US 11,842,880 B2
Estimation model generation method and electron microscope
Ryusuke Sagawa, Tokyo (JP); Shigeyuki Morishita, Tokyo (JP); Fuminori Uematsu, Tokyo (JP); Tomohiro Nakamichi, Tokyo (JP); and Keito Aibara, Tokyo (JP)
Assigned to JEOL Ltd., Tokyo (JP)
Filed by JEOL Ltd., Tokyo (JP)
Filed on Feb. 15, 2022, as Appl. No. 17/671,896.
Claims priority of application No. 2021-022197 (JP), filed on Feb. 16, 2021.
Prior Publication US 2022/0262595 A1, Aug. 18, 2022
Int. Cl. H01J 37/153 (2006.01); H01J 37/26 (2006.01); H01J 37/28 (2006.01)
CPC H01J 37/153 (2013.01) [H01J 37/265 (2013.01); H01J 37/28 (2013.01); H01J 2237/1534 (2013.01); H01J 2237/2487 (2013.01); H01J 2237/2802 (2013.01)] 8 Claims
OG exemplary drawing
 
1. A method of generating an estimation model, the method comprising:
generating a plurality of calculated Ronchigrams by repeatedly executing a simulation while changing a simulation condition; and
improving an estimation model in a learner by sequentially supplying, to the learner, a plurality of sets of training data formed from the plurality of calculated Ronchigrams and a plurality of sets of correct answer data corresponding to the plurality of calculated Ronchigrams, wherein
the estimation model is a model for estimating one or a plurality of aberration values which are referred to in aberration correction in an electron microscope, and
each set of correct answer data is data indicating one or a plurality of hypothetical aberration values included in each simulation condition.