COS Delivers Sensitive Ultraviolet Spectroscopy

Rachel Osten, osten@stsci.edu, for the COS team

The Cosmic Origins Spectrograph (COS) is Hubble’s third-generation ultraviolet spectrograph. It was installed on Hubble on May 16, 2009, during the third extra-vehicular activity of Servicing Mission 4. COS is designed to perform high-sensitivity, medium- and low-resolution spectroscopy of astronomical objects in the 1150–3200 Å wavelength range.  COS significantly expands the spectroscopic capabilities of Hubble at ultraviolet wavelengths, providing observers with unparalleled opportunities for observing faint point sources of ultraviolet light.  The primary science objectives of the mission are the study of the origins of large-scale structure in the Universe, the formation and evolution of galaxies, the origin of stellar and planetary systems, and the cold interstellar medium.  The COS near-ultraviolet (NUV) and far-ultraviolet (FUV) detectors have now been released for science, and the early observations illustrate COS’s capabilities (Figures 1–2).

Figure 1. COS observation of supernova remnant N132D–P3, obtained with the G130M and G160M modes. The full spectrum is shown. Several ionization states of oxygen are observed, and emission from Si IV is detected for the first time near the O IV lines. The oxygen and silicon exhibit a common velocity. A velocity offset of the carbon species suggests that the carbon is spatially distinct from the O-rich knot. The supernova explosion occurred nearly 3000 years ago, and yet the COS spectrum detects material that still has not mixed with interstellar material. The geocoronal lines of H I and O I have been removed. Credit: Kevin France and James C. Green (CASA/University of Colorado).

Figure 1. COS observation of supernova remnant N132D–P3, obtained with the G130M and G160M modes. The full spectrum is shown. Several ionization states of oxygen are observed, and emission from Si IV is detected for the first time near the O IV lines. The oxygen and silicon exhibit a common velocity. A velocity offset of the carbon species suggests that the carbon is spatially distinct from the O-rich knot. The supernova explosion occurred nearly 3000 years ago, and yet the COS spectrum detects material that still has not mixed with interstellar material. The geocoronal lines of H I and O I have been removed. Credit: Kevin France and James C. Green (CASA/University of Colorado).

Figure 2. COS spectrum of absorption from intervening gas clouds along the line of sight towards quasar PKS0405–123. The exposure time was about one-fourth as long—and the signal-to-noise ratio is higher—as other UV spectrographs on the same target. This advantage has enabled the detection of 3–5 times as many lower-density filaments of hydrogen in the cosmic web than previously detected along this line of sight. Credit: Charles Danforth (CASA/University of Colorado), Keith Noll (STScI), James C. Green, Cynthia Froning, Brian Keeney (CASA/University of Colorado).

Figure 2. COS spectrum of absorption from intervening gas clouds along the line of sight towards quasar PKS0405–123. The exposure time was about one-fourth as long—and the signal-to-noise ratio is higher—as other UV spectrographs on the same target. This advantage has enabled the detection of 3–5 times as many lower-density filaments of hydrogen in the cosmic web than previously detected along this line of sight. Credit: Charles Danforth (CASA/University of Colorado), Keith Noll (STScI), James C. Green, Cynthia Froning, Brian Keeney (CASA/University of Colorado).

The Institute’s COS team, in collaboration with the COS Instrument Definition Team, led by Principal Investigator James Green of the University of Colorado, participated in the on-orbit verification and initial calibration of the instrument. By the end of September 2009, a total of nearly 2800 individual exposures in 415 (259 external) orbits of COS observations had been executed in the process of activating the instrument.

Several early COS programs were dedicated to characterizing the on-orbit performance of the FUV and NUV detectors. The measured FUV dark rate meets expectations, and the NUV dark rate is 11 counts cm-2 s-1, which is about three times lower than pre-launch predictions. Other programs obtained flat-field images and verified the ability to achieve high signal-to-noise ratios.

We are characterizing the on-orbit line-spread function (LSF).  The line-spread function quantifies the response of the mirror, grating, and detector to the wavelength of an input photon.  On-orbit observations indicate that the LSF has a narrow core with a non-Gaussian contribution extending into wings of many pixels. We will provide the observers with more quantitative information in the near future.

We have performed initial, on-orbit measurements of the sensitivity of the COS detectors. The sensitivities of all FUV modes and detector segments are within 10–20% of the values expected from ground testing. On-orbit measurements with the B segment (short wavelength segment) of the detector with the G140L grating have confirmed some modest sensitivity at wavelengths shorter than 1150 Å, because the reflectivity of the MgF2 coatings is still non-zero in that wavelength range. Assessment of COS sensitivity of this regime will be made available to the user community in updates to the instrument handbook (IHB) and instrument science reports (ISRs).

The measured sensitivities of most modes in the NUV range are within 10–15% of the values predicted from ground tests. Two of the NUV gratings displayed time-dependent sensitivity loss on the ground; initial on-orbit measurements of the efficiencies of these gratings are consistent with pre-launch trends.  It is expected that this loss of efficiency will not continue on orbit. Also, the NUV detector suffers from some vignetting when observing external targets, which can decrease the sensitivity by as much as 20% at one edge of the detector. Vignetting is uniform for all three stripes, and the effect is now included in the NUV flat-field reference file.

All the COS target-acquisition modes work as planned, and have demonstrated the ability to center the target to within 10–20 milli-arcseconds. For acquisitions using the bright-object aperture, we now recommend exposing to achieve a signal-to-noise ratio of 60, rather than 40 as originally prescribed.

We are delivering new reference files, updating the IHB, and preparing new ISRs to document the COS instrument as it commences science operations. Updates based on the on-orbit performance of the instrument will be made available through our web pages.

Calibration activities will continue through Cycle 17. We will monitor dark frames, sensitivity, internal/external wavelength scales, and NUV imaging performance. Also, we will continue to improve the calibration software and reference files.

Observers who obtain COS data prior to the completion of these on-orbit calibrations should be aware that they may need to revise their analysis once the final performance characteristics are determined. Areas of active investigation and characterization include the COS line-spread function, flat fields, and wavelength solutions. If you have any specific concerns about your COS data, please contact help@stsci.edu with questions.

We expect the Call for Proposals for Cycle 18 to be released on December 4, 2009.