Molecular dynamics investigations of slow relaxations in supercooled liquids

Molecular dynamics investigation of benzoic acid in confined spaces

Classical molecular dynamics simulations are carried out to investigate the aggregation of supercooled benzoic acid in confined spaces. Nanocavities, nanotubes and nanolayers are defined by restricting the periodicity of the simulation to zero, one or two dimensions, with boundaries set by adjustable, general, and computationally cheap van der Waals barriers. The effect of different confinement geometries is explored. It is found that the confinement impacts the liquid collective dynamics, strengthening the correlations that affect the motion of distant molecules. Overall, confinement determines up to a tenfold increase of the viscosity of the liquid and strongly slows down the rotational correlation times. Aggregation mediated by interactions with the walls and partial polarization of the liquid are observed. Additionally, transitions to high-density liquid states occur when stiffer barriers are used. In general, a reduced accessible amount of phase space fosters the struggle for a closer packing to relieve unfavorable atom-atom contacts, while maximizing the attractive ones. In benzoic acid, this implies that the hydrogen bond network is organized more efficiently in high density states.

Critical temperature T(c) and memory kernel in molecular-dynamics-simulated glass-forming Ni(0.2)Zr(0.8)

The mode-coupling theory (MCT) of dense liquids marks the dynamical glass transition by a critical temperature T(c) that is reflected in the temperature dependence of a number of physical quantities. Here, molecular-dynamics simulation data of a model adapted to Ni0.2Zr0.8 are analyzed to deduce T(c) from different quantities and to check the consistency of the estimated values. Analyzed are the critical temperature T(c) from (i). the nonvanishing nonergodicity parameters as asymptotic solutions of the MCT equations in the arrested state, (ii). the g(m) parameters describing the approach of the melt towards the arrested state on the ergodic side, (iii). the diffusion coefficients in the melt, and (iv). the alpha-relaxation time. The resulting T(c) values are found to agree within about 10%. In addition, the time dependent memory kernel is calculated from the MCT for the incoherent intermediate scattering function around T(c) and compared with the kernel obtained by inverting the molecular dynamics data for the corresponding correlator.

Fast Monte Carlo algorithm for supercooled soft spheres

We present a nonlocal Monte Carlo algorithm with particle swaps that greatly accelerates thermalization of soft sphere binary mixtures in the glassy region. Our first results show that thermalization of systems of hundreds of particles is achievable, and find behavior compatible with a thermodynamic glass transition.