Reaction dynamics of Diels-Alder reactions from machine learned potentials.

Recent advances in the development of reactive machine-learned potentials (MLPs) promise to transform reaction modelling. However, such methods have remained computationally expensive and limited to experts. Here, we employ different MLP methods (ACE, NequIP, GAP), combined with automated fitting and active learning, to study the reaction dynamics of representative Diels-Alder reactions. We demonstrate that the ACE and NequIP MLPs can consistently achieve chemical accuracy (±1 kcal mol-1) to the ground-truth surface with only a few hundred reference calculations. These strategies are shown to enable routine ab initio-quality classical and quantum dynamics, and obtain dynamical quantities such as product ratios and free energies from non-static methods. For ambimodal reactions, product distributions were found to be strongly dependent on the QM method and less so on the type of dynamics propagated.

Computational Modeling of Supramolecular Metallo-organic Cages-Challenges and Opportunities.

Self-assembled metallo-organic cages have emerged as promising biomimetic platforms that can encapsulate whole substrates akin to an enzyme active site. Extensive experimental work has enabled access to a variety of structures, with a few notable examples showing catalytic behavior. However, computational investigations of metallo-organic cages are scarce, not least due to the challenges associated with their modeling and the lack of accurate and efficient protocols to evaluate these systems. In this review, we discuss key molecular principles governing the design of functional metallo-organic cages, from the assembly of building blocks through binding and catalysis. For each of these processes, computational protocols will be reviewed, considering their inherent strengths and weaknesses. We will demonstrate that while each approach may have its own specific pitfalls, they can be a powerful tool for rationalizing experimental observables and to guide synthetic efforts. To illustrate this point, we present several examples where modeling has helped to elucidate fundamental principles behind molecular recognition and reactivity. We highlight the importance of combining computational and experimental efforts to speed up supramolecular catalyst design while reducing time and resources.

Heterometathesis of diphosphanes (R2P-PR2) with dichalcogenides (R'E-ER', E = O, S, Se, Te).

The reactions of R2P-PR2 with R'E-ER', (where E = Se, S, O, Te) to give R2P-ER' have been explored experimentally and computationally. The reaction of Ph2P-PPh2 with PhSe-SePh gives Ph2P-SePh (1) rapidly and quantitatively. The P-P/Se-Se reaction is inhibited by the addition of the radical scavenger TEMPO which is consistent with a radical mechanism for the heterometathesis reaction. Compound 1 has been fully characterised, including by X-ray crystallography. A range of other Ar2P-SeR (R = Ph, nBu or CH2CH2CO2H) have also been prepared and characterised. The reaction of 1 with [Mo(CO)4(nbd)] (nbd = norbornadiene) gives two products which, from their characteristic 31P NMR data, have been identified as cis-[Mo(CO)4(Ph2PSePh-P)2] (8) and the mixed-donor complex cis-[Mo(CO)4(Ph2P-SePh-P)(Ph2P-SePh-Se)] (9). It is deduced that the P and Se atoms in ligand 1 have comparable capacity to coordinate to Mo(0). The reaction of Ph2P-PPh2 with PhS-SPh gives Ph2P-SPh (2) quantitatively but no reaction was observed between Ph2P-PPh2 and PhTe-TePh. Heterometathesis between Ph2P-PPh2 and tBuO-OtBu does not occur thermally but has been observed under UV irradiation to give Ph2P-OtBu along with P(V) oxidation by-products. DFT calculations have been carried out to illuminate why heterometatheses with dichalcogenides R'E-ER' occur readily when E = S and Se but not when E = O and Te. The calculations show that heterometathesis is predicted to be thermodynamically favourable for E = O, S and Se and unfavourable for E = Te. The fact that a metathesis reaction between Ph2P-PPh2 with tBuO-OtBu is not observed in the absence of UV radiation, is therefore due to kinetics.

A transferable active-learning strategy for reactive molecular force fields.

Predictive molecular simulations require fast, accurate and reactive interatomic potentials. Machine learning offers a promising approach to construct such potentials by fitting energies and forces to high-level quantum-mechanical data, but doing so typically requires considerable human intervention and data volume. Here we show that, by leveraging hierarchical and active learning, accurate Gaussian Approximation Potential (GAP) models can be developed for diverse chemical systems in an autonomous manner, requiring only hundreds to a few thousand energy and gradient evaluations on a reference potential-energy surface. The approach uses separate intra- and inter-molecular fits and employs a prospective error metric to assess the accuracy of the potentials. We demonstrate applications to a range of molecular systems with relevance to computational organic chemistry: ranging from bulk solvents, a solvated metal ion and a metallocage onwards to chemical reactivity, including a bifurcating Diels-Alder reaction in the gas phase and non-equilibrium dynamics (a model SN2 reaction) in explicit solvent. The method provides a route to routinely generating machine-learned force fields for reactive molecular systems.

Radical-initiated P,P-metathesis reactions of diphosphanes: evidence from experimental and computational studies.

By combining the diphosphanes Ar2P-PAr2, where Ar = C6H5, 4-C6H4Me, 4-C6H4OMe, 3,5-C6H3(CF3)2, it has been shown that P,P-metathesis generally occurs rapidly under ambient conditions. DFT calculations have shown that the stability of unsymmetrical diphosphanes Z2P-PZ'2 is a function of the difference between the Z and Z' substituents in terms of size and electronegativity. Of the mechanisms that were calculated for the P,P-metathesis, the most likely was considered to be one involving Ar2P˙ radicals. The observations that photolysis increases the rate of the P,P-metatheses and TEMPO inhibits it, are consistent with a radical chain process. The P,P-metathesis reactions that involve (o-Tol)2P-P(o-Tol)2 are anomalously slow and, in the absence of photolysis, were only observed to take place in CHCl3 and CH2Cl2. The role of the chlorinated solvent is ascribed to the formation of Ar2PCl which catalyses the P,P-metathesis. The slow kinetics observed with (o-Tol)2P-P(o-Tol)2 is tentatively attributed to the o-CH3 groups quenching the (o-Tol)2P˙ radicals or inhibiting the metathesis reaction sterically.

autodE: Automated Calculation of Reaction Energy Profiles- Application to Organic and Organometallic Reactions.

Calculating reaction energy profiles to aid in mechanistic elucidation has long been the domain of the expert computational chemist. Here, we introduce autodE (https://github.com/duartegroup/autodE), an open-source Python package capable of locating transition states (TSs) and minima and delivering a full reaction energy profile from 1D or 2D chemical representations. autodE is broadly applicable to study organic and organometallic reaction classes, including addition, substitution, elimination, migratory insertion, oxidative addition, and reductive elimination; it accounts for conformational sampling of both minima and TSs and is compatible with many electronic structure packages. The general applicability of autodE is demonstrated in complex multi-step reactions, including cobalt- and rhodium-catalyzed hydroformylation and an Ireland-Claisen rearrangement.

Carbonylative C-C Bond Activation of Aminocyclopropanes Using a Temporary Directing Group Strategy.

Temporary directing groups (TDGs) underpin a range of C-C bond activation methodologies; however, the use of TDGs for the regiocontrolled activation of cyclopropane C-C bonds is underdeveloped. In this report, we show how an unusual ring contraction process can be harnessed for TDG-based carbonylative C-C bond activations of cyclopropanes. The method involves the transient installation of an isocyanate-derived TDG, rather than relying on carbonyl condensation events as used in previous TDG-enabled C-C bond activations.

Synergistic Noncovalent Catalysis Facilitates Base-Free Michael Addition.

Carbon-carbon bond-forming processes that involve the deprotonation of a weakly acidic C-H pro-nucleophile using a strong Brønsted base are central to synthetic methodology. Enzymes also catalyze C-C bond formation from weakly C-H acidic substrates; however, they accomplish this at pH 7 using only collections of noncovalent interactions. Here, we show that a simple, bioinspired synthetic cage catalyzes Michael addition reactions using only Coulombic and other weak interactions to activate various pro-nucleophiles and electrophiles. The anion-stabilizing property of the cage promotes spontaneous pro-nucleophile deprotonation, suggesting acidity enhancement equivalent to several pKa units. Using a second noncovalent reagent-commercially available 18-crown-6-facilitates catalytic base-free addition of several challenging Michael partners. The cage's microenvironment also promotes high diastereoselectivity compared to a conventional base-catalyzed reaction.

cgbind: A Python Module and Web App for Automated Metallocage Construction and Host-Guest Characterization.

Metallocages offer a diverse and underexplored region of chemical space in which to search for novel catalysts and substrate hosts. However, the ability to tailor such structures toward applications in binding and catalysis is a challenging task. Here, we present an open-source computational toolkit, cgbind, that facilitates the construction, characterization, and prediction of functional metallocages. It employs known structural scaffolds as starting points and computationally efficient approaches for property evaluation. We demonstrate the ability of cgbind to construct libraries of cages with varied topologies and linker functionalities, generate accurate geometries (RMSD < 1.5 Å to crystal structures), and predict substrate binding with accuracy on par with semiempirical QM, all in seconds. The cgbind code presented here is freely available at github.com/duartegroup/cgbind and also via a web-based graphical user interface at cgbind.chem.ox.ac.uk. The protocol described here paves the way for high-throughput virtual screening of potential supramolecular structures, accelerating the search for new hosts and catalysts.

Rhodacyclopentanones as Linchpins for the Atom Economical Assembly of Diverse Polyheterocycles.

We outline a conceptual blueprint that provides direct and atom economical access to a wide range of complex polyheterocycles. Our method capitalizes on the ambiphilic reactivity of rhodacyclopentanones that arise upon exposure of cyclopropanes to Rh(I) catalysts and CO. Using this approach, a wide array of polycyclizations are achieved, including variants that involve powerful dearomatizations and medium ring formations.

Host-Guest-Induced Electron Transfer Triggers Radical-Cation Catalysis.

Modifying the reactivity of substrates by encapsulation is a fundamental principle of capsule catalysis. Here we show an alternative strategy, wherein catalytic activation of otherwise inactive quinone "co-factors" by a simple Pd2L4 capsule promotes a range of bulk-phase, radical-cation cycloadditions. Solution electron-transfer experiments and cyclic voltammetry show that the cage anodically shifts the redox potential of the encapsulated quinone by a significant 1 V. Moreover, the capsule also protects the reduced semiquinone from protonation, thus transforming the role of quinones from stoichiometric oxidants into catalytic single-electron acceptors. We envisage that the host-guest-induced release of an "electron hole" will translate to various forms of non-encapsulated catalysis that involve other difficult-to-handle, highly reactive species.

Rationalizing the Activity of an "Artificial Diels-Alderase": Establishing Efficient and Accurate Protocols for Calculating Supramolecular Catalysis.

Self-assembled cages have emerged as novel platforms to explore bioinspired catalysis. While many different size and shape supramolecular structures are now readily accessible, only a few are known to accelerate chemical reactions under substoichiometric conditions. These limited examples point to a poor understanding of cage catalysis in general, limiting the ability to design new systems. Here we show that a simple and efficient density-functional-theory-based methodology, informed by explicitly solvated molecular dynamics and coupled cluster calculations, is sufficient to accurately reproduce experimental guest binding affinities (MAD = 1.9 kcal mol-1) and identify the catalytic Diels-Alder proficiencies (>80% accuracy) of two homologous Pd2L4 metallocages with a variety of substrates. This analysis reveals how subtle structural differences in the cage framework affect binding and catalysis. These effects manifest in a smaller distortion and more favorable interaction energy for the catalytic cage compared to the inactive structure. This study gives detailed insight that would otherwise be difficult to obtain from experiments, providing new opportunities in the design of catalytically active supramolecular cages.

Catalytic Asymmetric Synthesis of Cyclohexanes by Hydrogen Borrowing Annulations.

Hydrogen borrowing catalysis serves as a powerful alternative to enolate alkylation, enabling the direct coupling of ketones with unactivated alcohols. However, to date, methods that enable control over the absolute stereochemical outcome of such a process have remained elusive. Here we report a catalytic asymmetric method for the synthesis of enantioenriched cyclohexanes from 1,5-diols via hydrogen borrowing catalysis. This reaction is mediated by the addition of a chiral iridium(I) complex, which is able to impart high levels of enantioselectivity upon the process. A series of enantioenriched cyclohexanes have been prepared and the mode of enantioinduction has been probed by a combination of experimental and DFT studies.

Stereospecific Alkene Aziridination Using a Bifunctional Amino-Reagent: An Aza-Prilezhaev Reaction.

In situ deprotection (TFA) of O-Ts activated N-Boc hydroxylamines triggers intramolecular aziridination of N-tethered alkenes to provide complex N-heterocyclic ring systems. Synthetic and computational studies corroborate a diastereospecific aza-Prilezhaev-type mechanism. The feasibility of related intermolecular alkene aziridinations is also demonstrated.

Inorganic Triphenylphosphine.

A completely inorganic version of one of the most famous organophosphorus compounds, triphenylphosphine, has been prepared. A comparison of the crystal structures of inorganic triphenylphosphine, PBaz3 (where Baz=B3 H2 N3 H3 ) and PPh3 shows that they have superficial similarities and furthermore, the Lewis basicities of the two compounds are remarkably similar. However, their oxygenation and hydrolysis reactions are starkly different. PBaz3 reacts quantitatively with water to give PH3 and with the oxidizing agent ONMe3 to give the triply-O-inserted product P(OBaz)3 , an inorganic version of triphenyl phosphite; a corresponding transformation with PPh3 is inconceivable. Thermodynamically, what drives these striking differences in the chemistry of PBaz3 and PPh3 is the great strength of the B-O bond.

A Simple and Broadly Applicable C-N Bond Forming Dearomatization Protocol Enabled by Bifunctional Amino Reagents.

A C-N bond forming dearomatization protocol with broad scope is outlined. Specifically, bifunctional amino reagents are used for sequential nucleophilic and electrophilic C-N bond formations, with the latter effecting the key dearomatization step. Using this approach, γ-arylated alcohols are converted to a wide range of differentially protected spirocyclic pyrrolidines in just two or three steps.