Davide Avagliano

Nonadiabatic molecular dynamics (NAMD) simulations are highly sensitive to the underlying electronic structure description. Within the complete active space second-order perturbation theory (CASPT2) framework, several multistate formulations, namely, multistate (MS)-, extended multistate (XMS)-, and rotated multistate (RMS)-CASPT2, are available, yet their relative robustness for on-the-fly NAMD remains an open question. In this work, we systematically assess the sensitivity of these three CASPT2 variants with respect to state averaging by performing NAMD simulations on two representative molecular test systems, fulvene and ethylene, using both trajectory surface hopping and ab initio multiple spawning approaches. For fulvene, pronounced differences are observed between the CASPT2 flavors. MS-CASPT2 shows a strong dependence on the number of averaged states, affecting not only excited-state lifetimes but also the nuclear evolution along key reaction coordinates. XMS-CASPT2 displays improved stability, with only minor variations in population dynamics and structural evolution upon increasing the number of states. RMS-CASPT2 proves to be the most robust formulation for this system, yielding consistent results across different state-averaging schemes. In contrast, ethylene represents a well-behaved case where all three CASPT2 formulations provide consistent dynamical results largely independent of state averaging. Overall, this work confirms the general reliability of CASPT2-based NAMD while highlighting that the choice of CASPT2 flavor and number of states can significantly influence both population dynamics and nuclear motion. Careful preliminary electronic-structure analysis and chemical intuition remain essential for meaningful CASPT2-based dynamics simulations. The benchmarking protocol presented here provides a foundation for future systematic assessments on more complex photochemical processes, aiming toward community-wide guidelines for the application of CASPT2 in nonadiabatic dynamics.

Nitrobenzochalcogenadiazole derivatives are emerging candidates for photodynamic therapy (PDT), yet the precise mechanisms governing their excited‐state deactivation and triplet generation remain to be fully elucidated. This study employs nonadiabatic dynamics simulations, using the trajectory surface hopping method within a linear vibronic coupling (LVC) framework, to unravel the intersystem crossing (ISC) pathways in these systems. We systematically investigate two design factors: the heavy‐atom effect (substituting S with Se and Te) and the influence of “push–pull” electronic architectures (donor–acceptor vs. donor–donor/acceptor–acceptor). Our results demonstrate that replacing sulfur with selenium and tellurium monotonically accelerates ISC, reducing excited‐state lifetimes from 6.1 ps to sub‐picosecond timescales ( 0.9 ps) via a dominant relaxation channel driven by enhanced spin–orbit coupling. Furthermore, we reveal that structural modifications that disrupt this push–pull nature (D–D or A–A) result in kinetic bottlenecks, trapping the population in the singlet manifold. These dynamical insights establish clear structure–property relationships, guiding the rational design of photosensitizers with optimized triplet quantum yields.

We report a new workflow to model time-resolved transient absorption spectra based on propagating non-adiabatic dynamics with ab initio multiple spawning. By explicitly including a laser field in the molecular Hamiltonian, we can model both the effect of a pump pulse in depopulating the ground state and a series of probe pulses to monitor the evolution of the nuclear density. We developed a flexible and efficient workflow that enables the inclusion of probes of different energies at different time delays, matching real experimental conditions. In this way, we are able to reconstruct the spectroscopic signal from a single set of trajectories. We included our workflow in the open-source code PySpawn to exploit all of the implementations recently added to this software and to ensure wider usability of our method. Finally, we showed the applicability and strength of our development in simulating and interpreting the time-resolved transient absorption spectrum of azomethane.

We have recently demonstrated the ability of using self-driving laboratories for AI-driven searches of organic emitters for solid-state lasing devices. Our past workflow featured solubility challenges for such large molecular moieties. In this next-generation study, we return to the drawing board to explore a family of compounds that are much solution processable and composed of a set of electronic cores that provide a broader color response. Out of 252 potential candidates, and with guidance from DFT calculations, we selectively perform a comprehensive study exploring 51 fluorene-based A-B-A-type organic laser oligomers, armed with our self-driving lab. The candidates range from simple hydrocarbon molecules to complex heteroatom-mixed molecules. As a result of this study, we highlight diketopyrrolopyrrole and benzodiazole derivatives for their largely red-shifted emissions. Furthermore, we investigate the effect of color change arising from heteroatom permutation, fluorine addition, thiophene coupling, and a combination of fluorine addition and thiophene coupling. Amplified spontaneous emission (ASE) measurements in the solid state further corroborate the lasing potential of selected candidates, reinforcing their suitability for future device applications. The computational study with density functional theory confirms the experimental results.

Explicitly modelling different solvent environments and the incident pulse influences the ultrafast photochemistry of furan.

Shared, structured representations can link excited-state dynamics studies into knowledge that is easier to understand, compare, and reuse. PhotoChem Explorer explores what it takes to support more cumulative knowledge building in computational photochemistry.

Mol2Raman is a graph neural network that predicts Raman spectra from molecular SMILES. Trained on >31k DFT-calculated spectra, it localizes peaks within 15 cm ⁻¹ with 64% accuracy, outperforming current SOTA deep learning algorithms on Raman spectra.

We have recently demonstrated the ability of using self-driving laboratories for AI-driven searches of new emitters for organic solid-state lasing devices. Our past workflow featured solubility challenges for such large molecular moieties. In this next-generation study, we returned to the drawing board to explore a family of compounds that were much solution processable and composed of a set of electronic cores that would give a broader color response. Out of 252 potential candidates, we selectively performed a comprehensive study exploring 51 fluorene-based A-B-A type organic laser oligomers, armed with our self-driving lab. The candidates ranged from simple hydrocarbon molecules to complex hetero atom-mixed molecules. As a result of this study, we highlight diketopyrrolopyrrole and benzodiazole derivatives for largely red-shifted emissions. Furthermore, we investigated the effect of color change from hetero atom permutation, fluorine addition, thiophene coupling, and a combination of fluorine addition and thiophene coupling. The computational study with density functional theory confirmed the experimental results.

We report a new workflow to model time-resolved transient absorption spectra, based on propagating nonadiabatic dynamics with ab initio multiple spawning. By explicitly including a laser field in the molecular Hamiltonian, we are able to model both the effect of a pump pulse in depopulating the ground state and a series of probe pulses to monitor the evolution of the nuclear density. We developed a flexible and efficient workflow that enables the inclusion of probes of different energies at different time delays, matching real experimental conditions. In this way, we are able to reconstruct the spectroscopic signal from a single set of trajectories. We included our workflow in the open-source code PySpawn to exploit all the implementations recently added in this software and to ensure wider usability of our method. Finally, we showed the applicability and strength of our development in simulating and interpreting the time-resolved transient absorption spectrum of azomethane.

An exhaustive benchmark of density functional approximations (DFAs) for nonadiabatic dynamics is reported on the trans-cis photoisomerization of the protonated Schiff base 3 (PSB3), which presents numerous challenges for time-dependent density functional theory (TD-DFT). We introduce a rigorous protocol for benchmarking DFAs for nonadiabatic dynamics regarding the initialization, the dynamics, and its evaluation. Different families of DFAs were compared with a high-level reference, highlighting that electronic populations are an unsuitable metric for evaluating the accuracy of dynamics. We found that several local functionals showed the best agreement of the population decay with the reference RMS-CASPT2, but strictly passed through a deactivation channel dominated by a single-bond torsion that is not accessible in the reference and in the literature. While using 100% Hartree-Fock exchange in the functional yields the only correct isomerization behavior, the time scales and quantum yields are far off the reference values, due to an artificial local minimum being predicted along the wrong torsion coordinate. Static energy scans suggest that this issue can be circumvented by double hybrid functionals, in particular those balancing nonlocal exchange and correlation combined with range-separation. Indeed, they predict energy profiles along the two torsion coordinates in close agreement with the RMS-CASPT2 reference. This emphasizes the impact these DFAs will have on the field of nonadiabatic dynamics once analytical gradients are introduced.