3 Line identifications

  

Figure 3: Titanium oxide absorption bands detected in the spectrum of RR Tel at 6651Å, 7052Å, 7666Å and 8206Å

The velocity-corrected measurement for each line of our RR Tel spectrum is listed, in angstroms, in Col. (1) of Table 2. In Col. (5) the suggested identities of the major contributors to these lines are listed, where the ion is followed by the laboratory wavelength. A superscript at one of these wavelengths refers to a note at the end of the table.

The methods employed in identifying the many emission lines were the same as those used by McKenna et al. ([1997]), and are described therein. However, several of their listed identifications were found to be erroneous; our greater spectral resolution allowed us to discard some of their lines, and to re-identify others. One finding list not used by McKenna et al. which was employed in the present study is the near-UV and optical line list of van Hoof ([1998]). However, it is still apparent that not all of the lines in this study are fully resolved. For example, the line at $\lambda$3724.99Å appears to be a blend of MnII $\lambda$3724.81Å, OII $\lambda$3725.30Å and [OII] $\lambda$3726.08Å.

The line widths and shapes were also considered in identifications - most lines of a given ion and multiplet were of similar width and shape. Also, the presence of several lines from a given multiplet makes it likely that other lines from that multiplet will be present. Furthermore, lines were wider for the higher stages of ionisation; for example, the average velocity width for the FeII and [FeII] lines was about $\overline{\Delta v}$=14kms^-1; the average width for [FeV] was 41kms^-1, and an average width of 62kms^-1 was found for the [FeVII] lines. Similarly, OII lines were found to have an average velocity width of 30kms^-1 and the average width of OIV lines was found to be 44kms^-1. We notice the similarity in velocity widths of the [FeV] ($\overline{\Delta v}$=41kms^-1) and OIV ($\overline{\Delta v}$=44kms^-1) lines, where the ionisation potentials are 56eV and 54.93eV, respectively.

In cases where it was unclear which of a number of possible ions were contributing to lines, the average velocity width for each ion was compared to the width of the line in question. For example, the line at $\lambda$3668.99Å was thought to be a blend of HI $\lambda$3669.46Å and [CaVI] $\lambda$3669.10Å. The measured line was significantly broader ($\Delta v$=62kms^-1) than the other HI (multiplet 5) lines which had an average velocity width $\overline{\Delta v}$=50kms^-1, so [CaVI] is listed as a possible co-contributor. On the other hand, HI (multiplet 3) lines were found to have an average velocity width $\overline{\Delta v}$=53kms^-1, and the line at $\lambda$3721.51Å agreed well with this, so the listing of [SIII] at $\lambda$3721.70Å by McKenna et al. ([1997]) was discarded. In all cases where blends are listed, the contributors are listed in decreasing order of perceived importance.

It is possible that some OVI emission has been detected at $\lambda\lambda$3811.36 and 3834.24Å. The 3811.36Å line was listed by McKenna et al. ([1997]), although its doublet partner at 3834.24Å was not. Attempts were made to ascertain whether or not these lines are real, as both features are blends, by comparing the profiles of the OIII (multiplet 2) lines at $\lambda\lambda$3759.87, 3774.00 and 3791.26Å to that of the line at 3811.21Å, which we have identified as a blend of OIII $\lambda$3810.96Å and OVI $\lambda$3811.36Å. However, this method proved inconclusive.

There is evidence of OVI emission at the red end of the spectrum. We find extremely broadened features at $\lambda\lambda$6825 and 7082Å, which have been identified by Espey et al. ([1995]) as the Raman-scattered ultraviolet OVI $\lambda\lambda$1032 and 1038Å resonance lines (Schmid [1989]). These lines have split profiles, probably due to relative gas motions in the red giant's wind.

The line listed in Table 2 at 4446.29Å is thought to be a blend of FeII $\lambda$4446.25Å and [NiVIII] $\lambda$4446.20. The [NiVIII] line is important as it has the highest ionisation potential (I.P.$\sim$136eV) of all the lines measured in this study. It proved impossible to resolve the [NiVIII] totally for the table, but the line width can be roughly measured as $\Delta v$$\simeq$29kms^-1.

Thackeray ([1977]) noted that the presence of the late type star in the RR Tel system is confirmed by the detection of TiO bandheads in the near infrared and red regions of the spectrum. These features are known to be prominent in mid M type stars (Kirkpatrick et al. [1991]; Vardya [1992]) and we have made definitive identifications of the TiO absorption bands (see Fig.3).