Nanoscale limits of angular optical scatterometry

Abstract

Angular scatterometry is a fast, in-line, noncontact, and nondestructive nanoscale metrology tool that is widely used in manufacturing processes. As scatterometry is a potential metrology technique for next generation semiconductor manufacturing and for other emerging large-area (roll-to-roll) nanotechnology products such as wire grid polarizers (WGPs) and nanostructured metamaterials, it is necessary to study its fundamental sensitivity and accuracy limitations. Two different samples are simulated using rigorous coupled-wave analysis. One is a high index contrast aluminum WGP structure, and the other is a low-index contrast resist grating on a polycarbonate substrate. During modeling, the sample structure is scaled by simultaneously scaling both the line width and the height of the grating with a fixed pitch and all linear dimensions, including pitch, line width, and grating height, of the structure. Two metrics are chosen to define the limits: the first is the comparison with experimental limits, that is, if the reflection difference for a 5% scaling variation is larger than the experimental noise floor, scatterometry has sufficient resolution to recover the metrology information; the second is the comparison with effective medium models, that is, if the simulated angular scatterometry signature differs from an effective medium model signature, again within experimental noise limits, scatterometry is judged to have sufficient resolution to determine the feature parameters. Using a 405 nm source, scatterometry provides sufficient information to analyze a 20 nm pitch WGP structure using a 405 nm laser source (wavelength/pitch = 20), while the minimum pitch resist grating is ∼24 nm (wavelength/pitch = 16.8).