1. S. Baggett, J. Anderson, J. MacKenty, J. Biretta, K. Noeske, and the WFC3 team (STScI) HST/WFC3 UVIS Detectors: Radiation Damage Effects and Mitigation Abstract The Wide Field Camera 3, a fourth-generation instrument installed on the Hubble Space Telescope in May 2009, has been operating well on-orbit. As expected, however, due to the harsh low-earth orbit environment the CCDs are accumulating radiation damage in the form of charge traps in the silicon. The result is a progressive loss in charge transfer efficiency (CTE) over time, which reduces the detected source flux as the traps intercept charge during readout and systematically shifts source centroids as the trapped charge is released during readout. The CTE losses depend not only on the number of transfers (distance of the source from the amplifier), but on source brightness as well as image background. We present the options available for WFC3 CTE loss mitigation, including the newly-commissioned post-flash mode which is highly effective particularly for faint sources in low backgrounds. We provide an update on the point source aperture photometry correction formulae and status of the empirical pixel-based correction. Summary  Low-earth orbit environment damages CCDs, creating hot pixels, increasing dark current, & decreasing charge transfer efficiency.  After >3.5 years on-orbit, WFC3/UVIS detectors exhibit – About 1000 new hot pix/day; monthly anneals fix ~80-90%. – Dark current of ~7e-/hr/pix, increasing by ~1.5 e-/hr/pix/yr. – Signal losses due to radiation damage: 2-4% (bright sources, high background) to ~50% (faint sources, low background).  CTE loss mitigation options – Placing small targets close to readout amp. – Applying post-flash to images during acquisition. – Adjusting photometry using formulaic corrections. – Correcting images using pixel-based image correction in post-observation processing, available in 2013. Mitigation option: formulaic corrections Characteristics of the UVIS channel  Two 2 K x4 K e2v thinned, back-illuminated CCDs at -83C  Field of view160 x 160 arc sec, covering 200-1000 nm  Read noise ~3 e-; dark ~7e-/pix/hr (~2 e-/pix/hr in 2009)  Standard readout: full-frame via 4 amps  Post-flash available for CTE loss mitigation Hot pixels and dark current Increasing over time. Histograms of WFC3 darks show an extended tail of hot pixels typical for CCDs. At a threshold of 54e-/hr/pix, ~1000 new hot pixels/day appear, now covering ~1.2% of each chip. Median dark current level is ~7 e-/pix/hr, increasing ~1.5 e-/hr/year. For comparison, the average sky background is ~200, ~3.6, and ~1.3 e-/pix/hr in the F606W, F225W, and F656N filters, respectively. See Baggett & Anderson for the backgrounds of other filters. Hot pixels controlled via anneals: Once a month, the detectors are warmed to +20C, fixing 80-90% of the hot pixels. Long-term growth in hot pixels is ~60/chip/day (Fig 1). Dithering removes remainder: Acquiring images in a dither pattern (small shifts between images) allows for removal of remaining hot or bad pixels during post-processing. References and Further Information \  WFC3 main page: www.stsci.edu/hst/wfc3 STScI help desk: help@stsci.eduwww.stsci.edu/hst/wfc3help@stsci.edu  For CTE advice: www.stsci.edu/hst/wfc3/ins_performance/ CTEwww.stsci.edu/hst/wfc3/ins_performance/ CTE  Anderson, J., & Bedin, L., An Empirical Pixel-Based Correction for Imperfect CTE. I. HST’s Advanced Camera for Surveys, PASP 122, 1035, 2010.  Anderson, J., et al., “Fitting a Pixel-Based Model to a CCD Detector: WFC3/UVIS and Low Background Issues”, in prep (2013).  Baggett, S., & Anderson, “WFC3/UVIS Sky Backgrounds ”, WFC3 ISR 2012-12.  MacKenty, J.W., & Smith, L. J., CTE White Paper, June 29,2012.  Noeske, K., et al., “ WFC3/UVIS CTE Monitor Results ”, WFC3 ISR 2012-09.  Massey, R., et al., “Pixel-based correction for Charge Transfer Inefficiency in the HST/ACS”, MNRAS 401, 2010. Signal losses from CTE degradation depend on 1) Distance from amplifier: more transfers=more traps 2) Signal level: higher fractional loss in faint sources 3) Image background: higher background fills traps 4) Observing scene: sources preceding target source can fill traps and make charge transfer more efficient Fig 5. CTE losses as a function of observation date and source flux (symbols: data, lines: model). The measured CTE losses as a function of MJD and signal level have been fit with 2-parameter polynomials. Separate fits are done for four different background levels. Fig 5 shows results for two levels. The derived fit coefficients are available to observers who wish to either estimate CTE losses in their data or apply them as a correction to their photometric results. Further details are available in Noeske et al. The efficacy of a higher background level is shown in the plot at right: flux losses as a function of background level (1, 9, 15 e- from left to right) for faint and bright sources (bottom to top, 250, 1600e- total within 3x3 pix). X-axis is the number of transfers. An empirical assessment of CTE losses is shown in Fig 3. The left panel shows the “truth” from a scaled-down deep exposure. The middle panel shows a short exposure of the same scene with low background and severe CTE losses while the right panel shows the same exposure, but with 16e- post-flash and marginal CTE losses. Fig 4. Post-flash illumination. Fig 3. Mitigation of CTE losses using a modest increase in background level. The WFC3 post-flash can be used to elevate the image background. For low-level sources on low backgrounds, this will increase the source signal faster than it increases noise. The illumination pattern (shown at left) varies by +/-20% across the full FOV but the pattern and intensity are quite repeatable (<<1% and <2%,). Fig 2. Measured flux losses as a function of background level, source brightness, and distance from the amplifier. CTE losses especially for faint targets can be significantly reduced via a relatively small increase in image background. Current recommendation: 12e- total (dark+sky+post-flash). Charge transfer efficiency One method of assessing CTE losses is via photometry of a stellar field placed close to and far from the readout amps. With CTE traps present, the target signals decrease as the distance from the readout amp increases. Fig 5 summarizes the WFC3 CTE losses as a function of source signal level and epoch: currently, bright sources suffer ~2- 4% losses while faint sources show ~10% and ~50% losses in low and high backgrounds, respectively. Mitigation option: post-flash Mitigation option: pixel-based correction A technique successfully applied to ACS data has been to use hot pixels in dark frames to empirically model CTE losses on a pixel- by-pixel basis (Massey 2010; Anderson & Bedin, 2010). The model is then applied to science images to restore charge to its original location in the image. WFC3 has acquired the necessary data for constraining the pixel- based model and work is underway to optimize the algorithm for UVIS. A standalone version should be available spring 2013. Fig 6 shows the results of applying a preliminary version of the pixel-based correction to a stellar image. Further refinements and testing of the model are underway (Anderson et al. 2013). A standalone version of the correction software is expected to be released for use in the spring of 2013. Fig 6. WFC3 image subsections farthest from the amp before (top) and after (bottom) correction. While the correction is expected to work well, the nature of the algorithm is such that it may not be able to completely recover what was lost, particularly at the faintest levels. In addition, in order to avoid amplification of read noise, the algorithm must be conservative in its reconstruction at the low background levels where losses are non-linear. ~1-2 e-/pix background ~20-30 e-/pix background “truth”no background16e- background Launch +31 daysNov 2012 Fig 1. Percentage of hot pixels (>54 e-/hr) as a function of time.