A portion of the disclosure of this patent document contains material which is subject to copyright protection. The copyright owner has no objection to the facsimile reproduction by anyone of the patent document or the patent disclosure, as it appears in the Patent and Trademark Office patent file or records, but otherwise reserves all copyright rights whatsoever. 37 CFR 1.71(e).
1. Field of the Invention
This invention generally relates to digital image processing and, more particularly, to systems and methods for detecting defects in a semiconductor device using image comparison techniques.
2. Description of the Related Art
Image comparison techniques are used to detect defects in a semiconductor wafer. Typically, a test image is acquired and then compared to a reference image. A defect-detection algorithm is then used to detect variations between the images and to determine whether such variations are real defects. In the so-called random-logic inspection mode, an image of a first die is acquired and then compared to the image of a second die in the same wafer. Array-inspection mode is similarly performed except that a section of a die is compared to another section in the same die having an identical structure. Array-inspection mode is used, for example, in testing devices with repeating structures such as memory cells. In lieu of comparing images from a wafer being tested, defects may also be detected by comparing an acquired test image with a known good image from a database.
A pixel from the test image can be different from a corresponding pixel in the reference image even if there are no defects in the two images. Intensity variations can be caused by, for example, differences in the physical layer structures, noise in the image acquisition electronics and signal paths, and varying noise modulation level within a single image across different gray levels. Thus, pixels in the difference image do not necessarily indicate that a defect exists. To differentiate real defects from false or “nuisance” defects, each pixel in the difference image is compared to a threshold window (
Finding the optimum threshold value for a given test image is an important but imprecise task. The threshold value must be chosen such that real defects are detected while differentiating nuisance defects. The narrower the threshold value, the more nuisance defects will be declared. Nuisance defects adversely affect production throughput because each defect event must be checked and verified. On the other hand, widening the threshold window will reduce nuisance defect events at the expense of letting real defects go undetected. Thus, a method for evaluating the effectiveness of a threshold or thresholding scheme is highly desirable.
The invention provides for a method and associated apparatus for relating a test image with a reference image. In an embodiment of the invention, the test and reference images are aligned. A two-dimensional scatter plot is then created by plotting the gray level of a test image pixel against the gray level of a corresponding reference image pixel for each aligned pixel location. The invention is applicable to electron-beam, bright-field, dark-field, laser, and atomic-force microscopy (“AFM”) inspection systems.
c show an alignment step in accordance with the present invention.
The present invention provides for a method and associated apparatus for relating the pixel of a test image with the corresponding pixel on a reference image. The invention can be used in determining the effectiveness of a threshold or thresholding scheme. The invention is also useful in other image processing applications such as those disclosed by the same inventor in the related co-pending U.S. patent application Ser. No. 09/365,503 filed Aug. 2, 1999, “Adaptive Mask Technique For Defect Inspection,” which is incorporated herein by reference in its entirety. Other uses for the invention are in electron-beam, bright-field, dark-field, laser, and atomic-force microscopy (“AFM”) inspection systems.
FIG. 3. illustrates the steps of an embodiment of the present invention. In step 310, a test image and a reference image of, for example, semiconductor structures are acquired using conventional image acquisition techniques. The images can also be acquired using the step-and-image acquisition system disclosed in commonly-owned U.S. patent application Ser. No. 09/226,967, “Detection of Defects In Patterned Substrates,” filed Jan. 8, 1999, which is incorporated herein by reference in its entirety.
In step 320, the test and reference images are aligned to match up corresponding pixels between the two images. A variety of alignment techniques can be used with the present invention including the technique disclosed in commonly-owned U.S. patent application Ser. No. 09/227,747, “Feature-Based Defect Detection,” filed Jan. 8, 1999, which is incorporated herein by reference in its entirety.
Step 320 is further illustrated in
Once the reference and test images are aligned, the pixel-to-pixel correspondence between the test image and the reference image is known. For each aligned pixel location, the gray level of a pixel from the test image is plotted against the gray level of the corresponding pixel in the reference image (
Table 3 shows that locations 434, 435, and 436 have varying gray levels and, thus, indicate the presence of possible defects. Locations 431, 432, and 433 are free of defects because the test image and the reference image have the same gray levels in said location. Scatter plot 500 (
Scatter plot 500 may be used to evaluate the effectiveness of a threshold or thresholding scheme. For example, a threshold window of ±40 gray level units may be plotted and superimposed on scatter plot 500 as shown by lines 502 and 503 in FIG. 5B. Line 502 represents all aligned pixel locations wherein the gray level of the test image is greater than the gray level of the reference image by 40 units. Similarly, line 503 represents aligned pixel locations wherein the gray level of the reference image is greater than that of the test image by 40 units. Aligned pixel locations outside lines 502 and 503, such as locations 435 and 436, will be declared as defect events. In
A pseudo code for implementing an embodiment of the invention in computer software is shown below. In the pseudo code, the gray level values are plotted in a memory array variable (“Scatter”). Appendix A lists the source code of a function written in the C programming language. On page 2 of Appendix A, “hist2D8” creates a two-dimensional scatter plot in accordance with the present invention. The code would be executed by a computer or processor which is conventionally coupled to or a part of a defect inspection system. Of course, such a system would typically store this source code and the resulting plots in a computer-readable medium (memory).
It is to be understood that the description given above is for purposes of illustration and is not intended to be limiting. Numerous variations are possible without deviating from the scope and spirit of the invention. The invention is set forth in the following claims.
Number | Name | Date | Kind |
---|---|---|---|
4511929 | Maeda et al. | Apr 1985 | A |
4783826 | Koso | Nov 1988 | A |
4823194 | Mishima et al. | Apr 1989 | A |
4886958 | Merryman et al. | Dec 1989 | A |
5033096 | Morrison et al. | Jul 1991 | A |
RE33894 | Bradley | Apr 1992 | E |
5291535 | Baker et al. | Mar 1994 | A |
5506793 | Straayer et al. | Apr 1996 | A |
5515453 | Hennessey et al. | May 1996 | A |
5568563 | Tanaka et al. | Oct 1996 | A |
5592236 | Rosenbaum et al. | Jan 1997 | A |
5638465 | Sano et al. | Jun 1997 | A |
5808735 | Lee et al. | Sep 1998 | A |
5943441 | Michael | Aug 1999 | A |
6178257 | Alumot et al. | Jan 2001 | B1 |
6252981 | Guest et al. | Jun 2001 | B1 |
6256093 | Ravid et al. | Jul 2001 | B1 |
6282309 | Emery | Aug 2001 | B1 |
6285397 | Webb et al. | Sep 2001 | B1 |
6288782 | Worster et al. | Sep 2001 | B1 |
6512849 | Yair et al. | Jan 2003 | B1 |
6539106 | Gallarda et al. | Mar 2003 | B1 |