Image analysis – Applications – Manufacturing or product inspection
Reexamination Certificate
2000-02-01
2003-12-16
Tran, Phuoc (Department: 2621)
Image analysis
Applications
Manufacturing or product inspection
C382S278000
Reexamination Certificate
active
06665432
ABSTRACT:
FIELD OF THE INVENTION
The present invention is related to the inspection of a first device pattern by comparing that device pattern with a second device pattern where those patterns may be on different dies, different device patterns on the same die, repeating patterns within the same device, on paper, or stored in memory. More particularly, the present invention relates to device pattern inspection where the pattern to be inspected is a random pattern, a repeating pattern, or a combination of the two.
BACKGROUND OF THE INVENTION
The types of patterns that are found on devices, particularly those in the production of semiconductor devices, are categorized as either random, repeating, or a combination of the two. This categorization is important in that, historically, random and repeating patterns have been inspected optically in the spatial domain, whereas repeating patterns are more easily inspected in the frequency domain. Also, in the early semiconductor technologies the repeating pattern was not readily used. More recently, with the advent of VLSI technologies, memory arrays have become quite common, and therefore of interest in the rapid inspection of devices with repeating patterns.
There are numerous techniques used for spatial optical inspection of a first device by means of comparison of that device with a second device. These techniques use either a real second device, or the desired attributes of the device to be inspected stored in memory, against which the first device is compared. Where a real device is used for the comparison, there are a variety of techniques that permit one to compare the first device against a separate second device, to compare two dies on the same device, or to compare repeating arrays within the same die.
In each comparison it is noted whether the devices are the same, within any selected tolerances, or whether they are different. Typically, the second device then becomes the first and a third becomes the second and another comparison is performed again noting whether they are the same or different. In this way it is possible to determine which of the devices are good and which ones are bad since it is presumed that the majority of devices will be good, therefore those which are not in that group are presumed to be defective.
Most of the high speed inspection systems that are currently available inspect the pattern in the spatial domain, no matter what the characteristics of the pattern are. However, the inspection of repeating patterns, but not random patterns, is greatly facilitated by the performance of the inspection in the frequency domain.
In order to improve the inspection time of repeating patterns, In-systems, Inc. developed an entirely optical technique that uses a special lens system that is device specific (U.S. Pat. No. 4,806,774, issued Feb. 21, 1989). This system projects a two dimensional image through a lens which yields a Fourier transform of that image in the back aperture plane. Then, through the use of a hologram that is specific to the repeating pattern of the device being inspected, the In-Systems method filters out the harmonic frequencies from the resulting frequency domain image of the device, thus removing the frequency domain attributes of any repeating pattern from the other features of the device that is being inspected.
The In-Systems inspection system does the inspection entirely optically. In-Systems method passes the 2-dimensional image through a lens resulting in a two dimensional Fourier transform of the device image. That image is then directed to a photographic plate that has been developed by shining light on it from a test device, so that the higher the light intensity of the Fourier transform of the image the more that develops on the plate. The photographic plate serves as a filter that depends on the intensity of the image on the photographic plate, i.e. there is a direct correlation between the amount of light that fell on the photographic plate and the density of the image on the photographic plate. So in areas where there is a lot of density on the plate, it filters to the same extent. The In-Systems filter is limited to being just in the areas where the Fourier transform is strong which may include off harmonic areas. It is meant for harmonics but nevertheless because of the way that the filter is made it will filter out those frequencies.
The filter is developed from the light shining on it from a test device. During the inspection process, light from a device that is to be analyzed is directed through a Fourier transform structure as described above and stored in a hologram. A laser is then shone through the hologram and the same lens to cancel out any aberrations introduced by the lens. The point is that when In-Systems filters in the frequency domain they are constrained to filtering everywhere that the Fourier transform is significantly strong.
In the In-Systems optical approach the image of the device is not scanned and digitized to perform the inspection. This means that the In-Systems method described above can only look at arrays, since the transform techniques are not suitable for non-repeating patterns. If non-repeating patterns are inspected using the In-Systems approach, the spectral components in the frequency domain will be scattered and not produce a meaningful spectrum that presents a frequency pattern that can be processed.
It would be desirable to have an inspection, method and apparatus that combines inspection domains to inspect each type of pattern in the domain that is more favorable to the inspection of that type of pattern. That is use the frequency domain for the inspection of the repeating patterns (which is superior to spatial domain inspection of repeating patterns) and spatial domain inspection of the random patterns (for which frequency domain inspection is unsuitable). It would be of particular interest to have an inspection method for arrays with the increased precision, flexibility and reliability of electronic digital arithmetic without the inconvenience of holograms and chemical development for the particular pattern. The present invention provides such an inspection method and apparatus.
SUMMARY OF THE INVENTION
The hybrid technique of the present invention is basically a method for finding defects on digitized device images using a combination of spatial domain and frequency domain techniques. The two dimensional spectra of two images are found using Fourier like transforms. Any strong harmonics in the spectra are removed, using the same spectral filter on both spectra. The images are then aligned, transformed back to the spatial domain, and subtracted. The resulting spectrally-filtered difference image is thresholded and analyzed for defects.
Use of the hybrid technique of the present invention to process digitized images results in the highest-performance and most flexible defect detection system. It is the best performer on both array and random devices, and it can cope with problems such as shading variations and the dark-bright problem that no other technique can address.
The hybrid technique of the present invention also uses frequency domain techniques to align the images with more precision than spatial domain techniques such as the cubic shift. Further, the relative offsets of the pairs of images are determined by frequency domain techniques—and this method may be the most accurate and the least expensive.
There are three additional major benefits from the hybrid technique of the present invention:
1. The hybrid technique automatically processes both array and random areas on the device;
2. The hybrid technique is well-suited to the bright-dark problem (the array appears dark and low-contrast if the illumination is set for random, and the random saturates the camera, or sensor, if the illumination is adjusted for repeating) because the hybrid technique is much more powerful for repeating than random; and
3. The hybrid technique can do additional frequency-domain processing virtually for free—for example, the shading problem can be solved (low fre
Evans David M. W.
Lin Jason Z.
Tsai Bin-Ming Ben
LandOfFree
Inspection method and apparatus for the inspection of either... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Inspection method and apparatus for the inspection of either..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Inspection method and apparatus for the inspection of either... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3132247