Infrared signature of active massive black holes in nearby dwarf galaxies
Francine R. Marleau, Dominic Clancy, Rebecca Habas, Matteo Bianconi
AAstronomy & Astrophysics manuscript no. marleau_AGNdwarfs_corr_printer c (cid:13)
ESO 2018July 20, 2018
Infrared signature of active massive black holesin nearby dwarf galaxies (cid:63)
Francine R. Marleau , Dominic Clancy , Rebecca Habas , and Matteo Bianconi , Institute of Astro and Particle Physics, University of Innsbruck, Technikerstraße 25, 6020 Innsbruck, Austriae-mail: [email protected]; [email protected]; [email protected] Astrophysics and Space Research Group, School of Physics and Astronomy, University of Birmingham,Edgbaston, Birmingham B15 2TT United Kingdom, e-mail: [email protected]
ABSTRACT
Context.
We investigate the possible presence of active galactic nuclei (AGN) in dwarf galaxies and other nearby galaxies to identifycandidates for follow-up confirmation and dynamical mass measurements.
Aims.
We identify candidate active central massive black holes (CMBH) using their mid-infrared emission, verify their nature usingexisting catalogues and optical line emission diagnostics, and study the relationship between their mass and the mass of their hostgalaxy.
Methods.
We use the Wide-field Infrared Survey Explorer (WISE) All-Sky Release Source Catalog and examine the infrared coloursof a sample of dwarf galaxies and other nearby galaxies in order to identify both unobscured and obscured candidate AGN by applyingthe infrared colour diagnostic. Stellar masses of galaxies are obtained using a combination of three independent methods. Black holemasses are estimated using the bolometric luminosity of the AGN candidates and computed for three cases of the bolometric-to-Eddington luminosity ratio.
Results.
We identify 303 candidate AGN, of which 276 were subsequently found to have been independently identified as AGN viaother methods. The remaining 9% require follow-up observations for confirmation. The activity is detected in galaxies with stellarmasses from ∼ to 10 M (cid:12) ; assuming the candidates are AGN, the black hole masses are estimated to be ∼ − M (cid:12) , adopting L bol = . L Edd . The black hole masses probed are several orders of magnitude smaller than previously reported for centrally locatedmassive black holes. We examine the stellar mass versus black hole mass relationship in this low galaxy mass regime. We find that itis consistent with the existing relation extending linearly (in log-log space) into the lower mass regime.
Conclusions.
These findings suggest that CMBH are present in low-mass galaxies and in the Local Universe, and provide new impetusfor follow-up dynamical studies of quiescent black holes in local dwarf galaxies.
Key words. galaxies: general – galaxies: Seyfert – galaxies: active – galaxies: dwarfs – galaxies: Local Group – infrared: galaxies
1. Introduction
Following pioneering work in the late 1960s (Zel’dovich &Novikov 1965; Salpeter 1964; Lynden-Bell 1969; Bardeen1970; Lynden-Bell & Rees 1971), a paradigm has graduallyemerged in which central massive black holes (CMBH) havecome to be regarded as an integral component of “most, if notall, massive galaxies,” intimately linked to their formation andevolution (Merritt & Ferrarese 2001; Bender & Kormendy2003; Ho, Kormendy & Murdin 2000; Peterson 2008; Peter-son, Somerville & Storchi-Bergmann 2010; Heckman & Kau ff -mann 2011; Schawinski 2012; Kormendy & Ho 2013). How-ever, despite a great deal of research and progress on manyfronts towards a better understanding of this paradigm (see e.g.Merloni 2015, for a recent review), essentially all of the fun-damental questions concerning the formation, growth, and hostco-evolution of CMBH remain unanswered. One such question,which is the focal point of this article, is whether CMBH aregeneric to all galaxies. It is interesting to note that while it isoften stated in the literature that black holes are to be found atthe centres of most galaxies, until recently there has been insu ffi - (cid:63) Tables 2 and 3 are only available in electronic form at theCDS via anonymous ftp to cdsarc.u-strasbg.fr (130.79.128.5) or viahttp: // cdsweb.u-strasbg.fr / cgi-bin / qcat?J / A + A / cient evidence to support this claim, even for the case of massivegalaxies not possessing a significant bulge component, let alonethe general population of galaxies, most of which are not mas-sive. In other words, it is not generally known whether CMBHare only a generic feature of certain galaxy types, present onlyin galaxies above a certain mass threshold, or subject to a com-bination of both of these restrictions.The question of CMBH genericity is strongly related to theissues of formation, growth and galaxy co-evolution. Its rela-tionship with the subject of galaxy formation can be understoodby considering the current observational constraints on high-redshift quasars (e.g. Fan 2001; Mortlock et al. 2011; Wu etal. 2015; Matsuoka et al. 2016), which imply that CMBH werepresent at very early times ( z >
7) and therefore must haveformed either concurrently with their host galaxies or prior tothem. While these constraints clearly highlight the close con-nection between the formation of CMBH and their hosts, theyalso conversely imply that galaxies which do not host a CMBHwill likely have undergone distinct formation processes. Simi-larly, in order to understand the link between CMBH genericityand evolution, one need only consider the evidence of the signifi-cant e ff ects that active galactic nuclei (AGN) have on their hosts’evolutions (Silk & Rees 1998; King 2003; Fabian 2012; Silk2013; Ishibashi & Fabian 2014) in order to realise that galax- Article number, page 1 of 14 a r X i v : . [ a s t r o - ph . GA ] J u l & A proofs: manuscript no. marleau_AGNdwarfs_corr_printer ies which do not host CMBH may have undergone a markedlydi ff erent evolution from those that do.It is apparent from these simple observations that the ques-tion of CMBH genericity has important implications for the the-ory and modelling of galaxy and large-scale structure evolution(see e.g. Benson 2010; Silk 2013, for recent reviews). Indeed,the importance of modelling AGN feedback has been under-stood for some time (Springel, Di Matteo & Hernquist 2005;Di Matteo, Springel & Hernquist 2005; McNamara & Nulsen2007) and various approaches are now incorporated as standardin simulations of galaxy and structure evolution (Sijacki et al.2007; Di Matteo et al. 2008; Khalatyan et al. 2008; Booth &Schaye 2009; Power, Nayakshin & King 2011; DeGraf et al.2012; Newton & Kay 2013). Feedback from AGN has also beenshown to play a significant role up to the largest scales of struc-ture Haider et al. (2016) (see also Wurster & Thacker 2013, fora comparison of models). In general, it is expected that theoriesand models which assume that black holes are generic will di ff ersignificantly in their predictions from those that do not. For ex-ample, while most models and simulations at present include thee ff ects of AGN feedback from massive galaxies, including thee ff ects of AGN feedback from low-mass galaxies may accountfor the observed low baryon fraction in Milky Way-type galaxiesat the present epoch (see e.g. Peirani et al. 2012).While the broad question of the genericity of CMBH remainsopen, recent findings have been strongly supportive of CMBHbeing generic for the subset of massive galaxies, i.e. for galaxieswith stellar masses above ∼ M (cid:12) . In particular, in a recentwork Marleau et al. (2013) studied a sample of 15 991 galax-ies and found the fraction of galaxies containing a CMBH to beapproximately the same for each morphological type. This studywas itself prompted by the recent discovery of CMBH in galax-ies lacking any substantial spheroidal component (Reines et al.2011; Simmons et al. 2013). Prior to this there was only strongevidence for the generic presence of CMBH in galaxies witha significant bulge component (Kormendy & Richstone 1995;Magorrian et al. 1998; Ferrarese & Merritt 2000; Gebhardt etal. 2000; Kormendy & Ho 2013). While evidence continues toaccrue to firmly establish the generic existence of CMBH withinmassive galaxies, attention has now begun to shift to the ques-tion of whether CMBH are also generic in the low-mass regime,i.e. in galaxies with stellar masses below ∼ M (cid:12) , or in otherwords, whether CMBH are generic to dwarf galaxies.Until very recently it was not even known whether any dwarfgalaxies contained CMBH, let alone whether they were generic.However, during the last few years results in this area have beenrapid. It is now known that a significant number of galaxies withmasses ∼ − M (cid:12) , i.e. at the high end of the dwarf galaxymass regime and the low end of the massive galaxy regime,likely contain active CMBH. In particular, Moran et al. (2014),Reines et al. (2013), Reines et al. (2014), Barth et al. (2008),Greene et al. (2007), Greene et al. (2004), Dong et al. (2012)and Dong et al. (2007) have detected the optical signatures ofmainly unobscured (type 1) AGN in 768 galaxies with total stel-lar masses in the range ∼ − M (cid:12) , while Marleau et al.(2013) have identified the first mid-infrared signatures of bothobscured and unobscured candidate AGN in 73 dwarf galaxieswith masses in the range ∼ − M (cid:12) .Even though these findings are encouraging, they barelyprobe the dwarf galaxy mass range, and it is also the case that theblack hole masses derived from the optical spectra may be sub-ject to large systematic uncertainties. Additionally, one wouldideally also like to have dynamically derived masses. At thepresent time, there is only one dynamical measurement of a black hole in a galaxy classified as a dwarf galaxy, i.e. with amass below 3 × M (cid:12) , namely the CMBH in NGC 4395, whichhas a mass of 4 × M (cid:12) (den Brok et al. 2015). Additionally,Seth et al. (2014) found a black hole of mass 2 . × M (cid:12) inthe ultra-compact dwarf galaxy M60-UCD1 of total stellar mass1 . × M (cid:12) , though this is believed to be a tidally strippedgalaxy whose progenitor had a mass of ∼ M (cid:12) , which wouldthen be consistent with the known black hole mass to host totalstellar mass scaling relation (Marleau et al. 2013).As dwarf galaxies exhibit a number of properties that di ff er-entiate them from other galaxies (see review in Mateo 1998) andare known to have di ff erent evolutionary histories from massivegalaxies, this could be a consequence of them not having CMBHor of having a di ff erent relationship with their CMBH. If it is thecase that dwarf galaxies generically contain CMBH, it is there-fore not necessarily the case that what has been learned aboutthe relationships between CMBH and their hosts in the massiveregime, as exhibited in their scaling relations, will necessarilyhold for dwarf galaxies. However, the extension of current scal-ing relations suggests that dwarf galaxies should host an inter-mediate mass CMBH, if they host one at all. The low surfacebrightness of dwarfs necessitates studying them in the nearbyuniverse; consequently, this population of galaxies provides theopportunity to discover very nearby, low-mass CMBH ( < M (cid:12) ), which would be ideal for follow-up dynamical studies.Motivated by our recent infrared (IR) study of active mas-sive black holes in galaxies of all morphological types (Mar-leau et al. 2013) and wide range of stellar masses, we haveundertaken a census in the very nearby Universe, specificallytargeting low-mass dwarf systems that should be ideal hostsfor CMBH in the mass range of intermediate mass black holes(IMBH; M BH ∼ − M (cid:12) ). Additionally, we are interestedin detecting AGN in nearby galaxies regardless of their mass sothat we can identify targets for future dynamical mass measure-ment. The structure of our paper is as follows. In Section 2, wedescribe our dwarfs and other nearby galaxies sample. In Sec-tions 3 and 4, this sample is matched to an infrared catalogue andthe AGN candidates are identified using the infrared colour di-agnostic. The detection of these AGN are verified using existingcatalogues and optical line emission diagnostic in Section 5 andtheir distance distribution is presented in Section 6. In Sections 7and 8, we derive stellar masses for the host galaxies of our AGNcandidates and compute the AGN fraction as a function of stel-lar mass based on our AGN selection method. In Sections 9 and10, we evaluate the black hole masses for our AGN candidates,present the black hole mass versus stellar mass scaling relation,and compare our black hole mass estimates to those derived us-ing other methods. In Section 11, we discuss and summarize ourresults. We believe that by firmly establishing the presence of ac-tive IMBHs not only at the low end of the galaxy mass functionbut also in the nearby Universe, it will be possible to follow-uptheir quiet counterparts with dynamical studies and hence pro-vide further support for their existence.
2. Nearby galaxy sample
Our primary sources were selected from the Updated NearbyGalaxy Catalogue of Karachentsev et al. (2013) and the cat-alogue of Local Group (LG) galaxies of McConnachie et al.(2012). The first is a recently updated all-sky catalogue con-taining 869 nearby galaxies with distance estimates within 11Mpc or corrected radial velocities less than 600 km s − . The lat-ter contains all known galaxies with distances determined frommeasurements of resolved stellar populations that place them Article number, page 2 of 14arleau et al.: CMBH in Nearby Dwarf Galaxies
Fig. 1.
Left: W − W W − W black open triangles ). The 303 sources above our IR colour cutare shown as red filled triangles . The two outliers with very red colours in the top right-hand corner of the diagram ( magenta filled circles ) arethe low-metallicity and heavily obscured BCDs that were originally identified by Gri ffi th et al. (2011). The two famous dwarf Seyfert 1 galaxies(NGC 4395 and POX 52), as well as a Seyfert 1 galaxy hosting a low-mass BH (UM 625) ( yellow filled circles ), have WISE colours above ourselection cut-o ff . Middle:
Same as left, but showing the AGN that have been previously identified in the MRBGD optical samples split into 549type 1 (BL; green filled triangles ) and 152 type 2 (NL; blue filled triangles ). Although a large fraction of these optically identified AGN fall inour IR selected sample of 303 galaxies, many also have WISE colours below our selection cut-o ff ( dashed line ). The 43 candidate AGN, selectedusing the IR diagnostic, that are not in the samples of MRBGD are shown as red filled triangles . Right:
Same as left, but showing the 182 BCDs( cyan filled triangles ) and the 2 low-metallicity and heavily obscured BCDs that were originally identified by Gri ffi th et al. (2011) ( magenta filledcircles ). The BCD MRK 709 S ( yellow filled circle ), one of the most metal-poor BCDs with evidence of an active galactic nucleus (Reines et al.2014), also has a WISE colour below our selection cut-o ff . within 3 Mpc of the Sun. We also searched the NASA Ex-tragalactic Database (NED) under the following three classifi-cations: dwarfs (dwarf, nucleated, dwarf elliptical (dE), dwarflenticular (S0), dwarf spiral (dS), dwarf irregular (dI) and bluecompact dwarf (BCD)); E peculiar; and compact E. Dwarfs areill defined in the literature (see e.g. Dunn 2010, for a reviewon the various definitions), and some known dwarfs, such asNGC 185 and 147, are labelled in NED as elliptical galaxies.Thus, we expanded our search parameters to ensure that we didnot lose any potential dwarfs in the nearby Universe. Dwarf spi-rals are a contentious subject in and of themselves, and we didnot search NED for spiral galaxies to add to the sample. In orderto define our dwarf sample in a consistent manner – regardlessof their previous classification in the literature – in the followingwork, we apply our own mass cut (see Sections 4 and 7).This primary sample was augmented by the surveys of galax-ies, including dwarfs at the high end of the mass range, that hadalready been identified as having the optical spectroscopic sig-nature of low-mass actively accreting black holes (Moran et al.2014; Reines et al. 2013, 2014; Barth et al. 2008; Greene etal. 2007, 2004; Dong et al. 2012, 2007, hereafter MRBGD).These AGN candidates will be used to verify the IR selectiontechnique. The final list, cleaned of all duplicates, consists of atotal of 5897 galaxies (see Table 1 for a summary of the varioussamples and their sizes discussed throughout this paper).
3. Mid-infrared colours of nearby galaxies
For this work, we used mid-infrared colours to identify theAGN in our sample. One advantage of using images in the mid-infrared is that both unobscured (type 1) and obscured (type 2)AGN are detected. The mid-infrared selection of unobscuredAGN relies upon distinguishing the approximately power-lawAGN spectrum from the black-body stellar spectrum of galax-ies (which peaks at rest-frame 1.6 µ m; Assef et al. 2010, Fig-ure 3) using its red mid-infrared colours. It is important to note that most of the previous surveys of AGN in dwarf galaxies havebeen carried out in the optical. Optical studies are biased towardsobserving primarily type 1 AGN, even though the unified AGNmodel (Antonucci 1993; Urry & Padovani 1995) predicts thattype 2 AGN should outnumber type 1 by a factor of ∼ µ m for 563,921,584 point-like and resolved objects de-tected on the Atlas Intensity images. The photometry in this cata-logue was performed using point source profile-fitting and multi-aperture photometry and the estimated sensitivities are 0.068,0.098, 0.86 and 5.4 mJy (5 σ ) at 3.4, 4.6, 12 and 22 µ m in uncon-fused regions on the ecliptic plane. J2000 positions and uncer-tainties were reconstructed using the 2MASS Point Source Cat-alog as astrometric reference. Astrometric accuracy is approx-imately 0.2 arcsec root-mean-square on each axis with respectto the 2MASS reference frame for sources with signal-to-noiseratio (S / N) greater than forty.We cross-matched our nearby galaxy catalogue with theWISE All-Sky Release Source Catalog using the US Virtual As-tronomical Observatory (VAO) cross-comparison tool. Using amatching radius of 6 arcseconds, corresponding to the resolu-tion of WISE in W1, we obtained 5042 matches. For each of thegalaxies, WISE photometry for W1, W2, W3 and W4 was gath-ered and a S / N cut of 3 was imposed on the first three bands sinceonly these three bands are needed for the AGN diagnostic dia-gram and to obtain an estimate of the BH mass (see Section 9).We used the point source profile-fitting photometry (w1mpro,w2mpro, w3mpro and w4mpro) for sources with goodness-of-fit ≤ Article number, page 3 of 14 & A proofs: manuscript no. marleau_AGNdwarfs_corr_printer
Table 1.
Sample descriptions and sizes.
Name Content SizePrimary sample Catalogue of Karachentsev et al. (2013); 5143catalogue of McConnachie et al. (2012);NED dwarfs, E peculiar, compact EMRBGD samples Catalogues from MRBGD 754Total sample Primary and MRBGD 5897WISE sample WISE matches to Primary sample 5042WISE S / N sample WISE matches to Primary sample with S / N > + BL AGN with S / N > a WISE MRBGD BL AGN sample WISE matches to MRBGD BL AGN with S / N > a WISE MRBGD NL AGN sample WISE matches to MRBGD NL AGN with S / N > a WISE BCD AGN sample WISE matches to BCD AGN with S / N > ffi th AGN sample WISE matches to Gri ffi th AGN with S / N > / N sample with W − W > . + BL AGN sample 43IR selected BCD AGN sample WISE BCD AGN sample with W − W > . + BL AGN sample with W − W > . a IR selected MRBGD BL AGN sample WISE MRBGD BL AGN sample with W − W > . a IR selected MRBGD NL AGN sample WISE MRBGD NL AGN sample with W − W > . a IR selected Gri ffi th AGN sample WISE Gri ffi th AGN sample with W − W > . a IR selected AGN sample with redshift / distance IR selected AGN sample with redshift / distance 300IR selected new AGN sample with redshift / distance IR selected new AGN sample with redshift / distance 40WISE S / N sample in MPH / JHU catalogue WISE S / N sample with MPH / JHU catalogue matches 968WISE S / N sample in BPT diagram WISE S / N sample with MPH / JHU BPT emission line fluxes 954WISE S / N sample with stellar masses WISE S / N sample with MPH / JHU stellar masses 934IR selected AGN sample with stellar masses IR selected AGN sample with MPH / JHU stellar masses 264IR selected dwarf AGN sample IR selected AGN sample with log M stellar < . / distance IR selected AGN sample with MPH / JHU stellar masses and redshift / distance 264IR selected MRGBD AGN sample with stellar masses and redshift / distance IR selected BL + NL AGN sample with MPH / JHU stellar masses and redshift / distance 239 a IR selected MRGBD BL AGN sample with stellar masses and redshift / distance IR selected BL AGN sample with MPH / JHU stellar masses and redshift / distance 227 a IR selected MRGBD NL AGN sample with stellar masses and redshift / distance IR selected NL AGN sample with MPH / JHU stellar masses and redshift / distance 14 a IR selected Gri ffi th AGN sample with stellar masses and redshift / distance IR selected Gri ffi th AGN sample with MPH / JHU stellar masses and redshift / distance 2IR selected new AGN sample with stellar masses and redshift / distance IR selected new AGN sample with MPH / JHU stellar masses and redshift / distance 23IR selected NED BCD AGN sample with stellar masses and redshift / distance IR selected NED BCD AGN sample with MPH / JHU stellar masses and redshift / distance 15IR selected AGN sample in BPT diagram IR selected AGN sample with MPH / JHU BPT emission line fluxes 135IR selected MRBGD AGN sample in BPT diagram IR selected MRBGD NL + BL AGN sample with MPH / JHU BPT emission line fluxes 126 a IR selected MRBGD NL AGN sample in BPT diagram IR selected MRBGD BL AGN sample with MPH / JHU BPT emission line fluxes 111 a IR selected MRBGD BL AGN sample in BPT diagram IR selected MRBGD NL AGN sample with MPH / JHU BPT emission line fluxes 16 a IR selected new AGN sample in BPT diagram IR selected new AGN sample with MPH / JHU BPT emission line fluxes 9IR selected NED BCD AGN sample in BPT diagram IR selected NED BCD AGN sample with MPH / JHU BPT emission line fluxes 83IR selected AGN sample with IR and [OIII] λ λ Notes. a Two objects in this sample are classified as both NL and BL. point source and the source is not associated with or superim-posed on a 2MASS Extended Source Catalog (XSC) source. Forsources with goodness-of-fit > W − W W − W / N >
4. Infrared colour diagnostic for AGN candidates
Several mid-infrared colour diagnostics have been used in theliterature to select AGN, starting with the pioneering work of(Lacy et al. 2004, 2007) and the so-called “Lacy wedge”. Forlow redshift galaxies (i.e. z < . µ m) andW2 (4.6 µ m) bands alone. According to these authors, a colourcut of W − W > .
8, 0.7, 0.6, and 0.5 is able to identify AGNwith a reliability of 95%, 85%, 70% and 50%, respectively. Notethat although reliability drops with lower colour cut, complete- ness increases and reaches the 95% level for W − W > .
5. Ascan be seen in their Figure 2, even the least stringent colour cutof W − W > . ≥
50% to z = . W − W W − W ff .The 3326 galaxies with WISE matches and S / N > W W W W − W > .
5, we identify 303 candidate AGN (see Figure 2).These final numbers come after we conservatively removed can-didates that were found close to an image artefact that could pos-sibly contaminate the source and give an erroneous photometry.Of the 303 candidate AGN, 62 are classified as dwarf galaxiesbased on a log M stellar cut of 9.5 (similar to Reines et al. 2013). Article number, page 4 of 14arleau et al.: CMBH in Nearby Dwarf Galaxies
Fig. 2.
Mosaic of the 43 candidate AGN, selected using the IR diagnostic, which are not in the samples of MRBGD. They are ordered, from left to right and top to bottom , by increasing distance with the exception of the first three candidates, which have no known redshift / distance (seeTable 2). The top images are either optical g’,r’,i’ colour images from SDSS, which are 76.8” on a side (except for OBJ 36, which is 153” on aside), or false colour images from the DSS-2-red catalogue, which are 1’ on a side (except for OBJ 6 and 8, which are 4’ on a side). The bottom images are W1,W2,W3 colour images from WISE, ∼
9’ on a side (except for OBJ 14, 23 and 34, which are ∼
4’ on a side).
The list of these galaxies, ordered by increasing distance, can befound in Tables 2 (dwarfs) and 3 (non-dwarfs).In our analysis, we use the colour cut of W − W > . W − W ff ect of metallicity on the intensity of the polycyclic aromatic hydrocarbon (PAH)emission has been quantified in detail using Spitzer data (e.g.Calzetti 2011, and references therein). Analyses of galaxies witha range of metallicities show that a factor of ∼
10 decrease inmetallicity is accompanied by an order of magnitude decreasein the 8 µ m to total infrared luminosity (i.e. redder W − W Article number, page 5 of 14 & A proofs: manuscript no. marleau_AGNdwarfs_corr_printer
Fig. 3.
Left: [OIII] / H β vs. [NII] / H α diagnostic diagram for the 954 out of the 3326 galaxies with both line flux ratio measurements from theMPA / JHU catalogue. As in Figure 1, the 135 sources above our IR colour cut are shown as red filled triangles . The dashed line is the demarcationline between normal star-forming galaxies and AGN from Stasinska et al. (2006).
Middle:
Same as left, but showing the 126 AGN that werepreviously identified in the MRBGD optical samples split into 111 type 1 (BL; green filled triangles ) and 16 type 2 (NL; blue filled triangles ).One-third of this MRBGD sample are located in the region for HII galaxies. The 9 candidate AGN, selected using the IR diagnostic, which arenot in the samples of MRBGD are shown as red filled triangles . Right:
Same as left, but showing the 83 BCDs ( cyan filled triangles ) and the twolow-metallicity and heavily obscured BCDs that were originally identified by Gri ffi th et al. (2011) ( magenta filled circles ). The BCD MRK 709 S( yellow filled circle ), one of the most metal-poor BCDs with X-ray and radio emission indicative of AGN activity, is also located in the region forHII galaxies. Table 2.
List of dwarf galaxies with IR signatures of active massive black holes.Name RA Dec Redshift Distance log(M stellar ) W1-W2 W2-W3 log(M
BH IR ) ImageJ2000 J2000 Mpc M (cid:12) [mag] [mag] M (cid:12)
AM 1906-621 19h11m33.43s -62d10m22.4s - - - 0.784 2.950 - 1AM 1238-405 12h41m15.40s -41d09m34.0s - - - 0.628 3.626 - 2ESO 184- G 050 19h16m17.23s -54d20m40.8s - - - 0.586 3.252 - 3HIPASS J1247-77 12h47m32.60s -77d35m01.0s 0.001378 3.160 - 2.086 4.026 2.192 4UGC 04459 08h34m07.20s + a + + + + + + Notes.
Col. (1): Object Name. Cols. (2) and (3); Right ascension and declination (J2000). Col. (4): Redshift. Col. (5): Distance in Mpc. Col. (6):Log of host galaxy stellar mass measured from SED fit. Cols. (7) and (8): WISE colours. Col. (9): Log of central black hole mass estimates basedon IR luminosity (assuming L bol = . L Edd ). Col. (10): Image number in Figure 2. a These sources are extended and larger than the largest WISE aperture (24.75 arcsec radius aperture); therefore, the infrared black hole massestimates for these sources are only lower limits.Table 2 is published in its entirety in the electronic edition of the A&A. A portion is shown here for guidance regarding its form and content. colours), with a transition at 12 + log(O / H) ≈ M stellar ∼ .
0. Hence many dwarfs, based on our masscut-o ff (log M stellar = . W − W × M (cid:12) central black hole is embedded in anuclear star cluster of mass 2 × M (cid:12) (den Brok et al. 2015),and is detected via its infrared signature (Satyapal et al. 2014).Hence, strong levels of star formation do not negate the possiblepresence and IR detection of an AGN.If we were to apply a cut on the W − W W − W < . W − W Article number, page 6 of 14arleau et al.: CMBH in Nearby Dwarf Galaxies
Table 3.
List of non-dwarf galaxies with IR signatures of active massive black holes.Name RA Dec Redshift Distance log(M stellar ) W1-W2 W2-W3 log(M
BH IR ) ImageJ2000 J2000 Mpc M (cid:12) [mag] [mag] M (cid:12)
CIRCINUS 14h13m09.30s -65d20m21.0s 0.001448 4.207 - 1.038 2.325 6.306 a + a + a + a + a + + a + Dong164 11h53m41.77s + + + + + a - Notes.
Col. (1): Name. Cols. (2) and (3); Right ascension and declination (J2000). Col. (4): Redshift. Col. (5): Distance in Mpc. Col. (6): Log ofstellar mass measured from SED fit. Cols. (7) and (8): WISE colours. Col. (9): Log of central black hole mass estimates based on IR luminosity(assuming L bol = . L Edd ). Col. (10): Image number in Figure 2. a These sources are extended and larger than the largest WISE aperture (24.75 arcsec radius aperture); therefore, the infrared black hole massestimates for these sources are only lower limits.Table 3 is published in its entirety in the electronic edition of the A&A. A portion is shown here for guidance regarding its form and content.
Fig. 4.
Distance distribution of our IR selected sample of 300 AGNcandidates with known redshift / distance. The subsample of 40 galaxiesselected via the IR diagnostic, shown here in red, contains the nearestAGN candidates known today. ffi th et al. (2011), but 8 are eithertype 1 (2) or type 2 (6) optically identified AGN. We note thatthe remaining 6 that were not previously identified as AGN havecolours similar to these 8 optically identified AGN and hence arevalid candidate AGN. We also note that AGN detected in dwarfgalaxies via other methods, such as the dwarf galaxy Henize 2-10 (Reines et al. 2011), have W − W > . W − W ∼ .
5. Verification using existing catalogues and opticalline emission diagnostic
To verify the validity of our method, we compare our final listof AGN candidates with AGN candidates determined by othermethods. The results of this comparison are listed below.1. We find that 258 of the previously optically identified AGNin the MRBGD samples are also above our IR colour cut (seeFigure 1). Of the 43 that are not in the MRBGD samples, aliterature search revealed that 16 had previously been identi-fied as hosting an AGN (see Table 4), further supporting ourIR colour selection criterion for a total of 276 out of 303 (or91%) AGN candidates.2. By splitting the MRBGD samples into type 1 (broad-line,hereafter BL) and type 2 (narrow-line, hereafter NL), we cansee that type 1 AGN appear on average to have redder W − W ff , ascompared to the type 2 AGN which lie preferentially below.This simply demonstrates what was discussed in Section 3above, that the IR colour diagnostic does not exhaustivelypick out AGN as their signatures can be washed out by anappreciable level of star-forming activity.3. In Figure 1, we highlight three famous Seyfert 1 galaxies,NGC 4395, POX 52, and UM 625, to show where they fall onthe diagram ( yellow circles ). As discussed above, NGC 4395has the only dynamical mass measurement of a CMBH in adwarf galaxy (den Brok et al. 2015), POX 52 has a robustCMBH mass estimate (Barth et al. 2004) and UM 625 isa Seyfert 1 galaxy hosting a low-mass CMBH (Jiang et al.2013). It should be noted that UM 625 was added manu-ally, and is not part of our sample as it is neither a dwarf nornearby, though its low-mass CMBH still makes it an objectof interest. We see that the AGN candidates for these threegalaxies have colours above our selection cut-o ff .4. It can be noted that there are four outliers with very redcolours, i.e. W − W > .
7, in the top right-hand cor-ner of the diagram. Two of these ( magenta circles in Fig-ure 1) are the low-metallicity and heavily obscured BCDs
Article number, page 7 of 14 & A proofs: manuscript no. marleau_AGNdwarfs_corr_printer
Table 4.
List of 16 known AGN with IR signatures of active massive black holes.Name RA Dec Classification log(M
BH IR ) log(M BH ) ReferenceJ2000 J2000 M (cid:12) M (cid:12) NGC 5253 13h39m55.80s -31d38m24.0s Sy2 5.602 a - Koulouridis (2014)CIRCINUS 14h13m09.30s -65d20m21.0s Sy2 6.306 a + a a a + a - SIMBAD info pageNGC 3690 NED01 11h28m31.02s + a + a + a + + +
437 14h17m01.41s + a - Veron-Cetty & Vernon (2010)FCSS J033846.0-352252 03h38m45.97s -35d22m52.3s Sy2 7.281 - Veron-Cetty & Vernon (2010) Notes.
In Col. (5), M BH IR denotes the central black hole mass estimates based on IR luminosity (assuming L bol = . L Edd ), while in Col. (6), M BH refers to the central black hole mass estimates based on other conventional (non-infrared) methods. a These sources are extended and larger than the largest WISE aperture (24.75 arcsec radius aperture); therefore, the infrared black hole massestimates for these sources are only lower limits. that were originally identified by Gri ffi th et al. (2011) asalready discussed above. These have been flagged in Ta-ble 2 as they are outliers and have optical diagnostics consis-tent with HII regions. The other two are HIPASS J1247-77( W − W ∼ . W − W ∼ . ffi thet al. 2011) ( magenta circles found in the top left corner ofFigure 1, right) and appear to be dominated by star forma-tion. Others, like MRK 709 S (red circle in Figure 1, right),have X-ray and radio emission indicative of AGN activity(Reines et al. 2014). The WISE colours of MRK 709 S are W − W = .
323 and W − W = . We explored the distribution of our sample of 3326 galaxieswith WISE matches and S / N > W W W ff mann et al. 2003). Diagnostic line-intensityratios, such as [OIII] / H β and [NII] / H α , corrected for reddening,are e ff ective at separating populations with di ff erent ionisationsources. The ionising radiation field found in active galaxies isharder than in star-forming galaxies, and this gives higher val-ues of [NII] / H α and [OIII] / H β . Hence, the narrow line AGN are found in the upper right portion of the [OIII] / H β versus [NII] / H α diagnostic diagram. The lines in these ratios are also selected tobe close in wavelength space in order to minimize the e ff ect ofdust extinction on the computed line ratios.We obtained emission line fluxes and stellar mass esti-mates for our sample from the Max Planck Institute for Astro-physics / Johns Hopkins University (MPH / JHU) collaboration ,which contains 927552 SDSS galaxies. Of the 968 matches, wefound that 954 (9 of the 43 AGN candidates without prior identi-fication) had all four [OIII], H β , [NII], and H α line fluxes and wewere therefore able to compute the line-intensity ratios shownin Figure 3, left. As in Figure 1, left, our 135 nearby galaxiesare shown as red triangles and the sample of 126 galaxies, in-cluding dwarfs, that had already been identified as having theoptical spectroscopic signature of AGN activity (MRBGD) areshown as black open triangles . These additional optically identi-fied AGN are split into 111 type 1 (BL; green filled triangles ) and16 type 2 (NL; blue filled triangles ). The dashed line is the de-marcation line between normal star-forming galaxies and AGNfrom Stasinska et al. (2006).The main results of this comparison is summarized below.1. As expected, the majority of the optically classified AGNare located in the conventional region of Seyfert galaxies, interms of the semi-empirical demarcation line of Stasinska etal. (2006) (the dashed line in Figure 3, left) in the [OIII] / H β versus [NII] / H α diagram. This also includes four objects thathad not previously been identified as AGN, if we conserva-tively include the two objects close to the demarcation line(red triangles in Figure 3, left).2. The remaining one-third of the objects (five that had not pre-viously been identified as AGN) are located in the region forHII galaxies in the same diagram. However, this does notnecessarily mean that they are not AGN. Indeed, Sartori etal. (2015) compared AGN selected via three methods (using http: // / SDSS / Article number, page 8 of 14arleau et al.: CMBH in Nearby Dwarf Galaxies
Fig. 5.
Left : SED-based stellar mass distribution of our sample of 264 AGN candidates. The subsample of 23 galaxies that are not in the samplesof MRBGD, shown here in red, contains the lowest stellar mass galaxies hosting an AGN candidate known today. The SED-based stellar massdistribution of our sample of 264 AGN candidates is also shown scaled down by a factor of 5 ( blue dotted line ). Right : Black hole mass distributionof our sample of 300 AGN candidates with estimated black hole mass. The black hole masses were estimated from their IR luminosity, assuming L bol = . L Edd , the mean of our calibration sample. The subsample of 40 galaxies selected via the IR diagnostic, shown here in red, contains thelowest mass BH candidates known today. As in the left plot, the black hole mass distribution of our sample of 300 AGN candidates is also shownscaled down by a factor of 5 ( blue dotted line ). the classical BPT diagram, a similar optical emission line di-agnostic based on the He II 4686Å line, and mid-IR colourcuts) and found that only 3 of their 336 sources fulfilled allthree criteria and that di ff erent criteria selected host galaxieswith di ff erent physical properties such as stellar mass andoptical colour. It is therefore likely that our IR colour cut isselecting a di ff erent subsample of AGN to that selected bythe optical diagnostic diagram. Also, as is reported in Donget al. (2012), this di ff erence can be explained by the pos-sible inclusion in the SDSS fibre aperture of emission fromstar formation regions in the host galaxies. Hence, althoughthe optical spectra undeniably reveal the presence of a broad-line AGN, the narrow-line ratios may still be dominated bythe characteristic emission of an HII region. We also notethat MRK 709 S (red circle in Figure 3, right), which has X-ray and radio emission indicative of AGN activity (Reines etal. 2014), is located very close to the demarcation line.3. Our comparison reveals some possible issues regarding us-ing the usual demarcation line between normal star-forminggalaxies and AGN. Indeed, it is possible that the demarcationmay not apply to the low-mass galaxy regime and for nearbygalaxies. Given that the BPT diagram is empirically derived,and to date there has been little data on AGN in dwarfs, thisis perhaps not surprising. In relation to this, it has also beenknown for some time that star formation behaves di ff erentlyin some dwarf galaxies where there is a steepening of theKennicutt–Schmidt law (Bigiel et al. 2008; Elmegreen etal. 2011; Roychowdhury et al. 2015). Additionally, recentwork by Kewley et al. (2015) has shown that at high red-shift, the demarcation line changes as a function of redshift.It is therefore perhaps possible that such a dependence couldalso extend to very low redshifts and / or low masses.
6. Distance distribution
One of the main goals of this study is to identify the closest dwarfgalaxy hosting an AGN. Hundreds of active massive black holecandidates have already been detected at the centres of low-massgalaxies (e.g. Marleau et al. 2013; Moran et al. 2014; Reines etal. 2013). The major limitation of these works, however, is thatthey do not provide an unambiguous confirmation of the exis-tence of these IMBHs as they rely solely on detecting the radia-tive signatures of AGN. Moreover, the dwarf galaxies in theseexisting samples are too far away to carry out dynamical stud-ies (e.g. the Reines et al. 2013, sample has a redshift range of z = . − . / distance. The first threecandidates listed in Table 2 have no known redshift / distance.The sample of 40 galaxies selected via the IR diagnostic, shownin red, contains the nearest AGN candidates known today . Be-low 11 Mpc, we find 11 AGN candidates, including NGC 4395(4.5 Mpc) which has been until now the closest one identifiedin a dwarf galaxy. Our method detected AGN in five galaxiesthat are closer than NGC 4395: HIPASS J1247-77 (3.2 Mpc),UGC 04459 (3.2 Mpc), NGC 5253 (3.6 Mpc), IC 2574 (3.8 Mpc)and CIRCINUS (4.2 Mpc). Of these, four can be classified asdwarf galaxies based on their stellar mass and / or absolute mag-nitude: HIPASS J1247-77 (Ryan-Weber et al. 2002) has a mag-nitude M B = -12.91 (Karachentsev et al. 2004), UGC 04459 hasa stellar mass of ∼ . × M (cid:12) and a magnitude M B = -13.43 Article number, page 9 of 14 & A proofs: manuscript no. marleau_AGNdwarfs_corr_printer
Fig. 6.
Left : WISE ( black curve ) / AGN ( red curve ) fraction as a function of stellar mass derived from the MPA / JHU catalogue.
Right : Same as leftbut from the catalogue of Mendel et al. (2014). (Zhang et al. 2012), NGC 5253 (Turner et al. 2015) has a stellarmass of ∼ . × M (cid:12) (Martin 1998) and IC 2574 has stellarmass of ∼ . × M (cid:12) (Lee et al. 2011) and a magnitude M B = -16.8 (Walter & Brinks 1999). The fifth galaxy, CIRCI-NUS, has a stellar mass of ∼ . × M (cid:12) and is therefore notconsidered a dwarf galaxy (For et al. 2012). In the context ofbeing able to follow-up and confirm the presence of a IMBH ina dwarf galaxy with dynamical measurement, this implies thatnearby dwarf galaxies should be targeted for dynamical obser-vations.
7. Stellar mass estimates
We are interested in determining whether low-mass galaxies har-bour a CMBH, and if so, whether or not the correlation betweenBH mass and total stellar mass derived in Marleau et al. (2013)extends to the low-mass regime. Therefore, in our calculationof stellar and BH masses, we did not only consider galaxies inour sample of dwarfs but also included the other galaxies in ournearby galaxies sample.We used a combination of three independent methods to es-timate the stellar masses for the 300 out of 303 galaxies withredshift measurement. The first method consisted of fitting theSEDs of our galaxies, constructed from SDSS photometry, us-ing the MAGPHYS package (da Cunha et al. 2008). For thesecond method, we estimated the stellar masses following theempirical relation of Taylor et al. (2011, see their equation 8).This method combines a galaxy’s luminosity ( M i expressed inthe AB system) with a mass-to-light ratio derived from a colourmeasurement ( g − i ). We transformed the SDSS magnitudes intoAB mag, applied K -corrections using the method of Chilingar-ian et al. (2010), which, for the low-redshift ( z < .
01) galaxiesin our sample, typically a ff ect the g and i values by a few hun-dredths of a magnitude. The absolute magnitude M i was com-puted using the distances given in Table 2. This method onlyapplies for galaxies with z < . K s -band magnitude taken fromboth the PSC and the XSC of 2MASS. A general calibration fac-tor was applied to the
K s fluxes by comparing the stellar massesto the SED derived stellar masses.We compared the stellar masses measured from our threemethods with those listed in the MPA / JHU and Mendel et al.(2014) catalogues and those found in the literature. For the 261galaxies with stellar masses from both the SED fit and the colormeasurement method, we found in general very good agreementbetween the two estimates, with only two exceptions wherein thetwo mass estimates di ff er by an amount larger than the scatter.We find excellent agreement between our SED fit stellar massesand those of the MPA / JHU and Mendel et al. (2014) catalogueswhich confirms that our SED fit stellar mass estimates are ro-bust. We also find in general good agreement between the SEDfit stellar masses and the few stellar masses which were reportedby MRBGD.As can be seen in Figure 5, we sample evenly a broad distri-bution of stellar masses, including dwarf galaxies in the stellarmass range ∼ − M (cid:12) .
8. WISE/AGN fraction
As discussed in the Introduction, dwarf galaxies exhibit manyproperties that distinguish them from massive galaxies (e.g.mass-to-light ratios, star formation histories, metallicities) andquestions exist about the nature of their evolution with respectto massive galaxies. In this regard, it is interesting to considerwhether AGN have evolved di ff erently in dwarf galaxies than inmassive galaxies. A measure of this is given by the AGN fractionat a given time and mass.Starting from the MPA / JHU catalogue of galaxies with stel-lar masses and from the catalogue of Mendel et al. (2014), weexamine the fraction of WISE matches and IR selected AGN as afunction of stellar mass in each of these samples. Each cataloguewas cross-matched independently to the WISE All-Sky ReleaseSource Catalog using the same method as described in Section 3
Article number, page 10 of 14arleau et al.: CMBH in Nearby Dwarf Galaxies
Fig. 7.
Left : Total stellar mass from SED fit vs. black hole mass obtained using the bolometric luminosity for our infrared sample of 264 galaxieswith both known redshifts / distances and stellar masses ( red filled triangles ). The data points and bisector linear regression fit of Marleau et al.(2013) ( dashed line ) are plotted for L bol / L Edd = .
1. Also shown are the fit for L bol / L Edd = . dotted line ) and for L bol / L Edd = .
01 ( solid line ). Middle : Same as left, but showing previously identified AGN of type 1 and type-2, respectively shown as green filled triangles and blue filledtriangles , and the AGN candidates identified from the IR diagnostic only, shown as red filled triangles . Also shown are the two low-metallicityand heavily obscured BCDs that were originally identified by Gri ffi th et al. (2011) ( magenta filled circles ). Right : Same as left, but showing theBCDs ( cyan filled triangles ), the two low-metallicity and heavily obscured BCDs that were originally identified by Gri ffi th et al. (2011) ( magentafilled circles ), and the BCD MRK 709 S with X-ray and radio emission indicative of AGN activity ( yellow filled circle ). and the same colour cut as described in Section 4. The errorswere computed assuming Poisson statistics. As can be seen inFigure 6, the fraction of IR selected AGN shows a signaturebump at a stellar mass ∼ . × M (cid:12) . Also, the fraction ofAGN appears to increase as a function of decreasing stellar massat stellar masses below ∼ M (cid:12) , i.e. in the low-mass regime ofdwarf galaxies, an e ff ect also reported in Satyapal et al. (2014,see their Figure 5). Note, however, as can also be seen in the fig-ure, that this increase is accompanied by an increase in the WISEfraction. This is not the case for the bump seen at ∼ . × M (cid:12) , which is displaced from the peak in the WISE fraction. Ifreal, this behaviour at low mass could be due to 1) the fact thatBH accretion activity is higher in nearby dwarf galaxies than intheir more massive counterparts (similar to the star-forming ac-tivity, i.e. the “downsizing” e ff ect); 2) the fact that it is easierto detect WISE sources / AGN in nearby low-mass (low surfacebrightness) galaxies; or 3) the fact that at these low masses, thefraction of AGN candidates contaminated by star formation maybe higher. This is a result that should be further explored andconfirmed.
9. Black hole mass estimates
Black hole masses were estimated using the bolometric lumi-nosity of our AGN candidates. Bolometric luminosities, takento be the 100 µ m to 10 keV integrated luminosity (Richardset al. 2006), are typically obtained using corrections to themid-infrared bands where the AGN emission dominates. The 12(W3) and 22 µ m (W4) k-corrected flux densities from WISE canbe used to compute the bolometric luminosities of the WISEcolour-selected AGN by applying the bolometric corrections L bol (cid:39) × L µ m (W3) and L bol (cid:39) × L µ m (W4) from Richardset al. (2006), which are not strongly dependent on AGN lumi-nosity (see their Figure 12). However, note that Richards et al.(2006) only considered type 1 (BL) AGN in their analysis so thatapplying these corrections to type 2 AGN may lead to additionaluncertainties in the BH mass estimates. Here, we only computethe BH mass estimates based on the 12 µ m luminosities, as the Fig. 8.
Total stellar mass from all methods used for stellar mass mea-surements vs. black hole mass obtained using the bolometric luminosityfor our infrared sample of 300 galaxies: from SED fit ( red filled trian-gles ), from colour measurement ( cyan filled triangles ), from Ks -bandfrom 2MASS / PSC and XSC ( magenta filled triangles ), from the liter-ature ( blue filled triangles ), from Mendel et al. (2014) ( yellow filledtriangles ) and MPA / JHU catalogues ( green filled triangles ). S / N of the WISE data in the 12 µ m images is higher than in the22 µ m images. By comparison, the bolometric luminosity of anAGN can also be estimated using the [O III] λ L bol (cid:39) × L )(Heckman et al. 2004) to a factor of ten ( L bol (cid:39) × L )(Greene et al. 2007) larger than the IR bolometric correction. Article number, page 11 of 14 & A proofs: manuscript no. marleau_AGNdwarfs_corr_printer
For point source AGN candidates, we used the w3mpro mag-nitudes to compute BH masses. For our 30 extended source AGNcandidates, we used the aperture magnitude corresponding to themeasured extent of the central source in W3 (e.g. for a galaxywith a radial extent of 19 arcsec in W3, we used w3mag_6). As11 of our AGN candidates are spatially larger than the largestWISE aperture (24.75 arcsec radius aperture), the BH mass esti-mates for these sources is only a lower limit. Given the bolomet-ric luminosity derived from the IR, making the assumption thataccretion is at the Eddington limit yields a lower limit on theblack hole mass. However, given that the growth rates of AGNcan span several orders of magnitude, from super-Eddington ac-cretion to 10 − L Edd (Simmons et al. 2013; Steinhardt & Elvis2010), we computed black hole masses assuming the followingthree cases: L bol = L Edd , L bol = . L Edd and L bol = . L Edd .In a separate work (López, K.M. 2015), the IR CMBHmass estimates were calibrated using 113 galaxies with robustCMBH mass measurements. Of the 113 galaxies, 51 have esti-mates based on the reverberation mapping method and 62 basedon dynamical measurements. After following the procedure de-scribed in Sections 3 and 4, 50 of these galaxies were identifiedas AGN candidates based on our IR colour diagnostic and theirbolometric luminosity was computed as described in the para-graphs above. By comparing the CMBH mass estimates based onthe other two methods to the estimate based on the IR method,the distribution of L bol / L Edd was computed. A Gaussian fit tothis distribution, in log space, yielded a mean of -1.0 and a σ of0.6, in agreement with the range of Eddington ratios found in theliterature (Woo & Urry 2002; Mushotzky et al. 2008).The histogram of BH masses computed using L bol = . L Edd , the mean value derived from our calibration sample (-1.0 in log space), is shown in Figure 5, right. This choice of L bol / L Edd was also used in Marleau et al. (2013) and also ap-pears to be a good fit to this data set (see Section 10). Theblack hole masses associated with the dwarf galaxies rangefrom ∼ − M (cid:12) (see Section 10), which is a significantlylower range than has previously been probed. However, as statedabove, note that the Eddington ratio is known to vary by as muchas three orders of magnitudes for di ff erent AGN so our results arealso presented for values of L bol / L Edd equal to 0.01 and 1.0.For NGC 4395, since an Eddington ratio has been measuredwith a value of 0.0012 (Peterson 2005), we can compute its IRCMBH mass directly. We calculate a mass of 3 . × M (cid:12) , invery good agreement with the dynamical mass of 4 × M (cid:12) given by den Brok et al. (2015). POX 52 also has a reportedEddington ratio of 0 . − . . × M (cid:12) , in agreementwith the range of CMBH masses of 2 . − . × M (cid:12) obtainedby Thornton et al. (2008) using other CMBH mass estimatormethods.
10. BH mass versus stellar mass scaling relation
Using the stellar masses of the galaxies and the black holemasses estimated from the bolometric luminosity of the AGNcandidate, we present in Figure 7, left, the correlation betweenblack hole mass and total SED stellar mass for our sample of264 galaxies. We find active BH in nearby dwarfs, as well as indwarf galaxies with lower stellar masses ( ∼ − M (cid:12) ) andcorrespondingly lower BH masses ( ∼ − M (cid:12) ) than the pre-vious works of MRBGD, which only probed to 10 − stellarmasses. We also find that the current results are consistent withthe existing correlation (Marleau et al. 2013) extending linearly Fig. 9.
Comparison between the black hole mass estimates computedfrom their IR luminosity and the black hole mass computed from the[OIII] λ black opentriangles ). Previously optically identified type 1 and type 2 AGN areshown as green filled triangles and blue filled triangles , respectively. (in log-log space) into the lower mass regime . However, we can-not rule out with the current data that the correlation could beweakly non-linear in the low-mass regime.In Figure 7, middle, we examine the stellar mass versus BHmass relationship of the BCDs. We see that in the low-massregime, the BCDs tend to be located mostly above the relation.This may indicate that a possible upturn at low mass is not realbut due to contamination from star formation activity. Indeed, wefind that the actively star-forming BCDs of Gri ffi th et al. (2011)( magenta circles ) fall above the relation. Also, the AGN hostingBCD MRK 709 S (Reines et al. 2014) ( red circle ) is expectedto be a ff ected by the host galaxy light based on its IR colour( W − W = . σ error). In Figure 8, weexplore the e ff ect of the uncertainty introduced by using di ff erentmethods on the scatter of the stellar mass versus BH mass rela-tion. The di ff erent colours are associated with the di ff erent meth-ods and catalogues used in estimating the stellar masses. Exceptfor the colour measurement method, we find that the agreementbetween the methods is quite good (see also Section 7) and doesnot significantly change the scatter seen in Figure 7, left.The errors on the BH mass were calculated from the un-certainties in the bolometric luminosity measurements and werefound to be ∼ L bol / L Edd , which we haveassumed to be the same for all sources but instead most likelyvaries for each source. As can be seen in Figure 7, simply allow-ing this value to vary from 0.01 to 1.0 reproduces most of thescatter seen in the data.We compare in Figure 9 the black hole mass estimates com-puted from the IR luminosity and those computed from the
Article number, page 12 of 14arleau et al.: CMBH in Nearby Dwarf Galaxies [OIII] λ ff of Greene et al. (2007) was 2 × M (cid:12) (6.3 in log space, see their Figure 1) and this cut-o ff is easilyseen in the diagram. Although some BH mass estimates are inagreement, the majority are not. The BH mass estimates com-puted from the [OIII] λ
11. Discussion and conclusions
We undertook a census of central massive black holes in thevery nearby Universe, targeting specifically low-mass dwarf sys-tems with BH masses in the IMBH mass range. Additionally, wewere interested in detecting AGN in nearby galaxies regardlessof their mass so that we could identify targets for future dynam-ical mass measurement. The results of our paper are as follows:1. Using the WISE All-Sky Release Source Catalog, we exam-ined the IR colours of a sample of known low-mass and othernearby systems in order to identify candidate AGN by apply-ing the infrared colour diagnostic W − W > .
5. We findthat 303 nearby galaxies have WISE colours consistent withgalaxies containing an AGN.2. We validate our IR detection method. Of the 303 candidateAGN, 276 (or 91%) are subsequently found to have been in-dependently identified as AGN via other methods. The re-maining 9% require follow-up observations to confirm thattheir red IR colours ( W − W > .
5) are not due to starformation. We find that if we applied an additional cut in W − W W − W < .
2, we would reject 16AGN candidates, 8 of which are optically identified AGN (2type 1 and 6 type 2) and only 2 are known low-metallicityand heavily obscured BCDs. The remaining 6 have colourssimilar to these 8 optically identified AGN and hence arevalid candidate AGN. We also point out that AGN detectedvia other methods in dwarf galaxies, such as the dwarf galaxyHenize 2-10, have W − W > . W − W ∼ . W − W ff , as compared to the type 2 AGN whichlie preferentially below. We show that NGC 4395, POX 52and UM 625 have WISE colours above our selection cut-o ff . We point out that although some low-metallicity and ex-tremely obscured BCDs have very red colours, others are notso red, such as MRK 709 S with W − W = .
323 and W − W = . ∼ − M (cid:12) , extending far beyondthe dwarf galaxy mass demarcation.6. We show that the fraction of IR selected AGN shows a sig-nature bump at a stellar mass ∼ . × M (cid:12) . Also, thefraction of AGN appears to increase as a function of decreas-ing stellar mass at stellar masses below ∼ M (cid:12) , i.e. in thelow-mass regime of dwarf galaxies. 7. Black hole masses are estimated using the bolometric lumi-nosity of the AGN candidates and computed for three casesof bolometric-to-Eddington luminosity ratio. Using the pre-viously measured Eddington ratio of 0.0012, we calculatefor NGC 4395 an IR CMBH mass of 3 . × M (cid:12) , in agree-ment with the recently measured dynamical mass estimateof 4 × M (cid:12) . Similarly, using a previously measured Ed-dinton ratio of 0 .
2, we compute for POX 52 an IR CMBHmass of 2 . × M (cid:12) , in agreement with the mass range of2 . − . × M (cid:12) obtained using other methods.8. Assuming all of our candidates are AGN, we find that ac-tivity is detected in dwarf galaxies with stellar masses fromapproximately 10 to 10 M (cid:12) and that this activity is due toblack holes with masses in the range ∼ − M (cid:12) , assum-ing L bol = . L Edd , the mean of our calibration sample. Theblack hole masses probed here are several orders of magni-tude smaller than previously reported for centrally locatedmassive black holes.9. We examine the stellar mass versus black hole mass relation-ship in this low galaxy mass regime. The current results areconsistent with the existing correlation extending linearly (inlog-log space) into the lower mass regime. However, we can-not rule out with the current data that the correlation couldbe weakly non-linear in the low-mass regime.These results suggest that central massive black holes arepresent in low-mass galaxies and in the Local Universe, and pro-vide new impetus for follow-up dynamical studies of quiescentblack holes in local dwarf galaxies.
Acknowledgements.
This work is a revised version of Marleau et al. (2014).This research has made use of the NASA / IPAC Extragalactic Database whichis operated by the Jet Propulsion Laboratory, California Institute of Technology,under contract with the National Aeronautics and Space Administration. Thisresearch has made use of the SIMBAD database, operated at CDS, Strasbourg,France.
References
Alonso-Herrero A., et al., 2014, arXiv:1301.4015v1Antonucci R., 1993, ARA&A, 31, 473Assef R.J., Kochanek C.S., Brodwin M., et al., 2010, ApJ, 713, 970Baldwin J.A., Phillips M.M., Terlevich R., 1981, PASP, 93, 5Ballantyne D.R., Draper A.R., Madsen K.K., Rigby J.R., Treister E., 2011, ApJ,736, 56Bardeen J.M., 1970, Nature, 226, 64Barth A.J., Greene J.E., Ho L.C., 2008, AJ, 136, 1179Barth A.J., Ho L.C., Rutledge R.E., Sargent W.L.W., 2004, ApJ, 607,90Beifiori A., et al., 2012, MNRAS, 419, 2497Bender R., Kormendy J., 2003, acfp.conf, 262Benson A.J., 2010, PhR, 495, 33Bigiel F., et al., 2008,AJ, 136, 2846Booth C.M., Schaye J., 2009, MNRAS, 398, 53Boselli A., Lequeux J., Gavazzi G., 2004, A&A, 428, 409Bukhmastova Y.L., 2001, Arep, 45, 581Calzetti D., 2011, PAHs and the Universe, C. Joblin and A.G.G.M. Tielens (eds),EAS Publications Series, 46, 133-141Chilingarian, I.V., Melchior, A.-L., Zolotukhin, I.Y., 2010, MNRAS, 405, 1409Comastri A., Setti G., Zamorani G., Hasinger G., 1995, A&A 296, 1da Cunha E., Charlot S., Elbaz D., 2008, MNRAS, 388, 1595Davis B., et al., 2014, ApJ, 789, 124Dunn J., 2010, MNRAS, 408, 392DeGraf C., Di Matteo T., Khandai N., Croft R., 2012, ApJ, 755, LL8den Brok M., Seth A.C., Barth A.J., et al., 2015, ApJ, 809, 101Di Matteo T., Springel V., Hernquist L., 2005, Nature, 433, 604Di Matteo T., Colberg J., Springel V., Hernquist L., Sijacki D., 2008, ApJ, 676,33Dong X.-B., et al., 2012, ApJ, 755, 167Dong X.-B., et al., 2007, ApJ, 657, 700Draine B.T., Dale D.A., Bendo G., et al., 2007, ApJ, 633, 866Elmegreen B.G., 2011, EAS Publications Series, 51, 3, arXiv:1101.3108
Article number, page 13 of 14 & A proofs: manuscript no. marleau_AGNdwarfs_corr_printer