C60 - C60 Collisions

In 1996, I performed some simulations of C60 - C60 collisions. The two buckyballs were each given an initial translation velocity corresponding to 100 eV per cage. To simulate an initial temperature, random velocities (with total mean zero) was given to the inidividual atoms. These random velocities corresponded to about 2% of the total initial kinetic energy. The potential energy was modelled with a semi-empirical* potential function for carbon-clusters. This function requires a term to be computed for every triple of atoms. This results in ~ 120^3 operations per step. To accurately resolve the collision, variable timestepping is used. The numerical integration scheme is a 4th-order time-reversible variable-stepsize method, based on the Adaptive Verlet method**.

* T. Takai, C. Lee, T. Halicioglu, and W.A. Tiller, A Model Potential Function for Carbon Systems, J. Phys. Chem. 94 , 4480--4482 (1990).

** W. Huang and B. Leimkuhler, The Adaptive Verlet Method, SIAM J. Sci. Comput. 18 , 239--256 (1997).

Here is a movie showing the collision. Bonds are drawn between any two atoms within bonding distance. The red bonds signify new bonds created via the high-speed collision.

Here are some results in the form of graphs (click to enlarge):

Energy Error

Variation in Step

Kinetic Energy

Potential Energy

Here are two movies showing the same collision at 115 eV and 45 eV .