Transcript The BeppoSAX view of Galaxy Clusters:

CLUSTERS OF GALAXIES
From Cooling-Flows
to Cool Cores
Forward
About 7 years ago 2 new X-ray
satellites have been launched
X-ray Observatories
After the rocket experiments during the
1960s, the first X-ray Earth-orbiting
explorers were launched in the 1970s:
Uhuru, SAS 3, Ariel5
 followed in late 1970s early 1980s by
larger missions:
HEAO-1, Einstein, EXOSAT, and Ginga.

X-ray Observatories



In the 1990s the ROSAT survey
detected more than 100,000 X-ray
objects
the ASCA mission made the first
sensitive measurements of the X-ray
spectra from these objects
BEPPOSAX contributed along this line
Current X-Ray Missions
Chandra
XMM-Newton
The X-ray Telescope Chandra
Chandra detectors
PSF
 DISPERSIVE SPECTROMETERS
All convert  into dispersion angle and hence into focal plane position in an X-ray
imaging detector
•BRAGG CRYSTAL SECTROMETERS (EINSTEIN, SPECTRUM X-GAMMA): Resolving
power up to 2700 but disadvantages of multiplicity of cristals, low throughput, no
spatially resolved spectroscopy
n x  = 2d x sin
•TRANSMISSION GRATINGS (EINSTEIN, EXOSAT, CHANDRA)
m x  = p x sin
where m is the order of diffraction and p the grating period
•REFLECTION GRATINGS (XMM)
m x  = p (cos - cos)
The resolving power for gratings is given by , assuming a focal lenght f and a position X
relative to the optical axis in the focal plane
X = f tan  f sin X = f  so
 is constant
X  1 p

 
X

R
m
Chandra Versus Previous Generation
X-ray Satellites
Previous X-ray telescopes had either good spatial resolution or spectral resolution
Rosat
Good Spatial resolution
Low or no Spectral resolution
ASCA
Low Spatial resolution
Good Spectral resolution
Chandra got both
Chandra Versus Previous Generation X-ray
Satellites
An Imaginary Test
RosatChandra
view of
ASCA view of
view of “the Creation”
“the creation” of
“the creation” of
Michelangelo of Michelangelo
Michelangelo
The RGS Result
A1795 Tamura et al. (2001a);
A1835 Peterson et al. (2001);
AS1101 Kaastra et al. (2001);
A496 Tamura et al. (2001b);
sample of 14 objects Peterson
et al. (2003)
There is a remarkable
lack of emission lines
expected from gas cooling
below 1-2 keV.
The most straightforward
interpretation is that gas
is cooling down to 2-3 keV
but not further.
Peterson et al. (2001)
Standard CF model predicts gas with T down to at least 0.1 keV!
The EPIC Result
EPIC has a spectral resolution ~ 10 times
worse than RGS.
It cannot resolve individual lines.
However it can discriminate btwn. models
with and without a minimum temperature
The major discriminant is the Fe L Shell
blend profile
Comparison btwn. multi-temperature
models
Spectra above ~1.3 keV are
similar.
Model spectra degraded to the EPIC resolution
Tmin=0.9 keV
Below we observe a
prominent line-like feature:
Fe-L shell line complex.
In the spectrum with
Tmin=0.1 keV we see a
shoulder down to ~ 0.8, this
is due to low ionization lines
from gas colder than 0.9
keV.
In the spectrum with
Tmin=0.9 keV the shoulder is
absent because the low
ionization lines are missing
Tmin=0.1 keV
Molendi & Pizzolato (2001)
EPIC vs. RGS Tmin
EPIC minimum
temperatures are
in good agreement
with RGS minimum
temperatures.
The result on Tmin
is a solid one!
All clusters observed so far show a Tmin
Values range between ~1 and ~3 keV
Way out:
hiding the flow
An attempt to save CF models (Fabian et al.
2001)
How can we have gas cooling below ~1 keV
without observing low ionization lines?
Way out:
Hiding the flow
Gas Mixing
For r< 20 kpc gas with T~103 K is present. Mixing of
hot, T ~3•107 K with cold gas may rapidly cool the hot
gas to T ~ 3 • 105 K
Gas at temperatures of T~103 K is seen in the
innermost ~ 20-30 kpc of only some clusters, thus it
may only work for the inner regions of some objects.
Way out:
hiding the flow
Differential Absorption
The absorber could be patchy and concentrated near the
center and perhapes absorb the gas producing the lines which
are not seen.
A few of the clusters observed so far have at their center an
AGN visible at X-ray wavelengths. The spectra of these AGN
do not show any evidence of absorption.
Way out:
hiding the flow
Bi-modal metallicity (very ingenuous!)
If metal distribution is highly bimodal (e.g. 10% of gas with
Z=3 and 90% of gas with Z=0), with Z rich gas in small clumps
(r < 1 kpc), for T> 2 keV Z rich and poor gas would cool
togheter, for T< 2 keV, when line emission becomes an
important coolant, Z rich gas would cool much more rapidly
than Z poor gas
Observed spectra are not well fit with bi-modal spectral
models.
Consequences
Multiphaseness
Gas is NOT multiphase, at least not
in the sense required by the standard
multi-phase CF model
Multiphaseness is or was a fundamental
ingredient of the CF model, without it
the model falls!
Intrinsic Absorption
•Cooling flow spectrum characterized by intense soft emission
•Spectrum with Tmin~ 1 keV not as much
•If you want to reproduce a spectrum with a Tmin~ 1 keV
with a CF model you have to get rid of the soft emission
•A possible way is to assume intrinsic absorption
•The large absorption column depths inferred by previous
analysis are an artifact resulting from the application of an
incorrect model to the data
Mass Deposition
•Little evidence of gas cooler than 1-2 keV
anywhere
•If gas does not cool below 1-2 keV it will not be
deposited as cold gas
•The gas could still be multi-phase on scales we do
not resolve with XMM/EPIC
•However at least in the case of M87 our resolution
is of a few kpc
Mass Deposition
RGS detects only very weak lines from gas cooler
than ~ 1 keV (Peterson et al. 2003)
occurs on scales which are not spatially resolved
by EPIC
Mass deposition is smaller and
confined to smaller scales than
previously thought
The old problem of not finding
abundant cooled gas is solved
Now that we have brought the house down it
is time to think about rebuilding
Let’s try with a fresh point of view, let us go
to Chandra observation.
Chandra has a very sharp eye, PSF better
than 1 arcsec about 10 times better than
XMM-Newton EPIC
The Chandra View
Hydra A
Chandra finds what
appear to be holes
“cavities”.
Radio lobes are conicident
with X-ray cavities
Radio lobes inflated by
jets appear to be making
their way pushing aside
the X-ray emitting
plasma
McNamara et al. (2001)
The Chandra View
Abell 2052 Blanton et al. (2001)
Radio lobes fill X-ray cavities
Cavities are surrounded by denser & cooler gas
The Chandra View
Centaurus, Sanders et al. (2001), Taylor et al. (2001)
Radio X-ray interaction produces an unusual
radio source with small bent lobes
The Chandra View
Perseus, Fabian et al. (2000)
Radio lobes fill X-ray cavities. Inner cavities surrounded by denser &
cooler gas. Holes appear to be devoid of ICM, Schmidt et al. (2002)
If we assume that the radio lobes are in pressure equilibrium with the
surrounding ICM, this is reasonable as no shocks are observed, then it is
easy to show that the lobes filled with B field and relativistc particles
have a smaller specific weight than surrounding ICM and should therefore
detach and rise buoyantly.
The Chandra View
Abell 2597, McNamara et al. (2001)
Expanded view of the central region of Abell 2597 after
subtracting a smooth background cluster model. The
8.44 GHz radio contours are superposed
VLA 1.4 GHz image of Abell 2597 at 11’’×6’’ resolution
Cavities in Abell 2597 are not coincident with bright radio lobes. Instead,
they are associated with faint extended radio emission seen in a deep Very
Large Array radio map. Ghost cavities are likely buoyantly rising relics of a
radio outburst that occurred between 50 and 100 Myr ago.
The overall picture
The core of CF clusters is far from being
relaxed, interaction with AGN is clearly present.
Radio Jets open up as radio lobes plowing their
way through the ICM.
Radio lobes are in pressure equilibrium with
ICM, no strong shocks observed.
Bubbles of low density plasma rise through
cluster atmosphere. Rising speed ~ a fraction of
vs, rise timescales of the order of 108 yrs, ~ to
cooling timescale
Higher density gas at rims, either dragged out
from center or compressed by uprising bubbles
Heating the flow
Feedback from AGN may provide a valid
heating mechanism
1.Mixing of relativistic plasma with ICM
2.Adiabatic expansion of bubbles as they
rise through cluster atmosphere
3.Dragging out of dense gas from core
which expands adiabatically becomes less
dense and eventually mixes with ambient
gas
Heating the flow
Feedback from AGN
• Gas cools everywhere in the core, cavities
are not seen everywhere
• Some cores do not host an active radio
source (e.g. A2597)
A possible solution is to assume a duty
cycle, note that the timescale over which
radio lobes evolve must be ~ a few times
shorter than the cooling time scale.
The energetic requirements, at least in
some cases could be met.
Heating the flow
Feedback from AGN
The total energy required to
quench a flow can be considerable.
Take total cooling energy,
determined from L(< rcool )•tHubble
for a set of clusters and compare
it with the total energy emitted
by an AGN over tHubble.
The more luminous cores imply
very large black-hole masses
Fabian et al. (2002)
Heating the flow
Conduction
Large heat reservoir in outer regions of cluster
Conduction and B Field. If the B field is chaotic over a
wide range of length scales as might happen with MHD
turbulence conductivity can be as high as 1/3 Spitzer
(Narayan & Medvedev 2001).
Large ΔT/T in cores => conduction should be efficent
Heating the flow
Conduction
Effective conduction
 eff
L ( r )

4r 2 dT / dr
Most clusters have
κeff between κs and
1/10κs
Conduction is
potentially important
Fabian et al. (2002)
Heating the flow
Conduction
From 3d structure it
is possible to derive
estimate of κeff(r)
For outer regions
κeff is between κs
and 1/3κs , for
innermost regions
κeff exceeds κs.
Conduction cannot
quench CF in the
very center
Ghizzardi et al. (2003), Voigt et al. (2002)
Heating the flow
Mixed models
Some authors have considered “mixed” models
Heating from conduction in outerparts
Heating from AGN within 20-30 kpc
Summary
The old revered Cooling-Flow model has fallen under
the weight of new observations obtained mainly with
XMM-Newton.
Chandra images show that the cores of clusters far
from being relaxed, are the sight of much dynamical
activity.
Interaction between radio lobes and ICM
Current efforts are concentrated on finding plausible
heating sources to balance the CF
AGN play a role in the innermost 20-30 kpc
and conduction may operate at larger radii