A review of simulation, characterization, and informatics in inorganic materials

Abstract

Ceramic materials exhibit a wide range of functional properties that are strongly influenced by their atomic structures, chemical bonding, and lattice defects, including interface and surface. A deep understanding and precise control of these factors are essential for advancing materials design. In this review, I summarize recent progress from my group on computational, experimental, and informatics approaches aimed at elucidating structures and defects, as well as guiding the development of inorganic materials. First, I describe defect simulations accelerated by machine learning. Bayesian optimization was shown to drastically reduce the computational cost of interface structure determination, identifying stable interface configurations with only a fraction of the calculations required by conventional approaches. In addition, adsorption properties at surfaces were effectively predicted using descriptors derived from the individual electronic structures of isolated systems, demonstrating a new pathway to high-throughput surface analysis. Next, I highlight nano-scale characterization of oxides using electron microscopy and spectroscopy. We demonstrated that the nano-scale characterization enables quantitative determination of coordination number distributions in disordered glass systems. High-temperature in-situ imaging was used to track the evolution of phase separation in glass, and a high-resolution analysis revealed the local thermal expansion behavior at the crystalline interface at the nano-scale. Finally, I introduce the data-driven materials discovery of layered intercalated compounds. By analyzing more than 9,000 host–guest combinations, we established a quantitative stability rule based on hard-soft acid-base interactions and Coulombic size effects, which can also predict the stability of previously unreported compounds. Furthermore, we identified promising graphite intercalation compounds expected to exhibit exotic properties such as superconductivity. Consequently, I emphasize the importance of methodological developments as essential foundations for advancing ceramic science. Looking forward, the convergence of simulations, characterizations, and informatics, together with emerging artificial intelligence (AI) technologies, is expected to open new directions for materials research.