Dmitry A. Aksyonov | Searcharxiv

arXiv2026

Active learning of collinear magnetic Moment Tensor Potentials using the spin-MLIP package from soft-constrained spin-polarized DFT calculations: a case study of Fe-Pd

Explicit incorporation of magnetic degrees of freedom in machine-learning interatomic potentials (magnetic MLIPs) plays a crucial role in the correct description of magnetic materials and their properties. An important ingredient for fitting of magnetic MLIPs is spin-polarized density functional theory (DFT) calculations with non-equilibrium magnetic moments, i.e. DFT calculations with constraints on magnetic moments. In this study, we present a workflow for active learning of magnetic Moment Tensor Potential (mMTP) during molecular dynamics (MD) simulations. Magnetic MTP and its active learning algorithm were implemented in the open-source spin-MLIP code, DFT soft-constrained spin-polarized calculations were performed with the VASP code, and MD simulations were conducted in the open-source LAMMPS code. We test our workflow on the Fe-Pd crystal. The dependencies of magnetization and phonon density of states (DOSs) on the volume of a supercell (or, pressure) are in good agreement with those calculated with DFT. Furthermore, the calculated DOSs correspond to the experimental ones.

cond-mat.mtrl-sci↗

arXiv2026

Machine learning assisted molecular dynamics of charge-transfer mechanisms at Li/Ga-doped Li$_7$La$_3$Zr$_2$O$_{12}$ (LLZO) interfaces

Interfacial charge transfer between solid electrolytes (SEs) and Li metal is a key factor limiting all-solid-state battery performance. Conventional density functional theory and nudged elastic band calculations are performed at 0 K along a single minimum-energy path and therefore neglect concerted multi-ion motion and finite-temperature effects, which can lead to inaccurate activation barriers. Here, we trained moment tensor potentials (MTPs) for garnet LLZO systems (t-LLZO, c-LLZO, and Ga-LLZO) and Li metal, enabling machine-learning molecular dynamics (MLMD) simulations of Li$^+$ diffusion in the bulk and across Li/SE interfaces. We also introduce a residence-time window method that filters out ion rattling at the interface and isolates genuine charge-transfer events. The resulting charge-transfer activation energy at the Li/Ga-LLZO interface is only $167 \pm 18$ meV, below the 200 meV barrier for vacancy-mediated Li migration in bulk Ga-LLZO. The corresponding intrinsic charge-transfer resistance is $\sim 10^{-5}\ Ω \mathrm{cm}^{2}$, almost four orders of magnitude below the lowest experimentally reported $R_{\mathrm{ct}}$. These results indicate that intrinsic charge transfer across the Li/Ga-LLZO interface is not rate-limiting, and that Li transport within the electrolyte instead governs the overall kinetics. Taken together, our findings clarify the fast interfacial kinetics in Li/LLZO systems, and the proposed methodology can aid further interface optimization in solid-state batteries.

cond-mat.mtrl-sci↗