The introduction of microkinetic modeling in catalysis more than ten years ago promised to improve our understanding of the chemistry and eventually lead to better catalysts and reactors. However, this promise has been met with limited success only, despite several research efforts. A major obstacle in microkinetic model development is the huge number of parameters that need to be extracted from experimental data or to be estimated. First principles density functional theory (DFT) emerges as a powerful tool to assist with parameter estimation. However, the common materials and pressure gaps along with the uncertainty of parameter values make direct use of DFT a daunting task. Furthermore, the computational burden associated with parameter estimation for large reaction networks renders use of DFT impractical. Even if parameters are estimated, the reliability of a microkinetic model in extrapolation is questionable.

In this talk, we propose a hierarchical multiscale simulation framework for development of microkinetic models. The idea of hierarchical multiscale modeling and simulation is to start with the simplest possible “sound” model at each scale and identify the important scales and (‘active') model parameters at each scale. Once this is accomplished, one assesses the model accuracy by comparison with data and potentially improves the model of the important scale(s) and the associated active parameters using a higher-level model or theory. For example, the simplest identification tool employed extensively and successfully in chemical kinetics is local sensitivity analysis. Upon improvement of models and parameters, another iteration is taken until convergence is achieved, i.e., the important scales and parameters do not change between successive iterations. We will demonstrate the power of the approach with the specific examples of water gas shift, preferential oxidation, and steam reforming reactions on noble metals.

In order for these models to become useful, we discuss multiscale model-based design of experiments to optimize the chemical information content of a reaction mechanism in order to improve the fidelity and accuracy of reaction models. Extension of this framework to catalyst design will be touched upon. We illustrate this approach using the examples of ammonia decomposition on Ru, to produce hydrogen for PEM fuel cells, and the water-gas shift reaction on Pt, for converting syngas to hydrogen.