4 The dependence of the ultraviolet spectral energy distribution on $\vec{T}_\mathsf{eff}$, log$\vec{g}$ and metallicity

As was shown by Holweger et al. (1994) and by Allard et al. (1998), the ultraviolet flux of the metal-poor A-type stars in the region 1250-1900 Å is strongly dependent on the model parameters. Castelli & Kurucz (2001) showed the dependence of the computed fluxes on the parameters for no $\alpha$-enhanced models and for models computed for the particular abundances of $\lambda$ Boo. In this paper we show the dependence of the computed ultraviolet fluxes on the parameters for the $\alpha$-enhanced models. Figures 3a and 3b illustrate, for the range 1250-3000 Å, the variation of the synthetic fluxes as a function of effective temperature and gravity, respectively, when ODFs computed for $\rm [M/H]=-1.50$a are used. The fluxes computed for metallicities ranging from -2.50a to -1.00a at steps of 0.5 dex, and $T_{\rm eff}$=8500 K, $\log\,g$=3.0 are shown in Fig. 3c for the same wavelength range. In each panel of Fig. 3 the fluxes are normalized to 5556 Å in order to be consistent with the usual comparison between observed and computed fluxes. All the models displayed in Fig. 3 are computed with ODFs corresponding to a microturbulent velocity $\xi =2$ km s^-1.

Figure 3 points out the different behaviour of the energy distribution in the IUE short- and long-wavelength regions. In the range 1250-2000 Å the energy distribution depends on $T_{\rm eff}$, on $\log\,g$, and also on the metallicity owing to the C I and Si I discontinuities at 1444 Å and 1525 Å respectively. On the contrary, the energy distribution in the 2000-3000 Å region weakly depends on the metallicity and it is hardly useful to fix both $T_{\rm eff}$ and $\log\,g$ at the same time, owing to the similar dependence of the energy distribution on $T_{\rm eff}$ and on $\log\,g$.

The comparison of Fig. 3 with Fig. 7 in Castelli & Kurucz (2001) indicates that the general behaviour is the same for the two sets of models, which were computed with different metallicities. The differences are mostly related with the different carbon and silicon abundances adopted in the model computations. For instance, when Fig. 3 is compared with Fig. 7 of Castelli & Kurucz (2001), it shows that the 0.4 dex larger silicon abundance and the -1.1 dex lower carbon abundance, adopted in this paper, increase the Si I discontinuity at 1525 Å and decrease both the C I discontinuity at 1444 Å and the intensity of all the C I lines, in particular that of the blend observed at 1657 Å.

Figure 3: The dependence of the ultraviolet flux on the model parameters in the region 1250-3000 Å. a) $\log\,g$=3.0, $\rm [M/H]=-1.5$a and $T_{\rm eff}$=9000 K, 8750 K, 8500 K, and 8250 K. b) $T_{\rm eff}$=8500 K, $\rm [M/H]=-1.5$a, and $\log\,g$=2.50, 3.00, 3.50, and 4.0. c) $T_{\rm eff}$=8500 K, $\log\,g$=3.00, and $\rm [M/H]=-2.5$a, -2.0a, -1.5a and -1.0a. The ordinate is the absolute flux $F_{\lambda }$ at the star surface. The fluxes are normalized at 5556 Å.