A High-Dimensional Neural Network Potential for Co$_3$O$_4$

Amir Omranpour,Jörg Behler

2024-09-17

Abstract:The Co$_3$O$_4$ spinel is an important material in oxidation catalysis. Its properties under catalytic conditions, i.e., at finite temperatures, can be studied by molecular dynamics simulations, which critically depend on an accurate description of the atomic interactions. Due to the high complexity of Co$_3$O$_4$, which is related to the presence of multiple oxidation states of the cobalt ions, to date \textit{ab initio} methods have been essentially the only way to reliably capture the underlying potential energy surface, while more efficient atomistic potentials are very challenging to construct. Consequently, the accessible length and time scales of computer simulations of systems containing Co$_3$O$_4$ are still severely limited. Rapid advances in the development of modern machine learning potentials (MLPs) trained on electronic structure data now make it possible to bridge this gap. In this work, we employ a high-dimensional neural network potential (HDNNP) to construct a MLP for bulk Co$_3$O$_4$ spinel based on density functional theory calculations. After a careful validation of the potential, we compute various structural, vibrational, and dynamical properties of the Co$_3$O$_4$ spinel with a particular focus on its temperature-dependent behavior, including the thermal expansion coefficient.

Materials Science

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is to develop a high - dimensional neural - network potential (HDNNP) to accurately describe the inter - atomic interactions in the Co₃O₄ spinel structure, especially the behavior at finite temperatures. Specifically, the article aims to:

1. **Improve simulation efficiency and accuracy**: Traditional ab - initio methods (such as density functional theory, DFT) can provide high - precision results, but due to the high computational cost, they limit the time and length scales that can be simulated. Empirical potential functions, although computationally efficient, have limited accuracy and it is difficult to construct models suitable for complex materials (such as Co₃O₄ containing cobalt ions with multiple oxidation states). By using a machine - learning - based high - dimensional neural - network potential, the simulation efficiency can be significantly improved while maintaining close to ab - initio accuracy.

2. **Study the properties of Co₃O₄ under catalytic conditions**: Co₃O₄, as an important oxidation catalyst, exhibits excellent performance in oxidation reactions. In order to better understand its performance in practical applications, especially the temperature - dependent behavior (such as the coefficient of thermal expansion), detailed molecular - dynamics simulations of its structure, vibration, and kinetic properties are required.

3. **Verify and develop new simulation tools**: By comparing with DFT reference data, verify the effectiveness and reliability of the constructed HDNNP and demonstrate its potential in describing materials with complex electronic structures.

### Specific objectives
- Construct a high - dimensional neural - network potential that can accurately describe the complex electronic structure of Co₃O₄.
- Through molecular - dynamics simulations, calculate various static, vibrational, and dynamic properties of Co₃O₄, especially the temperature - dependent behavior.
- Verify the applicability and accuracy of HDNNP at different temperatures to ensure that it can be used for further catalytic and electrochemical studies.

### Method overview
- **Reference electronic - structure calculation**: Use VASP software to perform DFT - based electronic - structure calculations, and use the optPBE - vdW+U functional to handle exchange - correlation effects.
- **Machine - learning - potential construction**: Use RuNNer code to construct HDNNP and continuously optimize the training set through active learning.
- **Molecular - dynamics simulation**: Use LAMMPS to perform molecular - dynamics simulations to evaluate the performance of HDNNP at different temperatures.
- **Phonon calculation**: Calculate the phonon - band structure and phonon - density of states through the Phonopy code to verify the ability of HDNNP to describe the vibrational characteristics of Co₃O₄.

Through these steps, the researchers hope to develop an efficient and accurate simulation tool to gain an in - depth understanding of the atomic - level mechanisms of Co₃O₄ in catalysis and electrochemistry.