Figure 3 shows the distribution of IAB magnitudes of the 13 radio sources in the HDF. Of the seven sources found in the complete sample, all have optical counterparts brighter than IAB = 21.2. At least half of the optical identifications are with spiral or irregular galaxies, many of which appear to be merging or are in small groups. The remaining optical counterparts are composed of nearby field spirals, red ellipticals, and possibly a few late type AGN. There are no identifications with quasars or with galactic stars.
Figure 3. Magnitude distribution for radio sources in the HDF. The black areas refer to the complete sample of seven sources and the lightly shaded areas the six additional radio sources not found in the complete sample. IAB magnitudes are taken from Williams et al. 1996.
Spectroscopic or photometric redshifts are available for all of the galaxies in the HDF with detected radio emission. All have redshifts less than 2.01 and the seven identified sources in the HDF complete sample have redshifts less than 1.01 (Figure 4). Although these microjansky radio sources are a million times weaker than Jansky level sources, such as found in the all sky 3CR or Parkes surveys, it is important to note that the redshift distribution of the microjansky sources does not differ appreciably from that of the strong radio source population.
Figure 4. The redshift distribution for radio sources in the HDF. The black areas refer to the complete sample of seven sources and the open areas the six additional radio sources not found in the complete sample which are identified with galaxies having measured redshifts. Redshifts are taken from the Keck/HDF Consortium (Moustakas et al. 1997).
This is because the radio luminosity function is relatively steep. Unlike the optical galaxy counts which reach to successively more distant galaxies at faint magnitudes, the radio counts, at microjansky levels, sample roughly the same part of redshift space as the strong source samples, but reflect the distribution of a much lower luminosity group of radio sources.
None of the galaxies in the HDF are strong FRII radio galaxies. As shown in Figure 5, the radio luminosity of all of the galaxies in the HDF is less than 6 x 1024 W/Hz at 8.5 GHz, typical of FRI radio galaxies, weak AGN found in some elliptical and Seyfert galaxies, and other galaxies with active star formation. The weakest source in the field, HDF 3648+1427 (we denote radio sources in the HDF and HFFs by their coordinates truncating the 12 hours and +62 degrees), which is identified with a bright (IAB = 18.4) relatively nearby (z ~ 0.02) galaxy which has a luminosity of only 2 x 1019 W/Hz, comparable with that of normal spiral galaxies such as M31.
Figure 5. The luminosity distribution of radio sources in the HDF calculated assuming H0 = 50 km/sec/Mpc and q0 = 1/2. The dark areas refer to the complete sample of seven sources and the open areas to the six additional radio sources not in the complete sample which have measured redshifts.
Only one source in the Hubble Deep Field (HDF 3646+1404) shows clear evidence of variability over the approximately 18 month period covered by our observations. It is a relatively powerful radio source with a luminosity at 8.5 GHz of 3 x 1024 W/Hz and has an unresolved radio core less than 0.1 arc seconds in diameter and a flat radio spectrum. It is identified with a face on Sb galaxy at a redshift of 0.96 which is very red in color and contains a compact optical nucleus with broad emission lines (Phillips et al. 1997). It is bright in the infrared H and K bands (Cowie et al. 1997). Rowan-Robinson (1997) interpret the ISO 6.7 micron flux density of 52 µJy as evidence for a massive starburst with a SFR of about 200 M0/yr. However, the observed radio variability combined with its flat radio spectrum and small size suggests that the radio emission in HDF 3644+1404 originates in an AGN, and is not primarily due to star formation.
The strongest radio source (S = 600 µJy) in the Hubble Deep Field, HDF 3644+1133, is identified with a bright (MB = -23) red elliptical galaxy at a redshift of 1.01. The radio luminosity is 6 x 1024 W/Hz, typical of the stronger FRI radio galaxies. It has a strong flat spectrum radio core which is unresolved at our highest angular resolution and is less than 0.1 arc seconds in diameter. A double lobe structure reaches some 15 arc seconds to the north and south of the unresolved core. HDF 3644+1133 has also been observed by ISO and has a 6.7 micron flux density of 50 µJy (Goldschmidt et al. 1997). Curiously, there is a very blue apparently elongated chain of galaxies, or possibly merging system, which lies within the northern radio lobe of HDF 3644+1133 and is at essentially the same redshift. This feature is similar in appearance to the "chain" galaxies described by Cowie et al. (1995), although it appears much larger in extent than the galaxies discussed by Cowie et al. It is not clear what relation exists, if any, between HDF 3644+1133 and the "chain" galaxy which has multiple apparently unresolved "hot spots." Possibly, the "chain" galaxy is undergoing active star formation induced by the jet emerging from the red elliptical. We have detected radio emission from another "chain" galaxy, HDF 3652+1354, but which is not in our complete radio sample.
Interestingly, there is one source which appears just below our completeness level with a snr of 4.7 which is also seen in our 20 cm data with a snr of 3.5, but which has no optical counterpart down to the limiting magnitude of IAB = 27.6 of the HDF. If real, it is unlikely that this unidentified source is an intrinsically faint galaxy at moderate redshift as no other radio sources are known to be identified with such faint galaxies. The observed radio emission could be the displaced radio lobe of an extended source with asymmetric structure and no detectable radio emission from the parent galaxy, but we consider this also unlikely as other microjansky sources all appear coincident with their optical counterpart. More likely, it may be a very high redshift (z > 6) I dropout galaxy, in which case H and J band observations with NICMOS may be able to detect the parent galaxy.