Optimization of Gas Ambient for High Quality β-Ga₂O₃ Single Crystals Grown by the Optical Floating Zone Technique

5N purity Ga₂O₃ powder (ABSCO Ltd, U.K) was used as the starting material for the preparation of feed and seed rods. The Ga₂O₃ powder was packed tightly in a latex balloon with uniform pressure. The air inside the balloon was removed using a vacuum pump and a hydrostatic pressure of ∼ 70 MPa was applied to densify the rod uniformly. The compacted rods were sintered in a box type furnace at 1450°C for 24 hours. The crystal growth was carried out in an optical floating zone furnace (Fig. 1). This furnace has 4 halogen lamps of 1 kW power each that are focused using four elliptical mirrors at the center of the furnace. Both the feed and seed rods can be independently rotated (usually in opposite directions) for better uniformity. The rotation speed was varied from 5-20 rpm. The crystal was pulled at linear speeds ranging from 5-20 mm/hr, in the downward direction.

Zoom In Zoom Out Reset image size

Initially, the as-sintered rods were used as feed and seed to grow the Ga₂O₃ crystal. It was observed that the grown crystal had several cracks, due to the polycrystalline seed which resulted in multiple nucleation sites, as shown in 2a. However, the use of an oriented seed crystal helped in getting much better results. Since no high-temperature glues can be used to attach the seed crystal to the rod, a (100) oriented seed crystal was first cleaved from a polycrystalline sample grown earlier, and then attached to the rod from top side by melting the seed rod (Fig. 2b). Then the seed rod with the oriented seed Ga₂O₃ crystal along [100]-direction was used for the growth of Ga₂O₃ single crystal. After attaching the single crystalline seed with the (100)-plane, the seed rod was fixed at the bottom and the feed rod was hung from the top. For the growth, the top portion of the feed rod was melted and gently attached to the (100)-oriented seed. Although, we started with the oriented seed crystal, we performed a necking process by translating the seed relatively faster compared to the feed, so that only one grain grows into a larger crystal. The growth details and the growth parameters are given in Table I.

Zoom In Zoom Out Reset image size

Another factor limiting the growth of long Ga₂O₃ crystals is the formation of bubbles inside the molten zone due to decomposition of Ga₂O₃ especially when growing in a 100% oxygen ambient. It has been previously observed that Ga₂O₃ starts decomposing above 1200°C, with the rate of decomposition increasing with temperature.¹⁶ The decomposition reaction¹⁶ is given by

This disturbs the growth process since the gaseous products due to decomposition of the Ga₂O₃ form a bubble inside the molten zone (Fig. 2c). In general, since the oxygen partial pressure is typically kept very high while growing oxides, such bubbles cannot easily diffuse out of the melt into the ambient. The gases accumulate inside bubbles over time and increase its size. This causes instability in the molten portion while rotating and beyond some critical size, the bubbles burst and as a result, the feed and seed rods get detached with each other. Fig. 2d shows the optical image of the crystal just taken after such a bubble explosion. This interrupts the growth of long Ga₂O₃ crystals, to a large extent. We find that the ambient atmosphere plays an important role in controlling the formation of bubbles. If air, or a mixture of argon and oxygen is used as the ambient the partial pressure of the oxygen is reduced. As a result, it is easier for any bubble that is formed to diffuse out of the molten zone. However, there is another critical trade-off. The reduction of the ambient oxygen partial pressure increases the decomposition of Ga₂O₃ resulting in gallium oxide powder being deposited on the quartz tube (Fig. 3). This whitish deposit reduces the transparency of the quartz tube, diminishing the lamp power reaching the focus and hence necessitates more power to the lamp. If oxygen partial pressure is very less or zero the deposition of powder on the quartz tube can be so severe that the opacity renders the crystal growth impossible. As a result, a proper optimization of the air or argon/oxygen mixture, and the flow rate is needed to grow long, good quality crystals. Fig. 4 shows the optical image of Ga₂O₃ crystals grown in different gas ambient. It was observed that the quality of the crystal is better for a flowing gas environment compared to that of a closed ambient. The crystals grown in an oxygen rich ambient are transparent and colorless in nature. The presence of argon in the ambient creates oxygen vacancies in the crystal, which give it a bluish color which gets deeper with increasing argon concentration in the ambient.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Single crystal Ga₂O₃ is prone to cleave along the (100) plane. Some small pieces of Ga₂O₃ of diameter ∼ 1 cm and thickness 300 μm were mechanically cleaved from the Ga₂O₃ crystals grown under different conditions. The Ga₂O₃ samples were placed on top of a goniometer in front of a white X-ray source to obtain a Laue diffraction pattern. Fig. 5a shows the Laue diffraction pattern in the reflection geometry for the crystal grown under optimized growth conditions. The sharp spots in the Laue pattern confirm good crystallinity of the grown Ga₂O₃ crystal. The stereographic projections of the Ga₂O₃ crystal about various symmetry axes were simulated to match with the obtained Laue pattern to determine the orientation of the single crystal. Fig. 5b shows the simulated Laue pattern corresponding to (100) direction which matches with the experimentally obtained Laue pattern reasonably well. It confirms the sample was cleaved along (100) surface. The powder XRD pattern of the Ga₂O₃ crystal (Fig. 6b) also has sharp peaks corresponding to(400), (600), and (800) reflections showing evidence of the good crystallinity. To examine the structural quality further, single crystal XRD measurements (Rigaku SmartLab) were done on cleaved Ga₂O₃ slices. The rocking curve shown in Fig. 6b having an FWHM of ∼ 22 arcsec for the (400) reflection points to a high degree of crystallinity of the Ga₂O₃ single crystal, comparable to that measured for commercially-available (100)Ga₂O₃ substrates⁹ under similar instrumental conditions (see supporting information for details).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Polarization-dependent optical transmission, photoluminescence, and Raman scattering measurements were done to investigate the optical properties of the Ga₂O₃ crystals grown under different gas ambient. Fig. 7a compares the optical transmission of four Ga₂O₃ samples grown under different conditions, measured on a Cary 5000 UV-VIS-NIR spectrophotometer with unpolarized light in the wavelength range from 200 to 2500 nm. The transmission curve is flat beyond the band edge for the sample grown in the pure oxygen ambient. A slight decrease of the sub-band-gap transmission curve is observed for the Ga₂O₃ crystal grown in Air atmosphere. However, for the sample grown in the 90% Ar and 10% oxygen atmosphere, there is significant absorption in the visible and infrared range, as also evident from the bluish color of the crystal. This is attributed to the free carrier absorption caused by the oxygen vacancies in the Ga₂O₃ crystal. As expected, the sample grown in a pure argon atmosphere shows enhanced absorption. Due to the monoclinic crystal structure of β- Ga₂O₃ the optical transmission is anisotropic. Fig. 7b shows the polarization dependent transmission spectrum with 0° and 90° coresponding to E//b and E//c respectively . The fundamental absorption edge has different values for the incident light field (E) direction being parallel to the c-axis or b-axis of the crystal. At intermediate angles of the polarizer, the optical absorption curve shows two steps, arising from the two band edges along the two orthogonal crystal axes.¹⁷

Zoom In Zoom Out Reset image size

Room temperature photoluminescence measurements were carried out using 266 nm UV laser excitation. No band edge luminescence was seen, however, a broad defect luminescence band was observed (Fig. 7c) which can be deconvolved into several peaks (dashed) corresponding to emission from defect complexes emitting at ∼ 353 (UV), 389 (blue/violet), 430 (blue), 457 (blue) nm respectively. Among these peaks, the UV peak does not depend on the impurity concentration and sample history, therefore, it is reported to be arising due to the intrinsic transition corresponding to the recombination of acceptor-bound exciton .¹⁸ The blue emissions are reported to be due to the donor-acceptor recombination. The splitting of the blue peaks is assumed to be due to the complex structure of β-Ga₂O₃ having two crystallographic sites for gallium and 3 crystallographic sites for oxygen.¹⁹

Raman measurements were done at room temperature on the best-grown Ga₂O₃ sample (FZ17014) using a WITec α-300 monochromator with a 1800 rulings/mm grating and a 532 nm laser source. β-Ga₂O₃ has a space group symmetry C2/m. The unit cell consists of two types of gallium (Ga_I, Ga_II) atoms and three types of oxygen (O_I, O_II, O_III,) atoms. Hence the primitive unit cell of β-Ga₂O₃ consists of 10 atoms which results in 3 × 10=30 vibrational modes. Among the 30 vibrational modes, 3 are acoustic and 27 are optical. The irreducible representations for the acoustic and the optical vibrational modes are as follows.²⁰

Among the optical modes the A_g and B_g modes are Raman active and due to odd parity, the modes A_u and B_u are infrared active.²¹ Fig. 7d shows the Raman spectrum of the grown Ga₂O₃ single crystal with all 10 peaks corresponding to the A_g modes. The peaks corresponding to the B_g modes except for B⁽²⁾_g are very near to the peaks corresponding to A_g modes. So the B_g modes except the B⁽²⁾_g are not distinguishable. To identify A_g and B_g modes separately, polarization-dependent Raman measurements were carried out in the back-scattering geometry (Fig. 8). The relative orientation between the crystallographic axes (a, b, c) and the observed X, Y, Z axes is shown in the inset of Fig. 8a. The X-axis is taken along (100) direction and is perpendicular to the cleaved surface. The Y-direction is taken along (010)-direction. The angle between the crystallographic a and c axis is 103.7^o resulting in an offset of 13.7^o between crystallographic c and Z axis. The excitation of the vibrational modes depends on the polarization selection rules. The A_g modes are allowed in XX, YY, ZZ and ZX configurations, and the B_g modes are allowed in XY, YZ configurations.²⁰ The first letter of the configuration denotes the polarization of the incident light and the second one denotes the polarization of the scattered light. In our Raman measurements, the direction of propagation of the incident and scattered light is along X and -X directions, respectively. Fig. 8 shows the polarization dependent Raman spectra taken in different polarization configurations. Figs. 8a and 8b are the Raman spectra taken in YY and ZZ configurations, respectively. As a result, only A_g modes are activated in this configuration. On the other hand, Figs. 8c and 8d are the spectra taken in YZ and ZY configurations. In this configuration, the intensity of the B_g modes is enhanced while the intensity of the A_g modes is suppressed, thus allowing the B_g modes to be seen clearly.

Zoom In Zoom Out Reset image size

The effect of the gas ambient on the growth of β- Ga₂O₃ was studied. A pure oxygen ambient atmosphere leads to the formation of bubbles inside the melt while a pure argon ambient leads to deposition of gallium oxide powder on the walls of the quartz tube that reduces the power delivered to the melt. Both extreme conditions are not desirable for the growth of high-quality β- Ga₂O₃. The optimized ambient of flowing 90% argon and 10% oxygen mixture allows high-quality light bluish color long β- Ga₂O₃ crystal to be grown. The structural and the optical properties of β- Ga₂O₃ and their dependence on the gas ambient and growth conditions were also studied.

The authors thank Prof. Sandip Ghosh for the help and discussion in the optical measurements. The authors are also thankful to Nilesh Kulkarni, Mahesh Gokhale and Vilas Mhatre for discussion and help in XRD measurements, Kushal Darje and TIFR central workshop for making accessories for the growth system, and Amit Shah for the help in electrical measurements.

Emroj Hossain 0000-0002-6556-0856

Rajib Mondal 0000-0001-7395-6634

A. Azizur Rahman 0000-0002-9006-474X

Arumugam Thamizhavel 0000-0003-1679-4370

Arnab Bhattacharya 0000-0003-4187-5025