Exploring the frontiers of condensed-phase chemistry with a general reactive machine learning potential.

Atomistic simulation has a broad range of applications from drug design to materials discovery. Machine learning interatomic potentials (MLIPs) have become an efficient alternative to computationally expensive ab initio simulations. For this reason, chemistry and materials science would greatly benefit from a general reactive MLIP, that is, an MLIP that is applicable to a broad range of reactive chemistry without the need for refitting. Here we develop a general reactive MLIP (ANI-1xnr) through automated sampling of condensed-phase reactions. ANI-1xnr is then applied to study five distinct systems: carbon solid-phase nucleation, graphene ring formation from acetylene, biofuel additives, combustion of methane and the spontaneous formation of glycine from early earth small molecules. In all studies, ANI-1xnr closely matches experiment (when available) and/or previous studies using traditional model chemistry methods. As such, ANI-1xnr proves to be a highly general reactive MLIP for C, H, N and O elements in the condensed phase, enabling high-throughput in silico reactive chemistry experimentation.

Synthesis, Isolation, and Study of Heterobimetallic Uranyl Crown Ether Complexes.

Although crown ethers can selectively bind many metal cations, little is known regarding the solution properties of crown ether complexes of the uranyl dication, UO22+. Here, the synthesis and characterization of isolable complexes in which the uranyl dication is bound in an 18-crown-6-like moiety are reported. A tailored macrocyclic ligand, templated with a Pt(II) center, captures UO22+ in the crown moiety, as demonstrated by results from single-crystal X-ray diffraction analysis. The U(V) oxidation state becomes accessible at a quite positive potential (E1/2) of -0.18 V vs Fc+/0 upon complexation, representing the most positive UVI/UV potential yet reported for the UO2n+ core. Isolation and characterization of the U(V) form of the crown complex are also reported here; there are no prior reports of reduced uranyl crown ether complexes, but U(V) is clearly stabilized by crown chelation. Joint computational studies show that the electronic structure of the U(V) form results in significant weakening of U-Ooxo bonding despite the quite positive reduction potential at which this species can be accessed, underscoring that crown-ligated uranyl species could demonstrate unique reactivity under only modestly reducing conditions.

Linear versus Bent Uranium(II) Metallocenes─A Local Vibrational Mode Study.

Uranium metallocenes have recently attracted attention driven by their use as catalysts in organometallic synthesis. In addition to bent U(IV) and U(III), an U(II) metallocene [(η5-C5i Pr5)2U] was synthesized with an unusual linear Cp-U-Cp angle. In this work, we investigated 22 U(II) metallocenes, (i) assessing the intrinsic strength of the U-ring interactions in these complexes with a novel bond strength measure based on our local vibrational mode analysis and (ii) systematically exploring what makes these U(II) metallocenes bent. We included relativistic effects through the NESCau Hamiltonian and complemented the local mode analysis with natural bonding orbital (NBO) and quantum theory of atoms in molecules (QTAIM) data. Our study led to the following results: (i) reduction of bulky U-ring ligand substituents does not lead to bent complexes for alkyl substituents (iPr and iBu) in contrast to SiMe3 ring substituents, which are all bent. (ii) The most bent complexes are [(η5-C5H4SiMe3)2U] (130°) and [η5-P5H5)2U] (143°). (iii) Linear complexes showed one hybridized NBO with s/d character, while bent structures were characterized by s/d/f mixing. (iv) We did not observe a correlation between the strength of the U-ring interaction and the amount of the ring-U-ring bend; the strongest interaction was found for [η5-Cp)2U] and the weakest for [η5-P5H5)2U]. In conclusion, our results provide a foundation for the design of U(II) metallocenes with specific physicochemical properties and increased reactivity.

Generative adversarial networks for transition state geometry prediction.

This work introduces a novel application of generative adversarial networks (GANs) for the prediction of starting geometries in transition state (TS) searches based on the geometries of reactants and products. The multi-dimensional potential energy space of a chemical reaction often complicates the location of a starting TS geometry, leading to the correct TS combining reactants and products in question. The proposed TS-GAN efficiently maps the space between reactants and products and generates reliable TS guess geometries, and it can be easily combined with any quantum chemical software package performing geometry optimizations. The TS-GAN was trained and applied to generate TS guess structures for typical chemical reactions, such as hydrogen migration, isomerization, and transition metal-catalyzed reactions. The performance of the TS-GAN was directly compared to that of classical approaches, proving its high accuracy and efficiency. The current TS-GAN can be extended to any dataset that contains sufficient chemical reactions for training. The software is freely available for training, experimentation, and prediction at https://github.com/ekraka/TS-GAN.

Theoretical Insights into [NHC]Au(I) Catalyzed Hydroalkoxylation of Allenes: A Unified Reaction Valley Approach Study.

Hydroxylation is an effective approach for the synthesis of carbon-oxygen bonds and allylic ethers. The [NHC]Au(I) catalyzed intermolecular hydroalkoxylation of allene was studied at the DFT and Coupled Cluster level of theory. Using the Unified Reaction Valley Approach (URVA), we carry out a comprehensive mechanistic analysis of [NHC]Au(I)-catalyzed and noncatalyzed reactions. The URVA study of several possible reaction pathways reveal that the [NHC]Au(I) catalyst enables the hydroalkoxylation reaction to occur via a two step mechanism based upon the Au ability to switch between π- and σ-complexation. The first step of the mechanism involves the formation of a CO bond after the transition state with no energy penalty. Following the CO bond breakage, the OH bond breaks and CH bond forms during the second step of the mechanism, as the catalyst transforms into the more stable π-Au complex. The URVA results were complemented with local vibrational mode analysis to provide measures of intrinsic bond strength for Au(I)-allene interactions of all stationary points, and NBO analysis was applied in order to observe charge transfer events along the reaction pathway. Overall, the π-Au C═C interactions of the products are stronger than those of the reactants adding to their exothermicity. Our work on the hydroxylation of allene provides new insights for the design of effective reaction pathways to produce allylic ethers and also unravels new strategies to form C-O bonds by activation of C═C bonds.