|Variation of the MegaCam u' band zeropoints since 2003. The highest values are usually measured right after the aluminization of the mirror. However, the 2014 aluminization brought the zeropoints about 0.2 magnitudes bellow what we usually get after aluminizing the mirror. The red points are the are more reliable points that have more than 5 images spread over more than one night with a scatter of less than 0.05 mag. The grey points are less reliable measurements. The number of images/nights is labeled next to the CRUNID label on the plot. The error bars indicate scatter, not error. Credits: Stephen Gwyn, NRC-H, Canada.|
During the November 2014 SAC meeting in Waimea, CFHT staff were made aware of an unexplained zero-point drop of sensitivity in MegaCam images dating back to late 2013. A subsequent review of the data confirmed an anomalous (~0.5 mag) shift in photometric zero-points over time and lost sensitivity that was not fully restored after recoating the primary mirror in August 2014.
A comprehensive search for the source of MegaCam's sensitivity loss was started on November 18, 2014. The scientific and technical staff investigated software bugs, the most obvious being contamination or poor normalization of the master flat or bias frames. Flats and biases recorded prior and following the sensitivity drop were compared by taking the median pixel value of each CCD, then averaging them across the CCD mosaic. The flats were found to vary by 0.1% and the biases varied by 1%. When MegaCam went off the telescope, an inspection of the cryo vessel window, the top surfaces of the image stabilizer plate and L4 in the wide Wide Field Corrector was instigated. No contamination consistent with a ~0.5 mag sensitivity loss was identified. A cleaning of the bottom, down looking face of the lowest and largest lens in the wide field corrector (L1) was also performed.
During the subsequent December MegaCam run, all the images showed extended halos around bright stars. Analysis showed that there was a loss of flux in the core of the PSF and that a significant portion of the lost flux could be found in the extended diffuse halo, suggesting the phenomenon is primarily scattered light rather than absorption or reflection. It was decided on December 18, 2014 to remove MegaCam from the telescope and identify the source of this scattering problem.
Left: Two images of the D1 CFHTLS deep field, in g-band, with a 120s exposure time. The image on the left was taken with the defective AR coating on L1 and the image on the right shows the same field after the AR coating was chemically stripped from L1.|
Right: Radial profile of the extended halos compared to the halos regularly seen on MegaCam. The bump seen in the red circle corresponds to the usual footprint of the MegaCam halo.
Credits: Simon Prunet, CFHT.
Visual inspection of the upper and lower optical elements of the MegaCam Wide Field Corrector showed an unexpected haziness on the down looking face of L1. A chemical analysis by an independent lab of a sample of the L1 contaminant identified it as silicone lubricant. All efforts at removing the haze with a wide range of organic and other solvents proved futile.
Further detailed inspection of the L1 surface with a microscope showed clearly that the scattering observed was the result of extensive pitting of the anti-reflection (AR) coating on the lens. The decision was made to remove the coating of the lens via chemical stripping and proceed with the scheduled January MegaCam run. Early images from that run do not show any signs of extended halos and a preliminary analysis indicates that the zero-points are closer to their nominal values.
At this point, we have not identified the cause of the L1 AR coating failure, though we continue to explore various possibilities. Given the implied MegaCam downtime and marginal gains associated with applying a new AR coating to L1, we do not plan to have a new AR coating applied to that optical surface.