Dynamic diversity-driven hierarchical particle swarm optimization method for edge distortion compensation in digital lithography mask optimization

Digital micromirror device (DMD) lithography suffers from edge distortions caused by micromirror discretization and diffraction effects. To address this challenge, this study proposed a diversity-driven hierarchical particle swarm optimization (PSO) method for high-fidelity mask optimization. The algorithm integrated tri-level population dynamics with adaptive diversity control to overcome premature convergence. Validation on CEC2005 and CEC2022 benchmarks demonstrated exceptional accuracy, including near-zero errors on multimodal functions. For 30*30 test patterns masks, pattern error (PE) reduced 91.8% with structural similarity index measure (SSIM) exceeding 0.99 at 5.59 s average runtime. For 100*100 complex masks, PE decreased 67.3% with SSIM above 0.97 at 72.57 s average runtime. This framework provided an efficient and effective solution for achieving high quality image in digital lithography.