Highly efficient and transferable interatomic potentials for α-iron and α-iron/hydrogen binary systems using deep neural networks

Shihao Zhang,Fanshun Meng,Rong Fu,Shigenobu Ogata

2023-12-01

Abstract:Artificial neural network potentials (NNPs) have emerged as effective tools for understanding atomic interactions at the atomic scale in various phenomena. Recently, we developed highly transferable NNPs for {\alpha}-iron and {\alpha}-iron/hydrogen binary systems (Physical Review Materials 5 (11), 113606, 2021). These potentials allowed us to investigate deformation and fracture in {\alpha}-iron under the influence of hydrogen. However, the computational cost of the NNP remains relatively high compared to empirical potentials, limiting their applicability in addressing practical issues related to hydrogen embrittlement. In this work, building upon our prior research on iron-hydrogen NNP, we developed a new NNP that not only maintains the excellent transferability but also significantly improves computational efficiency (more than 40 times faster). We applied this new NNP to study the impact of hydrogen on the cracking of iron and the deformation of polycrystalline iron. We employed large-scale through-thickness {110}<110> crack models and large-scale polycrystalline {\alpha}-iron models. The results clearly show that hydrogen atoms segregated at crack tips promote brittle-cleavage failure followed by crack growth. Additionally, hydrogen atoms at grain boundaries facilitate the nucleation of intergranular nanovoids and subsequent intergranular fracture. We anticipate that this high-efficiency NNP will serve as a valuable tool for gaining atomic-scale insights into hydrogen embrittlement.

Materials Science

What problem does this paper attempt to address?

This paper aims to solve the problem of atomic - scale understanding of the hydrogen embrittlement phenomenon in the iron - hydrogen system. Specifically, the research objective is to develop a new neural network potential (NNP), which not only maintains high transferability but also significantly improves computational efficiency (more than 40 times faster than the previous NNP). Through this efficient NNP, researchers can explore in more detail the influence of hydrogen atoms on the crack propagation in iron and the deformation of polycrystalline iron, especially the behavior at the crack tip and grain boundaries in the presence of hydrogen.

The main contributions of the paper are as follows:
1. **Improving computational efficiency**: The newly developed NNP greatly improves the computational speed while maintaining an accuracy comparable to that of density functional theory (DFT), making large - scale molecular dynamics simulations possible.
2. **Studying the hydrogen embrittlement mechanism**: Using the new NNP, it is studied how the aggregation of hydrogen atoms at the crack tip promotes brittle fracture and crack propagation, and how the role of hydrogen atoms at the grain boundaries promotes the nucleation of intergranular nano - voids, ultimately leading to intergranular fracture.
3. **Verifying the model performance**: Through a series of verification experiments, including phonon dispersion, stacking fault energy, vacancy cluster formation energy, self - interstitial energy, trapping energy of hydrogen atoms in vacancies, migration barrier of hydrogen - vacancy complexes, hydrogen adsorption energy on low - index surfaces, hydrogen - grain - boundary interaction, hydrogen - screw - dislocation interaction, screw - dislocation slip barrier, dislocation kink formation energy and hydrogen diffusion rate in iron, etc., the reliability and accuracy of the new NNP are proved.

In conclusion, through the development of an efficient and high - precision NNP, this paper provides an in - depth understanding of the hydrogen embrittlement phenomenon at the atomic scale, which is of great significance for designing new iron - based alloys with better hydrogen embrittlement resistance.