Curvature based fragmentation for curvilinear mask process correction

Curvilinear mask process correction: status quo and outlook

“…Figure 7 shows this in more detail, where fragment lengths of the original input polygon (shown in Figure 7a) change abruptly and are not strongly correlated to curvature. We have shown that this impacts correction accuracy and runtime negatively [15]. Also, note that the total number of fragments in the input polygon is large and fragment lengths change abruptly with low correlation to local curvature.…”

“…Basic details of the curvilinear mask process correction and verification steps have been presented extensively in our previous work [1,13,14,15]. Curvilinear masks are beneficial not only for DUV masks but also for EUV masks.…”

“…This presents the need for innovative approaches that can potentially reduce the CLMPC computation complexity without any adverse impact on the correction accuracy of the flow. Recently, we have developed two such approaches, namely curvature-based pre-bias (CBB) and curvature-based fragmentation (CBF) and demonstrated promising results which will be summarized next in this section [14], [15].…”

“…This is because everywhere, where the edge movement and shape distortion from CLMPC is large, fragment density is high and can follow the desired output shape accurately. More conceptual details and results of our CBF methodology can be found in our previous work published in 2021 [15].…”

A simulation-based methodology to analyze the impact of edge-length on curvilinear mask accuracy

“…While the curvilinear mask shapes generated by ILT improve the wafer lithography process window, there is a direct relationship between the efficiency of the mask data process, the MBMW data-path turnaround time and the number edges used to represent the curvilinear mask shapes. Many shape simplification methods such as curvature-based fragmentation (CBF) [1] and edge-oriented simplification [2] have recently been studied and have been shown to be effective in reducing the number of edges used to represent curvilinear mask data. However, a quantitative approach towards deciding the optimal edge length to be used for representing curvilinear shapes for a given mask process is not yet reported in literature.…”