OrbNet Denali: A machine learning potential for biological and organic chemistry with semi-empirical cost and DFT accuracy.

We present OrbNet Denali, a machine learning model for an electronic structure that is designed as a drop-in replacement for ground-state density functional theory (DFT) energy calculations. The model is a message-passing graph neural network that uses symmetry-adapted atomic orbital features from a low-cost quantum calculation to predict the energy of a molecule. OrbNet Denali is trained on a vast dataset of 2.3 × 106 DFT calculations on molecules and geometries. This dataset covers the most common elements in biochemistry and organic chemistry (H, Li, B, C, N, O, F, Na, Mg, Si, P, S, Cl, K, Ca, Br, and I) and charged molecules. OrbNet Denali is demonstrated on several well-established benchmark datasets, and we find that it provides accuracy that is on par with modern DFT methods while offering a speedup of up to three orders of magnitude. For the GMTKN55 benchmark set, OrbNet Denali achieves WTMAD-1 and WTMAD-2 scores of 7.19 and 9.84, on par with modern DFT functionals. For several GMTKN55 subsets, which contain chemical problems that are not present in the training set, OrbNet Denali produces a mean absolute error comparable to those of DFT methods. For the Hutchison conformer benchmark set, OrbNet Denali has a median correlation coefficient of R2 = 0.90 compared to the reference DLPNO-CCSD(T) calculation and R2 = 0.97 compared to the method used to generate the training data (ωB97X-D3/def2-TZVP), exceeding the performance of any other method with a similar cost. Similarly, the model reaches chemical accuracy for non-covalent interactions in the S66x10 dataset. For torsional profiles, OrbNet Denali reproduces the torsion profiles of ωB97X-D3/def2-TZVP with an average mean absolute error of 0.12 kcal/mol for the potential energy surfaces of the diverse fragments in the TorsionNet500 dataset.

Analytical gradients for molecular-orbital-based machine learning.

Molecular-orbital-based machine learning (MOB-ML) enables the prediction of accurate correlation energies at the cost of obtaining molecular orbitals. Here, we present the derivation, implementation, and numerical demonstration of MOB-ML analytical nuclear gradients, which are formulated in a general Lagrangian framework to enforce orthogonality, localization, and Brillouin constraints on the molecular orbitals. The MOB-ML gradient framework is general with respect to the regression technique (e.g., Gaussian process regression or neural networks) and the MOB feature design. We show that MOB-ML gradients are highly accurate compared to other ML methods on the ISO17 dataset while only being trained on energies for hundreds of molecules compared to energies and gradients for hundreds of thousands of molecules for the other ML methods. The MOB-ML gradients are also shown to yield accurate optimized structures at a computational cost for the gradient evaluation that is comparable to a density-corrected density functional theory calculation.

Small Nuclear Quantum Effects in Scattering of H and D from Graphene.

We study nuclear quantum effects in H/D sticking to graphene, comparing scattering experiments at near-zero coverage with classical, quantized, and transition-state calculations. The experiment shows H/D sticking probabilities that are indistinguishable from one another and markedly smaller than those expected from a consideration of zero-point energy shifts of the chemisorption transition state. Inclusion of dynamical effects and vibrational anharmonicity via ring-polymer molecular dynamics (RPMD) yields results that are in good agreement with the experimental results. RPMD also reveals that nuclear quantum effects, while modest, arise primarily from carbon and not from H/D motion, confirming the importance of a C atom rehybridization mechanism associated with H/D sticking on graphene.

Embedded Mean-Field Theory for Solution-Phase Transition-Metal Polyolefin Catalysis.

Decreasing the wall-clock time of quantum mechanics/molecular mechanics (QM/MM) calculations without sacrificing accuracy is a crucial prerequisite for widespread simulation of solution-phase dynamical processes. In this work, we demonstrate the use of embedded mean-field theory (EMFT) as the QM engine in QM/MM molecular dynamics (MD) simulations to examine polyolefin catalysts in solution. We show that employing EMFT in this mode preserves the accuracy of hybrid-functional DFT in the QM region, while providing up to 20-fold reductions in the cost per SCF cycle, thereby increasing the accessible simulation time-scales. We find that EMFT reproduces DFT-computed binding energies and optimized bond lengths to within chemical accuracy, as well as consistently ranking conformer stability. Furthermore, solution-phase EMFT/MM simulations provide insight into the interaction strength of strongly coordinating and bulky counterions.

Imaging covalent bond formation by H atom scattering from graphene.

Viewing the atomic-scale motion and energy dissipation pathways involved in forming a covalent bond is a longstanding challenge for chemistry. We performed scattering experiments of H atoms from graphene and observed a bimodal translational energy loss distribution. Using accurate first-principles dynamics simulations, we show that the quasi-elastic channel involves scattering through the physisorption well where collision sites are near the centers of the six-membered C-rings. The second channel results from transient C-H bond formation, where H atoms lose 1 to 2 electron volts of energy within a 10-femtosecond interaction time. This remarkably rapid form of intramolecular vibrational relaxation results from the C atom's rehybridization during bond formation and is responsible for an unexpectedly high sticking probability of H on graphene.

Linear-Response Time-Dependent Embedded Mean-Field Theory.

We present a time-dependent (TD) linear-response description of excited electronic states within the framework of embedded mean-field theory (EMFT). TD-EMFT allows for subsystems to be described at different mean-field levels of theory, enabling straightforward treatment of excited states and transition properties. We provide benchmark demonstrations of TD-EMFT for both local and nonlocal excitations in organic molecules, as well as applications to chlorophyll a, solvatochromic shifts of a dye in solution, and sulfur K-edge X-ray absorption spectroscopy (XAS). It is found that mixed-basis implementations of TD-EMFT lead to substantial errors in terms of transition properties; however, as previously found for ground-state EMFT, these errors are largely eliminated with the use of Fock-matrix corrections. These results indicate that TD-EMFT is a promising method for the efficient, multilevel description of excited-state electronic structure and dynamics in complex systems.

Embedded Mean-Field Theory with Block-Orthogonalized Partitioning.

Embedded mean-field theory (EMFT) provides a simple, flexible framework for describing subsystems at different levels of mean-field theory. Subsystems are defined by partitioning a one-particle basis set, with a natural choice being the atomic orbital (AO) basis. Although generally well behaved, EMFT with AO partitioning can exhibit unphysical collapse of the self-consistent solution. To avoid this issue, we introduce subsystem partitioning of a block-orthogonalized (BO) basis set; this eliminates the unphysical collapse without significantly increasing computational cost. We also investigate a non-self-consistent implementation of EMFT, in which the density matrix is obtained using BO partitioning and the final energy evaluated using AO partitioning; this density-corrected EMFT approach is found to yield more accurate energies than BO partitioning while also avoiding issues of the unphysical collapse. Using these refined implementations of EMFT, previously proposed descriptions of the exact-exchange coupling between subsystems are compared: although the EX1 coupling scheme is slightly more accurate than EX0, the small improvement does not merit its substantially greater computational cost.

Doping Versatile n-Type Organic Semiconductors via Room Temperature Solution-Processable Anionic Dopants.

In this study, we describe a facile solution-processing method to effectively dope versatile n-type organic semiconductors, including fullerene, n-type small molecules, and graphene by commercially available ammonium and phosphonium salts via in situ anion-induced electron transfer. In addition to the Lewis basicity of anions, we unveiled that the ionic binding strength between the cation and anion of the salts is also crucial in modulating the electron transfer strength of the dopants to affect the resulting doping efficiency. Furthermore, combined with the rational design of n-type molecules, an n-doped organic semiconductor is demonstrated to be thermally and environmentally stable. This finding provides a simple and generally applicable method to make highly efficient n-doped conductors which complements the well-established p-doped organics such as PEDOT:PSS for organic electronic applications.

Aromatic thiol-mediated cleavage of N-O bonds enables chemical ubiquitylation of folded proteins.

Access to protein substrates homogenously modified by ubiquitin (Ub) is critical for biophysical and biochemical investigations aimed at deconvoluting the myriad biological roles for Ub. Current chemical strategies for protein ubiquitylation, however, employ temporary ligation auxiliaries that are removed under harsh denaturing conditions and have limited applicability. We report an unprecedented aromatic thiol-mediated N-O bond cleavage and its application towards native chemical ubiquitylation with the ligation auxiliary 2-aminooxyethanethiol. Our interrogation of the reaction mechanism suggests a disulfide radical anion as the active species capable of cleaving the N-O bond. The successful semisynthesis of full-length histone H2B modified by the small ubiquitin-like modifier-3 (SUMO-3) protein further demonstrates the generalizability and compatibility of our strategy with folded proteins.

Highly Sensitive Built-In Strain Sensors for Polymer Composites: Fluorescence Turn-On Response through Mechanochemical Activation.

A new class of rationally designed mechanophores is developed for highly sensitive built-in strain sensors in polymer composites. These mechanophores are designed to regenerate the π-conjugation pathway between the electron donor and electron acceptor by force-induced cleavage of the covalent bond to form a fluorescent dipolar dye.

Approximate singly excited states from a two-component Hartree-Fock reference.

For many molecules, relaxing the spin symmetry constraint on the wave function results in the lowest energy mean-field solution. The two-component Hartree-Fock (2cHF) method relaxes all spin symmetry constraints, and the wave function is no longer an eigenfunction of the total spin, spin projection, or time-reversal symmetry operators. For ground state energies, 2cHF is a superior mean-field method for describing spin-frustrated molecules. For excited states, the utility of 2cHF is uncertain. Here, we implement the 2cHF extensions of two single-reference excited state methods, the two-component configuration interaction singles and time-dependent Hartree-Fock. We compare the results to the analogous methods based off of the unrestricted Hartree-Fock approximation, as well as the full configuration interaction for three small molecules with distinct 2cHF solutions, and discuss the nature of the 2cHF excited state solutions.

Ab initio two-component Ehrenfest dynamics.

We present an ab initio two-component Ehrenfest-based mixed quantum/classical molecular dynamics method to describe the effect of nuclear motion on the electron spin dynamics (and vice versa) in molecular systems. The two-component time-dependent non-collinear density functional theory is used for the propagation of spin-polarized electrons while the nuclei are treated classically. We use a three-time-step algorithm for the numerical integration of the coupled equations of motion, namely, the velocity Verlet for nuclear motion, the nuclear-position-dependent midpoint Fock update, and the modified midpoint and unitary transformation method for electronic propagation. As a test case, the method is applied to the dissociation of H2 and O2. In contrast to conventional Ehrenfest dynamics, this two-component approach provides a first principles description of the dynamics of non-collinear (e.g., spin-frustrated) magnetic materials, as well as the proper description of spin-state crossover, spin-rotation, and spin-flip dynamics by relaxing the constraint on spin configuration. This method also holds potential for applications to spin transport in molecular or even nanoscale magnetic devices.

Stability of the complex generalized Hartree-Fock equations.

For molecules with complex and competing magnetic interactions, it is often the case that the lowest energy Hartree-Fock solution may only be obtained by removing the spin and time-reversal symmetry constraints of the exact non-relativistic Hamiltonian. To do so results in the complex generalized Hartree-Fock (GHF) method. However, with the loss of variational constraints comes the greater possibility of converging to higher energy minima. Here, we report the implementation of stability test of the complex GHF equations, along with an orbital update scheme should an instability be found. We apply the methodology to finding the local minima of several spin-frustrated hydrogen rings, as well as the non-collinear molecular magnet Cr3, illustrating the utility of the broken symmetry GHF method and some of its lesser-known nuances.

Time-dependent non-equilibrium dielectric response in QM/continuum approaches.

The Polarizable Continuum Models (PCMs) are some of the most inexpensive yet successful methods for including the effects of solvation in quantum-mechanical calculations of molecular systems. However, when applied to the electronic excitation process, these methods are restricted to dichotomously assuming either that the solvent has completely equilibrated with the excited solute charge density (infinite-time limit), or that it retains the configuration that was in equilibrium with the solute prior to excitation (zero-time limit). This renders the traditional PCMs inappropriate for resolving time-dependent solvent effects on non-equilibrium solute electron dynamics like those implicated in the instants following photoexcitation of a solvated molecular species. To extend the existing methods to this non-equilibrium regime, we herein derive and apply a new formalism for a general time-dependent continuum embedding method designed to be propagated alongside the solute's electronic degrees of freedom in the time domain. Given the frequency-dependent dielectric constant of the solvent, an equation of motion for the dielectric polarization is derived within the PCM framework and numerically integrated simultaneously with the time-dependent Hartree fock/density functional theory equations. Results for small molecular systems show the anticipated dipole quenching and electronic state dephasing/relaxation resulting from out-of-phase charge fluctuations in the dielectric and embedded quantum system.

Ab initio non-relativistic spin dynamics.

Many magnetic materials do not conform to the (anti-)ferromagnetic paradigm where all electronic spins are aligned to a global magnetization axis. Unfortunately, most electronic structure methods cannot describe such materials with noncollinear electron spin on account of formally requiring spin alignment. To overcome this limitation, it is necessary to generalize electronic structure methods and allow each electron spin to rotate freely. Here, we report the development of an ab initio time-dependent non-relativistic two-component spinor (TDN2C), which is a generalization of the time-dependent Hartree-Fock equations. Propagating the TDN2C equations in the time domain allows for the first-principles description of spin dynamics. A numerical tool based on the Hirshfeld partitioning scheme is developed to analyze the time-dependent spin magnetization. In this work, we also introduce the coupling between electron spin and a homogenous magnetic field into the TDN2C framework to simulate the response of the electronic spin degrees of freedom to an external magnetic field. This is illustrated for several model systems, including the spin-frustrated Li3 molecule. Exact agreement is found between numerical and analytic results for Larmor precession of hydrogen and lithium atoms. The TDN2C method paves the way for the ab initio description of molecular spin transport and spintronics in the time domain.

Quantum coherent plasmon in silver nanowires: a real-time TDDFT study.

A plasmon-like phenomenon, arising from coinciding resonant excitations of different electronic characteristics in 1D silver nanowires, has been proposed based on theoretical linear absorption spectra. Such a molecular plasmon holds the potential for anisotropic nanoplasmonic applications. However, its dynamical nature remains unexplored. In this work, quantum dynamics of longitudinal and transverse excitations in 1D silver nanowires are carried out within the real-time time-dependent density functional theory framework. The anisotropic electron dynamics confirm that the transverse transitions of different electronic characteristics are collective in nature and oscillate in-phase with respect to each other. Analysis of the time evolutions of participating one-electron wave functions suggests that the transverse transitions form a coherent wave packet that gives rise to a strong plasmon resonance at the molecular level.

Density of States Guided Møller-Plesset Perturbation Theory.

An integral formalism using a density-of-state framework has been developed for Møller-Plesset perturbation theory. This method is designed to compute the correlation energy correction for large systems with high density of states, such as polymers and nanostructures. The framework has the potential to lower the computational cost of perturbation theory, and such perspectives are discussed in this paper. This method has been implemented for the second- and third-order perturbation theory. Applications of the new methods to test cases of conjugated molecules show very good accuracy and significant savings in computational cost.

Doping of fullerenes via anion-induced electron transfer and its implication for surfactant facilitated high performance polymer solar cells.

Simple and solution-processible tetrabutyl-ammonium salts (TBAX) can dope fullerene and its derivatives to achieve conductive thin films (σ as high as 0.56 S/m). The electron transfer between the anions of TBAXs and n-type semiconductors induces doping without encountering any harsh activation. These provide valid support for the surfactant interfacial doping of fullerene in polymer solar cells for enhanced device performance.

Solution-processible highly conducting fullerenes.

n-Doping of solution-processible organic semiconductors: highly conductive fullerenes are demonstrated through solution-processed fulleropyrrolidinium iodide (FPI) and FPI-doped PCBM to reach a high conductivity (3.2 S/m). The n-doping proceeds via anion-induced electron transfer between the iodide on FPI and the fullerene in the solid state.

An efficient method for calculating dynamical hyperpolarizabilities using real-time time-dependent density functional theory.

In this paper we present a time-domain time-dependent density functional theory (TDDFT) approach to calculate frequency-dependent polarizability and hyperpolarizabilities. In this approach, the electronic degrees of freedom are propagated within the density matrix based TDDFT framework using the efficient modified midpoint and unitary transformation algorithm. We use monochromatic waves as external perturbations and apply the finite field method to extract various orders of the time-dependent dipole moment. By fitting each order of time-dependent dipole to sinusoidal waves with harmonic frequencies, one can obtain the corresponding (hyper)polarizability tensors. This approach avoids explicit Fourier transform and therefore does not require long simulation time. The method is illustrated with application to the optically active organic molecule para-nitroaniline, of which the frequency-dependent polarizability α(-ω; ω), second-harmonic generation ß(-2ω; ω, ω), optical rectification ß(0; -ω, ω), third-harmonic generation γ(-3ω; ω, ω, ω), and degenerate four-wave mixing γ(-ω; ω, ω, -ω) are calculated.