MD Simulation for Head-on Collision of Liquid Nanodroplets Obeying Modified L-J Potential

Abstract

This project models and studies the `head-on' collision of liquid helium nanodroplets within a vacuum, using molecular dynamics simulation techniques. Programs written in MATLAB and C are utilized in tandem to facilitate computer experimentation that achieves this goal. The most expensive computation, that of collision simulation, is handled by a HPC cluster `ALICE' at the University of Leicester. Colliding droplets are modelled as roughly spherical collections of points, cut from a simple cubic lattice, obeying a modified Lennard-Jones potential, with average velocities initialized to ensure a `head-on' collision. These point-sets are then allowed to collide within a cuboid region, designed to take advantage of the observed angular distribution of post-collision fragmentation (favoring a plane orthogonal to `collision axis'). To implement the developed theoretical model, an existing C script by D. C. Rapaport, for modelling a homogeneous liquid state, is edited by the author to fit the given, highly heterogeneous scenario. To do analysis on the resulting positions and velocities of points in time, a precise definition of `droplet' was required. An object is defined, from the perspective of metric space, to satisfy this need. An existing cluster analysis algorithm, DBSCAN, is used to apply this definition to the points in simulation. The results presented here focus on three distinct properties of post-collision droplets, these being size, speed and temperature; for collision speed varying across ten equidistant averages chosen based on visual examination of collisions. Quantitative evidence of post-collision droplet speed being inversely proportional to droplet radius is presented, droplet temperature distribution post-collision is noted, and qualitative change in collision behavior across a certain threshold of collision speed is observed.