Use of IR Spectroscopy for Structure Determination

USE OF IR SPECTROSCOPY FOR STRUCTURE DETERMINATION

As seen above, IR spectroscopy is most commonly used to identify functional groups and bonding patterns in molecules from the higher energy portion of the spectrum (1200 – 4000 cm⁻¹) where absorptions are primarily due to bond-stretching vibrations. Some information on atom connectivity in the molecule can also be deduced from the frequency shifts caused by structural factors. In general, however, it is not possible to completely deduce the structure of a molecule by examination of its IR spectrum. However, IR spectroscopy is a powerful complement to NMR spectroscopy for structure determination.

For example, the reaction of 3-chlorocyclohexanone with DBU in toluene gives a product which is seen to have two vinyl protons by NMR and thus is an elimination product, probably either A or B. Now while one might rationalize by both chemical intuition and by the splitting pattern that conjugated isomer A is the product, examination of the IR spectrum shows a carbonyl group (at 1680 cm⁻¹) and an olefin band (at 1630 cm⁻¹). A typical cyclohexanone comes at 1710 cm⁻¹ and cyclohexene comes at 1643 cm⁻¹.

Clearly the observed frequencies of the product are at lower frequencies than a simple ketone or olefin and are indicative of a conjugative interaction between these two functions. Thus A and not B is the product.

Treatment of propiophenone with m-CPBA (m-chloroperbenzoic acid) in dichloromethane gives a single product. The carbonyl absorption of pro-piophenone is at 1695 cm⁻¹ whereas the product has a carbonyl absorption at 1745 cm⁻¹. This information reveals that the carbonyl group is intact but is no longer a ketone. The shift to higher frequency is consistent with the conversion to an ester, and thus the product could be either C or D.

Since ethyl benzoate C has a carbonyl stretch at 1725 cm⁻¹, the likely product is D, phenyl propionate. This is confirmed by the NMR spectrum, which has the methylene group as a quartet at 2.42δ. This chemical shift is typical for a methylene group next to an ester carbonyl but is much too high field for the –CH₂–group in ethoxy ester C, which comes at about 3.6δ.

Reaction of cinnamic acid, which has the IR and ¹H NMR spectra shown in Figures 11.36 – 11.38, with BH₃·THF gives rapid consumption of the starting material. A single product P₄ is formed that has the spectra shown. Comparison of the IR spectra shows that the double bond in the reactant (1630 cm⁻¹) is intact in the product (1654 cm^{− 1}) but moved to higher frequency. The product P₄ also has both vinylic (3026 cm⁻¹) and saturated (2861 cm⁻¹) C–H bonds. Both reactant and product have an O–H adsorption, but the strong H bonding in the acid (broad absorption at 3200 – 2600 cm−1) is replaced by a shift to higher frequency in the product (3349 cm−1) indicative of weaker H bonding found in an alcohol. 

Furthermore the carbonyl group in the reactant acid (1681 cm⁻¹) is missing in the product. The IR data suggest that the carboxylic acid has been reduced by BH₃·THF in preference to hydroboration of the double bond.

The ¹H NMR corroborates this conclusion since two vinyl protons are observed both in the reactant and product; however, a new two-proton doublet appears at 4.15δ for the newly produced allylic methylene group. The acid O–H pro-ton is moved far upfield as well. The coupling constants of the vinyl protons (J = 16 Hz) show the starting compound to be trans, and the large splitting for the downfield vinyl doublet of the product (J = 16 Hz) shows the trans stereochemistry to be maintained in the unsaturated alcohol product. Moreover the splitting between the methylene group and the upfield vinyl proton clearly supports its allylic position.

The ¹³C spectrum (Figure 11.39) is consistent with these structural assign-ments as the carbonyl carbon in the reactant (172.5δ) is gone and the product contains a new signal at 63.3δ, typical for a change to sp³ hybridization and an allylic group.

In the preceding example several types of spectroscopy are brought to bear. While the product structure could probably be deduced from IR spectroscopy or NMR (either ¹H or ¹³C), the use of all three methods confirms the assignment. It is often prudent to use more than a single technique for structure determination so that the results reinforce each other. If a structure assignment is not consistent with all the data, the structure is probably incorrect.

This lesson is brought home in the following example taken from a recent experiment. Ample chemical precedent suggested that the treatment of E with methyl amine should give F:

The spectra of the product are shown in Figures 11.40 – 11.42. As is clear, all of the appropriate resonances are present. In the ¹H NMR (Fig. 11.40) the amide N-methyl group is a doublet (J = 4 Hz) at 2.76δ due to weak splitting by the amide NH proton and the amino N–CH₃ group is the sharp singlet at 2.89δ which is not split by the amine N–H proton due to rapid exchange. The C–H methine proton is a singlet at 4.03δ and the ethoxy group is evident by the quartet– triplet A₂X₃ pattern.

However, several pieces of data just did not seem to fit. First the chemical shift of the amino N -methyl group is at 2.90δ whereas several known compounds of similar structure had the amino N -methyl group at 2.48δ. 

The 0.5-ppm shift might be due to the electron-withdrawing properties of the carbethoxy group, but that shift appears to be too large. In fact the chemical shift of 2.9δ is exactly that expected for an amide N -methyl, but in this case it is not split and there is already an amide N -methyl signal at 2.76δ, where it should be. Next, the integrated area of the N–H peak at 4.6δ is only 1H whereas it should account for two N–H’s which exchange. Moreover, the C–H signal at 4.03δ integrates for two protons rather than one. While one might discount the discrepancies in the integrated areas as being due to the fact that the sample was not analytically pure and while one might force fit the change in chemical shift of the amine N–CH₃ group, the compound did not dissolve in acid, as was expected for the amine. Adding to the difficulty was the IR spectrum of F (Fig. 11.41). In addition to the normal ester C=O stretch at 1745 cm⁻¹, the IR had a C=O stretch at 1636 cm⁻¹, which is at a much lower frequency than a normal secondary amide (≈1685 cm⁻¹).

Based on these discrepancies, the assigned structure had to be discarded. Start-ing once again from the beginning, the carbethoxy group (a) is set by the A₂X₃ spin system and the ester C=O stretch. The 2H singlet at 4.03δ is assigned as a two-proton CH₂ group next to the ester group (b) since it has the right chem-ical shift and integrates for two protons. Another fragment which is indicated is the C(O)NHCH₃ group by the chemical shift of the methyl group, its small splitting by the N–H amide proton, and the rather low C=O stretching frequency which suggests an amide group (c). What remains is an N–CH₃ fragment, which can only be placed between fragments b and c. That gives the unsymmetric urea as the assigned structure of F. 

This structure fits the data in every way. The methylene group (2H) and the single N–H proton fit the integration. Both N–CH₃ groups are amide types rather than one amino and one amide type, and both should appear at ∼2.8 – 3.0. The urea group has greater resonance stabiliza-tion, and thus the carbonyl group has more single-bond character and comes at a lower IR frequency than an amide. Finally F is not an amine and should not behave chemically like one, that is, it should not dissolve in acid.

The structure is confirmed by the off-resonance decoupled ¹³C spectrum (Figure 11.42), which shows the CH₂ group as a triplet 61.7δ in addition to the other expected carbon resonances.

It is clear that all the data must fit the structure and vice versa. If not, the assigned structure is probably not correct. It is important to let the data indicate the structure and not make the data fit a preconceived structure. Often the chem-istry will suggest a structure and the data will support that structure. However, that need not always be so, and it is necessary to always check that the data fit the structure. New chemistry is often discovered precisely because expected products are not supported by structural data. Consequently new structures resulting from new chemistry are revealed.