Active Site of Catalytic Ethene Epoxidation: Machine-Learning Global Pathway Sampling Rules Out the Metal Sites

Inverse design of chemoenzymatic epoxidation of soyabean oil through artificial intelligence-driven experimental approach

This paper presents an inverse design methodology that utilizes artificial intelligence (AI)-driven experiments to optimize the chemoenzymatic epoxidation of soyabean oil using hydrogen peroxide and lipase (Novozym 435). First, experiments are conducted using a systematic 3-level, 5-factor Box-Behnken design to explore the effect of input parameters on oxirane oxygen content (OOC (%)). Based on these experiments, various AI models are trained, with the support vector regression (SVR) model being found to be the most accurate. SVR is then used as a fitness function in particle swarm optimization, and the suggested optimal conditions, upon experimental validation, resulted in a maximum OOC of 7.19 % (∼98.5 % relative conversion of oil to epoxy). The results demonstrate the superiority of the proposed approach over existing methods. This framework offers a general intensified process optimization strategy with minimal resource utilization that can be applied to any other process.

In Situ Surfaced Mn-Mn Dimeric Sites Dictate CO Hydrogenation Activity and C2 Selectivity over MnRh Binary Catalysts

Massive ethanol production has long been a dream of human society. Despite extensive research in past decades, only a few systems have the potential of industrialization: specifically, Mn-promoted Rh (MnRh) binary heterogeneous catalysts were shown to achieve up to 60% C2 oxygenates selectivity in converting syngas (CO/H2) to ethanol. However, the active site of the binary system has remained poorly characterized. Here, large-scale machine-learning global optimization is utilized to identify the most stable Mn phases on Rh metal surfaces under reaction conditions by exploring millions of likely structures. We demonstrate that Mn prefers the subsurface sites of Rh metal surfaces and is able to emerge onto the surface forming MnRh surface alloy once the oxidative O/OH adsorbates are present. Our machine-learning-based transition state exploration further helps to resolve automatedly the whole reaction network, including 74 elementary reactions on various MnRh surface sites, and reveals that the Mn-Mn dimeric site at the monatomic step edge is the true active site for C2 oxygenate formation. The turnover frequency of the C2 product on the Mn-Mn dimeric site at MnRh steps is at least 107 higher than that on pure Rh steps from our microkinetic simulations, with the selectivity to the C2 product being 52% at 523 K. Our results demonstrate the key catalytic role of Mn-Mn dimeric sites in allowing C-O bond cleavage and facilitating the hydrogenation of O-terminating C2 intermediates, and rule out Rh metal by itself as the active site for CO hydrogenation to C2 oxygenates.

Carbon-metal versus metal-metal synergistic mechanism of ethylene electro-oxidation via electrolysis of water on TM2N6 sites in graphene

Acetaldehyde (AA) and ethylene oxide (EO) are important fine chemicals, and are also substrates with wide applications for high-value chemical products. Direct electrocatalytic oxidation of ethylene to AA and EO can avoid the untoward effects from harmful byproducts and high energy emissions. The most central intermediate state is the co-adsorption and coupling of ethylene and active oxygen intermediates (*O) at the active site(s), which is restricted by two factors: the stability of the *O intermediate generated during the electrolysis of water on the active site at a certain applied potential and pH range; and the lower kinetic energy barriers of the oxidation process based on the thermo-migration barrier from the *O intermediate to produce AA/EO. The benefit of two adjacent active atoms is more promising, since diverse adsorption and flexible catalytic sites may be provided for elementary reaction steps. Motivated by this strategy, we explored the feasibility of various homonuclear TM2N6@graphenes with dual-atomic-site catalysts (DASCs) for ethylene electro-oxidation through first-principles calculations via thermodynamic evaluation, analysis of the surface Pourbaix diagram, and kinetic evaluation. Two reaction mechanisms through C-TM versus TM-TM synergism were determined. Between them, a TM-TM mechanism on 4 TM2N6@graphenes and a C-TM mechanism on 5 TM2N6@graphenes are built. All 5 TM2N6@graphenes through the C-TM mechanism exhibit lower kinetic energy barriers for AA and EO generation than the 4 TM2N6@graphenes through the TM-TM mechanism. In particular, Pd2N6@graphene exhibits the most excellent catalytic activity, with energy barriers for generating AA and EO of only 0.02 and 0.65 eV at an applied potential of 1.77 V vs. RHE for the generation of an active oxygen intermediate. Electronic structure analysis indicates that the intrinsic C-TM mechanism is more advantageous than the TM-TM mechanism for ethylene electro-oxidation, and this study also provides valuable clues for further experimental exploration.

Efficient Machine Learning Model Focusing on Active Sites for the Discovery of Bifunctional Oxygen Electrocatalysts in Binary Alloys

The distinctive characteristics of alloy catalysts, encompassing composition, structure, and modifiable adsorption sites, present significant potential for the development of highly efficient electrocatalysts for oxygen evolution/reduction reactions [oxygen evolution reactions (OERs)/oxygen reduction reactions (ORRs)]. Machine learning (ML) methods can quickly establish the relationship between material features and catalytic activity, thus accelerating the development of alloy electrocatalysts. However, the current abundance of features presents a crucial challenge in selecting the most pertinent ones. In this study, we explored seven intrinsic features directly derived from the material's structure, with a specific focus on the chemical environment of active sites and their nearest neighbors. An accurate and efficient ML model to predict potential bifunctional oxygen electrocatalysts based on the intrinsic features of AB-type alloy active sites and intermediate free energies in the OERs/ORRs was established. These features possess clear physical and chemical meanings, closely linked to the electronic and geometric structures of active sites and neighboring atoms, thereby providing indispensable insights for the discovery of high-performance electrocatalysts. The ML model achieved R2 scores of 0.827, 0.913, and 0.711 for the predicted values of the three intermediate (OH, O, OOH) free energies, with corresponding mean absolute errors of 0.175, 0.242, and 0.200 eV, respectively. These results indicate that the ML model exhibits high accuracy in predicting the intermediate free energies. Furthermore, the ML model exhibited a prediction efficiency 150,000 times faster than traditional density functional theory calculations. This work will offer valuable insights and a framework for facilitating the rapid design of potential catalysts by ML methods.

Explainable machine-learning predictions for catalysts in CO2-assisted propane oxidative dehydrogenation

Propylene is an important raw material in the chemical industry that needs new routes for its production to meet the demand. The CO2-assisted oxidative dehydrogenation of propane (CO2-ODHP) represents an ideal way to produce propylene and uses the greenhouse gas CO2. The design of catalysts with high efficiency is crucial in CO2-ODHP research. Data-driven machine learning is currently of great interest and gaining popularity in the heterogeneous catalysis field for guiding catalyst development. In this study, the reaction results of CO2-ODHP reported in the literature are combined and analyzed with varied machine learning algorithms such as artificial neural network (ANN), k-nearest neighbors (KNN), support vector regression (SVR) and random forest regression (RF)and were used to predict the propylene space-time yield. Specifically, the RF method serves as a superior performing algorithm for propane conversion and propylene selectivity prediction, and SHapley Additive exPlanations (SHAP) based on the Shapley value performs fine model interpretation. Reaction conditions and chemical components show different impacts on catalytic performance. The work provides a valuable perspective for the machine learning in light alkane conversion, and helps us to design catalyst by catalytic performance hidden in the data of literatures.

Bonding and 13 C-NMR properties of coinage metal tris(ethylene) and tris(norbornene) complexes: Evaluation of the role of relativistic effects from DFT calculations

The π-complexes of cationic coinage metal ions (Cu(I), Ag(I), Au(I)) provide useful experimental support for understanding fundamental characteristics of bonding and ¹³ C-NMR patterns of the group 11 triad. Here, we account for the role of relativistic effects on olefin-coinage metal ion interaction for cationic, homoleptic tris-ethylene, and tris-norbornene complexes, [M(η² -C₂ H₄ )₃ ]⁺ and [M(η² -C₇ H₁₀ )₃ ]⁺ (M = Cu, Ag, Au), as representative case of studies. The M-(CC) bond strength in the cationic, tris-ethylene complexes is affected sizably for Au and to a lesser extent for Ag and Cu (48.6%, 16.7%, and 4.3%, respectively), owing to the influence on the different stabilizing terms accounting for the interaction energy in the formation of coinage metal cation-π complexes. The bonding elements provided by olefin → M σ-donation and olefin ← M π-backbonding are consequently affected, leading to a lesser covalent interaction going down in the triad if the relativistic effects are ignored. Analysis of the ¹³ C-NMR tensors provides further understanding of the observed experimental values, where the degree of backbonding charge donation to π₂ *-olefin orbital is the main influence on the observed high-field shifts in comparison to the free olefin. This donation is larger for ethylene complexes and lower for norbornene counterparts. However, the bonding energy in the later complexes is slightly stabilized given by the enhancement in the electrostatic character of the interaction. Thus, the theoretical evaluation of metal-alkene bonds, and other metal-bonding situations, benefits from the incorporation of relativistic effects even in lighter counterparts, which have an increasing role going down in the group.

Methanol Synthesis from CO2/CO Mixture on Cu-Zn Catalysts from Microkinetics-Guided Machine Learning Pathway Search

Methanol synthesis on industrial Cu/ZnO/Al₂O₃ catalysts via the hydrogenation of CO and CO₂ mixture, despite several decades of research, is still puzzling due to the nature of the active site and the role of CO₂ in the feed gas. Herein, with the large-scale machine learning atomic simulation, we develop a microkinetics-guided machine learning pathway search to explore thousands of reaction pathways for CO₂ and CO hydrogenations on thermodynamically favorable Cu-Zn surface structures, including Cu(111), Cu(211), and Zn-alloyed Cu(211) surfaces, from which the lowest energy pathways are identified. We find that Zn decorates at the step-edge at Cu(211) up to 0.22 ML under reaction conditions with the Zn-Zn dimeric sites being avoided. CO₂ and CO hydrogenations occur exclusively at the step-edge of the (211) surface with up to 0.11 ML Zn coverage, where the low coverage of Zn (0.11 ML) does not much affect the reaction kinetics, but the higher coverages of Zn (0.22 ML) poison the catalyst. It is CO₂ hydrogenation instead of CO hydrogenation that dominates methanol synthesis, agreeing with previous isotope experiments. While metallic steps are identified as the major active site, we show that the [-Zn-OH-Zn-] chains (cationic Zn) can grow on Cu(111) surfaces under reaction conditions, which suggests the critical role of CO in the mixed gas for reducing the cationic Zn and exposing metal sites for methanol synthesis. Our results provide a comprehensive picture on the dynamic coupling of the feed gas composition, the catalyst active site, and the reaction activity in this complex heterogeneous catalytic system.

Automated search for optimal surface phases (ASOPs) in grand canonical ensemble powered by machine learning

The surface of a material often undergoes dramatic structure evolution under a chemical environment, which, in turn, helps determine the different properties of the material. Here, we develop a general-purpose method for the automated search of optimal surface phases (ASOPs) in the grand canonical ensemble, which is facilitated by the stochastic surface walking (SSW) global optimization based on global neural network (G-NN) potential. The ASOP simulation starts by enumerating a series of composition grids, then utilizes SSW-NN to explore the configuration and composition spaces of surface phases, and relies on the Monte Carlo scheme to focus on energetically favorable compositions. The method is applied to silver surface oxide formation under the catalytic ethene epoxidation conditions. The known phases of surface oxides on Ag(111) are reproduced, and new phases on Ag(100) are revealed, which exhibit novel structure features that could be critical for understanding ethene epoxidation. Our results demonstrate that the ASOP method provides an automated and efficient way for probing complex surface structures that are beneficial for designing new functional materials under working conditions.