The identification of two z3.8 QSOs
in a deep CCD survey

Observations

We begin by discussing the photometric data used to identify the two QSO candidates, as well as archival information on these sources, before detailing our spectroscopic observations which have confirmed that both objects are high-redshift QSOs.

 
Figure 1: The (U-B)-(B-I) colour-colour plot for the I<21 objects with stellar morphologies from the combined fields of S98. The two QSOs are remarkable for their relatively blue (B-I) and very red (U-B) colours, and are marked as lower limits on this plot. The dotted line shows the expected colours of a QSO with a UV spectral slope of with , as a function of redshift. This prediction includes the effects of blanketing by the Lyman- (Madau 1995). The structure in this locus arises from the approximations used in the blanketing estimate. We also show the rough boundaries of the region in the colour-colour plane within which stars fall (taken from the distribution in Landolt 1992).

Palomar Observations

 


Figure 2: Deep CCD imaging of a arcsec regions around the two QSOs. the individual panels show the regions in three passbands , the strong detections of both Q1722+3211 and Q1322+3211 in I contrast with their non-detection in the U band to . The tick marks are every 5 arcsec on the panels and all have North top and East to the left.

The details of the observations and analysis of a CCD survey undertaken with the Hale 5.1-m of the Palomar Observatory are given in Smail et al. (1998, S98) and we only provide a very brief overview here. The aims of the survey were to study the galaxy populations of the clusters and to catalogue candidate lensed features within their cores (Edge et al. 1998, E98). The survey consists of deep imaging of arcmin regions centered on 10 z=0.2-0.3 luminous, X-ray clusters selected from the ROSAT All-sky Survey. The imaging discussed here was acquired under good conditions during the nights of 1994 June 9-12 using the COSMIC imaging spectrograph (Kells et al. 1998) and a thinned, blue-sensitive 2048 TEK detector. The total exposure times were 500s in both B and I and 3000s in U, reaching an 80% completeness limit of and photometry of these detections in with 20% accuracy to and . This survey is thus slightly deeper than that discussed in Hall et al. (1995), although their survey covers an area a little over 3 times larger than that used here (our survey covers 0.25 sq. degrees) and also includes much more extensive photometric information.

The seeing measured off the I frames discussed in this paper was 1.05 arcsec FWHM, with frames having typical seeing of arcsec. Observations of Landolt (1992) provide photometric calibration of the observations. Standard reduction techniques were applied to the science frames and objects were then detected and characterised from the final stacked I frames using the SExtractor package (Bertin & Arnouts 1996), with colours measured in 3 arcsec diameter aperture from the seeing-matched frames. The estimated reddenings in the two fields discussed here are E(B-V)=0.02 for A1758 and E(B-V)=0.07 for A2261, no reddening corrections have been applied to any of the photometry presented.

As part of the calibration and analysis we produced (U-B)-(B-I) colour-colour plots of those objects with I<21 which were classed as stars on the basis of their ellipticities and FWHM. 621 objects were selected from the complete survey covering 0.25 sq. degrees. For bright samples these plots show a strong stellar track which allowed us to confirm the accuracy of our photometric observations (S98). This procedure highlighted two stellar sources which were well-detected in our B and I frames with relatively blue colours, , but had only upper limits to their fluxes in the U-band (, Table 1 and Fig. 1). Visual inspection of the CCD frames confirmed that these two objects are bright in the BI frames, but are undetected in our deep U frames (Fig. 2).

Madau (1995) shows that due to blanketing in U by the Lyman- forest at rest-frame wavelengths shortward of 1215Å, a flat-spectrum source at z>3 will have (U-B)>2.5, and for would have -5 and (see also Hall et al. 1995). Thus the very red (U-B) colours for the two sources suggest that they lie at , while their relatively blue (B-I) colours indicate that the Lyman-limit hasn't moved completely through the B-band (Smith et al. 1994), providing an upper limit to their redshifts of -4.5. These redshift limits, combined with the compact morphologies of the two sources, support their identifications as QSOs at -4. The first candidate, Q1332+5034, lies on the outskirts of the cluster A1758 (z=0.280), while the other, Q1722+3211, lies in the field of A2261 (z=0.225). The large projected radii of the two sources in these clusters means that they are not significantly gravitationally amplified, with amplifications expected to be .

Archival data

By virtue of their presence in the fields of luminous X-ray clusters we have observations of the sources at X-ray wavelengths from the ROSAT X-ray satellite. A 16ks HRI image of A2261 provides only an upper limit for the X-ray flux of Q1722+3211 (Table 1). However, the 29ks HRI observation of A1758 shows that Q1332+5034 is coincident with an unresolved X-ray source with a flux erg cm s (unabsorbed, assuming a power law index of 1.5).

We have also searched the 1.4GHz VLA FIRST survey (White et al. 1997) for radio emission associated with the two QSOs. We find no detectable emission in the FIRST survey for Q1332+5034 or Q1722+3211. These upper limits correspond to a rest-frame 7.1GHz (4.3cm) power of less than WHz so neither object can be classed as radio loud.

Spectroscopic observations

 


Figure 3: The spectra of Q1332+5034 and Q1722+3211 obtained with ISIS on the WHT. Each panel shows the fluxed spectrum of the source, as well as an arbitrarily scaled spectrum of the night sky. In addition, in Q1722+3211, we also show at the top the spectrum smoothed to the instrumental resolution. The identifications of the main features are shown in both spectra (Table 2). In Q1332+5034 the features marked with -1/-2 are associated with the foreground absorption systems. The lower scale on each plot gives the observed wavelengths, while the upper scale indicates the wavelengths in the rest-frame of the QSO. The hatched regions mark areas affected by atmospheric absorption or the dichroic of the spectrograph.

The spectroscopic observations of the two QSO candidates were undertaken during an observing run to spectroscopically confirm the nature of a number of weak and strongly-lensed background sources identified in the cluster fields (E98). Details of these observations can be found in E98 and we repeat some of that material here. The observations were obtained on the nights of 1997 July 4-6 with the ISIS dual-beam spectrograph on the 4.2-m William Herschel Telescope (WHT) on La Palma. Using the 158 lines mm gratings in both the red and blue arms provides 2.9Å pixel sampling and Å resolution with a 1.5 arcsec slit, across a continuous wavelength range from 3500-9000Å. Conditions during the run were good, with a median seeing of 0.9 arcsec and good transparency. The total exposure times were 1200s for Q1332+5034 and 600s on Q1722+3211. Standard IRAF reduction procedures were used to process both the science frames and calibration data. We show the fluxed spectra of the two QSOs in Fig. 3.

© Copyright Astronomical Society of Australia 1997