Seiki Saito, Keisuke Takeuchi, Hiroaki Nakamura, Yasuhiro Oda, Kazuo Hoshino, Yuki Homma, Shohei Yamoto, Yuki Uchida
4 min
Abstract
Understanding the long-term transport of hydrogen isotopes in plasma-facing materials, such as tungsten, is critical for the steady-state operation of magnetic confinement fusion reactors. However, dynamically updating the transition parameters for kinetic Monte Carlo (kMC) simulations as the atomic structure evolves under continuous plasma irradiation remains a severe computational bottleneck. Conventionally, calculating these migration barriers requires the iterative and computationally expensive Nudged Elastic Band (NEB) method. To overcome this limitation, this article presents a highly efficient surrogate model for predicting migration barriers using a three-dimensional Convolutional Neural Network (3D-CNN), establishing the final component necessary to realize on-the-fly molecular dynamics (MD) and kMC hybrid simulations. The proposed deep learning model takes a two-channel volumetric input, the local three-dimensional potential energy distribution and the voxelized spatial coordinates of the initial and final trapping sites, to directly output the migration barrier as a scalar value. Trained on a comprehensive dataset of tungsten-hydrogen configurations evaluated using the Embedded Atom Method (EAM) potential, the model demonstrated robust predictive accuracy, achieving a Mean Absolute Error (MAE) of 0.124 eV and a high coefficient of determination of 0.890. Furthermore, utilizing GPU acceleration, the inference time is reduced to approximately 2.7 milliseconds per barrier, achieving a speed-up ratio of over 23,000 compared to conventional NEB calculations. This extraordinary acceleration effectively resolves the computational barrier of transition rate evaluations, paving the way for large-scale, dynamic modeling of plasma-wall interactions.
Alex: Huh—so instead of grinding through springs every time, train something to spot patterns in energy hills?
Sam: Yes. Trained on results from tungsten-hydrogen setups using an Embedded Atom Method potential—which models metal bonds by embedding atoms in electron clouds plus pairwise pushes and pulls—the network sees a 3D grid of energy around two sites and outputs the barrier directly. Paths weave through the full volume of the lattice, influenced by nearby atoms in all directions—like navigating a 3D maze of hills.
Alex: Voxels—like the 3D pixels in a video game world?
Sam: Exactly—a fine grid cropped around the start, limited to short hops. Convolutions then scan with small 3D filters, layering features like spotting ridges or valleys step by step. From thousands of examples, it achieves predictions that closely match the true barriers.
Alex: So the network learns the hill shapes implicitly...
Sam: Yes—this completes their trio of models: prior ones predict energy landscapes and trap sites; now barriers flow fast enough for real-time hybrids. The paper suggests this enables simulations to finally model years of wall degradation under plasma, meaning targeted reactor designs to manage hydrogen buildup.
Alex: The evidence shows it's a meaningful advance in handling dynamic structures.
Sam: Thanks for listening to ResearchPod.