2025年12月15日、Professor Abhishek K. Singh, Dr. Hiroshi Mizuseki, ：Professor N. S. Venkataramananによるご講演が
第182回構材ゼミ From fundamental understanding through atomistic simulations
to hypersonic technology development として開催されます。
日時：2025年12月15日（月曜日）14：00-16：50
場所：NIMS千現地区 先進構造材料研究棟5階カンファレンスルーム
1.講演題目：AI-based Hierarchical Representations of Materials for Structure–Property Prediction
1.講演者：Professor Abhishek K. Singh,Materials Research Centre, Chair, Office of International Relations,
Indian Institute of Science (IISc), Bangalore, India
2.講演題目：Ordered Phases in Mixed Crystals: High-entropy Alloys and III-V Compounds
2.講演者：Dr. Hiroshi Mizuseki, Principal Research Scientist, Computational Science Research Center,
Korea Institute of Science and Technology (KIST), Korea
3.講演題目：Computational Insights into Noncovalent Interactions Across Scales: From Molecules to Materials
3.講演者：Professor N. S. Venkataramanan, School of Engineering, Dayanada Sagar University, India
開催者：佐原 亮二（計算構造材料グループ）
Abstract #1:
Materials representation across multiple length scales is essential for enabling AI models to solve structure-to-property prediction problems in complex systems such as superalloys. Properties like Vickers hardness are primarily governed by microstructural features and formation energy correlates to atomic arrangements and, making it crucial to capture relevant information from different structural hierarchies. Accurate and interpretable representations across these scales allow machine learning approaches to accelerate materials discovery and design. At the microstructural level, three frameworks are used to represent image-based information: (1) statistical representations using 2-point spatial correlations to capture phase distribution patterns, (2) geometry-driven image processing techniques that extract morphological descriptors such as area, perimeter, and shape of precipitates, and (3) deep learning models like convolutional neural networks that automatically learn hierarchical features directly from raw SEM images. These image-derived features are combined with metadata such as composition and processing history to predict mechanical properties like Vickers hardness. At the atomic level, graph-based representations like the CLEAR (Chemistry and Local Environment Adaptive Representation) descriptor-based model represents crystal structures by combining elemental properties with interatomic distances through Voronoi based neighbors. By applying pooling operations, these graph features are transformed into fixed-size vectors that enable predictive modelling of formation energy and phase stability.
1. Acta Mater., 196, 295-303, (2020)
2. J. Mater. Sci. 55, 15845 (2020)
3. Acta Mater., 276, 120-122 (2024)
4. Comput. Mater. Sci (2025), 113854
Abstract #2:
The physical properties of mixed crystals and alloyed materials are fundamentally determined by their chemical composition and atomic configurations. In this study, we employ first-principles calculations to elucidate the principles governing atomic configurations, focusing on high-entropy alloys (HEAs) and III–V compounds as representative systems. Although HEAs have traditionally been considered to exhibit random atomic configurations, recent theoretical studies suggest the possibility of ordereddisordered coexistence in structures such as L12, D023, and D022. In this study, we applied first-principles calculations to determine the formation energies of 15 kinds of equiatomic quaternary alloys with 3 kinds of semi-ordered phases, selected from Al, Fe, Co, Ni, Cu, and Zn, as well as non-equiatomic CrFeCoNi alloys with varying composition ratios of the remaining constituent elements (Fe, Co, Ni), excluding the element forming the semi-ordered phase (Cr). Both equiatomic quaternary alloys and CrFeCoNi non-equiatomic alloys showed a tendency for decreasing formation energy and increasing stability in the order of L12, D023, and D022 as the value of Valence Electron Concentration (VEC) increased. The formation energies of all semi-ordered structures were lower than that of the random solid solution (RSS). [1, 2] Finally, we present results on VEC-dependent crystal structures of HEAs obtained through metallurgically informed and high-throughput screening approaches. [3]For group III nitrides with the wurtzite crystal structure, we identify energetically favorable ordered atomic configurations governed by “atomistic distancing rules” derived from pairwise interaction energies between two cation atoms. These configurations exhibit significantly lower formation energies than their disordered counterparts. [4-6] In zinc blende III–V compounds, we systematically investigate ordered configurations of group III cations at compositions x = 0.25, 0.5, and 0.75 across 12 quasibinary systems (e.g., AlₓGa₁₋ₓN, GaₓIn₁₋ₓAs). Since the spatial distribution of group III cations in the zinc blende structure is equivalent to that in a face-centered cubic (FCC) lattice, we evaluate FCC-based ordered phases to compare formation energies across different configurations. Our results reveal that for AlₓGa₁₋ₓP, AlₓGa₁₋ₓAs, and AlₓGa₁₋ₓSb at x = 0.25 and 0.75, the RSS structure yields the lowest formation energy, indicating a thermodynamic preference for disordered configurations. In contrast, at x = 0.5, the L11 structure is the most stable for these systems. For the remaining quasibinary systems, the D022 and chalcopyrite structures are energetically favored. We also analyze how pairwise cation interactions in zinc blende structure influence formation energies. [7] Reference:
[1] H. Mizuseki, R. Sahara, and K. Hongo, Sci. Technol. Adv. Mater.: Methods, 3, 2153632 (2023).
[2] H. Mizuseki, R. Sahara, and K. Hongo, Comput. Mater. Sci., 259, 114114 (2025).
[3] H. Mizuseki, R. Sahara, and K. Hongo, in preparation.
[4] H. Mizuseki, J. S. Gueriba, M. J. F. Empizo, N. Sarukura, Y. Kawazoe, and K. Ohkawa, J. Appl.Phys., 130, 035704 (2021).
[5] J. S. Gueriba, H. Mizuseki, M. Cadatal-Raduban, N. Sarukura, Y. Kawazoe, Y. Nagasawa, A.Hirano, and H. Amano, J. Phys.: Condens. Matter, 36, 135001 (2024).
[6] H. Mizuseki, J. S. Gueriba, M. Cadatal-Raduban, N. Sarukura, E. Tamiya, and Y. Kawazoe, J.Appl. Phys., 135, 145701 (2024).
[7] H. Mizuseki, N. Sarukura, N. Chikumoto, M. Cadatal-Raduban, T. Shimizu, and Y. Kawazoe, in preparation.
Abstract #3:
Noncovalent interactions play a fundamental role in determining the structure, stability, and function of molecular and material systems. Unlike covalent bonds, these interactions—encompassing hydrogen bonding, van der Waals forces, π–πstacking, electrostatic interactions, and hydrophobic effects—are relatively weak and highly tunable, enabling reversible assembly and dynamic behavior across chemical and biological contexts. From molecular recognition and protein folding to crystal engineering, catalysis, and the design of advanced functional materials, noncovalent forces govern key physicochemical properties and performance. Understanding these interactions from a computational perspective provides critical insights into the balance of energetic and entropic contributions that dictate molecular organization. However, accurately modeling noncovalent interactions (NCIs) remains a significant challenge. These interactions are typically weak (1–10 kcal·mol⁻¹) and arise from subtle electron correlation effects that standard DFT functionals often fail to capture. Here we present our work on the new wave functional methods that can address and benchmark for quantifying and interpreting noncovalent interactions. This talk will present the use of various wavefunction methods to understand intermolecular interactions in molecules to materials and provide quantitative predictions on the NCI which offer deep insight into the fundamental physical nature of intermolecular forces.