Recording obscured accretion across cosmic time

1
Recording obscured accretion across cosmic time
Roberto Gilli (INAF - OABo) Andrea Comastri
(INAF OABo) Guenther Hasinger
(MPE) Alessandro Marconi (INAF OAA) Cristian
Vignali (UniBo)
2
Rationale
Obscured accretion difficult to detect but leaves
(at least) two traces 1) X-ray background 2)
Local Supermassive Black Holes mass
function Use both to determine the size of
the obscured AGN population
3
Current Status on obscured AGN census
Low z High z
e.g Maiolino Rieke 1995
Low L High L
Trends Rabs/unabs decreases with Lx
generally agreed (Ueda03, La Franca05,
TreisterUrry05, Ballantyne06, GCH07, Della
Ceca talk, Sazonov06) Interpretation receding
torus model (e.g. Lawrence 1991, Simpson 2005) R
increases with redshift still debated (YES La
Franca05,Treister05,Ballantyne06 NO
Ueda03,GCH07) Interpretation AGN obscured at
their birth (Fabian 1998, Hopkins 06) Number of
C-thick AGN largely unknown but see talks by
Fabian Comastri
e.g Martinez-Sansigre2005
4
XRB model scheme

• Use available constraints at Elt10 keV to lock the
  properties of moderately obscured AGN and
  accurately estimate their contribution to the
  XRB. The Compton-thin/unabs ratio is
• obtained by comparing hard vs soft XLFs.
• Add Compton thick AGN to fit the 30 keV bump
• Verify assumptions/make predictions on Compton
  Thick AGN

5
Number of Compton-thin AGN from XLF comparison
Ueda03
La Franca05
Hard XLF C-Thin unobs. (Ueda03, La
Franca05, Silverman07) Soft XLF unobscured
(Hasinger05)
6
Dependence of Rthin/unabs AGN ratio on L and z

• Volume density ratio between soft and hard X-ray
  selected AGN
• Model independent
• XLF approach - only trend with L

Best fit thin/unabs ratios 4 for Log Lx lt 44
1 for Log Lx gt 44
2.6-4.8 0.6-1.5
7
Obscured AGN fraction vs luminosity observed
fractions
Hasinger07 in prep Lanzuisi07 in prep
Possible overestimate of number of luminous
QSOs. Caveats observed datapoints are likely
lower limits because of incompleteness
8
Obscured AGN fraction vs luminosity intrinsic
fractions
Decrease of R with luminosity generally found.
The steepness of this decrease still to be
determined with good accuracy larger samples of
luminous sources needed
Ueda03 Akylas06
Behaviour with z is model independent
9
The XRB spectrum
10
AGN X-ray spectral templates
Mildly thick (log NH 24-25) e.g. NGC 6240,
Circinus, NGC 5728 (new, Suzaku, Comastri et al.
in prep.) Heavily thick (log NH gt25) e.g. NGC1068
Assumption N heavily N mildly
Normalization of the NGC1068-like spectra (i.e.
reflection fraction f ) highly uncertain 2
assumed - average of literature results
11
Uncertainties on the C-thick number
Lower no heavily-thick Upper heavily 4 x
mildly (f 2)
Number of heavily c-thick poorly constrained by
XRB models. Also, XRB models only constrain the
product between the number of heavily thick AGN
and the reflection fraction f very low f ? very
high N_heavy
12
Local BHMF vs Relic BHMFradiative efficiency
and Eddington ratio
Once the total number of AGN is known the only
free parameters are the average accretion
efficiency and Eddington ratio.
e0.06 ?0.2 which are consistent with common
beliefs on AGNs
About 80 of the local SMBH Mass density has been
produced during obscured growth phases
13
Uncertainties on the C-thick number
Lower no heavily-thick Upper heavily 4 x
mildly
Use the maximum theoretical value allowed for e
and get an upper limit to the total number of
heavily thick e lt 0.4 ? heavily lt 26 x
unabs loose constraints
When constraints on f x N_heavy are combined with
the SMBH mass function one gets a hard limit of
heavily lt 26 x unabs (at z1). From mid-IR excess
sources heavily 1-5 x unabs at z2 (Daddi et
al. 2007, Fiore et al. 2007)
14
Summary

• XRB synthesis models self consistently explain
  the observational contraints
• below 10 keV and can be used to estimate the
  number of C-thick AGN
• unabs thin thick 1 4 4 at low L
• 1 1 1 at high L
• Decrease of abs/unabs AGN ratio with luminosity
  rather robust, steepness of the
• decrease still to be determined accurately.
  Behaviour with redshift debated.
• C-thick AGN are expected to be C-thin AGN
  heavily c-thick AGN poorly
• determined from both XRB and SMBH in principle
  there is room for a very
• large population of heavily obscured AGN ? ?
  Check IR-background and counts

15
X-ray counts
16
Soft and hard counts are measured with good
accuracy over several flux decades ? These are
primary constraints for XRB synthesis models

17
C-thick AGN in X-ray surveys
logN-logS
Fraction
10-40 keV
Simbol-X
2-10 keV
XEUS
CDFS Tozzi et al. 06 Swift Markwardt et al.
06
18
The AGN BH Mass Function

• Assume accretion onto BH as powering mechanism of
  AGN to link LAGN with MBH L ? MBHc2/tE
  e (dM/dt)c2
• Use the continuity equation (Cavaliere et al.
  1971) to relate the BH Mass function N(MBH) to
  the AGN Luminosity function F(L)
• Critical issues
• L is the TOTAL accretion luminosity ? need
  bolometric corrections
• F(L) is the luminosity function of ALL AGNs ?
  need C-thick AGN