This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
(S*)(p (A) |He| cp (S) q> (A*) >*^(4) The matrix elements for Coulomb interaction represent the repulsive electrostatic interaction between the electronic charge distributions in the initial and final state, respectively, and have the following shape: Qi C = cp s* (D
(5)
The matrix elements for exchange interaction represent the repulsive electronic interaction of the electronic charge distributions, and have the following shape Qi e = cp s* d)cp
A
(2) and Q f e = cp s (2)cp A* (1)
(6)
In fig. 10, electronic energy transfer by Coulomb and exchange interaction are compared. In case of Coulomb interaction, the electrons initially on the excited donor stay there whereas the electrons initially on the acceptor also remain there. This interaction does not require physical contact between the interacting partners, it is sufficient that the excited sensitiser ion induces a dipole oscillation on the activator ion. In case of energy transfer governed by exchange interaction, the way the energy is transferred can be visualised by a double electron substitution reaction: the excited electron on S* travels to A, whereas an electron on A goes to S. This type of interaction requires overlap of the electronic charge distribution, i.e. physical contact between the sensitiser ion and the activator ion.
386
0„.<1>
0.(2)
_e£_ R
0„ O 0
0 . I2) A
0(1)
0..(Z)
D
A
electron (l^exchangeX^
0(2) D
0
(D A
Fig. 10. Visualisation of energy transfer by Coulomb interaction (a) and exchange interaction (b)
3.b.l. Energy transfer governed by electrostatic interaction In this section, we will discuss, using a simple model, the factors governing the magnitude of the electrostatic interaction and how they relate to the rate of energy transfer. The interaction energy ESA between two dipoles is given by the magnitude of the two dipoles (us and UA, respectively) and the distance RSA between them: ESA00 USUA/R3SA
(7)
Forster \ 2 , 3 has identified u s and uA, respectively, with the oscillator strength for the radiative transitions S* <-> S and A* <-» A, respectively. We will now treat energy transfer more quantitatively. The power irradiated by an oscillating dipole u cos cot is given by: P= 4co|u|2/3c3
(8)
The result obtained is a factor of 4 larger than in classical electrodynamics, to differentiate between photons absorbed and emitted. The rate of decay A is given by (energy emitted per unit time divided by the photon energy #co): A = 1/TO = (4co|up/(3c3))/(/i co)
(9)
387 in which To is the radiative lifetime. We obtain for ||i| 2 : \\i\t = 3hc3/(&im>Ho)
(10)
The oscillator strength for an optical transition is related to |u|2 in the following way: f=2m0)|u| 2 /(3fie 2 )
(11)
|u|2 is thus given by: |u|2 = (3fte 2 )f/(2m(o)
(12)
Apart from some numbers we are now able to calculate the rate of energy transfer from sensitiser ions to activator ions, being the product of equation (9) for the sensitiser and (12) for the activator ion divided by the distance RSA6 (see above): WSA - 3hc3/(87H03Tos) (3 h e2)fA / (2mco RSA6)
(13)
or after some rearrangements and inclusion of the numbers mentioned above (which originate from the expansion of the interaction Hamiltonian in spherical harmonics), the following equation is obtained : WSA = 2 n I h 1/RSA6. 3e2c*/t 614m. fA / x s . x\. J gs(E) gA(E) / E 4 .dE
(14)
In this equation, Ts is the decay time of the sensitiser ion and r| is the quantum efficiency of the sensitiser ion. Equation (14) can also be written as: WSA=l/Ts.(Ro/RSA)6
(15)
where Ro is the distance at which the transfer rate to the activator is equal to the decay rate of the sensitiser and is given by: R06 = r|f A (3e 2 c 3 >J 5 7i/2m)j" gs(E) gA(E) / E 4 .dE
(16)
388 Inspection of the equations derived shows that energy transfer, governed by Coulomb interaction is favoured by a large spectral overlap, a small value of the intrinsic decay time of the sensitiser ion, a large absorption strength of the activator ion and a small distance between the sensitiser and activator ion. In practice, it is of more importance to determine the efficiency of the energy transfer process than the rate. The decay of the sensitiser ion is given by: 1/TS=1/XO+1/TS.RO6/RSA6
(17)
The yield T|SA for energy transfer is given by: T1SA=1/-CS.RO6/RSA6/(1/I;S)
( 18 )
being the energy transferred per unit time, divided by the total amount of energy transferred per unit time. For the efficiency of the energy transfer process T)SA, we therefore find the following proportionality: TISA = ( R O / R S A ) 6
(19)
Please note that the quantum efficiency of the sensitiser ion is contained in Ro (see equation (16)). For a distance RSA between the sensitiser ion and the activator ion smaller than Ro, energy transfer will dominate, in the opposite case, inherent decay of the sensitising ion S* is the most important process. With a distribution of sensitiser and activator ions, the rate constant for sensitiser deactivation is given by: 1/ is = 1/xo + l/xs. Z Ro6 / RSA6
(20)
The decay curve for emission (|)(t) originating from the sensitiser ions is therefore given by: <|>(t) =
(21)
where IT is determined by the transfer rate As seen from equation (21), in case of sensitiser ion - activator ion, the decay curve of the sensitiser ion is likely to be non exponential, due to transfer to activator ions close to the sensitiser ions and a slower transfer in case of more distant activator ions.
389 In principle, a number of regimes can be distinguished for which the expressions describing the decay will be given: -
Excitation into the sensitiser ion is followed by emission from the same S ion or migrates only in the S ion sublattice.
I(t) = l(0)exp(-t/x) -
(22)
Some transfer from sensitiser ions to activator ions occurs, but no energy transfer from S ions to S ions occurs:
I(t) = I(0)exp(-t/T-At 3 / n )
(23)
in which the parameter A describes the concentration of activator ions and the sensitiser ion - activator ion interaction strength. In this equation, n gives the nature of the multipolar interaction (see below). -
Energy transfer between sensitiser ions and from sensitiser ions to activator ions occurs: - The rate between sensitiser ions is much faster than between sensisiter and activator ions (fast diffusion) I(t) = 1(0) exp (-t / T) exp(-AA.PSA.t)
(24)
in which PSA is the transfer rate from sensitiser to activator. The rate between sensitiser ions is much smaller than between sensitiser and activator ions (diffusion limited energy migration). For a three dimensional lattice and after a long time t, the decay curve is given by: I(t) = 1(0) exp (-t/ T - AA.D3/4.t)
(25)
in which D is the diffusion constant for energy migration. 3.b.2. Energy transfer by higher order Coulomb interaction Apart from dipolar interaction, also higher order interaction may result in energy transfer. In table I, the distance dependence of interaction involving dipoles and quadrupoles is summarised for the case that the optical transitions involved are spin allowed:
390 Table I. Equations for energy transfer governed by Coulomb interaction as a function of the type of interaction for spin allowed transitions.
Interaction type electric dipole - electric dipole electric dipole - electric quadrupole electric quadrupole - electric quadrupole
Equation form WSA = (1/ts) (RO/RSA) 6 WSA = (1/ts) (RO/RSA) 8 WSA = (l/^s) (RO/RSA) 1 0
Range (nm) 3.5 0.8 0.2
A more quantitative analysis, using the expressions derived above, shows that in case of electric dipole interaction: Energy transfer from a broad band emitter to a line absorber only occurs between nearest neighbours - Energy transfer from a line emitter to a broad band absorber is possible for distances up to about 2 nm - Energy transfer from a broad band emitter to a broad band absorber is possible for distances as large as about 3.5 nm. 3.3. Energy transfer governed by magnetic interactions Dexter has formulated a theory describing energy transfer by exchange interaction5. The rate constant for energy transfer is written as: WSA = KJexp(-2RsA/L)
(26)
In equation (26), K is a constant determined by the interaction between the orbitals involved in the energy transfer process, J is determined by the spectral overlap integral, normalised for the absorption strength of the activator ion and L is determined by the van der Waals radii of the interacting ions. The exponential dependence is due the fact that the electron density in general falls of exponentially on the distance between the electron and the nucleus. As the energy transfer process does not involve electric dipole interaction, no dependence of the magnitude of the electric dipoles on the sensitiser and activator ions is expected. In view of the relation between the magnitude of the electric dipoles and the oscillator strength (see above), therefore no relation between the oscillator strength of the optical transitions on sensitiser and activator ion is expected. In fact, for exchange interaction, there is no relation between the interaction between the ions and any spectroscopic characteristic of the sensitiser or activator ions. Another difference between energy transfer governed by Coulomb and exchange interaction is the distance dependence. The rate of energy transfer shows an RSA" dependence in case of Coulomb interaction and an exp(-RsA/L) dependence in case of energy transfer by exchange interaction. In case of energy transfer by exchange interaction, the rate drops very quickly for distances RSA larger than about lnm.
391 Finally, exchange interaction, due to the requirement of wave function overlap, has a pronounced angular dependency. 3.4. Competition between energy transfer by exchange and Coulomb interaction The arguments developed above show that Coulomb interaction can induce energy transfer over relatively large distances, whereas exchange interaction can only result in energy transfer in case of overlapping wave functions. About 10 years ago, this has been studied very nicely by Buijs et al. on the system (Gdi.xEux)MgB50io, ref . These authors studied the integrated luminescence intensity of the 5Do —»7Fo _ 4 emission of the Eu3+ ion on excitation within the f-manifold of the Eu 3+ ion (7F0,i -* 5L6 excitation). In view of the small oscillator strength of the parity forbidden 4f* transitions, a linear dependence of the emission intensity on the Eu3+ concentration is expected. A deviation from the linear dependence indicates energy migration to killer centres resulting in concentration quenching. At 4.2 K, concentration quenching is almost absent, at 300 K concentration quenching sets in at values for x larger than approximately 0.85. In two- or three dimensional systems, concentration quenching generally sets in at lower concentrations. The probability that the migrating exciton encounters a non-radiative recombination centre is largely reduced by a decrease in dimensionality of the sub-lattice in which the exciton is mobile. In general, the emission intensity of the activator ions after a long time t, where trapping is the only decay mechanism for the migrating energy, can be written as 7,8: I(t) = exp (-t/T0 - A x^d+2)
f*^ 2 ')
(27)
in which xa is the activator ion concentration and d is the dimensionality of the system. A is a constant depending on the activator ion - activator ion transfer probability PAA according to: A = 3(7i 2 P A A /4) 1 / 3
(28)
In fig. 11, taken from the dissertation of M. Buijs9, the luminescence decay curve of the 5Do -> F 2 transition on Eu3+ recorded at 112 K has been given for the compound Euo.999Ndo.ooiMgBsOio,. The solid line is a fit to: I(t) = exp(-t/-c 0 -Bt 1/3 )
(29)
In this equation, B contains both the activator ion - activator ion transfer probability and the activator ion concentration. The good fit implies one-dimensional energy transfer, see equation (27).
392
105-
* 1°4to 1
l«'»»J s
-* 1
o
103-
o
Relative
1
10°-
* **r
I 1500
I I 3000 4500 Time (jis)
I 6000
,., 7500
Fig. 11. Decay curve of the 5D0 —> 7F2 transition on Eu + has been given for the compound Euo.999Ndo.ooiMgB5Oio recorded at 112 K. The solid curve is a fit to equation (25).
The analysis discussed above points to exchange interaction as the mechanism leading to energy transfer. The EuMgBsOio structure consists of linear zig-zag chains along the crystallographic b-axis with intra chain Eu-Eu distances of about 4A and inter chain Eu-Eu distances of 6A. The good quality of the fit to the equation for one dimensional energy transfer points to the absence of three dimensional energy transfer. Only for exchange interaction, the Eu-Eu distance dependence is strong enough to completely suppress three dimensional energy transfer. For more details, the reader is referred to the original paper6. 3.5. Practical implications The arguments developed above of course have practical implications. In general, the activator ions used in luminescent materials are rather expensive. This implies that the concentration of the activator ions should be as low as possible. When energy transfer processes are important, the smallest activator concentrations are possible for the case of broad band emitting sensitisers and broad band absorbing activators. On the other hand, for high quality fluorescent lamps, line emitters are required to obtain an optimised adaptation of the emission spectrum to the human eye with respect to the amount of visible light generated and the need to reproduce all colours in a natural way. As discussed above, this requires that sensitiser ion and activator ion are nearest neighbours, i.e. at least one of the two kinds of ions should be present in relatively high concentrations.
393 When optical absorption on the activator ions is forbidden at the energy where the sensitiser ion emits (the activator ions, however, should have an excited state there), in principle there are two options to nevertheless obtain energy transfer: - Coulomb interaction exploiting the quadrupole of the activator ion. This is the mechanism underlying the fluorescent lamp phosphors BaMgAl10Oi7:Eu,Mn10 and (Ce,Tb)MgAln0 19 " . Due to the vanishingly small spectral overlap between the Ce3+ absorption and emission bands no energy migration between the Ce-ions takes place. This requires not only short Ce-Tb distances in this luminescent material, but also relatively high Tb concentrations, to prevent Ce emission to occur. - Exchange interaction between sensitiser and activator ion. As shown above, this mechanism does not require allowed optical transitions. This is the mechanism which is operative in the one component white fluorescent lamp phosphor Ca5(P04)3(F,Cl):Sb,Mn, as deduced from an analysis of the decay curve for some Mn2+ concentrations 12. The same study did not reveal evidence for energy transfer between antimony ions, indicating the necessity of nearest neighbour Sb-Mn interaction, which is a prerequisite for energy transfer via exchange interaction. Please note in addition, that in view of the large Stokes Shift between absorption and emission on the Sb 3+ ion in this lattice, no energy transfer between the antimony ions is expected. Both for electric dipole - electric quadrupole and exchange interaction, the distance between sensitiser ion and activator ion has to be rather small, not larger than about lnm. 4. Excitation with high energy particles After absorption of electrons or high energy photons (e.g. X-ray quanta), impinging on the phosphor material, secondary charge carriers, i.e. electron-hole pairs are generated in the lattice. The electron-hole pairs thermalise. After thermalisation, the excitation is transferred to an activator (or sensitiser), eventually leading to emission. For each absorbed electron or high energy photon, a large number of electron-hole pairs will be generated. Each electron-hole pair can give rise to emission of one photon on the activator ion. Robbins has treated these processes more quantitatively13. The energy efficiency of the overall process is given by: TI = ( l - r ) < h v > / ( P E g ) . r | t r | a c t r | e s c
(30)
in which r is the backscatter coefficient, expressing that not all high energy particles are entering the material,is the mean photon energy of the photons emitted, PEg the energy needed to generate a thermalised electron-hole pair (P being a pure number and E g being the band gap energy), r|t is the transfer efficiency of electron-hole pairs to activators or sensitisers, T)act is the quantum efficiency of the activator ions and finally T|esc is the ratio between photons leaving the material and generated in the material (escape probability).
394
Backscattering is neglectable for X-rays, whereas for electrons, the backscatter coefficient r generally adopts values between 0.1 and 0.2. For applications, phosphors operating at physical limits are of interest only. This means that the transfer efficiency, the activator efficiency and the escape probability have to be unity. Equation (30) then simplifies to: r| = (1 - r)/ PEg
(31)
Inspection of this equation shows that the energy efficiency is determined by the mean energy of the photons emitted by the activator ions and by the product PEg. The primary particles, eventually generating the luminescence, lose their energy by impact ionisation and generation of optical phonons. In what follows, we will discuss this mechanism in some more detail. The average energy needed to create an electron-hole pair can be written as: 3E g =E i + Eop + 2Ef
(32)
In this expression, E; is the ionisation threshold, E the average energy lost in generating optical phonons and Ef is the threshold energy for the generation of electronhole pairs The ratio of the energy needed to generate optical phonons with frequency tim LO and impact ionisation is proportional to R, which in turn is given by: R = (1/ e~ - 1 /
E„)
(ha LO)1 5 / (1.5 Eg)
(33)
in which e„, and e0 are the optical and the static dielectric constants of the phosphor host lattice, respectively. The dependence of P on R is given in fig. 12. The number P is found to vary between about 2.5 and 10 for a number of host lattices. We observe that in order to obtain host lattices with small P, resulting in highly efficient phosphors, the value of R should be small as well. This condition implies a low optical phonon frequency or a small difference between the optical and the static dielectric constant. In table II, for a number of well-known phosphor materials, the relevant data are given. We observe a good agreement between the energy efficiencies observed and the maximum efficiencies predicted.
395
Index of efficiency
R = l . —— E 1.5 Eg
Fig. 12. Dependence of p on R Table II. Physical constants of a number of efficient host lattice excited phosphors. r)the is the maximum efficiency calculated, r)eXp is the energy efficiency observed experimentally.
Phosphor CsLTl ZnS:Ag ZnS:Cu CaS:Ce CaS:Mn La 2 0 2 S:Eu Y 2 0 3 :Eu YV0 4 :Eu
Urn LO (eV) 0.011 0.044 0.044 0.047 0.047 0.057 0.068 0.116
E g (eV) 6.4 3.8 3.8 4.8 4.8 4.4 5.6 3.7
(eV) P 2.25 2.5 2.75 2.9 2.3 2.9 2.3 3.0 2.1 3.0 2.0 3.9 2.0 4.6 2.0 7.5
Tithe
Tlexp
0.14 0.25 0.21 0.16 0.15 0.12 0.07 0.07
0.14 0.20 0.17 0.22 0.16 0.11 0.08 0.07
The mechanism discussed above involves efficient energy transfer from host lattice states to localised states. This mechanism also occurs in some luminescent materials, applied in Xe-discharges. The Xe-discharge generates radiation in the VUV part of the spectrum. In these materials, the host lattice itself acts as sensitiser. 5. Electroluminescence The direct generation of emission out of electrical energy is called electroluminescence. Two kinds of electroluminescence will be treated here: - Thin film AC electroluminescence - Thin film DC electroluminescence.
396 The mechanism, leading to electroluminescence in inorganic Light Emitting Diodes (LEDs) will not be discussed in this contribution. 5.1. Thin film AC Electroluminescence In thin film AC electroluminescent (ACTFEL) devices, an inorganic electroluminescent material is sandwiched between two inorganic electrodes. Like in the case of most inorganic phosphors, the emission is generated on activator ions. On applying an alternating voltage with values typically above 100 V, a current breakthrough is realised in the device. Charge carriers in the luminescent material are created which excite activator ions in this material, resulting in electroluminescence. The charge carriers can easily lose there energy by excitation of optical phonons. Therefore, a strong electric field is required as only in this case the amount of energy, taken up from the field, can be larger than losses due to phonon emission. Based on the ACTFEL principle, displays and even electroluminescent lamps can be realised. AC electroluminescent devices generally have a very low efficiency. Now, we will treat this principle in some detail and compare the equations obtained to the ones obtained in the case of cathode-ray or X-ray excitation. It will be shown that the low energy efficiency is due to a very small transfer efficiency, i.e. a very small value for the transfer of energy from host lattice states to activator states. The mechanism, underlying electroluminescence, is clear for a few materials only. For the well known material ZnS:Mn (which is applied in displays), the luminescent centres are excited by impact excitation. The maximum energy efficiency obtainable is given by: Tlmax=em oN/eF
(34)
whereem is the mean photon energy of the emitted radiation, o the crosssection for impact excitation, N the optimum activator ion concentration and F is the electric field applied. Inspection of this equation shows that 1/ (0 N) is the mean distance that an electron travels through the luminescent material between two impact excitation events. The cross-section for impact excitation cannot be extracted from first principles. In case of ZnS:Mn, we estimate it to be 10"16 cm2, i.e. atomic dimensions as the electric charge of the Mn and the Zn ion are equal and the size of the ions is almost equal. The other values for ZnS:Mn are em = 2 eV, N = 1020 cm"3 and F = 106V/cm. Insertion in equation (34) shows that r) = 2%, in excellent agreement with the experimentally observed values. However, we have used a number of simplifications: we did not account for the Stokes Shift (the energy loss on the activator ion) and we neglected light trapping effects in the thin electroluminescent layers. These phenomena further reduce the energy efficiency. It is not very likely that the energy efficiency can be improved significantly. This is
397 mainly due to the low value for the trapping cross-section as N cannot be chosen arbitrarily high in view of concentration quenching. The mean energy hvexc the charge carrier has taken up from the electric field between two excitation events (our simple treatment is not only valid for impact excitation) equals:= eF / (o- N)
(35)
In this equation, any losses due to phonon emission have been neglected. The minimum pathway Lcrit the charge carrier has to travel equals: Lcrit =/eF
(36)
Note that Lcrit depends on the electric field strength. Insertion of equation (36) into the equation for the energy efficiency results in: Tl = (em / ) (T N Lcri,
(37)
Comparing this expression to the one obtained for the case of cathode-ray excitation shows o N Lent to be the transfer efficiency. In case of cathode-ray phosphors, the transfer efficiency can be unity, for ZnS:Mn in case of electroluminescence, the transfer efficiency is only about 0.02. The low transfer efficiency is the main reason for the low energy efficiency values of electroluminescent materials. The low transfer efficiency, in turn, is due to the low value of the cross section for impact excitation and the relatively low activator ion concentration. 5.2. Thin film DC Electroluminescence In the last decade, thin film DC electroluminescence has gained considerable interest. Here luminescence is generated by applying a small voltage over either an electroluminescent polymeric material or a so-called low molecular weight material (LMW), which is completely organic or is co-ordinated to rare-earth metals. In the latter case, eventually the rare-earth metal has to be excited. In this way the, for many applications, superior spectral properties of many rare earth emitters (line emission) can be utilised. In addition, one can expect higher energy efficiencies: in case of electric excitation three times as many triplet states as singlet states are generated (based on spin-statistics). Only emission from the singlet states is allowed in view of the spin selection rule. This immediately implies a maximum quantum efficiency (on a photon per electron basis) of only 25% (neglecting interaction between triplet states). In case of incorporation of heavy ions, like rare-earth ions, spin-orbit interaction can relax the spin selection rule.
398 In general, electroluminescent devices based on polymers (Poly-LEDs) are prepared by spin coating whereas electroluminescent devices based on LMWs (OLEDs) are prepared either by evaporation or spin-coating. In Poly-LEDs, the polymeric material is contacted to an electrode with a high work function at the one side and an electrode with a low work function at the other side, respectively. In this way holes and electrons can be injected into the polymer, respectively, which generate luminescence on recombination. The voltages to be applied can be very low, values as low as 2 V have been published for red emitting Poly-LEDs. OLEDs generally have a more complicated structure, like the one given in fig. 13.
Alq
Eu(DBM)3(Phen)
PBD
""•
—II
""•
1
• ^ • • l
1
TPD
- — Mg:AI (Ca:AI) —— Alq - — —— - — ——
Eu(DBM)3(Phen): PBD TPD ITO Glass Substrate
Fig. 13. OLED structure with electron and hole transporting layer.
The light emitting material is sandwiched between electron and hole transporting layers. These charge transporting layers, in turn, are contacted to the electrodes. The electron transporting layer to an electrode with a low work function, the hole transporting layer to an electrode with a high work function. We will now briefly illustrate electroluminescence in OLEDs, based on a publication of one of the pioneers in the field: J. Kido14. Kido used a Eu-complex, the chemical structure of which is given in fig. 13 in a layer consisting of PBD (also given in fig. 13). The excitation mechanism is energy transfer from the host (PBD) to the Eucomplex. Without doping of the PBD layer with the Eu-complex, luminescence is obtained from either the hole transporting layer or the electron transporting layer, dependent of the thickness of the PBD layer. The recombination of the electrons and the holes presumably
399 takes place in the PBD layer, followed by excitonic energy transfer to either of the layers. Please note that though PBD is an electron transport layer, it does not possess hole blocking properties. On adding Eu(DBM)3(Phen) (see fig. 13), Eu3+ luminescence is obtained, the underlying process being energy transfer from PBD to the Eu-complex. 6. Energy transfer in action I, the emission spectrum of Y2C>3:Eu In Y 2 0 3 , two sites for the Y-ions are available, one site with S 6 symmetry, the other one with C2 symmetry. In both sites the surrounding oxygen ions occupy the corners of a slightly distorted cube, however two of the available eight positions are not occupied by oxygen ions. The S6 site, which occurs three times as many as the site with C2 symmetry, has inversion symmetry. As a consequence for Eu3+ ions incorporated at this site, an emission spectrum with its main emission at about 590 nm is expected, originating from the 5Do—> Fi optical transition, the only allowed optical transition for the Eu3+ ion at a site with inversion symmetry. This is not observed, instead we observe an emission spectrum with its maximum at about 611 nm (see fig. 5), originating from the 5Do —> 7 F 2 optical transition on the Eu3+ ion. Detailed investigations15 have revealed that efficient energy transfer occurs from Eu3+ ions incorporated at the S6 sites to Eu3+ ion at C2 sites. The emission spectrum of the red emitting material Y 2 0 3 :Eu is almost ideally suited for high quality fluorescent lamps with respect to colour rendering and adaptation to the eye sensitivity, despite the fact that only one quarter of the Eu3+ ions are located at sites with the appropriate symmetry, the underlying mechanism being energy transfer.
7. Energy transfer in action II: blue and green emission from the Tb
ion
A phenomenon not discussed until now is cross-relaxation. In such a process, which can also be looked upon as energy transfer, the excited ion transfers only part of its energy to another ion. For two interacting Tb 3+ ions, the process is depicted in fig. 14. In this case, the energy difference between the 5 D 3 and 5 D 4 excited states matches approximately the energy difference between the in the 7 F 6 ground state and higher 7Fj states. As in the energy transfer processes discussed above, at large Tb-Tb distances, the process of crossrelaxation has a low rate. In many host lattices, therefore, at low Tb concentration emission from both the 5 D 3 and S D 4 excited states is observed (unless the gap between these two states is bridged by phonon emission, for which relatively high energy phonons are required). The resulting emission spectrum has emission from the near UV into the red part of the optical spectrum. At higher Tb concentrations (in the order of five percent), crossrelaxation quenches the emission from the 5 D 3 level in favour of emission originating from the D 4 level, implying that it is not possible to obtain blue Tb 3+ emission in luminescent materials with higher Tb 3+ concentrations.
400
40 E (103cnV1) - 30
=D„
- 20
10
7
Fa
Fig. 14. Cross-relaxation between two Tb3+ ions.
Cross-relaxation also occurs for other ions. It quenches blue Eu3+ emission at already relatively low Eu3+ concentrations (<1%), in favour of the well known red emission. In case of ions like Sm3+ and Dy3+, cross-relaxation leads to quenching of the emission. This seriously limits the applicability of these ions. 8. Energy transfer in action III: the luminescence properties of (La,Gd)OBr:Ce, Tb The luminescence properties of (La,Gd)OBr:Ce,Tb have been studied a few years ago . The material is of interest for application in devices in which a very high excitation density is applied, like in cathode-ray tubes for projection television. LaOBr, doped with 1% of Ce3+ ions, is a cathode ray phosphor with moderate energy efficiency only, approximately 5%. On increasing the Ce-concentration, the energy efficiency does not increase significantly. On co-doping this material with Gd, the energy efficiency increases to values of about 12%, see fig. 15.
401
13
I
/
-
1
1
1 1
A
0
11
f
A
A
'
A
-
I' ID C LU
1
1
A 1
1
-
X
+ 0
>+ /
A
*
X
0.02 %Ce 0.2 %Ce 1.0 %Ce _ 2.0 %Ce
1
1
i
0.4
0.6
^
/
X
1
0
1
0.2
1
I
0.8
i
1.0
X Fig. 15. Energy efficiency of (La,Gd)OBr:Ce as a function of the Gd concentration.
Interestingly, the energy efficiency increases already at very small Gd3+ concentration, in the order of 1%. An activator (or sensitiser) ion concentration of about 1% corresponds to a mean distance of about 2 nm. Therefore one might think of Gd-Ce energy transfer as origin of the energy efficiency enhancement and that energy transfer from host lattice states to Gd states is more efficient that to Ce states. However, in the excitation spectra of the Gd and Ce emission (not shown), no host lattice sensitised emission of Gd is observed. In addition, co-doping LaOBnCe with Y also results in a higher energy efficiency and Y has no levels which could participate in an Y to Ce energy transfer process. In ref.17, this phenomenon has been dealt with in detail. The interested reader is referred to that paper for details. In the present contribution, we will deal with the results only. Both for co-doping with Y and Gd, an increase of the cathode-ray efficiency is observed. In case of Y, Y to Ce energy transfer is not possible and in case of Gd, the Gd emission is hardly sensitised by host lattice states in this lattice. The only factor which remains is the host lattice to Ce energy transfer, which is enhanced by co-doping with either Gd or Y. In , this has been studied, using time resolved luminescence measurements to calculate the transfer efficiency from host lattice states to Ce states as a function of the Gd-concentration, see fig. 16.
402 1.0
0.8
0.6-
0.4
La^xGdyOBriCe —i
0.0
|
0.2
i
|
i
0.4
|
0.6
i
|
0.8
r
1.0
X Fig. 16. Transfer efficiency from host lattice states to Ce-states in (La,Gd)OBr:Ce as a function of the Gd concentration
We observe an increase in transfer efficiency from host lattice states to Ce states on increasing Gd concentration, in this way leading to a higher energy efficiency (see equation (30)). A tentative explanation is the following: La3+ and Ce + have the same charge and almost the same ionic radii. Therefore, incorporation of Ce3+ in LaOBr gives an only slight distortion of the host lattice. As a result, the capture cross-section will be determined mainly by the oscillator strength of the Ce3 ion. Both Gd3+ and Y3+ are much smaller than La3+. This is expected to result in an extra contribution to energy transfer from host lattice states to Ce-states because of the larger lattice relaxation (in the ground state) around the Ce-ion, in view of the decreasing mean metal-ligand distances on increasing the Gd (or Y) concentration. In this way the trapping cross-section is not only determined by the oscillator strength of the Ce3+ ion but also by the lattice deformation around the Ce3+ ion and consequently the transfer efficiency of host lattice states to Ce-states increases. For application in displays, the emission of (La,Gd)OBr:Ce is too deeply blue. Codoping with Tb3+ results in a.o. additional green emission, originating from emission within the Tb3+ 4f-manifold. Surprisingly, the energy efficiency increases further, despite the emission of green emitting photons (which have a lower photon energy than the blue ones originating from Ce3+). LaOBnTb and also (La,Gd)OBr:Tb are very efficient cathoderay phosphors, with energy efficiencies even exceeding the one of (La,Gd)OBr:Ce. We therefore assign the increase of the energy efficiency in (La,Gd)OBr:Ce,Tb to energy transfer from Tb3+ to Ce3+ (please note that this is the opposite energy transfer as encountered usually), the further increase in energy efficiency being due to the even more
403 efficient energy transfer from host lattice states to Tb states than to Ce states in these materials. 9. Energy transfer in action IV: luminescent materials showing a quantum efficiency larger than unity The search for luminescent materials with a quantum yield exceeding unity has a long history and can be looked upon as the holy grail of luminescence jphysics. Already in the seventies, the Pr3+ ion has been shown to show a quantum efficiency of 140% in YF3, see 18,19. In the nineties, this ion was shown to be able to show quantum cutting in oxide hosts as well, see e.g. 20 , 21 . The mechanism underlying the quantum yield larger than unity is a so-called cascade emission on the Pr3+ ion, involving the 4f-levels of this ion, see fig. 17. It has been reported (based on Judd-Ofelt calculations) that the quantum efficiency of this cascade emission on the Pr3+ ion can reach quantum efficiencies of 199%, see e.g. . Despite the interesting effect and quantum efficiency, its usefulness is restricted. This is due to the fact that one of the f-f- emissions is located at about 408 nm, i.e. in a spectral region where the eye sensitivity is low. In addition, white light generated with this blue emission is of low quality, see e.g.23. Because of the physical nature of the f-electrons (they do not participate in the chemical bonding and are therefore still almost atomic like), one cannot shift the spectral position of the emissions significantly by varying the host lattice. The second emission in the cascade is located near 620 nm. As such this emission is useful, both in terms of eye sensitivity and obtainable white light quality.
404
I ,i
50
40
~ E o
5d band
-
30
3P
"
or
3
20
2
Pn
10 —1
,~"'
•<
'
}3Hj
Fig. 17. Cascade mechanism underlying the emission process resulting in quantum efficiencies larger than 100% on the Pr3+ ion.
Therefore, there has been a search for other ions and other cascade schemes. To the best of our knowledge, no other ion has been identified showing a cascade emission resulting in a quantum efficiency larger than unity in the visible part of the spectrum. As an alternative, luminescent materials showing the cascade in host lattice states have been investigated. Our analyses have shown that this process does not result in energy efficient quantum cutters, see e.g.22,23. The underlying physical reason is that the quantum yield starts exceeding unity only at excitation energies larger than approximately three times the band gap energy (note that at least twice the band gap energy is required), resulting in materials less energy efficient than currently applied fluorescent lamp phosphors. Quite recently, it has been shown that potentially useful quantum cutting materials can be obtained, based on physically interacting rare-earth ions: an ion pair showing the cascade emission. The most striking example is LiGdTvEu. After excitation of the Gd'3+ ion, part of the energy is transferred to Eu3+ by cross-relaxation. Thereafter part of the remaining energy is transferred to Eu3+ by energy transfer, see fig. 18.
405
5040
£
30
o
CO
o
100 EiP+
Gd3+
Gd3+
Gd3+
Gd 3 +
Gd3+
Fig. 18. Processes leading to quantum cutting in LiGcffVEu.
In this way, an internal quantum efficiency of about 190% is obtained, see e.g. , . Materials like these cannot be used in standard fluorescent lamps, based on a Hgdischarge, as the excitation, leading to quantum cutting (corresponding to an optical transition on the Gd3+ ion), is located in the vacuum ultraviolet part of the spectrum, at about 200 nm. The potential application area of these materials is in devices based on a Xe-discharge, like plasma display panels or Hg free fluorescent lamps. 10. Energy transfer in action V: energy transfer and phosphor degradation Transfer of energy is not in all cases a useful process. Energy transfer to nonradiative centres is a well known destructive process in luminescence science. Nonradiative centres in the bulk or at the surface of luminescent materials are responsible for the generally lower energy efficiencies on increasing the activator ion concentration in the phosphor. The presence of defects at the very surface of luminescent materials can easily explain the efficiency loss frequently encountered on decreasing the penetration depth of the excitation in e.g. low-voltage or vacuum ultraviolet excitation schemes, see e.g. 26,27. Such centres might be present in the as prepared material, but might also originate from interaction of the luminescent materials with the ionising radiation used to generate the luminescence. In this section, we will deal with this latter phenomenon using vacuum ultraviolet spectroscopy on BaMgAlioOn:Eu (BAM) as example28.
406
BAM is applied in fluorescent lamps and Plasma display panels (PDPs). In PDPs it shows degradation under prolonged VUV excitation, i.e. the quantum efficiency decreases. The quantum efficiency loss is much larger than encountered in fluorescent lamps. In fig. 19 the spectral light output of BAM is given as a function of the excitation wavelength for virginal BAM and BAM degraded for 600 hours under VUV excitation in our set up.
1 ,H
non irradiated 3*
1,0-
#1/1
/
1 0,84-.
I
FVVI
0,6-
CJ)
0,4-
CD
02-
ath
fJ
'
A
JV
d(
sv,v
i
H/
/
/
§
rr
/A
yJ600 h\ irradiated
'
0,0- L - '
T
150
'•'
1 —
200
'
'- 1
250
'•'
1
300
1
350
Wavelength [nm] Fig. 19. Spectral light output of virginal BAM and BAM degraded for 600 hours under VUV excitation.
We observe a decrease in spectral light output, due to a decrease in quantum efficiency in the VUV part of the spectrum only. In the UV part (A > 200 nm), almost no loss in spectral light output is encountered. This rationalises the applicability of this materials in fluorescent lamps. There are two other experimental observations which give as a clue to unravel the phosphor degradation mechanism. - Though BAM is a relatively efficient cathode-ray phosphor (with an energy efficiency of about 10% which is high for an oxide, see above), application is hampered by the extremely fast degradation (seconds) when excited with an electron beam. This very fast degradation to our opinion points to an electronic origin of the underlying mechanism, rather than an ionic one. - Thermal treatment of BAM leads to light output losses. For VUV excited BAM, these losses already set at thermal treatments above 300 °C, for UV excited BAM light output losses occur only after thermal treatments above 700 °C.
407 The fact that prolonged VUV irradiation leads to a loss in light output on excitation with VUV radiation and an, if at all, smaller effect is found on excitation with UV light points to the importance of host lattice excitation (band to band absorption), followed by energy transfer to Eu2+. In the degraded material, energy is lost because of energy transfer from host lattice states to killing centres. UV radiation is absorbed by the Eu2+ ions, consequently no decrease on quantum efficiency is encountered in this case. Which, however, are the VUV excitation induced killing centres? To our opinion, VUV irradiation induces photo ionisation of Eu2+. In this way Eu3+ is formed, together with an F-centre (consisting of an oxygen vacancy which captures the electron obtained by photo ionisation). It is expected, based on ionic radii arguments (Eu3+ is much smaller than Ba2+), that Eu3+ is a very effective killing centre in BAM, due to the huge relaxation of the excited Eu3+ ion which is possible in this lattice. The fact that the killing centres are produced purely electronically also rationalises the observation of an extremely fast degradation of BAM when excited with cathode-ray excitation. Diffusion processes are not expected to be that fast. Cathode-ray excitation, on the other hand, produces many host lattice excited states, similar to the ones generated by VUV excitation, which in turn can easily induce photo ionisation. Finally, thermal treatments lead to a faster loss in light output of BAM for VUV than for UV excitation. This, to our opinion, is due to the decreased penetration depth of the VUV radiation, compared to UV excitation, as VUV radiation is absorbed by host lattice states, whereas UV radiation is absorbed by the Eu2+ ions. Only when a relatively large number of Eu3+ ions is formed (as a consequence of the higher temperature), this is reflected in a reduced light output under UV excitation, the underlying mechanism being energy transfer from Eu + ions to Eu + ions. 11. Summary and outlook In this contribution, examples are given of mechanisms, resulting in visible emission in quite a number of luminescent materials, with emphasis on the processes underlying transport of energy in these materials. Many of these mechanisms and processes are quite well understood. Despite this high level of understanding, many topics remain of considerable interest, both from a scientific and an industrial point of view: - The use of as such known energy transfer processes has resulted in exciting new developments like luminescent materials with a quantum yield larger than unity. - New classes of luminescent materials, being completely organic in nature or consisting of an organic host and rare-earth ion activators have been developed. - New developments require alternative excitation schemes: - Short wavelength UV excitation for devices based on Xe-discharges - Long wavelength UV excitation or even excitation by blue light for phosphor-LEDs or emissive LCD displays - Very low voltage excitation (< 10 eV) for organic (or polymeric) LEDs
408 These developments clearly show that research on luminescent materials is alive and exciting, the driving forces originating from the availability of new materials and new applications. Quite frequently heard statements that research in this field has effectively come to an end therefore do not reflect reality and solely show a lack of imagination and creativeness. 12. References 1. T. Forster, in Fluoreszenz Organischer Verbindungen (Vandenhoech and Ruprech, Gottingen, 1951) 2. T. Forster, Disc. Faraday Soc. 27 (1959) 7 3. T. Foster, Ann. Physik 2 (1948) 55 4. B. DiBartolo, in Energy Transfer Processes in Condensed Matter, edited by B. DiBartolo (NATO ASI Series, Series B: Physics 114, Plenum Press, New York, 1984) 5. D.L. Dexter, J. Chem. Phys. 21 (1953) 836 6. M. Buijs and G. Blasse, J. Lum. 34 (1986) 263 7. J. Klafter, A. Blumen and G. Zumofen, /. Stat. Phys. 36 (1984) 561 8. B. Movaghar, G.W. Sauer and D. Wttrtz, J. Stat. Phys. 27 (1982) 473 9. M. Buijs, Dissertation, State University of Utrecht (1987) 10. A.L.N. Stevels and J.M.P.J. Verstegen, J. Lum. 14 (1976) 207 11. J.M.P.J. Verstegen, J.L. Sommerdijk and J.G. Verriet, /. Lum. 6 (1973) 425 12. T.F. Soules, R.L. Bateman, R.L. Hewes andE.R. Kreidler, Phys. Rev. S7 (1973) 1657 13. D.J. Robbins, J. Electrochem. Soc. 127 (1980) 2694 14. J. Kido, H. Hayase, K. Hongawa, K. Nagai and K. Okuyama, Appl. Phys. Lett. 65 (1994) 2124 15. M. Buijs, A. Meijerink and G. Blasse, J. Lum. 37 (1987) 9 16. C.R. Ronda and T. Welker, Eur. J. Solid State Inorg. Chem. 28 (1991) 545 17. T. Welker, C.R. Ronda and KJ.B.M. Nieuwesteeg, J. Electrochem. Soc. 138 (1991) 602 18. W.W. Piper, J.A. de Luca and F.S. Ham, J. Lum. 8 (1974) 344 19. J.L. Sommerdijk, A. Bril and A.W. de Jager, J. Lum. 8 (1974) 341 20. A.M. Srivastava and W.W. Beers, J. Lum. 71 (1997) 285 21. A.M. Srivastava, D.A. Doughty and W.W. Beers, J. Electrochem.Soc. 143 (1996) 4113 22. C.R. Ronda, J. Alloys and Compounds 225 (1995) 534 23. C.R. Ronda in, Spectroscopy and Dynamics of Collective Excitations in Solids, edited by B. DiBartolo (NATO ASI Series, Series B: Physics 356, Plenum Press, New York, 1997) 24. R.T. Wegh, H. Donker and A. Meijerink in, Book of Extended Abstract of the 192nd Meeting of the Electrochemical Society, Paris, France (1997) 1766 25. R.T. Wegh, H. Donker, K.D. Oskam and A. Meijerink, Science 283 (1999) 663 26. D.B.M. Klaassen, H. Mulder and C.R. Ronda, Phys. Rev. B39 (1989) 42
409 27. H. Bechtel, W. Czamojan, M. Haase, W.Mayr and H. Nikol, Philips J. of Research 50 (1996) 433 28. T. Justel, H.Bechtel, H. Nikol, C.R. Ronda and D.U. Wiechert in, Proceedings of the Seventh International Symposium on Physics and Chemistry of Luminescent Materials, edited by C.W. Struck, K.C. Mishra and B. Di Bartolo, (Electrochemical Society, 1999)
This page is intentionally left blank
FOUR-WAVE MIXING STUDIES OF ENERGY TRANSFER PROCESSES GEORGES BOULON Laboratoire de Physico-Chimie des Materiaux Luminescents Universite Claude Bernard Lyon 1 Unite Mixte de Recherche 5620 CNRS Domaine scientifique de la Doua 69622 Villeurbanne cedex FRANCE ABSTRACT The main objective of the lecture is to investigate the energy transfer processes among dopant ions in inorganic crystals by both four-wave mixing spectroscopy and fluorescence dynamics techniques. Four-wave mixing has been demonstrated to be an effective technique for characterizing long range energy migration and optical dephazing phenomena. In addition, measurements of the fluorescence decay dynamics of active ions are used to characterize the total energy transfer properties between sensitiser and activator ions. The set-up used in these experiments will be described in the approach of the production of the laser-induced grating signal in solids. Two nonlinear laser beams are focused inside the sample and the path difference is adjusted in such a way that the two beams interfere creating a sinusoidal intensity pattern. Since the wavelength of the laser beam is resonant with an absorption transition of the active ions, optical absorption creates a similar spatial sinusoidal distribution of the excited ions that acts like a population grating. The laser-induced grating signal carries all of the informations about the physical processes influencing the population grating. Signal strengths and decay times are measured as functions of laser-beam crossing angles and temperature. We shall show the examples of two different activator ions in Garnets: Cr3+ transition metal ion and Nd3+ rare earth ion which are characterized by two kinds of spectroscopy. The results will be discussed.
411
412
1. Theory and Experimental Technique of Four-Wave Mixing The nonlinear mixing of two laser "write" beams and a laser "probe" beam to produce a fourth "signal" beam is a technique known as four-wave mixing (FWM). This can be treated theoretically as a third order nonlinear optical process or, alternatively as Bragg scattering from a laser-induced refractive index grating 1 ' 2 . The major significance of this technique is its application to optical devices involving phase conjugation, demultiplexing, beam switching and amplification. However, FWM can also be used as a spectroscopic tool for elucidating fundamental physical properties of materials relevant to optical applications. In order to develop systems for optical technology applications, it will be necessary to tailor-make materials with specific FWM characteristics. This can only be accomplished if there is a full understanding of the FWM processes in the material, and this is the goal of the research described here. There are many different types of physical processes that can lead to FWM through laser-induced modulation of the real or imaginary parts of the complex refractive index. These include direct electro-optical effects, thermal effects, concentration effects, structural changes, and population effects. Each of these has different characteristics in terms of response times and sensitivities. In general, population effects may be the most useful source of dynamically programmable FWM signals since they encompass a wide range of response characteristics and can be controlled through alteration of the defect content of the material. These include populations of free carriers, excitons, trapped charges, and excited ions. For permanent holographic information storage, induced structural changes may provide the best stability. Many of the most interesting nonlinear properties leading to FWM are associated with defects in the material. Thus the characteristics of the FWM signal can be changed by altering the defect distribution in the material. This provides a convenient method for tailoring the material to give a desired FWM response and makes it imperative to understand the effects of different types of impurities in a specific host material. Several types of experimental setups have been used for FWM studies depending on the time regime of interest. All of these involve crossing two laser beams in the sample to form an interference pattern in the form of a sine wave (Fig. 1). The light intensity in the peak regions of the pattern modulates the complex refractive index of the material compared to its value in the valley regions of the pattern. This creates a laser-induced refractive index grating. The spacing and position of the grating are many times exactly the same as the light interference pattern. However, in some cases the effects of spatial diffusion and long range interaction processes can cause the grating spacing of position to be different than that of the light interference pattern. A probe beam passing through the sample in the presence of a refractive index grating will be diffracted by the grating. By measuring the properties of the diffracted signal beam, all of the information concerning the physical properties of the grating can be determined.
413
Fig. 1.- Experimental setup for Four-Wave Mixing (FWM)
Figure 2 shows the typical alignment of laser beams used for degenerate FWM and the condition for Bragg diffraction. The probe beam is aligned counter-propagating to one of the write beams and the signal beam appears as a phase-conjugate beam to the other write beam. For non degenerate FWM mixing, the alignment changes slightly. In Fig. 3, the grating pattern is shown schematically. The grating decays with time as the modulated refractive index relaxes to its initial condition or as the induced nonlinear effect migrates from the peak to the valley region of the grating.
Bragg Condition f s =f p+ iT,-
'"•J
Fringe spacing at the gnuing is A-—
Fig 2.- Laser beam alignement for degenerate Four-Wave Mixing
Distance
Fig 3.-Laser-Induced Grating pattern (LIG)
414
1.0
z0.5
1^4—d
0.0
-1.0
=
500 jim
w2^i
Lai
0.0 ANGLE (OEG.
+ 1.0
Fig 4.- Computer-simulated angular scattering patterns for An = 3x10 .
Figure 4 shows typical angular scattering patterns for laser-induced gratings3. The general shape consists of a central Bragg peak and Bessel function side-lobes. The width and shape of the central peak and the frequency and size of the side lobes are sensitive to the width and modulation index of the grating. The narrow angular region of high scattering efficiency allows these laser-induced gratings to be useful in demultiplexing applications. Measurement techniques with FWM include: (1) scattering efficiency at the Bragg angle; (2) computer fitting of angular scattering patterns; (3) dynamics of signal buildup and decay; (4) write beam energy transfer; and (5) anisotropic self-diffraction. Each of these provides complementary information about the underlying physical processes involved in the formation and erasure of the refractive index gratings3. The variable parameters for experimental measurement include laser power, temperature, sample orientation, the defect content of the sample, write beam crossing angle, and laser wavelength. The latter two parameters alter the spacing of the induced grating through the relationship A = M2sin(9/2)}
(1)
415 For a simple sine wave grating, the scattering efficiency at the Bragg angle is given by Tl = exp (-2ccL) {sin2 (drcAn/2A.) + sinh2 (dAa/4)}
(2)
where a and L are the absorption coefficient and thickness of the sample and d is the thickness of the grating. For typical conditions, the functions in Eq. (2) can be expanded so that the FWM signal efficiency is proportional to the square of their arguments. For population gratings, the modulation index is proportional to the product of the difference in the complex refractive index for the ground and excited states and the population density of excited states in the peak region of the grating. For a simple twolevel atomic system under equilibrium pumping conditions, the latter is given by i
N2p = NIo7{hv/x + ol}
(3)
where N is the total concentration of atoms, ois the ground state absorption cross section, T is the excited state lifetime, and v and I are the frequency and intensity of the laser. This predicts a direct relationship between the FWM signal efficiency, the laser power and the fluorescence lifetime and predicts the occurrence of saturation at high intensity levels. For pulsed excitation the signal will built up and decay exponentially with a rate dependent on twice the fluorescence decay rate and on contribution from excited state migration dynamics. In the limit of incoherent energy migration, the FWM signal decay is given by 2 ^Jn\2 . 26 Ic(t) = IG(0)exp- — + 32a — sin T„ U; 2
(4)
where id is the fluorescence lifetime of excited ions, D is the energy diffusion coefficient and X is the write beam wavelength. In these experiments, the crossing angle 0 which governs the value of A given by Eq. (1) was varied from 2.5° to 22°. The grating decay rates were measured as a function of the write beam crossing angle for temperatures between 10K and room temperature. The intercept is found to be equal to twice the fluorescence decay rate of Cr3+ while the slope yields in Eq. (4) to obtain the diffusion coefficient. In the following sections, we present examples of using FWM techniques to investigate different types of laser-induced gratings in Cr3+-doped garnets and Cr3+Nd3+-codoped garnets.
416
2. Main Spectroscopic Properties and Transient Gratings in Cr3+-Doped Laser Crystals 2.1. Spectroscopy and dynamics ofCr3+ centers in garnet crystals Cr3+ ions have now been made to lase in a variety of different host crystals including ruby, alexandrite, emerald and various garnets. We had focused our efforts on the spectroscopic properties of Cr3+-doped Garnets with a general configuration coordinate diagram scheme shown in Fig. 5 only for 4A2 ground state and 2E and 4T2 excited states which are involved by the dynamics. We had selected (Ca,Zr,Mg)substituted Gd3Ga50i2 (GGG)7' 8, Gd3Sc2Ga30i2 (GSGG) and GGG9 and YAG (Y3AI5O12) ' ". It has been shown that some disorder exists in such garnet crystals due to a partial occupation of the octahedral sites (Ga3+, Sc3+, Al 3+ ) by Y3+ or Gd3+, then the luminescence of Cr3+ ions introduced as dopant in the hosts is affected.
4v 15•AE
10--
5--
"fS co O
1 Cr-
Fig 5.- Main energy levels of Cr3+ drawn by configuration coordinate diagram.
417
•
•
T = 10 K A GGG (CA,If):Cr'* •
T = 3Q0K AGGG
ICa.ZitCr3"
• GGGiCr 3 *
GGG:Cr"
•
• A A
*
i
J
A
A *
•
A
•
A 8000
10000
4
4
* 1
1
•
A
12000 2000
4000
6000
12000
14000
Fig 6.- Cr3+ emission probabilities at 10 K and 300 K. in GGG and GGG(Ca, Zr) as a function of the desexcitation rate from 2E+4T2 excited levels.
Our approach was to study how as an example, Gd3+-Ga3+ substitutions or Ca-Zr ion pairs modify both the main spectroscopic properties and the average crystal field strength in both GGG and substituted-GGG. A simple model based on the local dilatation of the lattice around each defect has been presented in order to describe Cr3+ luminescence at any temperature as can be seen in Fig. 612. Another similar approach has consisted to assume also a random substitution and to use statistical calculation method to interpret luminescence lines in Cr3+-doped YAG11. The presence of multisites makes more complex the study of decay profiles and time-resolved spectroscopy which may be associated to different crystal field strengths Absorption s p es e? fcixr-m of Cr:GGGM
E n e r g y l e v e l s of Nd i n G-GGM
Fig 7.- Energy levels of Cr and Nd in GGGM.
3
418
A
\_=579.52 o n
^_a579.l5 nm
/n 872
873
874
/ U_=57S.St nm
875
876
872
873
874.
B75
876
WAVELENGTH (nm)
WAVELENGTH ( n m )
Fig. 8.- Emission spectra of F3/2(a)-> 19/2(1) transition at 10 K under different excitation wavelengths in GGG:Nd3+ (left) and GGG:Nd3+-Cr3+ (right). A new peak e is appearing due to Cr3+-Nd3+ neighbour ions.
and we have presented detailed investigations of the spin-orbit and electron-lattice couplings in the different Cr3+ sites leading to a description of the experimental spectroscopic data13. Complementary studies of the different crystal field sites have also been investigated on Nd3+-Cr3+-codoped YAG or GGG10"14. The broad absorption bands of the Cr3+ ions allow effective excitation by the pump light and the positions of energy levels of the Cr3+ and Nd3+ in garnets allow the possibility of efficient nonradiative energy transfer from Cr3+ ions to Nd3+ions (Fig. 7). Decay-time measurements show that Cr3+—>Nd3+ energy transfer takes place in several types of Nd3+ sites (Fig. 8, 9). In the same way, energy transfer between multisites has been analyzed in Cr3+-Tm +codoped YAG15"16 where it has been shown a selective energy transfer within these Cr3+-Tm3+ pairs. In addition, under both time-resolved spectroscopy and decay time analysis, we were able to point out that Tm + ions which are the first
* > < O
1 QOO
2000
3000 4.000 Time QJLLS~)
SOOO
SOOO
Fig. 9.- Decay time curves of Cr3+ principal site in Cr3+-doped and Cr3+-Nd3+ codoped GGG. Energy transfer from Cr3+ sensitiser to Nd3+ activators is clearly seen.
419
neighbors of one Cr3+ ion perturb the spectral and dynamical characteristics of the 2 E—»4A2 Cr3+ sharp line, as we show it in Fig. 10 a and b17. This article shows that the dipole-quadrupole interaction was the more reliable one to describe the Cr3+-Tm3+ interaction at 10K. We wish to mention the special interest of this study of Cr3+-Tm3+ transfers in GGG by using time-resolved spectroscopy and analysis of decay profiles because we have observed Cr3+-Tm3+ pairs with the shortest distance of 3.5 A within the lattice for which the Cr3+ emission is shifted from 693.6 to 694.0 run. The knowledge of this distance allows us to determine the nature of the interaction without any assumption of the average values in donor-acceptor distances as it is usually done. In this compound and also in Cr3+-Nd3+-codoped YAG18, Rotman's theory has been very useful to analyze decays of fluorescence when a part of the donor ions has an enhanced fluorescence. It means that with Rotman's theory19"20, the probability distribution of acceptor ions around a donor ion is not uniform and depends on the radial distance R from the donor for a multipolar interaction (Fig. 11). If R
l i t IS 28 36 « S3 51S! 77 K 94102111119127135144152160169177185193202210
lejis) Fig. 10.- Time resolved emission spectra of Cr3+ emissions in both Cr3+-doped and Cr3+-Tm3+codoped GGG crystals (left) and their fluorescence decays (right)
420
All experimental data have been analyzed from decay profiles, time-resolved spectroscopy and site-selective spectroscopy of donors and acceptors. We would like now to show how other technique as Four-Wave Mixing Technique is helpful not only in understanding the pumping dynamics, ion-ion interaction and energy transfer characteristics, but also the radiationless relaxation processes. ^r
i' unjform
2-
1°
?
!
I
{
!
Fig. 11.-(a) The radial distribution for uniform, excluded and enhanced volume distribution of acceptors around donors, (b) Sample decay curves for excited donor concentrations for various donor-acceptor distributions (I-H means Inokuti-Hirayama).
2.2. Application of the Four-Wave Mixing Technique to Cr + centers 2.2.1. Energy transfer Long range energy migration among the Cr3+ ions can alter the mode structure, quantum efficiency and saturation properties of laser operation. In order to investigate the properties of energy transfer, FWM signal decay rates were monitored as a function of the crossing angle of the write beams22. Typical laser induced grating signal has been shown in Fig. 12 as a function of crossing angle for the excitation into 4T2 band of the main site and the linear part in Fig. 13 of the FWM signal decay rate as a function of sin20/2 allows us to measure the diffusion parameters and, in addition, to check that at 6 = 0, the intercept is found to be equal to twice the fluorescence decay rate of Cr +. The theory developed by Kenkre et al.23 for transient gratings shows that the observed transient grating signal exhibits exponential decay in the presence of exciton migration which can be described by : I(t) = expf-2{[(l/T + a) 2 + b2]'2 -a}tj
(5)
421
1
Cr3*-GGG X exc .=590 nm • Exp. - Theo.
0,8
|
0,6
I
•V
•
0,4
0.2 -
1
• •"^"-^^^
•
^*~»-_*
0 CROSSING ANGLE (c
Fig. 12.- LIG scattering efficiency as a function of crossing angle for the excitation into 4T2 band.
where x is the fluorescence lifetime a is the exciton scattering rate b = 4Vsin[(27ta/A,)sin9/2] being V the nearest neighbour ion-ion interaction rate and a is the average distance between active ions. I
i
1
i
Cr*3 - doped GGGM
15
slope(8jt? / \2}D
^
^
^
o
~
10
o
O 1 —
^^^
o^^ ] 2 / t at 6 = 0
5
i
0
5
i
10 ss ii nn 22((66 / 2 )
Fig. 13.- FWM signal decay rate as a function of sin2 8/2.
i
15 x10*
1
20
422 Table 1. Energy migration parameters for Cr3+-doped laser crystals. MATERIAL AND TRANSITION ""Alexandrite (invers.) 4 T 2
Dq AE (cm"1) (cm"1)
D (cm2/s x10"8)
V (sec"1 xlO 5 )
a (sec"1 xlO 5 )
T (K)
a (A)
N (cm" 3 xl0 1 8 )
Ld (cm xlO"6)
2200
6400
0
NA*
NA
ALL
41
15
NA
Ruby 2 E
1820
2300
0
NA
NA
ALL
59-108
79-490
3
Alexandrite (mirror) 2 E
1680
800
3
12
200
25
27
8.9
12
Emerald 4 T 2
1620
400
28
1.9
2.6
12
10
177
31
GGG 4 T,
1597
298
1
NA
NA
18
NA
140
NA
GGGM
1567
100
6410
42
12
18
15
54
16
GSGG 4 T!
1565
50
10
NA
NA
230
NA
NA
NA
LLGG 4 T,
1480
-1000
0
NA
NA
ALL
15
50
NA
NA Dq
AE D V a T a N Ld
Not available Universal notation of the strength of the crystal field. lODq measures the energy separation between the twofold degenerate e state and threefold degenerate t2 of transition metals energy gap between the bottom of each parabola associated to 2E and T2 states respectively diffusion coeficient nearest neighbour ion-ion interaction rate exciton scattering rate temperature average distance between active ions density of ions diffusion length
423
The exciton dynamics can be characterized by these parameters in terms of the diffusion coefficient D, the diffusion lenth Ld, the mean free path Lm and the number of sites visited between successive scattering events Ns. These parameters are given by : D = 2VV/ct Ld = (2Dx)1/2 Lm=42Va/a Ns = LJa
(6) (7) (8) (9)
Fitting Kenkre's theory to these data gives the value of energy migrating parameters. Figure 14 shows the temperature dependences obtained for D for several different laser crystals. No long range energy migration was observed for Cr + ions in ruby and in the inversion symmetry sites in alexandrite. For Cr + ions in the mirror symmetry sites in alexandrite, the diffusion coefficient was found to increase at low temperatures while in emerald the value of D was found to decrease at low temperatures. The former results are consistent with long mean free path exciton migration limited by acoustic phonon scattering while the latter results are consistent with phonon assisted exciton migration. Both types of temperature dependences were observed in different Cr3+-doped garnet crystals 4"25. Table 1 summarizes all energy migration parameters for Cr3+-doped laser crystals which are only associated to the most populated site in each crystal due to the channel of the excitation into 4T2 band by an allowed electric dipole transition whereas the signal intensity under selective 2E excited level pumping is not enough high to yield pumping operation within each of multisites. IO
4
E
8|
o
'
— ,
'
1
,
,
1
—
r
|
100 T(K)
:
ln....l.n .!•..
-[ W
200
Fig. 14.- Temperature dependence of the exciton diffusion coefficients for the Cr3+-doped laser crystals determined from FWM experiments.
424
2.2.2 - Dynamics of the nonradiative decay from 4T2 level to the 2E level ofCr and T2 dephasing time measurements.
+
ions
One important piece of information concerning the pumping dynamics of Cr3+ laser materials is the pump-band to metastable state relaxation rate. This can be determined by measuring the dephasing time of the FWM signal while pumping into the 4 T2 band. The dephasing of this pumping transition is dominated by radiationless relaxation to the 2E level. The model used to analyze the dynamics of the nonradiative decay from the 4T2 level to the 2E level of Cr3+ ions is shown in Fig. 15. Two channels are possible: AB'C: 4T2-2E intersystem crossing mentioned by ISC in Fig. 15, followed by internal conversion of phonons within the 2E state ABC: internal conversion of phonons mentioned by IC in Fig. 15, within the 4T2 state followed by 4T2-2E intersystem crossing.
q
(arb. units)
Fig. 15.- Model used to analyze the dynamics of the nonradiative decay from the T2 level to the E level of Cr3+ions.
425
The detail of this process are described in 26. It is interesting to note that for all the materials considered, the ratio Knr(ISC)/Knr(IC) of nonradiative relaxation rate by ISC and IC is greater than 1, confirming the conclusion drawn in 26 that, after pumping into the 4T2 level, the dominant relaxation path for the excitation is (AB'C) and not (ABC). Using a simple two-level model for the atomic system, the value of the dephasing time T2 can be determined from a theoretical fit of the FWM signal efficiency versus crossing angle of the write beams22. The results of doing this for ruby, emerald, the two different crystal field sites in alexandrite, and two garnet crystals give values of the radiationless relaxation rate between the 4T2 and 2E levels ranging from a few tenths of a picosecond to tens of picoseconds (Fig. 16). These values appear to vary exponentially with the crystal field splitting between the two levels with no substantial variation with temperature.
0
1000
2000
2000
3000
AE Fig. 16.- T2-dephasing time and the ratio K^ISQ/Kn^IC) as a function of AEpp (the energy difference between the peak of the 4T2 band and the peak of the 2E line in the absortion spectrum).
426
2.3. Application of the Four-Wave Mixing Technique to Ncf+ centers Four-Wave Mixing Techniques were used to characterize the long range excitation migration and measurements of the fluorescence decay dynamics of both types of active ions were used to characterize the total Cr3+-Nd3+ energy transfer properties. In Fig. 17, typical results are shown for the decay rates of Cr3+ population gratings in Cr-Nd-codoped GGGM, plotted as a function of sin20/2 at 3 OK. The intercept is found to be equal of twice of the fluorescence decay rate of Cr3+, while the slope has been used in Eq. (4) to obtain a diffusion coefficient of 3.47xl0"6 cm2/s. Fig. 17 also shows typical results of FWM signal decay rate as a function of write beam crossing angle at 3 OK for the excited state population grating of Nd3+ ions in Nd3+ -doped GGGM. The slope of these data analyzed using Eq. (4) yields a diffusion coefficient of 3.86 x 10"6 cm2/s
Sio
<e=/2) *
!0
S in
2
2
(e /2)
x
10*
3+
Fig. 17.- LIG decay rate as a function of sin 6/2 for Cr ions in doped and Nd-codoped GGGM at 30 K.
A theoretical estimate for the diffusion coefficient of the excitation migration among the Nd3+ ions was also calculated25 assuming electric dipole-dipole interaction between the Nd3+ ions and a random distribution of active ions. The value obtained for D was 8.9xl0"8 cm2/s which is significantly lower than the experimentally measured value. The measured value of D was compared to those obtained by similar measurements mode on other Nd -doped garnet crystals27 and the results are given in Table 2. To understand the sample-to-sample variation, a normalized diffusion
427
coefficient was again defined to eliminate active ion concentration differences. For electric dipole-dipole interaction, this is given by Dnorm(cm6/s) = (nNC)'4/3 Dmeas
(10)
Comparing the values obtained for Dnorm shows that the variation in the measured excitation diffusion coefficient values in different garnet crystals is not entirely due to the differences in the concentration of Nd3+ ions. Again the differences in the ion-ion interaction rates and the active ion distribution must be considered. Table 2. Measured and normalized values for the excitation diffusion coefficient for Nd3+ ions at 25 K.
MATERIAL Y3A15012 (YAG) Y3(Alo,5Gao,5)50,2 (YAGG) Y 3 Ga 5 0,2 (YGG) GGGM
N
(xirPW3) 1,1 1,1 0,27 3,02
Dmeas ( X l 0 - 1 0 2
cm /sec) 3,5 9 2 38600
D^TcxiF7 cm2/sec) 7 17 25 17679
The excitation diffusion coefficient has been measured as a function of temperature for both samples. Fig. 18a shows the results for the Cr3+ ions in Cr-Nd codoped GGGM and Fig. 18b shows similar results for the Nd3+ ions in Nd3+-doped GGGM. In both the cases, the experimental data can be fit to an expression of the form28: D = A+B/VT
(ii)
Here T is the temperature of the sample, A is a constant which contains the ratio of the rates of exciton scattering by acoustic phonons and by all other centers, and B is a constant involving the exciton velocity and the matrix elements for exciton-acoustic phonon scattering. These results are consistent with phonon scattering limiting the mean free path of the excitation migration28 which agrees with earlier results24. The higher value (0.085) in the case of excitation migration among the Cr3+ ions as compared to the lower value (0.02) for the case of excitation migration among the Nd3+ ions is consistent with the higher value of D for Cr3+ ions as compared to the Nd3+ ions. 3. CONCLUSION Under both standard spectroscopy, time-resolved spectroscopy, site-selective spectroscopy and decay time analysis, we have shown the presence of Cr3+ multisites in Cr3+-doped garnet crystals. As a matter of fact, these multisites make more complex the
428 D(X10-7cm2/sec)
•
Cr3+-Nd3+-codoped GGGM • Nd3+-doped GGGM
x /
0
mf 0
f
Jr C w 0
* 0
=
0
7 0
0
0
* ' '
yt% 0
5
10
15
20 2
25
30
35
2
1/VT (X10" K'" ) Fig. 18.-Temperature dependence of the excitation diffusion coefficient for Cr3+ and Nd3+ ions.
spectroscopic studies. In a complementary approach, the characteristics of the FourWave Mixing Technique signal of Cr3+ ions and also of Nd3+ ions in the same crystals were investigated only for the main site of each type of hosts. The results are used to determine the properties of energy transfer and of the 4T2-2E radiationless relaxation processes among Cr3+ ions. The energy migration properties are affected by the active ion concentration, the distribution of active ions, ion-ion interaction rates and temperature. The variation of each of these parameters from host to host makes it difficult to establish a simple trend for energy transfer in Cr3+-doped crystals. Each of them have to be studied separately. Especially measurements of the diffusion coefficient D are very useful in the understanding of energy migration processes. Energy migration involving Cr3+ transition metal ions with unshielded outer shell electrons interacting through short range exchange interactions are more sensitive to the positions of the nearest active ions than in the case for energy migration among rare-earth ions with shielded outer shell electrons interacting through longer range electric dipole-dipole interactions which are more sensitive to the intrinsic strength of the interaction. The temperature dependence of the excitation diffusion coefficient D among the Nd + ions again shows that the excitation migration is limited by phonon scattering. Finally, the optical dephasing studies indicate that anharmonicity of the vibronic potentials is required to explain the relationship between the T2 -dephasing time of the T2 level and the crystal field splitting when pumping into the higher vibrational levels of the absorption transition. Under these pumping conditions, the dominant relaxation
429
path appears to be direct intersystem crossing from 4T2to the 2E level followed by internal conversion within this level. Acknowledgements. The author would like to thank Professor R.C. Powell (Center for Laser Research. Oklahoma State University - Stillwater and now Optical Sciences Center - Tucson) for a successful cooperation between our two laboratories. References 1. H.J. Eichler, P. Gunter and D.W. Pohl, "Laser-inducedDynamic Gratings", SpringerVerlag, Berlin (1986). 2. R.A. Fisher, "Optical Phase Conjugation", Academic, NY (1983). 3. J.K. Tyminski and R.C. Powell, J. Opt. Soc. Am. B ~, 440 (1985) 4. H.C. Chow, M.J. Kliewer, J.K. Tyminski and R.C. Powell, J. Opt. Soc. Am. B 3, 688 (1986) L. Arizmendi, M.J. Kliewer, and R.C. Powell, J. Appl. Phys. 61,1682 (1987) L. Arizmendi and R.C. Powell, J. Appl. Phys. 61, 2128 (1987) 5. H. Kogelnik, BellSyst. Tech. J. 48, 2909 (1969) 6. J. Salcedo, A. Siegman, D. Dlott, M. Fayer, Phys. Rev. Letters 41, n° 2, 131 (1978) 7. G. Boulon, C. Garapon, A. Monteil, Optical Science and Engineering - Series 8 Advances inLaser Science ~ Proceedings of the Second International Laser Science Conference - Seattle (1986); ed. M. Lapp, W. Stwalley, G. Kenney-Wallace, American Institute of Physics - Conference Proceedings n°160, New York (1987), p. 87 8. A. Monteil, C. Garapon, G. Boulon, J. of Luminescence 39, 167 (1988) 9. A. Monteil, W. Nie, C. Madej, G. Boulon, Optical and Quantum Electronics 22, S247-S257(1990) 10. W. Nie, A. Monteil, G. Boulon, Optical and Quantum Electronics 22, S227-S245 (1990) 11. V. Lupei, L. Lou, G. Boulon, A. Lupei, J. of Phys. .Condensed Matter 5, 35 (1993) 12. A. Brenier, G. Boulon, C. Pedrini, C. Madej, J. Appl. Phys. 71 (12), 6062 (1992) and Optical Properties of Excited States in Solids, ed. by B. di Bartolo, Plenum Press, New York, NATO ASI Series. Series B: Physics 301, 445 (1992) 13. M. Grinberg, A. Brenier, G. Boulon, C. Pedrini, C. Madej, J. of Luminescence 55, 303(1993) 14. I. Vergara, A. Monteil, G. Boulon, C. Madej, J. Garcia-Sole, Materials Chemistry and Physics 26,181 (1990) 15. Y. Kalisky, S. Rotman, E. Luria, A. Brenier, C. Pedrini, G. Boulon, M. Kokta, Journal of Electroceramics 1 (1), 91 (1997) 16. V. Lupei, L. Lou, G. Boulon, A. Lupei, C. Tiseanu, J. Phys. I (France) 3, 1245 (1993) 17. A. Brenier, G. Boulon, C. Pedrini, C. Madej, Optical Materials 1, 299 (1992) 18. J. Mares, Z. Khas, W. Nie, G. Boulon, J. Phys. I (France) 1, 881(1991) 19. S. Rotman, F. Hartman, Chem. Phys. Lett. 152, 311(1988) 20. S. Rotman, E. Luria, N. Yitzhaki, A. Eyal, Optical Materials 5, 1 (1996)
430
21. M. Inokuti and F. Hirayama, J. Chem. Phys. 43, 1978 (1965) 22. A. Suchocki, G.D. Gilliland, and R.C. Powell, Phys. Rev. B 35, 5830 (1987) 23. V.M. Kenkre and D. Schmid, Phys. Rev. B 31, 2430 (1985) 24. F. Hashmi, K. Ver Steeg, F. Durville, R.C. Powell, G. Boulon, Phys. Rev. B 42, n°7, 3818(1990) 25. F. Hashmi, R. C. Powell, G. Boulon, Optical Materials 1, 281(1992) 26. G. Gilliland, A. Suchocki, K. ver Steeg, R.C. Powell, D. Heller, Phys. Rev B 38, 6227(1988) 27. M. Zokai, R.C. Powell, G. Imbusch, B. di Bartolo, J. Appl. Phys. 50, 5930 (1979) 28. V. Agranovitch and M. Galanin, Electronic Excitation Energy Transfer in Condensed Matter (North-Holland, Amsterdam) (1982)
UNCONVENTIONAL LIGHT EMISSIONS IN R A R E - E A R T H D O P E D SOLIDS.
F. A U Z E L Groupe Optique des Terres-Rares, UPR 211, CNRS, 1, Place 92195 MeudonCedex
A-Briand,
ABSRACT As for other fluorescence light emitters, Rare-Earth (RE)-doped solids usually follow two well known principles: i) Emission takes place generally through both incoherent spontaneous emission and coherent stimulated emission with a ratio between them given by the statistical approach of Einstein. In absence of population inversion, incoherent spontaneous emission normally overcomes the stimulated one. ii) Both type of above emission obey Stokes law stating that excitation photons are at higher energy than emitted photons. In the following lecture, we want to discuss what we shall call "unconventional cases" where at least one of the two above principle is not respected. It shall be shown that REdoped solids may deviate rather easily from above principle giving rise under strong excitation density to unconventional emissions of the coherent spontaneous or of the anti-Stokes types. In a first part, the conditions necessary for observation of coherent spontaneous emission are discussed. Comparison with stimulated emission are described in order to discriminate both types one from the other. Recent results obtained on RE-doped crystalline powders are presented as examples and discussed in terms of the coherent back scattering effect. In a second part, anti-Stokes emissions in RE-doped solids are finally described with emphasis on the distinction between cooperative and energy transfer effects.
1.Introduction Conventional light emitting processes in Rare-Earth-doped solids usually follow the basic well known rules: 1) Incoherent spontaneous and coherent stimulated emissions at energy hv0, even in powders, are in the ratio given by the well known statistical approach of Einstein: 4i=Ai(pfo'o/Av) (1) Where p, is the volumic density of modes sustained by a spectral width Av, and is given by: p = Sm^Av/c3 (2)
431
432
2) Excitation of such emissions follow Stokes' Law which simply states that excitation photon energy, hvPi has to be at an higher energy than the photon emitted energyftvo: hvp>hv0 (3) In the following we shall consider the unconventional opposite cases for REdoped solids: a) Coherent superfluorescence light sources from powders in comparison with stimulated emission. b) Anti-stokes emissions of the following types: APTE effect, Cooperative effects, Photon avalanche. It can be noted that both types of unconventional emission could eventually be entangled. 2 Super-radiance and superfluorescence in RE-doped solids [1]. 2.1 From spontaneous emission to superfluorescence. Before going any further, we have to recall that under a strong laser excitation, the applicability of Golden Fermi's Rule: Wn = ^ | <2| V(t)\ 1)| 2S(E2-E,-
E)
(4)
may be questioned because the population in the ground state 1 should stay about constant, and condition W12.t « 1 has to be fulfilled which means that a short time interaction, adapted to W12, is necessary. It is clearly no longer the case when looking for highly populated states through laser excitations. Then the perturbative approach is no longer valid. In the following, the induced and spontaneous emission in the completely quantized case (both field and atom ) shall be considered for a two-level atom model. To do that we have to consider the time dependant Shrodinger equations for the two state system constituted of the two-level "atom + field" system; they become: ihCi(t) = C2(t)(
ihC2(t) = C1(0(^0-)|F(Ok1(Oy' ' (6) where C 2 2 ( 0 and C 2 (/) describe the probability for the complete system to be in states:2 with:
433
|C, 2 (0|+ |C 2 2 (0|= 1
(7)
The 2-levels atom itself is described by states I g> and | e> linked by rising and lowering operators: a+|g>=|e> and o"|e>=|g> Defining the following projection: system is written:
(8)
OZ=O+O"-CJ"O+,
the steady state of the total
H0 = —hcoQaz + hcoa+a 2
(9) It represents the "atom+field"system before coupling. The corresponding initial and final states of the total system being: |(pl>=|l> and \q>2>=\2>, Then emission and absorption shall respectively represent the following transitions:
\l) = \e,n)-*\2) = \g,n+l) and\l) = \g,n)^\2) = \e,n-l)
(1Q)
with n the field mode and with a total energy change (coo for atom and co for field): ha2X=h{a>0-co) (H) the time dependent interaction between field and atom may be written [1]:
HAF = Hg°+a ~ g*°~a+)
(12)
where g describing the electric dipole transition strength is proportional to the its matrix element. Assuming a strong external resonant field (n>l and coi2=0), it can be shown [1] that the probability to have the system in state 12> or to go from 11> to 12> is: |C 2 (0| 2 =
Sin2^2 (13) It oscillates ad infinitum with the elastic exchange between field and atom at Q0, the "Rabi" pulsation at resonance: Q=Q0=2gn1/2 (14) Out of resonance (©o^co), it becomes: 1
1
|c2(0f = § i - s , « ^ n
2
With the general form for Rabi pulsation:
>u+w>+r^
(15)
(16)
I C2(t) I is found to be reduced in the ratio of the Rabi pulsations at resonance to non resonance and the oscillation is accelerated being dephased with respect to the resonant case. Such is the case when an inhomogeneous transition is considered; only one frequency of the transition can be in resonance; all the others are dephased. See Fig. 1:
434
£2 2 =co 2 , 2 +C2 2 0 J
'
n=i.nn0 («.,2=05n0)
„;o,
1.0-
f\\
0.8-
/
A / \
0.6-
J 0.40.2-
0.01
..
2%
An
... .
I
nt Figure 1. Rabi oscillation of |C|2 for a resonant (to12=0) and a non-resonant case (co^O).
In order to know the field radiated by the dipole during this exchange with the atom, one has to calculate the expectation value of the electric dipole moment during the process:
(HoMkco)
(17)
Where the wave function for the mixed state of the whole system, y/(t), is given from the definition of C] and C2by: ¥(t) = Cx{t)(pl + C2(t)2 + Cos—^-q)^ (19) It shows that the system is in a mixed state except at extrema t=0; t=7i/£2o ; t=27t/Qo;
Since interaction HAF was the dipole moment: ( \ed\ ) - —Sinnote"a"'(l|^|2)
+ compl• conjugate (20)
435
Or assuming g to be real: ( le.rfl ) = ^^-Sina)0tSinnot
=
2
^Sinco0tSinnot 4
(21)
It shows that the dipole oscillates at the transition frequency
51 = -(C1*C2V"°' + C, C2Via,°') 52 = i(-C;C2et(a°' + Q C2V""°') 5 3 = | C I | a - | C 2 '2
(22)
With previous values for the resonant case: C 2 (f) = »'S»» —
and
C^O^Cos^(23)
It comes: 51 = SinQ.otSmco0t 52 = SinQ,0(Cosa>0t 53 = CosQ.0t
(24)
From above it can be noted that S2 is proportional to <e.d>. The locus of the Bloch vector extremity as given by above coordinates, (Equ.(24), describes a sphere as shown on Fig.2. No damping whatsoever has been considered up to now. It means that once, by some unspecified way, field or excitation of the atom have been created, the Bloch vector extremity is going to describe a parallel circle of the "Bloch Sphere" ad infinitum, keeping the Q0 existing when the field was suppressed. Assuming that at t=0 the Bloch vector is at the "north pole", a pulsed excitation of duration t=7i/2fi0» leaves the Bloch vector circling at the "equator" of the sphere. It is called a "JI/2" pulse. This kind of excitation gives the maximum <ed>. The field existing through <ed> after the excitation.has been stopped, is "free induction" emission.
436
0.5,
-0.5,
Figure 2. Bloch sphere for the resonant case (co12=0)
Now we shall consider some damping in the exchange process between field and atom. The most basic one is spontaneous emission that shall be introduced here through the "Weisskopf-Wigner Equation". By definition, and since we know of its existence since the statistical approach of Einstein, we shall try to introduce it in the previously considered "atom+field" system. Spontaneous emission is the one in absence of external field: n=0, and the initial state and final state of our two states system are respectively: |l>=|e,0> and |2>=|g,lx> However all modes (k) with different polarization and orientation with n=l are possible and Shrodinger equations become:
inCl(t)^(e,0\HAF\g,lx)C2(t)ei^'
(25)
(26) Solving for CI gives, after some calculations (see [1], p.315), the "Weisskopf-Wigner equation" is written: • , . %T + iAE _ , , c,(0 + — Z T — c , ( o + o 2fc (27) Where:
437
^\(s^\HAF\efi)\2Scol2
nT/2 = n
"
*
X
(28)
(.®12
{0fora12=0}
^
With same initial conditions as before (Ci(0)=l; C2(0)=0; C^t)*!), the solution to Weisskopf-Wigner equation is: Q(t) = e
e
(30)
And:
\CM\2=e~r'
(31)
Where T is the Weisskopf-Wigner damping originating from spontaneous emission. The probability of being in initial state |e,0> is a decreasing exponential with time: t=r1 (32) T is the natural width of a level connected with its lifetime by uncertainty relations. AE is a shift in level position induced by the field when con^O; it is usually negligible. In the definition of T, replacing the matrix element by its value: r =
^ZfiV(o+ij^12 «
(33)
A
gives for 1 mode: r
A = 2zg2Sco
12
(34)
which is the Fermi golden rule for spontaneous emission. Replacing n=0 in Rabi nutation definition, at resonance (co12=0), leads formally to Q=0: There is no Rabi nutation for spontaneous emission alone. In this instance it can be remarked that:
lim|C 2 (0r=4g 2 ^^ 1 2 =^spont 2
r
T
v w2 ° x = Wspont. = 2xg $c°n 12 (35} spon,. —"& which is just the spontaneous emission probability per unit time for on mode as given above by equ.(34).
438
Because we see that pure spontaneous emission is an asymptotic behaviour, in the following we shall consider the merging of coherent emission as defined by the existence of a Q*0 with spontaneous decay. We have also seen (equ.30) that the coupling of the atom in its excited state with an empty field provides a damping constant (neglecting shift): C,(t) = e-T>"2
(36)
Deriving: j»q(0=-i»^q(0
(37)
Specialising to the 2-level system, the Schrodinger equations become:
ih C i ( 0 =
C2(t)hgJn'e"°"'
ihC2(t) = C,(/)/& V»e*"»'
(38)
-ih^-C2(t) 2
(39)
Posing: (40)
The system is solved giving now for the Rabi pulsation: iT.
a' = (a21—p2+4g2n
(41)
and the probability becomes: 1 2
'
Q2
2
(42)
It shows that spontaneous emission behaves as a damping for Rabi oscillation. That the memory of excitation is kept through (n) as long as Rabi oscillation exists. We may conclude that induced processes (n) are not separated from spontaneous emission (TV) as may be thought when expressed as 1 in (n+1). Figure 3 shows this damping. During the spontaneous decay, in absence of any other external driving field and of any other damping process, field and atom may continue to exchange energy quasi-elastically through the induced process; this behavior is the root of "superfluorescence" that we shall discuss in the following chapter.
439 10 -i
0.8
0.6-
O
0.4-
0.2-
0.0 -I—i—|—i—|—i—|—i—|-
-i—i—i—[—i—,—i—|—i—r
2%
4K
n0t Figure 3. Damping of Rabi oscillation by spontaneous emission given by r 2 .
2.2 Field coupled cooperative multi-ions effects, super-radiance (SR) and superfluorescence (SF). Before going further we need to recall a number of relaxation times and their links with what has been just discussed above. In absence of damping, we have seen that the pseudo-spin vector movement is described by Bloch's equation: d§ dt
•= n A s
(43)
Where the vector -Q is defined by the coordinates (Qo, 0, ©12). With damping from only spontaneous emission, Bloch wrote it phenomenologically.
~ =
(&/,§)-[§
-S(\,i)]f->
i = 1,2 (44)
With T defined by the co-ordinates (T2, T2, T,). T2 is the so-called "Bloch transverse relaxation time"; it characterises perturbations on the phase of the dipole not its energy. Ti is "Bloch longitudinal relaxation time; it characterises perturbations on the energy of the system. The names
440
come from the fact that T2 applies to movement perpendicular to Q whereas Tl applies to the movement longitudinal with Q. In case the damping comes only from spontaneous emission, because S3 is associated with |Q(t)|2, it is also Weiskopf-Wigner damping. Then, Ti is identified with x the spontaneous radiative lifetime. Si, S2 being linked with C2(t) not its square, they are associated with 172. The corresponding damping is 172 linked with T2 by T2"'=r/2 In the general case: S], S2, being associated with Ci.Ck*, are connected with dephasing from any origin and T2 is the "dephasing time" In real sample, ions are in different surroundings giving rise to inhomogeneous spreading of con bringing contributions T2*to T2: 1 1
1 1
2
1 +
2
1
(45)
2
other interactions (ion-ion, ion-phonon) may give, besides radiative emission, homogeneous contributions T2' to T2: — ~—— + W rpi
h
OT ll
+W.
ion-ion
ion-phonon
\
(46)
Now let us consider a system of several two-level systems in a volume
N2)^-Sinw0t 2
(47)
(N r N 2 ) is population inversion density. The power radiated by the dipole is: P = E(t)-^
'& (48) where E(t)=EoCosco0t is the "reaction field" created by the dipole itself. It provides, in a cavity of quality Q, the amplitude E0: EI
P=
= %TCQP
ltoa
^Qo0(Nl-N2)2h2g2
(49)
or: -^ pU) in term of the dipole matrix element g of the atomic transition Note here the square behavior with population inversion density (N r N 2 ), instead of the linear one for usual emission. Also the emission starts without any delay because
441
the phasing of the atoms is obtained by the pump, T is found to be reduced by a factor (N]-N2) and becomes TSR as can be seen by its definition: stored energy TSR =
2(N, - N2 )tico =
. . (51)
radiated power P The condition to observe SR is that the macro-dipole keeps on existing during transition that is:
H
I s
" (52) Even when the pumping step involves some dephasing as is the case when a nonradiative step takes place during the pumping process, the spontaneous emission may exist in a somewhat coherent way. This is called super-fluorescence(SF) for which we shall now describe the basic process. Here we consider the cooperative emission of a large number of initially inverted atoms, without initial macro-dipole left by the excitation pulse (* SR) and instead of a 7t/2-pulse, now rather a 71-pulse is used, generally disturbed in its effects on the emitting level because the preparation of excited states is through phonon emission steps so avoiding a coherent pumping which would lead to SR. Emission starts by spontaneous emission with x decay and intensity oc N, the number of inverted atoms. After a delay, td, emission starts to be coherent with peak intensity « N2 and with decay TSF <* t/N. The macro-dipole has been created by the selforganization brought by the reaction field. We can review the conditions for SF: i)
dephasing has to be limited as for SR:
ii)
phase has to be kept during self-synchronizing step:
T
(53)
ta
442 They can also be given in term of a threshold critical lengths (Bonifacio's LT)[2]: LT
(57)
With: LT=LT S F/T 2 and 1C=CTC
3 Super-fluorescence versus amplification by stimulated emission (ASE) in powders. 3.1 Amplification by stimulated emission (ASE) and lasers with powders. Ideas about laser effects in scattering media are already old, since they can be traced back to the theoretical paper in 1966 by Letokhov[3] and to a first experiment in 1971 by Varsanyi[4] on the single grain of PrCl3, a fully concentrated material. However, real powder systems have been considered in a more systematic way only recently[4,5,6,7,8,9]. In parallel with this field, recent results obtained for laser action in dye solutions with passive scattering powder[10] have pushed forwards new theoretical investigations on gain in scattering medium in connection with photon localization^ 1] by the disordered medium and on coherent back scattering (CB) of light from amplifying medium[12]. In the following, we shall first review amplification by stimulated emission (ASE) and laser effects with active powders as observed under high peak power at low (77K)[5,6] and room temperature[8,9]. Such kind of results needs rather high concentration of about 1021 to 1022 cm"3 of active ions in order to obtain thresholds in the range of 2 105-109 W/cm2. We shall then discuss the superfluorescence effect, in the Bonifacio sense, that we have obtained on powder[13]. This effect is in marked contrast with ASE, since here low concentrations, below 1020 cm"3, are required providing, under continuous excitation (CW), very low density thresholds of about 103 W/cm2 [13] fromlOK to 60K[14]. Finally we shall present recent results for the super-radiance of Er 3+ , observed on a single layer of powder with grain diameter of about 30|am. 3.1.1 Laser effect in powders of stoechiometric and Nd doped laser materials. The first experiments of laser on powders have been obtained on so-called stoechiometric laser materials of Na 5 Nd(Mo0 4 ) 4 , LiNd(P0 3 ), NdP 5 Oi 4 [5,6]. Such family of materials, with the active rare earth ion as a constituent of the host, was well known in the late 70's as laser materials with tens of micron size active length[15]. Such effect has then been extended by the Russian group to a wider range of Nd-doped materials such as La 2 0 3 , La 2 0 2 S 5 , La 3 Nb0 7 , SrLa 2 W0 7 [7]. All these results have been obtained at 77K, with threshold from about 2 10 to 3 106 W/cm2 for grain diameters ranging from 250 to lOjim and Nd 3+ concentrations between 1021 and 4 1021 cm"3. Fig. 4 shows the various typical experimental set-ups.
443 YAG:Nd-» 2ra - *
dye laser
> \ . TEMPORAL SHAPE
lens filter dichroic mirror
lens
.optical fibre
CRYSTAL - .
C ^
/ OPTICAL FIBER [O P
0
CYLINDRICAL LENS
b)
a)
Figure 4. Experimental scheme for powder laser; a) from Markushev et al.[ 6]; b) from Gouedard et al.[8]
More recently, room temperature laser effect have been obtained under pulse excitation for Nd again in highly doped or stoechiometric powder materials with concentration above 1021 cm"3 in Lai.xP50M :Ndx, NdCl3 ;6H20[8] with threshold of 109 W/cm2 and in NdAl3(B03)4, NdSc3(B03)4, Sr5(P04)3F :Nd, with threshold above 1.7 107 W/cm2 [9]. Example of the pulsed behavior of the emission below and above laser threshold for NdCl3 ;6H20 is given on figure 5. ' •
powder laser
1
spontaneous decay (x2)
10ns i
•
.
•.
•
i
i
•
:
i .
•
i
Figure 5. Emission at 1.06um of NdCl3:6 H 2 0 in powder form under pulse excitation at 0.532pjn below and above laser threshold at 109 W/cm2; after Gouedard et al. [8].
444
For all these results the laser criterion is that above a pump threshold, a strong line narrowing is observed together with a pulsed time decay with time constants shorter than the spontaneous emission lifetime. The integrated output signal is linear with pumping above the threshold. 3.1.2 ASE in materials with coherent backscattering. Besides gain in powder medium, the so-called coherent backscattering effect on passive scattering powder has been discovered and investigated[16,17,18] at first independently of active powders and now in conjunction with active powders[12]. The coherent backscattering (CB) or case of photon " weak localization " can be understood by recognizing the fact that any random path of light inside a scattering medium, coming back to its impinging point can be " time reversed " that is can be followed the other way round. Those two waves interfere constructively whatever the randomness of the path. All other paths, not respecting the k -vector conservation law: -km=ko (58) vanish by destructive interference according to the phase mismatch caused by angle 0 between km and ko, where ko is the incident wave k-vector and km is the k-vector after the mth scattering event. The error on the phase mismatch is given by the average distance between the first and last scatterer that is also the mean free path between two successive scattering events, /. The corresponding dephasing is given by: AO = 2;ry0
(59)
where f?is the scattering angle with respect to the normal incidence, see Figure 6.
Figure 6. Dephasing in coherent back-scattering for an angle 6 from incident beam.
445
Then it can be shown that the coherently enhanced backscattered " cone " which is a gaussian function shall have an angular width at half intensity given by[16]:
24* = f * J
(60)
where X is the light wavelength. It has been experimentally shown that when the medium between the scattering objects is absorbing, for instance Rhodamine 6G solution with scattering polystyrene spheres, the CB cone flattens[19]. In fact, because the cone maximum corresponds to the longest paths, they are just the ones, which are the most reduced by absorption, though the scattering mean free path keeps the same. It is not a surprise that the opposite behavior is obtained when gain is provided inside the scattering composite medium. Both theory [20] and an experiment [12] have recently shown that the CB cone is narrowed by gain together with a rising of the background for larger angles as shown on Figure 7. This last experiment has been obtained in a pump-probe configuration in order to measure the CB cone at the probe wavelength with or without pumping of the scattering powder made of 0.15 wt % Ti 2 0 3 in A1203 (Titanium sapphire powder) either dry or in a water suspension. The ASE gain and CB cone are measured using a common frequency doubled Nd :YAG laser pumping both a regular Ti :Sapphire laser providing the probe signal, and the amplifying active scattering powder itself. This first experiment of controlled ASE amplification in a scattering powder does not present of course the threshold of a laser effect. Shortly afterwards, this experiment has been extended by the same group to a systematic study of the gain in Ti: Sapphire powder and of other random media such as ruby powder in air or in glycerol[21].
71K gain 36%gsm
t>Bfl08oU000|jO
I
L_
anglsfmrad)
Figure 7. Coherent back scattering in AI2O3 : 0.15%Ti powder under pulsed excitation at RT. Atter[10].
446
3.1.3 Super-radiance and superfluorescence in powders. A somewhat analogous behavior to ASE has been obtained, for emission at 2.7^i (I11/2—> I13/2 transition), on powder of L1YF4 :Er with large grain (>100iim)[13] and now extended down to 30um[22]. However the experimental conditions which prevail for the effect we are going now to describe are rather different from the one which were necessary for ASE in scattering media as just seen. The experimental set-up is simply the one usually used for recording a luminescence spectrum. The super-fluorescence effect can be viewed as the radiative coupling during T2 of one ion, S,, with another one, Sj, nearby as depicted on Fig. 8 by their Bloch sphere. When all ions in the samples are in phase, they emit coherently as a single "macro-dipole", the size of which is proportional to the number of coherently coupled ions. This provides an intensity proportional to the square of a dipole that is now proportional to the square of the number of excited ions in phase.
Figure 8. Ion Sj is synchronised by the spontaneous emission of ion Sj. Both ions are represented by their Bloch's sphere. From [23].
Here, contrary to ASE, concentration of active ions have to be kept low, in fact for RE practically below 1020 cm'3, in order to maintain T2 as large as possible. The observed threshold for line narrowing and spiking behavior for the output signal is very low (less than 103 W/cm2) and can be obtained by CW Krypton laser excitation; temperature cannot exceed about 60K and the thermal behavior for threshold is linked with a phonon direct process reducing the T2 of the emitting lowest Stark level (see Fig 9)[14]; integrated output signal is quadratic with pumping above threshold. All these facts point to a superfluorescence or super-radiance effect in the DickeBonifacio sense[2], that is to a collective effect of excited states for which the principal condition is the existence of a T2, the dephasing time, larger than the superfluorescence lifetime. On the contrary, conditions on T2 are not a prerequisite for ASE. This can be understood simply by remarking that lasers exist at room-temperature showing that T2 play no decisive role in a laser threshold .On the other hand, one knows that large T2 are
447
linked with low temperature and weak concentration. This is opposite to researching large gain in optical short paths as for ASE amplification. By direct excitation into the In/2 of Er, with a Ti-Sapphire CW laser, we have recently obtained the superfluorescence effect ( super-radiance ) down to grain diameter of 30u.m on a single layer of grains [22]. Fig. 10 gives the emission spectrum below and above threshold which at 9666 A, is 827 W/cm2 for T=l IK.
Figure 9. Threshold temperature dependence for superfluorescence of LiYF4 :Er. Activation energy corresponds to the first Stark splitting of 4In/2. The effect cannot be obtained above 60K. After [14]
2700
2750
2900
Figure 10. Emission spectra below (a) and above threshold (b) of a single layer of LiYF4 : Er powder with grains diameter of less about 30um Threshold is 827W/cm2 under Ti-Sapphire excitation at 9666A. After [22].
448
984965988967968968970 971 972973 974 975 976977
Wavelength (nm)
Figure 11. Excitation spectrum for superfluorescence at 2.7um in a LiYF4 iEr3* powder. After [22].
The observed excitation spectrum of Figure 11 reveals the six Stark levels of \u2 in the electric field of the local site of Er in LiYF4 plus two replicas at 9742 and 9755 A corresponding to the \5n first Stark level slightly populated at 1 IK. Chopping the pump beam with a long period of 80 ms, below and above threshold, allows the observation of the temporal behavior of the super-radiant emission, see Figure 12.
Figure 12. Dynamical behavior of the 4In/2 emission below (lower trace) and above threshold (upper trace). Trace with dots is the pump. Note superfluorescence peak. After [22].
449
l11/2 population (a.u.) Figure 13. Integrated intensity of the superfluorescent pulse from *lun of Er3*, versus relative population density measured by the spontaneous intensity level. Note quadratic fit, showing that superfluorescence is taking place rather than ASE. After [22].
Below threshold, only the usual spontaneous emission is observed; above threshold, a much shorter spike is observed the intensity of which shows a quadratic behavior shown on Fig. 13. Though the threshold power density is about the same for the large grain and the small grain cases, the fact that now we observe super-radiance on the powder with a smaller diameter grain than before, is attributed to the fact that the pumping photons are now directly used to create the excited state density. With the Argon laser excitation of ref. [13]), the ^9/2 state was first excited at 5110cm"1 above the useful excited state ; this energy had to be dissipated as phonons so heating the sample. Also the self-phasing of the spontaneous emission had to start from itself (super-fluorescence) whereas now the phasing of excitation can be used to start the phasing of emission from the lowest emitting Stark level (super-radiance). All these explain the better efficiency of the TiSapphire excitation we observe now. In particular the fact that at a concentration of 2.1 10 cm'3, super-radiance cannot be observed nor in powder nor in bulk whereas this is the optimized concentration for cw laser[24], clearly demonstrates the difference between the two processes described above in chapters 3.1 and 3.1.3
450
3.1.4 Discussion. Following ref. [7], we can estimate the average necessary gain path la to explain the line narrowing observed in laser effects in powder. It is given by: l
'~a*8vlhv
(61)
where a0 is the maximum gain coefficient, 8v the laser emission width, Av the line width below threshold. In ref.[7], with Svl Av=0.18, 0^=10cm"1, the average gain path by (3) has to be at least 2.3 cm. In our own case of powder laser[8], we had Svl Av = 5.3 10*4, ao=800cm" giving /a= 1.63cm above threshold. In both cases, one finds the same order of magnitude for the necessary path length. In ref. [7] it was concluded that the laser effect was taking place with several grains along a close ring path each grain keeping the photons for a few total reflection paths within the micro-cavity constituted by the grain. Such close rings of cm size are clearly not compatible with the close rings advocated in strong localization of light[24], because in the recent first demonstration of strong localization[25], the requested scattering mean free path, 1, is such that: , 1.5A l<
~^Tn
(62)
where -Jn is the average index of refraction of the composite material made of air and particles of index n. For rare earth doped materials this comes to l<0.\9\m\ which is much less than geometrically allowed for grain diameters between 5 and 50 pm at close contact. In our view a more realistic way to see the laser effect in powder is to consider the " time reverse " paths considered in CB as the effective closed rings necessary to explain the observed spectral narrowing. By analogy with the theory for a scattering medium with absorption [19], we consider the scattering angle 0a corresponding to a given gain length, /„, for a weak localization loop without dephasing :
For small angles such loops may be very long. For example, taking a scattering length 1=10^, A,=1.06n, a single path gain of 1cm is compatible with an angle #a of 0.92 mrd, typical value for CB experiments with gain (see Fig.7). On the other hand, all laser experiments scheme described to date on rare-earth doped powder laser[7,8], have in fact made use of a back-scattering detection scheme with an optical fiber. Then clearly 0„ is within the accepting angle of the detection path. For all these reasons we like to propose that powder lasers have to involve closed ring paths of the " time reverse " type. In order now to discuss the super-radiance case, we have to first recall the dimensional conditions for super-radiance[13]. They are the following:
451
L r « L « Lc (64) where LT is a threshold critical length, L is the sample length, Lc is the ArecchiCourtens cooperative length[26]. In case of pulsed pumping, LLC, ASE would overcome super-radiance and output should be proportional to p. In ref [28], we had found, for the same material, a super-radiance decay time xR<150ns. We can estimate Lf=LTR/T2 [29]. It means that the left-hand side of equ.(64) will be fulfilled for TR 150ns. This condition should be respected since T2 values of about 300ns to 2us have been reported for same host with various rare-earth dopants[30]. The right hand part of equ.(64) is always fulfilled in our case [13]. This discussion shows that, for our super-radiance experiment, individual grains of the powder are involved, and that scattering paths between different grains are not necessary. Not looking for gain paths, in turn explains the low acceptable concentration. Also, a large T2 necessitates a low concentration and a low temperature to avoid dephasing in the photon-ion interaction due to ion-ion and ion-phonon interactions. 4 Anti-Stokes emissions of light or up-conversion processes in RE-doped solids. 4.1 The APTE effect or sequential energy transfers between excited states. When active ions are situated at a sufficiently short distance for interactions between them to take place, two types of up-conversion processes may occur: i) summation of photon by energy transfers, [31], called the APTE effect (for Addition de Photon par Transferts dEnergie) [32]; ii) cooperative effects either by sensitization [33] or emission [34]. Both types, i and ii, are often mistaken one for the other because they present several similarities and may be simultaneously present in a given system for a given excitation. Until 1966, all energy transfers between R.E. ions were considered to be such that the activator ion receiving the energy from a nearby sensitizer (S), was in its ground state before interaction. Then it was proposed [31] to consider cases where activators (A) were already in an excited state as shown on fig. 14. This is evident afterwards because one has to realize that we can exchange only energy differences between ions and not absolute energy. The reason for proposing such an up-going transfer was to point out that energy transfers then used [35] to improve the laser action of Er3+ by pumping Yb3+ in a glass matrix could also have the detrimental effect to increase reabsorption [31,36]. The simple proof of such an effect was to look for an up-converted green emission (from 4S3/2 of Er3+) while pumping Yb3+ (2F7/2-»2Fs/2), which was effectively observed [14,20]. Of course the situation in fig. 14 could repeat itself several times at the activator site. This means that n-photon up-conversion by energy transfer is possible as demonstrated
452
by the 3-photon up-conversion of 0.97 um into blue light (0.475 um) in the Yb3+-Tm3+ couple [31]. W, Jtresonant transf. in excited state 'I P-a = 0
S
I
> I - —« '
phonon-assisted transf. in excited state | P-a = c„
|P
S
A
provided WSA > 1/t. Figure 14. Energy transfers towards an activator already in an excited state: the basic process for upconversion by energy transfer; effect called APTE (Adition de Photon par Transfer d'Energie)[32].
Independently, such IR->blue up-conversion was interpreted by Ovsyankin and Feofilov [33] as a 2-photon effect connected with a cooperative sensitization of Tm3+ by two Yb3+ ions. In fact, probably because of the saturation in an intermediate step, they had observed a quadratic law for emission versus excitation density instead of a cubic one. In order to clarify the terminology, a schematic comparison between the APTE effect and other 2-photon up-conversion processes namely: 2-step absorption, cooperative sensitization, cooperative luminescence, second harmonic generation (SHG) and 2photon absorption excitation is presented on Fig. 15 , together with typical efficiencies. Since we are dealing with nonlinear processes, the usual efficiency, as defined in percentage, has no meaning because it depends linearly on excitation intensity. Values are then normalized for incident flux and given in (cm2/W) units.
cooperative luminescence
S.H.6, 2-photon absorption excitation
YF,:Yb:fb
Yb PO,
KDP CaF2:Eu2+
n»10'«
n=10'\
n*io-" n=io-"
APTE effect
2.steps cooperative absorption sensitization
YF,:Yb:Er
SrFj:Er
n=lfr'
S
n=10-
Figure 15. Energy schemes for different n-photon up-conversion processes with quantum efficiency in cm2/W for specified materials. After [32].
453
In confined structures, because some of the most efficient processes may start saturating due to the very high pump density, lower efficiency processes may dominate. A simple inspection of the energy schemes involved shows that they differ at first sight by the resonances involved for in- and out-going photons: for the highest efficiency, photons have to interact with the medium a longer time which is practically obtained by the existence of resonances. As shown, the APTE effect is the most efficient because it is the process which most approaches the full resonance case; consequently it is the more likely to be encountered. Many times in literature up-conversion involving coupled ions are referred to as cooperative effects [38,39,40] without demonstration, when in fact as can be guessed from the relative positions of involved energy levels, APTE effects are involved [41,42]. The fact that the APTE effect and cooperative ones are often mistaken one for the other is linked to a number of common properties. For instance, for 2-photon up-conversion, both processes show a quadratic increase on excitation and on absorber concentrations; both show an emission lifetime equal to half the absorber lifetime. However, as shown below, the difference is more basic, though sometimes difficult to establish experimentally except in special cases: when single ion resonances clearly do not exist or when diffusion between ions is prohibited by a too small concentration with still an interaction as in clusters of R.E. ions. 4.2 Up-conversion in single ion level description and in a pair-level one for cooperative effects. As seen in the introduction, up-conversion by energy transfer is just a generalization of Dexter energy transfer [43] to the case of the activator being in a metastable state instead of being in its ground state; this requires that the interaction between S and A (HSA) be smaller than the vibronic interaction of S and A in order that both ions be described by single-ion levels coupled to the lattice. It is generally the case since for fully concentrated RE crystals or for cluster, pairs level splitting is of the order of 0.5 cm"1 [44]; in a host with a smaller concentration this interaction can even be weaker, whereas one-phonon or multiphonon sidebands may modulate the level positions by several 100 cm'1. Further, up-conversion requires that the transfer probability for the second step WSA be faster than radiative and non-radiative decay from the metastable level that is WSA > v * with xe, the measured lifetime. WSA is obtained from
WSA = ^ | < ¥s* A |H S A |*PX >fp(E), n ' '
(65)
where the wave functions are simple products of single ion wave functions ; p(E) describes, the dissipative density of states due to the coupling with the lattice.
454
On the contrary, all cooperative processes including up-conversion can be considered as transitions between a pair-level for both ions as a whole. A dipolar electric transition would be forbidden for such a two-center transition and one needs product wave functions corrected to first order to account for the interaction for electrons of different centers [45], M |00>. l^>=l^ o (s)«p o M))-i2: <s"a"|H %,(S)Va,(A)) £„-o+*,„-o'
,
(66)
for example for the ground state; s", a" denotes intermediate states for S and A; 5^, ea, denote their corresponding energies. Then any one-photon transition in the cooperative description involves already four terms in the matrix element which cannot be reduced to eq. (65). Figure 16 schematically presents tliese four tenns for the simplest case of cooperative luminescence of Fig. 15. The 4 terms in+ Dm, I < „ > : perturbations initial (ej
H„ H„ •s"
_J
D','L, ^
Is"
LJ_
1 2 a"-
^_ 1
2
_i_ j ^_ _m 1 1 2
V
2
1
T 1
final (o)
2
1
2
•-s"
-#~ —#—
III
—>hv
_; 1
2
—Y
1
2
1
2
cY'"'
-m—i-
-hv
IV
[_
;>
4 1
/" I i
Figure 16. The four terms of the two-center electric dipole matrix element for cooperative luminescence- S and A of equ. (66) are respectively here identical ion land ion 2. Because of the symmetry between 1 and 2 , terms I and II as well as III and IV are physically identical.
455
APTE up-conversion does not correspond to the same order of approximation as cooperative processes: the later have to be considered practically only when the first type cannot take place. Such is the case when real levels do not exist to allow energy transfers ; this is the case for Yb3+-Yb3+ in YbP04 crystallites [34] or Yb3+-Tb3+ up-conversion [44] or when concentration is to small to allows efficient transfer by energy diffusion between sensitisers. Then cooperative up-conversion is likely to occur within clusters [46,47]. One may also look for crystal structures where the pair clustering is built in [48]. As an example Figure 17 shows the cooperative emission of Yb3+ as created by Yb3+ clusters in a phophate glass for various doping precursors [49]. In order to illustrate the difference between APTE and cooperative up-conversion we shall discuss an example of a line-narrowing effect in n-photon summation as a mean to distinguish between both processes [41,50], Irradiating Er3+ doped samples with IR photons at 1.5 um leads to various visible emissions . Room temperature IR F-center laser excitation between 1.4 and 1.6 urn of Er3+: YF3 leads to emission bands from near IR to U.V. Such emissions may be ascribed to multiphoton excitation respectively of order 1 to 5, either by the cooperative or by the APTE type as depicted respectively with pair energy levels or with levels of single ions (APTE) (Fig. 18 a and b).
500
520
wavelength (nm)
Figure 17. Cooperative luminescence at 500nm of Yb3+ clusters in a phosphate glass of constant composition and doping concentration but for different doping precursors. Relative intensities reveal the relative amount of clustering. Bands in the 520-550nm range is APTE effect from Er impurities. After [49].
456
"EflO'cm-') 4
S 3 ,2 '1,3,2-*fvi r S/2
20
HIMJ
*--'-H«
Il5/2-»l 13/2 1,5/2-*"
'1,5,2
i*-*F 5 «
*I,5,2
^7/2
*Il5,2
'*" San
'ism
'1,1/2 ' 1,3/2 r
'un Inn-*; 10
11112
l,5fl
9,2
M5,2
—'1*2
'1,5/2
-
1,3/2
Il5,2
'1,5,2 'l,5/2 * T H I - t (a) Figure 18. (a) is energy scheme for the cooperative hypothesis with successive absorptions between "pair" levels; (b) is energy scheme for APTE effect. After [41].
Successive absorptions in Fig. 18a involve a combination of several J states. The APTE effect, because of self-matching by multiphonon processes, involves only J = 15/2 and J = 13/2 states (Fig. 18b). Excitation spectra depicted in Fig. 19 show a striking behavior: each spectrum presents the same spectral structure, but an increasing narrowing is clearly observed with process order. The structure reproduces the Stark structure of the 4Iis/2 —>4Ii3/2 first excited terms as can be obtained by a diffuse reflectance spectra. This then is a direct proof of the validity of the APTE explanation, since a cooperative effect should show the convolution of all J states involved in the multiple absorption between pair-levels [50].
457
Figure 19. Excitation spectra for up-conversion of order n; Xa„ is the wavelength at which the detection is operated. After [41].
The spectral narrowing can be understood by a rate equation treatment where higher excited populations are neglected with respect to the lower ones in order to obtain a tractable development (weak excitation assumption). The emitted power from an n-photon summation is then given by : W
W
(67)
(°Vl)-G>2)
with symbols of Fig. 18, and where Pj (X) is the line shape of 4Iis/2 -»4In/2 absorption [41]. 4.3 The photon avalanche effect in RE-doped solids. While looking for 2-steps absorption (ESA) in Pr3+ doped LaCb and LaBr3 at low temperature (<40K) as a mean to detect an IR photon by its energy summation with a more energetic photon so performing excited state absorption (ESA), it was found that
458
the higher energy photon alone could, in the same time, give rise to up-conversion and reduce the transmission of the sample above a given intensity threshold [51]. The effect was attributed to an increase of population on an excited state resulting from a crossrelaxation process. The starting process was initially not completely determined. In the Pr3+case, the 3Hs ->3Pi absorption is initially very weak at low temperature because 3Hs is about 2000 cm-1 above ground state (see Figure 20); however above about 1 mW of excitation this transition is increased ; the cross-relaxation (3H6, 3H4) -K3Hs, 3H5) increasing the 3Hs population which in turn reduces the transparency of the sample at the (3Pi-3H5) energy. Since the more the (3Pi-3Hs) energy is absorbed the more the 3Hs population is increased, the process was termed "photon avalanche"[51]. It is clearly a way to increase ESA in a sample.
non resonant pump(1) resonant pump (2)
cross-relaxation
i-
S
H Figure 20. Energy scheme and cross-relaxation process for avalanche in LaC^Pr3*. After [51].
Afterwards, similar effects have been observed in Sm3+, Nd3+, Ni2+, Tm 3+ doped halide crystals [52,53,54,55]. Recently we obtained the photon avalanche effect at room temperature for the Er3 + ion in a ZBLAN glass both in bulk and in fiber form [5659] and in a IJYF4 crystal [60]. There are three distinct aspects for the photon avalanche process non-linear behavior: i) transmission, ii) emission and iii) rise time on the pump power intensity with generally the existence of a critical pump threshold. Particularly long rise times, from seconds to minutes [60,61], have been observed.
459 At this point it is worth discussing the notion of threshold for avalanche. Because of the complexity of the phenomena, it has been usually modeled by a simplified threelevel system [62,63,64] 4.3.1. The avalanche process as a positive feedback system. Using the 3-level simplified model of [62] or [63] and adding to the initial ground state absorption (cr,m) we may write the following set of equations (see Figure 21 for explanation of symbols which except for the trigger a0OIR are the same as in [62]). Being interested in the steady state initial step of avalanche we assume : v dnx = dn2 = dn3 ^ J ' dt ' dt dt H ' ) « , = l - n, 1 Then the rate equations are simplified to : 0 = -Rj - a 0 O K + bW3n3 + W2n2 - Cn3 0 = (1 - b)W3n3 - W2n2 + 2Cn3 + R, + a0
i
iFR,
3,n,
L
1
i
' 4
o.®» 0**i
2, n,
\
Cn,n.
' 1,n, Figure 21. 3-level energy scheme for avalanche
(68) (69) (70)
460
with C the cross-relaxation parameter and b the branching ratio, the following relationships exist between the transition probabilities: (1 - b)W3 = W32 ; bW3 = W„ ; W3 = W32 + W31 Equ.(70) can be written : R, W 3 +C
(71)
Or considering amplitude and using the symbolic representation for feed-back systems equ.(71) gives block A of fig 22-a. In the same way, equ.(68) is written : ^ R . + q ^ C - b W ^ W,
w,
(72)
R2 W,+C
(R,+,R)
w,
C-bW3
w,
-=p
(a) Figure 22. Symbolic feed-back system "black-box" representation of equation system (68-70) (a) and gain behavior versus R2; R2c is threshold value (b). After [60].
which can be symbolized in Fig 22-a by block p and an adder with input source {Rt+a^m)/W2 Combining (72) and (71) gives the classical feedback systems scheme of fig 22-a. Such system is known to be unstable for: AyS=l. One can define a "gain" of the closed loop feed-back system ,G, by the ratio between the green output to a pump related input signal (Rj) plus an eventual trigger (O m ): G=-
(R.+Oo*,,)/^ It is well known that in term of the A and p of Fig 22-a, one has :
(73)
461
G=— ^ 1-A>9 The stability condition is then written:
JR^C-bW^ w 3 +c w2
(74)
(75 )
its limit is just the threshold condition given by Joubert et al. [62] obtained here in a simplified way: _W2(W3+C)
* * - c-bw3
(76)
with C > bW, for a positive feedback. The behavior of our feedback system below threshold can be described by the behavior of G(R2), Fig 22-b. The feedback "black box" approach has also been considered in studying the dynamics of the above three levels system [65]. It is based on the fact that the general feedback linear theory used to solve algebraically time variable differential equation systems by using the Laplace transform of the time dependent functions. 4.3.2 Necessary conditions to observe a threshold. With neglect of the first non-resonant absorption step (R^) and taking into account only the second resonant absorption step (R2) when calculating the population of the third level (03) versus R2; (the pumping excitation), leads to a well defined nonlinearity in n3 for the asymptotic curve ( R ^ K ) ) , as shown on Figure 23. When the first step (R1) is explicitly taken into account [64], the "threshold" nonlinearity is progressively smooth out when the ratio Rj/R2 is increased as shown on Fig. 23. This corresponds to a progressively more resonant first step. Practically a clear avalanche threshold can be expected only for R)/R2 ratios < 10"4 [64]. Some of the features of the avalanche effect have been observed at room temperature in Tm3 + : YAIO3 [19] and in Pr3 + in silica glass fibers [66]. The lack of a clear threshold in these two systems can certainly be related to the above prediction. The region on Fig. 23, where 10"4
462
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
Figure 23. Normalized output for avalanche up-conversion versus normalized pumping and for different Ri/R2 values (here p) as parameter. After [64],
i
-
^
N
-
v
4
,i E c
.
8 CO
Onm
579
E
E lO
CO
E
690n
1
^
*i i
\ '\
k
\' \
i i
4
i
A
F„
U 4 s* F„
1i
\ ' \
I
'
\
i 'r
Figure 24. Er*+ levels involved in the avalanche processes under 579 and 690nm excitation.
463
Up to now, only Er 3+ ,with (R^/R^slO-G, has shown at room temperature, all the three characteristic features of avalanche as well when doping a LiYF4 crystal or a ZBLAN fluoride glass both in bulk and in a fiber shape (see 4.3.4.); even the long delay of several seconds to minute was observed [56-60]. The involved energy schemes for Er3+ avalanche processes are presented on Figure 24. For comparison, the following values for the critical parameter have been given: for Nd3+-LiYF4, (R,/R2) =1.7 10' 4 for avalanche at T<40K [53,62]; for Tm 3+ -Ho 3+ Gd2Ga50i2, (R1/R2) =3 -6 10_2 for the two-ions looping process [69]; for a Tm 3+ BIGaZYTZr glass, (Ri/R2) =1.2 10"2 [70] for what was claimed to be "avalanche" at 100K. In this last case because the delay reaches only sixteen times [70] the metastable state lifetime (W2"*), we believe it to be a looping process case. In Er 3+ , as it shall be possible to see from 4.3.3 and 4.3.4 , the avalanche delay reaches 6 102 to 104 times W2_l respectively for Er-doped fluoride glass and crystal. 4.3.3 Experiments in Er^^-LiYF^ In the case of Er3+> the first step for photon avalanche has been clearly identified and attributed to anti-Stokcs multiphonon sidebands absorption [56,60] (sec Figure 24). Calculating the (R1/R2) ratio from mutiphonon absorption, allows to estimate a value of 5 10-6 [4;5]; from Fig.25 which displays the multiphonon side-bands absorption in the avalanche excitation regions. As observed, such experimental situation provides a clear threshold behavior in the Erbium case [56].
'500
550
600
650 700 750 WAVELENGTH (run)
800850
Figure 25. Absorption cross-section for Er3+:LiYF4 taking into account the multiphonon contribution; the heavier lines show the anti-Stokes zones which contribute to W) for the avalanche process in Er3 \ After [60].
464
The simple theory of 4.3.1 has been verified by experimentally measuring G(R2) (see Figure 22b). This was done using the following method (see the experimental set-up on figure 26): having obtained a given green output for a pump R2 with <1>,R = 0, R2 is reduced while increasing O m in order to maintain a constant green output. O m is an infrared signal at 0.94 um in resonance with the metastable state 2 (here "Ina) absorbing with a cross-section cx0. Each point is obtained after waiting for a steady state. Because of the large ratio for a^la^ this experiment provides a good description of G(R2), as shown by results on Figure 27 for three temperatures. We can define an R.2C asymptote only at 300K and 220K, respectively 120 mW and 240 mW. At 163K one cannot reach asymptote at our maximum available power of 250 mW. So the effect of lowering temperature is essentially to increase R2C . monochromator
}-<S^signal
chopper
Ar* 4.W- - Coherent innova 90
Dye
laser Spectraphysics 375
4}
—Tchopper
J
•t-* Til
LiYF4 / Acryostat Er3+sample 10-300 K
pump at 0.58 and 0,69 /jm
Ar* lOW Spectraphysics _ 171
Ti3+AI2Cv laser
silica fiber
Coherent 899
trigger at u"94/vm Figure 26. Experimental set-up for recording GCR2) and for the triggered avalanche experiment. After [60]
465
160
100 150 200 PUMP AT 578nm (incident mW)
250
Figure 27 Experimental G(R2) for three temperatures: 300K, 220K, 163K, the residual signal near R2=0 comes from the direct up-conversion under 0.94(m After [60].
The part in R2C which is most sensitive to temperature is C because it is related to phonons energy of only 100 cm"1 whereas W2 and W3 are related to phonons covering the energy gap below level 3 and 2, that is energies > 2000 cm1. However this requires C be of same order as W3 or bW3, otherwise, as long as C » W3, bW3, one has R2C = W2 and its temperature dependence is just the same as W2. Comparing theoretical threshold as given by equ.(76) with our experimental conditions, we can verify our simple model. We assume level 1 to be 4Iis/2, level 2 to be 4In/2 and level 3 to be the aggregation of levels between 2Cfe/2 and 4S3/2 with the emission properties of 4S3/2 (see Fig.24). Taking the room temperature values given by [71,72] we have the following parameters: W3 = 2500 s-1 ; b = 0.5 ; C = 0.5 106s-1;W2 = 140 s"1; because corresponding oscillators strength are about equals (« 0.4 10"6) [72], we assume 21 2 o 2 = a 0 = 4 lO"-" cm car4. Using reduced population units (pure number): we have R2C =140(2500 + 510 5 )/(510 5 -1250) = 141s"1 = W2 (at room temperature) from which we estimate: * * - * * = 141/410-21 = 3.51022s-'cm-:
466
SflUt
green output
>1B«U
Sfimil
5»
green output
<1B«U
SftnPLE
?»
green output
Figure 28. Avalanche behavior just below the threshold without (a) and with (b) a trigger at 0.94um feeding the metastable state and (c) much below the threshold. After [60].
At 0.578 urn it gives, for a 50 um diameter spot, a threshold power of Pt,, = 222 mW; this value is of the same order as observed threshold values for 0.578 um pumping. G(R2) shows (Fig.22) that we have a "marginally stable" positive feed-back system: even below the R2C asymptote, we know from feed-back system behavior that a strong input signal can drive the system otherwise stable into its instability state (existence of a "gain stability margin"). In order to verify this behavior we experimentally consider the case where, besides the input signal R1/W2 given by the pump, we add a pulsed trigger of amplitude CT OG>IR / W2; the experimental set-up is the same as already presented on Figure 26. The results at room temperature are given on Fig 28: In absence of trigger, with PP = 114 mW at 578 nm incident on sample, threshold is reached after a very long time
467
(> 50 s) (Fig 28-a). With same pump intensity (PP = 114 mW) and with a short trigger of 0.6 s, the avalanche state is obtained more quickly and maintained after trigger extinction (Fig 28-b). With same trigger but with a reduced pump (PP = 99 mW), the avalanche state is not reached. The behavior on Fig .28 a-b-c is obtained down to 180K. At temperature lower than 180 K, observed threshold certainly increases (Fig. 27), however our temperature scan having a relatively short time constant (3s / K from 10 to 50 K, then 21s/k from 50 to 150 K), we are not sure that threshold could not be reached for avalanche delay time>50s. From this experiment, we understand that measuring an avalanche threshold depends on the time we are ready to wait for before its observation. This time depends not only on excited state pumping but also on the ground state absorption conditions. In any case, below 180K, being then limited by the pump laser at a much lower power than threshold, we obtain the result of fig 28-c. Such triggering effect realizes an optical analogue of a thyratron. The time delay behavior of the avalanche process, has been studied theoretically quite recently within the general model of Landau for phase transition [70].The time delay at threshold, in fact a "critical slowing time", IQ, proved itself to be the most sensitive experimental data when looping or avalanche takes place. It has been shown to be given by[70]:
te=K^577^7
(77)
where K depends on other spectroscopic parameters. Eq.(77) is rather well verified in our experiments for which avalanche delay times have been determined for two different excitation wavelengths of known multiphonon anti-Stokes cross-sections: at X=688nm, with ai=10"24 c m 2 ; delay is found to be 0.4s; whereas at A,=579nm, with a\=2 10"26, the observed delay is 4s. Assuming for c*2 the same value in both cases of excitation, the ratio of delays is 0.1 which is rather well in accord with the value of 0.14 as given by equ.(77). Above threshold, the delay for avalanche has been given by [73] as: tc = k W 2 - ' / ( 0 / O c - l )
(78)
with k = W,"1 / (1 + b)(2c + W3), and where O c = Rc/02 is the pumping flux at threshold and O, the effective pumping flux. 4.3.4. Photon avalanche in Er^+-Fluoride glasses infiberand bulk shape. Recently, the photon avalanche effect has been observed in a Pr3+-doped silica fiber [66] and in an Er3+-doped fluoride glass fiber [57,59] in both cases at room temperature. In the first case, only the transmission non-linearity is observed and not the up-conversion emission threshold. It was believed that the threshold was so low that it could not be observed .We think that this is explained by the too strong non-resonant to
468
resonant absorption ratio as mentioned in 4.3.2. On the opposite in the second case, clear thresholds at 5mW and at 4mW of incident power at respectively 579nm [59]and 690 nm [57] are observed because in these last two cases, the first step is a weak anti-Stokes multiphonon absorption giving again a (R1/R2) ratio of about 1(H> much below the critical value of 10"4. The involved energy scheme for both excitations is essentially the same as on Fig.24; it shows both pumpings and the two types of involved crossrelaxations. Figure 29a presents the typical threshold behavior for the up-conversion emission. The long delay behavior is displayed on Figure 29b showing, near threshold, the very long time of 3.5 s. which is largely in excess of any of the life-times of the metastable states of Erbium. The observed delay follows rather well the behavior predicted by eq.(78). Same results can be obtained for glass and bulk samples [56,57]. Because the first absorption step, being of a multiphonon nature, is featureless, the excitation spectrum for
I'-
.
T"
1
-
-1
-V
•\ • \ - \ - \
ll)
v.
l"**
20
40 60 80 Pump power (mW)
100
20
•*
—J
m
, 1
40 60 60 Pump power (mW)
.
100
Figure 29. a) Up-conversion at 550nm .showing the avalanche threshold in a ZBLANiEr3* fibre observed from its extremity; b) is time delay for avalanche establishment versus incident pump power at 579nm in a ZBLAN:Er3+ fibre. After[49].
avalanche provides directly with a single excitation beam the ESA spectrum of the resonant second absorption as shown on Fig.30 for the 4In/2-^G9/2 transition of Er 3+ ; this gives a new method [58] to reach ESA spectra otherwise difficult to obtain.
469 Excitation spectrum for avalanche.at 549mm
Figure 30. Excited state absorption (ESA) spectrum for transition %m-2Gm of Er3+obtained with a single beam from the avalanche emission at 549nm excitation spectrum. After [48],
The main difference with bulk samples, is that in the fluoride fiber case, upconversion periodic spatial domains appear with periods ranging from few cm to mm and 100 p a [58,59], as shown on Fig.31,
200 pro
' * J ' ', s&&
Figure 31. Negative side view of the ZBLAN:Er fiber under avalanche regime. The 4 S 3 2~>4Io/2 green emission appears as black dashes; (a) shows the larger structure with average period of 824nm; (b), (c) and (d) show the iner structure with average period of 133,lum when respectively pumped at 579 12 579.32, and 579.53nm. After [75].
470
This behavior has been explained by the high contrast provided by the avalanche effect; It then optically reveals the internal electric field modes beating structure of the fiber waveguide[74, 75], 5. Conclusion In this lecture, we have presented a number of more or less not yet generally known emissive properties of rare-earth doped solids. They all are based on recognized basic processes though they may appear at variance with some famous physics laws: Stokes' law and Einstein's relationships. Of course such laws are not violated; they have limits in their own field of application. Outside such limits we have what has been called here, "unconventional" emissions: When stimulated and spontaneous emissions are still mixed (at times
References: 1. F.Auzel, in Properties of highly populated excited states in solids: superfluorescence, hot luminescence, excited state absorption, in: Optical Properties of Excited States in Solids, ed. B.Di Bartolo, (Plenum Press, New York, 1992), p.305. 2. R. Bonifacio and L.A.Lugiato, Phys.Rev.A, 11, (1975), 1507. 3. V.S.Letokov, Soviet Physics JETP,26,(1968),835. 4. F.Varsanyi, Applied Physics Let.,19,(1971),169.
471
5. V.M.Markushev, V.F.Zolin, and C.M.Briskina, Sov.J.Quatum Electron. 16,(1986),281. 6. V.M.Markushev,N.E.Ter-Gabrielyan, C.M.BriskinA, V.R.Belan, and V.F.Zolin, SovJ.Quantum Electron. 20,(1990), 773. 7. N.E.Ter-Gabrielyan, V.M.Markushev, V.R.Belan, C.M.Briskina, O.V.Dimitrova, V.F.Zolin, and A.V.Lavrov, SovJ.Quantum Electron. 21,(1991), 840. 8. C.Gouedard, D.Husson, C.Sauteret, F.Auzel, A.Migus, IAEA Conf. On Drivers for Inertial Confinement Fusion, Osaka, Japan, 15-19 April (1991), paper E04, and J.Opt.SocAm. B, 10,(1993), 2358. 9. M.A.Noginov, N.E.Noginova, HJ.Caufield, P.Venkateswarlu, T.Thomson, M.Mahdi, V.Osrroumov, J.Opt.SocAm. B, 13, (1996), 2024. 10. N.M.Lawandy, R.M.Balanchandran, A.S.L.Gomez, and E.Sauvain, Nature, 368, (1994)436. 11. J.Sageev andP.Gendi, Phys.Rev.A, 54, (1996), 3642. 12. D.S.Wiersma, M.P.Van Albada, and A.Lagendijk, Phys.Rev.Let. 75,(1995), 1739. 13. F.Auzel, S.Hubert, and D.Meichenin, Europhysics Let.l, (1988), 459. 14. S.Hubert, D.Meichenin, and F.Auzel, J.Lumin. ,45, (1990), 434. 15. F.Auzel, Processes in Heavily Doped Rare-Earth Materials and their Applications to Optoelectronic Devices, in .Semiconductor Optoelectronics, ed. M.A.Herman, Wiley NYC, 1980. 16. M.P.Van Albada and A.Lagendijk, Phys.Rev.Lett.,55, (1985, 2696. 17. P.E.Wolf, and G.Maret, Phys. Rev. Lett., 55, (1985), 2696. 18. P.E.Wolf, G.Maret, E.Akkermans, and R.Maynard, J.Phys.France, 49, (1988), 63. 19. Y.Kuga and A.Ishimaru, J.Opt.Soc.Am.,1, (1984), 831. 20. A.Y.Zyuzin, Europhys Lett., 26, (1994), 517. 21. D.S.Wiersma and A.Lagendijk, Phys.Rev.E, 54, (1996), 4256. 22. F.Auzel and P.Goldner,in Advances in Science and Technology, Vol.27, Innovative Light Emitting Materials, ed. P.Vincenzini, and G.C.Righini,(Techna, Faenza, Italia, 1999), p. 113. 23 L.O.Schwan, J.Lumin.,48/49, (1991), 289. 24. F.Auzel, S.Hubert, and D.Meichenin, Appl.Phys.Lett.54,(l9S9), 681. 25. P.Sheng, Scattering and Localization of Classical Waves in Random Media, (World Scientific Singapore, 1990). 26. D.S.Wiersma, P.Bartolini, A.Lagendijk, and R.Righini, Nature, 390, (1997), 671. 27. F.T.Arecchi and E.Courtens, Phys.Rev.A 2, (1970), 1730. 28. F.Auzel, Properties of highly populated excited states in solids : superfluorescence, hot luminescence, excited states absorption, in Optical Properties of Excited States in Solids, Ed. B.DiBartolo, Plenum Press, NYC (1992), 305. 29. H.M.Gibbs, Q.H.F.Vrehen, and H.M.J.Hikspoors, Phys.Rev.Lett.,39,(l977), 547. 30. G.K.Liu, M.F.Joubert, R.L.Cone, and B.Jacquier, J.Lumin.,38,(l987), 34. 31. F.Auzel, C.R. Acad. Sc. (Paris), 262, ,(1966)1016 and 263, ,(1966)819. 32. F.Auzel, Proc. IEEE, 61,(1973), 758. 33. V.V.Ovsyankin, and P.P. Feofilov, Sov. Phys. JEPTLett., 4,(1966), 317
472
34. E.Nakazawa, and S. Shionoya, Phys. Rev. Lett., 25,(1970, 1710). 35. E.Snitzer, and R. Woodcock, Appl. Phys. Lett., 6,(1965), 45. 36. F.Auzel, Ann. Telecom. (Paris), 24,(1969), 363. 37. F.Auzel, and O. Deutschbein, Z Nat. (Germany), 24a,(1969), 1562 38 J.P.Van der Ziel, L.G. Van Uitert, W.H. Grodkiewicz, and R.M. Mikulyak, J. Appl. PAjw., 60,(1986), 4262. 39. G.J.Kintz, R. Allen, and L. Esterowitz, Appl. Phys. Lett., 50,(1987)., 1553 40. E.Desurvire, J.R. Simpson, and P.C. Becker, Optics Lett., 12,(1987)., 888 41. F.Auzel, J. Lumin.., 31/32,(1984)., 759 42. F.Auzel, S. Hubert, and D. Meichenin, Appl. Phys. Lett., 54,(1989)., 681 43. D.L.Dexter, J. Chem. Phys., 21,( 1953), 836 44. V.V.Ovsyankin,, "Spectroscopy of collective states and cooperative transitions in disordered rare-earth activated solids", in "Spectroscopy of Solids Containing RareEarth Ions", Eds A.A. Kaplyanskii and R.M. Macfarlane, (North-Holland Amsterdam 1987.), p. 405. 45. M.Stavola, and D.L. Dexter, Phys. Rev., B20, (1979), 1867 46. J.C.Vial, R. Buisson, F. Madeore, and M. Poirier, J. de Phys., 40,(1979), 913. 47. F.Auzel, D. Meichenin, F. Pelte, and P. Goldner, Optical Mrf.,4,(1994), 35 48. N.J.Cockroft, G.D. Jones, and R.W.G. Syme, J. Lumin., 43,(1989), 275. 49. F.Auzel and P.Goldner, Material Science Forum, 315/317, (1999), 34. 50. F.Auzel, Rare-Earth Spectroscopy, Ed. B. Trzebiatowska, J. Legendziewicz and W. Strek, (World Scientific, Singapore 1985), p. 502. 51. A.W.Kueny, W.E. Case, and M.E. Koch, J. Opt. Soc. Am., B6,(1989), 639 52. N.J. Krasutsky, J. Appl. Phys., 54,(1983)., 1261 53. W.Lenth, and R.M. Macfarlane, /. Lumin., 45,(1990)., 346 54. U.Oetliker, M.J. Riley, P.S. May, and H.U. Giidel, J. Lumin., 53,(1992), 553 55. H.Ni, and S.C. Rand, Opt. Lett., 17,(1992), 1222 56. F.Auzel, Y.H. Chen, and D. Meichenin, ICL'93, Storrs, CN, USA, 9-12 August,(l993), and J. Lumin., 60/61,(1994), 692 57. Y.H.Chen, and F. Auzel, Electron. Lett., 30,(1994), 323 58. F.Auzel, and Y.H.Chen, J. Non-Crystal. Sol.,184, (1995), 57 59. Y.H.Chen, and F.Auzel, J. Phys.D: Appl. Phys.,28, (1995), 207 60. F.Auzel,and Y.H.Chen, J.Lumin, 65, (1995),207. 61. N.Pelletier-AHard, and R. Pelletier, Phys. Rev., B26,(1987), 4425 62. M.F.Joubert, S.Guy and B.Jacquier, Phys. Rev. B48, (1993), 10031 63. A.W.Kueny, W.E.Case, and M.E.Koch, J. Opt. Soc. ^m.,B10, (1993), 1834 64 P.Goldner and F.Pelle, Optical Materials, 5, (1996), 239. 65. A.Brenier, G.Boulon, C.Madej, C.Pedrini, and L.Lou, J. Lumin., 54, (1993), 271 66 A.S.L.Gomes, G.S. Maciel, R.E. de Araujo, L.H.Acioli, and C. B. de Araujo, Opt. Com. 103, (1993), 361 67. A.Brenier, L.C.Courrol, C.Pedrini, C.Madej, G.Boulon, J.Lumin.,58, (1994), 285
473
68. J.MDyson, S.M.Jaffe, H.Eilers, M.L.Jones, W.M.Dennis,and W.M.Yen, J.Lumin., 60/61,(1994),668 69. A.Brenier, L.C.Courrol, C.Pedrini, C.Madej, G.Boulon, Opt. Mat., 3, (1994), 25 70. S.Guy, "L'avalanche de photons; application a I'ion Tm3+ dans dijferents materiaux", These, Universite Claude-Bernard-Lyon 1,14 Decembre 1995. 71. J.Rubin, A.Brenier, R.Moncorge, and C.Pedrini, J. Lumin., 36, (1986), 39. 72. S.Hubert, D.Meichenin, B.Zhou, and F.Auzel, J. Lumin. 50, (1991), 7. 73. W.E.Case,M.E.Koch and AW.Kueny, J.Lumin. 45, (1990), 351 74. Y.H. Chen, and F.Auzel, Electron. Lett., 30, (1994), 1602 75. P.Goldner, M.Fesquet and F.Auzel, J.Opt.Soc.Am. B15, (1998), 2668. 76. F.Auzel, Coherent emission in Rare-Earth materials, in Handbook on the Physics and Chemistry of Rare Earths, Vol.22, ed. by K.A.Gscheidnerjr. and L.Eyring, (Elsevier Science B.V., 1996), Chapter 151, p.507.
This page is intentionally left blank
SPECIAL LECTURE
PHOTONIC MOLECULAR AND SUPRAMOLECULAR DEVICES
J. M. LEHN Universite Louis Pasteur Laboratoire de Chimie Supramoleculaire 4, rue Blaise Pascal 67000 Strasbourg, FRANCE
Molecular and supramolecular devices may be defined as structurally organized and functionally integrated chemical systems built on supramolecular architectures. The development of such devices requires the design of molecular components performing a given function (e.g. photoactive, electroactive, ionoactive, thermoactive or chemoactive) and suitable for assembly into an organized array. Photonic devices involve energy transfer or photoinduced electron transfer processes. Of special interest is the possibility to design devices that may form by molecular selfassembling, i.e. by the spontaneous generation of the desired molecular architecture from its components in a given set of conditions. Inorganic systems are particularly well suited for the programmed generation of specific arrangements of photochemically active coordination centers from metal ions and suitably designed ligands. The photochemical properties and the self-assembly of several types of architectures of potential interest as inorganic photochemical devices will be discussed. They concern in particular: — luminescent lanthanide cryptates and their use as energy transfer devices in medical diagnostics; — self-assembled arrays of metal ions of grid type; — mixed architectures resulting from the operation of several assembly subprograms. Developments in molecular and supramolecular design and engineering open perspectives towards the realization of molecular photonic devices based on energy transfer and electron transfer processes, and capable of building-up by self-organization from their components. Such systems may be able to perform highly selective operations for signal and information processing at the nanometric level.
General reference: J.-M. LEHN, Supramolecular Chemistry - Concepts and Perspectives, VCH Weinheim, 1995, Ch. 8.
475
This page is intentionally left blank
INTERDISCIPLINARY LECTRUE
REFLECTIONS ON THE THEORY OF EVERYTHING
G.COSTA Department of Physics, University of Padova Istituto Nazionale di Fisica Nucleare, Padova Via Marzolo, 8 - 35131 Padova (Italy)
ABSTRACT The present status of physics is compared with that at the end of the 19th century. A general issue seconded by mass media is that all fundamental laws of Nature have been discoverd, so that we are not far from the theory of everything, in terms of which all phenomena can be understood. Is this the end of fundamental science? We shall show by specific examples in particle physics and astrophysics that we are far from this presumption; in fact, many fundamental questions are still open and the present theoretical models cannot be tested completely, either because the experimental devices are not available, or because the appropriate mathematical tools are not known.
1. Introduction The idea of comprehending all different phenomena of Nature in a unifying picture goes back to ancient times. About 2500 years ago, the first greek philosophers proposed some unifying schemes based on the assumption that everything could be reduced to a few fundamental elements. In modern times, at the end of the 17th century, Newton showed that, with the law of gravitation, one could interpret all mechanical phenomena and astronomical observations. The impressive synthesis of physics, as it appeared in the treatise "Philosophiae naturalis principia mathematica", indicated that, following the newtonian approch, everything could be explained. Explicitly, in the preface of the Principia, Newton wrote: "I wish we could derive the rest of the phenomena [i.e. those not treated in the Principia] by the same kind of reasoning as for mechanical principles. For I am induced by many reasons to suspect that they may depend on certain forces." At the end of the 19th century, electric and magnetic phenomena were united in a beautiful theory: with his famous equations Maxwell achieved a syntesis of all properties of electromagnetism. A mechanical description was available for each sector of physics, including optics and thermodynamics. The general belief was that all the physical phenomena were understood: very little remainded to be done in physics! Determinism and reductionism were the adopted paradigms: in principle, the past and future states of the Universe could be determined exactly from the knowledge of the present state. As an anectode, we recall what is told about Max Planck: when he was a student at the University of Munich in 1875, he was ad viced to address
477
478 his interest to biology, because all important problems in physics had been solved already. However, at the same time, some symptoms of crisis appeared: specifically, Henri Poincare showed that, in many cases, contrary to what expected, it was not possible to make precise predictions even for simple mechanical systems(e.g. a solar system with two planets). In general, in the study of non-linear mechanical systems, dynamical instability emerges, to-day known as deterministic chaos, so that it is no longer possible to predict the evolution of the system. On the other hand, at the beginning of the 20th century, revolutionary discoveries and developments occurred, which changed completely the previous picture. Severe restrictions appeared to be imposed on the physical knowledge, which before was considered, at least in principle, unlimited. 1) Contrary to what previously believed, there is a finite value for the maximun speed of all physical signals, i.e. the speed of light. This was the basis on which special relativity was founded by Einstein in 1905. As a consequence, only a finite part of the Universe which is around us (inside the light cone) can be explored. This limitation, on the other hand, offers some advantages: specifically, the possibility of looking backwards in time, exploring the remote ages of the Universe. 2) The other restriction emerges from the properties of Quantum Mechanics, which reached the complete form in the period 1925-26, mainly through the work of Schoedinger, Heisenberg, and Dirac. It rules all the phenomena occurring at the scale of atomic physics. It implies that there is a finite disturbance in the observation of the phenomena at that scale: according to the Heisenberg uncertainty principle, the disturbance of the act of measurement cannot be neglected and it cannot be reduced below a certain value. This fact imposes strong limitations on the possibility of determining with sufficient accuracy the initial state of a system, and therefore of predicting its future development. On one hand, the structure of the physical laws of microcosmos appear very far from the common perception of the world, and it can be approached very indirectly through rather sophisticated experiments. On the other hand, the new principles have destroyed the belief that the physical laws lead to exact predictions. However, these new limitations did not reduce the progress of physics; instead, they stimulated intense research activities. In fact, there was a golden age for physics, during which the main features of microcosmos were unveiled. Now, at the end of the 20th century, the idea that we are close to the end of fundamental physics is maybe stronger than ever. On the optimistic side, we may think that we are approaching the theory of everything. On the other side, we are aware of other restrictions, which are: a) the limits of the mathematical tools in the description of Nature; b) the limits of technology which cannot provide all the sophisticated instruments needed for exploring the hidden aspects of reality; c) the limits on the research programs imposed by the allocated budget, Even the expression "theory of everything" can have different meanings. The most ambitious appoach is based on reductionism: one should be able to build a
479 single unified picture for all physical phenomena, and each level of hierarchy would be explained in terms of the lowest one. With a more modest attitude one would consider separately the different levels of complexity: each should be understood independently of the others; moreover, new laws could emerge at each level. In the following, I shall not attempt to present an epistemological discussion, but limit myself to a few examples taken from the physics of the seconf half of this century. Physics developed in various directions, but here examples will be taken from particle physics and cosmology. Specifically, we shall consider the present picture of the consituents of matter and their fundamental interactions. 2. Highlights and open questions in particle physics and cosmology The elementary constituents are listed in Table I. They can be grouped into three generations or families, each of which contains two quarks (each in three different states, denoted by three "colours") and two leptons; altogether there are six different kinds of quarks, each distinguished by a specific attribute called "flavour", three charged leptons and three neutrinos.
Table 1. The three families of quarks and leptons Leptons Families Symbol 1st e
2nd
3rd T
Name
Quarks Charge Symbol
Name
Charge
Electron neutrino
0
u
Up
2/3
Electron
-1
d
Down
-1/3
Muon neutrino
0
c
Charm
2/3
Muon
-1
s
Strange
-1/3
Tau neutrino
0
t
Top
2/3
Tau
-1
b
Bottom
-1/3
Each family is a replica of the others two and the only difference is in the mass scale, which increases in going from the first to the second, and from the second to the third family. The part of the Universe which is around us can be described essentially in terms of the members of the first family, while the role of the two heavier families for the structure of the Universe is not understood. There are four kinds of interactions acting among quarks and leptons: - strong, which act only among quarks; they cannot interchange flavour, but only colour;
480 - weak, which act among quarks and leptons; they interchange quark flavour; - electromagnetic, which act among all charged particles; - gravitational; they act among all constituents. All quarks and leptons have spin | , therefore they are fermions, i.e. they obey the Fermi-Dirac statistics, which requires completely antisymmetric states. Their mutual interections are carried by the quanta of the fields which have spin 1 (with the exception of gravitons which have spin 2): they are bosons, i.e. they obey the Bose-Einstein statistics and their states are completely symmetric. The carriers of the fundamental interations are listed in Table II.
Table I I . Carriers of t h e fundamental interactions
Name
Symbol
Interaction
Photon
7
Electromagnetic
Gluon
g
Strong
W-boson
W
±
Weak(charged)
Z-boson
z°
Weak(neutral)
Graviton
Gravitational
At first sight, only the electromagnetic and gravitational interactions appear to have a role in the world around us; however, the weak and strong forces, which were completely unknown to the physicists of the last century, are fundamental for life: they are responsible for the energy produced in the sun through a chain of nuclear reactions. Moreover, the strong forces are responsible for the confinment of the quarks inside the compound particles called "hadrons", of which there are two different kinds: baryons (fermionic bound states of three quarks) and mesons (bosonic quark-antiquark bound states). The great development in particle physics was made possible by two main theoretical tools: the relativistic quantum field theory (RQFT), which combines quantum mechanics and special relativity, and the use of symmetry and group theory. While RQFT imposes very stringent constraints and it gives rise to completely new features (e.g. particle-antiparticle creation and destruction), the use of symmetry is very powerful in discovering hidden properties of the world of particles. The present theoretical picture is provided by the so-called Standard Model, which is a RQFT based on the gauge symmetry group SU{Z) x SU(2) x U(l): it includes strong and electroweak interactions. (For an elementary introduction see ref. 1 ). The gauge symmetry is a local symmetry which associates a vector field to each generator of the group transformations. The "colour" symmetry group 5(7(3) describes the eight massless gluons mediating the strong interactions among the quarks. The symmetry group SU(2) x £/(l) unifies electromagnetic and weak interactions, which are mediated
481 by the massless photon and the heavy W± and Z° bosons. The masses of the latter (Mw a S0GeV/c2,Mz ^ QOGeV/c2) are generated through the Higgs mechanism, which requires that a scalar field acquires a non-vanishing vacuum expectation value. Thus, the physical vacuum contains not only fluctuations (e.g. particle-antiparticle virtual pairs), but it corresponds to finite values of average physical quantities. A physical signal of this picture is the appearence of a scalar (spin 0) boson, the so-called Higgs boson. The Standard Model has been very successful in reproducing with great accuracy all the features of the present phenomenological situation. However, from the theoretical point of view, it is not completely satisfactory. In fact, it represents only a partial unification of the fundamental interactions; it does not include gravitation; it contains too many arbitrary parameters (in particular, the masses of quarks and leptons are free parameters). Moreover, even if the theory is renormalizable, it is necessary to impose unnatural fine-tuning to make the Higgs boson mass stable under higher order corrections. The only way discovered up to now to eliminate the last difficulty is supersymmetry. This is the only symmetry, consistent with relativity, which has nontrivial commutation properties with the generators of the Poincare group: in fact, it interchanges half-integer with integer spin, and viceversa. Of course, the introduction of this new symmetry gives rise to new phenomena: specifically, it predicts the existence of a bosonic partner for each quark and lepton, and of a fermionic partner for each carrier of the interactions. With this new ingredients, the strong and electroweak interactions can be further united, in a Grand Unified Theory (GUT), keeping a good agreement with the phenomenological data. This would be the first step toward a theory of everything 2 (TOE), since only the gravitational interactions are not included in a supersymmetric GUT. One could imagine to be able to build a TOE, in the sense that it contains the fundamental laws to which the ultimate constituents obey. But, on the other hand, an impossible task would be to find all the solutions allowed by the TOE. One can imagine to be able to decompose a very large molecule of DNA into its uttermost constituents (electrons and quarks), but how to proceed from these constituents -even if only the known fundamental forces should be sufficient to reproduce the double helix of the DNA molecule? In fact, as we go from a level of complexity to one of greater complexity, new phenomena may emerge which have no counterpart at the simpler level. A less ambitious aim is to build a TOE which explains at least the phenomena at the lowest level. But one may wonder whether it is possible to build a self-contained theory within a single level. However, I do not want to go into philosophical discussions about a reductionistic versus a holistic point of view, but remain on more heuristic level. Anyhow, looking at the present situation, in order to confirm that we are following the right track, we must find experimental evidence for the ingredient which is still lacking in the Standard Model: i.e. the Higgs boson (the present lower limit for its mass is about 100 GeV). Moreover, one would like to get some evidence of supersymmetry. In 1993 the Congress of USA cut definitely the project of the SSC (Supercon-
482 ducting Super Collider), a proton-proton collider with maximun beam energy of 20 TeV, when part of the tunnel and buildings were already constructed. Fortunately, in a few years (in 2005) the LHC (large hadron collider), a proton-proton collider with maximun beam energy of 7 TeV, will be operating at CERN; it is a European project, but it receives finantial and scientific support also from other non-european contries, such as USA and Japan. Hopefully, the existence of the Higgs boson will be confirmed, and maybe we shall have some indication of supersymmetry. One may wonder if the LHC will be the last big accelerator for particle physics, and if it represents the last big high energy enterprise. Maybe new technology will be developed and higher energies will be reached through other means. But the cost will be so high to forbid further development. In any case, the research will continue also in other directions. The so-called "passive physics", carried out in large undergroud laboratories, is developing very rapidly and it has given already very important results in neutrino physics. We refer to the Kamiokande (Japan), to the Gran Sasso (Italy), and to the Homestake and Soudan laboratories (USA), to the Sudbury (Canada), and to the Baksan Neutrino observatories (Russian Federation). We should also mention the projects in which neutrino beams are produced in a site and are detected after a long distance; the first experiment is already operating in Japan (from KEK located near Tsukuba to Kamikande, about 250 Km apart). Other projects are in Europe (from CERN to Gran Sasso, about 730 Km) and in USA (from Fermilab to Soudan mine, about 730 km). In these laboratories, not only the elusive properties of the neutrinos, such as their masses and mixing, will be elucidated, but probably rare processes like the proton decay will be detected; moreover, the seaches of supersymmetric particles and of the possible candidates for dark matter will be performed in these laboratories. We know that the study of the motion of the galaxies and globular clusters show that there must be "dark matter", i.e. matter which does not emit detectable EM (electromagnetic radiation in the form of light, infrared or radio waves). In fact, astrophysical observations indicate that the ordinary matter (made of baryons, i.e. quark compounds) represent only a small fraction (about 1/10) of the total matter density. Strangely enough, most of the baryonic matter is dark, and most of the matter is not baryonic. Which are the candidates for this (non-baryonic) dark matter? The neutrinos seem to have too low masses to contribute appreciably to this form of matter; maybe the missing mass is of supersymmetric origin. Also cosmology can give us important information on elementary particles. In fact, according to the Big-Bang hypothesis, extremely high energies and temperatures were relevant at the first instants 3 : all the exotic particles we are looking for with high-energy accelerators were all present in a cosmic soup. Due to the final speed of light, by exploring very far away region in space we are at the same time exploring objects very remote in time. Light decoupled from matter when atoms were formed, i.e. about 106 years after the Big-Bang (when the temperature was T ~ 3 x 103 °K); now we detect a cosmic background EM radiation, which provides a picture of the Universe when it became transparent. The spectrum of this fossile radiation appears to be that of a black body at the temperature of 2.7 °K; its small fluctuations are
483 important to understand the formation of the galaxies. Maybe in the future, through the observation of fossile neutrinos (or maybe gravitation waves), we can have information on the previous periods of the Universe. In fact, neutrinos decoupled from other particles just after about a second (at the temperature T ~ 1010 °I<) and, in principle, they could provide information about the first instants of the Universe. A more precise knowledge of the origin of the Universe will give also information on its future. We know that there are three kinds of solutions of the Einstein equations of the general relativity, which correspond to different possibilities for the fate of the Universe: a) flat space-time: endless slowing-down expansion; b) hyperbolic geometry: expanding for ever; c) spherical geometry: closed Universe, i.e. big-bang followed by big cranch. To determine in what kind of Universe we are living in, it is crucial to determine accurately the value of the matter density pc and compare it with the "critical density"/?; the three solutions a,b,c correspond to the cases p = pc, p < pc, p > pc respectively. 3. Outlook After a very intense period of discoveries in particle physics, we have reached a period in which the new experimental findings require very big and long enterprises. In this period, abstract theories are produced, which unfortunately cannot be confirmed or disproved. On the other hand, the internal consistency and beauty can be an indication of the truth (of course, experimental tests will be always required). We recall the case of the general theory of relativity, which was formulated by Einstein starting from general mathematical and theoretical requirements. Only afterward, it was successfully tested. An extremly difficult task is the combination of general relativity with quantum mechanics, which is required if one wants to understand the properties of matter at very high energies and very short distances. The difficulty is somewhat related to the problems of local quantum field theories, where particles are described as point-like objects. This create the presence of infinities, which worried so much the builders of RQFT. The problem was solved by means of a mathematical procedure, called renormalization, giving up at the same time the possibility of explaining the parameters of the theories (masses, electric charges, coupling constants). In fact, even a GUT contains a certain number of arbitrary parameters; at the fundamental level, it mantains its validity even if one modifies the values of these parameters. However, at higher level, the structure of the Universe can change completely. A real TOE must predict also the value of its parameters, but then it should include all levels of complexity! The procedure of renormalization does not work in the case of gravitational interactions. In the eighties, a way out was found going beyond ordinary local field theories, by replacing point-like by extended objects: " strings". A consistent theory of strings has been built: its fundamental property is that it includes gravity
484 automatically. In fact, for the first time we have a theory which incorporates gravity and quantum mechanics; it includes also the other fundamental interactions; for its consistency, it requires supersymmetry and it must be formulated in 10 dimensions. At present, we do not know if string theory provides a good description of our world, since no experimental tests have been performed. Hopefully, indirect tests and detectable predictions will be found, without invoking impossible experiments at the Planck scale. I am sure that the future reserves for us big surprises; new facts can emerge also in apparently well established domains: for instance, the behaviour of gravity at short distances, where it is masked by other stronger interactions, may exhibit new features. I would like to end with a sentence taken from Freeman Dyson who, referring to the theorem of Godel which shows that the world of mathematics is inexhaustible, writes 4 ; "If my view is correct, it means that the world of physics and astronomy is also inexhaustible; no matter how far we go into the future, there will always be new things happening, new information coming in, new worlds to explore, a constantly expanding domain of life, consciousness, and memory". 4. References 1. G. Costa, in Spectroscopy and Dynamics of Collective Excitations in Solids, ed. B. Di Bartolo (Plenum Press, New York 1997) p. 581. 2. S. Weinberg, Dreams of a Final Theory (Hutchinson Radius, London 1993). 3. S. Weinberg, Tie First Three Minutes (Basic, New York 1977). 4. F.J. Dyson, Review of Mod. Physics 51 (1979) 447.
EARTHQUAKES, MEASUREMENTS, AND MITIGATION OF SEISMIC RISK
RODOLFO CONSOLE Institute) Nazionale di Geofisica Via di Vigna Murata, 605 00143 Roma, ITALY
Unlike in the past ages, in the twentieth century there has been a development of measures for mitigation of damages produced by natural hazards, based on a better knowledge of the phenomena giving rise to such damages. This is the case, with particular evidence, of seismic risk. For any type of risk, a cost-benefit balance could be performed in principle, plotting the curve of the increasing cost for mitigation measures and of the decreasing cost of the expected damages versus a parameter characterising the level of protection. If a minimum of the sum of these two costs does exist, the corresponding level of protection should be chosen as the optimal strategy against that type of natural hazard. In practice, this is difficult to be realised, because it requires the accurate estimate of these costs, implying a detailed knowledge of the natural phenomena and of the vulnerability of the objects subject to damage, versus the level of protection. With regard to the problem of seismic risk, the above mentioned process includes: (1) the assessment of the return time of all the earthquakes exceeding a given size in a specific area, assuming a Poisson time distribution; (2) the intensity and frequency spectrum of the ground shaking corresponding to an earthquake of given size; (3) eventual local site effects on ground shaking; (4) the vulnerability of buildings and other objects in terms of percentage of loss in value (equivalent to the cost of reparation) versus the intensity of ground shaking, and the input parameters characterising the level of protection chosen for their design; (5) the cost of new buildings or the reinforcement of the existing ones versus the input parameters chosen for their design. Items 1 to 3 are dealt with in the environment of geophysics or seismology, while items 4 to 5 are typical engineering problems. Seismology is a discipline for which the galileian concept of "experiment" is hardly applicable. Earthquake can not be reproduced in laboratory tests, and the natural duration of a full earthquake cycle can exceed by orders of magnitude the duration of human lives. Moreover, the information available on earthquakes from historical or pre-historical sources is typically inhomogeneous and unreliable. The assessment of the cost worth being spent for achieving extra safety of people, by the increase of the protection level with respect to the minimum cost level, is a particularly delicate matter. It is in fact a decision-making problem, which implies
485
486
political and social considerations, as well as comparisons with other kinds of risks for the best use of the resources available to the community. When earthquake prediction is considered as a possible element of earthquake risk mitigation, the matter becomes even more complicated. The effectiveness of earthquake precursors has almost never been yet tested by rigorous statistical methods. This depends on the very sparse and almost always occasional observations carried out in the world, in connection with the rarity of strong seismic events. Physical models are not yet available for a reliable deterministic approach. So, with the possible exception of earthquake properties like time and space short-term clustering, or apparent quasi-periodicity of longterm recurrence of large earthquakes, no time-dependent model has been yet proposed with a realistic chance to be implemented in a system of civil protection activities.
LONG SEMINARS SITE SELECTIVITY OF DEFECTS IN III-V COMPOUNDS BY LOCAL MODE SPECTROSCOPY AND MODEL CALCULATIONS DEVKI N. TALWAR Department of Physics, Indiana University of Pennsylvania, Indiana, PA 15705-1087, USA ABSTRACT In semiconductors, localized vibrational mode spectroscopy has been used successfully in identifying the sites occupied by intrinsic defects, light impurities and complexes formed by them. The measurements are most valuable when combined with other complementary techniques. We present an overview of both theoretical and experimental requirements for the observation of localized vibrational modes of impurities and defects. Our discussion is restricted to the infrared absorption and Raman scattering techniques with emphasis on the characteristics of modes caused by light impurities such as beryllium-, boron-, silicon-, and carbon- in GaAs and AlAs grown by molecular beam epitaxy (MBE). Direct evidence for C^ acceptors, and the pairing of intrinsic defects with Si and Be in MBE grown GaAs at low temperatures is demonstrated. A theoretical treatment based on Greens Ofunction method is illustrated by the calculations of isotopic fine structure of BAs lines in GaAs and the modes of nearest-neighbor and second nearestneighbor pair defects in GaAs, AlAs and AlxGa,.xAs. 1.
Introduction
In recent years, the increasing technological importance of defects in compound semiconductors for optical devices has stimulated considerable interest in the electronic1"4 and vibrational properties5"7. Despite numerous recent experimental efforts made by using Shubnikov de Haas (SdH),8 electron paramagnetic resonance (EPR),9"10 electron spin resonance (ESR),11 electron nuclear double resonance (ENDOR),12 deep-level transient spectroscopy (DLTS),13"16 photoluminescence (PL)17, secondary ion mass spectroscopy (SIMS), etc., there is still much controversy and mystery surrounding the identification of defects which produce non-stochiometry and deep localized levels. An alternative way of investigating the behavior of impurity centers in semiconductors is to study the vibrational spectrum through optical experiments.5"7 Any imperfection introduced into a perfect crystal causes the disruption of periodic structure and thus alters the electrical and optical properties of the material. In semiconductors, the disruption in the periodic structure can be caused by the presence of native defects (such as anion- or cation- vacancies; anti-site defects; or self interstitials)18 or by intentionally added impurities during the growth of bulk or epitaxial materials.19
487
488 The impurities may also be present as inadvertent contamination such as boron in GaAs during the liquid encapsulated Czochralski (LEC) growth, where boric oxide (B 2 0 3 ) is used as the encapsulant.20 Hydrogen is another unintentional impurity which can be incorporated during the diverse procedures necessary in wafer and device production such as plasma etching and cleaning.21"22 In many cases hydrogen deactivates the shallow acceptors and donors by forming complex centers.23 In semiconductors, the information about stable sites and charge states is very essential for understanding not only the role of isolated hydrogen but also its reactions with other impurities. Accurate measurements using high resolution probe are now available for the frequencies of LVMs from a wide range of isolated light substitutional impurities as well as complex defects in semiconductors. 24_45 The electrical data has been used in conjunction with the expected defect symmetries to assign the sites of defects responsible for the observed absorption bands in the optical experiments. These results are more rigorously tested and interpreted in terms of a sophisticated Green's function theory. 1.1
Basics of impurities and defects
Table 1.1a depicts the classification of possible point defects in III-V compounds. The lattice location of impurity atoms is generally indicated by a subscript added to the chemical symbol for the atom. Substitutional defects in semiconductors are classified Table 1.1a
Classification of simple point defects in III-V (AB : A1P, GaAs, InSb) compound semiconductors.
Type
Impurity
Native Defect VA, VA
: VG,
v,„
/ Vacancy N
VB :
Substitutional /
AB :
Charged : defect
Az = 1 single donor Az = 2 double donor Az = -1 single acceptor Az = -2 double acceptor
VAs Vsb Alp GaAs Insb
Antisite defect ^
Interstitial
P* BA : AsGa Sb,„ Aj or B,
Isoelectronic : Az = 0 defect
Li, or CUJ
according to their position in the periodic table of the elements relative to the host atom. The atoms from same column of the periodic table as the host atom (Az = 0, z being the number of valence electrons) are known as isovalent (isoelectronic) impurities.
489 Substitutional impurities with Az * 0 are termed as charged defects. The atoms with Az > 0 (Az < 0) are called donors (acceptors) as they have more (less) valence electrons than the host atoms. In order to make the enormous amount of experimental data on impurity modes transparent, it is important to classify point defects with a clear distinction made between intrinsic (native) defects and impurities. The native defects in compound semiconductors include vacancies, antisites and selfinterstitials. In GaAs, the anion and cation vacancies are represented by VAs and VGa , respectively. An arsenic (or gallium) atom occupying the Ga (or As) site AsGa (or GaAs) is an antisite defect, while a self interstitial Ga (or As) atom is represented by Gat (or Ast). It is true that intentionally added isoelectronic aluminum (or phosphorous) impurities in GaAs (see Table 1.1b) occupy the gallium (or arsenic) AlGa (PAs) sites and do not give rise to electrical activity. However, boron atoms can replace either gallium (BGa : isoelectronic) or arsenic (BAs: impurity antisite) lattice sites. Of particular interest in III-V compounds is to understand the role of amphoteric impurities (e.g., Si) of column IV. Silicon in GaAs behaves like donor when it occupies Ga-site (SiGa) and acceptor if substituted for the Assite (SiAs)-33 Besides forming nearest neighbor donor-acceptor pairs, silicon also enters into a number of complexes, especially when other impurities (e.g., Li, Be, B, etc.) are present in the lattice.5"6 Again, the relative concentration of BGa, BAs, or SiGa, and SiAs in doped GaAs can change during material processing including heat treatments.
Table 1.1b A portion of the periodic table of interest for substitutional impurities in GaAs.
1 H Li
II
HI
IV
V
VI
VII
Be
B A) Ga In
Si Ge Sn
c
N P As Sb
o s
Se Te
F CI Br I
Isoelectronic
Ampho - teric
Isoelectronic (i) Az=0
Charged defect (d+1) Az=l
Charged defect (d+2) Az= 2
MB
Cu
Zn Cd
Charged defect (a 2 ) Az=-2
Charged defect (a 1 ) Az = -1
(0 Az = 0
Unlike Si, the carbon impurities in GaAs behave differently. The high solubility of C and incomplete dissociation of metal-carbon bond are responsible for the high efficiency of its incorporation on the As sublattice (CAs: acceptor).33 However, Hall and secondaryion-mass spectroscopy data have revealed the hole concentration [p] at the high C-doping level to be only 40% of the total C-contents. This reduction in the hole concentration is
490 speculated to carbon atoms being transferred either to interstitial (C,) or to Ga-lattice sites (CGa : donors). Hydrogen passivation of C acceptors is also suggested for the reduction in [p] and in carrier mobility. As the incorporation of H often happens unintentionally or accidentally,21"22 the passivation proceeds in many cases by a poorly understood reaction with a defect whose structure is unknown. Gallium arsenide layers grown by molecular beam epitaxy (MBE) at low temperatures (-200 - 350 °C) on GaAs substrates (LT-GaAs) exhibit electrical and optical properties which are distinct from those observed when the layers are grown at the usual high temperature range T ~ 600 °C, HT-GaAs.46"49 The LT-GaAs materials are extremely non-stoichiometric and^-rich (containingup to 1.5% excesses atoms). These layers exhibit high resistivity after a 600 °C annealing treatment. Despite numerous efforts made in recent years, the nature of intrinsic defects responsible for the high resistivity of LTGaAs is not fully understood. 46"50 Among the various possible intrinsic defects, only arsenic antisite (AsGa) related-defects are identified by EPR.48 Hall and positron annihilation measurements have recently suggested that the dominant acceptors in LTGaAs are the Ga-vacancies.50 It is also implied that close 'donor-acceptor' pairs (e.g., Sioa-VGa o r Asoa-Beoa) involving intrinsic defects can be formed in Si- or Be- doped LT GaAs. In Be-doped LT GaAs, we have confirmed the possibility of BeGa-AsGa pairs by infrared LVM spectroscopy45 and a similar study in Si-doped materials has identified the presence of SiGa-VGa pairs.44 There is no doubt that the local mode spectroscopy has been used successfully in identifying the site selectivity of isolated defects with low atomic masses. The indormation provided by this technique is equally valuable in understanding the impurity interactions responsible for the formation of complex centers in semiconductors. The impurity vibrational features are generally detected by the infrared absorption spectroscopy, but Raman scattering has also been used. In Sec. 2, an overview of the infrared absorption and Raman scattering is presented with emphasis on the characteristics of LVMs of isolated (e.g., beryllium-, boron-, silicon-, and carbon, etc.) and complex defects in GaAs and AlAs grown by molecular beam epitaxy (MBE). Recent infrared studies of LVMs on LT GaAs doped with Be and Si are presented in Sec. 2.3. Although, resonance modes, overtones, absorption by local plus lattice modes, temperature dependence of the integrated absorption, line shapes and widths are all important topics these are not presented here. Discussions are available on these topics in several recent monographs and review articles.5"7 A theoretical treatment of LVMs based on Greens function theory is described in Sec. 3. This methodology is used to study not only isolated defects, and the nearest neighbor isotopic shifts of LVMs caused by SiAs, CAs or BAs (antisite), but also to understand the vibrational behavior of impurity complexes in GaAs and AlAs. Our approach emphasizes primarily the delineation of chemical trends and the attainment of a simple physical understanding of the bonding situation through the magnitude of impurity-host parameters (see : Sec. 3.2). A general theory for studying the vibrations of impurity complexes with a maximum of three
491 substitutional defects in semiconductors is outlined. In describing the perturbation models, we have considered mass change at the impurity sites and assumed that the force constant variations about the defects are significant only up to the first nearest neighbors (c. f. Sec. 3.3). We do not consider perturbations in the Coulomb interactions since their long-range would render the Greens function approach intractable. For simplifying numerical calculations, the use of Group theory is outlined in Sec. 3.4. The study of vibrational modes is illustrated by the detailed calculations of isolated defects, isotopic fine structure of the BAs (SiAs, CAs) line in GaAs and of the nearest-neighbor and second nearest-neighbor pair defects. Comparison of the calculated results with the experimental data is given in Sec. 5. The results are discussed in Sec. 6 with concluding remarks presented in Sec.7.
2.
Localized vibrational mode spectroscopy
2.1
Infrared absorption
The development of lattice dynamics has always been parallel to the advances made in understanding of the interaction of radiation field with matter. In the harmonic approximation, the interaction of radiation field with phonon field corresponds to the creation or annihilation of one phonon and hence to the existence of one band, centered at a single phonon frequency in the absorption spectrum of a diatomic crystal (e.g., GaAs). The dispersion relation obtained clearly demonstrates that the infrared dispersion originates from the interaction of incident photons with transverse optical phonons of the crystal. In the strong coupling limit, the normal modes of the coupled systems, the polaritons, have radiation and phonon like character. This strong interaction in GaAs makes the material appear opaque to infrared radiation in the region of restrahlen (just below the maximum phonon frequency comax -295 cm"1 at 4.2 K).51 However, one also observes on both sides of the fundamental band sets of weaker absorption lines corresponding to the multi-phonon transitions due to anharmonic terms in the electric dipole moment (see : Fig. 1).
30 —
1
I — I —
—i
1 —
r — '
-i—r-
0
z <
19
20
J
1
I
< If / h 10 h
o 1 c111
0!iZ 100
2 PHONON ABSORPTION
1
,
|
L i 1 • I i .1 200 300 400 500 WAVENUMBERSCM-1
2U>MAX
1 600
FIG. 1. Intrinsic vibrational absorption of GaAs at 4.2 K.
492 These absorption bands are found on the sum and difference of normal mode frequencies. In the two-phonon additive process, the absorption has a cutoff at 2comax (-590 cm"1) and is reduced in strength as the temperature is lowered to 4.2 K. Such absorption bands appear when two branches of the phonon dispersion curves will have a critical point for the same wave vector or when the slopes of the two branches will be equal for a subtractive process and opposite for an additive process. The LVMs (coL) of light impurities (except H) lie generally in the spectral range from comax to 2comax. The resonance states of heavier impurity atoms fall in the low energy side (co < comax) and in the region where the density of phonon states is low. If the impurity vibration is superimposed in the two-phonon background, as in the case of l2CAs, the intrinsic absorption for pure GaAs has to be subtracted to reveal weak LVM line more clearly.
2.2
Impurity modes of isolated defects
In Table 2.1, we have compiled the existing experimental data of LVMs for several isolated defects (ranging from isoelectronic to charged) in GaAs of comparable masses. Table 2.1 Localized vibrational modes (in wave numbers (cm"1)) of isolated defects in GaAs Ref. [5].
Impurity 6
Li0a(a2-) LiGa(a2-) "BecCa") 10 BGa® "B 0 ,(i)
Frequency coL
"PMQ
482 450 482 540 517 628 601 582 561 480 836 790 715, 730 679 331 326 322 362 384 379 373 399 389 355
"Ma*)
-
7
"KCa2-) "B^a 2 -) "CA.M
"CJa") "N*® "Oj 18
0,
16 os 18 o s 24
MgG.(a') Mg0,(a) Mg0,(a) 27 AlGa(i) 28 Sio.(d+) "Si^d*) 30 SiG,(d+) "SizuM "Si*, (a)
25
26
Reference Theis and Spitzer (1984) Murray etal. (1989a); Nandhara et al. (1988) Newman etal. (1972) Collins et al. (1989) Thomson et al. (1973) Gledhill et al. (1984); Woodhead et al. (1983) Newman et al. (1972) Theis et al. (1982) Kachare eta/. (1973) Akkerman etal. (1976) Schneider etal. (1989) Song etal. (1987) Schneider etal. (1989) Leung etal. (1972)
Lorimor and Spitzer (1966) Spitzer and Allred (1968) Theis and Spitzer (1984) Murray etal. (1989b) Theis and Spitzer (1984) Leung et al. (1974) Spitzer (1967)
493 When impurities with heavier mass are present, the modified lattice modes involving significant displacement of the impurity fall within the bands of optic or acoustic modes of the perfect crystal. In general, broad infrared absorption features occur at a < comax with a corresponding low sensitivity for their detection. Only useful information is obtained when resonance occurs in the region where the density of phonon states is low. The perusal of Table 2.1 reveals that significant change in coL is produced if mass of the impurity is replaced by an isotopic substitution. Small shifts of coL occur, however, if the isotopic masses of nearest neighbor host atoms to the substitutional defect are changed.33 In GaAs, the isotopic changes cannot occur for light impurities occupying the gallium sites (e.g., BGa or SiGa) as the neighboring 7SAs atoms are 100% naturally abundant. However, for impurities occupying the arsenic sites (e.g., BAs and CAs) there can be five different centers involving two naturally occurring 69Ga (60%) and 7IGa (40%) isotopes. The statistical combinations of these isotopic centers (see Fig. 2) have caused closely spaced lines, leading to the positive identification of the site selectivity of defects.
2? "E •e •3. c .2 u> to
'E
i
6 0 2 601 6 0 0 Wave number (cnf1)
Fig. 2
High resolution (0.03 cm"1) spectrum of the LVM absorption from "BM in GaAs (Gledhill et al. (1984) upper panel). In the lower panels, the numbers 1, 2, 3....9 represent the position and the relative heights of the calculated lines according to their probability of occurrence (Talwar et al. (1986)).
494 This effect was first noted to explain the larger linewidth (full width at half maximum FWHM) of the 28SiAs LVM compared to 2SSiGa under conditions of low instrumental resolution. Subsequently, the LVM of ,2Cm was resolved into five fine-structure components using high resolution Fourier transform infrared (FTIR) spectroscopy.33 Theoretically, nine such components are expected. Our Greens function calculations have shown that the three highest frequency components are close together and are unlikely to be resolved experimentally. Another experiment that supported our theoretical analyses of the spectra in 12CAs and provided direct evidence for the llBAs center is the work of GledhiU et a/.31 where a 601 cm'1 LVM is resolved by FTIR into four or possibly five fine structure lines (see Fig. 2). Theoretical pattern (as discussed in Sec. 5.2) shown in Fig. 2 is clearly a fingerprint for the substitutional impurity nB occupying the As site with Gaisotopic neighbors.
2.3
Complexes involving intrinsic defects in doped L T GaAs
Among the various possible intrinsic defects in LT-GaAs (e.g., AsGa, Ash VGa) only arsenic antisite related defects have been identified by EPR.48 Recently, infrared absorption due to localized vibrational modes of Si and Be has been investigated in LTGaAs grown by MBE.44"45 In S7-doped materials, a spectrum from SiGa-VGa has been identified (see Fig. 3). (a) Grown at 250°C
380 390 Wavenumber (cm-1)
Fig. 3
Infrared LVM absorption at (20 K) due to Si impurities in LT-GaAs samples grown by MBE at (a) 250 °C and (b) 210 °C, respectively (McQuaid et al. (1993)).
495 Based on the Greens function analysis of our FTIR data in Be-doped LT-GaAs, we strongly suggest (c. f. Sec. 5.5) the possibility of BeGa pairing with AsGa. 2.4
Raman scattering
Raman scattering is a complementary technique to IR absorption in the sense that selection rules are more specific and the symmetry of the modes coupled to the radiation field can be different. It is true that IR has a very high instrumental resolution (~ 0.03 cm"'), but it requires relatively thick samples and the need to compensate materials containing high concentration of shallow donors or acceptors.5"6 This disadvantage of IR spectroscopy is overcome in Raman spectroscopy where only microscopic amounts of materials are needed. The laser spot can be focused to - 1 urn and its penetration depth in a semiconductor material is ~ 10 nm at a photon energy of 3 eV. Raman spectra (77 K) recorded43 from layers of AlAs (-500 nm) grown by MBE on semi-insulating GaAs (001) at a temperature of 600 °C is shown in Fig. 4.
AlAs : C T = 77K
U
500
Fig. 4
i
I
i
I
i
l
.
l
.
i
600 7O0 Raman Shift (cm-')
Raman scattering spectra of AlAs : C sample (-500 nm) grown on GaAs (001) (Davidson etal{\ 994)) for a range of excitation energies and for an undoped reference sample.
496 The local mode of CAs in AlAs near 637 cm"1 observed by Raman spectroscopy has a linewidth (FWHM) of A = 20 cm"1. Infrared absorption measurements have also been reported on the same samples, after subtraction of the intrinsic phonon spectra of GaAs substrates. Because of the greater sensitivity, the IR data recorded at a much higher spectral resolution revealed several weaker features.43 In Sec. 5.3, we compare the measurements with LVM frequencies obtained from a Green's function theory.
3.
Theoretical
To extract the nature of impurity-host bonding in semiconductors from the existing optical data, two equally reliable theoretical efforts have been made in recent years : (a) A microscopic analysis based on the local-density approach - with the use of ionic pseudopotentials, the phonon energies for perfect or imperfect systems can be determined from first principles.53 (b) A macroscopic analysis based on the general treatment of the lattice dynamics in terms of interatomic forces - with the use of the Greens function for the perfect or imperfect lattice, the displacement response to the sinusoidal driving forces may be estimated. 52 ' 54 It is worth mentioning that the former type of calculation requires heavy computational work for the isoelectronic defects and presumably it would be even more cumbersome for charged defects. Except for a few ab initio calculations, the nature of impurity-host bonding has yet to be extracted by first principles for donors or acceptors in semiconductors. On the other hand, the Greens function theory has been successfully applied to understand the vibrational properties of imperfect solids, including the impurity modes, impurity induced IR absorption, Raman scattering, phonon sidebands, etc. 55 ' 56 The advantage of the second technique over the first one is that it allows the coupling of the vibrations of the defect to the bulk crystal, and one can easily visualize which type of vibrational modes remain localized around the defect. A comprehensive account of the dynamical properties of imperfect lattices using Greens function technique has been discussed in several review articles, monographs, and books.57'58 The discussion here is thus very brief for the purpose of establishing notations to be used throughout this article. 3.1
Greensfunction method
In the harmonic approximation, the lattice dynamics of the perfect crystal in Greens function notation is given by57 (Mco 2 -£)G(G)) = I
,
(1)
where co is the phonon frequency, M is the diagonal matrix containing the masses of the constituent atoms MK ( K = 1, 2), I is the unit matrix, <J> is the force-field matrix, and
497 G(co) is the lattice Greens-function matrix. The eigenfrequencies of the host system can be easily found by solving the equation det(Icu 2 -D) = detG- 1 /detM
,
(2)
where D =_M'm $ M"1/2 is called the dynamical matrix. The eigenvectors of the dynamical matrix D are the phonon normal modes \q,J>, where (q,j) labels the momentum and branch index: B\qj>=w\qj)\q,J>
.
(3)
The relation between the basis vectors I IKCC > of the crystal lattice site representation and the vectors | q,j~> of the momentum space representation can be expressed concisely by the transformation function=
N~U2exip{iq.R(lK)ea(K\qj)
,
(4)
where the set of vectors ea{ KI q ,j> are eigenvectors of the Fourier-transformed dynamical matrix Dafi(q\KK')=^xp{-iq.[R(lK)-R{rK')]}D^(lK,rK') , (5) X Dal) (q I KK' )ep (K' \qj) = (02 (qj)ea (K I qj)
.
(6)
The component form of the perfect lattice Greens function matrix takes the form 1
GaflWK': a) = W
l / U
_ , eJK\qj)e'jK'\qj)
,
) a
£'
2
'
exp{iq[Rm-R(rK*)]}, (7)
N(MKMK.yl2f< co2-co2(qj) where N denotes the number of wavevectors in the first Brillouin zone, R(ltc) is the equilibrium position vector of the atom (IK). For each eigenvalue co(qj), the eigenvectors satisfy the familiar orthogonality and closure relations.57 If P is the perturbation on the dynamical matrix of the host lattice, then the Greens function for the imperfect lattice U is defined as (Mco2 - *_- P) U(co) = I
,
(8a)
,
(8b)
or equivalently, (I-GP)U(ro) = G with the perturbation matrix P = -AJVI co2 + AO
(8c)
Here, A M and A_S> are the mass-change and force constant-change matrices. To make the problem tractable, P is usually assumed to be very localized around the defect. One, therefore, needs to solve the small sub-block of Eqs. (8 a-c) corresponding to the nonzero elements of the perturbation matrix.
498 3.2
Bulk phonons
In order to evaluate the Greens function and thereby the defect properties, it is necessary to construct a realistic model of the host lattice. We used an eleven parameter rigid-ion-model (RIM11) of Kunc59 for treating the phonons of GaAs and AlAs. In this model the crystal is regarded as consisting of a lattice of rigid, non-polarizable ions which interact through long-range electrostatic and short-range repulsive forces. The short-range interaction terms, whose physical origin is a repulsion between overlapping electronic distributions, fall off rapidly with distance, and are included only up to second-nearest neighbors. The Coulomb forces between ions are assumed to be electrostatic interactions between rigidly spherical charge distributions (i.e., point-charges) that are multiples of an effective electronic charge Ze. A more complete description of the physics of the rigidion-model, and of the success it has for predicting various experimental quantities, can be found in the original source. 59 The most compelling reason for our choice of RIM11 is that the application of the Greens function method is quite straightforward. The local basis set required by the Greens function method is simply the x, y, and z vibrations of the rigid ions. There is no need of introducing an extra degree of motion such as the one between an electron shell and a rigid core (shell model), or between a bond charge and an ion (bond-charge model). 59 In the RIM11, the quantities of interest for our calculations are the force-constant matrix 3> and the dynamical matrix D (= M""2 <5 M"la)» where the former can be expressed as
$ = o s + 3>c
(9)
,
where O s and <1>C correspond, respectively to the short-range Coulomb terms. Because of the tetrahedral (Td) symmetry of GaAs, the off-diagonal Cartesian blocks of the shortrange force constant matrix can be shown to have the form ^A £ s (Ga,As) = $ s (As,Ga)
B
B^
B A B B
B
(10a) AJ
A s
* (Ga,Ga) =
A
C,
-E,
-E,
^ (10b) Ei V
A -E^ s
£ (As,As)
:
A
(10c)
499 for interactions between nearest-neighbors (NNs) and second-nearest-neighbors (2NNs), respectively. One parameter Ze (the effective charge) is used to determine O c giving a total of 11 parameters. The evaluation of the involved 11 parameters is usually made using the neutron scattering data of eigenfrequencies at critical points as an input with the three experimentally known elastic constants as constraints. The calculated phonon dispersions based on RIM 11 for GaAs are shown in Fig. 5, along with the experimental results taken from the work of Strauch and Dorner60. •
i
i
GaAs
FREQUENCY
Fig. 5
(THx)
Rigid-ion-model calculations for the one-phonon density of states (upper panel) and the phonon dispersion curves along high symmetry directions (lower panel) for GaAs. The calculated results are compared with the inelastic neutron scattering data (solid circles) from Strauch and Dorner (1990).
500
This serves to illustrate the level of agreement obtained by RIM11. The corresponding one-phonon density of states for GaAs are also shown. Similar calculations for AlAs have provided a wide gap (-216-347 cm"1) between the acoustical and optical phonon branches. By incorporating the lattice phonons generated by RIM11, we have computed the bulk Greens function matrix elements {G} following the method discussed at length elsewhere.52 A general computer program has been written which included perturbation around three (maximum) substitutional impurities at specified atomic sites in a zincblende lattice (see : Fig. 6).
Fig. 6
Impurity models for the vibrations of one, two, and three defect centers in semiconductors : The number of atoms 1, 2, ...11 and their respective coordinates are described in Ref. [52] i) Td case : the isolated defect loa (las) located in site 1 (2) bound to NN As (Ga) atoms at 2 (1), 3 (6), 4 (7), and 5 (8) [see : Fig. a], ii) C3v case: the pair defect Ii(Ga) - l2(As) with the I[ impurity occupying the Ga site 1 and the I2 impurity occupying NN As site 2 [Fig. a)], iii) The most extended case of three substitutional defects occupying sites 1, 2, and 6 as described in the text in the NN approximation [see: Fig b]. If Ij = I6 the impurity center exhibit C2v symmetry; otherwise, Cs.
501 Besides studying the impurity vibrations for isolated defects with tetrahedral (TJ) symmetry, this program is also capable of modeling complexes involving defect centers with C3m C2v and Cs point-group symmetries. The main hypotheses adopted in defining the perturbation matrices {P} are those of Ref. [52]. The effects of local distortions around impurity atoms are studied by using a simple but first principles bond-orbital model of Harrison61 The values of the radial force constants between the impurity and the host-lattice atoms are estimated approximately from the second derivative of the bond energy and are included appropriately in the perturbation matrix P. The frequencies of the impurity vibrations are obtained by setting the determinant of the dynamical matrix equal to zero, i.e., Det|I-GP| = 0
3.3
(11)
Impurity perturbation
The simplest crystal defect considered here is a substitutional impurity replacing either a Ga atom (K= 1) or an As atom (K = 2) in the GaAs lattice. It involves only five atoms if the impurity space is confined to its nearest neighbors. We assume both the atomic mass at the impurity site and the NN force constants (A, B) to be changed. The changes are described by the parameters £,= (Af 1 -M,')/M 1 and
t = (A-A)/A
= (B-B)/B
= \-a
,
(12a)
,
(12b)
,
(12c)
or £2={M2-M'2)IM2 and
u=(A-A)/A
= (B-B")/B
= \-b
,
(12d)
when the impurity M[ or M[ occupies the Ga (1) or the As (2) site, respectively (see : Fig. 6a). For studying the vibrations of a NN pair defect of C}V symmetry we need a minimum of two (t and u) parameters, whereas for a complex defect with three impurity atoms (see : Fig. 6b) an additional parameter is required. Following Eqs. (12 a-d), the mass change and force constant change parameters for the impurity occupying the Ga site (6) can be obtained as: £*=(M6-M6')/M6, and
v=(A-A'")/A = (S-S"')/B = l - c
(12e) ,
(12f)
To make the perturbation models meaningful, we have also considered direct interactions in between the impurities by two additional parameters Fn and r26- The force variations are given by
502
and
FI2=\-ah + Tn = u + t-ut+r12
,
(12g)
F26=\-bc + T26 = u + v-uv+r26
,
(12h)
Although no explicit consideration of the Coulomb interactions between the charged impurities occupying sites 1, 2 or 6 has been made, their effects causing change in the electronic charge densities and hence the variation of forces in the 1 - 2 and 2 - 6 bonds, are contained in F12 and F26 parameters. A complete perturbation matrix P (33x33) is formed for a special case of an impurity complex with three impurity atoms at the 1, 2, and 6 sites, respectively.
3.4 Group-theoretical simplifications Single substitutional defect: Td symmetry The simple case of an isolated substitutional defect in zinc-blende type crystals retains the full point-group symmetry Tj and involves only five atoms if the impurity space is confined to its nearest neighbors. It is convenient to express the full size (15x15) G and P matrices in terms of a basis of symmetry coordinates which transform according to the Aj, E, Fj, and F2 irreducible representations of the tetrahedral group. The A, coordinate consists of a 'breathing-type' motion in which the impurity atom remains stationary and the NN host atoms move radially. Again, in the Fj and E coordinates, the impurity remains at rest while the NN atoms move. The impurity vibrates only in the triply degenerate F2 mode and we expect LVMs of this representation. Nearest-neighborpair defect: C3v symmetry The point group symmetry of the NN pair defect is C?v with the axis along the bond joining the defects. In the NN approximation the size of the G and P matrices is (24x24). The vector spaces formed by the displacement of the impurity molecule and its NNs transform according to Aj, A2 and E irreducible representations. The impurity center does not move in the A2 representation and only A] and E types of modes are optically active. Impurity complexes: C2v or Qsymmetry A 2NN pair defect with orthorhombic symmetry can be considered by assuming an impurity atom on the Ga site (1) and a second impurity occupying the nearest Ga site (6) in the [100] direction. The point group symmetry is C2v if Ii = I6; otherwise it is Cs. The defect space involving 11 atoms increases to 33x33 and the vector spaces formed by the displacements of the impurity molecule and its neighbors transform to A,, A2, Bj and B2 irreducible representations for C2v and to Ai and A2 representations for Cs symmetry, respectively. The impurity complex with C2v symmetry can give rise to A,, B1 and B2
503
types of vibrations, whereas in the Q-symmetry case both ^4, and A2 modes are optically allowed. 4.
Numerical Computations and Results
4.1 Isolated defects The solutions of Eq. (11) for A,, E, Fh and F2 irreducible representations for single substitutional defects in GaAs and AlAs were carried out with the appropriate perturbation parameter t (or u). The results of LVMs in the F2 representation for impurities occupying the Ga site in GaAs are displayed in Fig. 7. It is worth mentioning that the force-constant change parameters estimated from the local mode study have established52 an important physical fact that the bonding around charged impurities in semiconductors results from the equilibrium between the repulsion of the cores and the attraction due to valence electrons and is associated with the charge overlap (see: Sec 6.1). 700
'g 630 W
& 560 GI
3
I 490
I 420 an
I 350 280
• " ' ' ' " ' ' "
" '
• " • ' * • " ' '
••
• ' " '
•'
• • • •
-1 -0.75-0.5-0.25 0 0.25 0.5 0.75 1 t (font constant change parameter) Fig. 7 Greens function calculations of the localized vibrational mode frequencies (cm"1) as a function of the force constant change parameter t for impurities occupying the Ga site in GaAs.
504
Since a wide phonon gap (~216 - 347 cm"1) exists between the acoustical and opticalphonon branches of AIAs, one expects gap modes to emerge from the minimum of optical band for heavier impurity atoms occupying the Al site (e.g., SiAh and GaAl in AIAs). Light impurity atoms on the Al site (e.g. BM) in ^4/As will cause only LVMs. Besides the highfrequency LVM for a lighter impurity atom on As site in A\As, one also expects a gap mode to emerge from the top of the acoustical band. In our study, we found that a force variation u = -0.62 (see : Figs. 8 (a -b)) for uCAs in A\As provides a local mode near -630 cm"1 as well as a gap mode near 283 cm"1. It may be noted that the value of u estimated for A\As:C is almost identical to the value deduced for GaAs.C. We also noticed that the gap mode of CAs has a smaller amplitude on the C atom than that of the local mode. Therefore, it shifts only slightly to a value of -280 cm"1 by carbon isotopic (,3CAs)
! !
-1 -0.75-0.5-0.25 0 0.25 0.5 0.75 1 320
1 1 1 ' ,
i • ' '
i ' '
'
i ' '
. . . , . , ,
..,....,
1
AIAs
300
-
280
-
^\\o 260
"
240
-^
220
-
200
Fig. 8
^v\ ^
>v^SI
'
\ \
-
*
-
1 -0.75 -0.5 -0.25 0 0.25 0.5 0.7 5 u (force constant change parameter)
a) Green's function calculations of the localized vibrational modes (upper panel) as a function of the force-constant change parameter u for impurities occupying the As site in AIAs. b) Same key (lower panel) but for the gap modes due to impurities occupying the As site in AIAs.
505
substitution, whereas the shift of the LVM to 610 cm'1 is much higher. Despite numerous speculations, there is still no direct evidence of the existence of C donors in GaAs and AlAs. If they existed, the C donors should have caused force-constant changes similar to those of Si donors in GaAs (AlAs) with minor corrections due to the smaller sizes of their covalent radii (see : Table 4.1). This assertion is based on our earlier comprehensive analyses of LVMs due to various charged impurities in III-V and II-VI compound semiconductors.52 Table 4.1 Comparison of the calculated LVMs (cm"') due to several donor, isoelectronic, acceptor, and complex defect centers in GaAs, AlAs, and Alo.02Gao.98As.
System
Experimental
Theoretical
Relative force constant change parameters
362 331 384 482 540
362 F2 331 F2 384 F2 482 F2 540 F2 565 F2 355 F2 399 F2 582 F2 331 F2 535 F2 630 (283) F2 610 (280) F2
t = 0.18 t = 0.5 t = -0.03 t = 0.63 t = 0.33 t = -0.07 u =-0.31 u =-0.54 u = -0.61 t = 0.15 t = -0.10 u = -0.62 u = -0.62
464.2 Ai 392. 7 E 323.03 E 301.2 A,
t = -0.03 u = -0.54 r, 2 = -0.6
778.25 A, 666.47 E 536.03 E 489.06 A,
t = 0.15 u = -0.61 r 12 = -0.8
646.47 A, 641.53 B2 627.23 B, 626.67 Ai 614.79 B2 606.19 B,
u = -0.61 t = -0.06 v = -0.61 r 1 2 = 0.15
631.17 Ai 573.40 E 369.69 E 359.97 A,
t = 0.24 u = -0.51 r, 2 = -0.73
(a) Isolated GaAs:Al GaAs:Mg GaAs: Si GaAsiBe GaAs:B GaAs: ,2C G<Us:P GaAs:Si GaAs:'2C AI'As: Si ^lAs: l2C Al4s:,2C A]As:aC
-
355 399 582 331
-
630
-
(b) Complex SlGa-SlAs
464 393
CATCAS
v^As -v ^As
AlGa-CAs
631.6 573.6
r26 = o.i5
506
If the corrected value of t is used for nCM in ,4/As, we predict its LVM to be near 535 cm"1. A similar study of 12CGa in GaAs suggested the local mode to occur near 565 cm"1. The value for the local mode for the C donor is indeed very close to the satellite line F(~563 cm"1) observed in highly C-doped GaAs films grown by metal-organic molecular beam epitaxy (MOMBE).62 A simple model calculation in GaAs for the passivated H complex (H-CAs) in the bond centered configuration (C3v symmetry) also correlates the observed line Yto the carbon E mode (c. f. Sec. 6.2). 4.2 Carbon complexes in GaAs, AlAs and AlxGaj.^is For a low Al composition (e.g.,* < 0.04) in AlxGa!.xAs:C samples, one expects four vibrational modes due to a NN CAs-AlGa pair of C3v symmetry: two because of the movement of the C-Alpak along the line of joining, 0),( A,+) <—>,ra4(A~) ->->, and two as a result of movement perpendicular to it, co2(£+) T4- and cn^E') t t . Our calculation of such a NN pair defect predicts the vibrational modes to be at 0^(631 cm'1), > co2(573 cm"1), > co3(370 cm"1), and > oo4(360 cm"1), respectively. For a higher Al content in AlxGai_xAs:C samples (say x > 0.05), the presence of more than one Al atom in the vicinity of the CAs acceptor is expected to affect the carbon local vibrational modes. For one, two, three, and four Al neighbors in the vicinity of CAs in AlxGa!.xAs:C, our calculations predict the frequencies of C-local modes to lie near 631 cm"1 (Ai) and 573 cm"1 (E) (C3v symmetry); 636 cm"1 (A,), 610 cm"1 (B,), 570 cm"1 (B2) (C2v symmetry); 640 cm"1 04;) and 600 cm"1 (E) (C3v symmetry); and 630 cm"1 (F2) (Td symmetry), respectively (see : Table 4.1). 4.3 Intrinsic impurity centers in dopedLT-GaAs As mentioned before, a very high concentration of intrinsic defects exists in GaAs layers grown by MBE at low temperatures (T~ 200°C). Measurements of VGa using slow positron annihilation50 showed that the concentration of Go-vacancies is as high as 1019 cm"3. The presence of ^5-antisite defects (AsGa) has also been confirmed by EPR.48 In GaAs, the intrinsic defect AsGa resembles with an impurity for which there is no significant mass change but the 'impurity-host-bonding' may be modified. Due to the variation of 'impurity-host-bonding' it is feasible that AsGa may give rise to an inband resonance mode similar to phosphorous impurity in silicon lattice {PSi}.5 If BeGa-AsGa or (SiGa-VGa) forms acceptor-donor (donor-acceptor) pair, as might be the case in Be- (Si-) doped LT-GaAs, then the point group symmetry at the BeGa (SiGa) site will be reduced from Td -> Cs. Such a reduction in the symmetry will lift the degeneracy of the BeGa (SiGa) LVM and can modify the absorption spectrum. One can use this modification in the IR spectrum and monitor the possible evolution of the formation of complex defect centers involving impurities and intrinsic defects.
507
Recently, we performed such measurements for Be acceptors in LT-GaAs grown by MBE at temperatures rangingfrom200 to 325 °C.45 Absorption spectra were recorded by BOMEM FTIR spectrometer with a resolution of -0.25 cm"1 after cooling the samples at 77 K. Similar studies were reported earlier on Si doped LT-GaAs.44 Although the symmetry of complex centers involving impurities and intrinsic defects can be determined, their identification relies strongly on supplementary data including realistic model calculations. The IR data on LVMs (see : Table 4.2) when analyzed by the Greens function theory provided direct evidence of the 'donor-acceptor' pairs involving intrinsic defects (e.g., SiGa-VGa and AsGa-BeGa) in Si and 5e-doped LT-GaAs.
Table 4.2 Summary of the calculated and experimental LVMs and inband modes of VGa, AsGa, BeGa-AsGa, SiGa-VGa in Be and Si-doped LT GaAs.
5.
Impurity center
Symmetry
vG.
T„
Asa,
T„
Mode Freq. (cm"1) Experimental Theoretical
Af/f(/orv)
47 Fa
40.4 A,
1.0
186.9 Fa 198.4 A, 220.1 Fa
0.23
200 F2 223 A,/F2 482 Fa
482 F2
0.63
480.3 A, 481.8 A, 482.3 Aa
same as isolated case
384 F2
-0.03
366.3 A, 367.0A, 397.3 A2
-0.03 1.0
BeGa
T„
Bee - AsGa
Cs
SlGa
Td
384 Fa
SlGa - VGa
C,
366.7 A, 367.5Ai 397.8 A2
Comparison with experimental results
5.1 Si donors, C and Be acceptors in AlAs Electrical and optical measurements63 are reported recently for AlAs:Si/GaAs samples grown by solid-source MBE at 520 °C, in which the Si doping levels were lxlO19, 5xl018, and2xl018 cm"3, respectively. Transmission FTIR data recorded with a Bruker IFS 120 interferometer at a resolution of -0.1 cm"1 at liquid-He temperature revealed a small
508
absorption peak near 331 cm"1. The width of the peak A (= 0.6 - 0.7 cm"1) is found comparable to the usual impurity LVMs, and its strength is clearly in good proportion to the dopant concentration, although the peak with the 2x1018 cm"3 Si dopant level could not be distinguished from the noise in the sample. The nature and concentration of the charge carriers estimated for each sample from Hall-effect and C-V measurements suggested that the impurity responsible for the -331 cm"1 peak is to be related to Si donor substituted for Al site (SiAi) in AlAs. The gap mode observed by FTIR spectroscopy for SiM in AlAs is in excellent agreement with our theoretical calculations (see : Table 4.1). More recently, Ono and Baba64 also studied the effects of charge states on the LVMs of Be acceptors and Si donors in AlAs. For the neutral state of a BeAI acceptor, a local mode is observed near -481 cm"1 which shifts to a lower value of 470 cm"1 for the ionized case BeAl. For Si donor substituting for the Al atom in ,4/As, a gap mode (near 331 cm"4) is observed irrespective of its charge state. We believe that as the beryllium acceptor is transformed from its neutral to ionized state (fie°;—»Be~Al) the extra electronic charge is transferred primarily to the Be atom. This charge will repel the electronic charge density more towards the As site, causing softening in the Be'-As force constant (c. f. Sec. 6.1) and consequently resulting in the shift of the LVM to a lower frequency, as observed experimentally. For the case of SiA/ gap mode, the heavier impurity atom Si does not vibrate alone - in fact, its vibrations are coupled with the vibrations of As host lattice atoms in AlAs. Therefore, the charge state dependent gap modes of Si donors are not significantly affected by a small force-constant variation around the impurity atom. Recent Raman study43 of heavily C-doped AlAs/GaAs(001) samples grown by MBE has revealed a LVM of CAs (acceptor) near 635 cm"1. Infrared absorption measurements on the same samples has shown a corresponding line at 630 cm". A small discrepancy in the two values of the LVMs is probably due to the fact that IR measurement relates to the total layer thickness, whereas Raman spectroscopy study with above-band-gap excitation, primarily probes the near-surface region. Again, the A\As:C layers are not perfectly lattice-matched to the GaAs substrate because of the carbon doping so that residual strains are expected. For the layer thickness of - 500 nm this strain would be partially relaxed which could cause a variation of the CAs local mode. Our theoretical result for the LVM of CAs is in excellent agreement with the IR data. Besides the local mode, the theory also suggests the possibility of a gap mode near - 283 cm"1. For an isotopic l3CAs substitution, a shift to lower values in the local (-610 cm'1) and gap (-280 cm"1) modes is predicted (see : Table 4.1). There is no experimental data available for the carbon-isotopic shift in the LVM for comparison with our calculations, and it is not known to us whether the absorption due to the gap mode of the C acceptor in AlAs was present or not in the FTIR study.43,65 From the calculated value of the gap mode [-283 cm"1 (-280 cm"1)], however, we strongly believe that this mode could have been obscured by the very intense restrahlen absorption of the GaAs substrate.
509 5.2
Ga isotopicfine structure ofL VMs
Van Vechten18 was the first to predict theoretically the existence of both self- (GaASj AsGa) and foreign (BAs) antisite defects in GaAs. Using EPR measurements on samples grown from^i-rich or stoichiometric melts, Wagner et al.9 have provided support for the ASQC centers. In recent years, a great deal of effort has been made in seeking experimental evidence for the Ga antisite defects and to examine the role of 5 impurities in GaAs.31"32 Detailed IR measurements and its analyses have led Woodhead et al.32 to argue that in addition to the isoelectronic BGa defect lines (-540 and 517 cm"1) in GaAs, the LVMs detected earlier at (626 and 601 cm"1) due to B(2) centers actually arise from defects with antisite wBAs (I!BAs) structure. The experiment that supported the analyses of Woodhead et al.32 and provided direct evidence for the BAs centers is the work of Gledhill et al?1 where the 601 cm"1 (628 cm"1) transition has been resolved by FTIR into four or possibly five fine-structure lines. Comparison with similar measurements by Theis et al33 for 2S SiAs and 12CAs in GaAs concluded that the line at ~ 601 cm"1 originates from the Gaisotopic masses in different NN configurations (see : Fig. 2 (upper panel)). In support of the experimental observations, the calculated relative band strengths (see : Fig. 2) of the lines are found directly proportional to the natural abundance of the Ga isotopes involved in various arrangements (Table 5.1).
Table 5.1 Ga isotopic masses involved in different NN configurations of the GaA?: u B system. Considering the abundance of 69Ga (71Ga) at 60.4 (39.6) at %, the probability of the LVMs appearing in each configuration is also given.
System Ga4- B A , 69~„
11.-,
71~
Ga 3 -
BAJ-
Gai
ua2-
BAS-
ua2
Gar
BAS-
'Ga4-
Ga3
BAS
Symmetry
Mode
Probability factor for each LVM line
T„
F2
C, v
Ai, E
-C3(l-C),-C3(l-Q 3 3
0.116, 0.233
C-2v
A,, B,, B 2
2C 2 (1-C) ! ,2C ! (1-C) 2 , 2C 2 (I~C) 2
0.114, 0.114 0.114 0.05. 0.10
c4
c3v
A,,E
|C(1~C)3,|C(1-C)3
Td
F2
(1-C) 4
with C = 0.604 0.133
0.0246
510 5.3
C-complexes in AlAs
In highly C-doped AlAs samples, besides the CAs acceptor local mode near 630 cm"1, Davidson et al.43 have also observed several satellite lines near ~ 616.7, -632.8, -633.7, and -645.9 cm"1. The intensity of the line near -617 cm"1 was stronger than that at -646 cm"1 and it was probably even stronger than the intensity of the partially resolved feature near -633 cm"1. These observations led the authors of Ref. [43] to suspect that the mode at - 617 cm"1 is a donor CAI and that the weaker lines arise either from pairs of Cacceptors, complexes, or donor-acceptor pairs in close proximity. From theoretical standpoint, on the other hand, our calculation for an isolated C donor in ^4/As has provided a LVM frequency of -535 cm"1 at a much lower value than the speculated feature near -617 cm"1. Therefore, we believe that other possibilities for defect centers involving C and /or impurities need to be invoked to understand the origin of the four satellite lines observed in highly C-doped AlAs. We consider the following cases. Donor-acceptor CA/- CM pair If C occupies both Al and As sites in AlAs, then there might be a tendency for the two types to aggregate and form nearby donor-acceptor pairs. Four impurity vibrations are expected for this NN complex of Cj v symmetry. Our study of lattice relaxation effects for the isolated CAs acceptor and CAI donor using bond-orbital model give C-Al and C-As bond lengths of 2.09 and 2.13 A, respectively. The C-C bond length in the donoracceptor pair is rather small (1.6 A). In fact it is very close to the C-C bond length in the diamond lattice. The perturbation parameters for the C donor-acceptor pair in AlAs are calculated by using the force variation correlation with the bond length. These parameters are used in the Greens function theory to obtain the four LVMs (see : Table 4.1). Perusal of Table 4.1 reveals that the stretching (A*) mode of the NN C-pair is far from the CAs(F2) mode. As a matter of fact, none of the calculated local modes of the NN C pair are closer to the observed satellite lines. We strongly believe that the IR features (~ 616.7, -632.8, -633.7, and -645.9 cm'1) cannot be assigned to the NN C-donor-acceptor pairs. Carbon acceptor CAS-AI-CAS pair The second possible defect center that we have considered here is a 2NN C-acceptor (CAS-AI-CAS) pair of C2v symmetry. As the degeneracies are lifted, six vibrational modes are expected for the C-acceptor pair in AlAs. These modes, in the Greens function frame work, are obtained with the set of parameters given in Table 4.1. We will discuss these results in Sec. 6.3. 5.4
Silicon pairs in GaAs
The NN donor-acceptor 'SiGa-SiAs' pairs in GaAs exhibit C3v symmetry and four LVMs are expected. Earlier IR measurements 5,6 ' 41 have identified two symmetric modes near 464 cm"1 (A,+) and 393 cm"1 (E*~), while it is speculated that the bands observed near
511 367 and 369 cm"1 belong to two other unidentified but S;-related defects which will be considered later in this section. Although anti-symmetric modes were not observed, it was, however, erroneously thought41 at one time that the line near 367 cm'1 was another Si-pair mode. Using the same values of force variations for the isolated cases of Si-defects (c. f. Table 4.1) in the Greens function theory we were able to reproduce the A,+and E? modes in excellent agreement with the observed ones. Moreover the isotopic shifts of the A,+mode for 28SiGa-30SiAs,30SiGa-28SiAs, and30SiGa-30SiAs pairs to 455.2, 457.3 and 448. cm"1 are in good accord with high resolution measurements of Theis and Spitzer.24 With the present choice of the perturbation model, our calculations predict the E mode near 323 cm"1 and A," mode close to the maximum phonon frequency of GaAs. Let us now return to the absorption bands (~ 367 and 369 cm'1) which we believe are originating from separate centers, each involving one silicon and a native (vacancy or antisite) defect. Although there are no definitive interpretations for these modes, the earlier low resolution IR data showed that the FWHM for the 367 cm"1 band is significantly smaller (~1.5 cm"1) than that of 2 cm"1 for the 369 cm"1 line. By correlating the absorption strengths of these lines to pre-compensation values of carrier densities, the 367 cm"1 line was best fitted by assuming it to be a donor, whereas the 369 cm"1 line was fitted by assuming it to be an acceptor, although this evidence is felt to be statistically inconclusive. High resolution FTIR measurements of Theis and Spitzer24 find no splitting in the former (367 cm"1 : A band) line, which retained a width of ~ 1 cm"1, while the later band is split into three resolved components (368.31 cm"1 : B line; 369.53 cm"1 : C line; and 370.7 cm"': D line). From a variety of evidence, including the large FWHM of the A band near 367 cm"1, it was concluded that the SiAS-Asaa defect was the most probable, as this would create an (unresolved) broadening of the A line due to combinations of the three NN Ga isotopes to such a center and, furthermore, would have a donor behavior (see : Table 5.2). Newman 5 has, however, claimed that he has resolved the 367 cm"1 band into a doublet which is compatible with its identity being Sioa-Ga^ with two isotopes of Ga creating the two features in the doublet. Table 5.2 Ga isotopic fine structure of LVMs (bands B-D) in cm"1 observed in the high resolution FTIR measurements Ref. [24] (probably) due to 28SiAs-Asca pair defect in GaAs (see text)
Defect Center 69
M
Ga 3 - Si A ,-As G ,
69
28
Ga 2 - Si A ,-As Q , + 71 Ga,
Experimental
Calculated
D: 370.7
370.3
C: 369.53
370.12, 369.91
6<,
Gar28SiA,-AsG, +"Ga 2 Ga3-
SJAS-ASG.
369.69, 369.50 B: 368.31
368.94
512 Contrary to the observed electrical nature, such a pair would have a net acceptor behavior. Finally very little has been speculated about the identity of the B-C lines (the former S7-y center at 369 cm"1), except that since they occur only in electron-irradiated or Cu-compensated silicon-doped samples, involvement of native vacancies is possible. The 367 cm"1 LVM appears in all samples regardless of compensation methods.
5.5
Siar VGQ and BeGO-ASaa complexes in LT GaAs
Sica-^Ga complexes: IR absorption measurements in MBE grown 57-doped LT-GaAs samples show LVM of 28SiGa donors (c. f. Fig. 3) with an integrated absorption coefficient corresponding to [SiGa] = 1.5x 1019 cm"3. No absorption was detected44 from SiAs acceptors, SiGa-SiAS 'donor-acceptor pairs' or from the complex Si-X, thought to be a S'AS- VGO complex, so that, there was no evidence for Si impurities occupying ^s-lattice sites. However, weak LVM lines are observed at 366.7, 367.5 and 397.8 cm"1 from a defect labeled Si- Y and ascribed to SiGa-VGa second-neighbor pairs. As the samples are found almost compensated, it is inferred44 that the concentration of the unpaired VGa is ~10 1 9 cm"3. BeGa-AsGa complexes: We measured45 the LVMs from five GaAs samples grown at different temperatures (200 °C - 325 °C), all nominally doped with ~ 7xl0 19 cm"3 of Be. In an earlier study of LEC GaAs doped with high concentration of Be we observed a sharp and pronounced absorption near-482 cm"1. To carry out IR measurements, the sample was irradiated with 2 MeV electrons to remove the free carrier absorption. A similar study of LVM induced by Be near 482 cm"1 in high temperature MBE GaAs has been reported by Wagner et al.66 The absorption data for LT GaAsrBe samples revealed that for samples grown at increasingly lower temperature, the total integrated absorption of BeGa LVM is also decreased. Also at lower temperature (-275 °C) a low frequency shoulder near BeGa LVM started to emerge (spreading over ~2 cm"1 in the LVM structure) which could not be fully resolved.45 However, our secondary ion mass spectroscopy results indicate that the concentration of Be impurities are approximately constant for all the samples. The additional structure in the LVM suggests that the symmetry at BeGa is lowered. This lowering of symmetry may arise from the interaction of BeGa with either non-substitutional Be or intrinsic defects (Ast or AsGa). The stress-strain in the GaAs layers is ruled out as being the cause of the structure in the BeGa LVM. Based on our Green's function calculations (see : Table 4.2) we have shown that the isolated intrinsic defects (e.g., VGa, AsGa, etc.) in GaAs can only produce low frequency inband modes. However, these intrinsic defects can cause drastic modifications in the LVMs of BeGa or SiGa if involved in the formation of 2NN pairs. In order to explain the additional structure observed in Se-doped LT GaAs, we have considered two models involving either Ast or AsGa with BeGa. In the first case of a BeGa-Ash complex involving interstitial, the Ast occupies a site which is displaced from
513 BeGa acceptor along a [100] crystallographic direction. In this configuration of C2V symmetry, two of the vibrational modes of the complex would have displacement vectors in the plane defined by the Be atom and its two As NN on normal lattice site. Both of these modes would involve bond stretching and so they are expected to have much higher frequencies. On the other hand, the third mode with displacements of the Be out of this plane would have involved primarily bond bending and so a much lower frequency is expected. The calculated spread in the three frequencies was found much larger for the BeGa-Ash pair in our Greens function calculations. However, for SiGa-VGa and BeGa-AsGa complexes of C, symmetry, the theory provided features of the LVMs in reasonably good agreement with the observed values in Si- and Be- doped LT GaAs.
6.
Discussion
6.1 Physical significance of the perturbation parameters In Table 4.1, we have reported the results of our calculations for the LVMs in the F2 representation due to substitutional impurities in GaAs and AlAs. In each case, the table lists the relative change in force constant [t for defects occupying the Ga {AT) site or u for defects occupying the As site in GaAs (AlAs), respectively] between the impurity and its NN, which reproduces the experimental mode frequency. In contrast to the earlier findings,54 our results for isolated defects in semiconductors suggest that the relative variations of NN force constants are larger even for the isoelectronic impurities and quite appreciable for the charged defects. Bellomonte67 has suggested a possible size effect for the large softening in GaAs:B, as covalent radius of B (0.88 A) is much smaller than the replaced Ga (1.26 A) atom. Although we do find appreciable softening in GaAs, stiffening is indicated for B in InP.68 Further evidence that size of the substitutional atoms does not dominate bond strength can be found for N and C in III-V compounds, both of which have radii smaller than B.52 In other words, our study finds no correlation with the signs of t or u and the size of the impurity-host atoms. However, it reflects a large variation in the values of t or u as we pass from isoelectronic (/) to charged [donor (ct) or acceptor (a')] substituents of comparable masses. After analyzing nearly eighty cases of LVMs in fifteen III-V (II-VI and IV-IV) semiconductors, we have established the following trends for the closest mass a, i, and ct defects occupying III(II) and V(VI) in III-V(II-VI) compounds : At
( annm ~'///(«)}> ° (softening)
At
( d!,«m ~ W > ) < ° (stiffening)
,
(13a)
,
(13b)
- ivm}
< 0 (stiffening)
,
(13c)
Au{ dym - iym}
> 0 (softening)
.
(13d)
Au{ dvm
514
These trends in force constants are found independent of the long range Coulomb interactions, and we strongly argue that charged impurities in semiconductors affect only the short range forces via the redistribution of the electron charge density. To further stress this point, let us examine the bonding mechanism in unperturbed host lattices. For iso-row compounds Ge-GaAs-ZnSe (c.f. Table 1.1b) (or Sn-InSb-CdTe), the covalent bonding is apparent in Ge where the charge is centered midway between the two atoms. Vox partially covalent GaAs, the bonding charge is displaced towards the As atom and, in partially ionic ZnSe, nearly all the charge is centered near the Se atom (see : Fig. 9 a). If 27Al (0 is substituted for the Ga site in GaAs, the Al-As bond is represented by a cloud of valence electrons located near the As region. Replacing 27Al by an acceptor 26Mg (a') will cause the electrostatic forces to shift the electron cloud towards the As core, the bond becomes more ionic, and we observe its softening (see Fig. 9 b). On the contrary, if 27 Al is replaced by a donor 28Si {a*), its added additional positive charge will attract the electron cloud, increase the covalency of the bond and consequently we observe its stiffening (see : Table 4.1). It can be noted that for a', i, and dt defects occupying the V
Ge
Gfe GaQ
a a 4
d "®
-^DAs Zn
Se
C^oAs
As
Ga O—^^—® a'
Ga (b)
Fig. 9
Force-variation correlation with bond ionicity (covalency) in semiconductors: (a) perfect system (b) imperfect systems (our observations)
(VI) site in III-V (II-VI) compounds, the shift in the electronic charge density and thus the change in the covalency of the bond occurs in just the opposite way- a result in complete corroboration with our calculated trends of force variations. Typically, the absolute values of the relative variations of At and Au for single charge (a", c?) and isoelectronic (i) defects producing LVMs in all III-V (II-VI) compounds are found within 15-30%. On the
515 other hand, for double acceptor (a2' , e.g., Li) and isoelectronic (i, e.g., B) defect, we find At to be approximately twice as large compared to single acceptor (a e.g., Be) and isoelectronic B occupying Ga site in GaAs. The present simple physical understanding of the bonding mechanism in terms of the general magnitude of the 'impurity-hostparameters' ( t or u) has provided us with the most productive means of establishing and identifying the microstructure features of defects and their relationship to the optical experiments. From Eqs. 13 (a-d), it can be noted that there is a net correlation between the force perturbation and modification of the bond ionicity (or covalency). It is already shown qualitatively that the increase or decrease of the force variation due to charged impurities varies in the same sense as doe the increase or decrease of the covalency of the bond. The impact of the increasing or decreasing bond ionicity (covalency) on the interatomic forces in semiconductors may also be determined from its comparison with the short range coupling between first nearest neighbors (parameters A and B) as obtained from the lattice dynamics. In Fig. 10, we have plotted the parameter A in our RIM 11 against the interatomic distance r0 for various elemental and compound semiconductors. Similar
2.3
2.4
2.5
2.6
2.7
2.8
r.(X> Fig. 10 Variation of the NN force constant ,4 of the RIM 11 versus interatomic separation r0 for various I-VII, II-VI, III-V and IV-IV crystals.
516 behavior is noticed if B versus r0 is plotted. From Fig. 10, it is clear that two parameters are important for characterization of the bonding mechanism in semiconductors : bond ionicity (covalency) and bond length. It can be noted that (a) in groups II-VI, III-V and IV-IV the coupling A decreases slowly with increasing bond length rm and (b) for iso-row compounds, e.g., Ge-GaAs-ZnSe-CuBr, where the bond length r0 is practically constant but the coupling A decreases slowly with increasing ionicity or decreasing covalency. This behavior provides an added support to the fact that the increase or decrease of force perturbation due to charged impurities varies in the same way as the increase or decrease of the covalency of the bond.
6.2
Role of silicon, carbon in GaAs, AlAs and AlxGai^/ls
In contrast to the acceptor nature of carbon defects, silicon impurity in GaAs, depending upon its doping level, acts like an amphoteric defect, i.e., a donor occupying the Ga site (SiCa) for the lower doping level and acceptor replacing the As atom (SiAs) for the higher doping level. The IR study in ^/As:Si samples grown by MBE has shown an absorption line near 331 cm"1. The peak is found at a much lower frequency than the LVM of SiCa (-384 cm"1) in GaAs. Based on our Greens function calculations along with correlation of an integrated absorption of the line to the Si dopant and to the free-electron concentration, the observed satellite feature near 331 cm'1 is assigned to the gap mode of SiM donors (Tj symmetry) in AlAs. Similar to GaAs, Si at higher doping level is expected to act as an acceptor in AlAs. However, to the best of our knowledge, there exist no IR measurements demonstrating SiAs in AlAs. For lower doping level, while Si impurities in GaAs and AlAs behave like shallow donors, in AlxGai_xAs they are known to cause deep donors or DX centers for x > O.2.69 Since DX centers tightly bind electrons, their physical properties are expected to be very sensitive to their local environment. In fact, the transformation of shallow silicon donors in GaAs to deep DX centers in AlxGai_xAs has been suggested to be related to the changes in the local atomic configurations.70 Although an exact relationship between electronic states and atomic configurations has yet to be established, it is believed that Si atoms in AIxGa|.xAs are transformed into DX centers by the influence of neighboring^/ atoms.71 Such local atomic configurations around Si atoms can be revealed by measuring the LVMs of impurity atoms. Since the impurity LVMs in AlxGai.xAs:Si exhibit peak splitting and peak broadening, it is very difficult to detect impurity vibrations especially at higher Al composition.72 For low-^4/ composition in AlxGa!_xAs:Si the LVMs have been predicted theoretically.73 Since carbon is a group-IV element, an amphoteric nature in III-V compounds was expected. Despite some recent debates,43 there is still no direct experimental evidence for the existence of C donors in GaAs and AlAs. In GaAs, a line from l2CGa donor (if it exists) is expected to lie at a frequency lower than the CAs acceptor (582 cm'1). This
517 conjecture of a C-donor line is based on LVM frequencies for S7Ga donors at 384 cm and 399 cm"1, but it will show no fine structure since the 75As nearest neighbors are 100% abundant. In highly doped C in GaAs films grown by MOMBE, Woodhouse et al62 observed two lines: one at -452 cm"1 line X (FWHM = 1.4 cm"1) and the other at -563 cm"1 line F(FWHM = 3.2 cm"1) - one of them was thought to be related to I2Caa donors. It should be noted that as-grown MOVPE C-doped GaAs films containing hydrogen also show these lines. Introducing deuterium into the samples has led to the observation of a satellite feature near 440 cm"1. Based on polarization dependent Raman measurements, the authors of Ref. [62] have assigned the mode near 440 cm"1 to the A,"vibrations of CAsD passivated pair with C3v symmetry - an isotopic analog of the X line. On the other hand, similar measurements have not yet revealed the symmetry of line Y. This is believed to be due in part to the background resonant scattering from the 2LO phonons of intrinsic GaAs. Thus it is not known at the moment whether line Y is due to E? mode of the CAsH pair or to the F2 mode of the Coa donor. It is clear, though, that line Y is very clo se to the calculated LVM of CGa in GaAs. On the other hand, the absorption line associated with C donors is expected to have a FWHM significantly smaller compared to the CAs acceptor, as it should show no fine structure due to its 7SAs NN with 100% natural abundance. Since the FWHM of line Y is nearly five times larger than that of the C acceptors, its assignment to C donors precludes discrimination on the basis of observing NN isotopic fine structure. On the other hand, based on the observed variation in the strength of line Y followed by various heat treatments of the samples, the possibility of relating line Y to the CGa donors in GaAs has not been completely ruled out. We have also calculated the LVMs of CAs-H(D) pairs in GaAs with hydrogen (deuterium) in the bond-centered (BC) site. The results of our study provided H(D) and C modes in the complex to be at -2633 (1921 cm"1) (A,: stretch), 541 (528 cm"1) (E : wag), -491 (481 cm"1) (A,) and -272 (272 cm"1) (£), respectively. The details of this calculation will be reported elsewhere. 6.3
Carbon pairs in AlAs
The IR measurements performed in heavily C-doped AlAs materials have revealed four C-related satellite lines near 616.7, 632.8, 633.7 and 645.9 cm"1. From the observed number of bands, one expects a complex center with Cj v symmetry; however, the bands are assigned to a nearby CAs-Al-CAs acceptor pair. Our calculations for the CArCAs pair provides a stretching mode at a very high frequency due to the large stiffening between the C-C bond. The frequency of the mode falls far away from the observed bands, and we exclude this center as a possible candidate for any of these satellite lines. A pair of C acceptors in close proximity exhibits a point group symmetry of C2v. Since the degeneracies at the sites are lifted, six modes are expected. Using the parameter values of Table 4.1, we have predicted all six local modes for the C-acceptor pair. Although the difference between the highest and lowest modes (-40 cm"1) in our calculation compares
518 (-29 cm"1) well with the experimental data, the extra two lines predicted by the theory have not been observed. More IR measurements of the isotopic shift of local modes due to a pair of 13C acceptors are needed to confirm this assignment. At present, we are also looking into the possibilities of other centers with C3v symmetry, including CAs paired with C interstitials (CAs-C,).
7. Summary In conclusion, we have demonstrated that in semiconductors, the LVM spectroscopy is a valuable tool in providing information about the sites occupied both by intrinsic and light impurity atoms, as well as the properties of complexes incorporating these atoms. The experimental measurements have been made possible by the high resolution and low noise FTIR spectrometers. Equally important in these studies are the availability of high-quality, strain-free bulk and epitaxially grown materials. The most significant step forward has been the identification of the number and type of nearest neighbors of impurities deduced from the fine structure (or lack of it) produced by the various combinations of 69Ga and 71Ga host lattice atoms in GaAs. It is true that the symmetry of such complexes can be determined, their identification relied strongly on supplementary data including realistic model calculations. In this paper, we have shown that both infrared and Raman techniques can be used to identify the site selectivity of light isoelectronic, donor, acceptor impurities in materials (GaAs, AlAs and AlxGai_xAs) grown by MBE. More recently, these techniques have been applied to study the role of Si donors, Be acceptors and intrinsic defects in epitaxially grown group Ill-nitrides, thereby broadening the scope of LVM spectroscopy even further.
8. Acknowledgments The author gratefully acknowledges useful discussions on the subject matter with Dave Fischer of the Materials Directorate, and Dave Look, Keith Evans, Ron Kaspi, C. E. Stutz, B. Jogai, Don Reynolds and John Loehr of the Solid State Electronics Directorate, Wright Laboratories, Wright Patterson Air Force Base, Ohio. Part of this work was performed at Solid State Electronics Directorate, Wright Laboratories, Wright Patterson Air Force Base, Ohio and was funded in part by the National Research Council Associateship Program and by the American Chemical Society (Petroleum Research Fund # PRF 30145-B3). The work at Indiana University of Pennsylvania was supported by the Cottrell College Science Award (Research Corporation) and by the National Science Foundation Grant # ECS 9906077.
519 9.
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.
A. G. Milnes, Deep Impurities in Semiconductors (Wiley, New York, 1973). A. M. Stoneham, Theory of Defects in Solids (Clarendon, Oxford, 1975). S. T. Pantelides, Rev. Mod. Phys. 50, 797 (1978). M. Jaros, Adv. Phys. 29, 409 (1980). R. C. Newman, Infrared Studies of Crystal Defects (Taylor and Francis, London, 1973); ibid Semiconductor andSemimetals 38,117(1993). W. G. Spitzer, Festkorperprobleme XI, edited by O. Madelung (Pergamon, New York, 1971), pp. 1-44. A. S. Barker and A. J. Sievers, Rev. Mod. Phys. 47, S1 (1975). B. Hamilton, Landolt-Bornstein (O. Madelung and M. Schulz, eds.), vol 22b, (Springer Verlag, Berlin 1989) pp. 583-594. R. J. Wagner, J. J. Krebs, G. H. Stauss, and A. H. White, Solid State Commun. 36, 15 (1980). T. A. Kennedy and N. D. Wilsey, Phys. Rev. Lett. 41, 977 (1978); ibid Phys. Rev. B23, 6585 (1981). G. D. Watkins, in Defects and Radiation Damage in Semiconductors (J. E. White, ed.) (IOP, London, 1973) p. 228. P. van Engelen, Phys. Rev. B22, 3144 (1980). U. Kaufman, J. Schneider, and A. Rauber, ,4pp/. Phys. Lett. 29, 312 (1976). D. V. Lang, R. A. Logan, and L. C. Kimerling, Phys. Rev. B15,4874 (1977). R. H. Wallis, A. Zylbersztejn, and J. M. Besson, Appl. Phys. Lett. 38, 431 (1981). D. K. Maude, J. C. Portal, R. Murray, T. J. Foster, L. Dmowski, L. Evans and R. C. Newman, P. Basmaji, P. Gibart, J. J. harris and R. B. Beall, Physics ofDXCenters in GaAs Alloys (J. C. Bourgoin, ed.) Sci. Tech . Pub., Liechtenstein p. 121 (1990). L. Pavesi and M. Guzzi, J. Appl. Phys. 75, 4779 (1994). J. A. Van Vechten, Handbook on Semiconductors (T. S. Moss and S. P. Keller, eds.) (North Holland, Amsterdam, 1980) vol. 3, pp.1-111; ibid J. Phys. C: Solid State Phys. 17, L933 (1984). see: The Technology and Physics of Molecular Beam Epitaxy (E. H. C. Parker, ed.) (Plenum, London 1985). G. Roos, A. Shoner, G. Pensl, J. Wagner, B. K. Myer and R. C. Newman, J. Appl. Phys. 69, 1454 (1991). J. I. pankove, D. E. carlson, J. E. Berkeyheiser and R. O. Wance, Phys. Rev. Lett. 51, 222 (1983). A. J. Tavendale, A. A. Williams and S. J. Pearton, Appl. Phys. Lett. 48, 590 (1986). J. Chevallier, B. Clerjayd and B. Pajot, Semiconductors and Semimetals vol. 34, pp. 447 (1991). W. M. Theis and W. G. Spitzer, J. Appl. Phys. 56, 890 (1984). R. Murray and R. C. Newman, Landolt-Bornstein (O. Madelung and M. Schulz,
520
26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43.
44.
45. 46.
eds.), vol 22b, (Springer Verlag, Berlin 1989) pp. 105-113 and 576-583; ibid R. Murray et al, J. Appl. Phys. 66, 2589 (1989a). R. Murray, R. C. Newman, R. S. Leigh, R. B. Beall, J. J. Harris, M. R. Brozel, A. Mohades-Kassi and M. Goulding, Semicond. Sci. & Technol. 4, 423 (1989b). P. S. Nandhara, R. C. Newman, R. Murray, B. Pajot, J. Chevallier, R. B. Beall, and J. J. Harris, Semicond. Sci. & Technol. 3, 356 (1988). R. C. Newman, F. Thompson, M. Hyliands, and R. F. Peart, Solid State Commun. 10, 505 (1972). J. D. Collins, G. A. Gledhill, R. Murray, P. S. Nandhara, and R. C. Newman, Phys. Stat. Sol. B151,469 (1989). F. Thompson, S. R. Morris, and R. C. Newman, Inst. Phys. Conf. Ser. 16, 371 (1973). G. A. Gledhill, R. C. Newman, and J. Woodhead, J. Phys. C: Solid State Phys. 17, L301(1984). J. Woodhead, R. C. Newman, I. Grant, D. Rumsby and R. M. Ware, J. Phys. C: Solid State Phys. 16, 5523 (1983). W. M. Theis, K. K. Bajaj, C. W. Litton, and W. G. Spitzer, Appl. Phys. Lett. 41, 70 (1982); ibid Physica 117 & 118B, 116 (1983). A. A. Kachare, W. G. Spitzer, A. Kahan, F. K. Euler and T. A. Whatley, J. Appl. Phys. 44, 4393 (1973). Z. L. Akkerman, L. A. Borisova, and A. F. Kravchenko, Sov. Phys. Semicond. 10,590(1976). J. Schneider, B. Dischler, H. Seelewind, P. M. Mooney, J. Lagowski, M. Matsui, D. R. Beard and R. C. Newman, Appl. Phys. Lett. 54, 1442 (1989). C. Song, W. Ge, D. Jiang, and C. Hsu, Appl. Phys. Lett. 50,1666 (1987). P. C. Leung, L. H. Skolnik, W. P. Allred, and W. G. Spitzer, J. Appl. Phys. 43,4096(1972). O. Lorimor and W. G. Spitzer, J. Appl. Phys. 37, 2509 (1966). W. G. Spitzer abd W. Allred, J. Appl. Phys. 39,4999 (1968); ibid. App. Phys. Lett. 12, 5 (1968). P. C. Leung, J. Fredrickson, W. G. Spitzer, A. Kahan, and L. Bouthillette, J. Appl. Phys. 45, 1009(1974). W. G. Spitzer, J. Phys. Chem. Solids 28, 33 (1967). B. R. Davidson, R. C. Newman, D. A. Robbie, M. J. L. Sangster , J. Wagner, A. Fischer and K. Ploog, Semicond. Sci. Technol. 8, 611 (1993); B. R. Davidson, R. C. Newman, R. E. Pritchard, D. A. Robbie, M. J. L. Sangster, J. Wagner, A. Fischer and K. Ploog, Mat. Sci. Forum, vol. 143-147,247 (1994). S. A. McQuaid, R. C. Newman, M. Missouss and S. O'Hagan, Appl. Phys. Lett. 61, 3008 (1992); S. A. McQuaid, R. c. Newman, M. Missous and S. O.'Hagan, J. Crys. Growth 127, 515 (1993). D. N. Talwar, M. O. Manasreh, C. E. Stutz, R. Kaspi and K. R. Evans, J. Elect. Mat. 22,1445(1993). M. Kaminska, Z. Liliental-Weber, E. R. Weber, T. George, J. B. Kortright,
521
47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57.
58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73.
W. W. Smith, B. -Y. Tsauer and A. R. Calawa, Appl. Phys. Lett. 54,1881 (1989). F. W. Smith, A. R. Calawa, C. L. Chen, M. J. Manfra and L. J. Mahoney, IEEE Electron Device Lett. 9, 77 (1988). M. Kaminska, E. R. Weber, Z. Liliental-Weber, R. Leon and Z. U. Rek, J. Vac. Sci. Technol. B7,710 (1989). H. -J. Sun, G. D. Watkins, F. C. Rong, L. Fotiadis and E. H. Poindexter, Appl. Phys. Lett. 60, 718 (1992). D. E. Bliss, W. Walukiewicz, J. W. Ager, E. E. Haller, K. T. Chan, and S. Tanigawa, J. Appl. Phys. 711699 (1992). A. Mooradian and A. L. McWhorter, Phys. Rev. Lett. 19, 849 (1967). D. N. Talwar, M. Vandevyver, K. K. Bajaj and W. M. Theis, Phys. Rev. B33, 8525 (1986). K. Kline, Physica 116B, 52 (1983) and references therein. A. Grimm, A. A. Maradudin, I. P. Ipatova, and A. V. Subashiev, J. Phys. Chem. Solids 33, 775 (1972). D. N. Talwar and Bal K. Agrawal, J. Phys.Chem. Solids 39,207 (1978). D. N. Talwar, M. Vandevyver, and M. Zigone, J. Phys. C: Solid State Phys. C13, 3775 (1980); ibid Phys. Rev. B23,1743 (1981). A. A. Maradudin, E. W. Montroll, C. H. Weiss, and I. P. Ipatova, in Solid State Physics, 2 nd ed., (F. Seitz, D. Turnbull, and H. Ehrenreich, eds.) (Academic, New York, 1971). R. J. Elliott, J. A. Krumhansl, and P. L. Leath, Rev. Mod. Phys.46, 465 (1974). K. Kunc, Ann. Phys. (Paris) 8, 319 (1973-74). D. Strauch and B. Dorner, J. Phys. Condens. Matter 2, 1457 (1990). W. A. Harrison, Electronic Structure and the Properties of Solids (Freeman, San Francisco, 1980). K. Woodhouse, R. C. Newman, T. J. de Lyon, J. M. Woodall, G. J. Scilla, and F. Cardone, Semicond. Sci. Technol. 6, 330 (1991). H. Ono and T, Baba, Phys. Rev. B44, 10895 (1991). H. Ono (private communication). B. Clerjaud, F. Gendron, M. Krause, and W. Ulrici, Phy. Rev. Lett. 65,1800 (1990). J. Wagner, M. Meir, R. Murray, R. C. Newman, R. B. Beall and J. J. Harris, J. Appl. Phys. 69,971 (1991). L. Bellomonte, J. Phys. Chem. Solids 38, 59 (1977). M. Vandevyver and P. Plumelle, Phys. Rev. B17, 675 (1978). P. M. Mooney, J. Appl. Phys. 67, Rl (1990). D. J. Chadi and K. Chang, Phys. Rev. B39,10063 (1989). T. Baba, M. Mizuta, T. Fujisawa, J. Yoshino, and H. Kukimoto, Jpn. J. Appl. Phys. 28,L891(1989). H. Ono and T. Baba, Phys. Rev. B42,11423 (1990). D. N. Talwar and M. Vandevyver, Phys. Rev. B40, 9779 (1989).
This page is intentionally left blank
THE GENERAL NON-RADIATIVE ENERGY TRANSFER MASTER EQUATIONS FOR CRYSTALLINE MATERIALS, THE EXACT SOLUTION AND CURRENT MODELING LUIS A. DIAZ-TORRES, ORACIO BARBOSA-GARCIA, M. ANTONIO MENESESNAVA Centro de Investigaciones en Optica A.C., Loma del Bosque 115, Lomas del Campreste, Leon Gto. C.P. 37150, Mexico C.W. STRUCK Osram Sylvania, 71 Cherry Hill Drive, Beverly MA 01915, USA and B. DI BARTOLO Physics Department, Boston College, Chestnut Hill, MA 02167, USA.
ABSTRACT The general non-radiative energy transfer coupled equations are solved for the case of doped crystalline materials. In this work the exact solution is reported and the microscopic and macroscopic dopant, donor and acceptor, emissions are calculated. Our calculations are compared to major models such as the Forster-Dexter, Yokota-Tanimoto and Burshtein. It is shown that our modeling unifies and extends the predictions obtained by those authors. Our calculations consider the discreteness of the crystalline materials, and more than one interaction can be assumed to drive each of the processes that are taking place among pair of dopants.
1. Introduction Since the 1920's, research groups have been modeling the non-radiative energy transfer processes in different materials. "9 There are many technological and fundamental reasons to study the mechanism of the transfer of energy. For example, during the past decades those processes had played a major role in increasing the pumping efficiency of the active ions in some solid-state laser materials. Another example is the leading role played by glasses doped with rare earth elements in communications. A typical example is also seen in the fluorescent lamp phosphor Ca5(P04)3(F,Cl): Mn+2, Sb+3, chlorofluoroapatite. In this phosphor, Sb+3 absorbs the Hg radiation and partially emits in
523
524
the blue and partially transfer to Mn+2, which emits in the yellow, so that the combination is whitish. It is clear that by getting a better understanding on the nature of such processes new luminescent materials could be developed. Modeling the non-radiative energy transfer processes is not an easy task since for any given host material more than one process can take place. In addition, each process can be driven by more than one interaction among the impurities, i.e. dopants. Macroscopically, such interactions modify the dopant emissions from a straight line in a semilog scale of intensity and time. In fact, new measurements have shown that dopant transients are very complex showing that the interactions among dopants can be complicated. In principle, to model the non-radiative energy transfer processes one assumes that dopants are randomly distributed in a host material, and distinguishes two kind of ions. One kind absorbs the pumping energy and gets excited (Donor, D) while the other kind is in its ground state (Acceptor, A). As result of the transfer, the excitation moves from the D to the A ions. By assuming an interaction among dopant ions to drive the energy transfer, the individual D and A emissions have to be calculated. An averaging over these emissions has to be performed to be compared with experimental data. Current models use different approximations at different time stages of the donor emission transient. That is, the transient is divided into three temporal regions, at short, at intermediate and long times. Thus, with these temporal divisions, it is claimed that it is possible to understand the nature of the processes that are taking place and of the interaction between dopant pairs. Further, major models consider that only two kind of transfer processes take place among dopants at the same time, for example direct energy transfer and migration of the energy among donors. Furthermore, it is assumed that only one single interaction drives each of the transfer processes. In this work, the general master equations for the non-radiative energy transfer processes among dopants in crystalline media are solved in their exact form. That is, in addition to direct energy transfer and migration among donors, the back transfer from acceptors to donors and the migration of the energy among acceptors are considered in our solution. The corresponding analytical expressions for the macroscopic donor and acceptor emissions are reported. A comparison between our solution and the most important models in the literature is presented. It is shown that our modeling accounts for experimental data with no time stage divisions as current models do. Contrary to current models, our model approaches experimental data at any time and gives the nature of the interactions that drive the different transfer processes. The interaction that drives the transfer of energy can be assumed as a single or as a sum of multipolar interactions, or any sum of interactions. One further feature of our model is that the acceptor emission is calculated from the coupled master equations. That is, in our calculation the donor and acceptor emissions are dependent from each other. The general energy transfer master equations (GETME) are defined in Sec. 2. The main features of two of the most used models for the direct energy transfer are introduced
525
in Sec. 3.1.1 and 3.1.2, i.e., the models developed by Forster and Dexter and GolubovKonoveev-Sakun. The exact solution of GETME for the direct energy transfer is obtained in Sec. 3.1.3. Some experimental data are fitted with our solution in Sec. 3.1.4, and it is shown that a sum of interactions to drive the transfer of energy among dopants gives the best fit. The migration of the energy among donors and the main current models are presented in Sec. 3.2.1. The transfer function for current models is introduced in Sec. 3.2.1. In Sec. 4, the general solution to GETME is introduced. With the general solution one can assume that more than two processes can take place and that, more than one interaction drives the transfer of energy among dopants. In particular, the solution for two simultaneous processes, i.e., direct energy transfer and migration among donors is considered in this Sec 4.1. For these two processes it is shown how well our calculations predict the donor transients calculated from the Yokota-Tanimoto, Burshtein and supermigration models. Identical free parameters were used for the calculations of the donor emissions. That is, it is shown that with a single model one is able to calculate the donor emission predicted under different approximations developed by different authors. Finally, our concluding remarks are given in Sec. 5.
2. The General Energy Transfer Master Equations The non-radiative energy transfer processes take place between pair of ions, e.g., D-to-D, D-to-A, A-to-A ion pairs. It is assumed that dopants are randomly distributed in the host material, and for our modeling, it is considered that the host is a crystalline material. Therefore, each dopant is placed at a specific crystal site, and it is labeled with an index. That is, D; and Aj represent the ith and jth donor and acceptor, respectively. In principle, when the pumping energy shines the crystal sample at t=0, donor ions are excited from their ground level. From this excited state, they partially emit and partially transfer energy to donors and to acceptors as shown in Fig. 1. Thus, the excitation dynamics for Dj and Aj ions at any other time is given by the solution of the General Energy Transfer Master Equations (GETME):
<(0_ dt
- +2X1+ •DO
<(0_ dt
/=1
±wL # « + Z < ^ « + Z ^ % ( o (i) Dm Di
m=\,m*i
ND Aj '•AO
i=\
IK i=u*j
%«)+ iw^(o+i>#£(o (2)
526
In the above equations P^,{t) [P^(f) ] is the probability that the Dj [Aj] ion is still excited at time t. TDO [^AO] is the donor [acceptor] lifetime measured when there is no transfer process. NA and ND correspond to the number of acceptor and donor ions, respectively. The transfer rates, involved in both equations and in Fig. 1, are as follows. For Eq. (1), 1/TDO is the transfer rate due to the donor radiative emission. W$ is the non-radiative transfer rate from Di to Aj and defines the process known as direct energy transfer. W^m is the non-radiative transfer rate from Di to Dm and defines the process known as migration among donor ions. WAj' is the transfer rate from Aj to Di and defines the back transfer process from acceptor to donor ions. Similar definitions can be followed for Eq.(2).
Fig. 1.
Different non-radiative energy transfer processes between two species of dopants randomly placed in a crystalline lattice. Donors are in black and acceptors in light gray.
All transfer rates considered in Eqs. (1) and (2) only depend on the position of dopants at their corresponding crystal sites and the kind of interaction among dopant ions is shown below. For a shorted and general notation, the transfer rate from the rth ion of species X to they'th ion of species Y (with X,Y=D,A), W^ , is given by 2'
» ^ ( ^ ) = —exp 2R J
T
(R
Rf
XY\6
(Ri,) 6
(R|T
(Ro x , Y o) 10
(R„) 10
(3)
527 The first term on the right corresponds to an exchange transfer rate and the second to a multipolar transfer rate. An exchange interaction and a multipolar interaction drive these two kinds of transfer rates, respectively, between any pair of dopants. The former interaction is short-range interaction meanwhile the later is an expansion of the Coulomb's potentials among pair of ions. The free parameters for the exchange interaction are R0 and L , the critical transfer distance and the effective Bohr's radius, respectively. For the multipolar interaction there is only one free parameter for each term in Eq. (3), and corresponds to Ros (S = 6, 8, 10,...), the critical transfer distance. The S values define the dipole-dipole, dipole-quadrupole and quadrupole-quadrupole... interactions as the S value increases. Ry is the separation between the \th X ion and the )th Y ion; and txo is the lifetime of species X when no transfer process is present. It will be shown later that the parameters defined for each interaction in Eq. (3) are indeed the ones used by major models2_4'6"7 and by us. Let us remark that Eqs. (1) and (2) define a set of coupling equations, i.e., when calculating the donor and acceptor emissions from such equations, one emission is not independent from the other. Furthermore, one should take into account that any assumed process could be driven by a single term defined in Eq. (3) or a sum of terms if the experimental data looks complicated. Several models have been developed to get a better understanding of the non-radiative energy transfer processes. However, it is shown below that, to our knowledge, GETME have not been solved yet and any available model solves them in a restrictive way under different approximations. In the following sections the major current models are presented and compared to our solution to GETME.
3. GETME and current models Before we go through the main current models for the non-radiative energy transfer processes, let us comment some review articles concerning the solution of GETME. Powell 5 in 1984 considers Eq. (1) of the GETME as the very general expression for the evolution of energy with time away from an initially excited donor ion. It is claimed that the ideal way to theoretically interpret experimental results of energy transfer studies would be to find the exact solution to that equation. It is added that the general analytical solution has proven to be impossible and recalls that all reported solutions have involved highly restrictive assumptions. Further, it is pointed out that the most comprehensive theoretically treatment of this type is the work of Huber and coworkers.6"7 Henderson and Imbusch in their book published in 1989 also refer to Huber's work to give an account for the modeling of the energy transfer processes10. In such work, Huber used the average T-matrix formalism to solve Eq. (1) in a restricted form. For his modeling, the back transfer process is neglected and the acceptor ions are defined as
528 energy traps. That is, the coupling among GETME is broken and the calculated donor emission is independent of any acceptor emission. The main advantage of his results is that the same formalism can be used to treat, under different considerations, several cases such as those reported by Forster-Dexter,2 Yokota-Tanimoto3 and Burshtein.4 That is, his formalism could be used instead of the three models that are widely in use by the experimentalist. However, besides the restrictive solution to the GETME, all transfer rates calculated by Huber6 were done by considering a single interaction among dopants. This fact is also extensive to the other three major models.2"4
3.1. Direct energy transfer The direct non-radiative energy transfer process takes place between donoracceptor pairs and the transfer goes only in one direction, from donors to acceptors. Other processes considered in GETME, and in Fig. 1, are neglected. The most used model for this process is the one developed by Forster and Dexter. Inokuti and Hirayama12 reported similar calculations that extend the model for the exchange interaction. It is shown below that this model is reported in an analytical form by considering a continuos medium within the host material. In principle this model should not be applied to crystalline materials, however, the model has shown to be very useful when used in solid state laser materials such as Nd: YAG.
3.1.1 The Forster and Dexter model The pioneering work of Forster and Dexter (FD)2 was developed for the direct energy transfer by keeping only the first two terms on the right hand side of Eq. (1), and discarding Eq. (2). Therefore, the model reports no acceptor emission. The transfer process is driven by only one isotropic interaction. In principle, this interaction takes place among any pair of atoms in the host material, and can be single multipolar (dipoledipole, dipole-quadrupole, quadrupole-quadrupole,) or exchange. Further, to get the analytical expression for this model one has to consider several mathematical approximations.2'6'12 The main characteristic of this model resides in considering a uniform continuos distribution of acceptors around a single donor. However, the FD model is used for the analysis of the fluorescence emission of donors in both kinds of host materials, crystalline and amorphous. Further, this model has been extended to study other processes,3"6 but those models always start from a restricted form of the GETME as the FD does. Some other models assume a non-random distribution of ions, however, the continuos approximation for the acceptor distribution around donors still remains.13 Few models take into account the discrete character of the distribution of dopants in the crystalline host."' 14 " 19 Thus, under the FD model the GETME are reduced to solve the following equation,
529 1
NA
D0
j=\
l
Pmit)
(4)
Di
(5)
The solutions to this system of equations are,
^ ( 0 = ^/°exp -t
1 V ' DO
N
*
j=\
J
In Eq. (5) P^f are the initial conditions of excitation at t=0 for each donor ion. Once the individual excitation probability is known for each donor ion in the host sample, the macroscopic emission is determined by averaging over a large sample assumed to fill all the space. For the FD model it is assumed that the acceptor ions are uniformly distributed around each donor ion and that No « NA. If the sample is large enough, one can assume that each donor will see the same acceptor distribution. In consequence, the average of Eq. (5) over all donor ions within the host material is given by
PD{t) = P°D{t)exV- /
-M
(6)
Notice that Eq. (6) holds because the individual donor emissions are identical to each other. In order to perform the summation in Eq. (6) over the large number of acceptors within the sample, it is assumed that acceptors are distributed in a continuous way around a single donor Do. By considering a uniform continuous distribution function f(R) of acceptors around the donor, where R is the distance from donor to acceptor, the normalized average of Po(t) is obtained through a volume integral as follows,
™A
B=exp
Urn
\An f{R)R2 expi[-tWDA(R)\iR[
(7)
l
D0 J
Anf(R)R dR is the probability of finding an acceptor at a distance between R and .#+<#? from the donor, and RNA is the radius of a sphere containing NA acceptors. The solution to Eq. (7) has been reviewed by other authors.6,12,19 For the case of a single multipolar interaction the FD analytical expression is given by,
530
D = e x p ^ x
\
exp
DO J
3C,
Mc) 3^ s
(8) V '•ooy
CA is the acceptor concentration and T(l-3/S) is the gamma function where S takes the values 6,8,10,... for the dipole-dipole, dipole-quadrupole, quadrupole-quadrupole,... interactions, respectively. For the case of exchange interaction, the FD model gives the following solution12
(KAJ g\e
47t
exp
(9)
where y=2(Ro/L) and for ev(t/TDo)=z>0 g(z) can be approximated as, g(z) = 6zY —
{
^ ~
(10)
It should bee noted that the solutions expressed by Eqs. (8)-(9) do not allow to consider more than one interaction at a time. That is, the FD approximation does not allow the mixture of two or more interactions among dopants. Further, for the case of laser crystal materials and phosphors the FD model does not consider the discreteness of the host material.
3.1.2 Other models for the direct energy transfer Other models have been developed as an extension of the pioneering work of Forster and Dexter. Such is the case of the continuos Rotman-Hartmann model13 and the discrete model developed by Golubov, Konobeev, and Sakun (GKS).14"15 The former model assumes a nonrandom spatial distribution of dopants and solves the integral of Eq. (7). In order to have such distribution of dopants, it was introduced a positional correlation between donors and acceptors.13 An enhancing positional correlation increases the number of near acceptors, and thus increases the transfer rate among dopants. An excluding correlation decreases the number of nearby acceptors, and thus decreases the transfer rate. This extended model has been used by Rotman to interpret experimental results that cannot be explained by the FD model.20"21 However, the functional forms used in this model to obtain either enhanced or excluded correlation are somewhat nonphysical. Such positional correlation among dopants is not easy to justify, and Rotman had serious problems when applied to some experimental data.
531 Golubov Konobeev, and Sakun (GKS)14"15 reported a general donor emission decay expression for such process. Later, in a series of articles Blumen and Klafter made a generalization of the work reported by the Russian authors. With this modeling, it is shown that the FD analytical expressions are valid for low acceptor concentrations and that fails to predict the transient at very short times of the donor emission. The GKS model gives an exact statistical average for §o(t) for a crystal sample containing NA acceptors placed at the N possible lattice sites within the sample. Similar to the FD model, the GKS model takes Eq. (5) as the solution for the donor emissions of Eq. (4). Then the average over donors and acceptors is performed considering the probability P(K) for each possible configuration K of NA acceptors to occur within the possible N lattice sites, thus,0 =exp
'-t^
Z\P(K)Y[exp[-tW*}
(11)
V lD0 J
Notice that similar to the FD model, the GKS model implies to sum over the distribution of acceptors around one single donor ion, that is, in Eq. (11) around the donor DQ. Thus, if p is the probability that a particular site is occupied by an acceptor, then a specific distribution of NA acceptors placed within the N available lattice sites occurs with a probability given by P{K) =
pN'{\-p)h
(12)
then, K =exP (13) (
t \ N
• exp V ''DO J i=l
In order to compute expression (13) one can image a series of concentric spheres (shells) centered at the site where donor D0 is placed. Thus, the radius R| goes from donor D0 to the ith sphere where some acceptors are placed around that donor. Each shell has certain number of acceptor sites, thus, the ith shell contains Uj acceptor sites. From crystallographic tables, one can calculate the number of shells and the number of acceptors in each shell. Then, the analytical expression for the GKS model can be modified to, D =
ex
P
'-f^ V 'BO J
Y\{(\-p) + pexp[-tW*(Rj}}
(14)
532
From Eq. (14), it can be seen that the GKS model considers the discreteness of the host material and can be used for crystalline materials. However, the sum in this equation runs over all sites of the crystal sample that can be occupied by acceptors, and for practical calculations, one cuts it on an arbitrary way. From this equation, it can be seen that the model handles a single or a sum of interactions according to Eq. (3). That is, when calculating the donor emission from Eq. (14), one can assume that the direct energy transfer process can be driven by any sum of interactions. Stalder and Bass23 used this model to study the direct energy transfer from Nd to Er ions in garnet crystals. In particular, for the fixed neodymium concentration at 1.0 at. % two erbium concentrations, at 15 and 30 at. %, were considered in a YAG sample. The donor, neodymium ions, emission was recorded and it was found a fast initial decay, that was not measured in a previous work,24 followed by a long term non exponential decay. Therefore, it was assumed that the fast initial decay was driven by an exchange interaction among first neighbors, and the rest of the transient was due to a dipole-dipole interaction among Nd-Er pairs. To handle the sum of two interactions in Eq. (14), those authors introduced an arbitrary distance cut off. In their published work, they considered that the energy transfer rate was given by f nDA
V R°J J
»/
Roi>R,cutoff
(15)
W*(Raj) = -exp
1-
if
R0J
Rcuiqff is the introduced arbitrary cut off distance. For distances, from donor D0 to acceptors, larger (shorter) than Rcutqff, the driven interaction of the transfer of energy is only due to dipole-dipole (exchange) interaction. Although the fit obtained to the experimental data was satisfactory, we will show below that the arbitrary cut off distance is not necessary. It will be shown below that our solution of the GETME for the direct energy transfer mimics, under similar conditions, the GKS model at short and large times. It should be noticed that the GKS model does not give an expression to calculate the acceptor emission. 3.1.3 GETME for the direct energy transfer (DETME) For our modeling a great number of crystal samples are numerically generated. For each of these samples, dopant ions are randomly placed at their corresponding lattice sites. For the kth generated crystal sample there are ND donors and NA acceptors randomly placed by using a random number generator. Each donor ion has its particular acceptor distribution. Thus, after the excitation of the kth sample at time t=0, the dynamics for the excitation at the \th donor Di and the \th acceptor Aj is given by the solution of the Direct
533 Energy Transfer Master Equations (DETME). From Eqs. (1) and (2), those equations can be written as,
dt
T
.
<(0_ dt
1
2±
D0
.7=1
N„
1
— pf.frt+Yn
*£(')
(16)
yAjpk
(17)
Notice that the sums in DETME run through the number of donors and acceptors, and that the direct transfer rates from D-to-A ions, W£[, can be calculated by assuming a multipolar and/or an exchange interaction as shown by Eq. (3). By rewriting the DETME in a matrix form, their exact solutions are obtained straightforward. Thus, the individual probabilities of excitation, P^ (t) and PkA] (t) , for the ith donor and )th acceptor within the kth sample are given by, NA
Po,(t)-
-tfw*?
J
PkAj{t) =
J
\\ -exp •AOjJ
1
1
N
1
l
+ ^'°exp
(19)
*
-1^
' AO
P^f
(18)
'5>B ' Di *=1
WAj P*'° exp " Di Di
\ rAk
A ? exp
m=l
and P*f are the initial conditions of excitation after t=0 for each dopant ion. We
should point out that Eq. (19) is reported for the first time, and that the individual emission of acceptors is not obtained nor discussed within the FD approximation or any other FDlike model. Further, it is customary to consider that the acceptors are not initially excited, however, Eqs. (18) and (19) can handle acceptors initially excited or initially not excited. The former case is very important for the energy transfer up-conversion process. Once the individual excitation probabilities are known for each dopant in the kth crystal sample, we can determine the mean excitation probability that the donor ensemble remains excited at time t. The mean excitation probability value gives the average number of donors in the excited state and therefore is proportional to the normalized donor fluorescence from the kth crystalline sample,
534
ND
^ ( 0 = ^ ( 0 ; =(No
K,w NA
p*,0
p*,0
(20)
\
^y
V'- 1
J-l
/
If this probability is averaged over a large number of randomly generated crystal samples, the resulting average converges to the macroscopic donor fluorescence decay <j>D (t) , N„ Di
V ' DO
",<
y=i
y
«.
(21)
I^;° + I ^ 4/
V/=i
An analogous argument is valid for the acceptor fluorescence, and the normalized macroscopic acceptor emission is given by, AJ w pk-° n r
Di Di
1
t
N
-<
1
<
t^
^•°«
-exp
exp "A
\[ "a
NA
—---2X" IC+SC
^-r^ (22)
K-
IC+IC
In summary, to calculate the donor and acceptor emissions for the direct energy transfer process we have solved Eqs. (16)-(17), see Eqs. (18)-(19). For this calculation, donor and acceptor ions are randomly placed at their corresponding lattice sites within a numerically generated crystal sample of 10x10x10 unit cells. By using boundary conditions, our crystal sample is extended to fill the whole space. To do the placement of dopants, we consider the corresponding lattice sites coordinates reported in crystallographic tables for the unit lattice cell. The number of dopants within the crystal sample corresponds to the reported dopant concentrations. Then, with such crystal sample the transfer rates among all dopants are calculated according to the assumed interaction that drives the transfer of energy, Eq. (3). The microscopic individual emissions are calculated and averaged over a large number of numerically generated crystal samples according to Eqs. (21) and (22). We have developed a numerical code in MatLab®, called DETME, to grow numerically crystalline samples and to compute Eqs. (21)-(22). To do numerical calculations one needs current parameters used by major models. " ' ' " That is, i) the lifetime of each specie of dopants, ii) the dopant concentrations, iii) the site lattice positions for each species and the lattice constant, and iv) the free parameters for each of the assumed interaction among dopants. As reported in Sec. 2 the free parameters for the
535
multipolar interactions are R™ with S=6,8,10,..., and R0 and L for the exchange interaction, see below Eq. (3) for details. 3.1.4 Direct Energy Transfer driven by a single or multiple interactions. Let us now compare the predictions for the donor emission §D(t) that is obtained from each of the models presented for the direct energy transfer. To do this and due to the restriction of the FD model to handle only a single interaction to drive the energy transfer, we consider the following conditions for the calculations, i) The lifetime for donors and acceptors has the same value of 260usec. ii) The donor and acceptor concentrations are equal to 1.0 and 15.0 at. %, respectively, iii) A garnet host with dopants placed at the 24c sites, iv) The free parameter for each model will be Ro6=8.5xlO"10m. Notice that the information for iii) can be obtained from crystallographic Tables,25 and it used for the GKS and for our calculations. The calculated donor emissions are reported in Fig. 2 in a semilog scale, where the vertical axis is the log of the intensity in arbitrary units, and the horizontal axis is time in seconds. The dotted, the dashed and the solid curves correspond, respectively, to the FD, the GKS and our model. On the right of these curves, there is an enlargement for short times. As expected, between the FD and the other two models there are significant differences.
Time [usee]
Fig. 2.
Comparison among the donor emissions 0D(t) predicted by the FD model (dotted), the GSK (dashed) model, and our DETME solution (solid). All predictions consider a direct energy transfer driven by a dipole-dipole interaction. The upper solid curve corresponds to the donor emission with no energy transfer.
536
If one observes Fig. 2 for the plot at long times it could be said that the main differences between the dotted and the other two curves are at very long times. Indeed, the dotted curve is above the dashed and solid curves in this region. However, the enlargement at short times clearly shows that the FD calculation predicts a faster decay at short times. This is so because the FD does not consider the discreteness of the lattice. Let us now discuss the differences between the GKS and our calculation. The differences shown in both plots in Fig. 2 between the GKS model and our model are minor. However, it should be pointed out that the calculated curve for both models depend on the number of shells and on the number of generated crystal samples for the GKS and our modeling, respectively. From Eq. (14), one sees that the sum is over all lattice sites of a given lattice structure. For a realistic calculation, one cuts the sum up to some arbitrary shell. The number of shells that were used to calculate the dashed curve in Fig. 2 was 120. It can be seen in this figure that it approaches very close to our calculation, solid curve. This solid curve was obtained from 200 generated crystal samples as described in the previous section. Let us now consider a more complicated interaction among donor-acceptor pairs that drives the direct energy transfer. The example we like to present is for the case of the Cr(lat.%), Nd(lat.%), Er(15at.%): YAG for the direct energy transfer from Nd-to-Er ions. For this system the donor corresponds to the neodymium ions and its emission was measured by us and reported somewhere else.26 The experimental data is shown in Fig. 3 in a semilog scale, and corresponds to the fluorescence decay of the 4¥^n. Nd3+ state. The considered luminescent system is similar to the reported by other authors.20' 23"24 However, to approach the experimental data, we used a sum of interactions to drive the transfer of energy, as shown in Eq. (3). The interactions and the used free parameters for our calculations were dipole-dipole, dipole-quadrupole, quadrupole-quadrupole and exchange with R™=8.65A, R™=6.2A, R™=5.$A, and R0=5.6A, LDA=0.5A, respectively. The donor lifetime was 260|isec. The solid curve in Fig. 3 corresponds to the predicted donor emission by our modeling. As shown in this figure, our solid curve mimics the experimental data and assuming a sum of a dipole-dipole, dipole-quadrupole, quadrupole-quadrupole, and an exchange interactions did the calculation. All these interactions are taking place among any single donor-acceptor pair within the crystal sample, they drive the transfer of energy, and each of them vanishes naturally according to the separation among the considered pair. That is, it is not introduced any cut off distance to do the calculation of the donor emission as was introduced by Stalder and Bass.23 In Fig. 3, we also show the donor emission for the cases when each of the interactions were assumed to drive the energy transfer with the same values for the free parameters used for the solid curve. It can be seen in Fig. 3 that the dipole-dipole interaction is the closest curve to the experimental data at long times. The other two multipolar interactions are really off the experimental data; however, the exchange interaction approaches very well the data at very short times.
537
*-V q-q interaction
exchange interaction d-q interaction
13
1E-1
- d-d interaction 4-fold mixture of interactions
c
"**» *»,£ £Experimental data
1E-2
10
13
15
18
20
23
25
28
30
Time [u.sec]
Fig. 3.
Normalized emission ofthe 4 F 3/2 state of Nd 3+ in Cr(l%),Nd(l%),Er(15%):YAG. The solid curve corresponds to the calculated donor emission by assuming a four-fold mixture of interactions, see text. The dashed, dot-dashed, dot-dot-dashed, and dotted curves correspond to the emissions by assuming only one of the interactions in the solid curve, that is, to d-d, d-q, q-q, or exchange, respectively.
One may assume, as Stalder and Bass did that by assuming two interactions (the dipole-dipole and exchange) could be enough to approach the experimental data. This, however, is not true since our modeling shows that the other two multipolar interactions smooth the calculated emission about 1.0 to 8.0 usee. That is, among the time regime where the exchange vanishes naturally and the dipole-dipole interaction takes off. Indeed, we show that it is the sum of the four interactions that mimics the experimental data in Fig. 3, and that each of the assumed interactions by itself does not fit the experimental data in the whole time window of the transient.
Notice that by using more than one interaction to drive the energy transfer in the above luminescent system, or for any other system, one makes use of the corresponding free parameters defined in Eq. (3). No new free parameters are used as other models do when fitting complicated experimental data.13 Indeed, the non-radiative energy transfer
538 processes might be very complicated, but one should build up a model based on physical grounds and try to solve the more general problem. In summary, we have shown in this section that our modeling solves complicated donor transients assuming that more than one interaction drives the transfer of energy. The FD model does fail in handling more than one interaction whereas the GKS model does not. However, the GKS is restricted to calculate donor emission and omits the calculation for the acceptor emission. Furthermore, the GKS was developed to handle only the direct energy transfer, i.e., no further energy transfer process can be considered. In the following section, the migration of the energy among donors is added to the direct energy transfer, and thus, new terms are included of the GETME.
3.2. Direct energy transfer with migration of the energy among donors So far, we have presented some models for the non-radiative direct energy transfer and shown that they are focused to give analytical expressions for the donor emission. The FD model makes use of continuos distribution of acceptors around donors and thus, it has been applied to amorphous and crystalline host materials. Furthermore, this model cannot handle a sum of interactions and for some complex donor emissions, it fails to predict the whole transient. On the other hand, the GKS model was developed to consider the discreteness of the crystalline structure of the host material and can handle a sum of interactions to drive the transfer of energy. None of the models gives an analytical expression for the acceptor emission. Contrary to these models and to others in the literature, our model is based in the solution of the restrictive Eqs. (16) and (17) which are obtained from the GETME, and from this solution the donor and acceptor emission are calculated. Now in this section, we present the major models for the case when the donorto-donor energy transfer cannot be neglected.
3.2.1 Current models for migration of energy Two major models can be found in the literature. One assumes that migration occurs as a diffusion process and the other model considers that migration is a random walk problem. The former process was developed by Yokota and Tanimoto (YT) and the second by Burshtein. Let us first to consider the YT model. The differential equation that solves this model is not for the probability P*t(t) in Eq. (4), but for the probability p(R,t) that satisfies
1
4 P ( / ? , O = DV 2 PCR,0- — + L W $ ( R - R A J ) too
J=l
P(R,0
(23)
539 This differential equation is an extended expression of Eq. (4) that was discussed for the FD model. That is, Eq. (4) is modified to consider the spatial diffusion of p(R,t) through the first term in Eq. (23), whereas the second term gives the temporal evolution of p(R,t). The coefficient D is the diffusion constant that characterizes the migration of the energy among donors as a diffusion process. The solution to Eq. (23) for the excitation density p(R,t) is related with the average normalized emission <j>df) trough,
^D(t)=-^lp(R,t)d'R
(24)
N,
No general solution to Eq. (23) has been reported, and YT obtained an expression for <j>dij) for the specific case when the transfer of energy is driven by a dipole-dipole interaction. The reported solution, obtained by using the method of Pade approximates, is
(|)D(r) = exp
'
t ^
exp
3C,
xl/2
•ni
**<&
l + 10.87y + 15.5y2 ^ l + 8.743y
(25)
where
y=D
Mil
(26)
For the case that S=6 in Eq. (25) and when the diffusion process is neglected, y=0, the FD expression can be recovered. This result is not surprising since the YT model is an extension of the FD model. The diffusion constant D in this model depends on both the donor-to-donor critical transfer distance, R^D, and the donor concentration, Co- For the case of low donor concentration and for a multipolar process, Huber6"7 reported the D value as
D=\{n-MUcD
(/e)
(27)
For the particular case of a dipole-dipole interaction n=6 and one obtains
B.ifaD-tm
(28)
540 Let us now make some remarks about the predictions of this model at short and long times of the calculated donor emission, Eq. (25). At early times, t« td =
D
> °6 J^=
J*w
the diffusion is not important 0 « 1 ) , and Eq. (25) approaches Eq. (8) with S=6. That is, at early times the FD model predicts the decay of the donor and the transient is nonexponential. This is so because the time is not sufficient for the excitation to diffuse among the donors before being transferred to the acceptors. At sufficiently long times, <(>D(t) becomes asymptotically exponential and the corresponding decay rate is given by27 1
1
T
T DO
2
+Wdif = J-+ l
J££^^y(Ro,r l
D0
(29)
D0
Burshtein considered the migration of excitation among donors as a random walk process. Under this random walk approach the decay function for donors,
D(0exp
(_t\
-f
1 ',
+ - JW-f'JMOexp
\
dt'
(30)
For this expression 0[>(t) is the decay function of the donors in absence of donor-to-donor transfer, i.e,. Eq. (4). zo is the average time that the excitation resides on a donor before hopping to another donor ion. The ro value is given in terms of the concentration of donors, Co, and of the critical transfer distance among donors, R™. For the case of a dipole-dipole interaction, 1/ ro is given by27
to
\ 3J
xm
The solution to Eq. (30) can be obtained by numerical methods. Thus, it is found that at early times 0(0 is non-exponential. That is, for times t«To,D{f), i.e., to the FD model. Again, this is not a surprising result since the model is an extension of the FD model. Contrary to this behavior for times long enough
- = — + »W = — + ^ % o ? ) ^ C ) t
Ton
"^nn
"^ no
(32)
541
Therefore, from above discussion, it should be clear that both models do different approximations to solve the problem for the migration of the energy among donor ions. However, both models give similar predictions for the donor emission at early and long times. At early times, up to a characteristic time (tj or To), the donor emission is nonexponential and it is predicted through the FD model. For times long enough, the donor emission becomes exponential with a characteristic decay rate W&J or Whop- Furthermore, the expressions for the decay rates, Eqs. (29) and (32), both linearly depend on the donor and acceptor concentrations, Co and CA respectively, but the dependency on the critical transfer distance for donor-to-donor and for donor-to-acceptor energy transfer is different. That is, having identical free parameters R^6D and R^, each predicted exponential decay will be different one from the other. As a rule, it is claimed that the YT model is valid when the donor-to-donor transfer mechanism is much weaker that the donor-to-acceptor transfer. Contrary to this, the hopping model is valid when the donor-to-donor energy transfer is greater than the donor-to-acceptor. Thus, for the case that each process is driven by a dipole-dipole interaction one should follow the rule given by the following criteria,6"7'10 if R&D « Rof, use the diffusion model. (33) if R™ > R°A, use the hopping model. According to these two models, four cases can be distinguished when the donor concentration is changed from low to higher values. At low donor concentrations, the donor-donor separation is large enough and one can assume that migration is negligible. Therefore, the decay of the donor luminescence is predicted by the FD model and fait) is non-exponential. As the separation becomes smaller by increasing the donor concentration, the decay is limited by the diffusion rate and is described by an exponential given by the YT model. At higher concentrations the migration among donors becomes faster, and the decay is limited by the migration rate with the decay described by an exponential decay accordingly to the Burshtein model. At much higher concentrations, the transfer of energy is faster among donors {super-migration regime) than the donor-toacceptor.7 In this case the excitation migrates between so many donor ions that it senses the average environment of acceptor ions. The average donor-to-acceptor transfer rate is now ^QfFoo where the summation is over all the sites in the crystalline sample, and the predicted decay of donors is exponential with a rate given by
-
r
~
+W
^
t e
~
+c
^W^
(34)
542
3.2.2 Current models and the transfer function The above modeling, with no back transfer and no migration among acceptors is summarized in the literature into three temporal stages for the transfer function of the whole donor emission as shown in Fig. 4.6"7' 8 This figure is in a log-log scale where the vertical and the horizontal axis correspond to the transfer function I7(t) for donors in arbitrary units and time in seconds, respectively. The transfer function is defined through the observed donor luminescent intensity as28
•fl(0=
(35)
The energy transfer function gives the channel of de-excitation for donor ions due to nonradiative losses at any other time after the pumping excitation. The temporal stages are divided by two transition stages at t\ and h as shown in Fig. 4.
1E3
~
1E2 super migration
3
1E1
f
,M
transition stages
tC XW 1E-2 1E-8
V
A
1E-7
j
1E-6
DO
1E-5
1E-4
1E-3
1E-2
Time [sec] Fig. 4.
Energy transfer function lift) in a log-log scale. The temporal window is divided in three temporal stages by tj and t2. In such scale lift) is represented always by straight lines, the slope and y-coordinate to the origin depend on the stage and the strength of the interactions that drive the energy transfer processes.
For the first temporal stage, t < t\, it is considered that because of the random distribution of the acceptors some donors will find acceptors in very close proximity. Therefore, these donors will decay rapidly by energy transfer causing an initial fast decay
543 in the donor emission that corresponds to the lowest straight line with a slope with unity see Fig. 4. At this time stage, the FD model does not give the transfer value, function since it does not consider the discreteness of the host material. Instead, the transfer function is given by,
W) = tcAT.w£
(36)
where CA is the acceptor concentration and the sum is up to some convenient nearest acceptors. That is, there is no rule to indicate the cut off distance for the calculation of the sum. The second temporal stage, t/< t < (2, is defined by the FD model. In this model the transfer function is defined by
n(0 =£-r 1-1
(37)
S
where Cos = (4^-/3)^?^)
is the critical concentration defined through the critical
transfer distance R^ , with S=6, 8, 10, and T[x] is the gamma function. Then, in the loglog scale II(t) is represented by a straight line with a slope of (3/S), which becomes 0.5 for the particular case of a dipole-dipole interaction, as shown in Fig. 4. The third stage, ^ < t, is defined by either the diffusion, td< t, or the hopping, TO< t, model as discussed above. The transfer function for a dipole-dipole interaction among dopants, is defined by
1 + 10.87^ + 15.5y2 1 + 8.7433/
if
WA> > WLn,
(38)
n(o = c
+ Whopt
if
w* < wLn
544
For short and intermediate times, t < t2, both expressions in Eq. (38) behave like the FD approach. However, at long times, ?2 < t, both models predict an exponential decay with a rate given by Wdij or Whop, Eqs. (29) or (32), respectively. Therefore, J7(t) is represented by a straight line of unity slope as shown in Fig. 5. For the extreme case W£j « WDim a single exponential decay curve for the donor emission is predicted, Eq. (34). This case corresponds to a single straight line in our loglog scale in Fig. 5 with a slope of unity value that corresponds to the super-migration regime. That is, the transfer function is linear with time and at very short times, t
4. Exact solution to the GETME Instead of looking for the solution for the direct energy transfer and for the migration of the energy among donors, we report in this section the exact solution to the GETME reported as Eqs. (1) and (2) above. To solve such equations, we rewrite them in a matrix form as d
^-^dt
where,
= WkPk{t),
(39)
545
Pm«) Pk(t) =
PLD(t) PkAD
''
and
Wk =
JDND DN D
W,D\
w,
PLC)
n
TirDNc A\
Al
ANA
WDNa
ANA
M
ur
W,
w
wh T
w
(40)
w:
Pk(t) is a vector that contains the individual probabilities, Px,{t), that the ith ion of specie X (D or A) still is in an excited state at time t. Notice that each probability is defined for any dopant within the kth generated crystal sample as it was described for the case of the direct energy transfer. W* is the energy transfer matrix of the kth crystalline sample. This matrix contains all the information about the driving interactions of the different energy transfer processes, as well as the distribution of the dopant ions and their positions within the kth crystalline sample. The diagonal elements of W are defined by,
1
NA
ND
i = 1,2,.... TVb
(41.1)
j = l,2
(41.2)
and
7- + 2K+k.i.ktj 1<
JVA
Eq. (39) represents a typical differential rate equation, and it has a well-known exact solution: k
(t) = ljzxp((Vl,ipk
(42)
where Pk' = Pk (t = 0) are the initial conditions. That is, at time t=0 the crystal sample is shined by the pumping source and the donors are excited, however, for some cases it can be considered that acceptors also can be in an excited state. This case could be considered for the energy transfer up-conversion process. This vector for the initial conditions is not found in any other modeling. Thus, to compute Eq. (42), we have to evaluate the exponential function, and we do this by diagonalizing W. In this way one finds that the exact solution for each individual P^, (t) [-P<,(0 ] in the kth sample is given by,
546 N„ fND+N,
^,(0 =
:
>*,0 Dm
(43)
where [/J and £/,*, correspond to the [ij]-elements of the similarity matrix Uk and its inverse {jk y , respectively, that diagonalize W = (z*} 1 ffD""«k j / * . The W/f*'* are the diagonal elements of the diagonal matrix WDiag-k. Expression (43) and a similar expression for the acceptors give the exact solution of the GETME system. These two solutions give the microscopic individual emissions for each D and A ion within the k\h crystalline sample, and the mean excitation probability is given by
2>A« Pn(0-
ND
NA
(=1
;=i
(44)
If this probability is averaged over a large number of generated crystal samples, the resulting average converges to the macroscopic probability for the macroscopic donor ensemble that remains excited at time /. Thus, the normalized macroscopic donor emission >D(t) is given by
2%(o (45)
Dit) = [ 0+
k,0
2>«
4/ J~'
I
overk
with a similar expression for acceptor ions. In summary, the solution to the GETME, Eqs. (1) and (2), is given by Eq. (43) and a similar expression for acceptor ions. In the whole calculations, one has defined the nonradiative energy transfer processes involved in Eqs. (1) and (2). The interactions that drive the transfer of energy among dopants are defined by Eq. (3). Notice that in Eq. (3) a single term, or any sum of terms or any sum with new types of interaction can be assumed to drive the transfer of energy. Our general solution with Eq. (3) makes use of identical
547
free parameters as those used by the main models, i.e., FD, YT and Burshtein models. On the other hand, to carry out the calculation of the macroscopic emissions one generates a large number of crystal samples. Each crystal sample is generated from the unit cell and from the knowledge of the sites where dopants are placed. The placement of dopants is done with a random number generator. In this way each donor has its own distribution of acceptors around it. The unit cell is replicated to a certain number to build up any crystal sample. By using boundary conditions, we make sure that the crystal extends over the whole space. The numerical code calculates the dopant emissions and it was implemented to be used in a friendly environment as a toolbox within the MatLab® software. Therefore, the program asks to the user for the free parameters, dopant concentrations and dopants' lifetimes. The numerical code can be extended to be used to any kind of crystal structure.
4.1 GETME and the models for the migration of energy among donors In the previous section we have reported the exact solution to the GETME, and described how the donor and acceptor emissions can be calculated for the non-radiative energy transfer processes involved in such equations. Now, if we like to compare our calculation with the models described above for the migration among donors, one should give zero values to the other free parameters defined for other processes. For the comparison between our modeling with the models of YT and Burshtein, we refer to the calculations reported by Henderson and Imbusch in Fig. 10.5 in Ref 10. In that figure the transfer function is reported for those models plus the super-migration case and the FD model, this of course is just for reference. It should be clear that the calculation for the transfer function from our modeling does not represent any difficulty and it is calculated as defined above, Eq. (35). Fig. 5 is in a log-log scale similar to Fig. 4, and it shows six temporal orders in the x-axis and five orders for lift) in the y-axis. In Table I, we report the used free parameters for the calculation of the curves shown in Fig. 5. The non-solid curves correspond to the calculation from the FD model (stars), YT model (full-circles), Burshtein model (crosses), and super-migration (triangles). The solid curves correspond to our calculations using the same parameters used for the other models and reported in Table I. Further, for our model we assumed the YAG crystal structure and dopants were placed, randomly, at the 24c lattice sites. From Fig. 4, it should be clear that the transfer function for the FD model is a straight line with a slope value of 0.5, see the star curve in Fig. 5. It was also said that at very short times the IJ(t) is described by Eq. (36) since the FD model does not considers the discreteness of the lattice. Therefore, the first temporal stage ti is defined in Fig. 5 as the intersection of the triangles with the stars. The full-circle curve in Fig. 5 is for the case of the YT model, and for very short times it overlaps, as discussed above, the FD curve.
548 Table I Free parameters values used for the transfer functions reported in Fig. 5. Identical free parameter values were used for our modeling, and reported as solid curves in Fig. 5.
Model Direct donor to acceptor energy transfer, R^ Donor to donor energy transfer, R^D
F-D
Y-T
Burshtein
Supermigration
10.0 A
10.0 A
10.0 A
10.0 A
8.32 A
12.97 A
38.7 A
The second temporal stage for the YT model is defined at the point where it changes the slope value of 0.5. That is, it is defined at the time where the full-circle curve separates from the stars curve in Fig. 5. At very long times the YT predicts an exponential decay and it is represented in Fig. 5 as the straight line with a slope value of unity.
1E-2 1E-8
1E-7
1E-6
1E-5
1E-4
1 E-3
1 E-2
Time [sec] Fig. 5.
Energy transfer function lift) for the current models: the Forster-Dexter model (direct donor to acceptor energy transfer) in stars, the Yokota-Tanimoto model (diffusion among donors) in circles, the Burshtein model (migration among donors) in crosses, and the supermigration approach in triangles. Solid curves correspond to our simulations with the same parameters used for the corresponding cases of the current models, i.e., direct energy transfer, diffusion, migration, and supermigration.
549 The second stage time for the Burshtein model is defined similar to the YT model, but it occurs at different time. That is, t2 is defined at the time where the cross curve separates from the FD model. This occurs at earlier times than for the YT model as shown in Fig. 5. An exponential decay is predicted for the Burshtein model at long times, and it is represented by a straight line with a slope value of unity in Fig. 5. For the supermigration case the donor emission decay is exponential and corresponds to a straight line with slope value of unity as shown in that figure. The calculations with our model are represented as solid lines in Fig. 5, and at intermediate and long times they mimic very well the corresponding transfer function for the other two models. At short times, our solid curves do not follow the lift) curves predicted by the FD, YT and Burshtein models. This is so because those models do not consider the discreteness of the crystal lattice. Instead, our simulations are described by a straight line with a slope value of unity. However, as time increases the solid curves evolve smoothly to a straight line with a slope value of 0.5; then, they remain with that value for a while; and depending on how strong the migration is, the slope value evolves smoothly to a straight line with a slope value of unity. Thus, if migration is weak, our simulation is followed by the predicted curve of the YT model. If migration is strong, then the Burshtein model follows our simulations. For the super-migration case, our calculation does not show differences respect to the triangle curve. Thus, Fig. 5 and the above discussion clearly show that our model can predict the same predictions given by other models. The temporal stages considered by those models to get the understanding of the transfer processes are not needed in our modeling. Further, we could assume that more than one interaction among dopants drives the transfer of energy, and none of the above models can do this assumption. Besides this, more transfer processes can be included to calculate the donor and acceptor emissions or the transfer function for the donor. The other models are not able to do this. As claimed, our model also calculates the acceptor emission. In Fig. 6, we present our GETME predictions for the acceptor emission when the migration among donors is increasingly dominant. The curves presented in Fig. 6 correspond to the same cases pictured in Fig. 5 and the corresponding parameters are listed in Table I. The curve in stars corresponds to the acceptor emission when only the direct donor to acceptor energy transfer is considered. The curve in circles corresponds to the case when migration among donors is considered, but the donor to donor transfer rate is lower that the donor to acceptor transfer rate, i.e. the YT model regime. As the migration critical parameter increases, RQ6 > R^ , we obtain the acceptor decay under the hopping regime, crosses. And for an even faster migration, we observe, in triangles, the supermigration regime. From these curves, it is clear that as the migration among donors increases the acceptor fluorescence presents a higher maximum at shorter times. That is, for a stronger migration the donor subsystem transfers more rapidly the excitation to the acceptor subsystem.
550
ity [a.u.'
1-, 0.9 0.8 0.7-
11
i
£ .2
< -©^
w "—*
•|5
* * . • • • • ^\»
0.6-
GETME FOR * Forster-Dexter Yokota-Tanimoto Burshtein " J ^ ^ A Supermigration * ^ ^ >^^*.
0.5
*> * 0.4
a)
ptor
**^*.
0.3-
$ * _ x F
§ 3
s
IK \ ^
processes.
*i n9 u.z
* 0
50
100
Time Fig. 6.
^^^^^
/ life time decay of acceptors in absence of energy transfer
150
200
250
[usee]
Acceptor emissions predicted by the GETME model. The used parameters are in Table I, and correspond to the same cases presented in Fig. 5.
5. Concluding remarks We have compared the exact solution to the general non-radiative energy transfer master equations, GETME, with the major models for the direct energy transfer process and for the migration of the energy among donors. It was shown that our calculations for the donor emission unify and extend the calculations reported by those models, under identical conditions. Furthermore, our calculations show other features that are not considered by other models. For example, for any transfer process the acceptor emission can be calculated. Our modeling considers the discreteness of the host material and thus some differences are shown with other models that neglect it. When our calculations were compared with the GKS, which considers the discreteness of the host material, no major differences are shown, however, this model was developed only for the direct energy transfer process, and no acceptor emission can be calculated. For any calculation, we can consider that more than one process is taking place and that, more than on interaction among dopants drives the transfer of energy. Our model may include calculations for the back energy transfer and for migration of the energy among acceptors but the comparison was not included because of the lack of experimental data.
551 6. Acknowledgements We like to thank to the graduate students J. T. Vega-Duran and F. Mosino for their contributions to this manuscript. We are also indebted to the CONACyT, Mexico, for the partial support through the Grant 26434-E.
References 1. L. J. Dowell, Los Alamos National Laboratory report LA-11873-MS (1990). 2. T. Forster, Ann. Phys. (Leipzig) 2, 55 (1948). D. L. Dexter, J. Chem. Phys. 21, 836 (1953). 3. M. Yokota and O. Tanimoto, J. Phys. Soc. Japan 22, 779 (1967). 4. A. I. Burshtein, Sov. Phys. JETP 31, 882 (1972). 5. Powell in Energy Transfer Processes in Condensed Matter, edited by B. Di Bartolo (Plenum Press, New York, 1984) p. 103. 6. D. L. Huber, in Laser Spectroscopy of Solids, edited by W. M. Yen and P. W. Selzer, (Springer-Verlag, Berlin, 1986) p. 83. 7. D. L. Huber, Phys. Rev. B20, 2307 (1979). 8. Yu. K. Voron'ko, et al.; Sov. Phys. JETP. 44, 251 (1976). 9. B. Di Bartolo, in Energy Transfer Processes in Condensed Matter, edited by B. Di Bartolo (Plenum Press, New York, 1984) p. 103. 10. B. Henderson, and G. F. Imbusch, Optical Spectroscopy of Inorganic Solids, (Oxford University Press, New York, 1989) p. 445. 11. O. Barbosa-Garcia, C.W. Struck, J. Chem. Phys. 100(6), 4554 (1994). 12. M. Inokuti, F. Hirayama, J, Chem. Phys. 43 (1978). 13. S.R. Rotman; Appl. Phys. B 49, 59-64 (1989). 14. S.I. Golubov, Y. V. Konobeev, Sov. Phys. Solid State 13, 2679 (1972). 15. V.P. Sakun, Sov. Phys. Solid State 14, 1906 (1973). 16. A. Blumen, II Nuovo Cimento 63 (1) B, 50 (1981). 17. G. Zumofen, J. Klafter, A. Blumen, J. Chem. Phys. 79(10), 5131 (1983). 18. L.A. Diaz-Torres D. Sumida, O. Barbosa-Garcia, M.A. Meneses-Nava; SPIE Proceeding Vol. 3610 69-77(1999). 19. O. Barbosa-Gracia, PhD Thesis, Physics Department, Boston College (1993). 20. S.R. Rotman, Appl. Phys. Lett. 54, 5053 (1989). 21. S.R. Rotman, Opt. Lett., 15, 230 (1990). 22. L.A. Diaz-Torres, O. Barbosa-Garcia, C.W. Struck, R.A. McFarlane, J. Lumin. 78, 69 (1998). 23. M. Stalder, M. Bass, J. Opt. Soc. Am. B 8, 177 (1991). 24. W.Q. Shi, R. Kurtz, J.Machan, M. Bass, M. Birnbaum, Apply. Phys. Lett. 51 (16), 1218(1987). 25. R.W.G. Wyckoff, Crystal Structures, vol. 3, 2 nd ed. Wiley Interscience, New York, 1965.
552
26. O. Barbosa-Garcia, R.A. McFarlane, M. Birnbaum, L.A. Diaz-Torres, J. Opt. Soc. Am. B 14(10), 2731 (1997). 27. R.K. Watts, in Optical Properties oflOns in Solids, Ed. B. Di Bartolo, Plenum Press, New York (1975). 28. A. A. Kaminskii, Laser Crystals, (Springer-Verlag, Berlin, 1981) pp. 361.
SHORT SEMINARS
OPTICAL AND STRUCTURAL PROPERTIES OF ALUMINA AND RARE EARTH DOPED ALUMINA PLANAR WAVEGUIDES ELABORATED BY SOLGEL PROCESS AND BY PULSED LASER DEPOSITION ANNE PILLONNET Laboratoire de Physico-Chimie des Materiaux Luminescents Universite Claude Bernard Lyon 1 69622 Villeurbanne Cedex, FRANCE Pure and doped optical waveguides are of interest for integrated optics applications. Furthermore, fundamental studies about structural and optical properties of materials in thin films form take advantage of both the guiding of the light and the fluorescence properties of the dopants. Sol-gel process and pulsed laser deposition (PLD) were used to elaborate alumina and europium doped alumina films. All the films are deposited on silica substrate and present waveguiding properties. The refractive index, measured by m-lines spectroscopy, are higher for the PLD films than for sol-gel films, indicating a more dense structure. From grazing incidence X-ray diffraction, Waveguide Raman Spectroscopy, and transmission electron microscopy, we deduce the effect of elaborated parameters on the structural properties of the films. Furthermore, we use the fluorescence characteristics of Cr3+ doping ions to analyze the films' microstructure, and to compare these with those of the bulk material.
OPTICAL AMPLIFICATION IN ERBIUM-DOPED FIBER AMPLIFIERS LAURENT BIGOT Laboratoire dePhysico- Chimie des Materiaux Luminescents Universite Claude Bernard Lyon 1 43. Blvd 11 Novembre 1918 69622 Villeurbanne Cedex, FRANCE The beginning of the 1990's saw the advent of the rare-earth doped fiber amplifier in optical communication systems. Erbium-doped fiber amplifiers, which operate in the
553
554 important 1.55um third telecommunication window, appeared as the ideal medium: low loss, high gain and possibility of multiplexing, etc. Nowadays, the increasing needs (Internet, cellular network, etc.) impose a new step in performances and more specifically in the gain bandwidth, to permit the amplification of more than one hundred signals simultaneously in only one fiber. Using data from literature or supplied by Alcatel, we study some relevant parameters that have to be taken into account in view of the expected improvement, and what are the relevant techniques that permit their study.
CHARGE TRANSFER LUMINESCENCE OF YB 3+ L. VAN PIETERSON Debye Institute, Department of Condensed Matter Utrecht University, P. O. Box 80 000 3508 TA Utrecht, THE NETHERLANDS In this century, luminescence of rare earth ions has been well studied. Since the application of luminescence from rare earth ions in fluorescent tubes, colour television and X-ray phosphors, numerous papers have appeared on 4fa-4fn and 4fn"15d-4fn emissions. One kind of rare earth luminescence is still relatively unknown: charge transfer luminescence. This transition is the reverse of the well-known charge transfer absorption. Until now only three papers have reported on luminescence from the charge transfer state (CTS) of a rare earth ion [2, 3, 4] although the CTS is important for applications. For example in the red phosphor used in fluorescent tubes (Y203:Eu3+), UV radiation from the Hg-discharge is efficiently absorbed in the CTS of the Eu3+-ion. After nonradiative decay to the lower 5Dj states, luminescence occurs. In Eu3+, luminescence from the CTS can not be observed, because there will always be fast relaxation to the lower 4f levels. Yb 3+ (4f") is an ion for which CT luminescence can be expected. In this ion the only excited 4f state, 2Fs/2, is located 10 000 cm"1 above the ground state, 2Via. There are no levels of energy comparable to the CTS, which makes it possible for the ion to relax radiatively from this state. The influence of host lattice and temperature on the emission from the CTS is expected to be large. We present a systematic study of charge transfer luminescence of the Yb3+-ion. The emission in 25 different host lattices has been studied. As an example we would like to show the results obtained on Yb 3+ CT luminescence in the orthophosphates REPO4 (RE=Sc, Lu, Y, La). In these lattices, the size of the site available to the rare earth ion is varied, increasing from Sc to La. Temperature dependent measurements are performed to gain insight into the influence of the host lattice on the CT luminescence efficiency. The results are compared with the results obtained for Eu3+ luminescence under CT excitation in the same host lattices.
555
It has been found that Yb shows luminescence from the CTS in the orthophosphates of cations with a relatively small radius (Sc, Lu, Y). In LaPC>4, no evidence for CT luminescence could be obtained. In this lattice the large site available to Yb3+ probably facilitates nonradiative relaxation. The CT emission shifts to longer wavelengths with increasing cationic site, just as the CT absorption. The decay time is typically 150 ns ± 25 ns, and the quenching temperature is around 250 K. The Eu + luminescence under CT excitation quenches at much higher temperature, but the quenching order is observed to be the same as for Yb +.
PIGMENTATION OF CRT PHOSPHORS CLAUS CLEMENS FELDMANN Philips GmbH - Forschungslaboratorien P. O. Box 500145 D-52085 Aachen, GERMANY To improve the contrast of a TV screen, not only the phosphor brightness, but also the daylight reflectivity of a screen has to be considered. The relation of contrast on the one hand and luminance, as well as reflectivity on the other hand, can be quantified by the Luminance-Contrast-Performance (LCP): Luminance VDiffiise reflectance
In order to reduce the reflectivity of the white phosphor powders (e.g. ZnS:Ag,Al, Y202S:Eu), each individual phosphor particle is covered with a pigment. The pigment colour should correspond to the emission of the respective type of phosphor, for instance, C0AI2O4 for blue emitting ZnS:Ag,Al or Fe203 for red emitting Y2C>2S:Eu. The pigment particles (e.g. C0AI2O4, Fe203) have to be quite small in size (80 to 120 nm) to guarantee a sufficient adhesion on the phosphor particles. In addition, other ingredients like inorganic oxides or organic polymers can be required to achieve a good adhesion and to establish a homogeneous thin layer of pigment particles on the phosphor surface.
556
THE INFLUENCE OF THE PENETRATION DEPTH ON THE COLOUR POINT OF (Y,Gd)B03:Eu THOMAS JUSTEL Philips GmbH- Forschungslaboratorien Weisshansstrasse 2 D-52066 Aachen, GERMANY Almost all PDP manufacturers use (Y,Gd)B03:Eu as red primary because it has a high efficiency under VUV excitation. A drawback of this phosphor is that the orange colour point for appropriate video operation is too large. This is caused by the relatively intense line at 594 nm, which corresponds to the magnetic dipole transition 5Do-»7Fi while the emission lines of the electric dipole transition 5Do—>7F2 at 612 and 627 nm are weaker. The colour point of (Y,Gd)B03:Eu is x = 0.638 and y = 0.360 if the phosphor is excited into the charge-transfer band of Eu3+ which is located at 220 nm. Under band gap excitation, i.e. with energies above 7.0 eV in a PDP, we observe a somewhat more reddish colour point, viz. x = 0.646 and y = 0.349. A similar finding was made by the accurate analysis of the emission spectra obtained under cathode-ray excitation with electron energies between 2 and 10 kV. Decreasing the acceleration voltage results in a colour point shift to the red (fig. 2). The parameters determining the penetration depth R of electrons in matter are the density (in g/cm3) and the electron energy, e.g. in terms of acceleration voltage (U in kV). R = 0.046U5/3/p [nm] [1] According to equation [1], an acceleration voltage of 10 kV results in a penetration depth of more than 400 nm which can be described as bulk excitation. In contrast, the penetration depth for 2 kV electrons is only around 30 nm. From the comparison of the observed colour point shift for photo-excitation (chargetransfer excitation vs. band gap excitation) on one hand, and for cathode-ray excitation (2 kV vs. 10 kV) on the other hand, it is derived that the penetration depth of VUV radiation is also in the range of some ten nanometers. The observed colour point shift of (Y,Gd)B03:Eu from orange to red in dependence on the penetration depth can be explained as follows: In the (Y,Gd)B03:Eu lattice, Eu + ions occupy two sites in the lattice with C3 symmetry whereby the deviation from the ideal local S& symmetry is rather small. Due to the small deviation a relatively high 5 Do—»7Fi/5D0—>7F2 intensity ratio is observed since the electric dipole transition 5Do—Jfi is forbidden for inversion symmetry. The deviation from inversion symmetry becomes larger if the crystallinity is reduced which is, in general, the case at the grain boundaries of phosphor particles. Consequently, a lower penetration depth, resulting in the excitation
557 of Eu3+ ions at the very surface, gives an emission spectrum with a lower Do-> 7 Fi/ 5 Do-» 7 F 2 intensity ratio and thus a more reddish colour point.
5
SECOND-ORDER NONLINEAR OPTICAL BEHAVIOUR OF DENDRIMERS FUNTIONALIZED WITH NITRO ANILINES ABRAHAM W. (BRAM) GERBRANDD Leiden Institute of Chemistry, Gorlaeus Laboratories Leiden University - P. O. Box 9502 2300 RA Leiden, THE NETHERLANDS Organic materials are suitable candidates for nonlinear optics (NLO), because of their synthetic diversity, ease of fabrication and low cost. Second-order NLO materials, used for sum-frequency generation and electro-optical switching, often consist of NLO chromophores which are oriented with respect to each other in a fashion to avoid centrosymmetry. In this study, wedge-shaped amino-acid based dendnmers are used as a backbone for NLO-molecules. Since they are synthezised via the convergent method, the periphery can be functionalized with different NLO chromophores having a controlled mutual orientation. The second-order NLO properties of several generations of different types of dendnmers are investigated using hyper-Rayleigh scattering in solution.
SUB-PICOSECOND PULSE BREAK-UP IN AN InGaAsP OPTICAL AMPLIFIER FRANCIS ROMSTAD Center for Communications, Optics and Materials Technical University of Denmark, Bldg. 349 DK-2800 Lyngby, DENMARK For high speed optical communication system with data speed higher than 200 Gb/s sub-picosecond pulse dynamics, coherent effects become important. We have, experimentally and theoretically, investigated the pulse distortion of a 150 fs pulse due to amplification in a 250 um long InGaAsP ridge waveguide, working at 1.52 mm, for different input pulse energies. Amplitude and phase in both frequency and time domain of the initial and amplified pulse were measured using a XFROG (Cross-Frequency-Resolved Optical Gating) technique based on sum-frequency cross-correlation. Measurement show a pulse broadening and eventual break-up for input pulse energies on the order of picoJoules. This break-up is present in the gain region (6-14 dB), while for absorption (-6 dB) and transparency, pulse narrowing by a factor of two is
558 evidenced. We observe that not only the amplitude is modulated, but also the linear chirp of the initial pulse is strongly modified. According to a numerical model, two-photon absorption and gain dispersion are responsible for the broadening and break-up. Kerr-nonlinearity at high intensities modulates the phase of the pulse and thereby the spectrum, which is evidenced in a spectral break-up. Due to gain dispersion, the spectral break-up results in a break-up in time domain.
QUANTUM DOT SPECTROSCOPY KRISTJAN LEOSSON Center for Communications, Optics and Materials Technical University of Denmark, Bldg. 349 DK-2800 Lyngby, DENMARK Semiconductor quantum dots ("solid-state atoms") are promising candidates for quantum computers and future electronic and optoelectronic devices. Quantum dots are zero-dimensional electronic systems and therefore have discrete energy levels, similar to atoms or molecules. The size distribution of quantum dots, however, results in a large inhomogeneous broadening of quantum dot spectra. Work on self-assembled InGaAs/GaAs quantum dots will be presented. Properties of atom-like single-dot states are investigated optically using high spatial and spectral resolution. Single-dot spectra can be used to probe coherence times of exciton states and relaxation processes, both of which are important for future applications.
DOPED SEMICONDUCTOR NANOPARTICLES - A NEW CLASS OF LUMINESCENCE MATERIALS? AGEETH BOL Debye Institute, Department of Condensed Matter, Utrecht University P. O. Box 80 000, 3508 TA Utrecht, THE NETHERLANDS In 1994 Bhargava et al. [1] reported that nanocrystalline ZnS:Mn2+ can yield both high luminescence efficiencies and a decay time shortening from ms to ns at the same time. These spectacular results suggested that doped semiconductor nanoparticles form a new class of luminescent materials with a wide range of applications in, for example, displays, sensors and lasers. However, the absence of a ms decay in nanocrystalline ZnS:Mn2+ was never verified. From lifetime measurements and time-resolved spectroscopy we concluded [2] that the
559 Mn + emission does not show a spectacular shortening of the decay time upon decreasing particle size as reported earlier. The luminescence of nanocrystalline ZnS:Mn2+ has a short decay time indeed, but has a ms range decay time as well. The short decay time (-100 ns) is ascribed to a defect-related emission of ZnS and is not originating from the decay of the " T i - 6 ^ transition on the Mn2+ impurity as suggested by Bhargava. The Mn 2+ emission of nanocrystalline ZnS:Mn2+ has a decay time of about 1.9 ms, like in bulk ZnS:Mn2+. Based on our observations we concluded that doped semiconductor nanoparticles do not form a new class of luminescent materials, since there is not a combination of a high luminescent efficiency and a short (ns) decay time. Still, efficiently luminescing doped nanocrystals can be important for other applications and the luminescence of doped semiconductor nanoparticles remains an interesting field of research.
DIRECT EXTRACTION OF POTENTIALS AND DYNAMICS FROM CONDENSED-PHASE PUMP-PROBE SPECTROSCOPY MATIAS BARGHEER AG Schwentner ExperimentalphysikFU- Berlin; Arnimallee 1414195 Berlin, GERMANY We consider I2 molecules isolated in a Kr-matrix as a model system for condensed phase dynamics. The h. molecule is weakly coupled to the surrounding Kr-lattice serving as a bath held at 15 K. From pump-probe spectra we directly derive the anharmonicity of the excited-state (B-state) potential, the energy dissipation and predissociation rate. The pump pulses are provided by a home-built optical parametric amplifier, tunable from 540 to 720 nm, with a typical pulse duration of 60 fs. The pulse characterization is accomplished by a polarization-gating FROG (frequency resolved optical gating) crosscorrelation. Preparing the wavepacket on the B-state leads to oscillations with a typical round-trip time of 350 fs, persisting for 17 periods. As the probe, we promote the wavepacket at its outer turning point to ionic states from which we observe laser induced fluorescence (LIF). From the dependence of the oscillation frequency v on the excitation energy E we read the anharmonicity 3v/dE. The energy relaxation rate 9E/3t can be derived from the increase of the oscillation frequency dv/3t with time 3E/3t =(3E/dv)(3v/3t). The predissociation rate can be measured by fitting the increasing LIF-signal for a probe window in the predissociated state. Our results can be verified by comparing them to molecular dynamics simulations. The knowledge of the anharmonicity is exploited further to predict the optimal focusing time toPt =Pv/(27tv3v/3E) of the wavepacket with a simple analytic expression
560 for excitation with chirped pulses (p is the quadratic spectral phase). We show evidence for focusing the wavepacket after 6 oscillations. Despite the collisions of the molecule with the bath atoms, the information on the chirp (spectral phase) of the preparing pulse survives longer than 17 oscillations.
SELECTIVE HEAT-PULSE STUDIES OF POLYMERIC ELECTROLYTES JOHN E. FURNEAUX Department of Physics and Astronomy University of Oklahoma Norman, Oklahoma 73019 - 2061, USA Recently we have started a project to study the dynamics of polymeric electrolyte materials. We will concentrate on the system of polyethylene oxide), CH 3 0(CH2CH 2 0) n CH3 (PEO) with dissolved LiCF 3 S0 3 (lithium triflate, LiTf). In particular we are pumping these materials with a selective heat pulse in the form of an intense infrared pulse tuned to the n(CH)-stretch region near 2900 cm"1 using a Nd:YAG based optical parametric oscillator (OPO) system with wavelength extension. We will then observe changes in absorption of the system in the entire infrared region to determine the energy transfer dynamics of this system using step-scan Fourier-transform time-resolved infrared spectroscopy with a 10 ns resolution. In particular we will be observing the triflate bands at 750 cm"1 to see changes in coordination between the triflate" ions and Li+ ions. We will also be concentrating on conformational changes observed in the PEO bands from 770-910 cm"1. These studies will be greatly facilitated by our extensive previous studies of the PEO-LiTf system including smaller PEO oligomers [Freeh 1994]. We are particularly interested in studying conformational changes in the polymeric system which are induced by the infrared pulse and the resulting changes in ionic conductivity of the Li+ ions.
NEAR-INFRARED TO VISIBLE UPCONVERSION IN Er J+ DOPED Cs3Lu2Cl9, Cs3Lu2Br9, AND Cs3Y2I9 EXCITED AT 1.54 MM STEFAN LUTHI Department of Chemistry, University of Queensland St. Lucia, Brisbane QLD 4072, AUSTRALIA A detailed study of upconversion processes in Cs3Er2X9 (X = CI, Br, I) crystals and in the diluted systems Cs 3 Lu 2 Cl 9 : 1% Er3+, Cs 3 Lu 2 Br 9 : 1% Er3+, and Cs3Y2I9 : 1% Er3+ is presented. Efficient two-, three-, and four-step upconversion excitation along the
561
sequence 4Ii5/2 —> 4Ii3/2 —> %/2 —> 4^3/2 —> 2H9/2, leading to luminescence throughout the visible and near-ultraviolet is demonstrated using a 1.54 mm excitation wavelength. This stepwise excitation is possible due to the low phonon energies and, consequently, the significantly longer lifetime of the 4Ig/2 intermediate state in these systems relative to oxides and fluorides. The absorption and upconversion luminescence intensities increase along the isostructural series X = CI, Br, I as a result of the decreasing energy of the electric-dipole allowed 4f-5d transitions and, thus, their increasing influence on the parity forbidden 4f-4f transitions. The excitation mechanisms in the chloride systems are investigated by time-resolved spectroscopy and the respective dynamics is studied by a rate-equation model. In the diluted sample 4I9/2 —> 4S3/2, excited-state absorption plays a major role and occurs within 3 cm"1 of the ground-state absorption, whereas the dynamics in the concentrated system is dominated by energy-transfer upconversion in all excitation steps. Of the 35 most likely energy-transfer upconversion processes, eight are found to contribute significantly to the excitation mechanisms in the concentrated system.
NEW VUV PHOSPHORS USING DOWNCONVERSION KOERT OSKAM Debye Institute, Department of Condensed Matter Utrecht University P. O. Box 80 000 3508 TA Utrecht, THE NETHERLANDS For applications like mercury-free fluorescent lamps and plasma display panels, new VUV phosphors are needed to convert VUV radiation into visible light. To enhance the efficiency of VUV phosphors 'quantum cutters' can be used: a quantum cutter has a quantum efficiency higher than 100%. Some rare earth ions are known to have such properties, but losses in the IR and/or UV parts of the spectrum are too large to use them for practical purposes. To obtain sufficient high visible quantum yields the concept of downconversion is used. In a downconversion process, the quantum cutting ion transfers all or part of its excitation energy to other ions. An example of a downconversion system is the Gd3+ - Eu3+ couple. Upon excitation of Gd3+ in the VUV levels energy is transferred to Eu3+ in two steps, followed by the emission of two photons in the visible part of the spectrum. In a sample of LiGdF4 doped with 0.5% Eu3+, we derived from the emission spectra upon VUV excitation a quantum efficiency of ±190%.
562 LATTICE DYNAMICS OF CdS/ZnSe SUPERLATTICES ALEXANDER DINGER Institutfur Angewandte Physik der Universitdt Karlsruhe Kaisestr. 12 D-76128 Karlsruhe, GERMANY We investigate the lattice dynamics of cubic CdS/ZnSe superlattices (SLs) by means of Raman scattering and Far-Infrared spectroscopy. This superlattices are the special interest because it is one of a few superlattice systems formed of II-VI semiconductor materials with no common anion ore cation, also called AB/CD superlattices . The superlattices of usually 200 periods have been grown on a GaAs(OOl) substrate by molecular-beam epitaxy using CdS and ZnSe compound sources. Due to the good lateral and in-depth homogeneity of the superlattice, period folded longitudinal acoustic phonons (FLAPs) up to the third order are observed in Raman scattering. The experimentally determined energies are in good agreement with the calculated values using an elastic continuum model (Rytov-model) and show the expected dependence on the SL period. Furthermore, these experiments give information about the sound velocity of cubic CdS along the (00Indirection, which has not been determined experimentally so far. Besides strain induced energy shifts due to the lattice mismatch of 2.6% between CdS and ZnSe, pronounced confiment effects for the optical phonons are observed. Plotting the experimentally determined energies versus the effective wave vector and taking strain into account, we are able to reproduce the dispersion relation of the ZnSe longitudinal optical phonons over the first half of the Brillouin zone. Therefore, by means of Raman scattering it is possible to determine the dispersion relations of the bulk materials. This holds in general and provides, besides neutron scattering, an alternative and even more accurate method. In addition to the above mentioned folded and confined modes, we observe Zn-S and Cd-Se interface modes caused by the AB/CD character of our superlattices. Due to the type II electronic band-alignment these modes become even very strong in Raman scattering experiments. The energies of these modes are in very good agreement with the calculations based on a linear-chain model. This holds also for the confined optical modes.
563
EXCITON-EXCITON INTERACTION IN ZnSe/ZnMgSSe MULTIPLE QUANTUM WELLS SVEN WACHTER Institutfur Angewandte Physik der Universitdt Karlsruhe Kaiserstr. 12 - Postfach 6980 D-7 6128 Karlsruhe, GERMANY II-VI semiconductors provide an excellent system for studying exciton-exciton (X-X) interaction. In particular, ZnSe quantum wells with a quarternary ZnMgSSe barrier are interesting because both electrons and holes are confined in the well leading to a very large exciton binding energy. We are investigating multiple quantum wells with wellwidths in the range of 5-8 nm. The X-X interaction is studied by degenerate two beam four wave mixing (FWM) spectroscopy using laserpulses with a halfwidth of about 100 fs and a spectral width of about 20 meV. Despite the broad excitation we can selectively excite the heavyhold exciton since the hh-lh splitting is larger than 25 meV. Furthermore, signal-contributions of the exciton and the biexciton are easily distinguished, because of the large biexciton binding energy. In order to study the influence of the exciton density on the homogeneous linewidth (extracted from the measured dephasing time) we create an exciton background density by applying an additional prepulse and varying either the power of the prepulse and/or the delay between prepulse and actual FWM-measurement. As can be expected we find for linearly polarized pulses a linear dependence of the homogeneous linewidth of the hh-exciton on the additional background density (ABD). The most interesting point is that the homogeneous linewidth depends on the time of creation of the ABD. The interaction of the excitons is strongly enhanced if they are created within 5-10 ps before the experiment. On this timescale one can be sure that there are no lifetime effects involved. In further investigations we learned more about the reasons of the enhancement. If circularly polarized s+ (s-) prepulse is used, then the particular enhancement is only observed, if the experiment is conducted also with circularly polarized s+(s-) pulses. This indicates that in order to observe the effect the exciton background created by the prepulse must have the same spin orientation as the excitons created during the FWM experiment. Thus we could show that the interaction of spin-coherent excitons is much stronger than of spin-incoherent excitons due to the Pauliexclusion principle.
564 SELF-ORGANIZED II-VI SEMICONDUCTOR QUANTUM ISLANDS Elisabeth Kurtz Institutjur Angewandte Physik der Universitdt Karlsruhe Kaiserstr. 12 - Postfach 6980 D-76128 Karlsruhe, GERMANY II-VI semiconductors provide an excellent system for studying exciton-exciton (X-X) interaction. In particular, ZnSe quantum wells with a quarternary ZnMgSSe barrier are interesting because both electrons and holes are confined in the well leading to a very large exciton binding energy. We are investigating multiple quantum wells with wellwidths in the range of 5-8 run. The X-X interaction is studied by degenerate two beam four wave mixing (FWM) spectroscopy using laserpulses with a halfwidth of about 100 fs and a spectral width of about 20 meV. Despite the broad excitation we can selectively excite the heavyhold exciton since the hh-lh splitting is larger than 25 meV. Furthermore, signal-contributions of the exciton and the biexciton are easily distinguished, because of the large biexciton binding energy. In order to study the influence of the exciton density on the homogeneous linewidth (extracted from the measured dephasing time) we create an exciton background density by applying an additional prepulse and varying either the power of the prepulse and/or the delay between prepulse and actual FWM-measurement. As can be expected we find for linearly polarized pulses a linear dependence of the homogeneous linewidth of-the hh-exciton on the additional background density (ABD). The most interesting point is that the homogeneous linewidth depends on the time of creation of the ABD. The interaction of the excitons is strongly enhanced if they are created within 5-10 ps before the experiment. On this timescale one can be sure that there are no lifetime effects involved. In further investigations we learned more about the reasons of the enhancement. If circularly polarized s+ (s-) prepulse is used, then the particular enhancement is only observed, if the experiment is conducted also with circularly polarized s+(s-) pulses. This indicates that in order to observe the effect the exciton background created by the prepulse must have the same spin orientation as the excitons created during the FWM experiment. Thus we could show that the interaction of spin-coherent excitons is much stronger than of spin-incoherent excitons due to the Pauliexclusion principle.
565 ZnS:Mn 2+ NANOCRYSTALS: INFLUENCE OF S2" DURING SYNTHESIS FREEK SUYVER Debye Institute, Department of Condensed Matter Utrecht University - P. O. Box 80 000 3508 TA Utrecht, THE NETHERLANDS Fundamental research on nanocrystalline semiconductors in the past 20 years have yielded very interesting fundamental results, for example on quantum size effects. In many articles potential applications are mentioned, but until now no applications have been realized. A possible application is to use luminescent nanoparticles in electroluminescent devices. For this application efficiently luminescent nanoparticles are required. The system that is investigated here are ZnS nanocrystals doped with Mn2+. These nanocrystals are fabricated using inorganic synthesis route, in which a solution of Zn(CH3COO)2 (with some M n C y in water is mixed with a solution of Na2S in water, in the presence of a capping polymer. Using the right concentrations of Zn2+, Mn 2+ , S2" and the capping polymer yields luminescent ZnS:Mn particles of typically 3-5 nm diameter. In the emission spectra ZnS host lattice luminescence is observed at 420 nm and the Mn 2+ ion luminesces at 590 nm, which is attributed to the 4Ti —> 6Ai transition. In this study, data are presented on the influence of the S2" concentration on the particle diameter, reflectivity, luminescence emission spectra and the luminescence lifetime of the Mn2+. The goal of this work is to maximize the Mn 2+ related luminescence quantum yield. It was found that an excess of S2" with respect to the amount of Zn2+ used, leads to crystals with a typical diameter of 5 nm as determined by X-ray powder diffraction. Using an excess of Zn 2+ results in crystals with a diameter of 3.5 nm. Reflectivity measurements agree with the X-ray diffraction results and show a sudden shift in the onset of absorption (i.e. the bandgap energy) due to the change in particle diameter. The room temperature photoluminescence emission spectra (excitation at 300 nm) show a relative increase of Mn + related emission with respect to the ZnS luminescence intensity for increasing S2" concentration. A strong increase of three orders of magnitude in this ratio was observed in the transition region from an excess Zn2+ to an excess of S2~. The lifetime of the luminescence at 590 nm was found to be constant within the experimental uncertanties and was equal to the bulk value of the Mn 2+ lifetime: 1.9 ms. This agrees with other experimental evidence that the Mn 2+ lifetime should be size independent. The increase of the particle diameter as well as the luminescence intensity is explained as follows: Sulphur vacancies in the nanoparticles can act as recombination centers for the electron-hole pairs. The visible emission around 420 nm from the ZnS
566 could very well be due to sulphur vacancies, in analogy with the visible emission in ZnO, which has been related to oxygen vacancies. Preparation of particles using excess sulphur will strongly reduce the sulphur vacancy concentration. This results in a decrease of the recombination at sulphur vacancies (reducing the 420 nm intensity) and an increase of the (desired) competing process, recombination at Mn2+, yielding the well-known orange emission. The influence of the S27Zn2+ ratio on the particle size is not as clear. It may be related to differences in the nucleation and growth of the ZnS particles as a function of the concentration of Zn2+ and S2".
A METAL/INSULATOR TRANSITION IN A 2D ELECTRONIC SYSTEM AT B=0 JOHN E. FURNEAUX Department of Physics and Astronomy University of Oklahoma Norman, Oklahoma For the past few years we have been studying the transport properties of Si metaloxide-semiconductor field-effect transistors (MOSFETs) which have unusually high mobilities (up to 70,000 cm2/Vs). Surprisingly, in this system we observe clear evidence for a metal insulator transition at temperatures below 1 K. This transition is occurs at a critical 2D-electron density (n) of around nc=1011 cm" . By studying a number of samples, we have found that the critical parameter is the resistivity (p) of the 2D electronic system (2DES) rather than the density. There seems to be a universal critical parameter of about 3 h/e 2 . Furthermore, we see evidence for scaling not only with temperature but also with the electric field used for excitation. In this case scaling refers to the functional form of p(n,T) = p(T/T0) where T 0 = a&V2 and 5 n = In-nd/iv The scaling exponents (z,y) are the same on both sides of the transition. A similar scaling, p(n,E) = p (E/Eo) where Eo = aSn1*2"1"1' allows us to determine that z = 1 and y= 1.6. The overall transition is similar to the superconductor to insulator transition seen in thin metal films as a function of thickness and to quantum Hall transitions as a function of magnetic field. It is clear that the origin of this metal/insulator transition is electron-electron interactions. This does not cause a contradiction with the theory of AALK [Abrahams 1979] because interactions were not included there and there are clearly interactions present in the MOSFET system at low n and T. This can be particularly seen in the behavior of p(T) in the insulating regime. Here Coulomb gap behavior is seen such that p(T) = h/e exp{(T 0 /T) 1/2 }, a clear indication that interactions are important.
567
OPTICAL MICROCAVITIES BASED ON COLOR CENTERS IN LiF FRANCESCA MENCHINI ENEA Centro Richerche Frascati Via E.Fermi, 45 C.P. 45 00044 Frascati (Roma), Italy Recently, a lot of attention has been paid to the investigation of modifications in the spontaneously emitted radiation inside small-sized resonators. Starting from the first structures working in the millimetric and infrared ranges, the interest has turned towards the scale of optical wavelengths, which is widely used in the field of communications and optoelectronics. An optical microcavity is a periodic structure of different refractive index materials in which one or more spatial dimensions are of the order of the wavelength of the radiation propagating inside it. The optical confinement achieved in such a peculiar configuration changes the distribution of the electromagnetic field, thus modifying the emission properties of active materials placed inside the structure, such as the radiatedpower spatial and spectral distributions and the spontaneous emission rate. Within this frame, we have realized a half-cavity as a multi-layer stack of dielectric layers of submicrometric thickness. The active layer is constituted by a thermally evaporated LiF film, subsequently irradiated by low energy electrons to form point defect photoemitting in the visible, well known as color centers. The structure is projected to perform a constructive interference along its axis for the visible radiation emitted from the centers. This kind of resonators is interesting both for theoretical tests of cavity-QED and for the realization of novel kinds of solid-state miniaturized optical devices.
INVESTIGATING MORPHOLOGY AND PHASE-BEHAVIOR OF POLYMERS UNDER TENSILE DEFORMATION USING SINGLE MOLECULE MICROSCOPY WERNER TRABESINGER Lab. fur Physikalische Chemie Universitdtsstrasse 22 ETH-Zentrum CH-8092 Zurich, SWITZERLAND It has been demonstrated recently that the solid-state tensile deformation of flexiblecoil polymer matrices leads to excellent uniaxial orientation of embedded rodlike
568
conjugated polymer molecules. Due to their highly polarized fluorescence, these binary blend films are interesting candidates for applications in efficient display devices. We studied such systems at the level of single conjugated guest molecules by means of roomtemperature Scanning Confocal Optical Microscopy. The high sensitivity of this methodology in combination with its diffraction limited spatial resolution provides important insights into the deformation-induced morphology changes of the binary blend. Due to the fact that the guest chromophores are rigid linear molecules whose emission dipole moments coincide with the polymer backbone, they constitute ideal probes for the local morphology of the blend. The orientation of the guest molecules is dictated by direct physical interactions with their nanoscopic environment and can easily be determined in polarization measurements. This allows for visualizing the local orientation of the polymer blend at the positions of the probe chromophores. It is further possible to build up distributions for the orientation of the molecules in order to quantify the degree of film orientation induced by stretching. Another vital issue related to the use of polymer blends in device applications is the homogeneity of the systems. The small entropy of mixing of different polymers mostly leads to unwanted phase separation. With the help of our experimental technique it is possible to address individual spots on the sample with high lateral resolution and to count the number of molecules therein. Measurements on large sample areas allow for constructing a distribution that specifies the probability for finding clusters with a given number of molecules. By comparing such distributions for films with different draw ratios (equal extents of stretching) we demonstrate that tensile deformation induces the transition from initially phase-separated polymer blends to nearly perfect molecular solutions.
POSTERS
ENERGY TRANSFER AND FREQUENCY UPCONVERSION IN Pr3+ DOPED SYSTEMS ROLINDES BALDA Departamento de Fisica Aplicada I Universidad del Pais Vasco Alda - Urquijo S/N 48013 Bilbao, SPAIN We have investigated the process of fluorescence quenching from the 'D2 state of Pr3+ ions in fluoride and oxide crystals and in fluorophosphate glasses due to energy transfer among Pr3+ ions. The fluorescence of the rD2 level shows a strong concentration quenching even at moderate Pr3+ concentrations. The time evolution of the decays from the 'D2 state is consistent with a dipole-dipole energy transfer mechanism. Anti-Stokes emission from the 3P0 level following excitation of the 'D2 state is also studied for different Pr3+ concentrations. Analysis of the upconverted fluorescence decays supports that the mechanism responsible for the upconversion process is energy transfer. However, this process seems to be complex enough to allow for the use of a single model which could explain the behavior observed.
ORGANIC MOLECULAR BEAM DEPOSITION OF l,3,5(TRIS(6,7-DIMETHYL-3-PHENYL)QUINOXALINE-2-YL)BENZENE SYLKE BLUMSTENGEL Dipartimento di Scienza dei Materiali Universita degli Studi di Milano "Bicocca " Via R. Cozzi, 53 20125 Milano, ITALY Introduction of molecular order into the active layer of an organic light-emitting diode (OLED) is a promising approach to control luminescence as well as charge transport properties. For instance charge carrier transport is very sensitive to the degree of molecular order, packing and orientation, and mobilities can be considerably enhanced by proper molecular engineering. Organic molecular beam deposition (OMBD) has been shown to be a very powerful tool to build structures with well-defined molecular order if
569
570
the deposition conditions are properly chosen. Especially the choice of the substrate, substrate temperature and deposition rate is crucial. We report about OMBD of l,3,5(tris(6,7-dimethyl-3-phenyl)quinoxaline-2yl]benzene) (Me-TPQ), a starshaped molecule with electron deficient quinoxaline groups. This molecule belongs to the class tris(phenylquinoxalines) (TPQ). The properties of the TPQs depend strongly on the substituents on the quinoxaline core. Small substituents like the methyl groups of the TPQ used in this study lead to materials which have a tendency to crystallise. If bulky substituents or mixtures of isomers are introduced to the quinoxaline moiety the materials form glasses. Thermal analysis indicated that the TPQ used in this study has a crystalline phase and therefore, might yield crystalline films upon vacuum sublimation. We have studied the evaporation conditions in order to achieve control over the molecular order in the grown layer. X-ray diffraction (XRD) measurements have been performed to determine the type and degree of molecular order. Absorption and photo luminescence (PL)spectra of the grown films have been recorded in dilute solution and of the grown films in order to detect any influence of the molecular packing on the optical properties. The results of the XRD measurements reveal that the degree of order and orientation of the molecules with respect to the substrate surface can be varied by the choice of substrate. Sharp XRD patterns, i.e. well-ordered films, have been obtained for films grown on ruby-mica, the ME-TPQ molecules stand perpendicular with respect to the substrate surface. Ordered Me-TPQ films have been also grown on KAP, KBr and graphite substrates. On freshly cleaved KAP as well as on graphite the Me-TPQ molecules prefer a flat alignment on the substrate probably as a result of the 7C-7t interaction with the substrate, whereas on freshly cleaved KBr crystals the molecules do not lay completely flat on the surface but are inclined by an angle of roughly 20°. However, the XRD peaks are rather broad suggesting a poor overall order. On the other hand thin films grown on fused silica, ITO, Au and PPV are disordered even if the substrates are kept at elevated temperatures up to 280° during the deposition of the MeTPQ molecules. The shape of the absorption and PL spectra of the vacuum deposited films coincide with those taken in solution, indicating that the interaction between molecules in solid state is weak. No dependence of the shape and position of the absorption peaks on the packing of the molecules have been found. Absorption spectra of Me-TPQ deposited on ruby-mica, KBr and KAP, taken with linearly polarised light at normal and oblique incidence, reveal that the films are optically isotropic. In order to study the influence of the degree and type of molecular order on the charge transport and EL properties it is necessary to fabricate crystalline films on substrates which are not very suitable to induce epitaxial growth like thin vacuum sublimed metal electrodes, ITO or amorphous organic emitter layers. To overcome this problem we consider two techniques. Prior to deposition, the substrate can be covered with an oriented monomolecular layer, produced for instance by LB technique, which will induce ordered growth of the incident molecules. The second approach takes advantage of the fact that alkali halides as KBr are suitable substrates for OMBD to
571
obtain ordered thin films as we have shown above and therefore wet transfer method can be applied which allows the transfer of the vacuum sublimed organic film on any substrate required to study its electrical properties and fabricate OLEDs. Both approaches are currently studied and results will be published later. ENERGY TRANSFER ASSISTED BY LOCALIZED EXCITATIONS IN AMORPHOUS SOLIDS JOAQUIN FERNANDEZ Department de Fisica Aplicada Universidad del Pais Vasco Alda - Urquijo S/N 48013 Bilbao, SPAIN It has been known for some time that amorphous solids exhibit anomalous behaviors when compared with their crystalline counterparts. Among the various models proposed to explain these anomalous behaviors, it is worth pointing out the Soft Potential Model (SPM), which gives a unified description of these properties in terms of soft anharmonic potentials, the so-called Soft Potentials. In this work, our attention has been focused on the influence that characteristic lowenergy excitations of amorphous solids in the low temperature regime may have in the optical excitation transfer between Nd3+ ions in heavy metal fluoride glasses. We have found that the contribution of these excitations to the transition probability per unit time of an optical excitation transfer between Nd3+ ions follows a T1 law (T is the temperature), in full agreement with experimental data previously obtained in our laboratory for Nd3+ doped fluoride glasses. Our model allows to obtain independent estimations of the impurity-soft mode coupling constant (once the microscopic nature of the impurityimpurity interaction is known), which is usually estimated from a power law fit to the temperature dependence of the homogeneous linewidth of rare earth doped glasses. Thus, once these constants are known, we can discriminate among the various models which try to explain the temperature dependence of the homogeneous linewidth in glasses.
PHOTOLUMINESCENCE OF HYBRID ORGANIC/INORGANIC XEROGELS DOPED WITH Eu3+ MARIA RUTE DE AMORIM E SA FERREIRA Departamento de Fisica, Universidade de Aveiro 3810Aveiro, PORTUGAL The achievement of full color displays is one of the main challenging tasks on the field of luminescent materials, such as the stable and efficient white photoluminescence
572 sol-gel based silicates and the set of white light emission complexes based on lightemitting diodes recently reported. This work reports on the photoluminescence and local structure of sol-gel derived organic/inorganic hybrids, so-called Ureasils. Their organic/inorganic host matrix is a siliceous network to which short oxyethylene units (**) are covalently grafted by means of urea linkages. Three ureasil hybrids containing different numbers of oxyethylene units were synthesised. These new classes of luminescent ormosils were doped with different europium triflate concentrations, which allowed the combination of narrow yellow-red lines, assigned to intra 4f6 transitions, with the luminescence of the undoped hybrid itself. These hybrids are so multiwavelength phosphors (full-color or white light emitters).
PHOTOLUMINESCENCE OF DOPED GaN FILMS CRISTINA MARIA CARVALHO GASPAR Departamento de Fisica, Universidade de Aveiro 3810Aveiro, PORTUGAL GaN, its allows, QWs and MQWs have gained an important place among shortwavelength optical emitters and high temperature electronic devices. A major obstacle for the realisation of such devices from Metal-Organic Vapour Phase Epitaxy (MOVPE) grow GaN film is an apparent difficulty to achieve p-type material due to highly n-type residual conductivity. Several groups reported their ability to dope GaN using Mg as a dopant, in nitrogen environment to active Mg dopants. A strong blue emission is present when GaN films are heavily doped with Mg. Controlled n-type conductivity can be obtained using Si as dopant. It is now that Si replaces Ga in the lattice and acts as a single donor. Besides the near-band-gap emission centred at 3.461 eV, which becomes broader with increasing Si concentration also transitions at lower energies are observed. In the present work we analyze these emissions by time-resolved spectroscopy and steady-state spectroscopy
573 THE INFLUENCE OF CHEMICAL VARIATION ON THE SPECTROSCOPIC PROPERTIES OF CsCdCLB3.,:Ni2+ (x=0.3) STEFAN LUTHI Department of Chemistry, University of Queensland St. Lucia, Brisbane QLD 4072, AUSTRALIA In rare-earth doped compounds the non-linear optical behavior "photon avalanche" has been observed [11. For the first time in a transition-metal doped compound the phenomenon has been found in CsCdCl3:Ni2+ pl. Here we study the effect of chemical substitution (CI to Br) on the spectroscopic properties of Ni2+ to predict and optimize photon-avalanche behavior and to later use it as an excitation mechanism in NdCl3 and CsCdBr3. The mixed halide compound CsCdBrCl2 however does not fit into the trend. The increase in es along the series CsCdCl3->CsCdBrCl2-»CsCdBr3 indicates an increasing strength of the Ni-X s bonds (X = CI, Br). For the first time to our knowledge, the d-d transitions above 20,000 cm'1 are assigned in CsCdBr3:Ni2+ and reproduced by energy-level calculations. From our results we predict the excited-state transition 3T2g —> "A,, observed at 14250 cm"' in CsCdCl3:Ni2+, to occur -2000 cm'1 lower in energy in CsCdBr3:Ni2+. Excited-state excitation of this transition led to the photon-avalanche behavior observed in CsCdCl3:Ni2+.
LINEAR OPTICAL PROPERTIES AND STIMULATED EMISSION OF ZnO EPITAXIAL LAYERS MARTIN SCHMIDT Institutfur Angewandte Physik der Universitdt Karlsruhe Kaiserstr. 12 - Postfach 6980 D-76128 Karlsruhe, GERMANY Although the optical properties of ZnO bulk crystals are well understood, there has been a great interest to study the optical properties of thin ZnO films, because they reveal novel luminescence properties and show lasing at room-temperature at moderate excitation intensities. We investigate these thin films by means of transmission, luminescence and reflection spectroscopy. Our samples were grown in an ultra high vacuum chamber at 550 C on a sapphire substrate by ablating a highly pure ceramic ZnO target. The low temperature luminescence is dominated by two bound exciton emission lines. With increasing temperature the free exciton emission line appears, which is observed up to room-temperature. The stimulated emission, observed normal to the growth plane, is quite different to the luminescence spectra obtained by cw-excitation. At low pump intensities, we see a
574 broad band at 3.372 eV, which could be attributed to an exciton-exciton scattering process. With increasing excitation power a second emission band appears at the low energy side of the first band, which shows a red shift with increasing excitation intensity. This band could be due to an electron-hole plasma transition. Further experiments must be done to clarify the origin and physical mechanisms behind the two emission bands, using the "variable stripe length method."
PHOTOELECTROCHEMICAL BEHAVIOR OF CHLOROPHYLL a IN THIN FILMS LAURA TUGULEA Department of Electricity and Biophysics Faculty of Physics University of Bucharest P. O. BoxMG-11 Bucharest - Magurele 76900 ROMANIA Chlorophyll a is the major pigment in higher plant photosynthesis, being responsible for both light absorption and light induced charge separation. The photoelectrochemical behavior of chlorophyll a species P740 (polymerized water adduct of chlorophyll a, absorbing at 740nm) was investigated using chlorophyll a thin films, electrodeposited on both Sn02 and Sn02/Ti02 electrodes. Anodic photocurrents have been observed at both Sn02 and Sn02/Ti02 electrodes with chlorophyll a film electrodeposited, under potential-controlled condition. Generation of anodic photocurrents was interpreted schematically in terms of the electron donation from the excited P740-chlorophyll a to the the Sn02 or the Sn02/Ti02 electrode, followed by the reduction of Chi a cation by the free electrons from the electrolyte solution. The spectral behavior of the photoelectrochemical cells was investigated and compared with the absorption spectrum of chlorophyll a films. The photocurrent action spectra of chlorophyll a in electrochemical devices are used to explain the chlorophyll a behavior in the process of charge generation under illumination and in the transfer of electrons. The present study of photoelectrochemical devices based on P740 chlorophyll a, electrodeposited in thin films on large band gap semiconductors, lead to the conclusion that chlorophyll a is the photoactive component. The nature of the semiconductor electrode, i.e. nanoparticulate, porous or uniform film, at the interface with P740 chlorophyll a thin film proved to be an important factor in controlling the photoresponse. Optimization of the photoelectrochemical cell with respect to the efficient charge photogeneration and reduction of recombination losses can be achieved by controlling the parameters of both the chlorophyll a sensitized electrode and the electrolyte solution. The use of nanoparticles coated electrodes is promising for fabricating of efficient devices for light conversion, based on chlorophyll a.
FIRST ROUNDTABLE DISCUSSION
The first roundtable discussion was held on June 22, 1999 - the fifth day of the lectures. Professor Di Bartolo opened the discussion floor by inviting everybody to express their opinion and give positive and negative feedback about all aspects of the 16th course 'Advances in Energy Transfer Processes' including the lectures, the facilities, and the organization and to suggest ways of improvement. The first comments were on the structure of the lectures. The students stated that the lectures did not begin with enough basic introductory material to prelude to discussing more complex concepts. The students felt that, given the diversity in background of the participants, that the lectures should be more didactic, starting from the basics and building to the more complex aspects, as opposed to just a display of information. There was a suggestion that the lecturers try to keep to a schedule, allowing for a fifty minute lecture followed by a ten minute break. It was pointed out by several participants that some lecturers had too many transparencies per session, such that the lectures often ran overtime and cut short the break time. It was felt that sufficient break time was necessary to absorb the material and discuss some issues germane to the lectures before sitting down for another hour. Another issue raised was that no precedent had been set which invited questions during the lectures, so that discussion may take place on issues which merit further consideration. Some advice was offered regarding the preparation of transparencies. The concern was centred around the issue of colours to indicate or highlight experimental results or graphical data. Since the transparencies are Xerox copied to free the participants from taking notes, some opinion was voiced that the transparencies in colour do not show up on Xerox copies to indicate contrast or specific instances for which they were intended. Additional advice was offered regarding the format of the roundtable discussion itself. It was suggested that the participants separate into a few groups to discuss relevant issues in a smaller group setting first. The meeting could then reconvene and discuss the pertinent issues in a more focused and helpful way. Finally, a point was raised concerning the desire on the part of the students to be informed ahead of time of the possibility of presenting a poster or short seminar at the meeting. Professor Di Bartolo pointed out that this is a school, and we should be careful not to give the impression of being a conference.
575
SECOND ROUNDTABLE DISCUSSION
The second roundtable discussion began at 6:00pm on the evening of June 30, the last day of the school. Professor Mazur opened the meeting upstairs in the San Domenico building. He remarked that, while the first roundtable discussion served the purpose to steer the course in the right direction, the purpose of this second discussion was for the benefit of future meetings. He noted that it is always impossible to make everybody happy, but constructive comments on what was done well, what could be done better, and what was accomplished at this course would contribute to making future meetings better. With this said, the discussion was open to comments from the participants. The first comments were acknowledgements that the course had improved since the first roundtable discussion, because those later lectures had started from the fundamental aspects, and because drafts of the articles based on the lectures were made available. This led to a lengthy discussion about the logistics of preparing the article before the course and the usefulness of this endeavour. Some suggestions were as follows: 1) Lecturers could bring a representative journal article for distribution. 2) Lecturers could write a short introduction to provide an overview of their work. 3) Information could be posted on a website before the course. 4) Instead of providing more auxiliary material, lecturers could simply start from the basics. This last suggestion led to further discussion about returning to basics in the lectures, not just basics in the field of speciality, but basics in physics itself. A suggestion was made that the first day or two of the course should be devoted to very basic aspects, including definitions and vocabulary in the fields of study. The possibility of having a physics lecture on Solid State theory was mentioned. There was some concern that if all the lectures started from basics then there would be repetition. The counter-argument was that repetition is good, that sometimes seeing something two or three times brings better understanding and allows for time to absorb the material. The discussion then turned in the direction of improving the schedule for future meetings. The belief was very strong by the majority of participants that the schedule should be followed closely. The consensus was that lectures should be 45-50 minutes and the break time should be 10-20 minutes. In addition, it was felt that time should be scheduled for announcements before the meeting convenes in the morning or afternoon. It was also suggested that the afternoon breaktime should be shorter in the beginning of the course and become longer toward the end, when people have less energy. A concern that resurfaced several times was the fact that some lecturers presented too many transparencies per 50-minute talk. Someone mentioned that, as a general rule, 15-20 viewgraphs per 45-minute lecture was appropriate, allowing two to three minutes per slide.
576
577
Finally, the direction of the discussion turned to social aspects. It was suggested that the first excursion take place earlier, so that people can interact sooner and become familiar with each other. There was also a belief expressed that the second excursion was too many days from the first, such that six straight days of lectures resulted between free days. In closing, Professor Mazur thanked the participants for a stimulating experience and for the conversation and good questions inside and outside the lecture hall. He expressed a feeling that many friendships had started, and he encouraged everyone to keep in touch. The meeting adjourned at 7:00 pm on this note.
SUMMARY OF THE COURSE (C. Klingshirn)
The topic of this School was essentially a sequel to the one from 1983. A comparison of the contributions to the two schools shows which topics are still subjects of active research, which ones are mature and approaching the status of text-book knowledge, and which ones are essentially new. When considering the wide variety and interdisciplinarity of the topics treated this time, it becomes obvious, that this was indeed a school and not a workshop or a conference. In the following we shall shortly address the various fields and the speakers which presented them. For details see the included abstracts and later the book. Organic and Biological Systems Nobel Laureate J.M. Lehn gave a lucid introduction to the subject of energy transfer in organic systems with an amazingly wide variety of combinations of absorbing species, luminescent centers and electron "transfer tubes" between them. S. Weiss explained the transfer from organic light absorbers to emitters and showed how this interaction can be used e.g. to monitor the activities of organic devices like enzymes. M. Graetzel finally demonstrated in his lecture and in a real experiment the performance of a photovoltaic cell which consists essentially of two (transparent) electrodes one of which is coated with small semiconductor spheres (here Ti02), which in turn are covered with organic dyes as absorber, and the whole system is immersed in a suitable electrolyte. Localized Centers The main part of the school was, as usual, devoted to localized centers. B. Di Bartolo gave an extended theoretical overview of the quantum mechanical description of van der Waals interaction and of the energy transfer processes resulting from such interactions. W. Demtroeder explained the energy transfer between the electronic and vibronic excitations of atoms and molecules. The participants were then exposed to a wide and detailed overview of luminescent centers in various host materials. Most of them have technical relevance in solid state lasers or phosphors. Consequently both fundamental and applied aspects have been addressed. In the following we just want to list some highlights: G. Baldachini gave an overview of colour-centers. J. Barnes reviewed the application of lasers in sensing projects of NASA and N. Barnes took up this topic and illustrated, how the design of solid state laser materials can be optimized for these purposes. C. Ronda reported on luminescence enhancement via energy transfer in phosphors, and D. Huguenin on industrial applications of energy transfer processes in luminescent insulators and semiconductors, while L. Diaz deepened some of the aspects of energy transfer from the contribution of Di Bartolo.
578
579
Some contributions were devoted to nonlinear optical phenomena: X. Chen reported on upconversion in luminescence. G. Boulon presented four wave mixing experiments and F. Auzel unconventional emission in solids, like superfluorescence or anti-Stokes emission. Semiconductors and Related Topics A field of research, which has gained over the years an increasing weight in the series of schools organized by the International School of Sepctroscopy, consists of semiconductors and related topics. R. von Baltz gave an excellent theoretical introduction into various aspects of this field with strong links to the topics of most speakers treating among others the ways to describe a loss of phase in a quantum-mechanical system. This topic was taken up by M. Wegener, who showed, for example, how to measure the time it takes until an electron-phonon scattering process is completed (< lOOfs). J.M. Hvam treated the loss of coherence of excitons and biexciton on a slightly longer time scale between lOOfs and a few ps, while E. Mazur showed that pulses with the same duration but much higher energy density induce a breakdown of the band gap in GaAs and melting. D. Hulin and E. Mazur gave overviews of various experimental techniques to characterize the laser pulses and to use them for the determination of characteristic times of electronic excitations in solids. K. Nelson introduced the concept of phonons and of phonon polaritons and showed that the investigation of these particles allows monitoring of phase-transitions. C. Klingshirn extended the concept of polaritons to other collective excitations in solids, but also to transitions in gases or to the Moessbauer line in 57Fe. Another line of arguments lead to the concept of photonic crystals. Many of the contributions highlighted cross links between the various disciplines. We want to mention here only two of the many examples, namely the contribution by R. Reisfeld, who described how semiconductor nanocrystals can be produced by organic methods and by D. Talwar, who reported on the spectroscopy of localized centers in semiconductors. Contribution to the general education in Science There were two parts of the school, which increased the knowledge of all participants beyond the topic of the school namely the overviews of G. Costa and R. Console on elementary particle physics and on earth quake monitoring in Italy, respectively. Last but not least the posters and short seminars from many participants should be mentioned which gave an excellent overview of their various activities.
580 Some more aspects of this school In the round table discussion a new physical quantity has been introduced, the socalled TPM-value (transparencies per minute) of the speakers. There was general agreement, that this value should be ideally around 0.3 but must not exceed unity. It is an important aspect of a school that the participants learn also about the history, the culture and the lifestyle of the host country. This aim was accomplished by shorter outlines of the history given by Rino and by a more detailed one by our guide during an excursion to Selinunte and Segesta. The participants learned also about real life and how fragile our human existence is, by the sudden death of Dr. A. Gabriele, the director for Organizational Affairs of the Majorana Centre, when he was still on duty on a Saturday night. Though everybody has the theoretical knowledge that such things happen, it is a different thing, if they happen in reality and they demonstrate us the transience of our scientific work, of the hunting for grants, of the pride over prestigious publications or competition with colleagues. Some of the older participants who knew Dr. Gabriele and appreciated his valuable, reliable and unpretentious work for the Center, participated in the funeral in Trapani celebrated by the archbishop of Monreale. To demonstrate how close life and death are, one of the participants got a phone call, that his grandchild was born in the very night when Dr. Gabriele died. To conclude the author of these lines expresses the thanks on behalf of all the participants to Prof. Dr. B. Di Bartolo and his team and the one of the Majorana Center for organizing and running this excellent school.
LIST OF PARTICIPANTS Field of Work 1. Daniela Alba Centra Educativo Ignaziano Via Piersanti Mattarella, 38/42 90142 Palermo, ITALY
English Language
Tel.: (39) 091 625-6313; FAX (39) 091 625-6998 e-mail: [email protected] 2. Maria Rute de Amorim e Sa Ferreira Departamento de Fisica Universidade de Aveiro 3810 Aveiro, PORTUGAL e-mail: [email protected]
Photoluminescence and structural properties of hybrids organic/inorganic xerogels doped with rare earth ions.
3. Francois Auzel Group Optique des Terres-Rares UPR211,CNRS 1 Place Aristide Briand 92195 Meudon Cedex, FRANCE [email protected] (home) [email protected] (lab.)
Rare-earth doped glasses for optical amplifiers. Non-radiative transitions Up-conversion Powder coherent sources
4. Giuseppe Baldacchini ENEA, Centra Ricerche Frascati Via E. Fermi, 45 00044 Frascati (Roma), ITALY
Optical spectroscopy of solid-state laser materials, color centers and lasers. High resolution molecular spectroscopy and trace gas detection.
Tel. (39) 6 9400-5365; FAX (39) 6 9400-5334 e-mail: [email protected] 5. Rolindes Balda Departamento de Fisica Aplicada I Universidad del Pais Vasco Alda - Urquijo S/N 48013 Bilbao, SPAIN Tel.: (34) 946014258 FAX: (34) 94 6014178 e-mail: [email protected]
Linear and nonlinear spectroscopy of solid-state laser materials.
6. Oracio Barbosa-Garcia
Non-radiative energy transfer among impurity ions in crystalline materials.
Centra de Investigaciones en Optica Apartado Postal 1-948 37000 Leon, GTO. - MEXICO Tel.: (52) 47 -18 4625; FAX: (52) 47 17 5000 e-mail: [email protected] 7. Matias Bargheer AG Schwentner Experimentalphysik FU • Berlin;
581
Control of chemical reactions in the Condensed phase with ultrafast lasers.
Amimallee 1414195 Berlin, GERMANY FAX: (49) 30 838 3050 e-mail: bargheer® physik.fu-berlin.de James Barnes Head , Laser Systems Branch NASA Langley Research Center - MS 474 Hampton, VA 23681-0001, USA Tel.: (1) 757 864-1637; FAX (1) 757 864-8809 e-mail: [email protected]
Solid state lasers and non- linear materials.
Norman P. Barnes National Aeronautics and Space Administration Langley Research Center, MS 474 Hampton, VA 23681-0001, USA Tel. (1) 757 864-1630; FAX (757) 864-8809 e-mail: [email protected]
Develop solid state lasers and nonlinear optical systems for remote sensing applications. Spectroscopy and basic measurements of optical material develop laser and quantum mechanical models.
Mohamed Benatsou Laboratoire de Phisico-Chimie des Materiaux Luminescents, Bat 205 Universite Claude Bernard Lyon 1 69622 Villeurbanne Cedex, FRANCE Tel:(33) 4 72 44 83 21; FAX:(33) 4 72 43 11 30 e-mail: [email protected] .fr
Rare earth spectroscopy. Optical Amplification in optical fibers.
Laurent Bigot Laboratoire dePhysico- Chimie des Materiaux Luminescents Universite Claude Bernard Lyon 1 43. Blvd 11 Novembre1918 69622 Villeurbanne Cedex, FRANCE Tel:(33) 4 72 43 29 71; FAX:(33) 4 72 43 11 30 e-mail: [email protected]
Optical Amplification in erbium-doped silica fiber.
Sylke Blumstengel Dipartimento di Scienza dei Materiali Universita degli Studi di Milano "Bicocca" Via R. Cozzi, 53 20125 Milano, ITALY Tel.: (39) 02 644 85101/2/3: FAX: (39) 02 644 85400/1 e-mail: [email protected]
Fabrication of thin films by Organic molecular beam deposition. Investigation of optical and electrical properties of the grown films. Organic electroluminescence
Ageeth Bol Debye Institute, Department of Condensed Matter Utrecht University P. O. Box 80 000 3508 TA Utrecht, THE NETHERLANDS Tel.: (31) 30 253-3545; FAX: (31) 30 253-2403 e-mail: [email protected]
Luminescence of doped semiconductor nanoparticles.
583 14. Georges Boulon Laboratoire de Phisico-Chimie des Materiaux Luminescents Universite Claude Bernard Lyon 1 UMR 5620 CNRS 69622 Villeurbanne, FRANCE Tel:(33) 4 72 44 8271; Secretariat (33) 4 72 44 8321 FAX (33) 4 72 43 1130 [email protected]
Crystal growth and spectroscopic properties of doped-solid-state laser materials and nonlinear crystals.
15. Laura Bussotti LENS, Largo Fermi, 2 50125 Firenze, ITALY Tel.: (39) 055 2307-820, or 811; FAX: (39) 055 224-072 e-mail: [email protected]
Femtosecond transient absorption spectroscopy on molecular systems in solution.
16. John A. Capobianco Department of Chemistry and Biochemistry Concordia University 1455 de Maisonneuve Blvd., West Montreal, Quebec CANADA H3G 1M8 Tel.: (514) 848-3350; FAX: (514) 848-2868 e-mail: [email protected]
Laser Spectroscopy. Molecular Dynamics. Laser Materials.
17. Xuesheng Chen Department of Physics - Wheaton College Norton, MA 02766, USA Tel.: (1) 508 286-3977; FAX (1) 508 285-8278 e-mail: [email protected]
Spectroscopy of Laser Solids.
18. Rodolfo Console Instituto Nazionale di Geofisica Via di Vigna Murata, 605 00143 Roma, ITALY Tel: (39) 05 51860-417; Fax: (39) 06 5041181 e-mail: [email protected]
Seismology, earthquake statistics, identification of nuclear explosions.
19. Giovanni Costa Universita degli Studi Istituto di Fisica "Galileo Galilei" Via F. Marzolo, 8 35100 Padova, ITALY Tel.: (39) 049 757-323 (home) e-mail: [email protected]
Elementary particles. High energy physics.
20. Wolfgang Demtroder Erwin-Scrodinger Strasse 67663 Kaiserslautern, GERMANY Tel. (49) 631 205-2344 or 2281; FAX (49) 631 205-3903
Laser Spectroscopy of Molecules and Its applications.
584 e-mail: demtroed @ physik.uni-kl.de 21. Luis Armando Diaz-Torres Centra de Investigaciones en Optica Apartado Postal 1-948 37000 Leon, GTO. - MEXICO Tel:(52) 47 17 58 23 Fax:(52) 47 17 50 00 e-mail: [email protected]
Non-radiative energy transfer processes in solid state materials.
22. Baldassare (Rino) Di Bartolo Department of Physics, Boston College Chestnut Hill, MA 02167, USA Tel. (1) 617 552-3601; home (1) 781 483-3993 FAX: (1)617 552-8478 e-mail: [email protected]
Luminescence Spectroscopy; Photoacoustic, spectroscopy; Optical interactions in solids.
23. Daniel L Di Bartolo 1124 22nd Street North Bergen NJ 07047 Tel.: (1)201 271-0908 e-mail: [email protected]
Psycology; Cancer Research.
24. S. John Di Bartolo Department of Applied Mathematics and Physics Brooklyn Polytechnic University 6 Metrotek Center Brooklyn NY 11201, USA Tel.: (1)718 260-3614 e-mail: [email protected]
Theory of Superconductors.
25. Alexander Dinger Institut fur Angewandte Physik der Universitat Karlsruhe Kaisestr. 12 D-76128 Karlsruhe, GERMANY Tel. (49) 721 608-7479; FAX: (49) 721 607-593 dinger @ ap-pc513 a.physik.uni-karlsruhe.de
Investigations of the lattice dynamics and photoluminescence properties of CdS/ZnSe quantum well structures.
26. Claus Clemens Feldmann Philips GmbH - Forschungslaboratorien P.O. Box 500145 D-52085 Aachen, GERMANY Tel. (49) 241 6003-616; FAX (49) 241 6003-465 e-mail: [email protected]
Solid State Chemistry. Colloid Chemistry. Luminescent Materials.
27. Joaquin Fernandez Department de Fisica Aplicada Universidad del Pais Vasco Alda - Urquijo S/N 48013 Bilbao, SPAIN Tel.: (34) 94-6014044; Fax : (34) 94-6014178 e-mail: [email protected]
Linear and nonlinear spectroscopy of solid state laser materials
585 28. John E. Furneaux Department of Physics and Astronomy University of Oklahoma Norman, Oklahoma 73019 - 2061, USA Tel. : (1) 405 325-3961 (ext. 36103); Fax: (1) 405 325-7557 e-mail: [email protected] 29. Vidal Garner 251 Gerdensi#16 Cambridge MA 02138 Tel.: (1) (617) 617-4018
Transport in 2D electron systems (T< 1K); Fractal Quantum Hall, Quantum Hall etc.; Polymer electrolytes; (Li polymer batteries)
Business contractor
30. Cristina Maria Carvalho Gaspar Departamento de Fisica Universidade de Aveiro 3810 Aveiro, PORTUGAL Tel.: (351) 34 370356 e-mail: [email protected]
Optical characterization (timeresolved spectroscopy and photoluminescence spectroscopy) of semiconductors ( GaN, InGaN, MQW and ZnO).
31. Abraham W. (Bram) Gerbrandij Leiden Institute of Chemistry Gorlaeus Laboratories Leiden University - P. O. Box 9502 2300 RA Leiden, THE NETHERLANDS Tel.: (31) 71 527 4513; Fax: (31 )71 527 4488 e-mail:gerbrand@chem. Ieidenuniv.nl
Second-order nonlinear optics of dendrimers functionalized with NLO chromophores.
32. Michael Gratzel Ecole Polytechnique Federale de Lausanne Departement de Chimie - LPE CH-1015 Lausanne, SWITZERLAND Tel. (41) 21 693 3112; FAX (41) 21 693 6100 e-mail: [email protected]
Photovoltaic Solar Cells.
33. Kashawna Harling Department of Physics, Boston College Chestnut Hill, MA 02167, USA Tel:(1) 617 552-3596 (office); (1) 617 661-4018 e-mail: [email protected]
Spectroscopy of Laser materials.
34. Denis Huguenin IMRA-Europe Institut Minoru des Recherches Avancees BP 213 - Rue Albert Caquot 06904 Sophia-Antipolis, FRANCE Tel. (33) 4 9395 7375; FAX (33) 4 9395 7383 e-mail: [email protected]
Industrial applications of energy transfer processes
586 35. Arnaud Huignard Laboratoire de Physique de la Matiere Condensee Ecole Polytechnique 91128 Palaiseau, FRANCE Tel.: (33) 1 69 33 46 50; FAX: (33) 169 33 30 04 e-mail: [email protected] 36. DanieleHuhn ENSTA - Ecole Polytechnique Centre de I'Yvette 91761 Palaiseau, FRANCE Tel. (33) 1 6931 9708; FAX (33) 1 6931 9996 e-mail: [email protected]
Colloidal synthesis of inorganic luminescent nanoparticles.
Ultrafast phenomena and non-linear optics in semiconductors (quantum wells, quantum dots, bulk, amorphous, organics).
37. Jorn M. Hvam Research Center COM DK - 2800 Lyngby, DENMARK Tel. (45) 4525-5758 (direct); FAX (45)4593 6581 e-mail: hvam ©com. dtu.dk
Ultrafast dynamics / spectroscopy in semiconductor nano-structures and devices.
38. Thomas Justel Philips GmbH - Forschungslaboratorien Weisshausstrasse 2 D-52066 Aachen, GERMANY Tel. (49) 241 6003-375; FAX (49) 241 6003-442 e-mail: [email protected]
VUV Phosphors VUV Spectroscopy Solid State Chemistry Organic Phosphors
39. Claus Klingshirn Institut fur Angewandte Physik- Universitat Karlsruhe Kaiserstr. 12 D-76128 Karlsruhe, GERMANY Tel. (49) 721 608-3410; FAX (49) 721 607-593 home: (49) 721 848-920 e-mail: claus.klingshirn @ phys.uni-karlsruhe.de
Linear, non-linear and time-resolved spectroscopy of semiconductors and of semiconductor structures of reduced dimensionality.
40. Elisabeth Kurtz Institut fur Angewandte Physik der Universitat Karlsruhe Kaiserstr. 12 - Postfach 6980 D-76128 Karlsruhe, GERMANY Tel. (49) 721 608-6476; FAX: (49) 721 607-593 email:[email protected]
Molecular beam epitaxy. Fabrication of low dimensional II - VI heterostructures, quantum wells and self-organised quantum dots. Optical + structural properties of II VI quantum dots / wells.
41. Jean-Marie Lehn Universite Louis Pasteur Laboratoire de Chimie Supramoleculaire 4,rue Blaise Pascal 67000 Strasbourg, FRANCE Tel. (33) 3 8841 6056; FAX (33) 3 8841 1020 e-mail: [email protected]
Photonic molecular and supramolecular devices.
42. Kristjan Leosson Center for Communications,
Spatio-temporal dynamics of localized excitons in semiconductor
587 Optics and Materials Technical University of Denmark, Bldg. 349 DK-2800 Lyngby, DENMARK Tel.: (45) 4525 5785 e-mail: [email protected]
nanostructures.
43. Stefan Luthi Department of Chemistry University of Queensland St. Lucia, Brisbane QLD 4072, AUSTRALIA Tel.: (61) 7 3365 4107; FAX: (61) 7 3365 4299 e-mail: [email protected]
Optical spectroscopy and non-linear optical phenomena in transition-metal and rare-earth doped inorganic compounds.
44. Eric Mazur Division of Applied Sciences Harvard University - Pierce Hall 225 Cambridge, MA 02138, USA Tel. (1) 617 495-8729; Lab (617) 405-9616; FAX: (617) 405-9837 e-mail: [email protected]
Ultrafast optics: Femtosecond Spectroscopy of Materials, Optoelectronics and Material Processing.
45. Francesca Menchini ENEA Centra Richerche Frascati Via E.Fermi, 45 C.P. 45 00044 Frascati (Roma) Italy e-mail:menchini©frascati.enea.it
Spectroscopy of point defects in alkali halides as bulk crystals and thin films. Dielectric optical microcavities.
46. Antonino Messina Dipartimento di Scienze Fisiche ed Astronomiche Universita degli Studi; Via Archirafi, 36 90123 Palermo, ITALY Tel.:(39) 091 623-4243; FAX:(39) 091 617-1617 e-mail: [email protected]
Quantum Optics; nonclassical states (properties and generation) spin -boson interaction.
47. Anna Napoli Dipartimento di Scienze Fisiche ed Astronomiche Universita degli Studi; Via Archirafi, 36 90123 Palermo, ITALY Tel.: (39) 091 623-4237; FAX: (39) 091 617-1617 e-mail: [email protected]
Nonclassical states generation and detection; Micromasers.
48. Keith A. Nelson Department of Chemistry, Room 6-231 Massachusetts Institute of Technology Cambridge, MA 02139, USA Tel. (1) 617 253-1423; FAX (1) 617 253-7030 e-mail: [email protected]
Time-resolved spectroscopy of condensed matter; collective structural rearrangements; structural phase transitions; solid state chemistry; liquidglass transitions; thin films; fs pulse shaping; multiple pulse fs spectroscopy; optical control.
49. Stefan ia Nicolosi Dipartimento di Scienze Fisiche ed
Energy transfer among impurity ions in crystalline environment.
588 Astronomiche Universita degli Studi; Via Archirafi, 36 90123 Palermo, ITALY Tel.:(39) 091 623-4243; FAX: (39) 091 617-1617 e-mail: [email protected] 50. Koert Oskam Debye Institute, Department of Condensed Matter Utrecht University P. O. Box 80 000 3508 TA Utrecht, THE NETHERLANDS Tel.: (31) 30 253-3545; FAX: (31) 30 253-2403 e-mail: [email protected]
Solid state chemistry. Luminescence of solids. VUV spectroscopy.
51. T. Peres Concordia University Dept. of Chemistry 1455 de Maison Neuue Blvd.West Montreal, Quebec, Canada H3GIM8 Tel: 514-848-4268, Fax: 514-848-2868
Molecular Dynamics.
52. Anne Pillonnet Laboratoire de Physico-Chimie des Materiaux Luminescents Universite Claude Bernard Lyon 1 69622 Villeurbanne Cedex, FRANCE Tel.: (33) 04 72 43 29 71 e-mail: [email protected]
Alumina and Europiun doped alumina planar waveguides elabued by Sol.Gel process and pulsed laser deposition (PLD). Optical and structural properties ( M-lines, TEM, Raman, fluorescence, RX RBS AFM).
53. Giorgio Pozza ICTIMA, CNR Corso Stati Uniti,4 35127 Padova, ITALY Tel.(39) 049 8295-939; FAX: (39) 049 8295-649 e-mail:[email protected]
Laser induced photoluminescence of insulating inorganic materials doped with transition-metal and RE ions.
54. Renata Reisfeld Faculty of Sciences Department of Inorganic Chemistry The Hebrew University of Jerusalem Jerusalem, ISRAEL Tel.: (972) 2 658-5323: FAX: (972) 2 658-5319 e-mail: [email protected]
Nano-particles and interaction between quantum dots in glasses; rare earths. Energy transfer in glasses, sol-gel glasses, electrochromic, photochromic and gasochromic glasses.
55. Francis Romstad Center for Communications, Optics and Materials Technical University of Denmark, Bldg. 349 DK-2800 Lyngby, DENMARK e-mail: [email protected]
Ultra-fast femtosecond time resolved spectroscopy on InGaAsP semiconductor devices.
589 56. Cees Ronda Philips GmbH - Forschungslaboratorien P. O. Box 500145, D-52085 Aachen, GERMANY Tel. (49) 241 6003-397; FAX (49) 241 6003-442 e-mail: [email protected]
Hg consumption in fluorescent lamps. Luminescence of rare - earth and transition metal doped compounds. Light propagation in powder layers. Thermionic emission.
57. Martin Schmidt Institut fur Angewandte Physik der Universitat Karlsruhe Kaiserstr. 12 - Postfach 6980 D-76128 Karlsruhe, GERMANY Tel. (49) 721 608-7477; FAX: (49) 721 607-593 e-mail:[email protected]. uni-karlsruhe.de
Linear Spectroscopy of II - VI semiconductors and measurements of the absolut external quantum efficiency.
58. FreekSuyver Debye Institute Department of Condensed Matter Utrecht University P. O. Box 80 000 3508 TA Utrecht, THE NETHERLANDS Tel.: (31) 30 253-2241; FAX: (31) 30 253-2403 e-mail: [email protected]
Luminescence of doped semiconductor nanoparticles and device applications.
59. Devki N. Talwar Department of Physics Indiana University of Pennsylvania Weyandt Hall, Room 56 975 Oakland Avenue Indiana, Pennsylvania 15705-1087, USA Tel.: (1) 724 357-2190; FAX: (1) 724 357-5700 e-mail: [email protected]
IR and Raman Spectroscopy to study lattice dynamics of defects in semiconductors. Electronic bandstructure of semiconductors (bulk and superlattics). Infrared detectors based on quantum well and superlattices.
60. Werner Trabesinger Lab. fur Physikalische Chemie Universitatsstrasse 22 ETH-Zentrum CH-8092 Zurich, SWITZERLAND e-mail: [email protected]
Single Molecule Spectroscopy Comfocal Scanning Optical. Atomic Force Microscopy.
61. Laura Tugulea Department of Electricity and Biophysics Faculty of Physics University of Bucharest P. O. Box MG-11 Bucharest - Magurele 76900 ROMANIA e-mail: [email protected] 62. Liesbeth van Pieterson Debye Institute, Department of Condensed Matter Utrecht University
Photovoltric and photoelectrochemical devices. Model systems based on photosynthetic pigm.- spectroscopic studies.
Charge transfer luminescence of Yb3*
590 P. O. Box 80 000 3508 TA Utrecht, THE NETHERLANDS Tel.: (31) 30 2533545; FAX: (31) 30 2532403 e-mail: [email protected] 63. Ralph von Baltz Institut fur Theorie der Kondensierten Materie Fakultat fur Physik, Kaiser Str. 12 Universitat Postfach 6980 D 76128 Karlsruhe, GERMANY Tel. (49) 721 608-3367; FAX (49) 721 698150 e-mail: [email protected]
Optical properties of solids, plasmons, excitons, electronic properties of quasi - crystals. Photogalvanic effect.
64. Sven Wachter Institut fur Angewandte Physik der Universitat Karlsruhe Kaiserstr. 12 - Postfach 6980 D-76128 Karlsruhe, GERMANY Tel. (49) 721 608-7477; FAX: (49) 721 607-593 wachter® ap-pc513b.physik.uni-karlsruhe.de
Ultrafast Spectroscopy. Exciton-exciton-scattering. Time and spatially resolved photoluminescence of II - VI semiconductor quantum wells and quantum islands.
65. Brian Walsh NASA Langley Research Center, MS 474 Hampton, VA 23681, USA Tel.:(1) 757 864-7112; FAX: (757) 864-8809 e-mail: [email protected]
Solid-state laser spectroscopy of RE ions in crystals and glasses. Laser engineering.
66. Martin Wegener Institut fur Angewandte Physik- Universitat Karlsruhe Kaiserstr. 12 76128 Karlsruhe, GERMANY Tel. (49) 721 608-3400; FAX (49) 721 607-593 e-mail:[email protected]
Non-Markovian scattering proceses in semiconductors. Various spectroscopic techniques. Coherent control. Nearfield microscopy. Nanostructuring.
67. Shimon Weiss Materials Sciences and Physical Biosciences Division Lawrence Berkeley Laboratory Berkeley, CA 94720, USA Tel. (1) 510 486-5202; FAX (1) 510 486-5530 e-mail: [email protected]
Energy transfer molecules.
between
individual
INDEX Cavity polaritons, 196 Coherent energy transfer, 7 Collisional energy transfer, 140, 141 Defects in III-V compounds,
487
Earthquakes, 485 Electroluminescence, 395 Energy transfer Hamiltonian, 13 Energy transfer in color centers, 359 F centers, 363 FA centers, 365 F2 centers, 368 F3+centers, 368 Energy transfer in solids, 73 Energy transfer master equations, 523 Energy transfer processes, in atoms, 123 in doubly excited atoms, 124 in molecules, 124 irreversible dynamics, 101 reversible dynamics, 96 Energy transfer with migration among donors, 50 no migration among donors, 35 Exchange interactions, 26 Four-wave mixing,
308,411
Incoherent energy transfer, 10 Industrial applications of energy transfer processes, external control of energy transfer, 289 transfer between dopants and defects, 296 transfer from host lattice to dopants, 283 transfer via surface states, 293 591
592
Interactions among atoms, 1 in solids, 1 Interactions leading to energy transfer, Intramolecular transitions, 134
1
Lifetime measurements of molecules, 138 Luminescent materials, 377 energy transfer in, 384, 399-405 Memory function, 112 Migration as diffusion process, 50 Migration as random walk, 58 Modes of energy transfer, 30, 73 diffusion, 80 collective excitations, 87 scalar waves, 77 Multipolar electric interactions, 19, 384 Multipolar magnetic interactions, 23 Nanostructures, 310 Non Markovian scattering processes, in semiconductors, 215 experimental results, 233, 247 experimental techniques, 226 theory, 220 Optical Bloch equations, 222 Optoacoustic spectroscopy, 133 Overtone spectroscopy, 128 absorption spectroscopy, 128 optothermal spectroscopy, 132 Participants in the meeting, 581 Phonon-assisted energy transfer, 28 Photonic atoms, 207 Photonic crystals, 200 Photonic molecules, 209 Photovoltaic devices, 299 sensitization of electrochemical photovoltaics,
299
dyes and materials, 299, 301 semiconductor layers characteristics, 303 Polaritons, 174, 178 in crystals, 179 in disordered matter, 188 in systems of reduced dimansionality, 190 Posters presented at the meeting, 569 Quantum dots, 321 Quantum wells, 191,312 Quantum wires, 191,317 Rare earth ions in amorphous sol-gel films, 341 Round table discussions, 575, 576 Solid state lasers, 273 compositional tuned lasers, 276 energy transfer in, 331 for ozone measurements, 275 for remote sensing, 330 for wind velocity measurements, 275 Summary of the course, 578 Superluminal pulses, 84 Surface polaritons, 195 Techniques for the study of energy transfer, 147 pump-probe spectroscopy: absorption, 147 pump-probe spectroscopy: refractive index, time resolved luminescence, 158 Theory of everything, 477 Ultrafast spectroscopy, 307 quantum dots, 321 quantum wells, 312 quantum wires, 317 transient four-wave mixing, 308 Unconventional light emission, of rare-earth doped solids, 431 anti-Stokes emission, 451
594 superfluorescence, 432 superradiance, 432 upconversion processes, 451 Upconversion phenomena, 261, 451 excited state absorption, 263 energy transfer upconversion, 265 photon avalanche upconversion, 267 van der Waals Hamiltonian, 5 Vibrational energy in molecules,
124
ISBN 981-02-4728-1
www. worldscientific. com 4792 he
9 '78981o"24728r