Understanding and Breaking Scaling Relations in Single-Site Catalysis: Methane to Methanol Conversion by FeIV═O

Long-range electrostatic effects from intramolecular Lewis acid binding influence the redox properties of cobalt-porphyrin complexes

A CoII-porphyrin complex (1) with an appended aza-crown ether for Lewis acid (LA) binding was synthesized and characterized. NMR spectroscopy and electrochemistry show that cationic group I and II LAs (i.e., Li+, Na+, K+, Ca2+, Sr2+, and Ba2+) bind to the aza-crown ether group of 1. The binding constant for Li+ is comparable to that observed for a free aza-crown ether. LA binding causes an anodic shift in the CoII/CoI couple of between 10 and 40 mV and also impacts the CoIII/CoII couple. The magnitude of the anodic shift of the CoII/CoI couple varies linearly with the strength of the LA as determined by the pKa of the corresponding metal-aqua complex, with dications giving larger shifts than monocations. The extent of the anodic shift of the CoII/CoI couple also increases as the ionic strength of the solution decreases. This is consistent with electric field effects being responsible for the changes in the redox properties of 1 upon LA binding and provides a novel method to tune the reduction potential. Density functional theory calculations indicate that the bound LA is 5.6 to 6.8 Å away from the CoII ion, demonstrating that long-range electrostatic effects, which do not involve changes to the primary coordination sphere, are responsible for the variations in redox chemistry. Compound 1 was investigated as a CO2 reduction electrocatalyst and shows high activity but rapid decomposition.

An Active Site Tyr Residue Guides the Regioselectivity of Lysine Hydroxylation by Nonheme Iron Lysine-4-hydroxylase Enzymes through Proton-Coupled Electron Transfer

Lysine dioxygenase (KDO) is an important enzyme in human physiology involved in bioprocesses that trigger collagen cross-linking and blood pressure control. There are several KDOs in nature; however, little is known about the factors that govern the regio- and stereoselectivity of these enzymes. To understand how KDOs can selectively hydroxylate their substrate, we did a comprehensive computational study into the mechanisms and features of 4-lysine dioxygenase. In particular, we selected a snapshot from the MD simulation on KDO5 and created large QM cluster models (A, B, and C) containing 297, 312, and 407 atoms, respectively. The largest model predicts regioselectivity that matches experimental observation with rate-determining hydrogen atom abstraction from the C4-H position, followed by fast OH rebound to form 4-hydroxylysine products. The calculations show that in model C, the dipole moment is positioned along the C4-H bond of the substrate and, therefore, the electrostatic and electric field perturbations of the protein assist the enzyme in creating C4-H hydroxylation selectivity. Furthermore, an active site Tyr233 residue is identified that reacts through proton-coupled electron transfer akin to the axial Trp residue in cytochrome c peroxidase. Thus, upon formation of the iron(IV)-oxo species in the catalytic cycle, the Tyr233 phenol loses a proton to the nearby Asp179 residue, while at the same time, an electron is transferred to the iron to create an iron(III)-oxo active species. This charged tyrosyl residue directs the dipole moment along the C4-H bond of the substrate and guides the selectivity to the C4-hydroxylation of the substrate.

A density functional theory analysis of the C-H activation reactivity of iron(IV)-oxo complexes with an 'O' substituted tetramethylcyclam macrocycle

In this article, we present a meticulous computational study to foresee the effect of an oxygen-rich macrocycle on the reactivity for C-H activation. For this study, a widely studied nonheme Fe(IV)O molecule with a TMC (1,4,8,11-tetramethyl 1,4,8,11-tetraazacyclotetradecane) macrocycle that is equatorially attached to four nitrogen atoms (designated as N4) and acetonitrile as an axial ligand has been taken into account. For the goal of hetero-substitution, step-by-step replacement of the N4 framework with O atoms, i.e., N4, N3O1, N2O2, N1O3, and O4 systems, has been considered, and dihydroanthracene (DHA) has been used as the substrate. In order to neutralise the system and prevent the self-interaction error in DFT, triflate counterions have also been included in the calculations. The study of the energetics of these C-H bond activation reactions and the potential energy surfaces mapped therefore reveal that the initial hydrogen abstraction, which is the rate-determining step, follows the two-state reactivity (TSR) pattern, which means that the originally excited quintet state falls lower in the transition state and the product. The reaction follows the hydrogen atom transfer (HAT) mechanism, as indicated by the spin density studies. The results revealed a fascinating reactivity order, in which the reactivity increases with the enrichment of the oxygen atom in the equatorial position, namely the order follows N4 < N3O1 < N2O2 < N1O3 < O4. The impacts of oxygen substitution on quantum mechanical tunneling and the H/D kinetic isotope effect have also been investigated. When analysing the causes of this reactivity pattern, a number of variables have been identified, including the reactant-like transition structure, spin density distribution, distortion energy, and energies of the electron acceptor orbital, i.e., the energy of the LUMO (σ*z2), which validate the obtained outcome. Our results also show very good agreement with earlier combined experimental and theoretical studies considering TMC and TMCO-type complexes. The DFT predictions reported here will undoubtedly encourage experimental research in this biomimetic field, as they provide an alternative with higher reactivity in which heteroatoms can be substituted for the traditional nitrogen atom.

Interpreting chemisorption strength with AutoML-based feature deletion experiments

The chemisorption energy of reactants on a catalyst surface, [Formula: see text], is among the most informative characteristics of understanding and pinpointing the optimal catalyst. The intrinsic complexity of catalyst surfaces and chemisorption reactions presents significant difficulties in identifying the pivotal physical quantities determining [Formula: see text]. In response to this, the study proposes a methodology, the feature deletion experiment, based on Automatic Machine Learning (AutoML) for knowledge extraction from a high-throughput density functional theory (DFT) database. The study reveals that, for binary alloy surfaces, the local adsorption site geometric information is the primary physical quantity determining [Formula: see text], compared to the electronic and physiochemical properties of the catalyst alloys. By integrating the feature deletion experiment with instance-wise variable selection (INVASE), a neural network-based explainable AI (XAI) tool, we established the best-performing feature set containing 21 intrinsic, non-DFT computed properties, achieving an MAE of 0.23 eV across a periodic table-wide chemical space involving more than 1,600 types of alloys surfaces and 8,400 chemisorption reactions. This study demonstrates the stability, consistency, and potential of AutoML-based feature deletion experiment in developing concise, predictive, and theoretically meaningful models for complex chemical problems with minimal human intervention.

Automated de Novo Design of Olefin Metathesis Catalysts: Computational and Experimental Analysis of a Simple Thermodynamic Design Criterion

Methods for computational de novo design of inorganic molecules have paved the way for automated design of homogeneous catalysts. Such studies have so far relied on correlation-based prediction models as fitness functions (figures of merit), but the soundness of these approaches has yet to be tested by experimental verification of de novo-designed catalysts. Here, a previously developed criterion for the optimization of dative ligands L in ruthenium-based olefin metathesis catalysts RuCl2(L)(L')(═CHAr), where Ar is an aryl group and L' is a phosphine ligand dissociating to activate the catalyst, was used in de novo design experiments. These experiments predicted catalysts bearing an N-heterocyclic carbene (L = 9) substituted by two N-bound mesityls and two tert-butyl groups at the imidazolidin-2-ylidene backbone to be promising. Whereas the phosphine-stabilized precursor assumed by the prediction model could not be made, a pyridine-stabilized ruthenium alkylidene complex (17) bearing carbene 9 was less active than a known leading pyridine-stabilized Grubbs-type catalyst (18, L = H2IMes). A density functional theory-based analysis showed that the unsubstituted metallacyclobutane (MCB) intermediate generated in the presence of ethylene is the likely resting state of both 17 and 18. Whereas the design criterion via its correlation between the stability of the MCB and the rate-determining barrier indeed seeks to stabilize the MCB, it relies on RuCl2(L)(L')(═CH2) adducts as resting states. The change in resting state explains the discrepancy between the prediction and the actual performance of catalyst 17. To avoid such discrepancies and better address the multifaceted challenges of predicting catalytic performance, future de novo catalyst design studies should explore and test design criteria incorporating information from more than a single relative energy or intermediate.

Thermodynamic principle to enhance enzymatic activity using the substrate affinity

Understanding how to tune enzymatic activity is important not only for biotechnological applications, but also to elucidate the basic principles guiding the design and optimization of biological systems in nature. So far, the Michaelis-Menten equation has provided a fundamental framework of enzymatic activity. However, there is still no concrete guideline on how the parameters should be optimized towards higher activity. Here, we demonstrate that tuning the Michaelis-Menten constant ([Formula: see text]) to the substrate concentration ([Formula: see text]) enhances enzymatic activity. This guideline ([Formula: see text]) was obtained mathematically by assuming that thermodynamically favorable reactions have higher rate constants, and that the total driving force is fixed. Due to the generality of these thermodynamic considerations, we propose [Formula: see text] as a general concept to enhance enzymatic activity. Our bioinformatic analysis reveals that the [Formula: see text] and in vivo substrate concentrations are consistent across a dataset of approximately 1000 enzymes, suggesting that even natural selection follows the principle [Formula: see text].

Iterative multiscale and multi-physics computations for operando catalyst nanostructure elucidation and kinetic modeling

Modern heterogeneous catalysis has benefitted immensely from computational predictions of catalyst structure and its evolution under reaction conditions, first-principles mechanistic investigations, and detailed kinetic modeling, which are rungs on a multiscale workflow. Establishing connections across these rungs and integration with experiments have been challenging. Here, operando catalyst structure prediction techniques using density functional theory simulations and ab initio thermodynamics calculations, molecular dynamics, and machine learning techniques are presented. Surface structure characterization by computational spectroscopic and machine learning techniques is then discussed. Hierarchical approaches in kinetic parameter estimation involving semi-empirical, data-driven, and first-principles calculations and detailed kinetic modeling via mean-field microkinetic modeling and kinetic Monte Carlo simulations are discussed along with methods and the need for uncertainty quantification. With these as the background, this article proposes a bottom-up hierarchical and closed loop modeling framework incorporating consistency checks and iterative refinements at each level and across levels.

Computational Discovery of Stable Metal-Organic Frameworks for Methane-to-Methanol Catalysis

The challenge of direct partial oxidation of methane to methanol has motivated the targeted search of metal-organic frameworks (MOFs) as a promising class of materials for this transformation because of their site-isolated metals with tunable ligand environments. Thousands of MOFs have been synthesized, yet relatively few have been screened for their promise in methane conversion. We developed a high-throughput virtual screening workflow that identifies MOFs from a diverse space of experimental MOFs that have not been studied for catalysis, yet are thermally stable, synthesizable, and have promising unsaturated metal sites for C-H activation via a terminal metal-oxo species. We carried out density functional theory calculations of the radical rebound mechanism for methane-to-methanol conversion on models of the secondary building units (SBUs) from 87 selected MOFs. While we showed that oxo formation favorability decreases with increasing 3d filling, consistent with prior work, previously observed scaling relations between oxo formation and hydrogen atom transfer (HAT) are disrupted by the greater diversity in our MOF set. Accordingly, we focused on Mn MOFs, which favor oxo intermediates without disfavoring HAT or leading to high methanol release energies─a key feature for methane hydroxylation activity. We identified three Mn MOFs comprising unsaturated Mn centers bound to weak-field carboxylate ligands in planar or bent geometries with promising methane-to-methanol kinetics and thermodynamics. The energetic spans of these MOFs are indicative of promising turnover frequencies for methane to methanol that warrant further experimental catalytic studies.

Transition metal oxide complexes as molecular catalysts for selective methane to methanol transformation: any prospects or time to retire?

Transition metal oxides have been extensively used in the literature for the conversion of methane to methanol. Despite the progress made over the past decades, no method with satisfactory performance or economic viability has been detected. The main bottleneck is that the produced methanol oxidizes further due to its weaker C-H bond than that of methane. Every improvement in the efficiency of a catalyst to activate methane leads to reduction of the selectivity towards methanol. Is it therefore prudent to keep studying (both theoretically and experimentally) metal oxides as catalysts for the quantitative conversion of methane to methanol? This perspective focuses on molecular metal oxide complexes and suggests strategies to bypass the current bottlenecks with higher weight on the computational chemistry side. We first discuss the electronic structure of metal oxides, followed by assessing the role of the ligands in the reactivity of the catalysts. For better selectivity, we propose that metal oxide anionic complexes should be explored further, while hydrophylic cavities in the vicinity of the metal oxide can perturb the transition-state structure for methanol increasing appreciably the activation barrier for methanol. We also emphasize that computational studies should target the activation reaction of methanol (and not only methane), the study of complete catalytic cycles (including the recombination and oxidation steps), and the use of molecular oxygen as an oxidant. The titled chemical conversion is an excellent challenge for theory and we believe that computational studies should lead the field in the future. It is finally shown that bottom-up approaches offer a systematic way for exploration of the chemical space and should still be applied in parallel with the recently popular machine learning techniques. To answer the question of the title, we believe that metal oxides should still be considered provided that we change our focus and perform more systematic investigations on the activation of methanol.

New Strategies for Direct Methane-to-Methanol Conversion from Active Learning Exploration of 16 Million Catalysts

Despite decades of effort, no earth-abundant homogeneous catalysts have been discovered that can selectively oxidize methane to methanol. We exploit active learning to simultaneously optimize methane activation and methanol release calculated with machine learning-accelerated density functional theory in a space of 16 M candidate catalysts including novel macrocycles. By constructing macrocycles from fragments inspired by synthesized compounds, we ensure synthetic realism in our computational search. Our large-scale search reveals that low-spin Fe(II) compounds paired with strong-field (e.g., P or S-coordinating) ligands have among the best energetic tradeoffs between hydrogen atom transfer (HAT) and methanol release. This observation contrasts with prior efforts that have focused on high-spin Fe(II) with weak-field ligands. By decoupling equatorial and axial ligand effects, we determine that negatively charged axial ligands are critical for more rapid release of methanol and that higher-valency metals [i.e., M(III) vs M(II)] are likely to be rate-limited by slow methanol release. With full characterization of barrier heights, we confirm that optimizing for HAT does not lead to large oxo formation barriers. Energetic span analysis reveals designs for an intermediate-spin Mn(II) catalyst and a low-spin Fe(II) catalyst that are predicted to have good turnover frequencies. Our active learning approach to optimize two distinct reaction energies with efficient global optimization is expected to be beneficial for the search of large catalyst spaces where no prior designs have been identified and where linear scaling relationships between reaction energies or barriers may be limited or unknown.

Breaking adsorption-energy scaling limitations of electrocatalytic nitrate reduction on intermetallic CuPd nanocubes by machine-learned insights

The electrochemical nitrate reduction reaction (NO₃RR) to ammonia is an essential step toward restoring the globally disrupted nitrogen cycle. In search of highly efficient electrocatalysts, tailoring catalytic sites with ligand and strain effects in random alloys is a common approach but remains limited due to the ubiquitous energy-scaling relations. With interpretable machine learning, we unravel a mechanism of breaking adsorption-energy scaling relations through the site-specific Pauli repulsion interactions of the metal d-states with adsorbate frontier orbitals. The non-scaling behavior can be realized on (100)-type sites of ordered B2 intermetallics, in which the orbital overlap between the hollow *N and subsurface metal atoms is significant while the bridge-bidentate *NO₃ is not directly affected. Among those intermetallics predicted, we synthesize monodisperse ordered B2 CuPd nanocubes that demonstrate high performance for NO₃RR to ammonia with a Faradaic efficiency of 92.5% at -0.5 V_RHE and a yield rate of 6.25 mol h^-1 g^-1 at -0.6 V_RHE. This study provides machine-learned design rules besides the d-band center metrics, paving the path toward data-driven discovery of catalytic materials beyond linear scaling limitations.

Computational Scaling Relationships Predict Experimental Activity and Rate-Limiting Behavior in Homogeneous Water Oxidation

While computational screening with first-principles density functional theory (DFT) is essential for evaluating candidate catalysts, limitations in accuracy typically prevent the prediction of experimentally relevant activities. Exemplary of these challenges are homogeneous water oxidation catalysts (WOCs) where differences in experimental conditions or small changes in ligand structure can alter rate constants by over an order of magnitude. Here, we compute mechanistically relevant electronic and energetic properties for 19 mononuclear Ru transition-metal complexes (TMCs) from three experimental water oxidation catalysis studies. We discover that 15 of these TMCs have experimental activities that correlate with a single property, the ionization potential of the Ru(II)-O₂ catalytic intermediate. This scaling parameter allows the quantitative understanding of activity trends and provides insight into the rate-limiting behavior. We use this approach to rationalize differences in activity with different experimental conditions, and we qualitatively analyze the source of distinct behavior for different electronic states in the other four catalysts. Comparison to closely related single-atom catalysts and modified WOCs enables rationalization of the source of rate enhancement in these WOCs.

DABCO-promoted photocatalytic C-H functionalization of aldehydes

Herein we present a direct application of DABCO, an inexpensive and broadly accessible organic base, as a hydrogen atom transfer (HAT) abstractor in a photocatalytic strategy for aldehyde C–H activation. The acyl radicals generated in this step were arylated with aryl bromides through a well stablished nickel cross-coupling methodology, leading to a variety of interesting aryl ketones in good yields. We also performed computational calculations to shine light in the HAT step energetics and determined an optimized geometry for the transition state, showing that the hydrogen atom transfer between aldehydes and DABCO is a mildly endergonic, yet sufficiently fast step. The same calculations were performed with quinuclidine, for comparison of both catalysts and the differences are discussed.

Light alkane oxidation over well-defined active sites in metal–organic framework materials

We review structure–catalytic property relationships for MOF materials used in the direct oxidation of light alkanes, focusing specifically on the elucidation of active site structures and probes for reaction mechanisms.

Electrostatic perturbations in the substrate-binding pocket of taurine/α-ketoglutarate dioxygenase determine its selectivity

Taurine/α‐ketoglutarate dioxygenase is an important enzyme that takes part in the cysteine catabolism process in the human body and selectively hydroxylates taurine at the C¹‐position. Recent computational studies showed that in the gas‐phase the C²−H bond of taurine is substantially weaker than the C¹−H bond, yet no evidence exists of 2‐hydroxytaurine products. To this end, a detailed computational study on the selectivity patterns in TauD was performed. The calculations show that the second‐coordination sphere and the protonation states of residues play a major role in guiding the enzyme to the right selectivity. Specifically, a single proton on an active site histidine residue can change the regioselectivity of the reaction through its electrostatic perturbations in the active site and effectively changes the C¹−H and C²−H bond strengths of taurine. This is further emphasized by many polar and hydrogen bonding interactions of the protein cage in TauD with the substrate and the oxidant that weaken the pro‐R C¹−H bond and triggers a chemoselective reaction process. The large cluster models reproduce the experimental free energy of activation excellently.

Biodegradation of herbicides by a plant nonheme iron dioxygenase: mechanism and selectivity of substrate analogues

Aryloxyalkanoate dioxygenases are unique herbicide biodegrading nonheme iron enzymes found in plants and hence, from environmental and agricultural point of view they are important and valuable. However, they often are substrate specific and little is known on the details of the mechanism and the substrate scope. To this end, we created enzyme models and calculate the mechanism for 2,4-dichlorophenoxyacetic acid biodegradation and 2-methyl substituted analogs by density functional theory. The work shows that the substrate binding is tight and positions the aliphatic group close to the metal center to enable a chemoselective reaction mechanism to form the C 2 -hydroxy products, whereas the aromatic hydroxylation barriers are well higher in energy. Subsequently, we investigated the metabolism of R - and S -methyl substituted inhibitors and show that these do not react as efficiently as 2,4-dichlorophenoxyacetic acid substrate due to stereochemical clashes in the active site and particularly for the R -isomer give high rebound barriers.

Single-Atom (Iron-Based) Catalysts: Synthesis and Applications

Supported single-metal atom catalysts (SACs) are constituted of isolated active metal centers, which are heterogenized on inert supports such as graphene, porous carbon, and metal oxides. Their thermal stability, electronic properties, and catalytic activities can be controlled via interactions between the single-metal atom center and neighboring heteroatoms such as nitrogen, oxygen, and sulfur. Due to the atomic dispersion of the active catalytic centers, the amount of metal required for catalysis can be decreased, thus offering new possibilities to control the selectivity of a given transformation as well as to improve catalyst turnover frequencies and turnover numbers. This review aims to comprehensively summarize the synthesis of Fe-SACs with a focus on anchoring single atoms (SA) on carbon/graphene supports. The characterization of these advanced materials using various spectroscopic techniques and their applications in diverse research areas are described. When applicable, mechanistic investigations conducted to understand the specific behavior of Fe-SACs-based catalysts are highlighted, including the use of theoretical models.

Electronic Structure of RhO2+, Its Ammoniated Complexes (NH3)1-5RhO2+, and Mechanistic Exploration of CH4 Activation by Them

High-level electronic structure calculations are initially performed to investigate the electronic structure of RhO²⁺. The construction of potential energy curves for the ground and low-lying excited states allowed the calculation of spectroscopic constants, including harmonic and anharmonic vibrational frequencies, bond lengths, spin-orbit constants, and excitation energies. The equilibrium electronic configurations were used for the interpretation of the chemical bonding. We further monitored how the Rh-O bonding scheme changes with the gradual addition of ammonia ligands. The nature of this bond remains unaffected up to four ammonia ligands but adopts a different electronic configuration in the pseudo-octahedral geometry of (NH₃)₅RhO²⁺. This has consequences in the activation mechanism of the C-H bond of methane by these complexes, especially (NH₃)₄RhO²⁺. We show that the [2 + 2] mechanism in the (NH₃)₄RhO²⁺ case has a very low energy barrier comparable to that of a radical mechanism. We also demonstrate that methane can coordinate to the metal in a similar fashion to ammonia and that knowledge of the electronic structure of the pure ammonia complexes provides qualitative insights into the optimal reaction mechanism.

Identifying the Critical Surface Descriptors for the Negative Slopes in the Adsorption Energy Scaling Relationships via Density Functional Theory and Compressed Sensing

Adsorption energy scaling relationships have progressed beyond their original form, which was primarily focused on optimizing catalytic sites and lowering computational costs in simulations. The recent rise in interest in adsorption energy scaling relations is to investigate surfaces other than transition metals (TMs) as well as interactions involving complex compounds. In this work, we report our extensive study on the scaling relation (SR) between oxygen (O), with elements of neighboring groups such as boron (B), aluminum (Al), carbon (C), silicon (Si), nitrogen (N), phosphorus (P), and fluorine (F) on magnetic bimetallic surfaces. We observed that only O versus N and F seems to have a positive slope; the other slopes are negative. We present new theoretical model in terms of multiple surface descriptors using density functional theory and compressed sensing, whereas the original scaling theory was based on a single adsorbate descriptor: adsorbate valency.

Recent advances in heterogeneous catalysis for the nonoxidative conversion of methane

The direct conversion of methane to high-value chemicals is an attractive process that efficiently uses abundant natural/shale gas to provide an energy supply. The direct conversion of methane to high-value chemicals is an attractive process that efficiently uses abundant natural/shale gas to provide an energy supply. Among all the routes used for methane transformation, nonoxidative conversion of methane is noteworthy owing to its highly economic selectivity to bulk chemicals such as aromatics and olefins. Innovations in catalysts for selective C-H activation and controllable C-C coupling thus play a key role in this process and have been intensively investigated in recent years. In this review, we briefly summarize the recent advances in conventional metal/zeolite catalysts in the nonoxidative coupling of methane to aromatics, as well as the newly emerging single-atom based catalysts for the conversion of methane to olefins. The emphasis is primarily the experimental findings and the theoretical understanding of the active sites and reaction mechanisms. We also present our perspectives on the design of catalysts for C-H activation and C-C coupling of methane, to shed some light on improving the potential industrial applications of the nonoxidative conversion of methane into chemicals.

Electronic structure of the dicationic first row transition metal oxides

Multi-reference electronic structure calculations combined with large basis sets are performed to investigate the electronic structure of the ground and low-lying electronic states of the MO²⁺ diatomic species with M = Ti-Cu. These systems have shown high efficiency in the activation of the C-H of saturated hydrocarbons. This study is the first systematic and accurate work for these systems and our results and discussion provides insights into the reactivity and stability of MO²⁺ units. We find that they can be divided in three groups. The early transition metals (Ti, V, Cr) have very stable and well separated oxo (M⁴⁺O^2-) character ground states, the middle transition metals (Mn, Fe) have oxyl (M³⁺O˙^-) ground states with low-lying oxo excited states, and the late transition metals (Co, Ni, Cu) have well separated oxyl states. The reported spectroscopic constants will aid future experimental investigations, which are sparse in the literature. Periodic trends for the bond lengths, energetics, excitation energies, and wavefunction composition are discussed in detail. Complete basis set limit results indicate the high accuracy of the quintuple-ζ basis sets.

Computational Discovery of Transition-metal Complexes: From High-throughput Screening to Machine Learning

Transition-metal complexes are attractive targets for the design of catalysts and functional materials. The behavior of the metal-organic bond, while very tunable for achieving target properties, is challenging to predict and necessitates searching a wide and complex space to identify needles in haystacks for target applications. This review will focus on the techniques that make high-throughput search of transition-metal chemical space feasible for the discovery of complexes with desirable properties. The review will cover the development, promise, and limitations of "traditional" computational chemistry (i.e., force field, semiempirical, and density functional theory methods) as it pertains to data generation for inorganic molecular discovery. The review will also discuss the opportunities and limitations in leveraging experimental data sources. We will focus on how advances in statistical modeling, artificial intelligence, multiobjective optimization, and automation accelerate discovery of lead compounds and design rules. The overall objective of this review is to showcase how bringing together advances from diverse areas of computational chemistry and computer science have enabled the rapid uncovering of structure-property relationships in transition-metal chemistry. We aim to highlight how unique considerations in motifs of metal-organic bonding (e.g., variable spin and oxidation state, and bonding strength/nature) set them and their discovery apart from more commonly considered organic molecules. We will also highlight how uncertainty and relative data scarcity in transition-metal chemistry motivate specific developments in machine learning representations, model training, and in computational chemistry. Finally, we will conclude with an outlook of areas of opportunity for the accelerated discovery of transition-metal complexes.

A framework for constructing linear free energy relationships to design molecular transition metal catalysts

A computational framework for ligand-driven design of transition metal complexes is presented in this work. We propose a general procedure for the construction of active site-specific linear free energy relationships (LFERs), which are inspired from Hammett and Taft correlations in organic chemistry and grounded in the activation strain model (ASM). Ligand effects are isolated and quantified in terms of their contribution to interaction and strain energy components of ASM. Scalar descriptors that are easily obtainable are then employed to construct the complete LFER. We successfully demonstrate proof-of-concept by constructing and applying an LFER to CH activation with enzyme-inspired [Cu2O2]2+ complexes. The key benefit of using ASM is a built-in compensation or error cancellation between LFER prediction of interaction and strain terms, resulting in accurate barrier predictions for 37 of the 47 catalysts examined in this study. The LFER is also transferable with respect to level of theory and flexible towards the choice of reference system. The absence of interaction-strain compensation or poor model performance for the remaining systems is a consequence of the approximate nature of the chosen interaction energy descriptor and LFER construction of the strain term, which focuses largely on trends in substrate and not catalyst strain.

Structure and Functional Differences of Cysteine and 3-Mercaptopropionate Dioxygenases: A Computational Study

Thiol dioxygenases are important enzymes for human health; they are involved in the detoxification and catabolism of toxic thiol-containing natural products such as cysteine. As such, these enzymes have relevance to the development of Alzheimer's and Parkinson's diseases in the brain. Recent crystal structure coordinates of cysteine and 3-mercaptopropionate dioxygenase (CDO and MDO) showed major differences in the second-coordination spheres of the two enzymes. To understand the difference in activity between these two analogous enzymes, we created large, active-site cluster models. We show that CDO and MDO have different iron(III)-superoxo-bound structures due to differences in ligand coordination. Furthermore, our studies show that the differences in the second-coordination sphere and particularly the position of a positively charged Arg residue results in changes in substrate positioning, mobility and enzymatic turnover. Furthermore, the substrate scope of MDO is explored with cysteinate and 2-mercaptosuccinic acid and their reactivity is predicted.

General synthesis of single-atom catalysts with high metal loading using graphene quantum dots

Transition-metal single-atom catalysts present extraordinary activity per metal atomic site, but suffer from low metal-atom densities (typically less than 5 wt% or 1 at.%), which limits their overall catalytic performance. Here we report a general method for the synthesis of single-atom catalysts with high transition-metal-atom loadings of up to 40 wt% or 3.8 at.%, representing several-fold improvements compared to benchmarks in the literature. Graphene quantum dots, later interweaved into a carbon matrix, were used as a support, providing numerous anchoring sites and thus facilitating the generation of high densities of transition-metal atoms with sufficient spacing between the metal atoms to avoid aggregation. A significant increase in activity in electrochemical CO₂ reduction (used as a representative reaction) was demonstrated on a Ni single-atom catalyst with increased Ni loading.

A general QSPR protocol for the prediction of atomic/inter-atomic properties: a fragment based graph convolutional neural network (F-GCN)

In this study, a general quantitative structure-property relationship (QSPR) protocol, fragment based graph convolutional neural network (F-GCN), was developed for the prediction of atomic/inter-atomic properties. We applied this novel artificial intelligence (AI) tool in predictions of NMR chemical shifts and bond dissociation energies (BDEs). The obtained results were comparable to experimental measurements, while the computational cost was substantially reduced, with respect to pure density functional theory (DFT) calculations. The two important features of F-GCN can be summarised as: first, it could utilise different levels of molecular fragments for atomic/inter-atomic information extraction; second, the designed architecture is also open to include additional descriptors for a more accurate solution of the local environment at atomic level, making itself more efficient for structural solutions. And during our test, the averaged prediction error of 1H NMR chemical shifts is as small as 0.32 ppm, and the error of C-H BDE estimation is 2.7 kcal mol-1. Moreover, we further demonstrated the applicability of this developed F-GCN model via several challenging structural assignments. The success of the F-GCN in atomic and inter-atomic predictions also indicates an essential improvement of computational chemistry with the assistance of AI tools.

Thermal Treatment Effect on CO and NO Adsorption on Fe(II) and Fe(III) Species in Fe3O-Based MIL-Type Metal-Organic Frameworks: A Density Functional Theory Study

The properties of metal–organic frameworks (MOFs) based on triiron oxo-centered (Fe₃O) metal nodes are often related to the efficiency of the removal of the solvent molecules and the counteranion chemisorbed on the Fe₃O unit by postsynthetic thermal treatment. Temperature, time, and the reaction environment play a significant role in modifying key features of the materials, that is, the number of open metal sites and the reduction of Fe(III) centers to Fe(II). IR spectroscopy allows the inspection of these postsynthetic modifications by using carbon monoxide (CO) and nitric oxide (NO) as probe molecules. However, the reference data sets are based on spectra recorded for iron zeolites and oxides, whose structures are different from the Fe₃O one. We used density functional theory to study how the adsorption enthalpy and the vibrational bands of CO and NO are modified upon dehydration and reduction of Fe₃O metal nodes. We obtained a set of theoretical spectra that can model the modification observed in previously reported experimental spectra. Several CO and NO bands were previously assigned to heterogeneous Fe(II) and Fe(III) sites, suggesting a large defectivity of the materials. On the basis of the calculations, we propose an alternative assignment of these bands by considering only crystallographic iron sites. These findings affect the common description of Fe₃O-based MOFs as highly defective materials. We expect these results to be of interest to the large community of scientists working on Fe(II)- and Fe(III)-based MOFs and related materials.

Thermal treatment of triiron oxo-centered (Fe₃O)-based metal−organic frameworks is a common postsynthetic method to determine the material performances in many applications: we used density functional theory methods to study how the efficacy of the treatment modifies the energetics and the vibrational bands of nitric oxide (NO) and carbon monoxide. The obtained data set is meant to be part of the characterization toolboxes aimed at the assessment of thermal treatment protocols.

Molecular DFT+U: A Transferable, Low-Cost Approach to Eliminate Delocalization Error

While density functional theory (DFT) is widely applied for its combination of cost and accuracy, corrections (e.g., DFT+U) that improve it are often needed to tackle correlated transition-metal chemistry. In principle, the functional form of DFT+U, consisting of a set of localized atomic orbitals (AOs) and a quadratic energy penalty for deviation from integer occupations of those AOs, enables the recovery of the exact conditions of piecewise linearity and the derivative discontinuity. Nevertheless, for practical transition-metal complexes, where both atomic states and ligand orbitals participate in bonding, standard DFT+U can fail to eliminate delocalization error (DE). Here, we show that by introducing an alternative valence-state (i.e., molecular orbital or MO) basis to the DFT+U approach, we recover exact conditions in cases for which standard DFT+U corrections have no error-reducing effect. This MO-based DFT+U also eliminates DE where standard AO-based DFT+U is already successful. We demonstrate the transferability of our approach on representative transition-metal complexes with a range of ligand field strengths, electron configurations (i.e., from Sc to Zn), and spin states.

The Genesis of Molecular Volcano Plots

ConspectusFor the past two decades, linear free energy scaling relationships and volcano plots have seen frequent use as computational tools that aid in understanding and predicting the catalytic behavior of heterogeneous and electrocatalysts. Based on Sabatier's principle, which states that a catalyst should bind a substrate neither too strongly nor too weakly, volcano plots provide an estimate of catalytic performance (e.g., overpotential, catalytic cycle thermodynamics/kinetics, etc.) through knowledge of a descriptor variable. By the use of linear free energy scaling relationships, the value of this descriptor is employed to estimate the relative energies of other catalytic cycle intermediates/transition states. Postprocessing of these relationships leads to a volcano curve that reveals the anticipated performance of each catalyst, with the best species appearing on or near the peak or plateau. While the origin of volcanoes is undoubtedly rooted in examining heterogeneously catalyzed reactions, only recently has this concept been transferred to the realm of homogeneous catalysis. This Account summarizes the work done by our group in implementing and refining "molecular volcano plots" for use in analyzing and predicting the behavior of homogeneous catalysts.We begin by taking the reader through the initial proof-of-principle study that transferred the model from heterogeneous to homogeneous catalysis by examining thermodynamic aspects of a Suzuki-Miyaura cross-coupling reaction. By establishing linear free energy scaling relationships and reproducing the volcano shape, we definitively showed that volcano plots are also valid for homogeneous systems. On the basis of this key finding, we further illustrate how unified pictures of C-C cross-coupling thermodynamics were created using three-dimensional molecular volcanoes.The second section highlights an important transformation from "thermodynamic" to "kinetic" volcanoes by using the descriptor variable to directly estimate transition state barriers. Taking this idea further, we demonstrate how volcanoes can be used to directly predict an experimental observable, the turnover frequency. Discussion is also provided on how different flavors of molecular volcanoes can be used to analyze aspects of homogeneous catalysis of interest to experimentalists, such as determining the product selectivity and probing the substrate scope.The third section focuses on incorporating machine learning approaches into molecular volcanoes and invoking big-data-type approaches in the analysis of catalytic behavior. Specifically, we illustrate how machine learning can be used to predict the value of the descriptor variable, which facilitates nearly instantaneous screening of thousands of catalysts. With the large amount of data created from the machine learning/volcano plot tandem, we show how the resulting database can be mined to garner an enhanced understanding of catalytic processes. Emphasis is also placed on the latest generation of augmented volcano plots, which differ fundamentally from earlier volcanoes by elimination of the use of linear free energy scaling relationships and by assessment of the similarity of the complete catalytic cycle energy profile to that for an ideal reference species that is used to discriminate catalytic performance.We conclude by examining a handful of applications of molecular volcano plots to interesting problems in homogeneous catalysis and offering thoughts on the future prospects and uses of this new set of tools.

High throughput virtual screening of 230 billion molecular solar heat battery candidates

The dihydroazulene/vinylheptafulvene (DHA/VHF) thermocouple is a promising candidate for thermal heat batteries that absorb and store solar energy as chemical energy without the need for insulation. However, in order to be viable the energy storage capacity and lifetime of the high energy form (i.e., the free energy barrier to the back reaction) of the canonical parent compound must be increased significantly to be of practical use. We use semiempirical quantum chemical methods, machine learning, and density functional theory to virtually screen over 230 billion substituted DHA molecules to identify promising candidates. We identify a molecule with a predicted energy density of 0.38 kJ/g, which is significantly larger than the 0.14 kJ/g computed for the parent compound. The free energy barrier to the back reaction is 11 kJ/mol higher than the parent compound, which should correspond to a half-life of about 10 days—4 months. This is considerably longer than the 3–39 h (depending on solvent) observed for the parent compound and sufficiently long for many practical applications. Our paper makes two main important contributions: (1) a novel and generally applicable methodological approach that makes screening of huge libraries for properties involving chemical reactivity with modest computational resources, and (2) a clear demonstration that the storage capacity of the DHA/VHF thermocouple cannot be increased to >0.5 kJ/g by combining simple substituents.

Statistical analysis of C-H activation by oxo complexes supports diverse thermodynamic control over reactivity

Transition metal oxo species are key intermediates for the activation of strong C-H bonds. As such, there has been interest in understanding which structural or electronic parameters of metal oxo complexes determine their reactivity. Factors such as ground state thermodynamics, spin state, steric environment, oxygen radical character, and asynchronicity have all been cited as key contributors, yet there is no consensus on when each of these parameters is significant or the relative magnitude of their effects. Herein, we present a thorough statistical analysis of parameters that have been proposed to influence transition metal oxo mediated C-H activation. We used density functional theory (DFT) to compute parameters for transition metal oxo complexes and analyzed their ability to explain and predict an extensive data set of experimentally determined reaction barriers. We found that, in general, only thermodynamic parameters play a statistically significant role. Notably, however, there are independent and significant contributions from the oxidation potential and basicity of the oxo complexes which suggest a more complicated thermodynamic picture than what has been shown previously.

Shifting the scaling relations of single-atom catalysts for facile methane activation by tuning the coordination number

We investigate oxidative methane activation on a wide range of single transition metal atom catalysts embedded on N-doped graphene derivatives using density functional theory calculations. An inverse scaling relationship between *O formation and its hydrogen affinity is observed, consistent with a previous report. However, we find that the latter scaling line can be shifted towards a more reactive region by tuning the coordination number (CN) of the active metal sites. Specifically, we find that lowering the CN plays an important role in increasing the reactivity for methane activation via a radical-like transition state by moving the scaling lines. Thus, in the new design strategy suggested here, different from the conventional efforts focusing mainly on breaking the scaling relations, one maintains the scaling relations but moves them towards more reactive regions by controlling the coordination number of the active sites. With this design principle, we suggest several single atom catalysts with lower C-H activation barriers than some of the most active methane activation catalysts in the literature such as Cu-based zeolites.

Effect of Metal Ions on Oxidation of Micropollutants by Ferrate(VI): Enhancing Role of FeIV Species

This paper investigated the oxidation of recalcitrant micropollutants [i.e., atenolol (ATL), flumequine, aspartame, and diatrizoic acid] by combining ferrate(VI) (Fe^VIO₄^2-, Fe^VI) with a series of metal ions [i.e., Fe(III), Ca(II), Al(III), Sc(III), Co(II), and Ni(II)]. An addition of Fe(III) to Fe^VI enhanced the oxidation of micropollutants compared solely to Fe^VI. The enhanced oxidation of studied micropollutants increased with increasing [Fe(III)]/[Fe^VI] to 2.0. The complete conversion of phenyl methyl sulfoxide (PMSO), as a probe agent, to phenyl methyl sulfone (PMSO₂) by the Fe^VI-Fe(III) system suggested that the highly reactive intermediate Fe^IV/Fe^V species causes the increased oxidation of all four micropollutants. A kinetic modeling of the oxidation of ATL demonstrated that the major species causing the increase in ATL removal was Fe^IV, which had an estimated rate constant as (6.3 ± 0.2) × 10⁴ M^-1 s^-1, much higher than that of Fe^VI [(5.0 ± 0.4) × 10^-1 M^-1 s^-1]. Mechanisms of the formed oxidation products of ATL by Fe^IV, which included aromatic and/or benzylic oxidation, are delineated. The presence of natural organic matter significantly inhibited the removal of four pollutants by the Fe^VI-Fe(III) system. The enhanced effect of the Fe^VI-Fe(III) system was also seen in the oxidation of the micropollutants in river water and lake water.

Unravelling the cooperative role of lattice strain on MnO2/TiO2 and MnO2/ZnO catalysts for the fast decomposition of hydrogen peroxide

The rate of decomposition of hydrogen peroxide was directly correlated with lattice strain, as derived from strain calculations.

Influence of lattice strain on Fe3O4@carbon catalyst for the destruction of organic dye in polluted water using a combined adsorption and Fenton process

In this study, 8, 25 and 50 wt% Fe3O4@activated carbon (AC) catalysts were prepared by simple coprecipitation method. The efficiency of the catalysts for the advanced Fenton's oxidation process using methylene blue (MB) as a model substrate was tested. Both modified and unmodified activated carbon catalysts exhibited similar activity towards the Fenton's oxidation process. Therefore, it is difficult to identify the role of the catalyst in this dye removal process. Hence, we proposed a new methodology to remove the MB by adopting the adsorption process initially, followed by the Fenton's oxidation process. The proposed process significantly improved the methylene blue decomposition reaction over the 25 wt% Fe3O4@AC catalyst. However, this trend was not seen in pure activated carbon and Fe3O4@AC (8 and 50 wt%) catalysts due to the instability of the material in the oxidizing medium. The possible reason for the deactivation of the catalysts was evaluated from lattice strain calculations, as derived from the modified W-H models (Uniform Deformational Model (UDM), Uniform Stress Deformation Model (USDM) and Uniform Deformation Energy Density Model (UDEDM)). These results provided a quantitative relationship between the experimentally calculated lattice strain values and Fenton's catalytic activity. Furthermore, the optimized strain value and crystalite size of Fe3O4 on the activated carbon matrix are responsible for the high catalytic activity.