Abstract:Predicting crystal structure has been a challenging problem in physics and materials science for a long time. A reliable energy calculation engine combined with an efficient global search algorithm, such as particle swarm optimization algorithm or genetic algorithm, is needed to conduct crystal structure prediction. In recent years, machine learning-based interatomic potential energy surface models have been proposed, potentially allowing us to perform crystal structure prediction for systems with the accuracy of density functional theory (DFT) and the speed of empirical force fields. In this paper, we employ a previously developed Deep Potential model to predict the intermetallic compound of the aluminum–magnesium system, and find six meta-stable phases with negative or nearly zero formation energy. In particular, Mg 12 Al 8 shows excellent ductility and Mg 5 Al 27 has a high Young's modulus. Based on our benchmark results, we propose a relatively robust structure screening criterion that selects potentially stable structures from the Deep Potential-based convex hull and performs DFT refinement. By using this criterion, the computational cost needed to construct the convex hull with ab initio accuracy can be dramatically reduced.
What problem does this paper attempt to address?
This paper aims to solve the problem of crystal structure prediction in alloy systems. Specifically, the author uses a multi - body potential energy surface model based on deep learning (Deep Potential, abbreviated as DP model), combined with the Particle Swarm Optimization algorithm (Particle Swarm Optimization, abbreviated as PSO), to predict the crystal structure of the aluminum - magnesium binary alloy system. This research not only covers a wide range of concentration ranges, but also discovers six new metastable phases. These phases have negative or near - zero formation energies, indicating that they may be synthesized in experiments.
### Research Background and Objectives
1. **Problem Definition**:
- Crystal structure prediction is a long - standing problem in physics and materials science.
- A reliable energy calculation engine and an efficient global search algorithm, such as the Particle Swarm Optimization algorithm or the genetic algorithm, are required to perform crystal structure prediction.
- In recent years, machine - learning - based inter - atomic potential energy surface models (PES) have been proposed, which can find a balance between the accuracy of density functional theory (DFT) and the speed of empirical force fields.
2. **Research Objectives**:
- Use the previously developed Deep Potential model to predict intermetallic compounds in the aluminum - magnesium binary alloy system.
- Discover and verify the thermodynamic stability, dynamic stability, and mechanical stability of these compounds.
- Propose a relatively robust structure screening criterion to reduce the computational cost required to construct the convex hull.
### Main Methods and Results
1. **Methods**:
- **Deep Potential Model**: Use a neural network architecture to parameterize two networks. One is an embedding network, which is used to define a set of symmetry - preserving descriptors; the other is a fitting network, which is used to map these descriptors to local energy contributions.
- **Particle Swarm Optimization Algorithm (PSO)**: Combine with the CALYPSO software package to generate random initial structures and gradually optimize the structures through local relaxation and global search.
- **DFT Verification**: Use DFT to further verify the screened structures to ensure their accuracy.
2. **Results**:
- Successfully predict six new metastable aluminum - magnesium alloy phases: Mg12Al8, Mg7Al9, Mg14Al18, Mg6Al10, Mg8Al16, and Mg5Al27.
- These phases have negative or near - zero formation energies, indicating that they may be synthesized in experiments.
- In particular, Mg12Al8 exhibits excellent ductility, while Mg5Al27 has a high Young's modulus.
### Conclusions
This research demonstrates the effectiveness of the DP + CALYPSO scheme in predicting the crystal structures of binary alloy systems over a wide concentration range. Through this method, six new metastable aluminum - magnesium alloy phases are successfully predicted, and their thermodynamic stability, dynamic stability, and mechanical stability are verified. In particular, Mg12Al8 exhibits excellent ductility, and Mg5Al27 has a high Young's modulus, which makes them have potential application values in the automotive, aerospace, electronics, and equipment industries.