4. Results and discussion
4.1. Comparison of model calculations with DLA observations
In Fig. 7a-h we present abundances in DLA systems for eight elements Fe, Si, Zn, Cr, Ni, S, Al, and Mn. The data are compiled from the 29 papers listed in Table 3. Reliable and less reliable data are marked with filled and open circles, respectively. For several abundances the authors report only upper or lower limits which are indicated by open triangles pointing downwards or upwards, respectively. Those data, of course, are classified as less reliable .
In Figs. 7a to 7d some pairs of data points representing abundances measured for the same DLA system by two different authors are connected by heavy vertical lines. They show how abundance measurements by different authors still can differ and give us a means to estimate the observational errors. In Fig. 7a the DLA system at has two reliable values differing by about 0.2 dex in [Fe/H] and it lies just beneath the Sd curve. The second set of independent measurements in Fig. 7a belongs to the DLA system with . The lower value is indicated as a lower limit which is in agreement with the larger reliable value. Generally reliable measurements of different authors agree to within dex. Errors reported by a few authors for some (mostly reliable ) data are indicated as weak vertical lines in Fig. 7 and likewise are about dex.
The model curves in Fig. 7 represent the redshift evolution of Sa and Sd type galaxies as calculated from our chemically consistent chemical evolution models (heavy lines) and models using exclusively input physics of solar abundance (weak lines). In the legend they are indicated as "(chem.cons.)" and "(solar)", respectively. Sb and Sc galaxies are omitted to avoid overcrowding of the figures. Their curves always lie between those of Sa and Sd as demonstrated in Fig. 3. Furthermore, for the chemically consistent calculations, we also present the redshift evolution of element abundances using Woosley & Weaver's SN II yields from their model C, which has larger explosion energies (cf. Sect. 2.2), indicated as "model C" in Fig. 7. Model B curves are omitted because they always fall between our heavy lines (which use yields from model A) and the curves for model C (cf. Fig. 4, Sect. 2.3).
4.2. General implications for the models
For all eight elements under consideration almost all data points lie between our chemically consistent model curves for Sa and Sd galaxies. We find particularly good agreement between models and observations for the elements Zn and Ni where many observations are available and for Al, Mn and S with a smaller number of (reliably ) observations. Having in mind the fact that Sb and Sc models lie within the region outlined by the Sa and Sd curves (cf. Fig. 3 in Sect. 2.3) we can establish nearly perfect conformity between element abundances observed in DLA systems and our model calculations for spiral galaxies spanning the whole redshift range from 0 to 4.5. And since our models for z = 0 agree well with observed average ISM abundances of nearby spiral galaxies it is clear that DLA galaxies may well evolve into the full range of present-day spiral galaxies , although we cannot exclude the possibility that a few DLA systems might be LSB galaxies or (starbursting) dwarfs.
4.2.1. Implications for chemically consistent models
Differences for chemical consistent vs. pure solar metallicity models on the one hand and different SN II explosion energies (model A vs. model C by Woosley & Weaver 1995) on the other hand are small compared to differences due to the variation of the star formation rate characterizing our spectral galaxy types Sa Sd. Consequently none of the models can be excluded or is clearly favored by comparison with the DLA data. They all are in conformity with the observations. The range of dex of element abundance in DLA systems at any given redshift is naturally explained by the range of star formation rates among early to late type spirals as outlined in Sect. 2.1.
4.2.2. Scatter in observational data
Additionally, some observational scatter is expected in DLA element abundances: The column densities observed in DLAs could depend upon the (unknown) impact parameter (if abundance gradients already exist in (proto-) galactic disks at high redshift), on inclination effects, and on local inhomogeneities along the line of sight (cf. the different abundances determined for the cold and warm disk component of our Galaxy (Savage & Sembach 1996)).
4.2.3. Observed abundances exceeding our Sa model
In case of Fe, Si, Cr, Zn and S four reliable abundance measurements clearly exceed the values calculated for our Sa model. The corresponding DLAs are listed in Table 4 and the elements with abundance in excess of our Sa model prediction are marked with "X" ("O" indicates conformity of observations with models and for "-" no observations are available). The small error bars attached to the data points indicate that these deviations cannot be due to observational errors.
It is seen that in two of the four cases (QSO 0216+0803 and 0528-2505) only the typical SNII elements Si and S show abundances higher than those of our Sa model whereas the iron group elements Fe, Zn and Cr which have important SNI contributions are not enhanced. For these two DLAs a temporarily enhanced SFR or a small star burst in an early type spiral galaxy could easily explain their abundance pattern.
While, of course, we cannot exclude that some dwarf or LSB galaxies may also be present among the DLA absorber sample, our models indicate that the bulk of DLA abundances and, in particular, their redshift evolution, are consistent with them being normal spiral galaxy progenitors.
For the other two systems in QSO 2206-199 and 0201+365 both the SNII product Si as well as elements with important SNI contributions like Fe have abundances higher than those of our Sa model. For these we conclude that the enhancement probably is not due to a temporarily enhanced SFR or star burst but rather to a characteristic timescale of star formation shorter than the adopted for our Sa model. Our models cannot tell if these two DLA systems are the progenitors of S0 galaxies or of a bulge component, which both are believed to form the bulk of their stars on a short timescale. Alternatively the high abundances might result from a very early star formation enhancement, a formation of these systems at or a local overabundance where the line of sight is passing through the disk.
4.2.4. Chemically consistent versus purely solar models
In the case of Zn, Ni and Al we find a significant difference between chemically consistent models and the comparison models using solar metallicity stellar yields only. In Fig. 7c and 7e, i.e. for Zn and Ni, a considerable number of observations lie below the Sd curve (dashed line) of the solar metallicity model (weak line) but within the region outlined by our chemically consistent models (heavy line). In the case of Al (Fig. 7g), almost all data points lie between the chemically consistent evolution models (heavy line), but most of the observed abundances drop below the Sd curve of the solar metallicity model.
To conclude: For some elements the chemically consistent models do not differ very much from the solar metallicity ones, for other elements they do yield significantly lower abundances, and in these cases the chemically consistent calculations do fit the observations much better than the solar metallicity models.
4.3. Discussion of model parameters IMF and SFR
In Fig. 8 we present observed Fe abundances [Fe/H] in DLA galaxies together with model calculations using different IMFs (Scalo vs. Salpeter) and for Sd galaxies different star formation rates (SFR). Resulting curves using a Salpeter IMF are plotted with weak lines whereas those for the Scalo IMF are displayed by heavy lines as before. We find that abundances calculated with a Salpeter IMF are larger than those resulting from models using a Scalo IMF (cf. Fig. 6 in Sect. 2.3.5). Some large [Fe/H] values observed in DLA systems can now be reached with the Salpeter IMF but then, in turn, some low abundance data fall outside the region outlined by our models. This drawback can be compensated by lowering the star formation rate of Sd galaxies. Because the SFR of Sd galaxies is constant this can be done without changing the photometric properties of the galaxies.
It should be mentioned that while the SFHs of nearby galaxies seem to be very homogeneous for early type spirals the scatter around the average SFH used in our models significantly increases towards later types. This can be seen e.g. in the very small range of equivalent widths or colors among Sa galaxies as opposed to the much larger scatter both in equivalent widths and colors seen among local Sd galaxies (cf. Kennicutt & Kent 1983, Buta et al. 1994).
Very small SFRs are characteristic for low surface brightness galaxies (LSBs) and hence we conclude that a few observed DLA systems with particularly low abundances could be either late type spirals with a lower than average SFR or else LSB galaxies. As discussed in the previous subsection an extremely low element abundance measured in DLA systems could also be due to a large (unknown) impact parameter.
© European Southern Observatory (ESO) 1999
Online publication: December 16, 1998