Saulo A. Vázquez

Automated reaction discovery is often limited by the cost of high-level refinement of large reaction networks. Here, we extend AutoMeKin, an automated reaction-discovery program, by presenting a complete integration of ORCA 6 into its high-level refinement workflow, providing access to a broad range of electronic-structure methods through a package freely available to academic users. We also integrate the UMA and MACE-OMOL-0 machine-learning interatomic potentials and ∆-ML energy corrections, enabling automated refinement of reaction networks using conventional electronic-structure methods, foundation ML potentials, or ML-corrected quantum-mechanical approaches within a unified framework. The workflow is assessed for five systems, monoethanolamine, three cyclobutadiene-derived radicals, and N-[bis(2H-triazol-4-yl)methylidene]hydroxylamine, spanning both in-distribution and challenging out-of-distribution chemistry. Comparisons with reference data show that UMA and MACE-OMOL-0 generally recover the relevant reaction mechanisms and provide energy profiles of comparable quality to DFT at substantially reduced cost. RRKM analyses further reveal that vibrational frequencies can significantly affect product population fractions, especially for loose transition states. These results show that the new implementation increases the accessibility and flexibility of AutoMeKin and further support the view that foundation ML potentials can serve as practical surrogates for DFT in automated reaction-network discovery.

Analytical corrections were developed to improve the accuracy of the PM6 and GFN2-xTB semiempirical quantum mechanical methods for the evaluation of noncovalent interaction energies in alkanes and alkenes. We followed the approach of functional group corrections, wherein the atom–atom pair corrections depend on the nature of the interacting functional groups. The training set includes 21 alkane and 13 alkene complexes taken from the Donchev et al.’s database [Sci. Data 8, 55 (2021)], with interaction energies calculated at the CCSD(T)/CBS level, and our own data obtained for medium-size complexes (of 100 and 112 atoms). In general, for the systems included in the training and validation sets, the errors obtained with the PM6-FGC and xTB-FGC methods are within the chemical accuracy.

Electrostatic penetration shapes the total interaction energy in aromatic dimers.

Recently, we reported a new approach to develop pairwise analytical corrections to improve the description of noncovalent interactions, by approximate methods of electronic structures, such as semiempirical quantum mechanical (SQM) methods. In particular, and as a proof of concept, we used the PM6 Hamiltonian and we named the method PM6-FGC, where the FGC acronym, corresponding to Functional Group Corrections, emphasizes the idea that the corrections work for specific functional groups rather than for individual atom pairs. The analytical corrections were derived from fits to B3LYP-D3/def2-TZVP (reference). PM6 interaction energy differences, evaluated for a reduced set of small bimolecular complexes, were chosen as representatives of saturated hydrocarbons, carboxylic, amine and, tentatively, amide functional groups. For the validation, the method was applied to several complexes of well-known databases, as well as to complexes of diglycine and dialanine, assuming the transferability of amine group corrections to amide groups. The PM6-FGC method showed great potential but revealed significant inaccuracies for the description of some interactions involving the –NH₂ group in amines and amides, caused by the inadequate selection of the model compound used to represent these functional groups (an NH₃ molecule). In this work, methylamine and acetamide are used as representatives of amine and amide groups, respectively. This new selection leads to significant improvements in the calculation of noncovalent interactions in the validation set.

Vibrationally excited deuterium fluoride (DF) formed by fluorine atom reaction with a solvent was found (Science, 2015, 347, 530) to relax rapidly (less than 10 ps) in acetonitrile-d₃ (CD₃CN) and dichloromethane-d₂ (CD₂Cl₂). However, insights into how CD₂Cl₂ facilitates this energy relaxation have so far been lacking, given the weak interaction between DF and a single CD₂Cl₂. In this work, we report the results of reactive simulations with a two-state reactive empirical valence bond (EVB) potential to study the energy deposited into nascent DF after transition-state passage and of nonequilibrium molecular dynamics simulations using multiple different potential energy functions to model the relaxation dynamics. For these second simulations, we used the standard Merck molecular force field (MMFF) potential, an MMFF-based covalent-ionic empirical valence bond (EVB) potential (EVB^CI), a newly developed potential [referred to as MMFF(rDF)] which extends upon the MMFF potential by making the DF/CD₂Cl₂ interaction depend on the value of the D—F bond stretching coordinate and by taking the anisotropic charge distribution of the solvent molecules into account, the polarizable atomic multipole optimized energetics for biomolecular applications (AMOEBA) potential, and the quantum mechanics/molecular mechanics (QM/MM) potential. The relaxation is revealed to be highly sensitive to the potential used. Neither standard MMFF nor EVB^CI reproduces the experimentally observed rapid relaxation dynamics, and they also fail to provide a good description of the interaction potential between DF and CD₂Cl₂ as calculated using CCSD(T)-F12. This is attributed to the use of a point-charge model for the solute and to failing to model the anisotropic electrostatic properties of CD₂Cl₂. The MMFF(rDF), AMOEBA, and QM/MM potentials all reproduce the CCSD(T)-F12 two-body DF---CD₂Cl₂ interaction potential rather well but only with the QM/MM approach is fast vibrational relaxation obtained (lifetimes of ∼288, ∼186, and ∼8 ps, respectively), which we attribute to differences in the solute–solvent local structure. With QM/MM, a unique “many-body” interaction pattern in which DF is in close contact with two solvent Cl atoms and more than three solvent D atoms is found, but this structure is not seen with other potentials. The QM/MM dynamics also display enhanced solute–solvent interactions with vibrationally excited DF that induce a DF band redshift and hence a resonant overlap with solvent C—D modes, which facilitate the intermolecular energy transfer. Our work also suggests that potentials used to model energy relaxation need to capture the fine structure of solute–solvent interactions and not just the two-body part.

The nature of the substituent effect depends markedly on the relative position of the interacting rings. Contributions not directly related to the π–π interaction can significantly affect the observed value.