Finite temperature string by K-means clustering sampling with order parameters as collective variables for molecular...

Area of Science:

• Molecular Solid-State Chemistry
• Computational Materials Science

Background:

• Polymorphic transformation in molecular crystals is crucial for various industries but poorly understood at the molecular level.
• Simulating these transformations is challenging due to difficulties in enhanced sampling and defining order parameters (OPs) for complex structures.

Purpose of the Study:

• To elucidate the molecular-level mechanism of polymorphic transformation between β-CL-20 and ε-CL-20.
• To develop and apply effective computational methods for studying phase transitions in caged molecular crystals.

Main Methods:

• Constructed two types of order parameters (OPs) using bond distances, bond orientations, and relative orientations.
• Employed a K-means clustering algorithm to refine the finite temperature string (FTS) for improved convergence.
• Utilized the string method in collective variables, Markovian milestoning, and Voronoi tessellations to determine the minimum free energy path and nucleation kinetics.

Main Results:

• The K-means clustering algorithm demonstrated a faster convergence rate for FTS compared to average-based sampling.
• Simulations indicated that the transformation is influenced by finite-size effects but independent of the specific OP types used.
• Confirmed a surface-mediated local nucleation mechanism, identifying the critical nucleus at the shoulder of the potential of mean force.

Conclusions:

• The study provides an effective computational approach for exploring polymorphic transformations in caged molecular crystals.
• The findings highlight a surface-mediated nucleation mechanism and the importance of intermolecular interactions in critical nucleus formation.
• This work advances the understanding of molecular-level phase transitions in crystalline materials.