📝 SummarXiv

Abstract

The size and morphology of defect clusters formed during primary damage play a crucial role in the subsequent microstructural evolution of irradiated materials. Molecular dynamics (MD) simulations of collision cascades in tungsten (W) were performed using five interatomic potentials (IAPs): the quantum-accurate machine-learned Spectral Neighbor Analysis Potential (W-SNAP), the machine learning-based tabGAP potential, and three embedded-atom method (EAM) potentials. A total of 3,500 MD simulations were conducted with primary knock-on atoms (PKAs) at energies of 5, 10, 20, 50, 75, 100, and 150 keV. PKAs were launched in 100 random directions at each energy to ensure statistical validity. Analysis was performed using CSaransh , a web-based tool for large-scale collision cascade databases, to quantify: (i) the number of defects (isolated and clustered), (ii) defect cluster morphologies, (iii) defect cluster size distributions and (iv) the number of sub-cascades formed. We show that the difference in the formation energy of self interstitial atom dumbells along the <1 1 0> and <1 1 1> directions critically influence defect cluster morphology. Our results indicate that IAP stiffness and interaction range independently do not affect defect count. However, these parameters combined with defect formation energies, threshold displacement energies, and other factors significantly influence defect production.