Astronomical Science

Angular Momentum of Galaxies in the Densest ­Environments: A FLAMES/GIRAFFE IFS Study of the ­Massive Cluster Abell 1689 at z = 0.18

Francesco D’Eugenio1 of these empirical laws — alongside their bined with the collecting power of the Ryan C. W. Houghton1 very small scatter — imposes strong ESO Very Large Telescope (VLT), are the Roger L. Davies1 constraints on the structure and evolution ideal tools for this task. Elena Dalla Bontà2, 3 of ETGs (Bower et al., 1992). This makes them ideal testing grounds for any galaxy formation theory. Their study, important FLAMES/GIRAFFE observations of 1 Sub-department of Astrophysics, in its own right, is also fundamental Abell 1689 Department of Physics, University of for our understanding of the process of Oxford, United Kingdom structure formation in the Universe. Abell 1689, a massive galaxy cluster at 2 Dipartimento di Fisica e Astronomia redshift z = 0.183, has regular, concen- “G. Galilei”, Università degli Studi di The advent of integral field spectroscopy tric X-ray contours suggesting that it Padova, Italy (IFS) started a revolution in the study of is­ re­laxed. Its X-ray luminosity outshines 3 INAF–Osservatorio Astronomico di ETGs. The SAURON survey discovered Coma by a factor of three, and Virgo Padova, Italy the existence of two kinematically distinct by over an order of magnitude. Its comov- classes of ETGs, slow and fast rotators ing distance is 740 Mpc, giving a scale (SRs and FRs, see Emsellem et al. [2007]) of 1 arcsecond per 3.0 kpc. Alongside its Early-type galaxies (ETGs) exhibit kine- and ­Cappellari et al. (2007). The former physical properties, Abell 1689 was an matically distinct slow and fast rotator have little or no rotation, exhibit kinemati- ideal target for FLAMES because the (SR, FR) morphologies. The former cally decoupled cores and misalignment spatial resolution of the GIRAFFE-deploy- are much less common (10% of ETGs), between kinematics and photo­metry. The able integral field units (IFUs) samples up but their incidence is higher in the core latter are flattened systems, compatible to one effective radius (Re) for most gal- of the Virgo Cluster (25%). Here we pre- with rotational symmetry and a disc-type axies. Finally, a wealth of archival data, sent FLAMES/GIRAFFE integral field origin. The new division c­ rucially crosses including imaging from the Hubble Space spectroscopy of 30 galaxies in the mas- the boundary between Es and S0s, in Telescope (HST) Advanced Camera for sive cluster Abell 1689 at z = 0.183. that FRs populate both morphological Surveys (ACS), is available and vital to a Abell 1689 has a density 30 times classes. ATLAS3D (the ­volume-limited fol- study of this nature. higher than that of Virgo, making it the low-up survey to S­ AURON; Cappellari ideal place to test the effects of envi- et al., 2011a; Emsellem et al., 2011), F625W-band imaging from the HST ACS, ronment, such as local density and established that 66% of morphological combined with g:- and r:-band GEMINI cluster properties. We find 4.5 ± 1.0 ellipticals are FRs, and thus share the imaging were used for the photometry

SRs (or an average ETG fraction, ƒSR, of same internal structure as S0s. This is (Houghton et al., 2012). The spectroscopic 0.15 ± 0.03) in Abell 1689, identical to evidence for a new classifi­cation para- data (spectral resolution, R = 11 800, the value for field/groups in ATLAS3D. digm, based on kinematics rather than spectral range 573 nm < λ < 652 nm Within Abell 1689 ƒSR increases towards morphology (Cappellari et al., 2011b). [486 nm < λ < 552 nm rest frame]) covers the centre, ex­ceeding the value found in standard V-band absorption features. the core of Virgo. This work is the high- The ATLAS3D team presented the kine- GIRAFFE provides 15 independent mini est redshift study of its kind. matic morphology–density (kT–Σ) relation, IFUs, deployable anywhere on the focal analogous to the morphology–density plane; each IFU is positioned by a mag- relation (Dressler, 1980). It links the frac- netic button and contains an array of

Kinematical classification of ETGs tion of SRs in the ETG population (ƒSR) 20 square microlenses. They are arranged with the local number density of galaxies: in four rows of six (with four “dead” cor-

ETGs comprise morphologically distinct they found that ƒSR is independent of ners) for a total field of view of 3 by 2 arc- elliptical (E) and lenticular (S0) galaxies. the environment density over five orders seconds. Each lenslet is then connected Despite their differences, Es and S0s of magnitude from field to group environ- to the spectrometer with a dedicated opti- have lots in common. They are both char- ments. But they noticed a sharp increase cal fibre bundle. Alongside the 15 IFUs, acterised by old stellar populations, which in ƒSR in the inner core of the Virgo Clus- the instrument also provides 15 fully de­ has earned them the attribute “early- ter, the highest density probed by the ployable sky fibres. type”. The average ETG has little or no ATLAS3D survey. Virgo is an unrelaxed, cold gas, which is reflected in the star low-density cluster, but what would be Since the magnetic buttons are larger formation rate. Its light profile is smooth measured in the denser environments (10 arcseconds) than the IFU field of view, and its shape fairly regular. One of the beyond the local Universe? Addressing they cannot be deployed closer than a most puzzling facts about these galaxies this question gives further insight on minimum distance of 11 arcseconds. is how, with masses and luminosities that the kT–Σ relation, and on the processes GIRAFFE permits the observer to target span several orders of magnitude, they that drive galaxy formation and evolution. 15 objects simultaneously and we chose obey a number of tight scaling relations. to target 30 galaxies as a compromise These include the colour–magnitude Since rich, relaxed clusters are rare, they between sample statistics and integration ­relation, the colour–σ and Mg–σ relations can only be found at higher redshifts. time. In order to gain the maximum pos- (where σ is the velocity dispersion) and the The multiplexing capabilities of FLAMES/ sible signal-to-noise ratio, we initially fundamental plane. The mere existence GIRAFFE (Pasquini et al., 2002), com- selected the 30 ETGs with the highest

The Messenger 151 – March 2013 37 Astronomical Science D’Eugenio F. et al., Angular Momentum of Galaxies in the Densest ­Environments

01 02 03 Figure 1. Kinematic –150/150 0/225 –100/100 0/125–150/1500/225 maps of the Abell 1689 sample. For each of the 30 galaxies in the sam- ple we present a set of four images. The first 04 05 06 –100/100 0/125 –100/100 0/300–100/1000/325 one shows FLAMES/ GIRAFFE IFU footprints superimposed on HST imaging (Gemini imaging for target 20). The sec- ond plot shows the 07 08 09 –100/100 0/150 –100/100 0/225–100/1000/150 reconstructed image from VLT integral field spectroscopy, where each square is a spaxel and corresponds to a 10 11 12 lenslet of the IFU. Also –100/100 0/225 –100/100 0/150–100/1000/350 shown is an isophote at

either Re, or the closest integer fraction that fits into the IFU footprint. The four black corners 13 14 15 –100/100 0/200 –175/175 0/325–100/1000/200 correspond to unavaila- ble “dead” spaxels, while other black spax- els (seen in 11, 15 and 30) correspond to bro- ken or unused fibres. 16 17 18 –100/100 0/175 –250/250 0/325–125/1250/200 Velocity and velocity dispersion maps are depicted in the third and fourth plots. The black compass arrows show 19 20 21 north and east direc- –100/100 0/275 –100/100 0/150–100/1000/150 tions. Colour-bar limits are given in km/s.

22 23 24 –200/200 0/250 –100/100 0/200–100/1000/200

25 26 27 –200/200 0/225 –100/100 0/275–100/1000/250

28 29 30 –175/175 0/175 –150/150 0/200–100/1000/200

surface brightness within a 3-arcsecond was exposed five times for two hours, for tions of the Virgo Cluster (for which the radius. This sample was then subject a total of ten hours per galaxy. The ob­ SR population is known from ATLAS3D) to two practical constraints. We needed servations were carried out in visitor using the luminosity function of our sam- all of our targets to have high-resolution mode, which proved to be both efficient ple. This showed that we could recover

HST imaging, which limited our choice to and accurate in terms of object acquisi- the true value of ƒSR for Abell 1689. candidates in the innermost regions of tion. Excellent seeing of 0.60 arcseconds Although we did not do a colour selec- the cluster. The 11-arcsecond proximity reduced the correlation between neigh- tion, our sample falls entirely on the Red constraint ruled out some targets in the bouring spaxels. Sequence (RS). However, the number most crowded regions, forcing us to re- of galaxies that do not fall on the RS in select from a reserve list. This left us with Our sample is biased towards bright Abell 1689 is extremely low, and we 29 galaxies inside the HST field of view ob ­jects, which in turn could bias us to found the bias to be minimal. and one outside (galaxy 20). Each plate detect more SRs. We simulated observa-

38 The Messenger 151 – March 2013  

           

               (%4  (%4   1 1    h h  

                                                       

¡D ¡D

Figure 2. λR(IFU) vs. εe for the simulated observations in the Virgo core (Cappellari et al., 2011b). Figure 3. λR(IFU) vs. εe for all target galaxies in Abell of the SAURON sample of galaxies. The green lines However, contamination from face-on 1689. The green line separates FRs (blue dots separate SRs (below) from FRs. Red and blue circles above) from SRs (red dots below). The magenta and denote SRs and FRs respectively, according to the discs, which may appear as SRs, can dashed lines represent the view of an axisymmetric original SAURON classification (Emsellem et al., bias the estimate. galaxy, edge-on and at various inclinations. Details 2007). Due to the different observing setup, some about these models are given in Cappellari et al. galaxies have been misclassified: these are the red Five more objects, despite exhibiting (2007) and Emsellem et al. (2011). dots above the green line and the blue dots below. large-scale rotation, have misaligned ­kinematic and photometric axes, a fea- To reduce the FLAMES/GIRAFFE data we ture more common in SRs than in FRs introduces additional differences used the official ESO pipeline1, following ( ­Krajnovíc et al., 2011): these are galaxies between our setup and that of ATLAS3D, the guidelines ESO offers2. Stellar kine- 1, 3, 5, 9, 17 and 25 (Figure 1). Galaxies 3 but we used existing SAURON data to matics were extracted using pPXF3; we and 17 have very high ellipticities, and determine how our observations and def- used a line-of-sight velocity dispersion are thus unlikely to be SRs, which have inition compare to those of ATLAS3D. (LOSVD) expressed by a Gaussian func- ellipticity ε < 0.4. Galaxy 5 has high We made models of the SAURON galax- tion, obtaining just the velocity V and velocity dispersion, and also contains ies using kinemetry5 and, after projecting velocity dispersion σ. The stellar template an inner disc (R = 1.5 kpc) in the HST at redshift z = 0.183 and convolving with library of choice was the high-resolution imaging. the seeing, we sampled them using the version (R = 40 000) of the ELODIE tem- FLAMES/GIRAFFE setup. The resulting plate library4. simulated data, after adding noise, have

λR measurements and kinematic classifi- been used to measure λR, which we then cation compared with the original SAURON Slow and fast rotators at z = 0.18 ­values to estimate both bias and system- Emsellem et al. (2007) introduced the atic error.

Figure 1 shows the resulting kinematic estimator λR to measure the projected maps. Although the spatial resolution is specific angular momentum of galaxies; We corrected λR(IFU) according to the low (compared to SAURON) rotation can and Emsellem et al. (2011) further show bias measured and included the system- be clearly seen in some galaxies, and not how the combination of λR and ε con­ atic error in quadrature with the random in others. veniently captures the kinematic bound- error. This correction takes into account ary between SRs and FRs. They define both the different apertures between

Since our spatial resolution is coarse, we λR(Re) as the value of λR computed inside λR(IFU) and λR(Re) and the dif­ferent spatial cannot detect kinematically decoupled Re, and use it in their λR(Re) vs. ε diagram. resolutions between λR(IFU) and εe. In cores (KDCs) and double σ peaks (2σ) as Figure 2 we plot simulated v­ alues of in Krajnovíc et al. (2011). If we try to In our study the galaxies are not sampled λR(IFU) against published values of εe detect SRs from the velocity maps by eye evenly, because Re varies while the size (Emsellem et al., 2007). Despite the afore- (as done by the SAURON and ATLAS3D of the IFUs is fixed. We cannot follow the mentioned differences, there is little teams) we identify at most six: galaxies 4, ATLAS3D prescription precisely and there- (< 10%) misclassification in our diagram,

8, 12, 20, 26 and 27 (see Figure 1). The fore introduced λR(IFU); defined as the especially at high values of Re. Given overall value of ƒSR in the sample would value of λR computed using all the availa- the known uncertainties, we can calcu- then be 0.20, in line with what was found ble spaxels in the IFU field of view. This late the probability distribution for the

The Messenger 151 – March 2013 39 Astronomical Science D’Eugenio F. et al., Angular Momentum of Galaxies in the Densest ­Environments

measured number of SRs in the SAURON Figure 4. Fraction of slow rotators ƒSR         over the ETG population, as a function survey (galaxies below the green line in    of the environment density. The green Figure 2 defined by 0.31√ε and the green   circles and line are from the ATLAS3D line in Figure 2, Emsellem et al. [2011]). survey (Cappellari et al., 2011b), red We adopt a Monte Carlo approach (for   symbols are for Abell 1689. The num- each galaxy we assumed Gaussian bers at the top give the total number of   galaxies in each bin. The uncertainty errors in λR).­ The re­sulting probability dis- in the SR classification is reflected 21 tribution is Gaussian-like and we find c   in the error bars. The green square is 12.3 ± 1.7 slow rotators, where the true the value of ƒSR that we measure, value is 12.   ­resampling Virgo using our sample 5HQFNBNQD luminosity function. The error bars are   smaller than the point size. The aver- When we correct the values for our Abell age fraction for Abell 1689 is shown by 1689 data in the same way (Figure 3),    the red square. we can similarly calculate the probability   distribution for the number of SRs in l  KNF- ,OBl Abell 1689. This analysis finds 4.5 ± 1.0  slow rotators, corresponding to ƒSR = 3D 0.15 ± 0.03. starts exactly where ATLAS finished. age particle mass. Hence the higher ƒSR Abell 1689 shows a sharp increase in ƒSR in the core of clusters is consistent with Emsellem et al. (2007) warn about using with projected density, from ƒSR = 0.01 in the effects of dynamical friction. only λR to assign a galaxy to either the least dense environment to ƒSR = 0.58 the slow or fast rotator class. The dis- in the innermost regions. The densest On account of the importance of the ­ crepancy between the classifi­cation “by bin in Abell 1689 has a higher fraction of SR/FR division in understanding galaxy eye” and the classification we adopted SRs than the core of Virgo (Figure 4). formation/evolution, it is crucial to expand here underscores that warning. However, The intermediate bin has a value of ƒSR this study, in particular by increasing the when studying galaxies beyond the local compatible with both the field/group number of observed clusters, to quantify

Universe, a detailed analysis such as that ­envi ­ronments and the overall Virgo clus- the scatter in ƒSR and the radial variation carried out by the ATLAS3D team is not ter value, but is less than the Virgo core. within different clusters. Sampling clus- feasible. We are thus forced to rely on a ƒSR in the least dense bin is lower than ters with different densities and dynami- statistical approach. the ATLAS3D field and group values. cal states would give us great insight on the topic. Observations of the Abell 1650 Considering the whole Abell 1689 sam- cluster, with an intermediate mass, lying

Slow rotators and environment density ple, we find an average value of log10 Σ3 = between that of Virgo and Abell 1689, are 2.77 and a SR fraction of 0.15 ± 0.03 (red already scheduled at the VLT. Interesting For each galaxy in the sample we esti- square in Figure 4), which is the same times are ahead! mate the local environment density as the overall SR fraction in the Virgo ­following Cappellari et al. (2011b). We cluster, when sampled in the same way References defineΣ 3 as the number density of galax- (green square). Furthermore, both values ies inside the minimum circular area, are similar to the field and group samples Bower, G., Lucey, J. R. & Ellis, R. S. 1992, MNRAS, 3D ­centred on the target galaxy, and encom- in ATLAS , suggesting little to no dif­ 254, 613 passing three other galaxies (down to ference in ƒSR when it is averaged over an Cappellari, M. et al. 2007, MNRAS, 379, 418 a magnitude limit). Interlopers were dealt entire cluster population. Cappellari, M. et al. 2011a, MNRAS, 413, 813 with by subtracting everywhere a con- Cappellari, M. et al. 2011b, MNRAS, 416, 1680 Dressler, A. 1980, ApJ, 236, 351 stant mean value of Σ3 = 0.49 galaxies Abell 1689 has a higher average density Emsellem, E. et al. 2007, MNRAS, 379, 401 per square arcminute. In Figure 4 we than Virgo, but the same average slow Emsellem, E. et al. 2011, MNRAS, 414, 818 Houghton, R. C. W. et al. 2012, MNRAS, 423, 256 show ƒSR versus log10 Σ3 for Abell 1689 rotator fraction ƒSR. Inside the cluster, ƒSR (red), compared to the results of the rises with projected density. In the least Krajnovíc, D. et al. 2011, MNRAS, 414, 2923 3D Pasquini, L. et al. 2002, The Messenger, 110, 1 ­ATLAS survey (green; Cappellari et al., dense region, ƒSR is significantly smaller 2011b). The densest environment in than the ATLAS3D field/group value. Given ­ATLAS 3D (i.e., the core of the Virgo Clus- the low number of galaxies per bin, we Links ter) has ƒ = 0.25, double that typi- cannot rigorously claim that this is repre- SR 1 ESO Giraffe Pipeline: http://www.eso.org/sci/soft- cally found in less dense environments sentative. However, a similar “depletion” ware/pipelines/giraffe/giraf-pipe-recipes.html 2 (ƒSR ≈ 0.12). is observed in the outskirts of the Virgo Guidelines for Giraffe Pipeline: ftp://ftp.eso.org/ cluster (Cappellari et al., 2011b). SRs are pub/dfs/pipelines/giraffe/giraf-manual-2.8.7.pdf 3 In this study we probed e­ nvironments ­uniformly distributed across a range of pPXF: http://www-astro.physics.ox.ac.uk/~mxc/idl/ 4 Elodie template library: www.obs-u-bordeaux1.fr/ with values of log10 Σ3 between 2.06 environments; we know that they are on m2a/soubiran/elodie_library.html and 3.75: the minimum is comparable to average more massive than FRs, and 5 Kinemetry: www.davor.krajnovic.org/idl the core of Virgo, and the maximum is that dynamical friction is more efficient in 1.7 dex higher. In this respect our work objects with mass higher than the aver-

40 The Messenger 151 – March 2013