2. Near infrared (nir) photometry

Near infrared Kn band images of NGC 4650a were obtained at the 4m. Anglo-Australian telescope on the of April 1994, with the IRIS infrared camera at the f / 15 Cassegrain focus. The detector is a hybrid array of 128 128 pixels: our image scale is 0.61”/pix. The Kn filter was chosen for its high sensitivity and because it greatly reduces the effect of water vapour. Several

standard stars from the “IRIS Users’ Guide” were observed during the night to transform the images into the standard Kband system. Since the polar ring in NGC 4650a subtends an angle on the sky bigger than 1', a mosaic of 3 fields was required to image the whole system: total integration time for each field is 450 seconds. The images for each field were acquired using the offset technique: a cycle was defined so that 5 sub-images were taken together with sky frames (before and after each exposure on the galaxy), and bias frames (at the beginning of each cycle in order to linearise the CCD frames). Linearisation and bias subtraction were done using the FIGARO task irislin at the end of each cycle. Flatfielding was done using the “light-on, lightoff” dome flats, after differencing and normalising them. Each sub-frame was sky subtracted using an averaged sky, derived from the sky frames taken before and after it; the resulting image for each field of the mosaic was finally derived by registering and co-adding all subframes (see Fig. 1). The luminosity profiles were extracted along the position angles of the PR major axis and the host galaxy major axis (as listed in the PRC).

© European Southern Observatory ' Provided by the NASA Astrophysics Data System

F. Combes & M. Arnaboldiz The dark halo of polar-ring galaxy NGC 4650a

3. Modelling

3.1. Fitting the light distribution

The B,I broad band brightness profiles along the major axis of the lenticular, and along the polar ring are available in the literature (WMS, S94); the K band profile comes from this work. The lenticular light profile is exponential in the three bands, with a characteristic radial exponential scale of rd R: 4.5”. S94 have shown that the profiles could be well reproduced by both a thin and a thick exponential disk with respectively 4.7” and 4.4” radial scale (0.9” and 1.8” exponential scale height), provided that the inclination is then adjusted to 68° for the thin disk, and 78° for the thick disk. However, we stress that the observed axis ratio, 0.42, is still compatible with the typical thickness of a lenticular galaxy seen edge-on, so the inclination and the thickness parameters are not uniquely determined. We choose to model the lenticular with a thick disk (4.4” radial scale and 1.8” z-scale), which better corresponds to the galaxy morphological type (see Fig. 2). The fits to the luminosity profiles in the two cases are not very different, as shown by S94, but a thick disk gives a slightly better fit. We also add a small bulge, with a luminosity of 4.3 - 109 LG (SS), and a scale rb z 1". This bulge is detected only with good seeing conditions, and its low luminosity does not correspond to what can be expected for a typical bulge in a SO. It could correspond to a region of recent star formation in the nucleus, triggered by the gas accretion event following the polar ring formation, as suggested by S94. Moreover, S94 fits to the observed radial velocities and the velocity dispersion indicate a very low ratio (0.5) for this bulge. In any case, this component is not adding uncertainty to the parameters, since it has negligible effect on the various fits.

The fit to the PR luminosity profile gives more problems, because its inclination and position angles are slightly varying as function of radius. WMS pointed out that it is not possible to have the whole polar ring in the same slit, due to its twisting and S-shape bending. The underexposed blue image from WMS, with a seeing of‘ 1.6”, reveals clearly that the polar ring is not edge-on in the central part until roughly 30", where it becomes edge-on. This behaviour confirms that the polar ring in NGC 4650a is indeed a ring whose surface density is maximum at 1' 30". The I-band image from S94 (seeing 2.25”), and our K-band image (1 ) do not allow us to distinguish clearly the central hole of the polar ring, possibly because the dust lane is less prominent at larger wavelength, and there is a strong effect caused by the seeing conditions. The inclination of the central region of the polar ring is not far from edge-on, of the order of 85°. The velocities will not be affected, as far as their projections are concerned; but the luminosity profile, and the apparent velocity profile are significantly modified even in the case of a 5° departure from edge-on, due to the integration along the line of sight, as will be shown later on. Because of all these uncertainties, a fit with varying inclinations and position angles is not worthwhile. We choose a constant inclination of 85°, as suggested by the blue image, and an exponential thickness of 300 pc (corresponding to an axis ratio of w 0.05) to reproduce