System for automatic inspection of periodic patterns

• US 4,969,198 A
• Filed: 05/16/1988
• Issued: 11/06/1990
• Est. Priority Date: 04/17/1986
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Dom, Byron E., Bonner, Raymond E., Batchelder, John S., Jaffe, Robert S.
• Primary Examiner(s): Boudreau, Leo H.
• Assistant Examiner(s): Cammarata, Michael R.
• Application Number:
  US07/194,610
• Time in Patent Office: 904 Days
• Field of Search: 382/8, 382/27, 382/50, 358/106, 356/237
• US Class Current: 382/147
• CPC Class Codes: G03F 7/7065 Defects, e.g. optical inspection of patterned layer for defects
• CPC Class Codes: G06T 2207/10056 Microscopic image
• CPC Class Codes: G06T 2207/30141 Printed circuit board [PCB]
• CPC Class Codes: G06T 2207/30148 Semiconductor; IC; Wafer
• CPC Class Codes: G06T 7/001 using an image reference approach

Abstract

A method and apparatus for automatic inspection of periodic patterns typically found on patterned silicon wafers, printed circuit board, and the like is disclosed herein. The method comprises an inspection algorithm of two parts: a low-level algorithm and a higher level algorithm. The low-level algorithm utilizes the known periodically of the pattern to find defects by comparing identical cells in the periodic array. The high-level algorithm applies the low-level algorithm, some number of times (N) in succession on the image; accumulates defective pixels to form a separate image; and then applies a threshold-sort operation on a neighborhood to determine center pixel defectiveness.

The apparatus for implementing the above method comprises a parallel/pipeline architecture for high speed processing and RAM LUT'"'"'s to implement a plurality of subtract and compare functions.

Claims

1. In a system for the automatic inspection of two dimensional periodic patterns on printed circuit boards, photolithographic masks, patterned semiconductor wafers and the like wherein said pattern is scanned by image producing means, the improvement to said inspection system comprising:

(a) acquiring digital images of the same optical image pattern some predetermined plurality of times, N;

(b) applying a low-level defect test to each of said digital images by comparing pixel greyscale values a repetition period away in either direction to produce a binary image where ones correspond to suspected defect pixels and all other pixels are set to zero;

(c) adding the results of said low-level test to form an accumulator array corresponding pixel-wise to the digital images such that for a given optical image (N digital images) the pixels in the accumulator array will contain values ranging from 0 to N;

(d) applying a preset first threshold level to each pixel level of the accumulator array;

(e) sorting a neighborhood of the accumulator array for central pixels above said first threshold limit;

(f) summing the value in the accumulator array of each sorted neighborhood to obtain a sum of values of each of the pixels considered for each sorted neighborhood associated with a respective central pixel, and

(g) applying a second threshold to said sum such that if said sum is greater than or equal to the second threshold, the associated respective central pixel is considered a defect.

2. A system according to claim 1 wherein the compared pixel greyscale value comprise three neighborhoods of pixel greyscale values a repetition period away in either direction from a center designated neighborhood.

3. A system according to claim 2 wherein the neighborhood of greyscale pixel values comprise a central pixel and its connected pixel neighbors.

4. A system according to claim 3 wherein there are four connected pixel neighbors to a central pixel

5. A system according to claim 3 wherein there are eight connected pixel neighbors to a central pixel.

6. A system according to claim 1 wherein a suspected defect pixel is a pixel either blacker than or whiter than both of its left and right neighbors.

7. A system according to claim 6 wherein the determination of a pixel neighbor as being blacker than or whiter than the central pixel comprises comparing selected predetermined pixels of the neighboring pixel group to selected predetermined pixels of the central pixel neighborhood.

8. A system according to claim 7 wherein the comparison of the selected predetermined pixels comprises a high sensitivity comparison test, a horizontal row test, a vertical row test and a missing pattern test.

9. A system according to claim 8 wherein the high sensitivity comparison test comprises:

(a) a difference test wherein one of the neighboring pixel groups to the central pixel are compared to said central pixel and a first preset threshold, and

(b) a gradient test wherein the center pixel of the neighbor is compared to its connected neighbors and a second preset threshold.

10. A system according to claim 8 wherein the horizontal row test comprises:

(a) a side row test comparing each of the pixels in a horizontal row of the neighboring pixel group to the central pixel and

(b) a central row test wherein the central pixel of the center neighborhood is compared to its horizontal row connected neighbors.

11. A system according to claim 8 wherein the vertical row test comprises:

(a) side row test comparing each of the pixels in a vertical row of the neighboring pixel group to the central pixel and

(b) a central row test wherein the central pixel of the center neighborhood is compared to its vertical row connected neighbors.

12. A system according to claim 8 wherein the missing pattern test comprises, comparing the left and right neighbors of the central pixel for a predetermined number of horizontally consecutive blacker than pixel results, netting a counter to mark each blacker than result and comparing said counter result to a preset threshold such that a missing pattern is indicated if the counter value is greater than or equal to the threshold.

13. In a system for automatic inspection of two dimensional periodic patterns on printed circuit boards, photolithographic masks, pattern semiconductor wafers and the like wherein said pattern is scanned by image producing means such that digital images of the same optical image pattern is acquired some predetermined plurality of times, N, and a greyscale value is assigned to said pixels so that greyscale values of the pixels in one pattern can be compared to the greyscale values of the respective pixels a repetition period away, the improvement to said inspection system comprising:

(a) establishing three neighborhoods of pixel greyscale values a repetition period away in either direction from a center designated neighborhood

(b) comparing said neighborhoods to said center designated neighborhood to test said respective pixels as being blacker than or whiter than said center pixels

(c) labelling by binary code to identify a neighbor as being blacker than or whiter than its respective center neighbor, and

(d) accumulating said labelled-by-binary-code neighbors.

14. A system according to claim 13 wherein the neighborhood of greyscale pixel values comprise a central pixel and its connected pixel neighbors.

15. A system according to claim 14 wherein there are four connected pixel neighbors to a central pixel.

16. A system according to claim 14 wherein there are eight connected pixel neighbors to a central pixel.

17. A system according to claim 13 wherein a suspected defect pixel is a pixel either blacker than or whiter than both of its left and right neighbors.

18. A system according to claim 17 wherein the determination of a pixel neighbor as being blacker than or whiter than the central pixel comprises comparing selected predetermined pixels of the neighboring pixel group to selected predetermined pixels of the central pixel neighborhood.

19. A system according to claim 18 wherein the comparison of the selected predetermined pixels comprises a high sensitivity comparison test, a horizontal row test, a vertical row test and a missing pattern test.

20. In a system for automatic inspection of two dimensional periodic patterns on printed circuit boards, photolithographic masks, pattern semiconductor wafers and the like wherein said pattern is scanned by image producing means such that digital images of the same optical image pattern is acquired some predetermined plurality of times, N, and a greyscale value is assigned to said pixels so that greyscale values of the pixels in one pattern can be compared to the greyscale values of the respective pixels a repetition period away, the improvement to said inspection system comprising:

(a) means for establishing three neighborhoods of pixel greyscale values a repetition period away in either direction from a center designated neighborhood

(b) means for comparing said neighborhoods to said center designated neighborhood to test said respective pixels as being blacker than or whiter than said center pixels

(c) means for labelling by binary code to identify a neighbor as being blacker than or whiter than its respective center neighbor, and

(d) means for accumulating said labelled-by-binary-code neighbors.

21. A system according to claim 20 wherein said means for comparing said neighborhoods comprises providing more than one pixel in parallel to a network that routes the values to the inputs of subtract and compare units.

22. A system according to claim 21 wherein the routing network multiplexes said pixels provided in parallel to create a predetermined number of greyscales values to be routed to the subtract and compare units.