研究目的
To assess the transferability and performance of various interatomic potentials for molybdenum and silicon by comparing them with DFT reference calculations and experimental data across a wide range of materials properties.
研究成果
No single interatomic potential performs best across all properties for Mo or Si; each has strengths and weaknesses. Transferability varies, and potentials fitted to specific data (e.g., MD trajectories) show better overall transferability. Property correlations from DFT are not always reproduced by potentials. Careful assessment is essential before using a potential for specific simulations. The data provided aids in selecting appropriate potentials for Mo or Si simulations.
研究不足
Interatomic potentials have limited transferability and may not perform well for all properties or structures. DFT itself has deviations from experimental data (e.g., elastic constants). The study focuses on basic materials properties (level II validation) and does not extensively cover complex applications (level III) due to computational costs. The set of random structures may not fully represent disordered or liquid phases.
1：Experimental Design and Method Selection:
High-throughput calculations using DFT and interatomic potentials for Mo and Si. DFT calculations performed with VASP using LDA and GGA-PBE functionals. Interatomic potentials sourced from OpenKIM and NIST repositories, including Morse, EAM, MEAM, ADP, SW, Tersoff, EDIP, and BOP types. Computational management framework pyiron used for handling simulations and protocols.
2：Sample Selection and Data Sources:
Focus on molybdenum and silicon elements. Structures include ground-state crystal structures (bcc for Mo, diamond for Si), prototypes from Strukturbericht group A, and random structures with one or two atoms per unit cell to cover diverse atomic environments.
3：List of Experimental Equipment and Materials:
Software: VASP for DFT, LAMMPS, ASE, BOPfox for interatomic potential calculations, PHONOPY for phonon calculations. Computational resources for high-throughput simulations.
4：Experimental Procedures and Operational Workflow:
For each structure, calculate energy, forces, stresses. Relax structures to equilibrium. Compute properties: cohesive energy, atomic volume, elastic constants, phonon spectra, thermodynamic properties (heat capacity, thermal expansion), surface energies, vacancy formation energies, transformation paths. Use small displacement method for phonons, energy-volume fitting for bulk modulus, molecular dynamics for thermal expansion.
5：Data Analysis Methods:
Compare predictions to DFT and experimental data. Use Spearman correlation for property correlations. Define Boltzmann-weighted mean-absolute deviation to assess transferability. Analyze deviations and errors across properties.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
