Authors
Brian R. Smith
Topics
Surface Roughness and Optical MeasurementsOptical Coatings and GratingsAdvanced Measurement and Metrology Techniques View Online Export CitationCrossMarkRESEARCH ARTICLE | SEPTEMBER 09 2005Brian Smith; Scott Bushman; Chris BaumAIP Conf. Proc. 788, 437–441 (2005) https://doi.org/10.1063/1.2063000Texas Instruments, Inc., 13560 N. Central Expressway, MS3736, Dallas, TX, 75243. Tel: 972-995-2240, Fax: 972-995-6383, E-mail: b-smith8@ti.comAbstract. Scatterometer-based control in semiconductor processing requires careful thin-film model selection to obtain the best repeatability, reproducibility, linearity, and bias. A “working” standard based on process capability is used to gather data that is then re-matched with different models for comparison. This study provides a framework for model comparisons and examines the effects of the index of refraction and model complexity, or the number of trapezoids, on measurement system capability for control of a polysilicon gate etch process. Measurement capability for the gate profile is needed in addition to gate length. Results from a single-wavelength, multiple angle-of-incidence scatterometer are compared to reference data from atomic force and transmission electron microscopy. The polysilicon index of refraction has a strong effect on measurement scaling and repeatability. A counter-intuitive result is that a single trapezoid model can perform better than a double-trapezoid model for linearity and bias. Keywords: scatterometry, statistical-process-control, gauge study PACS: PACS numbers: 07.79.Lh, 07.85.Jy, 06.30.Bp
About
PublishedJan 1, 2005
Citations1
References1