No abstract

Large-scale atomistic computer simulations of materials heavily rely on interatomic potentials predicting the energy and Newtonian forces on atoms. Traditional interatomic potentials are based on physical intuition but contain few adjustable parameters and are usually not accurate. The emerging machine-learning (ML) potentials achieve highly accurate interpolation within a large DFT database but, being purely mathematical constructions, suffer from poor transferability to unknown structures. We propose a new approach that can drastically improve the transferability of ML potentials by informing them of the physical nature of interatomic bonding. This is achieved by combining a rather general physics-based model (analytical bond-order potential) with a neural-network regression. This approach, called the physically informed neural network (PINN) potential, is demonstrated by developing a general-purpose PINN potential for Al. We suggest that the development of physics-based ML potentials is the most effective way forward in the field of atomistic simulations.

Ni-Al-Co is a promising system for ferromagnetic shape memory applications. This paper reports on the development of a ternary embedded-atom potential for this system by fitting to experimental and first-principles data. Reasonably good agreement is achieved for physical properties between values predicted by the potential and values known from experiment and/or first-principles calculations. The potential reproduces basic features of the martensitic phase transformation from the B2-ordered high-temperature phase to a tetragonal CuAu-ordered low-temperature phase. The compositional and temperature ranges of this transformation and the martensite microstructure predicted by the potential compare well with existing experimental data. These results indicate that the proposed potential can be used for simulations of the shape memory effect in the Ni-Al-Co system.

We report on the development of an embedded-atom interatomic potential representing basic properties of both the hcp and the fcc phases of cobalt with nearly equal accuracy. The potential also reproduces the structural phase transformation between the two phases at a temperature close to the experimental value. The proposed potential can be used for large-scale atomistic simulations of cobalt microstructures over a wide range of temperatures. In a more general context, it offers a model for studying thermodynamic and kinetic properties of hcp/fcc interfaces and microstructure evolution in two-phase materials.

No abstract