Astron. Astrophys. 339, L13-L16 (1998)

3. Results

3.1. R-band images: the lensing galaxy

Fig. 1 (left) shows an R-band single exposure of HE 2149-2745 A,B with FWHM = 1:000. The subtraction of the scaled PSF at the position of both QSO images clearly leaves a diffuse, ellipsoidal object lying between and more or less equidistant to A and B, as is shown in the right-hand panel. The simultaneous scaling of the PSF overfitted the fluxes at the A and B peaks, thus leaving small regions with relative negative residuals around both QSOs. This is expected given the contribution from the underlying object outside a few seeing radii; consequently, small corrections of roughly 0.01 (A) and 0.02 (B) magnitudes were made to recover the lacking flux. The mean flux values in the central region deviate from the background sky level. The stability of the PSF leads us to conclude that we have undoubtedly detected a galaxy in the LOSs to HE 2149-2745 A and B, most probably the gravitational lens.

Fig. 1. Left: R-band 300 s exposure of HE 2149-2745 A,B. North is on the top, east to the right. Right: Same as in left-hand panel but after scaling and subtracting the PSF at the positions indicated with the crosses. Two exposures with similar effective seeing of 1:000 have been combined (600 s).

Simple aperture photometry yields for the galaxy. The error bar reflects the tolerance range allowed by the fine-tuning described in the above paragraph, and includes the zero-point uncertainty. For the sum of both QSO components we find . These magnitudes are based on the photometric results by Wisotzki et al. (1996) for star 3 (assuming it has not varied), and consider a color-term correction of 0.05 which reduces them to the Landolt (1992) system ¹.

The relatively deep R-band images reveal many red non-stellar objects in the SUSI field of view, but no evident overdensity is observed close to the lensing galaxy to a limiting magnitude of . However, we cannot exclude the possibility that the galaxy belongs to a cluster since the small field of view implies transverse distances still compatible with cluster sizes. In particular, two of the objects, not much fainter than 20.4, lie within of the galaxy position. It remains therefore unclear whether the galaxy belongs to a cluster or not.

Table 1 also gives the calibrated magnitudes of both QSO components. As the V magnitude of star 3 is not available we assumed using a selected sample of stars with similar colors. We note that the magnitude difference between A and B is consistent with a flux ratio A to B of for all three band filters. Fig. 2 shows isophotal contours of the lensing galaxy. The isophotes are separated by 0.2 magnitudes, and the lowest contour level is approximately at above the background count. Note the distorted region near QSO A, probably due to photon-shot noise near the A peak (see below). The positions of A and B are marked with filled squares, since their errors are too small to appear at this scale. The angular separation between A and B is , and the position angle is . The position of the galaxy was derived using the isophotes at brighter levels than half the peak intensity: , and relative to A. There is a small misalignment between this position and the QSO images, but alignment is possible within the measurement uncertainties. Its orientation coincides surprisingly well with the line joining A and B (), with an ellipticity of . The galaxy center lies nearly equidistant to both QSOs, with and .

Table 1. Photometry and astrometry of HE 2149-2745 A,B and the lensing galaxy. Approximate magnitude limits.

Fig. 2. Isophote contours of the lensing galaxy in intervals of 0.2 magnitudes. The lowest contour level is at above the sky background.

The good spatial sampling of our data allows an estimate of the galaxy parameters. A section aligned with the south semimajor axis was used to fit isophotal ellipses to the flux values. Fig. 3 shows the averaged surface brightness profile (error bars). A de Vaucouleurs law and an exponential disk were fitted to the profile within , thus including only values at the level but not significantly affected by the seeing smoothing. The final fit models result from re-constructing the isophotal ellipses with the fitted profiles and smoothing them with the PSF. We see that the data are better modeled by a law () than by an exponential disk (), especially at the core. Furthermore, images taken under better seeing conditions should accentuate the concave shape of the observed profile. We take this result, though cautiously, as evidence for an elliptical galaxy.

Fig. 3. Surface brightness profile of the lensing galaxy. Solid line: law fitted to the data within , and convolved with the PSF.Dashed line: Same for an exponential disk. The fitting range is indicated by the vertical dotted lines.

3.2. B and V band images

The galaxy is not detected in the PSF-subtracted B and V band images. However, the region surrounding QSO image A is slightly overfitted at both sides of . Such residuals are not observed for B. This "symmetric" overfit, though not significant ( in B and only marginal in V), suggests A might be composed of two or more fainter point sources lying on . We investigated this possibility (cf., Bade et al. 1997; Burud et al. 1998) in the B-band, as one expects here less contamination by a hypothetical foreground galaxy (in spite of the fact that the PSF is broader in the blue). However, attempts to re-fit A with two sources of nearly half its intensity failed at recovering the background level. It is difficult to establish on the basis of the present data what causes the low-quality fits in the B images. From an observational point of view, splitting of QSO image A in very close sub-components cannot be ruled out, but an underlying object could also contribute to slightly distort the QSO images. An explanation of this must await better quality data.

To put upper limits on the galaxy B and V brightness a variance frame was created which considers photon statistics (dominated by the QSO fluxes), readout noise, and the uncertainties introduced by background subtraction. Integration of the variances in a 3:005 radius aperture yields detection limits for B and V. In addition, we consider the fraction of possible galaxy light hidden in the QSO seeing disks and hence subtracted with the PSF. Analysis of R images with similar seeing as in B and V shows that this fraction can be as large as 0.7 (B) and 0.5 (V). The estimated () detection limits are listed in Table 1.

© European Southern Observatory (ESO) 1998

Online publication: September 30, 1998
helpdesk.link@springer.de