Khandakar Aaditta Arnab1,Isaac Maxfield1,Chennyung Lee2,Elif Ertekin2,Joel Varley3,Ymir Frodason4,Michael Scarpulla1
University of Utah1,University of Illinois at Urbana-Champaign2,Lawrence Livermore National Laboratory3,Centre for materials Science and Nanotechnology Physics4
Khandakar Aaditta Arnab1,Isaac Maxfield1,Chennyung Lee2,Elif Ertekin2,Joel Varley3,Ymir Frodason4,Michael Scarpulla1
University of Utah1,University of Illinois at Urbana-Champaign2,Lawrence Livermore National Laboratory3,Centre for materials Science and Nanotechnology Physics4
β-gallium oxide (β-Ga<sub>2</sub>O<sub>3</sub>) is of intense current interest because of its ultra-wide bandgap, high critical field, and availability of melt-grown substrates. Point defects and complexes determine the properties of bulk crystals as well as epitaxial layers, thus, predictive models of defect concentrations under various impurity and processing scenarios are of very high value. First-principle calculations of defect energetics have provided critical insights into the defect system in β-Ga<sub>2</sub>O<sub>3, </sub>but translating computed enthalpies into defect concentrations corresponding to real-world crystal growth requires additional steps. Material processing in terms of growth or annealing typically controls the sample’s thermochemical trajectory in terms of temperatures and partial pressures, while computational papers frequently present results holding chemical potentials constant.<br/>Here we report quantitative modelling of equilibrium defect concentrations in Ga<sub>2</sub>O<sub>3</sub>, considering especially the temperature dependence of the bandgap and temperature-dependent chemical potentials from the Ga-O binary system’s known thermochemistry. Additionally, we compute results for realistic sample types such as Fe- or Sn-doped wafers accounting for the fixed concentrations of these impurities as opposed to their fixed chemical potentials. Results are presented for various background n-type doping and for equilibrium and quenching, corresponding respectively to 0 or infinite cooling rates. We find significant departures from prior simpler predictions, especially in the case of the bandgap temperature dependence which tends to suppress V<sub>Ga</sub>. We compare our predicted results to experimental cases such as annealing in O<sub>2</sub> or Ga<sub>2</sub>O vapors.<br/>Finally, to give semi-quantitative insight into defect concentrations expected in finite-sized samples subjected to finite cooling rates without full-fledged defect reaction-diffusion simulations, we introduce the concept of generalized quenching as a 3<sup>rd</sup> type of computation. At the heart of generalized quenching is the insight that, because of their different diffusion constants, different types of defects located at different distances from free surfaces will be “frozen-in” at different temperatures. By combining the correct series of equilibrium and quenching calculations, it is possible to predict defect concentrations present in real-world samples e.g. as a function of radius within a boule or for thin films of different thicknesses. We compare these results to the known phenomena from bulk crystal growth, indicating differences in carrier density between the center and periphery of CZ-grown boules.