Fringing in the WFC3/UVIS detector

1. Fringing in the WFC3/UVIS detector Mike Wong UC Berkeley

2. Outline • Intro to fringing • Magnitude of fringing in the WFC3/UVIS filters • The fringe model • thanks: Eliot Malumuth • The ground test data sets • thanks: DCL staff, Howard Bond, Elizabeth Barker, S. Rinehart, Bob Hill, Bryan Hilbert, Howard Bushouse, Jen Mack, Ray Lucas, Megan Sosey, André Martel, Linda Dressel • Using data and model to solve for detector thickness • Future work: improvement and verification of fringing model solutions

3. Intro to fringing • Silicon grows transparent at long wavelengths • Multiple internal reflections • Interference effects (constructive/destructive) • strong sensitivity to wavelength • strong sensitivity to detector layer thickness • The curse becomes the cure: • Data: measure fringe patterns at multiple wavelengths • Model: determine thickness of detector layer • Model: predict fringe patterns for any wavelength or SED, create “fringe flats”

4. Monochromatic fringe flat TV3 data 977 nm

5. Filters affected by fringing WFC3 ISR-2010-04

6. Assessing fringe amplitude WFC3 ISR-2010-04

7. The fringe model • Model described in Malumuth et al. (2003 Proc. SPIE 4854, 567-576) • used to correct STIS slitless spectroscopic data • Solves the Fresnel equations: • continuity of electromagnetic field components across layer boundaries • multi-layer model • Model inputs: • light wavelength and incidence angle • layer thicknesses and roughnesses • layer indices of refraction (n + ik), based on composition

8. Model schematic Table: Malumuth et al. (2003)

9. Test data • DCL data • 2001-12-06 to 2001-12-12 • detector chips tested separately, not integrated • incident light angle 0° ± 1° • 146-151 (<0.5 nm FWHM) wavelengths/chip, nominally 700–1060 nm • TV3 data • 2008-03-28 to 2008-04-12 • detectors integrated into the instrument • flight-like incidence angle of 21° ± 1° • 77 (2-nm FWHM) wavelengths/chip, 845–990 nm WFC3 TIR-2010-01, ISR-2010-05

10. Test data • Basic processing • DCL chip 2, commanded wavelength = 760.26 nm • overscan/bias • flatfield • CR/hot pixels WFC3 TIR-2010-01

11. Test data Data for 1 pixel in Quad A, Bandpasses of UVIS filters affected by fringing WFC3 ISR-2010-05

12. Modeling the data • Single-pixel test data and best-fit model • Model tests 8000 thickness values and 6 sets of auxiliary parameters • Best fit minimized residuals between data and model WFC3 ISR-2010-05

13. Deriving thicknesses • For 1 pixel, best thickness minimizes residuals between model and data at all wavelengths • Problem: DCL and TV3 data sets give different answer !! WFC3 ISR-2010-05

14. Thickness maps • Thickness map based on TV3 data WFC3 ISR-2010-05

15. Reconciling TV3/DCL data sets • Order errors? No. • Basic processing, or normalization methods? No. • Errors in DCL and TV3 incident angles? No. • Anti-reflective coating index of refraction? No. • Wavelength error in DCL data? • Malumuth: DCL wavelengths could be off by 2–3 nm (But, no.) • comprehensive test of wavelength error provided surprising result... • actual wavelengths shorter than commanded wavelengths by about 20 nm • scale factor of 0.972 ± 0.003 gives best result

16. Optimal  determination • For this frame, com-manded  = 997.35 nm (black point) • Calculate whole-chip residuals between: • this DCL data frame at 0° incidence • 0° model with TV3-derived parameters • Minimum residual yields chip-averaged optimal wavelength, in this case 969.4 nm (red point) • Procedure repeated for each frame to get full spectrum of optimal vs. commanded wavelengths WFC3 ISR-2010-05

17. Optimal  spectrum • Strong systematic relationship between commanded and optimum wavelengths • Best parameterization: • constant scale factor at all wavelengths • higher-order fits not justified • scatter in data • lack of physical explanation for wavelength errors WFC3 ISR-2010-05

18. Constant scale factor... • Order errors cycle periodically through mean scale factor • This behavior is expected • Fun note:If opt / cmd =  , then:opt / cmd = ncmd /nopt • So finding a constant scale factor is like finding an error in the index of refraction for the DCL experiments...aerogel ?!?! WFC3 ISR-2010-05

19. Thickness maps • Thickness map based on DCL data WFC3 ISR-2010-05

20. Thickness maps • Map of TV3 thickness – DCL thickness • Difference is reduced by about a factor of 4 by wavelength correction to DCL data • Difference gradient indicates potential problem in our understanding of TV3 tests WFC3 ISR-2010-05

21. Test data format • In DCL tests, Chip 1 was rotated 180° (amplifiers at “bottom”) with respect to TV3 and flight configuration (amplifiers at “top”) • Something fishy in DCL tests would produce inverted thickness gradients between Chips 1 and 2

22. On-orbit test data

23. Cycle 17 calibration data to be collected in all filters affected by fringing Photometry in Omega Cen Data will allow comparison of TV3 and DCL models On-orbit test data is best way to verify fringe corrections extrapolated beyond ground test data range (11922 Sabbi, 12091 Wong) Ideas for new model solutions: Combine TV3 and DCL data together explored, but unlikely to be successful Create fringe models based on subsets of test data (Kalirai) may compensate for uncertainty in silicon index of refraction as a function of wavelength Incorporate ground flats in fringe-affected filters as additional test data wavelength-targeted approach Future work