Implicit in the previous discussion is the assumption that the "birth of a galaxy" is operationally equivalent to the epoch of its first major burst of star formation. This is usually the first chance we have to detect objects which will eventually evolve into modern galaxies. However, damped Ly absorbtion systems (DLAs), detected as neutral hydrogen troughs (NHI ~ 1020 - 23 cm-2) in the spectra of high-z quasars, detect baryonic concentrations independent of star formation or nonstellar nuclear activity. DLAs contain most of the H I in the universe.
Imaging searches for galaxies associated with DLAs are not new, but they have been almost impossible to carry out from the ground . They are best performed in the near-IR, where the DLA starlight should be relatively bright compared to the background quasar. Unfortunately, a large enough H I column density to be detected as a DLA is only expected to occur at small impact parameters through a galaxy (or protogalaxy) - within 10 or 20 kpc of its center (). This corresponds to a separation between the galaxy and the line-of-sight to the background quasar of less than 1 to 2 ". Detecting a faint galaxy this close to a bright source is exceedingly difficult from the ground. Adaptive optics is not yet the solution (beset by poor Strehl ratios). Proposed interferrometric techniques are more suitable for detecting a compact source at a known position angle with respect to the quasar (and still the required dynamic range may be unattainable). Ground-based imaging can discover DLA galaxies only at large impact parameters, greater than 25 kpc (for Ho = 75). To do better, an extremely sharp and stable PSF is required, preferably near the diffraction limit. This means going into space.
We therefore used the NICMOS-2 camera on the HST in coronographic imaging mode to search for near-IR continuum emission from starlight associated with two dozen DLAs. Our results are shown in Figure 14. Despite the approval of 95 targets, only two dozen quasar observations were attempted, and many were not as sensitive as they could have been. This low success rate was because: 1) the coronograph was not fully operational during the first half of Cycle 7, 2) SNAPSHOT proposals utilizing the coronograph were not scheduled for a few months after the coronograph became operational, and 3) failure of the coronograph to autonomously acquire several of the quasars when they were scheduled resulted in a loss of those observations. Despite these difficulties, we detected two likely DLA galaxies of ~ 0.7L, and obtained upper limits for nearly twenty DLAs (). Our 5 upper limits for the continuum fluxes of the DLAs are plotted in terms of L* versus redshift in Figure 14. Note that the vertical axis has logarithmic luminosity increasing downwards-higher observations are more sensitive to fainter galaxies. The only detections of possible DLA galaxies are shown by the two asterisks. The lowest curve is our measured sensitivity for Cycle 7; filled triangles represent direct imaging, while open triangles represent coronographic observations. The middle curve shows our predicted sensitivity in a 1300-second NICMOS integration, a half magnitude gain over our current limits. Finally, the upper curve is the predicted sensitivity of for "limiting" 10,000-second exposures, would push detection limits 2 mags fainter.
Figure 14. Upper limits to the 1.6 µm continuum emission from starlight associated with Damped Ly Absorbers, from Colbert and Malkan (). Since they are all limits (except for the two plausible detections shown by asterisks), they are pointing upwards (towards less luminous galaxies). These upper limits would be a factor of two tighter for any galaxies more than 1" away from the quasar (ie., they understate the true sensitivity of the observations). The dashed, dotted and solid lines illustrate typical 5 - upper limits obtained in our Cycle 7 program, and future possible NICMOS observations with somewhat longer, and much longer integration times. For twenty observed DLA systems, our HST coronographic imaging rules out the possibility of a galaxy having typically a luminosity of 0.7-1.0 L*.
There were only two plausible DLA detections and the best upper limits were H ~ 21.5 mag. When we account for the fact that more luminous galaxies are likely to have larger HI cross sections, we would have expected to detect 5 or 6 of the galaxies. If the DLAs were drawn from the general galaxy population, we should have detected 2 or 3 with 1.0L* or brighter, instead of zero. This discrepancy indicates that the luminosity function of DLAs is not consistent with the normal galaxy population. The objects associated with DLAs are on average fainter than normal galaxies we measure at either z = 0 or 3. Deeper NICMOS/HST imaging can test the possibility that they are produced by dwarf galaxies such as the Magellanic Clouds.
If deeper proposed NICMOS imaging of DLA systems continues to set even stricter upper limits on their continuum emission, it may be that DLAs trace a substantial baryon reservoir not associated with normal galaxies.
I thank James Colbert, Harry Teplitz and Erin Hicks for some of the figures from our ongoing research collaborations.