JP3114684B2 - Appearance inspection method and device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an appearance inspection method and apparatus, and more particularly to an appearance inspection method and apparatus for detecting foreign matter on a photomask or reticle for a semiconductor device lithography or a minute shape defect of a chrome pattern using an image. About.

[0002]

2. Description of the Related Art Defects in photomasks and reticles include:
Foreign matter that adheres after the completion of the manufacturing process and is relatively easy to remove by washing, etc., and chrome pattern formed in a shape different from the design in the manufacturing process such as drawing and etching, pattern shape defects that are relatively difficult to remove is there.

[0003] As a typical method of a conventional visual inspection method and apparatus for detecting foreign matter using images, US Pat. No. 5,556,702 to David G. Emery et al. is there. This is a method of detecting foreign matter by focusing on the optical difference between chromium and glass and foreign matter. Specifically, in a feature amount space in which pixel values of a filtered image such as a transmitted image and a reflected image, and a pixel value of an image subjected to filtering such as a differential image, a region of glass or chrome is obtained in advance. Pixels that come outside the area are regarded as foreign substances.

As a typical technique for detecting a shape defect of a chrome pattern using an image, Donald F.
US Patent No. to Donald F. Specht et al.,
According to No. 4805123, there are a die-to-die method in which an adjacent chip is compared with a template by template matching, and a die-to-database method in which a reference image is generated from CAD data and compared with the reference image.

In the die-to-die method, if there is a similar defect at the same coordinates of adjacent chips, it cannot be found. The die-to-database method was developed to detect defects even in such a case. However, since it is difficult to generate a reference image from CAD data with high accuracy, the die-to-database method has a defect detection capability.
It may be inferior to the die-to-die method.

[0006]

Each of the above-described conventional visual inspection methods and apparatuses has problems. The first method of David G. Emery et al. Has a problem that foreign substances having the same optical properties as the chromium pattern, such as chromium chips on glass, cannot be detected. In the second method of Donald F. Specht et al., It is necessary to simulate the process of generating a reticle image from CAD data. However, the actual process of generating a reticle is complicated. Since the imaging optical system is also complicated, it is very difficult to generate a reference image from CAD data, and there is a high possibility that a pseudo defect will be mistakenly recognized as having a defect even if the reticle has no defect. In spite of no defect, the density gradient near the edge between the inspection target image and the reference image is shown in FIG.
In the case where the distribution diagram (a) is different from the distribution diagram (b) in FIG.
A pseudo defect occurs as shown in FIG. Further, the method of Donald F. Specht et al. Has a problem that a large amount of calculation time and a storage area are required for generating and holding an image from CAD data.

An object of the present invention is to enable detection of a foreign substance having the same optical properties as a chromium pattern, such as chromium chips on glass, and that the density gradient near the edge differs for each image to be obtained. It is an object of the present invention to provide a visual inspection method and apparatus which can reduce the frequency of occurrence of pseudo defects, and can reduce the required calculation time and storage area without the need to generate and hold an image from CAD data.

[0008]

According to a visual inspection method of the present invention, an input image taken from the outside is defined as a pattern composed of a plurality of edge pieces, the area of a closed region defined by the plurality of edge pieces, and The edge piece includes a distance between edges for each set of parallel edge pieces among the plurality of edge pieces, and a distance between a bending point that is a connection point of the edge pieces and an adjacent bending point. A foreign substance or a pattern shape defect is detected by determining the presence or absence of a foreign substance or a defect based on a predetermined reference value for a partial shape characteristic of the pattern.

In the appearance inspection method of the present invention, when a foreign substance or a pattern shape defect is detected, the coordinates of the foreign substance or the pattern shape defect are marked on the input image, and the input image marked with the coordinates is stored. Good.

According to the appearance inspection method of the present invention, since an input image taken from the outside is defined as a pattern composed of a plurality of edge pieces, a pixel having a high differential intensity is detected as an edge element from the input image and the edge element image is detected. Generating and giving the same label to each edge element on the edge element image whose mutual distance is within a predetermined reference value range, connecting the edge elements to each other, and providing the same label with the same edge An edge image is generated by regarding a set of elements as one edge, a curvature is calculated for each of the edge elements on the edge image, and an edge element having a curvature larger than a predetermined reference value among the edge elements is defined as a bending point. A label different from the edge element which is the bending point is assigned to the edge elements on both sides of the edge element which is the bending point. By dividing the edge, each divided may be an edge piece.

According to the present invention, there is provided an appearance inspection apparatus comprising: edge detection means for detecting a pixel having a high differential intensity as an edge element from an input image input by an external image input means to generate an edge element image; Edge connecting means for connecting the edge elements of the same label to generate an edge image as one edge by giving the same label to those whose distance between the edge elements is within a predetermined reference value range; Comparing the shape characteristic of the pattern on the edge image determined from the distance between the edge element portions parallel to the curvature of each edge element and the area of the closed region by the closed loop edge with a predetermined reference value,
A shape identification unit that identifies the position of the edge indicating a value outside the range of the predetermined reference value as the existence position of the foreign matter or the shape defect, identifies the coordinates of the existence position of the foreign matter or the shape defect, and outputs the type of the defect; A mark on the coordinates of the location of the foreign matter or shape defect output by the shape identification means on the input image, and an image holding means for storing the input image marked with the coordinates.

[0012] In the visual inspection apparatus of the present invention, the shape discriminating means calculates a curvature for each edge element on the edge image, and, among the edge elements, an edge element whose curvature is larger than a predetermined reference value. Edges are divided by giving different labels to the edge element that is the inflection point and the edge element sandwiching the inflection point, and each of the fragments of the edge divided for each inflection point is used as an edge piece. Dividing means; calculating a distance between bending points whose distances are within a predetermined reference value for each bending point on the edge image; and storing a distance between the bending points; and calculating a distance between bending points on the edge image. The angle formed between edge pieces whose distance is within a predetermined reference value for each of the edge pieces is determined, and if this angle is smaller than a predetermined limit angle, the angle is determined. A parallel edge detecting means for storing coordinates of the edge pieces assuming that they are parallel to each other, and an edge for obtaining a distance between the edge pieces for each set of parallel edge pieces detected by the parallel edge detecting means and storing the value. Inter-distance calculating means, closed area detecting means for performing edge tracking for each edge on the edge image to detect edges forming closed areas, and calculating an area for each closed area detected by the closed area detecting means An area calculating means for storing this value, a bending point having a distance between the bending points stored in the distance between bending points calculation means smaller than a threshold value given from the outside, and a storage of the distance between edge calculation means. The parallel edge whose distance between edges is smaller than a predetermined reference value and the closed region whose area stored by the area calculating means is smaller than a predetermined threshold value are parallel to the bending point. Regarded di and closed areas to be foreign matter or pattern defect may comprise a shape parameter determining means for storing the type of coordinates and defect areas were considered to be foreign object or pattern shape defect.

[0013]

Next, embodiments of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing a first embodiment of the present invention.

The visual inspection apparatus 1 according to the first aspect of the present invention detects a pixel whose differential intensity is larger than a predetermined threshold from an input image input from the external image input means 6 as an edge element. The edge detecting means 2 for generating an image and the edge elements on the edge element image are connected to each other by giving the same label to each of the edge elements whose distances are within a predetermined reference value range. Edge connecting means 3 for generating an edge image as an edge
And, the shape characteristic of the pattern on the edge image determined from the distance between the edge element portions parallel to the curvature of each edge element and the area of the closed region by the closed loop edge is compared with a predetermined reference value, and A shape identification means 4 for identifying the position of the edge indicating a value outside the range of the determined reference value as the existence position of the foreign matter or the shape defect, identifying the coordinates of the existence position of the foreign matter or the shape defect and the type of the defect, and outputting the input; An image holding means 5 for marking the coordinates of the position of the foreign matter or shape defect output by the shape identification means on the image and storing the input image with the coordinates marked.

FIG. 2 is a block diagram showing a second embodiment of the present invention.

In the second embodiment of the present invention, the function of the shape identifying means 4 in the first embodiment of the present invention is subdivided and clarified. For this reason, the same portions as those in the configuration of FIG.

Next, the function of the visual inspection apparatus according to the second embodiment will be described in detail with reference to FIG. The description of the same parts as those in the first embodiment is omitted.

In the appearance inspection apparatus 7 of the second embodiment, the shape identification means 8 calculates the curvature for each edge element on the edge image, and the curvature calculation means 9 determines the curvature among the edge elements in advance. An edge element larger than the threshold value is defined as an inflection point, and the edge element that is the inflection point and the edge element sandwiching the edge element are given different labels to divide the edge. An edge dividing unit 10 that uses each of them as an edge piece, a bending point distance calculating unit 11 that calculates a distance between bending points whose mutual distance is within a predetermined reference value for each bending point on the edge image, For each edge piece on the image, the distance between each other is within a predetermined reference value, the angle between them is determined, and if this angle is smaller than a predetermined limit angle, The edge pieces are regarded as parallel, and the coordinates are stored. The parallel edge detecting means 12 stores the coordinates, and the distance between the edge pieces is determined for each set of the parallel edge pieces detected by the parallel edge detecting means 12, and the value is calculated. An inter-edge distance calculating means 13 for storing, a closed area detecting means 14 for performing edge tracking for determining an edge shape for each edge on the edge image and detecting an edge forming a closed area; An area calculating means 15 for calculating an area for each detected closed region and storing this value; and a bending point distance calculating means 1
1 and a distance between the inflection points stored in the inflection point smaller than a threshold value given from outside;
For a parallel edge whose distance between edges stored is smaller than a predetermined reference value and a closed region whose area stored by the area calculating means 15 is smaller than a predetermined reference value, a bent point, a parallel edge, and a closed region Is regarded as a foreign matter or a pattern shape defect, and the shape parameter determining means 16 stores the coordinates of the place where the foreign matter or the pattern shape defect is considered to exist and the type of the defect.

Here, it is assumed that the closed area detecting means 14 detects an edge forming a closed area by tracing an adjacent edge element on the edge image for each edge on the edge image and performing edge tracing. How to follow the edge element of time,
That is, various methods have been proposed as a method of tracking an edge.

As a method applicable to this embodiment, for example, as a method that can be summarized as a tracking method, first, the magnitude of the threshold is set large so as not to detect noise, and an edge having a strong contrast is set. Only detect. Next, the threshold is lowered to detect weak edges.
At this time, instead of searching the entire screen, a two-step processing method in which only the extension on the already detected strong edge is searched for, or only the strong edge is traced by increasing the threshold value of the edge detection There is a method of applying a scaling method to detect a noise component and detect an edge. Further, since the object to be detected is formed of a geometrically regular shape such as a straight line or an arc, it is possible to use the Hough transform method. As another applicable method, the above-mentioned tracking method locally determines the next point to be advanced, but on the other hand, finds a globally optimal line considering the characteristics of the line to be extracted A method is also conceivable. For example, a dynamic programming method introduced by Montanari to find this optimal line may be used. Still another method is that a feature value obtained by a local operation is not always reliable due to the influence of noise or the like. There is a relaxation method as a way to obtain consistent results. Although the relaxation method itself has been used as a numerical solution of simultaneous equations, it is currently used in various fields of image processing.

Next, the function and operation of the embodiment of the present invention will be described in detail. 1 to 8 are appropriately used for the description. 1 to 3 have already been used for the description, but FIGS. 4 to 8 are not described yet, so will be briefly described. FIG. 4 is an explanatory view showing an example of an isolated defect or a foreign substance for explaining a case where a pattern shape defect or a foreign substance is present in a glass region where there is no pattern. FIG. 5 is an explanatory view showing an example in which a hook / projection defect occurs on a boundary between a chrome region and a glass region which is originally a straight line. FIG. 6 is an explanatory diagram showing an example of a bridge / disconnection defect generated between chromium regions that are not originally connected. FIG. 7 is an explanatory diagram showing a line defect generated by protruding the screen into the glass area. FIG. 8 shows a bridge / disconnection defect having a smaller curvature than that of FIG. Note that these drawings (FIGS. 4 to 8) all have a relationship between a negative image and a positive image.
For example, what is shown as a chipped defect in FIG. 5 is a projection defect from a different viewpoint, and FIGS. 6 and 8 are explanatory diagrams showing a disconnection defect. Both are equivalent as shape recognition procedures.
There is no need to distinguish.

The visual inspection apparatus 1 of the present invention convolves, for example, a Laplacian Gaussian filter with an input image obtained from an image input means 6 constituted by an external camera or the like to generate an image in which edges are emphasized, and generates a luminance of this image. A pixel whose value crosses a value of zero is stored as an edge element and the coordinates of that pixel are stored. For example, an integer of 1 or more is given as a pixel value to each edge element so that the pixel value differs for each edge element. An edge detecting means 2 for generating an edge element image in which a pixel value 0 is given to a pixel; and for each edge element in the edge element image, for example, when the distance between the edge elements is smaller than a predetermined value. By making the pixel values the same, the edge elements are connected, and a group of edge elements given the same pixel value becomes 1
Edge connection means 3 for generating an image which can be regarded as one edge as an edge image, and for example, the curvature of the pattern is
As shown in FIG. 5 and FIG. 6, a point having a large curvature is closer to a distance smaller than a predetermined threshold value, for example, as shown in FIG. 6, FIG. 7, and FIG. The distance between edges is smaller than a predetermined threshold value as shown in a portion indicated by a circle, or the area of a closed region is smaller than a predetermined threshold value such as a portion shown as a small area in FIG. If an abnormality such as small is observed,
Assuming that there is a foreign substance or a shape defect at the position of the edge, the shape identification means 4 for transferring the coordinates and the type of the defect to the image holding means 5, and the presence of the foreign substance or the shape defect given from the shape identification means 4 on the input image An image holding means 5 is provided for marking the pixel having the coordinates to be set, for example, by setting the pixel value to a specific pixel value given from the outside which is different for each type of defect, and storing the marked image. You.

Next, in the appearance inspection apparatus 7 of the present invention, the shape discriminating means 8 includes a curvature calculating means 9 for calculating a curvature for each edge element which is a pixel constituting an edge on the edge image;
Among the edge elements, an edge element having a curvature larger than a predetermined threshold is defined as a bending point, and the edge is divided by giving different labels to the edge element that is the bending point and the edge element sandwiching the edge element, thereby dividing the edge element. An edge dividing means 10 for making each edge piece, a distance between bending points having a mutual distance within a predetermined reference value for each bending point on an edge image, and calculating a distance between bending points for storing this value Means 11 and, for each edge piece on the edge image, determine the angle between the edge pieces whose distance is within a predetermined reference value, and if this angle is smaller than a predetermined limit angle, Parallel edge detecting means 12 which regards the pieces as parallel and stores their coordinates
For each set of parallel edge pieces detected by the parallel edge detection means 12, for example, the distance between an edge element selected from one edge piece and an edge element selected from the other edge piece is determined, and all edges are determined. Edge distance calculating means 13 for storing the minimum distance among the distances determined for combinations of elements as the distance between edge pieces.
And a closed area detecting means 14 for tracing an adjacent edge element on the edge image for each edge on the edge image to detect an edge forming a closed area, and a closed area detecting means 14 The distance between the bending points stored in the area calculating means 15 for calculating the area for each closed region and storing the calculated value and the bending point shown in FIG. If the distance between the two points is smaller than a predetermined threshold value, as in the case of a point, the distance between the edges of the inflection point and the distance between the edges stored in the distance between the edges calculating means 13 are, for example, as shown in FIGS. In the case where the distance between the edges is smaller than a predetermined threshold value as in the case of the edge having a small edge-to-edge distance shown in FIG. Shown For a closed area smaller than a predetermined threshold value, such as a portion where the edge is bent, the inflection point or the parallel edge or the closed area is regarded as a foreign substance or a pattern shape defect. And the shape parameter determining means 16 for storing the coordinates and the type of the defect.

[0025]

As described above, the present invention does not use optical features such as light transmittance and reflectance, and is constituted by edges such as the area of a closed region, the distance between edges, and the distance between bending points. Since the presence or absence of a defect is determined only based on the shape of the pattern, a foreign substance having the same optical characteristic amount as the chrome pattern, such as a chromium foreign substance on glass, can be detected.
Also, unlike the traditional method of comparing multiple images,
In one image, the presence or absence of a defect is determined based only on the shape of a pattern composed of edges, such as the area of a closed region, the distance between edges, and the distance between bending points. Even if the concentration gradient is different, there is no effect at all,
There is an effect that a pseudo defect due to a difference in concentration gradient does not occur. In addition, by setting a threshold value for the area of the closed region, there is an effect that an isolated defect or a foreign substance can be easily detected. Further, by setting a threshold value for the distance between the bending points, there is an effect that a chipping / projection defect, a bridge / disconnection defect can be detected. Similarly, by setting a threshold value for the distance between parallel edges, there is an effect that a line defect protruding from the screen and a bridge / disconnection defect having a relatively small curvature at the base can be detected. INDUSTRIAL APPLICABILITY The present invention can be applied not only to detection of a foreign matter on a photomask or reticle for lithography of a semiconductor device or a minute shape defect of a chrome pattern but also to defect inspection of a semiconductor product such as a wafer with the same apparatus configuration.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a first embodiment of the present invention.

FIG. 2 is a block diagram showing a second embodiment of the present invention.

FIG. 3 is an explanatory diagram illustrating an example of an edge having a different density gradient near an edge.

FIG. 4 is an explanatory view showing an example of an isolated defect or a foreign substance for explaining a case where a pattern shape defect or a foreign substance is present in a glass region where there is no pattern.

FIG. 5 is an explanatory diagram showing an example in which hooking / projection defects occur on a boundary between a chrome region and a glass region which is originally a straight line.

FIG. 6 is an explanatory view showing an example of a bridge / disconnection defect generated between chromium regions that are not originally connected.

FIG. 7 is an explanatory diagram showing a line defect generated by protruding a screen out of a glass area.

FIG. 8 is an explanatory diagram showing a bridge / break defect having a small curvature.

[Explanation of symbols]

 1,7 Appearance inspection device 2 Edge detecting means 3 Edge connecting means 4,8 Shape discriminating means 5 Image holding means 6 Image input means 9 Curvature calculating means 10 Edge dividing means 11 Distance between bending points calculating means 12 Parallel edge detecting means 13 Edge Distance calculating means 14 closed area detecting means 15 area calculating means 16 shape parameter determining means

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁷ , DB name) G01N 21/84-21/958 G03F 1/00-1/16 G06T 1/00-9/40 H01L 21 / 64-21/66

Claims (5)

(57) [Claims] 1. An input image to be fetched from the outside is defined as a pattern composed of a plurality of edge pieces, an area of a closed region defined by the plurality of edge pieces, and an area of a parallel edge piece among the plurality of edge pieces. Predetermined for a partial shape characteristic of a pattern composed of the edge pieces including a distance between edges of each pair and a distance between a bending point which is a connection point of the edge pieces and an adjacent bending point. An appearance inspection method characterized by detecting foreign matter or a pattern shape defect by determining the presence or absence of a foreign matter or a defect based on a reference value. 2. The method according to claim 1, wherein upon detecting a foreign substance or a pattern shape defect, a mark is placed on the coordinates of the foreign substance or the pattern shape defect on the input image, and the input image with the coordinates marked is stored. Visual inspection method described. 3. An edge element image is generated by detecting a pixel having a high differential intensity as an edge element from the input image in order to define an input image captured from the outside as a pattern composed of a plurality of edge pieces. The edge elements are connected by giving the same label to each of the edge elements on the image whose mutual distance is within a predetermined reference value range. Generate an edge image as an edge, calculate a curvature for each of the edge elements on the edge image, and among the edge elements, use an edge element whose curvature is larger than a predetermined reference value as a bending point. Dividing the edge by giving a label different from the edge element being the inflection point to the edge elements on both sides with a certain edge element as a boundary, The appearance inspection method according to claim 1, wherein each of the divided pieces is an edge piece. 4. An edge detecting means for detecting a pixel having a high differential intensity as an edge element from an input image inputted by an external image input means to generate an edge element image; Edge connection means for generating an edge image as one edge by connecting the edge elements of the same label by giving the same label to those whose distance is within a predetermined reference value range;
The shape characteristic of the pattern on the edge image determined from the distance between the edge element portions in parallel with the curvature of each edge element and the area of the closed region by the closed loop edge is compared with a predetermined reference value, Shape identifying means for identifying the position of the edge indicating a value outside the range of the predetermined reference value as the existence position of the foreign matter or the shape defect, identifying and outputting the coordinates of the existence position of the foreign matter or the shape defect and the type of the defect, and And an image holding means for marking the coordinates of the position of the foreign matter or shape defect output by the shape identification means on the input image, and storing the input image marked with the coordinates. Inspection equipment. 5. A shape discriminating means for calculating a curvature for each edge element on an edge image, and among the edge elements, an edge element having a curvature larger than a predetermined reference value is defined as a bending point, Edge dividing means for dividing an edge by giving different labels to an edge element that is a point and an edge element sandwiching the point, and using each of the edge fragments divided for each bending point as an edge piece; For each inflection point on the image, determine the distance between the inflection points whose mutual distance is within a predetermined reference value and store this value between the inflection point distance calculating means, and for each edge piece on the edge image, For the edge pieces whose distance between each other is within a predetermined reference value, the angle formed by the edge pieces is determined. If this angle is smaller than a predetermined limit angle, the edge pieces are considered to be parallel. Parallel edge detection means for storing the deemed coordinates, edge distance calculation means for obtaining the distance between edge pieces for each set of parallel edge pieces detected by the parallel edge detection means and storing the value, and Closed area detecting means for performing edge tracking for each edge on the edge image to detect an edge forming a closed area, and calculating an area for each closed area detected by the closed area detecting means and storing the value Calculating means, a bending point where the distance between the bending points stored by the bending point distance calculating means is smaller than a threshold value given from the outside, and the edge distance stored by the edge distance calculating means is For a parallel edge smaller than a predetermined reference value and a closed area whose area stored by the area calculating means is smaller than a predetermined threshold, the bent point, the parallel edge and the closed area are separated by a foreign substance or Regarded as a pattern shape defect, the appearance inspection apparatus according to claim 4, characterized in that it comprises a shape parameter determining means for storing the type of coordinates and defect areas were considered to be foreign object or pattern shape defect.