The balances of interactions were studied by computational methods in the translational isomers of a solvent switchable fullerene-stoppered [2]rotaxane (1) manifesting unexpected behavior, namely that due to favorable dispersion interactions the fullerene stopper becomes the second station upon change of the solvent. For comparison, another system, a pH switchable molecular shuttle (2), was also examined as an example of prevailing electrostatic interactions. Tested for 1 were five global hybrid Generalized Gradient Approximation functionals (B3LYP, B3LYP-D3, B3LYP-D3BJ, PBEh1PBE and APFD), one long-range corrected, range-separated functional with D2 empirical dispersion correction, ωB97XD, the Zhao-Truhlar's hybrid meta-GGA functional M06 with double the amount of nonlocal exchange (2X), and a pure functional, B97, with the Grimme's D3BJ dispersion (B97D3). The molecular mechanics method qualitatively correctly reproduced the behavior of the [2]rotaxanes, whereas the DFT models, except for M06-2X to some extent, failed in the case of significant dispersion interactions with participation of the fulleropyrrolidine stopper (rotaxane 1). Unexpectedly, the benzylic amide macrocycle tends to adopt preferentially ‘boat’-like conformation in most of the cases. Four hydrogen bonds interconnect the axle with the wheel for the translational isomer with the macroring at the succinamide station (station II), whereas the number of hydrogen bonds vary for the isomer with the macroring at the fulleropyrrolidine stopper (station I) depending of the computational model used. The B3LYP and the PBEh1PBE results show strong preference of station II in the gas phase and in the model solvent DMSO. After including empirical dispersion correction, the translational isomer with the macroring at station I has the lower energy with B3LYP, both in the gas phase and in DMSO. The same result, but with higher preference of station I, was estimated with APFD, ωB97XD and B97D3. Only M06-2X presented qualitatively correct behavior for the relative stability of the two translational isomers, namely, slight preference of station II for the isolated molecule and higher relative energy of the same isomer with the model solvent DMSO. The electrostatic interactions in 2 have the decisive contribution both when the macroring is positioned at the dipeptide residue for the neutral form, and at the N-benzylalanine fragment after protonation, and the observed behavior of the [2]rotaxane is correctly reproduced by the methods used.
Read full abstract