The use of machine learning to develop neural network potentials (NNP) representing the interatomic potential energy surface allows us to achieve an optimal balance between accuracy and efficiency in computer simulation of materials. A key point in developing such potentials is the preparation of a training dataset of ab initio trajectories. Here we apply a deep potential molecular dynamics (DeePMD) approach to develop NNP for silica, which is the representative glassformer widely used as a model system for simulating network-forming liquids and glasses. We show that the use of a relatively small training dataset of high-temperature ab initio configurations is enough to fabricate NNP, which describes well both structural and dynamical properties of liquid silica. In particular, we calculate the pair correlation functions, angular distribution function, velocity autocorrelation functions, vibrational density of states, and mean-square displacement and reveal a close agreement with ab initio data. We show that NNP allows us to expand significantly the time-space scales achievable in simulations and thus calculating dynamical and transport properties with more accuracy than that for ab initio methods. We find that developed NNP allows us to describe the structure of the glassy silica with satisfactory accuracy even though no low-temperature configurations were included in the training procedure. The results obtained open up prospects for simulating structural and dynamical properties of liquids and glasses via NNP.

We investigate glassy dynamical properties of one-component three-dimensional system of particles interacting via pair repulsive potential by the molecular dynamic simulation in the wide region of densities. The glass state is superfragile and it has high glass-forming ability. The glass transition temperature $T_g$ has a pronounced minimum at densities where the frustration is maximal.

Using molecular dynamics simulations with embedded atom model potential, we study structural evolution of Cu_64.5Zr_35.5 alloy during the cooling in a wide range of cooling rates γ ∈ (1.5 ⋅ 10⁹, 10¹³) K/s. Investigating short- and medium-range orders, we show that the structure of Cu_64.5Zr_35.5 metallic glass essentially depends on cooling rate. In particular, a decrease of the cooling rate leads to an increase of abundances of both the icosahedral-like clusters and Frank-Kasper Z16 polyhedra. The amounts of these clusters in the glassy state drastically increase at the γ_min = 1.5 ⋅ 10⁹ K/s. Analysing the structure of the glass at γ_min, we observe the formation of nano-sized crystalline grain of Cu₂Zr intermetallic compound with the structure of Cu₂Mg Laves phase. The structure of this compound is isomorphous with that for Cu₅Zr intermetallic compound. Both crystal lattices consist of two types of clusters: Cu-centered 13-atom icosahedral-like cluster and Zr-centered 17-atom Frank-Kasper polyhedron Z16. That suggests the same structural motifs for the metallic glass and intermetallic compounds of Cu–Zr system and explains the drastic increase of the abundances of these clusters observed at γ_min.

The local order units of dense simple liquid are typically three-dimensional (close packed) clusters: hcp, fcc, and icosahedrons. We show that the fluid demonstrates the superstable tetrahedral local order up to temperatures several orders of magnitude higher than the melting temperature and down to critical density. While the solid-like local order (hcp, fcc) disappears in the fluid at much lower temperatures and far above critical density. We conclude that the supercritical fluid shows the temperature (density)-driven two-stage “melting” of the three-dimensional local order. We also find that the structure relaxation times in the supercritical fluid are much larger than ones estimated for weakly interactive gas even far above the melting line.

Using molecular dynamics simulations we show that a one-component system with purely repulsive two-length-scale potential can form decagonal quasicrystals.