A harmonic adiabatic approximation to calculate vibrational states of ammonia

In the present work, we study the vibrational spectrum of ammonia in full dimensionality (6D), with special emphasis on the tunneling splitting. The inversion motion, i.e., the active mode ν2, is indeed relatively well decoupled from the other five inactive modes. Therefore, an adiabatic separation in an active wave function and an inactive one, approximated with a 5D-harmonic basis function, is certainly well adapted to describing this motion. This separation leads to several 1D-effective Hamiltonians for each 5D-harmonic basis function or adiabatic channel. Two models have been tested: the harmonic adiabatic approximation (HADA), when only one channel is used and the coupled HADA (cHADA), when several channels are coupled. In order to get reliable values for tunneling splitting, our calculations have shown that: (i) the calculation of the electronic potential has to be performed with a large atomic basis set (up to quintuple zeta) with a method including core-valence correlation; (ii) the cHADA is required since the HADA overestimates the energy levels. Furthermore, our values for the tunneling splitting are in good agreement with the experimental data of ammonia and several isotopomers