Learning Design Rules for Selective Oxidation Catalysts from High-Throughput Experimentation and Artificial Intelligence

Abstract

The design of heterogeneous catalysts is challenged by the complexity of materials and processes that govern reactivity and by the fact that the number of good catalysts is very small in comparison to the number of possible materials. Here, we show how the subgroup-discovery (SGD) artificial-intelligence approach can be applied to an experimental plus theoretical data set to identify constraints on key physicochemical parameters, the so-called SG rules , which exclusively describe materials and reaction conditions with outstanding catalytic performance. By using high-throughput experimentation, 120 SiO 2 -supported catalysts containing ruthenium, tungsten, and phosphorus were synthesized and tested in the catalytic oxidation of propylene. As candidate descriptive parameters, the temperature and 10 parameters related to the composition and chemical nature of the catalyst materials, derived from calculated free-atom properties, were offered. The temperature, the phosphorus content, and the composition-weighted electronegativity are identified as key parameters describing high yields toward the value-added oxygenate products acrolein and acrylic acid. The SG rules not only reflect the underlying processes particularly associated with high performance but also guide the design of more complex catalysts containing up to five elements in their composition.

Platinum Group Metal-Doped Tungsten Phosphates for Selective C–H Activation of Lower Alkanes

“…Small amounts of CO and acetic acid were identified for Ru/P-ratios of 1/3 (Supporting Information Figure S18). Foppa et al used the high-throughput data set of these 109 catalysts and their catalytic results as the target for the subgroup discovery (SGD) artificial intelligence approach. They combined the experimental results with theoretical data sets from DFT calculations and could identify that temperature, phosphorous, composition-weighted electronegativity are critical parameters in order to achieve high yields of acrolein and acrylic acid.…”

A Route Map of Machine Learning Approaches in Heterogeneous CO2 Reduction Reaction

“…This can also be called another type of supervised learning method, as it uses labeled data. Scheffler and co-workers have explored the various aspects of materials science including catalysis with different data sets using this technique. − Recently, they have also employed this strategy to identify the potential indicators of CO 2 activation on metal oxide-based catalytic surfaces such as elongation of the C–O bond and bending of the O–C–O angle with respect to 180° and many others . The varying cutoff values of these two parameters along with other parameters lead to various subgroups, among which, in some cases, strong overlap between different subgroups has been observed.…”

Machine Learning: A New Paradigm in Computational Electrocatalysis

“…Therefore, sometimes it would be better to identify the physically similar subgroups and obtain the local models to work well in each subdomain. For descriptive and exploratory data mining, subgroup discovery (SGD) could be used to explore descriptions of the subgroups of a data set that exhibit intriguing behaviors in light of certain interestingness criteria. , Recently it has been applied in catalysis to reveal the similarity between the interesting catalysts. ,− An example is that Mazheika et al used SGD to identify one or more distinct combinations of catalyst features that trigger, facilitate or hinder the activation of CO 2 . Their results indicate that the surfaces of experimentally identified promising catalysts consistently exhibit combinations of features leading to strong elongation of C–O bonds.…”