VLA Limits for Intermediate Mass Black Holes in Three Globular Clusters
aa r X i v : . [ a s t r o - ph ] O c t D RAFT VERSION N OVEMBER
3, 2018Preprint typeset using L A TEX style emulateapj v. 11/12/01
VLA LIMITS FOR INTERMEDIATE MASS BLACK HOLES IN THREE GLOBULAR CLUSTERS
F.N. B
ASH , K. G EBHARDT , W.M. G OSS , P.A. V ANDEN B OUT Draft version November 3, 2018
ABSTRACTThe observational evidence for central black holes in globular clusters has been argued extensively, and theirexistence has important consequences for both the formation and evolution of the cluster. Most of the evidencecomes from dynamical arguments, but the interpretation is difficult, given the short relaxation times and old agesof the clusters. One of the most robust signatures for the existence of a black hole is radio and/or X-ray emission.We observed three globular clusters, NGC6093 (M80), NGC6266 (M62), and NGC7078 (M15), with the VLA inthe A and C configuration with a 3- σ noise of 36, 36 and 25 µ Jy, respectively. We find no statistically-significantevidence for radio emission from the central region for any of the three clusters. NGC6266 shows a 2- σ detection.It is difficult to infer a mass from these upper limits due to uncertainty about the central gas density, accretion rate,and accretion model. Subject headings: globular clusters INTRODUCTION
Although we do not understand how the nuclei of galaxiesform or why they have black holes (BH) at their centers, the cor-relation between BH mass and bulge velocity dispersion doesshed light on their formation and evolutionary histories (Geb-hardt et al. 2000a, 2000b: Ferrarese and Merritt 2000). A num-ber of different theories (e.g., Silk & Rees 1998; Haehnelt &Kauffmann 2000; Robertson et al. 2006) predict a BH massbulge-velocity-dispersion relation, although they predict differ-ent slopes and intercepts for this relation. Exploration of theextreme ends of this relationship will help illuminate the un-derlying physical model, and in this paper we focus on thelow mass end. Black holes at the low end of the relations,with masses between 100 and 10 M ⊙ , are generally referredto as intermediate-mass black holes (IMBHs). There is signif-icant evidence that black hole masses from 10 - M ⊙ existfrom the work of Barth, Greene & Ho (2005) and Greene & Ho(2006). To go to yet smaller black hole masses, an extrapolationof the correlation between black hole mass and stellar velocitydispersion suggests studying stellar systems with velocity dis-persions of 10–20 km s - . These dispersions are characteristicof globular clusters. Whether the existence of black holes inglobular clusters could shed light on the formation and correla-tions of supermassive black holes is unknown, but clearly it isa possibility. Furthermore, the existence of massive black holesin clusters will have a significant effect on the cluster evolution.Thus, quantifying the demographics of black holes in clustersmay be related to how supermassive black holes grow, and willdefinitely yield useful information about the evolution of clus-ters.Theoretical work suggests that we might expect IMBHs atthe centers of steller systems (Ebisuzaki et al. 2001; PortegiesZwart & McMillian 2002; Miller & Hamilton 2002), althoughit appears to be difficult to make black holes more massive than100 M ⊙ . Gurkan et al. (2004) suggest that IMBHs may be easyto form through runaway collisions with massive stars. Discov-eries of BHs in globular clusters have been claimed — G1 inM31 (Gebhardt, Rich & Ho 2002) and M15 (van der Marel et al. 2002; Gerssen et al. 2002). In fact, the M15 claimhas been made for the past 30 years, starting with the resultof Newell, da Costa & Norris (1976) and subsequently chal-lenged by Illingworth & King (1977). The basic issue is beingable to distinguish a rise in the central mass-to-light ratio beingdue to either a black hole or the expected stellar remnants (neu-tron stars, massive white dwarfs and solar mass black holes).The most recent M15 result has been challenged by Baumgardtet al. (2003a). The result in G1 has also been challenged byBaumgardt et al. (2003b) but Gebhardt, Rich & Ho (2005) in-clude additional data and analysis that support the black holeinterpretation.There has been two further observations which strongly sup-port the existence of a black hole in G1. Trudolyubov & Pried-horsky (2004) measure X-rays from G1 using the Chandra Ob-servatory, centered to within 2 ′′ of the center of G1. Subse-quently, Pooley & Rappaport (2006) suggest the X-ray emmis-sion is from accretion onto a black hole, and Maccarone & Ko-erding (2006) point out that if a black hole is present then a 30 µ Jy radio source may be expected. The most significant obser-vation comes from Ulvestad, Greene & Ho (2007) who find a28 µ Jy (4.5 σ ) emission centered on G1. Other interpretationsare a pulsar wind or a planetary nebula. The pulsar wind seemsunlikely given the age of G1 and the point-like radio source (anold pulsar would have a large size). A planetary nebula wouldshow optical emission lines which are not seen in the HST orKeck spectra of Gebhardt et al. (2003).Other studies of the existence of black holes in globularclusters have been less compelling. Colpi, Mapelli, & Pos-senti (2003) use indirect dynamical arguments to suggest afew hundred solar mass black hole in NGC 6752. McLaugh-lin et al. (2006) provide an estimate of black hole in 47Tucof 900 ± M ⊙ . To date, there are no published upper lim-its of black hole masses that are significantly below that ex-pected from an extrapolation of the correlation between blackhole mass and stellar velocity dispersion.While the dynamical arguements strongly support the blackhole interpretation in at least G1, the radio emission provides aclear and obvious result. Unfortunately, it is difficult to predict Department of Astronomy, University of Texas at Austin 1 University Station C1400, Austin, TX 78712; [email protected], [email protected] National Radio Astronomy Observatory, P.O. Box 0, Socorro NM 878701; [email protected] National Radio Astronomy Observatory, 520 Edgemont Road, Charlottesville, VA 22903; [email protected] TARGET SAMPLE AND RADIO FLUX DENSITYPREDICTIONS
We selected three globular clusters for observation. First,using the stellar velocity dispersion at the center of M15(NGC7078) suggests a revision of the mass of the BH at thecluster’s center to 1000 M ⊙ , higher than Maccarone’s assumedvalue of 440 M ⊙ , making M15 a promising candidate. Second,noting that Baumgardt et al. (2004) argue that highly centrallycondensed globular clusters, as seen from their luminosity pro-files, are unlikely to harbor central IMBHs, we selected twoglobular clusters with large cores that are more likely to havecentral BHs. These clusters, NGC6093 and NGC6266, alsohave large central stellar velocity dispersions.In order to predict radio flux density, the first step is to use anexpected black hole mass. The black hole masses can be esti-mated using an extrapolation of the correlations seen in galax-ies, namely either the black hole mass/velocity dispersion or theblack hole mass/galaxy bulge luminosity relations.A precise prediction of the expected radio flux densitiesbased the black hole mass is quite uncertain. Merloni et al.(2003) use radio flux densities, X-ray luminosities, and mea-sured black hole masses from both galactic and galaxian blackholes to derive a fundamental plane for the three parameters.They argue that using any one parameter to predict another isquite uncertain. Unfortunately, X-ray luminosities do not existfor the three clusters studied here. Furthermore, the Merloniet al. study do not include any black holes with masses from10 to 10 , making any use of the fundamental plane suspectfor the three globular clusters. Therefore, instead of directlyusing expected black hole mass and measured X-ray luminos-ity to predict the radio flux density, we simply use use the lo-cation between the 10 M ⊙ black holes and the galactic blackholes in the fundamental plane. In this region, the expected5 Ghz radio power ranges from 10 - ergs/s. Indeed, forthe G1 radio emission from Ulvestad, Greene & Ho (2007) cor-responds to 10 ergs/s, which is consistent with the measuredblack hole mass of 2 × from Gebhardt, Rich, & Ho (2005).Thus, in order to predict expected flux densities, we adopt thisrange in radio power and use the known distances of the globu-lar clusters. A significant assumption in these estimates is thatthe physical conditions are similar; if, for example, the gas den-sity where much lower in globular clusters, the predicted radiopower would be much less.Alternatively, Maccarone (2004) estimate expected radioemission based on the expected gas density and the correlationof Merloni et al. (2003). The gas density in the cluster comefrom the estimate of Freire et al. (2001) who use differences incolumn densities measured from pulsars in the front and backsides in the globular cluster 47 Tuc. While there is no reasonto expect similar gas densities from cluster to cluster, it is thebest measure we have of gas density in a cluster and thereforewe adopt that value. Maccarone (2004) further assumes thatthe BH is accreting intra-cluster gas at 0.1 and 1% of the Bondiaccretion rate. He assumes the BH mass to be 0.1% of the glob-ular cluster mass, which he estimates from the cluster’s total lu-minosity and an assumed mass-to-light ratio, and computes the expected 5 GHz flux density from the vicinity of the central BHfor 15 globular clusters. Six of the globular clusters in Macca-roni’s list lie north of the southern declination limit of the VeryLarge Array (VLA) and have an estimated 5 GHz flux densityof 40 µ Jy or greater (at 1% of the Bondi rate). We searchedthe VLA archive for observations of the centers of these clus-ters with noise levels low enough to have allowed a detectionat Maccarone’s predicted levels. No VLA archive data werefound which had the required sensitivity.There have been two similar studies to the one presentedhere. Maccarone et al. (2005) provide upper limits for omegaCen using ATCA observations and for M15 from archival VLAobservations. De Rijcke et al. (2006) provide upper limits for47 Tuc and NGC 6397 based on ATCA observations.Table 1 shows the computed flux densities for NGC6266,NGC7078, and NGC6093 at a frequency of 8.6 GHz. We as-sume a spectral index, α = - . OBSERVATIONS
Source positions, integration times on source, beam dimen-sions and position angles, and our 3 σ limits are given in Table2. We used the position of the center of M15 determined byNoyola & Gebhardt (2006) using the optical surface bright-ness profile from the Hubble Space Telescope (HST), which aregood to less than 1 ′′ . We observed that position using the VLAfor 7.5 hours on October 13, 2004 in the A configuration at 8.6GHz ( λ µ Jy/beam andcovers ≈ µ Jy. We do not detectthe other known low mass X-ray binary, M15 X-2 (White &Angelini 2001), even at 1-sigma. The image also contains theplanetary nebula K 648, for which we get: R.A. 21:29:59.39,Dec. 12:10:26.46 (J2000). The measured flux density is 4.2 ± µ Jy /beam.The positions of the centers of NGC6093 and NGC6266 alsocome from Noyola & Gebhardt (2006) (with a similar accuracyof around 1 ′′ ). These clusters were observed on August 11, 20,and 26, 2005 using the VLA in the C configuration at 8.6 GHz.Each map has an rms noise of 12 µ Jy/beam. These maps coveran area of 5.4 arcmin centered on each cluster’s center. No ra-dio source is seen at or above a level of 36 µ Jy/beam near eithercluster’s center. In the case of NGC6093 a source is detected 82arcsec to the SE of the cluster center at R.A. 16:17:05.00, Dec.-22:59:47.3 (J2000) with a flux density of 0.32 ± The Very Large Array is a facility of the National Radio Astronomy Observatory, operated by Associated Universities, Inc., under a cooperative agreement with theNational Science Foundation. ter. However, NGC 6266 (M62) shows about a 2- σ peak at thecenter. F IG . 1.— HST optical image of M15 overlayed with the VLA contours ofthe central 10 ′′ . Positive 1,2, and 3 σ noise contours (8.5, 17, and 25.5 µ Jy)are shown in green and negative are shown in red. The blue circle marks thecenter determined from Noyola & Gebhardt (2006), with a diameter of 0.5 ′′ .North is up and East to the left. The radio source, AC211, is easily seen justnorth of the center. The other known X-ray source, M15 X-2, is not detected.F IG . 2.— HST optical image of NGC 6093 (M80) overlayed with the VLAcontours of the central 10 ′′ . Positive 1,2, and 3 σ noise contours (12, 24, and 36 µ Jy)are shown in green and negative are shown in red. The blue circle marksthe center determined from Noyola & Gebhardt (2006), with a diameter of 1 ′′ .North is up and East to the left. F IG . 3.— HST optical image of NGC 6266 (M62) overlayed with the VLAcontours of the central 10 ′′ . Positive 1,2, and 3 σ noise contours (12, 24, and 36 µ Jy)are shown in green and negative are shown in red. The blue circle marksthe center determined from Noyola & Gebhardt (2006), with a diameter of 1 ′′ .North is up and East to the left. There is about a 2- σ positive 8Ghz signal atthe center. DISCUSSION
Failure to detect radio radiation at 8.6 GHz from the centersof three globular clusters does not prove that no globular clus-ters have IMBHs at their centers. Besides not having a blackhole, other interpretations include 1) accretion by the BH couldbe episodic and we happened to observe the BHs in an “off-state", 2) the gas density could be much lower compared togalaxies, 3) the radiative efficiency may be lower than assumed(although the assumed efficiencies are already quite low), 4) orthe accretion model may not be adequate in general. We wouldpredict, using the relation of Merloni et al. (2003) or usingstandard accretion models and gas density estimates (as done inMaccarone 2004), that we should have detected radio radiationat 8.6 GHz if accretion is steady and the accretion rate timesthe Bondi rate is 10 - × or higher. We would not have been ableto detect the flux density predicted by a rate of 10 - × or less.Ulvestad et al. (2007) estimate the fraction of the Bondi rate ofjust under 1% for G1, but it is difficult to interpret due to theunknown radiative efficiency. For galactic black holes, the ra-diative efficiencies appear to vary greatly with some lower than10 - (Lowenstein et al. 2001), although consistent with rates ofaround 10% of the Bondi rate.Models which predict 8.6 GHz flux densities from centralBHs in globular clusters above about 25 µ Jy/beam can be testedwith the VLA currently. The EVLA should produce, for contin-uum observations, a sensitivity improvement of about a factorof 15, making 8.6 GHz flux densities above about 2 µ Jy/beamdetectable.
REFERENCESBarth, A., Greene, J., & Ho, L.C. 2005, ApJ, 619, L151Baumgardt, H., Hut, P., Makino, J., McMillan, S., & Portegies Zwart, S. 2003a,ApJ, 582, L21Baumgardt, H., Makino, J., Hut, P., & Portegies Zwart, S. 2003b, ApJ, 589, L25Colpi, M., Mapelli, M., & Possenti, A. 2003, ApJ, 599, 1260De Rijcke, S., Buyle, P., & Dejonghe, H. 2006, MNRAS, 368, 43Ebisuzaki, T., et al. 2001, ApJ, 562, L19Ferrarese, L., & Merritt, D. 2000, ApJ, 539, L9Freire, P., Kramer, M., Lyne, A., Camilo, F., Manchester, R. & D’Amico, N.2001, ApJ, 557, L105Gebhardt, K., et al. 2000, ApJ, 539, L13Gebhardt, K., Rich, R.M.R., & Ho, L.C. 2002, ApJ, 578, L41Gebhardt, K., Rich, R.M.R., & Ho, L.C. 2005, ApJ, 634, 1093 Greene, J. & Ho, L.C. 2006, ApJ, 641, L21Gurkan, M., Freitag, M. & Rasio, F. 2004, ApJ, 604, 632Haehnelt, M. G., & Kauffmann, G. 2000, MNRAS, 318, L35Illingworth, G. & King, I. 1977, ApJ, 218, L109Jonhston, H., Kulkarni, S., & Goss, W.M. 1991, ApJ, 382, L89Lowenstein, M., Mushotsky, R., Angelini, L., Arnaud, K., & Quataert, E. 2001,ApJ, 555, L21Maccarone, T. 2004, MNRAS, 351, 1049Maccarone, T. & Koerding 2006, Astronomy & Geophysics, 47, 29Maccarone, T., Fender, R., & Tzioumis, A. 2005, Ap&SS, 300, 247McLaughlin, D., Anderson, J., Meylan, G., Gebhardt, K., Pryor, C., Minniti,D., & Phinney, S. 2006, ApJS, 166, 249Merloni, A., Heinz, S., & di Matteo, T. 2003, MNRAS, 345, 1057
Miller, M.C., & Hamilton, D.P. 2002, MNRAS, 330, 232Newell, B., Da Costa, G., & Norris, J. 1976, ApJ, 208, L55Noyola, E. & Gebhardt, K. 2006, AJ, 132, 447Pooley, D & Rappaport, S. 2006, ApJ, 644, L45Portegies Zwart, S., & McMillan, S. 2002, ApJ, 576, 899Robertson, B., Hernquist, L., Cox, T., Di Matteo, T., Hopkins, P., Martini, P., &Springel, V. 2006, ApJ, 641, 90 Silk, J., & Rees, M. J. 1998 A&A, 331, L1Trudolyubov, S. & Priedhorsky, W. 2004, ApJ, 616, 821Ulvestad, J., Greene, J., & Ho L.C. 2007, ApJ, 661, 151van der Marel, R.P., Gerssen, J., Guhathakurta, R., Peterson, R., & Gebhardt,K. 2002, AJ, 124, 3255White, N., & Angelini, L. 2001, ApJ, 561, L101 T ABLE Z F LUX D ENSITY V ALUES
Cluster M BH M BH Distance Flux DensityMaccarone, M sun this paper, M sun kpc µ JyNGC6093 (M80) · · · × - NGC6266 (M62) 450 3000 6 3 × - NGC7078 (M15) 440 1000 10 1 × - T ABLE BSERVATIONS
Cluster RA DEC Integration Beam; Pos. Ang 3 σ LimitJ2000 J2000 hours arcseconds; deg µ JyNGC6093 16:17:05.00 - . × .
3; -6 36NGC6266 17:01:12.96 - . × .
2; -6 36NGC7078 21:29:58.35 + ..
2; -6 36NGC7078 21:29:58.35 + .. × ..