COS2025: A strategy to extend the lifetime of the FUV detector on the Cosmic Origins Spectrograph

The Cosmic Origins Spectrograph (COS) has been collecting data since soon after its installation on the Hubble Space Telescope in May 2009. The two-segment microchannel plate-based detector for the FUV channel is subject to gain sag, and eventually a permanent loss of efficiency at the locations where the largest number of counts have fallen. The initial strategy for the use of the detector was to maximize the scientific productivity of the instrument over the five-year design lifetime. This has been accomplished by periodically adjusting the high voltage and moving the spectra to a different location on the detector in order to spread the damage on the detector and thus minimize gain sag. The instrument is now well past this point and performing well, so in late 2016 we began to investigate ways to extend the life of the detector for as long as possible without seriously affecting the scientific performance. As a result of these studies, we adopted a new lifetime extension strategy when the spectra were moved to Lifetime Position 4 (LP4) in October 2017, and placed restrictions on the G130M observing modes that put Lyman-a airglow lines on the detector. Central wavelengths 1300, 1309, 1318, and 1327 are no longer permitted to illuminate Segment B of the detector, and G130M/1291 is permitted on that segment only for FP-POS values 3 and 4 in order to concentrate the damage to the detector. These changes limit the damage from gain sag " holes" due to airglow to only two locations on the detector, rather than the twenty at the previous LPs. In addition, we modified our previous approach of increasing the high voltage or changing LPs before any hole experienced a sensitivity loss of 5%, and will permit the two G130M/1291 holes to become permanently sagged, thus creating a new detector gap. Science programs that require wavelength coverage near the rest-frame Lyman-a region (1216 angstrom) can be executed at LP3. Observations with other gratings remain unaffected, but the wavelength coverage on Segment B for a single exposure will now include gaps due to the gain sag holes from the G130M/1291 observations. Models of gain loss as a function of exposure suggest that by adopting this strategy and giving up these small regions of the detector, we will be able to use LP4 productively for six or more years, as opposed to the similar to 2.5 years that was available at previous positions.