Intrinsic pixel size variation in an LSST prototype sensor

Abstract

The ambitious science goals of the Large Synoptic Survey Telescope (LSST) have motivated a search for new and unexpected sources of systematic error in the LSST camera. Flat field images are a rich source of data on sensor anomalies, although such effects are typically dwarfed by shot noise in a single flat field. After combining many (0∼50) such images into ‘ultraflats’ to reduce the impact of shot noise, we perform photon transfer analysis on a pixel-by-pixel basis and observe no spatial structure in pixel linearity or gain at light levels of 100 ke− and below. At 125 ke−, a columnar structure is observed in the gain map—we attribute this to a flux-dependent charge-transfer inefficiency. We also probe small-scale variations in effective pixel size by analyzing pixel-neighbor correlations in ultraflat images, where we observe clear evidence of intrinsic variation in effective pixel size in an LSST prototype sensor near the ∼ .3% level.