HOME > Awards
2025.05.14
The paper I. Lobzenko, T. Tsuru, H. Mori, D. Matsunaka, Y. Shiihara, "Implementation of Atomic Stress Calculations with Artificial Neural Network Potentials", Materials Transactions, 64 (10), 2023, pp. 2481-2488 DOI https://doi.org/10.2320/matertrans.MT-M2023093 received the Paper Award for Physical Sciences (論文賞物性部門) from the Japan Institute of Metals (日本金属学会)
The concept of atomic stress in a discrete system is a powerful tool for studying complex processes such as heat flux, crack propagation, and void growth. The connection between the continuous stress field and the atomic stress has been established by the Irvin-Kirkwood procedure, but the existence of multiple variations of the procedure makes the definition of atomic stress unclear. In addition, special attention should be paid to the implementation of the atomic stress calculations for each type of interatomic potential. In principle, it is necessary to use pair forces to satisfy the linear momentum balance. Artificial neural network potentials (ANN potentials) have recently attracted wide attention due to their high accuracy. Before this study, no formulation of atomic stress existed that was suitable for the framework of ANN potentials. In the awarded paper, the rigorous formulation of the atomic stress was derived for the ANN potentials. Specifically, for two types of descriptors - Behler-Parinello functions and Chebyshev polynomials - the contributions to the pair forces were written. The requirements needed to preserve the balance of momentum are discussed and implemented. Also, the formulation was implemented in software that can be used in the widely known classical molecular dynamics code 'lammps'. That ensured the possibility of atomic stress calculations with ANN potentials to be used for a large variety of atomic systems. The verification was done by calculating stress distribution near the low-Miller-index surfaces in Fe and Al. It was shown that the distribution of stress in the direction perpendicular to the surface exhibits oscillations. In previous studies by first-principles calculations, such oscillations were associated with the Friedel-type oscillations of the charge distribution. In the awarded paper, the comparison of results obtained with the conventional EAM potentials and the ANN potentials shows that only the latter can reproduce the stress oscillations. This allows authors to claim that the ANN potentials can capture electronic properties despite their classical nature. Even though electronic properties are not included in the training, and therefore ANN potentials do not deal with electrons, there is an indirect influence of electronic properties. It can be understood as an influence on energies of structures in the dataset used for ANN potentials training, calculated from first principles (and therefore with the electronic sub-system fully accounted for).
【Awarded paper】https://doi.org/10.2320/matertrans.MT-M2023093
2025.04.30
Dr. Osamu Iwamoto, Dr. Nobuyuki Iwamoto, and Dr. Kenichi Tada of the Nuclear Science and Engineering Center received the "Minister of Education, Culture, Sports, Science and Technology (MEXT), Science and Technology Award" for the "Development of the Evaluated Nuclear Data Library JENDL."
The evaluated nuclear data library is a core database for numerical simulation technology involving nuclear reactions. Previous libraries had issues with completeness and reliability in meeting various needs such as the development of new reactors for GX, the treatment and disposal of radioactive waste, and the use of radiation in accelerators. In this development, a unique nuclear reaction model calculation code was constructed and used for evaluation to improve the reliability of nuclear reaction data for various radiation. For neutron-induced reactions, the completeness of the data was improved, including almost all nuclides with half-lives longer than one day, and the accuracy of nuclear analysis calculations in reactors was enhanced. This will enable use in research and development in a wide range of fields, including nuclear power, medicine, space, and basic science, while also improving the reliability of numerical simulations of radiation, such as by significantly enhancing the accuracy of predicting the criticalities for plutonium-containing cores and fast reactors. It is expected that this result will contribute to the development of new nuclear reactors toward carbon neutrality and the promotion of various radiation uses with accelerators.
2025.01.27
Dr. Kayo Yanagisawa of the Research Group for Nuclear Chemistry received the “New Century Award for Young Researchers” from the Kanto Branch of the Japan Society for Analytical Chemistry for “Analysis of trace elements and radioisotopes using ICP-MS combined with flow injection analysis”.
Inductively coupled plasma-mass spectrometry (ICP-MS) is widely used in various fields, including nuclear engineering. However, conventional methods for radioisotopes require complicated preparation, skilled operators, and constant attention to avoid radioactive contamination and exposure when handling radioactive standards. In this study, a new automatic analysis that does not require standards was developed by combining an automated sample preparation system with isotope dilution method. This method was successfully applied to the automated analysis of difficult-to-measure Sr-90 and to the quantitative mapping analysis of trace elements. These achievements contribute to making the analysis of trace elements and radioactive isotopes using ICP-MS simpler, faster, and less skilled, while also enhancing safety in the decommissioning of the Fukushima daiichi nuclear power plant, radioactive waste disposal, and environmental radiation monitoring.
Copyright © Japan Atomic Energy Agency. All Right Reserved.
