1: \documentclass[useAMS,usenatbib, usegraphicx]{mn2e}
2: \topmargin -0.5in
3: \usepackage{epsfig}
4: %\usepackage{lscape}
5: %\usepackage{graphicx}
6: %\usepackage{color}
7: \usepackage{amsmath}
8: \usepackage{amssymb}
9: %\usepackage{amsfonts}
10: %\usepackage{amstext}
11: %\usepackage{amsbsy}
12: \usepackage{natbib}
13: %\loadbold
14: \bibliographystyle{mn2e}
15: %\newcommand{\tbfrac}[2]{\genfrac{}{}{0pt}{0}{#1\strut}{#2\strut}}
16: \newcommand{\kms}{\ensuremath{{\rm km\,s}^{-1}}}
17: \newcommand{\msun}{\ensuremath{{\rm M}_{\odot}}}
18: \newcommand{\lsun}{\ensuremath{{\rm L}_{\odot}}}
19: \newcommand{\mbh}{\ensuremath{m_{\rm BH}}}
20: \newcommand{\yr}{\ensuremath{\rm yr}}
21: \newcommand{\Myr}{\ensuremath{\rm Myr}}
22: \newcommand{\Gyr}{\ensuremath{\rm Gyr}}
23: \newcommand{\pc}{\ensuremath{\rm pc}}
24: \newcommand{\kpc}{\ensuremath{\rm kpc}}
25: \newcommand{\Mpc}{\ensuremath{\rm Mpc}}
26: \newcommand{\Gpc}{\ensuremath{\rm Gpc}}
27: \newcommand{\new}{}
28: \newcommand{\infinity}{{\infty}}
29: \newcommand{\apj}{ApJ}
30: \newcommand{\apjl}{ApJ}
31: \newcommand{\mnras}{MNRAS}
32: \newcommand{\aj}{AJ}
33: \newcommand{\apjs}{ApJS}
34: \newcommand{\nat}{Nat}
35: \newcommand{\pasj}{PASJ}
36: \newcommand{\aap}{A\&A}
37: \newcommand{\letter}{{letter}}
38: \newcommand{\sag}{Sgr~A*}
39: \newcommand{\rmd}{{\rm d}}
40: \newcommand{\physrep}{Physics Reports}
41: \newcommand{\aaps}{A\&AS}
42: \newcommand{\prd}{PhRvD}
43: \newcommand{\msmbh}{\ensuremath{M_{\bullet}}}
44: %\begin{document}
45: \title[Recoiled BHs in the Milky Way]{Star clusters around recoiled
46: black holes in the Milky Way halo}
47:
48: \author[O'Leary \& Loeb]{Ryan M.\
49: O'Leary\thanks{E-mail:roleary@cfa.harvard.edu} and Abraham
50: Loeb\thanks{E-mail:aloeb@cfa.harvard.edu}\\Harvard University,
51: Department of Astronomy, 60 Garden St., Cambridge, MA 02138, USA\\}
52:
53: \begin{document}
54: \maketitle
55:
56: \begin{abstract}
57:
58: Gravitational wave emission by coalescing black holes (BHs) kicks
59: the remnant BH with a typical velocity of hundreds of $\kms$. This
60: velocity is sufficiently large to remove the remnant BH from a
61: low-mass galaxy but is below the escape velocity from the Milky Way
62: (MW) galaxy. If central BHs were common in the galactic building
63: blocks that merged to make the MW, then numerous BHs that were
64: kicked out of low-mass galaxies should be freely floating in the MW
65: halo today. We use a large statistical sample of possible merger
66: tree histories for the MW to estimate the expected number of
67: recoiled BH remnants present in the MW halo today. We find that
68: hundreds of BHs should remain bound to the MW halo after leaving
69: their parent low-mass galaxies. Each BH carries a compact cluster of
70: old stars that populated the core of its original host galaxy.
71: Using the time-dependent Fokker-Planck equation, we find that
72: present-day clusters are $\lesssim 1\,$pc in size, and their central
73: bright regions should be unresolved in most existing sky
74: surveys. These compact systems are distinguishable from globular
75: clusters by their internal (Keplerian) velocity dispersion greater
76: than one hundred km~s$^{-1}$ and their high mass-to-light ratio
77: owing to the central BH. An observational discovery of this relic
78: population of star clusters in the MW halo, would constrain the
79: formation history of the MW and the dynamics of BH mergers in the
80: early Universe. A similar population should exist around other
81: galaxies and may potentially be detectable in M31 and M33.
82:
83: \end{abstract}
84:
85: \begin{keywords}
86: galaxies:kinematics and dynamics--galaxies:nuclei--black hole
87: physics--gravitational waves--star clusters
88: \end{keywords}
89:
90:
91: \section{Introduction}
92:
93: During the final coalescence of two black holes (BHs), gravitational
94: waves (GWs) are emitted unisotropically and carry away linear
95: momentum, thus kicking the merger BH remnant in the opposite direction
96: \citep{1962PhRv..128.2471P,1973ApJ...183..657B,1983MNRAS.203.1049F}.
97: The resulting kick velocity of typically hundreds of $\kms$ depends on
98: the mass ratio of the BHs as well as the spin and orientation of the
99: binary before coalescence
100: \citep{2006ApJ...653L..93B,2007PhRvL..98w1102C,2007ApJ...659L...5C,2007PhRvD..76f1502T}. Such
101: kicks can alter both the the population of nuclear BHs in galactic
102: bulges \citep{2004ApJ...606L..17M,2006MNRAS.368.1381L,2007ApJ...663L...5V,2007ApJ...667L.133S,2008arXiv0805.1420B} as well
103: as the core of stars in the bulge itself \citep{2008ApJ...678..780G}.
104: The discovery of a remnant population of recoiled BHs in the
105: present-day Universe
106: can provide a new window into the merger statistics and spin
107: distribution of the BHs \citep{2006MNRAS.368.1381L,2007ApJ...663L...5V}, as well as test
108: general relativity in the strong regime.
109:
110: Recoiling BHs may be detected through their unique flaring
111: \citep{2008ApJ...676L...5L,2008ApJ...682..758S,2008arXiv0802.3556S}
112: or observed as spatially and kinematically offset quasars
113: \citep{2004ApJ...606L..17M}. If the BH binary was surrounded by
114: an accretion disk, the ejected remnant BH would carry the disk with it
115: and shine as a quasar \citep{Loeb}. A search for spectral shifts in the broad
116: lines of quasars relative to the narrow emission lines of their parent
117: galaxies resulted mainly in upper limits \citep{2007ApJ...666L..13B}
118: and possibly one suggested candidate \citep{2008ApJ...678L..81K} for
119: an ejected quasar. Unfortunately, these quasars lose their source of
120: gas for accretion, and have short lifetimes
121: \citep{Loeb,2008arXiv0805.1420B,2008ApJ...687L..57V}.
122:
123: Past studies of the observational consequences of GW recoil have
124: generally focused on massive BHs with a mass $\msmbh \gtrsim
125: 10^7\,\msun$. Such BHs reside in the most massive galaxies, and
126: require the greatest and rarest kick velocities in order to escape
127: their parent galaxies \citep{2007ApJ...662L..63S}. But since the kick
128: velocity from BH mergers is independent of the total mass of the
129: system, the dynamical consequences are most prominent for less massive
130: BHs which tend to reside in low-mass galaxies with shallower potential
131: wells and lower escape velocities
132: \citep{2004ApJ...606L..17M,2005MNRAS.358..913V,2006ApJ...653L..93B,2006MNRAS.368.1381L,2007ApJ...663L...5V,2008arXiv0807.4702T}. In fact, most BHs with
133: $\msmbh \lesssim 10^5\,\msun$ will likely be kicked out of their
134: parent galaxies as a result of major mergers
135: \citep{2006ApJ...653L..93B}. During the hierarchical build-up of the
136: Milky-Way (MW) galaxy, mergers of low-mass galaxies were common and
137: should have resulted in a population of freely floating BHs \citep{2005MNRAS.358..913V}.
138: Since the gravitational potential of an overdense region does not
139: evolve dramatically during the growth of cosmological structure
140: \citep{2006astro.ph..3360L} the region that eventually collapsed to
141: make the MW was able to trap those BHs with the most common kick
142: velocities ($\lesssim 500~{\rm km~s^{-1}}$) even before the MW had
143: formed. Each of the ejected BHs carries with it a star cluster that
144: used to populate the core of its parent galaxy
145: \citep{2008ApJ...678..780G, 2008ApJ...678L..81K}. In this paper we
146: examine the observational signatures of the ejected star clusters
147: which are expected to be floating in the MW halo. The discovery of a
148: relic population of the star clusters attached to recoiled BHs would
149: provide a new window for cosmology, allowing one to constrain the
150: merger history of the MW as well as the formation history of the first
151: population of massive BHs.
152:
153: {\new Previous studies have looked at populating the MW halo with
154: recoiling BHs as analysed again here
155: \citep{2004ApJ...606L..17M,2005MNRAS.358..913V,2006MNRAS.368.1381L},
156: through three body encounters \citep{2005MNRAS.358..913V}, as well
157: as a variety of other processes. Other sources of wandering BHs are
158: the remnants of Population III stars
159: \citep{2003MNRAS.340..647I,2005PhRvL..95a1301Z} as well as the the
160: direct collapse of baryons into a BH
161: \citep{2005PhRvD..72j3517B,2006MNRAS.368.1340M}. In both of these
162: cases, however, the BH either is presently located in the center of
163: a galaxy or is ``naked'' without any stellar companions. Such BHs
164: are likely to only be located by their interaction with the matter
165: that surrounds them
166: \citep{2004MNRAS.354..629I,2004MNRAS.354..443I,2005MNRAS.358..913V,2005ApJ...628..873M,2005PhRvD..72j3517B}.
167: Additionally, dwarf galaxies may be tidally stripped of their outer
168: stars, leaving behind a BH surrounded by a massive cluster of stars
169: similar to some globular clusters
170: \citep{1988IAUS..126..603Z,1994ApJ...431..634B}. In contrast, the
171: new source of BHs presented in this paper uniquely have a population
172: of bound stars that are only a fraction of the mass of the central BH. }
173:
174: In \S~\ref{sec:hier} we consider the formation history of the MW to
175: estimate the number and distribution of ejected BHs that may be found
176: in its halo today. We then follow the long term evolution of the star
177: clusters that are attached to them in \S~\ref{sec:relax}. In
178: \S~\ref{sec:observe} we discuss the observational signatures of these
179: systems. Finally, \S~\ref{sec:disc} summarizes our main results and
180: their implications.
181:
182: \section{The Merger History of the MW and Ejected BHs in the Halo }
183: \label{sec:hier}
184:
185: In the standard cosmological context of hierarchical galaxy formation
186: \citep{2008arXiv0809.0898S} the MW formed from an overdense region of
187: the universe with a present-day virial mass $M_{\rm vir} \approx
188: 1$--$2\times 10^{12}\, \msun$
189: \citep{2002ApJ...573..597K,2008MNRAS.384.1459L}. The threshold for
190: cooling by atomic transitions limits the minimum mass of star forming
191: galaxies at high redshifts to $M_{\rm gal}\sim 10^8\,\msun$
192: \citep{2006Natur.441..322W}. If the earliest population of
193: star-forming galaxies had BHs at their center, then their shallow
194: potential would be insufficient to retain the recoiling BHs after
195: mergers with a BH mass ratio $q\equiv (M_{\bullet,1}/M_{\bullet,2})
196: \gtrsim 0.1$. Nevertheless, since most kicks are modest ($\lesssim
197: 500\,\kms$), the ejected BHs remain confined to the MW.
198:
199: We estimate the number of BHs in the MW halo, $N_{\rm BH}$, by
200: analyzing a large statistical ensemble of possible merger tree
201: histories for the MW using analytic models for the distribution of
202: kick velocities and galaxy profiles. In particular, we generate
203: $\gtrsim 10^3$ Monte-Carlo realizations of the merger history of the
204: MW galaxy with the merger tree code made publicly available
205: by \citet{2008MNRAS.383..557P}. This code follows a modified form of
206: the extended \citet{1974ApJ...187..425P} formalism and is normalized to
207: match the Millennium run numerical simulation
208: \citep{2005Natur.435..629S,2008MNRAS.383..546C}. We assume that the
209: MW has a mass of $1.5\times 10^{12}\,\msun$, and restrict our
210: simulations to formation scenarios in which the MW undergoes no major
211: mergers ($q>1/3$) since a redshift $z=1$ \citep{2007ApJ...657..262D}.
212:
213: We determine the properties of the halo from its circular velocity at
214: the virial radius \citep{2001PhR...349..125B},
215: \begin{equation}
216: \label{eq:vcirc}
217: v_{\rm c} = 24 \left( \frac{M_{\rm gal}}{10^8 h^{-1}\,\msun}
218: \right)^{1/3} \left(\frac{\Omega_m}{\Omega_m^z}
219: \frac{\Delta_c}{18\pi^2}\right)^{1/6} \left(\frac{1+z_{\rm
220: merge}}{10}\right)^{1/2} \,\kms,
221: \end{equation}
222: where $\Delta_c=18\pi^2+82d-39d^2$, $d=\Omega_m^z-1$, $\Omega_m^z =
223: (\Omega_m (1+z)^3/ (\Omega_m (1+z)^3+\Omega_\Lambda)$, evaluated at
224: the merger redshift $z$ and we adopt the values for the
225: cosmological density parameters, $\Omega_m$ and $\Omega_\Lambda$, used
226: in the Millennium run \citep{2005Natur.435..629S}.
227: In each galaxy, we assume that the dark matter follows an NFW radial
228: profile \citep{1996ApJ...462..563N} with a concentration parameter
229: $c=4$ out to the virial radius, as expected for a newly formed
230: dark-matter halo \citep{2002ApJ...568...52W}. In this case, the
231: escape velocity from the galaxy's center is $v_{\rm esc} \approx 2.8
232: v_{\rm c}$. We scale the stellar velocity dispersion of the bulge in
233: the low-mass galaxies similarly to the MW and adopt $\sigma_\star =
234: v_{\rm c}/2$. The BH mass in the center of each galaxy is then
235: dictated by the observed $\msmbh$--$\sigma_\star$ relation
236: \citep{2002ApJ...574..740T}
237: \begin{equation}
238: \label{eq:msigma}
239: \msmbh = 8.1\times 10^6 M_\odot
240: \left(\frac{\sigma_\star}{100\,\kms}\right)^4 .
241: \end{equation}
242: We assume that the merger of the BHs is efficient, occurring before a
243: third BH is introduced into the system, and that the remnant BH mass
244: is $M_{\rm \bullet,final} = M_{\bullet,1}+M_ {\bullet,2}$. Although
245: the $\msmbh$--$\sigma_\star$ relation was initially determined from
246: the analysis of more massive BHs,
247: \citet{2005ApJ...619L.151B} and \citet{2006ApJ...641L..21G} have found that it
248: consistently extends to active galactic nuclei with BHs with masses
249: $\sim 10^5\,\msun$, {\new even without classical bulges \citep{2008ApJ...688..159G}.}
250:
251: For each merger tree, we randomly assign a kick velocity from analytic
252: models of the kick velocity distribution of
253: \citet{2007ApJ...662L..63S} assuming that all the BHs have the same
254: spin with two different values: $a = 0.1$ and $0.9$. A BH
255: is ejected from its host galaxy if the kick velocity is larger than
256: the escape velocity, $v_k > v_{\rm esc}$, and remains in the larger
257: potential of the MW if the final velocity is less than the escape
258: velocity of the MW halo, $v_{\rm ej} = \sqrt{v_{\rm k}^2 - v_{\rm
259: esc}^2} \lesssim 500\,\kms$. Although the MW escape velocity
260: evolves by tens of percent as the MW halo assembles\footnote{The
261: gravitational potential does not evolve during the linear growth of
262: perturbations at redshifts $z\gtrsim 1$, and is only modestly
263: enhanced in the final collapse of galaxy halos
264: \citep{2006astro.ph..3360L}.}, we neglect this evolution in our
265: analysis.
266:
267: In our realizations of the MW merger tree, there were an average of
268: $\approx 1500$ galaxy mergers for each run. Of these, $\approx 700$
269: were major mergers with $q > 0.1$. For BHs with spin $a = 0.9$
270: ($a=0.1$) gravitational wave recoil ejected $\approx 570$ ($\approx
271: 440$) BHs from their parent galaxy, and $N_{\rm BH} \approx 330$
272: ($\approx 440$) remained within the MW halo. In all of our
273: simulations, the average kick velocity, even for the most massive BHs
274: ejected, is comparable to or less than the present day velocity
275: dispersion of the halo $\lesssim 200\,\kms$. Thus, we expect that
276: most of the BHs trace the distribution of the dark matter in the halo.
277: {\new This is consistent with the results of
278: \citet{2006MNRAS.368.1381L} who used $N$-body simulations to follow
279: the kicked BHs in MW halo. The BHs which were not ejected from
280: their host galaxies settle back into the core of galaxies owing to
281: dynamical friction, however, the vast majority of these galaxies
282: have since merged to form the Milky Way.}
283:
284: In Figure~\ref{fig:dist}, we plot the cumulative flux distribution of
285: the remnant BHs, assuming that the BHs follow the present-day NFW
286: profile of the MW with $c = 12$ and $r_{\rm vir} = 200\,\kpc$
287: \citep{2002ApJ...573..597K,2008MNRAS.384.1459L}. We determine the
288: luminosity from Eq.~(\ref{eq:nbound}) in \S~\ref{sec:relax}, and
289: report the flux in units of the flux from the Sun if it were at a
290: distance of $1\,\kpc$, $f_{\odot,{\rm kpc}}$, and also the apparent
291: bolometric magnitude (top axis). Assuming that these clusters lose
292: little mass, we expect that nearly all remnant BHs with mass $\msmbh
293: \gtrsim 2\times 10^3\,\msun$ in the MW halo would be visible to the
294: depth of {\new {\it Sloan Digital Sky Survey} }\footnote{http://www.sdss.org} (SDSS).
295:
296: In all our models, we find the mass distribution of the ejected BHs to
297: scale roughly as $\rmd N_{\rm BH} / \rmd \msmbh \propto \msmbh^{-1}$
298: (with the $a=0.1$ case showing a slightly steeper distribution on the
299: high mass end of BHs). The most massive BH present in the halo is $
300: 6.3^{+5.7}_{-2.8}\times 10^5\,\msun$ for $a=0.9$, and $
301: 1.7^{+0.8}_{-0.5}\times 10^5\,\msun$ for $a=0.1$, where the quoted
302: uncertainty represents the one standard deviation. The closest BH that
303: has an apparent magnitude $< 21^{\rm m}$, the approximate depth of
304: SDSS, has an average mass of $\sim 3\times 10^3\,\msun$ and is $\sim
305: 1\,\kpc$ away. {\new For the typical distances and masses of the BHs}, dynamical friction is
306: negligible over the age of the universe.
307: {\new For example, the dynamical friction timescale of a $\sim 10^5\,\msun$
308: BH is $\gtrsim 10^{11}\,\yr$ at even $\sim 1\,$kpc from the galactic center \citep{1987gady.book.....B}. }
309:
310: {\new Given the assumptions used here and the uncertainties in the
311: lower limit for the BH mass, our results are consistent with the
312: findings of previous analyses of BHs in the MW halo, even though
313: only a few papers
314: \citep[i.e.,][]{2005MNRAS.358..913V,2006MNRAS.368.1381L} explicitly
315: account for GW recoil. Previous studies that focused on BHs that
316: formed in the smallest overdensities in the early universe ($\sim
317: 10^6\,\msun$ at $z\gtrsim 20$) found that there may be up to
318: $10^3-10^4$ BHs with masses $\gtrsim 10^2\,\msun$
319: \citep{2004MNRAS.354..427I,2005PhRvL..95a1301Z,2005PhRvD..72j3517B,2006MNRAS.368.1340M}.
320: \citet{2005MNRAS.358..913V} and \citet{2006MNRAS.368.1381L} used
321: more conservative estimates for the threshold of forming seed BHs,
322: and accounted for the growth of the BH owing to accretion. They
323: concluded that there are more likely only $\sim 10^2$ such BHs in
324: the halo. Here, we sidestepped the issue of how and when the seed
325: BHs formed, and assumed that all galaxies (with mass $\gtrsim
326: 10^8\,\msun$) have a central BH that follows the
327: $\msmbh-\sigma_\star$ relationship, which consistently extends to
328: the lowest mass BHs observed \citep{2006ApJ...641L..21G}. With
329: these conditions for the BH mass, our results are most consistent
330: with \citet{2005MNRAS.358..913V} and \citet{2006MNRAS.368.1381L},
331: however we do not follow the evolution of BHs with mass $\lesssim
332: 10^3\,\msun$.
333:
334: }
335:
336: \begin{figure}
337: \centering \includegraphics[width=\columnwidth]{massdist3.ps}
338: \caption{The cumulative distribution of ejected star clusters in the
339: MW Halo. Plotted is the flux distribution associated with BHs
340: masses greater than $10^3\,\msun$ (black), $10^4\,\msun$ (blue),
341: and $10^5\,\msun$ (red), in our models with BH spin $a=0.1$
342: (solid) and $a=0.9$ (dashed) lines, plotted in units of the flux
343: of the Sun at a distance of $1\,\kpc$ ($f_{\odot,{\rm kpc}}$).
344: The top axis is labelled with the apparent bolometric magnitude of
345: the clusters. Nearly all BHs with $\msmbh \gtrsim 2\times
346: 10^3\,\msun$ have apparent magnitudes greater than 21, the rough
347: magnitude limit of SDSS. The mass distribution of the ejected BHs
348: has approximately equal number per $\log\msmbh$ interval, with $\rmd
349: N_{\rm BH}/ \rmd \msmbh \propto \msmbh^{-1}$. }
350: \label{fig:dist}
351: \end{figure}
352:
353: \section{Structure and Evolution of Ejected Star Clusters}
354: \label{sec:relax}
355:
356: {\new The number of stars that remain bound to the ejected BH will
357: depend on the stellar distribution immediately before merger and the
358: magnitude of the kick from GWs. The stellar distribution is
359: determined by the} relaxation timescale at the radius of influence of
360: the BH binary, $r_i = G \msmbh / \sigma_\star^2$,
361: \citep{1976ApJ...209..214B}
362: \begin{equation}
363: \label{eq:relaxtime} t_r = \frac{3(2\pi \sigma_\star^2)^{3/2}}{32 \pi^2 G^2
364: m_\star^2 n_\star \ln{\Lambda}},
365: \end{equation}
366: where $\sigma_\star$ is the stellar velocity dispersion after the
367: galaxies merge, $m_\star$ is the average stellar mass, $n_\star$ is the number
368: density of stars at $r_i$, and $\ln \Lambda \approx \ln (\msmbh/M_\star)$
369: is the Coulomb logarithm. From the $\msmbh$--$\sigma_\star$ relation
370: (Eq.~\ref{eq:msigma}) the relaxation timescale can be written as
371: \begin{equation}
372: \label{eq:relaxsig}
373: t_r \approx 10^{9} \left(\frac{\msmbh}{ 10^5 \,\msun}\right)^{5/4} \,\yr,
374: \end{equation}
375: if we assume that $m_\star = 1\,\msun$, and that the total mass in
376: stars interior of $r_i$ is $\sim 2\msmbh$. Within a relaxation
377: timescale, the stars form a density cusp within the radius of
378: influence around the central BH, with $n_\star(r) \propto
379: r^{-\alpha}$, where $\alpha = 1.75$ for a population of equal mass
380: stars and $\alpha \approx 1.5-2.0$ for varying mass distributions of
381: the stars and compact remnants \citep{1976ApJ...209..214B,
382: 1977ApJ...216..883B, 2006ApJ...649...91F, 2006ApJ...645L.133H}. If
383: the BHs coalesce on a time much shorter than the relaxation timescale
384: of the stars {\new owing to a gas rich merger}, then the stellar
385: density profile is likely much shallower with $\alpha \approx 1$ due
386: to the scattering off the inspiraling binary
387: \citep{2006ApJ...648..890M, 2007ApJ...671...53M}. However, new stars
388: might form out of the gas and introduce a new cusp of stars. In
389: either case, however, the relaxation timescale of the stellar systems
390: is much shorter than a Hubble time, and the present-day distribution
391: of stars is different than its initial condition.
392:
393: Assuming that the stars follow a power-law density profile and that
394: the total stellar mass within $r_i$ is $2\msmbh$, the stellar density
395: profile before the BH ejection is
396: \begin{equation}
397: \label{eq:nstar}
398: n_\star(r) = \frac{\msmbh}{m_\star}\frac{3-\alpha}{2\pi r_i^3}
399: \left(\frac{r}{r_i} \right)^{-\alpha}.
400: \end{equation}
401: The kick is imparted to the BH merger remnant on a timescale much
402: shorter than the dynamical time of the star cluster. In the frame of
403: the BH, all stars receive kicks with a reflex velocity $-v_k$.
404: Stars with total energies $\gtrsim -m_\star v_k^2/2$ will become unbound
405: to the BH. In the Keplerian potential of the BH, this roughly
406: corresponds to stars at $r \gtrsim r_k = \sqrt{G \msmbh/v_k^2} =
407: (\sigma_\star/v_k)^2 r_i$. The total number of stars that remain
408: bound the the BH is then,
409: \begin{equation}
410: \label{eq:nbound}
411: N_{\rm cl} \approx \frac{2 \msmbh}{m_\star}
412: \left(\frac{v_k}{\sigma_\star}\right)^{2a - 6}.
413: \end{equation}
414: For $\alpha = 1.75$ the total number of stars that remain bound for a
415: minimally ejected BH with $v_k = 5.6\sigma_\star$ is then $\sim
416: 4\times 10^{3} (\msmbh/10^5\msun)$. The actual number of stars may be
417: less due to the ejection of stars by the BH binary inspiral
418: \citep{2006ApJ...648..890M, 2007ApJ...671...53M}. These stars will
419: mainly remain within a radius $r_k \approx (\sigma_\star/v_k)^2 r_i
420: \ll 1\,\pc$, and have a $\alpha = 4$ profile for $r \gtrsim r_k$
421: \citep{2008ApJ...683L..21K}. Given the short relaxation timescale of
422: the star cluster and the lack of a source of new stars, the star
423: cluster will expand.
424:
425: \begin{figure*}
426: \centering
427: \includegraphics[width=\columnwidth]{Scombo-normal.ps}
428: \includegraphics[width=\columnwidth]{Scombo-flat.ps}
429: \caption{The total number of projected stars interior to $r$,
430: $N_p(<r)$, for $\msmbh = 10^5\msun$ and $\alpha = 1.75$ (left)
431: {\new and $\alpha = 1$ (right)}. The solid line corresponds to
432: the cluster immediately after being ejected from its parent galaxy
433: with $v_k = 5.8 \sigma_\star$,{\new normalized to have the same number
434: density at the radius of influence}. The alternating dashed and
435: dotted lines correspond to the projected number of stars after
436: every $10 t_r \approx 650\,\Myr$. Immediately after ejection, the
437: cluster rapidly expands until its relaxation time becomes
438: comparable to the age of the Universe, with very little mass
439: loss. n For a $10^5\,\msun$ BH, the circular velocity of the stars
440: is $\approx 66\,\kms (r/0.1\,\pc)^{-1/2}$. }
441: \label{fig:densityevolution}
442: \end{figure*}
443:
444: We determine the long term evolution of the ejected cluster by
445: numerically solving the time dependent, angular momentum averaged,
446: Fokker-Planck equation for stars around a central massive object
447: \citep{1976ApJ...209..214B,1977ApJ...216..883B} {\new
448: \begin{equation}
449: \label{eq:fokkerplanck}
450: \frac{\partial g(x,\tau)}{\partial \tau} = -x^{5/2}
451: \frac{\partial}{\partial x} Q(x) - R(x),
452: \end{equation}
453: where $x = -E/(m_\star \sigma_\star^2)$ is the dimensionless energy,
454: $\tau = t / t_r$ is the dimensionless time, $g(x,\tau) = [(2 \pi
455: \sigma_\star^2)^{3/2} n_\star^{-1}]f(E)$ is the dimensionless
456: distribution function of the stars, $Q(x)$ is the rate stars flow to
457: energies larger than $x$, and $R(x)$ is the tidal disruption rate of
458: stars by the BH \citep{2008ApJ...683L..21K}. In these units,
459: \begin{eqnarray}
460: \label{eq:flowrate}
461: Q(x) = \int_{-\infinity}^{x_{\rm td}}
462: \rmd y [{\rm max} (x,y)]^{-3/2}
463: \left(g(x) \frac{\partial
464: g(y)}{\partial y} - g(y)\frac{\partial
465: g(x)}{\partial x}\right).
466: \end{eqnarray}
467: and
468: \begin{equation}
469: \label{eq:losscone}
470: R(x) = \frac{g(x)^2}{ \ln[J_c(x)/J_{\rm LC}]},
471: \end{equation}
472: where $J_c(x) = G \msmbh \sigma_\star^{-1} (2 x)^{-1/2}$ is the
473: angular momentum of a circular orbit at a radius $r \approx r_i / (2 x)$,
474: and $J_{\rm LC}$ is the maximum angular momentum for the star to be
475: destroyed or consumed by the BH. For a star of radius $R_\star$ $J_{\rm
476: LC} = G \msmbh \sigma^{-1}x_{\rm td}^{-1} (2(x+x_{\rm
477: td}))^{1/2}$, where $x_{\rm td} \approx (\msmbh / m_\star)^{-1/3} r_i/R_\star$.
478: We assume that far from the BH the stars initially follow a Maxwellian
479: distribution with dispersion $\sigma_\star$. In dimensionless units,
480: the distribution function of the unbound stars is just $g(x)=exp(x)$
481: for $x>0$. For a power-law distribution function with $g(x)\propto
482: x^{p}$, the density profile of the stars will also follow a power-law
483: $n_\star \propto r^{-\alpha}$, with $\alpha = -p - 3/2$. }
484:
485: We have performed two sets of calculations, {\new first for a binary
486: that merges because of stellar relaxation, and then for a gas rich
487: merger. For the first case}, we evolved the system for one
488: relaxation timescale, at which point the system has reached steady
489: state, and then we removed all unbound stars by setting $g(x) = 0$ for
490: $x>0$. The initial kick on the system breaks the spherical symmetry
491: assumed here. However, after a relaxation timescale we expect the
492: system to reach spherical symmetry again. To represent the kick on
493: the bound stars, we scale the distribution function of stars as $g(x)
494: \rightarrow g(x) z^{2.5}/(1+z^{2.5})$, where $z = x /
495: (v_k/\sigma_\star)^2$. This yields an asymptotic density profile with
496: $n \propto r^{-4}$ for $r \gtrsim r_k$, as expected immediately after
497: the kick \citep{2008ApJ...683L..21K}. To examine the evolution of the
498: star system if the BHs coalesce through a gas rich merger
499: \citep{2008arXiv0808.3088C}, we adopt $g(x) \propto x^{-1/2}
500: z^{3}/(1+z^{3})$, normalized by the same number density and relaxation
501: timescale used in Eq.~(\ref{eq:relaxtime}). {\new This yields $\alpha
502: \sim 1$ for $r \lesssim r_k$ and $\alpha \sim 4$ for $r \gtrsim
503: r_k$, as expected following a merger
504: \citep{2006ApJ...648..890M,2007ApJ...671...53M}.} This run results
505: in $\sim 10\%$ the number of bound stars as with $\alpha = 1.75$. Its
506: importance, however, is uncertain given the short relaxation
507: timescales for these systems.
508:
509: Shortly after the GW recoil kicks the BH, the star cluster expands as
510: a result of two-body relaxation and evolves until the relaxation
511: timescale of the cluster becomes comparable to the age of the
512: system. Figure~\ref{fig:densityevolution} shows the evolution of the
513: projected number of stars interior to a radius $r$ for a BH with mass
514: $10^5\,\msun$ for both $\alpha = 1.75$ and $\alpha = 1$ density
515: profiles. Since our simulations indicate little mass loss, the
516: condition that the relaxation timescale of the system
517: (Eq.~\ref{eq:relaxtime}) be comparable to the age of the Universe
518: implies that the cluster should expand by a factor of order
519: $(t_h/t_r)^{1/3}\sim 10$. Even then, the present day cluster size is
520: less than the initial radius of influence of the BH, $r_i \lesssim
521: 1\,\pc$, and much smaller than the tidal radius of the system,
522: distinguishing them from globular clusters with a similar stellar mass
523: (owing to the gravitational binding provided by the central BH).
524:
525: In our simulations there is very little mass loss due to the tidal
526: disruption of stars or the ejection of stars from the system. Since
527: the cluster expands so rapidly, the rate of tidal disruption of stars
528: rapidly decreases, and is much lower than the rate associated with a
529: comparable BH in the nucleus of a galaxy.
530:
531: Our simulations only treat the evolution of the cluster due to many
532: small-angle scatterings (two-body relaxation), and does not include
533: the strong scattering that might occur in a dense stellar
534: environment. Such scatterings would launch stars on eccentric orbits
535: that take them outside of the cluster to about the tidal radius, thus
536: enhancing the mass loss from the system. The existence of a halo of
537: such stars provides an additional signature unique to the compact star
538: clusters associated with recoiled BHs. {\new We also ignore the
539: effects of resonant relaxaation, which can deplete the number of
540: stars both before and after the ejection of of the BH
541: \citep{1996NewA....1..149R,1998MNRAS.299.1231R}. Although resonant
542: relaxation has been approximately accounted for in
543: Eq.~\ref{eq:fokkerplanck} by averaging over angular momentum
544: \citep{2006ApJ...645.1152H}, a full multidimensional analysis
545: accounting for angular momemtum is best suited for these clusters.}
546:
547: \section{Observing Ejected Systems}
548: \label{sec:observe}
549:
550: \noindent {\em Size and Structure:} Based on the results in
551: \S~\ref{sec:hier}, we estimate that there should be $\sim 300$
552: recoiled BHs in the MW halo. Equation~(\ref{eq:nbound}) implies that
553: these BHs are surrounded by $\sim 4\times 10^{3} (\msmbh/10^5
554: m_\star)$ stars with a half mass radius, $r_h \lesssim 1\,\pc$. At the
555: typical distance to the most massive recoiled BH, $\sim 100\,\kpc$,
556: the cluster will have a typical angular size $\lesssim 1$ arcsec,
557: below the resolution limit of most existing sky surveys. Thus, the
558: cluster will appear as a point source and might be confused as a
559: single foreground star with unusual colors. The closest clusters are
560: $\sim 1\,\kpc$ away and may be a few arcmin in size. However, these
561: clusters surround the smallest BHs ($\sim 10^3\,\msun$) with the
562: fewest number of stars, and their present day distribution is
563: uncertain owing to strong encounters between the stars in the cluster
564: as well as stars in the disk. To date, no star cluster has been
565: identified in the MW with such a small size; the smallest star cluster
566: discovered has a size of $\sim 3\,\pc$
567: \citep{2007ApJ...669..337K}. Clusters at distances of tens of kpc
568: could in principle be distinguished from point sources by comparing
569: their extended image to the point-spread-function of the telescope.
570: {\new In comparison, the stellar mass of globular clusters and the tidally
571: stripped cores of dwarf galaxies completely outweights the total
572: mass of any central BH in such systems. By measuring the total mass of
573: stars one can approximately determine how large the kick velocity
574: may have been.}
575:
576: \noindent {\em Color:} A recoiled star cluster can be distinguished from
577: single foreground stars by its anomalous color. The cluster is
578: likely old, $\sim 1-10\,\Gyr$, and to first order should have colors
579: similar to globular clusters of a similar age and metallicity.
580:
581: \noindent {\em Spectra:} The velocity dispersion profile of the
582: ejected star clusters should be Keplerian. The unusually large width
583: of their spectral lines (greather than one hundred km~s$^{-1}$) can be
584: used to distinguish these clusters from foreground stars, which
585: possess much lower rotational velocities.
586:
587: The compact star clusters around ejected BHs are distinguishable from
588: cores of globular clusters by their internal (Keplerian) velocity
589: dispersion of order one hundred km~s$^{-1}$ and their high
590: mass-to-light ratio owing to the central BH.
591:
592: \noindent {\em Proper Motion:} The most massive systems that reside near the
593: virial radius of the galaxy and have a relatively small proper motion.
594: Nevertheless, future surveys like
595: GAIA\footnote{http://www.esa.int/esaSC/},
596: Pan-STARRS\footnote{http://pan-starrs.ifa.hawaii.edu/}, and
597: LSST\footnote{http://www.lsst.org/}, could aim to
598: measure the proper motion of such clusters, and be used in conjunction
599: with expected cluster colors to distinguish the clusters from
600: foreground stars. The expected proper motion of these clusters is
601: $\sim 10^{-3}$ arcsec~yr$^{-1}$.
602:
603:
604: \noindent {\em The Local Group:} Given the constrained distances to M31 and M33,
605: color-magnitude diagrams could be used to identify candidate star
606: clusters around ejected BHs, and spectroscopic follow up could reveal
607: the high dispersions expected from the stars very near the BH. At the
608: distance of M31, a cluster with $r_{\rm h} \sim 0.5\,\pc$ would have
609: an angular size of $\sim 0.1$\,arcsec, and could in principle be
610: resolved by HST\footnote{http://www.stsci.edu/hst/}, JWST\footnote{http://www.stsci.edu/jwst/}, or ground-based telescopes with
611: adaptive optics.
612:
613: {\new
614: \noindent {\em Gas Accretion:} A faint radio glow from the central
615: black hole may also be detectable as it accretes material from
616: stellar winds in the surrounding cluster of stars
617: \citep{2004MNRAS.350..725L}. Rarely, these BHs may accrete
618: gas from the interstellar medium or molecular clouds as they pass
619: through the disk of the galaxy
620: \citep{1998ApJ...495L..85F,2003MNRAS.340..647I,2005ApJ...628..873M,2005MNRAS.358..913V,2006MNRAS.368.1340M}, however, for the number
621: of BHs expected in the MW, such a scenario is unlikely to be observed
622: \citep{2005MNRAS.358..913V}.
623: }
624:
625: \noindent {\em Dark Matter Annihilation:} The recoiled BHs may be surrounded by a
626: dense cusp of dark matter, which was adiabatically compressed by the
627: baryonic condensation and BH growth in its parent low-mass
628: galaxy. These high density cusps can dramatically increase the rate of
629: dark matter annihilation, and may be visible through the relativistic
630: by products of the annihilation
631: \citep{2005PhRvL..95a1301Z,2005PhRvD..72j3517B, 2008arXiv0809.0894S}.
632:
633:
634: \section{Summary and Discussion}
635: \label{sec:disc}
636:
637: Based on a large statistical ensemble of merger tree histories for the
638: MW, we have found that hundreds of GW recoiled BHs should reside
639: within the MW halo today. The BHs have a mass $\msmbh \gtrsim
640: 10^3\,\msun$, and should be surrounded by a cluster of stars that were
641: tightly bound to the BH when it was ejected. The most massive BH
642: weighs $\sim 1-6 \times 10^5\,\msun$, and is surrounded by a compact star
643: cluster $\lesssim 1\,$pc in size with a stellar velocity dispersion of
644: hundreds of km~s$^{-1}$. High-resolution adaptive optics imaging
645: along with spectroscopy of the star clusters can constrain the
646: distance and mass of the BH in the center of the cluster based on
647: orbits of individual stars for the closest clusters \citep{2005ApJ...620..744G,
648: 2005ApJ...628..246E}.
649:
650:
651: The number of recoiled BHs in the MW is most sensitive to the fraction
652: of low mass galaxies that harbored BHs in their centers, as well as to
653: the merger history of such galaxies during the formation of the MW. We
654: emphasize that most of these galactic ``building blocks'' were made at
655: high redshifts and were disrupted during the assembly process of the
656: MW; hence, they likely had different properties than the remaining
657: dwarf satellites of the MW today. The future discovery of the
658: population of ejected star clusters will provide a unique probe of the
659: early history of the Milky Way, as well as the distribution and
660: evolution of low mass BHs. It would also open a new window to
661: exploring the low-mass end of the population of nuclear BHs in
662: high-redshifts galaxies. \citet{2008arXiv0809.5046M}, whose work we
663: learned of after submission of this manuscript, arrived at a similar
664: conclusion independently, with a slightly different application to the
665: nearby Virgo cluster.
666:
667:
668: \section*{Acknowledgements}
669: We thank the GALFORM team for making their merger tree code
670: available. This work is supported in part by NASA grant NNX08AL43G, by
671: FQXi, and by Harvard University funds.
672:
673:
674: \bibliography{p}
675:
676:
677:
678: \end{document}
679:
680:
681: