Atomic and Electronic Structure of the Al₂O₃/Al Interface during Oxide Propagation Probed by Ab Initio Grand Canonical Monte Carlo

Abstract

Identifying the structure of the Al2O3/Al interface is important for advancing its performance in a wide range of applications, including microelectronics, corrosion barriers, and superconducting qubits. However, beyond the study of a few select terminations of the interface using computational methods, and top–down, laterally averaged spectroscopic and microscopic analyses, the explicit structure of the interface and the initial stages of propagation of the interface into the metal are largely unresolved. In this study, we utilize ab initio grand canonical Monte Carlo to perform a physically motivated, unbiased exploration of the interfacial composition and configuration space. We find that at equilibrium, the interface is atomically sharp with aluminum vacancies and propagates in a layer-by-layer fashion, with aluminum excess in the oxide layer at the interfacial plane. Oxygen incorporation, aluminum vacancy formation, and aluminum vacancy annihilation are the building blocks of Al2O3 formation at the interface. The localized interfacial mid-gap states from under-coordinated aluminum atoms from the oxide and the immediate depletion of aluminum states near the Fermi level upon oxygen incorporation prevent oxygen dissolution ahead of the interface front and result in the layer-by-layer propagation of the interface. This is in sharp contrast to the ZrO2/Zr system, which forms interfacial sub-oxides, and also explains the favorable self-healing nature of the Al2O3/Al system. The occupied interfacial mid-gap states also increase the calculated n-type Schottky barrier heights. Additionally, we identify that interfacial aluminum core-level shifts linearly depend on the aluminum coordination number, whereas interfacial oxygen core-level shifts depend on long-range ordering at the interface. The detailed geometric and electronic insights into the interface structure and evolution expand our understanding of this fundamental interface and have important implications for the engineering and design of Al2O3/Al-based corrosion coatings with enhanced barrier properties, controllable transistor technologies, and noise-free superconducting qubits.

Computational Modeling for Cr and S Poisoning Pathways on La0.6Sr0.4Co0.2Fe0.8O3 Surfaces

“…The GCMC method is a physically motivated approach to explore the space of surface configurations (24)(25)(26). It enables the reconstruction process to move towards configurations that typically have lower surface energy.…”

Surface phase diagrams from nested sampling

“…31 The methods above guide the search for structures that minimize the zero-Kelvin surface energy as a function of composition, which are used as inputs for ab initio thermodynamics in the grand canonical ensemble. Recent studies have increasingly employed (Markov chain) Monte Carlo (MC) simulations in various ensembles, such as canonical, 32 grand canonical, [33][34][35][36] and semi-grand canonical, 37 to facilitate the discovery of operando surface phases. These simulations leverage the Metropolis-Hastings algorithm, allowing efficient sampling from specific probability distributions within defined thermodynamic constraints.…”