The ammonia dimer potential energy surface: resolution of the apparent discrepancy between theory and experiment?

Abstract

Seven different structures of the ammonia dimer have been investigated by ab initio MO techniques. Geometries have been optimized and vibrational frequencies calculated using SCF MP2 and QCISD methods, with basis sets as large as 6–311 +G(3d′, 2p). Final energies were obtained using QCISD(T) theory. Several stationary points have been vibrationally characterized; the only minimum with large, balanced bases possesses an almost linear H bond, and is qualitatively different from the asymmetric cyclic structure deduced from spectroscopic data. Our best value for D e is 13.27 kJ mol −1 (11.60 kJ mol −1 after BSSE correction). We could not find a stationary point resembling the spectroscopic model. The overall potential surface is extraordinarily flat. The ZPE for a rocking motion of the two NH 3 units is comparable to the barrier separating two equivalent forms of the dimer. Spectroscopic ground-state rotational constants are thus grossly contaminated by vibrational averaging, and cannot be directly compared with the computed equilibrium values.