Optical Proximity Correction, Methodology and Limitations

Since the early 2000's, model based Optical Proximity Correction (MB-OPC) has been used by the semiconductor industry to improve the linewidth uniformity and pattern fidelity in photolithography. Designed to be improved from its predecessor, the rule based OPC (RB-OPC), which relies on a table of biases to correct linewidth variation due to Optical Proximity Effect (OPE), it uses aerial image to calculate pattern edge deviation from the design. The flow of the MB-OPC includes the model data collection, model setup and calibration, recipe setup, OPC correction and post-OPC verify check. Since the OPC process also include the addition of assist patterns, such as serif, Sub-Resolution Assist Features (SRAF), and hammer heads, etc., the OPC process can also help improve lithography process window. Albeit above advantages, OPC can have modeling errors which may cause pattern failures and re-toolings. The modeling error is understood to basically originate from non-perfect physical modeling of the lithography process, which implies the need for better modeling. Better modeling includes better photoresist characterization and modeling, better Mask 3D (M3D) scattering effect modeling, and better developing process characterization, etc. It is also related to the quality of patterning process setup, such as the optimization of the substrate film stack, mask bias, and the photoresist process, and Reactive Ion Etch (RIE) bias, etc. Once the OPC model is optimally setup, the correction recipe setup will be less challenging and can focus on difficult areas, such as few line structures, complicated 2D features, etc. In this paper, we propose a methodology in model calibration and recipe setup and will provide recommendation on the effective use of optical proximity correction.