Kinetic modeling is a fundamental aspect of chemical process design, with the accuracy of the models depending heavily on the approach used to determine the mechanism and rate constants. The methods of kinetic exploration can be categorized as bottom-up and top-down approaches, where the former defines kinetic parameters on the microkinetic scale from first-principles calculations or surface science experiments and the latter constructs simpler but less physically relevant kinetic models at industrially relevant operating conditions.  Transient experiments straddle the line between these approaches, providing microkinetic detail on industrially relevant materials. Although identifying deterministic values of model parameters is crucial, quantifying uncertainty can be just as crucial to process development and optimization. The confidence in quantities of interest (QoI) (e.g. TOFs, apparent activation barriers, rate limiting steps, etc.) are often low when the associated uncertainty present in the complex microkinetic models have been propagated. However, uncertainty quantification techniques are less developed for transient kinetic methods.<br/>In this presentation, we explore the uncertainty of temporal analysis of products (TAP) reactor systems, one of the tools available in the transient kinetic repertoire. TAP is a reaction-diffusion system generating data on the milli-second time scale, which leads to the expression and elucidation of multiple elementary processes in a narrow experimental timespan. Two primary sources of uncertainty present in TAP are outlined, including the outlet flux (signal noise) and initial states (i.e. pulse intensities and delays, surface coverages, and void fractions). The outlet flux is explored using a Hessian-based approach present in the objective function space. The initial states are analyzed either through a sensitivity or ensemble-based approach, with the sensitivity providing rapid approximations of parameter uncertainties and the ensemble method providing more robust, but computationally expensive, uncertainties. We conclude by highlighting potential approaches that can be used to reduce these sources of uncertainty, including the observation of the surface coverages during the evolution of the catalyst to further constrain fitted parameters, as well as the quantitative selection of additional experiments based on the prior understanding of the microkinetic model.

Exploring sources of uncertainty in transient kinetic experiments

ACS Spring 2022

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.