Date post: | 24-Nov-2023 |
Category: |
Documents |
Upload: | khangminh22 |
View: | 0 times |
Download: | 0 times |
X-RAY SENSITMTY AND X-RAY INDUCED
CHARGE TRANSPORT CHANGES IN STABILIZED
a-Se FILMS
A Thesis
Submitted to the College of Graduate Studies and Research
in Partial Fulfillment of the Requirements
for the Degree of
Doctor of Philosophy
in the Department of Electrical Engineering
University of Saskatchewan
Saskatoon
MARK TIMOTHY ALEXANDER NESDOLY
Saskatoon, Saskatchewan
Spring 2000
Copyright O 1999: Mark T. A. Nesdoly
National Library BM d c m BiblioWque nationale du Canada
Acqu~tiors and Acquisitions et Bibliographic S e ~ c e s wnrices bibliographiques 395 wdhglon Stroet 395. rue WeUingbon -ON K l A W OtEamON K 1 A W Canada Canada
The author has granted a non- L'auteur a accorde une licence non exclusive licence allowing the exclusive permettant a la National Ij'brary of Canada to Bibliotheque nationale du Canada de reproduce, loan, distniute or sell reproduire, p r k , disttibuer ou copies of this thesis in microform, vendre des copies de cette these sous paper or electronic formats. Ia forme de rnicrofiche/film, de
reproduction sur papier on sur format electronique .
The author retains ownership of the L'auteur conserve la propriete du copyright in this thesis. Neither the droit d'auteur cpi prot&ge cette these. thesis nor substantial extracts from it Ni la these ni des -ts substantiels may be printed or otherwise de celle-ci ne doivent 6tre imprim& reproduced without the author's ou autrement reproduits sans son permission, autorisation.
COPYRIGHT
The author has agreed that the Ll'brary, University of Saskatchewan, may make
this thesis fieely available for inspection. Moreover, the author has agreed that
permission for extensive copying of this thesis for scholarly purposes may be granted by
the professor who supervised the thesis work recorded herein, or in his absence, by the
Head of the Department or the Dean of the Coilege in which this thesis work was done.
It is understood that due recognition wilI be given to the author of this thesis and to the
University of Saskatchewan in any use of the material in this thesis. Copying or
publication or any other use of this thesis for financial gain without approval by the
University of Saskatchewan and the author's written permission is prohibited.
Requests for permission to copy or make any other use of the material in this
thesis is whole or in part should be addressed to:
Head of the Department of Electrical hgineering
University of Saskatchewan
Saskatooxi, Canada S7N OW0
ABSTRACT
Stabilized amorphous selenium (a-Se:0.2-0.5%As7 I0 - 20 ppm C1) is currently an x-ray photoconductor that is in use in medical diagnostic radiography. Amorphous selenium is attractive because it may be deposited in large areas with relative ease. It also hoIds promise for decreasing x-ray exposure for the patient and increasing the resolution of the radiographic image when compared with conventional filmbased radiography. The x-ray sensitivity and charge transport properties of a-Se are far from ideal, however it still performs acceptably as an x-ray detector. Even though existing commercial digital radiographic units employ a-Se as the x-ray detector, there are still several aspects regarding how x-rays affect the charge transport of a-Se that had not been investigated prior to this work. For example, the behaviour of electron transport following exposure to x-rays was unknown prior to this investigation. Knowledge of how exposure to x-rays adversely affects the charge transport properties of a-Se may ultimately lead to some means of minimizing or counteracting these same changes. This one aspect-minimization of x-ray induced charge loss-is a very important factor in decreasing patient dose.
The samples used throughout this work were studied via the timesf-flight (TOF) and interrupted field time-of-flight (IFTOF) transient photoconductivity techniques. The samples were of commercial device quality as evidenced by their low charge loss via deep trapping. This is quantified by a quantity known as the Schubweg, which is a measure of the average distance a charge carrier will travel before becoming trapped and, therefore, lost to conduction. In general, the Schubweg of a charge carrier must be longer than the thickness of the a-Se film in order to ensure that little charge will be lost due to deep trapping-
The detection of x-rays depends on the generation of fiee etectrons and holes within an a-Se film, and the subsequent collection of those charges once they reach the electrodes. This study investigates the mobility and trapping of charges and the recombination of x-ray induced charges in a-Se. Changes in these parameters, caused by exposure to x-rays, were also investigated. Prior to exposure to x-rays, the mobilities and deep trapping lifetimes of both holes and electrons were found to be constant and would not vary over time. No change in the mobility of holes or electrons was found upon exposure to x-rays. However, immediately following exposure to x-rays, the hoIe deep trapping lifetime would fall - 30% while the electron deep trapping lifetime would change oniy - 10%. Mer that initial exposure, both the hole and electron deep trapping lifetimes continued to change ovet time-in some cases rising, while in others falling. These changes were tracked ovet as much as a I2 hour period. Following a 24 to 48 hour rest period after that initial x-ray expomre, the hole and electron lifetimes within the a-Se
film wouId return to a stable, unchanging state. However, these rested lifetimes would not necessarily be equal to the initial lifetime prior to the x-ray exposure one or two days earlier. It is proposed that these changes occur because of a relaxation or reordering of the atoms in a-Se, similar to changes thought to occur as a result of exposure to visible light.
Analysis of the experimental evidence suggests that intimate valence alternation pair (IVAP) charged defects are created by x-ray irradiation; the concentration of these defects was found to be proportional to the absorbed x-ray dose with higher doses creating a higher concentration of these defects, These same defects were observed to be relatively unstable, disappearing within two hours after irradiation. Since these IVAP defects were observed to disappear within two hours after irradiation while the hole and electron lifetimes continued to change for at Ieast 12 hours, it was concluded that the traditiond view of deep charge trapping into these charged IVAP defects cannot be the cause. Indeed, some researchers have long suspected that these charged defects were not the cause of the deep traps in a-Se. A new charge trapping theory, consistent with published optically induced effects, is proposed in this work to explain these observations.
Annealing, i&-red soaking and ultrasonic treatment were investigated as possible methods to restore the original charge transport properties of a-Se films that were exposed to x-rays. However, none of these methods were found to aid this recovery.
The energy required to create a ike electron-hole pair in a-Se by exposure to x- rays was calculated by observing the total charge liberated by an x-ray beam when it impinges a biased a-Se film, and dividing that charge by the energy that the x-ray beam deposits in the h. This energy, Wmp, was found to exhibit a very pronounced field and temperature dependence, while almost no dependence on the mean energy of the x-ray beam was observed. These findings are consistent with the geminate recombination theory? generally agreed to be the dominant charge loss mechanism with optical photons in a-Se.
The persistent x-ray photocmt was investigated and was found to be thermally activated below - -20 "C with an activation energy of 0.16 eV. This energy does not co~espond to the level of any known traps in the bandgap of a-Se, and this finding lends further support to the charge trapping theory developed in this work to explain the charge trapping lifetime behaviour upon exposure to x-rays.
ACKNOWLEDGEMENTS
I would first like to thank my wife Nicole whom I met at the very start of this
endeavour three years ago. Even though I was a poor student at the time, she still found
it in her heart to do me the honour of marrying me. Her support (and patience while I
monopoIized the computer) was very much appreciated. My mother was also there when
I needed it, and for that I will always be gratefid. Chris Haugen also deserves a big
''thank you" for the support and stimulating (not to mention lively) conversations
regarding the measurements. I would like to thank Dr. Hugh Wood for giving me the
opportunity to earn a little extra cash during this study by allowing me to teach at the
undergraduate level. I greatly value the experience. Dr. Narinder Sidhu deserves
mention for the many occasions that he graciously offered his time and assistance, and
for that I am grateful. I would also like to express my appreciation to NSERC and the
University of Saskatchewan for the 6nancial support during the past three years. Finally,
and certainly not least, I would like to thank Dr. Safa Kasap for his support during this
work. His insistence that I continually delve deeper for the underlying truth is
appreciated and is a lesson well leaned.
"In the field of observation, chance favors only the prepared mind" L. Pasteur
TABLE OF CONTENTS
COPYRIGHT ....................................................................................................................... i . . ABSTRACT ........................................................................................................................ n
............................................................................................... ACKNOWLEDGEMENTS iv .................................................................................................... TABLE OF CONTENTS v ...
LIST OF FIGURES ........................................................................................................ vul LIST OF TABLES ......................................................................................................... xvi
......................................................................................... LIST OF ABBREVIATIONS xvii
1 . Diagnostic Radiography and Amorphous Semiconductors ........................................... I ............................................................................................................... 1.1 Introduction 1
1 2 Introduction to Radiographic Imaging ...................................................................... 2 1.3 Advantages of Solid State Detection .................................................................... 6 1.4 Solid State Imaging Systems .................................................................................... 7
1.4. i Xerographic Mode Detection ............................................................................. 7 1.4.2 Active Matrix Array Imaging System ....................................................... 9
1.5 Research Objectives ................................................................................................ 12 ................................................... 1.5.1 Carrier Drift Mobilities and Trapping Times 12
1.5.2 Changes in Charge Transport with X-ray Exposure ........................................ 13 1.5.3 Electron Hole Pair (EHP) Creation Energy ..................................................... 14
.................................................... 1 5 4 Persistent X-ray Photocurrent Investigation 14 ......................................................................................................... 1.6 Thesis Outline 15
2 . Electronic Band Structure and Electrical Behaviour of Amorphous Selenium ............ 16 ............................................................................................................. 2.1 Introduction 16
2.2 Atomic Structure of Amorphous Solids .................................... 16 ......................................................... 2 3 Band Theory of Amorphous Semiconductors 17
2.4 Bonding Structure of Amorphous Selenium ........................................................... 22 2.5 Band Model of Amorphous Selenium .................................................................... 27 2.6 Charge Transport in a-Se ........................................................................................ 28 2.7 OpticaI Properties of a-Se ....................................................................................... 30 2.8 Summary ................................................................................................................ 32
3 . The Timed-Flight Transient Photoconductivity Technique ....................................... 34 ............................................................................................................. 3.1 Introduction 34
............................................................. 3.2 Principle of the T i f - F l i g h t Technique 34 ................................. 3.3 Principle of the Interrupted Field Time-of-Flight Technique 42
. . 3.4 Transient Trap Lmted Theory .................................................................. 43
.................................................................... 3.4.1 Monoenergetic Trap Distn'bution 13 ............................................................................................. 3.4.1.1 Low Field Case 47 ................................. 3.4.1.2 High Field Case ... 4 8
.............................................................................. 3.4.2 Extended Trap Distriiution 48 ............................... 3.5 Summary .. .......................... 52
4 . X-ray Photoconductors ................................................................................................ 53 4.1 Introduction ........................................................................................................ 53 4.2 Ideal X-ray Photoconductive Material ............ .., .................................................. 53 4.3 Practical X-ray Photoconductors ............................................................................ 55
.............................. 4.3.1 Amorphous SeIenium (a-Se) ... .......................... 56 4.3.2 Hydrogenated Amorphous Silicon (a-SiW .................................................... 57 4.3.3 Cadmium Telluride (CdTe) .............................................................................. 57 4.3 -4 Lead Oxide (PbO) ............................................................................................ 58
........................................................................................ 4.3.5 Crystalline Materials 59 4.4 X-ray Sensitivity ..................................................................................................... 60
............................................................... 4.4.1 The Energy Absorption Coeficient 6 0 .................................................... 4.4.2 Energy Absorption and Detector Thickness 63
...................................................... 4.4.3 Electron-Hole Pair Creation Energy: Wm 65 4.5 Summary ................................................................................................................. 70
. ..........................................................*............. 5 Experimental Apparatus and Procedure 71 5.1 Introduction ......................................................................................................... 71 5.2 a-Se Thin Film Preparation ..................................................................................... 71 5.2.1 Substrate Reparation ....................................................................................... 72 . . 52.2 Vacuum Deposlnon System ......................................................................... 73
.................................................................... 5.2.3 Transparent Electrode Deposition 74 5.3 TOF/IFTOF Apparatus ........................................................................................... 76
................................................................................................. 5.3.1 Nitrogen Laser 79 .................................................................... 5.3.2 MOSFET High Voltage Switches 81
5.3.3 Voltage Follower ............................................................................................. 87 . .................................................................................. 5.3.4 FPGA Tlrmng Generator 89 . . .............................................................................................. 5.3.5 Data Aquisltlon 92 5.4 X-ray Photocurrent (Wm) Measuremmts ............................................................. 93 5.4.1 X-ray Exposure System .......................... .. ............................................... 94 5.4.2 Pulse Height Spectroscopy ........................................................................... 97 5.4.3 Spectral MtswumnmbProcedure ...................................................... 100
................... ....................................................... 5.4.4 Measured X-ray Spectra .. 102 .................................................... 5.4.5 Energy Absorbed in a-Se Layer and W m 106
................................................................................................. 5.4.6 I-V Converter 110 .......................................................... 5.4.7 Low Temperature Protective Chamber 112
5.4.8 X-ray ColIimatodFilta Housing .................................................................... 113 ....................................................................... 5.5 Miscellaneous Experimental Tools 115
.................................................................................................. 5.5.1 Sample Oven 115 5.5.2 IR Light Source .......................................................................................... 116
5.5.3 Ultrasonic Waves ........................................................................................... 1 16 5.6 Summary ............................................................................................................... 117
............................................................................................... 6 . Results and Discussion 118 6.1 Introduction ................................................................................ ........................ 118
......................................................................... 6 2 Charge Transport Measurements 118 ...................................................................................... 62. L TOF Measurements 118
6.2.2 lFTOF Measurements .................................................................................... 123 6.2.3 a-Se Film Quality ........................................................................................... 134
....................................................... 6.3 X-ray Induced Changes in Charge Transport 137 ...................................................................... 6.3.1 Charge Carrier Transit Time t~ 138
6.3.2 Charge Carrier Lifetimes .......................................................................... 139 6.3.3 Light Induced Structlrral Changes ia a-Se ..................................................... 154
....................................... 6.3.3.1 Trapping Mechanisms and the Observed Results 161 ............................................................ 6.3.3 2 Irreversiile X-ray Induced Damage 171
........................................................................ 6.3.4 Electron L i m e Recovery 171 6.3.4.1 Anneding .................................................................................................... 173
.................................................................................................. 6.3.4.2 IR Soaking 175 .................................................................................. 6.3.4.3 Ultrasonic Treatment 176
6.4 Electron Hole Pair (EHP) Creation Energy - WEHP ............................................. 179 ................................................................. 6.5 Persistent X-ray Induced Photocurrent 187
6.6 Summary ............................................................................................................... 190
........................................................................... 7 . Conclusions and Recommendations 192 ........................................................................................................... 7.1 Intraduction 192
7 2 Charge Transport Study ........................................................................................ 192 ...................................... 7.3 Changes in Charge Transport Upon Exposure to X-rays 193
............................................................................................ 7.4 EHP Creation Energy 194 ................................................................. 7.5 Persistent X-ray Induced Photocurrent 194
................................................................................ 7.6 Suggestions for Future Work 194
8 . References ................................................................................................................... 197
LIST OF FIGURES
F i e 1.1. An x-ray image of the hand of the wife of Wielm Roentgen. The image was obtained with an exposure of more than 30 minutes; an eternity by modem standards. [Image obtained fiom httDJ/~~~.accessexceilence.ordAElAEClCC/hist~rical back~round.html] ........... 3
Figure 1.2, TypicaI projection radiographic medical imaging system .............................. 4 Figure 13. (a) Schematic of a-Se plate with active matrix readout and (b) a
cross-section through one pixel (not to scale). [After Rowlands J., and Kasap S.O., "Amorphous semiconductors usher in digital x-ray imaging", Physics Today, 50, 1997, pp. 24-30.] ..................................................................................... 10
Figure 1.4. An x-ray image of a wrist phantom obtained by an experimental active matrix array imaging system (reduced h m original; courtesy of Dr. J. Rowlands, Sunnybrook Health Science Centre). ................................................. 1 1
F i i 2.1. Two dimensional representatioa of the structure of (a) a crystalline solid; (b) an amorphous solid. Spheres marked " 0 and 'W' represent over-coordinated and under-coordinated atoms, respectively. ................................. 1 7
F i i e 22. Density of states for a crystatline semiconductor. ........................................ 18 F i e 23, DOS models as proposed by (a) Mott; (b) Cohen, Fritzsche and
Ovshinski (CFO); (c) Marshall and Owen. The hatched regions denote localized energy states. ............................................................................................. 20
Figure 2.4. Selenium chain and the definition of the dihedral angle 9. The dihedral angle is defined as the angle separating the two planes defined by atoms 123 and 234. It may be observed by looking down the bond connecting atoms 2 and 3 [26]. ................................................................................ 23
Figure 25. The random chain model of the structure of a-Se showing localized regions that are ring-like and chain-like [26] ............................................................ 24
F i e 2.6. Structure and energy of simple bonding configurations for seienium atoms. Straight lines represent bonding orbitals, lobes represent lone-pair (nonbonding) orbitals, and circles represent antiinding orbitals. The energy
................................. of a lone-pair is taken as the zero energy. [Adapted h m 381 25 ......................... Figure 2.7. Illustration of the a-Se structure detailing an IVAP defect. 26
F i e 2.8. Experimentally determined density of states function for amorphous selenium [39]. .......................................................................................................... 28
F i i e 2.9. Absorption coefficient a (solid line) and quantum efficiency q (broken lines) as a function of the incident photon energy hv for various applied electric fields [I 7,461. ................................................................................. 3 1
F i e 3.1. (a) Simplified schematic and (b) small signal ac equivalent for the TOF transient photoconductivity technique. C, is the sample capacitance. ............ 35
F i e 3.2. The motion of a charge q through a distance dx in the sample induces a charge dQ to flow in the external circuit ................................................................ 38
Figure 33. Signals from the TOF experiment where (a) is the I-mode signal and (b) is the V-mode signal. Solid lines represent expected signals in a trap-free solid while dashed lines represent signals in a solid with deep traps only. .............. 41
F i e 3.4. (a) A typical TOF waveform and (b) a typical ETOF waveform. The interruption time is ti; the magnitude of the photocurrent immediately before and after interruption is denoted il and iz .............................................................. 42
Figure 3.5. Current flow with trapping and reIease processes in a thin slice of semiconductive material. ....................................................................................... 44
Figure 3.6. Sketch of a Gaussian distribution of shallow energy traps that lay immediately below the conduction band edge. ........................................................ 5 1
Figure 4.1 An attenuator is placed in an x-ray beam. ..................................................... 6 I Figure 4 3 A number of different interactions are possible when an x-ray photon
enters a material. ................................................................................................ 6 1 Figure 4 3 Energy absorption coefficient vs. photon energy for various
................................................................................. photoconductive materials [8 11 64 F i e 4.4. EHP creation energy vs. bandgap for a selection of photoconductive
materials [8 11. ........................................................................................................... 67 F i e 45. Schematic representation of the different types of recombination that
are posslible in an a-Se photoconductor. The cylinders represent the tracks of primary electrons. Bulk recombination occurs outside the tracks between charges that originated in different tracks. Geminate recombination occurs between the original hole and electron. Columnar recombination takes place between electrons and holes h m different pairs, but within the same trac k........... 69
Figure 5.1. Schematic diagram of vacuum deposition system ....................................... 74 F i r e 52. Schematic diagram of the electrode sputtering system. ................................. 75 F i e 53. SimpIified view of the TOFAFTOF apparatus. Two PCs were used;
the 8088 PC connected to the CCD m e r a (anaiog oscilloscope) was necessary because the antiquated video h e storage card could not be used in the modem pentium PC. ......................... ,., .......................................................... 78
Figure 5.4. Potential energy mes for the lowest triplet states in the N2
molecule [98]. .................................................................................................... 80 F i r e 5.5. Blumlein circuit used for rapid excitation of Nz laser. ................................. 8 1
....................... F i r e 5.6. Trigger timing requirements of the LN 103 C nitrogen laser 82 F i e 5.7. Application and removal of the high voltage bias during the lFTOF
........... experiment induces large switching transients across the sampling resistor. 82 Figure 5.8. The MOSFETloptoisolator gate driver pair that forms the basic
building block of the switches built for this work. MOSFET: Motorola MTP3NIOOE n-channel enhancement-mode rated at 1000 V, 3 A, R,,, = 4 R, Q,, = 32.5 nC. Optoisolator Hewiett Packard HCPL-3101 Power MOSFETnGBT Gate Drive Optocoupler rated at 5000 Vac isolation and 0.4
........... A peak output current. A calculated switching time of 8 1 ns is achievable. 84 ............................................................... Figare 5.9. Schematic of the high side switch, 85
Figure 5.10. Schematic of the low side switch ..................... .. ............................. 86 F i i 5.11. Unloaded switching transients of the high side switch. ............................ .. 88
Figare 5.12. Photograph of the low side switch ............... ,. .................................... 89 Figure 5.13. Circuit schematic of the voltage follower ................................................ 89 Figure 5.14. Gain vs. hquency response of the voltage follower. ................. -....... ....... 90 Figure 5.15. A simplified diagram of the connections to the Ahera UP1
Education Board showing the fIow of data into and out of the board. .........-......... .. 90 Figure 5.16. A block diagram of the chained counters responsible for generating
the timing waveforms necessary to control the TONIFTOF apparatus, and the timing waveforms thus created (not to scale). ................................................... 9 1
Figure 5.17. Sample connections for measurement of (a) hole transport and (b) ..................................................................................................... electroa transport. 93
Flgure 5.18. Schematic diagram iilustrating the major components of a rotating anode x-ray tube ............................ .. ........,,..... 94
Figure 5.19. Typical electron interactions with a target. (a) Electron suffers ionizational Iosses, giving rise to delta rays and heat. (b) Electron ejects K shell electron leading to characteristic radiation. (c) Collision between nucleus and electron of energy E leads to bmnstmhlung radiation of energy hv. The electron recedes h m the collision with energy E - hv. (d) Collision where electron is completely stopped by a collision with the nucleus. The full energy of the electron is released as bremsstrahlung radiation. ......................... 96
F i i e 5.20. The principle of pulse height spectroscopy. The photon with the highest energy is emitted fiom the radiation source first and strikes the detector first, generating the highest photocurrent pulse. The process is repeated, in turn, for the lowest and intermediate energy photons, which create the lowest and intermediate photocurrent pulses, respectively. ..................... 98
F i e 5.21. A photograph of the pulse height spectroscopy measurement unit used in the course of this work. The photograph on the left shows the DART A/D unit and the Pb shielded amplifier and detector. The photograph on the right is a top view of the amplifier and detector alone, with the top portion of the Pb shielding removed to reveal the cylindrical detector atop the amplifier. ....... 99
Figure 5.2 2. Measured spectrum of the x-ray tube at 50 kVp and 15 mA. The e6ect of PPU completely obscures the real output spectnrm of the x-ray unit . - ......................*.. ..............*...*.*..........................**..*. when no filtettng rs present. ,., 10 1
F i e 5.23. Raw spectral data obtained fiom the pulse height spectroscopy unit for the 39.2 keV beam. PPU is present, but it may be used as a check on the
......................................................................................................................... data 104 ..................... Figure 5.24. (a) Raw data and (b) filtered spectra of the 39.2 keV beam 105
Figure 5.25. The manually filtered spectra of the four beams used throughout the ............ course of this work with mean energies of 32.8,39.2,47.1 and 58.2 keV. 106
.................. Figare 5.26. Diagram of the apparatus used to measure the WUrp of a-Se. 1 10 F i e 5.27. Circuit schematic of the I-V converter built for this work with a
.................... variable conversion gain selectable in decades from 16 to 10' V/A. I 1 1 Figare 5.2%. Diagram of the low temperature protective chamber flooded with
Ar ............................................................................................................................ 113 Fie 5.29. A cutaway view of the modified collimator attached to the x-ray
unit to restrict scattered radiation from reaching the samp1es. ............................... 1 14
Figure 530. Simple oven for sample annealing using a temperature controller and a heating element. A thermocouple connected to tbe temperature controller provides a means to measure the temperature of the aluminum block .................................................................................................................. 1 15
Figure 531. Equipment employed to create ultrasonic waves in the a-Se samples .................................................................................................................. 1 16
F i i e 6.1. Typical I-mode TOF waveforms in a-Se for (a) holes and (b) electrons. The transit times are indicated as tr. [(a) F = 0.98 VIP, (b) F = 5.69 Wpm] ............. ,., ............................................................................................ 120
Figure 6.2. (a) Typical electron TOF photocurrent showing the location of the photocurrent's knee and co~~esponding !4 value point. (b) The photocurrent of (a) is differentiated revealing a local maxima corresponding to the location of the knee in (a). Only a portion of the differentiated waveform (the tail) has been shown for clarity .............................................................. 121
Figure 6.3. Plot of TOF tramit time vs. 1N for (a) holes showing = 0.132 2 2 ....................................... cm N s and (b) electrons showing = 0.003 18 crn Ns. 122
Figure 6.4. (a) Hole and (b) electron mobility plotted as a hc t ion of applied field, The hole mobility shows very little field dependence while the
............. electron mobility shows a slight field dependence of the form * a FO.". 123 F i e 6.5. Typical electron lFTOF waveform in an a-Se film showing il, i2, and
............................................................................................................................... ti 124 Figure 6.6. Fractional recovered hole photocunent as a hction of intamption
time for a-Se. Measured hole lifetime h m the plot is 132 p. The charge packet was halted at a depth of 0263 L where L is the thickness of the film. ....... 125
F i e 6.7. Fractional recovered electron photocurrent as a function of interruption time for a-Se. Measured electron lifetime h m the plot is 657 p. The charge packet was halted at a depth of 0.263 L where L is the
............................................................................................... thickness of the film. I26 F i e 6.8. Fractional recovered photocurrent at a depth of 0.537 L for the same
sample as in Figure 6.7. The higher y-intercept indicates that less charges are tost at this depth than at the shallower halt depth of Figure 6.7. ...................... 127
Figure 63. Two possiiilities for the lost charge observed in hctiond recovered electron photocurrent lFTOF plots: (a) electron diffusion while the packet is halted and (b) large dispersion during driR The latter case is more likely, as
.............................................. ........... supported by the experimental evidence. .... 1 28 F i e 6.10. Effective electron fransit time tr' plotted as a fimction of
interruption time ti in a typical IFTOF experiment. The quantity plotted is the incremental difference in the effective transit time compared to the transit time obtained &om a TOF experiment .................................................................. 130
Figure 6.11. PIots of differentiated electron lFTOF photocurrent tails vs. time for an a-Se sampie. The intermption time was varied fiom 50.8 p (bottom
................................................................................. curve) to 699. I ps (top curve) 13 3 F i i e 6.12. Incremental eIectron dispersion h m an IFTOF experiment relative
to the dispersion h m a TOF experiment, ATOF, pIotted vs. the interruption ....................................................................................................................... time ti 134
Figure 6.13. Measured electron lifetime, normalized to the initial (t = 0) Lifetime, for an a-Se film over the course of a 6 hour period. The measured lifetime did not significantly vary with time, therefore the measurement technique did not alter the trapping characteristics of the a-Se film under study. The electron lifetime was initially determined to be 5 I0 ps. ......................................... 135
F i i e 6.14. (a) Electron and (b) hole TOF transit times, normalized to the initial t~ and plotted as a function of cumulative dose. Irradiation by 58.2 keV beam; sample shorted to ground during irradiation. .............................................. 138
F i i e 6.15. Normalized electron lifetime at two depths within an a-Se film tracked over time. The film was initially irradiated with a 32.8 keV beam giving an absorbed dose of 26.2 mGy, and the sample was shorted to ground during irradiation. [Sample 97 1002 121 ................................................................. 140
Figure 6.16. Normalized electron lifetime at two depths within the same film as that in Figure 6.15. Irradiation specifics: 32.8 keV, 26.2 mGy, F = 1.96 Wpm. ...................................................................................................................... 141
Figure 6.17. The a-Se film of Figures 6.15 and 6.16 is exposed to an unfiltered x- ray beam of 90 kVp, 15 mA and 1 second duration. The sample was shorted during irradiation. .................................................................................................. 142
Figure 6.18. Nonnaiized electron lifetime at two depths tracked over time. Irradiation specifics: 32.8 keV, 28.1 mGy, shorted during irradiation. [Sample 97 1205 621 ................................................................................................ 143
Figure 6.19. Same sample as that of Figure 6.18; irradiation specifics: 32.8 keV, 28.1 mGy, F = 2.16 V / p . ............................................................................. I44
Figure 630. Same sample as that of Figure 6.18; irradiation specifics: 58.2 keV, 25.4 mGy, sample completely open circuit during irradiation ....................... 144
Figure 6.21. Normalized electron lifetime at two depths tracked over time. Irradiation specifics: 58.2 keV, 25.4 mGy, F = 2.16 Wpm. [Sample 971205 621 ........................................................................................................................... 145
Figure 682. Same sample as that of Figure 6.21, except that the absorbed dose is 1.8 mGy ................................................................................................................... 145
Figure 6.23. Normalized hole and electron Lifetime at the same depth tracked over time. M a t i o n specifics: 582 keV, 1.8 mGy, F = 1.96 Vlpn.
................................................................................................ [Sample 97 1002 121 147 Figure 634. Same sample as that of Figure 6.24, except the dose is now 24.9
mGy ......................................................................................................................... 147 F i e 635. Effective electron transit time t-r' vs, ti for the data of Figure 6.23.
(a) Immediately before irradiation, (b) immediately following irradiation, and (c) two hours after irradiation. At this low dose (1.8 mGy), there is no
.................................... significant change in the behaviour of tT' with irradiation. 148 F i i e 626. Effective electron transit rime tT' vs. ti for the data of Figure 6.24.
(a) Immediately before irradiation, (b) immediately folIowing irradiation, and (c) two hours after irradiation At this high dose (24.9 mGy), the behaviour of tr' changes noticeably with irradiation, but recovers within two hours ........................................................................................................................ 149
Figure 637. Changes in the electron lifetime at two depths within an a-Se I lm as a function of absorbed dose. The samples were shorted to ground while they
were being irradiated. (a) 32.8 keV beam [Sample 971205 621. (b) 32.8 keV beam [Sample 971002 121. (c) 58.2 keV beam [Sample 960521 46 SE 151. (d) 58.2 keV beam [Sample 97 1205 621. ................................................................ 15 1
Figure 6.28. Changes in the electron lifetime at two depths within an a-Se fi as a function of absorbed dose. The samples were biased with a field of - 2 V@n while they were being irradiated. (a) 58.2 keV beam, F = 1.96 V l p [Sample 971002 121. (b) 32.8 keV beam, F = 2.16 V l p [Sample 97 1205 621. (c) 58.2 keV beam, F = 2.16 V l p [Sample 971205 621. (d) 58.2 keV beam, F = 2.35 Wpm [Sample 980622-31. ............................................................. 152
Figure 6.29. Changes in the electron lifetime at two depths within an a-Se film as a fimction of absorbed dose. The samples were biased with strong fields while they were being irradiated. (a) 58.2 keV beam, F = 7.84 V l p [Sample 971002 121. (b) 58.2 keV beam, F = 8.66 V / p [Sample 971205 621. (c) 58.2 keV beam, F = 17.1 V1p.u [Sample 980622 - 31. ............................. 153
Figure 630. (a) A schematic diagram of Tanaka's 1980 [I211 model of bistable local bonding geometries and (b) corresponding double well potential. ................ 156
Figure 6.31. Bond twisting model in a-Se. The equilibrium state (a) is altered when a LP electron is excited from atom A (b). A then feels a strong coulombic attraction to B, which twists A into a new position (c). The excited electron then recombines (d) and the structure is "fiozen in". ................... 157
Figure 6.32. Schematic diagram of the transformation of an exciton in Se into an NAP pair accompanied by atomic distortion. The resulting IVAP is sometimes referred to as a "self-trapped exciton". The resulting Se,+ and
Se; defects are commonly referred to as D' and Dm defects, respectively. ............ 1 58 Figure 633. Formation of metastable triply-coordinated defects which serve to
cross-link adjacent Se chains. Antibonding electrons are denoted as "e". The metastable triply-coordinated defects may decay into (I) their ground state, (11) new bonds, or (111) into an NAP. ......................................................... 159
Figwe 63. Traditional view of hole trapping in a-Se. (a) Electrically neutraI a- Se with an isolated VAP defect and a drifting photoinjected hole. (b) Hole trapped by the D' defect resulting in an electtically neutral defect with a dangling bond .......................................................................................................... 1 63
Fignre 635. Traditional view of electron trapping in a-Se. (a) Electrically neutral a-Se with an isolated VAP defect and a drifting photoinjected electron. (b) Electron trapped by the D" defect resulting in an electrically neutral defect with a dangling bond. ....................................................................... 162
Figure 636. (a) The single trapped hole of Figure 6.34 (b) that results in a neutral singly coordinated defect may rid itself of its dangling bond by approaching a nearby chain and forming a neutral threefold defect. (b) The. unpaired electron of the neutral threefold coordinated defect could be
....... immediately neutralized by a nearby photoinjected hole to form a Dt d e f m 163 F i e 637. (a) The single trapped electron of Figure 6.35 (b) that results in a
neutral triply coordinated defect may rid itself of its dangling bond by releasing a bond and forming a neutral twofold bond and a nearby neutral singly coordinated defect. (b) The unpaired electron of the neutral singly
coordinated defect could be immediately paired by a nearby photoinjected electron to form a D- defect. .................................................................................. 164
F i r e 638. Two Se chains in which two twofold Se atoms are physically quite close to one another. (a) A photoinjected hole neutralizes a lone pair electron on one atom, leading to (b) a transient bond that cross-links the two chains as well as a single unpaired electron. (c) A second photoinjected hole neutralizes the unpaired electron, fonning a pair of D' defects .............................. 165
F i r e 639. Two Se chains in which there exist stretched and compressed bonds. (a) A photoinjected electron nears a stretched bond to form (b) a D- defect and a neutral singly coordinated defect that has an unpaired electron. (c) The neutral singly coordinated defect d i m a short distance before meeting a second photoinjected electron to form another D* defect. ...................................... 166
Figure 6.40. The shallow rested electron lifetime (closed circles) of Sample 971002 12 tracked over the course of the work. The absorbed dose on the occasion immediately preceding each experiment is indicated with the open diamonds. The dose preceding experiment 6 is not known, as that was the
........... exposure without any filtering placed in the beam (mentioned previously) 172 Figure 6.41. The shallow rested electron lifetime (closed circles) of Sample
971205 62 tracked over the course of the work. The absorbed dose on the occasion immediately preceding each experiment is indicated with the open diamonds. ............................................................................................................... 172
Figure 6.42. (a) Normalized electron (solid circles) and hole (open diamonds) lifetimes as a function of total time spent at 35 "C. (b) The room temperature recovery of those l i f h e s tracked over a three day period following the heat treatment. Both lifetimes were measured at the same
............................................ depth within the film: 0.263 L. [Sample 97 1002 121 174 F i 6.43. (a) Normalized electron mobility as a function of total time spent at
35 OC. (b) The room temperature recovery of the mobility tracked over a ..................... three day period following the heat treatment. [Sample 97 1002 121 175
Figure 6.44. Normalized electron lifetime at two depths within an a-Se sample before irradiation (stage I), immediately after irradiation (stage 2), and following 3 hours spent at 35 "C (stage 3). Irradiation specifics: 58.2 keV, 35.7 mGy, sample shorted during irradiation. [Sample 96052 1 46 SE 1 53 ............ 176
Figure 6.45, Normalized electron lifetime at two depths within two different a-Se samples. (a) Before irradiation (stage l), immediately after irradiation (stage 2), following 30 seconds of IR soaking (stage 3). Irradiation specifics: 58.2 keV, 96.6 mGy, sample shorted during irradiation and during soaking. [Sample 971002 121 (b) Before irradiation (stage I), immediately after itradiation (stage 2), following 30 seconds of IR soaking (stage 3), foUowing 90 seconds total IR soaking (stage 4). Irradiation specifics: 582 keV, 28.6 mGy, sample shorted during irradiation, but F = 4 V l p during IR soaking.
...................................................................................... [Sample 960521 46 SE15] 177 Figure 6.46. Normalized electron and hole lifetimes at the same depth (0.40 1 L)
within an a-Se sample initially (stage I), following 15 minutes of ultrasonic treatment (stage 2), two hours thereafter (stage 3). [SampIe 960521 46 SE15J ...................................................................................................................... 178
Figure 6.47. A typical x-ray induced photocment in a biased a-Se film, detailing the "spikes" due to the self-rectifying nature of the tube and the "rising baseline". Data obtained with a 58.2 keV beam and Sample 1463-3 biased at 1.77 Vlpn ............................................................................................................... 180
Figure 6.48. The energy required to create a fiee EHP as a function of the reciprocal electric field for four different mean x-ray beam energies spanning the range 32.8 - 58.2 keV. Linear regressions for the four sets of data converge to within 0.5 eV to yield an expected WL = 5.9 eV at infinite field. ........................................................................................................................ 182
Figure 6.49. Dependence of WEHP on the mean x-ray beam energy at a constant field of 10 V l p . .................................................................................................... 183
Figure 6.50. Plot of the energy required to create a collected EHP, Wm, as a function of temperature. The mean beam energy was 58.2 keV and the electric field was held constant at 1.77 V l p . [Sample 1463-31 ........................... 185
Figure 651. Photogeneration efficiency as a hc t ion of temperature. Mean beam energy: 58.2 keV, sample biased at 1.77 V l p . [Sample 1463-31 ............. 186
F i e 6.52. Plot of the charge contained in the baseline and photocurrent spikes vs. temperature. 58.2 keV, 1.77 V l p , Sample 1463-3. The two exhibit markedly different temperature dependencies. ...................................................... 1 88
Figure 6.53. Baseline charge as a function of the reciprocal of the temperature. The amount of charge released in the persistent photocwent becomes thermally activated below approximately -20 O C . The activation energy is 0.16 eV, which does not comespond to the deep hole or electron traps in a-
LIST OF TABLES
Table 4.1 A concise comparison of candidate x-ray photoconductive materials ............ 59 Table 5.1. Filter combinations and tube settings for the four different spectra ........... 103 Table 53. The four diffetent x-ray beams and their Kph values. ................................... 109 Table 6.1. Charge transport properties of the four a-Se films involved in the . * . carner hfebme study. ....................................................................................... 136 Table 6.2. Charge transport properties of the two a-Se films involved in the x-ray
photoconductive experiments. ... .... . . .. . ... .. . .. . .. . . .. . . .. . . . . .. . .. . . . . . . . . . . . . . . . . . . . . . . . . 1 36 Tabb 63. Minimum operating field for the rested (best case) and damaged
(worst case) electron lifetime conditions observed during the study ...................... 154
LIST OF ABBREVIATIONS
a-Se
a-Si:H
Am
AB
CB
CCD
CFO
CZT
DEC
DOS
DSC
EHP
ESR
FET
FPGA
FWHM
HVL
IFTOF
IR
IT0
I-v
IV AP
~ V P LP
MCA
amorphous selenium
hydrogenated amorphous silicon
analog-todigital converter
antiinding
conduction band
charge coupled device
density of states model proposed by Cohen, Fritzsche and Ovshinski
cadmium zinc telluride
deviant electron configurations
density of states
differential scanning calorimetry
electron-hole pair
electmn spin resonance
field effect transistor
field programmable gate array
W-width half-maximum
half value layer
intermpted field timesf-flight
&-red
indium-tin-oxide
merit-to-voltage
intimate valence alternation pair
peak kilovolts
lone pair
multichannel analyzer
MOSFET
NB
NSB
PACS
PC
PPM
PPU
PZT
SNR
SPECT
SSR
TEA
TFT.
TOF
TP
metal oxide semiconductor fieId effect transistor
nonbonding
normal structure bonding
picture archival and communications system
personal computer
parts per million
pulse pileup
lead zirconate titanate
signal-to-noise ratio
single photon emission computed tomography
solid state relay
transversely excited atmospheric
thin film transistor
time-of-flight
transient photoconductivity
transistor-transistor logic
valence alternation pair
valence band
1. Diagnostic Radiography and Amorphous
Semiconductors
1.1 Introduction
To the best of our present knowledge, we live in a quantum universe on a world
governed by the laws of quantum mechanics. This theory, pioneered over 60 years ago
by physicists such as Heisenberg, Shriidinger and Dirac, has helped us to understand the
intricacies of semiconductor behaviour. Indeed, it has catapulted us headlong into our
modem technological age mainly through the development of the transistor. However,
there is one caveat to quantum mechanics; the complex mathematical application of the
theory to a solid may be greatly simplified only if the material is crystalline in nature (i.e.
periodic). The nonperiodic spacing of the atoms in an amorphous material renders the
application of quantum mechanical theory to these materials quite difficult This is due
to a dramatic rise in the mathematical complexity when dealing with these nonperiodic
mctures.
Electronic devices based on amorphous materials did not experience the same
explosive growth that their c q d k e counterparts enjoyed for a number of reasons.
First, it was Iong suspected that amorphous semiconductors did not share the same charge
transport properties of their crystalline forms. When it was discovered that this was not
the case, modeling the behaviour of an amorphous semiconductor became a daunting task
because of the nonperiodic nature of the amorphous solid. Only since the 1960s has
much of the theoretical understanding of amorphous semiconductors been derived,
mainly through experimez~tal observation. Despite this late start, it is quite probable that
amorphous semiconductors will be the basis of the next era of dramatic growth in the
microelectronics i n d w [I]. Amorphous semiconductors are an attractive alternative to
crystalline semiconductors because they possess a greater diversity of physical properties
and the preparation (or growth) of amorphous solids does not usually require the same
carefully c o n t r o l 1 W d s l o ~ c h i q u e s . This amounts to a potentially tremendous
economic savings for the industry.
In addition, amorphous semiconductors are aiready in majar commercial use in
such applications as amorphous silicon thin film transistors (TFTs), photoconductors and
solar cells. They are also starting to make an impact in the field of medical diagnostic
radiography as an x-ray detector material. The development of radiography and modem
radiographic image formation through the use of photographic film will be discussed in
the next section.
1.2 Introduction to Radiographic Imaging
The discovery of x-rays by the German physicist Wilhelm Conrad Roentgen
(1845 - 1923) in 1895 was an accident. In spite of their accidental discovery, he found
that his newly termed "x-rays" could be used to view the internal structures of the human
body and aid in medical diagnosis. This was very quickly followed by the development
of the field of medical radiography.
Figure 1.1 is an x-ray image of the hand of Frau Roentgen. It is one of the 6rst
known x-ray images of the internal structure of the buman body. Although the image is
very fuzy and the contrast is poor by modem standards, the potential practical use of x-
rays was immediately apparent In June 1896, only 6 months after Roentgen announced
his discovery, x-rays were being used by battlefield physicians to locate bullets in
wounded soldiers.
Radiography st i l l counts itseIf as one of the most usefial and widely used tools to
aid physicians in making a patient diagnosis. It should be noted that diagnostic
radiography refers to those processes that form medical images through the use of x-rays
as the infomation carrier [2]. However, for the most part, modem radiography differs
very little fiom the first systems developed over 100 years ago in that it is still a
photographic film based analog technology.
F i e 1.1. An x-ray image of the hand of the wife of Wilhelm Roentgen. The image was obtained with an exposure of more than 30 minutes; an eternity by modern standards. [Image obtained h m htt~:llwww.accessexcellence.orelAEIAEUCC/historicai backmund.html]
Radiographic imaging systems rely on the differential attenuation of ionizing
radiation (x-rays) by the different s t m ~ and tissues in the body to produce a
radiological image. A typical configuration is shown in Figure 1.2, and consists of the
f d a r projection radiographic system with its x-ray source and an x-ray detector that
are placed in h n t of and behind the patient, reqectiveIy. In present systems, the
detector is almost exclusively based on photographic film.
X-ray Detector
n
X-ray S o w
F i e 1.2. Typical projection radiographic medical imaging system.
The x-rays that pass through the patient undergo differential attenuation
(governed by the body tissues through which they must pass) and this modulates the
radiation intensity that reaches the detector. This radiologicai image must then be
detected and stored until a usable optical image can be produced for the medical
practitioner. Conventional detect& consist of a cassette of photographic film
sandwiched between two fluorescent screens; the screens convert the incident x-ray
photons into visible light which then in turn exposes the film. To recover the image, the
film is &ally developed using standard photographic techniques.
This traditional method has numerous drawbacks. First, there is the problem of
how and where to store the x-ray films once they have been developed, as they occupy
considerable space. Further, these images must be developed using conventional
photographic techniques which are tedious and involve hazardous chemicals. Second,
the information on these films is difficult to share; in order for a doctor to consult a
colleague, the films must either be shipped to their destination or they must somehow be
scanned into digital form in order to be shipped electronically. Another much less
desired option is for the patient to travel to the site of the second doctor and have another
set of x-rays taken there. Third, this method suffers fiom image blwing, a natural
consequence of the fiWphosphor screen combination. Fourth, and perhaps most
important, this method of forming a radiological image requires a relatively high x-ray
dosage (intensity) in order to satisfactorily expose the film so that an image may be seen
afier development, This is due to the fact that the phosphor layers are rather insensitive
to x-ray radiation. Since radiation damage in living tissue is cumulative [3], it is strongly
desirabIe to reduce the x-ray intensity required for a given radiological image. Reduction
of x-ray exposure is tantamount to increased patient safety. The interaction between x-
ray radiation and the human body can lead to biological effects that were first noticed
soon after the discovery of x-rays. The hazard of burns due to overexposure soon became
evident; the risk of cancer was realized Iater. The damaging effects of x-ray radiation
have been summarized as follows:
"Unfortunately, x-rays, like other forms of ionizing radiation, affect living tissue, and exposure to x-rays must be limited to minimize tissue damage. At Iow doses, the biological effect is to change the cell metabolism and structure. This effect appears after only some latency period, which decreases as the radiation dose increases. Higher radiation doses result in cell death. The effect of radiation is cumulative, with small doses received at long intervals having an additive effect. Hence it is very important that x-ray imaging systems use the minimum amount of x-ray energy consistent with obtaining a good image [3]."
The idea of a solid state planar detector for use in x-ray imaging is not a new one;
the concept was first investigated in the 1940s and interest has been fervently renewed of
late [4]. The various advantages promised by a solid state detection scheme will be
introduced in the next section.
1.3 Advantages of Solid State Detection
Any effort to reduce the x-ray exposure of patients (and physicians--as in the
case of some fluoroscopic medical procedures like angiography) would increase overall
safety for all concerned. This particular area is where solid state detection promises the
greatest advantage; with proper selection of the detector material it is possiile to reduce
the exposure level relative to fiIm based radiography and still obtain an image of
comparable, if not superior, quality.
Solid state detection also promises to produce sharper images than photographic
film based techniques. Since a solid state detector directly converts incident x-ray
photons to mobile charge carriers, these carriers may be collected through the use of an
applied electric field. The carriers, for the most patt, travel along the straight lines of the
applied electric field before they are collected. An image may, for instance, be
constructed by ''reading" the amount of charge collected for each given area of the
detector. In the case of an active matrix array detector, careful selection of pixel size,
detector thickness and applied bias field will heip to minimize charge carrier dispersion
between adjacent pixels in the detector, contributing to a sharp image. Conversely, the
visl%le light created by an x-ray photon striking a phosphor in a conventional film
cassette cannot be "steered" as in the case of charge carriers in an electric field. The light
has to propagate through the phosphor layer to the film in order to expose it and create a
useable image. That light difbes appreciably due to scattering in the phosphor layer,
and thus reduces the sharpness of the recorded image.
Electronic x-ray image reconstruction holds many other advantages over present-
day film based techniques. First and foremost are the savings in the quite substantial
physical space needed to store conventional x-rays once they are taken. Approximately
80% of the imaging data in modem hospitals are x-ray images recorded through the use
of photographic film [S]. The need for the chemicals and the time to develop the film are
also eliminated. Further, the need fbr film itself is negated with a reusable solid state
detector.
Digital radiography lends itself to such advantages as computer aided diagnosis
161, and also permits the possibiiity of remote access and teleradiology where highly
qualified personnel could provide service to remote regions from a central location.
1.4 Solid State Imaging Systems
There are two broad classes of solid state x-ray imaging systems, delineated by
how they collect the charge created by the x-ray photons incident upon them. In general,
the detector may be used with a perm~ulentIy applied bias (short circuit mode) or in an
open circuit fashion. The open circuit (xerographic) mode is discussed first.
1.4.1 Xerographic Mode Detection
The xerographic mode was the basis of early solid state radiographic systems,
probably because of the similarity with the xerographic process used in commercial
photocopiers. The detector consists of a layer of amorphous selenium (a-Se) deposited
onto an aluminum substrate. The surface of the detector is then electrostatically charged
to a high potential with a scorotron prior to x-ray exposure. This uniformly distriiuted
surface charge generates a strong bias electric field within the a-Se layer. The detector is
then placed behind the patient and an x-ray exposure is performed.
The x-rays incident on the a-Se layer Liierate charge carriers in the bulk of the
layer, which are then swept either to the stlrfilce or to the A1 substrate, depending both on
the species of carrier (electrons or holes) and the polarity of the applied field. The net
effect is that some of the d a c e charge of the a-Se layer is proportionately discharged in
relation to how much x-ray radiation was incident at each location on the !ayer, greater
discharge occurring where exposure was higher (i.e. 'hugh tissue) and lesser discharge
occurring where exposure was lower (i.e. through bone). This leaves a latent eiectrostatic
image on the d a c e of the detector, which can be transformed into a viewable image in a
number of ways.
Again, this particular method of forming an x-ray image is very similar to the
xerographic process, so it should come as no surprise that the first development method
to be used was powder cloud development [5 ] . Powder cloud development is an andog
image formation technique. The deveIoped image could be viewed directly on the
selenium plate or transferred to paper for a permanent record. The selenium plate is
reusable, but must be cleaned of residual powder after each cycle.
A variety of methods to electronically develop the latent electrostatic image have
been developed. One such method involves the use of an array of micro-electrometers to
sequentially scan the charge distniution present on the surface of the selenium plate [7-
91. The general idea is that if a small sensing electrode is held very close to the surface of
the selenium plate, then the current flowing through the probe as the plate is slowly slid
past the probe will be proportional to the latent electrostatic image (charge) on the plate.
The probe current versus position data must be processed to yield an image, and this data
is generally converted to digital form to be processed and recorded by computer. This
method is advantageous in that the image is not altered by the readout process itself, and
repetitive scans of the same image may be made to improve recognition of fine detail and
verify previous measurements. However, this method suffers h m the fact that it must
employ mechanical parts to position the a-Se plate and precisely hold the electrometer
probes just above its surface. Obviously a I00036 electronic solution would be preferred,
especially since the a-Se plate must be transported to and inserted into the
scanning/analysis unit after every exposure; a tedious and an unnecessary task. Philips
has introduced a commercial digital x-ray imaging system called Thoravision that is
based on the electrostatic readout technique in which the latent electrostatic image on a
photoconductor drum is read by using a series of electrostatic probes. The measurement
is then digitized to provide an electronic image.
Another electronic deveIopment technique involves the sequential discharge of
the latent image through the use of a scanning pulsed laser [lo-1 I]. A transparent
electrode is coupled to the surf'ace of the detector and assumes a potential related to the
d a c e charge of the latent image. As the laser is scanned across the d a c e of the
detector, the charge under the electrode is locally discharged, inducing a current to flow
in the electrode. The image may be reconstructed by correlating the change in the
electrode signal to the position of the laser. This method is capable of high resolution
provided that the laser is precisely focused and p o s i t i o n e k immense tecbnologicd
feat in and of itself,
The principal disadvantage of the xerographic mode is that in the absence of a
permanent applied bias field, areas that are exposed to high x-ray fluences (intensities)
will experience a weakening of the static bias field. This is due to the fact that the
surfice charge will be increasingly depleted as more charge is liberated by x-rays. If the
field becomes too small, then charge carriers will no longer be drifted straight to either
the surface or the substrate, resulting in a blurring of the image. This same phenomenon
effectively negates xerographic mode detection systems h m continuous (real time)
acquisition of data, as in the case of fluoroscopy, since there is no practical method to
replenish the surface charge on the detector in real time.
1.4.2 Active Matrix Array Imaging System
A number of similar imaging systems based on externally biased a-Se plates with
active ma& readouts have recently been proposed 112-151. A schematic representation
of these systems is shown in Figure 1.3.
An amorphous selenium layer is used to convert the incident x-ray photons into
eiectron-hole pairs (EHPs) that are then swept apart by the externally applied bias field
and collected on pixel electrodes. As in the xerographic mode of operation, the charge
released in each area of the a-Se layer is proportional to how much x-ray radiation was
incident on that area, which in turn is proportional to the radiographic image. The active
matrix array consists of an array of thin 6lm ttaasistor (TFT) switches. As shown in
Figure 1.3 (b), the charges that reach the bottom of the a-Se layer (holes in this case) are
collected by pixel electrodes and stored by a capacitor Cij. The TFT switches are used to
control the reading of the image charge, one line at a time. The TFTs in each row have
their gates connected and the TFTs in each column have their sources connected. During
readout, a gate line is activated and the charge fiom the individual pixel electrodes in that
line are read in a parallel fashion and multiplexed into a serial data stream which is fed to
Multiplexer - Digitizer - Computer
Data (Source) Lines
. 1 b
E l & '. i Shield v - sia
Figure 1.3. (a) Schematic of a-Se plate with active matrix readout and (b) a cross- section through one pixel (not to scale). [Afta Rowlands J., and Kasap S.O., "Amorphous semiconductors usher in digitd x-ray imaging", Physics Today, 50, 1997, pp* 24-30.]
a computer for processing and display. The next row is then read, and the process
continues sequentially until the whole matrix has been scanned.
The active matrix m y system offers a number of advantages over traditional
film based radiography and xerographic radiographic methods. First, it offers a
potentially high resolution imaging method, governed by the size of the pixel electrodes
and the lateral spread of the charge carriers. The pixel size in current systems is
approximately 150 pm, which translates to an image quality approaching that of
conventional film based systems. However, the pixel size of future systems can be easily
reduced, leading to higher resolution in the radiographic image, Unlike a xerographic
mode system, no handling of the detector is required, and there are no moving parts
involved in the readout of the image. Further, the readout of the image is limited only by
how fast the TFT array can be scanned (several times per second). The x-ray image in
Figure 1.4 was obtained by the experimental active matrix system developed at the
Sunnybrook Health Science Centre in Toronto.
Figare 1.4. An x-ray image of a wrist phantom obtained by an experimental active matrix array imaging system (reduced Erom origid; courtesy of Dr. J. Rowlands, Sunnybrook Health Science Centre).
Another advantage lies in the permanently applied bias which allows for a
radiographic technique called fluoroscopy. Fluoroscopy is an x-ray imaging technique
that produces red time x-ray images. Charges are continuously generated and collected
as long as x-ray photons strike the a-Se plate. By continuously polhg the active matrix
array, a real time image can be generated at every complete scan time. FIuoroscopy finds
extensive use in angiography; a medical procedure to examine the blood vessels of the
heart. Angiography requires that a red time image of the heart be available to the
physician in order to properly guide the necessary instnrment(s) to the site of any
coronary blockage and mpen the artery, as with heart attack victims.
1.5 Research Objectives
The areas of investigation that will be undertaken in this work involve the charge
transport parmeters of biased a-Se pIates and how these are influenced by x-ray
exposure. The basis of this experimental examination will be the Time-of-Flight and
Intermpted Field Time-of-Flight (IFTOF) transient photoconductivity techniques. Both
are pow& ptoven methods for studying the charge transport kinetics in a host of
materials. The following sections will provide a brief outline of the research objectives
for this work.
13.1 Carrier Drift Mobilities and Trapping Times
Pure a-Se is not used as an x-ray photoconductor because pure a-Se crystallizes
over rime. Crystalline selenium is unsuitable as an x-ray photoconductor because it has a
much lower dark resistivity than a-Se, which leads to a dark m e n t that is orders of
magnitude greater than in the amorphous solid. It was found that alloying pure a-Se with
mall amounts of arsenic (0.2 - 0.5% As) greatly improved the stability of the composite
tllm and heiped to prevent crystaliization. However, the addition of arsenic was also
found to adversely affect the hole lifetime p because the arsenic introduces deep hole
traps. If the alloy is doped with 10 - 20 parts per million @pm) of a halogen (such as
chlorine), the hole Hetime is r e s t d to its initial value. Thus, the photoconductive
plates typically in use consist of a-Se that has been alloyed with 02-0.5% As (nominal
03% As) and doped with 10 - 20 ppm Cl.
Charge transport parameters such as electron and hoIe mobiIity (k and h) and
electron and hole lifetime (G and rb) in these alloyed a-Se plates depend quite strongly on
the relative amounts of the doping species, the original purity of selenium, and also on
the sample preparation conditions. The carrier Lifetimes demonstrate a higher sensitivity
to these factors than the mobilities. As mch, each and every caudidate sample that will
be used in the course of this study must be examined to determine these critical charge
transport fictors, Any samples found to possess poor charge transport will be excluded
ftom the balance of the study-
These initial measurements provide a means of comparing the suitability of the a-
Se plates as an x-ray detector with other possiile caudidate maten'als such as cadmium
zinc te1Iuride (CdI-,Zn,Te or more simply, CZT), PbO, CdTe, etc. The comparison of
charge transport parameters may be performed by analyzing the Schubwegs, the product
of the charge carrier mobility, the carrier lifetime and the applied field. The Schubweg is
a measure of the average distance a carrier will travel before becoming trapped.
1.5.2 Changes in Charge Transport with X-ray Exposure
Studies of hole transport in a-Se have shown that when a-Se radiographic plates
are exposed to x-rays, the hole lifetime becomes affected. For example, Kasap et al. [16]
have demonstrated that x-ray exposure of unbiased a-Se xerographic photoconductive
films decreases the hole lifetime.
Electron Wetime, and how it may be affected by x-ray exposure has not been
previoudy reported. A set of experiments using the IFTOF transient photoconductivity
technique to monitor changes in electron lifetime induced by x-ray exposure will be
performed (the photoinjected charge is supplied by a laser pulse). Furthermore, the
difference between exposures made with the photoconductor under bias and no bias
conditions must be investigated, corresponding to the two possl%le modes of operation of
a digital radiographic detector. Three methods to restore the electron lifetime will be
investigated; annealing, Sa - r ed (IR) soaking and ultrasonic treatment. These results
will be monitored to search for any indication of x-ray induced unrecoverable damage to
the a-Se samples. Ultrasonic recovery and permanent x-ray induced damage have not
been previously reported.
1.53 Electron Hole Pair (EHP) Creation Energy
The x-ray sensitivity of a material is the true measure of its worthiness as an x-ray
imaging receptor, and the EHP creation energy is one of the factors that determines the x-
ray sensitivity of a material. The study involves monitoring the x-ray induced
photocurrent in a biased a-Se film. By integrating the x-ray induced photocurrent, the
total amount of charge liberated by the incident x-ray photons in the film can be found. If
the amount of energy that the x-ray beam deposits in the a-Se film is known, then a
determination of the energy to kee an electron-hole pair, WmP, may be made. W ~ P is
the average energy required per collected (freed) electron-hole pair, it is not the average
energy per created electron-hole pair, Wk . The difference arises because some of the
created electron-hole pairs quickly recombine and do not contribute to the photocurrent,
effectively raising the energy required to generate collectable EHPs. This process is well
understood in the optical regime [I71 but is still a point of controversy at x-ray energies.
The competing theories predict different bias field, temperature, and x-ray energy
dependencies. A set of experiments detailing how Wmp varies as a function of these
parameters will be performed and will thus assist in resolving this dispute.
1 S.4 Persistent X-ray Photocurrent Investigation
[mmediately following x-ray irradiation there exists a penistent photocurrent in a-
Se whose origin has traditionally been attniuted to the thermal release of trapped charge
carriers. This persistent photocurrent limits the speed of an a-Se radiographic detector
becase the detector must be allowed to rest between exposures to M y discharge this
persistent photocurrent to prevent image ghosting. An experiment that investigates how
this persistent photocurrent changes with sample temperature will be performed in an
effort to identify the true source of this phenomenon.
1.6 Thesis Outline
This thesis is divided into a total of seven chapters. FoIIowing this introductory
chapter, a brief overview of the various properties of amorphous selenium will be given
in Chapter 2. Chapter 3 will provide the theoretical principle of the TOF and EFTOF
transient photoconductivity technique. A discussion of the properties of x-ray
photoconductors will be presented in Chapter 4, along with a detailed discussion of the
method of calculating the x-ray energy absorbed in an a-Se layer. A description of the
compIete experimental apparatus and a brief description of the sample preparation
procedure will be provided in Chapter 5. Chapter 6 will present the results of the
experimental work. The conclusions drawn fiwm the experimentit1 results will be
presented in Chapter 7, along with some recommendations for future work.
2. Electronic Band Structure and Electrical Behaviour
of Amorphous Selenium
2.1 introduction
To gain some measure of understanding of the ekctronic properties of
amorphous selenium (a-Se), a theoretical grasp of its electronic energy band structure is
required. Unfortunately, the quantum mechanical framework that so effectively predicts
the behaviour of crystalline semiconductors cannot be applied to the atomic arrangement
found in an amorphous material due to the nonperiodicity of the structure. As such, the
behaviour of amorphous semiconductors cannot be ptecisely predicted in the same
manner as crystalline semiconductors. Therefore it must suffice to map their electronic
band structure mainly through observations obtained through rigorous experimentation.
2.2 Atomic Structure of Amorphous Solids
Amorphous materials are best compared to a crystalline solid so as to accentuate
the diffetences in their structures, but also to reflect the similarities. A perfect elemental
crystal consists of a regular spatial arrangement of atoms, with precisely defined
distances (the interatomic spacing) separating adjacent atoms. The bonds of each atom
are also arranged at identical angular intervals. The result of this perfect ordering is a
periodic structure, as shown schematically in Figure 2.1 (a).
At &st glance, the amorphous material pictured in Figure 2.1 (b) bears little
resemblance to its orderly crystalline cormterpart. However, even though long range
order is absent in this solid, there is still a high degree of short range order. This short
range order manifests itself because the atoms of the amorphous solid must still satisfy
their individual valence bonding requirements, which leads to little deviation in the
interatomic spacing of the atoms relative to the crystalline case [If!]. However, there is
some small deviation in the bonding angle between adjacent atoms, leading to a
disruption of periodicity in the material.
Figure 2.1. Two dimensional representation of the structure of (a) a crystalline solid; (b) an amorphous solid. Spheres marked "0" and "U" represent over- coordinated and under-coordinated atoms, respectively.
Figure 2.1 (b) also depicts two possible defects that can be found in amorphous
materials: atoms that are either over- or under-coordinated b r n their normal structure
bonding (NSB). The inherently random nature of amoxphous materials and the existence
of these over- and under-coordinated defects has a profound impact on the electronic
band structure of the material.
2 3 Band Theory of Amorphous Semiconductors
Application of quantum mechanical theory to crystalline solids leads to the band
theory; the regular atomic structure of the crystal manifests in bands of allowable
electmn energies when large numbas of atom (i.e. lo2') are brought into close
proximity to form a solid, as shown in Figure 2.2. These bands describe the number of
electron states per unit energy per electron at energy E, through a function called the
density of states, g(E). In the case of most semiconductors, the two principal bands
(denoted the valence and conduction bands) are separated by a gap in allowable electron
energies, called the bandgap E,. This seemingly simple allowable electron energy
shucture determines the electronic properties of all crystalhe semiconductors.
c Conduction Band 1.
(extended states)
Band Gap t v
Valence Band (extended states) //
Densrty of States, g(E)
Figure 23. Density of states for a crystalline semiconductor.
The band theory was first derived for the case of periodic atomic structures (i-e.
crystalline solids) because the rigid periodicity made possr3le some mathematical
simplifications. Because amorphous materials did not possess any obvious order, it was
long believed that their electronic band structure was vastly different from the crystalline
case. However, when it was discovered that amorphous solids possessed the same basic
electronic and optical properties as their crystalline brethren [19], it was concluded that
only short range order in the atomic structure was necessary for the band theory to be
applicable.
The first attempt to quantify the energy band structure of amorphous materials
was undertaken by Mott [20]. He noted that all semiconductor crystals have two things
in common:
each individual electron within the crystal is descnied by extended Bloch wavefunctions that possess long range order in both phase and magnitude; the allowable electron energies fall into bands which are separated by a well delineated (ie. "sharp") energy gap.
Mott's work was based on the assumption that amorphous materials, despite their
obvious structural difference with aystalhe materials, will have certain similarities in
their band structure. He postulated that the Bloch wavefunctions of the amorphous solid
would have long range order in their amplitudes, but only shoa range order in their
phases; this results in a dissolution of the sharp band edges of the crystal to tails of
highly localized states that extend into the energy gap. This concept is illustrated in
Figure 2.3 (a) where there is an energy E, (the conduction band edge) above which the
eIectmnic states are extended and below which they are localized (and vice-versa for the
valence band edge &).
The first quantitative assessment of disorder and its effect on the solutions of the
Schrijdinger equation was performed by Anderson [21]. In an amorphous solid, the bond
angles between adjacent atoms are slightly distorted h m their ideal value, and long
range order is destroyed. Anderson demonstrated that these conditions of random
variations in the electron potential lead to IocaIized tail states which lie within the
forbidden energy gap, now dubbed Anderson localization.
A Conduction ,' I ,
A Conduction
Band Band
Conduction , /'
Band b
Figure 23 . DOS models as proposed by (a) Mott; (ti) Cohen, Fritzsche and Ovshinski ((30); (c) Marshall and Owen. The hatched regions denote localized energy states.
These tail states in an amorphous semiconductor have a profound effect on h e r
conduction. In a crystal, the carriers travel in the conduction or valence band via
electronic energy states which extend throughout the entire crystal. Due to the transition
between the extended and tail states in an amorphous solid, there is a corresponding
transition in the carrier mobility. Carriers can either tunnel between localized states,
controlled by thermal activation, or, given sufficient energy, they can travel in the
extended states, as in the crystalline case. This change in the mobility of charge carriers
led to the idea of a mobility gap in amorphous materials, much akin to the bandgap of
crystalline solids.
M. H. Cohen, H. Fritzsche and S. Ovshinski suspected that Mott underestimated
the amount of disorder in an amorphous solid. As such, the localized tail states would
extend throughout the energy gap of the material and actually overlap in the region of the
Fermi level, as shown in Figure 2.3 (b) [22]. This hypothesis later became known as the
CFO model. It is important to note that even though there exists a continuum of energy
levels throughout the gap, metallic conduction in this model is not possible because the
widened tail states are still highly localized in space, as in Mott's model.
The DOS model proposed by Marshall and Owen [23] is shown in Figure 2.3 (c).
They noted that all solids, crystalline or amorphous, contain defects such as dangling
bonds, chain ends, interstitials, vacancies, substitutional impurities, etc. These defects
lead to localized energy states within the bandgap of the material, in addition to the
disorder induced tail states. It was previously assumed that the disorder induced states
would be of sufficient number to mask these defect states, but Marshall and Owen
proposed that there would be significant mid gap states caused by these defects. It is
important to note that, in this model, the Fermi level is determined by these defect
statdonor-l ike and acceptor-like in the upper and lower half of the gap, respectively.
The Fermi level remains near the middle of the gap because self compensation adjusts the
concentrations of the donor and acceptor states. This has ramifications with respect to
extrinsic conduction, as doping the amorphous solid with donors or acceptors would have
little effect on the Fermi level; thus extrinsic conduction would be difficult to achieve.
Obviousiy since the electronic properties of a semiconductor are linked--through
the DOS-to any deviations fiom the NSB present in the solid, these deviations hold
particdar interest. In the following section, the various defects that may arise in a-Se are
discussed.
2.4 Bonding Structure of Amorphous Selenium
Selenium is a member of the group VI column of the periodic table; the elements
from this column have the family name of chalcogens. The atomic number (2) of
selenium is 34, and it has six valence (outer shell) electrons. The two valence electrons
in the s state form a lone pair (LP) and do not participate in bonding. Of the remaining
four p state electrons, only two are available for covalent bonding because there is
normally another lone pair in the p state. These lone pairs are sometimes referred to as
nonbonding states. This two-fold coordination bonding configuration for selenium has an
optimal bond angle between the bonds of 105" [34] and represents the lowest energy
configuration of the atom. Selenium therefore exhibits a chainlike structure because of
this divalent bonding scheme. In addition to native defects, elements from groups N or
V of the periodic table may be introduced to cross-link the chains to achieve three
dimensional stability in the solid.
Selenium has two crystalline fonns, monoclinic Se (a-Se) and trigonal Se (y-Se).
a-Se is composed of Ses rings while y-Se consists of parallel spiral Se chains. Given
these two possible ring and spiral chain structures in crystalline Se, it is natural to assume
that amorphous selenium would have ring-like and chain-like structures randomly
distriiuted throughout the solid. However, recent structural studies of a-Se and its alloys
favour a random chain model in which almost all the individual atoms of the solid are
bonded in a twofold coordinated chain structure. The dihedral angle @ of this chain
remains constant in magnitude but changes sign randomly [24,25]. The dihedral angle @,
as shown in Figure 2.4, is defined as the angle separating adjacent bonding planes and
may be seen by aligning the observer on the bond that connects atoms 2 and 3.
In a-Se, the dihedral angle is random so that regions that are ring-We or regions
that are chain-like are possible. If + or - are used to denote the relative phase of the
dihedral angle, then a sequence of +-+- has been tenned ring-like and a sequence of +t+i
or -- chain-like [24]. The local order shown in Figure 2.5 may then be characterized as
tt+-+-+-. This random chain model has successfully been employed to explain the
Figure 2.4. SeIenium chain and the definition of the dihedral angle 4. The dihedral angle is defined as the angle separating the two planes dehed by atoms 123 and 234. It may be observed by looking down the bond connecting atoms 2 and 3 [26].
vlibrational spectra of a-Se to account for Sea-like spectral features in the inbred
absorption and Raman scattering spectra without invoking a mixture of distinct ring-like
and chain-like portions present in the strumre. Other structural studies of a-Se generally
support the random c h i n model [27,28].
The structure of an amorphous solid is not compktely random, and there is, in
fact, a degree of order to the stnrcture-at least between individual atoms. As stated
above, each individual atom in the structure strives to fill its vdence states, thereby
normally defiulting to a random chain structure. However, not ail the atoms can satisfy
their individuai valency requirement due to the lack of periodicity in the structure.
Consequently, some ofthe atoms become over- or under-coordinated.
Figure 2.6 pictures the bonding schemes possiiIe in a-Se; for the purpose of
discussion, only the lowest energy bonding state, Se,O (which represents the twofold
Se, - Fragment
Chain Segments ' L , . , -. x . ,
Figure 2.5. The random chain model of the structure of a-Se showing localized regions that are ring-like and chain-like [26].
coordinated structure discussed above) will be considered to be the "normal" structure;
the other posslible states will forthwith be considered defects.
The lowest energy electrically neutral defect is the trigonally coordinated atom,
~ 4 , as shown in Figure 2.6 (d). Three of the pshell electrons enter into bonding states,
and the fourth electron enters into an antibonding state. Another common electrically
neutraI defect is a chain end, denoted ~4 as in Figure 2.6 (b). These defects possess
three pshell electrons which reside in nonbonding states, and one electron available for
boding.
Notation Structure Bonding States
A,B
s4 A A A ; 1 I , i v .
::-.A- '4 '+. B A
v
Energy
E = -2 E b
E = -Eb
E = -Eb+ Uc
E = - 2 E b + A
Figure 2.6. Structure and energy of simple bonding con6ptions for selenium atoms. Straight lines represent bonding orbitals, lobes represent lone-pair (nonbonding) orbitals, and circles represent antibonding orbitals. The energy of a lone-pair is taken as the zero energy. [Adapted from 381.
a-Se 61ms contain a large number (estimates range as high as 10'" 1Id //an3 of
thermodynamically derived charged structural defects called valence alternation pairs
(VAPs), which correspond to some of the Se atoms being over- or under-coordinated [29-
3 11. The absence of a detectable electron spin resonance (ESR) signal is indicative of no
unpaired electrons (dangling bonds) in the structure [32, 331. This means that there
cannot be a singly bonded neutral Se atom, ~ 4 , or a triply bonded neutral atom, ~ 4 , but
a pair of charged centers of the type Se; and (VAP). If the atoms of the pair are in
close proximity? they are termed an intimate valence alternation pair (NAP).
It is energetically more favourable to form the pair of over- and under-coordinated
atoms than it is to form singly bonded defects, as these singly bonded defects are
somewhat unstable. For example, a chain end, ~ 4 , can lower its energy by approaching
the lone pair on a normdly coordinated ~ e i atom and generate an NAP. The diffusion
of the resulting NAP pair away h m each other can furtfier serve to reduce the energy of
the solid. Thus the reaction + ~e,' + S< + Se; is exothermic because the lone pair
electrons have been absorbed into dative bonding. Figure 2.7 is a schematic illustration
of a typical a-Se structure with an IVAP defect.
Figure 2.7. I11usttation of the a-Se structure detailing an NAP defect.
Many photoelectric properties of a-Se and its alloys can at least be qualitatively
explained by using concepts based on VAP and WAP defects and interconversions
between the normally bonded (twofold coordinated) atoms present in the structure and
these defects. The physics of such processes has been extensively discussed in the
literature [34-363. Their existence and the possible defect reactions that can occur in the
structure have led to many important prerllctions and much insight into the behaviour of
chalcogenide semicoaductors, For example, the linear dependence of the steady state
photoconductivity on the light intensity in a-Se has been interpreted via photoinduced
NAP-type centers [37].
2.5 Band Model of Gmorphous Selenium
The preceding sections may act as the foundation to understand the energy band
structure of amorphous selenium. As stated earlier, the bond lengths of a-Se do not
significantly differ &om the aystalline solid, but the bond angles randomly vary in their
orientation. This, in large part, Ieads to tails of localized energy states that bridge the
energy gap that would not otherwise be present in the energy band of a crystalline
semiconductor. Further, the large concentration of IVAPs in the solid leads to two
localized energy states in the energy gap that are donor- (due to SG defects) and
acceptor-like (due to %- defects), as discussed in section 2.3.
With this in mind, the currently accepted DOS model for a-Se is presented in
Figure 2.8 [39]. It was developed through various transient photoconductivity and
electrophotographic measurements of cycled-up residual and dark discharge [39-421.
Experimental evidence suggests that the localized states (both shallow and deep) in the
energy gap are due to structmd defects of various types that are stable at room
temperature [43]. The near exponentially decaying shallow trap densities with discrete
manifolds at energies of - 0.29 eV above E, and - 0.35 eV below Ec determine the hole
and electron drift mob&@ through a shallow trapconttolled transport mechanism 1441.
These traps are known to be native dekcts, but their exact nature has not been
determined. However, it has recently been proposed that these defects are due to dihedral
angle distortions in the structure of a-Se where the lone-pair orbitals on adjacent Se atoms
approach parallel alignment [45].
. Deep Electron Traps
h 1.0- - - r - - - - - - 5
- ;/CIeep Hole Traps
. EF = 1.06 8V . . - - , Shallow Hole Traps --.
g(E) (cm *3e~'1)
F i 2.8. Experimentally determined density of states function for amorphous selenium 1393.
2.6 Charge Transport in a-Se
In a crystalline semiconductor, conduction occurs mainly due to charge carriers
moving in the extended energy states (the valence and conduction bands). Both the
valence and conduction bands caa carry charges; electrons travel in the conduction band,
while positive holes travel in the valence band. Free electrons and holes are able to drift
in the extended states under the influence of an appIied electric field. In the absence of
some external excitation (i-e. light or x-ray photons), fiee electrons and holes are created
by random thermal viirations of the crystal lattice. These vi%rations can excite electrons
h m the vaIence to the conduction band, provided the vibrations have sufficient energy.
Photoconductivity, on the other hand, depends on incident photons to excite electrons
across the gap-again, if the photons possess sufficient energy. The electron and hole
mobilities in a crystal are limited by the mean time between scattering events (e.g. from
lattice vibrations and defects). These scattering events are rare (relative to the amorphous
case), and as such, mobilities on the order of - lo3 cm2/Vs are commonplace in
crystallie semiconductors.
In an amorphous semiconductor such as a-Se, the localized states ptay a very
important role in conduction. These locdized states that lie in the energy gap act as
trapping centers that can remove drifting charge carriers from conduction, resulting in
effective drift mobilities of - lo-' c&s for holes, and - 10" cm2/Vs for electrons in a-
Se. There are two general classes of traps: shallow and deep, as identified in Figure 2.8.
There is a nearly discrete set of shallow traps at - 0.29 eV above E, and - 0.35 eV beiow
Ec that mainly serve to slow the progress of drifting charge carriers. In addition, Figure
2.8 demonstrates two sets of deep traps that are distributed in energy near the midgap;
these traps are sufliciently energetically deep that a carrier, once trapped, is effectively
forever removed from conduction.
Each trap is described by both a capture and a release lifetime. The capture
lifetime is defined as the mean time that a mobile carrier can drift in the extended states
before beaming trapped. The release lifetime is defined as the mean time that a camer
will remain in a trap before being released back into the emended states (be. a trap dwell
time). The capture lifetime of a trap is inversely proportional to the concentration of
unfilled traps. Obviously, the carrier lifetime will be short if the trap concentration is
high, and vice versa Once a carrier is trapped, it will remain immobile until a lattice
vlihration imparts enough energy to the carrier to excite it back into the extended states,
where it can drift once again. Release h m shallow traps is appropriatety much faster
than release h m deep traps since low energy lattice vibrations are much more EeIy than
the high energy vibrations required to impart enough energy to h e a deeply trapped
carrier. The shallow trap release time is very short, and a typical carrier may experience
many shallow capture and release events while traversing the solid during, say, a timesf-
flight experiment. The deep trap reIease time, however, is very Iong, and a deeply
trapped carrier is essentially permanently removed fiom condution when considering the
timescaie of a typical time-of-flight experiment.
2.7 Optical Properties of a-Se
A photoconductor is simply a transducer for converting incident photons to
mobile charge carriers which can be collected and detected. An important measute of the
worthiness of a photoconductor is the optical absorption coefficient a Optical
absorption in crystalline semiconductors is dictated by the probability that a photon will
excite an electron across the bandgap and generate a fke electron-hole pair (EHP).
Therefore, a depends both on the incident photon energy and the DOS at the band edges.
If the photon energy is less than the bandgap, no absorption will occur.
The optical absorption coefficient of a-Se exhibits an Urbach edge of the form
a = 7.35~10-12e /an, corresponding to excitation of carriers h n the midgap
localized states into the extended states [46]. At higb photon energies, the absorption
coefficient has been found to obey 4hv)-(hv-k) [47], where E, = 2.05 eV is the optical
bandgap at room temperature. This behaviour has been attniuted to a sharp rise in the
density of states at the band edges.
Even though the optical absorption coefficient shows considerable absorption at
photon energies above 2 eV, the quantum efficiency has been found to exhibit a strong
fieId and photon energy dependence, as shown in Figure 2.9. The quantum efficiency
determines the probability that optically generated EHPs will dissociate in the presence of
an applied field to form free electrons and holes. The mechanism behind the fieId
dependent quantum eficiency observed in a-Se bas been explained by the Onsager theory
[17]. The Onsager theory calculates the probability that an EHP will dissociate in the
presence of an appIied field. The quantum ef6cimy is a hct ion of the eIectric field F,
the temperature T, and the initial separation of the EHP r,, the thermalization length.
Thus, the quantum efficiency is dictated by
v =%f (F,T,~), 2.1
where f (F,T,~) is the probability that an EHP will dissociate and q. is the quannun
efficiency of the intrinsic photogeneration process.
Photon Energy h v (eV)
Figure 2.9. Absorption coefficient a (solid h e ) and quantum efficiency q (broken lines) as a hction of the incident photon energy hv for various applied electric fields [17,46].
For crystalline semiconductors, the quantum &ciency is largely detesmined by
recombination kinetics and is generally independent of the electric field As Figure 2.9
shows, this is obviously not the case with a-Se (and many other low mobility solids).
Light exposure has been observed to induce changes in the structure of
amorphous semiconductors such as a-Si:H (amorphous hydrogenated silicon) and a-Se
[36, 37, 48-51]. Specifically, in a-Si:H, light will induce under-coordinated Si atoms
which then act as efficient recombination centers. This causes problems regarding, for
example, solar pane1 efficiency. The nature of the photoinduced structural changes in a-
Se are thought to arise through two different mechanisms: the formation of IVAPs and
the further randomization of the soIid through bond twisting and the resultant relocation
of Se atoms. Photoinduced NAPS wodd affect the lifetime of photoinjected carriers
measured via the ETOF technique, since they would alter the number of native defects
thought to be responsible for the deep traps in the solid. Conversely, the M e r
randomization of the solid wouId alter the tail states in the gap, possibly leading to an
altered W e r mobility or increased carrier dispersion. Any observed changes in the
carrier transport parameters in a-Se upon exposure to x-rays would have to be interpreted
with these mechanisms in mind.
2.8 Summary
This chapter introduced the physical structure and properties of amorphous
semiconductors in general, and amorphous selenium in particular. It is widely thought
that the disorder found in an amorphous semiconductor leads to localized tails of energy
states in the bandgap, and that defects in the material lead to donor- and acceptor-like
states within the gap. These localized gap states govern the electrical conduction
properties of the material.
Charge transport in a-Se is dominated by the presence of shallow traps near the
band edge. This reduces the overall mobility of the charge carriers through multiple
trapping and reIease events fhm these shallow traps. In addition there are distriiutioos
of deep electron and hole traps Iocated near the midgap which can permanently remove a
carrier h m conduction.
While the optical absorption co&cient of a-Se is dependent on the energy of the
incident photons, with an optical gap of - 2.05 eV, the quantum efficiency is dependent
on both the photon energy and the appIied electric field. This field dependence has been
explained by the Onsager theoy, as a consequence, the quantum efficiency only reaches
acceptable levels at high electric fields. In addition, light has proven to induce defects in
a-Se; however, how those defects alter charge transport is not known.
3. The Time-of-Flight Transient Photoconductivity
Technique
3.1 Introduction
As stated earlier, the disordered nature of amorphous materials renders the
thearetical treatment of their charge transport difficult. As such, the study of amorphous
semiconductors heavily relies on measurements of their physicaI and electricd properties.
The Time-of-Flight (TOF) transient photoconductivity technique provides a powerful,
and proven, means of studying the nature of charge transport in low mobility materials in
general, and amorphous semiconductors in particular. This chapter presents the principle
of both the TOF and Intermpted Field Tie-of-Flight (IFTOF) transient
photoconductivity techniques, and the principles and theories involved in the
interpretation of the photocurrent signals for various transport and trapping conditions.
3.2 Principle of the Timesf-Flight Technique
The TOF transient photoconductivity technique simply involves the measutement
of the transient response caused by the drift of injected excess charge carrim through a
high resistivity solid, A simplified representation of the TOF measurement technique is
shown in Figure 3.1 (a). Consider a thin plate of a traphe material of thickness L
which is sandwiched between two metallic electrodes, A and B. Electrode A is
connected to a bias source of voltage V, while electrode B is connected through a
sampling resistor R to ground. In this particular instance, A is kept at a positive potential
with respect to ground. Free carriers are injected into the sample directly under electrode
A by some means of external excitation (i.e. visl%le light, energetic electrons, x-rays,
etc.), and these carriers are then induced to drift across the material by the applied bias.
This transient photocurrent may be monitored by observi~~g the voltage induced across
the sampIing resistor R. It should be noted that if visible light is employed as the
excitation source, then electrode A must be transparent, or semitransparent, if light is to
reach the sample material under test.
Figure 3.1. (a) Simplified schematic and (b) small siguaI ac equivalent for the TOF transient photoconductivity technique. Cs is the sample capacitance.
Analysis of the photocurrent is greatly simplified if only one species of charge
carrier (electrons or holes) are swept across the sample; this may be accomplished by
choosing an excitation source (for instance, visible light) which will be strongly absorbed
near the surface of the material. Proper selection of the photon wavelength to ensure that
the absorption depth 6 is much less than the sample thickness L will prevent bulk
generation of carriers, and hence only one species will be swept across the sample. kt the
present discussion, photogenerated electrons will be immediately removed b m the
sample by the positive potential at electrode A, leaving only photogenerated holes to drift
across the sample to electrode B. Electron transport may be studied by simply reversing
the polarity of the applied bias; in that case, photoinjected holes will be immediately
removed by electrode A while the photoinjected electrons will be fiee to drift across the
sample to eIecErode B.
Just as it is important to limit photogeneration to the region immediately beneath
the surface of the sample, it is equally important to ensure that the duration of the
extmal excitation t, be much Iess than the transit time tr of the injected charge carriers
across the sample. The reason for this particular requirement is that the initial width of
the sheet of photoinjected charge must be kept as narrow as possible in order to facilitate
an acceptable spatial resolution within the sample as the injected cbarge sheet can be
viewed as being akin to a measurement probe whose spatid resolution is limited by its
width w.
The material being studied via the TOF transient photoconductivity technique
must not have a high thermal equilibrium concentration of charge carriers as these could
potentidy recombine with and neutralize the photoinjected charge carriers. This
requirement is quantified by the dielectric relaxation time z d and its relation to the transit
time tr of the charge carriers under study. ~d represents the time required for any excess
(i-e. photogenerated) charge carriers in a material to decay to their thermal eqdi'brium
concentration, and in general rEi must be much longer than rr. By virtue of their high
resistivity, amorphous semiconductors have a low intrinsic charge carrier concentration at
room temperature which g e n d l y means that TA >> t ~ .
Interpretation of the TOF photocmt signal requires that the electric fieId F
within the material under study be constant-not just from one instant to the next-but
also at every location within the sample. Unfortunately, the photoinjected charge carriers
perturb the applied electric field Fa = VdL leading to an enhanced field, F?, in front of the
charge sheet, and a decreased field, Fl, behind the sheet, as shown in Figure 3.1 (a). The
application of Gauss' law and simple eIectrostatics leads to the following relations for Ft
and F2 at any position x' [52] :
and
Here, p, is the concentration of photoinjected charge carriers within the charge sheet, w is
the width of the sheet, and e is the dielectric permittivity of the sample. However, if the
amount of photoinjected charge p,wA (where A is the area of injection) is restrained to be
a small amount such that %wk << Fa, then the internal field at every point within the
sample can be approximated as being equal to the applied field F,, E F, = Fz. This is
known as the small signal condition, and it corresponds to Q, (injected) << V&. This
condition imposes a limit on the number of charge carriers which may be injected into a
sample during a TOF experiment. If there is no adherence to the small signal condition,
then the analysis of the TOF waveform must be modified to take this field e a t i o n
into account 1531.
Ramo's Theorem [54] dictates that the transient signal observed across the
sampling resistor R in the external circuit of Figure 3.1 (a) is due to the photoinjected
charge carriers moving from electrode A to electrode B in the materid. Consider a
charge being induced to flow in a material under the influence of an applied electric fieid
as in Figure 3.2. The work done in moving a positive charge q a &stance & is
dw = Fqdx, 3.3
where dW is the work done in moving the charge and F is the applied electric field. The
energy required to do this work must be supplied by the external source and is
dE = CdQ. 3.4
Therefore, dE must be equal to 6W; it then follows that
qdx m o ' s Theorem]. dQ=- 3.5 L
From Equation 3.5, it follows that for a trapfree solid, the induced photocurrent through
the sampling resistor is
F i e 32. The motion of a charge q through a distance dx in the sample induces a charge dQ to flow in the external circuit.
Figure 3.1 (b) depicts the small signal ac equivalent of the circuit of Figure 3.1
(a), where C, is a combination of the sample capacitance and any other stray capacitance
added by cables, electronics, etc. Obviously, the photocunent signal i&) must flow
through the pardel combination of the sampling resistance R and Cs. Since Cs will
introduce a frequency dependence into the transient response to the induced photocurrent,
there arises two distinct methods of detecting this photocurrent signal, based on the cutoff
fiequency of this RC circuit, since it is, in essence, a Iow pass filter. If V(s) and IpH(s) are
the Laplace transforms of the voltage signal and the p h o t o m t , it can be shown that
In the folIowing discussion, the bandwidth of the photocurrent signal is arbitrarily
defined as the reciprocal of the carrier transit time r ~ .
The first method of measuring the photocurrent signal relies on the bandwidth of
the photocurrent signal being much less than the cutoff fiequency of the parallel
combination of R and C,. This translates to RCs << t ~ , and the inverse LapIace transform
of Equation 3.7 becomes
Y(t ) = ~i~ @) for RC, << rr . 3.8
Equation 3.8 is known as the Emode signal because the observed output signal is directly
proportional to the induced photocurrent. In a trapfree solid, the Emode signaI d l
instantaneously rise to a constant level when the charge carriers are first injected and will
remain at that level until the charge sheet reaches the opposite electrode, at which point
the signal wilI fall to zero.
The second method of measuring the photocurrent signal relies on the cutoff
kequency of R and C., being much less than the bandwidth of the photocurrent signal
(RCs >> tr). Under this condition, the low pass filter formed by R and C, will integrate
the photocurrwt signal, leading to the inverse Laplace transform of Equation 3.7
becoming
Equation 3.9 is known as the V-mode signal, and it is the integral of the I-mode signal.
In a trap-fiee solid, it will rise linearly upon charge injection until the carriers reach the
opposite electrode, at which time it will flatten and remain constant,
The abrupt and obvious change in the I-mode signal when the carriers reach the
opposite electrode renders the determination of the transit time tr in drift mobility
calculations quite easy, and the I-mode technique is thus favoured for this purpose.
Traditionally, the V-mode technique was favoured for a determination of the mount of
charge that was injected into a sample, and it is sometimes referred to as the charge
transient technique for this reason. However, with the advent of digitizing oscilIoscopes
and powerfhl soha re analysis techniques, numerical integration of the I-mode signaI is
both simple and fast-elirninating the need for m m e m e n t of the V-mode signal.
Many semiconductors have deep traps situated in their bandgaps. A carrier, once
trapped, is effectively permanently removed fiom conduction given the large energy
barrier over which the trapped carrier must be thermally excited. The existence of these
deep traps can significantly reduce the number of free carriers in the charge sheet as it
drifts across the sample. For a given species of trap with a mean capture lifetime T~ and
assuming that carrier release firom the trap is negligible during the transit time of the
carriers, the number of h e carriers wiU decrease in an exponential manner. Equation 3.6
t&m becomes i @ ( r ) = e ~ ~ ~ to reflect the time dependence of the total number of tT
charge carriers. The resulting I-mode signal will then be modified to become
Similarly, the corresponding V-mode signal may be found by integrating the photocuxrent
signal according to Equation 3 -9 to yield
Equation 3.1 1 is the well known Hecht relationship which was extensively used to
estimate the trapping time of charge carriers fiom TOF photocurrent signals for many
years [55]. Figure 3.3 presents a comparison of photocurrent signals for a trap ftee solid
and a solid with a set of deep traps. Although the TOF technique may be employed to
measure the deep trapping lifetime of a material, the interrupted fieId TOF (IFTOF)
method is far better suited to the purpose.
Figure 33. Signals h m the TOF experiment where (a) is the Emode signal and (b) is the V-mode signal, Solid lines represent expected signals in a trapfree solid while dashed lines represent signals in a solid with deep traps only.
3.3 Principle of the Interrupted Field Time-of-Flight Technique
The ETOF technique is similar to the TOF transient photoconductivity method
with one obvious difference: in contrast to the TOF technique, where the charge packet
is allowed to traverse the entire sample without interruption or delay, the LFTOF method
will hdt the charge packet somewhere (virtually anywhere) within the bulk of the sample
by disconnecting the applied bias at the appropriate time. After some interval (called the
intermption time t ~ ) the bias is reapplied and this interrupted photocurrent is recorded.
Figure 3.4 depicts a typicaI IFTOF waveform and a representative TOF waveform for
comparison. The photocurrent within the sample is directly proportiond to the
concentration of fiee photoinjected carriers via i@(o = eppFO, where e is the electronic
charge, is the charge carzier mobility, p is the charge carrier concentration (or density),
and F, is the applied electric field. The magnitude of the photocureat immediately
before interruption is denoted il and immediately after intermption, i2. In the presence of
deep traps h m which the reIease time 7, is much Ionger than the transit time r~ the
concentration of free carriers within the packet is decreased by the presence of those deep
traps in the bandgap of the sample. Thus the ratio of i2/il may be written as [56-621
Figure 3.4. (a) A typical TOF waveform and (b) a typical IFTOF waveform. The interruption time is ti; the magnitude of the photocurrent immediately before and after intermption is denoted it and i2.
-'% i2/iI = e ? 3.12
where z is the effective trapping time of that species of deep trap. The determination of .t
is straightfornard: it may be determined h m the slope of a semilogarithmic plot of the
ratio of i2/i1 verms interruption time.
3.4 Transient Trap Limited Theory
Section 3.2 introduced the principie of the TOF experiment, but the expressions
deveIoped for the I- and V-mode transient responses of photoinjected w r i e s do not
reflect those found in amorphous semiconductors. Dispersion (spreading) of the canier
packet is appreciable as charge carrier kinetics in these materials are heavily influenced
by the large concentration of localized states within their bandgap. The transient trap
limited theory is developed in this section for two common situations: a monoenergetic
Level of traps and a distribution of traps.
3.4.1 Monoenergetic Trap Distribution
Consider a semiconducting materid as shown in Figure 3 -5 with a current due to a
photoinjected electron packet flowing within the sample. The number of fiee electrons
within a thin slice of thickness dx may increase due to the flow of electruns into the slice
or due to t h d release of trapped electrons within that slice. Ef the material is assumed
to have a high resistivity (as is the case with a-Se), then the effect of recombination may
be ignored, since the number of intrinsic charge carriers within the sample will be
negligible. This particular situation may be mathematically desm'bed by the following
continuity equation:
Here, n(x,,t) is the concentration of fiee electrons in the slice, J(x,t) is the net current
density flowing into the slice, n&,r) is the density of trapped eIectrons in the she, and e
is the electronic charge.
F i e 3.5. Current flow with trapping and release processes in a thin slice of semiconductive material.
The net current density J(x,t) is comprised of two components: the conduction
current JC{x,t) and the diffusion cumnt Jdx,t). The conduction current is due to the drift
of electrons under the influence of the applied field; this is expressed mathematically as
J,(x,t) = ew(~,f)F(x,t) . Here, is the microscopic mobility of the eIectr~ns and F(x.0 is
the appIied electric field. The diffusion crrrrent is due to spatial variations in the
concentration of charge carriers and is given by Jo(x,,t) = eD6n(x,t)/&, where D is the
diffusion coeficient of the electrons. The total current density is the sum of these two
components,
w x , t ) ~ ( x , t ) = e ~ ~ , n ( x , t ) ~ ( x , r ) + e.- . 3.14 &
The substitution of Equation 3.14 into Equation 3.13 leads to the following one
dimensional continuity equation:
A similar derivation may be made for the case of holes being the charge carriers.
Equation 3. I5 may be solved if the rate of detrapping is known. The rate equation, as the
expression Gndx,t)lSt is known, is determined by the difference in the charge trapping
and release rates. Given that rc and rr are the mean capture and release times for the
carriers hrn the traps, the rate equation becomes
where it is assumed that there is no trap saturation effect.
A number of assumptions may be made so that the simuItaneous solution of
Equations 3.15 and 3.16 for the fiee electron charge density n(x,t) is simplified. For
instance, the electric field for small signal TOF measurements is approximately uniform
at any time t or position x, as discussed in section 3.2. Thus, the electric fieId will not
vary with either time or position, so that the 6F(x.,t)/& term is zero. Another
simplification that may be made is to ignore the diffusion term in the continuity equation
since the magnitude of the diffusion current is usually considerably smaller than the
conduction current.
Initially, an impulse of No eIectrons is injected into the sample at time t = 0.
Mathematically, this initial condition translates to
n(x,0) = N06 (x,~) 3.17
and
%(x,0)=0 fo rx r0 . 3.18
Finally, the boundary conditions which account for the finite length of the sample are
also necessary to solve Equations 3.1 5 and 3.16. These are
n(x,t)=0 f o r x > l 3.19
and
Equations 3.15 and 3.16 have been solved by applying the initial and boundary
conditions above and by using Laplace transform techniques [63, 641. This leads to the
following expression for the k e electron charge density
where z = x l v , I& is the first order hyperbolic Bessel function, U(x) is the unit step
There are two components to Equation 3.21. The first texm quantifies that portion
of the injected charge in the packet that drifts unhindered across the sample. These
carriers do not endure any trapping and release events, but the number of these untrapped
charges decreases exponentially as they drift toward the opposite electrode. Any charges
that have been trapped and then released back into the conduction band at some time t are
represented by the second term in Equation 3.21. These carriers will obviously lag
behind the untrapped carriers (due to at least one trapping and release event), and will
contribute to an appreciable photocurrent beyond the unhindered carrier transit time
corresponding to t~ = LIW.
The concentration of free carriers within the sample at any time t may be found by
integration of Equation 3.21 over the length ofthe sample. There is no simple explicit
expression for the time dependence of the charge carrier concentration, so particular
(limiting) cases are used to evaluate the trapping parameters. The prediction of the
transient response of charge carriers during TOF experiments is not adversely affected by
these limiting cases, since these conditions may be achieved through proper selection of
such parameters as sample thickness, bias voltage and temperature. The I-mode current
response for both low and high field conditions is examined in the following sections.
3.4.1.1 Low FieId Case
The conditions T, << T, < t~ state that the capture time is much shorter than the
time it takes for the fastest (untrapped) wriers to traverse the sample, as given by tr =
WpJ. The release time fiom these traps is also less than tr. These partidar conditions
mean that the charge carriers will likely undergo many trapping and release events before
finally reaching the opposite electrode. It may be presumed that since the release time
h m these traps is less than the transit time of the carriers that these traps are shallow in
nature, and energetically lie very close to the band edge. When appIied to Equations 3.15
and 3.16, these restrictions cause the time dependence of the total fke electron
concentration to become very small (vanish) over a long t h e interval. By virtue of
conservation of charge, the total number of free electrons is given by
The I-mode photocurreat resulting fiom this special case may be found by substitution of
Equation 3.22 into Equation 3.6 to yield
e x =c =-- 't, +=r ips for f C ct err-.
t~ '=c +=r %
Equation 3.23 closely resembles the trap-he I-mode case presented in section
3.2, except that the transit time has beem increased by a factor of (T~ + T,)/G. This
implies a reduction in the effective carrier mobility from p,, to p, where p is defined as
The scalar 8 is referred to as the SWOW trapantroiled transport factor or mobility
reduction fhctor, When carrier mobility is reduced by a dwell time in the traps, the
transport mechanism is texmed shallow trapantrolIed transport. NegIecting the h e
derivatives of the h e and trapped electrons in Equations 3.15 and 3.16 has the
unfortunate r d t that an expression that rdates the relative amount of spreading of the
charge packet as it crosses the sample cannot be obtained. The spread of the packet
(dispersion) is significant (as evidenced by the long tails found in typical TOF
waveforms), and is due to the random nature of the trapping and release events. This
spreading is considerably higher than that due to simple diffusion done [65].
3.4.1.2 High Field Case
If the applied field is sufliciently strong, it is possible for charge carriers to cross
the sample without any trapping and release events taking place (tT < z,). The
photocment signal may then be separated into two t m s ; one for the drift of carriers for
t c tr, and the second for t > rr. The first term has been derived as [66]
The second term represents the response for those carriers that were trapped and later
released back into the conduction band, and most will cross the sample without becoming
trapped again. The photocurcent signal in this case is [66]
r~ e-x. fort > r, . ifi =-- 2 T J ,
3.42 Extended Trap Distribution
Analytically analyzing the motion of charges through a range of traps distributed
kt energy is far more complex than is the case for a monoenergetic set of traps. However,
Rudenko and Arkhipov [65] adcicessed the issue by considering a set of traps,
continuously distriiuted in energy, and characterized by q(E). The total density of traps
in the material is then
The total density of carriers in the system, n(i,t), may be broken into those &ers that
The carrier lifetime is dependent on a number of material factors; the caphrre
cross-section at energy E, a(E), the thermal velocity of the fiee carriers, ;, and the trap
density N, through r = . The capture coefficient is defined as the product of
the capture cross-section and the thermal velocity of the h e carriers, c ( E ) = ~ (E); .
The probability per unit time that a fiee carrier will become captured in a trap which lies
in the energy interval h m E to E + dE is then
If the trap release process is thermally activated, the probability per unit time that
a carrier which is trapped at an energy iE below Ec is released is given by
where 1/~,@) is the probability per unit time of release h m a trap at energy E, N, is the
energy density of conduction states, and AE is the height of the energy barrier over which
the wrier must be excited to become h e . The density of trapped carries may be
d e h d as
where p(x.t,E) is the density of trapped carriers in the energy range from E to E + dE.
The rate equation for the trapped carriers in the energy range from E to E + dE may be
written as
or by substitution of Equations 3.29 - 3.3 1 into Equation 3.32,
The rate of change of the trapped carrier density may be found by rearranging Equation
3.33 to get
If there exists a state of thermal equilibrium between fiee and trapped charge
carriers, the time derivative of Equation 3.34 will vanish (Le dnd = d%t = 0 ).
Therefore, the trapped charge density at energy E may be related to the concentration of
f?ee carriers through
The total trapped charge concentration may be found by substituting Equation 3.35 into
Equation 3.31; if Equation 3.28 is considered, then the total concentration of carriers
may be related to the concentration of free carriers:
If the trap-controlled transport factor 8 is defined as
then Equation 3.36 reduces to n/rx,t) = Bn(x,,r) since 8 cc 1.
One last piece is missing &om the puzzle in order to complete the analysis: the
one dimensional continuity equation. If the mutual coulombic repulsion of the carriers in
the charge packet is ignored (i.e. F is uniform throughout the material), and assuming
thennal equiliIbrium, it may be written as
Substitution of the approximation n h t ) = Bn(x,.t) into Equation 3.38 yields the foilowing
equation which is only dependent on n(xt):
where p = Op,, and D = OD,. Note how the reduction in the carrier mobility is similar to
the low field condition of a monoenergetic trap distribution.
As an example, consider the Gaussian trap distribution shown in Figure 3.6. The
distribution is described by
Figure 3.6. Sketch of a Gaussian distnhtion of shallow energy traps that lay immediately below the conduction band edge.
Consider the case where E, >> kT; the solution of Equation 3.37 Ieads to the
foilowing [65]:
Consequently, the carrier mobility becomes
where the mobility has a temperature dependent activation energy of E, = ExkT. On
the other hand, shallow trap distn'butions with a power law dependence on energy (as
found in a-Se) do not lead to a temperature dependent activation energy.
3.5 Summary
The principle of both the Timesf-Flight (TOF) and Interrupted Field TOF
transient photoconductivity experiments were presented in this chapter. Both the TOF
and IFTOF techniques are powefil methods for studying the charge transport parameters
of low mobility solids. The transient TOF waveforms were predicted for common
trapping conditions. The effect that an energetically shallow set of traps has on a species
of charge carriers is to lengthen the amount of time it takes the carriers to cross the
sample, and this is due to the average time the wriers dwell in the traps before being
released. Energetically deep traps, however, will effectively pexmanently remove charge
carriers from conduction, given the very large energy barriers over which they must be
excited in order to become h e .
4. X-ray Photoconductors
4.1 Introduction
The x-ray photoconductive material of a digitaI x-ray imaging system functions as
an x-ray photon-to-electrical charge transducer, and thus plays the single most important
role in the entire digitai x-ray imaging system. The properties of an ideal x-ray
photoconductor will be introduced in this chapter, and several different candidate
materials will be compared with the idea1 case. FindIy, some concepts necessary to
understand the x-ray sensitivity of a material are discussed.
4.2 Ideal X-ray Photoconductive Material
If one were able to find the perfect x-ray photoconductive material, it would have
a number of unique electrical and mechanical properties. A digital x-ray imaging system
relies on the absorption of x-rays that h'berate free charge carriers that may be collected
and then analyzed by some peripheral electronics. Since charge coUection is the
dominant electrical process being employed to detect x-rays, the x-ray photoconductor
must not allow a large dark current to flow. Ideally, the dark current should be zero in
order to maximize the signal-to-noise ratio ( S N R ) of the device. The dark current has
two components: one arising h m the injection of charge carriers at the electrodes, and
another caused by the thermal generation of free carriers in the bulk of the
photoconductor itself. It is weII established that a metallic electrode deposited directIy
onto the i?ee srzrface of a semiconductor can behave as a blocking Schottky contact
Therefore, carefd selection of eIectrode material may aid in reducing the dark current. A
relatively wide bandgap in the photoconductor will aid in reducing thermal generation of
free carriers, since thermal generation of h e carriers occurs from defect states in the
middle of the bandgap.
Again, sin= charge collection is the dominant electrical process being employed,
the x-ray photoconductor shodd liberate as many free charge carriers (electron-hole
pairs, EHPs) as possiile per unit energy incident on the material as x-rays in order to
maximize the detected signal. This translates to a low EHP creation energy in the
photoconductor itself.
Once the free EHPs are created within the x-ray photoconductor, none should be
lost; this ensures that the detected signal remains strong. Free E m s may be lost by two
methods: bulk recombination or deep trapping. Bullc recombination occurs when a
drifting electron and hoie meet and recombine with each other. Free charge carriers may
also become deeply trapped as described in detail in Chapters 2 and 3. Therefore, the
electron and hole Schubweg must each be much longer than the thickness of the
photoconductor itself ( ie. ptF >> L, where p is the drift mobility, r is the trapping time,
F is the electric field and L is the layer thickness) so that deep trapping of charges
becomes unlikely.
During fluoroscopic procedures the detector is continually irradiated and polled
by the accompanying electronics to form a real time image. Therefore, the longest carrier
transit time must be sufficiently short so that all charges in a pixeI are collected by the
time that pixel is next accessed.
In addition to the ideal electrical characteristics above, it would be desirous of an
x-ray photoconductor to strongly absorb x-ray photons within as small a detector
thickness as possibIe. The x-ray absorption of a material increases with increasing
atomic number, therefore, a materia1 with a high atomic number (Z) would be preferred
for this purpose. Strong x-ray absorption is necessary to make use of the maximum
number of x-ray photons that pass through the patient; the higher the absorption, the
lower the x-ray exposure for the patient consistent with an image of high quality. This
characteristic also has ramifications as regards to detector speed; the thinner the detector,
the faster charges are dected, and thus the faster its response.
Deterioration in the properties of the photoconductor with repeated x-ray
exposures should be absent or small so that overall detector performance is not adversely
affected.
Finally, the ideal x-ray photoconductor should be easily grown on or coated on to
any underlying electronics to form the detector itself. This must be done in large areas
(typically 30 x 30 cm and larger) and without damaging the underlying electronics (i.e.
an active matrix array).
To summarize, an ideal x-ray photoconductor should have the folIowing qualities:
Dark current should be absent or at a very low level. Low EHP creation energy in order to maximize detectable signaI per unit exposure (but not so low as to cause problems with thermal excitation). Carrier losses due to recombination or deep trapping should be absent or very low. Carrier transit times must be faster than the polling time to maximize detector speed. Material should be highly x-ray "absorbent"-nearly all x-ray photons incident on the detector should be absorbed within a practical detector thickness. X-ray induced material fatigue should be absent or be so s m d as ro have no effect on detector performance- X-ray photoconductive material must be easily grownlcoated over large areas without damaging any underlying electronics.
4 3 Practical X-ray Photoconductors
In the field of digital radiography, there is, unfortunately, no x-ray
photoconductive material that meets every idealized characteristic as set out in section
4.2. Some materids may be attractive for some properties, but davourable for others.
This section examines a number of different candidate materials for their suitability as a
practical x-ray photoconductor.
4.3.1 Amorphous Selenium (a-Se)
The reason why amorphous selenium finds quite widespread use as an x-ray
photoconductor is that, in many respects, it approaches the characteristics of an ideal x-
ray photoconductor. a-Se is perhaps the most developed of the possible x-ray
photoconductors; it has been extensiveIy studied since it was first used as a xerographic
photoreceptor in the 1940s through to the 1970s. The stabilized a-Se alloys in use today
have excellent charge transport properties, with typical hole and electron ranges (p
products) being 30x10' c m 2 ~ and 5x10' cm2/V rqectively for device grade a-Se
alloys [67]. At typical operating fidds (Le. 1 10 VIP), the hole Schubweg is 1 30 mrn
and the electron Schubweg is 1 5 mm. Since most a-Se detectors are at most - 500 pm
thick, these large Schubwegs will ensure that virhralIy no flee charges that are created by
x-ray irradiation will be lost to trapping.
a-Se is also favourable because of its very low dark current, usually on the order
of -1 nNcm2 at low elecaic fields with conventional metal electrodes. Recent work at
Noran& Advanced Materials 1681, has provided a method for producing a multilayer pi-
n diode-like a-Se detector s t r u m that r e d s in dark currents c 100 p ~ , ~ ' at fields as
high as 20 V l p .
X-ray induced fatigue of a-Se has not been investigated in detail prior to this
work, and electron lifetime was found to exhibit no apparent decrease with repeated x-ray
exposure.
The deposition of a-Se alloys onto substrates is fast and easy to perform using
simple vacuum deposition techniques [69], and deposition rates of - 2 @minute are
typical. The ease and speed with which large Iayers are deposited are distinct advantages
in favour of a-Se as an attractive x-ray photoconductor over competing materials.
Another factor in favour of a-Se is the fact that the substrate temperature during
deposition is kept low (- 60°C) and this will not damage any underlying detector
electronics (cg. an active matrix m y ) .
Where a-Se suffers somewhat in comparison to other materials is in two areas: x-
ray absorption and x-ray sensitivity (EHP creation energy). First, the atomic number (Z)
of selenium is 34; compared with other materials (cadmium zinc telluride Zff - 50, lead
oxide Z,E- 82), a-Se is a rather poor absorber of x-rays and a thick detector must be used
to absorb the same amount of x-ray radiation as compared to a thin layer of a material
with a higher atomic number. Second, the EHP creation energy of a-Se is highly field
dependent; it decreases with increasing field, but at typical operating fields (- 10 V / p )
the energy required to create a collected EHP is reported to be approximately 35 - 55 eV
over the diagnostic beam energy range [70-751. When compared to other competing
materials with EHP creation energies in the range - 1 - 10 eV, a-Se is not particularly
sensitive to x-ray radiation,
4.3.2 Hydrogenated Amorphous Silicon (a-Si:EI)
The attractive qualities of a-Si:H are that it possesses excellent charge transport
properties; it may be doped so that multilayer devices (like p-i-n structures) may be
fabricated, and it may be deposited in large areas. However, it also has a very Iow atomic
number which makes it a very poor x-ray absorber. This material is fi,uther hindered by
the fact that it requires high wbstrate and meding temperatures which would surely
cause damage to any underlying electronics that would be necessary in the detector. Add
to this the very slow deposition rate of a-Si:H, and it becomes a very unfavourable
alternative to a-Se.
433 Cadmium Telluride (CdTe)
CdTe has been widely used as a radiation detector for many years. It is attractive
because it is highly efficient at absorbing and converting x- and gamma-radiation to
mobile charge -a; a co~l~e~uence of its moderate atomic number (& - 50) and low
electron-hole pair creation energy of 4.5 eV [76]. Although bombardment with high
energy protons has been found to induce fatigue through charge trappiag in CdTe
detectors [77], it appears that exposure to diagnostic x-rays does CdTe little or no harm.
As opposed to a-Se, where both holes and electrons contribute to the
photocurrent, CdTe has very poor hole transport. This means that the Schubweg of holes
in CdTe is very small which makes it unlikely for them to cross even thin CdTe layers
without becoming trapped. Another issue is the rather high dark current, which is on the
order of - 10 nA/cm2. Further, the deposition of large area polycrystallim CdTe layers is
usually done by vacuum deposition techniques; they may be sputtered (which is very
slow) or thermally evaporated. As is the case with a-Si:H, the high substrate and
annealing temperatures which are necessary would undoubtedly damage any underlying
electronics present on the substrate.
43.4 Lead Oxide (PbO)
PbO is a polycrystalline semiconductor that has found use for a number of years
in both visiile Light and x-ray sensitive television tubes [78]. Since the & of PbO is
approximately 82, it is an excellent absorber of x-ray photons, but PbO is porous in
structure which offsets its high effective atomic number. Dark currents are capabIe of
being < 1 nNcm2 since PbO may be doped to form pi-n structures. The electron-hole
pair creation energy is reported to be 8 eV [76], which means that PbO is quite sensitive
to x-rays.
Growth of PbO layers tends to be difficult; they are d y vacuum deposited,
but relatively easy to manufacture PbO crystallites suspended in a resin binder have also
been successfblly investigated [79]. However, PbO will adversely react with air, causing
an increase in dark current and a decrease in x-ray sensitivity. Funhermore, thick PbO
layers will degrade h m prolonged x-ray exposure; this effect has not been noticed in
thin layers, but thin layers do not absorb d c i e n t x-ray radiation to make them feasibie
in medical imaging applications.
4.3.5 Crystalline Materials
There are a number of crystalline semiconductors which are popular materials for
use in radiation detectors and pulse height spectroscopy applications, notable among
them being germanium (Ge) and cadmium zinc telluride (Cdl-,Zn,Te or CZT). They are
highly sensitive to x-ray radiation, and in general have good charge transport properties;
however, a Ge detector must be cooled with liquid nitrogen before its charge transport
becomes acceptable. Their main drawback lies in the fact that they are crystalline and it
has so far proven not feasible to grow the very large crystals that are needed for flat panel
x-ray image detectors. However, there has been much recent work in development of
gamma-ray medical imaging systems based on many discrete CZT detectors or arrays of
detectors [80], and there exist commercially available SPECT (Single Photon Emission
Computed Tomography) imaging units by Siemens and Diguad, to name two.
Of all the materials discussed thus far, a-Se is the best choice at present since it
possesses many of the qualities of an ideal x-ray photoconductor, even if it is slightly
lacking in its x-ray absorption and sensitivity. Table 4.1 summarizes the properties of
these materials and compares them with the ideal case.
Table 4.1 A concise comparison of candidate x-ray photoconductive materials. I Desired I Ideal I aSe I aSkH I CdTe I PMI I
Property
X-ray smsitivity Carrier losses X-ray absorption X-ray
Material Low High
Low
ktigue Deposition
Hi&
None
Low Highonlyat high fields
Low
Easy
Modcraft
Small
Low High
Low
Vacuum Coating (Easy)
Low
Unknown
High High
Hi&
Easy, but high substrate
temperatures
Low Hi&
Low
Hi&
None
High
High
Easy, but high substrate
tempetanrres
Difficult
4.4 X-ray Sensitivity
The sensitivity of a material to x-rays is of major concern in the field of
diagnostic radiography since, in general, the more sensitive the detector, the less radiation
to which the patient is expased. Specifically, the x-ray sensitivity of a material is a
measure of the amount of energy required to Liberate mobile charge carriers in that
material, and depends on two interreIated processes: the absorption of x-ray photons, and
the creation of charge carriers by those photons. In order to calculate the x-ray sensitivity
of a given material, the amount of energy that an x-ray beam deposits in that material
must be known.
4.4.1 The Energy Absorption Coefficient
The linear attenuation coefficient of a material is used when dealing with the
absorption of visible (or near vislMe) light (usually by a semiconductor), and should be
familiar to the reader. However, at high photon energies, a term called the energy
absorption coefficient must be used to calculate the amount of energy that a beam of
photons (e.g x-ray photons) wiIl impart to a material.
The linear attenuation coefficient is a measure of the &action of photons that
interact per unit thickness of an attenuator (a photoconductive material in this case). For
example, if an attenuator is placed in an x-ray beam as in Figure 4.1, then the number of
photons transmitted by the atteauator.is given by
N = N,e-*, 4.1
meaning that
AN= N , ( L - ~ - " ) 4.2
photons interacted with it. No is the number of primary photons incident on the
attenuator, N is the number of photons transmitted by the attenuator, AT is the thickness
of the attenuator, and p is the Iinear attenuation coefficient of the material (the field of
nuclear science generally employs the symbol p as opposed to a with which most
engineers are familiar).
-c S J X .-
Scattered Photon S b _--- /
Incident Primary Beam I Attenoator Scattered Photon S
Figure 4.1 An attenuator is placed in an x-ray beam.
When an x-ray beam passes into an absorbing medium, several different processes
may occur, as detailed in Figure 4.2- The entire process is random; for instance the
photon scattered fhm the primary interaction at A in Figure 4.2 may or may not interact
Radiation enters attenuator in the form of a beam of x-rays
X Primary interaction occurs with A an electron
High speed electron giving
Ionization, excitation, breaking molecular bonds, heat
More like A & B
Figure 4 3 A number of different interactions are possiile when an x-ray photon enters a material.
with the attenuator again before leaving it, Also, the recoiling high speed electron from
the interaction at A in Figure 4.2 is not Iikely to transfer all of its kinetic energy to the
attenuator, as some energy may be lost (radiated) as bremsstrahtung (braking) radiation.
Whereas the linear attenuation coefficient is a measwe of the fraction of photons
that interact per unit tfiichas of attenuator, the energy absorption coefficient is a
measure of how much energy is absorbed by the attenuator through those photons that
interact with it (also per unit thickness of attenuator, as with the hear attenuation
coefficient). The attenuation of a photon beam produced by a layer of a given material
depends on the number of electrons and atoms present in that layer. It is for this reason
that the hear attenuation coefficient of a material is divided by its density, so that clear
comparisons may be made between the attenuation introduced by differat materials,
independent of density. This coefficient, represented by (p/p), is called the mass
attenuation coefficient and has units of m2kg, SimilarIy, the energy absorption
coefficient, W p ) , is also normally divided by the density of the material and also has
the unit mt/kg.
The energy absorption coefficient may be related to the linear attenuation
coefficient; imagine a beam of photons incident on an attenuator in which N photons
reach the tayer Ax. The number of interactions that occur in this layer is given by
n = p N A x . 4.3
If the average energy absorbed pcr interaction is c, then the energy absorbed in Ax
through interactions with photons of energy hv is
To summarize, the bear attenuation coefficient is a measure of the fraction of
photons that will interact with a given thickness of material. A number of diffnent
interactions are possible, and may or may not impart energy to the material. For
example, it is possible that a scattered photon resulting h m the primary interaction with
an electron in the material may exit the material without fUtther interacting with the
material. It is also possible (though not likely) that the high speed electron that recoils
fiom the primary interaction may Iose all of its energy through radiation (bremsstrahlung)
without imparting any energy to the material. Thus it becomes quite obvious that the
linear attenuation coefficient of a material cannot be used to caIcuIate the amount of
energy that an x-ray beam deposits in that material.
4.4.2 Energy Absorption and Detector Thickness
As in section 4.4.1, the absorption of energy depends on the atomic number Z and
the density of the material. However, the energy of the photons incident upon the
materid also strongly determines the energy absorption. Figure 4.3 shows the energy
dependence of the energy absorption coefficient of a number of common
photownductors.
Most of the energy absorption coefficients in Figure 4.3 exhibit sharp vertical
edges where the energy absorbed by the material suddenly increases sharply as the
energy of the incident photons climbs. This is caused by a strong onset of absorption
when an x-ray photon ejects an inner core electron, such as from the K shell, into the
conduction band. The incident x-ray photon imparts all of its energy to the electron in
this instance. This is known as the photoelectric effect, and is the dominant absorption
process for photons with energy below approximately 50 keV. Another possible
interaction between an x-ray photon and an electron will result in both an energetic recoil
electron and a scattered photon of less energy than the incident photon. This absorption
process is known as Compton scatteriug, and is the dominant process above - 50 keV
(and into the MeV range). The probability of Compton scattering decreases as the photon
energy increases, thus the decrease in the energy absorption weEcients in Figure 4.3 as
the photon energy climbs. Compton scattering is almost independent of the atomic
number of a material, but it is strongly dependent on the electron density of that material.
1 00
Photon Energy (keV)
Figure 4 3 Energy absorption coacient vs. photon energy for various photoconductive materials [8 11.
X-ray photoconductors are evaluated through comparison of their K and L
photoelectric edges and their absorption over ceaain photon energy ranges as to whether
they arr suitable for uses such as mammography (mean photon energy - 20 keV) or chest
radiology (mean photon energy - 40 keV). In general, higher absorption is preferred
since the minimization of patient dose requires that most of the x-ray radiation incident
on the detector be absorbed within it, and materiais which highly absorb x-rays at a given
energy are preferred over those which are less absorbing at that same energy. A rule of
thumb is that the absorption depth 6 = I l k must be less than the thickness of the
detector, L. Heace the required detector thiclcness depends on the photon energy and the
particular radiographic imaging application, If the required detector thickness is taken as
3~ then an a-Se detector wodd have to be about 80 pn for mammography and about
660 pm for chest radiology. The same two applications would require a CdTe detector to
be 160 and 180 pm respectively. Obviously it would be preferable to be able to use the
same detector for a number of imaging applications, and for this purpose CdTe would be
more attractive than a-Se as the x-ray photoconductor.
It should be emphasized that the detector cannot be made to be very thick;
whereas a thick detector would absorb more x-ray radiation, the fiee wriers generated
within it would have a correspondingly longer distance to travel in order to be colIected
and thus detected. Therefore, there is a compromise to be made when choosing an
appropriate detector thickness: one must balance the requirement for maximum energy
absorption which dictates a thick detector with the charge transport properties (the
Schubwegs) of the materid, which favour thinner detectors. The response speed of the
detector must also be considered if it is meant to operate in the fluoroscopic mode, which
requires that the detector be thin.
4.43 Electron-Hole Pair Creation Energy: WWIP
Since an x-ray photoconductor is essentially an x-ray photon-to-electrical charge
transducer, it is desirous that the photoconductor be as ef6cient as possible in converting
x-ray photons to mobile charges. Therefore, the amount of x-ray energy needed to create
a single, collected, electron-hole p a i r - W m u s t be as low as possible in order to
maximize the amount of detected charge AQ produced by incident radiation of energy
M.
When an x-ray photon enters a photoconductor, an energetic primary electron is
either ejected from an inner core shell via the photoeiectric effect, or is set in motion
through the result of Compton scattering (as discussed in section 4.4.2). This energetic
electron travels within the solid and will cause many additional collision ionizations
along its path (or track). Thw the initial interaction with an x-ray photon of energy of
several keV is capable of producing thousands of EHPs; a potentially very large signal if
all the charges are collected.
Klein [82] first showed that, for many semiconductors, the energy Ww to cTeate
an EHP depends on the energy bandgap E, of that material via what has become known
as Klein's rule: WWp = 2.8 Eg + Ephonon. The phonon energy term is expected to be small
(1 0.5 eV) so that WEHP is typically - 2.8 E,. Figure 4.4 shows this correlation between
WEHP and the bandgap energy E,; the solid line represents the WUIP = 2-8 Eg + 0.5 eV
behaviour f?om Klein's rule. There are many solids that are accurately represented by
&in's rule, but there are also a number of solids such as a-Si:H, PbIz and AgCI that
have WUrp values substantially less than that predicted by Klein's rule. It should be noted
that many crystalline semiconductors also exhibit an electric field independence in their
w,, value.
Alig and Bloom [83] were able to intuitively explain Klein's WEHP = 2-8 Eg rule
as follows. First they assumed that the masses of the primary energetic electron and the
masses of the secondary electron and hole that it creates are all equal. Just before the
collision, the primary electron has a momentum P and the kinetic energy threshold for
ionization is Er. From conservation of linear momentum, each particle has a momentum
of PI3 immediately after the collision. Since the kinetic energy of a particle is
proportional to the square of its momentum, each particle must have an energy of Ed9;
therefore, the minimum total kinetic energy of the EHP is then 2E#. Since the creation
of an EHP involves the excitation of an electron across a bandgap of energy E,
conservation of energy dictates that the total energy of the EHP, E, + 2Ed9, and the
recoiling primary electron, Ei9, must be equal to the kinetic energy El of the incident
primary electron just prior to the collision. Therefore,
Diamond 1 . 4 /
/,
y'
Klein's Rule / /'
\ Que and Rowlands
0 1 2 3 4 5 6
Bandgap (eV)
F i e 4.4. EHP creation energy vs. bandgap for a selection of photoconductive materials [8 11.
In tenns of the bandgap of the material, the minimum kinetic energy 2Ed9 of the
EHP is then E#3. The average energy to create an EHP, WEHp, must be equal to the
bandgap energy E, plus the average kinetic energy of the EHP. Alig and Bloom assumed
that the kinetic energy of the EHP may vary from 0 to EI; assuming that the density of
states varies as E:'* where Ek is the kinetic energy of the secondary electron or hole, then
where the factor of 2 accounts for the generation of two particles-the EHP. Substitution
of Equation 4.5 in Equation 4.6 leads to
Obviously a more rigorous derivation should include the effective masses of the
different particles and the effect of the generation of phonons in the impact ionization
process, but this fematkably simple derivation and expression accurately predicts byEHP
for a vast amber of (crystalline) semiconductors, as attested to by Figure 4.4. However,
some materials exhibit an electric field dependent WWp whose origin is still a point of
controversy.
Since amorphous materials are inherently disordered, Que and Rowlands (701
have argued that conservation of linear momentum in the ionization process may be
somewhat relaxed. This line of reasoning then Ieads to the lowest or saturated WmP,
W&, at very high electric fields: Wk = 2.2E, + Ea, where Epk is again a small
phonon energy term This is illustrated as the dashed line of Figure 4.4. Therefore, for a-
Se with E, = 2 2 eV, application of Que and Rowlands' theory would lead to wLP = 5
eV. The field dependence of Wm for a-Se and many other low mobiIity solids has
proven difficult to understand.
It is well known that the primary electron will generate many EHPs, but only a
tiaction of those EHPs will be collected. In general, the higher the electric field, the
more charge is collected. If it is assumed that practically no carriers are lost due to
trapping, as is the case for device quality photoconductive materid, then the losses can be
attrtiuted to onIy three sources: bulk or bimolecular recombination between drifting
holes and electrons, geminate (Onsager) recombination, or columnar recombination.
These three cases are illustrated in Figure 4.5.
' ' : m q n a r - A ', '\'\ \ Recombination
/--1 3 \ h L \ '>,
Electric Field
Figure 4.5. Schematic representation of the different types of recombination that are possible in an a-Se photoconductor. The cylinders represent the tracks of primary electrons. Bulk recombination occurs outside the tracks between charges that originated in different tracks. Geminate recombination occurs between the original hole and electron. Columnar recombination takes place between electrons and holes h m diffetent pairs, but within the same track.
In bulk recombination, the recombination rate is proportional to the concentration
of both species of charge carrier, therefore, the collected charge would not increase
linearly with the intensity of the incident x-ray photons-the collected charge would
exhibit a square root dependence on the x-ray intensity. However, since experiments
show that the amount of collected charge increases linearly with the x-ray photon
intensity, bulk recombination cannot be a factor in the charge loss mechanism in a-Se-
The simultaneously generated electron and hole it leaves behind have a strong
codombic attraction and may eventually recombine; hence the term geminate
recombination. This is the accepted model for the optical quantum efficiency of a-Se,
and the number of EHPs that escape geminate recombination is governed by the Onsager
model.
Columnar recombination involves the recombination of nongeminate electrons
and holes from within the columnar track of a primary electron. As the intensity of the x-
ray photons increases, the number of tracks also increases but these tracks rarely overlap
so that recombination within a track remains unaffected by the intensity of the radiation.
This means that the collected charge will increase linearly with the x-ray photon
intensity, in agreement with observations.
4.5 Summary
This chapter introduced the concept of an ideal or perfect x-ray photoconductor so
that a comparison of a number of different candidate materials could be performed to
assess their worthiness as an x-ray photoconductive material. a-Se is an attractive
candidate except for its rather poor x-ray absorption and field-dependent Wmp.
Although a-Se is not an ideal x-ray photoconductor, it is the best choice at present.
X-ray sensitivity was defined, and the various factors that affect the x-ray
sensitivity of a material were discussed. At present, there is no consensus as to how
EHPs, created by x-ray photons, are removed from conduction in a-Se. The two
competing charge loss theories, geminate and columnar recombination, predict different
temperature and x-ray beam energy dependencies.
5. Experimental Apparatus and Procedure
5.1 Introduction
A complete description of the equipment and procedures employed in the
measurement of the various charge transport and x-ray photoconductive properties of a-
Se are given in this chapter. The preparation of the a-Se samples is descriied first,
followed by the TOF/IFTOF apparatus and then the x-ray photoconductive measurements
are finally discussed. Some miscellaneous experimental tools are introduced last.
5.2 a-Se Thin Film Preparation
As mentioned in Chapter 4, a-Se is an attractive x-ray photoconductive material
because it may be quickly and easily deposited as a uniform film over large areas. The
film is grown by thermally evaporating seIenium pellets onto a conductive substrate
(electrode); the substrate is usually either oxidized aluminum or conventional glass
coated with a transparent indium-tin-oxide (lTO) conductive layer. The main difference
between a glass/lTO substrate and an aluminum substrate is in the amount of charge
injection that each wiU i n t m d u c ~ IT0 layer will inject considerably less charge into
the a-Se film under reverse bias than will an aluminum substrate. This point becomes
important when electrons are being studied, as the a-Se films must be reverse biased, as
first introduced in Chapter 3, This Iarge reverse bias dark w e n t made the study of
electron transport in samples with an aluminum substrate very difficult, and this point
will be explained in more detail in section 5.3.
Noranda Advanced Materials of Saint-Laurent, Quebec provided the necessary
materids for the films, including the electronic grade liquid quenched vitreous selenium
pellets. Optical emission spectroscopy was employed prior to delivery of the source
material to m e its purity, which is specified as being 99.999%. To coat a typical film
(a square of 3 cm x 3 cxn and - 60 pn in thickness) would require - 230 mg of selenium,
with more materid required to create a thicker film.
5.2.1 Substrate Preparation
Aluminum will spontaneously react with oxygen to form an electrically insulating
oxide layer - A1203. Tbis oxide layer is necessary to act as an insulator between the
substrate and overlying a-Se film to help prevent charge injection tiom the A1 electrode
into the film [84]. This oxide layer is also required because it f o m an amorphous base
for the a-Se f k . However, if the growth of this oxide layer is -sed, it will not
be uniform and its electrical characteristics will be unreliable. To provide some
protection against scratches, the duminum is coveted by a thin poiper coating. This
coating must be removed by ullrasonically cleaning the aluminum in a sequence of
acetone, distilled water, methanol, and W l y distilled water. A heated (65°C) caustic
etch solution of sodium carbonate, sodium phosphate, and distilled water is used to
partially remove any preexisting oxide layer before the sample is cleaned for a final time.
This 6nal cleaning consists of a nitric acid dip and repeated washings in distilled water
and detergent solutions. The oxide layer was rhea regrown under controUed conditions to
ensure its uniformity; this simply involved placing the substrate in a 300°C finnace for 4
- 5 hours. The preparation of the iTOighss substrates only required repeated washings,
as glass may be purchased with the IT0 conductive layer already in place. Once the
substrates were properly prepared, they were then ready for a-Se deposition.
5.2.2 Vacuum Deposition System
It has been previously reported [85] that the hole range ( p ~ product) of a-Se films
is strongly dependent on the substrate temperature during the sampIe
preparationfevaporation process. If the substrate temperature is held above the gIass
transition temperature of the a-Se alloy being deposited, the hole range in the a-Se film is
then maximized. Electron ranges are not particularly sensitive to the substrate
temperature and are mainly dependent on the purity of the source material. To find the
glass transition temperature of the selenium source material, Differentia1 Scanning
Calorimetry @SC) measurements were performed prior to evaporation. For typical a-Se
alloys used in the course of this work, glass transition temperatures varied fiom 40 -
5S°C, depending on the heating andlor cooling rates used during the DSC analysis.
The a-Se films were evaporated through the use of an NRC 31 t 7 vacuum
deposition system, as illustrated in Figure 5.1. The chamber is evacuated first by a
mechanical vacuum pump, followed by a diffbsion pump to a high vacuum of - 10' Torr. The substrate@) are heated and maintained above the glass transition temperature
of the a-Se alloy being deposited. A thermocouple provides a means of monitoring the
substrate temperature. A molybdenum boat contains the selenium pellets, which are
melted by passing a large (100 - 150 A) ac current through the boat. The boat
temperature is maintained at - 270°C, which, like the substrate temperature, is monitored
by a thermocouple. A deposition rate of - 2 @minute is achieved at this boat
temperature. When the h ( s ) reached the desired thickness, both the boat and substrate
heaters were turned off and the h ( s ) were allowed to slowly cool under vacuum. The
completed h ( s ) were then allowed to age in the dark at room temperature (no Ianger
under vacuum) to allow their physical properties to stabilize. A few days is all that is
required for the aging period.
Substrate Heater i - - < M I Jar
,, ,' ,- , ,, Substrate ,
Moveable Shutter I
' , </' f' ,/ , , , -. ' - - . d
Gas Inlet ----A-
Thermocouple - Molten Selenium r -
/' 1 ,*
' , MolyWenum Boat / '
.' /
' , Canying lSOA
v Diision Pump
Fie 5.1. Schematic diagram of vacuum deposition system.
5.23 Transparent Electrode Deposition
As previously stated in Chapter 3, TOF/l'FTOF studies necessitate a transparent
electrode through which some external excitation (visible light in this study) may be
eransmitted to the a-Se layer underneath the electrode. A Hummer VT sputtering system,
depicted in Figure 5.2, was used to sputter transparent metallic electrodes onto the
surface of the a-Se films. TOF studies have been proven to be independent of electrode
material [8q, so the choice of contact materid was not critical to the work. Gold (Au)
was chosen as the electrode material because the transparency of the electrode is
relatively easy to control; in addition, a suitably transparent Au electrode is also
acceptably conductive (meaning that the resistance of the electrode itself is quite iow)
and may be mated in a relatively short time.
Wmum Chamber /----- Insulation
/----- Cathode
CahodeShisld ,/ -- - Au A t m s - '.
' . . . '* Gold Target
- Anode
Figure 5.2. Schematic diagram of the electrode sputtering systesn.
To place a transpareat Au eIatmde on a sample, an aluminum mask having a
firevlar apmrre with an area of eithe~ 0.5 or 1.0 cm2 was first placed in contact with the
sample and secured with ordinary househoId scotch tape. The sample and mask were
then loaded into the vacuum chamber of the Hummer VI sputtering system. Evacuation
of the chamber could take upwards of - 45 minutes, during which time the vibration of
the mechanical vacuum pump would shift the position ofthe mask were it not secured to
the sample with scotch tape. Once evacuated of air (to a pressure of - 50 mTor), the
chamber was flushed with argon (As) for several minutes to ensure that little air remained
in the chamber. The Ar needle valve was then adjusted to maintain a chamber pressure
of I00 mTorr, at which point a large dc bias (3 - 4 kV) was applied between the anode
and cathode. This causes the Ar to ionize and form a plasma; the positively charged Ar
ions are accelerated toward the cathode where the Au target is also [mated. The collision
between these Ar ions and the target causes Au atoms to be dislodged; once disIodged
they settle on every exposed surface in the chamber. The unmasked portion of the a-Se
sample is coated in this fashion and thus forms the electrode. An acceptable electrode
would require a deposition of - 12 minutes at an Ar pressure of - I00 mTorr and a
plasma current of - 13 rnA. The side-to-side resistance of an acceptable contact ranged
from - 30 - 300 R.
Electrical connections to the substrate and top electrode for either TOFAFTOF
studies or for x-ray photoconductivity studies were made in one of two equivalent ways.
Contact could be established through a thin wire bonded with high purity silver paint (SPI
#5001), or by pressure contacts. Although either method is acceptable, the contacts made
with silver paint were the easiest to use, as the pressure contacts were at times difficult to
establish an acceptable electrical connection.
During TOFIIFTOF measurements, the photoinjected charge packet must be
confined to the area directly beneath the transparent electrode. If charge is photoinjected
in a region adjacent to the transparent electrode, it will drift under the influence of the
weak fXnging field at the edge of the electrode, leading to problems in interpreting the
TOFATOF waveform. During the course of this work, heavy black paper with a small
circular aperture was used to expose only the transparent electrode itself to the laser
pulse.
53 TOFAFTOF Apparatus
The TOF transient photoconductivity technique was pioneered in the late 1950s
and early 60s by Brown [87], Spear [88], and Kepler [89]-who used it to study the
electrical characteristics of a number of different materials. The TOF transient
photoconductivity technique is attractive because it allows the direct measurement of
either the electron or hole drift mobility of the material under scrutiny. As first
mentioned in Chapter 3, the TOF technique consists of measuring the transient
photocurrent that results when photoinjected charge carriers are induced to drift across a
high resistivity solid under the influence of an externally applied electric field-
A pulse of light of very short duration is the most common method of injecting
charge carriers into the sample; as such, care must be taken when choosing the
wavelength of the light pulse so that it is strongly absorbed by the material under study.
A number of different light sources have been used by different authors, including xenon
1901 and N2 [91] flash lamps to study a-Se, nitrogen pumped dye Iasers to study silicon
backbone polymers [92] and Q-switched ruby lasers to study organic polymers [93].
Variations of the versatile TOF technique have been employed to study a number
of different charge transport characteristics. For example, advance application of the bias
bas been used to study the effect of negative bulk space charge on the transport
characteristics in a-Se films [94]. Delaying application of the bias until a h charges are
photoinjected allows the study of charge carrier surface recombination [95]. Finally,
double injection (of holes on one side of the sample and electrons on the other) has been
used by Dolezalek and Spear [96] and by Haugen and Kasap [97] to study bulk
recombination in orthorhombic sulphur crystals and a-Se respectively.
The TOF/IFTOF apparatus employed throughout this work is illustrated in Figure
5.3. MOSFET high voltage switches apply the bias to the sample and connect the
floating voltage follower to ground and to the oscilloscopes; photoinjection is achieved
by a nitrogen laser. The transient photocurrent is captured and displayed by two
osciIloscopes: the andog scope captures the first portion of the IFTOF waveform (prior
to intemption) and the digital scope captures the second portion (immediately foUowing
the reapplication of the bias field). The entire process is contded by an FPGA-based
(Field Programmable Gate Array) timing generator which itself is controlled by a
computer. Transient photoconductivity measurements must be performed in the dark (for
obvious reasons) and the Pb cage thus served a dual purpose; to prevent operator x-ray
exposure and to isolate the a-Se sample being studied from any stray light in the room.
I X-ray Unit
Figure 53. Simplified view of the TOFFTOF apparatus. Two PCs were used, the 8088 PC connected to the CCD camera (analog oscilloscope) was necessary because the antiquated video b e storage card could not be used in the modern pentiurn PC.
5.3.1 Nitrogen Laser
A Laser Photonics LN 103C Transversely Excited Atmospheric (TEA) nitrogen
Iaser provided the light pulse that was used to photoinject charge carriers in the a-Se
samples under study. This laser provides a pulse with a duration of 300 ps at a
wavelength of 337.1 nm and a peak power of 250 kW. The light pulse from the laser was
coupled to an optical 6ber pigtail (Newport Optical Fiber adapter MM-2A) which was
fed into the Pb cage and held in place above the a-Se sample by a clamp.
As specified in Chapter 3, one requirement for the light source is that its duration
be much less than the transit time of the carriers across the sample. The thinnest sample
examined in this work had a thickness of 170 E.L~; this translates to a minimum possl3Ie
hole transit time of - 0.74 p (this minimum transit time is limited by the maximum
possible bias voltage of 3 kV h m the high voltage supply). Indeed, 300 ps is much less
than this minimum transit time. Another requirement for the light source is that it be of a
wavelength that is strongly absorbed near the surface of the a-Se sample. Light photons
of wavelength 337.1 nm have an energy of 3.68 eV; from Figure 2.9 the absorption
coefficient of a-Se at this energy is - 5.5 x 10' Icm which corresponds to an absorption
depth of - 0.02 p. Therefore, the nitrogen laser easily meets the requirements of the
light source for the TOFmOF measurements.
The TEA N2 laser is based on a 3-level laser system where the laser transitions
occur between the &Iu (C-state) and the B%, (B-state) electronic states of the Nz
molecule, as shown in Figure 5.4 [98, 991. A gas discharge excites electrons h m the
ground to the C-state by extmal electron impact collisions. The lifetime at the C-state is
relatively short (- 2 ns) at atmospheric pressures, while the B-state is metastable. This B-
state metastability prevents a nitrogen laser from continuous operation; as a result, TEA
N2 lasers may only be operated in a pulsed mode. An electrical discharge is employed to
achieve the rapid population inversion between the C- and B-states.
15 -
Electron Collisions - ' 337nm
Nuclear Separation (Angstroms)
F i e 5.4. Potential energy culves for the lowest triplet states in the N2 molecule 1981.
The Blumlein excitation method, pictured in Figure 5.5, is used to excite the Nz
molecules and produce a laser discharge. Initially, CI and Cz are charged to a high
potential (- 10 - 11 kV for this work). The laser channel electrodes are shorted by an
inductor. When the spark gep is triggered, C1 is rapidly discharged through the gap and a
damped oscillation occurs in the right hand loop of the circuit, causing a voltage reversal
to appear on Cr. This causes an overvoltage-and electrical breakdown-in the laser
channel. Discharge occurs very quickIy after the gap is triggered (on the order of -
Charged to High Voltage w
I' '., Spark
' . G a p laser v -: Head <.
A 4'
Figure 55. B I d e i n circuit used for rapid excitation of N2 laser.
The LN 103C Iaser is configured to be triggered by two separate logic high 'TTL
pulses. The firs& pulse (of at least 100 ns duration) to the Trigger Reg input will charge
the excitation circuit of Figure 5.5. A second pulse (of at least 100 ns duration) to the
Trigger Low input must fallow the first pulse between 30 and 50 ps later to fire the laser.
If the second puke does not arrive within 50 p of the ht, the laser will self fire. If the
spark g;ap is optimally adjusted, h e LN 103 C's operating manual claims that the jitter
between the arrival of the Trigger Low pulse and the firing of the laser is - 2 ns. Figure
5.6 depicts the Trigger Reg and Trigger Low timing requirements of the laser.
53.2 MOSFET High Voltage Switches
The single lagest impediment to performing UTOF measurements is the Iarge
transients that appear across the sampling resistor when the bias is applied to and
removed h m the sample under study, as pichued ia Figure 5.7. These transients are
problematic because they could potentially be sufficient to harm the voltage follower
amplifier and/or the osciUoscop, and they can obscure the small photocurrent si@.
Past work [58, 62, 100, 1011 usad various means to null the effect of these transients;
among them a Schering4ype bridge, a complimentary pulsed bias technique, and a
resistance ratio-type bridge. These methods had numerous drawbacks; the
comphentary pulsed bias technique was only useM for low bias voltages and the
bridge techniques required carell and tedious balancing of the bridge. The act of
baIancing the bridge could take several hours to accomplish; very often, some samples
simply could not be balanced.
Laset Self Fires --- Firing Windm -
Trigger Reg
Trigger Lm
Figure 5.6. Trigger timing requirements of the LN t 03 C nitrogen laser.
This work employed a MOSFET high vaitage switch to simply avoid the
switching transients induced in the sampling resistor. The high side switch applied and
removed the high voltage bias to the sample, and the low side switch
comecteddiswmected the floating reference of the voltage follower to ground and its
High Side Switch
F i e 5.7. Application and removal of the high voltage bias during the IFTOF experiment induces large switching transients across the sampling resistor.
output to the oscilloscopes. Slightly delaying (usually on the order of a few p) the
reconnection of the voltage follower after reapplication of the bias allowed the transient
on the sampling resistor to decay to a level that would no longer threaten the voltage
follower or oscilloscopes. Although a slight switching transient was still superimposed
on the photocurrent signal, this could be digitally subtracted through software: a
waveform without photoexcitation would be recorded and then subtracted fiom the
recovered photocurrent to yield a transient-fiee waveform. This method introduced a
sIight "dead time1' into the measurement of the second part of the IFTOF photocurrent; a
period of time following reapplication of the bias to the sample when the photocurrent
signal could not be recorded. However? this dead time effect wuld be minimized by
choosing a relatively low bias field so that the carrier transit time would be relatively
long compared to this dead time. For example, electron transit times for the samples
analyzed in this work were typically chosen to be - 500 p, while typical dead times were
- 10 - 30 p, depending on the sample, the polarity of the applied bias, and the value of
the sampling resistor. With the dead time much less than the carrier transit time,
interpolation of the magnitude of the photocurrent (i3 at the reapplication of the bias is
straighdbrward.
There were two requirements for the MOSFET high voltage switches: they had to
be able to switch 3 kV (the maximum available from the EG&G Ortec 556 high voltage
supply) in less than 100 ns. MOSFETs rated at up to 1.2 kV were available when the
switches were constructed, but 1 kV models were more readily available at lower wst. In
order to switch 3 kV, at least three 1 kV MOSFETs co~ected in series were required.
However, this minimal configuration was rather dangerous: if one FET became
damaged, it would trigger a cascade failure of the other two and would likely damage the
rest of the test equipment. To provide some redundancy should a FET fail, six I kV
MOSFETs connected in series were chosen for each switching chain.
The switching speed of a FET is mainly controlled by one factor. the speed with
which the gate capacitance may be charged or discharged. Therefore, to achieve the high
switching speeds required, the gate of each FET had to be driven by a high current, low
impedance circuit. Optoisolators with an integrated npdpnp transistor pair were chosen
to drive each FET to facilitate this requirement. A detailed schematic of one MOSFET
with its optoisolator gate driver is found in Figure 5.8. This FETfdriver pair f o m the
basic building block of the two switches built for this work. The role of the optoisolator
is twofold: it provides the high drive current necessary to quickly switch the FETs, and it
also provides electrical isolation between the high voltages being switched and the timing
generator driving the switches themselves.
+6!! - - x o . 1 pF
-5 - - 9 v + -
/ Drain -,
-1 , - MOSFEr +-- I
0-- -=.. 4
nn ',4 A
Source
Figure 5.8. The MOSFEWoptoisolator gate driver pair that forms the basic building block of the switches built for this work. MOSFET: Motorola MTP3NlOOE n- channel enhancement-mode rated at 1000 V, 3 A, bn = 4 Q, Q* = 32.5 nC. Optoisolator: Hewlett Packard HCPL-3101 Power MOSFETflGBT Gate Drive Optocoupler rated at 5000 Vac isolation and 0.4 A peak output current. A calcuiated switching time of 8 1 ns is achievable.
In addition to connecting or disconnecting the bias voltage to tbe a-Se samples,
the high side switch also had to ground the sample during those intervals when the bias
was disconnected fiom it. Grounding the sample was necessary so that there would be no
externd bias field during the interruption time ti of the IFTOF experiment. If a bias
remained across the sampIe during this intermption time, the charge packet would
continue to drift in the sample under the influence of the external field and would not be
haIted. It is for this reason that the high side switch has two switching chains of six FETs
each to alternately apply the bias to the sample and to short the sampIe to ground when
the bias is removed. A schematic of the complete high side switch is presented in Figure
5.9. To simplify the operation of the switch, only one TTL trigger is used for control;
the trigger is used to control the state of one chain while an inverted copy of the trigger
controls the state of the other chain. The TTL trigger to SimultaneousIy connect the bias
to the sample and disconnect the sample h m ground is active low. When the trigger is
high, the bias is disconnected h m the sample and the sample is internally shorted to
ground. The high side switch had one additional layer of optoisolation to protect the
timing generator and pentium PC just in case of an equipment failure.
!TngprlLL*.CPI I I , h r p g r ~ r a l
~oprorsolstod I , m d ! opaPisorarod 61FEfPalr - - siFEf Pair & s : F l T Pair A -
To Sample *-2
Figure 53. Schematic of the high side switch.
A schematic of the Iow side switch is given in Figure 5.10. Like the high side
switch, it had to be buiIt to withstand transients that could potentidly reach 3 kV, but its
two chains--one to connect the floating ground of the voltage folbwer to true ground
and another to connect the output of the follower to the oscilloscopes-had to be
switched in unison. Like the high side switch, this was triggered by a single active low
TTL signal and it also had one additional layer of optoisolation in case of an equipment
failure.
D lopoibolatoc . .. LOp(oisolatod Z - 1 2 - Loptoisolator/ 2
FET Pair 'FET Pair FET Pair
Amplifier Fleeting Ground
V a l (4 V)
Figare 5.10. Schematic ofthe low side switch,
A chain of six FETs has an appreciable stray capacitance to ground of - 400 pF
when turned on. This is the main reason that a floating amplifier arrangement was used
in conjunction with the low side switch to feed the transient photocment signal to the
oscilloscopes. By floating the amplitier, tbe effect of this large stray capacitance was
minimized. The stray capacitance of the chain connecting the floating ground of the
amplifier to true ground could be charged quickly because the only limiting resistance is
introduced by the chain itself (- 40 R when on), The stray capacitance of the chain
connecting the output of the amplifier to the oscilloscopes was limited only by the output
resistance of the amplifier (- 300 Q), which meant that it couId also be charged quickly.
The high side switch has easily withstood its rated switching voltage of 3 kV on
numerous occasions throughout the work with no failures to date. The speed of the
switches is also faster than the specified switching speed of 100 ns. Figure 5.11
illustrates the unloaded "on" and "off' voltage transients of a 100 V dc bias. 10 - 90%
and 90 - 10% switching speeds are 52 ns and 28 ns respectively; well within the
specified limit. Due to the inductive leads of the switch, there is a small amount of
ringing present when the switch is turned on/ofK This ringing could conceivably cause
problems when the bias is reapplied to the sample during an IFTOF procedure. However,
this ringing extinguishes in - 300 ns, and this is negligiile when compared to the cartier
transit time of a typical IFTOF experiment, The switching times of the low side switch
are identical to that of the high side switch. Figure 5.12 is a photograph of the low side
switch.
5.33 Voltage Follower
The voltage folIower is necessary to drive the relatively large capacitance
introduced by the coaxial cables that comect the oscilloscopes to the sampling resistor.
A very simple battery driven follower based on the Harris HA-5033-5 Video Buffer is
pictured in Figure 5-13. The batteries, power switch, 0.1 pF bypass capacitors and f 12 V
voltage regulators have been omitted for clarity. The HA-5033-5 has a low output
0 200 400 600 800 1000
Time (ns)
F i e 5.11. Unloaded switching transients of the high side switch.
resistance of 5 GI and a wide small signal bandwidth of 250 MHz. However, since the
buffer was being used to drive the 500 kR input impedance of the two oscilloscopes (two
paralleled I MQ resistances to ground) instead of the low impedance matched load for
which it was designed, stability b&e a problem. The series 270 R output resistor
made the amplifier unconditiodly stable, but limited its bandwidth to 20 MHz. This 20
MHz bandwidth was d c i e n t for the work, especially when considering that previous
projects [loo, 1021 used voltage followers with bandwidths of less than 10 MHz. Figure
5.14 depicts the gain vs, Ereqwncy response of the voltage follower.
Figure 5.12. Photograph of the low side switch.
F i r e 5.13. Circuit schematic of the voltage follower.
5.3.4 FPGA Timing Generator
The success11 completion of a TOF or IFTOF measurement requires strict
control over the timing of the switches, the triggering of the oscilloscopes, and the fixing
of the laser. To control these devices, a timing generator was constructed fiom an Altera
UP 1 Education Board and a National Instruments DIO-3WS Digital InputlOutput card
mounted inside the pentium PC.
The Altera UP1 Education Board interfaces to the pentium PC via the parallel
port and may be programmed via Altera's Max+Plus II student edition software. The
evaluation board contains an EPFlOK20 FPGA and a 25.175 MHz crystal oscillator.
Figure 5.14. Gain vs. frequency response of the voltage follower.
Other components are present, but are not relevant to the following discussion. External
connections via headers on the UP1 Board were made to the DIO-32HS card (inputs for
the timer functions) and to the test equipment (?TL timing outputs). A simplified
diagram of the connections to the UP1 Board are given in Figure 5.15.
Anera UP1 Education Board
25.175 MHz Cfystal Oscillator
EPFlOKM FPGA
Configuration Information via Parahel Port
(M~X+PIUS II Software)
F i e 5.15. A simplified diagram of the connections to the Altera UP1 Education Board showing the flow of data into and out of the board.
90
The EPFlOK20 FPGA was configured via the Max+Plus [I software to function
as a liming generator based on a number of chained single shot counters. These counters
were loaded and initially triggered via the pentium PC by the DIO-32HS card. The
25.175 MHz crystal oscillator on the UP 1 Board functioned as the timing reference. The
chained counter structure that the FPGA is programmed to emulate is shown in Figure
5.16. The input data and clock signals have been omitted from each counter to relieve
clutter. Operation of the timing generator was very simple. LabView software running
on the pentium PC would conveniently load each counter in the FPGA via the DIO-32HS
card according to the TOFIFTOF timing requirements set by the operator. Once aH
counters were loaded with their appropriate values, the software would trigger (via the
DIO-32HS) the first counter to begin counting down; once it completed its countdown, it
- . . A-
Laset Rag - Laser R e g - Scow 1 Laser LOW -- k" , (4
scopa 1 Lassr Low - Amp I
TrigBer (dl Amp uf (I) L Amp-Bias
(0) L m - B i a s
a Bias On (h) 'I
A m ~ o f f O
A m - Bias
5.16. A block diagram of the chained counters responsibIe for generating the timing waveforms necessary to control the TOF/IFTOF apparatus, and the timing waveforms thus created (not to scale).
would simultaneously disable itself and trigger the next counter in the chain to begin its
countdown via an active high signal, and so on. These signals that each counter passed to
the next in the chain were used to trigger the switches, osciIIoscopes and the laser, as
shown in Figure 5-16.
5.3.5 Data Acquisition
Data acquisition was performed via two oscilloscopes; one analog, one digital. A
Tektronix 2467B 400 MHz analog oscilloscope was used to capture the fht part of the
IFTOF waveform. Those waveforms were then captured and stored in the 8088 PC by a
Tektronix DCSO 1 CCD Camera Data Acquisition System. The DCSO I system consists
of a Tektronix C1002 charge coupled device (CCD) camera mounted to the fiont of the
oscilloscope and connected to a PC DXOl video frame storage card which resided in the
8088 PC. The digitized waveforms were transferred to the pentium PC via floppy disk
for subsequent analysis.
A Tektronix TDS210 1GSaIs (1 gigasample per second) 60 MHz digitizing
oscilloscope was used to capture the second part of the lFTOF waveform. The
waveforms were then directly transfened to the pentiurn. PC via a GPIB bus connection to
the oscilloscope and a N a t i d Instruments AT-GPlBtTNT card mounted in the pentium
PC.
The andog oscilloscope was used in this work strictly out of necessity. The CCD
camera waveform acquisition system is both slow and inaccurate, and the 2467B was the
cause of a sizeable (-2 mV) noise signal in all the recovered waveforms. Further, the use
of the 2467B negated the porn'bility of measuring the electron transport properties of
samples with duminum substrates, as first mentioned in section 5.2. The reason for this
is as fcllows. For this work, electron transport was measured not by revetsing the bias of
the high voltage supply, but by reversing the leads on the a-Se samples, as illustrated in
Figure 5.17. This method is functionally equivalent to reversing the polarity of the
applied bias.
Figure 5.17. Sample connections for measurement of (a) hole transport and (b) electron transport.
Since the sampling resistor and the oscilloscopes are dc coupled by the voltage
follower when the low side switch is closed, any dc ofiet across the sampling resistor
arising due to a dark current in the a-Se sample will be passed to the oscilloscopes as
well. In order to not perturb the applied bias field, the injected charge packet must be
small, as first discussed in Chapter 3. Under these coaditions, the maximum size of a
photocurrent wavefonn is at most 10 - 20 mV, and must be displayed on a small
voltageJdivision setting (i.e. 2 mV/div) so that the overall resolution of the photocurrent
remains high However, at 2 mV/div, the adjustable vertical position (dc offset) of the
waveform on the 24678 is very limited. The electron photocurrent of a sample with an
aluminum substrate fkquently could not be displayed by the 2467B oscilloscope because
of this limitation.
5.4 X-ray Photocurrent (Wm) Measurements
The measurement of Ww is straightforward, provided that the amount of energy
that the x-ray beam deposits in the sample is known. In very basic terms, the procedure
involves biasing an eIectroded a-Se detector, irradiating it with x-rays, and monitoring the
resuItant photoamrent Wmp is then simply the ratio of the x-ray energy deposited in the
sample to the amount of charge collected in the photocurrent. However, the energy that
the x-ray photons deposit in the sample is rather difficult to determine because the output
spectrum of an x-ray tube is not monoenergetic, and is difficult to measure.
5.4.1 X-ray Exposure System
X-radiation is electromagnetic radiation produced either when hi@ speed
electrons are rapidly decelerated, which results in the production of bremsstrahlung
(braking) radiation, or when an electron transitions from a high to a low energy state in
an atom, which results in characteristic (monoenergetic) radiation. An x-ray beam
produced by an x-ray tube contains both types of radiation.
A typical x-ray tube is pictured in Figure 5.18. An x-ray tube is a temperature-
limited vacuum tube diode; electrons are emitted by a hot filament cathode and are
accelerated by a high (- 100 kV or more) voltage to strike a small focused area on an
anode. The anode is typically fabricated from tungsten or a tungsten alloy because only
about 1% of the energy of the electrons is emitted as x-ray photons (1031. The rest of the
Vacuum Tube \ ,
I \
Tungsten Anode j 1.. Farget)
i-----'
- - -
- -
\ Fdcusing Cup !,'
Figure 5.18. Schematic diagram illustrating the major components of a rotating anode x-ray tube.
energy is dissipated as heat in the anode-thus the need for a high melting point materid
such as tungsten. Heat dissipation may be aided by a rotating anode assembly such as
that pictured in Figure 5.18, so that the electron beam does not burn one spot on the
anode.
X-ray photons are emitted fiom an x-ray tube when electrons are absorbed in the
anode; their energy depends on the kinetic energy of the incident electrons. Like the
visl%le light photons produced by an ordinary incandescent light bulb, the x-ray photons
produced by an x-ray tube are not monoenergetic and consists of a continuum of photons
with different energies. Bremsstrahlung or braking radiation is produced by the
deflection of an energetic electron by the nuclei of the anode (target) atoms.
Bremsstrahlung radiation has a wide distniution with an upper limit equal to the energy
of the incident electrons. Characteristic radiation results when the vacancies in the inner
shells of an atom (produced by the collision of an incident energetic electron) are filled
by outer shell electrons. The energy that the outer shell electron loses when it fills the
inner shell position forms well defined characteristic peaks in the output energy spectrum
of the x-ray tube. The various mechanisms leading to x-ray photon production are
illustrated in Figure 5.19.
The various x-ray spectra employed during the course of this study were produced
by a Gendex GX-1000 dental x-ray unit. This particular model had an adjustable tube
voltage, variable h m 50 - 100 kVp. The tube current had two settings: 10 or 15 mA.
The output x-ray intensity of this unit is not constant with time; the output occurs in
1/120h second bcspikes" corresponding to the self-rectifying nature of the tube and the 60
Hz suppiy voltage. The unit is equipped with an impulse timer which may be set to
output anywhere fiom 3 spikes to an output train of spikes of 5 second duration. The x-
ray head itself is equipped with 2.7 mrn of inherent A1 fltering and a threaded collimator
which could be removed if needed. The x-ray head was mounted inside a coxnmercidly
produced lead-lined cabinet for safety.
Characteristic Incident Electrons
F i e 5.19. Typical electron interactions with a target. (a) Electron suffers ionizationid losses, giving rise to delta rays and heatt (b) Electron ejects K shell electron leading to characteristic radiation. (c) Collision between nucleus and eiectron of energy E leads to bremsstrahlung radiation of energy hv. The electron recedes tiom the collision with energy E - hv. (d) Collision where electron is completely stopped by a collision with the nucleus. The 111 energy of the electron is released as bremsstrahlung radiation.
The output x-ray photon spectral distribution is affected by three factors: tube
c m t , tube voltage, and filtration. The number of energetic electmns emitted tiom the
cathode is dependent on the filament current; the higher the filament current, the higher
the number of electrons ejected from the cathode and accelerated to the anode (i-e. the
tube current). The higher the number of electrons that strike the target, the more x-ray
photons emitted. As the tube voltage is increased, the maximum energy of the output x-
ray spectral distniution is also increased. This is due to the increased kinetic energy of
the electrons impinging the anode. The resulting beam will also have a higher mean
energy than one produced with a lower tube voltage (all other factors being equaI). The
spectral components of an x-ray beam may also be changed by adding filtration- to the
beam. In general, higher energy photons of an x-ray beam will be less affkcted by a
given thickness of filter material (eg. Al, Cu, Pb, etc.) than lower energy photons. Thus
filtration will preferentialIy pass higher energy photons and block lower energy photons
and will have the overall effect of increasing the mean energy of the beam (again, all
other factors being equal).
5.4.2 Pulse Height Spectroscopy
A major chdenge in determining the energy deposited in the a-Se detector layer
was the determination of the exact spectral distnition of the x-ray beam. One approach
is to formulate an estimate of the effective energy of the beam through a measurement of
the halfvalue layer (HVL) of the beam [73,102]. The estimate of the mean energy of the
beam, in conjunction with a measurement of the exposue (defined as the amount of
ionizing radiation required to Iiberate 2 . 5 8 ~ 1 0 ~ C of charge in 1 kg of dry air) of the
beam is used to calculate the energy that the beam deposits in the a-Se samples. This is a
very crude method and is susceptible to error. Another very involved approach is to
measure the spectral components of an x-ray beam by employing an x-ray diffraction
device [104]. This method does offer an accurate means of measuring the spectrum of an
x-ray unit, but was unsuitable for this work because of a lack of equipment, the energies
involved, and the contined space inside the x-ray cabinet.
The method of measuring the spectral components present in the beam of the x-
ray unit utilized in this work involved the use of a commercially available pulse height
spectroscopy measurement unit. The basic premise of pulse height spectroscopy is very
straightforward: a crystalline material with a well defined, field independent, EHP
creation energy is used as the detector. This crystal is biased at some field, and the
photocurreat resulting fiom any incident radiation is fed to an analog-todigital converter
(AID), as seen in Figure 520. Since the detector material has a well defined field
independent Ww, ,the amount of fiee charge carriers h i e d by an incident photon will
be directly proportional to the energy of that photon. The amplitude (or height) of the
resulting photocurrent "spike" will in turn be directly proportional to the energy of the
incident photon, and the AID channel number to which the pulse height corresponds will
also be directIy proportional to the energy of that photon. A typical measurement
involves a summing of the number of pulses that fall into the various A/D channels.
-, -.. - Radiation Source
Intermediate Energy Photon /' .
- Lowest Energy Photon
\ / Highest Energy Photon \
Detector -- -
b
To Amplifier@) 8 AID Photocurrent Pulses
F i 530, The principle of pulse height spectroscopy. The photon with the highest energy is emitted from the radiation source k t and strikes the detector first, generating the highest photocurrent pulse. The process is repeated, in turn, for the lowest and intermediate energy photons, which create the lowest and intermediate photocment pdses, respectively.
Therefore, a determination of the energy (spectral distriiution) of the radiation striking
the detector simply involves counting the number of photocurrent spikes with a given
height. By graphing the cumulative number of pulses in each A/D channel vs. the A/D
channel number, a graphical representation of the spectral distriiution of the radiation
impinging the detector may be formed.
The pdse height spectroscopy unit used during the course of this work consisted
of an eV Products Model 180 CZT detector, an EG&G Ortec Model 297 CZT Detector
Probe Amplifier, and an EG&G Ortec DART MCA (muitichannel analyzer) Portable
Gamma-Ray Spectrometer equipped with MAESTRO for Windows control software. A
photograph of the unit is shown in Figure 5.2 1. The amplifier and DART unit both had
to be wrapped in - 1 cm of Pb sheeting (as seen on the photograph) to prevent any
unforeseen and undesired effects arising from any stray x-ray photons interacting with the
sensitive electronics inside the units. The only component which was exposed to x-ray
radiation during the measurements was the CZT detector itself.
Figure 531, A photograph of the pulse height spectroscopy measurement unit used in the course of this work. The photograph on the left shows the DART A/D unit and the Pb shielded amplifier and detector. The photograph on the right is a top view of the amplifier and detector alone, with the top portion of the Pb shielding removed to reveal the cylindrical detector atop the ampfifier.
Pulse height spectroscopy was originally developed to be used in "low count"
situations-situations in which the x-ray or gamma radiation striking the detector is
rather infrequent so that the resulting photocunent pulses are well separated in time and
may be easily ddineated. Such applications usually involve the detection and
identification of radioactive isotopes in various materids or locales. When the arrival
time between each photon that strikes the detector becomes smalI, the photocurrent pulse
due to each photon will then add; a phenomenon known as pulse pileup (PPU). These
additive pulses can appear to have been caused by a single photon with an energy equal
to the sum of the energies of the individual photons. Even though the measurement unit
was equipped with circuitry to detect and reject severe PPU conditions, this one
phenomenon proved difficult to overcome in the spectral measurements.
5.43 Spectral Measurements-Procedure
Because the pulse height spectroscopy unit is specifically tailored for low photon
fluences (intensities), using it to measure the output spectrum of the x-ray unit was very
difficult. The x-ray unit is designed to output a great number of x-ray photons, and the
number of photons that reached the detector had to be significantly decreased so as to
reduce PPU. Figure 522 is a plot of the measured spectrum of a single 10 impulse burst
of the x-ray unit without any filtering or apertures placed between the tube head and the
CZT detector. The tube settings were 50 kVp and 15 mA; the lower energy cutoff of the
detector was set to approximately 15 keV. As can be seen h m the plot, the measured
spectrum approximately exponentially decays with increasing energy, but the spectrum
has components that extend all the way up to about 440 keV (the upper limit of the
detector). Since the tube voltage was set to 50 kVp, it is physically impossible that the
tube emitted photons of energy greater than 50 keV. As was discovered, extensive A1
filtering, occasional f b filtering and the use of very small apertures to limit the number of
x-ray photons that reached the CZT detector were necessary to limit (but not completely
eliminate) the effect of PPU.
It was first necessary to determine how much Pb shielding was necessary to
completely block ail x-ray photons h m reaching the detector. Trial and error indicated
that - 9 mrn of Pb was necessary to block all x-ray photons fiom reaching the detector
practically all of the h e . This step was crucial to the measurement of the spectrum of
the x-ray beam-if my x-ray photons reached the detector by any other path other than
through the allowed aperture, then the measured and actual spectra would not be the
same. This is also the reason why the entire pulse height spectroscopy unit had to be
wrapped in - 1 cm o f Pb shielding.
300 400 500
Energy (keV)
F i e 532. Measured spectrum of the x-ray tube at 50 kVp and 15 mA. The effect of PPU completely obscures the real output spectrum of the x-ray unit when no filtering is present.
Once the necessary shielding had been determined and installed, apertures of
various sizes were drilled through the top Pb shield directly above the CZT detector. The
h a 1 aperture that was employed during the measurements had a diameter of just 0.016"
(406.4 p), corresponding to a #78 drill bit. Even with this very small aperture severely
restricting the number of x-ray photons reaching the detector, PPU still obscured the
spectrum of the x-ray nit.
The number of photons reaching the detector was fuaher cut by pIacing extensive
A1 and occasiond Pb fiIters in the beam. Trial and error resulted in the discovery of four
combinations of different iilter thicknesses and material combinations, and tube kVp and
mA settings that gave satisfactory spectral measurements and minimal PPU. The filters
were placed directly into the collimator of the x-ray unit, where a slot was cut for their
insertion.
The measurement of each spectrum occupied the better part of one day because x-
ray tube heating necessitated cooling intervals. To gather enough counts to produce a
satisfactorily smooth spectrum requires many exposures of the x-ray unit. The x-ray head
used in this work was purchased new at the beginning of 1999 because the pulse height
spectrometer revealed that the previous head was defective, most Likely due to
overheating caused by overuse. To limit tube heating, a short exposure of 116 second ( I 0
impulses) was selected as the "x-ray quanta" that would be used throughout the work. At
most, 30 116 second bursts, each 20 seconds apart, would be performed before the tube
was allowed to cool for 15 - 20 minutes. The measured spectra were the accumulated
counts resulting fiom as many as 360 of these individual 10 impulse bursts.
5.4.4 Measured X-ray Spectra
The four combinations of A1 and Pb filters and tube kVp and mA settings that
resulted in satisfactory spectra are summarized in Table 5.1. It was desired to obtain
mean beam energies that spanned a range typical of mammography, chest, etc.
radiographic energies. It was desirable to have a beam with a mean energy - 20 - 25 keV
(i.e. mammography), but the lower energy limit of the pulse height spectroscopy unit was
- 15 keV. Any attempt to set the lower energy limit of the unit < 15 keV resulted in
severe noise corruption of the spectrum and the complete characterization of a Iow
energy beam would have proven difficult, if not impossible. The uncertainty in the tube
voItage setting for the 32.8 keV beam in Table 5.1 arises h m the x-ray unit itself There
is a common tube voltage control knob with two scales: one for a tube current of 10 mA
and another for a tube current of 15 mA. Neither scale extends below 50 kVp, but 50
kVp on the 10 mA scale would correspond to - 42 kVp on the 15 mA scale, shodd it
extend that low. To obtain the 32.8 keV beam, the voItage was set to 50 kVp on the LO
mA scale, even though the tube cuxrent was set to 15 mk
Again, PPU could not be completely eliminated, but its presence served as a
helpful c o ~ a t i o n of the measured spectra. For example, consider the raw spectrum of
the 39.2 keV beam as shown in Figure 5.23. An interesting aspect to note fiom Iooking
at the raw data is the nature of the pulse pileup "tail" that extends to higher energies. For
example, photons of energy -38 - 42 keV are much more plentill or likely than photons
at other energies. Therefore, it would be reasonable to expect that a number of these
photons would likely arrive at the detector at the same time leading to puke pileup.
Indeed, the high energy tail has a plateau that extends fiom about 70 keV to about 85 keV
or so (- 2x the energy of the incident photons). Also, since the maximum energy of the
beam is 50 keV, it would be reasonable to expect that the high energy tail should
"extinguish"at around 100 keV. This is because two photons, each with energy 50 keV,
are not likely to arrive at the detector at exactly the same time siuce 50 keV photons are
not in abundance relative to photons of other energies. That is exactly what is observed
in the plot; the tail extinguishes at about 100 keV. Another conclusion that can be drawn
h m this analysis is that the energy response of the CZT crystal is indeed linear, that is,
a pulse caused by an 80 keV photon is twice the height of a pulse caused by a 40 keV
photon. Further supporting evidence is seen in the form of the very subtle, tiny peak
observed in the data at about 120 keV- This peak cannot be seen on the graph, but is
discernable if the raw data itself is examined. Again, -40 keV photons are the most
plentill, and therefore, more likely to arrive at the detector at the same time causing
pulse pileup. The instance where three 40 keV photons arrive at once is not overly likely,
but the miniature peak at 120 keV attests to its existence.
Table 5.1. Filter combinations and tube settings for the four different spectra.
The presence of spectra1 components with energy above the maximum tube
voltage of the x-ray unit in the measured spectra is a sure sign of PPU, and to accurately
caicdate the energy that the beam deposits in an a-Se detector, these erroneous
Mean Beam . Envg~y(keV)
32.8 39.2 47.1 5 8 2 *
Tube Voltage (kvp) - 42
50 60 80
Tube C.urrent ( m ~ )
15 10 10 10
A1 Filtering (mm)
12.7 22.05 18.7
31.75
Pb Filtering @m)
250 250
0 20 40 60 80 100 120 140
Photon Energy (keV)
F i i 5.23. Raw spectral data obtained from the pulse height spectmscopy unit for the 39.2 keV beam. PPU is present, but it may be used as a check on the data,
components had to be discarded. There was also a fair amount of false low energy
photon counts in the data, but their presence may be easily explained. The pulse height
spectroscopy unit as a whole was not noise-free and any low amplitude noise would
wunt as false occurrences of low energy photons. Therefore, the raw data accumulated
fiom the puke height spectroscopy unit had to be manually filtered. Before and after
pictures of the 39.2 keV beam are presented in Figure 5.24. Asymptotes were simply
estabhshed in the raw data to reflect the fact that the x-ray spectrum did not, in fact,
extend above 50 keV nor extend below - 27 keV.
25 30 35 40 45 50 55
Photon Energy (keV)
25 30 35 40 45 50 55
Photon Energy (keV)
F i i e 5.24. (a) Raw data and (b) t i l t e d spectra of the 39.2 keV beam.
The final filtered spectra of the four beams with mean energies of 32.8,39.Z, 47-1
and 58.2 keV are presented in Figure 525. The cwves were normalized to aI1 have the
same peak count for comparison. It should be noted that the mean energies of the beams
were evaluated through
where g(E) is the measured spectrum which extends fiom energy El to energy E?.
As a final note, the issue of accuracy in the pulse height spectrometer should be
addressed. A pulse processing system, when operated over a long period of time7 will
exhibit some drift that broadens pulses of a single amplitude into a distribution of
diffmt heights. By definition, the resolution R of a system is defied as the ratio of the
quadrature sum FWHM (MI width half maximum) energy drift from all somes in the
system to the energy of interest. The reported total drift of the MCA I 75 ppm. The
reported drift of the CZT detector is 4.3% @ 59.5 keV 241~m peak, 3.5% @ 122 keV
n ~ o peak, and 1.2% $662 keV %s peak. Therefore, the warst resolution over the
energies measured in this work is Rwmt I &75x 1 od)'+ (0.~3)' = 4.3% . It should be
noted that pulse drift and resolution are not a Iarge concern in this work, since the x-ray
5000 - - 39.2 keV Beam 47.1 keV Beam
10 20 30 40 50 60 70 80
Photon Energy (keV)
Figure 525. The manually fdtered spectra of the four beams used throughout the course of this work with mean energies of 32.8,39.2,47.l and 58.2 keV.
spectra being measured were continuous; therefore, pulse spreading is not a significaut
issue. Pulse spreading really only becomes important when examining the characteristic
emission peaks of radioactive i s o t o ~ w h i c h again reIates to the intended use of the
pulse height spectroscopy unit: radioactive isotope detection and identification.
5.4.5 Energy Absorbed in a-Se Layer and WEHP
To calculate the energy absorbed in the a-Se layer being irmdiated, one had to
know how many x-ray photons were present in the beam. e.g.
Here, E is the photon energy, d@(E)ldE is the number of photons with energy between E
and E + dE, A is the area of the detector, L s is the detector thickness, and A ~ L is the
energy absorption coefficient of selenium for photons of energy E.
That information was not possible to obtain born the pulse height spectroscopy
unit since the efficiency curve of the system was not known; this point will be discussed
below. The concept of "counts" vs. energy does not directly relate to the actual number
of photons that struck the detector without the intrinsic efficiency correction. Therefore,
the need for a calibrator x-ray photoconductor became apparent. The general idea of this
approach is that a detector material with a well deibed WW and energy absorption
coefficient could be irradiated by the four different beams characterized thus far. If the
resulting photocurrent was monitored to find the charge liberated in the calibrator
material, then the true amount of energy that the x-ray beam deposited in it could be
found via
Here e is the electronic charge and Q is the charge collected through integration of the
photocurrent (coulombs). Equations 52 and 5.3 codd then be used together to relate the
measured "counts" vs. energy spectrum to the actual number of photons vs. energy h m
the x-ray unit
Therefore,
where Kph is a dimensionless constant of proportionality that relates the measured
spectrum of CN counts at energy E in energy intmal dE to the actual energy absorbed in
the calibrator material. q(E) is the intriosic &ciency of the detector for photons with
energy between E and E + dE and will be discussed in detail beIow. The measured
spectrum is in units of keV, thus the need for the fictor of 1000. If Kfi could be found
for each of the four different beams, then the energy absorbed in the a-Se layer could be
easily calculated.
The intrinsic efficiency, q(E), of a radiation detector is a measure of the
probability that the detector will release charge in response to being struck with a photon
of energy E. Thus the measured spectra, which consists of counts vs. photon energy,
must be divided by the intrinsic efficiency of the detector in order to arrive at the true
spectra. This is required to accurately calculate the energy absorbed in the detector, as in
Equation 5.4. Since a CZT crystal was readily available fiom the pulse height
spectroscopy unit, it was chosen as the caIiirator material. However, the intrinsic
efficiency of the CZT detector employed in this work was not known and was not readily
measurable. It was expected that the intrinsic efficiency of a CZT detector should at least
resemble that of a CdTe detector, and such data is available [105]. Since the intrinsic
efficiency of CdTe varies at most - 10% over the range of photon energies employed in
this work (-80% at 30 keV, rising to 90% at 60 keV, and falling to - 82% at 80 keV), the
impact of this term on the measured spectra would be small. Indeed, a good
approximation would consider the efficiency to be constant over the energy range seen in
this work. This was the approach haHy taken; q was set to be unity and the constant
Kph would then account for geometric efficiency variations. The validity of this
technique was tested through the comparison of previously reported WEHP values with
those obtained by this method. At a constant bias field of 10 Vlpm, Wmp has been
reported to be - 35 - 55 eV over the diagnostic x-ray energy range [70 - 751; this study
found that WEHP was approximately 50 eV for the four different beams descri i earlier.
In addition, a simple check was performed. The measured spectra were corrected by the
efficiency curve of a CdTe detector (from [1051) and the Wmp values (with and without
the correction) for a sample irradiated by the 58.2 keV beam at F = 9.4 V / p differed by
only 0.41 %. This agreement with published results and the small difference in WUIP with
and without an efficiency correction were taken as Mer support that approximating the
intrinsic eEciency of the detector as constant was valid. The complete results are found
in Chapter 6.
The Wm for the Cdo.&.rTe detector tiom eV Products is 5.0 eV f 0.1 eV
[106], and the energy absorption coefficient was obtained from the Nationd Institute of
Standards and Technology [81]. The density of C&&,Te is given as 5.9 g/cm3 [107].
The CZT crystal itself is a 5 mm cube. The CZT detector was biased with a lOOOV dc
source, and the photocurrent resulting fiom x-ray irradiation was displayed and captured
by a Tektronix TDSZlO 1GSals (I gigasample per second) 60 MHz digitizing
oscilloscope. The photocurrent was transferred to computer and numerically integrated
to reveal the total charge h i t e d by x-ray irradiation. The complete experimental
apparatus and procedure involved in this task will be discussed in fiuther detail in the
following section.
Table 5 2 presents the Kph values of each of the four x-ray beams. Since the
output of the x-ray tube inevitably varies slightly fkom exposure to exposure, the charge
collected from 30 ten impulse '%burstsw was averaged to arrive at an average charge,
Qrnmge, liberated in the CZT crystal when irradiated by the different beams. Each Kph is calculated using this 30 measurement average. It should be noted that when performing
Wmp measurements on the a-Se layers, the distance between the x-ray unit and the a-Se
layers had to be equal to the distance between the x-ray unit and the CZT crystal when
the Kfi values were measured and calculated. Care was taken in this regard throughout
the measurements; if the two distances were not the same, the x-ray intensities seen by
the CZT crystal and a-Se layer would not be the same, and the Kph values would no
longer be accurate.
Once the Kfi values were found, the energy absorbed in the a-Se layer was then
trivial to accurately calculate. The energy absorbed in the a-Se layer is found by
Table 5.2. The four different x-ray beams and their Kh values.
and the electron-hole pair creation energy is then
M u n Beam Energy (IrcV) 32.8 39.2 47.1 58.2
6, (dhe~ionless) 55792 248.94 145.5 1 267.63
where Q is tfie charge collected from the a-Se layer when irradiated by an x-ray beam-
5.4.6 EV Converter
The layout of the x-ray induced photocurrent measurement system is shown in
Figure 5.26. These measurements were the most straightforward undertaken in this work.
An a-Se sample is b i d by a dc supply and the photocment resulting h m x-ray
irradiation is amplified by an I-V converter and captured by the TDS21O digital
oscilloscope. The captured waveforms are transferred to computer and numerically
integrated to iind the charge Liberated in the a-Se sample by the x-rays that strike it. The
energy that the x-ray beam deposits in the sample may be calculated as described above,
meaning that the energy to mate a coI1ected EHP, WMP, may be easily calculated.
X-ray Unit
Figure 5.26. Diagram of the apparatus used to measure the WEHP of a-Se.
StrictIy speaking, a Astor couId be considered to be a ~lrrent-to-voltage (I-V)
converter, and in many respects a simple resistor was the major component of the I-V
converter used in the come of this work. The I-V converter was necessary to amplify
the very Iow (- nA) x-ray induced photocurrents in the a-Se samples to a readily
detectable voltage. It was also necessary for it to provide a virtual ground at its input to
prevent the effect of sample loading. A schematic of the I-V converter built for this work
is pictured in Figure 5.27. The batteries, power switch, 0.1 pF bypass capacitors and f
15 V voltage regulators have been omitted for clarity.
Figure 527. Circuit schematic of the I-V converter built for this work with a variable conversion gain selectable in decades h m lo5 to 10' VIA.
The principle of operation of the I-V converter is straightforward. When an
operational amplifier (opamp) has a negative feedback path from its output to its
inverting input, it will strive to keep the two inputs at the same potential. Therefore,
since the noninverting (+) input of the first stage is grounded, the opamp will keep the
inverting input (-) at ground potential as well (a virtual ground input). The very low
input bias current (30 PA) of the TLO8 1 opamp ensures that the opamp itself will not load
the small (- nA) photocurrent signal, and its gain-bandwidth product of 3 MHz ensures
that all significant spectral components of the x-ray induced photocurrent will be passed.
Once the amplified photocurrent emerges h m the h t stage, it is inverted, that
is, the x-ray induced photocurrent pulses are negative. To flip the photocurrent pulses
positive, an inverting stage (the 3" stage) was added A voltage follower 2d stage is
present to isolate the 1" and 3d stages. An ac coupling capacitor is available in case the
dark current in the a-Se sample being studied is unacceptably high. The I-V converter
was calibrated through the use of a Keithley 616 Digital Electrometer. The maximum
conversion gain available from the I-V converter was lo9 VIA.
Low pass filtering of the x-ray induced photocurrent is performed by the 3" stage
of the I-V converter, with a cutoff frequency of 5 kHz. This filtering is necessary to
reduce high fkquency noise present in the high field dark current of an a-Se sample, and
to help force the I-V converter to be unconditionally stable. A passband of 5 kHz was
sufficient to adequately pass all significant components of the photocurrent pulses.
5.4.7 Low Temperature Protective Chamber
A previous study into the effect of temperature on the EHP creation energy of a-
Se [73, 1021 was extremely limited in its achievable temperatures because water vapour
would condense on the cooled sample and form a conductive path h m the electrode to
ground with disastrous results.
To prevent water vapour h m reaching the sample, a simple protective chamber
was fashioned h m the remnants of a 1 liter plastic soft drink container and an aluminum
"flwf' as pictured in Figure 5.28. The a-Se sample was held securely against the
aluminum "floor" with pressure contacts so that good thermal contact with the alumhum
would be maintained Connections to the electrode and substrate of the sample were
made with silver paint and to coaxid connectors mounted in the aluminum. A
thermocouple was placed in direct contact with the top of the sample near the electrode
so that the temperature of the sample codd be monitored. AU joints and holes in the
container, except one bleed hole, were then sealed with caulking. During the
experiments, the chamber was continually flooded with argon gas via the cap of the
container so that the Ar, being more dense than air, would displace any and all moisture
laden air present in the container.
Bleed Hole
Figure 538. Diagram of the low temperature protective chamber flooded with Ar.
The temperature dependence of the EHP creation energy of a-Se was investigated
by initially cooling the sample (inside its protective chamber) with dry ice (solid COz at - 195 K or - 7g°C). The WEHP of the sample was then measured as the sample warmed to
near room temperature.
5.4.8 X-ray Collimator/Fiiter Housing
Aluminum a d lead fltm were extensively used to create x-ray beams with
certain energy disttibutions, as described previously. However, x-ray photons that are
scattered by these filters can cause problems in that these low energy scattered photons
can increase the apparent dose seen by the detector [lo81 (the a-Se samples in this case).
To reduce the effect of scattered radiation, collimators are frequently employed to limit
the beam to a narrow cross section. Disks of Pb sheeting with a circular aperture of 2.5
cm diameter were placed inside the collimator of the x-ray unit as shown in Figure 529.
The collimator had a slot cut in its side for the insertion of up to 6 mrn of AI sheet filters,
but the majority of the filters were secured to the bottom of the collimator using duct
tape.
Threads into X-ray Unit 7 - Slot (for up to 6 mm /- &--- of additional filters)
I 5 mrn
I
_I 5 mm
-r-
Pb Disks
Pb Disks
F i e 5.29. A cutaway view of the modified collimator attached to the x-ray unit to restrict scattered radiation fiom reaching the samples.
With the collimator in place, care had to be taken to ensure that the a-Se samples
were aligned beneath the aperture in the collimator. Alignment was not absolutely
critical, as the x-ray unit had a sizeable (approx. 2 cm x 3 crn) spot where the measured
x-ray photon intensity was uniform. The a-Se samples were placed at a distance of - 25.5 cm from the focal spot of the x-ray unit during all measurements.
5.5 Miscellaneous Experimental Tools
The memement of the electron lifetime re and how it was affected by annealing
at elevated temperatures, IR soaking and exposure to uItrasonic vi'bration employed three
additional experimental tools, as discussed below.
5.5.1 Sample Oven
To elevate and control the temperature of the a-Se samples during annealing, a
small simple oven was built as shown in Figure 5.30. An Omega CN2012 Programmable
Temperature Controller used a thermocouple to measure the temperature of an aluminum
block with a corded heating element threaded through it. A solid state reIay (SSR) inside
the temperature controller would connect the heating element to h e voltage (120 V, 60
Hz) in order to raise the temperature of the aluminum block to the operator specified
Corded Heater Element ' ,' /'- , - /,. ' . -
,'- I-
/
Foam Insulation
120 Vac Wi State Relay
F i e 530. Simple oven for sample annealing using a temperature controller and a heating eiement. A thermocouple connected to the temperature controller provides a means to measure the temperature of the duminum block
setpoint. To prevent stray light in the room from reaching a sample while it was being
annealed, the unit was fitted with a Iid.
5.53 IR Light Source
The IR light source used in this work was identical to that used by Haugen [102],
and it consists of an ordinary household lamp with a standard incandescent tungsten Light
bulb. A thin silicon wafer was secured to the bell shade of the lamp with electrid tape;
the wafer completely covered the opening of the lamp. With the lamp turned on, only
photons of energy I 1 .I2 eV (the bandgap of silicon) wodd be transmitted by the silicon
wafer, forming a good IR light source.
5.53 Ultrasonic Waves
Ultrasonic waves were created by the apparatus shown in Figure 5.3 1 [I 091. A
Philips PM5134 Function Generator created a 4 MHz sine wave which was amplified and
fed to a Matec ICS-SR PZT (lead zirconate titanate) ultrasonic transducer with a nominal
resonant frequency of 4 MHz. Viscous automotive grease served as an acoustic matching
fluid so that the acoustic waves created by the traducer would be adequately coupled
into the a-Se sample being tested. The unit was fitted with a removable Iid to shield the
a-Se sample undergoing ultrasonic treatment fkom any stray room light.
Signal Generator 552 \ --.
4-00 MHz ' .z 6 6
Ultrasonic Waves \
L'
- :\>/ - PZT Transducer
"
L' ArnpTir
F i 531. Equipment employed to create ultrasonic waves in the a-Se samples.
5.6 Summary
The experimental systems used in the course of this work were introduced; a
brief description of the sample preparation procedure led the discussion. After selenium
pellets are thermally evaporated onto conductive substrates, semitransparent Au
electrodes are sputtered onto the samples to facilitate transient photoconductivity and
Wmp measurements.
A TOF/IFTOF transient photoconductivity apparatus was descnied; this system
allows the charge transport properties of high resistivity solids to be examined.
irradiation of the samples by x-rays allows the study of x-ray induced properties such as
changes in electron lifetime.
A system to measure the x-ray induced photocunent in an a-Se film was
introduced. This system measures the number of free carrim produced by irradiating an
a-Se sample with x-rays, and provides a means to calculate the EHP creation energy in a-
Se. The sample must be isolated from atmospheric water vapour when performing low
temperature measurements to prevent the buildup of condensation and thus the
destruction of the sample.
Finally, some miscellaneous experimental tools used in the course of this work to
investigate the electron lifetime dependence of a-Se on annealiug, IR soaking and
dtrasonic treatment were discussed
6. Results and Discussion
6.1 Introduction
Chapter 5 presented the experimental details of the study undertaken in this work,
and a comprehensive summary of the results are discussed in this chapter. The samples
used throughout the course of this work were chlorinated a-Se:0.2-O.S%As; otherwise
known as stabilized amorphous selenium, and henceforth simply called a-St.
This chapter first presents a description of the TOF and IFTOF experiments
performed on the samples to determine their suitability as an x-ray detector. h general, a
detector must possess suitably good charge transport such that a charge carrier that is
created by x-ray irradiation will not become trapped as it traverses the sample. The
results of these preliminary measurements are presented £irst, followed by extensive
IFTOF measurements to gauge any changes in charge transport upon exposure to x-rays.
This subject is dealt with in detail in section 6.3. Section 6.4 discusses the measurement
of WUp and how it varies with temperahue, bias field and mean x-ray beam energy. The
penistent x-ray photocufient is examined in section 6.5 in an effort to M e r understand
this phenomenon.
6.2 Charge Transport Measurements
6.2.1 TOF Measurements
The small signal condition, introduced in Chapter 3, must be met when
performing TOF or IFTOF measurements. Under small signal conditions (ie. when the
photoinjected charge is much less than C,V'), the electric field F = VJL inside the sample
may be approximated as being uniform at alI points within it. This may be sew
experimentally by an almost rectangular photocurrent pulse in samples with relativeIy
long camier lifetimes. The transition h m the small signd condition to charge
overinjection is easy to determine, as the photocurrent pulse will suddenly change h m
being rectangular to almost triangular in shape. Charge overinjection was avoided during
all measurements through the use of neutral density optical attenuators placed in the path
of the laser pulse.
It has been shown that the internal bias field will remain constant and uniform
following application of the bias until a transition time tsc is reached f 1101. A h this
transition, a space charge will build in the bulk of the sample due to charge injection from
the contacts. The internal field then becomes non-uniform as dictated by the Poisson
equation dF/& = pk, where p is the buik charge density and E is the permittivity of the
materid. If this bulk space charge is allowed to build in the sample prior to
photoexcitation, then the photocurrent will either increase or decay depending on the
species of bulk charge [94]. The transition time of a-Se is typically on the order of - 100
ms, and in the present measurements photoexcitation took place within 20 ms of bias
application to avoid these field pexturbations.
Figure 6.1 presents typical TOF photocurrent waveforms of holes and electrons
(the amount of photoinjected charge is not equal for both cases). Examination of both
p h o t o ~ ~ e n t s reveals that minimal deep trapping occurs during the transit time, as
evidenced by the reIatively flat plateaus of the waveforms. The eIectron photocurrent
displays a rather sharply decaying Ieading edge and it is thought that this behviour may
be due to carrier thermhtion (the time necessary for the photoinjected carriers to reach
equiliium with the shallow traps). This behaviour wdd also be caused by
nonuniformities within the sampIe itsec and it is not conclusiveIy known which may be
the cause. It should be noted that all samples employed in the study exhiiited this
electron transport behaviour,
Time (ps)
Time (ps)
F i i e 6.1. Typical I-made TOF waveforms in a-Se for (a) holes and (b) electrons. The m i t times are indicated as t ~ . [(a) F = 0.98 V/CL~, (b) F = 5.69 V l p ]
For this study, the transit time of the charge caniers was defined as the time when
the photocurrent drops to half its value at the knee of the waveform, as shown in Figure
6.2 below. This definition was necessary because electron TOF waveforms can undergo
quite significant trapping during their trip across the sample, making a determination of
the nominal photocurrent during drift difficult. It was found that there is a maxima in the
Wefetztiated photocurrent corresponding to the knee of the photocurrent (as shown in
Figure 6.2), making automated location of this point through software very easy. This
maxima is a mathematical consequence of the cliffamtiation algorithm.
The charge carrier mobility p and the measured transit time are related through
the following equation
Where L is the sampIe thickness and V is the appIied voltage. Thus a plot of the
F i e 63. (a) Typical electron TOF photocurrent showing the location of the photocurrent's knee and corresponding '/4 value point (b) The photocurrent of (a) is differentiated reveaIing a local maxima corresponding to the Iocation of the knee in (a). M y a portion of the diffimitiated waveform (the tail) has been shown for clarity.
measured transit time t~ VS. 1/V for various applied voltages will yield a straight h e
whose slope is the mobility of that charge carrier. Figure 6.3 (a) and (b) are plots of the
transit time vs. 1/V for holes and electrons respectively. The calculated hole mobility
from Figure 6.3 (a) is p+, = 0.132 c m h s and the calculated electron mobility from Figure
6.3 (b) is ~ l , = 0.003 18 cm2Ns.
F i e 63. Plot of TOF transit time vs. IIV for (a) holes showing = 0.132 c m 2 ~ s and (b) electrons showing 14 = 0.003 18 an2Ns.
When Equation 6.1 is used to calculate the drift mobilities of both holes and
electrons and these mobilities are plotted as a hc t ion of applied eIectric field, the plor
of Figure 6.4 (a) and (b) are obsenred. As can be seen, the hole mobility shows tittle
observable field dependence, while the electron mobility has a very slight power law field
dependence of the type = F where n was experimentally determined to be - 0.17,
which is quite small. This small field dependence is not I l l y understood at present, but
for the Iimited field ranges employed in this work, both the hok and electron drift
mobilities were assumed to be constant and field independent.
108 i 07 IOB 1 o7 Field (Vim) Field (Vlm)
[a) (b)
Figure 6.4. (a) Hole and (b) etectron mobiIity plotted as a function of applied field. The hole mobility shows very little field dependence while the electron mobility shows a slight field dependence of the form ~4 = l?",
6.2.2 IFTOF Measurements
The charge transport propexties of an x-ray photoconductor are not completeIy
characterized by the drift mobilities of its charge carriers. Some carriers may be Iost due
to deep trapping and will thus decrease the effective sensitivity of the detector.
As described earlier, the FTOF experiment provides a convenient means of
measuring the mean canier lifetime z by measuring the hctional recovered photocurrent
as a hction of the interruption period. Figure 6.5 shows a typical FTOF waveform for
the case of drifting eIectrons. The field is removed at $1 and reapptied at t2, corresponding
to an intenuption time q = t2 - tl. Since the photocurrent at any particular instant is
pro@onaI to the number of drifting charge carriers at that same instant, the slope of a
semilogarithmic plot of i j i r vs. ti yieIds the charge carrier trapping lifetime 7.
During the IFTOF experiment, charge carriers are trapped under zero applied
field, and the measured carrier lifetime should thus be independent of applied voltage,
provided that the charge packet is halted at the same depth in the sample. A recent study
bas c o d b e d this fact [6 11.
Figure 6.5. Typical electron IFTOF waveform in an a-Se film showing i,, i2, and ti.
By measuring ir and i2 for several different interruption times ti, the plots found in
Figures 6.6 and 6.7 for the case of holes and electrons respectively were obtained. The
hole lifetime was determined to be T,J, = 132 ps and the corresponding electron lifetime
was T= = 657 ps. These lifetimes seem reasonable when compared to the TOF waveforms
of Figure 6.1; neither photocurrent exhibits any discemable decay since f~ << r. The
exponentid decay of the fhctional recovered photocurrent for both cases (holes and
eiectrons) may be used as evidence that there is no release of trapped carriers fiom deep
traps during the timescale of the experiment.
0 20 40 60 80 100 120 140 160
Intemption Time (ps)
Figure 6.6. Fractional recovered hole photocurrent as a function of interruption time for a-Se. Measured hole lifetime b m the plot is 132 p. The charge packet was halted at a depth of 0.263 L where L is the thickness of the fi.
One immediately obvious feature of Figure 6.7 is the fact that the plot of
hctional recovered electron photocurrent does not pass through the origin; in other
words, a zero interruption time does not correspond to all injected charges being
subsequently collected. This peculiar feature of electrons was noted for all samples
examined during the course of the work and for all depths at which the charge packet was
halted. Since there is no applied field while the charge packet is halted, any motion of
charge carriers must be due solely to the self-field drift alone, caused in part by the
mutual wulombic repulsion of the charge carriers themselves. This motion of electrons
then resuIts in a hction of the charge packet which reaches the top electrode and is
removed h m the sample, resulting in the behaviour seen in Figure 6.7. Figure 6.8 is a
plot of the M o n a 1 recovered electroil photocurrent as in Figure 6.7, except that the
charge packet was halted at a depth of 0.537 L. Since the y-intercept of this plot shows
that less charge was lost during intemption at 0.537 L than at 0.263 L that serves as
confirmation that !he charges are indeed being lost through the top contact. A recent
Interruption Time (ps)
Figure 6.7. Fractional recovered eIectron photocurrent as a hct ion of intemption time for a-Se. Measured eiectron lifetime from the plot is 657 p. The charge packet was halted at a depth of 0.263 L where L is the thickness of the fi.
study [I 111 also discovered that electron dispersion in a-Se films is anomalous in that the
high amount of dispersion evidenced by an electron charge packet in a TOF measurement
cannot be reconciled through conventional diffusion, multiple trap and release events
from a monoenergetic set of traps, or hopping transport.
The exact mechanism for this behaviour is unknown, but two theories are
presented schematically in Figure 6.9 below. The first possibility is that the zero field
diffusion of electrons in a-Se is considerably higher than previously thought, leading to
the considerable spreading of what was once a narrow charge packet upon its halting.
Some charge may eventually reach the top electrode and be lost from the sample, as in
Figure 6.9 (a). The second possibility is that the electron charge packet undergoes a very
large amount of dispersion as it drifts across the sample. This may leave an appreciable
number of electrons behind the main ''fotmati0n7' as it is swept across the sample; a large
kction of these sIow carriers may Iie within a diffusion length of the top contact and can
be removed from the sample while the packet is halted, as shown in Figure 6.9 (b). The
0 100 200 300 400 500 600 700 800 900
Interruption Time (p)
F i e 6.8. Fractional recovered photocurrent at a depth of 0.537 L for the same sample as in Figure 6.7. The higher y-interce~t indicates that less charges are lost at this depth than at the shallower halt depth of Figure 6.7.
fact that more charges are lost when the packet is halted at a relatively shallow depth in
contrast to when the charge packet is haIted relatively deeply suggests that the latter case
is the cause. If pure diffusion were responsible, some charges could reach the bottom
contact as well as the top, and here should be relatively equal numbers of charges lost at
both shallow and deep haking depths. The shape of the differentiated photocurrent tail
(see Figure 6.2 (b)) corresponds to the shape of the charge carrier packet [loo, 1 1 11 and it
shows a very long tail region corresponding to ''slow" carriers, also in agreement with a
large amount of dispersion. The width of the tail of a typical TOF electron photocurrent is
also suggestive of dispersion being the cause, as the total spread of the tail is comparable
to the transit time of the carriers.
An alternative and convenient method of determining the charge carrier deep
trapping lifetime was also used throughout the course of this study. The ratio qTlql,
comsponding to the amount of charge present in the photocment pulse immediately
prior to and following interruption will also yield the same carrier Lifetime when plotted
semilogarithmidly versus the interruption time [6O, LOO]. This method was preferred
because it could be performed quickIy and automatically via software. This method dso
offered the advantage of being relatively immune to any noise present on the recovered
photocurrent pulses [60, loo], and it offered a "neutd' analysis of the data in that it did
not have any preconceived notions of how the data should appear, as opposed to human
eyes.
Photoexcitation
a-Se Sample -
During Interruption Packet Spreads /7- . ,
, , , i '. .
, I i .. . , " %, +b
,:Packet When ,, _ - - - / . Halted -
> Direction of Drift
+ Direction of Drift
Figure 6.9. Two possibilities for tbe lost charge observed in fractional recovered electron photocurrent IFTOF pbts: (a) ekctmn diffusion while the packet is halted and (b) large dispersion during driR Tbe latter case is more likely, as supported by the experimental evidence.
The charge carrier Hetime is found via the slope of a linear regression applied to
the various vdues of fractional recovered photowent (or charge) when plotted
semilogarithmically against the interruption time. Hence the error in the carrier lifetime
is directly related to the uncertainty in the slope of the linear regression. For a given set
of N ordered pairs of data, y&), the equation of a least squared error line that best fits
that data is given by [I 12- 1 141
y=mx+b, 6.2
where m is the slope and b is the y-intercept of the line. m is found through
where
and
where ; is the average of all xi and ; is the average of all fi The standard deviation of
the slope is
where
Thus the error in the charge carrier lifetime is simply + a,. This relation was used
throughout the course of the work to determine the error in the calculated charge carrier
lifetimes.
As mentioned previously, there is ample evidence to suggest that the electron
charge packet undergoes very significant dispersion during its drift across the a-Se
sample. This is aIso reflected in a pIot of the effective carrier transit time vs. intenuption
time (ti) for the case of electrons, as in Figure 6.10. The effective transit time tr' is
defined as the time it takes for the photoinjected carriers to be swept out of the sample
during an IFTOF experiment, minus the interruption time itself. As can be seen in Figure
6.10, initially the effective transit time increases linearly with increasing ti, with a
pronounced corner at ti - 200 p after which point the linear increase in tr' with t i is not
as strong. This phenomenon is unique to electrons in a-Se, as hole transport does not
display this behaviour [IOO]. This seemingly odd behaviour-that the eIectron transit
time is affected by and depends on an intmption in the field-may be explained in
Interruption Time (ps)
Figure 6.10. Eff'ective electron transit time tT' plotted as a function of intmption time ti in a typical IFTOF experiment. The quantity plotted is the incremental difference in the effective transit time compared to the transit time obtained from a TOF experiment.
terms of carrier dispersion and the non-symmetric shape of the electron charge packet.
The eIectron charge carrier packet is highly non-symmetric (refer to Figure 6 2
(b)), with a region near the "head" of the packet in which most of the carriers reside.
There is also a long tail region which contains a significant number of carriers that tags
behind the main 'Tormation". For this study, the transit time (or effective transit time)
was defined as the point in time at which the tail of the photocurrent decayed to !4 its
value at the %nee'' of the waveform. This may be equivalently stated as the time at
which half the photoinjected carriers left the a-Se sample. Therefore, since the
photoinjected electron charge packet is highly non-symmetric, self-fieid drift while the
packet is halted will significantly alter the overall shape and electron distribution within
the packet. This would also alter the location of the half charge point within the carrier
packet, leading to a change in the observed effective transit time which depended on the
intenuption time as seen in Figure 6.10. This also explains why this behaviour is not
observed for the case of bole transport in a-Se, since the hole packet is highly symmetric
[IOO] and its half charge point would be relatively immune to self-field drift while the
packet is halted.
The apparent coma at - 200 p seen in Figure 6.10 may be due to the mutual
coulombic repulsion of ike carriers. As time progresses, the packet will spread more and
more, but the spread should slow somewhat as ibe carriers become deeply trapped. The
electron Lifetime corresponding to the data of Figure 6.10 was measured to be 707 p.
Therefore, at 200 p, the number of charge carriers remaining in the packet will be
-200 e -0.1 1 = -64 where the 0.1 1 is the hction of charge lost at zero interruption time
because a plot of hctional recovered photocurrent vs. intemption time does not pass
through the origin for electrons, as discussed d e r . It is quite remarkable that the
corner of Figure 6.10 corresponds to 64% of the charge remaining in the packet; in other
words, to 36% ofthe charge lost from the packet Since = 0.36, it may be surmised
that the spread of the carrier packet initially has an exponential dependence on the
number of free Wers in the charge pack% with a characteristic T = 200 p.
The observed dependence of the effective transit time for electrons on the
intermption time during an IFTOF experiment coincided with a similar dependence of the
measured electron dispersion on the interruption time. For this study, the dispersion of
charge caniers was measured by examining the differentiated tail of the TOF or IFTOF
waveforms. The precise method has been discussed elsewhere [86, 100, 1021, so only a
very brief outline of the procedure will be presented here.
In general, differentiation tends to be a very noisy process which can be prone to
large errors depending on the initial "smoothness'* of the waveform to be differentiated.
In an effort to smooth any noise present in TOF or IFTOF waveforms, a least squared
error parabolic curve is fit to a portion of the data, hereby r e f d to as the window.
That parabola is then diffaentiated, yielding one data point corresponding to the slope of
the original TOF or lFTOF waveform at the middle of the window. The window is then
"slid" one point along in time and the process is repeated until the entire waveform has
been analyzed.
The measured dispersion for this study corresponded to the FWHM of the
differentiated tail; in other words, to the 95 magnitude width of the differentiated tail,
consisteat with other studies [86, I#, 102, 11 11. Figure 6.1 1 is a plot of the
differentiated tails of recovered electron IFTOF waveforms, with ti = 50.8, 99.1, 150,
200.8, 300, 399.1, 500.8, 600 and 699.1 p. It can be seen that dispersion generally
increases with increasing interruption time ti-
The measured FWHM dispersion vs. intermption time corresponding to the
cwes of Figure 6.1 1 is presented in Figure 6.12. The data is presented in Figure 6. I2 as
an incrementd dispersion relative to the dispersion present in a TOF waveform, ATOF.
Just as with the case of the effective transit time (Figure 6.10), there is a comer in the
curve at an interruption time of - 200 p. Again, the mutual wdombic repulsion of free
photoinjected charge carriers would tend to strongly spread the carrier packet for small
interruption times; as more free charges become deeply trapped, the observed increase in
I t I L I L
0 200 400 600 800 1000
Time (ps)
Figure 6.1 1. PIots of differentiated electron IFTOF photocunent tails vs. time for an a-Se sample. The intemption time was varied from 50.8 ps (bottom curve) to 699.1 P (top m e ) .
dispersion would tend to slow for large interruption times. The comer at - 200 p lends
firrther evidence that the spread of the packet is strongly dependent on the concentration
of free photoinjected carriers.
The bulk of the work done in this study concerned any x-ray induced changes in
the electron deep trapping lifetime r, and it was therefore essentiai to etlfllre that the
measurement of & itself had no effect on the charge transport properttes of the material
under study. Of particular concern was the pssiiility of trap filling, whereby the
measurement of the trapping lifetime itseIf would reduce the number of avaiIabIe d i e d
traps, ieading to an increase in h e apparent lifetime. To determine if the measurement of
the trapping lifetime itself was af5ecting the materid, a set of four measurements at two
0 100 200 300 400 500 600 700 800
Intemption Time (ps)
Figure 6.12. Incremental electron dispersion h m an EFTOF experiment relative to the dispersion h m a TOF experiment, ATOF, plotted vs. the interruption time ti.
hour intervals was performed. The results of this study are presented in Figure 6.13. The
charge packet was halted at a depth of 0.287 L for every measurement. As can be seen,
the measured electron Hetime does not significandy vary from measurement to
measurement. It was therefore assumed that the method of measuring the electron
lifetime did not alter the material itself, and was herefore a valid meamremat tool.
6.23 a-Se Film Quality
As stated before, the average distance that a charge carrier will travel before
becoming deeply trapped is an important measure of the suitability of an a-Se fiIm as an
x-ray detector. This distance, denoted the Schubweg, is equd to the product of the
0 1 2 3 4 5 6 7
Time (h)
Figure 6.13. Measured electron lifetime, normalized to the initial (t = 0) lifetime, for an a-Se fiIm over the course of a 6 hour period. The measured lifetime did not significantly vary with time, therefore the measurement technique did not alter the trapping characteristics of the a-Se film under study. The electron lifetime was initially determined to be 5 10 p.
charge carrier mobility, deep trapping lifetime and applied electric field. In general, the
Schubwegs (both electron and hoIe) of a film must be greater than the film thickness in
order to ensure that all charges that are Ir'berated by x-rays will be collected before they
become trapped.
The a-Se samples used in the course of this work were selected on the basis of
their excellent charge transport properties so that a study of films simiIar to those in use
in commercially available systems could be made. The study of charge carrier Iifetime
and how it was affected by exposure to x-rays involved the use of four samples, each
detailed in Table 6.1 below. Every sample exhiiited a variance in its charge carrier
lifetime which was dependent on the depth at which the lFTOF analysis was performed.
For this reason, the carrier lifetimes presented below represent the lowest measured
lifetimes in the sample. The concept of the minimum operating field is introduced so that
a clear comparison among the samples may be made; it is defined (for this study) as the
smallest fieId which will result in a Schubweg equal to twice the Nm thickness for the
charge carriers with the smallest ~ L T product (the range of the carrier). This ensures that
the carriers with the lowest range will likely be able to traverse the sample without
becoming trapped. In general, the sample with the lowest minimum operating field will
have the best charge transport.
Table 6.1. Charge transport properties of the four a-Se films involved in the carrier
The two samples used in the measurement of WEHP and the persistent
lifetime study.
photocurrent study are each detailed in Table 6.2.
Table 6.2. Charge transport properties of the two a-Se films involved in the x-ray
Minimum Operating
Field (Vim) 5.56 7.18 4.05 537
~ ( p s )
504 447 697 185
Typical operating fields in a commercial digital x-ray imaging system employing
z b w )
167 186 178 - 100
a-Se as the detector exceed 10 V l p - As can be seen in Tables 6.1 and 6.2, the necessary
(cm2ffs)
0.00357 0.003 18 0.00327 0.00342
Sample
960521 46 SE 15 971002 12 971205 62 980622 - 3
photoconductive ex-ents.
minimum operating field for good charge collection is below this threshold for every
Sample
~un#30
f 463-3
sample. Thus it may be stated that the samples employed in this study were each
L (Clm)
500 5 10 462 170
Minh- Operating
Field (Vim) 1.n 3.0 1
L W)
64
565
commercial device quality.
CIL (cm2ffs)
0.132 0.132 0.133 0.126
% W)
- uo 952
zb(Cu)
- 15 - 300
(cm2ffs)
0.1 I
0.129
~ ( c m 2 f f s )
0.0029
0.00394
63 X-ray Induced Changes in Charge Transport
This section will first introduce the experimental results of the study into the
effect that x-ray irradiation has on the charge transport properties of a-Se. The subject of
Light induced structural changes in amorphous semiconductors in general, and a-Se in
particular were briefly introduced in Chapter 2. This topic will be discussed in detail
folIowing the experimental results, and the applicability of these light induced effects to
the present study involving x-rays will be covered next. How annealing, IR soaking and
ultrasonic treatment affect the charge transport of a-Se will be introduced last.
All of the measurements that follow are presented as a function of absorbed dose;
a quantity in common use in the field of radiography. Absorbed dose is simply defined
as the energy absorbed by an object (as x-rays for example) per unit mass of the object.
Dose has units of Grays, with one Gray (Gy) equal to one Joule per kilogram.
In all but one special case, the a-Se samples were inadiated by the heavily filtered
x-ray beams described in Chapter 5. Since the spectra of the x-ray unit was known for
those four beams, the amount of energy that the beam deposited in each sample was
straightforward to calculate. Once the energy deposited in the sample was found through
Equation 5.5, the dose was then calculated by dividing that energy by the mass of the
electroded sample.
The a-Se samples used during the course of this work were usually subjected to a
maximum cumulative absorbed dose of - 30 mGy at any one time before the samples
were allowed to rest in the dark for several days to recover. For comparison, a patient
undergoing a traditional film-based diagnostic x-ray procedure absorbs a dose which is
typically less than 2 mGy, depending mainly on the area of the body being examined.
6.3.1 Charge Carrier Transit Time tr
Since the charge carrier mobility is controlled by the shallow states present in the
bandgap of a-Se, any x-ray induced changes in the carrier mobility would be due to the
alteration of either the concentration or the capture cross-section of these states. To
determine if x-ray exposure had any effect on the charge carrier mobility, two
experiments were perfonn&ne for the case of electron transport and one for the case
of hole transpod These straightforward experiments simply consisted of monitoring the
TOF transit time tr of both electrons and holes as an a-Se sample was irradiated with x-
rays. The results are found in Figure 6.14, and are expressed as a normalized transit time.
The measured electron transit time was found to change negligibly (< 1.5%) as the dose
was varied from 0 to 25.4 mGy. This change fell within the t~ measurement error of 2%;
therefore, it was concluded that x-ray irradiation has no effect on electron mobility. Hole
transit was also found to be independent of x-ray irradiation, as shown in Figure 6.14 (b).
Cumulative Dose (mGy)
0 5 10 15 20 25 30
Cumulative Dose (mGy)
F i e 6.14. (a) Electron and (b) hole TOF transit times, normalized to the initial t~ and plotted as a function of cumulative dose. Irradiation by 58.2 keV beam; sample shorted to ground during irradiation.
6.3.2 Charge Carrier Lifetimes
The bulk of the measurements performed during the course of this work involved
the measurement of the charge carrier lifetimes of a-Se fiIms and how those lifetimes
change upon exposure to x-rays. Most measurements were of the electron Lifetime, but a
number of hole lifetimes were also obtained. The data that follows is the cuImination of
over 3500 FTOF measurements resulting in almost 500 individual lifetimes.
The measurements were performed under a number of different conditions,
including different bias levels during irradiation, mean energy of the x-ray beam used for
irradiation, and absorbed dose. These parameters were varied so that the reaction of the
carrier lifetimes to these irradiation conditions could be observed. A single lifetime
usually involved 8 - 10 individual IFTOF measurements and took approximately 15 - 20
minutes to complete. In all thefigures that follow, carrier lifetimes are nomlized to the
initial lifetime obtainedjust prior to x-ray irradiation (i.e. the t = 0 point).
It was initially expected that irradiating an a-Se film with x-rays would drastically
lower the electron lifetime within that film; hthermore it was expected that a recovery
would be observed as the material relaxed back to its initial state. With this in mind, a
nmbex of experiments were performed whereby the electron lifetimes at two depths
within an a-Se £iim were initially measured. The film was then irradiated with x-rays and
the Lifetimes at those same two depths were again immediately measured Quite
surprisingly, the measured lifetimes immediately before and after irradiation usually did
not differ by more than lo%, as seen in Figures 6.15 and 6.16. Even more surprising was
the fact that as time went on, the measured lifetimes continued to change-in some cases
rising and others falling.
This u n d behaviour--that the measured electron lifetimes initially hardly
changed upon irradiation by x-rays, but then continued to change as time progressed-
was completely unexpected. It was thought that perhaps the heavily f3tered nature of the
x-ray beams used in this work could be the cause of this phenomenon. Figure 6.17 is a
32.8 keV 26.2 mGy Shorted during irradiation -
0.5 1 I I I I I /A 0 2 4 6 8 10 72
Time (h)
Figure 6.15. Normalized electron lifetime at two depths within an a-Se film tracked over time. The film was initially irradiated with a 32.8 keV beam giving an absorbed dose of 26.2 mGy, and the sample was shorted to ground during irradiation. [Sample 97 1002 121
plot of the measured electron lifetimes of the same film as that in Figures 6.15 and 6.16,
except that all filtering was removed from the beam when the sample was exposed. The
x-ray unit was set to a 90 kVp, 15 mA, 1 second exposure. The exact absorbed dose due
to this exposure is not known, but it can be safely said that it was much higher than that
in Figures 6.1 5 and 6.16. As with the previous two figures, the electron lifetimes before
and immediately after irradiation hardly changed; however, as time progressed, the
measured lifetimes continued to change. Therefore, these observed changes in the
electron lifetime within these films were assumed to be unrelated to the specifics of the
experimentd procedure itself (i-e. the heavily filtered beams).
o 6 i 32.8 keV 26.2 mGy F = 1.96 Vlpm during irradiation
Time (h)
Figure 6.16. Normalized electron lifetime at two depths within the same film as that in Figure 6.15. Irradiation specifics: 32.8 keV, 26.2 mGy, F = 1.96 Wpm.
These changes in the measured electron lifetime have been observed in all
samples examined in this work. Furthermore, the changes show no discernable
dependence on the irradiation wnditio*i.e. electric field strength during irradiation or
beam energy. Essentially, these changes do not appear to follow a recognizable pattern
and the fact that the lifetimes at two different depths act independently (they do not
necessady "track" each other) testifies that they can be quite localized within the film.
Figures 6-18 - 6.20 are further plots illustrating how the measured electron lifetime
within an a-Se film varies with time after the initid x-ray irradiation.
Another interesting feature of the eIectron lifetime is that its behaviour with time
after i d a t i o n showed no observable dependence on the absorbed dose over the range
i No Filtering 90 kVp, 15 mA, 1 second exposure Shorted during irradiation
0 2 4 6 8 10 144
Time (h)
Figure 6.17. The a-Se 6lm of Figures 6.15 and 6.16 is exposed to an u t e r e d x-ray beam of 90 kVp, 15 mA and 1 second duration. The sample was shorted during irradiation.
examined in this study. Figures 6.21 and 6.22 depict how the measured electron lifetime
changed with time after irradiation. A11 variables for these two plots were identical
except for the absorbed dose. In Figure 6.21, the absorbed dose was 25.4 mGy whereas it
was 1.8 mGy for Figure 6.22. In both plots it may be seen that the electron lifetimes
immediately before and after inadiation hardly change (< - 10°?), but as time progresses,
the measured lifetimes vary somewhat. Although the absorbed dose of Figure 622 was
l a than 1110~ that of Figure 6.21, the measured lifetimes evince greater variance as time
progresses. At ht glance, this may seem to suggest that a low dose induces a bigger
change in the electron iifetime, but an examination of the other figures presented thus far
proves that the changes lie within the general range observed under other widely varying
irradiation conditions.
"4 32.8 Ice" , , , 28.1 mGy Shorted during irradiation
0.5
0 2 4 6 8 10 96
l7me (h)
Figure 6.18. Normalized electron lifetime at two depths tracked over time. Irradiation specifics: 32.8 keV, 28.1 mGy, shorted during irradiation. [Sample 971205 621
The samples, once irradiated, were allowed to rest in the dark for at least two days
before they were again used. This allowed the samples to recover fiom any x-ray
induced changes. Sometimes the original dark-rested electron lifetimes would be
recovered, as in Figures 6.15, 6.19 and 6.20; however, sometimes the datk-rested
electron lifetimes would be as much as 20% higher than that of the previous study, as in
Figures 6.16 - 6.18 and 6.2 1. As with the changes in electron lifetime as time progressed
after the sample was irradiated, these dark-rested lifetimes seem to be random in that
there is no discernable pattern between the irradiation conditions and the resulting rested
lifetime a few days later.
This odd electron lifetime behaviour'hat the lifetime would hardly change
immediately following irradiation, but could change quite dramatically hours Iater-was
quite unexpected, as stated earlier. Since the electron lifetime was not affected by the
IFTOF measurement itself (Figure 6.13), therefore, initially-upon irradiation--some
change within the a-Se films had to be driving the later changes in electron lifetime.
0.6 4 28.1 mGy F = 2.16 Vlpm during irradiation
Time (h)
F i e 6.19. Same sample as that of Figure 6.18; irradiation specifics: 32.8
58.2 keV 25.4 mGy Open circuit during irradiation
0 2 4 6 8 1 44
Time (h)
Figure 6.20. Same sample as that of Figure 6.18; irradiation specifics: 58.2 keV, 25.4 mGy, sample completely open circuit during irradiation.
0 2 4 6 8 1 44
Time (h)
/+ 0.287L
0 0.573 L 1.1 -
r" 1.0
= 0.9
P I t2
Figure 6.21. Normalized electron lifetime at two depths tracked over time. Irradiation specifics: 58.2 keV, 25.4 mGy, F = 2.16 V l p . [Sample 971205 621
0.8 - z
0.7 - OS6 - 0.5
58.2 keV 25.4 mGy F = 2.16 Vlpm during irradiation
I I I I /A
Time (h)
1.3
Figure 6.22. Same sample as that of Figure 6.21, except that the absorbed dose is 1.8 mG y.
l2 - 1 .I =-
r'
In an effort to find some cause for these changes, both the hole and electron
lifetime at the same depth within a sample were examined as a fimction of time following
irradiation. As can be seen in Figures 6.23 and 6.24, the electron lifetime behaves in the
58.2 keV l.8rnGy F = 2.16 Vlpm during irradiation
0 0.573 L
p 1.0
z 0.7 - 0.6 - P
same manner as seen earlier. Immediately before and after irradiation, it is hardly
affiected, but goes on to change unpredictably hours later. However, the hole lifetime
dramatically fdls 20 - 30% immediately after irradiation; as with the observed changes
in electron lifetime, it proceeds to change unpredictably as time progresses. These
observations allayed some of the previous concern; irradiation by x-rays did induce
immediate changes in the a-Se films, but those changes had more of an effect on hole
transport than electron transport, As time progressed and the a-Se samples relaxed, other
changes occurred, changes which affected the electron lifetime.
As Figures 6.23 and 6.24 attest, there is no significant difference between the two
plots in their electron and hole lifetime behaviour over time, even though the absorbed
dose in Figure 6.24 was in excess of ten times that in Figure 6.23. However, if the
behaviour of the effective electron transit time tr' as a function of IFTOF intermption
time ti is examined for these two cases, a difference clearly emerges, Figure 6.25 depicts
the dependence of the electron effective transit time on the interruption time for the data
of Figure 6.23 just prior to irradiation (a), just after irradiation (b), and two hours after
irradiation (c). At this low dose (1.8 mGy), the effective transit time exhibits no
significant change, with aU the plots exhibiting the same characteristic comer at ti - 200
ps as seen previously. Figure 6.26 shows the dependence of the effective electron transit
time on the intermption time for the data of Figure 6.24. Obviously, this higher dose
(24.9 mGy) has m effect as seen in Figure 6.26 (b) immediately following irradiation, as
the corner disappears and the dependence of tr' on ti becomes more-or-less linear. The
characteristic corner is restored two hours later.
It was stated earlier that the existence of this comer is evidence that the self-fieId
Qift of the photoinjected electron charge packet is strongly dependent on the number of
fke charges remaining in the packet. As some are removed due to deep trapping, the
spread of the packet (and the increase in tT3 slows accordingly. Therefore, the
disappearance of the comer immediately following irradiation could be due to one of two
things: either the irradiation altered the concentration or capture cross-section of the
shallow traps, or a significant bulk space charge interfefed with the mutual codombic
F = 1.96 V/mm during irradiation 1.8 mGy 0.263 L
0 2 4 6 8 10 12 120
Time (h)
Figure 6.23. Normalized hole and electron lifetime at the same depth tracked over time. Irradiation specifics: 58.2 keV, 1.8 mGy, F = 1.96 Wpm. [Sample 97 1002 121
- 582 keV F = 1 96 Vlmm during irradiation /'+
24.9 mGy 0.263 L T
Time (h)
Figure 6.24. Same sample as that of Figure 6.24, except the dose is now 24.9 mGy.
repulsion of the photoinjected electrons. Since the TOF electron transit time was
147
measured to be independent of x-ray irradiation, that rules out the possibility of a change
in the shallow traps somehow being the cause.
lnterruption Time (ps) lnterruption Time (ps)
0 200 400 600 800
Interruption Time (p)
Figure 6.25. Effetive electron transit time f ~ ' vs. ti for the data of Figure 6.23. (a) Immediately before irradiation, (b) immediately following irradiation, and (c) two hours after irradiation. At this low dose (1.8 mGy), there is no significant change in the behaviour of tr' with irradiation.
A bulk space charge of a single sign (positive or negative) will alter the electric
field within the a-Se sample, leading to a region where the photoinjected charge packet
will travel under the influence of a slightIy enhanced field, and another region with a
slightly decreased field. This condition results in photocments that either increase or
decay significantly, depending on the sign of the photoinjected carriers and the sign of
the bulk charge itself [102, 1151. Since neither the electron nor hole IFTOF
photocurrats noticeably changed shape during the course of the measurements, that d e s
Interruption Time (ps)
Interruption Time (p)
Interruption Time (p)
F i e 626. Effective electron transit time rT' us. ti for the data of Figure 6.24. (a) Immediately before irradiation, (b) immediately foIlowing irradiation, and (c) two hours after irradiation. At this high dose (24.9 mGy), the behaviour of rr' changes noticeably with irradiation, but recovers within two hours.
out the possibility of a bulk space charge of a single sign.
Nevertheless, the change in the effective transit time behaviour from being
noticeably dependent on the fiee photoinjected canier concentration before irradiation to
being relatively independent of it as in Figure 6.26 (a) and (b) would seem to indicate that
a bulk space charge is present to interfere in the zero-field diffusion of the photoinjected
electrons. The absence of a shift in the shape of the LFTOF waveforms rules out a bulk
charge of a single sign. Therefore, it is likely that x-ray irradiation induces a bulk space
charge within the a-Se film consisting of an approximately equal number of both positive
and negative charges. The experimental evidence shows that the concentration of this
bulk charge is proportional to the absorbed dose obtained fiom the x-ray beam, and that it
dissipates within two hours after irradiation.
Knowing how the charge carrier lifetimes within an a-Se radiographic detector
change over time after being irradiated with x-rays has scientific merit, but little practical
merit since a detector will likely be used many times throughout the course of a day in a
typical diagnostic setting. Therefore, if the carrier lifetimes of the a-Se samples are
continually polled as the samples are irradiated, the plots in Figures 627 - 6.29 are
obtained. Figure 6.27 shows how the electron lifetime within the samples varied as a
function of dose for two different beam energies. The data of Figure 6.27 was obtained
for the case of the samples being short circuited to ground while they were irradiated-
The plots of Figure 6.28 similarly detail how the electron lifetimes vary as a bct ion of
dose for two beam energies, except the samples were biased with an electric field of - 2
V l p while they were irradiated. The data found in Figure 6.29 was obtained while the
samples were biased with relatively strong electric fields.
The individual plots of Figures 6.27 - 6.29 do not substantially differ in their
behavior, in general it can be seen that the electron lifetime can be erratic at absorbed
doses < 5 mGy. At higher absorbed doses, - > 5 - 10 mGy, the electron lifetime
generally settles to a relatively constant level which is generally within the range of 70 - 100% of the lifetime before irradiation. Also, factors such as mean x-ray beam energy
and bias level during irradiation generally have little effect on the behaviour of the
electron lifetime vs. absorbed dose.
(a) Absorbed Dose (mGy) (b) Absorbed Dose (mGy)
0 5 10 15 20 25 30 0 5 10 25 20 25 30 35
Absorbed Dose (mGy) (dl
Absorbed Dose (mGy) (c)
Figure 627. Changes in the electron lifetime at two depths within an a-Se film as a function of absorbed dose. The samples were shorted to ground whiIe they were being irradiated. (a) 32.8 keV beam [Sample 971205 621. (b) 32.8 keV beam [Sample 971002 121. (c) 58.2 keV beam [SampIe 960521 46 SE15j. (d) 58.2 keV beam [Sample 971205 621.
58.2 keV F = 1-96 Vlpm
0.5
32.8 keV F = 2.16 Vlpm
0.5 0 5 10 15 20 25 30 0 10 20 30 40 50
Absorbed Dose (mGy) Absorbed Dose (mGy) (a) (b)
Absorbed Dose (mGy) (a
0.6 -( 52.8 keV - I F = 2.35 Vlpm
Absorbed Dose (mGy) (a
Figure 6.28. Changes in the electron lifetime at two depths within an a-Se film as a fimction of absorbed dose. The samples were biased with a field of - 2 V/pm while they were being irradiated. (a) 58.2 keV beam, F = 1-96 V/pm [Sample 971002 121- (b) 32.8 keV beam, F = 2-16 Vlpm [Sample 971205 621. (c) 58.2 keV beam, F = 2.16 V@n [Sample 971205 621. (d) 5 8 2 keV beam, F = 2.35 Vlpm [Sample 980622-31.
As stated earlier, the measurement of one l i f i i e took approximateIy 15 - 20
minutes; therefore, to M a t e a sample and measure the lifietimes at two depths within it
generally occupied the better part of an hour. The plots of Figures 627 - 6.29 therefore
each took about one 8 hour day to accumulate, and the changes in electron lifetimes over
time were undoubtedly also contributing to the observed changes.
0 2 4 6 8 1 0
Absorbed Dose ( m a ) (c)
F i i e 6.29. Changes in the electron lifetine at two depths within an a-Se film as a function of absorbed dose. The samples were biased with strong fields while they were being irradiated. (a) 58.2 keV beam, F = 7.84 Vim [SampIe 971002 121. (b) 582 keV beam, F = 8.66 Wpm [Sample 97 1205 621. (c) 5 8 2 keV beam, F = 17.1 V@n [Sample 980622 - 31.
How these changes in electron lifetimes would affect the performance of a
radiographic detector would be minimal at worst. The electron lifetimes observed in this
study never fell below - 60% of their rested value; therefore, the Schubweg would also
fall by - 60% in the worst case. This translates to a change in the minimum operating
field (introduced in Table 6.1) of 110.6 = 1.67; therefore, the minimum operating field of
the four samples studied would have to rise 67% to ensure adequate charge colIection.
These new minimum operating fields under the worst case electron lifetime condition are
presented in Table 6.3.
Table 63. Minimum operating field for the rested (best case) and damaged (worst case) electron lifetime conditions obierved during the study.
-
Sample I Best Cue Min. Openting ( Worst Cue Min. I
Again, since typical operating fields in a commercial digital x-ray imaging system
employing a-Se as the detector exceed 10 VICL~, the expected impact of the observed
decrease in electron l i f h e on charge collection would be minimal.
96052 1 46 SE 15 97 LO02 12 97 1205 62 980622-3
6.33 Light Induced Structural Changes in a-Se
Photo-induced effects in chalcogenide glasses (and a-Se in particular) are of
considerable technological importance, having applications in optical imaging, hologram
recording and optical mass memories. A number of different physical manifestations
resulting &om light exposure have been documented, among them light induced
crystallization [SO, 1161, suppression of photocrystallization by simdtaneous use of two
different lasers [117], polarization dependent photocrystallization [118], and revmi le
photo-amorphization [119,120].
Field (Vlw) 5.56 7.18 4.05 5.37
Perhaps the first photo-induced effect to be noticed was the phenomenon of
reversiile photodarkening, peculiar to the amorphous state of a material. Photodarkening
Operatha Field (V/pm) . 9.27 12.0 6.75 8.95
is the photoinduced increase in the low energy opticai absorption of a materid, induced
by light with an energy approximately equal to that of the bandgap of the material. It is
most often characterized by a nearly paraLleI shift of the low energy absorption edge (the
Urbach edge) to lower energy. Photodarkening can be reversed with time, annealing, or
by exposure to light of a different energy (photobleaching). This shift of the Urbach edge
must therefore be accompanied by an extension in the disorder-induced tail states into the
energy gap, thereby reducing the bandgap of the material.
These photostructural changes were originally presumed to be caused primarily
by the creation of IVAPs [49,51]; however when Tanaka (in 1980) [I211 examined all
available experimental evidence, he concluded that these structural changes could not be
explained by NAPS. He noted that the observed changes in the optical absorption
(which also coincided with changes in the volume and x-ray difhction pattern of the
material) must then be due to a bulk oriented feature of the disordered network; a
quantitative disorder such as fluctuations in bond length, angle or closed shell distance.
He introduced the concept of the double well potential to explain the phenomenon of
atom displacement, as pictured in Figure 630. He theorized that through some as-yet
unknown process, an atom originally in position A could be transformed to a new
metastable position A' in the solid. A relativeIy Iarge number of atoms being moved into
new positions in the solid would then significantly alter the overall randomness of the
material, thereby accounting for the observed changes in sample volume, x-ray
diffiction, and IR spectra.
Tanaka subsequently expanded his 1980 theory [122, 1231 with a possible
mechanism that would result in the movement of atoms to new positions through bond
twisting. Refer to Figure 6.31. Suppose the equili'brium configuration of the atoms
cotresponds to configuration (a). If a photon having an energy comparable to the
bandgap excites an electron fiom the lone pair (LP) of atom A (it is generally accepted
that the LP electrons fonn the top of the valence band while the anttinding orbitals
(AB) form the bottom of the conduction band [123]) to an electronic configuration
corresponding to configuration (b), there will exist a strong coulombic attractive force
F i r e 630. (a) A schematic diagram of Tanaka's 1980 [12 t ] model of bistable local bonding geometries and (b) corresponding double well potential.
between atom A and the LP electrons of atom B. This force has been estimated as being
- 1 eV [123], which may be enough to surmount the energy barrier faced by A in moving
to a new position as pictured in wdguration (c). This twisting motion can be completed
within lo-" xc [l23]. At that point, the excited electron recombines snd the resulting
twisted structure becomes 'Ybzen in" (d). This new structure is distorted in its
intermolecular bonding distance, giving rise to an increase in the randomness of the solid,
and a corresponding narrowing of the bandgap. It is very interesting to note that this
model relies on the presence of geminate EHPs to cause these bond flipping
photostructural changes.
A photoinduced EHP does not necessarily have only two posslile fates:
dissolution or recombination. An excited electron and the hole it leaves behind can
sometimes travel in unison; bound by their mutual wulombic attraction and contributing
nothing to the net dc current When an electron and hole act in this manner, tbey are
referred to as an exciton [124], and have hydrogenic-like energies. Street hypothesized
that an exciton could cause an WAP pair, as shown in Figure 6.32--he calIed this
situation a ''self-happed exciton" [125]. Biegelsm and Street [126] subsequently
supplied experimental evidence to support this theory.
Figure 631. Bond twisting model in a-Se. The equilibrium state (a) is altered when a LP electron is excited Erom atom A (b). A then feels a strong coulombic attraction to B, which twists A into a new position (c). The excited electron then recombines (d) and the structure is "hzen in".
It is important to realize that the shift in the optical absorption coefficient
coincident with photodarkening is not of sufficient magnitude for the phenomenon to be
caused by charges excited from the deep midgap traps associated with isolated VAP
defects, thus shallower energy states must be the cause. There is evidence to suggest,
however, that the close proximity of the S4 and S< defects in an IVAP wodd lead to
energy levels that lay shallower in the bandgap of a-Se than the deep traps due to isdated
VAP defects [126, 1271, Charges excited from these shallow traps would help to account
for the photodarkening phenomenon.
Se Chain
Before Self-Trapping
After Self-Trapping
F i e 632. Schematic diagram of the transformation of an exciton in Se into an IVAP pair accompanied by atomic distortion. The resulting NAP is sometimes referred to as a "self-trapped exciton". The resulting Sg and Ski- defects are commonly referred to as D' and D' defects, respectively.
Other photoinduced defects, stable at low temperature, have been reported. These
are primarily the fonnation of a pair of neutral threefold coordinated defm [128-13 11
which serve to cross-link Se chains, as shown in Figure 6.33. These defects are
metastable and tend to disappear quickly after the light is turned off and at elevated
temperatures. They may decay into either their original (ground) state, into stable new
twofoId bonds, or into an NAP as in Figure 6.33. Observations reported by Roy et al.
[132] support the second case (path I1 in Figure 6.33).
Other models for the photoinduced changes have been suggested, among them a
symmetry reversal of the pyramid centered at an overcoordinated atom to explain the
photoinduced optical anisotropy found in chalcogenide glasses [1333.
r(
Dark .4 \
tight Off
Figure 6.33. Formation of metastable triply-coordinated defects which serve to cross- Link adjacent Se chains. Antibonding electrons are denoted as "e". The metastable triplyaordinated defects may decay into (I) their ground state, (11) new bonds, or (m) into an NAP.
Oscillatory behaviour in the optical properties of amorphous GeSez films under
cw illumination h m a He-Ne laser has even been reported [134-1361. It was found that
under strict conditions, depending mainly on the incident power density of the laser, the
amount of light reflezted and transmitted by the a-GeSq films would oscillate at a rate
between - 3 - 50 Hz. These oscillations in the optical properties of the films were
accompanied by a simultaneous oscillation in the induced photocurrent. The oscillations
were subsequently explained as follows [137]: incident light fiom the He-Ne laser
induces photostructural changes leading to photodarkening. Once the transition to the
dark state has been achieved, the amorphous film absorbs considerable energy h m the
laser beam, causing it to undergo localized heating. This heating causes thermal
relaxation of the light induced defects, leading to a bleached state. Once the film is
bleached, the process begins anew. The photoinduced changes were assumed to be of the
type proposed by Tanaka [12 1-1231.
In 1993, Fritzsche [I383 examined all the evidence available at that time and
finally concluded that there is ample evidence that amorphous chalcogenides possess a
degree of medium range order (MRO). illumination by light tends to induce a large
amount of disorder, in agreement with Tanaka [I21 -1231, but also, to a lesser extent,
induces VAP or VAP-like defects. However, despite all the effort that has gone into
trying to understand this phenomenon over the last 20 years, it is likely that the true
origin of the photoinduced defects will remain unclear for some time.
Even though x-rays are far more energetic than the visible light usually employed
to induce structural changes in a-Se, it is stdl conceivable that they could induce those
same changes. As descriied in Chapter 4, an x-ray h t interacts with a solid by
producing an energetic '"hot" electron. This hot electron then travels in the solid, ionizing
many other e lemns before coming to rest. The ionizations produced by that hot
electron will undoubtedly have a range of energies, with some falling roughly in the same
range as those that induce photostructural changes-that is, energies on the order of the
bandgap. Thus x-ray induced photostructural changes in a-Se should be possible, as
confirmed by Tai et al. [139] (but with an eye toward employing these effects for VLSI
mimlithography). Relatively weak ionizations wodd likely not be solely resportstile for
the observed changes; energetic electrons or x-ray photons could also induce changes
through bond rupturing. It is unfortunate that the two broad fields of optics and
electronics do not mix for there is no published infomation as to how these
photostructural changes (ie. changes in the optical properties) of a material will affect the
charge transport properties of that material; this aspect of a-Se has not been
systematically investigated prior to this work.
633.1 Trapping Mechanisms and the Observed Results
In order to begin to grasp the structural mechanisms at the heart of the observed
changes in electron and hole trapping lifetimes upon x-ray irradiation, it is critical that the
trapping mechanisms themselves first be examined.
The first fact that mwt be stressed is that in the dark and at room temperature,
there exists no detectable ESR signal in a-Se [32, 33, 5 1, 121, 126, 140-1421. Since
dangling bonds (unpaired electrons) are the cause of an ESR signal, there therefore are
negIigible dangling bonds present in a-Se at room temperature and in the dark. Optically
induced ESR signals can be observed in a-Se at room temperature [32, 331, but they
disappear quickly after the light is turned off. Unfortunately, exactly how quickly they
disappear has been omitted fbm the literature. However, it can be estimated that the
relaxation would be on the order of seconds at most since Kolobov et al. [MO- 1421 found
that an optically induced ESR signal in a-Se at 20 K completely disappeared within 20
minutes when the sample was warmed to 150 K..
The traditional view is that in a-Se, VAPs or NAPS would be a possible cause of
deep charge trapping. This situation is presented in Figure 6.34 for the case of hote
trapping and Figure 6.35 for the case of electron trapping. Suppose that initially a
particular region of the a-Se material contains an isolated D' and D' defect. A
photoinjected hole would be electrostatically attracted to the singly coordinated D- defect,
creating an electrically neutral Se,O atom with an unpaired electron. Silarly, a
photoinjected electron would be electrostatically attracted to the triply coordinated D'
d e f a creating an electrically neutral ~e,' atom with an unpaired electron.
The immediately obvious problem with this traditional view of charge trapping is
that a single trapped charge will result in a smgle unpaired electron, contrary to dI
published observations involving ESR studies [32,33,51, 12 1, 126, 140-1421. It may be
argued that the concentration of photoinjected charge in a typical TOF or IFTOF
F i e 6.34. One possible view of hole trapping in a-Se. (a) Electrically neutraI a-Se with an isolated VAP defect and a drifting photoinjected hole. (b) Hole trapped by the D' defect resulting in an eIectrically neutral defect with a danghg bond.
experiment is very small and the resulting concentration of unpaired electrons would then
be correspondingly difficult to measure via an ESR test. However, it is very dear h m
the published data that these unpaired electrons are extremely unstable at elevated
temperatures, and it would be expected that very few unpaired electrons would exist at
room temperature, if any at all.
Figure 635. One possl'bIe view of electron trapping in a&. (a) Electrically neutral a-Se with an isolated VAP defect and a drifhg photoinjected eIectron. (b) Electron trapped by the D* defect resulting in an ekctrically neutral defkct with a dangling bond.
In order to resolve the observed absence of unpaired electrons with the traditiod
charge trapping model, it is necessary that it be modifled slightly. For instance, the
singly coordinated neutral defect resulting from a single trapped hole in Figure 6.34 (b)
could rid itself of its dangling bond by approaching a nearby chain, as in Figure 6.36.
The resulting neutral threefold coordimted defect has an unpaired electron which could
be immediatefy neutralized by a second photoinjected hoke in the vicinity, as in Figure
6.36 (b).
Figure 636. (a) The single trapped hole of Figure 6.34 (b) that results in a neutral singly coordinated defect may rid itself of its dangling bond by approaching a nearby chain and forming a neutral threefold defect. (b) The unpaired electron of the neutral threefold coordinated defect could be immediately neutratized by a nearby photoinjected hole to form a D' defect.
Similarly for electron trapping, the neutral threefold coordinated defect of Figure
6.35 (b) could rid itself of its dangling bond by releasing one of its bonds, resulting in a
neutral twofold coordinated atom at the site of the original D+ defect, and a nearby
electrically neutral singly coordinated defect as in Figure 637. The resulting neutral
singly coordinated defect has a single unpaired electron which couid be immediately
paired by a second photoinjected electron in the vicinity, as in Figure 6.37 (b).
This model is consistent with the published experimental evidence in that it
explains tbe absence of unpaired electrons in a-Se at room temperature. One can
hypothesize the foIlowing model: that charge trapping in a-Se does not involve a singIe
charge carrier, but involves a pair of carriers in order to maintain '8ond neutrality".
F i e 637. (a) The single trapped electron of Figure 6.35 (b) that results in a neutral triply coordinated defect may rid itself of its dangling bond by releasing a bond and forming a neutral twofold bond and a nearby neutral singly coordinated defect. (b) The unpaired electron of the neutral singly coordinated defect could be immediately paired by a nearby photoinjected electron to form a D* defect.
From xerographic and IFTOF experiments, Kasap et al. found that the capture
radii of deep hole traps in a-Se are about 2 - 3 A; on the onda of the Se-Se bond length
[143]. Since the capture radius of an IVAP or isolated VAP defect is expected to be at
least 2 - 3 times that of the bond length [143], deep hole trapping into these defects
would appear unlikely. indeed, it has long been suspected by some researchers that
trapping into VAPs or IVAPs was unlikely and that low cross-section electrically neutral
traps of unknown origin were the cause [5 1 1.
A possible hok trapping mechanism not involving a VAP or N A P defect is
illustrated in Figure 6.38. This new model builds on the observations of Kolobov et al.
[128-13 11 that a-Se tends to form a large number of neutral, paired, mpIy coordinated
defects when illuminated by light, Since illurninzition ha^ :-.- nhserved to create these
bonds, then perhaps hoIe trapping could accomplish the same thing. If a lone pair
electron on one of two relatively close twofold coordinated atoms on adjacent chains
were to be neutralized by a photoinjected hole as in Figure 6.38 (a), then those two Se
atoms could form a bond that Iinks the two chains, as in Figure 6.38 (b). This transient
situation would Ieave one atom with an unpaired electron which wouId have to be
neutralized by a second photoinjected hole to create a D' defect, as in Figure 6.38 (c).
This model is consistent with the observation that the hole trap cross-section is of the
order of a Se-Se bond length.
Figure 6.38. Two Se chains in which two twofold Se atoms are physically quite close to one another. (a) A photoinjected hole neutralizes a lone pair electron on one atom, leading to (b) a transient bond that cross-links the two chains as well as a single unpaired electron. (c) A second photoinjected hoIe neutralizes the unpaired electron, forming a pair of D' defects.
The deep trapping cross-section of electrons in a-Se has not been measured, so
there are no published results on which to base a rigorous theory. However, if it is
assumed that electrons are trapped not by VAPs or NAPS, but by traps of a small cross-
section as with holes, then the situation illustrated in Figure 6.39 couid be a posslhIe deep
electron trapping mechanism. In a-Se, the bond lengths between adjacent atoms is not
uniform; as a result, some bonds will be shorter than the average, and some will be
longer. Consider the case of two adjacent Se chains, as in Figure 6.39 (a). Some bonds
are relatively short while others are relatively long. If a photoinjected electron is in the
vicinity of one of these stretched bonds, it could be detained by one of the previously
twofold coordinated Se atoms to form a D' defect and a neutral singly coordinated defect
with one unpaired electron, as in Figure 6.39 (b). Then a second photoinjected electron
could be trapped by the neutral defect to form another D- defect with no unpaired
electrons.
F i i e 639. Two Se chains in which there exist stretched and compressed bonds. (a) A photoinjected electron nears a stretched bond to form (b) a D- defect and a neutral singly coordinated defect that has an unpaired electron. (c) The neutral singly coordinated defect diff ies a short distance before meeting a second photoinjected electron to form another D- d e f a
It should be noted that this new charge trapping theory--that trapping involves
the capture of two carriers-is simply the reverse of a previously reported emission
process. Both Melnyk [I441 and Baxendde and Jubasz [I453 speculated that an a-Se D*
defect could decompose into a D' defect through the emission of two h e holes.
With the above possible electron and hole trapping mechanisms in mind, the
results of the present study into how charge carrier lifetimes are affected by x-ray
irradiation may be summarized as follows:
neither electron nor hole mobility is affected by x-ray irradiation; hole lifetimes are immediately and dramatically affected by x-ray irradiation, independent of the irradiation conditions; electron lifetimes exhibit very little change immediately following irradiation by x-rays, independent of the irradiation conditions; both electron and hole lifetimes continue to change hours after being irradiated; these changes appear to be unpredictable in nature (sometimes rising, sometimes falling); the rested electron and hole lifetimes days after being irradiated are not necessarily equal to their pre-irradiated values, in some cases being 20% hisher, electron lifetimes sometimes behave erratically if the sample is continually irradiated, but generally exhibit little change with continued irradiation; behaviour of the IFTOF effective transit time (and thus dispersion) is disrupted by high doses of itradiation but unaffected by low doses-likely cause is a bulk space charge of equal numbers of positive and negative charges, and this charge disappears within two hours after irradiation.
There are several clues as to the trapping mechanisms at work in the a-Se samples
when subjected to x-rays. The most interesting is the observation that the FTOF
effective electron transit time changes from being highly dependent on the number of tiee
photoinjected carriers remaining in the packet to being independent of it when the
samples are irradiated with high x-ray doses (Figures 6.25 & 6.26). Since the carrier
mobilities were found to be unaffected by irradiation, the most likely mechanism for this
dependence is that the x-rays form NAPS witbin the a-Se film that interfkre with the
mutual coulombic repulsion of the free photoinjected carriers; the concentration of
NAPS is strongly dependent on the absorbed dose, as evidenced by the observation that
low doses do not change the behaviour of the effective IFTOF electron transit time upon
irradiatioa The formation of IVAPs is thought to be dependent on geminate EHPs [12S]
and the number of EHPs that escape geminate recombination is very low in a-Se, even at
moderate electric fields [73].
Therefore, the IFTOF effective electron transit time behaviour points to x-ray
irradiation initially creating a high concentration of lVAPs in the a-Se flhs, and this
concentration increases with the absorbed x-ray dose. Intuitively, this point makes sense
in that more x-ray irradiation would be expected to cause a larger change in the a-Se film.
The restoration of the original effective transit time behaviour within two hours is
evidence that these NAPS decay back to their equilibrium concentration fairly quickly
and must therefore be quite unstable at room temperature.
The next observation is that hole trapping is very dramatically and immediately
affected by irradiation, and this change is independent of dose for the limited range
employed in this study (refer to Figures 6.23 & 6.24). Electron lifetime was initially
found to change very little (C lOa/a), if at all, upon irradiation. The deep trapping
lifetimes of both electrons and holes were observed to continue to change over a period
of several hours after irradiation, and were approximately restored to their original values
several days later.
Since the hole lifetime immediately f d s upon irradiation, it could be possible that
the x-ray induced lVAPs detected through the electron IFTOF e f f d v e transit time
experiment would be the cause of this dramatic decline in lifetime. However, the hole
lifetime f d s just as dramatidly for low doses as for high, and the presence of NAPS
was detected only at high doses. It is important to remember that these NAPS vanish
within two hours, but the observed hole lifetime continues to fall well beyond two hours
later.
These facts support the theory that deep charge trapping does not occur at charged
defect centers. If charged defects were the cause, then both the hole and electron
lifetimes should dramatically fall upon irradiation, since irradiation produces a high
concentration of IVAPs. These NAPS disappear within two hours after irradiation, while
the electron and hole lifetimes continue to change; in some cases rising, in others falling.
Tanaka's bond twisting model [ 121 -1231 relies on geminate EHPs to achieve the
relocation of atoms to new metastable locations in the solid, and as stated previously,
geminate EHPs are in abundance in a-Se. Therefore, it is reasonable to expect that a
large number of atoms will be twisted into new positions by x-ray irradiation, and that as
time progresses some may be able to overcome the potential barrier and relax back to
their ground state position.
If, as a result of bond twisting, some atoms find themselves in relatively close
proximity to a neighbour, then the bond formation hole trapping mechanism of Figure
6.38 would likely become dominant. Indeed, since the hole lifetime dramatically falls
upon irradiation, this trapping mechanism wodd seem to be the cause. Another
interesting point is the fact that the hole Iifetime continues to change hours after the
IVAPs have relaxed. This would point to these metastable twisted bonds being more
stable (relatively speaking) than WAPs at room temperature.
Since the electron Iifetime does not drarnaticaIIy fall immediately upon
irradiation, electron trapping by charged defects (NAPS) cannot be the dominant
trapping mechanism, for the reasons cited above. If the stretched bond mechanism of
Figure 6.39 is dominant, then the bond twisting model of Tanaka [121-1231 cannot
significantly increase the number of stretched bonds in the solid. The reason for this
statement is the fact that the hole Lifetime significantly falls upon irradiation, likely due to
these twisted bonds, but the cotresponding electron lifetime is hardly affected.
As time progresses after the sample has been irradiated, both the electron and hole
lifetimes continue to change; sometimes rising, sometimes falling. This is indicative of a
mdom structural relaxation as the atoms of the a-Se film strive to tearrange themselves
into an energetically agreeable arrangement, but not necessarily their original
configuration; Le. those very same flipped atoms flipping back. At any one instant the
concentration of "close neighbod atoms may increase--Ieading to an increase in the
deep hole traps and a decrease in the hoIe Wedme, and at any other instant this
concentration may decreaseleading to an increase in the hole lifetime.
Correspondingly, the electron lifetime initially is hardly affected by irradiation itself, but
goes on to change hours later. At any one instant, the concentration of stretched bonds
may increas-leading to a decrease in the electron lifetime, and at any other instant this
concentration may decreas-leading to an increase in the electron Lifetime. Again, this
relaxation is a random process which depends on the local atomic arrangement of the
atoms, and that explains the observed behaviour of the electron and hole lifetimes over
time--sometimes falling, sometimes rising.
This theory also explains why the electron lifetime of a sample can sometimes
behave erratically as the sample is continually irradiated. M a t e the sample once, and
the atomic arrangements are disrupted. Irradiate it again, and the arrangements are
"shaken up" once more. The resulting atomic arrangements are random, depending on
which atoms were twisted where, and the corresponding lifetime wodd then also be
unpredictable since the concentration of traps would be highly dependent on the local
atomic arrangement.
Two very limited studies into the effect of x-rays on the hole lifetime in a-Se
tilms were done prior to this work. In the first [16], xerographic measurements of the
cycled up and first residual of a-Se plates were performed and it was concluded that the
number of hole traps increased as a result of exposure to x-rays. It was speculated that
the accompanying decrease in the hole lifetime was most Iikely due to trapped electroas
which could recombine with the photoinjected holes, thus leading to the reduced lifetime.
It shodd be pointed out that these xerographic measurements do not minimally disturb
the very traps they are meant to detect; instead, these measurements work by filling d l
available traps by repeated cycling until no change in the amount of charge retained by
the a-Se flh can be detected.
The second published study [60, 1001 observed that the hole lifetime in an a-Se
film, measured via the lFTOF technique, decreased by a factor of - 5 as a result of
exposure to x-rays. The large displacement currents during switching were eliminated
via balanced bridge techniques [58, 6062, 1001. These measurements can take a long
time to perfonn since the bridge is difficult to properly balance, as mentioned in Chapter
3. No indication of when the measurements were obtained relative to the initial
irradiation of the a-Se film are found in the references.
These previous results are consistent with this study. In both, it was found that
the hole lifetime decreases upon exposure to x-rays, consistent with the findings of this
study. However, it is known that minimal (if any) filtering was employed in the x-ray
beam of these studies, which could explain why the observed hole lifetime decreased by a
factor of - 5 in the second study (an 80% decrease), whereas the maximum decrease in
hole lifetime noted in this study was - 60%.
6.33.2 Irreversible X-ray Induced Damage
It is desirable that a radiographic x-ray detector not be adversely affected by
continual x-ray irradiation over time; in other words, that it not be fatigued by the x-rays
themselves. This requirement is desirable for obvious reasons. The rested electron
trapping lifetimes measured in this work were monitored over time to gauge the
posslhility of irreversible x-ray induced damage. Figures 6.40 and 6.41 illustrate how the
rested electron lifetimes at a depth of 0.263 L for Sample 971002 12 and 0.287 L for
Sample 971205 62 respectively varied h r n expaiment to experiment. The total
absorbed dose deposited in each sample just prior to each rested lifetime is also listed (the
open diamonds in the plots). The samples were allowed to rest in the dark at room
temperature for at least two days before being firrther irradiated. It is obvious fhm the
figures that there is no downward trend indicative of permanent damage. Instead, these
plots lend further credence to the earlier theory that the charge carrier lifetimes in an a-Se
6lm are dependent only on their local atomic arrangement. As such, both the behaviour
of the carrier lifetimes with time after irradiation and the rested lifetimes days later would
not be correlated and would be difficult to predict.
63.4 Electron Lifetime Recovery
One of the original gods of this study was to not only investigate how x-ray
irradiation affected the charge transport of a-Se films, but aIso if annealing, IR soaking
and ultrasonic treatment could be employed to recover the original properties of the film.
Experiment
800
h rn a. 750 - w
E .- 700 - 1
t 2 u
650 - iil u Q) u
600 -
550
Figure 6.40. The rested electron lifetime at a depth of 0.263 L (closed circles) of Sample 971002 12 tracked over the course of the work. The absorbed dose on the occasion immediately preceding each experiment is indicated with the open diamonds. The dose preceding experiment 6 is not known, as that was the exposure without any filtering placed in the beam (mentioned previously).
2
I 0 Dose
f 0 0
0 I 1 I I 1 I 1
1 2 3 4 5 6 7 8
Experiment
Figure 6.41. The rested electron lifetime at a depth of 0.287 L (closed circles) of Sample 971205 62 tracked over the course of the work. The absorbed dose on the occasion immediately preceding each experiment is indicated with the open diamonds.
6.3.4.1 Annealing
Annealing the samples at 35 OC had a dramatic effect on their charge transport. In
general, electron transport deteriorated greatly and displayed a decline in both the deep
trapping lifetime and the mobility. The hole mobility displayed no discemable change,
but the hoIe lifetime was affected to a greater degree than the electron lifetime. This data
is presented in Figures 6.42 and 6.43.
All three qualities-the electron and hole lifetimes and the electron mobility-do
not totally recover back to their original values. This would be indicative of irreversible
changes induced by the time spent at an elevated temperature. The elevated temperature
could conceivably alter the local atomic arrangement of the a-Se film, leading to the
observed changes in the carrier lifetimes. These observations are in agreement with
previous studies which reported that the drift mobility in a-Se is strongly temperature
dependent [I461 and that the bandgap decreases at elevated temperatures [147]. This
latter study implies that the decrease in the bandgap is due to an increase in disorder
within the d i d , consistent with the bond-flipping theory of Tanaka [12 1 - 1231. Thus the
obsmed decrease in the hole and electron Hetimes upon anneding lends hther
credence to the carrier trapping theory introduced earlier.
When a sample was first irradiated by x-rays and then annealed at 35 OC, the plot
of Figure 6.44 was obtained. The electron lifetimes at two depths within the sample were
initially measured (stage 1 of Figure 6.44) and then the sample was irradiated with x-rays.
The lifetimes at those same depths were again measured (stage 2) before the sample was
annealed for 3 hours at 35 "C (stage 3). It appears that annealing does not aid electron
lifetime recovery after a sample is irradiated with x-rays.
Total Time Spent at 35 C (h) I 4
0.6 ' I I
24 48 72
(b) Recovery at Room Temperature (h)
Figure 6A2. (a) Normalized electron (solid circles) and hole (open diamonds) lifetimes as a hction of total the spent at 35 "C. (b) The room temperature recovery of those lifetimes tracked over a three day period following the heat treatment. Both Iifetimes were measured at the same depth within the film 0.263 L. [Sample 971002 121
Total Time Spent at 35 C (h)
(b) Recovery at Room Temperature (h)
Figure 6.43. (a) Normalized electron mobility as a hction of total time spent at 35 "C. (b) The mom temperature recovery of the mobility tracked over a three day period following the heat treatment. [Sample 971002 121
6.3.4.2 IR Soaking
Soaking an a-Se sample with IR light has proven usefid in discharging x-ray
induced bulk space charges [102, IZTJ, so an investigation into whether IR soaking
would aid electron lifetime recovery was wananted.
Rested d Exposed P Anneal
Stage
Figure 6.44. Nomdized electron lifetime at two depths within an a-Se sample before irradiation (stage i), immediately after irradiation (stage 2), and following 3 hours spent at 35 O C (stage 3). Irradiation specifics: 58.2 keV, 35.7 mGy, sample shorted during irradiation. [Sample 96052 1 46 SE 151
Even short periods (- 30 seconds) of IR soaking would noticeably raise the
temperature of the sample through radiant heat transfer. It is therefore quite difficult to
discern if the observed effects were due to the action of the IR photons alone, or a
combination of the IR photons and sampIe heating. In general, the electron lifetimes
decreased as a result of IR soaking, but this was expected given that other researchers
have reported that sub-bandgap light decreased the deep trapping lifetime of both
electrons and holes in a-Se [51]. The d t s are presented in Figure 6.45.
63.43 Ultrasonic Treatment
Very early in the study, ultrasonic treatment (at 4 MHz as mentioned in Chapter
5) was conceived as a very simple prucedure to perform on an a-Se sample to investigate
whether or not the ultrasonic treatment did indeed have any effect on the charge transport
properties of the film. As with IR soaking, it was difficult to establish whether the
observed changes were caused by the ultrasonic waves themselves, or due to sample
heating. Unfortunatelyy during the 15 minute ultrasonic matment, the ultrasonic
P pa 0.9 Rested
Exposed
30 sec Soak
0.5 - (a)
0.4 r I
1 2 3
Stage
1 .l
1.0 - !! Exposed 0.401 L
0 0.603 L 0s9 - Rested w
8 0.8 - .-
E l -
0.7 - 90 sec Soak 5 Z 0.6 - @
0.5 - (b) P
Stage
45. Normalized electron lifetime at two depths within two differen samples. (a) Before hradiation (stage I), immediately after irradiation (stage 2), following 30 seconds of IR soaking (stage 3). Irradiation specifics: 58.2 keV, 96.6 mGy, sample shorted during irradiation and during soaking. [Sample 97 1002 121 (b) Before irradiation (stage l), immediately after irradiation (stage 2), following 30 seconds of IR soaking (stage 3), following 90 seconds total IR soaking (stage 4). Irradiation specifics: 582 keV, 28.6 mGy, sample shorted during irradiation, but F = 4 Vljm during IR soaking. [Sanq.de 960521 46 SE I S]
transducer noticeably warmed (to - 40 "C), and this undoubtedly had an effect on the a-
Se sample.
The electron and hole lifetimes of an a-Se film at a single depth in the sample
were first measured, and then the sampIe was subjected to a 15 minute ultrasonic
treatment. The electron and hole lifetimes were again measured at the same depth in the
sample immediately foIIowing the ultrasonic treatment, and two hours thereafter. The
results are found in Figure 6.46.
F i r e 6 Normalized electron and hole lifetimes at the same depth (0.401 L) within an a-Se sample initially (stage l), following 15 minutes of ultrasonic treatment (stage 2), two hours thereafter (stage 3). [Sample 96052 1 46 SE I S]
P 2 h Later
It is clear fiom the figure that the electron and hole lifetimes were adversely
affened by the treatment, but it is difficult to discern whether this directly resuIted f?om
the dtrasonic treatment itseIf or the inadvertent heating of the sample during treatment.
However, it is felt that the observed changes are likely caused by sample heating. This
part idar aspect of the behaviour of aSe should be investigated W a if an apparatus
could be designed that maintained the sample at a constant temperature while it was
being subjected to uItrasouic waves.
a 0.9- P" o 0.8 - 8
Rested Z
= 0.7 (P
0.6 pi 0
0.5 - P 15 min Ultrasonic
6.4 Electron Hole Pair (EBP) Creation Energy - WEHp
The mechanism(s) underlying the liberation of mobile charge carriers in a-Se by
x-rays has both a practical and a scientific significance. Practically, it is desired to
achieve the maximum possible collected charge when irradiating a-Se with x-rays in
order to achieve high quality radiographic images with low x-ray doses. This particular
quality is arguably the biggest controlling factor in the sensitivity of a-Se radiographic
detectors. Scientifically, one desires to understand exactly how x-rays come to liberate
free charges within a-Se, and wbat changes take place in the solid when this occurs.
The integration of the x-ray induced photocurrent when a biased a-Se film is
irradiated with x-rays reveals the number of carriers keed by the incident x-rays. Figure
6.47 pictures a typicd x-ray photocurrent produced by the ten impulse x-ray "quanta"
used throughout the work. There are two fatures of Figure 6.47 that are immediately
apparent: the photocurrent "spikes" caused by the self-rectifying x-ray tube, and a subtle
''rising baseline" which increases slightly with every x-ray burst. This rising baseline
corresponds to a persistent x-ray photocurrent. The exact cause behind the rising baseline
is not known, though it is unlikely to simply be carriers released &om traps, as previously
proposed [67, 681. There may be a contribution arising from increased contact injection
[73], although it is unlikely that this is the primary hctor. WEHP was evaluated by
excluding the charge in this rising baseline, consistent with other studies [67, 681. The
rising baseline will be examined in detail in section 6.5.
Another feature of Figure 6.47 are the nmt spikes at tie beginning of the
pulsetrain. This aspect of the x-ray photoments was a phenomenon unique to the x-ray
unit employed in the work, and not the r d t of the heavily filtered beams used
throughout the study. It is suspected that these smaller leading spikes are the
consequence of the x-ray tube somehow ''warming up" because these spikes were still
observed even if the filtering were removed* These small spikes were also observed in
the x-ray photocurrents induced in the CZT calkator detector, so their existence being
unique to a-Se is also ruled out.
Photocurrent "Spikes"
Rising Baseline
0.10 0.15 0.20 0.25
Time (s)
F i e 6.47. A typical x-ray induced photocurrent in a biased a-Se film, detailing the "spikes" due to the self-rectifying nature of the tube and the "rising basehe*. Data obtained with a 58.2 keV beam and Sample 1463-3 biased at 1.77 V l w .
The Gendex x-ray unit employed in this work is not the highest quality x-ray unit
available; as such, its output varied slightiy as evidenced by the non-uniform spikes
comprising the latter part of the puIsetrain shown in Figure 6.47. To reduce the effect
this variance in the beam output had on the charge collected h m the x-ray photocurrent
(and thus, the error in WEHP), all measurements of WmP were based on the average of 10
individual measurements, each of a 10-impulse burst Once this average collected charge
was found, the energy deposited in the a-Se layer was found via the method described in
Chapter 5 and the electron hole pair creation energy, Wmp, was calculated.
a-Se is well known for its field dependent photogeneration. A large fraction of
photogenerated carriers are lost via various recombination processes thought to exist in a-
Se, which means that the measured WEHP is an effective value needed to create collected
EHPs. The intrinsic energy to create an EHP, wLp, corresponds to no recombination
losses which is expected to be the case at very high fields. The theoretical intrinsic
energy bYO, via Klein's rule [82, 831 may be used as a starting point to establish a value
for w&. As desmied in Chapter 4, this energy will be about 2.8 E, plus a small (0.5
eV) phonon texm. With E, = 2.2 eV for a-Se, whP = 6.7 eV. If the conservation of
momentum rule is relaxed in a-Se, as argued by Que and Rowlands [70], wLp = 2.2 E,
plus a small phonon texm. This leads to wLP = 5.3 eV in a-Se. An estimate of WL in
a-Se may be found by extrapolating the WUIP VS. l lF behaviour to 11F = 0.
The energy to create a collected EHP as a hct ion of the reciprocal electric field
for four different beam energies is presented in Figure 6.45. The strong field dependence
is very evident fiom the plot, and the number of collected charges increases with
increasing field strength. The linear regressions for each data set all remarkably converge
to within 0.5 eV at infinite field, or the expected value of wL,, . This convergence in the
data gives an average WL = 5.9 eV, in reasonable agreement with the theoretical values
of 5.3 - 6.7 eV introduced above, This value is also in close agreement with the previous
reported result of 5.9 eV [73, 1023. Asswning an operating field of 10 Vlpn, Wmp has
been reported to be in the rage of - 35 - 55 eV over the diagnostic beam energy range
[70-751, also in close agreement with the present results of - 50 eV.
It was found that Wm did not si&cantly vary as a hct ion of the mean x-ray
beam energy over the limited range employed in this work. Figure 6.49 is a plot of Wm
at a constant electric field of I0 VIpm (0.1 CMN) vs. the mean x-ray beam energy.
There is a very slight energy dependence which tends to lower values of WEHP at higher
beam energies. If a hea r regression is applied to the data (as shown), the slope of the
h e yields a WUIP dependence upon the mean beam energy of -0.05 eVkeV. Given the
Figure 6.48. The energy required to create a 6ee EHP as a function of the reciprocal electric field for four different mean x-ray beam energies spanning the range 32.8 - 58.2 keV. Linear regressions for the four sets of data converge to within 0.5 eV to yield an expected W&,, = 5.9 eV at infinite field. [Sample Run #30]
limited energy range of this study, this value is comparable to Fieidler's reported
dependence of 4.24 eVkeV which was obtained over a range spanning h m - 30 keV
to 200 keV [71]. It should be noted that a similar analysis was performed on the
remainder of the data found in Figure 6-48. It revealed that all dopes of the linear
regressions were negative. The average d u e of this energy dependence was found to be
4 - 0 2 1 eV/keV.
I Slope: -0.05 eV/keV
30 35 40 45 50 55 60
Mean X-ray Beam Energy (keV)
Figure 6.49. Dependence of Wmp on the mean x-ray beam energy at a constant field of 10 V / p .
The a-Se sample which was used to obtain the WwP vs. reciprocal field plot of
Figure 6.48 has a minimum operating field (Table 6.2) of 1.77 Wpm. FieIds above this
minimum value will generally ensure that no charges are lost to trapping, therefore,
trapping may be neglected as a charge loss mechanism.
There are essentially two expected recombination processes that may be
responsible for the charge Ioss and would thus explain the strong field dependence of
WW. These are geminate and columnar recombination, mentioned previously in Chapter 4. In geminate recombination the twin excited eIectron and the hole it leaves
behind recombine with each other [17,70,73, 1483, whereas in columnar recombination,
electrons and holes within the track of a primary electron drift and recombine within that
m k [70, 1493.
Que and Rowlands set forth a number of requirements for both geminate and
columnar recombination in order to determine which is the likely controlling process for
x-ray photogeneration in a-Se [70]. Geminate recombination is strongly temperature
dependent because the largest controlling factor which determines whether the
photogenerated electron and hole dissociate is the initial separation of the two charges, r,.
This initial separation increases with increasing temperature, meaning that the probability
of dissociation, and thus the yield, will decrease with decreasing temperature. On the
other hand, columnar recombination is relatively insensitive to temperature.
Figure 6.50 is a plot of WUIP US. temperature which was measured at a constant
electric field of F = 1.77 Vlpn and a mean beam energy of 58.2 keV. The measurements
span a temperature raage from - -60 "C to room temperature. It is very obvious that
WEHP exhibits a very strong temperature dependence that decreases the yield (and thus
increases WEHP) as the temperature is lowered. These results are generally indicative of
geminate recombination being the dominant recombination mechanism in a-Se when
irradiated by x-rays. Recently, Haugen et ti/. [73] reported no observable WEHp
temperature dependence in a-Se, but the temperature range of the study was extremely
limited (-10 OC to room temperature). However, their results are consistent with the
behaviour observed in this study, as WEHP shows only a weak temperature dependence
down to - -30 OC, after which the association becomes much stronger.
It is common to refer instead to the photogeneration efficiency as opposed to the
EHP creation energy, Wm. The photogeneration efficiency, q, can be defined as the
ratio of generated EHPs that escape recombination relative to the total number of created
EHPs. This may be expressed as a ratio of intrinsic and apparent EHP creation energies
200 220 240 260 280 300 320
Temperature (K)
Figure 630. Plot of the energy required to create a collected EHP, WW, as a function of temperature. The mean beam energy was 58.2 keV and the electric field was held constant at I.77 Vlpn. [Sample 1463-31
The large value of Wm and its field dependence lead to a field dependent
photogeneration efficiency with a value less than unity. Figure 6.51 plots the
photogeneration efficiency vs. temperature for the data of Figure 6.50. The
photogeneration efficiency is very small at low temperature, about 0.2% at - -60 "C,
rising to - 4% at room temperature.
200 220 240 260 280 300 320
Temperature (K)
Figure 6.51. Photogeneration eFticiency as a function of temperature. Mean beam energy: 58.2 keV, sample biased at 1.77 V/w. [Sample 1463-31
As stated earlier, the initid separation of the EHP controls the probability that
they will escape their mutual wulombic attraction and become free. An optical photon
can only produce a single EHP, and this initial separation will therefore be dependent on
the photon energy. In other words, the initial separation will be the same for each EHP.
When an x-ray photon interacts with a sotid, thousands of EHPs can be created and their
initial separations will span a range. The distriiution of these separations shouId have
little or no dependence on the x-ray photon energy, as the EHPs are created randomly.
Therefore, WmP should be independent of the mean x-ray beam energy if geminate
recombination were the dominant recombination mechanism. As was seen in Figure
6.49, WmP shows very M e , if any, dependence on the x-ray beam energy. Again, this
fact may be indicative of geminate recombination being the dominant recombination
mechanism in a-Se.
Columnar recombination should exhiiit a pronounced decrease in Wmp with
increasing photon energy. Columnar recombination in a-Se was modeled by S a h p
[l SO], and he found that the charge density within the primary electron track decreased as
the x-ray photon energy increased, and was due to an increase in the mean separation of
EHPs within the column. This reduction in the concentration of h e carriers within the
column then directly results in a decrease in the recombination which in turn reduces
WmP. A direct comparison between the present results and Sahyun's predicted values
cannot be made since Sahyun's values extend only up to fields of - I V i m and this
study started at a field of 10 V i p . However, it is readily obvious that Sahyun's
calculated Wmp at - 1 V/pn of - 13 - 27 eV for beam energies in the range of 26 - 42
keV is far too low; this does not agree with the reported measurements of WEHP at 10
Vlpm of - 35 - 55 eV [70-751, nor with the results of this study.
There has existed a controversy over whether geminate or columnar
recombination is dominant in a-Se, but the results of this study seem to favour geminate
over columnar recombination. The very pronounced temperature dependence of Wmp
and the lack of a strong dependence on the x-ray beam energy is strongIy suggestive that
geminate recombination is the controlling charge loss mechanism in a-Se when irradiated
by x-rays, consistent with the accepted mechanism at optical energies [17].
6.5 Persistent X-ray Induced Photocurrent
Immediately fbIlowing x-ray irradiation, there exists a persistent photocurrent in
a-Se whose origin has been attributed to the thermal retease of deeply trapped holes [67,
681. This persistent photocurrent is problematic in a practicaf setting because it places a
limit on how quickly the a-Se detector may be again used following an exposure. If the
detector is not allowed to rest for a sufficient length of t h e (about three minutes [67,
68]), image ghosting may occur, whereby a remnant of the previous image will appear in
the present image.
This study of the persistent photocurrent was triggered quite by accident when it
was discovered that the magnitude of the rising baseline mentioned earlier remained quite
large at low temperatures, even though the x-ray induced photocurrent spikes decreased
dramatically. Figure 6.52 is a plot of the charge contained in the rising baseline vs. the
sample temperature. The details of the irradiation are the same as the WEHP VS.
temperature study presented in section 6.4: 58.2 keV, 1.77 V/pm, and Sample 1463-3. It
is immediately apparent that the basetine exhi'bits a temperature dependence which is
quite different &om that displayed by the photocurrent spikes, which have also been
plotted for comparison.
Baseline 0 Photocurrent Spikes
200 220 240 260 280 300
Temperature (K)
Figure 6.52. Plot of the charge contained in the baseline and photocurrent spikes vs. temperature. 58.2 keV, 1.77 Wpm, Sample 1463-3. The two exhibit markedly different temperature dependencies.
It was subsequently discovered that the charge contained in the rising baseline
was thermally activated. Figure 6.53 is a semi-Iog pIot of the rising baseline charge as a
fimction of 1/T. The baseline charge saturates at higher temperatures, but shows an
activated type of behaviour at temperatures below approximately -20 OC. This activation
energy is - 0.16 eV; since the deep hole traps in a-Se lie at - 0.8 eV, the persistent
photocurrent cannot be caused by the thermal release of deeply trapped holes (or
electrons for that matter).
Figure 6.53. Baseiine charge as a fimction of the reciprocal of the temperature. The amount of charge released in the persistent photoment becomes thmal ly activated below approximately -20 O C . The activation energy is 0.16 eV, which does not correspond to the deep hole or electron traps in a-Se.
The rather surprising result that the persistent photocurrent is thermally activated
fiom a previously unknown level in the bandgap lends fbrther credence to the theory of
photoinduced structural changes and that charge trapping is not into charged defects as
put forward in section 6.3. Intuitively, however, this makes sense. Irradiation by x-rays
creates transient metastable bonds or can rupture previously existing bonds leading to
metastable d e f m with unpaired electrons [128-132, 140-1423. Those unpaired electrons
(dangling bonds) are highly unstable and are released as these metastable defects
decompose into more stable forms: either an Se,' or an Se; defect. It is likely that the
release of these charges is the origin of the persistent photocunent, and is a temperature
activated process, in agreement with the observation that an ESR signal may be “frozen"
into a-Se at very low temperatures by Light [ 140-1 421.
6.6 Summary
The charge transport properties of various stabilized a-Se i3ms were evaluated via
the TOF and ETOF techniques to ensure that they had acceptably long carrier
Schubwegs so that charge trapping would not be a factor in the studies that followed.
It was found that x-ray irradiation has no effect on the TOF transit time of holes
or elmons, meaning that x-rays do not affect the shallow gap states in a-Se. CFTOF
studies of both electrons and holes suggest that irradiation produces NAPS, and their
concentration increases with the absorbed dose of the Miation. These NAPS disappear
quickly (within two hours after irradiation) and do not play a part in the deep !rapping of
holes or eiectrons, which was proposed to take place through bond rearrangements. The
electron and hole lifetime of an a-Se film continues to change hours after being irradiated
by x-rays, and this can only occur by way of a structural rearranging of the soIid as it tries
to assume a configuration consistent with a low stmctmd energy. This behaviour was
previously observed as a result of optical excitation, but the corresponding cbarge
transport properties of a-Se with these photostructural changes in pIace had not been
investigated prior to this work. La general, hoIe tramport deteriorates much more with x-
ray irradiation than electron transport. The observed changes in carrier lifetimes were not
of a sul3cient magnitude to affect the performance of a commercial radiographic
detector.
Both hole and electron transport was found to deteriorate as a result of amding,
IR soaking and ultrasonic treatment. Except for the case of ultmonic treatmeat (which
had not been investigated previously), these observations were in agreement with
reported results.
The e f f d v e EHP creation energy, WEHP, was determined by measuring the
number of free carriers generated through irradiation of a biased a-Se film and dividing
that number by the energy deposited in the a-Se layer by the x-ray beam. WEHP has a
pronounced field and temperature dependence which suggests geminate recombination as
being the dominant charge loss mechanism. The intrinsic EHP creation energy, w&, was found to be 5.9 eV, in close agreement with theoretical and previously published
values.
The persistent x-ray induced photocurrent was examined over a wide temperature
range, and it was found to exhibit a thermally activated behaviour, with an activation
energy of 0.16 eV. This energy does not correspond to a level of any known Craps in the
bandgap of a-Se. It is probably associated with the x-ray induced photostructural changes
that were used to explain the behaviour of the charge lifetimes upon exposure to x-rays,
or perhaps changes in the metaVa-Se contact behaviour.
7. Conclusions and Recommendations
7.1 Introduction
The objective of this work was to study the charge generation, transport and
trapping propexties of stabilized a-Se films as they pertain to its use as an x-ray
photoconductor. The TOF and IFTOF transient photoconductivity techniques were
initially employed to study the carrier drift mobility and deep trapping lifetimes of the
films. The x-ray induced photocurrent was also investigated using a high gain cunent-to-
voItage (I-V) converter.
The implementation of practical selenium-based solid-state x-ray imaging systems
requires a sound understanding of the charge carrier generation and transport behaviour
in the presence of x-rays. The optical properties and behaviour of opticalIy generated
charge carriers in a-Se films have been extensively studied over the last four decades,
primarily due to the use of a-Se in xerography in the 1960s and 70s. This study strove to
answer some of the questions raised by the use of selenium in the presence of x-rays.
7.2 Charge Transport Study
The a-Se samples were investigated to ensure that their charge transport
properties were acceptable for use in x-ray radiographic image detectors. Electron
transport was found to be the main controlling fictor in the determination of the
Schubweg of the sample. Each sample employed in the study had charge transport
properties comparable to that of a commercial device quality film.
Electrons were discovered to undergo very significant dispersion during either
TOF or IFTOF studies. The total spatial width of the electron charge packet is
comparable to the thickness of the film itself. Consequently, a plot of the hctionai
recovered electron photocurrent vs. LFTOF intemption time did not pass through the
origin for all samples used in the study. The dispersion of the electron charge packet
during an IFTOF procedure was found to depend very strongly on the concentration of
free carriers remaining in the packet. This suggests that the self-field of the photoinjected
carrier packet contniuted substantially to the dispersion.
7.3 Changes in Charge Transport Upon Exposure to X-rays
It was found that x-ray irradiation has no effect on the TOF transit time of holes
or electrons, meaning that x-rays do not affect the shallow gap states in a-Se. IFTOF
studies of both electrons and holes suggest that irradiation produces IVAPs, and their
concentration increases with the absorbed dose of the irradiation. These IVAPs disappear
quickly (within two hours after irradiation) and do not play a part in the deep trapping of
holes or electrons, which was proposed to take place through structural rearrangements.
The electron and hole lifetime of an a-Se 6lm continues to change hours after being
irradiated by x-rays, and this can only occur by way of a structural remangement of the
solid as it tries to assume a configuration consistent with a low structural energy. This
behaviour was previously observed as a result of optical excitation, but the corresponding
charge transport properties of a-Se yith these photostructural changes in place had not
been investigated prior to this work. En geueraI, hole transport deteriorates much more
with x-ray irradiation than electron transport. The observed changes in carrier lifetimes
were not of a sufficient magnitude to significantiy affect the performance of a
commercial a-Se based radiographic detector.
Both hole and eIectron ttansport were found to degrade as a result of annealing,
IR soaking and ultrasonic treatment. Except for the case of ultrasonic treatment (which
was not previously investigated), these observations were in agreement with reported
results.
7.4 EBP Creation Energy
The effective EHP creation energy, Wm, was determined by measuring the
number of fiee carriers generated through irradiation of a biased a-Se film and dividing
that charge by the energy deposited in the a-Se layer by the x-ray beam. WW has a
pronounced field dependence consistent with previously reported Watp vs. l/F
behaviour. Below - 240 K, WEHP was observed to exhibit a strong temperature
dependence which seems to suggest geminate recombination as being the dominant
charge loss mechanism. The intrinsic E W creation energy, w & ~ , was found by
extrapolation to be about 5.9 eV, in close agreement with theoretical and previously
published values.
7.5 Persistent X-ray Induced Photocurrent
The persistent x-ray induced photocurrent was examined over a wide temperature
range, and it was found to exhiiit a thermally activated behaviour below - 240 K, with an
activation energy of 0.16 eV. This energy does not correspond to a known level of traps
in the bandgap of a-Se. It is possl%Ie that it is related to the nature of charge trapping
proposed to be coincident with x-ray induced structural changes in a-Se, or x-ray induced
changes in the metda-Se interface (e.g. enhanced contact injection).
7.6 Suggestions for Future Work
It was discovered during the course of this work that the ekctron and hole
lifetimes depended very strongly on the temperature of the a-Se fib itself. To exclude
any lifetime variations imposed by the varying ambient temperature of the laboratory, a
proper cryostat should be employed which could be used in conjunction with the existing
lead-lined x-ray cabinet. This cryostat would also make the measurement of WEHP and
the baseline charge (persistent photocurrent) over a wide temperature range much easier
to perform, These particular measurements were the most difficult to perform during this
work, as condensation from atmospheric water vapour made the measurements very
challenging.
Although the MOSFET switches performed quite satisfactorily, their stray
capacitances slightly limited the recovery speed of the IFTOF waveforms. CommercialIy
available MOSFET switches have lower stray capacitances and should improve
performance of the system. Commercial switches are also attractive for the simple fact
that they do not have 12 individual power switches and they do not require time-
consuming battery changes every three weeks of continuous operation.
In order to vet+@ that the release of unpaired electrons is the true cause of the
persistent x-ray photocurrent, an ESR study must be performed. However, these ESR
measurements would be difficult to perform given that the persistent photocurrent, and
thus the unpaired efectrons, disappear very quickly after irradiation. This means that tbe
ESR study will have to be performed immediately following irradiation (within seconds).
Essentially, an apparatus that would allow the simultaneous x-ray irradiation of the
sample and an ESR measurement is needed.
The theory underlying photostructural changes in a-Se is quite well developed and
the results of this study would seem to indicate that the same structural changes can also
be induced by x-rays. In order to have a direct comparative study, it would be instructive
to irradiate not just the thick a-Se samples used for the charge transport studies, but aIso
some very thin samples so that the optical properties of a-Se could be studied as well.
This would be useful to explore if x-ray irradiation induces such phenomena as
photodarkening. When used in conjunction, these two studies could provide clarification
as to the cause of both photodarkening and charge trapping in a-Se.
It is apparent that although a-Se was extensively studied by researchers at Xerox
during the 1960's and 7OYs, that there are still many aspects of its properties that are not
understood. Most theories concerning amorphous semiconductors were developed
mainly during the 1980's through to the present day. During this time a-Si:H has usurped
most of the attention, but this study would indicate that a reexamination of a-Se is
warranted.
8. References
Ovshinski S.R, "Amorphous semiconductors for microelectronics", Proceedings &om SPIE, 617, 1986, pp. 1-9.
Cowen A.R., "Digital x-ray imaging", Measurement Science and Technology, 2, 199 1, pp. 69 1-707.
Bronzino J.D., Biomedical Engineering and Imtmmentation: Basic Concepts and Applications, PWS Engineering, Boston USA, pp. 403.
Boag J.W., 'Xeroradiography", Physics in Medicine and Biology, 18, 1973, pp. 3-37.
Thourson T.L., 'Xeroradiography", Symposium on Medical X-ray Photo-optical System Evaluation, SPIE, 56,1975, pp. 225-235.
Chan HI., Doi IC, Galhotra S., Vybomy C.J., MacMahon H., and Jokich P.M., "Image feature analysis and computer-aided diagnosis in digital radiography. I, Automated detection of microcalcifications in mammography", Journal of Medical Physics, 14, 1987, pp. 538-548.
Jeromin L.S. and Klynn L.M., "Electronic recording of x-ray images", Journal of Applied Photographic Engineering, 5,1979, pp. 183-189.
Schiebel U., Hillen W. and Zaengel T., "Image quality in selenium-based digital radiography", SPLE Proceedings . Medicine MV, SPIE, 626,1986, pp. 176- 184.
Neitzel U., Maack I. and Giinther-Kohfahi S., 'lmage quality of a digital chest radiography system based on a selenium detector", Journal of Medical Physics, 21, 1994, pp. 509-516.
Davis H.J., Araj N., Rowlands JA. and Hunter DM., "A novel, digitaI radiographic imaging system using selenium plates", Proceedings of the Fourth International Symposium on Uses of Selenium and Tellurium, edited by S. Carpello, Selenium-Tellurium Development Association, New York, 1989, pp. 267-280.
1 I] de Monts H. and Beaumont F., "A new photoconductor imaging system for digital radiography", Journal of Medical Physics, 16,1989, pp. 105-1 08.
Zhao W., Rowlands LA, Germann S., Waechter D. and Huang Z., "Digital radiography using self-scanned readout of amorphous selenium: design considerations for mammography", SPIE Proceedings: Medical Imaging 1995: Physics of Medical Imaging, SPIE, 2432, 1995, pp. 250-259.
Lee D.L., Cheung L.K. and Jeromin L.S., "A new digital detector for projection radiography", SPLE Proceedings: Medical Imaging 1995: Physics of Medical Imaging, SPIE, 2432, 1995, pp. 237-249.
Schiebel U., Wieczorek H. and Brauers A., "X-ray image detector", United States Patent Office, Patent Number 5,396,072, 1995.
Rowlands J., and Kasap S.O., "Amorphous semiconductors usher in digital x-ray imaging", Physics Today, 50,1997, pp. 24-30.
Kasap S.O., Aiyah V., Baillie A. and Leiga A.G., "X-ray induced hole trapping in electroradiographic plates", Journal of Applied Physics, 69, 199 1, pp. 7087-7096.
Pai D.M. and Enck RC., "Onsager mechanism of photogeneration in amorphous selenium", Physical Review B, 11, 1975, pp. 5163-5174.
Shevchik N.J. and Paul W., ''Voids in amorphous semiconductors", JournaI of Noncrystalline Solids, 16, t 972, pp. 55-71.
Madan A. and Shaw M.P., fie Physics and Applications of Amorphous Semiconductors, Academic Press Inc., San Diego, 1988, pp. 4- 10,47 1-476.
Mott N.F., "Electrons in disordered structures", Advances in Physics, 16, 1967, pp. 49-57.
Anderson P.W., "Absence of diffbsion in certain random lattices", Physical Review, 109,1958, pp. 1492-1505.
Cohen M.H., Fritzsche H. and Ovshinski S.R, "Simple band model for amorphous semiconductor alloys", PhysicaI Review Letters, 22, 1969, pp. 1065- 1072.
Marshall J.M. and Owen GE., "Drift mobility studies in vitreous arsenic triselenide", Philosophical Magazine, 24, 1972, pp. 128 1 - 1290.
Lucovsky G., 'Thysics of selenium and telIwiumY', edited by E. Gerlach and P. Grosse, Springer-Verlag, New York, 1979.
1251 Lucovsky G. and Galeener F.L., 'Intermediate range order in amorphous solids", Journal of Non-Crystalline Solids, 35-36,1980, pp. 1209- 12 14.
Kasap S.O., "Photoreceptors: the selenium alloys", Handbook of Imaging Materials, edited by A.S. Diamond, Marcel Dekker Inc., New York, 1991, pp. 329-372.
Meek P.E., "StructuraI interpretations of the vibrational spectra of models of amorphous As and Sew, Proceedings ofthe Symposium on the Structure of Non- Crystalline Solids, edited by P.H. Gaskell, Taylor & Francis Ltd., London, 1977, pp. 235-237.
Robertson J., "Electronic structure of amorphous semiconductors", Advances in Physics, 32, 1983, pp. 361-452.
Kastner M., Adler D. and Fritzsche H., bbValence-altemation model for localized gap states in lone-pair semiconductorsn, Physical Review Letters, 37, 1976, pp. 1504-1 507
Adler D., b'Amorphous-semiconductors~', Scientific American, 236, 1977, pp. 36- 48.
Kastner M., "Amorphous and liquid semiconductors", Seventh International Conference on Amorphous and Liquid Semiconductors, edited by W.E. Spear, University of Edinburgh, Edinburgh, 1977, pp. 504.
Abkowitz M., "Changes in the photoelectric properties of glassy chalcogenides induced by chemical doping, irmdiation, and thermal history", Journal of Chemical Physics, 46,1967, pp. 4535.
Agarwal S.C., "Nature of localized states in amorphous semiconductors-a study by electron spin resonance", Physical Review B, 7,1973, pp. 685-691.
Mott N.F. and Davis E.A., Electronic Processes in Non-Crystalline Solids, Clarendon Press, Oxford, 1979.
Elliot S.R, Physics ofAmolpholls Materials, Longman, New York, 1984.
Elliot S.R, "A unified model for reversibIe photostructural effects in chalwgenide glasses", Journal of Non-Crystallice Solids, 81, 1986, pp. 71-98.
Carles D., LeErancois G. and Lannagnac J.P., "A model for steady-state photoconductivity in amorphous seleniumy', Journal of Physics Letters, 45, 1984, p ~ . L901-L906.
Adler D. and Yoffa EJ., ''Localized electronic states in amorphous semiconductors", Canadian hurnal of Chemistry, 55, 1977, pp. 1920- 1929.
Abkowitz M., "Density of states in a-Se f h m combined analysis of xerographic potentials and transient transport data", Philosophical Magazine Letters, 58, 1988, pp. 53-57.
Abkowitz M., "Evidence of the defect origin of states which control photoelectric behavior of amorphous chaIcogenides", Journal of Non-Crystalline Soiids, 66, 1984, pp. 3 15-320.
Abkowitz M., "ReIaxation induced changes in electrical behavior of glassy chalcogenide semiconductors", Polymer Engineering Science, 24, 1984, pp. 1 149- 1 154.
Kasap S.O. and Juhasz C., 'Time-of-flight drift mobility measurements on chlorinedoped amorphous selenium films", Journal of Physics D: Applied Physics, 18,1985, pp. 703-720.
Abkowitz M. and Matkovics JM., Tvidence of equilibrium native defect populations in amorphous chalcogenides fiom analysis of xerographic spectra", Philosophical Magazine B, 49,1984, pp. L3 1-L36.
Orlowski T.E. and Abkowitz M., 'Microstripline transient photocurrents in a-Se - structure resolved in shallow band tail states", Solid State Communications, 59, 1986, pp. 665-668.
Wong C.K., Lucovsky G. and Bemholc I., "Intermediate range order in amorphous selenium - a novel approach based on IR absorption", Journal of Non- Crystalline Solids, 97-98, 1985, pp. 1 171 -1 174.
Hartke J.L. and Regensburger P. J., '%Electronic states in vitreous seleniumy', Physical Review, 139,1965, pp. A970-A980.
Davis E.A., "OpticaI absorption, transport and photoconductivity in amorphous seleniumy', Journal of Non-Crystalline Solids, 4, 1970, pp. 107-1 16.
Fritzsche H., "Light-induced metastable structural changes in amorphous semiconductors", httDJ/www.er.doe. eov/nroductio~chmI~hotocheml~tzsch. html, 1998.
DeNeufville IS., Moss S.C- and Ovshinsky S.R, bbPhotostructural transformations in amorphous AszSe and As2S3 films", Journal of Non- Crystalline Solids, 13,1974, pp. 191-223.
Hansen S.G. and Robitaille T.E., "Structural phototransformations in thin seienium films", Thin Solid Films, 151,1987, pp. 1 1 1-120.
Abkowitz M. and Enck RC., "Photoenhanced metastable deep trapping in amorphous chalcogenides near room temperature", PhysicaI Review B, 27, 1983, pp. 7402-741 1.
Spear W.E., "Drift mobility techniques for the study of electrical transport properties in insulating solids", Journal of Non-Crystalline Solids, 1, 1969, pp. 197-2 14.
Papadakis A.C., 'Theory of space-charge perturbed currents in insulators", Journal of Physics and Chemistry of Solids, 28, 1967, pp. 641-647.
Ehrenberg W. and Gibbons D.J., Electron Bombardment Induced Conductivity, Academic Press Inc., London, 198 1, pp. 141 -1 94.
Hecht K., "Zum mechanismus des lichtekktrischen primirstromes in isdierenden kristallen", Zeitschrift Fur Physik, 77,1932, pp. 235-245.
Gibbons D.J. and Spear W.E., "Electron hopping transport and trapping phenomena in orthorhombic sulfur crystals", Journal of Physics and Chemishy of Solids, 27, 1966, pp. 19 1 7- 1925.
Spear W.E., Steemers HL. and Mannsperger H., "Carrier lifetimes in amorphous silicon junctions h m delayed and interrupted field experiments", Philosophical Magazine B, 48,1983, pp. L49-L54.
Kasap S.O., Thakur R.P.S. and Dodds D., "Method and apparatus for interrupted transit time transient photoconductivity measurements", Journal of Physics E: Scientific Instnrmmts, 21, 1988, pp. 1 1954202.
Nheth-Buhin A. and Juhasz C., "Study of canier transport in molecularly doped polymers by interrupted time-of-flight measurements", Journal of Physics: Condensed Matter, 9, 1997, pp. 483 1-4839.
Polischuk B., Kasap S.O., Aiyah V. and Baillie A., "The interrupted field time-of- flight transient photoconductivity technique for studying charge transport and trapping in high resistivity semiconductors", International Journal of Electronics, 76,1994, p ~ . 1029-1041.
Polischuk B. and Kasap S.O., "A high-voltage interrupted-field time-of-flight transient photoconductivity apparatus", Measurement Science and Technology, 2, 1991, pp. 75-80.
[62] Kasap S.O., Polischdc B., and Dodds D., "An intempted field time-of-flight W O F ) technique in transient photoconductivity measurements", Review of Scientific Instruments, 61, 1990, pp. 2080-2087.
Zanio ICR., Akutagawa W.M., and Kikuchi R., "Transient currents in semi- insulating CdTe characteristic of deep traps", Journal of Applied Physics, 39, 1968, pp. 28 1 8-2828.
Akutagawa W. and Zanio K, "The possibility of using CdTe as a gamma spectrometer", IEEE Transactions on Nuclear Science, 15, 1968, pp. 266-274.
Rudenko A.I. and Atkhipov V.I., "Drift and diffusion in materials with traps. I. Quasi-equiIiium transport regime", PhiIosophical Magazine B, 45, 1982, pp. 177-187.
Martini M., Mayer J.W. and Zanio K.R., "DriDrift veIocity and trapping in s e m i c o n d u c t o ~ e n t charge technique", Applied Solid State Science: Advances in Materials and Device Research, edited by R. Wolfe, Academic Press, 1972, pp. 181-261.
Shukri Z., PoIischuk B., Coia C. and Rougeot H., "Characteristics of a selenium flat panel x-ray detector", Proceedings of the Sixth International Symposium on the Uses of Selenium aad TeIlurium, edited by Y. Palmieri, Selenium-Tellurium Devebpment Association, Grimbergen (Belgium), 1998, pp. 249-255.
Polischuk B., Shukri Z., Legtos A. and Rougeot H., "Selenium direct converter structure for static and dynamic x-ray detection in medical imaging applications", SPIE ROC&~IQS, 3336,1998, p ~ . 494-503.
Kasap S.O., "Photoreceptors: the selenium alloys", Handbook of Imging Materials, edited by A.S. Diamond, Marcel Dekker Inc., New York, 1991, pp. 329-372.
Que W. and Rowlands J.A., 'X-ray photogeneration in amorphous selenium: geminate versus columnar recombination", Physical Review 8, 51, 1995, pp. 1 OSOO-IO507.
Fiedler H. and Laugwitz F., "Zur qmtenausbeute elektroradiopphischer selenschichten", J. Signal AM, 9, 1981, pp. 229-235.
Donovan J.L, 'X-ray sensitivity of selenium*', Journal of Applied Physics, 50, 1979, pp. 65006504.
Haugen C., Kasap S.O. and Rowlands J., "Charge transport and electron-hole-pair creation energy in stabilized a-Se x-ray photoconductors", JournaI of Physics D: Applied Physics, 32,1999, pp. 200-207.
[74] Kasap S.O., Haugen C., Nesdo ty M. and Rowlands J., "Properties of a-Se for use in flat panel x-ray image detectors", Journal of Non-Crystalline SoIids [in press].
Rowlands J.A., DeCrescenzo G. and Araj N., "X-ray imaging using amorphous selenium: Determination of x-ray sensitivity by pulse height spectroscopy", Medical Physics, 19,1992, pp. 1065-1069.
Deamley G. and Nodrop D.C., Semiconductor Countersfor Nuclear Radiations, 1974, E.F. Spon Ltd., London, pp. 19.
Hull E.L., Pehl RH. and Varnell L.S., "The effects of 199 MeV proton radiation damage on CdZnTe photon detectors", IEEE Transactions on Nuclear Science, 44,1997, pp. 870-873.
Heijne L., Schagen P. and Buining H., "An experimental photoconductive camera tube for television*', Philips Technical Review, 16, 1954, pp. 23.
Clasen R., "Low-dose electroradiographic multilayer system with PbO binder layers", Journal of Photographic Science, 28, 1980, pp. 226-230.
Singh M. and Mumcuoglu E., "Design of a CZT based breastSPECT system", IEEE Transactions on Nuclear Science, 45, 1 998, pp. 1 1 58- 1 165.
Hubbel J.R. and Seltzer S.M., *%m~/~hvsics.nist.gov/PhvsRefData! XravMassCoef/wvererhtml", 1997.
Klein C.A., 'Bandgap dependence and related features of radiation ianization energies in semiconductors", Joumal of Applied Physics, 39, 1968, pp. 2029- 2038.
Alig RC. and Bloom S., "Electron-hole pair creation energies in semiconductors", Physical Review Letters, 35, 1975, pp. 1522- 1525.
Zhang Q. and Champness C.H., "Charge decay in amorphous selenium layers with an intermediate aluminum oxide layer", Canadian Journal of Physics, 69, 1990, pp. 278-283.
Poiischuk B., Kasap S.O., Dodds D., and Yannacopoulas S., "Charge trapping studies in amorphous selenium films by the interrupted transit time time-of-fight technique", Proceedings jma the Fourth International Symposium on uses of Selenium and Tellurium, edited by S . Carapello, Selenium-Telluium Development Association, New York, 1989, pp. 202-218.
Tanha R., M. Sc. Thesis, Electron and Hole Transport in Stabilized a-Se for X-ray Imaging, University of Saskatchewan, Saskatoon, Canada, 1998.
Brown F.C., Temperature dependence of electron mobility in AgCI", Physical Review, 97,1955, pp. 355-362.
Spear W.E., "Transit time measurements of charge carriers in amorphous selenium films", Proceedings from the Physical Society of London, B70, 1957, pp. 669-675.
Kepler RG., "Charge carrier production and mobility in anthracene crystals", Physical Review, 119,1960, pp. 1226-1229.
Tabak W.D. and Warter P.J., "Field-controlled photogeneration and free carrier transport in amorphous selenium film", Physical Review, 173, 1968, pp. 899- 907.
Oda O., Onufllka A. and Tsuboya I., "Effect of oxygen on electrophotographic properties of selenium", Journal of Non-Crystalline Solids, 83, 1986, pp. 49-62.
Abkowitz MA., Rice M.J. and Stolka M., "Electronic transport in silicon backbone polymers", Philosophical Magazine B, 61, 1990, pp. 25-57.
Gill W.D., "Drift mobilities in amorphous charge-transfer complexes of trinitrofluorenone and poly-n-vinylcarbazole", Journal of Applied Physics, 43, 1972, pp. 5033-5040.
Kasap S.O. and Juhasz C., "Transient photoconductivity probing of negative bulk film space charge evolution in halogenated amorphous selenium films", Solid State Communications, 63,1987, pp. 553,556.
Mort J., Troup A., Morgan M., Grammatica S., Knights J.C. and Lujan R., "Geminate recombination in a -S i r , Applied Physics Letters, 38, 1981, pp. 277- 279.
Dolezalek F.K. and Spear W.E., "Carrier recombination in orthorhombic sulphur", Journal of Physics and Chemistry of Solids, 36,1975, pp. 8 19-825.
Haugen C. and Kasap S.O., "Langevin recombination of drifting electrons and holes in stabilized a-Se (C1-doped a-Se:0.3% As)", Philosophical Magazine B, 71, 1995, pp. 91-96.
Svelto O., Principles ofbers, Plenum Press, New York, 1989, pp. 320-327.
Gerry E.T., 'Tukd-molecular-nitrogen Iaser theory", Applied Physics Letters, 7, 1965, pp. 6-8.
[I001 Polischuk B., Ph. D. Thesis, Interrupted Field Erne-of-Flight Transient Photoconductivity Technique and irs Applications to Amorphous Semiconductors, University of Saskatchewan, Saskatwn, Canada, 1993.
[loll Helfrich W. and Mark P., "Space charge limited currents in anthracene as a means of determining the hole drift mobility", Zeitschrifi Fur Physik, 166, 1962, pp. 370-385.
[I 021 Haugen C.J., Ph. D. Thesis, Charge Transpon in Stabilized a-Se Film Used in X- ray Image Detector Applications, University of Saskatchewan, Saskatoon, Canada, 1999.
[103] Hollins M., Medical Physics, University of Bath Science, Thomas Nelson and Sons Ltd., UK, 1992.
[I041 Boone, JM., Yu T. and Saiert A,, "Mammography spectrum measurement using an x-ray diffhction device", Physics of Medical Biology, 43, 1998, pp. 2569- 2582.
[I051 Kato H., Tsuzaka M., Koyarna S., Maekoshi H., Suzuki S. and Fujii S., "Energy- dependent responses of cadmium-telluride (CdTe) detectors to x-ray photon beams", Japanese Journal of Radiological Technology, 15, 1996, pp. 16 1 9- 1 626.
[106] P m n a l Communication with L d J., Principal Member of the Technical Staff, Sandia National Laboratories, Livermore, CA 94551, email: jlund@sandiagov, March 22, 1999.
[I071 t' Zand J., '%m:/~e8~~~.nsfc.~~~:ov/docs/~m~0sra~ne~~ASIS/cai/basis detector-html", 1997.
[I081 Johns H.E. and Cunningham J.R., The Physics of Radiology, Charles Thomas Publisher, Springfield USA, 1983, pp. 139-140.
[I091 Forsyth D.S., M.Sc. Thesis, Ultrasonic Mmrements for the Evaluation of Thermal Fatigue Damage, University of Saskatchewan, Saskatoon, Canada, 1992.
[I 101 Tab& M.D. and Scharfe M.E., 'Transition h m emission-limited to space- charge-Limited photoconductivity", Journal of Applied Physics, 41, 1970, pp. 21 14-21 18.
[11 I] Kasap SO., Haugen C., Tanha R and PoIischuk B., 'Tieid dependence of the dispersion of carriers in the transient photocurrent in time-of-flight measurements: stabilized a-Se (Cl doped Se:0.30/&)" [to be submitted]
Kreyszig E., Advanced Engineering Mathematics, John Wiley & Sons, Toronto, 1988, pp. 1285-1289.
Stoodley K., Basic Statistical Techniques for Engineering and Science Students, Bradford University Press, London, 1974, pp. 121-1 34.
Haugen C., Kasap S.O. and Rowlands J., "X-ray irradiation induced bulk space charge in stabiIized a-Se x-ray photoconductors", Journal of Applied Physics, 84, 1998, pp. 5495-5501.
Owen A.E., Firth AS. and Ewen P.J.S., "Photo-induced structural and physico- chemicaI changes in amorphous cMcogenide semiconductors", Philosophical Magazine B, 52, 1985, pp. 347-362.
Roy A., Kolobov A.V. and Tanaka EL, "Laser-induced suppression of photocrystallization rate in amorphous selenium films", Journal of Applied Physics, 83,1998, pp. 495 1-4956.
Poborchii V.V., Kolobov A.V. and Tanaka K., "An in situ raman study of polarization-dependent photocrystallization in amorphous selenium films", Applied Physics Letters, 72, 1998, pp. 1 167-1 169.
Kolobov A.V. and Elliott S.R., "Reverstile photo-amorphization of a crystaed Ass&e~ alloy", Philosophical Magazine B, 71,1995, pp. 1-10.
Kolobov A.V. and Elliott S.R, '%eversiIble photo-amorphization of crystalIine films of As&%", Journal of Non-Crystalline Solids, 189,1995, pp. 297-300.
Tanaka K., 'Revm'bIe photostructural change: mechanisms, properties and applicationsy', Journal ofNon-Crystalline Solids, 35 & 36, 1980, pp. 1023- 1034.
Tanaka K. and Odajima A,, Thotodarkening in amorphous selenium", Solid State CommImications, 43,1982, pp. % 1-964.
Tanaka K., "Configurational and stmctd models for photodarkening in glassy chalcagenides", Japanese loumal of Applied Physics, 25,1986, pp. 779-786.
Sapoval B- and Hetmann C., Physics of Semicon&ctors, Springer-Verlag, New YO&, 1995, pp. 155-157.
Street RA., 'Won-radiative recombiaation in chaIcogenide glasses", Solid State Communications, 24,1977, pp. 363-365.
Biegelsen D.K. and Street R.A., 'Thotoinduced defects in chalcogenide glassesy', Physical Review Letters, 44,1980, pp. 803-806.
[I271 Vanderbilt D. and Joannopoulos J., "Calculation of defect states in amorphous selenium*', PhysicaI Review Letters, 42, 1979, pp. I0 12-1 0 15.
[I281 Oyanagi H., Kolobov A. and Tanaka K., "Pump and probe x-ray absorption fine structure using high-brilIiance photon sources", JoumaI of Synchrotron Radiation, 5, 1998, pp. 1001-1003.
[129] Kolobov A.V., Oyanagi H., Tanaka K. and Tanaka Ke, bThotostructural changes in amorphous selenium: an in situ EXAFS study at low temperature", Journal of Non-Crystdhe Solids, 19&200,1996, pp. 709-7 13.
[I301 Kolobov A.V., Oyanagi H., Roy A. and Tanaka R, "A nanometer scale mechanism for the reversible photostructural change in amorphous chalcogenides", Journal of Non-Crystalline Solids, 232-234, 1998, pp. 80-85.
[13 11 Kolobov A.V., Oyanagi H., Tanaka K. and T d a Ke, "Structural study of amorphous selenium by in situ W S : observation of photoinduced bond alteration", Physical Review B, 55, 1997, pp.726-734.
[I321 Roy A., Kolobov A.V., Oyanagi H. and Tanaka K., "Photo-induced ring-to-chain conversion in as-evaporated 6lms of amorphous selenium", Philosophical Magazine B, 78, 1998, pp. 97-94.
[I331 Tikhomirov V K , Adriaensens GJ. and Elliott SK, "Temperature dependence of the photoinduced anisotropy in chalcogenide glasses: activation energies and their inteqmtation", Physical Review B, 55, 1997, pp. R660-R663.
11341 Hajto J. and Apai P., 'Investigation of laser induced Iight absorption oscillation", Journal of Non-Crystalline Solids, 35 & 36,1980, pp. 1085- 1 O9O.
[13S] Hajto J. and Fiistijss-Wdgner M., 'Zaser induced oscillatory phenomena in a- GeSe films", Journal de Physique Colloque, 42,198 1, pp. C4-3 13-C4-3 16.
[I361 Hajto J. and Jhossy I., "Optical bistability observed in amorphous semiconductor films", Philosophical Magazixie B, 47,1983, pp. 347-366.
[I371 Hajto I., Jiinossy J. and Firth A., ''Explanation of the laser-induced oscillatory phenomenon in amorphous semiconductor film", Philosophical Magazine B, 48, 1983, p ~ . 31 1-321.
[I381 Fritzsche H., ''The origin of reversiile and ineversiile photostructural changes in chalcogenide glasses", Philosophical Magazine B, 68,1993, pp. 561-572.
[I391 Tai ILL, Ong E. and Vadimsky RG., %organic resist systems for VLSI microlithography", Proceedings - The Electrochemical Society, 82, 1982, pp. 9- 35.
[ l a ] KoIobov A.V., Kondo M., Matsuda A. and Tanaka K., "Negative correlation energy in amorphous selenium: experimental evidence", Journal of Non- Crystalline Solids, 227-230, 1998, pp. 842-846.
[I411 Kolobov A.V., Kondo M., Oyanagi H., Dumy R., Matsuda A. and Tanaka IC, 'Txperirnental evidence for negative correlation energy and valence alternation in amorphous selenium", Physical Review B, 56,1997, pp. R485-R488.
[I421 Kolobov A.V., Kondo M., Oyanagi H., Matsuda A. and Tanaka IC, 'Wegative correlation energy and valence altmation in amorphous selenium: an in situ optically induced ESR study", PhysicaI Review By 58, 1998, pp. 12004- 12010.
11431 h a p S.O., Aiyah V. and Polischuk B., "Determination of the deep-hole capture crowsection in a-Se via xerographic and intenupted-field time-of-flight techniques", Philosophical Magazine Letters, 62,1990, pp. 377-382.
[144] Mehyk A.R., "Hole emission defect states in amorphous AstSe3", Journal of Non-Crystalline Solids, 35 & 36,1980, pp. 837-842.
[145] Baxendale M. and Juhasz C., 'Temperaturedependent xerographic depletion discharge studies of a-Se:Te alloys", SPIE, 1253,1990, pp. 212-222.
[I461 Kasap S.O., Polischuk B., Aiyah V. and Yannacopoulos S., '?)rift mobility relaxation in a-Se", Journal of Applied Physics, 67, 1990, pp. 1 9 1 8- 1922.
[I471 Chaudhrrri S., Biswas SK, Choudhury A. and Goswami K., "Variation of optical gap of thick amorphous selenium film on heat treatment", Journal of Non- Crystalline Solids, 54,1983, pp. 179-1 82.
[I481 Onsager L., "Initial recombination of ions", Physical Review, 54, 1938, pp. 554- 557.
[I491 Jaffe G., "Zur theorie ionisation in kolonnen", Annalen der Physik, 42, 19 13, pp. 303-344.
[I 501 Sahyun M.R.V., "Monte carlo modeling of electrophotographic x-ray detecto~s'~, Journal of Applied Physics, 53,1982, pp. 6253-626 1.