1 Introduction

Transitions within the n=2 complex of ions in the boron isoelectronic sequence are of great importance in the study of laboratory and astrophysical plasmas. Their spectral lines are observed in a wide variety of astronomical sources where they are frequently used as temperature, density and abundance diagnostics (Peng & Pradhan 1995). For example, lines belonging to C II, N III and O IV have been identified in the spectrum of planetary nebulae (Stencel et al. 1981), giant stars (Judge 1986a,b; Carpenter et al. 1991), symbiotic stars (Nussbaumer & Schild 1981; Hayes & Nussbaumer 1986), in the solar chromosphere (Doschek et al. 1977) and the solar transition region (Doschek et al. 1976; Brekke et al. 1991). Sandlin et al. (1976) have observed the five components of the 2s$\sp2$2p $\sp2$P$\sp0$ - 2s2p$\sp2$ $\sp4$P multiplet of Fe XXII in solar-flare spectra.

Reliable laboratory lifetime measurements based on the beam-foil excitation technique have been reported for the short-lived states of C II and N III (Reistad et al. 1986; Bengtsson et al. 1995; Nandi et al. 1996) and for the $\sp4$P$_{\rm J}$ metastable levels of Fe XXII by Hutton et al. (1997). Also, by recording the spontaneous emission time from metastable ions stored in a cylindrical radio-frequency ion trap, Fang et al. (1993a,b) succeeded in determining radiative rates for the $\sp4$P$_{\rm J}$ levels in C II and N III. These experimental data offer a good opportunity for comparison with theory.

Due to their importance in astronomy, ions of the boron sequence have received a high priority in the IRON Project (IP, see the general description by Hummer et al. 1993), an on-going international collaboration concerned with the computation of accurate atomic data for ions of astrophysical interest (a complete list of papers in the IP series can be found at the URL http://www.am.qub.ac.uk/projects/iron/papers /papers.html). In the context of the IP, electron impact excitation rates for boron-like ions have been calculated by Zhang et al. (1994) as a continuation to the work by Blum & Pradhan (1992) on C II, N III and O IV. The present study can be regarded as a complement to this work by generating an accurate ($\sim$10%, say) dataset containing the corresponding radiative-decay rates. It is primarily intended to increase the content and usefulness of the IP public databases, but an attempt has also been made to provide accuracy ratings for the listed transition probabilities.

Radiative datasets for the boron sequence have been previously computed by Dankwort & Trefftz (DT, 1978) in the multiconfiguration Hartree-Fock (MCHF) approximation; by Cheng et al. (CKD, 1979) using the multiconfiguration Dirac-Fock method; and by an approach based on many-body perturbation theory by Merkelis et al. (MVGK, 1995). These calculations are mainly concerned with electric dipole transitions (E1) although Cheng et al. have also considered the forbidden electric quadrupole (E2) and magnetic dipole (M1) transitions within the ground term. The latter transitions have also been calculated for the sequence by Froese Fischer (1983) in the MCHF approximation. The aim of the present work is to generate an A-value dataset for the n=2 transitions in B-like ions more complete and of higher statistical accuracy than previous work. In this respect it is worth mentioning that the level of agreement between the published datasets is not satisfactory: for some transitions the quoted rates show discrepancies as large as an order of magnitude. We are therefore interested in establishing the physical effects that cause such differences. Since most of the problems appear towards the low-Z end of the sequence due to correlation effects, we make extensive comparisons with experiment and with the following detailed single-ion calculations: C II by Nussbaumer & Storey (1981), Lennon et al. (1985) and Froese Fischer (1994); N III by Nussbaumer & Storey (1979), Brage et al. (1995) and Bell et al. (1995); and O IV by Brage et al. (1996).

For the present work we have made use of the atomic structure program SUPERSTRUCTURE (Eissner et al. 1974; Nussbaumer & Storey 1978; Eissner 1991). This code allows for the inclusion of configuration interaction (CI) and Breit-Pauli (BP) relativistic effects, and has proven to be a practical platform for the computation of extended radiative datasets. Within the IP it has recently been used in a similar fashion for the carbon and oxygen isoelectronic sequences by Galavís et al. (1997).

The numerical method is described in Sect. 2, followed by analyses of the energy levels (Sect. 3) and transition probabilities (Sect. 4). Conclusions are summarised in Sect. 5.