Identifying the true reaction pathway of a prototypical ring-opening isomerization

The photochemical ring-opening reaction of 1,3-cyclohexadiene (CHD) to 1,3,5-hexatriene (HT) is a textbook example of a pericyclic reaction. In this process, according to the Woodward-Hoffmann rules, the doubly occupied LUMO of the CHD reactive state becomes the doubly occupied HOMO of HT. The evolution of photoexcited CHD has been studied with a variety of tools, possibly making this reaction the most explored with isolated-molecule fundamental methods. In general, photochemical reactions are known to be driven by conical intersections (CoIns). These are geometry points at which the energy separation between two potential energy surfaces (PESs) becomes smaller, and which act as effective funnels for transfer of population between different PESs. Near a CoIn, the adiabatic PESs, which are the eigenvalues of the electronic Hamiltonian and thus the result of electronic structure calculations, conserve their energetic ordering but not their chemical character. PESs that retain their character and cross at CoIns are known as diabatic states. Since electronic properties change smoothly only in the diabatic representation, the dynamics of the diabatic electronic populations is the one monitored in time-resolved spectroscopy. The generally accepted pathway for the CHD-HT isomerization is a valence excitation to the first bright state, labelled 1¹B, followed by a passage through a CoIn to a dark state, labelled 2¹A⁻. A branching between two pathways follows at a second CoIn, either the return to the ground state of CHD or the actual ring-opening reaction leading to HT. Here we show that this evolution is more complex than usually described.

We performed pump-probe experiments on the LDM beamline of the FERMI free electron laser, exploiting the fact that time-resolved photoemission spectra can be obtained there with resolution high enough to precisely characterize ionization from electronic states even if they are weak and/or close in energy. The pump was the seed laser for users (SLU) at a wavelength of 267 nm to excite CHD to the first bright state, and the probe was valence photoemission with a photon energy of 19.23 eV. The delay time range was from –1 to 2 ps, spanned in steps of 50-100 fs. The reaction dynamics was simulated with non-adiabatic surface-hopping trajectories. In this method, an ensemble of classical trajectories is propagated starting in the excited state, and the stochastic fewest switches algorithm is used to allow each trajectory to “hop” to different electronic states based on the non-adiabatic couplings. The energies and forces needed for propagation of the trajectories are computed using the (XMS(3)-CASPT2(6,6)) method. In Figs. 1a,b we show valence photoelectron spectra and in Figs. 1c,d intensity of the spectral features marked A,B,C,D in Figs. 1a,b as a function of pump-probe delay. The spectral features A and B correspond to the photoionization of excited states created by the pump and then evolving in time, so their identification is the key of our work.

Figure 1: Valence photoionization spectroscopy probes the time evolution of photoexcited CHD. a) experimental and b) computed photoelectron spectra plotted as a function of pump-probe delay; c) experimental and d), computed time evolution of the spectral intensity integrated over the area marked by colored rectangles in a).

In Fig. 2a we show the calculated excitation energies and charge distributions of the most relevant electronic states of CHD, in particular electron density increasing (in blue) or decreasing (in red) with respect to the ground state. The states 2¹A^–, 1¹A⁺ and 3¹A^–  have a dominant double excitation character  (see Fig. 2b). Thus, each of them satisfies the necessary condition to be the reactive state in the reaction. However, only 3¹A^– has the proper characteristic in terms of charge distribution, namely a significant reduction of electron density upon the important carbon-carbon bond. In Fig. 2c we show a cut through the multi-state PESs, both adiabatic (in black) and diabatic (in colors) along the carbon-carbon distance of the bond which breaks during the isomerization process. We can see clearly the smooth decrease in energy of the 3¹A^– state along the reaction path, which is given correctly only by the diabatic PESs.

Figure 2: The chemical character of the excited states explains their reactivity. a): excitation energies and relative absorption intensities of the 1¹B, 2¹A⁻, 1¹A⁺ and 3¹A^– states (sticks) and the corresponding maps showing the electron density difference with respect to the electronic ground state. Areas of increased and reduced electron density are shown in blue and red, respectively; b): leading configurations of the 2¹A⁻, 1¹A⁺ and 3¹A^– states; c): coordinate dependence of the potential energy of the lowest adiabatic electronic states (black) and diabatic electronic states: 1¹A⁻ (blue), 1¹B (red), 2¹A⁻ (green), 2¹B (dark violet), 1¹A+ (light violet) and 3¹A^– (orange).

In conclusion, by a combined experimental and theoretical effort, we show that the usual description of the reaction pathway in three steps is oversimplified, and the overall picture is much more consistent if diabatic rather than adiabatic states are analyzed. In particular, we show that the doubly excited dark state, labeled 2¹A^–, which is considered in the literature the gateway to the isomerization process, does not play a significant role. Instead, an initially high-lying state, labeled 3¹A^–, is the reactive state whose temporal evolution drives the reaction.

© 2022 The Authors. This work is abridged from the original, by the same Authors, under License CC BY 4.0.

This research was conducted by the following research team:

Oksana Travnikova^1,*, Tomislav Piteša^2,*, Aurora Ponzi2, Marin Sapunar², Richard James Squibb³, Robert Richter⁴, Paola Finetti⁴, Michele Di Fraia⁴, Alberto De Fanis⁵, Nicola Mahne⁶, Michele Manfredda⁴, Vitali Zhaunerchyk³, Tatiana Marchenko¹, Renaud Guillemin¹, Loic Journel¹, Kevin Charles Prince⁴, Carlo Callegari⁴, Marc Simon¹, Raimund Feifel³, Piero Decleva⁷, Nađa Došlić² and Maria Novella Piancastelli^1,8

¹ Sorbonne Université, CNRS, Laboratoire de Chimie Physique-Matière et Rayonnement, LCPMR, Paris, France
² Institut Ruđer Bošković, Zagreb, Croatia
³ Department of Physics, University of Gothenburg, Gothenburg, Sweden
⁴ Elettra-Sincrotrone Trieste, Trieste, Italy
⁵ European XFEL, Schenefeld, Germany
⁶ IOM-CNR, Trieste, Italy
⁷ Dipartimento di Scienze Chimiche e Farmaceutiche, Università di Trieste, Trieste, Italy
⁸ Department of Physics and Astronomy, Uppsala University, Uppsala, Sweden

^* These authors contributed equally

Contact person:

Local contact person: