Journal of Non-Crystalline Solids 358 (2012) 860–868

Contents lists available at SciVerse ScienceDirect

Journal of Non-Crystalline Solids

j ourna l homepage: www.e lsev ie r .com/ locate / jnoncryso l

Study of As―Se―Te glasses by neutron-, X-ray diffraction and opticalspectroscopic methods

M. Fábián a,⁎, E. Sváb a, V. Pamukchieva b, A. Szekeres b, P. Petrik c, S. Vogel d, U. Ruett e

a Research Institute for Solid State Physics and Optics, H-1525 Budapest, PO Box 49, Hungaryb Institute of Solid State Physics, Bulgarian Academy of Sciences, Tzarigradsko Chaussee 72, 1784 Sofia, Bulgariac Research Institute for Technical Physics and Materials Science, Hungarian Academy of Sciences, Konkoly Thege Miklos u. 29–33, H-1121 Budapest, Hungaryd Los Alamos Natl. Lab., Lujan Neutron Scattering Ctr, Los Alamos, NM 87545, USAe Deutsches Elektronen-Synchrotron, Notkestr. 85, D-22603 Hamburg, Germany

⁎ Corresponding author. Tel.: +36 1 3922222; fax: +E-mail address: fabian@szfki.hu (M. Fábián).

0022-3093/$ – see front matter © 2012 Elsevier B.V. Alldoi:10.1016/j.jnoncrysol.2011.12.076

a b s t r a c t

a r t i c l e i n f o

Article history:Received 14 October 2011Received in revised form 16 December 2011Available online 12 January 2012

Keywords:Disordered solids;Optical properties;RMC simulation;Neutron diffraction;X-ray diffraction

The atomic structures of amorphous As40Se(60−x)Tex (x=10 and 15) and As40Se60 glasses have been investigat-ed by neutron and high energy X-ray diffraction methods. The two datasets were modeled simultaneously byreverse Monte Carlo (RMC) simulation technique. The RMC simulations revealed a glassy network built-upfrom As(Se, Te)3 pyramids in which Te atoms substitute Se atoms. The As―Se correlation function shows astrong and sharp first peak at 2.4 Å and two broad and much less intense peaks at 3.7 and 5.6 Å, related to1st, 2nd and 3rd neighbor distances of the As―Se bonds, respectively. They are an evidence for existence ofshort and medium ordering in the studied glasses. The similarity of ΘTe―As―Te and ΘSe―As―Se bond distributionssuggests that Te atoms have a similar role in the structure formation as Se atoms. The FTIR spectra analysisrevealed impurity bonds of Se―H, As―O, Se―O, and Te―O in the glasses which contributed to enhanced ab-sorption in visible spectral range. From the ellipsometric data analysis the optical constants and the energeticparameters of the studied glasses were established. The compositional variation of these parameters isexplained in terms of chemical bonds formation and change in the density of charged defects.

© 2012 Elsevier B.V. All rights reserved.

1. Introduction

Chalcogenide glasses exhibit a variety of interesting propertiesmaking them applicable in a wide range of fields. The exploitationpossibilities of chalcogenide glasses are usually treated dependingon their atomic structure and optical properties. In recent years thestructure of glassy materials and in particular the short-range orderin a number of glasses, such as Se, GeSe, As2Se3, and GeSe2, hasbeen extensively studied [1–7]. Since there is no universal approachto studying glassy structure, one can obtain a consistent picture ofthe glass structure only by combining several investigationtechniques.

As―Se and As―Se―Te glassy compositions have great potentialfor manufacturing optical fibers because of their low phonon energy,good transparency, low optical losses and good thermal and chemicalstability. The structures of As2Se3 glasses have been the subject ofsome debate for many years and concerning the ternary As―Se―Teglasses structural studies are rather scanty. The absence of

36 1 3959162.

rights reserved.

translational symmetry in these glasses suggests that the particlesare randomly arranged around each atom. While the local nearest-neighbor order in As―Se glasses is well established [4,6,7], the inter-mediate range order on a scale of several interatomic spacing is stillcontroversial. Yet, there is no answer how the glassy structurechanges with the addition of Te.

In this work we focused on the atomic-scale structural character-ization of ternary As40Se50Te10 and As40Se45Te15 and binary As40Se60chalcogenide glasses using neutron- and X-ray diffraction methods.For structure modeling the reverse Monte Carlo (RMC) simulationwas applied. Studying the structure of As40Se60 glasses by thesemethods gives a possibility to get better insight on the structuralchanges caused by the addition of Te to this composition. As far aswe know, results on the structural study of these glasses by meansof Neutron-Diffraction (ND) method are not reported yet. The possi-ble contamination of the synthesized materials during technologicalprocedures was examined by applying the Fourier Transform infra-red (FTIR) spectroscopy. The optical properties of these glasseswere studied by applying spectroscopic ellipsometry (SE) in UV–VIS–NIR spectral range of 190–1700 nm. Still, there are only a fewliterature data concerning optical studies of these glasses by ellipso-metric method and they are performed in a narrow VIS spectralrange.

-1

0

1

2

ND

Q (Å-1)

As40Se60

As40Se50Te10

As40Se45Te15

S(Q

) -1

A

0 5 10 15 20 25

0 5 10 15 20 25-1

0

1

2

XD

As40Se60

As40Se50Te10

As40Se45Te15

S(Q

) -1

Q (Å-1)

B

Fig. 1. Experimental structure factor data obtained from the neutron diffraction (ND)(A) and X-ray diffraction (XD) (B) measurements (empty squares) and from RMCmodeling (solid line) for the studied glasses.

861M. Fábián et al. / Journal of Non-Crystalline Solids 358 (2012) 860–868

2. Experimental details

2.1. Sample preparation

The glassy samples with compositions of As40Se60, As40Se50Te10and As40Se45Te15 were synthesized from 5 N purity elements by theconventional melt-quenching method. The components of a propercomposition were placed in a quartz ampoule which was evacuatedto a residual pressure of 10−3 Pa. The syntheses were performed ina rotary furnace as the ampoules were heated up to 950 °C and keptat this temperature for 24 h, rotating the furnace for homogeneousmelting. After finishing the synthesis, the ampoules were pulled outand were quenched in air.

The density of the synthesized materials was measured by apply-ing the Archimedes method using the hydrostatic weighing in xylene,the measurement procedure is written in detail elsewhere [8]. Afterthat, part of the samples were powdered for the neutron diffractionmeasurements, while the other part was cut into ~4 mm thick slicesand polished on one side for optical measurements.

2.2. Neutron and X-ray diffraction measurements

Neutron diffraction (ND) measurements were performed in a rel-atively broad momentum transfer range, combining the datameasured by the 2-axis ‘PSD’ monochromatic neutron diffractometer(λ0=1.068 Å; Q=0.45–9.8 Å−1) [9] at the 10 MW Budapest re-search reactor and by the time-of-flight ‘HIPPO’ instrument at theLANSCE pulsed neutron source Q=0.9–50 Å−1 [10].

The high-energy X-ray diffraction (XD) measurements were car-ried out at the BW5 experimental station [11] at HASYLAB, DESY.The powdered samples were filled into quartz capillary tubes of2 mm in diameter (wall thickness of ~0.02 mm). The energy of the ra-diation was 109.5 keV (λ0=0.113 Å). The XD structure factor for theAs40Se60 sample was obtained up to 19 Å−1, while for the As40Se50Te10and As40Se45Te15 samples it was up to 25 and 23 Å−1, respectively.

The structure factors, S(Q)'s were evaluated from the raw experi-mental data, obtained separately from the ND and XD measurements,using the program packages available at these two facilities. The rawdata was corrected for detector dead time, background, absorptionand variations in detector solid angle.

2.3. Reverse Monte Carlo (RMC) simulation

The diffraction experimental data was treated by the reverseMonte Carlo (RMC) simulation [12] in order to get structural informa-tion about the possible atomic configurations. As an RMC startingmodel, for each composition a disordered atomic configuration wasbuilt up with a simulation box containing 10,000 atoms. In the RMCsimulation procedure the atomic densities of 0.0355, 0.035 and0.034 atoms/cm3 and half-box lengths of 34.668, 32.931 and32.931 Å were taken for the As40Se60, As40Se50Te10 and As40Se45Te15glasses, respectively. Two types of constraints were used, namelythe minimum interatomic distances between atomic pairs (cut-offdistances) and connectivity constraints. In order to check the qualityof the experimental data, the neutron and X-ray diffraction data-setswere modeled separately and also both measurements in the RMCsimulation run were combined.

2.4. Optical measurements

The optical properties of the studied glasses were investigated indetail by Fourier Transform infrared (FTIR) spectroscopy and spectro-scopic ellipsometry (SE).

The FTIR measurements were performed using a SHIMADZU FTIRPrestije-2 apparatus working with a high energy ceramic light source(ω=350–7800 cm−1) and high signal to noise ratio of 40000:1. The

measurements were performed at room temperature in the spectralrange of 400–4000 cm−1 with a resolution of 4 cm−1.

The ellipsometric measurements were performed on a J.A. Wool-lam Co. ellipsometer in the spectral range of 190–1700 nm at anglesof light incidence 50, 60 and 70°. The measure of the goodness in a fit-ting procedure was expressed in root mean squared error (MSE)values being the sum of differences between the measured andmodel-generated data. The software used a standard iterative non-linear regression algorithm of Levenberg–Marquardt method to min-imize the MSE values (MSEb10). The values of the refractive indexand extinction coefficient were calculated with an accuracy of±0.005.

3. Results

3.1. Neutron and X-ray diffraction data analysis

The ND and XD results of the experimental structural factor S(Q)are summarized in Fig. 1A and B, respectively. At first sight, the S(Q)spectra look similar for all three compositions, but several fine detailsindicate the existing differences. The S(Q) values are obtained with agood signal-to-noise ratio up to Qmax=29 Å−1, which allows to fur-ther perform the high-resolution r-space analyses.

The constituent elements (As, Se, Te) possess rather similarneutron scattering amplitudes b (bAs=6.58 fm, bSe=7.97 fm,bTe=5.8 fm [13]) and X-ray scattering amplitudes. Consequently,the atomic correlation functions are overlapped, which makes it diffi-cult to separate the corresponding atomic positions. That is why weperformed different program simulations. In order to avoid possibleprogram memory effects of initial configurations, in the first

0

4

8

12

As-Se

g As-

Se(

r)

r (Å)

A

1 2 3 4 5 6 7 8

862 M. Fábián et al. / Journal of Non-Crystalline Solids 358 (2012) 860–868

modeling run random atomic positions were considered. Then,unconstrained runs were carried out, in which only the density, theminimal interatomic distances and experimental data were consid-ered in the simulations. In this case the partial correlation functionsof gAsTe(r) and gSeTe(r) fully overlapped with the partial correlationfunction of gAsSe(r). Therefore, a series of simulations were per-formed, in which we tried to fit the experimental structure factorusing the cut-off distances and coordination constrains, reported inour previous paper [14]. These results are considered below in details.

Using reasonable cut-off distances and connectivity constraints inthe RMC simulation procedure, excellent fitting of the experimentalS(Q) data for both the neutron and X-ray spectra was achieved. Inthe RMC simulation the cut-off distances were 2.2 Å for As―Se,2.45 Å for As―Te and Se―Te, and 3.1 Å for As―As, Se―Se andTe―Te [14]. In the first approach of the RMCmodeling we consideredonly the possible heteropolar bonds, since homopolar bonds are notlikely present. We applied connectivity constraints for As atoms sur-rounded by three Se neighbors, and for Se atoms surrounded by twoAs neighbor atoms, as reported in most of the corresponding refer-ences [6,7,15–20]. The convergence of the RMC simulation was goodand the final S(Q) matched very well the experimental structure fac-tor, as is seen in Fig. 1, where the experimental S(Q) spectra are givenwith open symbols and the calculated ones are given by solid lines.The corresponding total atomic pair correlation functions gtotal(r),calculated from the neutron diffraction measurements, are illustratedin Fig. 2 for the As40Se60 and As40Se50Te10 compositions. The reason-able agreement between the experimental data calculated from theSexp(Q) by Fourier transformation (FT) and one calculated by RMCmodeling is well seen.

From the RMC modeling the partial correlation functions (g(r)),the coordination number (CN) and bond angle (Θ) distributions

0

2

4 ND As

40Se

60 - RMC

As40

Se60

- FT

g tot

al(r

)g t

otal

(r)

r(Å)1 2 3 4 5 6

1 2 3 4 5 6

0

1

2

3ND

As40

Se50

Te10

- RMC As

40Se

50Te

10 - FT

r(Å)

A

B

Fig. 2. Total atomic distribution functions calculated from RMC (square) and for Fouriertransformation (crosses), for As40Se60 (A) and As40Se50Te10 (B) glasses.

with good reproducibility were elucidated. The results are summa-rized in Figs. 3–8.

The partial atomic correlation functions obtained from the RMCmodeling for As―Se, As―As and Se―Se network former atomicpairs are presented in Fig. 3. In general, the curves for all three com-positions are very similar. The As―Se correlation (Fig. 3A) shows astrong and sharp first peak at 2.4±0.01 Å and two broad and muchless intense peaks at 3.7±0.1 and 5.6±0.1 Å, related to 2nd neighbordistance and 3rd neighbor distance of the As―Se bonds, respectively.The characteristic peaks for the possible As―As and Se―Se homopo-lar bonds (Fig. 3B and C, respectively) appear at 3.7±0.1 Å in the bi-nary As40Se60 glass, while these peaks are slightly shifted to 3.6±0.1 Å in the ternary As40Se50Te10 and As40Se45Te15 glasses.

g As-

As(

r)g S

e-S

e(r)

0

1

2

3

As-As

r (Å)

B

1 2 3 4 5 6 7 8

1 2 3 4 5 6 7 8

0

1

2

3Se-Se

r (Å)

C

Fig. 3. Partial atomic correlation functions obtained from the RMC modeling of As―Se(A), As―As (B) and Se―Se (C) atom pairs for As40Se60 (full squares), As40Se50Te10(empty circles) and As40Se45Te15 (crosses) glasses. The first peak position is with an ac-curacy of ±0.01 Å, while the second and third peaks are with an accuracy of ±0.1 Å.

0

2

4

6

As-Teg A

s-T

e(r)

g Se-

Te(

r)g T

e-T

e(r)

r (Å)

A

0

1

2

3

4Se-Te

r (Å)

B

1 2 3 4 5 6 7 8

1 2 3 4 5 6 7 8

1 2 3 4 5 6 7 8

0

1

2

3

4

5

Te-Te

r (Å)

C

Fig. 4. Partial atomic correlation functions, obtained from the RMC modeling of As―Te(A), Se―Te (B) and Te―Te (C) atomic pairs in As40Se50Te10 (empty circles) and As40-Se45Te15 (crosses) glasses. The peaks position is with an accuracy of ±0.1 Å.

0

1000

2000

3000 As-Se

Num

ber

of a

tom

s

Coordination number

A

1 2 3 4

1 2 3 40

1000

2000

3000

4000

Coordination number

Se-As

Num

ber

of a

tom

s

B

Fig. 5. Coordination number distribution for As40Se60 (black), As40Se50Te10 (gray) andAs40Se45Te15 (crosses) glasses obtained from the RMC modeling of As―Se (A) andSe―As (B) atomic pairs.

863M. Fábián et al. / Journal of Non-Crystalline Solids 358 (2012) 860–868

The partial atomic correlation functions (g(r)) of the possible Te-containing chemical bonds are presented in Fig. 4. For the As40Se50-Te10 composition, the first neighbor peak appears at 2.5±0.1 Å,while the second peak appears at 3.5±0.1 Å in the heteropolarAs―Te and Se―Te pair correlation functions. With increasing the Tecontent in As40Se45Te15 glasses, these peaks shift slightly to larger dis-tances, namely to 2.55±0.1 Å and 3.7±0.1 Å, respectively. In the ho-mopolar Te―Te pair correlation functions (Fig. 4C), a well-pronounced peak appears at 3.55±0.1 Å, which also shifts to largerdistance of 3.7±0.1 Å by increasing the Te content in the As40Se45-Te15 composition.

In Fig. 5 the coordination number (CN) distributions for the basicAs―Se bonds in the compositions are presented. The average coordi-nation numbers CNij i.e. the average number of j atoms around i atom,

are CNAsSe=2.8±0.1, 2.4±0.1 and 2.2±0.1 and CNSeAs=1.8±0.05,1.9±0.05 and 1.9±0.05 for theAs40Se60, As40Se50Te10 andAs40Se45Te15glasses, respectively.

For the As40Se50Te10 and As40Se45Te15 compositions, for whichthe first peak of gAsTe(r) (Fig. 4) is located at 2.5 and 2.55 Å, re-spectively, the corresponding coordination number (CN) distri-butions of the Te atomic environments are presented in Fig. 6.They are CNAs―Te=0.5±0.05 and 0.44±0.05, respectively, andCNTe―As=1.91±0.1 and 1.65±0.1, respectively. Without any co-ordination constraint applied for Te atom, we obtained a coordi-nation number very close to 2.0 for the Te―As bond and acoordination number near to 0.5 for the As―Te bond.

Further we calculated the angular distribution Θ (cos Θ) of thebonds between first neighbor atoms. The Θ(cos Θ) functions of thebasic bonds for the three glassy compositions are shown in Figs. 7and 8. All these functions were calculated considering the positionof the first minimum after the peak of the first shell withrmax≈4.0 Å, obtained from the final atomic configuration. In mostcases, they are very similar to each other and the distribution peaksappear almost at the same positions.

The three-atom-bond angle distributions of As―Se―As andSe―As―Se bonds, obtained from RMC modeling of the studied com-positions are presented in Fig. 7. More stable and reproducible valuesare obtained for ΘAs―Se―As=97±2° and ΘSe―As―Se=99–92±2°,which are in good agreement with our previous results [14] andthose reported in [4].

The bond angle distributions of ΘSe―Te―Se(cosΘ), ΘTe―As―Te(cosΘ),ΘTe―As―Se(cosΘ) and ΘTe―Se―As(cos Θ) for the two ternary glasses aregiven in Fig. 8. As is seen, an average ΘSe―Te―Se bond angle of92°±3°/56±3° (Fig. 8a) and an average ΘTe―As―Te bond angle of

0

200

400

600

800

1000

Coordination number

Te-As

Num

ber

of a

tom

sA

0 1 2 3

0 1 2 30

800

1600

2400 As-Te

Num

ber

of a

tom

s

Coordination number

B

Fig. 6. Coordination number distribution for As40Se50Te10 (gray) and As40Se45Te15(crosses) glasses obtained from the RMC modeling of Te―As (A) and As―Te (B) atom-ic pairs.

0,0

0,5

1,0

1,5

As-Se-As

97o 2o

Pro

babi

lity

(a.u

.)

cos θ

+-

A

-1,0 -0,5 0,0 0,5 1,0

-1,0 -0,5 0,0 0,5 1,0

0,0

0,5

1,0

1,5

2,0

Se-As-Se

Pro

babi

lity

(a.u

.)

cos θ

99o 2o+-

92o 2o+-

B

Fig. 7. Three-atom-bond angle distribution of As―Se―As (A) and Se―As―Se(B) bonds, obtained from RMC modeling of As40Se60 (full squares), As40Se50Te10(open circles) and As40Se45Te15 (crosses) glasses.

864 M. Fábián et al. / Journal of Non-Crystalline Solids 358 (2012) 860–868

92±3° (Fig. 8B) are determined. Further, the calculations made for thethree-atom distributions of ΘTe―As―Se (Fig. 8C) indicated two maximaat 96°±2° and 60±2°. For the Te―Se―As bonds, the ΘTe―Se―As angledistributions (Fig. 8D) also indicate well-defined and similar structureswith two maxima at 88±2°(92±2°) and 60±2°. The only differ-ence is that the angle distribution for As40Se50Te10 compositionmoves from 88° to a larger value of 92° by increasing the Te contentin the As40Se45Te15 composition.

3.2. FTIR reflectance spectra

The IR reflectance spectra of the studied glasses are given in Fig. 9.The detected bands were identified on the basis of the reported datain [21–25] and their positions were summarized in Table 1.

The broad band, centered around 665 cm−1 and with multitudeweak superimposed peaks and the weak shoulder around 695 cm−1

can be assigned to the symmetric stretching vibrational modes of ox-ides bonds, such as Te―O [25], As―O [24] and Se―O [23] bonds. Thebroad band centered around 785 cm−1 can be also connected withthe asymmetric stretching modes of As―O and Se―O bonds [23,24].However, this band is absent from the spectrum of As40Se50Te10composition.

The well-pronounced double band centered at around 2337 and2360 cm−1 can be assigned to the stretching mode of Se―H chemicalbonds [21].

There are two regions in the IR spectra of the studied glasses,where noisy signal appears (Fig. 9). They can be attributed to thepresence of water related bonds. In the literature characteristic vibra-tional bands of O―H hydroxyl groups and molecular H2O are situated

in the 3500–3700 cm−1 range and around 1600 cm−1, respectively[26].

3.3. Ellipsometric data analysis

In Fig. 10 the dispersion curves of the refractive index n and ex-tinction coefficient k of the synthesized glasses are presented. Bothoptical constants tend to increase in the whole studied spectral regionwhen Se is substituted by Te. The glasses have a good transparency inthe 800–1700 nm spectral region, where the k values are low.

The refractive index dispersion curves were analyzed using theconcept of single-oscillator [27], considering the single oscillator atinfinitive wavelength λ0. It includes the high frequency region,where a simple classical dispersion equation (n02−1)/(n2−1)=1−(λ0/λ)2 can be used, where n0 is the refractive index of an empty lat-tice at infinite wavelength (at zero energy). The quantity n0 also de-termines the high frequency dielectric constant ε∞=n0

2, in otherwords the high-frequency component of the relative permittivity ε∞for a given material. The wavelength λ0 is the average oscillatorwavelength and it is related to the energy at which direct interbandelectron transitions start. It is equal to λ0=hc/E0, where E0 is the os-cillator energy and represents the average bandgap energy. By plot-ting 1/(n2−1) versus 1/λ2 (Fig. 11A) these parameters weredetermined, as the intercept of the approximated line with the verti-cal axis at zero 1/λ2 gave n0 and, the slope of this line — λ0. Thesevalues are summarized in Table 2.

The same refractive index data were further analyzed below theinterband absorption edge using the equation n2−1=E0Ed/(E20−E2) [27]. By plotting 1/(n2−1) versus E2 (Fig. 11B) the oscillator ener-getic parameters E0 and Ed, where E0 is the oscillator energy and Ed isthe oscillator strength, were calculated. Their values are also given in

0

1

2

3

Se-Te-Se

Pro

babi

lity

(a.u

.)

cos θ

92o 3o+-

56o 3o+-

0,0

0,5

1,0

1,5

2,0

Te-As-Te

Pro

babi

lity

(a.u

.)

cos θ

92o 3o+-

0

1

2Te-As-Se

Pro

babi

lity

(a.u

.)

cos θ

96o 2o+-

60o 2o+-

-1,0 -0,5 0,0 0,5 1,0

-1,0 -0,5 0,0 0,5 1,0

-1,0 -0,5 0,0 0,5 1,0

-1,0 -0,5 0,0 0,5 1,0

0,0

0,5

1,0

1,5

Te-Se-As

Pro

babi

lity

(a.u

.)

cos θ

60o 2o+-

88o 2o+-

92o 2o+-

A B

DC

Fig. 8. Three-atom-bond angle distribution of Se―Te―Se (A), Te―As―Te (B), Te―As―Se (C) and Te―Se―As (D) bonds from RMC modeling for As40Se50Te10 (open circles) andAs40Se45Te15 (crosses) glasses.

865M. Fábián et al. / Journal of Non-Crystalline Solids 358 (2012) 860–868

Table 2. As can be seen, there is a good coincidence between the cor-responding parameters as fluctuation in the values is within the ap-proximation errors.

The absorption coefficient α=4πk/λ was calculated from the kvalues given in Fig. 10. Using the Tauc's expression [28] and assumingindirect type electron transitions, the energy band gap Eog was deter-mined from the plots (αE)1/2 versus E, presented in Fig. 12. Consider-ably broad absorption tails are observed, which are related to tails oflocalized states, originated from structural defects and energeticallylying within the bandgap of glasses. By extrapolating the linear partof the curves toward zero absorption, the interception with the pho-ton energy axis provides the Eog value. For the As40Se60 compositionEog is 1.73±0.1 eV, while for the As40Se50Te10 Eog is 1.51±0.1 eV.

400 600 800 1000 1200 140050

60

70

80

90

1400 2100 2800 3500

50

60

70

80

90 As40

Se60

As40

Se50

Te10

Ref

lect

ance

(%

)

Wavenumber (cm-1)

As40

Se60

As40

Se50

Te10

Ref

lect

ance

(%

)

Wavenumber (cm-1)

Fig. 9. FTIR reflectance spectra of As40Se60 and As40Se50Te10 glasses.

4. Discussion

The ND and XD patterns of the studied compositions showed thatthe specimens were fully amorphous. The similarity of the partialatomic correlation functions in Fig. 3 suggests that substituting apart of Se by Te atoms does not change the basic network structurewithin the limit of experimental and modeling errors. The presenceof three well-pronounced peaks in Fig. 3A is an evidence for structuralordering. The rather sharp first peak at 2.4 Å is related to first neigh-bor distance and shows the existence of short rage ordering in theglassy structure. The second peak at 3.7 Å and the third peak at5.6 Å are the second and third neighbor distances of As―Se bondsand they are an evidence for the presence of medium range orderingin both the binary and ternary glassy structures. The position of thesepeaks does not change with changing the composition, i.e. in the ter-nary As―Se―Te system the As―Se bond distance remains the sameas in the binary As―Se system, which points out that the additionof Te to As―Se network has negligible effect on As―Se bonds.

In the ternary compositions, for all three As―Te, Se―Te andTe―Te partial atomic pair correlations (Fig. 4) the highly intensive

Table 1Vibrational bands in the FTIR spectra of As40Se60 and As40Se50Te10 glasses.

Vibrational modes As―Se Se―H As―O Se―O Te―O O―H H2O

Band position(±4 cm−1)

As40Se60 composition480 2338 796 778 ~3600 ~1600

785 785665 665616

As40Se50Te10 composition480 2357 665 665 665 ~3600 ~1600

616 695

200 400 600 800 1000 1200 1400 16000

1

2

3

0

1

2

3

As40

Se50

Te10

As40

Se60

Ref

ract

ive

inde

x

Wavelength (nm)

Ext

inct

ion

coef

ficie

nt

Fig. 10. Dispersion spectra of the refractive index (n) and extinction coefficient (k) ofAs40Se60 and As40Se50Te10 glasses. The n and k values were determined with an accu-racy of ±0.005.

Table 2Parameters derived from the dispersion spectra in Fig. 10 for the studied glasses.

Single-oscillatorapproximation

Opticalconstants

Glass composition

As2Se3 As40Se50Te10

(n02−1)/(n2−1)=1−(λ0/λ)2 n0 2.65±0.01 2.80±0.01ε∞=n0

2 7.02 7.84λ0 339±5 nm 395±5 nmE0=hc/λ0 (eV) 3.66 3.14

n2−1=E0Ed/(E20−E2) n0 2.64±0.01 2.8±0.01E0 (eV) 3.67±0.05 3.15±0.05Ed (eV) 22.09±0.05 21.6±0.05

Tauc plot (αE)1/2 vs. E Eog (eV) 1.73±0.1 1.51±0.1

866 M. Fábián et al. / Journal of Non-Crystalline Solids 358 (2012) 860–868

peak at 3.5–3.7 Å is an evidence for second neighbor distance of thesebonds. This suggests the presence of a well-definedmedium range or-dering in the glassy structure of the studied ternary systems. The ob-served increase of the first and second neighbor distances in theheteropolar As―Te and Se―Te bonds with the increasing Te content(Fig. 5) proves the possible replacement of Se atoms by Te atoms dur-ing the incorporation of Te atoms.

The RMC modeling revealed that in the binary glass nearly 92% ofthe bonds are heteropolar. The average coordination numbers for thebasic As―Se bonds (Fig. 5) are somewhat less than the expected 3 forAs and 2 for Se atomic neighbors, which may be a consequence of a

0,04

0,08

0,12

0,16

λ-2 (μm)-2

(n2 -

1)-1

(n2 -

1)-1

As40Se50Te10

As40Se60

0 1 2 3

0 1 2 3 40,04

0,08

0,12

0,16

E2 (eV)2

As40

Se50

Te10

As40

Se60

A

B

Fig. 11. Plots of 1/(n2−1) versus 1/λ2 (A) and 1/(n2−1) versus E2 (B) for the As40Se60and As40Se50Te10 glasses.

slight distortion from the ideal AsSe3 trigonal pyramid units. The ad-dition of Te to the As―Se glasses leads to the reduction of CNAsSe to2.4 and 2.2 for As40Se50Te10 and As40Se45Te15 glasses respectively,while CNSeAs increase slightly to 1.9 for both ternary compositions.Therefore, the main structure units in the studied glasses can be con-sidered as [AsSe3] pyramids connecting with homopolar Se―Se andAs―As bonds, which also participate in the network formation.With the increase of the Te content and decrease of the Se content ac-cordingly, the glassy network becomes more distorted, as it is seen inFig. 5 from the systematic changes of the coordination numberdistributions.

The obtained coordination number of Te related bonds, i. e.CNAs―Te and CNTe―As (Fig. 6) are well correlated with thosegiven in the literature. Since the Te―As (Fig. 6A) and Se―As(Fig. 5B) coordination number distributions are very similar, wesuggest that the Se and Te atoms surrounding environments arealso similar. From these results it can be concluded that the in-troduction of Te atoms into the As―Se glass does not destroythe covalent network formed by the threefold coordinated Asand twofold coordinated Se atoms. The Te atoms replacing Seatoms behave similarly as Se atoms. The shape and very closevalues of the bond angle distributions obtained from RMC calcu-lation for ΘSe―Te―Se, ΘTe―As―Te, ΘTe―As―Se and ΘTe―Se―As (Fig. 8)are also evidence that substitution of Se by Te does not modifythe basic network structure and thus we have a characteristicdistribution for both ternary compositions. The functions aresimilar to each other as the value of ΘTe―As―Te alters only slightlyby increasing the Te content. The latter can be explained by thefact that the difference of the Te content in the studied glassesis not large and that we increased the Te content only from 10to 15 at.%.

In the case of ΘTe―As―Te bond distribution (Fig. 8B), the obtainedvalues are very similar to those obtained for the ΘSe―As―Se bond an-gles (Fig. 7B). This supports the suggestion that the Te atom couldhave a similar role in the structure formation as Se atom. The bondangle distributions for ΘTe―As―Se and ΘTe―Se―As shows a possible

0 1 2 3 4 5 60

400

800

1200

1600

2000

2400

As40

Se60

As40

Se50

Te10

(αΕ)

1/2

(cm

-1eV

)1/2

E (eV)

Fig. 12. Plots (αE)1/2 versus photon energy E for the As40Se60 and As40Se50Te10 glasses.

867M. Fábián et al. / Journal of Non-Crystalline Solids 358 (2012) 860–868

polyhedron configuration, for which a relatively narrow peak at 60°(close to the ideal tetrahedral units) and an asymmetric broadenedpeak are observed in the As40Se50Te10 glass with lower Teconcentration.

In general, strong bands related to vibrational modes of basicchemical bonds in the studied glasses are below the measurementlimit of 400 cm−1 and, therefore, they cannot be detected. However,the observed broad band with multiple very weak features in the IRspectra (Fig. 9) peaking around 480 cm−1 can be assigned to thetwo-phonon modes of the As―Se bonds [22]. Hence, the dominantfeatures in the recorded FTIR spectra are vibrational bands relatedto extrinsic impurities being present in the synthesized glasses.They could be assigned to oxide and hydride impurities, which aremost probably introduced by the initial substances and during thesynthesis of glass-forming melts. Moreover, the appearance of noisysignals in the spectral regions around 1600 cm−1 and in the3500–3700 cm−1 range is an evidence that during technological pro-cesses and IR measurements some amount of moisture from the envi-ronment is absorbed onto the samples surface.

The refractive index dependences follow the normal dispersionlaw (Fig. 10). The estimated n values of As40Se60 composition areclose to those reported in [29]. For the As40Se50Te10 composition,higher index values are obtained, which can be explained by the en-hanced polarizability of the As40Se50Te10 glass because of larger po-larizability of Te atom due to its larger atomic radius (rAs=1.33 Å,rSe=1.22 Å and rTe=1.42 Å). Although the glasses are transparentin the studied spectral region, the estimated k values for theAs40Se60 composition (Fig. 10) are somewhat higher than thosegiven in the literature [30]. Most probably, the reason for that is thedistortion of the main structural units in the glassy network and thepresence of defective homopolar bonds, as was detected by the NDmeasurements on one side. On the other side, extrinsic impuritiespresent in the glasses, as detected by IR spectroscopy, contribute tostronger light absorption.

In comparison to As40Se60 composition, in the As40Se50Te10 com-position the absorption edge shifts toward longer wavelengths indi-cating for smaller optical bandgap energy. The obtained values ofEog=1.73 eV for the As40Se60 composition and Eog=1.51 eV for theAs40Se50Te10 composition are close to those given in the literature[29].

It is empirically established that the optical bandgap of a chalco-genide material is approximately twice smaller than its oscillator en-ergy (E0~2Eog) [31]. As is seen from Table 2, this relation is alsoobserved for the studied compositions, as the ratio E0/Eog is veryclose to 2. Since the oscillator energy represents the average bandgapenergy, its value follows the same compositional dependence as theone for the optical bandgap, namely decreasing by substituting Sefor Te in the ternary composition. Both E0 and Eog quantities are relat-ed to the average molar bond energy of different bonds present in thematerial and their value depends on the cohesive energy in the glassysystem. The latter represents the stabilization energy of an infinitelylarge cluster of the material per atom and it is a sum of energies ofchemical bonds, expected in the system.

The results obtained from the RMC modeling showed that the Teatoms can be integrated in the As40Se60 glassy structure withoutchanging the basic network. However, the replacement of Se atomsby Te atoms and formation of As―Te bonds lead to the reduction ofcohesive energy of ternary glasses because the bond energy ofAs―Te (136,86 kJ/mol) is smaller than that of As―Se (174.35 kJ/mol). This results in a decrease of bandgap energy in ternary glasses,as was observed. Although Te atom have similar surrounding atomicenvironments as Se atoms have, as followed from the similarity of co-ordination number distributions of Te―As and Se―As bonds (Figs. 5Band 6A), incorporation of Te into As―Se glass causes some distortionof the glassy covalent network. The reason is that Te atoms have atendency to hybridization of 5d, 6s and 6p electron orbitals in

chemical bonding, which leads to a compositional disordering. Atthe same time, Te atoms create charged diamagnetic coordination de-fects, i.e. one D+ and two D− centers, which highly increase the den-sity of D− states in the system [32,33]. All these contribute to anincrease in both structural and compositional disorders and thuscausing further reduction of the optical bandgap energy Eog and aver-age bandgap energy E0 values.

5. Conclusions

The performed neutron and X-ray diffraction measurements andreverse Monte Carlo modeling of As40Se50Te10, As40Se45Te15 andAs40Se60, glasses have revealed that addition of Te to the binaryglass does not change the basic glassy network structure. The firstneighbor distance of As―Se bond is determined and it is equal to2.4 Å. The average coordination numbers of As and Se atoms areestablished as As atoms are surrounded accordingly by 2.8 Se atomsand Se atoms are surrounded with 1.8 As atoms. The nearestTe―As bond length in the As40Se50Te10 glass is shorter than that inthe As40Se45Te15 glass. The simulations have shown a glass networkbuilding up by As(Se, Te)3 pyramids in which the Te atoms can besubstitute for the Se atoms.

The FTIR spectra analysis has detected vibrational bands related toimpurity bonds of Se―H, As―O, Se―O and Te―O in the glasses con-tributing to enhanced absorption in VIS–NIR spectral range. Theymight be created during technological procedures for synthesis and/or measurements.

By analyzing the ellipsometric data in a wide spectral range of190–1700 nm, the optical constants (n, k, α, n0, λ0, ε∞) and the ener-getic parameters (E0, Eog, Ed) of the studied As40Se60 and As40Se50Te10glasses are established. The compositional variation of these parame-ters is explained in terms of chemical bonds formation and change inthe density of charged defects.

Acknowledgements

This research project has been supported by the EC under the FP7Grant Agreement N226507-NMI3 and No. 226716. This study hasbenefited from the use of HIPPO at the LANSCE, by US-DOE contractW-7405-ENG-36.

References

[1] B. Bureau, X.H. Zhang, F. Smektala, J.-L. Adam, J. Troles, H.-L. Ma, C. Boussard-Ple`del, J. Lucas, P. Lucas, D. Le Coq, M.R. Riley, J.H. Simmons, J. Non-Cryst. Solids345&346 (2004) 276.

[2] S.A. Dembovsky, J. Non-Cryst. Solids 353 (2007) 2944.[3] Philip S. Salmon, J. Non-Cryst. Solids 353 (2007) 2959.[4] Y. Iwadatea, T. Hattori, S. Nishiyama, K. Fukushima, Y. Mochizuki, M. Misawa, T.

Fukunaga, J. Phys. Chem. Solids 60 (1999) 1447.[5] E. Mytilineou, A. Kolobov, Short-, Medium-, and Long-Order Structural Transfor-

mations in Amorphous Semiconductors, Photo-Induced Metastability in Amor-phous Semiconductors, Wiley, New York, 2003, p. 43.

[6] K. Shirao, T. Hattori, S. Nishiyama, K. Fukushima, Y. Iwadate, M. Misawa, T.Fukunaga, T. Nakazawa, K. Noda, J. Phys. Chem. Solids 60 (1999) 1453.

[7] B. Bureau, J. Troles, M. Floch, F. Smektala, J. Lucas, J. Non-Cryst. Solids 326–327(2003) 58.

[8] V. Pamukchieva, A. Szekeres, K. Todorova, M. Fabian, E. Svab, E. Svab, E. Zs. Revay,L. Szentmiklosi, J. Non-Cryst. Solids 355 (2009) 2485.

[9] E. Sváb, Gy. Mészáros, F. Deák, Mater. Sci. Forum 228 (1996) 247 http://www.bnc.hu/.

[10] Th. Proffen, S.J.L. Billinge, T. Egami, D. Louca, Z. Kristallogr. 218 (2003) 132.[11] H. Poulsen, J. Neuefeind, H.B. Neumann, J.R. Schneider, M.D. Zeidler, J. Non-Cryst.

Solids 188 (1995) 63.[12] R.L. McGreevy, L. Pusztai, Mol. Simul. 1 (1988) 359.[13] A.C. Hannon, ISIS Disordered Materials Database, 2006, 2006 http://www.isis.rl.

ac.uk/disordered.[14] M. Fabian, E. Svab, V. Pamukchieva, A. Szekeres, S. Vogel, U. Ruett, J. Phys. Conf.

Ser. 253 (2010) 012053.[15] A. Eungho, G.A. Williams, P.C. Taylor, D.G. Georgiev, P. Boolchand, B.E. Schwickert,

R.L. Cappelletti, J. Non-Cryst. Solids 299–302 (2002) 958.[16] B. Bureau, J. Troles, M. Floch, F. Smektala, G. Silly, J. Lucas, Solid State Sci. 5 (2003)

2194.

868 M. Fábián et al. / Journal of Non-Crystalline Solids 358 (2012) 860–868

[17] J.C. Mauro, A.K. Varshneya, J. Non-Cryst. Solids 353 (2007) 1226.[18] I. Kaban, W. Hoyer, T. Petkova, P. Petkov, B. Beuneu, A. Schöps, M.A. Webb, J.

Ovonic Res. 3 (2007) 67.[19] M.R. Balboul, S.S. Fouad, S.A. Fayek, M.S. El-Bana, J. Alloys Cmpd. 460 (2008) 570.[20] P. Jóvári, B. Bureau, I. Kabanc, V. Nazabal, B. Beuneu, U. Rütt, J. Alloys Cmpd. 488

(2009) 39.[21] S.P. Love, A.J. Sievers, B.L. Halfpap, S.M. Lindsay, Phys. Rev. Lett. 65 (1990) 1792.[22] C.T. Moynihan, P.B. Macedo, M.S. Maklad, R.K. Mohr, R.E. Howard, J. Non-Cryst.

Solids 17 (1975) 369.[23] A. Andriesh, M. Iovu, O. Shpotiuka, I. Culeac, J. Optoelectron. Adv. Mater. 11

(2009) 2172.[24] G.E. Snopatin, V.S. Shiryaev, V.G. Plotnichenko, E.M. Dianov, M.F. Churbanov,

Inorg. Mater. 45 (2009) 1439.

[25] D. Munoz-Martín, M.A. Villegas, J. Gonzalo, J.M. Fernandez-Navarro, J. Eur. Ceram.Soc. 29 (2009) 2903.

[26] T.S. Kavetskyy, A.P. Kovalskiy, V. Pamukchieva, O.I. Shpotyuk, Infrared Phys. Tech-nol. 41 (2000) 41.

[27] S.H. Wemple, M. DiDomenico, Phys. Rev. B 3 (1971) 1338.[28] J. Tauc, Amorphous and Liquid Semiconductors, Plenum Press, London, 1974.[29] J. Orava, J. Šik, T. Wágner, M. Frumar, J. Appl. Phys. 103 (2008) 083512.[30] D.C. Sati, R. Kumar, R.M. Mehra, Turk. J. Phys. 30 (2006) 519.[31] R. Prieto-Alcón, J.M. González-Leal, R. Jiménez-Garay, E. Márquez, J. Optoelectron.

Adv. Mater. 3 (2001) 287.[32] S.A. Fayek, M.R. Balboul, K.H. Marzouk, Thin Solid Films 515 (2007) 7281.[33] N.F. Mott, E.A. Davis, Electronic Processes in Non-Crystalline Materials, Clarendon

Press, Oxford, 1971.