3. Data reduction
3.1. Continuum fitting and identification of other spectral lines
We binned the data in the echelle order of the O VI line into elements containing 7 pixel (of optical resolution), equivalent to 0.21 Å and fitted a 5th order polynominal to define the interstellar continuum. For the galactic targets, a straight line is a good approach to the continuum of the narrow interstellar features (Fig. 1b). A multi gaussian or even single gaussian fit (see HD 93521 in Fig. 1a) was applied to the spectral structures, to define the equivalent width of the explored absorption.
In the vicinity of the O VI line at 1031.92 Å absorption structures due to H2 Lyman P(3) at 1031.19 Å and H2 Lyman R(4) at 1032.35 Å are present. However, these are well separable from the O VI line. Further possible interference may arise from the R(0) 6-0 interstellar line of HD at 1031.91 Å. We determined the contributions from this feature by looking for the 7-0 and 5-0 R(0) HD lines at 1021.453 and 1042.847 Å, respectively. Taking these results into account, an upper limit of 0.04 Å and 0.03 Å was subtracted from the O VI equivalent width to HD 116852 and HD 93840. HD 116852 and HD 93840 were the two only targets with measurable absorptions of HD. Following the data in Morton's (1991) list some metals may show here absorption lines, too. Since these particular elements have small intrinsic abundance and since these lines are from excited states we can ignore any contribution. In most of our spectra, the weaker component of the O VI doublet at 1037.61 Å is blended with the line of excited C II at 1037.02 Å and by two H2 Lyman absorptions at 1037.15 Å and 1038.16 Å, both of level J = 1. Therefore, we used the weaker O VI line only to verify the result derived from the stronger component.
The velocity of the O VI absorption lines usually agree well with those of the C IV lines as seen in IUE and HST spectra. Apparently, the O VI ions exist in gas well related with the gas containing C IV . The complexity of the stellar background spectrum near the O VI absorption together with the intrinsic uncertainty in the IUE velocity scale does not warrant a more detailed comparison at this time.
3.2. Galactic targets
In all spectra of galactic stars which we have included to derive the O VI scaleheight, the H2 Lyman P(3) line at 1031.19 Å was clearly separated from the O VI absorption. The H2 Lyman R(4) at 1032.35 Å blended with O VI for several targets. A two-component fit was used to separate the H2 contribution to the equivalent width for HD 77770, HD 93840 (Fig. 1b), HD 116852 and HDE 214930. In HD 93521 (Fig. 1a) and HD 49798 the R(4) absorption is clearly separated. In these cases a single gaussian was used to determine the equivalent width.
For HD 18100, HD 146813 and HDE 217505 no measurable absorptions from neither H2 Lyman P(3) nor H2 Lyman R(4) was found between 990 and 1120 Å, hence the contribution of any H2 to the O VI line can be neglected in these cases.
3.3. Magellanic Cloud targets
Magellanic Cloud spectra have a poorer signal to noise ratio than those of the galactic stars. Yet, three velocity components can be clearly identified in each absorption. This means, that the H2 lines mentioned above will blend (due to the complex velocity structure of the gas on these lines of sight) with the O VI line, and decomposition may be problematic.
3.3.1. LMC stars
In the spectra of HD 36402, HDE 269546, and Sk 166 neither measurable galactic H2 absorption with rotational levels J 4 nor an LMC component from any absorption with a level of J 3 are present. Thus, for these three targets the zero velocity component was used to derive the galactic O VI column density. We did not include the high negative velocity component we found in all our extragalactic stars. This feature seems to be composed of galactic H2 Lyman P(3) absorption and a second component not yet defined (Fig. 2).
LH 10:3120 : In the O VI line region de Boer et al. (1998) found H2-absorption at LSR velocity as well as at +270 km s-1. The LMC component of H2 P(3) 1031.19 Å and the galactic component of the H2 R(4) 1032.35 Å line blend with O VI . Due to a lower S/N ratio, the separation of the galactic and LMC components of these two H2 absorptions was very inaccurate. Therefore we just fitted the three velocity components mentioned above (negative velocity, zero velocity and LMC component) to the absorption profile. Referring to the results of de Boer et al. (1998) we subtracted 0.2 Å in equivalent width from our zero velocity component.
3.3.2. HD 5980 in the SMC
The SMC star HD 5980 shows a first positive velocity component at +147 km s-1, the velocity of the SMC (Westerlund 1997). In additon, we found at +300 km s-1 a velocity component clearly separated from the normal SMC absorption. It has also been seen in many other ions in IUE spectra of this star (Fitzpatrick & Savage 1983) and the authors suggested an expanding SNR in the foreground to the SMC star as a possible origin of this feature. The equivalent width of our derived zero velocity component leads to an exceptionally small O VI column density. The FWHM of 0.35 Å is also exceptionally small compared to the LMC targets (1.2 Å). It seems reasonable to suspect that galactic and SMC components are blended. However, given the uncertainties we decided not to include the O VI information from the line of sight to HD 5980 in the scale height fit procedure.
© European Southern Observatory (ESO) 1998
Online publication: September 8, 1998