JP4537144B2 - Mask defect classification method and classification apparatus - Google Patents

Description

  The present invention relates to a defect classification method and classification apparatus for an EB (Electron Beam) mask and a photomask used for manufacturing a semiconductor device, for example.

  In the manufacturing process of an EB mask used for electron beam exposure and a photomask used for light exposure, an appearance inspection is performed to inspect whether or not there is a defect in each created mask, and there is no defect. While confirming that it is a mask, the defective part is corrected if necessary.

  Currently, with regard to EB mask appearance inspection devices such as LEEPL (Low Energy Electron-Beam Proximity Projection Lithography) related to 1X exposure mask and EPL (Electron Projection Lithography) related to reduced exposure mask, a dedicated transmissive EB inspection device is used. It is done using.

Usually, in such a transmission type EB inspection apparatus, image data of a pattern to be inspected is obtained and created for an actually created mask pattern, and this image data is compared with design data, so-called Die To DB (Data Base ) Method is adopted. In this method, the presence or absence of a defect is detected from the difference between the image data of all the patterns of the created mask and the corresponding design data. Then, the coordinate information, design data, and image data of the mask pattern can be stored and output for the pattern regarded as a defect. The inspection apparatus generally stores each defective mask pattern in a state where there is a pattern determined to be defective at the center position of the screen to be stored (see FIG. 4).
JP 2000-88563 A

  However, since the inspection apparatus only determines whether or not the pattern is a defect, information on what type of defect is found cannot be obtained. It is also impossible to quantify the degree of defect such as how much it differs from the design data.

  In an actual EB mask manufacturing process, a defect found by inspection is corrected by a correction machine, but it is necessary to determine whether or not the defect can be corrected depending on the type and degree of the defect. At present, this determination is made qualitatively by displaying both the design data output from the EB inspection apparatus and the image data of the mask pattern on the display of the terminal device and comparing them visually.

  However, since some of the masks created include pseudo defects that do not actually need to be corrected, the number of defects for which correction is determined may be several thousand per mask. The amount of work may be enormous. Further, there are many cases where it cannot be determined whether or not the image is defective by image comparison using human eyes.

  On the other hand, since the square hole pattern becomes a basic shape (unit shape) of various mask patterns formed in the EB mask, defect inspection and defect classification of the square hole pattern are the basis for maintaining the quality of the mask pattern.

  In addition, with the high integration of semiconductor devices such as VLSI, internal wiring becomes extremely complex, and inevitably multilayering of wiring is being promoted. In order to interconnect the multilayered internal wiring, it is necessary to form a large number of normally square through holes in each of the plurality of interlayer insulating films. High quality is also required for the hole pattern of the EB mask used to form these through-holes, and rapid inspection is required to maintain high quality.

  For this reason, it is necessary to establish a method for automatically quantifying and classifying defects and to provide a classification device for a hole pattern which is mainly used in an EB mask, usually designed as a square shape, for example.

  Therefore, in view of the above problems, the present invention performs image processing on the defect image data found by the EB inspection apparatus for the hole pattern, and extracts pattern feature amounts such as area comparison, circularity, and center of gravity. It is an object of the present invention to provide a defect classification method and a defect classification apparatus for classifying and quantifying pattern defects with high speed and high accuracy. By classifying and quantifying the pattern defect, it becomes easy to correct the defect in the subsequent process, and the mask manufacturing process can be easily managed, so that high quality can be maintained.

  The mask defect classification method according to the present invention described in the following embodiments acquires an image of design data and an image of an inspection target mask used for manufacturing a mask, and determines the pattern position and inspection range of the inspection target mask. Set, calculate the area ratio of the mask image to the design data image within the inspection range, compare the area ratio with the predetermined upper limit value and lower limit value, calculate the pattern circularity of the mask image to be inspected, and By comparing the degree with the predetermined first and second standard values, calculating the centroid of each pattern of the design data image and the inspection target mask image, and comparing the deviation of each centroid with a predetermined allowable value This is a method for classifying mask defects.

  In addition, the design data image used for manufacturing the mask and the image of the inspection target mask are acquired, the pattern position of the inspection target mask and the inspection range are set, and the area of the mask image relative to the design data image within the inspection range When the area ratio is 0, it is determined as an unopened defect. When the area ratio is not 0, the area ratio is compared with a predetermined lower limit value, and the area ratio is smaller than the predetermined lower limit value. The pattern circularity of the image of the mask to be inspected is calculated, and when the circularity is equal to or higher than a predetermined first standard value, it is determined as an undersize defect, and when it is smaller than the first standard value, a chip defect is detected. On the other hand, if the area ratio is greater than or equal to a predetermined lower limit value, it is compared with a predetermined upper limit value. If the area ratio is greater than the predetermined upper limit value, the pattern circularity of the image of the inspection object mask is determined. Calculated and circularity is a predetermined second standard value In the above case, it is determined as an oversize defect. When the circularity is smaller than the second standard value, it is determined as a protrusion defect. On the other hand, when the area ratio is equal to or lower than a predetermined upper limit value, an image of design data is determined. The center of gravity of each pattern of the mask image to be inspected is calculated, and if the deviation of each center of gravity is larger than a predetermined allowable value, it is determined as an IP error, and if it is less than the predetermined allowable value, it is determined as a pseudo defect. This is a method of classifying mask defects.

  Further, the mask defect classification apparatus according to the present invention described in the following embodiments includes means for acquiring an image of design data and an image of an inspection target mask used for manufacturing the mask, and a position of the pattern of the inspection target mask. Means for setting the inspection range, means for calculating the area ratio of the mask image to the design data image within the inspection range, means for calculating the pattern circularity of the image of the inspection target mask, the image of the design data, and the inspection target Means for calculating the center of gravity of each pattern of the mask image, based on data calculated by means for calculating the area ratio, means for calculating the pattern circularity, and means for calculating the center of gravity of each pattern. A device having means for determining the type of defect.

  Also, a means for obtaining an image of the design data used for manufacturing the mask and an image of the inspection target mask, a means for setting the position of the inspection target mask pattern and the inspection range, and a design data image within the inspection range. The means for calculating the area ratio of the mask image, the means for determining the type of defect, and the means for determining determine that the area ratio is 0 and determine that the defect is an unopened defect, and set the area ratio to a predetermined upper limit value or The means for comparing with the lower limit value, the means for calculating the pattern circularity of the image of the mask to be inspected, and the means for determining are determined as undersize defects when the circularity is equal to or greater than a predetermined first standard value. If it is smaller than the first standard value, it is determined that the defect is a chip defect, and the means for calculating the pattern circularity of the image of the inspection target mask, and the means for determining the second standard value with a predetermined circularity If this is the case, Means for determining a bar-size defect, determining that the defect is a protrusion defect when the circularity is smaller than the second standard value, and calculating the center of gravity of each pattern of the design data image and the inspection target mask image The judging means is an apparatus for classifying a defect that judges a P error when the deviation of each center of gravity is larger than a predetermined allowable value, and judges a false defect when it is less than the predetermined allowable value.

  For hole patterns, image data with defects is processed, pattern area comparison and pattern feature quantities such as circularity and center of gravity of image patterns are extracted, and pattern defects can be classified at high speed and with high accuracy. It has become possible to provide a pattern defect classification method and a pattern defect classification apparatus that perform quantification.

  Further, by classifying and quantifying the pattern defects, it becomes easy to correct a mask pattern defect in a later process, and the mask manufacturing process can be easily managed, thereby enabling high-quality mask manufacturing.

  Hereinafter, the contents of the present invention will be described in detail with reference to the accompanying drawings. FIG. 1 shows a cross section of a stencil mask 1 used as an EB mask. The stencil mask 1 is usually formed using an SOI (Silicon On Insulator) substrate. The beam 3 formed at a substantially constant interval from the support frame 2 that supports the membrane 5 in the peripheral portion of the mask is made of, for example, silicon that is a support substrate having a thickness of 725 μm. A single crystal having a thickness of, for example, about several hundreds of nm (in the case of a LEEPL mask) or about 2 μm (in the case of an EPL mask) is formed on the bottom thereof via a silicon oxide film 4 having a thickness of, for example, 1 μm or less. A membrane 5 made of a silicon film is disposed. As a material for the membrane 5, for example, a diamond thin film can be used in addition to silicon.

  The mask pattern is determined by the shape of the hole 6 formed in the membrane 5. Note that exposure of a semiconductor wafer (not shown) forming an LSI or the like is performed by an electron beam passing through the hole 6. The shape of the hole 6 determines, for example, the shape of a hole pattern in the EB mask. The present invention relates to a method and apparatus for classifying the shape defects of the holes 6.

  FIG. 2 is a flowchart showing an embodiment of the defect classification method according to the present invention, and FIG. 3 shows a mask defect according to the present invention formed using, for example, a CPU (Central Processing Unit) capable of image data processing. It is a figure which shows embodiment of the classification device. The defect classification method and defect classification apparatus for the EB mask will be described with reference to FIGS.

  First, a measured mask image and a design data image are acquired for a pattern determined as a defect by the mask appearance inspection apparatus 20 that compares the measurement data and the design data, and recorded in the image data recording unit 22.

  Various types of mask appearance inspection apparatuses 20 are used. As an example, the Die to CAD-Data method (not shown) will be briefly described. In this method, an EB mask to be inspected is irradiated with an electron beam while being sequentially scanned, and an image based on the electron beam sequentially passing through holes 6 for forming a mask pattern is received by a CCD camera. Acquired as image data (Die-Data) corresponding to the hole. The acquired image data and the design data (CAD-Data) used for manufacturing the EB mask are compared, and if there is a difference between the two, it is determined that the manufactured EB mask has a defect, Output defect data. As the defect data, for example, both the position of the pattern in which the defect has occurred and its image data (Die-Data) and design data (CAD-Data) are output. These data are transmitted directly to the mask defect device 21 according to the present invention through an appropriate transmission line (not shown) such as a LAN, or via a predetermined recording medium such as a CDROM (not shown). Sent.

  The data output from the mask appearance inspection apparatus 20 is, for all patterns determined to be defective, for example, one file in a TIFF (Tagged-Image Files Format) format in which the data relating to the mask image and the design data image is not compressed. Are output together. In this case, image data of a pattern for which a defect is to be classified is sequentially extracted from the TIFF file and recorded in a predetermined recording unit (not shown), and the following classification process is performed.

  An example of the acquired mask image and design data image is shown in FIG. The designed hole pattern 41 is usually a square of about 70 to 200 nm □ in LEEPL but usually about 280 to 800 nm □ in EPL, but the created hole pattern 42 is shown in FIG. As shown in FIGS. 4 and 5 (b), it is generally nearly circular.

  Next, the inspection range setting means 23 sets the position of the pattern to be classified and the measurement range 43. The images (FIGS. 4A and 4B) output from the inspection apparatus 20 are output so that, for example, as shown in FIG. In this case, first, in the design data image, a pattern in the vicinity of the center is surrounded by a rectangle, for example, as a measurement range. Also for the mask image, a rectangle is similarly generated at the same position and the same size, and is set as the measurement range. An example in which the measurement range 43 is set in the design data image and the mask image is shown in FIG.

  Next, the area ratio of the design data image and the mask image with respect to the hole pattern 45 in the measurement range 43 is calculated by the area ratio calculating means 24, and the subsequent processing is branched depending on the value of the area ratio. The area ratio is obtained, for example, by using the image binarizing means 25, and obtaining the number of pixels having a density value equal to or higher than a predetermined threshold for the mask design data image and the mask image in the measurement range 43, These ratios (the pattern area of the mask image / the pattern area of the design data image) can be calculated. The threshold value applied to binarize the grayscale image as shown in the mask image of FIG. 4 or 5 can be calculated by, for example, a discriminant analysis method (so-called Otsu's method).

  When this area ratio is 0, it means that there is no hole pattern in the mask image, so the defect determination means 35 can determine the defect as an unopened defect. An example of an unopened defect is shown in FIG.

  When the area ratio is not 0, the area ratio comparison means 26 compares the area ratio with the upper limit value and lower limit value determined in advance. For example, if the allowable range is approximately ± 10%, the upper limit value may be set to a specific value in the range of 1.0 to 1.1, and the lower limit value may be set to a specific value in the range of 0.8 to 0.9. it can. Here, as shown in FIGS. 4 and 5, for example, the actual mask pattern is rounded with rounded corners as a whole with respect to the square design data, so the area ratio is 1 even in the case of a normal pattern. Less than.

  Since the upper limit value and the lower limit value of the area ratio for determining whether or not there is a defect in the pattern vary depending on the hole pattern formation conditions, it is necessary to investigate and set appropriate values for each mask in advance. If the area ratio is smaller than the lower limit value, it means that the hole pattern of the mask image is smaller than the design value for some reason. Conversely, if the area ratio is larger than the upper limit value, the hole pattern becomes larger than the design value. It means that there is.

  When the area ratio is smaller than the lower limit value or larger than the upper limit value, the circularity calculation means 27 calculates the circularity of the pattern of the mask image. In order to calculate the circularity, the contour extraction means 28 first extracts the contour of the pattern that is the target of the mask image. The contour line is binarized by the discriminant analysis method, then edge extraction processing is performed by the edge extraction processing 29, and the thinning processing means 30 performs the thinning processing. An example in which the outline of the pattern is extracted is shown in FIG.

  The circularity is a ratio of the difference area between this pattern and the approximate circle and the area of the approximate circle (difference area / area of the approximate circle). The circularity is a real number of 1 or less, and approaches 1 as the pattern is closer to a perfect circle. The area of the approximate circle is calculated from the approximate radius obtained by averaging the distance between the center of the pattern and each point of the contour line by the approximate circle calculating means 31.

  Next, the defect determination means 35 determines and classifies the contents of the defect. Defects to be judged include, for example, an undersize defect having a normal shape and an area smaller than the design value, a chip defect having a pattern area smaller than the design value and a distorted shape, a normal shape and an area smaller than the design value Large oversized defects, pattern defects whose shape is distorted larger than the design value, misaligned defects in which the pattern is misaligned, or falsely judged as a defective pattern even though there is no actual defect There are defects.

  The circularity is compared with a predetermined standard value by the circularity comparison means 32. As this standard value, for example, 0.8 can be set. If the area ratio is smaller than the lower limit value and the circularity is equal to or greater than a certain standard value, it means that only the area is small while the pattern is close to a perfect circle, so the defect determination means 35 determines that it is an undersize defect. An example of an undersize defect is shown in FIG. In the example shown in FIG. 8C, the circularity = 0.91.

  On the other hand, if the area ratio is smaller than the lower limit value and the circularity is smaller than the standard value, the pattern has a small area and a distorted shape, so the defect determination means 35 determines that the defect is a chip defect. . An example of a chip defect is shown in FIG. In the example shown in FIG. 9C, the circularity = 0.71.

  If the area ratio is larger than the upper limit value and the circularity is equal to or greater than a certain standard value, it means that only the area is large in a state where the pattern is close to a perfect circle, so the defect determining means 35 determines that it is an oversized defect. To do. An example of an oversize defect is shown in FIG. In the example shown in FIG. 10C, the circularity = 0.97, which is almost a perfect circle.

  On the other hand, when the area ratio is larger than the upper limit value and the circularity is smaller than the standard value, it means that the pattern has a large area and a distorted shape, so the defect determining means 35 determines that it is a protrusion defect. . An example of the protrusion defect is shown in FIG. In the example shown in FIG. 11C, the circularity = 0.73.

Then if the area ratio is in the range between the lower limit and upper limit, it means that the size of the hole pattern is suitable, IP are misaligned pattern (Image Placement; image placement) Error Or a false defect erroneously detected by the mask appearance inspection apparatus 20.

  In order to determine this, centroid coordinates (x, y) are calculated for each pattern of the design data image and the mask image by the centroid coordinate detection means 33 using a normal centroid coordinate detection method, and the centroid coordinate comparison means 34 x , Y is obtained. The deviation of the barycentric coordinates is calculated, and when this deviation is larger than a predetermined allowable value, the defect judging means 35 judges that the pattern position of the mask image is displaced and makes an IP error. An example of an IP error is shown in FIG. The barycentric coordinates of the design data image are (7524 nm, 7515 nm), whereas the barycentric coordinates of the mask image are (7535 nm, 7490 nm), and the barycentric coordinates are shifted by (11 nm, −25 nm).

  On the other hand, if the deviation of the center of gravity is less than or equal to the allowable value, it does not belong to any defect, so the defect determining means 35 determines that it is a false defect erroneously detected by the mask appearance inspection apparatus 20.

  Such a defect classification method is not limited to the EB mask but can also be used for defect classification of a photomask. Further, although the case of a square pattern has been described in the above embodiment, the present invention can be applied to the case of a mask having another shape such as a circle or a polygon.

  The embodiment of the present invention described here is merely an example, and the defect classification apparatus can be variously modified without changing the gist of the present invention.

It is a figure which shows the cross section of the electron beam exposure mask of mask defect classification | category object by this invention. It is a flowchart figure of the defect classification method which concerns on this invention. It is a figure which shows the mask defect classification | category apparatus 21 based on this invention formed using CPU etc. which can process image data. It is a figure which shows the mask image (b) and design data image (a) by which the mask was acquired by the defect inspection apparatus. It is a figure which shows the example which set the measurement range to the design data image (a) and the mask image (b). It is a figure which shows the example of a non-opening defect. It is a figure which shows the example which extracted the outline of the pattern. It is a figure which shows the example of an undersize defect. It is a figure which shows the example of a chip defect. It is a figure which shows the example of an oversize defect. It is a figure which shows the example of a protrusion defect. It is a figure which shows the example of IP error.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Stencil mask, 2 ... Strut, 3 ... Beam, 4 ... Silicon oxide film, 5 ... Membrane, 6 ... Hole, 20 ... Mask external appearance inspection apparatus, 21 ... Mask defect classification apparatus, 22 ... Image data acquisition means, 23 ... Inspection range setting means, 24 ... Area ratio calculation means, 25 ... Image binarization means, 26 ... Area ratio comparison means, 27 ... Circularity calculation means, 28 ... Outline extraction means, 29 ... Edge extraction means, 30 ... Thin lines Processing means, 31 ... approximate circle calculation means, 32 ... circularity comparison means, 33 ... centroid coordinate detection means, 34 ... centroid coordinate comparison means, 35 ... defect determination means, 41 ... hole pattern design data, 42 ... hole pattern image Data, 43 ... Measurement range, 45 ... Hole pattern

Claims (4)

A method of classifying mask defects,
Obtaining an image of design data and an image of an inspection target mask used for manufacturing the mask,
Set the pattern position and inspection range of the mask to be inspected,
Calculate the area ratio of the mask image to the design data image within the inspection range,
When the area ratio is 0, it is determined as an unopened defect,
If the area ratio is not 0, the area ratio is compared with a predetermined lower limit value,
When the area ratio is smaller than the predetermined lower limit value, calculate the pattern circularity of the image of the inspection target mask,
When the circularity is equal to or greater than a predetermined first standard value, it is determined as an undersize defect, and when it is smaller than the first standard value, it is determined as a chip defect.
On the other hand, if the area ratio is equal to or greater than the predetermined lower limit value, it is compared with a predetermined upper limit value,
If the area ratio is greater than the predetermined upper limit value, calculate the pattern circularity of the image of the inspection target mask,
When the circularity is equal to or greater than a predetermined second standard value, it is determined as an oversize defect, and when the circularity is smaller than the second standard value, it is determined as a protrusion defect,
On the other hand, when the area ratio is less than or equal to the predetermined upper limit value, the center of gravity of each pattern of the design data image and the inspection target mask image is calculated,
A method of classifying a defect of a mask when it is determined that an image placement error occurs when the deviation of each center of gravity is larger than a predetermined allowable value, and when it is equal to or less than the predetermined allowable value. An apparatus for classifying mask pattern defects,
Means for obtaining an image of design data and an image of an inspection target mask used for manufacturing the mask;
Means for setting the pattern position and inspection range of the mask to be inspected;
Means for calculating the area ratio of the mask image to the design data image within the inspection range;
A means of determining the type of defect;
Means for comparing the area ratio with a predetermined upper limit or lower limit;
Means for calculating a pattern circularity of an image of the inspection object mask;
Means for calculating the center of gravity of each pattern of the design data image and the inspection target mask image,
It means for pre-SL judgment,
When the area ratio is 0, it is determined as an unopened defect ,
The area ratio is smaller than the lower limit, and the circularity is determined to undersize defects in the case of more than a predetermined first standard value, it is determined that the chipping defect if the less than the first standard value ,
The area ratio is greater than the upper limit value, and before Symbol determines that oversize defect when circularity is equal to or larger than the predetermined second standard value, if the circularity is smaller than the second standard value Judge as a protrusion defect,
When the area ratio is between the upper limit value and the lower limit value, and the deviation of each center of gravity is larger than a predetermined allowable value, it is determined as an image placement error, and when it is equal to or lower than the predetermined allowable value, it is determined as a pseudo defect. To
A device that classifies defects. 3. The apparatus for classifying defects according to claim 2, wherein the means for calculating the area ratio includes means for binarizing image data. The apparatus for classifying defects according to claim 2, wherein the means for calculating the circularity includes contour extraction means for extracting a contour line from a mask image.

Images