(329e) Automated Screening of Catalytic Reaction Mechanisms: A Tool for Novel Chemistries

The development of kinetic models for novel heterogeneously-catalyzed reactions can be significantly boosted by automating the screening of potential mechanisms against experimental data. To the best of our knowledge, no tool nor methodology exists to automatically generate and screen reaction mechanisms without a priori knowledge of the catalytic system. The goal of this work was, hence, to conceptualize, develop and validate such a tool.

The tool is divided into three parts: (1) the features of the experimental data are extracted; (2) the theoretical curves are generated from the rate equations in the library and, for the specific operating conditions of the dataset, their features are extracted; and (3) the features of the theoretical curves are compared to the features in the experimental data and rate equations that cannot describe the data are eliminated. This tool was implemented in Python and is available as user-friendly software.

To evaluate the performance of the tool, a total of five case studies with literature data were evaluated. Here, the most representative one is presented, i.e. on ethanol dehydrogenation to acetaldehyde ¹ (see figure). As identified by the tool in part 1, there are two features: first an increase in reaction rate with pressure and then, after reaching a maximum, a decrease with pressure increase above ca. 5 bar (see figure).

In part 2, six different mechanisms were generated which correspond to four rate equations. As it is visually clear in the figure, the features in the experimental data can only be mathematically reproduced by the rate equation for which all species adsorb molecularly and the rate-determining step is the surface reaction (middle left). In part 3, the tool effectively proposed only this mechanism which corresponds precisely to the one determined by the researchers ¹.

1. J. Franckaerts, G. F. Froment, 1964, 19, 807-818.