Defect is one of the issues that limit the present performance of β-Ga₂O₃ devices. For example, several defects have been observed at 0.6, 0.8. and 1.1 eV below the conduction band edge of β-Ga₂O₃, which are considered to affect doping compensation, leakage current, and threshold stability in transistors. Conventional detection of β-Ga₂O₃ defects (in various forms such as Hall, conductivity, admittance spectroscopy, and deep-level transient spectroscopy (DLTS)) is accomplished by inspecting the electrical charge response, which is based on the Arrhenius behavior of the carrier emission rate from a defect determined by the activation energy E_a and the attempt-to-escape frequency ν₀. All thermally activated electrical charge response measurements are conventionally analyzed by the Arrhenius plot procedure, where one fits the Arrhenius plot of ln(ν) versus T^-1 to a line and extract E_a from the slope and ν₀ from theintercept. Improvement of the measurement expediency for extracting E_a and ν₀ is desirable to understanding their physiochemical origins and devising mitigation strategies in β-Ga₂O₃ material and device engineering.

We investigate a ~0.8 eV defect in β-Ga₂O₃ using a technique that offers substantial improvement over the conventional DLTS technique, specifically in the analytical processing of electrical signal and the extraction of E_a and ν₀. The technique bypasses both the rate-window treatment and the Arrhenius plot. First, only the raw capacitance transients in the time domain are needed, which can be readily acquired by general-purpose instruments such as impedance analyzers and lock-in amplifiers. Next, the capacitance transients are projected between the temperature and time domains, as well as to the E_a and ν₀ domains. Extraction of E_a and ν₀ is accomplished by matching the projected and experimental capacitance transients to each other. The efficient utilization of information from the 2D temperature-time domain allows operation in a smaller temperature/voltage range and extraction of the temperature and electric-field dependence of E_a and ν₀.

Li₂O-Ga₂O₃ is an oxide system of broader interest. LiGaO₂ (LGO) and multiple phases of Ga₂O₃ (GO) are ultra-wide bandgap metal oxides for future electronic and optoelectronic applications [1], and both LGO, which is orthorhombic, and the orthorhombic phase of GO are expected to be piezoelectric due to the lack of inversion symmetry [1]. While both GO and LGO have recently been identified to most likely trap holes which makes the achievement of sufficient p-type conductivity difficult [2], LGO is particularly promising as a substrate for heteroepitaxial growth of GaN due to very small lattice mismatch (<1%), and a composite LGO/β-GO substrate has also been demonstrated [3]. Here, we provide a thorough study of the fundamental optical and phonon mode properties of high-quality single-crystals of LGO using generalized spectroscopic ellipsometry in combination with hybrid-level density functional theory calculations to investigate the optical properties in the mid- to far-infrared spectral range. From this, all 33 infrared-active pairs of transverse and longitudinal optical phonon modes are observed. We derive the anisotropic midband gap indices of refraction and static dielectric constants.

[1] A review of band structure and material properties of transparent conducting and semiconducting oxides: Ga₂O₃, Al₂O₃, In₂O₃, ZnO, SnO₂, CdO, NiO, CuO, and Sc₂O₃, Joseph A. Spencer, Alyssa L. Mock, Alan G. Jacobs, Mathias Schubert, Yuhao Zhang, and Marko J. Tadjer , Applied Physics Reviews 9, 011315 (2022)

[2] Self-trapped holes and polaronic acceptors in ultrawide-bandgap oxides, John L. Lyons, Journal of Applied Physics 131, 025701 (2022)

[3] Composite substrate LiGaO₂ (0 0 1) β-Ga₂O₃ (1 0 0) fabricated by vapor transport equilibration, Jungang Zhang, Changtai Xia, Shuzhi Li, Xiaodong Xu, Feng Wu, Guangqing Pei, Jun Xu, Shengming Zhou, Qun Deng, Wusheng Xu, Hongsheng Shi, Mater. Lett. 60. 3073-3075. (2006)

Zinc gallate (ZnGa₂O₄) is shown to be a promising alternative to gallium oxide. This is due to the material’s larger bandgap of 5.27(3) eV, compared to that of β-Ga₂O₃ (5.04 eV), which is linked to a higher Baliga’s figure of merit. [1,2] ZnGa₂O₄ also contains an isotropic structure, which is advantageous compared to both the monoclinic β-phase and uniaxial α-phase of Ga₂O₃ in simplifying device design. [2,5] Additionally, ZnGa₂O₄ growth has rapidly developed to where bulk single crystals can be melt-grown with controllable n-type conductivity. [4]

In this work, the optical properties of ZnGa₂O₄ are modeled using a spectroscopic ellipsometry approach at temperatures between 22°C and 600°C, where material properties drastically change in elevated temperatures. At each 50°C interval a Cauchy dispersion equation is applied to the transparent region of the data where the refractive index and high-frequency refractive index is derived. Furthermore, a critical point model is implemented across the spectral range of 1 eV to 6.5 eV. This allows for the determination of the bandgap, which is found to red-shift linearly with temperature with a slope of -0.72(4) meV K^-1, resulting from the thermal expansion of the lattice. [3] The linear decrease in the bandgap energy when exposed to increasing elevated temperatures is in congruence with behavior shown by common wide bandgap metal oxides. In particular, the reduction of bandgap width as a function of temperature is comparable to that of the ultrawide bandgap material β-Ga₂O₃, further justifying ZnGa₂O₄ as a suitable high-power device material. [2,3]

References:

[1] M. Hilfiker et al. Appl. Phys. Lett. 118, 132102 (2021).

[2] A. Mock et al. Appl. Phys. Lett. 112, 041905 (2018).

[3] M. Hilfiker et al. Appl. Phys. Lett. 120, 132105 (2022).

[4] Z. Galazka et al. APL Materials 7, 022512 (2019).

[5] S. J. Pearton et al. Appl. Phys. Lett. 5, 011301 (2018).

Large interest in Ga₂O₃ originates from the possibility to build devices with high breakdown voltage. Understanding electronic defects is essential to utilize the material. As β-Ga₂O₃ is monoclinic, the effect of the low symmetry needs to be studied. Electron (paramagnetic) spin resonance (EPR) spectroscopy gives access to local site symmetry of spin-carrying defects. Deploying THz ellipsometry [1], we can measure high-field EPR at arbitrary variable frequency by reflection a free beam [2]. This allows true distinction of anisotropic g-factor and zero-field spin splitting, and hence examining the local site symmetry of electronic defects.

Iron incorporated on gallium sites can act as compensating acceptor and facilitate semi-insulating material. Here, we investigate the spin Hamiltonian of the neutral Fe³⁺ state with spin s=5/2. It is characterized by large zero-field splitting that differs for Fe on octahedral Ga_ii site (preferential) and Fe on tetrahedral Ga_i site. Different, partially incorrect, reports exist about the nature of the spin Hamiltonian [3,4]. In contrast to standard EPR measurements at X or Q band with limited access to allowed spin transitions, we obtain EPR scans in the frequency range 110-170GHz at magnetic field between 3 and 7T that capture all five resonances for each Fe site at the same time. Modeling the spin Hamiltonian reveals a slight anisotropy of the g-factor and shows that zero-field splitting up to fourth order is relevant. We will discuss how the monoclinic s=5/2 spin Hamiltonian differs from orthorhombic and/or s=3/2 approximations.

[1] P. Kühne et al., IEEE Trans. Terahertz Sci. Technol. 8(3), 257 (2018).
[2] S. Schubert et al., Appl. Phys Lett. 120, 102101 (2022).
[3] R. Büscher et al., Z. Naturforsch. 42a, 67 (1987).
[4] M. L. Meil’man, Sov. Phys. Sol. State 11(6), 1403 (1969).

The supplemental material features exemplary experimental data and the form of the spin Hamiltonian.

Efficient wafer size scaleup of quality (010) β-Ga₂O₃ substrates and epi development require extensive materials characterization. Over the past two years, the Applied Research Laboratory (ARL) Electronic Materials and Devices Department (EMDD) has partnered with Northrop Grumman’s SYNOPTICS division in support of the Air Force Research Laboratory (AFRL)’s initiative to produce 2-inch epi-ready (010) β-Ga₂O₃ substrates via the Czochralski (Cz) method. As part of this effort, ARL has developed a multitude of techniques specifically for characterizing β-Ga₂O₃ substrates at various stages of processing. In this poster, we highlight these techniques and describe the rapid feedback loop that ARL enables between those working on crystal growth, substrate fabrication, epi growth, and device processing within the expanding β-Ga₂O₃ community.

This poster covers defect mapping and identification, surface metrology, and x-ray characterization. Each of these areas provide essential information about the quality of β-Ga₂O₃ substrates for subsequent epi growth and ultimately device fabrication. We show how etch pit analysis is used to automatically map defects in up to 2-inch wafers in conjunction with the focused ion beam (FIB) approach to prepare cross-sections for imaging the defect structures, including nanopipes. White light interferometry/profilometry and atomic force microscopy (AFM) provide three-dimensional topography information about the wafers, i.e., after polishing, thermal annealing, or free etching. This poster also describes our method for in-situ high-resolution x-ray characterization during fabrication in which a decrease in full-width at half-maximum (FWHM) was correlated with removal amount to optimize β-Ga₂O₃ polishing. Finally, we discuss how x-ray characterization provides information about curvature and substrate/epi layer quality with grazing incidence x-ray diffraction (GIXRD) being utilized to provide extremely surface-sensitive monitoring.

Alloying β-Ga₂O₃ with Al₂O₃ to create (Al_xGa_1-x)₂O₃ enables ultra-wide band gap material deep into the UV. Here, photoluminescence (PL) spectra of Cr³⁺ dopant is compared between monoclinic single crystals β-Ga₂O₃, and 10 mol.% Al₂O₃ alloyed with β-Ga₂O₃, denoted β-(Al_0.1Ga_0.9)₂O₃ or AGO. Temperature dependent PL properties were studied for Cr³⁺ in AGO and β-Ga₂O₃ from 285 to 16 K. For β-Ga₂O₃at room temperature, the red-line emission doublet R₁ and R₂ occurs at 696 nm (1.78 eV) and 690 nm (1.80 eV) respectively along with a broad emission band at 709 nm (1.75 eV). For both AGO and β-Ga₂O₃ the R₁ line increases in intensity with decreasing temperature. This can be explained by the thermal depopulation effect of the R₁ state, which occurs with increasing temperature. The R₁ and R₂ lines of both materials were observed to blue-shift with decreasing temperature. Additional emission lines emerge at lower temperatures, with β-Ga₂O₃ showing more peaks than AGO. R₁ and R₂ peak parameters such as energy, intensity, width, and splitting were studied as a function of temperature, with significant differences between the two materials.

We have used X-ray Photoelectron Diffraction (XPD), low-energy electron diffraction (LEED), atomic force microscopy (AFM) and X-ray photoelectron spectroscopy (XPS) to determine the surface structure, chemistry and interplanar relaxation and rumpling in single crystal, Sn-doped b-Ga₂O₃ (010). XPD is a powerful technique, which combines information on local chemistry and atomic structure. By measuring the angular anisotropy of core level intensity one can, by comparison with simulations, deduce the local atomic and chemical environment around each type of emitting atom. The XPS measurements show typical spectra for stoichiometric Ga₂O₃(010). Annealing at 823 K yielded a well-ordered surface with sharp (1x1) low-energy electron diffraction (LEED) pattern. AFM shows unique surface termination with root mean square roughness of 0.1-0.15 nm. The XPD measurements were performed using a laboratory based setup with a monochromatic Al Ka (1486.7 eV) source and a high precision angular manipulator capable of scanning both polar (q) and azimuthal (f) angles. The XPD patterns collected for the Ga 2p_3/2 and O 1s emission. Surface interlayer relaxation up to 8% of the bulk interplanar distance and 0.11–0.14 Å rumpling are observed at the β-Ga₂O₃(010) surface. At the surface, the oxygen atoms shift toward the vacuum with respect to the gallium atoms. The rumpling decreases to zero and and the interplanar distance reaches the bulk value of 1.52 Å by the sixth atomic layer. The surface structure agrees with that predicted by first-principles density functional theory calculations which, in addition, suggest a significant band gap narrowing of ≈1 eV in the surface layer, due to surface states spatially localized on surface oxygen atoms of O_II type.

A. Pancotti et al. Phys. Rev. B102, 245306 (2020)

The rhombohedral alpha and monoclinic beta phases of gallium oxide both make promising candidates for ultra-wide bandgap semiconductor technology. Of particular interest are alloyed films and the evolution of anisotropic optical properties with respect to both alloy composition and strain induced effects. Here, we study alpha and beta phase (Al_xGa_1-x)₂O₃ via a combined density functional theory and generalized spectroscopic ellipsometry approach across a range of alloying. Infrared-active phonon properties, static dielectric constants and midband gap indices of refraction are quantified.[1,2,3] Strain and alloying effects are shown and compared to previous theoretical works.[4] Band-to-band transitions, excitons, and high-frequency dielectric constants are also investigated in the visible to vacuum-ultra-violet (VUV) spectral range.[5,6,7,8] We identify a switch in band order where the lowest band-to-band transition occurs with polarization along the ordinary plane in α-Ga­­₂O­₃ whereas for α-Al­₂O₃ the lowest transition occurs with polarization in the extraordinary direction. With this, we present the most comprehensive picture of optical properties’ evolution along composition and strain currently available.

[1] M. Stokey, et al., Phys. Rev. Materials 6, 014601 (2022)

[2] M. Stokey, et al., Appl. Phys. Lett. 120, 112202 (2022)

[3] The influence of strain and composition on the infrared active phonons in epitaxial β-(AlxGa1-x)2O3 deposited onto (010) β-Ga2O3; M. Stokey, et al., In Preparation

[4] R. Korlacki, et al., Rev. B 102, 180101(R) (2020)

[5] M. Hilfiker, et al., Appl. Phys. Lett. 118, 062103 (2021)

[6]Anisotropic dielectric function, direction dependent bandgap energy, band order, and indirect to direct gap cross over in α-(AlxGa1-x)2O3 (0≤x≤1); M. Hilfiker, et al., Appl. Phys. Lett. XX, XX (2022)

[7] M. Hilfiker, et al., Appl. Phys. Lett. 119, 092103 (2021)

[8] M. Hilfiker, et al., Phys. Lett. 114, 231901 (2019)

The mechanisms generating photoluminescence (PL) emissions from gallium oxide (Ga₂O₃) and aluminum gallium oxide (AGO) have been under intense scrutiny. In general, spectrally-resolved PL is used to characterize the defects leading to radiative recombination processes within a specific material. In this regard, the PL spectra for β-Ga₂O₃ has generally been deconvoluted in three emission bands: UV, blue, and green. So far, the intense debate in defining the defects and phenomenological explanations of electronic processes that cause Ga₂O₃ and AGO emissions have only explored point defects as the potential source for the PL emission leaving out whether extended defects could affect PL. Because of the strong electron phonon coupling, emission peak shapes from any defect are expected to be very broad and not have a simple functional peak shape; these attributes make it challenging to fit spectra uniquely. Because of this, there is little chance of directly assigning PL spectral features unambiguously to specific β-Ga₂O₃ point defects from PL alone.

Here, a systematic PL study on multiple series of β-Ga₂O₃ and AGO epitaxial thin films and bulk single crystals is performed. Spectrally-resolved PL, PL intensity mapping, scanning electron microscopy (SEM), atomic force microscopy (AFM), and transmission electron microscopy (TEM) were used along with literature to show that extended structural defects largely determine the PL emission from many samples of β-Ga₂O₃ and AGO. Homogeneous films with no extended defects or stacking faults and bulk crystals yield PL emission with a dominant UV peak, while samples of lesser crystalline quality exhibiting stacking faults, rotation domain boundaries, and other such extended defects do not exhibit a UV emission but rather exhibit blue.

Si-doped homoepitaxial (010) β-Ga₂O₃ samples yield homogeneous crystalline films with a low density of extended defects and an unshifting dominant UV emission in PL. A bulk (-201) β-Ga₂O₃ sample shows a dominant UV emission while heteroepitaxial and homoepitaxial (-201) β-Ga₂O₃ films show dominant blue PL emission (due to the films’ poor quality as seen in PL mapping, AFM and SEM). An AGO series shows consistent blue centered PL for AGO grown on both sapphire and β-Ga₂O₃ (also due to the films’ poor quality as seen in TEM, PL mapping, AFM and SEM). Lastly, an improved (reduced number of extended defects) 10% AGO film grown on (010) bulk β-Ga₂O₃ show a shift in the PL spectrum with a now UV dominant emission. PL mapping shows that areas of extended defects emit blue PL while the crystal film emits UV PL within the sample.

Selective doping by ion implantation is critical for small-scale device fabrication in wide-bandgap materials such as β-Ga₂O₃, requiring understanding of both lattice damage due to implantation and subsequent lattice recovery during thermal annealing. Carrier activation, mobility, and diffusion are known to be critically coupled to annealing temperature, time and ambient as well as intrinsic and extrinsic film properties. Electrical measurements provide one measure of annealing behavior, but are limited due to the need for direct metal contacts, the intrinsic spatial averaging, and the inability to directly measure lattice recovery or observe associated defects. To address these limitations, we present Raman spectroscopy and photoluminescence (PL) measurements of β-Ga₂O₃ implanted and annealed samples, evaluating their ability to local carrier activation and lattice recovery non-destructively.

Recent studies of bulk doped β-Ga₂O₃ [1] have identified additional Raman peaks in samples with carrier concentrations above the Mott criterion. To determine if these peaks are directly linked to carrier activation, we examined a wide range of Si-implanted (5x10¹⁹ cm^-3) and annealed samples using a laterally localized Raman probe. A peak at 285 cm^-1 was observed above the noise floor in several samples, with the intensity increasingly linearly with sheet carrier density (N_s). The effect of the lattice quality (recovery) on the relative intensities of other Raman peaks was also analyzed.

Previous β-Ga₂O₃ studies suggest that an increase in activated dopants decreases the total PL, due to the higher defect density and the resultant probability of electrons reaching non-radiative recombination centers [2]. PL behavior of the Si-implanted samples was thus measured as a function of lattice recovery with (N₂ ambient) and without (O₂ ambient) carrier activation. The Pl signal decreased dramatically after implantation, with partial recovery after thermal anneals and lattice damage recovery. Negative correlation was observed between the PL intensity and carrier activation. Correlations between PL and cathodoluminescence were also measured.

[1] Fiedler, A., Ramsteiner, M., Galazka, Z., and K. Irmscher. Raman scattering in heavily donor doped B-Ga2O3. Appl. Phys. Lett. 117, 152107 (2020)

[2] Shimamura, K., Villora, E.G., Ujiie, T., and K. Aoki. Excitation and photoluminescence of pure and Si-doped B-Ga2O3 single crystals. Appl. Phys. Lett. 92, 201914 (2008).