We note first the enormous increase of the averaged collision strength effected by resonant coupling to dipole-allowed transition channels. This increase extends from the excitation threshold to energies of roughly 1.2 Ryd. The early distorted wave calculations (Krüger & Czyzak 1970) neglected channel coupling completely and the present results for the averaged collision strength differ from theirs by factors three to five. As can be seen from Figs. 2 to 4 the distorted wave results do not even come close to the background values of the collision strengths. Earlier CC calculations have attempted to address this problem, see Johnson et al. (1986), and have obtained an increase by a factor of four for the averaged collision strength compared to the distorted wave calculations. It must be mentioned that the earlier calculations were limited by the computer power that was available at the time.
Secondly, we find that inclusion of resonances alone does not necessarily give the full answer. The proper delineation of the resonance structure is crucial for accurate determinations of the thermally averaged collision strength, particularly at low temperatures, as discussed by Storey et al. (1996) and noted by Johnson et al. (1986).
A third non-negligible effect is caused by any shift of these resonances, especially if they are large features that compare in width with the Maxwell velocity distribution function. The extent of the overlap will determine the magnitude of the thermally averaged collision strength at a given temperature. Significant resonance shifts can result from the use of calculated rather than experimental term energies, from the inclusion or neglect of relativistic effects, or from an early cutoff of the CC expansion, (see the detailed discussion for Ca VIII and Ar VI in Saraph & Storey 1996). We have shown that the discrepancy between the current results and those of Johnson et al. for S IV is due to the difference in the cutoff of the CC expansion, (Fig. 6). By cutting off the present 21 state CC expansion after only six terms we obtain results that are remarkably close to those of Johnson et al. (1986).Acknowledgements
The authors wish to acknowledge the support of the SERC through grant GR/H94979.
Copyright The European Southern Observatory (ESO)