(429e) Extracting Knowledge from Transient Kinetic Datasets

Transient kinetic techniques such as temporal analysis of products (TAP) experiments provide rich datasets that contain intrinsic information about multiple elementary steps of a catalytic reaction. However, the resulting data is governed by multiple physical processes arising from transport, thermodynamics, and reaction kinetics. Even in the case of TAP reactors with well-defined transport, it can be difficult to disentangle these phenomena, especially for complex multi-step reaction mechanisms. This talk will introduce the TAPSolver code, an open-source finite-element code for numerical simulation of TAP data that includes the ability to simulate and fit data based on complex reaction mechanisms. The talk will highlight the capabilities of TAPSolver and show results for several case studies including CO oxidation and propane dehydrogenation. In addition, the impact of uncertainty in both experimental measurements and initial conditions on the intrinsic kinetic parameters extracted from TAP data will be discussed. The findings show that kinetic parameters extracted from TAP data have numerical uncertainty that is significantly lower than typical computational chemistry techniques such as DFT, at least in the case where the mechanism is well-defined and involves relatively few elementary steps. Finally, the talk will include discussion of current and future directions in the analysis of TAP datasets, including the use of "kinetic-informed neural networks" for fitting multi-pulse and spectro-kinetic data. The results indicate that the combination of TAP data and numerical methods is a powerful strategy for elucidating the intrinsic kinetic parameters of complex catalytic reactions.