SpringerLink
Forum Springer Astron. Astrophys.
Forum Whats New Search Orders


Astron. Astrophys. 326, 1069-1075 (1997)

Previous Section Next Section Title Page Table of Contents

3. Reduction and analysis

The spectra were reduced using standard IRAF software running on a SUN SPARC station at the Dublin Institute for Advanced Studies. The overscan values and the averaged bias exposures were subtracted, the data trimmed to remove the overscan and edge columns of the CCD, and the data divided by averaged and normalized flat fields. One dimensional spectra were extracted, and were wavelength calibrated using ThAr arc lamp spectra. The different spectra of each star were weighted and combined to form single spectra. The continua were set automatically using a low order polynomial fit to the uncontaminated continua. On occasions when this procedure was clearly unsatisfactory, the continua were set manually. The line strengths were measured by fitting Gaussians to the line profiles. Example spectra are illustrated in Fig. 1, where a Nd II and a FeI line are marked.

[FIGURE] Fig. 1. Example spectra

Line transition probabilities were taken from the literature, using the highest quality, and most up to date measures available. These are reported, together with the measured equivalent widths in Tables 9 and 10 in the appendix (available in electronic format from CDS - Centre de Données astronomiques de Strasbourg). The abundances were determined using the fine abundance analysis program WIDTH6 (Bessell & Norris 1984). This programme uses iterative spectrum synthesis, whereby abundances are adjusted until equivalent widths of model lines match those of the measured lines. The results were so good in nearly all cases, that little additional adjustment of the gf -values was required.

In the majority of cases, only line strengths of less than 100mA were used in the analysis. The notable exceptions were the lines of BaII, which in most cases, were still less than 200mA. No corrections were made for hyperfine splitting, as these were not important in determining the abundances of the n-capture elements, of interest in this work.

The only comparison that could be made with equivalent widths quoted in the literature, was with the work of Kraft et al. (1992). These authors measured lines in 4 stars in common with this work, including M13_I_13. The comparison is shown graphically in Fig. 2, where the line represents strict equivalence. The agreement in excellent, with the average difference (this work minus Kraft et al. (1992)) of -2.2 [FORMULA] 4.2mA. It is clear that abundance errors due to measurement errors are negligible except for the very weakest lines.

[FIGURE] Fig. 2. Comparison of equivalent widths between this work and Kraft et al. (1992)

The atmospheric parameters are shown in Table 2. Although adequate photometry was available in some cases, for consistency, all stars were analysed using only spectroscopic criteria to determine the atmospheric parameters. The consistency of the results with those from the literature was checked to determine the efficacy of the procedures.

[TABLE]

Table 2. Derived physical parameters of the stars in this study
2 Suntzeff et al. (1988)
3 Kraft et al. (1992)

The effective temperatures were determined by requiring that the iron abundances determined from FeI, were independent of excitation potential. This was accurate to within the model atmosphere grid spacing of [FORMULA] 125K (Kurucz 1992). The gravities were determined by requiring the abundances of iron derived from FeI and FeII lines, should match (log g [FORMULA] 0.5 cm s-2). The microturbulent velocities [FORMULA], were determined by requiring the iron abundances derived from FeI be independent of equivalent width([FORMULA] [FORMULA] 0.5 kms-1). This required the use of stronger Fe I lines, to ensure that an accurate value of the parameter was measured for use, especially, with elements having only strong lines. Metallicities of the model atmospheres were kept at a constant value of [Fe/H] = -1.5, since this was taken to be unknown a priori. The resulting errors (see Table 4) were not significant enough to require reanalysis at more exact metallicities. The abundances were calculated using the WIDTH6 modeling code, with recent hydrostatic, plane-parallel model atmospheres, assuming LTE (Kurucz, 1992). Table 3 details the physical parameters derived for the standard stars, and compares them with the most recent values quoted in the literature. The agreement is very close, and gives grounds for believing the methods used here are sound.

[TABLE]

Table 3. Physical parameters of the standard stars
4 K92 = Kraft et al. (1992); G84 = Gratton et al. (1984); G89 = Gratton (1989); G88 = Gilroy et al. (1988); S91 = Sneden at al. (1991); GS91 = Gratton and Sneden (1991); GS94 = Gratton and Sneden (1994); SC88 = Sneden and Crocker (1988); LB85 = Luck and Bond (1985)
5 Transformation from b-y according to Crawford (1995)

Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 1997

Online publication: April 8, 1998
helpdesk.link@springer.de