Training a learning based defect classifier
First Claim
Patent Images
1. A system configured to train a learning based defect classifier, comprising:
- an inspection subsystem comprising at least an energy source and a detector, wherein the energy source is configured to generate energy that is directed to wafers, wherein the detector is configured to detect energy from the wafers and to generate output responsive to the detected energy, and wherein the wafers comprise at least one training wafer known to have an abnormally high defectivity and at least one inspection wafer expected to have normal defectivity; and
one or more computer subsystems configured for;
detecting defects on the at least one training wafer and the at least one inspection wafer by applying a defect detection method to the output generated for the at least one training wafer and the at least one inspection wafer, respectively;
identifying defects of interest on the at least one training wafer and the at least one inspection wafer by determining which of the defects detected on the at least one training wafer and the at least one inspection wafer, respectively, are the defects of interest;
identifying nuisances on the at least one training wafer and the at least one inspection wafer by determining which of the defects detected on the at least one training wafer and the at least one inspection wafer, respectively, are the nuisances;
generating a training set of defects comprising the identified defects of interest and the identified nuisances by combining information for the identified defects of interest and the identified nuisances;
identifying one or more locations on the at least one training wafer and the at least one inspection wafer at which none of the defects are detected;
adding images generated for the identified one or more locations to the training set of defects as one or more non-defective locations on the at least one training wafer and the at least one inspection wafer; and
training a learning based defect classifier with the training set of defects.
1 Assignment
0 Petitions
Accused Products
Abstract
Methods and systems for training a learning based defect classifier are provided. One method includes training a learning based defect classifier with a training set of defects that includes identified defects of interest (DOIs) and identified nuisances. The DOIs and nuisances in the training set include DOIs and nuisances identified on at least one training wafer and at least one inspection wafer. The at least one training wafer is known to have an abnormally high defectivity and the at least one inspection wafer is expected to have normal defectivity.
31 Citations
52 Claims
-
1. A system configured to train a learning based defect classifier, comprising:
-
an inspection subsystem comprising at least an energy source and a detector, wherein the energy source is configured to generate energy that is directed to wafers, wherein the detector is configured to detect energy from the wafers and to generate output responsive to the detected energy, and wherein the wafers comprise at least one training wafer known to have an abnormally high defectivity and at least one inspection wafer expected to have normal defectivity; and one or more computer subsystems configured for; detecting defects on the at least one training wafer and the at least one inspection wafer by applying a defect detection method to the output generated for the at least one training wafer and the at least one inspection wafer, respectively; identifying defects of interest on the at least one training wafer and the at least one inspection wafer by determining which of the defects detected on the at least one training wafer and the at least one inspection wafer, respectively, are the defects of interest; identifying nuisances on the at least one training wafer and the at least one inspection wafer by determining which of the defects detected on the at least one training wafer and the at least one inspection wafer, respectively, are the nuisances; generating a training set of defects comprising the identified defects of interest and the identified nuisances by combining information for the identified defects of interest and the identified nuisances; identifying one or more locations on the at least one training wafer and the at least one inspection wafer at which none of the defects are detected; adding images generated for the identified one or more locations to the training set of defects as one or more non-defective locations on the at least one training wafer and the at least one inspection wafer; and training a learning based defect classifier with the training set of defects. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25)
-
-
26. A non-transitory computer-readable medium, storing program instructions executable on a computer system for performing a computer-implemented method for training a learning based defect classifier, wherein the computer-implemented method comprises:
-
detecting defects on at least one training wafer and at least one inspection wafer by applying a defect detection method to output generated for the at least one training wafer and the at least one inspection wafer, respectively, by a detector of an inspection subsystem, wherein the at least one training wafer is known to have an abnormally high defectivity, wherein the at least one inspection wafer is expected to have normal defectivity, wherein the inspection subsystem comprises at least an energy source and the detector, wherein the energy source is configured to generate energy that is directed to the wafer, and wherein the detector is configured to detect energy from the wafer and to generate the output responsive to the detected energy; identifying defects of interest on the at least one training wafer and the at least one inspection wafer by determining which of the defects detected on the at least one training wafer and the at least one inspection wafer, respectively, are the defects of interest; identifying nuisances on the at least one training wafer and the at least one inspection wafer by determining which of the defects detected on the at least one training wafer and the at least one inspection wafer, respectively, are the nuisances; generating a training set of defects comprising the identified defects of interest and the identified nuisances by combining information for the identified defects of interest and the identified nuisances; and training a learning based defect classifier with the training set of defects.
-
-
27. A computer-implemented method for training a learning based defect classifier, comprising:
-
detecting defects on at least one training wafer and at least one inspection wafer by applying a defect detection method to output generated for the at least one training wafer and the at least one inspection wafer, respectively, by a detector of an inspection subsystem, wherein the at least one training wafer is known to have an abnormally high defectivity, wherein the at least one inspection wafer is expected to have normal defectivity, wherein the inspection subsystem comprises at least an energy source and the detector, wherein the energy source is configured to generate energy that is directed to the wafer, and wherein the detector is configured to detect energy from the wafer and to generate the output responsive to the detected energy; identifying defects of interest on the at least one training wafer and the at least one inspection wafer by determining which of the defects detected on the at least one training wafer and the at least one inspection wafer, respectively, are the defects of interest; identifying nuisances on the at least one training wafer and the at least one inspection wafer by determining which of the defects detected on the at least one training wafer and the at least one inspection wafer, respectively, are the nuisances; generating a training set of defects comprising the identified defects of interest and the identified nuisances by combining information for the identified defects of interest and the identified nuisances; and training a learning based defect classifier with the training set of defects, wherein said detecting, said identifying the defects of interest, said identifying the nuisances, said generating, and said training are performed by one or more computer subsystems coupled to the inspection subsystem. - View Dependent Claims (28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52)
-
Specification