Image analysis – Applications – Manufacturing or product inspection
Reexamination Certificate
2002-07-23
2004-12-07
Mehta, Bhavesh M. (Department: 2625)
Image analysis
Applications
Manufacturing or product inspection
C438S014000, C356S237100
Reexamination Certificate
active
06829381
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to detection of defects, especially for detection of defects on the surface of semiconductor devices.
BACKGROUND ART
Current demands for high density and performance associated with ultra large scale integration require submicron features, increased transistor and circuit speeds and improved reliability. Such demands require formation of device features with high precision and uniformity, which in turn necessitates careful process monitoring, including frequent and detailed inspections of the devices while they are still in the form of semiconductor wafers.
Conventional in-process monitoring techniques employ a two phase “inspection and review” procedure. During the first phase the surface of the wafer is inspected at high-speed and relatively low-resolution. The purpose of the first phase is to produce a defect map showing suspected locations on the wafer having a high probability of a defect. During the second phase the suspected locations are more thoroughly analyses. Both phases may be implemented by the same device, but this is not necessary.
The two phase inspection tool may have a single detector or multiple detectors. A multiple detector two phase inspection device is described in U.S. Pat. Nos. 5,699,447, 5,982,921 and 6,178,257B1 of Alumot (hereinafter collectively referred to as the Alumot system) whose contents are hereby incorporated herein by reference.
During the first phase (also referred to as the inspection phase) the Alumot system (a) obtains an inspected pixel, neighboring inspected pixels and a reference pixel, (b) determines the type of the inspected pixel and/or determined the reference pixel type, (c) compares between the inspected pixel and the reference pixel and a threshold that depends upon the inspected pixel type, (d) and determines a presence of a defect in response to said comparison. The step of determining the type of a pixel involves a first stage of determining the following parameters: (i) local maxima—whether the pixel is a local maxima (if the pixel is a maximum relative to his neighbors), (ii) intensity—if the pixel is intense (if the intensity of the pixel is significant relative to a threshold), (iii) ratio—what is the ratio between the intensity of a pixel and the intensity of its neighbors relative to a threshold) and (iv) gradient—whether the pixel is located in a slope area relative to a threshold. The second stage involves classifying the pixel to one of the following types, in response to the parameters: (I) isolated peak (if the pixel is a local maxima with significant intensity and ratio), (II) multipeak (if the pixel is not an isolated peak, it has significant intensity and none of its neighbors is an isolated peak), (III) slope—if either one of the pixel's neighbors is an isolated peak or has significant gradient, or (IV) background—if the pixel has no significant intensity or gradient and none of its neighbors is an isolated peak.
Accordingly, the defect detection is responsive to the typing of the inspected pixel as the thresholds are responsive to the type of the pixel. Various errors in the measurement process, especially in noisy environment, and/or measurement inaccuracies may effect the type determination and may result in erroneous defect detections. Furthermore, the set of a pixel's types are fixed. Thus the types cannot be dynamically adjusted or tailored according to an end-user requirement.
The reference pixel is usually obtained from a previous inspection of another die, either from the same wafer or not. A comparison between pixels that were obtained from different dies, especially when the inspection tool parameters could change, may also result in erroneous determinations.
There exists a need for an improved and more robust method for inspecting a substrate, and especially a semiconductor substrate, for defects.
There exists a further need for a method for inspecting a substrate that allows a dynamic definition of pixel types.
SUMMARY OF THE INVENTION
The invention provides a method for inspecting a substrate for defects, the method includes the steps of: (a) obtaining an inspected pixel and a reference pixel; (b) calculating an inspected value and a reference value, the inspected value representative of the inspected pixel and the reference value representative of the reference pixel; (c) selecting a threshold in response to a selected value out of the inspected value and the reference value; and (d) determining a relationship between the selected threshold, the reference value and the inspected value to indicate a presence of a defect. The relationship may reflect a displacement between a pair of inspected pixel and reference pixel to the threshold. Step (d) may also include comparing between the selected threshold and a difference between the inspected value and the reference value. Ideally, at the absence of a defect, the inspected pixel and the reference pixel are equal.
The invention provides a method for inspecting a substrate for defects, the method including the steps of: (a) obtaining an inspected pixel; (b) calculating an inspected value representative of the inspected pixel; (c1) determining whether to obtain a reference pixel from a same die of the inspected pixel or from a reference pixel from another die; (c2) obtaining a reference pixel in response to the determination; (c3) calculating an inspected value representative of the inspected pixel; (d) selecting a threshold in response to a selected value out of the inspected value and the reference value; and (e) comparing the selected threshold to a difference between the reference value and the inspected value to determine the presence of a defect. Steps (a)-(b) allow for determining whether the reference pixel may be obtained from the same die. Obtaining a reference pixel from the same die as the inspected pixel and utilizing the same inspection tool and especially during the same inspection may compensate for changes in the working environment, for measurement inaccuracies and the like. It is also noted that obtaining the reference pixel from the same die may simplify the process of obtaining the reference pixel, as the reference pixel may be obtained during the same scan of the measurement unit. Usually a reference pixel may be obtained from the same die, or even from the vicinity of the inspected pixel if the inspected pixel is located in a background area. Changes of gray levels at neighboring pixels within a background area usually indicate a defect.
The step of selecting a value out of the inspected value and the reference value can be responsive to an evaluation of the noise associated with each of the inspected and reference values such to select the value that is less noisy. In many cases the less noisy value is the lower value. The selection may involve evaluating a first probability that the inspected value is erroneous and evaluating a second probability that the reference value is erroneous and selecting the value that is associated with a lower probability out of the first and second probabilities.
The inspected value is representative of the inspected pixel and/or neighboring inspected pixels. The reference value is representative of the reference pixel and/or neighboring reference pixels. The inspected value may represent a comparison between the inspected pixel and neighboring inspected pixels. The reference value may represent a comparison between reference pixel and neighboring reference pixels.
The inspected value can represent at least one of the following parameters: a gray level of the inspected pixel; a ratio between the gray level of the inspected pixel and between a combination of gray levels of at least one inspected neighboring pixel; a difference between the gray level of the inspected pixel and a combination of gray levels of at least one inspected neighboring pixel. The inspected value can reflect a difference between a maximal gray level of a neighboring inspected pixel and a minimal gray level of a neighboring inspected pixel. The inspected value can
Cohen Yehuda
Levin Evgeni
Applied Materials Inc.
Carter Aaron
McDermott & Will & Emery
LandOfFree
Method for detecting defects does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for detecting defects, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for detecting defects will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3330132