MULTIDIMENSIONAL HYDROGEN TUNNELING DYNAMICS IN THE GROUND VIBRATIONAL-STATE OF THE AMMONIA DIMER

We have measured and assigned more than 800 new far-infrared absorption lines and 12 new microwave absorption lines of the ammonia dimer. Our data are analyzed in combination with all previously measured far-infrared and microwave spectra for this cluster. The vibration-rotation-tunneling (VRT) states of the ammonia dimer connected by electric-dipole-allowed transitions are separated into three groups that correspond to different combinations of monomer rotational states: A + A states (states formed from the combination of two ammonia monomers in A states), A + E states, and E + E states. We present complete experimentally determined energy-level diagrams for the K(a)=0 and K(a)=1 levels of each group in the ground vibrational state of this complex. From these, we deduce that the appropriate molecular symmetry group for the ammonia dimer is G144. This, in turn, implies that three kinds of tunneling motions are feasible for the ammonia dimer: interchange of the "donor" and "acceptor" roles of the monomers, internal rotation of the monomers about their C3 symmetry axes, and quite unexpectedly, "umbrella" inversion tunneling. In the K(a)=0 A + E and E + E states, the measured umbrella inversion tunneling splittings range from 1.1 to 3.3 GHz. In K(a)=1, these inversion splittings between two sets of E + E states are 48 and 9 MHz, while all others are completely quenched. Another surprise, in light of previous analyses of tunneling in the ammonia dimer, is our discovery that the interchange tunneling splittings are large. In the A + A and E + E states, they are 16.1 and 19.3 cm-1, respectively. In the A + E states, the measured 20.5 cm-1 splitting can result from a difference in "donor" and "acceptor" internal rotation frequencies that is increased by interchange tunneling. We rule out the possibility that the upper state of the observed far-infrared subbands is the very-low-frequency out-of-plane intermolecular vibration predicted in several theoretical studies [C. E. Dykstra and L. Andrews, J. Chem. Phys. 92, 6043 (1990); M. J. Frisch, J. E. Del Bene, J. S. Binkley, and H. F. Schaefer III, ibid. 84, 2279 (1986)]. In their structure determination, Nelson et al. assumed that monomer umbrella inversion tunneling was completely quenched and that "donor-acceptor" interchange tunneling was nearly quenched in the ammonia dimer [D. D. Nelson, G. T. Fraser, and W. Klemperer, J. Chem. Phys. 83, 6201 (1985); D. D. Nelson, W. Klemperer, G. T. Fraser, F. J. Lovas, and R. D. Suenram, ibid. 87, 6364 (1987)]. Our experimental results, considered together with the results of six-dimensional calculations of the VRT dynamics presented by van Bladel et al. in the accompanying paper [J. Chem. Phys. 97, 4750 (1992)], make it unlikely that the structure proposed by Nelson et al. for the ammonia dimer is the equilibrium structure.