1. Calibration of the Cosmic Origins Spectrograph
James C. Green
University of Colorado

2. The CU COS Instrument Team
Project Scientist: Cynthia Froning
Instrument Scientist: Steve Osterman
Steve Penton
Stéphane Béland
Detector Scientist: Jason McPhate (Berkeley)
Gone – but not forgotten
Erik Wilkinson
Ken Brownsberger
John Andrews
Jon Morse
Dennis Ebbets
Dave Sahnow

3. COS has 2 channels to provide low and medium
NUV MAMA
Detector
(STIS spare)
Calibration
Platform
OSM2: G185M, G225M,
G285M, G230L, TA1
FUV XDL
Detector
Aperture Mechanism:
Primary Science Aperture,
Bright Object Aperture
OSM1: G130M,
G160M, G140L,
NCM1
COS has 2 channels to provide low and medium
resolution UV spectroscopy
FUV: Å, NUV: Å
FUV gratings: G130M, G160M, G140L
NUV gratings: G185M, G225M, G285M, G230L
M gratings have spectral resolution of R ~ 20,000
Optical bench
(not shown):
re-use of GHRS
bench

4. COS FUV Spectroscopic Modes
Nominal Wavelength Resolving Power
Grating Wavelength Range (R = l/Dl) b
Coverage a per Exposure
G130M Å Å 20, ,000
G160M Å Å , ,000
G140L Å > 820 Å
a Nominal Wavelength Coverage is the expected usable spectral range delivered by each grating mode. The G140L grating disperses the Å region onto one FUV detector segment and Å onto the other. The sensitivity to wavelengths longer than 2050 Å or shorter than 1150 Å will be very low.
b The lower values of the Resolving Power shown are delivered at the shortest wavelengths covered, and the higher values at longer wavelengths. The resolution increases roughly linearly between the short and long wavelengths covered by each grating mode.

5. * N2 purge data through FUV detector door window.
* Portion of FUV detector flat-field obtained during component-level testing.

6. COS NUV Spectroscopic Modes
Nominal Wavelength Resolving Power
Grating Wavelength Range (R = l/Dl) b
Coverage a per Exposure
G185M Å x 35 Å 16, ,000
G225M Å x 35 Å 20, ,000
G285M Å x 41 Å 20, ,000
G230L Å (1 or 2) x 400 Å
a Nominal Wavelength Coverage is the expected usable spectral range delivered by each grating mode, in three non-contiguous strips for the medium-resolution modes. The G230L grating disperses the 1st-order spectrum between Å along the middle strip on the NUV detector. G230L also disperses the Å region onto one of the outer spectral strips and the Å region onto the other. The shorter wavelengths will be blocked by an order separation filter and the longer will have low thruput on the solar blind detector. The G230L 2nd-order spectrum between Å will be detected along the long wavelength strip.
b The lower values of the Resolving Power shown are delivered at the shortest wavelengths covered, and the higher values at longer wavelengths. The resolution increases roughly linearly between the short and long wavelengths covered by each grating mode.

7. NUV spectra projected onto the NUV MAMA detector
Increasing Wavelength
Point source Spectra
Calibration Stripes
25.60 square
2.80
NCM3a
NCM3b
NCM3c
1.45
2.25
5.75
4.95
All dimensions in mm
NUV spectra projected onto the NUV MAMA detector

9. Wavelength Accuracy Target Acquisition
Extragalactic moderate resolution programs generally require absolute wavelength accuracy of ~ +/- 1 resel ( = +/- 15 km/s), with relative accuracy of 1/3 resel rms across the spectrum.
Some programs that require higher accuracy can use “tricks” to obtain needed calibration - e.g., using known wavelengths of ISM lines along sight-line.
The aberrated HST PSF centered in the COS Primary Science Aperture.
Target Acquisition
COS is a “slitless” spectrograph, so the precision of target acquisition (placement of target relative to calibration aperture) is the largest uncertainty for determining the absolute wavelength scale.
Goal is to center targets routinely in science apertures to a precision of
+/- 0.1 arcsec (= +/- 10 km/s).
Throughput is relatively insensitive to centering due to large size of science apertures; centering of +/- 0.3 arcsec necessary for >98% slit throughput.

10. Fundamental Design Priorities
1: Sensitivity
2: Bandpass
3: Spectral Resolution

11. High Sensitivity
High throughput achieved through a minimum of reflections - ion etched blazed holographic gratings - windowless, opaque photocathode (NUV)
COS throughput measured by direct comparison to RASCAL reference detector – effective areas calculated by multiplying by HST area and reflectivity

13. For Comparison - COS Throughputs: G130M @ 1216 Å (peak) = 19
For Comparison - COS Throughputs: 1216 Å (peak) = Å = 9.0% 1430 Å = Å = 5.8% 1248 Å = Å = 0.65% Å = Å = 2.3% G Å = 2497 Å = 2.8% 2659 Å = 2998 = 1.0% 1846 Å = 2998 Å = 0.8%

14. Spectral Resolution
Spectral Resolution determined by measuring the FWHM of best Gaussian fit to emission lines:
FWHM (m) * dispersion (Å/m) = 
/ = Resolution
This technique will underestimate the true resolution if the emission line is not monochromatic

20. COS has an imaging mode
TA1 : implemented for target acquisition and correction of initial installation errors
Broad Band Imaging in the NUV – no filters – response dictated by the MAMA QE curve

24. COS Imaging Plate scale = 0.024″ / pixel
FWHM resolution (radial)  2.7 pixels = 0.064″ (FWHM is ~ 78% encircled energy)
Field of View = 2.5″ - 3″ (fuzzy edge)
(40 resol diameter pixel diameter image)
For comparison, ACS WFC/HRC has 80/82% encircled energy in 0.25″

25. COS Flat Fielding
NUV MAMA – classic flat filed techniques yield high signal/noise capabilities

28. COS Flat Fielding
FUV detector show hex pattern consistent with MCP manufacturing process –
Stable geometric distortion map can be developed and applied
Classic flat field techniques do not fully remove the structure - redistribution

30. COS Bonus Feature Potential capability from 912 - 1150Å
G140L places these wavelengths on the detector with spectral resolution of ~ 1000
Grating efficiency at 1048Å measured – implying cm2 effective area at 1048Å – assuming that HST mirror coatings behave as grating coating