Draftversion January22,2014 PreprinttypesetusingLATEXstyleemulateapjv.04/17/13 CAN AGN FEEDBACK BREAK THE SELF-SIMILARITY OF GALAXIES, GROUPS, AND CLUSTERS? M. Gaspari1,2,5, F. Brighenti2,3, P. Temi4, S. Ettori5,6 1MaxPlanckInstitute forAstrophysics,Karl-Schwarzschild-Strasse1,85741Garching,Germany;[email protected] 2AstronomyDepartment, UniversityofBologna,ViaRanzani 1,40127Bologna,Italy 3UCO/LickObservatory,DepartmentofAstronomyandAstrophysics,UniversityofCalifornia,SantaCruz,CA95064, USA 4AstrophysicsBranch,NASA/AmesResearchCenter,MS245-6,MoffettField,CA94035 5INAF,OsservatorioAstronomicodiBologna,viaRanzani 1,40127Bologna,Italy 4 6INFN,SezionediBologna,vialeBertiPichat6/2,40127Bologna,Italy 1 Draft version January 22, 2014 0 2 ABSTRACT n It is commonly thought that AGN feedback can break the self-similar scaling relations of galaxies, a groups, and clusters. Using high-resolution 3D hydrodynamic simulations, we isolate the impact J of AGN feedback on the L − T relation, testing the two archetypal and common regimes, self- x x 0 regulated mechanical feedback and a quasar thermal blast. We find that AGN feedback has severe 2 difficulty in breaking the relation in a consistent way. The similarity breaking is directly linked to the gas evacuation within R , while the central cooling times are inversely proportionalto the core 500 ] density. Breaking self-similarity implies thus breaking the cool core, morphingallsystemstonon-cool- O core objects, which is in clear contradiction with the observed data populated by several cool-core C systems. Self-regulated feedback, which quenches cooling flows and preserves cool cores, prevents . the dramatic evacuationand similarity breakingat any scale; the relationscatter is also limited. The ph impulsivethermalblastcanbreakthecore-includedLx−TxatT500 <∼1keV,butsubstantiallyempties and overheats the halo, generating a perennial non-cool-core group, as experienced by cosmological - o simulations. Even with partial evacuation, massive systems remain overheated. We show the action r of purely AGN feedback is to lower the luminosity and heating the gas, perpendicular to the fit. t s Keywords: galaxies: active — galaxies: clusters: intracluster medium — galaxies: groups: general — a galaxies: jets — hydrodynamics — methods: numerical [ 1 1. INTRODUCTION ening in the group regime, ∼4-5 (Mulchaey et al. 2003; v Osmond & Ponman 2004; Helsdon & Ponman 2000a,b; 7 In the last decade, feedback due to active galactic nuclei Sun et al. 2009; Sun 2012; see Fig. 1). 6 (AGN) has allowed to solve crucial astrophysical prob- In recent years, different authors have studied the 0 lems. The supermassive black hole (SMBH) at the cen- scaling relations by means of large cosmological sim- 5 ter of galaxies,groups, and clusters can indeed release a . terrific amount of energy (>1061 erg), providing an effi- ulations with AGN feedback (e.g., Sijacki et al. 2007; 1 Fabjan et al. 2010; McCarthy et al. 2010; Short et al. cient source to quench cooling flows and star formation 0 2010). In general, they find that the implemented AGN 4 (McNamara & Nulsen 2007). In particular, mechanical feedback is able to break the self-similarity, lowering lu- 1 AGNfeedbackintheformofjets/outflowsisabletoreg- minosities by orders of magnitude and, surprisingly, de- : ulate for several Gyr the thermodynamical state of the v creasing the global temperature (Puchwein et al. 2008, systemcore(Gaspari et al.2013aforareview). However, i fig.2). Oftenoverlooked,thesimulatedsystemsarehow- X it is far from clear if AGN feedback is able to strongly ever non-cool-core objects (Planelles et al. 2013, fig. 7, modify the large-scale gas halo, as the total X-ray lu- r foracriticaldiscussion). Evenno-feedbackrunsproduce a minosity and temperature, in other words, breaking the negativetemperaturegradients,duetoextremeadiabatic self-similar scaling relations. heating,whicharenotpresentinhigh-resolutionsimula- If gravity were the single driver of the evolution tions(e.g.,Li & Bryan2012). Besidestheunder-resolved (Kaiser 1986; Kravtsov& Borgani 2012), all systems black hole/feedback physics and subgrid numerics (see would scale only with mass, M = (4π/3)∆ρ R3, ∆ c ∆ the analysis in Barai et al. 2014), it remains difficult to where ∆ is the chosen overdensity. The critical den- disentangleandisolatetheactionoffeedbackinthecom- sity of the universe evolves in redshift as ρ (z) ∝ c plex evolutionshaped by mergers,filaments, star forma- E2(z), where E2(z) ≃ Ω (1 + z)3 + Ω , giving a m Λ tion, sink particles, and other prescriptions. characteristic radius R∆ ∝ M∆1/3E−2/3(z). Via hy- The objective of this study is to critically examine drostatic equilibrium (M∆ ∝ T R∆), we can retrieve if AGN feedback itself can break the self-similarity of M ∝ T3/2E−1(z). Since the bolometric X-ray lumi- galaxies, groups, and clusters, in a way consistent with ∆ nosityscalesasL ∝n2T1/2R3 (inthe Bremsstrahlung observations. Via controlled high-resolution 3D (mesh) x x ∆ simulations, we study how the main scaling, L −T is regime), using gas number density n ∝ M /R3 ∝ x x ∆ ∆ shaped by the two archetypal and commonly adopted ρ (z) and the above relations, we find the well-known c feedback models, self-regulated kinetic feedback and a self-similar scaling L ∝ T2E(z). However, cluster x x quasarthermalblast. Inaforthcomingwork,weexplore observations show a slope steeper than 2 (∼3; e.g., othermodelsanddifferentrelations(Gasparietal. 2014, Pratt et al. 2009; Maughan et al. 2012), further sharp- 2 M. Gaspari et al. in prep.). The normalization of the density profile is set by the Acrucialconstraintdrivenbyobservationsisthepres- gas fraction at the virial radius, f ≃ 0.15. The gas,vir ence of a (strong or weak) cool core in the majority initial gas fraction is intentionally high, near the cosmic of observed systems (>∼ 65 per cent, Peres et al. 1998; value (Planck Collaboration et al. 2013), since we want Mittal et al. 2009; Sun et al. 2009; Hudson et al. 2010; to testif AGN feedbackis the originalcauseof gasevac- Zhao et al. 2013). Such systems show, in the core, uation and hence self-similarity breaking. Using lower cooling times < 7 Gyr, positive temperature gradients, values (e.g., 0.1) does not change the conclusions. and low gas entropy (<10s keV cm2). Moreover, cool cores appear to be long lived and in place since z > 1 2.2. Hydrodynamics, cooling and heating (McDonald et al. 2013). AGN feedback, or inside-out Using FLASH4 code, we integrate the 3D equations heating, intrinsically evacuates the central regions, be- of hydrodynamics in conservative form, including to- fore touching the periphery of the system. Although tal gravity, gas radiative cooling, and feedback heat- the AGN feedback energetics is in principle capable ing. The latter two source terms are implemented fol- to breaking the group self-similarity (Cavaliere & Lapi lowing the unified self-regulation model, as described in 2008;Giodini et al.2010),theenergydepositionandhy- Gaspari et al. (2012b). Transport mechanisms, as con- drodynamics is crucial. We show that breaking self- duction,arenotincludedsincedatasuggestastrongsup- similarity via AGN feedback implies disrupting the cool pression (e.g., Gaspari & Churazov 2013). The cubical core, morphing the system into perennial non-cool-core box fully coversthe virial radius, ∼1.2-4.6Mpc (groups objects;viceversa,self-regulationpreservesthecoreand to clusters). We use concentric grid levels with radius the large-scale structure. of ∼60 cells, centered on the BCG, where the maximum resolution reaches ≈ 290 pc. The system is integrated 2. PHYSICS&NUMERICS for at least 5 Gyr. Boundary conditions are set in diode 2.1. Initial conditions mode. Inordertofullyisolatetheroleoffeedbackinalteringthe 2.2.1. Self-regulated mechanical feedback scalingrelations,westartwithavirializedgroup/cluster In Gaspari et al. (2011a,b, 2012a,b, 2013a,b) was found havingaformedcoolcore,whichcharacterizesthemajor- ity of observed systems (§1). Groups and clusters share that the most consistent model able to solve the cool- ing flow problem is mechanical feedback, self-regulated many common properties, allowing to build an initial bycoldaccretion. Inturbulentregionswherethecooling ‘universal’ system defined only by its mass. Following Vikhlinin et al. (2006)1, the observed average tempera- time dropsbelow∼10×the free-falltime, thermalinsta- bilities become quickly nonlinear,leadingto the conden- ture profile can be modeled as sationofcoldgas outof the hot phase. Such coldclouds 0.45+(rˆ/0.045)1.9 1 andfilamentscollideinaninelasticandchaoticwaywhile T(r)=T0 1+(rˆ/0.045)1.9 (1+(rˆ/0.6)2)0.45, (1) rainingonto the blackhole, boostingthe accretionrate. Bipolar massive sub-relativistic outflows are then trig- where rˆ= r/R ; the normalization is T ≃ 1.4 T ≃ geredwithkineticpowerproportionaltothecentralcool- 500 0 500 3 keV(M500/1014M⊙)0.6 (cf., Sun et al. 2009). Eq. 1 ing rate, Pjet = ǫM˙coolc2 (Gaspari et al. 2012b for the models the positive gradient of the cool core and the numerical details), with optimal mechanical efficiencies gentle decrease at large r; the peak temperature (r ∼ ǫ ∼ 5 × 10−4 − 5 × 10−3. The self-regulated outflow 0.15R ) is ∼2× the central value, which is reached generates the cocoon shock, two buoyant bubbles, and 500 again at r ∼ R . Albeit some groups have slightly gas/metal uplift. The kinetic feedback rises the central 500 higher T peak and steeper decrease (Sun et al. 2009), gas entropy, quenching cooling and stifling the accretion such minor differences have no impact on the results. rate; the self-regulated loop starts then over again. The system is initially in hydrostatic equilibrium The gentle self-regulation with either kinetic or ther- within the gravitationalpotential φ, dominated by dark mal injection (the latter commonly used in cosmologi- matter,modeledviatheusualNFWprofileintheconcor- calsimulations2)producesanalogousimpactonthescal- dance ΛCDM universe. The halo concentrationis linked ing relations, although thermal feedback induces again to the virial mass as c ≃ 8.5(Mvir/1014 M⊙)−0.1 (e.g., excessive core overheating (Brighenti & Mathews 2003; Bullock et al. 2001). In addition, each group/cluster is Gaspari et al. 2011b). dominatedbyacentralmassiveellipticalgalaxy(‘BCG’), modeled with a de Vaucouleurs stellar density profile. 2.2.2. Quasar thermal blast The BCG K-band luminosity increases with the halo In the opposite spectrum of feedback models resides the mass as LK ≃ 4.7 ×1011(Mvir/1014 M⊙)0.39L⊙ (e.g., sudden and powerful release of thermal energy. This Lin & Mohr 2004). The stellar mass is then retrieved can be justified by a quasar event, emitting large ra- adoptingM∗/LK ∼1(e.g.,Mannucci et al.2005). Since diative power absorbed by highly dense clouds, or in BCGs are large ellipticals, we keep the effective radius alternative, by an Eddington wind fully thermalizing Reff ≃9kpc. AsdescribedinGaspari et al.(2012b),the in the inner core. Numerically, the thermal energy is BCG injects a low amount of energy and mass due to injected in the inner ∼4 kpc, with Eddington power SNIa and stellar winds; however, the energetics is domi- PEdd ≃1.5×1047(Mbh/109M⊙)ergs−1 lasting∼6Myr. nated by the AGN feedback. ThetotalreleasedenergyisE ≡ηM c2 ≃3×1061 AGN bh 1 Note a typo in the published version, missing the 0.45 expo- 2Insomecosmologicalworks,the‘quasarmode’issimplyquasi- nent;A.Vikhlinin,privatecommunication(seeastro-phversion). continuous thermalfeedback. Breaking galaxies, groups, and clusters with AGN feedback 3 Figure 1. X-ray bolometric luminosity versus X-ray temperature, including (top) or excising (middle) the core, r < 0.15R500. The bottom panels show the central coolingtime(∼15-20 kpc), withthecool-core thresholdtcool∼7Gyr. Left: self-regulatedkineticmodels (§2.2.1;ǫ=5×10−3). Right: quasarthermalblast(§2.2.2);theblackpointsshowthesimulated5Gyrevolutionevery500Myr(theinitial point has magenta contour). The observational data are from Maughanetal. (2012, Chandra; red), Prattetal. (2009, XMM; green), Sunetal. (2009,Chandra;blue),Mulchaeyetal.(2003)andOsmond&Ponman(2004;magenta),Helsdon&Ponman(2000a,b,ROSAT; cyan); wealways useh=0.7(e.g., forLx ∝h−2). Theself-regulatedAGN feedback prevents overheating, at thesametimeavoidingthe self-similaritybreaking. Conversely,thepowerfulthermalblastcanbreaktheLx−Txrelationatthegroupscale,butmorphingthesystem intoaperennialnon-cool-coreobject. 4 M. Gaspari et al. erg, i.e., the characteristic energy of a SMBH with typi- points towards lower luminosities and higher tempera- calMbh ∼109M⊙ andradiativeefficiencyη ≃1.5×10−2 tures. Similarly, radiative cooling moves the system to- (Novak 2013). The isotropic blast triggers once the nec- wards higher L and lower T (cf., Ettori & Brighenti x 500 essary mass of cold gas has been accreted; due to the 2008). In other words, the L −T secularly moves due x x powerful heating, no second event ever occurs. Conclu- to heating/cooling perpendicular to the fit, particularly sionsareunalteredwithdifferentE andcompactde- as the core is included. Works based on cosmological AGN position windows. simulations (§1) show instead a decrease of T , adding 500 Boosting ǫ above the optimal values (e.g., ∼ 0.1) subgrid AGN feedback. This could be linked to the re- transformsthe previousgentle self-regulatedfeedback in duction of the extreme adiabatic heating present in the the impulsive blast. Conversely, significantly lowering under-resolved pure cooling flow. A more serious prob- E morphs the feedback in the self-regulated regime. lem is that, although the similarity breaking occurs, no AGN The presented models constitute thus the two opposing object can be observationally described as a cool core archetypes of inside-out feedback. with the positive T gradient depicted in §2.1. Theself-regulatedmodelsshowthatavoidingthecom- 3. RESULTS pleteself-similaritybreakingimpliespreservingthe cool- Figure 1 presents the key results of the high-resolution core structure – at the same time reducing the cool- hydrodynamic simulations, testing kinetic or thermal ing rate below 10 per cent of that of the pure cooling AGN feedback in the range of systems with T ≃0.5- flow. In the bottom panel (left), the central cooling 500 6 keV (Mvir ∼ 1013-1015 M⊙). In the top and mid- timestaysinanyhalobelow∼7Gyr,thecommonupper dle panels, we show the X-ray luminosity versus X-ray limitusedtodefinecool-coresystems(e.g.,Hudson et al. temperature3 within R , including or excisingthe core 2010). These systems also preserve the positive tem- 500 (r < 0.15R ), respectively. The bottom panel depicts perature gradient and low central entropy. As found in 500 the gas central cooling time (in the shell ∼15-20 kpc, Gaspari et al. (2011b, 2012b), ǫ ≃ 5×10−3 is the best contained within <∼ 0.06R500). The Lx − Tx relation value for clusters; indeed, the two groups switch to a is shaped by the global amount of cooling and heating, weaker cool core after the initial heating. Overall, op- while t ∝T/nΛ assesses the core thermal state (Λ is timal self-regulation induces the system to oscillate be- cool the cooling function; see Gaspari et al. 2012b). tween a state of weak and strong cool core, preventing The self-similar relation is expected to be L ∝ T2, both the cooling and heating runaway. The duty cycle x x even shallower in the group regime due to line emission is very efficient with cold accretion, while much weaker (L ∝ T3/2Λ ∝ T ). Figure 1 (top) reveals that the ob- with hot Bondi regulation (Gaspari et al. 2011b). Since x x x most of the observed systems host a cool core (§1), self- servationaldatarelativetotheclusterregime(red,green) regulated mechanical heating represents the long-term are already deviating, with a slope α∼3. Below 2 keV, maintenance mode of AGN feedback, while avoiding the i.e., for small and massive groups, the relation steepens break of the scaling relations. further,reachingα∼ 4-5,with amuchmoresignificant Wetestnowthe otherextremeofAGNfeedbackmod- scatter. Excising the core in both quantities reduces the els, i.e., the impulsive thermal blast (Fig. 1, right). The scatter and the steepness (α ∼ 2.5), avoiding an abrupt suddenisotropicenergyrelease(∼3×1061 erg,compara- decline. The relation becomes nearly self-similar consid- bletothatofatypicalSMBH)candramaticallyevacuate ering only cool-coreclusters (Maughan et al.2012). Un- the atmosphere of the compactgroup,decreasing the X- fortunately, the excised relation for small groups is not ray luminosity by 2.5 orders of magnitude; T initially covered by observational data. 500 increases by ∼2× (the rightmost circle). The strong In the left column, the simulations (black; the ini- breakingoccursbecause the grouptotalbinding energy4 tial state has magenta contour) show that the impact of self-regulated jet feedback (ǫ = 5 × 10−3) on the is ≃ 1061 erg. After the initial blast, the BCG is slowly replenished by the stellar mass loss, mildly increasing L − T relation is limited. The remarkable aspect is x x the luminosity (<R ) to ∼ 1041 erg s−1, while restor- that no break – a deviation of orders of magnitude – 500 ing T near the initial value (this is not due to the ac- occurs, even at the scales of small halos. In the mas- 500 tionofAGNfeedback). Thesystemhascompletelymor- sivegroup(pentagons),the maximumdeviationinlumi- phed. The central t (bottom) is always >60 Gyr, in nosity/temperature is ∼30/10 per cent (0.1/0.05 dex), cool a perennial non-cool-core state. Excising the core (mid- strongly diminishing to ∼10/1 per cent in massive clus- dle), aggravates the breaking, since the compact group ters. In the compact group (circles), the luminosity de- is substantially devoid of gas outside the BCG, with an creases by ∼2.8×, while temperature increases by ∼80 unrealistic drop down to ∼1040 erg s−1 (cf., Sun 2012). per cent. T showsthe weakerscatter;the mainaction 500 Overall, the inside-out heating able to fully break self- of feedback, especially kinetic, is to evacuate gas and, similarity in the core-included L −T , violently alters secondarily, to heat the global atmosphere. x x the excised L −T relation, which is instead observed Excising the core within 0.15R (middle panel), sig- x x nificantly reduces the scatter to5<∼001/3 of the previous to have tighter scatter. In the massive group (pentagons), the total binding values. By removing the core separately for each vari- energy is ≃ 1062 erg (> E ). The thermal blast able, we see that internal heating can only move the AGN can thus only partially evacuate the gas from the core 3 Wecomputedboththeemission-weightedT500 withChandra (∼ 0.1R500), halted by the extended atmosphere before sensitivity(Tx>∼0.3keV;Gasparietal.2012b)andspectroscopic- reaching R500. The result is a decrease in luminosity by liketemperature(Vikhlinin2006);sinceourflowisnotmultiphase, tthheeydaiffreerveenrcyesiwmitihlart.heAsspinheorbicsaerlvvaatliuoensi,swmeinuoser.theprojectedT500; 4 Equaltothegravitational energyEb=R0RvirρgasφdV. Breaking galaxies, groups, and clusters with AGN feedback 5 maximum 0.9 dex and T oscillating within 0.1 dex. one order of magnitude at large radii implies decreas- 500 The extended evacuation is confirmed by the excised re- ing the core density n by at least 10× more, inducing 0 lation. Both simulated L −T are consistent with the t ≫ t . The problem is further aggravated by the x x cool H observed data. However, the similarity deviation occurs increaseoftemperature(t ∝T1/2). Thedirectaction cool againattheexpenseofthecoolcore. Infact,thecentral of AGN feedback is to lower the luminosity and heating cooling time stays above tens Gyr, signaling a strong thegas,notmovingthesystemparalleltotheL −T fit. x x non-cool-core group. Analyzing the poor and massive Overall,AGNfeedbackappearsnaturallysuitedtoregu- cluster simulations (squares, triangles), we see no sim- latethethermodynamicalstateofcosmicsystems,inthe ilarity breaking, due to the larger Eb. The maximum core, but not over large radii (r >∼0.2R500). We remark deviation in the core-included relation is highly limited, that any feedback mechanism, that is able to break the ∼0.17/0.08 dex in Lx and ∼0.03/0.015 in T500, for the self-similarity, needs to properly solve the cooling flow poorandmassivecluster,respectively. Excisingthe core problem. stifles the scatter by at least 1/3. The powerful AGN In a forthcoming work, we discuss other heating mod- heating has again the side effect of destroying the core els, parameters, and scaling relations. We found, nev- (tcool ≫ tH). Only the massive cluster can partially re- ertheless, that AGN feedback models fall in the two cover after several Gyr. Cool cores are common in the archetypalcategoriespresentedhere: self-regulatedheat- universe,hencethistypeofbreakingshouldberare. No- ing,preventingthebreaking,orstrongimpulsiveheating, ticethatcombiningthe twofeedbackmechanismsaggra- which breaks the scaling relations but destroys the core. vates the core overheating. For instance, using either thermal or kinetic feedback The global Lx−Tx property seems overallmore likely with self-regulation has the same minor impact on the linked to a primordial imprint that the group/cluster scaling relations (§2.2.1). Injecting energy in the center experienced, rather than an internal breaking after for- or at a distance of a few 10 kpc (<∼ 0.05R500) has also mation. However, we note that when the external ‘pre- thesameeffect,consideringthatthefeedbackmustaffect heating’athighredshift–whoseagencyisstillunclear– extremely large regions (R , several 100s kpc); a too 500 is high enough to bring Lx consistent with observations, distant injection allows instead central runaway cooling. thegasentropybecomesusuallytoohigh,inhibitingcool Further, boosting ǫ transforms the self-regulated feed- corestoformabinitio(e.g.,Brighenti & Mathews2001). back into the quasar-like blast; vice versa, diminishing the impulsive E slightly lowers t , but prevents 4. CONCLUSIONS AGN cool the similarity breaking. In other words, even with a dif- We showed that AGN feedback has severe difficulty in ferent parametrization of the archetypal AGN feedback breaking the self-similarity of galaxies,groups,and clus- models,breakingself-similarityimpliesbreakingthecool ters, in a consistentway. Via high-resolution3D simula- core. tions,weisolatedtheimpactofthetwocommonregimes ofAGNfeedbackontheprincipalscalingrelationLx−Tx. ACKNOWLEDGMENTS • Self-regulated kinetic feedback prevents the simi- The FLASH code was in part developed by the DOE larity breaking, inducing a limited scatter (<∼0.1 NNSA-ASC OASCR Flash center at the University dex). Self-regulation allows to properly quench of Chicago. MG is grateful for the financial sup- cooling flows preserving the cool-core structure; port provided by the Max Planck Fellowship. SE avoiding overheating translates thus in a modest and FB acknowledge financial contribution from ASI- central gas evacuation, maintaining low core INAF I/009/10/0, PRIN INAF 2012, PRIN MIUR cooling times (t < 7 Gyr) and avoiding the 2010LY5N2T. High-performance computing resources cool L − T breaking at any halo scale. Since the wereprovidedbytheNASA/AmesHECProgram(SMD- x x majority of observed systems display a cool core 13-3935, SMD-13-4373, SMD-13-4377; Pleiades). We (§1), this mode should represent the long-term thank M. Sun for providing the groups data, E. Chu- maintenance phase of AGN feedback. razovandtheanonymousrefereeforinterestinginsights. REFERENCES • An impulsive quasar thermal blast, injecting the totalenergyofatypicalSMBH,isabletobreakthe Barai,P.,Viel,M.,Murante,G.,Gaspari,M.,&Borgani,S.2014, core-includedLx−TxatscalesT500 <∼1keV(where MNRAS,437,1456 EAGN >∼ Eb). However, after full breaking, the B—r.ig2h0e0n3t,i,AFp.J,,&58M7,at5h8e0ws,W.G.2001,ApJ,553,103 systemisalmostdevoidofgas,alsoatlargeradii,in Bullock,J.S.,Kolatt,T.S.,Sigad,Y.,etal.2001,MNRAS,321, contradiction with the core-excised relation. Even 559 with partial evacuation (Mvir >∼5×1013 M⊙), the Cavaliere,A.,&Lapi,A.2008, ApJ,673,L5 central t is raised to several times the Hubble Ettori,S.,&Brighenti,F.2008,MNRAS,387,631 time. Inccoloulsters,thescatterisagainlimited,<∼0.2 Fa1b6j7a0n,D.,Borgani,S.,Tornatore,L.,etal.2010, MNRAS,401, dex. The imprint of the thermal blast is indelible, Gaspari,M.,Ruszkowski,M.,&Sharma,P.2012a, ApJ,746,94 morphingthesystemintoaperennialnon-cool-core Gaspari,M.,Brighenti,F.,&Temi,P.2012b,MNRAS,424,190 object. If existent, such a mechanism should be Gaspari,M.,Brighenti,F.,&Ruszkowski,M.2013a, rare or occurring at very high redshift. AstronomischeNachrichten, 334,394 Gaspari,M.,&Churazov,E.2013,A&A,559,A78 Breaking self-similarity via inside-out heating means Gaspari,M.,Ruszkowski,M.,&Oh,S.P.2013b, MNRAS,432, to evacuate most of the gas from the region <∼R500. Ga3s4p0a1ri,M.,Melioli,C.,Brighenti,F.,&D’Ercole,A.2011a, Since central tcool ∝ n−01, lowering the gas density by MNRAS,411,349 6 M. Gaspari et al. Gaspari,M.,Brighenti,F.,D’Ercole,A.,&Melioli,C.2011b, Mulchaey,J.S.,Davis,D.S.,Mushotzky,R.F.,&Burstein,D. MNRAS,415,1549 2003,ApJS,145,39 Giodini,S.,Smolˇci´c,V.,Finoguenov, A.,etal.2010, ApJ,714, Novak,G.S.2013,arXiv:1310.3833 218 Osmond,J.P.F.,&Ponman,T.J.2004,MNRAS,350,1511 Helsdon,S.F.,&Ponman,T.J.2000a, MNRAS,319,933 Peres,C.B.,Fabian,A.C.,Edge,A.C.,etal.1998,MNRAS, —.2000b,MNRAS,315,356 298,416 Hudson,D.S.,Mittal,R.,Reiprich,T.H.,etal.2010,A&A,513, PlanckCollaboration,Ade,P.A.R.,Aghanim,N.,etal.2013, A37 arXiv:1303.5076 Kaiser,N.1986, MNRAS,222,323 Planelles,S.,Borgani,S.,Fabjan,D.,etal.2013,MNRAS,2889P Kravtsov,A.V.,&Borgani,S.2012, ARA&A,50,353 Pratt,G.W.,Croston,J.H.,Arnaud,M.,&B¨ohringer,H.2009, Li,Y.,&Bryan,G.L.2012,ApJ,747,26 A&A,498,361 Lin,Y.-T.,&Mohr,J.J.2004, ApJ,617,879 Puchwein,E.,Sijacki,D.,&Springel,V.2008,ApJ,687,L53 Mannucci,F.,DellaValle,M.,Panagia,N.,etal.2005,A&A, Short,C.J.,Thomas,P.A.,Young,O.E.,etal.2010,MNRAS, 433,807 408,2213 Maughan,B.J.,Giles,P.A.,Randall,S.W.,Jones,C.,& Sijacki,D.,Springel,V.,DiMatteo,T.,&Hernquist,L.2007, Forman,W.R.2012, MNRAS,421,1583 MNRAS,380,877 McCarthy,I.G.,Schaye, J.,Ponman,T.J.,etal.2010, MNRAS, Sun,M.2012,NewJournalofPhysics,14,045004 406,822 Sun,M.,Voit,G.M.,Donahue, M.,etal.2009,ApJ,693,1142 McDonald,M.,Benson,B.A.,Vikhlinin,A.,etal.2013,ApJ, Vikhlinin,A.2006,ApJ,640,710 774,23 Vikhlinin,A.,Kravtsov,A.,Forman,W.,etal.2006,ApJ,640, McNamara,B.R.,&Nulsen,P.E.J.2007,ARA&A,45,117 691 Mittal,R.,Hudson,D.S.,Reiprich,T.H.,&Clarke,T.2009, Zhao,H.-H.,Jia,S.-M.,Chen,Y.,etal.2013,ApJ,778,124 A&A,501,835