Simulating Phase-Transitions of Hard Spherocylinders in the pe Physics Engine

Mario Heene, Matthias Hofmann, Paul Manstetten, David Staubach

The project team realised this honours project at the (ZISC) in cooperation with Klaus Mecke (Theo.Phys. I).

The aim of this project is the verification of former simulation results in for fluids consisting of nonspherical hard particles using the pe Physics Engine, a simulator for rigid body dynamics. The nematic to isotropic phase-transition for hard sphreocylinders is simulated using a simulation domain with periodic boundary conditions divided into parallel processes with some 10K particles in total.

Figure 1: Phase configurations for different densities of Hard Spherocylinders. Beginning at low density in the isotropic phase (top left) the nematic phase (top right) follows before the smectic phase (bottom left) and finally the solid phase (bottom right) established for high densities.

To resolve the collisions between the rigid bodies, each representing a hard spherocylinder, a FFD (Fast Friction Dynamics) solver is used. The ensemble of hard spherocylinders starts on a regular grid with random initial velocities. Phase boundaries are identified by computing a nematic order parameter for each simulation run. Simulation runs are done for a certain length-to-diameter ratio of the hard spherocylinders and different volume densities in the domain. The nematic order parameter provides information about the orientational distribution of the ensemble of particles.

To conserve the energy of the ensemble in long simulation runs with a large number of hard spherocylinders, an energy correction is applied. The new simulation results validate the existing simulation results for a length-to-diameter ratio of 30 that were produced using different simulation approaches. Also results for a length-to-diameter ratio of 10 are produced that have not been reported in the existing simulation results.

Figure 2: Converged nematic order parameters for L/D = 30, revealing a constant, a linear and a logarithmic region. Additionally the green and red bars indicate the bounds reported in Bolhuis’ work.

A detailed project report can be found here.