Evaluation of protonation sites in two MacMillan catalysts in solution by gas phase predissociation spectroscopy and electronic structure calculations

The protonation sites of first and second generation MacMillan catalyst were evaluated using cryogenic ion vibrational predissociation spectroscopy. The comparison with calculated spectra showed that when isolated in the gas phase, the species are protonated at the secondary amine, as would be expected by the difference in free energy of the calculated protomers. The relative energies of the conformers were also calculated for the species in solution using the SMD model, in water, methanol and in a 1:1 (v/v) water:methanol mixture. This comparison indicates that the trends observed in the gas phase are retained in solution


Introduction
New challenges posed by the modern world have been pushing the boundaries of Science in general and Chemistry in particular because of its central role. 1 Advances in diverse fields ranging from nanostructured materials, renewable energy and feedstock to new drugs and treatments are a few examples where transformational change is currently underway. 2In this context, Chemistry plays a fundamental role by providing insights into the pathways that drive chemical processes at the molecular level, thus enabling the understanding and development of new processes.Paramount among these processes is the catalytic activation and transformation of chemical bonds. 3,4It is estimated that roughly 85 to 90% of the products generated by chemical industry relies on catalysts at least in one step along the reaction sequence. 5symmetric catalysts are especially important because the absolute configuration of enantiomeric centers are essential for the pharmaceutical industries. 6,7][10][11][12] The use of small organic molecules made it possible to obtain outstanding enantiomeric excess with high reaction yield, while having with much less impact on the environment than previous metal-based strategies due to their lower toxicity to living organisms. 13Moreover, these "green" organocatalysts promise to use reactants more efficiently as well as lower costs associated with purification and disposal.
Here we are concerned with the class of organocatalysts based on the imidazolidinones, also known as the MacMillan catalysts, [14][15][16][17][18] with two structures illustrated in Figure 1.7][28][29] One way to monitor the structures of ionic species in solution is to extract them into a mass spectrometer using ambient ionization, where a highly refined arsenal of tools are available to identify structures.1][32] In the past decade, nonetheless, hybrid methods that combine mass spectrometry with cryogenic ion vibrational spectroscopy have been developed that are capable of structure determination. 33,34In that approach, vibrational spectra are obtained for ions that are cooled after they are extracted from solution to cryogenic temperatures.Once cold (~20K), weakly bound neutral species such as H 2 or N 2 are condensed onto the ions.These adducts or "mass tags" are then photodissociated upon resonant absorption of a single photon from a tunable IR laser.Measurement of the photodissociation yield of the tagged species as a function of the laser wavelength yields the linear vibrational spectra of the mass selected ions in the gas phase.By comparing the experimental band pattern with theoretical calculations for candidate structures of the protonated imidazolidinones, it is possible to establish their protonation sites and thus infer the speciation of these catalysts in solution. 30,35,36n this work, we evaluated the protonation sites of two imidazolidinones known as the first and second generation MacMillan catalysts with structures indicated in Figure 1.
The cations of both neutral compounds corresponding to addition of a proton from solution were extracted by electrospray (ESI) from a 1:1 methanol:water mixture, which is typically used as solvent for these species.Their structures were then interrogated by analyzing the pattern of vibrational bands obtained using cryogenic predissociation spectroscopy with DFT calculations as described below.To gauge whether the gas phase results obtained correlate with the solution phase, solvation effects were modeled by the SMD method, as suggested in the literature. 37UTHOR(S)

Results and Discussion
Figure 2 presents the predissociation spectra of [Cat1+H] + and [Cat2+H] + compared with the calculated spectra of the same species with three different protonation sites: i) secondary amine; ii) amide oxygen and, iii) amide nitrogen, at the B3LYP/6-311+G(3df,2p) level of theory.For both catalysts, the experimental spectra are dominated by mainly two vibrational features at the high frequency range (above 2000 cm -1 ) and a distinctive set of bands in the crowded fingerprint region below 1500 cm-1.This preliminary observation suggests that, despite the change in structure between first and second generation MacMillan catalysts, the protonation site and/or conformation preferences are similar.
For [Cat1+H] + , the relative intensities and splitting between the two bands in the higher frequency range at 3286 and 3051 cm -1 (Table 1) suggest their assignment to the NH 2 + asymmetric and symmetric stretching fundamentals, respectively.This pattern is in best agreement with the theoretical results obtained for the secondary amine protonation, whose scaled vibrational frequencies were calculated at 3369 and 3160 cm -1 , respectively.The calculated spectra of the carbonyl protonated (4.6 kcal mol -1 higher in energy) and the amide protonated species (18.2 kcal mol -1 higher in energy), on the other hand, are in poor agreement, most noticeably in that they completely miss the strong band at 3051 cm -1 observed experimentally.(See Table S1 in Supplementary Information).Moreover, the theoretical spectrum of the isomer protonated at the secondary amine nicely accounts for the pattern of observed bands in the lower frequency region, matching particularly well the carbonyl stretch, (calculated at 1753 cm -1 vs the experimental value of 1770 cm -1 ).On the other hand, the C=O fundamentals for the carbonyl and amide protonated species are almost 100 cm -1 lower and higher in energy with values of 1684 and 1866 cm -1 , respectively, after the same scaling.
The remaining features of the fingerprint region are less definitive.The highest energy amide protonated species show the poorest match in this region, while the other two protomers show a similar profile, with predicted bands red or blueshifted from the experimental bands at 1394 and 1370 cm -1 , assigned to multiple C-H bending motions.
The spectrum of the second generation catalyst [Cat2+H] + (Figure 2 bottom) is similar to that observed for [Cat1+H] + .Once again, the calculated NH 2 + bands of the isomer protonated at the secondary amine best recover the relative intensities and splitting of the strong features observed in this frequency range, (3290 and 3057 cm -1 ) but are significantly blueshifted with bands predicted at 3383 and 3175 cm -1 , (Table 1).The carbonyl stretch predicted for this lowest energy structure (1756 cm -1 ) is also in better agreement with the experimental band at 1770 cm -1 , while the calculated C=O fundamentals of the other protomers do not recover the observed location of this feature (Table S1 in Supporting Information).In summary, the theoretical analysis of the spectra strongly support the identification of the protonated secondary amine structures as the isomeric forms generated for both the Cat1 and Cat2 in an ESI ambient ion source.It is worth noting that, at the B3LYP/6-311+G(3df,2p) level of theory, for both [Cat1+H] + and [Cat2+H] + , although the scaled low frequency bands values are in good agreement with the observed bands, the higher frequency modes are still ca.100 cm -1 blueshifted relative to the experimental values.We note, however, that these modes involve acidic protons that are typically much more anharmonic than the skeletal modes and NH bends.This, in turn, suggests the necessity of the use of dual scale factors, as suggested for calculations carried out with more limited basis sets. 38o compare the effect of different methodologies on the predicted vibrational spectra, the most stable protomer (secondary amine protonation) was also calculated in CAM-B3LYP and M06-2X functionals, as suggested in previous works of our group 37,39 (Table 1), maintaining the same basis set.The use of these functionals led to similar improvements in the calculated spectra for both catalysts (Table 1).In the high frequency bands, the calculated bands were redshifted, reducing the observed shift for B3LYP from approximately 90 cm -1 to 45 cm -1 for the asymmetric NH 2 stretching mode.For the symmetric stretch, the improvement was a little less pronounced as the shift was reduced from approximately 115 to 80 cm -1 (see Figure S1 in the Supporting Information and Table 1).In the lower frequency range, the CAM-B3LYP functional led to no difference on the carbonyl stretching mode when compared to B3LYP functional.For M06-2X, however, the calculated frequency is blueshifted, but only by 6 and 9 cm-1 from the experimental values for [Cat1+H]+ and [Cat2+H]+ respectively.It is also worth noting that for [Cat1+H] + , the intensity of the features observed for the bands 1394 and 1370 cm -1 were better represented by the CAM-B3LYP and M06-2X functionals, despite of the blueshift in the frequency values.

AUTHOR(S)
For both catalysts studied, the secondary amine structure is calculated to be the most stable among the candidate protomers (ΔΔG= 0 kcal/mol).The formation of the protonated species at the secondary amine would also be expected considering the basicity of the functional groups in this species.If the protonation at the amide would occur, for example, it would be preferable to do so at the oxygen atom (ΔΔG= 4.6 or 5.9 kcal/mol, for [Cat1+H] + and [Cat2+H] + , respectively) given that the lone pair of the amide nitrogen is less basic, with it being resonance-stabilized in this structure.For that reason, these protomers have a higher free energy (ΔΔG= 18.2 and 18.3 kcal/mol for [Cat1+H] + and [Cat2+H] + , respectively) and are therefore not expected to be observed in the gas phase.
We recognize that, despite the ample evidence for the protonation site to be located at the secondary amine, the results discussed so far are carried out in the gas phase and indirectly probe the speciation of the cations in solution prior to extraction into the mass spectrometer by electrospray ionization (ESI).][42] To evaluate if the results observed in the gas phase could be transferred to solution, we carried out the continuous solvation calculation using the SMD method in water, methanol and a methanol:water 1:1 mixture at the M06-2X/cc-pVDZ level of theory, as this is the preferred functionals for SMD calculations.The relative energy obtained for each catalyst in these different condensed phases are collected in Table 2.These relative free energy values show that, for both catalysts and solution compositions, the same trend observed in the gas phase is predicted to be preserved in solution, with the secondary amine protonated protomer being more stable than the amide protonation at oxygen and nitrogen, respectively.In fact, when implicit solvation effects are considered, the relative free energy for the amide oxygen protonated protomer increases from approximately 5 to 10 kcal mol -1 when compared to gas-phase results, while the amide protonation relative energy increases from 2 to 4 kcal mol -1 .This suggests that the protonated secondary amine is even more stabilized than the other protomers in solution, what is in line with what is expected for amine protonation trends from gas phase to solution. 43This trend is in agreement with the pK a values expected for the different protonation sites of imidazolidinones in water 44,45 (Figure S2 in Supporting Information) and can be explained by the fact that the secondary amine charge is more localized than the charge in the other two protomers, which are likely more delocalized by the amide resonance.This hypothesis is supported by the NBO charge calculated at the B3LYP/6-31+G(d,p) level of theory for [Cat1+H] + and [Cat2+H] + , as shown in Figure 3.As can be seen, the secondary amide shows the lowest NBO charge in comparison to the other protonation sites.Therefore, the increased charge density in the secondary amine is better solvated than the other two protomers.
These results suggest that the protonation tendency observed in the gas phase may be transferred to solution, as the relative energy values of the calculated protomers are retained. 37

Conclusions
The MacMillan catalysts were evaluated using Predissociation Spectroscopy and showed that the protonation site for these species occurs at the secondary amine in the gas phase, for both catalyst generations.This species is also the most stable protomer by 4.6 and 5.9 kcal mol -1 for [Cat1+H + ] and [Cat2+H + ], respectively.The predicted absorptions at the B3LYP/6-311+(3df,2p) level of theory accurately reproduced the bands at the lower frequency region of the spectra but underestimated the red-shifts in the NH 2 stretching bands in the highly anharmonic region above 3000 cm -1 .This discrepancy was less significant, however, with calculations using the CAM-B3LYP and M06-2X functionals.Solvation effects were modeled by the SMD method and showed that for both catalysts, the energy trends observed in the gas phase are conserved in solution for water, methanol and water:methanol 1:1 mixture, as expected by comparison with pK a values in water for similar species.Therefore, we suggest that the protonation site characterized in the gas phase can be extrapolated to the situation in solution.These considerations indicate that the secondary amine is the preferential protonation site for the imidazolidinones, and the preference for this arrangement can be rationalized due to the amide resonance that reduces the amide basicity and charge density, thus stabilizing the secondary amine protonation site in the gas phase and in solution.
At this temperature, the ions are "tagged" with molecular nitrogen, forming an adduct that is mass selected and probed by infrared predissociation spectroscopy.An OPO/OPA parametric system (Laservision) was used to dissociate the tagged ions in a range from 1000 cm -1 to 3600 cm -1 .The photodissociation yield as a function of the laser wavelength provides de vibrational spectrum of the ion.Experimental results were compared to electronic structure and vibrational spectrum calculations carried out at the Gaussian09 computational package. 48All reported species were confirmed to be true minima by displaying no imaginary frequency upon vibrational analysis.DFT was used to describe the electronic structure, and different functionals, B3LYP, 49 CAM-B3LYP 50 and M06-2X 51 were tested with the same basis set, 6-311+G(3df,2p).Scaling factors were used to correct the vibrational frequencies due to anharmonicity effects.The theoretical spectra were scaled by 0.967, 0.947, 0.952 at B3LYP, CAM-B3LYP and M06-2X, respectively. 52NBO charges were calculated at the B3LYP/6-31+G(d,p) level of theory in NBO 3.1 as implemented in Gaussian09. 53MEP plots were calculated at by Gaussian G09 at this same level of theory and plotted with Chemcraft visualization program. 54A gaussian profile of 10 FWHM was used to represent the band width in the calculated spectra reported.All optimized geometries and comparison between different methodologies can be found in the Supporting Information Material.SMD calculations were carried out for methanol and water using standard G09 parameters.Methanol:water 1:1 mixture parameters were obtained by using the experimental values for a 1:1 mixture of methanol:water (~ 0.4:0.6 methanol:water molar fraction) for refraction index 55 and surface tension 56 and taking the molar fraction weighted average for the Abraham's hydrogen bond acidity and basicity, and dielectric constant, 57 necessary for the SMD calculations.These parameters can be found in the supplementary information.

Figure 3 .
Figure 3. Representation of a) Cat1 and b) Cat2 NBOcharges at the protonation sites at the B3LYP/6-31+G(d,p) level of theory overlaid to a molecular electrostatic potential surface mapped on a 0.013458 iso-surface value.C -yellow, H -blue, N -purple and O -red.