We present the characterization of optical proximity effects and their correction for 0.18 mu m deep-ultraviolet (UV) lithography processes using a semiempirically derived "behavior" model. Since critical dimension (CD) measurement data for deriving the model are taken after resist pattern transfer into the underlying layer (alpha-Si), the model incorporates all of the different proximity effect contributors: optics, mask, resist bake, etching, etc. The modeling technology we use allows for user defined model forms. It was empirically determined that the CD behavior could be adequately described by the diffusion of the aerial image with one Gaussian, and an adjustment to the signal threshold based on signal slope. The validity of the model for random geometry was confirmed by comparing contours drawn on Prospector(TM) with two-dimensional configurations of the uncorrected and corrected parts of a 0.18 mu m test circuit. The model is then used for proximity effects correction of the gate level of a 0.18 mu m test design and for the feasibility study for printing 0.15 mu m features using deep-UV (lambda=248 nm) lithography. Also, some process window enhancement techniques are evaluated in the correction recipe to enlarge the limited process window for printing 0.175 mu m features.