4 Iron abundances

Fe abundances were derived using a rather large number of lines, typically $\sim$20 and $\sim$30-40 for TO-stars and subgiants respectively. Full details about the line list, including gf's, EWs and abundances from individual lines, will be given in separate papers (in preparation).

The Fe abundance obtained for dwarfs in NGC 6397 ( ${\rm [Fe/H]}=-2.02\pm 0.01$, where this error bar is the internal error as given by the standard deviation of the mean) is in good agreement with that determined from subgiants ( ${\rm [Fe/H]}=-2.05\pm 0.03$). The average of the two groups is ${\rm [Fe/H]}=-2.03\pm 0.02$ (this is the internal error of our analysis, obtained by combining results for TO-stars and subgiants; systematic errors, within the abundance scale defined by observations of the field stars, are mainly due to uncertainties in the adopted temperature; again combining results for TO stars and subgiants, that may be considered as independent each other, they are $\pm 0.04$ dex). This value is less than that derived from giants in Carretta & Gratton (1997): ${\rm [Fe/H]}=-1.82\pm 0.04$), and Zinn & West (1984: ${\rm [Fe/H]}=-1.91$). However it agrees very well with the value obtained by Minniti et al. (1993: ${\rm [Fe/H]}=-1.96\pm 0.04$), and the recent, comprehensive analysis of giants and subgiants by Castilho et al. (2000: $-2.00\pm 0.05$).

The equilibrium of ionization is not well reproduced: abundances from neutral Fe lines are 0.11 dex larger than those from singly ionized Fe lines. The same result is obtained for NGC 6752 and our field stars. Note that a smaller difference (0.07 dex) in the same sense is also present in our solar reference analysis (in that case we obtained $\log n({\rm Fe})=7.52$ from Fe I lines, and 7.45 from Fe II lines using the solar model atmosphere from Kurucz 1993, with no overshooting). We then think most of the difference for the Sun is due to either the adopted gf's values (these are laboratory values taken from recent literature) or to the model atmospheres (models might underestimate the temperature gradient in real atmospheres, perhaps due to an inappropriate consideration of convection). The residual difference for the program stars might be due to the adoption of slightly too high $T_{\rm eff}$'s ($\sim$40-50 K) or too low gravities (by $\sim$0.1 dex). Note that it cannot be due to departures from LTE, because the expected dominating effect (overionization: Idiart & Thévenin 1999; Gratton et al. 1999) would lead to larger abundances from Fe II lines than from Fe I ones (opposite to observations).

The star-to-star scatter in [Fe/H] values is extremely small: the rms scatter is only 0.04 dex (i.e. 10%) for NGC 6397. This seems a very homogeneous cluster as far as Fe abundances are considered.

Figure 4: Spectral region including the 8183-94 Å Na I doublet in NGC 6752 TO-stars (stars are ordered according to decreasing Na abundances). The position of the Na lines is marked. Note that all these stars essentially have the same temperature, gravity, overall metal abundance and microturbulent velocity, so that observed variations in the line strengths can be directly interpreted as spread in the abundances

Figure 5: Same as Fig. 3, but for the region including the 7771-74 Å OI triplet in NGC 6752 TO-stars (stars are ordered according to decreasing Na abundances). The position of the O lines is marked

On the other side, the [Fe/H] value for NGC 6752 ( ${\rm [Fe/H]}=-1.42\pm 0.02$, internal error; systematic error is again $\pm 0.04$ dex), obtained both from TO and subgiant stars, which agree completely, coincides with that derived from giants by Carretta & Gratton (1997: ${\rm [Fe/H]}=-1.42\pm 0.02$); it is somewhat larger than the value quoted by Zinn & West ( ${\rm [Fe/H]}=-1.54$) and Minniti et al. (1993: ${\rm [Fe/H]}=-1.58\pm 0.04$). The spectra of NGC 6752 have a S/N lower than those in NGC 6397, since we chose to observe more stars, even at a lower S/N. The scatter of abundances for individual lines (from 0.12 to 0.27 dex) is larger than that obtained for stars in NGC 6397, roughly in agreement with the lower S/N.

For field stars we may compare the present Fe abundances with those derived by Carretta et al. (2000). Limiting ourselves to only those stars for which Carretta et al. considered high dispersion abundances, the present Fe abundances are smaller on average by $-0.05\pm 0.02$ dex (11 stars, ${\rm rms}=0.08$ dex). The slightly lower metal abundances are due to lower $T_{\rm eff}$'s adopted in the present paper.