy Italy Switzerland Italy

Open Systems srl Universidad Autonoma Madrid INFN - Napoli CERN Univ. Di Roma La Sapienza CERN University of Manchester Open Systems srl Universit dell'Insubria INFN Milano INFN Sezione di Genova University of Alberta GSI Czech Technical University in Prague

Italy Spain

USA Italy Czech Republic U.K. Belgium Romania Italy U.K. Italy Germany Italy Italy Switzerland Italy Italy

Italy Switzerland Italy Switzerland U.K. Italy Italy Italy Canada Germany Czech Republic

1143 PRICE

Lawrence

PRYDDERCH

Mark

PUERTA-PELAYO PUGMIRE

Jesus Gregg

RAINO' RANCOITA RANDO RASEL

Silvia Pier Giorgio Riccardo Ernst M.

REBUZZI

Daniela

REGLER

Meinhard

RIBONI RUCHTI

Pier Luigi Randy

RUPPI RUSSO RUTH SAKHAROV SAKURAI SALA SALERNO

Martino Stefano Thomas J. Alexander Kunitomo Silvano Roberto

SALZBURGER SANGIORGIO SANI

Andreas Samuele Gabriele

SANTAGATA SANTIN SCAPPARONE SCHLEICH SCHMIDT SCHRAM SCHUTZ SCIORTINO SEGUINOT SELLER SERFON SMYRSKI

Alfonso Giovanni Eugenio Wolfgang Rudiger Malachi Bernard Silvio J. Paul Cedric Jerzy

Argonne National Laboratory CCLRC/Rutherford Appleton Lab. CERN Nexsan Technologies Limited INFN - Bari INFN - Milano INFN - Padova Institute for Quantum Optics, Hanover University INFN and Universit di Pavia Institute for High Energy Physics ETHZ-Zurich University of Notre-Dame, National Science Foundation INFN - Bari INFN - Napoli TRIUMF CERN Waseda University INFN - Milano INFN Milano and Universit di Milano Bicocca CERN Universit dell'Insubria Universit degli Studi di Pavia ENEA European Space Agency INFN - Bologna Universitaet Ulm CERN University of Carleton Albert Einstein Institute Universit di Firenze Collge de France, Paris Rutherford Lab CPPM Institute of Physics, Jagellonian University

USA U.K. Switzerland U.K. Italy Italy Italy Germany

Italy Austria Canada USA

Italy Italy Canada Switzerland Japan Italy Italy

Switzerland Italy Italy Italy The Netherlanc Italy Germany Switzerland Canada Germany Italy France U.K. France Poland

1144 SOPCZAK SOSSI

Andre Vesna

SOZZI

Federica

SPADA

Francesca

SPARVOLI

Roberta

SZYMANOWSKI

Hanitra

TARASOVA TARKOVSKY THOMPSON TITOV

Oxana Eugueny Richard Maxim

TLUSTOS TOPKAR

Lukas Anita

UCHE

Nnona

ULYANOV

Alexei

USLENGHI VAN NIEUWENHUIZEN VENTURA VILLASENOR VOLONTE' VOSS VYKYDAL WALLRAFF WEBER WELLISCH

Michela G.

WENG

Andrea Luis Sergio Kai Zdenek Wolfgang Michael Johannes Peter Joanna

YAMASHITA ZANEVSKY

Naoyuki Yuri

ZHU

Chaofeng

ZONA

Crist iano

Lancaster University University of British Columbia INFN and Trieste University Univ. Di Roma La Sapienza INFN and University of Roma Tor Vergata DKFZ German Cancer Research Center DESY ITEP General Electric Research Freiburg University /ITEP Moscow CERN Bhabha Atomic Research Centre National Academy of Science Institute for Theoretical and Experimental Physics INAF-IASF Milano Massachusetts Institute of Technology INFN - Lecce University of Michoacan ESA Paris University of Victoria NIKHEF Physikalisches Institut B Physikalisches Institut B Privatgelhrter CERN/University of Karlsruhe Waseda University Joint Institute for Nuclear Research Institute of Inorganic Materials INFN Milano and Universit di Milano

U.K. Canada Italy Italy Italy Germany Germany Russia USA Germany Switzerland India Nigeria Russia Italy USA Italy Mexico France Switzerland The Netherlands Germany Germany France Switzerland Japan Russia China Italy

Proceedings of the 9th Conference

Astroparticle, Particle and Space Physics, Detectors and Medical Physics Applications The exploration of the subnuclear world is done through experiments

increasingly

covering

a

complex

wide

range

of energies and in a large variety of environments-from particle accelerators, underground detectors to satellites and space

laboratory.

For

these

research

programs to succeed novel techniques, new materials and new instrumentation need to be used in detectors, often on a large scale. Therefore, particle physics is at the forefront of technological advance and also leads to many

applications.

Among these, medical applications have a particular importance due to health and social benefits they bring to the public. I his book reviews the advances made in all technological aspects of the experiments at various stages.

6099 he ISBN 981-256-798-4

'orld Scientific YEARS or B i -

PUBLISHING 2 o o 6

www.worldscientitic.com