A SMART MONTE-CARLO TECHNIQUE FOR FREE-ENERGY SIMULATIONS OF MULTICONFORMATIONAL MOLECULES, DIRECT CALCULATIONS OF THE CONFORMATIONAL POPULATIONS OF ORGANIC-MOLECULES

Metropolis Monte Carlo (MMC) can be a highly inefficient simulation te chnique when only a small fraction of an energy surface is populated a nd barriers between low-energy regions are high. In such cases, previo us knowledge of the surface (e.g. low-energy conformations of molecule s) can be used to preferentially sample the significantly populated re gions. In this work we present a new MC method for accomplishing this goal. We term the method JBW for Jumping Between Wells. The JBW proced ure operates by locating the various conformations of a molecule and s ubsequently driving an MMC-like simulation to jump repeatedly between them. Using simulations on 1- and 2-dimensional potential surfaces and on n-pentane, the JBW method is shown to generate ensembles of states that are indistinguishable from the canonical ensembles generated by classical MMC in the limit. Integration of JBW into the recently descr ibed MC/SD hybrid simulation algorithm enables rapidly converged simul ations of conformationally flexible molecules including cyclic molecul es in all degrees of freedom. The new method (MC(JBW)/SD) gives conver ged comformational populations at a rate that is essentially independe nt of the energy barriers between conformations, We use the method to evaluate free energy differences between the conformers of various sub stituted cyclohexanes and of the larger ring hydrocarbons cycloheptane , cyclooctane, cyclononane and cyclodecane on several widely used pote ntial energy surfaces. Such conformational free energies are compared with simple molecular mechanics steric energies both with and without rigid rotor-harmonic oscillator free energy corrections. In general, w e find that assumptions of harmonicity do not lead to good approximati ons of the actual anharmonic free energies. In the case of cyclohexane derivatives at room temperature, the MC(JBW)/SD method is estimated t o generate converged ensembles of all conformations at a rate similar to 10(6) times faster than methods based on simple molecular or stocha stic dynamics.