Statistical Analysis and Discovery of Heterogeneous Catalysts Based on Machine Learning from Diverse Published Data

Abstract

The literature provides insights for catalyst design and discovery. Effective analysis of reported data using machine learning (ML) methods offers the ability to gain valuable information. However, utilizing the literature in this way has obstacles such as lack of compositional overlaps, bias from prior published data, and low sample counts for many elements. The present study describes an ML approach that considers elemental features as input representations instead of inputting catalyst compositions directly. This ML method has the potential for catalyst discovery, including catalytic reactions with limited catalyst composition overlap in the available data. Oxidative coupling of methane (OCM), water gas shift (WGS), and CO oxidation reactions were chosen to confirm the effectiveness of the proposed method by analysis using several state‐of‐the‐art ML methods. Among the ML methods tested, gradient boosting regression with XGBoost (XGB) provided the best results, and prediction accuracy was improved by the proposed approach for all three reaction types. In addition, a quantitative value of “feature importance score” was calculated to evaluate the most influential input variables on catalyst performance. Finally, catalyst optimization was explored using ML as a “surrogate” model, and the top 20 promising candidate catalysts were identified for the OCM reaction based on the optimization. The advantages of ML in catalysis analysis as well as the difficulties and limitations originating from the complexity of heterogeneous catalysis were explored.

Analysis of Updated Literature Data up to 2019 on the Oxidative Coupling of Methane Using an Extrapolative Machine‐Learning Method to Identify Novel Catalysts

“…In our previous study, [49] we employed the catalytic OCM reaction data originally reported by Baerns and coworkers [42] for ML analysis. A large number of new OCM reaction data have been reported since this database was published.…”

“…The OCM reaction is one of the most studied heterogeneous catalytic reactions using ML and other relevant statistical analysis techniques with datasets obtained from high‐throughput experiments, published data, and computational studies [36–49] . Many of these studies have utilized the database of Baerns and coworkers, which consists of 1868 OCM reaction datapoints, includes the catalyst composition, experimental conditions, and the catalytic performance of the reactions, and was compiled from a wide range of data published before 2010 [42] .…”

Revisiting Machine Learning Predictions for Oxidative Coupling of Methane (OCM) based on Literature Data

“…Takigawa et al recently proposed the ML prediction model for OCM reaction with the crosscutting viewpoint of experimental C1 chemistry results including OCM (1833 datasets), water-gas shift (4185 datasets), and CO oxidation (5567 datasets), while representing differently promising catalyst candidates for OCM. [21] Such multiple views of literature data are expected to be interesting for future research. In summary, the authors investigated the ML predicted catalyst based on literature data for OCM with actual experimental conditions.…”

Reaction condition optimization for non-oxidative conversion of methane using artificial intelligence

“…[7][8][9] Although these approaches have been successfully applied to design new heterogeneous catalysts, machine learning (ML) methods have recently attracted considerable attention as they require smaller datasets and lower computational costs than traditional methods. 10 Moreover, in chemical reaction engineering, ML methods can play an essential role, for example, in self-optimizing platforms 11 and suggesting reaction conditions. 12 ML can match input variables (or features) to target properties and thus can be generally employed to available datasets for numerous catalytic reactions.…”

“…4 With the advancements in data science, chemical reaction data can be investigated with several approaches such as statistical analysis, 4 machine learning. 10,16,17 micro-kinetic simulations, 18 and meta-analysis. 19 Moreover, a recent experimental study demonstrated that the combination of high-throughput experiments and ML techniques can provide catalyst compositions with improved catalytic efficiencies for the OCM with less effort.…”