Raman spectra of MCl-Ga₂S₃-GeS₂ (M = Na, K, Rb) glasses

Introduction

Alkali halide doped sulfide and selenide glasses are promising functional materials for optical applications, portable ion sources, electronic devices and reliable chemical sensors [1], [2], [3], [4], [5]. However, the systematic studies of these glasses as a function of the cation nature and the content of alkali halide are limited. Recently, we have investigated the glass-forming regions and macroscopic properties in the MY-Ga₂S₃-GeS₂ systems (M = Na, K, Rb; Y = Cl, Br, I) [6, 7]. The structural analysis is necessary for a better understanding of the properties of glass.

Raman scattering is a very powerful spectroscopic technique for detecting the local order in the amorphous solids. Several Raman spectroscopy studies have been carried out in lithium, potassium and caesium halide doped Ga₂S₃-GeS₂ glasses [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18] but only few reports are devoted to sodium and rubidium containing counterparts [19], [20], [21]. It is generally accepted that the incorporation of metal halides into chalcogenide glassy matrix does not change the Raman spectra significantly. Nevertheless, the comparative study of the Raman spectra in the alkali halide doped Ga₂S₃-GeS₂ glasses is still missing and deserves to be carried out.

In this paper we present the structural analysis of the (MCl) _x (Ga₂S₃)_0.2−0.2x (GeS₂)_0.8−0.8x glasses (M = Na, K, Rb) using Raman spectroscopy. The main reason to choose this composition line is the constant Ga₂S₃/GeS₂ ratio in the host matrix whatever the amount of the incorporated salt. Thus, the only effect of MY addition on the (Ga₂S₃)_0.2(GeS₂)_0.8 glass is observed. In the case of sodium containing glasses, the compositions with 30 mol% NaBr and NaI are also investigated. The main purpose of this work is to compare the Raman spectra of Ga₂S₃-GeS₂ glass doped with light and heavy alkali halides of different concentrations.

Materials and methods

Results and discussion

All the measured raw Raman spectra were analyzed over the 70–600 cm⁻¹ range. The spectral background was approximated by a Voigt function and then subtracted from the experimental data with further normalization to the most intense feature. Typical raw Raman spectra treatment is shown in Fig. S1 (Supplementary Material). The resulting Raman spectra of the (NaCl) _x (Ga₂S₃)_0.2−0.2x (GeS₂)_0.8−0.8x glasses, 0.0 ≤ x ≤ 0.25, are shown in Fig. 1(a). The Raman spectra of the vitreous matrix (Ga₂S₃)_0.2(GeS₂)_0.8 and (NaY)_0.3(Ga₂S₃)_0.14(GeS₂)_0.56 glasses, where Y = Br or I, are visualised in Figs. 2 and 3. The results for the potassium (0.0 ≤ x ≤ 0.30) and rubidium (0.0 ≤ x ≤ 0.40) containing counterparts are presented in Figs. 4 and 5, respectively. Before discussing the structural changes in glassy matrix (Ga₂S₃)_0.2(GeS₂)_0.8 related to alloying with alkali halides, it seems reasonable to understand what happens in glassy GeS₂ network with Ga₂S₃ additions.

Fig. 1:

(a) Experimental Raman spectra of (NaCl) _x (Ga₂S₃)_0.2−0.2x (GeS₂)_0.8−0.8x glasses, 0.0 ≤ x ≤ 0.25; (b) the difference Raman spectra of NaCl-containing glasses obtained by subtraction of the spectrum for vitreous matrix (Ga₂S₃)_0.2(GeS₂)_0.8.

Fig. 2:

Experimental Raman spectra of pure glassy matrix (Ga₂S₃)_0.2(GeS₂)_0.8 and a NaBr-Ga₂S₃-GeS₂ glass containing 30 mol% NaBr.

Fig. 3:

Experimental Raman spectra of pure glassy matrix (Ga₂S₃)_0.2(GeS₂)_0.8 and a NaI-Ga₂S₃-GeS₂ glass containing 30 mol% NaI.

Fig. 4:

Experimental Raman spectra of (KCl) _x (Ga₂S₃)_0.2−0.2x (GeS₂)_0.8−0.8x glasses, 0.0 ≤ x ≤ 0.30.

Fig. 5:

Experimental Raman spectra of (RbCl) _x (Ga₂S₃)_0.2−0.2x (GeS₂)_0.8−0.8x glasses, 0.0 ≤ x ≤ 0.40.

Fig. 6(a) shows the Raman spectra of g-GeS₂ (taken from [22]) and vitreous matrix (Ga₂S₃)_0.2(GeS₂)_0.8. Gallium sulphide broadens the entire spectrum and slightly shifts to lower frequencies the main spectroscopic feature of glassy GeS₂, the A 1 in-phase breathing mode of corner-sharing CS-GeS_4/2 tetrahedra at 344 cm⁻¹ [23], [24], [25]. In addition, a new structural feature appears at ≈270 cm⁻¹. The frequency of the last mode is characteristic for Ge–Ge or Ga–Ga stretching in chalcogenide glasses and suggests the appearance of Ge–Ge or Ga–Ga homopolar bonds. Glassy GeS₂ reveals a weak chemical disorder [26, 27] reflected by the Ge–Ge stretching mode at 259 cm⁻¹, Fig. 7(a). The intensity of this mode is strongly enhanced in Ge-rich glasses, e.g. (GeS)_0.1(GeS₂)_0.9, shown in Figs. 6(b) and 7(b) [28]. Nevertheless, the peak position and line shape imply the 270 cm⁻¹ peak is rather related to Ga–Ga homopolar bonds. The DFT modelling of Raman spectra for glassy Ga₂S₃-GeS₂ alloys [28] confirms this suggestion and also reveals the asymmetric low-frequency broadening of the A 1 breathing mode at 344 cm⁻¹ is associated with the appearance of Ga-S triclusters in the GeS₂ host glass.

Fig. 6:

Experimental Raman spectra of (a) glassy GeS₂ [22] and (Ga₂S₃)_0.2(GeS₂)_0.8, and (b) glassy GeS₂ [22] and (GeS)_0.1(GeS₂)_0.9 [28].

Fig. 7:

Experimental Raman spectra of (a) glassy GeS₂ [22] and (Ga₂S₃)_0.2(GeS₂)_0.8, and (b) glassy GeS₂ [22] and (GeS)_0.1(GeS₂)_0.9 [28] (detail).

The origin of the Ga-S triclusters, i.e., the structural units containing three-fold coordinated sulfur, is directly related to tetrahedral gallium coordination and Ga₂S₃ stoichiometry. Two tetrahedral Ga atoms are forming 8 Ga–S bonds. Consequently, three sulfur species should have the average S-Ga coordination N S − Ga  = 2⅔, or ⅓ of sulfur is two-fold coordinated and ⅔ of sulfur has trigonal local environment. The crystal structure of monoclinic α-Ga₂S₃, space group C c , fully corresponds to this scenario [29]. Fig. 8 shows the lattice and structural motifs in α-Ga₂S₃. The interatomic distances are also different for the two-fold S_2F and three-fold S_3F coordinated sulfur: Ga-S_2F = 2.196 ± 0.005 Å and Ga-S_3F = 2.315 ± 0.008 Å.

Fig. 8:

Lattice and structural motifs in monoclinic Ga₂S₃ [29]: (a) selected two-fold S_2F and three-fold S_3F coordinated sulfur species in the lattice are indicated by red circles, (b) a Ga-S tricluster, (c) a corner-sharing CS-Ga₂S₇ dimer.

Nevertheless, the origin of homopolar Ga–Ga bonds and the 270 cm⁻¹ stretching mode remains unclear in stoichiometric Ga₂S₃-GeS₂ glasses, except for a possible rather weak chemical disorder. We should note that the Ga-S triclusters are over-coordinated, N i j  = 3.2, and too rigid for a glass network with ideal average coordination number N i j  = 2.4 [30, 31], as in glassy As₂S₃ or GeSe₄. Consequently, some Ga-S triclusters should be transformed into less rigid units allowing the gallium accommodation within the glass network. The ethane-like S_3/2Ga-GaS_3/2 units, ETH-Ga₂S_6/2, with homopolar Ga–Ga bond, tetrahedral gallium and two-fold coordinated S_2F species, having acceptable average coordination N i j  = 2.8, represent thus the best choice to satisfy the rigidity requirements, stoichiometry and Ga local coordination. The only problem seems to be a higher energy of ETH-Ga₂S_6/2 compared to Ga-S triclusters. The structural equilibrium

where (Ga₂S₃)* represents the average structure with ⅔ of Ga-S triclusters, should depend from the glass chemical composition, synthesis temperature and quenching rate. The observed differences in the amplitude of the 270 cm⁻¹ mode, reported by different research groups [19, 32, 33], are consistent with this suggestion.

The Raman spectra of NaY-Ga₂S₃-GeS₂ glassy alloys (Y = Cl, Br, I), Figs. 1 –3, exhibit similar composition trends: (i) the Ga–Ga stretching mode at 270 cm⁻¹ decreases with increasing sodium halide content x; (ii) a new feature at ≈320 cm⁻¹ emerges and grows with x; and (iii) broad unresolved high-frequency modes at ≈380 cm⁻¹ and ≈435 cm⁻¹ are also decreasing. The difference spectra in Fig. 1(b) reveal clearly these composition trends. The halogen nature does not affect significantly the Raman spectra of the glasses.

The 270 cm⁻¹ and partly ≈380 cm⁻¹ modes belong to the ETH-Ga₂S_6/2 units decreasing with increasing x. A fraction of the last poorly resolved feature is related to symmetric and asymmetric Ga-S stretching in the ethane-like units [28]. The 435 cm⁻¹ peak seems to be the high-frequency F 2 mode in ES-GeS_4/2. The edge-sharing ES-GeS_4/2 tetrahedra remain largely intact in the vitreous matrix (Ga₂S₃)_0.2(GeS₂)_0.8, Fig. 6(a), evidenced also by the A 1 c mode at 372 cm⁻¹. However, sodium halide additions are changing the network topology, diminishing both the Ga-related entities, ETH-Ga₂S_6/2, and the Ge-centred structural units, ES-GeS_4/2.

The emerging and growing ≈320 cm⁻¹ feature seems to be related to Ga–Cl stretching in mixed GaCl _m S_4-m tetrahedra, where m = 1 or 2. Similar modes were reported for CsAlSCl₂ (at 325 cm⁻¹) and CsGaSCl₂ (at 315 cm⁻¹) glasses and attributed to vibrations in the m = 2 mixed tetrahedra, forming corner-sharing chains (CS-ACl₂S_2/2) _k , where A = Al or Ga [34]. Later, similar hypothesis was formulated for MCl-Ga₂S₃ glasses [19], where M = Ag, Tl, Rb, and Cs, but for different tetrahedral stoichiometry, GaClS_3/2, i.e., m = 1. Neutron and high-energy X-ray diffraction as well as the DFT modelling of Raman spectra were used to verify the gallium local environment in the CsCl-Ga₂S₃ glasses [13, 35], essentially confirming the m ≈ 1 stoichiometry.

The suggested appearance of the mixed GaYS_3/2 tetrahedra in the NaY-Ga₂S₃-GeS₂ glasses compensates the excessive rigidity of Ga-S triclusters without their transformation into ETH-Ge₂S_6/2, since the average coordination in the mixed tetrahedron, N i j  = 2.2, appears to be even lower than that in the ETH-unit, N i j  = 2.8. Ideally, the Ga-S triclusters disappear completely when equimolar ratio Y/Ga = 1 is achieved at x = 0.2857. The Raman data roughly confirm this limit, at least for ETH-Ge₂S_6/2, whose intensity appears to be negligible at x = 0.3.

The Raman spectra of potassium and rubidium chloride doped Ga₂S₃-GeS₂ glasses, Figs. 4 and 5, follow the similar changes with the increase of MCl molar content: the peak at ≈270 cm⁻¹ attributed to the ETH-Ga₂S_6/2 units decreases in intensity and a ≈320 cm⁻¹ feature related to the mixed GaClS_3/2 tetrahedra emerges and grows with x. It should be noted that the glass-forming region in the NaY-Ga₂S₃-GeS₂ systems broadens with the increasing alkali radius and the largest amount of dissolved alkali halide can be reached in the (MCl) _x (Ga₂S₃)_0.2−0.2x (GeS₂)_0.8−0.8x systems (M = Na, K, Rb) for rubidium-doped glasses with 40 mol% RbCl [7]. Moreover, it is generally assumed that MY additions interact essentially with Ga₂S₃ in the pseudo-ternary MY-Ga₂S₃-GeS₂ glasses. Consequently, the gradual changes in the relative intensity of the peak at ≈320 cm⁻¹ can be seen clearly in Raman spectra for cation-rich glasses from 25 mol% NaCl to 40 mol% RbCl. In the case of (RbCl)_0.4(Ga₂S₃)_0.12(GeS₂)_0.48 composition, Ga–Cl stretching in mixed GaClS_3/2 tetrahedra becomes the most intense spectral feature, Fig. 9. The same spectral evolution was observed for (CsCl) _x (Ga₂S₃)₂₅(GeS₂)_75−x glasses [14]. The dominant GeS₂ peak at ≈340 cm⁻¹ broadens as the CsCl content increases over 20 mol% in the system and separates at 40 mol% CsCl.

Fig. 9:

Experimental Raman spectra of (MCl) _x (Ga₂S₃)_0.2−0.2x (GeS₂)_0.8−0.8x glasses: x = 25 for M = Na; x = 30 for M = K; x = 40 for M = Rb.

A gradual blue-shift of the vibration mode between 390 and 420 cm⁻¹ is observed for potassium-rich (KCl) _x (Ga₂S₃)_0.2−0.2x (GeS₂)_0.8−0.8x glasses with the addition of KCl, Fig. 4. Similar Raman spectral evolution has already been reported for potassium and caesium-doped glasses in this system [11, 14, 16]. The growing shoulder located at ≈390 cm⁻¹ and the peak positioned at ≈420 cm⁻¹ were attributed to the symmetric stretching vibrations of the outer Ga–Cl bonds in mixed GaClS_3/2 and Ga₂Cl₂S_4/2 subunits, respectively. Indeed, the above assumption was supported by the study of the vibrational spectrum of molten dimeric Ga₂Cl₆, where the strongest Raman feature at ≈413 cm⁻¹ was assigned to the symmetric stretching vibrations of the outer Ga–Cl bonds [36]. The origin of the peak at ≈254 cm⁻¹ for the (KCl)_0.3(Ga₂S₃)_0.14(GeS₂)_0.56 glass is rather controversial. This feature was not observed for the same composition in the previous study on KCl-Ga₂S₃-GeS₂ glasses [11]. Nevertheless, the authors present the Raman spectra for 2KCl – Ga₂S₃ solid with the most intense spectral band located at ≈250 cm⁻¹.

The rubidium-rich (RbCl)_0.4(Ga₂S₃)_0.12(GeS₂)_0.48 composition presents the low intensity Raman feature at ≈475 cm⁻¹, Fig. 5. This peak could be assigned to the contribution of S–S homopolar bonds, which is present in S₈ ring or S_n chains [26] and appears due to enhanced chemical disorder in the glasses. Indeed, the (RbCl)_0.4(Ga₂S₃)_0.12(GeS₂)_0.48 glass is on the limit of the glass-forming region and seems to be phase-separated [7]. Further studies of local and intermediate-range order in MY-Ga₂S₃-GeS₂ glasses will have been carried out using pulsed neutron and high-energy X-ray diffraction.

The decreasing connectivity of the Ga-Ge-S glass network, related to increasing fraction of the mixed tetrahedra, results in lower glass transition temperatures [7]. The diminishing rigidity decreases also the crystallisation ability; the glasses become more resistant to crystallisation. The mobile M⁺ cations are giving rise to ionic conductivity in the glasses. The ionic conductivity increases with NaY content but remains nearly invariant to the halide nature; glasses containing NaI have roughly the same conductivity as the glasses with NaCl additions. The origin of the last phenomenon resides probably in the glass network compression. Covalent Ge-rich subnetwork prevents expansion of structural regions containing mixed tetrahedra and sodium species, when smaller halide as Cl is replaced by large iodine species. Moderate changes of Na-Y interatomic distances compared to those in crystalline NaY compounds are consistent with this hypothesis [37].

Conclusions

The changes in Raman spectra with the addition of alkali halides into the chalcogenide glass were investigated by analyzing (MY) _x (Ga₂S₃)_0.2−0.2x (GeS₂)_0.8−0.8x systems (M = Na, K, Rb; Y = Cl, Br, I). The Ga₂S₃-GeS₂ vitreous matrix is different from GeS₂-like disordered network in two aspects: (i) the presence of Ga-S triclusters containing 3-fold coordinated sulfur S_3F, and (ii) the ethane-like ETH-Ga₂S_6/2 units with homopolar Ga–Ga bonds. The Ga-S triclusters, forming ⅔ of the crystalline lattice in monoclinic Ga₂S₃, probably exist in the melt. The ETH-Ga₂S_6/2 units appear in stoichiometric Ga₂S₃-GeS₂ glasses on cooling and vitrification because of excessive rigidity of the Ga-S triclusters (the average coordination N i j  = 3.2 vs. the optimum glass value of 2.4). They are structurally acceptable, N i j  = 2.8, and probably energetically unfavourable but frozen in the glass network. Alloying with alkali halides proposes another solution decreasing the network rigidity: formation of mixed GaYS_3/2 tetrahedra, N i j  = 2.2. Consequently, the ETH-Ga₂S_6/2 units disappear while the mixed GaYS_3/2 tetrahedra grow with increasing MY content.