Parameterization of a reactive force field using a Monte Carlo algorithm

Area of Science:

• Computational Chemistry
• Materials Science
• Chemical Physics

Background:

• Accurate molecular dynamics simulations require precise force field parameterization.
• ReaxFF, a reactive force field, uses numerous empirical parameters, making optimization challenging.
• Current parameter optimization methods, like single-parameter search, are often inefficient.

Purpose of the Study:

• To develop and implement a robust optimization method for ReaxFF force field parameters.
• To optimize ReaxFF parameters for MgSO4 hydrates using quantum chemical data.
• To assess the transferability of the optimized ReaxFF force field.

Main Methods:

• Utilized a metropolis Monte Carlo algorithm with simulated annealing for high-dimensional parameter space exploration.
• Optimized ReaxFF parameters against quantum chemical data for MgSO4 hydrates.
• Conducted transferability tests to evaluate the force field's applicability to different systems.

Main Results:

• The optimized ReaxFF force field accurately reproduced chemical structures, equations of state, and water binding curves for MgSO4 hydrates.
• The Monte Carlo approach demonstrated superior efficiency in navigating the complex parameter landscape.
• Transferability tests indicated limitations in the general applicability of the optimized force field.

Conclusions:

• The enhanced Monte Carlo optimization method significantly improves ReaxFF parameterization for specific molecular systems.
• The optimized ReaxFF force field provides reliable predictions for MgSO4 hydrates.
• The study underscores the importance of validating force field transferability, as ReaxFF is not universally applicable.