3.2. Details of the WFPC2 F410M search for Ly emitting z 2.4 candidates
The HST data are presented in the color-color diagram of Fig. 4, which shows photometry for 115 objects detected simultaneously in the Cycle 5 broad-band WFPC2 filters V606 and B450, and in the medium-band filter F410M. Results from the two Cycle 6 HST F410M fields are also plotted in Fig. 4. (For consistency with the Cycle 5 F410M field, the (B450-V606) colors were interpolated from the (B450-I814) colors using the effective central wavelengths of the three broad-band filters involved, following Ho95b). As with the ground-based data, the F410M passband is contained entirely inside the B450 filter, so that proper continuum subtraction is possible. At z 2.39, these bands sample the emitted wavelengths of 1200-2100 Å, where the spectra of young star-forming galaxies are relatively flat (W91). The solid line in Fig. 4 shows the expected relation for objects with a flat power-law spectrum (F -) across these three adjacent filters. Indeed most of the general field objects are distributed around it. There is a group of 18 objects which are located 2 below this line, showing that they have significant emission in F410M, which is likely Ly emission at z 2.39, as argued by P96b and below. Eight of these z 2.4 candidates are spectroscopically confirmed with the MMT and the KPNO 4m (P96a, P96b, Co97, K97), out of ten objects for which spectra of high enough quality were obtained, indicating this method of finding compact z 2.4 candidates is at least 70% reliable. Apart from the three brighter z 2.4 objects, which were easily seen in ground-based photometry (P96a) and contained weak active galactic nuclei (AGN), most of the sample has Ly emission with average restframe equivalent widths (estimated from the F410M photometry) of only 40-50 Å - more typical of ionization arising purely from star formation (cf. S96a, S96b).
Figure 4. (F410M-B450) versus (B450-V606) color-color diagram from 51 WFPC2 orbits on both the Cycle 5 53W002 field (circles, P96b) and the two Cycle 6 parallel F410M fields (squares, P97). The solid line shows the expected relation for a featureless power-law (F -) labeled with values of , around which most of the general field objects (open symbols) are distributed. Approximate 2 error boundaries for the WFPC2 photometry are indicated by the dotted lines. Error bars are plotted only for objects 2 below the power-law line (filled symbols), which are significant candidate Ly emitters at z 2.39. The five large filled triangles are spectroscopically confirmation of P96b at z 2.39. Approximate implied rest-frame Ly equivalent widths are indicated on the right-hand vertical axis. Most z 2.4 candidates have WLy < 100 Å.
Two open circles in the region below the lower 2 error boundary populated by the z 2.4 candidates were spectroscopically confirmed not to be at z 2.39 but are stars (M stars can turn out to be apparent F410M-emitters given their complicated spectra around 4000 Å). Another ~ 250 objects were detected in V606 and B450 (with B450 26.0 mag), but not in the F410M image. They have 2 lower limits in (F410M-B450) -0.2 mag and are not plotted here. For 26.0 B450 27.5 mag, the B450-band sample is 90% complete (O96), but the underexposed F410M image does not provide useful upper limits for such faint objects. The Cycle 6 fields are shallower, and have fewer objects, but have similar fractions of lower limits in F410M-B450 ( 0.0 mag). These limits are indicated by the dotted arrows at the right axis of Fig. 4. Both the observed and the restframe ultra-violet reddening vector expected at z 2.39 are indicated by the arrows, and suggest that the few reddest candidates may have a visual absorption of AV 2-3 mag. Most of the compact z 2.4 Ly emitting candidates, however, have rather blue colors and are likely not very reddened internally.
The differential volume element at z 2.39 is ~ 20x larger than at z 0.097, making it less likely that F410M traces [O II] at z 0.097 than Ly at z 2.4. Any other emission lines between 1216 Å and 3727 Å strong enough to be detected (e.g., C-IV or Mg II) arise from objects known to have much lower surface densities (e.g., powerful luminous radio galaxies or QSOs). No other strong emission lines exist in this range for star-forming objects (K96), implying that most of the 18 significant candidates are likely at z 2.39. The reliability of the 18 z 2.39 candidates (Fig. 4) is further strengthened by the fact that three of the spectroscopically confirmed members are at the lower boundary in the (Ly410-B450) color of the general field population.