Learning Catalyst Design Based on Bias-Free Data Set for Oxidative Coupling of Methane

MonteCat: A Basin-Hopping-Inspired Catalyst Descriptor Search Algorithm for Machine Learning Models

Proposing relevant catalyst descriptors that can relate the information on a catalyst's composition to its actual performance is an ongoing area in catalyst informatics, as it is a necessary step to improve our understanding on the target reactions. Herein, a small descriptor-engineered data set containing 3289 descriptor variables and the performance of 200 catalysts for the oxidative coupling of methane (OCM) is analyzed, and a descriptor search algorithm based on the workflow of the Basin-hopping optimization methodology is proposed to select the descriptors that better fit a predictive model. The algorithm, which can be considered wrapper in nature, consists of the successive generation of random-based modifications to the descriptor subset used in a regression model and adopting them depending on their effect on the model's score. The results are presented after being tested on linear and Support Vector Regression models with average cross-validation r2 scores of 0.8268 and 0.6875, respectively.

Automatic feature engineering for catalyst design using small data without prior knowledge of target catalysis

The empirical aspect of descriptor design in catalyst informatics, particularly when confronted with limited data, necessitates adequate prior knowledge for delving into unknown territories, thus presenting a logical contradiction. This study introduces a technique for automatic feature engineering (AFE) that works on small catalyst datasets, without reliance on specific assumptions or pre-existing knowledge about the target catalysis when designing descriptors and building machine-learning models. This technique generates numerous features through mathematical operations on general physicochemical features of catalytic components and extracts relevant features for the desired catalysis, essentially screening numerous hypotheses on a machine. AFE yields reasonable regression results for three types of heterogeneous catalysis: oxidative coupling of methane (OCM), conversion of ethanol to butadiene, and three-way catalysis, where only the training set is swapped. Moreover, through the application of active learning that combines AFE and high-throughput experimentation for OCM, we successfully visualize the machine's process of acquiring precise recognition of the catalyst design. Thus, AFE is a versatile technique for data-driven catalysis research and a key step towards fully automated catalyst discoveries.

Machine Learning Descriptors for Data-Driven Catalysis Study

Traditional trial-and-error experiments and theoretical simulations have difficulty optimizing catalytic processes and developing new, better-performing catalysts. Machine learning (ML) provides a promising approach for accelerating catalysis research due to its powerful learning and predictive abilities. The selection of appropriate input features (descriptors) plays a decisive role in improving the predictive accuracy of ML models and uncovering the key factors that influence catalytic activity and selectivity. This review introduces tactics for the utilization and extraction of catalytic descriptors in ML-assisted experimental and theoretical research. In addition to the effectiveness and advantages of various descriptors, their limitations are also discussed. Highlighted are both 1) newly developed spectral descriptors for catalytic performance prediction and 2) a novel research paradigm combining computational and experimental ML models through suitable intermediate descriptors. Current challenges and future perspectives on the application of descriptors and ML techniques to catalysis are also presented.

What do we talk about, when we talk about single-crystal termination-dependent selectivity of Cu electrocatalysts for CO2 reduction? A data-driven retrospective

We mine from the literature experimental data on the CO₂ electrochemical reduction selectivity of Cu single crystal surfaces. We then probe the accuracy of a machine learning model trained to predict faradaic efficiencies for 11 CO₂ reduction reaction products, as a function of the applied voltage at which the reaction takes place, and the relative amounts of non equivalent surface sites, distinguished according to their nominal coordination. A satisfactory model accuracy is found only when discriminating data according to their provenance. On one hand, this result points at a qualitative agreement across reported experimental CO₂ reduction reactions trends for single-crystal surfaces with well-defined terminations. On the other, this finding hints at the presence of differences in nominally identical catalysts and/or CO₂ reduction reaction measurements, which result in quantitative disagreement between experiments.

Catalysts informatics: paradigm shift towards data-driven catalyst design

Designing catalysts is a challenging matter as catalysts are involved with various factors that impact synthesis, catalysts, reactor and reaction. In order to overcome these difficulties, catalysts informatics is proposed as an alternative way to design and understand catalysts. The underlying concept of catalysts informatics is to design the catalysts from trends and patterns found in catalysts data. Here, three key concepts are introduced: experimental catalysts database, knowledge extraction from catalyst data via data science, and a catalysts informatics platform. Methane oxidation is chosen as a prototype reaction for demonstrating various aspects of catalysts informatics. This work summarizes how catalysts informatics plays a role in catalyst design. The work covers big data generation via high throughput experiments, machine learning, catalysts network method, catalyst design from small data, catalysts informatics platform, and the future of catalysts informatics via ontology. Thus, the proposed catalysts informatics would help innovate how catalysts can be designed and understood.

Biomacromolecule-Assisted Screening for Reaction Discovery and Catalyst Optimization

Reaction discovery and catalyst screening lie at the heart of synthetic organic chemistry. While there are efforts at de novo catalyst design using computation/artificial intelligence, at its core, synthetic chemistry is an experimental science. This review overviews biomacromolecule-assisted screening methods and the follow-on elaboration of chemistry so discovered. All three types of biomacromolecules discussed─enzymes, antibodies, and nucleic acids─have been used as "sensors" to provide a readout on product chirality exploiting their native chirality. Enzymatic sensing methods yield both UV-spectrophotometric and visible, colorimetric readouts. Antibody sensors provide direct fluorescent readout upon analyte binding in some cases or provide for cat-ELISA (Enzyme-Linked ImmunoSorbent Assay)-type readouts. DNA biomacromolecule-assisted screening allows for templation to facilitate reaction discovery, driving bimolecular reactions into a pseudo-unimolecular format. In addition, the ability to use DNA-encoded libraries permits the barcoding of reactants. All three types of biomacromolecule-based screens afford high sensitivity and selectivity. Among the chemical transformations discovered by enzymatic screening methods are the first Ni(0)-mediated asymmetric allylic amination and a new thiocyanopalladation/carbocyclization transformation in which both C-SCN and C-C bonds are fashioned sequentially. Cat-ELISA screening has identified new classes of sydnone-alkyne cycloadditions, and DNA-encoded screening has been exploited to uncover interesting oxidative Pd-mediated amido-alkyne/alkene coupling reactions.

Designing transformer oil immersion cooling servers for machine learning and first principle calculations

A transfomer oil immersion cooling server is designed and constructed for machine learning applications and first principle calculations that are carried out for materials-related research. CPU, motherboard, random access memory, hard disk drive, solid state drive, graphic card, and the power supply unit are submerged into the transformer oil in order to cool the entire system. Benchmark tests reveal that overall performance is improved while performance times for multicore calculations are dramatically improved. Furthermore, calculation times for machine learning with large data sets and density functional theory calculations are shortened during single core calculations. Thus, a transformer oil immersion cooling server is proposed to be an alternative cooling system used for improving the performance of first principle calculations and machine learning.

High-Throughput Screening and Literature Data Driven Machine Learning Assisting Investigation of Multi-component La2O3-based Catalysts for Oxidative Coupling of Methane

Multi-component La2O3-based catalysts for oxidative coupling of methane (OCM) were designed based on high-throughput screening (HTS) and literature datasets with multi-output machine learning (ML) approaches including random forest regression (RFR),...

Constructing catalyst knowledge networks from catalyst big data in oxidative coupling of methane for designing catalysts

Designing high performance catalysts for the oxidative coupling of methane (OCM) reaction is often hindered by inconsistent catalyst data, which often leads to difficulties in extracting information such as combinatorial effects of elements upon catalyst performance as well as difficulties in reaching yields beyond a particular threshold. In order to investigate C₂ yields more systematically, high throughput experiments are conducted in an effort to mass-produce catalyst-related data in a way that provides more consistency and structure. Graph theory is applied in order to visualize underlying trends in the transformation of high-throughput data into networks, which are then used to design new catalysts that potentially result in high C₂ yields during the OCM reaction. Transforming high-throughput data in this manner has resulted in a representation of catalyst data that is more intuitive to use and also has resulted in the successful design of a myriad of catalysts that elicit high C₂ yields, several of which resulted in yields greater than those originally reported in the high-throughput data. Thus, transforming high-throughput catalytic data into catalyst design-friendly maps provides a new method of catalyst design that is more efficient and has a higher likelihood of resulting in high performance catalysts.

Catalyst data created through high-throughput experimentation is transformed into catalyst knowledge networks, leading to a new method of catalyst design where successfully designed catalysts result in high C₂ yields during the OCM reaction.

Catalysis Gene Expression Profiling: Sequencing and Designing Catalysts

Identification of catalysts is a difficult matter as catalytic activities involve a vast number of complex features that each catalyst possesses. Here, catalysis gene expression profiling is proposed from unique features discovered in catalyst data collected by high-throughput experiments as an alternative way of representing the catalysts. Combining constructed catalyst gene sequences with hierarchical clustering results in catalyst gene expression profiling where natural language processing is used to identify similar catalysts based on edit distance. In addition, catalysts with similar properties are designed by modifying catalyst genes where the designed catalysts are experimentally confirmed to have catalytic activities that are associated with their catalyst gene sequences. Thus, the proposed method of catalyst gene expressions allows for a novel way of describing catalysts that allows for similarities in catalysts and catalytic activity to be easily recognized while enabling the ability to design new catalysts based on manipulating chemical elements of catalysts with similar catalyst gene sequences.

Functional and Material Properties in Nanocatalyst Design: A Data Handling and Sharing Problem

(1) Background: Properties and descriptors are two forms of molecular in silico representations. Properties can be further divided into functional, e.g., catalyst or drug activity, and material, e.g., X-ray crystal data. Millions of real measured functional property records are available for drugs or drug candidates in online databases. In contrast, there is not a single database that registers a real conversion, TON or TOF data for catalysts. All of the data are molecular descriptors or material properties, which are mainly of a calculation origin. (2) Results: Here, we explain the reason for this. We reviewed the data handling and sharing problems in the design and discovery of catalyst candidates particularly, material informatics and catalyst design, structural coding, data collection and validation, infrastructure for catalyst design and the online databases for catalyst design. (3) Conclusions: Material design requires a property prediction step. This can only be achieved based on the registered real property measurement. In reality, in catalyst design and discovery, we can observe either a severe functional property deficit or even property famine.