CLUSTER SOFT EXCESS: NEW FACES OF AN OLD ENIGMA

CLUSTER SOFT EXCESS: NEW FACES OF AN OLD ENIGMA

Richard Lieu University of Alabama, Huntsville

Jonathan Mittaz Max Bonamente

The discovery of cluster soft excess as extra photon emission in the 0.2 – 0.5 keV range above the level expected from the low energy tailof the virialized intracluster gas at X-ray temperatures was made bythe EUVE mission in 1995

Coma Cluster in the EUV

Coma Cluster 6’ – 9’ ROSAT and EUVE DS

Solid line is the expected emission spectrum of the hot ICM at kT = 8.7 +/- 0.4 keV and A = 0.3 solar, as measured by ASCA.

ROSAT PSPCEUVE

XMM-Newton has definitively confirmed the cluster soft excess

Fit with a single temperature acrossthe whole 0.3-7 keV band showssignificant residuals indicating acluster soft excess

Fit to the hot ICM (1-7 keV) showingthe cluster soft excess at energiesbelow 1 keV. The excess is seen at > 20% level above the hot ICM model

45.12

Fit to PKS2155-304 to demonstrate the systematic errors in extrapolating to lower energies from a 1-7 keV fit. Note the maximum residuals are at the 8% level, much less than residuals seen showing the presence of a cluster soft excess

Note for A3112, cluster soft excess is not a background effect

One of the recent claims regarding the CSE is the detection ofOVII line emission

Kaastra et al. (2003)

AS1101 (2’-5’) with ICM model (fitted from 2-7 keV) and backgrounds

Intrinsic background

Kaastra sky average background

SOFT EXCESSREMAINS ROBUST(after subtractingthe higher background)

Isothermal modelkT = 3.08 keVA = 0.194 solar

However, the importance of good background subtraction cannot be overstated

AS1101 10’-13’ background spectrum. OVII+OVIII linesconsistent with Galactic emission and not associated with the cluster redshift (z=0.058)

OVII+OVIII lines positioned at the cluster redshift in AS1101 background

And little evidence for redshifted OVII line emission

Line fit to the OVII+OVIII complex with no constraint on the energy of the line

Line completely consistent with zero redshift i.e. Galactic origin

As yet no strong evidence for redshifted line emission

Best thermal candidate is emission from the WHIM (Warm Hot Intergalactic Medium)

Gas between KTK 75 1010 from a cosmological simulation (from R. Cen’s homepage)

The CSE and thermal models

There is emission from the WHIM but it is very weak

Maximum emission forCells with T < 1keV

Emission from atypical sightline

Simply not enough material to give the observed cluster soft excess

(Mittaz et al. 2004)

Non-thermal interpretation of the Cluster Soft Excess

Abell 1795Region Power-law

Luminosity

(0.2-1leV)

( )

Photon Index

Hot Gas kT

(keV)

Relativistic

Electron

Energy

(total ergs)

Relativistic

Electron pressure

( )

Gas Pressure

( )

Gas density

( )

0’-1’

2’-5’

3/ cmergs

3/ cmergs 3cm

4310

42107.9

sergs /

30.2

81.2

0.5

3.6

60102.1 601018.1

12108 14107

101048.1 11107.1

21085.1 3107.1

In the center 0’-1’ region, the central galaxy may quite easily supply cosmic rays of total electronenergy of a few ergs. As mentioned before, the ratio of proton to electron pressure in theCR population is a few x100. Thus the CR protons can obtain (or surpass) equipartition with the gas

REASON FOR THE ABSENCE OF A COOLING FLOW?

In the outer parts, the CR’s have to come from supernovae within the member galaxies. Based onthe best fit adundance of 0.32 solar for the 2’-5’ region, the amount of iron in the gas is SN’s in the past. Assuming each SN outputs ergs of CRs, one estimates ergs,mostlyin protons. Within the < 3Gyrs of loss time against inverse-Compton scattering these protons produce ergs of secondary electrons: NOT ENOUGH

6010

sunM9107.5 10107.5

50103~ 61107.1)'5'2( CRE

58108

Leading annihilation channels

Rate

Cosmological relic density

Cross-section =0.3, h=0.5)

annihilation in galaxy clusters

Secondary electrons with Ee M are produced in situ

[Colafrancesco & Mele 2001, ApJ, 562, 24; Colafrancesco 2004, A&A, 422, L23 ]

EUV/soft X-ray ICS emission is produced by the secondary electrons - created by annihilation -which scatter the CMB photons (Colafrancesco 2004)

The EUV/soft X-ray excess in Coma is best fitted by a neutralino with:

scmV A /104 326

GeVM 30

quite independent of the model.

The EUV/soft X-ray excess provides the bound

121327 )(10330

hscm

VGeVM A

121327 )(103

30

hscm

VGeVM A

CONCLUSION

•XMM-Newton Observations of Clusters in the 0.4-7 keV range were made. They confirm ‘beyond reasonable doubt’ the existence of a soft excess first noted by the EUVE and ROSAT teams. In particular the soft excess:

•is strong and in XMM extends to 1-2 keV energy, at a level beyond the uncertainties of the detector responses•is seen in regions where the cluster low energy flux is 10-100 times above background•may be explained as a Galactic absorption anomaly only by asserting that the HI column should actually be zero.•At the outskirts of clusters, the OVII line claimed by Kaastra Lieu et al. is due to Galactic foreground emission•Non-thermal models require an unrealistic shock acceleration mechanism• Signature of dark matter not excluded

The discovery of cluster soft excess as extra photon emission in the 0.2 – 0.5 keV range above the level expected from the low energy tailof the virialised intracluster gas at X-ray temperatures was made bythe EUVE mission in 1995

Coma Cluster in the EUV

3.0,4.07.8 AkeVkT as measured by ASCA

R

GMv

R

GMmmvVT 202

Virial speed of mass in the cluster is given by

Virial theorem

For a rich cluster like Coma,

skmvMpcRMM sun /10002,105 14

A proton moving in this potential has kinetic energy

kTkeVvmp 52

1 2

Coma Cluster 6’ – 9’ ROSAT and EUVE DS

Solid line is the expected emission spectrum of the hot ICM at kT = 8.7 +/- 0.4 keV and A = 0.3 solar, as measured by ASCA.

ROSAT PSPCEUVE

XMM-Newton fit to Coma 0’-5’ region

Fit across the whole XMM-Newton band with 1 temperature mode092.0086.0336.7

kT014.0013.0176.0

A

Limiting the fit to Coma between 2 – 7 keV yields a more consistent fit26.025.000.9

kT018.0018.0197.0

A

21.023.079.8

kT019.0024.0219.0

A

This temperature is inconsistent with other estimates e.g. ASCA

Since the temperature of the full band fit is low – suggests the presenceof soft excess emission < 2keV. If we model this with a power-law

27.013.022.2

This model has a temperature consistent with the 2-7keV/ASCA fitF-Test indicates a significant improvement at > 99.99% over singletemperature fit.

XMM-Newton spectra of the Coma Cluster 0’-5’

EPIC PN Spectrum EPIC MOS1 + MOS2 Spectra

(Nevalainen et al., 2003, ApJ, 584, 716)

Solid line is the best fit single temperature hot ICM emission model (kT = 9.6 keV, A = 0.22 solar) as obtained by fitting the XMM data between2.0 and 7.0 keV

IS THE SOFT EXCESS AT A LEVEL ABOVE THE KNOWN SYSTEMATICUNCERTAINTIES IN THE CALIBRATION OF XMM?

XMM Calibration uncertainties for PM/MOS ~ 5%Soft excess is above the calibrational uncertainties

Taken from “EPIC status of calibration and data analysis” Kirsch et al.XMM-SOC-CAL-TN-0018

Could the cluster soft excess be due to an

incorrect subtraction of the background?

BUT THE BACKGROUND IS ~30 TIMES BELOW COMA

COULD THE SOFT EXCESS BE DUE TO AN INCORRECT ESTIMATIONOF THE FOREGROUND GALACTIC HI ABSORPTION ?

Coma Cluster

Average sky backgroundat high latitudes

Bregman et al., 2003, ApJ, 597, 399

BUT THE SOFT EXCESS WITNESSED BY XMM-NEWTON IS ABOVETHE RANGE OF THIS PLOT!

DO YOU BELIEVE IN ZERO GALACTIC COLUMN?

Best 3 Temperature model for the soft excess of Coma’s 6’-9’ region

IS THIS A PHYSICALLY SENSIBLE MODEL?

Physical constraints on the model

For intracluster origin of the WHIM

w

hhwhotwarm T

TcmnnPP )10/( 33

If we take 3210,10 cmnTT wwh

Radiative cooling time is important

yrscmnKT w1335.069 )10/()10/(106

For 326 10,10~ cmnKT ww years8106

WHAT SUSTAINS THE WARM GAS AGAINST SUCH RAPID RADIATIVECOOLING?

Thermal (mekal) model kT K610

Giant ¼ keV Halo centered at Coma (as detailed by the ROSAT sky survey)

ROSAT/PSPC Radial surface brightness of Coma(Bonamente, Joy, & Lieu, 2003, ApJ, 585, 722)

¼ keV ¾ keV

ROSAT/PSPC data of the Coma cluster (50’-70’ annulus)

X-ray thermal model (kT~8keV) Fitting the excess with a 2nd component

Hot ICM + power-law

Hot ICM + warm component

Unlike the cluster core, strong soft excess at the outskirts of Coma. Statistically the thermal model is preferred (to a power law).

The warm gas here may be part of the WHIM (e.g. Cen & Ostriker 1999)not in physical contact with the hot ICM. XMM-Newton confirmation ofthe Coma soft excess halo.

Coma cluster 0.5 – 2 keV with XMM-Newton pointings

XMM-Newton spectrum of the Coma 11 region(Finoguenov, Briel & Henry, 2003, A&A, 410, 777)

What are the PROS and CONS if a thermal intercluster filament modelfor the outer soft excess?

If emission is caused by warm intercluster filaments:Can avoid the need for pressure balance with the hot intracluster gas so the warm gas density can be 3210 cmnw

BEWARE of constraints specific to this scenario. For example the emission measure EM is given by

LAnEM 2

Which is fixed by observations (i.e. lower n implies higher L). For a given valueof EM, the line-of-sight COLUMN DENSITY of warm filaments is

nnL

1~ which can be observationally constrained!

XMM-Newton PN Spectrum of X-Comae (z=0.091 28 arcminutes from Coma center). Shown is a continuum model together with OVII 21.6A line with an equivalent width of 28eV

This can be used to set a limit on the absorbing column (Nicastro et al. 1999)222108.1 cmNH

Coupled with the EM of the soft excess which is non-negotiable this means34105.5 cmn

For a warm gas appropriate for the cluster halo (A=0.1, kT = 0.2 Bonamente et al. 2003,Finoguenov et al. 2003) this gives a possible lifetime of or yrsw

91012.2 18.0z

Arguments against thermal origin of the softexcess at cluster outskirts

•Minimum required warm gas column density is contradicted by absorption line measurements of quasar spectra

•Is the outer soft excess really associated with the Warm Hot Intergalactic Medium?

Cen & Ostriker, 1999, ApJ, 514, 1

Comparing emission from the WHIM with XMM-Newton observations

Emission weighted temperature map of one projection of the simulation showing the simulatedCluster at the top right hand corner (Mittaz et al. 2004)

Simulated ‘Coma’ X-ray spectrum from Cen’s datacube for a typical sightline in region 5’-13’ assuming the redshift of Coma

No cluster soft excess seen

(Mittaz et al. 2004)

Arguments against thermal origin of the softexcess at cluster outskirts

•Minimum required warm gas column density is contradicted by absorption line measurements of quasar spectra•Is the outer soft excess really associated with the Warm Hot Intergalactic Medium? •Are the OVII emission lines found on top the soft excess spectra at the outskirts of some cluster real?

NO!

Thermal line detections from Kaastra et al., 2003, A&A, 397, 445

¼ ROSAT all sky survey image around AS1101

Arguments against thermal origin of the softexcess at cluster outskirts

•Minimum required warm gas column density is contradicted by absorption line measurements of quasar spectra.•Is the outer soft excess really associated with the Warm Hot Intergalactic Medium? NO!

•Are the OVII emission lines found on top of the soft excess spectra at the outskirts of some clusters real? NO!

CLUSTER SOFT EXCESS DIAGNOSTICS

•The soft excess outside clusters’ cores is now open toquestion, because the effect is real.•Inside a cluster’s core the thermal model is even moredisqualified. If the origin is outlying filaments seen inprojection, the required column density will be enormous.if intracluster warm gas – problem with cooling time.

THERE IS, HOWEVER, AN ALTOGETHER DIFFERENTAPPROACH.

A1795 Goodness of Fit0 – 1’ region (MOS1 + MOS2)

027.0199.4 kT 789,316.12

Null probability9102.5

Broad band (0.4-7keV) fit

Limited (2-7keV) fit

12.001.5 kT

2 – 5’ region (MOS1 + MOS2)

Broad band (0.4-7keV) fit

05.052.5 kT 835,095.12

Null probability21094.2

Limited (2-7keV) fit

21.026.6 kT

A1795: single temperature fit (2-7 keV) for two annuli

Background 100x below clusterBackground 10x below cluster

kT = 4.88 +/- 0.08 keV A = 0.43 +/- 0.02 solar

kT = 6.05 +/- 0.15 keV A = 0.27 +/- 0.03 solar

Comparison of backgrounds comparing Lumb data set (red) and Lumb + double subtraction background in the offset field 2’-5’

CONCLUSION

•XMM-Newton Observations of Clusters in the 0.4-7 keV range were made. They confirm ‘beyond reasonable doubt’ the existence of a soft excess first noted by the EUVE and ROSAT teams. In particular the soft excess:

•is strong and in XMM extends to 1-2 keV energy, at a level beyond the uncertainties of the detector responses•is seen in regions where the cluster low energy flux is 10-100 times above background•may be explained as a Galactic absorption anomaly only by asserting that the HI column should actually be zero.

•At the cluster cores, a population of cosmic rays accelerated at the ‘central engine’ could account for the phenomena. The proton pressure of these CR’s attains equipartition with the hot ICM i.e. sufficient to choke a cooling flow.

•At the cluster outskirts the soft excess does not appear to be thermal in origin, because:

•requisite column density of warm gas is inconsistent with the spectrum of background quasars;•the excess is 100-1000 times brighter than the level predicted by WHIM models;•the OVII line claims by Kaastra Lieu et al. is due to Galactic foreground emission.

•On the other hand, non-thermal models also do not seem to be relevant here, because there is no obvious source of relativistic particles.

THUS, AFTER A FULL DECADE OF RESEARCH, WHILE CLUSTERSOFT EXCESS HAS BECOME AN OBSERVATIONAL REALITY, ITREMAINS A THEORETICAL ENIGMA.

Simulation of A1795 (2’-5’) hot ICM with a thermal soft excess in Astro-E. Model is a single temperature fit to the spectrum. Lines due to the soft excess component can be clearly seen. Exposure time is 40 ksec.

Backgroundz = 0

It can also be demonstrated that the reduction in best-fit temperature whenthe 0.4 – 2 keV band is included in the fitted procedure is NOT because A hard tail is present in the Coma spectra (thereby biaing upwards the temperature inferred from the 2-7 keV band). Reason is that when the 2-7 keV data are modeled with hot isothermal bremsstrahlung plus a powerlaw (to account for any possible hard tail) the resulting improvement ismarginal

740.0

1098,7.812

21.0

60.9

2

2

A

kT

761.0

1100,6.837

20.0

0.9

2

2

A

kTMekal + Power-law Mekal only

F-Test gives likelihood of improvement at only 32%

Coma XMM-Newton MOS1+2 and PN fits to 0’-5’ region

Single temperature Single temperature + Power-law

kT=7.34 keV kT=8.79 keV

Assuming a soft excess in Coma dramatically improves the fit

Coma in the Radio

Coma has relativistic electrons!

Coma Hard tail : data from Fusco-Femiano

Cosmic rays in equipartition with the hot ICM?Where do these particles come from?

Still, at least within the central 20’ radius of Coma the non-thermalapproach to cluster soft excess does not suffer from any obvioussetbacks.

Detection of very extended EUV excess in A2199 (Lieu et al., 1999,ApJ, 527, L77

Most of the CR pressure is in the protons

In above plots curve 1: 0 Gyrs, 2: 0.5 Gyrs, 3: 1.5 Gyrs 4: 3.0 Gyrs, 5: 4.5 GyrsInitial pressure ratio all protons : soft excess emitting electrons ~ 200:1.After ~ 3 Gyrs this ratio > 1,000:1

• At the core of a cluster, the central galaxy is an obvious source of cosmic rays and pairs. Normally they are transported to larger radii by Bohm diffusion – very slow.

• At the outskirts, one can only expect cosmic rays (i.e. no pairs) generated in-situ by supernova remnants within member galaxies of the same region.

Do we have enough SN’s to account for therequired level of cosmic rays’ presence? This canbe answered by looking at the iron content of the gas.

We will use the Newton data of Abell 1795 to test the non-thermal scenario.

XMM-Newton fit to Coma 0’-5’ region

Fit across the whole XMM-Newton band with 1 temperature mode092.0086.0336.7

kT014.0013.0176.0

A1733,2.17702

02.12

Limiting the fit to Coma between 2 – 7 keV yields a more consistent fit26.025.000.9

kT018.0018.0197.0

A1100,6.8372

761.02

21.023.079.8

kT019.0024.0219.0

A1731,2.14042

811.02

This temperature is inconsistent with other estimates e.g. ASCA

Since the temperature of the full band fit is low – suggests the presenceof soft excess emission < 2keV. If we model this with a power-law

27.013.022.2

This model has a temperature consistent with the 2-7keV/ASCA fitF-Test indicates a significant improvement at > 99.99% over singletemperature fit.

Non-thermal interpretation of the cluster soft excess

Hwang, C.-Y., 1997, Science, 278, 191Ensslin, T.A. & Biermann, P.L., 1998, A&A, 330, 20Sarazin, C.L. & Lieu, R., 1998, ApJ, 494, L177

Proposed the origin of the cluster soft excess emission as due to inverse-Compton scattering between intracluster cosmic rays (relativistic electrons with Lorentz factors of a few hundred)and the cosmic microwave background

HOW LARGE A COSMIC-RAY (CR) POPULATION DO WE NEEDTO ACCOUNT FOR THE SOFT EXCESS BRIGHTNESS?

NB. Center can be e+/e- pairs, but outside has to be CR’sfrom supernova events.