Image inspection system for defect detection
First Claim
1. An image inspection system, comprisingcomparison means automatically responsive to the said image and to a desired form thereof whereby to detect abnormalities in the image and to produce data relating thereto,storage means operative to store the said data,first classifying means responsive to the stored data to classify each abnormality represented thereby as either being an abnormality of addition or an abnormality of omission, andsecond classifying means responsive to the said first classifying means and operative to classify the abnormalities of addition and the abnormalities of omission as being particular types of such abnormalities.
1 Assignment
0 Petitions
Accused Products
Abstract
A defect detection system for classifying defects in reticles (such as used in the manufacture of integrated circuit wafers) is described. An inspection unit, scanning the reticle in real time, compares it with the correct form and inputs signals representing each defect and its position in a store. The stored defects are then used to direct the inspection unit to inspect each defect more slowly. The inspection unit classifies each defect as either being an "excess metal" or a "missing metal" defect, and corresponding signals are stored in respective additional stores. The signals stored in these stores are then further processed and compared with the desired form of the reticle so as to classify the "excess metal" defects as either being a "pin spot" defect, an "extension" defect or a "bridge-type" defect. Similarly, the "missing metal" defects are classified as either being a "pinhole" defect an "intrusion" defect or a "break-type" defect. A size measuring unit measures the size of each defect.
70 Citations
13 Claims
-
1. An image inspection system, comprising
comparison means automatically responsive to the said image and to a desired form thereof whereby to detect abnormalities in the image and to produce data relating thereto, storage means operative to store the said data, first classifying means responsive to the stored data to classify each abnormality represented thereby as either being an abnormality of addition or an abnormality of omission, and second classifying means responsive to the said first classifying means and operative to classify the abnormalities of addition and the abnormalities of omission as being particular types of such abnormalities.
-
3. An image inspection system, comprising
comparison means responsive to the said image and to a desired form thereof and operative to carry out a rapid inspection of the whole image so as to detect the existence of any abnormalities therein and at least the approximate position of those abnormalities by comparing the actual image with the desired form thereof, storage means responsive to the comparison means and connected to store signals corresponding to the detected abnormalities and at least their said approximate positions, and inspection means responsive to the stored signals and operative less rapidly to inspect the said detected abnormalities and to classify those abnormalities as to type and to measure their sizes, the said inspection means comprising means operative to classify each abnormality as either being an abnormality of addition or an abnormality of omission, and means operative to classify the abnormalities of addition and the abnormalities of omission as being particular types of such abnormalities.
-
6. An image inspection system for inspecting an image in the form of a desired predetermined pattern of image paths with spaces between them, comprising
comparison means for automatically inspecting the image, the comparison means comprising means for comparing the said image with the desired form of the image and producing data identifying the position and size of each abnormality of addition and of each abnormality of omission, storage means for storing the said data, first classifying means responsive to the data identifying each abnormality of addition for automatically classifying each such abnormality as an abnormality of addition in a said space and unconnected with any said path, an abnormality of addition in a said space and connected to one said path, or an abnormality of addition in a said space and connected to two said paths, and second classifying means responsive to the said data identifying each abnormality of omission for automatically classifying each abnormality of omission as an abnormality of omission within a said path but not completely breaking the path, an abnormality of omission in an edge, only, of a said path, or an abnormality of omission within and completely breaking a said path.
-
9. A method of inspecting an image, comprising the steps of
rapidly comparing the whole image with a desired form thereof so as to detect the existence of any abnormalities therein and at least the approximate position of those abnormalities and producing data relating to the existence of those abnormalities and their said approximate positions, storing the said data, processing the stored data so as to carry out a less rapid inspection of those parts of the image where abnormalities have been detected by the comparing step to classify those abnormalities as to type and to measure their sizes, classifying each detected abnormality as either being an abnormality of addition or an abnormality of omission, and classifying the abnormalities of addition and the abnormalities of omission as being particular types of such abnormalities.
-
11. A method of inspecting an image in the form of a predetermined pattern of image paths with spaces between them, comprising the steps of
automatically comparing the image with a desired form of the image whereby to produce data identifying the position and size of each abnormality of addition and of each abnormality of omission, storing the said data, automatically processing the said data to classify each abnormality of addition as being either an abnormality of addition in a said space and unconnected with any said path, an abnormality of addition in a said space and connected to one said path, or an abnormality of addition in a said space and connected to two said paths, and automatically processing the said data to classify each abnormality of omission as being either an abnormality of omission within a said path but not completely breaking the path, an abnormality of omission in an edge, only, of a said path, or an abnormality of omission within and completely breaking a said path.
Specification