Here’s a nice example of the productive interplay between experiment and computations.1 The dipeptide N-Acyl-Ala-Ala-Benzyl was prepared and subjected to UV and IR/UV analysis. The IR showed two separate structures with distinctly different environments for the NH bonds: one structure showed intramolecular hydrogen bonding while the other did not.

B97/TZVPP computations revealed two structures. The first is a linear dipeptide with intramolecular hydrogen bonding occurring in a 5,5 relationship. (There are actually three conformers of this but all have similar energy, only one is shown in Figure 1.) The second structure displays a bent shape with a NH-π interaction, also shown in Figure 1. The computed vibrational spectra for each structure matches up well with the NH region of the experimental IR.

Figure 1. B97-D/TZVPP optimized structures of N-Acyl-Ala-Ala-Benzyl.

The authors spend a great deal of time noting that the 0 K energies predict that the second structure, being 4 kcal mol-1 more stable, should be the only one observed. However, since the jet cooling will likely trap the structures at their 300 K distribution, this could account for the existence of two structures. However, when the computations include entropy corrections, so now we’re looking at ΔG(200 K), B97-D and MO6-2x suggest that the two structures are very close in energy. But they caution that MP2 predicts a large energy gap unless atomic counterpoise corrections are used to account for intramolecular basis set superposition (see this post), a problem that appears to be much less severe with the DFT methods.


(1) Gloaguen, E.; de Courcy, B.; Piquemal, J. P.; Pilme, J.; Parisel, O.; Pollet, R.; Biswal, H. S.; Piuzzi, F.; Tardivel, B.; Broquier, M.; Mons, M. J. Am. Chem. Soc, 2010, 132, 11860-11863, DOI: 10.1021/ja103996q