3. Atmospheric parameters
The simultaneous fitting of Balmer and He line profiles by a grid of synthetic spectra (see Saffer et al. 1994) has become the standard technique to determine the atmospheric parameters of sdB stars. The Balmer lines (H to H 12), HeI (4471 Å, 4026 Å, 4922 Å, 4713 Å, 5016 Å, 5048 Å) and HeII 4686 Å lines are fitted to derive all three parameters simultaneously.
The analysis is based on grids of metal line blanketed LTE model atmospheres for solar metalicity and Kurucz' ATLAS6 Opacity Distribution Functions (see Heber et al. 1999). Synthetic spectra are calculated with Lemke's LINFOR program (see Moehler et al. 1998).
The results are listed in Table 1 and compared to published values obtained from low resolution spectra. The formal errors of our fits are much smaller than the systematic errors (see below). The agreement with the results from low resolution spectra analysed with similar models (Koen et al. 1998) as well as from our own low resolution spectrum for PG 1605+072 is very encouraging.
Table 1. Atmospheric parameters for PG 1605+072 from different methods, see text
Four species are represented by two stages of ionization (HeI and HeII , CII and CIII , NII and NIII , SiIII and SiIV ). Since these line ratios are very temperature sensitive at the temperatures in question, we alternatively can derive and abundances by matching these ionization equilibria. Gravity is derived subsequently from the Balmer lines by keeping and fixed. These two steps are iterated until consistency is reached. CII is represented by the 4267 Å line only, which is known to give notoriously too low carbon abundances. Indeed the carbon ionization equilibrium can not be matched at any reasonable . The ionization equilibria of He, N and Si require to be higher than from the Saffer procedure, i.e. 33 200 K (He), 33900 K (N) and 32 800 K (Si). Table 1 lists the result for the He ionization equilibrium.
This difference could be caused by NLTE effects. Therefore we repeated the procedure for and using a grid of H-He line blanketed, metal free NLTE model atmospheres (Napiwotzki 1997), calculated with the ALI code of Werner & Dreizler (1999). NLTE calculations for N and Si are beyond the scope of this letter.
Applying Saffer's procedure with the NLTE model grid (see Fig. 1) yields almost identical to that obtained with the LTE grid. Evaluating the He ionization equilibrium in NLTE, indeed, results in being consistent with that from Saffer's procedure (see Table 1). We therefore conclude that the higher derived above from the ionization equilibrium in LTE is due to NLTE effects.
However, a systematic difference in persists, the LTE values being higher by 0.06-0.08 dex than the NLTE results (see Table 1). Since its origin is obscure, we finally adopted the averaged atmospheric parameters given in Table 1. Helium is deficient by a factor of 30 as is typical for sdB stars.
© European Southern Observatory (ESO) 1999
Online publication: July 26, 1999