| This thesis presents a technique for obtaining the surface geometries of patterned wafers from optical metrology data during microelectronic manufacturing processes. Our emphasis is on the use of appropriate optical modeling to extract pattern profile information, which includes feature depth, critical dimension (CD) and sidewall slope, from spectroscopic ellipsometry data.; Specular reflected light techniques, including both single wavelength and spectroscopic versions of ellipsometry and reflectometry, have been used for both etch and growth rate control for blanket wafers. However, use of these techniques for process control on products has been limited due to the problems inherent in the analysis of reflected light from patterned structures. In this thesis work, we examine techniques for the quantitative analysis of data from both highly regular grating structures and patterns with only local order.; We use rigorous coupled wave analysis (RCWA), a numerical solution to electromagnetic boundary value problems, to model the polarization dependent reflection from periodic patterns with micron or sub-micron feature size. We find good quantitative agreement of this vector diffraction theory to specular reflection data. Our simulations show that this technique can be applied to features in the deep sub-micron regime, and a better than 10 nm resolution can be obtained. However, the RCWA algorithm is computationally intensive, especially for structures with large linewidth. Moreover, it is inherently limited to the solution of periodic structures.; We use a much simpler scalar method for modeling reflection from patterns with large linewidth, and discover that while simple and fast, this method can yield accurate simulation for structures with feature size above 10 μm. We also find that the limited spatial coherence length of the broadband probe light should be taken into account in the model for the analysis of the data from relatively large structures. We finally extend the application of the vector analysis to some non-periodic structures with local order by combining RCWA with the scalar model.; We conclude that this reflection-based technique is of significant promise for in situ IC production applications. |
