An accidental resonance mediated predissociation pathway of water molecules excited to the electronic [C with combining tilde] state

Zhigang He a, Dongyuan Yang ab, Zhichao Chen a, Kaijun Yuan a, Dongxu Dai ac, Guorong Wu *ac and Xueming Yang *ac
aState Key Laboratory of Molecular Reaction Dynamics, Dalian Institute of Chemical Physics, 457 Zhongshan Road, Dalian 116023, Liaoning, China. E-mail: wugr@dicp.ac.cn; xmyang@dicp.ac.cn
bUniversity of Chinese Academy of Sciences, Beijing 100049, China
cSynergetic Innovation Center of Quantum Information & Quantum Physics, University of Science and Technology of China, Hefei, Anhui 230026, China

Received 14th September 2017 , Accepted 18th October 2017

First published on 20th October 2017


The predissociation dynamics of water molecules in the electronic [C with combining tilde] state were studied using the time-resolved photoelectron imaging method. Both vibrationless and vibrationally excited states in the [C with combining tilde] electronic state were studied, with an emphasis on the vibrational excitation effects on the predissociation dynamics of the [C with combining tilde] state. Besides the well-known rotationally and non-rotationally mediated predissociation pathways (Proc. Natl. Acad. Sci. U. S. A., 2008, 105, 19148), an accidental resonance mediated predissociation pathway for the first bending mode excited state in the [C with combining tilde] electronic state of H2O is revealed, providing an excellent example of competition between non-adiabatic decay pathways involving at least five electronic states.


Introduction

Non-adiabatic processes occur throughout chemistry when electronically excited states are involved. These processes are key to our understanding of important processes in vision chemistry, atmospheric chemistry, interstellar chemistry, etc.1–3 Photochemistry of water in the gas phase has served as a model polyatomic system for the understanding of non-adiabatic dynamics induced by the coupling between different potential energy surfaces. In the last thirty years, numerous experimental and theoretical studies have been devoted to this system from which many details of photochemical and photophysical processes of water molecules in electronically excited states, especially the four lowest singlet states, have been revealed.4 Among these states, the [C with combining tilde] state is bound adiabatically with a fully resolved rotational structure, but subject to weak predissociation.5–10 Photodissociation of water in the [C with combining tilde] state shows rotational-state dependent predissociation dynamics: for 〈Ĵa2 = 0〉 rotational levels (〈Ĵa2〉 is the expectation value of the square of the rotational angular momentum about the molecular a-axis), the [C with combining tilde] state predissociates exclusively through the à state via electronic coupling and vibrationally hot and rotationally cold ground state OH(X2Π) products are produced; while for 〈Ĵa2 ≥ 1〉 rotational states, another channel opens through Coriolis-type coupling to the [B with combining tilde] surface, generating extremely rotationally hot and vibrationally cold ground state OH(X2Π) products, as well as rotationally hot electronically excited state OH(Ã2Σ+) products. Accordingly, the [C with combining tilde] state shows a lifetime with a clear dependence on 〈Ĵa2〉: with an increase of 〈Ĵa2〉, the lifetime of the [C with combining tilde] state rapidly decreases.

The relatively long lifetime and ro-vibrationally resolved absorption band of the [C with combining tilde] state provide a good opportunity for vibrational state-selective excitation of this state and therefore investigation of the vibrational excitation effects on the predissociation dynamics of water in the [C with combining tilde] state. However, this has not been studied as thoroughly to date. In this paper, we present an experimental study of the predissociation dynamics of water molecules in the [C with combining tilde] state using the time-resolved photoelectron imaging (TRPEI) technique. Both vibrationless and vibrationally excited states in the [C with combining tilde] electronic state of H2O and D2O are studied to investigate the vibrational excitation effects on the predissociation dynamics of the [C with combining tilde] state. All vibrationally excited states show a moderate enhancement on the predissociation rate of the [C with combining tilde] state, except for the first bending mode excited state in the [C with combining tilde] state of H2O which shows an “abnormal” high enhancement in the predissociation rate of the [C with combining tilde] state. Further analysis reveals that this is due to an accidental resonance mediated pathway through the [D with combining tilde] state, suggesting a new decay mechanism of the [C with combining tilde] state of H2O.

Experimental

The experiment was carried out on a velocity map imaging (VMI) spectrometer, constructed recently in our lab and detailed elsewhere.11 Four different pump wavelengths were used: 248.0 nm (∼3.5 μJ), 244.2 nm (∼3.1 μJ), 240.1 nm (∼2.5 μJ) and 239.0 nm (∼2.5 μJ). The bandwidth of the pump pulses was about 400 cm−1. The probe wavelength was fixed at 332.9 nm (6.2–6.7 μJ). Femtosecond laser pulses were obtained from a fully integrated Ti:Sapphire oscillator/regenerative amplifier system (<50 fs, 800 nm, 3.8 mJ and 1 kHz, Coherent, Libra-HE). Two optical parametric amplifiers (OPA, Coherent, OperA Solo), each pumped by a fraction of the fundamental (1.3 mJ per pulse), were used to generate the pump and probe pulses, respectively. UV pulses were combined collinearly on a dichroic mirror without further compression, and then focused using an f/75 lens to intersect a seeded water molecular beam in the interaction region of the VMI spectrometer. Time delays between pump and probe pulses were scanned using a computer-controlled stepper motor which was located upstream of one of the OPAs to change the delay of the input fundamental. The H2O and D2O molecular beams were generated by bubbling He of 2.5 bar through the H2O or D2O sample at room temperature using a pulsed valve (General Valve) operated at 50 Hz. The pump pulse excited the water molecules from their ground state into a vibrational level of the [C with combining tilde] state by two-photon absorption whereupon the delayed probe pulse produced photoelectrons via one-photon ionization (Fig. 1). Photoelectron images arising from the pump or the probe laser alone were also recorded. The sum of the single color photoelectron images were subtracted in order to correct for background photoelectrons generated from single color multiphoton ionizations. Pump–probe time delays were scanned back and forth multiple times to minimize any small hysteresis effects, and effects caused by the fluctuations and drifts in the laser pulse energies, pointing and molecular beam intensity, etc. The 2D photoelectron images were transferred to the 3D distribution using the basis-set expansion method.12 The time-dependent photoelectron 3D distributions were further integrated along the recoiling angle to derive the kinetic energy distributions of the photoelectron, i.e., time-resolved photoelectron spectra (TRPES). Electron kinetic energy calibration was achieved using multi-photon ionization of Xe atoms. This also served to measure the cross-correlation (i.e., instrumental response function) between the pump and the probe laser pulses. The time-zero overlap was checked before and after the TRPEI measurement of the water molecules. The delay-dependent curve of the electron yield was fitted, based on the approximation that both pump and probe laser pulses have a Gaussian profile. It has to be noted here that only photoelectrons generated by absorption of two pump photons and a single probe photon were taken into account in this analysis, thanks to the capability of the VMI technique in differentially analyzing the outgoing photoelectron as a function of delay with respect to kinetic energy. The derived cross-correlations are summarized in Table 1. The error bars of the cross-correlation represent one standard deviation derived from the fit of the total electron yield to a Gaussian distribution.
image file: c7cp06286a-f1.tif
Fig. 1 Schematic view of the pump and probe schemes used in the present experiment.
Table 1 2D globally fitted time constants from the TRPES data. The pump laser wavelengths used, the corresponding vibronic states excited,5,13–16 and the corresponding cross-correlations are also indicated
λ (nm) H2O D2O Cross-correlation (fs)
Vibronic state Lifetime Vibronic state Lifetime (ps)
248.0 [C with combining tilde](000) 1.42 ± 0.15 ps [C with combining tilde](000) 5.76 ± 0.35
244.2 [C with combining tilde](010) 327+23−47 fs [C with combining tilde](010) 4.0 ± 0.4 173 ± 7 (H2O)
155 ± 4 (D2O)
240.1 [C with combining tilde](100) 3.6 ± 0.1 149 ± 14
239.0 [C with combining tilde](100) 643+160−183 fs 153 ± 13


Results and discussion

The experimental results and data analysis

Water molecules were excited into the [C with combining tilde] state by two-photon absorption whereupon another delayed laser pulse probed the excited state water molecules via one-photon ionization (Fig. 1). By tuning the pump wavelength, three different vibrational states of the [C with combining tilde] electronic state, [C with combining tilde](000), [C with combining tilde](010) and [C with combining tilde](100), were excited. (Vibronic states are denoted by (ν1ν2ν3), where ν1, ν2 and ν3 specify the quantum numbers in the symmetric stretching, bending and asymmetric stretching modes, respectively.) Fig. 2(a)–(c) show the TRPES data of H2O at pump wavelengths of 248.0, 244.2 and 239.0 nm, respectively. At 248.0 nm, the TRPES data is dominated by a single sharp peak centered at 1.1 eV. The delay dependence of this peak seems to show a simple exponential decay with a lifetime of the order of magnitude of 1 ps. At 244.2 nm, the TRPES data is composed of a strong, sharp peak at around 1.25 eV showing a very fast decay and a weak feature peaked at 1.1 eV with a relatively slow decay dynamics. A feature at 0.3 eV is also presented in this TRPES data and that at 239.0 nm and some of those for D2O (see Fig. 3(b) and (c)). During the time delay implemented in these measurements, no decay of this feature was observed. The TRPES data at 239.0 nm is similar to that at 244.2 nm, except that the fast decaying component becomes very diffuse over the kinetic energy range from 0.2 to 1.5 eV.
image file: c7cp06286a-f2.tif
Fig. 2 TRPES of H2O at pump wavelengths of 248.0 (a), 244.2 (b) and 239.0 nm (c), after subtracting background photoelectrons generated from single color multiphoton ionization. Note that a combination of linear and logarithmic scales is used in the ordinate. (d–f) The photoelectron kinetic energy distributions derived from a 2D global least-squares fit to the data shown in (a–c), respectively. The energetic limit to the ground state of the cation is indicated by the black arrow. (g–i) The delay dependent photoelectron signal integrated over the kinetic energy range 1.0–1.2 eV where the main feature of the [C with combining tilde] state is located. The transients from the 2D global least-squares fit are also included.

image file: c7cp06286a-f3.tif
Fig. 3 Same as Fig. 2, but for D2O at pump wavelengths of 248.0 (a), 244.2 (b) and 240.1 nm (c).

In order to extract more detailed information from these TRPES data, a 2D global least-squares method was employed to simultaneously fit data at all time delays and photoelectron kinetic energies. Only photoelectrons with kinetic energies ≥0.5 eV were included in order to avoid the interference from the 0.3 eV feature which was found to be associated with the H atom (detailed in the ESI). The kinetic model used to fit the TRPES data is the same as that used in our previous studies and will not be reiterated here.17 Cross-correlation functions between two photons of the pump pulse and one photon of the probe pulse, measured independently, were invoked in the fit except that at 248.0 nm. At 248.0 nm, resonance-enhancement in the multiphoton ionization of Xe was observed which prevented a reliable determination of the cross-correlation between 2 × 248.0 and 332.9 nm. Therefore, the cross-correlation function at 244.2 nm was used in the fit of the TRPES data at 248.0 nm and it was varied in a small range to find a best fit (minimum χ2 value). The photoelectron kinetic energy distribution(s) of component(s) involved at each pump wavelength, derived from the fit, are shown in Fig. 2(d)–(f). The energetic limit for ionization to the ground state of the cation (D0), calculated using the previously reported ionization potential of 12.62 eV,18 is indicated with a black arrow. The derived time constants for the [C with combining tilde] electronic state are summarized in Table 1. The photoelectron transients over the kinetic energy range 1.0–1.2 eV where the main feature of the [C with combining tilde] state is located (vide infra) are shown in Fig. 2(g)–(i), together with the curves from the 2D global fit. In the data analysis, time-zeros were varied in the range of the time-zero drift experimentally measured and the cross-correlations were also varied in the range of their uncertainties to obtain a best fit. This analysis also served to estimate the confidence intervals of these time constants.

At 248.0 nm, it is straight forward to assign the strong, single peak feature to the [C with combining tilde](000) vibronic state, with a lifetime of 1.42 ± 0.15 ps. The [C with combining tilde] state is a Rydberg state with an electron from the highest occupied molecular orbital (HOMO) excited to the 3pa1 Rydberg orbital, and as such it may be expected to possess a minimum energy geometry very close to that of the D0 state of the cation, resulting in the photoionization process being dominated by diagonal Franck–Condon (FC) factors (Δν = 0). In the present case, the D0(000) vibronic state of the cation is expected which is consistent with the observations in the photoelectron kinetic energy distribution: a single peak located at the energetic limit. At 244.2 nm, there are two components involved, with a lifetime of 327+23−47 and 52+12−6 fs, respectively. The fit also suggests that these two components have a parallel relationship, i.e., these two components are excited simultaneously and decay independently. According to the assignment in the absorption spectrum, there are two vibronic states covered by the bandwidth of the 244.2 nm pump pulse (∼400 cm−1), [D with combining tilde](000) and [C with combining tilde](010), with the former having a much larger absorption cross section than the latter. Similar to the [C with combining tilde](000) TRPES data, the photoelectron from the [D with combining tilde](000) vibronic state should also correspond to a single peak at the energetic limit (D is also a Rydberg state), i.e., the 52 fs component corresponds to the [D with combining tilde](000) vibronic state. In the photoionization of the [C with combining tilde](010) vibronic state, the ν2 = 1 vibrational excitation will be conserved in the cation, leading to the formation of the D0(010) vibronic state of the cation, due to the Δν = 0 FC propensity in the photoionization step. Therefore, the photoelectrons from the [C with combining tilde](010) vibronic state will have the same kinetic energy as that from the [C with combining tilde](000) vibronic state when the same probe wavelength is used. This is indeed what was observed in the experiment (Fig. S1, ESI), confirming that the other component corresponds to the [C with combining tilde](010) vibronic state, with a lifetime of 327+23−47 fs. There is a negative dip in the kinetic energy distribution of the photoelectron of the [C with combining tilde](010) state. This is an artificial structure from the 2D global fit, due to the much strong neighboring peak in the photoelectron kinetic energy distribution of the [D with combining tilde](000) state. As for TRPES data at 239.0 nm, similar analysis was performed and the 643+160−183 fs component is attributed to the [C with combining tilde](100) vibronic state. According to the absorption spectrum, the component with a lifetime less than 50 fs should be the [D with combining tilde](010) vibronic state. However, the corresponding photoelectron kinetic energy distribution is very diffuse, covering the whole range from 1.5 down to 0.2 eV, inconsistent with the expectation of the photoionization of the [D with combining tilde] Rydberg state. Very likely, this is due to contamination from the non-resonant two-color multiphoton ionization. The very weak pump–probe signal and the short lifetime of the [D with combining tilde](010) state make it indistinguishable from the non-resonant background (it is therefore labelled as τnon in Fig. 2(f)). The TRPES data for the D2O molecule, shown in Fig. 3(a–c), are fitted and analyzed in an analogous manner and lifetimes of 5.76 ± 0.35, 4.0 ± 0.4 and 3.6 ± 0.1 ps are derived for the [C with combining tilde](000), [C with combining tilde](010) and [C with combining tilde](100) states, respectively.

The predissociation dynamics of water excited to the [C with combining tilde] state and the vibrational excitation effects

The lifetimes of the [C with combining tilde](000) state derived in the current study are in reasonable agreement with the previous studies, but obvious differences also exist. By measuring the line width in the 3 + 1 REMPI spectra, Ashfold et al. estimated the upper limit for the lifetime of the [C with combining tilde] state.5 They obtained 2.5 ps for H2O and 6 ps for D2O at the lowest rotational quantum number 〈Ĵa2 = 0〉. In a time resolved study, Radloff and co-workers directly measured the lifetime of the [C with combining tilde] state, leading to 0.5 ± 0.1 and 1.2 ± 0.1 ps for H2O and D2O, respectively.8 The lifetimes derived from the current study are between the results from those two previous studies for both H2O and D2O. This might be due to the different rotational temperatures of the water samples used in these experiments. It was concluded that the lifetime of the [C with combining tilde] state decreases rapidly with increasing rotational excitation of the water molecules, scaled with 〈Ĵa2〉. The experiment of Radloff and co-workers was performed with water vapor at room temperature, and therefore a relatively high average 〈Ĵa2〉 and a smaller (state averaged) lifetime are expected for both H2O and D2O. The current experiment was performed with a molecule beam, and therefore only a very small, but non-zero 〈Ĵa2〉 is anticipated. Therefore, the lifetimes from the current experiment are larger than those from Radloff and co-workers, but smaller than ones for 〈Ĵa2 = 0〉 derived by Ashfold et al. As such, the observed difference in the lifetimes of the [C with combining tilde](000) state for both H2O and D2O is due to the temperature difference of the water samples used in three experiments which in turn confirms the previous conclusion: the lifetime of the [C with combining tilde](000) state decreases rapidly with the increase of 〈Ĵa2〉.

Besides the vibrational ground state, bending and symmetric stretching mode excited states of the [C with combining tilde] electronic state were also accessed, enabling an investigation of vibrational excitation effects on the predissociation dynamics of the [C with combining tilde] state. From Table 1, the vibrationally excited states all show faster predissociation. This might be due to the following two factors: (1) energetic factor: The vibrational excited states locate energetically higher than the vibrational ground state and higher densities of state of the à and [B with combining tilde] states are expected, resulting in faster predissociation rates (Fermi Golden Rules); (2) FC factor: Compared to the vibrational ground state, the structured wavefunction of the vibrationally excited states on the bound [C with combining tilde] state potential energy surface (PES) might have a larger overlap integral with the outgoing wavefunction on the à or [B with combining tilde] state PES, which is highly oscillatory, resulting in a faster predissociation. In particular, the bending mode excited wavefunction should have a larger overlap with the outgoing wavefunction on the [B with combining tilde] state PES which contains extensive nodes along the bending angle as shown by Dixon and coworkers.9 Therefore, the [C with combining tilde](010) state should be more efficiently predissociated by the B state. As for the à state, the outgoing wavefunction is highly oscillatory along the symmetric stretching coordinate and the [C with combining tilde](100) state should have a larger FC overlap with and is more easily predissociated by the à state. Both energetic and FC factors might play a role in promoting the predissociation of the [C with combining tilde] state. It is hard to separate the contributions of these factors solely from the current experimental data. High level theoretical work is definitely needed to examine these proposals and provide further details into vibrational excitation effects on the predissociation dynamics of the [C with combining tilde] state of water.

From Table 1, the most remarkable observation is that the bending mode excitation of H2O shows an “abnormal” enhancement in the predissociation rate of the [C with combining tilde] state by a factor of about 4.3, with an additional vibrational energy of about 1400 cm−1. The symmetric stretching excitation enhances the predissociation rate only by a factor of 2.2 with more than two times additional vibrational energy. While for D2O, only a factor of 1.4 is observed in the predissociation rate enhancement in the [C with combining tilde] state by the bending mode excitation. Therefore, it seems hard to rationalize the enhancement of the predissociation rate by the bending mode excitation in H2O by only considering the two factors mentioned above and a third factor should play a role. A closer look at the vibronic levels of the [C with combining tilde] state and the neighboring electronic states reveals that, as suggested by Ashfold et al. in their REMPI spectroscopy studies,5,6 the state mixing between different vibronic levels has to be invoked. As shown in Fig. 4, the [C with combining tilde](010) and [D with combining tilde](000) vibronic states are isoenergetic and a strong wavefunction mixing is expected.15 The latter is strongly predissociated by the [B with combining tilde] state, with a lifetime of 52 fs (vide supra). This wavefunction mixing results in an additional predissociation pathway opening for the [C with combining tilde](010) state, besides the two well-known predissociation pathways via the à or [B with combining tilde] state: the [C with combining tilde](010) state predissociates through the [D with combining tilde] state, followed by a fast decay of the latter to the [B with combining tilde] state. The [D with combining tilde] and [B with combining tilde] states are strongly coupled at bent geometries.19,20 The initial excitation in the bending mode of the [C with combining tilde](010) state might also facilitate the decay of the wavepacket on the [D with combining tilde] state PES to the [B with combining tilde] state. From the dramatic increase of the predissociation rate, the new predissociation pathway should play a dominant role in predissociation of the [C with combining tilde](010) state. Accordingly, the decay mechanism of the [C with combining tilde](010) state of H2O is proposed as:

image file: c7cp06286a-t1.tif

image file: c7cp06286a-t2.tif

image file: c7cp06286a-t3.tif

image file: c7cp06286a-t4.tif
As such, the predissociation of H2O in the [C with combining tilde](010) state provides an excellent example of competition between different non-adiabatic decay pathways with at least five electronic states involved. As for the [C with combining tilde](010) state of D2O, there is a much larger energy gap (282 cm−1)15 with the [D with combining tilde](000) state and therefore a negligible mixing between these two states exists and the predissociation of the [C with combining tilde](010) state of D2O is barely affected.


image file: c7cp06286a-f4.tif
Fig. 4 Schematic view of the energy levels of water molecules for the related electronic states and the predissociation pathways of the [C with combining tilde] state. The inset panels show the lifetimes of the [C with combining tilde] state with different vibrational excitation.

Conclusions

In summary, the predissociation of water molecules in the electronic [C with combining tilde] state was studied using the time-resolved photoelectron imaging method. The [C with combining tilde](000), [C with combining tilde](010) and [C with combining tilde](100) states of H2O and D2O were excited by two-photon absorption in the wavelength range of 248.0–239.0 nm. All these states were unambiguously identified in the time-resolved photoelectron spectra data and their lifetimes were measured. The derived lifetimes of the [C with combining tilde](000) state for both H2O and D2O are in reasonable agreement with previous experimental results, but with certain difference due to the different temperatures of the water samples used in various experiments. Finally, the vibrational excitation effects on the predissociation dynamics of water are discussed in detail. Vibrationally excited states all show faster predissociation which is attributed to the energetic and Franck–Condon factors. The bending mode excitation of H2O shows an “abnormal” high enhancement in the predissociation rate with a factor of 4.3 which is rationalized by invoking an additional accidental resonance mediated pathway, highlighting a new decay mechanism of H2O in the [C with combining tilde] state.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (Grant No. 21573228), the Ministry of Science and Technology of China (2012YQ12004704) and the National Basic Research Program of China (973 Program, 2013CB922200). We would also like to thank Prof. Michael N. R. Ashfold for very helpful discussions.

References

  1. R. Schinke, Photodissociation dynamics, Cambridge University Press, Cambridge, 1993 Search PubMed.
  2. B. P. Bonev, M. J. Mumma, M. A. DiSanti, N. Dello Russo, K. Magee-Sauer, R. S. Ellis and D. P. Stark, Astrophys. J., 2006, 653, 774 CrossRef CAS.
  3. B. P. Bonev and M. J. Mumma, Astrophys. J., 2006, 653, 788 CrossRef CAS.
  4. K. Yuan, R. N. Dixon and X. Yang, Acc. Chem. Res., 2011, 44, 369 CrossRef CAS PubMed and references therein.
  5. M. N. R. Ashfold, J. M. Bayley and R. N. Dixon, Chem. Phys., 1984, 84, 35 CrossRef CAS.
  6. M. N. R. Ashfold, J. M. Bayley, R. N. Dixon and J. D. Prince, Ber. Bunsenges. Phys. Chem., 1985, 89, 254 CrossRef CAS.
  7. G. Meijer, J. J. Termeulen, P. Andresen and A. Bath, J. Chem. Phys., 1986, 85, 6914 CrossRef CAS.
  8. O. Steinkellner, F. Noack, H.-H. Ritze, W. Radloff and I. V. Hertel, J. Chem. Phys., 2004, 121, 1765 CrossRef CAS PubMed.
  9. K. Yuan, Y. Cheng, L. Cheng, Q. Guo, D. Dai, X. Wang, X. Yang and R. N. Dixon, Proc. Natl. Acad. Sci. U. S. A., 2008, 105, 19148 CrossRef CAS PubMed.
  10. K. J. Yuan, Y. A. Cheng, L. N. Cheng, Q. Guo, D. X. Dai, X. M. Yang and R. N. Dixon, J. Chem. Phys., 2010, 133, 134301 CrossRef PubMed.
  11. Z. He, Z. Chen, D. Yang, D. Dai, G. Wu and X. Yang, Chin. J. Chem. Phys., 2017, 30, 247 CrossRef CAS.
  12. V. Dribinski, A. Ossadtchi, V. A. Mandelshtam and H. Reisler, Rev. Sci. Instrum., 2002, 73, 2634 CrossRef CAS.
  13. J. H. Fillion, R. van Harrevelt, J. Ruiz, N. Castillejo, A. H. Zanganeh, J. L. Lemaire, M. C. van Hemert and F. Rostas, J. Phys. Chem. A, 2001, 105, 11414 CrossRef CAS.
  14. J. W. C. Johns, Can. J. Phys., 1963, 41, 209 CrossRef CAS.
  15. S. Bell, J. Mol. Spectrosc., 1965, 16, 205 CrossRef CAS.
  16. P. Gurtler, V. Saile and E. E. Koch, Chem. Phys. Lett., 1977, 51, 386 CrossRef.
  17. G. R. Wu, A. E. Boguslavskiy, O. Schalk, M. S. Schuurman and A. Stolow, J. Chem. Phys., 2011, 135, 164309 CrossRef PubMed.
  18. R. H. Page, R. J. Larkin, Y. R. Shen and Y. T. Lee, J. Chem. Phys., 1988, 88, 2249 CrossRef CAS.
  19. D. M. Hirst and M. S. Child, Mol. Phys., 1992, 77, 463 CrossRef CAS.
  20. R. van Harrevelt and M. C. van Hemert, J. Chem. Phys., 2000, 112, 5787 CrossRef CAS.

Footnotes

Electronic supplementary information (ESI) available: Analysis of the 0.3 eV feature in the TRPES data and the comparison of the photoelectron kinetic energy distributions of the [C with combining tilde] state with different vibrational excitations. See DOI: 10.1039/c7cp06286a
These authors contributed equally.

This journal is © the Owner Societies 2017