MLatom Software Ecosystem for Surface Hopping Dynamics in Python with Quantum Mechanical and Machine Learning Methods.

We present an open-source MLatom@XACS software ecosystem for on-the-fly surface hopping nonadiabatic dynamics based on the Landau-Zener-Belyaev-Lebedev algorithm. The dynamics can be performed via Python API with a wide range of quantum mechanical (QM) and machine learning (ML) methods, including ab initio QM (CASSCF and ADC(2)), semiempirical QM methods (e.g., AM1, PM3, OMx, and ODMx), and many types of ML potentials (e.g., KREG, ANI, and MACE). Combinations of QM and ML methods can also be used. While the user can build their own combinations, we provide AIQM1, which is based on Δ-learning and can be used out-of-the-box. We showcase how AIQM1 reproduces the isomerization quantum yield of trans-azobenzene at a low cost. We provide example scripts that, in dozens of lines, enable the user to obtain the final population plots by simply providing the initial geometry of a molecule. Thus, those scripts perform geometry optimization, normal mode calculations, initial condition sampling, parallel trajectories propagation, population analysis, and final result plotting. Given the capabilities of MLatom to be used for training different ML models, this ecosystem can be seamlessly integrated into the protocols building ML models for nonadiabatic dynamics. In the future, a deeper and more efficient integration of MLatom with Newton-X will enable a vast range of functionalities for surface hopping dynamics, such as fewest-switches surface hopping, to facilitate similar workflows via the Python API.

Accelerating Molecular Vibrational Spectra Simulations with a Physically Informed Deep Learning Model.

In recent years, machine learning (ML) surrogate models have emerged as an indispensable tool to accelerate simulations of physical and chemical processes. However, there is still a lack of ML models that can accurately predict molecular vibrational spectra. Here, we present a highly efficient multitask ML surrogate model termed Vibrational Spectra Neural Network (VSpecNN), to accurately calculate infrared (IR) and Raman spectra based on dipole moments and polarizabilities obtained on-the-fly via ML-enhanced molecular dynamics simulations. The methodology is applied to pyrazine, a prototypical polyatomic chromophore. The VSpecNN-predicted energies are well within the chemical accuracy (1 kcal/mol), and the errors for VSpecNN-predicted forces are only half of those obtained from a popular high-performance ML model. Compared to the ab initio reference, the VSpecNN-predicted frequencies of IR and Raman spectra differ only by less than 5.87 cm-1, and the intensities of IR spectra and the depolarization ratios of Raman spectra are well reproduced. The VSpecNN model developed in this work highlights the importance of constructing highly accurate neural network potentials for predicting molecular vibrational spectra.

Artificial-Intelligence-Enhanced On-the-Fly Simulation of Nonlinear Time-Resolved Spectra.

Time-resolved spectroscopy is an important tool for unraveling the minute details of structural changes in molecules of biological and technological significance. The nonlinear femtosecond signals detected for such systems must be interpreted, but it is a challenging task for which theoretical simulations are often indispensable. Accurate simulations of transient absorption or two-dimensional electronic spectra are, however, computationally very expensive, prohibiting the wider adoption of existing first-principles methods. Here, we report an artificial-intelligence-enhanced protocol to drastically reduce the computational cost of simulating nonlinear time-resolved electronic spectra, which makes such simulations affordable for polyatomic molecules of increasing size. The protocol is based on the doorway-window approach for the on-the-fly surface-hopping simulations. We show its applicability for the prototypical molecule of pyrazine for which it produces spectra with high precision with respect to ab initio reference while cutting the computational cost by at least 95% compared to pure first-principles simulations.

Reduction of CO2 with Hydrated Electrons: Ab Initio Computational Studies for Finite-Size Cluster Models.

In the present work, the mechanisms of the reduction of the CO2 molecule with hydrated electrons to the hydroxyl-formyl (HOCO) radical were studied with ab initio computational methods. Hydrated hydronium radicals, H3O(H2O)n (n = 0,3,6), are considered as finite-size models of the hydrated electron in liquid water. The investigation of cluster models allows the application of high-accuracy electronic-structure methods, which are not computationally feasible in condensed-phase simulations. Reaction paths and potential-energy (PE) profiles of the proton-coupled electron-transfer reaction from hydrated H3O radicals to the CO2 molecule were explored on the ground-state PE surface. The computationally efficient unrestricted second-order Møller-Plesset method is employed, and its accuracy has been carefully benchmarked in comparison with complete-active-space self-consistent-field and multi-reference second-order perturbation calculations. The results provide insights into the interplay of electron transfer from the diffuse Rydberg-type unpaired electron of H3O to the CO2 molecule, the contraction of the electron cloud by the re-hybridization of the carbon atom of CO2, and proton transfer from the nearest water molecule to the CO2- anion, followed by Grotthus-type proton rearrangements to form stable clusters. Starting from local energy minima of hydrogen-bonded CO2-H3O(H2O)n complexes, the reaction to form HOCO-(H2O)n+1 complexes is exothermic by about 1.3 eV (125 kJ/mol). The reaction is barrier controlled with a barrier of the order of a few tenths of an electron volt, depending on size and conformation of the water cluster. This barrier is at least an order of magnitude lower than the barrier of the reaction of CO2 with any closed-shell partner molecule. The HOCO radicals can recombine by H-atom transfer (disproportionation), resulting in formic acid or a dihydroxycarbene product, as well as by the formation of a C-C bond, resulting in oxalic acid. The strong exothermicity of these radical-radical recombination reactions likely results in the fragmentation of the closed-shell products formic acid and oxalic acid, which explains the strong specificity for CO formation observed in recent experiments of Hamers and co-workers.

Ab Initio Electronic Structure Study of the Photoinduced Reduction of Carbon Dioxide with the Heptazinyl Radical.

The photocatalytic conversion of carbon dioxide to liquid fuels with electrons taken from water with solar photons is one of the grand goals of renewable energy research. Polymeric carbon nitrides recently emerged as metal-free materials with promising functionalities for hydrogen evolution from water as well as the activation of carbon dioxide. Molecular heptazine (Hz), the building block of polymeric carbon nitrides, is one the strongest known organic photo-oxidants and has been shown to be able to photo-oxidize water with near-visible light, resulting in reduced (hydrogenated) heptazine (HzH) and OH radicals. In the present work, we explored with ab initio computational methods whether the HzH chromophore is able to reduce carbon dioxide to the hydroxy-formyl (HOCO) radical in hydrogen-bonded HzH-CO2 complexes by the absorption of a photon. In remarkable contrast to the high barrier for carbon dioxide activation in the electronic ground state, the excited-state proton-coupled electron transfer (PCET) reaction is nearly barrierless, but requires the diabatic passage of three conical intersections. The possibility of barrierless carbon dioxide activation by excited-state PCET has so far not been taken into consideration in the interpretation of photocatalytic carbon dioxide reduction on carbon nitride materials.

Triangular boron carbon nitrides: an unexplored family of chromophores with unique properties for photocatalysis and optoelectronics.

It has recently been shown that cycl[3.3.3]azine and heptazine (1,3,4,6,7,9,9b-heptaazaphenalene) as well as related azaphenalenes exhibit inverted singlet and triplet states, that is, the energy of the lowest singlet excited state (S1) is below the energy of the lowest triplet excited state (T1). This feature is unique among all known aromatic chromophores and is of outstanding relevance for applications in photocatalysis and organic optoelectronics. Heptazine is the building block of the polymeric material graphitic carbon nitride which is an extensively explored photocatalyst in hydrogen evolution photocatalysis. Derivatives of heptazine have also been identified as efficient emitters in organic light emitting diodes (OLEDs). In both areas, the inverted singlet-triplet gap of heptazine is a highly beneficial feature. In photocatalysis, the absence of a long-lived triplet state eliminates the activation of atmospheric oxygen, which is favourable for long-term operational stability. In optoelectronics, singlet-triplet inversion implies the possibility of 100% fluorescence efficiency of electron-hole recombination. However, the absorption and luminescence wavelengths of heptazine and the S1-S0 transition dipole moment are difficult to tune for optimal functionality. In this work, we employed high-level ab initio electronic structure theory to devise and characterize a large family of novel heteroaromatic chromophores, the triangular boron carbon nitrides. These novel heterocycles inherit essential spectroscopic features from heptazine, in particular the inverted singlet-triplet gap, while their absorption and luminescence spectra and transition dipole moments are widely tuneable. For applications in photocatalysis, the wavelength of the absorption maximum can be tuned to improve the overlap with the solar spectrum at the surface of earth. For applications in OLEDs, the colour of emission can be adjusted and the fluorescence yield can be enhanced.