STIS First Order Spectroscopic PSF s. Ted Gull, Don Lindler Nick Collins Don Tennant Bob S. Hill Eliot Malumuth

Transcription

1 STIS First Order Spectroscopic PSF s Ted Gull, Don Lindler Nick Collins Don Tennant Bob S. Hill Eliot Malumuth

2 Example: Eta Carinae

3 Balmer Alpha

4 150 s Eta Carinae and Ejecta 52 x0.2 F1 (0.5 ) 10 s 0.3 s

5 Eta Carina A

6 HST s distinctive advantage: excellent angular resolution to ~0.1. Many astronomical sources have spatial structure of interest on the scale of 0.1 How do we obtain accurate spectroscopy of detailed structure at scales approaching HST s diffractive limit? What is the level of scattered light and can we correct for it? The Problem:

7 Measuring Instrument Contribution to PSF Reference fiducials were included in each long aperture format as aid to determine wavelength, aperture position. Grating select wheel encoded sufficiently to precisely position aperture with F1 or F2 fiducial near normal aperture center (done at the expense of small angle tilt of aperture) By placing a point source behind the F1 (0.5 width) or F2 (0.7 width), peak flux cut by at least 4 orders of magnitude. Likewise the scattered light internal to the instrument is cut, most noticably in the red and PSF outside of F1, F2 closer to HST PSF. Selected stars either coronagraphic standards or flux standard: HD HD HD BD+75D325 K G V A WD Standard

8 G750L Exposure

9 Fringes in CCD Malumuth, et al. PASP, accepted We must not ignore fringing effects when observing extended sources, especially with G750L grating

10 Example of Model-Fringe-Corrected Spectrum

11 HD4748 (Courtesy Angela Coterie) Observed Spectrum 1 offset Simulated Spectrum

12 Normalized to 1 at peak F1 rejection in core > 10 4 F2 (0.7 ) does not cut wings as this is ~PSF of Telescope. Outer wings suppressed further G V Coronagraphic Standard Slice for G750L

13 G750M Comparison of G750L for Spectral Type

14 Internal scatter is greatly decreased. Zero flux level is hard to determine as bias change across chip in y direction. Wings for F1 and F2 similar. Slice for G430L

15 Comparison of G430L for Spectral Type

16 As photons absorbed near surface of CCD, far less scatter. G V star faint, so measured wings are much less dependable. Slice for G230LB

17 Comparison of G230MB with Spectral Type

18 Comparison of G750L w/ G750M & 52x0.2 w/52x0.1

19 Comparison of G430L w/ G430M

20 Comparison of G230LB w/ G230MB (Limited data due to long exposure times.)

21 Scattered light vs wavelength: 52 x0.2 x0.2

22 Scattered Light with F1 (0.5 ) Fiducial

23 Scattered Light with F2 (0.7 ) fiducial

24 Ghosts and Ghoulies (and things that go bump in the night)

25 Window Ghosts in STIS Direct Imagery Ghosts due to reflection off of the CCD housing window. ~1.5% total flux

26 How the Window Ghost Affects a Point-Source Spectrum At Center of field, ghost is just below in Y, and shifted towards center.

27 Spectroscopic Ghosts

28 Geometry of Ghost Reflection

29 How do we get the best photometry at diffraction limited spatial resolution 1) dither (takes 2x 4x as long) 2) deconvolution? What about fiducial observation Requires grating dither - - not an option in the STIS planning. - - not obvious that it would indeed improve things How would one detect a planet spectroscopically with 10^-4 to 10^-8 flux of star?

30 Conclusions Fiducials F1 and F2 are very useful for rejecting the light of a point source. Currently only 52 x0.2 F1 is supported. These results (Proposal 8844) demonstrate that 52x0.2F2, 52x0.1F1 and 52x0.1F2 yield good science. They too should be supported. Beyond 6500A, CCD internal scattering is a major contributant. A good observing/ extraction method at the sampling/ diffraction limit of HST/STIS is not yet developed. You don t get sumthin for nuthin.

31 Acknowledgements: Charles Proffitt and the schedulers HST GO and STIS GTO funding Don Lindler, Keith Feggans, and Don Tennant (future astronaut) Bob S. Hill Ghosties and ghoulies and things that go bump in the night.. And below the steller spectrum Eliot Malumuth the Fringe Model- Maker

32

33

34 HST/STIS vs. VLT/UVES

35 Extractions: Bilinear vs quadratic also 7 pixel vs 3 pixel extraction.