Plasma spectral emission codes have been used to analyze high energy astrophysical spectra for two decades. With their implementation into the XSPEC spectral fitting process, two codes (under the long-time leadership of John Raymond and Rolf Mewe, respectively) have become critical to the analysis of collisionally ionized X-ray emitting plasmas. Both codes (now Brickhouse-Raymond-Liedahl and SPEX) are currently in the state of major revision and rewrites (Brickhouse et al 1995; Kaastra & Mewe 1994).
While modern computing capabilities mean that spectral codes can be
made more efficient and user-friendly, the chief motivation for the
upgrades has been to take advantage of improvements in the atomic data
over the last decade. These improvements include not only higher
accuracy for some important atomic rates, but also increased
completeness in the data included. For some strong lines, we now have
rates that may be as accurate as 10 to 20%. Furthermore, we have a
lot more lines -- as an example, the version of Raymond-Smith now
available in XSPEC lists 13 lines of Fe XIX, while we now compute
36,000 emission lines from that ion alone.
Figure 1. ASCA SIS0 spectrum of Capella observed 1996 March 3 to 4. The poor fit at 10 Å is due to lines missing in the plasma emission codes.
Yet, unbelievable as it may seem, we now have good reasons to believe that 36,000 emission lines from Fe XIX is not enough! Figure 1 shows the poor fit to the ASCA spectrum because lines are missing in the plasma codes. These models are being improved with the help of higher resolution spectra from EUVE (Fig. 2). More precisely, we do not have calculations that adequately represent the power radiated by Fe XIX in all X-ray spectral bins. Since an ion has an infinite number of energy levels, all atomic models are necessarily cut off. The usual art of atomic physics is to approximate this infinity of levels in such a way as not to jeapordize the accuracy of a specific transition rate. For CCD-resolution spectra, however, a somewhat different approach is needed to address the issue of completeness. For the complex ions often encountered in X-ray spectra, there can be a lot of power radiated from very high energy levels.
Figure 2. The better resolved EUVE spectrum of Capella observed 1996 March 3 to 7 provides the basis for improving the spectral models.
Duane Liedahl is beginning to generate models to represent the complete power radiated, and we plan to provide these data to the community through the plasma spectral codes as soon as possible. But even in the most optimistic scenarios, we will still only have theoretical data.
In the absence of experimental data, estimating the accuracy of the theoretical data is an educated guess. Different codes and methods can be compared; and the effects of truncation can be explored by increasing the size of the atomic model (currently limited by supercomputer capability). While comparisons with experiments are the only unequivocal tests, experiments to measure collisional cross sections are rare, and most of these have been done for low Z elements and low ionization states. For the higher Z elements and higher ionization states of great interest to AXAF, very little work exists.