A first Chandra view of the cool core cluster A1668: offset cooling and AGN feedback cycle
Thomas Pasini, Myriam Gitti, Fabrizio Brighenti, Ewan O'Sullivan, Fabio Gastaldello, Pasquale Temi, Stephen Hamer
DDraft version February 24, 2021
Typeset using L A TEX twocolumn style in AASTeX63
A first Chandra view of the cool core cluster A1668: offset cooling and AGN feedback cycle
T. Pasini , M. Gitti ,
2, 3
F. Brighenti , E. O’Sullivan , F. Gastaldello , P. Temi , andS. L. Hamer Hamburger Sternwarte, Universit¨at Hamburg, Gojenbergsweg 112, 21029 Hamburg, Germany Dipartimento di Fisica e Astronomia (DIFA), Universita‘ di Bologna, via Gobetti 93/2, 40129 Bologna, Italy Istituto Nazionale di Astrofisica (INAF) – Istituto di Radioastronomia (IRA), via Gobetti 101, I-40129 Bologna, Italy Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA02138, USA INAF-IASF Milano, via E. Bassini 15, I-20133 Milano, Italy Astrophysics Branch, NASA/Ames Research Center, MS 245-6, Moffett Field, CA 94035 Department of Physics, University of Bath, Claverton Down, BA2 7AY, UK (Accepted 2021/02/18)
Submitted to ApJABSTRACTWe present a multi-wavelength analysis of the galaxy cluster A1668, performed by means of newEVLA and
Chandra observations and archival H α data. The radio images exhibit a small centralsource ( ∼
14 kpc at 1.4 GHz) with L ∼ · W Hz − . The mean spectral index between 1.4GHz and 5 GHz is ∼ -1, consistent with the usual indices found in BCGs. The cooling region extendsfor 40 kpc, with bolometric X-ray luminosity L cool = 1 . ± . · erg s − . We detect an offset of ∼ ∼ α andthe X-ray peaks. We discuss possible causes for these offsets, which suggest that the coolest gas is notcondensing directly from the lowest-entropy gas. In particular, we argue that the cool ICM was drawnout from the core by sloshing, whereas the H α filaments were pushed aside from the expanding radiogalaxy lobes. We detect two putative X-ray cavities, spatially associated to the west radio lobe (cavityA) and to the east radio lobe (cavity B). The cavity power and age of the system are P cav ∼ × erg s − and t age ∼ Keywords: galaxy clusters, AGN, AGN feedback, offset, A1668, cooling flow INTRODUCTIONIn the last two decades, our understanding of theevolution of cool core galaxy clusters has led to a pic-ture in which the cooling of the
Intra-Cluster Medium (ICM), the cold gas accreting onto the
Brightest ClusterGalaxy (BCG), and the feedback from the central radiosource give birth to a tightly-connected cycle, knownas Active Galactic Nuclei (AGN) feedback loop (for re-views see e.g. McNamara & Nulsen 2007; Gitti et al.
Corresponding author: Thomas [email protected] a r X i v : . [ a s t r o - ph . GA ] F e b Pasini et al.
Recently, a number of studies have revealed stronglinks between the central BCG, the X-ray core and thecluster dynamics (Sanderson et al. 2009; Hudson et al.2010; Rossetti et al. 2016). In particular, spatial off-sets between the BCG, the H α line emission and theX-ray emission peak (e.g. Haarsma et al. 2010; Hameret al. 2012, 2016; Barbosa et al. 2018) suggest that ICMsloshing and offset cooling, together with the AGN, canhave a significant influence on the cluster evolution. In-deed, all these elements affect the activity of the centralSupermassive Black Hole (SMBH) through motions ofthe gas, that could be able to regulate the cavity pro-duction and, consequently, the feedback cycle, since theICM oscillates back and forth with respect to the centralSMBH.This was recently discussed in Pasini et al. (2019) forthe cool core cluster A2495. Spatial offsets have beenobserved in this cluster, with the X-ray peak being sep-arated by ∼ ∼ α line emission peak. The analysis presented bythe authors on two putative systems of X-ray cavities,hinted at in the shallow ( ∼ Chandra observation,suggests that even if cooling is not depositing gas ontothe BCG core, the coupling between the AGN poweroutput and the cooling rate is still consistent with theobserved distribution for cluster samples. In a forth-coming publication we will present the detailed analysisof the deeper
Chandra observations of A2495, recentlyallocated ( ∼
130 ks, P.I. Gitti ), which will be key toprobe the presence of two pairs of ICM cavities and testthe proposed scenario that the feeding-feedback cycle isnot broken.A1668 was selected, along with A2495, from theROSAT Brightest Cluster Sample (BCS; Ebeling et al.1998) by choosing objects with X-ray fluxes greater than10 − erg cm − s − and, among these, by selecting thosecharacterized by logL H α >
40 from the catalogue ofCrawford et al. (1999). Of the obtained sample of 13objects, A2495 and A1668 still lacked
Chandra obser-vations, that were obtained jointly with new VLA data(P.I. Gitti ). Pasini et al. (2019) have presented the re-sults for A2495, making also use of H α line emission dataand Hubble Space Telescope (HST) archival images. Inthis work we combine the A1668 VLA and
Chandra newobservations in order to study the interactions betweenthe radio source hosted in the BCG and the ICM. Aswell as for A2495, we included H α line emission datafrom Hamer et al. (2016); on the other hand, no HST Proposal Number 22800391 Proposal Number 12800143 data are available for this cluster.A1668 was previously observed in the radio band byTGSS (TIFR GMRT Sky Survey), which gives an esti-mate for the 150 MHz flux density of 1589 ±
159 mJy;Hogan et al. (2015) performed a 5 GHz radio analy-sis (the data they used are not the same presented inthis work), estimating a flux density of 21.0 ± ± (cid:39) · M (cid:12) (Andreon 2016) and M =3.9 ± . . · M (cid:12) (Pulido et al. 2018). The cluster’sBCG, IC4130, shows a Star Formation Rate (SFR),estimated from extinction-corrected H α luminosity ob-tained from long-slit observations, of SFR = 2.5 ± (cid:12) yr − (Pulido et al. 2018), and extends for ∼
85 kpc(diameter at the isophotal level of 25 mag/arcsec in theB-band, Makarov et al. 2014). Edwards et al. (2009)also presented IFU observations of the H α emission closeto the BCG, finding a clear velocity gradient from posi-tive values north of the centre to negative values at thesouth. They also argued that the line emitting gas islikely not at rest with respect to the BCG.In this work, we adopt a ΛCDM cosmology with H = 73 km s − Mpc − , Ω M = 1 − Ω Λ = 0.3. The BCGredshift is z = 0.06355 (Hamer et al. 2016) and the lu-minosity distance is 273.7 Mpc, leading to a conversionof 1 arcsec = 1.173 kpc. RADIO ANALYSIS2.1.
Observations and data reduction
IC4130, the BCG of A1668, was observed with theEVLA on 2011 June 17th in the 1.4 GHz band, andon 2011 March 9th in the 5 GHz band, in A and Bconfigurations respectively. Details of the observationsare shown in Table 1.The sources J1331+3030 (3C286) and J1327+2210were used for both the observations as flux and phasecalibrators, respectively. The data reduction was per-formed using the NRAO Common Astronomy SoftwareApplications package (CASA, version 5.3), applying thestandard calibration procedure after carrying out an ac-curate editing of the visibilities with the CASA task
FLAGDATA . We removed about 6% of the target visibili-ties at 5 GHz, whereas at 1.4 GHz the data were highly HyperLEDA catalog. first Chandra view of the cool core cluster A1668 Frequency Number of spw Channels Bandwith Array Total exposure time5 GHz (C BAND) 2 (4832 MHz - 4960 MHz) 64 128 MHz B 3h59m21s1.4 GHz (L BAND) 2 (1264 MHz - 1392 MHz) 64 128 MHz A 2h59m28s
Table 1.
Radio observations properties (project code SC0143, P.I. M. Gitti). contaminated by Radio Frequence Interferences (RFI),thus producing a visibility loss of ∼ CLEAN task on a 7” ×
7” region centered onthe radio source. We took into account the sky curva-ture by setting the gridmode=WIDEFIELD parameter andused a two-terms approximation of the spectral modelexploiting the MS-MFFS algorithm (Rau & Cornwell2011). 2.2.
Results
We produced total intensity radio maps by setting weighting = BRIGGS , corresponding to
ROBUST 0 . Thisbaseline weighting provides the best compromise be-tween angular resolution (determined by long baselines)and sensitivity to extended emission (provided by shortbaselines). The uncertainty on the flux density measure-ments is 5%, estimated from the amplitude calibrationerrors.At 5 GHz (Fig. 1), the radio source exhibits a to-tal flux density of 19.9 ± = (1 . ± . · W Hz − . The rms noiseis 6 µ Jy beam − . The source stretches Eastwards for ∼
11 kpc, with a minor axis of ∼ eq (5 GHz) = 8.7 ± µ G.The 1.4 GHz map (
ROBUST 0 , Fig. 2) shows no sig-nificant differences with respect to the 5 GHz emission.The source flux density is 70.2 ± ∼ µ Jy beam − . The radio source scale is slightlylarger ( ∼
14 kpc for the major axis, ∼ Figure 1.
ROBUST 0 ) of the radio sourcehosted in IC4130, the BCG of A1668. The resolution is 1.14” × µ Jy beam − . Contours are at-3,3,6,12,24,48 · rms. The source flux density is 19.9 ± cal BCG, and the small scale suggests that it can all beaccounted to the AGN/radio galaxy. It is possible thatdiffuse emission larger than our sensitivity scale exists;however, given the extended double-lobe morphology ofthe LOFAR 150 MHz image presented in Bˆırzan et al.(2020), the presence of a mini-halo in A1668 looks un-likely. The equipartition field is H eq (1.4 GHz) = 10.3 ± µ G. Radio properties can be found in Table 2.The radio source hosted in the centre of A1668 can beclassified as a FRI galaxy, as demonstrated by both themorphology (asimmetric lobes, no hotspots) and the 1.4GHz luminosity (L = (6 . ± . · W Hz − ),that place IC4130 in the 70 th percentile of the BCG ra-dio luminosity function presented in Hogan et al. (2015).2.2.1. Spectral index map
The synchrotron spectrum follows a power law S ν ∝ ν α , where α is the spectral index. The spectral in-dex map (Fig. 3) was generated using the CASA task Pasini et al.
Band Flux density rms beam Luminosity Volume Brightness Temperature Equipartition Field[mJy] [ µ Jy beam − ] [arcsec] [10 W Hz − ] [kpc ] [K] [ µ G]5 GHz 19.9 ± ± ±
22 39.6 ± ± ± ± ±
30 1129.3 ± ± Table 2.
Radio properties of A1668 in the two bands observed. The axes of the radio galaxy are a = 10.9 ± b = 5.7 ± a = 14.0 ± b = 7.1 ± σ contours,while for the volume we assumed a prolate elissoid shape. Figure 2.
ROBUST 0 ) of the radio sourcehosted in IC4130. The resolution is 1.44” × µ Jy beam − . Contours are at -3,3,6,12,24,48 · rms. The source flux density is 70.2 ± IMMATH , combining 1.4 GHz and 5 GHz maps producedwith matched weighting=UNIFORM (to enhance the res-olution),
UVRANGE=6.5-152 , and a resolution of 1.4” × UVRANGE was set in order to be sensitive tothe same baselines (thus, physical scales) for both ob-servations.Table 3 lists the peak, the extended and the total ra-dio emission flux densities at 5 and 1.4 GHz, togetherwith the estimated spectral index between the two fre-quencies. The radio core exhibits a flat index ( α (cid:39) α (cid:39) − . ± Figure 3.
Spectral index map between 5 GHz and 1.4 GHzof the radio source hosted in IC4130. Contours are the sameas Fig. 2, and typical errors range from ∆ α (cid:39) α (cid:39) S C ± ∆ S C S L ± ∆ S L α ± ∆ α [mJy] [mJy]Peak 7.5 ± ± ± ± ± ± ± ± ± Table 3.
The first column shows the flux density valuesat 5 GHz (C band), while the second displays the 1.4 GHz(L band) values. The third column presents the corrispon-dent spectral index values. The extended flux density wasestimated as the difference between the total and the peakfluxes. tent with the typical values found in BCGs (Hogan et al.2015). Table 3 summarizes the spectral index properties. X-RAY ANALYSIS3.1.
Observation and data reduction first Chandra view of the cool core cluster A1668
30 kpc . . : : . : . Right ascension D e c li na t i on Abell 1668
Figure 4.
Chandra image of A1668 in the 0.5-2 keV band,smoothed with a gaussian filter with a 3 pixel radius.
A1668 was observed with the
Chandra Advanced CCDImaging Spectrometer (ACIS), with the focal point onthe S3 CCD, in cycle 12 (ObsID 12877, P.I. Gitti) fora total exposure of ∼
10 ks. Data were reprocessedwith CIAO 4.9 (Fruscione et al. 2006) using CALDB4.2.1. We ran the
Chandra repro script to perform thestandard calibration process. After background flare re-moval, we used the
Blanksky template files, filtered andnormalized to the count rate of the source in the hardX-ray band (9-12 keV), in order to subtract the back-ground. The final exposure time is 9979 s, with roughly ∼ ∼
120 kpc radius region(0.5-2 keV) centered on the cluster.Point sources were identified and removed using theCIAO task
WAVDETECT . Making use of optical catalogues,we found that no astrometry correction was necessary.Unless otherwise stated, the reported errors are at 68 %confidence level (1 σ ).3.2. Results
Surface Brightness Profile
In Fig. 4 we show the smoothed 0.5-2 keV image ofA1668. The ICM exhibits a roughly circular and regularmorphology on large scales ( > ∼
35 kpc), while thecluster core shows a region with enhanced emission inthe NE-SW direction. Using the tool
SHERPA (Freemanet al. 2001), a surface brightness profile was producedfrom a background-subtracted, exposure-corrected im-age, making use of 2”-width concentric annuli centered
Figure 5. β -Model fit per-formed in the external 30”-100” interval and extrapolatedto the center, while the red line is the double β -Model fitperformed on every radius. on the X-ray peak. The profile was then fitted with asingle β -Model (Cavaliere & Fusco-Femiano 1976) overthe external 30”-100” (35-120 kpc) interval, in order toexclude the whole core region . The result of the fit( χ /DoF ∼ r0 =10.0 ± . . arcsec ( ∼ beta =0.43 ± . . and central surface bright-ness ampl =0.64 ± . . counts s − cm − sr − .The central brightness excess with respect to the β -Model is a strong indication of the presence of a coolcore in A1668, as we expected from the selection crite-ria described in Sec. 1. This will also be confirmed bythe spectral analysis (see Sec 3.2.2). We therefore fittedthe same profile on the entire radial range with a double β -Model (Mohr et al. 1999; LaRoque et al. 2006), rep-resented with the red line in Fig. 5 ( χ /DoF ∼ r0 =15.6 ± . . arcsec ( ∼ beta =0.67 ± . . and ampl = 0.59 ± . . counts s − cm − sr − forthe first and r0 =0.64 ± . . arcsec, beta =0.42 ± . . and ampl =1.78 ± . . counts s − cm − sr − for the sec-ond β -Model. 3.2.2. Spectral Analysis The assumption of 30”, that was already justifiable through vi-sual inspection, will be furtherly supported, in Sec. 3.2.2, by theestimate of the cooling radius
Pasini et al.
Spectra were extracted with the CIAO task specextract in the 0.5-7 keV band; the extraction wasmade from a series of concentric rings centered on theX-ray peak. Each region contains at least ∼ Blanksky files of each region. We individually fit-ted every spectrum via
Xspec (Arnaud 1996, vv.12.9.1)using a phabs*apec model, approximating an absorbed,collisionally-ionized diffuse gas. The redshift was fixedat z=0.06355 and the hydrogen column density was fixedat N H = 2.20 · cm − (estimated from Kalberla et al.2005). The normalization parameter and the tempera-ture kT were left free to vary. The observation was tooshallow to allow us to fit metallicity, which was insteadkept fixed at a value of 0.3 Z (cid:12) , . Note that these fitsdo not take in to account projection effects. The best-fitting parameters are listed in Table 4. The projectedtemperature profile of A1668 is shown in blue in Fig 6. r min - r max r min - r max Counts kT ± σ kT χ / DoF[arcsec] [kpc] [keV]0 - 12 0 - 14 1346 (99.6 %) 1.74 ± . . ± . . ± . . ± . . ± . . ± . . ± . . Table 4.
Fit results for the projected analysis. The firstand second columns show the lower and upper limits of theextraction rings in arcsec and kpc, while the third columnrepresents the number of source photons coming from eachring, with the percentage indicating their number comparedto the total photons of the same region. In the last twocolumns we report the values of kT, with associated errors,and the χ / DoF.
Projection effects were then taken into account ex-tracting spectra from concentric rings centered on theX-ray peak, containing more than 1500 counts, and fit-ting them with a projct*phabs*apec model. Temper-ature and normalization were left free to vary, while col-umn density, redshift and abundance were frozen at thesame values of the projected analysis above. Results arelisted in Table 5. The deprojected temperature profileof the cluster is shown in black in Fig. 6. This value was assumed after we tried to leave the metallicity freeto vary. However, errorbars were too large to keep it thawed. The exploited abundance table is from Anders & Grevesse (1989) r (kpc)23456 k T ( k e V ) Figure 6.
Projected (blue) and deprojected (yellow) tem-perature profile of A1668. Bars in the x-axis represent therange of the extraction rings, while in the y-axis are the er-rors for the temperature values.
Following the same method described in Pasini et al.(2019) , we estimated the electronic density as : n e = (cid:115) (cid:18) π · N ( r ) · [ D A · (1 + z )] . · V (cid:19) (1)where N(r) is the apec normalization of the depro-jected model, V is the shell volume and D A is the angulardistance of the source, estimated as D A = D L /(1+ z ) .Table 5 lists the density values for each ring, with theresults showed in Fig. 7.Making use of the deprojected temperature and den-sity values, we can derive the cooling time, the pressureand the entropy for each bin. Table 5 presents the pres-sure values, calculated as p = 1 . n e kT , while the en-tropy, that was estimated as S = kT n − / e , is presentedin Fig. 8.The cooling time is defined as: t cool = H Λ( T ) n e n p = γγ − kT ( r ) µXn e ( r )Λ( T ) (2) Note the typo in Eq. 4 of that paper first Chandra view of the cool core cluster A1668 r min - r max r min - r max Counts kT N(r) (10 − ) Electronic Density Pressure Entropy t cool [arcsec] [kpc] [keV] [10 − cm − ] [10 − dy cm − ] [keV cm ] [Gyr]0 - 20 0 - 23.5 2506 (99.4 %) 1.68 ± . . ± . . ± . . ± . . ± . . ± . .
20 - 40 23.5 - 46.9 2162(97.4 %) 2.51 ± . . ± . . ± . . ± . . ± . . ± . .
40 - 65 46.9 - 76.2 1862 (93.7 %) 4.18 ± . . ± . . ± . . ± . . ± . . ± . .
65 - 90 76.2 - 105.5 1653 (89.1 %) 3.36 ± . . ± . . ± . . ± . . ± . . ± . . Table 5.
Fit results for the deprojected analysis. The first two columns report the limits of the annular regions and the numberof source photons from each ring, with the percentage indicating their number compared to the total photons of the same region.The remaining columns report temperature, normalization factor, electronic density, pressure, entropy and cooling time. Thefit gives χ / DoF = 1.46. r (kpc) n e [ c m ] Figure 7.
Density radial profile of A1668 derived from thedeprojected analysis. Each bin defines an extraction region. where γ =5/3 is the adiabatic index, H is the en-thalpy, µ (cid:39) (cid:39) T ) is the cooling function (Sutherland & Dopita1993). Results are listed in Table 5, while the coolingtime radial profile is shown in Fig. 9.We thus estimated the cooling radius of the cluster,i.e. the radius within which the ICM cooling is efficient,assuming t age ∼ age (red line) defines the cooling radiusof A1668, being r cool ≈ ≈
40 kpc.The bolometric X-ray luminosity emitted within thisradius was estimated by extracting a spectrum froman annular region centered on the X-ray peak with r= r cool . Projection effects were taken into account by r (kpc)100 E n t r o p y ( k e V c m ) Figure 8.
Entropy radial profile of A1668 derived from thedeprojected analysis. Each bin defines an extraction region. using a second annular region with internal radius co-incident with r cool and external radius ∼ projct*phabs*apec model, thebolometric luminosity inside the cooling region results L cool = 1 . ± . · erg s − . Assuming a steadystate cooling flow model, the Mass Deposition Rate ofthe cooling flow of A1668 can be estimated as:˙M (cid:39) µm p kT · L cool (3)In this way, we obtain ˙M (cid:39) . ± . (cid:12) yr − .As a different approach, we performed a further fit ofthe spectrum of the cooling region with a phabs*(apec+ mkcflow) model, where the apec component approx-imates the ICM emission along the line of sight out-side of the cooling region, while mkcflow is a multi-phase component reproducing a cooling flow-like emis- Pasini et al. r (kpc)10 C oo li n g t i m e ( G y r ) Figure 9.
Cooling time profile of A1668. Each bin definesan extraction region. The blue line represents the best-fitfunction f(x)=(0.57 ± . ± . , while the red line is t age = 7.7 Gyr. sion inside the cooling radius. As above, the abundancewas fixed at 0.3 Z (cid:12) , while the temperature of the apec model was left free to vary and bounded to the hightemperature parameter of mkcflow . Redshift and ab-sorbing column density were fixed at the Galactic val-ues (see above), while the low temperature parame-ter of mkcflow was fixed at the lowest possible value, ∼ χ /DoF = 105/100 and pro-vides an upper limit of ˙M < (cid:12) yr − . The bolo-metric luminosity associated to the mkcflow model isL mkcflow = 3 . ± . · erg s − . The difference be-tween the two estimates of the mass deposition rate re-flects the Cooling Flow (CF) problem : observed mass de-position rates do not match expectations from the stan-dard CF model, and heating contribution, likely pro-duced by the central AGN, is required to balance theICM radiative losses. DISCUSSION4.1.
Radio-X-ray combined analysis
In order to investigate the interactions between thecooling ICM and the BCG, we overlaid the 1.4 GHzradio contours on the X-ray 0.5-2 keV cluster image.Since we are interested in the core region, in Fig. 10 weshow the resulting image, zoomed in the central 30 × RA =13 h m s , DEC =+19 ° m s ), defined as the center of theisophotes, lies within this region, too. On theother hand, the X-ray peak ( RA =13 h m s , DEC =+19 ° m s ), is found to the south of the nu-cleus of the BCG, exhibiting a significant offset of ∼ ∼ ∼
4” circular re-gion centered on the peak, and fitted it with two models: phabs*apec and phabs*(apec+powerlaw) . The first fitgave χ / DoF ∼ χ / DoF ∼ powerlaw component pro-vided a significant improvement of the fit. We obtainedan F-value of 3.4 and p=0.035, correspondent to a nullhypothesis probability of 1-p=0.965. This suggests thatthe addition of the point-source emission componentis not statistically significant. As a further check, welooked for possible point-sources in high energy, opticaland infrared catalogues, as well as in a harder band (4-7keV) X-ray image; however, we did not detect any pointsource coincident with the X-ray peak. We thus con-clude that the peak detection is likely not biased, andtherefore the offset is real.This is analogous with what was found in A2495, thatpresents a similar-scale offset between these two com-ponents, and with a number of recent works that foundthe same feature in other clusters (e.g., Sanderson et al.2009; Haarsma et al. 2010; Hudson et al. 2010; Rossettiet al. 2016, and others; for a brief review of the state-of-the-art literature about BCG/cool core offsets, seePasini et al. 2019). We will return on this in Sec. 4.3.A two-dimensional temperature map is often used inorder to further investigate on the cluster structure andits thermodynamical state, but the small number of pho-tons prevents us from producing such map. We also es-timated the softness ratio as (S-H)/(S+H), where S andH are the number of counts in the soft (0.5-2 keV) andhard (2-7 keV) band, respectively. However, the statis-tics are still too poor and errors are too large to drawany conclusion from such analysis.4.2. H α analysis The presence of optical line emitting nebulae in galaxyclusters is linked to the thermodynamical conditions ofthe cluster core; observational studies (e.g., Cavagnolo first Chandra view of the cool core cluster A1668 . . . : : . . Right ascension D e c li na t i on Abell 1668 k p c
20 kpc kpc
20 kpc
X-RAY RADIO 1.4 GHzH 𝛼 OPTICAL
Figure 10.
Left Panel : 1.4 GHz radio (green) and H α (black) contours overlaid on the 0.5-2 keV X-ray image, zoomed towardsthe cluster centre. The cyan cross represents the X-ray emission centroid, coincident with the BCG centre; the red and whitecrosses are the X-ray and H α peaks, respectively. Right Panel : From left to right, from top to bottom: 0.5-2 keV, 1.4 GHz, H α and optical (SDSS) images of A1668. All images are centered on the X-ray peak. The cyan cross represents the BCG. et al. 2008; McNamara et al. 2016) have argued thatsuch warm structures are only found if the central ( ∼
10 kpc) entropy falls below 30 keV cm or, alternatively,when t cool /t ff < ff = (cid:112) R /GM is the freefall time. The estimated entropywithin the central bin of our spectral analysis ( r < . ∼ , thus satisfying the cri-terion for the presence of such nebulae in A1668. Hameret al. (2016) presented VIMOS observations of the H α line emission of a sample of 73 BCGs, including A1668,whose image is shown in Fig. 11.The H α structure presents a rather compact shape,extending for ∼
9” ( ∼ H α = 3.85 ± · erg s − , and Hamer et al.(2016) classified it as a quiescent object, showing a sim-ple, centrally-concentrated morphology . To better vi-sualize the interplay of the three components, in Fig. 10we also overlay the H α contours on the X-ray 0.5-2 keV This estimate differs from Pulido et al. (2018) since it is notextinction-corrected and is obtained from IFU observations. The morphology looks slightly different when compared to Ed-wards et al. (2009), whose IFU image shows a somehow better’resolved’ shape. . : : . . . . Right ascension D e c li na t i on A1668
Figure 11.
VIMOS H α image (Hamer et al. 2016). The mean seeingis 1.21”, and the map units are in 10 − erg s − cm − ˚A.The white contours are the optical isophotes from SDSS. Thecyan cross represents the radio galaxy centre, coincident withthe BCG core. Pasini et al. image.The line emission lies entirely within the BCG, butalthough it is also within the cool core as defined fromthe cooling time profile, it only overlaps one end ofthe bright X-ray ridge. This highlights a differencewith A2495, in which the H α structure connects thegalaxy with the X-ray peak, and with other systems(e.g., Bayer-Kim et al. 2002; Hamer et al. 2012, 2016),in which the line emission seems to be mostly associatedto the cooling ICM, rather than to the BCG.A significant offset of ∼ ∼ α ( RA =13 h m s , DEC =+19 ° m s )and the X-ray peaks. The same feature, albeit smaller,was found in A2495 by Pasini et al. (2019); offsetsbetween these two peaks were also detected in A1795(Crawford et al. 2005) and in a number of systems byHamer et al. (2016). We will return on this in Sec. 4.3.As described in more details in Hamer et al. (2012),the ratio of the H α plume extent and the total velocitygradient of the warm gas provides an estimate of theprojected offset timescale. Pasini et al. (2019) showedthat, for A2495, such timescale is comparable to the agedifference between two putative cavity pairs, suggestingthat, for systems where the BCG oscillates back andforth through the cooling region, this measure could bea good indicator for the AGN cycle intermittency. InA1668, the H α structure extends for D’ ∼ ∼ ∼ +500 km s − betweenthe gas at the BCG centre and that at the tail of theH α structure. This indicates a projected timescale ofT’= D’/V ∼
18 Myr. In order to correct for the pro-jection effects we assumed a most likely inclination of ∼ ° , with a range of 30-75 ° (Hamer et al. 2012);the corrected timescale can thus be estimated with T= T’ × cos(i)/sen(i), with i being the inclination. Weobtained in this way an offset timescale of ∼ ∼ ∼ ∼ M H α (cid:39) L H α µm p n H α (cid:15) H α (4) where (cid:15) H α ∼ . · − erg cm s − is the H α line emis-sivity, while n H α is obtained assuming pressure equilib-rium with the local ICM: n H α T H α (cid:39) n ICM T ICM (5)where T H α ∼ K, T ICM is the first value reportedin Tab. 4 ( ∼ α structure is lo-cated within the first spectral bin), while n ICM ∼ . n e ,where n e is the electronic density reported in the firstrow of Tab. 5. In this way, we obtain M H α ∼ (2.4 ± · M (cid:12) .The [SII] λ λ ± K, this gives an electron densityof n e, H α =350 ±
270 cm − which corresponds to a to-tal density of n H α =640 ±
495 cm − . These values arecomparable, within the (large) uncertainties, with thevalue of n H α =100 ± cm − derived from Eq. 5. Itis important to consider the impact of the assumedionised gas temperature ( T H α ) on the two measurementsthough. A lower T H α would result in a higher n e, H α fromEq. 5 but a lower n e, H α from the [SII] λ λ T H α would havethe opposite effect on both measurements. The VI-MOS data from (Hamer et al. 2016) are not sensi-tive enough to provide a reliable estimate of T H α forAbell 1668, but deep observations of other objects havefound upper limits to T H α that are much lower thanexpected (e.g. T H α < T H α = 5000 K, wefind n e = 230 ±
180 cm − and n H α =420 ±
330 cm − fromthe [SII] ratio and n H α =200 ± cm − derived from Eq.5, indicating that the two measurements are consistentwithin the limits of the available data and assumed val-ues. 4.3. Putative cavities and AGN feedback cycle
From the 0.5-2 keV X-ray image (see Fig. 12) it ispossible to identify a number of ICM surface brightnessdepressions. The present observation is very shallow ( ∼
10 ks), thus the reader shall be warned about the signif-icance of these deficits, that could possibly be artifacts.However, we focused our attention on three of these fea-tures (showed in Fig. 12) that, due to their position,could possibly represent real ICM cavities. One of them(A) lies within the radio galaxy West lobe; another sur-face brightness depression (B) is detected in the radiogalaxy East lobe, while a symmetrical (with respect tothe X-ray peak), similar-shaped brightness depression(C), is found at the opposite side of the cluster core, first Chandra view of the cool core cluster A1668
10 kpcC B A . . . : : . Right ascension D e c li na t i on Abell 1668
Figure 12. not associated with the radio galaxy. It is noteworthyto mention that cavity A is also coincident with the H α line emission peak.To investigate on these features, we estimated theirsignificance as N M -N C / √ N M + N C , where N M and N C are the number of counts in regions of equal area close tothe candidate cavity and within the cavity, respectively.The number of counts is different depending on the sizeof the elliptical region chosen to cover the putative cav-ity and, as previously stated, the current observationrequires us to be cautious, since the shape and extent ofthe depression observed ’by eye’ can slightly change us-ing different color scales. For this reason, the upper andlower limits for the dimensions of these regions were es-timated by varying the axes of the ellipse until reachinga significance of 2 σ and 3 σ , respectively. The assumed’true’ size of the bubble is the mean between these twolimits. Cavity A exhibits a circular shape, with a diam-eter of 4.8 ± ± ± L cool (10 ergs s ) P c a v ( e r g s s ) pV4pV16pV Bîrzan et al. 2017Pasini et al. 2019A1668
Figure 13.
Blue points are the data from Bˆırzan et al.(2017), the orange ones are the values for A2495 (Pasini et al.2019), while red represents the cavity system detected inA1668. Dashed lines represent, from left to right, P cav =L cool assuming pV , 4 pV or 16 pV as the deposited energy. Following the method described in Bˆırzan et al. (2004),we determined the cavity power: P cav = E cav t cav = 4 pVt cav (6)where t cav is the age of the cavity, calculated as t cav = R/c s , with R being the cavity distance from theBCG center, p is the pressure at the distance of thecavity, V is the cavity volume and c s is the sound ve-locity. The volume was estimated assuming an oblateelipsoidal shape for both cavities, while for pressure andtemperature we assumed the values listed in Table 5 cor-responding to the annular bin the cavities lie in.We obtained the same age for both cavities: t cav =5.2 ± . cav, A = 5.1 ± . · erg s − for cavity A, and P cav, B = 3.8 ± . · ergs − for cavity B. The age is consistent with the offsettimescale estimated in Sec. 4.2. Finally, we comparedthe estimated values of P cav (in the hypothesis that thecavities are real) and L cool with the typical distributionobserved for cool core clusters (Bˆırzan et al. 2017). Theresult is presented in Fig. 13.The L cool -P cav relationship in A1668 is consistentwithin the scatter of the expected distribution, despitebeing on its lower edge. The detected offsets, therefore,2 Pasini et al. do not seem to affect the feeding-feedback cycle, that isstill maintained. The same result was found in A2495,where the two cavity systems (i.e., the AGN) haveenough energy to balance the radiative losses within thecooling region. We then argue that small offsets are notable to break the AGN feeding-feedback cycle.4.4.
Offsets, cooling and H α emission: are sloshingand AGN activity shaping the core of A1668? As described in the previous sections, the core ofA1668 contains a complex set of structures whose ori-gin is not immediately clear. It is worth reiterating thatthe available
Chandra data is only a snapshot, provid-ing somewhat limited information on the ICM. It shouldalso be remembered that all of the structures we observeare contained within the central ∼
20 kpc, inside thestellar body of the BCG.The offsets between the peaks of the radio, X-ray andH α emission raise the question of how these componentscame to be separated. The peak of emission from thehot ICM lies not in the BCG nucleus, but ∼ α peaks in the region of the west ra-dio jet/lobe, extending around the western half of theradio structure and overlapping the optical centroid ofthe BCG.We suggest a qualitative scenario which might explainthe relative morphologies of the different components.At some point in the past, A1668 may have been a re-laxed cluster with a cool core. In that core, centredon the BCG, gas had begun to cool and condense outof the ICM, forming the kind of filamentary H α neb-ula observed in other cool core cluster. The small ve-locity gradient in the H α emission is consistent withsuch an origin. At that stage, A1668 underwent a mi-nor merger, which caused the core to begin sloshing,oscillating around the centre of the cluster gravitationalpotential. Sloshing motions in the plane of the sky aretypically visible as a spiral pattern in the ICM, but ifthe plane of motion is aligned along the line of sight,the motions produce pairs of nested cold fronts, and thecool ICM gas drawn out from the core can appear as atail to one side of the BCG. Our Chandra observation istoo short for fronts to be visible, but we do see the tail:the ridge structure.At this stage, cooling (and H α emission) would stillhave been centred in the core of the BCG. About 5 Myrago ( ∼ cavity age), sufficient cooled material reached thecentral SMBH to trigger an outburst. This produced theradio jets/lobes we observe, and as these expanded theypushed aside the pre-existing H α filaments and hotter ICM gas. This produced a correlated H α /radio mor-phology, with much of the H α wrapped around the westjet/lobe. It also disrupted the centre of the cool core,reducing the X-ray surface brightness as a large part ofthe volume in the core of the BCG was filled by the radiolobes, producing the apparent cavities in the ICM. Thisbrings us to the current situation, where the brightestX-ray emission is in the tail to the south of the BCGnucleus.The expansion timescale of the radio lobes is onlya few 10 yr. This is very short compared to typicalsloshing timescales. The hot ICM oscillates with slosh-ing timescale given by t slosh = 2 π/ω BV where ω BV =Ω K (cid:16) γ d ln Sd ln r (cid:17) / . Here d ls S / d ln r is the logarithmic en-tropy gradient, Ω K = (cid:112) GM/r and γ =5/3 for the ion-ized ICM plasma. We also know that the free-fall timein the cluster is t ff = (cid:112) r/g . Our Chandra data arenot sufficient to accurately model the mass profile, butwe know that the stellar velocity dispersion, σ ∗ , in theinner regions of the BCG will follow the gravitationalpotential, so that t ff (cid:39) r/σ ∗ (Voit et al. 2015). We cantherefore approximate ω BV as t ff (cid:113) d ln Sd ln r .Our data do not allow the calculation of the entropyprofile on scales r (cid:46)
25 kpc (Fig. 8), thus we use theaverage slope d ln S/d ln r = 0 .
67 given by Hogan et al.(2017b) (see also Panagoulia et al. 2014). This returns t slosh ∼ t ff . We do not know the scale of the sloshing,but it must be greater than the length of the X-ray tail( ∼
16 kpc). Based on the measured σ ∗ =226 ± − (Pulido et al. 2018), at 16 kpc t ff =70 Myr, and thus t slosh may be as much as ∼
490 Myr. As expected this isconsiderably longer than the AGN expansion timescale,confirming that if sloshing is occurring, it cannot yethave affected the structure of the radio lobes.As argued by Olivares et al. (2019), the fact that fil-amentary nebulae in cool core clusters generally lack asignificant velocity gradient indicates that the cool H α or CO-emitting gas they contain is at least partially tiedto the surrounding ICM. It is unclear how the differentphases are connected, but it has been suggested that thedenser material may be enveloped by many diffuse layersof warmer gas (Li et al. 2018), or threaded through bymagnetic fields (McCourt et al. 2015), either of whichcould increase drag forces. We would thus expect theH α emission to trace the regions in which gas has mostrecently cooled from the ICM, after modification by theexpanding radio lobes. If the inflation of cavities hasdisrupted the cooling region, we might expect the locusof any future cooling to be at the new X-ray peak ofsurface brightness, source of the BCG nucleus. How-ever, the lack of H α emission at that location suggests first Chandra view of the cool core cluster A1668 α emitting gas mighthave formed as a result of the AGN outburst, with theexpanding lobes triggering condensation (e.g., Qiu et al.2020). If the cluster is sloshing, we cannot know thescale or alignment of the motion without deeper data.However, our scenario explains several basic facts: theradio and H α emission are correlated because the ra-dio has at least partly determined the morphology ofthe H α -emitting gas. The X-ray offset is the result ofsloshing, which has not affected the radio sources or H α because the radio source expansion timescale is shortcompared to the sloshing timescale. The BCG is nolonger the centre of ICM cooling because the AGN haspushed aside the dense gas which fuelled the outburst.The scenario also makes testable predictions. If thecluster is sloshing, we should expect deeper Chandra data to reveal nested cold fronts, and the X-ray ridgeshould contain relatively cool, high abundance gas.Deeper imaging should also allow us to more accuratelymeasure the morphology of any cavities, which shouldbe correlated with the rado jets and lobes. Higher res-olution radio data may be required to make this com-parison. Lastly, we might expect higher resolution H α imaging to reveal complex structure in the cooled mate-rial, consistent with a filamentary nebula disturbed byan AGN outburst.4.5. Alternative explanations for the origin of the H α emission and spatial offsets An alternative hypothesis that could explain the ob-served displacement of the H α emission is cooling in situ ,perhaps stimulated by the same AGN outburst whichoriginated the X-ray cavities. Inhomogeneous coolingscenarios in clusters have been the object of a long, livelydebate (see early reviews by Fabian et al. 1991, 1994, andreferences therein). Nowadays, many lines of evidencesuggest that hot gas cools at a (mean) low rate and ina spatially distributed fashion, when the ISM/ICM con-ditions are appropriate (see Hogan et al. 2017b; Pulidoet al. 2018; Lakhchaura et al. 2018 for a quantitative dis-cussion). The primary trigger of this localized cooling(that is, the origin of thermally unstable perturbations)might be turbulence (e.g. Chaotic Cold Accretion, CCA,see Gaspari et al. 2012, 2013; Voit et al. 2015), liftingof low entropy gas by X-ray cavities (Revaz et al. 2008;Brighenti et al. 2015) or the sloshing itself. CCA implies the trigger of thermal instabilities, thatis favoured by a central ( ≤
10 kpc) cooling time t cool ∼ t cool /t ff ≤ −
20 (Voit et al. 2015). It is easier to ensueat the position of the X-ray peak, but it can be triggeredwherever these conditions are respected. Therefore, thedisplacement observed for the H α gas could be the resultof CCA detached from the emission peak. CCA at thecurrent position of the peak could still be happening,but it could have not built yet enough material to bedetected in H α . However, given the available X-ray ob-servation, we are not in a position to accurately estimatethe cooling time in the central region of the cluster.We can explore in some more detail the scenario wherethe warm gas derives from a cool component of the ICM,originally located close to the nucleus of the BCG, up-lifted by the cavities and then cooled to 10 K. FollowingArchimedes’ principle, cavities can lift an amount of gasequal to their displacement, though simulations suggestthat the maximum amount is only ∼
50% of this value(Pope et al. 2010). This corresponds to M uplift = 9 · M (cid:12) for cavity A, and M uplift = 3 · M (cid:12) for cavity B.The mass of the H α plume is lower ( M H α ∼ · M (cid:12) , see Sec. 4.2). However, given the state-of-the-artcorrelation between L H α and molecular gas mass M mol (see e.g., Edge 2001; Salom´e & Combes 2003; Pulidoet al. 2018), that is usually found to be co-spatial withH α , we would expect to have M mol ∼ M (cid:12) (lowerthan the upper limit 1 . × M (cid:12) quoted by Salom´e &Combes 2003). The total amount of gas would thereforebe too large for the cavities to uplift and this, along withthe radio/H α morphology, would imply the need for anearlier cycle of AGN jet activity if uplift is responsible.The observed H α line emission could also be the rem-nant of the ISM of a gas-rich galaxy which merged withthe BCG. To test this hypothesis, we examined SDSSand DSS optical images and catalogs in order to checkwhether a member galaxy could be interacting with theBCG. However, the closest system lies more than 40 kpcaway from the BCG, not showing any hint of interplay.Therefore, the merging hypothesis looks unlikely withthe current data.Finally, the warm gas could originate from the stellarmass loss in the BCG (Mathews 1990; Li et al. 2019).With a total B-band luminosity L B ∼ . × L B, (cid:12) (Makarov et al. 2014) and a stellar mass to ratio for anold population ( M/L B ) ∼ M ∗ ∼ .
35 M (cid:12) /yr (Mathews1989). Thus, the observed amount of emission line gascan be accumulated in less than 2 Myr. However, thedisplacement with respect to the BCG center and itsfilamentary and disturbed distribution (Edwards et al.4
Pasini et al. CONCLUSIONSWe performed a multi-wavelength analysis of the coolcore cluster A1668, by means of new radio (EVLA) andX-ray (
Chandra ) observations and of H α line emissiondata from Hamer et al. (2016). The results can be sum-marized as follows: • The radio analysis at 1.4 (L ∼ · WHz − ) and 5 GHz (L ∼ · W Hz − )shows a small ( ∼
11 -14 kpc) and elongated FRIradio galaxy, with no hints of larger scale emissionat these frequencies. The mean spectral index is α = 0 . ± .
06, consistent with the usual valuesfound in BCGs. • The X-ray analysis confirms the classification ofA1668 as a cool core cluster, with a cooling radiusof ∼