Why must measurement system analysis (MSA), i.e., gauge study, be redefined? The existing literature does not make clear which metric should be used to compare measurement system performance against the requirements or specifications. The requirements are straightforward to determine because they are either pulled from the Metrology ITRS roadmap, which typically specifies the requirements for only the most challenging situations, or they are derived from the target value or process tolerance of the given application. Expanding on the latter point, a good rule of thumb is to seek 1% of the target value or 10% of the process tolerance, ideally whichever is smaller. 2% of the target value or 20% of the process tolerance (whichever is smaller) is acceptable for automated process control, but continuous improvement should be pursued to achieve 1% of target or 10% of the process tolerance over time.
As for which metric to use to compare against the requirements, there are two scenarios to consider. For a single tool, the recommendation is to use the precision. For a fleet of n tools, where n.1, the answer is not obvious - should each tool’s precision be used? What about the offset between tools? This lack of clarity drove the creation of a new metric. It is desirable to represent a fleet of tools with one value that captures the measurement variation of the entire fleet; this value is called the fleet measurement precision (FMP), and it should be used to compare against requirements when n.1. A fleet of tools is used to measure the various processes in most situations. Very rarely is one measurement tool used to control a given process except in certain circumstances, such as an integrated metrology module on a process tool, or when there is no other way to meet the requirements other than with a single tool. The FMP is the key metric used to describe the entire fleet, whereas the tool-matching precision (TMP) describes each tool in the fleet. TMP is a composite metric created from each tool’s precision, offset, slopeinduced- shift offset (SISoffset), and nonlinearity. This breakdown is different than the existing literature and previous chapter, which use terms like random and systematic variations but do not go far enough to better categorize the contributors. By categorizing the components in this manner, the assignable cause is easier to determine. The metrology toolset owner or supplier typically knows which “knobs” affect a given component(s) that is the dominant contributor. An offset issue is likely explained by a handful of “knobs,” whereas a precision issue is likely explained by a different set of “knobs,” and so forth. The TMP metric is a fundamental building block for improving the measurement performance of the fleet. It provides critical diagnostic information about where to focus efforts to improve the FMP, which is especially important when the requirements are not being met.
Online access to SPIE eBooks is limited to subscribing institutions.