ProjectsEducational / Academic Type

Research Overview

Our research aims to transform conventional “first-principles calculations,” which often oversimplify many-electron systems, into “true first-principles methods” capable of reliably predicting the properties of industrially relevant materials. The foundation of this work is TOMBO (TOhoku Mixed-Basis Orbitals ab initio simulation package), a unique computational framework we have been developing for over 30 years. TOMBO enables the calculation of absolute energy values of material systems and the prediction of chemical reaction dynamics without relying on experimental data or phenomenological parameters. As a result, it provides highly reliable predictions of the time evolution of chemical processes.
In parallel, we promote science and technology education for companies, students, local governments, and the broader public through the “DA-TE Graduate School.” This platform hosts more than 130 recorded lectures delivered by university emeritus professors, contributing to knowledge dissemination across society.

Research Features

First-principles calculations based on density functional theory (DFT) are widely used across condensed matter physics, theoretical chemistry, pharmacology, and materials science. However, their accessibility through commercial software often obscures fundamental limitations, particularly the reliance on experimentally fitted parameters. In recent years, many AI-driven approaches have focused on accumulating large datasets and applying machine learning to predict new materials. While effective in some contexts, these approaches often lack a rigorous theoretical foundation. In contrast, our research emphasizes fundamental advances in computational theory. A key feature of TOMBO is its ability to compute absolute energies, whereas most existing methods provide only relative values. Moreover, TOMBO uniquely enables highly accurate treatment of excited states, which are essential for describing chemical reaction pathways.

Expected Outcomes and Developments

Density functional theory, proposed by Professor Kohn more than 60 years ago, revolutionized computational science by reducing the complexity of many-electron problems through the use of electron density rather than wave functions. However, DFT is fundamentally limited to ground-state properties, and the exchange–correlation functional cannot be uniquely determined. Consequently, most practical calculations rely on parameterization to reproduce experimental results. Our work seeks to eliminate this dependence. We have already demonstrated parameter-free simulations of hydrogen molecule formation during methane decomposition on catalytic surfaces, contributing to a deeper understanding of combustion processes and hydrogen production. We are also investigating the fundamental mechanisms of DNA recombination, with the goal of advancing our understanding of complex biological systems. Although current supercomputing capabilities restrict simulations to relatively small systems, computational power continues to grow rapidly. We therefore anticipate extending our methods to large-scale systems, such as hydration reactions on cement surfaces, in the near future.

References
1. ”Non-Adiabatic Excited-State Time-Dependent GW (TDGW) Molecular Dynamics Simulation of Nickel-Atom Aided Photolysis of Methane to Produce a Hydrogen Molecule”, Nanomaterials, 2024,
https://doi.org/10.3390/nano1010000.
2. 日刊工業新聞2024年6月21日「第一原理計算に新手法ー東北大など電子励起状態に対応」