Photomask linewidth measurements are needed for vendor/buyer communication, for developing specifications and ensuring that products meet specifications, and sometimes for legal compliance. Linewidths are also measured for process monitoring for either optimization or stability, and for process modeling. The value of such measurements lies in their total uncertainty and in the user's confidence in the uncertainty estimate. Uncertainty in linewidth measurement (and in linewidth standards) is dominated by a systematic component, as much as 70 nm, due to edge runout or roughness of the chrome film which constitutes the line, and to a lesser extent from light transmitted through chrome/chrome oxide whose index of refraction is unknown. This edge roughness appears in SEM micrographs but is poorly resolved by the optical measuring instrument. Even so, the same poor quality photomask feature prints a clean and sharp line with well defined edges on photoresist because of the spatial low-pass filtering in the stepper optical system and of the nonlinear response of photoresist to light intensity. For some measurement purposes, such as process monitoring, it is the difference in linewidth from time to time during etching or from one unit to the next which is important, and not the absolute linewidth. In this case these major systematic uncertainty components can largely cancel out and much higher accuracy is possible. Measurement by topside or backside reflection can be useful here, but not for absolute measurement unless modeled. Measurement traceability generally is not needed here. The major components of the uncertainty of this differential measurement are usually the tool's precision and the accuracy of its scale. Other cases require an absolute measurement of linewidth. But, it is a feature's optical image which ultimately prints on the wafer, and an optical image which is measured in the metrology tool. A feature's image can be measured with far greater accuracy than its geometric width. For this reason photomask CD measurement accuracy can be greatly improved by emulating in the metrology tool the aerial image formed by the stepper, and measuring this emulated stepper image. The geometric linewidth can be determined from these data, but this is optional.
Proceedings of SPIE
September 16, 1994
Santa Clara, CA
14th Annual BACUS Symposium on Photomask Technology and Management, William L. Brodsky, Gilbert V. Shelden, Editors December 1994
aerial image, CD measurement, exposure emulation, linewidth, metrology, photomask