On 10 January 1998 the Phoenix grating surface developed a series of fine surface cracks. This event happened suddenly during an observing run. We believe that this resulted solely from thermally induced stresses in the grating. No part of the spectrograph shows evidence of physical damage or contamination, and all the mechanisms were functioning correctly. At the time the grating may have been cooling slightly, but was close to the equilibrium value of 50 K, and no sudden temperature shocks had occurred. As a result of the surface cracks in the grating, Phoenix experienced a decrease in throughput by an order of magnitude combined with a dramatic increase in scattered light.
The grating in Phoenix is a replica on a silicon substrate. It has been known since the first time the grating was cooled to cryogenic temperatures that the grating substrate and the replication epoxy had sufficiently different coefficients of expansion to be incompatible. During the design the difference of the coefficients of expansion was roughly known (although somewhat underestimated). Silicon had been selected for the substrate because of its dimensional stability and high thermal conductivity. It was incorrectly assumed that the strength of the (thin) epoxy replication layer was low so that it would comply with thermal expansion and contraction of the the silicon substrate. On the first cool down, small coils of epoxy still bonded to flecks of the silicon substrate popped off the edges of the grating. This flecking reoccurred on every cool down and indicated that the epoxy did not have negligible strength compared than the silicon substrate. Since the damage had been limited to the mainly unilluminated edges of the grating, this had little impact on the performance. However, optical modeling this past year indicated that the differential expansion of the epoxy and silicon was also warping the grating, causing astigmatic images. For this reason the grating was scheduled to be replaced. Evidently on 10 January stress in the grating replication layer reached the point where an overall failure took place.
The grating will be replaced with a replica grating on an aluminum substrate. Aluminum has a coefficient of expansion that much more closely matches the replication epoxy than did the coefficient of expansion of the silicon. Aluminum also has a proven performance history, with aluminum substrate replica gratings having been used in other cryogenic spectrographs. The new grating should be available for the 1998B semester. A positive note is that the optical performance of Phoenix should be improved. The astigmatic images with the silicon substrate grating could be focused spectrally with the collimator focus, but were spread out spatially. Better spatial focus with the new grating will increase the sensitivity as well as improve the spatial mapping capabilities.
The Users Committee strongly endorsed sending Phoenix to CTIO in early calendar 1999. Therefore, the 1998B semester will be the last for which Phoenix will be available in the Northern Hemisphere for several years.