astro-ph0605732/ms.tex
1: %\documentclass[12pt,preprint]{aastex}
2: \documentclass{emulateapj} % use deluxetable* on wide tables
3: %\usepackage[hypertex]{hyperref}
4: %\usepackage{graphicx}
5: %\usepackage{color,calc}
6: \usepackage{amsmath}
7: %\usepackage{natbib}
8: 
9: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
10: \newcommand{\Fewbody}{{\em Fewbody\/}}
11: %\newcommand{\note}[3]{{\color{red}\bf [#1: #2 -- #3]}}
12: \newcommand{\privcomm}[1]{(#1, private communication)}
13: 
14: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
15: \begin{document}
16: 
17: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
18: \title{Observing IMBH--IMBH Binary Coalescences via Gravitational Radiation}
19: \shorttitle{Observing IMBH--IMBH Binary Coalescences via Gravitational Radiation}
20: \submitted{Accepted for publication in \apjl}
21: \author{John M. Fregeau$^1$, Shane L. Larson$^2$, M. Coleman Miller$^{3,4}$, 
22:   Richard O'Shaughnessy$^1$, Frederic A. Rasio$^1$}
23: \shortauthors{Fregeau, et al.}
24: \affil{$^1$Department of Physics and Astronomy, Northwestern University, Evanston, 
25:   IL 60208; fregeau@alum.mit.edu, oshaughn@northwestern.edu, rasio@northwestern.edu}
26: \affil{$^2$Center for Gravitational Wave Physics, The Pennsylvania State University, 
27:   University Park, PA 16802; shane@gravity.psu.edu}
28: \affil{$^3$University of Maryland, Department of Astronomy, College Park, MD 20742;
29:   miller@astro.umd.edu}
30: \affil{$^4$Goddard Space Flight Center, Greenbelt, MD}
31: 
32: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
33: \begin{abstract}
34: Recent numerical simulations have suggested the possibility of forming double
35: intermediate mass black holes (IMBHs) via the collisional runaway scenario
36: in young dense star clusters.  The two IMBHs formed would exchange into
37: a common binary shortly after their birth, and quickly inspiral and merge.
38: Since space-borne gravitational wave (GW) observatories such as LISA will be able 
39: to see the late phases of their inspiral out to several Gpc, and LIGO will be able
40: to see the merger and ringdown out to similar distances, they represent
41: potentially significant GW sources.  In this Letter we estimate 
42: the rate at which LISA and LIGO will see their inspiral and merger in young star 
43: clusters, and discuss the information that can be extracted from the observations.
44: We find that LISA will likely see tens of IMBH--IMBH inspirals per year, while 
45: advanced LIGO could see $\sim 10$ merger and ringdown events per year,
46: with both rates strongly dependent on the distribution of cluster masses and
47: densities.
48: \end{abstract}
49: 
50: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
51: \keywords{stellar dynamics --- black hole physics --- gravitational waves}
52: 
53: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
54: \section{Introduction}\label{sec:intro}
55: 
56: Observations suggesting the existence of intermediate-mass black holes (IMBHs)
57: have mounted in recent years.  Ultra-luminous X-ray sources (ULXs)---point X-ray
58: sources with inferred luminosities $\gtrsim 10^{39}\,{\rm erg}/{\rm s}$---may
59: be explained by sub-Eddington accretion onto BHs more massive than the maximum
60: of $\sim 10M_\sun$ expected from stellar core collapse \citep{2004IJMPD..13....1M}.
61: Similarly, the cuspy velocity dispersion profiles in the centers of the globular clusters
62: M15 and G1 may also be explained by the dynamical influence of a central IMBH
63: \citep{2002AJ....124.3255V,2002AJ....124.3270G,2005ApJ...634.1093G}, although 
64: this conclusion remains somewhat controversial \citep{2003ApJ...582L..21B}.
65: 
66: The most likely formation scenario for an IMBH is the collapse of a very
67: massive star (VMS), which was formed early in the lifetime of a young star
68: cluster via a runaway sequence of physical collisions of massive 
69: main-sequence stars
70: \citep{1999A&A...348..117P,2001ApJ...562L..19E,2002ApJ...576..899P,2004ApJ...604..632G}.  
71: This scenario has been studied in detail for star clusters without primordial
72: binaries, with recent work showing that runaway 
73: growth of a VMS to $\sim 10^3\,M_\sun$ occurs generically in clusters with deep core collapse 
74: times shorter than the $\sim 3\,{\rm Myr}$ main-sequence lifetime of the most
75: massive stars \citep{2006MNRAS.368..141F}.
76: 
77: Due to the computational cost of simulating the more realistic case of star clusters 
78: with primordial binaries, it is only recently that such simulations have been performed
79: \citep{2004Natur.428..724P,2006ApJ...640L..39G}.  The work of \citet{2006ApJ...640L..39G}
80: was the first to systematically study the influence of primordial binaries on the
81: runaway growth process.  They showed that stellar collisions during binary
82: scattering interactions offer an alternate channel for runaway growth, with the main
83: result that clusters with binary fractions larger than $\approx 10\%$ generically produce 
84: {\em two} VMSs, provided the cluster is sufficiently dense and/or centrally concentrated
85: to trigger the runaway earlier than $\sim 3\,{\rm Myr}$ in the absence of primordial binaries.
86: Observations and recent numerical calculations suggest that star clusters may be
87: born with large binary fractions \citep[$\gtrsim 30\%$;][]{1992PASP..104..981H,2005MNRAS.358..572I},
88: implying that {\em all} sufficiently dense and massive star clusters could form
89: multiple VMSs. 
90: 
91: The VMSs formed will undergo core-collapse supernovae and likely become IMBHs 
92: on a timescale of $\sim 4\,{\rm Myr}$ after cluster formation 
93: \citep[the lifetime of a VMS is extended slightly by collisional rejuvenation; see, e.g.][]{2006MNRAS.368..141F}.  
94: After their separate formation,
95: the two IMBHs will quickly exchange into a common binary via dynamical
96: interactions.  The IMBH--IMBH binary (IMBHB) will then shrink via dynamical friction
97: due to the cluster stars, on a timescale
98: $\sim t_r \langle m \rangle / M_{\rm IMBH} \lesssim 10\,{\rm Myr}$, where $t_r$ is the
99: core relaxation time, $\langle m \rangle$ is the local average stellar
100: mass, and $\langle m \rangle / M_{\rm IMBH} \lesssim 10^{-2}$.
101: Note that since $t_r$ scales inversely with $\langle m \rangle$
102: for fixed core velocity dispersion and mass density, the dynamical
103: friction timescale is independent of $\langle m \rangle$ \citep{1987gady.book.....B}.
104: The IMBHB will then shrink further via dynamical encounters with cluster stars
105: \citep{1996NewA....1...35Q,2003ApJ...599.1129Y,2005ApJ...618..426M}, until it merges quickly via 
106: gravitational radiation, on a timescale 
107: $\approx 1\,{\rm Myr}\,(\sigma_c/20\,{\rm km}\,{\rm s}^{-1})^3(\rho_c/10^5\,M_\sun\,{\rm pc}^{-3})^{-1}(M_{\rm IMBH}/10^3\,M_\sun)^{-1}$,
108: where $\sigma_c$ is the cluster core velocity dispersion and $\rho_c$ is the core mass density
109: \citep[][eqs.~{[29]} and {[30]}]{1996NewA....1...35Q}.
110: This timescale has also been confirmed by numerical scattering calculations 
111: \privcomm{G\"ultekin}.
112: 
113: Only the more massive IMBHBs merge in the LISA band
114: of $10^{-4}$--$1\,{\rm Hz}$ (redshifted binary mass 
115: $M_z \equiv (1+z)M \gtrsim 4 \times 10^3 \, M_\sun$, where $M$ is the 
116: total binary mass).  Fig.~\ref{fig:fs} shows the final
117: gravitational wave (GW) frequency $f_f$ (the frequency at
118: the inner-most stable circular orbit if within the LISA frequency range (large $M_z$), otherwise
119: the maximum LISA frequency of $\approx 1\,{\rm Hz}$ (small $M_z$), 
120: as in \citet{2004ApJ...611.1080W}),
121: and the frequency 1 yr prior, $f_i$, as a function of redshifted
122: mass $M_z$, for the reduced mass parameters $\eta=0.25$ (equal-mass binary)
123: and $\eta=0.1$ (mass ratio $0.13$) \citep[see, e.g.,][]{2004ApJ...611.1080W}.
124: For a wide range in $M_z$, the late stages of inspiral clearly
125: span the LISA ``sweet spot'' (roughly a decade centered on $10^{-2.2}\,{\rm Hz}$),
126: implying that LISA could easily detect the chirp signal, enabling 
127: a measurement of the masses of the binary members.  Such an observation would 
128: be direct evidence for an IMBH.
129: 
130: In Sec.~\ref{sec:lisa} we estimate the rate at which LISA will observe inspiral of
131: IMBHBs in young star clusters.  In Sec.~\ref{sec:ligo} we estimate the rate at which 
132: LIGO will observe their merger and ringdown.  Finally, in Sec.~\ref{sec:discussion} 
133: we discuss the observational consequences.
134: 
135: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
136: \begin{figure}
137:   \begin{center}
138:     \includegraphics[width=\columnwidth]{f1.eps}
139:     \caption{The final GW frequency $f_f$ (see text), and the frequency 
140:       1 yr prior, $f_i$, for an IMBHB with
141:       total mass $M$ and reduced mass parameter $\eta$,
142:       as a function of redshifted binary mass $M_z$, for
143:       $\eta=0.25$ (equal-mass binary) and $\eta=0.1$ (mass ratio $0.13$).
144:       (The final frequency is roughly independent of $\eta$.)
145:       \label{fig:fs}}
146:   \end{center}
147: \end{figure}
148: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
149: 
150: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
151: \begin{figure}
152:   \begin{center}
153:     \includegraphics[width=\columnwidth]{f2.eps}
154:     \caption{Luminosity distance, $d_L(z)$, to which an IMBHB of
155:       total mass $M$ and reduced mass parameter $\eta$ can be seen
156:       via its inspiral with LISA with $S/N=10$ for a 1 yr integration,
157:       and via its merger and ringdown
158:       with $S/N=8$ for iLIGO and AdLIGO,  as a function
159:       of the redshifted mass $M_z$.  The corresponding redshift
160:       (calculated using the WMAP year 3 cosmological parameters,
161:       as discussed in the text) is shown on the right vertical axis.
162:       \label{fig:dlvsm}}
163:   \end{center}
164: \end{figure}
165: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
166: 
167: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
168: \section{Estimating the LISA detection rate}\label{sec:lisa}
169: 
170: We first need to know the distance to which LISA can see IMBHB inspirals.  
171: Following the techniques in \citet{2004ApJ...611.1080W} and 
172: \citet{1998PhRvD..57.4535F}, we adopt the latest LISA sensitivity curve \citep{SCG},
173: including confusion noise from Galactic white dwarf binaries \citep{benderhils1997},
174: and calculate the maximum luminosity distance, $d_L(z)$, to which an IMBHB
175: of total mass $M$ and reduced mass parameter $\eta$ can be
176: seen with $S/N=10$ for a 1 yr integration.
177: The results are shown in Fig.~\ref{fig:dlvsm} as a function of $M_z$, for $\eta=0.25$ and $\eta=0.1$.
178: Note that the results of \citet{2006ApJ...640L..39G} show that the masses of the 
179: IMBHs never differ by more than a factor of a few ($\eta \gtrsim 0.15$).  
180: Thus LISA will be able to see typical IMBHBs ($M \sim 10^3\,M_\sun$) out to a few Gpc.
181: 
182: With this information in hand, we first make a crude
183: estimate of the LISA event rate.  Following \citet{2002ApJ...581..438M}, 
184: we write for the total rate
185: \begin{equation}\label{eq:a}
186:   R \equiv
187:   \frac{dN_{\rm event}}{dt} = 
188:   \left(\int_0^{z_{\rm max}} \frac{dV_c}{dz} dz \right)
189:   \frac{dN_{\rm cl}}{dV}
190:   g
191:   \frac{1}{t_U} \, .
192: \end{equation}
193: The first factor, $\int_0^{z_{\rm max}} (dV_c/dz) dz$, is the integrated comoving
194: volume of space in which LISA is sensitive to the events.  The second factor, $dN_{\rm cl}/dV$,
195: is the number density of star clusters sufficiently massive
196: to form IMBHBs.  Since the {\em globular} clusters we currently
197: see were likely at least a few times more massive at formation \citep{2001ApJ...550..691J},
198: we set this factor to the current density of globular clusters in the local
199: universe, $dN_{\rm cl}/dV \approx 8 h^3\,{\rm Mpc}^{-3}$ \citep{2000ApJ...528L..17P}.  
200: The third factor, $g$, is the fraction of sufficiently massive clusters that have a large
201: enough binary fraction and initial central density to produce IMBHBs.  
202: Since initial cluster structural parameters are largely unknown, we treat $g$ as a parameter.
203: The fourth factor is the event rate per IMBHB-producing cluster, taken to be 
204: one divided by the age of the universe, since only one IMBHB
205: is formed per cluster over its lifetime.  We adopt a $\Lambda$CDM cosmology,
206: with parameters $\Omega_M=0.24$, $\Omega_\Lambda=0.76$, and $h=0.73$, for
207: which $t_U=13.8\,{\rm Gyr}$ \citep{astro-ph/0603449}.  Putting this together for 
208: $d_L=4.9\,{\rm Gpc}$ ($z_{\rm max}=0.79$), the distance to which LISA can see
209: IMBHBs with $M=2\times 10^3\,M_\sun$, eq.~(\ref{eq:a}) gives 
210: $R \approx 1(g/0.1)\, {\rm yr}^{-1}$.
211: 
212: Writing down a generalized form of the rate integral in eq.~(\ref{eq:a}) is 
213: straight-forward.  Since the time between cluster formation and IMBHB merger 
214: is $\ll t_U$, we assume that the merger is coincident with cluster formation.  
215: Thus the rate integral is
216: \begin{multline}\label{eq:b}
217:   R \equiv
218:   \frac{dN_{\rm event}}{dt_o} = 
219:   \int_0^{z_{\rm max}}
220:   \frac{d^2M_{\rm SF}}{dV_c dt_e}
221:   g_{\rm cl}
222:   g\\
223:   \times \frac{dt_e}{dt_o} 
224:   \frac{dV_c}{dz}
225:   \int_{M_{\rm cl,min}(z)}^{M_{\rm cl,max}}
226:   \frac{d^2N_{\rm cl}}{dM_{\rm SF,cl} dM_{\rm cl}}
227:   \, dM_{\rm cl} \, dz \, .
228: \end{multline}
229: Here $R \equiv dN_{\rm event}/dt_o$ is the event rate observed at $z=0$
230: by LISA, $d^2M_{\rm SF}/dV_c dt_e$ is the star formation rate (SFR) in 
231: mass per unit of comoving volume per unit of local time, 
232: $g_{\rm cl}$ is the fraction of star forming mass that goes
233: into star clusters more massive than $10^{3.5}\,M_\sun$ 
234: (generally a function of $z$), $g$ is as above, and
235: $d^2N_{\rm cl}/dM_{\rm SF,cl} dM_{\rm cl}$ is the distribution function 
236: of clusters over individual cluster mass $M_{\rm cl}$ and
237: total star forming mass in clusters $M_{\rm SF,cl}$.  Finally, $dt_e/dt_o$ is simply
238: $(1+z)^{-1}$, and $dV_c/dz$ is the rate of change of comoving volume
239: with redshift, which is a function of cosmological parameters \citep{astro-ph/9905116}.
240: Note that we set $z_{\rm max}=5$, since this is roughly the limit to which the
241: cosmic SFR can be traced.  Thus the integral in eq.~(\ref{eq:b})
242: should be considered a mild lower limit to the true rate.  We now discuss each 
243: element in eq.~(\ref{eq:b}) in more detail.
244: 
245: Following \citet{2001ApJ...548..522P}, we adopt three different choices
246: for the SFR:
247: \begin{equation}\label{eq:sfr}
248:   \left(\frac{d^2M}{dV_c dt}\right)_{{\rm SF}i} = C_i h_{65} F(z) G_i(z) M_\sun\,{\rm yr}^{-1}\,{\rm Mpc}^{-3} \, ,
249: \end{equation}
250: with $i=1,2,3$ denoting the different rates, 
251: $C_i$ a constant, $G_i(z)$ a function of $z$, $h_{\rm 65}=h/0.65$, and 
252: $F(z)=[\Omega_M(1+z)^3+\Omega_k(1+z)^2+\Omega_\Lambda]^{1/2}/(1+z)^{3/2}$.
253: The first is from \citet{2000MNRAS.312L...9M}, with 
254: $C_1=0.3$ and $G_1(z)=e^{3.4z}/(e^{3.8z}+45)$, which peaks between $z=1$ and 2 
255: and decreases at larger redshift.  The second is from \citet{1999ApJ...519....1S}, with
256: $C_2=0.15$ and $G_2(z)=e^{3.4z}/(e^{3.4z}+22)$, which is roughly constant for 
257: $z \gtrsim 2$.  The third is from \citet{1999ApJ...512L..87B}, with
258: $C_3=0.2$ and $G_3(z)=e^{3.05z-0.4}/(e^{2.93z}+15)$, 
259: which increases above $z \approx 2$.
260: 
261: Measuring the fraction of star-forming mass in clusters is 
262: difficult for anywhere but the local universe.
263: Similarly, while we know reasonably well the initial cluster conditions required 
264: to form an IMBHB \citep{2006ApJ...640L..39G},
265: we know much less well the distribution of cluster properties at birth.
266: We therefore treat $g_{\rm cl}$ and $g$ as parameters, taking $g_{\rm cl} = 0.1$ and
267: $g=0.1$ somewhat arbitrarily as canonical values.
268: 
269: Assuming that the spectrum of cluster masses is neither a function of cosmic
270: epoch nor the total star forming mass available for clusters, the factor
271: $d^2N_{\rm cl}/dM_{\rm SF,cl} dM_{\rm cl}$ can be separated as
272: \begin{equation}
273:   \frac{d^2N_{\rm cl}}{dM_{\rm SF,cl} dM_{\rm cl}} = 
274:   \frac{f(M_{\rm cl})}{\int M_{\rm cl} f(M_{\rm cl}) dM_{\rm cl}} \, ,
275: \end{equation}
276: where $f(M_{\rm cl})$ is the (normalized) distribution function of cluster masses.  For this
277: we adopt the power-law form observed for young star clusters in the Antennae, which is 
278: thought to be universal: $f(M_{\rm cl}) \propto M_{\rm cl}^{-2}$ \citep{1999ApJ...527L..81Z},
279: with a lower limit of $10^{3.5}\,M_\sun$ and an upper limit of $10^7\,M_\sun$.
280: 
281: It is the limit $M_{\rm cl,min}(z)$ in eq.~(\ref{eq:b}) that 
282: encodes all information about the detectability of an IMBHB inspiral by 
283: LISA.  Specifically, the redshift to which LISA can see the inspiral is a function
284: of the binary mass, which is itself a function of the host cluster mass.  Adopting
285: an efficiency factor $f_{\rm GC}$ for the fraction of
286: cluster mass going into the IMBHB, this
287: relationship is inverted to obtain $M_{\rm cl,min}(z)$.  
288: Recent numerical work shows that the efficiency factor is 
289: $f_{\rm GC}\approx 2\times 10^{-3}$, independent of cluster initial 
290: conditions \citep{2004ApJ...604..632G}, which we take as our canonical value.  
291: At low redshift, $M_{\rm cl,min}(z)$ is clamped at the value $M_{\rm cl}=200\,M_\sun/f_{\rm GC}$, 
292: set by adopting the definition that an IMBH have mass $\geq 10^2\,M_\sun$.
293: At high redshift ($z > 5$, so not relevant to our calculation), 
294: $M_{\rm cl,min}(z)$ is clamped at the value of $10^7\,M_\sun$ from the cluster
295: mass function; in other words no cluster is sufficiently massive to produce an
296: IMBHB massive enough to be observable by LISA, so the integral is zero.
297: 
298: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
299: \begin{figure}
300:   \begin{center}
301:     \includegraphics[width=\columnwidth]{f3.eps}
302:     \caption{Integrand of the rate integral in eq.~(\ref{eq:b}) for
303:       the three different SFRs in eqs.~(\ref{eq:sfr}), 
304:       for $\eta=0.25$ and $\eta=0.1$.
305:       \label{fig:dRdz}}
306:   \end{center}
307: \end{figure}
308: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
309: 
310: We numerically integrated eq.~(\ref{eq:b}) for the different SFRs in eqs.~(\ref{eq:sfr}),
311: for $S/N=10$ and an integration time of 1 yr, to find that the rate is 
312: \begin{equation}\label{eq:rate1}
313:   R(\eta=0.25)\approx 40\mbox{--}50 
314:   \left(\frac{g_{\rm cl}}{0.1}\right) \left(\frac{g}{0.1}\right) 
315:        {\rm yr}^{-1} \, ,
316: \end{equation}
317: with the spread in the coefficient from the different
318: SFRs.  The coefficient decreases to $20\mbox{--}25$ for 
319: $\eta=0.1$.  The rate is dominated by clusters in the mass range 
320: $10^6$--$10^{6.5}\,M_\sun$ (IMBHB mass $2 \times 10^3$--$6\times 10^3\,M_\sun$), 
321: with more than half the contribution to the rate coming 
322: from this mass range, for both $\eta=0.1$ and $\eta=0.25$, and for all 
323: three SFRs in eq.~(\ref{eq:sfr}) (except SF3 for $\eta=0.1$).
324: Note that eq.~(\ref{eq:b}) is only strictly valid when the source is 
325: visible by the instrument for less than the integration
326: time.  This turns out not to be precisely correct.  A typical IMBHB
327: with mass $M=f_{\rm GC} 10^{6.25}\,M_\sun$ takes roughly 4 years to cross
328: the LISA band from the edge of the white dwarf confusion knee at $\approx 2\,{\rm mHz}$ 
329: to the upper edge of the band at $\approx 1\,{\rm Hz}$.  Thus the rate presented in
330: eq.~(\ref{eq:rate1}) is an underestimate by of order a factor of a few.
331: 
332: Fig.~\ref{fig:dRdz} shows
333: the integrand of the rate integral in eq.~(\ref{eq:b}) for
334: the three different SFRs in eqs.~(\ref{eq:sfr}), 
335: for $\eta=0.25$ and $\eta=0.1$.  Most events originate
336: from $z \sim 1$.  Unfortunately, neither $R$ nor $dR/dz$ is particularly
337: sensitive to the cosmic SFR, with $dR/dz$ decreasing
338: quickly above $z \approx 2$ even when the SFR is increasing
339: (as in SF3).  Thus observations of IMBHB inspirals will not be very
340: informative about the cosmic SFR.
341: However, they will likely yield a handle on the fraction of star formation 
342: that is in compact massive clusters.
343: 
344: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
345: \section{Estimating the LIGO detection rate}\label{sec:ligo}
346: 
347: Shortly after the two IMBHs merge, the merger product can be well
348: described as a single perturbed black hole, emitting GWs at its quasinormal frequencies.
349: Largely falling within the initial and advanced LIGO (iLIGO and AdLIGO) sensitivity
350: bands, the merger and ringdown waves will likely carry a few percent of the rest mass
351: energy of the hole \citep[see, e.g.][]{1998PhRvD..57.4535F}.  
352: Numerical simulations suggest that a merging pair of nonspinning equal-mass
353: black holes will emit a fraction $\epsilon \simeq 0.03$ of their rest mass in merger and
354: ringdown GWs, forming a black hole with spin parameter $a \simeq 0.7$ 
355: \citep{Lazarus,NR:Brownsville,gr-qc/0602026}.  Under these conditions, the
356: ringdown frequency is given by \citep[see {eq.~[3.17]} of][]{1998PhRvD..57.4535F}
357: \begin{equation}
358:   f \approx \frac{c^3}{2 \pi G M_z} (1-0.63(1-a)^{3/10})
359:   \approx 180 \left(\frac{M_z}{10^{2}M_\odot}\right)^{-1} {\rm Hz} \, .
360: \end{equation}
361: We can express the distance to which we are
362: sensitive to ringdown waves at signal-to-noise ratio $\rho$ as
363: \begin{equation}
364: d_L(z) = \left(\frac{2 \epsilon M_z}{5 \pi^2\rho^2 f^2 S(f)}\right)^{1/2} \, ,
365: \end{equation}
366: where $S(f)$ is the spectral noise density of LIGO.
367: Combining this expression with the concordance cosmological model and
368: iLIGO and AdLIGO sensitivity curves, we find the range to
369: which LIGO can detect ringdown shown in Fig.~\ref{fig:dlvsm}.
370: 
371: To obtain a conservative estimate for the rate at which iLIGO and AdLIGO
372: could detect these mergers with a ringdown-only search,
373: %\footnote{While
374: %  one can conjoin an inspiral search with the ringdown search, we gain
375: %  no additional range for IMBH-size masses.  Because of the different
376: %  characteristic frequencies of the two waveforms and the fairly
377: %  narrow LIGO peak sensitivity band, the two waveforms probe different
378: %  mass ranges.  See, e.g., \citet{1998PhRvD..57.4535F}.
379: %}
380: we use eq.~(\ref{eq:a}) with a
381: moderately optimistic range of $d_L \approx 100\,{\rm Mpc}$ for iLIGO and 
382: $d_L = 2\,{\rm Gpc}$ for AdLIGO.  The expected detection rate is then
383: $10^{-4} (g/0.1)\,{\rm yr}^{-1}$ and $1 (g/0.1)\,{\rm yr}^{-1}$, 
384: for iLIGO and AdLIGO, respectively.  More detailed estimates using 
385: machinery analogous to eq.~(\ref{eq:b}) increase these estimates by roughly 
386: an order of magnitude, making the rate for AdLIGO 
387: $10 (g_{\rm cl}/0.1) (g/0.1)\,{\rm yr}^{-1}$.
388: 
389: %However, because the range is \emph{not monotonic}
390: %with mass, for redshifts below a critical redshift, a \emph{band} of
391: %masses is allowed, running from  very low
392: %mass IMBHs (visible because they occur at great frequency) or
393: %very high mass IMBHs (visible because they can be seen to great
394: %range).  Thus, the rightmost portion of equation \ref{eq:b} must be
395: %generalized to
396: %\begin{equation}
397: %  \int_{M_{\rm cl,-}}^{M_{\rm cl,+}}
398: %  \frac{d^2N_{\rm cl}}{dM_{\rm SF,cl} dM_{\rm cl}}
399: %  \, dM_{\rm cl}  .
400: %\end{equation}
401: %where $M_{cl,\pm}(z)$ arises from inverting
402: %\ref{fig:dlvsm}.   Performing a detailed calculation, we conclude
403: % AdLIGO could easily see several ringdowns a year from the most
404: % low-mass IMBH binaries:
405: %\begin{equation}
406: %R_{L2,\text{ring}}=10 \text{yr}^{-1}(g_{cand}/0.1)(g_{cl}/0.1) \; .
407: %\end{equation}
408: 
409: %\begin{figure}
410: %  \begin{center}
411: %    \includegraphics[width=\columnwidth]{figIntegrand_clean.epsi}
412: %    \caption{\label{fig:rate-advLIGO-ringdown} Integrand
413: %      of the rate integral in eq.~(\ref{eq:b}) for ringdown searches, for
414: %      the star formation history SF1.
415: %    }
416: %  \end{center}
417: %\end{figure}
418: 
419: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
420: \section{Discussion}\label{sec:discussion}
421: 
422: It appears likely that LISA will see tens of IMBHB inspiral events per year,
423: while AdLIGO could see $\sim 10$ merger and ringdown events per year,
424: with both rates strongly dependent on the distribution of cluster masses and
425: densities.  Detection of an IMBHB would have profound implications.  A match-filtered
426: observation of the inspiral would yield the redshifted masses of the black
427: holes, directly confirming the existence of IMBHs.  It would also yield the
428: luminosity distance to the source; with enough observations, constraints could
429: be placed on the cosmic history of star formation in dense, massive clusters.
430: Detection of the ringdown signal from the merger product will additionally 
431: yield its spin, which may provide insight into its formation history.
432: 
433: Typical IMBHBs spend $\gtrsim 10^6\,{\rm yr}$ inspiraling
434: through the LISA band, with nearly all of that time spent at low frequencies 
435: ($\lesssim 10^{-3}\,{\rm Hz}$).  In the low frequency
436: region they will thus appear as a large number of monochromatic
437: sources, possibly contributing to confusion noise and increasing the noise 
438: floor \citep[e.g.,][]{2003MNRAS.346.1197F}.  A detailed calculation of this
439: is beyond the scope of this Letter.  However,
440: we note that if their contribution is similar in magnitude to that of Galactic
441: compact object binaries \citep{benderhils1997}, the rates predicted in
442: eq.~(\ref{eq:rate1}) would decrease by about 20\%.
443: 
444: %% % Cole's text from his email:
445: 
446: %% Although our analysis has been primarily concerned with the inspiral and merger of
447: %% IMBHBs while they are still in their host clusters, it is
448: %% useful to consider the evolution of an IMBHB in a cluster
449: %% near the Galactic center.  In this case it is possible that the host
450: %% cluster will shrink in its orbit due to dynamical friction, to the point
451: %% that the IMBHB is ``deposited'' into a close orbit about the central
452: %% supermassive black hole.  Two questions naturally arise: (1) how likely
453: %% is this to occur, and (2) will the orbit of an
454: %% IMBH-IMBH binary around an SMBH be an interesting LISA source?
455: %% Answering the first question requires a level of analysis that is beyond
456: %% this Letter.  The second, however, is more straight-forward.
457: 
458: %% For a binary IMBH to survive in orbit around an SMBH requires that
459: %% the tidal acceleration from the SMBH be smaller than the internal
460: %% acceleration of the binary IMBH.  In analogy with the condition for
461: %% Roche lobe overflow (or Hill spheres for planetary applications), this
462: %% implies that the orbital frequency of the binary center of
463: %% mass about the SMBH must be less than the orbital frequency of the
464: %% two IMBHs about each other.  Thus a given 
465: %% frequency for the orbit around the SMBH yields a lower limit on
466: %% the binary IMBH frequency, and hence an upper limit to how long it will
467: %% take for the IMBHs to spiral into each other.
468: 
469: %% If we write the IMBH masses as $M_1$ and $M_2$ and define the
470: %% total mass $M=M_1+M_2$, the reduced mass $\mu=M_1M_2/M$, and the
471: %% reduced mass parameter $\eta=\mu/M$ (which has a maximum of 1/4), then
472: %% the inspiral time for a circular orbit (which maximizes the
473: %% inspiral time for a fixed orbital frequency) is
474: %% \begin{equation}
475: %%   \tau_{\rm inspiral}\approx 3\times 10^8~{\rm yr}
476: %%   \left(\frac{10^9~M_\odot^3}{\eta M^3}\right)\left(\frac{a}{1~{\rm AU}}\right)^4 \, .
477: %% \end{equation}
478: %% Since the GW frequency is 
479: %% $f_{\rm GW}=2f_{\rm orb}=\sqrt{GM/a^3\pi^2}$, this implies
480: %% \begin{equation}
481: %% a=0.01\,{\rm AU}\left(\frac{M}{10^3~M_\sun}\right)^{1/3}
482: %% \left(\frac{f_{\rm GW}}{2\,{\rm mHz}}\right)^{-2/3} \, ,
483: %% \end{equation}
484: %% so
485: %% \begin{equation}
486: %% \tau_{\rm inspiral} \approx 3\,{\rm yr}\,\eta^{-1}\left(\frac{M}{10^3~M_\odot}\right)^{-5/3}
487: %% \left(\frac{f_{\rm GW}}{2~{\rm mHz}}\right)^{-8/3} \, .
488: %% \end{equation}
489: %% The quantity $f_{\rm GW}$ is scaled by 2~mHz to get to the lower end of LISA's sweet spot, 
490: %% but it is clear that for $\eta\sim 0.25$, even at $f_{\rm GW}=10^{-4}$~Hz
491: %% the duration is only a few tens of thousands of years.  It will take much
492: %% longer than that to sink the binary IMBH towards the center.
493: %% We thus expect that if a binary IMBH sinks towards an SMBH, it will
494: %% either be tidally separated or will merge with itself rather than 
495: %% undergoing multiple orbits in the LISA frequency band.
496: 
497: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
498: \acknowledgments
499: 
500: We thank Kayhan G\"ultekin for providing the numerically-calculated timescale for an IMBHB to shrink
501: by stellar dynamical encounters after formation.
502: We also thank the anonymous referee for many helpful comments on the manuscript.
503: JMF and FAR acknowledge support from NASA grant NNG04G176G and NSF grant PHY-0245028.
504: SLL acknowledges support from NASA award NNG05GF71G.
505: MCM was supported in part by NASA grant NAG 5-13229 and by the Research Associateships
506: Programs Office of the National Research Council and Oak Ridge Associated Universities.
507: ROS acknowledges support from NSF grants PHYS-0121416 and PHYS-0353111.
508: 
509: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
510: \bibliographystyle{apj}
511: \bibliography{apj-jour,main}
512: 
513: \end{document}
514: