Abstract
Fully optimized structures, normal vibrational modes, and thermodynamic functions were obtained at the MP2/6‐311++G(d,p) level for the following ionic hydrogen‐bonded systems: H3O+/H2O, NH+4/H2O, NH+4/NH3, CH3NH+3/H2O, CH3NH+3/NH3, OH/H2O, CH3O/H2O, cyanide/H2O, HCC/H2O, and formate/H2O. The calculated ΔH fell within error limits of experimental values for all but OH/H2O and the two NH3 complexes. In the latter cases expansion of the polarization basis to (2d,2p) satisfactorily improved the model, but for OH/H2O increasing the level of correlation to MP4 was more important. Where comparison was possible, optimized structures were quite similar to those obtained at MP2/6‐31+G(d,p) by other workers. All the complexes examined here have a hydrogen bond stretching mode in the 200–350 cm−1 range which may contribute to proton transfer, and whose frequency correlates qualitatively with the acid strength of the H‐bond donor. For the cyanide/H2O system, it is not possible to choose between CN/H2O and NC/H2O as the dominant complex in the gas phase: the thermodynamic results suggest a mixture. In formate/H2O a C2V ‘‘cyclic’’ complex is thermodynamically favored but a ‘‘linear’’ complex is apparently observed at ∼500 K. This might well be an entropy effect, but the C2V complex was found here to have a vibrational mode at 92 cm−1 which could facilitate conversion to the linear complex at all but very low temperatures. Counterpoise calculations with ghost functions were done at MP2 for all complexes, and HF/6‐311++G(d,p) for three of the more strongly bound cases. Comparison of the MP2 and self‐consistent‐field results, along with a close examination of the virtual orbitals obtained with and without ghost functions for the components of NH+4/NH3, supports a recent contention that counterpoise calculations with correlation are highly suspect.