US 12,292,389 B2
Defect inspection apparatus
Toshifumi Honda, Tokyo (JP)
Assigned to HITACHI HIGH-TECH CORPORATION, Tokyo (JP)
Appl. No. 17/924,237
Filed by Hitachi High-Tech Corporation, Tokyo (JP)
PCT Filed Jun. 5, 2020, PCT No. PCT/JP2020/022391
§ 371(c)(1), (2) Date Nov. 9, 2022,
PCT Pub. No. WO2021/245937, PCT Pub. Date Dec. 9, 2021.
Prior Publication US 2023/0175982 A1, Jun. 8, 2023
Int. Cl. G01N 21/95 (2006.01); G01N 21/47 (2006.01)
CPC G01N 21/9505 (2013.01) [G01N 21/47 (2013.01)] 10 Claims
OG exemplary drawing
 
1. A defect inspection apparatus for detecting a defect in a semiconductor substrate, the defect inspection apparatus comprising:
a sample stand that supports a sample of the semiconductor substrate;
an illumination optical system that irradiates the sample placed on the sample stand with illumination light;
a scanning device that drives the sample stand to change a relative position between the sample and the illumination optical system;
a plurality of detection optical systems that condense illumination scattered light from a surface of the sample;
a plurality of sensors that convert the illumination scattered light condensed by a corresponding detection optical system into an electric signal and outputs a detection signal;
a storage device that stores a plurality of feature vectors for each defect type with respect to an instruction defect for defect determination; and
a signal processing device that processes detection signals input from the plurality of sensors,
wherein the signal processing device;
calculates a plurality of actual measurement vectors, each of the plurality of actual measurement vectors is a feature vector of a detection defect on the surface of the sample detected by the plurality of sensors,
wherein a size of each of the plurality of actual measurement vectors is a discrete value,
calculate expected values for feature values of a virtual data with a same defect type as the plurality of actual measurement vectors,
calculate a variation in each of the expected values,
calculates, using the variation in each of the expected values, a plurality of feature vectors of the virtual defect having the same defect type, and
stores each the plurality of feature vectors of the virtual defect in the storage device instruction defect corresponding to the same defect.