The Optical/Near-IR Colours of Red Quasars
Paul J Francis , Matthew T. Whiting , Rachel L. Webster, PASA, 17 (1), 56.

Next Section: Multicolour Selection of Red
Title/Abstract Page: The Optical/Near-IR Colours of
Previous Section: OBSERVATIONS
Contents Page: Volume 17, Number 1

Results and Discussion

The Colour Distribution

The results are listed in Table 4. Quoted errors are $1 \sigma$ ; upper limits are $3 \sigma$ .

Our data confirm the basic result of Webster et al: the Parkes quasars have very different B-K colours from optically selected QSOs (Fig 3). The difference is significant: a Kolmogorov-Smirnov test shows that the the probability of getting two samples this different from the same parent population is only

9.1 x 10^-5. The bluest Parkes sources have colours very similar to those of optically selected QSOs, but the distribution of colours extends much further into the red.

**Figure 3:** The distribution of B-K_n colours for the Parkes sample (top panel), and the optically selected LBQS sample (bottom panel). Sources with spatially extended images (radio galaxies) have been excluded, as have sources with redshift z>3 (as Ly $\alpha$ forest depressed the B-band flux). Only Parkes sources within the complete sub-sample have been used. The LBQS data from this paper have been supplemented by data from Francis (1996).
$\begin{figure} \begin{center} \psfig{file=bkhist.eps,height=10cm}\end{center}\end{figure}$

**Figure 4:** The optical and infra-red colours of the complete sub-set of the Parkes sample (triangles and crosses), compared with a small sample of optically selected LBQS QSOs (circles). Solid triangles denote unresolved sources: crosses are galaxies. The solid line shows where a pure power-law continuum slope would lie: it runs from
$F_{\nu } \propto \nu ^{0}$ on the left end, to

$F_{\nu } \propto \nu ^{-2}$ on the right end. Error bars are not shown for the unresolved Parkes sources, but are comparable to those of the optically selected QSOs. The reddening vector is for an extinction E(B-V)=0.2, a redshift of one, and dust extinction as in equation 1. The direction of the reddening vector is independent of redshift.
$\begin{figure} \begin{center} \psfig{file=colours.eps,height=12cm}\end{center}\end{figure}$

The `Main Sequence'

Are the Parkes sources uniformly red everywhere between B and K_n? In Fig 4 we plot a measure of the optical colour (B-I) against a measure of the near-IR colour (J-K_n) for the complete sub-sample. Objects whose continuum shape approximates a featureless power-law all the way from B to K_n should lie close to the solid line in this plot.

$\sim 90$ % of all the Parkes sources do indeed lie close to the power-law line in Fig 4. These sources form a `main sequence' of quasar colours, stretching from blue objects with

$F_{\nu} \propto \nu^{\sim 0}$ to red objects with

$F_{\nu} \propto \nu^{\sim -2}$ . Examples of quasars from both ends of this `main sequence' are shown in Fig 5. Note that these quasars can lie on either side of the power-law line: ie. they can have both `n'- and `u'-shaped continuum spectra. The majority, however, lie above the line, consistent with slightly `u'-shaped spectra (redder in the near-IR than in the optical). This supports the synchrotron model for these sources. We defer discussion of this point to the detailed synchrotron modelling of the companion paper Whiting et al.

**Figure 5:** Spectral energy distributions of representative Parkes quasars from the blue (left six plots) and red (right six plots) ends of the `main sequence', as defined in the text. Sources on the left have J-K_n < 1.5 and B-I < 1.5; sources on the right have J-K_n > 1.8 and 3 > B-I > 1.8.
$\begin{figure} \begin{center} \psfig{file=mainseq.eps,height=12cm}\end{center}\end{figure}$

Optically Selected QSOs

As Fig 4 shows, the optically selected QSOs all have very similar colours, and lie at the blue end of the `main sequence'. They lie systematically below the power-law line, however, indicating that they have `n' shaped spectra: ie. they are redder in the optical than in the near-IR. This can be seen in their spectra energy distributions, shown in Fig 6.

**Figure 6:** Spectral energy distributions of all six optically selected QSOs with complete photometric data.
$\begin{figure} \begin{center} \psfig{file=lbqs.eps,height=12cm}\end{center}\end{figure}$

This spectral curvature matches the predictions of the dust model. Wills, Netzer & Wills (1985), however, suggested that it may be partially due to blended Fe II and Balmer-line emission, though Francis et al. (1991) argued that this curvature is too large to be plausibly explained by emission-line contributions.

The position of the optically selected QSOs at the blue end of the `main sequence' would be expected if the cause of redness in the Parkes quasars is the addition of a red synchrotron component to an underlying blue continuum which is identical to that in radio-quiet QSOs (Whiting et al.).

Galaxies and Extremely Red Objects

**Figure 7:** The spectral energy distributions of three representative galaxies from the Parkes sample.
$\begin{figure} \begin{center} \psfig{file=gals.eps,height=10cm}\end{center}\end{figure}$

The spectra of the spatially extended sources in the Parkes sample are sharply peaked in the red, as would be expected from moderate redshift galaxies (Fig 7). They therefore lie far below the `main sequence' in Fig 4, the one exception being PKS 1514-241, which is a galaxy at z=0.049 with a BL Lac nucleus, which is presumably diluting the galaxy colours. Higher redshift galaxies lie further to the right on this plot, as would be expected due to the 400 nm break reducing the B-band flux.

**Figure 8:** The spectral energy distributions of three representative Parkes quasars with redshifts z>3, showing the dip in the B-band caused by Ly $\alpha$ forest absorption.
$\begin{figure} \begin{center} \psfig{file=highz.eps,height=10cm}\end{center}\end{figure}$

What are the other, red, highly `n'-shaped objects lying far below the `main sequence' which are not spatially resolved? A few are high redshift QSOs, in which the B-band flux has been reduced by Ly $\alpha$ forest absorption (Fig 3.4). The reddest objects, however, with B-I>3 (Fig 9), do not lie at high redshifts. We have obtained spectra of four of these very red objects (Francis et al. 2000, in preparation). Three show hybrid spectra: they look like galaxies at short wavelengths, but at longer wavelengths a red power-law continuum component is seen, along with broad emission-lines. The ratios of H $\alpha$ to H $\beta$ are around 20: far above those seen in normal AGN ( $\sim 5$ ) and evidence of substantial reddening (Fig 10). Note that these hybrid objects all have radio spectra indices near the steep spectrum cut-off of our sample, as do the galaxies in the sample.

**Figure 9:** The spectral energy distributions of the six Parkes sources with B-I>3. The data for PKS 1706+006 have been adjusted for galactic dust extinction of E(B-V) = 0.23 (Schlegel, Finkbeiner & Davis 1998), assuming a dust extinction law as described in the text.
$\begin{figure} \begin{center} \psfig{file=red.eps,height=10cm}\end{center}\end{figure}$

**Figure 10:** Optical spectra of four extremely red Parkes sources. With the exception of PKS 0131-001, the spectra show features both of galaxy light (the 400 nm break and narrow [O II] 372.7 nm and [O III] 495.9/500.7 lines) and of dust-reddened quasar light (a red continuum at long wavelengths, broad H $\alpha$ 656.3 nm line emission, and the notable weakness of the broad H $\beta$ 486.1 nm line with respect to H $\alpha$ ).
$\begin{figure} \begin{center} \psfig{file=specplot.eps,height=12cm}\end{center}\end{figure}$

The reddest objects are thus a heterogeneous group: some are high redshift quasars, some are galaxies, and some are heavily dust-reddened quasars.

Unidentified Objects

Four Parkes sources were not detected in any band. After correction for galactic foreground absorption (Schlegel et al.), our non-detections impose 3 $\sigma$ upper limits of H> 19.61 for PKS 1532+004, H > 19.76 & K > 19.29 for PKS 1601-222, H>17.22 and K>16.61 for PKS 1649-062 (which is subjected to substantial galactic reddening) and H > 19.82 for PKS 2047+098.

If unified schemes for radio-loud AGN are correct, the host galaxies of our flat-radio-spectrum sources should be very similar to those of steep-radio-spectrum radio galaxies. This enables us to place a lower-limit on the redshift of these unidentified sources: even if their AGN light is completely obscured, we should still see the host galaxy, which should lie on the K-band Hubble diagram for radio galaxies (eg. McCarthy 1992). To be undetected at our magnitude limits, therefore, all these sources must lie above redshift 1, and apart from PKS 1649-062, probably lie above redshift 3.

Anomalous Objects

**Figure 11:** Spectral energy distributions of three anomalous Parkes sources.
$\begin{figure} \begin{center} \psfig{file=weirdo.eps,height=10cm}\end{center}\end{figure}$

Three sources have colours that do not fit any of these categories (Fig 11). We discuss these in turn.

PKS 1648+015 shows a smooth optical power-law rising into the red, until at around $1.4 \mu$ m, the flux abruptly decreases. As all the IR data points were taken within minutes of each other in good weather conditions, we believe that this near-IR turn-over is real. We obtained a somewhat noisy optical spectrum of this source (Drinkwater et al.) which shows a featureless, very red power-law, in excellent agreement with the photometry. We cannot explain this source.

PKS 1732+094 is blue longwards of around $0.6 \mu$ m, but drops dramatically at shorter wavelengths. Our spectrum of this source (Drinkwater et al.) is too poor to be of any use. We hypothesise that this may be a very high redshift z>4 quasar, and that the drop in the blue is due to Ly $\alpha$ absorption.

PKS 2002-185 has optical colours typical of the bluest Parkes sources, but in the near-IR is bluer still: far bluer than any other source at these wavelengths. An optical spectrum, covering a very restricted wavelength range (Wilkes et al. 1983) shows only a single broad emission-line: on the assumption that this is Mg II (279.8 nm) a redshift of 0.859 is determined.

Next Section: Multicolour Selection of Red
Title/Abstract Page: The Optical/Near-IR Colours of
Previous Section: OBSERVATIONS
Contents Page: Volume 17, Number 1

ASKAP

Public