Low-mass Black Holes in Active Galaxies jenny greene Yanfei Jiang, Aaron Barth, Ting Xiao, Luis Ho Malte Schramm, John Silverman Chien Y. Peng the Megamaser Cosmology Project: James Braatz, Cheng-Yu Kuo, Fred Lo, Jim Condon, Mark Reid, Lincoln Greenhill Monday, August 22, 2011 What is a “low-mass” black hole? 204 GU¨LTEKIN ET AL. Vol. 698 <106 M (< than the MW) ⦿ What are the smallest supermassive 109 BHs? Is a bulge required to host a BH? 108 What are the formation mechanisms of seed BHs? What might we see with a LISA-like 107 gravity wave experiment? Gultekin et al. 2009 Monday, August 22, 2011 Figure1.M–σ relationforgalaxieswithdynamicalmeasurements.ThesymbolindicatesthemethodofBHmassmeasurement:stellardynamical(pentagrams),gas dynamical(circles),masers(asterisks).Arrowsindicate3σ upperlimitstoBHmass.Ifthe3σ limitisnotavailable,weplotitatthreetimesthe1σ orat1.5times 68 68 68 the2σ limits.Forclarity,weonlyploterrorboxesforupperlimitsthatareclosetoorbelowthebest-fitrelation.ThecoloroftheerrorellipseindicatestheHubble 68 typeofthehostgalaxy:elliptical(red),S0(green),andspiral(blue).Thesaturationofthecolorsintheerrorellipsesorboxesisinverselyproportionaltotheareaof theellipseorbox.Squaresaregalaxiesthatwedonotincludeinourfit.Thelineisthebestfitrelationtothefullsample:M 108.12 M (σ/200kms 1)4.24.The BH − mass uncertainty for NGC 4258 has been plotted much larger than its actual value so that it will show on this plot. For clarit=y, we omit!labels of some galaxies in crowdedregions. relationfromsampleS.Thedistributionoftheresidualsappears M–σ or even that a power-law form is the right model. The consistentwithanormalorGaussiandistributioninlogarithmic only assumption required is that the ridge line of any M–σ mass, although the distribution is noisy because of the small relation that may exist does not change substantially across numbers.Foramoredirecttestofnormalitywelookatlog(M ) the range of velocity dispersion. The mean of the logarithmic BH ingalaxieswithσ between165and235 km s 1,corresponding mass in solar units is 8.16, and the standard deviation is e − to a range in log(σ /200 km s 1) from approximately 0.075 0.45. The expected standard deviation in mass is 0.19, based e − to 0.075. The predicted masses for the 19 galaxies−in this on the rms dispersion of log(σ /200 km s 1) (0.046) in this e − narrow range differ by at most a factor of 4.3, given our rangetimestheM–σ slopeβ;thusthevariationintheridgeline best-fit relation. The power of having a large number of of the M–σ relation in this sample is negligible compared to galaxies in a narrow range in velocity dispersion is evident the intrinsic scatter. We perform an Anderson–Darling test for here, as there is no need to assume a value for the slope of normality with unknown center and variance on this sample of What is a “low-mass” black hole? 204 GU¨LTEKIN ET AL. Vol. 698 <106 M (< than the MW) ⦿ What are the smallest supermassive 109 BHs? Is a bulge required to host a BH? 108 What are the formation mechanisms of seed BHs? What might we see with a LISA-like 107 gravity wave experiment? Gultekin et al. 2009 Monday, August 22, 2011 Figure1.M–σ relationforgalaxieswithdynamicalmeasurements.ThesymbolindicatesthemethodofBHmassmeasurement:stellardynamical(pentagrams),gas dynamical(circles),masers(asterisks).Arrowsindicate3σ upperlimitstoBHmass.Ifthe3σ limitisnotavailable,weplotitatthreetimesthe1σ orat1.5times 68 68 68 the2σ limits.Forclarity,weonlyploterrorboxesforupperlimitsthatareclosetoorbelowthebest-fitrelation.ThecoloroftheerrorellipseindicatestheHubble 68 typeofthehostgalaxy:elliptical(red),S0(green),andspiral(blue).Thesaturationofthecolorsintheerrorellipsesorboxesisinverselyproportionaltotheareaof theellipseorbox.Squaresaregalaxiesthatwedonotincludeinourfit.Thelineisthebestfitrelationtothefullsample:M 108.12 M (σ/200kms 1)4.24.The BH − mass uncertainty for NGC 4258 has been plotted much larger than its actual value so that it will show on this plot. For clarit=y, we omit!labels of some galaxies in crowdedregions. relationfromsampleS.Thedistributionoftheresidualsappears M–σ or even that a power-law form is the right model. The consistentwithanormalorGaussiandistributioninlogarithmic only assumption required is that the ridge line of any M–σ mass, although the distribution is noisy because of the small relation that may exist does not change substantially across numbers.Foramoredirecttestofnormalitywelookatlog(M ) the range of velocity dispersion. The mean of the logarithmic BH ingalaxieswithσ between165and235 km s 1,corresponding mass in solar units is 8.16, and the standard deviation is e − to a range in log(σ /200 km s 1) from approximately 0.075 0.45. The expected standard deviation in mass is 0.19, based e − to 0.075. The predicted masses for the 19 galaxies−in this on the rms dispersion of log(σ /200 km s 1) (0.046) in this e − narrow range differ by at most a factor of 4.3, given our rangetimestheM–σ slopeβ;thusthevariationintheridgeline best-fit relation. The power of having a large number of of the M–σ relation in this sample is negligible compared to galaxies in a narrow range in velocity dispersion is evident the intrinsic scatter. We perform an Anderson–Darling test for here, as there is no need to assume a value for the slope of normality with unknown center and variance on this sample of Evolution of massive black hole seeds 1083 of mass M in a galaxy with velocity dispersion σ : have distinct consequences for the properties of the MBH popula- BH c tion at z 0. Below, we present theoretical predictions of the various d f (t) f 1 α(t) #M˙ c2 −α = Edd Ed−d Edd seed models for the properties of the local SMBH population. , (7) dt = α t 109 L Q | | ! " " where here t is the time elapsed from the beginning of the accretion event. Solving this equation gives us the instantaneous Eddington 4.1 SMBHs in dwarf galaxies ratio for a given MBH at a specific time, and therefore we can The repercussions of different initial efficiencies for seed formation self-consistently grow the MBH mass. We set the Eddington ratio for the overall evolution of the MBH population stretch from high f 10 3 at t 0. This same type of accretion is assumed to Edd − redshift to the local Universe. Obviously, a higher density of MBH = = occur, at z > 15, following a major merger in which an MBH is not seeds implies a more numerous population of MBHs at later times, fed by disc instabilities. which can produce observational signatures in statistical samples. In a hierarchical universe, where galaxies grow by mergers, MBH More subtly, the formation of seeds in a $CDM scenario follows mergers are a natural consequence, and we trace their contribution the cosmological bias. As a consequence, the progenitors of massive to the evolving MBH population (cf. Sesana et al. 2007, for details galaxies (or clusters of galaxies) have a higher probability of host- on the dynamical modelling). During the final phases of an MBH ing MBH seeds (cf. Madau & Rees 2001). In the case of low-bias merger, emission of gravitational radiation drives the orbital decay systems, such as isolated dwarf galaxies, very few of the high-z pro- of the binary. Recent numerical relativity simulations suggest that genitors have the deep potential wells needed for gas retention and merging MBH binaries might be subject to a large ‘gravitational cooling, a prerequisite for MBH formation. We can read off directly recoil’: a general relativistic effect (Fitchett 1983; Redmount & from Fig. 1 the average number of massive progenitors required for Rees 1989) due to the non-zero net linear momentum carried away a present-day galaxy to host an MBH. In model A, a galaxy needs by gravitational waves in the coalescence of two unequal-mass BHs. of the order of 25 massive progenitors (mass above 107 M ) to Radiation recoil is a strong field effect that depends on the lack ∼ " ensure a high probability of seeding within the merger tree. In model of symmetry in the system. For merging MBHs with high spin, in C, instead, the requirement drops to four massive progenitors, in- particular orbital configurations, the recoil velocity can be as high as creasing the probability of MBH formation in lower bias haloes. afewthousandsofkilometerspersecond(Campanellietal.2007a,b; ThesignatureoftheefficiencyoftheformationofMBHseedswill Gonza´lez et al. 2007; Herrmann et al. 2007; Schnittman 2007). consequently be stronger in isolated dwarf galaxies. Fig. 3 (bottom Here, we aim to determine the characteristic features of the MBH panel) shows a comparison between the observed M –σ relation BH populationderivingfromaspecificseedscenario,anditssignaturein and the one predicted by our models (shown with circles) and, in present-day galaxies; we study the case without gravitational recoil. particular, from the left-hand to right-hand panel, the three models We discuss this issue further in Section 4. based on the Lodato & Natarajan (2006, 2007) seed masses with Q 1.5, 2 and 3, and a fourth model based on lower mass c = Population III star seeds. The upper panel of Fig. 3 shows the frac- 4 RESULTS tion of galaxies that do not host any MBHs for different velocity Detection of gravitational waves from seeds merging at the redshift dispersion bins. This shows that the fraction of galaxies without an of formation (Sesana et al. 2007) is probably one of the best ways to MBH increases with decreasing halo masses at z 0. A larger frac- What do we want to know? = discriminate among formation mechanisms. On the other hand, the tion of low-mass haloes are devoid of central BHs for lower seed imprint of different formation scenarios can also be sought in ob- formation efficiencies. Note that this is one of the key discriminants servations at lower redshifts. The various seed formation scenarios between our models and those seeded with Population III remnants. Volonteri et al. 2008 Figure 3. The M –velocity dispersion (σ ) relation at z 0. Every circle represents the central MBH in a halo of given σ . Observational data are marked BH c c = by their quoted error bars, both in σ and in M (Tremaine et al. 2002). Left-hand to right-hand panels: Q 1.5, Q 2,Q 3, Population III star seeds. c BH c c c = = = Top panels: fraction of galaxies at a given velocity dispersion which do not host a central MBH. C 2007 The Authors. Journal compilation C 2007 RAS, MNRAS 383, 1079–1088 $ $ Monday, August 22, 2011 Evolution of massive black hole seeds 1083 of mass M in a galaxy with velocity dispersion σ : have distinct consequences for the properties of the MBH popula- BH c tion at z 0. Below, we present theoretical predictions of the various d f (t) f 1 α(t) #M˙ c2 −α = Edd Ed−d Edd seed models for the properties of the local SMBH population. , (7) dt = α t 109 L Q | | ! " " where here t is the time elapsed from the beginning of the accretion event. Solving this equation gives us the instantaneous Eddington 4.1 SMBHs in dwarf galaxies ratio for a given MBH at a specific time, and therefore we can The repercussions of different initial efficiencies for seed formation self-consistently grow the MBH mass. We set the Eddington ratio for the overall evolution of the MBH population stretch from high f 10 3 at t 0. This same type of accretion is assumed to Edd − redshift to the local Universe. Obviously, a higher density of MBH = = occur, at z > 15, following a major merger in which an MBH is not seeds implies a more numerous population of MBHs at later times, fed by disc instabilities. which can produce observational signatures in statistical samples. In a hierarchical universe, where galaxies grow by mergers, MBH More subtly, the formation of seeds in a $CDM scenario follows mergers are a natural consequence, and we trace their contribution the cosmological bias. As a consequence, the progenitors of massive to the evolving MBH population (cf. Sesana et al. 2007, for details galaxies (or clusters of galaxies) have a higher probability of host- on the dynamical modelling). During the final phases of an MBH ing MBH seeds (cf. Madau & Rees 2001). In the case of low-bias merger, emission of gravitational radiation drives the orbital decay systems, such as isolated dwarf galaxies, very few of the high-z pro- of the binary. Recent numerical relativity simulations suggest that genitors have the deep potential wells needed for gas retention and merging MBH binaries might be subject to a large ‘gravitational cooling, a prerequisite for MBH formation. We can read off directly recoil’: a general relativistic effect (Fitchett 1983; Redmount & from Fig. 1 the average number of massive progenitors required for Rees 1989) due to the non-zero net linear momentum carried away a present-day galaxy to host an MBH. In model A, a galaxy needs by gravitational waves in the coalescence of two unequal-mass BHs. of the order of 25 massive progenitors (mass above 107 M ) to Radiation recoil is a strong field effect that depends on the lack ∼ " ensure a high probability of seeding within the merger tree. In model of symmetry in the system. For merging MBHs with high spin, in C, instead, the requirement drops to four massive progenitors, in- particular orbital configurations, the recoil velocity can be as high as creasing the probability of MBH formation in lower bias haloes. afewthousandsofkilometerspersecond(Campanellietal.2007a,b; ThesignatureoftheefficiencyoftheformationofMBHseedswill Gonza´lez et al. 2007; Herrmann et al. 2007; Schnittman 2007). consequently be stronger in isolated dwarf galaxies. Fig. 3 (bottom Here, we aim to determine the characteristic features of the MBH panel) shows a comparison between the observed M –σ relation BH populationderivingfromaspecificseedscenario,anditssignaturein and the one predicted by our models (shown with circles) and, in present-day galaxies; we study the case without gravitational recoil. particular, from the left-hand to right-hand panel, the three models We discuss this issue further in Section 4. based on the Lodato & Natarajan (2006, 2007) seed masses with Q 1.5, 2 and 3, and a fourth model based on lower mass c = Population III star seeds. The upper panel of Fig. 3 shows the frac- 4 RESULTS tion of galaxies that do not host any MBHs for different velocity Detection of gravitational waves from seeds merging at the redshift dispersion bins. This shows that the fraction of galaxies without an of formation (Sesana et al. 2007) is probably one of the best ways to MBH increases with decreasing halo masses at z 0. A larger frac- What do we want to know? = discriminate among formation mechanisms. On the other hand, the tion of low-mass haloes are devoid of central BHs for lower seed imprint of different formation scenarios can also be sought in ob- formation efficiencies. Note that this is one of the key discriminants servations at lower redshifts. The various seed formation scenarios between our models and those seeded with Population III remnants. Volonteri et al. 2008 Figure 3. The M –velocity dispersion (σ ) relation at z 0. Every circle represents the central MBH in a halo of given σ . Observational data are marked BH c c = by their quoted eSrropr baacrs,eb odtheinnσsaitnideinsM (Tremaine et al. 2002). Left-hand to rightN-haundmpabneels:rQ of o1.5f,fQ-nu2c,lQear3 ,BPoHpuslation III star seeds. c BH c c c = = = Top panels: fraction of galaxies at a given velocity dispersion which do not host a central MBH. C 2007 The Authors. Journal compilation C 2007 RAS, MNRAS 383, 1079–1088 $ Scaling relation$s Monday, August 22, 2011 MW What is a “low-mass” black hole? Origin of BH-galaxy scaling Elliptical relations: Scatter in relations as function of BH mass AND galaxy morphology S0 Spiral Then our interest is in all systems with M < 107 M where the BH ⦿ widest range of galaxy morphology is found M32 Gültekin et al. 2009 small elliptical Monday, August 22, 2011 MW What is a “low-mass” black hole? Origin of BH-galaxy scaling Elliptical relations: Scatter in relations as function of BH mass AND galaxy morphology S0 Spiral Then our interest is in all systems with M < 107 M where the BH ⦿ widest range of galaxy morphology is found M32 Gültekin et al. 2009 small elliptical Monday, August 22, 2011 How do we find BHs with M < 106 M ? BH ⦿ Monday, August 22, 2011 Dynamical BH masses for <106 M BHs only ⦿ possible within a few Mpc Gebhardt et al. (2001) Gebhardt, Rich, & Ho NGC 404 (2005) M ~ 105 M BH ⦿ Seth et al. 2010 2 BARTH ET AL. however, the black hole content of nuclear clusters in late- type spirals is very poorly constrained, and M33 provides a TABLE 1 HUBBLE SPACE TELESCOPE ARCHIVAL DATA convincing demonstration that some nuclear clusters do not C G1 contain a black hole of dynamically significant mass. The o Camera Filter Exposure Time (s) Observation Date best evidence that black holes can occur in at least some very M ~ 104 M n BH ⦿ t late-type disk galaxies comes from the detection of a small WFPC2/PC F606W 2×80 1994-08-17 r × o M33 ACS/WFC1 F435W 3 360 2003-02-03 number of active galactic nuclei (AGNs) in Scd and Sd-type × v F555W 3 360 2003-02-03 M < 103 M e spirals. The best-studied case is the BSHd galaxy N⦿GC 4395, F814W 3×360 2003-02-03 r which contains a Seyfert 1 nucleus (Filippenko & Sargent NICMOS/NIC3 F190N 6×224 2007-03-31 Omega CGen1 si 1989; Filippenko & Ho 2003); it remains the only clear exam- a M ~ 4x104 M l BH ⦿ ple of a broad-lined AGN in a bulgeless galaxy. In addition, a Noyola et al. 2010 Valluri et al. (2005) few examples of Type 2 AGNs in very late-type spirals have σ 25 km s-1 = been detected recently in optical spectroscopic surveys, such * NGC 205 as NGC 1042 (Seth et al. 2008a; Shields et al. 2008) and UGC M = 2 x 104 M M < 104 M BH ⦿ 6192 (Barth et al. 2008). BH ⦿ A recent Spitzer spectroscopic observation of the Sd galaxy NGC 3621 by Satyapal et al. (2007) led to the discovery of an NGC 3621 AGN in this galaxy, based on the detection of [Ne V] emission Barth et al. 2008 lines at 14.3 µm and 24.3 µm. Since the ionization potential for ionization of Ne+3 to Ne+4 is 95 eV, ordinary H II regions MBH < 3x106 M⦿ are not expected to be significant sources of [Ne V] emission, FIG. 1.— HST images of the nucleus of NGC 3621. Left panel: A portion but a hard AGN continuum canMeoandsaiyl,y Aupgruostv 2i2d, e201th1e necessary of the WFPC2/PC F606W image. The rectangle surrounding the nucleus ionizing photons. As a result, the relative strengths of neon shows the position and size of the spectroscopic aperture used in the Keck observation. Right panel: A portion of the NICMOS/NIC3 F190N image. In infrared fine-structure lines from different ionization states of both panels, north is up and east is to the left. neon are useful as diagnostics of the ionization conditions view. One ACS pointing placed the nucleus on the WFC1 within AGN narrow-line regions (e.g., Spinoglio & Malkan CCD and the other pointing placed it on the WFC2. We se- 1992; Voit 1992; Sturm et al. 2002; Groves et al. 2006). Abel lected the WFC1 pointing since this image included a larger & Satyapal (2008) used the results of new photoionization region surrounding the nucleus. The NICMOS F190N ob- models to argue that the strength of the [Ne V] emission in servation was taken as the continuum image for an F187N NGC 3621 could only be plausibly explained by the presence (Paschen α) image; we do not use the F187N emission-line of an AGN, and not by an ordinary burst of nuclear star forma- image. tion. The detection of an AGN in NGC 3621 is of significant The NICMOS and ACS images were retrieved from the interest since it is one of the latest-type spirals known to host HST data archives and we use the standard pipeline-processed an active nucleus, making it an important target for further versions of these images. For the WFPC2 data, we use the observations to constrain its black hole mass and AGN ener- cosmic-ray cleaned and co-added image from the WFPC2 As- getics. Satyapal et al. (2007) note that there is no previously sociations archive. published optical spectrum of the nucleus of NGC 3621 suit- Figure 1 displays the central regions of the WFPC2 and able for emission-line classification. NICMOS images. The images show that the galaxy contains In this paper, we use archival HST images to show that a very compact and photometrically distinct nuclear star clus- NGC 3621 contains a well-defined and compact nuclear star ter. Surrounding the cluster is a smooth and nearly featureless cluster. A new optical spectrum of this star cluster is used to !! region of radius ∼ 1. 5 (or ∼ 50 pc). At larger radii, dust lanes examine the classification of the active nucleus, and to mea- and young star clusters become very prominent in the optical sure the stellar velocity dispersion of the cluster. We describe images, especially in the ACS F435W band. dynamical modeling of the nuclear cluster and the resulting constraints on the masses of both the cluster and the central 2.1.1. GALFIT Modeling black hole. We also examine the structure of the galaxy us- ing near-infrared images from 2MASS in order to determine To determine the structure of the nuclear cluster, we use whether a bulge is present in this late-type disk galaxy. For the 2-dimensional modeling package GALFIT (Peng et al. the distance to NGC 3621, we adopt D = 6.6 Mpc, based on 2002). Unfortunately, the nuclear star cluster is saturated in the Cepheid measurements of Freedman et al. (2001). At this the ACS/WFC F555W and F814W images, making it impos- !! distance, 1 corresponds to 32.0 pc. sible to use these images to derive structural parameters for the cluster. The ACS F435W image is not saturated, but in this 2. IMAGING DATA blue passband the dust lanes and massive stars in the circum- 2.1. Archival HST Data nuclear region are more prominent, making this band less eas- NGC 3621 has been observed numerous times with HST, ily suited to modeling. In the NICMOS image (with a scale of although many of the observations were of outer fields that do 0.!!2 pixel!1), the cluster is unresolved. Therefore, we use the not include the nucleus, taken as part of the HST Key Project WFPC2/PC image for the GALFIT decomposition. WFPC2 on the Cepheid distance scale (Freedman et al. 2001). We magnitudes listed below are on the Vegamag system, using a found three sets of images that did cover the galaxy nucleus: zeropoint of 22.887 for the F606W filter (Baggett et al. 1997). one with WFPC2, one with ACS, and one with NICMOS. The We used the Tiny Tim package Krist (1993) to create a parameters for each of these observations are listed in Table model point-spread function for the WFPC2/PC camera. The 1. The galaxy was observed twice with ACS/WFC, with iden- model was created for the F606W filter at 2× oversampling. tical exposure sequences taken in two different pointings that For the GALFIT modeling, we extracted a 101×101 pixel2 each placed the nucleus near the edge of the WFC field of region from the PC image, centered on the nuclear cluster. 278 WOO ET AL. Vol. 716 Reverberation Mapping...Hasn’t quite reached 1 Figure 6. Dependence of residuals from the M –σ relation (∆ log M (log VP+log f ) (α +β log σ/(200 km s )) on parameters related to the accretion state: BH BH − = − ∗ V /σ (top left); Eddington ratio (top right); V (bottom left); line dispersion σ (bottom right) of the Hβ line. In this plot, we adopt the local relation FWHM line FWHM such low mlinae sses yet (except for NGC 4395) with α 8.12, β 4.24 taken from Gu¨ltekin et al. (2009), and log f 0.72 as determined in Section 6.2. = = = the dependence since the line width of most of our objects is relatively small, σ < 2000 km s 1. By dividing our sample line − at σ 1500 km s 1 into two groups of similar sample size, line − = we separately measured the virial coefficient for narrower-line and broader-line AGNs. The difference in the virial coefficient is ∆ log f 0.1–0.2, which is not significant given the uncertainty see Aaron’s poster for = of 0.15 dex on the virial coefficient. LAMP II results W 5. DISCUSSION o o e We present the M –σ relation of the reverberation sample t BH a ∗ in Figure 7, using the slope, β 3.55 0.60, determined in l . = ± 2 Section 4.1 and the average virial coefficient, log f 0.72 0 1 # $ = ± 0.10, determined in Section 4.2. Compared to the local quiescent 0 galaxies, active galaxies follow a consistent M –σ relation BH ∗ with a similar slope and scatter. Note that the mean black hole mass of the reverberation sample ( log M /M 7.3 0.74) BH # $ = ± % is an order of magnitude smaller than that of quiescent galaxies ( log M /M 8.2 0.79). The slightly shallower slope BH # $ = ± % of the reverberation sample is consistent with the trend in the quiescent galaxies that the slope is shallower for galaxies with lower velocity dispersion (σ < 200 km s 1); see Gu¨ltekin − ∗ et al. (2009). In contrast, the slope of late-type quiescent galaxies (4.58 1.58) seems higher than the slope of our active Figure 7. M –σ relation of active galaxies with reverberation black hole BH ± Monday, August 22, 2011 ∗ galaxies, which are mainly late-type galaxies. However, given masses (blue), compared with non-active galaxies with dynamical black hole masses. The reverberation masses were determined assuming the virial coef- the uncertainty in the slope (β 3.55 0.60) of the active = ± ficient, log f 0.72 0.10. The solid line is the best-fit slope of the active galaxy M –σ relation, the difference between quiescent and = ± BH galaxies, while the dashed line is the best fit to the inactive galaxy samples from ∗ active galaxies is only marginal. We did not attempt to divide Gu¨ltekin et al. (2009). our sample into various morphology groups or a few mass bins (A color version of this figure is available in the online journal.) to test the dependence of the slope, since the sample size is still small and biased toward lower mass objects.