Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.stsci.edu/~golim/papers/mei2006.pdf Äàòà èçìåíåíèÿ: Wed May 3 17:43:32 2006 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 07:13:14 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï

The Astrophysical Journal, 639:81 ­ 94, 2006 March 1
# 2006. The American Astronomical Society. All rights reserved. Printed in U.S.A.

EVOLUTION OF THE COLOR-MAGNITUDE RELATION IN HIGH-REDSHIFT CLUSTERS: BLUE EARLY-TYPE GALAXIES AND RED PAIRS IN RDCS J0910+5422
S. Mei,1 J. P. Blakeslee,1 S. A. Stanford,2, 3 B. P. Holden,4 P. Rosati,5 V. Strazzullo,5, 6 N. Homeier,1 M. Postman,1, M. Franx,8 A. Rettura,5, 9 H. Ford,1 G. D. Illingworth,4 S. Ettori,10 R. J. Bouwens,4 R. Demarco,1 A. R. Martel,1 M. Clampin,7 G. F. Hartig,7 P. Eisenhardt,11 D. R. Ardila,1 F. Bartko,12 N. Benitez,13 L. D. Bradley,1 ´ T. J. Broadhurst,14 R. A. Brown,7 C. J. Burrows,7 E. S. Cheng,15 N. J. G. Cross,16 P. D. Feldman,1 D. A. Golimowski,1 T. Goto,1 C. Gronwall,17 L. Infante,18 R. A. Kimble,19 J. E. Krist,7 M. P. Lesser,20 F. Menanteau,1 G. R. Meurer,1 G. K. Miley,8 V. Motta,18 M. Sirianni,7 W. B. Sparks,7 H. D. Tran,21 Z. I. Tsvetanov,1 R. L. White,7 and W. Zheng1
Received 2005 June 7; accepted 2005 October 26
7

ABSTRACT The color-magnitude relation has been determined for the RDCS J0910+5422 cluster of galaxies at redshift z ¼ 1:106. Cluster members were selected from the Hubble Space Telescope Advanced Camera for Surveys (HST ACS) images, combined with ground-based near-IR imaging and optical spectroscopy. The observed early-type color-magnitude relation (CMR) in i775 þ z850 versus z850 shows an intrinsic scatter in color of 0:060 ô 0:009 mag , within 10 from the cluster X-ray emission center. Both the elliptical and the S0 galaxies show small scatter about the CMR of 0:042 ô 0:010 and 0:044 ô 0:020 mag, respectively. From the scatter about the CMR, a mean luminosity ­ weighted age t > 3:3 Gyr (zf % 3) is derived for the elliptical galaxies, assuming a simple stellar population modeling (single-burst solar metallicity). Strikingly, the S0 galaxies in RDCS J0910+5422 are systematically bluer in i775 þ z850 ,by0:07 ô 0:02 mag, than the ellipticals. The ellipticity distribution as a function of color indicates that the face-on S0s in this particular cluster have likely been classified as elliptical. Thus, if anything, the offset in color between the elliptical and S0 populations may be even more significant. The color offset between S0 and E galaxies corresponds to an age difference of %1 Gyr for a single-burst solar-metallicity model. A solar-metallicity model with an exponential decay in star formation will reproduce the offset for an age of 3.5 Gyr ; i.e., the S0s have evolved gradually from star-forming progenitors. The early-type population in this cluster appears to be still forming. The blue early-type disk galaxies in RDCS J0910+5422 likely represent the direct progenitors of the more evolved S0s that follow the same red sequence as elliptical galaxies in other clusters. Thirteen red galaxy pairs are observed, and the galaxies associated in pairs constitute $40% of the CMR galaxies in this cluster. Subject headings: galaxies: clusters: individual ( RDCS J0910+5422) -- g galaxies: elliptical and lenticular, cD -- galaxies: evolution

1. INTRODUCTION The Advanced Camera for Surveys (ACS; Ford et al. 2003), by virtue of its high spatial resolution and sensitivity, allows us to study galaxy clusters in great detail up to redshifts of unity and beyond. At these redshifts, galaxy clusters are still assembling, and galaxies are evolving toward the populations that we observe today. Recent results from our ACS Intermediate Redshift Cluster Survey ( Blakeslee et al. 2003b; Lidman et al. 2004; Demarco et al. 2005; Goto et al. 2005; Holden et al. 2005a, 2005b; Homeier et al. 2005; Postman et al. 2005) have shown
1 Department of Physics and Astronomy, Johns Hopkins University, Baltimore, MD 21218; smei@pha.jhu.edu. 2 Department of Physics, University of California, Davis, CA 94516. 3 Institute of Geophysics and Planetary Physics, Lawrence Livermore National Laboratory, Livermore, CA 94551. 4 Lick Observatory, University of California, Santa Cruz, CA 95064. 5 European Southern Observatory, Karl-Schwarzschild-Strasse 2, D-85748 Garching, Germany. 6 ` Dipartimento di Scienze Fisiche, Universita Federico II, I-80126 Naples, Italy. 7 Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218. 8 Leiden Observatory, Postbus 9513, 2300 RA Leiden, Netherlands. 9 ´ Universite Paris-Sud 11, rue Georges Clemenceau 15, Orsay, F-91405, France. 10 INAF-Osservatorio Astronomico, via Ranzani 1, 40127 Bologna, Italy. 11 Jet Propulsion Laboratory, CalTech, 4800 Oak Grove Drive, Pasadena, CA 91125.

that galaxy clusters at redshift around unity show many similarities to local clusters in terms of galaxy populations and their distribution, but also significant differences in galaxy morphology, ellipticity, and mass-luminosity ratios. The strongest evolution observed in the early-type population is a deficit of a S0 population in this sample when compared to lower redshift samples ( Postman et al. 2005). This would be evidence that the formation of the S0 population is still under way in clusters at redshift unity. One of the most striking similarities is that the tight relation between early-type galaxy colors and luminosities that applies
Bartko Science and Technology, 14520 Akron Street, Brighton, CO 80602. ´ Instituto de Astrof ´sica de Andalucia (CSIC ), Camino Bajo de Huetor 50, i ´ Granada 18008, Spain. 14 School of Physics and Astronomy, Tel Aviv University, Tel Aviv 69978, Israel. 15 Conceptual Analytics, LLC, 8209 Woburn Abbey Road, Glenn Dale, MD 20769. 16 Royal Observatory Edinburgh, Blackford Hill, Edinburgh, EH9 3HJ, UK. 17 Department of Astronomy and Astrophysics, Pennslyvania State University, University Park, PA 16802. 18 ´ Departamento de Astronom´a y Astrofisica, Pontificia Universidad Catolica, i ´ Casilla 306, Santiago 22, Chile. 19 NASA Goddard Space Flight Center, Code 681, Greenbelt, MD 20771. 20 Steward Observatory, University of Arizona, Tucson, AZ 85721. 21 W. M. Keck Observatory, 65-1120 Mamalahoa Highway, Kamuela, HI 96743.
13 12

81


82

MEI ET AL.

Vol. 639

locally (the color-magnitude relation [CMR]) is already in place at redshifts as high as z $ 1:3 (e.g., Stanford et al. 1997; Mullis et al. 2005). The CMR in local samples of galaxy clusters presents universal properties in terms of scatter and zero point ( Bower et al. ´ 1992; van Dokkum et al. 1998, Hogg et al. 2004; Lopez-Cruz et al. 2004; Bell et al. 2004; Bernardi et al. 2005; McIntosh et al. 2005) that evolve back in time, in agreement with passively evolving models ( Ellis et al. 1997; Stanford et al. 1998; van Dokkum et al. 2000, 2001; Blakeslee et al. 2003b, 2005; Holden et al. 2004; De Lucia et al. 2004). ACS enables accurate measurement of the scatter around the CMR, with enough precision to seriously constrain galaxy formation age, which is impossible to obtain from ground-based data (see, for example, Holden et al. 2004). The measurement of the CMR scatter of the first cluster in our ACS cluster survey, RX J1252.9þ292, permitted us to constrain the mean luminosity ­ weighted age for the elliptical galaxies to >2.6 Gyr (z > 2:7; Blakeslee et al. 2003b), based on simple modeling. In this paper, we extend the results obtained in Blakeslee et al. (2003b) to RX J0910+5422. RX J0910+5422 is part of the ACS cluster survey (guaranteed time observation, GTO, program 9919) that includes eight clusters in the redshift range at 0:8 < z < 1:3, selected in the X-ray, optical, and near-IR ( Ford et al. 2004). RX J0910+5422 was selected from the Rontgensatellit (ROSAT ) Deep Cluster Survey ¨ ( Rosati et al. 1998) and confirmed with near-IR and spectroscopic observations by Stanford et al. (2002). Extensive follow-up spectroscopy at the Keck Observatory has been carried out in a magnitude-limited sample reaching Ks ¼ 20:0 mag in the central 30 (S. A. Stanford et al. 2006, in preparation). The mean redshift of the cluster was measured to be z ¼ 1:106 (Stanford et al. 2002). In this paper, we combine ACS imaging with ground-based spectroscopy and near-IR imaging to constrain galaxy ages and formation histories from the study of their CMR. We discuss the properties of the elliptical ( E ) and lenticular (S0) populations separately in the light of simple galaxy formation scenarios. 2. OBSERVATIONS RX J0910+5422 was observed in 2004 March with the ACS WFC ( Wide Field Camera) in the F775W (i775) and F850LP (z850) bandpasses, with total exposure times of 6840 and 11440 s, respectively. The ACS WFC scale is 0B05 pixelþ1, and its field of view is 210 00 ; 204 00 . The APSIS pipeline ( Blakeslee et al. 2003a), with a Lanczos 3 interpolation kernel, was used for processing the images. The ACS photometric zero points (AB system) are 25.654 and 24.862 mag in i775 and z850, respectively (Sirianni et al. 2005). A Galactic reddening of E (B þ V ) ¼ 0:019 toward RX J0910+5422 was adopted (Schlegel et al. 1998), with Ai 775 ¼ 0:039 and Az850 ¼ 0:029 (Sirianni et al. 2005). The ACS WFC field covers an area that at the redshift of this cluster, z ¼ 1:106, corresponds to %1Mpc2 in the Wilkinson Microwave Anisotropy Probe (WMAP) cosmology (Spergel et al. 2003; m ¼ 0:27, ö ¼ 0:73, and h ¼ 0:71, adopted as our standard cosmology hereafter). Figure 1 shows the ACS color image with X-ray contours from Chandra ACIS (Advanced CCD Imaging Spectrometer) data that have been adaptively smoothed (Stanford et al. 2002). Near-IR JKs and optical i-band images were obtained at Palomar Observatory as described in detail by Stanford et al. (2002). Optical spectroscopy of galaxies in RX J0910+5422 was obtained using the Low Resolution Imaging Spectrometer ( LRIS; Oke et al. 1995) on the Keck 1 and 2 telescopes (S. A. Stanford et al. 2006, in preparation). Our typical errors in redshift correspond to errors in velocity between 100 and 300 km sþ1. Objects for spectroscopy were chosen initially from the catalog of ob-

Fig. 1.--ACS color image with X-ray contours overlaid. The X-ray observations are from Chandra ACIS-I, over the energy range 0.5 ­ 2 KeV, and were smoothed with a 500 FWHM Gaussian. North is up, and east is to the left. In the enlargement, galaxy pairs in the central filamentary structure are shown. Two main groups of interacting galaxies lie along the filamentary structure at the center of the cluster. The first group is shown by the yellow arrow, the second one by the red arrows. Three other red sequence early-type galaxies are shown by the green arrows.

jects with Ks < 20:0 mag ( Vega magnitudes) within the IR imaging area; outside of this area objects were chosen with i > 21 mag from the i-band image to fill out masks. Our final sample included 66% of the objects with Ks < 20:0 mag. Spectra were obtained using the 400 lines mmþ1 grating for all runs except for the initial two discovery masks as reported in Stanford et al. (2002). Nine more masks were observed using LRIS during four runs between 2001 January and 2003 February. Usually each mask was observed in a series of four 1800 s exposures, with small spatial offsets along the long axis of the slits. On average, the seeing was 0B9. The blue-side data were generally not used, since the rest-frame wavelengths probed at z ¼ 1:1fall far to the blue of the spectral features of interest for galaxies in the cluster. In total, 149 redshifts were obtained. The slit mask data were separated into slitlet spectra and then reduced using standard long-slit techniques. A fringe frame was constructed for each exposure from neighboring exposures, each offset from the previous by 300 in an observing sequence for each mask and then subtracted from each exposure to greatly reduce fringing in the red. The exposures for each slitlet were reduced separately and then co-added. One-dimensional spectra were extracted for each targeted object, as well as the occasional serendipitous source. Wavelength calibration of the one-dimensional spectra was obtained from arc lamp exposures taken immediately after the object exposures. A relative flux calibration was obtained from long-slit observations of the standard stars HZ44, G191B2B, and Feige 67 ( Massey & Gronwall 1990). 3. OBJECT SELECTION AND PHOTOMETRY SExtractor ( Bertin & Arnouts 1996) was used to find objects in the i775 and z850 images and measure their magnitudes. Threshold and deblending settings were used as in Benitez et al. ´ (2004). Although we have extensive spectroscopy, the ACS imaging reaches considerably deeper along the cluster luminosity function. Thus, we have chosen to use the colors i775 þ z850 and J þ Ks to isolate a set of probable cluster members. In Figure 2, the i775 þ z850 and J þ Ks colors are shown as a function of galaxy age, using Bruzual & Charlot (2003, hereafter


No. 1, 2006

COLOR-MAGNITUDE RELATION FOR RDCS J0910+5422

83

larger. Our median Re is %5.5 pixels (%13 kpc at z ¼ 1:106). Our final results do not change (within the uncertainties) if the effective radii are calculated via a two-component (Sersic bulge + exponential disk) surface brightness decomposition technique using GIM2D ( Marleau & Simard 1998; A. Rettura et al. 2006, in preparation), which permits us to better fit the galaxy light profile. The photometric uncertainties due to flat-fielding, PSF variations, and the pixel-to-pixel correlation for ACS (Sirianni et al. 2005) were estimated by measuring the standard deviation of photometry in the background for circular apertures in the range of the measured effective radii. These photometric errors were added in quadrature to the Poisson uncertainties in the measured fluxes for each object. The derived errors in the colors are between 0.01 and 0.03 mag, down to z850 ¼ 24 mag. SExtractor MAG_ AUTO were used for the z850 magnitude in the CMR; these are fairly robust, although they may systematically miss a small fraction of the light ( Benitez et al. 2004). ´ We finally color-selected 34 early-type ( E, S0, and S0/a) galaxies with 0:8 mag < i775 þ z850 < 1:1 mag within 20 from the cluster center, taken as the center of the X-ray emission (Stanford et al. 2002). Images of the color-selected galaxies are shown in Figures 3, 4, and 5. Of the 34 color-selected galaxies, 15 are spectroscopically confirmed cluster members, one (S0/a, with magnitude z850 ¼ 24:2 mag) is a confirmed nonmember, and the others were not targeted for spectroscopy. The selection in i775 þ z850 at z ¼ 1:1 therefore appears to be robust: only 1 of the 16 selected galaxies with measured redshifts is a nonmember. We expect few of the other 18 to be interloper field galaxies. 4. COLOR-MAGNITUDE RELATION The color-magnitude relation (CMR) for the final colorselected objects is shown in Figure 7. Orange circles are elliptical galaxies, and orange squares and stars are S0 and S0/a galaxies, respectively. Smaller black symbols represent earlytype galaxies that do not lie on the red sequence. Small triangles are late-type galaxies. Boxes are plotted around confirmed cluster members. Confirmed interlopers are circled in the figure. Surprisingly, the two brightest cluster members are not ellipticals, but S0s. The brightest of these two galaxies lies %700 kpc (%1A2) from the cluster center, and the other bright S0 lies at %300 kpc (%0A6). Moreover, there are late-type galaxies with luminosities that are similar to the red sequence bright early-type galaxies ( Fig. 6 ). Two of them lie on the red sequence and are confirmed cluster members, at %80 kpc from the cluster center (see also below in the discussion of the color and morphology distribution as a function of distance from the cluster center). We fitted the following linear CMR to various subsamples of the galaxies: i
775

Fig. 2.--The i775 þ z850 and J þ K colors as a function of galaxy age. BC03 models with solar metallicity and age equal to 4 Gyr are shown with a solid line. The dashed lines are for half-solar metallicity, and the dot-dashed lines are for twice-solar metallicity. We considered a sample of color-selected galaxies with 0:8 mag < i775 þ z850 < 1:1 mag and J þ K > 1:45.

BC03) stellar population models, redshifted to z ¼ 1:106. Earlytype cluster members would have ages of at least 0.5 Gyr, corresponding to i775 þ z850 > 0:8 mag and J þ Ks > 1:45 mag. At first, we give a larger color margin and select as potential cluster members all morphologically classified early-type galaxies with 0:5 mag < i775 þ z850 < 1:2mag and J þ Ks > 1:45 mag, down to z850 ¼ 24 mag (the limiting magnitude of the Postman et al. [2005] morphological classification, which included all clusters in our sample at redshift unity). Our results in this paper are based on this morphological classification, and a detailed discussion of the uncertainties in this classification can be found in that work. This selected sample includes 38 galaxies within the ACS field. Our final colors were measured within galaxy effective radii (Re) to avoid biases due to galaxy internal gradients, following the approach in Blakeslee et al. (2003b) and van Dokkum et al. (1998, 2000). The Re values were derived with the program GALFIT ( Peng et al. 2002), constraining the Sersic index n 4 (as in Blakeslee et al. 2003b). To remove differential blurring effects (the point-spread function [ PSF] is $10% broader in the z850 band ), each galaxy image in both i775 and z850 was deconvolved using the CLEAN algorithm ( Hogbom 1974). The i775 þ ¨ z850 colors were measured on the deconvolved images within a circular aperture of radius equal to Re or 3 pixels, whichever is

þ z850 ¼ c0 × slope (z

850

þ 23):

Ï 1÷

The solid line in Figure 7 is the fit to the CMR for the elliptical galaxies, the black dotted line is the fit to the CMR for the S0s, and the dashed-dotted line is the fit to the full sample, within 20 from the cluster center (see discussion below). The dashed line is the fit to the full sample of early-type galaxies in RX J1252.9þ292 from Blakeslee et al. (2003b), scaled to this redshift with BC03 evolved stellar population models with solar metallicity and a formation age of 2.6 Gyr (since Blakeslee et al. 2003b obtains elliptical mean ages >2.6 Gyr). The dashed vertical line is the magnitude limit of the morphological classification z850 ¼ 24 mag. The results for different morphological samples are given in Table 1.


84

MEI ET AL.

Vol. 639

Fig. 3.-- Color images of the 20 CMR elliptical galaxies within 20 from the cluster center. Two of the CMR faint elliptical galaxies, which are satellites of brighter galaxies, are shown in the same postage stamp image as their bright companion.

A robust linear fit based on bisquare weights ( Tukey's biweight; Press et al. 1992) has been used to fit the CMR. Uncertainties on the parameters were estimated by bootstrapping on 10,000 simulations. The scatter around the fit was estimated from a biweight scale estimator ( Beers et al. 1990) that is insensitive to outliers, in the same set of bootstrap simulations. The internal color scatter (int ) was measured in two ways: (1) from the scatter around the fit, we have subtracted in quadrature the average uncertainty due to the galaxy color error and (2) we have calculated the internal scatter for which the 2 of the fit would be unity. Both methods give us internal scatters consistent to within a few 0.001 mag. All galaxies in this sample lie within 3 of the fit. The X-ray distribution appears to be very symmetric and largely confined within 1A5 from the cluster center. We calculated the CMR zero point and scatter within 10 (which corresponds to a

scale of %0.5 Mpc at this redshift), within 1A5 (%0.7 Mpc), and within 20 (%1 Mpc, the scale used for the analysis of RX J1252.9þ292). According to the results in Table 1, the internal color scatter increases when adding populations between 10 and 20 , especially for the S0 and S0/a populations, as was also observed in local samples (e.g., van Dokkum et al. 1998), with only a small increase in sample size. We therefore focus on the results obtained for color-selected galaxies within 10 from the cluster center (where 90% of the color-selected galaxies lie). The slope of the elliptical CMR (þ0:033 ô 0:015) is slightly steeper than the observed slope in RX J1252.9þ292 (þ0:020 ô 0:009) and in Coma when the latter are shifted to the observed colors at z $ 1:1, using nonevolving BC03 stellar population models, but still consistent within the uncertainties. We do not find a flatter (with respect to Coma) slope as in Stanford et al. (2002).

Fig. 4.-- Color images of the 11 CMR S0 galaxies within 20 from the cluster center.


No. 1, 2006

COLOR-MAGNITUDE RELATION FOR RDCS J0910+5422

85

Fig. 5.-- Color images of the three CMR S0/a galaxies within 20 from the cluster center.

However, the S0 sample shows a much shallower slope (0:005 ô 0:023) than the ellipticals, resulting in a much flatter slope for ellipticals and S0s together (þ0:024 ô 0:020). This can explain why a shallower slope was found in that work, in which elliptical and S0 galaxies were not separated. The spectroscopically selected elliptical plus S0 slope (þ0:010 ô 0:034) is also flattened by the S0 population, while the spectroscopically selected elliptical galaxies have a slope (þ0:021 ô 0:046) similar to that of RX J1252.9þ292 and Coma. All the difference in slope are, however, within the uncertainties and are statistically insignificant. Using BC03 stellar population models, as in Blakeslee et al. (2003b), we derive a constraint on the age of the stellar populations in the galaxies from galaxy colors and the scatter of the CMR (van Dokkum et al. 2001; Blakeslee et al. 2003b). Two simple models have been considered, and our conclusions depend on the model chosen. The first model is a single-burst model, in which galaxies form in single bursts at random times tf , between the age of the cluster and the recombination epoch. The second is a model with constant star formation in a range of time between t1 and t2, randomly chosen to be between the age of the cluster and the recombination epoch. Colors for 10,000 galaxies were simulated with their scatter around the CMR to be dependent on the burst age. In Figure 8, we show the simulated scatters as a function of burst age, with solar-, half-solar-, and twice-solar metallicity models. We assume solar metallicity in what follows. From the scatter (int ¼ 0:042 ô 0:010) in the colors of the galaxies classified as ellipticals, we obtain ages >2.1 Gyr (z > 2), with a mean luminosity ­ weighted age t > 3:31 Gyr (zf % 3:1), assuming the random single-burst model. From the constant ­ star formation model, we obtain ages >1.6 Gyr (z > 1:7), with a mean luminosity ­ weighted age t > 3:26 Gyr (zf % 3). This agrees with the conclusion (e.g., Blakeslee et al. 2003b; Holden et al. 2004; Lidman et al. 2004; De Lucia et al. 2004) that the elliptical population in clusters of galaxies formed at zf > 2:3 and has evolved mainly passively until z ¼ 1:1. In the U þ B rest frame (using BC03 stellar population models with solar metallicity and age equal to 4 Gyr), a scatter in

i775 þ z850 of 0:042 ô 0:010 corresponds to 0:050 ô 0:011. As pointed out in van Dokkum (2000) and Blakeslee et al. (2003b), CMR scatters vary little with redshift. The Blakeslee et al. (2003b) scatter for the elliptical CMR in RDCS 1252þ2927 (0:024 ô 0:008) corresponds to a scatter of 0:042 ô 0:014 in the U þ B rest frame, indistinguishable within the uncertainties from our result. The scatter in the CMR for galaxies classified as S0 (int ¼ 0:044 ô 0:020) is comparable to the one in the E CMR, but the galaxy colors are bluer and are not compatible with a population that is as old as the ellipticals. All the S0s lie below the elliptical CMR. In fact, between the elliptical and the S0 CMR fits there is a zero-point difference of 0:07 ô 0:02 mag, with the S0s being bluer than the ellipticals. One of the three S0/a galaxies has a color that is 0.07 mag redder than the elliptical CMR, and another has a color that is %0.15 bluer than the CMR relation for Es. The inclusion of the S0/a galaxies does not significantly change the fitted CMR for the S0s. When we consider all galaxies within 20 from the cluster center, the S0 and total early-type slopes are similar, while the color offset in the CMR is still present (0:05 ô 0:02 mag). The S0 population of this cluster has a very peculiar CMR with respect to the average cluster of galaxies. In fact, for several other studies the CMR of the S0 population has a similar zero point and on average a larger scatter with respect to the elliptical population (van Dokkum et al. 1998; Blakeslee et al. 2003b; Holden et al. 2004; De Lucia et al. 2004), quite different from our results. We discuss this peculiar behavior in detail in the rest of the paper, including an examination of orientation effects on the classification. In Figure 9, the near-IR ( Vega magnitudes) and i775 þ z850 (AB magnitudes) colors are shown compared with single-burst stellar population model predictions from BC03. The S0 colors are consistent with young (<2 Gyr), solar-metallicity populations or older (<3.5 Gyr), half-solar ­ metallicity populations. If the difference in E and S0 mean colors is mainly due to metallicity, then even if the two populations were formed at the same epoch, elliptical galaxies must have been able to retain more metals than the S0s; i.e., they were more massive at a given luminosity (given the observed mass-metallicity relation for earlytype galaxies; e.g., Tremonti et al. 2004, Bernardi et al. 2005, and references therein). This would imply higher mass-to-light ratios for the elliptical galaxies with respect to the S0s. However, the lack of strong evolution in the slope and scatter of the CMR from the present out to z $ 1 suggests that the CMR is mainly the result of a metallicity-mass (i.e., metallicity-magnitude) relation (e.g., Kodama & Arimoto 1997; Kauffman & Charlot 1998; Vazdekis et al. 2001; Bernardi et al. 2005). So, at a given magnitude we do not expect large metallicity variations.

Fig. 6.-- Color images of the four bright spiral galaxies with z850 brighter than 22.5 mag and i775 þ z850 between 0.5 and 1.3 mag, within 20 from the cluster center. These spiral galaxies have luminosities that are similar to those of the red sequence bright elliptical galaxies.


86

MEI ET AL.

Vol. 639

Fig. 7.-- Color-magnitude relations (CMRs) for RX J0910+5422. Orange circles and orange squares are, respectively, E and S0 candidates on the red sequence. Orange stars are S0/a candidates on the sequence. Smaller black circles, squares, and stars are used for the corresponding early-type galaxies that do not lie on the red sequence. Small triangles are late-type galaxies. Boxes are plotted around confirmed cluster members. Open circles are plotted around confirmed interlopers. The solid line is the fit to the ellipticals, the black dotted line is the fit to the S0s, and the dashed-dotted line is the fit to the entire sample within 20 . The green dotted line is fitted to the S0s within 10 . The dashed line is the RX J1252.9þ292 CMR scaled to the redshift and colors for RX J0910+5422. The dashed vertical line is the magnitude limit of the morphological classification z850 ¼ 24 mag. The S0 CMR is bluer than the elliptical CMR by 0:07 ô 0:02 mag.

If the offset is due to a different star formation history, a model with solar metallicity and an exponential decay of the star formation will reproduce the offset at a galaxy mean age of %3.5 Gyr. This age is consistent with the small scatter observed in the S0 CMR. We would then be observing galaxies that followed
TABLE 1 Color-Magnitude Rela tions c0 (mag) 0.99 1.00 1.02 1.02 0.95 0.95 0.99 0.99 1.02 1.02 0.95 0.96 0.99 0.99 1.01 1.02 0.96 0.96 ô ô ô ô ô ô ô ô ô ô ô ô ô ô ô ô ô ô 0.01 0.02 0.01 0.04 0.02 0.02 0.01 0.02 0.01 0.04 0.02 0.02 0.01 0.02 0.01 0.04 0.02 0.02 int (mag) 0.060 0.054 0.042 0.047 0.044 0.057 0.060 0.054 0.042 0.047 0.051 0.065 0.059 0.054 0.044 0.047 0.053 0.065 ô ô ô ô ô ô ô ô ô ô ô ô ô ô ô ô ô ô 0.008 0.009 0.011 0.022 0.02 0.015 0.008 0.009 0.011 0.022 0.018 0.015 0.008 0.009 0.010 0.022 0.015 0.013 Fig. 8.-- CMR scatter using BC03 single-burst ulation models for solar line), and twice-solar me it is possible to estimate as a function of galaxy age from a simulation obtained (top) and constant star formation (bottom) stellar popmetallicity (solid line), half-solar metallicity (dashed tallicity (dash-dotted line). From the measured scatter the mean age of the stellar bursts.

Sample E+S0+S0/aa ............... E+S0a,b ...................... Ea ............................... Ea,b............................. S0a ............................. S0+S0/aa .................... E+S0+S0/ac ............... E+S0b,c ...................... Ec ............................... Eb,c............................. S0c ............................. S0+S0/ac .................... E+S0+S0/ad ............... E+S0b,d ...................... Ed............................... Eb,/d............................ S0d ............................. S0+S0/ad ...................
a b c d

N 31 14 19 10 9 12 32 15 19 10 10 13 34 15 20 10 11 14

Slope þ0.030 þ0.010 þ0.033 þ0.020 0.005 þ0.007 þ0.032 þ0.021 þ0.033 þ0.020 þ0.012 þ0.015 þ0.036 þ0.022 þ0.032 þ0.021 þ0.022 þ0.024 ô ô ô ô ô ô ô ô ô ô ô ô ô ô ô ô ô ô 0.020 0.033 0.015 0.044 0.023 0.027 0.019 0.034 0.015 0.044 0.036 0.033 0.018 0.035 0.015 0.046 0.038 0.034

Within 10 . Only confirmed members. Within 1A5. Within 20 .


No. 1, 2006

COLOR-MAGNITUDE RELATION FOR RDCS J0910+5422

87

Fig. 10.--Left: Asymmetry (A) vs. concentration (C) parameters for elliptical (circles), S0 (squares), S0/a (stars), and spiral (triangles) galaxies in the i775 þ z850 color range between 0.5 and 1.3 mag for RX J0910+5422. Large diamonds identify red sequence (i775 þ z850 color between 0.8 and 1.1 mag) early-type galaxies. Right: Same parameters, but for for all galaxies in our low-redshift sample, using the same symbols for different galaxy types. E and S0 galaxies in RX J0910+5422 show parameters characteristic of an early-type population, A < 0:2 and C > 0:3, as are most of the low-redshift early-type galaxies. This means that S0s were not likely misclassified late-type galaxies. Fig. 9.--The i775 þ z850 (AB) vs. J þ K (Vega) color. Filled circles are elliptical galaxies, squares are S0 galaxies, and stars are S0/a galaxies. BC03 single-burst models are shown for half-solar metallicity (dashed line), solar metallicity (solid line), and twice-solar metallicity (dotted line). Ages run from 1.5 to 3.5 Gyr, in steps of 0.5 Gyr, as diamonds from left to right. The bluer S0 colors might be due to younger ages or lower metallicities.

5.1. Asymmetry and Concentration In Figure 10 (left), we compare the asymmetry A and the concentration C for ellipticals, S0s, and spirals with i775 þ z850 colors between 0.5 and 1.2 mag. The asymmetry parameter is obtained by subtracting a 180 rotated image from each original galaxy image, summing the residuals, and including a correction for the background. The concentration parameter is defined as in Abraham et al. (1996) as the sum of the galaxy flux within an aperture r0.3 divided the total flux, and r0.3 is calculated using the SExtractor fit to the galaxies at 1.5 above the background. The obtained semimajor and semiminor axes from this fit were multiplied by 0.3 to derive the r0.3 aperture (see also Homeier et al. 2005). Early-type and late-type galaxies lie on different regions in this A versus C plane. All our red sequence S0s have A < 0:3 and C > 0:3. All but one have A < 0:2 and C > 0:3. This is the same locus in the A-C plane that is occupied by most early-type galaxies in Abraham et al. (1996) and in our Postman et al. (2005) low-redshift sample ( Fig. 10, right). This last sample includes five strong-lensing clusters observed as part of our ACS GTO program ( Zw 1455+2232 [z ¼ 0:258], MS 1008þ 1224 [z ¼ 0:301], MS 1358+6245, Cl 0016+1654 [z ¼ 0:54], and MS J0454þ0300 [z ¼ 0:55]). Visual and automated classification for this cluster in the i775 band for all galaxies with i775 < 22:5 was performed by Postman et al. (2005). We conclude that the S0 population presents statistical parameters typical of an early-type population ( low asymmetry and high compactness). 5.2. Sersic Indices Figure 11 plots Sersic index n as a function of galaxy effective radius Re , both from our GALFIT modeling, for elliptical, S0, and spiral galaxies with i775 þ z850 colors between 0.5 and 1.3 mag. Red sequence (i775 þ z850 color between 0.8 and 1.1 mag) galaxies are shown by large diamonds. Most spiral galaxies have n < 2, and most early types have n > 2, as expected. However, the n-values do not permit us to discriminate between S0s and ellipticals in a unique way, unless they are combined with goodness-of-fit information for the Sersic model (e.g., the ``bumpiness'' parameter introduced by Blakeslee et al. 2005).

different star formation: single-burst, passive evolution for the elliptical galaxies and exponentially decaying star formation for the S0s. An exponential decay in the star formation is observed in field spiral galaxy samples ( Rowan-Robinson 2001). If this is the case, our S0 population might be the evolved product of an old spiral population that was already in place in this cluster when the ellipticals formed and then gradually lost available gas for star formation. If the E versus S0 color difference is mainly due to a difference in age, for a solar-metallicity and a single-burst BC03 template with age 4 Gyr, the color difference corresponds to an age difference of $1 Gyr. For clusters of galaxies at z > 1, the cluster members on the red sequence are only a part of all the progenitors of present-day early-type galaxies. Some of today's galaxy progenitors would have been bluer than the red sequence at these redshifts (van Dokkum & Franx 2001). In the S0 population of this cluster, we may be seeing the transitional progenitor population that in $1 Gyr will evolve onto the same red sequence as now occupied by the elliptical galaxies. Either of the latter two scenarios would be consistent with the Postman et al. (2005) observed deficit of the S0 population of our ACS cluster sample when compared to lower redshift samples, implying that part of the S0 population is still forming in clusters at redshifts around unity. 5. GALAXY SHAPE PROPERTIES Since the galaxies classified as S0 in RX J0910+5422 are found to be systematically bluer (with respect to the red sequence) than the S0 populations observed in previous studies, we wish to examine further the properties of these galaxies in terms of their shapes and light distributions and how they compare to the elliptical and spiral samples in this and other clusters. The shape parameters that we consider are concentration and asymmetry (Abraham et al. 1996; Conselice et al. 2004), Sersic index n, and galaxy axial ratios.


88

MEI ET AL.

Vol. 639

Fig. 11.--Sersic index n as a function of effective radius Re from GALFIT for elliptical (circles), S0 (squares), S0/a (stars), and spiral (triangles)galaxies with i775 þ z850 colors between 0.5 and 1.3 mag. Red sequence (i775 þ z850 color between 0.8 and 1.1 mag) galaxies are shown by large diamonds. E and S0 galaxies in RX J0910+5422 show a Sersic index 2. This means that S0s were not likely misclassified late-type galaxies.

Fig. 12.--Axial ratios vs. effective radii for elliptical (circles) and S0 galaxies (squares) in the i775 þ z850 color range between 0.5 and 1.3 mag. Large diamonds identify red sequence galaxies with i775 þ z850 color between 0.8 and 1.1 mag. Stars are S0/a galaxies, and triangles are spiral galaxies.

5.3. Axial Ratios
5.3.1. Axial Ratio Distribution

In Figure 12 we compare the apparent axial ratio from SExtractor versus effective radii for ellipticals and S0s with i775 þ z850 colors between 0.5 and 1.3 mag. The axial ratios have been verified by using ELLIPROF (the isophotal-fitting software that is used for surface brightness fluctuation analysis in Tonry et al. 1997 and Mei et al. 2005) on each galaxy image, after the cleaning procedure. As above, red sequence galaxies are shown by large diamonds. The red sequence elliptical galaxies and the bluer S0s have different axial ratio distributions, with all red sequence S0s showing axial ratios b /a P 0:7, and nearly all red sequence elliptical galaxies with b /a > 0:7. Assuming axisymmetric disks (oblate ellipsoids) viewed with random orientation and with a Gaussian-distributed intrinsic axial ratio (with mean equal to 0:3 ô 0:1 [extreme thin disk], 0:5 ô 0:1 [early-type galaxy], and 0:75 ô 0:1 [elliptical galaxy]; Jorgensen & Franx 1994), one would expect k40%, k60%, and k90%, respectively, of the S0s to have axial ratios above 0.7. Just 9% (1 out of 11) of the red sequence S0s are observed to have an axial ratio this large (or 22% for the full S0 sample in this cluster field; Fig. 13; top). The random probability that the S0 axial ratios would show such a low fraction with b /a > 0:7 is less than 1%. This is a very simple model, but it points out a lack of round S0s, indicating either that there is some orientation bias in the classification or that this class of objects is intrinsically prolate in shape. Jorgensen & Franx (1994) found a similar deficit of round S0s in the center of the Coma Cluster. They concluded that part of the face-on S0s were classified as elliptical galaxies. Fabricant et al. (2000) also found a deficit of round S0s in the cluster Cl 1358+62, at z ¼ 0:33. Their analysis shows that ellipticities and

the bulge ­ to ­ total light ratio do not allow us to distinguish elliptical from S0 galaxies. The other two ACS GTO clusters at z > 1 ( RX J1252.9þ292 and RX J0848+4452) do not show a similar lack of round S0s, as 90% of the red sequence elliptical galaxies (out of %70) and 47% of the S0s (out of %35) have b /a > 0:7 ( Fig. 13, bottom). This bias is also not observed in other clusters of our ACS Intermediate Redshift Cluster Survey (see Fig. 5 of Postman et al. 2005), for which more than 40% of the S0 galaxies have axial ratios above 0.7. The observed peculiarity of the RX J0910+5422 S0 axial ratio distribution might call into question our result above that the S0s have a significant color offset with respect to the ellipticals. For instance, if there is a bias in our color measurement procedure that causes elongated objects to have colors that are too blue, then the color offset found above may be artificial. Such a color bias might occur if the high inclination angles bias our Re measurements to higher values and if the S0s become progressively bluer at larger radii. We first examine this possibility, then proceed to discuss resolutions to the peculiarity of the S0 axial ratio distribution, with the aim of establishing whether a misclassification of face-on S0s as ellipticals would bias the measurement of the offset between the elliptical and S0 CMR zero points.
5.3.2. Internal Color Gradients

It is conceivable that our i775 þ z850 colors could be biased by aperture effects in the nearly edge-on S0s, for which the ( possibly) bluer outer disks might contribute more to the galaxy colors than in the rounder elliptical galaxy population. If this effect were severe enough, it might mimic the offset in color of the S0s and ellipticals found above. We test this possibility here. S0 and elliptical internal color profiles are shown in Figures 14 and 15. The gradients have been calculated with aperture photometry on the same images used to calculate our i775 þ z850 colors. Circles represent i775 þ z850 colors at different radii, and


No. 1, 2006

COLOR-MAGNITUDE RELATION FOR RDCS J0910+5422

89

Fig. 14.--Red sequence S0 color gradients as a function of distance from the galaxy center. The cross marks the color calculated at the effective radius.

Fig. 13.--Top: Histogram of axial ratios for the elliptical (red line) and S0 (blue line) galaxies. Dashed lines are for red sequence galaxies (i775 þ z850 between 0.8 and 1.1 mag). We note a lack of high axial ratio (low ellipticity) S0s in RX J0910+5422. Bottom: Elliptical (red ) and S0 (blue) distributions in RX J1252.9þ292 and RX J0848+4452. The median of their S0 distribution is consistent with that expected for nearly axisymmetric disks viewed at random inclination angles.

the cross represents the i775 þ z850 color at the effective radius used for the CMR. The S0 colors profiles do not show strong gradients. In particular, the i775 þ z850 colors calculated at the effective radii are not systematically bluer than colors determined at smaller radii. Most of the S0 galaxies have flat profiles; one has a blue inward gradient (ACS ID 1621; z850 ¼ 23:72 mag, i775 þ z850 ¼ 1 mag). In two galaxies (ACS ID 1393 and 3177) the colors in the central 0.15 arcsec2 are redder than the color at the effective radius. When compared with elliptical gradients, on average S0 colors do not appear biased toward higher effective radii and bluer colors than the ellipticals. We note that three elliptical galaxies show blue inward gradients (ACS ID 1753, 1519, and 3323).
5.3.3. Orientation or Intrinsic Shape: Axial Ratios versus i775 þ z850 Colors

Orientation biases are known to occur in the classification of elliptical galaxies and S0s in local galaxy samples. For instance, Rix & White (1990, 1992) showed, based on both isophotal and

dynamical modeling, that a large fraction of elliptical galaxies contain a disk component with at least $20% of the light, but which is hidden due to projection effects. Jorgensen & Franx (1994) found a strong deficit of round S0s in a sample of 171 galaxies in the central square degree of the nearby Coma Cluster and concluded that inclination angle played a large part in the classification of Es and S0s. Michard (1994) proposed that, except for the bright boxy elliptical galaxies without rotational support , earlytype galaxies constitute a single class of oblate rotators, with orientation being the main criterion for classification as either E or S0. On the other hand, van den Bergh (1994) explained the predominance of flattened S0s by invoking two distinct subpopulations: bright disky objects intermediate between elliptical and spiral galaxies, and a fainter population of prolate objects. We now address the question of whether the galaxies classified as S0 in RX J0910+5422 are preferentially flattened in shape because of an orientation bias in the classifications or intrinsically prolate shapes. If it is an orientation bias, then this could mean either that (1) face-on S0s have been misclassified as elliptical galaxies because their disks are not apparent or (2) edge-on spiral galaxies tend to be called S0s because the spiral structure is obscured. Either would result in a predominance of flattened S0s. However, in the former case, the misclassification of faceon S0s as ellipticals would tend to blur any color separation between the two classes, while in the latter case, a color offset might be introduced between the two classes because of contamination by bluer spiral galaxies. Because we do observe a color offset between galaxies classified as E and S0, with the S0s being bluer, it is possible to look for the ``missing'' population of face-on blue galaxies by examining ellipticity versus color. If a population of round blue objects is found, we can then determine the nature of the classification bias and whether it biases our color offset measurement. Histograms of the axial ratio distributions for ellipticals and S0s are shown in Figure 13 (top). We find that 60% of the earlytype red sequence galaxies have b /a > 0:7, but 95% of these low-ellipticity galaxies are classified as Es. However, if we split


90

MEI ET AL.

Vol. 639

Fig. 16.--Histogram types with i775 þ z850 < ratio (low ellipticity) S0 the total early-type or in

of the axial ratio of all early types (black) and early 0:99 mag (blue). Whereas there is a lack of high axial galaxies in RX J0910+5422, there is no similar lack in the blue early-type sample.

From our axial ratio simulations and statistics from our z % 1 ACS GT0 sample, the number of blue S0 galaxies might be between 25% and 40% higher than estimated in x 5.3.1. Moreover, the inclusion of a few blue S0s in the elliptical red sequence will have the effect of increasing very slightly the observed CMR
Fig. 15.--Red sequence elliptical color gradients as a function of distance from the galaxy center. The cross marks the color calculated at the effective radius.

galaxies instead by color, using i775 þ z850 < 0:99 mag as the separation point, then we find that 54% of all galaxies bluer than this separation have b /a > 0:7, while 43% of the red sequence galaxies bluer than this (0:8 mag < i775 þ z850 < 0:99 mag) have b /a > 0:7 ( Fig. 16). Thus, the deficit of round S0 galaxies (which are also significantly bluer than the mean of the E class) is not found when the early-type galaxies are split based purely on color. This suggests that some of the bluer round galaxies are the face-on counterparts of those classified as S0. Figure 17 shows axial ratios versus i775 þ z850 color residuals with respect to the total early-type galaxy CMR relation, for all galaxies in the RX J0910+5422 red sequence. Galaxy types are coded with different symbols; using the color residuals in this way takes out the effect of the magnitude dependence of the colors. There are five round blue elliptical galaxies, i.e., with b /a > 0:7 and on the blue side of the early-type CMR. If these are the face-on counterparts of the blue S0s, then the S0 axial ratio distribution becomes much more in line with expectations for a randomly oriented disk population (35% are rounder than b /a ¼ 0:7). Furthermore, we note that four other elliptical galaxies were classified as E/S0 in Postman et al. (2005); if these are also taken as face-on S0s, then the axial ratio distribution is in very close agreement with expectations. We conclude that the face-on S0s in RX J0910+5422 are classified as elliptical galaxies, just as in local early-type galaxy samples.

Fig. 17.--The i775 þ z850 color residuals (with respect to the mean of full early-type CMR fit) vs. axial ratios for red sequence (i775 þ z850 color between 0.8 and 1.1 mag) elliptical (circles), S0 (squares), S0/a (stars), and spiral (triangles) galaxies. While there is a lack of S0s with b /a > 0:7, we find five ellipticals and one spiral that have blue colors (negative CMR residuals) similar to most S0s, but round shapes (b /a > 0:7), unlike the classified S0s. These may be the face-on counterparts of the elongated blue S0 population (see text).


No. 1, 2006

COLOR-MAGNITUDE RELATION FOR RDCS J0910+5422

91

scatter for the elliptical galaxies. The uncertainty in the Postman et al. (2005) S0 fraction in RX J0910+5422 is significantly larger than 40%, and hence a systematic change by this amount for one cluster does not alter any of the results or conclusions in that paper. Throughout the paper, we continue to use elliptical and S0 classifications from Postman et al. (2005), but keep in mind the presence of this possible projection effect. 6. COLOR TRENDS, VELOCITY DISPERSION, AND MERGING ACTIVITY Stanford et al. (2002) have analyzed the X-ray and near-IR properties of this cluster. This system appears to be fairly relaxed, based on its regular X-ray profile; however, they find indications that the cluster is in an early phase of formation. In fact, the Chandra ACIS data show evidence for temperature structure, possibly due to an infalling group or mass streaming along a filament. The soft component of the X-ray emission (0.5 ­ 2 keV ) dominates the X-ray center of the cluster, while to the south there is a harder component (2 ­ 6 keV; see Fig. 6 from Stanford et al. 2002). The cluster does not have a central bright cluster galaxy ( BCG) or cD galaxy, and the X-ray emission center does not correspond to an optical grouping of galaxies; rather, a number of luminous confirmed cluster members are linearly distributed, at least as projected on the sky (as shown in Fig. 1). We do not observe any strong trend of the galaxy color i775 þ z850 with distance from the cluster X-ray center ( Fig. 18). Surprisingly, although not statistically significant, the bluer galaxies are concentrated toward the cluster center, instead of the outskirts, as in the other ACS intermediate-redshift cluster sample ( Demarco et al. 2005; Goto et al. 2005; Homeier et al. 2005; Postman et al. 2005). This tendency might support the hypothesis that we are observing subgroups of galaxies in an edge-on sheet, e.g., a group of bluer disk galaxies on a redder, older population of elliptical galaxies. There are 10 confirmed elliptical and 5 confirmed S0 members in the center of the cluster (R < 500 kpc). The average S0 redshift is 1:102 ô 0:002, and the average elliptical redshift is 1:105 ô 0:007 (the given uncertainties are standard deviations about the mean). This indicates small relative velocities (the two redshifts are indistinguishable given the errors) between the classes and is true regardless of the possible classification bias discussed above. Unfortunately, our spectroscopic sample does not permit us to track the cluster central structure in detail. The average relative velocity between the confirmed E and S0s of %500 km sþ1 (which also corresponds to the cluster velocity dispersion; see below) is fairly small. If merging of two distinct groups of galaxies is happening along the line of sight, we expect much higher velocity dispersions and/or relative velocities between the infalling S0s and the ellipticals. Stanford et al. (2002) also suggest that active galaxy-galaxy merging should be observed, based on the X-ray temperature structure. To investigate any ongoing dynamical activity, we calculated the cluster velocity dispersion and the merger rate. From the 25 spectroscopically confirmed members (all galaxy types included ) the line-of-sight rest-frame velocity dispersion is ¼ 675 ô 190 km sþ1, using the software ROSTAT from Beers et al. (1990). The available Chandra data give X-ray temperatures ranging from kT ¼ 7:2×2::2 keV (Stanford et al. 2002) þ1 4 to kT ¼ 6:6×1::7 keV ( Ettori et al. 2004). Wu et al. (1999; see þ1 3 also Rosati et al. 2002) give the relationship between kT and for relaxed clusters, which predicts that the velocity dispersion corresponding to the measured X-ray temperature should be %1000 km sþ1, considerably higher than we have found. Again,

Fig. 18.--Red sequence galaxy colors, corrected by the elliptical CMR, as a function of the distance from the cluster center, showing elliptical (circles), S0 (squares), S0/a (stars), and spiral (triangles) galaxies. Spectroscopically confirmed members are indicated by boxes. Known nonmembers are omitted from the plot. The continuous and dashed lines show the elliptical and the S0 CMR zero points, respectively. Most of the blue S0 lie close to the cluster center, not on the outskirts.

in the case of merging groups (along the line of sight) we would also have expected a higher velocity dispersion. 7. RED GALAXY PAIRS The quality of the ACS data allows us to discern merging activity among the cluster galaxies. If we assume that galaxy pairs with projected separations less than 20 hþ1 kpc are physically 70 associated, we observe 13 associated early-type galaxies, of which 9 galaxies lie on a filamentary structure about %100 kpc from the cluster center ( Fig. 1). As noted above, RX J0910+5422 lacks any cD galaxy near the center of the X-ray emission (see also Fig. 1), but rather has a filamentary group of galaxies around the X-ray center. The nine early-type interacting galaxies within this filamentary structure (at radius of $100 hþ1 kpc from the X-ray center) 70 include three unique pairs ( yellow arrows in Fig. 1), plus a galaxy triplet (the three components marked with red arrows in Fig. 1). Each of the three pairs consists of a bright elliptical galaxy with a smaller companion (all closer than 10 hþ1 kpc), 70 while the triplet is a large E galaxy with two smaller S0s (also closer than 10 hþ1 kpc; one of these two S0s is ACS ID 1621, the 70 S0 with an inward blue gradient). One of the pairs (the middle pair in the figure) and the two nearest galaxies in the triplet have essentially zero relative velocity and thus are likely merger candidates. Also two of the elliptical galaxies with blue inward gradients lie on the central filaments, and they are both small satellites of a larger galaxy. The other two pairs, which do not lie on the filamentary structure, are at 250 hþ1 kpc (two E galaxies 70 of similar size, both with weak O ii emission [Stanford et al. 2002]) and 300 hþ1 kpc (one S0/a and one E galaxy of similar 70 size) from the X-ray center and have relative velocities of 10,000 þ1 and 2000 km s , respectively. The presence of the low-velocity pairs is consistent with the low velocity dispersion and provides evidence for the ongoing hierarchical growth of the cluster (e.g., van Dokkum et al. 1999).


92

MEI ET AL.

Vol. 639

The pairing of the bright ellipticals with smaller S0s also argues against the view that we are observing a blue S0 ­ dominated group infalling into a red cluster elliptical population, but rather, a complex stellar population evolution within a filamentary structure. The red sequence S0/a confirmed members also lie within this filamentary structure. The observations of a significant number of red galaxy pairs in a cluster at z $ 1 is interesting in the context of the recent findings by van Dokkum (2005) of red galaxy interactions in %70% of 86 early-type galaxies in a selected sample of nearby red galaxies from the MUSYC ( Multiwavelength Survey by Yale-Chile; Gawiser et al. 2006) and the NOAO Deep WideField Survey. This work concluded that most of the elliptical galaxies in local samples were assembled by red galaxy-galaxy mergers, denominated dry mergers because they would involve gas-poor early-type galaxies. At higher redshift , Tran et al. (2005a) confirmed red galaxy mergers first observed by van Dokkum et al. (1999) in MS 1054þ03 at z ¼ 0:83. Tran et al. selected mergers as associated pairs with projected separation less than 10 hþ1 kpc 70 and relative line-of-sight velocities less than 165 km sþ1. As in RX J0910+5422, the red early-type galaxies involved in these mergers are among the brightest cluster members. Their results suggest that most early-type galaxies grew from passive red galaxygalaxy mergers. In our sample we observe a triplet and three red galaxy pairs with projected distances less than 10 hþ1 kpc. 70 Of those, the triplet and one galaxy pair show zero relative velocity. Of the two other pairs, composed of a bright and a fainter companion, redshifts are not available for the faint companions. 8. CLUSTER LUMINOSITY FUNCTION To obtain a deeper understanding of the RX J0910+5422 galaxy population, we constructed galaxy luminosity functions in the following way. We start with the original SExtractor catalog (described in x 3). All objects with magnitudes brighter than iF775W ¼ 21:1 mag were considered foreground objects. Nine of these bright objects are confirmed nonmembers. The remaining seven are objects that do not belong to the red sequence and whose sizes and luminosities are much larger than those of the confirmed members, in particular those of the bright red sequence galaxies; therefore, they are very unlikely to be at the cluster redshift. The contribution to the luminosity function from both foreground and background field galaxies ( hereafter, the field ) has been estimated from the galaxy counts in a reference field. The control region is taken from the GOODS-S (Great Observatories Origins Deep Survey ­ South; Giavalisco et al. 2004) ACS field, observed in the same filter as the cluster field. Pointlike objects were eliminated in a consistent way in the cluster and in the control field, by identification of the stellar locus in the diagnostic plot of the SExtractor parameters MAG _ AUTO versus FLUX _ RADIUS (the selected objects have FWHM equal to the PSF in the image). Cluster and control field luminosity functions were normalized to the cluster area. Both cluster and field counts were binned with a bin size of 0.5 mag. For each bin, the field counts were subtracted from the cluster counts, taking into account the extensive spectroscopic sample (more than 60% of the objects used for the luminosity function [ LF ] determination brighter than M ö have measured redshifts). Known interlopers were excluded from the analysis. The uncertainties in the cluster counts after subtraction of the field contribution are calculated by adding in quadrature Poissonian uncertainties. The luminosity functions are shown in Figure 19. The circles with errors are the total background-corrected cluster luminosity function. We do not include errors from cosmic

Fig. 19.--The z850 luminosity functions for all galaxies (black circles with errors), as well as red sequence early-type (red ), elliptical (blue), and S0+S0/a ( green) galaxies. The faint end of the red sequence luminosity function is dominated by S0s and S0/a galaxies. The solid line is the fit to the total luminosity function. We obtain M ö ¼ 22:6×0::6 mag and ¼ þ0:75 ô 0:4. Galaxies have þ0 7 been classified morphologically down to z850 ¼ 24 mag. The histogram of red sequence galaxies in RDS1252.9þ292 is shown as the dashed red line for comparison. The red arrows show the histogram values after background subtraction.

variance due to the choice of the background control region. The red sequence elliptical and spheroidal (S0 and S0/a) luminosity functions are shown, respectively, in blue and green. The red histogram is the luminosity function of all early-type galaxies with color 0:8 mag < i775 þ z850 < 1:1 mag, excluding confirmed nonmembers. The histogram of red sequence galaxies in RX J1252.9þ292 is shown as the dashed red line. The background contribution is very small for the early-type sample. The red arrows show the histogram values after background subtraction. The rest-frame B magnitudes are shown along the top of the plot, calculated from colors obtained from the BC03 stellar population model and templates (Sbc and Scd ) from Coleman et al. (1980). The solid black line is a Schechter function fit to the total cluster luminosity function. It is obtained by calculating the C-statistic (Cash 1979; a maximum likelihood statistic to fit data with Poissonian errors) on a grid in the M ö - plane for each combination of M ö and : first, the normalization (ö ) is calculated in order to reproduce the observed number of galaxies in the observed magnitude range, then the C-statistic is computed as C ¼ þ2ôibin ni ln mi þ mi þ ln ni !, where ni is the observed number of galaxies in the ith bin and mi is the number of galaxies predicted in that bin by the Schechter function with parameters M ö , , and ö . The combination M ö , that minimizes the C-statistic is taken as the best fit. If the C-statistic is defined as above, the 1, 2, and 3 confidence levels for M ö and can be estimated from àC ¼ 2:3, 6.17, and 11.8. We obtain M ö ¼ 22:6×0::6 mag and þ0 7 ¼ þ0:75 ô 0:4. Most of the faint-end population is composed of S0 and S0/a galaxies. The two brightest galaxies in the red sequence are S0s. With respect to RX J1252.9þ292, a bright population of red sequence elliptical galaxies is missing in RX J0910+5422. However, the large Poissonian errors on the bright end of the cluster population prevent us from definitively excluding the hypothesis that the two clusters could be drawn from the same parent population. Similarly, small number statistics do not permit us to study the luminosity functions of the different red and blue faint populations in this cluster.


No. 1, 2006

COLOR-MAGNITUDE RELATION FOR RDCS J0910+5422

93

9. DISCUSSION AND CONCLUSIONS In this paper we have studied the CMRs of galaxies in RX J0910+5422 to constrain their ages and formation histories. Our results show that the CMR for the elliptical galaxies is consistent both in slope and scatter with that of RX J1252.9þ292 ( Blakeslee et al. 2003b; Lidman et al. 2004) and recent results from Holden et al. (2004) and De Lucia et al. (2004), confirming that elliptical galaxies in galaxy clusters show a universal CMR consistent with an old passively evolving population even at z $ 1. From the CMR relation of the elliptical galaxies, we derive a mean luminosity ­ weighted age t > 3:3 Gyr (zf % 3). We find that the S0s in RX J0910+5422 define a colormagnitude sequence with a scatter similar to that found for the elliptical galaxies, but shifted bluer by 0:07 ô 0:02 mag. This is peculiar with respect to previous cluster studies, which more typically found that the S0s followed the same CMR as the ellipticals, but with somewhat larger scatter ( Bower et al. 1992; Ellis ´ et al. 1997; Stanford et al. 1997, 1998; Lopez-Cruz et al. 2004; van Dokkum et al. 2000, 2001; Blakeslee et al. 2003b; Holden et al. 2004). Only one earlier study, that of van Dokkum et al. (1998), found a significantly bluer S0 population. We examine this population of blue S0s in some detail, noting that there is a strong predominance of flattened systems with axial ratios b /a > 0:7, and conclude that the face-on members of the population have likely been classified as ellipticals. If so, the color offset between the two classes would be even more significant, and the true CMR scatter for the ellipticals would be slightly lower than we have estimated. This peculiarity is not observed in other clusters of our ACS Intermediate Redshift Cluster Survey, and its amplitude is smaller than the uncertainties adopted in Postman et al. (2005). If the observed color difference between the ellipticals and S0s is mainly due to metallicity at the same age, this would imply that the redder ellipticals were able to retain more metals than the S0s; i.e., they are more massive. However, current data suggest that the CMR is mainly the result of a metallicity-mass (i.e., metallicity-magnitude) relation (e.g., Kodama & Arimoto 1997; Kauffman & Charlot 1998; Vazdekis et al. 2001; Bernardi et al. 2005). This implies that we do not expect large metallicity variations at a given magnitude. If, instead, the offset is mainly due to age, then the implied age difference would be $1 Gyr for single-burst solar-metallicity BC03 models. It could also result from different star formation histories, with the S0s experiencing a more extended period of star formation. A model with solar metallicity and with an exponential decay of the star formation reproduces the offset at a galaxy mean age of %3.5 Gyr. The blue S0s may constitute a group infalling from the field onto a more evolved red cluster population, or they may be a transitional cluster population not yet evolved all the way onto the elliptical red sequence (van Dokkum & Franx 2001). Assuming passive evolution, they will reach this red sequence after about 1 Gyr. High fractions of faint blue late-type galaxies were observed in substructures infalling in a main cluster (e.g., Abraham et al. 1996; Tran et al. 2005b), and were proposed as the pro-

genitors of faint S0s in clusters. The view of this cluster as a structure still in formation is supported by X-ray observations of the cluster temperature structure (Stanford et al. 2002), the lack of a cD galaxy, and its filamentary structure, which suggests merging of substructures. However, we derive a small cluster velocity dispersion, unusual for merging substructures. Moreover, the blue S0s in this sample span the same luminosity range of the bright ellipticals and are distributed toward the center of the cluster, and some of the faintest ones are physically associated with brighter ellipticals belonging to the central filamentary structure. These elements would argue against the bluer S0s being a young group merging with an existing red cluster population and support the hypothesis that we are observing a transitional blue S0 population in a cluster core that is still evolving onto the elliptical red sequence. This result is also consistent with the deficit of S0s observed in our ACS cluster sample when compared to lower redshift samples, which implies that the S0 population is not yet in place but still forming in clusters at redshifts around unity ( Postman et al. 2005). Interestingly, we also observe in this cluster potential progenitors for bright S0 galaxies: four bright spiral galaxies (spectroscopically confirmed cluster members) with z850 brighter than 22.5 mag and i775 þ z850 between 0.5 and 1.3 mag. Red galaxy pairs are also observed. A triplet and three red galaxy pairs have projected distances less than 10 hþ1 kpc, and 70 of those the triplet and one pair show zero relative velocity. This would be the evidence of red galaxy mergers at z $ 1. Van Dokkum (2005) and Tran et al. (2005a) have observed mergers of red galaxies in a nearby elliptical sample and in MS 1054þ03 at z ¼ 0:83, respectively. They suggested a scenario in which most of the early-type galaxies were formed from passive red galaxy-galaxy mergers, called dry mergers, because they involve gas-poor early-type galaxies. Future papers will analyze the ages and masses of the cluster members using our optical spectroscopy along with newly obtained Spitzer IRAC imaging. A larger sample would be needed to draw firmer conclusions about the formation of S0s.

ACS was developed under NASA contract NAS 5-32865, and this research has been supported by NASA grant NAG5-7697 and by an equipment grant from Sun Microsystems, Inc. The Space Telescope Science Institute is operated by AURA, Inc., under NASA contract NAS5-26555. We are grateful to K. Anderson, J. McCann, S. Busching, A. Framarini, S. Barkhouser, and T. Allen for their invaluable contributions to the ACS project at Johns Hopkins University. We thank W. J. McCann for the use of the FITSCUT routine for our color images. S. M. thanks Tadayuki Kodama for useful discussions. S. A. S.'s work was performed under the auspices of the US Department of Energy, National Nuclear Security Administration, at the University of California, Lawrence Livermore National Laboratory, under contract W-7405-Eng-48.

REFERENCES Abraham, R. G., van den Bergh, S., Glazebrook, K., Ellis, R. S., Santiago, Bertin, E., & Arnouts, S. 1996, A&AS, 117, 393 B. X., Surma, P., & Griffiths, R. E. 1996, ApJS, 107, 1 Blakeslee, J. P., Anderson, K. R., Meurer, G. R., Benitez, N., & Magee, D. ´ Beers, T. C., Flynn, K., & Gebhardt, K. 1990, AJ, 100, 32 2003a, in ASP Conf. Ser. 295, Astronomical Data Analysis Software and Bell, E. F., Day, M. S., Rendleman, C. A., Woosley, S. E., & Zingale, M. 2004, Systems, ed. H. E. Payne et al. (San Francisco: ASP), 257 ApJ, 608, 883 Blakeslee, J. P., et al. 2003b, ApJ, 596, L143 Benitez, N., et al. 2004, ApJS, 150, 1 ´ ------. 2005, ApJ, submitted Bernardi, M., Sheth, R. K., Nichol, R. C., Schneider, D. P., & Brinkmann, J. Bower, R. G., Lucey, J. R., & Ellis, R. S. 1992, MNRAS, 254, 589 2005, AJ, 129, 61 Bruzual A., G., & Charlot, S. 2003, MNRAS, 344, 1000 ( BC03)


94

MEI ET AL.
Postman, M., et al. 2005, ApJ, 623, 721 Press, W. H., Teukolsky, S. A., Vetterling, W. T., & Flannery, B. P. 1992, Numerical Recipes in C (2nd. ed.; Cambridge: Cambridge Univ. Press) Rix, H., & White, S. D. M. 1990, ApJ, 362, 52 ------. 1992, MNRAS, 254, 389 Rosati, P., Borgani, S., & Norman, C. 2002, ARA&A, 40, 539 Rosati, P., della Ceca, R., Norman, C., & Giacconi, R. 1998, ApJ, 492, L21 Rowan-Robinson, M. 2001, ApJ, 549, 745 Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ, 500, 525 Sirianni, M., et al. 2005, PASP, 117, 1049 Spergel, D. N., et al. 2003, ApJS, 148, 175 Stanford, S. A., Eisenhardt, P. R., & Dickinson, M. 1998, ApJ, 492, 461 Stanford, S. A., Elston, R., Eisenhardt, P. R., Spinrad, H., Stern, D., & Dey, A. 1997, AJ, 114, 2232 Stanford, S. A., Holden, B., Rosati, P., Eisenhardt, P. R., Stern, D., Squires, G., & Spinrad, H. 2002, AJ, 123, 619 Tonry, J. L., Blakeslee, J. P., Ajhar, E. A., & Dressler, A. 1997, ApJ, 475, 399 Tran, K.-V., van Dokkum, P. G., Franx, M., Illingworth, G. D., Kelson, D., & Schreiber-Foerster, N. M. 2005a, ApJ, 627, L25 Tran, K.-V., van Dokkum, P. G., Illingworth, G. D., Kelson, D., Gonzalez, A., & Franx, M. 2005b, ApJ, 619, 134 Tremonti, C. A., et al. 2004, ApJ, 613, 898 van den Bergh, S. 1994, AJ, 107, 153 van Dokkum, P. G. 2005, AJ, 130, 2647 van Dokkum, P. G., & Franx, M. 2001, ApJ, 553, 90 van Dokkum, P. G., Franx, M., Fabricant, D., Illingworth, G. D., & Kelson, D. D. 2000, ApJ, 541, 95 van Dokkum, P. G., Franx, M., Fabricant, D., Kelson, D. D., & Illingworth, G. D. 1999, ApJ, 520, L95 van Dokkum, P. G., Franx, M., Kelson, D. D., Illingworth, G. D., Fisher, D., & Fabricant, D. 1998, ApJ, 500, 714 van Dokkum, P. G., Stanford, S. A., Holden, B. P., Eisenhardt, P. R., Dickinson, M., & Elston, R. 2001, ApJ, 552, L101 Vazdekis, A., Kuntschner, H., Davies, R. L., Arimoto, N., Nakamura, O., & Peletier, R. 2001, ApJ, 551, L127 Wu, X.-P., Xue, Y.-J., & Fang, L.-Z. 1999, ApJ, 524, 22

Cash, W. 1979, ApJ, 228, 939 Coleman, G. D., Wu, C.-C., & Weedman, D. W. 1980, ApJS, 43, 393 Conselice, C. J., et al. 2004, ApJ, 600, L139 De Lucia, G., et al. 2004, ApJ, 610, L77 Demarco, R., et al. 2005, A&A, 432, 381 Ellis, R. S., Smail, I., Dressler, A., Couch, W. J., Oemler, A., Jr., Butcher, H., & Sharples, R. M. 1997, ApJ, 483, 582 Ettori, S., et al. 2004, MNRAS, 354, 111 Fabricant, D., Franx, M., & van Dokkum, P. 2000, ApJ, 539, 577 Ford, H. C., et al. 2003, Proc. SPIE, 4854, 81 ------. 2004, preprint (astro-ph /0408165) Gawiser, E., et al. 2006, ApJS, 162, 1 Giavalisco, M., et al. 2004, ApJ, 600, L93 Goto, T., et al. 2005, ApJ, 621, 188 Hogbom, J. A. 1974, A&AS, 15, 417 ¨ Hogg, D. W., et al. 2004, ApJ, 601, L29 Holden, B. P., Stanford, S. A., Eisenhardt, P., & Dickinson, M. 2004, AJ, 127, 2484 Holden, B. P., et al. 2005a, ApJ, 620, L83 ------. 2005b, ApJ, 626, 809 Homeier, N. L., et al. 2005, ApJ, 621, 651 Jorgensen, I., & Franx, M. 1994, ApJ, 433, 553 Kauffman, G., & Charlot, S. 1998, MNRAS, 294, 705 Kodama, T., & Arimoto, N. 1997, A&A, 320, 41 Lidman, C., Rosati, P., Demarco, R., Nonino, M., Mainieri, V., Stanford, S. A., & Toft, S. 2004, A&A, 416, 829 ´ Lopez-Cruz, O., Barkhouse, W. A., & Yee, H. K. C. 2004, ApJ, 614, 679 Marleau, F. R., & Simard, L. 1998, ApJ, 507, 585 Massey, P., & Gronwall, C. 1990, ApJ, 358, 344 McIntosh, D. H., Zabludoff, A. I., Rix, H.-W., & Caldwell, N. 2005, ApJ, 619, 193 Mei, S., et al. 2005, ApJ, 625, 121 Michard, R. 1994, A&A, 288, 401 Mullis, C. R., Rosati, P., Lamer, G., Bohringer, H., Schwope, A., Schuecker, P., & Fassbender, R. 2005, ApJ, 623, L85 Oke, J. B., et al. 1995, PASP, 107, 375 Peng, C. Y., Ho, L. C., Impey, C. D., & Rix, H. 2002, AJ, 124, 266