Lattice kinetic Monte Carlo simulations have become a vital tool forpredictive quality atomistic understanding of complex surface chemical reactionkinetics over a wide range of reaction conditions. In order to expand theirpractical value in terms of giving guidelines for atomic level design ofcatalytic systems, it is very desirable to readily evaluate a sensitivityanalysis for a given model. The result of such a sensitivity analysisquantitatively expresses the dependency of the turnover frequency, being themain output variable, on the rate constants entering the model. In the past theapplication of sensitivity analysis, such as Degree of Rate Control, has beenhampered by its exuberant computational effort required to accurately samplenumerical derivatives of a property that is obtained from a stochasticsimulation method. In this study we present an efficient and robust three stageapproach that is capable of reliably evaluating the sensitivity measures forstiff microkinetic models as we demonstrate using CO oxidation on RuO2(110) asa prototypical reaction. In a first step, we utilize the Fisher InformationMatrix for filtering out elementary processes which only yield negligiblesensitivity. Then we employ an estimator based on linear response theory forcalculating the sensitivity measure for non-critical conditions which coversthe majority of cases. Finally we adopt a method for sampling coupled finitedifferences for evaluating the sensitivity measure of lattice based models.This allows efficient evaluation even in critical regions near a second orderphase transition that are hitherto difficult to control. The combined approachleads to significant computational savings over straightforward numericalderivatives and should aid in accelerating the nano scale design ofheterogeneous catalysts.
展开▼
