[2N13-16] Fundamental Properties of Fuel, FP and Fuel Debris

3:50 PM - 4:05 PM JST(6:50 AM - 7:05 AM UTC)

[2N14] Evaluation of physical properties of plutonium dioxide using machine-learning molecular dynamics

*Hiroki Nakamura¹, Keita Kobayashi¹, Masahiko Okumura¹, Mitsuhiro Itakura¹, Masahiko Machida¹, Masashi Watanabe¹, Masato Kato² (1. JAEA, 2. Inspection Development)

Keywords:

First-principles calculations,PuO2,Machine-learning molecular dynamics,Bredig transition

In the development of nuclear fuels, it is important to obtain knowledge on the physical properties of nuclear fuel materials. However, sufficient data on fuel materials are not yet available due to handling restrictions and other factors. In such cases, it is necessary to obtain more precise data by interpolating physical properties through highly reliable numerical simulations such as first-principles calculations. However, first-principles calculations are computationally expensive and difficult to evaluate physical properties at high temperatures. Recently, machine-learning molecular dynamics that learn from first-principles calculation results have been developed, making it possible to simulate large-scale systems. In this study, we applied this method to plutonium dioxide to evaluate its high-temperature properties, and evaluated the effectiveness of this method.