A giant sloshing cold front in Abell 2142 (+Bonus)

Dominique Eckert

Department of Astronomy, University of Geneva

Collaborators: Mariachiara Rossetti, Sabrina De Grandi, Fabio Gastaldello, Simona Ghizzardi, Elke Roediger, Silvano Molendi, Larry Rudnick, Damon Farnsworth

D. Eckert January 15, 2013 The case of Abell 2142

A2142 (z=0.09): first cluster where CFs have been observed (Markevitch et al. 2000)

Markevitch & Vikhlinin 2007

D. Eckert January 15, 2013 Tittley & Henriksen (2005) classify the CFs in A2142 as sloshing →

Owers et al. 2011 ... However this interpretation is hard to reconcile with the velocity distribution

The riddle of the dynamical state

In the discovery paper M00 interpreted A2142 as a merging cluster and the CFs as remnant cores

D. Eckert January 15, 2013 Tittley & Henriksen (2005) classify the CFs in A2142 as sloshing →

MINOR MERGERS IN A2142 AND RXJ1720 9

The riddle of the dynamical state Figure 4. Phase space diagrams used for defining cluster membership forA2142(left panel)andRXJ1720(right panel). The red curves show the cluster limits in phase-space and are determined from the cluster mass profile (see the text for a description). Black crosses In the discoveryshow paper cluster membersM00 interpreted which lie on the red A2142 sequence, as while a mergingbluestarswithinthephasespacelimitsshowthoseclustermemb cluster ers which lie blueward of the red sequence. Red boxes show fore- and background . Red boxes filled with blue show non-members which and the CFs asare remnant bluer than the cores cluster red sequence.

Owers et al. 2011 ... However this interpretation is hard to reconcile with the galaxy velocity distribution

D. Eckert January 15, 2013

Figure 5. The velocity distribution for all 956 cluster members (top left panel), 463 members with 1500

In the discovery paper M00 interpreted A2142 as a merging cluster and the CFs as remnant cores MINORMERGERSINA2142ANDRXJ1720 17

Owers et al. 2011 ... However this interpretation is hard to reconcile with the galaxy velocity distribution Tittley & Henriksen (2005) classify the CFs in A2142 as sloshing → D. Eckert January 15, 2013

Figure 12. Top panel: The black circles show regions selected for the KMManalysesbasedonsubstructurerevealedinFigure7and 10. Black contours show galaxy surface density as in Figure 7.Thestarsarecolorcodedtomatchthecorrespondingvelocity components shown in the lower panels. Black points reveal the positions of the remaining cluster members. The riddle of the dynamical state

Unlike all known sloshing clusters, A2142 is not a cool core 2 (K0 = 68 keV cm )

Ghizzardi et al. 2010

D. Eckert January 15, 2013 Sloshing ! Concentric edges as in simulations ! Optical subcluster identified as candidate perturber (Matt’s talk) % Not a relaxed, cool core cluster, flat entropy profile % Hosts a giant radio halo like major merger clusters (Damon’s talk) The origin of the CFs is still uncertain! We obtained a new 50 ks XMM observation to constrain the dynamics→

The riddle of the dynamical state

Pros and cons Remnant core ! Cometary shape of the edges ! Overall elongation in X-rays and in the optical in the SE-NW direction ! Not a relaxed, cool core cluster % Problem with the optical distribution: where are the galaxies associated to the NW CF?

D. Eckert January 15, 2013 The origin of the CFs is still uncertain! We obtained a new 50 ks XMM observation to constrain the dynamics→

The riddle of the dynamical state

Pros and cons Remnant core Sloshing ! Cometary shape of the edges ! Concentric edges as in ! Overall elongation in X-rays simulations and in the optical in the ! Optical subcluster identified SE-NW direction as candidate perturber ! Not a relaxed, cool core (Matt’s talk) cluster % Not a relaxed, cool core % Problem with the optical cluster, flat entropy profile distribution: where are the % Hosts a giant radio halo like galaxies associated to the major merger clusters NW CF? (Damon’s talk)

D. Eckert January 15, 2013 The riddle of the dynamical state

Pros and cons Remnant core Sloshing ! Cometary shape of the edges ! Concentric edges as in ! Overall elongation in X-rays simulations and in the optical in the ! Optical subcluster identified SE-NW direction as candidate perturber ! Not a relaxed, cool core (Matt’s talk) cluster % Not a relaxed, cool core % Problem with the optical cluster, flat entropy profile distribution: where are the % Hosts a giant radio halo like galaxies associated to the major merger clusters NW CF? (Damon’s talk) The origin of the CFs is still uncertain! We obtained a new 50 ks XMM observation to constrain the →dynamics

D. Eckert January 15, 2013 Abell 2142 as seen by XMM

A2142 was observed by XMM for 50 ks on July 13, 2011

D. Eckert January 15, 2013 We discovered a third discontinuity 1 Mpc SE of the cluster core ∼

Abell 2142 as seen by XMM

The two inner CFs are clearly visible

D. Eckert January 15, 2013 Abell 2142 as seen by XMM

The two inner CFs are clearly visible We discovered a third discontinuity 1 Mpc SE of the cluster core ∼ D. Eckert January 15, 2013 SB profiles in NW and SE directions SB discontinuity detected 10 arcmin from the∼ cluster center

A2142 ROSAT/PSPC data

The third discontinuity was not unexpected to us Clearly visible in the ROSAT residual image

D. Eckert January 15, 2013 SB discontinuity detected 10 arcmin from the∼ cluster center

A2142 ROSAT/PSPC data

The third discontinuity was not unexpected to us Clearly visible in the ROSAT residual image SB profiles in NW and SE directions

D. Eckert January 15, 2013 A2142 ROSAT/PSPC data

The third NW sector SE sector discontinuity was not NW CF SE CF unexpected to us 10-1 ]

Clearly visible in the -2 ROSAT residual 10-2

image arcmin -1 SB profiles in NW 10-3 and SE directions

SB discontinuity SB [counts s 10-4 detected 10 arcmin ∼ from the cluster 1 10 center Distance [arcmin] Rossetti et al. subm.

D. Eckert January 15, 2013 Third discontinuity: XMM/PN SB profile

SE sector

-1

] 10 -2 arcmin -1 SB [counts s

10-2

10 Radius [arcmin] Rossetti et al. subm.

Density jump: nin/nout = 1.79 0.06 Distance to the center: 947 3± kpc Linear size of the discontinuity:± 1.2 Mpc

D. Eckert January 15, 2013 The shock hypothesis is excluded at > 5σ The third discontinuity is the largest and outermost cold front ever detected

Third discontinuity: temperature jump

+0.45 +2.4 kTin = 6.13 0.37 keV kTout = 9.0 1.5 keV − −

Source

Source

NXB

NXB

Halo CXB CXB QSP

QSP Halo

Local Local Bubble Bubble

Rossetti et al. subm.

D. Eckert January 15, 2013 Third discontinuity: temperature jump

+0.45 +2.4 kTin = 6.13 0.37 keV kTout = 9.0 1.5 keV − −

Source

Source

NXB

NXB

Halo CXB CXB QSP

QSP Halo

Local Local Bubble Bubble

Rossetti et al. subm. The shock hypothesis is excluded at > 5σ The third discontinuity is the largest and outermost cold front ever detected

D. Eckert January 15, 2013 The CF is asymmetric, more extended toward the NE than toward the SW This is unlike the case of the remnant core in A3667 (Vikhlinin et al. 2001)

Rossetti et al. subm.

Rossetti M. et al.: Abell 2142 at large scales: An extreme case for sloshing?

Source

Source "'"! "'! "'"! "'! ! ! ! !

NXB 2708 2708 ! !

! NXB ! & & ! ! !" !"

Halo CXB CXB QSP

% QSP % ! ! Halo !" !" ()*+,-./0123)4(5626 ()*+,-./0123)4(5626

Local Local Bubble Bubble $ $ ! !

!" ! #$!" !" ! #$!" 9(0*:;2<708= 9(0*:;2<708=

Figure 3: XMM-Newton/MOS spectra of theStability IN (left) and OUT of (right) the regions NW and CF best-fit models. MOS1 data are shown in black, MOS2 in red. The contribution from the various background components (sky: local bubble, and CXB; internal: NXB and QSP) is compared to the best-fit source model.We studied the stability of the NW CF by measuring the SB jump in several sectors 27.250 27.200 27.150 IN Declination OUT 27.100 27.050

239.850 239.800 239.750 239.700 239.650 239.600 Right ascension

Figure 4: Zoom on the SE (left) and NW (right) cold fronts, with the regions used for spectral and imaging analysis. details of the background modeling are described in Appendix. Hugoniot conditions predict an outer temperature of 3.9 keV. Spectral fitting to the spectra in the IN and OUT regions was per- Fixing the temperatureD. Eckert of the plasmaJanuary to 3.9 15, keV, 2013 we find a di↵er- formed using XSPEC v12.7.0 and the modified C-statistic. The ence in C-statistic C = 28.1 with respect to the best-fit temper- appropriate e↵ective area was applied to the sky components, ature of 9.0 keV, which translates into a rejection of this hypoth- but not to the NXB and QSP components. The source spectrum esis at a confidence level of 5.3. Therefore, we can conclude was modeled with an absorbed APEC model with metal abun- with good confidence that this discontinuity is a cold front. At dance fixed to 0.3Z and fixed to that of the cluster. All almost 1 Mpc from the center this is the most distant cold front the free parameters were then fitted simultaneously. The result- ever detected in a (see Sec. 4). ing spectra are shown in Fig. 3, together with the various model components. In the IN region, our model gives a best-fit temper- +0.45 3.2. The NW cold front ature of kTIN = 6.13 0.37 keV, while for the OUT region we find kT = 9.0+2.4 keV. The errors are given at the 1 confidence OUT 1.5 level. The spatial resolution of XMM-Newton is su cient to analize The temperature of the OUT region appears higher than that the surface brightness and temperature profile also across of the IN region, so the SE discontinuity is likely a cold front. the NW cold front and compare this to results obtained with Indeed, in the shock scenario, with an inner temperature of 6.1 Chandra. keV and a density jump of 1.8 (see Table 1), the Rankine- We extracted the surface brightness profile across the NW cold front in the elliptical sector shown in Fig. 4 (magenta

4 Stability of the NW CF

Rossetti M. et al.: AbellWe 2142 studied at large scales: the stabilityAn extreme case of for the sloshing? NW CF by measuring the SB jump sector centered at 15 : 58 : 21, +27 : 13 : 41)in We several fitted the sectors profile between 1 and 7 arcminute from the center with an elliptical broken power-law model (Owers et al. 2009) and we found a density jump n /n = 1.97 0.03 with the IN OUT ± MOS image and nIN/nOUT = 2.00 0.04 with the pn. These values are consistent with the measurement± by Owers et al. (2009), nIN/nOUT = 1.9 0.1 , obtained with higher resolution Chandra data. We extracted± spectra in elliptical annular sectors to measure temperature inside and outside the edge, finding kTIN = (6.9 0.1) keV and kTOUT = (10.0 0.3) keV, a significant temperature± discontinuity. With this± analysis we could also measure a mildly significant (1.8) metal abun- dance discontinuity across the front: ZIN = (0.32 0.02)Z , ± ZOUT = (0.24 0.04)Z . Since the cold± front is detected over a large angular scale (see Sec. 3.2.1), we chose a sector extending from 20 to 190 degrees. corresponding to the whole angular range where a density discontinuity is detected. The best fit density jump is nIN/nOUT = 1.76 0.02. The temperature discontinuity is Figure 5: Density (upper panel) and temperature (lower panelRossetti et al. subm. smaller but still significant:± kT = (7.42 0.08) keV and IN ± The CF is asymmetric,discontinuities as more a function extended of the o↵ axis toward angle from the the NE axis than toward kTOUT = (9.6 0.2) keV. inclination of the NW cold front) ± the SW This is unlike the case of the remnant core in A3667 (Vikhlinin et 3.2.1. Azimuthal extent of the cold front uals are apparent at the position of the two known cold fronts al. 2001) and a significant large scale excess is clearly detected to the SE, The North-Western cold front in A2142 looks sharp over a large corresponding to the new feature we discussed Sec. 3.1. Part of D. Eckert January 15, 2013 angular range. We extracted SB profile in narrow (20-30, Fig. this excess was indeed visible also in the Chandra residual im- 4 ) elliptical sectors and fitted them with the projected broken age shown by Owers et al. (2011). power law described above. A significant density discontinuity Since the overall geometry of the X-ray emission of A2142 is (nIN/nOUT > 1 at more than 3) is detected from the sector [- very elliptical (with an axes ratio of 1.58), we also extracted the 20,10] to the sector [160,190]. In Fig.5, we plot the density surface brightness profile in elliptical sectors and produced the jump as a function of the o↵-axis angle from the “tip” of the residual image with respect to this model (Fig. 6, right panel). front, that we assume coincident with the overall elongation of While the negative residuals in the NE-SW direction almost dis- the elliptical contours, i. e. 40. A similar analysis has been per- appear, confirming that they are only the signature of the elliptic- formed for the “remnant-core” cold front of A3667 by Vikhlinin ity, the concentric excesses corresponding to the three cold fronts et al. (2001), but the shape of the front is very di↵erent, less are still clearly visible, showing that they are robust against sym- extended and more symmetric while in A2142 the discontinu- metry assumptions. With this geometry, it becomes more appar- ity extends more to the North than to the South. The very large ent a possible connection from the NW to the SE cold front in angular range (180) over which the density discontinuity can the northern part of the image which may recall a spiral struc- provide information on the geometry of the fronts with respect ture. to the line of sight (Sec. 4). The presence of multiple concentric excesses or spiral-like fea- We also measured the temperature discontinuity across the cold tures in the residual map is often observed in residual images front in sectors of 60width. The temperature jump is significant of simulated clusters featuring sloshing (see Sec. 4 for more de- at more than 2 in the sector extending from 10 to 130 degrees tails). from the R.A. axis, at larger angles the discontinuity is not sig- nificant. 3.4. Thermodynamic maps and X-ray morphology

3.3. Residual image We used the technique described in Rossetti et al. (2007) to pre- pare two dimensional temperature and projected density maps Residual images highlight the surface brightness deviations from starting from X-ray images. We then combined them to produce a symmetrical model and thus are an e↵ective way to detect and pseudo-pressure and pseudo-entropy maps. While obtained with identify cold fronts, both in real and simulated images (Roediger a simple background subtraction, that do not allow us to study et al. 2012). We prepare residual images starting from the EPIC the outskirts, our maps , shown in Fig. 7, can be used to infer the flux image: we mask point sources and calculate the average sur- gas properties in the more central region, up to the SE cold front. face brightness in concentric annuli (circular or elliptical). We The temperature map shows the presence of cool gas (around 7 then subtract this symmetric model from the image, and divide keV) in the denser regions of all three cold fronts detected in the the resulting di↵erence image by the model. This method allow X-ray image. The pseudo-entropy map shows that the lowest en- us to provide a quantitative information on the relative varia- tropy gas is located close to the more central SE cold front, as ob- tion of the source with respect to the symmetric model. It is also served in sloshing simulation, and shows an extension of lower less model-dependent than calculating residuals with respect to entropy gas to the SE in the direction of the new cold front. Both a beta-model fitting the surface brightness profile of the source. maps are highly asymmetrical at all scales, as usually observed In Fig. 6, middle panel, we show the residual image of A2142, in non cool core un-relaxed clusters. The pseudo-pressure map with respect to the spherical azimuthal average. Positive resid- is more symmetric but elliptical, possibly reflecting the shape of

5 The residual map uncovers concentric large-scale excesses

XMM residual map

We prepared residual maps by subtracting the azimuthal average at Rossetti M. et al.: Abell 2142 at large scales: An extreme case for sloshing? each radius

D. Eckert January 15, 2013 Figure 6: Left panel: EPIC flux image 0.4 2 keV. Central panel: Residual image with respect to the azimuthal average in concentric annuli. Right panel: Residual image with respect to the azimuthal average in elliptical annuli. the underlying potential well of the cluster. Table 2: Morphological and thermodynamical indicators of To better characterize the dynamical state of A2142 we also A2142 measured di↵erent morphological estimators starting from X- ray images (Table 2). We calculated the centroid shift w, both Indicator value class references 2 within R500 (Maughan et al. 2008) and within 500 kpc (Cassano w(< R500) (1.4 0.1) 10 intermediate 1 ± ⇥ 2 w(< 500 kpc) (1.3 0.1) 10 intermediate 2 et al. 2010): in both cases the values are at the limit between ± ⇥ the relaxed and disturbed class . The value of the surface bright- c100 0.247 0.001 relaxed 2 . ± . ness concentration parameter c depends on the scales used to c40 0 069 0 001 non-cool core 3 ✏ 0.37 ± 0.01 unrelaxed 4 quantify it: at the intermediate scales mapped with the definition ± 2 K0 (68 2)keV cm non-cool core 5 in Cassano et al. 2010 (c100=SB(< 100 kpc)/SB(< 500 kpc)) ±(0.5 ??) intermediate 6 the cluster apper relaxed, while at the smaller scales mapped ± by the original definition of Santos et al. 2008 (c40 =SB(< 40 References. (1) Maughan et al. (2008); (2) Cassano et al. (2010); kpc)/SB(< 400 kpc)) A2142 is not as concentrated as cool core (3) Santos et al. (2008); (4) Hashimoto et al. (2007); (5) Cavagnolo objects. The ellipticity is very high: with respect the sample of et al. (2009); (6) Leccardi et al. (2010). Hashimoto et al. (2007), A2142 shows one of the largest values (✏ = 1 b/a = 0.37). We can also study the thermodynamical state of A2142 using in- manually the regions with respect to the residual surface bright- dicators of the cool core state. The central entropy (K0 = 68 2 ness map (Fig. 8) and we extracted spectra (as in Sec. 3.1.2) from 2 ± each region. We then fitted spectra within XSPEC in the 0.7 10. keV cm ) is larger than the typical values of cool core objects (Cavagnolo et al. 2009) and the pseudo-entropy ratio classify it keV band with a vapec model multiplied by the Galactic hy- as an intermediate objects (Leccardi et al. 2010). drogen column density,NH, through the wabs absorption model. According to these results (summarized in Table 2), A2142 can- In the fitting procedure we fixed redshift and NH, whereas we not be unambiguously classified either as a relaxed cool core ob- left free all the elements with emission lines at energies above ject or as an un-relaxed merging non-cool core cluster. Its char- 0.7 keV . The metal abundances are measured relative to the acteristics are intermediate between the two classes. solar photospheric values of Anders & Grevesse (1989), where Fe = 4.68 105. We calculated the barycenter of every poly- gon and measure⇥ its distance from the surface brightness peak 3.5. Metal abundance distribution then we plot the temperature and metal abundance as a func- tion of the e↵ective distance. Regions corresponding to the SB We selected regions for spectral analysis starting from the sur- excess associated to the central and north-western cold front are face brightness residual map. This selection allow us to char- shown in red in Fig. 8: they feature lower temperature and higher acterize the thermodynamic properties of the ICM featuring a metal abundance than other regions located at the same distance surface brightness excess with respect to the other region of the from the center. Although the metal abundance excess is small, clusters. More specifically, if the SB excess is due to the ICM of Z = (0.35 0.1)Z with respect to the mean value in the other the core “sloshing” around the cluster, we expect these regions regions Z =±(0.29 0.1)Z , it is interesting to show that the gas to be cooler and possibly metal richer with respect to other re- with the highest ± (and lower temperature) of the clus- gions at the same distance from the cluster center. We selected ter is located in the regions corresponding to the central and NW

6 XMM residual map

Rossetti M. et al.: Abell 2142 at largeWe scales: prepared An extreme residual case for sloshing? maps by subtracting the azimuthal average at each radius

The residual map uncovers concentric large-scale excesses

D. Eckert January 15, 2013 Figure 6: Left panel: EPIC flux image 0.4 2 keV. Central panel: Residual image with respect to the azimuthal average in concentric annuli. Right panel: Residual image with respect to the azimuthal average in elliptical annuli. the underlying potential well of the cluster. Table 2: Morphological and thermodynamical indicators of To better characterize the dynamical state of A2142 we also A2142 measured di↵erent morphological estimators starting from X- ray images (Table 2). We calculated the centroid shift w, both Indicator value class references 2 within R500 (Maughan et al. 2008) and within 500 kpc (Cassano w(< R500) (1.4 0.1) 10 intermediate 1 ± ⇥ 2 w(< 500 kpc) (1.3 0.1) 10 intermediate 2 et al. 2010): in both cases the values are at the limit between ± ⇥ the relaxed and disturbed class . The value of the surface bright- c100 0.247 0.001 relaxed 2 . ± . ness concentration parameter c depends on the scales used to c40 0 069 0 001 non-cool core 3 ✏ 0.37 ± 0.01 unrelaxed 4 quantify it: at the intermediate scales mapped with the definition ± 2 K0 (68 2)keV cm non-cool core 5 in Cassano et al. 2010 (c100=SB(< 100 kpc)/SB(< 500 kpc)) ±(0.5 ??) intermediate 6 the cluster apper relaxed, while at the smaller scales mapped ± by the original definition of Santos et al. 2008 (c40 =SB(< 40 References. (1) Maughan et al. (2008); (2) Cassano et al. (2010); kpc)/SB(< 400 kpc)) A2142 is not as concentrated as cool core (3) Santos et al. (2008); (4) Hashimoto et al. (2007); (5) Cavagnolo objects. The ellipticity is very high: with respect the sample of et al. (2009); (6) Leccardi et al. (2010). Hashimoto et al. (2007), A2142 shows one of the largest values (✏ = 1 b/a = 0.37). We can also study the thermodynamical state of A2142 using in- manually the regions with respect to the residual surface bright- dicators of the cool core state. The central entropy (K0 = 68 2 ness map (Fig. 8) and we extracted spectra (as in Sec. 3.1.2) from 2 ± each region. We then fitted spectra within XSPEC in the 0.7 10. keV cm ) is larger than the typical values of cool core objects (Cavagnolo et al. 2009) and the pseudo-entropy ratio classify it keV band with a vapec model multiplied by the Galactic hy- as an intermediate objects (Leccardi et al. 2010). drogen column density,NH, through the wabs absorption model. According to these results (summarized in Table 2), A2142 can- In the fitting procedure we fixed redshift and NH, whereas we not be unambiguously classified either as a relaxed cool core ob- left free all the elements with emission lines at energies above ject or as an un-relaxed merging non-cool core cluster. Its char- 0.7 keV . The metal abundances are measured relative to the acteristics are intermediate between the two classes. solar photospheric values of Anders & Grevesse (1989), where Fe = 4.68 105. We calculated the barycenter of every poly- gon and measure⇥ its distance from the surface brightness peak 3.5. Metal abundance distribution then we plot the temperature and metal abundance as a func- tion of the e↵ective distance. Regions corresponding to the SB We selected regions for spectral analysis starting from the sur- excess associated to the central and north-western cold front are face brightness residual map. This selection allow us to char- shown in red in Fig. 8: they feature lower temperature and higher acterize the thermodynamic properties of the ICM featuring a metal abundance than other regions located at the same distance surface brightness excess with respect to the other region of the from the center. Although the metal abundance excess is small, clusters. More specifically, if the SB excess is due to the ICM of Z = (0.35 0.1)Z with respect to the mean value in the other the core “sloshing” around the cluster, we expect these regions regions Z =±(0.29 0.1)Z , it is interesting to show that the gas to be cooler and possibly metal richer with respect to other re- with the highest metallicity± (and lower temperature) of the clus- gions at the same distance from the cluster center. We selected ter is located in the regions corresponding to the central and NW

6 Spiral in an elliptical geometry; however, the scales are 3 times larger than in A496! ∼ Sloshing extends out to large radii (0.5R200) as in Perseus (Simionescu et al. 2012)

Comparison with simulations Comparison with simulations tailored for A496 (Roediger et al. 2012) 3640 E. Roediger et al.

one axis. At this strong inclination, the clear spiral pattern would be absent. The second option is to consider the stage at 1.8 Gyr after pericentre passage and rotate the merger geometry by 180◦, such that the previously outer northern front replaces the outer southern one, and the previously inner southern one replaces the outer northern one. As discussed in Section 3.3.2.3, this configuration does neither match the overall brightness distribution, especially at large scales, nor all CF radii simultaneously. Thus, our fiducial configuration is the best match. The single minor merger scenario we suggest for the history of A496 is a very simple one, and an almost complete match between predicted and observed features can be regarded as a success. Very likely the history of A496 has been more complex, e.g. there may be remnant bulk flows from an earlier merger. The interaction of outer southern CF/fan with this remnant perturbation may transfer the cold fan into a sharp CF.

Figure 8. X-ray brightness residual map for the initially spherical cluster. SpiralThe sloshing in a spiral spherically-symmetric is more regular. The fiducial subcluster potential moved on well a 4.3 Identifying the responsible subcluster diagonal orbit (see black line). Flin & Krywult (2006) analysed the galaxy distribution in A496 out to large radii and detected a subcluster or towards the N-NW at the outer edge of their field of view. They applied The brightness excess spiral is, however, less elongated compared their substructure detection method to a large number of clusters, to the elliptical cluster and much more regular (Fig. 8). With a and their general description of the method implies that each cluster diagonal SW–NE orbitD. Eckert orientation (seeJanuary black 15, line 2013 in Fig. 8), these is studied out to 1.5 Mpc from the cluster centre. Specifically for models also reproduce the large-scale brightness excess towards the A496, they compare their result to the ones of Durret et al. (2000). N-NW, but the outer brightness deficit is located towards the SE. This comparison reveals that in the case of A496, the field of view This difference to the elliptical models arises because the elliptical was larger, and that the subcluster in question is about 55 arcmin or structure alone leads to a characteristic brightness enhancement 2.1 Mpc from the cluster centre towards the N-NW. We have marked towards N and S and deficits towards E and W. the subcluster position in Fig. 5 along with the predicted position Overall, the elliptical cluster model results in a better match. of our simulated subcluster, i.e. at 1 Gyr after pericentre passage. Our prediction of the current subcluster position requires two more considerations. First, the rigid potential method slightly over- 4.1.2 Influence of initial temperature profiles estimates the CF ages. Secondly, the employed subcluster orbit is From the observations, the temperature profile of A496 is not well a simple test mass orbit, which does not include dynamical fric- constrained. Hence we tested several possible fits to the observations tion. While this is accurate enough for the orbit prior to pericentre (see Fig. B1). Our results are not sensitive to the details of the passage, dynamical friction will slow down the subcluster signif- temperature profile. The best match is achieved with profile 2, which icantly after pericentre distance (see Roediger & ZuHone 2012). is also the basis of our elliptical cluster model. Both considerations require that we expect the subcluster at a much smaller distance to the cluster centre than naively expected in our first estimate, i.e. around 1 Mpc towards the N-NE. While there is 4.2 The nature of the outer southern cold front some intrinsic uncertainty in the direction of the orbit due to the In our simulations, the outer southern CF never is a true discon- ellipticity of the cluster, in Section 3.3.2.2 we have excluded orbits tinuity, while the observations suggest that it is. This is the only which bring the subcluster towards the N-NW. Furthermore, dy- aspect the simulations do not reproduce. In the stronger merger namical friction makes it unlikely that a subcluster that passed the simulation, the gradients across it are steeper in all quantities, but cluster centre only about 1 Gyr ago or less can already have reached it is still continuous. R11 found the same structure in their Virgo the observed distance of the Flin–Krywult subcluster. These rea- sloshing simulations and named it cold fan to distinguish from a sons argue against this subcluster being responsible for the sloshing proper CF. We verified that this is not an artefact of the rigid poten- signatures in A496. tial approximation employed here (Roediger & ZuHone 2012). A The only other subcluster detected towards the N is LDCE0308 different orientation between the LOS and the orbital plane would about 1.3 Mpc to the N-NW. While its distance is favourable, our not make the feature appear sharper. simulations clearly favour a position towards N-NE instead of Alternatively, we could alter our fiducial merger geometry and N-NW. Furthermore, this subcluster is detected in the Two Micron history in order achieve a true outer southern CF, but we would All Sky Survey only and probably of low mass. sacrifice the good match we achieve in all other properties. After This leaves the identity of the subcluster still an open question, about 4.5 Gyr after pericentre passage, the now inner southern CF because in the region where we expect the subcluster, there is no ob- would have moved out far enough to reach the position of the vious galaxy concentration. However, it may be impossible to ever observed outer front and could replace it. At this stage, the northern identify the responsible subcluster due to the effect of tidal forces. CF would have reached a distance of about 270 kpc, where no CF During the pericentre passage, a subcluster is tidally compressed. is observed. The same is true for the western CF. An inclination Afew 100 Myr afterwards, however, it suffers substantial mass × of 77◦ between the orbital plane and the LOS would shorten the loss due to tidal decompression or tidal stripping, and may well be apparent distances of the CFs to the observed values, but only along dispersed and not recognizable as a compact structure anymore.

C # 2012 The Authors, MNRAS 420, 3632–3648 C Monthly Notices of the Royal Astronomical Society # 2012 RAS Spiral in an elliptical geometry; however, the scales are 3 times larger than in A496! ∼ Sloshing extends out to large radii (0.5R200) as in Perseus (Simionescu et al. 2012)

Rossetti M. et al.: Abell 2142 at large scales: An extreme case for sloshing?

17

18 26 10 28 16 8 6 4 12 0 22 11 2 75 15 91 3 19 13 14 23 20 21 27 Comparison with simulations

Comparison with simulations tailored-3.1 for-2.4 -1.7 A496-1 -0.31 (Roediger0.4 1.1 1.8 et2.5 al.3.2 3.9 Rossetti M. et al.: Abell 2142 at large scales: An extreme case for sloshing? Figure 8: Left panel: Surface brightness residual map, with the regions used for the spectral analysis. Right panel: temperature and 2012) metal abundance profile as a function of the e↵ective radius. Red and blue points correspond to the red and blue regions in the left panel (i.e. SB excess associated to the central cold fronts and to the SE one, respectively.)

The simulations by Roediger et al. (2012) are already based on a merger with a mass ratio of 5 and with an impact parameter in the range 100-400 kpc: a more' violent event than the minor mergers simulated by Ascasibar & Markevitch (2006), which in- duced only gentle perturbations in the core. The merger that in- duced the cold fronts in A2142 may be even more extreme than in A496. The metal abundance and temperature distributions in A2142 (Sec. 3.5) are also in agreement with prediction of simulations (Roediger et al. 2011): the richer and cooler gas is associated to the regions featuring a surface brightness excess. The gas in the large scale excess associated to the SE cold front is cooler but not significantly metal-richer than other regions. If we as- sume the metal abundance to remain constant during the slosh- ing process, the fact that the SE excess feature a lower metallic- ity may suggest that its gas did not originate in the more central regions but at larger scales. This is indeed expected: Ascasibar & Markevitch (2006) showed that during sloshing the ICM does not change its distance from the center by more than a factor of three (see also Sec. 4.6). Figure 9: Simulated SB residual map for A496, after 1 Gyr from Spiral in an ellipticalthe perycenter potential passage, from Roediger well et al. (2012) rotated to match the geometry of A2142 4.3. Comparison with optical results Recently, Owers et al. (2011) presented results based on multi- object spectroscopy of A2142. They selected 956 spectroscopi- to match the geometry of A2142. The morphological similar- cally confirmed cluster members and provided evidence of sig- Figure 6: Left panel: EPIC flux image 0.4 2 keV. Central panel: Residual image with respect to the azimuthal average in concentric ity between the observed and simulated maps is striking: they nificant substructure in the galaxy distribution. They identified both show concentric excesses corresponding to the central cold several group-sized substructures and suggest as the most likely annuli. Right panel: Residual image with respect to the azimuthal average in elliptical annuli. fronts and a third larger scale excess. Large scale excesses never perturber, which could have induced the sloshing, the subclus- show a sharp discontinuity in the simulations (Roediger et al. ter S1. It consists of a small group of galaxies (7-10) around the 2011, 2012), and indeed these features were dubbed “cool fan” second BCG (180 kpc NW of the first BCG), with a mean pe- D. Eckert January 15, 2013 1 the underlying potential well of the cluster. Table 2: Morphological and thermodynamical indicators of to distinguish them from cold fronts. On the contrary, we could culiar velocity vpec = 1760 km s and a dispersion v = 224 1 To better characterize the dynamical state of A2142 we also A2142 clearly detect the SE cold front both in ROSAT/PSPC and XMM- km s , as returned by the KMM test. Owers et al. (2011) in- Newton data. fer that its motion is mainly aligned with our line of sight and measured di↵erent morphological estimators starting from X- The main di↵erence between observed and simulated maps is that this orbit is consistent with the geometry of the two concen- ray images (Table 2). We calculated the centroid shift w, both Indicator value class references the larger scale of the sloshing phenomenon in A2142: cold tric cold fronts discovered by Chandra. However, the discovery 2 fronts are located at about twice the distance from the center of the third cold fronts and the large scale residual map (Fig. 6, within R500 (Maughan et al. 2008) and within 500 kpc (Cassano w(< R500) (1.4 0.1) 10 intermediate 1 ± ⇥ 2 w(< 500 kpc) (1.3 0.1) 10 intermediate 2 predicted by simulations. Cold fronts can be reproduced in sim- right panel), shows a possible connection between the NW cold et al. 2010): in both cases the values are at the limit between ± ⇥ ulations at such large scales if the sloshing phenomenon per- front and the SE excess, similar to the spiral like features in sim- the relaxed and disturbed class . The value of the surface bright- c100 0.247 0.001 relaxed 2 c 0.069 ± 0.001 non-cool core 3 sist for more than 2 Gyr or if they move outwards with a larger ulated images when the orbit of the perturber is in the plane of ness concentration parameter c depends on the scales used to 40 ± velocity following a “not very minor” initial merger (in terms the sky (Fig. 9). Therefore, we are probably observing the slosh- ✏ 0.37 0.01 unrelaxed 4 of impact parameter, mass and infall velocity of the subcluster). ing in an intermediate inclination. quantify it: at the intermediate scales mapped with the definition ± 2 K0 (68 2)keV cm non-cool core 5 in Cassano et al. 2010 (c100=SB(< 100 kpc)/SB(< 500 kpc)) ±(0.5 ??) intermediate 6 the cluster apper relaxed, while at the smaller scales mapped ± 8 by the original definition of Santos et al. 2008 (c40 =SB(< 40 References. (1) Maughan et al. (2008); (2) Cassano et al. (2010); kpc)/SB(< 400 kpc)) A2142 is not as concentrated as cool core (3) Santos et al. (2008); (4) Hashimoto et al. (2007); (5) Cavagnolo objects. The ellipticity is very high: with respect the sample of et al. (2009); (6) Leccardi et al. (2010). Hashimoto et al. (2007), A2142 shows one of the largest values (✏ = 1 b/a = 0.37). We can also study the thermodynamical state of A2142 using in- manually the regions with respect to the residual surface bright- dicators of the cool core state. The central entropy (K0 = 68 2 ness map (Fig. 8) and we extracted spectra (as in Sec. 3.1.2) from 2 ± each region. We then fitted spectra within XSPEC in the 0.7 10. keV cm ) is larger than the typical values of cool core objects (Cavagnolo et al. 2009) and the pseudo-entropy ratio classify it keV band with a vapec model multiplied by the Galactic hy- as an intermediate objects (Leccardi et al. 2010). drogen column density,NH, through the wabs absorption model. According to these results (summarized in Table 2), A2142 can- In the fitting procedure we fixed redshift and NH, whereas we not be unambiguously classified either as a relaxed cool core ob- left free all the elements with emission lines at energies above ject or as an un-relaxed merging non-cool core cluster. Its char- 0.7 keV . The metal abundances are measured relative to the acteristics are intermediate between the two classes. solar photospheric values of Anders & Grevesse (1989), where Fe = 4.68 105. We calculated the barycenter of every poly- gon and measure⇥ its distance from the surface brightness peak 3.5. Metal abundance distribution then we plot the temperature and metal abundance as a func- tion of the e↵ective distance. Regions corresponding to the SB We selected regions for spectral analysis starting from the sur- excess associated to the central and north-western cold front are face brightness residual map. This selection allow us to char- shown in red in Fig. 8: they feature lower temperature and higher acterize the thermodynamic properties of the ICM featuring a metal abundance than other regions located at the same distance surface brightness excess with respect to the other region of the from the center. Although the metal abundance excess is small, clusters. More specifically, if the SB excess is due to the ICM of Z = (0.35 0.1)Z with respect to the mean value in the other the core “sloshing” around the cluster, we expect these regions regions Z =±(0.29 0.1)Z , it is interesting to show that the gas to be cooler and possibly metal richer with respect to other re- with the highest metallicity± (and lower temperature) of the clus- gions at the same distance from the cluster center. We selected ter is located in the regions corresponding to the central and NW

6 Rossetti M. et al.: Abell 2142 at large scales: An extreme case for sloshing?

17

18 26 10 28 16 8 6 4 12 0 22 11 2 75 15 91 3 19 13 14 23 20 21 27 Comparison with simulations

Comparison with simulations tailored-3.1 for-2.4 -1.7 A496-1 -0.31 (Roediger0.4 1.1 1.8 et2.5 al.3.2 3.9 Rossetti M. et al.: Abell 2142 at large scales: An extreme case for sloshing? Figure 8: Left panel: Surface brightness residual map, with the regions used for the spectral analysis. Right panel: temperature and 2012) metal abundance profile as a function of the e↵ective radius. Red and blue points correspond to the red and blue regions in the left panel (i.e. SB excess associated to the central cold fronts and to the SE one, respectively.)

The simulations by Roediger et al. (2012) are already based on a merger with a mass ratio of 5 and with an impact parameter in the range 100-400 kpc: a more' violent event than the minor mergers simulated by Ascasibar & Markevitch (2006), which in- duced only gentle perturbations in the core. The merger that in- duced the cold fronts in A2142 may be even more extreme than in A496. The metal abundance and temperature distributions in A2142 (Sec. 3.5) are also in agreement with prediction of simulations (Roediger et al. 2011): the richer and cooler gas is associated to the regions featuring a surface brightness excess. The gas in the large scale excess associated to the SE cold front is cooler but not significantly metal-richer than other regions. If we as- sume the metal abundance to remain constant during the slosh- ing process, the fact that the SE excess feature a lower metallic- ity may suggest that its gas did not originate in the more central regions but at larger scales. This is indeed expected: Ascasibar & Markevitch (2006) showed that during sloshing the ICM does not change its distance from the center by more than a factor of three (see also Sec. 4.6). Figure 9: Simulated SB residual map for A496, after 1 Gyr from the perycenter passage, from Roediger et al. (2012) rotated to Spiral in an elliptical geometry;match thehowever, geometry of A2142 the scales are 34.3. Comparison with optical results ∼ Recently, Owers et al. (2011) presented results based on multi- times larger than in A496! object spectroscopy of A2142. They selected 956 spectroscopi- to match the geometry of A2142. The morphological similar- cally confirmed cluster members and provided evidence of sig- Figure 6: Left panel: EPIC flux image 0.4 2 keV. Central panel: Residual image withSloshing respect to the extends azimuthal outaverage to in concentriclarge radiiity between (0.5 theR observed) andas simulated in Perseus maps is striking: they nificant substructure in the galaxy distribution. They identified 200 several group-sized substructures and suggest as the most likely annuli. Right panel: Residual image with respect to the azimuthal average in elliptical annuli. both show concentric excesses corresponding to the central cold (Simionescu et al. 2012) fronts and a third larger scale excess. Large scale excesses never perturber, which could have induced the sloshing, the subclus- show a sharp discontinuity in the simulations (Roediger et al. ter S1. It consists of a small group of galaxies (7-10) around the 2011, 2012), and indeed these features were dubbed “cool fan” second BCG (180 kpc NW of the first BCG), with a mean pe- D. Eckert January 15, 2013 1 the underlying potential well of the cluster. Table 2: Morphological and thermodynamical indicators of to distinguish them from cold fronts. On the contrary, we could culiar velocity vpec = 1760 km s and a dispersion v = 224 1 To better characterize the dynamical state of A2142 we also A2142 clearly detect the SE cold front both in ROSAT/PSPC and XMM- km s , as returned by the KMM test. Owers et al. (2011) in- Newton data. fer that its motion is mainly aligned with our line of sight and measured di↵erent morphological estimators starting from X- The main di↵erence between observed and simulated maps is that this orbit is consistent with the geometry of the two concen- ray images (Table 2). We calculated the centroid shift w, both Indicator value class references the larger scale of the sloshing phenomenon in A2142: cold tric cold fronts discovered by Chandra. However, the discovery 2 fronts are located at about twice the distance from the center of the third cold fronts and the large scale residual map (Fig. 6, within R500 (Maughan et al. 2008) and within 500 kpc (Cassano w(< R500) (1.4 0.1) 10 intermediate 1 ± ⇥ 2 w(< 500 kpc) (1.3 0.1) 10 intermediate 2 predicted by simulations. Cold fronts can be reproduced in sim- right panel), shows a possible connection between the NW cold et al. 2010): in both cases the values are at the limit between ± ⇥ ulations at such large scales if the sloshing phenomenon per- front and the SE excess, similar to the spiral like features in sim- the relaxed and disturbed class . The value of the surface bright- c100 0.247 0.001 relaxed 2 c 0.069 ± 0.001 non-cool core 3 sist for more than 2 Gyr or if they move outwards with a larger ulated images when the orbit of the perturber is in the plane of ness concentration parameter c depends on the scales used to 40 ± velocity following a “not very minor” initial merger (in terms the sky (Fig. 9). Therefore, we are probably observing the slosh- ✏ 0.37 0.01 unrelaxed 4 of impact parameter, mass and infall velocity of the subcluster). ing in an intermediate inclination. quantify it: at the intermediate scales mapped with the definition ± 2 K0 (68 2)keV cm non-cool core 5 in Cassano et al. 2010 (c100=SB(< 100 kpc)/SB(< 500 kpc)) ±(0.5 ??) intermediate 6 the cluster apper relaxed, while at the smaller scales mapped ± 8 by the original definition of Santos et al. 2008 (c40 =SB(< 40 References. (1) Maughan et al. (2008); (2) Cassano et al. (2010); kpc)/SB(< 400 kpc)) A2142 is not as concentrated as cool core (3) Santos et al. (2008); (4) Hashimoto et al. (2007); (5) Cavagnolo objects. The ellipticity is very high: with respect the sample of et al. (2009); (6) Leccardi et al. (2010). Hashimoto et al. (2007), A2142 shows one of the largest values (✏ = 1 b/a = 0.37). We can also study the thermodynamical state of A2142 using in- manually the regions with respect to the residual surface bright- dicators of the cool core state. The central entropy (K0 = 68 2 ness map (Fig. 8) and we extracted spectra (as in Sec. 3.1.2) from 2 ± each region. We then fitted spectra within XSPEC in the 0.7 10. keV cm ) is larger than the typical values of cool core objects (Cavagnolo et al. 2009) and the pseudo-entropy ratio classify it keV band with a vapec model multiplied by the Galactic hy- as an intermediate objects (Leccardi et al. 2010). drogen column density,NH, through the wabs absorption model. According to these results (summarized in Table 2), A2142 can- In the fitting procedure we fixed redshift and NH, whereas we not be unambiguously classified either as a relaxed cool core ob- left free all the elements with emission lines at energies above ject or as an un-relaxed merging non-cool core cluster. Its char- 0.7 keV . The metal abundances are measured relative to the acteristics are intermediate between the two classes. solar photospheric values of Anders & Grevesse (1989), where Fe = 4.68 105. We calculated the barycenter of every poly- gon and measure⇥ its distance from the surface brightness peak 3.5. Metal abundance distribution then we plot the temperature and metal abundance as a func- tion of the e↵ective distance. Regions corresponding to the SB We selected regions for spectral analysis starting from the sur- excess associated to the central and north-western cold front are face brightness residual map. This selection allow us to char- shown in red in Fig. 8: they feature lower temperature and higher acterize the thermodynamic properties of the ICM featuring a metal abundance than other regions located at the same distance surface brightness excess with respect to the other region of the from the center. Although the metal abundance excess is small, clusters. More specifically, if the SB excess is due to the ICM of Z = (0.35 0.1)Z with respect to the mean value in the other the core “sloshing” around the cluster, we expect these regions regions Z =±(0.29 0.1)Z , it is interesting to show that the gas to be cooler and possibly metal richer with respect to other re- with the highest metallicity± (and lower temperature) of the clus- gions at the same distance from the cluster center. We selected ter is located in the regions corresponding to the central and NW

6 The temperature is systematically lower in the spiral There is a metallicity excess in the inner regions of the spiral, disappears farther out

Metallicity and temperature across the spiral

We extracted temperature and abundance profiles comparing inside and outside the spiral Rossetti M. et al.: Abell 2142 at large scales: An extreme case for sloshing?

17

18 26 10 28 16 8 6 4 12 0 22 11 2 75 15 91 3 19 13 14 23 20 21 27

-3.1 -2.4 -1.7 -1 -0.31 0.4 1.1 1.8 2.5 3.2 3.9

Figure 8: Left panel: Surface brightness residual map, with the regions used for the spectral analysis. Right panel: temperature and metal abundance profile as a function of the e↵ective radius. Red and blue points correspond to the red and blue regions in the left panel (i.e. SB excess associated to the central cold fronts and to the SE one, respectively.) D. Eckert January 15, 2013

The simulations by Roediger et al. (2012) are already based on a merger with a mass ratio of 5 and with an impact parameter in the range 100-400 kpc: a more' violent event than the minor mergers simulated by Ascasibar & Markevitch (2006), which in- duced only gentle perturbations in the core. The merger that in- duced the cold fronts in A2142 may be even more extreme than in A496. The metal abundance and temperature distributions in A2142 (Sec. 3.5) are also in agreement with prediction of simulations (Roediger et al. 2011): the richer and cooler gas is associated to the regions featuring a surface brightness excess. The gas in the large scale excess associated to the SE cold front is cooler but not significantly metal-richer than other regions. If we as- sume the metal abundance to remain constant during the slosh- ing process, the fact that the SE excess feature a lower metallic- ity may suggest that its gas did not originate in the more central regions but at larger scales. This is indeed expected: Ascasibar & Markevitch (2006) showed that during sloshing the ICM does not change its distance from the center by more than a factor of three (see also Sec. 4.6). Figure 9: Simulated SB residual map for A496, after 1 Gyr from the perycenter passage, from Roediger et al. (2012) rotated to match the geometry of A2142 4.3. Comparison with optical results Recently, Owers et al. (2011) presented results based on multi- object spectroscopy of A2142. They selected 956 spectroscopi- to match the geometry of A2142. The morphological similar- cally confirmed cluster members and provided evidence of sig- ity between the observed and simulated maps is striking: they nificant substructure in the galaxy distribution. They identified both show concentric excesses corresponding to the central cold several group-sized substructures and suggest as the most likely fronts and a third larger scale excess. Large scale excesses never perturber, which could have induced the sloshing, the subclus- show a sharp discontinuity in the simulations (Roediger et al. ter S1. It consists of a small group of galaxies (7-10) around the 2011, 2012), and indeed these features were dubbed “cool fan” second BCG (180 kpc NW of the first BCG), with a mean pe- 1 to distinguish them from cold fronts. On the contrary, we could culiar velocity vpec = 1760 km s and a dispersion v = 224 1 clearly detect the SE cold front both in ROSAT/PSPC and XMM- km s , as returned by the KMM test. Owers et al. (2011) in- Newton data. fer that its motion is mainly aligned with our line of sight and The main di↵erence between observed and simulated maps is that this orbit is consistent with the geometry of the two concen- the larger scale of the sloshing phenomenon in A2142: cold tric cold fronts discovered by Chandra. However, the discovery fronts are located at about twice the distance from the center of the third cold fronts and the large scale residual map (Fig. 6, predicted by simulations. Cold fronts can be reproduced in sim- right panel), shows a possible connection between the NW cold ulations at such large scales if the sloshing phenomenon per- front and the SE excess, similar to the spiral like features in sim- sist for more than 2 Gyr or if they move outwards with a larger ulated images when the orbit of the perturber is in the plane of velocity following a “not very minor” initial merger (in terms the sky (Fig. 9). Therefore, we are probably observing the slosh- of impact parameter, mass and infall velocity of the subcluster). ing in an intermediate inclination.

8 Metallicity and temperature across the spiral

We extracted temperature and abundance profiles comparing inside Rossetti M. et al.: Abell 2142 at large scales: An extreme case for sloshing? and outside the spiral

17

18 26 10 28 16 8 6 4 12 0 22 11 2 75 15 91 3 19 13 14 23 20 21 27

-3.1 -2.4 -1.7 -1 -0.31 0.4 1.1 1.8 2.5 3.2 3.9

Figure 8: Left panel: Surface brightness residual map, withThe the temperature regions used for the is spectralsystematically analysis. Right lower panel: in temperature the spiral and metal abundance profile as a function of the e↵ective radius. Red and blue points correspond to the red and blue regions in the left panel (i.e. SB excess associated to the central cold frontsThere and to the is SE a one,metallicity respectively.) excess in the inner regions of the spiral, disappears farther out The simulations by Roediger et al. (2012) are already based on a merger with a mass ratio of 5 and with an impact parameter in the range 100-400D. Eckert kpc: a more' January violent 15, event 2013 than the minor mergers simulated by Ascasibar & Markevitch (2006), which in- duced only gentle perturbations in the core. The merger that in- duced the cold fronts in A2142 may be even more extreme than in A496. The metal abundance and temperature distributions in A2142 (Sec. 3.5) are also in agreement with prediction of simulations (Roediger et al. 2011): the richer and cooler gas is associated to the regions featuring a surface brightness excess. The gas in the large scale excess associated to the SE cold front is cooler but not significantly metal-richer than other regions. If we as- sume the metal abundance to remain constant during the slosh- ing process, the fact that the SE excess feature a lower metallic- ity may suggest that its gas did not originate in the more central regions but at larger scales. This is indeed expected: Ascasibar & Markevitch (2006) showed that during sloshing the ICM does not change its distance from the center by more than a factor of three (see also Sec. 4.6). Figure 9: Simulated SB residual map for A496, after 1 Gyr from the perycenter passage, from Roediger et al. (2012) rotated to match the geometry of A2142 4.3. Comparison with optical results Recently, Owers et al. (2011) presented results based on multi- object spectroscopy of A2142. They selected 956 spectroscopi- to match the geometry of A2142. The morphological similar- cally confirmed cluster members and provided evidence of sig- ity between the observed and simulated maps is striking: they nificant substructure in the galaxy distribution. They identified both show concentric excesses corresponding to the central cold several group-sized substructures and suggest as the most likely fronts and a third larger scale excess. Large scale excesses never perturber, which could have induced the sloshing, the subclus- show a sharp discontinuity in the simulations (Roediger et al. ter S1. It consists of a small group of galaxies (7-10) around the 2011, 2012), and indeed these features were dubbed “cool fan” second BCG (180 kpc NW of the first BCG), with a mean pe- 1 to distinguish them from cold fronts. On the contrary, we could culiar velocity vpec = 1760 km s and a dispersion v = 224 1 clearly detect the SE cold front both in ROSAT/PSPC and XMM- km s , as returned by the KMM test. Owers et al. (2011) in- Newton data. fer that its motion is mainly aligned with our line of sight and The main di↵erence between observed and simulated maps is that this orbit is consistent with the geometry of the two concen- the larger scale of the sloshing phenomenon in A2142: cold tric cold fronts discovered by Chandra. However, the discovery fronts are located at about twice the distance from the center of the third cold fronts and the large scale residual map (Fig. 6, predicted by simulations. Cold fronts can be reproduced in sim- right panel), shows a possible connection between the NW cold ulations at such large scales if the sloshing phenomenon per- front and the SE excess, similar to the spiral like features in sim- sist for more than 2 Gyr or if they move outwards with a larger ulated images when the orbit of the perturber is in the plane of velocity following a “not very minor” initial merger (in terms the sky (Fig. 9). Therefore, we are probably observing the slosh- of impact parameter, mass and infall velocity of the subcluster). ing in an intermediate inclination.

8 Rossetti M. et al.: Abell 2142 at large scales: An extreme case for sloshing?

Pseudo-entropy map

We produced a pseudo-entropy map using adaptive binning (Rossetti et al. 2007)

There is a tail of low-entropy gas in direction of the new SE 5.71 6.45 7.18 7.93 8.67 CF9.41 10.1 10.9 11.6 12.4 13.1

Figure 7: EPIC SB image (upper left), temperature map(upper right), pressure map(lowerD. left) Eckert andJanuary entropy 15, 2013 map (lower right). Maps were obtained with a target S/N = 15. cold fronts. The regions corresponding to the SE large scale ex- excesses and “crossing” of profiles in opposite directions is a cess and to the third cold front, shown in blue in Fig. 8, also fea- generic feature of the sloshing scenario. ture a lower temperature with respect to regions at the same dis- The SE cold front of A2142 is a record breaking feature: at 1 tance from the center but do not show a significant metal abun- Mpc from the center, is the outermost cold front, detected both dance excess. as a surface brightness and temperature discontinuity and not obviously associated to a moving subcluster, ever observed in a galaxy cluster (see also Sec. 4.5). 4. Discussion 4.1. The SE cold front 4.2. Comparison with simulations As discussed in Sec. 3.1, the south-eastern edge clearly visible in Fig. 1 is likely a cold front. The presence of another cold front The residual surface brightness images of A2142 shown in Fig. 6 in this cluster, located at almost 1 Mpc from the center, makes can be compared with the prediction of numerical simulations the overall picture described in Sec. 1 even more intriguing. to test the possibility that the fronts in A2142 are due to slosh- We tested the possibility that this cold front may belong to the ing. For instance Roediger et al. (2011, 2012) show predicted “remnant-core” class and therefore be the boundary of a merging residual images from ad-hoc simulations of the Virgo cluster and subcluster traveling towards south-east in the ICM of the main A496 to reproduce the observed features and to infer the char- cluster, and possibly inducing the sloshing of the core to produce acteristics of the minor mergers which induced the sloshing. To the other small scale cold fronts. However, we could not find ei- perform a similar analysis on A2142, new tailored simulations ther a galaxy concentration in the analysis of Owers et al. (2011) are needed. A full hydro-dynamical + N-body treatement will either a clump in the mass distribution in the weak lensing anal- be required, since in the outskirts (at 1 Mpc from the center, ysis of Okabe & Umetsu (2008), which could be associated to where we observe the SE cold front) the rigid potential approxi- this feature. mation (Roediger & Zuhone 2012) may not be completely valid. The presence of this newly detected discontinuity cannot be un- Moreover, if the ellipticity of A2142 reflects the ellipticity of derstood without a global view of the cluster. A2142 is one of the the potential well, the position and shapes of cold fronts may few clusters hosting four cold fronts, with A496 (Ghizzardi et al. be di↵erent with respect to spherical simulation. Therefore, we 2010) and possibly Perseus (Simionescu et al. 2012). Moreover, limit the comparison with simulations at a qualitative level and as noticed also in Perseus by Simionescu et al. (2012), the large we refer for a more complete discussion to a forthcoming paper scale excess looks connected to the smaller scale structures: the (Roediger et al. in prep.). SE excess begins, in terms of radius, just at the same distance In Fig.9, we show a simulated residual image for A496, with from the center as the NW discontinuity. This alternation of the orbit of the perturber in the plane of the sky, that we rotated

7 If the process is adiabatic by comparing the (pseudo-)entropy profiles we can estimate the original location of the gas before the sloshing mechanism set in

Pseudo-entropy profiles

We defined a “spiral” and an “unperturbed” region and extracted pseudo-entropy profiles (using projected T and n)

0 0.03 0.089 0.21 0.44 0.92 1.9 3.7 7.5 15 30

D. Eckert January 15, 2013 If the process is adiabatic by comparing the (pseudo-)entropy profiles we can estimate the original location of the gas before the sloshing mechanism set in

Pseudo-entropy profiles

We defined a “spiral” and an “unperturbed” region and extracted pseudo-entropy profiles (using projected T and n) ] -2

Projected entropy [keV cm 100

1 Radius [arcmin]

D. Eckert January 15, 2013 Pseudo-entropy profiles

We defined a “spiral” and an “unperturbed” region and extracted pseudo-entropy profiles (using projected T and n) ] -2

Projected entropy [keV cm 100

1 c(r) r Radius [arcmin] If the process is adiabatic by comparing the (pseudo-)entropy profiles we can estimate the original location of the gas before the sloshing mechanism set in

D. Eckert January 15, 2013 correcting for intrinsic ellipticity of the halo

Determination of the sloshing energy Corresponding radius c(r) where the gas at radius r was originally located before the sloshing set in

6

5

4

3

2

Corresponding radius [arcmin] 1

1 2 3 4 5 6 7 8 Radius in the spiral [arcmin] The total energy inputted by sloshing is the difference of potential energy: ≈

Mtot (< r) Mtot (< c(r)) ∆E = GMgas (r) dr spiral r − c(r) Z  

D. Eckert January 15, 2013 Determination of the sloshing energy Corresponding radius c(r) where the gas at radius r was originally located before the sloshing set in correcting for intrinsic ellipticity of the halo

6

5

4

3

2

Corresponding radius [arcmin] 1

1 2 3 4 5 6 7 8 Radius in the spiral [arcmin] The total energy inputted by sloshing is the difference of potential energy: ≈

Mtot (< r) Mtot (< c(r)) ∆E = GMgas (r) dr spiral r − c(r) Z  

D. Eckert January 15, 2013 13 Corresponds to the kinetic energy of a 10 M at vrel 1,000 1 ∼ km s− Represents 5% of the thermal energy of the cluster within ∼ R500, 10% within the sloshing region ∼ Same order of magnitude as the energy dissipated by radiation in 1Gyr in the cooling region

Determination of the sloshing energy

The total energy inputted by sloshing is the difference of potential energy: ≈

Mtot (< r) Mtot (< c(r)) ∆E = GMgas (r) dr spiral r − c(r) Z  

62 Result: Esloshing = (0.8 1.5) 10 ergs − ×

D. Eckert January 15, 2013 Represents 5% of the thermal energy of the cluster within ∼ R500, 10% within the sloshing region ∼ Same order of magnitude as the energy dissipated by radiation in 1Gyr in the cooling region

Determination of the sloshing energy

The total energy inputted by sloshing is the difference of potential energy: ≈

Mtot (< r) Mtot (< c(r)) ∆E = GMgas (r) dr spiral r − c(r) Z  

62 Result: Esloshing = (0.8 1.5) 10 ergs − ×

13 Corresponds to the kinetic energy of a 10 M at vrel 1,000 1 ∼ km s−

D. Eckert January 15, 2013 Same order of magnitude as the energy dissipated by radiation in 1Gyr in the cooling region

Determination of the sloshing energy

The total energy inputted by sloshing is the difference of potential energy: ≈

Mtot (< r) Mtot (< c(r)) ∆E = GMgas (r) dr spiral r − c(r) Z  

62 Result: Esloshing = (0.8 1.5) 10 ergs − ×

13 Corresponds to the kinetic energy of a 10 M at vrel 1,000 1 ∼ km s− Represents 5% of the thermal energy of the cluster within ∼ R500, 10% within the sloshing region ∼

D. Eckert January 15, 2013 Determination of the sloshing energy

The total energy inputted by sloshing is the difference of potential energy: ≈

Mtot (< r) Mtot (< c(r)) ∆E = GMgas (r) dr spiral r − c(r) Z  

62 Result: Esloshing = (0.8 1.5) 10 ergs − ×

13 Corresponds to the kinetic energy of a 10 M at vrel 1,000 1 ∼ km s− Represents 5% of the thermal energy of the cluster within ∼ R500, 10% within the sloshing region ∼ Same order of magnitude as the energy dissipated by radiation in 1Gyr in the cooling region

D. Eckert January 15, 2013 Summary and conclusions

We discovered a third CF in A2142 at 1 Mpc from the cluster center, it is the largest and outermost CF detected so far We uncovered a giant spiral structure extending out to half of the virial radius of the system ∼ The morphology is similar to sloshing simulations, although on a much larger scale A2142 is a large-scale sloshing cluster, the second with Perseus→ The (pseudo-)entropy is systematically lower in the spiral The gas has been transported outwards adiabatically → We estimate a total sloshing energy of 1062 ergs, which ∼ corresponds to 5% of the thermal energy within R500 ∼ Stay tuned for the next talks on this cluster (Damon, Mariachiara, and Matt)!

D. Eckert January 15, 2013 Bonus: γ-ray upper limits with Fermi/LAT

CR protons hitting ICM ions trigger particle cascades (if ECR > 2mπ : p + p p + p + nπ → Neutral pions decay directly into two γ-ray photons The γ-ray emission (E > 100 MeV) can be used to constrain the CR→ energy density

D. Eckert January 15, 2013 Bonus: γ-ray upper limits with Fermi/LAT

CR protons hitting ICM ions trigger particle cascades (if ECR > 2mπ : p + p p + p + nπ → Neutral pions decay directly into two γ-ray photons The γ-ray emission (E > 100 MeV) can be used to constrain the CR→ energy density

D. Eckert January 15, 2013 No significant signal detected, Flux UL 1 order of magnitude below UL for individual systems ∼

Cluster stacking analysis

We selected a subsample of 53 clusters from the complete HIFLUGCS sample and stacked the data

sorted by descending M / (z*z)

24 photon index −0.8 photon index −1.2 photon index −1.6 21 photon index −2.0 photon index −2.4 photon index −2.8 18

15

12 TS value 9

6

3

0

1 6 11 16 21 26 31 36 41 46 51 number of stacked clusters

Huber et al. subm.

D. Eckert January 15, 2013 Cluster stacking analysis

We selected a subsample of 53 clusters from the complete HIFLUGCS sample and stacked the data

sorted by descending M / (z*z)

photon index −0.8, UL photon index −1.2, UL photon index −1.6, UL photon index −2.0, UL −9 10 photon index −2.4, UL photon index −2.8, UL s) 2

−10 10 flux (ph/cm

−11 10

1 6 11 16 21 26 31 36 41 46 51 number of stacked clusters

Huber et al. subm.

No significant signal detected, Flux UL 1 order of magnitude below UL for individual systems ∼

D. Eckert January 15, 2013 Q: Are we starting to probe hadronic models for mini-halos?

Upper limits on the CR energy density

B. Huber et al.: Probing the cosmic-ray content of galaxy clusters by stacking Fermi-LAT count maps Modeling the π0 emission spectrum and using the density profiles with the gas (i.e. we set ! =1)anddeterminethegamma-of targetTable particles 2. Upper from limits X-rays, on the we CR-to-thermal computed UL onenergy the CRratio energy (in ray flux that should be observed under this assumption.density.percent) We foralso di splitfferent the power-law sample into indices CC and of NCC the parent systems proton The ratio of the observed UL to the flux expected when population, computed using Eq. 10 and assuming the same ra- assuming equipartition then gives the value of !. dial distribution for the CR and for the gas (Eq. 7). Observationally, we obtained upper limits on the Γp Γph CC NCC ALL stacked cluster sample for photon indices Γph=2, 2.4, and 2.8. Such effective spectra are produced by CR populations 2.10 2.0 8.11.73.0 2.55 2.4 6.50.81.0 with power-law indices Γp=2.1, 2.55, and 3, respectively. We discard here the results obtained for very hard photon 3.00 2.8 5.50.80.8 indices, since the π0-decay process cannot produce such hard spectra. Under the assumption of equipartition, we Huber et al. subm. Table 3. Upper limits on the CR-to-thermal energy ratio (in derive for each cluster the emitted photon spectrum per Radial CR distribution identical to the gas unit time and volume. The equipartition flux of cluster i percent) for different power-law indices of the parent proton produced within these assumptions is then given by population, computed using Eq. 10 and assuming that the CR have a steeper radial distribution than the thermal gas, !(r) 0.5 ∼ 100 GeV r− . 1 3 dNγ (Γ) Fi(Γ)= 2 d r Eγ dEγ . (9) 4πd dV dEγ dt Γ Γ CC NCC ALL L,i !Vi !200 MeV " #i p ph D. Eckert January 15, 2013 To compare with our upper limits on the stacked popula- 2.10 2.0 12.02.34.1 2.55 2.4 10.01.01.4 tions, we define the expected equipartition flux Fequip as 3.00 2.8 8.61.11.2 the mean of the equipartition fluxes of individual systems,

N 1 F (Γ)= F (Γ), (10) Table 4. Upper limit on the CR-to-thermal energy ratio for dif- equip N i i=1 ferent power-law indices of the parent proton population, com- $ puted using Eq. 10 and assuming that the CR have a flatter where N is the total number of galaxy clusters in the sam- radial distribution, !(r) r0.5. ple. Then, an upper limit on the CR-to-thermal energy ratio ∼ can simply be obtained by taking the ratio of the observed Γp Γph CC NCC ALL flux to the equipartition flux, 2.10 2.0 4.71.32 2.55 2.4 3.80.60.7 ULFermi(Γ) !(Γ)= , (11) 3.00 2.8 3.20.60.6 Fequip(Γ) where ULFermi is the upper limit obtained in the previous section. regions, leading to an increasing radial profile (Vazza et al. 2012b). In Table 3 we give upper limits on the average CR- 5.2.2. Results to-thermal energy ratio, assuming a profile increasing like !(r) r0.5.Conversely,ifCRareinjectedbyoutbursts We compute the equipartition flux for each galaxy clus- of the∼ central AGN, a steeper CR profile is expected with ter by solving Eq. 9 using the parameters derived by Chen respect to the gas (Fujita & Ohira 2011). In Table 4 we et al. (2007). We note that among the 53 systems compris- 0.5 provide the results assuming !(r) r− . ing our sample, 32 are classified as NCC, while 21 exhibit ∼ CC properties. As an estimate of the virial radius Rvir, we use the scaling relations of Arnaud et al. (2005), mak- 6. Discussion ing the approximation Rvir R200.Forindication,inour sample of 53 galaxy clusters,∼ the average values for the 6.1. Comparison with previous observational results 2 3 parameters are R200 1.5Mpc,ngas0 10− cm− , rc Using data obtained by the EGRET experiment on board ∼ ∼ ∼ 210 kpc, β 0.63, and kT 4.6keV.Wenotethatamong the Compton Gamma-Ray Observatory,Reimeretal. ∼ ∼ our sample CC clusters typically have a lower temperature (2003) performed a similar study to the one presented in (TCC 3.2keV),andthusalowermass,thanNCCsystems this paper, reaching upper limits on the CR-to-thermal en- ∼ (TNCC 5.5keV). ergy ratio of the order of 10-20%. Thanks to the excellent ∼ Our results on the CR-to-thermal energy ratio for the sensitivity of Fermi /LAT, our upper limit lies more than different subsamples and photon indices are given in Table an order of magnitude below the upper limit obtained in 2. The limits provided here can be readily transformed into this study. Recently, Ackermann et al. (2010) performed a the pressure ratio (χ)usingtherelation UCR =2PCR in Uth Pth similar analysis on individual systems, always resulting on the fully relativistic case. So for the CC sample, our best gamma-ray upper limits. This study allowed to constrain limit in term of the pressure ratio is χ =2.3%, obtained the CR energy density at the level of a few percent of the for a proton index of 3. For the NCC sample, the best limit thermal energy density, in the best cases. Similar results is χ =0.4%, obtained for proton indices softer than 2.55, were recently obtained on the Coma cluster through a com- while for the total sample, the best limit is of χ =0.4%, bination of Fermi and VERITAS data (Arlen et al. 2012), again for the softest spectral index. which led to an upper limit of the order of 2% on the CR- In addition to the isobaric case, we also used different to-thermal pressure ratio. Compared to these studies, our CR radial distributions, representing a decreasing and an analysis brings constraints on the average population which increasing pressure ratio. Indeed, injection of CR at cos- are slightly below the few best cases for individual systems. mological shocks would occur preferentially in the outer The stacking method therefore allows us to bring the con-

8 Q: Are we starting to probe hadronic models for mini-halos?

B. Huber et al.: Probing the cosmic-ray content of galaxy clusters by stacking Fermi-LAT count maps with the gas (i.e. we set ! = 1) and determine the gamma- Table 2. Upper limits on the CR-to-thermal energy ratio (in ray flux that should be observed under this assumption. percent) for different power-law indices of the parent proton The ratio of the observed UL to the flux expected when population, computed using Eq. 10 and assuming the same ra- assuming equipartition then gives the value of !. dial distribution for the CR and for the gas (Eq. 7). Observationally, we obtained upper limits on the Γp Γph CC NCC ALL stacked cluster sample for photon indices Γph=2, 2.4, and 2.8. Such effective spectra are produced by CR populations 2.10 2.0 8.11.73.0 2.55 2.4 6.50.81.0 with power-law indices Γp=2.1, 2.55, and 3, respectively. We discard here the results obtained for very hard photon 3.00 2.8 5.50.80.8 indices, since the π0-decay process cannot produce such hard spectra. Under the assumption of equipartition, we derive for each cluster the emitted photon spectrum per Table 3. Upper limits on the CR-to-thermal energy ratio (in unit time and volume. The equipartition flux of cluster i percent) for different power-law indices of the parent proton produced within these assumptions is then given by population, computed using Eq. 10 and assuming that the CR have a steeper radial distribution than the thermal gas, !(r) 0.5 ∼ 100 GeV r− . 1 3 dNγ (Γ) Fi(Γ)= 2 d r Eγ dEγ . (9) 4πd dV dEγ dt Γ Γ CC NCC ALL L,i !Vi !200 MeV " #i p ph To compare with our upper limits on the stacked popula- 2.10 2.0 12.02.34.1 Upper limits on the2.55 CR 2.4 energy 10.01 density.01.4 tions, we define the expected equipartition flux Fequip as 3.00 2.8 8.61.11.2 the mean of the equipartition fluxes of individual systems, 0 N Modeling the π emission spectrum and using the density profiles 1 F (Γ)= F (Γ), (10)of targetTable particles 4. Upper from limit X-rays, on the CR-to-thermal we computed UL energy on the ratio CR for energy dif- equip N i i=1 density.ferent We power-law also split indices the sample of theinto parent CC proton and NCC population, systems com- $ puted using Eq. 10 and assuming that the CR have a flatter where N is the total number of galaxy clusters in the sam- radial distribution, !(r) r0.5. ple. Then, an upper limit on the CR-to-thermal energy ratio ∼ can simply be obtained by taking the ratio of the observed Γp Γph CC NCC ALL flux to the equipartition flux, 2.10 2.0 4.71.32.0 2.55 2.4 3.80.60.7 ULFermi(Γ) !(Γ)= , (11) 3.00 2.8 3.20.60.6 Fequip(Γ) Huber et al. subm. where ULFermi is the upper limit obtained in the previous section. regions, leadingRadial CR to andistribution increasing steeper radial than profile the (Vazza gas et al. 2012b). In Table 3 we give upper limits on the average CR- 5.2.2. Results to-thermal energy ratio, assuming a profile increasing like !(r) r0.5. Conversely, if CR are injected by outbursts We compute the equipartition flux for each galaxy clus- of the∼ central AGN, a steeper CR profile is expected with ter by solving Eq. 9 using the parameters derived by Chen respect to the gas (Fujita & Ohira 2011). In Table 4 we et al. (2007). We note that among the 53 systems compris- 0.5 provide the results assumingD. Eckert !(Januaryr) r 15,− 2013. ing our sample, 32 are classified as NCC, while 21 exhibit ∼ CC properties. As an estimate of the virial radius Rvir, we use the scaling relations of Arnaud et al. (2005), mak- 6. Discussion ing the approximation Rvir R200. For indication, in our sample of 53 galaxy clusters,∼ the average values for the 6.1. Comparison with previous observational results 2 3 parameters are R200 1.5Mpc,ngas0 10− cm− , rc Using data obtained by the EGRET experiment on board ∼ ∼ ∼ 210 kpc, β 0.63, and kT 4.6keV.Wenotethatamong the Compton Gamma-Ray Observatory, Reimer et al. ∼ ∼ our sample CC clusters typically have a lower temperature (2003) performed a similar study to the one presented in (TCC 3.2 keV), and thus a lower mass, than NCC systems this paper, reaching upper limits on the CR-to-thermal en- ∼ (TNCC 5.5keV). ergy ratio of the order of 10-20%. Thanks to the excellent ∼ Our results on the CR-to-thermal energy ratio for the sensitivity of Fermi /LAT, our upper limit lies more than different subsamples and photon indices are given in Table an order of magnitude below the upper limit obtained in 2. The limits provided here can be readily transformed into this study. Recently, Ackermann et al. (2010) performed a the pressure ratio (χ) using the relation UCR =2PCR in Uth Pth similar analysis on individual systems, always resulting on the fully relativistic case. So for the CC sample, our best gamma-ray upper limits. This study allowed to constrain limit in term of the pressure ratio is χ =2.3%, obtained the CR energy density at the level of a few percent of the for a proton index of 3. For the NCC sample, the best limit thermal energy density, in the best cases. Similar results is χ =0.4%, obtained for proton indices softer than 2.55, were recently obtained on the Coma cluster through a com- while for the total sample, the best limit is of χ =0.4%, bination of Fermi and VERITAS data (Arlen et al. 2012), again for the softest spectral index. which led to an upper limit of the order of 2% on the CR- In addition to the isobaric case, we also used different to-thermal pressure ratio. Compared to these studies, our CR radial distributions, representing a decreasing and an analysis brings constraints on the average population which increasing pressure ratio. Indeed, injection of CR at cos- are slightly below the few best cases for individual systems. mological shocks would occur preferentially in the outer The stacking method therefore allows us to bring the con-

8 B. Huber et al.: Probing the cosmic-ray content of galaxy clusters by stacking Fermi-LAT count maps with the gas (i.e. we set ! = 1) and determine the gamma- Table 2. Upper limits on the CR-to-thermal energy ratio (in ray flux that should be observed under this assumption. percent) for different power-law indices of the parent proton The ratio of the observed UL to the flux expected when population, computed using Eq. 10 and assuming the same ra- assuming equipartition then gives the value of !. dial distribution for the CR and for the gas (Eq. 7). Observationally, we obtained upper limits on the Γp Γph CC NCC ALL stacked cluster sample for photon indices Γph=2, 2.4, and 2.8. Such effective spectra are produced by CR populations 2.10 2.0 8.11.73.0 2.55 2.4 6.50.81.0 with power-law indices Γp=2.1, 2.55, and 3, respectively. We discard here the results obtained for very hard photon 3.00 2.8 5.50.80.8 indices, since the π0-decay process cannot produce such hard spectra. Under the assumption of equipartition, we derive for each cluster the emitted photon spectrum per Table 3. Upper limits on the CR-to-thermal energy ratio (in unit time and volume. The equipartition flux of cluster i percent) for different power-law indices of the parent proton produced within these assumptions is then given by population, computed using Eq. 10 and assuming that the CR have a steeper radial distribution than the thermal gas, !(r) 0.5 ∼ 100 GeV r− . 1 3 dNγ (Γ) Fi(Γ)= 2 d r Eγ dEγ . (9) 4πd dV dEγ dt Γ Γ CC NCC ALL L,i !Vi !200 MeV " #i p ph To compare with our upper limits on the stacked popula- 2.10 2.0 12.02.34.1 Upper limits on the2.55 CR 2.4 energy 10.01 density.01.4 tions, we define the expected equipartition flux Fequip as 3.00 2.8 8.61.11.2 the mean of the equipartition fluxes of individual systems, 0 N Modeling the π emission spectrum and using the density profiles 1 F (Γ)= F (Γ), (10)of targetTable particles 4. Upper from limit X-rays, on the CR-to-thermal we computed UL energy on the ratio CR for energy dif- equip N i i=1 density.ferent We power-law also split indices the sample of theinto parent CC proton and NCC population, systems com- $ puted using Eq. 10 and assuming that the CR have a flatter where N is the total number of galaxy clusters in the sam- radial distribution, !(r) r0.5. ple. Then, an upper limit on the CR-to-thermal energy ratio ∼ can simply be obtained by taking the ratio of the observed Γp Γph CC NCC ALL flux to the equipartition flux, 2.10 2.0 4.71.32.0 2.55 2.4 3.80.60.7 ULFermi(Γ) !(Γ)= , (11) 3.00 2.8 3.20.60.6 Fequip(Γ) Huber et al. subm. where ULFermi is the upper limit obtained in the previous section. regions, leadingRadial CR to andistribution increasing steeper radial than profile the (Vazza gas et al. 2012b). In Table 3 we give upper limits on the average CR- to-thermal energy ratio, assuming a profile increasing like 5.2.2. Results Q: Are we starting to probe hadronic models for mini-halos? !(r) r0.5. Conversely, if CR are injected by outbursts We compute the equipartition flux for each galaxy clus- of the∼ central AGN, a steeper CR profile is expected with ter by solving Eq. 9 using the parameters derived by Chen respect to the gas (Fujita & Ohira 2011). In Table 4 we et al. (2007). We note that among the 53 systems compris- 0.5 provide the results assumingD. Eckert !(Januaryr) r 15,− 2013. ing our sample, 32 are classified as NCC, while 21 exhibit ∼ CC properties. As an estimate of the virial radius Rvir, we use the scaling relations of Arnaud et al. (2005), mak- 6. Discussion ing the approximation Rvir R200. For indication, in our sample of 53 galaxy clusters,∼ the average values for the 6.1. Comparison with previous observational results 2 3 parameters are R200 1.5Mpc,ngas0 10− cm− , rc Using data obtained by the EGRET experiment on board ∼ ∼ ∼ 210 kpc, β 0.63, and kT 4.6keV.Wenotethatamong the Compton Gamma-Ray Observatory, Reimer et al. ∼ ∼ our sample CC clusters typically have a lower temperature (2003) performed a similar study to the one presented in (TCC 3.2 keV), and thus a lower mass, than NCC systems this paper, reaching upper limits on the CR-to-thermal en- ∼ (TNCC 5.5keV). ergy ratio of the order of 10-20%. Thanks to the excellent ∼ Our results on the CR-to-thermal energy ratio for the sensitivity of Fermi /LAT, our upper limit lies more than different subsamples and photon indices are given in Table an order of magnitude below the upper limit obtained in 2. The limits provided here can be readily transformed into this study. Recently, Ackermann et al. (2010) performed a the pressure ratio (χ) using the relation UCR =2PCR in Uth Pth similar analysis on individual systems, always resulting on the fully relativistic case. So for the CC sample, our best gamma-ray upper limits. This study allowed to constrain limit in term of the pressure ratio is χ =2.3%, obtained the CR energy density at the level of a few percent of the for a proton index of 3. For the NCC sample, the best limit thermal energy density, in the best cases. Similar results is χ =0.4%, obtained for proton indices softer than 2.55, were recently obtained on the Coma cluster through a com- while for the total sample, the best limit is of χ =0.4%, bination of Fermi and VERITAS data (Arlen et al. 2012), again for the softest spectral index. which led to an upper limit of the order of 2% on the CR- In addition to the isobaric case, we also used different to-thermal pressure ratio. Compared to these studies, our CR radial distributions, representing a decreasing and an analysis brings constraints on the average population which increasing pressure ratio. Indeed, injection of CR at cos- are slightly below the few best cases for individual systems. mological shocks would occur preferentially in the outer The stacking method therefore allows us to bring the con-

8