Charged particle imaging methods for CD metrology of sub 22nm 3D device structures.

摘要

Critical dimension scanning electron microscopes (CD-SEMs) are used to perform highly accurate dimensional metrology on patterned features. In order to ensure optimal feedback for process control, it is necessary that these tools produce highly reproducible measurements. As the smallest device features continue to shrink, and new challenging high aspect ratio (HAR) structures are being introduced, gaps are appearing between process control measurements that are necessary for high volume manufacturing and the capabilities of the CD-SEM. Two possible routes for solving this problem include improvement of the existing CD-SEM technology or the replacement of the CD-SEM.;With improved tool monitoring techniques, the uncertainty in the tool measurements may be reduced, leading to an improvement in the tool performance. By using a carefully designed test structure (such as a pseudorandom dot array), the Contrast Transfer Function (CTF) of a given tool can be decoupled from the specimen information, allowing for characterization of the imaging system itself. Test samples are fabricated using nanoimprint lithography and are imaged in a variety of CD-SEMs in order to measure the performance of the microscopes. This technique is used successfully to identify when the tool is not performing optimally, as well as to monitor the performance of a tool over time and match the performance of different tools.;Research is being made into CD-SEM replacement technologies, among them, ion microscopy. The Helium Ion Microscope's (HeIM) higher depth of focus than the CD-SEM could be advantageous for the imaging of HAR structures. Studies were conducted in order to determine what imaging signals will be the most useful for CD-metrology and to evaluate the damage that the beam will do to the sample. A technique was developed to determine the depth which that signals were able to escape from the HAR structures, using a series of images acquired with varied tilts. This allows the abilities of the SEM and HeIM to image the bottoms of deep contact holes is compared, and Monte Carlo modeling is used to gain further insights into the process. In our tests, the HeIM outperformed the SEM, but it was unable to detect signals from the bottoms of all of the deep contact hole structures fabricated for this study. Modeling results show for SEM imaging of contact holes, signals should be able to escape from the hole, but they will almost completely be obscured by noise.