Conformer-Specific Photoelectron Spectroscopy of Carbonic Acid: H2CO3

Carbonic acid (H2CO3) is a fundamental species in biological, ecological, and astronomical systems. However, its spectroscopic characterization is incomplete because of its reactive nature. The photoionization (PI) and the photoion mass-selected threshold photoelectron (ms-TPE) spectra of H2CO3 were obtained by utilizing vacuum ultraviolet (VUV) synchrotron radiation and double imaging photoelectron photoion coincidence spectroscopy. Two carbonic acid conformers, namely, cis–cis and cis–trans, were identified. Experimental adiabatic ionization energies (AIEs) of cis–cis and cis–trans H2CO3 were determined to be 11.27 ± 0.02 and 11.18 ± 0.03 eV, and the cation enthalpies of formation could be derived as ΔfH°0K = 485 ± 2 and 482 ± 3 kJ mol–1, respectively. The cis–cis conformer shows intense peaks in the ms-TPES that are assigned to the C=O/C–OH stretching mode, while the cis–trans conformer exhibits a long progression to which two C=O/C–OH stretching modes contribute. The TPE spectra allow for the sensitive and conformer-selective detection of carbonic acid in terrestrial experiments to better understand astrochemical reactions.

S 2

1/ H2CO3 production
We attempted to produce H2CO3 (1) by heating of NH4HCO3 25 and di-tert-butyl carbonate (2) 23 as precursors.In addition to 2, diethyl carbonate (3) was also used as a precursor to produce 1 via twofold ethylene (C2H4) loss as suggested by Bucher et al. 44 However, 1 was not sufficiently produced either with the method using NH4HCO3, similarly to the results of Reisenauer et al., 23 or by the flash pyrolysis of 3, the latter being likely due to a competitive and more preferable reaction channel producing CO2 and ethanol over a barrier 20-25 kJ mol −1 below that of 1 formation. 44,45The results shown in the present study are therefore based on flash pyrolysis of 2. shows molecular beam (MB) as well as room temperature background (RT-BG) ions.The ms-TPES of the photoions at m/z 62 (b), obtained by integrating the MB (blue lines, lower trace), shows a broad and unstructured band, indicative for extensive hot-and sequence band transitions and a rovibrational temperature close to the one of the reactor. 62The room temperature (RT-BG, red lines b)), on the other hand, shows a clearly pronounced vibrational structure, as the molecules collide with the ion optics and chamber wall to efficiently rethermalize after only a few collisions to a room temperature internal energy distribution.The neutrals diffuse back into the ionization region and contribute to the RT-BG component.This leads to less hot and sequence band transitions being excited upon photoionization and to well resolved peaks in the ms-TPES, justifying a Franck-Condon simulation at 300 K. c) ms-TPES of the dissociative ionization products at m/z 59 and 60, taken under the same conditions, do not show any overlapping features with the spectrum of m/z 62.
To further test the sturdiness of our fit, the 0-0 transition of the 1ct FC simulation was set to 11.21 eV (a) and 11.24 eV (b), respectively.Together with the fit at 11.18 eV (Figure S3c), the three simulations span a range of 11.21 ± 0.03 eV around the composite method calculations (Table S1).From the three simulations the ones at 11.21 and 11.24 eV led to inferior fits, providing further evidence that the AIE of 1ct is 11.18 eV.
11.21 eV a) Figure S5.Franck-Condon (FC) simulations at different levels of theory.The weights of 1cc and 1ct in the weighted sum lines (magenta) are fixed in all cases.The B3LYP result (a) shows generally good agreement with the ms-TPES.The MP2 (d) and CCSD (e) results reproduce the band at 11.39 eV well but overestimate the intensity of the main peaks, e.g., the peak at around 11.64 eV in the ms-TPES.The slight blue-shift in the FC simulation at the higher photon energies, e.g., the peak at around 11.81 eV, is due to anharmonicity at high vibrational excitation being disregarded in the double harmonic approximation.

Figure S1 .
Figure S1.(left) Ion velocity map images (VMI) of m/z 62 and 59, which indicate that the m/z 62 signal results from direct photoionization of pyrolysis products and is not affected by dissociative photoionization (DPI) signal of the precursor 2 or any other pyrolysis intermediate.The VMI of m/z 59, on the other hand, shows broadening perpendicular to the MB axis, indicative of kinetic energy release in the dissociative ionization of the precursor 2. (right) Mass spectrum of di-tert-butyl carbonate (2) pyrolysis at 790 K diluted with helium at photon energy of 11.5 eV.

Figure S3 .
Figure S3.Residuals of the 1cc and 1cc+1ct fit (a).The comparison of the FC simulation (300 K) of 1cc with the experimental spectrum shows that the bands at 11.39 and 11.55-11.64eV are underestimated (a and b).The addition of the 1ct simulation (c) leads to a better agreement of the model with the experimental spectrum (R 2 0.123 vs. 0.289 / lower residuals a).Especially the peak at 11.39 eV is much better represented by the model, which is also closer to the uncertainty bars of the experiment.Experimental error bars are defined as ± √ ! .

Table S2 .
Unscaled vibrational frequencies in cm -1 of neutrals, cations of both 1cc and 1ct conformers at the B3LYP/6-311++G(d,p) level of theory.Mulliken numbering of the vibrational modes is given in parentheses.Franck-Condon active modes are highlighted in boldface.

Table S3 .
Bond lengths and angles of 1cc and 1ct calculated at the B3LYP/6-311++G(d,p) level of theory.See Figure3in the manuscript for the atom numbering.