Correcting STIS Point-Source CCD Spectra for CTE loss

1. SPACE TELESCOPE SCIENCE INSTITUTE Correcting STIS Point-Source CCD Spectra for CTE loss Operated for NASA by AURA Paul Goudfrooij* Space Telescope Science Institute • Methods to measure CTE of STIS CCD • (Visual) effects of CTE loss • CTE correction formula for point-source spectroscopy • Plans for the near future * With contributions of Ralph Bohlin, Jesús Maíz-Apellániz, & Randy Kimble

2. A Brief History of STIS CTE Measurements CTE: “Fraction of charge transferred per pixel transfer”(< 1 due to traps in the silicon). Typically, one measures CTI  1 – CTE. • Significant parallel CTI discovered on STIS in 1998 (Gilliland, Goudfrooij & Kimble,1999, PASP, 111, 1009). • Apparent non-linearity, most significant for faint sources. • Review of effects of radiation damage on STIS CCD: Kimble, Goudfrooij & Gilliland, 2000, Proc. SPIE, vol. 4013, p. 532. • Empirical determination of functional form for imaging CTE: Goudfrooij & Kimble, 2002, in Calibration Workshop Proceedings • First empirical determination of functional form for spectroscopic CTE: Bohlin & Goudfrooij, STIS ISR 2003-03 • Alternative determination using modelinvolving published parameters of ‘known’ traps (Bristow, STIS-CE ISRs, 2002-2004)

3. Amp D Nominal Readout Direction Axis2 (Y) Nominal Clocking Direction 1 2 (fluxD / fluxB) Y CTI ~ Parallel (virtual) overscan Amp A Amp B Amp C Axis1 (X) Serial overscan Serial overscan STIS CCD Architecture; Measurement Method STIS CCD: • 4 Readout Amps (1 / corner) • Nominal Amp: D (lowest RN) • Bi-directional Clocking yields CTI  1 – CTE: Measured using “Sparse Field Tests”

4. “Internal” Sparse Field Test “Sparse Field” Test • Sparse fields to ensure that sources do not overlap, in which case (e.g.) PSF wings would fill traps for sources along the readout direction • Typical situation for point source spectroscopy • Lamp images along narrow, cross-dispersion slits, projected at 5 positions along columns (or rows) • Representative of “worst-case” point source spectroscopy (essentially no background to fill traps) • Test conducted before launch and throughout STIS life

5. Visual effects of CTE loss: Dependence on Y position “External” Sparse Field Test: Outer Field in NGC 6752 (Oct 2001) Exp. time = 20 s

6. Visual effects of CTE loss: Effects of Intensity and Sky “External” Sparse Field Test: Outer Field in NGC 6752 (Oct 2001) Exp. time = 20 s Exp. time = 100 s

7. “Internal” Sparse Field Test Visual effects of CTE loss: Centroid shifts • Charge trailing and centroid shift measurable; most significant at low signal. • Impacts ‘shape’ measurements (surface photometry), especially for faint objects.

8. Spectroscopy Imaging CCD Column Number CCD Row Number Functional Dependence on Signal and Background Levels • To be done separately for imaging and spectroscopy: • For given “extracted” signal level, spectroscopy has higher “per column” signal and hence: • Lower CTE loss • Different dependence on background level

9. Internal Sparse Field Test: CTIAnalysis • Clear signatures of CTE effect: • Fractional loss decreasing with increasing signal • Absolute loss increasing with increasing signal • Measurements identical every epoch (and one measurement before launch) • Hence used to derive time constant of CTI degradation B amp / D amp ratio  Rows from B amp 

10. Insights from Monitoring of Spectrophotometric Standards • Wide-slit spectra of flux standard AGK+81D266 taken every few months • Division of (early, middle, late) data by overall average reveals CTE effect (even for such high-S/N spectra) • CTI dependence on background B and gross signal G modeled (in spring 2003) as CTI  G–a exp (–b(B/G)c)

11. 0.21 Insights from Monitoring of Spectrophotometric Standards Other important constraints on functional form of CTI: – Observed flux ratios between spectra of flux standards: • MAMA/G230L vs. • CCD/G230LB or CCD/G430L – Functional form producing best fit to the data as of Spring 2003 (cf. ISR 2003-03 and implemented in CALSTIS pipeline): Example: LDS 749B, 90 – 1800 e– B G CTI = 0.0355 G–0.75 (1 + 0.243 (t –2000.6))  exp (–2.97 )

12. Final Close-out Effort • 2003 solution overestimated CTI somewhat for G750L spectra redward of about 8500 Å • Likely due to presence of extended PSF (“red halo”), which acts as effective “extra background”, filling traps • Working on implementation as follows: CTI  G–a exp (–b (cB+H)/G)d) [where H is the ‘halo’ signal in the PSF beyond (above) the standard 7-pixel extraction box for point source spectra] • H values to be calculated by pipeline, using available encircled energy tables as function of wavelength (the PCT reference files)

13. Improvement to CTI corrections (P. Goudfrooij & R. Bohlin) • Inclusion of ‘halo’ component yields significant improvement beyond 8000 Å

14. Future Plans on Empirical CTE corrections • Finalize correction formula for spectroscopic data • Write ISR (Nov 2006) • Implement in CALSTIS pipeline (Spring 2006) • Write IRAF (post-observation) script to correct imaging photometry of compact sources for CTI (Spring 2006) • Analysis of extended source CTI measurements (Spectroscopy & Imaging; Spring 2006)

15.  bckgr counts External Sparse Field Test: CTIAnalysis Clear dependence on background level (“sky”) • Slope systematically flatter with increasing flux • “Sky” presumably fills traps in bottoms of potential wells, mostly affecting transfer of small charge packets. • Suggests CTI  exp –