As mask complexity has increased and design rules continued to shrink, the manufacturing cost per mask has steadily
increased as well. Studies also show that defects are the number one issue for mask yield. Smaller defects are typically
addressed through process development, or through photomask repair. The occurrence of large defects often may only
be further reduced through use of expensive clean room improvements, like SMIF handling systems. The impact of each
large defect therefore increases while the feasibility in their repair decreases as they can span a large number of adjoining
densely packed patterns. The presence of sub-resolution features such as scatter bars and the increasing use of embedded
phase-shifting masks also complicates the timely repair of such defects.
Existing mask repair techniques such as nanomachining, electron beam, or focused ion beam are challenged to produce
high yield repairs on such large defects within a reasonable timeframe. Often very complex repairs may in fact take
longer than a rewrite of the mask! Deep UV (DUV) femtosecond pulse laser mask repair provides a unique solution to
this defect repair need.
Methods and results are discussed for the process optimization for the removal of large (5μm) area repair on both Cr and
MoSi absorber films on quartz. Additionally, high repair throughput results are shown for unknown contamination
removal, and reproduction of ≥1 μm complex unconnected patterns in a single repair run lasting a matter of minutes.
Closed-loop CD feedback in-situ with the iterative repair process for such structures can readily result in an edge
placement control within ±15 nm. Prior iterative CD closed-loop repairs on specific structures have reliably yielded
results within ±10 nm, as confirmed by AIMS CD error, even after aggressive mask wet clean. The nanometer scale
dimensional resolution and repeatability of such repairs is shown with the use of sub-pixel resolution automated pattern
reconstruction using integrated high-NA DUV microscope imaging.