Bifunctional hydroformylation on heterogeneous Rh-WOx pair site catalysts.

Metal-catalysed reactions are often hypothesized to proceed on bifunctional active sites, whereby colocalized reactive species facilitate distinct elementary steps in a catalytic cycle1-8. Bifunctional active sites have been established on homogeneous binuclear organometallic catalysts9-11. Empirical evidence exists for bifunctional active sites on supported metal catalysts, for example, at metal-oxide support interfaces2,6,7,12. However, elucidating bifunctional reaction mechanisms on supported metal catalysts is challenging due to the distribution of potential active-site structures, their dynamic reconstruction and required non-mean-field kinetic descriptions7,12,13. We overcome these limitations by synthesizing supported, atomically dispersed rhodium-tungsten oxide (Rh-WOx) pair site catalysts. The relative simplicity of the pair site structure and sufficient description by mean-field modelling enable correlation of the experimental kinetics with first principles-based microkinetic simulations. The Rh-WOx pair sites catalyse ethylene hydroformylation through a bifunctional mechanism involving Rh-assisted WOx reduction, transfer of ethylene from WOx to Rh and H2 dissociation at the Rh-WOx interface. The pair sites exhibited >95% selectivity at a product formation rate of 0.1 gpropanal cm-3 h-1 in gas-phase ethylene hydroformylation. Our results demonstrate that oxide-supported pair sites can enable bifunctional reaction mechanisms with high activity and selectivity for reactions that are performed in industry using homogeneous catalysts.

Programmable heating and quenching for efficient thermochemical synthesis.

Conventional thermochemical syntheses by continuous heating under near-equilibrium conditions face critical challenges in improving the synthesis rate, selectivity, catalyst stability and energy efficiency, owing to the lack of temporal control over the reaction temperature and time, and thus the reaction pathways1-3. As an alternative, we present a non-equilibrium, continuous synthesis technique that uses pulsed heating and quenching (for example, 0.02 s on, 1.08 s off) using a programmable electric current to rapidly switch the reaction between high (for example, up to 2,400 K) and low temperatures. The rapid quenching ensures high selectivity and good catalyst stability, as well as lowers the average temperature to reduce the energy cost. Using CH4 pyrolysis as a model reaction, our programmable heating and quenching technique leads to high selectivity to value-added C2 products (>75% versus <35% by the conventional non-catalytic method and versus <60% by most conventional methods using optimized catalysts). Our technique can be extended to a range of thermochemical reactions, such as NH3 synthesis, for which we achieve a stable and high synthesis rate of about 6,000 µmol gFe-1 h-1 at ambient pressure for >100 h using a non-optimized catalyst. This study establishes a new model towards highly efficient non-equilibrium thermochemical synthesis.

Effect of Dynamic and Preferential Decoration of Pt Catalyst Surfaces by WOx on Hydrodeoxygenation Reactions.

Catalysts containing Pt nanoparticles and reducible transition-metal oxides (WOx, NbOx, TiOx) exhibit remarkable selectivity to aromatic products in hydrodeoxygenation (HDO) reactions for biomass valorization, contrasting the undesired aromatic hydrogenation typically observed for metal catalysts. However, the active site(s) responsible for the high selectivity remains elusive. Here, theoretical and experimental analyses are combined to explain the observed HDO reactivity by interrogating the organization of reduced WOx domains on Pt surfaces at sub-monolayer coverage. The SurfGraph algorithm is used to develop model structures that capture the configurational space (∼1000 configurations) for density functional theory (DFT) calculations of a W3O7 trimer on stepped Pt surfaces. Machine-learning models trained on the DFT calculations identify the preferential occupation of well-coordinated Pt sites (≥8 Pt coordination number) by WOx and structural features governing WOx-Pt stability. WOx/Pt/SiO2 catalysts are synthesized with varying W loadings to test the theoretical predictions and relate them to HDO reactivity. Spectroscopy- and microscopy-based catalyst characterizations identify the dynamic and preferential decoration of well-coordinated sites on Pt nanoparticles by reduced WOx species, consistent with theoretical predictions. The catalytic consequences of this preferential decoration on the HDO of a lignin model compound, dihydroeugenol, are clarified. The effect of WOx decoration on Pt nanoparticles for HDO involves WOx inhibition of aromatic ring hydrogenation by preferentially blocking well-coordinated Pt sites. The identification of preferential decoration on specific sites of late-transition-metal surfaces by reducible metal oxides provides a new perspective for understanding and controlling metal-support interactions in heterogeneous catalysis.

Applications of Artificial Intelligence and Machine Learning Algorithms to Crystallization.

Artificial intelligence and specifically machine learning applications are nowadays used in a variety of scientific applications and cutting-edge technologies, where they have a transformative impact. Such an assembly of statistical and linear algebra methods making use of large data sets is becoming more and more integrated into chemistry and crystallization research workflows. This review aims to present, for the first time, a holistic overview of machine learning and cheminformatics applications as a novel, powerful means to accelerate the discovery of new crystal structures, predict key properties of organic crystalline materials, simulate, understand, and control the dynamics of complex crystallization process systems, as well as contribute to high throughput automation of chemical process development involving crystalline materials. We critically review the advances in these new, rapidly emerging research areas, raising awareness in issues such as the bridging of machine learning models with first-principles mechanistic models, data set size, structure, and quality, as well as the selection of appropriate descriptors. At the same time, we propose future research at the interface of applied mathematics, chemistry, and crystallography. Overall, this review aims to increase the adoption of such methods and tools by chemists and scientists across industry and academia.

Tuning Active Site Flexibility by Defect Engineering of Graphene Ribbon Edge-hosted Fe-N3 Sites.

Nitrogen-doped, carbon-supported transition metal catalysts are excellent for several reactions. Structural engineering of M-Nx sites to boost catalytic activity is rarely studied. Here, we demonstrate that the structural flexibility of Fe-N3 site is vital for tuning the electronic structure of Fe atoms and regulating the catalytic transfer hydrogenation (CTH) activity. By introducing carbon defects, we construct Fe-N3 sites with varying Fe-N bond lengths distinguishable by X-ray absorption spectroscopy. We investigate the CTH activity by density-functional theory and microkinetic calculations and reveal that the vertical displacement of the Fe atom out of the plane of the support, induced by the Fe-N3 distortion, raises the Fe 3 d z 2 ${3{d}_{{z}^{2}}{\rm \ }}$ orbital and strengthens binding. We propose that the activity is controlled by the relaxation of the reconstructed site, which is further affected by Fe-N bond length, an excellent activity descriptor. We elucidate the origin of the CTH activity and principles for high-performing Fe-N-C catalysts by defect engineering.

CKineticsDB─An Extensible and FAIR Data Management Framework and Datahub for Multiscale Modeling in Heterogeneous Catalysis.

A great advantage of computational research is its reproducibility and reusability. However, an enormous amount of computational research data in heterogeneous catalysis is barricaded due to logistical limitations. Sufficient provenance and characterization of data and computational environment, with uniform organization and easy accessibility, can allow the development of software tools for integration across the multiscale modeling workflow. Here, we develop the Chemical Kinetics Database, CKineticsDB, a state-of-the-art datahub for multiscale modeling, designed to be compliant with the FAIR guiding principles for scientific data management. CKineticsDB utilizes a MongoDB back-end for extensibility and adaptation to varying data formats, with a referencing-based data model to reduce redundancy in storage. We have developed a Python software program for data processing operations and with built-in features to extract data for common applications. CKineticsDB evaluates the incoming data for quality and uniformity, retains curated information from simulations, enables accurate regeneration of publication results, optimizes storage, and allows the selective retrieval of files based on domain-relevant catalyst and simulation parameters. CKineticsDB provides data from multiple scales of theory (ab initio calculations, thermochemistry, and microkinetic models) to accelerate the development of new reaction pathways, kinetic analysis of reaction mechanisms, and catalysis discovery, along with several data-driven applications.

OpenMKM: An Open-Source C++ Multiscale Modeling Simulator for Homogeneous and Heterogeneous Catalytic Reactions.

Microkinetic modeling is invaluable for coupling "microscale" atomistic data with "macroscale" reactor observables. We introduce an Open-source Microkinetic Modeling (OpenMKM) multiscale mean-field microkinetics modeling toolkit targeting mainly heterogeneous catalytic reactions but applies equally to homogeneous reactions. OpenMKM is a modular, object-oriented, C++ software, built on top of the robust open-source Cantera built mainly for homogeneous reactions. Reaction mechanisms can be input from human-readable files or automatic reaction generators, avoiding tedious work and errors. The governing equations are also built automatically, unlike Matlab and Python manual implementations, providing speed and error-free models. OpenMKM has built-in interfaces with numerical software, SUNDIALS, for solving ordinary differential equations and differential-algebraic equations. Users can choose various ideal reactors and energy balance options, such as isothermal, adiabatic, temperature ramp, and an experimentally measured temperature profile. OpenMKM is tightly integrated with pMuTT for thermochemistry input file generation from density functional theory (DFT), streamlining the workflow from DFT to MKM and eliminating tedious work and human errors. It is also seamlessly integrated with the RenView software for visualizing the reaction pathways and performing the reaction path or flux analysis (RPA). OpenMKM includes local sensitivity analysis (LSA) by solving the augmented system of equations or using the one-at-a-time finite difference (first or second order) method. LSA can identify not only kinetically influential reactions but also species. The software provides two techniques for large reaction mechanisms for which LSA is too expensive to run. One is the Fischer Information Matrix, which is approximate but comes at nearly zero cost. The other is a new method that we term RPA-guided LSA, which is a finite difference-based method but uses RPA to select kinetically relevant reactions instead of exploiting the entire reaction network. Users can quickly set up and conduct microkinetic simulations without writing code. The user inputs are conveniently divided into reactor setup files and thermodynamic and kinetic definition files to set up different reactors. The source code and documentation are openly available at https://github.com/VlachosGroup/openmkm.

Transition-state correlations for predicting thermochemistry of adsorbates and surface reactions.

Estimating thermochemical properties from linear correlations may provide a pathway to circumvent expensive density functional theory (DFT) calculations for quantities such as pre-exponentials and temperature corrections to DFT energies. Here, we construct thermochemical scaling relations between C1-C6n-alkanes in the gas phase and adsorbed alkyl chains extending from several transition metal surfaces, and examine changes in the slope and fit between metals and adsorption sites. We subsequently add -OH, -NH2, CO, and CC functional groups to the C1-C6 molecules and demonstrate strong linear correlations for thermochemistry across all species. We broaden the correlations to incorporate transition states of C1-C6n-alkane dehydrogenation reactions, where thermochemistry for computationally prohibitive transition-state calculations can be quickly assessed. Additionally, we rationalize the linearity of thermochemical correlations based on the composition of the homologous series and theoretical assessments. As an application of the correlations, we estimate pre-exponentials for elementary surface reactions of ethane and propane hydrogenolysis on Ru(0001), which is of relevance to plastic hydrogenolysis. Depending on kinetically important steps, entropic contributions may be necessary to include in certain reaction mechanisms; in contrasting examples, entropies are found to be relatively insignificant for ethane hydrogenolysis but pertinent for propane hydrogenolysis.

Exploring Structure-Sensitive Relations for Small Species Adsorption Using Machine Learning.

Accurate prediction of adsorption energies on heterogeneous catalyst surfaces is crucial to predicting reactivity and screening materials. Adsorption linear scaling relations have been developed extensively but often lack accuracy and apply to one adsorbate and a single binding site type at a time. These facts undermine their ability to predict structure sensitivity and optimal catalyst structure. Using machine learning on nearly 300 density functional theory calculations, we demonstrate that generalized coordination number scaling relations hold well for oxygen- and high-valency carbon-binding species but fail for others. We reveal that the valency and the electronic coupling of a species with the surface, along with the site type and its coordination environment, are critical for small species adsorption. The model simultaneously predicts the adsorption energy and preferred site and significantly outperforms linear scalings in accuracy. It can expose the structure sensitivity of chemical reactions and enable enhanced catalyst activity via engineering particle shape and facet defects. The generality of our methodology is validated by training the model with transition metal data and transferring it to predict adsorption energies on single-atom alloys.

Unleashing the Power of Knowledge Extraction from Scientific Literature in Catalysis.

Valuable knowledge of catalysis is often hidden in a large amount of scientific literature. There is an urgent need to extract useful knowledge to facilitate scientific discovery. This work takes the first step toward the goal in the field of catalysis. Specifically, we construct the first information extraction benchmark data set that covers the field of catalysis and also develop a general extraction framework that can accurately extract catalysis-related entities from scientific literature with 90% extraction accuracy. We further demonstrate the feasibility of leveraging the extracted knowledge to help users better access relevant information in catalysis through an entity-aware search engine and a correlation analysis system.

The impact of differential lignin S/G ratios on mutagenicity and chicken embryonic toxicity.

Lignin and lignin-based materials have received considerable attention in various fields due to their promise as sustainable feedstocks. Guaiacol (G) and syringol (S) are two primary monolignols that occur in different ratios for different plant species. As methoxyphenols, G and S have been targeted as atmospheric pollutants and their acute toxicity examined. However, there is a rare understanding of the toxicological properties on other endpoints and mixture effects of these monolignols. To fill this knowledge gap, our study investigated the impact of different S/G ratios (0.5, 1, and 2) and three lignin depolymerization samples from poplar, pine, and miscanthus species on mutagenicity and developmental toxicity. A multitiered method consisted of in silico simulation, in vitro Ames test, and in vivo chicken embryonic assay was employed. In the Ames test, syringol showed a sign of mutagenicity, whereas guaiacol did not, which agreed with the T.E.S.T. simulation. For three S and G mixture and lignin monomers, mutagenic activity was related to the proportion of syringol. In addition, both S and G showed developmental toxicity in the chicken embryonic assay and T.E.S.T. simulation, and guaiacol had a severe effect on lipid peroxidation. A similar trend and comparable developmental toxicity levels were detected for S and G mixtures and the three lignin depolymerized monomers. This study provides data and insights on the differential toxicity of varying S/G ratios for some important building blocks for bio-based materials.

Learning Chemistry of Complex Reaction Systems via a Python First-Principles Reaction Rule Stencil (pReSt) Generator.

Complex reaction networks can be generated with automated network generators from initial reactants and reaction rules. Reaction rule specification is central to network generation. These reaction rules are, at present, user-defined based on (intuitive) expert knowledge of chemistry and are often transferred from gas-phase to surface processes. The catalyst active site geometry is usually left out but is often responsible for selectivity. We propose a first-principles-based reaction mechanism generation framework using density functional theory (DFT) data of published reaction mechanisms. The framework "learns the chemistry" from published mechanisms. It can generate reaction networks not studied before, "flag" reactions not seen before for further DFT convergence tests, and easily reconcile differences between catalysts and reactants that may introduce new pathways never seen before. As such, it can be a diagnostic tool for data (mechanism) quality assessment and novel pathway discovery to new molecules. A software, the Python Reaction Stencil (pReSt), was developed for this purpose to wrap around automatic mechanism generation software. Multiple catalytic chemistries are considered to show the efficacy of the proposed framework.

NEXTorch: A Design and Bayesian Optimization Toolkit for Chemical Sciences and Engineering.

Automation and optimization of chemical systems require well-informed decisions on what experiments to run to reduce time, materials, and/or computations. Data-driven active learning algorithms have emerged as valuable tools to solve such tasks. Bayesian optimization, a sequential global optimization approach, is a popular active-learning framework. Past studies have demonstrated its efficiency in solving chemistry and engineering problems. We introduce NEXTorch, a library in Python/PyTorch, to facilitate laboratory or computational design using Bayesian optimization. NEXTorch offers fast predictive modeling, flexible optimization loops, visualization capabilities, easy interfacing with legacy software, and multiple types of parameters and data type conversions. It provides GPU acceleration, parallelization, and state-of-the-art Bayesian optimization algorithms and supports both automated and human-in-the-loop optimization. The comprehensive online documentation introduces Bayesian optimization theory and several examples from catalyst synthesis, reaction condition optimization, parameter estimation, and reactor geometry optimization. NEXTorch is open-source and available on GitHub.

Thermochemical Data Fusion Using Graph Representation Learning.

Large databases are required for "Big Data" applications in catalysis and materials science. Thermochemical databases can be created by combining data from various sources and by correcting low-fidelity data sets to higher accuracy with minimal computation. To achieve this "data fusion", thermochemical quantities of interest, calculated at various levels of density functional theory (DFT), need to be mapped to the same, high levels of theory. In this work, a graph theoretical, statistical framework is proposed for such tasks. Subgraph frequencies are shown to provide a natural representation for learning these fusion maps. The maps are linear and are learnt with automated descriptor selection. Using a data set of as few as ∼1% from the QM9 database of 133 885 molecules, these models can predict multiple thermochemical quantities at a higher level of theory with an accuracy of 1 kcal/mol. The method is explainable, generalizable, and provides a diagnostic tool for outlier identification.

Phosphonate-Modified UiO-66 Brønsted Acid Catalyst and Its Use in Dehydra-Decyclization of 2-Methyltetrahydrofuran to Pentadienes.

Phosphorus-modified all-silica zeolites exhibit activity and selectivity in certain Brønsted acid catalyzed reactions for biomass conversion. In an effort to achieve similar performance with catalysts having well-defined sites, we report the incorporation of Brønsted acidity to metal-organic frameworks with the UiO-66 topology, achieved by attaching phosphonic acid to the 1,4-benzenedicarboxylate ligand and using it to form UiO-66-PO3 H2 by post-synthesis modification. Characterization reveals that UiO-66-PO3 H2 retains stability similar to UiO-66, and exhibits weak Brønsted acidity, as demonstrated by titrations, alcohol dehydration, and dehydra-decyclization of 2-methyltetrahydrofuran (2-MTHF). For the later reaction, the reported catalyst exhibits site-time yields and selectivity approaching that of phosphoric acid on all-silica zeolites. Using solid-state NMR and deprotonation energy calculations, the chemical environments of P and the corresponding acidities are determined.

Conjugation-Driven "Reverse Mars-van Krevelen"-Type Radical Mechanism for Low-Temperature C-O Bond Activation.

C-O bond activation on monofunctional catalysts (metals, carbides, and oxides) is challenging due to activity constraints imposed by energy scaling relationships. Yet, contrary to predictions, recently discovered multifunctional metal/metal oxide catalysts (e.g., Rh/ReOx, Rh/MoOx, Ir/VOx) demonstrate unusually high C-O scission activity at moderate temperatures. Herein, we use extensive density functional theory calculations, first-principles microkinetic modeling, and electronic structure analysis to elucidate the metal/metal oxide synergy in the Ru/RuO2 catalyst, which enables up to 76% yield of the C-O scission product (2-methyl furan) in catalytic transfer hydrogenolysis of furfural at low temperatures. Our key mechanistic finding is a facile radical-mediated C-O bond activation on RuO2 oxygen vacancies, which directly leads to a weakly bound final product. This is the first time the radical reduction mechanism is reported in heterogeneous catalysis at temperatures <200 °C. We attribute the unique catalytic properties to the formation of a conjugation-stabilized furfuryl radical upon C-O bond scission, the strong hydroxyl affinity of oxygen vacancies due to the metallic character of RuO2, and the acid-base heterogeneity of the oxide surface. The conjugation-driven radical-assisted C-O bond scission applies to any catalytic surface that preserves the π-electron system of the reactant and leads to C-O selectivity enhancement, with notable examples including Cu, H-covered Pd, self-assembled monolayers on Pd, and oxygen-covered Mo2C. Furthermore, we reveal the cooperativity of active sites in multifunctional catalysts. The mechanism is fully consistent with kinetic studies and isotopic labeling experiments, and the insights gained might prove useful more broadly in overcoming activity constraints induced by energy scaling relationships.

Adsorption in zeolites using mechanically embedded ONIOM clusters.

We have explored mechanically embedded three-layer QM/QM/MM ONIOM models for computational studies of binding in Al-substituted zeolites. In all the models considered, the high-level-theory layer consists of the adsorbate molecule and of the framework atoms within the first two coordination spheres of the Al atom and is treated at the M06-2X/6-311G(2df,p) level. For simplicity, flexibility and routine applicability, the outer, low-level-theory layer is treated with the UFF. We have modelled the intermediate-level layer quantum mechanically and investigated the performance of HF theory and of three DFT functionals, B3LYP, M06-2X and ωB97x-D, for different layer sizes and various basis sets, with and without BSSE corrections. We have studied the binding of sixteen probe molecules in H-MFI and compared the computed adsorption enthalpies with published experimental data. We have demonstrated that HF and B3LYP are inadequate for the description of the interactions between the probe molecules and the framework surrounding the metal site of the zeolite on account of their inability to capture dispersion forces. Both M06-2X and ωB97x-D on average converge within ca. 10% of the experimental values. We have further demonstrated transferability of the approach by computing the binding enthalpies of n-alkanes (C1-C8) in H-MFI, H-BEA and H-FAU, with very satisfactory agreement with experiment. The computed entropies of adsorption of n-alkanes in H-MFI are also found to be in good agreement with experimental data. Finally, we compare with published adsorption energies calculated by periodic-DFT for n-C3 to n-C6 alkanes, water and methanol in H-ZSM-5 and find very good agreement.

Stochastic averaging and sensitivity analysis for two scale reaction networks.

In the presence of multiscale dynamics in a reaction network, direct simulation methods become inefficient as they can only advance the system on the smallest scale. This work presents stochastic averaging techniques to accelerate computations for obtaining estimates of expected values and sensitivities with respect to the steady state distribution. A two-time-scale formulation is used to establish bounds on the bias induced by the averaging method. Further, this formulation provides a framework to create an accelerated "averaged" version of most single-scale sensitivity estimation methods. In particular, we propose the use of a centered ergodic likelihood ratio method for steady state estimation and show how one can adapt it to accelerated simulations of multiscale systems. Finally, we develop an adaptive "batch-means" stopping rule for determining when to terminate the micro-equilibration process.

Insights into the Cr(III) catalyzed isomerization mechanism of glucose to fructose in the presence of water using ab initio molecular dynamics.

The mechanism of glucose ring opening and isomerization to fructose, catalyzed by the Lewis acid catalyst CrCl3 in the presence of water, is investigated using Car-Parrinello molecular dynamics with metadynamics. Minimum energy pathways for the reactions are revealed and the corresponding free energy barriers are computed. Addition of glucose replaces two water molecules in the active [Cr(H2O)5OH](+2) complex, with two hydroxyl groups of glucose taking their place. Ring opening and isomerization reactions can only proceed if the first step involving the deprotonation of glucose is accompanied by the protonation of the OH(-) group in the partially hydrolyzed metal center ([Cr(C6H12O6)(H2O)3OH](+2) → [Cr(C6H11O6)(H2O)4](+2)). This provides further evidence that the partially hydrolyzed [Cr(H2O)5OH](+2) is the active species catalyzing ring opening and isomerization reactions and that unhydrolyzed Cr(+3) may not be able to catalyze the reactions. After the ring opening, the isomerization reaction proceeds via deprotonation, followed by hydride shift and the back donation of the proton from the metal complex to the sugar. Water molecules outside the first coordination sphere of the metal complex participate in the reaction for mediating the proton transfer. The hydride shift in the isomerization is the overall rate limiting step with a free energy barrier of 104 kJ mol(-1). The simulation computed barrier is in agreement with experiments.

Solvent-induced frequency shifts of 5-hydroxymethylfurfural deduced via infrared spectroscopy and ab initio calculations.

Solvent-induced frequency shifts (SIFS) of the carbonyl stretching vibration ν(C═O) of 5-hydroxymethylfurfural were measured in protic, polar aprotic, and nonpolar solvents. The Gutmann acceptor number (AN) was found to correlate with the measured frequency shifts. The SIFS in six solvents were investigated using ab initio electronic structure calculations, treating the solvent implicitly and with an explicit solvent ligand interacting with the carbonyl. The conductor-polarizable continuum model (CPCM) of solvation predicted that ν(C═O) shifted according with the dielectric constant as (ε - 1)/(2ε + 1), in agreement with the analytical predictions of the Kirkwood-Bauer-Magat (KBM) theory for a dipole in a dielectric continuum, but in disagreement with the experimental trend. The experimental SIFS were best predicted using gas-phase complexes of HMF and explicit solvent-ligand. Natural bond orbital (NBO) analysis and Bader's atoms in molecules theory were used to investigate the electronic structure of these complexes. Strong SIFS were found to arise from stronger H-bonding interactions, as observed in delocalization of carbonyl lone-pair electrons by H-bonding solvent σ*(X-H) orbitals, and an increase in charge density and a decrease in local potential energy at the H-bond (3, -1) critical point. Consequently, by predicting the experimental SIFS and examining the electronic structure, we find the first theoretical evidence for treating Gutmann's solvent AN as a measure of solvent Lewis acidity.