Semiconductor device image inspection utilizing image subtraction and threshold imaging

• US 5,949,901 A
• Filed: 03/21/1996
• Issued: 09/07/1999
• Est. Priority Date: 03/21/1996
• Status: Expired due to Term

• Current Assignee: Cognex Corporation
• Original Assignee: Cognex Corporation
• Inventors: Scola, Joseph, Nichani, Sanjay
• Primary Examiner(s): Boudreau, Leo H.
• Assistant Examiner(s): Werner, Brian P.
• Application Number:
  US08/621,190
• Time in Patent Office: 1,265 Days
• Field of Search: 382/144, 382/145, 382/146, 382/147, 382/148, 382/149, 348/87, 348/126, 437/8
• US Class Current: 382/149
• CPC Class Codes: G06T 2207/30148 Semiconductor; IC; Wafer
• CPC Class Codes: G06T 7/001 using an image reference approach
• CPC Class Codes: G06T 7/174 involving the use of two or more images

Abstract

Machine vision methods for inspection of semiconductor die surfaces include the steps of generating a first image of the die surface, generating a second image of the die surface and any defect thereon, and subtracting the second image from the first image. The methods are characterized in that the second image is generated such that subtraction of it from the first image emphasizes the defect with respect to the die surface.

Claims

1. A machine vision method for inspecting a semiconductor device, including either a semiconductor die surface or a semiconductor lead frame, the method comprising the steps of:

illuminating the semiconductor device with an illumination source selected from a group of illumination sources including a direct on-axis light source and a diffuse off-axis light source;

generating a first on-axis image of the semiconductor device while it is so illuminated;

illuminating the semiconductor device with another illumination source selected from the aforesaid group;

generating a second on-axis image of the semiconductor device while it is so illuminated;

subtracting the second image from the first image to form a third image that emphasizes any defect on the semiconductor device;

isolating the defect in the third image;

the isolating step including

(i) thresholding the third image to distinguish at least one of a defect and its edges from the semiconductor device, wherein the thresholding step includes the step of generating a threshold image from at least one of the first and second images, the threshold image having pixels representing local threshold intensity values; and

wherein the step of generating the threshold image includes the step of mapping image intensity values in at least one of the first and second images to generate the threshold image,

(ii) using the threshold image to distinguish, in the third image, at least one of the defect and its edges from the semiconductor device.

2. A method according to claim 1, wherein the step of generating the second image includes the step of generating that image such that subtraction of the second image from the first image increases a contrast between the defect and the semiconductor device.

3. A method according to claim 1, where the isolating step comprises the step of performing connectivity analysis on the third image to distinguish the defect from the semiconductor device.

4. A method according to claim 1, wherein the isolating step comprises the step of detecting and tracking edges in the third image to isolate the defect.

5. A method according to claim 1, wherein the thresholding step comprises the step of determining an intensity threshold value that distinguishes at least one of the defect and its edges from the semiconductor device.

6. A method according to claim 1, wherein the step of using the threshold image includes the step of subtracting the threshold image from the third image.

7. A method according to claim 1, comprising the step of normalizing at least one of the first and second images before the subtracting step.

8. A method according to 7, wherein the normalizing step includes the steps of

determining distributions of intensity values of each of the first and second images;

generating a mapping function for matching extrema of those distributions; and

transforming the intensity values of at least one of the first and second images with that mapping function.

9. A method according to claim 1, including the step of generating the first and second images by illuminating the semiconductor device with different respective emission sources.

10. A method according to claim 1, including the step of generating the first and second images with light of different respective polarizations.

11. A method according to claim 1, including the step of generating the first and second images by illuminating the semiconductor device with emissions in different respective wavelengths.

12. A method according to claim 1, including the further step of registering the first and second images with one another before the subtracting step.

13. A machine vision method for inspecting a semiconductor device, including either a semiconductor die surface or a semiconductor lead frame, the method comprising the steps of:

generating a first image of the semiconductor device;

generating a second image of the semiconductor device and any defect thereon such that subtraction of the second image from the first image emphasizes the defect with respect to the semiconductor device;

subtracting the second image from the first image to form a third image;

generating a threshold image from the second image, the threshold image having pixels representing local threshold intensity values;

subtracting the threshold image from the third image to generate a fourth image; and

isolating the defect in the fourth image by any of segmentation, edge detection and tracking, connectivity analysis, and thresholding.

14. A machine vision method for inspecting a semiconductor device, including either a semiconductor die surface or a semiconductor lead frame, comprising the steps of:

generating a first image of the semiconductor device;

generating a second image of the semiconductor device and any defect thereon such that subtraction of the second image from the first image emphasizes the defect with respect to the semiconductor device;

at least one of registering and normalizing the first and second images with respect to one another;

subtracting the second image from the first image to form a third image;

generating a threshold image from the second image, the threshold image having pixels representing local threshold intensity values;

subtracting the threshold image from the third image to generate a fourth image; and

isolating the defect in the fourth image by any of segmentation, edge detection and tracking, connectivity analysis, and thresholding.