Method for automatic semiconductor wafer inspection

• US 5,129,009 A
• Filed: 06/04/1990
• Issued: 07/07/1992
• Est. Priority Date: 06/04/1990
• Status: Expired due to Term

• Current Assignee: Freescale Semiconductor, Inc. (NXP Semiconductors NV)
• Original Assignee: Motorola, Inc. (Motorola Solutions, Inc.)
• Inventors: Lebeau, Christopher J.
• Primary Examiner(s): Mancuso, Joseph
• Application Number:
  US07/533,207
• Time in Patent Office: 764 Days
• Field of Search: 382/8, 382/32, 382/34, 382/41, 382/49, 382/22, 358/101, 358/106, 358/107, 358/388
• US Class Current: 382/149
• CPC Class Codes: G01N 21/956 Inspecting patterns on the surface of objects contactless testing of electronic circuits G01R31/308; testing currency G07D; manufacturing processes per se of semiconductor devices implementing a measuring step H01L22/10
• CPC Class Codes: G06T 2207/10024 Color image
• CPC Class Codes: G06T 2207/10152 Varying illumination
• CPC Class Codes: G06T 2207/20192 Edge enhancement; Edge preservation
• CPC Class Codes: G06T 2207/30148 Semiconductor; IC; Wafer
• CPC Class Codes: G06T 7/001 using an image reference approach
• CPC Class Codes: Y10S 706/902 Application using ai with detail of the ai system

Abstract

An automatic integrated circuit inspection method is provided wherein an image of an integrated circuit is obtained and a direction edge enhancement is performed. An image of an integrated circuit under inspection is then obtained and the direction edge enhancement performed. The second edge enhanced image is then logically compared to the first edge enhanced image. Preferably, the first edge enhanced image is dilated while the second edge enhanced image is skeletonized to improve robustness of the system allowing for magnification and rotation errors in either the sample image or the image under inspection. Further, defects which are located are then classified by obtaining a plurality of images of the defect while changing light conditions. The plurality of defect images are combined to form a feature matrix which is then compared against an expert system database having a large number of feature matrices associated with defect classifications.

Claims

1. A method of inspecting a semiconductor integrated circuit (IC) comprising:

obtaining an image of the IC;

performing direction edge enhancement of the image to form a direction edge shape;

skeletonizing the direction edge shape;

testing the skeletonized direction edge shape for correlation to a predetermined shape which has been previously stored;

identifying an anomalous shape which cannot be correlated to the stored predetermined direction edge shape and classifying the anomalous shape wherein the step of classifying the anomalous shape further comprises;

obtaining a plurality of images of the anomaly under varied lighting angles and lighting colors;

building a feature matrix using the plurality of images; and

comparing the feature matrix to an expert data base having feature data associated with defect classification data.

2. A method for automatically inspecting a patterned integrated circuit (IC) comprising:

analyzing a first IC including the steps of obtaining an image of a pattern formed on a first IC;

deriving a first direction edge image from the image of the first IC'"'"'s pattern;

analyzing a second IC including the steps of obtaining an image of another pattern formed on a second IC;

deriving a second direction edge image from the image of the second IC'"'"'s pattern; and

logically comparing the second direction edge image to the first direction edge image to identify anomalies in the second IC'"'"'s pattern which do not correspond to the first IC'"'"'s pattern;

storing a location of each anomaly identified after the logical comparison step;

illuminating an identified anomaly with a variety of colors of light;

obtaining a number of color images of the anomaly under each color of illumination;

storing each of the color images;

illuminating the anomaly at a variety of incident light angles;

obtaining a number of on/off axis images of the anomaly at each incident light angle;

storing each of the on/off axis images;

building a feature vector for the anomaly;

building a feature matrix for the anomaly;

providing a data base having defect classification associated with anomaly feature information; and

comparing the feature vector and feature matrix to the data base to classify the anomaly.

3. The method of claim 2 further comprising manually determining a defect classification for the anomaly and storing the features matrix data in the data base along with its associated defect classification.