Andrew E. Sifain

Fewest Switches Surface Hopping in Liouville Space.

The novel approach to nonadiabatic quantum dynamics greatly increases the accuracy of the most popular semiclassical technique while maintaining its simplicity and efficiency. Unlike the standard Tully surface hopping in Hilbert space, which deals with population flow, the new strategy in Liouville space puts population and coherence on equal footing. Dual avoided crossing and energy transfer models show that the accuracy is improved in both diabatic and adiabatic representations and that Liouville space simulation converges faster with the number of trajectories than Hilbert space simulation. The constructed master equation accurately captures superexchange, tunneling, and quantum interference. These effects are essential for charge, phonon and energy transport and scattering, exciton fission and fusion, quantum optics and computing, and many other areas of physics and chemistry.

Communication: Proper treatment of classically forbidden electronic transitions significantly improves detailed balance in surface hopping.

Surface hopping is the most popular method for nonadiabatic molecular dynamics. Many have reported that it does not rigorously attain detailed balance at thermal equilibrium, but does so approximately. We show that convergence to the Boltzmann populations is significantly improved when the nuclear velocity is reversed after a classically forbidden hop. The proposed prescription significantly reduces the total number of classically forbidden hops encountered along a trajectory, suggesting that some randomization in nuclear velocity is needed when classically forbidden hops constitute a large fraction of attempted hops. Our results are verified computationally using two- and three-level quantum subsystems, coupled to a classical bath undergoing Langevin dynamics.

Mixed quantum-classical equilibrium in global flux surface hopping

Global flux surface hopping (GFSH) generalizes fewest switches surface hopping (FSSH)—one of the most popular approaches to nonadiabatic molecular dynamics—for processes exhibiting superexchange. We show that GFSH satisfies detailed balance and leads to thermodynamic equilibrium with accuracy similar to FSSH. This feature is particularly important when studying electron-vibrational relaxation and phonon-assisted transport. By studying the dynamics in a three-level quantum system coupled to a classical atom in contact with a classical bath, we demonstrate that both FSSH and GFSH achieve the Boltzmann state populations. Thermal equilibrium is attained significantly faster with GFSH, since it accurately represents the superexchange process. GFSH converges closer to the Boltzmann averages than FSSH and exhibits significantly smaller statistical errors.

Communication: Global flux surface hopping in Liouville space

Recent years have witnessed substantial progress in the surface hopping (SH) formulation of non-adiabatic molecular dynamics. A generalization of the traditional fewest switches SH (FSSH), global flux SH (GFSH) utilizes the gross population flow between states to derive SH probabilities. The Liouville space formulation of FSSH puts state populations and coherences on equal footing, by shifting the hopping dynamics from Hilbert to Liouville space. Both ideas have shown superior results relative to the standard FSSH in Hilbert space, which has been the most popular approach over the past two and a half decades. By merging the two ideas, we develop GFSH in Liouville space. The new method is nearly as straightforward as the standard FSSH, and carries comparable computational expense. Tested with a representative super-exchange model, it gives the best performance among all existing techniques in the FSSH series. The obtained numerical results match almost perfectly the exact quantum mechanical solutions. Moreover, the results are nearly invariant under the choice of a basis state representation for SH, in contrast to the earlier techniques which exhibit notable basis set dependence. Unique to the developed approach, this property is particularly encouraging, because exact quantum dynamics is representation independent. GFSH in Liouville space significantly improves accuracy and applicability of SH for a broad range of chemical and physical processes.

Two-Photon Absorption in Conjugated Energetic Molecules.

Time-dependent density functional theory (TD-DFT) was used to investigate the relationship between molecular structure and the one- and two-photon absorption (OPA and TPA, respectively) properties of novel and recently synthesized conjugated energetic molecules (CEMs). The molecular structures of CEMs can be strategically altered to influence the heat of formation and oxygen balance, two factors that can contribute to the sensitivity and strength of an explosive material. OPA and TPA are sensitive to changes in molecular structure as well, influencing the optical range of excitation. We found calculated vertical excitation energies to be in good agreement with experiment for most molecules. Peak TPA intensities were found to be significant and on the order of 10(2) GM. Natural transition orbitals for essential electronic states defining TPA peaks of relatively large intensity were used to examine the character of relevant transitions. Modification of molecular substituents, such as additional oxygen or other functional groups, produces significant changes in electronic structure, OPA, and TPA and improves oxygen balance. The results show that certain molecules are apt to undergo nonlinear absorption, opening the possibility for controlled, direct optical initiation of CEMs through photochemical pathways.

Cooperative enhancement of the nonlinear optical response in conjugated energetic materials: A TD-DFT study

Conjugated energetic molecules (CEMs) are a class of explosives with high nitrogen content that posses both enhanced safety and energetic performance properties and are ideal for direct optical initiation. As isolated molecules, they absorb within the range of conventional lasers. Crystalline CEMs are used in practice, however, and their properties can differ due to intermolecular interaction. Herein, time-dependent density functional theory was used to investigate one-photon absorption (OPA) and two-photon absorption (TPA) of monomers and dimers obtained from experimentally determined crystal structures of CEMs. OPA scales linearly with the number of chromophore units, while TPA scales nonlinearly, where a more than 3-fold enhancement in peak intensity, per chromophore unit, is calculated. Cooperative enhancement depends on electronic delocalization spanning both chromophore units. An increase in sensitivity to nonlinear laser initiation makes these materials suitable for practical use. This is the first study predicting a cooperative enhancement of the nonlinear optical response in energetic materials composed of relatively small molecules. The proposed model quantum chemistry is validated by comparison to crystal structure geometries and the optical absorption of these materials dissolved in solution.

Correction Scheme for Comparison of Computed and Experimental Optical Transition Energies in Functionalized Single-Walled Carbon Nanotubes

Covalent functionalization of single-walled carbon nanotubes (SWCNTs) introduces red-shifted emission features in the near-infrared spectral range due to exciton localization around the defect site. Such chemical modifications increase their potential use as near-infrared emitters and single-photon sources in telecommunications applications. Density functional theory (DFT) studies using finite-length tube models have been used to calculate their optical transition energies. Predicted energies are typically blue-shifted compared to experiment due to methodology errors including imprecise self-interaction corrections in the density functional and finite-size basis sets. Furthermore, artificial quantum confinement in finite models cannot be corrected by a constant-energy shift since they depend on the degree of exciton localization. Herein, we present a method that corrects the emission energies predicted by time-dependent DFT. Confinement and methodology errors are separately estimated using experimental data for unmodified tubes. Corrected emission energies are in remarkable agreement with experiment, suggesting the value of this straightforward method toward predicting and interpreting the optical features of functionalized SWCNTs.

Photoexcited Nonadiabatic Dynamics of Solvated Push-Pull π-Conjugated Oligomers with the NEXMD Software

Solvation can be modeled implicitly by embedding the solute in a dielectric cavity. This approach models the induced surface charge density at the solute-solvent boundary, giving rise to extra Coulombic interactions. Herein, the Nonadiabatic EXcited-state Molecular Dynamics (NEXMD) software was used to model the photoexcited nonradiative relaxation dynamics in a set of substituted donor-acceptor oligo( p-phenylenevinylene) (PPVO) derivatives in the presence of implicit solvent. Several properties of interest including optical spectra, excited state lifetimes, exciton localization, excited state dipole moments, and structural relaxation are calculated to elucidate dependence of functionalization and solvent polarity on photoinduced nonadiabatic dynamics. Results show that solvation generally affects all these properties, where the magnitude of these effects vary from one system to another depending on donor-acceptor substituents and molecular polarizability. We conclude that implicit solvation can be directly incorporated into nonadiabatic simulations within the NEXMD framework with little computational overhead and that it qualitatively reproduces solvent-dependent effects observed in solution-based spectroscopic experiments.

Discovering a Transferable Charge Assignment Model Using Machine Learning

Partial atomic charge assignment is of immense practical value to force field parametrization, molecular docking, and cheminformatics. Machine learning has emerged as a powerful tool for modeling chemistry at unprecedented computational speeds given accurate reference data. However, certain tasks, such as charge assignment, do not have a unique solution. Herein, we use a machine learning algorithm to discover a new charge assignment model by learning to replicate molecular dipole moments across a large, diverse set of nonequilibrium conformations of molecules containing C, H, N, and O atoms. The new model, called Affordable Charge Assignment (ACA), is computationally inexpensive and predicts dipoles of out-of-sample molecules accurately. Furthermore, dipole-inferred ACA charges are transferable to dipole and even quadrupole moments of much larger molecules than those used for training. We apply ACA to dynamical trajectories of biomolecules and produce their infrared spectra. Additionally, we find that ACA assigns similar charges to Charge Model 5 but with greatly reduced computational cost.

Transferable Dynamic Molecular Charge Assignment Using Deep Neural Networks

The ability to accurately and efficiently compute quantum-mechanical partial atomistic charges has many practical applications, such as calculations of IR spectra, analysis of chemical bonding, and classical force field parametrization. Machine learning (ML) techniques provide a possible avenue for the efficient prediction of atomic partial charges. Modern ML advances in the prediction of molecular energies [i.e., the hierarchical interacting particle neural network (HIP-NN)] has provided the necessary model framework and architecture to predict transferable, extensible, and conformationally dynamic atomic partial charges based on reference density functional theory (DFT) simulations. Utilizing HIP-NN, we show that ML charge prediction can be highly accurate over a wide range of molecules (both small and large) across a variety of charge partitioning schemes such as the Hirshfeld, CM5, MSK, and NBO methods. To demonstrate transferability and size extensibility, we compare ML results with reference DFT calculations on the COMP6 benchmark, achieving errors of 0.004e- (elementary charge). This is remarkable since this benchmark contains two proteins that are multiple times larger than the largest molecules in the training set. An application of our atomic charge predictions on nonequilibrium geometries is the generation of IR spectra for organic molecules from dynamical trajectories on a variety of organic molecules, which show good agreement with calculated IR spectra with reference method. Critically, HIP-NN charge predictions are many orders of magnitude faster than direct DFT calculations. These combined results provide further evidence that ML (specifically HIP-NN) provides a pathway to greatly increase the range of feasible simulations while retaining quantum-level accuracy.