Effect of rare-earth doping on the free-volume structure of Ga-modified Te₂₀As₃₀Se₅₀ glass

Abstract

By exploring the positron–electron annihilation technique in positron lifetime measuring mode, it is shown that principal rare-earth (RE) induced structural reconfiguration in Ga-codoped TAS-235 glass (that is a glassy Te₂₀As₂₉Ga₁Se₅₀ alloy) is related to occupation of intrinsic free-volume voids by embedded RE ions tightly connected with Ga-based tetrahedrons via strong covalent RE-Se/Te–Ga links. A gradual decrease in the intensity of the second component of two-term decomposed lifetime spectra of annihilating positrons accompanied with a detectable increase in the defect-related positron lifetime (thus inducing essentially a depressed rate in positron trapping) is evidenced by the example of Pr³⁺ ions added homogeneously to Te₂₀As₂₉Ga₁Se₅₀ glass in the amount of 500 ppmw. Observed changes in positron lifetime spectra are explained in terms of the competitive contribution of different occupancy positions in Ga-codoped glass available for RE ions and trapped positrons.

---

Introduction

Rare-earth (RE) doping in IR-transmitting chalcogenide glasses (ChG) is of high importance as a promising technological resolution to produce prospective mid-IR laser sources for compact photonics.^1–6 Therefore, the development of a highly reliable structure-sensitive tool for identification and characterization of RE doping has attracted great attention in the photonics research community dealing with the implementation of new functional media.

In contrast to oxide glasses, where a large variety of experimental techniques can be employed to study RE doping (such as nuclear magnetic and electron paramagnetic resonances, neutron diffraction, fluorescence line-narrowing and decay analysis),^7,8 the structural probes to be successfully applied to ChG are very restricted. In order to ensure optimum low phonon energy environment for positively charged RE-ions, the ChG should be additionally modified by local negatively charged sites, which act like cation vacancies in crystals providing a total electrical charge compensation.⁸ Because of anomalously high electronegativity proper to oxygen O atom,⁹ an intrinsic atomic arrangement in oxide glasses can be adapted respectively to create own electrical charge misbalance needed to accommodate homogeneously the embedded RE ions without their clustering (and thus to overcome a parasitic concentration quenching of fluorescence).^6–8 In ChG, which are characterized by close electronegativities for constituting atoms, the RE-incorporation induced structural reconfiguration cannot be simply over-balanced by intrinsic covalent-bonded atomic environment obeying full saturation in respect to the Mott's 8-N rule.⁹ Charge-compensation equilibrium in such a case is achieved only extrinsically, e.g. by additional sites randomly distributed in a glass matrix possessing locally uncompensated excess of negative electrical charge. In this respect, the ChG preliminary doped with Ga or In seem most efficient,^6,10–15 where charge compensation for incorporated RE ions is satisfied due to local over-coordinated environment around these atoms whatever their main valence state. Under a given rate of these codopants, which do not disturb the glass-forming ability of the hosting ChG, the RE ions can reside homogeneously without forming tightly bounded atomic clusters.⁸ Thus, the most efficient structural tools to probe successful RE doping in ChG should be sensitive to hierarchical atomistic transformations squeezing (Ga,In)-codoped and RE-doped sites.

That is why the phenomenon of positron–electron annihilation in positron lifetime measuring mode, e.g. positron annihilation lifetime (PAL) spectroscopy known as high-informative free-volume probing method suitable for different solids despite their structural nature,^16,17 is expected to be useful to characterize RE-doping in ChG. Indeed, the positron (the positive antiparticle of electron) can be imagined as highly electropositive probe in its interaction with matter, whatever the nature of the interaction (trapping in extended defects or due to decaying of bound positron–electron states known as positronium Ps atoms in structurally-intrinsic voids).^16,17 Under condition of atomic-accessible free volume in Ga/In-modified ChG, the negatively charged sites related to incorporated codopants are most preferential occupancy positions for both positrons and RE-ions. So competitive contribution from corresponding annihilation paths parameterized through histogram of elementary positron–electron annihilation events forming PAL spectrum, is expected as numerical measure of RE doping.

In this research, the method of PAL spectroscopy will be first employed to study structural changes in Ga-modified TAS-235 glass (the known glassy-like Te₂₀As₃₀Se₅₀ alloy widely used in IR chalcogenide photonics) caused by doping with Pr³⁺ ions.^18–21

Experimental

The studied ChG alloys of parent Te₂₀As₃₀Se₅₀ (TAS-235), Ga-codoped Te₂₀As₂₉Ga₁Se₅₀ and this glass further doped with 500 ppmw of Pr³⁺ ions were prepared from high-purity elemental precursors, e.g. Ga (7N), As (5N), Se (5N), Te (6N) and Pr₂Se₃ (3N), the ingredients being specially purified by distillation with low evaporation rate to remove impurities such as O, C, H₂O, and SiO₂. Appropriate amounts of ingredients with total weight close to 30 g were put into silica tube of 10 mm diameter. Then, the ampoules were sealed under a vacuum and heated to 650 °C with 2 °C min⁻¹ rate in a rocking furnace for 10 h with further quenching in water from 500 °C. To remove mechanical strains appeared during rapid quenching, the alloys were annealed during 6 h at temperature of 10 °C less than glass transition (∼120 °C). The obtained rods were cut into ∼1.5 mm thick disks and polished. A more detailed description of samples preparation can find elsewhere.^20–22 The glassy state and high purity of the samples prepared were confirmed by XRD measurements showing wide-stretched halos typical for amorphous state and absence of any impurity-related signatures in the IR absorption spectra.²⁰

The PAL measurements were performed using a fast–fast coincidence system of 230 ps resolution (the full width at half maximum of single Gaussian determined by measuring ⁶⁰Co isotope) based on two Photonis XP2020/Q photomultiplier tubes coupled to BaF₂ scintillator 25.4A10/2M-Q-BaF-X-N detectors (Scionix, Bunnik, Holland) and ORTEC® electronics (ORTEC, Oak Ridge, TN, USA). The reliable PAL spectra were detected in a normal-measurement statistics (∼1 M coincidences) under stabilized temperature (22 °C) and relative humidity (35%). The channel width of 6.15 ps allows a total number of channels to be 8000. The radioactive ²²Na isotope of relatively low ∼50 kBq activity prepared from aqueous solution of ²²NaCl wrapped by Kapton® foil (DuPont™, Circleville, OH, USA) of 12 μm thickness was used as positron source sandwiched between two identical tested samples.

The raw PAL spectra were processed with LT 9.0 program.²³ Accepting unchanged contribution intensity from a source with two inputs (with 372 ps and ∼2 ns lifetimes), these spectra were adequately decomposed into two components with τ_1,2 lifetimes and normalized I_1,2 intensities (I₁ + I₂ = 1). Under above spectrometer resolution, this allows an error-bar for such arranged measuring protocol not worse than ±0.005 ns in lifetimes and ±0.01 in intensities. Introducing third component in the envelope of fitting curves did not improve goodness of fitting significantly. So Ps formation is not proper for studied ChG in full agreement with previous results.²¹ The positron trapping modes, e.g. average positron lifetime τ_av., lifetime in defect-free bulk τ_b, trapping rate in defects κ_d and fraction of trapped positrons η were calculated exploring a formalism of known two-state positron trapping model.^{16,17,21,24–26} In addition, the (τ₂ − τ_b) difference was accepted as a size measure for extended free-volume defects where positrons were trapped, as well as the τ₂/τ_b ratio was taken as direct signature of nature of these trapping defects in terms of equivalent number of monovacancies.¹⁶

Results and discussion

The PAL characteristics of pure TAS-235 (Te₂₀As₃₀Se₅₀) glass and Ga-modified partially-crystallized Te₂₀As_30−xGa_xSe₅₀ alloys (x = 2, 5) were preliminary studied in previous work.²¹ The changes observed in positron trapping modes under crystallization were ascribed to appearance of cubic Ga₂Se₃ phase, the explanation being given in terms of agglomeration of intrinsic free-volume voids under the same chemical environment in glass and partially crystallized states. It was also shown that TAS-235 glass doped with 1 at% of Ga added instead of As (within Te₂₀As_30−xGa_xSe₅₀ cut-section) is not subjected to any devitrification influences, forming suitable host matrix for further RE-doping.^20,22 In this research, the Te₂₀As₂₉Ga₁Se₅₀ glass was used as parent matrix for doping with 500 ppmw of Pr³⁺ ions.

The PAL spectra of both parent and Pr³⁺-doped Te₂₀As₂₉Ga₁Se₅₀ glass reconstructed from x²-component fitting at the general background of standard source contribution are respectively depicted on Fig. 1a and b. The limited values of statistical scatter of variance tightly grouped around 0-axis testify that PAL measurements are adequately described within this fitting procedure. So the decaying behavior of PAL spectra on Fig. 1 can be reflected by sum of two exponents with different time constants inversed to positron lifetimes.¹⁶ The best-fit positron trapping modes for these samples along with experimental data for TAS-235 glass calculated within two-state model are given in Table 1.

Fig. 1 Raw PAL spectra of glassy Te₂₀As₂₉Ga₁Se₅₀ (a) and Te₂₀As₂₉Ga₁Se₅₀ doped with 500 ppmw of Pr³⁺ (b) reconstructed from two-component fitting at the general background of source contribution (bottom inset shows statistical scatter of variance).

Table 1 Fitting parameters and positron trapping modes describing two-term reconstructed PAL spectra of glassy Te₂₀As₃₀Se₅₀,²¹ as well as parent and Pr³⁺-doped (500 ppmw) Te₂₀As₂₉Ga₁Se₅₀

Sample, state	Fitting parameters			Positron trapping modes
	τ1 ns	τ2 ns	I2 a.u.	τav. ns	τb ns	κd ns^−1	τ2 − τb ns	τ2/τb	η
Te20As30Se50	0.202	0.360	0.400	0.264	0.244	0.86	0.12	1.47	0.17
Z = 2.30; ρ = 4.888 g cm^−3
Te20As29Ga1Se50	0.208	0.358	0.370	0.263	0.246	0.75	0.11	1.44	0.16
Z = 2.30; ρ = 4.912 g cm^−3
Te20As29Ga1Se50 + 500 ppm Pr^3+	0.213	0.363	0.330	0.262	0.246	0.64	0.12	1.47	0.14
Z = 2.30; ρ = 4.910 g cm^−3

---

It is obvious that Ga-codoping does not cause increase in defect-specific positron lifetime τ₂ (like under crystallization),²¹ but rather leads to an opposite and very slight decaying tendency (decrease from 0.360 ns in TAS-235 glass to 0.358 ns in Ga-codoped Te₂₀As₂₉Ga₁Se₅₀ glass). In contrast, the intensity of second component I₂ is subjected to more pronounced changes, e.g. rough dropping, thus resulting in gradual decrease in positron trapping rate in defects κ_d and, correspondingly, the fraction of trapped positrons η (Table 1). It means that Ga-codoping does not approach fragmentation of free-volume voids in TAS-235 glass as it occurs in crystallized Te₂₀As₂₈Ga₂Se₅₀ alloy.²¹

Then, it is found that doping of Te₂₀As₂₉Ga₁Se₅₀ glass with 500 ppmw of Pr³⁺ further depresses the process of positron trapping (in part, the κ_d and η values are respectively reduced to 0.64 ns⁻¹ and 0.14), this tendency being realized via more essential increase in defect-related positron lifetime τ₂ (from 0.358 ns to 0.363 ns) accompanied by strong decrease in I₂ intensity (from 0.370 to 0.330). Noteworthy, neither defect-free bulk positron lifetime τ_b, nor average positron lifetime τ_av. are subjected to detectable changes (within an error-bar of PAL measurements) in these successive Ga-codoping and Pr³⁺-doping processes. Let's clarify physical meaning of these structural reconfiguration processes grounded on previous results for similar chalcogenide glass-forming systems.^{20–22,26–30}

Thus, it is well justified that among huge diversity of expected positron trapping sites possible in different ChG matrices, the preferential process of positron capturing is defined by extended free-volume defects in the nearest vicinity of chalcogen atoms neighboring with main glass-forming structural units (network-composing polyhedrons).^26–30 In view of extra-low two-fold coordination and strong directionality of covalent bonding, the chalcogen atoms form low-electron density spaces, termed also as bond free solid angles (BFSA) by Kastner.³¹ Such BFSA undoubtedly contribute to neighboring geometrical free-volume spaces, ensuring them to possess an effective negative electrical charge due to proximity with more electronegative chalcogen atom (Ch = S, Se, Te) linked with more electropositive cation-type neighbor (As). The Ch atoms form an outer wall for innermost free-volume voids, which can be identified in view of their preferential electric state as counterparts of cation vacancies in crystals. Therefore, the most efficient positron traps in ChG can be imagined as geometrical voids within cycle-type formations of Ch-interlinked polyhedrons, such as As(Se/Te)_3/2 pyramids, surrounded preferentially by Se- and/or Te-based BFSA. These voids are composed of atomic-accessible geometrical cores arrested by surrounding atomic-inaccessible shells formed by Se- and/or Te-based BFSA.

Specifically, in the case of chalcogen-rich TAS-235 glass (as compared with stoichiometric ChG built of corner-shared As(Se/Te)_3/2 pyramids),^18–21 an essential contribution of –Se–Se– bridges between pyramids somewhat depresses the positron trapping like in Se-rich glassy As–Se.²⁸ But just these homonuclear links can be affected by small Ga additions, as it follows from energetic balance of covalent chemical bonding in quaternary Ga–As–Se–Te system (see Table 2), composed by assuming the homonuclear Ga–Ga, As–As, Se–Se and Te–Te bond energies as 34.1 kcal mol⁻¹, 32.1 kcal mol⁻¹, 44.0 kcal mol⁻¹ and 33.0 kcal mol⁻¹, respectively,^32–34 and calculating the heteronuclear bond energies in respect to known Pauling method.³² Thus, in view of this analysis and in excellent agreement with previous structure study,²⁰ it follows that GaSe_4/2 tetrahedrons (not GaTe_4/2) mostly appear in TAS-235 glass under Ga codoping instead of homonuclear Ch–Ch (e.g. Se–Se, Se–Te and/or Te–Te) links. So corner-sharing polyhedrons directly linked through common Ch atom become dominating building blocks in Te₂₀As₂₉Ga₁Se₅₀ glass, respectively modifying an arrangement of free-volume voids.

Table 2 Mean molar energies E (kcal mol⁻¹) of covalent bonds in Ga-codoped TAS-235 glass

Bond	E, kcal mol^−1	Bond	E, kcal mol^−1
Ga–Ga	34.1	As–Se	41.7
As–As	32.1	As–Te	32.7
Se–Se	44.0	Ga–Se	55.2
Te–Te	33.0	Ga–As	37.2
Se–Te	44.2	Ga–Te	36.1

---

Following to well-known graphical presentation of idealized structure of RE-doped Ge–As–S glass given by Aitken,¹² this specificity of free-volume arrangement in TAS-235 glass can be presented as shown on Fig. 2a. The pyramidal AsSe_3/2, AsTe_3/2 and mixed As(Se/Te)_3/2 units interlinked directly through common Ch atom (–Se– and/or –Te– bridges) or through two Ch atoms (e.g. –Se–Se–, –Se–Te– and/or –Te–Te– bridges) create characteristic cycle-type arrangement with some amount of positron trapping sites in the form of agglomerated inner free-volume voids. Under transition to Ga-codoped Te₂₀As₂₉Ga₁Se₅₀ glass, the Ga-based GaSe_4/2 polyhedrons with energetically favorable Ga–Se bonds (Table 2) appear in the network of interlinked As(Se/Te)_3/2 pyramids (Fig. 2b). In preferential Ch environment (as in TAS-235 glass), the Ga behaves as typical metal tending to be four-fold coordinated by Ch atoms to form mostly GaSe_4/2 tetrahedrons (despite its main valence state +3), as it was well argued in previous research.²⁰ From charge compensation standpoint, it means that some excess of negative electrical charge is stabilized in a vicinity of these tetrahedrons. Created deficit in Ch is counterbalanced by disappearing of some –Se–Se–, –Se–Te– and –Te–Te– bridges at the cost of direct corner-shared links between polyhedrons. Under small amount of Ga added (not too far to disturbed preferential ring statistics in a glassy network), these structural reconfiguration processes do not change essentially the volume of existing positron traps (at least, near average atomic coordination 2.3 character for TAS-235, where rather smooth compositional dependence of defect-related positron lifetime τ₂ is expected for glassy As–Se).^26,28,35 Disappearing of some two-atomic –Ch–Ch– bridges along with formation of GaSe_4/2 tetrahedrons makes cycle-type formations in glassy Te₂₀As₂₉Ga₁Se₅₀ narrower (Fig. 2b), and free-volume voids arrested by interlinked polyhedrons become respectively smaller resulting in slightly reduced defect-related positron lifetime τ₂ (Table 1). So network of Ga-codoped glass becomes denser. Indeed, as it follows from Archimedes displacement measurements (in a distilled water with ±0.005 g cm⁻³ error-bar), the atomic density of glassy Te₂₀As₂₉Ga₁Se₅₀ reaches 4.912 g cm⁻³ instead of 4.888 g cm⁻³ for TAS-235 glass (thus giving in recalculation to molar volume, correspondingly, 17.80 cm³ mol⁻¹ and 17.90 cm³ mol⁻¹).²¹

Fig. 2 Sketch of plane projection of glass structure built of As(Ch)_3/2 pyramidal units interlinked by –Ch– or –Ch–Ch– bridges showing structural cycle-type arrangement with free-volume voids in Ga-free network (a), Ga-codoped network (b) and additionally RE-modified Ga-codoped network (c).

Such densification of atomic structure of Te₂₀As₂₉Ga₁Se₅₀ glass shifts balance between available positron annihilation paths from defect-specific trapping towards defect-free bulk trapping. Under such condition, the intensity of second component I₂ is not further a good parameter, describing realistic concentration of positron traps (since positron trapping now is essentially disturbed by increased probability of annihilation from bulk defect-free states). Even under unchanged defect-free bulk positron lifetime τ_b, which can be caused by counterbalanced inputs from changes in positron annihilation channels, this effect leads to depressed trapping rate in defects κ_d and fraction of trapped positrons η, as it is demonstrated in Table 1.

Effect of RE-doping on PAL modes in such modified structure of Ga-codoped Te₂₀As₂₉Ga₁Se₅₀ glass can be explained in terms of competitive contribution of changed occupancy positions available for RE ions and trapped positrons. Such free-volume sites possessing simultaneously an effective negative electrical charge are placed in a vicinity of Ga-based tetrahedrons as schematically illustrated in Fig. 2b. The Pr³⁺ ions are stabilized in the network of Te₂₀As₂₉Ga₁Se₅₀ glass due to strong covalent bridging Pr³⁺–Se/Te–Ga links,^7,8,12 thus eliminating corresponding void as potential positron trapping center (Fig. 2c). Thereby, the depressed positron trapping in Te₂₀As₂₉Ga₁Se₅₀ glass doped with 500 ppmw of Pr³⁺ (as it demonstrated by positron trapping modes gathered in Table 1) results from changed occupancy of these Ga-related free-volume positron traps.

As to the concentration of these extended positron trapping sites, it can be roughly estimated by accepting their analogy with negatively charged vacancies in elemental and compound semiconductors giving trapping coefficient at the level of 10¹⁵ atom per s.^36,37 With atomic densities and experimental trapping rates character for different glasses in Table 1, it gives defect concentration near 5 × 10¹⁶ cm⁻³. Of course, these defects can be affected by RE additions even at such low level as a few tens of ppmw, provided the same sites are responsible for RE occupation and positron trapping. Undoubtedly, just this specificity is most essential in application of PAL spectroscopy to study RE doping effects in ChG, especially in a view that conventional atomic structure sensitive probes (X-ray, electron or neutron diffraction) are ineffective because of under-margin level of added ions, which is typically beyond reliably detectable limits of these methods.^7,8 Noteworthy unchanged (τ₂ − τ_b) difference and τ₂/τ_b ratio (see Table 1) testify also in a favor of the same nature for positron traps in all studied ChG.

Conclusions

The PAL spectroscopy in positron lifetime measuring mode treated in terms of two-state trapping model was utilized firstly to study effect of 500 ppmw of Pr³⁺ ions doping on free-volume structure of Ga-modified Te₂₀As₃₀Se₅₀ (TAS-235) glass. The positron trapping is shown to be mostly depressed in Ga-codoped glassy Te₂₀As₂₉Ga₁Se₅₀, this process occurring without essential changing in the volume of positron trapping defects. Appearance of GaSe_4/2 tetrahedrons based on energetically favorable Ga–Se bonds possessing local excess of negative electrical charge within network of interlinked As(Se/Te)_3/2 pyramids is expected as principal response in the structure of this glass, the created deficit in chalcogen being counterbalanced by disappearing of some two-atomic chalcogen bridges at a cost of direct corner-sharing links between polyhedrons. It is found that doping of Te₂₀As₂₉Ga₁Se₅₀ glass with 500 ppmw of Pr³⁺ further depresses the positron trapping, this tendency being realized via more detectable increase in defect-related positron lifetime and strong decrease in I₂ intensity. This effect is explained in terms of competitive contribution of different occupancy positions available for RE ions and trapped positrons. The Pr³⁺ ions are stabilized in network of Te₂₀As₂₉Ga₁Se₅₀ glass in vicinity of GaSe_4/2 tetrahedrons due to strong bridging Pr–Se/Te–Ga links, thus eliminating respective free-volume voids as potential positron traps.

Acknowledgements

This research is supported by POLONIUM common actions program for years 2015-2016 realized in respect to bilateral Agreement on scientific-technical cooperation between Polish and French governments from 1966.

References

1. B. J. Eggleton, B. Luther-Davies and K. Richardson, Nat. Photonics, 2011, 5, 141–148.
2. Chalcogenide Glasses: Preparation, Properties and Applications, ed. J.-L. Adam and X. Zhang, Woodhead Publishing, Oxford, Cambridge, New Dehli, 2014.
3. S. Cui, R. Chahal, Ya. Shpotyuk, C. Boussard, J. Lucas, F. Charpentier, H. Tariel, O. Loreal, V. Nazabal, O. Sire, V. Monbet, Z. Yang, P. Lucas and B. Bureau, Proc. SPIE, 2014, 8938, 893805.
4. B. Bureau, X. Zhang, F. Smektala, J.-L. Adam, J. Troles, H. Ma, C. Boussard-Pledel, J. Lucas, P. Lucas, D. Le Coq, M. R. Riley and J. H. Simmons, J. Non-Cryst. Solids, 2004, 345–346, 276–283.
5. A. B. Seddon, Z. Tang, D. Furniss, S. Sujecki and T. M. Benson, Opt. Express, 2010, 18, 26704–26719.
6. E. R. Barney, Z. Tang, A. Seddon, D. Furniss, S. Sujecki, T. Benson, N. Neate and D. Gianolio, RSC Adv., 2014, 4, 42364–42371.
7. T. H. Lee, S. I. Simdyankin, L. Su and S. R. Elliott, Phys. Rev. B: Condens. Matter Mater. Phys., 2009, 79, 180202.
8. T. H. Lee, S. I. Simdyankin, J. Hegedus, J. Heo and S. R. Elliott, Phys. Rev. B: Condens. Matter Mater. Phys., 2010, 81, 104204.
9. A. Feltz, Amorphous and vitreous inorganic solids, Mir, Moscow, 1986.
10. B. Cole, L. B. Shaw, P. C. Pureza, R. Mossadegh, J. S. Sanghera and I. D. Aggarwal, J. Non-Cryst. Solids, 1999, 256–257, 253–259.
11. J. Hu, C. R. Menyuk, C. Wei, B. Shaw, J. S. Sanghera and I. D. Aggarwal, Opt. Lett., 2015, 40, 3687–3690.
12. B. G. Aitken, C. W. Ponader and R. S. Quimby, C. R. Chim., 2002, 5, 865–872.
13. M. F. Churbanov, I. V. Scribachev, V. S. Shiryaev, V. G. Plotnichenko, S. V. Smetanin, E. B. Kryukova, Yu. N. Pyrkov and B. I. Galagan, J. Non-Cryst. Solids, 2013, 326–327, 301–305.
14. R. Golovchak, Ya. Shpotyuk, V. Nazabal, C. Boussard-Pledel, B. Bureau, J. Cebulski and H. Jain, J. Chem. Phys., 2015, 142, 184501.
15. R. Golovchak, Ya. Shpotyuk, C. M. Thomas, V. Nazabal, C. Boussard-Pledel, B. Bureau and H. Jain, J. Non-Cryst. Solids, 2015, 429, 104–111.
16. R. Krause-Rehberg and H. Leipner, Positron annihilation in semiconductors: defect studies, Springer, Heidelberg, 1999.
17. Y. C. Jean, P. E. Mallon and D. M. Schrader, Principles and Application of Positron and Positronium Chemistry, World Sci. Publ. Co. Pte. Ltd., New Jersy-London-Singapore-Hong Kong, 2003.
18. G. Delazir, M. Dussauze, V. Nazabal, P. Lecante, M. Dolle, P. Rozier, E. I. Kamitsos, P. Jovari and B. Bureau, J. Alloys Compd., 2011, 509, 831–836.
19. P. Jovari, B. Bureau, I. Kaban, V. Nazabal, B. Beuneu and U. Rutt, J. Alloys Compd., 2009, 488, 39–43.
20. Ya. Shpotyuk, B. Bureau, C. Boussard-Pledel, V. Nazabal, R. Golovchak, P. Demchenko and I. Polovynko, J. Non-Cryst. Solids, 2014, 398–399, 19–25.
21. O. Shpotyuk, A. Ingram, B. Bureau, Ya. Shpotyuk, C. Boussard-Pledel, V. Nazabal and R. Szatanik, J. Phys. Chem. Solids, 2014, 75, 1049–1053.
22. Ya. Shpotyuk, C. Boussard-Pledel, V. Nazabal, R. Chahal, J. Ari, B. Pavlyk, J. Cebulski, J. L. Doualan and B. Bureau, Opt. Mater., 2015, 46, 228–232.
23. J. Kansy, Nucl. Instrum. Methods Phys. Res., Sect. A, 1996, 374, 235–244.
24. F. Tuomisto and I. Makkonen, Rev. Mod. Phys., 2013, 85, 1583–1631.
25. A. Seeger, Appl. Phys., 1974, 4, 183–199.
26. O. Shpotyuk and J. Filipecki, Free volume in vitreous chalcogenide semiconductors: possibilities of positron annihilation lifetime study, WSP, Czestochowa, 2003.
27. M. Hyla, J. Filipecki, O. Shpotyuk, M. Popescu and V. Balitska, J. Optoelectron. Adv. Mater., 2007, 9, 3177–3181.
28. A. Ingram, R. Golovchak, M. Kostrzewa, S. Wacke, M. Shpotyuk and O. Shpotyuk, Phys. B, 2012, 407, 652–655.
29. O. Shpotyuk, R. Golovchak, A. Ingram, V. Boyko and L. Shpotyuk, Phys. Status Solidi C, 2013, 10, 117–120.
30. O. I. Shpotyuk, J. Filipecki and V. O. Balitska, J. Optoelectron. Adv. Mater., 2008, 10, 3193–3197.
31. M. Kastner, Phys. Rev. B: Condens. Matter Mater. Phys., 1973, 7, 5237–5252.
32. L. Pauling, The nature of the chemical bond, Cornell Univ. Press, New York, 1960.
33. J. Bicerano and S. R. Ovshinsky, J. Non-Cryst. Solids, 1985, 74, 75–84.
34. L. Tichy and H. Ticha, J. Non-Cryst. Solids, 1995, 189, 141–146.
35. O. K. Alekseeva, V. I. Mihajlov and V. I. Shantarovich, Phys. Status Solidi A, 1978, 48, K169–K173.
36. R. Krause-Rehberg and H. S. Leipner, Appl. Phys. A, 1997, 64, 457–466.
37. P. Hautojarvi, Mater. Sci. Forum, 1995, 175–178, 47–58.

---