The photoinduced relaxation dynamics of pyrimidine nucleobases (uracil, thymine, and cytosine) was studied using the surface-hopping approach at the semiempirical OM2/MRCI level of theory. The relevant potential energy surfaces were characterized by performing geometry optimizations of the energy minima of the lowest electronic states and of the most important conical intersections and by computing excitation energies at each configuration. Surface-hopping molecular dynamics simulations were performed to describe the nonadiabatic dynamics after excitation into the optically active state. In each of the molecules, the two lowest excited singlet states are involved in the dynamics, and there are competing relaxation paths. The dynamics is dominated by a two-step relaxation mechanism in uracil and thymine, while the direct decay to the ground state is most important in cytosine. For all three molecules, the simulations yield ultrafast S(2)-S(1) deexcitation within 50 fs and internal conversion to the ground state in less than 1 ps, consistent with recent experimental results from time-resolved photoelectron spectroscopy.

The photoinduced hydrogen elimination reaction in phenol via the conical intersections of the dissociative 1pi sigma* state with the 1pi pi* state and the electronic ground state has been investigated by time-dependent quantum wave-packet calculations. A model including three intersecting electronic potential-energy surfaces (S0, 1pi sigma*, and 1pi pi*) and two nuclear degrees of freedom (OH stretching and OH torsion) has been constructed on the basis of accurate ab initio multireference electronic-structure data. The electronic population transfer processes at the conical intersections, the branching ratio between the two dissociation channels, and their dependence on the initial vibrational levels have been investigated by photoexciting phenol from different vibrational levels of its ground electronic state. The nonadiabatic transitions between the excited states and the ground state occur on a time scale of a few tens of femtoseconds if the 1pi pi*-1pi sigma* conical intersection is directly accessible, which requires the excitation of at least one quantum of the OH stretching mode in the 1pi pi* state. It is shown that the node structure, which is imposed on the nuclear wave packet by the initial preparation as well as by the transition through the first conical intersection (1pi pi*-1pi sigma*), has a profound effect on the nonadiabatic dynamics at the second conical intersection (1pi sigma*-S0). These findings suggest that laser control of the photodissociation of phenol via IR mode-specific excitation of vibrational levels in the electronic ground state should be possible.

The van der Waals (vdW) interfaces of two-dimensional (2D) semiconductor are central to new device concepts and emerging technologies in light-electricity transduction where the efficient charge separation is a key factor. Contrary to general expectation, efficient electron-hole separation can occur in vertically stacked transition-metal dichalcogenide heterostructure bilayers through ultrafast charge transfer between the neighboring layers despite their weak vdW bonding. In this report, we show by ab initio nonadiabatic molecular dynamics calculations, that instead of direct tunneling, the ultrafast interlayer hole transfer is strongly promoted by an adiabatic mechanism through phonon excitation occurring on 20 fs, which is in good agreement with the experiment. The atomic level picture of the phonon-assisted ultrafast mechanism revealed in our study is valuable both for the fundamental understanding of ultrafast charge carrier dynamics at vdW heterointerfaces as well as for the design of novel quasi-2D devices for optoelectronic and photovoltaic applications.

We report a theoretical study of the photoisomerization step in the operating cycle of a prototypical fluorene-based molecular rotary motor (1). The potential energy surfaces of the ground electronic state (S0) and the first singlet excited state (S1) are explored by semiempirical quantum-chemical calculations using the orthogonalization-corrected OM2 Hamiltonian in combination with a multireference configuration interaction (MRCI) treatment. The OM2/MRCI results for the S0 and S1 minima of the relevant 1-P and 1-M isomers and for the corresponding S0 transition state are in good agreement with higher-level calculations, both with regard to geometries and energetics. The S1 surface is characterized at the OM2/MRCI level by locating two S0-S1 minimum-energy conical intersections and nearby points on the intersection seam and by computing energy profiles for pathways toward these MECIs. Semiclassical Tully-type trajectory surface hopping (TSH) simulations with on-the-fly OM2/MRCI calculations are carried out to study the excited-state dynamics after photoexcitation to the S1 state. Fast relaxation to the ground state is observed through the conical intersection regions, predominantly through the lowest-energy one with a strongly twisted central C═C double bond and pyramidalized central carbon atom. The excited-state lifetimes for the direct and inverse photoisomerization reactions (1.40 and 1.79 ps) and the photostationary state ratio (2.7:1) from the TSH-OM2 simulations are in good agreement with the available experimental data (ca. 1.7 ps and 3:1). Excited-state lifetimes, photostationary state ratio, and dynamical details of the TSH-OM2 simulations also agree with classical molecular dynamics simulations using a reparametrized optimized potentials for liquid simulations (OPLS) all-atom force field with ad-hoc surface hops at predefined conical intersection points. The latter approach allows for a more extensive statistical sampling.

We discuss a theoretical approach that employs machine learning potential energy surfaces (ML-PESs) in the nonadiabatic dynamics simulation of polyatomic systems by taking 6-aminopyrimidine as a typical example. The Zhu-Nakamura theory is employed in the surface hopping dynamics, which does not require the calculation of the nonadiabatic coupling vectors. The kernel ridge regression is used in the construction of the adiabatic PESs. In the nonadiabatic dynamics simulation, we use ML-PESs for most geometries and switch back to the electronic structure calculations for a few geometries either near the S/S conical intersections or in the out-of-confidence regions. The dynamics results based on ML-PESs are consistent with those based on CASSCF PESs. The ML-PESs are further used to achieve the highly efficient massive dynamics simulations with a large number of trajectories. This work displays the powerful role of ML methods in the nonadiabatic dynamics simulation of polyatomic systems.