Filaments, Bubbles, and Weak Shocks in the Gaseous Atmosphere of M87

Draft version February 5, 2008

Preprint typeset using I4TgX style emulateapj v. 10/09/06

O

o

(N

;h

<
(N

>

m
oo
in

o
\o
o

Oh

6

%

FILAMENTS, BUBBLES, AND WEAK SHOCKS IN THE GASEOUS ATMOSPHERE OF M87

W. Forman', C. Jones', E. Churazov-', M. Markevitch', P. Nulsen', A. Vikhlinin'-', M. Begelman"*, H. Bohringer^

J. EILEK^ S. Heinz', R. Kraft', F. Owen** M. Pahre'
Short Title: Filaments, Bubbles, and Weak Shocks in M87

Draft version February 5, 2008

ABSTRACT

We present the first results from a 500 ksec Chandra ACIS-I observation of M87. At soft energies (0.5-
LO keV), we detect filamentary structures associated with the eastern and southwestern X-ray and radio arms.
Many filaments are spatially resolved with widths of ^ 300 pc. This filamentary structure is particularly striking
in the eastern arm where we suggest the filaments are outer edges of a series of plasma-filled, buoyant bubbles
whose ages differ by ~ 6 x 10^ years. These X-ray structures may be influenced by magnetic filamentation.
At hard energies (3.5-7.5 keV), we detect a nearly circular ring of outer radius 2.8' (13 kpc) which provides
an unambiguous signature of a weak shock, driven by an outburst from the SMBH. The density rise in the
shock is Pshockl Pq~ 1-3 (Mach number, M w 1.2). The observed spectral hardening in the ring corresponds
to a temperature rise Tsi,ock/To w L2, or M w L2, in agreement with the Mach number derived independently
from the gas density. Thus, for the first time, we detect gas temperature and density jumps associated with
a classical shock in the atmosphere around a supermassive black hole. We also detect two additional surface
brightness edges and pressure enhancements at radii of ~ 0.6' and ^^ 1'. The ^ 0.6' feature may be over-
pressurized thermal gas surrounding the relativistic plasma in the radio cocoon, the "piston", produced by the
current episode of AGN activity. The over-pressurized gas is surrounded by a cool gas shell. The ^ 1' feature
may be an additional weak shock from a secondary outburst. In an earlier episode, the "piston" was responsible
for driving the 2.8' shock.
Subject headings: galaxies: active - galaxies: individual (M87, NGC4486) - X-rays: galaxies

1. INTRODUCTION

M87 (NGC4486), the dominant central galaxy in the Virgo
cluster, hosts a 3.2 x 10'' M© supermassive black hole
(SMBH, Harms et al. 1994, Ford et al. 1994, Macchetto
et al. 1997) and its well-studied jet (e.g.. Sparks, Biretta, &
Macchetto 1996, Perlman et al. 2001, MarshaU et al. 2002,
Harris et al. 2003). On larger scales M87 has been the subject
of detailed radio observations showing remarkable structures
on scales up to 40 kpc (Owen et al. 2000; see also Hines et
al. 1989). In soft X-rays, M87 is the second brightest ex-
tragalactic source (after the Perseus cluster) and the emission
is dominated by thermal radiation from its ~ 2 keV gaseous
atmosphere (e.g., Gorenstein et al. 1977, Fabricant & Goren-
stein 1983, Bohringer et al. 2001, Matsushita et al. 2002,
Belsole et al. 2001, Molendi 2002).

M87 is a classic example of a "cooling flow" cluster (e.g.,
Fabian 1994). XMM-Newton and Chandra observations (e.g.,
David et al. 2001, Peterson et al. 2003 and references therein)
limit the amount of cooling gas in cluster cores to 10-20% of
that predicted by the standard cooling flow model and suggest
that the average core temperature does not fall below ^ 30%
of the gas temperature at large radii. The dramatic reduction

' Smithsonian Astropliysical Observatory, Harvard-Smitlisonian Cen-
ter for Astrophysics, 60 Garden St., Cambridge, MA 02138; wfor-
man @ gmail.com

- Space Research Institute (IKl), Profsoyuznaya 84/32, Moscow 117810,
Russia

^ MPI ffir Astrophysik, Karl-Schwarzschild-Strasse 1, 85740 Garching,
Germany

* JILA, 440 UCB, University of Colorado, Boulder, CO 80309

^ MPI fUr Extraterrestrischephysrk, Giessenbachstrae, 85748 Garching,
Germany

^ New Mexico Tech., Socorro, NM 87801
' University of Wisconsin, Madison, WI

* National Radio Astronomy Observatory, Socorro, NM 87801

in the amount of cooling gas, compared to the standard model,
requires a considerable energy supply to compensate for the
observed radiative losses. Thus, instead of seeking the repos-
itory of cold gas, we are now seeking the energy source to
(re)heat the radiating gas.

Suggestions for heating include thermal conduction (e.g..
Tucker & Rosner 1983, Bertschinger & Meiksin 1986, Gaetz
1989, David, Hughes & Tucker 1992, Zakamska & Narayan
2003, Voigt & Fabian 2004) and cluster mergers (e.g., Fu-
jita et al. 2004, but see Bohringer et al. 2004). One of the
most promising heating mechanisms has been SMBH out-
bursts. Strong shocks are rare and most heating probably
arises from less violent mechanisms to transfer AGN mechan-
ical power into heating the thermal plasma. Buoyant bubbles,
inflated by SMBH activity, were first modeled for M87 by
Churazov et al. (2001; see also Reynolds, Heinz & Begelman
2001, Kaiser & Binney 2003, De Young 2003). These bub-
bles can be a significant energy source for the cooling gas and
can explain much of the radio and X-ray morphology (Chu-
razov et al. 2001, QuiHs et al. 2001, Churazov et al. 2002,
Briiggen et al. 2002). The energy of jets and buoyant bubbles
can be transferred to the radiating gas by via shocks, gravity
waves, or turbulence (e.g., Churazov et al. 2002; Ruszkowski,
Briiggen & Begelman 2004a, b; Begelman 2004; Omma et
al. 2004; Roychowdhury et al. 2004, 2005; Heinz & Chura-
zov 2005; Mathews, Faltenbacher & Brighenti 2006; Binney,
Alouani Bibi & Omma 2007). From deep Chandra observa-
tions of the Perseus cluster, Fabian et al. (2006) showed that
pressure ripples could balance radiative cooling in the Perseus
core, if the viscosity is high. Dunn et al. (2005) showed that
buoyant bubbles may be a significant heat source in 70% of a
sample of cooling core clusters (see also Birzan et al. 2004).

As the nearest cooling core cluster with an active nucleus,
M87 is an ideal system for studying the energy input from the

Forman et al.

AGN to the hot, cooling gas. The first suggestion of struc-
ture in the gaseous halo of M87 and its possible relation to
the radio structure was made by Feigelson et al. (1987). Sub-
sequently, with ever-improving angular resolution, the rela-
tionship has been explored in more detail (Bohringer et al.
1995, Churazov et al. 2001, Belsole et al. 2001). In an ear-
lier 40 ksec Chandra observation. Young, Wilson & Mundell
(2002) reported X-ray cavities and edges in the surface bright-
ness profile. In a longer (^ 100 ksec) Chandra observation,
Forman et al. (2005) suggested that the edges or rings of en-
hanced emission at 13 and 17 kpc were likely shock fronts as-
sociated with AGN outbursts that began 1-2 x 10^ years ago.
From these same observations, Jordan et al. (2004) identified
a population of ~150 LMXBs, 40% of which are associated
with globular clusters.

We report here on the 500 ksec observation of M87 with the
Chandra Observatory. We focus on two major results:

• the soft emission (0.5-l.OkeV) from M87, especially in
the eastern and southwestern arms, forms a web of re-
solved filamentary structures (broader in extent than a
point source) of width ^ 300 pc (see Fig.|2]i. These fea-
tures, especially those seen in the eastern arm may arise
from a series of buoyant bubbles at different heights in
the atmosphere and different stages of evolution.

• the hard emission (3.5-7.5 keV) shows a ring of emis-
sion (see Fig.|7]i that is nearly circular with an outer ra-
dius ranging from 2.5' to 2.85' (11.6 - 13.3 kpc). This
ring of hard emission provides an unambiguous signa-
ture of a weak shock. Hardness ratios (deprojected)
show that the gas temperature in the ring rises from
^2.0 keV to ^2.4 keV implying a Mach number of
M -- 1.2 (shock velocity v = 880 km s"' for a 2 keV
thermal gas). At the shock, the density jump is 1.33
which yields a Mach number of 1.22, consistent with
that derived from the temperature jump. The age of the
outburst that gave rise to the shock must be approxi-
mately the radius of the shock divided the shock veloc-
ity, toiitbiirsi ^ Rshock/v shock = 14 Myr.

In addition to these results, the Chandra image provides a
wealth of information on structures with scales from the jet
to large scale cavities. We will discuss these other features in
more detail in future papers and focus here on the two new
results from the soft and hard band images from the deep ob-
servation.

2. OBSERVATIONS AND DATA PROCESSING

2.1. Chandra Observations

The new Chandra observations (OBSIDS 5826, 5827, 5828,
6186, 7210, 7211, and 7212) were taken at a variety of in-
strument roll angles from February to November 2005 using
the ACIS-I detector (CCDs 10-13) in Very Faint (VF) mode
to minimize the background. We reprocessed all observa-
tions applying the latest CTI and time dependent gain cali-
brations (acisD2000-01-29gain_ctiN0005.fits). We incorpo-
rated non-uniformity of the quantum efficiency and the time
and spatial dependence of the contamination on the optical
blocking filter. We performed the usual filtering by grade, ex-
cluded bad/hot pixels and columns, removed cosmic ray "af-
terglows", and applied the VF mode filtering. We also re-
processed two Faint mode ACIS-S OBSIDS (3717 and 2707)
and treated the front (S2) and back (S3) illuminated CCDs in-
dependently. We compared the images from the Very Faint

and Faint mode observations to verify that no artifacts were
introduced near the bright jet by use of the large 5x5 pixel
event regions in VF mode. We examined the data for back-
ground flaring and found moderate flaring in OBSIDS 3717
and 2707 (the back illuminated CCDs only) for which we ex-
cluded approximately half the duration. The total effective
remaining observation time varied over the field, since the ob-
servations were taken at many different roll angles. A typical
effective exposure time is ~ 525 ksec. The background files
(see Markevitch 2001 for details) were processed in exactly
the same manner as the observations.

Images were generated by combining the M87 observa-
tions after subtracting background and correcting for expo-
sure (which included all the effects mentioned above). When
generating merged images, to exclude the readout stripe from
the bright nucleus, we excluded rectangular regions extend-
ing from the M87 nucleus to the edge of the ACIS chip for
each pointing (since the orientation of the readout stripe varies
from pointing to pointing). The backgrounds were normalized
by time and small corrections were made using the counts in
the 10-12 keV energy band in each image and corresponding
background file. Exposure maps in each energy band were
computed separately and included vignetting and detector re-
sponse appropriate to the observed spectrum, assumed to be
emission from thermal gas with the redshift, galactic absorp-
tion and elemental abundance of the Virgo cluster For quan-
titative imaging and spectral analysis we relied on the more
recent ACIS-I observations alone.

To investigate the properties of the thermal gas that domi-
nates the Virgo core, we selected three energy bands that pro-
vide 1) an image of the soft thermal gas ((). 5-1.0 keV), 2) a
density map (1.2-2.5 keV), and 3) a pressure map (3.5-7.5
keV). Using the full Chandra response, we computed the pre-
dicted count rate for a unit volume of gas of unit density for
different energy bands as a function of gas temperature (see
Fig. [U. For the 0.5-1.0 keV band. Fig. [T] (upper dashed line)
shows that this energy band (Fig.|2l[3]i focuses on the the soft
emission.

For the 1.2-2.5 keV band. Fig. [T] (upper solid line) shows
that the count rate for a unit volume of unit density is inde-
pendent of gas temperature, for gas temperatures above about
1 ke V. If we express the count rate as

C oc / nle(T)dl

(1)

where n^, is the electron density, e{T) is the volume emissiv-
ity of the gas convolved with the Chandra response, and I is
the path length along the line of sight, the independence of
the 1.2-2.5 keV band emissivity on temperature implies that
C(1.2-2.5keV)cx/«2c//. Thus, the 1.2-2.5 keV band image
(Fig. [8^, b) is a "density" image (actually the square of the
density integrated along the line of sight).

The properties of the 3.5-7.5 keV band can be determined
by rewriting eq.[T]using the ideal gas law, p ex n^T to eliminate
the gas density. We find

Ccx p^eiT)/T^dl

(2)

The lower dashed curve in Fig.[T]is e(T)/T^ for a unit den-
sity and unit volume of gas. For gas temperatures from 1 to 3
keV, Fig.[T]shows that this expression varies only weakly with
temperature. Hence, eq.|2]in the hard band can be written as
C(3.5-7.5keV) ex J p^dl. Thus, the hard band (Fig. Eli, b)

Filaments, Bubbles, and Weak Shocks in M87

0.0001

Fig . 1 . — The emissivity of an optically thin plasma as a function of temper-
ature as observed by the Chandra ACIS-I detector (in units of counts s^' with
arbitrary normalization). For the 0.5-1.0 keV soft band (upper dashed line),
the emissivity strongly increases for temperatures below 1 keV. For the 1.2-
2.5 keV band (upper solid curve), the emissivity is nearly independent of gas
temperature for temperatures above ^ 0.75 keV. Hence, surface brightness
maps in this band are maps of f n^dl where n is the electron density and / is
the path length. In the 3.5-7.5 keV energy band (lower solid curve), the pho-
ton flux per unit volume F, can be expressed as F oc p^e(T)/T^ where t(T)
is the gas emissivity, p is the pressure, and T is the temperature (see text for
details). In the hard 3.5-7.5 keV energy band, e{T)/T^ (lower dashed curve)
depends only weakly on T (for temperatures from 1-3 keV), and hence, the
3.5-7.5 keV band image is approximately an image of the square of the pres-
sure (projected along the line of sight), i.e., total photon flux oc J p^dl. For
this figure, we assumed an abundance of 0.75 of solar. Note that for abun-
dances in the range of 0- 1 of solar, only the soft band curve (upper dashed
line) changes qualitatively. It shows no peak at low temperatures.

provides a "pressure" map for the gas (actually the square
of the pressure integrated along the line of sight). Hence,
bright regions in the hard band indicate pressure enhance-
ments which may be characteristic of shocks. In the analysis
and discussion below, we utilize these three energy bands to
investigate the gas around M87.

2.2. Spitzer Observations

M87 also was imaged with the IRAC instrument (Fazio et
al. 2004) on the Spitzer Space Telescope on 2005 June 1 1
under Guest Observer Program ID 3228. The exposure time
was 300 seconds for the central region of the galaxy (Spitzer
dataset # 0012673792) but only 150 seconds for a larger re-
gion surrounding the target (Spitzer dataset # 0010483200).
To create the images in each waveband, the Basic Calibra-
tion Data (BCD) were mosaiced using the Spitzer Science
Center software MOPEX (Makovoz et al. 2005). Optical
distortion was removed in the process, and the images were
subpixellated to 0.86 arcsec, which is a linear reduction by
a factor of V2 in each dimension. Fig. |6};, the 4.5/im im-
age divided by an azimuthally symmetric /3-model to remove
the strong stellar light gradient, shows the features associated
with the jet and counterjet. The synchrotron emission is visu-
ally clearest at A = 4.5/im because this bandpass falls between
the starlight emission (strongest at 3.6/im and decreasing with
wavelength) and the strong PAH emission features (which can
dominate the 5.8 and 8.0/im bands).

3. IMAGING RESULTS

Fig. 121 the soft (0.5-1.0 keV) band image (see also Fig. [3]),
shows that the familiar soft X-ray structures coincident with
the radio arms (Fig.|5h, b) are dominated by a filamentary web
that extends into a series of bubbles to the south of the east-
ern arm (Fig. ^p, |3^). The hard band (3.5-7.5 keV) image.
Fig. |7^, shows a shell with the characteristic pressure signa-
ture of a weak shock. In addition to these prominent features,
the images (see Fig.|5]for an overview of the large scale struc-
tures) show:

• cavities at the end of the jet that correspond to radio
emitting plasma (seen at 6 cm; see Fig. |6^ labeled
"Counter Jet Cavity" and "Jet Cavity" and compare to
6 cm image in Fig. ^p and 4.5/im Spitzer IRAC im-
age in Fig. |6]:). On the counter jet side, small scale
structures in the IR and radio (Fig. |6j5, c labeled with
arrows) coincide. The counter jet structures probably
lie beyond the end of the beamed jet. Shi et al. (2006)
found that several of the knots visible in the IRAC im-
ages, including the counterjet structures, have spectral
indices f^ ex i/"" with a ranging between 0.7 and 0.9.

• a very clear demarcation of the counter jet cavity
(marked "Counter Jet Rim" in Fig.|6t).

• an over-pressurized egg-shaped region (most likely
thermal gas since there is no corresponding non-
thermal radio emission, e.g., see Fig. ^p) of radius
~ 0.6' indicated by the lower arrow in Fig.|7j3 and outer
(cyan) contour in Fig. ^p. This over-pressurized gas is
being driven by the relativistic plasma in the radio co-
coon (Fig.|6}5 and magenta contours in Fig. [8})) that has
been produced in the current outburst.

• a rim of soft X-ray emission (see shell labeled "Cool
Rim" in Fig. ^p) surrounding the overpressurized gas
that represents gas swept up by the expanding "piston"
(relativistic plasma of the cocoon).

• a second weaker pressure ring at a radius of ^ 1' (indi-
cated by the upper arrow in Fig.|7j5) may be the result
of a secondary outburst, weaker than that which formed
the main 13 kpc shock.

• the very fine structure of the southwestern arm that may
indicate a braided/twisted structure (see region marked
"Filament" in Fig. |2] and also Fig. ^p). Although the
arm itself is resolved, there appear to be finer filaments
suggesting the importance of magnetic fields.

• a sharp boundary at the southern edge of the eastern arm
that does not correspond to any radio feature (marked
"Edge" in Fig. [5^).

• a new large X-ray cavity (labeled "Outer Cavity" in
Fig. |5h) at a radius of 4.8' -6.2' at an azimuth of
9 = 145° lying just beyond the eastern X-ray arms
(J2000, 12:31:07.0, H-12:27:0.0). The minimum energy
required to inflate the cavity is ~ 10^^ erg and its age
(rise time) is about 7 x 10^ years. This feature is coin-
cident with 90 cm radio emission (see Fig. ^p, the 90
cm radio image).

• ~ 500 point sources of which most are LMXBs and ^
100 are background AGNs (e.g. some are seen in Fig.|2]
andlSi).

Forman et al.

^^_l

Eastern Arm

^1

• .^_^- .

■^\^.-..M.^^^^^^^K^

Filament

2'

m 11/ :

Bubbles

/

m

^^M

1

SMHBi^^^B 2'

Fig. 2. — Images of the 9 ACIS-I pointings after background subtraction and "flat fielding" in the energy band 0.5-1.0 keV with 1 pix=3" at two different
scales, a) The image at left shows the very prominent eastern and southwestern arms (see also Fig.|5ji, b where the are labeled in both the X-ray and radio images)
b) The image at right shows a more detailed view of the tracery of filaments and suggests the similarity between the structures in the eastern and southwestern
arms. The long southwestern arm appears to be composed of several intertwined filaments (see location indicated by the label "Filament". The eastern arm can be
interpreted as a series of bubbles (several labeled in right panel) at different evolutionary stages as they rise in the atmosphere of M87. The filamentary structures
are very soft and are not apparent at energies above 2 keV (e.g., see Fig.|7j.

;J^'' .■"■■'■'

R2

Point

\ R3

'te.-::

Fig. 3. — (a) The central region of the 0.5-1.0 keV band image showing the multiple buoyant bubbles rising in the M87 atmosphere. A series of four "bubbles"
are outlined (see Fig.|2j) where the first three bubbles are indicated with arrows), (b) The 0.5-1.0 keV image of a portion of the southwestern arm indicating the
locations of projections and a comparison point source.

^

-10

-5
Dis

5

e (arcsec)

5 5

DisLance (arcse

10 10

DisLance (arcsec)

5 10 15 20 25

Dislance (aresec)

Fig. 4. — The 0.5-1.0 keV projections across the southwestern arm in the regions marked in Fig. [5}). The figures show the complex structure along the arm.
a) The first panel (region Rl in Fig.fSj)) shows two peaks which may arise either from two separate, close, filaments or from the limb-brightened edges of a
cylindrical filament, b) The second panel shows (region R2) a continuation of the southwestern arm consisting of two broad filaments, c) The third panel (R3)
shows three resolved structures (and a point source below, labeled "Point" in Fig.[3j)). Comparison to Fig. |3p shows that the westernmost filament appears to
wrap around the central filament, d) The final panel shows a projection across a boundary of the outer bubble of the series in the eastern arm along with a scaled
projection (lower histogram) of a point source 4' from the nucleus of M87, twice as far as the eastern arm filamentary region.

Filaments, Bubbles, and Weak Shocks in M87

In the following discussion of the new X-ray observation of
M87, the intimate relationship between the thermal gas and
the relativistic radio emitting plasma should be kept in mind.
Fig.|5]and |6]compare the X-ray and radio images and empha-
size this complex relationship.

3.1. The Soft Filamentary Web

The soft band image (Fig. |2] |3]l transforms our X-ray view
of M87. With the increased statistics of the longer observation
and the focus on the energy band below 1 keV, we see a web
of filamentary structures with length-to-width ratios of up to
^ 50. This web is especially pronounced in the eastern arm.

The eastern arm appears to be produced by a series of buoy-
ant bubbles. The youngest is associated with a clearly visi-
ble "bud" in the X-ray image (westernmost bubble labeled in
Fig.|2] marked "Bud" in Fig.|6^, and outlined circle in Fig.[3t;
see also Forman et al. Fig. lb, c). The "bud" has a radius of 0.9
kpc and is filled with radio plasma (Fig. |6h and|6j) compare
the X-ray and radio images). The minimum energy required
to inflate the "bud" is ~ lO^*" ergs for a spherical cavity (For-
man et al. 2005).

Three bubbles, including the "bud", are marked "Bubbles"
in Fig. ^p and four are highlighted in Fig. |3^. We estimate
the rise time of the four features using the approximation in
Churazov et al. (2000; section 3.2) to derive the velocity, v,-,se
of buoyant bubbles. Equating the buoyancy force to the drag
force and taking the galaxy mass from Cote et al. (2001), we
find Vrise ~ 400 km s"' over the whole inner 40 kpc of M87.
With an average separation of the features of approximately
0.5', the features differ in age by roughly 6 x 10^ years.

The series of bubbles, whose outer rims we only partially
see in Fig.|3^, may continue into the eastern arm which shows
additional filamentary features. These filaments may be in-
dependent buoyant bubbles elongated by the rise of the large
buoyant bubble that produced the torus (cap of the column
forming the eastern radio arm seen and labeled in Fig.lSji).

The X-ray filamentary structure is likely to be influenced
by magnetic filamentation or magnetic flux ropes. We know
from Faraday rotation data of the inner radio halo (Owen et
al. 1990) and radio emission of the large scale halo (Owen et
al. 2000), that the magnetic fields are highly filamented. The
radio filaments are likely magnetic flux ropes. The similarity
in scale and geometry of the X-ray and radio filaments argues
for the importance of magnetic fields in their formation, de-
spite their different emission mechanisms.

The X-ray filaments that form the eastern arm are individ-
ually very similar in width to the long, thin X-ray filaments
that form the southwestern arm (see Fig. |2] |3] |5^). Thus,
while the eastern arm is significantly broader than the south-
western arm, the structures which compose the eastern arm
are individually thin (see Fig. [3^, Fig. IHi. We have used the
marked regions in Fig. ^p to extract projections, shown in
Fig. m across the southwestern arm and a point source for
comparison. Fig. |4^, the projection closest to the center of
M87 (Rl in Fig.[3j5) shows a resolved arm (broader in extent
than a point source) with two subpeaks. These could be ei-
ther two separate filaments or the limb-brightened edges of
a cylindrical tube. Fig. ^p shows the onset of a second fila-
mentary structure about 10" east of the main filament. Fig.|4};
shows the continuation of the eastern filament, the main fila-
ment, and a third filament that in Fig. |3] appears to be twisting
around the central filament. Comparison of the point source
projections (Fig.H};, d) to those of the filaments shows that the
filamentary structures have widths (FWHM) of about 4", 300

pc.

The integrated spectra of the arms are well characterized
by a mean temperature of '--^ 1.5 keV (see Belsole et al. 2001
and the temperature map in Young et al. 2002). However,
Molendi (2002) argued that the spectral data in the arms, on
scales as small as 4" x 4", require multiple components. This
is at least partly caused by the blending of the thin filaments
with intra-filament emission. Also, it is likely that the fila-
ments are composed of still finer components that remain un-
resolved. We will analyze the detailed spectral properties of
the arms, using the Chandra data, in a later paper.

The Chandra observations highlight the differences seen in
the structure of the eastern and the southwestern arms. While
both arms show long, narrow, spatially resolved filaments, the
southwestern arm has only a single set of filaments, while the
eastern arm shows multiple sets of filaments and bubbles that
increase in scale with distance from the center In addition, the
radio plasma and X-ray thermal plasma are co-spatial in the
eastern arm while in the southwestern arm, the X-ray and ra-
dio emitting plasmas avoid each other with the X-ray plasma
interior to the surrounding radio plasma (see Fig.|5]l.

3.2. Weak Shocks

The hard band (3.5-7.5 keV) image, Fig.|7^, is no less strik-
ing than the soft band. We see, for the first time, the unam-
biguous signature of a shock - a nearly complete ring (indi-
cated by arrows in Fig. |7^) of higher pressure gas. The outer
radius of the hard ring varies slightly as a function of azimuth,
9 (measured counter clockwise from west). From the south-
west to the northeast {9 = -45° to 135°), over nearly 180°,
the outer radius is R„,irer = 2.85' (13 kpc). In the eastern and
southeastern directions {9 = 135°-315°), the outer edge of the
ring is slightly smaller, R„uier ^ 2.5' (1 1.5 kpc).

Evidence for two additional features is seen in Fig. |7j5 at
radii of 0.6' and ^1'. The region of radius 0.6' (outer bound-
ary indicated by the lower arrow) is egg-shaped with the nar-
rower end aligned with the jet (barely visible in Fig. ^).
This region of high pressure was first reported by Young et
al. (2002) who noted a surface brightness "front" at a radius
of 40". This over-pressurized region extends to the north, be-
yond the inner bright radio cocoon seen in the 6 cm radio
emission (see Fig. |6}5; Hines et al. 1989 and Fig. [S}?). We
suggest that the plasma in the radio cocoon is the "piston"
that mediates the activity of the central supermassive black
hole and drives the pressure waves into the surrounding gas.
The 0.6' over-pressurized region likely arises from the current
episode of AGN activity. In an earlier episode of activity, the
"piston" drove the 13 kpc shock. The second weak feature at
~ 1' (5 kpc), indicated by the upper arrow in Fig.|7j3, may be
the result of a secondary outburst, weaker than that which pro-
duced the primary 13 kpc ring. Both the 0.6' and 1' features
are seen in the surface brightness profiles discussed below.

In our previous study, we detected only the surface bright-
ness enhancement associated with the main 13' shock, but
lacked the statistical precision to measure changes in gas tem-
perature. We now detect the increased pressure and spectral
hardening characteristic of a weak shock and can measure the
temperature increase in the 13 kpc shock. Fig.|9h shows the
surface brightness profile in the 1.2-2.5 keV (upper curve)
and 3.5-7.5 keV (lower curve) energy bands azimuthally aver-
aged over an 80° wedge centered on north where the surface
brightness enhancement in the hard band lies at nearly the
same radial distance from the M87 nucleus. The most pro-
nounced feature is the shock between 2' and 3'. Also seen

Forman et al.

Cmt/ar C.

Outer Ring

Fig. 5. — (a) The relative deviations of tlie surface briglitness from a radially averaged surface brightness model i.e., [Data— Model]/Model over a broad energy
band (0.5-2.5 keV). The shock, an outer cavity beyond the eastern arm, a sharp edge in the eastern arm, and an outer partial ring are seen. We have excised
the prominent point sources from this image my substituting a local background, (b) The 90 cm VLA image from Owen et al. (1990) at the same scale as the
Chandra image shows the relationship between the X-ray and radio structures. In particular, the eastern and southwestern arms are apparent in both X-ray and
radio, the outer X-ray cavity corresponds to an enhancement in the radio and the outer ring (enhancement in X-ray image) lies just beyond the edge of the large
scale radio emission. The radio torus, at the end of the eastern arm, is connected by the arm to the center of M87. The torus and arm produces a "mushroom"
shaped structure (cap and stem).

Fig. 6. — (a) The 0.5-2.5 keV band full resolution (1 pix = 0.492 ) image of the entire data set after background subtraction and "flat fielding" of the center of
M87. b) The 6 cm VLA radio image from Hines et al. (1989) showing the radio jet, and the synchrotron emission from the cocoon. The cocoon of relativistic
plasma is the "piston" that mediates outbursts from the central SMBH and drives shocks into the sun'ounding X-ray emitting, thermal gas. c) IRAC 4.5^m image
divided by a /3 model to remove the strong gradient of emission from the galaxy light. Prominent X-ray features of the central region show the counterjet cavity
surrounded by a very fine rim of gas and cavities to the west and southwest of the jet after the jet passes the sonic point and the radio emitting plasma bends
clockwise. The innermost buoyant bubble (X-ray cavity, labeled "bud" in left panel) coincides with the radio synchrotron emission extending south from the
cocoon (center panel). The IRAC image shows the emission from the nucleus and the jet. The IR jet emission ends just before the feature "Jet Cavity" in the
X-ray image. On the counter jet side of the nucleus, two bright IR patches (labeled with arrows in the IRAC image) lie within a "C" shaped region. The two
bright IR patches coincide with brighter regions of 6 cm emission (also marked with arrows in the center panel) and associated with structures and r) in Hines et
al. (1989). The IR emission (and the coincident radio emission) lie at nearly 90° from the direction of the jet (in projection) and arise from unbeamed emission.

in the profiles is a decrease at 0.6' (most clearly seen in the
hard band between the fourth and fifth bins) which was noted
above as being over-pressurized gas produced by the present
episode of activity (and seen in the hard band image Fig.|7J)).
The second enhancement, possibly a weaker shock, is barely
seen at ^ 1'.

To derive quantitative parameters of the main 1 3 kpc shock,
we have deprojected the surface brightness profiles. In our
analysis we assume spherical symmetry but make no specific
assumption about the form of the underlying gravitational po-

tential. We first calculate the surface brightness (in a given
energy band) in a set of annuli (or wedges) and choose a cor-
responding set of spherical shells. The gas parameters are
assumed to be uniform inside each shell. Outside 8' the emis-
sivity was assumed to decrease with radius as a power law.
The projection can then be written as a convolution of the
emissivities in each shell with the projection matrix. The so-
lution for emissivities minimizing the x^ deviation from the
observed surface brightness in the set of annuli can be easily
found (see e.g. Churazov et al. 2003). The emissivities are

Filaments, Bubbles, and Weak Shocks in M87

Fig. 7. — Images from the 3.5-7.5 keV energy band at different scales. We have excised the prominent point sources from this image my substituting a local
background, a) The 3.5 to 7.5 keV band image of the 9 ACIS-I pointings after background subtraction and "flat fielding". The fine tracery seen in the soft band
image, Fig.|2]is replaced by a nearly azimuthally symmetiic ring (identified by arrows) of outer radius ~ 2.75' (12.8 kpc). As discussed in the text (see also
Fig.Q, this hard ring, a map of the square of the pressure projected along the line of sight, is the characteristic signature of a shock driven by an outburst from the
central SMBH. b) The central region of the hard band image showing two additional regions of enhanced pressure. An inner egg-shaped region of radius ~ 0.6'
(outer edge marked with the lower arrow) with the narrow end of the "egg" aligned with the jet (barely visible in the image) was generated during the cuirent
AGN outburst from M87's SMBH. An earlier episode of activity is responsible for the 13 kpc shock. A second region of enhanced pressure has an outer edge of
~ 1 ' (marked with the upper arrow) may be a weak, secondary shock.

2'

Fig. 8. — a) Image from 1.2-2.5 keV derived from 9 ACIS-I pointings after background subtraction and "flat fielding". As noted in the text (see also Fig.[TJ,
this image shows the the density squared integrated along the line of sight. The pressure shock is seen along with the two prominent arms. Many features, labeled
in other images are seen including the outer cavity, sharp edge, eastern and southwestern arms labeled in Fig. [3^, b and the "bud" labeled in Fig. |6] We have
excised the prominent point sources from this image my substituting a local background, b) Central region of the 1 .2-2.5 keV band image. The two inner contours
(magenta) show the 6 cm VLA synchrotron emission which represents the inner cocoon (the "piston" of relativistic plasma that mediates the energy output of the
SMBH and drives the shocks into the atmosphere of M87). This outer contour is derived from the 0.6' high pressure region seen in Fig.|7j) and represents the
over-pressurized gas, that is being driven by the "piston" during the cuirent outburst. The figure shows the cool rim (labeled "Cool Rim") that almost completely
surrounds the over-pressurized region. The cool rim is gas that has been displaced by the piston as it has expanded into M87's atmosphere. Also marked are the
bud (arrow denotes its outer cool shell) and the base of the southwestern arm.

then converted to electron densities using the Chandra spectral
response, evaluated for the spectrum with a given temperature
and abundance of heavy elements (see the discussion of un-
certainties introduced by the assumption of fixed temperature
and abundance at the end of this section). The resulting emis-
sivity profiles for the northern wedge are shown in Fig. |9j5.
We note that we are observing a disturbance in the gas that is
only approximately spherical. Hence, any shock features are
broadened by deviations from spherical symmetry.
We have used the deprojected emissivity profiles in the two

energy bands to derive the gas temperature profile. We ap-
plied the full Chandra response with all telescope and detec-
tor effects included (as enumerated above for correcting the
exposure maps) to thermal gas models. The resulting tem-
perature profile for the northern wedge is shown in Fig. [TO]
In addition to the surface brightness profiles and deprojection
(Fig. [9^, b) for the northern wedge, we repeated the analysis
for the full 360° range of azimuths. The results for both az-
imuthal ranges, shown in Fig. [TOl demonstrate a marked rise
in the gas temperature at ^ 2.5' radius. We note that there is

Forman et al.

Radius, arcmin

Radius, arcmin

Fig. 9. — (a) The azimuthally averaged (sector of width 80° centered on north) surface brightness profiles in the 1.2-2.5 keV (upper curve) and 3.5-7.5
keV (lower curve) bands show a shaip edge at ^^ 0.6' (most distinctly seen in the hard band as a decrease between the fourth and fifth data points), a moderate
flattening of the profile at about 1', and a strong excess at 2— 3'. (b) The deprojected emissivity profiles, derived from the azimuthally averaged surface brightness
profiles in the two energy bands 1.2-2.5 keV (upper curve) and 3.5-7.5 keV (lower curve). The deprojection shows the very pronounced feature at 2' — 3' which
represents the 13 kpc shock. In addition, for a spherical shock one expects a rarefaction region trailing the shock front (e.g. Zel'dovich & Raizer 2002, p. 100)
characterized by a density and temperature decrease below the upstream values. Such a rarefaction region is probably seen in Fig.|9}) at r ~ 1.5 — 2', where the
emissivity shows a clear depression. The coiTesponding temperature decrease can also be identified in Fig. 1101

a variation in temperature as a function of azimuth; the north-
ern region (large round symbols) appears slightly hotter than
the mean (small square symbols). As we discuss below, de-
spite these differences, the temperature jumps, and hence the
derived shock properties (using the Rankine-Hugoniot shock
jump conditions), are consistent.

To estimate the magnitude of the temperature rise, we fit
the data points in Fig. [TO] on either side of the shock. For
the 360° averaged profile, we consider three data points to
be affected by the shock (the peak and the one point on ei-
ther side) while for the profile to the north, we assume that
only two points are affected (the peak and the next point
at larger radius; for the northern profile with less data, the
uncertainties are larger and the radial range of each bin is
larger). The temperature rise is computed from a linear fit
to the temperature profile (using three points before the shock
and four points beyond) to the peak in each curve. For the
two sets of data, the 360° average and the northern wedge,
we find Tshock/To = 1.18 ±0.03 and 1.24 ±0.06 respectively
(Tq = 1.90 ±0.04 and 1.98 ±0.04 keV; all uncertainties are
statistical only). The temperature jumps, Tshock/To, are con-
sistent and yield a Mach number, M w 1.2. We have verified
that the deprojected hardness ratios yield correct temperatures
using direct spectral fits (following the deprojection approach
described in David et al. 2001). However, these detailed tem-
perature fits, although they agree with the deprojection analy-
sis results described above, have unacceptably large values of
X^ and hence the resulting error bars are difficult to interpret.
We defer a detailed discussion of spectral fitting to a later pa-
per, but only note that the best fit values are in good agreement
with those derived from the hardness ratios. Within these lim-
itations, the temperature jump at the shock front is consistent
with expectations for the Mach 1 .2 shock model of Forman et
al. (2005), using the Rankine-Hugoniot jump conditions for a
monatomic gas with 7 = 5/3.

We have also analyzed the density jump at the 13 kpc shock
to derive an independent estimate of the shock Mach num-
ber. Using the azimuthally averaged, deprojected emissivity
profile for the 1.2-2.5 keV band (Fig.|9]3), we derived the un-
perturbed density distribution at the location of the shock by
fitting a power law to three points before and four after the
shock. The resulting density jump is p shock / Pq ~ 1.33 ±0.02
which yields a Mach number M = 1.22 ±0.02 (statistical
errors only). This independent measurement of the shock
strength from the density distribution agrees with that derived
from the temperature measurement. Thus, for the shock at 13
kpc around M87, both density and temperature profiles exhibit
the properties of a classical shock in a gas with 7 = 5/3. The
age of the outburst that gave rise to the shock must be approxi-
mately the radius of the shock divided the shock velocity. This
age, to,„b„rs, ^ Rshock/v shock = 14 Myr, should slightly overesti-
mate the age of the shock since the velocity was higher in the
past (see also Forman et al. 2005 who used a simple model to
derive an estimate of t outburst ~ 1 1 Myr).

The gas temperature and densities derived above are subject
to systematic uncertainties. We mentioned above that projec-
tion and asphericity introduce uncertainties. Projection effects
will undoubtedly reduce the temperature peak while any rar-
efaction region behind the shock will increase it. While strong
shocks can dramatically alter the gas properties behind the
shock, the 13 kpc shock in M87 is weak and, hence, behaves
almost like an adiabatic wave leaving little disturbance in the
gas in the region behind the shock that we use to derive the
properties of the shock itself. Also, abundance gradients (see
Gastaldello & Molendi 2002 and Matsushita et al. 2003 for
detailed discussions of the heavy element abundance profiles
in M87) can affect the radial profiles in different energy bands.
The effects on temperature determinations for the bands we
have selected are small. Over the temperature range from 1-3
keV, abundance variations from to 100% of solar, change

Filaments, Bubbles, and Weak Shocks in M87

I

2 3

Radius, arcmin

Fig. 10. — The temperature profile derived from the 360° azimuthally av-
eraged data (small square symbols) and the wedge of width 80° (large round
symbols) centered on north. We used the deprojected emissivity profiles in
the 1.2-2.5 keV and 3.5-7.5 keV energy bands to derive the gas temperatures.
All telescope and detector effects have been included (as enumerated in the
text for correcting the exposure maps). There is a clear variation in the tem-
perature as a function of azimuth (the northern region appears hotter) which
arises partly from the absence of a perfectly azimuthally symmetric distii-
bution. The temperature rise associated with the shock is apparent in the
temperatures derived for both the full 360° average and the northern wedge.
We have not computed the temperatures at radii less than 1 ' due to the very
complex structures seen at smaller radii (see Fig.|6).

the temperature by at most ±0.1 keV. We expect any abun-
dance variations to occur gradually as a function of radius (for
our azimuthally averaged and deprojected profiles). Further-
more, over the radial range from l'-4' where we are deriv-
ing the temperature and density properties of the shock, the
abundance gradients are small (see Matsushita et al. 2003).
Detailed spectral fits, where the abundance and temperature
are both free parameters, confirm that any effects on the de-
rived temperatures from abundance variations are small. For
the gas density, the radial profile would also vary only mod-
estly. For the observed range in abundance in the radial range
from 2' to 3' (e.g., Gastaldello & Molendi 2001), the change
in the density would be less than 5%. Thus, uncertainties in
the shock temperature and density jumps are modest.

4. CONCLUSIONS

M87 and its gaseous halo provide a unique laboratory for
investigating the interaction between a SMBH and the sur-
rounding intracluster medium. Because of its low mean tem-
perature of about 2.5 keV, the energy range, 3.5-7.5 keV,
where the pressure enhancements of modest shocks are read-
ily observed (Fig. |7]l, lies within the Chandra energy band.
Using the 500 ksec Chandra observation of M87, we find:

• a direct image of a weak shock, a region of enhanced
pressure, at a radius of 13 kpc ((Fig.|^).

• a central over-pressurized region (radius 0.6', seen in
the hard band image Fig.|7]3, see also Fig.lSJ)) that sur-
rounds the "piston"(radio emitting relativistic plasma)
responsible for mediating the SMBH activity. The 0.6'
over-pressurized region is itself surrounded by a cool
rim of gas (Fig.HJ)). During an earlier phase of activity,
the "piston" drove a shock which we presently observe
at 13 kpc.

• at the shock, gas density and temperature jumps
{Pshock/po ~ 1.33 ±0.02 and r,,,„,,/7b » 1.18 ±0.03,
respectively), yield consistent values of the shock Mach
number, M w 1.2, characteristic of a classical shock in
a gas with 7 = 5/3.

• a web of soft filaments that may arise from a series of
buoyant bubbles produced by small outbursts from the
supermassive black hole at the center of M87. The fil-
aments, the rims of the buoyant bubbles, are resolved
by Chandra and have widths of approximately 300 pc,
although still finer, unresolved structures are likely to
exist.

M87 shows a remarkably "organized" and coherent struc-
ture in the eastern arm characteristic of a rising buoyant bub-
ble transformed into a torus in both the X-ray and radio im-
ages (see Owen et al. 2000 and Churazov et al. 2001) and a
second southwestern arm in both radio and X-ray. However,
although X-ray cavities are common in cooling core clusters
(e.g., Fabian et al. 2000, McNamara et al. 2000, Blanton et
al. 2001, Heinz et al. 2002, Mazzotta et al. 2003), the asso-
ciated thermal gas and radio emission characteristic of a large
buoyant bubble (Owen et al. 2000, Churazov et al. 2001;
"mushroom shaped" feature) are, so far, only seen in M87.

While progress has been made in understanding galaxy
clusters through both Chandra and XMM-Newton observa-
tions, questions remain. The nature of the filaments and im-
portance of magnetic fields in their topology requires addi-
tional modeling. Differences between the isothermal "shock"
in Perseus (Fabian et al. 2006) and the clear hardening of the
shock in M87 need to be understood and observations of ad-
ditional clusters are required to place the M87 and Perseus
observations in a larger context.

We acknowledge stimulating discussions with T. Bastian.
We thank the anonymous referee for helpful and construc-
tive comments which significantly improved the paper. This
work was supported by conti-acts NAS8-38248, NAS8-01 130,
NAS8-03060, the Chandra Science Center, the Smithso-
nian Institution, MPI fiir Astrophysik, and MPI fiir Ex-
traterrestrische Physik. M. Begelman acknowledges sup-
port from NSF grant AST-0307502. M.A.P acknowledges
NASA/LTSA grant # NAG5-10777. This work is based
partly on observations made with the Spitzer Space Telescope,
which is operated by the Jet Propulsion Laboratory, California
Institute of Technology under NASA contract 1407.

REFERENCES

Begelman, M. & Ruszkowski, M. 2005 _Royal__Socie tv of London
Transactions Series A, 363, 655-665 (and astro-ph/0403128 1

Begelman, M. 2004, in "Coevolution of Black Holes and oalaxies" from
the Carnegie Observatories Centennial Symposia. Cambridge University
Press, Ed. L. C. Ho, 374

Belsole, E. et al. 2001, A&A, 365, L188

Bertschinger, E. & Meiksin, A. 1986, ApJL, 306, LI

Binney, J. Alouani Bibi, F. & Omma, H. 2007, MNRAS, in press

Birzan. L., Rafferty, D., McNamara, B.. Wise, M., Nulsen, R 2004, ApJ, 607,

800
Blanton, E. L., Sarazin, C. L., McNamara, B. R., & Wise, M. 2001, ApJ, 558,

L15
Bohringer, H., Nulsen, R, Braun, R. & Fabian, A. 1995, MNRAS, 274, L67
Bohringer, H., et al. 2001. A&A, 365, L181
Bohringer, H., Matsushita, K., Churazov, E., Finoguenov, A., Ikebe, Y. 2004,

A&A, 416, L21

10

Forman et al.

Bruggen, M., Kaiser, C, Churazov, E., Ensslin, T. 2002, MNRAS, 331, 545
Churazov, E., Forman, W., Jones, C. & Bohringer, H. 2000, A&A, 366, 788
Churazov, E., Bruggen, M., Kaiser, C, Bohringer, H. & Forman, W. 2001,

ApJ, 554, 261
Churazov, E., Sunyaev, R., Forman, W. & Bohringer, H. 2002, MNRAS, 332,

729
Churazov, E., Forman, W., Jones, C. & Bohringer, H. 2003, MNRAS, 590,

225
Cote, D. et al. 2001, ApJ, 559, 828
David, L. et al. 2001, ApJ, 557, 546
David, L., Hughes, J. & Tucker, W. 1992, ApJ, 394, 452
De Young, D. 2003, MNRAS, 343, 719
Dunn, R., Fabian, A. & Taylor, C. 2005, MNRAS, 364, 1343
Fabian, A. 1994, ARA&A, 32, 277
Fabian, A., Sanders, J., Ettori, S., Taylor, G., Allen, S., Crawford, C,

Iwasawa, K., Johnstone, R. & Ogle, R 2000, MNRAS, 318, L65
Fabian, A., Sanders, J., Taylor, G., Allen, S., Crawford, C, Johnstone, R. &

Iwasawa, K. 2006, MNRAS, 366, 417 318, L65
Fabricant, D. & Gorenstein, R 1983, ApJ, 267, 535
Fazio, G. G., et al. 2004, ApJS, 154, 10

Feigelson, E., Wood, R Schreier, E., Harris, D., Reid, M. 1987, ApJ, 312, 101
Ford, H. et al. 1994, ApJL, 435, L27
Forman, W., Nulsen, R, Heinz, S., Owen, R, Eilek, J., Vikhlinin, A.,

Markevitch, M., Kraft, R., Churazov, E. & Jones, C. 2005, ApJ, 635, 894
Fujita, Y., Matsumoto, T. & Wada, K. 2004, ApJ, 612, L9
Gaetz, T 1989, ApJ, 345, 666
Gastaldello, F & Molendi, S. 2001, A&A, 375. L14
Gastaldello, F & Molendi, S. 2002, ApJ, 572, 160
Gorenstein, P. , Fabricant, D., Topka, K., Tucker, W., Hamden, F. 1977, ApJ,

216, L95
Harms, R. J., et al. 1994, ApJ, 435, L35
Harris, D., Biretta, J., Junor, W. , Perlman, E., Sparks, W. & Wilson, A., 2003,

ApJL, 586, L41
Heinz, S. & Churazov, E. 2005, ApJ, 634, L141

Heinz, S., Choi, Y, Reynolds, C. & Begelman, M. 2002, ApJ, 569, L79
Hines, D., Eilek, J. & Owen, F 1989, ApJ, 347, 713
Jordan, A., Cote, P., FeiTarese, L., Blakeslee, J. P., Mei, S., Memtt, D.,

Milosavljevic, M., Peng, E., Tonry, J.L., & West, M.J. 2004, ApJ, 613,

279
Kaiser, C. & Binney, J. 2003, MNRAS, 338, 839
Macchetto, F., Marconi, A., Axon, D. J., Capetti, A., Sparks, W., & Crane, P.

1997, ApJ, 489, 579

Makovoz, D., & Khan, 1. 2005, ASP Conf. Ser 347: Astronomical Data

Analysis Software and Systems XIV, 347, 8 1
Markevitch, M. 2001, Chandra calibration memo,

http://asc.harvard.edu/cal/ "ACTS," "ACIS Background" and

http ://cxc . harvard. edu/contrib/maxim/acisbg/COOKB OOK
Mai-shall, H., Miller, B., Davis, D., Perlman, E., Wise, M., Canizai'es, C,

Harris, D. 2002, ApJ, 564, 683
Matsushita, K., Belsole, E., Finoguenov, A. & Bohringer, H., 2002, A&A,

386, 77
Matsushita, K., Finoguenov, A. & Bohringer, H., 2003, MNRAS, 401, 443
Mathews, W., Faltenbacher, A. & Brighenti, F 2006, 638, 659
Mazzotta, R, Edge, A. & Markevitch, M. 2003, ApJ, 596, 190
McNamara, B. R., Wise, M., Nulsen, R E. J., David, L. P, Sarazin, C. L.,

Bautz, M., Markevitch, M., Vikhlinin, A., Forman, W. R., Jones, C, &

Harris, D. E. 2000, ApJ, 534, L135
Molendi, S. 2002, ApJ, 580, 815

Nulsen, R, McNamara, B., Wise, M. & David, L. 2005, ApJ, 628, 629
Omma, H., Binney, J., Bryan, G. & Slyz, A. 2004, MNRAS, 348, 1 105
Owen, R, Eilek, J. & Keel, W. 1990, ApJ, 362, 449
Owen, R, Eilek, J. & Kassim, N. 2000, ApJ, 543, 61 1
Perlman, E. S., Biretta, J. A., Sparks, W. B., Macchetto, F D., & Leahy, J. R

2001, ApJ, 551,206
Peterson, J. et al. 2003, Proceedings of "The Riddle of Coohng Flows i n

Clusters of Galaxies" eds. Reiprich, Kempner, Soker, astro-ph/03 1 0008 1
Quilis, v.. Bower, R., Balogh, M. 2001, MNRAS, 328, 1091
Reynolds, C, Heinz, S., Begelman, M. 2001, ApJL, 549, L179
Roychowdhuiy, S., Ruszkowski, M., Nath, B., & Begelman, M. 2004, ApJ,

615,681
Roychowdhury, S., Ruszkowski, M., & Nath, B. 2005, ApJ, 634, 90
Ruszkowski, M., Bruggen, M. & Begelman, M. 2004a, ApJ, 611, 158
Ruszkowski, M., Bruggen, M. & Begelman, M. 2004b, ApJ, 615, 675
Shi. Y Rieke, G., Hines, D., Gordon, K. & Egami, E. 2007, ApJ, 655, 781
Sparks, W. B., Biretta, J. A., & Macchetto, F 1996, ApJ, 473, 254
Tucker, W. & Rosner, R. 1983, ApJ, 267, 547
Voigt, L. & Fabian, A. 2004, MNRAS, 347, 1130
Werner, M. W., et al. 2004, ApJS, 154, 1

Zel'dovich, Y & Raizer, Y 2002 "Physics of Shock Waves and High-
Temperature Hydrodynamic Phenomena" (NY: Dover Publications)
Young, A., Wilson, A. & Mundell, C. 2002, ApJ, 579, 560
Zakamska, N. & Narayan, R. 2003, ApJ, 582, 162