Computing Surface Reaction Rates by Adaptive Multilevel Splitting Combined with Machine Learning and

Area of Science:

• Computational Chemistry
• Surface Science
• Chemical Kinetics

Background:

• Calculating catalytic reaction rates is crucial for chemical applications but remains challenging.
• Traditional methods often oversimplify complex surface phenomena and entropic contributions.

Purpose of the Study:

• To develop and evaluate a computational approach combining adaptive multilevel splitting (AMS) and ab initio molecular dynamics for accurate catalytic rate constant computation.
• To investigate the role of reaction coordinates and entropic effects in surface catalysis.

Main Methods:

• Employed adaptive multilevel splitting (AMS), a rare event sampling technique.
• Integrated AMS with ab initio molecular dynamics and density functional theory (DFT) calculations.
• Explored various reaction coordinate identification methods, including support vector machines and path collective variables.
• Studied water molecule adsorption and dissociation on a γ-alumina surface as a case study.

Main Results:

• The AMS approach successfully computed rate constants and transition mechanisms for surface catalytic events.
• Identified significant entropic effects contributing to rate constants.
• Calculated rate constants were up to two orders of magnitude smaller than those from static approaches (Eyring-Polanyi equation).

Conclusions:

• The combination of AMS and ab initio molecular dynamics provides a more accurate method for calculating catalytic rate constants.
• This approach is essential for understanding complex surface reactions where entropic factors are significant.
• The findings offer improved insights into material catalysis and surface chemistry.