JP4262288B2 - Pattern matching method and apparatus - Google Patents

Description

  The present invention relates to, for example, a pattern matching method and apparatus for performing pattern matching on a semiconductor wafer using design data, and more particularly to a system configuration, design data, and SEM that automatically generate a pattern photographing / inspection condition from design data. (Abbreviation for Scanning Electron Microscope) This invention relates to a method and apparatus for stably performing matching processing with an image.

  In recent years, the semiconductor industry has shifted from memory production to system LSI (abbreviation for Large Scale Integrated Circuit) production. When viewed as a pattern on a semiconductor wafer, the pattern of the system LSI is not created as a simple repetitive pattern, unlike the pattern of the memory. Therefore, when measuring a pattern of a system LSI with a length measuring SEM which is one of semiconductor evaluation apparatuses, it is necessary to frequently change a measurement position, that is, a matching template. Pre-design data such as CAD (Computer Aided Design) is used because taking images for template registration frequently during actual measurement significantly reduces the overall throughput. It is desirable to generate a template directly from In addition, since the wafer size is 300mm and it can no longer be transported by hand, and inspection in a high-purity clean room is becoming necessary, it is desired that the semiconductor factory be completely robotized. . Therefore, not only the template for positioning, but also all the conditions necessary for inspection, from design information to shooting conditions, points to measure, and length measurement algorithms, and operator-free to perform actual inspection under those conditions There is a need for a fully automated semiconductor inspection system.

  In the conventional length measurement SEM, registration of the point for image recognition, the length measurement position, and the length measurement algorithm is performed by photographing an actual wafer once and using it. In other words, an actual wafer is required, and the length measuring SEM must be temporarily occupied in order to take an SEM image and register various conditions. Moreover, the matching technology between the design data and the SEM image is not sufficient, and the matching cannot be performed with high accuracy. For example, when design data is used as a template in the prior art to specify the pattern position on a SEM image of a semiconductor wafer, the edge image is detected by filtering the SEM image using a Sobel filter or the like to detect edge components. It was created and matching was performed between the edge image and the design data, such as normalized correlation processing.

  Fig. 1 shows an outline flow of conventional processing and an image example used for processing. And FIG. Shown in First, in 101, a template of a pattern desired to be obtained from design data is registered. A pattern 701 registered from the design data is shown. Next, an SEM image is acquired at 102. The acquired SEM image is shown at 702. An edge enhancement filter such as a Sobel filter is applied to the SEM image acquired in 103. In 104, the image in which the edge is enhanced is binarized to obtain a line image in which only the edge is extracted. Reference numeral 703 denotes a line image extracted from the SEM image 702. In 105, matching processing such as normalized correlation with the design data registered in 101 is performed.

JP-A 64-54305 JP 2000-266706 A JP 2000-236007 A

  In the conventional semiconductor inspection system, registration of the point for image recognition, the measurement position, and the measurement algorithm is performed by photographing an actual wafer once and using it. For this reason, there is a problem that throughput is not improved because it takes time for registration and occupies the time device. In addition, there is a problem in that an operator is always required for a person to judge and register by looking at an actual SEM image, and an operator-free fully automated semiconductor inspection system cannot be constructed. Furthermore, in the matching technology between design information and SEM image, the conventional technology cannot cope with changes in the shape of CAD data and SEM image, and the S / N (signal / noise ratio) of image when extracting edge information from SEM image. ) Cannot sufficiently extract edge information, and when a line image is created by binarization, it is difficult to determine a threshold value, and an optimal value may not be obtained. Therefore, there has been a problem that the correlation coefficient becomes very small in the subsequent matching processing by normalized correlation.

  The present invention has been made in view of such a situation, creating all conditions necessary for inspection from design information such as CAD data to shooting conditions, length measurement points, and length measurement algorithms. Realizes an operator-free fully automated semiconductor inspection system that performs actual inspections at the same time, and performs matching processing with high correlation values when matching processing with SEM images using design data as a template in that system The method and apparatus which can be implemented are realized.

  In order to achieve the above object, a pattern matching method of the present invention includes a pattern matching method for performing pattern matching between design data and an image acquired by a scanning electron microscope, the edge of the design data, and the scanning. A matching process for each direction is performed between edges of an image acquired by a scanning electron microscope.

  Furthermore, the pattern matching apparatus of the present invention is a pattern matching apparatus that performs pattern matching between design data and an image acquired by a scanning electron microscope, and is acquired by an edge of the design data and the scanning electron microscope. Edge acquisition means for acquiring an edge of the image to be obtained, and pattern matching means for performing matching processing according to direction between the edge of the design data and the edge of the image.

  According to the present invention, it is possible to realize a method and an apparatus capable of executing a matching process having a high correlation value and stable.

  FIG. 3 is a block diagram of a schematic configuration of the scanning electron microscope system of the present invention. Reference numeral 301 denotes a mirror body part of an electron microscope, and an electron beam 303 emitted from an electron gun 302 is converged by an electron lens not shown in the drawing, and is irradiated onto a sample 305. By the electron beam irradiation, the intensity of secondary electrons or reflected electrons generated from the sample surface is detected by the electron detector 306 and amplified by the amplifier 307. A deflector 304 moves the position of the electron beam, and raster scans the electron beam 303 on the sample surface by the control signal 308 of the control computer 310. The signal output from the amplifier 307 is AD converted in the image processor 309 to create digital image data. Reference numeral 311 denotes a display device that displays the image data. The image processor 309 includes an image memory that stores digital image data, an image processing circuit that performs various image processing, and a display control circuit that performs display control. Input means 312 such as a keyboard and a mouse is connected to the control computer 310.

  An electron microscope apparatus is used to measure the line width of a fine pattern drawn on a wafer when creating a semiconductor device. At this time, the normalized correlation method is currently used as a method for finding a portion for measuring the line width on the wafer, and in this case, an optimal template selection is required. Since the image processor 309 of the present invention is configured to be able to optimally perform template selection in template matching, it can be applied to an electron microscope apparatus.

  FIG. 2 shows a matching processing flow using design data and an SEM image according to an embodiment of the present invention. First, in 201, a pattern portion to be detected from design data is registered as a template. An SEM image is acquired at 202 and matching processing is performed at 203. For this matching process, there are various techniques. For example, the same techniques (edge enhancement filter process, binarization process, normalized correlation process) as 103 to 105 in FIG. 1 may be used. As a result, a position on the SEM image corresponding to the design data pattern is detected at 204. Next, at 205, the part of the SEM image corresponding to the position detected at 204 is re-registered as a template. Thereafter, an SEM image is acquired at 206, matching processing is performed using the SEM image re-registered at 205 as a template, and position detection is performed at 208. By doing so, since the re-registered template is the SEM image, the matching process between the SEM gray images is performed, so that the matching process with a high correlation value and a stable detection rate can be performed. When a plurality of detections are performed, the process from 206 to 209 is repeated. If a template to be registered first is set in advance, the subsequent processing is automatically advanced by a computer program.

  FIG. 4 is a processing flow for re-registering an SEM image as a template over time according to an embodiment of the present invention. 401 to 408 correspond to 201 to 208 in FIG. In step 409, it is determined whether or not template re-registration is performed at a fixed time interval or processing number interval. If so, the matching process using the design data and the SEM image is performed again. Execute. By doing so, even if the SEM image in the middle of photographing changes with time, matching processing with high correlation value and stable detection rate can be performed.

  FIG. 5 is a processing flow for re-registering an SEM image as a template when a correlation value larger than the previous correlation value is obtained according to an embodiment of the present invention. Reference numerals 501 to 504 and 506 to 510 correspond to 401 to 404 and 405 to 409 in FIG. In 510, it is determined whether or not template re-registration is performed at a constant time interval or processing frequency interval, and if so, the matching process using the design data and the SEM image is performed again. Execute. Next, if the correlation value of the position detected this time in 505 is larger than the correlation value of the template up to the previous time, template re-registration of 506 is performed, but if it is smaller, re-registration is not performed and processing from 507 is started. . Therefore, the template to be used can be optimized to the template having the highest correlation value.

  FIG. 6 is a processing flow in the case where the design data and the SEM image are matched any number of times according to an embodiment of the present invention, and the SEM image at the highest correlation value is re-registered as a template. is there. Reference numerals 601 to 604 and 606 to 610 correspond to 201 to 204 and 205 to 209 in FIG. The matching process using the design data and the SEM image is repeated an arbitrary number of times from 602 to 605, and in S606, the SEM image at the position with the highest correlation value among the detected positions is re-registered as a template. Therefore, an SEM image template having a high correlation value can be selected. When performing a plurality of detections, the process from 607 to 609 is repeated using the template. Note that the processing of FIGS. 4 to 6 can also be automated by a computer program.

  FIG. 8 is a processing flow of matching processing between a bitmap data template and an SEM image according to an embodiment of the present invention. In step 801, edge information is extracted from each of the bitmap data and the SEM image. In general, an edge enhancement filter such as a Sobel filter is used for this portion of processing. In this part, both images lose the contrast information and are easily matched. However, since the SEM image has a considerably different shape from the actual CAD data, the matching detection rate is low as it is. Therefore, smoothing processing is performed on both images that have become edge images in 802 to compensate for the shape change between the two images. A slightly stronger smoothing filter is applied to this portion of processing. In addition, it is necessary to change the smoothness of CAD data and SEM images and to make CAD data smoother. In this way, since the edge images corrected for the shape deformed portions are matched at 803, matching processing with a high detection rate can be performed. Note that if the edge information is set in advance from the bitmap data to be extracted first and the SEM image, matching processing by a computer program can be automated.

  FIG. 9 is a processing flow of a matching process between a bitmap data template and an SEM image according to another embodiment of the present invention. A difference from the processing flow of FIG. 8 is that edge extraction in 901 is performed for each direction. The edge extraction process for each direction generally uses a Sobel filter that can extract edges for each direction. As directions, two directions of X and Y or four directions of X, Y, XY and YX are used. In 902, smoothing processing for compensating for the shape change portion is performed on each edge image decomposed in each direction. In 903, the images divided in each direction are combined and integrated as shown in FIG. By synthesizing in this way, in 903, processing can be performed as matching between one image. Of course, it is also possible to perform matching processing for each direction separately without performing the integration of 903. Thus, by extracting the edge for each direction and performing matching processing, the matching accuracy in each direction can be improved. Note that the original image in FIG. 10 corresponds to the template and the input SEM image in FIG. 9, and when obtaining the X-direction differentiation of these images, the original image is divided into a plurality of lines in the Y direction. Further, when obtaining the Y-direction differentiation, the original image is divided into a plurality of lines in the X direction. This matching process can also be automated by a computer program.

  FIG. 11 is a configuration diagram of a semiconductor inspection system according to an embodiment of the present invention. Reference numeral 1101 denotes a navigation system that stores design information of a semiconductor chip such as CAD data and can arbitrarily extract an area to be inspected from the design information. Reference numeral 1102 denotes a scanning electron microscope system that performs a predetermined inspection by actually photographing a semiconductor wafer using the information. These systems 1101 and 1102 are connected by a network and can exchange information and data.

  FIG. 12 is a configuration diagram of a navigation system according to an embodiment of the present invention. The navigation system 1101 extracts desired design data from the stored design information, and takes a bitmap data creation unit 1201 having a function of creating bitmap data, and an imaging / photography used in the scanning electron microscope system 1102 from the design data. The imaging / inspection condition editing unit 1202 has a function of editing and transmitting inspection conditions. The navigation system 1101 may be configured by dividing the functional parts of the bitmap data creation unit 1201 and the imaging / inspection condition editing unit 1202 in one WS (workstation) or PC (personal computer). The function may be divided into two or multiple WSs or PCs.

  FIG. 13 is a configuration diagram of a semiconductor inspection system according to an embodiment of the present invention. If the navigation system 1302 has its own semiconductor pattern design function and does not have a design function, it receives design information from another system 1301 connected via the network and having the design function, and uses that information.

  FIG. 14 is a network configuration diagram of a semiconductor inspection system according to an embodiment of the present invention. In the semiconductor inspection system of the present invention, the navigation system 1401 can exchange data with other navigation systems 1402 to 1404 connected to the network of the installation facility, and further includes a plurality of scanning electron microscopes connected to the network. Imaging / inspection conditions can be sent to systems 1405-1406. By doing so, it is possible to share a shooting / inspection condition and simultaneously simultaneously operate a plurality of systems.

  FIG. 15 is a display example in the navigation system according to one embodiment of the present invention. In the navigation system, semiconductor design data 1501 is stored. When the operator inputs design information such as 1502 layers and cells, the designated portion 1502 is extracted from the design data 1501 and is displayed as 1503. Has a function to display on the display screen. In this case, 1501 may be on a design system connected by a network as shown in FIG.

  FIG. 16 is an example of bitmap data created by the navigation system 1302 according to one embodiment of the present invention. Reference numeral 1601 denotes an area extracted from the design data in FIG. 15, and a portion to be inspected / measured like 1602 is designated from within this area. In this case, the inspection / length measurement designation area 1602 is converted into bitmap data such as 1603 and sent to the scanning electron microscope system 1303. Here, black and white binary values are used as the bitmap data 1603, but this color can be set arbitrarily.

  FIG. 17 is a processing flow performed in the navigation system according to one embodiment of the present invention. At 1701, as shown in FIG. 15, the design layer, cell information, etc. are designated, and the designated data from the stored design data is displayed on the screen. Reference numeral 1702 denotes an imaging area. In 1703, pattern data and position information in the field of view (an area specified as the imaging area) are captured and converted into bitmap data 1603. This portion is the same as that shown in FIG. Next, a location to be inspected / measured is designated at 1704, the coordinate data is read, and a template for positioning is designated at 1705. The template is usually specified by the operator by selecting the part with the most specific feature, and such a part is automatically specified using image processing technology that evaluates the high-frequency component and specificity of the image. It is also possible to specify. Finally, in 1706, all information necessary for photographing / inspecting with the scanning electron microscope system is edited based on the information from 1701 to 1705, and transmitted to the operation type electron microscope system.

  FIG. 18 is an example of specifying the measurement points and templates in the navigation system according to one embodiment of the present invention. Reference numeral 1801 designates a measuring point, and 1802 designates a template. Here, the image to be specified is bitmap data, but it may of course be specified on design data before bitmap data conversion.

  FIG. 19 is a processing flow performed in the scanning electron microscope system according to one embodiment of the present invention. In steps 1901 to 1904, wafer alignment information, positioning template information, length measurement point information, and imaging conditions and length measurement methods are registered from the information sent from 1706 in FIG. Actual shooting is performed at 1905, and search processing (position detection) using the template registered at 1902 is executed at 1906. In 1907, a length measurement point is calculated from the positioning coordinates detected in 1906 and the length measurement is executed. Reference numeral 1908 denotes a determination as to whether or not the length measurement has been completed for all the length measurement points, in order to perform the length measurement for all the length measurement points.

  FIG. 20 is an automatic condition file in which imaging / inspection conditions are registered according to an embodiment of the present invention. The automatic condition file may be either a navigation system or a scanning electron microscope system. Actual photographing / inspection is performed by the scanning electron microscope system according to the conditions registered in the automatic condition file 2001. When the photographing / inspection conditions are determined from the information obtained by the navigation system, the condition creation process can be simplified by selecting the most suitable conditions from pre-registered recipes as in the present invention. It is also convenient for management and maintenance. Each recipe registered in the automatic condition file can be partially changed and deleted as in 2002, or can be registered with another name. Furthermore, it is possible to take statistics on how much each recipe is used, and to automatically delete recipes that are less frequently used.

  FIG. 21 is a processing flow when an automatic condition file according to an embodiment of the present invention is used. In 2101, it is determined whether or not a new recipe should be created. If there is no recipe that can be already corrected or partially corrected in the automatic condition file 2001, a new one is created in 2102. Once created, a new recipe is registered in the automatic condition file 2001 in 2106. After registration, it can be executed in 2108 with reference to the recipe. Sometimes only execution is performed without registration. If the same or partially amendable recipe already exists in the automatic condition file 2001, the existing recipe in the automatic condition file is referred to in 2103, and it is determined whether or not a partial change should be made in 2104 To do. If it is the same, there is no need to change a part, and the existing recipe is executed in 2108 as it is. Moreover, even if it is not the same, it is the same if the existing recipe can be substituted. In the case of partial modification, after partial modification in 2105, it is determined whether or not the modified recipe should be registered in the automatic condition file 2001 in 2106, and registered and executed in 2107 and 2008, or not registered. 2108. By registering a recipe that has been used once in the automatic condition file according to this flow, it becomes possible to refer to the condition next time. If a part of the imaging / inspection condition is changed, it is possible to register the condition as another condition in 2106. In this case, both the file before and after the change can be referred to.

  Since the present embodiment is configured as described above, the following effects can be obtained.

  In the conventional semiconductor inspection system, registration of the point for image recognition, the measurement position, and the measurement algorithm is performed by photographing an actual wafer once and using it. For this reason, there is a problem that throughput is not improved because it takes time for registration and occupies the time device. In addition, there is a problem in that an operator is always required for a person to judge and register by looking at an actual SEM image, and an operator-free fully automated semiconductor inspection system cannot be constructed.

  In response to these problems, the present invention creates all the conditions necessary for inspection from design information such as CAD data to shooting conditions, length measurement points, and length measurement algorithms, and performs actual inspection under those conditions. Therefore, an operator-free fully automated high-throughput semiconductor inspection system can be realized.

  Conventionally, when performing matching processing between design data and SEM images, the shape coefficient of the design data and SEM images cannot be dealt with, and the correlation coefficient becomes very small, making it impossible to perform stable matching processing. . In order to solve this problem, the present invention performs matching processing that compensates for shape change by edge information for each direction and its smoothing when matching processing between design data and SEM images. Further, the present invention performs a matching process between the edge image and the template of the design data, re-registers the part of the SEM image corresponding to the detected position as the template, and performs the matching process. Matching processing with a stable detection rate can be realized.

  The pattern matching method of the present embodiment relates to a pattern matching method for performing pattern matching between design data and an image acquired by a scanning electron microscope. In this pattern matching method, design data is converted into a bitmap, and the bitmapped design data is matched with an image acquired by a scanning electron microscope. As a result, it is possible to automate matching processing with high accuracy and high efficiency using design information.

  In addition, since the matching process is performed after edge enhancement is performed on the bitmap, the matching process is facilitated. Furthermore, the detection rate can be further increased by performing the matching process after the smoothing process is performed on the bitmap subjected to the edge enhancement.

Conventional design data and SEM image matching process flow Matching process flow using design data and SEM image according to one embodiment of the present invention Outline of Configuration of Semiconductor Inspection System as an Example of Image Processing Apparatus of the Present Invention Processing flow for re-registering an SEM image as a template over time according to an embodiment of the present invention Processing flow for re-registering an SEM image as a template when the correlation value is higher than the previous correlation value according to an embodiment of the present invention Processing flow for re-registering an SEM image at a position having the highest correlation value as a template in an arbitrary number of times of design data and SEM image matching processing according to an embodiment of the present invention Examples of images used for conventional processing Processing flow of matching processing between bitmap data template and SEM image according to one embodiment of the present invention Processing flow of matching processing between a bitmap data template and an SEM image according to another embodiment of the present invention Method of combining and integrating images divided in each direction The block diagram of the semiconductor inspection system which is one Example of this invention The block diagram of the navigation system which is one Example of this invention The block diagram of the semiconductor inspection system which is one Example of this invention 1 is a network configuration diagram of a semiconductor inspection system according to an embodiment of the present invention. Example of display in navigation system according to one embodiment of the present invention Example of bitmap data created by a navigation system according to an embodiment of the present invention Processing flow performed in navigation system according to one embodiment of the present invention Specification example of measuring point and template in navigation system according to one embodiment of the present invention Processing flow performed in a scanning electron microscope system according to an embodiment of the present invention Automatic condition file in which photographing / inspection conditions are registered according to an embodiment of the present invention Processing flow when an automatic condition file according to an embodiment of the present invention is used

Explanation of symbols

1101 ... Navigation system, 1102 ... Scanning electron microscope system

Claims (8)

In a pattern matching method for performing pattern matching between design data and an image acquired by a scanning electron microscope,
Extracting the edge for each direction of the design data and the edge for each direction of the image acquired by the scanning electron microscope , performing a smoothing process on these edges, and the design data subjected to the smoothing process A pattern matching method comprising integrating edges according to directions, integrating edges according to directions of images acquired by the scanning electron microscope, and performing matching processing between the integrated images . In a pattern matching method for performing pattern matching between design data and an image acquired by a scanning electron microscope,
Extracting the edge for each direction of the design data and the edge for each direction of the image acquired by the scanning electron microscope, performing a smoothing process on these edges, and the design data subjected to the smoothing process A pattern matching method, wherein matching processing is performed separately for each direction in each direction with respect to an edge and an edge of an edge of an image acquired by the scanning electron microscope. In claim 1 or 2,
A pattern matching method, wherein a Sobel filter is used for extracting the edge for each direction of the image acquired by the scanning electron microscope. In claim 1 or 2,
2. The pattern matching method according to claim 1, wherein the direction of edge extraction is two directions or four directions. In a pattern matching device that performs pattern matching between design data and an image acquired by a scanning electron microscope,
An edge for each direction of the design data, and an edge extraction means for extracting an edge for each direction of an image acquired by the scanning electron microscope;
Smoothing the edges of the design data and the edges of the image, integrating the edges for each direction of the design data subjected to the smoothing process, and for each direction of the image acquired by the scanning electron microscope Pattern matching means for integrating edges and performing matching processing between the integrated images ;
A pattern matching apparatus comprising: In a pattern matching device that performs pattern matching between design data and an image acquired by a scanning electron microscope,
An edge for each direction of the design data, and an edge extraction means for extracting an edge for each direction of an image acquired by the scanning electron microscope;
The edge of the design data and the edge of the image are subjected to a smoothing process, and the edge of the design data subjected to the smoothing process and the edge of the edge of the image acquired by the scanning electron microscope are classified by direction. Pattern matching means for performing matching processing separately from each other;
A pattern matching apparatus comprising: In claim 5 or 6,
A pattern matching apparatus, wherein a Sobel filter is used to extract the edge for each direction of an image acquired by the scanning electron microscope. In claim 5 or 6,
The pattern matching apparatus according to claim 1, wherein the edge extraction direction is two directions or four directions.

Images