We present a new methodology to study nucleation by means of molecular simulation. A conventional approach in the field has been to use umbrella sampling technique, in which a bias potential given in terms of a set of order parameters forces the system to visit high free energy states, including the one containing a critical nucleus, so that the free energy barrier of nucleation can be evaluated accurately. This approach, however, requires one to identify order parameters that can describe nucleation.

During the course of our study of bubble nucleation, we have tested many order parameters and found that they cannot control the free barrier crossing in a reversible fashion, rendering impossible the accurate evaluation of free energy barrier by this method. In fact, identification of a proper set of order parameters presents a considerable challenge in the conventional use of umbrella sampling technique.

The new approach we developed still uses the umbrella sampling technique, but only to drive the system out of the metastable free energy well. Thus, the bias potential is made to vanish well before the system reaches critical nucleus, and hence has no effect on nucleation behavior except to greatly accelerate the barrier crossing. A method has been devised to evaluate the unbiased first passage time, i.e., the time required for the system to cross over the free energy barrier in the absence of biasing potential, from the biased first passage time evaluated by simulation. We verified that the unbiased first passage time thus obtained is independent of the details of the imposed bias.

Our current implementation of the method is through Monte Carlo method, which allows for extraction of purely thermodynamic information on nucleation behavior decoupled from dynamical aspects of the process. However, we note that the method can also be implemented with molecular dynamics to yield dynamical information.

Results for both bubble nucleation and vapor phase condensation will be reported.