next previous
Up: Evidence for chemical evolution galaxies


   
2 Sample selection and observations

The data set used in this study combines spectra of high-redshift galaxies observed with the ESO VLT and spectra of local starburst galaxies taken from the IUE archive. 51 of the high-redshift spectra were selected (according to the criteria listed below) from about 300 low-resolution spectra observed mainly during the spectroscopic observing runs of the FDF program (in 3 nights in Sep. and Oct. 2000 and 3.5 nights in Oct. and Dec. 2001). For these observations we used FORS1&2 at the VLT in MOS and MXU mode with a slit width of 1 $^{\prime\prime}$ and the 150I grism. (For instrumental details see the FORS Manual at the ESO web page www.eso.org). A few additional spectra had already been obtained during the commissioning phases of FORS1&2 in 1998, 1999 and 2000 using the same spectroscopical setup as described above. All spectra cover a spectral range (in the observer's frame) from about 3400 Å to about 10 000 Å with a spectral scale of 5 Å/pixel and a spectral resolution of about 200. Although spectra of galaxies as faint as I = 26.0 mag have been observed successfully in the FDF, in the present investigation only galaxies with I $ \leq 24.5$ mag were included. Depending on the objects' magnitude and the seeing conditions ( $0\hbox{$.\!\!^{\prime\prime}$ }7$ on average) the integration times ranged between 2 and 12 hours. The data reduction (bias subtraction, flatfielding, cosmic ray elimination, sky subtraction, wavelength calibration, etc.), was performed using standard MIDAS routines. A detailed description of the FDF spectroscopic observing program and the data reduction procedures will be presented by Noll et al. (2002).

With respect to the photometric redshift catalogue of the FDF (see Bender et al. 2001) our spectroscopic sample is complete to about 85% for our limiting magnitude (I $ \leq 24.5$ mag) for photometric redshifts in the range $2.2 < z_{{\rm phot}} < 3.5$. The distribution of the spectroscopic redshifts is in good agreement with the photometric redshift distribution of the FDF which has peaks at redshifts of around 2.4 and 3.4 (see Fig. 3 in Bender et al. 2001). Exceptions are the redshifts in the range $1.4 \leq z \leq 2.2$ where a lack of strong spectroscopic features in our observed wavelength range makes a reliable spectroscopic redshift determination rather difficult, resulting in an artificial low number of objects in our spectroscopic sample.


  \begin{figure}
\par\includegraphics[width=17cm,clip]{ms2802f1.eps}
\end{figure} Figure 1: Examples of the low-resolution galaxy spectra obtained with FORS. The redshift of the objects is increasing from panel a) to c). The dotted line indicates the noise level which (due to the night sky spectrum and the wavelength dependent instrumental efficiency) varies with wavelength. The S/N of these spectra is about $\le $16, 43 and 12 for object 7029, 5903 and 6063, respectively. The object designations correspond to the catalog of Heidt et al. (2002). For orientation the expected positions of selected spectral lines are indicated by vertical dashed lines. The two apparent emission features (marked by arrows) longward of the C  IV absorption feature are artifacts and resulted from the increased noise at this wavelengths due to telluric OH band edges and atmospheric absorption. For the same reason Fig. 1c becomes unreliable beyond 7000 Å.

For the present investigation we selected those FDF galaxies showing absorption line spectra with an adequate S/N (>10 per resolution element) for a meaningful quantitative analysis of the C  IV resonance doublet. All these galaxies show typical starburst characteristics in their spectra such as intense (rest frame) UV continua and highly ionized metal absorption lines. Three examples of FDF spectra are displayed in Fig. 1. The 10 high-redshift galaxies which are dominated by their Ly$_{\alpha}$ emission were not included in our study. This leads to a limitation of the sample to galaxies with z < 4, since most of the few FDF galaxies with larger redshifts (up to 5.0) observed so far show strong Ly$\alpha $ emission (while pure absorption line spectra dominate at lower redshifts; see Noll et al. 2002). Also excluded from our study were 4 objects that show absorption lines with clear emission components, forming P-Cygni profiles. (The apparent emission peak redwards from the C  IV absorption feature in Fig. 1c is not a P Cygni emission component. The absorption line is unshifted and, as pointed out in the caption, the apparent emission component is an artifact.)

In order to enlarge our sample somewhat we added 6 additional FORS spectra (matching the criteria listed above) which had been observed with the same setup during the FORS commissioning runs (and are now available from the VLT archive). Four of these additional spectra were selected among the gravitationally amplified galaxies behind the cluster 1E0657-558 (Mehlert et al. 2001). We further included two spectra from the HDF-S and AXAF Deep Field follow-up studies (Cristiani et al. 2000), which met our criteria.

The comparison sample of IUE low resolution spectra from the IUE archive[*] consists of 36 local ( $z\approx 0$) starburst galaxies investigated by Heckman et al. (1998)[*]. All these IUE spectra were obtained with the Short Wavelength Prime Camera in the low dispersion mode and, therefore, cover a similar rest-frame spectral range with a slightly better spectral resolution as our FORS spectra of high redshift objects. The IUE spectra were reduced using the pipeline provided by the archive and smoothed to attain the same spectral resolution as the VLT spectra.


next previous
Up: Evidence for chemical evolution galaxies

Copyright ESO 2002