Electronic structure of norbornadiene and quadricyclane

TL;DR

This work investigates the ground and excited electronic structure of norbornadiene and quadricyclane as molecular photoswitches, focusing on valence-state control of photoisomerization. It introduces a new basis set and benchmarks multi-reference methods (CASSCF, MRCI, XMS-CASPT2, SHCI) for accuracy and efficiency in on-the-fly dynamics. It maps the $S_1$/$S_0$ conical intersection and analyzes the $S_2$/$S_1$ intersection with branching-plane analyses, highlighting the role of local linear representations. It shows that, with dynamic correlation, the low-energy valence manifold stays below the Rydberg manifold, enabling a three-state valence model for photodynamics of substituted energy-storage candidates and guiding method/basis choices for reliable simulation of photoinduced processes in these systems.

Abstract

The ground and excited state electronic structure of the molecular photoswitches quadricyclane and norbornadiene is examined qualitatively and quantitatively. A new custom basis set is introduced, optimised for efficient yet accurate calculations. A number of advanced multi-configurational and multi-reference electronic structure methods are evaluated, identifying those sufficiently accurate and efficient to be used in {\it{on-the-fly}} simulations of photoexcited dynamics. The key valence states participating in the isomerisation reaction are investigated, specifically mapping the important S$_1$/S$_0$ conical intersection that governs the non-radiative decay of the excited system. The powerful yet simple three-state valence model introduced here provides a suitable base for future computational exploration of the photodynamics of the substituted molecules suitable for \textit{e.g}.\ energy-storage applications.

Figures (23)

• Figure S1: Orbitals (isosurface value of 0.05) for CASSCF(4,4)/p-cc-(p)VDZ at the NBD geometry. We use the same labelling as Fig. 2 in the main text
• Figure S2: Orbital (iso-surface value of 0.05) for CASSCF(2,2)/p-cc-(p)VDZ at the NBD geometry. On the right, we show the state-averaged (SA) natural orbitals, which exhibit pronounced asymmetry. The state-specific orbitals for the S$_0$ state (left, S$_1$ state gives similar results), do not show this asymmetry, and look much closer to Fig. \ref{['fig:44orbs']}.
• Figure S3: Carbon-carbon distance coordinates for the two LIICs used in the $(r_{\mathrm{cc}},r_{\mathrm{rh}})$-plane, with key geometries labelled. All geometries are calculated at CASSCF(2,2)/p-cc-(p)VDZ level. The projection of the two branching plane coordinates $X$ and $Y$ into the plane are also shown (cf Fig. 12, main text).
• Figure S4: Branching plane X and Y vectors from S$_1$/S$_0$ MECI, optimised at XMS-CASPT2(2,2)/p-cc-(p)VDZ level. These vectors are very similar to the CASSCF(2,2) vectors shown in the main text, and all other methods tested here.
• Figure S5: S$_1$/S$_0$ MECI geometries for all combinations of CASSCF, MRCI, and XMS-CASPT2, and the (2,2) and (4,4) active spaces overlaid. All geometries show approximately the same distortions from the ground state minima.