2. Observations and results
The field of PSR1706-44 was observed with the Test Camera on VLT -UT1 on August 19, 1998, during SV. (The instrumentation is described in Leibundgut & Renzini 1999.) Six images of 600 seconds each were obtained in the band. All observations were made with 2x2 binning, a pixel thus corresponds to on the sky. The raw images were bias subtracted by determining the bias level in the overscan region of the CCD. The two-dimensional bias structure was removed with a master bias frame. Flatfielding was done using a flat obtained from the science observations on the previous night 1. The six images were aligned and combined into a final image (see Fig. 1). The quality of this image is very good, with a FWHM of .
Previous attempts to constrain the emission from the pulsar have been severely hampered by a bright nearby star. This star was named Star 1 by Chakrabarty & Kaspi (1998; henceforth CK98), and its magnitude was measured to . Due to poorer spatial resolution they could only obtain an upper limit for the pulsar of . The good seeing of the VLT image enables us to significantly improve upon this.
In Fig. 2 we show a blow-up of the region around Star 1. Using the radio position of the pulsar (Frail & Goss 1998; Wang et al. 1998), CK98 estimate that the pulsar should lie away from the star. The uncertainty in this position is a combination of errors in the radio position, errors in the astrometric solution to the optical image and a mismatch in aligning radio and optical frames. These are all of the order . We have adopted a combined error of . This error circle around the position away from Star 1 is shown in Fig. 2.
We carefully searched for the pulsar around that position in the VLT image, but no object was found. To estimate an upper limit to the pulsar's emission we constructed a PSF from ten bright stars in the field using the IRAF/DAOPHOT PSF task. We thereafter added a number of artificial stars with different magnitudes to the image. The magnitudes are all measured relative to Star 1 (, CK98). The colour terms for the configuration have been measured to be small and are neglected in our study.
To ensure similar backgrounds, the artificial stars were all positioned at a distance of from Star 1. We thus find that stars of magnitude should have been easily seen. Also is clearly visible but an artificial star with is rather faint. If the pulsar is positioned at the outer end of our error circle, even a star is easily seen. Measuring the background at this distance from Star 1 shows that an artificial star with has in fact a peak pixel value that is more than 3 above the background. This is thus a firm upper limit for the image.
However, if the pulsar would be positioned much closer to the star than , our method of measuring the limiting magnitude becomes more uncertain. In fact, we do see a region of brighter emission in the innermost part of our error circle. This might just be fluctuations in the PSF of the bright star. To estimate how bright a star one could hide in the PSF of Star 1 we instead subtracted artificial stars from this position until a hole appeared in the background. We find that it is possible to hide a rather bright point source () at a distance of from the star. As a firm upper limit for a pulsar this close to Star 1 we therefore claim .
We conclude that the pulsar is most likely fainter than V=26.0 magnitudes. Deeper exposures are needed to address this question. However, if the pulsar is indeed substantially closer to the bright Star 1, the PSF of that star limits our study. We claim an upper limit of V=24.5 inside our error circle. This is still much fainter than the previous upper limit of CK98. HST resolution would be required to improve upon this estimate.
© European Southern Observatory (ESO) 1999
Online publication: March 1, 1999