A fundamental problem in all designs was the changing order separation (like a step-ladder with the rungs getter further apart). The net effects of this are the reduction in slit length and the inefficient use of array area. This was solved in the prism designs by using two prisms made from different materials. The dispersion of the two materials are matched to even-out the the order spacing. thereby allowing longer slits (or more orders on the array - wider wavelength coverage for a given resolving power).
The best combination of materials for the 0.8-2.5 micron cross-dispersed mode was found to be ZnSe and Fused Silica. The best combination of materials for the 3-5.5 micron cross-dispersed mode was found to be Ge and LiF.