Mass-detection of a matter concentration projected near the cluster Abell 1942 Dark clump o

A&A manuscript no. (will be inserted by hand later) Your thesaurus codes are: 02 (12.03.4; 12.04.1; 12.07.1; 12.12.1)


Mass-detection of a matter concentration projected near the cluster Abell 1942: Dark clump or high-redshift cluster??
Thomas Erben1 , Ludovic van Waerbeke2,1 , Yannick Mellier3,6, Peter Schneider1 , Jean-Charles Cuillandre4 , Francisco Javier Castander5 Mireille Dantel-Fort6

arXiv:astro-ph/9907134v1 11 Jul 1999

Max-Planck-Institut f¨r Astrophysik, P.O. Box 1523, D-85740 Garching, Germany u CITA, University of Toronto, 60 St. George Street, Toronto, Ontario, Canada M5S 1A7, Canada 3 Institut d’Astrophysique de Paris, 98bis Boulevard Arago, F-75014 Paris, France 4 CFHT Corporation, P.O. Box 1597, Kamuela, Hawaii 96743, USA 5 OMP, 14 Av. Edouard Belin, 31400 Toulouse, France 6 Observatoire de Paris, DEMIRM, 77 avenue Denfert Rochereau, F-75014 Paris, France


February 1, 2008

Abstract. A weak-lensing analysis of a wide-?eld V -band image centered on the cluster Abell 1942 has uncovered the presence of a mass concentration projected ? 7 arcminutes South of the cluster center. From an additional wide-?eld image, taken with a di?erent camera in the Iband, the presence of this mass concentration is con?rmed. A statistical analysis, using the aperture mass technique, shows that the probability of ?nding such a mass concentration from a random alignment of background galaxies is 10?6 and 4 × 10?4 for the V - and I-band image, respectively. No obvious strong concentration of bright galaxies is seen at the position of the mass concentration, but a slight galaxy number overdensity is present about 1′ away from its center. Archival ROSAT-HRI data show the presence of a weak extended X-ray source near to the mass concentration, but also displaced by about 1′ from its center, and very close to the center of the slight galaxy number concentration. From the spatial dependence of the tangential alignment around the center of the mass concentration, a rough mass estimate can be obtained which depends strongly on the assumed redshift of the lens and the redshift distribution of the background galaxies. A lower bound on the mass inside a sphere of radius 0.5h?1 Mpc is 1 × 1014 h?1 M⊙ , considerably higher than crude mass estiSend o?print requests to: erben@mpa-garching.mpg.de ? based on observations with the Canada-France-Hawaii Telescope (CFHT) operated by the National Research Council of Canada (CNRC), the Institut des Sciences de l’Univers (INSU) of the Centre National de la Recherche Scienti?que (CNRS) and the University of Hawaii (UH) and on data obtained through the NASA/GSFC HEASARC Online archive.

mates based on the X-ray data; shifting the lens to higher redshift increases both the lensing and X-ray mass estimates, but does not resolve the mass discrepancy. Concerning the nature of the mass concentration, no ?rm conclusion can be obtained from the available data. If it were a high-redshift cluster, the weak X-ray ?ux would indicate that it had an untypically low X-ray luminosity for its mass; if the X-ray emission were physically unrelated to the mass concentration, e.g., coming from a relatively low-redshift group which shows up in the number density of galaxies, this conclusion would be even stronger. Since the search for massive halos by weak lensing enables us for the ?rst time to select halos based on their mass properties only, it is possible that new types of objects can be detected, e.g., halos with very little X-ray and/or optical luminosity, should they exist. The mass concentration in the ?eld of A1942 may be the ?rst example of such a halo. Possibilities to establish the nature of this mass concentration with future observations are brie?y discussed. Key words: Cosmology: dark matter, gravitational lenses

1. Introduction The abundance of clusters of galaxies as a function of mass and redshift provides one of the most sensitive cosmological tests (e.g., Richstone et al. 1992; Bartelmann et al. 1993). In particular, in a high-density Universe, the abundance of massive clusters strongly decreases with redshift,


Dark mass concentration near Abell 1942

so that the existence of a few massive high-redshift clusters can in principle rule out an ?0 = 1 model (e.g., Eke et al. 1996; Bahcall & Fan 1998). The reliability of the test depends on the detection ef?ciency and selection e?ects in existing samples of clusters whose understanding may be critical. Currently, clusters are selected either by their optical appearance as overdensities of galaxies projected onto the sky and/or in colormagnitude diagrams, or by their X-ray emission. Both selection techniques may bias the resulting sample towards high-luminosity objects, i.e. they would under-represent clusters with high mass-to-(optical or X-ray) light ratio. Furthermore, the observed properties have to be related to their mass in order to compare the observed abundance to cosmological predictions. The usual procedures consist in assuming a dynamical and/or hydrostatic equilibrium state as well as the geometry of the mass distribution, which in general may be questionable and fairly poorly justi?ed from a theoretical point of view. Indeed, whereas cosmological theories have made great progress in their ability to predict the distribution of dark matter in the Universe, either analytically or numerically (e.g., Lacey & Cole 1993; Jenkins et al. 1998), the luminous properties of matter are much more di?cult to model. For example, to relate the X-ray data of a cluster to its mass, a redshift-dependent luminosity-temperature relation needs to be employed (see Borgani et al. 1999 and references therein), in the absence of a detailed understanding of the physics in the intra-cluster gas. It would therefore be of considerable interest to be able to de?ne a sample of ‘clusters’ – or more precisely, dark matter halos – which can be directly compared with the predictions coming from N-body simulations. Weak gravitational lensing o?ers an attractive possibility to detect dark matter halos by their mass properties only. A mass concentration produces a tidal gravitational ?eld which distorts the light bundles from background sources. Owing to their assumed random intrinsic orientation, this tidal ?eld can be detected statistically as a coherent tangential alignment of galaxy images around the mass concentration. A method to quantify this tangential alignment was originally introduced by Kaiser et al. (1994) to obtain lower bounds on cluster masses, and later generalized and proposed as a tool for the search of dark matter halos (Schneider 1996). This so-called aperture mass method can be applied to blank ?eld imaging surveys to detect peaks in the projected density ?eld. Combining halo abundance predictions from Press & Schechter (1974) theory with the universal density pro?le found in N-body simulations (Navarro et al. 1997), Kruse & Schneider (1999) estimated the number density of dark matter halos detectable with this method (with a signal-to-noise threshold of 5) to be of order 10 deg?2 , for a number density of 30 galaxies/arcmin2, and depending on the cosmological model. These predictions were con?rmed (Reblinsky et al.

1999) in ray-tracing simulations (Jain et al. 1999) through numerically-generated cosmic density ?elds. In this paper, we report the ?rst detection of a dark matter halo not obviously associated with light, using the above-mentioned weak lensing technique. Using a 14′ ×14′ deep V -band image, obtained with MOCAM at CFHT, we aimed to investigate the projected mass pro?le of the cluster Abell 1942 on which the image is centered. We found a highly signi?cant peak in the reconstructed mass map, in addition to that corresponding to the cluster itself. This second peak, located about 7′ South of the cluster center, shows up in the alignment statistics of background galaxy images with a signi?cance > 99.99%, as obtained from Monte-Carlo simulations which randomized the orientation of these background galaxies. An additional deep I-band image, taken with the UH8K at CFHT, con?rms the presence of the mass peak. No obvious large overdensity of galaxies is seen at this location, implying either a mass concentration with low light-to-mass ratio, or a halo at substantially higher redshift than A1942 itself. Finally, an analysis of an archival ROSAT/HRI image of A1942 shows, in addition to the emission from the cluster, a 3.2σ detection of a source with position close to the peak in the projected mass maps; though this weak detection would be of no signi?cance by itself, the positional coincidence with the ‘dark’ clump suggests that it corresponds to the same halo, and that it may be due to a high-redshift (z > 0.5) cluster. ? The outline of the paper is as follows: in Sect. 2 we describe the observations and data reduction techniques, as well as the measurement of galaxy ellipticities which we employed. The aperture mass statistics is brie?y described in Sect. 3.1 and applied to the optical data sets, together with a determination of the peak detection signi?cance. Properties of the mass concentration as derived from the optical data sets and the X-ray data are discussed in Sects. 3.2 and 3.3, respectively, and a discussion of our ?ndings is provided in Sect. 4. We shall concentrate in this paper mainly on the ‘dark’ clump; an analysis of the mass pro?le of the cluster A1942 and the reliability of mass reconstruction will be published elsewhere (van Waerbeke et al., in preparation) 2. Summary of optical observations and image processing The V - and I-band observations were obtained at the prime focus of CFHT with the MOCAM and the UH8K cameras, respectively. Both observing procedures were similar, with elementary exposure time of 1800 seconds each in V and 1200 seconds in I. A small shift of 10 arcseconds between pointings was applied in order to remove cosmic rays and to prepare a super-?at?eld. The V -band images were obtained during an observing run in dark time of June 1995 with the 4K × 4K mosaic camera MOCAM (Cuillandre et al 1997). Each individual

Dark mass concentration near Abell 1942


Fig. 1. The geometry of the optical data used in this paper. The left-hand side shows the area of the V -band MOCAM ?eld ′ ′ (square) and the I-band UH8K-chip3 data (rectangle). The framed regions are 3.3 × 3.3 cutouts around the cluster center of A1942 and around our ‘dark clump’ candidate. These regions are zoomed in on the right-hand side. The ‘dark clump’ region is centered around α(J2000)=14h 38m 22.59s ; δ(J2000)=03? 32′ 32.22′′ .
′′ chip is a 2K × 2K LORAL CCD, with 0. 206 per pixel, so ′ ′ the total ?eld-of-view is 14 × 14 . Nine images have been re-centered and co-added, to produce a ?nal frame with a total exposure time of 4h30min. The seeing of the coadded ′′ image is 0. 74. The I-band images were obtained with the 8K × 8K mosaic camera UH8K (Luppino, Bredthauer & Geary, 1994). Each individual chip is a 2K × 4K LORAL CCD, ′′ also with 0. 206 per pixel, giving a ?eld-of-view of 28′ ×28′ . The ?nal centered coadded image resulting from 9 subimages has a total exposure time of 3h and a seeing of ′′ 0. 67. The V - and I-band images have been processed in a similar manner, using standard IRAF procedures and some more speci?c ones developed at CFHT and at the TERAPIX1 data center for large-?eld CCD cameras. None of these procedures had innovative algorithms, so there is basically no di?erence in the pre-processing and processing of the MOCAM and UH8K images. For the present paper, we use only Chip 3 of the UH8K I-band image which

contains the cluster A1942, and the additional mass concentration discussed further below. Fig. 1 shows the CCD images from both ?elds and their relative geometry. A ?rst object detection and the photometry have been performed with SExtractor2.0.17 (Bertin & Arnouts 1996). The MOCAM ?eld has been calibrated using the photometric standard stars of the Landolt ?eld SA110 (Landolt 1992), and the UH8K ?eld was calibrated using the Landolt ?elds SA104 and SA110. The completeness limit is V = 26 and I = 24.5. The lensing analysis was done with the imcat software, based on the method for analysing weak shear data by Kaiser, Squires & Broadhurst (1995), with modi?cations described in Luppino & Kaiser (1997) and Hoekstra et al. (1998; hereafter HFKS98). This method is based on calculations of weighted moments of the light distribution. Imcat is speci?cally designed for the measurement of ellipticities of faint and small galaxy images, and their correction for the smearing of images by a PSF, and for any anisotropy of the PSF which could mimic a shear signal.



Dark mass concentration near Abell 1942

Fig. 2. For all objects detected with the imcat method from the MOCAM frame (left panel) and the UH8K chip3 (right panel), the magnitude is plotted as a function of half-light radius rh , measured in pixels. Objects containing saturated pixels have been removed from the plots. The magnitudes are in an arbitrary system. In both cases we can clearly identify a prominent sequence of stellar objects at about rh = 2.2 for MOCAM, and rh = 1.75 for UH8K-chip3.

These corrections are employed by the relation χ = χ0 + P γ γ + P sm p , (1)

where χ is the observed image ellipticity (de?ned as in, e.g., Schneider & Seitz 1995), χ0 is the ellipticity of the unlensed source smeared by the isotropic part of the PSF, P γ is the response tensor of the image ellipticity to a shear, and P sm is the response tensor to an anisotropic part of the PSF, characterized by p. These tensors are calculated for each galaxy image individually. Since the expectation value of χ0 in (1) is zero, one obtains an unbiased estimate of the shear through γ = (P γ )?1 [χ ? P sm p] . ? (2)

second-order polynomial (see also Fig. 4). With these polynomials we performed the anisotropy correction in (1). We follow the prescription of HFKS98 for the calculation of P γ , and used the full tensors, not just their trace-part, in (2). The current version of imcat does not give information about the quality of objects; for this we produced a SExtractor (version 2.0.20) catalog containing all objects that had at least six connected pixels with 1-σ above the local sky background. From this catalog we sorted out all objects with potential problems for shape estimation (like being deblended with another object or having a close neighbour). This included all objects with FLAGS≥ 2 (internal SExtractor ?ag). The remaining catalog was matched with the corresponding imcat catalog, using a maximum positional di?erence of three pixels, and keeping only those objects for which the detection signal-to-noise of imcat was ≥ 7. This procedure left us with 4190 objects (V > 22.0) for the MOCAM and 1708 objects (I > 21.0) for the I-band chip3. With these ?nal catalogs all subsequent analysis

(? is in reality an estimate for the reduced shear γ/(1 ? κ) γ which reduces to the shear if κ ? 1.) The PSF anisotropy in our images is fairly small and regular over the ?eld. We selected bright, unsaturated stars from a size vs. magnitude plot (see Fig. 2) and determined their ellipticities. As Fig. 3 shows, the stellar ellipticity changes very smoothly over the ?elds so that its behaviour can be easily ?t with a

Dark mass concentration near Abell 1942


Fig. 3. The ellipticity ?elds for stars for the V -band MOCAM ?eld (left panel) and the I-band UH8K-chip3 containing the cluster. Both ?elds show a smooth variation and can be easily modelled by a low-order polynomial. The maximal ellipticity is about 5% for the MOCAM and 8% for the UH8K.

was done. We note that we did not cross-correlate the MOCAM and UH8K catalogs; hence, the galaxies taken from both catalogs will be di?erent even in the region of overlap. Due to the di?erent waveband used for object selection, the redshift distribution of the background galaxies selected on the MOCAM and the UH8K-chip3 frame can be di?erent. 3. Analysis of the ‘dark’ clump 3.1. Weak lensing analysis From the image ellipticities of ‘background’ galaxies, we have ?rst reconstructed the two-dimensional mass map of the cluster ?eld from the MOCAM data, using the maximum-likelihood method described in Bartelmann et al. (1996) and independently, the method described in Seitz & Schneider (1998). The resulting mass maps are very similar, and we show the former of these only. In the left panel of Fig. 5, we show the resulting mass map with the (mass-sheet degeneracy) transformation parameter λ chosen such that κ = 0 (see Schneider & Seitz

1995), together with contours of the smoothed number density of bright galaxies. In general, this number density correlates quite well with the reconstructed surface mass density. As can be seen, a prominent mass peak shows up centered right on the brightest cluster galaxy. In addition to this mass peak, several other peaks are present in the mass map. Such peaks may partly be due to noise coming from the intrinsic image ellipticities and, to a lesser degree, to errors in the determination of image ellipticities. In order to test the statistical signi?cance of the mass peaks, we used the aperture mass method (Schneider 1996). Let U (?) be a ?lter function which vanishes for ? ≥ θ, θ and which has zero mean, 0 d? ? U (?) = 0. Then we de?ne the aperture mass Map (?) at position ? as Map (?) = d2 ?′ κ(? + ?′ ) U (|?′ |) . (3)

|? |≤θ

Hence, Map (?) is a ?ltered version of the density ?eld κ; it is invariant with respect to adding a homogeneous mass sheet or a linear density ?eld, and is positive if centered


Dark mass concentration near Abell 1942

Fig. 4. The left panels show the raw imcat ellipticities from bright, unsaturated foreground stars in our ?elds (upper panels: MOCAM ?eld; lower panels: UH8K chip). The right panels show the ellipticities after they have been corrected with a second-order polynomial as described in the text. The rms of the ellipticities after correction is typically 0.015.

on a mass peak with size comparable to the ?lter scale θ. The nice feature about this aperture mass is that it can be expressed directly in terms of the shear, as Map (?) =
|? |≤θ

d2 ?′ γt (?′ ; ?) Q(|?′ |)


(Kaiser et al. 1994; Schneider 1996), where the ?lter func? tion Q(?) = 2??2 0 d?′ ?′ U (?′ ) ? U (?) is determined in terms of U (?), and vanishes for ? ≥ θ. The tangential shear γt (?′ ; ?) at relative position ?′ with respect to ? is de?ned as γt (?′ ; ?) = ?Re[γ(? + ?′ ) e?2i? ] ,


where ?′ is the polar angle of the vector ?′ . In the case of weak lensing (κ ? 1), the observed image ellipticities γ ? from (2) are an unbiased estimator of the local shear, and so the aperture mass can be obtained by summing over image ellipticities as
′ Map (?) =

where the sum extends over all N galaxy images with positions θi which are located within θ of ?, and the tangential component γti (?) of the image ellipticity relative ? to the position ? is de?ned in analogy to γt . In general, ′ Map (?) is not an unbiased estimator of Map (?) since the expectation value of γ is the reduced shear, not the shear ? itself. However, unless the aperture includes a strong mass ′ clump where κ is not small compared to unity, Map will approximate Map closely. But even if the weak-lensing approximation breaks down for part of the aperture, one can ′ consider the quantity Map (?) in its own right, representing the tangential alignment of galaxy images with respect to the point ?. This interpretation also remains valid if the aperture is centered on a position which is less than θ away from the boundary of the data ?eld, so that part of the aperture is located outside the data ?eld, in which case ′ Map (?) will not be a reliable estimator of Map (?). In order to determine the signi?cance of the peaks in ′ the mass map shown in Fig. 5, we have calculated Map on a grid of points ? over the data ?eld, for four values of the ?lter scale θ. Then, we have randomized the posi-

πθ2 N

γti (?) Q(|θ i ? ?|) , ?


Dark mass concentration near Abell 1942


Fig. 5. The ?gure shows mass reconstructions and galaxy number density from the MOCAM ?eld (left panel) and the UH8K-chip3 (right panel). The white contours show κ = 0.03, 0.05, 0.07, 0.1, 0.12, 0.15, 0.17 and 0.2. For the reconstruction the shear was smoothed with a Gaussian of σ = 40′′ width. The black contours show the smoothed galaxy distribution from all galaxies brighter than V = 21.0 and I = 20.0 (the smoothing kernel here was a Gaussian with σ = 20′′ ).
′ tion angles of all galaxy images, and calculated Map on the same grid for these randomized realizations. This has been repeated Nrand times. Finally, at each grid point the ′ fraction ν of randomizations where Map is larger than the measured value from the actual data has been obtained; this fraction (which we shall call ‘error level’ in the follow′ ing) is the probability of ?nding a value of Map at that gridpoint for randomly oriented galaxy images, but with the same positions and ellipticities as the observed galaxies.

Fig. 6 displays the contours of constant ν, for di?erent ?lter radii, varying from 80′′ to 200′′ . As can be seen, the cluster center shows up prominently in the ν-map on all scales. In addition, two highly signi?cant peaks show up, one at the upper right corner, the other ? 7′ South of the cluster center, close to the edge of the MOCAM ?eld. We have veri?ed the robustness of this Southern peak by using SExtractor ellipticities instead of those from imcat, and found both the cluster components and the Southern peak also with that catalog (although it should be much less suited for weak lensing techniques). After these ?ndings, we obtained the UH8K I-band image, on which both the cluster and the Southern mass peak are located on Chip 3. The mass reconstruction from galaxy images on Chip 3 are shown in the right panel of Fig. 5, from which we see that the cluster and this Southern mass peak also show up. Repeating the aperture mass

statistics for Chip 3, we obtain the error levels as shown in Fig. 7; again, this Southern peak shows up at very high signi?cance. Whereas the third peak in the signi?cance maps (considering the two larger ?lter scales) from Chip 3, about halfway between cluster and the Southern component and slightly to the West, is also quite signi?cant and is also seen in the corresponding MOCAM map (and most likely also corresponds to a mass peak, though a highly elongated one for which the aperture mass is less sensitive), we shall concentrate on the Southern peak, which we call, for lack of a better name, the ‘dark clump’. In fact, as can be seen from Figures 1 and 5, this mass peak does not seem to be associated with any concentration of brighter galaxies. This could mean two things: either, the mass concentration is in fact associated with little light, or is at much higher redshift than A1942 itself. Concentrating on the location of the dark clump, we ′ determined the probability distribution p0 (Map ) for the ′ 6 value of Map , obtained from 2 × 10 randomizations of the galaxy orientations within 160′′ of the dark clump. This probability distribution is shown as the solid (from MOCAM) and dashed (from Chip 3) curve on the left of Fig. 8. These two distributions are very well approximated by a Gaussian, as expected from the central limit ′ theorem. The value of Map at the dark clump is 0.0395 for MOCAM, and 0.0283 for Chip 3. The fact that these two values are di?erent is not problematic, since for Chip


Dark mass concentration near Abell 1942

′ Fig. 6. The four panels show the signi?cance ν (see text) of the Map maps of the MOCAM ?eld. We chose Nrand = 5000, the black contours mark areas with ν = 1, 10, 30/5000 and the white contours ν = 100, 180, 260/5000. The ?lter scales are 80′′ (upper left panel), 120′′ (upper right panel), 160′′ (lower left panel) and 200′′ (lower right panel). For the larger scales the cluster components and the dark clump are detected with a very high signi?cance

3, the whole aperture ?ts inside the data ?eld, whereas it is partially outside for MOCAM; hence, the two values of ′ Map measure a di?erent tangential alignment. Also, since the two data sets use galaxies selected in a di?erent waveband, their redshift distribution can be di?erent, yielding di?erent values of the resulting lens strength. The probability that a randomization of image orientations yields a ′ value of Map larger than the observed one is ? 10?6 for the MOCAM ?eld, and 4.2 × 10?4 for Chip 3. Next we investigate whether the highly signi?cant ′ value of Map at the dark clump comes from a few galaxy images only. For this, the sample of galaxy images inside the aperture was bootstrap resampled, to obtain the prob′ ability pboot (Map ) that this resampling yields a particular ′ value of Map . This probability is also shown in Fig. 8. The ′ probability that the bootstrapped value of Map is nega-

tive is 3.8 × 10?4 for Chip 3, and < 10?6 for the MOCAM peak. The radial dependence of the tangential image ellipticity is considered next. Fig. 9 shows the mean tangential image ellipticity in annuli of width 20′′ , both for the MOCAM and the UH8K data centered on the dark clump. The error bars show the 80% probability interval obtained again from bootstrapping. It is reassuring that the radial behaviour of γt is very similar on the two data sets. ? In fact, owing to the di?erent wavebands of the two data ?elds and the fact that the aperture does not ?t inside the MOCAM ?eld, this agreement is better than one might expect. The mean tangential ellipticity is positive over a large angular range; except for one of the inner bins (for which the error bar is fairly large), γt is positive in all ? bins for θ < 150′′ . This ?gure thus shows that the large ?

Dark mass concentration near Abell 1942


Fig. 7. The same as Fig. 6 for the UH8K-chip3 I-band data. The ?lter scales are, from left to right: 80′′ , 120′′ , 160′′ and 200′′ . A1942 and the dark clump are also detected here with a very high signi?cance
′ and signi?cant value of Map at the dark clump is not dominated by galaxy images at a particular angular separation.

rameters, this becomes σv = 1135 γ100 0.06 1 km/s , 3 Dds /Ds (8)

3.2. Properties of the dark clump We now investigate some physical properties of our dark clump candidate. We ?rst argue that it is very unlikely for our object to lie at a redshift higher than 1. For our magnitude limit of 24.5 in the I band we expect approximately 30 galaxies/(1′)2 . We used approximately half of them (see Sec. 2) as putative background galaxies for our analysis. The median of simulated redshift distributions that extend the CFRS data (Lilly et al. 1995) to fainter magnitude limits (Baugh, Cole & Frenk 1996) is at about z ≈ 0.7 ? 0.8. If we assume that all our galaxies lie in the extreme tail of these distributions, then z = 1.0 represents a good upper limit for the redshift of our clump. However, the lensing analysis of the high-redshift cluster MS1054?03 (Luppino & Kaiser 1997) may provide an indication for a somewhat larger mean source redshift. Next we use Fig. 9 to obtain a crude estimate of the mass of this object. Although the tangential shear appears to be fairly small close to the center position of the clump, there is a region between ? 50′′ and ? 150′′ where the tangential shear is clearly positive and decreases smoothly with radius. If we describe the mass pro?le by an isothermal sphere, its velocity dispersion σv would be given by σv c

where γ100 is the tangential shear 100′′ from the mass center. Alternatively, we can express this result in terms of 2 the mass within a sphere of radius R, M (< R) = 2σv R/G; ?1 for example, within R = 0.5h Mpc, we ?nd M (< 0.5h?1 Mpc) = 2.9 × 1014 h?1 M⊙ γ100 1 . 0.06 3 Dds /Ds (9) Whereas this model is quite crude, the largest uncertainty in quantitative mass estimates comes from the unknown redshift of the dark clump and the unknown redshift distribution of the background galaxy population. The mass is a monotonically increasing function of the lens redshift, and depends very strongly on the assumed mean source redshift, in particular for values of zd > 0.5. ? With the I band data we now estimate the light coming from the dark clump. For this we created a SExtractor catalog counting every connected area with at least 3 pixels 0.5-σ above the sky background as a potential object. The ?ux of all these objects (except from obvious stars) in a circle of 100′′ radius around the clump center was summed up. We did the same in 32 control circles around ‘empty’ regions in the other UH8K chips. It turned out that the ?ux within the clump region is compatible with the mean ?ux of the control annuli, i.e., there is no overdensity of light at the position of the dark clump. So we took the 1-σ ?uctuation of the ?uxes in the control circles as a reasonable upper limit for the light coming from the dark clump. For converting the ?ux into a total I band magnitude we assumed that we are dominated by elliptical galaxies, using K corrections for this galaxy type calculated with the


1 (γt θ) 2π

Dds Ds




where the product γt θ would be independent of θ for an isothermal sphere model, and the ?nal term is the ratio lens-source to observer-source distance, averaged over the background galaxy population. Introducing ?ducial pa-


Dark mass concentration near Abell 1942

′ Fig. 8. Probability distributions for Map , with the aperture centered on the peak position of the dark clump. Solid (dashed) curves correspond to the MOCAM (Chip 3) data set. For an aperture of 160′′ , the left of the two curves shows the probability ′ ′ distribution p0 (Map ) for values of Map obtained by randomizing the position angles of the galaxy images. These two curves ′ nearly coincide. The two curves on the right-hand side show the probability distribution pboot (Map ) obtained from bootstrap ′ resampling of the galaxy images inside the aperture. The two vertical lines show the measured values of Map

latest version of the Bruzual & Charlot stellar population synthesis models for the spectrophotometric evolution of galaxies (Bruzual & Charlot 1993). From the total I band magnitude we derived a bolometric magnitude and a bolometric luminosity using standard approximations. With a lower limit for the mass and an upper limit for the luminosity we can give lower limits for the mass to light ratio of our object. This is shown in Fig. 10 for di?erent source redshift distributions and two cosmologies. We see that the EdS universe gives fairly high M/L estimates in comparison to a ? = 0.3, Λ = 0.7 model. When we assume a redshift of z ≈ 0.8 for our clump we obtain a lower limit of M/L ≈ 300 in the Λ cosmology. This is a conservative lower limit which could be lowered signi?cantly only if one

assumes that the redshift distribution of the faint galaxies extends to substantially higher redshift. As the dark clump has a mass characteristic of massive clusters it is of interest to search for X-ray emission associated with it. 3.3. The X-Ray data analysis A1942 was observed by the ROSAT HRI in August 1995. The total integration time was 44,515 s. We retrieved the X-ray images from the public archive and reduced them using ESAS, Snowden’s code especially developed for the analysis of extended sources in ROSAT data (Snowden et al 1994; Snowden & Kuntz 1998). The region showing a signi?cant peak in the weak lensing reconstructed mass map is within the ?eld of view of

Dark mass concentration near Abell 1942


Fig. 9. Mean tangential image ellipticity in independent bins of width 20′′ around the dark clump, triangles show the mean, solid (dashed) error bars the 80% error interval obtained from bootstrapping, using the MOCAM (Chip 3) data. For better display, the points and error bars are slightly shifted in the θ direction.

the HRI image of A1942. We have searched for X-ray emission in this area. First of all, we have re?ned the astrometry in the X-ray image matching X-ray point sources to objects in our deep optical images. The astrometric o?set from the original instrument coordinates is 3.5”. There is a signi?cant X-ray emission peak centered at 14h 38m 22.8s , 3? 33′ 11′′ (J2000.0). This position is 60′′ away from the weak lensing mass peak. The X-ray source is detected at the 3.2-σ level using an aperture of 30′′ radius. Although the number of counts detected is low, its distribution is inconsistent with a point-like source, showing a pro?le elongated along the NW-SE direction that is broader than the instrumental PSF. We have measured the source count-rate using concentric circular apertures centered on the X-ray emission peak. We obtain a count-rate of 7.4 ± 2.5 × 10?4s?1 within a circular aperture of 45′′ radius. The counts still increase

somewhat at larger radii but the measurement is much noisier given the uncertainty in the sky determination. The total ?ux is thus approximately 10-30% larger than the above value. We convert the count-rate into a ?ux assuming an incident spectrum of T = 3 keV and a local hydrogen column density of NH = 2.61 × 1021 cm?2 . The resulting unabsorbed ?ux is 3.4±1.2×10?14 erg cm?2 s?1 in the 0.1-2.4 keV band. We have also ?tted a standard beta pro?le (Cavaliere & Fusco-Fermiano 1978) to the azimuthally averaged radial pro?le. We obtain best values for the core radius and beta parameter (slope decline at large radii) of 15′′ and 0.80, respectively, although these values are quite uncertain given the low total number of counts. The X-ray luminosity depends on the redshift of the source. Assuming an incident spectrum at the detector of T = 3 keV [T = 3(1 + z)keV at the source], the rest-frame


Dark mass concentration near Abell 1942

Fig. 10. Estimate of the lensing mass (upper left panel), an upper bound for the luminosity of the lens (upper right panel), and a lower limit on the mass-to-light ratio (lower panel), as a function of assumed lens redshift. All estimates are for an aperture size of 100′′ . The solid, short dashed and long dashed curves show the M/L ratio in an EdS universe for zs = 0.8, zs = 0.9 and zs = 1.0. The dotted, dot-short dashed and dot-long dashed curves show the same in an ? = 0.3 , Λ = 0.7 universe. We have assumed a redshift distribution ∝ z 2 exp[?(z/z0 )3/2 ] for the source galaxies; hence zs ≈ 1.5z0 . A value of γ100 = 0.06 was assumed.

X-ray luminosity in the 0.1-2.4 keV band would range from 1.9 ± 2.5 × 1042 h?2 erg s?1 if the redshift is the same as that of A1942 (z = 0.223) to 3.5 ± 0.5 × 1043 h?2 erg s?1 if z = 1.0 (qo = 0.5). We have also made a crude estimate of the mass of the system. On the one hand, if we assume an X-ray luminosity–temperature relation (e.g., Reichart et al 1999, Arnaud & Evrard 1999) and a temperature–mass relation (e.g., Mohr et al 1999), we can get mass estimates at a 0.5h?1 Mpc radius from 1.5 × 1013h?1 M⊙ at z = 0.223 to 1.6 × 1014 h?1 M⊙ at z = 1 (qo = 0.5). We can also assume a beta pro?le, ?xing the core radius and the beta parameter, and compute the normalization necessary to obtain the observed ?ux at the measured radius. Then we can

integrate the pro?le to obtain the gas mass. If we further assume a gas fraction, we can also obtain a total mass estimate. If we take the values obtained from our previous ?t of the X-ray surface brightness pro?le, we get total masses at a radius of 0.5h?1 Mpc, of 9.2 × 1012 h?1 M⊙ at z = 0.223 and 2.3 × 1013 h?1 M⊙ at z = 1 (qo = 0.5). Note the di?erence of a factor of 1.5 and 7 compared to the previous estimates. This gives an indication of the errors involved. If instead we were to use typical values of the core radius and beta parameter of most clusters of galaxies (e.g., rc = 0.125h?1 Mpc and β = 2/3) the mass estimates would be approximately a factor 3 larger and closer to the estimates using standard correlations.

Dark mass concentration near Abell 1942


Fig. 11. For the UH8K-chip3 ?eld, surface mass density (black) and X-ray (white) contours are plotted. The surface density contours are the same as in Fig. 5, whereas the X-ray contours correspond to 1.5 × 10?5 , 1.6 × 10?5 , 2.0 × 10?5 , 3.0 × 10?5 and 4.0 × 10?5 counts/s/pixel. The cluster A1942 itself is clearly seen in X-rays, centered on the brightest cluster galaxy. In addition, extended X-ray emission near the dark clump is detected

Although we have presented quantitative values for the mass of the system based on the X-ray data, these should be taken only as informative given the assumptions and errors involved. Our main point in presenting these estimates is to show that this system has the X-ray properties of a galaxy group if it is at the same redshift of A1942. The lensing shear signal measured would then be too large for such a group unless it had a remarkable unusually high mass-to-X-ray light ratio. It seems more plausible that the system is a more massive cluster of galaxies at a higher redshift if the X-ray and lensing signal do indeed come from the same source, although the X-ray derived mass is still lower than the one obtained from the shear signal. The small angular scale X-ray core radius (larger physical scale if at larger redshift) and the lack of bright galaxies also point towards the same conclusion.

As an alternative, the X-ray emission may be unrelated to the dark clump, but associated with the small galaxy number overdensity projected near it, as seen from the black contours in the right-hand panel of Fig. 5. In that case, both the local enhancement of the galaxy density and the X-ray emission may be compatible with a group of galaxies, rather than a massive cluster, as indicated by the weak lensing analysis. 4. Discussion and conclusions Using weak lensing analysis on a deep high-quality wide?eld V -band image centered on the cluster Abell 1942, we have detected a mass concentration some 7′ South of the cluster. This detection was con?rmed by a deep I-band image. No clear overdensity of bright galaxies spatially


Dark mass concentration near Abell 1942

associated with this mass concentration is seen; therefore, we termed it the ‘dark clump’. A slight overdensity of galaxies is seen ? 1′ away from the mass center of the dark clump, but it is unclear at present whether it is physically associated with the mass concentration. Archival X-ray data allowed us to detect a 3.2-σ X-ray source near the dark clump, separated by 60 arcseconds from its peak; it appears to be extended. The X-ray source is spatially coincident with the slight galaxy overdensity. We have estimated the signi?cance of the detection of this mass peak, using several methods. For the V -band image, the probability that this mass peak is caused by random noise of the intrinsic galaxy ellipticities is ? 10?6 ; a similar estimate from the I-band image yields a probability of ? 4 × 10?4 . Thus, the mass peak is detected with extremely high statistical signi?cance. A bootstrapping analysis has shown that the tangential image alignment is not dominated by a few galaxy images, as also con?rmed by the smooth dependence of the tangential shear on the angular separation from its center. Whereas these statistical tests cannot exclude any systematic e?ect during observations, data reduction, and ellipticity determination, the fact that this dark clump is seen in two independent images, taken in di?erent ?lters and with di?erent cameras, make such systematics as the cause for the strong alignment highly unlikely. Although we have accounted for the slight anisotropy of the PSF, the uncorrected image ellipticities yield approximately the same result. A simple mass estimate of the dark clump shows it to be truly massive, with the exact value depending strongly on its redshift and the redshift distribution of the faint background galaxies. The mass inside a sphere of radius 0.5h?1 Mpc is > 1014 h?1 M⊙ , if an isothermal sphere ? model is assumed; if the lens redshift is larger, this lower mass limit increases, by about a factor 2 for z ? 0.5 and a factor of about 10 for z ? 1. In any case, this mass estimate appears to be incompatible with the X-ray ?ux if the dark clump corresponds to a ‘normal cluster’, at any redshift. We therefore conclude that the mass concentration, though of a mass that is characteristic of a massive cluster, is not a typical cluster. This conclusion is independent of whether the X-ray emission is physically associated with the dark clump or not. The lack of an obvious concentration of galaxies near the mass peak has been transformed into an upper limits of the luminosity associated with the mass concentration, and therefore into a lower limit of the mass-to-light ratio. This M/L limit depends again strongly on the redshift distribution of the faint galaxies, as well as on the assumed clump redshift. Whereas values for M/L as low as ? 200 (in solar units) are theoretically possible if the clump has a redshift in excess of unity, the corresponding mass becomes excessively and unrealistically large; for more reasonable redshifts zd < 0.8, M/L > 450 for an Einstein-de ? ? Sitter Universe, and M/L > 300 for a low density ?at Uni? verse. We would like to point out, though, that estimates

of the M/L-ratio quoted in the literature practically never assume a Λ-dominated cosmology, so that the M/L ratio quoted above for the low-density Universe cannot be directly compared to literature values. We can only speculate about the nature of this dark clump. As argued above, a normal cluster seems to be ruled out, owing to the lack of bright X-ray emission. Whereas the estimated X-ray luminosity can be increased by shifting the putative cluster to higher redshifts, the corresponding lens mass also increases with zd , in a way which depends on the redshift distribution of the source galaxies. The spatial coincidence of the slight galaxy overdensity and the X-ray emission, both ? 1′ away from the mass center of the dark clump, may best be interpreted as a galaxy group or weak cluster at relatively low redshift and not associated with the dark clump. The dark clump itself may then be a mass concentration with either low baryon density or low temperature, or both. For example, it may correspond to a cluster in the process of formation where the gas has not yet been heated to the virial temperature so that the X-ray luminosity is much lower than expected for a relaxed cluster. The fact that the tangential shear decreases towards the center of the mass clump may indeed be an indication of a non-relaxed halo. Further observations may elucidate the nature of this mass concentration. Deep infrared images of this region will allow us to check whether an overdensity of IRselected galaxies can be detected, as would be expected for a high-redshift cluster, together with an early-type sequence in the color-magnitude diagram. A deep image with the Hubble Space Telescope would yield a higherresolution mass map of the dark clump, owing to the large number density of galaxies for which a shape can be measured, and thus determine its radial pro?le with better accuracy. Images in additional (optical and IR) wavebands can be used to estimate photometric redshifts for the background galaxies. In conjunction with an HST image, one might obtain ‘tomographic’ information, i.e., measuring the lens strength as a function of background source redshift; this would then yield an estimate of the lens redshift. The upcoming X-ray missions will be considerably more sensitive than the ROSAT HRI and will therefore be able to study the nature of the X-ray source in much more detail. And ?nally, one could seek a Sunyaev-Zel’dovich signature towards the dark clump; its redshift-independence may be ideal to verify the nature of a high-redshift mass concentration. But whatever the interpretation at this point, one must bear in mind that weak lensing opens up a new channel for the detection of massive halos in the Universe, so that one should perhaps not be surprised to ?nd a new class of objects, or members of a class of objects with unusual properties. The potential consequences of the existence of such highly underluminous objects may be far reaching: if, besides the known optical and X-ray luminous clusters,

Dark mass concentration near Abell 1942


a population of far less luminous dark matter halos exist, the normalization of the power spectrum may need to be revised, and the estimate of the mean mass density of the Universe from its luminosity density and an average massto-light ratio may change. We also remind the reader that already for one cluster, MS1224, an apparently very high mass-to-light ratio has been inferred by two completely independent studies (Fahlman et al. 1994; Fischer 1999).
Acknowledgements. We thank Emmanuel Bertin, Stephane Charlot, Nick Kaiser, Lindsay King and Simon White for usefull discussions and suggestions. We are grateful to Stephane Charlot for providing the K-corrections of elliptical galaxies in the I band. This work was supported by the TMR Network “Gravitational Lensing: New Constraints on Cosmology and the Distribution of Dark Matter” of the EC under contract No. ERBFMRX-CT97-0172, the “Sonderforschungsbereich 375-95 f¨r Astro–Teilchenphysik” der Deutschen Forschungsgemeinu schaft, and a PROCOPE grant No. 9723878 by the DAAD and the A.P.A.P.E.

Luppino, G. A., Kaiser, N. 1997, ApJ 475, 20 Mohr, J.J., Mathiesen, B. & Evrard, A.E. 1999, ApJ, 517, 627 Navarro, J., Frenk, C. S., White, S. D. M. 1997, ApJ 490, 493 Press, W. H., Schechter, P. 1974, ApJ 187, 425 Reblinsky, K., Kruse, G., Jain, B., Schneider, P. 1999, in preparation Reichart, D.E., Castander, F.J. & Nichol, R.C. 1999, ApJ, 516, 1 Schneider, P., 1996, MNRAS 283, 837 Schneider, P., Seitz, C. 1995, A& A 294, 411 Seitz, S., Schneider, P., Bartelmann, M. 1998, MPA report 1086 Snowden, S. & Kuntz, K.D., 1998 ESAS documentation, ftp://legacy.gsfc.nasa.gov/rosat/software/fortran/esas. Snowden, S.L., McCammon, D., Burrows, D.N. & Mendenhall, J.A. 1994, ApJ, 424, 714 Richstone, D., Loeb, A., Turner, E. L. 1992, ApJ 393, 477 van Waerbeke, L., Erben, T., Mellier, Y., Schneider, P., Cuillandre, J. C., in preparation

Arnaud, M. & Evrard, A.E. 1999, MNRAS, 305, 631 Bahcall, N.A. & Fan, X. 1998, ApJ 504, 1 Bartelmann, M., Narayan, R., Seitz, S., Schneider, P. 1996, ApJ 464, L115 Baugh, C.M., Cole, S., Frenk, C. S. 1996, MNRAS 282, L27 Bertin, E., Arnouts, S. 1996, A& A 117, 393 Borgani, S., Plionis, M., Kolokotronis, V. 1999, MNRAS 305, 866 Bruzual, G. A., Charlot, S. 1993, ApJ 405, 538 Cavaliere, A. & Fusco-Fermiano, R. 1978, A&A, 70, 677 Cole, S., Treyer, M-A., Silk, J. 1992, ApJ 385, 9 Dickey, J.M. & Lockman, F.J. 1990, ARAA, 28, 215 Eke, V. R., Cole, S., Frenk, C. S. 1996, MNRAS 282, 263 Erben, T., van Waerbeke, L., Bertin, E., Schneider, P., Mellier, Y., in preparation Evrard, A.E., Metzler, C. & Navarro, J.F. 1996, ApJ, 469, 494 Fahlman, G., Kaiser, N., Squires, G. & Woods, D. 1994, ApJ 437, 56 Fischer, P. 1999, stro-ph/9901407 Gorenstein, M. V., Shapiro, I. I., Falco, E. E. 1988, ApJ 327, 693 Hoekstra, H., Franx, M., Kuijken, K., Squires, G. 1998, ApJ 504, 636 Jain, B., Seljak, U., White S. D. M. 1999, astro-ph/9901191 Kaiser, N., Squires, G., Broadhurst, T. 1995, (KSB95), ApJ 449, 460 Kaiser, N., Squires, G., Fahlmann, G. G., Woods, D. 1994, Clusters of Galaxies (Proceedings of the XXIXth Rencontre de Moriond; ed. by Durret, F., Mazure, A., Tran Thanh Van, J.) Kruse, G., Schneider, P. 1999, MNRAS 302, 821 Lacey, C., Cole, S. 1993, MNRAS 262, 627 Landolt, A. U. 1992, AJ 104, 340 Lilly, S. J., Tresse L., Hammer F., Crampton, D., Lefevre, O. 1995, ApJ 455, 108 Luppino, G., Bredthauer, R., Geary, J., C. 1994, SPIE 2198, 810

This article was processed by the author using Springer-Verlag a L TEX A&A style ?le L-AA version 3.

...of Galaxy Cluster Abell 1689 Evidence for a Merg....pdf

Complex Structure of Galaxy Cluster Abell 1689 ...clusters with a high concentration of central X-...matter) of a galaxy cluster from its X-ray ...

Galaxy Cluster Formation from the Large-scale Structure A ....pdf

Galaxy Cluster Formation from the Large-scale ...concentration of galaxies in the Abell 2125 ...As our source detection algorithms are optimized ...

...Assembly of the Brightest Cluster Galaxy in a Cluster at z....pdf

of a rich, high concentration system resembling ...The projected separation between the peaks of the...Abell clusters have multiple nuclei within 19.2 ...

Study of the Core of the Shapley Concentration I.th....pdf

Study of the Core of the Shapley Concentration ...Abell clusters A3556, A3558 and the poor ...Finally, from an analysis of the projected and ...

The Toronto Red-Sequence Cluster Survey First Results.pdf

morphology in the form of a concentration index.... and o the mass of each cluster can be ...Abell Richness counts NFW core scale radius ...

Quasars and Galaxy Clusters Paired Across NGC 4410_图文.pdf

13.6 19.4g 18.9g 19.3g 18.5g 17.3 19.... a dense concentration of points streaming away ...4.1. Abell Clusters 1541, 1541A and 1541C The...