3. Spectroscopic data and non-LTE methodology
The spectra analysed in this paper were obtained using the ESO 3.6-m telescope and the Cassegrain-echelle spectrograph (CASPEC) with a Tektronix 512x512 27 m pixel CCD detector. The short camera was used with the 31.6 lines/mm echelle grating and a 2 arcsec slit resulting in a spectral resolution of 20 000 in the wavelength region covered, which was approximately 3830-5240Å. A formal signal-to-noise ratio in excess of 100 was obtained but unexpected fringing, probably due to the use of neutral density filters, was found to be present in the observations. This was most serious in the observations of Ori and increased towards the red, reaching 2% of the continuum level at around 5000Å. The echelle spectra were extracted, background corrected and rectified within the MIDAS environment (Ponz & Brinks, 1986) using standard techniques. The individual, normalised exposures were co-added and transferred to the DIPSO environment (Howarth & Murray 1991), where equivalent width measurements and line fitting were performed.
The analyses presented here are based upon non-LTE model atmospheres and non-LTE line formation computations; further details may be found in McErlean et al. (1997) and Lennon et al. (1991). Briefly, the model atmospheres used include the elements hydrogen and helium, and permit departures from LTE for the first five levels of H I and He I, and the first ten levels of He II. We note that for He I, the splitting of the states due to spin and angular momentum was ignored with these levels being treated as purely hydrogenic.
The subsequent line formation calculations were performed using the programs DETAIL and SURFACE (Giddings 1981 and Butler 1984, respectively), with the former solving the radiative transfer and statistical equilibrium equations and the latter computing the emergent flux. For these calculations the number of non-LTE levels was increased to 10 levels for H I, 27 levels for He I (including the explicit treatment of all LS -states up to and including n =4; states n =5 to n =8 have their spin series treated as degenerate) and 14 He II levels. As pointed out by Lennon et al. (1991), the use of hydrogenic levels above n =4 for neutral helium may be inadequate, as these states are the upper levels for many of the transitions considered in this paper.
Line formation calculations were performed for helium fractions of y = 0.1, 0.2 (Throughout this paper, the definition y = N [He] / N [H + He] is used). We have allowed for the effect of microturbulence on the line profile in the standard way by including an additional term in the usual Doppler width ();
where is the thermal velocity of the ion in question and is the microturbulent velocity. This modified Doppler profile is then convolved with the usual Stark profiles. As we pointed out however, the line formation calculations were performed in two steps. In the first step (DETAIL) we solved for the level populations (or departure coefficients) using purely Gaussian profiles for all lines, detailed numerical profiles only being included in the final profile calculations (SURFACE). Typically in non-LTE calculations of this kind for main sequence B-type stars, microturbulence is only included in the final profile calculation since the effect of values up to 5 kms-1 on the level populations is small (Kamp 1978, Mihalas 1972). However, for supergiants we have to consider values in excess of 10 kms-1 and test calculations showed that significantly different profiles resulted if the microturbulence was also included in the computation of the level populations. Similar results were found for Mg II by Snijders & Lamers (1975, see their Table 2). We have therefore adopted the same value of microturbulence in both DETAIL and SURFACE in all the line formation calculations discussed here.
© European Southern Observatory (ESO) 1998
Online publication: December 8, 1997