While the transferable potentials for phase equilibria-united atom (TraPPE-UA) force field has generally been successful at providing parameters that are highly transferable between different molecules, the polarity and polarizability of a given functional group can be significantly perturbed in small cyclic structures, which limits the transferability of parameters obtained for linear molecules. This has motivated us to develop a version of the TraPPE-UA force field specifically for five- and six-membered cyclic alkanes and ethers. The Lennard-Jones parameters for the methylene group obtained from cyclic alkanes are transferred to the ethers for each ring size, and those for the oxygen atom are common to all compounds for a given ring size. However, the partial charges are molecule specific and parametrized using liquid-phase dielectric constants. This model yields accurate saturated liquid densities and vapor pressures, critical temperatures and densities, normal boiling points, heat capacities, and isothermal compressibilities for the following molecules: cyclopentane, tetrahydrofuran, 1,3-dioxolane, cyclohexane, oxane, 1,4-dioxane, 1,3-dioxane, and 1,3,5-trioxane. The azeotropic behavior and separation factor for the binary mixtures of 1,3-dioxolane/cyclohexane and ethanol/1,4-dioxane are qualitively reproduced.

The aggregation-volume-bias Monte Carlo based simulation technique, which has led to our recent success in vapor-liquid nucleation research, was extended to the study of crystal nucleation processes. In contrast to conventional bulk-phase techniques, this method deals with crystal nucleation events in cluster systems. This approach was applied to the crystal nucleation of Lennard-Jonesium under a wide range of undercooling conditions from 35% to 13% below the triple point. It was found that crystal nucleation in these model clusters proceeds initially via a vapor-liquid like aggregation followed by the formation of crystals inside the aggregates. The separation of these two stages of nucleation is distinct except at deeper undercooling conditions where the crystal nucleation barrier was found to diminish. The simulation results obtained for these two nucleation steps are separately compared to the classical nucleation theory (CNT). For the vapor-liquid nucleation step, the CNT was shown to provide a reasonable description of the critical cluster size but overestimate the barrier heights, consistent with previous simulation studies. On the contrary, for the crystal nucleation step, nearly perfect agreement with the barrier heights was found between the simulations and the CNT. For the critical cluster size, the comparison is more difficult as the simulation data were found to be sensitive to the definition of the solid cluster, but a stringent criterion and lower undercooling conditions generally lead to results closer with the CNT. Additional simulations at undercooling conditions of 40% or above indicate a nearly barrierless transition from the liquid to crystalline-like structure for sufficiently large clusters, which leads to further departure of the barrier height predicted by the CNT from the simulation data for the aggregation step. This is consistent with the latest experimental results on argon that show an unusually large underestimation of the nucleation rate by the CNT toward deep undercooling conditions.

Monte Carlo simulations of homogeneous nucleation for a protein model with an exceedingly short-ranged attractive potential yielded a nonconventional crystal nucleation mechanism, which proceeds by the formation of fractal, low-dimensional aggregates followed by a concurrent collapse and increase of the crystallinity of these aggregates to become compact ordered nuclei. This result corroborates a recently proposed two-step mechanism for protein crystal nucleation from solution.

Could two like ions be attractive to each other in the presence of water? To address this question and to further interrogate the intriguing solvent effects at a molecular level on multiply charged species, a "bottom-up" simulation approach was formulated, from which the inter-ionic potential of mean force and other properties were monitored closely with the gradual addition of the water molecules. This approach was first tested on a commonly studied ion pair (namely, Na+ and Cl-), where excellent agreement with the published bulk-phase data was found. Further application of this approach to the like-ion pair indicated that an attractive interaction between two anions or two cations can be induced by the addition of an appropriate number of water molecules. This result corroborates a recent experimental report of an intriguing folding of a dianionic polymer into a more compact structure with the addition of water molecules in gas phase as well as previous theoretical findings of possible attraction between like-ion pairs in bulk aqueous phases.

Fragment methods have been widely studied for computing energies and forces, but less attention has been paid to nonenergetic properties. Here we extend the electrostatically embedded many-body (EE-MB) method to the calculation of cluster dipole moments, dipole moments of molecules in clusters, partial atomic charges, and charge transfer, and we test and validate the method by comparing to results calculated for the entire system without fragmentation. We also compare to calculations carried out by the conventional many-body (MB) method without electrostatic embedding. Systems considered are NH(3)(H(2)O)(11), (NH(3))(2)(H(2)O)(14), [Cl(H(2)O)(6)](-), (HF)(4), (HF)(5), (HF)(2)H(2)O, (HF)(3)H(2)O, and (HF)(3)(H(2)O)(2). With electrostatic embedding, we find that even at the pairwise additive level a quantitatively accurate description of a system's dipole moment and partial charge distribution and a qualitatively accurate description of the amount of intermolecular charge transfer can often be obtained.