(672b) Quantifying Error Propagation in Microkinetic Predictions of Dynamic Rate Enhancement

Programmable catalysis is an emerging catalyst design strategy in which temporal manipulation of a catalyst is predicted to enhance reaction rates by orders of magnitude. To date, there have been several microkinetic analyses of programmable catalysis by both our group and others.^[1-3] However, an understanding of how parameter uncertainty impacts microkinetic predictions of dynamic turnover frequency (TOF) enhancement is yet to be established. In this work, we elucidate the impact of error in linear scaling relations (LSR) and Brønsted-Evans-Polanyi (BEP) relations on predicted dynamic TOF enhancement. We begin our analysis by analyzing an A-to-B toy reaction under ideal differential conversion conditions. The uncertainty in model-predicted TOF is quantified using a distributed evaluation of local sensitivity analysis (DELSA); this method was chosen over a full variance-based global sensitivity analysis (e.g., Sobol’s method) due to its lower computational cost.^[4,5] The uncertainty space of LSR and BEP parameters are sampled quasi-randomly assuming a uniform distribution. After establishing how a simple reaction is affected by LSR and BEP uncertainty, we expand our analysis to more complex toy reaction mechanisms to investigate whether errors dampen out in series or parallel mechanisms. The microkinetic model and uncertainty analysis are implemented in Julia 1.8. Uncertainties in microkinetic predictions of rate for a static catalyst can be significant (ca. 3 orders of magnitude from the median prediction^[6]) and are not yet quantified for programmable catalysts. We present the first investigation of how LSR and BEP errors impacts microkinetic predictions of dynamic rate enhancement.

[1] ACS Catal. 2019, 9, 6929.

[2] Sci. Adv. 2022, 8, eabl6576.

[3] Chem Catalysis 2022, 2, 3497.

[4] AIChE J. 2022, 68, e17653.

[5] Adv. Theory Simul. 2022, 2200615.

[6] J. Catal. 2016, 338, 273.