Semiconducting Chalcogenide Glass III Applications of Chalcogenide Glasses SEMICONDUCTORS AND SEMIMETALS Volume 80
Semiconductors and Semimetals A Treatise
Edited by R.K. Willardson CONSULTING PHYSICIST
12722 EAST 23RD AVENUE SPOKANE, WA 99216-0327 USA
Eicke R. Weber DEPARTMENT OF MATERIALS SCIENCE AND MINERAL ENGINEERING UNIVERSITY OF CALIFORNIA AT BERKELEY
BERKELEY, CA 94720 USA
Semiconducting Chalcogenide Glass III Applications of Chalcogenide Glasses SEMICONDUCTORS AND SEMIMETALS Volume 80 ROBERT FAIRMAN Beaverton, OR, USA
BORIS USHKOV JSC ELMA Ltd Moscow, Russia
2004
ELSEVIER ACADEMIC PRESS Amsterdam
9 B o s t o n ~ L o n d o n ~ N e w Y o r k 9 O x f o r d ~ Paris
San D i e g o ~ San F r a n c i s c o 9 S i n g a p o r e
9 Sydney 9Tokyo
ELSEVIER B.V. Radarweg 29 P.O. Box 211, 1000 AE Amsterdam, The Netherlands
ELSEVIER Inc. 525 B Street, Suite 1900 San Diego, CA 92101-4495 USA
ELSEVIER Ltd The Boulevard, Langford Lane Kidlington, Oxford OX5 1GB UK
ELSEVIER Ltd 84 Theobalds Road London WC1X 8RR UK
9 2004 Elsevier Inc. All rights reserved. This work is protected under copyright by Elsevier Inc., and the following terms and conditions apply to its use: Photocopying Single photocopies of single chapters may be made for personal use as allowed by national copyright laws. Permission of the Publisher and payment of a fee is required for all other photocopying, including multiple or systematic copying, copying for advertising or promotional purposes, resale, and all forms of document delivery. Special rates are available for educational institutions that wish to make photocopies for non-profit educational classroom use. Permissions may be sought directly from Elsevier's Rights Department in Oxford, UK: phone (+44) 1865 843830, fax (+44) 1865 853333, e-mail:
[email protected]. Requests may also be completed on-line via the Elsevier homepage (http://www.elsevier.com/locate/permissions). In the USA, users may clear permissions and make payments through the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; phone: (+1) (978) 7508400, fax: (+1) (978) 7504744, and in the UK through the Copyright Licensing Agency Rapid Clearance Service (CLARCS), 90 Tottenham Court Road, London W1P 0LP, UK; phone: (+44) 20 7631 5555; fax: (+44) 20 7631 5500. Other countries may have a local reprographic rights agency for payments. Derivative Works Tables of contents may be reproduced for internal circulation, but permission of the Publisher is required for external resale or distribution of such material. Permission of the Publisher is required for all other derivative works, including compilations and translations. Electronic Storage or Usage Permission of the Publisher is required to store or use electronically any material contained in this work, including any chapter or part of a chapter. Except as outlined above, no part of this work may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission of the Publisher. Address permissions requests to: Elsevier's Rights Department, at the fax and e-mail addresses noted above. Notice No responsibility is assumed by the Publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be made. 1st. edition 2004 Library of Congress Cataloging in Publication Data A catalog record is available from the Library of Congress. British Library Cataloguing in Publication Data A catalogue record is available from the British Library. ISBN:0-12-752189-5 ISSN:0080-8784 (series) ~) The paper used in this publication meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper). Printed in Great Britian.
Working together to grow libraries in developing countries www.elsevier.com I www.bookaid.org I www.sabre.org
Contents LIST OF CONTRIBUTORS PREFACE
Chapter 1
Electronic Devices and Systems Based on Current Instability in Chalcogenide Semiconductors
Andrey S. Glebov 1.
SWITCHES AND MEMORY ELEMENTS
1
2.
NONLINEARSWITCHING ELEMENTS
3 3 8 11 13 14
2.1. CVS-based Thin Film Thermo-sensors Threshold Switch with Controlling Parameters The Acoustic-frequency Phase Inverter Voltage Stabilizers Based on SEs Elements of Calibration Microelectronics
2.2. 2.3. 2.4. 2.5. 3.
ELECTRONICDEVICES BASED ON SEs
3.1. Generators of Relaxation Oscillations 3.2. Control Circuits of Electro-Luminescence Indicators (ELI) 3.3. Converter of the Decimal Code into the Computer Code 3.4. Electroluminescence Matrix Displays
18 18 24 28 29
4.
MEMORY CELLS AND SWITCHES BASED ON HETERO-CONTACTS 'CRYSTAL-GLASS'
31
5.
RE-PROGRAMMEDMEMORY DEVICES ( R A M )
45
REFERENCES
54
Chapter 2
Heterostructures on Chalcogenide Glass and Their Applications
57
Dumitru Tsiulyanu 1.
METAL-GLASS JUNCTIONS
1.1. Formation of the Contact Barrier 1.2. Mechanism of Current Transport 1.3. Junction Capacitance 1.4. Effect of the Thick Insulator Layer at the Interface 2.
CRYSTAL-GLASS JUNCTIONS
2.1. Isotype Heterojunctions
57 57 63 67 70 75 76
Contents
vi 2.2. Glass~n-Crystal Heterojunctions
86
3.
GLASS-GLASS JUNCTIONS
90
4.
APPLICATIONSOF HETEROSTRUCTURESON CHALCOGENIDE GLASS
4.1. Photodetectors 4.2. Vidicon Camera Tubes and Electrophotographic Plates 4.3. Optical Information Recording Media and Lithography
91 91 92 96
REFERENCES
98
Chapter 3
Ion Conductivity and Sensors
103
E. B y c h k o v , Yu. T v e r y a n o v i c h a n d Yu. Vlasov 1.
ION TRANSPORT IN SILVER AND COPPER CHALCOGENIDE AND CHALCOHALIDE GLASSES
1.1. Introduction 1.2. Percolation Threshold and Critical Percolation Regime at Low Mobile Ion Content 1.3. Universal Trend of the Haven Ratio and Modifier-controlled Regime 1.4. Conclusions 2.
CHALCOHALIDE GLASSES BASED ON COMPOUNDS OF THE THIRD GROUP METALS
2.1. Introduction 2.2. Glass-formation Regions 2.3. Properties of Glasses 2.4. Diagrams of State 2.5. Systems with Exchange Reactions 2.6. Structural Investigations 2.7. Glass-forming Ability 2.8. Ion Conductivity and Ion Selectivity 3.
103 103 104 113 129 131 131 131 132 135 136 137 148 150
3.1. Selective and Low-selective Sensors: History 3.2. Sensor Fabrication 3.3. Mechanism of Chalcogenide Glass Sensors Operation 3.4. Sensor Applications 3.5. Conclusion
154 154 157 161 163 165
REFERENCES
165
RECENT ADVANCES IN THE FIELD OF CHALCOGENIDE GLASS CHEMICAL SENSORS
Chapter 4
Rare-earth Doped Chalcogenide Glass
169
Yu. S. T v e r ' y a n o v i c h a n d A. T v e r j a n o v i c h 1.
CHALCOGENIDE GLASS PROPERTIES AND LUMINESCENCE MATERIALS
169
2.
THE GLASSY SEMICONDUCTORS USED AS A MATRIX FOR INTRODUCTION OF LANTHANOIDS
171 171 172 173
2.1. The Basic Requirements for a Glass Matrix 2.2. The General Laws of Choice of Glass-forming System 2.3. Some of the Most Investigated Glass-forming Systems 3.
4.
STRUCTURE AND COMPOSITION OF THE ENVIRONMENT OF LANTHANOIDS IN CHALCOGENIDE GLASSES
178
STOKESLUMINESCENCE
183 183 190
4.1. Optical Excitation Through the Absorption Band of the Lanthanoid 4.2. Optical Excitation Due to Inter-band Absorption of the Glassy Matrbc 5.
ANTI-STOKES LUMINESCENCE
192
6.
OTHER OPTICAL PROPERTIES OF THE GLASSY SEMICONDUCTORS ACTIVATED BY LANTHANOIDS
204
REFERENCES
205
Contents
Chapter 5
Optical Fibers from High-purity Arsenic Chalcogenide Glasses
vii 209
M. F. Churbanov and V. G. Plotnichenko 1.
INTRODUCTION
209
2.
PREPARATION OF HIGH-PURITY CHALCOGENIDE GLASSES
210 210 210 216
2.1. Preparation of Vitreous Arsenic Chalcogenides 2.2. Transparency of Chalcogenide Glasses as an Impurity-sensitive Property 2.3. Preparation of High-purity Glasses Based on Arsenic Chalcogenides 3.
CHALCOGENIDE GLASS FIBERS WITH LOW OPTICAL LOSSES
3.1. Fabrication of Optical Fiber 3.2. Properties of Chalcogenide Glass Fibers 3.3. Some Applications of Chalcogenide Glass Fibers
219 219 222 226
REFERENCES
229
INDEX
231
CONTENTS OF VOLUMES IN THIS SERIES
237
This Page Intentionally Left Blank
List of Contributors
E. BYCHKOV (103), Argonne National Laboratory, Argonne, IL 60439, USA M. F. CHURBANOV (209), Institute of Chemistry of High-purity Substances, Russian Academy of Sciences, Nizhny Novgorod, Russia ANDRF.Y S. GLEBOV (1), Riazan State Radiotechnology Academy (RSRTA) V. G. PLOTNICHENKO (209), Fiber Optics Research Center at the General Physics Institute, Russian Academy of Sciences, Moscow, Russia DUMITRU TSIULYANU (57), Technical University, Department of Physics, Dacia Str. 41, Kishinau, MD-2060, Moldova A. TVERJANOVICH(169), Department of Chemistry, Saint-Petersburg State University, Saint-Petersburg, 198504, Russia Yu. TVERYANOVICH(103), St Petersburg University, 199034 St Petersburg, Russia Yu. S. TVER'YANOVICH (169), Department of Chemistry, Saint-Petersburg State University, Saint-Petersburg, 198504, Russia Yu. VLASOV (103), St Petersburg University, 199034 St Petersburg, Russia
ix
This Page Intentionally Left Blank
Preface
'Semiconducting Chalcogenide Glass III' is Volume 80 of the 'Semiconductors and Semimetals' series. Written by a group of Eastern European scientists, this volume is the result of comprehensive experimental results and theoretical analysis of problems related to the quite complicated and interesting subject matter of semiconducting chalcogenide glasses. This volume follows Volume 78, which treated abstract knowledge of the glass formation phenomenon as well as principles and paradigms of glass structure and chemical composition, and Volume 79, which discussed the analysis of properties of chalcogenide glasses dependent on the quintessence factors noted above. In Volume 80, the authors now treat the practical applications of semiconducting chalcogenide glasses in science and industry. In each chapter, descriptions of the practical applications of chalcogenide glasses are preceded by experiment descriptions, as well as discussions of the theoretical basis of the physical and chemical effects demonstrated in the research. These include electrical and magnetic fields, various kinds of radiation, and the influence on chalcogenide glass in an interface with other materials, including instances when it is included in a double-layer 'pie' ('glass-glass', 'glass-crystal', 'glass-metal'). Interested readers can compare the data presented here with the results of other, more abstract research that is the foundation of these efforts. The properties of chalcogenide glasses described here are dependent on chemical compositions and structures, in accordance with Kurnakov's main principle of physical-chemical analysis. These properties are reactions to the external impacts that provide the basis of phenomena that determine practical applications of any substance, any material, and chalcogenide glasses included. One of the important applications of semiconducting chalcogenide glasses described in Chapter 1 is their use in electronic devices and systems that are based on current instability, which was described in detail in Volume 79. Non-linear switching elements and memory elements are the main types of elements used in such systems. The former are used in thermosensors, threshold switches, acoustic-frequency phase inverters, and voltage stabilizers; the latter are the basis of reprogrammed memory devices and circuits. xi
xii
Preface
Following on from Chapter 1, Chapter 2 considers problems related with heterostructures 'metal-glass', 'crystal-glass', and 'glass-glass' involving the formation of contact barriers, mechanisms of current transition, junction capacitance, etc. The final part of the chapter is devoted to the practical application of chalcogenide glass-based heterostructures in photo-detectors, vidicons, and electrographic plates. Special attention is concentrated on developing fields of applications such as optical information recording and lithography. Chapter 3 is devoted to ion conductivity and sensors--devices that detect certain properties of the environment and are capable of transforming the response into electrical, optical, and other signals. Contributors have skillfully described the complicated physical-chemical situation in which the phenomena of detection, transformation, and registration of signals take place; signals which transmit measurements of pressure, temperature, mass, alterations of magnetic field, as well as concentrations of various chemical species (ions, molecules) in liquid solutions and gas phases. Optimal characteristics of sensors and their sensitivity are described, in connection with the optimal combination of factors necessary to indicate features of chalcogenide glasses such as chemical composition, structure and glass formation ability, and, of course, concentrations of the elements responsible for ion conductivity. The analysis of these features, as well as the solution of some experimental and theoretical problems caused by the necessity to provide their proper combination, has allowed the contributors to describe the technical basis for the creation of various types of chalcogenide glass-based chemical sensors with improved selectivity, high sensitivity, low detection limits, and also long-term stability and life expectancy in strongly acidic and aggressive media. Particular attention in Chapter 4 is devoted to a new group of luminescent materials that have been widely investigated recently--semiconducting chalcogenide glasses doped--with rare-earth metals La, Pr, Nd, Sm, Dy, Er, and others. Contributors have analyzed glass formation abilities, structures, chemical compositions and properties of chalcogenide glasses that are prospective candidates for matrices for doping with rareearth elements. Glass forming systems are proposed which are the most promising for the synthesis of effective luminescent materials, particularly the GeCh2-Ga2Ch3 system (Ch--S, Se, Te) doped with Nd, Pr, Sm, and Yb. Chapter 4 also contains descriptions of concrete structures and compositions of the nearest neighborhood of rare-earth ions in chalcogenide glasses as well as Stokes and anti-Stokes luminescence, including specific features and methods of excitation of rare-earth ions by the optical pumping. The final chapter of Volume 80, Chapter 5, is devoted to high-purity arsenic chalcogenide glasses that are used as optical fibers transparent in the 0.62-17.5 /zm range. Problems of transparency as a function of H, C, O and A1, Fe, Ni, Cr impurity contents are reviewed, as well as methods of preparation and purification of arsenic chalcogenide glasses and optical fiber production. Examples of applications of optical fibers for solving medical and industrial problems are provided, including non-contact temperature measurements in technological processes, and new possibilities for
Preface
xiii
applications in laser surgery are also described. The usage of the IR fibers allows qualitative and quantitative analysis of gases, vapors, and liquids to realize remote control of the environment. Optical fibers of chalcogenide glasses have increased capabilities of the IR spectroscopy allowing to channelize radiations of high power IR sources such as HF (2.7/xm), YAG-Er 3+ (2.94/xm), CO 2 (10.6 ~m), and other lasers used in medicine, industry, and other fields. In conclusion, it must be noted that not all leading scientists from Eastern European countries could take part in this work, and there are some resulting gaps in the descriptions of scientific and practical achievements in the field of chalcogenide glasses. Nevertheless, we hope this volume will demonstrate to Western readers the overall scope of topics investigated by Eastern European scientists. V.S. Minaev Editor-compiler
This Page Intentionally Left Blank
CHAPTER
1
ELECTRONIC DEVICES AND SYSTEMS BASED ON CURRENT INSTABILITY IN CHALCOGENIDE SEMICONDUCTORS Andrey S. Glebov RIAZAN STATE RADIOTECHNOLOGYACADEMY (RSRTA)
1. Switches and Memory Elements At present, the following thin film structures, as shown in Figure 1, are mainly used for production of switching elements (SEs) based on CVS in the microelectronic design: (a) (b) (c) (d)
the the the the
film-planar type; film of 'sandwich' type; film-isoplanar planar type; film of the 'well' type.
The film version of the 'sandwich' type (Fig. lb) is very promising for microminiaturization, and it was used for the first time in matrices RM-256 of Energy Conversion Devices company (Ovshinsky). In the film-planar design of the SE (Fig. 1a), the operation of film deposition of the vitreous semiconductor is the last one in the technological process that is very important because the properties of CVS films are very sensitive to various physical-chemical impacts. In our opinion, the most promising designs are those of the 'well' type of switch shown in Figure 1d and the isoplanar type (Fig. l c). The former combines technologies of both 'sandwich' and planar versions of elements, i.e., the film deposition operation is the last one. It is achieved by having the upper metal bus-bar deposited on the dielectric film by standard regimes with substrate heating, then a window is made in the bus-bar and the dielectric film until the lower electrode is open. After that, the deposition of the CVS film is carried out. In the isoplanar version of the element, reduction of the threshold voltages compared with the planar design is achieved through a sharp decrease of the inter-electrode distance being possible. This design allows combination of the integrated film switching element with a large class of integrated circuits without increase in the number of technological operations. Depending on properties of CVS films, SEs are divided into mono-stable (proper switches) and bi-stable (memory elements) switches. In the mono-stable switching, the element returns to the initial high-resistance state after the voltage is off. With bi-stable Copyright 9 2004 Elsevier Inc. All rights reserved. ISBN 0-12-752189-5 ISSN 0080-8784
2
A . S . Glebov (a)
(b) 3
4
2
3
(d)
(c) 3
,
3 2
2
1
1
FIG. 1. Layouts of switching elements (SE) and memory cells based on vitreous semiconductors symmetrical to the polarity of the applied voltage (bipolar): 1 - the dielectric substrate; 2 - the first metal electrode; 3 - the vitreous semiconductor film; 4 - the second metal electrode; 5 - the dielectric film.
I
3
Iopen
" iThr
_ I
ITr s I._____ ~ Ic1o -URem
_- _ UTr
1 _ _
~i
0.7 UThr UThr
U
FIG. 2. The static CVC of the S-diode based on chalcogenide glass: 1, the region of the low conductivity (HR); 2, the region of the negative resistance (NR), 3, the region of the high conductivity (LR). Ithr: the current running through the diode at the zero differential resistance (the transition from HR to NR), Uthr: the voltage on the diode corresponding to the current of switching, Itr: the current running through the diode at the zero differential resistance (the transition from LR to HR), Ut~: the voltage on the diode corresponding to the current of the r e v e r s e transition. Rclos e --- 0.7 Uthr//cl . . . . Ropen = Urem//open 9
Electronic Devices and Systems Based on Current Instability
3
switching, the high-conductivity (low-resistance) state can remain for an arbitrarily extended period. This state of SE can be obtained in the case of current transmission through the element with the value higher than a certain value (the recording current) during a certain time (recording time). To switch the memory element into a highresistance state is made possible by the transmission of a certain current (the erasing current) during a certain time (erasing time). The static CVC of mono-stable and bi-stable SE is shown in Figure 2.
2. Nonlinear Switching Elements 2.1.
CVS-BASED THIN FILM THERMO-SENSORS
Investigations of temperature influence on static CVC of SE have shown that the main parameters of SEsDRdose, Uthr, and lthr--are significantly dependent on temperature (Fig. 3). This allows using them to be used as temperature sensors in two operation modesDactive and passive (Petrov, Rozhkova, Rotsel, Shrainer, Raykin and Gorelova, 1985; Petrov and Gruznov, 1988). In the passive mode of the temperature sensor operation, the temperature dependence of the resistance (or the current) in the high-resistance state is used (Fig. 4), the temperature sensor operating as the usual thermistor. In this regime, the thermal sensitivity coefficient (B1) depends on the applied bias achieving saturation at some of its values. At low values of the electric field, the temperature dependence of the electrical resistance follows the expression: R = A 1exp(B1 / T),
(1)
L Ael0 6
300 90~ C 60~ C
200
\
t
,,
40~ C 7 20~ C
100 ~~,,~
i
50
"
lOO
lO~
150
u, v
FIG. 3. Static CVC of switching elements at different environmental temperatures.
4
A.S.
Glebov
Rclose, A. 10-6
30 I
12
20
f
4
10
-25
0
25
50
T, ~
FIG. 4. The dependence of the SE resistance in the close state on the temperature at different bias voltages: 1 10 V, 2--20 V, 3--30 V, 4--40 V.
where A~ is constant; B1 = Eo'/2k, the coefficient of temperature sensitivity of the passive TS; Eo-, the electroconductivity activation energy of CVS and k is the Boltzmann constant. In the active regime, the thermosensor operation is based on uneven transition from the high-resistance (Rclose ~ 10 6 - 1 0 7 ~-~) state to the low-resistance (Ropen ~ 102-10 3 ~-~) state on achieving of a certain threshold voltage, which follows the dependence in a wide temperature range:
Uthr = A2exp(B2/T) (2) where A2 is constant; B2 = Wn/2k, determines sensitivity of the SE and Wn is the activation energy of switching dependent on the composition of the active material and the design of the electrode system. The temperature dependence of the electrical resistance is presented, for convenience of thermosensor parameter calculations, as follows (Heivang, 1987): R T --
Roexp(B/T )
(3)
The value Ro is measured using the constant current (the static CVC, Fig. 3) at certain initial temperature, usually at 20 ~ R0 -- U / I - - 24 MI~. The constant B can be determined experimentally from the temperature dependence of the thermosensor's resistance (Fig. 4):
R T -- Roexp(Bo(T- To)/TTo) or
T - 1/(1/T - 1/Bo)(ln Rt/Ro)
(4)
E l e c t r o n i c D e v i c e s a n d S y s t e m s B a s e d on C u r r e n t Instability
5
Converting expression (4), we obtain the formula for the constant B: B-- TTo/(T-
(5)
T o ) l n ( R o / R t)
Substituting measured values of temperatures To and T and corresponding values Rt and R0, determined from the curves of Figures 3 and 4, in expression (5), we can determine B: Rt(600 ~
-- 24(B)/35 x 10-6(A) -- 686(k1~) B -- (293 x 333/333 - 293)1n 24 X 106/686 x 103 -- 8537
The temperature coefficient of resistance (TCR) is determined as the relative change of the thermosensor's resistance at the change of the environmental temperature:
TCR = (1/et)det/dT
= (1/et)(e
t - Ro)/(T
-
To)
(6)
Substituting expression (4) in expression (6) and differentiating, we obtain TCR = - B / T 2 = - 0 . 0 8 or - 8%/~ The power dissipation coefficient (H) can be calculated from the CVC and the temperature dependence of the resistance (Figs. 3 and 4). We determine the dissipation power for given temperatures To and T: PO -- IoUo = 24 ixW, P = I U = 840 txW H-
(P - P o ) / ( T - To) -- 2.04 x 10 -5 W/~
(7)
The coefficient of energy sensitivity (G) is determined as the input power necessary for reduction of the thermosensor's resistance by 1%: G-
H / T C R ~ 3 x 10 - 4
(8)
The time constant (~-)is a parameter characterizing the operating speed of the thermosensor and, as a thermal constant, is calculated from the ratio of its thermal capacity to the power dissipation coefficient. The time constant depends on the dimensions and the design of the thermosensors, as well as on environmental conditions, and is equal to several microseconds: "It
-
-
C t / H = 1.2 p~s
(9)
The maximal dissipation power (Pmax) at which the stable operation of the thermosensor is ensured can be obtained from the expression: Pmax = (Tmax - T o ) / R T - - 42 mW
(10)
The thermal sensitivity of the sensors in the passive regime is determined, in particular, by the activation energy of the electrical conductivity (Eo-) of the CVS (1), which depends on its chemical composition and the energy of its covalence-ion binding (Ecru) (Rozhkova, Glebov and Petrov, 1989). All this suggests that the thermal sensitivity of sensors can be controlled by selection of the CVS composition with large EcIB. In addition, the thermal sensitivity coefficient of the passive TS depends on the design of the electrode system. For example, reduction of the inter-electrode distance from 30 to 10 Ixm at a width of 100 txm increases the thermal sensitivity (in the range of the exponential dependence of the current vs. the voltage) (Tatarinov, 1976).
6
A . S . Glebov
The thermal sensitivity of the active thermosensors depends on the switching activation energy (Esw), which is also proportional to ECIB: Esw = AEcm - B
(11)
where A and B are constants dependent on the CVS system. Then, the threshold voltage follows the expression: Uth r = a z e x p ( E t h r / k T
(12)
) = azexp((AEci B - B)/kT).
It is therefore possible to change the thermal sensitivity of active thermosensors and to control it by selection of the CVS composition in accordance with calculated ECIB values. The current, passing through the SE at switching, changes from several txA to tens txA. It is sufficient, for example, for the commutation of relays or blocking devices to cut off a system from a power supply unit when its critical temperature is reached. The threshold temperature (Tthr) is determined by the Uth r value and set by the bias voltage (Ub) (Fig. 5). Temperature-response variations in the thermal relay regime do not exceed _ 10 ~ The geometry of the working area exerts a strong influence on the temperature dependences of the main parameters, as well as on the regimes of operations (active or passive) of the semiconductor elements as thermal sensors, particularly the width of electrodes proportional to the working area (Rozhkova et al., 1989). The influence of the width of the electrodes (l) on the threshold voltages and currents at different environment temperatures and constant inter-electrode distances (10 txm) is shown in Figures 6 and 7. Threshold voltages insignificantly and smoothly decrease with increasing of the width of the electrodes (Fig. 6), and there are three regions in the threshold current dependence: at the electrode widths below 30 ~m and above 300 txm, Ithr increases insignificantly, and in the range of 30-300 lxm, the dependence is more sharp (Fig. 7). This character of the dependence of the threshold voltage and current on the electrode's width is likely related to the fact that the whole active area of the elements works at narrow electrodes and the uniform current distribution exists in the working Tthr, ~
180
80
0
10
20
Ubias,
g
FIG. 5. The dependence of SE temperature vs. the bias voltage at different inter-electrode gaps (l = 100 ixm)--l: 30 ~m, 2 : 2 0 txm, 3 : 1 0 Ixm.
Electronic Devices and Systems Based on Current Instability
7
Uthr, V
3
120
90 2 60
30
0
2.5
5
In (1, gm)
FIG. 6. The dependence of the threshold voltage vs. the electrode's width at different environmental temperatures--l: 90 ~ 2:50 ~ 3" 30 ~
area. In increasing the electrode width, the uniformity of the current distribution gets broken, the formation of narrow current channels is observed, and the value of the current in the background area plays an important role. The ratio of the channel current to the background area current is determined by the electrode width and, with wide electrodes (more than 100 Ixm), the background current determines the switching process character and thus the regime of operation of CVS-based thermosensors. The active regime of the thermosensor operation is realized only at an electrode width lower than 20 Ixm. At an electrode width of about 100 ixm, active and passive regimes of operations are observed (depending on the temperature ramp). At a further increase of the electrode width, a great increase of the background current takes place that causes disappearance of the switching effect and the element operates as the thermistor.
Ithr,
f
gA
20 ~
~
I / 3
2
3
4
In (1, gm)
FIG. 7. The dependence of the threshold current vs. the electrode's width at different environmental temperatures--l: 90 ~ 2:50 ~ 3:30 ~
8
A.S. Glebov
Ie 103, A
/
114,5 J
I23,5
50
100 Uthr , V
FIG. 8. The CVC of the switch obtained in darkness (1) and at illumination by the filament lamp (2).
2.2.
THRESHOLD SWITCH WITH CONTROLLING PARAMETERS
Based on the 'nonuniformity' model of switching, the switching current value must be determined by the background current value. The observed effect of 'reverse' switching in CVS-based film-planar switches in the system P b - S e - A s T e - S , which possesses noticeable photosensitivity along with the switching effect (Guznov, Glebov and Petrov, 1984), can be considered as the experimental confirmation of this thesis. If the switch is illuminated by light, the value of the switching current (12) becomes higher than its dark value (I1) (Fig. 8). The effect considered above was used for the creation of CVS-based light-controlled switches. Film-planar structures on cover glass of 0.2 m m thickness were used for investigations. The width of the m o l y b d e n u m electrodes and the distance between them varied from 10 to 900 Ixm and from 10 to 200 txm, respectively. The relative resistance variation of the CVS-based film element at illumination is equal to a = AR/R
The values of relative conductivity variations are given in Table I: ,40-//o-= ce/1 - ce TABLE I RESULTS OF PHOTOSENSITIVITYMEASUREMENTSOF CVS FILMS Composition Ga 1eGe6.5Si6.sAsesTe5o Ga7.sPbloSi7.5As25Teso Ga4Cu7Ge 16.64Te76.36 PbalnaGe15.64Te76.36 GenSne.88Si14.4Te78.75 PbloAsazS6SeleTe3o Pb loAS40S6Se4Te40 Pb loAS44S2SeaTe40 Pb6As41.8S1o.4Se2o.9Te2o.9 Pb3As43,1Ge10.8Se21.55Te21.55 Ge4Pbl.a2Sila.aTe80.18
AR/R (%)
Ao-/o- (%)
25 22 19 24 23 15 17 13 28 50 25
33 28 24 31 30 18 21 15 39 100 33
Electronic Devices and Systems Based on Current Instability
9
Investigations have shown that the photoconductivity value does not depend on the electrical-field intensity, the electrode width on the inter-electrode distance that indicates on the volume character of photoconductivity. Barrier effects on the ' m e t a l - C V S ' interface can be neglected. The average value of the relative conductivity increase is in the range of 20% for the absolute majority of compositions. From Table I, the conclusion can be made that the most promising composition for creation of the light-controlled threshold switch is Pb3As43.1Se21.55Te21.55Gelo.8. To convert the switch into the highconductivity state, it is necessary either to exceed Uth r o r tO run the current I (11 > I > 12) through it and eliminate illumination. In other words, the switch can be converted from the high-conductivity to the low-conductivity state, not by illumination with light, but by elimination of illumination. Such an effect of the 'reverse' switching is observed in switches where the electrode width exceeds 10 ~xm. At that, the threshold current at illumination can vary on o n e - t w o orders of value that is quite sufficient for stable operation of the switch. For production of memory elements with increased photoconductivity, i.e. those whose threshold characteristics can be controlled by light, it is proposed to use CVS of the G a - P b G e - T e system (Guznov and Sazhin, 1985). The current-controlled switch can be produced in two versions. In one of version, two connected-in-series elements are located on a common dielectric substrate on which three metal-film electrodes are deposited, one electrode being common for both elements. Layers of vitreous semiconductor material are deposited between electrodes. To ensure different values of the electrical resistance of the semiconductor materials in a highresistance state, different geometrical dimensions of the active area of the device are used. For example, in the second element, the inter-electrode distance is less and the overlapping electrode area is larger than those of the first element. This ensures reduction of the electrical resistance of the second element in the high-resistance state compared with the first element. The connection of two similar SEs with different electrical resistances in the highresistance state in the series circuit gives a new technical effect: a quite small change of voltage on one element can control switching of another element with quite high threshold voltage. Figure 9 shows the electric circuit of the device. Figure 10 shows the current-voltage characteristics of two elements of the device, and Figure 11 shows one of the versions of the device design. The device consists of two elements, 1 and 2 connected in series (Fig. 9). In one version (Fig. 11), element 1 is located on common substrate 3 with element 2 as filmplanar structure. The electrical resistance of element 1 in the high-resistance state is significantly higher than the electrical resistance of element 2. This is achieved owing to the different geometry of the inter-electrode distance between electrodes 5, 6 and 6, 8. In particular, the inter-electrode distance in element 1 (electrodes 5, 6) is larger than that in element 2 (electrodes 6, 8) (Fig. 11), and the overlapping area of electrodes 5 and 6 in the active area is less than that of electrodes 6 and 8. It is also possible to create elements with different resistances in the high-resistance state by selection of the composition of CVS. Elements 1 and 2 (Fig. 9) are connected to the circuit of the constant current source 9. To control the voltage on element 2, the regulated source of constant voltage 10 is connected to the circuit. Resistor 11 is the circuit load. To restrict the current in element 2, resistor 12 is connected. If the device is
10
A. S. Glebov
L, 10
/
/ \ 9
I
I
FIG. 9. The circuit diagram of the device connection.
created as in the version in Figure 11, electrode 6 is common for elements 1 and 2 and connected to regulated constant voltage source 10. Electrode 5 is connected to load resistor 11, and electrode 8 is connected to the common bus-bar of sources 9 and 10. In the regulated voltage source, the voltage U1, dropped on element 1, is set slightly less than the threshold voltage (Uthrl) (Fig. 10). In voltage source 10, voltage U2 is set somewhat lower than the threshold voltage of element 2 (Uthr2). In this case, elements 1 and 2 are in the high-resistance state. On increasing the voltage on element 2 due to, for example, increasing voltage E2 of source 10 to the value U t h r 2 , a jump decrease of the electrical resistance of element 2 occurs and it converts into the low-resistance state. In this case, the voltage drop on element 1 will be increased, causing it to switch from the high-resistance state into the low-resistance state. To convert the switches into the highresistance state, it is necessary to decrease the current through element 2 to a value lower than I2, which will cause it to switch into the high-resistance state, and then the current
s
I I il
!l
/
_ i i
I UO
/ /
/
i
I I i
.
'
I I
I I
I i
I I
I
I
U 2 Uthr2
~ ~
"
Ul
gthrl
g
FIG. 10. Voltage-current characteristics of two elements of device.
Electronic Devices and Systems Based on Current Instability
7
8
11
,3
FIG. 11. The design of the device. through element 1 will be decreased to a value lower than I1 and element 2 will return to the initial high-resistance state. This current-controlled threshold switch has a higher control efficiency and allows electrical commutation of loaded circuits of significant power supply voltages by small voltages.
2.3.
THE ACOUSTIC-FREQUENCY PHASE INVERTER
The phase inverter is a four-terminal network which is connected before the load Z2 (Fig. 12) and creates the phase shift q~i between currents I2 and 11 or q~v between voltages U1 and U2. The sign is determined by the rule that q~l is phase I2 minus phase 11, and q~v is phase U2 minus phase U1. The positive phase shift means an advance of values on the load Z2 regarding the input of the four-pole network. Four-terminal networks giving either constant or smoothly changing phase shift are known (Aseev, 1959). They usually consist of links of constant and variable resistors, active and reactive, connected in series or as bridge circuits. The known circuits of phase inverters have a disadvantage such that variable active resistors, variable capacitors, or goniometers are used to obtain a smooth change of the phase shift. The presence of variable active and reactive elements makes the circuit complicated and does not allow creation of phase inverters with the smooth phase shift change in the purely film form because reactive parameters L and C have large values in the sound frequency region. For example, for 500 Hz frequency, the capacitance of the bridge-phase inverter has the value of several microfarads (Aseev, 1959).
I1
12
Z2 ~ U2
U1 0 FIG. 12. The phase inverter.
12
A. S. Glebov
1
.(~
.(~
2
3
5
FIG. 13. The circuit diagram of the phase inverter based on chalcogenide S-diodes.
It is known that chalcogenide switches have the inductance character of reactivity. The value of the inductance depends on the bias voltage. The detailed investigations of inductance of CVS-based SEs in various versions are presented, for example, in Baryshev (1971). Using this property of chalcogenide switches, the circuit of the inductance-phase inverter for the sound frequency range with the smooth phase shift change has been developed. The inductance-phase inverter (Fig. 13) consists of the SE (3) and the active resistor (4) connected in series. SEs are created on the basis of chalcogenide glass as a film-planar or a film-sandwich design. The value of resistor (4) is less than the total resistance of the switch at the close region of the current-voltage characteristic near the threshold voltage. The bias voltage (for the smooth phase shift change) is applied to terminal 1. The signal of the sound frequency whose phase should be shifted is applied to terminal 2. The phase-shifted voltage is taken from terminal 5. Tests have shown that, in the case of the SE in the film-planar design with the interelectrode distance of 5 txm, the circuit provides the smooth phase shift change in the range from 0 to 70 ~ at changing bias voltage from 0 to 30 V (Fig. 14). Using SEs
~D ~D
80
/
y
60
~ 40 9
.,I
~ 2o
I
J 20
22
24 Bias voltage, V
26
28
FIG. 14. The dependence of the phase shift vs. the bias voltage.
Electronic Devices and Systems B a s e d on Current Instability
13
of the bead design, the change of the bias voltage in the same range provides a phase shift change from 0 to 80 ~ At the stepwise connection of two and three phase inverter circuits, it is possible to achieve phase shifts of 140 and 210 ~, respectively. This circuit of the phase inverter is remarkable for its long-term durability because chalcogenide switches operate in closed states. In addition, this circuit can be realized in the film design with a single technological cycle, for example by thermal evaporation in a vacuum. 2.4.
VOLTAGESTABILIZERS BASED ON SEs
In investigations of film-sandwich structures A1-CVS-A1, it has been revealed that thin structures (with the CVS film not more than 1 txm thick) have the varistor-wise current-voltage characteristic (Fig. 15, curve 1) (Petrov, Sokolov, Semenov, and Balabanov). For thicker CVS films (for example, 3 txm, curve 2, Fig. 15), a region of negative resistance appears on the current-voltage characteristic. Further increase of the chalcogenide film thickness leads to the appearance of the switching effect on the CVC of diodes. Figure 15 (curve 3) shows CVC of diodes with a film thickness of 5 Ixm. Such a dependence of the CVC type vs. the CVS film thickness can be explained by assuming the presence of a transition A1203 layer between the aluminum electrodes and the chalcogenide glass (Petrov, Oreshkin, Glebov, Timofeev, Baryshev and Semenov, 1971). The barrier layers of Mott's type can localize on this transition layer (Stylobans, 1967), whose CVC is described by the formula: j = evan(O)(exp(eg/kT)
-
1)
(13)
Here, j is the current density; 1)d, the charge carrier drift velocity in the barrier layer; n(0), the concentration of current carriers in the barrier layer on the interface with the metal electrode; V, the part of the external voltage dropping on the barrier layer (V > 0 for the forward voltage); k, the Boltzmann constant and T is the absolute temperature. Because the A1203 layer (more exactly, the whole transition layer between the metal electrode and chalcogenide glass) has a small thickness, formula (13) can be used only as a qualitative illustration. Moreover, it is known that, in reality, because of the decrease
mA
12,
~3 1
0
1 q~
)
IP
I)
q~
|1
t|
10
20
30
40
50U, V
FIG. 15. Static CVC of switching elements with the electrode's area of 0.9 • thicknesses--l: 0.2 ~m, 2:3 p~m,3:5 ~m.
10 - 3 c m 2
and film's
14
A. S. Glebov
of the work function of the metal, the current can increase more sharply, and the tunneling through the layer can be observed at small thicknesses of the barrier layer (Stylobans, 1967). Small thicknesses of chalcogenide films, can lead to increase of the free carrier concentration in them, and the 'working element' in this case will be the barrier layer near the second electrode, which is open under the indicated polarity in the forward direction. Therefore, in accordance with Eq. (13), the exponential dependence of current vs. voltage should be observed, i.e. the starting region of CVC of diodes with the thin chalcogenide layer must have an exponential character. Taking into account the real case, which is different from the ideal Mott's barrier, it can be expected that the starting region of CVC of such a diode will have a somewhat nonlinear varistor-type curve. When the thicknesses of the chalcogenide films are increased, the effect of tunneling through the barrier layer (switched in the reverse direction) decreases because of the recombination of induced carriers and the presence of an increasing number of barrier layers of micro-diodes, connected in reverse-series directions in the chalcogenide film itself (Oreshkin, Semenov, Zolotaterv and Mitrofanov, 1969). With the increased thickness of the chalcogenide film, the role of near-electrode barriers will decrease and the negative resistance and switching effect can be observed. Film varistors with aluminum electrodes are sufficiently durable and stable devices. When a certain voltage is reached, almost vertical CVC, i.e. the voltage stabilization, is observed. Figure 16 shows static CVC of 'thin' switches (thicknesses of chalcogenide films of 0.5 txm) with different areas of electrodes. The stabilization voltage increases with decrease of the electrode areas. Tests of such stabilizers during 500 h of continuous operation in the dynamic regime showed the following technical characteristics: Fluctuations of the stabilized voltage--not more than 3%; The stabilization voltagem5-15 V; The dynamic resistance of the stabilizer--20-100 12; The static resistancem20-50 k~; The minimum stabilized current~0.1-1 mA; The maximum stabilized current--3-15 mA; The nonlinearity coefficient--up to 1000. The parameters presented show that chalcogenide glass-based film switches with aluminum electrodes can be used as voltage stabilizers in microelectronics circuits. At the same time, such diodes can be used as nonlinear semiconductor film resistors (NLSR) with symmetric CVC, i.e. as varistors (Udalov, 1965). 2.5.
ELEMENTS OF CALIBRATION MICROELECTRONICS
The technological process of the production of measuring electronics practically always contains a calibration stage, i.e. a complex of adjustment operations aimed at achieving the accuracy of measuring transformers in electronic channels corresponding to technical requirements. A film calibration resistor as the memory element allows only a single fixation of the adjustment value of the electrical resistance, but it is possible to create on its base
Electronic Devices and Systems Based on Current Instability
15
I, mA 30 ,3 25
20
10
J
5
10
15
20
u, v
FIG. 16. StaticCVC of switchingelements. The film's thickness0.5 txm,electrode's areas--l" 25 x 2:0.9 x 10 - 3 c m 2, 3" 0.1 • 10 - 3 c m 2.
10 - 3
cm2,
the calibration set of resistors (CSR) as a single microchip. However, such a design concept has a serious disadvantage related to the extreme difficulty of adjusting resistors through outer connectors of the CSR because it requires either a width or thickness change of the resistor film. The practical application of a similar CSR microchip is indicated by those electronic device manufacturers who have the necessary technological equipment for the corresponding treatment of the film and the technology for the production of hybrid ICs for a particular application. One of the radical solutions to this problem is the application for CSR of specific microelectronics materials whose physical properties allow, to a great extent, the combination of required consuming qualities. For example, in films of CVS, a peculiar property has been revealed, which consists in an electrically programmable creation and the memorization of electrical resistance whose value can vary within very wide limits, from tens ~ to hundreds k~). We have shown (Glebov and Petrov, 2000; Glebov, Petrov and Prihod'ko, 2000a; Glebov, Petrov and Prihod'ko, 2000b) that for CVS-based resistors it is possible to obtain a family of CVC curves with different values of resistance in the low-resistivity state, depending on the programming current amplitude (Fig. 17). On increasing the amplitude of the current flowing through the working area of the sample, memorization of the intermediate values of resistances is observed that is explained by the peculiarities of structural-phase transformations in glass-crystal junctions (Glebov and Petrov, 2000).
16
A. S. Glebov
(a) I,mA
(b) I,mA
n
I
0.3
~
l]j),
I
0.8 q
50 gm d = 200 gm 1 =
1= 500 ].tm d= 10 ~tm
_\
06/ 0.4 //
/
0.2
.~ 0
,
20 40
.
-..-,-~. ' J 60 80 U,V
2
0
4
6
8 U,V
FIG. 17. Typical static CVC of memory elements. Film-planar CVS-based resistors of As4Te81Ge~5 composition with different interelectrode distances and electrode widths have been investigated. These investigations have been carried out in the regime of the current generation with internal resistance of about 109 1). Use of the current generator as a power supply unit allows the setting and holding of a certain current value in the region of the negative differential resistance which is the necessary condition for continuous change of the electrical resistance in a given interval of values. The possibility of obtaining different resistances by adjusting resistors in the open state in this or that range of values depending on the values of inter-electrode distances and widths has been shown. The different character of CVC regions, dependent on the electrode geometry, can be explained on the basis of the nonuniformity model in which the main theses are stated, for example, in Heivang (1987). Dependences of resistance of adjusting resistors in the open state vs. the value of the recording current at different inter-electrode distances and the electrode widths of 500 lxm are shown in Figure 18. The influence of the geometry of the adjustment memory elements (the electrode width 1 and the inter-electrode gap d) at different recording curren[s is shown in Figures 19 and 20.
Ropen' 103 Ohm 2O
10
0
1
2
3
Iwr, mA
FIG. 18. The dependence of the resistance of the analog memory element vs. the writing current for different inter-electrode gaps; l = 500 Ixm; dl = 200 lxm; d2 -- 100 I~m; d3 - 50 t~m.
Electronic Devices and Systems Based on Current Instability
17
Ropen, 103Ohm
d= 10/.tm A
0.5 mA
9
1
O
2 mA
15
mA
10 L
~ ~
100
200
L____.______
300
1,/am
FIG. 19. The dependence of the resistance of the analog element vs. electrode's widths at different writing currents.
Ro3en' 103 Ohm 1= 500 gm
Y
1--1mA 2
1.2 mA
3
1.5mA
4
2 mA
~ t
/
2 f
5
r ~
2.5
''/I
J
~
l ~
3 4
~..- ~ v ' f ~ ' - - - - - - - - ~
0 50 100 d, gm FIG. 20. The dependence of the resistance of the analogelement vs. the inter-electrodegap at different writing currents.
18
A. S. Glebov
These presented dependences of the adjusting resistor resistances vs. their geometry and the programmed current of the open state allow for purposefully set design dimensions, and for obtaining the required characteristics and selecting the regimes of their operation. To create adjusting resistors, it is necessary to use CVS, allowing repeated conversion of the element from the low-resistance state to the high-resistance state or any intermediate state and vice versa by the action of the electric field, as well as reliable conservation of the given state for a long time after the power supply is off. Current pulses supplied on the level of the initial threshold current can be used for programming. Preliminary investigations of this phenomenon, being carried out presently, have revealed the following collection of useful properties for the adjustment technology: (a) the re-programming of the electrical resistance, i.e. the possibility of recording, erasing and re-recording of a new (including corrected) value; (b) the possibility of the changing of TCR in a wide range by changing the composition, dispersity and technology of the CVS films production; (c) the possibility of programming the CVS-based adjustment elements through external contacts of microcircuits both at separate two-pole connection to the current programmer and multi-pole connection at adjustment of the resistor ratio and measurement bridges based on them. All the above points open a new prospective direction--adjustment microelectronics where integrated products, CVC-based adjustment sets for example, can be used directly as re-programmable semi-products in the assembly production of adjustment electronics manufacturers together with other deliverable ICs. This direction is very promising because it allows simplified interface electronics for sensors, and it also leads to further improvement of consumer properties, including microprocessor correctors. In addition, these elements can be used as memory elements of analog signals that are required in teachable control systems (Petrov, Guznov, Menshov and Lobanev, 1989).
3.
3.1.
Electronic Devices Based on SEs GENERATORS OF RELAXATION OSCILLATIONS
The mechanism of the generation of oscillation in devices with the four-layer p - n - p - n structure having an S-like CVC has been considered in Gariainov and Abezgauz (1968), where the analogy of relaxation circuits with electron valves and transistors is used. There, the advantages of S-devices having an internal positive feedback on current have been pointed out. However, the features of the operation of CVS-based SEs (Glebov, Petrov, Vihrov, Tatarinov and Litvak, 1972) presume new possibilities in the creation of generators of various types of oscillations. Diagrams of dependences of the amplitude and the period duration of generation with the constant voltage power supply unit are shown in Figure 21. As is seen from
Electronic Devices and Systems Based on Current Instability (a)
19
(b) A,V
t, I
ms
40 ~p
35
I
R = 1,0 MOhm _ C 1 -" 220 pF C 2 -" 390 pF C 3 = 1000 pF C 4 = 6800 pF experimental theoretical -
"-4
30 25
I
2
I
20 15
0
400
800
1200
U, V
0
400
800
1200
i~60; U. V
FIG. 21. The dependence of the amplitude (a) and the period of the relaxation oscillations (b) on the applied voltage.
the diagrams, the generation gets suppressed at both small (less than g t h r voltage of SE) and large power supply voltages. In the latter case, the SE remains in the open state. The generation period decreases, as the first approximation, exponentially on increase of the power supply voltage and, with considerable accuracy, is directly proportional to the value of the capacitance connected. At first, the generation amplitude increases with increase of the power supply voltage, Ups, and then it decreases. Oscillograms of the generation process with the pulse power supply voltages are shown in Figure 22. Oscillograms of the generation with the sinusoidal power supply voltage with different characters of output load are presented in Figure 23. At the voltage increase (approaching to the sinusoid's maximum), decrease of the amplitude and
FIG. 22. The oscillograms of generations on the constant current.
20
A. S. Glebov
FIG. 23. The oscillograms of generations on the square pulses. duration of the generation is observed (Fig. 23). There, the period of generation on 'edges' of the half-period remains approximately constant. It is typical that the subsequent suppression, following the first one, has a lesser amplitude than the next ones. At increase of the power supply voltage amplitude, disappearance of the generation on the sinusoid's maximums takes place. At small values of resistance and capacitance of the load, complete suppression of generation takes place. Analysis of experimental data allows one to make the statement that mechanisms of generation with alternative, constant and pulsed voltages are the same. It is sufficient to describe the generation process with the constant voltage to explain, taking into account the dynamic characteristics of SE, the features of generation with alternative and pulsed voltages. Details of the generation process can be illustrated by Figure 24, where current (upper) and voltage (lower) oscillograms of the same generation process with the constant current in the micro-second region (Fig. 24a) and in the nano-second region (Fig. 24b) are presented. As seen in Figure 24b, the pattern of the switching-on process of the SE in the generation regime does not differ qualitatively from the similar pattern of the single switching-on with pulsed voltage. As has been already indicated, the switching effect,
FIG. 24. The oscillograms of generations on the sinusoidal voltage.
Electronic Devices and Systems Based on Current Instability
21
the step-like voltage change, is related to violation of the stability condition (Gariainov and Abezgauz, 1968). If the left side of the condition is not observed, conditions for its cut-off are created after every switch-on of the device that is followed by charging of the capacitor and consequent switching-on of the SE. Thus, the switching phenomenon is the periodical switching on and off of the SE. For calculations of the characteristics of generation, the equivalent electrical circuits presented in Figure 25a,b have been used. The equivalent electrical circuit of the measurement device is presented in Figure 25a with SE in the close state (the forward direction of generation--the charging) and in Figure 25b (the reverse direction of generation--the discharging). The equivalent circuit consists of the following elements: R~, the load resistance (see Fig. 17). The resistance Ri is neglected because R i << R1, C, the total capacitance (parasitic and connected); Ra, the resistance of the SE in the close state (positive, nonlinear, voltage-dependent); R, the negative differential resistance of the SE dependent on the current flowing; La and Ra constitute the equivalent electrical circuit of the SE in the open state; Lc, the parasitic inductivity of the capacitor; Rc, the resistance limiting the discharging current; Rin and Cin, the input resistance and capacitance of the oscilloscope (or the cathode follower), respectively. The current in the SE circuit is called the generation current. The voltage on terminals 2 - 2 is the generation voltage. The power supply is applied to terminals 1-1. Let us consider the generation process in the current and voltage coordinates (Fig. 26a) and in the voltage and time coordinates (Fig. 26b). i.e.
(a) 0
1
I-
I
"
Jz
[
2
"-'in.osc
(2
Ups
U
t
A 1
1
2
l
-
~-Ld
~in osc
2 Lc
Cin.osc
c
Ups
t
-Rd
~inosc ~Rd 1
-
Ugen
Ugen
Rc 2
FIG. 25. The equivalent circuit diagram of the measuring cell (a) the forward direction of generation (the charging); (b) the reverse direction of generation (the discharging).
22
A. S. Glebov
(a) Ups --ff
(b) Ugen
\\
/~ I/\~
Ups
\\ Ugen 2 \\Igen
',\\ \
Ubum
',
Ithr Uthr Uthr+ RIthr
~--i
FIG. 26. The general presentation of the generation process. (a) In coordinates 'voltage-current" 1, CVC of the switching element; 2, the generation cycle; (b) in coordinates 'voltage-time'.
At application of the constant voltage to terminals l - 1 (Fig. 25a), the exponential charging of capacitances C and Cin.oscthrough the resistor Rn takes place that corresponds to the charging region (Fig. 26b). If the influence of the pre-threshold current is not taken into account, the charging region (Fig. 24b) is described quite well by the expression: Ugen -- Usuppl(l - exp(t/R*(C + Cin))),
where R* -- (R1Rin.osc)/(RH +
Rin.osc)
(14)
if the condition (15)
Usuppl > Uthr --[-IthrR1
is observed. Upon reaching the threshold voltage (Fig. 26a), i.e. in the negative region of the CVC, conditions are created for the discharge of capacitances C and Cin through the negative differential resistance of the SE. Thus, the duration of the forward direction of generation (Zcharge) (Fig. 26b), neglecting the pre-threshold current and the residual voltage, is determined by the expression: "/'charge ----- R* ( C + Cin)ln( Usuppl/ ( Usupp1
-
-
Uthrl)) ,
(16)
where Uthrl is the threshold voltage maintained after the first switching. The amplitude of the discharge current at discharging proved to be a finite value due to the existence of Lc, Ld, Rd, and it is additionally restricted by the resistor Rc (Fig. 25b). The maximum amplitude of the generation current can exceed 10 times the threshold current. There, the generation voltage (the momentary value) consists of the voltage on the SE, the voltage on Ld and the voltage on Rd (Fig. 25b). Duration of the reverse stage of generation (Tdischarge)is determined by values C, Lc, Rc, Ld, and Rd (Fig. 25b), and equals several p~s. In U - I coordinates, the trend of the curve is complicated and determined by the amplitude and phase ratios of the reactivities of the circuit. In the process of discharge of the capacitance C in the working region of the SE, significant heat is released. On the approaching of the operating point to the point of origin (Fig. 26a), the power (released in the SE) decreases, and as early as in the end of the second stage of generation, the high-resistance state of the SE begins to be restored.
Electronic Devices and Systems Based on Current Instability
23
With significant thermal inertia, the state of the SE can be restored, even on the initial region of the first stage of the following generation period, because the rate of the exponential voltage increase is relatively small and the operating point of the SE is still located near the point of origin. With the beginning of the voltage increase on the SE, the cycle of generation begins if the SE was not blocked by the passing current pulse and had time to restore the highresistance state. It should be noted that discharging time, at Re << R~, and restoration time at determination of ~'genare neglected in practical cases compared with charging time. The next cycle of generation begins from the value Urem- Therefore, the amplitude of generation Agen (Fig. 26b) is equal to Agen--
Uth r -- Ure m
(17)
Taking this and the pre-threshold current into account, expression (16) can be defined more accurately: "/'gen --
R~(C +
Cin.osc)ln((Usuppl - Urem)/(Usupp 1 - Ure m -- Uthr)) -q- 'r(lthr),
(18)
where T(/thr) is the additional duration originating at branching of the part of the charge current in the SE circuit due to the pre-threshold current. At Usuppl > Uthr + (IthrR1), the influence of this term in expression (18) can be neglected. Figure 20b shows, along with the experimental dependence, the theoretical dependence of the duration of generation vs. the power supply voltage. As seen from Figure 20b, at Ups close to Uthr + (IthrR1), curve ~'gengoes higher than the theoretical one as a result of increase of the fraction of the pre-threshold current of the SE in the charge current. At increasing Ups, the average current in the unit time through the SE increases, the dissipated power increases, causing heating of the working region, and the Uth r decreases. There, in accordance with Eq. (18), the experimental curve goes lower than the theoretical curve and, in accordance with Eq. (17), the amplitude of generation decreases (Fig. 20a). The decrease of the generation amplitude at small power supply voltages (Fig. 20a) takes place due to decrease of Uth r as the result of the action of the law dU/dt. It should be noted that decrease of the generation amplitude due to the heating of SE by the average current leads to increase of the generation frequency, i.e. to increase of the number of discharges in the unit time causing additional heating of the working region. Suppression of the generation at large power supply voltages is explained by the simultaneous action of two systematic factors: the increase of the average current through SE in the unit time and the increase of the generation frequency. The action of the former causes increased heating of the working region and decrease of the burning voltage. In practice, the generation amplitude was changed by use of SE with a corresponding switching voltage, the frequency--by changing of the time constant of the capacitance charging, and, in small limits, by changing of the power supply voltage. The family of dependences of generation frequencies vs. the power supply voltage at different voltages is presented in Figure 27. From the standpoint of practical application, the light thermal regime which is characterized by high stability and reliability is advisable, although at that the coefficient of the power supply voltage usage somewhat decreases. Nevertheless, the efficiency of this device is significantly higher than that of auto-generation devices with transistors.
24
A. S. Glebov F, MHz
1.0
J
0.3 0.1
2
j
..o--
0.03 3
~
4
0.01
100
200
300
U, V
FIG. 27. The experimental dependence of the generation frequency vs. the power supply voltage at different active components of the load--l' R = 1.2 MI~, 2: R = 2.4 M~), 3: R -- 4.3 MI~, 4: R -- 10.0 MI~.
The simplest auto-generation cell in the film variant using the internal positive current feedback in SE (the circuit is presented in Fig. 28) allows one to obtain saw-tooth oscillations offrequencies from fractions of Hz up to several MHz depending on the values R and C. The amplitude can be increased by the series connection of SEs. The practical generator circuit (Fig. 28) has the following parameters: R1 = 430 kf~ and C - 360 pF. At change of the power supply voltage from 120 to 180 V, the generation frequency changes from 3 x 102 to 3 x 104 Hz and the amplitude from 110 to 100 V, respectively.
3.2.
C O N T R O L CIRCUITS OF E L E C T R O - L U M I N E S C E N C E INDICATORS ( E L I )
Due to the wide use of ELI in information displays, the need appeared for simple, portable and reliable control circuits (CC). For commutation of ELI, a series of CC are used whose parameters meet the required parameters to a greater or lesser extent. As such, auto-generator circuits, pumping (parametric) circuits, and contact and contactless key circuits are used (Derkach and Korsunskiy, 1966). Recently, opto-electronic circuits attracted the attention of ELI control circuit designers (Rodin et al., 1963; Saydaris, 1966; Svechnikov, 1971). However, with the increasing number of ELI elements and expanding dimensions of matrix displays, parameters of such types of CC fail to comply with more and more strict requirements related to the dimensions of vehicle-borne and stationary apparatus. In this
O
I
I
(, J)
Ups D
Ugen
J
FIG. 28. The simplest self-generator cell with the single switching element.
Electronic Devices and Systems Based on Current Instability
25
connection, it is interesting to consider the possibilities of using chalcogenide SEs for the creation of CC ELI. CC of ELI must provide the output voltage with the excitation amplitude U and frequency 400-1200 Hz. Using higher frequencies, the aging of ELI must be taken into account. Use of a sinusoidal voltage is not necessary at all for the power supply as a pulsed voltage can be used. For example, use of the rectangular pulse with an amplitude equal to the sinusoidal amplitude provides an indicator brightness greater by 15-30%. The CC of ELI can be in two states: the regime of supplying Uexci t to the indicator (the regime of luminescence) and the regime of absence of luminescence. It is necessary that, in the latter regime, the parasitic highlighting is absent. Usually, it is given that the following condition should be fulfilled to have the parasitic illumination significantly different in the contrast from the illuminated e l e m e n t s : U g e n / U e x c i t . p a r --- 5--6. CC of ELI are usually interfaced with either special computers or distributing-transforming devices, which are interface elements of ELI-based output devices with computers. Output parameters which characterize CC are as follows: 1. Sensitivity, i.e. voltage or current of actuation; 2. The minimum 'signal to noise' ratio; 3. Minimum duration of the control signal. The CC of ELI is also characterized by dimensions, weight, the operation temperature range and power consumption. In descriptions of the operation of chalcogenide SEs in the regime of the relaxation oscillation generation, it is shown that, at usage of the capacitance of the same order of value as the ELI capacitance, the frequency and amplitude of the oscillations generated satisfy the requirements for the excitation voltage of ELI. From a practical point of view, the circuit of the self-exciting multivibrator (Fig. 29) deserves special attention. The principle of its operation is as follows: in the initial state, one SE is open (it is in the low-resistance state), the voltage on the other is increased up
O
R~
Il
R2
ELC
I E~
FIG. 29. The circuit diagram of the multivibrator with two switching elements.
26
A. S. Glebov
FIG. 30. The generation voltage oscillogram of the multivibrator.
to the switching-on voltage at the expense of the electroluminescent capacitor. At that, the current through the capacitor and the opened SE decreases exponentially. The voltage on the opened SE increases to some extent (the region of the negative resistance). After the suppression of generation, circuit reversal takes place and the cycle is repeated. The voltage oscillogram of one SE is shown in Figure 30. Experimental dependences of the period and the generation amplitude vs. the power supply voltage are presented in Figure 31. The key control circuit of ELI with one chalcogenide SE is presented in Figure 32a. To switch such an EL cell on, it is necessary to supply the resistor with the pulse of any polarity and duration of about the period of the excitation voltage of ELI. Switching off the cell is performed by switching off the power supply voltage for a time longer than the recovery time of SE. Figure 32b shows the electrical schematic diagram of the key circuit with two SE. The power supply voltage here is also selected to be less than the total voltage of the first suppression of SEs. The switching on is performed by supply of a short gated control pulse of any polarity. Either an active resistor or a resistance-inductive load (ELI) can serve as the load of the switch. The control current of the switch, as is easily seen,
A, V
T, ms
12
40
20
0
200
300
Ups
FIG. 31. Experimental dependencies of the period and the generation amplitude of the multivibrator.
Electronic Devices and Systems Based on Current Instability
(a) ~
(b) ___r-a_ I
-400
27
__~
C1
O
-
H
I I Ucomm ~ T ELC Ucontr
D2
Ups
FIo. 32. Key control circuits of ELI: (a), with one SE; (b), with two SE. is the current of the SE suppression and equals several microamperes for some variants of SE. Due to the high reproducibility of the parameters of SEs in the planar variant, the tolerance of the suppression voltage has been significantly decreased that, in its turn, allows decrease of the control voltage to 50% of the suppression voltage in severe temperature conditions and to 20% at a small range of temperature change (10-30 ~ The differential resistance of the switch in the open state is negative. Therefore, the static resistance depends on the value of the running current and equals tens or hundreds ~. It is three orders of value less than the impedance of the segment of the typical electroluminescence alphanumerical indicator in the range of operational frequencies. At the same time, the differential resistance of the closed switch is about 107 ~. The maximum operation frequency for some variants of planar film switches is higher than 5 x 104 Hz. An interesting thing is consideration of switch operation in the trigger regime. This regime of operation with constant current is performed automatically at stability condition disruption (Garianinov and Tihodeev, 1977). However, from the standpoint of practical application (reliability improvement), the trigger regime of operation on the alternative current deserves attention. It can be obtained by three ways: 1. Use of the accumulation effect in SE (the dynamical memory). 2. Use of the 'circuit' memory at SE operation with the capacitance load. 3. Use of the physical memory of switches. The above-described key circuits operate in the trigger regime either with the capacitance load or with the resistance load, if the frequency of the power supply voltage
FIG. 33. Oscillogramsof currents (above) and voltages (below) of ELI.
28
A. S. Glebov
is higher than the frequency of accumulation of the SEs used (Oreshkin, Litvak, Liamichev, Petrov, Tatarinov and Glebov, 1973). At operation with the capacitance load the (ELI element), the effect of the 'circuit' memory appears due to summing up of the voltage on the capacitance and the power supply voltage. At increase of the voltage, the number of suppressions for the period increases (Fig. 33), the phenomenon of the voltage quantization appears, which is characterized by proportionality between the voltage amplitude and the number of switching on of the SE during one period of the sinusoidal or sawtooth voltage. There, at the voltage increase, up to five and more brightness and chroma gradations that can carry additional information in displays are observed (Nosov, 1989).
3.3.
CONVERTER OF THE DECIMAL CODE INTO THE COMPUTER CODE
In computer information output devices and distribution-transformation devices, code converters (Zazulin, 1970), decoders on crystalline diodes (Derkach and Korsunskiy, 1966), on ferrites (Chistiakov), and others are widely used. The disadvantages of known devices are their complexity, large dimensions and high cost. To simplify the circuit, as well as to obtain the effect of the internal memory and to provide the possibility of operation on alternative current, the decoder (the code converter) has been developed using threshold switches based on vitreous semiconductors (Fig. 34) (Tatarinov, Petrov and Glebov, 1975). The circuit uses the nonlinear properties of switches to perform the logic operation 'or'. The decoder is used for conversion of the decimal code into the electroluminescent alphanumeric indicator (ELANI) code or, conditionally, for conversion of the binary code into the ELANI code. Electrodes of switches, contact pads, and connectors can be produced by thin film technology and photolithography methods on the body of the indicator. Timely, limitless information storage is obtained thanks to the effect of the physical memory in the CVS-based SE. Common
1
2
3
4
5
6
7
8
9
0
ELANI
-Otto FIG. 34. The circuit diagram of the decoder-converter of the decimal code into the code of the seven-segment alphanumeric indicator.
Electronic Devices and Systems Based on Current Instability
3.4.
29
ELECTROLUMINESCENCE MATRIX DISPLAYS
Designers of matrix EL displays have encountered some significant difficulties that include excessively complicated sweep circuits and a small light output for the display. In addition, the effect of the internal memory is absent and the coefficient of usage of the excitation voltage is too small in known EL matrix displays. In Derkach and Korsunskiy (1966) and Liamichev, Orlov and Pershin (1971), matrix displays are described that are a system of vertical and horizontal conducting bars, between which a layer of electro-luminophore and a layer of a nonlinear element with the varistor characteristic are placed. On supply of the excitation voltage on one of the horizontal bars and one of the vertical bars, the electroluminescent cell on the intersection of selected bars is luminous. On other cells, belonging to the bars selected, the voltage appears, achieving half of the excitation voltage of the display. As a result, highlighting of the above cells of the display is observed. To decrease the parasitic highlighting, nonlinear semiconductor resistors (NSR) are used. The operation of such EL displays is clarified in Figure 37 where stroke signs (e.g. CVC~mthe current-voltage characteristic of the varistor) refer to known devices (Derkach and Korsunskiy, 1966; Liamichev et al., 1971). As seen in Figure 37, at the voltage drop equal to the half-voltage of the generation (1/2 Ugen ) on the cell consisting of electroluminescence layers and NSR layers connected in series, the significant part of it (tJvar) drops on the varistor and, therefore, the parasitic highlighting of the 'cross' decreases. However, in this case, rather significant highlighting current ~ runs through the cell that does not allow complete elimination of the visible highlighting of the display. Looked at another way, in the regime of luminescence of the EL cell, about half of the excitation voltage (Ulex.var)can drop on the varistor resistance that decreases the coefficient of usage of the excitation voltage, i.e. the efficiency of the display. These disadvantages can be eliminated to a significant extent by connecting in series the luminophore layer and devices with the S-type current-voltage characteristic. However, this seemed impossible until the present time because semiconductor devices with similar characteristics (dinistors, tiristors, etc.) do not have the symmetrical CVC that would allow them to use alternating current, and they cannot be produced on the body of EL displays in a single technological cycle. Figure 35 presents the design of the EL matrix display in which the electroluminescent cell, consisting of the lower
9
4
5~ A
3
6
A 6
/¢
y
;;'~
x - ~:::i:. FIG. 35. The design of the electroluminescent matrix display--l: the glass substrate, 2: the transparent electrode--X-bus-bar, 3: the electro-luminophore layer, 4: the protective layer, 5: the metal electrode, 6: the vertical Y-bus-bar, 7: the dielectric layer, 8: the working gap of the threshold switch, 9: the vitreous semiconductor film.
30
A. S. Glebov
~
__l--]--- Ucomm
-,---6
1
Uexc+ Ucomm
UELC' I I-~
"2
~
,~ ~'Uexc
Uthr
Ucomm FIG. 36. The circuit diagramof the cell of the EL matrixdisplay--l: the verticalbus-bar, 2: the horizontal busbar, 3: the threshold switch, 4: the electro-luminescent condenser, 5 and 7: the load resistors, 6 and 8: the decoupling condensers.
transparent electrode (2), serving as the X-bar of the matrix display, the electroluminophore (3) and the upper metal electrode (5), are connected in series with the chalcogenide switch, one electrode of which is the upper electrode of cell (5) and the second electrode constitutes Y-bar (6) of the matrix display. The film of chalcogenide glass (9) is deposited in working gap (3). The circuit diagram of the considered matrix display is shown in Figure 36. The technological process of the display production is quite effective because it consists mainly of operations of thermal evaporation in a vacuum and various photolithography operations. The principle of operation of the considered device in the voltage and current coordinate system is explained in Figure 37a,b. On supply of the excitation voltage Uex, to the system of vertical bus-bars (1) relative to the system of horizontal bus-bars (2, Fig. 36), the switching on of the display's cell does not take place because the voltage on switch (3, Usw) is less than the threshold voltage of the switch (Fig. 37b). On supply of commutating pulses on the vertical bus-bar and the horizontal bus-bar through coupling capacitors (6, 8) (Fig. 36), the transition of the cell, located at the intersection of selected bus-bars, to the switch-on state takes place. At that, the voltage Uon on EL capacitor (4) appears, which remains after switch (3) is off until the next half-period of the excitation voltage. The stored electric energy in EL capacitor (4) causes switching on of switch (3) at the next half-period because the excitation voltage applied to capacitor (3) is summed up with voltage Uon on capacitor (4). It is seen in Figure 37b that the voltage on switch (3) achieves the value Uthr, even at the excitation voltage equal to Uexl. Switching on of switch (3) and re-charging of EL capacitor (4) take place, predisposition to switching on of the EL cell at the next half-period of the excitation voltage is created, and so on. Switching off the EL cell is performed by switching off the excitation voltage for a certain time.
31
Electronic Devices and Systems Based on Current Instability (a)
(b)
I ~U~ .on Ion
~.
I IBKN
I'BKN
L'l--
Uc.on U'var.on
-
AIoff
7"
/
CVC'
Usw ~ U ? ,
-U,qH Uav
Uav
]~'
c.off
I'hl Uav
Ihl 1/2Uexc
Uex c
1/2U'exc
U'ex c
U
t '-I
FIG. 37. The diagram of the voltages and currents distribution (a) for explanation of the parasitic highlighting elimination effect in electro-luminescent matrix displays, (b) for explanation of the internal memory effects.
On supply to the cell of the half-value of the excitation voltage, the chalcogenide switch is in the close state (the resistance is about 107 f~), therefore the voltage on the EL capacitor (Ue) and the highlighting current (Fig. 37a) is negligible. Practically all Uon voltage drops on the EL capacitors of the switch-on cell because the resistance of the chalcogenide switch in the opened state is about tens of volts. As seen from Figure 37a, the current of the parasitic highlighting of known (Derkach and Korsunskiy, 1966; Liamichev et al., 1971) devices is much higher than that of the device considered. At the same value of the voltage on the EL capacitor of the switched on cell (Uon and/-/on), the required excitation voltage in the device considered (Uex) is approximately 2 times less than /_flex. Thus, use of chalcogenide switches in an EL matrix displays effective elimination of the parasitic highlighting in EL matrix displays of the 'cross', increasing the efficiency of the voltage use to practically 100% as well as to obtaining the effect of internal memory.
4.
Memory Cells and Switches Based on Hetero-contacts 'Crystal-Glass'
Figure 38 shows the memory cell based on the contact 'crystal-glass'. The feature of the design is that the substrate is ceramics (for example, glass ceramics) (1), and asymmetry of the CVC is provided by the contact of lower bus-bar (2), produced of a thin film of polycrystalline semiconductor, with amorphous conductor (3). Metal film (4) is used as the upper bus-bar.
4
13
FIG. 38. The design of the heterocell.
32
A. S. Glebov
5
~
4~
2" FIG. 39. The design of the heterocell.
Using a vitreous semiconductor with the switching effect without 'memory', the cell works as an asymmetric switch, i.e. at reaching the voltage on bus-bars higher than U + the transition of the cell in the low-resistance state occurs, which remains only in the case when the voltage on the bus-bars is present. The advantages of such cells are simplicity of technology, increased radiation resistance, and others inherent to CVS-based devices. The other design is shown in Figure 39. The following notations are used: 1, the substrate of high-resistivity Si; 2, the lower metal bus-bars with areas of exposed semiconductor; 3, the layer of blocking dielectric; 4, the upper metal bus-bars; 5, the film of vitreous semiconductor. The lower metal bus-bars are used for reduction of the resistance of the heterostructure as a whole because high-resistivity silicon must be used to obtain the asymmetrical CVC. The blocking dielectric allows limitation of the working region of the memory element, and it also creates decoupling between the upper and the lower metal bus-bars. The upper metal bus-bars can be deposited using heating of the substrate. This greatly improves adhesion of the metal film to the dielectric. The final operation is the glass-layer deposition. Such a sequence of layer depositions ensures conservation of the initial properties of the glassy film and simplifies to a great extent the whole technological cycle of the storage-device production. There is a known design of the heterocell based on vitreous semiconductors providing control of electric parameters with the help of the field effect. To this purpose, the third control electrode is introduced in the design of the above heterocell (Fig. 40).
(a) 4
(b) 5 ~ -
~
3
4
2
6
1
FIG. 40. The design of heterocells controlled by the electrical field--l: the semiconductor substrate, 2: the lower metal bus-bar, 3: the blocking dielectric film, 4: the amorphous semiconductor film, 5: the upper metal bus-bar, 6: the control electrode.
Electronic Devices and Systems Based on Current Instability
33
R"I",
kOhm
,< 1
40 20
0
20
40
60
80
100
T, ~
FIG. 41. The temperature dependence of the resistance in the open state--l" the substrate of (111) orientation, 2" the substrate of (100) orientation. Control electrode '6' is located in the layer of the blocking dielectric above the surface of the crystalline semiconductor substrate between the film of vitreous semiconductor and the lower metal bus-bar. The value of the resistance of memory elements in the open state ( R ' I ' ) (Fig. 41) depends to a great extent on the orientation of the crystal substrate. At (100) substrate orientation, the less distinctive temperature dependence vs. resistance, lying in the range of lesser resistances, is observed. To ensure weaker dependence of the resistance in the open state vs. temperature and the resistivity of the substrate, and a resulting lower error at reading the open state, it is necessary to select substrates of (100) orientation. Figure 42 shows the dependences of the resistance in the close state vs. temperature at different substrate orientations. For substrates of (111) orientation, the influence of temperature on the memory element resistance turns out to be less distinctive than for (100) orientation substrates. It is likely related to the fact that potential barriers of different height appear at the substrate interface and the barrier height is probably lower for (111) substrate orientation than for (100) substrate orientation.
R"0",
Mohm 1.5
A\
2
0.5
0
20
40
60
80
100
T, ~ FIG. 42. The temperature dependence of the resistance in the close state--l" the substrate of <111) orientation, 2: the substrate of <100) orientation.
34
A. S. Glebov
Iwr, mA
4 2
/'"
2
,
0
20
40
60
80
twr, ms
FIG. 43. The temperature dependence of the stable writing region '1'--1: the temperature of 20 ~ 2: the temperature of 60 ~ 3: the temperature of 80 ~ Figure 43 shows the dependences of the recording current amplitude '1' vs. duration of the recording signal. As is seen from the figure, a decline of the current amplitude with recording pulse duration increase is observed in accordance with the law close to the hyperbolic one. This is explained by the fact that, at reduction of the duration of recording signals, there is a shortage of the Joule energy to cause a phase transformation of the 'glass-crystal' type. Therefore, it is necessary to increase the amplitude of the recording pulses. At certain pulse durations, the recording current practically does not change, indicating the existence of a critical energy necessary for phase transition. Investigations of the region of the stable recording '1' in different temperature regions (Fig. 43) have shown that, with temperature increase, the region of the stable recording '1' shifts in the direction of lesser amplitudes and lesser durations. It is likely explained by the fact that the energy of the phase-transition 'glass-crystal' decreases owing to the increase of the external thermal energy supplied. Figure 44 shows the dependence of the erasing current amplitude vs. the erasing pulse duration. The minimum of the amplitude is observed at certain durations of the erasing pulse. Large amplitudes of short durations are explained by the necessity to reach a certain energy for melting Ist,
mA 80
60 40
r
20
0
4
16
64
FIG. 44. The region of the stable recording of " 0 " : i r e
tst, c --
2 mA, "~rec
~ts ----
35 ms.
Electronic Devices and Systems Based on Current Instability
35
Ier, mA 80 2 60 1
1
40
20
0
4
16
64
to, gs
FIG. 45. The dependence of the amplitude of the writing current '0' vs. the duration of the erasing pulse at increasing of temperature--l: the temperature of 20 ~ 2: the temperature of 60 ~ 3" the temperature of 80 ~
of the crystal, and at long durations to ensure glass-formation in the working area material by creation of a sufficient level of overheating of the melt. The temperature increase leads to the plateau-like appearance of the stable recording area '0' (Fig. 45), which is not typical for common symmetrical sandwich structures. Such behavior can likely be explained by the fact that, at the temperature increase, decrease of the substrate resistance takes place and a large portion of energy is accumulated in the filament that leads to the decrease of the amplitude of the erasing pulse. In 'glass-crystal' structures, the space charge limited current (SCLC) mechanism of the current flow is observed. The most interesting point, from the standpoint of use of this effect in various variants of the heterocell, is the presence of the super-linear region on the CVC where, even at insignificant voltage change on the structure, significant (several orders of the value) current change takes place. A series of thin film transistors designs, using the effect of the current limited by the space charge, are known (Zernov, Elinson and Sandomirskiy, 1966). Some of them are produced in the design similar to the vacuum triode, i.e. a metal grid is placed between the anode and the cathode inside the dielectric. The metal of the grid must have a blocking contact with the dielectric, i.e. the grid can work only under a negative potential relative to the cathode. The other known design uses isolation of the grid by the additional dielectric having blocking contacts with both the working dielectric and the metal of the grid. However, this creates additional technological difficulties during manufacture of such a device. There is also another design in which the role of the grid is performed by the electrode located above the gap between the anode and the cathode (made of different materials) like the MDS transistor. The disadvantages of the known designs are that, using metals as injectors (the anode or the cathode), it is difficult to obtain good and reliable injection of carriers in the film of active material and relatively small operation currents are obtained.
36
A. S. Glebov 4
s
3
II
!
3
1 Fro. 46. The analog thin-film triode.
In Zernov et al. (1966), a design is proposed which allows effective influence on the SCLC value with the help of the field effect, i.e. a thin-field design with transistor characteristics. The proposed device works as shown in Figure 46. At a certain voltage value on electrodes (6 and 2), the transition of the CVS film (4) in the high-conductivity state takes place. A CVS composition must be used which has the switching effect with memory. In this case, the film remains in the high-conductivity state after removal of the voltage due to formation of a polycrystalline filament between the electrodes. The location of control electrode (5) under the working region of the CVS film (under the polycrystalline filament in our case) allows change of the value of the volume charge in the polycrystalline filament with the help of the potential change, i.e. change of the level of the traps population and, ultimately, change of the current flowing through the structure. If the sign of the controlling potential coincides with the type of the injected carriers, the current through the structure will decrease; if it does not coincide, it will increase. The type of injected carriers in the polycrystalline filament is determined by the type of carriers in the crystal substrate. The silicon substrate is used as the cathode (1), upper metal electrode (6) serves as the anode, the grid is control electrode (5). The proposed design thus allows one to obtain the device with transistor characteristics. Memory devices based on heterostructures Si-CVS have a simpler design and electronic control circuit when compared with existing variants of memory devices based on CVS and known semiconductor memory devices. The cross-section of the heterostructure-based memory device is shown in Figure 47. The device consists of semiconductor substrate 1, isolating dielectric layer 2, shifting electrodes 3, sensitive layer (CVS) 4, erasing electrode 5 and access electrode 6. The device operates as follows. In layer 4, due to the local electron-beam or laser impact at recording, crystal state regions 7 are created. These regions form injecting heterojunctions in the area of the contact with the substrate, providing conversion of the coded information into packets of electric charges. The charges, injected by the heterojunctions, move, due to diffusion, to the shift register at synchronous supply of access pulses and voltage on
4
5
I \1 J
2 "
7
L_...
"',ll'lX',,"l
~3
FIG. 47. The memory device based on heterostructures Si/CVS.
Electronic Devices and Systems Based on Current Instability
37
electrodes 3. After access is finished, the shift voltage is supplied to electrodes 3 and injected charges, reflecting the structural information relief of the digital or analog coding, are supplied to the external circuit of the data processing. The injection ability can be increased by creation of injecting p - n junctions in the silicon substrate. There, the polycrystalline conducting filament will serve as the contact to the required injecting p - n junction. The effectiveness of the proposed device will be mainly determined by processes of the charge-pumping into the charge-coupling packets that require additional investigation, which has not presently been fulfilled. Investigation of the effect of the threshold switching in heterostructures ' m e t a l - C V S crystal semiconductor', together with the presence of the effect of CVC asymmetry allowed the discovery of the temperature stabilization of the threshold voltage on the reverse region of CVC. The region of the temperature stabilization of the threshold voltage begins after some initial proportionality region (Fig. 48). The voltage, applied to the heterostructure on switching in the reverse direction, is distributed between the film of the crystalline semiconductor and the area of the space charge (ASC) of the crystalline substrate. Temperature changes lead to redistribution of the total voltage applied to the device. At that, increase of the temperature accompanied by the resistance reduction of the CVS layer leads to increase of the voltage portion dropped on the ASC of the crystalline semiconductor and to stabilizing of the threshold voltage. Test results of four threshold elements produced on silicon substrates (boron-doped, 7.5 ~ cm resistivity) are presented in Table II. The significant role in stabilizing the threshold voltage is played by the resistivity of the crystalline substrate. For example, the best result has been achieved on p-type silicon substrates with resistivity of 1-10 ~ cm. There, the change of the resistivity led to the change of the region of the temperature stabilization of the threshold voltage (Table II). Outside this resistivity region, the positive effect becomes reduced. At the reduced resistivity of the crystalline substrate, the silicon contact gains the properties of a metal U,V
100
f
r
9
80
6o
! 20
40
60
8O
T, ~
FIG. 48. The dependence of the threshold voltage in the reverse-biased heterostructure Al/GalzGe6.sSi6.5 AszsTeso/pSi.
38
A. S. Glebov
TABLE II TEST RESULTS OF DEVICES BASED ON SILICON OF DIFFERENT RESISTIVITY
T (~
20
30
40
50
60
70
80
90
Uthrl, V (7.5 1)cm) Uthr2 , V (7.5 1) cm) Uthr3, V (7.5 1)cm) Uthr4, V (7.5 11 cm) Uthr, V (1 l)cm) Uthr, V (10 f~ cm)
57 61 60 56 98 47
76 77 78 74 106 53
91 93 89 88 110 65
96 98 96 91 112 75
96 100 98 93 116 80
99 99 97 94 114 83
98 101 100 97 116 85
101 102 99 98 119 88
electrode and, vice versa, use of a high-resistivity substrate leads to increasing of threshold voltages, and their values are to a greater extent determined by the breakdown in the depletion region of the crystalline semiconductor. In general, the absence of the long-range order in CVS exerts insignificant influence on proper semiconductor properties like photoconductivity, the optical absorption edge, etc., because the material preserves its properties as long as its short-range order is stable (Ioffe and Regel, 1975). From the standpoint of the optical energy absorption, amorphous semiconductors, compared with crystalline ones, are characterized by diffused and more extensive edges near the fundamental absorption (Fig. 49). This circumstance leads to expansion of the range and increase, in total, of the absorbed energy. For example, the optical energy absorption in amorphous silicon is one order of value higher than the level of absorption in monocrystalline semiconductors (Physics of Hydrogenised Amorphous Silicon, 1988). A similar situation takes place in CVS. L o o k e d at differently, the large amount of defects and states in vitreous semiconductors exert significant influence on carrier transfer (Ioffe and Regel, 1975), in particular on the mobility of electrons and holes. Typical values of mobility for some vitreous semiconductors are presented in Table III. As is seen in the table, the drift
a~
cm- 1
10 3
101
2.0
2.6
3.2
hv, eV
FIG. 49. The optical absorption at room temperature--l: amorphous Si, 2: crystalline Si.
Electronic Devices and Systems B a s e d on Current Instability
39
TABLE III DRIFT MOBILITY OF SOME NON-CRYSTALLINE SEMICONDUCTORS Material
Mobility (cm2/V s)
Se AS2Se3 Se97.3Te2.7
As6.6Se87.0Te6.4 GeSe4 Bil.5Gal.5CUl.5Inl.5Snl.5Pbl.5 Sbl.sSel.sSi6.sGe6.sAs25Teso
Electrons
Holes
6 • 10 .3 3-6 • 10.3 < 10.8 9.6 • 10 -4 1.2 • 10-4 1.0 x 10-I
1 x 10-~ 1-2 • 10-1 5 X 10.5 2 • 10-6 1 x 10-1 4.0 • 10-3 3.8 x 10-2
mobility in these materials is 4 - 9 orders of value less than that in crystalline silicon. This significantly decreases the effectiveness of passive C V S - b a s e d sensors operating similarly to photo-resistors based on crystalline semiconductors. One of the main parameters that determine the properties of the " m e t a l semiconductor" contact is the height of the potential barrier at the interface of two substances (Fig. 50). The height of the barrier can determine, for example, the open-circuit voltage in photo-elements. It is necessary, however, to consider also the charge-transfer m e c h a n i s m s because they can limit the flowing current due to the greater recombination rate in the barrier and neutral layers of the semiconductor included. Neglecting the influence of the interface, the height of the barrier is directly determined by the expression: (19)
qgb = q)m -- X s / q ,
where q9m is the electron work function for the metal; As, the electron affinity energy of the semiconductor and q is the electron charge. For experimental data processing, the following expression is used practically: q)b = aq0m -t- b,
(a) Evac.
.
E vac
(20)
(b)
~ E vac
qWm
Ec EF
m
d
q.
q-
EF Ev Ev
FIG. 50. Band diagrams of metal and semiconductor: (a) before being in contact; (b) after the Schottky barrier is formed.
40
A. S. Glebov
where a and b are constants different for different substances. For example, for GaP they are equal to 0.27 and - 0 . 0 1 , respectively (Madoly, Guseihanov and Boltovskiy, 1975). In the general case, the barrier height on the contact 'metal-semiconductor' is expressed as the difference between the work functions of each material (Fig. 51) and it can be regulated either by selection of the electrode's material or by changing the position of the Fermi level in the semiconductor as the result of additional doping of the basic material. At the same time, it is widely known that the doping of CVS leads rather to its modification than to a change of properties. Therefore, only the variant of usage of different materials remains. In practice, % almost does not depend on the type of metal used in combinations with semiconductors having covalent bonds (Si, Ge and the majority of I I I - V compounds). With increase of the role of the ion bond in semiconductors, the dependence % vs. q9m becomes more clearly apparent and can even be nonlinear. The absence of the dependence of experimentally measured ~ values vs. the type of metal reflects the constancy of the Fermi level at the deposition of metal layers in common conditions that are typical for the fixed position of the Fermi level at the presence of electron states on the interface. The photoelectric efficiency of devices based on the contact ' m e t a l - C V S ' is also insignificant due to the small carrier mobility and the high resistance of the base. The devices cannot be also used as current sensors because the dark resistance would be 104-101~ ~-~and photocurrents excited by the absorption of the optical radiation have the order of value 0.1-0.2 of the dark current. From the standpoint of the creation of high-sensitive light sensors, nonideal heterojunctions, of vitreous and crystalline semiconductors, for example, are of special interest. The main advantages of heterojunctions (HJ) are realized using a wide-gap semiconductor as a 'window', which is transparent for the light absorbed in the narrowgap semiconductor (Fig. 54). Such structures promote a suitable environment for the generation of electron-hole pairs taking place directly in the region of the space charge of the structure, which increases the operating speed of the devices and eliminates recombination losses on the illuminated surface, which in turn leads to decreased losses. The best separation of carriers by the contact field of the junction takes place
Ecl
hv
~_~ ~
EF
Ec2 EF
Evl
3W
~
E
v
2
FIG. 51. Photo-e.m.f.sensorsbased on heterostructures-- 1: the optical generation, 2: the electron, 3: the hole.
Electronic Devices and Systems Based on Current Instability
41
at the absence of band ruptures (for example, AEv = 0) and at high perfection of the interface (Panasyuk and Dementiev, 1980). The advantages of HJ compared with photo-elements based on p - n homojunctions are the higher efficiency and the higher radiation resistance if the first semiconductor layer is rather thick and the semiconductor has a wide bandgap. Application of HJ is promising for the creation of two types of devices: photo-e.m.f. sensors in a given spectrum range, particularly in the IR-range (Gaugash, Kasian and Katrush, 1980; Simashkevich, 1980), and threshold light sensors. The principle of action of photo-e.m.f, sensors is based on the separation of optically excited 'electron-hole' pairs on the interface of two materials generated due to the difference of work functions of the contacting semiconductors (Fig. 51). The idea of the creation of threshold light sensors is based on two theses: 1. It is experimentally determined that the current in the reverse branch of CVC sharply increases at illumination of 'CVS-crystalline semiconductor' heterostructures (Fig. 52) (Reinhard, Arntz and Adler, 1973). 2. Heterostructures 'CVS-crystalline semiconductor' have the effect of asymmetrical threshold switching on forward and reverse branches of the CVC. At that, the threshold voltage for the forward, relative to the crystalline substrate, branch has the value of 15 V order, and for the reverse branch - 2 5 0 V. The combination of these two effects gives the possibility of creating sensitive light sensors based on 'CVS-crystalline semiconductor' heterostructures. Anderson's model is presently used for explanation of electro-physical properties of 'metal-CVC-crystalline semiconductor' heterostructures (Anderson, 1962). Figure 53 shows the energy-band diagram of the contact of two semiconductors in the absence of electrical bias. The difference in positions of the bottom of the conductance bands is denoted as/lEe, and the difference in the position of the top of the valence bandsmAEv .
I, ~tA
... - . . . . . ." . . . . . . . . . . . . . .
6 4 2
~ The illumination | intensityincrease
2" ~ 1 7 6 ~o~176 e ~ 1 7.6. . . . . . . . . . . . . . . . ~ o'.-
I
2
-2
4
6
8
U,V
-2 -4 FIG. 52. The currency-voltage characteristic of the Mo/P3Ge7SilsAs35Te4o/pSi heterostructure: the solid line--the dark current; the dotted line--the photo current.
42
A. S. Glebov
vac tqw, t Ecj
Ec ,
Egl
x2
,
Wi.
Evl qU
d'l
I
I Xl ..._ II '-'-9 i
DEv
Eg2| I
~
I I... X2 ~ - - ~ i --", ,---
V
Ev2
FIG. 53. The band diagram of the Anderson's heterojunction model at DEc > 0, DEv > 0.
In production devices, the measured currents are significantly different from the calculated ones in both quantitative and qualitative aspects. One of the reasons for this discrepancy of crystalline heterojunctions is the appearance of energy levels near the interface related to the lattice mismatch of the selected materials as well as the presence of impurities and defects introduced in the process of structure production which can be in the electrically active or neutral state. Investigation of the influence of the integral illumination of heterostructures on the general appearance of their CVC has shown that the structures have a small value of photo-e.m.f. (up to 200 mV), and the threshold characteristics do not practically depend on illumination of the structure. The sharp current increase on the reverse branch of the CVC (up to two orders of value) was observed only in the region of small bias (up to 100 V). These experimental data completely comply with investigations of voltagecapacitance characteristics. The small value of band curvature in semiconductors leads to small values of the photo-e.m.f. The influence of the photocurrent at the large reverse bias is compensated by large injection-current densities. The barrier is reduced at the forward bias that leads to the photocurrent reduction (Fig. 54). Tl~ese conclusions have been confirmed by investigations of the spectral characteristics of structures. Spectral dependencies of the photo-e.m.f, and the photocurrent are presented in Figures 55 and 56. The general appearance of spectral dependencies indicates the optical charge generation
CVS
CC
CC
CVS
J
hv hv
J FIG. 54. Photo-e.m.f. sensors based on CVS.
Electronic Devices and Systems Based on Current Instability
43
I/I o
*ii' a 9 o A1-CVS-pSi
0
A A Sn-CVS-pSi
0.6
tP o
Cu-CVS-pSi A O hcv s= 1.4 lain O ~ hcv s=2.9 gm
0.4
9 hcv s=3.6 gm
0
I
0.2
1.0
1.5
hv, eV
FIG. 55. The dependence of the photo current of the Me/CVS/pSi heterostructure vs. the light wavelength. in CVS in the vicinity of the heterojunctions with their subsequent division by the junction field (Fig. 57). Application of an additional external bias to heterostructures has shown that the bending of bands in heterostructures corresponds to the formation of depleted layers in semiconductors near the interface. At that, on application of the forward bias, the value
U/U0 ,
I
v
n
O
0.8
I (~j
9 0 A1-CVS-pSi
OA A Sn-fVS-pSi 0.6-
9
I ""
9~ J Cu-CVS-pSi O
0.4
o
0.2
O A hcvs = 1.4 gm
O__
O ~ hcvs = 2.9 gm
~p,
J 9 A hcv s = 3.6 gm
o
I
O 0 1.0
1.5
hv, eV
FIG. 56. The dependence of the photo-e.m.f, of the Me/CVS/pSi heterostructure vs. the light wavelength.
44
A. S. Glebov
I, pA
!
2
,w--2 "~3
~ J x~4
j5
x
X -1
1.2
J
./ 1.3
1.4
1.5
hv, eV
Fro. 57. The spectral dependence of the photocurrent in the Sn/CVS/pSi heterostructure at hCVS = 2.9 t x m - - l " Ubias = -- 10 m V , 2: Ubias = -- 5 m V , 3" Ubias = 0 m V , 4: Ubias = 5 m V , 5" Ubias = 10 m V .
of the peak somewhat decreases (the barrier value decrease) and, in contrast, it increases on application of the reverse bias. The obtained set of results allows the construction of the approximate energy-band structure (Fig. 58) and proposes the equivalent energy-band structure for the heterostructure 'metal-CVS-crystalline semiconductor' (Fig. 59) where the following symbols are used: Cg--the geometrical capacitance of the CVS layer; Rg--the resistance of the CVS layer; C~ and C"mcapacitances accounting for charged layers near the metal electrode and the crystalline semiconductor, respectively, which are recharged at application of the external bias through resistors R~ and R", respectively; Lg--accounts for manifestation of inductive properties by such structures; Csi and Rsi--values accounting for the capacity and the conductivity of the heterojunctions. Despite the complicated appearance, this equivalent circuit takes into account many static and dynamic properties of heterostructures 'metal-CVS-crystalline semiconductor' including infra-low frequencies. Despite small values of photo-e.m.f., heterostructures have good photosensitivity in the region of the IR-spectrum. At that, spectral characteristics can be regulated by values
Electronic Devices and Systems Based on Current Instability
45
Vacuum level Evac
Evac CVS
Si
4.5 eV
4.05 eV
Ec E C
-0.6 eV
EFn
EF
t
.I --0.56eV -0.1 eV
0.55 eV
EFp
Ev
Ev FIG. 58.
The approximate band diagram of the heterostructure in absence of the electric contact.
of bandgaps. To improve the photoelectric properties of heterostructures, it is necessary to select materials with a larger difference of work functions.
5. Re-programmed Memory Devices (RAM) In 1970, published reports (Neel, Nelson and Moor, 1970) documented the creation and production of a new type of integrated circuits, the so-called Random Access Memory (RAM) devices with electric information re-writing. The analysis of existing RAM performed in Petrov et al. (1985) indicates that CVSbased RAMs exceed all existing semiconductor-based RAMs in operating speed, reliability, the ability to store information infinitely long time at power supply switching
A
C p
pp
Cg
Rg
i
Lg
CVS
R"
[ l Csi
Rsi
Crystalline semiconductor
FIG. 59. The equivalent circuit diagram of the Me/CVS/pSi heterostructure.
46
A. S. Glebov
off, high radiation and ultraviolet resistance, and in simplicity of electric control of the storage device. A variant of the RAM is presented below which, when compared with existing samples, is distinguished by the simplicity of its technological production process, a higher integration level, and the reliability of the matrix elements. This is achieved by the production of the memory element by isoplanar technology. Electrodes of the memory elements in this case are areas of the single crystal and the conducting (polycrystalline) silicon layers separated by the narrow (1.5-2 txm) dielectric SiO2 layer propagating to the surface of the substrate. The essence of the considered variant of the matrix is explained by drawings (Fig. 60), where: Step 1: p-type substrate 1 with n-type epitaxial layer 2. Step 2: substrate 1 with etched trenches 3 for separating insulation. Step 3: substrate 1 after simultaneous deposition of dielectric 4 and the conducting layers of bus-bars. Step 4: substrate 1 after lapping out of deposited layer 5. Step 5: substrate 1 after p-type diffused areas in windows 7, etched in oxide, are obtained. Step 6: substrate 1 after etching out of contacts for active elements and deposition of vitreous semiconductor 8 with subsequent photolithography. Step 7: substrate 1 with the deposited aluminum layer. Step 8: the cross-section of the matrix in the case of solid dielectric isolation 10. Step 9: the general view from above of the ready matrix (Steps 3, 6, 7 indicated). The technology of the matrix production is described below. The preliminary cleaned wafer 1 of high-resistivity silicon single crystal, p-doped, 10 ~ cm resistivity, (100) orientation, is placed in the epitaxial reactor, and the gas etching of the lapped surface is carried out at a temperature of 1180-1190 ~ Then, epitaxial growth of n-type silicon layer 2 on substrate surface 1 is performed by letting silicon tetrachloride and hydrogen pass through the reactor. The thickness of the epitaxial layer is set such that it makes it possible to form devices in high-resistivity silicon by diffusion processes. In the next step, wafer 1 undergoes thermal oxidation in the diffusion furnace at temperature 1100-1200 ~ during 1.5-2 h. As a result, the oxide layer with thickness of about 0.8-1.9 ~m is formed on the surface. Then, windows in the oxide layer are created by usual photolithography technology (not shown in Step 2) and the etching of trenches in the epitaxial layer is carried out by the selective etching of silicon, which can be by the alkaline etchant consisting of KOH and propanol. Then the wafer is placed in the reactor where oxidation of the relief surface of the structure is carried out by letting silicon tetrachloride and water vapor pass through at a temperature of 1200 ~ After insulating oxide 4 is formed, the water flow is stopped and hydrogen is allowed to pass through the solution, decreasing the temperature of the structure to 1140-1160 ~ and deposition of heavy-doped polycrystal 5 (Step 3) with a doping level up to 1020 cm -3 on oxidized surface 4 is carried out. The thickness of the oxide layer is 4 - - 1 . 5 - 2 ~m, and the polycrystal layer 5 - - 1 5 - 2 0 ~m. After that, polycrystal layer 5 is lapped away from the structure surface to isolating oxide layer 4. Further, in accordance with the standard
Electronic Devices and Systems Based on Current Instability
2 n
1 \
P
6
n Vb
V'
47
7
P Step 1
Uvw
Step 5 J
3 ill ~
7
Step 2
5
V
"lllll
I
7
Step 6
4 in Ip
9
V L~JV~_J ~ V Step 7
Step 3 4
"ll I ~
UWLU
9
8
v r
P
Step 4
Step 8
5 4
6~
9d Step 9 FIG. 60. The technological process of the integral RAM matrix production.
microelectronics technology, vitreous semiconductor layer 8 is deposited with subsequent photolithography for creation of local areas. In the final step of the matrix production, contact pads are opened in oxide 7, masking of the surface of the semiconductor and photolithography of aluminum layer 9 is carried out (the view from above of part of the matrix is shown in Step 9). To improve electrophysical parameters and radiation resistance, the 'islands' of single crystal 2 are separated from each other and substrate 1 by solid dielectric layer 10 (Step 8). The switching on voltage of memory elements of the matrix depends on the thickness of the isolating oxide in the well and is 10-15 V in the static regime. For reliable operation of the memory element, it is first necessary to form a filament in the working area. The process of the filament formation is carried out with the help
48
A. S. Glebov
of the current generator and consists in smooth current increase through the memory element. At that, the memory element state is monitored by the oscillograph. When the memory element turns to the '1' state, the current increases to 1.5-2 A. The memory elements formed in such a way operate in the following regime:
Uav--20 V; lw'l 'm0.5 mA; z'l ' m l 0 0 ms; Ie--23 mA; "r'O'--5 ~s. In the cycling regime, such memory elements stand about 200 cycles of re-writing without failures. Variations of element parameters in the matrix do not exceed 10% and are likely related to thickness variations of the oxide layer and irregularities of the filament formation regime. Possible variants of the electronic control system of the storage device are described in Andreev. RAMs of small ( < 50 bits) information capacity can find application, for example, in code locks, telephone sets, various counters, and other electronic devices. A high level of integration is not required for such devices, so it is expedient to produce them as hybrid integral circuits (HIC). The film-planar design variant is selected for such devices due to its simplicity and its compatibility with the VLSI technology. The main characteristics of planar memory elements are determined by the properties of the vitreous semiconductor film, the distance between electrodes and their width. Based on the requirements of the developed device and results of investigations obtained, we have selected two compositions, Ga4Ge~sTe81 and Cu2In6GelsTe77, as sufficiently heat resistant and possessing the stable effect of switching with memory. The values of the gap between the electrodes and their width has been obtained by the analytical method. The required value of the threshold voltage, which, according to requirements for this device, was not higher than 60 V at 20 ~ can be obtained by changing of the electrode width at the constant inter-electrode gap (10 txm). Figure 61 shows the environmental temperature dependences of the threshold voltage of the element produced on the base of the CVS composition Ge15Te81Ga4 for several values Uthr, V
240
\
160 5
'
/3 ~ 1
80
0
40
80
T,~
FIG. 61. Calculated dependencies of the threshold voltage of the memory element vs. the environmental temperature for CVS-based memory elements of the Ga4Ge15Te8~ composition with several values of electrode's widths--l: b = 10 ~m, 2: b = 50 ~m, 3: b = 100 ~m, 4: b = 200 txm, 5: b -- 300 lxm.
Electronic Devices and Systems Based on Current Instability
49
R, Ohm 106
AN, -"aN.
105
10 4
-40
0
40
T, ~
FIG. 62. The temperature dependence of the memory element resistance in the state '0'. of the electrode width at the constant inter-electrode gap (10 ~m) and thickness of the semiconductor film (2 I~m). Because the thickness of the CVS film is not expedient to increase due to the fast decrease of the adhesion strength with the thickness increase of the film, and taking into account some overestimation of calculated values of the threshold voltage compared with experimental ones, the electrode width must be no less than 100 I~m. Further increase of the electrode width is not expedient because the resistance of the element in the highresistance state increases but the resistance decrease in the low-resistance state is insignificant (Fig. 62). This influence is especially strong at increased temperatures when the film resistance in the high-resistance state becomes insignificant (Fig. 62). Figure 65 shows the temperature dependence on the resistance of the memory element in the lowresistance state. At small writing currents, the resistance of the device is still temperature dependent, but to a significantly lesser extent than that of the resistance of the highresistance state ('0'). At high writing currents, the resistivity of the low-resistance state
R, kOhm
•
3 ,
-40
x
>
0
x
40
T, ~
FIG. 63. The temperature dependence of the memory element resistance in the state '1'
50
A. S. Glebov
R, kOhm
0
50
1oo
150
200
B,
mm
FIG. 64. The dependence of the switching resistance vs. the electrode' s width at the room temperature (20 ~
('1') is not practically temperature dependent. This is explained by increase of the conductivity of the crystalline filament due to increase of its diameter and as a result of improving the crystallization conditions. Comparing Figures 62 and 63, we can see that the difference of resistances between low-resistance and high-resistance states changes from 103 to 4 - 5 in the operation temperature range. To increase this difference, it is necessary to select the minimum possible electrode width, i.e. to decrease the 'background' current component of the CVS film to the minimum (Fig. 64). The 'background' component is caused by the presence of the vitreous semiconductor around the working filament of the memory element whose resistance is electrically connected in parallel to the filament resistance. To increase the 'background' component of the resistance, the CVS film must be deposited only on the working area of the memory element (Fig. 65). Zirconium is recommended as the electrode material that provides reliable operation of the memory element and possesses the lowest thermal conductivity among refractory metals. This ensures the most favorable conditions at information re-writing. To reduce the electrode resistance of the memory elements, the aluminum layer must be deposited on the zirconium film, and the aluminum film must be removed in the contact area with CVS.
(a)
1~ 3 FIG. 65. Variants of memory elements designs.
Electronic Devices and Systems Based on Current Instability
51
I1
.
.
.
.
t
tls
tlf t
FIG. 66. The trapezoidal shape of the writing pulse '1'. The operational ability of RAM is determined not only by its design and the properties of the materials used, but also by the parameters of the re-recording pulses. For example, the shape of pulses influences the recording stability. Our investigations have shown that the best shape of the recording pulse is a trapezoidal one (Fig. 66). The recording pulse duration influences the reliability of the recording. The longer the pulse, the more reliable the recording. For the developed memory device, a pulse of 300 ms duration and 100 ms trailing edge duration ensured stable recording '1' in the memory elements of G e - T e - G A composition in the temperature range of - 6 0 to 4-85 ~ The amplitude of the recording pulse '1' (11) influences the resistance of the memory elements in the low-resistance state (R1) and the position of the characteristic Uth r = f ( T ) (Fig. 67). Figure 68 shows the temperature dependence of the low resistance of the memory elements for different recording currents at the voltage of the power supply of the write driver of Uwd = 60 V. It is seen in Figure 68 that, at a current of I1 = 2 mA and a temperature of 100 ~ the voltage U t h r becomes higher than Uwd and the writing of '1' in the memory element is impossible. At 11 = 8 mA and Uwd = 60 V, the writing of '1' is possible at temperatures up to - 6 0 ~ (Fig. 69).
T D, g s Uthr, V
60
60
40
40
20
20
%. J
-60
-20
20
60
T, ~
FIG. 67. The temperature dependencies of the threshold voltage and the delay time.
52
A. S. Glebov
R, kOhm
2
/
3
/
4
^
-40
0
40
80
T, ~
FIG. 68. The temperature dependence of the low resistance--l" I1 -- 1 mA, 2:I1 = 2 mA, 3:I1 = 5 mA, 4: I 1 --- 8 mA, T1 = 400 mS, U d r - - 60 V.
It should be noted that these diagrams are obtained on continuous cycling of the memory elements and the temperature change from the plus region to the minus one and back. In the case when the writing of '0' was carried out at + 20 ~ and the temperature was reduced to - 60 ~ the voltage Uwd = 60 V was not sufficient to switch the memory element and write ' 1' in it. At - 60 ~ it was necessary to increase U to 9 0 - 1 0 0 V. After the first switching of the memory element and passing the 8 mA current through it, it was possible to ensure the cycling of the memory element even at the voltage of U = 60 V. In this relation, the regime of the continuous cycling at temperature change and the regime of single re-writing at different temperatures should be distinguished. The reduction of R at large recording currents '1' can be explained by the presence of polycrystals in the filament of the memory element decreasing the effective length of the filament.
Iwr,,1,, , mA
Iwr,,0,,, mA
-7 3O
-6 -5
20
% r , , , , ~
10
2
1 -60
-20
20
60
T, ~
FIG. 69. The temperature dependence of the re-writing pulse amplitude: U = 60 V.
Electronic Devices and Systems Based on Current Instability
53
II
Io, mA
40
t
I
20
0
2
4
6
N, pulses
FIG. 70. The dependence of the amplitude of the erasing pulse vs. the number pulses in the pulse burst: t~ = const; I1 = const; to = 8 I~S; t = 100 I~S; T = 20 ~
At writing '0' (erasing), the pulse parameters also influence the reliability of the re-writing. In some works of Andreev (1976)and Malchenko (1980), it has been shown that the most reliable regime of erasing is erasing by pulse burst. In Malchenko (1980), investigations on the impact of the pulse burst on memory elements have been carried out. It has been shown that erasing (the writing of '0') can be ensured by the pulse burst with a lesser amplitude than that of the single pulse. Figure 70 shows the diagram of the dependence of the erasing pulse amplitude vs. the number of pulses in the burst. The mechanism of action of the erasing pulse burst can be explained as follows. As investigations have shown, the resistance of the memory element increases gradually after the first erasing pulses and, after all the burst passed, the memory element turns to the '0' state with high resistance (Fig. 71). At further increase of the erasing amplitude, the resistance does not practically change. Evidently, a kind of self-acting choice of optimal erasing regime at the action of the erasing pulse burst takes place because the heating up of the filament to the melting temperature occurs in the result of the heat accumulation. At that, no overheating takes place and the maximum cooling rate of the filament material is ensured. We have carried out investigations on the influence of the re-writing pulse parameters on the operation ability of the memory elements in the operation temperature region in the regime of continuous cycling. Figure 67 shows the temperature dependences of the threshold voltage, Uthr, and the delay time, ~'d. The investigations have shown that, by changing re-writing pulse parameters, it is possible to ensure the re-writing in the whole range of operation temperatures. Figure 69 shows diagrams of re-writing pulse amplitude change in the operation temperature range which ensured reliable re-writing in the regime of continuous cycling. The erasing was carried out by the pulse burst of 20-25 pulses with the duration of 5 - 7 Ixs and the off-duty factor of Q > 20. It is seen from the diagram that, by changing the pulse parameters, it is possible to ensure the re-writing at any temperatures and small values of power supply voltage. It should be noted that it is necessary to increase
54
A. S. Glebov
I~ Al
R, Ohm
I
I
I
130
230
330
t, Its
106 t
104
_S 102 I 30
t, Its
FIG. 71. The influence of the erasing pulse burst on the memory element. significantly the power supply voltage for the single re-writing at low temperatures to switch the device on. This is the essential disadvantage of such m e m o r y elements. At small information capacity, it is most expedient to use the 1.5D organizational system (Andreev, 1976). However, in our case, the m e m o r y device of 8 bits capacity with the possibility of parallel information read-out from all registers is necessary. R A M with the 1.5D organization only provides sequential information read-out. To provide parallel read-out, it is necessary to use intermediate registers that make the control electronics complicated. RAM, organized in accordance with the 1D system, provides parallel information read-out by simple schematic means despite more laborious production.
References Anderson, R.L. (1962) Experiments on Ge-GaAs heterojunctions, SoL-St. Electron., 5. Andreev, V.P. (1986) Re-programmable memory devices based on vitreous semiconductors. Moscow, 135 pp. Andreev, V.P. (1976) Investigations of principles of memory devices design with electrical re-writing based on memory elements from amorphous semiconductors. Candidate's Dissertation: 01.04.10. Riazan. Aseev, B.N. (1959) Phase relations in radio engineering, Cviazizdat Publishers Moscow, p. 121. Baryshev, V.G. (1971) Candidate' s dissertation: 01.04.10. Riazan. Chistiakov, V.B. (1965) The converting device of the binary-decimalcode into the code of the electroluminescent digital indicator. Inventors Certificate USSR, No. 204705 of 27.02.65, Class 42m3,3/14. Derkach, V.P. and Korsunskiy, V.M. (1966) Electroluminescent devices, Naukova Dumka Publishers, Kiev. Gariainov, S.A. and Abezgauz, I.D. (1968) Semiconductor Devices with Negative Resistance, Energy Publishers, Moscow, Leningrad. Garianinov, S.A. and Tihodeev, Yu.S. (1977) Physics Models of Semiconductor Devices with Negative Resistance, Radio i Sviaz Publishers, Moscow, 276 pp. Gaugash, P.V., Kasian, V.A. and Katrush, P.I. (1980) Heterojunctions Between AIIIB V and AIIB VI Compounds. Photoelectric Properties of Heterojunctions, Shtinnitsa, Kishinev. Glebov, A.S. and Petrov, I.M. (2000) Physics and Application of the Current Instability in Vitreous Semiconductors, Uzoroch'e, Riazan, 256 pp.
Electronic Devices and Systems Based on Current Instability
55
Glebov, A.S., Petrov, I.M. and Prihod'ko, P.S. (2000a) Thin-film calibration resistor. Certificate on useful model, RU 15419 V1 7H01C7/00. Glebov, A.S., Petrov, I.M. and Prihod'ko, P.S. (2000b) Physical-technological problems of creation of calibration microelectronics based on vitreous semiconductors, Proc. of H Int. Conf. Amorphous and Monocrystal Semiconductors, St. Petersburg University Publishers, St. Petersburg, p. 141. Glebov, A.S., Petrov, I.M., Vihrov, S.P., Tatarinov, V.G. and Litvak, I.I. (1972) Physics-Technological Problems of Cybernetics. Kiev. Guznov, A.A., Glebov, A.S. and Petrov, I.M. (1984) Study of possibility to control parameters of CVS-based switches by light. Proc. of III All-Union Scientific-Technical Seminar Ways of Improving Stability and Reliability of Microelements and Microcircuits, Riazan, June 14-15, 1984. Guznov, A.A. and Sazhin, B.N. (1985) Study of possibility to control parameters of CVS-based switches by light. Proc. of III All-Union Scientific-Technical Seminar Ways of Improving Stability and Reliability of Microelements and Microcircuits, Riazan, June 14-15, 1984. Riazan, 1985. Heivang, V. (1987) Amorphous and Polycrystalline Semiconductors, Mir Publishers, Moscow, 147 pp. Ioffe, A.F. and Regel, A.R. (1975) Non-crystalline, amorphous and liquid semiconductors, Vol. 2, Ioffe, Selected Works Nauka Publisher, Leningrad. Karlson, D. (1988), In Physics of Hydrogenised Amorphous Silicon, Vol. II (Eds., Goannopoulos, G.E. and Lukovsky, G.) Mir Publishers, Moscow, p. 259. Liamichev, I.Ya., Orlov, I.N. and Pershin, G.G. (1971), In Collection "Microelectronics", Vol. 4 (Ed., Lukon), Soviet Radio Publishers, Moscow. Madoly, S.G., Guseihanov, M.K. and Boltovskiy, V.V. (1975) Metal contacts to gallium phosphide, Reviews on Electronic Engineering, Series 2, 10, 328. Malchenko, S.I. (1980) Investigations of active elements from vitreous semiconductors and development of electronic devices on their base. Candidate's Dissertation: 01.04.10. Riazan. Neel, R.G., Nelson, D.L. and Moor, G.E. (1970) ROM with electrical re-writing conserving information at power supply cut-off, Electronics, 43(20), 3-9. Nosov, Yu.P. (1989) Optoelectronics, 2nd Ed. Radio I Sviaz Publishers, Moscow, 360 pp. Oreshkin, P.T., Litvak, I.I., Liamichev, I. Ya. Petrov, I.M., Tatarinov, V.G. and Glebov, A.S. (1973). On possibility of application of discrete S-diodes and matrixes of planar S-diodes, based on amorphous semiconductors, in control devices of electroluminescent displays. Proc. MIEM, 29, Moscow, 120-127. Oreshkin, P.T., Semenov, V.A., Zolotaterv, V.F. and Mitrofanov, O.V. (1969) Proc. Higher Ed. USSR, Physics, 3, 85-86. Ovshinsky, S.R. (1963) Symmetrical current controlling device. Patent USA Class, 307-885, (328, 1591), September 20, 1963. Panasyuk, L.M. and Dementiev, I.V. (1980) Heterostructures in Optical Information Registration Systems. Photoelectric Properties of Heterojunctions, Shtinnitsa, Kishinev. Petrov, I.M. and Gruznov, A.A. (1988) Active thermosensors based on vitreous semiconductors. Proceedings of VI All-Union Conf. Electrical Methods and Means of Temperature Measurements (Electrothermometry). Part III, Lutsk, September 13-15, pp. 349-350. Petrov, I.M., Guznov, A.A., Menshov, O.B. and Lobanev, E.V. (1989) Analog CVS-based memory elements. CAD and automatization of production processes, Lipetsk. Petrov, I.M., Oreshkin, V.P., Glebov, A.S., Timofeev, V.I., Baryshev, V.G. and Semenov, V.A. (1971) PhysicsTechnological Problems of Cybernetics. Kiev. Petrov, I.M., Rozhkova, G.V., Rotsel, G.P., Shrainer, Yu.A., Raykin, L.G. and Gorelova, M.Yu. (1985) Active thermosensors based on vitreous semiconductors. Proceedings of lII All-Union Scientific-Technical Seminar Ways of Improving Stability and Reliability of Microelements and Microcircuits. Riazan, June 14-16, Riazan, 1985. Petrov, I.M., Sokolov, B.M., Semenov, V.A. and Balabanov, E.T. (1972), Proc. of RRTI, Riazan, Vol. 40, pp. 111-114. Reinhard, D.K., Arntz, F.O. and Adler, D. (1973) Properties of chalcogenide glass-silicon heterojunctions, Appl. Phys. Lett., 23(4). Rozhkova, G.V., Glebov, A.S. and Petrov, I.M. (1989) Possibility of forecasting parameters of active thermosensors based on chalcogenide vitreous semiconductors. Proc. of Conf. Sensors Based Microelectronic Technology. Moscow. Rodin, O.A., Mishin, V.P. and Svetov, A.B. (1963) Problems of Radioelectronics, Series IX, TV Engineering, Vol. 3.
56
A. S. Glebov
Saydaris (1966) Electronics (Russian translation), 19, 71. Simashkevich, A.V. (1980) Photosensors Based on Heterojunctions Between AIIBV~Compounds. Photoelectric Properties of Heterojunctions, Shtinnitsa, Kishinev. Stylobans. (1967) Physics of Semiconductors, Soviet Radio Publishers, Moscow. Svechnikov, S.V. (1971) Elements of Opto-electronic Engineering, Nauka Publishers, Moscow. Tatarinov, V.G. (1976) Applications of S-diodes based on vitreous semiconductors. Candidate's Dissertation: 01.04.10. Riazan, 163 pp. Tatarinov, V.G., Petrov, I.M. and Glebov, A.S. (1975) Dynamic memory of switches, based on amorphous glasses, and some aspects of their application, Opto-electronics and spectroscopyuUlianovsk, 1, 54-57. Udalov, N.P. (1965) Semiconductor Sensors, Energy Publishers, Moscow. Zazulin, S.N. (1970) The control cell of ELI. Information-reference leaflet. Zernov, V.B., Elinson, M.I. and Sandomirskiy, V.B. (Eds.) (1966) Problems of Film Electronics, Soviet Radio Publishers, Moscow.
CHAPTER
2
HETEROSTRUCTURES ON CHALCOGENIDE GLASS AND THEIR APPLICATIONS Dumitru Tsiulyanu TECHNICALUNIVERSITY,DEPARTMENTOF PHYSICS,DACIASTR. 41, KISHINAU,MD-2060, MOLDOVA
Research conducted on electrical contact phenomena in chalcogenide glassy semiconductors (ChGS) took place simultaneously with advances in the electronic processes of these materials. The starting point for the research was the fact that non-crystalline chalcogenides have a well-defined energy gap, Eg, determined values of electronic affinity X and work function q~. As in crystalline materials, the space charge region (SCR) appears near the contacts. Parameters and properties of SCR depend on the contact materials, interface condition, and composition of CHGS. However, unlike crystalline semiconductors, ChGS exhibit a large density of localized states distributed quasicontinuously in the gap, yet the Fermi level is fixed near the middle of the gap independent of the magnitude of doping. These circumstances lead to low values of depletion length and other peculiarities of SCR. Therefore it is convenient to consider contact phenomena in ChGS for three basic types of structures: metal-glass, crystalglass, and glass-glass.
1. Metal-Glass Junctions 1.1.
FORMATION
OF THE CONTACT
BARRIER
Proceeding from the general theoretical concept that in ChGS the density of localized states N(E) near Fermi level is finite with temperature, Mott (1971) supposed that N(E) exhibits a peak near the middle of the gap, where the Fermi level was fixed. Near the contacts these states can trap the charge of both signs, forming SCRs and the corresponding potential barriers. Width of the SCR can be estimated using the wellknown relation for depletion length
q2NF
(1)
where NF is the density of states near Fermi level, e the permittivity, e0 the electric constant and q the electron charge. 57
Copyright 9 2004 Elsevier Inc. All rights reserved. ISBN 0-12-752189-5 ISSN 0080-8784
58
D. Tsiulyanu
According to the numerous data from experiments on ac conductivity, ArE is found to be 1018-1019 c m -3 eV -1, and as demonstrated by Barbe (1971) the depletion length A is less than 100 ,~. Measurements of the field effect in chalcogenide semiconductors reported by Marshall and Owen (1976) have shown that these values of the screening length are typical for many materials. However, as is clear from Eq. (1), the depletion length should be independent of temperature. This fact has not been confirmed experimentally. Marshall (1978) has shown that depletion length increases with decreasing temperature, hence the transport phenomena, as field effect, are not controlled by localized states close to the bulk Fermi level. He suggested that the formation of SCR can be due to shallow trapping centers located at 0.13 eV below this level. Therefore the depletion length should be calculated with the relation
A=(ee~
e2N0 )
1/2 { exp~ E0
(2)
where No is the density of localized trapping centers at E0 = 0.13 eV. At room temperature the screening length calculated with this equation is not larger than 100 A. Because SCR in ChGS is much less than in crystalline semiconductors, it allows charge tunneling, and the metal contacts of ChGS are expected to be ohmic. Many experiments confirm this notion, in fact. These are: 9 the independency of capacity of these structures on frequency and applied voltage (Lyubin and Maidzinski, 1964), 9 the absence of asymmetry of I-U characteristics of metal-ChGS-metal structures, the weak influence of material of the metallic electrode on I-U characteristics and on the switching effect (Petrillo and Kao, 1974; Wallace, Owen and Robertson, 1978), 9 the linear dependency of electrical resistance on the thickness of chalcogenide film (Wallace et al., 1978). Nevertheless, the influence of electrodes on electrical and photoelectrical properties of ChGS has been observed, even in some early works concerning this class of materials (Kolomiets and Lyubin, 1973). In particular, it has been found that a chosen pair of electrodes determines the value and polarity of photovoltaic effect (Andreichin and Kolomiets, 1962), the shape of I - V characteristics and potential distribution (Lyubin and Maidzinski, 1964, 1968), and the character of temperature dependence of conductivity (Andriesh and Ciornii, 1975). Two basic premises were developed to explain these peculiarities: (a) SCR on metal-chalcogenide contact can be much larger than that calculated from the equations of Debye depletion length. Thus, it is assumed that Schottky barrier can be formed near contacts, similar to barriers known for crystalline semiconductors, but showing peculiarities typical for CHGS. (b) SCR is absent in chalcogenide glasses (or is very short and carriers can readily tunnel through such barriers), and the above-mentioned properties are explained
Heterostructures on Chalcogenide Glass and Their Applications
59
with diffusion of the metal atoms into chalcogenide material, followed by formation of domains with a new chemical composition, or of insulator layer at the interface. Confirmation of the first premise appeared in one of the first works devoted to investigation of contact phenomena in vitreous layers of Sb2S3 with A1, Pt, Au and SnO2 electrodes (Lyubin and Maidzinski, 1968), in which the length of SCR was about 0.5/xm. According to the findings of this work, the capacitance of these structures does not depend on frequency. The authors suggested that it is due to the existence of deep centers of sticking, which control the kinetics of establishment of the stationary value of SCR length. This suggestion has been confirmed by several experimental works (Stotzel, Kottwitz and Leimer, 1976; Shiraishi, Kurosu and Iida, 1978; Simashkievici and Shutov, 1984a,b), where the mentioned dependence has been carried out at low frequencies, when period of applied voltage variation becomes comparable to the dielectric relaxation time. The capacity of the structures depends on temperature and illumination that confirms the conception closed with the formation of Schottky barrier at the interface. At the same time, the length of the depletion region, calculated from capacity measurements, is rather different for diverse structures: GeSe/A1, 5/xm (Stotzel et al., 1976); Ge16As35TezgSzl/ Au (Wey, 1976) and SilzTe48As3oGelo/Au, 50 nm (Shiraishi et al., 1978); SbzS3/A1, 300 nm and AszS3/A1, 800 nm (Simashkievici and Shutov, 1984a,b); AszSe3/A1, 150 nm (Sersembinov, Maximova and Fedorenko, 1982). Moreover, in some cases the dependence of structure capacitance on the applied voltage, frequency, and temperature appears if there is an insulator layer on the interface only. The insulator layer on the interface can appear in structures prepared by breaking the vacuum during technological process, i.e., when the layer electrode is put in contact with atmosphere before deposition of ChGS layer. These kinds of structures will be discussed later. The formation of contact barrier at the interface of metals with chalcogenide glasses has been studied by photoemission of carriers from metal into ChGS (Iovu, Iovu and Shutov, 1978; Andriesh, Iovu, Tsiulyanu and Shutov, 1981; Tsiulyanu, 1988) as well as by investigation of current-voltage and capacitance characteristics (Andriesh, Tsiulyanu and Triduch, 1983a; Andriesh, Verlan, Gumeniuk and Malkov, 1983b; Simashkievici and Shutov, 1984a,b). Metals such as Mg, A1, Bi, Cr, Sb, Au, Ni, Te, Pt and chalcogenide materials like AszS3, AszSe3, Sb2S3, AszS3Gel.46, AszS3Ge4.0 have been used for the preparation of thin film structures. The samples for photoemission measurements usually have been prepared by thermal vacuum (10 -5 Torr) deposition of ChGS and metal layers on glassy or quartz substrates with preliminary grown layer of transparent SnO2. The preparation of the structures has been performed without disclosing the vacuum chamber during technological process. Arrangement of the photoemission experiment is shown in Figure 1 together with measured spectral characteristics. The structures have been illuminated from the side of substrates through the SnO2 layer, and the photocurrent in the external circuit was detected. Because quartz and tin dioxide are transparent for light in the energy range of interest, carriers can only be excited in the metal base electrode and therefore be identified as holes or electrons from the direction of the current in the external circuit. In most cases the direction of the photocurrent corresponds to photoemission of the holes through blocking contacts.
60
D. Tsiulyanu
1,5
(b) ..i,.
m
1.0
t
2
-2
2
3
4
0.5
,', ,',' ,_/ t, 1.6
1.8
2.0
(a)
2.2
2.4 h v
(eV)
FIG. 1. (a) Arrangement ofphotoemission experiment: 1, transparent substrate; 2, SnO2; 3, ChGS; 4, metal. (b) Spectral characteristics of metal-AszS3 structures with electrodes from Mg (1), A1 (2) and Sb (3).
Figure lb shows the typical dependencies of photocurrent normalized to the number of ,1/2 , incident photons vs. photon's energy in Fowler coordinates, i.e., tf - n v . The data are given for AszS3-based structure with electrodes from several metals. It can be seen that in the wide range of energy Fowler's dependencies are straight lines. The intercept of the extrapolated straight line with the hv-axis indicates the barrier height. The heights of contact barrier for holes %, determined from Fowler' s plot, for AszS3, AszSe3 and SbzS3based structures are given in Table I. In the same table the values of diffusion potentials calculated as q~ = ~ + Ev - Ef (Ev represents the top of the valence band and Ef the energy of the Fermi level) as well as the data for work functions q9m of metals are given. The work functions of the metals are averaged from experimental data obtained by Michaelson (1977). It is interesting that Te is the only contact showing non-blocking properties, i.e., experimentally photoemission of electrons from electrode into chalcogenide glass has been observed. In accordance with Schottky-Mott model, the barrier height ~ for m e t a l semiconductor junctions with p-type of semiconductor should linearly decrease with increasing metal work function as --- Eg + X -
qgm
(3)
where Eg represents the band gap of the chalcogenide semiconductor and X its electron affinity.
61
Heterostructures on Chalcogenide Glass and Their Applications TABLE I CONTACT BARRIER HEIGHT AND BUILT-IN POTENTIAL AT THE M E T A L - C H G S INTERFACE N
Metal
q9m (eV)
As2S 3
(eV) 1 2 3 4 5 6 7 8 9
Mg A1 Bi Cr Sb Au Ni Te Pt
3.61 4.18 4.22 4.44 4.56 4.71 4.73 4.95 5.43
1.65 1.70 1.53 1.56 1.73 1.25 1.72 0.82 .
.
As2Se 3
8b283
qb (eV)
~ (eV)
@ (eV)
~ (eV)
qb (eV)
0.70 0.75 0.58 0.61 0.78 0.28 0.77 - 0.13 .
1.30 1.25 1.10 1.20 1.18 1.60
0.49 0.44 0.29 0.39 0.37 - 0.24
1.28 1.23 1.26 1.27 1.13 1.32 1.14 1.03
0.46 0.41 0.44 0.45 0.31 0.50 0.32 0.21
.
Figure 2 (solid lines) shows the dependencies of the barrier height on metal work function for junctions based on glassy AszS3, AszSe3 and Sb2S3. The broken lines show the expected dependencies, according to the relation (3). As can be seen, in spite of essential dispersion of % experimental data, the barrier height is a less sensitive function of q9m than that given by Eq. (3). Contact barriers do not show much dependence on the nature of metal. Independence (or weak dependence) of contact barrier height on work function of the metal has been observed in many covalent semiconductor-metal junctions (Rhoderick, 1978). Such weak dependence is usually explained by the formation of thin (10-20 A)
2.0-
\ M~, " Al
-
,._1.21.4
SbNi_.Te
Bi - , ~ ; r
.
o'"~~'~~l\
,~:u
Cr
1.4
1.0
9~
"
m
Mg I
3.5
b
%
Al ,~
-
a
Sb Ni
Mg
0.6,.1I
----
Sb N i =
c
I~i \ 5 A ~ ~ P t I
I
4.5
N
I
I
5.5
q)m, eV FIG. 2. Barrier height vs. metal work function for junctions based on glassy As2S3 (a), As2Se 3 (b) and Sb2S3 (c). Broken lines show the expected dependencies, according to Eq. (3).
D. Tsiulyanu
62
insulator layer on the interface and by a major role of semiconductor-localized surface states. Because of the neutrality of the system, these states are filled up to some level q~0 (neutral level) only. This model is known as Bardeen model. Thus, metal-chalcogenide semiconductor junctions should be considered by this model. The formation of thin insulating layer at the interface of this junction that can be transparent for carriers has been observed earlier by Wallace et al. (1978) and Ema and Hayashi (1982). Moreover, as demonstrated by Tsiulyanu, Ciumacov and Grinshpun (1983a) and Tsiulyanu, Kolomeyko and Bazik (1983b), in some junctions such as A1-ChGS, the thickness of this layer can be increased by the external voltage applied. It is assumed that the formation of thin insulator layer on the interface is caused by interaction of the surface of chalcogenide semiconductor with remnant gases in vacuum chamber during technological manipulations. The surface states that pin the Fermi level near neutral q~0 can be dangling bonds of the semiconductor that own atoms, and/or states which appear as a result of interaction with absorbates (Mamontova, Kochemirovskii and Pivovarova, 1988), or with atoms of the metal electrode (Spicer, Lindau, Skeath, Su and Chye, 1980). Figure 3a shows the possible band diagram of metal-ChGS structure with surface states at the interface. The neutral level has been estimated as 1.6 + 0.08 eV for A s 2 8 3 , 1.2 + 0.06 eV for AszSe3 and 1.2 _+ 0.08 eV for Sb2S3. Energetic distances between stationary Fermi level and top of the valence band, at room temperature, were estimated in the bulk as 0.95, 0.81 and 0.62 eV, respectively. These values are significantly lower than the neutral levels where the Fermi level at the interface is pinned. Therefore, for the equilibrium (electric neutrality) to be reached, the
1.0
al i2 I
1,5
2,0
2,5
(eV) FIG. 3. (a) Possible band diagram of metal-ChGS structure with surface states at the interface. (b) Barrier height (1) and neutral level q~0 (2) at the interface A1-ChGS vs. semiconductor gap.
Heterostructures on Chalcogenide Glass and Their Applications
63
(eV) 1.9
1.8
I
1.7
1.6
l
I
I
I
0
I
2
3
I 4
UOD
FIG. 4. Barrier height vs. applied voltage for structure A1-As2S3-SnO2.
energy bands in the chalcogenide semiconductor bend down, and blocking contacts are formed. Tellurium contacts are an exceptionmthey are ohmic. The reason is still not clear. Perhaps it is due to diffusion of Te atoms, which reduced the surface charge. Proof of this explanation is supported by the dependence of barrier height at the metalChGS interface on the forbidden gap of ChGS and on the applied voltage. Figure 3b shows the barrier height at the A1-ChGS interface vs. semiconductor gap. As can be observed, in spite of some dispersion, experimental data follow the linear dependence q~o - Eg with slope 2/3. The estimated values of the neutral level q~0 are very closed with this line. The Bardeen model shows the barrier height to be approximately proportional to the square root of applied voltage. Such dependence is much more significant than the dependence obtained taking into consideration the influence of image forces in the simple Schottky model, where ~ o c U 1/4. Figure 4 shows the dependence of barrier height on applied voltage for A1-As2S3SnO2 structure. The effect of voltage is significant but there is some deviation from the dependence q% cc U1/2 predicted by the Bardeen model. Besides, the asymmetry is observed if the polarity is changed. These peculiarities can be explained by the asymmetry of chosen contacts, i.e., A1 and SnO2.
1.2.
MECHANISM OF CURRENT TRANSPORT
The mechanism of current flow in metal-ChGS junctions strongly depends on the glass composition. As demonstrated by Wallace et al. (1978), the metal-glass junctions, where thin films of AszTe3 and G e - A s - T e were used as glass, are not rectifying and the recombination/generation mechanism of transport across the junction was shown to be dominating. As far as other structures are concerned (Tsiulyanu and Triduch, 1991), including those on the bases of typical ChGS as AszS3 and AszSe3, most metals form blocking barriers,
64
D. Tsiulyanu
which makes the investigation of current transport difficult. Figure 5a shows a typical I - U characteristic of the A1-As2Se3-A1 structure. The structure was prepared without
breaking the vacuum during the technological process. The formation of thick insulating layers as well as the electrically controlled chemical reactions at the interface observed by Andriesh and Tsiulyanu (1976), Wallace et al. (1978) and Simashkievici and Shutov (1984a,b), were excluded. Similar characteristics are observed for structures with symmetrical and asymmetrical electrodes of Mg, Sb, Cr, Ni, and SnO2. The bipolar saturation of the I - U characteristics indicates the presence of two back-to-back Schottky barriers, but for the current transport studies one of the contacts must be ohmic. Internal photoemission studies have shown that for the m e t a l - C h G S structures a tellurium electrode behaves like an ohmic contact. As such, structures with only one blocking contact can be prepared if the second one is made from tellurium. Figure 5b shows the typical I - U characteristics of symmetrical T e - A s 2 S e 3 - T e and asymmetrical Te-AszSe3-A1 structures (curves 1 and 2, respectively). The structures have been prepared by thermal vacuum deposition of Te (A1), AszSe3, and A1 (Te) layers without disclosing the chamber. The measurements were made in a vacuum of 10 -4 Torr. Obviously, the first curve shows that both contacts are ohmic, and the second one indicates the presence of only one ohmic contact. Figure 5c shows the possible band diagram of the asymmetrical Te-AszSe3-A1 structure plotted on the basis of internal photoemission data. Comparing this diagram I
I
T 5- I " 4 - T "]
I
t
I/
I
-
C :~ As2Se3
AI
W
j,1'
~ ---~ - F - . ~
0
-1 -2 -3
@
"
/
I
I "-0..8
4 0
' t,.. '
+'~'+"
/
-"
-8 ,I,
-4 .4ira
2
-o,s-6,4 o
I -0,6
I -0,4
I -0,2
I O
I 0,2
I 0,4
o'4 " , o,s
c9,
0,6
=,, 0,8
U(V) FIG. 5. I - U characteristics of the structures: (a) A1-As2Se3-A1 (+to upper electrode); (b) Te-As2Se3-Te (1); Te-AszSe3-A1 (2) (+to Te); (c) possible energy band diagram of a Te-AszSe3-A1 structure constructed by using photoemission data.
Heterostructures on Chalcogenide Glass and Their Applications
65
with the above-shown I - U characteristics (Fig. 5b, curve 2), one may conclude that the I - U characteristics are conditioned by current transport through the depletion region of the S c h o t t k y - M o t t barrier at the AsaSe3-A1 interface and a neutral region of the bulk. Using the variation of slope of the lg I - U characteristics at large forward bias, the resistance of the neutral region was found to be ohmic and it changes with temperature according to: (4)
p = p exp(-E/kT)
The experimentally obtained value of the activation energy (E ~ 0.8 eV) is close to half the band gap of AszSe3. Therefore the resistance of the neutral region of ChGS limits the current transport at forward bias and low temperature. At room temperature, p = 1.6 x 1012 ~'~ cm. Figure 6a shows the forward branch of the I - U characteristics of a T e - A s 2 S e 3 - A 1 structure, constructed by allowing for the resistance of the neutral region Rs at different temperatures. It is clear that the I - U characteristics are given by: I = Io[exp{q(U-
I R s ) / n k T } - 1]
(5)
Here n is the diode ideality factor and I0 a coefficient determined by the current transport mechanism. Figure 6b shows the temperature dependence of ideality factor n and the coefficient I 0. The values were found from the forward and reverse branches of the I - U characteristics by extrapolation of the lg I vs. ( U - IRs) dependence to the region where expression (5) is true, i.e., IU - IRsl > 3 n k T / q . |
|
|
10-~ a
10-7
,
~
10 -~
,
,
,!
(4
-
I
0
0,5
,
I
2,4 2,6
2,8
I
3,0
I
I
3,7 3,4
0
-
0
-
I
1,0
1,5
O-IRs (V) FIG. 6. (a) Forward branch of I - U characteristics at different temperatures (~ 1l0 (5); 135 (6). (b) Ideality factor and lg I0 vs. temperature.
2,0
= 20 (1)" 36 (2); 60 (3); 80 (4);
66
D. Tsiulyanu
High values of the ideality factor are due to the presence of a thin insulating layer at the A1-AszSe3 interface (Fig. 5c), on which the potential drops and through which the holes must tunnel. The formation of a minimum in the n vs. 103/T dependence indicates the presence of two competing mechanisms of current transport: the first one is present mainly at low temperatures and the second one at high temperatures. The same is indicated by the temperature dependence of the pre-exponential factor I0. From Figure 6b (curve 2) it follows:
Io = a exp(-clg/kT)
(6)
Here A is a coefficient of proportionality, q~ the activation energy which corresponds to qbb ~ 1.1 eV at high (T > 60 ~ and qbb ~ 0.2 eV at low temperatures. The first value of 05b is close to a corresponding value of the Schottky-Mott barrier height at the A1-As2Se3 interface determined by studying the internal photoemission of holes from metals to ChGS (Fig. 5c). The somewhat lower value of qbb could be explained by the inaccuracy of its determination by I - U characteristics because of the non-ideality of the diode (high value of n). Therefore one can conclude that at high temperatures the current transport both in forward and reverse direction is conditioned by thermionic emission over the barrier. As far as it is known, the thermionic current can be limited by diffusion and drift processes through a depletion region close to the contact (diffusion theory), and/or by the emission process through the interface (diode mechanism). Taking into consideration that the ChGS are low-mobility semiconductors the diffusion mechanism is more probable. In this connection the pre-exponential factor is given as A = qNvl~EsS, where Es is the maximum electric field strength next to the interface, S the diode area and Nv the density of states in the valence band. As Es depends on the bias voltage the reverse current must not saturate but increase roughly as IUI~ (Rhoderick, 1978). The experimental results show (Fig. 7) an increase of the reverse current by I ~ U ~ The decreasing slope of the lg I0 vs. 103/T dependence at temperatures lower than 60 ~ (Fig. 6b, curve 2) indicates the existence and essential influence of another current transport mechanism. In this region the diode ideality factor n increases when the temperature decreases and the reverse current weakly depends on the applied voltage. Such behavior of the I - U characteristics may be conditioned by recombination/ generation of carriers in the depletion region, or by charge tunneling to vacant trap states in the ChGS (Gupta and Overstraeten, 1975). The recombination/generation mechanism has been identified by Spear, Lecomber, Kinmond and Brodsky (1976) in junctions on the basis of amorphous silicon and by Wallace et al. (1978) in junctions on the basis of S i - T e - A s - G e glasses. In this case, the ideality factor is approximately 2 and the activation energy of the pre-exponential coefficient in expression (5) is close to Eg/2 (Eg is the gap of ChGS). The values of n ~ 5 and qbt ~ 0.2 eV obtained for A1-AszSe3 structures essentially differ from the above values. It is believed that the competing mechanism of current transport in this temperature interval is the tunnel mechanism of carriers to vacant trap states, or in the opposite direction, in dependence on the applied voltage polarity. For the process to proceed continuously, it is necessary to have an additional trap mechanism emptying on which the carriers tunnel. In ChGS there are two possibilities: thermionic activation to the valence band or hopping through the trap states. The experimentally found value of the activation
H e t e r o s t r u c t u r e s on C h a l c o g e n i d e Glass a n d T h e i r A p p l i c a t i o n s
10-2
10-1
67
u - I ~ (V) 10 o
10-6
10 .7
+Dip 2,-
10-8
.
/
lO-tO
0
1
2
U-IRs ( V ) ~ FIG. 7. Reverse branch of I - U characteristics at different temperatures (~ 110 (5)" 135 (6). Short arrows show the voltage (U - IR+) -- 3nkT/q.
20 (1); 36 (2); 60 (3); 80 (4)"
energy for junctions on the basis of As2Se3 (q~t = 0.2 eV) is greatly in accordance with the latter mechanism of trap emptying described by Mott and Davis (1971). Thus, the current transport in metal-ChGS structures with blocking Schottky-Mott barrier at the interface at low temperatures and high forward bias is limited by the ohmic resistance of the neutral region of the ChGS, but under moderate temperatures and biases, by the contact barrier. The current flow in the latter case is determined by thermionic emission or by tunneling to, or from, vacant trap states of CHGS.
1.3.
JUNCTION CAPACITANCE
The significant peculiarity of metal-ChGS junctions is the independence of their capacitance on bias voltage and frequency, at high frequencies of testing signals. Decreasing the frequency up to the level when period of applied voltage variation becomes comparable to the dielectric relaxation time Tc ~ ' r = eeop (e is the permittivity and p the specific resistance of the ChGS) leads to rapid increase of the capacitance, as well as to its dependence on applied voltage and temperature.
D. Tsiulyanu
68
ol
40 C) o
w 0
I
-3
-2
o
..~_
-I
I
I
0
I
1
2
l g f { s -1) FIG. 8. Capacitance-frequency dependence of Au-ChGS-A1 structures (Simashkievici and Shutov, 1984a,b). Figure 8 shows capacitance of A u - C h G S - A 1 vs. frequency of applied voltage for glassy As2S 3 and Sb2S 3 obtained by Simashkievici and Shutov (1984a,b). The constant value of capacitance at high frequency corresponds to a geometric capacitor C = eeoS/L, where L represents the thickness of ChGS layer and S the electrode area. Increasing the capacitance at low frequency indicates modulation of the depletion regions near contacts. In such conditions the capacitance corresponds to C(U)= eeoS/(Wl + w2), where Wl and w2 represent the width at the depletion regions. As demonstrated by Andriesh, Triduch and Tsiulyanu (1988), capacitance dependence on the bias voltage for the storage of structures and/or measured in air consists of two well-distinguished maximums, situated at both sides of zero bias. Figure 9 shows the dependence of capacitance of the A1-ChGS-A1 structure on the bias voltage at room temperature. The data are given for As2Se3 at frequency 102 Hz. Curve 1 is measured in air, after the relative long (2 days) storage of the structure at normal conditions. Curve 2 shows the same dependence after the evacuation of air from cryostat, i.e., at pressure 40 35 3o
a~ 25 0
2o 151-1
-8
-4
0
3
4
8
U{V)
FIG. 9. Capacitance-voltage characteristics of A1-As2Se3-A1 structures measured in: 1, air; 2, vacuum 10 -4 Torr; 3, in air again; 4, after electrostimulated chemical transformation of the upper electrode.
Heterostructures on Chalcogenide Glass and Their Applications
69
10 - 4 Torr. Curve 3 is carried out after the filling of the cryostat with ambient air again. As is seen, the structure capacitance strongly depends on the applied voltage, being very sensitive to the prehistory of the sample and to conditions in which the experiment provides. The simplest and most reproductive measurement is the capacitance-voltage characteristic carried out in vacuum (curve 2). The capacitance of the structure in this case follows the curve with one maximum only, and this maximum is shifted from zero voltage. Such behavior indicates that low-frequency capacitance is determined by two contacts barriers which can be considered in the framework of the two back-to-back diode model developed by Van Opdorp and Kanerva (1967) and Sze, Coleman and Lova (1971). The existing frequency is sufficiently low to realize the condition that variation of the barriers capacitance is more essential than variation of their differential resistance; i.e., for both polarities the capacitance of the reverse biased diode dominates. By neglecting losses in the forward biased barrier, the standard Schottky diode plots C-2(U) can be obtained. At the same time, calculation of the built-in potential and the density of charged states in the depletion region from the capacitance measurements cannot be made because capacitances of both depletion regions are present. Built-in potential (Vd), SCR width (w) and density of charged state Nt can be estimated from capacitance measurements of structures with only one blocking contact. As was mentioned above, internal photoemission and U - I characteristics studies indicate that for the metal-ChGS structures, a tellurium electrode behaves like an ohmic contact. Figure 10 shows the dependence C - 2 - U for the structure A1-AszSe3-Te measured at room temperature in v a c u u m 10 - 4 Torr by frequency 10 -3 Hz. As is seen, C - 2 - U dependence is a straight line that means a typical Schottky-Mott behavior. The cut-off voltage Ud = 0.4 V and SCR width w = 0.25/xm, calculated from this dependence, are in good agreement with the photoemission measurements. At higher bias (U > 6 V) the capacitance no longer
12'0t
4,0
-1 0
2
4
6
8
U {V]
FIG. 10. Reciprocalcapacitance squared vs. applied bias for structures A1-As2Se3-Te.
70
D. Tsiulyanu
depends on voltage. The width of the SCR in this case is equal to 1.0/xm and becomes comparable with sample thickness. Density of charged states, estimated from the slope of C - z - U in the depletion region, N t ~ 5 • 1015 cm -3, is more than one order less than in the bulk of AszSe3. This disagreement is not a confusion because evaluation of the charge density was made with an approximation of a single level of localized states. The contact barrier in glassy semiconductors is formed by localized states quasi-continuously distributed in the forbidden gap. The charge of the depletion region is formed by the change in localized state occupation, which is controlled by the thermal detrapping rate estimated by Snell (1980) "r = 70 exp
(Ec - Ef)-1-- e(x) kT
(7)
where z0 is the time constant of the traps and e(x) a profile of the depletion region. At a given frequency to, the condition for a localized state to contribute to the capacitance is to'/" ~
1.
Therefore, the contribution to the measured capacitance gives only the traps located between stationary Fermi level and quasi-level Fermi, defined by band bending: eb(Xb) =
The states, for which emax 1.4.
> e >
- k T ln(toz0) - (Ec - Ef)
(8)
eb~ do not influence the junction capacitance.
EFFECT OF THE T H I C K INSULATOR LAYER AT THE INTERFACE
The physical properties of metal-ChGS structures can be heavily influenced by environment and irreversible phenomena on the interface. As an example, it has already been shown (Fig. 9) that the capacitance of the A1-AszSe3-A1 structure is strongly influenced by environment. Two maximas are observed on C - U characteristics (curves 1 and 3) measured in air. The maximas completely disappeared if the measuring is provided in vacuum. This means that in normal conditions the absorption of the gas molecules occurs at the interface. Water vapor is shown, by Mamontova et al. (1988), to be the most active absorbate which leads to a dramatic increase of the surface conductivity of CHGS. As surface conductivity increases, the differential resistance of the junction and dielectric relaxation time of the carriers in SCL regions decrease. The last factor gives rise to the increase in the junction capacitance. The adsorbtion processes play an important role in the formation of a thick insulator layer at the interface. As demonstrated by Andriesh, Tsiulyanu and Kolomeyko (1977a,b), these phenomena are accompanied by outlay of the metal electrode and appearance of electromotive force. Figure 11 shows the typical variation of the transmittance spectra of the A1-AszS3Ge4 structure during storage at normal conditions. The outlying of the A1 electrode (the transmittance increases) by the electrode process can be seen. As the speed of the electrode processes can be varied by external voltage (Andriesh and Tsiulyanu, 1976), the phenomenon of electrostimulated chemical transformation (ESCT) is found to depend on direction and density of current flow through interface. Hence, ESCT are controlled by electrical field strength and other external factors, such as heating
Heterostructures on Chalcogenide Glass and Their Applications
71
T% 16 12
0
0,6
0,7
0,8
0,9 1,0 ~., ~rn
FIG. 11. Variationof the transmittance spectra of the A1-As2S3Ge4structure by storage at normal conditions: 1, virgin structure; 2, after storage for 2 days.
and illumination, leading to the ChGS conductivity variation. Matter transfer with the subsequent chemical interaction explains the phenomenon. This matter transfer can be realized in a dual manner: both by electrodiffusion of metal ions inside ChGS and by drift of ions, which appear in ChGS, towards the electrode and their subsequent interaction with metal. Careful investigation of the phenomena, using X-ray photoelectron spectroscopy (XPS), carried out by Tsiulyanu et al. (1983a,b) and Andriesh, Nefeodov, Tsiulyanu, Socolov and Triduch (1990), showed the first mentioned process to be realized when the positive polarity is applied on electrode (anode effect of ESCT), and the second to be realized when negative polarity is on this electrode (cathode effect of ESCT). At the anode process of ESCT at the A1-AszS3Ge4 interface the aluminum atoms diffuse through the whole thickness of ChGS, showing almost a linear decreasing profile. Simultaneously, the intensive oxidation of aluminum atoms occurs at the interface, and as a result, an essential part of aluminum electrode is transformed into A1203. Figure 12a shows the characteristic A1 2p spectra for structures in question, carried out before and after ESCT. It is seen that if the ESCT are performed, the relative intensity of the A1 2p line for A1 bonded in A1203 essentially increases with respect to the same line for free aluminum. Moreover, oxidation of the metal takes place from the side of A1-ChGS interface and enlarges to surface. Figure 12b gives the evidence of this statement. Curve 1 shows the ratio of intensities of the A12p line of A1 bonded in A1203 (I) and free one (I0) vs. etching time for fresh-fabricated structure. Curve 2 shows the same ratio for the structure with the total dissolved electrode after the anode ESCT. The etching was performed by Ar ion beam in vacuum 10 -8 Torr with a rate of about 100 A min-1. Before ESCT (curve 1) the ratio in question has a nearly constant value due to oxidation of the thin surface layer, because of remnant oxygen's presence in the chamber. For the structure with dissolved
72
D. Tsiulyanu A1203
20
1 2
)
2
---...
10
0
5 i__=
|
i
In
._
.',
FzG. 12. Characteristicspectra of A12p (a) and ratio I/Io vs. etching time (b) carried out before (1) and after (2) ESCT. electrode (curve 2), the mentioned ratio increases sharply at the drawing near the interface. The transition of the metal to its oxide at the A1-ChGS interface can be explained by taking into consideration that the anode electrode process strongly depends on the humidity of the environment. Therefore it can essentially be amplified if water vapor is adsorbed on the ChGS surface before the electrode is grown (breaking the vacuum before metal deposition). Existence of the water vapor on the interface leads to transformation of the metal into its hydroxide (Ema and Hayashi, 1982) which decomposes by two reactions 2Al(OH)3 ---, A1203 + 3H20 2Al(OH)3 ---' 2A13+ + 6(OH)Hence, at the interface a thin A1203 layer is formed, but free A1 ions and regenerated water are produced as well. As the Fermi level differs in the metal network and at the interface, a part of ions from the electrode moves to the interface until the electrical neutrality is established. In such an equilibrium, a double charged layer at the interface is formed and electromotive force is generated. An applied external electric field moves the system out of equilibrium and initiates the electrode process. The influence of other external factors (heating, illumination, etc.) on ESCT is explained by the semiconductor properties of CHGS. These factors lead to increasing the conductivity of ChGS only. When the redistribution of the applied voltage
Heterostructures on Chalcogenide Glass and Their Applications
73
drop occurs, more voltage drops on the interface, and as a consequence, the intensity of the electrode process increases. It should be mentioned that because of the small direct interaction of the chalcogenide with metal atoms at the interface, some molecules of A12S3 (AlzSe3) can appear. The hydrolysis of these compounds through reactions A12S3 + 6H20 ~ 2Al(OH)3 + 3H2S 1" is also possible. The emission of gaseous H2S or H2Se during ESCT was observed by Corsacov and Tsukerman (1976). Thus, if the metal-ChGS structure is fabricated in broken vacuum, and/or has a quite thin electrode and is kept in humid environment, the formation of a thick insulator layer of the interface occurs. Figure 13 shows the band diagram of the A1-ChGS junction by these conditions. Even if the ESCTs are stopped, the insulator layer on the interface dramatically influences the physical properties of the structure. Wallace et al. (1978) mentioned that A1203 layer, formed because of breaking the vacuum during structure preparation only, results in series resistance which is approximately two orders higher than that of the chalcogenide layer. Hence, at low bias the current is limited by this resistance. Figure 14 shows the influence of the insulating layer, created by ESCT, on the temperature dependence of electro conductivity of the A1-AlzSe3-SnO2 structure (Tsiulyanu et al., 1983a,b). This dependence has been carried out by cooling down the short-circuited sample in vacuum to 100 K, which was then uniformly heated up at the rate of 4 K m i n -1, with the applied voltage. Curve 1 shows the ln o--103/T for pure AszSe3 (before ESCT). It is an exponential dependence with activation energy AE~ -~ 0.6 eV, which is closed with a half of the AszSe3 gap. ESCT (curves 2 and 3) diminishes the conductivity by 3 - 4 orders of magnitude. Besides, the In o-- 103 / T curves exhibit one well-distinguished peak, with a position very weakly dependent on the quantity of A1 dissolved by ESCT. Uncommon characteristics of In o-- 103/T seems to be typical for thermostimulated polarization (TSP) of metal-insulator-semiconductor
H S';H Ef -/
A1 / ,
1
SnO2
"9, \ \ " \ \ "
b
'" ",
al(OI-I)a__ FIG. 13. The band diagram of the A1-ChGS junction by ESCT.
74
D. Tsiulyanu
1D-10 10 -B
7
B I0 -II 5"
/0-9
._~.. 1 0
10-10
-12
lO-t I
~r~ i0-13 10-]4 3
4
5
IO3/T(K- )
6
FIG. 14. Temperature dependence of electroconductivity (curves 1-3) and TSD (curve 4) of the A1-A12Se3SnO2 structure: 1, before ESCT: 2,3 after different times of ESCT.
(MIS) structures. As demonstrated by Muller (1976, 1981), at low temperatures the conductivity is so low that dielectric relaxation time is much higher than measurement time. Under these conditions the application of an external electric field results in a charging of the two-layer capacitor. During heating the current peak is measured in the external circuit, which corresponds to the charge transport inside the semiconductor and the transition from two-layer to one-layer capacitor. Vice versa, thermostimulated depolarization (TSD) occurs (Fig. 14, curve 4) with the peak in the same temperature position. Both TSP and TSD peaks are observed when the conductivity of ChGS reaches the value: ~
~~ ( d2 ) -- --if/~go" t31~l1 + ~2
(9)
el, dl, e 2 and d2 refer to the relative permittivities and the thicknesses of the insulating (A1203) layer and ChGS, respectively, fl the reciprocal heating rate and k the Boltzmann' s constant. An insulating layer at the interface influences the capacitance-voltage characteristics of metal-ChGS junction. Figure 9 (curve 4) shows the capacitance-voltage characteristics of A1-AszSe3-A1 structure after the partial 'dissolution' of the upper A1 electrode by ESCT. The increase of insulator thickness at one of the interfaces results in diminishing the whole capacitance of the structure and in a shift of the maximum. This influence is due to an additional series constant (not dependent on applied voltage) capacitance Cad > Co which can be found as Cad -- C o U / ( C o - Co), where C~ is the maximal value of capacitance (Fig. 9, curve 4). The formation of an additional thick insulator layer on the interface
Heterostructures on Chalcogenide Glass and Their Applications
75
l_~ (lO -19 F -3)
1,5 2'
1,0
-6
-4
-2
0
2
4
6
8 12(V)
FIG. 15. The reciprocal capacitance squared vs. applied bias for structure A1-As2Se3-A1 measured in vacuum before (1,1 l) and after (2,2 l) ESCT.
must result in the influence of applied voltage on the contact barrier height (Cowley, 1966) and the junction capacitance should depend on bias. Figure 15 shows the reciprocal capacitance squared vs. applied bias for A1-AszSe3-A1 structure, measured in vacuum before (1,11) and after (2,2/) the insulating layer on one of the interface was grown. It is clearly seen that the position and slopes of C-2-V characteristics are changed, which is evidence for the existence of insulating layer. Finally it should be mentioned that not only electrophysical but also optical properties of the metal-ChGS structures are drastically influenced by ESCT. Due to practical applications, these phenomena will be considered further in Section 4.3.
2.
Crystal-Glass Junctions
Hybrid glassy-crystalline heterostructures are of interest from both fundamental and applied viewpoints. It is well known that there is a possibility in heterojunctions to control the stream of carriers through interface, and "the window effect" allows the photons with necessary energy to reach the junction interface of optical radiation. Hybrid heterojunctions possess an apparent technological simplicity of preparation and low cost. The possibility to combine the crystalline and glassy solid states suggests great potential for their application. The earliest work in this field, carried out by Kolomiets, Grigorovici, Croitoru and Vescan (1970), dealt with heterojunction based on glassy TezSeAszTe3 and single crystals Ge, Si, InSb. This work found for the first time that the junctions with p-type crystals show rectification of room temperature, but capacity measurements under reverse bias indicated the formation of abrupt heterojunctions (Fig. 16). Research proceeded in two basic directions: the study of threshold switching (and memory) in heterostructures based on a combination of silicon single crystal with complex glassy compounds such as Te40As35GevP3, Te39As36SilvPl.o, and the study of transport phenomena, aimed to carry out the data about contact and bulk properties of the CHGS. These structures were also studied because of the necessity to improve the parameters
76
D. Tsiulyanu
FIG. 16. I - V characteristics of a p-Ge/Te2SeAs2Te3junction at 300 K (Kolomietset al., 1970). of hybrid heterojunction-based receptors for electrophotography and vidicon TV tubes. ChGS are p-type semiconductors and cannot be efficiently doped. Both isotype and p - n hybrid heterojunctions can be easily fabricated.
2.1.
ISOTYPEHETEROJUNCTIONS
Isotype heterojunctions are attractive because the majority carriers (holes) can be injected into ChGS through interface. The investigation of injected charge and its interaction with localized states of the ChGS allows to receive information about energy spectra of these states and their influence on the charge transport processes. Single crystal p-silicon-based hybrid heterostructures seem to be the most studied. In most works (Andriesh et al., 1977a,b; Tsiulyanu, Andriesh and Kolomeyko, 1978a,b; Andriesh, Tsiulyanu and Kolomeyko, 1979; Andriesh, Iovu, Kolomeyko, Tsiulyanu and Shutov, 1980a; Andriesh, Tsiulyanu and Kolomeyko, 1980b; Andriesh, Tsiulyanu and Kolomeyko, 1985; Tsiulyanu, Kolomeyko and Stamov, 1987), heterostructures were obtained by evaporation of ChGS in vacuum onto p-Si single crystal substrates (boron doped to 1015 cm-3). The cross-sectional view of the heterostructures is shown in Figure 17a. Figure 17b shows the typical current-voltage characteristics of heterostructures with different thickness of the ChGS layer, carried out at room temperature. The forward bias is realized by application of positive polarity to silicon. Glassy AszS3 has been used as CHGS. As can be seen all I - U characteristics, in this range of applied voltage, are clearly rectifying. The increase of the ChGS layer thickness leads to a parallel shift of the forward branch in the direction of higher biases. At the same time, the reverse branch shifts closely to voltage axis. Obviously, such behavior is due to influence of the series resistance of the ChGS layer, the fact being confirmed also by Figure 18a, where the rectifying coefficient vs. ChGS layer thickness is given. Part of the applied voltage on the heterojunction drops across the series resistance (bulk of ChGS). This resistance
77
Heterostructures on Chalcogenide Glass and Their Applications
.zl
I I l: 7
-5
-
-
3 - 5
-
7
FIG. 17. The cross-sectional view (a) and I - U characteristics (b) of p-Si/As2S3 heterojunctions. The thickness of As283 consists of (~m): 1, 0.65; 2, 1.57; 3, 3.7; 4, 7.7.
reduces the voltage appearing across the intrinsic junction to Vd -- U-IR, where Vd is the cut-off voltage equal to the diffusion potential, U the total voltage applied, R the resistance in series with junction and I the junction current. The high series resistance explains the high values of cut-off voltages. Hence the classic cut-off method of determination of the diffusion potential is not available for these heterojunctions. Figure 19 presents the typical plot of the logarithm of forward current, vs. both applied voltage (a) and logarithm of applied voltage (b) for heterostructure p-Si/AszS3 at room temperature. Two straight line regions are seen on these curves. At low values of applied voltage I - U characteristics are described by the relation I = I0 exp(B U)
(10)
where I0 ~ 10 -8 A and B ~ 14 V-1. Forward current increases with increasing temperature as
I = Io exp(E/kT)
(11)
with activation energy E ~ 0.4 eV. Hence, for applied low voltages U < 0.1 V the I - U characteristics have the general form I = I 0 exp(BU) exp(E/kT)
(12)
which looks like the relation for tunneling currents in isotype heavily doped heterojunctions (Sharma and Purohit, 1974). Taking into consideration the high value of the activation energy (~ 0.4 eV), it can be assumed that carrier transport is due to direct tunneling of holes from p-Si to localized states in As283, followed by their thermal activation into valence band.
D. Tsiulyanu
78
I0
_
10 4
10 3 0
0
----~'
1,0
10 2
0
1,0
10
I
I
FIG. 18. The rectifying coefficient (a) and forward current logarithm (b) vs. ChGS layer thickness logarithm.
u~
0.1
l
I(A) 10-3 10-4
10
/,
-5
I I I I
10 .6
I I
10-7
I I I
I
Iun
O.I
ILrk
I
0.3
0.~
u(v)
0.7
0.9
FIG. 19. Forward branch of I - U characteristics: (a) semi-logarithmic scale; (b) double logarithmic scale.
Heterostructures on Chalcogenide Glass and Their Applications
79
As will be shown later from band diagram of the p-S/As2S3 heterojunction, the energy barrier of about 0.7 eV height is formed at the interface. Coefficient B in Eqs. (10) and (12) determines the probability of tunneling, but energy E ~ 0.4 eV represents the depth of localized states in As283 on which the tunneling occurs. The thermal activation of the carriers from these levels leads both to increasing or concentrating of the free carriers, and also to increasing of probability of tunneling, as a result of traps releasing. At voltages higher than Un ~ 0.1 V the forward branch of the I - U characteristic follows the power law. The forward branch of the I - U characteristics over six orders of magnitude in current and more than two orders in voltage is described by the relation
I=DU m
(13)
where D is a proportionality factor and m an exponent having values of about 4.0 for applied voltages less than some value Vk and 2.0 for larger voltage values. Such dependence may be due to the limitation of the injection current by the space charge in AszS3. A direct argument for such an assumption could be obtained from the investigation of forward current variation with the thickness of the As283 film. On the other hand, increasing the thickness of the ChGS layer leads to the respective increasing of the series resistance of the junction and to redistribution of the voltage drop. Additional voltage drops on the serial resistance and smaller voltage drops on the junction. This diminishes the injection current. This suggestion is confirmed by the above-shown I - U characteristics and the effect of ChGS thickness on rectifying coefficient (Fig. 18a). In order to support the space charge limited current (SCLC), it is necessary to increase the applied voltage and to take into account the voltage drop on the series resistance. Figure 18b shows the forward current vs. ChGS thickness by taking into consideration the series resistance. It is seen to be in good accordance with the relation I--~ d -3 characteristic for SCLC. The part of the I - U characteristics, with exponent m > 3 implies an exponential trap distribution within a certain interval in the gap of As2S3. Then the I - U characteristics are given (Lampert and Mark, 1970) as
(e) J - NvetX ~
IUI+I d2/+1
(14)
where J is the current density, Nv the effective number of states in the valence band, e the electron charge,/x the drift mobility, e the dielectric constant of AszS3, 1 = A/kT, k being the Boltzmann's constant, T the absolute temperature, U the applied voltage, and d the thickness of the As2S3 film. From this part of the I - U characteristics (Fig. 19) and Eq. (14) the parameter of the trap distribution can be calculated as A = (ml - 1)kT. In AszS3 A = 0.07 eV. The existence of the localized states exponentially distributed in the forbidden gap of the glassy As2S3 is also confirmed by hyperbolic dependence of the power sign m 1 on temperature, in the range 7 7 - 3 8 0 K (Fig. 20a). The finite position of the quasi-Fermi level Fk, which shows the end of the exponential distribution of traps, can be found from the temperature dependence of current, for the voltage corresponding to the bending point of the I - U characteristics Uk. In fact, the increase of the applied voltage leads to the increase of the injected charge. The quasi-Fermi
80
D. Tsiulyanu
in]
I0
I (mA) 1,2
1,0
~
i-
d
0,8 0,6
5
,.~/~~,...._;_~
2 ___.....,,,._ 0,4
- - o - - - - - - -
.
0,2
x'0
1o3f Z( K -~ )
FIG. 20. Power sign ml (a) and current at U > Uk (b) vs. reciprocal temperature U (V): 1, 2.1; 2, 4.2; 3, 7.2.
level moves towards the valence band edge and the traps start filling. For voltage Uk, all the exponentially distributed traps are filled. Heating leads to the thermal emission of carriers from the least deep traps to the valence band resulting in a current increase. The activation energy of Fk = 0.23 eV, obtained from the temperature dependence of current at voltage Uk shows minimum energy Fk for the captured holes to be emitted into the valence band, i.e., a finite quasi-Fermi level position. For the voltages higher than Uk, over three orders of magnitude in current, the square law of the I - U characteristics is observed. Such dependence may be due to the presence of a trap level between Fk and the top of valence band, or a transition to Child's law after filling up the exponentially distributed traps. In the first case the I - U characteristics are given by 9 U2 I = ~ p,e0 d-T (15) where 0 is the constant, which shows the ratio of free to total carrier concentration. The increase of temperature leads to thermal emission of the carriers from the traps into the band. Therefore, the exponential increase of 0 and current should be expected. In the second case 0 ~ 1 and the temperature dependence of current is governed by the temperature variation of e and/z. Figure 20b shows current vs. temperature on this part of I - V characteristics. As is seen, the current-temperature dependence is weak and has a minimum at 360 K. This implies in favor of the second possibility, i.e., in the total filling of the exponentially distributed traps. Hence, from the part of I - U characteristics with m2 = 2 (Fig. 19) and Eq. (15), at 0 ~ 1 the hole drift mobility in ChGS studied can be calculated. For AszS3 with e = 6.3e0 (e0 is the electric constant) the drift mobility of holes at room temperature is 10 -5 cm 2 V-1 s-1. Such a small value of the drift mobility is due to hopping mechanism of conductivity through shallow traps of the valence band edge. These shallow traps
Heterostructures on Chalcogenide Glass and Their Applications
81
do not influence the space charge limitation of current, but do limit the drift mobility of holes. The minimum on the temperature dependence of the current observed at the 360 K is formed as a result of double effect. It consists in the decrease of current due to the drop of the carrier mobility, as a result of phonon scattering and its increase, due to increasing of concentration of the free carriers at their activation from deep levels. Energy arrangement of the deep levels can be estimated from a part of I - 103/T characteristics (Fig. 20b) corresponding to the sharp increase of current. Calculations show the above-mentioned value of E ~ 0.4 eV. In the considered range of forward-applied voltage, there are three mechanisms of current flow: tunneling, SCLC by exponentially distributed states, and the same mechanism occurring without participation of these traps. Figure 21 shows the experimental results (Andriesh et al., 1985), which illustrate these three mechanisms of forward current flow in isotype crystal-glass heterojunctions. The results are given for several As2S3-based heterojunctions, with different thicknesses of glassy layer and the devices being fabricated in several technologic conditions. It is clearly seen that at low biases U < Un ~ 0.1 V, the temperature dependence of the forward current at different voltages is a straight line in In I - 103/T coordinates, and
I (A,) '
10
1 0 -4
% I O -G
dL
l 0 -8 v n I
2
I
I
4
I
I
6
I
I
8
I
/ T, (K -1)
I
10
I
I
12
I
I
14
FIG. 21. Temperature dependence of forward current of the p-Si/As2 $3 heterojunction. Applied voltage increases from 1st to 10th curve.
82
D. Tsiulyanu
I - U characteristics are described by Eq. (12). At voltages higher than Un, the current transport starts to be controlled by space charge in ChGS and the second straight line region appears. The slope of the second region strongly depends on the applied voltage, and I - U characteristic follows Eq. (14). If the applied voltage draws near Uk the current temperature dependence becomes weak, i.e., the relation (15) with 0 ~ 1 is respected and the above-mentioned minimum appears at 360 K. In all three considered cases of current flow, the contribution of carriers activated from the deep levels (0.4 eV in AszS3.) is observed at high temperatures. The reverse I - U characteristics (positive polarity is applied to AszS3) of the heterojunctions studied is described by typical expression for generation-recombination currents: I-
Is[exp( eU
16,
where U is the applied voltage, Is the saturation current for large reverse bias, rl ~ 1. The saturation current Is is relatively independent of applied voltage but increases with temperature according to - A e x p ( - ~-~) gE
(17)
where A is a proportionality factor, ~E the activation energy, which is independent of, applied voltage. According to Reinhard, Arnts and Adler (1973) and Peterson and Adler (1974), the activation energy ~E is independent of the level of p-Si doping. Taking these facts into account, one can believe that the reverse current is due to injection of minority carriers (electrons) from silicon into As2S3 through the barrier formed as a result of the difference in the electron affinity of the contacting materials. These carriers are trapped on the states near the interface on which the recombination occurs. Chang and Dann (1977) have given the experimental confirmation for this suggestion. They have carried out the activation energy ~E by two independent methods: both from temperature dependence of reverse saturation current and capacitance measurements. In the last case the diffusion potential VO was evaluated and then the discontinuity of the conduction band edges has been calculated. Experiments have been made on heterojunctions based on the typical ChGS materials, i.e., p-Si/As2Se3; n-Si/As2Se3; p-Si/GeSe; n-Si/GeSe; p-Si/GeTe; n-Si/GeTe. The correlation of ~E, determined by these two independent methods, was found and it confirms the above-mentioned mechanism of reverse current flow to be correct. Hence, the activation energy ~E -- 0.51 eV, determined from the plot of In Is vs. 1/T represents the difference in electron affinity of silicon and As2S3. Taking the electron affinity of silicon equal to X1 = 4.01 eV (Milnes and Feucht, 1972), the electron affinity of As2S3 is found to be )(2 = 3.5 eV. Note that the reverse current may also be due to injection of majority carriers from As2S3 into silicon. However, as it will be shown below, for a hole moving in this direction there is no barrier to limit the current. Therefore ~E in this case must correspond to the activation energy taken from the temperature dependence of conductivity; i.e., 1.05 or 0.4 eV. Such a correspondence has not been observed. The absence of this mechanism of current flow is probably due to the small concentration of majority carriers within As2S3.
Heterostructures on Chalcogenide Glass and Their Applications
83
These experimental data were used by Tsiulyanu et al. (1978a,b) to construct the possible energy band model of the heterojunction p-Si/As2S3 under the assumption that the junction is abrupt and without any states at the interface. The silicon single crystals used in this work were boron doped to 1015 cm -3, therefore the Fermi level at room temperature is situated at 61 = 0.11 eV above the valence band edge. The optical gap of arsenic trisulphide is known as 2.35 eV (Mott and Davis, 1971). The position of the Fermi level in AszS3 at room temperature is 62 --0.95 eV, and it was determined from the temperature dependence of conductivity under the assumption that the conductivity at room temperature is due to carriers hopping through localized states on the valence band edge. The temperature coefficient of the optical gap was taken into consideration as 6.7 X l0 -4 eV K -1 (Andriesh and Tsiulyanu, 1973). Using the well-known value of the electron affinity of silicon X1 = 4.01 eV and the experimentally found value for AszS3 X2 = 3.5 eV, the work functions of these materials can be determined as q~ = X + Eg - 6. Thus, 91 -- 5.01 eV and q~2 4.9 eV for silicon and As2S3, respectively. The diffusion voltage is Ud -- q~l -- q~2 = 0.11 eV. The discontinuity in conduction band edges AEc and valence band edges AEv are, respectively, =
AEc -- X1 - X2 = 0.51 eV A E v = 82 -
~
-
Ud = O.73 e V
These parameters suggest the energy band diagram of the heterojunction p-Si/As2S3 shown in Figure 22. The peculiarity of the suggested band diagram is the presence of high barriers for electrons and holes formed by the discontinuities of band edges on the interface, but the emission limited current was observed for reverse bias only. Nevertheless, there is no disagreement between suggested band model and behavior of the forward current. At low biases the tunneling of the holes from the silicon valence band to localized states in the gap of As2S3 occurs, followed by their thermal activation in the extended states. The emission mechanism of forward current becomes possible at high biases only.
-_diP--"
vacuum level
anevl 4.01ev
3.SeV
Aszs3
I
b.SleV'f--L
1235,v
Si
i_
o77,v. _iT v
FIG. 22. Possible energy band diagram of the p-Si/As2S3 heterojunction.
84
D. Tsiulyanu
As has been mentioned, the band diagram model is constructed under assumption that the junction is abrupt without any states at the interface. Many works concerning hybrid heterojunctions indicate the non-essential role of the interface states. Jivoderov (1980), Oreshkin, Jivoderov and Glebov (1980) and Glebov and Petrov (2000), all using an original method of capacitance measurements, have shown the density of surface localized states in p-Si/ChGS heterojunctions to be less than 10 -12 eV -1 cm -2. Moreover, the direct evidences of the non-essential role of the surface states are the frequency independence of the junction capacity, as shown by Busmandrud (1975) and Chang and Dann (1977), as well as the linear dependence of C -2 on applied voltage. At low biases the same characteristics are observed, as with other chalcogenide-based hybrid isotype heterojunctions, but it is not observed for oxide-based ones. Photoelectrical properties of hybrid isotype heterojunctions were studied by Tsiulyanu et al. (1978a,b, 1987) and Krupanidi, Srivastava, Srinivas, Bhattacharya and Marsingh (1983). Figure 23a shows a typical I - U characteristic of the AszS3/p-Si heterojunctions in darkness and integral light illumination from the side of the AszS3 layer. One can see that the illumination mainly influences the reverse branch of the I - U characteristics. Along with the usual parallel shift of the branch of the characteristics, an increase of the reverse current with increasing bias voltage can be observed; and the short-circuit current does not coincide with the saturation current (Fig. 23b). This peculiarity can be explained by the division of the generated charge by the contact field (considering the local states in the forbidden gap of As2S3 that cause the localization of holes, excited in AszS3) and by the existence of a high potential barrier for the electrons, generated in p-Si. The latter appears due to the difference between the energies of the electron affinity of contacting materials, which is of the order of 0.51 eV. Actually, the dark current results from the injection of minority carriers from p-Si into AszS3 through the barrier mentioned above. In spite of the barrier absence, the stream of majority carriers from AszS3 to p-Si is very much limited due to their low concentration. Both As2S3 and p-Si carriers become excited by integral light illumination of a nonbiased heterojunction, but the contact field divides mainly the carriers excited in As2S3, and the short-circuit current results from their drifting to p-Si. Minority carriers excited in p-Si stick to the bottom of the potential well due to the presence of the barrier, located near the interface. At a reverse bias, the bottom of the p-Si conduction band rises and a direct injection of excited electrons in ChGS becomes possible (Fig. 23c). Consequently the current increases sharply. Figure 23a shows another peculiarity of the I - U characteristicsncrossing of the dark and light characteristics at forward bias. This peculiarity has been observed earlier by Gill and Bube (1970) in crystal heterojunctions, and it can be explained by a decrease of the ballast resistance of the ChGS layer, connected in series with the heterojunction. The I - U characteristics of the illuminated heterostructure can be described by the relation
I-
[
Is exp
rlkT
1]-,sc
(18)
Heterostructures on Chalcogenide Glass and Their Applications
I (IRA)
-4
-3
.
-2
~21
-I
!,
85
1
F
_
. m
!
u(v) 2
'--
31 -
-50
4
-I00
__.
,2 '
"
"
-
-
0,06
1
2f
-0,6 4
5
/
6
b
[-10'
FIG. 23. I - U characteristics of a p-Si/As2S3 heterojunction at various degrees of illumination: (a) (1) 0, (2) 540, (3) 4.2 X 103, (4) 7 x 103 lux; (b) (1) 0, (2) 540, (3) 1.4 x 103, (4) 2 x 103, (5) 4.2 x 103, (6) 7 x 103 lux; (c) heterojunction diagram with reverse bias.
where Is is the dark saturation current, R the resistance of the ChGS layer and Isc the short-circuit current. The short-circuit current increases linearly with the growth of the illumination level (Fig. 24), and it makes fractions of microampere at illumination of 103 lux. Such low values of the short-circuit current are the result of high resistance of the As2S3 layer and the decrease of the lifetime of charge carriers, generated near the interface because of their being trapped in As2S3. The open-circuit voltage also grows with the illumination level (Fig. 24). However, at illumination of 3 X 103 lux it reaches saturation at the level of 0.06 V. The open-circuit voltage saturation is less than the build-up potential Ud ~" 0.11 eV. This can be explained by the high density of localized states in the gap of
D. Tsiulyanu
86
15.=.(108A)
Vo.,.{102V)
,301
~ _ , . - - - - - 0 ~
o - | 6,0
j-
2o 101 /
t40
_--"'-
1,0
2,0
3,0
t2,0
4,0
5,0
6,0
7,0
FIG. 24. Short-circuit current lsc and open-circuit voltage Uoc dependencies on bias level. As2S3, which leads to the appearance of a strong non-radiation channel of recombination.
The fill factor of the I - U characteristics is about 40%. Isotype p-Si/ChGS heterojunctions are therefore not of great value from the point of view of energy photoconversion because of their small efficiency. More interesting is the photodiode regime of operation that will be considered further.
2.2.
GLASS/N-CRYSTAL HETEROJUNCTIONS
The first investigations of the junctions between glassy semiconductors (Te2SeAs2Te) and n-type single crystals (Ge, Si, InSb) carried out by Kolomiets et al. (1970) showed no significant rectification as the junction resistance was practically equal to the resistance of glassy materials. The same results has been obtained for others n-crystal/glass heterojunctions, such as n-Si/Te39As36SlvGeTP1 (Adler, 1973) and n-Si/AszSe3 (Alonso et al., 1976). There are several models, which explain absence of significant rectification in these devices. From the polarity of photoelectomotive force of n-Si/Te39As36S17GeTPl heterojunction, Peterson and Adler (1974) have concluded that at the silicon-ChGS interface, the silicon bands bend down and the chalcogenide bands bend up. The forward current initially is limited by the accumulation of the electrons on the Si side due to barrier on the interface. However, when the field in the Si is increased, after an offset voltage, the accumulation region stops growing and the conductivity becomes limited by glass itself. The reverse current in this kind of heterostructures is due to electron injection from the ChGS into Si being limited only by the small barrier at the interface. Increasing both Si doping concentration and reverse bias lowers the height of this barrier. Consequently, for heavy doped n-Si based heterostructures and/or high reverse bias, a significant increase of reverse current occurs. In these conditions, no real reverse current saturation occurs and I - U characteristics become almost symmetric. Brodsky and Dohler (1975) proposed another explanation of non-rectifying behavior of the n-crystal/glass heterojunctions. The proposed model is based on the fact that there is a high density of localized states in the ChGS near stationary Fermi level and
Heterostructures on Chalcogenide Glass and Their Applications
87
the hopping mechanism of carrier transport can occur. Hence, the surface of the chalcogenide semiconductor on the interface may be considered like a metal surface of Schottky barrier. The I - U characteristics in this case is described by relation
{[(
j ( U ) -- Aef exp(--/3Vb)exp /3
nd 2e )nd]iJ2
2 ~ 0 + Nf
Nff
• [exp(/3eU) - 1]
(19)
where U is the applied voltage, Aef the effective Richardson's constant, e the electronic charge, Vb the contact barrier height, nd the donor concentration in crystalline semiconductor, Nf the density of localized states near Fermi level in glassy semiconductor, ~ o the build-in up potential. It follows from Eq. (19) that at e U > k T the reverse current is strongly influenced by the ratio n d / N f . Increase of this ratio remotes reverse current from the saturation current of an ideal rectifier, but at its high value the rectifying becomes no significant. On the other hand, the rectifying behavior was observed in n-Si/ChGS heterojunctions based on glassy A s - S - G e (Andriesh, Tsiulyanu and Kolomeyko, 1977a,b), G e - S e - T e (Dann and Mackenzie, 1976; Tonge, Minami and Tanaka, 1979), As2Te3 (Krupanidi et al., 1983), A s - T e - G e (Persin and Mitra, 1980) and As2S3 (Korkinova and Andreichin, 1982). Figure 25 shows the current-voltage characteristics of n - S i / G e - S e - T e heterojunctions (Tonge et al., 1979) in dark and under illumination with integral light. ,,,..% cq
o 2,0 ~
p
~ 1,0 0 -0,2 L
,..I..-
l -1,0,
0
,1
.
u(v)
-0,2 ~ FIG. 25. I-U characteristics of n-Si/Ge-Se-Te heterojunctions: 1--in dark; 2,3--under illumination400 and 800 lx, respectively (after Tonge et al., 1979).
D. Tsiulyanu
88
Others mentioned n-Si/ChGS exhibit similar behavior. The relation like Eq. (16) describes I - U characteristics at low bias. At high forward bias a power-low behavior of I - U characteristics like Eq. (13) is observed. The exponent "m" varies for different heterostructures in limits m - 1 . 1 - 1 0 (Table II). Considering I - U characteristics, the main mechanism of current flow in n-crystal/glass junctions are emission-recombination at low voltages and space-charge-limited carrier transport at high forward biases. The capacitance-voltage measurements (Chang and Dann, 1977) shows the C - 2 - U dependence for hybrid heterostructures n-Si/ChGS to be linear, i.e., they behave like abrupt heterojunctions and the effect of the surface states on the interface can be neglected. The build-in up potential for n - S i / G e - S e - T e evaluated by Dann and Mackenzie (1976) and Tonge et al. (1979) from C - 2 - U characteristics vary from 0.2 to 0.58 V in dependence on ChGS composition. Calculation of concentration of the donor impurities in n-Si from C - z - U characteristics showed the values by 4 orders higher than the real value. Therefore, it was concluded that the space charge layer is located not only in the crystal but also in glass. It should be mentioned that capacitance-voltage characteristics of n-Si/AszTe3 heterostructure (Krupanidi et al., 1983) do not confirm the formation of an abrupt heterojunction on the interface. It looks more like the C - V characteristics of a conventional MIS structure, and is explained qualitatively by assuming a low breakdown voltage for the heterojunction in question. Photoelectrical properties of glass/n-crystal heterojunctions are very much influenced by properties of the sampled chalcogenide semiconductors and it is difficult to make a generalization of these properties. The spectral distribution of open-circuit voltage of the n - S i / G e - S e - T e shows two well distinguished peaks due to light absorption in n-Si and ChGS, that confirms the location of space charge inside both components (Tonge et al., 1979). Photoresponse of the n - S i / A s - T e - G e heterostructure exhibits two peaks, a positive peak at 0.84 ~m, and a negative peak near wavelength of about 0.7/xm. As it is assumed by Persin and Mitra (1980) this is due to formation, near the interface, of an SCR which induces an opposite electric field. The short-circuit photocurrent in n-Si/AszSe5 heterojunctions exhibits one visible and two infrared peaks (Alonso et al., 1976) and is explained by different optical transitions inside the components and at the interface. Table II summarizes some photovoltage parameters of n-Si/ChGS heterojunctions.
TABLE II SOME PROPERTIES OF GLASS/N-CRYSTALHETEROJUNCTIONS
Heterostructure
m
n-Si/As2Te3
2
n-Si/Ge2oSeso
4.1-9.1
n-Si/Ge2oTe3oSso
1.0-7.9
n-Si/Ge2oTe3oS20
1.1-3.9 depends on bias n-Si/As16Te83Gel.o -
q~(eV)
Isc (/zA)
0.58
0.03-0.045 (S is not indicated) 3.4, S - 1 cm2, E -- 800 lux 60, S -- 1 cm2, E = 800 lux 20, S -- 1 cm2, E = 800 lux 0.318, S = 1 cm2
0.24
Uoc (V)
Reference
0.15-0.22 Krupanidi et al. (1983) 0.32
Tonge et al. (1979)
0.22
Tonge et al. (1979)
0.09
Tonge et al. (1979)
0.115
Persin and Mitra (1980)
Heterostructures on Chalcogenide Glass and Their Applications
89
The hybrid crystal-glass heterostructures based on photoconductive AHB VI compounds have been studied because of their prospective vidicon camera tube application (Section 4.2). The most studied are thin film structures CdS/Sb2S3 (Kolomiets, Lyubin, Maidzinski, Plisova and Fedorova, 1971) and CdSe/As2Se3 (Cimpl, Jedelicka, Kosek and Schauer, 1979; Schauer, Heza, Jedlicka, Kosek and Cimpl, 1980). A key factor for obtaining rectifying characteristics with small reverse dark current has the thermal airbaking of CdSe layer before ChGS deposition. During this procedure the recrystallization of CdSe occurs and the surface of the layer is enriched with oxygen. The oxidation process of CdSe results in creation of new defects and the consequent redistribution of the surface states at the interface. The rectification factor increases by about 106 at U -- 10 V. The reverse dark current is a linear function of the applied voltage and can be described by the relation
Ir - AecenNeNsv e x p ( Es - Ee
(20)
where e is the electronic charge, o~n the effective cross-section for electronics, Ne the effective density of states for conduction band, Ns the interface concentration of states located at the energy Es, Ec the conduction band edge, and v the thermal velocity of electrons corresponding to the temperature T. Equation (20) describes the reverse branch of I - U characteristics if the current flow is due to thermal generation of the electrons from the surface states at the interface and their emission in CdSe. This mechanism of current flow has been experimentally confirmed by Schauer et al. (1980) and Zajic, Kosek, Cimpl, Schauer and Stourac (1991). They have studied the influence of technology of preparation (the thermal air-baking, pressure of residual gases in vacuum chamber, etc.) on the reverse dark current and have shown that increase of air-baking time results in decreasing of reverse dark current. This is due to both increase of the dominating energy level position AE -- Es - Ec and depletion region thickness of the CdSe layer. The junction capacitance decreases, which additionally leads to improving parameters of the vidicon camera target.
8/11o/phoII8
polyerystal
'w I
nW0
Wcl W ! i i
9
k.
W~(
i
I
i i I t
I i I
!
tW I f I I I I I
Xl X X0 X2 FIG. 26. The possible band model of the n-crystal/glass heterojunction CdSe/As2Se3 with artificially built-in electronic states at the (after Zajic et al., 1991).
D. Tsiulyanu
90
Zajic et al. (1991) proposed the possible band model of the n-crystal/glass heterojunction CdSe/As2Se3 with artificially built-in electronic states at the interface (Fig. 26). As a matter of fact this is a Schottky diode with the density of charge at the interface Q s s - eNs. The forward current of these heterojunctions exhibits typical dependence of SCLCs, probably due to the extracting of holes from AszSe3 by recombination process via localized states. 3.
Glass- Glass Junctions
Thin layer heterostructures based on two chalcogenide semiconductors compounds have attracted a great deal of attention due to real and prospective vidicon camera tube, electrophotography and photothermoplastic device application. It has been shown by Kolomiets et al. (1971), Dementev, Panasiuk and Tarabuhina, (1974) and Buzdugan, Zelenina, Ivashcenko, Iovu, Simashkievici and Shutov (1983), that these heterostructures allow not only to enlarge the spectral region of sensitivity but also to increase the effective resistance and to govern the response time of the :system. The spectral response consists of two peaks, in general agreement with light absorption and forbidden optical gap of the heterostructure components. Figure 27 shows the responsivity of the A s z S 3 - A s z S e 3 heterostructure at different biases and thicknesses (a)
4
I
3 2
'.2 0 1,5 1,7 1,9 2,1 2 , 3 2 , 5 2 , 7
hv(eV}
(b)
3' 2 1
01,5
1,71,92,1
2,3
2,5 2,7
hv(ev) FIO. 27. Responsivity of the SnO2-As2S3-As2Se3-A1 heterostructure at bias: a--10 V; b--80 V (Dementev et al., 1974). The thickness of the As2S3 and As2Se3 layers are 1.1 and 7.35/xm, respectively. Polarity: 1-SnO2" 4-"; 2-A1 " 4- ".
Heterostructures on Chalcogenide Glass and Their Applications
91
of the components (Dementev et al., 1974). In all cases two peaks are observed. The amplitudes of peaks are clearly dependent on the polarity and value of applied voltage, thickness of the components layers and the side (AszS3 or AszSe3) of the light incidence. These peculiarities indicate that heterostructure characteristics are explained mainly by bulk properties of the semiconductor layer components. The contribution of each layer in the total photoresponse depends on the share of the voltage drop on it. As I - U characteristics of the component layers are non-linear and different, it is clear that distribution of the applied voltage between these layers depends on the value of this voltage, particularly at high bias. This is one reason why the most general properties of glass-glass heterostructures appear to depend on the electrical and photoelectrical properties of component layers and their thickness. At the same time, the careful investigation of these heterostructures at low biases, shows their characteristics depend not only on bulk properties of the semiconductor components but also on the injection, accumulation and/or capture of the carriers on the interface. These factors result in increase of the dark resistance and light sensitivity as well as in governing the photoresponse time and in the realization of the phenomenon of injection sensibilization of the structure, demonstrated by Buzdugan, Zelenina, Iovu, Simashkievici and Shutov (1985). It should be mentioned that due to very complex character of the m e t a l - C h G S / C h G S metal heterostructures, there is no reliable model for describing the physical processes on the interface to evaluate the contribution of the bulk and contacts to heterostructure properties.
4. Applications of Heterostructures on Chalcogenide Glass Heterostructures on chalcogenide glass are of interest not only from the viewpoint of basic science but also from the applied viewpoint because of their great potential for applications. Threshold switches, memory units and hybrid glassy crystalline transistors have already been mentioned in Chapter 1. Solid-state photodetectors, targets of vidicon camera tubes, receptors for electrophotography and photolithography, as well as other media for optical information registration will be considered here. 4.1.
PHOTODETECTORS
Device structure and electrical characteristics of same crystal/glass diodes have been considered in Section 2. Although they are not attractive from energy photoconversion viewpoint, these heterojunctions are quite interesting in photodiode regime of operation. The spectral distribution of the photoresponse of the reversibly biased heterojunction depends on the bias level. Figure 28 shows this dependence for heterojunction p-Si/AszS3. One can see the photoresponse in the region, corresponding to p-Si light absorption, grows with increase of the bias level. This agrees well with I - U and C - U characteristics and proves that in the investigated heterojunction the contact field is mainly distributed in p-Si. Photoconductivity vs. excitation intensity ( L x - A characteristics) of the reverse biased diodes at A = 0.44 ~ m is shown in Figure 29. It is seen that the current increases linearly with excitation level. Such dependence can give evidence that all the generated carriers flow away through the junction in a time less than the carrier lifetime, i.e., all the generated carriers are collected and give a contribution to the current measured
92
D. Tsiulyanu
10 -7 10 -8 10-9~
~
-
~i0-Io 1 0 - I ? --
I0-14_/,
I
0 1,0 1,5
1 2,0
5 ~ 2,5 3,0 3,5
E(eV) --FIG. 28. Spectral distribution of photocurrent with respect to the reverse bias value: (1) 0, (2) 1, (3) 3, (4) 9 V. in the external circuit. In this approach, the quantum efficiency of the heterojunction can be evaluated from the expression
If fl = eN(1
-
R) -ad
(21)
where If is the photocurrent, N the light intensity, a and R are the absorption and reflection coefficients of the upper aluminum electrode for the used wavelength, respectively, d is its thickness. The calculation according to Eq. (21) gives for the quantum efficiency the value fl = 0, 3. It should be noted that ChGS-layer thickness and its composition could regulate the spectral region of the photoresponse also. For example, the ultraviolet photoresponse limit shifts toward lower photon energies by 1.0 eV introducing about 44 at.% Ge into the As2S3 melts. The photoresponse kinetics is characterized by times of the order of 1 ms and almost does not depend on the spectral composition of exciting light. 4.2.
VIDICON CAMERA TUBES AND ELECTROPHOTOGRAPHIC PLATES
One of the first applications of heterostructures based on ChGS were targets of TV camera tubes. The principle of the conversion of the optical image into electrical signal in these cameras consists of: 1. transformation of image carried by photons into charge carriers accumulated to the points of light incidence,
Heterostructures on Chalcogenide Glass and Their Applications
93
5 4 3 2 1 0
5 10 15 N (1013 ph/s era2)
20
FIG. 29. L x - A characteristics of p-Si/As2S3 for a wavelength 0.44/zm.
2. neutralization of accumulated positive charge (carrying information) by the electron beam, 3. acquisition of the videosignal and its processing. A target scanned by a focused electron beam carries out the first two steps of operation. The most primitive selenium-based vidicon target represents the heterostructure between n-type SnO2 and a-Se. In order to avoid the secondary electron emission, the porous Sb2 $3 layer is deposited on top of the a-Si. This layer acts also as a blocking layer for electrons from the scanning beam impinging into the conduction band of the a-Se. Inserting of the hole-blocking layer between the SnO2 and a-Se (for instance CeO2) leads to more complete saturation of the photocurrent; i.e., improves the target parameters. Similar heterostructure have been developed by using a-Se and many n-type materials such as CdSe, CdS, SnO2, In203, etc. Although amorphous selenium is considered to be very suitable for TV camera tube, it has two major drawbacks: crystallization of a-Se and insufficient red sensitivity. As demonstrated by Maruyama (1982), the addition of arsenic to a-Se increases the crystallization temperature, but the addition of tellurium enhances photosensitivity in the long wavelength region. However, these additives increase carrier traps in a-Se and lower the effective cartier mobility by several orders of the magnitude. The blocking characteristics of the heterostructure are destroyed and the stable operation of device is disturbed. In order to avoid the destroying of blocking characteristics Maruyama and Hirai (1983) suggested locating the addition of Te into a-Se. They have proposed a type of photosensitive heterostructure with graded composition, which became the basic structure of the photoconductive target in the Saticon, widely utilized as a TV camera tube for broadcast, industrial and home use.
94
D. Tsiulyanu
Two photodiode structures with graded composition have be,en proposed by these authors to improve the sensitivity and stability of Saticon targets: a graded-gap structure and a builtin field effect structure (Fig. 30). In both cases, the Te atoms are concentrated near the transparent electrode (where incident light is absorbed) and far from blocking contact. Hence, the red sensitivity is enhanced and blocking characterJtstics are not destroyed. In the second type of structure the tellurium-enriched region is followed by A s - T e region enriched in arsenic. Since arsenic forms deep electron traps in a-Se, a negative space charge is accumulated in this region, during the operation of the target. Thus, a built-in high electric field is established between this SCR and the hole blocking contact. This built-in field results in effective extraction of photogenerated carrier from the Te added region. Besides, arsenic region prevents the ther3~al diffusion of Te in a-Se. Figure 31 shows the target of Saticon TV pickup tube based on built-in-field effect structure. This structure is also used as xerographic plates for diode-laser printers. Besides, a number of other blocking photodiodes are proposed for vidicon targets and electrophotographic plates. Amorphous selenium in these heterostructures is substituted by other chalcogenide glasses such as S e - A s - T e (Maruyama, 1982), or As2 Se3 (Cimpl et al., 1979; Zajic et al., 1991). Polycrystalline CdSe and CdS are used as n-type component of heterojunction. Arrangement of the vidicon camera tube with CdSe-As2 Se3 heterostructure target is shown in Figure 32. As demonstrated by Schauer et al. (1980), target based on this heterojunction possesses all the parameter and properties required for usage in image camera tubes. These parameters are: 1. 2. 3. 4.
small dark current (I ~ 10 -10 A cm 2) at room temperature; capacitance 2900 pF cm -2 at 2 kHz; high quantum yield of photogeneration in visible region of spectra; high specific resistance (p > 1012 ~-~ c m ) of the photoconductive layer and its small (< 1) coefficient at secondary emission.
(a) t00j
-
-i
o,2 0,3
3,9
4
=
.2
..
11
#
11-:. . . . . . . . . .
03 0,2 0,:5 -
Film thickness
j_{_
~ r
9
5,9 4
(, ,~ ffl)
FIG. 30. Compositional distribution of (a) graded-gap structure and build-in-field effect structure blocking photodiode. (after Maruyama and Hirai, 1983)
Heterostructures on Chalcogenide Glass and Their Applications
95
Electron beam _. ~ ....
._
Sb2S3 iN
As
-
Sc ,
As
-
Se
--
Te
AS- Sc Cc02 ,
_a..
-
4 ~tm
a
iii
'
-" _----
Sn02 J_.
FIG. 31. Build-in-effect TV tube target.
The target consists of a semitransparent SnO2 layer, followed by the photoconductive layer of polycrystalline CdSe with a thin glassy AszSe3 layer deposited on it. CdSe layer is prepared by spray-method on the heated substrate, but AszSe3 layer by vacuum evaporation. The air baking of the CdSe-ChGS layer heterostructure is shown (Section 2.2) to have a decisive significance for function of vidicon target. The minimum baking time of train = 15 min at 400 ~ is found to be optimal in production procedure. Glass
SRO 2
CdSe
Electron beam
R
I
Cathode |T
" A s 2 Se 3
V
FIG. 32. Arrangement of the "4dicon camera tube with CdSe-As2 Se3 heterostructure target.
96 4.3.
D. Tsiulyanu OPTICAL INFORMATION RECORDING MEDIA AND LITHOGRAPHY
Optical memory effects in chalcogenide glass allow their wide application in holography and high resolution lithography. Among these effects there are two of them which in metal-ChGS structures occur: photo-induced doping and photo-ESCTs on the interface, which has already been considered in Section 1.4. The photo-doping effect consists of light-induced dissolution of the silver layer predeposited on ChGS film, into this film. If thickness of silver layer is complete photodissolution occurs. Various methods of investigation of concentration profiles movement during silver migration through ChGS, including Rutherford backscattering, have indicated that it is not a simple diffusion. An inhomogeneous chemical photo-induced reaction occurs between Ag and ChGS, which result in formation of solid mixture ChGS: Ag. The structural and compositional transformations by photo-doping lead to a sharp decrease of etching rate of films in alkaline solutions. This permits to use photo-doping effects in optical (electronic, X-ray) wet lithography. Figure 33 shows the pattern of lithography process where Ag-ChGS is used as negative photoresist. Another lithography process, the so-called "dry development process", has been realized by Yoshikawa, Ochi and Muzushima (1980) using the differential plasma etching rate between non-doped and photo-doped portions of the film. After the Ag photo-doping the plasma etch rate in CF 4 gas drastically decreases and the etch rate reaches as 370:1. Thus, photosensitive structures Ag-ChGS can be used as inorganic photoresists with resolving power more than 5000 mm-1 and light sensitivity ~ 1 J/cm 2 both in wet and dry processes.
FIG. 33. Patternof negative lithographic process with Ag-ChGS photoresist.
Heterostructures on Chalcogenide Glass and Their Applications
97
Effect of ESCTs is observed in metal-glass structures, based on metals (such as aluminum) which do not interact with ChGS at usual conditions. Yet, the electrical field applied to A1-ChGS structure can give rise to interaction between these two components accompanied by variation of many properties of the structure, including optical ones. The rate of interaction is influenced by direction and density of electrical current flowing through interface A1-ChGS, hence, it is controlled by electrical field strength, and by other external factors (including illumination), leading to the ChGS conductivity variation. One can distinguish the anode and the cathode effects of ESCT (Section 1.4) Figure 34 shows the absorption spectra of used ChGS film and the A1-ChGS structure before and after ESCT with anode and cathode effects. It is seen that ESCT by anode process shifts the optical absorption edge to large wavelength and by cathode process to short wavelength. Thus, in the region of ChGS fundamental light absorption, anode effect of ESCT leads to bleaching while the cathode effect of ESCT leads to darkening of A1ChGS structure. In both cases, the unselective absorption appears in the region of total transparency of ChGS due to diffused and non-interacted aluminum. As the effect of ESCT can be intensified by external illumination, an electrical controlled photosensitive process can be developed (Andriesh et al., 1983a,b). Figure 35a shows the characteristic curve (optical density vs. exposure) of A1-AszS3Ge4 structure by anode ESCT. The curve exhibits a linear part, which moves at lower exposures, when the applied voltage increases; i.e., the sensitivity depends on applied voltage. Figure 35b shows the sensitivity defined as absorbed power per unit area needed to reach some value of bleaching vs. applied voltage. It is seen as a linear increasing of sensitivity with applied voltage increase. This enhancement of sensitivity can be realized by 2 - 3 orders of magnitude because of structure breakdown at high voltages. The sensitivity was found to strongly depend on the thickness of metal electrode and light wavelength. Maximum sensitivity is observed to be in the region of fundamental absorption edge of ChGS, i.e., corresponding to its forbidden gap. Hence, information
1000 > E
750
~'~~500
250 I
1,6
I
2,0
I
I
2,4
I
I
2,8
I
I
3,2
hv (eV) FIG. 34. Optical absorption edge of A1-ChGS-SnO2 structures before deposition of A1 layer (2) and after its deposition and dissolution by cathode (1) and by anode (3) processes of ESCT.
98
D. Tsiulyanu
~"
3
~
2
\ T (ref. ua.)
2
b)
I iO
20
l
30 U(V)
I f
lg E t (rel. units) FIG. 35. Opticaldensityvs. exposureof A1-As2S3Ge4structureby anodeESCT (a) and the sensitivityvs. applied voltage (b). Inset shows the reconstructedimage of the grating with density 4201/mm written by cathode ESCT. storage materials (including holographic ones) can be developed both for visible and IR regions of spectra. The resolving power particularly in cathode ESCT is higher than 1000 lin/mm. Inset of Figure 33a shows the image of the grating with density 420 lin/mm written by Fourier holographic method. Diffraction efficiency of the holographic images is 1 - 2 % , being controlled by applied voltage. The abnormally high etching rate of exposed portion of multilayer film, e.g., in alkaline solutions, suggests additional application of this process in photolithography as well. The time necessary for complete etching of exposed portions is less than 30 s varying on composition and temperature of etching solution, film thickness and other technological factors. Technical features of this photoresists are still not completely elucidated. Nevertheless, it is clear that photosensitivity is 10-1_ 1.0 J cm -2 and covers UV, visible and IR spectral regions with resolving power more than 1013.0 mm-1.
References Adler, D. (1973) Electrical switching in amorphous semiconductors,J. Vac. Sci. Technol., 10, 728-738. Alonso, B., Pioqueras, J. and Minoz, E. (1976) Electro-optical properties of amorphous As2Ses, Appl. Phys. Lett., 28, 41-44. Andreichin, R. and Kolomiets,B.T. (1962) Photoelectromotiveforce in arsenic chalcogenides,Fiz. Tverd. Tela., 4, 814-815. Andriesh, A.M. and Ciornii, S.P. (1975) New Semiconductor Alloys and Their Properties, Stiinta, Chisinau, pp. 88-94. Andriesh, A.M., Iovu, M.S., Kolomeyko, E.P., Tsiulyanu, D.I. and Shutov, S.D. (1980a) Investigation of the hole transport in vitreous As2S3, J. Non-Cryst. Solids, 35/36, 981-986.
Heterostructures on Chalcogenide Glass and Their Applications
99
Andriesh, A.M., Iovu, M.S., Tsiulyanu, D.I. and Shutov, S.D. (1981) Glassy Arsenic Trisulphide and Its Alloys, Stiinta, Kishinau. Andriesh, A.M., Nefeodov, V.I., Tsiulyanu, D.I., Socolov, A.N. and Triduch, G.M. (1990) Anode electrode process in chalcogenide glassy semiconductors stimulated by external influences, J. Techn. Phys., 60, 148-151. Andriesh, A.M., Triduch, G.M. and Tsiulyanu, D.I. (1988) Capacitance of metal-chalcogenide glassy semiconductor structures by physical-chemical transformations on the interface, Phys. Chem. Glass, 14, 633-636. Andriesh, A.M. and Tsiulyanu, D.I. (1973) Influence of annealing and temperature on the absorption edge of AszS3-Ge amorphous films, Phys. Stat. Sol. A, 19, 307-312. Andriesh, A.M. and Tsiulyanu, D.I. (1976) Memory effect by electrostimulated chemical transformations in thin film structures metal-glassy semiconductor-metal, J.Tech. Phys. Lett., 2, 38-41. Andriesh, A.M., Tsiulyanu, D.I. and Kolomeyko, E.P. (1977a) Electrophysical properties of heterojunctions silicon single crystal-amorphous AszS3-Ge, Fiz. Tehn. Poluprovodn., 11, 664-668. Andriesh, A.M., Tsiulyanu, D.I. and Kolomeyko, E.P. (1977b) Electromotive force in thin film structures metal-chalcogenide glassy semiconductor, J. Tech. Phys. letters, 3, 135-137. Andriesh, A.M., Tsiulyanu, D.I. and Kolomeyko, E.P. (1979) Space charge limited currents in glassy materials AszS3-Ge, Fiz. Tehn. Poluprovodn., 13, 201. Andriesh, A.M., Tsiulyanu, D.I. and Kolomeyko, E.P. (1980b) Temperature dependence of the space charged limited currents in glassy As2S3, Fiz. Tehn. Poluprovodn., 14, 789-791. Andriesh, A.M., Tsiulyanu, D.I. and Kolomeyko, E.P. (1985) About mechanism of current flow in isotropic heterojunctions glassy AszS3/p-Si, Fiz. Tehn. Poluprovodn., 19, 1671-1675. Andriesh, A.M., Tsiulyanu, D.I. and Triduch, G.M. (1983a) Chalcogenide glassy semiconductor based electrical-controlled photosensitive structures, J. Techn. Phys., 54, 715-718. Andriesh, A.M., Verlan, V.I., Gumeniuk, N.A. and Malkov, S.A. (1983b) Physical Phenomena in Noncrystalline Semiconductors, Ujgorod University, Ujgorod, pp. 37-39. Barbe, D.F. (1971) Theory of field effect in amorphous covalent semiconductor films, J. Vac. Sci. Tehnol., 8, 102-107. Brodsky, M.H. and Dohler, G.H. (1975) A new type of junctions: amorphous-crystalline, Crit. Rev. Solid State Sci., 5, 591-595. Busmandrud, O. (1975) Properties of amorphous-crystalline silicon junctions, Phys. Stat. Sol., 28, 225-260. Buzdugan, A.I., Zelenina, L.I., Iovu, M.S., Simashkievici, A.A. and Shutov, S.D. (1985) Injection sensibilization of photoeffect in double heterostructure from glassy semiconductors. In Glassy Semiconductors, RTR, Leningrad, pp. 56-57. Buzdugan, A.I., Zelenina, L.I., Ivashcenko, Iu.N., Iovu, M.S., Simashkievici, A.A. and Shutov, S.D. (1983) Photothermoplastic material based on glassy semiconductor heterostructure, J. Sci. Appl. Photogr. Cinem., 5, 384-386. Chang, C.H. and Dann, B. (1977) Construction of energy-band diagrams for amorphous/crystalline heterojunctions, J. Appl. Phys., 48, 1751- 1752. Cimpl, Z., Jedelicka, M., Kosek, F. and Schauer, F. (1979) Electrical properties of CdSe-amorphous chalcogenide heterostructure. In Proceedings of International Conference on Amorphous Semiconductors '78, Pardubice, Czecho-Slovakia, pp. 635-638. Corsacov, V.V. and Tsukerman, V.G. (1976) Photoelectrochemical process in metal-chalcogenide glassy semiconductor structure, J. Tech. Phys. Lett., 2, 10-12. Cowley, A.M. (1966) Depletion capacitance and diffusion potential of GaP Schottky barrier diodes, J. Appl. Phys., 37, 3024-3032. Dann, B. and Mackenzie, J.D. (1976) Transport properties of glass-silicon heterojunctions, J. Appl. Phys., 47, 1010-1014. Dementev, I.V., Panasiuk, L.M. and Tarabuhina, N.B. (1974) Some photoelectrical properties of thin films heterojunctions AszS3/AszSe3. In Physical Processes in Heterostructures and Some AIIB vI Compounds, Stiinta, Kishinau, pp. 47-53. Ema, Y. and Hayashi, T. (1982) Aging effect of capacitance and related effects in Au/a-Se/A1 structure, Jpn. J. Appl. Phys., 21, 1665-1670. Gill, W.D. and Bube, R.H. (1970) Photovoltaic properties of Cu2S-CdS heterojunctions, J. Appl. Phys., 41, 3731-3738.
100
D. Tsiulyanu
Glebov, A.S. and Petrov, I.M. (2000) Physics and Application of Current Instabilities in Glassy Semiconductors, Uzorochie, Ryazani. Gupta, H.M. and Overstraeten, R.J. (1975) Role of the trap states in the insulator region for MIM characteristics, J. Appl. Phys., 46, 1682- 2675. Iovu, M.A., Iovu, M.S. and Shutov, S.D. (1978) Photoelectrical properties of contacts metal-glassy AszS3 or Sb2S3, J. Tech. Phys. Lett., 4, 1246-1250. Jivoderov, A.N. (1980) Method of investigation of the C - V characteristics of silicon glassy semiconductor heterojunctions. In Physics of Semiconductors and Microelectronics, Ryazani, p. 27. Kolomiets, B.T., Grigorovici, R., Croitoru, N. and Vescan, L. (1970) Rectifying properties of junctions between vitreous T12SeAszTe3 and Ge, Si and InSb single crystals, Rev. Roum. Phys., 15, 129-131. Kolomiets, B.T. and Lyubin, V.M. (1973) Photoelectric phenomena in amorphous chalcogenide semiconductors, Phys. Stat. Sol. A, 17, 11-46. Kolomiets, B.T., Lyubin, V.M., Maidzinski, V.S., Plisova, P.A., Fedorova, G.A. and Fedorova, E.I. (1971) Electrical and photoelectrical properties of some thin films amorphous heterostructures, Fiz. Tehn. Poluprovodn., 5, 1533-1540. Korkinova, Ts.N. and Andreichin, R.E. (1982) The properties of amorphous arsenic sulphide-crystalline silicon junction. In Proceedings of International Conference on Amorphous Semiconductors '82, Bucharest, pp. 253-255. Krupanidi, S.B., Srivastava, R.K., Srinivas, K., Bhattacharya, D.K. and Marsingh, A. (1983) I - V and C - V studies of evaporated amorphous arsenic telluride film on crystalline silicon, J. Appl. Phys., 54, 1383-1389. Lampert, M.A. and Mark, P. (1970) Current Injection in Solids, Academic Press, New York. Lyubin, V.M. and Maidzinski, V.S. (1964) Contact phenomena and carrier mobility in amorphous Sb2S3 films, Fiz. Tverd. Tela., 6, 3740-3742. Lyubin, V.M. and Maidzinski, V.S. (1968) The peculiarities of current flow and photoelectric processes in Sb2S3 by existence of injection and blocking contacts, Fiz. Tehn. Poluprovodn., 3, 1675-1679. Mamontova, T.N., Kochemirovskii, A.S. and Pivovarova, L.V. (1988) Electronic processes on the surface of As-Se chalcogenide glassy semiconductors, Phys. Stat. Sol. A, 107, 11-43. Marshall, J.M. (1978) Charge screening length in amorphous chalcogenide semiconductors, Phil. Mag., 38, 407-417. Marshall, J.M. and Owen, A.E. (1976) Field effect measurements in disordered As3oTeasSilzGelo and AszTe3, Phil. Mag., 33, 457-474. Maruyama, E. (1982) Amorphous build-in-field effect photoreceptors, Jpn. J. Appl. Phys., 21, 213-223. Maruyama, E. and Hirai, T. (1983) Physical aspects in amorphous image devices, J. Non-Cryst. Solids, 59/60, 1247-1254. Michaelson, H.B. (1977) The work function of the elements and its periodicity, J. Appl. Phys., 48, 4729 -4733. Milnes, A.G. and Feucht, D.L. (1972) Heterojunctions and Metal-Semiconduct'or Junctions, Academic Press, New York. Mott, N.F. (1971) Conduction in non-crystalline systems. VII, Nonohmic behavior and switching, Phil. Mag., 24(190), 911. Mott, N.F. and Davis, E.A. (1971) Electronic Processes in Non-crystalline Materials, Clarendon Press, Oxford. Muller, P. (1976) Direct proof of the conduction type in semi-insulating thin films, Phys. Stat. Sol. A, 33, 543-553. Muller, P. (1981) Electric relaxation in high-resistivity solids, Phys. Stat. Sol. A, 67, 11-60. Oreshkin, P.T., Jivoderov, A.N. and Glebov, A.S. (1980) Band structure and electric instability in crystal-glassy semiconductor heterojunctions. In Physical Processes in Non-crystalline Semiconductors, Stiinta, Kishinau, pp. 115-118. Persin, M. and Mitra, V. (1980) Electrical and photovoltaic properties of a heterojunction between A s - T e - G e film and crystalline silicon, Thin Solid Films, 70, 85-90. Peterson, S.K. and Adler, D. (1974) On state characteristics of amorphous/crystalline heterojunctions, Appl. Phys. Lett., 25, 211-213. Petrillo, G.A. and Kao, K.C. (1974) The effects of electrode materials on the switching behavior of the amorphous semiconductor Si 12GeloAS3oTe48, J. Non- Cryst. Solids, 16, 247-- 257.
Heterostructures on Chalcogenide Glass and Their Applications
101
Reinhard, D.K., Arnts, F.O. and Adler, D. (1973) Properties of chalcogenide glass-silicon heterojunctions, Appl. Phys. Lett., 23, 186-188. Rhoderick, E. (1978) Metal-Semiconductor Contacts, Clarendon Press, Oxford. Schauer, F., Heza, S., Jedlicka, M., Kosek, F. and Cimpl, Z. (1980) Influence of CdSe layer air-baking on physical properties of CdSe-As2Se3 heterostructures, Phys. Stat. Sol. A, 61, 349-356. Sersembinov, Sh.Sh., Maximova, S.Ia. and Fedorenko, T.A. (1982) Investigation of metal- amorphous AszSe3 contact, Akad. Sci. Kaz. SSR, Ser. Math-Phys., 4, 58-62. Sharma, B.L. and Purohit, R.K. (1974) Semiconductor Heterojunctions, Pergamon Press, New York. Shiraishi, T., Kurosu, T. and Iida, M. (1978) Some properties of Schottky barrier formed on chalcogenide amorphous semiconductor, Jpn. J. Appl. Phys., 17, 1883-1884. Simashkievici, A.A. and Shutov, S.D. (1984a) Evidence of Schottky barrier formation at contact of metal with chalcogenide glassy semiconductor, Phys. Stat. Sol. A, 84, 343-352. Simashkievici, A.A. and Shutov, S .D. (1984b) Investigation of contact phenomena on the metal-ChGS interface. In Proceedings of the Conference on Amorphous Semiconductors--84, Gabrovo, Bulgaria, pp. 71-73. Snell, A.I. (1980) The metal-amorphous silicon barrier, interpretation of capacitance and conductance measurements, J. Non-Cryst. Solids, 35/36, 593-598. Spear, W.E., Lecomber, P.G., Kinmond, S. and Brodsky, M.H. (1976) Amorphous silicon p - n junction, Appl. Phys. Lett., 28, 105-106. Spicer, W.E., Lindau, I., Skeath, P., Su, C.Y. and Chye, P. (1980) Unified mechanism for Schottky-barrier formation on III-V oxide interface states, Phys. Rev. Lett., 44, 420-423. Stotzel, H., Kottwitz, A. and Leimer, F. (1976) Electrical behavior of amorphous GeSe films with blocking contacts. In Electrical Phenomena in Non-crystalline Semiconductors (Ed., Kolomiets, B.T.) Nauka, Leningrad, pp. 315- 319. Sze, S.M., Coleman, D.J. and Lova, A. (1971) Current transport in metal-semiconductor-metal structures, Solid State Electron., 14, 1209-1218. Tonge, N., Minami, T. and Tanaka, M. (1979) The electrical and photovoltaic properties of heterojunctions between an amorphous G e - T e - S e film and crystalline silicon, Thin Solid Films, 56, 377-382. Tsiulyanu, D.I. (1988) Formation of the Schottky-Mott barrier on the contact metal-chalcogenide glassy semiconductor, Fiz. Tehn. Poluprovodn., 22, 1181- 1184. Tsiulyanu, D.I., Andriesh, A.M. and Kolomeyko, E.P. (1978a) Arsenic trisulphide in isotropic amorphouscrystalline heterojunctions, Phys. Stat. Sol. A, 50, 195-202. Tsiulyanu, D.I., Andriesh, A.M. and Kolomeyko, E.P. (1978b) Injection currents in chalcogenide glassy semiconductors AszS3-Ge. In Proceedings of the International Conference on Amorphous Semiconductors '78, Pardubice, Czecho-Slovakia, pp. 360-363. Tsiulyanu, D.I., Ciumacov, I.S. and Grinshpun, L.B. (1983a) Cathode effect of electrostimulated chemical transformation in metal-chalcogenide glassy semiconductor structures, Fiz. Tehn. Poluprovodn., 17, 2196-2198. Tsiulyanu, D.I., Kolomeyko, E.P. and Bazik, N.G. (1983b) The properties of chalcogenide glassy semiconductor-metal structures by electrostimulated chemical transformation on the interface, Fiz. Tehn. Poluprovodn., 17, 491-493. Tsiulyanu, D.I., Kolomeyko, E.P. and Stamov, V.N. (1987) Photoelectric phenomena in isotropic vitreous As2S3-silicon single crystal heterojunctions, Phys. Stat. Sol. A, 102, K103-K107. Tsiulyanu, D.I. and Triduch, G.M. (1991) Current transport in metal-chalcogenide glass structures with blocking barrier at the interface, Phys. Stat. Sol. A, 123, K13-K18. Van Opdorp, C. and Kanerva, H.K.I. (1967) Current-voltage characteristics and capacitance of isotropic heterojunctions, Solid State Electron., 10, 401-421. Wallace, A.M., Owen, A.E. and Robertson, J.M. (1978) Electrical contact properties of semiconducting chalcogenide glasses, Phil. Mag., 38, 57-70. Wey, H.Y. (1976) Surface of amorphous semiconductors and their contacts with metals, Phys. Rev. B, 13, 3495-35O5. Yoshikawa, A., Ochi, O. and Muzushima, Y. (1980) Dry development of Se-Ge inorganic photoresist, Appl. Phys. Lett., 36, 107-110. Zajic, J., Kosek, F., Cimpl, Z., Schauer, F. and Stourac, L. (1991) Reverse dark I - V characteristics of the heterojunction based on oriented polycrystalline CdSe and amorphous As2Se3 films, J. Non-Cryst. Solids, 128, 1-7.
This Page Intentionally Left Blank
CHAPTER
3
ION CONDUCTIVITY AND SENSORS E. Bychkov a'l aARGONNE NATIONALLABORATORY,ARGONNE, IL 60439, USA
Yu. Tveryanovich b and Yu. Vlasov b bST PETERSBURGUNIVERSITY, 199034 ST PETERSBURG,RUSSIA
1. Ion Transport in Silver and Copper Chalcogenide and Chalcohalide Glasses 1.1. INTRODUCTION Ionic transport in glasses was discovered in the 19th century following the classical work of Warburg (1884). Since then, considerable progress has been achieved in both theoretical understanding and practical applications of ion-conducting vitreous systems (see Frischat, 1975; Malugani and Robert, 1980; Ribes, Barrau and Souquet, 1980; Kennedy and Yang, 1987; Vlasov and Bychkov, 1987; Hayashi, Tatsumisago and Minami, 1999; Doremus, 1962 and references therein). Nevertheless, this topic and especially the ion-conducting mechanisms in disordered solids need additional study using traditional macroscopic methods (ac and dc electrical conductivity, tracer diffusion, and ion transport number measurements), as well as advanced structural techniques on third generation synchrotron light sources and spallation neutron sources over a large range of the scattering vector Q. This approach led to the discovery of important features: in particular, different transport regimes at low and high mobile ion content that are closely related to a competition between the stochastic scenario and a non-random distribution of the mobile ions in the glass network. Well-known experimental findings such as compositional dependence of the Haven ratio HR, interpreted earlier by a number of drastically different ion transport models, can also be explained using a unified approach. Many of the new experimental results were obtained for silver and copper chalcogenide glasses which appear to be useful model materials, in part because of a large accessible composition domain, as well as coverage of five orders of magnitude in the mobile cation content, and corresponding dramatic changes in the ionic transport up to 10 orders of magnitude. In the following sections, we will focus our attention on the observed phenomena. On sabbatical leave from the Universitd du Littoral, 59140 Dunkerque, France. 103
Copyright 9 2004 Elsevier Inc. All rights reserved. ISBN 0-12-752189-5 ISSN 0080-8784
104
E. Bychkov et al.
1.2. PERCOLATION THRESHOLD AND CRITICAL PERCOLATION REGIME AT LOW MOBILE ION CONTENT
Many authors have investigated electrical characteristics of Ag-doped chalcogenide glasses (Owen, 1967; Kolomiets, Rukhlyadev and Shilo, 1971; Kawamoto, Nagura and Tsuchihashi, 1974; Rykova, Borisova, Pasin and Kalinina, 1980; Robinel, Carette and Ribes, 1983; Proceedings of the International Conference on Solid State Ionicsm2001, 2002) but only few papers have been devoted to tracer diffusion measurements (Kawamoto and Nishida, 1977; Zhabrev and Kazakova, 1982; Vlasov, Bychkov and Seleznev, 1987), and the concentration ranges of silver doping were different. The effect of small Ag additives (< 1 at.%) on electronic properties of vitreous arsenic chalcogenides was extensively studied in the 1970s (Owen, 1967; Kolomiets et al., 1971). Ion-conducting glasses with much higher silver concentration (up to 50 at.%) were investigated later (Kawamoto et al., 1974; Rykova et al., 1980; Robinel et al., 1983; Proceedings of the International Conference on Solid State Ionicsm2001, 2002). It has also been reported (see Borisova, 1981; Borisova, Bychkov and Tveryanovich, 1991 and references therein) that As-based glasses doped wlith silver exhibit a steplike transition of the conductivity at ~ 6 - 8 at.% Ag, where conductivity increases by 3 - 5 orders of magnitude. Glasses with lower Ag concentrations have been suggested as amorphous semiconductors, and vitreous alloys with higher silver content are vitreous electrolytes. Phase separation was found to be responsible for these drastic changes. On the contrary, Kawamoto et al. (1974) reported that in homogeneous Ag2S-GeS-GeS2 glasses the ionic conductivity increases monotonically with increasing silver content. The minimal Ag concentration in these glasses was 0.85 at.%. More recently, systematic studies of silver chalcogenide glasses over an extremely large composition range, from a few ppm to 30-35 at.% Ag, have been completed (Bychkov, Tsegelnik, Vlasov, Pradel and Ribes, 1996; Bychkov, Bychkov, Pradel and Ribes, 1998; Drugov, Bolotov, Vlasov and Bychkov, 2000). The macroscopic transport properties were investigated using electrical measurements and 11~ tracer diffusion experiments. These two experimental techniques are perfectly complementary since the residual electronic conductivity of the host matrix, especially for selenide systems, could affect the ionic conductivity measurements and lead to erroneous conclusions. The results clearly show at least three distinctly different ion transport regimes: (i) below, (ii) just above, and (iii) far above the percolation threshold at Xc ~ 30 ppm Ag.
1.2.1.
Electronic vs. Ionic Insulators
At a doping of a few ppm (domain (i) at x < Xc), the glasses are essentially electronic insulators: the silver ion transport number tAg+ is between 0.1 and 0.2 (Fig. 1). The Ag diffusion coefficient DAg (Fig. 2) and the diffusion activation energy Ed (Fig. 3) do not depend on the silver content and are similar to those in the pure host matrix. Penetration profiles for glasses with x --- 120 ppm Ag obey the usual solution of Fick' s law for thinlayer geometry over the entire investigated temperature range. Figure 4 shows that
105
Ion Conductivity and Sensors 1.2
%
1.0
0.8
Percolation Threshold
I
0.6
r~
0.4
@
9 o
0.2
,
0 . 0
|
,
l,,,|l
10- 4
,
,
,,,,,I
,
10- 2
10-3
,
,
,,,,,I
,
,
,
Ag2S-As2S 3 Ag2S-GeS-GeS 2
,,,,,I
10-1
,
,
,
,,,,,I
10-~
,
,
,
,,,,
101
10 z
Silver Concentration (at.%) FIG. 1. Silver ion transport n u m b e r tAg+ as a function of the silver content for A g 2 S - G e S - G e S 2 et al., 1996) and A g 2 S - A s 2 S 3 (Drugov et al., 2000) glasses.
(Bychkov
the experimental points are rather well approximated by one erfc function A(X, tA) = No erfc
(1)
2 Dx/_D~ A ,
where A(X, ta) is the residual activity of the sample after a thickness X has been removed, D the diffusion coefficient and ta the diffusion anneal time. In some cases, the initial counting rate A0 was found to be slightly higher than expected from Eq. (1), but the strong
10-7
,
,
,
. ., . . i
,
,
Ag2S-As2S 3 Glasses , , i , ,i . . .,. . . , i . . .,. , , ,
i
,
169.1 ~ "7
......,
O
Percolation 10 . 9
,,,i
"ii-
10-8
e~0 <
,,,
Percolation Domain
Threshold
10-10 O
tD = - 0 . 3 8 ( 2 )
O .,=~ r~
10 TM
\ As2S
ModifierControlled
Domain 3
10-12 ........ I ....... I ........ I ........ I 10-4 10-3 10-2 10-1 100
...... I . . . . . . . . 101 102
Silver Concentration (at. %) FIG. 2. l l~ tracer diffusion isotherm at 169.1 ~ for the A g e S - A s 2 S 3 glasses: O, D r u g o v et al. (2000); 9 K a z a k o v a (1980). The shaded area corresponds to pure As2S3.
106
E. Bychkov et al. ''""1
'
' ''""1
'
'
''""1
'
''""1
'
'
''""
1.2 ModifierControlled Domain
Percolation-Controlled Domain
1.1 1.0 ~
0.9
I-i
m
0.8
i
@
.~ <
0.7
o
0.6
.,,~
/
.,..~
0.5
Percolation threshold 0.4 Ag2S-As2S 3 0.3 i
10-4
i
iiiiiii
i
10-3
iiiiili
i
10-2
i iiiinl
i
10-1
i iiiinl
I
100
IIIIIll
I
101
i
I iiiii
102
Silver Concentration (at.%) FIG. 3. Diffusion activation energy Eo for AgzS-As2S3 glasses: O, Drugov et ill. (2000); 9 Kazakova (1980). The shaded area corresponds to pure As2S3.
profile distortion in a near-surface layer has become much more evident for the pure host g l a s s ( A s z S 3 in this particular case) and glasses with x < 120 ppm Ag. Their profiles cannot be described by a single diffusion coefficient (Fig. 4). The two-erfc fitting was used to extract values of the near-surface low-D component DL and bulk high-D diffusion coefficient DH. The values of DL did not depend on temperature and glass composition. These observations were consistent with previous results for AszS3 (Buroff, Nebauer, SiJptitz and Willert, 1977). In contrast, the bulk diffusion coefficients Dr] are thermally activated. Further, we will discuss only these bulk diffusion coefficients and related parameters (Ed and HR). The ionic conductivity o'i becomes predominant above the percolation threshold at Xc even for highly diluted glasses in case of the sulphide systems (AgzS-GeS-GeS2, AgzS-AszS3). The silver ion transport number tAg+ exhibits a step-like increase from 0.1-0.2 (x < Xc) to 0.6-0.8 (x = 8 0 - 1 2 0 p p m Ag) and increases rapidly to 1 with further silver doping (Fig. 1). The selenide glassy systems ( A g - G e - S b - S e , A g I AszSe3) reveal an enhanced residual electronic conductivity but even in this case tAg+ tends to 1 at x --> 0.3 at.% Ag. These results contradict the usual hypothesis that only Agrich chalcogenide glasses belong to nearly pure ionic conductors. The reason for the observed discrepancies is probably the limited reversibility at the; glass/reversible contact interface which is usually used for emf or other electrical measurements to determine
107
Ion Conductivity and Sensors
105
'40C
ell
==
0.012 at.% Ag
"~ 104 0
rj
103 0
20
40
60
80
100
120
Depth (gtm) FIG. 4. ll~ penetration profiles for extremely diluted AgzS-AszS3 glasses below (12ppm or 0.0012 at.% Ag) and above (120 ppm or 0.012 at.% Ag) the percolation threshold at Xc ~ 30 ppm Ag (Drugov et al., 2000). The solid lines representthe erfc fitting of the profiles. Near-surfacelow-D componentDL and bulk high-D diffusion coefficientDH for the glass with 12 ppm Ag are shown by dashed and dotted lines, respectively.
the Ag + ion transport number. Experiments with radioactive 11~ solution clearly show rather slow tracer exchange dynamics and chemical reaction at the diluted glass/ solution interface (the DL component in Figure 4) which certainly leads to a strongly diminished value of tAg+ in the emf experiments.
1.2.2.
C h a r a c t e r i s t i c F e a t u r e s in the C r i t i c a l P e r c o l a t i o n D o m a i n
Just above the percolation threshold (domain (ii) at x > Xc), one observes a step-like increase of the diffusion coefficient (by a factor of 5 - 1 0 at high T and about two orders of magnitude at room temperature) with a simultaneous abrupt decrease of Ed by 0 . 1 - 0 . 2 eV. Furthermore, at Xc < x-< 1 - 3 at.% Ag, the parameters of the Ag + ion transport (o'i, DAg, Ea and Ed) exhibit very characteristic trends. First, a power-law composition dependence of oi(x) and OAg(X) was observed over 2.5 orders of magnitude in the Ag concentration x (Fig. 5) o"i (x, T) -- ~ri (1, T ) x t(T) ,
(2)
DAg(X, T) -- DAg(1 , T ) x t(r)-I ,
(3)
108
E. Bychkov et al. (a)
(b)
10-3
Ag-Ge-S Glasses
10-4
o
T-
~
10-5
Oo
"1o
C (D .m O
0
't D = 0 . 6 0 /
10-12. ~,~,, Z . . ~ ' "
-
"N 10-13 9
10-10
a
10-12. 10-13. 10-TM.
9
0
10-9
C O 1 0-11.
O
o o
r't~ 10-11.
o
~"ol;:: 1 0-7 "T, ~ lO 8 >
Ag-Ge-S Glasses
10-1~
0,,.%
o
10-6
10-9
I0-15 ~,1 9
298 ~ -
10-15 10"3
|
.
.
.
.
.
1:D = 1 . 0 2 ( 1 2 ) /
10-14.
~8K / .
.
.
10-2
.
.
.
.
.
.
.
.
.
10"1
.
.
.
.
.
.
.
.
.
.
10 0
.
.
.
.
.
.
.
.
.
101
.
.
.
.
.
.
.
10 2
10"16
..... . . . ~,, /, ,T~. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10-3
Silver Concentration (at.%)
10-2
10-1
10 0
101
10 2
Silver Concentration (at.%)
FIG. 5. (a) Ionic conductivity ~ri(x) and (b) diffusion coefficient DAg(X) isotherms at 298 and 373 K for Ag2SGeS-GeS2 glasses plotted on a log-log scale (Bychkov et al., 1996). The critical diffusion exponent tD(T) is less than the conductivity one, tD(T) = t(T) - 1, due to the Nemst-Einstein relation (DM ec o-ix-l).
with temperature-dependent power-law exponent t(T) t(T) = to + To/T -~ To~T,
(4)
where tri(l,T) and DAg(I, T) are the ionic conductivity and diffusion coefficient of a hypothetical phase at the mole fractionx = I, to is small and could be neglected in Eq. (4), To is the critical fictive temperature which will be discussed later. Accordingly, the conductivity E a and diffusion Ed activation energies are very similar and decrease linearly with increasing log x (Fig. 6). Three different models give nearly identical equations to describe the observed phenomena: modified classical percolation (Bychkov et al., 1996), the dynamic structure model (Bunde, Ingram and Maass, 1994) and a statistical (occupation) approach taking into account percolation effects on ionic conduction in glasses (Hunt, 1994). The modified classical percolation model predicts the effect of glass network dimensionality and seems to be more adapted to explain the obtained conductivity and diffusion results. Classical geometrical percolation (Kirkpatrick, 1973, 19"75, 1979) dealing with temperature-invariant conductances in a random resistor network gives a simple relation for the conductance G(p) as a function of site or bond fraction p just above the percolation threshold Pc G(p) oc (p - pc) t,
(5)
for site, bond or correlated bond percolation models. The model parameters Pc and t depend primarily on the resistor network dimensionality and are mutually consistent for these three geometrical models (Table I). One should note the Pc values of 0.1-0.7 for 3D and 2D network, distinctly smaller values of the critical exponent t for 2D models, and the absence of percolation in 1D case.
109
Ion Conductivity and Sensors
1.0
'
'''"1
'
'
'
'~
'
'
'
'''"1
'
'
'''"1
0.9
0.8
ModifierControlled Domain
v
~" 0.7 Ag2S-GeS-GeS2 o
0.6
> o,,,~ o
<
~
o.5
>
o L)
0.4 Percolation-Controlled Domain
0.3
0.2
, ,,,,,i
0.01
,
,
, ,,,,,i
,
,
, ,,,,,i
0.1
,
, ,,,,,i
1
10
Silver Concentration (at. %) FIG. 6. Composition dependencies of the conductivity activation energy Ea for Ag2S-GeS-GeS2 (Bychkov et al., 1996) and AgzS-AszS3 (Bychkov et al., 1998) glasses in the percolation and modifier-controlled domains.
The allowed volume approach (Bychkov et al., 1996) enables to understand the observed enormous difference of four orders of magnitude between Pc = 0.1-0.7 and xc ~ 3 x 10 -5. In contrast to macroscopic classical models, an average microscopic cage of 18-20 ~3 occupied by the Ag + ion itself could not represent a site of the 'infinite' percolation cluster due to a certain probability to find this ion outside its residence place. An allowed volume of the glass limited by local mean-square displacements of the mobile ion is hence larger than 1 8 - 2 0 ~3 (Fig. 7). The allowed volume namely plays a role of the percolation cluster site. In a spherical approximation taking into account
TABLE I PERCOLATION THRESHOLDPc AND CRITICAL EXPONENT t FOR SITE, BOND AND CORRELATED BOND GEOMETRICAL PERCOLATION MODELS OF DIFFERENT DIMENSIONALITY (KIRKPATRICK, 1973, ~978, ~979) Network 3D 2D 1D
Pc
t
0.12-0.43 0.35-0.70 1.00
1.6-1.7 1.0-1.2 0
110
E. Bychkov et al.
Fro. 7. Allowed volume approach: the atomic volume of the Ag + ion (18--20,3) is much less than the allowed volume of the glass, which is restricted by local mean-square displacements of the mobile cation. The allowed volumes play a role of sites in an 'infinite' percolation cluster, and hence their volume fraction should be of the order of Pc for the onset of percolation.
the theoretical values of Pc, the atomic volume of the Ag + ion and Xc, the radius of the allowed volume sphere appears to be 16-25 ,~ at the percolation threshold. This rather rough estimation is in good agreement with an average A g - A g distance in extremely diluted glasses at Xc assuming homogeneous distribution of the Ag + ions. Additionally, recent ac conductivity measurements carried out for Na20--GeO2 glasses (Roling, Martiny and Funke, 1999) showed that a characteristic transport distance, deduced from the local mean-square displacements {?2(c~)), depends on the mobile ion content approaching the values of 20-22 A in the limit of low sodium concentrations (60-70 ppm Na). Temperature dependence of the critical exponent t (Fig. 8 (Bychkov, 2001) and Eq. (4)) can easily be understood taking into account thermally activated o'i(T) and DAg(T). Having assumed that the pre-exponential factor o0 does not depend on x and the transport parameters tri(1 , T) and DAg(1, T) are also thermally activated, which is a good approximation for the investigated silver chalcogenide glasses, Eqs. (2)-(4) can be re-written giving for the ionic conductivity, for example, the following relations o-i(x, T) -- o-0T-1 exp - -k--T- ' E(x) -- Eo - kTo
ln(X),
(6) (7)
Xc
where E0 is the activation energy at the percolation thre,shold Xc. The critical temperature To is thus a unique parameter, which governs the ion transport properties in the critical percolation region and can be determined either from the conductivity and/or diffusion isotherms, or from the slope OE(x)/O(ln x), the two methods giving
111
Ion Conductivity and Sensors '
'
I
'
'
I
'
'
l
'
'
I
'
'
I
'
'
2.0-
1.8 Ag-Ge-S 1.6 1.4 O
1.2
-~o 0.8 0.6 --
T O= 242(10) K-
O.4 ,
1.8
,
I
2.1
,
i
I
2.4
,
,
I
,
,
2.7
I
3.0
,
,
I
3.3
,
,
3.6
1000/T (K -1) FIG. 8. Temperature dependence of the critical exponent t for a number of silver chalcogenide glassy systems (Bychkov, 2001). The t(T) dependence is perfectly described by Eq. (4) and in that respect differs significantly from T-invariant critical exponent t in classical geometrical percolation models (Kirkpatrick, 1973).
similar results. Consequently, the transport characteristics of the whole system in the critical percolation domain can be represented as a master plot (Bychkov, 2001). Figure 9 shows this approach for a number of Ag + and Cu + conducting chalcogenide and chalcohalide glasses, when the 11~ and 64Cu tracer diffusion data (Bychkov et al., 1996, 2001; Drugov et al., 2000) are plotted as log{DM(x,T)x 1-r~ vs. reciprocal temperature T -1 This figure allows two important points to be mentioned. First, the combination of Eqs. (2), (3) and (7) provide an appropriate scaling; i.e., the experimental data points for different glass compositions in the critical percolation region are well approximated by a single line, especially taking into account that the DM(x)'s for a particular system at the same temperature might vary by three orders of magnitude as a function of x (see, for example, Fig. 5). Second, master plots for the MI-AszSe3 and MzX-AszX 3 glass families, where M = Ag, Cu and X = S, Se, appear to be very close to each other within approximately half an order of magnitude, as they have very similar values of To, 273 K _< To--< 311 K. However, they differ significantly from the master plots for the Ge-based systems. The A g z S - G e S - G e S 2 glasses provide a master plot which is a factor of 100-1000 higher than those for the AszX3-based glasses. Basic conclusions from these observations can be formulated as follows. Neither the nature of the mobile cation (Ag + or Cu+), nor the chemical form of the dopant (metal halide or chalcogenide) play any important role in the critical percolation domain. The chemical form of the host matrix (sulphide or selenide) is not important either.
112
E. Bychkov et al. I
'
'
I
'
'
I
'
'
I
'
'
I
'
'
I
'
'
I
-8.0 -8.5 -9.0 O
-9.5 -10.0
| +
~
-GeS2
-10.5 -11.0
~ -11.5 O
-12.0
~
A
Ag-Ge-Sb-Se~,,, AgI-AszSe3
-12.5
gzS-AszS3
-13.0 I
1.8
,
,
I
,
2.1
,
I
,
2.4
,
I
~
2.7
~
I
3.0
,
~
I
3.3
,
~
I
3.6
1000/T (K -1) FIG. 9. Master plots of the l l~ and 64Cu tracer diffusion coefficients for a number of Ag + and Cu + conducting chalcogenide and chalcohalide glasses in the critical percolation domain (Bychkov et al., 1996, 2001" Drugov et al., 2000).
1.2.3. Critical Fictive Temperature To and Connectivity of the Host Matrix The critical fictive temperature To appears to be a primordial parameter to describe the percolation phenomena in the glass. The percolative ion transport depends on the number of 'infinite' percolation clusters, on one hand, and on their interconnectivity, on the other. The second term in Eq. (7) hence represents a configuration entropy term, where ln(x/Xc) is related to the number and To to the interconnectivity of conduction pathways frozen below Tg. The interconnectivity of clusters embedded in a host matrix should depend on the host network connectivity if the dopant does not change considerably the microstructural organization of the network. Plotting the critical fictive temperature To versus the average local coordination number of the host matrix (no), one obtains a nearly linear decrease of To with decreasing (no) (Fig. 10) (Alekseev and Bychkov, 2003). The most important conclusion, which could be drawn from this dependence, is a suggested absence of the percolative transport (To ~ 0) for chain structure.s ((no) = 2), i.e., TOcc (no) - 2.
(8)
This prediction is identical to the absence of percolation for 1D network in classical models (Table I). Neutron diffraction experiments (Bychkov and Price, 2000; Bychkov, Siewenie, Sampath and Benmore, 2004) show that in the percolation-controlled domain of silver chalcogenide glasses neither the short- nor the intermediate-range order exhibits any significant transformation, justifying the choice of the host network average coordination number (no) in Eq. (8) and Figure 10.
113
Ion Conductivity and Sensors 700
I
'
I
'
I
'
I
'
I
'
'
I
'
I
'
I
'
I
Ag2S-GeS-Ge /
ay
6O0 [--,
I
500
r
400 r
AgI-As2Se~ _ N a 2 S I B 2 S 3
300
f
200
a s-as s
100
.. 2.0
. 2.1
2.2
2.3
2.4
2.5
2.6
2.7
2.8
2.9
Average Host Coordination Number <no> FIG. 10. Critical fictive temperature To for a number of silver chalcogenide and chalcohalide glasses as a function of the average local coordination number (no) of the host matrix (Bychkov, 2001). The To value for NazS-B2S3 glasses was calculated using conductivity results Patel and Martin (1992). In the latter system, the average coordination number (no) changes with increasing sodium content because NazS additions transform triangular BS3 units into BS4 tetrahedra (Cho, Martin, Meyer, Kim and Torgeson, 2000); an enhanced value, (no) -- 2.47, was therefore used in this case, representing an average over the composition range of interest.
Intuitively, we could not expect any significant change in the short- and intermediaterange order of the glassy matrix in the critical percolation domain, simply because the concentration of the added mobile cations is too small. The underlying structural hypothesis also suggests a random distribution of the mobile cations in the host network, which explains the universal trend of the Haven ratio HR in glasses, discussed in Section 1.3. Small-angle neutron scattering experiments (Bychkov, Price and Lapp, 2001) show a more homogeneous mobile cation distribution in the critical percolation region (ii) compared to the modifier-controlled domain (iii). Taking into account the previously reported and new results, the predominant role of the host network connectivity appears to be completely reasonable. Structurally, infinite percolation clusters, assumed to be a conjugation of the allowed volumes for the mobile cations restricted by their mean-square displacements (Fig. 11 (Bychkov and Price, 2000)), represent a relatively intact host matrix. No preferential conduction pathways are yet formed. The diffusive motion occurs within the slightly modified skeleton and follows its connectivity reflected by the average coordination number (no). This situation changes entirely with further addition of mobile cations.
1.3. UNIVERSAL TREND OF THE HAVEN RATIO AND MODIFIER-CONTROLLED REGIME In the modifier-controlled domain (iii) at x > ~ 10 at.% Ag or Cu, dramatic changes in the ion transport properties are observed. First, the transport parameters depend essentially on the modifier (Ag, Cu) content and not any longer on the microstructural organization of the host matrix (see, for example, Figs. 6 and 12). Second, a significant differentiation in the diffusion coefficient between Ag + and Cu + conducting glasses,
114
E. Bychkov et al.
9
9
9
0
0
~,
9 0
. ,,..~
~a
.,..~
9,-..n
d,
o~
>
,,,.,~
~D
c..~ +,.a
o "~
,-~
Ei o
,
115
Ion Conductivity and Sensors 10-7
'
' '''"I
........
I
'
'
' '''"I
'
' '''"I
'
'
118.8 ~
'
9
10-8
O []
=a
Percolation-Controlled Domain
< 10-9
[] 0
~ 10-1C r~ ~E,2 ~'-'-'''~'-''''~'2
tD= 0.52(5) T ~ " ~
10TM
ModifierControlled Domain
< 10_12 Ag2S_As2S3 to =0.29(4) 10-13
. . . . . . .
I
. . . . . . . .
10-2
I
. . . . . . . .
10-1
I
. . . . . . .
100
I
,
,
101
Silver Concentration (at.%) FIG. 12. Composition dependencies of the l l~ tracer diffusion coefficient DAg for Ag2S-As2S3 (C),[-]) and AgzS-GeS-GeS2 ( O , I ) glasses in the percolation and modifier-controlled domains at 118.8 ~ (Bychkov et al., 1996; Drugov et al., 2000). One should note a decrease of DAg with increasing x for the AgzS-As2S3 glasses in the critical percolation region and an increase of DAg for the AgzS-GeS-GeS2 ones caused by distinctly different values of To (241 (9) and 546(6) K, respectively). In contrast, the values of DAg are very similar in the modifier-controlled domain (iii).
Ag2S- and Ag2Se-containing systems, silver or copper selenide and iodide glass families becomes evident (Figs. 13 and 14) in contrast to chemically invariant ion transport in the critical percolation domain (ii). The two ion transport regimes are also distinguished by the composition dependence of the Haven ratio. 1.3.1.
Haven Ratio and Diffusion Mechanism in Glasses
The Haven ratio HR is a simple experimental parameter easily accessible either from a combined (tracer diffusion D* and ionic conductivity oi) experiment, or from a single electrodiffusion or Chemla measurement Ha -
D* D,~
,
(9)
where D,~ is the diffusion coefficient calculated from oi using the Nernst-Einstein relation Do-
kToi x(ze)2 ,
(10)
where ze is the charge of the mobile ion and x its concentration. In a very simple case (non-interacting charge carriers in ionic crystals), the Haven ratio can be associated with
116
E. Bychkov et al. MI-As2Se 3 (M = Ag, Cu) I
10-8
'
I
'
I
'
I
'
I
'
298 K
I
fi2
10 - 9
~i
10-1~
~ 10-11 "~
AgI-AszSe3 1
"~ 10-13 10_14 10_15 10-16 I
0
,
I
10
,
I
20
,
I
30
,
I
40
,
I
50
MI Concentration (mol. %) FIG. 13. Room-temperature ll~ and 64Cu tracer diffusion coefficients for the MI-As2Se 3 glasses (M = Ag, Cu) (Bychkov et al., 2001). The DAg and Dcu diffusion coefficients are very similar in the critical percolation domain (x < 3 at.% M) but differ by a factor of 104 in the modifier-controlled region (iii).
a Bardeen-Herring tracer correlation factor f (Murch, 1982) reflecting geometrical aspects of successive atomic jumps in the lattice HR ~ f = 1 + 2(/__~1c o s 01,1+j) ,
(11)
where cos 01,1+j is the cosine of the angle between the first and (j + 1)th jumps. The tabulated values off are given in diffusion textbooks, for example in Manning (1968) and Le Claire (1970). Usually, they cover the range between 1 and 1/3, depending only on the lattice and diffusion mechanisms. In systems with interionic interactions, the Haven ratio can be represented by (Murch, 1982) Ha -- hf '
(12)
where j~ is the physical or conductivity correlation factor reflecting ion-ion and defection interactions. Both oxide and chalcogenide glasses exhibit a characteristic composition dependence of the Haven ratio. A typical example for oxide systems is presented in Figure 15, summarizing the experimental data for MzO-GeO2 glasses (M = Na, Rb) taken from Evstrop'ev (1970), Kelly, Cordaro and Tomozawa (1980) and Thomas and Peterson (1984). An extremely dilute glass (75 ppmNa) is characterized by HR =0.99(3). Between 747 ppm and 10 at.% Na, the Haven ratio decreases monotonically and remains
117
Ion Conductivity and Sensors 10
-11
,
,
,
,
,
,
,
,
,
,
,
10 -12 t-,q
E o
o
~
10
-13
l-" o
~ 0 0
10 -14
t-0 ffl
~= 10 D
o 0
1
-15
0-16 298
I
10 -17
I
,
0
I
,
5
I
,
10
,
I
15
K
.
20
i
25
Cu Concentration (at.%)
FIG. 14. Room-temperature 64Cu t r a c e r diffusion isotherms for CuI-As2Se3 and Cu2Se-As2Se3 glassy systems (Bychkov et al., 2001). The diffusion coefficient is nearly identical in the percolation-controlled region and differs by 3 - 5 orders of magnitude at higher x.
M20-GeO 2 Glasses I
'
I
'
I
1.0
'
9 [] 9 9
0.8
I
'
I
'
I
'
I
Na20-GeO 2, Evstrop'ev et al. (1966) Rb20-GeO 2, Evstrop'ev et al. (1963) Na20-GeO 2, Kelly et al. (1980) Na20-GeO 2, Thomas & Peterson (1984)
o
=~ 9 0.6
:~
0.4
0.2
0.0
I
0
,
I
5
,
I
10
,
I
15
,
I
,
20
I
25
,
I
30
M 2 0 Concentration (mol. %) FIc. 15. Composition dependence of the Haven ratio for M20-GeO2 glasses (M = Na, Rb). The experimental data were taken from Evstrop'ev (1970), Kelly et al. (1980) and Thomas and Peterson (1984).
118
E. Bychkov et al.
nearly constant (HR ~ 0.2) at x > 10 at.%. This universal trend has been observed for many different oxide systems (see, for example, Evstrop'ev, 1970; Kelly et al., 1980; Thomas and Peterson, 1984 and references therein). Two approaches were proposed to explain the HR(X) systematics: (A) A multidiffusion model (Haven and Verkerk, 1965), popular in the 1960s and 1970s, was based on a unique value of HR for a given diffusion mechanism and therefore suggests changes in the diffusion mechanism with increasing x (a direct interstitial mechanism at low x, a vacancy-predominant transport at intermediate x, and finally an interstitialcy or other cooperative ion motion at high x). (B) A single diffusion approach (Jain, Peterson and Downing, 1983) introduces ionion and/or defect-ion interactions in the ion and tracer dynamics, while the diffusion mechanism itself is considered to be unique. Neither approach has been verified experimentally. Basically, the same universal trend of HR(X) was found for chalcogenide systems but with two new features (Fig. 16). The Haven ratio decreases only in the critical percolation domain, and this decrease is nearly linear as a function of the reciprocal A g - A g separation distance (rAg_Ag)- 1 calculated on the basis of a homogeneous distribution of silver in the glass network: const H R = 1-
~ . rAg-Ag
(13)
In the modifier-controlled domain, the Haven ratio is rather constant and small, for example HR = 0.32(6) for the silver sulphide glasses, indicating a strongly correlated ion motion.
'
I
'
I
'
I
1.0
0.8 ~Z o 0.6
controlled domain
Modifiercontrolled domain
_a_ ~ , "-
~Z 0.4 "o-
0.2
9 Ag2S-As2S 3 o Ag2S-GeS_GeS 2 I
0.00
0.05
,
I
"r
,
0.10
I
0.15
,
I
0.20
,
-r
I
0.25
1/rAg_Ag (/~-1)
FIG. 16. The Haven ratio for Ag2S-GeS-GeS2 (Bychkov et al., 1996) and Ag2S-As2S3 (Drugov et al., 2000) glasses plotted as a function of reciprocal A g - A g separation distance calculated assuming a homogeneous distribution of silver in the glass network.
Ion Conductivity and Sensors 1.3.2.
119
Structural Changes in the Glass Network with Increasing Mobile Ion Content
Dramatic changes in the ion transport with increasing mobile ion content should be reflected in the structural organization of glasses. Nevertheless, an obvious structural correlation with the short- or intermediate-range order is lacking for silver and copper chalcogenide systems. Neither mobile (Cu +, Ag +) nor network-forming (As, Ge) cation local coordination is crucial for the ion transport. An enormous difference of 4 - 5 orders of magnitude between DAg and Dcu for metal chalcogenide and chalcohalide glasses in the modifier-controlled domain (Fig. 13) coexists with the same trigonal (NM-S(Se) = 3) or tetrahedral M + local coordination in the two glass families (Benmore and Salmon, 1993, 1994; Bychkov, Bolotov, Armand and Ibanez, 1998; Bychkov, Bolotov, Tsegelnik, Grushko and Vlasov, 2001a; Bychkov, Hannon and Lapp, 2001b). On the contrary, very similar a]~ tracer diffusion coefficients were observed for the AgzS-AszS3 (NAs-S = 3 (Bychkov and Price, 2000)) and A g z S - G e S - G e S 2 (NGe-~S,Ge)= 4 (Bychkov et al., 2004)) glassy systems in the modifier-controlled region, in marked contrast to DAg(X, T) in the critical percolation domain (Fig. 12). Intermediate-range order reflected by the first sharp diffraction peak (FSDP) in the structure factor S(Q) is not related to the ion transport characteristics either. The most prominent change in the Faber-Ziman S(Q) for the investigated silver and copper chalcogenide glasses is a drastic decrease of the FSDP at Q1 ~ 1.25 ~ - 1 (As-based systems) or at ~ 1.0 A-1 (Ge-based systems) with increasing x (Fig. 17). Such behavior has been reported for a number of metal chalcogenide glasses (Penfold and Salmon, 1990; Dejus, Susman, Volin, Montague and Price, 1992; Benmore and Salmon, 1993, 1994; Lee, Owens, Pradel, Hannon, Ribes and Elliott, 1996; Bychkov et al., 2001) and represents a characteristic feature of the host network structural evolution on MzX doping. However,
I
'
I
'
..-... o-
." ."
.~
Ag (at.%) 0.12-_~ O~~ o
o~-~ ~ ~ O,.o
9
~% o% ~,,,-i-~ ~
.ooO "++:o 9 9~
.
~+~ 9
9
~
o +
.~d."~
.u.~ .~
~.~o .~r
~o
r.13
I ooO~176176 o
~
~O~o
+, ~_
o o
". 'k'-.#.+o .,.
..~Oo .~-Oo -"
" ++~o -~.~-?~+~+4 ~ %
o
..+ o -..~-+~
o
"v"
,
o,',"
~ o4 .,~"
25 o+...-." o +
.,"
o~ - ~ , . ,
0
1
Ag2S-As2S3 3
2
4
_*
Q (A-1) FIG. 17. The Faber-Ziman structure factor S(Q) at low Q for selected Ag2S-AszS3 glasses with indicated values of the silver concentration (Bychkovand Price, 2000).
120
E. Bychkov et al. I
'
I
'
I
'
I
'
I
'
I
1.2 i
1.0 =
0.8
-t-~AgI-As-~,.Se3
O.6 r~
0.4
Cu2Se-As2Se3
0.2 0.0 I
0
,
I
5
;
I
l0
;
I
,
15
I
20
,
I
25
Metal Concentration (at. %) FIG. 18. Normalized to the host glass area of the first sharp diffraction peak (FSDP) for a number of silver (A,II) and copper (QO) chalcogenide and chalcohalide glasses (Bychkov et al., 2002). A very similar FSDP decrease should be noted in the critical percolation domain (x -< 2-3 at.% M). In contrast, drastically different behavior is observed for the modifier-controlled region.
a striking resemblance of the FSDP disappearance in the modifier-controlled domain was observed for the AgzS-AszS3 and CuzS-AszSe3 systems (Fig. 18), which are characterized by contrasting values of the diffusion coefficient (DAg/Dcu ~ 104 - 105 (Drugov et al., 2000; Bychkov et al., 2001)). Moreover, the FSDP change is different for metal chalcohalide glasses MI-AszSe3 (Bychkov et al., 2001; Bychkov, Price, Hannon and Benmore, 2002). More excitingly, the FSDP decreases in the same manner for all systems in the critical percolation region (this change is relatively small, ~20%); however, there is a significant difference between chalcogenide M z X - A s z X 3 and chalcohalide MI-AszX3 glasses in the modifier-controlled domain (Fig. 18). The FSDP reflects the intermediate-range ordering of the network-forming cations, i.e., A s - A s or G e - G e correlations at a characteristic distance L] ~ 2 7 r / Q ] - 5 - 8 A, confirmed by anomalous X-ray scattering (Fuoss, Eisenberger, Warburton and Bienenstock, 1981; Armand, Ibanez, Philippot, Ma and Raoux, 1992; Zhou, Sayers, Paesler, Boucher-Fabre, Ma and Raoux, 1993) and neutron diffraction with isotopic substitution (Benmore and Salmon, 1993, 1994; Lee et al., 1996; Petri, Salmon and Fischer, 2000; Bychkov et al., 2004). The lack of correlation between the FSDP amplitude and/or position and ion transport parameters means that the intermediate-range order in the host glassy matrix and its change with the mobile cation doping has no obvious influence on the M + ion transport. Finally, the free volume model (Swenson and B6rjesson, 1996), relating very well the conductivity to the network expansion for ternary superionic glasses MY-MzX-AnXm, where M = Ag, Li, Na; Y = C1, Br, I; X = O, S, appears to be inappropriate to the particular case of quasi-binary metal chalcogenide or chalcohalide systems. Figure 19 shows a remarkable increase of DAg by several orders of magnitude at a nearly invariant
Ion Conductivity and Sensors (a)
121
[Va(x)-Va(O)]Na(O) < 0 108
I " " '
'
'
'
I " " '
'
'
'
I " " '
'
'
' q
9
107
<>
106 o
9 9 [] o <>
Ag2S-GeS--GeS 2 Ag-Ge-Sb-Se Ag2S-As2S 3 AgC1-Sb2S 3 AgBr-Sb2S 3 Agi_As2Se3
e~o
< 105 < 104 D
e,
103
[]
#
o
D
102
.s..
10 3
9
<> $=
1000~1 . . . . .
(b)
9
<>
'. . . . . . . . I. . . . . . . . I. . . . . . . . 10-1 10-2 10-3 Network Contraction Wa(x)-Va(0)]/Va(0) > 0 ........ ' ........
109 108
[] 107 106
[]
@ 105
[]
9 9 9
10--4
1
9 AgI-Ag2S-GeS 2 [] AgI-SbaS 3 9 AgI-PbS-AszS 3
9149 9
~o
< 104 103 102 101 100
9 -
10-2
9
.
9
................ 10-1 Network Expansion
100
FIG. 19. Relative diffusion coefficient DAg(X)/DAg(O),where DAg(0 ) is the ll~ tracer diffusion coefficient of the undoped glass (x = 0), versus (a) the network contraction, [Va(x)- Va(O)]/Va(O) < 0, and (b) the network expansion, [Va(x) - Va(O)]/Va(O) > 0, for a number of Ag + conducting chalcogenide and chalcohalide glasses. The cubic scaling, expected in the free-volume model (Swenson and B6rjesson, 1996), is presented in (b) by the straight line. The diffusion data are taken from Dawed (1992), Vlasov, Bychkov, Tsegelnik and Ben-Shaban (1995), Bychkov et al. (1996, 2001), Ben-Shaban (1998) and Drugov et al. (2000).
E. Bychkov et al.
122
average atomic volume Va(x) or, more surprisingly, at a network contraction, [Va(x) Va(O)]/Va(O) < 0, where Va(0) is the average atomic volume of the host glass (x - 0). This exciting structural puzzle can be solved by analyzing the high-resolution real-space correlation functions obtained by Fourier transform of S(Q) over a large range of the scattering vector Q up to 40 A - 1, this Qmax being by a factor of 2.5-3.0 higher than usual reactor or laboratory X-ray diffractometer limit. Pulsed neutrons on a spallation source or hard X-rays on a third generation synchrotron allow these new possibilities to be explored. Typical total correlation functions for metal chalcogenide glasses obtained using time-of-flight neutron diffraction, TN(r), and high-energy X-ray diffraction, Tx(r), are shown in Figure 20, taking as an example the Ag2S-As2S3 glasses (Bychkov et al., 2002). As expected, the TN(r) and Tx(r) real-space functions are similar, but one should note that the neutron and X-ray coherent scattering cross-sections are different for the elements involved. In particular, Ag-related correlations (Ag-S, A g - A g , etc.) are more pronounced in the Tx(r). This difference and combined analysis of the two sets of the data allows more reliable results to be extracted. The first peak in the T(r) at ~ 2.3 ,a, and the second one at ~ 2 . 6 A correspond to A s - S and A g - S first neighbor correlations, respectively. The ~ 2.3 A peak decreases and the ~ 2.6 A peak increases with increasing x, but the local arsenic and silver trigonal coordination remains intact. A special attention should be paid to the third peak at ~ 3.0 ,~, which also increases with x and corresponds to A g - A g second neighbor contacts. Similar peaks were observed in many other Ag-rich chalcogenide and oxide glasses and identified as A g - A g correlations using neutron diffraction with isotopic substitution (Penfold and Salmon, 1990; Dejus et al., 1992; Lee et al., 1996; Bychkov et al., 2004). In the copper selenide systems, Cu2Se-AsSe and Cu2Se-As2Se3, a peak at ~- 2.7/k was found and identified as C u - C u contacts (Benmore and Salmon, 1994). These relatively short M - M correlations indicate direct contacts of MX3 pyramids, which become primordial in the modifier-controlled domain and explain dramatic differences in the ion transport between the critical percolation and modifiercontrolled regimes. A schematic representation of this hypothesis is shown in Figure 21, taking silver sulphide glasses as an example (Bychkov and Price, 2000; Bychkov et al., 2001). Simple geometrical considerations confirm that two edge-sharing ES-AgS3 pyramids with a A g - S first neioghbor distance of 2 . 5 - 2 . 6 / k would have a A g - A g second neighbor distance of 2.9-3.1 A. The proposed Ag2S4 dimer (NAg-Ag --- 1) was assumed to be a primary building block to construct oligomeric structural units (1 -< NAg-Ag < 2), chains or cross-linking chains (2 -< NAg-Ag < 3), sheets, tunnels and other 2D and 3D objects (NAg-Ag--> 3). The A g - A g coordination number appears to be an essential structural parameter for distinguishing between different types of the silver-related network structure. Similar results were found recently for the MI-As2Se3 glasses (Bychkov et al., 2001, 2002). The mobile cation coordination in the two systems is tetrahedral but there are two significant differences at the short- and intermediate-range scale. First, the copper local environment is mixed, consistent with previous EXAFS results at the Cu and Se K-edges (Bychkov et al., 1998). The CuIySe4_ystructural units are evidenced by C u - S e at ~ 2.4 A and C u - I correlations at ~ 2 . 6 A (Fig. 22(a)). The number of the iodine nearest neighbors, 2 -< y -< 3, increases with x but never reaches 4. These results indicate a molecular dispersion of CuI, and copper interaction with neighboring Se atoms coming from the AsSe3 host units. Hence it is not surprising to observe C u - A s second neighbor o
o
123
Ion Conductivity and Sensors
(a)
Ag2S-As2S3 II-
'
I
'
4
I
'
I
'
-: -..--...
2
t
-
t <
1~ ~ ) 2 2 . _ ~ 6 0 1
2
3
~.% ~g
]
4
5
r(~) (b)
Ag2S-As2S 3 12-
A
hv = 80.6 keY
11 10 9 8
[...• 6 5
; ; ;-"
l
4 |
|
|
2 1
0 1
2
3
4
5
r(A) FIG. 20. (a) Neutron TN(r) and (b) X-ray Tx(r) total correlation functions for As2S3 and silver-rich Ag2SAs2S3 glasses (Bychkov et al., 2002). The Ag-Ag correlations at ~ 3.0 A are clearly visible. The Ag-Ag coordination number N A g _ A g - - 2.0-2.5 indicates that the edge-sharing AgS3 pyramids form chains and/or cross-linking chains in the glass structure.
124
--
FIG. 21. Schematic representation of an Ag2S4 dimer formed by two edge-sharing AgS3 pyramids (Bychkov et al., 2001). This structural unit seems to be a primary building block to construct preferential conduction pathways in the modifier-controlled domain of silver chalcogenide glasses, which depending on the system, assumed to be chains, cross-linking chains, sheets, tunnels, etc.
correlations at ~ 3.0 A, and NCu_Se ~ NCu_As. Surprisingly, the C u - C u correlations at 2.7 ,~, characteristic of the copper selenide glasses, are absent in the CuI-As2Se3 system. Recent neutron diffraction results with isotopic substitution (Salmon and Xin, 2002) show, however, their appearance at ~ 4 / k in a similar 0.6CuI-0.4Sb2Se3 glass sample. Such a finding means that the C u I y S e 4 _ y tetrahedra share their comers and form 1D chains in the glass network since Ncu-Cu ~ 2 (Salmon and Xin, 2002). There are no mixed tetrahedral units in the AgI-As2Se3 glassy system. The A g - I correlations only were observed in this particular case at ~ 2 . 8 A (Fig. 22(b)), also in agreement with EXAFS results (Armand, unpublished results). In contrast to CuI-As2Se3, the usual A g - A g correlations appear to be at ~ 3 A indicating the edgesharing of AgI4's. The ES-AgI4 structural units form al: least 2D sheets or tunnels, NAg_Ag = 3.5(5).
1.3.3.
Haven Ratio, Ion Transport and Connectivity of the Mobile Cation Related Network
At least two direct implications of the proposed structural model are as follows: (I) the mobile cation distribution in the glass network is no longer random since rM-M -- 2 . 7 - 4 . 0 A < rM-M -- 5 - 7 A, calculated for a random mobile cation distribution
Ion Conductivity and Sensors (a)
0.5CuI-0.5As2Se 3 2.0
A
-
0"5CuI-0"5As2Se3 ....... scaled expanded As2Se3 - - Difference A T ( r )
]/
1:;I
"
1.5 9
~
125
As-Se I[il 3 " 0 ( 1 ~L i / ~ _ \ ~ / 2 C ~ ( - I )
~'~ "
~"
~
~
1.0 9
0.5
_
,
1.(32 ~ 1 . . : ~ , . \ /
/[
9
Cu-A .1.2(2)s C u - A s -".".. . . . .............. ...
:
.
0.0 2
3 r(A)
(b)
4
5
0.5AgI-0.5As2Se 3 '
I
'
I
2.0 -
As-Se 2.98112
"~" 1.0
'
I
'
0.5AgI--0.5As2Se3 ..... scaled expanded As2Se3 Difference AT(r)
Ag-I
A
i
t 9
0.5
0.0 . . . . . . . . . . . . . . . . . . . . . . ,
1
I
2
,
I
3
,
I
4
,
5
r(/~) FIG. 22. Total correlation functions TN(r) for (a) 0.5CuI-0.5As2Se3 and (b) 0.5AgI-0.5As2Se3 glasses (Bychkov et al., 2001, 2002). The dotted line represents the TN(r) of the scaled expanded AszSe3 host matrix, the filled squares show the difference function AT(r), and the dash-dotted lines the selected first and second neighbor correlations. The coordination numbers are written below the peak identification.
in the modifier-controlled domain; (II) if the mobile cation related network structure forms preferential conduction pathways, as was proposed in the modified random network model for alkali oxide glasses (Greaves, 1985), one can expect certain relationships between ion transport characteristics and specific structural parameters NVI-M and rM-M. More homogeneous distribution of silver in the critical percolation domain compared to modifier-controlled region was verified using small-angle neutron scattering (Bychkov et al., 2001). The differential Ag-related scattering functions AI(Q)/~rcoh, shown in Figure 23, exhibit a remarkable difference between the two domains. There is no
126
E. Bychkov et al. Ag2S-As2S 3 '
7 ~ 3
'
'
'
'
'
I
Glasses from the Modifier-Controlled Domain
31.6
,.Q
'
~2 25 at.% Ag
Glass from the Critical Percolation Domain
1.2 at.% Ag f m
,
,
,
I
0.01
i
I
I
I
I
Q (A_I)
I
I
I
I
I
0.1
FIG. 23. Ag-related difference scattering function AI(Q)/o'coh for homogeneous AgzS-AszS3 glasses from the critical percolation domain (1.2 at.% Ag) and modifier-controlled region (25 and 31.6 at.% Ag) (Bychkov et al., 2001).
excessive scattering for the x - 1.2 at.% Ag sample even at the lowest scattering vectors Q. In contrast, the AI(Q)/~rco h functions for the Ag-rich samples increase systematically with decreasing Q _< 0.03/k and increasing silver content. More quantitative estimates need to be obtained, but qualitatively the random silver distribution in the critical percolation region and Ag association and/or clustering in the modifiercontrolled domain are confirmed. One should emphasize that no short A g - A g correlations were found for Ag-diluted A g - A s - X glasses (x = 4 at.% Ag) using neutron diffraction with isotopic substitution (Salmon and Liu, 1996). The universal trend of the Haven ratio in glasses and silver chalcogenide systems in particular can be explained by using a single diffusion approach (Jain et al., 1983) and different Ag distributions in the two domains. In the critical percolation region, the random silver distribution means that the average A g - A g separation distance decreases with increasing x. Accordingly, interionic interactions become increasingly strong, leading to a monotonic decrease of HR. In the modifier-controlled domain, the interionic interactions are controlled by an invariant A g - A g second neighbor distance at ~ 3 A, and HR appears to be constant. Additionally, the low HR value, 20.32 for the silver sulphide glasses and 2 0 . 2 for their selenide counterparts are in excellent agreement with the short A g - A g correlations and hence, strong interionic interactions. A heavily correlated cooperative ion motion therefore has a clear structural explanation. Compelling evidence in favor of the proposed scenario was given by the Haven ratio itself: the intersection of the two HR branches in Figure 16 at HR ~ 0.32 gives an A g - A g separation distance of 3.5(4)/~,! In other words, the interionic interactions corresponding to the modifier-controlled domain are induced by ion-ion separations that are very similar to those :found in a direct diffraction
127
Ion Conductivity and Sensors
experiment. This result suggests that Ag-related network structure is indeed responsible for the change in an ultimate diffusion parameter, HR, and thus inevitably has to form preferential conduction pathways in the modifier-controlled region (point (II) above). Assumption (II) was also be validated by other examples. An important difference between silver-rich A g - A s - S and A g - A s - S e glasses is the value of NAg_Ag. Edgesharing ES-AgS3 pyramids form chains or cross-linking chains in the silver sulphide glasses (NAg_Ag- 2.0-2.5). In contrast, their silver selenide counterparts are characterized by NAg_Ag= 2.7-3.1. A higher connectivity of the Ag-related subnetwork giving rise to quasi-2D conduction pathways in the latter case suggest an enhanced mobility and diffusivity of the Ag + ions for the silver selenide glassy system. The al~ tracer diffusion data (Kazakova, 1980; Zhabrev and Kazakova, 1982; Vlasov et al., 1987; Drugov et al., 2000) are remarkably consistent with this suggestion (Fig. 24).
10-9
10-10
10_11
~ 10 -12
NAg_Ag -- 2.0--2.5
O
O .,..~
.,.,~
~
10-13
10 -14
/ / f
~" As-Se .
10 -15
10_16
--
.0
0 8 1 298 K
I 10
i
I 15
J
I i I i I 20 25 30 Metal Concentration (at. %)
i
I 35
FIc. 24. Room-temperature l l~ and 64Cu tracer diffusion coefficients in the modifier-controlled domain for a number of silver and copper chalcogenide glasses (Bychkov et al., 2001). The largest DM > 10 - l ~ cm e s -1 were observed for the A g - A s - S e glasses with the highest connectivity of the Ag-related network (NAg-Ag = 2.7--3.1), see text for details.
128
E. Bychkov et al. I
'
I
'
I
298K
10-1~
Ag2S-As2S3
~
110mAg~r
iF
8 10-11
10-12
64Cutracer
I
,
I
,
I
10
20 30 MobileCationConcentration(at.%) Fro. 25. Tracerdiffusion coefficientsfor the Ag+ and Cu+ conducting glasses with a chain-like connectivity of the cation-related structuralunits (ES-AgS3 for Ag2S-AszS3and CS-CuIySe4_yfor CuI-As2Se3) (Bychkov et al., 2002).
At x = 15 at.% Ag, the difference in DAg is about two orders of magnitude. The same scenario seems to be applicable to copper selenide glasses, which are characterized by a value of Dcu 4 - 5 orders of magnitude lower than that of DAg in corresponding silver chalcogenide families (Vlasov et al., 1987; Bychkov et al., 2001). A much lower connectivity of the Cu-related network is the most probable reason. In both the Cu2SeAsSe (Benmore and Salmon, 1994) and CuzSe-AszSe3 (Bychkov et al., 2001) systems, the ES-CuSe3 pyramids and/or CuSe4 tetrahedra with a short C u - C u second neighbor distance of ~ 2.7 A form predominantly dimers (Ncu-Cu - 0 . 8 - 1 . 0 ) . As a result, the mobility and diffusivity of the Cu + ions appears to be small. More excitingly, the diffusion coefficients in glasses with similar connectivity, i.e., Ag2S-AszS3 and CuI-AszSe3, appear to be similar as well (Fig. 25), in marked contrast to the usual difference of 4 - 5 orders of magnitude between the Ag + and Cu + conducting glasses. One should also note that an important role of the connectivity of the cation coordination polyhedra was found recently in the molecular dynamics simulations of alkali ion transport in oxide glasses (Cormack, Du and Zeitler, 2004). Recent combined neutron diffraction experiments with isotopic substitution, highenergy X-ray scattering measurements and reverse Monte Carlo modeling carried out for A g z S - G e S - G e S 2 glasses containing from 5 to 30 at.% Ag show the process of preferential conduction pathways formation in more detail (Bychkov et al., 2004).
Ion Conductivity and Sensors
129
FIG. 26. Characteristic snapshots of the structure of Ag2S-GeS-GeS2 glasses containing (a) 5.9 at.% Ag, (a) 8.3 at.% Ag, (a) 13.1 at.% Ag, and (a) 30.8 at.% Ag (Bychkov et al., 2004). Reverse Monte Carlo (RMC) modeling of neutron diffraction with isotopic substitution and high-energy X-ray scattering results was used to reproduce the formation of preferential conduction pathways in the glass structure, formed by edge- and cornersharing AgS3 pyramids. A 20 ,~ box was taken from the center of the RMC configuration. Ag, Ge and S are represented by black, red and yellow spheres, respectively; A g - A g correlations in the range 0 - 5 ,~ are also shown.
Characteristic snapshots of the glass structure are presented in Figure 26(a-d) visualizing the transition from mostly isolated Ag atoms in the low-concentration limit to a highly connected Ag-related sub-network at high silver content. 1.4.
CONCLUSIONS
Tracer diffusion and conductivity measurements for silver and copper chalcogenide and chalcohalide glasses over an extremely large composition range covering in some cases five orders of magnitude in the mobile ion content, from 4 ppm to 35 at.%, allow three
130
E. Bychkov et al.
FIG. 26. Continued.
distinctly different ion transport regimes to be distinguished: (i) below the percolation threshold at Xc ~ 30 ppm, the glasses are electronic insulators; (ii) in the critical percolation region atXc < x -< 1 - 3 at.% M +, the ionic conductivity and diffusion depends almost entirely on the connectivity of the host matrix, reflected by the average host coordination number (no): the nature of the mobile cations, ,;hemical form of the dopant or of the host matrix does not play any important role. The mobile cation distribution is random, the M + - M + separation distance decreases with increasing x, and no preferential conduction pathways are still formed; (iii) In the modifier-controlled domain at x > ~ 10 at.% M +, the connectivity of the cation-related structural units MY z, where Y = chalcogen and/or halide, z = 3 - 4 , evidenced by the short invariant M - M second neighbor correlations and reflected by the M - M coordination number, appears to be predominant. Highly connected edge- or corner-sharing MYz units, which form 1D chains, 2D sheets or tunnels, etc., serving as preferential conduction pathways, ensure high mobility of the M + ions.
131
Ion Conductivity and Sensors
2. 2.1.
Chalcohalide Glasses Based on Compounds of the Third Group Metals INTRODUCTION
The reason of selection of the indicated group of materials in the special partition is the peculiar role of complex-formation ability of metals of the third group in the formation of structure and properties of glasses containing halides and other chemical elements. Here is the sense of the title: "based on chalcogenide compounds of metals of the third group," in spite of that the content of these compounds in glasses does not usually exceed 20%, and they are not glass forming. At the same time the meaning of investigations of interaction of chalcogenides of metals of the third group with the other components of chalcogenide glasses is important not only for this chapter of the book, but as it will be seen from Chapter 4 in Volume 3 (Chalcogenide glasses doped with rare earth), these compounds also play a special role in the formation of the structure of glasses, doped with lanthanide. The development of chalcogenide glasses containing high amount of metal halides gives a possibility to create new solid electrolytes and optic materials with wide range of transparency.
2.2.
GLASS-FORMATIONRECIONS
Figure 27(a) demonstrates the glass formation regions obtained for the systems containing lithium, sodium, potassium, and cesium as well as thallium chlorides (Tver'yanovich, Nedoshovenko, Aleksandrov, Turkina, Tver'yanovich and Sokolov, 1996b). The boundaries of glass formation region in the rubidium chloride system lie between those in the systems with potassium and cesium chlorides. It is interesting that the concentration boundaries of all the glass-formation regions are similar in shape: the boundaries are extended from the glass formation region of the quasi-binary GeSz-GazS3 system toward the MGaS3/2C1 composition and the concentration of C1 does not significantly exceed the concentration of Ga.
(a)
Ga2S3
GeS2
(b)
MCI
GeS2
7 ..... Ga2S 3 20
40
60
80 AgHal
FIG. 27. (a) Glass-formationregions for the systems GeS2-Ga2S3-MC1; M = Li (1), Na (2), K (3), Cs (4), Ag (5), T1 (6). (b) The glass-formationregions for the systems with GeSz-Ga2S3 with AgC1(1), AgBr (2) and AgI (3).
132
E. Bychkov et al.
(a)
(b)
GeS2
GeS2
7w-J. 7I
,.)-, ~,.,. ,..zI
~0
\
7
.
Ga2S 3 20
.
.
4o
v
.
60
8o MnCI 2(PbCI2 )
Ga2S 320
v
40
v
60
k2o Y X
ao NaF(PbF:z )
FIG. 28. (a) The glass-formation regions with MnC12 (1) and PbC12 (2). (b) The glass-formation region of the system with NaF (1) and PbF2 (2).
The glass-formation regions for the systems with AgC1, AgBr and AgI are demonstrated in Figure 27(b). All these regions are similar in shape, but the chemical structures of glasses differ from each other dramatically because of the exchange reactions (see later), which are characteristic of the system wJ.th AgC1 and are not for the system with AgI. The glass formation region for the system with copper chloride is not depicted, because the glasses could be manufactured only when the CuC1 content was less than 5 mol% and when the quenching was performed into water, by pouring the melt onto the tube walls. This miserable glass-formation ability of alloys is also the result of exchange reactions and of formation of CuzS. The glass-formation regions with MnCI2 and FeCI2 have similar shape (Fig. 28(a)). The maximum content of C1 does not exceed the content of Ga. It is not true for glasses with PbCI2 (Fig. 28(a)). But the alloys with concentration of' PbCI2 more than of GazS3 are not homogeneous; they consist of red glass and oil-like yellow liquid. The glasses GazS3-GeSz-MeFn (where Me = Na, T1, Pb, Mn, Bi) were also elaborated (Orkina and Baidakov, 1993). The glass-formation region of the system with NaF has the same shape as the glass-formation region of the system with NaC1 (Fig. 28(b)). The glass-formation region of the system with PbF2 (Fig. 28(b)) has a slightly different shape. Glasses with other fluorides were elaborated for the following system (1 - x ) ( G a z S 3 ) a ( G e S z ) l - a - xMeFn where a = 0.20, 0.25. The maximum values of x are: 0.30 for T1F; 0.25 for MnF2; 0.35 for BiF3. Apart from the above-mentioned, there exist many other glass-formation regions which can be obtained by replacement of any element by their analogues. For example, it is possible to replace S by Se or Te. That is how the pale-yellow glass CsGaSe3/2C1 was prepared. But as a rule at such replacement it is necessary to replace C1 by its heavy analogues. Otherwise the possibility of obtaining a melt with immiscibility of components exists. The different glasses were synthesized with replacement of Ga by its analogues, with replacement of germanium by silicon. The glasses with halides of metals of the second group were also synthesized. 2.3.
PROPERTIES OF GLASSES
The synthesis of glasses with a small halide content as well as of glasses in the quasibinary GeS2-GazS3 system is complicated because of the considerable explosion hazard upon the interaction of components at 800-900 ~ The viscous melts are poorly
Ion Conductivity and Sensors
133
homogenized even at the maximum temperature of synthesis. Furthermore, for the glass to be obtained, higher cooling rates are necessary, even to the point of quenching into water. The melts with high halide content (more than 15-20mo1%) are most readily synthesized. In this case the interaction between the batch components is not accompanied by an abrupt heat evolution. This greatly decreases explosion hazard. The melts are more mobile and easily homogenized. If the melt composition does not lie at the boundary of the glass formation region, a high cooling rate is not required to obtain the glass; it may even be down to the point of the furnace-cooled regime. Glasses lose their moisture resistance with increase of chlorides (bromides or iodides) content. At maximum chloride concentrations, the glass ingots begin to crack after the quartz tubes are opened up. The moisture resistance of glasses with fluorides is significantly higher. The density of glasses (1 - x - y)GeS2 - xGazS3 - yLiC1 and (1 - x - y)GeS2 xGazS3 - yNaC1 is a linear function of their composition. These dependencies can be described by the following equations (Borisova et al., 1991): For the first system p = 2.84(1 - x - y) + 3.46x + 2.19y for the second system p = 2.84(1 - x - y) + 3.46x + 2.26y. The rise of partial input of NaC1 compared to LiC1 can be explained by the rise of an average atom weight. On the contrary, the rising of GazS3 average atom weight comparing with GeS2 average atom weight is not enough for the explanation of its partial inputs deference. The reason of that is the difference of structural units formed by Ge (GeS4/2) and by Ga (GAS3/2S1/3) and, as a result, the difference of average coordination numbers (2.7 and 3.2 correspondingly). It is known that the formation of additional chemical interactions in a system results usually in negative deviations from linear dependence of molar volume as functions of molar composition. Thus the maximum deviation should be near the composition, for which the relation of the molar parts of interacting compounds corresponds to stoichiometry of chemical interaction. If we assume that for the investigated glasses [(GeSz)l_a(GazS3)a]l_x(MFn)x where M = Na, Pb, Bi complex structural unites M [GaS3/zF],, are formed, the maximum deviation of molar volume from linear dependence should be at x0 -- 2a/(2a + n) (they are indicated by arrows in Fig. 29). For the systems with PbF2 and BiF3 the glasses with relation F/Ga more than 1 can be elaborated. Therefore, the discussed phenomenon can be observed (Aksenov, Gutenev, Orkina and Makarov, 1992). Tg is reduced at the introduction of metal chlorides in composition of glasses GazS3GeS2. So the introduction of 15-20 mol% of LiC1 or 2 5 - 3 0 mol% of NaC1 reduces the softening point on 100 ~ Simultaneously the difference between the softening temperature and crystallization temperature increases and achieves the magnitude of 150-200 ~. The fluorides of metals influence the characteristic temperatures of much smaller glasses. Furthermore, in the opposite chlorides the fluorides increment the microhardness of glasses from 180 up to 250 kg mm -2 (Gutenev, Orkina and Baidakov, 1990; Orkina and Baidakov, 1993).
134
E. B y c h k o v et al.
O,x,,,
V, cm3/mol :55
O, 50 9
'~
~O~O
:50
45
50
~~A~--~A
I 45
'~,
45
Vl~E! 50
45 i
i
0.0
0.1
i
i
0.2
0.3
0.4
PbF 2, mol. part V, cm3/mol 50 5:5
5c)
55
|174
50
I
0.0
45
~
'
~
I
0.1
6
'
I
'
0.2
I
0.3
'
0.4
MF n, mol.part FIG. 29. Molar volume as a function of fluoride molar part for glasses (1 - x)[ (1 - a)GeS2.aGa2S3] - xMF n (Gutenev et al., 1990): (1) a -- 0.15; M = Pb. (2) a = 0.2; M = Pb. (3) a = 0.33" M = Pb. (4) a = 0.4; M -- Pb. (5) a -- 0.2; M -- Na. (6) a = 0.2; M = Bi. Dotted arrows mark the compositions with equal concentrations of Ga and F.
135
Ion Conductivity and Sensors
The glasses that contain chlorides in small amounts are yellow. As the content of alkali metal chlorides increases, the color becomes lighter and then completely disappears for CsGaS3/2C1. The glass CsGaSe3/2C1 is slightly yellow. The blue transmission edge shifts toward the near UV region and becomes 0.35/xm for CsGaS3/2C1. The red transmission edge does not shift toward the long-wavelength range and is about 12 ~m for sulphide glasses and 14 ~m for selenide glasses with MC1. The red transmission edge for glasses containing MF,, is about 8/~m (Orkina and Baidakov, 1993). The refraction index changes from 1.8 for glasses enriched by MC1 up to 2.1 for glasses GeSz-GazS3.
2.4.
DIAGRAMS OF STATE
Let us discuss the interaction of the components in the systems containing chlorides of alkali metals. The interaction between components in the ternary GeSz-GazS3-NaC1 system exhibits the eutectic nature (Fig. 30) (Nedoshovenko, Turkina, Tver'yanovich and Borisova, 1986). The triple eutectic corresponds to the composition: 40 mol% NaC1, 25 tool% GazS3, 35 mol% GeS2 with melting point 580 ~ So, glass-forming region is situated along the line of simultaneous crystallization of GazS3 and GeS2 up to the field of NaC1 primary crystallization. Quasi-binary system GeSz-NaC1 is also eutectic with eutectic point 794 ~ and less than 2 mol% GeS2. But it has immiscibility region located between 2 and 38 mol% GeS2 with the temperature of monotectic reaction 800 ~ Quasibinary system GazS3-NaC1 is eutectic with eutectic point 708 ~ and 60 mol% NaC1. Quasi-binary system GazS3-LiC1 (Fig. 31(a)) is eutectic with eutectic point 595 ~ situated near LiC1. It has immiscibility region between 42 and 95 mol% LiC1 and with the temperature of monotectic reaction 880 ~ (Tver'yanovich et al., 1996b). Liquidus for the GazS3-KC1 and GazS3-CsC1 systems is like the one for the eutectic system. The alloys with the lowest liquidus temperatures are placed near composition GeS
Ga 2 S3
2
el
NaCI
FIG. 30. The interaction of components in the ternary system GeS2-Ga2S3-NaC1. Letters mark temperature (~ (Nedoshovenko et al., 1986).
136
E. Bychkov et al. (b) 1100
1000
\
\ \
900
\
E-"
,,,
800
"
7oo-5
(a) I
o
n
600
i.......... t
1 1 0~ 0 ~
Ga2S3
20
,i0
6'0
80
MnCI 2
LI+L2 ~ ~)
900 -
~f
~.,
I
700-
ii Ga2S3
I
iJ 20
I 40
~ 60
~ 80
8oo ..~
~"
, I
/
\
700
7=, n e
600 .
LiCI
\
GeS2
20
.
.
40
.
.
60
.
80
MnCI 2
FIG. 31. (a) Quasi-binary system Ga2Sa-LiC1 (Tver'yanovich et al., 1996b). (b) Diagrams of states for the systems Ga2Sa-MnC12 and GeS2-MnC12 (Baidakova et al., 1991).
with 67 mol% of MC1 (Tver'yanovich et al., 1996b). The lowest liquidus temperature is equal to 605 ~ for the system with KC1 and 535 ~ for the system with CsC1. Diagrams of states for the systems with MnC12 (GazS3-MnCI2 and GeSz-MnCI2, Fig. 31 (b)), are similar with the ones for the systems with LiC1 (the expansion of glassformation regions for the corresponding ternary systems is also similar) (Baidakova, Dvinova, Tver'yanovich and Turkina, 1991). Ga2S3-MnCI2 is eutectic system with eutectic point at 97 mol% of MnCI2 and 625 ~ The liquation region is absent but liquidus has inflection near 40 mol% ofMnC12. GeSz-MnC12 is eutectic system with eutectic point at 95 mol% of MnCl2 and 624 ~ The liquation region stretches from 12 up to 90 mol% of MnCI2. The alloys with less than 12 mol% MnCI2 can be elaborated in glassy state.
2.5. SYSTEMSWITH EXCHANGE REACTIONS The existence of exchange reactions of chloride compound with GeS2 is the peculiarity of the systems Ga2S3-GeSz-AgCI(PbCI2). The presence of GeC14 in glasses with AgC1 was established using Raman spectroscopy (Kuznetsov, Mikhailov, Pecheritsyn and Turkina, 1997). A similar clarification cannot be made in the case of glasses with AgBr and AgI. The investigation of exchange reactions was performed with the help of X-ray and thermal analysis of the crystalline phases produced during s,low cooling of the melt.
Ion Conductivity and Sensors
137
It was understood that as a result of the exchange reaction in the quaternary reciprocal system GazS3-GeSz-AgC1 liquid GeC14 appears. This liquid compound at feeble heat can be observed in quartz ampoules, together with other crystalline compounds, visually. Silver forms the crystalline ternary compound AgGaS2. The results of the thermodynamic estimation of the probability of the reactions GeS2 + 4AgHal -- GeHal4 + 2AgzS,
(14)
GazS 3 + 6AgHal = 2GaHal 3 + 3AgzS ,
(15)
shows that the exchange reactions of type (14) and/or (15) are most probable in AgC1 containing glasses and unlikely in case of AgI containing glasses (Kuznetsov et al., 1997). More complicated interaction of components is realized in the system GeSz-GazS3PbCI2. The following compounds were observed in different alloys of this system: GeS2, GazS3, PbCI2, PbS, GeCI4, PbGeS13, PbzGeS4 (Mikhailov, Tver'yanovich and Turkina, 1993).
2.6.
STRUCTURAL INVESTIGATIONS
The concentration dependencies of K~ chemical shift for Ga and Ge were measured using X-ray emission spectroscopy for [(GeSz)0.s(GazS3)0.z]a-x(NaF)x glasses (Aksenov et al., 1992). As it was mentioned above, Ga in glasses of the system GazS3-GeS2 is fourfold coordinated by sulphur. The replacement of sulphur by more electronegative fluorine should result in the rise of chemical shift. In fact, the chemical shift for K~ lines of Ga and Ge grows at magnification of the fluorine content. And this effect is much stronger for Ga atoms (Fig. 32). In the Raman spectrum of glassy GexSl-x (GeS2 glass for example, see Fig. 33) the peak at 258 cm-1 has been ascribed to the stretching of the direct G e - G e bond in the S
\
/
S
subunits of the S-Ge-Ge-S type (e.g., Ge2S6/2) (Lucovsky et al., 1974). In glassy / \ s
s
stoichiometric composition these homopolar bonds exist due to concentration fluctuations. Their content is relatively low and one depends on the glass-preparing regime. Certainly this peak is pronounced in glassy GexSl-x at x > 0.33 (i.e., where there is a chalcogen deficit relative to stoichiometric GeS2), at x = 0.37 for example (Lucovsky et al., 1974). The existence of G e - G e homopolar bonds in G e - S glasses was also confirmed by Feltz using ESCA (Feltz, Voigt, Burckhardt, Senf and Leonhardt, 1976). We propose to assign the 268cm -1 line in Raman spectra of glasses containing gallium sulphide (see Fig. 34, x - - 0 . 2 ) to the stretching of S
S
s
s
\ / the G a - G a bond in the complex subunit S-Ga-Ga-S (Ga2S6/2) in an analogous / \
138
E. Bychkov et al.
0.20
O
~9 o.15 0.10
J
0.05 0
0.00 --0.05 -2
2
2
4 6 F, wt.%
8
1'0
FIG. 32. Relativechemical shift 6 of (1) Ga KoL 1 line and (2) Ge KoL 1 line as a function of fluorine content for glasses 0.8GeSz-0.2GazS3-MFn. The arrow marks 6 for crystalline GazS3 and GeS2 (Aksenov et al., 1992). way. 2 A slightly higher frequency (by 3.8%) would be in accordance with the somewhat lighter mass of Ga compared to Ge (by 4.2%), so that 2.1% frequency shift is expected (to 263 cm-1). Small discrepancy can be due to the difference between the force constants. We have some additional reasons for such an assumption. Besides, the peak intensity depends on the ratio of Ga and S components as it is in the case of G e - S system. The increase of Ga contents relative to S results in the increase of the band intensity (Fontana, Rosi, Ivanova and Kirov, 1988; Barnier, Palazzi, Massot and Julien, 1990). In the gallium sesquisulfide-germanium disulfide-metal chloride glass systems, when the ratio of G a - X (X is chalcogen or halogen) components amounts to 1 "4, i.e., the possibility of fourfold coordination of gallium by chalcogen/haIogen takes place (GaX4), the peak disappears at 268 cm-1 (with the exception of the system with AgC1, Fig. 34). The disappearance of the above-mentioned mode of the Ga2S6/2 bond with the increasing T1C1 content (see in Fig. 35), can probably be explained by the formation of a complex \ anion: instead of Ga2S6/2 homopolar bonds, -Ga-C1-1 (GaS3/2C1) heteropolar bonds appear. / This statement is in agreement with the set of spectra shown in Figure 36, where the ratio of GeS2 to GazS3 is changed at approximately constant T1C1 content. Here one can see shift intensity from the band at 340 cm-1 (ratio 4" 1) to the position around 320 cm-1 (ratio 1.88 91). Summarizing both the observations we may conclude that, when C1-Ga bonds become possible due to the addition of metal chloride to GazS3-containing glass, fourfold coordination of Ga by sulphur and chlorine develops, and the breathing mode of such tetrahedra is about 320 cm-1. The idea of mixed halogen-chalcogen environment of Ga in the glassy state was also under consideration (Berg and Bjerrum, 1983; Le Toullec, Christensen, Lucas and Berg, 1987). The formation of such tetrahedra assists the glass formation. Therefore there is a possibility of glass formation even without GeS2 glass-forming matrix. Indeed we have found wide glass-forming regions in the GazS3-MC1 systems (Tve,r' yanovich et al., 1996b). 2 At any rate two sulfur atoms of this structure unit are threefold coordinated (Ga284/282/3).
139
Ion Conductivity and Sensors
(1-x)GeS z
.,-.
(x)Ga2S 3
-
~z clz
x=O x=O.2 I
-108,3
I
-2_.10.5
I
I
-414.9
-312.7
-517,1
Frequency s h i f t / c m -1 FIG. 33.
Raman spectra of glasses (1 - x)GeS2-(x)Ga2S3, where x -- 0, 0.2 (Tverjanovich et al., 1996a).
According to the previously mentioned facts, one would argue that the effects of the interaction of GazS3with T1C1 are similar both with and without the GeS2component. It is obvious from the example of T1GaS1.sC1 glass. The Raman spectrum shown in Figure 37 can be explained satisfactorily within the framework of the above-described model (superposition of the contributions from pseudo-molecular subunits G a 2 8 6 / 2 and
yGeS 2 - G a 2 S3 - xAgCI
j
/
)
!
y=400
"k
-'-, y=39o "L y=3.6_8 '~ x = 0 . 5 7
-2
-102
-202
-302
-402
-502
Frequency shift/cm -l FIG. 34.
Raman spectra ofyGeS2-Ga2S3-xAgC1 glasses (Tverjanovich et al., 1996a).
E. Bychkov et al.
140
4GeS2 - Ga2 ~ - xTICI
o
-1
-101
FIG. 35.
-201 -301 Frequency shift/cm -1
-401
-501
Raman spectra of 4GeS2-Ga2S3-xT1C1 glasses (Tverjanovich et al., 1996a).
y=4.00
-~
=.
~9
y=2.75
o
-2
FIG. 36.
-102
-202 -302 Frequency shift/cm -1
-402
-502
Raman spectra of yGeSa-Ga2S3-xT1C1 glasses (Tverjanovich et al., 1996a).
Ion Conductivity and Sensors
141
MGaS1.5 CI
.,..,
~
rm
I -11
-111
I -211
I -al 1
.....M=TI
I -411
I -511
Frequency shift/cm-1 FIG. 37. Ramanspectra of glassy MGaS15C1, where M = T1, Rb or Cs (Tverjanovich et al., 1996a).
GaS3/2C1). Both our Raman data of glassy T1GaS1.sC1 and the data for [GaC14]- (Hesler, 1971; Berg and Bjerrum, 1983) are represented in Table II. The exact agreement is obtained for both sets except for the discussed peak at 268 cm -1. GeS2 favors for formation of stable GaS3/2 C1-1 tetrahedrons, which correspond to composition 1GazS32T1C1. So, SzGa-GaS2 structural units (268 cm -1) are absent. These structural units appear when GeS2 is absent in composition of glass (Fig. 37). As said before, the increase of the metal ion radius at substitution of T1 by Cs in the glassy MGaS1.sC1 results in radical changes of Raman spectra. Table III presents the experimental Raman frequencies of CsGaS1.sC1 glass together with the known data and the assignments for GaC13 melt (Balls, Downs, Greenwood and Stranghan, 1966; Beattie, Gilson and Cocking, 1967) (these assignments are based on normal vibrations analysis of bridged X2Y6 molecules (Bell and Longuet-Higgins, 1945)). The observed bands are
TABLE II THERMODYNAMICESTIMATIONOF AG AT I2OO K FOR EXCHANGE REACTIONSOF TYPES (I4) AND (I5) Halide
AgC1 AgBr AgI
AG1200 K (kJ)
GeS2
Ga2S3
- 62 19 129
90 289 344
E. Bychkov et al.
142
T A B L E III
RAMAN FREQUENCIES (IN CM-X)
melt (Berg and Bjerrum, 1983)
T1GaS 1.5C1 glass (Tverjanovich et al., 1996a; Tver'yanovich et al., 2004)
115
120
--~ 107
122
153
153
--~ 174
172
346 380
343 370
--~ 271 --- 316 --~ 384
315 381
[GaC14] - l
melt (Hesler, 1971)
CsGaC14
C1
\
mostly assigned to the modes of the Ga2C16 dimer
/
C1
Ga /
C1
Ga . It is really striking that
\
C1
\ /
CsGaS l.sC1 melt, 650 ~ (Tverjanovich et al., 1996a; Tver'yanovich et al., 2004)
/\
C1
C1
all the Raman bands of CsGaS3/2C1 glass except the 2 1 5 - 2 3 0 cm-1 region correspond precisely to the band of GaC13 melt, i.e., to the vibration :frequencies of dimer Ga2C16. \
C1
/
Most modes of the dimer are assigned to the Ga /
\
/
Ga ring, which has two edge-shared
\
/\
C1
tetrahedra. Hence we may suppose (taking into account the similarity of S and C1 atom weights) that in the Ga2S3-CsC1 glass, tetrahedral GaS3/2C1 subunits are linked into chains by sharing common edges (Fig. 38(c)). The assumption about the chain-like structure in the system Ga2S3-Na2S was proposed in Barnier et al. (1990). As the temperature is raised up to 650 ~ (TL -- 550 ~ the Raman spectrum of CsGaS3/2C1 changes dramatically especially between 270 and 400 ~ (see Fig. 39). We think that in accordance with the proposed chain model the chains are destroyed and at about 500 ~ the spectrum is much more similar to the case of Tl-containing glass for which we propose the structure formed by the corner connected tetrahedra. However, in contrast to the glasses containing T1C1, the band at 268 cm -1 appears to be absent (unless it is drowned by very broad main peak at --~ 320 c m - 1). It points to the possibility that each Ga atom is saturated by the fourfold coordination of S plus C1. As it can be seen from the Raman data for GaC14~- (Hesler, 1971; Berg and Bjerrum, 1983) in Table II, there appears the same vibration energies as in the high-temperature spectrum of CsGaS1.sC1. So, it is reasonable to suppose that the structure of glass CsGaS3/2C1 is formed by polymer chains consisting of double edge-sharing tetrahedra [Ga2S2S2/2C12] 2-. C1
\
/
S
Ga \
/
S
S
\/
\
Ga
\
/\
S
C1
The squares formed by two atoms of Ga and two atoms of S (Ga2/8S2/2) are strong structural units linked into chain by soft, hinged sulphuric bridge. As a result,
Ion Conductivity and Sensors
,a,
143
@
(e
FIG. 38. Models of structural units proposed for Ga2S3-MC1 (M - metal, not depicted) glasses.
the deviations in both bond angles and bond lengths between the atoms forming these squares are negligible for all links of polymer chains. It leads to the small half-width of the Raman bands. A similar result can be obtained from the comparison of Raman spectra CsGaSe3/2C1 glass (Fig. 40, curve 3) and GaBr3 melt (Balls et al., 1966). These spectra are also
E. Bychkov et al.
144
CsGasl.sC1
Temperature
650 Oc
600
550 oo
500
c~ t..,
450
+.~
320
270
200
I
-11
-111
I
-211
I
-311
I
-411
I
-1 -511 cm
FIG. 39. Raman spectra of CsGaS~.sC1 in the glassy and liquid states al: the different temperatures from 20 to 650 ~ (Tverjanovich et al., 1996a).
145
Ion Conductivity and Sensors
similar (Table III). Here, as in the previous case, the atom weight of chalcogen and halogen are nearly equal (78.96 for Se and 79.904 for Br). But it is surprising that the difference of C1 atom weight (one atom per tetrahedron) and Se atom weight does not lead to principal deviations of CsGaSe3/2C1 spectrum from GaBr3 spectrum. Perhaps, it happens because the spectrum is defined mainly by vibrations of the discussed square (GaSe2/2Ga or GaBr2/2Ga) and also because the frequency of the breathing mode of the mixed tetrahedron GaSe3/2C1 (234 cm-1 as it will be discussed later) coincides with one of the vibration frequencies of the square GaSe2/2Ga (see Table III). We tried to simulate the Raman spectrum of CsGaSe3/2C1-CsGaS3/2C1 glass by mixing the spectra of CsGaSe3/2C1 and of CsGaS3/2C1. From comparison of the simulated spectrum with the experimentally measured spectrum (Fig. 40) we can conclude that the additional bands at 225-265 cm-~ correspond to vibration frequencies of mixed structural units, containing sulphur and selenium together. The substitution of C1 with Br or I leads to moderate transformations of the spectra (Fig. 41, curves 1, 2, 3, respectively). The basic shape of the spectra is invariable because in all the cases the main vibration unit is square Ga2/8S2/2. To find the influence of halide atoms substitution on Raman spectra we calculated the vibration frequencies of the breathing mode of the tetrahedrons GaX4 where X4 is, for example, 3 S and 1 Br (or I), 2 S and 2 Br (or I) and so on. The vibration breathing modes of tetrahedral GaBr4 and GaI4 correspond to Raman bands 210 and 145 cm -~,
d ~ r~ ~D
100
200 300 Frequency shift/cml
400
FIG. 40. Raman spectra of CsGaS3/2C1 (1), 0.45CsGaSe3/2C1-0.55CsGaS3/2C1 (2) and CsGaSe3/2C1 (3) (Tver'yanovich, Vlcek and Tverjanovich, to be published).
146
E. Bychkov et al.
3
~D
Y.
~
I
~
L
.
.
.
.
_
_
_
_
_
.
.
t
.
.
.
_
100
200
300 A
400
100
200
300 ~
400
J
1
100
200
300
L.____a.~
400
Frequency shifffcrn-1 FIG. 41. Raman spectra of glassy CsGaS3/2C1 (1), CsGaS3/2Br (2) and CsGaS3/2(0.85C1.0.15I) (3) (Tver'yanovich et al. to be published).
respectively (Anderson, 1977). The calculated frequencies are marked in Figure 41 with arrows where digit is number of substituted atoms. Indeed the difference between the spectra of CsGaS3/2C1 and CsGaS3/zBr or CsGaS3/2I is possible due to the vibration of tetrahedra where Br or I occupied some comers. It can be seen (Fig. 42, curve 1) that no shift or appearance of new bands take place but the redistribution of bands intensity exists at partial replacement of Ga by A1 (Tverjanovich, Tver'yanovich and Loheider, 1996a). Apparently such amount of doping (15 mol%) is not enough for the appearance of new
U
r~
3 100
200 300 400 Frequency shift/cm-1
500
Fla. 42. The comparison of spectra of glasses Cs(0.15A1.0.85Ga)S3/2C1 (1), CsGaS3/2C1 (2), and Cs(0.15In.0.85Ga)S3/2C1 (3) (Tver'yanovich et al., to be published).
147
Ion Conductivity and Sensors
vibration frequencies, but we can see the influence of doping on the host structure. The complex-forming ability of A1 is stronger than that of Ga. Thus, one can see that the shape of the spectra of the chain-like structure remains invariable. The increased intensity of band at 340 cm-1 can be caused by two reasons: first, the frequency of the breathing mode of tetrahedron A1C141- is 349 cm -1 (Anderson, 1977); secondly, the frequency of the strongest vibration of the dimer A12C16 is 340 cm -~ (Bell and Longuet-Higgins, 1945). In does not have such a strong ability to form the complex anions as Ga does. Therefore at partial substitution of Ga with In the polymer chain structure of glass disappears and we can see the domination of the tetrahedron breathing mode (320 cm-1 is the breathing mode of tetrahedron GaS3/2C11- (Tverjanovich et al., 1996a)), as it takes place at the heating of the CsGaS3/2C1 melt. The disappearance of chain structure leads to the rise of band half-width as well. The polymer-chain structure of CsGaS3/2C1 glass is also destroyed at the substitution of CsC1 with GeS2. The transformations of the Raman spectrum at the introduction of 0.15GazS3.0.85GeS2 alloy into glassy CsGaS3/2C1 are depicted in Figure 43 (Tverjanovich et al., 1996a). The three-dimensional structure of GazS3-GeS2 glasses is formed by the corner-shared tetrahedra in opposite to the chain structure of the glassy CsGaS3/2C1 where the basic vibration unit is the double tetrahedra. The introduction of 10 mol% of 0.15GazS3.0.85GeS2 leads to the formation of three-dimensional network with the single tetrahedron as the basic vibration unit (320 cm-1). So, the high resolution of CsGaS3/2C1 glass Raman spectra disappears as a result of transition from polymerchain structure to the three-dimensional network. The most intensive transformations of the shape of the spectrum are completed when the content of the 0.15GazS3.0.85GeS2 is equal to 5 - 1 0 mol%. This interpretation corresponds to the result of the glass transition temperature investigations. The low glass transition temperature is typical of the glasses with chain structure. One can see the rapid increase of Tg at the addition of 0-6.5 mol% 0.15GazS3.0.85GeS2 (Fig. 44(a), curve 1) (Tverjanovich et al., 1996a). The saturation of the concentration dependence of Tg takes place at high quantity of 0.15GazS3.0.85GeS2. A similar concentration dependence of Tg takes place at the transition from the chain
I
1O0 FIG. 43.
i
I
i
I
200 300 Frequency shift/cm -1
i
I
i
400
Raman spectra of the (1 - x)(0.333GazS3.0.666CsC1)-x(0.15Ga2S3.0.85GeS2) glasses. (1) x - - 0 ;
(2) x = 0.05; (3) x = 0.1; (4) x = 0.15; (5) x = 0.20.
148
E. Bychkov et al. 220
(a)
120 210 100 200
80 [..,
190
60
180
40 I
20 (b)
[.-,
,
0.00
I
,
0.05
I
0.10
,
I
,
--
0.20
0.15
X
170
4.4 4.2
,"
4.0
et0
3.8
-9
3.6
o..
3.4 3.2
I
0.00
,
I
0.05
,
I
0.10
,
I
0.15
,
.I_.
0.20
FIG. 44. (a) Concentration dependence of the glass transition temperature for glasses in the GexSel-x system (Fontana et al., 1988) (1) and ( 1 - x)(0.333Ga2S3.0.666CsC1)-x(0.15Ga2S3.0.85GeS2) system (2). (b) Concentration dependence of the thermal expansion coefficient for glasses in the (1 -x)(0.333Ga2S30.666CsC1)-x(0.15Ga2S3.0.85GeS2) system.
structure of Se to three-dimensional network of the G e - S e glasses (Fig. 44(a), curve 2) (Balls et al., 1966). Usually the substances with polymer chain structure have higher thermal expansion rate than the substances with three-dimensional network. So, the decrease in thermal expansion coefficient at the introduction of 0.15Ga2S3.0.85GeS2 into CsGaS3/2C1 (Fig. 44(b)) corresponds to the idea of the transition from the chain structure to the three-dimensional network.
2.7.
GLASS-FORMING ABILITY
The similarity of both the concentration boundary shape for tile glass formation regions and the nature of interaction between components in an the systems with Li, Na, K, Rb, Cs, T1 chlorides suggests the general mechanism of glass formation. It was noted above that, in the system containing NaC1, the glass formation region extended along the eutectic fold formed by the field of initial crystallization of GeS2 and Ga2S3 toward the ternary eutectic and was bounded by the field of initial precipitation of NaC1. One can assume that glass formation in these systems is determined by the 'freezing' of lowmelting chlorides into relatively high-melting glass matrix upon the eutectic interaction
149
Ion Conductivity and Sensors
of components. The decrease in the melting point of chloride leads to the reduction of the field of initial precipitation of this chloride and, hence, to the increase in the maximum chloride content in glasses. There is another explanation for why the melts of systems under consideration can be transformed into the vitreous state. For GeSz-GazS3 glasses the fourfold coordination of Ga with S (it will be demonstrated later), along with the same coordination of germanium, produces too high a degree of glass-network connectivity, which hinders glass formation (Fig. 45). For example the introduction of CsC1 into the glass composition and formation of complex structural units [GaS3/zC1]-Cs + retains the fourfold coordination of gallium but decreases the degree of glass-network connectivity (Fig. 45). Therefore, it is proposed that the formation of complex anion favors the glass formation. The insufficient stability of complex anions in the melt at the temperature, when the glass network is formed, is a problem for glass formation. It is known that the stability of the complex anion depends directly on the polar nature of the outer-sphere cation, or inversely, on the field strength of this cation. To put it differently, the stability of the complex anion should increase with a rise in the radius of the outer-sphere cation and should favor the incorporation of halides into glass network. The dependencies of the maximum content of halide in glass on the cation radii and on the melting point of halide are depicted in Figure 46. The systems containing AgC1 and CuC1 especially do not submit to both regularities. This is the result of participation of these chlorides in the exchange reactions with at least one sulphide. The systems containing LiC1 do not follow the dependence of maximum chlorides content in glasses on melting point. It is due to the following reason. The stability of the complex with Li is negligible. Thus, the interaction between GazS3 and LiC1 is absent and a liquation region appears (Fig. 31). Formally it means that at 50 mol% LiC1 two initially crystallized phases exist--one of them is LiC1. But all other systems GazS3-MC1 under consideration are eutectic and for the eutectic systems the expansion field of initial crystallization of any component up to 50 mol% is possible if the melting points of both the components are similar. Therefore, the point corresponding to LiC1 in Figure 46 must be translocated as demonstrated by the arrow. The concept of the formation of complex anions and their beneficial effect on the glassforming ability of melts may be used to explain the regularities of glass formation in the systems under consideration. It seems likely that the increase in the degree of covalence
|
9
.
. "I FIG. 45. The changing of glass-networkconnectivityfor the systemGaeS3-GeS2 at the introduction of CsC1 into the glass composition.
150
E. Bychkov et al. I
80
'
t
j
60 Na "6 ~- 40 Ag 20, Li 0
CU
. 0J)8
0.'12
0.]6
RM ~nm
80T1
Rb
Cs
9
60.
K
&, ~40.
Ag O
Na
20.
Li Cu
Q
400
5()0
6()0
700
8()0
Tmelting,~ FIG. 46. The dependencies of maximum content of different halides in glass on cation radii and on melting point of halides.
for the complexing agent metal bonds due to complex formation is also favorable for glass formation. It was of interest to find a system that would be a totally halide system, but close in composition to the studied compounds, and where the role of complexformation ability of the Ga +3 ions in the formation of glass network would be evident. Such a system, namely, (l-x~GaI3-a.xNaC1,was found. GaI3 does not form any glass. The glass-forming region extends from x = 0.05 to 0.25 and 2.8 <: a < 3 (Tver'yanovich et al., 1996b). The softening points for glasses were below room temperature. The rate of glass hydrolysis in air is very high. The glasses are pale yellow and become colorless with an increase in sodium chloride content.
2.8.
ION CONDUCTIVITY AND ION SELECTIVITY
The compositions of glasses (molar fraction), the conductivity of which was investigated, lie in the following sections: (1)(0.2GaeS3.0.8GeSe)-LiC1;
151
Ion Conductivity and Sensors
-4
LiC1 _ ~ .-~"-~~ "
I
i
0
iO
,
20
l
NaC1
,
i
9
30 40 MC1, mol.%
,
50
dO
FIG. 47. Examplesof the concentration dependence of ionic conductivity.
(2) (0.2Ga2S3-0.8GeS2)-NaC1; (3) (0.3Ga2S3.0.7GeS2)-(0.7NaC1-0.3Ga2S3); (4) (0.2Ga2_ S3.0.8GeSz)-(0.67KC1.0.33GazS3); (5)(0.11GazS3-0.89GeSz)-(0.67CsC1-0.33GazS3); (6) ( 0 . 2 G a z S 3 . 0 . 8 G e S z ) - A g C 1 ; (7) ( 0 . 2 G a z S 3 . 0 . 8 G e S z ) - ( 0 . 7 A g C 1 - 0 . 3 G a z S 3 ) ; (8) (0.2GazS3.0.8GeSz)-(0.65T1C1.0.35GazS3). The examples of the concentration dependence of ionic conductivity are presented in Figure 47. The main aim was to analyze the concentration dependencies of activation energy of the cation conductivity (E~). These dependencies are presented in Figure 48. It should be noted that the main orientation of all the sections ( 1 - 8 ) in glass-formed systems G a z S 3 - G e S z - M C 1 is closed. Due to this fact it became possible to compare the concentration dependencies of E~ for different cations. We propose that the conductivity is defined by the vacancy mechanism. The formation of vacancies is the result of the thermo-dissociation of the cation-vacancy pairs. The activation energy of the cationic conductivity consists of three terms. The Coulomb potential
1.0
0.8 K O
0.6
0.4 0.0
0'.2
' [M]l/3
0'.4
'
0.6
FIG. 48. Concentration dependence of the cation conductivity activation energy.
152
E. Bychkov et al.
of the dissociation of cation-vacancy pairs for vacancy generation has the following value: e 2
Ev =
87reeo(R c +
Ra),
(16)
where e and e0 are the dielectric susceptibility of glass-matrix and the dielectric constant, respectively; e is the charge of electron and Ra and Rc are anion (C1-) and cation radii correspondingly. The Coulomb barrier between two neighboring vacancies is equal to e2
Eb
-
(
1
4"tree0 Re
2
+ Ra
-
1)
-
+
r
2r
(17)
'
where 2r is the distance of jump (see Fig. 49). The elastic deformation of the glass network due to the movement of cations with large radius is determined by the multiplication of the deformed volume (8"rrRh(Rc- Rh) 2) and the shear modulus (G) (Frenkel, 1947), where Rh is the radius of the channels among the atoms through which the cation moves (Rh = 0.06 nm (Anderson and Stuart, 1954)). So, for the activation energy of the cation conductivity (E~) we can write the following expression:
i) E ~ --
Re -+- R a
r
+ 8rrGRh(R c
8~ree o
-
Rh)2.
(18)
We supposed that the effective equilibrium distance between the negative charge of the anion and the positive charge of the cation is approximately the same as the sum of their radii (Rc + Ra). Eq. (5) is similar to that in the paper of Anderson and Stuart (1954). But different suppositions are made at the spelling of these two equations. The authors (Anderson and Stuart, 1954) do not account for the process of generation of empty vacancies. They use the less-precise expression for the height of the Coulomb barrier between the two vacancies. Let us assume that 2r is equal to the average distance between
i
8g~0r .........t.~_ ' t
:.9
i
.
2r
i
...... p
:
FIG. 49. A schematic representation of the cation gap between two vacancies.
Ion Conductivity and Sensors
153
the cation positions in the glass network: 8r 3 ~
ti pNAM '
(19)
where A is the average weight of the atoms, p the glass density, M the atomic fraction of the conductive cations in glass. The dependence of the factor A / p N A on the composition is weak because the density of glasses is usually a linear function of an average mass of the atoms. Let us suppose that the dielectric constant e does not depend on the composition of glasses in the investigated systems. According to the model the E~(M 1/3) dependencies have the following peculiarities. All E~(M 1/3) dependencies under investigation should be linear. Their slope does not depend on the nature of the cation and is equal to dEcr 3e 2 ~ pNA - dM1/3 -- 4"tree0 _ A "
(20)
From this expression and according to the experimental data it can be estimated that e has the value of about 10. The values of E~ at M - - , 0(E ~ are equal to
E~ =
3e 2
8~reeo(Rc + Ra)
+ 8wGRh(Rc
-
(21)
g h ) 2.
We can find the value of E ~ by extrapolation of the linear dependence E~(M 1/3) (Fig. 48). So, Eq. (21) allows finding the part of E~ which is determined by the deformation of the glass network due to the motion of the cations (Edef). According to Eq. (21), it depends linearly on (Rc - Rh) 2 (Fig. 50). Thus, it can be found from the slope of this dependence
0.8
'
'
'
'Cs
O//~
0.6
O
~. 0.4 O
Ag 0.2
0.0
U 0.0
,
, 0.5
. 1'.0
(Rc-Rh)2,10-2nm2 FIG. 50. Deformationpart of the cation conductivity activation energy as a function of (Rc - Rh)2.
154
E. Bychkov et al.
that G ~ 101~ J m -3. According to the results of the sound velocity investigations it was established for various chalcogenide glasses that G changes between 0.97 and 2.06-101~ J m -3 (Borisova, 1981). Glasses Ga2S3-GeS2-MC1 can be used for the manufacture of membranes for potential ion-selective sensors. For example, glasses Ga2S3-GeS2-NaC1 with high content of NaC1 have no sufficient chemical stability not only in the water, but also in the moist air. But if the content of NaC1 is less than 10 mol%, the glasses withstand the contact with water for several months without any modifications of surface. The GeSz can be substituted partially by GeS for additional resistance to water. It is also possible to use the partial substitution of NaC1 by NaF. As a result, the membranes with the stable electrode function can be obtained. These electrodes have (Vlasov, Bychkov, Tver'yanovich, Nedoshovenko, Borisova, Avdeev and Antonov, 1987): low electrical resistance (107 ~-~), that allows use of more noise-resistant measuring devices; low limit of detection of Na ions (pNa ~ 6.1); linear range of electrode function extends from 0 up to 4.5 pNa; they are capable of working in dilute solutions of acids, including HF, so for example at pH 2 (solution of HNO3) the coefficient of selectivity is equal to 0.05.
3. 3.1.
Recent Advances in the Field of Chalcogenide Glass Chemical Sensors SELECTIVE AND LOW-SELECTIVE SENSORS: HISTORY
Chalcogenide glasses are convenient materials for solid membranes of various types of solid-state chemical sensors (conventional ion-selective electrodes (ISE), ion-selective field-effect transistors (ISFET), miniature silicon-based sensors), which can be developed both as selective discrete sensors and low-selective sensor arrays. It is possible to give the following definition of sensors: "A sensor is a device which can respond to certain properties of the environment and can transfer the response into an electrical/optical or some other signals" (Vlasov, 1990, 1997). Physical sensors respond to the changes of pressure, temperature, mass, magnetic field, etc. Chemical sensors respond to the changes of concentration of different chemical species (ions, molecules) in liquid solutions and gas phases. The needs of the industry (especially automobile and domestic equipment industry), agriculture, medicine, environmental control, etc., have stimulated R&D of the sensors which are widely used as a convenient analytical tool. The history of chemical sensors belongs to the last century (Table IV) (Vlasov, 1992; Vlasov and Legin, 1998), the selective discrete sensors having been developed at the first several decades of development and finally at the end of the century low-selective sensors arrays have been used to construct analytical devices such as 'electronic nose' and 'electronic tongue'. By now a great number of various sensors were developed as seen from the improved version of classification of potentiometric chemical sensors (Vlasov, 1993) (Fig. 51). Application of chalcogenide glasses as membranes of sensors had played an important role in the discovery of ionic transport and ionic conductivity in some chalcogenide glasses in 1973 have played the great role for application of chalcogenide glasses as membranes of sensors (Kawamoto, Nagura and Tsuchihashi, 1973).
155
Ion Conductivity and Sensors
T A B L E IV BRIEF HISTORY OF SENSOR DEVELOPMENT (I) 1906-1937. pH glass electrode and ion-exchange theory 1906 Cremer--dependence of the cell potential difference on pH (glass membrane) 1909 Haber, Klemensiewicz--development of a glass electrode 1936 Beckman--commercial of pH meter 1937 Nikolsky--Nikolsky equation and theory of operation of a glass electrode 1937 Kolthoff--crystalline 'electrode' 1937 Nikolsky---crystalline membrane (II) 1961-1969. Conventional ion-selective electrodes and biosensors 1957 Eisenman--properties of glass electrode and Eisenmann-Nikolsky equation 1958 Severinghaus, Bradley--gas-sensitive electrodes 1961 Pungor--heterogeneous solid ISE 1961 Eisenman--theory of glass electrode 1962 Seiyama, Taguchi--semiconductor gas sensor 1966 Frant, Ross--LaF3-electrode 1966 Simon--liquid ISE with neutral carrier 1967 Ross--ion-exchange membrane 1969 Guibault, Montalvo--potentiometric biosensor 1969 Baker, Trachtenberg---chalcogenide glass membrane for ISE 1971 Moody, Thomas--PVC-based ISEs (III) 1970-till present. Microelectronics in sensor development 1970 Bergveld--ISFET 1975 Lundstrom--gasFET 1976 Schenck--immunoFET 1986 Thorn EMI Microsensor--first commercial ISFET (IV) 1982-till present. Multisensor arrays and sensor systems 1982 Persaud, Dodd--'electronic nose' 1992 Toko--'taste' sensor 1995 Vlasov, Legin, D'Amico, Di Natale--'electronic tongue'
J
Chemical sensors
Selective
discrete
sel'~OP~
I
Low-celective
sen~ors for "electronic tongue"
I
i,ooo~
JElectr~ sensitive J
I
the 1st and J 2nd kind J
I
JMoleculesensitive J
I seIn-ire J
I
reduction systems
I ISE
I
1
I I I
ISFET
I I I
GasFET
I
Jo, ..... I I(BioFETetc)J I
FIG. 51. Chemical sensor classification. Various membrane materials (the lower level of classification) can be used for fabrication of different sensors. Chalcogenide glass materials can be used for membranes of ionselective electrodes (ISE) and ion-selective field effect transistors (ISFET) (middle level) to apply them both as discrete selective sensors and as low-selective cross-sensitive sensor arrays in electronic tongue (top level).
156
E. Bychkov et al.
From the very beginning of sensor development it was a great idea to find out such membrane materials for sensors which could respond sele.ctively to concentration of some specific species (ions, molecules). But in most cases it was a problem to get such membranes because usually most of the materials were sensitive to several species with different grade of sensitivity. First, they tried to solve this problem by introducing the coefficients of selectivity Kid into the famous Nikolsky equation (Nikolsky, 1937) for dependence of the sensor potential on activity (concentration) of the ion in solution. E-
RT
E0 + ~
ln(ai + kijaj) ,
(22)
where E is electrode (sensor) potential, E0 the standard potential, R the gas constant, T absolute temperature, n charge of ion i, F Faraday number, a i and aj are activities of the primary and interfering ions in solution, respectively. Eq. (22) expresses the linear dependence of E on In a i with the slope (which is the sensor sensitivity), R T / n F equal to 59 mV/ln a (n = 1) Or 28.5 mV/ln a (n = 2) at T = 298.2 K. In the case where there are several interfering ions in solution (multicomponent system) it is impossible from Eq. (22) to take into consideration the influence of all interfering species on the sensor potential E. The solution of the problem of selectivity was found out at the end of the last century; then application of low-selective sensor arrays together with pattern recognition methods of signal processing based on artificial neural networks (ANN), fuzzy logic, etc., was suggested and gas and liquid analyzers like electronic nose and electronic tongue were developed. Thus, depending on the sensor application it is necessary to consider important parameters of the chemical sensors, such as selectivity and cross-.sensitivity (low-selectivity). The latter parameter demonstrates the ability of the sensor material to respond to several different species in liquid or gas phases. To understand better the role of chalcogenide glasses as a membrane material for chemical sensors it is interesting to consider a brief history of chemical sensors (Vlasov, 1992; Vlasov and Legin, 1998). From Table IV it is seen that there are four periods in sensor history. At first of three periods it was a struggle of the scientists for developing the sensors with the highest level of selectivity. And at the second period the chalcogenide glasses were introduced as solid membrane material (Baker and Trachtenberg (1969)). It was the start for chalcogenide glass ion-selective sensor investigations. Jasinski and Trachtenberg (1972, 1973a,b), Owen (1980) and Tohge, Minami and Tanaka (1983) suggested some compositions for membranes of chalcogenide glass sensors. Wide investigations have been carried out by Vlasov and Bychkov (1982) (from 1982 till now), who studied both solid-state properties of chalcogenide glass membranes and their sensor behavior (see review paper (Vlasov and Bychkov, 1987)). Their extensive research resulted in the development of great number of bulk chalcogenide glass sensors. Later, at the third period of the history this material was applied as thin films for ISFETs (Vlasov and Tarantov, 1989) and at the fourth modem period of the history (Table IV) as low-selective cross-sensitive membranes for electronic tongue (Di Natale, D'Amico, Vlasov, Legin and Rudnitskaya, 1995; Vlasov, Legin, Rudnitskaya, Di Natale and D'Amico, 1996) (see also Fig. 51) and silicon-based thin films sensors (Mourzina, Schoning, Schubert, Zander, Legin, Vlasov and Luth, 2001a; Mourzina, Schubert, Zander, Legin, Vlasov
Ion Conductivity and Sensors
157
and Schoning, 2001b). The principles of sensors are widely described in many monographs (Morf, 1981; Gopel, Hesse and Zemel, 1991). In this chapter we will point out shortly the most important properties of the membranes of chemical sensors and general parameters of chemical sensors and will note the advantages of chalcogenide glass sensors. The most important parameters of the chemical sensors are: (i) selectivity or cross-sensitivity, (ii) sensitivity, (iii) potential stability, (iv) the concentration limit, (v) response time, (vi) pH region, and (vii) life time. From a practical point of view the membranes of solid-state chemical sensors for applications in the technology processes and natural liquid media must be unsoluble, have mechanical and chemical durability, must be ionic conductors with low electronic conductivity to fabricate all-solid-state sensors with solid internal contact (metallic wire). Using chalcogenide glass sensors of different types it is possible to get the improved sensor parameters mentioned above.
3.2.
3.2.1.
SENSORFABRICATION
Methods of Fabrication
The method of formation of bulk chalcogenide glass sensor membranes and targets for Pulse Laser Deposition (PLD) is described below. Thin films of chalcogenide glasses have been prepared by some physical and chemical methods, such as high-frequency (Vlasov and Tarantov, 1989) and r.f. co-sputtering glass and dopant (Bychkov, Bruns, Klewe-Nebenius, Pfennig, Hoffmann and Ache, 1995), PLD (Schoning, Mourzina, Schubert, Zander, Legin, Vlasov and Luth, 2001a), spincoating technique or sol-gel process (Vlasov, Legin and Baidakov, 1996) (using organic solutions). PLD is the most convenient method because of its compatibility with IC technology which makes it possible to fabricate microelectronic silicon-based sensors. PLD gives the possibility to prepare thin films with stoichiometric composition similar to the bulk target used.
3.2.2.
Chalcogenide Glass Conventional Sensors
The details of experimental methods of the preparations of chalcogenide glasses have been described elsewhere (Vlasov, Bychkov and Seleznev, 1990). The main features of their fabrication consist of glass formation from melted pure substances in evacuated silica ampoules (800-1200 K) followed by quenching in air or water. The homogeneity and amorphous state were confirmed by X-ray diffraction, while surface properties were studied by X-ray photoelectron spectroscopy, Auger electron spectroscopy, secondary ion mass spectrometry and radioisotope methods. The EXAFS and Mossbauer spectroscopy and some other methods were also used for investigation (Vlasov and Bychkov, 1987). Sensor membranes were prepared by cutting discs (6-10 mm in diameter) from glasses, followed by polishing and sealing into plastic tubes. Solid inner contacts were
E. Bychkov et al.
158
TABLE V CHALCOGENIDE GLASS ION-SELECTIVE SENSORS
Determination n(X)
Ag + Cu + pb2+
Membrane composition Ag-As-S, Ag-As-Se Cu-Ag-As-Se Pb12-Ag2S-As283, PbS-AgzS-AszS3 CdS-AgzS-AszS3, CdI2-Ag2S-As2S3 Ge28Sb12Se6o(Fe) AgBr-AgzS-AszS~ Tlwchalcogenide glass Hg--chalcogenide glass Cr---chalcogenide glass NaC1-GezS3-GeS2
Cd 2+ Fe 3+ BrT1+ Hg 2+ Cr(IV) Na +
Sensitivity (mV pX -1)
Detection limit (10 -7 mo1-1)
pH
59 29 29
5 1 1
Up to 6 mol 1-~ HNO3 Up to 1 mol 1-1 HNO3 2-6
28
1
1-7
58 59, 5 59 28 30-60** 58
500 5 50 5 1 400
1-2 2-10 0_5 t 0_6 t 1-2
*Glassy/crystalline membrane. *pH region depends on solution concentration. ** The sensitivity depends on solution concentration.
obtained by deposition of a silver layer on the inner side of the membrane and attaching a wire by micro-adhesive. The electrochemical behavior of the sensors was characterized by potentiometric measurements. Chalcogenide glass sensors for determination of various ions in solutions have been developed as selective for discrete sensors application (Table V) and as low selective (with cross-sensitivity) to use in sensor arrays for electronic tongue (Vlasov, Legin and Rudnitskaya, 1997). All the chalcogenide glass ion-selective sensors have long-term stability with steady potential during continuous measurements, all-solid state design, high sensitivity, low detection limits, long life time, and high stability in aggressive, strong acid and corrosive media. There is no necessity for preliminary treatment of the sensor before measurements; they are especially convenient for industrial and environmental control applications. The glass samples with membrane compositions shown in Table V were also used as targets in PLD for silicon-based sensors fabrications. The change of component concentration in the membranes resulted in changing parameter of the sensors such as selectivity and receiving cross-sensitivity. In this case it is possible to use them for electronic tongue fabrications. Many of the chalcogenide glass compositions were investigations from this point of view (Vlasov et al., 1997). 3.2.3.
ISFETs
Arsenic triselenide film doped with copper (As2Se3:Cu) or nickel (As2Se3:Ni) was obtained by high-frequency co-sputtering of the glass and a corresponding metal. Copper or nickel concentration in the chalcogenide film was determined by X-ray microanalyzer and varied in the range 0 . 5 - 1 0 at.%. By sputtering chalcogenide film on the surface of silicon SiO2 layer 70 nm thick was obtained (Vlasov and Tarantov, 1989).
Ion Conductivity and Sensors
159
N-channel silicon ISFETs were used. The substrate was (100) p-type monocrystalline silicon doped with boron to a resistivity of 10 1~ cm. Source and drain regions were formed by doping with phosphorus. The length and the width of the channel were 30/~m and 1 mm, respectively. An SiO2 layer 70 nm thick was used as a subgate insulator. Chalcogenide film 300 nm thick was sputtered onto the SiO2 layer. The sealing of end faces, bonding pads and current taps of ISFETs was achieved by the use of an epoxy compound. Ellipsometric measurements were made by a laser photoelectric ellipsometer with operational wavelength 632.8 nm in the air and in contact with electrolyte solutions. In the latter case a special cell in the form of a Teflon bag was used. Multiangle ellipsometer measurements were employed. Capacitance-voltage curves for an electrolyte-insulatorsemiconductor (EIS) system were measured at 30 kHz by conventional techniques. The ISFETs with chalcogenide membranes were investigated. To receive sensitivity to Cu, Ni (Vlasov and Tarantov, 1989), and Cd (Salardenne, Morcos, Ait Allad and Portier, 1990), the compositions like AszSe3(Cu), AszSe3(Ni) and AszSe3-AgzS-CdS have been used, respectively. The sensitivity of Cu-ISFET and Ni-ISFET with chalcogenide membranes (Vlasov and Tarantov, 1989) is shown in Figure 52.
3.2.4.
Silicon-based Thin Film Chalcogenide Glass Sensors (Schoning, Mourzina, Schubert, Zander, Legin, Vlasov and Luth, 2001b)
The targets for PLD were bulk monolith pieces of chalcogenide glasses prepared according the procedure described above. The typical size of the target was 10-12 mm in diameter and 10-12 mm in length. Single-crystal silicon plates of (100) n-type, phosphorus doped, were used as substrate for sensor fabrication. This substrate has a resistivity of 1000-5000 ~ cm, and a thickness of 350 _+ 40 ~m. Before depositing the chalcogenide films, different metallic contact layers of Ti/Pt, Ti/Pt/Au and Cr/Au were photo-lithographically structured on a substance. The metallic contact layers were cut into individual dies and the deposition of the thin film chalcogenide-based sensitive layers from the bulk chalcogenide glass targets was performed using an off-axis PLD technique. In this setup a KrF excimer laser with a pulse width of 20 ns and a pulse energy of approximately 1 J per pulse was used. The deposition cycles were carried out at N2 partial pressure of about 0.2 mbar at room temperature to prevent oxidation of the films. The deposition time varied from 300 to 600 s. When using an energy density of 5 J cm -2 and a repetition rate of 10 Hz, the layer thickness ranged between 700 and 1500 nm. After fabrication, the chalcogenide-based thin film sensors were mounted on printed circuit boards, followed by wire bonding and encapsulation. The component composition, structural features and thickness of the deposited layers were investigated by Rutherford backscattering spectrometry (RBS) using 1.4 MeV incident He + ions and transmission electron microscopy (TEM). The electrochemical behavior of the sensors was characterized using potentiometric measurements. The parameters of silicon-based thin films sensors developed are summarized in Table VI (Schoning et al., 2001b).
160
E. Bychkov et al.
//
(a)
E
,
7
I
,
8
I
,
I
5
,
I
4
,
I
3
,
I
2
,
1
0
pCu (b)
,
7
FIG. 52.
i
.
8
i
6
,
i
4
.
i
3
,
i
.
i
2
.
1
13
pNi (a) C u - I S F E T and (b) Ni-ISFET. Sensor potential (E) dependence on ion concentration in solutions.
T A B L E VI SILICON THIN FILMS CHALCOGENIDE GLASS SENSORS Parameter
Membrane material Pb-Ag-I-As-S
Ion Sensitivity (mV p X - 1) Response time (min) Stability (day) pH range Detection limit (mol 1-~) Drift (mV day -~)
Pb e+ 26-29 < 1 > 210 2-7 1 x 10 -7 < 1
Cd-Ag-I-As-S Cd e+ 25-28 < 1 > 180 2.5-8 4 x 10 -7 < 1
Cu-Ag-As-Se-Te Cu e+ 27-30 < 1 > 210 2-6 1 • 10 -7 < 1
T1-Ag-As-I-S TI+ 56-59 < 1 > 150 2-5 3 x 10 -5 < 1
Ion Conductivity and Sensors
161
These sensors are developed with methods compatible with IC technology and could be used in future for fabrication of multisensor arrays for microelectronic devices like electronic tongue. In addition, these methods of silicon-based microsensors fabrication were used to prepare ISFETs and new types of chemical microsensors with chalcogenide glass membrane (so-called the Light Addressable Potentiometric Sensors (LAPS)) (Ermolenko, Yoshinobu, Mourzina, Schoning, Vlasov and Iwasaki, 2001; Mourzina, Yoshinobu, Schubert, Luth, Iwasaki and Schoning, 2001c).
3.3.
MECHANISM OF CHALCOGENIDE GLASS SENSORS OPERATION
All the compositions of chalcogenide glass membranes (Table V) demonstrate high ionic conductivity and ion transport which is necessary for achieving ion sensitivity of the sensor membranes. Following the consideration of a chalcogenide glass sensor as a multiphase system (Fig. 53), it is evident that the sensor operation in solutions should depend on properties of the bulk and surface of the membrane. In the extensive research by Vlasov and Bychkov with co-workers (Vlasov and Bychkov, 1984, 1987, 1994) it was shown that ionic and electronic processes which govern ionic sensitivity and selectivity of the chalcogenide glass sensors depend on bulk compositions and formation of modified surface layers with intrinsic and induced exchange sites, the potential forming process being the direct ion exchange between solution and modified surface layers. It was found that there is a close relation between surface ionic processes, bulk ionic and electronic transport and durability of the glass surface (Vlasov and Bychkov, 1994). A considerable glass-forming ability with several elements makes it possible to vary the composition of chalcogenide glasses over a wide range changing physical, chemical, and electrochemical properties. From the typical experimental results (Figs. 54 and 55) the possibility to prepare chalcogenide glass membranes with different electrochemical
Properties
of the bulk
Electrolyte solution
Solid membrane
\
Metal
/
Properties of t~he surfacek~/ FIG. 53. The sketch of solid membrane of chemical sensor for solution concentration determination (Vlasov, 1990). The modified surface lager exists in solid membrane on the solution/solid interface.
162
E. Bychkov et al. Pbl 2
~ 6 z{o0/ /A\%,&\n-~ ~Z
Ag 2S
20
40
&40
60
80
As2S 3
FIG. 54. Composition dependence of some properties of chalcogenide :glasses: glass forming region (open circles) and sensor properties of chalcogenide glass membranes in the system PbIz-AgzS-AszS3 (Vlasov and Bychkov, 1987). Sensor properties received from separate experiments and expressed by numbers (arb. units) inside the glass region. The more is number, the more is cross-sensitivity.
. . . . .
o
I ~
r
l
'
CUxAg25-xAS37.sSe377. .-:_
C/I
io -4
Eo
i
I0
b
.
I0-~ 298
IO
-7
o
I 5
I 10
I 15
x { a t . % Cu}
,, I 20
K 25
FIG. 55. Conductivity dependence of chalcogenide glass solid membranes on dopant (Cu) concentration in the system CuxAgzs_xAs37.sSe37.5 (Vlasov and Bychkov, 1987). Types of conductivity: total (solid line), ionic (open circles), hole (closed circles).
Ion Conductivity and Sensors
163
and sensor properties is seen. Using the different compositions one can get the glass membranes with great selectivity as well as with cross-sensitivity and low selectivity. Thus, chalcogenide glasses could be used as membranes of selective discrete sensors and for low-selective sensor arrays to apply them in the electronic tongue. X-ray photoelectron and Auger spectroscopy measurements of the bulk membranes, radionuclide investigation of ion diffusion in solids and solution/surface ion exchange (Vlasov and Bychkov, 1987) as well as ellipsometry study of thin films (Vlasov and Tarantov, 1989) confirm the formation of modified surface layers of 1 0 0 - 1 2 0 n m thickness. 3.4.
SENSOkAPPLICATIONS
Taking into account the special properties of chalcogenide glass sensors such as longterm stability, all-solid-state design, stability in corrosive and strong acid media, improved sensitivity and selectivity and possibility to vary the cross-sensitivity one can apply these sensors in technological processes, for environmental monitoring, in medicine, for food quality control, etc. The conventional sensors are convenient for industrial application, and miniature silicon-based sensors, including ISFETs, can be used in medicine and microanalysis. Some examples of chalcogenide glass sensor applications (Vlasov, Bychkov and Legin, 1994, 1997) are described below. Chalcogenide glass sensors (Table V) are available for determination of heavy metal ions such as Ag +, Cu 2+, Pb 2+, Cd 2+, Fe 3+, T1+, Hg 2+, Cr(VI) as well as Na +, Br-, S 2- in the different solutions. Copper chalcogenide glass sensors have been used for copper concentration determination in natural water, waste waters of different factories, in process control, for example in high-purity nickel hydrometallurgical productions, 0.4-2.0 mg 1-1 of copper being measured during 2 - 3 months on the background of 10,000 times excess of nickel ions; the sensor was also applied to control the corrosion process at some Russian nuclear power plants. Lead-selective sensor has been used for a long-term monitoring of lead content in the waste water of lead battery factories. Chromium(VI) has been determined by chrome sensor in industrial waste of the machinery factory, in plating baths, in technological solutions during radioactive waste utilization. Sodium chalcogenide glass sensor is a perspective for Na + ions determination in fluoride solutions (Vlasov and Bychkov, 1989). Low-selective chalcogenide glass sensors in sensor arrays in the electronic tongue devices have opened new wide field of sensor application in analytical chemistry for qualitative and quantitative analysis. One of the advantages of electronic tongue is the possibility to determine the concentration of ions for which no sensor exists, but in this case the potential measurements of other sensors in array give the information about concentration of the ion of interest (Fig. 56). Electronic tongue usually consists of 5 - 3 0 various sensors, chalcogenide glass sensors having been used in combination with some other types of sensors. Electronic tongue was applied, for example, for quantitative measurements of heavy metals and other impurities in Neva river (St Petersburg, Russia) (Di Natale et al., 1995; Vlasov et al., 1996), for U(VI)
164
E. Bychkov et al.
FIG. 56. Dependence of sensor potentials E in three chalcogenide glass sensor array (Fe-Cu solution), the array having no Fe-selective sensor. The pattern recognition methods of sensor signal E processing (artificial neural network, principle component analysis, etc.) gives the information about Fe and Cu concentration in solution (unpublished results). and U(IV) determination in uranium mines in former East Germany (Legin, Seleznev, Rudnitskaya and Vlasov, 1999a; Legin, Seleznev, Rudnitskaya, Vlasov and Tverdochlebov, 1999b), for groundwater monitoring near city of Braunschweig, G e r m a n y (Rudnitskaya, Ehlert, Legin, Vlasov and Buttgenbach, 2001), and in flow-injection system for online measurements of the concentration of lead, chromium, copper, c a d m i u m in exhaust gases from waste incineration plants (Mortensen, Legin, Ipatov, Rudnitskaya, Vlasov and Hjuler, 2000). -3 I
-2 9
-1 ;
[
0 "
I
1 9
I
2 !
9
3 I
Nescafe Classic
Manhattan
1
e-
ei
Tchibo
0
0
e-
|
8
E -1 0 o
Nescafe Gold J l
.)Arabica l
-3
- -
I
-2
"
I
-1
'
I
0
9
I
1
I
2
9
!
/
~2
3
C o m p o n e n t 1, arb.units
FIG. 57. Result of the application of electronic tongue consisted of chalcogenide glass sensors for classification and taste determination of different types of coffee.
Ion Conductivity and Sensors
165
As an intelligent sensor system electronic tongue based on chalcogenide glass sensors have been used for classification and taste determination of many beverages (coffee, bear, mineral waters, juice, tea, wines), vegetable oil, milk, etc. (Legin, Rudnitskaya, Seleznev and Vlasov, 2002; Legin, Rudnitskaya, Lvova, Vlasov, Di Natale and D'Amico, 2003) (Fig. 57.). The information about chalcogenide glass sensors is available from some review papers (about discrete sensors see Vlasov and Bychkov (1987), ISFETs (Legin et al., 2003), electronic tongue (Vlasov and Legin, 1998; Vlasov, Legin and Rudnitskaya, 2002a)) and from the books of Vlasov, Legin and Rudnitskaya (2002b) and Legin, Rudnitskaya and Vlasov (2003).
3.5.
CONCLUSION
Chalcogenide glasses are convenient materials for development of various types of chalcogenide glass chemical sensors: (i) Conventional discrete ion-selective sensors, which have ion-conductive membranes and completely solid state design, improved selectivity, high sensitivity, low detection limit, long-term stability and life time in aggressive, strong acid and corrosive media. (ii) Microelectronic ISFETs and silicon-based miniature sensors, including LAPS, in which thin films of chalcogenide membrane can be formed by the methods compatible with IC-Technology; ISFETs have small size, completely solid-state design, low output resistance, possibility of developing multisensors on one chip to make miniature sensor arrays for electronic tongue and different analytical and bioanalytical applications. (iii) Low-selective, cross-sensitive all-solid-state sensors with high stability and durability in water and corrosive media for application in sensors arrays in analytical devices such as electronic tongue. The solid state and electrochemical investigation of chalcogenide glass membranes of chemical sensors indicate that there are definite relationships between ionic response and local structure, bulk ion transport, surface ion exchange. The main potential generating process is an exchange of primary ions between solution and modified surface layer of the glass. Chalcogenide glass chemical sensors can be widely used in industry, medicine, for environmental control for qualitative and quantitative analysis of liquids, including technology solutions, waste water, drugs, beverages (food), biological liquids, etc.
References
Aksenov, N.D., Gutenev, M.S., Orkina, T.N. and Makarov, L.L. (1992) Sov. J. Glass Phys. Chem., 18, 182. Alekseev, I. and Bychkov,E. (2003) Proceedings of the 14th International Conference on Solid State Ionics, Monterey, USA, June 2003, p. 333. Anderson, A. (Ed.) (1977)Application of Raman Spectra, Mir, Moscow,p. 62.
166
E. Bychkov et al.
Anderson, O.L. and Stuart, D.A. (1954) J. Am. Ceram. Soc., 37, 573. Armand, P. unpublished EXAFS results for the AgI-As2Se3 glasses. Armand, P., Ibanez, A., Philippot, E., Ma, Q. and Raoux, D. (1992) J. Non..Cryst. Solids, 150, 371. Baidakova, O.L., Dvinova, A.S., Tver'yanovich, Yu.S. and Turkina, E.Yu. (1991) Vestnik Leningradskogo Universiteta Ser., 4, 96. Baker, C.T. and Trachtenberg, J. (1969) US Office Saline Water Res. Dev. Progr. Rep., 469, 35. Balls, A., Downs, A.J., Greenwood, N.N. and Stranghan, B.P. (1966) Trans. Faraday Soc., 62, 521. Barnier, S., Palazzi, M., Massot, M. and Julien, C. (1990) Solid State Ionics, 44, 81. Beattie, I.R., Gilson, T. and Cocking, P. (1967) J. Chem. Soc., A4, 702. Bell, R.P. and Longuet-Higgins, H.C. (1945) Proc. R. Soc. A, 183, 357. Benmore, C.J. and Salmon, P.S. (1993) J. Non-Cryst. Solids, 156-158, 720. Benmore, C.J. and Salmon, P.S. (1994) Phys. Rev. Lett., 73, 264. Ben-Shaban, Y.M. (1998) PhD Thesis, St Petersburg University, St Petersburg. Berg, R.W. and Bjerrum, N.J. (1983) Polyhedron, 2, 179. Borisova, Z.U. (1981) Glassy Semiconductors, Plenum Press, New York. Borisova, Z.U., Bychkov, E.A. and Tveryanovich, Yu.S. (1991) Interaction of Metals with Chalcogenide Glasses, St Petersburg University Press, St Petersburg. Bunde, A., Ingram, M.D. and Maass, P. (1994) J. Non-Cryst. Solids, 172-174, 1222. Buroff, A., Nebauer, E., Siiptitz, P. and Willert, I. (1977) Phys. Stat. Sol., A40, K195. Bychkov, E. (2001) Defect Diff. Forum, 194-199, 909. Bychkov, E., Bolotov, A., Armand, P. and Ibanez, A. (1998) J. Non-Cryst. Solids, 232-234, 314. Bychkov, E., Bolotov, A., Tsegelnik, V., Grushko, Yu. and Vlasov, Yu. (2001a) Defect Diff. Forum, 194-199, 919. Bychkov, E., Bruns, M., Klewe-Nebenius, H., Pfennig, G., Hoffmann, W. and Ache, H. (1995) Sensors Actuators B, 24-25, 733. Bychkov, E., Bychkov, A., Pradel, A. and Ribes, M. (1998) Solid State Ionics, 113-115, 691. Bychkov, E., Hannon, A.C. and Lapp, A. (200 l b) XIX International Congress on Glass, Edinburgh, Scotland, July, p. 384. Bychkov, E. and Price, D.L. (2000) Solid State Ionics, 136-137, 1041. Bychkov, E., Price, D.L., Hannon, A.C. and Benmore, C.J. (2002) Solid State Ionics, 154-155, 349. Bychkov, E., Price, D.L. and Lapp, A. (2001c) J. Non-Cryst. Solids, 293-295, 211. Bychkov, E., Siewenie, J., Sampath, S. and Benmore, C.J. (2004) Ninth International Conference on the Structure of Non-Crystalline Materials, Coming, USA, July 2004. Bychkov, E., Tsegelnik, V., Vlasov, Yu., Pradel, A. and Ribes, M. (1996) J. Non-Cryst. Solids, 208, 1. Cho, J., Martin, S.W., Meyer, B., Kim, K.-H. and Torgeson, D.R. (2000) J. Non-Cryst. Solids, 270, 205. Cormack, A.N., Du, J. and Zeitler, T. (2003) J. Non-Cryst. Solids, 323, 147. Dawed, M. (1992) PhD Thesis, St Petersburg University, St Petersburg. Dejus, R.J., Susman, S., Volin, K.J., Montague, D.G. and Price, D.L. (1992) J. Non-Cryst. Solids, 143, 162. Di Natale, C., D'Amico, A., Vlasov, Yu., Legin, A. and Rudnitskaya, A. (1995) Proceedings of the International Conference on Eurosensors IX, Stockholm, Sweden, p. 512. Doremus, R.H. (1962) In Modern Aspects of the Vitreous State, Vol. 2 (Ed., Mackenzie, J.D.) Butterworth, London, p. 1. Drugov, Yu., Bolotov, A., Vlasov, Yu. and Bychkov, E. (2000) Solid-State lonics, 136-137, 1091. Ermolenko, Yu., Yoshinobu, T., Mourzina, Yu., Schoning, M., Vlasov, Yu. and Iwasaki, H. (2001) Anal. Sci., 17, 1777. Evstrop'ev, K.K. (1970) Diffusion Processes in Glass, Stroiizdat, Leningrad. Feltz, A., Voigt, B., Burckhardt, W., Senf, L. and Leonhardt, G. (1976) In Structure and Properties of NonCrystalline Semiconductors (Ed., Kolomiets, B.T.) Nauka, Leningrad, p. 88. Fontana, A., Rosi, B., Ivanova, Z. and Kirov, N. (1988) Philos. Mag. B, 56, 507. Frenkel, J. (1947) Kinetic Theory of Liquids, Oxford University Press, New York, p. 544. Frischat, G.H. (1975) Ionic Diffusion in Oxide Glasses, Trans Tech Publishers, Aedermannsdorf, 182pp. Fuoss, P.H., Eisenberger, P., Warburton, W.K. and Bienenstock, A. (1981) Phys. Rev. Lett., 46, 1537. Gopel, W., Hesse, J. and Zemel, J.N. (Eds.) (1991), Sensors VCH, Vol. 7/8 Weinheim, 395pp. Greaves, G.N. (1985) J. Non-Cryst. Solids, 71, 203. Gutenev, M.S., Orkina, T.N. and Baidakov, L.A. (1990) Izv. Akad. Nauk SSSR, Neorg. Mater., 26, 1737. Haven, Y. and Verkerk, B. (1965) Phys. Chem. Glasses, 6, 38.
Ion Conductivity and Sensors
167
Hayashi, A., Tatsumisago, M. and Minami, T. (1999)J. Electrochem. Soc., 146, 3472. Hesler, R.E. (1971) In Advances in Molten Salt Chemistry, Vol. 1 (Eds., Brannstein, J., Mamantov, G. and Smith, G.P.) Kluwer, Dordrecht, p. 1. Hunt, A. (1994) J. Non-Cryst. Solids, 175, 59. Jain, H., Peterson, N.L. and Downing, H.L. (1983) J. Non-Cryst. Solids, 55, 283. Jasinski, R. and Trachtenberg, J. (1972) Anal. Chem., 44, 2373. Jasinski, R. and Trachtenberg, J. (1973a)J. Electrochem. Soc., 120, 1169. Jasinski, R. and Trachtenberg, J. (1973b) Anal. Chem., 45, 1973. Kawamoto, Y., Nagura, N. and Tsuchihashi, S. (1973) J. Am. Ceram. Soc., 56, 289. Kawamoto, Y., Nagura, N. and Tsuchihashi, S. (1974) J. Am. Ceram. Soc., 57, 489. Kawamoto, Y. and Nishida, M. (1977) Phys. Chem. Glasses, 18, 19. Kazakova, E.A. (1980) PhD Thesis, Leningrad University, Leningrad. Kelly, J.E. III, Cordaro, J.F. and Tomozawa, M. (1980) J. Non-Cryst. Solids, 41, 47. Kennedy, J. and Yang, Y. (1987) J. Solid State Chem., 69, 252. Kirkpatrick, S. (1973) Rev. Mod. Phys., 45, 574. Kirkpatrick, S. (1978) AIP Conf. Proc., 40, 99. Kirkpatrick, S. (1979) In Ill-Condensed Matter (Eds., Balian, R., Maynard, R. and Toulouse, G.) North-Holland, Amsterdam, p. 321. Kolomiets, B.T., Rukhlyadev, Yu.V. and Shilo, V.P. (1971) J. Non-Cryst. Solids, 5, 389. Kuznetsov, S.L., Mikhailov, M.D., Pecheritsyn, I.M. and Turkina, E.Yu. (1997) J. Non-Cryst. Solids, 213-214, 68. Le Claire, A.D. (1970) In Physical Chemistry--An Advanced Treatise, Vol. 10 (Eds., Eyring, H., Henderson, D. and Jost, W.) Academic Press, New York, p. 261. Lee, J.H., Owens, A.P., Pradel, A., Harmon, A.C., Ribes, M. and Elliott, S.R. (1996) Phys. Rev. B, 54, 3895. Legin, A., Rudnitskaya, A., Lvova, L., Vlasov, Yu., Di Natale, C. and D'Amico, A. (2003) Anal. Chim. Acta, 484, 33. Legin, A., Rudnitskaya, A., Seleznev, B. and Vlasov, Yu. (2002) Int. J. Food Sci. Technol., 37, 375. Legin, A., Rudnitskaya, A. and Vlasov, Yu. (2003) In Integrated Analytical Systems (Ed., Alegret, S.) Elsevier, Amsterdam, p. 437. Legin, A.V., Seleznev, B.L., Rudnitskaya, A.M. and Vlasov, Yu.G. (1999a) Radiokhimia, 41, 86. Legin, A.V., Seleznev, B.L., Rudnitskaya, A.M., Vlasov, Yu.G. and Tverdochlebov, S.V. (1999b) Czech. J. Phys., 49(Suppl. S 1), 679. Le Toullec, M., Christensen, P.S., Lucas, J. and Berg, R.W. (1987) Mater. Res. Bull., 22, 1517. Lucovsky, G., Galeener, F.L., Keezer, R.C., Geils, R.H. and Six, H.A. (1974) Phys. Rev. B, 10, 5134. Malugani, J.P. and Robert, G. (1980) Solid State Ionics, 1, 519. Manning, J.R. (1968) Diffusion Kinetics for Atoms in Crystals, Van Nostrand, Princeton, NJ. Mikhailov, M.D., Tver'yanovich, Yu.S. and Turkina, E.Yu. (1993) Fiz. Khim. Stekla, 29, 2. Morf, W.E. (198 l) The Principles of Ion-Selective Electrodes and of Membrane Transport, Akademiai Kiado, Budapest, 43 lpp. Mortensen, J., Legin, A., Ipatov, A., Rudnitskaya, A., Vlasov, Yu. and Hjuler, K. (2000)Anal. Chim. Acta, 403, 273. Mourzina, Yu., Schoning, M., Schubert, J., Zander, W., Legin, A., Vlasov, Yu. and Luth, H. (2001 a) Anal. Chim. Acta, 433, 103. Mourzina, Yu., Schubert, J., Zander, W., Legin, A., Vlasov, Yu.G. and Schoning, M.J. (2001b) Electrochim. Acta, 47, 251. Mourzina, Yu., Yoshinobu, T., Schubert, J., Luth, H., Iwasaki, H. and Schoning, M.J. (2001 c) Sensors Actuators B, 80, 136. Murch, G.E. (1982) Solid State Ionics, 7, 177. Nedoshovenko, E.G., Turkina, E.Yu., Tver' yanovich, Yu.S. and Borisova, Z.U. (1986) Vestnik Leningradskogo Universiteta Ser., 4, 52. Nikolsky, B.P. (1937) Acta Phys.-Chim. URSS, 7, 597. Orkina, T.N. and Baidakov, L.A. (1993) Sov. J. Glass Phys. Chem., 19, 460. Owen, A.E. (1967) Glass Ind., 48, 637. Owen, A.E. (1980) J. Non-Cryst. Solids, 35-36, 999. Patel, H.K. and Martin, S.W. (1992) Phys. Rev. B, 45, 10292. Penfold, I.T. and Salmon, P.S. (1990) Phys. Rev. Lett., 64, 2164.
168
E. Bychkov et al.
Petri, I., Salmon, P.S. and Fischer, H.E. (2000) Phys. Rev. Lett., 84, 2413. For recent references see, for example, Proceedings of the International Conference on Solid State Ionics-2001, (2002) Solid State Ionics, 154-155 and references therein. Ribes, M., Barrau, B. and Souquet, J.L. (1980) J. Non-Cryst Solids, 38-39, 271. Robinel, E., Carette, B. and Ribes, M. (1983) J. Non-Cryst. Solids, 57, 49 Roling, B., Martiny, C. and Funke, K. (1999) J. Non-Cryst. Solids, 249, 21)1. Rudnitskaya, A., Ehlert, A., Legin, A., Vlasov, Yu. and Buttgenbach, S. (2001) Talanta, 55, 425. Rykova, T.S., Borisova, Z.U., Pasin, A.V. and Kalinina, L.A. (1980) Fiz. Khim. Stekla, 6, 419. Salardenne, J., Morcos, J., Ait Allad, M. and Portier, J. (1990) Sensors Actuators B, 1, 385. Salmon, P.S. and Liu, J. (1996) J. Non-Cryst. Solids, 205-207, 172. Salmon, P.S. and Xin, S. (2002) Phys. Rev. B, 65, 064202. Schoning, M., Mourzina, Yu., Schubert, J., Zander, W., Legin, A., Vlasov, Yu. and Luth, H. (2001a) Sensors Actuators B, 78, 273. Schoning, M.J., Mourzina, Yu., Schubert, J., Zander, W., Legin, A., Vlasov, Yu.G. and Luth, H. (2001b) Electroanalysis, 13, 727. Siiptitz, P. and Willert, I. (1975) Phys. Stat. Sol. A, 28, 223. Swenson, J. and B6rjesson, L. (1996) Phys. Rev. Lett., 77, 3569. Thomas, M.P. and Peterson, N.L. (1984) Solid State Ionics, 14, 297. Tohge, N., Minami, T. and Tanaka, M. (1983) Yogyo-Kyokai-Shi, 91, 42. Tverjanovich, A., Tver'yanovich, Yu.S. and Loheider, S. (1996a) J. Non-Cryst. Solids, 208, 49. Tver'yanovich, Yu.S., Nedoshovenko, E.G., Aleksandrov, V.V., Turkina, E.Yu., Tver'yanovich, A.S and Sokolov, I.A. (1996b) Sov. J. Glass Phys. Chem., 22, 9. Tver'yanovich, Yu.S., Vlcek, M. and Tverjanovich, A. (2004) J. Non-Crys,~.. Solids, 333, 85-89. Vlasov, Yu.G. (1990) Zh. Anal. Khim., 45, 1279. Vlasov, Yu. (1991) Mikrochim. Acta [Wien], 2, 363. Vlasov, Yu.G. (1992) Zh. Anal. Khim., 47, 114. Vlasov, Yu.G. (1993) Sensors Actuators B, 15-16, 6. Vlasov, Yu.G. (1997) Ann. Chim., 87, 261. Vlasov, Yu.G. and Bychkov, E.A. (1982) Hung. Sci. lnstrum., 53, 35. Vlasov, Yu.G. and Bychkov, E.A. (1984) In Ion-Selective Electrodes-5 (Ed., Pungor, E.) Pergamon Press, Oxford, p. 243. Vlasov, Yu. and Bychkov, E. (1987) Ion-Select. Electrode Rev., 9, 5. Vlasov, Yu.G. and Bychkov, E.A. (1989) Anal. Lett., 22, 1125. Vlasov, Yu.G. and Bychkov, E.A. (1994) J. Electroanal. Chem., 378, 201. Vlasov, Yu.G., Bychkov, E.A. and Legin, A.V. (1994) Talanta, 41, 1059. Vlasov, Yu.G., Bychkov, E.A. and Legin, A.V. (1997) Zh. Anal. Khim., 52, 1184. Vlasov, Yu.G., Bychkov, E.A. and Seleznev, B.L. (1987) Solid State Ionics, 24, 179. Vlasov, Yu.G., Bychkov, E. and Seleznev, B. (1990) Sensors Actuators B, 2, 23. Vlasov, Yu., Bychkov, E., Tsegelnik, V. and Ben-Shaban, Y.M. (1995) Vestnik St Peterburg. Univ. Ser., 4(25), 50. Vlasov, Yu.G., Bychkov, E.A., Tver'yanovich, Yu.S., Nedoshovenko, E.G., Borisova, Z.U., Avdeev, A.V. and Antonov, P.P. (1987) USSR Patent No. 1,402,913. Vlasov, Yu.G. and Legin, A.V. (1998) Fres. J. Anal. Chem., 361, 255. Vlasov, Yu.G., Legin, A.V. and Baidakov, D.L. (1996) Zh. Prikl. Khim., 69, 1436. Vlasov, Yu., Legin, A. and Rudnitskaya, A. (1997) Sensors Actuators B, 44,532. Vlasov, Yu., Legin, A. and Rudnitskaya, A. (2002a) Anal. Bioanal. Chem., 373, 136. Vlasov, Yu., Legin, A. and Rudnitskaya, A. (2002b) In Sensor Update, Vol. 10 (Eds., Fedder, G.K. and Korvink, J.G.) Wiley-VCH Verlag GmbH, Weinheim, p. 143. Vlasov, Yu.G., Legin, A., Rudnitskaya, A., Di Natale, C. and D'Amico, A. (1996) Zh. Prikl. Khim., 69, 958. Vlasov, Yu.G. and Tarantov, Yu.G. (1989) Chemical Sensor Technology, Vol. 2 (Ed., Seiyama, T.) Elsevier, Amsterdam, p. 173. Warburg, E. (1884) Ann. Phys., 21, 622. Zhabrev, V.A. and Kazakova, E.A. (1982) Fiz. Khim. Stekla, 8, 51. Zhou, W., Sayers, D.E., Paesler, M.A., Boucher-Fabre, B., Ma, Q. and Raoux, D. (1993) Phys, Rev. B, 47, 686.
CHAPTER
4
RARE-EARTH DOPED CHALCOGENIDE GLASS Yu. S. Tver'yanovich and A. Tverjanovich ASSISTANT PROFESSOR,DEPARTMENT OF CHEMISTRY, SAINT-PETERSBURGSTATE UNIVERSITY, SAINT-PETERSBURG, 198504, RUSSIA
This chapter is devoted to new group of luminescence materials which are now being actively investigated: glassy semiconductors (chalcogenide glasses) doped with rareearth ions (REI). We enumerate the unique properties of this group of materials, which attract interest for prospective use as a matrix for rare-earth doping. We describe the glassy systems, which are the most promising for the synthesis of effective luminescence materials. This chapter also contains results of investigations of the structure and composition of the nearest neighbors of REI in chalcogenide glasses as well as Stokes and anti-Stokes luminescence, including a specific method of REI excitation via optical pumping using band-band absorption of matrix.
1. Chalcogenide Glass Properties and Luminescence Materials When considered for their use as hosts for the introduction of REI, the properties of the glassy semiconductors can be divided into two groups. In the first group are properties of the glassy semiconductors that do not contain REI and particularly those properties that make them attractive for optoelectronics and fiber optics applications. These properties include the transparency of glassy semiconductors in the IR range, the optical compatibility of glass and crystal semiconductors when used in microelectronic devices for generation, processing and detecting of light signals. The glassy semiconductors possesses a high factor of optical non-linearity and has highresolution photoresist properties. They can be easily deposited as thin films by thermal evaporation and are capable of changing optical properties under light. This polyfunctionality opens an opportunity of unification of technological processes for manufacturing a wide spectrum of devices of optoelectronics and fiber optics. It is natural that in such situations further expansion of a spectrum of possible applications of semiconductors by introduction of REI additives in their structure is needed. Properties of the glassy semiconductors directly favorable to their use as luminescent hosts activated with lanthanoids concern the second group of properties. Chalcogenide glasses possess low phonon energy, caused by low frequency of vibration of metalchalcogen bonds. Frequency of atom vibration in chalcogenide glasses is even less than that in halogenide (fluoride) glasses known for low phonon energy. Low phonon energy results not only in the wide band of transparency in IR range but also in reduction 169
Copyright 9 2004 Elsevier Inc. All rights reserved. ISBN 0-12-752189-5 ISSN 0080-8784
170
Yu. S. Tver'yanovich and A. Tverjanovich
of probability of non-radiated transitions of REI from the excited levels. The relaxation from the excited level can occur due to the radiated transitions used for optical amplification or up-conversion, and on account of non-radiated transitions that reduce quantum efficiency of luminescence. One of the reasons of non-radiated transitions is the so-called multiphonon relaxation. The probability of non-radiated relaxations due to such multiphonon contribution depends on several factors and can be expressed by the following equation (van Dijk and Schuurmans, 1983) W NR "~ ~ exp[-
a(AE
-
2humax)]
(1)
where/3 and c~are the parameters of glass matrixes, AE the energy gap between REI levels and h vmax the greatest phonon energy of glassy matrix. Thus, the probability of nonradiated transitions for the account of multiphonon relaxation exponentially decreases on reduction of phonon energy. In the work by, e.g., Hector, Medeiros Neto, Samson, Brown, Jedrzejewski, Wang, Taylor, Laming, Wylangowski and Payne (1994), for Pr 3+ in oxide glasses the transition from level 1G4 to level 3H5 occurs practically completely nonradiatively (/~'max/C "~ 1100 cm-1), for Pr 3+ in ZBLAN glasses ( P m a x / C "-" 600 cm -1) quantum efficiency of the given transition is already about 5 % (Carter, Wang, Brady, Kluth, Hewak, Brocklesby and Payne, 1998), and in the case of glasses of G a - L a - S system for which maximal phonon energy corresponds to 400 cm- 1 quantum efficiency reaches 68% (Hewak, Wyatt, Szebesta and Davey, 1991). The further reduction of oscillatory energy of glassy matrix due to transition to heavier chalcogen increases the quantum efficiency up to 90% (emission of ion Dy 3+ on the wavelength 1.3/xm in glasses of A s - G e - S e system (Shaw and Izumitani, 1993)). Similarly, the quantum efficiency of luminescence Er 3+ (from l e v e l 453/2) changes from 0.076 up to 47.5% at reduction of the maximal phonon energy of glassy matrix from 1100 to 500 cm- 1 (Zou, Kim, Jo, Hahn and Choi, 1989). This feature of glassy semiconductors has led to their intensive research regarding their prospective use in fiber optical amplifiers. The major feature of the glassy semiconductors is the high degree of covalence of the chemical bonds that form them, which is not less than 87% according to Pauling. This circumstance determines both a high parameter of refraction and the semi-conductor nature of this group of materials. Indeed, glassy semiconductors possess the refraction index (n) changing in limits from 2.2 up to 3.4, while the refraction index (n) of oxide glasses usually does not exceed 2, and fluoride glasses have smaller values. Increase in the refraction index leads to increase in the probability of radiating transitions (W R) in power dependence W R "~ (Sedn(n 2 +
2)2/9 -k- n3Smd)
(2)
where Sea and Smd are the forces of electric and magnetic dipoles, respectively. The semiconductor nature of the chalcogenide glasses predetermines the affinity of their energy gap and energy of REI levels. This circumstance creates an opportunity of existence of various effects caused by the energy exchange between REI and the matrix of glassy semiconductor. Research in this interesting field is described in Section 4. The high share of the covalent component of chemical bonds in chalcogenide glasses opens an opportunity to realize laser transitions between closely located energy levels. An important attribute of chalcogenide glass is that limiting concentration of the REI added in glass and the uniformity of their distribution in some cases exceed these
Rare-Earth Doped Chalcogenide Glass 30 2826 24 2 22 201816. 14 12 2 1086420-2
171
As2Se3 ]
As2S3 I AszTe3I ........ As203 I ....... Si02 I
io I I
l
i
J
i i i
" "
I
,
:i
,j
,
i
i
:
........
0
/
i
,,:,~ ! -,.-..._L : ! ...... "~--.I
/ /
i
I
/
.~-, S ' S
:
J ,
I
,
;0
,
,
I
~tm FIG. 1. Absorption spectra of As2X 3 (X -- O, S, Se, Te) and SiO2.
parameters for oxide glasses and is comparable with halogenide glasses. This is reflected in the shift of the concentration quenching in the side of big concentration Er 3+ for chalcogenide glasses in comparison with halogenide glasses (Cho and Risbud, 1994), and even more so in comparison with oxide glasses (Rapp, 1986). Moreover, photo-induced changes of the refraction index and the coefficient of absorption allow to form the wave guides for integrated optics. To conclude a list of the advantages of chalcogenide glasses as a matrix for introduction of REI, it is necessary to note their chemical resistance, rather low temperatures of softening, and the opportunity of drawing fibers. At the same time, chalcogenide glasses as luminescent materials have a number of defects. The short-wave edge of absorption in chalcogenide glasses lies in the visible range of the spectrum, limiting their application as up-converters. Optical loss of a signal is increased in comparison with oxide glasses due to the higher coefficient of absorption in the range of transparency. Transmission spectra of fused quartz, arsenic oxide and arsenic chalcogenides are shown in Figure 1 for comparison (Kumpta, Cole, Sanghera, Aggarwal and Schaafsma, 1998). An additional feature of glassy semiconductors is the strong self-focusing of laser radiation due to change of refraction index. One of the reasons is the large temperature factor of width of the forbidden band. In glasses doped with REI this can interfere with laser generation at high pump power (Viana, Palazzi and LeFol, 1997).
2.
The Glassy S e m i c o n d u c t o r s Used as a Matrix for Introduction of L a n t h a n o i d s
2.1.
T H E BASIC REQUIREMENTS FOR A GLASS MATRIX
Certainly, the first requirement for a glassy matrix for introduction of lanthanoids is the ability of the system to form a glass with large concentration of REI as well as maximum
172
Yu. S. Tver' yanovich and A. Tverjanovich
uniformity of REI in a glass matrix. It is desirable that the temperature of synthesis and rate of quenching should not be extremely high. For use of glass in fiber optics there is an additional requirement: a significant difference of temperatures of crystallization and softening. The glass should also possess an interval of an optical transmission as required depending on the expected application. To increase luminescence efficiency, it is necessary that the refraction index should be as large as possible and the energy of phonons should be as small as possible. The latter requirements conflict with the condition of transparency in the visible range of a spectrum. For example, replacement of S with Se in a glass matrix essentially reduces the energy of phonons (at about --~ 150 cm- 1 or at 40%) and increases the refraction index, but the fundamental ,edge of absorption is shifted at --~300 nm in long-wave side falling outside the visible range. It is necessary to take into account that oscillator strength and intensity of luminescence of REI are dependent on its environment: the increase of the covalent share of the chemical bonds and growth of asymmetry of a crystal field increase the oscillator strength (Krupke, 1966). 2.2.
THE GENERAL LAWS OF CHOICE OF GLASS-FORMINGSYSTEM
Sulfides of rare-earth elements Ln2S3 (Ln -- La-Er) are not glass-formers but form glasses with chalcogenides of other metals such as GazS3, A12S3, GeS2 and AszS3. The basic component of glassy matrixes for introduction of REI is GazS3. GazS3 has the ability to form glass with significant concentrations of rare-earth elements in comparison with such typical glass-formers as AszS3 and GeS2. So the glass-forming region in GazS3-LazS3 system reaches 50-85 mol% GazS3. With an increase in the atomic number of a rare-earth element, the glass-forming region decreases (Kumpta and Risbud, 1994), which is due to a simultaneous decrease of the glass-forming temperature and an increase of the eutectic temperatures (Cervelle,, Jaulmes, Laruelle and Loireau-Lozac'h, 1980). For Dy glass-forming region does not exceed 4 mol% (it lies between 63 and 67 mol% GazS3). Up to 40 mol% LazS3 can be introduced in GeS2 (Kumpta and Risbud, 1994), but such large glass-forming regions are apparent. The initial introduction of rare-earth sulfide results in phase segregation (Kumpta and Risbud, 1990). Homogeneously no more than 0.01 mol% of REI can be introduced in GeS2 (Simons, Faber and De Waal, 1995a; Marchese and Jha, 1997). According to Kadono, Higuchi, Takahashi, Kawamoto and Tanaka (1995) it is not possible to synthesize homogeneous glasses in GeSz-LazS3 and AszS3-LazS3 systems. Solubility of REI in glasses of ternary system G e - A s - S is a little bit higher, than in corresponding binary system, nevertheless, the tendency of REI to cluster formation remains very high (Shim, Cho and Heo, 1996). The ability of GazS3 to reduce crystallization ability of lanthanoid sulfide is caused by high complex ability of Ga 3+ ion. In glasses Ga is four coordinated by atoms of chalcogens (Loireau-Lozac'h, Keller-Besrest and Benazeth, 1996; Tverjanovich, Tver'yanovich and Loheider, 1996). The superfluous charge of such complex structural unit is compensated by the ion of a rare-earth element that results in stabilization of the complex and increase of solubility of lanthanoid in GazS3. Ion A13+ possesses a greater complex-forming ability than Ga 3+. Therefore it was necessary to expect the increase in the concentration limit of solubility of the REI in glasses on this basis. Because of the very high hygroscopicity of these glasses and their interaction with walls of quartz ampoules it complicates not only their practical application but also research.
Rare-Earth Doped Chalcogenide Glass
173
Unlike GeS2 or As2S3, Ga2S3 in itself is not a glass-former. To increase the glassforming ability of glasses with GazS3 basis, the second components used are either the above-mentioned typical glass-formers, or halogenide of alkaline (alkaline-earth) metals to form gallium-containing complex GaX4 M+. The second method allows an additional feature, expansion of the transmission range to the blue part of the spectrum. So, e.g., CsGaS3/2C1 glass is transparent practically in all the visible range (Tver'yanovich, Nedoshovenko, Aleksandrov, Turkina, Tverjanovich and Sokolov, 1996). The systems belonging to the first group are such as GazS3-GeS2, GazS3AszS3. It is possible to attribute to the second group the following systems: GazS3NazS (Ishikawa, Tawarayama, Ito, Aoki, Yanagita and Toratani, 1996), GazS3-CsC1, GazS3-LazS3, etc. Frequently, both of these variants are used simultaneously for improving the characteristics of the glasses (e.g., GazS3-GeSz-CsBr system (Griscom, Adam and Binnemans, 1999)).
2.3. 2.3.1.
SOME OF THE MOST INVESTIGATED GLASS-FORMING SYSTEMS
Glasses on the Basis of Ga2S3-GeS2 System
One of the most investigated glassy matrixes for introduction of an REI is system GazS3-GeS2. Up to 10mol% NdzS3 and not less than 10mol% ErzS3 can be incorporated into this system (Fig. 2a and b, respectively) (Man'shina, Kurochkin, Degtyarev, Grigor'ev, Tverjanovich, Tver'yanovich and Smirnov, 2001). The glassforming region is extended in the direction from eutectic composition in quasi-binary system xGazS3-(1 - x ) G e S 2 (x = 0.15) to eutectic composition in system xErzS3(1 - x)Ga2S3 (x -- 0.33). The glass-formation in ternary system G a - G e - S without rareearth additives was investigated in a number of works, and we note the most recent work by Abe, Takebe and Morinaga (1997). The glass transition temperature (Tg) in GazS3GeS2 system regularly decreases with increases in GazS3 content, but introduction of the rare-earth sulfide results in reduction of this effect (Fig. 3). The deviation from stoichiometry corresponding to pseudo-binary section of the GazS3-GeS2 system leads to decrease of Tg (Abe et al., 1997). It is necessary to note that there are significant distinctions in the value of Tg measured by various authors. For instance, according to Loireau-Lozac'h and Guittard (1975) the glass transition temperature for all glasses of the GazS3-GeS2 system is identical and equal to 683 K; according to Messaddeq, Siu Li, Lezal, Ribeiro and Messaddeq (~2001) for 0.17GazS3-0.83GeS2 glass Tg -- 673 K, for 0.205GazS3-0.795GeS2 glass Tg = 706 K (Saffarini, 1994). The refraction index of glasses of the given system is in a range which is typical for sulfide glasses from --~2.0 up to --~2.2 depending on structure and conditions of the glass preparation. For example, refraction index of the glass composition 0.17GazS30.83GES2 with 0.4 wt% Nd is equal to 2.178 on the wavelength 587.6 nm (Abe et al., 1997). The increase of the Ga contents results in insignificant increase in the refraction index (Abe et al., 1997) that is caused on the one hand by increase of density of a glass and on the other hand by the greater polarization of Ga ion in comparison with Ge ion. Glasses of this system are transparent in a wide range of wavelengths--from the visible region (460-500 nm) (Fig. 4) up to IR region (--~ 12,000 nm) (Fig. 5), which
174
Yu. S. Tver' yanovich and A. Tverjanovich
(a)
O_ 100
20/
. ~ 8 0
(~) -
40 ,,/___.......
\ ,, ~,,~.. 60
~/
60 0
20
.40 6'0
40 Ga2S3
(b)
GeS2 0,~ 100 1 0 ~
60J 0 Er2S3
.... ~ / t
10
N~/! 20
90
"~A ,~o//'-.'~(/, 30
40
50
h,40 60 Ga2S3
FIc. 2. (a) Glass-forming region in the ternary GeS2-Ga2S3-Nd2S3 system. (b) Glass-forming region in the ternary GeS2-Ga2S3-Er2S3 system. Open circles mark glass, solid circles mark glass-crystal, solid squares mark crystal.
makes them of interest as optics materials in visible and IR ranges. Among glasses of the ternary G a - G e - S system the compositions belonging to the quasi-binary section Ga2S3-GeS2 are the most investigated; however, that they have been the focus of research is not necessarily related to the optimality of their properties. According to Shin and Heo (1999) deficiency of sulfur relatively of the stoichiometric section raises the solubility of REI. At the same time for drawing an optical fibre, a surplus of sulfur which essentially lowers Tg is more preferable. So for 0.02Ga2S3-0.98GeS3 glass Tg equals 275 ~ (Schimmel, Faber, de Waardt, Beerkens and Khoe, 2001), and for this composition concentration quenching is not revealed up to 700 ppm Pr. The introduction of 25 mol% CsX (X - C1, Br, I) in glass containing 25 mol% Ga2S3, 50 mol% GeS2 and 0.5 at.% Nd 3+ leads to reduction of the refraction index by 0.1, increase of Tg by 10 K and increase in optical width of the forbidden band (Griscom et al., 1999). At the same time, according to Marchese and Jha (1997), the absorption edge for 15Ga2S375GeS2-10CsI glass is located at 550 nm, while for a composition without cesium
Rare-Earth Doped Chalcogenide Glass
175
•
500 -
-r
480 460 440 420 400
38o 360 340 320 300
9
0.0 at % REI 0.4 at % Nd
9
1.0 at % N d
=
280 x
260 I
~
" []
[]
1.2 at % Er '
I
5
'
10
I
'
I
15
'
I
20
'
I
25
'
30
I
I
35
40
'
I
45
Ga2S 3, mol % FIG. 3. Dependences of Tcr and Tg on Ga2S3 contents for the 0.15GazS3-0.85GeS2 system.
chloride (15Ga2S3-85GeS2) it is located at 450nm. Both glasses were doped with 2000 ppm wt Pr 3+. 2.3.2.
Glasses Based on the G a 2 S 3 - L a 2 S 3
System
As has been described, the glasses in the Ga2S3-La2S3 system can be prepared by quenching in water with Ga2S3 contents from 50 up to 85 mol%. In comparison with
........ 0.65La2S30.35Ga2S3: 1.2at.% Er 0.15Ga2S30.85GeS2: 1.8at.% Er
""'"....~
r~9
...............
9
,
00
I
500
,
I
,
600
I
700
,
I
800
,
I
900
nm FIG. 4. Absorption spectra of 0.85GeSz-0.15GazS3 glass doped with 1.8 at.% Er and 0.65La2S3-0.35Ga2S3 glass doped with 1.2 at.% Er in the visible spectral range.
176
Yu. S. Tver'yanovich and A. Tverjanovich
100
r
80
6o ,.Q
4O ~.+...--
~ 0.15Ga2S3-0.85GES2: Er ......... 0.65Ga2S3-0.35La2S3: Er
20
o
,
I
5
,
1'0
,
1'5
,
I
20
,
)~,~tm FIG. 5. Absorption spectra of 0.85GeS2-0.15Ga2S3 glass doped with 1.8 at.% Er and 0.65La2S3-0.35Ga2S3 glass doped with 1.2 at.% Er in the IR spectral range.
glasses of the Ga2S3-GeS2 system, the absorption edge caused by electronic transitions is shifted in long-wave side (Fig. 4). The edge of absorption related to multiphonon absorption is somewhat shifted in short-wave regions, which can be due to the increased probability of formation of oxides during the synthesis of glasses from elements (Fig. 5). The glass transition temperature is essentially higher than in the germaniumcontaining system and is equal to about 890 K (Kumpta and Risbud, 1994). It is necessary to note the high value of the refraction index: n - - 2 . 4 - 2 . 8 (at A = 400-700 nm) (Kumpta and Risbud, 1994). Calculation of the minimal losses results in a value of 0.5 dB km-1 at wavelength 3500 nm (Brady, Schweizer, Wang and Hewak, 1998). High temperatures of synthesis and glass transition and high crystallization ability reduce the appeal of the glasses of this system to practical application. Addition of alkaline metal halogenide to glasses of the GazS3-LazS3 system results in decrease of Tg. If the alkaline metal halogenide is entered instead of lanthanum sulfide then the value of the refraction index decreases too. For example, the refraction index of the glass 0.65GazS30.05LazS30.30CsC1 doped with 500 ppm wt PrzS3 is 2.00 while the glass with inverse relative contents of sulfide lanthanum and cesium chloride (0.65GazS30.30LazS30.05CsC1) possesses a refraction index equal 2.40 (Hector et al., 1998). If the halogenide of the alkaline metal is entered instead of the gallium sulfide, then the refraction index increases from n = 2.253 (A = 1.5/zm) for the glass 0.70GazS30.20LazS30.10CsC1 up to n = 2.379 for the glass 0.40GazS30.20LazS3 0.40CsC1 (Marmolejo, Granado, Alves, Cesar and Barbosa, 1999). The other variant of improvement of the technical characteristics of glasses of the GazS3-LazS3 system is introduction of GeS2, i.e. pooling of the two systems: GazS3-LazS3 and GazS3-GeS2. Thus in the system (70 - x)GaS3/z-xGeSz-27LaS3/2 on increase of x from 0 up to 35 the Tg decreases, the fundamental absorption edge is shifted in long-wave range, the upconversion luminescence efficiency of ions Er 3+ (Kadono et al., 1995) decreases, and
Rare-Earth Doped Chalcogenide Glass
177
the refraction index varies from 2.408 up to 2.457 (Higuchi, Takahashi, Kawamoto, Kadono, Ohtsuki, Peyghambarian and Kitamura, 1998b).
2.3.3.
Glasses Based on the GeS2-As2S3-Ga2S3 System
For Ge25(ASlo-xGax)S65 section the color of glasses changes from yellow at x - 10 to red at x - 2 (position of the absorption edge changes from 480 to 580 nm) and with further increase of the AszS3 contents it is again displaced in short-wave region reaching up to 510 nm at x - 0 (Aitken and Quimby, 1997). The shift of absorption edge at the introduction of AszS3 was described in the work by Belykh, Glebov, Lerminiaux, Lunter, Mikhailov, Plyukhin, Prassas and Przhevuskii (1997), in which they remarked that color of some glasses of system G e - G a - A s - S varies from red to dark red. The increase of the AszS3 contents results also in a downturn of Tg, increased crystallization stability and in an increase of the refraction index from 2.20 to 2.29. However, such change of the refraction index could be caused by change of width of the forbidden band since the refraction index was measured at the wavelength 589 nm. Solubility of REI decreases on increase of the AszS3 contents as one would expect. For composition 13.9AszS3-2.8Ga2S3-83.3GeS2 the limiting contents of the Pr are about 4000 ppm. But the concentration quenching reduces the lifetime of the excited levels of Pr 3+ at concentrations already above 900 ppm (Aitken and Quimby, 1997). In glasses of the given system it can be homogeneously entered up to 0.4 at.% of Nd as well (Belykh et al., 1997). Just as in the case of the GeSzGazS3 system, attempts to improve the characteristic of glasses of the G e - A s - G a - S system by introduction of cesium halogenides have been undertaken (Shin, Yang and Heo, 2002). It is found that glass of composition 0.7(Ge0.zsASo.lSo.65)-0.15GaS1.50.15CsBr doped with 0.05 mol% Dy2S3 is characterized by high crystallization stability, and lifetime of the level 6Fll/2 -~- 6H9/2 achieves 1730/~s, which is 17 times higher than in glasses without CsBr. Thus the quantum efficiency grows approximately four times. The importance of the equi-molar parity GAS1.5 and CsBr is emphasized.
2.3.4.
Other Components for Improving Properties of the Glasses
The introduction of heavy element chalcogenide (such as lead or thallium chalcogenide) in a glass is a prospective means to optimize glassy matrix composition, but research is thin. The introduction of PbS in glasses of the GeSz-GazS3 system leads to the increase of the refraction index (Aitken and Quimby, 1997). However, the measured lifetime of the excited REI levels decreases, which is apparently due to the large concentration of oxygen in the samples because of the reaction of lead with the walls of quartz ampoule. In turn, the introduction of light alkaline-earth metals such as sodium results in a shift of absorption edge in short-wave range and a drop in melting temperature. Improvement of these given properties is accompanied by the reduction of the oscillator strength and essential concentration quenching (Viana et al., 1997).
178 0
Yu. S. Tver'yanovich and A. Tverjanovich
Structure and Composition of the Environment of Lanthanoids in Chalcogenide Glasses
The introduction of lanthanoids in chalcogenide glass-forming alloys results in a sharp decrease in their glass-forming ability. Chalcogenide of rare-earth element is the primary crystallizing phase in these glasses. Moreover, before the appearance of these crystal nucleases, microareas enriched with structural units containing lanthanoid atoms are formed in the melt. In the case of luminescence it results in increase of the concentration quenching. The structure and composition of the environment of the REI in a glass network also influence a number of parameters that determine luminescent properties (such as oscillator strength, spectral position of the bands corresponding to optical transitions of REI, energy position of REI levels in band structure of the glassy semiconductor, etc.). The aforesaid testifies the necessity of studying and managing the nearest and medium structural order around lanthanoids. The simplest method to determine the composition of the first coordination sphere of REI is based on the effect of spectral shift of absorption bands of lanthanoids at the change of the degree of ionicity which is caused by the change of intensity of local electric fields on the ion of lanthanoid (Tver'yanovich, Degtyarev and Pivovarov, 1998). The dependence of spectral location of the absorption band Nd 3+ in glasses of various compositions (in the range 7 9 0 - 8 2 0 nm, transitions 419/2-2H9/2 and 419/2-4F5/2) on electronegativity of anions of the glassy matrix is shown in Figure 6. It is seen that the absorption bands shift to the long-wavelength region of the spectrum on transition from anions with large electronegativity to anions with small electronegativity in the order: F ~ O ~ S ~ Se. The distinction in spectral location of the specified bands for aluminosilicate and phosphate glasses attracts attention. This is not surprising, as it is well known that in the case of phosphate glasses it is correct to speak not about the anions of oxygen but about the phosphate anions. However, the method of optical
12600 -
, /1
1255012500"Z 12450-
3
1240012350 4,
12300 12250 2.0
"
2'.5
'
310 " 3'.5 7, arb. units
'
4113
'
4'.5
FIG. 6. Dependence of spectral location of N d 3+ absorption band on electronegativity of anions for (1) fluoride; (2) phosphate; (3) silicate and (4) chalcogenide glasses.
Rare-Earth Doped Chalcogenide Glass
179
chemical shift can be sensitive not only to the structure of first but also of the second coordination sphere. This method has been applied to the study of environment of the ions Nd 3+, pr3+,Sm 3+, Yb 3+ in the glassy matrix of composition 0.85GeChz-0.15GazCh3 where Ch is (1 - x - y)S-(x)Se-(y)Te. As glass contained various anions which are chemical analogues (oxygen, sulfur, selenium, tellurium) the average value of electronegativity (c0 designed in view of the molecular fraction of each type of anions was used. In Figure 7 a - d the received dependences of spectral location of absorption band of Nd 3+, Pr 3+, Sm 3+, Yb 3+ on the average electronegativity of the elements forming anionic sublattice of the glass are depicted. Strictly speaking, the ionicity of the chemical bond is not a linear function of the difference of electronegativity at change of the latter in a wide range of values. However, taking into account the narrowness of the interval of change of electronegativity, it was supposed that linear approximation is carried out. The influence of structure of the second coordination sphere of REI can be ignored. The linearity of the dependence of spectral location of the absorption bands on the average electronegativity of anions in case of Nd 3+ ions (Fig. 7a) can be interpreted as a result of the coincidence of average composition of the first coordination sphere neodymium with the average anionic composition of the glass. At the same time the deviation from the linear dependence in the case of ions of the Pr 3+, Sm 3+ and Yb 3+ (Fig. 7b-d) can be explained by the increase of the fraction of sulfur (the more electronegative element) in the environment of the REI in comparison with the average anionic composition of the glass. However, the given conclusions demand additional substantiation. There is a question not only about the chemical composition of the first coordination sphere of REI but also about the coordination number of REI. The degree of ionicity of the chemical bonds formed by REI in chalcogenide compounds compared to chalcogenidesglass-formers is great enough. Therefore the rule of Pauling for structures of ionic crystals (Pauling, 1929; Evans, 1964) can be applied to a certain extent to the glasses containing REI (Wang, Brocklesby, Lincoln, Townsend and Payne, 1993). Coordination of the cation satisfying minimum of the free energy varies depending on the ratio of ionic radius of the cation and anion. So at the ratio of radii in a range from 0.414 to 0.732, the most probable coordination number of cation is equal to 6. With an increase of the value of the given parameter, the coordination number can increase up to 9. The ionic radius of REI varies depending on the atom number from 1.15 A (for La 3+) to 0.85 A, with an increase of atom number the ionic radius decreases due to the effect of compression (Topp, 1965). The ionic radius of anion S 2- is about 1.84 ,~. Thus, in sulfide glasses, according to Pauling rule, all REI should be six-coordinated. The chlorine anion has practically the same ionic radius as the anion of sulfur (1.81 A), therefore the replacement of sulfur by chlorine should not result in change of coordination number of a REI. However, the replacement of sulfur by oxygen (R(O 2-) -- 1.40 A) and particularly by fluorine (R(F-) = 1.36 A) should result in the increase of coordination number. Thus, high coordination contradicts Phillips's topological rule, which suggests that maximal stability of the covalent network of glass is observed at average coordination number 2.45 (Phillips, 1979). The given structural incompatibility of REI sulfides with typical glass-formers, combined with their very high melting temperatures when compared to germanium sulfides (and even more so with the arsenic sulfides) results in
Yu. S. Tver' yanovich and A. Tverjanovich
180 (a) 1.126 1.124 1.122 1.120 1.118
S
Nd 3+ 2.'40
2.'44
2.~48
2.'52
(b) 0.6340.633 0.632 0.631
pr 3+
7 "(c)
0.630 2.38
2.'40
2.'42
2.'44
2.'46
2.'48
2.'50
2.52
1.042 1.040 1.038 1.036 1.034 1.032
Sm 3+ 2.'40
2.'42
2.'44
2.'46
2.'48
2.'50
(d) 1.00
Y
0.99 0.98 0.97 0.96
yb 3+ 2.'40
2.'42
2.'44
2.'46
2.'48
2.'50
FIG. 7. Dependences of spectral location of absorption band Nd 3+, Pr 3+, Sm 3+, Yb 3+ on the average electronegativity of anionic sublattice.
Rare-Earth Doped Chalcogenide Glass
181
the formation of the micro-inhomogeneities enriched by REI on their introduction in glass (Kadono et al., 1995). Inhomogeneities in glasses lead to the occurrence of the first sharp diffraction peak (FSDP) on the dependence of the structural factor on a wave vector at scattering of X-rays (Moil and Arai, 1983; Elliott, 1991). Thus it is supposed that glass consists of internally self-coordinated microareas of the nanoscale size rather poorly connected with each other. It is considered that FSDP characterizes the size of these microareas--correlation length (Rc) according to the following relation (Bojesson, Torell, Dahlborg and Howels, 1989; Susman, Price, Volin, Dejus and Montague, 1989) (3)
Rc ~ 2,rr/AQ
where AQ is half-width at half-maximum of the peak. The dependence of the correlation length calculated according to relation (3) on the concentration of neodymium sulfide in the (0.15Ga2S3-0.85GeS2)(1-x~-(Nd2S3)x system is shown in Figure 8 (Tver'yanovich, Mamedov and Degtyarev, 2000). REI in chalcogenide glasses form the environment corresponding to their own coordination properties instead of intrinsic for network of glass. As has been mentioned, they realize coordination numbers approximately twice greater than the average coordination number characteristic for the glassy matrix. The model describing the average order in glasses containing REI has been based on this distinction (Tver'yanovich et al., 2000). Let us suppose that glass consists of structural elements of two different types, which differ in the nature and sizes. The initial glassy matrix is characterized by correlation length L. Around the rare-earth atoms the area with other structure described in the correlation length I is formed. If the volume fraction of/-areas is much less than volume fraction of L-areas the effective correlation length of glass-containing REI can be calculated as average value of correlation lengths of these two areas taking into account
20"
< d
15-
-%
10-
'
0.0000
I
'
0.0001
I
'
0.0002
'
0.0;03
'
0.0004
Nd concentration, -3A FIG. 8. Dependence of correlation length on Nd 3+ contents for the 0.15Ga2S3-0.85GeS2:Nd2S3 system. Points are values calculated from the experimental data. Curves 1 and 2 are approximations with Eqs. (4) and (5), respectively.
182
Yu. S. Tver' yanovich and A. Tverjanovich
their volume fractions R e -- 14n -+- L(1 - 13) = L - (L -- l)13n
(4)
where n = N A X p / M is the concentration of REI, x the atom fraction of REI, M the average atom weight, p the density of glass and NA the Avogadro number. On increase of concentration of the REI there is a moment when the average distance between /-areas becomes less than L. It corresponds to the performance of the following inequality n(l + L) 3 --> 1. Taking into account high coordination ability of the rare-earth elements and large energy of their bonds with chalcogen, it is natural to assume that /-areas start to destroy L-areas. In this case the correlation length of the matrix becomes equal to the average distance between/-areas that is essentially less than L. Therefore the average correlation length for the glass-containing REI can be calculated from Eq. (5) having only one free parameterml:
The experimental points corresponding to composition with concentration Nd more than 10-4/~-3 are well described by Eq. (5) at 1 ~ 6.5 A (Fig. 8, curve 2). At smaller concentration of Nd the experimental points are described by Eq. (4) (Fig. 8, curve 1). It is possible to estimate the value of correlation length for a glass matrix by inserting the value 1 ~ 6.5 ,~ in Eq. (4). The estimated value is equal to L ~ 19 A. This value of correlation length is characteristic for typical golass-formers such as SiO2, AszS3, B203 for which it is estimated in the range 15-20 A. The structural inhomogeneities having the characteristic: sizes in some nanometers are also responsible for the occurrence of the low-frequency 'boson' peak in Raman spectra (Martin and Brenig, 1974). Besides that, the conformity between values of the correlation length appreciated with the help of both methods is marked (Novikov and Sokolov, 1991; Sokolov, Kisliuk, Soltwisch and Quitmann, 1992). There are two approaches explaining the occurrence of tile low-frequency peak. One of them is on the basis that the boson peak is caused by acoustic modes in the non-uniform environment with correlation length defined by the following equation (Martin and Brenig, 1974) Rc = vr/~rcco
(6)
where v is the average rate of transversal (vT) and longitudinal (vL) sound waves, 1Iv 3 = (2/(vr) 3 + 1/(vL)3)/3
(7)
c is the speed of light and co the frequency corresponding to the maximum of the peak. That is for l l configuration. For I1• configuration vr must be instead of v in Eq. (6). The second approach connects the occurrence of the peak with optical m o d e s m with orientationally uncorrelated anisotropic scattering centers (Mazzacurati, Nardone and Signorelli, 1979). Introduction of Pr 3+ and Dy 3+ in such chalcogenide glassy systems as G e - S - I , G a - L a - S and G a - A s - S results in increases of intensity of the boson peak (Tikhomorov, Jha, Perakis, Sarantopoulou, Naftaly, Krasteva, Li and Seddon, 1999). From the particular level of REI concentration resulting from the glassy
Rare-Earth Doped Chalcogenide Glass
183
matrix composition, the intensity of boson peak decreases, which is caused by the processes of cluster-formation of REI. Moreover, the introduction of REI in the glassy matrix results in the shift of boson peak location in short-wavelength range which, according to expression (6) (at the invariance or increase of sound rate in glass), demonstrates the reduction of the correlation length. Thus, this data correlates with the aforesaid analysis of behaviour of the correlation length based on the results of research of the FSDP. 4.
Stokes Luminescence
The need for optical amplifiers for wavelengths 1.3 and 1.5/.~m in telecommunication applications has increased interest in chalcogenide glasses doped with REI. Chalcogenide glasses doped with the Pr 3+, Nd 3+, Dy 3+ and Er 3+ ions are potential materials for such lasers. 4.1.
OPTICAL EXCITATION THROUGH THE ABSORPTION BAND OF THE LANTHANOID
4.1.1. Praseodymium The scheme of energy levels of Pr 3+ ion is shown in Figure 9. The excited ion radiates 1.3/~m at relaxation from the 16 4 level to the 3H5 level and 1.5/~m at transition from the 1D2 level to the 1G4 level. The spectrum of luminescence Pr 3+ in the G a - G e - S glass at excitation by radiation with wavelength 1.02/~m is represented in Figure 10 (Wei, Machewirth, Wenzel, Snitzer and Sigel, 1995). At this excitation wavelength there is up-conversion luminescence also (Fernandez, Balda, Mendioroz and Garcia-Adeva, 2001). The absorption spectrum of the GezsGasS7o glass containing 1 wt% Pr is depicted in Figure 11 (Wei et al., 1995). Analysis of absorption spectra according to the Judd-Ofelt theory allows an estimation of the radiative lifetime (~'r), which for the 1G4 level is equal to 511 ~s (Wei et al. 1995), almost six times shorter than in ZBLAN glasses (3.2 ms) (Ohishi, Kanamori, Kitagawa, Takahashi, Snitzer and Sigel, 1991). The shortened radiating time in glasses on the basis of G a - G e - S system is caused by higher values of Judd-Ofelt parameters (~(~2,~(~4,~6) and higher refraction index. These parameters reflect the force and symmetry of a field in which there is REI. Among three parameters, J22 is the most sensitive to the composition of the matrix. Its value increases on increase of the degree of covalency of bonds. For glasses based on the G a - G e - S system the value of the ~2 parameter is almost 10 times more than in ZBLAN glasses (Wei et al., 1995). At the same time, the measured emission lifetime (~'m) in sulfide glasses is 360/~s, which is more than in fluoride glasses (110 p,s). The lessened probability of non-radiative losses accounts for decreasing of the multiphonon relaxations. The quantum efficiency determined as/3 -- "/'m/'/'r for the radiating transition of ion Pr 3+ from the 1G4 level to the 3H5 level in the G a - G e - S glasses is equal to about 70%. Slightly different values for the said parameters have been obtained for just the same composition of the matrix (Gez6GasS69) by Park, Heo and Kim (1999). Though the emission lifetime is just the same (~'m = 327 ~s), the value of quantum efficiency is essentially smaller than 49% that
Yu. S. Tver'yanovich and A. Tverjanovich
184
22-
3P21I6 3P 1
20
3P0
18 1D2 16 -
14 -
O
1.50 Bm
0.605
12
-
10
-
Bm
1G4
8
6
3F4 3F3
4
3F2 3H6 1.02
2
3H5
0
3H4
9. Energy level diagram of Pr 3+. Pumping and radiative transitions are marked with solid arrows. Crossrelaxation transitions are marked with dotted arrows. FIG.
100 -
80
== 60
~
4O
r~
"
20
I
1200
,
I
1250
,
I
1300
,
I
1350
,
I
1400
,
I
1450
,
I
,
1500
~,, nm Fro. 10. Luminescence spectrum of Pr 3+ pumped at 1.02/xm in G a - G e - S glasses.
185
Rare-Earth Doped Chalcogenide Glass
r
3 2 1
0 I
,
I
500
,
I
1000
~
I
1500
~
I
2000
,
I
2500
3000
~,, nm FIG. 11. Absorption spectra of GezsGasS7o glass doped with 1 wt% Pr.
is connected with the difference in the estimated values of the O2, ~4, ~-~6 parameters. The alteration of the composition of the glassy matrix in the G a - G e - S system results in the change of values of T m and ~'r. The reduction of the sulfur contents' relatively stoichiometric composition leads to reduction of 3"m and Zr. However, Zr decreases faster than ~'m, therefore the quantum efficiency (determined as/3 = Zm/Zr) at the same time increases (Park et al., 1999). For composition Ge32GasS63, the value of quantum efficiency is equal to 66%. Perhaps the downturn of quantum efficiency at surplus of sulfur is due to the increase of probability of multiphonon relaxation because of the occurrence of high-frequency vibrations of the S - S bonds (--~470 cm-1) in the phonon spectrum. An illustration of the given tendency is depicted in Figure 12 (Park et al., 1999), demonstrating the dependence of probability of multiphonon relaxation on sulfur contents for glasses GexGasS ~-x containing 300 ppm wt Pr 3+. The formation of clusters
1.6
~
1.5 _
0
01
1.4
0"3
o_
1.3 1.2 1.1 1.0 0.9 63
'
64
65
'
66
6~7
'
68
'
69
'
at.% S FIG. 12. Dependence of multiphonon relaxation rate on concentration of S in GexGasS~_x glasses doped with 300 ppm wt Pr 3+.
186
Yu. S. Tver'yanovich and A. Tverjanovich
400 350 E 300 250 ~D ~D
200
9
150 1 O0 ,I
0.01
|
|
|
,
,
|
|
I
0.1 wt.% Pr
i
i
i
i
t
i
i
i
[
1
FIG. 13. The emission lifetime of the pr3+IG4 state in G e - G a - S glass as function of Pr content.
enriched with Pr 3+ ions blocks the increase of luminescence intensity because of the increase of the REI concentration. The experimental dependence of emission lifetime of the 1G4 level for glass of composition Ge25GasS7o on concentration of Pr 3+ ions is shown in Figure 13 (Park et al., 1999). The maximum concentration of Pr 3+ ions in glasses of the G a - G e - S system at which it is yet possible to avoid the concentration quenching is no more than 500 ppm wt (Fig. 13; Wei et al., 1995; Park et al., 1999). The probable channels of cross-relaxation responsible for concentration quenching are marked with dotted arrows in Figure 9. Besides that, the emission lifetime essentially depends on the concentration of OH- impurity in the given systems (Simons et al., 1995b; Marchese, Kakarantzas and Jha, 1996). The spectroscopic properties of Pr 3+ ions in glasses of the G a - L a - S - C s C 1 and G e - G a - S e systems are investigated by Hector et al. (1998) and Nemec, Frumarova and Frumar (2000), respectively.
4.1.2. Dysprosium The diagram of energy levels of Dy 3+ ion is shown in Figure 14. The 1.3/xm emission corresponds to the transition of excited ion from the dual l e v e l 6F 11/2 + 6H9/2 at the 6H15/2 level. The spectrum of luminescence Dy 3+ in glass G a - G e - S on excitation by radiation with wavelength 0.81/xm is represented in Figure 15 (Wei, Machewirth, Wenzel, Snitzer and Sigel, 1994). Maximum luminescence intensity (near the 1.3/zm) is observed at the wavelength of 1.34/zm. For the G a - L a - S matrix the maximum is located at 1.33 ~m (Tanabe, Hanada, Watanabe, Hayashi and Soga, 1995). The absorption spectrum of glass Ge25GasS70 containing 1 wt% Dy is depicted in Figure 16 (Wei et al., 1994). The calculated value of the radiative lifetime for d u a l 6F 11/2 % 6H9/2 level of the Dy 3+ in glass of composition Ge25GasS7o is equal to 227/zs (Wei et all., 1994). It is approximately twice shorter than the radiative lifetime of the 1G4 level of Pr 3+ in the same matrix. The measured emission lifetime achieves only 38/xs (concentration of Dy is 0.1 wt%). It is almost nine times shorter than that for the 1G4 level of Pr 3+ (Wei et al., 1994). Therefore the quantum efficiency of the d u a l 6F 11/2 + 6H9/2 level of the Dy 3+ in Ge25GasS7o glass responsible for the luminescence with wavelength 1.3 ~m achieves only 16.8%, while for
Rare-Earth Doped Chalcogenide Glass
187
6F3/2 6F5/2
12
6F7/2 6H5/2
10
6F9/2-I-6H7/2 6F11/2+ 6H9/2 5" 6
6H11/2
6H13/2
0.8 aa 0.9 gm
1.3 gm
1.1 gm r
6H15/2
Fro. 14. Energy level diagram of Dy3+. Pumping and radiative transitions are marked with solid arrows. Pr 3+ the similar parameter is almost four times higher (see Section 4.1.1). It is necessary to note that the energy gap between the 6F 11/2 -q- 6H9/2 level and the nearest below-lying level (6H 1l/Z) is equal to about 1850 cm -1. It is much less than the corresponding value for the ion of Pr 3+. The decrease of the given energy gap results in the increase of probability of multiphonon relaxation. As in the case of glasses G e - G a - S : P r , reduction of sulfur concentration relative to the stoichiometric composition leads to a decrease of non-radiative losses for glasses of the G e - G a - S : D y system (Shin and Heo, 1999). It is caused both by the reduction of probability of up-conversion processes and by an increase of solubility of Dy in the glassy matrix.
1.0 0.8 .~ 0.6 ~9 0.4 0.2 0.0 1000
15;0'
2~0 ' 2 5 ; 0 ' ~,, nm
3~0
'
35;0
FIG. 15. Luminescence spectrum of Dy3+ pumped at 0.82/xm in Ga-Ge-S glasses.
188
Yu. S. Tver'yanovich and A. Tverjanovich
T 2
ooo 1 'oo o'oo
o'oo
~,, nm FIG. 16. Absorption spectra of Ge25GasS7o glass doped with 1 wt% Dy.
4.1.3.
Neodymium
The absorption spectrum of glass of the (1 - x ) G e S 2 - ( x ) G a 2 S 3 system containing 1 at.% Nd is shown in Figure 17 (Man' shina et al., 2001). The scheme of energy levels of Nd 3+ ion is illustrated in Figure 18 (Man'shina et al., 2001). The excited Nd 3+ ion radiates 1.3/xm at relaxation from the 4F3/2 level to the 4113/2 level. The spectrum of luminescence of Nd3+-doped 0.85GeSz-0.15GazS3 glass excited by radiation with wavelength 0.81 ~ m is shown in Figure 19 (Man'shina et al., 2001). The maximum luminescence intensity (near 1.3/,~m) is observed at wavelength 1.36/zm. For the given composition of a glass the reduction of luminescence, intensity is observed at
4
0 450
1
,
I
500
,
I
550
FIG. 17. Absorption spectra of (1 doped with 1 at.% Nd.
,
I
600
,
I
650
,
700 750 ~, nm
- x)GeS2-(x)Ga2S3
800
850
900
950
glass (x = 0.1, 0.2, I).35; curves 1-3, respectively)
Rare-Earth Doped Chalcogenide Glass
189
~G5/2
G7/2 2H11/2 4F9/2 453/~F7/2 2 4F5/2H9/2
16 14 12
4F3/2
10 7
o
8 1800 nm 6
4115/2 1350 n m
4113/2 1060 nm
4Ill/2
815 nm 900 nm
419/2 FIG. 18. Energy level diagram of Nd 3+.
1.0-
~2 0.5
(D
Y.
0.0
i
800
900
lO00
l lO0
12100 1300
!
1400
1500
~, nm FIG. 19. Luminescence spectrum of Nd 3+ pumped at 0.81/xm in 0.85GeS2-0.15Ga2S3 glass doped with 0.2 at.% Nd.
190
Yu. S. Tver' yanovich and A. Tverjanovich 300 O
250 o 0.3Ga2S3-0.7GeS 2 0 85GeS 2 200
"~
150
100
50
. 0.0
, 0.2
.
, 0.4
.
, 0.6
, 0.8
.
, 1.0
.
, , , . , . 1.2 1.4 1.6
at% Nd FIG.
20.
D e p e n d e n c e o f l u m i n e s c e n c e i n t e n s i t y on c o n c e n t r a t i o n o f N d in (1 - x ) G e S z - ( x ) G a z S 3
glasses.
concentration of Nd exceeding 0.2at.% (Fig. 20). The display of concentration quenching at such small concentration of the REI testifies that the non-uniform distribution of Nd 3+ ions takes place in the matrix of glass. The increase of the GazS3 contents in glass results in the increase of solubility of Nd in the glassy matrix. So, the maximum dependence of luminescence intensity on concentration is shifted from 0.2 up to 0.8 at.% Nd at the increase of GazS3 contents from 1:5 to 30 mol% (Fig. 20; Man'shina et al., 2001). Calculations according to the Judd-Ofelt theory have been executed for the glass of composition 0.3LazS3-0.7GazS3 containing 0.2 mol% NdzS3 (Lima, Sampaio, Catunda, de Camargo, Nunes, Baesso and Hewak, 2001). The estimated value of radiative lifetime of t h e 4F3/2 level is equal to 78/~s. The measured emission lifetime is equal to 75/zs. The reduction of concentration of NdzS3 till 0.05 mol% results in insignificant increase in radiative lifetime,, up to 78/xs. Thus, the quantum efficiency achieves 100% at a concentration of 0.05 tool% NdzS3 in glass and 96% at concentration 0.2 mol% (Lima et al., 2001). The replacement of GazS3 by CsX (X = C1, Br, I) results in the increase of radiative lifetime up 1:o 104 ~s for a glass of composition 50GeSz-25GazS3-25CsC1 containing 0.5 at.% Nd. At that the measured emission lifetime of t h e 4F3/2 level is 121/xs and quantum efficiency exceeds 100% accordingly (Griscom et al., 1999).
4.2.
OPTICAL OF THE
EXCITATION GLASSY
DUE
TO INTER-BAND
ABSORPTION
MATRIX
Apart from the direct excitation of REI by radiation with the energy coinciding with that of the corresponding REI level, a mechanism of excitation in the chalcogenide glassy matrix is also possible, in which the stimulating radiation is absorbed by the matrix of the glass with subsequent transfer of energy to the REI (Gu, Ramachandran, Reuter, Turnbull and Verdeyen, 1995; Turnbull, Aitken and Bishop, 1999; Bishop, Turnbull and Aitken, 2000). The forbidden band of chalcogenide glasses in many cases
191
Rare-Earth Doped Chalcogenide Glass - 16000 70 / / / ~
14000
......... Ix,a~bm_~nits
60
12000
i
50 40
10000 8000 7 6000
9~ 30 "
4000
20
2000
10 -0 , 500
I 600
,
I 700
,
I 800 nm
,
I 900
,
I 1000
,
-2000 1100
FIG. 21. Absorption spectra of Ge33ASlzSe55 glass and luminescence excitation spectra of Ge33ASlzSess glass doped with 0.2 wt% Er.
overlaps some 4f transitions of the REI that leads to strong energy exchange between the chalcogenide matrix and REI levels. The luminescence excitation spectrum and spectral position of absorption edge of the glass Ge33ASlzSess containing 0.2 wt% Er is shown in Figure 21 (Gu et al., 1995). The energy position of the maximum of the curve of excitation coincides with the absorption edge of a glass. However, the excitation energy is also transferred through the weak absorption in the range of the absorption edge tail caused by the states of defects and impurities (Fig. 21). At the same time it was found that the same effect is observed for high-purity samples. Hence it is possible to conclude that the states in the forbidden band through which the energy is transferred are the internal defects of a glass. For the excitation of REI through the absorption of stimulating energy by a matrix the following mechanism was offered (Turnbull et al., 1999; Bishop et al., 2000). The light with energy close to the optical gap is absorbed by the matrix, thus the electron from the valence band is excited into the conductivity band. The created hole is captured by the nearest defect (on the contrary, capture of electron is also possible). Then the excited defect relaxes, giving a part of energy to a lattice thus occupying the deeper level in the forbidden band. Then the photo-excited electron recombines with the hole. The energy formed at this process is transferred to the nearby REI, thus translating REI to the excited state. After the recombination and energy transfer, the energy level of the defect occupies the initial energy position at the edge of the forbidden band. The fact of the energy exchange between Er 3+ ion and glassy matrix in the GeSzGazS3 : ErzS3 system was experimentally confirmed in the work by Ivanova, Man'shina, Kurochkin and Tver'yanovich (2002a) while investigating up-conversion luminescence. Such energy exchange can lead to the increase of conductivity of the sulfide glass containing Er 3+ ions on irradiating them with wavelength being in the range of transparency of the glassy matrix but coinciding with the energy of Er levels.
192 5.
Yu. S. Tver' yanovich and A. Tverjanovich Anti-Stokes Luminescence
Ability of a substance to radiate light with frequency greater than the frequency of stimulating light has been discovered in the 1960s (Auzel, 1966; Ovsynkin and Feofilov, 1966), and has been named anti-Stokes luminescence in contradiction to the classical luminescence satisfying Stokes's rule. In the last decade, interest towards this phenomenon has sharply grown in connection with the transition from crystal matrixes doped with REI to glassy matrix. The phenomenon of the up-conversion is interesting not only because of the transformation of IR radiation into the visible spectrum range but also because the process lowers the efficiency of laser transitions. The occurrence of anti-Stokes luminescence is caused by the transitions inside nonfilled 4f shells of REI which, strictly speaking, are forbidden, as the rules of selection of the main and orbital quantum numbers are broken for them. REI have complex structure of energy levels that are insensitive to the introduction of REI in the crystal or glassy matrix. It occurs because of shielding of electrons of non-filled 4f shells by filled 5 s and 5p shells. The effective up-conversion processes are observed for the Er 3+ and Pr 3+ ions. The systems containing Er 3+ ions are the most intensively studied. The diagram of energy levels of Er 3+ ion and the scheme of excitation of up-conversion luminescence (the excitation by light with wavelengths 1550 and 813 nm) are shown in Figure 22. As was marked in Section 2, the glasses of the GeS2-Ga2S3 system possess a number of advantages as a matrix for introduction of the up-convertor ions. So in glasses
2.5
-
4F7/2
2.0
~H 11/2 53/2
1.5 7
> >
= 4F9/2
!>r
419/2 4111/2
-9 1.0 >
4113/2
0.5
1550 nm
813 nm
:'. O
O t"q
r---
O tt')
O tt%
0.0
4115/2
FIG. 22. Energylevel diagram of Er3+.
Rare-Earth Doped Chalcogenide Glass
193
of the given system it is possible to introduce a significant amount of Er 3+ ions, up to 4.7 at%, into glass of composition 0.85GeS2-0.15Ga2S3. The limiting concentration of the REI in glass is even higher on the increase of Ga2S3 contents (Tverjanovich, Grigoriev, Degtyarev, Kurochkin, Man'shina and Tver'yanovich, 2001). Besides that, these glasses are transparent in most of the visible range of the spectrum. The absorption spectrum of the glass of composition (0.85GeS2-0.15Ga2S3)-Er2S3 is represented in Figure 23a. Transitions with energy smaller than the transition from the level 4H 11/2 to the basic level 4115/2 are in the field of transparency. Anti-Stokes spectra of luminescence at excitation by radiation with wavelengths of 1550 and 813 nm are shown in Figure 23b and c (Tverjanovich et al., 2001). The irradiation of glass by light with wavelength of 813 nm causes visually detected green fluorescence. The intensity of a luminescence naturally increases on increase of the REI concentration but from a certain threshold, the process of concentration quenching caused by non-radiated energy transfer from excited atom by means of a dipole-dipole interaction begins. For glasses of x(0.85GeS20.15Ga2S3)-(1 - x)Er2S3 system the maximum intensity is observed at concentration of Er equal to 1.2 at.% (Fig. 24). The maximum is shifted aside increasing of Er concentration on relative increasing of Ga2S3 contents (Fig. 24, curves 1- 3, respectively)
20~ 18
4H11/2-4I15/2
~'~12
_
~
I I11/2-
'-'3/2- 15/2 9 4 ~ ( C4~ 41
..... 520 530 540 550 560 570 ~, nm
3
,
,
.
,
.
I15/2
,
| /
~ 419/2-4115/2/
5
I
I
,
500
/
4F9/2-4115/2 1~ ~
~ 4532/-41152/
lO
4
I
600
,
I
700
,
I
800
~
I
900
I ih
~
(b)
I
,
I
1000 1100
(a)
1 0
500
1000
,
~ 1500
, '
I 2000
~, am FI~. 23. (a) Absorption spectraof 0.85GeS2-0.15Ga2S3 glass doped with 1.8 at.% Er. (b and c) Luminescence spectra of Er3+ excited with 1550 and 813 nm, respectively.
194
Yu. S. T v e r ' y a n o v i c h a n d A. T v e r j a n o v i c h
7 -
o ~
9 0.15 Ga2S 3
~/
o 0.30 Ga2S 3 0.40 GazS 3
4
bar
O •
'
I
'
I
'
I
'
I
'
I
'
I
'
I
'
I
'
I
'
I
'
"1
0.0 0.5 l.O 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0:5.5 at.% Er Fro. 24. Dependence of luminescence intensity on erbium concentration for glasses of the (1 - x)GeS2(x)Ga2S3 system at x -- 0.15; 0.30; 0.40.
but does not exceed 2.2 at.%. For matrixes on the basis of silicate glasses the optimum concentration of Er 3+ ion does not exceed 1 at.% (Gu, Ong, Cai, Yang, Du and Yang, 1999) while for crystal system ( Y 2 0 2 S : Er) it is much above--about 6 at.% (Kurochkin, Manashirov, Sattarov, Smirnov and Tsyurupa, 1992). The dipole-dipole interaction is negligible at inter-ion distance exceeding 0.5 nm (Dgabotinsky et al., 1980). Indeed the calculation of distance between Er 3+ ions in polycrystalline Y202 S doped with 6 at.% Er 3+ in the assumption of uniform distribution of ions in a crystal lattice results in a value of 0.6 nm. The calculated value for the glass of composition 0.85GeSz-0.15GazS3 containing 1.2 at.% Er is 1.7 nm that exceeds the boundary condition by three times. For composition 0.60GeSz-0.40GazS3 containing 2.2 at.% Er (Fig. 24, curve 3) the distances between Er ions calculated in the assumption of uniform distribution is 1.3 nm. The presence of dipole-dipole interaction in glasses at distances formally exceeding the radius of the effective exchange can be explained by non-uniform distribution of the REI in the matrix of a glass. In this connection it is necessary to note two features. The first one is that the increase of GazS3 contents in glass results in the reduction of the tendency to cluster formation of Er atoms. The second is the optimization of chemical composition of the glassy host and conditions of synthesis of a glass with the purpose of increasing the degree of uniformity of REI distribution in the glassy matrix, which allows to increase the efficiency of the luminescence essentially.
Rare-Earth Doped Chalcogenide Glass
195
The non-uniform distribution of the REI in the glass matrix is caused by two reasons. The first one is statistical: fluctuation deviations from the uniform distribution. The second one is the tendency of association of structural units containing REI in the glass. We should try to divide these two factors when taking into account published research (Tver' yanovich and Murin, 1999), and we make the following simplifying assumptions. The luminescence of REI is completely quenched if there is another ion at a distance shorter than 0.6 nm. Luminescence is not reduced if the distance to another ion is longer than 0.6 nm. The specified bounds distance corresponds to two inter-atomic distances. Therefore, we consider only those REI radiate, the second coordination sphere of which is free from REI (we designate the fraction of such REI as go according to Tver'yanovich and Murin, 1999). The coordination number of REI in chalcogenide glasses usually is equal to 6. Therefore let us consider that in the second coordination sphere of REI there are six positions for atoms of metals. From the general reasons the intensity of a luminescence (I) is proportional to the concentration of Er atoms not participating in the dipole-dipole interaction. This concentration of them is equal to concentration of Er atoms (CEr) multiplied by the fraction of them which does not participate in the dipole-dipole interaction: I--~ CErg0
(8)
When statistical filling of the second coordination sphere of REI proceeding from the combination theory is assumed it is possible to write the following expression for go go -- (1
-
b) 6
(9)
where b is the fraction of Er atoms from the total number of atoms of metals (it is supposed that the second coordination sphere can be occupied only by atoms of Er, Ge or Ga). The sixth degree is caused by six possible places for atoms of metals in the second coordination sphere. On the other hand, the probability of occupation of the second coordination sphere by an erbium atom in relation to probability of occupation of it by any other atom of metal (P) (in the absence of correlation between type of the central ion and composition of its second coordination sphere) is equal to the relation b/(1 - b). In the case of the deviation from the statistical distribution, it is possible to enter the additional parameter S in the manner P = Sb/(1 - b ) . S is the so-called segregation factor. The segregation factor is equal to 1 if Er atoms are distributed statistically on the metal positions in the network of glass. It is more than 1 if there is a tendency for aggregation of the structural units containing rare-earth atoms. It is less than 1 if there is a deviation from the statisticalnuniform distribution aside the regular--uniform distribution. Let us suppose that b # is the fraction of Er atoms from the total number of atoms of metals with the account segregation factor: P = Sb/(1 - b ) - - b # / ( 1 - b # ) . From this it follows that b # -- Sb/(1 - b + Sb) and substituting b # instead of b in the expression for go we receive: go --
1 - b(1 - S)
(10)
The calculated dependencies of intensity of the luminescence on the concentration of Er for various values of segregation factors are shown in Figure 25. The maximum intensity located at 6 at.% (as in the case of Y2028: Er) corresponds to the tendency of statistical distribution. The maximum intensity at 2 at.% (chalcogenide glasses)
196
Yu. S. Tver' yanovich and A. Tverjanovich
0.025 0.020
~
8
.,..~
= 0.015 x5 ~= 0.010 0.005 0.000
,
I
,
5
I
,
10
I
15
at.% Er FIG. 25. The calculated dependences of luminescence intensity on erbium contents for various values of segregation factors.
corresponds to the tendency for association of the structural units containing Er (S = 4). It is interesting that the same value of the segregation factor has been found for the distribution of ions of iron in glasses of the S b - G e - S e system (Bychkov, Borisova and Vlasov, 1984). The maximal concentrations of Er and Fe in corresponding glassy matrixes are also close to each other (4.7 and 2.2 at.%, respectively). From the depicted curves it also follows that on realization of statistically uniform distribution of Er in the volume of glass, it is possible to expect the increase of intensity of luminescence,, by four times. The functional dependence of luminescence efficiency of REI on their concentration can be considered in a different way. It is possible to write down the following expression for probability of relaxation of the REI from the excited level (W) W = I/'r=
W R --~ W NR
(11)
where ~-is the lifetime of the excited level and W R and W NR the probabilities of radiating and non-radiating transition, respectively. W R does not depend on the concentration of REI. In turn, W NR can be divided into the probability of multiphonon relaxations which do not depend on the concentration of active ions W(0), NR and the probability of crossrelaxation processes which is square-law function of concentration of REI in the cases of their uniform distribution in the matrix of glass w(N~. Thus expression (11) can be copied in the following l/z-
W R -[- wiN~ + w ( N ~ - - l/z(0 ) + 1/,ric )
(12)
where Z(o) is the lifetime of the excited level at the extrapolation of REI concentration to 0. The following semi-empirical expression is offered for probability of cross-relaxation processes (Higuchi, Kadono and Kitamura, 1998a) w(N~ -" I/T(C ) -- (I/T(o))(C/Cq) 2
(13)
where Cq is the concentration at which, owing to concentration quenching, the probabilities of relaxation caused by the processes depending on the concentration
Rare-Earth Doped Chalcogenide Glass
197
and those not depending on the concentration become equal, i.e., W R -1- W ( ~ = W(c)NR. Thus the achievement of statistical distribution of the REI in glass which should increase the optimal concentration of REI from 2.2 at.% (Fig. 25) up to 6 at.% will allow reducing the probability of the cross-relaxation channel of the non-radiative relaxation by seven times. Uniting the expressions (12) and (13) we obtain the expression connecting the lifetime of the excited level with the concentration of the REI: ~--- ~-(0)/(1 + (C/Cq) 2)
(14)
The experimental values of the lifetime of the l e v e l 483/2 in glass of composition x(0.85GeSz-0.15Ga2S3)-(1 - x)ErzS3 with the various contents of Er ions and approximation of this dependence by Eq. (14) (Ivanova, Man' shina, Kurochkin, Tver'yanovich and Smirnov, 2002b) are represented in Figure 26. The value ~'(o) obtained by the approximation is 8/xs, i.e., W R -4- wiN~ - - 125 X 103 C - 1 . Calculated from the absorption spectrum on the basis of the Judd-Ofelt theory the probability of radiating relaxation is equal to 3.6 x 103 s -1 therefore W(~ - 121 x 103 c -1 and quantum efficiency (/3) determined as/3 = w R / ( w g + W(No~)reaches only 3%. On the other hand the probability of the multiphonon relaxations can be estimated on the basis of the following relation (Layne, Lowdermilk and Weber, 1977)
W(No~-- k(n(T)+ 1)p e x p ( - a A E )
(15)
where AE is the energy gap between the given level and the nearest one with smaller energy; n(T) the Bose-Einstein thermal population factor; p the ratio of phonon energy to AE; k and a the constants dependent on the matrix. The value of w(N~ calculated with Eq. (15) is equal to 400 s -1 (the following issue data were used (Heo and Shin, 1996): AE = 3070 cm -1 ( 4 8 3 / 2 - 4 F 9 / 2 ) , k = 106 C-~ and a = 2.9 X 10 -3 cm-1). The calculated value is 300 times less than the experimental datum. It indicates that there is an additional channel of a non-radiative relaxation.
=6 (D .,..,
~5
|
0
I
1
!
I
2
|
I
3
|
I
4
|
I
!
5
p, 102~ -3 FIG. 26. Lifetimeof the level 4S3/2for various erbium concentrations. Line is approximationby Eq. (14).
198
Yu. S. Tver'yanovich and A. Tverjanovic,h
The dependence of intensity of the up-conversion luminescence from t h e 483/2 level (it was excited by laser radiation with wavelength 813 nm) on the location of absorption edge (it was determined at the level ce = 10 cm -1) is shown in Figure 27a. It suggests the correlation of these two characteristics. It is caused by the overlapping of the above-stated 483/2 level by the tail of conductivity band of the glassy matrix (Fig. 23a). The lifetime of the 4S3/2 level decreases on reduction of the energy gap between the fundamental absorption edge and the given level (Fig. 27b), which is due to transfer of excitation energy to the electronic subsystem of the matrix (Ivanova et al., 2002b). This channel of the non-radiative relaxation of excitation of REI in the band of a matrix was noted by Higuchi et al. (1998b), and it is apparently responsible for the additional non-radiative losses of excitation energy. According to the scheme of excitation of up-conversion luminescences (Fig. 22, excitation wavelength 813 nm), the 4F7/2 level is involved in the mechanism of the luminescence. The energy of this level is close to
(a) 6x10 6-
5x10 6-
4x10 6-
3x1062x 106 lxl0 6
I
'
I
470
'
I
480
'
I
490
'
I
500 nm
'
I
510
'
520
I
530
(b)
6-
:I.
4
I
470
'
I
480
'
i
490
'
I
500
'
i
510
'
I
I
520
530
nm
FIG. 27. (a) Dependence intensity of up-conversion luminescence on the position of absorption edge. (b) Dependence lifetime of 4S3n level on the position of absorption edge.
Rare-Earth Doped Chalcogenide Glass
199
580 560 O
540 = 520 M 500 i 480 460
FIG. 28.
at.% Er Dependenceof fundamental absorptionedge on erbium concentration.
the optical width of the forbidden band. Hence the probability of the non-radiative relaxation from this level should be even higher than from the 483/2 level. At the fixed concentration of REI (the increase of the ErzS3 contents in glass results in the shift of absorption edge in long-wavelength range (Fig. 28)) the energy position of absorption edge for glasses of the GeSz-GazS3 system is determined by the concentration of the homo-bonds M - M (M = Ge, Ga). The increase of concentration of such bonds shifts the absorption edge in long-wavelength range (for GeS Eg -- 1.6 eV, while for GeS2 Eg -- 3.5 eV). The absorption spectra of various glasses of the GazS3GeSz:Er2S3 system are depicted in Figure 29. Each of these compositions was prepared in two different regimes in order to receive relatively lighter and darker samples. The Raman spectra of the same samples (intensity is normalized) are shown in Figure 30. The intensity of the peak located near 270 cm-1 in comparison with the intensity of the basic peak at 340 cm-~ is higher for dark glasses. The band 340 cm-~ is caused by full symmetric vibrations of the tetrahedrons M84/2 ( M - - Ge, Ga) while the vibrations with frequencies of about 270 cm -1 are characteristic for bond M - M (Tverjanovich et al., 1996). Hence the direct dependence between the concentration of the M - M bonds and position of absorption edge is traced. The occurrence of such bonds in the glass of stoichiometric composition GeS2 is due to both the fluctuation of composition and the thermal instability of GeS2 at high temperatures (GeS2 dissociates to GeS and S). The decrease of temperature of synthesis results in the shift of absorption edge aside large energy (Fig. 31; Tverjanovich et al., 2001) and therefore the increase of the anti-Stokes luminescence intensity (Fig. 32; Tverjanovich et al., 2001). The annealing of the glasses leads to additional increase of luminescence intensity (Fig. 32). The concentration of M - M (M -- Ge, Ga) bonds depends also on the concentration of impurities. So in glassy GeS2 the OH- groups borrow places of sulfur in tetrahedron GeS4/2 (Simons et al., 1995b; Marchese and Jha, 1997) and thus reduce the probability of formation of bonds Ge-Ge. The typical IR absorption spectrum of the given glasses is shown in Figure 33 in which the bands corresponding to OH-, SH- and - C H 2- impurities are marked.
200
Tverjanovich
Yu. S. Tver'yanovich and A. 5040
dark ........ light
30 20
40 mol.% Ga2S 3
10
'
400
7~I 6o
I
'
500
I
'
I
600
'
700
I
800
'
I
900
!
50 ........ light
'fi 40 30
dark
~
20
20 mol.% GazS 3
10 0 400
I
500
~
I
~
600
I
700
~,
~
I
I
800
900
n m
FIG. 29. Absorption spectra of glasses prepared in different regimes.
The concentration of the given impurities in the investigated samples is given in Table 1. The extinction coefficients are taken from the works of Kale, Liu and Jha (1998), Liu, Kale, Tikhomirov and Jha (1999) and Kobelke, Kirchhof, Schuster and Schwuchow (2001), respectively. All samples contain identical amount of SH- groups while light samples contain larger quantity of OH- and - C H 2 - impurities in comparison to the dark samples. The increase of the temperature of synthesis results in the following reaction in the melt (Kobelke et al., 2001)
-CH2-
+ OH- + S 2 r
C O 2 -~- H 2 S + C O S + S O 2 -~- C S 2
and hence leads to relative reduction of concentration of the OH- and - C H 2 - impurities. At the same time the increase of concentration of the OH- groups in glass should result in the reduction of efficiency of the luminescence due to the increase of probability of the non-radiative relaxation. So the energy of transition of the erbium ion from the 483/2 level to the 4F9/2 level just coincides with the vibration energy of the OH- group. Moreover the erbium ion can relax from the 4113/2 level to the ground 4115/2 level if OHgroups take two quanta of the vibration energy. Besides the increase of optical width of the forbidden band in glasses of the G e - G a - S system can be achieved by partial replacement of sulfur by bromine as specified in Section 2. One of the methods to increase the efficiency of up-conversion luminescence consists of the introduction of sensibilizators (for example Yb 3+ ions) in glass. The scheme
Rare-Earth
Doped
Chalcogenide
9
I
'
.,.,
........ light dark
," "),
I
100
'
I
200
'
I
300
'
I
400
.
'
500
-....... light
:'",
.,..,
201
Glass
.
~D
I
'
I
1O0
'
I
200
'
'
I
400
""
'
500
........ light .~ ~ dark
",,, 9,.
I
I
300
9
'
I
100
?
'
I
'
I
'
I
200 300 400 Raman shift, cm -1
'
500
Fro. 30. Raman spectra of glasses prepared in different regimes 9
30.......... 850 ~ ..... 20
d)
900 ~
950~ ........ 1000 ~
c) ............ ~.\
1050~
,,,b)
"-., a) 10-
".
.
\\,\
",, .
.
.
.
.
460
,
I
~, \,
.
i
,,
"'"'"'........ . . . . . . . . . . . . ..."*~
0
\
x ~',, ...... ....
,
I
480
500
"'..... . . . . . . . . . . . . ..... _
./l'l! ,qr .,,,,
--__
,
I
520
X, nm FIG. 31. Absorption spectra of glasses synthesized at various temperatures 9
,
202
Yu. S. Tver'yanovich and A. Tverjanovich
102
ta quenched ~ anneale._______~d
r~
4 ~D
8bo
'
8~0
'
9bo
'
9~0
'
1~)00'
1~)50
T, ~
FIG. 32. Dependenceof luminescenceintensity on synthesis temperaturefor quenched and annealed glasses.
of energy levels of the Yb 3+ and Er 3+ ions is represented in Figure 34. In this case the mechanism of population of the erbium 453/2 and 4F9/2 levels is the following. The excitation radiation with wavelength about 1/zm is absorbed by Yb 3+ ion, which excites at the 4F5/2 level. Further, as a result of interaction of excited Yb 3+ ion and ion of Er 3+ in the ground state 4115/2 Er 3+ ion transits at the 4Ill/2 level and Yb 3+ ion comes back to the ground level where it again absorbs a quantum of energy. At the second stage the excited Yb 3+ ion transfers the energy to the already excited Er 3+ ion transferring this Er 3+ ion to the 2H1~/2 level from which due to relaxation it transits at the 4S3/2 level from which it radiates in the green range of the spectrum. The population of the 4F9/2 level with subsequent radiation with wavelength 670 nm can occur in two ways. First is due to
SH-
"7
E
6
4
I
2.5
,
I
3.0
J
I
,
3.5
I
4.0
,
I
4.5
,
I
5.0
~,, ~tm FIG. 33. IR absorption spectra for glasses of the (1 - x)GeSe-(x)Ga2S3 system.
203
Rare-Earth Doped Chalcogenide Glass TABLE I CONCENTRATION OF O H - , S H - , - C H 2 - IMPURITIES IN THE GLASSES OF THE (xGazS3-(1 - x)GeS2) SYSTEM CONTAINING 1.8 AT.% ER Composition (x)
Index
Coil--
Csu--
C-cuT
ppm wt
at.%
ppm wt
at.%
ppm wt
at.%
0.2
'light' 'dark'
143 39
0.04 0.01
299 288
0.04 0.04
180 32
0.06 0.01
0.3
'light' 'dark'
323 237
0.09 0.07
261 272
0.04 0.04
313 58
0.11 0.02
non-radiative relaxation from the 4S3/2 level. But the following mechanism seems to be more effective. The first stage of energy transfer is similar to the one co/asjdered above. Then the Er 3+ ion relaxes non-radiatively from the 4Ill/2 level to the 4113/2 level at which it again receives energy from Yb 3+ ion and transits at the 4F9/2 level. According to the data of Oliveira, de Araujo, Gouveia-Neto, Medeiros Neto, Sombra and Messaddeq (1998), the efficiency of luminescence in the samples containing Er 3+ and Yb 3+ ions is 25 times higher than in glasses doped only with Er 3+ ions. However, it is necessary to note that the introduction of ytterbium in glasses of the G a - G e - S system results in reduction
2.0 -
2H11/2 Er 3+ l
1.5
~LI
4F9/2
.I 1
Yb 3+ "7 1.0
483/2
419/2
4Ill/2
4113/2 990 nm
0.5
o c~
0.0
-
T
2F7/2
t
FIc. 34. Energy level diagrams of Yb 3+ and Er 3+.
4115/2
204
Yu. S. Tver'yanovich and A. Tverjanovich
of optical width of the forbidden band which in turn should result in the increase of the non-radiative losses.
6. Other Optical Properties of the Glassy Semiconductors Activated by Lanthanoids A number of works remark the existence of non-linear optical properties of the chalcogenide glasses that are specifically connected with the presence of lanthanoids in their composition. For example, the photo-darkening of the chalcooxide glasses containing lanthanum and praseodymium ((Lal-xPrx)zS32Ga203) under light illumination in the wavelength range from 340 to 500 nm is described in the work by Zobov, Nasrulaeva and Sokolov (1996). Photo-darkening is indicated by displacement of fundamental absorption edge in the direction of long-wavelength part of the spectrum and in reduction of the transmission in all ranges of the transparency of glass (Fig. 35). The specified effect is absent completely if glass does not contain Pr (x -- 0) and it regularly grows on increase of x from 0.1 up to 1. However, it is necessary to take into account that the absorption within the spectral range of light processing is appreciably defined by the absorption bands of Pr 3+. The observed effect is reversible at room temperature. The activation energy of the photo-induced changes was estimated from kinetic investigation of low temperature annealing at different temperatures. It has a value of about 1 eV. The assumption of connection of the given effect with the change of the degree of oxidation of praseodymium from 3 + to 4 + (Zobov et al., 1996) c~mses serious doubts. The photo-induced optical anisotropy in chalcogenide glasses was investigated in the laboratory headed by Prof. V.M. Lyubin (FTI, St Petersburg) more than 10 years ago (Lyubin and Tikhomirov, 1989). Similar researches of chalcogenide glasses doped with Pr have been carried out by one of the employees (V.K. Tihomirov) of this laboratory
30 25
......... ~ni2-di2~2~ ted
20 15
.
.
.
.
.
.
10
,
0.36
I
0.38
,
I
0.40
,
I
0.42
~
l
,
I
0.44 0.46 )~, ~tm
,
I
0.48
,
I
0.50
,
I
0.52
FIG. 35. Transmissionspectra of (Lal-xPrx)2Sy2Ga203 glass before and after irradiation.
205
Rare-Earth Doped Chalcogenide Glass
1.61.4"
o/O
1.2
9 --'-'----A
_........~~
o/
-
~ ~
o / O . ~ ~ ' A ~ A ~ A
1.0
0.8 o , A~
~ 0.6
8
/)m~"
~/
0.4
/m~m-.---.-.---m ~
--'--
:..-.
0.2
2000 ppm
i
m
0.0 -0.2
i
0
'
~
5
'
I
10
,
ll5
'
i
20
t, min FIG. 36. Kinetics of photoinduced dichroism for Ge34.5S55.5Iloglasses with various concentrations of Pr3+.
(Tikhomirov, Krasteva, Hertogen, Adriaenssens and Sigel, 1997; Tikhomirov, 1999). It has been discovered that the value of induced optical dichroism of glasses of the G e - S - I system increases 10 times on introduction, in their composition, of 1000 ppm Pr (Tikhomirov et al., 1997). Further increase of concentration of Pr leads to saturation of the observable effect (Fig. 36; Tikhomirov, 1999). Photo-anisotropy in chalcogenide glasses is due to the polarization of C~-C~- centers (where C is an atom of chalcogen, the bottom index is coordination number, the top index is a charge; Tikhomirov and Elliott, 1995). Influence of an REI is attributed to increase of polarizability of C~-C~- centers because of electrostatic interaction of C~- and Pr 3+ (Tikhomirov, 1999). It is necessary to note that the estimated concentration of such centers in chalcogenide glasses is about some thousand ppm, which coincides with the concentration of REI necessary for saturation of the effect. It is also interesting that the specified concentration of REI approximately corresponds to the concentration at which the maximum intensity of luminescence is observed. The composition of the glass matrix has the same strong influence on the value of photo-induced anisotropy and on characteristic time of its output on saturation that is in the frameworks of the offered approach.
R e f e r e n c e s
Abe, K., Takebe, H. and Morinaga, K. (1997) J. Non-Cryst. Solids, 212, 143-150. Aitken, B.G. and Quimby, R.S. (1997) J. Non-Cryst. Solids, 213/214, 281-287. Auzel, F. (1966) C. R. Acad. Sci., 262, 819-821. Belykh, A., Glebov, L., Lerminiaux, C., Lunter, S., Mikhailov, M., Plyukhin, A., Prassas, M. and Przhevuskii, A. (1997) J. Non-Cryst. Solids, 213/214, 238-244. Bishop, S.G., Turnbull, D.A. and Aitken, B.G. (2000) J. Non-Cryst. Solids, 266-269, 876-883. Bojesson, L., Torell, L.M., Dahlborg, U. and Howels, W.S. (1989) Phys. Rev. B, 39, 3404. Brady, D.J., Schweizer, T., Wang, J. and Hewak, D.W. (1998) J. Non-Cryst. Solids, 242, 92-98. Bychkov, E.A., Borisova, Z.U. and Vlasov, Yu.G. (1984) Fiz. Khim. Stekla, 10, 390.
206
Yu. S. Tver'yanovich and A. Tverjanovich
Carter, S.F., Wyatt, R., Szebesta, D. and Davey, S.T. (1991) Proceeding~ oflOOC-ECOC91, Parts 1-4, Vol. 4, p. 21, Chapter 241. Cervelle, B.D., Jaulmes, S., Laruelle, P. and Loireau-Lozac'h, A.M. (1980) Mater. Res. Bull., 15(1), 159-164. Cho, W., Kim, M., Jo, J., Hahn, T. and Choi, S. (1989) Proceedings of the Sixh International Symposium on Halide Glasses, Clausthal-Zellerfeld, Germany, October, p. 325. Dgabotinsky, M.E., Alekseev, N.E., Gapontsev, V.P., Zhabotinsky, M.E. (1980) Laser Phosphate Glasses, (Ed., Zhabotinsky, M.E.) Nauka, Moscow, p. 352. Elliott, S.R. (1991) Phys. Rev. Lett., 67(6), 711-714. Evans, R.C. (1964)An Introduction to Crystal Chemistry, Cambridge University, Cambridge. Fernandez, J., Balda, R., Mendioroz, A. and Garcia-Adeva, A.J. (?001) J. Phys. Condens. Matter, 13, 10347-10358. Griscom, L.S., Adam, J.-L. and Binnemans, K. (1999) J. Non-Cryst. Solids, 256/257, 383-389. Gu, G., Ong, P.P., Cai, J., Yang, W., Du, Y. and Yang, S. (1999) J. Thin Solid Films, 340, 230. Gu, S.Q., Ramachandran, S., Reuter, E.E., Turnbull, D.A. and Verdeyen, J.T. (1995) Appl. Phys. Lett., 66(6), 670-672. Hector, J.R., Wang, J., Brady, D., Kluth, M., Hewak, D.W., Brocklesby, W.S. and Payne, D.N. (1998) J. NonCryst. Solids, 239, 176-180. Heo, J. and Shin, Y.B. (1996) J. Non-Cryst. Solids, 196, 162. Hewak, D.W., Medeiros Neto, J.A., Samson, B.N., Brown, R.S., Jedrzejewski, K.P., Wang, J., Taylor, E., Laming, R.I., Wylangowski, G. and Payne, D.N. (1994) IEEE Photon Technol. Lett., 6(5), 609. Higuchi, H., Kadono, K. and Kitamura, N. (1998a) J. Appl. Phys., 83, 19. Higuchi, H., Takahashi, M., Kawamoto, Y., Kadono, K., Ohtsuki, T., Peyghambarian, N. and Kitamura, N. (1998b) J. appl. Phys., 83(1), 19-27. Ishikawa, E., Tawarayama, H., Ito, K., Aoki, H., Yanagita, H. and Toratani, H. (1996) Technical Report IECIE, OPE96-111. Ivanova, T.Yu., Man' shina, A.A., Kurochkin, A.V. and Tver'yanovich, Yu.S. (2002a) J. Non-Cryst. Solids, 298, 7-14. Ivanova, T.Yu., Man'shina, A.A., Kurochkin, A.V., Tver'yanovich, Yu.S. and Smirnov, V.B. (2002b) J. NonCryst. Solids, 298, 7-14. Kadono, K., Higuchi, H., Takahashi, M., Kawamoto, Y. and Tanaka, H (1995) 184, 309-313. Kale, B.B., Liu, X. and Jha, A. (1998) Characterisation of OH- impurities in GeSz-based glasses. In: Proceedings of XVIII International Congress on Glass, CA, USA. Kobelke, J., Kirchhof, J., Schuster, K. and Schwuchow, A. (2001) J. Non-Cryst. Solids, 284, 123-127. Krupke, W.F. (1966) Phys. Rev., 145, 325. Kumta, P.N. and Risbud, S.H. (1990) Ceram. Bull., 69, 1977. Kumpta, P.N. and Risbud, S.H. (1994) J. Mater. Sci., 29, 1135-1158. Kurochkin, A.V., Manashirov, O.Ya., Sattarov, D.K., Smirnov, V.B. and Tsyurupa, O.V. (1992) Svetotekhnika, (5), 4. Layne, C.B., Lowdermilk, W.H. and Weber, M.J. (1977) Phys. Rev. B, 116, 10. Lima, S.M., Sampaio, J.A., Catunda, T., de Camargo, A.S.S., Nunes, L.A.O., Baesso, M.L. and Hewak, D.W. (2001) J. Non-Cryst. Solids, 284, 274-281. Liu, X., Kale, B.B., Tikhomirov, V.K. and Jha, A. (1999) J. Non-Cryst. Solids, 2561257, 294-298. Loireau-Lozac'h, A.M. and Guittard, M. (1975) Ann. Chim. (France), 10(2), 101-104. Loireau-Lozac'h, A.M., Keller-Besrest, F. and Benazeth, S. (1996) J. Solid State Chem., 123, 60-67. Lyubin, V.M. and Tikhomirov, V.K. (1989) J. Non-Cryst. Solids, 114, 133. Man' shina, A.A., Kurochkin, A.V., Degtyarev, S.V., Grigor'ev, Ya.G., Tw~,rjanovich, A.S., Tver'yanovich, Yu.S. and Smirnov, V.B. (2001) Proc. SPIE, 4429, 80-88. Marchese, D. and Jha, A. (1997) J. Non-Cryst. Solids, 213/214, 381. Marchese, D., Kakarantzas, G. and Jha, A. (1996) J. Non-Cryst. Solids, 196, 314- 319. Marmolejo, E.M., Granado, E., Alves, O.L., Cesar, C.L. and Barbosa, I_,.C. (1999) J. Non-Cryst. Solids, 247, 189-195. Martin, A.J. and Brenig, W. (1974) Phys. Stat. Sol., B64, 163. Mazzacurati, V., Nardone, M. and Signorelli, G. (1979) Mol. Phys., 38, 1379. Messaddeq, S.H., Siu Li, M., Lezal, D., Ribeiro, S.J.L. and Messaddeq, Y. (2001) J. Non-Cryst. Solids, 284, 282-287.
Rare-Earth Doped Chalcogenide Glass
207
Mori, T. and Arai, T. (1983) J. Non-Cryst. Solids, 59/60, 867. Nemec, P., Frumarova, B. and Frumar, M. (2000) J. Non-Cryst. Solids, 270, 137-146. Novikov, V.N. and Sokolov, A.P. (1991)Solid State Commun., 77, 243. Ohishi, Y., Kanamori, T., Kitagawa, T., Takahashi, S., Snitzer, E. and Sigel, G.H. Jr. (1991) Opt. Lett., 16, 1747. Oliveira, A.S., de Araujo, M.T., Gouveia-Neto, A.S., Medeiros Neto, J.A., Sombra, A.S.B. and Messaddeq, Y. (1998) Appl. Phys. Lett., 72(7), 753-755. Ovsynkin, V.V. and Feofilov, P.P. (1966) Pis'ma VZh. Eh. T. Ph., 4(11), 471. Park, S.H., Heo, J. and Kim, H.S. (1999) J. Non-Cryst. Solids, 259, 31-38. Pauling, L. (1929) J. Am. Chem. Soc., 51, 1010. Phillips, J.C. (1979) J. Non-Cryst. Solids, 34(2), 153-181. Rapp, C.P. (1986), Handbook of Laser Science and Technology Vol. 5 (Ed., Weber, M.J.) CRC, Boca Raton, FL, pp. 339. Saffarini, G. (1994) Solid State Commun., 91(7), 577-580. Schimmel, R.C., Faber, A.J., de Waardt, H., Beerkens, R.G.C. and Khoe, G.D. (2001) J. Non-Cryst. Solids., 284, 188-192. Shaw, L.B., Cole, B.J., Sanghera, J.S., Aggarwal, I.D. and Schaafsma, D.T. (1998) 0FC'98 Tech. Digest WG8, p. 141. Shim, Y.B., Cho, W.Y. and Heo, J. (1996) J. Non-Cryst. Solids, 208, 29-35. Shin, Y.B. and Heo, J. (1999) J. Non-Cryst. Solids, 256/257, 260-265. Shin, Y.B., Yang, C.K. and Heo, J. (2002) J. Non-Cryst. Solids, 298, 153-159. Simons, D.R., Faber, A.J. and De Waal, H. (1995a) Opt. Lett., 20, 468. Simons, D.R., Faber, A.J. and de Waal, H. (1995b) J. Non-Cryst. Solids, 185, 283-288. Sokolov, A.P., Kisliuk, A., Soltwisch, M. and Quitmann, D. (1992) Phys. Rev. Lett., 69, 1540. Susman, S., Price, D.L., Volin, K.J., Dejus, R.J. and Montague, D.G. (1989) J. Non-Cryst. Solids, 106, 26. Tanabe, S., Hanada, T., Watanabe, M., Hayashi, T. and Soga, N. (1995) J. Am. Ceram. Soc., 78(11), 2917-2922. Tikhomirov, V.K. (1999)J. Non-Cryst. Solids, 256/257, 328-336. Tikhomirov, V.K. and Elliott, S.R. (1995) J. Phys. Condens. Matter, 7, 1737. Tikhomorov, V.K., Jha, A., Perakis, A., Sarantopoulou, E., Naftaly, M., Krasteva, V., Li, R. and Seddon, A.B. (1999) J. Non-Cryst. Solids, 256/257, 89-94. Tikhomirov, V., Krasteva, V., Hertogen, P., Adriaenssens, G. and Sigel, G. (1997) J. Non-Cryst. Solids, 222, 296-303. Topp, N.E. (1965) The Chemistry of the Rare-Earth Elements, Elsevier, Amsterdam. Turnbull, D.A., Aitken, B.G. and Bishop, S.G. (1999) J. Non-Cryst. Solids, 244, 260-266. Tverjanovich, A., Grigoriev, Ya.G., Degtyarev, S.V., Kurochkin, A.V., Man' shina, A.A. and Tver'yanovich, Yu.S. (2001) J. Non-Cryst. Solids, 286, 89-92. Tverjanovich, A., Tver'yanovich, Yu.S. and Loheider, S. (1996) J. Non-Cryst. Solids, 208(1/2), 49-55. Tver'yanovich, Yu.S., Degtyarev, S.V., Pivovarov, S.S. et al., (1998) J. Non-Cryst. Solids, 256/257, 95-99. Tver'yanovich, Y., Mamedov, S. and Degtyarev, S. (2000) Solid State Commun., 115, 631-633. Tver'yanovich, Yu.S. and Murin, I.V. (1999) J. Non-Cryst. Solids, 256/257, 100-104. Tver'yanovich, Yu.S., Nedoshovenko, E.G., Aleksandrov, V.V., Turkina, E.Yu., Tverjanovich, A.S. and Sokolov, I.A. (1996) Fiz. Khim. Stekla, 22(1), 13-19. van Dijk, J.M.F. and Schuurmans, M.F.H. (1983) J. Chem. Phys., 78(9), 5317. Viana, B., Palazzi, M. and LeFol, O. (1997) J. Non-Cryst. Solids, 215, 96-102. Wang, J., Brocklesby, W.S., Lincoln, J.R., Townsend, J.E. and Payne, D.N. (1993) J. Non-Cryst. Solids, 163, 261-267. Wei, K., Machewirth, D.P., Wenzel, J., Snitzer, E. and Sigel, G.H. Jr. (1994) Opt. Lett., 19(12), 904-906. Wei, K., Machewirth, D.P., Wenzel, J., Snitzer, E. and Sigel, G.H. Jr. (1995) J. Non-Cryst. Solids, 182, 257-261. Zobov, E.M., Nasrulaeva, A.Kh. and Sokolov, V.V. (1996) Opticheskiy Zh., 63(3), 62-65. Zou, X. and Izumitani, T. (1993) J. Non-Cryst. Solids, 162, 68-80.
This Page Intentionally Left Blank
CHAPTER
5
OPTICAL FIBERS FROM HIGH-PURITY ARSENIC CHALCOGENIDE GLASSES M. F. Churbanov INSTITUTE OF CHEMISTRY OF HIGH-PURITY SUBSTANCES, RUSSIAN ACADEMYOF SCIENCES, NIZHNY NOVGOROD, RUSSIA
V. G. Plotnichenko FIBER OPTICS RESEARCH CENTER AT THE GENERAL PHYSICS INSTITUTE, RUSSIAN ACADEMYOF SCIENCES, MOSCOW, RUSSIA
1. Introduction Chalcogenide glasses have been known as optical materials for about 50 years (Frerics, 1953). The first report on their applications in fiber optics was published in the 1960s (Kapany and Simms, 1965). The most intensive research on chalcogenide glasses as materials for fiber optics for the mid-IR range has been carried out during the past 15-20 years (Katsujama and Matsumura, 1988; Devyatykh et al., 1991a; Nishii and Yamashita, 1998), and research included the testing of a set of chalcogenide glasses made from III-V group elements for prospective optical fibers production. The most promising results were obtained from the glasses based on arsenic chalcogenides. Their properties include a wide transparency range, low optical losses within 2 - 1 2 / x m range, and stability to atmospheric moisture. The transparency range of glasses measured at the absorption level of 1 cm-1 is 0.6211.5/xm for As2S3, 0.8-17.5/zm for AszSe3, and 0.75-12.25/xm for AszSel.sS1.5. Minimum optical losses in AszS3 and AszSe 3 are evaluated as (6-7) x 10 -2 dB km -1 at 4 - 6 / z m range (Lines, 1984). For this reason the vitreous arsenic chalcogenides are attractive and applicable for optical fiber manufacturing. Moreover, a number of technical problems in optics and optoelectronics may be efficiently solved through the use of chalcogenide glass fibers with low optical losses. As a result, there is a permanent scientific and applied interest to chalcogenide glass fibers. Several research groups from different countries are now at work. The results of investigation and different aspects of chalcogenide glass fibers are considered in original papers and reviews. This chapter presents the current status of high-purity chalcogenide glasses and fibers based on arsenic chalcogenide glasses. 209
Copyright 9 2004 Elsevier Inc. All rights reserved. ISBN 0-12-752189-5 ISSN 0080-8784
210
M. F. Churbanov and V. G. Plotnichenko
2. Preparation of High-purity Chalcogenide Glasses 2.1. PREPARATION OF VITREOUS ARSENIC CHALCOGENIDES Bulk samples of vitreous arsenic chalcogenides of optical grade are produced by solidification of the glass-forming melt. The initial charge with definite composition is melted in evacuated sealed ampoules made from silica glass at the time-temperature modes, which exclude the crystallization, liquation and stria formation. Unique for different glass systems, these conditions are additionally determined by mass of the melted charge and by stability of glasses to crystallization. For example, the values of critical cooling rate for the vitreous AszS3 and AszSe3 differ approximately by three orders of magnitude and are equal to 2.4 x 10 -6 and 9 x 10 -3 K s -1, respectively. To prepare the initial charge for glass-forming chalcogenide synthesis, different variants (Churbanov, 1998) may be used, from chemical elements (the traditional method); decomposition of volatile inorganic hydrides; or the use of the arsenic monosulfide as an arsenic-containing component of the charge. The layers of arsenic selenide may be prepared by plasmochemical decomposition of arsine and selenium hydride. The prepared glass contains a noticeable quantity (several atomic percents) of hydrogen. The potential for hydrogen bubbles to blow up the melt during the heat treatment of glass still presents problems for the manufacturing process. Nevertheless, this method may be promising for preparing arsenic selenide thin films. The dependence of the content of arsenic and selenium of different valence states in the layers was first established for this process during experimental conditions.
2.2. TRANSPARENCY OF CHALCOGENIDE GLASSES AS AN IMPURITY-SENSITIVE PROPERTY
In general, the quality of solid material of a given macrocomposition depends on the content of impurities and structural defects. The property value P may be represented as the sum P = P0 + Pi -q- Pd nt- Pii q- Pdd + Pid
(1)
where P0 is the value for impurity and defect-free hypothetical substance, Pi and Pd are the increments due to impurities and defects, respectively, Pii and Pdd are the increments caused only by mutual interaction of impurities and defects, respectively, and Pid is the increment caused by interaction between impurities and defects. P0 is usually calculated on the basis of model conception. Transparency of optical medium is quantitatively characterized via attenuation of light flux passing through the sample and is determined by a well-known Buger-Lambert-Beer equation J -- J0 e-/31
(2)
where Jo and J are the light fluxes at the sample input and output, respectively, I is the optical path length, and 13 is the absorption (attenuation) coefficient with dimension of inverse length. The notion of 'attenuation coefficient' is equivalent to the notion of 'optical losses'. For chalcogenide glasses the coefficient fl may be represented in the form
Optical Fibers from High-purity Arsenic Chalcogenide Glasses
211
of a sum with each component corresponding to a definite mechanism of interaction between light and medium = A o e a/a + B o e b/a + G / ~ 4 --t-F 0 e f/'~ + Do e - d / k B T + Zl?,iXi "~ n / l ~ 4 -~- n / ~ 2 -nt- J
(3)
where A is the wavelength, T is the temperature, x i is the content of the ith impurity, e i is the extinction coefficient of the ith impurity, and A0, Bo, G, Fo, Do, B, H, J, a, b, f and d are the constants. The components in the right-hand part of the equation characterize in series the electron and multi-phonon absorption; Rayleigh scattering from density fluctuations; flux attenuation due to the so-called 'weak absorption tail'; and absorption on free charge carriers. Due to these components, the optical losses, calculated for hypothetically impurity-free and structurally perfect glasses, are termed intrinsic (equivalent to P0 in Eq. (1)). The last four components represent the extrinsic optical losses due to impurity absorption and scattering from impurities and defects of the structure. In a priori estimation that considers only electron and multi-phonon absorption as well as Rayleigh scattering, the minimum optical losses for As2S3 and As2Se3 are equal to (6-7) x l0 -2 dB km -1 at 4-6/.~m (1 dB km -1 - 2.3 X 10 -6 cm-1). The calculated minimum optical losses will depend upon the assumed physical model of interaction between radiation and optical medium. Thus, accounting for 'weak absorption tail' increases the calculated minimum of optical losses in As2S3 up to 20 dB km -1 (Kanamori et al., 1985). The measured optical losses in chalcogenide glasses are noticeably higher than the intrinsic losses. This is due to a great number of impurities in the glasses obtained at present. The elemental composition, aggregation and chemical forms of impurity present in vitreous arsenic chalcogenides depend on their macrocomposition, and on conditions of their synthesis. The method of preparation and the degree of purity of the initial substances (arsenic, chalcogens) are of importance. A few impurity groups may be isolated in their classification in accordance with the element nature and the form of impurity present in glass (Devyatykh et al., 1998; Churbanov et al., 2001a) (Table I).
TABLE I IMPURITY GROUPS IN VITREOUS ARSENIC CHALCOGENIDES
Impurity group Light elements (gas-forming impurities) Metals Analogs of elements-macrocomponents Embedded in the glass network Dissolved compounds Heterogeneous inclusions
Impurity in the group Hydrogen, oxygen, carbon, nitrogen Transition and other metals, silicon Phosphorus, antimony, sulfur, selenium Hydrogen, oxygen, nitrogen, halogens (OH, SH, Sell, NH, AsH groups) CO2, COS, H20, N2 Carbon, silicon dioxide
Typical impurity content (ppm at.) 10-100 0.1-1.0 1- 100 0.1-10 0.01-10 106-109 cm-3
212
M. F. Churbanov and V. G. Plotnichenko
There are three basic sources of impurities in chalcogenide glasses. These include the initial substances used for the glass-forming compound synthesis. Arsenic, chalcogens as pure materials were developed for semiconductor applications. Electrically active impurities, e.g., metallic impurities, were the main object of attention. The commercial samples of As, S, Se, and Te contain 0.1-0.01 ppm wt of metallic impurities and a higher (1-10 ppm wt) content of hydrogen, oxygen, carbon and silicon. The second source of impurities is the material of a container used for the synthesis of glass-forming compounds. Impurities with a higher value of diffusion coefficients, especially hydrogen, enter the chalcogenide melt. It was found that hydrogen from silica glass, containing 100 ppm of hydrogen, enters the melt of chalcogenide glass starting from 650 ~ (Churbanov, 1995). The temperature dependence of hydrogen entry rate, v (g cm -2 s -1) is described by the expression In v = - 11.2 - (15,300/T),
8 7 3 K _ < T _ < 1023K
(4)
Chemical interaction of chalcogenides and some impurities (CS2, TeO2) with silica glass at high temperatures leads to thin layer formation of new compounds on the inner surface of a container, and to the appearance of heterogeneous inclusions in chalcogenide melt. The third source of impurities is the surrounding atmosphere and residual gas of vacuum used during the glass synthesis and glass treatment procedures. Arsenic and chalcogens have a tendency toward oxidation at high temperatures. As it follows from thermodynamic evaluation and from mass-spectrometric investigations of oxide vapors, the equilibrium partial pressure of oxygen at dissociation of arsenic, selenium and tellurium oxides at ---1000 K is at the level of 10 - 4 - 1 0 -6 Pa (10 - 6 10 -s mm Hg). This value will become lower with decreasing temperature. For this reason the oxidation of chalcogens and arsenic in the open vacuum system will be excluded for the vacuum higher than 10 - 4 - 1 0 -6 Pa. Impurities are present in vitreous arsenic chalcogenide in different forms (Table I). Atoms of impurity elements may be embedded in glass networks as non-bridging and bridging atoms. The stable impurity compounds (CO2, CS2, COS, N2, etc.) are present in dissolved state in the form of instant solution. Both these forms are well known. Heterogeneous inclusions, however, are the less studied form of impurity. They consist of substances which are hardly soluble in chalcogenide melts. The particle size depends on the glass sample origin, and it lies within 0.05 and several microns (see Fig. 1). Inclusions in arsenic chalcogenide mainly consist of carbon and silicon dioxide. Impurities in this form enter the glasses from initial substances and are also formed by interaction of chalcogen, arsenic, and chalcogenides with the apparatus material. Some impurities may present in glasses in several forms. Carbon compounds (oxides, sulfide and oxysulfide) are dissolved in glass. Elementary carbon has low solubility in arsenic chalcogenide melts and is present as heterogeneous inclusions (Churbanov, 2001 a). The oxygen in arsenic selenide present as a cage molecule As406 and as a polymer formed by structural units AsO3/2. In AszSe3 melt, the monomeric and polymeric forms are in equilibrium, i.e., frozen at the melt cooling. Taking into account the peculiarities of mutual solubility of polymers, it will be reasonable to assume the presence of polymeric arsenic oxide in the form of inclusions. At solidification of melt, the polymeric arsenic
Optical Fibers from High-purity Arsenic Chalcogenide Glasses E
100
. ............
,
0
3
-'
10
213
L
9
X
9==', ......... (D
'6 1E
1-
o
0.1-
.m
--
[,
| l ! .............
i
0
=
i :i
1
0.01 ,"
0 t-
-
O
o
0.001--
60
~
.
70
I
80
i= ....
90
?
I
1O0
,
.
110
I
120
130
9
I
140
Diameter of particles, nm
Histograms of the size distribution of impurity particles in glasses: (1) As2S 3" (2) As2Se 3" (3) As2S e 1.sTe1.5.
FIG.
1.
oxide may crystallize. The evidence for this follows from the observation of absorption bands characteristic of the transmission spectra of individual arsenic oxide in the form of glass, claudetite and A s 4 0 6 molecules in the transmission spectra of oxygen containing AszSe3 (Churbanov, 2001b). The impurities, embedded into the glass network or the dissolved impurities, manifest themselves in the total optical loss spectra of optical fibers as the bands of selective absorption. The position of the absorption band is determined by the nature of impurity. Table II gives the positions of the maxima of impurity absorption bands in vitreous arsenic sulfide and arsenic selenide. It follows from Table II that a considerable number of impurity absorption bands are present in the transparency range. The extinction coefficient is a quantitative measure for estimation of impurity effect on the optical loss in glasses, and may be given by BugerLambert-Beer law J = J0 e - ~ l
(5)
where e is the extinction coefficient, x is the impurity content, and 1 is the optical path length. The extinction coefficient values are known only for a limited number of impurities in chalcogenide glasses (Table III) (Churbanov, 1995, 2001b; Churbanov et al., 1999; Devyatykh et al., 1999). Figure 2 gives the spectral dependence of the extinction coefficient of Sell group, As203 and of sulfur in AszSe3. It follows from Tables II and III and from Figure 2 that the impurity effect of hydrogen-containing substances is mostly pronounced in 2.5-6.5/.~m spectral range, and that of the oxygen impurity is in the longer wavelength region ( 7 14/xm). The effect of sulfur impurity, embedded in the network of selenide and selenidetelluride glass is substantially weaker than that in the form of SH groups. In view of the experimental values of the extinction coefficient known for SH and Sell groups, and CO2, COS, CS2 and A s 2 0 3 , the content of these impurities in glasses with optical losses at the level of the intrinsic values should not exceed 0 . 1 - 1 0 ppb (see Table III).
214
M. F. Churbanov and V. G. Plotnichenko TABLE II THE MAXIMA OF ABSORPTION BANDS FOR THE MAIN IMPURITIES IN A s - S AND A s - S E GLASSES Compound or functional group leading to absorption
Position of the maximum of the absorption band (/xm)
OHS-H Se-H Ge-H As-H P-H H20 Ge-O P-O CO2 COS CSe2 CS2 Arsenic oxides (different forms) Se-O Si-O Non-identified bands presumably due to carbon presence
2.92 4.01, 3.65, 3.11, 2.05 7.8, 4.57, 4.12, 3.53, 2.32 4.95 5.02 4.35 6.31, 2.86, 2.79 12.8, 7.9 8.3 4.33, 4.31, 15.0 4.95 7.8 6.68, 4.65 15.4, 12.7, 9.5, 8.9, 7.9, 7.5 10.67, 11.06 9.1-9.6 4.65, 5.17, 5.56, 6.0
T h e p r e s e n c e o f i m p u r i t i e s in the f o r m o f h e t e r o p h a s e i n c l u s i o n s o f s u b m i c r o n size l e a d s to a d d i t i o n a l optical loss due to a b s o r p t i o n a n d scattering, the effect o f w h i c h d e p e n d s on the c o n t e n t o f p a r t i c l e s and their size distribution. M e c h a n i c a l strength, a n d p a r t i c u l a r l y the d a m a g e t h r e s h o l d o f o p t i c a l fibers, is also sensitive to the p r e s e n c e o f
TABLE III THE VALUES OF EXTINCTION COEFFICIENTS FOR THE IMPURITIES IN CHALCOGENIDE GLASSES Impurity compound or functional group
SH Sell
Glass
As203
As2S3 As2S3 As253 As2S3 As2S3 As2Se3
Se-O S S
As2Se3 AszSe3 As2Sel.sTel.5
CO2
COS C52
Maximum of absorption band ( / x m )
Extinction coefficient a (dB km -1 ppm -1)
Calculated content of impurity leading to the optical loss equal to the intrinsic lossa (ppb)
4.0 4.5 4.33 4.95 6.68 12.65 9.5 10.6 10.6 14.5
2500 1000 1.5 X 104 105 4.8 • 105 4.3 X 104 1030 380 0.52 32
0.3 0.1 0.05 0.008 0.2 100 400 2000 106 104
aThe intrinsic loss of glass is estimated accounting for 'the weak absorption tail'.
Optical Fibers from High-purity Arsenic Chalcogenide Glasses
10s r
'
'
'
'
'
'
'
' '
'
'
"
'
"
215
!J
10 4 2
10 s
o. o.
r
"o
li\
10 2
,
,,...
101-
100 10-1 ,
2
I
,
I
4
,,
6
I
,
I
8
,
I
10
i
|
12
,
14
i6
~,, I.ml
FIG. 2. Spectral dependence of extinction coefficient of Sell groups (1), oxygen (2) and sulfur (3) in AszSe3 glass.
impurities in the form of heterophase inclusions. The energy threshold of the volume damage of AszS3 glass under the effect of Y A G - E r laser pulse radiation increases by two orders of magnitude with lowering the content of submicron-size particles in glass from 106 to 104 c m - 3 (see Fig. 3). The As283 glass samples with the particle content lower than the limit of detection were not damaged by the pulses with energy more than 1 J (Kamensky et al., 1998).
50" 40" 30" Iii 20" 10' i
4'.o
"
4' .~
'
~'. o
"
~',~
"
6'. o
"
'6',~
"
IgN
FIG. 3. The dependence of volume damage threshold for As2S3 glass upon the content of submicron particles: N is the particle content (cm-3); Elmp is the pulse energy (mJ) of YAG-Er laser.
216
M. F. Churbanov and V. G. Plotnichenko
In the ideal case the glass intended for power optics applications should not contain any particle in the volume illuminated by intense radiation flux. The value of 103104 cm -3 may be accepted as the upper limit of particle content in this case.
2.3.
PREPARATION OF HIGH-PURITY GLASSES BASED ON ARSENIC CHALCOGENIDES
As follows from previous consideration, low impurity content is an unalterable condition for providing high transparency of chalcogenide glasses. At present there are two efficient ways of preparing arsenic chalcogenide melts with low impurity content, the solidification of which gives homogeneous high-purity glasses. First is the synthesis of glass-forming compounds or their mixture by vacuum melting the charge prepared by the above-mentioned methods. This variant can be used for preparation of all chalcogenide glasses. The existing problems are connected with a rather high (up to tens ppm) content of the impurities of oxygen, carbon, and hydrogen in the commercial samples of pure chalcogens and arsenic. Selenium, tellurium, and arsenic can oxidize at the contact with environment even at room temperature. That is why the starting high-purity substances are loaded into the container for glass synthesis by evaporation under oil-free vacuum. Ultrapurification of arsenic and selenium from submicron particles, consisting mainly of carbon, by the method of vacuum sublimation or distillation is of low efficiency at the acceptable evaporation rates (Devyatykh et al., 1998). Arsenic monosulfide As4S 4 w a s used as an arsenic-containing component of the charge to manufacture the glasses with the ratio of As/S equal to 1/1 and less. This compound is more suitable for the ultrapurification from submicron particles because of low viscosity of the melts (Devyatykh et al., 1999). Two- and three-component glasses were manufactured by melting the purified arsenic monosulfide with the required amount of chalcogens, e.g., As4S 4 -+- S 2 ~ 2As2S 3
(6)
As4S 4 -+- Se 2 ~ 2As2S2Se
(7)
The second method includes the purification of glass-forming arsenic compounds of technical quality by chemical and distillation methods, melting of some distillate fractions up to homogeneous state and melt solidification. Vacuum distillation in the open and closed system decreases the content of impurities of highly volatile substances and of submicron particles by 3 - 1 0 times (Churbanov, 1992). Particles behave themselves as hardly volatile impurities and their content decreased in the final distillate fraction. The efficiency of purification is very sensitive to the evaporation rate and to the melt viscosity. Depending on the type of distillate fraction taken for the subsequent treatment, it is possible to prepare glasses with different content of various impurities. The glass from the first and final distillate fraction is enriched with hydrogen, oxygen compounds, and with heterogeneous particles, respectively. To remove some impurities (carbon, oxygen), the preliminary chemical treatment of arsenic chalcogenide may be used. In order to remove oxygen, chemically bound with glass macrocomponents, a small amount of magnesium (aluminum, rare-earth elements) was added to the melt of the compound being purified with the goal to bind oxygen into the magnesium oxide. After heating at 700-800 ~
Optical Fibers from High-purity Arsenic Chalcogenide Glasses
217
50. . . . . . . . . . 48 r 46 (n-
<44'
42' 40' o
....
"
'
Part of distillated substance. %
FIG. 4. Average distillate composition as a function of evaporated substance at vacuum distillation of As2S3 (1) and AS484 (2).
the melt was subjected to vacuum distillation. To reduce the carbon content, similar procedures with addition of small amounts of arsenic and selenium oxides may be applied. A special feature of the process is the difference in macrocomposition of the distillate and of the initial chalcogenide melt. Arsenic chalcogenide evaporates with dissociation, and it leads to separation of vapor components during condensation (Churbanov et al., 2001a). In Figure 4 the average distillate composition is given as a function of a part of evaporated loading in the process of vacuum distillation of As2S3 in the isolated system. It is seen that AszS3 distillate is enriched with arsenic during vacuum distillation. This phenomenon should be taken into account while preparing arsenic chalcogenide layers by condensation of their vapors. Stable reproducibility of glass properties in samples prepared in different experiments is a practically important question. Variation in glass properties during the synthesis is due to high sensitivity of glasses to the impurities. Absolute and relative content of impurity compounds and groups in glasses may change due to variations in impurity content of the initial substances and due to difference in the conditions for glass synthesis and melt solidification (vacuum level, timetemperature modes). Relative content of the glass network defects, e.g., 'chalcogenchalcogen' and 'arsenic-arsenic' bonds, is also affected by variations in conditions. For example, the presence of sulfur at the level of several atomic percents in selenide glass will not noticeably affect its transparence at 7-11 ~m wavelength range. But the impurity hydrogen will change its chemical form. The main part of hydrogen will be bound with sulfur atoms due to higher strength of S - H bond as compared with that of S e - H bond. The intensity of Sell absorption band will diminish considerably, while the intensity of SH band will increase in the transmission spectra of glass samples (Churbanov et al., 2001b). Figure 5 shows the absorption spectra from 3000 to 1500 cm -1 for AszSe3 (1), AszS3 (2) and (AszSe3)o.95(AszS3)0.05 (3). The last glass was prepared from glasses 1 and 2.
218
M. F. Churbanov and V. G. Plotnichenko 1.0
|
!
|
0.8
i
9
i
!
|
~~H
0.6 8 0.4
j
3000 2800 2600 2400 2200 2000 1800 1600 Wavenumber. crn ~
FIG. 5. A hydrogen transfer from Se to S in the mixed As2S3-As2Se3 glasses: (1), (2), and (3) are the absorption spectra of As2S3, As2Se3, and (As2Se3)o.95(As2S3)o.05,respectively. TO prepare vitreous arsenic chalcogenide with the reproducible optical properties, a strict identity of the impurity content in the initial substances should be provided, as well as the conditions of the glass preparation. In Table IV the content of impurities in arsenic chalcogenide glasses is given. It is seen that the content of ordinary metals and silicon is 0 . 1 - 1 . 0 ppm wt. The content of oxygen and carbon determined by reaction gas-chromatography method and by activation method for the best glass samples is equal to 0 . 2 - 0 . 6 ppm at. The content of hydrogen in the form of SH, Sell and OH groups determined by IR spectroscopy of bulk samples and fibers is equal to 0 . 0 2 - 0 . 2 ppm at. The content of particles with diameter of 0 . 1 / x m is lower than 1 • 104 cm -3 for As2S3 glass as a result of additional efforts directed to the removal of particles. In A s - S e and A s - S e - T e glasses the particle content is equal to 1 • 105-5 • 106 cm -3. The level of transparency in the middle IR range corresponding to the achieved degree of purity in arsenic chalcogenide glasses is given in Table V. It is obvious that the difference between the predicted and actual optical losses is of two orders of magnitude.
TABLE IV IMPURITY CONTENT IN THE PREPAREDARSENIC CHALCOGENIDES Impurity
Method of determination
Impurity content in As2S3
Hydrogen in the form of Sell and SH groups Carbon Oxygen Submicron size particles, 0.06-0.1/xm A1, Fe, Ni, Cr, Mg, Mn, Ni
As2S~.sSe~.5 1 ppm at.
As2Se3
Spectroscopy of fiber
0.05 ppm at.
0.5 ppm at.
Activation and gas chromatography Activation Laser ultramicroscopy
---0.2 ppm wt 1 ppm wt -<2 • 104 cm- 3
5 • 106 cm -3
0.9 ppm wt 2 • 106 cm -3
Emission spectroscopy
1-100 ppb wt
< 0.2-2 ppm wt
1-1 O0 ppb wt
0.6 ppm wt
219
Optical Fibers from High-purity Arsenic Chalcogenide Glasses
TABLE V OPTICAL TRANSPARENCE OF THE BEST SAMPLES OF VITREOUS ARSENIC CHALCOGENIDES Glass
Minimum optical losses
As2S3 As2S 1.5Se1.5 AszSe3 AszSe 1.5Te1.5
3. 3.1.
At laser wavelength (dB
km-1)
/3 (dB km -1)
A (~m)
2.08/xm
2.9/xm
3.7/xm
5.6/xm
10.6/xm
23 60 78 160
2.4 4.8 4.2 6.6
400 1000 160
160 3800 160
140 400 80
230 200 85 1000
650 3500
Chalcogenide Glass Fibers with Low Optical Losses FABRICATION OF OPTICAL FIBER
Optical fiber, based on the complete internal reflection principle, is an extensive double-layer structure consisting from core and cladding (in the simplest case a doublelayer fiber of round cross-section). The refraction index of the core is higher than that of the cladding. The light, introduced into fiber at the angle less than a certain critical value, will propagate only through the core. The values of refraction index of glasses for the core (nl) and for the cladding (n2) with angle of complete reflection (Oc) are related by the equation Oc - 2 arcsin(n 2 -
n2) 1/2
(8)
The glasses used for manufacturing of optical fibers should meet a number of requirements. These requirements include the spectral transmission range and the level of optical losses at its different points, the refraction index and its spectral dependence, the mechanical strength, the chemical stability, and stability to crystallization. Due to this fact, quantitative data on the properties of glasses, required for manufacturing of optical fibers with their subsequent application, are quite voluminous. Structure, fiber characteristics, the level of optical loss in the required spectral range, aperture and other parameters are different in chalcogenide glass fibers depending upon various actual applications. Among them are the multi-mode fibers with and without reflecting cladding from the corresponding chalcogenide glass, the fibers with and without protective polymeric cladding, the fibers with reflecting polymeric cladding (see Fig. 6), and single-mode fibers. There are about 10 tested parameters for the fibers. They include the diameter of the core, the ratio of the core/cladding diameters, the concentricity of arrangement of core and cladding, the concentricity of the secondary coating, the numerical aperture, the mechanical strength, the minimum admissible radius of bending, the continuous length of fiber, the level of optical losses in the required range, etc. In the process of development of set-ups for fiber drawing from high-purity chalcogenide, the main problem was connected with the possibility to provide and control all the fiber parameters. Chalcogenide glass optical fibers are manufactured by drawing either from the cylindrical rod (fillet), or from the rod placed into the tube of cladding glass
220
M. F. Churbanov and V. G. Plomichenko
FIG. 6. Cross-sectionof chalcogenide glass fibers of different types: (1) a 'naked' (without glass cladding) fiber; (2, 3) a glass-polymer fiber; (3, 4) a glass-glass fiber; (5) a multi-core fiber.
(the 'rod-in-tube' method). The top end of the preform is fixed in the holder of set-up used to draw the fiber and the bottom end is located inside the tubular resistance heater (Fig. 7). The fiber is being drawn from the heated (softened) end of the preform and wound on the rotating intake drum. A purified inert gas is supplied to the drawing zone to protect the heated chalcogenide glass from oxidation. The rate of drawing of the optical fiber in this method is from 1 to 10 m min-1, the diameter of the optical fibers manufactured is from 0.1 to 0.8 mm. With this method the optical fibers with reflective and non-reflective glass cladding were manufactured from A s - S , As-Se, A s - S e - S , A s - S e - T e , G e - A s - S , and G e - A s - S e glass systems. The tubes for the reflective cladding are manufactured by the method of spun casting. The drawing of optical fibers from the melt is carried out by single and double-layer crucible method. The experimental technique is similar to that used for drawing optical fibers from silicate glasses. Figure 8 shows one of the variants of the double crucible used for drawing double-layer optical fibers from chalcogenide glasses (Scripachev et al., 1994). The glass for the core and the cladding, heated to plastic state, is supplied into the concentric double extrusion nozzle. The conical 'drawing cone' is formed at the outlet of the extrusion nozzle, and this cone is used to draw a double-layered optical fiber. The rate of drawing and the diameter of optical fiber are adjusted by variation
Optical Fibers from High-purity Arsenic Chalcogenide Glasses
f7
Clz
221
Ar*Clz |
i
7
i
q
, Av
tJ ttococ
9
B
ff
At
l
A
A-A
@
tg
Fic. 7. A scheme for set-up for drawing the fiber by the 'rod-in-tube' method: (1) tube; (2) rod; (3) preform holder; (4-7, 17) valves; (7) vacuum meter; (8) resistance heater; (9) humidity controller; (10) gas flow rate controller; (11) inert gas purification system; (12) inert gas reservoir; (13) vacuum pump; ( 14, 15) inert gas lines; (16) intake drum; (18, 19) applicator and furnace application and drying of polymeric coating; (20) fiber diameter controller. in the rotational rate of the intake drum and in the pressure of inert gas over the melt. The ratio of core/cladding diameter can be changed by changing the diameters of channels for the stream of core/cladding melt and by changing the v o l u m e rate of melt flow.
222
M. F. Churbanov and V. G. Plotnichenko ,4-,4
6
Pf~ --
J
q~02 i
5
J t g
FIG. 8. A double crucible scheme (the position before the beginning of fiber drawing): (1) core glass; (2) clad glass; (3, 4) vessel for core and clad glasses; (5) a sealing flange during supply of inert gas.
In the drawing process, the primary protective coating is applied to optical fiber from polymer or fusible metal. F-42 fluoroplastic, applied from methylethylketone solution, is often used as a polymer for the primary coating. Polyvinylchloride and photo-solidified polyacrylates are used as secondary polymer coating. Each of the methods for manufacturing optical fibers has its merits and limitations. The crucible method provides a high quality end surface of optical fiber, and as a result, a higher mechanical strength of optical fiber. It is easier to control the ratio of core/cladding diameters in the 'rod-in-tube' method. The crucible drawing is more preferable to manufacture optical fibers from glasses stable to crystallization. The required difference in refraction indices of core/cladding glass is provided by cation and anion doping of the base glass. For example, the replacement of a part of arsenic for germanium or a part of selenium for sulfur in AszSe3 leads to the decrease in refraction index. It is possible to control the refraction index in two-component glasses (As-S and A s - S e systems) by changing the ratio of cation and anion components.
3.2.
PROPERTIES OF CHALCOGENIDE GLASS FIBERS
Actual applications of chalcogenide glass optical fibers depend on their parameters (mainly of technical and operating nature) meeting the requirements of the specific technical task.
Optical Fibersfrom High-purity Arsenic Chalcogenide Glasses
223
Optical losses and their spectral dependence, i.e., the total optical loss spectra, are the main parameters of optical properties. Their experimental values are based upon the Lambert-Beer law (Eq. (2)) and comprise the measurement of Jo/J for two parts of optical fiber of different lengths. The conditions for input radiation into optical fiber should be equal for both parts. This requirement is met by the fact that initially the transmission spectrum of the longer part of optical fiber is registered and after that a part of the output end of optical fiber is cut off with subsequent registration of transmission spectrum. The optical losses are estimated by the equation
/3-
lg(J2/J1) At
(9)
where Al is the difference in length of the first and the second parts of optical fiber. This technique is called as a two-point or 'breaking-off' technique. This technique allows determination of the value of optical losses with the mean-square deviation of 5% at confidence level of 0.95. The spectrum of the total optical losses comprises the data on the spectral width of transparency region in optical fiber, on transmission in separate parts and on wavelength in this region, on the bands of selective absorption. Figures 9 - 1 2 give the spectra of the total optical losses in optical fibers based on arsenic chalcogenides (Vasiliev et al., 1993; Scripachev et al., 1994, 2001; Devyatykh et al., 2000; Churbanov et al., 2001b, 2002a; Dianov et al., 2003). It is seen from the figures that the transparency range of optical fibers is shifted into the longer wavelength region while shifting to heavier chalcogen. It is an essential factor that the lower optical losses are expected to be found while manufacturing optical fibers from
1000
g
8OO
E
600
400
200
2
3
4
5
6
;L. mkm
FIG. 9. Spectraldependence of the total optical losses in multi-modeAs2S 3 glass fiber. Core/claddingdiameter is 195/300/zm; numerical aperture is 0.27.
224
M. F. Churbanov and V. G. Plotnichenko
1000
100
%', ~~176 ; ,:,, "..:~-_23 d B / ~
10
7 .mkm
FxG. 10. Total loss spectra of As2S3 (1) and As2S].sSel.5 (2) glass fibers.
glasses of higher purity. Impurity absorption bands are present in the transmission spectra of all fibers. Mechanical strength of the fibers was determined in two-point bending between parallel plates at 20 ~ in air (Dianov et al., 1990). The fiber was bent into a loop and --1 fixed between the planes, which were then driven closer together at a rate of 1 mm s
IODO0
/" J
1000
/
i 100 2
;3
4
$
6
"t
3
9
10
11
Lmkm
FIG. 11. Total loss spectra of A s - S e - T e glass fiber. Core/cladding diameter is 200/300/xm; core glass is Ge 1As38Se41Te2o.
Optical Fibers from High-purity Arsenic Chalcogenide Glasses
225
0.8 E
t
"~ 0.6 o:
9
= O
.~%
f,
0.4
r.'.
.,..
E c~ =
0.2 t~ ~
I
1400
I
16100
I
t~
I
18;0
2000
'
2200
Wavelength. nm FIG. 12. Transmissionloss spectraof As2S3 single-modefibers with core diametersof (1) 3/xm, (2) 4/xm, and (3) 7/xm. until fiber failure. The corresponding distance D between the plates was determined using an optoelectric sensor with an accuracy of 1%. The ultimate strain e and fracture stress owere determined by the formulas 2r e -- 1 . 1 9 8 D ~ _d ,
o r - Ee
(10)
where E is Young's modulus of the glass, r is the radius of the fiber without coating, d is the total diameter of the fiber, and D is the distance between the plates. The fiber radius r was measured on both fracture surfaces using an optical microscope. Each strength value was the average of at least 30 repeated measurements. The results were used to construct Weibull plots, representing the probability of fiber failure, F, as a function of applied stress. The accuracy in mechanical tests was - 3 % . For some of the glasses, Young's modulus E is equal to 16.7 GPa for As4oS6o, 17.0 GPa for As40Se6o, 18 GPa for GesAs38Se57 and approximately four times lower than for silica (70 GPa) (Sanghera and Aggarwal, 1998). Figures 13 and 14 give the Weibull distributions for multi-mode optical fibers with diameter of 400/xm made of different glasses and for single-mode optical fiber with diameter of 125/xm made of arsenic sulfide. All optical fibers are manufactured by the crucible method. A higher strength of optical fibers from As2S3 is due to a higher degree of purity of the glass compared to the vitreous As2Se3 and As2S1.sSe]. 5. Table VI gives the generalized data on properties of optical fibers fabricated from highpurity arsenic chalcogenide glasses. The time stability of optical and mechanical properties is one of the most important operating parameters of optical fibers. Investigations demonstrated that optical and mechanical properties of optical fibers from As2S 3 glasses do not degrade in 3 - 5 years when their side surface is protected with a double polymeric or metal coating (Churbanov et al., 2002b). In general, the current quality of chalcogenide optical fibers is sufficient for their various production applications.
226
M. F. Churbanov and V. G. Plotnichenko _ I"-::-
:
. . . . . .
~
--
".
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
"'
9
-
"; . . . . . .
I
+
FD
+
J
9
/ 9
..1111/
,,
n
i
"
-1
4-
-2
J
.
I"
4
I n
/,
+
./;/
-3
t
II
-4 9 -~--
--~
......
-. . . . . . . .
:i
~
"
~
~ . . . .
0.2
-
:,
. . . . . .
0.4
i .........
,:
--
0.6
88
-
~ ....
|. . . . .
0.8
: :
-
- ......
1
-
.............
2
Bending strength, GPa
FIc. 13. Weibull plots for the bending strength of fibers. Core glass: As2Si.sSel.5 (1), As2Se3 (2), As2S3 (3). Fiber diameter is 400/zm; the primary coating is from fluoroplastic F-42.
3.3.
SOME APPLICATIONS OF CHALCOGENIDE GLASS FIBERS
Interest in the middle IR range optical fibers is largely due to their properties of the middle IR radiation. It may be considered as the information source on the presence and
102 L-
t.r " d~ dz
.,=.
~1 9
o ,,',1
"2
"
I t / ID~
9
9
{O 9
.,.
1130
I
800
,
t
1000
i
i
1200
,
,
1400
,
J
1600
Breaking stress. MPa FIG. 14. Weibull plots for the bending strength of single-mode As2S3 fibers with core diameters of (1) 2.7 Ixm and (2) 7.3 Ixm.
Optical Fibers from High-purity Arsenic Chalcogenide Glasses
227
T A B L E VI BASIC PERFORMANCES OF THE BEST FIBERS MADE OF HIGH-PURITY ARSENIC CHALCOGENIDE GLASSES Fiber characteristics
Parameter
Core diameter (/~m)
3 - 3 0 for single mode up to 800 for multi-mode 80-90 Tens-hundreds meters up to 1 km for AszS 3 fibers 0.5-1.2 Up to 100-120 23 at 2.4/zm (As2S3) 80 at 4.3/~m (AszSe3) 60 at 4.8/~m (As2Si.sSel.5) 160 at 6.6/~m (AszSel.sTel.5)
Core/clad concentricity (%) Continuous fiber length Bending strength (GPa) Working temperature (~ Minimum optical loss at wavelength (dB km -1)
Optical loss at laser wavelength (dB km -1) YAG:Er 3+ (A -- 2.94/~m) CO (A = 5.5-6.3/~m) CO2 (A = 9.2-11.3/~m) Numerical aperture Power of radiation transmitted through fiber YAG:Er 3+ laser CO laser CO2 laser
160 100-200 600-1600 0.12-0.5 1.5 kJ cm -2 -> 10 W 1.8 W
temperature of heated bodies. In accordance with Stephan-Boltzmann law and Planck's law, the radiation parameters (wavelength, spectral distribution of radiation intensity) determine the temperature of radiator. Application of optical fibers facilitates the solution of some technical problems in medical and technical diagnostics, including contactless temperature measurement in technological processes. The middle IR radiation is a convenient energy form for the treatment of materials and biological tissues. It leads to possibility of such fiber applications as laser surgery, mild intercavity heating of biological tissues. The main part of biological tissues contain up to 70-90% of water and can absorb radiation of YAG-Er, CO, CO2 and some other lasers. The delivery of laser power through the fibers may be considered an efficient solution in some cases. The vibration frequencies of different chemical bonds and functional groups lie at least in the middle IR region. The use of IR fibers permits to realize the qualitative and quantitative analysis of gases, vapors, and liquids, as well as the remote control of environment. A general picture of potential and actual applications of IR optical fibers is given in the book Infrared Fiber Optics (Sanghera and Aggarwal, 1998). Some examples of actual applications of chalcogenide glass optical fibers are given below.
3.3.1. InfraredPyrometry Application of optical fibers appears to be the most convenient in the case of strong electromagnetic disturbances as well as when the temperature is being measured in
228
M. F. Churbanov and V. G. Plotnichenko
the objects located in a hard-to-reach place or the intermediate environment is present between them with pyrometer affecting the reading of measurement data. The use of silica fibers for these purposes is restricted to the region of sufficiently high temperatures usually exceeding 200-300 ~ Thermal control is of special interest to biological objects, parts of operating electrical machines, units of microelectronics and other devices with temperature below 200~ or even close to the room temperature. The low-temperature pyrometer with a fiber made from GesAs38Ses7 glass of 2 m length is described to provide the reduction of the temperature boundary down to 40 ~ at 800 Hz modulation frequency with the use of non-cooled PbSe photoresist as a detector of radiation (Svet et al., 1985; Vasiliev et al., 1985). Possible application of chalcogenide fibers to the instruments for medical diagnostics is shown (Devyatykh et al., 1991b) where the bundle is made of 180 fibers with 250/xm diametermthe temperature of retina was controlled after surgical cataract and glaucoma treatment. 3.3.2.
Analytical IR Spectroscopy
Use of fiber for analytical purposes extends the capacity of IR spectroscopy, since it makes possible to investigate gaseous, liquid and solid samples that cannot be placed into the measuring camera of spectral instrument. Fibers eliminate many restrictions on sample sizes and forms. Depending on the character of registered signal, the measuring IR-fiber-assisted systems can be subdivided into four major types. These are systems for measuring the spectra of radiation, transmission, reflection and total internal reflection with multiple disturbances. Fibers made from glasses of As-S, As-Se and G e - A s - S e systems are applicable for the radiation output from semiconductor lasers operating at liquid nitrogen temperature, from cryostats in diode laser spectrometers of ultrahigh resolution (Zasavitsky et al., 1987) using multi-pass gas cells for the compositional analysis of high-purity volatile substances (Artjushenko et al., 1991; Kuznetsov et al., 1992). 3.3.3.
Power Fiber Optics
Transmission of radiation from high power IR sources, such as HF (2.7/xm), YAGEr 3+ (2.94/zm), DF (3.8/xm), CO (5.3-6.2/xm) and CO2 (10.6/xm) lasers for medical, technological and other special purposes, through the IR fibers is considered to be one of the major trends in their usage. It will allow the replacement of cumbersome and inconvenient mirror manipulators. For example, the minimum laser radiation power, required for surgical purposes, is measured in the tens of watts. To provide transmission, the optical fiber losses should not exceed 200-400dB km -1. The radiation of continuous CO laser with output power up to 10 W was delivered during several hours through the fiber from As35S65 glass of 480/xm diameter and 1.5 m length (Dianov et al., 1984). A system for YAG-Er laser microsurgery in ophthalmology is based on the use of fiber from high-purity arsenic sulfide glass is described elsewhere (Kamensky et al., 1998).
Optical Fibers from High-purity Arsenic Chalcogenide Glasses
229
Single-mode chalcogenide glass optical fibers are used as sensors for various physical fields (Plotnichenko et al., 1984) and for development of ultra-fast all-optical switches (Asobe, 1997).
References Artjushenko, V.G., Plotnichenko, V.G., Stepanov, E.V., Nadezshdinskii, A.I., Kuznetsov, A.I. and Moskalenko, K.L. (1991) Infrared diode laser chemical sensors with multipass cell based on silver halide and chalcogenide fibers, Proc. SPIE, 1591, 206-217. Asobe, M. (1997) Nonlinear optical properties of chalcogenide glass fibers and their application to all-optical switching, Opt. Fiber Technol., 3, 142-148. Churbanov, M.F. (1992) Recent advances in preparation of high-purity chalcogenide glasses, J. Non-Cryst. Solids, 140, 324-326. Churbanov, M.F. (1995) High-purity chalcogenide glasses as materials for fiber optics, J. Non-Cryst. Solids, 184, 25-29. Churbanov, M.F. (1998) Purification of chalcogenide glasses. In Properties, Processing and Applications of Glass and Rare Earth Doped Glasses for Optical Fibers (Ed., Hewak, D.) EMIS Data Reviews Series No. 22, pp. 340-343. Churbanov, M.F. (2001a) Formation of second-phase inclusions in molten As2Se3 via chemical transport of carbon, Inorg. Mater., 37, 339-341. Churbanov, M.F. (200 l b) Effect of oxygen impurity on the optical transmission of As2Se3.4 glass, Inorg. Mater., 37, 1388-1394. Churbanov, M.F., Shiryaev, V.S., Smetanin, S.V., Dianov, E.M., Plotnichenko, V.G., Qingheng, H., Guangping, L. and Hongfeng, S. (1999) Effect of sulfur on optical transmission of As2Se3 and As2Sel.sTel.5 glasses in the range 500-1110 cm- ~, Inorg. Mater., 35, 1229-1234. Churbanov, M.F., Scripachev, I.V., Snopatin, G.E., Shiryaev, V.S. and Plotnichenko, V.G. (2001a) High-purity glasses based on arsenic chalcogenides, J. Optoelectron. Adv. Mater., 3, 341-349. Churbanov, M.F., Shiryaev, V.S., Scripachev, I.V., Snopatin, G.E., Gerasimenko, V.V., Smetanin, S.V., Fadin, I.E. and Plotnichenko, V.G. (2001b) Optical fibers based on A s - S - S e glass system, J. Non-Cryst. Solids, 284, 146-152. Churbanov, M.F., Shiryaev, V.S., Skripachev, I.V., Snopatin, G.E., Pimenov, V.G., Smetanin, S.V., Shaposhnikov, R.M., Fadin, I.E., Pyrkov, Yu.N. and Plotnichenko, V.G. (2002a) High-purity As2S1.sSel.5 glass optical fibers, Inorg. Mater., 38, 193-197. Churbanov, M.F., Shiryaev, V.S., Gerasimenko, V.V., Pushkin, A.A., Skripachev, I.V., Snopatin, G.E. and Plotnichenko, V.G. (2002b) Stability of optical and mechanical properties of chalcogenide fibers, Inorg. Mater., 38, 1063-1068. Devyatykh, G.G., et al. (1991a) Fiber waveguides based on high-purity chalcogenide glasses, High-Purity Subst., 5, 1-27. Devyatykh, G.G., et al. (1991b) Optical fiber IR radiometer for medical diagnosis, High-Purity Subst., 5, 188-191. Devyatykh, G.G., Churbanov, M.F., Shiryaev, V.S., Snopatin, G.E. and Gerasimenko, V.V. (1998) Impurity inclusions in extra-pure arsenic and chalcogens, Inorg. Mater., 34, 902-906. Devyatykh, G.G., Churbanov, M.F., Scripachev, I.V., Snopatin, G.E., Dianov, E.M. and Plotnichenko, V.G. (1999) Recent development in As-S glass fibers, J. Non-Cryst. Solids, 256/257, 318-322. Devyatykh, G.G., Dianov, E.M., Plotnichenko, V.G., Pushkin, A.A., Pyrkov, Yu.N., Skripachev, I.V., Snopatin, G.E., Churbanov, M.F. and Shiryaev, V.S. (2000) Low loss infrared arsenic chalcogenide glass optical fibers, SPIE Proc., 4083, 229-237. Dianov, E.M., Masychev, V.I., Plotnichenko, V.G., Sysoev, V.K., Baikalov, P.I., Devyatyk, G.G., Konov, A.S., Scripachev, I.V. and Churbanov, M.F. (1984) Fiber optics cable for the delivery of CO-laser radiation, Electron. Lett., 20, 129. Dianov, E.M., Krasteva, V.M., Plotnichenko, V.G., Semenov, S.L., Scripachev, I.V. and Churbanov, M.F. (1990) Strength of optical fibers based on chalcogenide glasses obtained by the crucible method, HighPurity Subst., 4, 767-772.
230
M. F. Churbanov and V. G. Plotnichenko
Dianov, E.M., Plotnichenko, V.G., Pyrkov, Yu.N., Smol'nikov, I.V., Koleskin, S.A., Devyatykh, G.G., Churbanov, M.F., Snopatin, G.E., Skripachev, I.V. and Shaposhnikov, R.M. (2003) Single-mode As-S glass fibers, Inorg. Mater., 39, 627-630. Frerics, R.J. (1953) J. Opt. Soc. Am., 43, 197. Kamensky, V., Scripachev, I.V., Snopatin, G.E., Pushkin, A.A. and Churbanov, M.P. (1998) High power As-S glass fiber instrument for pulse YAG-Er laser radiation, Appl. Opt., 37, 5596-5599. Kanamori, T., Terunuma, Y., Takahashi, S. and Miyashita, T. (1985) Transmission loss characteristics of As4oS6o and As3sGesSes7 glass unclad fibers, J. Non-Cryst. Solids, 69, 231-242. Kapany, N.S. and Simms, R.J. (1965) Recent developments in infrared fiber optics, Infrared Phys., 5, 69-76. Katsujama, T. and Matsumura, H. (1988) Infrared Optical Fibers, Adam Hilger, Bristol. Kuznetsov, A.I., Nadezshdinskii, A.I., Moskalenko, K.L., Stepanov, G.V., Davarashvilli, O.I., Zasavitskii, I.I., Plotnichenko, V.G. and Artjushenko, V.G. (1992) Tunable diode laser spectroscopy accessories based on middle IR halide and chalcogenide fibers, Proc. SPIE, 1724, 104-118. Lines, M.E. (1984) Scattering losses in optical fiber materials. II. Numerical estimates, J. Appl. Phys., 55, 4058-4063. Nishii, J. and Yamashita, T. (1998) In Chalcogenide Glass Based Fibers in Infrared Fiber Optics (Eds., Sanghera, J.S. and Aggarwal, I.D.) CRC Press, Boca Raton, FL, pp. 143-184. Plotnichenko, V.G., et al. (1984) Estimation of sensitivity of fiber-optic sensors based on IR optical fibers, Kvantovaya Elektronika, 11, 194-196, in Russian. Sanghera, J.S. and Aggarwal, I.D. (1998) Summary. In Infrared Fiber Optics (Eds., Sanghera, J.S. and Aggarwal, I.D.) CRC Press, Boca Raton, FL, pp. 325-335. Scripachev, I.V., et al. (1994) Fabrication of double layer fibers based on high-purity As-S, As-Se, G e - A s - S e glasses, Vysokochistye Veshchestva, 4, 34-41, in Russian. Scripachev, I.V., Churbanov, M.F., Gerasimenko, V.V., Snopatin, G.E., Shiryaev, V.S., Pushkin, A.A., Fadin, I.E., Plotnichenko, V.G. and Pyrkov, Yu.N. (2001) Optical and mechanical characteristics of fibers made of arsenic chalcogenide, J. Optoelectron. Adv. Mater., 3, 351-360. Svet, D.Ya., et al. (1985) Low temperature pyrometer with flexible optical fiber, Pribory i Sistemy Kontrolya, 2, 18-19, in Russian. Vasiliev, A.V., et al. (1985) The use of IR optical fibers at pyrometric measurements, Zhurnal Prikladnoi Spectroskopii, 42, 82-84, in Russian. Vasiliev, A.V., et al. (1993) Double-layer chalcogenide glass fibers with optical losses lower 30 dB/km, Kvantovaya Elektronika, 20, 109-110, in Russian. Zasavitsky, I.I., et al. (1987) Cryostating of semiconductor laser with radiation output via IR optical fiber, Vysokochistye Veshchestva, 5, 202-204, in Russian.
Index barrier height 39-42, 60-3, 79 bi-stable switching elements 1-3 bias voltages 12-13, 67-70 binary code conversion 28 bond geometrical percolation 108-10 boson peaks 182 Buger-Lambert-Beer equation/law 210-11, 213-14 build-in effect, camera tubes 94-5 built-in potentials 60-1, 69-70 bus-bars 32
absolute impurity content 217-18 absorption coefficient 171, 210-11 crystal-glass hetero-contacts 38 high-purity glasses 210-11, 213-15 lanthanoids 183-91 metal-glass junction 70 rare-earth doped glass 171, 178-80, 183-91, 193, 199-201 acoustic-frequency phase inverters 11-13 activation energy crystal-glass junctions 82 ion transport 104, 106-11 metal compound glasses 151-4 metal-glass junctions 66 alkali metals, chlorides 131 - 54 allowed volume percolation 109-10 aluminum 71-5 amplitude 18-19, 22-4, 26, 53 analytical infrared spectroscopy 228 anions 149-50, 178-80 anisotropy 204-5 anti-Stokes luminescence 192- 204 applied voltage 63 arsenic glasses 209-29 arsenic selenide 212-15 arsenic sulfide 173, 215-16 attenuation coefficient 210-11
calibration microelectronics 14-18 calibration set of resistors (CSR) 15-18 camera tubes 92-6 capacitance 67-70, 74-5, 88 carbon compounds 212 carrier transfer 38-9, 66-7 cations 124- 30, 151-4 CC s e e control circuits cesium chloride 131, 135-6 chalcogenide systems chemical sensors 154-65 current instability 1- 54 heterostructures 57-98 high-purity arsenic optical fibers 209-29 ion conductivity 103-48 optical fibers 209-29 rare-earth doped glass 169-205 sensors 154-65 chalcohalide glasses 131-54 charge tunneling 58, 66-7
band diagrams s e e energy level diagrams Bardeen-Herring tracer correlation factor 116 231
232 charged state density 69-70 chemical bond covalence 170-1 chemical composition optimization 194 chemical interaction formation 133-5 chemical purification methods 216 chemical sensors 154-65 chemical shift 137-8 chlorides 131-54 cladding 220-1 classical geometrical percolation 108-10 classification of chemical sensors 154-6 complete reflection 219 complex anions 149-50 composition dependence 161-2, 178-83, 194 computer code-decimal code converters 28 concentration dependency 137, 151 conductivity activation energy 108-9, 151-4 dependence 161 - 2 ion transport 128- 9 threshold switches 8 - 9 connectivity 112-13, 124- 30 contact barrier formation 57-63 control circuits (CC) 24-8 controlling parameters 8-11 converters 28 coordination numbers 179 copper chalcogenides 103-30 copper concentration sensors 163 cores 220-1 correlation functions 122- 3 correlation lengths 181-2 Coulomb barriers 152-3 Coulomb potentials 151 - 2 covalence 170-1 critical exponent 110-11 critical fictive temperature 112-13 critical percolation regime 104-14 cross-relaxation 196-7 crucible drawing 220, 222 crystal-glass heterostructures 31-45, 75-90 CSR s e e calibration set of resistors current crystal-glass heterostructures 42-4, 77-90 density 13 instability 1- 54 limiting 35-6, 79, 81 metal-glass junctions 63-7 oscillations 27-8 relaxation oscillation generators 19, 21-4
Index
current-voltage characteristics (CVC) calibration microelectronics 15 - 16 crystal-glass junctions 76-80, 87-8 electroluminescence matrix displays 31 light sensors 41 metal-glass junctions 64-5 relaxation oscillation generators 2 1 - 2 S-diodes 2 - 3 thin film thermosensors 3-4, 8 threshold switches 9-11 voltage stabilizers 13-15 damage thresholds 215-16 dark current 89 Debye depletion length 58 decimal code-computer code converters 28 delay time 51 depletion lengths 57-9 device design 9-11 diagrams of state 135-6 diffusion 104- 9, 113-18, 120- 2, 128 dipole-dipole interactions 194- 5 discontinuity band edges 83 dissipation power 5 distillation purification methods 216-17 double crucible drawing 220, 222 drawing optical fibers 219-22 drift mobility 38-9, 80-1 dysprosium 186-8 ELANI
s e e electroluminescent alphanumeric indicator electric contact phenomena 57-91 electric field effect 32 electrochemical properties 161 - 3 electroconductivity 73-4 electrode width 16-17 electroluminescence indicators (ELI), control circuits 24-8 electroluminescence matrix displays 29-31 electroluminescent alphanumeric indicator (ELANI) code 28 electronic insulators 104-7, 114, 130 electronic tongues 163-5 electrophotography 92-6 electrostimulated chemical transformation (ESCT) 70-1, 73-5, 97-8 ELI s e e electroluminescence indicators emission lifetimes 186
233
Index
energy level diagrams crystal-glass heterostructures 41-2, 83-4, 89-90 dysprosium 186-7 erbium 192, 203 metal-glass junctions 62-5 neodymium 188-9 praseodymium 183-4 ytterbium 203 energy sensitivity coefficient 5 equivalent electrical circuits 21-2 erasing parameters 34- 5, 53-4 erbium 191 - 7, 202- 3 ESCT s e e electrostimulated chemical transformation exchange reactions 136-7 excitation 183-91 exhaust gas sensors 164 exponentially distributed states 81 extinction coefficients 200, 213-15 Faber-Ziman structure factor 119-20 fabrication chemical sensors 157-8 optical fibers 219- 22 Fermi levels 57-8, 62-3, 79-80 fiber optics s e e optical fibers filament formation 47-8 first coordination spheres 178-9 first sharp diffraction peak (FSDP) 119- 21, 181, 183 four-terminal networks 11-13 free volume model 120- 2 frequency 11-13, 67-70, 141-2 FSDP s e e first sharp diffraction peak gallium metal compound glasses 131-2, 135, 137-48 Stokes luminescence 183, 185-6, 188, 190-1 gallium sesquisulfide-germanium disulfidemetal chloride glass systems 131-2, 135, 138-48 gallium sulfide 131-2, 135, 138-48, 172-7 generators of relaxation oscillations 18-24 geometrical percolation, ion transport 108-10
germanium metal compound glasses 137-48 Stokes luminescence 183, 185-6, 188, 190-1 germanium sulfide 173-5, 177 glass network structural changes 119-24 glass properties 132-5, 169-71 glass transition temperature 173-6 glass-crystal heterojunctions 31-45, 75-90 glass-formation metal compound glasses 131 - 2, 148- 50 optical fibers 216-17 rare-earth doped glass 172-7 glass-glass junction heterostructures 90-1 glass-metal heterojunctions 57-75 glassy matrices 171 - 2, 190-1 Haven ratio 113- 29 heavy element chalcogenides 177 heavy metal ion sensors 163 hetero-contacts 31-45 heterojunctions 40-5 heterophase inclusions 214-15 heterostructures applications 91-8 crystal-glass junctions 75-90 glass-glass junctions 90-1 hetero-contacts 31-45 heterojunctions 40-5 metal-glass junctions 57-75 photodetectors 91 - 2 HIC s e e hybrid integral circuits high-purity arsenic 209-29 highlighting 29- 31 historical overviews 154-7 holography 96 host matrix connectivity 112-13, 124- 30 hybrid glassy-crystalline heterostructures 75-90 hybrid integral circuits (HIC) 48 hydrolysis 73 hydroxides 72, 200, 203 I - U characteristics s e e current voltage... ideality factor 65-6 impurities 200, 210-19 infrared (IR) optical fibers 226-7 infrared (IR) pyrometry 227-8 infrared (IR) spectroscopy 228
234 inhomogeneity 181 insulator layer effects 70-5 interband absorption 190-1 interface insulator layer effects 70-5 ion selectivity 150-4, 158-60 ion transport 103- 30 conductivity 103-48, 150-4 copper chalcogenide 103- 30 critical fictive temperature 112-13 critical percolation regime 104-13 Haven ratio 113- 29 host matrix connectivity 112-13 mobile cation network 124-9 mobile ion content 119- 24 modifier-controlled regime 113- 29 percolation threshold regime 104-13 silver chalcogenide 103- 30 ion-selective electrodes 154 ion-selective field-effect transistors (ISFETs) 154, 158-60 ionic insulators 104-7, 130 IR s e e infrared ISFETs 154, 158-60 isoplaner switches 1 isotype heterojunctions 76-86 junction capacitance 67-70 lanthanoids 169, 171-90 absorption bands 183-90 composition 178-83 optical excitation 183-90 rare-earth doped glass 169, 171 - 7 structure 178-83 lead 163- 5, 177 light sensor 40-5 limiting concentrations 170-1 lithium chloride 131, 135-6 lithography 96-8 localized states density 57, 79 low mobile ion content 104-13 low selective chemical sensors 154-7 luminescence 169-71,183-4, 186-90, 193-6 s e e also electroluminescence matrices connectivity 112-13, 124-30 electroluminescence displays 29-31 glassy 171-2, 190-1
Index
Random Access Memory 46-8 rare-earth doped glass 171-7 mechanical strength 224-5 membrane materials 156-8 memory elements calibration microelectronics 15-18 current instability 1- 3 hetero-contacts 31-45 reprogrammed devices 45-54 mercapto radicals 203 metal bus-bars 32 metal-glass junctions capacitance 67-70 contact barrier formation 57-63 current transport 63-7 heterostructures 57-75 insulator layer effects 70-5 interface insulator layer effects 70-5 junction capacitance 67-70 middle infrared optical fibers 226-7 miniature silicon-based sensors 154, 159-61 mobile cation network 124- 30 mobile ion content 119-24, 128-9 modifier-controlled regime 113- 30 moisture resistance 133 molar fraction 150-1 molar volume 134-5 mono-stable switching elements 1-3 multidiffusion models 118 multiphonon relaxations 185-6, 197-8 multivibrators 25-6 neodymium 179-182, 188-90 neutron diffraction 128- 9 neutron total correlation functions 122- 3, 125 Nikolsky equation 156 non-linear optical properties 204-5 non-linear switching elements 3-18 ohmic resistance 63-7 operation mechanisms, chemical sensors 161-3 optical anisotropy 204-5 optical excitation 183-91 optical fibers fabrication 219- 22 high-purity arsenic 209-29 optical losses 219- 29 parameters 219 optical information registration media 96-8
235
Index
optical losses 210-11, 213-15, 219-29 oscillation generators 18-24 oxidation 71-2 oxygen 212-13 parameters, optical fibers 219 parasitic highlighting 29-31 passive regime 5 - 6 Pauling rule 179 penetration profiles 104-7 percolation threshold regime 104-14 performance efficiency 225, 227 period duration 18-19, 22-3, 26 phase inverters 11-13 phonon energy 169-70 photo-anisotropy 204-5 photo-current 42-4 photo-darkening 204-5 photo-detectors 91 - 2 photo-doping effect 96-8 photo-electric efficiency 40-5 photo-electric properties 84-5, 88-90 photo-electromotive force 40-5, 86 photo-emission 59-60 photo-lithography 96-8 photo-sensitivity 8 - 9 PLD s e e Pulse Laser Deposition polymer chains 142- 3, 147 potassium chloride 131, 135-6 power dissipation coefficient 5 power fiber optics 228-9 praseodymium 179-80, 183-6, 192 preparation, high-purity glasses 210-19 pulse burst Memory 53-4 Pulse Laser Deposition (PLD) 157, 159 pulse power supply voltages 19-20 purification, high-purity glasses 216-17 pyrometry 227-8 quasi-Fermi levels 79-80 radicals 203 RAM s e e Random Access Memory Raman frequency 141 - 2 Raman spectra 137-48, 182, 201 Random Access Memory (RAM) 45-54 rare-earth doped glass 169-205 anti-Stokes luminescence 192-204 glass properties 169-71 lanthanoids 169, 171-90
luminescence materials 169-71 Stokes luminescence 183-91 real-space correlation functions 122- 3 recombination 66-7 recording current amplitude 34-5 rectification 76, 78, 86-8 reflection 219 refraction index 171, 219 relative conductivity 8-9 relative diffusion coefficient 120-2 relative impurity content 217-18 relaxation oscillation generators 18-24 reproducibility 217-18 reprogrammed memory devices 45-54 resistance calibration microelectronics 15-18 crystal-glass hetero-contacts 33-4 Random Access Memory 50-2 resistors 15-18 thin film thermosensors 3 - 6 threshold switches 9 resolving power 98 responsivity 90-1 rod-in-tube method 219, 221 S-diodes 2 - 3 Samarium 179-80 Saticon 93-4 scattering 125-6, 214-15 SCLC s e e space charge limited current SCR s e e space charge region SE s e e switching elements selective chemical sensors 154-7 selenium 212-16 self-generator cells 24 sensibilizators 200- 3 sensitivity 210-16 sensors 154-65 short-circuit current 85-6 silicon-based sensors 154, 159-61 silver chalcogenides 103- 30 sinusoidal voltage 20-1 sodium chloride 131, 135 solid-state photodetectors 91-2 space charge limited current (SCLC) 35-6, 79, 81 space charge region (SCR) 57-9 spectroscopy 228 state diagrams 135-6 Stokes luminescence 183-91
236 strain 225 stress 225 structural investigations 136-48 structural unit models 142-3 structure, lanthanoids 178-83 sulfides 172-3, 215-16 suppression 23-4 switching elements (SE) 1- 18, 31-45 TCR s e e temperature coefficient of resistance temperature coefficient of resistance (TCR) 5 temperature dependence critical exponent 110-11 crystal-glass heterostructures 33-4, 80-2 metal-glass junctions 65-6, 73-4 Random Access Memory 49-52 thin film thermosensors 3-6 tested parameters, optical fibers 219 tetrahedra formation 138-9 thallium 177 thallium chloride 131 thermal detrapping rate 70 thermal sensitivity 3-6 thermionic emission 66-7 thermosensors 3-8 thickness dependence 49-50 thin film sensors 3-8, 154, 159-61 thin film thermosensors 3-8 third group metal compounds 131-54 threshold currents 6-7 threshold switching 8-11, 37 threshold temperatures 6 threshold voltages crystal-glass hetero-contacts 37-8 Random Access Memory 48-9, 51 relaxation oscillation generators 22 thin film thermosensors 6-7 time constants 5 time coordinates 21-2 time stability 225 total correlation functions 122-5 total optical losses 223-5 total reflection 219 tracer diffusion coefficient 113-17, 120- 2, 128-30 transition temperature 147-8, 173-6 transmission losses 223, 225 transparency 209-16, 218-19, 223-4
Index
transport chalcogenide glasses 103-30 copper chalcogenide 103- 30 metal-glass junctions 63-7 numbers 104-6 silver chalcogenide 103- 30 trigger regime 27-8 tunneling 14, 58, 66-7, 81 two-point bending 224-5 up-conversion 192-204 vacancy mechanisms 151-4 vacant trap states 66-7 vidicon camera tubes 92-6 viscous melts 132-3 vitreous arsenic chalcogenides 210-16 vitreous semiconductors 32 voltage applied 63 bias 12-13, 67-70 crystal-glass junctions 37-8, 77-90 electro-luminescence indicators 25-8 metal-glass junctions 67-70, 74-5 pulse power supply 19-20 relaxation oscillation generators 18-24 sinusoidal 19- 21 stabilizers 13-15 threshold switches 9-11 s e e a l s o current-voltage...; threshold... volume damage threshold 215-16 volume fractions 182 water vapor 72 Weibull distributions 225-6 well type design switches 1-2 work function 61-2 writing currents 16-17, 34-5 writing pulse 51 - 2 X-ray scattering 128- 9 X-ray total correlation functions 122-3, 125 Young's modulus 225 ytterbium 179-80, 200, 202-3 zirconium 50
Contents of Volumes in This Series
Volume 1
Physicso f l l l - V Compounds
c. Hilsum, Some Key Features of III-V Compounds F. Bassani, Methods of Band Calculations Applicable to III-V Compounds E. O. Kane, The k-p Method V. L. Bonch-Bruevich, Effect of Heavy Doping on the Semiconductor Band Structure
D. Long, Energy Band Structures of Mixed Crystals of III-V Compounds L. M. Roth and P. N. Argyres, Magnetic Quantum Effects S. M. Puri and T. H. Geballe, Thermomagnetic Effects in the Quantum Region W. M. Becket, Band Characteristics near Principal Minima from Magnetoresistance E. H. Putley, Freeze-Out Effects, Hot Electron Effects, and Submillimeter Photoconductivity in InSb H. Weiss, Magnetoresistance B. Ancker-Johnson, Plasma in Semiconductors and Semimetals
Volume 2
Physics oflll-V
Compounds
M. G. Holland, Thermal Conductivity S. L Novkova, Thermal Expansion U. Piesbergen, Heat Capacity and Debye Temperatures G. Giesecke, Lattice Constants J. R. Drabble, Elastic Properties A. U. Mac Rae and G. W. Gobeli, Low Energy Electron Diffraction Studies R. Lee Mieher, Nuclear Magnetic Resonance B. Goldstein, Electron Paramagnetic Resonance T. S. Moss, Photoconduction in III-V Compounds E. Antoncik and J. Tauc, Quantum Efficiency of the Internal Photoelectric Effect in InSb G. W. Gobeli and L G. Allen, Photoelectric Threshold and Work Function P. S. Pershan, Nonlinear Optics in III-V Compounds
237
238
Contents o f Volumes in This Series
M. Gershenzon, Radiative Recombination in the III-V Compounds F. Stern, Stimulated Emission in Semiconductors
Optical Properties of III-V Compounds
Volume 3 M. Hass, Lattice Reflection
W. G. Spitzer, Multiphonon Lattice Absorption D. L. Stierwalt and R. F. Potter, Emittance Studies H. R. Philipp and H. Ehrenveich, Ultraviolet Optical Properties M. Cardona, Optical Absorption Above the Fundamental Edge E. J. Johnson, Absorption Near the Fundamental Edge J. 0. Dimmock, Introduction to the Theory of Exciton States in Semiconductors B. Lax and J. G. Mavroides, Interband Magnetooptical Effects H. Y. Fan, Effects of Free Carries on Optical Properties E. D. Palik and G. B. Wright, Free-Carrier Magnetooptical Effects R. H. Bube, Photoelectronic Analysis B. O. Seraphin and H. E. Benett, Optical Constants
Volume 4
Physics of III-V Compounds
N. A. Goryunova, A. S. Borchevskii and D. N. Tretiakov, Hardness N. N. Sirota, Heats of Formation and Temperatures and Heats of Fusion of Compounds of Arab v D. L. Kendall, Diffusion A. G. Chynoweth, Charge Multiplication Phenomena R. W. Keyes, The Effects of Hydrostatic Pressure on the Properties of III-V Semiconductors L. W. Aukerman, Radiation Effects N. A. Goryunova, F. P. Kesamanly, and D. N. Nasledov, Phenomena in Solid Solutions R. T. Bate, Electrical Properties of Nonuniform Crystals
Volume 5
Infrared Detectors
H. Levinstein, Characterization of Infrared Detectors P. W. Kruse, Indium Antimonide Photoconductive and Photoelectromagnetic Detectors M. B. Prince, Narrowband Self-Filtering Detectors L Melngalis and T. C. Harman, Single-Crystal Lead-Tin Chalcogenides D. Long and J. L. Schmidt, Mercury-Cadmium Telluride and Closely Related Alloys E. H. Putley, The Pyroelectric Detector N. B. Stevens, Radiation Thermopiles R. J. Keyes and T. M. Quist, Low Level Coherent and Incoherent Detection in the Infrared M. C. Teich, Coherent Detection in the Infrared F. R. Arams, E. W. Sard, B. J. Peyton and F. P. Pace, Infrared Heterodyne Detection with Gigahertz
IF Response H. S. Sommers, Jr., Macrowave-Based Photoconductive Detector R. Sehr and R. Zuleeg, Imaging and Display
Contents o f Volumes in This Series
Volume 6
Injection Phenomena
M. A. Lampert and R. B. Schilling, Current Injection in Solids: The Regional Approximation Method R. Williams, Injection by Internal Photoemission A. M. Barnett, Current Filament Formation R. Baron and J. W. Mayer, Double Injection in Semiconductors W. Ruppel, The Photoconductor-Metal Contact
Volume 7
Application and Devices
Part A J. A. Copeland and S. Knight, Applications Utilizing Bulk Negative Resistance F. A. Padovani, The Voltage-Current Characteristics of Metal-Semiconductor Contacts P. L. Hower, W. W. Hooper, B. R. Cairns, R. D. Fairman, and D. A. Tremere, The GaAs
Field-Effect Transistor M. H. White, MOS Transistors G. R. Antell, Gallium Arsenide Transistors T. L. Tansley, Heterojunction Properties
Part B T. Misawa, IMPATT Diodes H. C. Okean, Tunnel Diodes R. B. Campbell and Hung-Chi Chang, Silicon Junction Carbide Devices R. E. Enstrom, H. Kressel, and L. Krassner, High-Temperature Power Rectifiers of GaASl-x Px
Volume 8
Transport and Optical Phenomena
R. J. Stirn, Band Structure and Galvanomagnetic Effects in III-V Compounds with Indirect Band Gaps R. W. Ure, Jr., Thermoelectric Effects in III-V Compounds H. Piller, Faraday Rotation H. Barry Bebb and E. W. Williams, Photoluminescence I: Theory E. W. Williams and H. Barry Bebb, Photoluminescence II: Gallium Arsenide
Volume 9
Modulation Techniques
B. O. Seraphin, Electroreflectance R. L. Aggarwal, Modulated Interband Magnetooptics D. F. Blossey and Paul Handler, Electroabsorption B. Batz, Thermal and Wavelength Modulation Spectroscopy I. Balslev, Piezooptical Effects D. E. Aspnes and N. Bottka, Electric-Field Effects on the Dielectric Function of Semiconductors
and Insulators
239
240
Contents o f Volumes in This Series
V o l u m e 10
Transport Phenomena
R. L. Rhode, Low-Field Electron Transport J. D. Wiley, Mobility of Holes in III-V Compounds C. M. Wolfe and G. E. Stillman, Apparent Mobility Enhancement in Inhomogeneous Crystals R. L. Petersen, The Magnetophonon Effect
Volume
11
Solar Cells
H. J. Hovel, Introduction; Carrier Collection, Spectral Response, and Photocurrent; Solar Cell Electrical
Characteristics; Efficiency; Thickness; Other Solar Cell Devices; Radiation Effects; Temperature and Intensity; Solar Cell Technology
Volume 12
Infrared Detectors (II)
w. L. Eiseman, J. D. Merriam, and R. F. Potter, Operational Characteristics of Infrared Photodetectors P. R. Bratt, Impurity Germanium and Silicon Infrared Detectors E. H. Putley, InSb Submillimeter Photoconductive Detectors G. E. Stillman, C. M. Wolfe, and J. 0. Dimmock, Far-Infrared Photoconductivity in High Purity GaAs G. E. Stillman and C. M. Wolfe, Avalanche Photodiodes P. L. Richards, The Josephson Junction as a Detector of Microwave and Far-Infrared Radiation E. H. Putley, The Pyroelectric Detector - An Update
Volume
13
Cadmium Telluride
K. Zanio, Materials Preparations; Physics; Defects; Applications
Volume
14
Lasers, Junctions, Transport
N. Holonyak, Jr., and M. H. Lee, Photopumped III-V Semiconductor Lasers H. Kressel and J. K. Butler, Heterojunction Laser Diodes A. Van der Ziel, Space-Charge-Limited Solid-State Diodes P. J. Price, Monte Carlo Calculation of Electron Transport in Solids
Volume
15
Contacts, Junctions, Emitters
B. L. Sharma, Ohmic Contacts to III-V Compounds Semiconductors A. Nussbaum, The Theory of Semiconducting Junctions J. S. Escher, NEA Semiconductor Photoemitters
V o l u m e 16
Defects, (HgCd)Se,
(HgCd)Te
H. Kressel, The Effect of Crystal Defects on Optoelectronic Devices C. R. Whitsett, J. G. Broerman, and C. J. Summers, Crystal Growth and Properties of Hgl-x Cdx Se Alloys
Contents o f Volumes in This Series
241
M. H. Weiler, Magnetooptical Properties of Hgl-x Cdx Te Alloys P. W. Kruse and J. G. Ready, Nonlinear Optical Effects in Hgl-x Cd~ Te
Volume 17
CW Processing of Silicon and Other Semiconductors
J. F. Gibbons, Beam Processing of Silicon A. Lietoila, R. B. Gold, J. F. Gibbons, and L. A. Christel, Temperature Distributions and Solid Phase Reaction
Rates Produced by Scanning CW Beams A. Leitoila and J. F. Gibbons, Applications of CW Beam Processing to Ion Implanted Crystalline Silicon N. M. Johnson, Electronic Defects in CW Transient Thermal Processed Silicon K. F. Lee, T. J. Stultz, and J.F. Gibbons, Beam Recrystallized Polycrystalline Silicon: Properties, Applications,
and Techniques T. Shibata, A. Wakita, T. W. Sigmon and J. F. Gibbons, Metal-Silicon Reactions and Silicide Y. I. Nissim and J. F. Gibbons, CW Beam Processing of Gallium Arsenide
Volume 18
Mercury Cadmium Telluride
P. W. Kruse, The Emergence of (Hgl-x Cdx)Te as a Modern Infrared Sensitive Material H. E. Hirsch, S. C. Liang, and A. G. White, Preparation of High-Purity Cadmium, Mercury, and Tellurium W. F. H. Micklethwaite, The Crystal Growth of Cadmium Mercury Telluride P. E. Petersen, Auger Recombination in Mercury Cadmium Telluride R. M. Broudy and V. J. Mazurczyck, (HgCd)Te Photoconductive Detectors M. B. Reine, A. K. Soad, and T. J. Tredwell, Photovoltaic Infrared Detectors M. A. Kinch, Metal-Insulator-Semiconductor Infrared Detectors
Volume 19
Deep Levels, GaAs, Alloys, Photochemistry
G. F. Neumark and K. Kosai, Deep Levels in Wide Band-Gap III-V Semiconductors D. C. Look, The Electrical and Photoelectronic Properties of Semi-Insulating GaAs R. F. Brebrick, Ching-Hua Su, and Pok-Kai Liao, Associated Solution Model for Ga-In-Sb and Hg-Cd-Te E Ya. Gurevich and E V. Pleskon, Photoelectrochemistry of Semiconductors
Volume 20
Semi-Insulating GaAs
R. N. Thomas, H. M. Hobgood, G. W. Eldridge, D. L. Barrett, T. T. Braggins, L. B. Ta, and S. K. Wang,
High-Purity LEC Growth and Direct Implantation of GaAs for Monolithic Microwave Circuits C. A. Stolte, Ion Implantation and Materials for GaAs Integrated Circuits C. G. Kirkpatrick, R. T. Chen, D. E. Holmes, P. M. Asbeck, K. R. Elliott, R. D. Fairman, and J. R. Oliver, LEC GaAs for Integrated Circuit Applications J. S. Blakemore and S. Rahimi, Models for Mid-Gap Centers in Gallium Arsenide
Volume 21 Part A J. I. Pankove, Introduction
Hydrogenated Amorphous Silicon
242
Contents o f Volumes in This Series
M. Hirose, Glow Discharge; Chemical Vapor Deposition Y. Uchida, di Glow Discharge T. D. Moustakas, Sputtering I. Yamada, Ionized-Cluster Beam Deposition B. A. Scott, Homogeneous Chemical Vapor Deposition F. J. Kampas, Chemical Reactions in Plasma Deposition P. A. Longeway, Plasma Kinetics H. A. Weakliem, Diagnostics of Silane Glow Discharges Using Probes and Mass Spectroscopy L. Gluttman, Relation between the Atomic and the Electronic Structures A. Chenevas-Paule, Experiment Determination of Structure S. Minomura, Pressure Effects on the Local Atomic Structure D. Adler, Defects and Density of Localized States
Part B J. i. Pankove, Introduction G. D. Cody, The Optical Absorption Edge of a-Si: H N. M. Amer and W. B. Jackson, Optical Properties of Defect States in a-Si: H P. J. Zanzucchi, The Vibrational Spectra of a-Si: H F. Hamakawa, Electroreflectance and Electroabsorption J. S. Lannin, Raman Scattering of Amorphous Si, Ge, and Their Alloys R. A. Street, Luminescence in a-Si: H R. S. Crandall, Photoconductivity J. Tauc, Time-Resolved Spectroscopy of Electronic Relaxation Processes P. E. Vanier, IR-Induced Quenching and Enhancement of Photoconductivity and Photoluminescence H. Schade, Irradiation-Induced Metastable Effects L. Ley, Photoelectron Emission Studies
Part C J. I. Pankove, Introduction J. D. Cohen, Density of States from Junction Measurements in Hydrogenated Amorphous Silicon P. C. Taylor, Magnetic Resonance Measurements in a-Si: H K. Morigaki, Optically Detected Magnetic Resonance J. Dresner, Carrier Mobility in a-Si: H T. Tiedje, Information About Band-Tail States from Time-of-Flight Experiments A. R. Moore, Diffusion Length in Undoped a-S: H W. Beyer and J. Overhof, Doping Effects in a-Si: H H. Fritzche, Electronic Properties of Surfaces in a-Si: H C. R. Wronski, The Staebler-Wronski Effect R. J. Nemanich, Schottky Barriers on a-Si: H B. Abeles and T. Tiedje, Amorphous Semiconductor Superlattices
Part D J. i. Pankove, Introduction D. E. Carlson, Solar Cells
Contents o f Volumes in This Series
243
G. A. Swartz, Closed-Form Solution of I - V Characteristic for a s-Si: H Solar Cells I. Shimizu, Electrophotography S. Ishioka, Image Pickup Tubes P. G. Lecomber and W. E. Spear, The Development of the a-Si: H Field-Effect Transistor and its Possible
Applications D. G. Ast, a-Si:H FET-Addressed LCD Panel S. Kaneko, Solid-State Image Sensor M. Matsumura, Charge-Coupled Devices M. A. Bosch, Optical Recording A. D'Amico and G. Fortunato, Ambient Sensors H. Kulkimoto, Amorphous Light-Emitting Devices R. J. Phelan, Jr., Fast Decorators and Modulators J. I. Pankove, Hybrid Structures P. G. LeComber, A. E. Owen, W. E. Spear, J. Hajto, and W. K. Choi, Electronic Switching in Amorphous
Silicon Junction Devices
Volume 22
Lightwave Communications Technology
Part A K. Nakajima, The Liquid-Phase Epitaxial Growth of InGaAsP W. T. Tsang, Molecular Beam Epitaxy for III-V Compound Semiconductors G. B. Stringfellow, Organometallic Vapor-Phase Epitaxial Growth of III-V Semiconductors G. Beuchet, Halide and Chloride Transport Vapor-Phase Deposition of InGaAsP and GaAs M. Razeghi, Low-Pressure, Metallo-Organic Chemical Vapor Deposition of GaxInl-xAsP~ _y Alloys P. M. Petroff, Defects in III-V Compound Semiconductors
Part B J. P. van der Ziel, Mode Locking of Semiconductor Lasers K. F. Lau and A. Yariv, High-Frequency Current Modulation of Semiconductor Injection Lasers C. H. Henry, Special Properties of Semi Conductor Lasers F. Suematsu, K. Kishino, S. Arai, and F. Koyama, Dynamic Single-Mode Semiconductor Lasers with a
Distributed Reflector W. T. Tsang, The Cleaved-Coupled-Cavity (C 3) Laser
Part C R. J. Nelson and N. K. Durra, Review of InGaAsP InP Laser Structures and Comparison of
Their Performance N. Chinone and M. Nakamura, Mode-Stabilized Semiconductor Lasers for 0.7-0.8- and 1.1 - 1.6-pum Regions F. Horikoshi, Semiconductor Lasers with Wavelengths Exceeding 2 p~m B. A. Dean and M. Dixon, The Functional Reliability of Semiconductor Lasers as Optical Transmitters R. H. Saul, T. P. Lee, and C. A. Burus, Light-Emitting Device Design C. L. Zipfel, Light-Emitting Diode-Reliability T. P. Lee and T. Li, LED-Based Multimode Lightwave Systems K. Ogawa, Semiconductor Noise-Mode Partition Noise
244
Contents o f Volumes in This Series
Part D F. Capasso, The Physics of Avalanche Photodiodes T. P. Pearsall and M. A. Pollack, Compound Semiconductor Photodiodes T. Kaneda, Silicon and Germanium Avalanche Photodiodes S. R. Forrest, Sensitivity of Avalanche Photodetector Receivers for High-Bit-Rate Long-Wavelength
Optical Communication Systems J. C. Campbell, Phototransistors for Lightwave Communications
Part E s. Wang, Principles and Characteristics of Integrable Active and Passive Optical Devices S. Margalit and A. Yariv, Integrated Electronic and Photonic Devices T. Mukai, F. Yamamoto, and T. Kimura, Optical Amplification by Semiconductor Lasers
V o l u m e 23
Pulsed Laser Processing of Semiconductors
R. F. Wood, C W. White and R. T. Young, Laser Processing of Semiconductors: An Overview C. W. White, Segregation, Solute Trapping and Supersaturated Alloys G. E. Jellison, Jr., Optical and Electrical Properties of Pulsed Laser-Annealed Silicon R. F. Wood and G. E. Jellison, Jr., Melting Model of Pulsed Laser Processing R. F. Wood and F. W. Young, Jr., Nonequilibrium Solidification Following Pulsed Laser Melting D. H. Lowndes and G. E. Jellison, Jr., Time-Resolved Measurement During Pulsed Laser Irradiation of Silicon D. M. Zebner, Surface Studies of Pulsed Laser Irradiated Semiconductors D. H. Lowndes, Pulsed Beam Processing of Gallium Arsenide R. B. James, Pulsed CO2 Laser Annealing of Semiconductors R. T. Young and R. F. Wood, Applications of Pulsed Laser Processing
V o l u m e 24
Applications of Multiquantum Wells, Selective Doping, and Superlattices
c. Weisbuch, Fundamental Properties of III-V Semiconductor Two-Dimensional Quantized Structures: The
Basis for Optical and Electronic Device Applications H. Morkof and H. Unlu, Factors Affecting the Performance of (AI,Ga)As/GaAs and (A1,Ga)As/InGaAs
Modulation-Doped Field-Effect Transistors: Microwave and Digital Applications N. T. Linh, Two-Dimensional Electron Gas FETs: Microwave Applications M. Abe et al., Ultra-High-Speed HEMT Integrated Circuits D. S. Chemla, D. A. B. Miller and P. W. Smith, Nonlinear Optical Properties of Multiple Quantum Well
Structures for Optical Signal Processing F. Capasso, Graded-Gap and Superlattice Devices by Band-Gap Engineering W. T. Tsang, Quantum Confinement Heterostructure Semiconductor Lasers G. C. Osbourn et al., Principles and Applications of Semiconductor Strained-Layer Superlattices
Volume 25
Diluted Magnetic Semiconductors
w. Giriat and J. K. Furdyna, Crystal Structure, Composition, and Materials Preparation of Diluted Magnetic
Semiconductors W. M. Becker, Band Structure and Optical Properties of Wide-Gap A~XxMnxBxvAlloys at Zero Magnetic Field
Contents o f Volumes in This Series
245
S. Oseroff and P. H. Keesom, Magnetic Properties: Macroscopic Studies T. Giebultowicz and T. M. Holden, Neutron Scattering Studies of the Magnetic Structure and Dynamics of
Diluted Magnetic Semiconductors J. Kossut, Band Structure and Quantum Transport Phenomena in Narrow-Gap Diluted Magnetic
Semiconductors C. Riquaux, Magnetooptical Properties of Large-Gap Diluted Magnetic Semiconductors J. A. Gaj, Magnetooptical Properties of Large-Gap Diluted Magnetic Semiconductors J. Mycielski, Shallow Acceptors in Diluted Magnetic Semiconductors: Splitting, Boil-off, Giant Negative
Magnetoresistance A. K. Ramadas and R. Rodriquez, Raman Scattering in Diluted Magnetic Semiconductors P. A. Wolff, Theory of Bound Magnetic Polarons in Semimagnetic Semiconductors
V o l u m e 26
III-V Compound Semiconductors and Semiconductor Properties of Superionic Materials
z. Yuanxi, III-V Compounds H. V. Winston, A. T. Hunter, H. Kimura, and R. E. Lee, InAs-Alloyed GaAs Substrates for Direct Implantation P. K. Bhattacharya and S. Dhar, Deep Levels in III-V Compound Semiconductors Grown by MBE Y. Ya. Gurevich and A. K. Ivanov-Shits, Semiconductor Properties of Supersonic Materials
Volume 27
High Conducting Quasi-One-Dimensional Organic Crystals
E. M. Conwell, Introduction to Highly Conducting Quasi-One-Dimensional Organic Crystals I. A. Howard, A Reference Guide to the Conducting Quasi-One-Dimensional Organic Molecular Crystals J. P. Pouquet, Structural Instabilities E. M. Conwell, Transport Properties C. S. Jacobsen, Optical Properties J. C. Scott, Magnetic Properties L. Zuppiroli, Irradiation Effects: Perfect Crystals and Real Crystals
V o l u m e 28
Measurement of High-Speed Signals in Solid State Devices
J. Frey and D. Ioannou, Materials and Devices for High-Speed and Optoelectronic Applications H. Schumacher and E. Strid, Electronic Wafer Probing Techniques D. H. Auston, Picosecond Photoconductivity: High-Speed Measurements of Devices and Materials J. A. Valdmanis, Electro-Optic Measurement Techniques for Picosecond Materials, Devices and Integrated
Circuits J. M. Wiesenfeld and R. K. Jain, Direct Optical Probing of Integrated Circuits and High-Speed Devices G. Plows, Electron-Beam Probing A. M. Weiner and R. B. Marcus, Photoemissive Probing
Volume 29
Very High Speed Integrated Circuits: Gallium Arsenide LSI
M. Kuzuhara and T. Nazaki, Active Layer Formation by Ion Implantation H. Hasimoto, Focused Ion Beam Implantation Technology T. Nozaki and A. Higashisaka, Device Fabrication Process Technology
246
Contents o f Volumes in This Series
M. Ino and T. Takada, GaAs LSI Circuit Design M. Hirayama, M. Ohmori, and K. Yamasaki, GaAs LSI Fabrication and Performance
Volume 30
Very High Speed Integrated Circuits: Heterostructure
H. Watanabe, T. Mizutani, and A. Usui, Fundamentals of Epitaxial Growth and Atomic Layer Epitaxy S. Hiyamizu, Characteristics of Two-Dimensional Electron Gas in III-V Compound Heterostructures Grown by
MBE T. Nakanisi, Metalorganic Vapor Phase Epitaxy for High-Quality Active Layers T. Nimura, High Electron Mobility Transistor and LSI Applications T. Sugeta and T. Ishibashi, Hetero-Bipolar Transistor and LSI Application H. Matsuedo, T. Tanaka, and M. Nakamura, Optoelectronic Integrated Circuits
Volume 31
Indium Phosphide: Crystal Growth and Characterization
J. P. Farges, Growth of Discoloration-Free InP M. J. McCollum and G. E. Stillman, High Purity InP Grown by Hydride Vapor Phase Epitaxy I. Inada and T. Fukuda, Direct Synthesis and Growth of Indium Phosphide by the Liquid Phosphorous
Encapsulated Czochralski Method O. Oda, K. Katagiri, K. Shinohara, S. Katsura, Y. Takahashi, K. Kainosho, K. Kohiro, and R. Hirano, InP
Crystal Growth, Substrate Preparation and Evaluation K. Tada, M. Tatsumi, M. Morioka, T. Araki, and T. Kawase, InP Substrates: Production and Quality Control M. Razeghi, LP-MOCVD Growth, Characterization, and Application of InP Material T. A. Kennedy and P. J. Lin-Chung, Stoichiometric Defects in InP
Volume 32
Strained-Layer Superlattices: Physics
T. P. Pearsall, Strained-Layer Superlattices F. H. Pollack, Effects of Homogeneous Strain on the Electronic and Vibrational Levels in Semiconductors J. F. Marzin, J. M. Ger~rd, P. Voisin, and J. A. Bruin, Optical Studies of Strained III-V Heterolayers R. People and S. A. Jackson, Structurally Induced States from Strain and Confinement M. Jaros, Microscopic Phenomena in Ordered Superlattices
Volume 33
Strained-Layer Superlattices: Material Science and Technology
R. Hull and J. C. Bean, Principles and Concepts of Strained-Layer Epitaxy W. J. Shaft, P. J. Tasker, M. C. Foisy, and L. F. Eastman, Device Applications of Strained-Layer Epitaxy S. T. Picraux, B. L. Doyle, and J. Y. Tsao, Structure and Characterization of Strained-Layer Superlattices E. Kasper and F. Schaffer, Group IV Compounds D. L. Martin, Molecular Beam Epitaxy of IV-VI Compounds Heterojunction R. L. Gunshot, L. A. Kolodziejski, A. V. Nurmikko, and N. Otsuka, Molecular Beam Epitaxy of I-VI
Semiconductor Microstructures
Contents o f Volumes in This Series
Volume 34
247
Hydrogen in Semiconductors
J. I. Pankove and N. M. Johnson, Introduction to Hydrogen in Semiconductors C. H. Seager, Hydrogenation Methods J. I. Pankove, Hydrogenation of Defects in Crystalline Silicon J. W. Corbett, P. De~k, U. V. Desnica, and S. J. Pearton, Hydrogen Passivation of Damage Centers in
Semiconductors S. J. Pearton, Neutralization of Deep Levels in Silicon J. I. Pankove, Neutralization of Shallow Acceptors in Silicon N. M. Johnson, Neutralization of Donor Dopants and Formation of Hydrogen-Induced Defects in n-Type Silicon M. Stavola and S. J. Pearton, Vibrational Spectroscopy of Hydrogen-Related Defects in Silicon A. D. Marwick, Hydrogen in Semiconductors: Ion Beam Techniques C. Herring and N. M. Johnson, Hydrogen Migration and Solubility in Silicon E. E. Haller, Hydrogen-Related Phenomena in Crystalline Germanium J. Kakalios, Hydrogen Diffusion in Amorphous Silicon J. Chevalier, B. Clerjaud, and B. Pajot, Neutralization of Defects and Dopants in III-V Semiconductors G. G. DeLeo and W. B. Fowler, Computational Studies of Hydrogen-Containing Complexes in Semiconductors R. F. Kiefl and T. L. Estle, Muonium in Semiconductors C. G. Van de Walle, Theory of Isolated Interstitial Hydrogen and Muonium in Crystalline Semiconductors
V o l u m e 35
Nanostructured Systems
M. Reed, Introduction H. van Houten, C. W. J. Beenakker, and B. J. Wees, Quantum Point Contacts G. Timp, When Does a Wire Become an Electron Waveguide? M. Bizttiker, The Quantum Hall Effects in Open Conductors W. Hansen, J. P. Kotthaus, and U. Merkt, Electrons in Laterally Periodic Nanostructures
Volume 36
The Spectroscopy of Semiconductors
D. Heiman, Spectroscopy of Semiconductors at Low Temperatures and High Magnetic Fields A. V. Nurmikko, Transient Spectroscopy by Ultrashort Laser Pulse Techniques A. K. Ramclas and S. Rodriguez, Piezospectroscopy of Semiconductors O. J. Glembocki and B. V. Shanabrook, Photoreflectance Spectroscopy of Microstructures D. G. Seiler, C. L. Littler, and M. H. Wiler, One- and Two-Photon Magneto-Optical Spectroscopy of InSb and
Hgl-xCdxTe
Volume 37
The Mechanical Properties of Semiconductors
A.-B. Chen, A. Sher, and W. T. Yost, Elastic Constants and Related Properties of Semiconductor Compounds and
Their Alloys D. R. Clarke, Fracture of Silicon and Other Semiconductors H. Siethoff, The Plasticity of Elemental and Compound Semiconductors S. Guruswamy, K. T. Faber, and J. P. Hirth, Mechanical Behavior of Compound Semiconductors S. Mahajan, Deformation Behavior of Compound Semiconductors J. P. Hirth, Injection of Dislocations into Strained Multilayer Structures
248
Contents o f Volumes in This Series
D. Kendall, C. B. Fleddermann, and K. J. Malloy, Critical Technologies for the Micromatching of Silicon I. Matsuba and K. Mokuya, Processing and Semiconductor Thermoelastic Behavior
Volume 38
Imperfections in III/V Materials
U. Scherz and M. Scheffier, Density-Functional Theory of sp-Bonded Defects in III/V Semiconductors M. Kaminska and E. R. Weber, El2 Defect in GaAs D. C. Look, Defects Relevant for Compensation in Semi-Insulating GaAs R. C. Newman, Local Vibrational Mode Spectroscopy of Defects in III/V Compounds A. M. Hennel, Transition Metals in III/V Compounds K. J. Malloy and K. Khachaturyan, DX and Related Defects in Semiconductors V. Swaminathan and A. S. Jordan, Dislocations in III/V Compounds K. W. Nauka, Deep Level Defects in the Epitaxial III/V Materials
V o l u m e 39
Minority Carriers in III-V Semiconductors: Physics and Applications
N. K. Dutta, Radiative Transition in GaAs and Other III-V Compounds R. K. Ahrenkiel, Minority-Carrier Lifetime in III-V Semiconductors T. Furuta, High Field Minority Electron Transport in p-GaAs M. S. Lundstrom, Minority-Carrier Transport in III-V Semiconductors R. A. Abram, Effects of Heavy Doping and High Excitation on the Band Structure of GaAs D. Yevick and W. Bardyszewski, An Introduction to Non-Equilibrium Many-Body Analyses of Optical Processes
in III-V Semiconductors
Volume 40
Epitaxial Microstructures
E. F. Schubert, Delta-Doping of Semiconductors: Electronic, Optical and Structural Properties of Materials and
Devices A. Gossard, M. Sundaram, and P. Hopkins, Wide Graded Potential Wells P. Petroff, Direct Growth of Nanometer-Size Quantum Wire Superlattices E. Kapon, Lateral Patterning of Quantum Well Heterostructures by Growth of Nonplanar Substrates H. Temkin, D. Gershoni, and M. Panish, Optical Properties of Gal-x InxAs/InP Quantum Wells
Volume
41
High Speed Heterostructure Devices
F. Capasso, F. Be#ram, S. Sen, A. Pahlevi, and A. Y. Cho, Quantum Electron Devices: Physics
and Applications P. Solomon, D. J. Frank, S. L. Wright and F. Canora, GaAs-Gate Semiconductor-Insulator- Semiconductor FET M. H. Hashemi and U. K. Mishra, Unipolar InP-Based Transistors R. Kiehl, Complementary Heterostructure FET Integrated Circuits T. Ishibashi, GaAs-Based and InP-Based Heterostructure Bipolar-Transistors H. C. Liu and T. C. L. G. Sollner, High-Frequency-Tunneling Devices H. Ohnishi, T. More, M. Takatsu, K. Imamura, and N. Yokoyama, Resonant-Tunneling Hot-Electron Transistors
and Circuits
Contents o f Volumes in This Series
V o l u m e 42
Oxygen
in Silicon
F. Shimura, Introduction to Oxygen in Silicon W. Lin, The Incorporation of Oxygen into Silicon Crystals T. J. Schaffner and D. K. Schroder, Characterization Techniques for Oxygen in Silicon W. M. Bullis, Oxygen Concentration Measurement S. M. Hu, Intrinsic Point Defects in Silicon B. Pajot, Some Atomic Configuration of Oxygen J. Michel and L. C. Kimerling, Electrical Properties of Oxygen in Silicon R. C. Newman and R. Jones, Diffusion of Oxygen in Silicon T. Y. Tan and W. J. Taylor, Mechanisms of Oxygen Precipitation: Some Quantitative Aspects M. Schrems, Simulation of Oxygen Precipitation K. Simino and L Yonenaga, Oxygen Effect on Mechanical Properties W. Bergholz, Grown-in and Process-Induced Effects F. Shimura, Intrinsic/Internal Gettering H. Tsuya, Oxygen Effect on Electronic Device Performance
V o l u m e 43
Semiconductors for Room Temperature Nuclear Detector Applications
R. B. James and T. E. Schlesinger, Introduction and Overview L. S. Darken and C. E. Cox, High-Purity Germanium Detectors A. Burger, D. Nason, L. Van den Berg, and M. Schieber, Growth of Mercuric Iodide X. J. Bao, T. E. Schlesinger, and R. B. James, Electrical Properties of Mercuric Iodide X. J. Bao, R. B. James, and T. E. Schlesinger, Optical Properties of Red Mercuric Iodide M. Hage-Ali and P. Siffert, Growth Methods of CdTe Nuclear Detector Materials M. Hage-Ali and P. Siffert, Characterization of CdTe Nuclear Detector Materials M. Hage-Ali and P. Siffert, CdTe Nuclear Detectors and Applications R. B. James, T. E. Schlesinger, J. Lund, and M. Schieber, Cdl-x Znx Te Spectrometers for Gamma
and X-Ray Applications D. S. McGregor, J. E. Kammeraad, Gallium Arsenide Radiation Detectors and Spectrometers J. C. Lund, F. Olschner, and A. Burger, Lead Iodide M. R. Squillante and K. S. Shah, Other Materials: Status and Prospects V. M. Gerrish, Characterization and Quantification of Detector Performance J. S. Iwanczyk and B. E. Patt, Electronics for X-ray and Gamma Ray Spectrometers M. Schieber, R. B. James and T. E. Schlesinger, Summary and Remaining Issues for Room Temperature
Radiation Spectrometers
Volume 44
I I - I V Blue/Green Light Emitters" Device Physics and Epitaxial Growth
J. Han and R. L. Gunshor, MBE Growth and Electrical Properties of Wide Bandgap ZnSe-based II-VI
Semiconductors S. Fujita and S. Fujita, Growth and Characterization of ZnSe-based II-VI Semiconductors by MOVPE E. Ho and L. A. Kolodziejski, Gaseous Source UHV Epitaxy Technologies for Wide Bandgap II-VI
Semiconductors
249
250
Contents o f Volumes in This Series
C. G. Van de Walle, Doping of Wide-Band-Gap II-VI Compounds - Theory R. Cingolani, Optical Properties of Excitons in ZnSe-Based Quantum Well Heterostructures A. Ishibashi and A. V. Nurmikko, II-VI Diode Lasers: A Current View of Device Performance and Issues S. Guha and J. Petruzello, Defects and Degradation in Wide-Gap II-VI-based Structure and Light Emitting
Devices
Volume 45
Effect of Disorder and Defects in Ion-Implanted Semiconductors: Electrical and Physiochemical Characterization
H. Ryssel, Ion Implantation into Semiconductors: Historical Perspectives You-Nian Wang and Teng-Cai Ma, Electronic Stopping Power for Energetic Ions in Solids S. T. Nakagawa, Solid Effect on the Electronic Stopping of Crystalline Target and Application to Range
Estimation G. Miller, S. Kalbitzer, and G. N. Greaves, Ion Beams in Amorphous Semiconductor Research J. Boussey-Said, Sheet and Spreading Resistance Analysis of Ion Implanted and Annealed Semiconductors M. L. Polignano and G. Queirolo, Studies of the Stripping Hall Effect in Ion-Implanted Silicon J. Stoemenos, Transmission Electron Microscopy Analyses R. Nipoti and M. Servidori, Rutherford Backscattering Studies of Ion Implanted Semiconductors P. Zaumseil, X-ray Diffraction Techniques
Volume 46
Effect of Disorder and Defects in Ion-Implanted Semiconductors: Optical and Photothermal Characterization
M. Fried, T. Lohner, and J. Gyulai, Ellipsometric Analysis A. Seas and C. Christofides, Transmission and Reflection Spectroscopy on Ion Implanted Semiconductors A. Othonos and C. Christofides, Photoluminescence and Raman Scattering of Ion Implanted Semiconductors.
Influence of Annealing C. Christofides, Photomodulated Thermoreflectance Investigation of Implanted Wafers. Annealing Kinetics
of Defects U. Zammit, Photothermal Deflection Spectroscopy Characterization of Ion-Implanted and Annealed Silicon
Films A. Mandelis, A. Budiman, and M. Vargas, Photothermal Deep-Level Transient Spectroscopy of Impurities
and Defects in Semiconductors R. Kalish and S. Charbonneau, Ion Implantation into Quantum-Well Structures A. M. Myasnikov and N. N. Gerasimenko, Ion Implantation and Thermal Annealing of III-V Compound
Semiconducting Systems: Some Problems of III-V Narrow Gap Semiconductors
Volume 47
Uncooled Infrared Imaging Arrays and Systems
R. G. Buser and M. P. Tompsett, Historical Overview P. W. Kruse, Principles of Uncooled Infrared Focal Plane Arrays R. A. Wood, Monolithic Silicon Microbolometer Arrays C. M. Hanson, Hybrid Pyroelectric-Ferroelectric Bolometer Arrays D. L. Polla and J. R. Choi, Monolithic Pyroelectric Bolometer Arrays
Contents o f Volumes in This Series
251
N. Teranishi, Thermoelectric Uncooled Infrared Focal Plane Arrays M. F. Tompsett, Pyroelectric Vidicon T. W. Kenny, Tunneling Infrared Sensors J. R. Vig, R. L. Filler, and Y. Kim, Application of Quartz Microresonators to Uncooled Infrared Imaging Arrays P. W. Kruse, Application of Uncooled Monolithic Thermoelectric Linear Arrays to Imaging Radiometers
Volume 48
High Brightness Light Emitting Diodes
G. B. Stringfellow, Materials Issues in High-Brightness Light-Emitting Diodes M.G. Craford, Overview of Device Issues in High-Brightness Light-Emitting Diodes F. M. Steranka, A1GaAs Red Light Emitting Diodes C. H. Chen, S. A. Stockman, M. J. Peanasky, and C. P. Kuo, OMVPE Growth of A1GaInP for High Efficiency
Visible Light-Emitting Diodes F. A. Kish and R. M. Fletcher, AIGaInP Light-Emitting Diodes M. W. Hodapp, Applications for High Brightness Light-Emitting Diodes I. Akasaki and H. Amano, Organometallic Vapor Epitaxy of GaN for High Brightness Blue Light Emitting
Diodes S. Nakamura, Group III-V Nitride Based Ultraviolet-Blue-Green-Yellow Light-Emitting Diodes and Laser
Diodes
Volume 49
Light Emission in Silicon: from Physics to Devices
D. J. Lockwood, Light Emission in Silicon G. Abstreiter, Band Gaps and Light Emission in Si/SiGe Atomic Layer Structures T. G. Brown and D. G. Hall, Radiative Isoelectronic Impurities in Silicon and Silicon-Germanium Alloys
and Superlattices J. Michel, L. V. C. Assali, M. T. Morse, and L. C. Kimerling, Erbium in Silicon Y. Kanemitsu, Silicon and Germanium Nanoparticles P. M. Fauchet, Porous Silicon: Photoluminescence and Electroluminescent Devices C. Delerue, G. Allan, and M. Lannoo, Theory of Radiative and Nonradiative Processes in Silicon
Nanocrystallites L. Brus, Silicon Polymers and Nanocrystals
Volume 50
Gallium Nitride (GaN)
J. I. Pankove and T. D. Moustakas, Introduction S. P. DenBaars and S. Keller, Metalorganic Chemical Vapor Deposition (MOCVD) of Group III Nitrides W. A. Bryden and T. J. Kistenmacher, Growth of Group III-A Nitrides by Reactive Sputtering N. Newman, Thermochemistry of III-N Semiconductors S. J. Pearton and R. J. Shul, Etching of III Nitrides S. M. Bedair, Indium-based Nitride Compounds A. Trampert, O. Brandt, and K. H. Ploog, Crystal Structure of Group III Nitrides H. Morko9, F. Hamdani, and A. Salvador, Electronic and Optical Properties of III-V Nitride based Quantum
Wells and Superlattices K. Doverspike and J. I. Pankove, Doping in the III-Nitrides T. Suski and P. Perlin, High Pressure Studies of Defects and Impurities in Gallium Nitride
252
Contents o f Volumes in This Series
B. Monemar, Optical Properties of GaN W. R. L. Lambrecht, Band Structure of the Group III Nitrides N. E. Christensen and P. Perlin, Phonons and Phase Transitions in GaN S. Nakamura, Applications of LEDs and LDs L Akasaki and H. Amano, Lasers J. A. Cooper, Jr., Nonvolatile Random Access Memories in Wide Bandgap Semiconductors
Identification of Defects in Semiconductors
Volume 51A
G. D. Watkins, EPR and ENDOR Studies of Defects in Semiconductors J.-M. Spaeth, Magneto-Optical and Electrical Detection of Paramagnetic Resonance in Semiconductors T. A. Kennedy and E. R. Glaser, Magnetic Resonance of Epitaxial Layers Detected by Photoluminescence K. H. Chow, B. Hitti, and R. F. Kiefl, ptSR on Muonium in Semiconductors and Its Relation to Hydrogen K. Saarinen, P. Hautofiirvi, and C. Corbel, Positron Annihilation Spectroscopy of Defects in Semiconductors R. Jones and P. R. Briddon, The Ab Initio Cluster Method and the Dynamics of Defects in Semiconductors
Identification Defects in Semiconductors
V o l u m e 51B
G. Davies, Optical Measurements of Point Defects P. M. Mooney, Defect Identification Using Capacitance Spectroscopy M. Stavola, Vibrational Spectroscopy of Light Element Impurities in Semiconductors P. Schwander, W. D. Rau, C. Kisielowski, M. Gribelyuk, and A. Ourmazd, Defect Processes in Semiconductors
Studied at the Atomic Level by Transmission Electron Microscopy N. D. Jager and E. R. Weber, Scanning Tunneling Microscopy of Defects in Semiconductors
V o l u m e 52
SiC Materials and Devices
K. Jiirrendahl and R. F. Davis, Materials Properties and Characterization of SiC V. A. Dmitiriev and M. G. Spencer, SiC Fabrication Technology: Growth and Doping V. Saxena and A. J. Steckl, Building Blocks for SiC Devices: Ohmic Contacts, Schottky Contacts, and p-n
Junctions M. S. Shur, SiC Transistors C. D. Brandt, R. C. Clarke, R. R. Siergiej, J. B. Casady, A. W. Morse, S. Sriram, and A. K. Agarwal, SiC for
Applications in High-Power Electronics R. J. Trew, SiC Microwave Devices J. Edmond, H. Kong, G. Negley, M. Leonard, K. Doverspike, W. Weeks, A. Suvorov, D. Waltz, and C. Carter, Jr.,
SiC-Based UV Photodiodes and Light-Emitting Diodes H. Morkog, Beyond Silicon Carbide! III-V Nitride-Based Heterostructures and Devices
V o l u m e 53 C u m u l a t i v e S u b j e c t s a n d A u t h o r I n d e x Including T a b l e s o f C o n t e n t s for Volumes 1 - 5 0
V o l u m e 54
High Pressure in Semiconductor Physics I
w. Paul, High Pressure in Semiconductor Physics: A Historical Overview N. E. Christensen, Electronic Structure Calculations for Semiconductors Under Pressure
Contents o f Volumes in This Series
253
R. J. Neimes and M. L McMahon, Structural Transitions in the Group IV, III-V and II-VI Semiconductors
Under Pressure A. R. Goni and K. Syassen, Optical Properties of Semiconductors Under Pressure P. Trautman, M. Baj, and J. M. Baranowski, Hydrostatic Pressure and Uniaxial Stress in Investigations of the
EL2 Defect in GaAs M. Li and P. Y. Yu, High-Pressure Study of DX Centers Using Capacitance Techniques T. Suski, Spatial Correlations of Impurity Charges in Doped Semiconductors N. Kuroda, Pressure Effects on the Electronic Properties of Diluted Magnetic Semiconductors
Volume 55
High Pressure in Semiconductor Physics II
D. K. Maude and J. C. Portal, Parallel Transport in Low-Dimensional Semiconductor Structures P. C. Klipstein, Tunneling Under Pressure: High-Pressure Studies of Vertical Transport in Semiconductor
Heterostructures E. Anastassakis and M. Cardona, Phonons, Strains, and Pressure in Semiconductors F. H. Pollak, Effects of External Uniaxial Stress on the Optical Properties of Semiconductors and
Semiconductor Microstructures A. R. Adams, M. Silver, and J. Allam, Semiconductor Optoelectronic Devices S. Porowski and L Grzegory, The Application of High Nitrogen Pressure in the Physics and Technology of
III-N Compounds M. Yousuf, Diamond Anvil Cells in High Pressure Studies of Semiconductors
Volume 56
Germanium Silicon: Physics and Materials
J. c. Bean, Growth Techniques and Procedures D. E. Savage, F. Liu, V. Zielasek, and M. G. Lagally, Fundamental Crystal Growth Mechanisms R. Hull, Misfit Strain Accommodation in SiGe Heterostructures M. J. Shaw and M. Jaros, Fundamental Physics of Strained Layer GeSi: Quo Vadis? F. Cerdeira, Optical Properties S. A. Ringel and P. N. Grillot, Electronic Properties and Deep Levels in Germanium-Silicon J. C. Campbell, Optoelectronics in Silicon and Germanium Silicon K. Eberl, K. Brunner, and O. G. Schmidt, Sil-yCy and Sil-x-yGezCy Alloy Layers
Volume
57
Gallium Nitride (GaN) I I
R. J. Molnar, Hydride Vapor Phase Epitaxial Growth of III-V Nitrides T. D. Moustakas, Growth of III-V Nitrides by Molecular Beam Epitaxy Z. Liliental-Weber, Defects in Bulk GaN and Homoepitaxial Layers C. G. Van de Walle and N. M. Johnson, Hydrogen in III-V Nitrides W. GStz and N. M. Johnson, Characterization of Dopants and Deep Level Defects in Gallium Nitride B. Gil, Stress Effects on Optical Properties C. Kisielowski, Strain in GaN Thin Films and Heterostructures J. A. Miragliotta and D. K. Wickenden, Nonlinear Optical Properties of Gallium Nitride B. K. Meyer, Magnetic Resonance Investigations on Group III-Nitrides M. S. Shur and M. Asif Khan, GaN and AIGaN Ultraviolet Detectors C. H. Qiu, J. I. Pankove and C. Rossington, I I - V Nitride-Based X-ray Detectors
254
Contents o f Volumes in This Series
V o l u m e 58
Nonlinear Optics in Semiconductors I
A. Kost, Resonant Optical Nonlinearities in Semiconductors E. Garmire, Optical Nonlinearities in Semiconductors Enhanced by Carrier Transport D. S. Chemla, Ultrafast Transient Nonlinear Optical Processes in Semiconductors M. Sheik-Bahae and E. W. Van Stryland, Optical Nonlinearities in the Transparency Region of Bulk
Semiconductors J. E. Millerd, M. Ziari, and A. Partovi, Photorefractivity in Semiconductors
Volume 59
Nonlinear Optics in Semiconductors II
J. B. Khurgin, Second Order Nonlinearities and Optical Rectification K. L. Hall, E. R. Thoen, and E. P. lppen, Nonlinearities in Active Media E. Hanamura, Optical Responses of Quantum Wires/Dots and Microcavities U. Keller, Semiconductor Nonlinearities for Solid-State Laser Modelocking and Q-Switching A. Miller, Transient Grating Studies of Carrier Diffusion and Mobility in Semiconductors
Volume 60
Self-Assembled InGaAs/GaAs Quantum Dots
Mitsuru Sugawara, Theoretical Bases of the Optical Properties of Semiconductor Quantum Nano-Structures Yoshiaki Nakata, Yoshihiro Sugiyama, and Mitsuru Sugawara, Molecular Beam Epitaxial Growth of
Self-Assembled InAs/GaAs Quantum Dots Kohki Mukai, Mitsuru Sugawara, Mitsuru Egawa, and Nobuyuki Ohtsuka, Metalorganic Vapor Phase Epitaxial
Growth of Self-Assembled InGaAs/GaAs Quantum Dots Emitting at 1.3 I~m Kohki Mukai and Mitsuru Sugawara, Optical Characterization of Quantum Dots Kohki Mukai and Mitsuru Sugawara, The Photon Bottleneck Effect in Quantum Dots Hajime Shoji, Self-Assembled Quantum Dot Lasers Hiroshi Ishikawa, Applications of Quantum Dot to Optical Devices Mitsuru Sugawara, Kohki Mukai, Hiroshi Ishikawa, Koji Otsubo, and Yoshiaki Nakata, The Latest News
V o l u m e 61
Hydrogen in Semiconductors I I
Norbert H. Nickel, Introduction to Hydrogen in Semiconductors II Noble M. Johnson and Chris G. Van de Walle, Isolated Monatomic Hydrogen in Silicon Yurij V. Gorelkinskii, Electron Paramagnetic Resonance Studies of Hydrogen and Hydrogen-Related Defects in
Crystalline Silicon Norbert H. Nickel, Hydrogen in Polycrystalline Silicon Wolfhard Beyer, Hydrogen Phenomena in Hydrogenated Amorphous Silicon Chris G. Van de Walle, Hydrogen Interactions with Polycrystalline and Amorphous Silicon-Theory Karen M. McManus Rutledge, Hydrogen in Polycrystalline CVD Diamond Roger L. Lichti, Dynamics of Muonium Diffusion, Site Changes and Charge-State Transitions Matthew D. McCluskey and Eugene E. Haller, Hydrogen in III-V and II-VI Semiconductors S. J. Pearton and J. W. Lee, The Properties of Hydrogen in GaN and Related Alloys J6rg Neugebauer and Chris G. Van de Walle, Theory of Hydrogen in GaN
Contents o f Volumes in This Series
255
Intersubband Transitions in Quantum W e l l s : Physics and Device Applications I
V o l u m e 62
Manfred Helm, The Basic Physics of Intersubband Transitions Jerome Faist, Carlo Sirtori, Federico Capasso, Loren N. Pfeiffer, Ken W. West, Deborah L. Sivco, and Alfred Y. Cho, Quantum Interference Effects in Intersubband Transitions H. C. Liu, Quantum Well Infrared Photodetector Physics and Novel Devices S. D. Gunapala and S. V. Bandara, Quantum Well Infrared Photodetector (QWIP) Focal Plane Arrays
Volume 63
Chemical Mechanical Polishing in Si Processing
Frank B. Kaufman, Introduction Thomas Bibby and Karey Holland, Equipment John P. Bare, Facilitization Duane S. Boning and Okumu Ouma, Modeling and Simulation Shin Hwa Li, Bruce Tredinnick, and Mel Hoffman, Consumables I: Slurry Lee M. Cook, CMP Consumables II: Pad Frangois Tardif Post-CMP Clean Shin Hwa Li, Tara Chhatpar, and Frederic Robert, CMP Metrology Shin Hwa Li, Visun Bucha, and Kyle Wooldridge, Applications and CMP-Related Process Problems
V o l u m e 64
Electroluminescence I
M. G. Craford, S. A. Stockman, M. J. Peansky, and F. A. Kish, Visible Light-Emitting Diodes H. Chui, N. F. Gardner, P. N. Grillot, J. W. Huang, M. R. Krames, and S. A. Maranowski, High-Efficiency
AIGaInP Light-Emitting Diodes R. S. Kern, W. Gbtz, C. H. Chen, H. Liu, R. M. Fletcher, and C. P. Kuo, High-Brightness Nitride-Based
Visible-Light-Emitting Diodes Yoshiharu Sato, Organic LED System Considerations V. Bulovi(, P. E. Burrows, and S. R. Forrest, Molecular Organic Light-Emitting Devices
V o l u m e 65
Electroluminescence II
v. Bulovib and S. R. Forrest, Polymeric and Molecular Organic Light Emitting Devices: A Comparison Regina Mueller-Mach and Gerd 0. Mueller, Thin Film Electroluminescence Markku Leskel& Wei-Min Li, and Mikko Ritala, Materials in Thin Film Electroluminescent Devices Kristiaan Neyts, Microcavities for Electroluminescent Devices
Volume 66
Intersubband Transitions in Quantum Wells: Physics and Device Applications II
Jerome Faist, Federico Capasso, Carlo Sirtori, Deborah L. Sivco, and Alfred Y. Cho, Quantum Cascade Lasers Federico Capasso, Carlo Sirtori, D. L. Sivco, and A. Y. Cho, Nonlinear Optics in Coupled-Quantum- Well
Quasi-Molecules Karl Unterrainer, Photon-Assisted Tunneling in Semiconductor Quantum Structures P. Haring Bolivar, T. Dekorsy, and H. Kurz, Optically Excited Bloch Oscillations-Fundamentals and
Application Perspectives
256
Contents o f Volumes in This Series
Volume 67
Ultrafast Physical Processes in Semiconductors
Alfred Leitenstorfer and Alfred Laubereau, Ultrafast Electron-Phonon Interactions in Semiconductors:
Quantum Kinetic Memory Effects Christoph Lienau and Thomas Elsaesser, Spatially and Temporally Resolved Near-Field Scanning Optical
Microscopy Studies of Semiconductor Quantum Wires K. T. Tsen, Ultrafast Dynamics in Wide Bandgap Wurtzite GaN J. Paul Callan, Albert M.-T. Kim, Christopher A. D. Roeser, and Eriz Mazur, Ultrafast Dynamics and Phase
Changes in Highly Excited GaAs Hartmut Haug, Quantum Kinetics for Femtosecond Spectroscopy in Semiconductors T. Meier and S. W. Koch, Coulomb Correlation Signatures in the Excitonic Optical Nonlinearities of
Semiconductors Roland E. Allen, Traian Dumitricd, and Ben Torralva, Electronic and Structural Response of Materials to Fast,
Intense Laser Pulses E. Gornik and R. Kersting, Coherent THz Emission in Semiconductors
Volume 68
Isotope Effects in Solid State Physics
Vladimir G. Plekhanov, Elastic Properties; Thermal Properties; Vibrational Properties; Raman Spectra of
Isotopically Mixed Crystals; Excitons in LiH Crystals; Exciton-Phonon Interaction; Isotopic Effect in the Emission Spectrum of Polaritons; Isotopic Disordering of Crystal Lattices; Future Developments and Applications; Conclusions
V o l u m e 69
Recent Trends in Thermoelectric Materials Research I
H. Julian Goldsmid, Introduction Terry M. Tritt and Valerie M. Browning, Overview of Measurement and Characterization Techniques for
Thermoelectric Materials Mercouri G. Kanatzidis, The Role of Solid-State Chemistry in the Discovery of New Thermoelectric Materials B. Lenoir, H. Scherrer, and T. Caillat, An Overview of Recent Developments for BiSb Alloys Citrad Uher, Skutterudities: Prospective Novel Thermoelectrics George S. Nolas, Glen A. Slack, and Sandra B. Schujman, Semiconductor Clathrates: A Phonon Glass Electron
Crystal Material with Potential for Thermoelectric Applications
Volume 70
Recent Trends in Thermoelectric Materials Research II
Brian C. Sales, David G. Mandrus, and Bryan C. Chakoumakos, Use of Atomic Displacement Parameters in
Thermoelectric Materials Research S. Joseph Poon, Electronic and Thermoelectric Properties of Half-Heusler Alloys Terry M. Tritt, A. L. Pope, and J. W. Kolis, Overview of the Thermoelectric Properties of Quasicrystalline
Materials and Their Potential for Thermoelectric Applications Alexander C. Ehrlich and Stuart A. Wolf, Military Applications of Enhanced Thermoelectrics David J. Singh, Theoretical and Computational Approaches for Identifying and Optimizing Novel
Thermoelectric Materials Terry M. Tritt and R. T. Littleton, IV, Thermoelectric Properties of the Transition Metal Pentatellurides:
Potential Low-Temperature Thermoelectric Materials
Contents o f Volumes in This Series
257
Franz Freibert, Timothy W. Darling, Albert Miglori, and Stuart A. Trugman, Thermomagnetic Effects and
Measurements M. Bartkowiak and G. D. Mahan, Heat and Electricity Transport Through Interfaces
Volume 71
Recent Trends in Thermoelectric Materials Research III
M. S. Dresselhaus, Y.-M. Lin, T. Koga, S. B. Cronin, O. Rabin, M. R. Black, and G. Dresselhaus, Quantum Wells
and Quantum Wires for Potential Thermoelectric Applications D. A. Broido and T. L. Reinecke, Thermoelectric Transport in Quantum Well and Quantum
Wire Superlattices G. D. Mahan, Thermionic Refrigeration Rama Venkatasubramanian, Phonon Blocking Electron Transmitting Superlattice Structures as Advanced Thin
Film Thermoelectric Materials G. Chen, Phonon Transport in Low-Dimensional Structures
V o l u m e 72
Silicon Epitaxy
s. Acerboni, ST Microelectronics, CFM-AGI Department, Agrate Brianza, Italy V.-M. Airaksinen, Okmetic Oyj R&D Department, Vantaa, Finland G. Beretta, ST Microelectronics, DSG Epitaxy Catania Department, Catania, Italy C. Cavallotti, Dipartimento di Chimica Fisica Applicata, Politecnico di Milano, Milano, Italy D. Crippa, MEMC Electronic Materials, Epitaxial and CVD Department, Operations Technology Division,
Novara, Italy D. Dutartre, ST Microelectronics, Central R&D, Crolles, France Srikanth Kommu, MEMC Electronic Materials inc., EPI Technology Group, St. Peters, Missouri M. Masi, Dipartimento di Chimica Fisica Applicata, Politecnico di Milano, Milano, Italy D. J. Meyer, ASM Epitaxy, Phoenix, Arizona J. Murota, Research Institute of Electrical Communication, Laboratory for Electronic Intelligent Systems,
Tohoku University, Sendai, Japan V. Pozzetti, LPE Epitaxial Technologies, Bollate, Italy A. M. Rinaldi, MEMC Electronic Materials, Epitaxial and CVD Department, Operations Technology Division,
Novara, Italy Y. Shiraki, Research Center for Advanced Science and Technology (RCAST), University of Tokyo, Tokyo,
Japan
V o l u m e 73
Processing and Properties of Compound Semiconductors
s. J. Pearton, Introduction Eric Donkor, Gallium Arsenide Heterostructures Annamraju Kasi Viswanath, Growth and Optical Properties of GaN D. F. C. Lie and K. L. Wang, SiGe/Si Processing S. Kim and M. Razeghi, Advances in Quantum Dot Structures Walter P. Gomes, Wet Etching of III-V Semiconductors
258
Contents o f Volumes in This Series
Volume 74
Silicon-Germanium Strained Layers and Heterostructures
s. c. Jain and M. Willander, Introduction; Strain, Stability, Reliability and Growth; Mechanism of Strain
Relaxation; Strain, Growth, and TED in SiGeC Layers; Bandstructure and Related Properties; Heterostructure Bipolar Transistors; FETs and Other Devices
V o l u m e 75
Laser Crystallization of Silicon
Norbert H. Nickel, Introduction to Laser Crystallization of Silicon Costas P. Grigoropoulos, Seung-Jae Moon and Ming-Hong Lee, Heat Transfer and Phase Transformations in
Laser Melting and Recrystallization of Amorphous Thin Si Films Robert Cernf~ and Pete P?ikryl, Modeling Laser-Induced Phase-Change Processes: Theory and Computation Paulo V. Santos, Laser Interference Crystallization of Amorphous Films Philipp Lengsfeld and Norbert H. Nickel, Structural and Electronic Properties of Laser-Crystallized Poly-Si
Thin-Film Diamond I
V o l u m e 76
x. Jiang, Textured and Heteroepitaxial CVD Diamond Films Eberhard Blank, Structural Imperfections in CVD Diamond Films R. Kalish, Doping Diamond by Ion-Implantation A. Deneuville, Boron Doping of Diamond Films from the Gas Phase S. Koizumi, n-Type Diamond Growth C. E. Nebel, Transport and Defect Properties of Intrinsic and Boron-Doped Diamond Milo~ Neslddek, Ken Haenen and Milan Van~Oek, Optical Properties of CVD Diamond RolfSauer, Luminescence from Optical Defects and Impurities in CVD Diamond
Volume 77
Thin-Film Diamond II
Jacques Chevallier, Hydrogen Diffusion and Acceptor Passivation in Diamond Jiirgen Ristein, Structural and Electronic Properties of Diamond Surfaces John C. Angus, Yuri V. Pleskov and Sally C. Eaton, Electrochemistry of Diamond Greg M. Swain, Electroanalytical Applications of Diamond Electrodes Werner Haenni, Philippe Rychen, Matthyas Fryda and Christos Comninellis, Industrial
Applications of Diamond Electrodes Philippe Bergonzo and Richard B Jackman, Diamond-Based Radiation and Photon Detectors Hiroshi Kawarada, Diamond Field Effect Transistors Using H-Terminated Surfaces Shinichi Shikata and Hideaki Nakahata, Diamond Surface Acoustic Wave Device
V o l u m e 78
Semiconducting Chalcogenide Glass I
v. s. Minaev and S. P. Timoshenkov, Glass-Formation in Chalcogenide Systems and Periodic System A. Popov, Atomic Structure and Structural Modification of Glass V. A. Funtikov, Eutectoidal Concept of Glass Structure and Its Application in Chalcogenide Semiconductor
Glasses V. S. Minaev, Concept of Polymeric Polymorphous-Crystalloid Structure of Glass and Chalcogenide Systems:
Structure and Relaxation of Liquid and Glass