Towards a multi-dimensional picture of classical nucleation: The example of liquid condensation through theory and rare event sampling

Abstract: While very popular for its ability to provide an elegant conceptual framework, classical nucleation theory (CNT) still fails at quantitatively reproducing most of the experimental and numerical observations of first order transitions. In this paper, we extend CNT by explicitly incorporating the multi-dimensional aspect of nucleation. Focusing on the liquid condensation of a Lennard-Jones gas, we used state of the art rare event sampling simulations to precisely characterize the nucleation event insisting on the specific features of the critical cluster. Our numerical results indicate a simultaneous growth and densification in liquid condensation. We then exploit these insights to develop a 2-variable nucleation theory. Our model based on the capillary approximation is able to quantitatively retrieve the numerical results in nucleation rate and critical cluster properties. Furthermore, our model provides a qualitatively more accurate representation of nucleation near the spinodal regime. The effectiveness of this integrated numerical and theoretical framework highlights the limitations of CNT and more recent theories while offering a robust foundation for its refinement.