JOURNAL OF APPLIED PHYSICS 99, 094301 共2006兲

Photoinduced interdiffusion in nanolayered Se/ As2S3 films: Optical and x-ray photoelectron spectroscopic studies K. V. Adarsh and K. S. Sangunnia兲 Department of Physics, Indian Institute of Science, Bangalore, 560012, India

T. Shripathi UGC-DAE Consortium for Scientific Research, University Campus, Khandwa Road, DIndore 452 017, India

S. Kokenyesi Department of Experimental Physics, University of Debrecen, Bem ter 18/a, Debrecen 4026, Hungary

M. Shipljak Uzhgorod National University, Pidhirna 46, Uzhgorod 88000, Ukraine

共Received 13 November 2005; accepted 6 March 2006; published online 11 May 2006兲 Photoinduced interdiffusion was observed with above band gap light in nanolayered Se/ As2S3 films. It is discussed in terms of the optical parameters such as band gap, Urbach edge 共Ee兲 关F. Urbach, Phys. Rev. 92, 1324 共1953兲兴, and B1/2 共Tauc’s parameter兲 关J. Tauc et al., Phys. Status Solidi 15, 627 共1966兲兴. Experimental data of B1/2 and Ee for as-prepared samples do not show clear correlation implied by the Mott-Davis model 关N. F. Mott and E. A. Davis, Electronic Process in Non-crystalline Materials 共Clarendon, Oxford 1979兲, p. 287兴. It is also shown that the optical parameters can be changed with a change in the Se sublayer thickness. Variations of these optical parameters as a function of modulation period and photoinduced interdiffusion were discussed in terms of the quantum confinement effect and changes in the valence and conduction bands. We proposed a model to explain the mechanism of Se diffusion in As2S3, which suggests that diffusion takes place through the wrong bonds. X-ray photoelectron spectroscopy 共XPS兲 is used to investigate the chemical alternations in the bonding. The proposed model was supported by the XPS data. © 2006 American Institute of Physics. 关DOI: 10.1063/1.2193061兴 INTRODUCTION

Availability of amorphous semiconductors in the form of high quality multilayers provides potential applications in the field of micro- and optoelectronics.1,2 Among amorphous multilayers 共AMLs兲 chalcogenide multilayers are attractive because of the prominent photoinduced effects. Studies in chalcogenide AML have been directed towards two phenomena. One is photoinduced diffusion in short period multilayer systems, which finds potential applications in holographic recording and fabrication of phase gratings.2,3 The other is photodarkening or bleaching,4–8 which is also known in thick films.6,9 Studies on nanostructured chalcogenides are still at the infant stage. Although the misfit problems in AML are considerably reduced compared to crystalline superlattices, there are still some physical processes 共e.g., quantum confinement, diffusion兲 that are not well understood.4,10,11 Since most structural changes are related to atomic diffusion, understanding of the structural transformation must be based on the diffusion process. Moreover, in AML, the process of interdiffusion is not well understood. Photoinduced interdiffusion has been observed in amorphous Se/ As2S3 and similar multilayers.12–14 This is an india兲

Author to whom correspondence should be addressed; electronic mail: [email protected]

0021-8979/2006/99共9兲/094301/6/$23.00

cation that the illumination can enhance the atomic mobility. Interlayer diffusion by illumination is explained based on Fick’s second law15 as follows:

冉 冊

⳵C ⳵C ⳵ = D , ⳵x ⳵t ⳵x

共1兲

where D is the coefficient of interdiffusion. Assuming, for simplicity, a composition independent diffusion coefficient, the composition within a modulation period 共⌳兲 follows the analtytic expression

冉 冊

C共x兲 = 兺 A0i e−Dt共2␲i/⌳兲 sin i=0

2␲i x , ⌳

共2兲

where Ai is the Fourier coefficient of the initial concentration distribution in a multilayer along the x axis that is normal to the multilayer surface and t is the exposure time. It is assumed that the diffusion coefficient depends only on the temperature and light intensity. The thickness change in one pair of adjacent layers can be expressed through the concentration dependence of the density as follows: d = d0

冕

⌳

␥关C共x兲兴dx,

共3兲

0

where ␥关C共x兲兴 = dm / d0, where dm, d0, and d are the layer thickness of the intermixed multilayer, initial thickness before irradiation, and the change of total thickness after irra-

99, 094301-1

© 2006 American Institute of Physics

Downloaded 18 Aug 2006 to 203.200.43.195. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

094301-2

J. Appl. Phys. 99, 094301 共2006兲

Adarsh et al.

diation, respectively. The diffusional intermixing will affect several parameters of the multilayer, i.e., layer thickness, refractive index, optical band gap, and photoluminescence.12–14,16 Reported results showed that there is a change in optical absorption edge, layer thickness, photoluminescence 共PL兲 intensity, PL width, conductivity, and photoconductivity.12,14 These results indicate that light- and thermo-stimulated interdiffusion effects are rather similar. The changes in these properties are explained based on the formation of ternary solid solution due to interdiffusion. The remarkable difference in the photodiffusion in Ag/ As– S 共Ref. 17兲 multilayers 共MLs兲 and Se/ As2S3 MLs 共Ref. 18 are that Ag can easily diffuse into a depth of 1 ␮m, but for Se this depth has been estimated to be less than 100 Å. But the process by which Se diffuses in to As2S3 is still unclear. The difference in diffusion length may be due to ionized silver atoms migrating electrically. Since Se has chainlike molecular structures, photoinduced breakage of the molecules into isolated single atoms may occur with difficulty. So, Tanaka et al. suggested that neutral Se fragments, which could be produced by illumination, thermally diffuse into the neighboring amorphous regions.18 In this article, interdiffusion in Se/ As2S3 multilayered samples are studied by optical absorption and x-ray photoelectron spectroscopy 共XPS兲. Raman scattering and infrared spectroscopy techniques were used to study the interdiffusion, but the results were not satisfactory with regard to the intermixing.12 The characteristic spectra of components in the multilayer and those of the mixed layer were rather similar.5 In order to understand the diffusion mechanism of Se into As2S3 we mainly used the Tauc parameter 共B1/2兲 and Urbach energy. The main reason for using the above parameters is that they give information about the distribution of electronic states in the absorption edge region. XPS is used to analyze the new bonds formed between the components due to interdiffusion. XPS is a useful surface analytical technique to study the chemical state and local environment of an atom.19,20 The chemical bonding is often realized through correlation with chemical shifts in XPS binding energies of the corresponding elements. The effective optical band gap of the samples was determined using the equation

␣h␯ = B共h␯ − Eg兲2 ,

共4兲

where ␣, h, ␯, Eg, and B are the absorption coefficient, Plank’s constant, frequency, optical band gap, and a constant 共Tauc parameter兲, respectively.21 The constant B includes information on the convolution of the valence band and conduction band states and on the matrix element of optical transitions, which reflects not only the relaxed k selection rule but also the disorder induced spatial correlation of optical transitions between the valence band and conduction band.21 Moreover, B is highly dependent on the character of the bonding. At the energy levels where the Tauc model is used 共for photon energies corresponding to ␣ ⬎ 104兲, the joint density of states does not include tail states. The information at the band tails were obtained from the Urbach energy. The origin of the Urbach edge is still unclear, but two general mechanisms may be responsible: either the exponential dependence of ␣ 共absorption coefficient兲 arises from an expo-

nential energy dependence of the valence and conduction band densities at the band edges 共neglecting matrix element effects兲 or a universal absorption mechanism exists, which gives rise to the exponential behavior of ␣, e.g., the fieldbroadened exciton model of Dow and Redfield,22 Abel and Yoyozawa,23 and Soukoulis et al.24 Theoretically it has been shown that exponential band tails can result from potential fluctuations associated with structural disorder.24 Although there is experimental evidence in the case of a-Si: H that the magnitude of the Ee 共Urbach energy兲 is determined by the degree of disorder, that of chalcogenide is not very clear. EXPERIMENTAL PROCEDURES

Se/ As2S3-type ML1 共sublayer thickness of a-Se and As2S3 are 3–4 and 11– 12 nm兲 and ML2 共a-Se: 1 – 2 nm, As2S3: 11– 12 nm兲 were prepared by a cyclic thermal evaporation technique from bulk 共powdered兲 a-Se and As2S3. Deposition rates were 2 – 10 nm/ s in a vacuum of 5 ⫻ 10−4 Pa. Periodicity was monitored by the low angle x-ray diffraction method.2 To study photostimulated effects we irradiated the as-prepared samples with a diode pumped solid state laser of wavelength 532 nm and a power density of 1 W / cm2 up to 40 min at room temperature. The UVvisible-IR transmission spectrum was measured in the wavelength range from 400 to 1200 nm. Surface chemistry of the samples were studied by using electron spectroscopy for chemical analysis 共ESCA兲 at a vacuum of ⬃10−9 Torr. A monochromatic Al K␣ x-ray source 共h␯ = 1487 eV兲 was used for the analysis. For insulators such as glasses, the charging effect can change the binding energy 共BE兲 of the electrons from sample to sample. So the measurement of the absolute BE of electrons from a specified energy level is not reliable. The C 1s line from either adventitious carbon or intentionally added graphite powder on the surface has been widely used for charge referencing.25,26 For this study, the adventitious carbon was used as a reference and the BE of the reference C 1s line was set as 284.6 eV. For each sample, a calibration factor was calculated from the difference between the measured C 1s BE and the reference value 284.6 eV.27 The original BE data were corrected according to the calibration factor. RESULTS AND DISCUSSIONS

Photoinduced interdiffusion was observed with above band gap light 共2.37 eV兲. The optical absorption edge was measured and the results were interpreted based on the model of effective optical media.12,13 According to this model, narrow band gap “well” layers determine the absorption and the contribution from the wide band gap “barrier” layers is small. The effective optical band gap of our samples was determined using the equation 共␣h␯兲1/2 = B1/2共h␯ − Eg兲.

共5兲

Equation 共5兲 is valid for a number of amorphous materials in the spectral region of large ␣ 共104 ⬍ ␣ ⬍ 105兲, i.e., Tauc region.12,28 Se well layers determine the optical band gap in our AMLs and the band gap values are given in Table I. The optical band gap of ML1 is less than the optical band gap of

Downloaded 18 Aug 2006 to 203.200.43.195. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

094301-3

J. Appl. Phys. 99, 094301 共2006兲

Adarsh et al.

TABLE I. Slope of the Urbach edge and optical band gap of the samples. Here the slope of the Urbach edge is at room temperature.

Sample ML1 IML1 ML2 IML2

Urbach energy 共meV兲 共Ee兲

Optical band gap 共eV兲

B1/2 共cm eV−1/2兲

165± 1 160± 1 205± 3 149± 2

1.90± 0.002 2.06± 0.002 2.19± 0.002 2.24± 0.002

294± 1 329± 2 343± 2 478± 2

−1/2

ML2. This blueshift in optical band gap is due to the quantum confinement effect, which was known in amorphous superlattices.12,29 The optical band gap of irradiated samples clearly shows a blueshift compared to the corresponding asprepared samples. The blueshift can be explained by the creation of new bonds between the components due to interdiffusion and formation of continuous rows of ternary solid solutions.12 The parameter B1/2 of the Tauc region 关Eq. 共5兲兴 depends on the product of the oscillator strength of the optical transition, the deformation potential, and the mean deviation of the atomic coordinates.30 The values of B1/2 of the irradiated and as-prepared films are listed in Table I. The B1/2 value of ML2 is greater than that of ML1. This increase in the B1/2 value may be due to the changes in the electronic states because of the well-known quantum confinement effects. A decrease in Se sublayer thickness will result in an increase in structural disorder connected with possible bond angle distribution change.5 Based on the experiments in a-Ge– H, a-Si– H, a-SiN, and a-GeN based alloys, Zanatta and Chambouleyron proved that B1/2 is sensitive to topological disorder only when electronic structural changes occur.21 So we assume that the change in structural disorder will not affect the B1/2 values. B1/2 values of the irradiated samples are much greater than that of the as-prepared samples. During photodiffusion the density of Se–Se bonds decreases and changes in conduction and valence band states may occur. As a consequence, the absorption edge becomes steeper and B1/2 presents a high value, similar to the results observed by J. Robertson31 in a-Si and a-Ge alloys, a decreasing B1/2 has been observed with decrease in the Si–Si and Ge–Ge bond density. Before irradiation, the Se layers determine the B1/2 values. But after irradiation, it is determined by the solid solution of As2S3 – Se having Se–S or Se–As bonds. In the exponential part of the absorption edge 共where ␣ ⬍ 104兲, the absorption coefficient is governed by the socalled Urbach rule32

冉 冊

␣共h␯兲 = ␣0 exp

h␯ , Ee

共6兲

where the Urbach energy Ee characterizes the slope of this region. Plotting the dependence of log共␣兲 on photon energy will give a straight line. The calculated value of Ee, the inverse of the slope of the straight line, gives the width of the tails of the localized states into the gap at band edges.33 It is known that the Urbach edge is a useful parameter to evaluate the degree of disorder. Although there is experimental evidence in the case of a-Si: H that the magnitude of Ee is

FIG. 1. Optical absorption spectrum of 共1兲 ML1, 共2兲 IML1, 共3兲 ML2, and 共4兲 IML2.

determined by the degree of disorder, that of the chalcogenides is not very clear.34 Tanaka et al.35 have reported that the slope of the Urbach absorption edge does not change as a function of the quench temperature, even though the extended x-ray absorption fine structure 共EXAFS兲 measurements have indicated structural disorder.36 These contradictory results are explained in terms of the lone pair nature of the top of the valence band, i.e., tailing of the valence band is mainly responsible for the Urbach edge.37 Intramolecular fluctuations 共bond length and bond angle兲 have little effect. The values of the Ee of the irradiated and as-prepared films are listed in Table I. The Urbach energy Ee of ML2 is much greater than the Ee of ML1. The increase in Ee may be due to the quantum confinement effects that induce an increase in band gap, which may lead to an increase in the degree of tailing. A decrease in Se sublayer thickness results in a small increase in structural disorder connected with a possible bond angle distribution change,5 similar to that of aSi: H / a-SiNx superlattices.5,14,38 But this has very little effect on the change in the Urbach edge. A decrease in Urbach energy Ee is observed after photodiffusion. During photodiffusion the density of Se–Se bonds decreases and changes in conduction and valence band states may occur. We tried to associate the B1/2 with the Urbach energy based on the model proposed by Mott and Davis39 in which B⬀

关N共Ec兲兴2 , 共n0⌬Ec兲

共7兲

where N共Ec兲, n0, and ⌬Ec are the density of states at the conduction band edge, the index of refraction, and the width of the conduction band tail, respectively. Experimental data of B1/2 and Ee for as-prepared samples do not show the clear correlation implied by the above equation, i.e., both B1/2 and Ee increase with a decrease in Se thickness. In the case of ML1 and IML1, the Ee does not change much, but the B1/2 changes appreciably 共Fig. 1兲. But for ML2 and IML2, both the values change appreciably. The increase in B1/2 of the

Downloaded 18 Aug 2006 to 203.200.43.195. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

094301-4

J. Appl. Phys. 99, 094301 共2006兲

Adarsh et al.

FIG. 2. Tauc plots of optical absorption coefficient 关共␣h␯兲1/2 vs h␯兴 for IML2. The experimental spectra are represented by symbols “␱;” the solid line represents the theoretical fit.

FIG. 4. Different photoelectron and Auger peaks 共As, S, and Se兲 between 150– 280 eV. The detailed information about these peaks can be obtained from Ref. 33.

irradiated samples over the respective as-prepared samples can be explained based on the change in refractive index, changes in the density of states at the conduction band edge, and ⌬Ec. Palyok et al.2 and Kikineshi et al.15 found that the refractive index decreases with photodiffusion in Se/ As2S3 multilayers. But there are no experimental evidences available for the variation of the other two parameters. The maximum decrease in refractive index is only 4%. In the case of ML1and IML1 there is only a small change in the Urbach energy 共we assume that n0 and ⌬Ec are more or less constant for ML1 and IML1兲, but the B1/2 changes considerably. From this we infer that the N共Ec兲 increases with photodiffusion. Before irradiation, in Se/ As2S3 multilayers the valence band and conduction bands are formed by the lone pair electrons of Se and the empty antibonding orbital, but after irradiation it is determined by the solid solution of Se– As2S3. During photodiffusion the density of Se–Se bonds decreases and

changes in conduction and valence band states may occur. The higher values of B1/2 and lower values of Ee of the irradiated samples over the corresponding as-prepared samples clearly indicate that the irradiated samples are more ordered 共chemically兲 than the as-prepared samples, i.e., the removal of homopolar bonds and formation of heteropolar bonds after photodiffusion 共Fig. 2兲. A typical XPS spectrum of Se/ As2S3 multilayer is shown in Fig. 3. Se, As, and S have many photoelectron and Auger peaks 共Fig. 4兲. The detailed information about the BE of all those peaks can be found in Ref. 39. The BE of As 3d and Se 3d of all the samples are listed in Table. II. The BE of the Se 3d peak of ML1 and ML2 are at the same energy, i.e., 55.1 eV. Since the BE of elemental Se is 55 eV,35 we assume that this peak is due to the Se–Se bond. In addition to the peak at 55.1 eV, it has a satellite peak at 56.3 eV. This represents the S–Se bond. Because S has a higher electronegativity than Se,40 the BE of Se shifts to higher energy. It is evident that there are Se–Se and S–Se bonds in the asprepared samples. The BE spectra of As 3d contain two peaks in ML1 and three peaks in ML2 共Fig. 5兲. The peaks at 42.9 and 45 eV of ML1 and ML2 are due to the As–S bond and Se–Se bond. The peak at 41.8 eV present in ML2, which is absent in ML1, is due to the As–As bond, because the BE of the 3d peak of elemental As is at 41.9 eV. Even though As–As homopolar bonds exist in ML1 and ML2, the possible TABLE II. XPS core level BE values of As 3d and Se 3d for all the samples.

Sample ML1 IML1 ML2 IML2

As 共3d兲 共eV兲

Se 共3d兲 共eV兲

42.9, 45 42.1, 42.9, 45 41.8, 42.9, 45 42.9, 45

55.1, 56.3 54.3, 55.1, 56.5 55.1, 56.3 55.2, 56.5

FIG. 3. Typical XPS spectra of ML1.

Downloaded 18 Aug 2006 to 203.200.43.195. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

094301-5

J. Appl. Phys. 99, 094301 共2006兲

Adarsh et al.

FIG. 5. As 3d peak of IML2.

FIG. 6. Se 3d peak of IML2.

reason for the absence of this peak in ML1 may be due to the thick Se layer 共3 – 4 nm兲, which is on top of the As2S3 layer. The XPS signal will come mainly from the top 4 – 5 nm, i.e., from Se layer and the self-diffused boundary layers. We expect that As–As homopolar bonds exist only in the As2S3 layer and not in the diffused region. Nordman et al.41 suggested that irreversible photostructural changes take place only at the strained sites 共regions with homopolar bonds兲. So we assume that diffusion may take place through the strained sites. Therefore there is a very small probability of having the As–As bonds in the diffused region. Since in ML2 the top Se sublayer thickness is very small 共1 – 2 nm兲, we will get the XPS signal from both the Se and As2S3 layers. The BE spectra of As 3d of IML1 and IML2 contain three and two peaks, respectively. The peaks at 42.9 and 45 eV of IML1 and IML2 are due to the As–S and Se–Se bonds, respectively. The peak at 42.1 eV that is present only in IML1 may be due to the As–Se bond. Since Se’s electronegativity is greater than As but less than Se,40 the BE of the As–Se bond will lie between the As–As and As–S bonds. So it is evident that during photodiffusion As–Se bonds are formed. The BE spectrum of Se 3d contains three peaks in IML1 and two peaks in IML2 共Fig. 6兲. Peaks at 55.1 and 56.5 eV are due to Se–Se and S–Se bonds. The peak at 54.3 eV, which is present only in IML1, may be due to the As–Se bond. The As 3d spectra of IML1 shows the As–Se bond formation. The possible reason for the absence of this bond in IML2 may be due to the lower concentration of Se. If we compare the Se 3d BE of the as-prepared and irradiated samples, we can infer the following. It is clear that during photodiffusion Se is forming bonds with both As and S. As–As bonds are present in the as-prepared samples. These bonds are converted into As–S or As–Se bonds with light irradiation. Earlier studies on unannealed films have shown that the evaporated As2S3 sublayer contains a large number of species such As, As4, S2, S8, AsS3, As4S5, As4S4, As2, etc.42 This means that even in a stochiometric film such

as As2S3, where the stochiometry would only allow As–S bonds, a large number of so-called wrong bonds 共As–As, S–S兲 are present. Spectroscopic studies on unannealed films have shown that irreversible band gap light-induced photostructural changes are mainly due to the photoinduced As–As bond breaking 共As–As being the weakest bond兲 followed by the phonon-assisted creation of As–S bonds.41 2As30 + S20 + h␯ ↔ As2+ + As2− + S20 + phonon ↔ As30 + S3+ + As2− .

共8兲

Here the superscript indices designate the electric charge of atoms, and the subscript indices their coordination number. After that phonon-assisted As–S bond formation takes place by using the lone pair ␲ electrons43,44 of S20. We expect all these defects in our As2S3 sublayer. XPS data of our samples show that during light irradiation these defects are removed, i.e., As atoms will form As–S or As–Se and S forms S–Se or S–As bonds 共heteropolar bonds replace homopolar bonds兲. In our AMLs the above reaction takes place at the As2S3 rich region. But at the interfaces, in addition to the above reaction, the wrong sulphur bonds easily react with Se forming S–Se bonds. If the Se concentration in the multilayer is very high, Se can react to As–As wrong bonds and the reaction is as follows: 2As30 + Se20 + h␯ ↔ As2+ + As2− + Se20 + phonon ↔ As30 + Se3+ + As2− .

共9兲

The As–Se bond was observed in IML1, where the Se concentration is very high. If we compare the XPS data with the optical parameters of the as-prepared and irradiated samples, we will get a detailed picture about the chemical ordering 共heteropolar bonds replace homopolar bonds兲. An increase in B1/2 and a decrease in the Urbach energy were observed with photodiffusion. After photodiffusion the samples become chemically ordered.

Downloaded 18 Aug 2006 to 203.200.43.195. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

094301-6

J. Appl. Phys. 99, 094301 共2006兲

Adarsh et al.

Based on the above experimental data, we propose a simple model to explain the mechanism of Se diffusion in As2S3. The as-prepared As2S3 sublayer contains a large number of S–S and As–As wrong bonds. We assume that diffusion takes place through the reaction of Se20 with the wrong S–S 共S20兲 bonds and As–As bonds. The Se20 / S20 defects are known as annihilating defects 共ADs兲.41 It should be noted that diffusion takes place mainly in strained sites 共where wrong bonds are close to Se20 sites兲. During light irradiation, a decrease in photoinduced viscosity leads to a mechanical stress gradient and thus enables the directed motion of ADs. Se20 will react with S20, forming S–Se and As–As to As–Se. A cooperative 共because of mechanical stresses in a continuous glass network and strong electron-phonon interaction in amorphous chalocogenides兲 annihilation of ADs results in the formation of As–Se, S–Se, and As–S bonds. These bond rearrangements continue as long as the material is exposed to light and this dynamic state is comparable to that near the glass transition temperature. It is clear that the illuminated material will flow under uniaxial stress since the local bond changes and atomic motions tend to decrease the local strain energy. These cycles continue until the local stresses are released and the concentration of As–S, As–Se, and S–Se bonds is large enough to increase the viscosity and prevent the AD diffusion. This atomic diffusion occurs through the lone pair electrons of Se and S 关refer to Eqs. 共8兲 and 共9兲兴. The cumulative effect of local configuration changes produces changes in optical and electrical properties of the glasses. If we compare the As and Se 3d peaks of IML1 and IML2, we can see clearly that Se is forming bonds with both As and S after irradiation, which supports the above discussion. IV. CONCLUSIONS

We found that the Urbach edge 共Ee兲 and B1/2 共Tauc’s parameter兲 change with changes in Se sublayer thickness and also with photodiffusion. Experimental data of B1/2 and Ee for as-prepared samples do not show the clear correlation implied by the Mott-Davis model. The increase found in the B1/2 after photodiffusion is coherent with the corresponding decrease of the Urbach energy. This fact is discussed in terms of the structural changes induced by the photoinduced interdiffusion, i.e., creation of new bonds between components, which modifies the conduction and valence bands. XPS analysis shows that, during photodiffusion, homopolar bonds are replaced by heteropolar bonds, i.e., the irradiated samples are chemically ordered than the corresponding as-prepared samples. We proposed a model to describe the photodiffusion, which suggests that diffusion takes place through the wrong bonds. ACKNOWLEDGMENTS

One of the authors 共K.V.A.兲 thanks CSIR for financial support. The authors thank the bilateral Indo-Hungarian R&D and Hungarian OTKA 共N T046758兲 grants. The au-

thors also thank the Inter-University Consortium for DAE facilities for XPS measurements. 1

M. Hirose and S. Miyazaki, Japan Annual Reviews in Electronics, Computers and Telecommunications, edited by Y. Hamakawa 共OHM-NorthHolland, Amsterdam, 1987兲, Vol. 22, p. 147. 2 V. Palyok, A. Mishak, I. Szabo, D. L. Beke, and A. Kikineshi, Appl. Phys. A: Mater. Sci. Process. 68, 489 共1999兲. 3 T. Wagner and P. J. S. Ewen, J. Non-Cryst. Solids 266–269, 979 共2000兲. 4 A. Kikineshi and A. Mishak, NATO ASI Ser., Ser. 3 36, 249 共1997兲. 5 A. Kikineshi, Opt. Eng. 共Bellingham兲 34, 1040 共1995兲. 6 H. Eguchi, Y. Suzuki, and M. Hirai, J. Non-Cryst. Solids 95–96, 757 共1987兲. 7 K. Hayashi and N. Mitsuishi, J. Non-Cryst. Solids 299–302, 949 共2002兲. 8 K. Tanaka, S. Kyohya, and A. Odajima, Thin Solid Films 111, 195 共1984兲. 9 I. I. Turjanitsa and A. Kikineshi, D. G. Ukr. J. Phys. 24, 534 共1979兲. 10 D. Nesheva, D. Arsova, and Z. Levi, Philos. Mag. B 69, 205 共1994兲. 11 A. L. Greer, Defect Diffus. Forum 129–130, 163 共1996兲. 12 M. Malyovanik, M. Shiplyak, V. Cheresnya, T. Remeta, S. Ivan, and A. Kikineshi, J. Optoelectron. Adv. Mater. 5, 397 共2003兲. 13 M. Malyovanik, S. Ivan, A. Csik, G. A. Langer, D. L. Beke, and S. Kokenyesi, J. Appl. Phys. 93, 139 共2003兲. 14 K. V. Adarsh, K. S. Sangunni, S. Kokenyesi, S. Ivan, and M. Shipljak, J. Appl. Phys. 97, 044314 共2005兲. 15 A. Kikineshi, V. Paylok, I. A. Szabo, M. Shipljak, I. Ivan, and D. L. Beke, J. Non-Cryst. Solids 326–327, 484 共2003兲. 16 V. Paylok and M. Malyovanik, J. Optoelectron. Adv. Mater. 1, 77 共1999兲. 17 A. E. Owen, A. P. Firth, and P. J. S. Ewen, Philos. Mag. B 52, 347 共1985兲. 18 K. Tanaka, Y. Ichimura, and M. Komasaki, Thin Solid Films 189, 51 共1990兲. 19 P. I. K. Onorato, M. N. Alexander, C. W. Struck, G. W. Tasker, and D. R. Uhlmann, J. Am. Ceram. Soc. 68, 148 共1985兲. 20 J. Heo, J. S. Sanghera, and J. D. Mackenzie, J. Non-Cryst. Solids 101, 23 共1988兲. 21 A. R. Zanatta and I. Chambouleyron, Phys. Rev. B 53, 3833 共1996兲. 22 J. D. Dow and D. Redfield, Phys. Rev. B 1, 3358 共1970兲. 23 S. Abe and Y. Yoyozawa, J. Phys. Soc. Jpn. 50, 2185 共1981兲. 24 C. M. Soukoulis, M. H. Cohennd, and E. N. Economou, Phys. Rev. Lett. 53, 616 共1984兲. 25 J. F. Moulder, W. F. Sticker, P. E. Sobol, and K. D. Bomben, Hand Book of X-ray Photoelectron Spectroscopy 共Perkin Elmer, Eden Pairie, MN, 1992兲. 26 W. J. Stec, W. E. Morgan, R. G. Albridge, and J. R. Van Wazer, Inorg. Chem. 11, 219 共1972兲. 27 T. L. Barr and S. Seal, J. Vac. Sci. Technol. A 13, 1239 共1995兲. 28 R. Bindemann and O. Paetzold, Phys. Status Solidi B 160, K183 共1990兲. 29 B. Abeles and T. Tiedje, Phys. Rev. Lett. 51, 2003 共1983兲. 30 R. Tsu, P. Menna, and H. Mahan, Sol. Cells 21, 189 共1987兲. 31 J. Robertson, Philos. Mag. B 69, 307 共1994兲. 32 F. Urbach, Phys. Rev. 92, 1324 共1953兲. 33 M. M. Hafiz, A. H. Moharram, and A. A. Abu-Sehly, Appl. Phys. A: Mater. Sci. Process. 66, 217 共1998兲. 34 S. R. Elliot, Materials Science and Technology: A Comprehensive Treatment 共Wiley, New York, 1996兲, 19 and 375. 35 K. Tanaka., S. Godha, and A. Odajima, Solid State Commun. 56, 899 共1985兲. 36 C. Y. Yang, D. E. Sayers, and M. A. Paesler, Phys. Rev. B 36, 8122 共1987兲. 37 K. Tanaka, Disordered Systems and New Materials 共World Scientific, Singapore, 1989兲, p. 290. 38 N. Maley and J. S. Lanin, Phys. Rev. B 31, 5577 共1985兲. 39 N. F. Mott and E. A. Davis, Electronic Process in Non-crystalline Materials 共Clarendon, Oxford, 1979兲, p. 287. 40 CRC Handbook of Chemistry and Physics, edited by D. R. Lide, 78th ed. 共CRC, Boca Raton, FL, 1997兲. 41 O. Nordman, O. Ozols, and N. Nordman, J. Opt. Soc. Am. B 16, 631 共1999兲. 42 K. Shimakawa, A. V. Kolobov, and S. R. Elliot, Adv. Phys. 44, 475 共1995兲. 43 K. Schwartz, The Physics of Optical Recording 共Springer-Verlag, Berlin, 1993兲. 44 O. I. Shopotyuk, Phys. Status Solidi B 183, 365 共1994兲.

Downloaded 18 Aug 2006 to 203.200.43.195. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp