Communication: The ionization spectroscopy of mixed carboxylic acid dimers

Intra- and intermolecular hydrogen bonding interactions in water and carboxylic acids are of continuing interest to the scientific community.¹ Exemplary cases include the intramolecular H bond as in glycolic acid,² and the intermolecular H bond(s), as in water dimer³ or formic acid dimer.⁴ The complex of glycolic acid with formic acid has both intra- and intermolecular H bonds.⁵ Details of the H bonding can be revealed by vibrational spectroscopy, accomplished either by irradiation in the infrared region or by Raman scattering of light in the visible range. Vibrational frequencies are measured as well from IR-UV double resonance experiments. In principle these techniques could directly measure the frequency changes in, e.g., stretches of hydroxyl OH groups, attending changes in chemical environment. But in the carboxylic acid dimers, due to the substantial mode couplings arising from the double intermolecular H bonds, the OH stretches become very broad spanning the range from 2800 to 3200 cm⁻¹.⁶ This conceals valuable information. Quantum state resolved photoionization spectroscopies, e.g., resonance enhanced multiphotonionization (REMPI) spectroscopy and mass analyzed threshold ionization (MATI) spectroscopy can avoid this problem. The MATI spectroscopy has been applied to a wide range of systems from atoms to small molecules to polyatomic molecules to clusters of many types to study molecular spectroscopy and chemical dynamics.⁷ One would hope that such techniques could yield information on strength and directionality of H bonds, and the effect on H bonding of ionization. We have studied a number of fluorene derivatives and their H-bonding complexes and carboxylic acids and water complexes using spectroscopy such as REMPI, MATI, and IR-UV double resonance. Unfortunately, in many H bonded clusters ionization forces large geometry changes. For example, consider 9-fluorene acetic acid (9FCA) and its complex with water 9FCA-H₂O. This system and similar species display very poorly defined zero kinetic energy (ZEKE) photoelectron spectra. We attribute this to the substantial geometric reorganization induced by ionization. In Figure 1 B3LYP/6-311++G(d,p) modeling (all calculations were performed in Gaussian 09⁸) shows that upon ionization the torsional dihedral angle between the 9 position's CH bond and the carboxyl C=O bond changes from −25° to 119°. A number of similar monomer species that display extensive geometric changes upon ionization are shown in Figure 1. In contrast we observe well-defined MATI spectra for 9-hydroxy-9-fluorenecarboxylic acid (9HFCA) and more interestingly on its complexes with formic acid, acetic acid, and benzoic acid. Modeling shows that 9HFCA maintains C_s symmetry. That is, the torsional dihedral angle between the 9 position's OH bond and the carboxyl C=O bond remains zero upon ionization, which we may attribute to the effect of internal H-bonding between the 9-position's –OH and the 9HFCA carbonyl oxygen.

The molecular beam apparatus has been documented in previous publications.⁹ The 9HFCA and complexes were prepared in a pulsed supersonic expansion using 1 bar helium as a carrier gas. The 9HFCA (Aldrich) was placed in the body of the silicone coated pulsed valve (General Valve, series 9) which was heated to 120 °C to provide sufficient vapor pressure without any decomposition. The solvents were either seeded in the helium via a variable flow controller or mixed with the 9HFCA inside the valve. The pulsed beam (20 Hz) was skimmed and the REMPI and MATI spectroscopy was performed downstream in a second chamber at a distance of 10 cm from the nozzle. Two independently tunable UV laser pulses were used in the experiments. One laser is a pulsed nanosecond Nd:YAG operating at 20 Hz (Continuum NY-61) pumping a tunable dye laser (Lumonics HD500) with a visible bandwidth of 0.05 cm⁻¹. The second laser system is also a Nd:YAG (Quanta-Ray GCR-3) pumping a tunable dye laser (Quanta-Ray PDL). Both of the dye lasers are equipped with frequency doubling capabilities (Spectra-Physics WEX). These two lasers were focused using 1 m lens and spatially and temporally overlapped in the interaction region, and one functions as the pump laser, and the other as the probe. The MATI experiments were performed with pumping the S₁ origin band and scanning the probe toward the ionization potential (IP). Briefly, the molecules were excited in a field free environment and after ∼0.1 μs relative to the laser firing, a small pulse of +0.5 V/cm was applied to a grid in the interaction region for a duration of 20 μs to remove prompt ions away from the interaction region. A pair of pulsed high voltages were then applied to field ionize the surviving high n Rydberg states and push the resulting ions toward the detector.

The MATI ionization spectra in Figure 2 for the monomer and its formic, acetic, and benzoic acids clusters were plotted as a function of the total ionization energy. The lowest energy peak in each spectrum is considered the adiabatic IP. The IP for 9HFCA is 64 920 ± 8 cm⁻¹(32 767.8 + 32 152.0). In the formic, acetic, and benzoic acid clusters the well-resolved cation spectra allow accurate determination of the adiabatic IPs as 64 375 ± 8 cm⁻¹ (32 740.3 + 31 634.2), 63 790 ± 8 cm⁻¹ (32 758.3 + 31 031.6), and 63 485 ± 8 cm⁻¹ (32 757.6 + 30 727.0), respectively. In particular the IPs for the carboxylic acids are redshifted over a relatively small range, from 545 cm⁻¹ to 1435 cm⁻¹. For 9HFCA monomer we see a well-defined vibrational progression with interval 58 cm⁻¹ and a smaller feature at 88 cm⁻¹. All clusters show similar vibrational progressions of 44, 40, and 34 cm⁻¹, respectively. This significant low frequency progression in the threshold region could have an additional member to the red which is not observed due to very poor Franck-Condon factors. However, by comparison to the monomer, and given the good signal to noise, particularly for the formic and acetic acid dimers, we expect that all members have been observed. For the benzoic acid cluster it may not be as definitive but this would not have a major impact on the general conclusions of this work. As energy increases in each cluster, higher frequency vibrational modes are excited; the spectrum becomes so complex that individual transitions are not distinguishable. The measured small IP redshift of 545 cm⁻¹ from monomer to its formic acid dimer is particularly interesting compared to other systems such as the single H bonded phenol-H₂O and indole-H₂O clusters in which the IP redshifts are several thousands of wavenumbers.¹⁰ Our results show that the MATI spectroscopy is well suited to study certain kinds of multiple H bonded clusters.

These complexes possess two intermolecular H bonds in addition to the intra H bond already present in the monomer. In general complex H-bonding geometries are easily sabotaged and reformed upon ionization, so that the cation spectra are often unobservable due to extremely poor Franck-Condon overlap. The clear MATI spectra of 9HFCA-carboxylic acid complexes shown in Figure 2 strongly imply that the network of multiple H bonds in the cation retains the topology of the neutral complex. This suggestion is confirmed by density functional theory (DFT) based calculations. With B3LYP/6-311++G(d,p) modeling, we find that the covalent bond lengths in the H-bonding region change by no more than 0.05 Å; the weaker and more sensitive H⋯O lengths can change by as much as 0.2 Å, but are never broken. More significantly, the C_s symmetry of the neutral complex is retained in the cation, as verified by vibrational analysis. Alternative and thoroughly tested modern methods, M06-2X and wB97XD, agree.

According to B3LYP and M06-2X calculations ionization in the 9HFCA system has the effect of weakening the internal H-bonding slightly, but is primarily expressed in bond length adjustments in the fluorene fragment. The form of the HOMO in the neutral system reflects the bond length changes; the largest effect is a shortening of the single bond linking the phenyl fragments of fluorene from 1.472 to 1.427 Å. Ionization has a more dramatic impact on the geometry of the H-bonds formed in complexation. In every case, the ionization strengthens and shortens the H-bond in which 9HFCA is the H-donor, and weakens and lengthens the H-bond for which 9HFCA is the H-acceptor. In the neutral 9HFCA-formic acid cluster the calculated distances for the intermolecular H bonds (9HFCA donor and 9HFCA acceptor) are 1.688 and 1.704 Å, respectively. In the cation the corresponding values become 1.562 and 1.832 Å. The weaker internal H bond is almost unchanged, as 2.004 and 2.021 Å for neutral and cation.

Geometry optimizations with vibrational calculations permit identification of the vibrational modes excited as the ionized molecule relaxes toward its own equilibrium geometry. As noted above, ionization strengthens and shortens the H-bond in which 9HFCA is the H-donor, and weakens and lengthens the H-bond for which 9HFCA is the H-acceptor. Relaxation naturally follows the rocking motion of the H-bonding partner relative to 9HFCA. The decline in vibrational frequency in the series formic acid, acetic acid, and benzoic acid follows the trend suggested by the reduced mass of this mode. One does not expect quantitative agreement between the harmonic frequencies produced by DFT calculations and the anharmonic frequencies for the cation, but the computed values for the rock, 43, 38, and 35 cm⁻¹, are not in serious disagreement with the observed values 44, 40, and 34 cm⁻¹. In the monomer cation, we do find computed frequencies for the carboxyl rock at 58 cm⁻¹ and the fluorene butterfly wingflap at 94 cm⁻¹, consistent with our assignments of the observed 58 and 88 cm⁻¹ features of the vibrational progression.

DFT modeling can deepen our insight into the IP values and shifts attending formation of H-bonded complexes of 9HFCA by extending our study to a broader range of H-bonding partners. Adiabatic IP shifts for these complexes are listed in Table I. The calculated IP of the monomer itself is 62 594 cm⁻¹ (B3LYP/6-311++G(d,p)) and 65 153 cm⁻¹ (M06-2X/6-311++G(d,p)), to be compared with the experimental value 64 923 cm⁻¹. The correlation between B3LYP and M06-2X values is excellent (R² = 0.992) and both correlate very well with experimental measurements on complexes with benzoic, propionic, acetic, and formic acids (R² = 0.990 with B3LYP and 0.996 with M06-2X).

We have modeled other complexes with partners ranging from HF to NH₃. In Figure 3, remarkably, the adiabatic IP can be either red- or blueshifted in the clusters depending on the binding partner. Shifts may arise primarily from differences in properties of the neutral complexes or may be influenced by the nature and degree of relaxation after ionization. With the B3LYP/6-311++G(d,p) level of theory, we find that the Koopmans estimate of ionization potentials correlates closely (R² = 0.982) with the computed adiabatic ionization potential. This suggests that the explanation of shifts is to be found in the properties of the neutral complexes.

Consider the set of carboxylic acid partners; acetic acid displays a computed redshift in ionization of ca. 1200 cm⁻¹, in close agreement with our observed shift, while modeling suggests a blueshift of about 600 cm⁻¹ for trifluoroacetic acid. In the acetic acid neutral complex the stronger H bond is from the partner to 9HFCA's carboxyl OH, while for the trifluoroacetic acid neutral complex the stronger H bond is from the partner to 9HFCA's carbonyl group. More specifically, the distance from acetic acid's = O to 9HFCA's HO is 1.654 Å while the acetic acid's OH to 9HFCA O = distance is 1.720 Å. For the trifluoroacetic acid, the corresponding values are 1.744 and 1.637 Å. The most extreme redshift is for the ammonia complex which bonds primarily by accepting the 9HFCA carboxyl OH proton, while the most extreme blueshift is predicted for the HF complex which donates a proton to the 9HFCA carbonyl oxygen. For the ammonia neutral complex, the N⋯HO distance is 1.709 Å while the NH⋯O= distance is 2.666 Å. For the HF neutral complex the F⋯HO distance is 1.930 Å while the FH⋯O= distance is 1.672 Å. As the binding partner acts more effectively as a proton donor to the carbonyl oxygen of 9HFCA, as in the sequence acetaldehyde, formamide to thioformic acid, the shift becomes bluer. This simple geometric criterion accounts systematically for the results of modeling summarized in Table I and lends credence to the hypothesis that the proton donation from the partner is associated with an IP blueshift and proton donation from 9HFCA is linked to the IP redshift.

In conclusion, our application of MATI spectroscopy to 9HFCA and its carboxylic acid complexes reveals an effect on ionization potentials derived from the nature of H bonding in the complex. The well resolved photoionization cation spectrum of 9HFCA is a consequence of the structural stability conferred by the intramolecular H bond in all species. The two intermolecular H bonds in the complexes further inhibit drastic geometry changes attending ionization. Shifts in the ionization potentials of formic, acetic, and benzoic acids complexes relative to isolated 9HFCA can be ascribed to subtleties of H bonding. In short, dominant H donation by the partner tends toward blueshifts; strong H donation by 9HFCA tends toward redshifts.

We thank Wesleyan University for computer time. This work was supported in part by the NSF under Grant No. CNS-0619508.