New Astronomy Reviews 47 (2003) 211–214www.elsevier.com/ locate/newastrev

Q uasars are more luminous than radio galaxies—so what?*Chris Simpson

Subaru Telescope, National Astronomical Observatory of Japan, 650 North A’ ohoku Place, Hilo, HI 96720,USA

Abstract

Surveys to find high-redshift radio galaxies deliberately exclude optically-bright objects, which may be distant radio-loudquasars. In order to properly determine the space density of supermassive black holes, the fraction of such objects missedmust be determined within a quantitative framework for AGN unification. I briefly describe the receding torus model, whichpredicts that quasars should have more luminous ionizing continua than radio galaxies of similar radio luminosity, andpresent evidence to support it. I also suggest two further tests of the model which should constrain some of its parameters. 2003 Published by Elsevier B.V.

Keywords: Galaxies: active; Radio continuum: galaxies

1 . Introduction how orientation affects the observed properties ofextragalactic radio sources is therefore required to

As the cosmological applications of radio galaxies fully exploit these objects.become clearer, it is important to realize that they donot represent a fundamental class of object. Searchesfor high-redshift radio sources use spectral index and 2 . The receding torusoptical magnitude selection criteria which result inthe exclusion of radio-loud quasars even though The ubiquity of the ‘‘big red bump’’ longward ofthese are fundamentally the same objects, according 1mm in the spectra of QSOs (Elvis et al., 1994),to the standard AGN unification paradigm. While this together with the interpretation of this as thermaldoes not matter if one is using radio galaxies simply emission from hot dust (Barvainis, 1987; Clavel etas signposts to locate and study places where large-al., 1989), seems to indicate that dust will alwaysscale structure is developing (Venemans et al., 2002; exist as close to the active nucleus as physics allows.Simpson and Rawlings, 2002), it is possible that they If this is the same dust that is responsible for hidingcould be used to measure the rate of formation of the nucleus from direct view in narrow-line objects,(spinning) supermassive black holes and/or clusters. then it must lie further from the nucleus in objectsTo do this, however, requires not just an understand- with higher ionizing luminosities. The assumptioning of radio source physics (Blundell et al., 1999), that the height of the obscuring structure (‘‘torus’’ inbut also a renormalization to account for the missed unification parlance) remains constant leads to thequasar population. Aquantitative understanding of receding torus model (Lawrence, 1991; Hill et al.,

1996; Simpson, 1998). In this scenario, more lumin-ous objects have a higher probability of being*Corresponding author.

E-mail address: chris@naoj.org(C. Simpson). observed as quasars, and consequently the mean

1387-6473/03/$ – see front matter 2003 Published by Elsevier B.V.doi:10.1016/S1387-6473(03)00027-7

212 C. Simpson / New Astronomy Reviews 47 (2003) 211–214

ionizing luminosity of quasars in an orientationally-unbiased (radio-selected) sample will be higher thanthat of radio galaxies. It therefore follows that anyquantity which is more strongly correlated withionizing luminosity than with radio luminosity willalso be higher in quasars than radio galaxies (Simp-son, 1998). Indeed, studies of [OIII ] (Jackson andBrowne, 1990; Laing et al., 1994), mid-to-far-in-frared (Heckman et al., 1992; van Bemmel et al.,2000), and submillimetre (Willott et al., 2002)luminosities indicate that quasars are indeed moreluminous than radio galaxies in these properties byfactors of a few. Since the receding torus model wasfirst described at around the same time as the earliestof these studies, it has always surprised me that it did Fig. 1. Relationship between optical and near-IR spectral indicesnot gain greater acceptance as a way to explain these(open and solid points, respectively) and rest-frame 1mm lumin-differences, which were incompatible with a picture osity for a complete sample of 3CRR quasars (Simpson and

Rawlings, 2000). The solid line indicates the relationship pre-where the torus opening angle was the same in all0.5dicted by the receding torus model (L ~ L ) for the near-IR2 mm 1 mmobjects. The lack of a difference in [OII]

spectral indexonly (with arbitrary vertical normalization).luminosities between quasars and radio galaxies (Heset al., 1993) is due to the insensitivity of this line tothe strength of the ionizing continuum (Simpson,1998), while the apparent lack of a difference in [OIII ] luminosity at high redshift (Jackson and Rawl- increase with AGN luminosity as the inner wallings, 1997) is probably a result of large measurement of the torus is pushed away (Willott et al., 2000;uncertainties—the lack of a significant difference Grimes et al., 2003).does not equate to the two classes having the same (2) The fraction of lightly-reddened quasars (i.e.,luminosity, and the data are also consistent with those where broad wings are seen on Ha but notquasars being twice as luminous in this line, as is Hb ) should decrease with luminosity as theseen at lower redshift (Laing et al., 1994). solid angle over which lines of sight ‘‘graze’’

the edge of the torus decreases (Simpson et al.,1999).

3 . Observational evidence (3) The strength of ‘‘big red bump’’ relative to theionizing continuum should be less in more

Of course, the key assumption in the receding luminous objects as the solid angle subtendedtorus model is that the height of the torus is by the torus decreases (Simpson and Rawlings,independent of AGN luminosity. While this seems to 2000; Law-Green et al., 2003). This is thebe a reasonable zeroth-order assumption, it need not cleanest and most quantitative test and thebe true since the AGN luminosity is correlated with model fits the data well (Fig. 1).black hole mass, and therefore also with the mass ofthe host galaxy; it is quite possible that either or both Interestingly, there is also evidence that in theof these quantities might affect the height of the most luminous quasars, where the inner wall of thetorus. However, several pieces of observational torus is pushed beyond a few parsecs, that the torusevidence indicate that the height of the torus is not a disintegrates completely (Law-Green et al., 2003).strong function of luminosity. The most distant SDSS quasars therefore provide an

accurate census of the number of supermassive black(1) The quasar fraction in different samples should holes in the early Universe.

C. Simpson / New Astronomy Reviews 47 (2003) 211–214 213

4 . Implications

None of the pieces of evidence presented in theprevious section is consistent with what some au-thors refer to as the ‘‘simplest’’ unification scenario,where broad- and narrow-line objects are separatedby a single critical angle, independent of AGNproperties. This should not be a cause of concern,since such a scenario is unrealistic, requiring eitherthe inner walls of the torus to unphysically remain atthe same distance from the nucleus irrespective ofthe AGN luminosity, or the torus height to increasein a contrived manner so as to maintain a constantopening angle. In addition, the first and third pointsabove favour a torus rather than a warped disk since Fig. 2. The factor by which quasars are more luminous than radiotogether they indicate that the obscuration is caused galaxies in an orientation-independent sample, as a function of theby the hot dust close to the nucleus, rather than by a dispersion (assumed to be a Gaussian in log space) in the ionizing

luminosities. The solid black region is for the simple recedingwarp at larger distances; the strong correlationtorus model, where the spread represents samples with differentbetween nuclear extinction and viewing angle inquasar fractions from 30 to 80%. The other regions indicate what

narrow-line radio galaxies also favours a torus rather happens if there is a random spread in the heights of the tori in thethan a warped disk (Simpson et al., 2000). sample, of factors of 2, 5, and 10 (1s scatter), with 10 being the

As explained earlier, the receding torus model lightest gray colour.

results in the mean (ionizing) luminosity of quasarsin a sample being brighter than the mean (ionizing)luminosity of radio galaxies. The factor by which

1quasars are more luminous depends on two quan- this number is insensitive to the quasar fraction) .tities: (i) the mean opening angle of the torus in the These values are larger than the factors of 2–5 bysample, which is fixed by the observed quasar which quasars are observed to be more luminousfraction, and (ii) the spread in the distribution of than radio galaxies (Heckman et al., 1992; Laing etionizing luminosities in the sample. This second al., 1994; van Bemmel et al., 2000; Willott et al.,quantity is impossible to measure, but can be esti- 2002), but this is due to the simplistic assumptionmated as the convolution of the observed radio that all tori have the same height. If torus heights areluminosity distribution with a Gaussian representing drawn from a log-normal distribution, the overlumin-the scatter in the radio-ionizing luminosity correla- osity is reduced since it increases the likelihood oftion. Unfortunately, this scatter is also unknown; I low-L objects being seen as quasars (since someion

previously adopted 0.6 dex (Simpson, 1998) from the will have short tori) while decreasing the likelihoodscatter in the radio–optical correlation (Serjeant et of high-L objects being seen as quasars (sinceion

al., 1998), although the use ofM as a tracer of the some will have tall tori). This is represented by theB

total ionizing luminosity adds uncertainties due to gray regions inFig. 2 where it can be seen that adifferences in rest-frame wavelength, extinction, and significant difference in luminosities persists evenoptical /UV spectral index; tighter correlations do when the torus height is allowed to vary by 1 dex. Itexist (Rawlings and Saunders, 1991) and I believe is therefore inevitable that quasars will be, onthe true scatter is lower. For now, I shall leave this as average, more luminous in their ionizing radiationa free parameter.

The solid black region inFig. 2 indicates that1quasars exceed radio galaxies by a factor of 2–20 for Note that Fig. 4 inSimpson (1998) is incorrect due to improper

reasonable values of theL 2L dispersion (also, normalization.rad ion

214 C. Simpson / New Astronomy Reviews 47 (2003) 211–214

than radio galaxies, and hence more luminous in any fundamental class of object and their number den-related quantities (e.g. emission line and infrared sities must be related to the total (radio-loud) AGNluminosities, which arise from the reprocessing of number density by accounting for the missing popu-ionizing photons). This is exactly what is observed, lation (narrow- and broad-line, respectively).and adds additional support for the receding torus.

R eferences5 . Further studies

B arvainis, R., 1987. ApJ 320, 537.B lundell, K.M., Rawlings, S., Willott, C.J., 1999. AJ 117, 677.None of the observational results quoted in thisC lavel, J., Wamsteker, W., Glass, I.S., 1989. ApJ 337, 236.paper arose from a deliberate attempt to test theE lvis, M., Wilkes, B.J., McDowell, J.C., Green, R.F., Bechtold, J.,

receding torus model. However, the success of this Willner, S.P., Oey, M.S., Polomski, E., Cutri, R., 1994. ApJSmodel, together with the intriguing possibility that 95, 1.Fig. 2 could be used to estimate the scatter in the G rimes J.A., Rawlings S., Willott C.J., 2003. NewAR 47, this

issue [doi:10.1016/S1387-6473(03)00026-5].L 2L correlation and the range of torus heights,rad ionH eckman, T.M., Chambers, K.C., Postman, M., 1992. ApJ 391,makes further study worthwhile.

39.First, the correlation inFig. 1 should be extended H es, R., Barthel, P.D., Fosbury, R.A.E., 1993. Nature 362, 326.

to lower luminosities by studying fainter samples of H ill, G.J., Goodrich, R.W., DePoy, D.L., 1996. ApJ 462, 163.radio-loud quasars at the same redshift (e.g. from the J ackson, N., Browne, I.W.A., 1990. Nature 343, 43.

J ackson, N., Rawlings, S., 1997. MNRAS 286, 241.7C sample); a sufficiently large number can beL aw-Green J.D.B., Simpson C., Ward M.J., Boisson C., 2003.studied with an 8-m telescope in a few nights.

MNRAS, submitted.Second, direct evidence for an increase in the L awrence, A., 1991. MNRAS 252, 586.

opening angle of the torus should be sought by L aing, R.A., Jenkins, C.R., Wall, J.V., Unger, S.V., 1994. In:looking at the distributions of the core-to-lobe ratio, Bicknell, G.V., Dopita, M.A., Quinn, P.J. (Eds.), ASP 54: The

First Stromlo Symposium: The Physics of Active Galaxies.R, in different samples. This measurement can beASP, San Francisco, p. 201.converted into an approximate viewing angle (Simp-

R awlings, S., Saunders, R., 1991. Nature 349, 138.son et al., 2000) and the torus opening angle can be S erjeant, S., Rawlings, S., Maddox, S.J., Baker, J.C., Clements,inferred from the distributions of quasars and radio D., Lacy, M., Lilje, P.B., 1998. MNRAS 294, 494.galaxies. S impson, C., 1998. MNRAS 297, L39.

S impson, C., Rawlings, S., Lacy, M., 1999. MNRAS 306, 828.S impson, C., Rawlings, S., 2000. MNRAS 317, 1023.S impson, C., Ward, M., Wall, J.V., 2000. MNRAS 319, 963.

6 . Summary S impson, C., Rawlings, S., 2002. MNRAS 334, 511.v an Bemmel, I.M., Barthel, P.D., de Graauw, T., 2000. A&A 359,

The receding torus model provides a successful 523.V enemans, B. et al., 2002. ApJ 569, L11.quantitative and physical framework within which toW illott, C.J., Rawlings, S., Blundell, K.M., Lacy, M., 2000.unify broad- and narrow-line AGN. The quantitative

MNRAS 316, 449.aspect is important for studies of the high-redshift W illott, C.J., Rawlings, S., Archibald, E.N., Dunlop, J.S., 2002.Universe since neither optically-selected QSOs nor MNRAS 331, 435.radio /optically-selected radio galaxies represent a