Data-Centric Heterogeneous Catalysis: Identifying Rules and Materials Genes of Alkane Selective Oxidation

Abstract

Artificial intelligence (AI) can accelerate catalyst design by identifying key physicochemical descriptive parameters correlated with the underlying processes triggering, favoring, or hindering the performance. In analogy to genes in biology, these parameters might be called “materials genes” of heterogeneous catalysis. However, widely used AI methods require big data, and only the smallest part of the available data meets the quality requirement for data-efficient AI. Here, we use rigorous experimental procedures, designed to consistently take into account the kinetics of the catalyst active states formation, to measure 55 physicochemical parameters as well as the reactivity of 12 catalysts toward ethane, propane, and n-butane oxidation reactions. These materials are based on vanadium or manganese redox-active elements and present diverse phase compositions, crystallinities, and catalytic behaviors. By applying the sure-independence-screening-and-sparsifying-operator symbolic-regression approach to the consistent data set, we identify nonlinear property–function relationships depending on several key parameters and reflecting the intricate interplay of processes that govern the formation of olefins and oxygenates: local transport, site isolation, surface redox activity, adsorption, and the material dynamical restructuring under reaction conditions. These processes are captured by parameters derived from N2 adsorption, X-ray photoelectron spectroscopy (XPS), and near-ambient-pressure in situ XPS. The data-centric approach indicates the most relevant characterization techniques to be used for catalyst design and provides “rules” on how the catalyst properties may be tuned in order to achieve the desired performance.

Distilling Accurate Descriptors from Multi-Source Experimental Data for Discovering Highly Active Perovskite OER Catalysts

“…Such surface complexity may be explicitly incorporated to refine the data-driven models in the future, for example, by extracting surface information from experimental spectroscopic data . Recent work showed that the SISSO models can be improved if surface information obtained from in situ XPS measurements was included as input features …”

“…46 Recent work showed that the SISSO models can be improved if surface information obtained from in situ XPS measurements was included as input features. 47…”

Achieving Digital Catalysis: Strategies for Data Acquisition, Storage and Use

“…Neural networks and ensemble‐based methods, for example, offer little insight, [46, 47] but can help bridge the gap between different time and length scales of typical DFT calculations and reactor studies and enable fast screening in catalysts (materials) informatics [5h] . Interpretable ML algorithms, on the other hand, can reveal the physical laws underlying catalyst's properties and enable further development of catalyst systems based on hypotheses that follow from this analysis, which can accelerate catalyst discovery in a more sustainable manner [5a, 34b] . These transparent methods in catalysis informatics [5i] include, for example, tree‐based classification and regression methods [5a, 46] .…”

“…For this purpose, it is necessary to conduct systematic studies by using reliable experimental data. Interpretable machine learning can likewise contribute to identify relevant characterization methods [34b] …”

Tuning catalysis by surface-deposition of elements on oxidation catalysts via atomic layer deposition

“…With oxygenates as the target product (C 4 , C 3 ) the electronic properties of the bulk and the surface, as well as the adsorption properties of the catalyst are crucial. 40 Hence, the catalytic functionality of bulk catalysts is determined by the properties of the termination layer, which depends on the bulk composition and is formed under reaction conditions. 24,26,36,[41][42][43] However, the surface can additionally be modified using synthetic methods like atomic layer deposition (ALD).…”