Cosmological Evolution of Hard X-ray AGN Luminosity Function:Formation History of the SMBHs

Yoshihiro Ueda (ISAS)

Co-Is on construction of the HXLF:

Masayuki Akiyama (Subaru Telescope, NAOJ)

Kouji Ohta (Kyoto Univ.)

Takamitsu Miyaji (Carnegie Mellon Univ.)

Special thanks to ASCA extragalactic survey teams

Contents

Optical identification of the ASCA MSS

(Akiyama et al. 2003 ApJS, astro-ph/0307164)

• Nature of Hard X-ray population at bright fluxes (fx > 10-13 erg s-1 cm-2 in 2-10 keV)

Hard X-ray Luminosity Function (HXLF) (U et al. 2003 ApJ, astro-ph/0308140)

• NH function and the HXLF

• Population synthesis: Origin of the HXRB

Observational constraints on the SMBH formation: growth curve of SMBHs

MotivationAGN Luminosity Function:

a goal of X-ray surveys

the most fundamental measure to understand the cosmological evolution of AGNs

How many AGNs in the universe (as a function of time)?

How supermassive black holes form?

The X-ray Background : 40 years mistery

Ginga spectrum of Seyfert II galaxies(Awaki et al 1991)

Comastri et al. 1995

Importance of hard X-ray surveys

The XRB: all the integrated emission from acceting supermassive black holes (SMBHs) in galaxies To reveal the XRB origin is to reveal the evolution of

AGNs i.e., the accretion history of SMBHs in the universe

Majority of AGNs are obscured !• from the XRB spectrum• Ginga spectra of Seyfert II galaxies (Awaki et al 1991)• column density distribution of neaby Seyferts II Galaxies (Ris

aliti et al 1999) Surveys in hard band (above 2 keV) are crucial to detect a major population of AGNs

The ASCA Medium Sensitivity Survey

(U. et al. 2001; Akiyama et al.2003)

Optical Identification of the AMSS-north sample• GIS 2-10 keV selected sample

Fx > 3x10-13 erg cm-2 s-1 (2-10 keV )• 87 sources / 68 degree2

• UH 2.2m, KPNO 2.1m, (Calar Alto 3.5m, SUBARU)• 79 AGNs (3 BL Lacs), 7 Clusters, 1 star

Optical Identification of the AMSSn sample (Akiyama, U, Ohta et al. 2003)

X-ray absorbed AGNs (log NH >22)

(1) at z<0.6

do not show broad Hβline (x)

(2) at z>0.6

often show broad MgII, CIII, or Lyαlines

Scattered nuclear light?

Summary of optical properties of X-ray absorbed AGNs at Fx ~ 10-13

Low-redshift Low-luminosity AGNs• Classical Type 1.8-2 Seyfert galaxies• Very hard sources: optically “normal” galaxies (Watanabe et al 2002; cf. Mushotzky et al 2000 for Chandra)

High-redshift High-luminosity AGNs• Broad-lines are detected: X-ray type-II AGNs are not always optically type-II AGNs• We can not neglect the effect of scattered nuclear light

• Av/NH values of AGNs are smaller than Galactic one ?

The Hard X-ray AGNLuminosity Function

Construction of the HXLF We have performed statistical analysis from a combin

ed AGN sample in the wide z-Lx range to derive • NH function (probability distribution function of absorption)• luminosity function (spatial number densities)

as a function of luminosity and redshift

Construction of a sample 1. selected in the hard X-ray band (E >2 keV) 2. covering a wide flux range 3. with high completeness

is crucial, in particular to constrain the evolution of type-II AGNs, major populations of the XRB.

Sample: 247 Hard-band Selected AGNsTotal Completeness > 96 %

HEAO-1 A2 + MC-LASS(total 49, Fx>1.7x10-11)Shinozaki (Poster 77)

ASCA (142, Fx>2x10-14) ALSS (30) Akiyama et al (2000)

AMSS(75+20) Akiyama et al (2003)

Deep surveys (12+5)

CDFN flux limited(56, Fx>3.8x10-15)

Brandt et al (2001) Barger et al (2002,2003)

X

X-ray type II (NH>1022)

Optical type II

Analysis

Utilize the maximum likelihood method Search for a solution where the probability of finding the obse

rved set of (L, z, NH) is maximized Actual sensitivity is determined by count rate rather t

han “flux”, and hence depends on the spectral shape Take into account detector response in each survey Consider the spectrum of each source by referring to HEAO-1: ASCA/XMM follow-up results ASCA and CDFN: hardness ratio (assuming a photon index of 1.9 with a reflection component)

Compton-thick AGNs are (and can be) neglected

Observed Luminosity Distribution

Observed fraction of type-II AGNs (both in X-ray and optical) decreases with the hard X-ray luminosity

Optical Type II AGN

X-ray Type II AGN

Total

Results: the NH Function

The probability distribution function of absorption column density of AGNs, as a function of Lx (and z)

Quantitative test of the “unified scheme”

All

Log Lx <43

43<Log Lx <44.5

Log Lx >44.5

Fraction of X-ray Absorbed AGNs

The fraction of X-ray absorbed AGNs (Log NH >22) decreases with the luminosity.

Its redshift dependence is not significant.

Modification of the simplest unified scheme is necessary. claimed in e.g., Lawrence & Elvis (1982), Akiyama et al. (2000). See also Steffen et al (2003)

The HXLF of All the Compton-thin AGNs (Type-I + Type-II)

Best described with a luminosity dependent density evolution (LDDE) where the cut-off redshift increases with the luminosity

The AGN Number Density as a Function of Redshift

Luminous AGNs has a peak (cutoff redshift) earlier than less luminous AGNs

Similar results: see Miyaji et al (Poster 54) Fiore et al (2003)

Related to the star forming activity?

e.g., Francheschini et al. (1999)

• The evoluiton of the number density of high (low) luminosity AGNs is simlar to that of the star forming rate of early (late) type galaxies.

• A strong link between formation of the SMBH and that of the spheroid component of galaxies.

~(1+z) 4 (z>zc)

Population Synthesis Model

Integrated AGN emission from our HXLF and the absorption function (lower black) well reproduces the broad band XRB spectrum (blue) below 300 keV

High energy cutoff must be arround 500 keV in average

(red left: 400 keV, red right: 600 keV)

Presense of Compton-thick AGNs estimated by Risaliti et al (1999) is consistent with the XRB spectrum (upper black)

Reflection components are important

(green: no reflection)

Observed XRB spectrum

Composition of the XRBAGNs with Log Lx ~ 43.8 or at z~0.6most largely contribute to the 2-10 keV XRB

Comparison of the Quasar Optical LF with the HXLF

Importance of Hard X-ray surveys for AGN astronomy

Data: Optical LF (Boyle et al 2000)Line: Prediction from the HXLF

Formation history of the SMBHs: mass density

Total accreted mass (hence BH mass density) can be estimated by LF

assuming the mass-to-energy conversion factor ε(Soltan 1982).

Compare with a local BH mass density estimated from the relation between BH mass and velocity dispersion to constrain ε.

Previous studies:• Fabian and Iwasawa (1999) , Elvis et al (2002):

use the XRB intensity assuming z=2 for all the XRB source

• Yu and Tremaine (2002)

use the QSO optical LF from 2dF survey

The growth curve of SMBHs revealed from the HXLF

For ε=0.1 the total accreted mass estimated from the HXLF is marginally consistent with the present-day BH mass density from the SDSS, (2.9±0.6)x105 Mo Mpc-3 (Yu and Tremaine 2002), indicating that radiatively inefficiant accretion flow is not important for formation of SMBHs (F&I 1999).

Significant accretion has continued after z<1

From QSO OLF (Lx>1044)

With Compton-thick AGNs

Conclusion The fraction of X-ray absorbed AGNs decreases with

the intrinsic luminosity. The HXLF of total AGNs is best described by a LDDE where the cutoff redshift increases with the luminosit

y: the number density of luminous AGNs has a peak in earlier epochs than less luminous ones.

Purely “observation based” population synthesis model is constructed. Most of the XRB origin is now quantitatively resolved!

These results give: Strong constraints on galaxy/QSO formation history Trace of the accretion history of the universe of all th

e AGNs including obscured objects (i.e., RIAF is not important for SMBHs formation)