Abstract
In this paper we present the results of several theoretical models examined in order to explain the observed systematic deviations of the homogeneous nucleation of highly polar substances from the predictions of the classical nucleation theory (CNT). It is found that molecular association in acetronitrile or nitromethane vapor cannot explain the observed higher supersaturations. Similarly, the use of virial coefficient data to obtain a size dependent surface energy correction factor does not reproduce the experimental results. It is shown that orientation of the surface dipoles within the spherical droplets results in higher supersaturations in agreement with the experimental findings. We also generalize the CNT for elongated clusters and show that these clusters can allow for favorable end‐to‐end and antiparallel alignments of the dipoles. It is found that for reasonable agreement with experiment, acetonitrile, and benzonitrile clusters require a larger degree of elongation compared to nitromethane and nitrobenzene clusters. These results are discussed in relation to molecular simulations of bulk liquids and clusters.