about us
Research Overview
Our group focuses on the Phase-Field Method (PF method) and aims to establish a universal framework that enables a fundamental understanding of microstructure formation in materials, as well as the improvement and optimization of advanced materials through microstructure–property analysis across a wide range of systems. The PF method is a phenomenological simulation approach based on continuum models for microstructure evolution. Its applications now extend dendritic solidification, diffusion-controlled phase separation (including nucleation, spinodal decomposition, and Ostwald ripening), order-disorder phase transitions, various types of domain evolution (in dielectric and magnetic materials), structural phase transformations, martensitic transformations, shape-memory behavior, grain growth and recrystallization, dislocation dynamics, fracture and crack propagation, and many other phenomena; essentially covering the entire field of materials science. Because of this broad applicability, the PF method can naturally handle situations involving multiple simultaneous phase transformations and allows relatively straightforward modeling of microstructure formation under external fields such as stress, magnetic, or electric fields. In recent years, next-generation material design methods that combine image-based material property analysis with the PF method have also been attracting attention. In conventional image-based material property analysis, the image data representing a material’s internal microstructure typically consisted of experimentally measured images or idealized representations of actual microstructures. However, with the advancement of PF method, it has become possible to replace these image data with computational results from the PF method. Consequently, new research is currently underway in various fields to integrate the process analysis of microstructure formation based on PF method with image-based materials property calculations. Furthermore, with recent advances in AI technology, there has been a growing global trend toward integrating enormous amounts of experimental data with many computational methods in materials design to overwhelmingly accelerate materials development (and drastically improve the efficiency of trial-and-error testing processes). Our group is also working to apply accumulated knowledge from AI research to materials science and engineering, with a particular emphasis on the PF method and image-based property analysis described above.
Members
Specialty Fields and Research Interests
1. Understanding Microstructure Formation Mechanisms Using the Phase-Field Method
The PF method is a continuum-based phenomenological framework that describes a wide variety of microstructure formation processes in materials, where heterogeneous microstructures are represented using multiple order parameters (functions of space and time). The total free energy functional of the heterogeneous system is defined in terms of these order parameters, and microstructure evolution is computed by solving the governing equations that describe the reduction of total free energy. Because the PF method provides a rigorous computational framework that integrates energetic theory with kinetics, it offers a powerful and highly logical means of understanding the mechanisms underlying the formation of heterogeneous and complex microstructures.
2. Image-Based Material Property Analysis
Since the PF method provides detailed information on heterogeneous microstructures, including their temporal evolution, it enables image-based property calculations using PF-generated microstructures as boundary conditions. Various analytical approaches exist for evaluating the properties of multiphase microstructures, including mean-field approximations developed in micromechanics, micromagnetic calculations for magnetic materials, and methods based on Landau theory for dielectric materials. We are currently working to integrate these analytical techniques with microstructures obtained from PF simulations.
3. AI-Assisted Materials Design Based on Microstructural Information
These approaches allow us to construct datasets that combine microstructural information with material property data across diverse fields of materials science and engineering. Machine learning techniques can then be used to efficiently analyze relationships between microstructures and properties. Moreover, recent advances in generative AI have made it possible for AI agents to perform entire analysis and simulation workflows. Leveraging our strength in developing all program code in-house, we are currently building an integrated AI-assisted materials design system that incorporates all of the above components.