公开(公告)号：US12062583B2

公开(公告)日：2024-08-13

申请号：US17198913

申请日：2021-03-11

申请人： Applied Materials Israel Ltd.

发明人： Eric Chin Hong Ng ,  Edward Budiarto ,  Sergey Starik ,  Todd J. Egan

IPC分类号： H01L21/66 ,  G01B11/06 ,  G01N21/21 ,  G06N20/00

CPC分类号： H01L22/12 ,  G01B11/0625 ,  G01B11/0633 ,  G01B11/0641 ,  G01B11/065 ,  G01B11/0683 ,  G06N20/00 ,  H01L22/20 ,  G01B2210/56 ,  G01N21/211 ,  G01N2021/213

摘要： An optical metrology model for in-line thickness measurements of a film overlying non-ideal structures on a substrate is generated by performing pre-measurements prior to deposition of the film and performing post-measurements after the deposition. The pre- and post-measurements are performed at at least one of multiple polarization angles or multiple orientations of the substrate. Differences in reflectance between the pre- and post-measurements are determined at the multiple polarization angles and the multiple orientations. At least one of the multiple polarization angles or the multiple orientations are identified where the differences in reflectance are indicative of a suppressed influence from the non-ideal structures. The optical metrology model is generated using the identified polarization angles and the identified orientations as inputs to a machine-learning algorithm.

公开(公告)号：US20220290974A1

公开(公告)日：2022-09-15

申请号：US17198913

申请日：2021-03-11

申请人： Applied Materials Israel Ltd.

发明人： Eric Chin Hong Ng ,  Edward Budiarto ,  Sergey Starik ,  Todd J. Egan

IPC分类号： G01B11/06 ,  H01L21/66 ,  G06N20/00

摘要： An optical metrology model for in-line thickness measurements of a film overlying non-ideal structures on a substrate is generated by performing pre-measurements prior to deposition of the film and performing post-measurements after the deposition. The pre- and post-measurements are performed at at least one of multiple polarization angles or multiple orientations of the substrate. Differences in reflectance between the pre- and post-measurements are determined at the multiple polarization angles and the multiple orientations. At least one of the multiple polarization angles or the multiple orientations are identified where the differences in reflectance are indicative of a suppressed influence from the non-ideal structures. The optical metrology model is generated using the identified polarization angles and the identified orientations as inputs to a machine-learning algorithm