Investigating the Evolution of Quasar Host Galaxies with NICMOS

Marek J. Kukula¹ Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218, USA

James S. Dunlop and David H. Hughes Institute for Astronomy, University of Edinburgh, Royal Observatory, Edinburgh EH9 3HJ, UK

Lance Miller Nuclear and Astrophysics Laboratory, University of Oxford, Keble Road, Oxford OX1 3RH, UK

Christopher P. O'Dea and Stefi A. Baum Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218, USA

           

(1)on leave from the Institute for Astronomy, University of Edinburgh

 

Abstract:

We present the preliminary results of our Cycle 7 program to study the evolution of the host galaxies of radio-loud and radio-quiet quasars out to a redshift of 2. By careful choice of filters, and careful quasar sample selection we have ensured that (1) we sample the same spectral region at all redshifts (2) our images consist purely of starlight, uncontaminated by emission lines, and (3) our results are not biased by quasar luminosity correlating with redshift. With this data we can investigate the cosmological evolution of the luminosity, scale-length, and degree of disturbance in quasar hosts and thus obtain insights into several key questions about quasar activity including the mechanism behind the dramatic evolution in the quasar luminosity function. An initial, very crude analysis of our NICMOS images shows that we are detecting the hosts out to a redshift of $\sim2$. The estimated galaxy luminosities are consistent with the values obtained for low-z quasars of the same luminosity.

quasars, host galaxies

Introduction

In recent years the study of quasar host galaxies has become increasingly important and it is now recognized that an understanding of their properties and evolution has the potential to resolve some long-standing mysteries concerning nuclear activity in galaxies. Conversely, quasar surveys provide an easy method of locating galaxies at large redshifts, so knowledge of the properties of quasar hosts offers a means of investigating the history and evolution of galaxies themselves.

The problems inherent in observing quasar host galaxies from the ground are too well-known to require detailed explanation but considering the enormous difficulties involved in separating faint, diffuse galaxy light from the point spread function (PSF) of a bright quasar, it is perhaps surprising that so much progress has been made to date using ground-based techniques. Although such studies are effectively limited to $z \leq 0.3$, at these low redshifts a combination of ground-based and HST programs is beginning to piece together a clear picture of the properties of quasar hosts in the local universe.

However, the local universe is not the most representative region in which to study the quasar population: arguably the epoch of greatest importance to quasar research occurred at a redshift of $\sim2$, when quasars were 100 times more numerous than they are today. Although simulations show that it is impossible to derive the properties of quasar hosts at such high z from the ground with any degree of confidence, numerous attempts have been made with - not surprisingly - confusing results.

In this paper we describe the preliminary results of a project designed to exploit the unique capabilities of the HST+WFPC2+NICMOS combination. Coupled with rigorous sample selection our HST data should enable us to study how the luminosities, sizes and morphologies of the hosts of both RLQs and RQQs have evolved from the `golden era' of quasar activity at $z \sim 2$ to the present day.

Previous work

The advent of NICMOS offers the first realistic hope of determining the properties of quasar hosts at high redshifts, and so a brief summary of the current state of our knowledge of host galaxies at low and high z is perhaps worthwhile at this point. Since the subject of this paper is our own NICMOS program we will focus on work which we ourselves have carried out, and of which our NICMOS study is the natural extension. However this is not to diminish the impact of the many other ground-based and HST studies which have been carried out in recent years.

The host galaxies of low-z quasars

At low z, attention has been focused on determining the luminosities, scalelengths, morphologies and interaction histories of the host galaxies, and investigating the extent to which these properties are correlated to the optical and radio luminosity of the quasar (Véron-Cetty & Woltjer 1990; Dunlop et al. 1993; McLeod & Rieke 1994a,b; Bahcall, Kirhakos & Schneider 1995; Hutchings & Morris 1995; Disney et al. 1995; Bahcall et al. 1996). For example, it has long been known that low-luminosity AGN display marked preferences in terms of host type, with (radio-quiet) Seyferts favoring spiral hosts whilst (radio-loud) Radio Galaxies are found exclusively in massive ellipticals, but the evidence for or against such a distinction for the hosts of radio-loud and radio-quiet quasars has until recently been unclear.

In our own ground-based K-band study of RLQ and RQQ hosts at $z\simeq0.2$ we found that almost half of the RQQs, as well as all of the RLQs, lie in spheroidal galaxies rather than exponential discs (Dunlop et al. 1993, Taylor et al. 1996). We also found that all of the host galaxies are large, with half-light radii $\geq 10$ kpc, luminous, with $L \geq L^{\star}$ at K, and display a $\mu_{1/2} - r_{1/2}$ (surface brightness - half-light radius) relation identical in both slope and normalization to that displayed by the brightest cluster galaxies.

Through a follow-up program of off-nuclear spectroscopy (Kukula et al. 1997; Hughes et al. 1998) we have also shown that the SEDs of these host galaxies can be described by a two-component spectrum, consisting of a well-behaved, old stellar population and a strong blue component which can be interpreted as an ongoing starburst, scattered quasar light from the nucleus or some combination of both. But the result most relevant to our HST program is that, longwards of the 4000Å  break, our spectroscopy demonstrates that the host galaxy spectrum is almost entirely dominated by light from the passive, underlying stellar population - i.e. directly related to the galaxy itself rather than to processes which might be linked to the active nucleus.

R-band imaging of low-z quasar hosts with WFPC2

It was this fact which persuaded us to choose R-band (F675W) for a WFPC2 host galaxy imaging proposal in Cycle 6. With matched RLQ and RQQ samples covering the redshift range $0.1 \leq z \leq 0.25$ the F675W filter samples the starlight from the host galaxies longward of the 4000Å break, whilst simultaneously avoiding strong emission lines such as [O III] and H$\alpha$ (impossible to achieve with a wider filter). As a result, compared to previous HST studies of quasar hosts, our WFPC2 observations are uniquely sensitive to the stellar continuum of the host and exclude many contaminating sources of emission which may have caused confusion in the past.

At the time of writing our low-z WFPC2 observing program is half complete. The data obtained so far are of extremely high quality, and we have been able to follow the host galaxy light to within $\sim 1''$ of the quasar, and thus to place tight constraints on the luminosity of the galaxy. When the observations are complete we will be able to rigorously examine the relationship between quasar optical luminosity/radio-loudness and the properties of the host at redshifts of $\simeq 0.2$.

However, in the context of our Cycle 7 NICMOS study these Cycle 6 WFPC2 observations will allow us to determine the rest-frame R-band luminosities, scalelengths and morphologies of the starlight in low-redshift quasar hosts. We can therefore use them to establish a crucial `here-and-now' baseline against which to judge the results of our near-infrared observations at higher redshifts.

High-redshift quasars.

Given that the host galaxies of low-redshift quasars have properties consistent with the products of successive galaxy mergers, and given the dramatic cosmological evolution seen in the quasar population (see below) one of the key aims of new quasar research must be to investigate the cosmological evolution of their hosts. In particular we would like to determine how the luminosities, scalelengths, and the degree of morphological disturbance seen in quasar host galaxies varies with cosmological epoch between the peak of quasar activity at $z \simeq 2$ and the present day, and whether the hosts of radio-loud and radio-quiet quasars differ in their luminosity/morphological evolution.

However, the difficulties involved in such studies are far more formidable than at low z and this is reflected in the current state of confusion surrounding the interpretation of recent ground-based attempts to observe quasar hosts at $z \simeq 2$.

Several groups have attempted such observations and have succeeded in detecting extensions around high-redshift quasars (Hintzen, Romanishin & Valdes 1991; Heckman et al. 1991; Lehnert et al. 1992; Aretxaga, Boyle & Terlevich 1995). These studies suggest that the host galaxies of quasars at $z \simeq 2$ are $\sim 2.5$ to 3 magnitudes brighter than those of low-redshift quasars, a result which is at least consistent with common scenarios for elliptical galaxy evolution. However, the usefulness of any such direct comparison is hampered by the fact that these high-redshift quasars are typically 5 magnitudes more luminous than the low-redshift objects studied to date. In addition, the situation is somewhat confused since other workers have failed to detect extended emission around high-redshift RQQs (Lowenthal et al. 1995). Hutchings (1995) also concludes that the hosts of RQQs are considerably fainter than those of RLQs at high redshift, a result which may simply indicate that much of the light detected around high-redshift RLQs may not be due to stars, but rather to processes associated with the extreme radio activity (as is found for high-z radio galaxies (Tadhunter et al. 1992), and for low-z RLQs (Stockton & MacKenty 1987). This emphasizes the importance of avoiding emission lines and sampling the spectrum longwards of the 4000Å break.

Motivation: host galaxies and quasar evolution

One of the main motivating factors in the study of quasar hosts at high redshifts is the hope that a knowledge of these galaxies might help to us to understand the physical origin of quasar evolution.

The strong cosmological evolution in comoving space density of quasars of a given luminosity has been known quantitatively for 30 years. However, although much work has gone into empirical fits of the evolution (Schmidt & Green 1983; Boyle, Shanks & Peterson 1988; Hewett, Foltz & Chaffee 1993; Goldschmidt et al. 1998) its causes remain a mystery. Hierarchical growth models such as CDM may be able to explain the rapid rise in QSO numbers with time at early epochs (z>4) as a consequence of the rate of galaxy mergers, but as yet they cannot account for the decline by a factor 100 in quasar numbers from z=2 to the present day without invoking some rather ad hoc assumptions (Carlberg 1990; Haenhelt & Rees 1993).

Clearly more observational evidence is required. Host galaxy studies are currently one of the most promising avenues of inquiry since they hold out the hope of linking quasars into the latest models of galaxy formation and evolution. At low redshifts there does appear to be a link between quasar luminosity and galaxy mass, at least for AGN in the luminosity range -20 > M_H > -26, in the sense that there appears to be a lower limit to the host luminosity which increases with quasar luminosity (MacLeod & Rieke 1994a,b; 1995). With HST/NICMOS it will be possible for the first time to study quasar hosts at the key epoch of quasar activity: z=2, the period at which the rate at which quasars were being created was overtaken by the rate at which quasars fade or disappear.

There is one crucial question to be answered: is the decline in quasar numbers caused by a form of luminosity evolution (LE) of long-lived quasars, or does it in fact reflect a more complicated balance between birth and death of short-lived quasars which is a function of quasar luminosity (luminosity-dependent density evolution, LDDE)?

Host galaxies offer a means of distinguishing between these two hypotheses. In the LE model, individual quasars decrease in luminosity with time. If we create test samples covering a narrow range of quasar luminosity over the range of redshifts from z=2 to the present epoch, we should see quasars of a fixed luminosity occupying a progressively wider range in mass of host galaxy.

In the LDDE model with short-lived quasars, quasar evolution has implicitly little to do with the physics of the active nucleus, and is linked to cosmological evolution of massive structures. We expect that there should be no redshift-dependence of quasar host luminosity (other than passive evolution) for quasars of a given luminosity. This result would tell us that quasar evolution is indeed directly linked to galaxy formation and the subsequent merger and interaction probabilities. It would tell us that the epoch z=2 was a key epoch for galaxies, not just active galaxies, and we can use N-body simulations of galaxy formation and hierarchical merging to test whether different galaxy-formation models can explain the observed quasar evolution.

By comparing the hosts of RLQs and RQQs and the way in which they evolve with time we will also gain insights into the way in which environment and history influence the `radio loudness' of quasars.

This project: studying host galaxy evolution with WFPC2 and NICMOS

The potential rewards of a program to determine the properties of quasar hosts from redshifts of $z\simeq0.2$ out to the peak in the quasar number density at $z \simeq 2$ are clearly quite far reaching and provided the impetus for our Cycle 7 study. However, the lesson of previous ground-based attempts has clearly shown that the only way to succeed in such an endeavor is to adopt a thorough and carefully thought-out approach.

Our Cycle 7 program therefore relies heavily on our previous experience with ground-based and Cycle 6 HST studies of low-z quasar hosts, and uses many of the same strategies in order to avoid potential problems. Fundamental to this approach is the selection of appropriate quasar samples and the use of carefully chosen filter-redshift combinations to maximize the chances of detecting starlight from the host galaxies.

Sample Design

We therefore now list and explain the selection criteria which we have used to define the samples of RQQs and RLQs for our Cycle 7 study of host-galaxy evolution. We stress that such sample control is essential if the aim is to obtain clean, meaningful results and advance our understanding of the cosmological evolution of the hosts of RQQs and RLQs.

• From simulated images of quasars and their hosts it is clear that it is essential, particularly at high-redshift, to confine our study to quasars of moderate-low optical luminosity if we are to derive reliable luminosities and scalelengths for their host galaxies (see Taylor et al. 1996). Accordingly we have confined our selection of high-redshift quasars to absolute magnitudes -24 > M_V > -25.

• Having defined the luminosity bin at high redshift, it is vital to break the flux-density-redshift correlation which inevitably arises in flux limited samples, and to make sure we sample a comparable range of luminosities at all intermediate redshifts. In our Cycle 6 study of quasar hosts at $z \sim 0.2$ (using the F675W filter in WFPC2) there are 5 RQQs and 5 RLQs with -24 > M_V > -25. These 10 objects form the lowz baseline against which we will test the results of our high-z observations. Thus we will be able to compare the host galaxies of quasars of the same intrinsic optical luminosity from $z \sim 0.2$ to $\sim2$.

• To ensure that we detect starlight from the host galaxies at all redshifts it is necessary to make sure that, as in our Cycle 6 WFPC2 study at $z\simeq0.2$, the spectrum of the host galaxy is always observed at $\lambda_{rest} > 4000$Å  and yet is uncontaminated by either [OIII] or H$\alpha$ line emission. This cannot be done with the somewhat more sensitive `Wide' filters such as F110W and F160W on NICMOS, but can be achieved over a wide range of redshifts by using the F814W filter (for WFPC2), and the F110M and F165M filters in NICMOS Camera 1, provided the quasar redshifts are restricted to lie within the following redshift bins: 0.32 < z < 0.43, 0.83 < z < 1.00, 1.36 < z < 1.69 and 1.67 < z < 2.09.

• In order to perform a clean comparison of the hosts of RLQs and RQQs at each epoch it is necessary to ensure that the RLQs are genuinely radio-loud ( $P_{5 GHz} > {\rm 10^{25} W Hz^{-1} sr^{-1}}$), that the RQQs are genuinely radio quiet ( $P_{5 GHz} < 10^{24.5} {\rm W Hz^{-1} sr^{-1}}$), and that within each redshift bin their optical luminosity distributions are well-matched. This is not a trivial task. It requires that we select only RQQs which have been observed and have not been detected with sufficient sensitivity at the VLA, and requires that we confine our RLQ sample to steep-spectrum objects whose intrinsic radio luminosity cannot be artificially boosted by relativistic beaming.

  

Figure: Redshift regimes in the current sample of RQQs (filled circles) and RLQs (open circles), with the appropriate NICMOS/WFPC2 filters indicated. Also shown are the quasars from our Cycle 6 WFPC2 F675W imaging study, to emphasize that these objects will provide the low-z baseline against which to measure any cosmological evolution in the current sample. Each redshift-filter combination samples the same region of the galaxy's restframe spectrum (roughly restframe R-band) and avoids both H$\alpha$ and [O III] $\lambda 5007$ emission.

The simultaneous application of these 4 constraints is sufficiently stringent that the resulting quasar sample, illustrated in Figure 1, is virtually unique (primarily because there are relatively few optically-faint steep-spectrum RLQs, and because relatively few RQQs have been observed with sufficient sensitivity in the radio). Also shown in Figure 1 is the low-redshift ($z \sim 0.2$) sample from our Cycle 6 WFPC2 program. It can be seen that this low-z sample spans almost 3 magnitudes in optical luminosity in a narrow redshift range whereas the samples selected for the Cycle 7 study are confined to only 1 magnitude in luminosity but span the bulk of the history of the Universe. This figure thus emphasizes how these two complementary samples will allow us to unambiguously separate the effects of cosmological epoch from any relation between host galaxy properties and quasar luminosity.

PSF determination

Accurate knowledge of the form of the PSF is essential in order to decouple the quasar contribution from that of its host. Synthetic PSFs, though often an excellent match to the inner regions of the profile, sometimes fail to reproduce the shape of the PSF at larger radii and this can be disastrous for the detection of host galaxy light.

We therefore devoted two entire orbits of our Cycle 7 allocation to obtaining empirical PSFs using our chosen filter/camera combinations (F110M and F165M with NICMOS Camera 1). Two different stars were used, to safeguard against the possibility that one of them might possess Vega-like dust shells which would compromise the PSF, and by using a sequence of exposures in MULTIACCUM mode we were able to obtain deep, unsaturated images of very high dynamic range.

Preliminary NICMOS results

The acquisition of NICMOS data for our Cycle 7 program was completed in February 1998, and the remaining WFPC2 F814W observations of objects in our $z\sim 0.4$ sub-sample are currently scheduled for the first half of 1999.

Pedestal removal and recalibration using the most recent reference files have resulted in significant improvements in image quality, although several problems, such as CR persistence, remain to be dealt with. Full 2-D modeling of the images would be premature at this stage. However, we have been able to carry out a preliminary analysis of the data, the results of which we describe here.

We have limited ourselves to an aperture 20 pixels (0.86 arcsec) across, centered on the quasar. By comparing the total flux within this region with that of an appropriately scaled empirical PSF we have been able to estimate the contributions of the quasar and its host to the light within a radius of $\sim 4$ kpc of the quasar. This method is extremely crude and inevitably leads to an underestimation of the total galaxy luminosity. However, the initial results are extremely encouraging:

• The measured J and H magnitudes of the quasars in our $z\sim1$ (F110M) and $z \sim 2$ (F165M) samples are close to the predicted values.

• In almost every case we detect a significant excess flux with respect to the empirical PSF, which we attribute to the host galaxy. This component typically accounts for $\sim 25$% of the light in our 0.87-arcsec aperture. This implies quasar:host ratios and galaxy luminosities which are consistent with the values obtained for low-z objects, although the uncertainties are currently too large to allow the testing of specific models of galaxy evolution.

We are therefore confident that, as the remaining issues of image processing are resolved, our NICMOS images will begin to provide the first clean measurements of host galaxy sizes and luminosities for quasars spanning a large fraction of the history of the Universe.