4.4. Serendipitous Long-Slit Searches

Deep spectroscopy alone is also an efficient means to detect Ly emission from very high-redshift systems (see Fig. 8). Long spectroscopic integrations are sensitive to line-emitting sources which serendipitously fall within the slit, potentially out to z 6.5. This limit is set by the plummetting response of CCDs in the near-infrared and to some degree by the strong OH sky emissions at > 9300 Å; infrared spectrographs can potentially extend these surveys to still higher redshifts. The first confirmed object at z > 5 was the result of a serendipitous detection (Dey et al. 1998). Hu et al. (1998) also report the serendipitous detection of an isolated emission-line source which they interpret as Ly at z > 5. Serendipitous long-slit searches are fully complementary to narrowband work: whereas narrowband imaging probes a thin shell of redshift space, deep spectroscopy (admittedly over a smaller solid angle) probes a pencil beam of look-back time. The resolution of low-dispersion optical spectrographs are also better matched to typical Ly line widths than filters with widths of 3000 km s^-1.

Figure 8. Two-dimensional spectrogram of serendipitously discovered strong line emitters in the SSA 22 field. The source SSA 22-C17 is a Lyman-break galaxy at z = 3.299 discovered by Steidel and collaborators. During deep, moderate-dispersion Keck/LRIS follow-up spectroscopy of that source, two strong line emitters were serendipitously discovered on the same multislit slitlet. The high equivalent widths, lack of secondary emission features, and narrow velocity width of the lines argue that these are high-redshift Ly emitters. Figure courtesy Manning et al. (2000).

At Berkeley, we are conducting a careful analysis of our deepest archival spectroscopic exposures on the Keck telescopes obtained in good meteorological conditions (see Fig. 8). In a 1.5 hr spectrogram at moderate dispersion ( / 1000) with the LRIS camera, the limiting flux density probed for spectrally unresolved line emission in a 1 arcsec² aperture is S_lim(5 ) 1 × 10^-17 ergs cm^-2 s^-1 at 9300 Å (Stern et al. 1999a). The limit is strongly wavelength dependent. A single slit-mask exposure typically covers 600 arcsec², or one-sixth of a square arcminute. In a year, up to 2 arcmin² can be covered, implying reasonable statistics on the surface density of high-redshift line emitters, only somewhat compromised when portions of the mask are dedicated to photometric high-redshift candidates or when observations target rich clusters. This work is in progress (Manning et al. 2000) and the reader will appreciate that distinguishing Ly serendips from lower redshift interlopers can be challenging and is best done with follow-up spectroscopic and/or deep multiband imaging observations. Our observations typically are biased toward redder wavelengths, which has the unfortunate consequence that we are not sensitive to z ~ 3 serendips and thus cannot compare our results to well-documented results of Steidel and collaborators at that redshift interval (e.g., Steidel et al. 1996a, 1996b, 1999; Dickinson 1998). We are currently emphasizing serendipitous Ly-emitting galaxies at 4.5 z 5.5. We have four reasonable candidates in the 1.6 arcmin² analyzed thus far, not including candidates behind the Abell 2390 cluster. One source is confirmed at z = 5.34 (Dey et al. 1998). We also have a few candidates at z > 5.5 which are not included in the current discussion. The measured surface density is 2 ± 1 candidate Ly emitters arcmin^-2 at z = 5. Their continua are estimated to be quite faint, I 26.5 (AB). Serendipitously identified absorption-break galaxies at high redshift are also possible, but difficult at the faint levels we now probe.