Radiatively Inefficient Accretion in Nearby Galaxies
aa r X i v : . [ a s t r o - ph . GA ] J un T O APPEAR IN
The Astrophysical Journal .Preprint typeset using L A TEX style emulateapj v. 26/01/00
RADIATIVELY INEFFICIENT ACCRETION IN NEARBY GALAXIES L UIS
C. H O The Observatories of the Carnegie Institution of Washington, 813 Santa Barbara Street, Pasadena, CA 91101, USA
To appear in The Astrophysical Journal.
ABSTRACTWe use new central stellar velocity dispersions and nuclear X-ray and H α luminosities for the Palomar surveyof nearby galaxies to investigate the distribution of nuclear bolometric luminosities and Eddington ratios fortheir central black holes (BHs). This information helps to constrain the nature of their accretion flows and thephysical drivers that control the spectral diversity of nearby active galactic nuclei. The characteristic values ofthe bolometric luminosities and Eddington ratios, which span over 7–8 orders of magnitude, from L bol ∼ < to3 × erg s - and L bol / L Edd ≈ - to 10 - , vary systematically with nuclear spectral classification, increasingalong the sequence absorption-line nuclei → transition objects → low-ionization nuclear emission-line regions → Seyferts. The Eddington ratio also increases from early-type to late-type galaxies. We show that the verymodest accretion rates inferred from the nuclear luminosities can be readily supplied through local mass loss fromevolved stars and Bondi accretion of hot gas, without appealing to additional fueling mechanisms such as angularmomentum transport on larger scales. Indeed, we argue that the fuel reservoir generated by local processes shouldproduce far more active nuclei than is actually observed. This generic luminosity-deficit problem suggests thatthe radiative efficiency in these systems is much less than the canonical value of 0.1 for traditional optically thick,geometrically thin accretion disks. The observed values of L bol / L Edd , all substantially below unity, further supportthe hypothesis that massive BHs in most nearby galaxies reside in a low or quiescent state, sustained by accretionthrough a radiatively inefficient mode.
Subject headings: black hole physics — galaxies: active — galaxies: nuclei — galaxies: Seyfert INTRODUCTION
Simple considerations of the quasar population predict thatmassive black holes (BHs) should be common in a sizable frac-tion of present-day galaxies. From the integrated luminositydensity of quasars, one can estimate that a typical L ∗ galaxyshould, on average, contain a waste mass of ∼ M ⊙ lockedup in a BH (Sołtan 1982; Chokshi & Turner 1992). But whyare the quasar remnants so quiescent? Active galactic nuclei(AGNs) with quasarlike luminosities are absent at z ≈ η = 0.1 appropri-ate for geometrically thin, optically thick accretion disks (seeFrank et al. 1992), the BH has to consume 1–100 M ⊙ yr - inorder to generate luminosities of L bol = 10 –10 L ⊙ , as typ-ically seen in quasars. This level of gas supply is difficult tosustain in nearby galaxies. On the other hand, accretion ratesof ˙ M = 0.001–0.1 M ⊙ yr - do not appear implausible. Evenif angular momentum transport on nuclear scales is ineffectivein disk galaxies, this level of fueling can be supplied simplythrough local stellar mass loss (Ho et al. 1997c). Hence, for η = 0.1, there ought to be many nuclei shining as AGNs with L bol = 10 –10 L ⊙ . This is not observed. Only ∼ ˙ M and thus correspondingly larger luminosities, again contraryto observations (Fabian & Canizares 1988).The dilemma posed by the luminosity deficit in the nu-clei of nearby elliptical galaxies can be resolved by discardingthe premise that massive BHs are ubiquitous in these systems(Fabian & Canizares 1988). But this proposition is no longertenable in light of our current knowledge on the demography of central BHs based on direct dynamical searches (Magorrian etal. 1998; Ho 1999a; Kormendy 2004). Massive BHs appear tobe a generic component of galactic structure in most, if not all,systems with a bulge. Consistent with this picture, low-levelnuclear activity qualitatively resembling that of more luminousAGNs is found to be equally pervasive in nearby galaxies (Hoet al. 1997b; Ho 2008).Radiatively inefficient accretion flows (RIAFs; see Narayanet al. 1998; Quataert 2001 for reviews) provide an attractiveframework for solving the luminosity-deficit problem. In theregime when the mass accretion rate onto the central BH isvery low, the low-density, optically thin accreting medium can-not cool efficiently, and the accretion flow consequently puffsup into a quasi-spherical structure. Most relevant to the presentdiscussion, RIAFs attain radiative efficiencies much below thecanonical value of 0.1. RIAFs are characteristically dim. Opti-cally thin RIAFs are predicted to exist for accretion rates belowa critical threshold of ˙ M crit ≈ α ˙ M Edd ≈ . ˙ M Edd (Narayan et al.1998), where the Eddington accretion rate is defined by L Edd = η ˙ M Edd c , with η = 0 . L Edd = 1 . × (cid:0) M BH / M ⊙ (cid:1) ergs - ; the Shakura & Sunyaev (1973) viscosity parameter is takento be α ≈ . M BH - σ ∗ relation(Gebhardt et al. 2000; Ferrarese & Merritt 2000). The feeble1 HO TABLE 1Bolometric Corrections for Low-Luminosity AGNs
Galaxy Class D L L bol C H α C X Reference(Mpc) (erg s − ) Narrow Broad Total(1) (2) (3) (4) (5) (6) (7) (8) (9)NGC 1097 L2/L1.9 14.5 1.8 ×
529 474 250 34.4 1NGC 3031 S1.5/L1.5 3.6 2.1 ×
300 140 95 3.5 2NGC 4203 L1.9 9.7 9.5 ×
431 243 156 3.9 3NGC 4261 L2 30.0 1.7 × · · ·
355 14.2 3NGC 4374 L2 16.8 8.2 × · · ·
410 8.4 3NGC 4450 L1.9 16.8 3.4 × × · · ·
389 17.7 3NGC 4579 S1.9/L1.9 16.8 9.9 ×
353 380 184 7.7 3NGC 4594 L2 9.2 2.7 × · · ·
208 7.5 3NGC 6251 S2 92.0 8.0 × · · ·
36 2.9 4Arp 102B L1.2 96.6 2.9 ×
546 62 55 1.8 5Pictor A L1.5 140.1 1.3 × OTE .— Col. (1) Galaxy name. Col. (2) Spectroscopic classification of nucleus. Col. (3) Adopted distance. Col. (4)Bolometric luminosity. Col. (5) Ratio of bolometric to narrow H α luminosity. Col. (6) Ratio of bolometric to broad H α luminosity. Col. (7) Ratio of bolometric to total (narrow + broad) H α luminosity. Col. (8) Ratio of bolometric to X-rayluminosity in the 2–10 keV band. Col. (9) Reference for the H α data: (1) Storchi-Bergmann et al. (1993), narrow-lineluminosity corrected for an extinction of A V = 2 .
46 mag; (2) Ho et al. (1996); (3) Ho et al. (1997a); (4) Shuder &Osterbrock (1981); (5) Halpern et al. (1996); (6) Carswell et al. (1984). Data for cols. (2)–(4) and (8) come from Ho(1999b) and Ho et al. (2000). nuclear activity in nearby galaxies can be comfortably sustainedthrough stellar mass loss and spherical accretion of hot gas inthe inner regions of bulges. We show that the inferred accretionrates lie well within the critical threshold of RIAFs. Finally,we discuss the physical connection between accretion rates andnuclear spectral types. OBSERVATIONAL MATERIAL
Our analysis explicitly makes the following assumptions: (1)all galaxy bulges contain central BHs whose masses can bewell estimated using the recently established correlation be-tween BH mass and bulge stellar velocity dispersion; (2) thepresent level of activity of the BHs manifests itself as AGN-like emission-line nuclei; and (3) the accretion luminosity canbe extrapolated through the observed nuclear X-ray or opticalline luminosities.We focus on the sample of nuclei in the Palomar survey ofnearby galaxies, a magnitude-limited spectroscopic study of anearly complete sample of 486 bright ( B T ≤ . δ > ◦ ) galaxies (see Ho et al. 1997a, 1997b, and refer-ences therein). All of the galaxies have been assigned a nu-clear spectroscopic classification using a set of uniform crite-ria, into the following classes (see Ho et al. 1997a for details):absorption-line nuclei (A), H II nuclei (H), Seyfert nuclei (S), low-ionization nuclear emission-line regions (LINERs; L), andtransition objects (T; LINER/H II composites). This paper con-siders all classes except the H II nuclei. The Palomar surveycontains 277 galaxies classified as absorption-line (66), Seyfert(52), LINER (94), and transition nuclei (65).2.1. Nuclear Luminosities
Our study requires bolometric luminosities for a well-definedsample of AGNs covering a wide range of power. BecauseAGNs emit a very broad spectrum, their bolometric luminosi-ties ideally should be measured directly from their broadbandspectral energy distributions (SEDs). In practice, however,complete SEDs are not readily available for most AGNs, leastof all for low-luminosity sources such as LINERs, which aremost prevalent in nearby galaxies. The largest existing compila-tions of broadband SEDs for low-luminosity AGNs (Ho 1999b;Ho et al. 2000; Maoz 2007) contain only a limited number ofobjects.For this study, we circumvent this difficulty by using twomeasures of the nuclear power to estimate the AGN bolomet-ric luminosity, one based on the H α emission line and anotherbased on X-rays. Although the H α luminosity comprises onlya small percentage of the total power, its fractional contribu-tion to the bolometric luminosity, as shown below, turns outCCRETION IN NEARBY GALAXIES 3to be fairly well defined. Moreover, unlike most other spec-tral windows, H α measurements are readily available for large,relatively complete samples of nuclei. The Palomar databasegives H α luminosities measured through an aperture of 2 ′′ × ′′ centered on the nucleus, which corresponds to a linear scaleof ∼
200 pc ×
400 pc for a typical distance of 20 Mpc. Asexplained in Ho et al. (2003a), some of the H α luminositiespublished in Ho et al. (1997a) have been updated with moreaccurate values from the literature. We will also make use ofupper limits for the H α luminosity of the absorption-line nucleithat were not given explicitly in Ho et al. (1997a); the limitswere calculated from the equivalent-width detection limit of thesurvey in conjunction with the 6600 Å continuum flux densitymeasurements, assuming that the line has a typical full widthat half-maximum (FWHM) of 250 km s - . A supplementarylist of H α luminosities, including the upper limits, is given inHo et al. (2003a). In total, H α luminosities, or upper limitsthereof, are available for 246 objects, which account for 80%,98%, 89%, and 91% of the nuclear classes A, S, L, and T, re-spectively. Thus, H α measurements are available for the vastmajority of the objects considered in this study.The bolometric correction for H α , C H α = L bol / L H α , can beobtained in one of two ways. For luminous, type 1 AGNs, itis often expedient to estimate L bol from the optical continuumluminosity, frequently chosen at 5100 Å in the recent litera-ture: L bol = C λ L λ (5100). Now, the H α luminosity corre-lates strongly with the optical continuum, including λ L λ (5100)(Greene & Ho 2005); the correlation is slightly nonlinear.Choosing C = 9 . L bol = 2 . × ( L H α / erg s - ) . erg s - .The conversion pertains to the entire H α line, which in lumi-nous type 1 AGNs is dominated by the broad component. Be-cause of the reliance on the L H α - λ L λ (5100) correlation and theassumption that C is constant, the relation between L bol and L H α is formally slightly nonlinear. It is unclear how robust thisresult is and whether it can be extrapolated toward lower lumi-nosities. For the luminosity range of interest to us, it introducesan uncertainty of a factor of ∼ L H α = 10 , 10 , and 10 erg s - , C H α ≈ C H α empiricallyfrom the observed SEDs of low-luminosity AGNs, limitedthough they may be. We use the sample of 12 low-luminositySeyferts and LINERs with broadband SEDs studied by Ho(1999b) and Ho et al. (2000; see also Ho 2002b) as a guide. Thedata, summarized in Table 1, show that the median value of C H α ranges from 228 to 410, depending on whether we include onlythe narrow component of the line, only the broad component,or both. A reasonable compromise might be C H α ≈ ± β line has a significant scatter, especially for low-luminositysources. For sources with Eddington ratios below 0.1, and as-suming that on average H α /H β = 3.5 (Greene & Ho 2005), themedian value of C H α ≈
220 with an interquartile range of 160.This agrees reasonably well with the range of values obtainedfrom our calibration sample in Table 1. For concreteness, wewill simply adopt C H α = 300; none of the main conclusions inthis study depends critically on the exact value of the bolomet-ric correction.The H α luminosities of the narrow-line objects are subjectto a potential source of complication. From consideration of the relative strength of X-ray (2–10 keV) and H α emission indifferent classes of nearby AGNs, Ho (2008) shows that type 2sources (Seyfert 2s, LINER 2s, and essentially all of the tran-sition objects) have a general tendency to emit excess opticalline emission compared to their type 1 counterparts. He at-tributes this excess emission to extranuclear processes unasso-ciated with the active nucleus. If this interpretation is correct,then only a fraction of the narrow H α emission should be in-cluded in the budget for the nuclear luminosity. Ho (2008)finds that the median ratio of L X / L H α is 7.3 for Seyfert 1s and4.6 for LINER 1s, whereas Seyfert 2s, LINER 2s, and transi-tion objects have significantly lower values of 0.75, 1.6, and0.41, respectively. Assuming, for concreteness, that the in-trinsic L X / L H α ratio for the type 2 sources is equal to that ofLINER 1s, the H α luminosities for Seyfert 2s, LINER 2s, andtransition objects need to be reduced by a factor of 6.1, 2.9, and11.2, respectively.In light of the above source of ambiguity with the H α emis-sion, it would be desirable to have an alternative handle onthe nuclear luminosity. By far the most secure measure of nu-clear luminosity in AGNs comes from X-ray observations. Be-cause of the faintness of low-luminosity AGNs, the data mustbe of high enough angular resolution so that the nucleus can becleanly separated from the host galaxy (e.g., Ho et al. 2001;Flohic et al. 2006). As reviewed in Ho (2008), a substantialfraction of the galaxies in the Palomar survey have now beenobserved in the X-rays. The Appendix gives a compilationof all pertinent X-ray measurements taken from the literature,as well as from new analysis of data taken from the Chandra public archives. The X-ray data are not nearly as complete asH α . Nevertheless, X-ray luminosities, or upper limits thereof,are now available for 175 out of the 277 objects in the parentsample (63%), which account for 47% of the absorption nu-clei, 68% of the LINERs, 83% of the Seyferts, and 57% of thetransition objects. The incompleteness and heterogeneous na-ture of the X-ray measurements make it difficult to rigorouslyassess selection effects. However, if observational biases exist,they should be in the direction of missing very faint sources, aneffect that strengthens our main conclusions.As with the H α data, determining the appropriate bolometriccorrection for the X-ray band is not trivial. Since the SEDs ofAGNs vary strongly with accretion rate (Ho 1999b), we mustabandon the usual practice of adopting a single correction factorbased on the average SED of luminous quasars. Using, again,the small sample of low-luminosity AGNs with reliable broad-band SEDs, Table 1 shows that the median value of the bolo-metric correction in the 2–10 keV band is C X = L bol / L X ≈ L X is the luminosity in the 2–10 keV band, corrected, tothe extent possible, for absorption. The more extensive data setof L. C. Ho (in preparation) suggests a value larger by abouta factor of 2: sources with L bol / L Edd ∼ < C X = 15 .
8, with an interquartile range of 9.6. Because low-luminosity AGNs tend to be “X-ray-loud” (Ho 1999b) their val-ues of C X are significantly smaller than conventionally assumedfor luminous sources ( C X ≈
35; Elvis et al. 1994). This is con-sistent with the analysis of Vasudevan & Fabian (2007). For thepresent purposes, we will adopt C X = 15 .
8, noting, as before,that factors of a few variation in the bolometric correction donot affect the main conclusions of this study.The uncertainties on L bol are difficult to estimate. As dis-cussed in Ho et al. (1997a), the H α fluxes in the Palomar surveyhave typical errors of 30%–50%, reaching 100% in the worst HO F IG . 1.— Distribution of ( a ) bolometric luminosity, L bol , and ( b ) ratio of bolometric luminosity to the Eddington luminosity, L bol / L Edd , for all objects, Seyferts(L), LINERs (L), transition nuclei (T), and absorption-line nuclei (A). The hatched and open histograms denote detections and upper limits, respectively. Theoriginal sample is shown in blue, and the subsample restricted to Sab–Sbc ( T = 2 -
4) Hubble types is shown in red. The bolometric luminosity is based on the 2–10keV X-ray luminosity, assuming L bol = 15 . L X .F IG . 2.— Distribution of ( a ) bolometric luminosity, L bol , and ( b ) ratio of bolometric luminosity to the Eddington luminosity, L bol / L Edd , for galaxies binnedby Hubble type. The hatched and open histograms denote detections and upper limits, respectively. The bolometric luminosity is based on the 2–10 keV X-rayluminosity, assuming L bol = 15 . L X . cases. The largest source of uncertainty for the H α -based lumi-nosities, however, comes from our still-tentative knowledge of C H α (factor ∼
2) and the amount of extranuclear contaminationof the narrow-line emission (factor ∼ -
5, depending on thespectral class). In the X-ray band, we can be more confidentthat the flux is largely nuclear, and large-amplitude variability seems to be rather uncommon for the systems in question (Ho2008). Not all of the X-ray detections have sufficient counts forrigorous spectral fitting, but fortunately low-luminosity AGNsgenerally have small absorbing columns (Ho 2008). Still, at themoment we do not know C X to better than a factor of ∼
2. Weconservatively guess that the estimates of L bol based on H α CCRETION IN NEARBY GALAXIES 5 F IG . 3.— Distribution of ( a ) bolometric luminosity, L bol , and ( b ) ratio of bolometric luminosity to the Eddington luminosity, L bol / L Edd , for all objects, Seyferts(L), LINERs (L), transition nuclei (T), and absorption-line nuclei (A). The bolometric luminosity is based on the extinction corrected total (narrow + broad) H α luminosity, assuming L bol = 300 L H α . The blue histograms show the distributions after correcting the narrow-line sources for extranuclear contamination (see textfor details). The hatched and open histograms denote detections and upper limits, respectively. and X-rays measurements have uncertainties of 0.7 and 0.3 dex,respectively. 2.2. Black Hole Masses
The majority of the objects in our sample do not have di-rect, dynamically determined BH masses. We will estimate BHmasses using the tight correlation between BH mass and bulgestellar velocity dispersion (the M BH - σ ∗ relation: Gebhardt et al.2000; Ferrarese & Merritt 2000), as determined by Tremaine etal. (2002):log (cid:18) M BH M ⊙ (cid:19) = (4 . ± .
32) log (cid:18) σ ∗
200 km s - (cid:19) + (8 . ± . . (1)The intrinsic scatter of the above fit is estimated to be ∼ < σ e , the luminosity-weighted velocity disper-sion measured within the effective radius of the bulge. Sincewe do not have measurements of σ e for most of our galaxies,we use instead σ , the central velocity dispersion. Gebhardt etal. (2000) have shown that in general σ ≈ σ e within a scatterof ∼ II ] λ ∼
5% to 15%.An error of 10% in σ ∗ introduces an uncertainty of ∼ M BH . We assume that M BH has an uncertainty dominatedby the scatter of the M BH - σ ∗ relation, ∼ BOLOMETRIC LUMINOSITIES AND EDDINGTON RATIOS
Figure 1 shows the distributions of bolometric luminositiesand their values normalized with respect to the Eddington lumi-nosity, L bol / L Edd . To avoid the possible complication of extranu-clear contamination in the H α emission, we base the bolometricluminosities on the hard X-ray measurements. The statistics ofthe distributions are listed in Table 2. The four classes of nucleicomprise a sequence of increasing luminosity: A → T → L → S. Whereas LINER and transition nuclei have very similar H α luminosities (Ho et al. 2003a)—an effect that can be attributedto the H α emission in transition objects being boosted by non-nuclear sources (Ho 2008)—there is no doubt that in the hardX-ray band LINERs are more luminous than transition objects(median L bol = 3 . × vs. 6 . × erg s - ). Both LINERsand transition objects, in turn, are less powerful than Seyferts(median L bol = 2 . × erg s - ). This systematic trend be-comes even more sharply defined when we consider the Ed-dington ratios. As with L bol , the median value of L bol / L Edd forLINERs (6 . × - ) is a factor of 4 larger than for transition ob-jects (1 . × - ), both being 1–2 orders of magnitude smallerthan in Seyferts (1 . × - ). Notably, all galactic nuclei in thePalomar sample are highly sub-Eddington systems.The distribution of L bol is broadly similar for galaxies of dif-ferent morphological types (Fig. 2 a ). By contrast, L bol / L Edd in-creases mildly, but systematically, from early-type to late-typegalaxies (Fig. 2 b ). Since the various classes of emission-linenuclei in the Palomar survey are hosted by slightly differentHubble types (Ho et al. 2003a), it would be of interest to ex-amine the trends in L bol and L bol / L Edd after factoring out thedependence on Hubble type. This is illustrated by the red his-tograms in Figure 1, where we now restrict the comparison togalaxies with morphological types Sab–Sbc ( T = 2 - TABLE 2Sample Statistics a Sample
N N u L bol (erg s − ) L bol / L Edd
Mean Error Median Mean Error MedianAll (all types) 175 37 8.5 × × × × − × − × − S (all types) 43 3 3.4 × × × × − × − × − (Sab–Sbc) 15 0 5.9 × × × × − × − × − L (all types) 64 12 6.1 × × × × − × − × − (Sab–Sbc) 16 2 1.9 × × × × − × − × − T (all types) 37 10 3.0 × × × × − × − × − (Sab–Sbc) 17 6 3.5 × × × × − × − × − A (all types) 31 10 1.3 × × × × − × − × − E 41 10 4.6 × × × × − × − × − S0–S0/a 51 12 8.1 × × × × − × − × − Sa–Sbc 64 9 1.5 × × × × − × − × − Sc–later 19 6 6.1 × × × × − × − × − The bolometric luminosities are based on the 2–10 keV X-ray luminosity, assuming L bol = 15 . L X . Statistics forsubsamples containing upper limits (whose number is denoted by N u ) computed using the Kaplan-Meier product-limit estimator (Feigelson & Nelson 1985). Our sample contains four galaxies (NGC 1275, 4261, 4374, and 4486) withstrong FR I-type (Fanaroff & Riley 1974) radio jets, as noted in Table 3. The nuclear X-ray luminosities of theseobjects might be partly contaminated by emission from the jet component. Excluding these four galaxies does notsignificantly change the statistics of this table. distributions of morphological types for all three AGN sub-classes. The overall trends of the parent sample are preserved(see also Table 2).Although in this study we give preference to the X-ray lu-minosities over the H α luminosities because of concerns overextranuclear line contamination, the H α data have the advan-tage of being uniform and nearly complete. Figure 3 repeats theanalysis of Figure 1, but now using bolometric luminosities de-rived from H α . The black histograms show the H α luminositiesas observed, converted to L bol assuming a bolometric correctionof C H α = 300. The blue histograms plot the same data with astatistical correction for extranuclear line emission applied tothe Seyfert 2s, LINER 2s, and transition objects based on theirobserved L X / L H α ratio (see § 2.1). Not surprisingly, the overalltrends seen in Figure 1 are well mirrored in Figure 3, but nowthey are delineated with better statistics.Finally, we turn to the distribution of M BH and L bol vs. L bol / L Edd (Fig. 4). In these diagrams, we have marked the var-ious subclasses of nuclei, and we have included the large sam-ple of z < .
35 high-luminosity AGNs (Seyfert 1 nuclei andlow-redshift quasars) selected by Greene & Ho (2007a) fromthe Sloan Digital Sky Survey (SDSS). Greene & Ho derivedBH masses and bolometric luminosities using the width andstrength of the broad H α line. We point out several salient fea-tures.1. Considered collectively, there is no dependence of M BH on L bol / L Edd : at a given value of M BH , which mostly fallin the range 10 - M ⊙ , L bol / L Edd spans ∼ ∼ M BH and L bol / L Edd . Thisarises because L bol spans a narrower range of values than M BH .3. At a fixed value of M BH , L bol / L Edd increases systemati-cally along the sequence absorption-line nuclei → tran-sition objects → LINERs → low-luminosity Seyferts → high-luminosity Seyferts and quasars. There is consid-erable overlap among the classes. The apparent gap in L bol / L Edd between the Palomar and SDSS sample maybe an artifact of observational selection effects; the twosurveys have very different sensitivity limits (Ho 2008).4. All emission-line nuclei in nearby galaxies are sub-Eddington systems, with the vast majority having L bol / L Edd ≪
1. All LINERs and transition nuclei arecharacterized by L bol / L Edd < - .5. The combined distribution of L bol or L bol / L Edd for thePalomar sample shows no evidence for bimodality orother indications of an abrupt transition between low-ionization (LINERs and transition objects) and high-ionization (Seyferts) sources.6. Because L bol spans a much larger range than M BH , L bol broadly increases with increasing L bol / L Edd . Low lu-minosity generally corresponds to low Eddington ra-tios. But there are important exceptions. A minor-ity of AGNs have low luminosities because they havelow BH masses, not necessarily low Eddington ratios.Within the Palomar sample, NGC 4395 (Filippenko &Ho 2003) provides a good example (Fig. 4 b ), and sim-ilar types of low-mass AGNs have been discovered inSDSS (Greene & Ho 2004, 2007b).CCRETION IN NEARBY GALAXIES 7 F IG . 4.— Distribution of ( a ) BH masses and ( b ) L bol vs. L bol / L Edd for objects separated by spectral classification. The bolometric luminosity is based on the2–10 keV X-ray luminosity, assuming L bol = 15 . L X . The symbols are identified in the legend. The objects marked as “QSO” refer to the sample of high-luminositySeyfert 1 nuclei and quasars studied by Greene & Ho (2007a). Line segments denote upper limits. SOURCES OF FUEL
In this section we make some rough estimates of the mini-mum amount of fuel likely to be available in the central regionsof nearby galaxies. For the moment, let us neglect any contribu-tion due to dissipation from a large-scale disk, external acqui-sition from tidal interactions and infall, or to discrete, episodicevents such as tidal disruptions of stars. Galactic nuclei can befed, in a steady state manner from the inner bulge of the galaxy,by at least two sources: (1) ordinary mass loss from evolvedstars and (2) gravitational capture of gas from the hot interstel-lar medium. 4.1.
Stellar Mass Loss
Present-day elliptical galaxies and the bulges of S0s and spi-rals contain mostly old, evolved stars. Red giants and planetarynebulae return a significant fraction of their mass to the inter-stellar medium through mass loss. For a Salpeter stellar initialmass function with a lower mass cutoff of 0.1 M ⊙ , an uppermass cutoff of 100 M ⊙ , solar metallicities, and an age of 15Gyr, Padovani & Matteucci (1993) estimate ˙ M ∗ ≈ × - (cid:18) LL ⊙ , V (cid:19) M ⊙ yr - . (2)This result is consistent, within a factor of ∼
2, with the workof Faber & Gallagher (1976), Ciotti et al. (1991), and Jung-wiert et al. (2001). Athey et al. (2002) obtained mid-infraredobservations to probe more directly the mass-losing stars in el-liptical galaxies. They find ˙ M ∗ ≈ × - (cid:0) L / L ⊙ , B (cid:1) M ⊙ yr - .For B - V ≈ ˙ M ∗ ≈ × - (cid:0) L / L ⊙ , V (cid:1) M ⊙ yr - , againclose to the estimate by Padovani & Matteucci (1993). Thus, weuse Padovani & Matteucci’s relation to convert between V -bandluminosity and mass loss rate. Hubble Space Telescope (HST) images have offered an un-precedently detailed view of the morphological structure of thecentral regions of nearby galaxies. The majority of galaxiescontain central density concentrations, either in the form of nu-clear cusps or photometrically distinct, compact stellar nuclei(e.g., Lauer et al. 1995; Phillips et al. 1996; Carollo et al. 1997;Faber et al. 1997; Rest et al. 2001; Ravindranath et al. 2001;Böker et al. 2002; Ferrarese et al. 2006; Kormendy et al. 2009).The cusp profiles continue to rise to the resolution limit of
HST (0 . ′′ ∼
10 pc at a distance of 20 Mpc. The nuclearstellar population in most instances is old (Ho et al. 2003a;Sarzi et al. 2005; Zhang et al. 2008). How much gaseous ma-terial is available through stellar mass loss? We note that theuncertainties associated with the effectiveness of angular mo-mentum transport on large (1–10 kpc) or intermediate (0.1–1kpc) scales are bypassed by focusing only on nuclear ( ∼ <
10 pc)scales. Although the fate of the nuclear gas is not entirely clear,it is important to recognize that stellar mass loss confined tothe nuclear cusp or nuclear cluster does furnish a steady state, in situ supply of gas that is in principle available for accretion.Shortly after being shed, the stellar gaseous envelopes quicklybecome thermalized with the hot ambient medium of the bulge,but some of the gas remains cool (Parriott & Bregman 2008).Even after the stellar ejecta joins the hot phase, the cooling timeis sufficiently short in the inner region of the bulge that a cool-ing flow should develop (Mathews 1990).The three main Local Group galaxies (M31, M32, and M33)serve as instructive examples. From the work of Lauer etal. (1998), the central stellar densities of all three galax-ies rise steeply toward the center as ρ ∝ r - . ± . ; at r = 0 . ρ ≈ . ± . M ⊙ pc - . More typi-cally, for galaxies beyond the Local Group, HST data probe r ≈
10 pc, where ρ ≈ - L ⊙ , V pc - for the “core” el-lipticals and ρ ≈ - L ⊙ , V pc - for the “power-law” el-lipticals and bulges of early-type spirals and S0s (e.g., Faber HOet al. 1997). Within a spherical region of r = 10 pc, thediffuse cores have L ≈ × - × L ⊙ , V , which yields ˙ M ∗ ≈ × - - × - M ⊙ yr - ; for the denser power-lawcusps, L ≈ × - × L ⊙ , V , or ˙ M ∗ ≈ × - - × - M ⊙ yr - . Centrally dominant nuclear star clusters, present ina large fraction of disk galaxies, typically have luminosities L ≈ L ⊙ (Lauer et al. 1995; Carollo et al. 1997; Bökeret al. 2002), and hence ˙ M ∗ ≈ - M ⊙ yr - .4.2. Bondi Accretion
The inner regions of ellipticals and bulges contain X-ray-emitting plasma, sustained through thermalized ejecta fromstellar mass loss (Mathews 1990), with temperatures character-istic of the virial velocities of the stars, ∼ - K. This dif-fuse, hot gas holds another plentiful fuel reservoir for accretion,through the mechanism described by Bondi (1952). The Bondiaccretion rate depends on the gas density and temperature atthe accretion radius, R a ≈ GM / c s , where c s ≈ . T / km s - is the sound speed of the gas. From the continuity equation, ˙ M B = 4 π R a ρ a c s , where ρ a is the gas density at R a . Expressedin terms of typical observed parameters (see below), ˙ M B ≈ . × - (cid:18) M BH M ⊙ (cid:19) (cid:16) n . - (cid:17)
200 km s - c s ! M ⊙ yr - . (3)Recent high-resolution X-ray observations using Chandra find that the diffuse gas in the central few hundred parsec re-gions of elliptical galaxies typically has temperatures of kT ≈ . - n ≈ . - . - (Di Matteo etal. 2001; Loewenstein et al. 2001; Sarazin et al. 2000; Pelle-grini 2005; Soria et al. 2006). Data for the bulges of spiral andS0 galaxies are more fragmentary. Chandra has so far resolvedthe hot gas in the centers of a handful of bulges, including theMilky Way (Sbc; Baganoff et al. 2003), M81 (Sab; Swartz etal. 2003), NGC 1291 (Sa; Irwin et al. 2002), and NGC 1553(S0; Blanton et al. 2001). The center of M31 (Sb) has beeninvestigated using both
XMM-Newton (Shirey et al. 2001) and
Chandra (Garcia et al. 2005). These studies suggest that bulgestypically have gas temperatures of kT ≈ . - . n ≈ . - .If, for simplicity, we assume that the hot gas in most bulgesis characterized by n = 0 . - and kT = 0 . ˙ M B ≈ - - - M ⊙ yr - for M BH = 10 - M ⊙ . In ellip-tical galaxies M BH ≈ - M ⊙ , and for n = 0 . - and kT = 0 . ˙ M B ≈ - - - M ⊙ yr - . We note that theseestimates of the Bondi accretion rates are probably lower lim-its because the actual densities near R a are likely to be higherthan we assumed. For the above fiducial temperatures and BHmasses, R a ≈ -
10 pc for bulges and ∼ -
100 pc for ellip-ticals, roughly an order of magnitude smaller than the typicallinear resolution achieved by
Chandra for nearby galaxies. Inthe case of the Galactic center, n = 26 cm - and kT = 1 . ′′ (0.4 pc), rising to n = 130 cm - and kT = 2 keV at 1 ′′ (Baganoff et al. 2003).4.3. Other Sources
The two processes discussed above—ordinary stellar massloss and Bondi accretion of hot gas—can be regarded as aconservative, steady state supply of fuel for galactic nuclei.Other sources, however, can raise this minimum level. In termsof purely stellar sources, some possibilities include (1) stellarmass loss enhanced by dynamical heating (Allen & Hughes1987; Armitage et al. 1996) or AGN irradiation (Edwards 1980;Scoville & Norman 1988; Voit & Shull 1988), (2) stellar-stellarcollisions in a dense nuclear cluster (Spitzer & Saslaw 1966;Frank 1978; Rauch 1999), and (3) tidal disruption of stars bythe central BH (Hills 1975; Rees 1988; Milosavljevi´c et al.2006). It is difficult to evaluate quantitatively the contributionthese effects would make to the total fuel budget of nearby nu-clei; we merely note that cumulatively they may significantlyboost the “baseline” accretion rate estimated above.We have also neglected any contribution from the cold phaseof the interstellar medium. Nonaxisymmetric perturbations dueto galaxy-galaxy tidal interactions, large-scale bars, nuclearbars, or nuclear spirals are often invoked as mechanisms forangular momentum transport of the cold gas in disk galaxies(e.g., Wada 2004). The effectiveness of these processes for fuel-ing nearby, relatively low-luminosity AGNs, however, has beenunclear. With respect to the well-studied Palomar survey, AGNactivity seems to be affected neither by large-scale bars (Ho etal. 1997c) nor by local galaxy environment (Schmitt 2001; Hoet al. 2003a). In any event, if dissipation of the cold gas doesoccur on nuclear scales, as inevitably it must at some level in atleast some objects, it would further add to the fuel supply. IMPLICATIONS
Accretion Flow and Radiative Efficiency
The results presented in this paper provide some importantinsights into the nature of BH accretion in nearby galactic nu-clei. We have established, for the first time using a large, sta-tistically robust sample, that virtually all massive BHs in thenearby Universe share two common properties: they have lowluminosities and radiate well below the Eddington limit. Thisholds for galaxies spanning a wide range of Hubble types andnuclear spectral classes. The median value of the bolometricluminosities are only L bol ≈ - erg s - , and the medianEddington ratios range from L bol / L Edd ≈ × - to 3 × - .The extreme dimness of these nuclei strongly suggests thattheir accretion flows are radiatively inefficient. In the con-text of the class of accretion models commonly called opti-cally thin RIAFs the accretion luminosity is given by (Ma-hadevan 1997) L acc = (cid:0) η/ . (cid:1)(cid:2) . ˙ m /α ) (cid:3) ˙ Mc , valid in theregime ˙ m > - α , where ˙ M = ˙ m ˙ M Edd and ˙ M Edd = 2 . × - (cid:0) η/ . (cid:1)(cid:0) M BH / M ⊙ (cid:1) M ⊙ yr - . This expression adopts thecanonical values of the microphysics parameters used by Ma-hadevan (1997). In the notation used in this paper, ˙ m ≃ . (cid:16) α . (cid:17) (cid:18) L bol L Edd (cid:19) / . (4)Thus, for L bol / L Edd = 10 - - - , ˙ m ≈ × - - × - ,which lie comfortably within the regime of optically thin RI-AFs, ˙ m ≤ ˙ m crit ≈ α ≈ . M BH ≈ - M ⊙ , are ˙ M ≈ - - - M ⊙ yr - . As originally formulated by Narayan & Yi (1994, 1995), this class of accretion disk models was called advection-dominated accretion flows. Subsequent workhas shown that such flows are inherently unstable to outflows and convection. To avoid delving into the technical details, which are unimportant for the present levelof discussion, we simply follow Quataert (2001) and refer to this class of models as RIAFs.
CCRETION IN NEARBY GALAXIES 9The accretion rates estimated in § 4 provide a more directargument that the radiative efficiency of the accretion flow,whatever its form, is likely to be low. The median bolomet-ric luminosities of the Palomar emission-line objects rangefrom ∼ × erg s - for transition objects to ∼ × ergs - for Seyferts (Table 2). If this emission is produced by acanonical optically thick, physically thin disk, which radiatesat L acc = η ˙ Mc = 5 . × (cid:0) η/ . (cid:1)(cid:0) ˙ M / M ⊙ yr - (cid:1) erg s - , weexpect typical mass accretion rates of ˙ M ≈ × - to 4 × - M ⊙ yr - . These values of ˙ M are miniscule by comparison withthe minimum accretion rates likely to be available through stel-lar mass loss and Bondi accretion alone. The majority of thePalomar AGNs are hosted by early-type disk galaxies (S0s andSa–Sbc spirals; see Ho et al. 1997b, 2003a), whose bulges tendto have cuspy light profiles of the “power-law” type. As dis-cussed in § 4.1, the inner regions of such bulges should have ˙ M ∗ ≈ - - - M ⊙ yr - , and probably closer to the upperend of this range because of the additional contribution fromcentral star clusters, which add ˙ M ∗ ≈ - M ⊙ yr - . We havealso erred on the side of caution by assuming that all of thestars are old. In actuality, the central regions of many spirals inthe Palomar sample often show evidence for some contributionfrom composite populations (Ho et al. 2009), which will helpto boost the mass loss rates even further. Bondi accretion of hotgas contributes roughly the same amount as stellar mass loss, ˙ M B ≈ - - - M ⊙ yr - . Thus, ˙ M tot = ˙ M ∗ + ˙ M B ≈ - - - M ⊙ yr - , but more likely, ˙ M tot ∼ > - M ⊙ yr - .A similar exercise leads to an even stronger result for theabsorption-line nuclei, which are found predominantly in el-liptical and S0 galaxies. Here, the median L bol of 2 × ergs - requires only ˙ M ≈ × - M ⊙ yr - for η = 0.1. On theother hand, the centers of the host galaxies can supply at least ˙ M ∗ ≈ - - - M ⊙ yr - for low-density cores, a factor of10 higher still in ˙ M ∗ for power-law cusps, and yet another 10-fold increase for ˙ M B . Ho et al. (2003b) highlighted the acute-ness of the luminosity-deficit problem for the nearest of theabsorption-line objects, M32, whose 2.5 × M ⊙ BH emitsmerely 9.4 × erg s - in the 2–10 keV band at an Eddingtonratio of ∼ × - .These simple comparisons suggest that if η indeed is 0.1,then nearby galactic nuclei are 1–4 orders of magnitude under-luminous. For the accretion rates that we infer to be present,they should radiate far more prodigiously than actually ob-served. There are four possible interpretations of this finding.(1) First, our estimates of ˙ M tot could be too high by a largefactor, namely 1–4 orders of magnitude. We consider this tobe unlikely. Recall that ˙ M tot includes only normal mass lossfrom evolved stars and Bondi accretion of hot gas, either oneof which alone would violate the luminosity limit for η = 0.1;we have conservatively neglected other potential sources of fuel(§ 4.3). (2) Second, it could be argued that perhaps the gas re-leased through stellar mass loss manages to escape from the nu-cleus before it gets accreted. In actively star-forming galaxies,for example, the collective effects of strong winds and shocksfrom massive stars can expel gas to large galactocentric dis-tances. This mechanism of gas removal, however, appears to beextremely implausible given the typical ages of the nuclear stel-lar population (Ho et al. 2003a). Storing the gas in an inert colddisk or converting it to young stars violates other observationalconstraints (Ho 2008). (3) Third, η may be ≪ .
1, as expectedfrom RIAFs (Narayan et al. 1998; Quataert 2001). This is anargument that is frequently invoked to explain the apparent con- flict between the nuclear luminosities and Bondi accretion ratesin some early-type galaxies (e.g., Fabian & Rees 1995; Ma-hadevan 1997; Di Matteo et al. 2000, 2001; Loewenstein et al.2001; Ho et al. 2003b). (4) Lastly, an inherent ambiguity existsbetween inefficient radiation and inefficient accretion. A va-riety of physical effects, summarized in Ho (2008), can divertthe inflowing gas and severely curtail the amount of materialthat ultimately gets accreted. For instance, if RIAFs naturallydevelop outflows or winds, as recently shown in a number ofstudies (see Quataert 2001), the actual accretion rate would bemuch lower than that estimated at large radii. If so, then η maynot need to be so exceptionally low, although it should still besubstantially below 0.1 because the outflow models ultimatelyrely on the accreting gas to be radiatively inefficient. Energyfeedback from the AGN, associated with either disk outflowsor small-scale radio jets—a ubiquitous feature of low- L bol / L Edd systems (Ho 2002a)—may be another culprit for interruptingsmooth mass inflow. Ho (2009) presents quantitative evidencethat AGN feedback injects nongravitational perturbations to thekinematics of the ionized gas in the Palomar sources.5.2.
Accretion States of Massive Black Holes
By analogy with stellar BHs in X-ray binaries, supermas-sive BHs in galactic nuclei may evolve through different evo-lutionary phases, corresponding to distinct “states” (Narayan etal. 1998). The basic physical picture is that the structure ofthe accretion flow changes in response to variations in the ac-cretion rate. Parameterizing the accretion rate in terms of thedimensionless variable ˙ m , one can define three, possibly fourregimes. (1) When ˙ m ∼ >
1, an object is in the “very high”state. The high radiation density in these “super-Eddington”sources traps the photons, and the accretion flow is that of anoptically thick RIAF or slim disk (Begelman & Meier 1982;Abramowicz et al. 1988). An extragalactic analog of such sys-tems are the narrow-line Seyfert 1 nuclei (Pounds & Vaughan2000). (2) Objects satisfying ˙ m crit < ˙ m < ˙ m ≤ ˙ m crit , anoptically thin, geometrically thick, radiatively inefficient flowdevelops, and the luminosity of the source plummets. Depend-ing on how low the accretion rate drops, one might distinguishbetween objects in the “low” state (10 - ≤ ˙ m ≤ ˙ m crit ) versusthose in the “quiescent” state ( ˙ m < ). Objects in quiescenceare those dominated by, or which exclusively contain, a pureRIAF, such as the Galactic center source Sgr A ∗ . Low-stateobjects contain a hybrid structure consisting of an inner RIAFplus an outer thin disk, whose truncation radius recedes as ˙ m decreases. Such a configuration has been suggested for a num-ber of low-luminosity AGNs, especially LINERs (Lasota et al.1996; Quataert et al. 1999; Ho et al. 2000; Ho 2002b, 2008).We note that the boundary between the low and quiescent states( ˙ m ≈ - ) is purely illustrative; it remains to be demonstratedthat there are two distinct states, and if so, where the transitiontruly lies.If AGN activity is characterized by distinct states, asschematically sketched above, then this ought to be reflected inthe observed distribution of accretion rates for AGNs spanningthe full range of ˙ m . Since it is difficult to measure ˙ m directly, L bol / L Edd can be used as a surrogate for ˙ m . We stress, however,that analysis of this kind is only meaningful when performed on0 HOlarge, well-defined, statistically complete samples. It is danger-ous to combine samples with different selection criteria (e.g.,Marchesini et al. 2004; Hopkins et al. 2006). With this inmind, we recall that the distribution of L bol / L Edd for the entirePalomar sample (Figs. 1 b and 3 b )—a more or less completecensus of nearby galaxies—shows no obvious substructure thatmight be identified with physically distinct populations. Unfor-tunately, the volume sampled by the Palomar survey does notcontain sufficient luminous AGNs to properly cover the upperend of the L bol / L Edd distribution. This would be an importantgoal for future statistical studies of AGNs. SUMMARY
Nearly all of the objects in the Palomar survey of nearbygalaxies now have central stellar velocity dispersions, fromwhich BH masses can be inferred using the M BH - σ ∗ relation.A previously published collection of H α line fluxes, in combi-nation with a newly assembled database of nuclear X-ray mea-surements and a reevaluation of the appropriate bolometric cor-rections, provides estimates of the accretion luminosity of thenuclei. We use these resources to evaluate the distribution ofbolometric luminosities and Eddington ratios for a large, well-defined sample of galactic nuclei in order to investigate the na-ture of accretion onto massive BHs in nearby galaxies.Nearby galactic nuclei span at least 7 orders of magnitude innuclear bolometric luminosities, from L bol < to ∼ × erg s - , and an even broader range in Eddington ratios, from L bol / L Edd ≈ - to 10 - . Both L bol and L bol / L Edd , but espe-cially the latter, decrease systematically along the followingspectral sequence: Seyferts → LINERs → transition objects → absorption-line nuclei. The spectral diversity of emission-line nuclei reflects and is primarily controlled by variations inthe mass accretion rate. The characteristic value of L bol / L Edd also varies systematically along the Hubble sequence, increas-ing from galaxies with large to small bulge-to-disk ratios. The accretion rates inferred from the nuclear luminosities,assuming a standard radiative efficiency of η = 0 .
1, are verylow, typically ˙ M ∼ < - to 10 - M ⊙ yr - . Such tiny rates caneasily be supplied in situ by ordinary mass loss from evolvedstars in the nuclear stellar cusp or central star clusters, or byBondi accretion of hot gas in the inner bulges of galaxies de-tected in X-ray observations. Indeed, we argue that conserva-tive estimates of the gas mass potentially available for accretionalready far exceed the observational limits imposed by the lu-minosity measurements, suggesting that in many, if not most,nearby galaxies the radiative efficiency is likely to be much lessthan 0.1. The requirement for exceptionally low radiative ef-ficiencies, however, could be mitigated if the actual accretionrates are curtailed by AGN feedback in the form of disk out-flows or small-scale radio jets. The prevalence of RIAFs is fur-ther supported by the low Eddington ratios: all the objects inour sample are sub-Eddington systems, with the majority hav-ing values of L bol / L Edd that satisfy the theoretically predictedcriterion for RIAFs.We suggest that massive BHs in galactic nuclei evolvethrough distinct states in response to changes in the mass accre-tion rate. The nearby objects considered in this study are largelysystems in the low or quiescent state. We see no evidence of bi-modality in the distribution of L bol / L Edd , but this is probably aconsequence of the limited volume probed by the Palomar sur-vey. It would be of considerable interest to extend the analysispresented here to include objects of higher luminosity in orderto map out the full distribution of AGN luminosities and Ed-dington ratios.This work was supported by the Carnegie Institution ofWashington and by
Chandra grant GO5-6107X. I thank Louis-Benoit Desroches for help with analyzing some of the archival
Chandra data. An anonymous referee offered a positive andhelpful report.
REFERENCESAbramowicz, M. A., Czerny, B., Lasota, J.-P., & Szuszkiewicz, E. 1988, ApJ,332, 646Allen, A. J., & Hughes, P. A. 1987, ApJ, 313, 152Armitage, P. J., Zurek, W. H., & Davies, M. B. 1996, ApJ, 470, 237Athey, A. E., Bregman, J. N., Bregman, J. D., Temi, P., & Sauvage, M. 2002,ApJ, 571, 272Baganoff, F. K., et al. 2003, ApJ, 591, 891Begelman, M. C., & Meier, D. L. 1982, ApJ, 253, 873Blanton, E. L., Sarazin, C. L., & Irwin, J. A. 2001, ApJ, 552, 106Böker, T., van der Marel, R. P., Laine, S., Rix, H.-W., Sarzi, M., Ho, L. C., &Shields, J. C. 2002, AJ, 123, 1389Bondi, H. 1952, MNRAS, 112, 195Cappi, M., et al. 2006, A&A, 446, 459Carollo, C. M., Stiavelli, M., de Zeeuw, P. T., & Mack, J. 1997, AJ, 114, 2366Carswell, R. F., Baldwin, J. A., Atwood, B., & Phillips, M. M. 1984, ApJ, 286,464Chokshi, A., & Turner, E. L. 1992, MNRAS, 259, 421Ciotti, L., D’Ercole, A., Pellegrini, S., & Renzini, A. 1991, ApJ, 376, 380Dahlem, M., & Stuhrmann, N. 1998, A&A, 332, 449Deluit, S., & Courvoisier, T. J.-L. 2003, A&A, 399, 77Desroches, L.-B., & Ho, L. C. 2009, ApJ, 690, 267Di Matteo, T., Johnstone, R. M., Allen, S. W., & Fabian, A. C. 2001, ApJ, 550,L19Di Matteo, T., Quataert, E., Allen, S. W., Narayan, R., & Fabian, A. C. 2000,MNRAS, 311, 507Dudik, R. P., Satyapal, S., Gliozzi, M., & Sambruna, R. M. 2005, ApJ, 620, 113Edwards, A. C. 1980, MNRAS, 190, 757Elvis, M., et al. 1994, ApJS, 95, 1Eracleous, M., Shields, J. C., Chartas, G., & Moran, E. C. 2002, ApJ, 565, 108Faber, S. M., & Gallagher, J. S. 1976, ApJ, 204, 365Faber, S. M., et al. 1997, AJ, 114, 1771Fabian, A. C., & Canizares, C. R. 1988, Nature, 333, 829Fabian, A. C., & Rees, M. J. 1995, MNRAS, 277, L55 Fanaroff, B. L., & Riley, J. M. 1974, MNRAS, 167, 31PFeigelson, E. D., & Nelson, P. I. 1985, ApJ, 293, 192Ferrarese, L., & Merritt, D. 2000, ApJ, 539, L9Ferrarese, L., et al. 2006, ApJS, 164, 334Filippenko, A. V., & Ho, L. C. 2003, ApJ, 588, L13Flohic, H. M. L. G., Eracleous, M., Chartas, G., Shields, J. C., & Moran, E. C.2006, ApJ, 647, 140Frank, J. 1978, MNRAS, 184, 87Frank, J., King, A. R., & Raine, D. J. 1992, Accretion Power in Astrophysics(Cambridge: Cambridge Univ. Press)Fukazawa, Y., Iyomoto, N., Kubota, A., Matsumoto, Y., & Makishima, K. 2001,A&A, 374, 73Fukugita, M., Shimasaku, K., & Ichikawa, T. 1995, PASP, 107, 945Garcia, M. R., Williams, B. F., Yuan, F., Kong, A. K. H., Primini, F. A., Barmby,P., Kaaret, P., & Murray, S. S. 2005, ApJ, 632, 1042Gebhardt, K., et al. 2000, ApJ, 539, L13González-Martín, O., Masegosa, J., Márquez, I., Guerrero, M. A., & Dultzin-Hacyan, D. 2006, A&A, 460, 45Greene, J. E., & Ho, L. C. 2004, ApJ, 610, 722——. 2005, ApJ, 630, 122——. 2007a, ApJ, 667, 131——. 2007b, ApJ, 670, 92Halderson, E. L., Moran, E. C., Filippenko, A. V., & Ho, L. C. 2001, AJ, 122,637Halpern, J. P., Eracleous, M., Filippenko, A. V., & Chen, K. 1996, ApJ, 464,704Hills, J. G. 1975, Nature, 254, 295Ho, L. C. 1999a, in Observational Evidence for Black Holes in the Universe,ed. S. K. Chakrabarti (Dordrecht: Kluwer), 157——. 1999b, ApJ, 516, 672——. 2002a, ApJ, 564, 120——. 2002b, in Issues in Unification of AGNs, ed. R. Maiolino, A. Marconi, &N. Nagar (San Francisco, CA: ASP), 165
CCRETION IN NEARBY GALAXIES 11 ——. 2008, ARA&A, 46, 475——. 2009, ApJ, in pressHo, L. C., Filippenko, A. V., & Sargent, W. L. W. 1996, ApJ, 462, 183——. 1997a, ApJS, 112, 315——. 1997b, ApJ, 487, 568——. 1997c, ApJ, 487, 591——. 2003a, ApJ, 583, 159Ho, L. C., Greene, J. E., Filippenko, A. V., & Sargent, W. L. W. 2009, ApJS, inpressHo, L. C., Rudnick, G., Rix, H.-W., Shields, J. C., McIntosh, D. H., Filippenko,A. V., Sargent, W. L. W., & Eracleous, M. 2000, ApJ, 541, 120Ho, L. C., Terashima, Y., & Ulvestad, J. S. 2003b, ApJ, 589, 783Ho, L. C., et al. 2001, ApJ, 549, L51Hopkins, P. F., Narayan, R., & Hernquist, L. 2006, ApJ, 643, 641Huchra, J. P., & Burg, R. 1992, ApJ, 393, 90Irwin, J. A., Sarazin, C. L., & Bregman, J. N. 2002, ApJ, 570, 152Iwasawa, K., Fabian, A. C., & Matt, G. 1997a, MNRAS, 289, 443Iwasawa, K., Fabian, A. C., Ueno, S., Awaki, H., Matsushita, K., & Makishima,K. 1997b, MNRAS, 285, 683Jungwiert, B., Combes, F., & Palous, J. 2001, A&A, 376, 85Komossa, S., Böhringer, H., & Huchra, J. P. 1999, A&A, 349, 88Kormendy, J. 2004, in Carnegie Observatories Astrophysics Series, Vol. 1:Coevolution of Black Holes and Galaxies, ed. L. C. Ho (Cambridge:Cambridge Univ. Press), 1Kormendy, J., Fisher, D. B., Cornell, M. E., & Bender, R. 2009, ApJS, in pressLasota, J.-P., Abramowicz, M. A., Chen, X., Krolik, J., Narayan, R., & Yi, I.1996, ApJ, 462, 142Lauer, T. R., Faber, S. M., Ajhar, E. A., Grillmair, C. J., & Scowen, P. A. 1998,AJ, 116, 2263Lauer, T. R., et al. 1995, AJ, 110, 2622Liu, J.-F., & Bregman, J. N. 2005, ApJS, 157, 59Loewenstein, M., Mushotzky, R. F., Angelini, L., Arnaud, K. A., & Quataert,E. 2001, ApJ, 555, L21Magorrian, J., et al. 1998, AJ, 115, 2285Mahadevan, R. 1997, ApJ, 477, 585Maoz, D. 2007, MNRAS, 377, 1696Marchesini, D., Celotti, A., & Ferrarese, L. 2004, MNRAS, 351, 733Mathews, W. G. 1990, ApJ, 354, 468McLure, R. J., & Dunlop, J. S. 2004, MNRAS, 352, 1390Milosavljevi´c, M., Merritt, D., & Ho, L. C. 2006, ApJ, 652, 120Narayan, R., Mahadevan, R., & Quataert, E. 1998, in The Theory of BlackHole Accretion Discs, ed. M. A. Abramowicz, G. Björnsson, & J. E. Pringle(Cambridge: Cambridge Univ. Press), 148Narayan, R., & Yi, I. 1994, ApJ, 428, L13——. 1995, ApJ, 444, 231Padovani, P., & Matteucci, F. 1993, ApJ, 416, 26Panessa, F., Bassani, L., Cappi, M., Dadina, M., Barcons, X., Carrera, F. J., Ho,L. C., & Iwasawa, K. 2006, A&A, 455, 173Pappa, A., Georgantopoulos, I., Stewart, G. C., & Zezas, A. L. 2001, MNRAS,326, 995Parriott, J. R., & Bregman, J. N. 2008, ApJ, 681, 1215 Pellegrini, S. 2005, ApJ, 624, 155Pellegrini, S., Siemiginowska, A., Fabbiano, G., Elvis, M., Greenhill, L., Soria,R., Baldi, A., & Kim, D. W. 2007, ApJ, 667, 749Pfefferkorn, F., Boller, Th., & Rafanelli, P. 2001, A&A, 368, 797Phillips, A. C., Illingworth, G. D., MacKenty, J. W., & Franx, M. 1996, AJ,111, 1566Pounds, K. A., & Vaughan, S. 2000, New Astron. Rev., 44, 411Quataert, E. 2001, in Probing the Physics of Active Galactic Nuclei byMultiwavelength Monitoring, ed. B. M. Peterson, R. S. Polidan, & R. W.Pogge (San Francisco, CA: ASP), 71Quataert, E., Di Matteo, T., Narayan, R., & Ho, L. C. 1999, ApJ, 525, L89Rauch, K. P. 1999, ApJ, 514, 725Ravindranath, S., Ho, L. C., Peng, C. Y., Filippenko, A. V., & Sargent, W. L. W.2001, AJ, 122, 653Rees, M. J. 1988, Nature, 333, 523Rest, A., van den Bosch, F. C., Jaffe, W., Tran, H., Tsvetanov, Z., Ford, H. C.,Davies, J., & Schafer, J. 2001, AJ, 121, 2431Reynolds, C. S., Nowak, M. A., Markoff, S., Tueller, J., Wilms, J., & Young,A. J. 2009, ApJ, 691, 1159Roberts, T. P., & Warwick, R. S. 2000, MNRAS, 315, 98Sansom, A. E., O’Sullivan, E., Forbes, D. A., Proctor, R. N., & Davis, D. S.2006, MNRAS, 370, 1541Sarazin, C., Irwin, J. A., & Bregman, J. N. 2000, ApJ, 544, L101Sarzi, M., et al. 2005, ApJ, 628, 169Satyapal, S., Dudik, R. P., O’Halloran, B., & Gliozzi, M. 2005, ApJ, 633, 86Satyapal, S., Sambruna, R. M., & Dudik, R. P. 2004, A&A, 414, 825Schmitt, H. R. 2001, AJ, 122, 2243Scoville, N., & Norman, C. 1988, ApJ, 332, 163Shakura, N. I., & Sunyaev, R. A. 1973, A&A, 24, 337Shirey, R., et al. 2001, A&A, 365, L195Shuder, J. M., & Osterbrock, D. E. 1981, ApJ, 250, 55Sivakoff, G. R., Sarazin, C. L., & Irwin, J. A. 2003, ApJ, 599, 218Soldatenkov, D., Vikhlinin, A., & Pavlinsky, M. 2003, Astron. Lett., 5, 298Sołtan, A. 1982, MNRAS, 200, 115Soria, R., Fabbiano, G., Graham, A., Baldi, A., Elvis, M., Jerjen, H., Pellegrini,S., & Siemiginowska, A. 2006, ApJ, 640, 126Spitzer Jr., L., & Saslaw, W. C. 1966, ApJ, 143, 400Storchi-Bergmann, T., Baldwin, J. A., & Wilson, A. S. 1993, ApJ, 410, L11Swartz, D. A., Ghosh, K. K., McCollough, M. L., Pannuti, T. G., Tennant, A. F.,& Wu, K. 2003, ApJS, 144, 213Terashima, Y., Iyomoto, N., Ho, L. C., & Ptak, A. F. 2002, ApJS, 139, 1Terashima, Y., & Wilson, A. S. 2003, ApJ, 583, 145Tremaine, S., et al. 2002, ApJ, 574, 740Vasudevan, R. V., & Fabian, A. C. 2007, MNRAS, 381, 1235Voit, G. M., & Shull, J. M. 1988, ApJ, 331, 197Wada, K. 2004, in Carnegie Observatories Astrophysics Series, Vol. 1:Coevolution of Black Holes and Galaxies, ed. L. C. Ho (Cambridge:Cambridge Univ. Press), 186Wrobel, J. M., Terashima, Y., & Ho, L. C. 2008, ApJ, 675, 1041Zhang, Y., Gu, Q.-S., & Ho, L. C. 2008, A&A, 487, 177
APPENDIXNUCLEAR X-RAY LUMINOSITIES
This Appendix summarizes the X-ray luminosities used in this paper. We performed a comprehensive literature search of allpublished X-ray measurements for the AGNs (Seyferts, LINERs, and transition objects) and absorption-line nuclei in the Palomarsurvey. Because of the faintness of most nearby nuclei and potential confusion with circumnuclear emission, the most crucialconsideration is angular resolution. With a point-spread function (PSF) of FWHM ≈ ′′ and low background noise, the instrumentof choice is ACIS on Chandra . Although less ideal, the HRI imager on
ROSAT and the MOS camera on
XMM-Newton , both havingPSFs with FWHM ≈ ′′ , also yield acceptable data under most circumstances. For sources bright enough for rigorous spectralanalysis, even lower resolution observations (e.g., ASCA ) can be used if the nucleus can be isolated through spectral fitting.Of the 277 galaxies in the parent sample, acceptable literature data were located for 166. Most of the observations (75%) wereacquired with
Chandra /ACIS, and the rest were taken largely with
ROSAT /HRI or
XMM-Newton . Only a small handful come from
ROSAT /PSPC and
ASCA ; one object was observed with
BeppoSAX . We also performed a thorough search of the
Chandra archivesand included all useful, unpublished data that were nonproprietary as of 2007 October. A total of nine additional galaxies werelocated. We analyzed these data sets following standard techniques, as described in Ho et al. (2001) and Desroches & Ho (2009).Table 3 lists X-ray luminosities for the final sample of 175 galaxies. Because the literature data were acquired with a varietyof different instruments and analyzed using many different techniques, we converted all the luminosities to one standard bandpass,2–10 keV. When reliable spectral fits are available, we use the published best-fit spectral slope to extrapolate to the desired bandpass.Otherwise, we assume a photon index of Γ = 1 .
8, which is close to the typical values observed in low-luminosity AGNs (see Ho2008, and references therein). Likewise, we quote intrinsic (absorption-corrected) luminosities whenever possible. Although manyof the fainter sources do not have sufficient counts to constrain the absorbing column, many lines of evidence suggest that mostlow-luminosity AGNs do not suffer from much obscuration (Ho 2008).CCRETION IN NEARBY GALAXIES 13
TABLE 3Nuclear X-ray Luminosities
Galaxy D L (Mpc) log ( L X /erg s − ) Tel. Ref. Galaxy D L (Mpc) log ( L X /erg s − ) Tel. Ref.(1) (2) (3) (4) (5) (1) (2) (3) (4) (5)IC 239 16.8 < < < < < < < < < < < a X 16 NGC 3516 38.9 42.39 X 16NGC 1358 53.6 42.68 B 17 NGC 3607 19.9 38.63 C 8NGC 1667 61.2 40.55 C 15 NGC 3608 23.4 38.85 C 8NGC 1961 53.1 40.31 R 3 NGC 3610 29.2 38.79 C 1NGC 2273 28.4 40.02 R 18 NGC 3623 7.3 38.25 C 14NGC 2300 31.0 40.93 R 18 NGC 3627 6.6 < < < < < TABLE 3—
Continued
Galaxy D L (Mpc) log ( L X /erg s − ) Tel. Ref. Galaxy D L (Mpc) log ( L X /erg s − ) Tel. Ref.(1) (2) (3) (4) (5) (1) (2) (3) (4) (5)NGC 3941 18.9 39.27 X 11 NGC 4435 16.8 39.60 R 18NGC 3982 17.0 38.76 C 16 NGC 4438 16.8 39.21 C 8NGC 3998 21.6 41.34 R 3 NGC 4450 16.8 40.02 C 21NGC 4013 17.0 < < < < a C 8NGC 4150 9.7 < < < < a C 8 NGC 4579 16.8 41.15 C 8NGC 4278 9.7 39.64 C 13 NGC 4594 20.0 40.69 C 8NGC 4291 29.4 40.63 R 18 NGC 4596 16.8 38.65 C 8NGC 4293 17.0 < < < < a C 8 NGC 4698 16.8 38.69 C 8NGC 4382 16.8 37.87 C 25 NGC 4713 17.9 38.40 C 2NGC 4388 16.8 42.14 A 33 NGC 4725 12.4 39.11 C 13NGC 4394 16.8 < < CCRETION IN NEARBY GALAXIES 15
TABLE 3—
Continued
Galaxy D L (Mpc) log ( L X /erg s − ) Tel. Ref. Galaxy D L (Mpc) log ( L X /erg s − ) Tel. Ref.(1) (2) (3) (4) (5) (1) (2) (3) (4) (5)NGC 5005 21.3 39.94 C 10 NGC 5850 28.5 < < < < < < < OTE .— Col. (1) Galaxy name. Col. (2) Adopted distance. Col. (3) X-ray luminosity in the 2–10 keV band. Col. (4) Telescope:A =
ASCA ; B =
BeppoSAX ; C =
Chandra /ACIS; R =
ROSAT /HRI; X =
XMM-Newton . Col. (9) Reference for the X-ray data:(1) Desroches & Ho (2009); (2)
Chandra archives, analyzed in this paper; (3) Roberts & Warwick (2000); (4) Halderson et al.(2001); (5) Ho et al. (2003b); (6) Garcia et al. (2005); (7) Terashima & Wilson (2003); (8) Gonz´alez-Mart´ın et al. (2006); (9)Eracleous et al. (2002); (10) Dudik et al. (2005); (11) Cappi et al. (2006); (12) Pellegrini et al. (2007); (13) Ho et al. (2001); (14)Satyapal et al. (2004); (15) Pappa et al. (2001); (16) Panessa et al. (2006); (17) Deluit & Courvoisier (2003); (18) Liu & Bregman(2005); (19) Satyapal et al. (2005); (20) Terashima et al. (2002); (21) Komossa et al. (1999); (22) Dahlem & Stuhrmann (1998);(23) Soria et al. (2006); (24) Pfefferkorn et al. (2001); (25) Sivakoff et al. (2003); (26) Loewenstein et al. (2001); (27) Pellegrini(2005); (28) Wrobel et al. (2008); (29) Soldatenkov et al. (2003); (30) Sansom et al. (2006); (31) Iwasawa et al. (1997a); (32)Fukazawa et al. (2001); (33) Iwasawa et al. (1997b); (34) Reynolds et al. (2009). aa