Our Research Objective = “Paving the way for Designed Materials through a deep understanding of chemical bonding ”

For decades, materials development has been driven by trial and error. That paradigm is now changing. Our vision is the realization of Designed Materials — a new frontier in which atomic arrangements are deliberately engineered to achieve target functionalities, bringing into existence structures that nature alone would never produce. At the heart of this vision lies a deep understanding of structure-property relationships — the connections between atomic and electronic structure and material function. By deciphering, at the level of chemical bonding, exactly where and how function emerges, we transform materials from something to be discovered into something to be designed. The Mizoguchi Research Group integrates artificial intelligence, Physical AI, computational simulation, and materials characterization into a seamless research pipeline — driving the full cycle from the elucidation of structure-property relationships to the practical implementation of materials design.
The Mizoguchi Research Group is engaged in the following research themes:

AI for Materials: Toward Designed Materials through Data-Driven Science

In recent years, data science has been attracting attention as the fourth paradigm, and "Materials Informatics," which uses data science in material research, is advancing globally and rapidly.　Our laboratory integrates machine learning, large-scale data analytics, generative AI, and large language models (LLMs) with Physical AI and robotics to establish and implement a unified, data-driven research framework that seamlessly connects materials design, exploration, synthesis, measurement, and analysis. We further develop novel methodologies, build large-scale cross-modal databases spanning experiments and simulations, and advance autonomous research workflows powered by AI agents.
1) Paving the way for materials design with generative AI and large-scale language models (LLMs)
2) Multimodal materials characterization using machine learning
3) Materials developments through inverse design
4) Materials discovery and synthesis using AI Agents
5) Discovery of new materials and new physical properties using data-driven methods
6) Intelligent materials characterizations and developments via Physical AIs
For example, using our methods, it is possible to complete calculations that would normally take 22 years in just 3 hours. Additionally, by utilizing AI technology, we have succeeded in gaining new insights that even specialized researchers could not obtain before.

Keywords:Materials Informatics, Data-Driven, Artificial Intelligence, Generative AI, LLM, AI agent, Material Exploration, Spectral Analysis, Autonomous & Automated Analysis, Physical AI, Lattice Defects, Physics Extraction

“Accelerated inorganic materials design with generative AI agents” I. Takahara, T. Mizoguchi, and B. Liu, Cell Press Phys. Sci., 17 (2025) 103019-1-10.
“Band structure database of layered intercalation compounds with various intercalant atoms and layered hosts” N. Kawaguchi, K. Shibata, and T. Mizoguchi, Scientific Data, 11 (2024) 1244-1-10.
“Crystal orbital overlap population based on all-electron ab initio simulation with numeric atom-centered orbitals and its application to chemical-bonding analysis in Li-intercalated layered materials” I. Takahara, K. Shibata, and T. Mizoguchi, Modelling and Simulation in Materials Science and Engineering (MSMSE), 32 (2024), 055028-1-20.
“Prediction of the Ground State Electronic Structure from Core-loss Spectra of Organic Molecules by Machine Learning” PY. Chen, K. Shibata, K. Hagita, T. Miyata, and T. Mizoguchi, J. Phys. Chem. Lett., 14 (2023) 4858-4865.
“Learning excited states from ground states by using an artificial neural network S. Kiyohara, M. Tsubaki, and T. Mizoguchi, npj Comp. Mater., 6 (2020) 68-1-6.
“Prediction of interface structures and energies via virtual screening” S. Kiyohara, H. Oda, T. Miyata, and T. Mizoguchi, Science Adv., 2 (2016) e1600746-1-7.

High Precision and High Throughput Simulation of Lattice Defect Formation and Dynamics in Semiconductors and Two-Dimensional Materials　

With the dramatic improvement in computing speed in recent years, it is becoming possible to perform electronic state calculations for systems consisting of hundreds of atoms using parameter-free first-principles methods. Additionally, the environment for "Universal" machine learning potentials has improved, making it possible to optimize the structure of models with thousands of atoms with high precision on a laptop-level computer. Our laboratory is utilizing such cutting-edge simulation methods to conduct research on lattice defects, dopants, surfaces, and interfaces in advanced materials such as semiconductors, ceramics, two-dimensional compounds, and layered compounds, focusing on their formation behavior and dynamics.

Keywords:Semiconductors, Ferroelectric materials, Ceramics, Two-Dimensional Compounds, Moire materials, Machine learning potential (MLP), Unviersal MLP

“High Precision Machine Learning Force Field Development for BaTiO3 Phase Transitions, Amorphous, and Liquid Structures” P. Y. Chen, K. Shibata, and T. Mizoguchi APL Machine Learning, 3 (2025), 036115
“Unraveling the Stability of Layered Intercalation Compounds through First-Principles Calculations: Establishing a Linear Free Energy Relationship with Aqueous Ions” N. Kawaguchi, K. Shibata, and T. Mizoguchi, ACS phys. Chem. Au, 4 (2024) 281-291.
“A defect formation mechanism induced by structural reconstruction of a well-known silicon grain boundary” YS. Xie, K. Shibata, and T. Mizoguchi, Acta Mater., 250 (2023) 118827-1-11.
“The influence of neighboring vacancies and their charge state on the atomic migration of LaAlO3” T. Yamamoto and T. Mizoguchi , Appl. Phys. Lett., 102 (2013) 211910-1-4.
“Defect energetics in LaAlO3 polymorphs: A first principles study” T. Yamamoto and T. Mizoguchi , Phys. Rev. B, 86 (2012) 094117.

Understanding the Structure-Function Correlation in Energy-Related Materials　　

Developing high-performance power generation materials and storage materials is essential for achieving carbon neutrality. Our laboratory is conducting research aiming to understand the correlation between structure and function by analyzing the atomic and electronic structures near lattice defects such as interfaces and dopants, targeting energy-related materials like Li-Air Battery, Fuel cell, photovoltaic materials, secondary battery materials, ion conductors, and superconductors, using first-principles calculations, machine learning potential, and atomic-level analysis.

Keywords: Li-Air Battery, Li-ion Battery, Fuel Cell Materials, Photovoltaic Materials, Ion Conductors, Superconductors

“Possible New Graphite Intercalation Compounds for Superconductors and Charge Density Wave Materials: Systematic Simulations with Various Intercalants Using a van der Waals Density Functional Method” N. Kawaguchi, K. Shibata, and T. Mizoguchi, J. Phys. Chem. C, 127 (2023) 9833-9843.
“A valence state evaluation of a positive electrode material in a Li-ion battery with first-principles K- and L-edge XANES spectral simulations and resonance photoelectron spectroscopy” K. Kubobuchi, M. Mogi, M. Matsumoto, T. Baba, C. Sato, T. Yamamoto, T. Mizoguchi, H. Imai, J. Appl. Phys., 120, 142125-1-13 (2016)
“The atomic structure, band gap, and electrostatic potential at the (112)[1-10] twin grain boundary of CuInSe2” H. Yamaguchi, H. Hiramatsu, H. Hosono, and T. Mizoguchi, Appl. Phys. Lett., 104, 153904-1-5 (2014).

Simulation of Core-loss Spectrum and Developping “Spectrum Informatics” approaches

Electron energy-loss near-edge structure (ELNES) and X-ray absorption near-edge structure (XANES) are both core-level excitation spectra that occur when core electrons transition to unoccupied orbitals. ELNES and XANES are powerful material analysis techniques with high spatial resolution, time resolution, and detection sensitivity. However, interpreting these spectra requires theoretical calculations based on first-principles methods. Our laboratory is working on developing theoretical computational methods for ELNES/XANES and is currently engaged in research aimed at establishing comprehensive computational methods for all structures, all elements, and all absorption edges. Futhermore, based on the spectrum database through the simulations, we have attempted to extract new insight and physics from the spectrum via machine learning.

Keywords:Core-Level Excitation Spectroscopy, ELNES, XANES, Single Particle, Exciton, Multi-Electron, spectrum database, new physics for spectrum, Spectrum informatics
“Data-Driven ELNES/XANES Analysis: Predicting Spectra, Unveiling Structures, and Quantifying Properties” [Invited Review] T. Mizoguchi Microscopy, 2025 in press. DOI:10.1093/jmicro/dfaf038
“Quantification of the Properties of Organic Molecules Using Core-Loss Spectra as Neural Network Descriptors” K. Kikumasa, S. Kiyohara, K. Shibata, and T. Mizoguchi, Advanced Intelligent Systems, 4 (2022) 2100103-1-10. doi:10.1002/aisy.202100103
“Simulated carbon K edge spectral database of organic molecules” K. Shibata, K. Kikumasa, S. Kiyohara, and T. Mizoguchi, Scientific Data, 9 (2022) 214-1-11.
“Machine learning applications for ELNES/XANES” [Invited review] T.Mizoguchi and S. Kiyohara Microscopy, 69 (2020) 92-109.
“Basics and Applications of ELNES calculation” [Invited Review] H. Ikeno and T. Mizoguchi, Microscopy, 66 (2017) 305–327.
“Strong excitonic interactions in the oxygen K-edge of perovskite oxides” K. Tomita, T. Miyata, W. Olovsson, and T. Mizoguchi, Ultramicroscopy, 178 (2017) 105-111.
“Theoretical ELNES: one particle and many particle calculations”[Invited Review] T. Mizoguchi, W. Olovsson, H. Ikeno, and I. Tanaka, Micron 41 (2010) 695–709

Atomic-Resolution Material Analysis by “The Ultimate Analysis”　　

By combining spherical aberration-corrected scanning transmission electron microscopy (STEM) with electron energy loss spectroscopy (EELS), it is possible to obtain atomic-resolution information on the atomic and electronic structures of materials. This method is so powerful that it is referred to as "The Ultimate Analysis." Our laboratory is applying this "Ultimate Analysis" to advanced materials such as artificial superlattices and optical fibers, aiming to establish material design.

Keywords:Amorphous Materials, Artificial Superlattices, Li-Ion Battery Cathode Materials, Solar Cell Materials

“Charge Disproportionation at Twisted SrTiO3 Bilayer Interface Driven by Local Atomic Registry” Min-Su Kim, Kyoungjun Lee, Ryo Ishikawa, Kyung Song, Naafis Ahnaf Shahed, Kitae Eom, Mark Rzchowski, Evgeny Y Tsymbal, Naoya Shibata, Teruyasu Mizoguchi, Chang-Beom Eom, Si-Young Choi ACS Nano, 19 (2025) 39714-39724.
“Nanoscale Investigation of Local Thermal Expansion at SrTiO3 Grain Boundaries by Electron Energy Loss Spectroscopy” K. Liao, K. Shibata,and T. Mizoguchi, Nano Letters, 21 (2021) 10416-10422
“Controlling interface intermixing and property of SrTiO3 based superlattices” T. Mizoguchi, H. Ohta et al., Adv. Funct. Mater. 21, (2011) 2258–2263.
“Atomic Scale Identification of Individual Lanthanide Dopants in Optical Glass Fiber” T. Mizoguchi et al., ACS Nano, 7 (2013) 5058-5063.
“Site dependence and Peak assignment of YBa2Cu3O7 O-K ELNES” T. Mizoguchi et al.,Phys. Rev. B, 77 (2008) 024504-1-5.

Material Design of Glasses, Liquids, and Soft Materials Utilizing Atomic-Resolution Measurements　　

Glasses, liquids, gases, and soft materials are widely used in daily life and industrial activities. The properties of these materials are often determined by local atomic structure disorder and dynamic behavior changes, but local structure analysis is challenging due to their complexity. Our laboratory is developing methods to analyze glasses, liquids, gases, and soft materials with high spatial resolution using transmission electron microscopy and is applying these methods to actual materials. We have successfully observed the behavior of individual atoms in liquids in real space, observed dynamic behavior of gas molecules with high spatial resolution, and quantitatively elucidated high-temperature phase separation phenomena in glasses in real space and real-time.


Keywords:Glass, Ionic Liquids, Electrolytes, Fuel Cells, Surfactants, Soft Materials, Gases

“Revealing Spatial Distribution of Al-Coordinated Species in a Phase-Separated Aluminosilicate Glass by STEM-EELS” K. Liao, A. Masuno, A. Taguchi, H. Moriwake, H. Inoue, and T. Mizoguchi, J. Phys. Chem. Lett., 11 (2020) 9637–9642. here
“In situ observation of the dynamics in the middle stage of spinodal decomposition of a silicate glass via scanning transmission electron microscopy”K. Nakazawa, S. Amma, and T. Mizoguchi, Acta Mater. 200 (2020) 720-726. here
“Real-space analysis of diffusion behavior and activation energy of individual monatomic ions in a liquid” T. Miyata, F. Uesugi, and T. Mizoguchi, Science Advances, 3 (2017) e1701546-1-5. here
“Estimation of the molecular vibration of gases using electron microscopy” H. Katsukura, T. Miyata, M. Shirai, H. Matsumoto, and T. Mizoguchi, Scientific Reports, 7 (2017), 16434-1-9. here
“An estimation of molecular dynamic behaviour in a liquid using core-loss spectroscopy” Y. Matsui, K. Seki, A. Hibara, T. Mizoguchi, Scientific Reports, 3 (2013) 3503-1-7. here