Principal Investigator Brian Salter-Duke Project g46

School of Mathematical and Physical Sciences, Machine VP

Northern Territory University

Computational Quantum Chemistry of Group 13 Hydrides

Since 1989 I have been studying new hydrides of group 13 elements. These have included the simple and mixed hydrides ABH6 (A or B = B, Al or Ga), the mixed hydride-chlorides of Ga (Ga2HnCl6-n), a new isomer of B3H9, and various borohydrides of both Ga and Al (e.g. MB3H10, M2B2H10, M3BH10, M4H10 and MB2H9 (M = Al or Ga).

Recently a collaboration with the experimental group of A. J. Downs at the University of Oxford has been developing strongly. One project concerns the structure of [AlH(CH3)2]n with the objective of determining whether the oligomer that appears with the dimer is either the trimer or the tetramer. A second project seeks higher level calculations on the Ga2Cl2H4 system than are available from my earlier study. The principal objective here is to predict vibrational frequencies and infra-red and Raman intensities. A third project seeks similar data for GaH3PH3.

This work was carried out primarily on a Dec Alpha 3000/800 in Darwin using Gaussian92, but some calculations required a modest use of the VP2200.

What are the basic questions addressed?

What is the accurate geometry of the molecule? This assists the experimental group in resolving complex electron diffraction data. What are the observable vibration frequencies along with the infra-red and Raman activities? This assists the experimental group in assigning the spectrum.

What are the results to date and future of the work?

All the above questions have been answered and have been sent to the experimental group in Oxford. The results are proving to be valuable and I expect joint publications in the near future.

What computational techniques are used and why is a supercomputer required?

Ab initio quantum chemistry techniques using the Gaussian programs at Hartree-Fock, Moller-Plesset and Density Functional Theory levels have been applied to all systems studied. It appeared to be not possible to use ACESII for coupled cluster calculations on molecules containing Gallium, although this question will be revisited on a visit to ANU later in 1996. With limited 3 hours runs and limited disk scratch space, the largest calculations were carried out in Darwin. However, the VP2200 proved important in using Gaussian94 (which I did not have available in Darwin at that time) for resolving some serious problems of numerical instability (rotational invariance) with these compounds and the need to use very fine grids for the numerical integration over the functionals of Density Functional Theory.