News for the Cosmic Origins Spectrograph

Justin Ely,, for the COS team

The Cosmic Origins Spectrograph (COS) on Hubble continues to produce exciting science and accounts for nearly 18% of all guest-observer prime orbits allocated in Cycle 19. With such an abundance of interest and data, the calibration and characterization of the COS detectors continues at a fast pace in order to provide optimal science.


Gain sag

The COS far-ultraviolet (FUV) channel employs a type of micro-channel plate detector known as a crossed delay line (XDL).  This type of detector converts each UV photon into a shower of electrons, for which the detector electronics calculate the x and y coordinates and the total charge in the electron cloud, or the pulse height amplitude (PHA).  Because XDL detectors need to operate at significantly higher gain than traditional micro-channel plate detectors, and thus each photon event extracts much more charge, prolonged exposure to light causes the detectors to become less efficient in converting photons to electrons.  This is a phenomenon called “gain sag.” As a result, the peak of the PHA distribution slowly decreases. As it approaches the minimum threshold imposed by the COS data-reduction pipeline, CalCOS, target photons may be rejected as background events. Eventually, photons may be permanently lost. Gain sag appears first in the regions of the detector illuminated by bright airglow lines, as higher count rates translate into accelerated gain sag, but eventually it affects the entire spectrum.

An additional consequence of gain sag is less accurate registration of photon events in the cross-dispersion (XD) or y direction, commonly referred to as “y walk.” While y walk does not adversely affect the extracted one-dimensional science spectrum, it can reduce the accuracy of target acquisitions obtained in dispersed light—the PEAKXD mode. If the target is centered in the aperture but the y walk shifts its spectrum in the XD direction, then the PEAKXD algorithm will miscalculate its centroid and move the target away from the aperture center. Without a correction, we estimate that FUV dispersed-light target acquisitions would mis-center the target by ~0.4–0.5 arcsec.  To prevent this, the COS team has implemented a change in the PEAKXD algorithm for acquisitions.  The algorithm now ignores data from one of the two FUV detector segments, Segment B, when computing the spectral centroid.  This change is advantageous because, due to the numerous Lyman-alpha airglow lines from observations taken with the G130M grating, the gain sag on Segment B is worse than on Segment A.  As an additional change, the flight software now shifts the computed centroid up two pixels to account for the y walk on Segment A.  Cycle 18 programs have been checked and modified as necessary to execute with this change, and for Cycle 19, version 19.2 of the COS Exposure Time Calculator (ETC) computes the signal-to-noise ratio (SNR) for segments A and B separately for PEAKXD target acquisitions.

A correction for the y walk caused by gain sag has also been added to the upcoming version of CalCOS, which applies a correction to the y location based on the PHA for each photon event.  This correction assumes a linear relation between the observed PHA and the y location of an event. The early results are quite encouraging.  The effect of the correction on a highly sagged region of the FUV detector Segment B is shown in Figure 1.

Figure 1

In addition to correcting data that have already been affected by gain-sag, considerable effort is ongoing to decrease the impact on future data.  As a first step to suppress the effects, on March 6, 2011, the FUV high voltage (HV) level in Segment B was successfully increased.  This increase alleviated much of the effects of gain sag due to both bright airglow lines and the continuum itself.  Analysis of data taken at the new HV value indicates that no recalibration is needed to accommodate this change, and that future adjustments to the HV may be used to mitigate gain-sag effects on the detectors. Nevertheless, there are limits to the amount by which the HV can be raised, and the long-term solution to gain sag will be to move the spectrum to a new position in the XD direction.  Preparations are currently underway to make this move, which will involve a recalibration of the instrument, by mid-to-late 2012.  Between such position changes and HV adjustments, we expect to significantly extend COS’s operational lifetime.