4 Possible explanations for unusual abundance ratios

The results of Fig. 4 are striking. It is particularly noteworthy that both DLA A and B exhibit unusually low $\alpha$/Fe-peak ratios, as does the DLA in the group towards Q0201+1120 (Ellison et al. 2001). However, before drawing the conclusion that this is evidence that the environment has a significant impact on star formation histories at $z \sim 2$, we explore other factors that may affect our abundance determinations.

We first consider the possibility of partial photo-ionization. This is unlikely to be an issue for the relatively high N(H I) DLA A, but may have an effect on DLA B. Evidence that low N(H I) DLAs and sub-DLAs may be increasingly affected by photo-ionization comes from the increasing fraction of N(Al III)/N(Al II) found by Vladilo et al. (2001). However, the upper limit we determine for the ratio of N(Al III)/N(Al II) < -0.91 is significantly lower than that predicted for a log N(H I) = 20.1 from the trend found by Vladilo et al. (2001). This provides direct evidence that photo-ionization does not have a significant impact on these abundance determinations. Specifically, in the case of [O/Fe], the calculations of Vladilo et al. (2001) show that Fe II is converted to Fe III as effectively as O I is ionized to O II, so that partial ionization is highly unlikely to be the reason for the low O abundance. In fact, according to Vladilo et al. (2001), a ionization correction would have a greater effect on [Si/Fe] pushing it to a lower value and enhancing the effect seen in Fig. 4.

Next, we consider the effects of saturation. As mentioned in the previous section, it is plausible that the O I $\lambda$1302 transition is mildly saturated, requiring an upward revision of N(O). In fact, it is possible to achieve a fit with the same $\chi^2$ statistic if N(O) is increased by 0.4 dex and the b-values allowed to vary freely. A reduction of the b-values by only $\sim$1.0 km s^-1 in the two strongest components is sufficient to achieve this. The additional corollary of increasing N(O) is to bring the O/Si ratio in closer agreement with solar values. On the other hand, the high spectral resolution of these data make the possibility of "hidden" saturation in other absorption lines highly unlikely. Indeed, the excellent agreement between column densities of Fe and Ni determined by using transitions with different oscillator strengths shows that there is no hidden saturation effect for the other absorption lines used here.

Finally, we consider the possibility of atypical ammounts of dust depletion. The low [S/Fe] abundance shown in Fig. 4 may be due to anomalously low amounts of dust in these DLAs, as supported by the low [Zn/Cr] ratio in DLA A. However, dust is unlikely to be the explanation for the low [Si/Fe] ratios, since this would require that Si be more depleted than Fe, contrary to what is observed locally (Savage & Sembach 1996). The middle panel of Fig. 4 is therefore the most convincing evidence that the intrinsic abundances of this double DLA are anomalous compared with other systems in the literature.