arXiv subjects

Vadim Sotskov

Publications and source records attributed to Vadim Sotskov.

4 recordsLinked to original sources

arXiv2024

Fitting to magnetic forces improves the reliability of magnetic Moment Tensor Potentials

We developed a method for fitting machine-learning interatomic potentials with magnetic degrees of freedom, namely, magnetic Moment Tensor Potentials (mMTP). The main feature of our method consists in fitting mMTP to magnetic forces (negative derivatives of energies with respect to magnetic moments) as obtained spin-polarized density functional theory calculations. We test our method on the bcc Fe-Al system with different compositions. Specifically, we calculate formation energies, equilibrium lattice parameter, and total cell magnetization. Our findings demonstrate an accurate correspondence between the values calculated with mMTP and those obtained by DFT at zero temperature. Additionally, using molecular dynamics, we estimate the finite-temperature lattice parameter and capture the cell expansion as was previously revealed in experiment. Furthermore, we demonstrate that fitting to magnetic forces increases the reliability of structure relaxation (or, equilibration), in the sense of ensuring that every relaxation run ends up with a successfully relaxed structure (the failure may otherwise be caused by falsely driving a configuration away from the region covered in the training set).

cond-mat.mtrl-sci↗

arXiv2023

A machine learning potential-based generative algorithm for on-lattice crystal structure prediction

We propose a method for crystal structure prediction based on a new structure generation algorithm and on-lattice machine learning interatomic potentials. Our algorithm generates the atomic configurations assigning atomic species to sites of the given lattice, and uses cluster expansion or low-rank potential to evaluate their energy. We demonstrate two benefits of such approach. First, our structure generation algorithm offers a ``smart'' configurational space sampling, targeting low-energy structures which significantly reduces computational costs. Second, the application of machine learning interatomic potentials significantly reduces the number of DFT calculations. We discuss how our algorithm resembles the latent diffusion models for image generation. We demonstrate the efficiency of our method by constructing the convex hull of Nb-Mo-Ta-W system, including binary and ternary Nb-W and Mo-Ta-W subsystems. We found new binary, ternary, and quaternary stable structures that are not reported in the AFLOW database which we choose as our baseline. Due to the computational efficiency of our method we anticipate that it can pave the way towards efficient high-throughput discovery of multicomponent materials.

cond-mat.mtrl-sci↗

arXiv2022

Machine-learning Driven Synthesis of TiZrNbHfTaC5 High-Entropy Carbide

Synthesis of high-entropy carbides (HEC) requires high temperatures that can be provided by electric arc plasma method. However, the formation temperature of a single-phase sample remains unknown. Moreover, under some temperatures multi-phase structures can emerge. In this work we developed an approach for a controllable synthesis of HEC TiZrNbHfTaC5 based on theoretical and experimental techniques. We used canonical Monte Carlo (CMC) simulations with the machine learning interatomic potentials to determine the temperature conditions for the formation of single-phase and multi-phase samples. In full agreement with the theory, the single-phase sample, produced with electric arc discharge, was observed at 2000 K. Below 1200 K the sample decomposed into (Ti-Nb-Ta)C and a mixture of (Zr-Hf-Ta)C, (Zr-Nb-Hf)C, (Zr-Nb)C, and (Zr-Ta)C. Our results demonstrate the conditions for the formation of HEC and we anticipate that our approach can pave the way towards targeted synthesis of multicomponent materials.

cond-mat.mtrl-sci↗

arXiv2022

Modelling Surface Segregation in Compositionally Complex Alloys with Ab-Initio Accuracy

Compositionally complex alloys or concentrated solid solutions are the latest frontier in catalyst design, but mixing different elements in one catalyst may result in surface segregation. Atomistic simulations can predict segregation patterns, but standard approaches based on mean-field models, cluster expansion, or classical interatomic potentials are often limited for the description of multicomponent alloys. We present machine learning potentials that can describe surface segregation with near DFT accuracy. The method is used to study a complex Co-Cu-Fe-Mo-Ni quinary alloy. For this alloy, an unexpected segregation of Co, which has a relatively high surface energy, is observed. We rationalize this surprising mechanism in terms of simple transition-metal chemistry.

cond-mat.mtrl-sci↗