Prsentation PowerPoint

1
UNCERTAINTIES ON THE BLACK HOLE MASSES AND
CONSEQUENCES FOR THE EDDINGTON RATIOS Suzy
Collin Observatoire de Paris-Meudon,
France Collaborators T. Kawaguchi (Tokyo), B.
Peterson (Ohio U), M. Vestergaard (Steward
obs), C. Boisson, M. Mouchet (Paris)
2

• Uncertainties on direct mass determinations
• Indirect mass determinations super-Eddington
  accretion rates

3
I. UNCERTAINTIES ON THE MASS DETERMINATIONS
IN 35 SEYFERT AND LOW REDSHIFT QUASARS BH MASSES
ARE DETERMINED DIRECTLY BY THE REVERBERATION
MAPPING METHOD It consists in measuring the
time delay between the continuum and the line
variations which respond to them it gives an
(approximate) size of the broad Line Region
. Assuming that the BLR is gravitationally
bound (certainly true for the Balmer line
emitting region), the mass of the BH,
, is then where is the dispersion
velocity, and a scale factor. It is usually
assumed (Peterson Wandel 1999, Kaspi et al.
2000) that
which correspond to an isotropic BLR with a
random distribution of orbits.
?
R(BLR)
M(BH),
M(BH) f (c ? V2/G) f (Virial Product)
V
f
V FWHM, and f 3/4
by-product? an empirical relation between R(BLR)
and L(optical)
IN ALL OTHER OBJECTS (EXCEPT ONE) THE MASSES ARE
DETERMINED INDIRECTLY USING THIS EMPIRICAL
RELATION
4
VARIOUS UNCERTAINTIES

• The scale factor f depends on the geometry and on
  the kinematics of the BLR, and most probably on
  the Eddington rate (Collin et al. 2006).
• What is the best choice for measuring V? The
  FWHM (all works) or ?line? ?line seems more
  reliable (Peterson et al. 2004), but it is
  generally not measured.
• Is it better to use the RMS or the mean spectrum?
  
• Etc

5
Example of systematic uncertainties
Sample of reverberation mapped objects
Peterson et al. 2004 use ?(line) instead of
FWHM, RMS spectrum instead of mean spectrum, and
changed the factor f (scaled on the bulge masses
by Onken et al. 2004)
6
Collin, Kawaguchi, Peterson, Vestergaard (2006)
THE SCALE FACTOR IS NOT A CONSTANT
broad flat topped lines
Gaussian profile
NGC5548
narrow peaked lines
Reverberation mapped objects, all datasets
7
Scale factor determined by fitting M(BH) to
M(?), M(BH) being FWHM-based
Pop 2
f 0.85 ? 0.15
Remember f 0.75 is used
Collin et al.
f 2.4 ? 1
Pop 1
Pop A
Pop B
similar to Sulentic et al.
8
THE INCLINATION OF THE BLR PLAYS A ROLE
Assuming that the velocity includes a plane
rotational plus an isotropic part Vobs VKep (a
2 sin?2)1/2 and using the distribution of M(?)
/M(RM), we found that
a fraction of Pop1 objects should be seen at
small inclination their masses can be
underestimated by factors up to ten (NGC4051,
Mrk590, NGC7469)
9
BH MASSES DETERMINED IN LARGE SAMPLES
INDIRECTLY BY THE EMPIRICAL RELATION
Kaspi et al. 2000, revised by Kaspi et al. 2005
Allows to determine M(BH) for single epoch
observations, simply by measuring Lopt (?L? at
5100A) and the FWHM
Determination of Lbol/Ledd Lbol generally
deduced from Lopt, assuming Lbol 10 Lopt
10

• CAUTION!
• IT ELIMINATES THE INTRINSIC DISPERSION OF THE L-S
  RELATION
• THE SCALE FACTOR CAN STILL BE WRONG
• THE INCLINATION CAN STILL AFFECT THE WIDTHS

11
Example from the SDSS, McLure Dunlop 2004
L/Ledd1
Narrow line objects
The relation leads to very high M(BH) for
luminous quasars (1010Mo, Netzer 2003,
Vestergaard 2002).
12
II. EXISTENCE OF SUPER-EDDINGTON ACCRETION RATES
Are quasars accreting at super-Eddington rates?
Collin, Boisson, Mouchet, et al.
2002 Reverberation mapped sample of Kaspi et al.
2000
13
BUT

• We have used Ho50, overestimating the
  luminosities by a factor 2
• 2. The masses of the sample have been revised by
  Peterson et al. (2004)

BUT THERE ARE OTHER SAMPLES, WITH BH MASSES
DETERMINED INDIRECTLY
STRONG DECREASE OF
14
THE ACCRETION RATE MUST NOT BE CONFUSED WITH THE
LUMINOSITY!
Let assume that Lopt is due to a thin accretion
disk (as usually accepted) In the optical, the
AD radiates locally like a BB (Hubeny et al. 2001)
cos ? 0.7, Corr-bol10
?
L/Ledd0.1
L/Ledd1
FOR SMALL MASSES AND LARGE LOPT, THE EFFICIENCY ?
SHOULD BE VERY SMALL
15
example of super-Eddington accretion rates
BH masses deduced from the size-luminosity
relationship
Collin Kawaguchi, 2004
16
Another example of super-Eddington accretion rates
Collin Kawaguchi, 2004
17
Super-Eddington accretion rates are found not
only for NLS1s, but generally for low mass samples
18
McLure Dunlop 2004
L/Ledd1
After correction for the efficiency
19

• COULD THERE BE ANOTHER EXPLANATION?
• 1. Super-Eddington accretion rate (not large, 1
  Mo per year) at large distance from the BH, but
  super-Eddington relativistic wind at small
  distance (Pounds et al. 2004, Gierlinski Done
  2004, Chevallier et al. 2006).
• 2. The optical-UV emission is not due to the
  accretion disk, even taking into account a
  non-gravitational external heating but to what
  else?

3. The empirical L-R(BLR) relation may not be
valid at large L/Ledd and small masses. 4.
Alternatively, we observe really super-Eddington
accretion rates. Indeed
20
Super-Eddington accretion rates are well
explained by slim disks
Collin Kawaguchi, 2004
21
COSMOLOGICAL CONSEQUENCES OF SUPER-EDDINGTON
ACCRETION RATES

• During their low mass phase, the growth time of
  the BHs is not Eddington limited (but most
  probably mass supply limited) it is thus much
  smaller than the Eddington time (Kawaguchi et al.
  2004).

Super-Eddington accretion can explain the rapid
early growth of BHs
2. It implies that the BH/bulge mass relationship
for NLS1s may be more dispersed than for other
objects.
22
SUMMARY

• There are both random (factor 3) AND systematic
  uncertainties (in particular in the scale factor)
  in the determination of M(BH) using reverberation
  mapping technique.
• These uncertainties are exported to the masses
  determined indirectly through the L-R(BLR)
  relationship in the other AGN. Moreover it is not
  clear whether this relationship can be
  extrapolated to large and small masses and to
  large Eddington factors.
• If it can be extrapolated to large Eddington
  factors (1), it implies that the accretion rates
  should be strongly super-Eddington in low mass
  objects ( below 108Mo).
• It can have important consequences for the
  cosmological growth of BHs.

23
Example Collin Kawaguchi, 2004
A very tight correlation appears, due to the
neglect of the error bars on the empirical
relation
RM objects Boroson et al 04
Grupe et al 99 Grupe et al 03 Veron
et al 01 Lopt lt 5 1043 ergs/s
NLS1
x