Hubble Space Telescope Cycle 19 GO Proposal 909

Ghost, Dark, Stripped, and Bullet Clusters Unleashed by Pandora’s Cluster, Abell 2744

Principal Investigator: Dr. Dan Coe Institution: Space Telescope Science Institute Electronic Mail: [email protected] Scientific Category: COSMOLOGY Scientific Keywords: Clusters Of Galaxies, Dark Matter, Gravitational Lensing, Intracluster Medium Instruments: ACS

Proprietary Period: 0

Orbit Request Prime Parallel Cycle 19 8 0

Abstract

The Bullet Cluster presents us with a simple paradigm regarding collisions between galaxy clusters: gas is self- collisional causing it to be stripped from galaxies and dark matter which are not. However some cluster mergers appear to deviate from this prescription, likely because of more complicated merger physics yet to be fully understood or perhaps, as some suggest, due to dark matter self-collisionality. Abell 2744 is a merger of four galaxy clusters, making it one of the most active mergers known, and the only to feature a Mach ~3 shock front aside from the Bullet Cluster. Our recent analysis of this merger helps reveal several features which challenge our current understanding of galaxy cluster mergers. We find gas apparently leading rather than trailing mass in our "ghost" cluster, perhaps by far the largest "ram-pressure slingshot" yet observed. Mass appears to be offset from galaxies in our "dark cluster" (perhaps the first of its kind). And gas appears to be stripped more cleanly and to a greater distance (> 250 kpc) from one of our clusters than any other yet known. As we demonstrate below, HST imaging and additional analysis are required to verify and understand these strange behaviors unleashed by "Pandora's Cluster". We waive any proprietary period to this data. Dr. Dan Coe : Ghost, Dark, Stripped, and Bullet Clusters Unleashed by Pandora’s Cluster, Abell 2744 Investigators:

Investigator Institution Country PI Dr. Dan Coe Space Telescope Science Institute USA/MD CoI Dr. Renato A. Dupke (Co-PI) Eureka Scientific Inc. USA/CA CoI Dr. Julian Manuel Merten Universitat Heidelberg Germany CoI Dr. Richard J. Massey University of Edinburgh, Institute for Astronomy UK CoI Mr. Adi Zitrin Tel Aviv University - Wise Observatory Israel CoI Dr. Matt Owers Swinburne University of Technology Australia CoI Dr. Leonidas Moustakas Jet Propulsion Laboratory USA/CA CoI Dr. Jason Rhodes Jet Propulsion Laboratory USA/CA CoI Dr. Massimo Meneghetti INAF, Osservatorio Astronomico di Bologna Italy CoI Dr. Narciso Benitez Instituto de Astrofisica de Andalucia (IAA) Spain CoI Prof. Brenda L. Frye University of San Francisco USA/CA CoI Dr. Laerte Sodre Universidade de Sao Paulo Brazil CoI Dr. Jessica Krick California Institute of Technology USA/CA CoI Prof. Joel N. Bregman University of Michigan USA/MI Number of investigators: 14

Target Summary:

Target RA Dec Magnitude ACO-2744CLUSTER 00 14 6.7724 -30 22 47.38 V = 16.7 +/- 0.5

Observing Summary:

Target Config Mode and Spectral Elements Flags Orbits ACO-2744CLUSTER ACS/WFC Imaging F435W 3 ACO-2744CLUSTER ACS/WFC Imaging F606W 2 ACO-2744CLUSTER ACS/WFC Imaging F814W 3 Total prime orbits: 8

● Scientific Justification Each “bullet cluster” presents us with a unique and rare opportunity to improve our understanding of structure formation and test our understanding of dark matter The Bullet Cluster gave us for the first time direct empirical evidence of the existence of dark matter as well as upper limits on its self-collisional cross-section (Markevitch04, Clowe06, Randall08). In the aftermath of this galaxy cluster merger, we find gas stripped from galaxies due to collisional pressure, while dark matter and galaxies appear to have passed cleanly through. Abell 2744 (hereafter, A2744; z = 0.308) appears to include a very similar cluster collision of Mach ~3, including a gas shock front, but as viewed from an angle closer to our line of sight (based on analysis of Chandra images by Owers11). Our analysis of this collision yields constraints on dark matter particle self-interaction cross section (σ/m < ~3 cm2/g; (Merten11) which are very similar to those obtained from the Bullet Cluster. Two additional clusters also appear to have participated in this merger, making it one of the most complex known. As revealed in part by our analyses, these additional clusters exhibit puzzling and rich phenomenology yet to be fully explained. While the Bullet Cluster provided a straightforward explanation regarding the interplay between galaxies, gas, and dark matter, other cluster mergers do not always follow this narrative. The “Baby Bullet” (Bradac08) is one example that does appear to behave as the Bullet Cluster with gas being stripped from galaxies and mass. A possible counterexample was found in Abell 520 “the cosmic train wreck” (Mahdavi07), which appeared to reveal a dark matter core devoid of, and perhaps collisionally stripped from, gas and galaxies (although see Okabe08). A recent paper (Williams11) has also suggested that collisional dark matter may have been stripped from a galaxy merging with the core of the nearby cluster Abell 3827. These suggestions of collisional dark matter are tempered with possible alternative explanations. In each case, we are challenged us to either attain a new understanding of how mergers may proceed (perhaps by recreating the observed behavior in simulations) or else question our underlying theories about dark matter as a virtually collisionless particle. A2744 appears to exhibit several such challenging cases, yet further observations are required to confirm this. Collisions which exhibit shock fronts are especially valuable, providing us with information about merger dynamics and plasma physics, as well as perhaps dark matter collisionality. Such shock fronts have only been detected so far in a handful of clusters (~5- 8; Markevitch10 and references therein). Thus each presents a rare and unique opportunity to study such an energetic event. Ghost, Dark, Stripped, and Bullet Clusters in A2744 Only A2744 so far appears to exhibit a shock front with a velocity similar to that of the Bullet Cluster (Mach ~3 in three dimensions, Owers11). Shocks detected in other cluster mergers are significantly weaker (Mach ~1.6 -- 2, Markevitch10). In addition to

this “bullet”, A2744, which we dub “Pandora’s Cluster”, appears to have unleashed a “ghost” cluster, a “dark” cluster, and perhaps the most significantly stripped cluster yet observed (see Fig. 1). These configurations, which reside in the less well-studied Western half of the merger (where we propose imaging), have yet to be well explained, though progress has been made (including Owers11, Merten11). 14 We detected four separate mass clumps of ~10 M (within 250 kpc) with our gravitational lensing analysis of recently acquired HST/ACS images supplemented by VLT and Subaru images (Merten11). Participating in this bullet merger are two or three of these clumps, including our “dark cluster” in the NW. Contratry to expectations, we find the brightest galaxies in this NW region appear to be leading this mass clump by significant distances (~150 and ~300 kpc). Just as surprisingly, a gas clump, our “ghost cluster”, is leading still further ahead (~450 kpc). Though long assumed to be trailing the galaxies, Owers11 present evidence that this gas clump (which they call the “interloper”) is indeed leading rather than trailing, and they suggest it has undergone a “ram-pressure slingshot”. In this scenario, the gas previously trailing the mass has since caught up and been flung around to the other side (Markevitch07, Ascasibar06). Such an effect appears to have been observed but on a much smaller scale in Abell 168 (Hallman04). If confirmed, the A2744 “slingshot” would be by far the largest observed to date. HST observations and dynamical simulations will reveal whether some mechanism, perhaps involving multiple mass clump accretion may have enhanced this separation. 14 Directly West of the core we detect a fourth ~10 M mass clump (“W” in Fig. 1). This clump is coincident with bright galaxies but significantly stripped of gas. Gas appears to have been stripped > 250 kpc off of this mass clump, the largest separation between gas and mass yet observed (compared to ~200 kpc for the Bullet Cluster; see Shan10). A faint X-ray trail does appear to lead back from this clump back to the core of dense X-ray gas (Merten11). HST imaging significantly improves our lensing-based mass models Toward understanding this mysterious behavior, we propose HST imaging of the Western half of this complex cluster merger. This region has received less attention in the literature to date, but our new lensing analysis has revealed significant mass clumps in this region, warranting follow-up study. As we clearly demonstrate in Fig. 3 (and discuss below), HST imaging significantly improves our mass models in both recovered amplitude and resolution. Where we lack this coverage, our lensing signals may well be diluted, yielding underestimated masses or even undetected mass clumps. Our NW galaxies, for example, lie just outside our Cycle 17 HST FOV. Our proposed HST coverage of these galaxies will either a) confirm their mass deficit and the strange offset, or b) reveal previously underestimated mass clumps well aligned with the galaxies. Using the mass map reconstruction method of Merten09 (see Bradac05 for a similar method), we compare our results with and without HST data. The HST images yield

improved constraints from weak lensing measurements as well as strong lensing features which could not be identified in the ground-based images. Each individually contributes additional robust signal which increases the amplitude of our mass model. Without this data, our estimate of the mass of the central core, for example, is ~60% lower. This lower mass density is clearly incorrect, as it would fail to reproduce the strong lensing features we observe. Our Cycle 17 ACS images yield ~60 galaxies / arcmin2 for our weak lensing analysis, a three-fold improvement compared to our deep VLT images (Cypriano04 reanalyzed). Space-based galaxy shape measurements are also higher precision due to the smaller, more stable PSF (Kasliwal08). With our Cycle 17 ACS imaging, we also identified 34 multiple images of 11 strongly lensed galaxies (Fig. 4). None of these had been previously identified in ground- based images. As shown in Fig. 3, this strong lensing data also significantly improves our mass model recovery. Similar improvement has been demonstrated in analyses of simulated clusters with known input masses (Meneghetti10b). Resolving the mysteries surrounding this complex and active cluster merger Our analysis of Cycle 17 HST observations, combined with data from other facilities, raised important new questions regarding the history of this merger. By obtaining HST images of the Western half of “Pandora’s Cluster” we will learn: ● Does the NW “dark cluster” in fact lag (surprisingly) behind its associated galaxies? ● Does it have a double mass peak which might help explain a large ram-pressure slingshot of the “ghost cluster”? ● Does the “ghost cluster” have any significant associated mass? ● What is the precise location of the W mass clump and has its gas been stripped more cleanly and to a greater distance than any other such mass clump known? Once these facts have been established, we will work to explain any confirmed odd behaviors. Dynamical simulations will be performed in Heidelberg in an attempt to recreate and better understand the physics behind the merger (as in Springel07). We note we will also be proposing for data from other facilities to help further establish these facts. Deeper X- ray observations will further probe the interplay between gas and mass. Spectra of the lensed arcs will help normalize our mass models. And deeper spectroscopy of cluster members will enable us to detect and characterize dynamical substructure in the western region. The unique opportunity of Pandora An HST Multi-Cycle Treasury program (CLASH) is currently underway to study 25 galaxy clusters. The majority of these clusters are relaxed, as one of the main science goals is to study their mass profiles. Five of the CLASH clusters were instead selected based on their strength as gravitational lenses. One of these, MACS J0717, is a merger nearly as complex as A2744, though lacking a “bullet” shock front. We stress the uniqueness of A2744 and thus the importance of our proposed observations. There exist only a handful of opportunities in our universe to study these rare glimpses into such actively merging clusters.

Fig. 1 -- A2744 is likely a quadruple cluster merger. Shown here are mass contours from our gravitational lensing analysis (Merten11). Mass is also shaded blue and (Chandra) X- ray gas red, overlaid on a VLT color image (VRI filters). A2744 includes a Bullet Cluster configuration (S -- N and/or NW), as well as clumps which we call “dark” (dark matter only) and “ghost” (gas only). All clumps appear to be stripped of gas, though the W clump appears to have its gas stripped by >250 kpc, the greatest such distance yet observed. The NW configuration is especially puzzling, as gas appears to be leading galaxies, which in turn appear to be leading mass.

Fig. 2 -- We propose to image the Western half of the merger, as outlined in green at right. This region has received less attention to date in the literature, yet we have detected significant mass clumps there warranting further study. These new observations will enable us to recover significantly greater lensing signals, improving our mass models in these regions. For example, HST coverage will enable us to make more Orange: Cycle 17 ACS imaging definitive claims about any Green: proposed ACS imaging separation between mass and Yellow: objects of interest galaxies in the NW clump. Color image as in Fig. 1; Black & white: Subaru i-band

Each square field is 10′ (2.7 Mpc) on a side Fig. 3 -- HST/ACS imaging significantly improves our mass models of A2744. Shown are our mass reconstructions (in units of critical lensing density) based on various subsets of our data: 1) weak lensing from VLT and Subaru; 2) adding weak lensing from ACS (observed area outlined in white); 3) adding strong lensing from ACS (all data); 4) As in 3, but higher resolution, as enabled by the additional data (and requiring significantly more CPU time). Note the mass map from ground-based weak lensing alone is significantly underestimated, yielding, for example, a ~60% lower mass of the cluster core. This lower mass density is clearly incorrect as it would be insufficient to produce the strong lensing features we observe. Similar results have been obtained applying this analysis method to simulated clusters of known mass (Meneghetti10b).

Fig. 4 -- In our HST/ACS BVi imaging, we identified 34 multiple images (labeled here) of 11 strongly-lensed galaxies using the method of Zitrin09. As shown in Fig. 3, this information significantly improves our ability to resolve mass clumps in this region, both in amplitude and positional precision. We expect similar gains from strong lensing features revealed in our proposed HST imaging of the W and NW clumps. Note that none of these multiple image systems could be identified in lower resolution ground- based imaging. Multiband HST imaging is essential to provide ACS BVi color image of the cluster core resolved color information 100″ ~ 450 kpc on each side toward correctly matching these Multiple images of strongly lensed galaxies faint objects. and lensing critical curve labeled in white

References Markevitch10 arXiv:1010.3660 Ascasibar06 ApJ 650, 102 Meneghetti10a A&A 519, 90 Benitez09 ApJL 692, 5 Meneghetti10b A&A 514, A93+ Bradac05 A&A 437, 39 Merten09 A&A 500, 681 Bradac06 ApJ 652, 937 Merten11 MNRAS, submitted Bradac08 ApJ 687, 959 Okabe08 PASJ 60, 345 Clowe06, ApJ 648, L109 Owers11 ApJ 728, 27 Cypriano04 ApJ 613, 95 Randall08 ApJ 679, 1173 Hallman04 ApJ 610, L81 Shan10 MNRAS 406, 1134 Kasliwal08 ApJ 684, 34 Springel07, MNRAS 380, 911 Mahdavi07 ApJ 668, 806 Williams11 arXiv:1102.3943 Markevitch04 ApJ 606, 819 Zitrin09 MNRAS 396, 1985 Markevitch07 Phys. Rep 443, 1

● Description of the Observations

In Cycle 17 we obtained ACS observations in two pointings with ~50% overlap each with 8 orbits divided among 3 filters. We have demonstrated that these observations yield significant improvements in our ability to map the mass of these clusters. Here we aim to extend the area over which we can obtain consistent results. Thus we propose one additional pointing to image the remaining (and more intriguing) massive components in the Western half of this cluster merger (see Fig. 2). For consistency, we propose the same filter set to the same depths as in Cycle 17: F435W (3 orbits), F606W (2.5 orbits), F814W (2.5 orbits). (Apologies that depths are rounded to full orbits in the APT file.) As we argued in our previous proposal, this filter set allows one to efficiently identify cluster galaxies as they are well segregated in color-color space. And in practice, we find these filters yielded surprisingly good photometric redshifts for galaxies at z < 0.7. Of 118 galaxies with spectroscopic redshifts (all z < 0.7) within our FOV, 99 had confident photo-z and these proved accurate to Δz ~0.06(1+z) RMS with no significant outliers (all Δz < 0.3).

As with our Cycle 17 data, key analysis tasks will be led by the following people: ● D Coe -- image reduction, photometric redshifts ● R Massey -- weak lensing analysis of ACS images, including CTE/I corrections ● A Zitrin -- strong lensing analysis: image identification and modeling ● J Merten -- strong + weak lensing mass modeling

R Dupke and M Owers will lead analysis and interpretation of the Chandra X-ray data in context with our mass model results.

● Special Requirements

None.

● Coordinated Observations

None

● Justify Duplications

A2744 has been previously observed with WFPC2 in Cycles 4 and 16 and by our group with ACS in Cycle 17 (see Fig. 2). The WFPC2 observations are centered on the cluster core and do not overlap significantly with our proposed observations. Our previous ACS observations imaged the Eastern half of this complex merger, and here we propose to image the Western half. Small areas of overlap are necessary to include all regions of interest.

● Past HST Usage and Current Commitments

As a member of the ACS GTO science team, the P.I. (Coe) has worked on multiband Hubble ACS images of galaxy clusters since the inception of that instrument. Specifically he worked on A1689 (GO 9289), MS1358 (GO 9717, 9292, 10325), and CL0024 (GO 10325). He also analyzed ACS (and NICMOS) images of the UDF.

Coe supervised and contributed to the analysis of our Cycle 17 ACS observations of A2744 (GO 11689). He is currently a member of CLASH (GO 10265). The analysis tools for the proposed project are largely in place, as described above. If this proposal is accepted we will allot time and resources to rapidly perform and publish these analyses.

Selected recent HST publications: Coe10, ApJ 723, 1678 “A High-resolution Mass Map of Galaxy Cluster Substructure: LensPerfect Analysis of A1689” Coe06, AJ 132, 926 “Galaxies in the Hubble Ultra Deep Field. I. Detection, Multiband Photometry, Photometric Redshifts, and Morphology” Merten11, MNRAS, submitted (based on our Cycle 17 observations) Merten09, A&A 500, 681, “Combining weak and strong cluster lensing: applications to simulations and MS 2137” Massey10, MNRAS 409, 109 “Charge transfer inefficiency in the Hubble Space Telescope since Servicing Mission 4” Massey07, Nature 445, 286 “Dark Matter Maps Reveal Cosmic Scaffolding” Zitrin11, MNRAS 410, 1939 “Strong-lensing analysis of a complete sample of 12 MACS clusters at z > 0.5: mass models and Einstein radii” Zitrin09, MNRAS 396, 1985, “New multiply-lensed galaxies identified in ACS/NIC3 observations of Cl0024+1654 using an improved mass model”