3. HST/WFPC2 observations
3.1. The HST/WFPC2 data set
The HST data of the Cloverleaf used in this analysis have been provided by the ESO/ST-ECF Science Archive Facility (Garching). Two data sets obtained with the WFPC2 are presented here. The first one was obtained by Turnshek (1994 July 12 [ID:5442] & 1994 December 23 [ID:5621]) and the second one by Westphal (1995 April 27 [ID:5772]). Turnshek's observations aimed at discovering the lensing galaxy. They consist in WFPC-2 images and FOS spectra of the four quasar images and of the suspected lensing object. These observations were not successful in finding the lensing galaxy, but the FOS spectra of the four quasar spots have been discussed by Turnshek (1995). A recent preprint by Turnshek et al (1997) discussed the PC observations (ID:5442 and 5621 as well as archival pre-costar observations), giving astrometry and color variation of the different spots (the latter which they explained in terms of dust absorption). They also put some limits on the detection of the primary lensing galaxy. Observations of the Cloverleaf by Westphal are part of a larger programme to study multiple quasars and their environments. Standard reduction procedures using IRAF/STSDAS packages have been applied. The absolute photometry was obtained using magnitude zero-points given in Holtzmann et al. (1995). Information about the final images in filter F336W (central rest wavelength 945 Å), F555W (central rest wavelength 1560 Å), F702W (central rest wavelength 1975 Å), and F814W (central rest wavelength 2290 Å) are summarized in Table 2.
Table 2. Summary of the HST data sets. Note: the data set of 1994 December shows a weird PSF, though the FGS status was on the FINE guiding mode: it may be due to some guiding problem resulting from the paucity of guide stars in this area.
3.2. Properties of the Cloverleaf images
Two types of information are relevant for our present goal of modelling the gravitational lens: the relative astrometry (with respect to spot A, as shown on Fig. 3) as well as the shape of the four spots and their intensity ratios.
Regarding the relative astrometry and spot sizes, the PC observations in filters F336W, F555W, F702W and F814W provide similar results displayed in Table 3. The four spots are stellar-like with a FWHM 1.5 pixel or .
Table 3. Absolute and relative astrometry of the four quasar spots. The absolute position of the brightest spot A comes from CFHT/FOCAM images (coordinates are given within a rms accuracy), while the relative positions of B, C and D with respect to A are from the WFPC2/PC1 data.
The absolute photometry and the relative intensity ratios (Table 4) have been computed with the Sextractor software (Bertin & Arnouts 1996) using total and isophotal magnitudes when images of the quasar were not saturated. No PSF fitting was applied as the four different spots are well separated on the PC; furthermore a PSF fitting would not have been possible for the December 1994 data as the PSF was not circular for this observation (see caption of Table 2). The important variation of the intensity ratios in U compared to V, R and I band could probably be only explained by absorption along the line of sight by intervening galaxies (HI clouds at redshift will act as very efficient absorbers in the U band). This absorption has to be larger for the spots A & B which display the biggest relative change to the C & D spots. Turnshek et al (1997) proposed that dust extinction can be the explanation with a preference for an SMC-like dust extinction at the redshift of the quasar. No obvious evidences of microlensing are detected from the HST data set, though small variations in the quasar spot intensity ratios are consistent with the ESO/NOT monitoring of the Cloverleaf (Ostensen et al 1997).
Table 4. Total magnitude and relative intensity ratios of the four quasar spots from the WFPC2/PC1 data, compared to previous observations found in the literature.
3.3. The Cloverleaf environment
Fig. 4 shows the deep F814W image of the Cloverleaf. The four quasar spots lay at the center of the WF3 chip. The presence of numerous faint objects over a region in the environment of the Cloverleaf is striking. In order to quantify this effect we have computed the object number density, N , and its dispersion, , in the magnitude range I=23 to 25. We have estimated this density in different regions across the image: in a diameter region around the Cloverleaf where the density contrast is clearly visible by eye and in various randomly selected areas of similar size. On the whole frame we find a mean value N objects/arcmin2 (1 ). The region around the Cloverleaf has N objects/arcmin2, with a peak at 130 objects/arcmin2. The detection of a peak is therefore significant at a 4 level. This brings us to suspect the presence of a cluster of galaxies (of unknown redshift), which could contribute significantly to the gravitational lens effect. We provide in Fig. 5 the deep F702W image of the Cloverleaf where the four quasar spots lay at the center of the PC1 chip. Similarly to the F814W image we have plotted the number density isocontours of faint objects between R=23 to 25. There is no overdensity larger than 2 . Therefore, we conclude that the 4 overdensity around the Cloverleaf is significant. Fig. 6 shows a zoom of this area from the deep F814W image: the photometry and color of the related objects (obtained by combining the deep WF and PC observations) are given in Table 5. The bulk of them are red objects with (from 0.7 to 1.2). While their morphology cannot be obtained from the HST images, their red color suggest that we might be dealing with E/S0 galaxies in a high-redshift cluster.
Table 5. Relative position from the quasar spot A, total magnitude and color of the galaxies detected in I (numbered as in Fig. 6). Typical error for the faintest object is 0.2 mag.
The small size (of the order of which hampers any strong morphological classification), faint magnitude () and red color of these galaxies are suggestive of high-redshift E/S0 galaxies (as can be found for example in the CFRS sub-sample of galaxies with 1 z 2 (Lilly et al 1995)). Moreover, the significant concentration of these galaxies near the particular line of sight to the multiple quasar is a convincing indicator of the presence of a high-redshift cluster. This cluster could either be linked physically to the quasar at redshift z 2.56 or along the line of sight. Hereafter, we assume this cluster to be at a redshift close to that of the narrow absorption systems observed by Turnshek et al (1988), and Magain et al (1988): either 1.438, 1.661, 1.87 or 2.07. For the sake of simplicity, we shall adopt hereafter a value of 1.7. Since no observations are publicly available yet in the near infrared, it is practically impossible to obtain secure photometric redshifts in this redshift range (e.g. the 4000 Å break would fall at 1 m at z=1.7). Dedicated high-resolution IR imaging and/or optical/IR spectroscopy should be carried out on the objects forming this overdensity in order to determine their distance and nature.
We have examined the alternative possibility that this ensemble of faint sources corresponds to a population of HII regions in a nearby galaxy with low surface brightness, extending over in diameter. The intrinsic size of such a dwarf galactic system, around 10 kpc, requires that its distance should be less than 50 Mpc (or its redshift less than 0.01). However, the observed R-I 0.9 color of the objects forming the overdensity is far too red to fit the R-I color of giant extragalactic HII regions observed in a sample of nearby HII galaxies, R-I -0.17 (Telles et al 1997) or as modelled, R-I -0.2 (e.g. Bica et al 1990). Furthermore, the observed I magnitudes are too faint by at least 7 magnitudes with respect to expected values of such an HII population located in the local Universe. Therefore, this interpretation should be discarded.
3.4. The lensing object
Although the cluster candidate mentioned earlier is acting as an additional lensing agent, the small angular separation between the four spots implies that most of the lensing effect is likely produced by a galaxy located amid the four spots of the quasar. This lensing galaxy has not been detected so far on either "deep" visible and near infrared ground based observations (Angonin et al 1990; Lawrence 1996) nor on the PC observations by Turnshek et al (1997).
Yet, the detection of the lensing galaxy would be of great importance to model accurately the mass distribution of the lens as we could then assume that the mass follows the geometry of the light, as in other successful lensing model calculations and as expected in the inner regions of galaxies.
We tentatively used the deepest WFPC2/F814W images in order to detect the central lensing galaxy. The four spots of the quasar have been subtracted using the PSF model provided by the nearby star. Due to the significant light residual related to the quasar photon noise, the telescope diffusion and the first diffraction rings, the limiting surface brightness in the center of the Cloverleaf is at least 1 mag lower than across the whole image. This reduces the limiting magnitude to a value of I 24, though the faintest objects detected in the field have I 25. Even at such a faint magnitude, there is no evidence of a signal spread over few pixels. We conclude that the lensing galaxy must have . We can set an upper limit on its velocity dispersion, assuming the extreme case where the lens is an elliptical galaxy fainter than I=24 with a redshift close to that of the narrow absorption lines seen on the quasar spectrum at . In order to infer the absolute magnitude of the corresponding lens, we have run a galaxy evolution model for an elliptical galaxy with standard parameters (power law Initial Mass Function with one segment ranging from 0.1 to 125 M , single burst of star formation, =1., =50 km/sec/Mpc) from Bruzual & Charlot (1993). We find that in the most favorable case, the galaxy should have an absolute magnitude larger than in order not to be detected. Using the Faber-Jackson relation (Fall 1981), we derive an upper limit of its velocity dispersion km/sec. A similar study can be performed using the 2000 sec. K-band exposure obtained at Keck (Lawrence 1996). Given the expected throughput of the Keck camera, we estimate the central lensing galaxy to have , which is a constraint not as stringent as that derived from the HST images. Hence, the visible and near infrared data are not sufficient to constrain the mass of the lensing galaxy. But in fact, if we take into account the likely presence of the distant cluster, it is no longer necessary to invoke a large mass (i.e. high brightness) for the lensing galaxy since a fraction of the convergence and the deflection angle would be contributed by the cluster.
© European Southern Observatory (ESO) 1998
Online publication: December 16, 1997