Method for detection of defects in the inspection of structured surfaces

• US 6,075,880 A
• Filed: 02/28/1995
• Issued: 06/13/2000
• Est. Priority Date: 03/26/1994
• Status: Expired due to Term

• Current Assignee: Leica Microsystems Gmbh (Danaher Corporation)
• Original Assignee: Jenoptik Technologie GmbH
• Inventors: Kempe, Heiko, Kollhof, Dietmar, Graef, Michael, Wienecke, Joachim, Franke, Karl-Heinz
• Primary Examiner(s): Bost, Dwayne
• Assistant Examiner(s): WYCHE, MYRON
• Application Number:
  US08/395,668
• Time in Patent Office: 1,932 Days
• Field of Search: 382/141, 382/144, 382/145, 382/146, 382/147, 382/148, 382/149, 382/150, 382/151
• US Class Current: 382/141
• CPC Class Codes: G01N 21/8851 Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges G01N21/8806 and G01N21/93 - G01N21/95692 take precedence; optical measurement of dimensions G01B11/00; optical scanning G02B26/10; image transformation G06T3/00; computerised image enhancement G06T5/00; image processing per se for flaw detection G06T7/0002
• CPC Class Codes: G01R 31/308 using non-ionising electromagnetic radiation, e.g. optical radiation investigating or analysing materials by the use of optical means G01N21/00; image analysis G06T7/00
• CPC Class Codes: G06T 2207/20076 Probabilistic image processing
• CPC Class Codes: G06T 2207/30141 Printed circuit board [PCB]
• CPC Class Codes: G06T 2207/30148 Semiconductor; IC; Wafer
• CPC Class Codes: G06T 7/0004 Industrial image inspection

Abstract

The object of a method for detecting defects in the inspection of structured surfaces is to ensure a detection of defects which is not dependent on the number of structuring planes and includes structure features in real-time operation for separating defects from good structures. From image point classification in which zones of a recorded image which have similar image point features are assembled, a gray-value intermediate image containing edge structures and corner structures is generated from the image and the behavior of the image point features of every image point in the intermediate image is analyzed with respect to its neighboring image points. The method is used predominantly in statistical process control in the production process of masks, LCD'"'"'s, printed circuit boards and semiconductor wafers.

Claims

1. In a method for detecting defects in inspecting structured surfaces, in particular masks, LCD'"'"'s, printed circuit boards and semiconductor wafers, using specific image point features of structures and defects of a recorded image of a surface in which zones of the image having similar image point features are compiled by image point classification and a behavior of image point features of an image point is analyzed with reference to its neighboring image points by structure classification, the improvement comprising the step of:

generating a gray-value intermediate image containing edge structures and corner structures and serving for structure classification in an image point classification of the recorded image, the generating of the intermediate image including a step of describing distributions of image point features of selected random samples which correspond at least to a number of defect-free surface portions with distinct features in a feature space plotted by a number of image point features by means of probabilities which are transformed into grayscale values.

2. The method according to claim 1, wherein the grayscale values are contents of a first addressable storage for revaluation of every image point feature of the recorded image into a grayscale value and distributions of all random samples form an overall distribution, the method further comprising storing results of an evaluation of surroundings for each image point of the intermediate image carried out on a basis of at least two uncorrelated characteristics in a second addressable storage in binary form so as to be separated into defect-free regions and defect regions, wherein size of the second addressable storage is determined by a second feature space plotted by a number of evaluated characteristics and the storing in the second addressable storage corresponds to a distribution of characteristics of every image point in the second feature space.

3. The method according to claim 2, including the step of taking the random samples from the recorded image.

4. The method according to claim 2, including the step of taking the random samples from stored image contents.

5. The method according to claim 2, including the step of taking the random samples from the recorded image as well as from the stored image contents.

6. The method according to claim 2, including the step of using random samples of surface portions with defects in addition to the selected random samples which correspond at least to the number of defect-free surface portions with distinct features.

7. The method according to claim 2, including the step of effecting the evaluation of surroundings by at least two filters with variable core and uncorrelated characteristics by noise suppression and highlighting self-contained zones lying within the evaluated surroundings.

8. The method according to claim 7, including the step of moving a pair of filters, one of said pair being a low-pass filter and the other being a bandpass filter.

9. The method according to claim 8, including the step of using a Gauss filter as the low-pass filter and a spot filter as the bandpass filter whose core is a second derivative of a two-dimensional Gaussian function.

10. The method according to claim 9, including the step of using lookup tables as addressable storages.

11. The method according to claim 10, including the step of using color features of the image points as image point features.

12. The method according to claim 10, including the step of using texture features which are associated with each image point by its surroundings as image point features.

13. The method according to claim 11, including the steps of first selecting an image cutout which corresponds to dimensions of the recorded image and in which detected defects once again serve as random samples from a first inspection zone of different inspection zones in order to determine contents of the first addressable storage and the second addressable storage and testing applicability of results of the contents determination to at least one image cutout in at least one of the different inspection zones other than the first inspection zone.

14. The method according to claim 12, including the steps of first selecting an image cutout which corresponds to dimensions of the recorded image and in which detected defects once again serve as random samples from a first inspection zone of different inspection zones in order to determine contents of the first addressable storage and the second addressable storage and testing applicability of results of the contents determination to at least one image cutout in at least one of the different inspection zones other than the first inspection zone.

15. In a method for detecting defects in inspecting structured surfaces, in particular masks, LCD'"'"'s, printed circuit boards and semiconductor wafers, using specific image point features of structures and defects of a recorded image of a surface in which zones of the image having similar image point features are compiled by image point classification and a behavior of image point features of an image point is analyzed with reference to its neighboring image points by structure classification, the improvement comprising the step of:

generating a gray-value intermediate image containing edge structures and corner structures and serving for structure classification in an image point classification of the recorded image, the generating of the intermediate image including a step of describing separate continuous distributions of image point features of randomly selected sample regions of the recorded image which correspond at least to a number of defect-free surface portions with distinct features in a feature space, the generating of the intermediate image further including steps of approximating each of the separate continuous distributions by a respective probability function that is normalized to 1, calculating an overall probability function for any image point feature to be present in any one of the structures by use of all the probability functions of the separate continuous distributions and transforming each image point feature into a grayscale value by the overall probability function.

16. The method according to claim 15, wherein the grayscale values are contents of a first addressable storage for revaluation of every image point feature of the recorded image into a grayscale value and distributions of all randomly selected sample regions form an overall distribution, the method further comprising storing results of an evaluation of surroundings for each image point of the intermediate image carried out on a basis of at least two uncorrelated characteristics in a second addressable storage in binary form so as to be separated into defect-free regions and defect regions, wherein size of the second addressable storage is determined by a second feature space plotted by a number of evaluated characteristics and the storing in the second addressable storage corresponds to a distribution of characteristics of every image point in the second feature space.

17. The method according to claim 16, including the step of taking the randomly selected sample regions from the recorded image.

18. The method according to claim 16, including the step of taking the randomly selected sample regions from stored image contents.

19. The method according to claim 16, including the step of taking the randomly selected sample regions from the recorded image as well as from stored image contents.

20. The method according to claim 16, including the step of using random samples of surface portions with defects in addition to the randomly selected sample regions which correspond at least to the number of defect-free surface portions with distinct features.

21. The method according to claim 16, further including the step of effecting the evaluation of surroundings by at least two filters with variable core and uncorrelated characteristics by noise suppression and highlighting self-contained zones lying within the evaluated surroundings.

22. The method according to claim 21, including the step of moving a pair of filters, one of said pair being a low-pass filter and the other being a bandpass filter.