In close collaboration with experimental researchers, we advance the design of structural and functional materials through multiscale simulations based on first-principles calculations.

Phase diagrams are indispensable tools throughout the alloy development process. Conventionally, phase diagrams are calculated using the CALculation of PHAse Diagrams (CALPHAD) method with reference to thermodynamic (experimental) databases. However, constructing such databases requires considerable time and cost, and the reliability of the resulting phase diagrams depends strongly on the accuracy and availability of the limited experimental data incorporated into the databases.
To overcome these limitations, we have demonstrated that alloy phase diagrams can be determined entirely from first-principles calculations, without relying on empirical thermodynamic databases. The calculated phase diagrams successfully reproduce all key features observed in their experimental counterparts. Furthermore, the computational framework has been significantly accelerated through the use of machine-learning interatomic potentials.

To establish a multiscale modeling framework through coarse-graining of first-principles calculations, we are developing methodologies that appropriately account for boundary conditions between different length scales and enable first-principles calculations to be seamlessly coupled with more macroscopic approaches, including lattice Monte Carlo and phase field methods. The key concept is to integrate out degrees of freedom at length scales shorter than the lattice constant, thereby enabling interactions in continuous space to be rigorously mapped onto effective interactions in discrete space.
Through a series of studies, we have proposed the first-principles phase field (FPPF) method and successfully demonstrated parameter-free prediction of alloy microstructures, including those of Ni–Al-based heat-resistant superalloys. We are currently extending this framework to multicomponent alloy systems.

Density functional theory (DFT) may not always provide sufficient accuracy for detailed analyses of the electronic properties of materials. For example, DFT is well known to underestimate the band gaps of semiconductors. To overcome this limitation, we employ an all-electron GW approach based on quasiparticle theory. The calculations are performed using TOMBO (Tohoku Mixed Basis Orbitals ab initio program), a first-principles computational code based on the all-electron mixed basis approach.
This approach enables the quantitative prediction of band gaps and the elucidation of the underlying physical mechanisms, thereby facilitating the development of advanced materials, including structural biomaterials. We are also extending this methodology to time-dependent calculations (time dependent GW, TDGW).