Defect review sampling and normalization based on defect and design attributes

Patent metadata

Field	Value
Publication number	US-10204290-B2
Application number	US-201715426138-A
Country	US
Kind code	B2
Filing date	Feb 7, 2017
Priority date	Oct 14, 2016
Publication date	Feb 12, 2019
Grant date	Feb 12, 2019

Abstract

A decision tree and normalized reclassification are used to classify defects. Defect review sampling and normalization can be used for accurate Pareto ranking and defect source analysis. A defect review system, such as a broadband plasma tool, and a controller can be used to bin defects using the decision tree based on defect attributes and design attributes. Class codes are assigned to at least some of the defects in each bin. Normalized reclassification assigns a class code to any unclassified defects in a bin. Additional decision trees can be used if any bin has more than one class code after normalized reclassification.

First claim

What is claimed is:

1. A system comprising:
a defect review system, wherein the defect review system includes:
a stage configured to hold a wafer; and
an image generation system configured to generate an image of the wafer, wherein the image generation system includes at least one of an electron beam source, a broadband plasma source, a laser source, or a lamp; and

a controller in electronic communication with the defect review system, wherein the controller includes a processor, an electronic data storage unit in electronic communication with the processor, and a communication port in electronic communication with the processor and the electronic data storage unit, wherein the controller is configured to:
bin a plurality of defects into a plurality of bins using a decision tree based on defect attributes and design attributes;
assign one of one or more class codes to at least some of the defects in each of the bins thereby resulting in classified defects, wherein each of the class codes represents a different defect type, and wherein unclassified defects remain after the one or more class codes are applied; and
perform normalized reclassification on each of the bins with the unclassified defects, wherein each of the unclassified defects in the bins is assigned one of the one or more class codes of its respective bin by randomly assigning one of the one or more class codes to each of the unclassified defects according to a ratio of each quantity of the classified defects assigned the one or more class codes in the bin that contains the unclassified defect.

2. The system of claim 1 , wherein the image generation system includes the broadband plasma source, and wherein the defect review system is a broadband plasma tool.

3. The system of claim 1 , wherein one of the bins includes two of the class codes after the normalized reclassification, and wherein the controller is further configured to:
determine that one of the bins includes two of the class codes after the normalized reclassification;
bin the defects in the bin with two of the class codes into a plurality of secondary bins using a secondary decision tree based on secondary defect attributes and secondary design attributes;
assign one of one or more secondary class codes to at least some of the defects in each of the secondary bins thereby resulting in the classified defects in the secondary bins, wherein each of the secondary class codes represents a different defect type, and wherein the unclassified defects remain in at least one of the secondary bins after the one or more secondary class codes are applied; and
perform normalized reclassification on each of the secondary bins with the unclassified defects, wherein each of the unclassified defects in the secondary bins is assigned one of the one or more secondary class codes of its respective bin by randomly assigning one of the one or more secondary class codes to each of the unclassified defects in the secondary bin according to a ratio of each quantity of the classified defects assigned the one or more secondary class codes in the secondary bin that contains the unclassified defect.

4. The system of claim 1 , wherein the defect attributes include one or more of attributes extracted from patch image processing algorithms, inspector optical attributes, inspector recipe attributes, wafer level signature attributes, zonal attributes, care area information, metrology attributes, process conditions, process equipment, and user-defined attributes, and wherein the design attributes include one or more of design-based class, design-based grouping, pattern grouping, hotspot grouping, design criticality index, pattern complexity index, and region of interest.

5. The system of claim 1 , wherein the controller is configured to proportionally distribute a number of the defects per bin after the binning.

6. A method comprising:
binning, using a controller, a plurality of defects from a semiconductor wafer into a plurality of bins using a decision tree based on defect attributes and design attributes, wherein the controller includes a processor, an electronic data storage unit in electronic communication with the processor, and a communication port in electronic communication with the processor and the electronic data storage unit;
assigning, using the controller, one of one or more class codes to at least some of the defects in each of the bins thereby resulting in classified defects, wherein each of the class codes represents a different defect type, and wherein unclassified defects remain after the one or more class codes are applied; and
performing, using the controller, normalized reclassification on each of the bins with the unclassified defects, wherein each of the unclassified defects in the bins is assigned one of the one or more class codes of its respective bin by randomly assigning one of the one or more class codes to each of the unclassified defects according to a ratio of each quantity of the classified defects assigned the one or more class codes in the bin that contains the unclassified defect.

7. The method of claim 6 , wherein the defect attributes include one or more of attributes extracted from patch image processing algorithms, inspector optical attributes, inspector recipe attributes, wafer level signature attributes, zonal attributes, care area information, metrology attributes, process conditions, process equipment, and user-defined attributes, and wherein the design attributes include one or more of design-based class, design-based grouping, pattern grouping, hotspot grouping, design criticality index, pattern complexity index, and region of interest.

8. The method of claim 6 , wherein a number of the defects per bin is distributed proportionally after the binning.

9. The method of claim 6 , wherein one of the bins includes two of the class codes after the normalized reclassification, and further comprising:
determining, using the controller, that one of the bins includes two of the class codes after the normalized reclassification;
binning, using the controller, the defects in the bin with two of the class codes into a plurality of secondary bins using a secondary decision tree based on secondary defect attributes and secondary design attributes;
assigning, using the controller, one of one or more secondary class codes to at least some of the defects in each of the secondary bins thereby resulting in the classified defects in the secondary bins, wherein each of the secondary class codes represents a different defect type, and wherein the unclassified defects remain in at least one of the secondary bins after the one or more secondary class codes are applied; and
performing, using the controller, normalized reclassification on each of the secondary bins with the unclassified defects, wherein each of the unclassified defects in the secondary bins is assigned one of the one or more secondary class codes of its respective bin by randomly assigning one of the one or more secondary class codes to each of the unclassified defects in the secondary bin according to a ratio of each quantity of the classified defects assigned the one or more secondary class codes in the secondary bin that contains the unclassified defect.

10. The method of claim 6 , further comprising inspecting a wafer for defects and communicating the plurality of defects on the wafer to the controller for binning.

11. A non-transitory computer-readable storage medium, comprising one or more programs for executing the following steps on one or more computing devices:
binning a plurality of defects from a semiconductor wafer into a plurality of bins using a decision tree based on defect attributes and design at

Assignees

• Kla Tencor Corp

Inventors

• Yong Poh Boon

Classifications

• G06V10/764

  using classification, e.g. of video objects · CPC title

• G06F18/24323

  Tree-organised classifiers · CPC title

• G06T7/0006 Primary

  using a design-rule based approach · CPC title

• G06V10/44

  Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components · CPC title

• G06T2207/30148

  Semiconductor; IC; Wafer · CPC title