摘要
光学邻近校正（OPC）技术已经成为纳米级半导体工艺技术中的一个关键.目前在OPC中多边形的切分算法均基于配方（recipe）,但随着特征线宽减小及版图越来越复杂,用于切分的配方难以覆盖所有的情况;不完备的配方引发或加剧了芯片上的纹波、断线和桥连等现象.论文提出了一种新的基于光刻模型的动态自适应切分算法,根据不同的光刻模型和几何环境可以给出不同的切分,并且可在校正循环中动态改变切分方式和采样点的放置位置.通过90nm工艺下版图设计的验证,这种切分不仅减少了被切分出的小线段（segment）数量的10%-15%,节省了调试切分规则的时间,而且提高了OPC的质量,使PRV（post RET verification）错误率降低了35%.
optical proximity correction （OPC） is a key step in nanometer scale lithography technology. Currently,dissection in OPC is recipe-based. However, as the critical features shrink and the layout becomes more complicated, it is hard to debug and handle all possible cases in the layout ; Incomplete recipe-based dissection will introduce or worsen the effects of ripple, breaking, bridging, and lineend shortening. This paper presents a new dissection method,which is lithographic model-based and can dynamically change the dissection and sampling point positions during OPC correction loops. According to experiments on 90nm designs, the new dissection method can reduce 10%-15% of segments, save considerable time during recipe debugging,improve OPC quality, and reduce hot spot errors rates by 35%.
