The indispensable first step to getting the things you want out of life is this: decide what you want. - Ben Stein
Scatter specifications are for the most part the main point for this book. You need them in order to qualify parts and/or systems. You even need scatter specifications to build a scatterometer. And, to be appropriate, specifications need to address the issue at hand: they must be application specific. Generating the right specification requires knowledge of the system (or process) under design (or test), as well as knowledge of scatter measurement and analysis. The preceding chapters have presented the definitions and techniques for quantifying, measuring, and analyzing optical scatter. The issue is now approached from the other direction. How can meaningful scatter specifications be found?
The 1970s and 80s generated considerable concern over scatter in optical systems. Although it was often recognized that low-scatter optics were required for a given application, actual specifications were seldom given. The easiest, most available, and cheapest scatter measurement was the TIS. Most of the specifications written to handle scatter concerns are either TIS (usually given without angle or frequency limits) or rms roughness found from profile data and often given without spatial bandwidth limits. Surface roughness was often specified to control scatter, even though it was recognized that it would be difficult, futile, and sometimes impossible to attempt to relate the roughness parameter σ to actual component scatter. But at least the direction was right (no sign error), as smoother surfaces do generally mean less scatter. In the late 1980s, serious work began on BSDF standards in an ASTM committee, funded in part by the United States Air Force. The result (as described in the previous chapter) was a set of written standards that not only detailed measurement requirements, but also gave a data-format system. This enabled the easy transfer of data between laboratories, and increased the ease with which specifications could be written and checked. Unfortunately, the industry has made very limited use of these documents. Roughness, and occasionally TIS, are still by far the most common choice to specify low-scatter requirements.