You wouldn't think space would be a problem in a spectrographic telescope the size of a small car. That is, until you try to pack 22 infrared mirrors inside into a precise optical path. But for National Optical Astronomy Observatories researchers, the real challenge came not from arranging the multiple mirrors, but from minimizing scattering and thermal deformation.AlumiPlate plating enabled the mirrors in this new 8-m spectrograph atop Cerro Pachon in Chile to provide surface roughnesses as low as 30 Å root mean square (rms). Also, researchers at the National Optical Astronomy Observatories noted that the mirrors' optical surface figure changed less than 0.05 wave rms during temperature cycling between 65 K and 300 K. Courtesy of National Optical Astronomy Observatories. After unsuccessful attempts to circumvent these problems by shaping or contouring the backs of the mirrors, making them thicker and sampling complicated electroless-nickel-balancing techniques, the researchers opted to use aluminum-plated mirrors from AlumiPlate Inc. that were diamond-turned by II-VI Inc. The two 8-m spectrographs are under construction atop mountains in Chile and Hawaii as part of the groups' Gemini project. They will perform an estimated 40 percent of the project's observations, according to Daniel Vukobratovich, the senior engineer developing the mirrors. "They're very versatile," he said. "Sort of the Swiss army knife of spectrographs." Specifications for the mirrors in these instruments required a surface roughness of less than 50 Å root mean square (rms) to keep scattering losses below 1 percent per surface. Vukobratovich evaluated the surface finish of two diamond-turned sample mirrors with an AlumiPlate layer: a 50-mm sphere and an 80-mm parabola. The surface roughness of the 50-mm sphere was about 30 Å rms -- about two to four times better than sampled electroless- nickel-plated Al6061 mirrors. Similarly, AlumiPlate's 80-mm parabola had a surface finish of about 40 Å rms. Vukobratovich also found that AlumiPlate's material more closely matched the coefficient of thermal expansion of the mirrors' Al6061 substrate. This allowed him to avoid surface deformation and figure loss common in aluminum mirrors plated with nickel, which has a coefficient of thermal expansion approximately half that of Al6061. The AlumiPlate mirrors evaluated showed a change in optical surface figure that was less than 0.05-wave rms during temperature cycling between 65 and 300 K, he said. While Vukobratovich emphasized that AlumiPlate's aluminum plating and II-VI's diamond-turning process were essential to his project's success, he noted that the high purity of the material makes its surface very soft. Consequently, mirrors can't be polished and can be difficult to clean. According to AlumiPlate, some manufacturers have enabled the surface to be cleaned by applying a wear-resistant coating over it.