Fluorescein and its derivatives bear several functional groups, which may serve as sites for protonation. Furthermore, the carboxyl-group of the benzoic acid moiety is not rigid, resulting in a large number of possible rotamers for both the cationic and anionic species. Jockusch et al. performed IRMPD measurements on isolated fluorescein [F] and 2’,7’-dichlorofluorescein ions in an ion trap. Surprisingly, they found the favored monoanionic isomer in the gas phase to be the phenolate instead of the benzoate form.[26] Introducing functional groups into the phenyl ring may change this behavior, especially in the case of [5-NF]-, as benzoic acid functionalized with electron
withdrawing groups has typically a lower pKa value.[27] For a first qualitative insight into the conformers that are present in the ion trap, we recorded IRMPD spectra in a frequency region of 2800-3760 cm-1, which is sensitive to O-H-stretching vibrations and should allow for discrimination between tautomeric forms by comparison with theoretical IR spectra.
The geometries and respective relative Gibbs free energies of the five lowest energy conformers of [5-AF]- and [5-NF]- are shown in Figure 3. Note that for the sake of
conciseness, we omitted rotamers resulting from rotation of the O-H-group at X (in Anion X and Anion X’; lactone), as the energy differences are negligible and, more
87 importantly, their excited state properties are expected to be indistinguishable from those of the conformers we took into account.
Figure 3 DFT (ωB97XD/6-31+G(d)) optimized ground state geometries and
relative Gibbs free energies of the five lowest energy conformers of [5-AF]-
(left) and [5-NF]- (right).
Anion B1 and Anion B2 are calculated to be the two most stable conformers at the B3LYP/6-31+G(d) level of theory for both [5-AF]- and [5-NF]-. In Anion B1, the
hydrogen of the carboxylic acid group points towards the center of one of the terminal phenyl rings in X, whereas in Anion B2 the carboxylic O-H group points away from X. Anion B3 is slightly less stable for both derivatives. Another conformer would result from rotating the O-H group of Anion B3 (hydrogen pointing “downward”). However, it was not found as the structure obtained from geometry search is identical to Anion B1, which implies that a long range hydrogen-bond like interaction between the carboxylic
88
hydrogen and the X provides significant stabilization to this conformer. This is also suggested by the unusual orientation of the carboxylic O-H found in the geometry of Anion B1. The benzoates are predicted to be less stable, by +29.4 kJ/mol for [5-AF]- and
(noteworthy) only +11.1 kJ/mol for [5-NF]-, than the respective lowest energy
conformer. The electron withdrawing property of the nitro-group apparently stabilizes the excess charge after deprotonation at the carboxylate site, whereas the amino substituent has the opposite effect, resulting in a less stable conformer Anion X. The lactone forms of [5-AF]- and [5-NF]- (Anion X’) both exhibit relative energies (24.4 and
11.9 kJ/mol, respectively) comparable to the non-lactone form. The lactone form has a slightly skewed geometry, as the central C-atom in the xanthene ring is forced from a planar to a tetrahedral configuration, which impairs conjugation within the π-system. IRMPD of [5-AF]- (m/z 346) yields only two major fragment ions at m/z 302 and 301,
attributed to the loss of CO2 and CO2H (Figure S2), whereas for [5-NF]- (apart from the
loss of CO2 and CO2H at m/z 332 and 331, respectively) additional fragmentation channels (m/z 315, 302 and 285) were observed (Figure S3). In general, it is quite difficult to unravel the fragmentation behavior of purely organic compounds and usually a thorough mechanistic study is required. At this point, we refrain from providing a definite identification of the occurring fragment ion signals. However, we would like to point out that the additional signals observed for [5-NF]- probably involves the nitro-
substituent as a formal leaving group (resulting in e.g. the signal at m/z 302 and 285) whereas for [5-AF]-, no fragment signal originates from a cleaving of the C-NH2 bond in
the substituted benzoic acid sub-unit. This results probably from a stronger C-N bond between the benzoic acid carbon atom and the amino nitrogen atom with respect to the C-NO2 bond strength. In a recent study on the bond lengths and dissociation enthalpies in substituted benzene compounds it was found that the C-NO2 bond in nitrobenzene is indeed weaker than the C-NH2 bond in aniline. This was explained by a weaker hyperconjugation in nitrobenzene, due to unfavorable π-π* electrostatic repulsion[28] thus opening up additional fragmentation routes resulting in a more complex fragmentation behavior of the nitro-compound. Recording the fragment ion intensity as a function of photon energy enables us to record gas phase IR spectra. The experimental IRMPD and calculated IR spectra of [5-AF]- are shown in Figure 4.
89
Figure 4 Experimental IRMPD (top, red) and IR spectra (below, black)
calculated for the five considered conformers of [5-AF]-. Calculated IR
intensities are shown as sticks and broadened with 7 cm-1 fwhm Lorentzian functions (black lines). A scaling factor of 0.95 was employed.
90
Two broad absorption bands (~3040 cm-1 and 3420 cm-1) and a more narrow, low intensity band (~3650 cm-1) can be identified in the experimental IRMPD spectrum of
[5-AF]-. By comparison with the calculated spectra of the respective conformers, the
absorption centered at 3040 cm-1 (green) can be assigned to aromatic C-H stretching vibrations. The position of the intense band at ~3420 cm-1 matches better the predicted weakened COO-H stretching vibration of Anion B1, whereas the same stretching vibration (in Anion B2 and B3) exhibits reduced calculated absorption intensity and additionally is shifted to higher energies (to ~3590 cm-1). The expected N-H stretching (symmetric and asymmetric) vibrations were not observed in the experimental spectrum. This is either due to them being low in intensity or the bands coincide with the absorption band at ~3420 cm-1, as suggested by the calculated spectrum of Anion B1. Lastly, the weak absorption band at ~3650 cm-1 can probably be attributed to a free O-H stretching vibration, located at the X moiety, which is exclusive to Anion X and X’. Clearly, the experimental spectrum agrees best with the calculated spectrum of Anion B1. From the relative intensities of the observed absorption bands in the IRMPD spectrum, one may deduce that the population of trapped anions consists mainly of Anion B1 and probably, to a far lesser degree, of Anion X and X’. Anion B2 and B3 cannot be completely excluded, due to their low relative energies (4.7 kJ/mol and 15.3 kJ/mol, respectively). However, the frequencies of the calculated O-H stretching vibrations (~3590 cm-1) do neither fit the broad, intense band (3420 cm-1) nor the low intensity band (3650 cm-1) observed in the experiment. A summary of the calculated vibrational frequencies and intensities is given in Table S6.
91
Figure 5 Experimental IRMPD (top, red) and IR spectra (below, black)
calculated for the five considered conformers of [5-NF]-. Calculated IR
intensities are shown as sticks and broadened with 7 cm-1 fwhm Lorentzian functions (black lines). A scaling factor of 0.95 was employed.
92
Similar to [5-AF]-, the experimental IRMPD spectrum of [5-NF]- (Figure 5) exhibits two
broad IR bands (~3040 cm-1 and ~3400 cm-1) and a narrow band at ~3650 cm-1, which has a drastically increased intensity compared to the absorption band of similar energy in the spectrum of [5-AF]-. Additionally, a low-intensity band in the O-H stretching
region is observed at ~3570 cm-1. By comparison of the experimental spectrum with the calculated spectra, it is obvious that the broad absorption at ~3400 cm-1 is best described by the hydrogen bonded COO-H stretching vibration, present in Anion B1. The narrow absorption band at ~3650 cm-1 agrees well with the calculated free phenolic O-H stretching vibration of Anion X and X’. Lastly, the free COO-H stretching vibrations calculated for Anion B2 and B3 coincide with the low intensity band observed at ~3570 cm-1 in the experimental spectrum. The spectral comparison implies that no conformer can definitely be excluded for a complete theoretical description of the experimental spectrum. Anion B2 and B3 should, however, not be significantly populated, whereas the intense absorption bands at ~3400 cm-1 and ~3650 cm-1 provide strong evidence for a high contribution of Anion B1 and Anion X/X’ like conformers to the ion population present under the given experimental conditions. A summary of the calculated vibrational frequencies and intensities is given in Table S7. It is not clear as to why there is no impactful spectral evidence for Anion B2 and B3 in the experiments, although these conformers are predicted to be the two most stable structures next to the lowest energy conformer. Anion X and X’, on the other hand, have calculated relative energies similar to that of Anion B3, however, the high intensity of the free O-H stretching vibration in the IRMPD spectrum implies that this tautomer contributes significantly to the ion population. It is conceivable that due to the lose nature of the carboxylic acid group, the O-H may rotate more or less freely at room temperature, allowing for interconversion between the geometries of the phenolate and thus in average a higher population of the lowest energy conformer, within the timeframe of the experiment. A change between the two tautomeric forms, on the other hand, is not possible in the gas phase, which implies that already a considerable amount of the benzoate must be present in solution and is kinetically trapped in this form during the transfer from solution to the gas phase.
93 In summary, the IRMPD spectra in the O-H stretching vibration region strongly point to a population of anions mainly consisting of Anion B1 like geometries in the case of
[5-AF]-, whereas for [5-NF]- both conceivable tautomers contribute equally to the ion
population and thus lead to the observed spectral features in the experiment.
For the cations, only four distinct isomers were identified. Stable lactonic structures were not obtained from geometry search. Again, for the sake of convenience, we omitted rotamers resulting from rotation of the two O-H-groups at X, as their spectroscopic behavior is not expected to be drastically different. The geometries and respective relative Gibbs free energies of the four lowest energy conformers of [5-AF]+ and [5-NF]+
are presented in Figure 6.
Figure 6 DFT (ωB97XD/6-31+G(d)) optimized geometries and relative Gibbs
94
The four lowest energy cationic conformers of [5-AF]+ and [5-NF]+ solely differ in the
orientation of the carboxylic O-H group. In Cation 1, the hydrogen is bound to the oxygen atom farthest away from X, whereas in Cation 2 the hydrogen is in closer proximity to X. However, it points in the opposite direction, probably to minimize interaction with the increased positive charge density in the protonated fluorophore X. The geometry of Cation 3 results from rotation of the carboxylic O-H bond starting from the geometry of conformer Cation 1. The hydrogen atom, however, is not pointing straight downward, but rather out of the phenyl-ring plane to avoid the adjacent phenyl-hydrogen atom. Such a conformer was not found for the anionic species, as structure optimization starting form a similar geometry resulted in Anion B1. Surprisingly, the geometry denoted as Cation 4, which is reminiscent of the lowest energy conformer found for the anions (Anion B1), is the energetically highest conformation. The deviation from the mutual orientation of 90° between X and B (~27° and ~19° for [5-AF]+ and [5-NF]+) is
highest for this geometry and additionally the carboxyl group is twisted out of the phenyl-plane by >50°, avoiding a geometry in which the hydrogen atom points directly towards the xanthene unit. A summary of the changes in geometry upon structure optimization of the respective conformers is given in the supporting information section (section 4.9.2; Table S1 and Figure S1).
The fragmentation of the cations induced by IR irradiation is quite complex compared to the respective monoanions. A similar behavior was observed by Jockusch et al. in their IRMPD and UV/Vis photodissociation studies on [F]+ and its two halogenated
derivatives.[26, 29-30] The main fragmentation channels for [5-AF]+ and [5-NF]+ are the
loss of the carboxylic acid group (m/z 302 and 332, respectively) and additionally the concomitant loss of the respective functional group (m/z 286; -HCOOH,-NH2 for [5-AF]+;
-HCOOH,-NO2 for [5-NF]+). The relative signal intensity of the fragment at m/z 286 is
higher for [5-NF]+, indicative of a weaker bond strength between the carboxylic acid
ring and the formal leaving group. The experimental IRMPD and calculated IR spectra of
95
Figure 7 Experimental IRMPD (top, red) and IR spectra (below, black)
calculated for the four considered conformers of [5-AF]+. Calculated IR
intensities are shown as sticks and convoluted with 7 cm-1 fwhm Lorentzian functions (black lines). A scaling factor of 0.95 was employed.
96
Figure 8 Experimental IRMPD (top, red) and IR spectra (below, black)
calculated for the five considered conformers of [5-NF]+. Calculated IR
intensities are shown as sticks and convoluted with 7 cm-1 fwhm Lorentzian functions (black lines). A scaling factor of 0.95 was employed.
97 The experimental spectra of the cations exhibit less spectroscopic features than those of the monoanions, although more IR active chromophores are present. In the IRMPD spectrum of [5-AF]+ only a single narrow band at ~3610 cm-1 is distinguishable, whereas in the spectrum of [5-NF]+ a low intensity band at ~3560 cm-1 and an intense band at ~3610 cm-1 can be identified. No spectral features, which can be associated with aromatic C-H stretching vibrations, are observed for both cationic species. These, however, have a smaller calculated IR intensity for all considered conformers compared to calculated IR intensities for the anionic species (nearly an order of magnitude smaller than for [5-AF]-). Furthermore, the expected N-H stretching (symmetric and
asymmetric) vibrations are not observed in the experimental spectrum of [5-AF]+.
Either the absorption intensity is severely overestimated by the level of theory applied, or the bands are spectroscopically dark, due to e.g. a poor vibrational energy redistribution (IVR) process. The calculated IR spectra of Cation 1 and 2 are nearly indistinguishable regarding band composition and energetic spacing between the calculated vibrational modes. Although theory predicts (for Cation 1 and 2 like geometries) two separate absorption bands for the free phenolic O-H and the carboxylic O-H stretching vibration, presumably only one of those can be assigned to the single band observed in the experimental spectrum of [5-AF]+, unless the energy of the COO-H
vibration is underestimated by theory. Hence it would lie closer to the free O-H vibration, merging into one absorption feature, as predicted, e.g. by the calculated IR spectra for Cation 3 and 4 like geometries. This, however, is questionable, as the experimental IRMPD spectrum of [5-NF]+ exhibits two distinct absorption bands
(~3560 cm-1 and ~3610 cm-1), which can be assigned either to the free xanthene O-H or the COO-H vibration, agreeing reasonably well with the calculated IR spectra for Cation 1 and Cation 2. It is thus conceivable that two separate bands (associated with either the O-H or COO-H stretching vibration) are more likely to appear in the experimental spectrum of [5-NF]+, as the electron withdrawing property of the nitro-
substituent should weaken the O-H bond in the carboxylic acid group, shifting it to lower energies, with respect to the phenolic O-H vibration. In [5-AF]+, the amino-substituent
possesses electron donating properties. Hence, the opposite effect is expected, i.e. a shift of the COO-H vibration towards higher energies and closer to the free xanthene O-H vibration. If this is indeed the case, then the level of theory applied may not be sufficient to account for such an effect. The weakening of the COO-H bond, due to hydrogen like
98
bonding is predicted by theory for Cation 3 and 4 like structures so that the frequency of the COO-H stretching vibration virtually coincides with the free xanthene O-H vibration. However, these two geometries have calculated relative energies of 35.7 (35.4) kJ/mol and 44.6 (52.7) kJ/mol for [5-AF]+ ([5-NF]+) and are hence not expected to contribute
significantly. An overview over the calculated vibrational frequencies and intensities for
[5-AF]+ and [5-NF]+ is given in Table S8 and S9, respectively.
In summary, the IRMPD spectra of the anionic compounds imply that the ion population of [5-AF]- consists mainly of the lowest energy isomer Anion B1, whereas for [5-NF]- no
isomer can be definitely excluded. The relative intensities of the appearing gas phase IR bands, however, point to the coexistence of mainly Anion B1 like geometries and its tautomeric form (either Anion X or the lactonic Anion X’). The rotamers (Anion B2 and B3), although not lying significantly higher in energy, seem not to contribute significantly, unless their spectral features remain dark for a yet unknown reason.
The experimental spectra of the cations exhibit less spectral features than the spectra of the anions. The main absorption band in the calculated spectra of the isomers fits the experiment in either case, making identification of the ion population questionable. However, due to the high relative energies of Cation 3 and 4 isomers, mainly Cation 1 and 2 are expected to contribute to the ion ensemble in the trap.