The unique response of amorphous ionic oxides to changes in oxygen stoichiometry is investigated using computationally intensive ab initio molecular dynamics simulations, comprehensive structural analysis, and hybrid density-functional calculations for the oxygen defect formation energy and electronic properties of amorphous In2O3-x with x = 0-0.185. In marked contrast to nonstoichiometric crystalline nanocomposites with clusters of metallic inclusions inside an insulating matrix, the lack of oxygen in amorphous indium oxide is distributed between a large fraction of undercoordinated In atoms, leading to an extended shallow state for x < 0.037, a variety of weakly and strongly localized states for 0.074 < x < 0.148, and a percolation-like network of single-atom chains of metallic In-In bonds for x > 0.185. The calculated carrier concentration increases from 3.3 x 1020cm-3 at x = 0.037 to 6.6 x 1020 cm-3 at x=0.074 and decreases only slightly at lower oxygen content. At the same time, the density of deep defects located between 1 and 2.5 eV below the Fermi level increases from 0.4 x 1021 cm-3 at x = 0.074 to 2.2 x 1021 cm-3 at x = 0.185. The wide range of localized gap states associated with various spatial distributions and individual structural characteristics of undercoordinated In is passivated by hydrogen that helps enhance electron velocity from 7.6 x 104 to 9.7 x 104 m/s and restore optical transparency within the visible range; H doping is also expected to improve the material's stability under thermal and bias stress.
J. E. Medvedeva et al., "Metallic Networks and Hydrogen Compensation in Highly Nonstoichiometric Amorphous In₂O₃₋ₓ," Physical Review Materials, vol. 6, no. 2, article no. 25601, American Physical Society (APS), Feb 2022.
The definitive version is available at https://doi.org/10.1103/PhysRevMaterials.6.025601
International Standard Serial Number (ISSN)
Article - Journal
© 2022 American Physical Society (APS), All rights reserved.
01 Feb 2022
The authors acknowledge the support from the National Science Foundation (NSF) DMREF Grants No. DMR-1729779 and No. DMR-1842467. The computational resources were provided by Missouri S&T and NSF-MRI Grant No. OAC-1919789.