Image persistence in 1.7 μm cut-off HgCdTe focal plane arrays for SNAP

Image persistence, an elevated dark current caused by prior illumination, has been a chronic problem in near infrared focal plane arrays. It has the potential to interfere with precision photometric surveys for astronomy such as the Supernova/acceleration probe (SNAP) and other missions to study dark energy. By presenting a plausible model and supporting data, we address widespread misconceptions that persistence is a consequence of saturation and that it can be mitigated by multiple resets. Persistence increases monotonically with stimulus amplitude until the detector diodes approach forward bias, and, for the Teledyne detectors tested, is linear over most of that range, significantly improving the accuracy of calibration possible. We characterize the functional form of the temporal response, superposition of the decay curves from multiple stimuli and suppression of persistence by signal in the current frame. The effect of stimulus duration on response time constants, and the resulting scope for overly optimistic interpretation of manufacturer's data are described. Spatial variability in the amplitude and shape of the decay curve are examined. Worst case requirements for tracking prior exposure history, and some mitigation options are discussed together with practical compromises for correcting the complex behavior of persistence, given that the time required to calibrate individual pixels to sufficient accuracy is prohibitive.