The Infall of the Virgo Elliptical Galaxy M60 toward M87 and the Gaseous Structures Produced by Kelvin-Helmholtz Instabilities
R. A. Wood, C. Jones, M. E. Machacek, W. R. Forman, A. Bogdan, F. Andrade-Santos, R. P. Kraft, A. Paggi, E. Roediger
DD RAFT VERSION O CTOBER
14, 2018
Preprint typeset using L A TEX style AASTeX6 v. 1.0
THE INFALL OF THE VIRGO ELLIPTICAL GALAXY M60 TOWARD M87AND THE GASEOUS STRUCTURES PRODUCED BY KELVIN-HELMHOLTZ INSTABILITIES
R. A. W
OOD , C. J ONES , M. E. M ACHACEK , W. R. F ORMAN , A. B OGDAN , F. A NDRADE -S ANTOS , R. P. K RAFT , A. P AGGI , E.R OEDIGER University of Southampton, Southampton, SO17 1BJ United Kingdom Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138 USAand School of Mathematics & Physical Sciences, University of Hull, Hull HU6 7RX, UK
ABSTRACTWe present
Chandra observations of hot gas structures, characteristic of gas stripping during infall, in the Virgocluster elliptical galaxy M60 (NGC4649) located 1 Mpc east of M87. 0 . − Chandra
X-ray images showa sharp leading edge in the surface brightness 12 . ± . . ± .
02 keV for abundance 0 . Z (cid:12) inside the edge and 1 . + . − . keV for abundance 0 . Z (cid:12) in the Virgo ICM free stream region. We find that the observed jump in surface brightness yields a densityratio n in / n out = 6 . + . − . between gas inside the edge and in the cluster free stream region. If the edge is acold front due solely to the infall of M60 in the direction of M87, we find a pressure ratio of 4 . + . − . and Machnumber 1 . + . − . . For 1 .
37 keV Virgo gas, we find a total infall velocity for M60 of v M60 = 1030 ±
180 km s − .We calculate the motion in the plane of the sky to be v tran = 1012 + − km s − implying an inclination angle ξ = 11 + − degrees. Surface brightness profiles also show the presence of a faint, diffuse gaseous tail. We identifyfilamentary gaseous wing structures caused by the galaxy’s motion through the ICM. The structure and dimen-sions of these wings are consistent with simulations of Kelvin-Helmholtz instabilities as expected if the gasstripping is close to inviscid. Keywords: galaxies: clusters: general, Virgo — galaxies: individual(NGC 4649, NGC 4647, M60, M87) —galaxies: intergalactic medium — X-rays: galaxies INTRODUCTIONThe earliest X-ray images taken with
Einstein revealedgalaxy clusters to be far from dynamically relaxed systems(Jones et al. 1979). Extensive evidence now exists for sub-cluster and galaxy infall into clusters. As galaxies movethrough the intracluster medium they undergo hydrodynami-cal and tidal interactions of varying severity. Surface bright-ness edges (contact discontinuities in density and tempera-ture) may be caused by a variety of physical mechanisms.When seen in cluster galaxies moving through the intra-cluster gas, these surface brightness edges may be the resultof ram-pressure stripping as the galaxy moves through theICM. These ‘cold fronts’ are contact discontinuities at theedge of two gas regions of different densities and tempera-tures (e.g. Markevitch & Vikhlinin 2007). [email protected]
From X-ray observations of diffuse galaxy and cluster gas,and the galaxy redshift, the total galaxy velocity and di-rection of motion can be measured, providing one of thefew ways that galaxy velocities in the plane of the sky canbe inferred. The gas densities and temperatures are deter-mined from X-ray observations, from which the componentsof thermal pressure on both sides of the contact discontinuityare derived. Approximating the flow to that of uniform gasabout a blunt body, the total galaxy velocity follows fromthe ratio of thermal pressures at the stagnation point and freestream region (Vikhlinin et al. 2001). These cold fronts fromram pressure stripping are often accompanied by additionalgas features such as wings and stripped gaseous tails (seee.g. Machacek et al. 2005; 2006; Kraft et al. 2017). Themorphology of these features may allow one to constrain themicrophysical properties of the surrounding gas (Roedigeret al. 2015a,b). Cold front edges may also be induced bygas bulk motions (sloshing) caused by ongoing galaxy merg- a r X i v : . [ a s t r o - ph . GA ] D ec W OOD ET AL .ers. In this case multiple surface brightness edges or a spiralpattern are often observed, depending on the orientation ofthe merger with respect to the observer’s line of sight (e.g.Markevitch & Vikhlinen 2007). The measured ratio of ther-mal pressures across a sloshing cold front is 1. Finally, sur-face brightness edges may be due to shocks from episodicnuclear activity, lifting galaxy gas higher in the galaxy grav-itational potential making it easier to be stripped. Several orall of these processes may act in concert on galaxies in galaxyclusters as the galaxies and clusters evolve.The Virgo cluster of galaxies is located 17 . ± . α = 12 h m s . , δ = 11 ◦ (cid:48) (cid:48)(cid:48) ), is a massive ellip-tical galaxy located in projection 195 arcmin (0 .
97 Mpc) eastof M87 ( α = 12 h m s . δ = 12 ◦ (cid:48) . (cid:48)(cid:48) ). M60’s line ofsight velocity ( v rad = 1117 ± − ; Trager et al. 2000)compared to that of M87 ( v rad = 1307 ± − ; Smithet al. 2000) implies that M60 is moving through the VirgoICM with at least ∆ v rad = 190 ±
13 km s − towards us rela-tive to M87. We obtain key measurements of the gas in themassive elliptical galaxy M60 using archival Chandra
X-rayobservations (see Table 2) with a cleaned coadded exposuretime of 262 ks. These deep data sets reveal a wealth of gas-stripping related structures, showing that M60 is undergoingram pressure stripping as it falls through the ICM towards thecluster center. In the
Chandra image (Fig. 1), we see the gasremoved from the core of M60 has formed wings, a west-ern edge and a faint eastern filamentary tail. Understandingthe nature of these edges and filamentary gaseous structuresand a determination of the velocity of M60 through the VirgoICM are the focus of this paper.The discussion in this paper is presented as follows: In §2we list the five
Chandra observations used, explaining thedata reduction and processing methods. §3 documents thecritical steps undertaken to analyze the
Chandra data, high-lighting the observational results from the reprocessed, coad-ded data and discussing the results drawn. Our conclusionsare presented in §4. All coordinates are J2000 and, unlessotherwise indicated, all errors are 90% confidence levels. Us-ing the Surface Brightness Fluctuation (SBF) distances fromMei et al. (2007), the distance to M87, is 17 . ± . (cid:48) = 4 .
98 kpc. These measurements assume a Hubble con-stant H = 73 km s − Mpc − in the Λ CDM cosmology and areconsistent with the SBF measurements by Tonry et al. (2001)(16 . ± . OBSERVATIONS AND DATA REDUCTION
Figure 1 . Exposure-corrected, background-subtracted, co-added
Chandra
X-ray image of M60 in the soft band (0 . − . × (cid:48)(cid:48) × (cid:48)(cid:48) ) is used and the image has beensmoothed with a 2 (cid:48)(cid:48) Gaussian kernel to highlight faint diffuse fea-tures. Both the instrumental readout and cosmic X-ray backgroundshave been subtracted and point sources have been excluded. Northis up and east is to the left. Note the leading edge to the north andwest in the direction of M87 and gaseous wings extending from theedge to either side. The small region of bright emission northwestof the surface brightness edge is the spiral galaxy NGC 4647.
We reprocessed and then co-added five observations ofM60 (Table 2) taken with the
Chandra
X-ray Observatory us-ing the Advanced CCD Imaging Spectrometer array (ACIS)with ACIS-S (chip S3) at the aimpoint, giving a total expo-sure of 270 ks. A sixth, 38 ks observation from the archive(ObsID 785) was badly flared and so was excluded from ouranalysis. All the data were taken in VFAINT mode and an-alyzed using the standard X-ray processing packages, CIAO4.7 (CALDB 4.6.8), FTOOLS 6.15, Sherpa 4.4 and XSPEC12.9.0. The data were filtered with lc_clean and deflare toremove events in periods of abnormally low or high counts,where the count rate deviates more than 20% above or belowthe mean. This gave a useful exposure time of 262 ks.
Table 1 . Chandra observations of M60ObsID Date Exposure Cleaned Exposure(ks) (ks)8182 2007 Jan 30 52 .
37 48 . Table 1 continued HE I NFALL OF
M60 T
OWARD
M87 3
Table 1 (continued)
ObsID Date Exposure Cleaned Exposure(ks) (ks)8507 2007 Feb 1 17 .
52 17 . .
93 80 . .
04 101 . .
97 13 . OTE —The cleaned exposure is after removal of periods ofanomalously high and low count rates.
Some events from M60’s bright core are redistributedalong the ACIS readout direction during readout. Theseout-of-time events may contaminate both imaging and spec-tral analyses of the faint diffuse X-ray emission of interest.We model the readout contribution to the background usingCIAO tool readout_bkg , based on the algorithm developedby Vikhlinin et al. (2005). The cosmic X-ray contributionto the background was taken from the blank-sky background(BSB) files; a series of source-free background sets obtainedat high galactic latitude to avoid contamination from ourGalaxy. For each observation the relevant blank-sky back-ground for that date was downloaded and reprocessed. Thesewere the period D source-free datasets for the S3 CCD withexposure 400 ks. Identical energy filters and the same clean-ing process using CIAO were applied to the blank-sky back-ground files. The normalization of the background was setby the ratio of exposure times between the M60 and blank-sky background datasets, with a last adjustment to accountfor the time variability of the particle background componentmade by matching the detected count rate for background andsource in the 10 −
12 keV energy band, where particle back-grounds dominate. Both the readout and cosmic X-ray back-ground contributions were subtracted from our subsequentimage analysis. Exposure-corrected, background-subtractedflux images were created for each observation in the soft(0 . − . . − . . (cid:48)(cid:48) × . (cid:48)(cid:48) Chandra observations used in our analysis withthe identified point sources excluded is shown in Fig. 1. Thedynamical gas features studied within this paper are marked.The reader may view a less processed coadded mosaic of thefive data sets that includes point sources in Fig. 5. RESULTSM60 shows several features evident of gas dynamics.Fig 1, in the 0 . − . ∼
12 kpc north and west from the galaxycenter in the direction of M87, a north wing arcing to the northeast and a less prominent, possibly bifurcated, wing ex-tending to the south. We will also show evidence for a faintdiffuse tail downstream. These features are similar to thosein M89 caused by ram-pressure stripping as M89 interactswith the Virgo ICM during infall (Machacek et al. 2006,Kraft et al. 2017). We suggest that M60 is similarly under-going gaseous stripping as it moves through the Virgo ICM,forming the upstream edge, wings to the northeast and south,and a faint diffuse eastern tail. A leading edge, faint tailand wings caused by Kelvin-Helmholtz Instabilities (KHI)are expected for ram-pressure stripping through an inviscidmedium (Roediger et al. 2015a). We study the gas propertiesin these features in M60 and compare with the simulations ofM89 (Roediger et al. 2015a,b), a similar Virgo galaxy, to testthis scenario.
Figure 2 . Sectors for the profile analysis from Table 2 overlaid onthe background subtracted, exposure corrected 0 . − . . (cid:48)(cid:48) × . (cid:48)(cid:48) , , , Surface Brightness Profiles
To study the leading edge, we construct circular surfacebrightness profiles from the background subtracted, exposurecorrected, coadded 0 . − α = 12 h m s . , δ = 11 ◦ (cid:48) . (cid:48)(cid:48) ) and confined tosectors to the north and west that avoid the wings and ex-clude the interacting spiral galaxy NGC 4647. We similarly W OOD ET AL .construct the circular surface brightness profile in the east-ern (downstream) direction to search for a diffuse, faint tail.These sectors are defined in Table 2 and are shown in Fig. 2.
Table 2 . Profile SectorsProfile Label A B(degree) (degree)N 63 88 .
3W 330 27 .
6E 152 214N
OTE —The sector is centered at the X-ray peak and is defined as subtending theangle measured counterclockwise fromangle A to angle B. All angles are mea-sured counterclockwise from West.
Our results are shown in Fig. 3. Note that the profilesin the N and W sectors are the same in the radial range ofoverlap. This suggests that they are part of the same lead-ing edge caused by ram-pressure, only interrupted in the ex-cluded region by the impending merger of NGC 4647. Dueto our Chandra coverage we can study the morphology of thesurface brightness profile to larger radii to the north (profileN) than to the west (profile W). We thus confine our analy-sis to the profile of the N sector of the leading edge, shownin Fig 3, to measure gas properties to larger radii and probefarther from the galaxy into the surrounding Virgo ICM.We see three distinct regions, the characteristic edge profilefor galaxy gas for r < (cid:48)(cid:48) (12 kpc), a very slowly varying,nearly flat surface brightness profile for r > (cid:48)(cid:48) (22 kpc),and a steeply falling region of excess emission in between.At the 0 .
97 Mpc distance of M60 from M87, taken to be thecenter of the Virgo cluster, emission from undisturbed Virgocluster gas would appear to be flat over the ∼
50 kpc scalemeasured by the profile, consistent with the profile behaviorat r >
22 kpc. We take this to be representative of the ’freestream region’ from Vikhlinin et al. ( 2001). We suggest thesharply falling surface brightness profile at 12 < r <
22 kpcmay be Virgo gas gravitationally attracted to M60’s deepgravitational potential at close radii, i.e. the pile-up region(Vikhlinin et al. 2001). These regions are also shown in Fig.2. Vikhlinin et al. (2001) showed the edge of a cool densegas cloud moving through surrounding hot gas can be seenin radial X-ray surface brightness profiles as a discontinuity(‘edge’) and that the relative velocity of the gas cloud and itssurrounding ICM can be determined by the ratio of thermalgas pressures between gas at the stagnation point and clus-ter gas in the free stream region. If the dense M60 gas wereat rest relative to the ICM, the thermal pressures would be equal on both sides of the edge. However, for M60 infallingthrough the Virgo ICM, where it is subjected to both ramand thermal pressures, this equality is only true at the stagna-tion point where the relative gas velocity is zero. FollowingVikhlinin et al. (2001), we use the gas pressure just insidethe edge as a proxy for the pressure at the stagnation point.We then use the pressure ratio between gas at the stagnationpoint and in the free stream region to derive the relative ve-locity between M60 and the Virgo ICM, as detailed in §3.4.To calculate pressures, characterize the nature of the edge,and, in the case of ram pressure stripping, calculate the ve-locity of the galaxy, the relative temperatures and electrondensities must be obtained for the gas on both sides of theedge. 3.2.
Gas Densities
As in Machacek et al. (2006), we assume spherically sym-metric electron power law density models inside and outsidethe edge ( r = r e ) n i ( r < r e ) = J d n (cid:16) rr e (cid:17) − α i n o ( r > r e ) = n (cid:16) rr e (cid:17) − α o (1)with normalization n , inner and outer power law indices α i and α o , respectively, and a discontinuous density ’jump’ J d = n i ( r e ) / n o ( r e ) across the edge. We then determine thedensity by integrating the X-ray emissivity along the line ofsight to fit the surface brightness profile, using a multi-variate χ -minimization scheme with the position of the edge ( r e ),the density power law indices α i and α o , and the square rootof the surface brightness discontinuity √ J sb ∝ J d across theedge allowed to vary. The measured surface brightness dis-continuity J sb is related to the density jump J d at the edgethrough the ratio of X-ray emissivities J sb = Λ i Λ o ( J d ) (2)where Λ i ( Λ o ) are the cooling functions for gas inside (out-side) the edge, respectively.At a projected distance of ∼ r >
22 kpc to the observed edge and fit the resulting profileto determine the gas density ratio between galaxy gas insidethe edge and undisturbed Virgo gas (the free stream region)outside the edge. Our results are shown in Fig. 4 and listedin Table 3. We find an edge position at 12 . ± . √ J s = 10 . + . − . .For massive galaxy clusters where ICM gas temperaturesare high, the cooling functions in Eq. 2 are largely indepen-dent of gas abundances and only a weak function of the tem-perature such that Λ i / Λ o ∼ √ ( J sb ) ∼ ( J d ). The densityjump can be inferred from the surface brightness profile fits HE I NFALL OF
M60 T
OWARD
M87 5
Figure 3 . Surface brightness profiles from the regions shown in Fig. 2. The N and W profiles (denoted by black circles and red squares,respectively) are the same within uncertainties in their radial regions of overlap, beyond which ( r >
22 kpc) the N profile flattens, consistentwith that expected for Virgo emission. The E profile (blue diamonds) shows excess emission beyond r ∼
11 kpc, consistent with a diffuse tail.
Figure 4 . Fit to the radial surface brightness profile across the north edge. The vertical dashed line indicates the best-fit edge location. SeeTable 3. W OOD ET AL .alone. However, for gas at lower temperatures ( ≤ − Table 3 . X-ray Surface Brightness Profile Model r e α i α o √ J sb (kpc)12 . ± . − . + . − . − . + . − . . + . − . N OTE — The density model extrapolates the slowlyvarying Virgo ICM over the pile-up region to theedge. Columns are edge location r e , power law in-dices α i ( α o ) inside and outside th edge, and squareroot of the surface brightness jump. The densitymodel extrapolates the slowly varying Virgo ICMover the pile-up region to the edge. Spectral Modeling: Gas Temperatures and Abundances
To determine the mean spectral properties of the gas inM60, we use CIAO tool specextract to extract a mean spec-trum in a 200 (cid:48)(cid:48) circular region centered on the observed X-raypeak. We also extract spectra for the regions N1, N2, and N3along the northern profile shown in Fig. 2 to characterize theVirgo emission (N3) at the position of M60 and to completethe cold front analysis across the leading surface brightnessedge (see Fig. 4). Point sources above a detection thresh-old of ∼ × erg s − (see Appendix), as well as emissionfrom the interacting companion galaxy NGC 4647, are ex-cluded from the data. The resulting spectra are modeled us-ing XSpec 12.9.0 (See Arnaud 1996).3.3.1. Mean X-ray Spectral Properties of M60
We first consider the average spectral properties of M60as a whole using the blank sky background files as in §3.1for backgrounds. This will allow us to determine the meanabundance as well as temperature of gas within M60. ACISread-out effects redistribute 1 .
3% of the source counts alongthe read-out direction. Since most of these photons are fromthe bright core, which is included in our mean spectrum, andinspection of the read-out map shows that only 0 .
3% of thetotal source counts are redistributed outside the 200 (cid:48)(cid:48) circularspectral region, read-out will not significantly affect the meanspectral fit. We use an absorbed VAPEC model, that assumesX-ray emission from collisionally ionized diffuse gas with emission rates calculated from the most current atomic tran-sition data tabulated in AtomDB. We fix the hydrogen col-umn density at the Galactic value ( N H = 2 . × cm − ). Formore information on the VAPEC model, please see XSpec:An X-ray Spectral Fitting Package User’s Guide by Arnaud,Gordon, and Dorman. The temperature and abundances for Fe, Mg, Si, and Owere allowed to vary. All other abundances were fixed atsolar. We find a mean temperature kT = 0 . ± .
004 keVwith Fe, O, Mg and Si Anders & Grevesse (1989) abun-dances of 0 . ± . Z (cid:12) , 0 . ± . Z (cid:12) , 0 . + . − . Z (cid:12) , and0 . ± . Z (cid:12) for χ / (dof) = 1989 / . Chandra data measure the X-ray emis-sion from all sources in M60 and along the line of sight. Thisincludes X-ray emission from any unresolved point sources,such as cataclysmic variables (CV’s), accreting white dwarfs(AB’s) and low mass X-ray binaries (LMXBs), and VirgoICM emission along the line of sight, as well as the dif-fuse galaxy gas. We model the relative contribution of eachof these unresolved stellar X-ray sources in the Appendixto determine whether these sources significantly affect ourspectral measurements of flux, temperature and metallicityof M60’s diffuse gas. We find that CV’s and AB’s contributeonly 1 .
5% and LMXB’s 2% of M60’s total 0 . − . − Table 4 . M60 Intrinsic X-ray fluxes and luminositiesBand Range Flux Luminosity(keV) (10 − erg s − cm − ) (10 erg s − )Soft 0 . − . .
27 11 . . − . .
27 0 . . − . .
54 12 . OTE — Flux and luminosities from the best absorbed VApecmodel fit to the mean spectrum of M60 using a 200 (cid:48)(cid:48) radiuscircular region centered on M60’s X-ray peak. http://atomdb.org https//heasarc.nasa.gov/docs/xanadu/spec/manual/manual.html HE I NFALL OF
M60 T
OWARD
M87 73.3.2.
Virgo ICM
The final component to model in fitting the spectra is thecontribution from the cluster emission. Virgo is a young, dy-namically active cluster. Urban et al. (2011) found from
XMM-Newton observations that beyond 500 kpc from M87,the X-ray surface brightness of the Virgo ICM is highly vari-able, varying as much as a factor 2 at 0 . . Z (cid:12) abundance (Urban et al. 2011).We find best-fit temperature, model normalization and in-trinsic 0 . − . + . − . keV, 1 . × − cm − , and 8 . × − erg s − cm − (see Table 5). X-ray emissionin the cluster outskirts is faint, contributing only 0 .
3% ofthe 0 . − (cid:48)(cid:48) circular region emcompass-ing M60. From the VAPEC model normalization we find amean electron density for this region of 3 × − cm − with ∼
30% uncertainties, yielding a thermal gas pressure of ∼ . × − ergs cm − . These results are consistent with Ur-ban et al. (2011), given the measured variability observedat these large radii. Since we directly measure the thermalproperties of the Virgo cluster gas, i.e. gas density, temper-ature and thermal gas pressure, from the spectral fitting ofthe X-ray emission of the gas, any evolutionary effects of theambient cluster magnetic field on determining those thermalproperties have implicitly been accounted for in that mea-surement. Ambient magnetic fields could also contribute anadditional non-thermal component to the total pressure in theVirgo Cluster. However, using the simulated magnetic fieldprofile for the Virgo Cluster from Pfrommer & Dursi (2010),the expected non-thermal pressure due to ambient magneticfields is at most ∼
1% of the thermal pressure at these radiiand likely could be much less. Thus it does not significantlyaffect our analysis.
Table 5 . Northern Edge Spectral ModelsRegion kT norm flux Λ χ / (dof)(keV) (10 − cm − ) (10 − erg s − cm − ) 10 − ergs cm s − )N1 1 . ± .
02 2 .
34 3 .
19 1 .
36 112 / . + . − . .
82 0 .
93 0 .
51 186 / . + . − . .
75 0 .
89 0 .
51 293 / OTE —Spectra were modeled using an absorbed Apec model with Galactic absorption and abun-dance fixed at 0 . Z (cid:12) for M60 galaxy gas (N1) and at 0 . Z (cid:12) for regions N2 and N3 consistentwith Virgo Cluster gas Spectral Fitting Across the Edge
To determine the infall velocity of M60 into the Virgogalaxy cluster, we need to measure the temperatures and den-sities of gas on either side of the edge. We chose the northernprofile sector for our analysis because detector coverage al-lows us to measure gas properties out to greater distancesconsistent with undisturbed Virgo gas (N3). Additionally thenorthern sector lies perpendicular to the readout direction,and the inner radius of region N1, the region just inside theedge, lies outside the bright core region, where readout wouldbe greatest, minimizing the contribution of these out-of-timeevents in these regions. Thus readout in the spectral analysisof the northern regions (N1,N2,N3) may be neglected. Wealso note that a comparsion of X-ray and kinematical massmeasurements for M60, show that magnetic fields do not con- tribute significantly to the pressure inside the edge in regionN1 (6 . < r < . . Z (cid:12) for region N1 in M60, consistent with ourmeasured mean Fe abundance for M60 galaxy gas, and at theVirgo value (0 . Z (cid:12) ) for gas in the pile-up region N2. Ourresults are given in Table 5.Since both galaxy and Virgo gas have temperatures ∼ Λ ( A , T ) and thus emissivity Λ n e n p for each depends sensitively on the metal abundance A of the gas. We determine the cooling functions for each of W OOD ET AL .the northern spectral regions (N1, N2, N3) using Λ = 10 − FD L N APEC [ D A (1 + z )] (3)where F is the unabsorbed model flux in the 0 . − N APEC is the APEC model normalization, z is the redshift, D L ( D A ) are the luminosity (angular size) distances, respec-tively, and D L ∼ [ D A (1 + z )] for z <<
1. Values for N APEC , F , and Λ are also given in Table 5.3.4. Constraining M60’s dynamical motion
Using the fitted surface brightness jumps √ ( J sb ) from Ta-ble 3 and the cooling functions given in Table 5 in Equation2, we calculate the density ratio between galaxy gas insidethe edge and free streaming Virgo gas and multiply these bythe respective temperature ratios from the spectral fits to ob-tain the pressure ratio between the stagnation point and theVirgo free stream region. Uncertainties in derived ratios areestimated using the extreme values of the 90% CL uncertain-ties for measured properties. Our results are listed in Table6. Table 6 . M60 Velocity Analysis T i / T o Λ i / Λ o n i / n o p i / p o M a v v t ξ ( km s − ) ( km s − ) (deg)0 . + . − . .
66 6 . + . − . . + . − . . ± . ±
180 1012 + − ± OTE —Velocities assume a sound speed of 604 km s − for 1 .
37 keV Virgo gas and M60radial velocity v r = − ±
15 km s − . Uncertainites for derived values assume extremes inthe 90% CL uncertainties for measured properties. The Mach number M a = v / c s (where c s is the speed ofsound in the cluster free stream region) for the cold gas cloudmoving through the hot ICM is determined from the ratio ofpressures (see Eqs. 4 and Eq. 5 ). p i p o = (cid:18) + γ − M a (cid:19) γγ − , M a ≤ p i p o = (cid:18) γ + (cid:19) γ + γ − M a (cid:18) γ − γ − M a (cid:19) − γ − , M a ≥ γ = 5 / . ± . .
37 keV Virgogas in the free stream region is c s = 604 km s − , such that wefind the speed v = M a c s of M60 relative to the Virgo ICM is1030 ±
180 km s − .The physical separation of M60 and M87 is 971 kpc,confirmed with the distance moduli measurements in bothTonry et al. (2001) and Mei et al. (2007). The radial ve-locity difference between M60 and M87 is ∆ v rad = − ±
15 km s − (Trager et al. 2000). Taking the Virgo cluster ICMto be at rest relative to M87, we determine the relative trans-verse component of M60’s velocity as v t = 1012 + − km s − .The components of M60’s motion through the Virgo ICM im-ply an inclination angle with respect to the plane of the skyof ξ = 11 ± t pp for peri-center passage assuming a constant infall velocity and directprojected path towards M87, such that t pp ∼ . M pc / v t =0 .
95 Gyr. If the motion of M60 in the plane of the sky is di-rectly towards M87, this will be an upper bound on the infalltime.MHD simulations suggest that magnetic fields may wraparound a galaxy during infall, forming a thin draping layer.In this layer ambient cluster magnetic fields may become am-plified to a maximum value such that their magnetic pressureequals the ram pressure of the ICM on the galaxy (Lyutikov2006; Pfrommer & Dursi 2010). Recent simulations alsosuggest these fields may become tangled and not suppressram pressure stripping (Ruszkowski et al. 2014). For M60infalling at 1030 km s − through Virgo gas of electron density3 × − cm − , this would imply a possible maximum mag-netic field of ∼ . µ G within a 1 (cid:48)(cid:48) . Chandra exposure afactor 2 . Wings and Tail Structures of M60
The motion of M60 through the Virgo ICM causes the gason the northwest side to be stripped and pushed behind thegalaxy in the wake of its passage through Virgo, forming HE I NFALL OF
M60 T
OWARD
M87 9
Figure 5 . (upper) Coadded
Chandra image of M60 with the dou-ble wing structures highlighted by blocking to a bin size of 8 × ∼ (cid:48)(cid:48) × (cid:48)(cid:48) ). (lower) Simulation of gas stripping and Kelvin-Helmholtz instabilities in M89, which form "wings" similar to thosein the Chandra image of M60. The frame is a low energy band(0 . − . the observed ‘wings’ and sharp northwest edge. The coad-ded images show there are a pair of wings in both the north-east and south directions. These wings align with the direc-tion of the outburst of the AGN at the core of M60 (Paggiet al. 2014).The double wing structure is shown more clearly by bin-ning to 8 × (cid:48)(cid:48) × (cid:48)(cid:48) bins) in Fig 5. The filamentarywings are thin and long, extending at least 150 (cid:48)(cid:48) in both direc-tions and as wide as the galaxy atmosphere radius (150 (cid:48)(cid:48) spanacross each double wing structure). The northeastern wingsare of approximately equal length (150 (cid:48)(cid:48) ) and separated by a cavity 28 (cid:48)(cid:48) wide. The wings to the south are less symmetric,with the front wing being shorter than the downstream wingby a factor of 2.The dimensions of these filamentary wings were comparedwith simulations by Roediger et al. (2015a,b) of the Virgogalaxy M89 (NGC4552). The lower panel of Fig 5 showsa soft band X-ray projection of M89 that demonstrates howturbulent stripping can create wings of diffuse gas to the sidesof the galaxy, as we observe for M60. The size of the wingsscales with the size of the gas atmosphere, hence larger wingsare seen in M60, compared to the simulations of the Virgoelliptical M89, since most of M89’s gas has already beenstripped.The double wing structure is seen in the simulation image(Fig 5 lower panel). The wings are attached directly to thesides of the galaxy as is the case for M60, another indicatorof motion nearly in the plane of the sky. This can also beexplained by the inclination angle; the simulations assumemotion purely in the plane of the sky, while M60 has an in-clination of 11 ± r >
10 kpc in theeastern surface brightness profile (Fig. 3) over that observedin the northern or western profiles, suggestive of such a tail.The simulations of M89 predict a downstream edge mark-ing the boundary between the galaxy atmosphere and diffusetail. We performed our edge analysis as outlined in §3.2, onthe eastern profile of M60 to look for this downstream edge(See Figs. 2 and 3 and Table 2).Using the power law model from Eq. 1, the eastern profileis well fit with upstream and downstream power law slopesof α i = 1 . + . − . and α o = 1 . + . − . , respectively. We find asmall jump √ ( J sb ) = 1 . + . − . at r = 12 . + . − . kpc downstreamfrom the center of M60. Since we expect from simulationsthat the near tail is composed of galaxy gas displaced but notyet completely stripped from M60 (Roediger etal 2015a), the0 W OOD ET AL . Figure 6 . Surface brightness radial profile for the northern upstream(black dots) and eastern downstream (blue diamonds) regions ofM60. Note the excess emission to the east at r >
10 kpc suggestinga faint, diffuse gas tail. Vertical lines denote the edge locations forthe upstream northern profile (black dashes) and the eastern down-stream tail (blue dot-dashed). abundance across the eastern edges should be the same. Thus √ ( J sb ) = J d , the ratio of the gas density across the easterndownstream edge. Fig 6 shows the extracted surface bright-ness profiles for both the upstream (northern cold front) anddownstream (eastern tail) edges. The simulations generallyfind the downstream edge at a larger radius than the upstreamedge, even more so than is observed for M60.The surface brightness profiles are symmetric in both ra-dial directions until each respective edge. In the simulationsof M89, the upstream edge of the cold front and wings looksharper than the downstream edge, consistent with what wesee in Fig 6.There are several factors that could keep the tail of M60faint. A steep initial gas density profile for M60 would meanthere was less gas to strip at larger radii and the stripped gasmay mix quickly with the surrounding ICM. The large dis-tance of M60 from M87 (971 kpc) and the galaxy velocity of v M60 = 1030 ±
180 km s − suggest multiple possibilities. First,we could be witnessing the very early stages of the M60-Virgo interaction, thus sufficient gas has not yet been strippedto form a clear tail as observed in M89, which is located farcloser to M87 (390 kpc) and therefore has undergone a muchlonger gas stripping period. Alternatively, M60 may alreadyhave passed through the Virgo system once and we are ob-serving it shortly upon completing a turn around in its or-bit. This would suggest stripping is occurring primarily fromKHI, such that no large volumes of gas are being pushed be-hind the galaxy. In either case, stripping may be less efficientin the low density Virgo outskirts, requiring gas to first beuplifted by periodic AGN activity before being pushed backinto the tail. CONCLUSIONS Using archival data from the
Chandra
X-ray Observatory(total cleaned exposure time 262 ks), we identified a surfacebrightness discontinuity (edge) r e = 12 kpc from the centerof M60 to the north and west in the direction on M87, theVirgo Cluster center. The surface brightness edge is producedby the ram pressure stripping of the gas in M60 as it passesthrough the Virgo ICM. From the surface brightness profile,taken to the north to minimize instrumental effects and max-imize radial distance coverage, we measured gas tempera-tures within the edge in M60, in the Virgo gas pile-up region(12 . < r <
22 kpc) immediately outside the edge, and inthe Virgo free stream region ( r >
22 kpc), and measured thedensity ratio between gas inside the edge and in the Virgofree stream region to determine the physical motion of theelliptical galaxy M60 through the Virgo ICM.We find: • X-ray emission in M60 is gas dominated and well fitin the mean by a VAPEC spectral model with kT =0 . ± .
004 keV and Fe, O, Mg, and Si abundance of0 . ± . Z (cid:12) , 0 . ± . Z (cid:12) , 0 . + . − . Z (cid:12) , and 0 . ± . Z (cid:12) , respectively. • Fixing the abundance inside the edge at 0 . Z (cid:12) con-sistent with M60’s mean results and at 0 . Z (cid:12) forVirgo gas at 0 .
971 Mpc from the literature, gas tem-peratures along the northern profile for galaxy gas in-side the edge, in the pile-up region, and in the Virgofree stream region are 1 . ± .
02 keV, 1 . + . − . keV, and1 . + . − . keV, respectively. • The leading (northern) edge at r e = 12 . ± . n i / n o = 6 . + . − . between gas inside the edge and Virgogas in the free stream region. • The measured pressure ratio p i / p o = 4 . + . − . betweengalaxy gas inside the edge and in the Virgo free streamregion implies M60 is infalling with total velocity v M60 = 1030 ±
180 km s − (Mach 1 . ± .
3) relative tothe Virgo ICM. Given the relative radial velocity of ∆ v rad = − ±
15, this yields an inclination angle ξ = 11 ± v t = 1012 + − km s − , places an upper bound on thetime to pericenter passage of ∼ .
95 Gyr. • Extended wing-like features are observed to thenortheast and south of M60. Comparison of thesefeatures with simulations of ram-pressure strippingsuggests that these wings are likely produced byKelvin-Helmholtz instabilities caused by M60’s mo-tion through a nearly inviscid Virgo ICM. The thin andlong filamentary gas wings scale with the size of thegas atmosphere and are attached directly to the sides HE I NFALL OF
M60 T
OWARD
M87 11of M60, an indicator of motion in the plane of the sky,consistent with the small measured inclination angle. • Excess emission observed to the east (downstream) ofM60’s motion and confirmed in the eastern surfacebrightness profile is consistent with the existence ofa faint, diffuse tail, similar to those seen in simula-tions of ram pressure stripping. The faintness of thetail is due either to insufficient stripping of the gas dueto M60’s distance from the cluster core, or becausewe are observing M60 shortly after completing a turn-around of its orbit.ACKNOWLEDGMENTS Data reduction and analysis was supported by the CXCCIAO, Sherpa, XMM-ESAS software packages and CALDBv4.4.8. Archival data was extracted from the
Chandra
We-bchaser and XMM data archives. The NASA/IPAC Extra-galactic Database (NED), which is operated by JPL/Caltech,under contract with NASA was used throughout as werethe ADS facilities and arXiv for the literature. We thankGary Mamon, and Paul Nulsen for useful discussions of theVirgo cluster and gas stripping, respectively. This work wassupported by
Chandra grants GO1-13141X, GO1-12110X,GO0-1106X, NASA contract NAS8-03060, the Universityof Southampton and the Smithsonian Astrophysical Obser-vatory.
REFERENCES
Anders, E. & Grevesse, N. 1989, Geochimina et Cosmochimica Acta, 53,197Arnaud, K. A. 1996, Astronomical Data Analysis Software and Systems V,eds. Jacoby G. and Barnes, J., ASP Conference Series Vol 101, pg. 17Bell, E. F., McIntosh, D. H., Katz, N., & Weinberg, M. D. 2003, ApJS, 149,289Bogdan, A. & Gilfanov, M. 2011, MNRAS, 418, 1901Böhringer, H., Briel, U. G., Schwarz, R. A., et al. 1994, Nature, 368, 828de Vaucouleurs, G., de Vaucouleurs, A., Corbin, H. G.. et al. 1991, ThirdReference Catalogue of Bright Galaxies (RC3), version 3.9b, Springer-Verlag New YorkForman, W., Schwarz, J., Jones, C., Liller, W., & Fabian, A. C. 1979, ApJ,234, L27Gilfanov, M. 2004, MNRAS, 349, 146Jones, C., Mandel, E., Schwarz, J., et al. 1979, ApJ, 234, L21Kraft, R. P., Forman, W. R., Jones, C., et al. 2011, ApJ, 727, 41Kraft, R.P., Roediger, E., Machacek, M., et al. 2017, ApJ, 848, 27Lyutikov, M. 2006, MNRAS, 373, 73Machacek, M., Dosaj, A. Forman W., et al. 2005, ApJ, 621, 663Machacek, M., Jones, C., Forman, W. R., & Nulsen, P. 2006, ApJ, 644, 155Markevitch, M., & Vikhlinin, A. 2007, Phys. Rep., 443, 1 Mei, S., Blakeslee, J. P., Cote, P., et al. 2007, ApJ, 655, 144Paggi, A., Fabbiano, G., Kim, D., Pellegrini, et al. 2014, ApJ, 787, 134Pfrommer, C. & Dursi, L. J. 2010, Nature Physics, 6, 520Randall, S., Nulsen, P., Forman, W. R., et al. 2008, ApJ, 688, 208Revnivtsev, M., Churazov, E., Sazonov, S., Forman, W., & Jones, C. 2007,A&A, 473, 783Revnivtsev, M., Churazov, E., Sazonov, S., Forman, W., & Jones, C. 2008,A&A, 490, 37Roediger, E., Kraft, R. P., Machacek, M. E., et al. 2012, ApJ, 754, 147Roediger, E., Kraft, R. P., Machacek, M. E., et al. 2015a, ApJ, 806, 103Roediger, E., Kraft, R. P., Machacek, M. E., et al. 2015b, ApJ, 806, 104Ruszkowski, M. Brüggen, M., Lee, D. & Shin, M.-S. 2014, ApJ, 784, 75Smith, R. J., Lucey, J. R., Hudson, M. J., Schlegel, D. J., & Davies, R. L.2000, MNRAS, 313, 469Su, Y., Kraft, R. P., Roediger, E., et al. 2017, ApJ, 834, 74Tonry, J. L., Dressler, A., Blakeslee, J. P., et al. 2001, ApJ, 546, 681Trager, S. C., Faber, S. M., Worthey, G., & Gonzalez, J. J. 2000, AJ, 119,1645Urban, O., Werner, N., Simionescu, A., Allen, S. W., & Böhringer, H. 2011,MNRAS, 414, 2101Vikhlinin, A., Markevitch, M., & Murray, S. S. 2001, ApJ, 551, 160Vikhlinin, A. Markevitch, M., Murray, S.S., et al. 2005, ApJ, 628, 655
APPENDIXThe
Chandra data measure the X-ray emission from all sources in M60; this will encompass the diffuse galaxy gas, anyunresolved point sources, such as cataclysmic variables (CV’s), accreting white dwarfs (AB’s) and low mass X-ray binaries(LMXBs), as well as the Virgo ICM emission along the line of sight. We model the relative contribution of each of theseunresolved stellar X-ray sources to the X-ray luminosity of M60 below:
CATACLYSMIC VARIABLES AND ACCRETING BINARIES
Based on X-ray observations of M32, Revnivtsev et al. (2007) found that unresolved stellar objects may provide a largefraction of the diffuse X-ray emission in low mass galaxies. In their follow-up paper, Revnivtsev et al. (2008) show the oldstellar populations in galaxies can be characterized by a universal value of X-ray emissivity per unit stellar mass or per unitK-band luminosity.From studying the K-band images of M60 taken with the 2MASS survey, we determined the stellar mass and implied X-ray luminosity of the stars. The K-band luminosity measured in a 200 (cid:48)(cid:48) radius region centered on the X-ray peak is L K =2 . × L K , (cid:12) . Bell et al. (2003) showed that the relation between the galaxy’s stellar mass and the K-band luminosity can beexpressed as: log (cid:18) M ∗ L K (cid:19) = a K + b K × ( B − V ) (6)where M ∗ is in units of M (cid:12) and L K is in units of L K , (cid:12) . Using a K = − . b K = 0 .
135 (Bell et al. 2003) and the extinctioncorrected color for M60 B − V = 0 .
95 from RC3 data (de Vaucouleurs et al. 1991; NED) in Eq. 6, we find a stellar mass of M ∗ = 2 . × M (cid:12) , where systematic uncertainties in the M/L relationship may be as high as 25%.2 W OOD ET AL .Revnivtsev et al. (2007) give the relation between X-ray luminosity and stellar mass in the soft (0 . − . L . − . = 7 × (cid:18) M ∗ M (cid:12) (cid:19) erg s − (7)Using Eq. 7, the stellar contribution to the soft X-rays is therefore L . − . = 1 . × erg s − , such that the stellar compo-nent is 1 .
5% of the total 0 . − X - L K relation for these components in early type galaxies (see, e.g. Bogdan and Gilfanov 2011).In fitting the X-ray spectra, this component is modeled with a mekal model plus power law with fixed temperature kT = 0 . A = 1 . Z (cid:12) and power law exponent Γ = 1 . .
03 in the 0 . − .
5% of the 0 . − LOW MASS X-RAY BINARIES
A second component to the X-ray emission is required to account for unresolved LMXBs below the individual source detectionthreshold that contribute to the overall diffuse X-ray luminosity. The azimuthally averaged spatial distribution of the number ofLMXBs for most normal galaxies follows closely the distribution of the near-infrared light (Gilfanov 2004). The combinedluminosity functions of LMXBs for such galaxies are as follows:d N d L = K ( L L b , ) − α , L < L b , K ( L L b , ) − α , L b , < L < L b , K ( L L cut ) − α , L b , < L < L cut , L > L cut (8) K = K ( L b , L b , ) α , K = K ( L b , L cut ) α , L = L X erg s − (9)The average normalization is K = 440 . ± . M (cid:12) , α = 1 . α = 1 . L b , = 0 . L b , = 5 . Γ = 1 .
56, and the combined exposure time of 262 ks, the estimated source detectionsensitivity would be 6 × erg s − . However, due to the presence of copious diffuse emission in M60, the actual sourcedetection sensitivity is significantly higher than this. Based on a sensitivity map that was computed using the CIAO LIM _ SENS task, we estimate that in most regions of M60 we detect sources brighter than L lim = 2 × erg s − . Based on this sourcedetection sensitivity and equations 8 and 9,we predict that the X-ray luminosity of unresolved LMXBs in M60 is L (cid:48)(cid:48) LMXB =2 . × erg s − in the 0 . − (cid:48)(cid:48)(cid:48)(cid:48)