Parameter Estimation of Microkinetic Models for Heterogeneous Catalysis

Microkinetic models (MKMs) provide insights and understanding of surface kinetics, reaction mechanisms, and catalyst behavior for complex reaction systems. Estimation of the MKM parameters can be posed as a nonlinear optimization problem (NLP). However, it is often difficult to solve the NLP due to model stiffness, ill-conditioning, and nonconvexity. In this work, we first formulate an NLP that allows the incorporation of experimental and computational data, DFT calculations, and semiempirical relations, to estimate the MKM parameters. We also propose a robust method with an adaptive feasibility restoration to solve the NLP. To address the ill-conditioning commonly encountered in catalytic reaction networks, the algorithm uses a minimum-norm least-squares search direction with a projected backtracking line search that enforces physical bounds without introducing additional inequality multipliers. We utilize the algorithm to estimate rate parameters for multitemperature experimental data across different reaction systems. The strategy improves numerical robustness, solution quality, and computational efficiency and provides a robust approach for microkinetic parameter estimation in heterogeneous catalysis. The MKM parameter estimation code is made available here: such as https://github.com/SOULS-TAMU/MKMParameterEstimation.git.