|
Barend Thijsse1, Erik Neyts2, Maarten Mees3 1Department of Materials Science and Engineering, Delft University of Technology, The Netherlands (b.j.thijsse@tudelft.nl). 2Department of Chemistry, University of Antwerp, Belgium (erik.neyts@ua.ac.be). 3Department of Physics, Katholieke Universiteit Leuven, and IMEC, Heverlee, Belgium (meesm@imec.be). Force-biased Monte Carlo is a useful addition to a Molecular Dynamics code, since the force algorithm is already available so that a few extra lines are sufficient to give access to an entirely different and complementary simulation technique. MC and MD can take turns in an alternating sequence, where MD takes care of the violent changes and MC handles the equilibration processes. In this presentation we show how the particular uniform-acceptance MC version that we use - i.e., all stochastic moves are accepted - outperforms MD in diffusion, phase transformation (Si), and nanotube growth (C). The method appears to work always, does not depend on event detection, and is not sensitive to, e.g., very low activation barriers. New in this work are the proof of a rigid statistical mechanics underpinning of the method and the introduction of statistical time, allowing this MC technique to provide unbiased estimates of activation kinetics. We will also discuss recent indications that the method may in fact not "always" work. |
|