Mask calibration dominated methodology for OPC matching

Abstract

The effectacy of the OPC model depends greatly on test pattern data calibration that accurately captures mask and wafer processing characteristics. The CD deviation caused by an off-center mask process can easily consume the majority of the lithography process CD budget. Mask manufacturing variables such as write tools' resolution, etch process effects, and pre-bias of the fractured data have great impacts on OPC model performance. As a result, wafer performance using masks from different mask shops varies due to variations in the mask manufacturing process, even if the masks are written with the same data set and use the same manufacturing specifications. A methodology for mask manufacturing calibration is proposed in order to make an OPC model consistent between two mask manufacturing processes. The methodology consists of two parts: mask manufacturing calibration and wafer-level OPC accuracy verification. The mask manufacturing process and metrology are calibrated separately. The OPC model is built based on the database of the first-party mask shop, and OPC verification is carried out by wafer data using the newly calibrated mask from the second-party mask shop. By checking wafer performance of both OPC model matrix items and complicated 2D structures, the conclusion can be drawn that different mask shops can share the same OPC model with rigorous mask calibration. This methodology leads to lower engineering costs, shorter turn around time (TAT) and robust OPC performance.