(25c) Analyzing and Compressing the Parametric Space for Modeling and Design of Heterogeneous Catalytic Systems

Authors: 
Rangarajan, S., University of Wisconsin-Madison
Maravelias, C. T., University of Wisconsin-Madison
Mavrikakis, M., University of Wisconsin-Madison

Heterogeneous catalytic reactions can involve several gaseous and surface species and numerous interlinking reactions. The development of microkinetic models and identification of optimal catalyst properties requires detailed evaluation of the energetics of the reaction network at the atomic scale and relating it to the macroscopic observables such as rate and selectivity; typical approaches include using direct input or linear correlations developed from density functional theory. A universal property of nonlinear systems is that model response is dominated by a few important (or “stiff”) and several unimportant (or “sloppy”) parameters [1]. Heterogeneous catalytic systems, too, are expected to be similar.

In this study, we adopt a derivative-based global sensitivity analysis and hierarchical clustering based method to elucidate several heterogeneous catalytic systems to show that: (a) few parameters – binding energy of species and activation barrier of elementary reaction steps – are globally sensitive and form a superset of locally sensitive parameters, (b) the span of parametric sensitivity can vary several orders of magnitude, and (c) several parameters are statistically correlated and hence form clusters of “similar” parameters. This analysis shows that the dimensions of the microkinetic model can be significantly reduced and the catalyst design space compressed prior to applying further simplifying physico-chemical approximations such as semi-empirical correlations of the energetics. The resultant reduced model and compressed parameter space can then be the basis for both mechanistic analysis and catalyst design.

Our proposed method and application to representative heterogeneous catalytic systems will be discussed.

  1. Machta, B. B., Chachra, R., Transtrum, M. K., and Sethna, J. Science (2013), 342, 6158, 604-607