1 Introduction

In the two previous papers on Al-like ions, IP XI: Saraph & Storey (1996) and IP XIV: Storey et al. (1996), effective collision strengths for excitation of the ground state fine-structure transition ²P$^{\rm o}$$_{1\over 2}$ - ²P$^{\rm o}$$_{3\over 2}$in Ar VI, K VII, Ca VIII and Fe XIV, respectively, were presented. Of the three ions discussed in the present paper S IV is astrophysically the most interesting since the $\lambda 10.5~\mu$m line is an important diagnostic tool. The calculations by Johnson et al. (1986) showed that the effective collision strengths increased by a factor four compared to earlier determinations (Brocklehurst 1972) when the effects of resonances were included. We show in this paper that a still higher degree of sophistication is required to calculate the positions of these resonances accurately. Only then can reliable collision rates be obtained. Our results are 30% higher than those of Johnson et al. (1986) at 10000 K.

The three ions discussed in the present paper are only two to four times ionised and thus have a rather open target structure, with n=4 valence electron states lying among the states of the n=3 complex. The construction of the target model is therefore more complicated than for the heavier ions. The details are given in Sect. 2. The collision strengths were calculated on a very fine energy mesh. In the near threshold region the details of the resonances were accurately delineated. For higher energies a sampling technique was developed as described in Sect. 3. The results are presented in Sect. 4, where a detailed discussion is devoted to the differences between the current work and that of Johnson et al. (1986).

We refer readers to the paper by Hummer et al. (1993) for details of the theory, numerical methods and a description of the computer programs. All energies are given in Rydberg units.