2.4. Element Ratios
So far we have considered only the overall metallicity of DLAs as measured by the [Zn/H] ratio. However, the relative abundances of different elements offer additional insights into the chemical evolution of this population of galaxies, as we shall now see. This aspect of the work has really blossomed with the advent of efficient echelle spectrographs on both the Keck and VLT facilities, which have allowed the absorption lines of a wide variety of elements to be recorded simultaneously, often with exquisite precision.
The presence of dust in DLAs can be inferred by comparing the gas phase abundances of two elements which in local interstellar clouds are depleted by differing amounts. The [Cr/Zn] ratio is one of the most suitable of such pairs for the reasons described above. It became apparent from the earliest abundance measurements in DLAs that this ratio is generally sub-solar, as expected if a fraction of the Cr has been incorporated into dust grains. Figure 10 shows this result for a subset of the DLAs in Figure 7; similar plots are now available for larger samples of DLAs and for other pairs of elements, one of which is refractory and the other is not (e.g. Prochaska & Wolfe 1999; 2002).
 |
Figure 10. (Reproduced from
Pettini et al. 1997a).
Cr abundance relative to Zn in 18 damped
Ly |
From this body of data it is now firmly established that the depletions of refractory elements are generally lower in DLAs than in interstellar clouds of similar column density in the disk of the Milky Way. The reasons for this are not entirely clear. The question has not yet been addressed quantitatively; qualitatively the effect is probably related to the lower metallicities of the DLAs and the likely higher temperature of the interstellar medium in these absorbers (Wolfire et al. 1995; Petitjean, Srianand, & Ledoux 2000; Kanekar & Chengalur 2001). Figure 10 does seem to indicate a weak trend of decreasing Cr depletion with decreasing metallicity, also supported by the results of Prochaska & Wolfe (2002).
Typically, it is found that refractory elements are depleted
by about a factor of two in DLAs - a straightforward average of
the measurements in Figure 10 yields a
mean <[Cr/Zn]> = - 0.3+0.15-0.2
(1
limits).
When we combine this value with the mean metallicity of DLAs,
[<Zn / HDLA>] = - 1.13,
or <ZDLA> = 1/13
Z
,
we reach the conclusion that in damped systems the "typical"
dust-to-gas ratio is only about
1/30 of the Milky Way
value (although there is likely to be a large dispersion
from DLA to DLA, reflecting the range of metallicities
evident in Figure 7).
In the disk of our Galaxy, there a well determined relationship between the
neutral hydrogen column density and the visual extinction,
<N(H I)> / <AV> = 1.5 ×
1021 cm-2 mag-1
(Diplas & Savage
1994),
where AV is the
extinction (in magnitudes) in the V band. For the typical damped
Ly
system
with neutral hydrogen column density N(H I) =
1 × 1021 cm-2
and dust-to-gas ratio 1/30 that of the local ISM,
we therefore expect a trifling
AV 
0.02 mag in the rest-frame V band. Of more interest is the
far-UV extinction, since this is the spectral region observed at optical
wavelengths at redshifts z = 2 - 3. Adopting the SMC
extinction curve
(Bouchet et al. 1985)
- which may be the appropriate one to
use at the low metallicities of most DLAs - we calculate that a damped
Ly system will typically
introduce an extinction at 1500 Å of
A1500
0.1 mag in the spectrum of a background QSO.
Such a small degree of obscuration is consistent
with the mild reddening found in the spectra of QSO with DLAs,
compared to the average UV continuum slope of QSOs without
(Pei, Fall, & Bechtold
1991).
The moderate degree of depletion of refractory elements in DLAs has motivated a number of attempts to correct for the fractions missing from the gas phase (Vladilo 2002a and references therein) and thereby explore intrinsic (rather than dust-induced) departures from solar relative abundances. The basic idea, which is discussed extensively in the contribution to this volume by Francesca Matteucci, is that different elements are produced by stars of different masses and therefore different lifetimes. Thus, the relative abundances of two elements can, under the right circumstances, provide clues to the previous star formation history of the galaxy, or stellar population, under consideration (Wheeler, Sneden, & Truran 1989). Such clues are not always easy to decipher, however. For one thing, they rely on our incomplete, and mostly theoretical, knowledge of the stellar yields. Secondly, we must assume a `standard' initial mass function (IMF) because, if we were free to alter at will the relative proportions of high and low mass stars, then we would obviously be able to reproduce most element ratios, but we could scarcely claim to have learnt anything in the process. Fortunately, all available evidence (including that from DLAs, Molaro et al. 2001) points to a universal IMF as a reasonable first order approximation (Kennicutt 1998a).
One of the cornerstones of this kind of
approach is the well established overabundance
of the alpha-capture elements relative
to Iron in metal-poor stars of the Galactic halo.
Mg, Si, Ca, and Ti are generally overabundant by
factors of between two and three in stars where Fe is below
one tenth solar, i.e.
[/Fe] = +0.3 to +0.5
when [Fe/H]
- 1.0
(Ryan et al. 1996).
This result can be understood if approximately
two thirds of the Fe (and other Fe-peak) elements
are produced by Type Ia supernovae (SN)
and released into the ISM
with a time lag of about 1Gyr relative
the -capture elements
(and one third of the Fe) manufactured by the massive stars which
explode as Type II supernovae.
In this picture, 1 Gyr is therefore the time over which
the halo of our Galaxy became enriched to a metallicity
[Fe/H] = - 1, ultimately reflecting the
rate at which star formation proceeded in this stellar component
of the Milky Way. Clearly, the situation could be different
in other environments
(Gilmore & Wyse 1991;
Matteucci & Recchi
2001).
The thick disk, for example, evidently reached solar abundances of the
-elements in less than
1 Gyr, since
the overabundance - or
more correctly the Fe deficiency - seems to persist to this high
level of metallicity
(Fuhrmann 1998).
Do damped Ly systems,
which as we have seen are generally metal-poor, show an overabundance
of the elements? This
question has been addressed by several authors and there seems to be
a general consensus that there is not a unique answer.
As can be seen from Figure 11, while some DLAs
conform to the pattern seen in Galactic stars, many others
do not, in that they exhibit near solar values
of [Si/Fe] even when [Fe/H] is << - 1
(Molaro et al. 2000;
Pettini et al. 2000a;
Ledoux et al. 2002;
Prochaska & Wolfe
2002;
Vladilo 2002b).
Presumably, these are galaxies where star formation
has proceeded only slowly, or intermittently,
allowing the Fe abundance to `catch up' with that
of the Type II supernova products. The
Magellanic Clouds may be local counterparts of these DLAs
(Pagel & Tautvaisiene
1998).
Thus, the chemical
clues provided by the these element ratios are another
demonstration, together with the wide range in
metallicity at the same epoch (Section 2.3)
and the morphologies of the absorbers
(Section 2.1),
that DLAs trace a diverse population of galaxies,
with different evolutionary histories. Their common
trait is simply a large cross-section on the sky
at a high surface density of neutral hydrogen.
 |
Figure 11. (Reproduced from Ledoux et al. 2002). Dust-corrected abundance ratios of Si relative to Fe versus DLA metallicity, as measured from the dust-corrected Fe abundances. Errors are typically ± 0.1dex. Different symbols are used for different dust depletion patterns adopted when correcting the observed abundances. The shaded area shows the region occupied by Galactic stars in the disk and halo over this range of metallicities. |
2.4.3. The Nucleosynthesis of Nitrogen
A case of special interest is Nitrogen, whose nucleosynthetic origin is a subject of considerable interest and discussion. There is general agreement that the main pathway is a six step process in the CN branch of the CNO cycle which takes place in the stellar H burning layer, with the net result that 14N is synthesised from 12C and 16O. The continuing debate, however, centres on which range of stellar masses is responsible for the bulk of the nitrogen production. A comprehensive reappraisal of the problem was presented by Henry, Edmunds, & Köppen (2000) who compiled an extensive set of abundance measurements and computed chemical evolution models using published yields. Briefly, nitrogen has both a primary and a secondary component, depending on whether the seed carbon and oxygen are those manufactured by the star during helium burning, or were already present when the star first condensed out of the interstellar medium.
Observational evidence for this dual nature of nitrogen
is provided mainly from measurements of the N and O abundances
in H II regions.
(For consistency with other published work, we depart here from the
notation used throughout the rest of this article,
and use parentheses to indicate logarithmic ratios of number densities;
adopting the recent reappraisal of solar photospheric abundances by
Holweger (2001),
we have (N/H) =
- 4.07; (O/H) =
- 3.26; and
(N/O) = - 0.81).
In H II regions of nearby galaxies, (N/O)
exhibits a strong dependence on (O/H) when the latter is greater
than ~ 2/5 solar; this is generally interpreted
as the regime where secondary N becomes dominant.
At low metallicities on the other hand,
when (O/H) - 4.0
(that is, 1/5
solar), N is mostly primary and tracks O; this results in a
plateau at (N/O) - 1.5
(see Figure 12).
 |
Figure 12. Abundances of N and O in
extragalactic H II regions (small dots) and damped
Ly |
The principal sources of primary N are thought to be intermediate
mass stars, with masses
4
M /
M
7, during the
asymptotic giant branch (AGB) phase.
A corollary of this hypothesis is that the release
of N into the ISM should lag behind that of O which, as we have seen,
is widely believed to be produced by massive stars which explode
as Type II supernovae soon after an episode of star formation.
Henry et al. (2000)
calculated this time delay to be approximately 250 Myr; at low
metallicities the (N/O) ratio could then perhaps be used as a
clock with which to measure the past rate of star formation,
as proposed by
Edmunds & Pagel
(1978).
Specifically, in metal-poor galaxies which have
only recently experienced a burst of star formation one may expect to find
values of (N/O) below the primary plateau
at (N/O)
- 1.5, provided the
fresh Oxygen has been mixed with the ISM
(Larsen, Sommer-Larsen,
& Pagel 2001).
As pointed out by Pettini, Lipman, & Hunstead (1995), clues to the nucleosynthetic origin of nitrogen can also be provided by measurements of N and O in high redshift DLAs. Apart from the obvious interest in taking such abundance measurements to the distant past, when galaxies were young, one of the advantages of DLAs is that, thanks to their generally low metallicities, they probe a regime where local H II region abundance measurements are sparse or non-existent and where the effect of a delayed production of primary nitrogen should be most pronounced.
Figure 12 shows the most recent compilation of
data relevant to this question. The fact that all DLA measurements fall
within the region in the
(N/O) vs. (O/H) plot bounded by the primary
and secondary levels of N production
provides empirical evidence in support of
currently favoured ideas for the nucleosynthesis of
primary N by intermediate mass stars. The uniform value (N/O)
- 1.5
seen in nearby metal-poor star-forming galaxies
can be understood in this scenario if these
galaxies are not young, but contain older stellar populations,
as indicated by a number of imaging studies
with HST.
It is also somewhat surprisingly to find such a high proportion (40%)
of DLAs which have apparently not
yet attained the full primary level of N enrichment
at (N/O) - 1.5.
Possibly, the low metallicity regime - where
the difference between secondary and primary nitrogen enrichment
is most pronounced - preferentially selects
young galaxies which have only recently
condensed out of the intergalactic medium and begun forming
stars. A more speculative alternative, which needs to
be explored computationally, is that at low
metallicities stars with masses lower than
4M may
make a significant contribution
to the overall N yield (Lattanzio et al., in preparation;
Meynet & Maeder
2002).
The release of primary N may, under these circumstances,
continue for longer than 250 Myr, perhaps for a substantial
fraction of the Hubble time at the median <z> = 2.5
of our sample.
In concluding this section, it is evident that DLAs are a rich source of information on nucleosynthesis in the early stages of galaxy formation. Element abundances in DLAs are increasingly being taken into consideration, together with stellar and H II region data from local systems, in models of the chemical evolution of galaxies and in the calculation of stellar yields. The chemical clues they provide will be even more valuable once their connection to today's galaxies in the Hubble sequence is clarified.