Gamma ray irradiation induced optical band gap variations in chalcogenide glasses

Nuclear Instruments and Methods in Physics Research B 234 (2005) 525–532

www.elsevier.com/locate/nimb

Gamma ray irradiation induced optical band gapvariations in chalcogenide glasses

Fang Xia a,b, S. Baccaro b, Donghui Zhao a, M. Falconieri c, Guorong Chen a,*

a Key Laboratory for Ultrafine Materials of Ministry of Education, School of Materials Science and Engineering,

130 Meilong Road, Box 306, East China University of Science and Technology, Shanghai 200237, Chinab ENEA-FIS/ION, Via Anguillarese 301, 00060 S. Maria di Galeria (Roma), Italy

c ENEA-MAT/NANO, Via Anguillarese 301, 00060 S. Maria di Galeria (Roma), Italy

Received 15 December 2004; received in revised form 23 February 2005

Available online 12 April 2005

Abstract

In the present work c-irradiation induced optical band gap variations (DEopt) were investigated on Ge–As–Se andGe–As–Se–Te chalcogenide glasses. Higher doses of c-irradiation resulted in decreased Eopt, which was composition

dependent. Especially, glasses with stoichiometric compositions showed different DEopt from nonstoichiometric glassesunder the same irradiation conditions. There seemed existence of a threshold Eopt (TE) under the certain dose of irra-

diation below which DEopt hardly occurred. Results were interpreted from viewpoint of glass structure, Chemical BondApproach (CBA) and localized states density theory. Raman analysis supported well these discussions.

� 2005 Elsevier B.V. All rights reserved.

PACS: 61.82.Fk; 61.80.Ed; 61.43.Fs

Keywords: Gamma ray irradiation; Optical band gap; Chalcogenide glasses; Raman spectra

1. Introduction

Chalcogenide glasses exhibit high sensitivity to

irradiations due to their flexible structure [1]. In

the last decades, many potential applications

0168-583X/$ - see front matter � 2005 Elsevier B.V. All rights reserv

doi:10.1016/j.nimb.2005.02.019

* Corresponding author. Tel.: +86 216 425 2647; fax: +86 216

425 3395.

E-mail address: [email protected] (G. Chen).

based on these effects have been explored, for

examples, in the fields of submicron photoresists,optical memories, diffraction elements, optical

light-guide and optoelectronic element and devices

[2–5]. Recently, bulk chalcogenide glasses are con-

sidered as a good alternative candidate for high-

energy irradiation detector in dosimetric system

for wide industrial applications due to a tight

relationship of irradiation-induced effects with

ed.

526 F. Xia et al. / Nucl. Instr. and Meth. in Phys. Res. B 234 (2005) 525–532

absorbed doses [6]. Such kind of irradiation detec-

tors allows a low barrier of information bleaching

temperature (<350 �C) in comparison with widelyused colored oxide glasses (>500 �C). There havebeen quite a few investigations reported in thisrespect on such glass systems as As–S [7],

As–Se–Sb [8] and Ge–As–S [9]. Our present work

extended studies in this field to the Se-contained

Ge–As–Se and Ge–As–Se–Te glass systems which

possess advantages of higher thermal stability, eas-

ier production, and most important, the possible

higher sensitivity to irradiation due to their more

flexible structure. A group of stoichiometric andnonstoichiometric chalcogenide glasses in these

two systems were prepared, and c-irradiationtreatments were performed on these Samples for

the first time. The band gap Eopt was selected as

a parameter for detector applications, while its

dependence on irradiation doses was studied.

Raman spectrum analysis was made on some sam-

ples in order to get further information about glassstructure transformation under irradiation.

2. Experimental

Glass compositions used for the present work

are given in Table 1, which are divided into stoichi-

ometric and nonstoichiometric groups. The aim ofthis division is to investigate the different effects

between them under irradiation due to their differ-

ent structure and bond configuration as has been

confirmed by some authors [10]. Te was intro-

Table 1

Glass compositions and Eopt as well as its variations (DEopt) under g

No. Compositions (at%) 0 G

Eop

Stoichiometric 1 Ge30As4Se66/(GeSe2)0.94(As2Se3)0.06 1.80

2 Ge25As10Se65/(GeSe2)0.83(As2Se3)0.17 1.74

3 Ge20As16Se64/(GeSe2)0.71((As2Se3)0.29 1.70

Nonstoichiometric 4 Ge35As5Se60 1.76

5 Ge30As10Se60 1.78

6 Ge27As13Se60 1.71

7 Ge20As10Se70 1.69

8 Ge20As20Se40Te20 1.02

9 Ge20As20Se14Te46 0.78

duced for the purpose of investigating the effect

due to its substitution for Se.

A commonly used direct melting synthesis

method was applied for preparing bulk glass sam-

ples [11]. High purity elemental constituents Se, Ge(99.99%), As (99.999%) and Te (99.99%) were used

as the raw materials. They were weighed in appro-

priate atom percentage (at%) and put into preheat

treated quartz glass ampoules (12 mm diameter)

with low hydroxyl content. The charged ampoule

was sealed in a vacuum of 10�3 Pa. The glasses

were melted at 900–950 �C for more than 12 h ina specially designed rocking furnace to ensurehomogenization of the glass liquids. After melting,

melt-containing ampoules were removed from the

furnace for quenching in air at room temperature

to obtain the bulk glasses. Glasses were then

annealed at temperature between 185 and 350 �Cfor 2 h to relax the stress inside the glass matrix.

After melting and annealing, the glass samples

were cut, ground and double-face polished with2–3 mm in thickness for measurements.

The c-irradiation tests on glass samples wereperformed by Calliope 60Co source (ENEA-FIS/

ION, Italy) in air at room temperature with the ab-

sorbed dose rate of 800 Gy/h and a total dose of

50–115 kGy. The average energy of gamma ray is

1.25 MeV in this case.

For determining Eopt of glasses, VIS–NIRtransmission spectra were measured on glass sam-

ples before and after irradiation from 550 nm to

2000 nm using the high-resolution (0.1 nm) spec-

trometer (Perkin Elmer UV/VIS/NIR spectro-

amma ray irradiation

y 50 kGy 115 kGy

t (eV) Eopt (eV) DEopt (eV) Eopt (eV) DEopt (eV)

8 1.795 �0.013 1.795 �0.0138 1.742 �0.006 1.735 �0.0135 1.705 0 1.698 �0.007

3 1.745 �0.018 – –

0 1.774 �0.006 1.769 �0.0116 1.709 �0.007 1.700 �0.0164 1.694 0 1.688 �0.0068 1.028 0 1.023 �0.0053 0.783 0 0.783 0

5

61 Ge30As4Se662 Ge25As10Se653 Ge20As16Se644 Ge35As5Se605 Ge30As10Se606 Ge27As13Se607 Ge20As10Se70 7

F. Xia et al. / Nucl. Instr. and Meth. in Phys. Res. B 234 (2005) 525–532 527

meter, Lambda 900). Raman spectra were col-

lected using 532 nm excitation wavelength and a

550 cm focal length monochromator combined

with a notch filter. A LN2-cooled CCD camera

was used as detector.

0

1

2

3

4

0.6 0.8 1 1.2 1.4 1.6 1.8 2

8 Ge20As20Se40Te209 Ge20As20Se14Te46

(αhv

)1/2

(cm

-1/2

eV1/

2 )

Photon Energy hv (eV)

2

3

4

9

8

6

5

1

Fig. 1. A Plot of (ahm)1/2 as a function of hm for as prepared Ge–As–Se and Ge–As–Se–Te glasses.

-0.002

0

0.002

0.004

0.006

0.008

0.01

0.012

0.014

-20 0 20 40 60 80 100 120

x=0.94x=0.83x=0.71

∆Eop

t (-e

V)

Dose (kGy)

Fig. 2. A plot of DEopt as a function of absorbed doses forstoichiometric (GeSe2)x(As2Se3)1�x glasses with x = 0.94, 0.83

and 0.71.

3. Results

The optical band gap (Eopt) in Urbach region of

the glasses can be obtained according to its depen-

dence on absorption coefficients a and the energyhm of the incident photon [12], as expressed bythe following equation [13–15]

ahm ¼ Aðhm � EoptÞn

where A is a constant and n is a parameter associ-ated with both the type of transition and the pro-

file of the electron density in the valence and

conduction bands. In the present case an allowed

indirect transition process is involved for which

n = 2 would be the best fit [16]. The absorption

coefficients a values for the investigated glasseswere calculated from VIS–NIR transmission data

using the following formula:

a ¼ 1dlog10ð1=T Þ

where d is the thickness and T the transmission of

samples. Eopt values were then determined by plot-ting (ahm)1/2 as a function of photon energy hm andextrapolating the linear portion of the curve to

intersect the hm axis. Fig. 1 shows all plots for asprepared glasses and Eopt values obtained there-

from are listed in Table 1. The latter also includes

the Eopt values for c-irradiated glass samples ob-tained in the same way. It can be seen from Table

1 and Fig. 1 that Eopt was generally compositiondependent whereas c-irradiation induced Eopt vari-ations (DEopt) took on decreasing tendency as afunction of absorbed doses. In both cases different

regularities between glasses with stoichiometric

and nonstoichiometric compositions were

observed.

For glasses with the stoichiometric composi-

tion: (GeSe2)x(As2Se3)1�x (Samples 1–3), Eoptwas found to increase with the increased x. How-

ever, DEopt showed the different characters, as

illustrated in Fig. 2 where DEopt was plotted as afunction of absorbed doses. Firstly, the initial

DEopt due to 50 kGy c-irradiation increased pro-nouncedly with the increased x. Secondly, under

the further higher dose (115 kGy) of irradiation,a saturated DEopt was observed for Sample 1 withthe highest x, while for Samples 2 and 3 DEopt kept

528 F. Xia et al. / Nucl. Instr. and Meth. in Phys. Res. B 234 (2005) 525–532

increasing at nearly the same rate. Both phenom-

ena imply that the sensitivity of glass to c-irradia-tion is closely related to x, that is, to the content of

GeSe2.

As for glasses with nonstoichiometric composi-tions (Samples 4–9), their Eopt values did not show

straightforward variations with the change of com-

ponents. Taking GeyAs40�ySe60 (Samples 4–6) for

instances, with the decrease of y (Ge content), Eoptincreased first and then decreased, as shown in

Table 1. While Sample 7 containing the lowest

content of Ge displayed the lowest Eopt among

all Ge–As–Se glasses, as shown in Fig. 3, Samples8 and 9 with the fourth element Te exhibited fur-

ther reduction of Eopt. At the same time, DEoptoccurred to all samples except Sample 9 with

the higher Te which showed little sensitivity to

irradiation even at the dose of 115 kGy, as shown

in Fig. 3. Moreover, Sample 4 exhibited the high-

est DEopt among GeyAs40�ySe60 glasses (Samples

4–6).The above results are in good agreement with

Butkiewicz�s research work that focused on Ge–

-0.005

0

0.005

0.01

0.015

0.02

0.025

0.03

-20 0 20 40 60 80 100 120

4 Ge35As5Se60 5 Ge30As10Se60 6 Ge27As13Se60 7 Ge20As10Se70 8 Ge20As20Se40Te20 9 Ge20As20Se14Te46

∆Eop

t (-e

V)

Dose (Gy)

Fig. 3. A plot of DEopt as a function of absorbed doses fornonstoichiometric Ge–As–Se and Ge–As–Se–Te glasses.

Table 2

Strength of bonds possibly existing in Ge–As–Se and Ge–As–Se–Te g

Bonds Ge–Se Ge–Ge Ge–Te As–As

Bond Strength (kcal/mol) 49.42 37.6 35.47 32.1

As–S system [9]: increased GeS2 content in stoichi-

ometric composition (As2S3)y (GeS2)1�y gave rise

to increased Da/a0, where a is absorption coeffi-cient, while in nonstoichiometric compositions

(As2S3)x (Ge2S3)1�x, decreased Ge content resultedin a fluctuant change of Da/a0.To supply supporting evidences for DEopt,

Raman spectra were measured on Samples 4 and

7 before and after irradiation. As shown in Fig.

5, after 15 kGy c-irradiation, the Se-deficient glass(Sample 4) displayed the decreased intensities of

Raman bands due to Ge–Ge bond (183 cm�1)

[17] and As–As bonds (167 cm�1, 222 cm�1,235 cm�1, 374 cm�1) vibrations [18,19] while that

of Ge–Se bonds (191 cm�1) [20] remained un-

changed. For the Se-rich glass (Sample 7), besides

observations of the main bands corresponding to

As–Se bonds (230 cm�1) [21] and Ge–Se bonds

(195 cm�1) [20] vibrations as presented in Fig. 6,

there existed a series of bands due to Se–Se wrong

bonds vibrations before c-irradiation and thesebonds took the forms of C–Se–Se–C (207 cm�1)

[20](C@Ge and/or As), Se chains (238 cm�1) [22]

and Se rings (252 cm�1, 270 cm�1, 282 cm�1)

[21]. After 50 kGy irradiation, however, Se–Se as

well as As–Se bonds exhibited the weakened vibra-

tions while Ge–Se bonds kept unchanged.

4. Discussion

It is known that the stoichiometric (GeSe2)x-

(As2Se3)1�x glasses are mainly composed of

Ge(Se1/2)4 tetrahedral and As(Se1/2)3 triangular

units. The increased x corresponded to the enhan-

ced Ge(Se1/2)4 units which were constructed with

the stronger Ge–Se bonds than As–Se bonds(Table 2), the latter formed As(Se1/2)3 units. Fur-

thermore, Ge(Se1/2)4 tetrahedra were three-

dimensionally arranged and more compact than

layered As(Se1/2)3 triangles. Therefore, barrier of

electron transitions between valence and conduc-

lass systems [13,24]

As–Se Se–Se Te–Te As–Te Ge–As Se–Te

41.69 44 33.0 32.74 35.61 44.18

F. Xia et al. / Nucl. Instr. and Meth. in Phys. Res. B 234 (2005) 525–532 529

tion bands became higher with the increased x,

resulting in the broadening of Eopt.

On the other hand, Eopt value of semiconduct-

ing glasses is strongly influenced by the defect-

induced localized state (DILS) in line of the StatesDensity Theory (SDT) [12]. Normally a high con-

centration of DILS in the band structure of glass

favors Eopt narrowing. As the As(Se1/2)3 unit was

easer to destruct under irradiation due to its weak-

er structure in comparison to the Ge(Se1/2)4 unit,

the resultant defects (for example, dangling bonds)

would have been expected more that acted as the

defect state to narrow Eopt. The question nowarises as to why the glass with the lower x (higher

content of As2Se3) showed less DEopt than theglass with the higher x in the present work. This

was also the case for Butkiewicz�s work but noexplanation was given in his paper [9]. In my opin-

ion, it is most likely associated with the original

Eopt value of glasses. According to semiconducting

physics, the smaller Eopt corresponds to the highervalence band energy, which means the DILS

energy levels may locate nearly above valence

band or in the original localized state, contributing

less to the width of localized state, thus causing

less DEopt. In the case of the larger Eopt, however,the DILS energy is easier to be higher than that of

valence band and original localized state, resulting

in the broadening of localized state, and thereforethe more notable DEopt. In Shpotyuk�s work [8], itwas found that Ge–Sb–S glass had a poor sensitiv-

ity to ionizing irradiation in comparison with

Ge–As–S glass due to the low energetic barrier of

the coordination defects states in the former case,

but not the original Eopt value. According to our

work, however, both factors are suggested affect-

Table 3

Calculated proportions of bonds in Ge–As–Se(Se–Te) nonstoichiome

No. and compositions

(at%)

Proportion of bonds in glass struct

Ge–Se Ge–Ge As–Se

4 Ge35As5Se60 87.3 7.3 –

5 Ge30As10Se60 88.9 – –

6 Ge27As13Se60 80.9 – 9.0

7 Ge20As10Se70 66.7 – 25

8 Ge20As20Se40Te20 61.5 – –

9 Ge20As20Se14Te46 21.5 – –

ing the irradiation sensitivity of Sb-contained

glass.

The observed saturated DEopt might be attrib-uted to the possible equilibrium of destruction

and reconstruction process of As–Se bonds in theglass network during irradiation. Sample 1

reached the equilibrium state at the earliest oppor-

tunity under a certain dose of irradiation because

it had the lowest content of As(Se1/2)3 triangles

(6%).

The observed different behavior for nonstoi-

chiometric glasses might be assigned to the

different structure from their stoichiometric coun-terparts. According to the principle of Chemical

Bond Approach (CBA) [23], in the nonstoichio-

metric Ge–As–Se glass system Ge–Se bonds with

the highest bond strength (49.42 kJ/mol, Table 2)

are expected to form first, followed orderly by

As–Se bonds (41.69 kJ/mol), Se–Se bonds (44 kJ/

mol), Ge–Ge bonds (37.6 kJ/mol) and As–As

bonds (32.1 kJ/mol), if any. The reason for priorityof As–Se bond formation to Se–Se bond is due to

the principle [25] that atoms combine more favor-

able with atoms of different kinds than with the

same kind in glass structure. In accordance with

this concept, deficiency of Se in this system could

possibly lead to formation of homopolar Ge–Ge

and/or As–As bonds besides Ge–Se bonds and/or

As–Se. Strength of bonds possibly existing in thepresent glass systems were listed in Table 2 while

proportions of different bond types in the nonsto-

ichiometric Ge–As–Se and Ge–As–Se–Te systems

were calculated using this approach and summa-

rized in Table 3. It is seen that among glasses in

the GeyAs40�ySe60 system, Sample 5 possessed

the highest percentage of Ge–Se bonds and Sample

tric glasses (at%)

ural network (at%)

As–As Se–Se Ge–Te As–Te

5.4 – – –

11.1 – – –

10.1 – – –

– 8.3 – –

7.7 – – 30.8

7.7 – 40 30.8

530 F. Xia et al. / Nucl. Instr. and Meth. in Phys. Res. B 234 (2005) 525–532

6 the lowest, corresponding respectively to the

largest and the smallest Eopt. This is just identical

to our above discussions on stoichiometric glasses,

that is, the higher concentration of Ge(Se1/2)4tetrahedral units contributed to the larger Eoptbecause of the stronger Ge–Se bonds and the high-

er rigid structure. Similarly, the lowest Eopt for

Sample 7 could be attributed to the lowest amount

of Ge–Se bonds, as illustrated in Fig. 4 by the lin-

early fit curve of Eopt as a function of Ge–Se bonds

content among all nonstoichiometric Ge–As–Se

glasses.

Samples 8 and 9 introduced Te as the fourth ele-ment and exhibited the much smaller Eopt compar-

ing with Ge–As–Se system glasses. It could be

interpreted in term of Lone Pair Electron (LPE)

mechanism. The valence band energy level of chal-

cogenide semiconducting glasses is known as to be

determined by LPE from chalcogens [26]. Consid-

ering that Te has bigger atomic radii (142 pm [27])

than Se (116 pm [27]), its valence electrons wouldsuffer less attraction from positive nucleus. Thus,

LPEs in valence orbits would have higher energy

and become more sensitive to external influences.

The consequence would be that the valence band

energy as well as the localized states energy

became higher, approaching Fermi energy level

(Ef), and the Eopt was reduced.

1.68

1.7

1.72

1.74

1.76

1.78

1.8

65 70 75 80 85 90

y = 1.4423 + 0.0036555x R= 0.92303

E opt

(eV)

Ge-Se Bonds Content (%)

Fig. 4. A plot of Eopt as a function of Ge–Se bond proportion

in the nonstoichiometric glasses.

Unlike the stoichiometric glasses whose sensi-

tivity to irradiation was mainly determined by

Eopt, nonstoichiometric glasses showed the strong

dependence on defects in this respect due to pres-

ence of wrong homopolar bonds. In the light ofCBA, the wrong homopolar bonds are easier to

destruct under irradiation due to their lower bond

strengths and their role as the network modifiers.

As Sample 4 contained the highest concentration

of homopolar bonds (12.7 at%), more dangling

bonds would be expected after irradiation that

broadened the defect state according to SDT, lead-

ing to the largest DEopt. By comparison betweenSamples 5 and 6, it is found that the latter showed

the larger DEopt but a little lower concentration ofhomopolar As–As bonds than the former. We as-

sumed that presence of higher proportion of As–Se

bonds (9 at%) might play an important role. As we

discussed before, As–Se bonds have the lower

bond strength than Ge–Se bonds, thus being more

sensitive to the irradiation due to creation of moredefect states, especially when the irradiation dose

was higher (115 kGy). Butkiewicz ascribed the

fluctuant change of Da/a0 to the existing homopo-lar bonds in nonstoichiometric compositions

(As2S3)x(Ge2S3)1�x [9]. This is to some extent in

accordance with the above discussion.

Moreover, Samples 7 and 8 exhibited little sen-

sitivity to the 50 kGy irradiation while Sample 9did not show any Eopt variation even under the

115 kGy irradiation. Considering the same phe-

nomenon occurring to Sample 3, we assumed that

there seemed existence of a threshold Eopt (TE) be-

low which DEopt hardly occur under the certaindose of irradiation. In the present case, TE most

likely lied in around 1.710 eV.

Above interpretations are well supported bycomparison of Raman spectra before and after

irradiation. For example, decreased intensities of

Ge–Ge and As–As bonds vibrations for the Se-

deficient glass (Fig. 5) and of Se–Se bonds vibra-

tions for the Se-rich glass (Fig. 6) figure the

destruction of such bonds during irradiation,

demonstrating that these wrong homopolar bonds

exerted major effect on sensitivity to irradiationfor nonstoichiometric glasses. However, it is

opposite to the previously reported work on the

stoichiometric As2S3 glass which showed the

150 200 250 300 350 400 450 500

Inte

nsity

(arb

.un.

)

Raman shift (cm-1)

Ge20As10Se70

Unirradiated

Irradiated (50kGy)

270

(Ser

ings

)25

2 (S

erin

gs)

282

(Ser

ings

)

195

(Ge−

Ss)

207

(Se−

Se)

238

(Sec

hain

s)23

0 (A

s−Se

)

Fig. 6. Raman spectrum of glass Sample 7 (Ge20As10Se70)

before and after gamma ray irradiation (50 kGy) at room

temperature.

150 200 250 300 350 400 450 500

Inte

nsity

(arb

.un.

)

Raman shift (cm-1)

Ge35As5Se60

Unirradiated

Irradiated(15kGy)

222

(As−

As)

364

(As−

As)

167

(As−

As)

183

(Ge−

Ge)

191

(Ge−

Se)

235

(As−

As)

Fig. 5. Raman spectrum of glass Sample 4 (Ge35As5Se60)

before and after gamma ray irradiation (15 kGy) at room

temperature.

F. Xia et al. / Nucl. Instr. and Meth. in Phys. Res. B 234 (2005) 525–532 531

irradiation-induced enhancement of As–As vibra-

tion bands using far IR Fourier spectroscopy [7].

This might due to the pure 2D structure of As2S3glass which is different from that of (As2Se3)x-

(GeSe2)1�x system. On the other hand, unchanged

intensity of Ge–Se bonds vibrations after irradia-

tion contrasts sharply with the weakened As–Se

bonds vibrations (Fig. 6), which is consistent to

the fact that destruction of As(Se1/2)3 triangular

units accounted for DEopt with respect to bothstoichiometric and nonstoichiometric glasses.

5. Conclusion

The optical band gap (Eopt) of glasses in Ge–

As–Se system was proportional to the amount of

Ge(Se1/2)4 tetrahedron units in the stoichiometric

case or to the Ge–Se bonds concentration in thenonstoichiometric case. However, they showed dif-

ferent c-irradiation induced Eopt variation (DEopt)characters. For the former, DEopt was mainlydetermined by Eopt while for the latter, DEoptshowed the strong dependence on defects due to

presence of wrong homopolar bonds. Moreover,

there seemed existence of a threshold Eopt (TE)

below which DEopt hardly occurred under the cer-tain dose of irradiation. Comparison of Raman

spectra supported above statement by showing

the decreased intensities of Raman bands due to

wrong homopolar bonds (Ge–Ge, As–As, Se–Se)

and As–Se bonds vibrations after irradiation with

respect to the unchanged Ge–Se bonds vibrations.

Acknowledgments

The authors would like to express the thanks to

A. Piegari and A. Krasilnikova from ENEA-FIS/

OTT for their help in using IR spectrophotome-

ter. This work is partially supported by the

fellowship granted in ENEA (Italy) for foreign

researchers.

References

[1] O.I. Shpotyuk, Nucl. Instr. and Meth. B 166–167 (2000)

525.

[2] H. Fritzsche, Phys. Rev. B 52 (1995) 15854.

[3] M. Mitkova, T. Petkova, R. Markovski, V. Mateev, J.

Non-Cryst. Solids 164–166 (1993) 1203.

[4] H. Hisakuni, K. Tanaka, Science 270 (1995) 974.

[5] T. Ide, M. Suzuki, M. Okuda, Jpn. J. Appl. Phys. 34

(1995) L529.

[6] O.I. Shpotyuk, Radiat. Phys. Chem. 46 (1995) 1279.

532 F. Xia et al. / Nucl. Instr. and Meth. in Phys. Res. B 234 (2005) 525–532

[7] V.O. Balitska, O.I. Shpotyuk, J. Non-Cryst. Solids 227–

230 (1998) 723.

[8] O.I. Shpotyuk, R.Ya. Golovchak, T.S. Kavetsky, Nucl.

Instr. and Meth. B 166–167 (2000) 517.

[9] B. Butkiewicz, R. Golovchak, A. Kovalskiy, Radiat. Eff.

Defects Solids 153 (2001) 211.

[10] V. Balitska, J. Filipecki, O. Shpotyuk, J. Non-Cryst. Solids

287 (2001) 216.

[11] G. Chen, J. Cheng, W. Chen, J. Non-Cryst. Solids 220

(1997) 249.

[12] N.F. Mott, E.A. Davis, Electronic Processes in Non-

crystalline Materials, Clarendon, Oxford, 1971.

[13] L. Pauling, The Nature of the Chemical Bond, third ed.,

Cornell University Press, Ithaca, NY, 1960.

[14] E.A. Davis, N.F. Mott, Philos. Mag. 22 (1970) 903.

[15] M.M.A. Imran, N.S. Saxena, Y.K. Vijay, J. Non-Cryst.

Solids 298 (2002) 53.

[16] S.M. EI-Sayed, Nucl. Instr. and Meth. B 225 (2004) 535.

[17] Y. Wang, K. Tanaka, T. Nakaoka, J. Non-Cryst. Solids

299–302 (2002) 963.

[18] M. Frumar, Z. Polak, Z. Cernosek, J. Non-Cryst. Solids

256–257 (1999) 105.

[19] R. Ston, M. Vlcek, H. Jain, J. Non-Cryst. Solids 326–327

(2003) 220.

[20] Takeshi Usuki, Osamu Uemura, Kazuhiro Fujimura,

Yasuo Kameda, J. Non-Cryst. Solids 192–193 (1995)

69.

[21] V. Kovanda, M. Vlcek, H. Jain, J. Non-Cryst. Solids 326–

327 (2003) 88.

[22] P. Boolchand, X. Feng, W.J. Bresser, J. Non-Cryst. Solids

293–295 (2001) 348.

[23] Bicerano Jozef, R. Oushinsky Stanford, J. Non-Cryst.

Solids 74 (1985) 75.

[24] A.L. Allred, J. Inorg. Nucl. Chem. 17 (1961) 215.

[25] W.H. Zachariasen, J. Am. Chem. Soc. 54 (1932) 3841.

[26] Liu Enke, Zhu Bingsheng, Luo Jinsheng, Semiconducting

Physics, National defense industrial press, Beijing, 1994 (in

Chinese).

[27] J.A. Dean, Lange�s Handbook of Chemistry, 15th ed.,Vol. 4, McGraw-Hill Inc., 1999, p. 30.