Bing Gu

Electron donor–acceptor (EDA) complexes, formed through noncovalent interactions between electron‐rich and electron‐deficient species, can undergo photoinduced charge transfer to access reactive radical intermediates. However, rapid and unproductive back electron transfer (BET) limits their synthetic utility. Herein, we report a dual‐function ( η ⁶ –arene)rhodium(III) catalyst that enables contra ‐thermodynamic E ‐to‐ Z isomerization of stilbenes via selective EDA complexation. In contrast to well‐studied photoisomerization reactions driven by triplet photosensitizers via intermolecular energy transfer (EnT), photoexcitation of the self‐assembled EDA complex generates a charge‐separated singlet exciplex that undergoes back electron transfer coupled with intersystem crossing to populate a localized triplet‐state olefin intermediate. The catalyst plays two pivotal roles: the metal‐bound arene scaffold enables molecular recognition of planar ( E )‐stilbenes over twisted ( Z )‐isomers through noncovalent interactions, while the heavy rhodium atom promotes ISC.

Electron donor–acceptor (EDA) complexes, formed through noncovalent interactions between electron‐rich and electron‐deficient species, can undergo photoinduced charge transfer to access reactive radical intermediates. However, rapid and unproductive back electron transfer (BET) limits their synthetic utility. Herein, we report a dual‐function ( η ⁶ –arene)rhodium(III) catalyst that enables contra ‐thermodynamic E ‐to‐ Z isomerization of stilbenes via selective EDA complexation. In contrast to well‐studied photoisomerization reactions driven by triplet photosensitizers via intermolecular energy transfer (EnT), photoexcitation of the self‐assembled EDA complex generates a charge‐separated singlet exciplex that undergoes back electron transfer coupled with intersystem crossing to populate a localized triplet‐state olefin intermediate. The catalyst plays two pivotal roles: the metal‐bound arene scaffold enables molecular recognition of planar ( E )‐stilbenes over twisted ( Z )‐isomers through noncovalent interactions, while the heavy rhodium atom promotes ISC.

We investigate the influence of strong static disorder on the transition frequency of molecules to the superradiant quantum phase transition in the Dicke model. A spin–fermion mapping is employed to recast the Dicke model into a coupled fermion–boson problem, which is subsequently treated by the non-equilibrium Green’s function formalism. An analytical expression for the critical point as a function of the disorder strength is derived. In particular, there exists a regime where increasing the disorder strength reduces the critical coupling strength.

PyQED is an open-source Python package designed for the numerical simulation of strongly coupled electron-nuclear quantum dynamics, in particular, conical intersection dynamics. Besides conventional nona- diabatic wavepacket dynamics methods based on the Born–Huang representation and mixed quantum- classical Ehrenfest dynamics, PyQED implements the geometric quantum dynamics based on the local diabatic representation, which provides a numerically exact framework for nonadiabatic quantum molecular dynamics. It differs from the conventional Born–Huang representation in that all non-Born–Oppenheimer effects are accounted for by a single electronic overlap matrix between adiabatic states, therefore removing the singular derivative couplings. The complete workflow for ab initio modeling of conical intersection dynamics is illustrated through the internal conversion dynamics in the H3+ cation. PyQED provides a powerful and user-friendly computational platform for first-principles quantum dynamics, with applications to photochemical and photophysical processes.

The local diabatic representation is demonstrated to be highly accurate and robust for nonadiabatic eigenvalue problems, even with divergent nonadiabatic couplings at conical intersections.

Quantum geometrical molecular dynamics provides a quantum geometric picture for understanding reactive dynamics, especially excited-state conical intersection dynamics, and also a numerically exact method for strongly correlated electron-nuclear dynamics. However, there are substantial challenges in describing medium-sized molecules with tens of nuclear degrees of freedom. The main challenge is that it uses a discrete variable representation to discretize the molecular configuration space, and thus requires a tremendous number of quantum chemistry calculations to construct the electronic overlap matrix. Moreover, the expansion coefficients scale exponentially with molecular size for direct-product basis sets. We address these challenges by first introducing a coarse-grained local diabatic ansatz, followed by a tensor network representation of the expansion coefficients and the molecular time-evolution operator. With a full 24-dimensional demonstration using the pyrazine molecule, we show that such developments provide a highly accurate and computationally tractable method for high-dimensional, fully quantum, strongly coupled electron-nuclear dynamics from first principles.

We develop a unified and exact quantum geometric framework to understand and model molecular reactive quantum dynamics. The critical roles of quantum geometry of adiabatic electronic states in both adiabatic and nonadiabatic quantum dynamics are unveiled. A numerically exact geometric quantum molecular dynamics method is developed via discrete local trivialization of the projected electronic Hilbert space bundle over nuclear configuration space, eliminating all singularities from nonanalytic adiabatic electronic states. In it, the singular electronic quantum geometric tensor—Abelian for adiabatic dynamics and non-Abelian for nonadiabatic dynamics—is fully encoded in the global electronic overlap matrix. Numerical illustrations demonstrate that atomic motion, whether adiabatic or nonadiabatic, is governed not only by variations in electronic energies (potential energy surfaces) but also by variations in electronic states (electronic quantum geometry). Beyond quantum molecular dynamics, the strategy of discrete local trivialization can be extended to describe quantum dynamics, possibly non-Hermitian, on arbitrary, especially nondifferential fiber bundles.

The global many-electron wave function overlap matrix accounts for all effects beyond the Born-Oppenheimer approximation in the discrete variable local diabatic representation, a numerically exact framework for modeling nonadiabatic conical intersection wave packet dynamics. Nevertheless, calculating the electronic overlap matrix from the electronic structure is computationally expensive. Here, we introduce an approximation for constructing the electronic overlap matrix between any two long-range geometries by the product of nearest-neighbor overlap matrices (links) along a path connecting these two geometries. This approximation significantly reduces the computational effort by requiring electronic structure calculations only for the nearest-neighbor overlap matrices. The accuracy of this approximation is demonstrated through an exact simulation of a proton-coupled electron transfer model. Our results show that although the approximate overlap matrix can exhibit noticeable differences from the exact ones, the conical intersection dynamics is in almost exact agreement with those using the exact overlap matrix.

The discrete variable local diabatic representation (LDR) provides a divergence-free framework for exact conical intersection dynamics simulation. In this work, we investigate the convergence with respect to the number of "nuclear" grid points and "electronic" states of LDR for the eigenvalue problems using coupled oscillator models. The performance of LDR is compared with traditional approaches based on the Born-Huang ansatz and on the crude adiabatic representation. Our results demonstrate that for weak vibronic couplings, LDR shows similar convergence rate as the exact Born-Huang representation including not only the first-order derivative couplings but also the diagonal Born-Oppenheimer corrections and second-order derivative couplings. Surprisingly, for strong vibronic couplings, LDR shows a significant faster convergence rate with respect to the number of grid points, hence the number of electronic structure computations, than the exact Born-Huang representation. The crude adiabatic representation in generally shows a much slower convergence rate for all cases. The diagonal Born-Oppenheimer corrections and second-order derivative couplings are found to be important in the Born-Huang framework.