The observations took place at the ESO/MPI 2.2m telescope situated at La Silla Observatory, Chile, on the nights of April 14 and 15, 1997. The IR camera IRAC2b was used at the Cassegrain focus of the telescope. The detector of IRAC2b is a 256 256 NICMOS 3 HgCdTe array and the instrument has a variety of optical lenses available for imaging at different pixel scales. Lens LB (Lidman, Gredel & Moneti, 1997) was chosen since it gives a good compromise between the pixel scale (a small pixel is needed for the deconvolution) and the size of the field. During the observations, PG 1115+080 was also observed. A by-product of these observations is a more accurate estimate of the IRAC2b pixel size for images taken in lens LB. The new scale, 0.2762 /pixel, is based on the precise astrometry of this field given by Courbin et al. (1997). This results in a field of view of 71.
Numerous short exposures of HE 1104-1805 were obtained in J ( =1.25 micron) and ( =2.15 micron) under good meteorological conditions. The mean seeing was 0:007-0:008 and the sky was photometric. We set the Detector Integration Time (DIT) to 20 sec in and 60 sec in J. Each image taken in (resp. J) is the average of 3 (resp. 2) such integrations. The choice of the DIT is dictated by detector saturation. The number of DITs is dictated by the frequency at which the sky intensity varies. Since the field of HE 1104-1805 is uncrowded, we took sequences of 9 science exposures, dithered in a random manner by 5 to 10, always keeping the object and three PSF stars in the field.
Dome flat-fields were taken in order to correct for the pixel to pixel sensitivity variations of the detector. However, dome flat-fields do not accurately represent the large scale sensitivity variations of the array, in the J band. Fortunately, this variation does not show strong gradients and has a maximum amplitude of 3-4% over the whole field. It was modeled by observing a bright star over a grid of 9 different positions across the array and by fitting a two dimensional, third order polynomial to the flux of the star. The residuals of the fit were 1%. This fit, which is commonly called an illumination correction, was multiplied by the dome flat to produce the final flat-field containing both the low and high frequency sensitivity variations of the array. After subtraction of a dark frame, the flat-field correction is applied to all scientific frames.
The background is removed from every exposure. It is estimated for any particular exposure by averaging the 6 preceding and the 6 following exposures. Thanks to the dithering between exposures, objects in these frames could be rejected before the average was taken. This method allows us to accurately follow the background which varies on the time scale of a few minutes.
Standard stars were observed every two hours. The standard deviation in the zero points were 0.024 magnitudes in J and 0.012 magnitudes in K. The magnitude and colours in this paper are in the JHK system as defined by Bessell and Brett (1988).
© European Southern Observatory (ESO) 1998
Online publication: January 8, 1998