1: %version 0: 11/April, 2006 JMW (in IHEP)
2: %version 1: 28/April, 2006 JMW
3:
4: \documentclass{aastex}
5: \usepackage{emulateapj5}
6: \usepackage{apjfonts}
7: \usepackage{epsfig}
8: %\usepackage{lscape}
9:
10:
11: \slugcomment{Received 2006 May 15; accepted 2006 June 28 }
12: \journalinfo{The Astrophysical Journal Letters in Press}
13:
14: \shorttitle{Duty Cycle of Quasars}
15: \shortauthors{Wang, Chen \& Zhang}
16:
17:
18: \def\calcb{{\cal C}_{\rm B}}
19: \def\caln{{\cal N}(\mbh,z)}
20: \def\pdotm{{\langle \dot{M}(\mbh,t)\rangle}}
21: \def\kms{\ifmmode {\rm km~ s^{-1}} \else {\rm km~s^{-1}}\ \fi}
22: \def\dotmbh{\dot{M}_{\bullet}}
23: \def\mbh{M_{\bullet}}
24: \def\mmbh{m_{\bullet}}
25: \def\mbhc{M_{\bullet}^{\rm c}}
26: \def\mgii{\ifmmode Mg {\sc ii} \else Mg {\sc ii}\ \fi}
27: \def\oiii{\ifmmode O {\sc iii} \else O {\sc iii}\ \fi}
28: \def\feii{\ifmmode Fe {\sc ii} \else Fe {\sc ii}\ \fi}
29: \def\rhobh{\rho_{\bullet}}
30: \def\rmd{{\rm d}}
31: \def\rg{R_{\rm G}}
32: \def\sunm{M_{\odot}}
33: \def\tq{t_{\rm qso}}
34: \def\tp{t^{\prime}}
35: \def\tpi{t_i^{\prime}}
36:
37: \def\lax{{$\mathrel{\hbox{\rlap{\hbox{\lower4pt\hbox{$\sim$}}}\hbox{$<$}}}$}}
38: \def\gax{{$\mathrel{\hbox{\rlap{\hbox{\lower4pt\hbox{$\sim$}}}\hbox{$>$}}}$}}
39:
40: \begin{document}
41:
42: \title{Cosmological Evolution of the Duty Cycle of Quasars}
43:
44: \author{Jian-Min Wang\altaffilmark{1}, Yan-Mei Chen\altaffilmark{1,2} and Fan Zhang\altaffilmark{1}}
45:
46: \altaffiltext{1}{Key Laboratory for Particle Astrophysics, Institute of High Energy Physics,
47: Chinese Academy of Sciences, 19B Yuquan Road, Beijing 100049, China}
48:
49: \altaffiltext{2}{Graduate School, Chinese Academy of Science, 19A Yuquan Road, Beijing 100049, China}
50:
51:
52: \begin{abstract}
53: Quasars are powered by accretion onto supermassive black holes,
54: but the problem of the duty cycle related to the episodic activity of the black holes remains open as one of
55: the major questions of cosmological evolution of quasars. In this Letter, we obtain quasar
56: duty cycles based on analyses of a large sample composed of 10,979 quasars with redshifts $z\le2.1$
57: from the Sloan Digital Sky Survey (SDSS) Data Release Three. We estimate masses of quasar black holes
58: and obtain their mass function (MF) of the present sample. We then get the duty cycle
59: $\bar{\delta}(z)=10^{-3}\sim 1$ based on the So\l tan's argument, implying that black holes are undergoing
60: multiple episodic activity. We find that the duty cycle has a strong evolution.
61: By comparison, we show that evolution of the duty cycle follows the history of cosmic star
62: formation rate (SFR) density in the Universe, providing intriguing evidence for a natural
63: connection between star formation and triggering of black hole activity.
64: Feedback on star formation from black hole activity is briefly discussed.
65: \end{abstract}
66: \keywords{black hole physics --- galaxies: active ---
67: galaxies: evolution --- galaxies: nuclei --- quasar: general}
68:
69: \section{Introduction}
70: Supermassive black holes are relics of quasars in the Universe (So\l tan 1982; Rees 1984, 1990).
71: Evolution of quasars is led by switching on and off accretion onto the black holes.
72: During their entire evolution, how many times and how many black holes are triggered? what is
73: the triggering mechanism? and why do quasars switch off?
74:
75: The duty cycle, defined as the fraction of active black holes to their total number
76: is a key to tackling the above problems (Richstone et al. 1998; Martini 2004).
77: A popular method to get the duty cycle invokes the continuity equation and the MF of quasar
78: black holes. With an assumption that quasar black holes have the same Eddington ratio,
79: their MF can be obtained from the luminosity function and finally the duty cycle can be found
80: from the continuity equation (Small \& Blandford 1990; Marconi et al. 2004).
81: This is a convenient way to discuss evolution of the black holes, but
82: the degeneracy of the Eddington ratio and the duty cycle still holds.
83: Actually, not only are the Eddington ratios {\em not} constant for different quasars
84: at different epochs, but they appear to be quite scattered (Vestergaard 2004). %On the other hand,
85: The duty cycle is poorly understood as a statistical parameter tracing the evolution of quasar
86: populations.
87:
88: In recent years, there has been much progress in estimating black hole masses both in nearby
89: galaxies and distant quasars. The empirical relation of reverberation mapping allows us to conveniently
90: obtain the black hole masses from a large sample and directly get their MF. Thus it becomes
91: realistic to get new clues to understand the evolution of quasars. Fortunately,
92: by invoking the MF, we can decouple the degeneracy of the duty cycle and
93: the Eddington ratio to get the duty cycle.
94:
95: In this Letter, we use available SDSS data to directly get the MF of
96: the black holes so as to discuss the duty cycle based on the MF and find
97: that there is a strong cosmological evolution. Our calculations assume a cosmology with the
98: Hubble constant $H_0=70~{\rm Mpc^{-1}~km~s^{-1}}$, $\Omega_{\rm m}=0.3$, and $\Omega_{\Lambda}=0.7$.
99:
100: \section{Black Hole Evolution}
101: Active black hole evolution can be described by the MF $\Phi(\mbh,z)$, which is defined as
102: $\Phi(\mbh,z)=\rmd^2 N/\rmd\mbh\rmd V,$ where $\mbh$ is the black hole
103: mass, $\rmd N$ is the number of quasars within the comoving volume element
104: $\rmd V$ and mass interval $\rmd \mbh$. The number density of quasar black holes is then given by
105: $N_{\rm qso}(z)=\int_{\mbh^*}\Phi(\mbh,z)\rmd \mbh$ at redshift $z$, where $\mbh^*$ is the flux-limited
106: black hole mass in a survey. If $\caln$ is the MF of
107: all black holes including the active and the inactive ones, the total number density of black holes is
108: $N_{\rm tot}(z)=\int_{\mbh^*}\caln\rmd\mbh$ at redshift $z$. The relation between $\caln$ and
109: $\Phi(\mbh,z)$ can be written as (Small \& Blandford 1992; Marconi et al. 2004),
110: %
111: \begin{equation}
112: \Phi(\mbh,z)=\delta(\mbh,z)\caln.
113: \end{equation}
114: %
115: where the duty cycle, $\delta(\mbh,z)$ is a function of black hole mass and redshift. Integrating
116: equation (1) over all black hole masses, we have a mean duty cycle in terms of their number density
117: %
118: \begin{equation}
119: {\bar{\delta}}_1(z)=\frac{\int_{\mbh^*}\Phi(\mbh,z)\rmd \mbh}{\int_{\mbh^*}\caln\rmd \mbh}=
120: \frac{N_{\rm qso}(z)}{N_{\rm tot}(z)}.
121: \end{equation}
122: %
123: This duty cycle represents the relative number of the active black holes
124: to the total. Multiplying by $\mbh$ and integrating Eq. (1) over all black hole masses, we have an
125: averaged duty cycle weighted by the masses of the black holes
126: %
127: \begin{equation}
128: {\bar \delta}_2(z)=\frac{\int_{\mbh^*}\Phi(\mbh,z)\mbh\rmd\mbh}{\int_{\mbh^*}\caln\mbh\rmd \mbh}
129: =\frac{\langle\mbh(z)\rangle_{\rm qso}}{\langle\mbh(z)\rangle_{\rm all}}
130: {\bar{\delta}}_1(z),
131: \end{equation}
132: %
133: where $\langle{\mbh(z)}\rangle_{\rm qso}=\int_{\mbh^*}\Phi(\mbh,z)\mbh\rmd \mbh/N_{\rm qso}$
134: is the averaged mass of the quasar black holes at redshift $z$ and
135: $\langle{\mbh(z)}\rangle_{\rm all}=\int_{\mbh^*}\caln\mbh\rmd \mbh/N_{\rm tot}$ for all of active
136: and inactive black holes. We show below $\langle{\mbh(z)}\rangle_{\rm qso}=\langle{\mbh(z)}\rangle_{\rm all}$
137: for $\mbh>\mbh^*$.
138: %\gg\langle\mbh^{\rm S}\rangle$, where $\langle\mbh^{\rm S}\rangle$ is the mean mass of seed black holes.
139: So we have $\bar{\delta}_1(z)=\bar{\delta}_2(z)$.
140:
141: %The continuity equation of the black hole population reads (Small \& Blandford 1992)
142: %
143: %\begin{equation}
144: %\frac{\partial {\cal N}(\mbh,t)}{\partial t}+
145: %\frac{\partial}{\partial \mbh}\left[\langle \dotmbh(\mbh,t)\rangle {\cal N}(\mbh,t)\right]=0,
146: %\end{equation}
147: %where $\langle \dotmbh\rangle$ is the averaged accretion rates for all black holes with $\mbh$.
148: %Multiplying $\mbh$ and integrating eq (4), we have
149: %
150: %\begin{equation}
151: %\langle \mbh\rangle_{\rm all} N_{\rm tot}=\int_{t_0}^t\rmd t\int \langle \dotmbh(\mbh,t)\rangle
152: % {\cal N}(\mbh,t)\rmd \mbh,
153: %\end{equation}
154: %where we take
155: %$\int\rmd \mbh \langle\dotmbh (\mbh,t)\rangle{\cal N}(\mbh,t=0)\ll \langle \mbh\rangle_{\rm all} N_{\rm tot}$
156: %and $\langle \dotmbh(\mbh,t)\rangle{\cal N}(\mbh,t)=0$ at the two side of the MF.
157: %Considering the first integral of right side of eq. (5), it can be rewritten as
158: %$\int\langle\dotmbh(\mbh,t)\rangle {\cal N}(\mbh,t)\rmd \mbh=N_{\rm tot}{\dot{\cal{M}}}_{\bullet}(t)$,
159: %where ${\dot{\cal{M}}}_{\bullet}(t)={\bar{\delta}}_1(z)\dotmbh$ is the mean accretion rates of all
160: %black holes. Therefore the right side of eq. (5) can be rewritten by
161: %$\int_{t_0}^tN_{\rm tot}\bar{\delta}_1(z)\dotmbh(\mbh,t)\rmd t^{\prime}=
162: %\int_{t_0}^{t_{\rm acc}}N_{\rm tot}\dotmbh(\mbh,t)\rmd t=\langle\mbh\rangle_{\rm qso}N_{\rm tot}$,
163: %where we use $\bar{\delta}_1(z)\rmd t^{\prime}=\rmd t$, $\bar{\delta}_1(z)t^{\prime}=t_{\rm acc}$
164: %and $t_{\rm acc}$ is accretion time (Small \& Blandford 1992).
165: %We also take the initial mass is much less than that at time $t$.
166: %Finally equation (5) produces $\langle\mbh\rangle_{\rm all}=\langle\mbh\rangle_{\rm qso}$.
167:
168: The averaged mass of all black holes is
169: $\langle\mbh(t)\rangle_{\rm all}=\langle\mbh^{\rm S}\rangle+\int_0^t\langle\dotmbh(\tp)\rangle_{\rm all}
170: \rmd t^{\prime}$, where $\langle\mbh^{\rm S}\rangle$ is the mean mass of seed black holes and
171: $\langle\dotmbh(\tp)\rangle_{\rm all}$ is the averaged accretion rate of all
172: black holes larger than $\mbh^*$. Considering that
173: $\langle\dotmbh(\tp)\rangle_{\rm all}=\bar{\delta}_1(\tp)\langle\dotmbh(\tp)\rangle_{\rm qso}$
174: [see Eq. (9) in Small \& Blandford 1992; and the first equation in sentence with Eq. (14) in
175: Marconi et al. 2004], we have
176: $\langle\mbh(t)\rangle_{\rm all}=\langle\mbh^{\rm S}\rangle+
177: \int_0^t\bar{\delta}_1(\tp)\langle\dotmbh(\tp)\rangle_{\rm qso} \rmd \tp$. On the other hand,
178: the duty cycle can be written as $\bar{\delta}_1(t)=\Delta t_{\rm act}/\Delta t$
179: for a single episodic phase within a time interval $\Delta t$, where
180: $\Delta t=\Delta t_{\rm act}+\Delta t_{\rm dor}$; $\Delta t_{\rm act}$ and $\Delta t_{\rm dor}$ are
181: the active and the dormant times, respectively. Then, for the $i-$th episodic phase within $\Delta \tpi$,
182: the accreted mass is given by $\langle\dotmbh(\tpi)\rangle_{\rm qso}\bar{\delta}_1(\tpi)\Delta \tpi$. The
183: mean mass of quasar black holes
184: then reads $\langle\mbh(t)\rangle_{\rm qso}=\langle\mbh^{\rm S}\rangle+
185: \sum_i \langle\dotmbh(\tpi)\rangle_{\rm qso} \bar{\delta}_1(\tpi)\Delta t_i^{\prime}
186: =\langle\mbh^{\rm S}\rangle+\int_0^t\langle\dotmbh(\tp)\rangle_{\rm qso}
187: \bar{\delta}_1(\tp)\rmd t^{\prime}$ after multiple episodic phases. We then have
188: $\langle\mbh(t)\rangle_{\rm all}=\langle\mbh(t)\rangle_{\rm qso}$.
189: This relation can be understood more easily if we consider multiple episodic growth of black holes.
190: We thus have the duty cycle of the black holes in a concise form of
191: $\bar{\delta}(z)\equiv \bar{\delta}_1(z)=\bar{\delta}_2(z)$.%=N_{\rm qso}/N_{\rm tot}$.
192:
193: %For quasars,
194: %$\langle\mbh(t)\rangle_{\rm qso}=\langle\mbh^{\rm S}\rangle+\int_0^{t_{\rm acc}}
195: %\langle\dotmbh(t^{\prime})\rangle_{\rm qso}\rmd t^{\prime}$,
196: %where $\langle\dotmbh(t^{\prime})\rangle_{\rm qso}$ is the mean accretion rate of quasar black holes with
197: %$\mbh>\mbh^*$ and the accretion time is $t_{\rm acc}=\int_0^t {\bar{\delta}}_1(t^{\prime})\rmd t^{\prime}$.
198:
199: The total mass density of the black holes at redshift $z$ contributed from accretion is then given by
200: %
201: \begin{equation}
202: \rho_{\rm acc}(z)=\int_{\infty}^z\left(\frac{\rmd t}{\rmd z}\right){\rmd z}
203: \int_{L_*(z)}^{\infty}\frac{1-\eta}{\eta}\frac{L_{\rm Bol}}{c^2}\Psi(L,z)\rmd L,
204: \end{equation}
205: %
206: where $L_*(z)$ is the limit luminosity due to survey sensitivity, $\Psi(L,z)$ is
207: the luminosity function, the bolometric luminosity is given by $L_{\rm Bol}=\calcb L_{\rm B}$,
208: with $\calcb$ the correction factor relating $B-$band luminosity $L_{\rm B}$ to
209: $L_{\rm Bol}$, $\eta$ the
210: radiative efficiency and $c$ the light speed. This density includes {\em all} the black holes
211: that were brighter than $L_*(z)$.
212: The mass density of the active black holes at $z$ with $L>L_*$ is given by
213: %
214: \begin{equation}
215: \rho_{\rm qso}(z)=\int_{\mbh^{\rm C}}^{\infty}\Phi(\mbh,z)\mbh\rmd\mbh,
216: \end{equation}
217: %
218: where $\mbh^{\rm C}$ is the limit black hole mass due to the limit luminosity.
219: We apply the So\l tan's argument to the present sample at any redshift $z$
220: %
221: \begin{equation}
222: \rho_{\rm tot}(z)=\int_{\mbh^{\rm C}}^{\infty}{\cal N}(\mbh,z)\mbh\rmd\mbh=\rho_{\rm acc}(z).
223: \end{equation}
224: %
225: Though we do not know the distribution $\caln$, we know the mass density of all black holes
226: from So\l tan's argument. Finally, we have duty cycle $\bar{\delta}(z)$,
227: %
228: \begin{equation}
229: {\bar \delta}(z)=\frac{\rho_{\rm qso}(z)}{\rho_{\rm acc}(z)}.%;~~~~
230: \end{equation}
231: %
232: If we know $\Psi(L,z)$ and $\Phi(\mbh,z)$, we can easily get
233: the duty cycle of quasars at redshift $z$. We have to stress that Eq. (7) neither needs an initial
234: condition nor the assumption of a constant Eddington ratio, something which must be specified for in
235: the alternative approach solving the continuity equation (e.g. $\bar{\delta}=1$ at $z=3$ in Marconi et al. 2004).
236:
237: \section{Sample and black hole mass function}
238:
239: \subsection{Spectrum Analysis}
240: The largest quasar sample, given by Richards et al. (2006) from the SDSS DR 3,
241: consists of 15,343 quasars from $z=0$ to 5, which is complete and homogeneous for apparent
242: magnitude $i=15-19$. We only use those quasars (11,954) with $z\le 2.1$ in the present paper.
243: We subtract the continuum and iron emission (based on the iron template derived from
244: I Zw 1 spectra, which is kindly provided by R. J. McLure from private
245: communication 2006). We then fit
246: H$\beta$ and Mg {\sc ii} lines. For those with $z\le 0.7$, four components are used to model the
247: spectra: broad and narrow H$\beta$ plus narrow [O {\sc iii}] 4959\AA~ and [O {\sc iii}] 5007\AA.
248: For others, we use one broad and one narrow component to model the Mg {\sc ii} line.
249: Some objects are removed from the sample for one of three reasons; they have: 1) too poor spectra to fit
250: due to low signal-noise ratio; 2) only narrow lines ($<2000~\kms$)
251: or 3) serious absorption at the Mg {\sc ii} line.
252: Finally we have 10,979 quasars available for estimating black hole masses
253: with $z\le 2.1$. The removed quasars reduce the completeness of the
254: sample given in Table 1. The last bin ($z_{10}$) is poor because the quality of quasar spectra
255: is not good enough to measure width of Mg {\sc ii}. A future paper will give a detailed description of
256: estimating black hole masses and related issues (Chen et al. in preparation).
257:
258: \subsection{Black Hole Mass Function}
259: We apply the latest version of
260: the empirical relation of reverberation mapping (Kaspi et al. 2005; Vestergaard \& Peterson 2006)
261: to calculate the black hole mass in each quasar.
262: For low redshift quasars with $z\le 0.7$, we use the full-width-half-maximum of
263: H$\beta$ whereas we use Mg {\sc ii} for those with $0.7<z\le2.1$ (McLure \& Dunlop 2004).
264: The scatter of the re-calibrated relation for the black hole masses is
265: less than 0.4 dex, which is much improved. Figure 1{\em a} shows the mass
266: distribution of the present sample. We fit the mass distribution of the black holes via
267: least square method in a form of three power laws
268: %
269: \begin{equation}
270: {\cal F}(\mmbh)=f_0\mmbh^{\alpha_1}
271: \left(1+\frac{\mmbh}{m_{\bullet 1}}\right)^{-\alpha_2}
272: \left(1+\frac{\mmbh}{m_{\bullet 2}}\right)^{-\alpha_3},
273: \end{equation}
274: %
275: where $\mmbh$ is mass of black holes in unit of solar mass.
276: This expression has the limits ${\cal F}(\mmbh)\propto \mmbh^{\alpha_1}$ for $\mmbh\ll m_{\bullet 1}$,
277: ${\cal F}(\mmbh)\propto \mmbh^{\alpha_1-\alpha_2}$ for $m_{\bullet 1}\ll\mmbh\ll m_{\bullet 2}$ and
278: ${\cal F}(\mmbh)\propto \mmbh^{\alpha_1-\alpha_2-\alpha_3}$ for $\mmbh\gg m_{\bullet 2}$.
279: We obtain $f_0=(2.70\pm 0.35)\times 10^{-25}$, $m_{\bullet 1}=(4.14\pm 0.56)\times 10^7$,
280: $m_{\bullet 2}=(2.50\pm 0.22)\times 10^9$,
281: $\alpha_1=3.56\pm 0.28$, $\alpha_2=2.87\pm 0.25$ and $\alpha_3=2.71\pm 0.07$. A significant break appears at
282: $\mbh=2.5\times 10^9\sunm$ and then a steeper mass spectrum ${\cal F}(\mmbh)\propto \mmbh^{-2.02}$
283: follows the break mass, which is consistent with the maximum mass from the SDSS DR1
284: (McLure \& Dunlop 2004).
285:
286:
287: {\footnotesize
288: \begin{center}{\sc Table 1 The mass function }\end{center}
289: \begin{center}
290: \begin{tabular}{ccrcccc} \hline\hline
291: $z$-bin & $\Phi_* $ & $\beta_1$~~~& $\beta_2$ &$\mbh^*$ & $\chi_j^2/d.o.f.$ & $\cal{C}$ \\
292: & $(10^{-6})$ & & & $(10^{9}\sunm)$ & & ($\%$) \\ \hline
293: $z_1$ & 48.1 & $-$0.60 & 1.55 & 0.01 & 4.40/4 & 98 \\
294: $z_2$ & 4.74 & 0.32 & 2.48 & 0.16 & 7.90/9 & 95 \\
295: $z_3$ & 5.94 & 0.79 & 2.58 & 0.18 & 8.77/10 & 93 \\
296: $z_4$ & 0.79 & 0.18 & 2.89 & 0.58 & 9.89/10 & 94 \\
297: $z_5$ & 0.44 & 0.48 & 3.36 & 1.25 & 8.25/10 & 95 \\
298: $z_6$ & 0.27 & 0.15 & 3.21 & 1.81 & 8.16/9 & 96 \\
299: $z_7$ & 0.29 & 0.42 & 2.82 & 1.53 & 8.78/8 & 97 \\
300: $z_8$ & 0.33 & 0.68 & 2.86 & 1.55 & 8.58/8 & 96 \\
301: $z_9$ & 0.28 & 0.50 & 2.54 & 1.29 & 6.45/7 & 89 \\
302: $z_{10}$ & 0.15 & 0.63 & 2.30 & 1.26 & 6.71/7 & 67 \\ \hline
303: \end{tabular}
304: \parbox{3.1in}
305: {\baselineskip 9pt
306: The last column is the completeness at each redshift bin in our sample. The redshift bin $z_j$ is defined
307: by $z_j=(j-1/2)\Delta z$, where $j=1,..., 10$.
308: The fittings are not included one point in $z_1$, $z_4$, $z_6$ and $z_8-z_{10}-$bins,
309: which significantly deviate from the trend of double power-laws.}
310: \end{center}
311: }
312:
313:
314:
315: We get the MF by dividing our sample into 10 redshift bins
316: with an interval $\Delta z=0.21$ and then into 20 black hole mass bins in
317: each redshift bin. Figure 1{\em b} shows the MF.
318: We find that the function can be well fitted by double power laws in the following form
319: %
320: \begin{equation}
321: \Phi(\mbh,z)={\Phi_*}\left[\left(\frac{\mbh^*}{\mbh}\right)^{\beta_1}+
322: \left(\frac{\mbh}{\mbh^*}\right)^{\beta_2}\right]^{-1},
323: \end{equation}
324: %
325: where $\beta_1$, $\beta_2$, $\Phi_*$ and $\mbh^{\ast}$ are constants. The peak mass is given by
326: $\mbh^{\rm peak}=\left(\beta_1/\beta_2\right)^{1/(\beta_1+\beta_2)} \mbh^*$.
327: We get the four parameters from the least $\chi-$square method via minimizing
328: $\chi_j^2=\sum_i\left[\Phi(\mbh,z)-\Phi_{ij}\right]^2/\sigma_{ij}^2$, where
329: $\Phi_{ij}$ is the MF derived from the sample in mass $M_{\bullet,i}-$bin and
330: redshift $z_j-$bin. Since the averaged error
331: bar in the luminosity function is about $\Delta \Psi/\Psi\sim 0.1$ (Richards et al. 2006), we take
332: $\sigma_{ij}=0.1\Phi_{ij}$ for the MF. Table 1 lists all the parameters for each redshift bin.
333:
334:
335:
336: \figurenum{1}
337: \centerline{\psfig{figure=f1.ps,angle=270,width=8.0cm}}
338: \figcaption{{\em a} shows the mass distribution of the black holes in our sample.
339: The red line is the least square fit. The black hole mass function
340: $\Phi(\mbh,z)$ at each redshift bin is shown in {\em b}. The lines represent the best fits given in Tab 1.}
341: \label{fig1}
342: \vglue 0.3cm
343:
344: The MF shows an increasing peak mass toward high redshifts. The mass break at the low mass
345: side is not real. It is caused by the survey flux limit. We have to point out that the MF
346: is {\em not} complete for all black holes, but it is complete for our flux-limited criterion.
347:
348:
349: \section{Results}
350: The $B-$band luminosity is converted from the absolute magnitude $M_i(z=2)$ via a relation
351: of $M_B=M_i(z=2)+0.804$ (Richards et al. 2006).
352: It has been found that the factor $\calcb$ is not a function of redshift (%Strateva et al. 2005;
353: Steffen et al. 2006). A more
354: elaborate treatment gives $\calcb=6-7$ for quasars (Marconi et al. 2004).
355: We use $\calcb=6.5$ throughout the redshift range in this
356: paper. The luminosity function is taken from Richards et al. (2006).
357: We take $\mbh^{\rm C}$ from the minimum mass of the black holes in each redshift bin to calculate
358: their mass density, $\rho_{\rm qso}$(Eq. 5). This corresponds to the survey limit and is consistent with
359: the accretion density $\rho_{\rm acc}$ (Eq. 4).
360: The radiative efficiency $\eta\ge 0.1$ is reached by Yu \& Tremaine (2002), $\eta\ge 0.15$
361: by Elvis et al. (2002) and $\eta=0.15$ by Gammie et al. (2004), based on the So\l tan's argument,
362: the X-ray background and numerical simulations, respectively. Numerical calculations indicate that black
363: holes are rotating with their maximum spin all the time during their evolution if accretion
364: is included (Volonteri et al 2005), which is supported by studies of SDSS quasars (Wang et al. 2006);
365: then $\eta\approx 0.3$ without significant evolution. We thus calculate
366: the duty cycle for two different radiative efficiencies $\eta=0.1$ and $0.3$.
367:
368: \figurenum{2}
369: \centerline{\psfig{figure=f2.ps,angle=270,width=8.0cm}}
370: \figcaption{\footnotesize {\em a, b} show the duty cycle of quasars as a function of redshift in our
371: sample and the history of the SFR density, respectively. The SFR density
372: is taken from P\'er\'ez-Gonz\'alez et al. (2005).
373: }
374: \label{fig2}
375: \vglue 0.3cm
376:
377: Results are shown in Figure 2. First, quasars have duty cycle of $\bar{\delta}(z)=10^{-3}\sim 1$
378: as shown in Fig 2{\em a}, indicating that black holes are undergoing active and dormant phases, namely
379: episodic activity. The value of $\bar{\delta}(z)\rightarrow 1$ at $z\sim 2$ agrees with the assumption
380: of $\delta=1$ at $z=3$ in Marconi et al. (2004).
381: Second, the duty cycle is rapidly evolving from $z\approx 2$ to the local Universe. We find
382: $\bar{\delta}(z)\propto z^{\gamma}$, where $\gamma\sim 2.5$ until $z=1.5$, above which it tends to flatten.
383: Third, as shown in Fig. 2{\em a} and 2{\em b}, the duty cycle connects quite naturally with the history
384: of the SFR density as a consequence of co-evolution of galaxies and black holes.
385: Massive, gas-rich mergers account not only for most of the star formation at
386: $z\approx 2-3$, but they are probably also responsible for triggering major episodes of
387: black hole activity (Di Matteo et al. 2005). Figures 2{\em a} and 2{\em b} show that the duty cycle follows
388: the star formation history, implying evidence for
389: intrinsic star-formation-triggered black hole activity. The higher the SFR density,
390: the higher the triggering frequencies of the black hole activity. Last, we define
391: ${\cal R}=\Delta t_{\rm act}/\Delta t_{\rm dor}$, then have
392: ${\cal R}=\bar{\delta}(z)/\left[1-\bar{\delta}(z)\right]$.
393: Episodic activity of the black holes can be described by this parameter.
394: When $\bar{\delta}(z)\rightarrow 1$,
395: black holes have ${\cal R}\gg 1$, namely, the dormant black holes are frequently
396: triggered at $z\sim 2$ by star formation.
397: At that time, quasars look like long-lived phenomena because of ${\cal R}\gg 1$.
398:
399: \section{Discussion and Summary}
400: The evolution of quasars is jointly controlled by the triggering mechanism and accretion.
401: The duty cycle is a key parameter to unveil the evolution of quasars. The results of the present paper show
402: a very strong cosmological evolution of quasar's duty cycle.
403: The triggering history represented by $\bar{\delta}(z)$ is quite similar to the
404: evolution of cosmic SFR density. This indicates that star formation may be the direct
405: mechanism to trigger the activity of the black holes.
406:
407: The duty cycle can be roughly justified from the galaxy luminosity function.
408: According to the luminosity function of galaxies at $1.8\le z\le 2.0$ (Dahlen et al. 2005), the galaxy
409: number density is $n_{\rm G}\approx 826$Gpc$^{-3}$ for galaxies brighter than $R-$band magnitude
410: $M_{\rm R}=-24$. This corresponds to the number density of galaxies with black hole mass larger than
411: $10^9\sunm$ converted from $\log \left(M_{\rm BH}/\sunm\right)=-0.5M_{\rm R}-3$ (McLure \& Dunlop 2001).
412: Number density of quasars brighter than $M_i=-28$ (corresponding to a black hole with mass $>10^9\sunm$
413: if it is accreting at the Eddington limit) is $n_{\rm Q}\approx 175$ Gpc$^{-3}$ based
414: on the quasar luminosity function (Richards et al. 2006). We thus estimate a duty cycle of $\sim 0.18$, which is
415: roughly consistent with the present results (see Figure 2a).
416:
417: The SFR density rises with redshift out to $z=1.5$ and appears to be roughly flat between $z\approx 1.5$ and
418: $z\approx 3.0$. This tendency
419: could be explained by strong feedback from activity of black holes to their host galaxies
420: (Silk 2005; Di Matteo et al. 2005; Croton et al. 2006). With a balance between star formation and
421: feedback in $z\sim 1.5-3.0$, the violent star formation is then suppressed.
422: However, the SFR density is going to decrease with time due to
423: a shortage of gas and the BH duty cycle follows this trend. To further confirm this, future work
424: will focus on the dependence of the duty cycle on the black hole masses. It could show the
425: feedback-dependence on the BH growth itself. Additionally the total accretion time
426: (net lifetime) of black holes will be then obtained.
427:
428:
429: \acknowledgements{The anonymous referee is greatly acknowledged for pointing out a mathematical error in
430: the paper. The authors are grateful to M. Z. Kong, R. Wang and R. J. McLure for help in SDSS data
431: analysis, and L. C. Ho, S. N. Zhang, J. X. Wang and X. Y. Xia for useful discussions. This research is
432: supported by the Natural Science Foundation of China through NSFC-10325313, 10233030 and 10521001. }
433:
434:
435: \begin{thebibliography}{1}
436: \bibitem[]{373}Croton, D. J., et al., 2006, MNRAS, 365, 11
437: \bibitem[]{374}Dahlen, T., et al., 2005, ApJ, 631, 126
438: \bibitem[]{375}Elvis, M., Risaliti, G., \& Zamorani, G., 2002, {\apj}, 565, L75
439: %\bibitem[]{376}Hopkins, A. M., 2004, ApJ, 615, 209
440: \bibitem[]{377}Gammie, C. F., Sharpiro, S. L., \& McKinney, J. C., 2004, \apj, 602, 312
441: \bibitem[]{378}Di Matteo, T., Springel, V., \& Hernquist, L., 2005, \nat, 433, 604
442: \bibitem[]{379}Kaspi, S., et al., %Maoz, D., Netzer, H., Peterson, B. M., Vestergaard, M., \& Jannuzi, B. T.,
443: 2005, \apj, 629, 674
444: \bibitem[]{381}Marconi, A., Risaliti, G., Gilli, R., Hunt, L. K., Maiolino, R., \& Salvati, M.,
445: 2004, {\mnras}, 351, 169
446: \bibitem[]{383}Martini, P., 2004, in Coevolution of Black Holes and Galaxies, from the
447: Carnegie Observatories Centennial Symposia. by Cambridge University Press,
448: Ed. L. C. Ho, p. 169
449: \bibitem[]{387}Martini, P., \& Schneider, D., 2003, \apj, 597, L109
450: \bibitem[]{388}McLure, R., \& Dunlop, J. S., 2001, \mnras, 327, 199
451: \bibitem[]{389}McLure, R., \& Dunlop, J. S., 2004, \mnras, 352, 1390
452: \bibitem[]{390}P\'er\'ez-Gonz\'alez, P. G., et al. 2005, ApJ, 630, 82
453: \bibitem[]{391}Rees, M. J., 1984, \araa, 22, 471
454: \bibitem[]{392}Rees, M. J., 1990, Science, 247, 817
455: \bibitem[]{393}Richards, G. T., et al., 2006, \aj, 131, 2766
456: \bibitem[]{394}Richstone, D., et al., 1998, Nature, 395A, 14
457: \bibitem[]{396}Silk, J., 2005, MNRAS, 364, 1337
458: \bibitem[]{397}Small, T.A., \& Blandford, R. D., 1992, \mnras, 259, 725
459: \bibitem[]{398}So\l tan, A., 1982, \mnras, 200, 115
460: \bibitem[]{399}Steffen, A. T., et al. %Strateva, I., Brandt, W. N., Alexander, D. M., Koekemoer, A. M.,
461: %Lehmer, B. D., Schneider, D. P., Vignali, C.
462: , 2006, \aj, 131, 2826
463: %\bibitem[]{401}Strateva, I. V., Brandt, W. N., Schneider, D. P., Vanden Berk, D. G., \& Vignali, C.,
464: %2005, \aj, 130, 387
465: \bibitem[]{405}Vestergaard, M., 2004, ApJ, 601, 676
466: \bibitem[]{406}Vestergaard, M., \& Peterson, B. M., 2006, \apj, 641, 689
467: \bibitem[]{407}Volonteri, M., Madau, P., Quataert, E., Rees, M. J., 2005, \apj, 620, 69
468: \bibitem[]{408}Wang, J.-M., Chen, Y.-M., Ho, L. C., \& McLure, R. J., 2006, \apj, 642, L111
469: %\bibitem[]{409}York, D. G., et al., 2000, {\aj}, 120, 1579
470: \bibitem[]{410}Yu, Q., \& Tremaine, S., 2002, {\mnras}, 335, 965
471: \end{thebibliography}
472:
473:
474: \end{document}
475: