Optical Proximity Correction With Linear Regression

Abstract

An important step in today's Integrated Circuit (IC) manufacturing is optical proximity correction (OPC). In model based OPC, masks are systematically modified to compensate for the non-ideal optical and process effects of optical lithography system. The polygons in the layout are fragmented, and simulations are performed to determine the image intensity pattern on the wafer. If the simulated pattern on the wafer does not match the desired one, the mask is perturbed by moving the fragments.This iterative process continues until the pattern on the wafer matches the desired one. Although OPC increases the fidelity of pattern transfer to the wafer, it is quite CPU intensive due to the simulations performed at each iteration. In this paper, linear regression techniques from statistical learning are used to predict the fragment movements. The goal is to reduce the number of iterations required in model based OPC by using a fast, computationally efficient linear regression solution as the initial guess to model based OPC. Experimental results show that fragment movement predictions via linear regression model significantly decrease the number of iterations required in model based OPC, thereby decreasing the product development time in I.C. design and manufacturing.

Critical Review of Current Trends in ASIC Writing and Layout Analysis

“…The mask layout is adapted for the convergence of the final predicted image with the target pattern [38,39]. The performance of this procedure depends heavily on the model calibration and accuracy.…”

Fast pixel-based optical proximity correction based on nonparametric kernel regression

“…In the future, we may investigate the proposed algorithm using test layouts with smaller critical dimensions (CD). Similar to Gu et al, 18 the features on the training layout are separated into three classes of fragments: convex corners, concave corners, and edges. In Fig.…”

“…EBOPC decomposes the mask into edges and corners and optimizes their locations, while pixelbased OPC decomposes the mask into small pixels and optimizes their transmission coefficients. 18 Compared to EBOPC, PBOPC has more degrees of freedom during the optimization process, and may insert subresolution assistant features (SRAFs) around main mask features to further improve image fidelity.…”

“…34 Gu and Zakhor generalized this idea and proposed a method based on linear regression to provide an initial estimate for the movement of edges and corners in EBOPC. 18 Then, Gao et al improved upon Gu and Zakhor by using principle component regression. 35 These initialization strategies can effectively reduce the number of iterations required for the EBOPC algorithms to converge and thereby reduce the overall runtime.…”

“…35 These initialization strategies can effectively reduce the number of iterations required for the EBOPC algorithms to converge and thereby reduce the overall runtime. 18,35 These methods mentioned above opened a new line of research for efficient OPC solutions. However, it is desirable to use more advanced strategy to effectively improve the computational efficiency of PBOPC, which has many more degrees of freedom than EBOPC, for optical lithography systems.…”