What Can DP Do too?| Unveiling the Secrets of Radiation Damage Resistance and Intergranular Fracture in Superconducting Materials Using Deep Learning Potentials

Niobium (Nb), a superconducting metal, plays an indispensable role in cutting-edge technological fields such as superconducting radio frequency cavities and nuclear fusion reactors due to its excellent superconducting properties, corrosion resistance, and high melting point. However, in these extreme service environments, niobium inevitably faces challenges from high-energy particle irradiation and microcrack propagation, which can severely damage its performance and even lead to catastrophic failure. Therefore, how to suppress irradiation damage and resist crack propagation is crucial to ensuring the safe and stable operation of related equipment.

Recently, the research group of Professor Yong Huadong and Associate Professor Zhang Yajun from Lanzhou University published a research paper entitled "Suppression of irradiation defects and crack propagation in niobium via grain boundary engineering: A deep potential molecular dynamics study" in the internationally renowned journal Materials & Design. This study systematically revealed the intrinsic mechanism of improving the irradiation resistance and fracture toughness of superconducting niobium materials through "grain boundary engineering" using deep learning potentials (DP), providing a solid theoretical basis for the design of highly reliable polycrystalline niobium materials.

Research Background

Materials in irradiation environments generate a large number of point defects (vacancies and interstitial atoms), and the continuous accumulation of these defects leads to the deterioration of material properties. Theoretically, grain boundaries (GBs) inside materials can act as "defect absorption sinks" to effectively capture and annihilate these defects. However, there are many types of grain boundaries with different structures, and their ability to absorb defects varies greatly. Which kind of grain boundary has the best "healing" effect? This is an urgent question to be answered. To address this bottleneck, this study developed and applied a high-precision DP potential function that balances accuracy and efficiency. Figure 1 shows the prediction of some defect structures by this model, which is highly consistent with the DFT results.

Figure 1. Comparison of energies and forces predicted by DP and DFT for the test set deviating from equilibrium.

High-Precision DP Model: The Key to Unlocking Large-Scale Accurate Simulations

The constructed DP model has been specially optimized and combined with the ZBL (Ziegler-Biersack-Littmark) potential to accurately describe the interaction between atoms during high-energy collisions. As shown in Table 1, this model can not only accurately reproduce the basic physical properties such as lattice constants and elastic constants calculated by experiments and DFT but also precisely predict the point defect formation energy, stacking fault energy, and grain boundary energy, which are crucial for our research. Its comprehensive performance is significantly better than the traditional EAM and MEAM potential functions. This "key" allows us to explore the internal secrets of niobium materials with unprecedented accuracy on a larger spatiotemporal scale.

Table 1. Comparison of basic material properties of Nb calculated by DFT method, MD combined with machine learning potential, and MD combined with empirical potential with experiments.

Figure 2. Number of residual defects after irradiation in different types of bicrystal models simulated by the DP-ZBL model.

In Figure 2, the performance of different types of grain boundaries in suppressing irradiation defects is systematically compared. By simulating the 10 keV particle bombardment process, all models containing grain boundaries (symmetric tilt grain boundaries STGB, asymmetric tilt grain boundaries ATGB, twist grain boundaries TWGB) show stronger irradiation resistance compared to the perfect single crystal bulk material. This proves the universal effectiveness of grain boundary engineering.

Among the best: Among various grain boundaries, STGB Σ3{112}, ATGB Σ11{225}/{441}, and TWGB Σ3{111} perform particularly well, as they can absorb and annihilate irradiation-induced defects to the greatest extent.

Suppressing Crack Propagation: How Do Grain Boundaries Improve Material Toughness?

Irradiation not only generates point defects but also makes materials brittle and more prone to cracking. Our research further reveals the key role of grain boundaries in resisting crack propagation.

Negative effects of irradiation: Irradiation reduces the strength of niobium materials regardless of the presence of grain boundaries, which is a common consequence of irradiation damage.

Positive role of grain boundaries: Compared with single crystal materials, models containing specific grain boundaries (especially ATGB and TWGB) can withstand greater deformation before fracture, showing higher strength and fracture toughness.

Excellent performance of twist grain boundaries (TWGB): We found that the model containing TWGB Σ3{111} exhibits the best comprehensive mechanical properties both before and after irradiation, playing a crucial role in enhancing material strength and toughness. It dissipates energy by promoting dislocation emission at the crack tip, thereby effectively preventing the fatal propagation of cracks.

Figure 3(d). Distribution of internal defects and crack tips under different strain states in the TWGB model under irradiated and unirradiated conditions.

Conclusions and Prospects

This study systematically clarified the mechanism of grain boundary engineering and strain engineering in suppressing irradiation damage and crack propagation in superconducting niobium using high-precision deep learning potential functions. We screened out several specific grain boundary types with excellent performance in improving material irradiation resistance and fracture toughness, especially the twist grain boundary TWGB Σ3{111}. These findings not only deepen our understanding of the damage behavior of superconducting niobium but also provide important theoretical guidance and design ideas for the future design and manufacture of high-performance and high-reliability niobium materials used in extreme environments such as superconductivity and nuclear energy. This research was funded by the National Natural Science Foundation of China, and the relevant computing work was completed relying on the Supercomputing Center of Lanzhou University.

Paper link:
https://doi.org/10.1016/j.matdes.2025.114292