Comprehensive model for projection photolithography: Rigorous, fast and novel processing

The central theme of this thesis is comprehensively modeling projection photolithography and identifying how process factors such as the mask, lens and resist impact the lithographic resolution. In addition, a novel electric-filed enhanced post exposure bake process that improves resolution is introduced. First a methodology of modeling line-end shortening (LES) effects in KrF 248nm resists is described, which combines optics and resist post exposure bake (PEB) simulation and matches the experimental data to better than 10% of the nominal linewidth. The results show that LES in KrF resists results from optical proximity and acid diffusion. Then a fast algorithm is developed for solving the non-linear partial differential equations in resist reaction/diffusion. Using high order partial differential expansion, this algorithm achieves a time step 100 times larger than the Finite Difference Method and is more than 10 times faster than Finite Element Method. An image processing algorithm is also developed which retrieves the physically fine contours of patterns from noisy Scanning Electron Microscope (SEM) pictures. Then contours are compared with those from simulation to effectively extract the model parameters.; A main contribution of the thesis is the extension of the comprehensive model using a divide-and-conquer strategy to ArF lithography of 120nm critical dimension (CD) with the emerging exposure tools and resists. Extensive linewidth measurements and SEMs were made at Texas Instruments and merged with data on lens aberrations and resist models from SEMATECH. For the 100nm node, when adjusting for mask error, the model fit through dose, focus and feature types and size (DFS) CDs within 5% for lines and matched the SEM areas of 2D patterns within 15%. The most important factors were image position in the resist and threshold. A strong DFS variation for the low light intensity images of small gaps was observed. Adding trajectory dissolution model allowed the gap prediction error of 25nm over 80 DFS cases. This error is large but reasonable given that the SEM LES error in 193nm resists is greater than 10nm.; This thesis also presents a novel electric-field-enhanced post exposure bake technique which uses alternating electric field to aggressively modify or control the profile of photoresist. In PEB process the presence of an alternating field leads to an order of magnitude greater arc length of the proton in the preferred direction than in the unwanted lateral directions. In experiments a resist profile improvement effect was clearly observed and the resist resolution was significantly improved. This technique is of practical interest and a patent application was filed by the University of California.