Routine monitoring of the COS spectrographic sensitivity has shown that the throughput of a number of modes is declining over time. This includes all COS FUV modes as well as the bare aluminum NUV G225M and G285M gratings. In contrast, throughputs for the MgF2 coated G230L and G185M NUV gratings appear not to have changed. The mean trends derived for a number of routinely monitored central wavelength settings are shown in the figure. The G225M and G285M gratings are uncoated bare Aluminum gratings. Some degradation in the throughput of these gratings had been observed prior to launch and had been attributed to the formation of a surface oxide layer. It had been hoped that this degradation would cease on-orbit. However, periodic monitoring observations of external standard stars show that the on-orbit rates of throughput decline appear to be as big or bigger than the prelaunch estimates. The throughput of the G225M grating is dropping at a rate of about 3 %/yr, while the G285M's throughput is decreasing at about 10 %/yr. For each of these gratings, the throughput decline is roughly constant with wavelength. In contrast, the throughput of G185M and the first order parts of the G230L spectra have been changing at a rate of less than 1%/yr. Uncertainties for the stripes that contain the 2nd order G230L spectrum are larger. For the FUV modes, the rate of change appears to depend primarily on wavelength. The throughput is declining at about 3%/yr for wavelengths shorter than 1400 Angstroms, and by as much as 11%/yr at 1800 Angstroms. The possible causes of this decline are still being investigated, but the trend with wavelength appears to be consistent with some kind of degradation or contamination of the CsI photocathode that makes it more difficult for lower energy photoelectrons to make their way out of the photocathode material and into the micro-channel plate pores. The COS FUV detector, unlike the STIS FUV or ACS SBC detectors uses an unsealed windowless tube, and this may expose the photocathode to additional sources of contamination. Monitoring of these modes continues, but, given the limits on the visit-to-visit repeatability imposed by pointing and focus uncertainties, it is not yet possible to say whether the rate of throughput decline for any of these modes is slowing with time or not. A more detailed description of the throughput measurements and the derived trends will be shortly reported in a COS Instrument Science report by Osten et al (2010). Version 18.2 of the COS Exposure Time Calculator (ETC) has been updated under the assumption that these trends can be extrapolated to the middle of Cycle 18 (March 2011). While corrections for these time dependent sensitivity changes are not yet included in the CALCOS pipeline, in the near future, updated reference files will be delivered that to allow CALCOS to apply these corrections to the extracted fluxes. Note that even taking into account these projected decreases in throughput, the COS FUV sensitivity is still excellent, and COS should remain the instrument of choice for most spectroscopic observations of very faint FUV targets.
![](cos_tds_trends.jpg)
Figure 1: The mean rate of throughput change for each of the routinely monitored COS central wavelength settings is shown here. The vertical error bars show the uncertainty in the rates, while the horizontal bars show the range covered by that spectral segment or stripe.