Abstract:We present the development of machine-learning interatomic potentials for uranium dioxide UO2 . Density functional theory calculations with a Hubbard U correction were leveraged to construct a training set of atomic configurations. This training set was designed to capture elastic and plastic deformations, as well as point and extended defects, and it was enriched through an active learning procedure. New configurations were added to the training database using a multiobjective criterion based on predicted uncertainties on energy and forces (obtained using a committee of models) and relative distances between new configurations in descriptor space. Two machine-learning potentials were developed based on physically sound pairwise potentials, which include the Coulombic interaction: a neural network potential and a SNAP potential. These potentials were optimized to minimize the root mean square error on the training database. Subsequently, the SNAP potential was used to compute the stacking fault energy surface in multiple directions, and the stabilized configurations were employed for subsequent DFT minimizations. The final DFT stacking fault energy surfaces of UO2 are presented, and the associated configurations are included in the training database for a new optimization. Finally, the results obtained from both machine-learned potentials were compared to standard semiempirical ones, demonstrating their excellent predictive capabilities for solid properties. These properties include defect formation energies, γ surface, elastic properties, and phonon dispersion curves up to the Breidig transition temperature. https://doi.org/10.1103/PhysRevMaterials.8.025402 ©2024 American Physical Society

Atomistic Model of Uranium

Study on the Electronic Structure and Elastic Constants of Uranium Dioxide by First Principles

A comparative study on modeling of the ferromagnetic and paramagnetic states of uranium hydride using a DFT+U method

A new type angular-dependent interatomic potential and its application to model displacement cascades in uranium

Density functional theory study on the electronic structure of UAI(3) and USn3

Relativistic Embedded Equation-of-Motion Coupled-Cluster Approach to the Core-Ionized States of Actinides: A Case Study of Uranyl(VI) in Cs$_2$UO$_2$Cl$_4$

Structure, Formation Energies and Elastic Constants of Uranium Metal Investigated by First Principles Calculations

A study of U-isotope ground state properties with Covariant Energy Density Functional

Electron Affinity of Uranium and Bound States of Opposite Parity in Its Anion

Geometries, Electronic Structures, and Excited States of UN2, NUO+, and UO 2 2+ : a Combined CCSD(T), RAS/CASPT2 and TDDFT Study

Ab initio DFT+U study of He atom incorporation into UO2 crystals

A Long-Range U–Nb Potential for the Calculation of Some Chemical and Physical Properties of the U–Nb System

Mechanical and Thermal Properties for Uranium and U–6Nb Alloy from First-Principles Theory

Atom-Resolved Chemical States in the Multivalent U-TM-O (TM: Ti, V, Cr, Mn, Fe, Ni, Nb, Mo, W) Ternary Oxides from First-Principles

Atomistic simulations of nuclear fuel with machine learning interatomic potentials

First-principles study of elastic and thermodynamic properties of UO2, γ-UO3 and α-U3O8

Density Functional Theory Study on the Electronic Structure of UAl3 a Nd USn3

Molecular Dynamics Simulation of Solid/liquid Interfacial Energy of Uranium

Strong Electron Correlation In Uo2-: A Photoelectron Spectroscopy And Relativistic Quantum Chemistry Study

Phonon density of states for solid uranium: Accuracy of the embedded atom model classical interatomic potential

Structural and Electronic Properties of Uranium-Encapsulated Au-14 Cage