Count rate dependent non-linearity and pixel size variations in 1.7 micron cut-off detectors

Abstract

Over the past decade scientists have collected convincing evidence that the content of our universe is dominated by a mysterious dark energy. Understanding the nature of dark energy is a very difficult task, and requires a variety of independent experimental approaches. Most of these approaches rely on photometric calibrations over a wide range of intensities using standardized stars and internal reference sources, and hence on a complete understanding of the linearity of the detectors. The SNAP near infrared (NIR) instrument team has performed a comprehensive study of precision photometry on 1.7 micron cut-off HgCdTe detectors. Among those studies are the count rate dependent detector non-linearity that was recently discovered with the NICMOS array on the Hubble Space Telescope, and possible pixel size variations seen in quantum efficiency (QE) data. The nonlinearity on NICMOS exhibits an unexpected behavior, where pixels with high (low) count rates detect slightly more (less) flux than expected for a linear system. To test this count rate dependent non-linearity a dedicated setup was built that produces a known amount of light on a detector, and measures its response as a function of light intensity and wavelength. If the pixel response variations seen in QE data are due to pixel area variations, standard flat-fielding will degrade photometry precision for point sources in an undersampled telescope. Studies have been performed to estimate the magnitude of pixel area variations.