Arresting doubly excited electronic state mediating lightning speed ring-opening reaction of 1,3-cyclohexadiene

The photoinduced ring-opening reaction of 1,3-cyclohexadiene to produce 1,3,5-hexatriene is a well-known example of the Woodward-Hoffmann rule for stereochemical reactions governed by molecular orbital symmetry, and it plays an essential role in photobiological synthesis of vitamin D 3 in the skin. Since the photoexcited 1 1 B state of 1,3-cyclohexadiene is not electronically correlated to the ground state of 1,3,5-hexatriene, the reaction is expected to proceed via non-adiabatic transitions through a doubly excited 2 1 A electronic state. However, spectroscopic observation of this elusive state has been difficult. Here we present the results of a photoelectron spectroscopy study using table-top ultrafast deep and extreme UV lasers, based on filamentation four-wave mixing and high-harmonic generation, which enabled us to arrest the 2 1 A state. It is shown that the ring-opening reaction takes only 60 fs to complete, which is a considerably shorter time than previous experimental and theoretical estimates. The ballistic reaction creates vibrational coherence in the reaction products. cancel each other. The spectra overlain on the observed spectrum are the predicted spectra for CHD and HT reaction products at 2 ps after photoexcitation, and were obtained by semi-empirical CISD dynamical calculations. Static photoelectron spectra of cold CHD and HT( tZt ) are shown along with the band positions predicted by XMS-CASPT2 quantum chemical calculations for CHD and three HT isomers. Time profile of photoelectron intensities observed for eBE values of 8, 8.4 and 10 eV. Open symbols are and are The components were simulated population the reaction products. It is noted that the phase is shifted by π between the traces at and eV. The quantum beats are damped faster at 10 than eV. The trace at eV is for the photoelectron band center of the ground-state CHD, for which no oscillatory component is identified. Fourier transform.


Introduction
The photoinduced ring-opening reaction of 1,3-cyclohexadiene (CHD) to produce 1,3,5hexatriene (HT) is one of the best-known examples of electrocyclic chemical reactions, 1, 2 and it plays a central role in photobiological synthesis of vitamin D3 in the skin by facilitating photoisomerization of 7-dehydrochoestrerol to previtamin D3. 3 The ring-opening reaction of CHD follows the Woodward-Hoffmann rule, and C5-C6 bond rupture via the electronically excited state occurs with conrotatory motion of the end CH2 groups. 4 The reaction starts with a HOMO (12a)-LUMO (11b) transition in CHD induced by ultraviolet (UV) radiation, and the photoexcited 1 1 B state ultimately relaxes to the ground electronic state (1 1 A), 5 during which the system bifurcates into HT (a ring-opening reaction) and CHD (non-reactive internal conversion). However, it should be noted that the photoexcited 1 1 B state is not electronically correlated with the ground state of HT. Non-adiabatic transition(s) among the electronic states must be involved in order to facilitate the reaction. Van der Lugt and Oosterhoff performed a theoretical analysis of a related ring-closing reaction from butadiene to cyclobutene, and they pointed out that the reaction is mediated by a two-electron excited state. 6 The same mechanism is also expected here, as illustrated in Fig. 1. The potential energy surface of the 1 1 B state undergoes a conical intersection (CoIn) with that of the doubly excited 2 1 A state to induce a non-adiabatic transition, and the 2 1 A state mediates ultrafast internal conversion to the ground state to induce the ring-opening reaction. The essential feature of the electronic structure of this system is that the energy ordering of the HOMO and LUMO is reversed between the reactant and product, which places the 1 1 B, 2 1 A, and 1 1 A states in energetic proximity. Thus, it is the doubly excited 2 1 A state that plays a key role in the reaction; however, its spectroscopic detection has been difficult. (The C2 symmetry of CHD is lifted during the course of the reaction; however, we refer to the relevant electronic states and molecular orbitals using the symmetry notation under the C2 point group throughout this paper.) Ultrafast photoionization mass spectrometry by Fuß and coworkers reported decades ago 5,7,8 inspired a number of subsequent studies on this benchmark system. 5,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27 Fuß et al. employed strong field ionization of the transient species using 800 nm pulses, and suggested the wave packet propagation time on the 1 1 B surface, the time for transition from the 1 1 B to 2 1 A states, and the lifetime of the 2 1 A state to be 21, 35 and 80 fs, respectively. 7 Although the claims by Fuß et al. were intriguing, definitive spectroscopic assignments of the ionized species and electronic states were lacking in their mass spectrometric study, leaving a large degree of ambiguity in their arguments. The reaction has been studied more recently using photoelectron spectroscopy, 20,22,23,25 X-ray absorption spectroscopy 26 and electron diffraction; 27 however, the time resolution (120 -160 fs) of these experiments was insufficient for capturing the electronic dynamics in 3 real time. Meanwhile, computational studies provided a great deal of mechanistic insights into the reaction; however, the calculated reaction times were scattered, falling in the range of 70-280 fs.
The objective of this study is to experimentally identify non-adiabatic transitions via the doubly excited state of CHD and establish the mechanistic picture of this paradigmatic photoinduced ring-opening reaction. To this end, we employ ultrafast photoelectron spectroscopy using deep UV (DUV; 267.5 nm, 4.6 eV, 28 fs) pump and extreme UV (XUV; 57 nm, 21.7 eV, 43 fs) probe pulses. 28 Since XUV radiation ionizes all chemical species involved in the reaction, XUV photoelectron spectroscopy enables complete observation from the Franck-Condon region in the 1 1 B state of CHD until the final states of the reaction products. While XUV radiation produced by high-harmonic generation (HHG) 29,30,31,32 has been employed in ultrafast spectroscopy; 32,33,34 experiments with DUV pump pulses are scarce owing to the technical difficulties in generating ultrashort DUV pulses. 35 In the present study, we combined a filamentation four-wave mixing (FFWM) DUV light source 36 and a HHG XUV light source for the first time to achieve the time-resolution of 48 ± 2 fs.  Figure 2 shows photoelectron spectra measured as a function of the pump-probe delay time.

Results and Discussion
The pump pulses excite CHD to the 1 1 B state, and the probe pulses induce photoemission from all transient species in all electronic states. The spectra are plotted against the electron binding energy (eBE), given by the difference between the XUV photon energy and the measured electron kinetic energy (eKE). Since the XUV probe pulses cause photoemission from the ground electronic state of CHD for all delay times, its contribution has already been subtracted from the pump-probe spectra.
The three negative-intensity bands indicated in violet in Fig. 2 are the ground-state bleach (depopulation) induced by the pump pulses. We estimate that the ground-state bleach (the excitation efficiency) is 1.8 % under our experimental conditions. The positive signals at short pump-probe delay times (< 100 fs) are from the excited electronic state of CHD, while those at longer delay times are from reaction products. The ground-state bleach signal is never recovered, because the reaction products show different photoelectron spectra. In Fig. 2, it can be seen that the spectral features of the positive intensity bands, shown in green to red, rapidly change within 100 fs and become almost constant thereafter. This clearly indicates that the ring-opening reaction is completed in less than 100 fs. It can also be seen that the photoelectron signal intensities exhibit oscillations, most notably at around 8 and 10 eV. These are vibrational quantum beats of the reaction products observed for the first time for this reaction. which correspond to ionization to the 1 2 A [(10b) 2 (12a) 1 (11b) 0 ], 1 2 B [(10b) 1 (12a) 2 (11b) 0 ] and 2 2 B (or 6 3 2 B) [(10b) 2 (12a) 0 (11b) 1 ] cationic states. The 1 2 A←1 1 B band rapidly shifts from 3.75 to 5.25 eV within 30 fs, because the 1 1 B potential energy surface has a steep gradient for the C5-C6 stretching coordinate, while the 1 2 A cationic surface has a deep potential well. Therefore, eBE continuously increases as the vibrational wave packet propagates in the descending direction of the 1 1 B surface from the Franck-Condon region. The intensity of the b-band is initially weaker than that of the a-and c-bands owing to the electron configuration of 1 2 B [(10b) 1 (12a) 2 (11b) 0 ] unfavorable for one-photon transitions from the 1 1 B state. However, the b-band intensity grows with time and then exhibits a maximum, which is attributed to emergence of the 1 2 A← 2 1 A photoionization process. Quantum chemical calculations using XMS-CASPT2 (extended multi-state complete active space second-order perturbation theory) 37 estimate the eBE of the 2 1 A state in its equilibrium geometry to be 6.2 eV, which supports our assignment of the 1 2 A←2 1 A band.
We performed global fitting of the observed spectra to separate the contributions of different  fs is drastically shorter than the values so far reported. Fig. 3(D) illustrates the populations in the 1 1 B, 2 1 A and 1 1 A states without convoluting with the experimental time resolution. 7 Our model assumes that the population decay starts immediately after the system changes to the 2 1 A electronic state. In reality, the wave packet is expected to propagate on the 2 1 A surface for a finite time to arrive at the 2 1 A-1 1 A CoIn region and start decaying to the 1 1 A state. Therefore, it is likely that there is a time lag between the arrival time of the wave packet at the former and the latter CoIn. Thus, 32 fs is the upper bound for the arrival time of the wave packet at the first CoIn. and obtained the spectrum shown in Fig. 4(A). In this spectrum, the negative signal in the eBE region below 8 eV is primarily due to the excited state of CHD, whereas the positive signal at 8-10 eV is from the reaction products. Above 11 eV, ionization to a dense cationic excited state manifold occurs from both the excited states of CHD and the reaction products, and these negative and positive signals largely cancel each other. When we compare the observed reaction-product spectrum with those of cold CHD and HT, we find that the reaction product spectrum is more like the HT spectrum. However, it is noted that the reaction products are highly vibrationally excited, as the pump photon energy is entirely converted into the internal energy of the reaction products (4.6 eV for CHD and 3.5-3.8 eV for HT), and the energy cannot dissipate in the gas phase via intermolecular interactions. Thus, we performed semi-empirical dynamical calculations to predict the photoelectron spectra of vibrationally excited CHD and HT products at 1 ps after photoexcitation (See Supplementary Information for details). As shown in Fig. 4(A), the predicted spectra of CHD and HT are very similar to each other, and their differentiation in the photoelectron spectra is difficult. Thus, with our best efforts using ultrafast photoelectron spectroscopy, it is currently difficult to differentiate CHD and HT and estimate the reaction product branching ratio. Currently, the most common theoretical estimate for the CHD/HT branching ratio in the gas phase is 1:1, but this is yet to be experimentally verified.
As mentioned earlier, there are periodic intensity oscillations in the photoelectron spectra, which are ascribed to vibrational quantum beats in the reaction products. These oscillatory features are observed only in a narrow eBE window, where no strong signal occurs from cold CHD. It is noted that all transient signals arise from only 1.8 % of CHD molecules photoexcited to the 1 1 B state, and that a minute fluctuation of the photoelectron signal from unexcited cold CHD (98.2% remaining in the 9 ground state) obscures the weak quantum beat signals from the reaction products. Figure 4(B) shows photoelectron time profiles measured at an eBE of 8 and 10 eV, and the extracted oscillatory components. The two profiles show some interesting differences. One is that the quantum beat persists for a longer time period at 8 eV, and the other is that the phase of the oscillations is shifted by π between the two profiles. On the other hand, Fourier transforms of the oscillatory components shown in Fig.   4(C) indicate that the beat frequencies are similar for the two traces. In ultrafast spectroscopy, impulsive vibrational Raman scattering by pump pulses creates vibrational coherence in the ground state of the reactant, which must be differentiated from the coherence in the reaction products. In the present case, the following features suggest that the observed vibrational quantum beats are attributed to the reaction products. Firstly, the eBE region (for example, 9 -10 eV) where the beat appears is well separated from eBE of the ground-state CHD. Secondly, the beat intensity increases with increasing reaction product signal. And, finally the observed beat frequencies differ from the fundamental vibrational frequencies known for CHD. 38 Previously, vibrational coherence in the photochemical reaction product has been observed for the cis-trans photoisomerization of rhodopsin that occurs in 30 fs. 39 The phenomenon was indicated by the pioneering Raman spectroscopy studies by Mathies and coworkers 40,41,42,43 and unambiguously shown by the recent ultrafast transient grating spectroscopy study by Miller and coworkers. 39 The ringopening reaction of CHD has similar strong electronic coupling and an extremely short reaction time. 44 The 48-fs time resolution in our experiment provided a narrow frequency window for observation of vibrational coherence, but experiments with a higher time resolution will enable further examination of the vibrational dynamics in this paradigmatic reaction system.
In conclusion, we performed ultrafast photoelectron spectroscopy to examine a pericyclic ring-opening reaction of CHD using a deep UV ultrafast laser with filamentation four-wave mixing and an XUV ultrafast laser with high-harmonic generation. We unambiguously identified the formation and decay of the doubly excited 2 1 A electronic state that plays a key role in the reaction. The overall reaction time of 60 fs is strikingly shorter than previous estimates, and illustrates the ballistic nature 10 of this reaction. Vibrational wave packet motion was observed for the first time for the reaction products.

Methods
A one-box 1 kHz Ti:sapphire regenerative amplifier (Astrella, 35 fs, 800 nm, 1 kHz, 6 mJ) was used as a driving laser for non-linear optical processes to generate the DUV pump (267.5 nm) and XUV probe (57.1 nm) pulses. The XUV pulses were produced using high-harmonic generation in Kr gas with the second harmonic (2ω, 0.29 mJ) of the Ti:sapphire laser as a driving pulse. 28 The 2ω laser pulses were produced with a 0.3-mm thick beta barium borate (BBO) crystal, and focused into a Kr gas cell using a quartz lens (f = 500 mm). The 14th order (14ω: 57 nm, 21.7 eV) single harmonic was selected with a grating-based time-preserving monochromator. The DUV pump pulses were generated by a filamentation four-wave mixing light source in Ar gas using the ω (0.5 mJ) and 2ω (0.3 mJ) pulses. 36 The multi-color output from the filamentation cell was reflected and monochromatized using multilayer dielectric mirrors for 3ω in a vacuum chamber. The timing of the 3ω pulses was controlled 12 by a vacuum-compatible translation stage with 5-nm resolution. The pulse width and energy of the pump beam were sub-30 fs and 800 nJ/pulse, respectively. Since the entire optical path was kept under vacuum and no transmitting optical components were used, material dispersion was minimized. The cross correlation between the pump and probe pulses was determined to be 48 fs using non-resonant ionization of Xe.
The sample of 1,3-cyclohexadiene (CHD) was seeded by a He career gas and injected into a photoelectron spectrometer through a pinhole (⌀0.1 mm) with a stagnation pressure of 0.06 MPa at room temperature. The PKE distribution was measured using a magnetic bottle time-of-flight spectrometer. 45 The photoelectrons traveled through a 1300-mm-long flight tube and were detected using a microchannel plate detector (MCP, ⌀42 mm) mounted at the end of the flight tube. A retardation voltage of -6.0 V was applied to the flight tube in order to reject low-energy electrons produced by one-color two-photon ionization by the pump pulses. The measured PKE was calibrated using the photoelectron bands of 2 P3/2 and 2 P1/2 of Xe. The energy resolution was estimated to be 0.12 eV. The pressure in the photoionization chamber and TOF analyzer were 4.0×10 -5 and < 1.0×10 -7 Torr, respectively, during the experiment.