Addressing global uncertainty and sensitivity in first-principles based microkinetic models by an adaptive sparse...

Area of Science:

• Computational Chemistry
• Heterogeneous Catalysis
• Materials Science

Background:

• First-principles microkinetic modeling is crucial for understanding heterogeneous catalysis mechanisms.
• Approximate Density Functional Theory (DFT) introduces errors into these models.
• The impact of DFT errors on catalytic turnover frequency (TOF) is not well understood.

Purpose of the Study:

• To analyze the propagation of DFT errors to catalytic turnover frequency (TOF) in microkinetic models.
• To develop and apply efficient methods for global sensitivity and uncertainty analysis.
• To assess the reliability of mechanistic conclusions drawn from uncertain catalytic models.

Main Methods:

• Employed adaptive sparse grid quadrature for efficient high-dimensional integration.
• Utilized global sensitivity and uncertainty analysis to quantify error propagation.
• Applied a maximum entropy error model with defined bounds on DFT errors.

Main Results:

• DFT errors can introduce uncertainties of several orders of magnitude in TOF.
• Despite significant uncertainties, reliable conclusions about atomistic reactivity controls can be drawn.
• Adaptive sparse grids enable analysis with a modest number of function evaluations.

Conclusions:

• Global sensitivity analysis using adaptive sparse grids is effective for studying DFT error propagation in microkinetic models.
• This approach allows for drawing meaningful mechanistic insights even from models with substantial uncertainties.
• The methodology is extendable to more complex models, including kinetic Monte Carlo simulations.