A Complete Sample of Millijansky Radio Sources

Abstract

The results of an optical and infrared investigation of a radio sample drawn from the 1.4 GHz Leiden-Berkeley Deep Survey are presented. This is believed to be the most comprehensive sample of radio sources at millijansky flux limits that is currently available. Optical counterparts have been identified for all but four sources in the two Hercules fields, and 80% of them are identified in the near-infrared. Redshifts have been obtained for 49 of the identified sources, and photometric redshifts were computed from the g,r,i,K data for the remaining 20. The general properties of the sample are summarized. The use of this sample in measuring the 1.4 GHz radio luminosity function is discussed. Finally, HST/NICMOS images of two old, red galaxies at z=1.5 are presented which show that both galaxies are dominated by an r^{1/4} profile with a scale-length of 5 kpc.

Full text

Page 1

arXiv:astro-ph/9908174v2 16 Aug 1999
A Complete Sample of Millijansky Radio Sources
I. Waddington
Department of Physics & Astronomy, Arizona State University, Tempe,
AZ 85287–1504
Abstract.
The results of an optical and infrared investigation of a radio sample
drawn from the 1.4 GHz Leiden-Berkeley Deep Survey are presented.
This is believed to be the most comprehensive sample of radio sources at
millijansky flux limits that is currently available. Optical counterparts
have been identified for all but four sources in the two Hercules fields,
and 80% of them are identified in the near-infrared. Redshifts have been
obtained for 49 of the identified sources, and photometric redshifts were
computed from the griK data for the remaining 20.
The general properties of the sample are summarized. The use of this
sample in measuring the 1.4 GHz radio luminosity function is discussed.
Finally, HST/NICMOS images of two old, red galaxies at z ≃ 1.5 are
presented which show that both galaxies are dominated by an r1/4profile
with a scale-length of 5 kpc.
1.The sample
The Leiden-Berkeley Deep Survey (‘LBDS’) consists of nine high latitude fields
in the selected areas SA28, SA57, SA68 and an area in Hercules. They were
surveyed with the Westerbork Synthesis Radio Telescope at 21 cm (1.412 GHz),
reaching a 5-σ limiting flux density of 1 mJy (Windhorst et al. 1984a). Multi-
color prime focus photographic plates of the fields were used to find optical
counterparts to the radio sources. Identifications were found for 53% of the
sources in the full survey, whilst for the Hercules fields 47 out of 73 sources were
identified (Windhorst et al. 1984b; Kron et al. 1985).
The Hercules fields were subsequently observed on the 200 inch Hale Tele-
scope at Palomar Observatory between 1984 and 1988. Multiple observations
were made through Gunn g, r and i filters over six runs. After processing and
stacking of the multiple-epoch images, optical counterparts for 22 of the sources
were found, leaving only four sources unidentified to r ≃ 26. Near-infrared
observations have been made of the entire subsample at K, yielding 60/73 de-
tections down to K ≃ 19–21. Half of the sources have been observed in H and
approximately one-third in J. Observations of the brighter sources were made by
Thuan et al. (1984) and by Neugebauer et al. and Katgert et al. (priv. comm.).
K-band observations of the sample were completed by the present author and
collaborators at the UK Infrared Telescope.
1

Page 2

Figure 1.
S1.4 ≥ 2 mJy in the LBDS Hercules fields. The flux density limits
for the survey are shown for flat-spectrum (solid line, open circles)
and steep-spectrum (dashed line, solid circles) sources, together with
the limits for the PSR (dotted line) and 3C (dot-dash line) surveys.
(b) [right] The cumulative redshift distribution for the S1.4≥ 2 mJy
sample.The bold histogram corresponds to the best-fitting photo-
metric redshift distribution, the light histograms represent the 99%
confidence limits on the photometric redshifts.
the model luminosity functions of Dunlop & Peacock (1990), assuming
H0= 50 km s−1Mpc−1, Ω0= 1, Λ = 0.
(a) [left] The luminosity–redshift plane for sources with
The gray lines are
Prior to the start of the current work, only 16 of the 73 sources in the LBDS
Hercules fields had redshifts published in the literature. Another 16 sources had
unpublished redshifts. The author and collaborators have successfully observed
a further 17 sources during the past few years, using both the 4.2 m William
Herschel Telescope (Waddington 1999) and the 10 m Keck Telescope (Dunlop
et al. 1996; Spinrad et al. 1997; Dey 1997). This brings the total number of
spectroscopic redshifts to 49 out of 73 sources (67%). Photometric redshifts
were calculated for the remaining one-third of the sample, using the spectral
population synthesis models of Jimenez et al. (1998).
2. The 1.4 GHz radio luminosity function
Dunlop & Peacock (1990) used a sample of radio sources brighter than 0.1 Jy
at 2.7 GHz to investigate the evolution of the radio luminosity function (RLF).
They concluded that the comoving density of both flat- and steep-spectrum
sources suffers a cut-off at redshifts z ≃ 2–4. This conclusion was drawn from
the behavior of both free-form and simple parametric models (PLE/LDE), and
the model-independent, banded V/Vmaxtest. However, the results were crucially
dependent upon the accuracy of their redshift estimates in the Parkes Selected
Regions (PSR).
Figure 1a compares the flux limits of the LBDS with those of the PSR and
3C surveys. It is seen how the LBDS can be used to: (i) probe the faint end
of the RLF out to much greater redshifts than the brighter surveys; and (ii)
2

Page 3

Figure 2.
files for 53W069 at z = 1.432 [left] and 53W091 at z = 1.552 [right].
The images are 3′′on a side. Solid lines are the best-fitting model
profiles convolved with the NICMOS PSF. AB magnitudes are used.
HST/NICMOS F110W images and surface brightness pro-
detect powerful radio galaxies out to very high redshifts (∼> 10). Thus we are
able to use this millijansky sample to test the reality of the redshift cut-off. In
figure 1b the cumulative redshift distribution of the LBDS Hercules sample (only
those sources with S1.4≥ 2 mJy) is compared with the predictions of Dunlop
& Peacock (1990). It is seen that two of the free-form models (FF-4 and FF-5)
provide a reasonable fit to the data over all redshifts, suggesting that the high-z
decline in the RLF is real. The “bump” in the best-fit histogram at 0.4∼< z∼< 1
is due to two spikes in the redshift distribution, that may be the result of possible
large-scale structures (sheets) along the line of sight.
However, a more detailed investigation of the RLF reveals that the lumi-
nosity dependence of the data does not agree with the models. The observed
RLF shows some indication that it turns over at z ≃ 0.5–1.5, and that the red-
shift of this cut-off is a function of the radio luminosity. The small number of
sources makes it difficult to separate the redshift and luminosity dependence of
the RLF sufficiently to be certain of this trend, but work is ongoing to improve
the modeling of the data.
3.Surface brightness profiles of two old galaxies
Two of the most interesting individual sources in the sample are 53W069 and
53W091. Keck spectra of these galaxies revealed that their restframe ultraviolet
light was dominated by old stellar populations at ages of 4.5 Gyr at z = 1.432
and 3.5 Gyr at z = 1.552 respectively (Dey 1997; Spinrad et al. 1997). Although
there continues to be some debate over the accuracy of these ages, there can be
little dispute that they are two of the oldest galaxies observed at that redshift.
An important corollary to investigate was whether these galaxies were also
dynamically old objects. Peacock et al. (1998) showed that in order for the
number density of these objects to be consistent with the evolving power spec-
trum of primordial density fluctuations, then the galaxies must be 3–4 Gyr old
at z ≃ 1.5, having collapsed at z ≃ 6–8. This result is independent of the ages
of the sources derived from their Keck spectra. Further clues to the dynami-
cal history of the two galaxies can be gained by looking at their morphologies.
53W069 and 53W091 were observed using WFPC2 and NICMOS on the Hubble
Space Telescope in Cycle 7 (Waddington et al. 1999). Data were collected in
3

Page 4

F814W and F110W: these filters bridge the 4000˚ A break and are thus most
sensitive to the young and old stars in a galaxy respectively. Figure 2 shows the
F110W images and surface brightess profiles for the two objects.
53W069 has a regular de Vaucouleurs r1/4profile, with an effective radius
of 0.′′5 or 4 kpc (H0= 65 km s−1Mpc−1, Ω0= 0.2, Λ = 0). 53W091 is similarly
dominated by an r1/4profile of effective radius 0.′′5 (4 kpc), however there is
an additional component required to fit this galaxy. The form of this extra
emission has not been unambiguously identified, but initial results suggest that it
is consistent with an exponential profile of 0.′′2 (2 kpc) scale-length, contributing
∼ 40% of the F110W flux within a 1.′′5 diameter aperture.
The spectrum of 53W091 can be modeled by adding a young stellar pop-
ulation to the spectrum of 53W069. Similarly the surface brightness profile of
53W091 can be modeled by adding an exponential component to the profile of
53W069. A possible interpretation is that 53W091 has a star-forming disk-like
structure surrounding an otherwise passively evolving elliptical galaxy, such as
53W069.
Acknowledgments.
larly Rogier Windhorst, James Dunlop and John Peacock.
supported by a research studentship from the UK PPARC; and NASA grant
GO-7280.0*.96A from STScI under NASA contract NAS5-26555.
I thank my collaborators on this project, particu-
This work was
References
Dey, A. 1997, “The Hubble Space Telescope and the High Redshift Universe”,
N. Tanvir et al. (eds), World Scientific, p. 373
Dunlop, J.S., Peacock, J. A., Spinrad, H., Dey, A., Jimenez, R., Stern, D.,
Windhorst, R. A. 1996, Nature, 381, 581
Jimenez, R., Dunlop, J. S., Peacock, J. A., Padoan, P., MacDonald, J., Jørgensen,
U. G. 1999, MNRAS, submitted
Kron, R. G., Koo, D. C., Windhorst, R. A. 1985, A&A, 146, 38
Peacock, J. A., Jimenez, R., Dunlop, J. S., Waddington, I., Spinrad, H., Stern,
D., Dey, A., Windhorst, R. A. 1998, MNRAS, 296, 1089
Spinrad, H., Dey, A., Stern, D., Dunlop, J. S., Peacock, J. A., Jimenez, R.,
Windhorst, R. A. 1997, ApJ, 484, 581
Thuan, T. X., Windhorst, R. A., Puschell, J. J., Isaacman, R. B., Owen, F. N.
1984, ApJ, 285, 515
Waddington, I. 1999, PhD Thesis, University of Edinburgh
Waddington, I., Windhorst, R. A., Peacock, J. A., Dunlop, J. S., Cohen, S. H.,
McClure, R. 1999, in preparation
Windhorst, R. A., van Heerde, G. M., Katgert, P. 1984a, A&AS, 58, 1
Windhorst, R. A., Kron, R. G., Koo, D. C. 1984b, A&AS, 58, 39
4