Effectiveness of multipass and multirow writing methods for massively parallel e-beam systems

Massively parallel electron-beam systems are equipped with a large number of beams to improve the writing throughput. It is unavoidable that some of the beams are abnormal, e.g., always on or off, spatial and temporal fluctuations of beam current, beam-positioning error, etc. A practical approach to improve the writing quality is to spread the negative effects of abnormal beams spatially. The multirow writing (MRW) was introduced, which uses each beam to expose pixels over multiple rows in each writing path minimizing the localization of pixels affected by an abnormal beam. In another method, the single-row writing (SRW), each beam exposes pixels in one row in each writing path localizing the affected pixels in a row. To spread the negative effects, especially for the single-row writing, each row of pixels may be exposed through multiple passes, i.e., multipass writing. In this study, the multirow and multipass writing methods in various combinations with the MRW and SRW are compared in terms of their effectiveness in reducing the negative effects of abnormal beams. The results from an extensive simulation study are analyzed in detail.