A dislocation-aware active learned interatomic potential for Bi1−xSbx

We introduce a force-following active learning algorithm that integrates density functional theory (DFT) with the Gaussian Approximation Potential (GAP) framework to develop a robust interatomic potential (IP) for a dislocation in the topological insulator, Bi_1−xSb_x. Starting from an initial potential, IP₀, trained on unit cell data from strained Bi–Sb binaries, our active learning approach iteratively refines the IP during a structural relaxation. In each cycle, if the force uncertainty of any atom near the dislocation core exceeds a threshold value, the IP_i is efficiently retrained (IP_i→ IP_i+1) by incorporating DFT-computed total energies and atomic forces of representative structures that include the highest-uncertainty atom and its surrounding local environment. This strategy ensures that the relaxation process maintains a low force uncertainty until the full convergence is achieved. In this work, we demonstrate the framework on a [100] edge dislocation in Bi₇Sb₁. Consequently, the final IP, IP_f, has two capabilities: (1) it reproduces the relaxation pathway observed during the active learning process unlike the initial IP₀, which lacks prior dislocation core knowledge; and (2) it captures the lattice and elastic properties of Bi–Sb binaries across a range of Sb concentrations. We also evaluate dislocation properties (Peierls stresses and dislocation generation by compression) to assess the performance of the “dislocation-aware” IP_f.