GOX 2022 Session TM-TuP: Theory, Modeling and Simulation Poster Session

β-Ga₂O₃ has attracted high interest in the last decade and emerged as a promising material for next-generation power, GHz power switching, and RF applications. The large bandgap opens its possibilities for potential space applications, as the electron-hole pair generation in radiation (LET) is inversely proportional to the bandgap. Under radiation conditions in space, single event effects (SEE) are a potential reliability issue for devices in space-based RF and power systems. Previous reports have explored SEE burnout in SiC MOSFETs and thus SEE could also be an issue in Ga₂O₃ based devices. To date, there have been no reports of study of SEE in Ga₂O₃ based devices. In this work, we present the SEE effect in Ga₂O₃ Schottky diodes by simulation. 2-D simulations were performed using Silvaco TCAD to investigate the effects of ionizing radiation (measured in LET) under various bias conditions to understand the failure mechanism in the device. Breakdown simulations (V_br= 1400 V) show that the anode edge has the highest electric field; thus, it is chosen as the ion strike location in simulations. Under the radiation conditions, the time-dependent simulations show that the current recovers even at 1000 V and 40 MeV/mg/cm² LETs; even though the instantaneous peak field exceeds 8 MV/cm. It is noted that higher LETs take a longer time to recover. The heat generated due to ion strikes can also lead to the failure of devices due to thermal effects. The ion strike results in an instantaneous high current density within the device and leads to excessive Joule heating. The thermal effects were studied in Ga₂O₃ diodes with a diamond coating. Below radiation conditions of V=500V and LET=5 MeV/mg/cm² the device does not show any signatures of single event burnout (SEB) including thermal effects. At higher voltage bias and high LETs, there are signatures of possible thermal runaway where the temperature rises to a high value but only for a short amount of time and then recovers. The total energy dissipated under the ion strike condition is also calculated by integrating the power density spatially across the ion track. Our results indicate that under similar radiation conditions the power dissipated in b-Ga₂O₃ Schottky diodes is lower compared to the experimental data of SiC Schottky diodes. Future work includes investigating the threshold condition of SEB in b-Ga₂O₃ Schottky diodes for potential applications in harsh radiation conditions.

Polarization-dependent photoresponsivity measurements on (001) β-Ga₂O₃ are modeled by calculating the anisotropic absorption process via the dielectric tensor¹. Quasiparticle band structure calculations show that three different transitions from topmost group of O^2- 2p valance bands to the lowest Ga³⁺ 4s conduction band can occur that contribute significantly to the excitonic spectra². A linear polarizer is utilized in the excitation path to generate photocurrent in (001) β-Ga₂O₃ Schottky diode. Three excitonic transitions corresponding to P_x, mixed P_x/P_z and P_y valance band sates are observed as peaks in the measured photoresponsivity spectra at 4.9eV, 5.2eV and 5.5eV². The intensity of the peaks depends on the linear polarization of the excitation.

This strongly anisotropic absorption process in β-Ga₂O₃ cannot be modeled using a generic absorption coefficient; the Beer-Lambert law is not strictly accurate in an anisotropic dielectric. To model anisotropic absorption, we solve the electromagnetic wave equation using the Berreman³ matrix and the measured dielectric tensor of β-Ga₂O₃¹. The Poynting vector versus the linear polarization of the excitation with respect to the a- and b- axes and absorption depth is calculated and used to determine the generation rate and the photoresponsivity spectrum of β-Ga₂O₃. The modeled photoresponsivity matches the experimentally measured 3-peak spectrum and its variation with polarization angle. The simulations confirm that the photon flux in β-Ga₂O₃ over the energy range of interest (4-6eV) does not decay exponentially with depth over the photocarrier collection region, in violation with the Beer-Lambert law.

The exciton Franz Keldysh effect is observed in the photoresponsivity spectra, where exciton absorption peaks red shift with reverse bias⁴. The magnitudes of red shift for the three photoresponsivity peaks are inversely related to the corresponding interband transition energies i.e. the smallest energy transition peak shows the highest amount of red shift. We will discuss the polarization anisotropy of the field-dependent red-shift in terms of the exciton wave function anisotropy and its impact on the Stark shift.

¹C. Sturm et al., APL Mater. 3, 106106 (2015).
²J. Furthmüller et al., Phys. Rev. B 93, 115204 (2016).
³D. W. Berreman, J. Opt. Soc. Am. 62, 4, pp. 502-510 (1972).
⁴M. M. R. Adnan et al., Phys. Rev. Appl. 16, 034011 (2021).

Although Ga₂O₃ is widely believed to be the most promising ultrawide-bandgap semiconductors, its inability to be p-type doped hampers its future applications. Recently, other oxide materials have emerged as potential competitors to Ga₂O₃, but their propensity for hole conductivity is less well known. Here the stability of hole polarons in a set of ultrawide-bandgap oxides (Ga₂O₃, Al₂O₃, ZnGa₂O₄, MgGa₂O₄, LiGaO₂ and GeO₂) is examined and compared, both in pristine material and in the presence of acceptor impurities. Holes spontaneously self-trap in all oxides investigated here, with varying stability. Acceptor impurities further stabilize these trapped holes, leading to large acceptor ionization energies. Hole trapping also leads to characteristic distortions and distinct optical transitions, which may explain some experimentally-observed signals. These results indicate that achieving p-type conductivity in any of these oxides is unlikely.

This work was supported by the ONR/NRL 6.1 Basic Research Program.

The ultra-wide bandgap (UWBG) of Ga₂O₃ allows it to achieve over 10-times lower intrinsic carrier concentration than Si permitting Ga₂O₃ devices to operate at much higher temperatures. However, its low thermal conductivity and the associated self-heating could cause the device to exceed its safe operating temperature as prescribed by gate dielectric and packaging material limitations. The objective of this study is to develop a physics-based simulation and computation framework for the co-design of Ga₂O₃ devices and packaging.

Table I outlines the benefits and limitations of different models that are conventionally used for device and packaging design. In the design of the package and the micro-/nano-scale device structures, it is critical to observe the interactions between the two [1] [2]. The traditional package-level FEA simulation usually assumes a uniform power density and junction temperature in the devices, while neglecting the temperature variations in the sub-micron device structures; this variation can be up to tens of kelvin under high-power device operations. Conversely, the typical physics-based TCAD simulation accurately models the electrothermal behaviors within the device but the high computational power required (due to the large difference in length scales between the electrically active regions and the thermal diffusion regions) limits their use when package components need to be considered. These models instead simplify the packaging into nominal boundary thermal resistances, neglecting larger packaging elements in the heat flow path that are critical to overall thermal management and reliability [3].

The simple bottom-side cooled diode shown in Fig. 1. was modeled in ANSYS Workbench, Silvaco 2D TCAD, and Silvaco 3D TCAD. As can be seen in Fig. 2., there were stark differences in the heat distribution between the models. These hot spots are severe reliability liabilities and would need to be accounted for in the package design.

In order to build a platform that accounts for the additional electro-thermal effects while still being practical and efficient, a series of models were built to integrate the physics-based material/device-level simulation with a package-level FEA simulation. These models were then evaluated against more traditional methods of device-package simulation, seen in Table II, to understand the potential benefits of such a method. In end effect, this method would guide both device and package design with the hope of identifying and optimizing the thermal and electric field management needs.

[1] IEEE EDL, vol. 90, no. 6, p. 1065-1076, 2002.

[2] IEEE WiPDA, 2018 p. 287-294.

[3] APL, vol. 115, no. 173508, 2019.

The surface crystallographic and electronic structure of bulk single crystals of transparent, wide gap semiconductor gallium oxide β-Ga₂O₃(010) has been studied using Spot Profile Analysis Low-Energy Electron Diffraction (SPA-LEED) and X-ray Photoelectron Spectroscopy (XPS). The XPS measurements show typical spectra for Ga₂O₃(010). The surface structure of Sn doped single crystal β-Ga₂O₃ Using quantitative LEED I-V. The surface shows a (1 x 1) LEED pattern without reconstruction. The comparison between experimental I-V curves and theoretical LEED simulations using Green’s function formalism indicates the formation of a single surface termination layer. There are significant displacements in the first topmost slab surface, moreover, experiment and theory suggest important atomic rumpling between gallium and oxygen atoms in the topmost surface layers. The surface structure agrees with that predicted by first-principles density functional theory calculations and X-ray photoelectron diffraction.