Competitive Phase Separation to Controllable Crystallization in 80GeS₂·20In₂S₃ Chalcogenide Glass

Abstract

Glass–ceramics of 80GeS₂·20In₂S₃ were fabricated by heat-treating the base glass at 402°C (T_g + 30°C) for different durations. The glass–ceramics exhibited some improved mechanical properties such as hardness and resistance to crack propagation, and meanwhile remained an excellent infrared (IR) transmission. The XRD and Raman results showed that only In₂S₃ crystals were precipitated inside glassy matrix. The evolution of two crystallization peaks (CPs) in differential scanning calorimeter (DSC) curves were studied with samples heat-treated at 402°C for different durations. It was found that the precipitation of In₂S₃ crystal phase is responsible for the low-temperature (first) CP, whereas the high-temperature (second) CP shifts to a higher temperature with the elongation of the heat-treatment duration. The crystallization of the higher temperature phase was inhibited with the precipitation of In₂S₃. Furthermore, crystallization mechanism was investigated using the nonisothermal method. The computed results showed that strictly more energy (higher activation energy, E_c) is essential for the precipitation of the higher temperature phase, which is in accordance with the DSC study of crystallized samples. More noticeable, the crystallization rate constant (K) value of 6.639 × 10⁻⁸ s⁻¹ for the second CP is ~ 5 orders of magnitude smaller than that of the In₂S₃ phase, and this significant difference makes the crystallization of higher temperature crystal phase very hard. Consequently, controllable crystallization of 80GeS₂·20In₂S₃ chalcogenide glass–ceramics with sole In₂S₃ crystallites can be achieved easily.

I. Introduction

In the past decades, chalcogenide glasses have been studied extensively due to their practical and potential applications in the fields of infrared (IR) imaging, chemical sensing, and optical communications.[1-5] To satisfy the application requirements, their weaknesses such as poor resistance to crack propagation and thermal shocks should be improved. Fortunately, crystallization procedure is an effective way to improve the mechanical properties of the original glass.[6, 7] The attempt to obtain chalcogenide glass–ceramics by crystallization treatment has been made since 1976.[8] Since then, the crystallization behavior of chalcogenide glasses have been investigated in some different systems such as GeS₂–Ga₂S₃, GeSe₂–Ga₂Se₃, GeSe₂–Sb₂Se₃, and Ge–(Ga/Sb)–(Se/S)–MX (MX = alkali halide).[9-12] In these studies, chalcogenide glass–ceramics were obtained from base glasses using an appropriate heat treatment. Such glass–ceramics present improved mechanical properties, and kept an excellent IR transmittance. But the crystallization mechanism, which plays an important role in controllable crystallization, was not discussed thoroughly in previous researches.

According to previous researches, GeS₂-based chalcogenide glasses have been shown to provide excellent optical and thermal properties.[13, 14] In particular, the GeS₂–Ga₂S₃ glasses have augmented rare-earth solubility which is due to the structural modification by addition of gallium.[15, 16] Considering the similarity of chemical properties of indium and gallium, GeS₂–In₂S₃ glasses should have similar basic properties.[17] Furthermore, large third-order optical nonlinearity and efficient emission of rare-earth ions would benefit from the increase in linear refractive index with the addition of indium.[18-20] Especially, few literatures have focused on the crystallization mechanism in GeS₂–In₂S₃ system.

In this work, we investigated the precipitated crystal phase, optical and mechanical properties, and the crystallization behavior of 80GeS₂·20In₂S₃ glass–ceramics. Based on the differential scanning calorimeter (DSC), XRD, and Raman results, the crystallization mechanism is elucidated in detail. Our discussion will focus on the inhibited crystallization of the second crystal phase (high-temperature phase) which is caused by the precipitation of In₂S₃ (low-temperature phase), which leads to controllable crystallization for reproducible chalcogenide glass–ceramics containing In₂S₃ crystals.

Compared with previous work in 80GeS₂·20Ga₂S_3,[9] we found a totally different crystallization behavior in this studied glass, we think this difference may be due to the microstructure difference between these two glasses. The existence of [InS₆] octahedral leads to a much deeper sulfur deficiency in the 80GeS₂·20In₂S₃ glass than that in the 80GeS₂·20Ga₂S₃. But it still needs to be confirmed, and it is ongoing in our further studies.

II. Experimental Procedure

Bulk glass with a composition of 80GeS₂·20In₂S₃ was prepared by the conventional melt-quenching technique. The weighed highly pure germanium, indium, and sulfur (5N) were sealed into cleaned quartz glass ampoules under a vacuum of ~10⁻³ Pa and melted at 1000°C for 36 h in a rocking furnace to allow thorough reaction. Bulk samples were obtained by quenching the melts in water with subsequent annealing at a temperature of 30°C below the glass transition temperature (T_g) to remove inner constraints. The glass rod was cut and polished into disks of ~10 mm in diameter and 1 mm in thickness. Glass–ceramics were obtained after heat treatments at different temperatures above T_g for different durations.

To obtain the thermodynamic information of the samples, calorimetric measurements were performed using a DSC (TA Instruments Q2000, New Castle, DE) with a temperature accuracy of ±1°C. The transmission spectra in visible and near-IR region were recorded with a UV–Vis–NIR spectrophotometer (PerkinElmer Lambda 950, Waltham, MA). Crystalline morphology in the glass–ceramics was observed using a scanning electronic microscope (Tescan VEGA 3 SBH, Czech) with an accelerating voltage of 10 kV, performed on a fresh surface. The Vickers hardness values were obtained by a Vickers microindenter (Everone MH-3, Everone Enterprises. Ltd., Shanghai, China), and the marks and cracks made by indentation with a charge of 100 g for 5 s were observed by optical microscope (VHX-1000E; Keyence Corporation, Osaka, Japan). The crystalline phases precipitated in the heat-treated samples were identified by X-Ray diffraction (XRD). The XRD data were collected by a diffractometer (Bruker D2 phaser, λ = 0.15406 nm, 30 kV, 10 mA, CuKα) with a step width of 0.02°. Raman spectra were recorded at room temperature using back scattering configuration by the laser confocal Raman spectrometer (Renishaw InVia, Gloucestershire, UK) with excitation wavelength of 488 nm. The resolution in the frequencies is 1 cm⁻¹.

III. Results and Discussion

The DSC curve of the as-prepared glass (80GeS₂·20In₂S₃) is shown in Fig. 1. It clearly shows a glass transition at a temperature (T_g) of 372°C. Moreover, two obvious exothermic crystallization peaks (CPs) can be seen at the temperatures of T_p1 and T_p2. The difference (ΔT = T_x1−T_g) of onset CP temperature (T_x1) and T_g is a critical parameter to evaluate the thermal stability. To control the crystallization, a suitable ΔT is important. Herein, a suitable ΔT value of 91°C indicates that it is a sub-stable glass, which is appropriate for controllable process of nucleation and crystal growth.[21] The characteristic temperatures of T_g, T_x, T_p, and ΔT are listed in Table 1. Based on the DSC results and previous studies,[6, 22, 23] a fairly low temperature of 402°C (T_g + 30°C) was chosen as heat-treatment temperature for crystallization process.

Table 1. Characteristic Temperatures of T_g, T_x, T_p, and ΔT for 80GeS₂·20In₂S₃

Heating rate, β(°C/min)	Tg (°C)	Tx1 (°C)	Tp1 (°C)	Tx2 (°C)	Tp2 (°C)	ΔT (°C)
3	359	447	460	509	546	88
5	361	449	465	505	554	88
10	372	463	476	547	560	91
15	375	469	480	552	565	94
20	374	472	485	559	569	98
25	377	476	488	560	571	99

Figure 1. The DSC curve for bulk glass of 80GeS₂·20In₂S₃ at a heating rate of 10°C/min.

It is well known that the optical transmittance of glass–ceramics will decrease as the size of the precipitated crystals comes close to the wavelength.[24] Thus, the heat treatment must keep crystals precipitated from the glass matrix small enough to ensure good optical transmittance in the required IR region. The Vis–NIR transmission spectra, which are very sensitive to the presence of crystals, are shown in Fig. 2. The sample heat-treated at 402°C for 10 h showed a decline of transparence from 520 to 950 nm, but kept the original transmission level in the longer wavelength region. It can be concluded that the presence of small crystals does not impair the transparency in the application IR window. The cut-off edge of short wavelength red-shifted with the prolongation of heat-treatment times from 10 to 60 h, indicating the nucleation and growth of the crystals in the glassy matrix. In addition, there was a slight decrease in maximum transmittance when the heat-treatment durations were prolonged to 20 h, which is possibly due to the growth and agglomeration of the precipitated crystallites. The transmittance scarcely changed when the duration extended from 60 to 80 h, suggesting that no more crystallites were further precipitated.

Figure 2. Transmission spectra of 80GeS₂·20In₂S₃ base glass and glasses heat-treated at 402°C for 10, 20, 30, 40, 60, and 80 h, respectively.

Scanning electron microscope (SEM) images of the fresh surface of base glass and glass–ceramics are shown in Fig. 3. For the sample heat-treated at 402°C for 20 h as indicated in Fig. 3(b), the precipitated crystals are of size about 50 nm, and the degree of crystallinity is about 40%. With the prolongation of heat-treated times, the size of crystals become larger, and result in an increase in scattering loss, which is in good agreement with the transmission spectra.

Figure 3. SEM images for samples crystallized at 402°C for different durations: (a) base glass; (b) 20 h; (c) 30 h; and (d) 40 h.

To investigate whether the mechanical properties would be improved after crystallization procedure, Vickers indentations were performed on the base glass and glass–ceramics. The Vickers hardness values are listed in Table 2. Compared with the base glass, the hardness of the glass–ceramics was increased. The marks and cracks made by indentation were observed with an optical microscope. As shown in Fig. 4, a decrease in crack length has been observed with increasing heat-treatment durations, indicating a progressive strengthening of the glass by precipitating crystallites inside the glassy matrix.

Table 2. The Vickers Hardness of the 80GeS₂·20In₂S₃ Glass and Glass–Ceramics

Samples	base	7 (h)	10 (h)	20 (h)	30 (h)	40 (h)	60 (h)
Hardness, Hv (±2 Kg mm^−^2)	215	223	227	237	248	253	265

Figure 4. Optical microscope pictures of the different samples after Vickers indentation: (a) base glass and glasses crystallized at 402°C for (b) 10 h, (c) 20 h, and (d) 60 h, respectively.

To identify the precipitated crystalline phase, XRD patterns of the samples were collected using the bulk glass and glass–ceramics, as shown in Fig. 5. The XRD patterns reveal that only diffuse humps appear for the base glass sample, evidencing the amorphous nature of the original structure. While after heat treatment, several peaks appear at the location of 2θ = 27.6°, 33.5°, 43.9°, and 48.0°, and then grow slowly and continuously. The position of these diffraction peaks are in good agreement with the In₂S₃ (JCPDF card of no. 73-1366), indicating only In₂S₃ phase was precipitated during the crystallization procedure. Moreover, the Raman spectra of the glass–ceramics and In₂S₃ crystal were obtained by laser confocal Raman spectrometer. As shown in Fig. 6, the dominating Raman band of 80GeS₂·20In₂S₃ base glass locates at about 342 cm⁻¹, which is assigned to the symmetrical stretching vibrations of GeS₄ tetrahedra.[25, 26] With the increase in the heat-treatment time, the bands around 308 and 245 cm⁻¹ which correspond to the symmetric stretching vibration of InS₄ and InS₆ units[27] are gradually enhanced, and the band at 308 cm⁻¹ (InS₄) becomes the dominating peak instead of the band at 342 cm⁻¹ (GeS₄). Based on the above-mentioned Raman analysis, the crystallites in the glass–ceramics have the structural units of InS₄ and InS₆, which is in good agreement with the XRD analysis. In a word, the evolution of the Raman spectra in Fig. 6 exposits the structural change in the glass ceramics, and further proved the precipitation of In₂S₃ crystallites.

Figure 5. XRD patterns of 80GeS₂·20In₂S₃ base glass and glasses heat-treated at 402°C for 7, 10, 20, 30, 40, and 60 h, respectively. The JCPDF card (No. 73-1366) of In₂S₃ standard crystal is also present.

Figure 6. Raman spectra of 80GeS₂·20In₂S₃ base glass and glasses heat-treated at 402°C for 10, 20, 30, 40, 60, and 80 h, respectively.

Analogous study of crystallization behavior can be found in similar chalcogenide glass with a composition of 80GeS₂·20Ga₂S₃.[9] In that case, the GeS₂ phase is precipitated in the glassy matrix following the Ga₂S₃ phase. In this case, however, only In₂S₃ crystallites were precipitated after heat treatment in 80GeS₂·20In₂S₃ glass. It means that the In₂S₃ crystallites inhibit the precipitation of GeS₂·phase. To prove the supposition, DSC curves of glasses heat-treated at 402°C for different durations were recorded. As shown in Fig. 7, the DSC curves of crystallized glass samples provide the information of crystallization procedure. After heat treatment, the low-temperature (first) CP shifts toward lower temperature. It is possibly due to the increasing number of nuclei in the glass matrix, and leading to an easy crystallization. In addition, the first CP obviously decreases with the prolongation of heat-treatment durations and almost vanishes after 60 h heat treatment, indicating the total crystallization of the associated crystal phase. Based on the XRD results and the transmittance analysis, no more In₂S₃ crystallites were further precipitated when the heat-treatment durations at 402°C prolonged from 60 to 80 h, which is in good agreement with the evolution of the first CP. Consequently, the In₂S₃ crystal phase is responsible for the first CP in the 80GeS₂·20In₂S₃ glass. The high-temperature (second) CP showed a different evolution. With the progress in heat treatment, it shifts to the higher temperature, suggesting that the precipitation of crystal phase associated with second CP is inhibited by the competitive phase separation of In₂S₃ crystalline phase.

Figure 7. DSC curves of the bulk glasses heat-treated at 402°C for different durations.

To investigate why the second phase is inhibited to be precipitated, a nonisothermal DSC method was employed to evaluate the kinetic parameters for crystallization of 80GeS₂·20In₂S₃ glass. Table 1 also lists the characteristic temperatures of the studied chalcogenide glass at different heating rates, β = 3°C, 5°C, 10°C, 15°C, 20°C, and 25°C/min. The crystallization behavior of the studied chalcogenide glass can be understood by the following kinetic parameters: the activation energy for the crystallization E_c, the frequency factor K₀, and the crystallization rate constant K. To obtain these kinetic parameters, the kinetic model of Bansal and Hyatt[28] is used and expressed as follows:

• (1)

where R is the ideal gas constant (8.314 J mol⁻¹ K⁻¹), T_c is the crystallization temperature in Kelvin, and β is the heating rate, respectively. The plots of against 1000/T_c are shown in Fig. 8. From the slope of these lines, the values of the activation energy for crystallization, E_c, is obtained. The values of K₀ can also be deduced from the vertical axis intersection as given by Eq. (1). The calculated values of E_c and K₀ for the first and second CP are 331.58 KJ/mol, 9.89 × 10²² s⁻¹ and 482.57 KJ/mol, 1.44 × 10³⁰ s⁻¹, respectively. These kinetic parameters are listed in Table 3. In addition, the crystallization rate constant K is employed to reveal the reason for the inhibited crystallization of second phase. In general, the crystallization rate constant K increases exponentially with the temperature, indicating that the crystallization is a thermally activated process. It can be computed with E_c and K₀, and mathematically expressed as

• (2)

The values of K for the first CP (lower temperature, In₂S₃ phase) and the second CP (higher temperature) are listed in Table 3. The K value of 6.64 × 10⁻⁸ s⁻¹ for the second CP is ~ 5 orders of magnitude smaller than that of the In₂S₃ phase, suggesting a strictly inhibited crystallization mechanism of the second phase. Thus, controllable crystallization to transparent chalcogenide glass–ceramics with sole In₂S₃ crystallites can be achieved.

Table 3. Crystallization Kinetics Parameters for 80GeS₂·20In₂S₃ Glass

CPs	Ec (KJ/mol)	K0 (s^−1)	K402°C (s^−1)
First (lower temperature)	331.58	9.89 × 10^22	2.19 × 10^−3
Second (higher temperature)	482.57	1.44 × 10^30	6.64 × 10^−8

Figure 8. Plots of vs (1000/T_c).

IV. Conclusion

In this study, the glass–ceramics were obtained from 80GeS₂·20In₂S₃ chalcogenide glass by appropriate heat treatment. The obtained glass–ceramics are highly transparent in IR region. Mechanical properties such as hardness and resistance to crack propagation are improved by generating crystallites. The XRD results showed that only In₂S₃ crystals were precipitated inside the glassy matrix, and the evolution of the DSC curves for the glass–ceramics indicates that the precipitation of In₂S₃ crystal phase is responsible for the first (lower temperature) CP. Furthermore, the precipitation of crystal phase associated with second CP is inhibited. The kinetic parameters for crystallization such as E_c, K₀, and K, were calculated from a series of DSC curves collected with variable heating rates. The computed results showed that the K value of 6.639 × 10⁻⁸ s⁻¹ for the second CP is ~ 5 orders of magnitude smaller than that of the In₂S₃ phase, illustrating an inhibited crystallization mechanism of the second phase. This research also indicates that controllable crystallization of 80GeS₂·20In₂S₃ chalcogenide glass can be easily achieved.

Acknowledgments

This work was partially supported by the International Science & Technology Cooperation Program of China (grant no. 2011DFA12040), National Program on Key Basic Research Project (973 Program) (grant no. 2012CB722703), the NSFC (grant no. 61108057), the Zhejiang Provincial NSF (grant nos. R1101263 and Y4110322), the NSF of Ningbo City (grant no. 2011A610091), and was also sponsored by K. C. Wong Magna Fund in Ningbo University.