Machine-Learning Microstructure for Inverse Material Design.

Metallurgy and material design have thousands of years' history and have played a critical role in the civilization process of humankind. The traditional trial-and-error method has been unprecedentedly challenged in the modern era when the number of components and phases in novel alloys keeps increasing, with high-entropy alloys as the representative. New opportunities emerge for alloy design in the artificial intelligence era. Here, a successful machine-learning (ML) method has been developed to identify the microstructure images with eye-challenging morphology for a number of martensitic and ferritic steels. Assisted by it, a new neural-network method is proposed for the inverse design of alloys with 20 components, which can accelerate the design process based on microstructure. The method is also readily applied to other material systems given sufficient microstructure images. This work lays the foundation for inverse alloy design based on microstructure images with extremely similar features.

Iterative spectral method for solving electrostatic or magnetostatic problems in complex and evolving heterostructures.

Electrostatic or magnetostatic problems involving complex heterogeneity are nontrivial for modeling and simulation. Most existing numerical methods focus on sharp interface models and the computational cost increases with increasing complexity of the geometry. Here we develop an iterative spectral method, the bound charge successive approximation algorithm, to solve electrostatic or magnetostatic heterogeneity problems in the context of diffuse-interface modeling. As tests and verifications, this algorithm is applied to calculation of the depolarization factor of an ellipsoid, and simulation of random dielectric mixtures and the dielectophoretic motion of multiple particles. The algorithm shows excellent efficiency and the computational cost mainly depends on the permittivity or permeability contrast in the whole system, regardless of the complexity of the geometry. In particular, for evolving heterostructures the solution of bound charge in one time step can be used as input for the next, which could further significantly shorten the iteration (approximation) process, making it practical to simulate the long-range electrostatic or magnetostatic interaction in complex and evolving heterostructures.

Toward a Quantitative Understanding of the Electric Field in Thermal Metal Oxidation and a Self-Consistent Wagner Theory.

The electric field in the growing oxide film is important to the kinetics and mechanism of metal oxidation. However, understanding of the essential characteristics of the electric field during oxidation remains insufficient. A special-case analytical model is presented that provides a unified understanding for the electric field from the viewpoints of kinetics and thermodynamics. More general cases are studied by computer simulations that show similar characteristics in the electric field. In particular, simulations indicate that in many situations, the electrostatic potential drop across the bulk oxide is limited to ∼kBT/e, which means that the total electrostatic potential drop across the oxide film, if on the order of 1 V by rough estimation, should have contributions mostly from the electrified interfaces. Finally, regarding the Gibbs-Duhem relation, the commonly used isobaric assumption for the diffusing species is refuted. The results contained herein also provide a self-consistent understanding of Wagner's oxidation theory.