2.1 Fabrication of Glass and Crystalline Samples
As mentioned above, SbSI composition does not form glass readily, but the addition of GeS2 facilitates glass formation.[14, 15] Nevertheless, we attempted to form pure SbSI glass by emulating the conditions previously established for the SbSI–GeS2 system. An elemental powder mixture (about 5 g) of stoichiometric SbSI composition was prepared directly from Sb, S, and I elemental powders. The mixture was sealed in an evacuated 11 mm inner diameter and 10 cm long silica (quartz) tube under 10−2 torr pressure. To prevent explosion, the ampoule was slowly heated in a rocking furnace at 1 K/min to 128°C, 250°C, 450°C, and 650°C, and kept for an hour at each temperature. It was then heated to the final step at 730°C and maintained there for 12 h. After slowly cooling down to 650°C, the ampoules were quenched in cold water to form glass.
To eliminate stresses arising from nonuniform cooling during quenching, the as-cast samples were annealed in a convection oven preheated to 70°C. After 2 h, the ampoule was slowly cooled to room temperature and then cut to obtain solid castings. We found that even though stoichiometric SbSI composition had been considered to be outside of the conventional glass-forming region of Sb–S–I system, when the batch size was small and the quench rate was sufficiently fast (~100K/s–200K/s or faster), we could obtain glassy samples. Visual inspection and further X-ray powder diffraction (XRPD) analysis showed that the regions of sample next to ampoule walls formed the amorphous phase, whereas the interior was partly or fully crystalline (see Fig. 1). Decreasing the ampoule temperature from 650°C to 500°C before quenching in cold water increased the size of the glassy region but did not eliminate completely the crystalline region. It was determined that the low thermal conductivity of the ampoule walls and the SbSI compound limited the cooling rates within the SbSI melt. To achieve faster heat transfer, 6 mm ID ampoules were used to increase the surface area relative to the volume. The as-cast samples [see Fig. 2(a)] obtained in this manner were completely glassy as shown by X-Ray Powder Diffraction (XRPD).
To make SbSI crystal as a reference, we prepared the melt as before, except that the ampoules were cooled slowly from 730°C to room temperature at the last step of preparation. XRPD analysis shows that the druse of needles, as seen in Fig. 2(b), is simply the SbSI crystalline phase; there are no signs of other likely phases such as Sb2S3 and SbI3.
For energy-dispersive X-ray (EDS) analysis a small “needle” shaped crystal with length 5 mm and thickness 0.3–0.4 mm was selected and was mounted in epoxy resin. Samples were polished successively using 600-, 1000- and 1200-grit SiC abrasive papers with water as a polishing medium, followed by polishing with the suspension of Al2O3 powder of 3, 1, 0.3, 0.1, and 0.05 μm particle sizes.
The powdered sample was divided into fractions of different particle sizes by sieving. We selected six sieves with different mesh sizes (66, 178, 251, 354, 500, and 710 μm) and placed them on the base of the shaker in increasing order. Glass sample was crushed in an agate mortar and transferred to the upper sieve with 710 μm mesh size. After shaking the sieve's stack for 5 min, the particles from the upper sieve were crushed again and the shaking operation was repeated.
A CW diode laser operating at wavelength λ = 520 nm was used for writing crystal lines. The laser beam was first focused on the polished surface of the glass sample by a microscope objective (numerical aperture 0.75) to a spot of 5 μm diameter. We had previously observed selective evaporation of SbI3 from SbSI surface under irradiation from an Argon ion laser operating at 488 nm whose intensity was modulated by a combination of a wave plate and polarizer.[17, 18] In an attempt to avoid this undesirable change of composition and gain a more precise control over the laser intensity, the 520 nm diode laser was employed. This laser provided the ability to precisely control the intensity through the application of an analog voltage, and simultaneously eliminated a significant amount of the noise present in the Argon laser. Thus, we were able to create the spots by slowly (5–10 s) ramping the power density from 0 to 0.05 mW/μm2. This procedure minimized surface evaporation of SbI3 and induced crystallization under the surface of glass sample. In the second stage, the lines were fabricated by translating the sample at a speed of 0.1 μm/s while maintaining the final power density of 0.05 mW/μm2.
2.2 Methods of Characterization
XRPD and DSC were used for identifying crystalline phases and phase transformations, respectively. The XRPD analyses were performed on a Rigaku “MiniFlex II” diffractometer (Tokyo, Japan). The diffraction data were recorded between θ = 10° and 50°, with 0.02° scan step and 0.5 s step time. The glass transition (Tg) and maximum crystallization (Tc) temperatures were determined with a DSC system (model Q2000; TA Instruments, New Castle, DE). The measurements were conducted with a heating rate from 3 to 20 K/min on powders with five different-sized particles: 66–178, 178–251, 251–354, 500–710 μm and >2 mm from room temperature to 250°C.
The chemical composition of glass and crystal samples was determined by an EDS spectroscopy device attached to a scanning electron microscope (SEM) Hitachi 4300 SE (Dallas, TX) in low vacuum environment to eliminate the charging effects usually observed on insulating samples. For EDS analysis, an acceleration voltage of 20 kV and water vapor pressure of 30 Pa were chosen. The spectra were collected and analyzed using EDAX-Genesis software package (EDAX Inc., Mahwah, NJ). The parameters for data acquisition (time, full scale for intensity, and pulse processing time) were kept the same for all the samples. The EDS spectra for crystal and glass samples are shown in Fig. 3, which compare Sb, S, and I in crystal and glass samples. The chemical composition, as calculated following the ZAF procedure [corrections for atomic number effects (Z), absorption (A) and fluorescence (F)] for the three major elements, is summarized in Table 1.
Table 1. Energy-Dispersive X-Ray Data from Different Regions in Crystal and Glass
|Sample||Sb (at.%)||S (at.%)||I (at.%)|
|Crystal||34.9 ± 2.0||32.0 ± 2.0||33.1 ± 2.0|
|34.8 ± 2.0||31.8 ± 2.0||33.4 ± 2.0|
|35.2 ± 2.0||31.4 ± 2.0||33.4 ± 2.0|
|Glass||35.7 ± 2.0||33.4 ± 2.0||30.9 ± 2.0|
|35.4 ± 2.0||33.6 ± 2.0||31.0 ± 2.0|
|35.7 ± 2.0||32.9 ± 2.0||31.4 ± 2.0|
The laser-induced lines were observed with scanning electron (Hitachi 4300 SE) and optical microscopies. The optical images were received using a PAX-IT camera attached to an Olympus BH-2 light optical microscope with PAX-IT digital imaging software (Center Valley, PA). Sample topography was determined by atomic force microscopy (Solver Next AFM/STM, NT-MDT, Zelenograd, Russia) with a conductive Pt-coated NSG10 tip in semicontact mode. The amorphous or crystalline nature of the lines was checked by electron back-scatter diffraction analysis.