www.elsevier.com/locate/jnoncrysol

Journal of Non-Crystalline Solids 352 (2006) 562–566

An XPS study of the early stages of silver photodiffusionin Ag/a-As2S3 films

Himanshu Jain *, Andriy Kovalskiy, Alfred Miller

Materials Science and Engineering Department, Lehigh University, 5 East Packer Avenue, Whitaker Lab. 5, Bethlehem, 18015 PA, USA

Available online 21 February 2006

Abstract

X-ray induced structural changes at the Ag/As2S3 interface are investigated using X-ray photoelectron spectroscopy on the samplesprepared within the spectrometer. The as-prepared film consists of stable heteropolar As–S bonds as well as �16% S (and As) atoms inlower (higher) electron density configurations such as the –S–S– (–As–As–) segments with ‘wrong’ homopolar bonds. Two distinct stagesof the X-ray induced diffusion are revealed. At first, silver reacts with atoms within –S–S– like segments to form Ag–S bonds. In thesecond stage, the Ag–S bonds decompose due to the reaction of S with As atoms within the –As–As– ‘wrong’ segments to formAs–S heteropolar bonds, and silver diffuses away from the interface into the film. The results provide guideline for enhancing silverphotodiffusion in chalcogenide glass. The irradiation of the (Ag–Te)/As2S3 sample with X-rays shows that not only Ag, but Te alsodiffuses away from the surface.� 2006 Elsevier B.V. All rights reserved.

PACS: 79.60.Ht; 68.35.Fx; 61.43.Fs

Keywords: Diffusion and transport; Chalcogenides; Photoinduced effects; XPS

1. Introduction

It is well known that the diffusion of silver into chalco-genide glass (ChG) matrix is enhanced by increasing thetemperature, or by exposure to visible light (photodoping)[1,2], X-rays [3,4], electrons [5] and ions [6]. The unusualradiation enhanced diffusion, accompanied by the changein chemical reactivity, is very useful in modern lithography,programmable metallization cell devices, as well as in thedevelopment of holographic materials, relief images, dif-fractive optical elements, MEMS structures, submicronpatterns, etc. [7–10]. In spite of numerous investigationsduring the past three decades [1,2,7,11–17], the mechanismsof irradiation induced silver diffusion in ChG matrixremain unclear, especially with regard to ionizing radiation(X-rays, electrons and ions). There is a disagreement in theliterature even about the number and characteristics of the

0022-3093/$ - see front matter � 2006 Elsevier B.V. All rights reserved.

doi:10.1016/j.jnoncrysol.2005.11.044

* Corresponding author. Tel.: +1 610 758 4217; fax: +1 610 758 4244.E-mail address: h.jain@lehigh.edu (H. Jain).

stages of silver photodiffusion. Very few investigations aredevoted to the compositional features of this phenomenon.Therefore, in the present work we have employed X-rayphotoelectron spectroscopy (XPS) to characterize in situthe early stages of X-ray induced diffusion as well ascorresponding changes of electronic structure at theAg/As2S3 interface. Similar observations are also made atthe (Ag–Te)/As2S3 interface.

2. Experimental

As2S3 films (�500 A thick) were deposited on Si sub-strate (Wacker Siltronic Corp., 525 ± 20 lm thickness) bythermal evaporation of the bulk glass (Amorphous Materi-als Inc.) of same composition, followed by the depositionof an �70 A thick film of silver metal (Puratronic,99.9999%) on top of the ChG film. In a few samples thefilm of silver–tellurium alloy (86% Ag + 14% Te, as mea-sured by XPS) was deposited, in place of pure silver, ontop of the As2S3 layer.

1000 800 600 400 200 00

10000

20000

30000

40000

1 - Ag/As2S3

2 - Ag-Te/As2S3

2

1

Cou

nts

Binding energy (eV)

Fig. 1. XPS survey spectra of the surfaces of Ag/As2S3 (curve 1) and(Ag–Te)/As2S3 (curve 2) thin film multilayers.

45 44 43 42 410

1000

2000

3000

4000

5000

6000

7000

a

15.5 %

84.5 %

As2S3

As3d

Cou

nts

Binding energy (eV)

experimental curve sum of deconvoluted components As atoms within AsS3/2 pyramidal units As atoms within units containing As-As bonds

165 164 163 162 161

2000

4000

6000

8000

10000

b

84 %

16 %

As2S3

S2p

experimental curvesum of deconvoluted components

S atoms within AsS3/2 pyramidal unitsS atoms within units containing S-S bonds

Cou

nts

Binding energy (eV)

Fig. 2. Deconvoluted As3d (a) and S2p (b) core level XPS spectra ofthermally evaporated As2S3 thin film.

H. Jain et al. / Journal of Non-Crystalline Solids 352 (2006) 562–566 563

Considering that the ChG films are photosensitive andreadily contaminated by exposure to oxygen [18], the sam-ples were prepared by thermal evaporation in darknessinside the XPS spectrometer chamber under high vacuum(�10�7 Torr for As2S3 and �10�8 Torr for Ag). So thepresent samples had no contact with ambient atmospheretill the end of measurements.

In our XPS experiment, the X-ray beam served both asthe excitation probe of core level electrons and the radia-tion that may produce photodiffusion and structuralchanges. The XPS core level spectra were obtained usingScienta ESCA-300 spectrometer with monochromatic AlKa X-rays (1486.6 eV). For all measurements the anglebetween the surface and the detector was 90�, so that thedepth of analysis was �100 A.

The XPS data analysis was conducted with standardESCA-300 software package. Concentrations of variouselements were determined from the area of respective corelevel peaks after subtracting the background. All core levelspectra were analyzed using Voigt function for peak shape.The parameters of analysis such as full-width-at-half-max-imum (FWHM), relative mix of Gaussian and Lorentziancomponents, and the asymmetry were chosen to be thesame for the different spin-orbit components of a core levelpeak of given chemical element. To study the influence ofX-ray induced Ag diffusion on the electronic structure ofsamples, the XPS data were collected with the start of irra-diation, during the first hour of irradiation and after 16 hof X-ray exposure. For measurements during the first hourof irradiation, the spectra were recorded in single sweep,whereas the spectra at the very beginning of X-ray expo-sure and after 16 h were recorded three times for better sta-tistics needed for their deconvolution and determination ofconcentrations with sufficient precision. For reference, thecore level spectra of freshly deposited As2S3 films beforethe deposition of silver layer were also recorded andanalyzed.

3. Results

The XPS survey spectra of Ag/As2S3 and (Ag–Te)/As2S3

samples, as shown in Fig. 1, verify that no detectable oxy-gen was present on the sample surface. The As3d and S2pcore level spectra (Fig. 2) of the freshly deposited As2S3

film indicate slight deviation of film composition(S/As = 58/42 (±2%)) from the stoichiometric ratio ofS/As = 60/40.

Both As3d and S2p spectra of the as-prepared As2S3

film consist of two peaks corresponding to 5/2 and 3/2 spinstates of 3d orbitals (intensity ratio 5:3), and 3/2 and 1/2spin states of the 2p orbitals (intensity ratio 3:2), respec-tively. From the curve fitting of the spectra we find thatit is not possible to describe either the S2p or the As3dspectrum of freshly prepared As2S3 film as arising from sin-gle species pair of peaks. Therefore, both spectra have beendeconvoluted assuming two pairs of peaks representingtwo different bonding states of these two elements. The

45 44 43 42 410

2000

4000

6000

8000

10000

12000

14000

16000

18000

17 %

83 %

Ag/As2S3

As3d4 min X-ray

a experimental curvesum of deconvoluted components

As atoms within AsS3/2 pyramidal unitsAs atoms within units containing As-As bonds

Cou

nts

Binding energy (eV)

165 164 163 162 161 160

4000

6000

8000

10000

12000

14000

16000

18000

20000

22000

24000b

14.5 %

85.5 %

Ag/As2S3

S2p4 min X-ray

experimental curve sum of deconvoluted componentsS atoms within AsS3/2 pyramidal unitsS atoms bonded with Ag

Cou

nts

Binding energy (eV)

Fig. 3. Deconvoluted As3d (a) and S2p (b) core level XPS spectra ofAg/As2S3 surface at the beginning of X-ray irradiation.

168 166 164 162 160 158 1560

5000

10000

15000

20000

25000

30000

Ag/As2S3

S2p

4 min X-ray irradiation16 h X-ray irradiation

Cou

nts

Binding energy (eV)

Fig. 4. S2p core level XPS spectra of Ag/As2S3 sample surface at thebeginning and after 16 h of X-ray irradiation.

45 44 43 42 410

2000

4000

6000

8000

10000

12000

14000

16000

18000

20000

13.5 %

86.5 %

a

Ag/As2S3

As3d16 h X-ray

experimental curve sum of deconvoluted components

As atoms within AsS3/2 pyramidal unitsAs atoms within units containing As-As bonds

Cou

nts

Binding energy (eV)

165 164 163 162 161 160

5000

10000

15000

20000

25000

30000

7 %

93 %

b

Ag/As2S3

S2p16 h X-ray

experimental curvesum of deconvoluted components

S atoms within AsS3/2 pyramidal units S atoms bonded with Ag

Cou

nts

Binding energy (eV)

Fig. 5. Deconvoluted As3d (a) and S2p (b) core level XPS spectra of theAg/As2S3 surface after 16 h of X-ray irradiation.

50 49 48 47 46 45 44 43 42 41 400

5000

10000

15000

20000

Te4d5/2

As3dTe4d3/2

Cou

nts

Binding energy (eV)

As2S3

(Ag-Te)/As2S3 4 min X-ray (Ag-Te)/As2S3 55 min X-ray

Fig. 6. As3d core level XPS spectra of the surface of (Ag–Te)/As2S3

multilayer.

564 H. Jain et al. / Journal of Non-Crystalline Solids 352 (2006) 562–566

H. Jain et al. / Journal of Non-Crystalline Solids 352 (2006) 562–566 565

results of deconvolution indicate the presence of �84% Asand S atoms in one chemical state or configuration and�16% in another configuration. The binding energy ofthe smaller peak of As and S representing minority config-urations shows the chemical shifts of �0.6 eV and +0.7 eV,respectively, from their major component.

The As3d and S2p spectra for the Ag/As2S3 sample havebeen analyzed following the above procedure for the refer-ence films without silver in Fig. 2. The two components ofAs3d spectrum agree well with those observed for theAs2S3 film (Fig. 3(a)), but the S2p core level spectrumreveals important differences (Fig. 3(b)). The latter can stillbe analyzed as made of two components representing twodifferent bonding states of sulfur, but the position of theminority component is significantly different than seen inthe reference As2S3 spectrum. The smaller peak initially sit-uated at +0.7 eV from the main peak has disappeared.Instead, we observe a different peak of similar magnitudebut shifted by �0.6 eV with respect to the main peak.

As a result of X-ray irradiation the Ag concentration onthe surface decreases from 12.5 to 6.2 at.% during 16 h.During the same period the shape of Ag3d5/2 core levelpeak does not show any appreciable change. The X-rayinduced silver diffusion, as noted from the decrease of silverXPS signal from the surface region, is accompanied by thenarrowing of S2p FWHM (Fig. 4). At the same time theshape of As3d peak remains relatively unchanged (notshown), indicating that the narrowing of sulfur peak isnot from any change in the sample condition (like charg-ing). Another important change caused by X-ray irradia-tion is the relative decrease of the minority peaks in boththe deconvoluted As3d and S2p spectra (Fig. 5). The effectof X-ray irradiation on the As3d and S2p core level spectrafor the (Ag–Te)/As2S3 sample is similar to that observedfor the Ag/As2S3 sample. The only notable observation isthat during 55 min of X-ray irradiation, there is a veryrapid decrease in the area of Te4d5/2 XPS peak, and hencethe surface concentration of Te (see Fig. 6).

4. Discussion

The photoinduced effects of bandgap light and corre-sponding changes in the XPS spectra of chalcogenide glassfilms are well documented [19–21]. Our earlier studiesshowed negligible effect of X-ray irradiation on the atomicor electronic structure of simple chalcogenide glass films(i.e. without silver). Therefore, one might anticipate thatX-rays would not cause significant changes in the electronicstructure of the Ag/As2S3 films either; for the same reasonone may expect negligible effect of X-ray irradiation on sil-ver diffusion. Numerous previous XPS studies of photo andthermally induced silver diffusion in chalcogenide glassesdid not take into consideration of any X-ray induced effects[11,22–27]. So contrary to general expectation, we haveobserved significant and very interesting X-ray inducedchanges in the As3d and S2p core level spectra of theAg/As2S3 samples. We find that not only the near-bandgap

light but also much higher energy penetrating radiation,specifically the 1486.6 eV X-rays in the present work, signif-icantly induces silver diffusion into ChG matrix, and alterthe chemical structure of the surface region.

From a comparison of the S2p core level spectrum forthe As2S3 and Ag/As2S3 films (Figs. 2(b) and 3(b)) we con-clude that as soon as the latter sample is exposed to X-rayirradiation, Ag atoms interact with S atoms forming theAg–S bonds. The as-prepared As2S3 film consists of twokinds of S atoms as indicated by two peaks in Fig. 2(b).The large peak at higher binding energy (BE) representsS in stable As–S–As configuration. The As–S bonds in suchsegments should be the only type of bonds in a stoichiom-etric material without any bonding defects. The other peakrepresents 16% of S atoms such as in –S–S– chain segmentsat a higher BE as expected from their relatively lower elec-tron density than on S in As–S–As bonds. The observedhigher binding energy configurations represent ‘wrong’homopolar bonds, which are formed due to the rapid for-mation of film from vapor state into non-equilibriumglassy state. The concentration of these wrongly bondedatoms is similar to that found in As2Se3 films [20]. We findthat the S2p peak due to –S–S– type configurations van-ishes in the S2p spectrum of Ag/As2S3 sample shown inFig. 3(b). At the same time more ionic Ag–S bonds withhigher electron density around S form, which give rise toanother S2p peak at a lower binding energy than the mainpeak due to the S atoms in As–S bonds. Evidently, Agreacts essentially with those S atoms that are present in–S–S– like configurations. So if one wishes to enhance(retard) the first stage of silver diffusion, the recommenda-tion from present observation would be to increase(decrease) the concentration of wrong bonds/configura-tions such as by a faster (slower) formation of the film.

On exposure to X-rays for 16 h, three significant changesin the surface layer are noted from the XPS spectra (com-pare Fig. 3(a) with 5(a)): (a) The total concentration of sil-ver decreases from 12.5 to 6.2 at.%. (b) The fraction of S inAg–S bonds with respect to total S (in Ag–S and As–S–Asconfigurations) decreases from 14.5% to 7%, and (c) thefraction of As in higher electron density configuration thanin S–As–S bonds, such as –As–As– containing segmentswith respect to total arsenic decreases from 17 to 13.5at.%. These observations indicate that in time As atoms inthe –As–As– segments and Ag–S bonds break up to formstable As–S bonds with the silver atoms no longer bondedto any other atom. We have found that the diffusion of sil-ver is sensitive to the presence of electric field in the sample,indicating that upon breaking from Ag–S the silver atomsare actually present as Ag+ ions at this stage.

Based on the above experimental data we propose thefollowing mechanism of X-ray induced silver diffusion inAg/As2S3 samples: to begin with the starting film containssignificant concentration of low electron density –S–S– seg-ments with homopolar bonds. When silver is deposited onthis film and the sample is tested under X-ray irradiation,in Stage I Ag–S bonds such as in Ag2S form by reaction

566 H. Jain et al. / Journal of Non-Crystalline Solids 352 (2006) 562–566

of Ag with S in these configurations, as described by thefollowing reaction:

4Ag + [SAS]! 2Ag2S ð1ÞIn Stage II, as X-ray irradiation continues, the –As–As-containing segments are activated to react with the recentlyformed Ag2S molecules, leading to the formation of stableAs–S bonds as in As2S3 and liberation of silver. This step isconfirmed by the decrease of the concentration of As in‘wrong’ homopolar As–As bonds

3Ag2S + [AsAAs]!As2S3 + 6Agþ+ 6e� ð2ÞThe net result of the two stages is that the wrong S–S andAs–As bonds are replaced by energetically more stableAs–S bonds and the metallic silver atoms are convertedto Ag+. Subsequently, the so formed silver ions then diffuserapidly deeper into the sample, presumably by one of themechanisms proposed previously [7,16,28]. The presentexperiments elucidate the initial stages of photoinducedsilver diffusion in arsenic sulfide amorphous films. Onceinside the film, the silver ions may react to form againAg2S by reaction (1). It is plausible that when the concen-tration of both Ag2S and As2S3 is high, as would be thecase if silver diffusion rates were high, stable ternary phasesof Ag, As and S are formed as below [16,27]:

Ag2S + As2S3! 2AgAsS2 ð3Þor

3Ag2S + As2S3! 2Ag3AsS3 ð4ÞIt is noteworthy that during X-ray irradiation of

Ag/As2S3 sample, the S/As ratio remained stable. It showsthat during the whole period of X-ray induced diffusionthere is no significant migration of As or S near the surfaceof the sample, although different bonds or structural unitsmay break and form as a result of irradiation. The S2p peakafter the long irradiation of 16 h is sharper than at thebeginning (see Fig. 4), indicating that a chemically moreordered structure is restored by the radiation induced diffu-sion of silver.

The overall structure of (Ag–Te) thin layer onto As2S3 isrevealed in the survey spectrum of Fig. 1. A close up of tel-lurium signal is seen more clearly in the Te4d peak, whichoccurs next to the As3d peak (Fig. 6). As in the case ofAg/As2S3, X-rays induce silver diffusion into the As2S3 film.However, we also note a rapid decrease of Te peak withX-ray irradiation time. It suggests that Te atoms, unlikesulfur atom, diffuse into the sample body. Additional exper-iments are needed to further understand the role of Te.

5. Conclusions

The XPS of Ag/As2S3 amorphous thin film samplesprepared within the spectrometer has revealed new insightof photodiffusion of silver in chalcogenide glasses. TheX-ray induced diffusion of silver, being relatively slowerthan under bandgap light illumination, has helped identifytwo distinct stages of photodiffusion. In Stage I, a metasta-

ble phase containing Ag–S bonds is formed due to the reac-tion of Ag atoms with S in ‘wrong’ homopolar bonds.Therefore, we predict that silver photodiffusion will beslower when the concentration of wrong bonds/segmentsis reduced such as by thermal annealing of the films. In stageII, prolonged X-ray irradiation causes breaking of Ag–Sbonds and formation of As–S heteropolar bonds; Ag+ ionsare liberated at the same time, which migrate rapidly intothe film. The X-ray irradiation of (Ag–Te)/As2S3 sampleshows similar stages of photodiffusion; in addition theresults indicate rapid diffusion of Te atoms into the film.

Acknowledgments

Separate parts of this work were supported by a LehighUniversity – Army Research Lab (ARL) collaborativeresearch program, and the National Science Foundation(DMR 03-12081, DMR 04-09588).

References

[1] T. Wagner, A. Mackova, V. Perina, E. Rauhala, A. Seppala, S.O.Kasap, M. Frumar, Mir. Vlcek, Mil. Vlcek, J. Non-Cryst. Solids 299–302 (2002) 1028.

[2] M. Mitkova, M.N. Kozicki, H.C. Kim, T.L. Alford, Thin Solid Films449 (2004) 248.

[3] Y. Utsugi, M. Kakuchi, Rev. Sci. Instrum. 60 (1989) 2295.[4] K.D. Kolwicz, M.S. Chang, J. Electrochem. Soc. 127 (1980) 135.[5] K. Balasubramanyam, A.L. Ruoff, J. Vac. Sci. Technol. 19 (1981)

1374.[6] W. Beyer, R. Klabes, A. Thomas, G. Kluge, R. Grotzschel,

P. Suptitz, Phys. Status Solidi A 101 (1987) K9.[7] A.V. Kolobov, S.R. Elliott, Adv. Phys. 40 (1991) 625.[8] A. Zakery, C.W. Slinger, P.J.S. Ewen, A.P. Firth, A.E. Owen, J. Phys.

D: Appl. Phys. 21 (1988) 78.[9] R.G. Vadimsky, J. Vac. Sci. Technol. B 6 (1988) 2221.

[10] M. Mitkova, M.N. Kozicki, J. Non-Cryst. Solids 299–302 (2002)1023.

[11] J.H. Horton, C. Hardacre, C.J. Baddeley, G.D. Moggridge, R.M.Ormerod, R.M. Lambert, J. Phys.: Condensed Matter 8 (1996) 707.

[12] M. Aniya, J. Non-Cryst. Solids 198–200 (1996) 762.[13] T. Kawaguchi, J. Non-Cryst. Solids 345&346 (2004) 265.[14] Y. Yamamoto, T. Itoh, Y. Hirose, H. Hirose, J. Appl. Phys. 47 (1976)

3603.[15] J.M. Lavine, S.A. Dumford, J. Appl. Phys. 74 (1993) 5135.[16] K. Tanaka, J. Non-Cryst. Solids 137&138 (1991) 1021.[17] T. Wagner, G. Dale, P.J.S. Ewen, A.E. Owen, V. Perina, J. Appl.

Phys. 87 (2000) 7758.[18] J.T. Bloking, S. Krishnaswami, H. Jain, M. Vlcek, R.P. Vinci, Opt.

Mater. 17 (2001) 453.[19] H. Jain, S. Krishnaswami, A.C. Miller, P. Krecmer, S.R. Elliott,

M. Vlcek, J. Non-Cryst. Solids 274 (2000) 115.[20] K. Antoine, J. Li, D.A. Drabold, H. Jain, M. Vlcek, A.C. Miller,

J. Non-Cryst. Solids 326&327 (2003) 248.[21] K. Antoine, H. Jain, J. Li, D.A. Drabold, M. Vlcek, A.C. Miller,

J. Non-Cryst. Solids 349 (2004) 162.[22] A.G. Fitzgerald, C.P. McHardy, Surf. Interface Anal. 9 (1986) 334.[23] T. Ueno, A. Odajima, J. Physique – Colloque 42 (1981) 899.[24] T. Ueno, A. Odajima, Jpn. J. Appl. Phys. 21 (1982) 230.[25] S. Rajagopalan, K.S. Harshavardhan, B. Singh, K.L. Chopra,

J. Physique – Colloque 42 (1981) 911.[26] S. Maruno, J. Non-Cryst. Solids 59&60 (1983) 933.[27] T. Kawaguchi, Jpn. J. Appl. Phys. 37 (1998) 29.[28] A.V. Kolobov, G.E. Bedel’baeva, Philos. Mag. B 64 (1991) 21.