JP4174486B2 - Image generation apparatus, image generation method, and sample inspection apparatus - Google Patents

Description

  The present invention relates to an image generation apparatus, an image generation method, or a sample inspection apparatus. For example, the present invention relates to a sample inspection apparatus used for semiconductor manufacturing and a reference image generation method used for the sample inspection apparatus.

  In recent years, the circuit line width required for a semiconductor element has been increasingly narrowed as LSI is highly integrated and has a large capacity. These semiconductor elements use an original pattern pattern (also referred to as a mask or a reticle, hereinafter referred to as a mask) on which a circuit pattern is formed, and the pattern is exposed and transferred onto a wafer by a reduction projection exposure apparatus called a stepper. It is manufactured by forming a circuit. Therefore, a pattern drawing apparatus capable of drawing a fine circuit pattern is used for manufacturing a mask for transferring such a fine circuit pattern onto a wafer. A pattern circuit may be directly drawn on a wafer using such a pattern drawing apparatus. The electron beam drawing apparatus is also described in the literature (see, for example, Patent Document 1). Alternatively, development of a laser beam drawing apparatus for drawing using a laser beam in addition to an electron beam has been attempted and disclosed in the literature (for example, see Patent Document 2).

  However, as represented by a 1 gigabit class DRAM (Random Access Memory), a pattern constituting a large scale integrated circuit (LSI) is going to be on the order of submicron to nanometer. One of the major causes of a decrease in yield in the manufacture of LSI is a defect of a mask used when an ultrafine pattern is exposed and transferred onto a semiconductor wafer by a photolithography technique. In particular, with the miniaturization of the pattern dimensions of LSIs formed on semiconductor wafers, the dimensions that must be detected as pattern defects have become extremely small. Therefore, an apparatus for inspecting such a defect has been developed.

  On the other hand, with the development of multimedia, LCDs (Liquid Crystal Display) are increasing in size of the liquid crystal substrate to 500 mm × 600 mm or more, and TFTs (Thin Film Transistors) formed on the liquid crystal substrate. : Thin film transistors) and the like are being miniaturized. Therefore, it is required to inspect a very small pattern defect over a wide range. For this reason, there is an urgent need to develop a sample inspection apparatus for efficiently inspecting defects of a photomask used in manufacturing such a large area LCD pattern and a large area LCD in a short time.

  Here, as a mask defect inspection method, “die to die (die-to-die) inspection” that compares the same pattern at different locations on the same mask, or CAD data (design data) used when drawing the mask pattern is compared. There is “die to database inspection”. In the inspection method in such an inspection apparatus, for example, the inspection area of the mask is divided into a plurality of stripe-shaped inspection areas (inspection stripes) having overlapping portions in the Y direction, and the inspection is performed sequentially for each inspection area, and finally all The defects in the entire inspection area are integrated to detect defects in the entire mask.

  For example, first, the XY stage on which the mask is mounted is moved to the inspection start position of the first inspection stripe. Then, each time a movement of a constant pitch is detected by the laser interferometer while the XY stage is fed at a constant speed in the X direction, the laser beam is scanned by the laser scanning optical device in the Y direction, and the transmitted light is detected to a predetermined size. A two-dimensional image is acquired for each area. And the optical image of the area is compared with a reference image, and defect detection is performed (for example, refer patent document 3).

  Here, while the die-to-die comparison may overlook common defects, the die-to-database comparison is a reliable method, but it is necessary to generate reference image data corresponding to an image pattern from design data. For this purpose, for example, a binary or gray level (multilevel) raster image is generated by painting design data expressed in a polygon such as a rectangle or a triangle in units of pixels, and an optical point response function is convoluted ( Convolution) is widely performed. In order to further improve the degree of approximation, it may be possible to adopt a physical model with a large amount of calculation such as estimating a point response function from an image pattern. It is also conceivable to adopt a partially coherent imaging model, which is a general model in lithography simulation. Next, as a technique for increasing the inspection speed, a high-speed image processing algorithm using run-length coding is known. In run-length encoding, an image is encoded with a continuous number of white and black binary images on the same line. As this image processing algorithm, in addition to two-dimensional convolution (for example, see Non-Patent Document 1), morphology erosion, opening, and the like are known (for example, Non-Patent Document 2). reference).

Based on Non-Patent Document 1, a two-dimensional convolution algorithm will be briefly described. Assume that the output y (t) is obtained by convolving f (t) with the input u (t). This can be expressed in frequency space as Y (s) = F (s) · U (s).
Here, Y (s), F (s), and U (s) are Laplace transforms of y (t), f (t), and u (t), respectively. However, since Y (s) = (1 / s) (F (s) .sU (s)) holds, the input U (s) is first differentiated from this equation, and then f (t) and convolution. Thus, it can be seen that the same result can be obtained even if integration is performed at the end. Since the input u (t) given by the run length is the sum of the step functions, if this is differentiated, it becomes a Dirac delta function, and the convolution of the delta function and an arbitrary function g (t) is simply parallel of g (t). Since it is a movement, convolution can be omitted. This can significantly reduce the labor of product-sum operation.

  In addition, as described in Non-Patent Document 2, if a morphology technique is used, resizing and corner rounding processing can be realized. As described in Non-Patent Document 3, run-length encoding itself is one of encodings that are widely used especially for binary images such as faxes and can perform information compression without loss of information. . Furthermore, as described above, the processing algorithm can be made more efficient.

  In addition, techniques for generating reference images are disclosed in the literature (see, for example, Patent Documents 4 and 5).

Here, even if any of the above-described conventional techniques is used, a reference image for a die-to-database comparative inspection of a multilayer mask created using, for example, two-layer data such as a tritone or a Levenson mask is generated. I couldn't. Therefore, conventionally, a pseudo defect has occurred at the boundary between the first layer and the second layer. Conventionally, in order not to cause such a problem, it has been unavoidable to deal with the problem by reducing the comparative sensitivity at the boundary position between the first layer and the second layer.
Japanese Patent Laid-Open No. 2002-237445 US Pat. No. 5,386,221 JP 2000-147749 A JP 2002-107307 A JP 2003-107669 A Patrick Simard, Leon Vonto, Patrick Haffner, and Yann Le Cun, “Boxlets: a fast convolution al ce n s e n s e n s e n s e n s e n s e n s e n e n s e n e n e n e n e n e n e n e n e n e n e n e n e n e n e n e n e n e n e n e n F. Ercal, F.M. Bunyak, F.M. Hao, and L.H. Zheng, “A Fast Modular RLE-based Inspection Scheme for PCBs”, Proc. of SPIE-Architectures, Networks, and Intelligent Systems for Manufacturing Integration, Pittsburg, Oct. 1997, Vol. 3203, pp. 49-59. Tomohiro Oshi, Hideo Kuroda, "Image compression technology understood by illustration", Nihon Jitsugyo Publishing (1999)

  As described above, the conventional defect inspection apparatus can generate a reference image for die-to-database comparison inspection of a multi-layer mask created using, for example, two-layer data such as tritone and Levenson mask. There wasn't. Therefore, conventionally, a pseudo defect has occurred at the boundary between the first layer and the second layer. Conventionally, in order not to cause such a problem, it has been unavoidable to deal with the problem by reducing the comparative sensitivity at the boundary position between the first layer and the second layer.

  The present invention overcomes the above-mentioned problems and generates a reference image that can be inspected without reducing the comparison sensitivity, which is used for die-to-database comparative inspection of a multi-layer mask created using two-layer data. Objective.

An image generation device according to one embodiment of the present invention includes:
When the first and second multi-value images forming the first and second pattern layers are input to the substrate and the input first and second multi-value images are combined, An extraction unit for extracting an area ratio occupied by the base layer, the first pattern layer, and the second pattern layer;
A correlation calculation unit that performs a correlation calculation for each pixel using the area ratio of two pixels that are respectively point-symmetric with each pixel from which the area ratio is extracted;
A point response function is convolved and integrated with the result of the correlation calculation performed on each pixel to calculate the gradation value of each pixel, and an image obtained by combining the first and second multi-valued images is generated. A convolution integrator to
Equipped with a,
The correlation calculation unit is located in a predetermined region centered on each pixel, and all the two pixels that are point-symmetric with each pixel are
The product of the area ratios occupied by the base layer, the total value of the product of the area ratio occupied by the base layer and the area ratio occupied by the first pattern layer, and the first pattern layer A first calculation unit for calculating a product value of the area ratios;
The calculated product value of the area ratio occupied by the base layer and the first weighting factor, and the product value of the area ratio occupied by the base layer and the area ratio occupied by the first pattern layer. A second calculation unit that calculates a product value of the total value of the first and second weighting factors, a product value of the area ratios occupied by the first pattern layer, and a product value of the third weighting factor;
The calculated product value of the area ratio occupied by the base layer and the first weighting factor, and the product value of the area ratio occupied by the base layer and the area ratio occupied by the first pattern layer. A third calculation unit for calculating a sum of a product value of the total value of the first and second weighting factors, a product value of the area ratios occupied by the first pattern layer, and a product value of the third weighting factor When,
It is characterized by having .

And the image generation method of one aspect of the present invention includes:
An input step of inputting the first and second multi-value images for forming the first and second pattern layers on the substrate;
An extraction step of extracting an area ratio occupied by the base layer, the first pattern layer, and the second pattern layer for each pixel when the input first and second multi-value images are combined. When,
A correlation calculation step of performing a correlation calculation for each of the pixels using the area ratio of the two pixels that are in point symmetry with the respective pixels from which the area ratio has been extracted;
A point response function is convolved and integrated with the result of the correlation calculation performed on each pixel to calculate the gradation value of each pixel, and an image obtained by combining the first and second multi-valued images is generated. A convolution integration step,
Equipped with a,
In the correlation calculation step, for all two pixels located in a predetermined area centered on each pixel, each corresponding to a point-symmetrical position with each pixel,
The product of the area ratios occupied by the base layer, the total value of the product of the area ratio occupied by the base layer and the area ratio occupied by the first pattern layer, and the first pattern layer Calculate the product of the area ratios,
The calculated product value of the area ratio occupied by the base layer and the first weighting factor, and the product value of the area ratio occupied by the base layer and the area ratio occupied by the first pattern layer. A product value of the total value of the first and the second weighting factor, a product value of the area ratios occupied by the first pattern layer and a product value of the third weighting factor,
The calculated product value of the area ratio occupied by the base layer and the first weighting factor, and the product value of the area ratio occupied by the base layer and the area ratio occupied by the first pattern layer. The sum of the product value of the total value of the first and the second weighting factor and the product value of the area ratio occupied by the first pattern layer and the product of the third weighting factor is calculated. .

In addition, the sample inspection device of one embodiment of the present invention includes:
When the first and second multi-value images as design data for forming the first and second pattern layers on the substrate are input and the input first and second multi-value images are combined, each pixel And an extraction unit for extracting an area ratio occupied by the base layer, the first pattern layer, and the second pattern layer,
A correlation calculation unit that performs a correlation calculation for each pixel using the area ratio of two pixels that are respectively point-symmetric with each pixel from which the area ratio is extracted;
A reference image obtained by combining the first and second multi-valued images by calculating the gradation value of each pixel by convolving and integrating a point response function with the result of the correlation calculation performed on each pixel. A convolution integrator to generate,
An optical image acquisition unit for acquiring an optical image of a sample to be inspected produced based on the design data;
A comparison unit that compares the optical image of the sample to be inspected with the reference image;
Equipped with a,
The correlation calculation unit is located in a predetermined region centered on each pixel, and all the two pixels that are point-symmetric with each pixel are
The product of the area ratios occupied by the base layer, the total value of the product of the area ratio occupied by the base layer and the area ratio occupied by the first pattern layer, and the first pattern layer A first calculation unit for calculating a product value of the area ratios;
The calculated product value of the area ratio occupied by the base layer and the first weighting factor, and the product value of the area ratio occupied by the base layer and the area ratio occupied by the first pattern layer. A second calculation unit that calculates a product value of the total value of the first and second weighting factors, a product value of the area ratios occupied by the first pattern layer, and a product value of the third weighting factor;
The calculated product value of the area ratio occupied by the base layer and the first weighting factor, and the product value of the area ratio occupied by the base layer and the area ratio occupied by the first pattern layer. A third calculation unit for calculating a sum of a product value of the total value of the first and second weighting factors, a product value of the area ratios occupied by the first pattern layer, and a product value of the third weighting factor When,
It is characterized by having .

  According to the present invention, for each pixel, by extracting the area ratio occupied by the base layer, the first pattern layer, and the second pattern layer, the area ratio after overlapping the two-layer data is obtained. Can be extracted. Then, using the area ratio of the two pixels that are in point symmetry with the respective pixels, the correlation calculation is performed for each of the pixels, and the gradation value of each pixel is calculated by convolving and integrating the point response function. be able to. As a result, it is possible to generate a highly accurate reference image using two-layer data that has not been possible in the past. Therefore, the generation of pseudo defects in the die-to-database comparison inspection can be reduced without reducing the comparison sensitivity at the boundary position between the first layer and the second layer.

Embodiment 1 FIG.
In the first embodiment, as an example of a multilayer mask, a halftone (HT) film as a phase shift layer and a chromium (Cr) film as a light-shielding film layer are formed on a glass substrate surface (Qz: quartz) as a transmission layer. An apparatus for generating a reference image for a die-to-database comparative inspection of a tritone mask and an algorithm of the method will be described.

FIG. 1 is a diagram illustrating a main part of each step of the reference image generation method according to the first embodiment.
In FIG. 1, the algorithm in the reference image generation method according to the first embodiment is Qz / HT multiple, which is a process of developing two-layer drawing data into multi-valued images for forming a three-layer phase shift mask. Value image development step (S102), HT / Cr multivalue image development step (S104), resizing step (S106), resizing step (S108), corner rounding step (S110), corner rounding step (S112), shift A processing step (S114), a Qz / HT / Cr region extraction step (S116) for extracting a multi-level image region of three layers based on a drawing process, and pixels at symmetrical positions around the point of interest in advance Select two pixels that are less than or equal to the specified distance, and add an image intensity determined by any two layers to the two pixels, weighted by the area when the two pixels are included in the layer. A correlation calculation step of (S118), carrying out the convolution step (S120) for convolution integration. Then, the reference image data is generated by this process and compared with the actual image.

FIG. 2 is a conceptual diagram showing the configuration of the sample inspection apparatus according to the first embodiment.
In FIG. 2, a sample inspection apparatus 100 that samples a substrate such as a mask or a wafer and inspects defects of the sample includes an optical image acquisition unit and a control system circuit. The optical image acquisition unit includes an XYθ table 102, a light source 103, a magnifying optical system 104, a photodiode array 105, a sensor circuit 106, a laser length measurement system 122, and an autoloader 130. In the control system circuit, a control computer 110 serving as a computer is connected via a bus 120 to a position circuit 107, a comparison circuit 108, a development circuit 111, a reference circuit 112, an autoloader control circuit 113, a table control circuit 114, a magnetic disk device 109, A magnetic tape device 115, a flexible disk device (FD) 116, a CRT 117, a pattern monitor 118, and a printer 119 are connected. The XYθ table 102 is driven by an X-axis motor, a Y-axis motor, and a θ-axis motor.

  A photomask 101 to be inspected is placed on an XYθ table 102 provided so as to be movable in a horizontal direction and a rotation direction by motors of XYθ axes, and an appropriate light source for a pattern formed on the photomask 101. 103 is irradiated with light. The light that has passed through the photomask 101 forms an optical image on the photodiode array 105 via the magnifying optical system 104 and is incident thereon.

FIG. 3 is a diagram for explaining an optical image acquisition procedure.
As shown in FIG. 3, the inspection area is virtually divided into a plurality of strip-shaped inspection stripes having a scan width W in the Y direction, and each of the divided inspection stripes is continuously scanned. Thus, the operation of the XYθ table 102 is controlled, and an optical image is acquired while moving in the X direction. In the photodiode array 105, images having a scan width W as shown in FIG. 3 are continuously input. Then, after acquiring the image of the first inspection stripe, the image of the scan width W is continuously input in the same manner while moving the image of the second inspection stripe in the opposite direction. When an image in the third inspection stripe is acquired, the image moves while moving in the direction opposite to the direction in which the image in the second inspection stripe is acquired, that is, in the direction in which the image in the first inspection stripe is acquired. To get. In this way, it is possible to shorten a useless processing time by continuously acquiring images.

  The pattern image formed on the photodiode array 105 is photoelectrically converted by the photodiode array 105 and further A / D (analog-digital) converted by the sensor circuit 106. These light source 103, magnifying optical system 104, photodiode array 105, and sensor circuit 106 constitute a high-magnification inspection optical system.

  The XYθ table 102 is driven by the table control circuit 114 under the control of the control computer 110. The movement position of the XYθ table 102 is measured by the laser length measurement system 122 and supplied to the position circuit 107. The photomask 101 on the XYθ table 102 is transported from an autoloader 130 driven by an autoloader control circuit 113.

  The measurement pattern data output from the sensor circuit 106 is sent to the comparison circuit 108 together with data indicating the position of the photomask 101 on the XYθ table 102 output from the position circuit 107.

  On the other hand, the design data used when forming the pattern of the photomask 101 is read out from the magnetic disk 109 which is an example of a storage device to the development circuit 111 through the control computer 110. The development circuit 111 converts (develops) the design image data of the photomask 101 that is the read sample to be inspected into binary or multivalued image data, and sends this image data to the reference circuit 112. The reference circuit 112 performs an appropriate filter process on the sent graphic image data to generate a reference image.

  Then, the comparison circuit 108 cuts out the measurement pattern data to be a real image and the reference image as an image of a predetermined area (area). Then, the measurement pattern data and the reference image are aligned (area alignment), and the area-aligned measurement pattern data and the reference image are compared. For example, if the difference is equal to or greater than a set threshold value, it is determined as a defect. To do.

FIG. 4 is a block diagram showing an internal configuration of the reference circuit.
In FIG. 4, a reference circuit 112 includes a resizing circuit 312, a resizing circuit 314, a corner rounding circuit 322, a corner rounding circuit 324, a shift processing circuit 330, a Qz / HT / Cr region extraction circuit 340, a correlation calculation circuit 350, and a convolution circuit. 360 is provided. The correlation calculation circuit 350 includes an area ratio calculation circuit 352, a weighting calculation circuit 354, and a total sum calculation circuit 356.

FIG. 5 is a process sectional view of a full tritone mask.
A general drawing method of a full tritone mask which is an example of the photomask 101 will be described below. First, as shown in FIG. 5A, the mask blank has a three-layer structure of a chromium film 206 as a chromium layer, a halftone film 204 as a halftone layer, and a glass substrate 202 as a quartz layer from the air side. Suppose you are.

  As shown in FIG. 5B, using the resist pattern formed by drawing using the drawing data A of the first layer as a mask, the chromium layer is first etched and then the halftone layer is etched. .

  Next, as shown in FIG. 5C, only the chromium layer is etched using the resist pattern formed by drawing using the drawing data B of the second layer as a mask. Through the above steps, a full tritone mask composed of three layers of a chromium layer, a halftone layer, and a quartz layer can be manufactured.

A process for comparing and inspecting the tritone mask with the database will be described below.
In S (step) 102, as a Qz / HT multilevel image expansion process, the expansion circuit 111 inputs the first layer of drawing data A from the magnetic disk device 109, and converts the drawing data into multivalued image data for each pixel. Expand (render).

  In S (step) 104, as the HT / Cr multilevel image expansion process, the expansion circuit 111 inputs the drawing data B of the second layer from the magnetic disk device 109, and converts the drawing data into multivalued image data for each pixel. Expand (render).

  In S106, as a resizing process, the resizing circuit 312 inputs the drawing data A of the first layer expanded into multi-value image data, and matches the drawing data A of the first layer with the mask drawing process characteristics. To resize the multi-valued image. By performing the resizing process, it is possible to approximate the actual image drawn by the drawing data A of the first layer.

  In S108, as the resizing process, the resizing circuit 314 inputs the drawing data B of the second layer expanded into the multi-value image data, and matches the drawing data B of the second layer with the mask drawing process characteristics. To resize the multi-valued image. By performing the resizing process, it is possible to approximate the actual image drawn by the drawing data B of the second layer.

  In S110, as the corner rounding step, the corner rounding circuit 322 inputs the drawing data A of the first layer expanded into the multivalued image data, and the mask drawing process characteristics with respect to the drawing data A of the first layer. In accordance with, the corner rounding processing of the multi-valued image is performed. By performing the corner rounding process, it is possible to approximate the actual image drawn by the drawing data A of the first layer.

  In S112, as the corner rounding step, the corner rounding circuit 324 inputs the drawing data B of the second layer expanded into the multi-valued image data, and the mask drawing process characteristics with respect to the drawing data B of the second layer. In accordance with, the corner rounding processing of the multi-valued image is performed. By performing the corner rounding process, it is possible to approximate the actual image drawn by the drawing data B of the second layer.

  In S114, as a shift processing step, the shift processing circuit 330 aligns the second layer with respect to the first layer in consideration of the overlay accuracy of the second layer with respect to the first layer. If the shift processing step is performed before the resizing step, the resized regions may overlap. Therefore, by carrying out the shift processing step after the resizing step, overlapping of the resized portions can be prevented. Furthermore, if the shift processing step is performed before the corner rounding step, the corner rounding amount and the value that are originally required change because the position changes. Therefore, by performing the shift processing step after the corner rounding step, the corner rounding processing can be performed with the originally required corner rounding amount. Further, re-extraction of the Qz / HT / Cr region can be prevented by performing it before the Qz / HT / Cr region extraction step described later. And recalculation of the calculation after the Qz / HT / Cr region extraction step can be prevented.

  In S116, as the Qz / HT / Cr region extraction step, the Qz / HT / Cr region extraction circuit 340, which is an example of an extraction unit, extracts the quartz portion from the two-layer multivalued image that has been subjected to the resizing process and the corner rounding process. A multi-valued image separated into three layers of a halftone part and a chrome part is regenerated. In other words, the Qz / HT / Cr region extraction circuit 340 includes the drawing data A and the first layer of the first layer of the multilevel image forming the pattern layer of the halftone film 204 and the pattern layer of the Cr film 206 on the glass substrate 202. When the drawing data B of the second layer is input and the input drawing data A of the first layer and the drawing data B of the second layer are combined, a quartz part and a halftone part are provided for each pixel. Extract the area ratio occupied by the chrome part.

  In the Qz / HT / Cr region extraction step, it is necessary to predefine what the combination of the two-layer multi-valued image and the actual layer correspond to in consideration of the mask drawing process. Here, as an example, Qz / HT / Cr region extraction in the mask drawing process described in FIG. 5 will be described. The pixel X1, X2, X3, X4 corresponding to the position shown in FIG. 5C will be described as an example.

FIG. 6 is a diagram for explaining Qz / HT / Cr region extraction in the pixel X1.
The left figure of Fig.6 (a) is sectional drawing which cut out the area | region of the pixel X1 shown in FIG.5 (c). And the right figure of Fig.6 (a) is a top view of the area | region of the pixel X1. The region of the pixel X1 is a region in which three layers of a quartz portion, a halftone portion, and a chrome portion are mixed.

  In such a region, the area ratio occupied by the quartz portion can be represented by a region ratio etched by the pattern of the drawing data A of the first layer. In FIG. 6B, this is indicated as A. That is, in the pattern of the drawing data A of the first layer, since both the Cr film 206 and the halftone film 204 are etched, the etched region is always a quartz part.

  The area ratio occupied by the halftone portion is indicated by a difference obtained by subtracting the area ratio etched with the pattern of the drawing data A of the first layer from the area ratio etched with the pattern of the drawing data B of the second layer. be able to. Here, when the difference value becomes negative, the area ratio becomes the ratio value “0”. In FIG. 6B, (B-A, 0) is shown. That is, in the pattern of the drawing data B of the second layer, only the Cr film 206 is etched out of the Cr film 206 and the halftone film 204. Therefore, in the region where the Cr film 206 and the halftone film 204 remain, The etched area always becomes a halftone part. Here, in order to prevent the remaining etching of the Cr film 206, it is desirable that the pattern of the drawing data B of the second layer also includes a region that has already become a quartz part. Therefore, when viewed in pixel units, a difference obtained by subtracting the pattern area ratio of the drawing data A of the first layer from the pattern area ratio of the drawing data B of the second layer is the halftone portion.

  The area ratio occupied by the chromium portion is the remainder of the area ratio occupied by the quartz portion and the area ratio occupied by the halftone portion.

  Here, as shown in FIG. 6C, for example, when the resolution is obtained by dividing one pixel by an area of 0 to 255, the first layer that becomes the Cr layer formation pattern in the region where the pixel X1 is located. Assume that the pattern area ratio of the drawing data A for the eye is 100/255, and the pattern area ratio of the drawing data B for the second layer that is the HT layer formation pattern is 200/255. In this case, as shown in FIG. 6D, the area ratio occupied by the quartz portion, the halftone portion, and the chrome portion in the pixel X1 can be extracted as (100/255, 100/255, 56/255). .

FIG. 7 is a diagram for explaining Qz / HT / Cr region extraction in the pixel X2.
The left figure of Fig.7 (a) is sectional drawing which cut out the area | region of the pixel X2 shown in FIG.5 (c). And the right figure of Fig.7 (a) is a top view of the area | region of the pixel X2. The region of the pixel X2 is a region in which the chrome portion occupies the entire surface among the three layers of the quartz portion, the halftone portion, and the chrome portion.

  In this region, the area ratio occupied by the quartz part is the ratio of the pattern area of the drawing data A of the first layer indicated by “A” in FIG. 7B, and the area ratio occupied by the halftone part is FIG. ), The difference obtained by subtracting the pattern area ratio of the drawing data A of the first layer from the pattern area ratio of the drawing data B of the second layer indicated as “B−A, 0”, and the area ratio occupied by the chromium portion is In FIG. 7B, it can be shown by the remaining area ratio occupied by the quartz portion indicated as “remaining” and the area ratio occupied by the halftone portion.

  Here, as shown in FIG. 7C, for example, when the resolution is obtained by dividing one pixel by an area of 0 to 255, the first layer that becomes the Cr layer formation pattern in the region where the pixel X2 is located. Assume that the pattern area ratio of the drawing data A of the eye is 0/255, and the pattern area ratio of the drawing data B of the second layer that becomes the HT layer formation pattern is 0/255. In this case, as shown in FIG. 7D, the area ratio occupied by the quartz part, the halftone part, and the chrome part in the pixel X2 can be extracted as (0/255, 0/255, 255/255). .

FIG. 8 is a diagram for explaining Qz / HT / Cr region extraction in the pixel X3.
The left figure of Fig.8 (a) is sectional drawing which cut out the area | region of the pixel X3 shown in FIG.5 (c). And the right figure of Fig.8 (a) is a top view of the area | region of the pixel X3. The region of the pixel X3 is a region where the halftone portion occupies the entire surface among the three layers of the quartz portion, the halftone portion, and the chrome portion.

  In such a region, the area ratio occupied by the quartz part is the ratio of the pattern area of the drawing data A of the first layer indicated by “A” in FIG. 8B, and the area ratio occupied by the halftone part is FIG. ), The difference obtained by subtracting the pattern area ratio of the drawing data A of the first layer from the pattern area ratio of the drawing data B of the second layer indicated as “B−A, 0”, and the area ratio occupied by the chromium portion is In FIG. 8B, it can be shown by the remaining area ratio occupied by the quartz portion indicated as “remaining” and the remaining area ratio occupied by the halftone portion.

  Here, as shown in FIG. 8C, for example, when the resolution is obtained by dividing one pixel by an area of 0 to 255, the first layer that becomes the Cr layer formation pattern in the region where the pixel X3 is located. Assume that the pattern area ratio of the drawing data A of the eye is 0/255, and the pattern area ratio of the drawing data B of the second layer that becomes the HT layer formation pattern is 255/255. In this case, as shown in FIG. 8D, the area ratio occupied by the quartz portion, the halftone portion, and the chrome portion in the pixel X3 can be extracted as (0/255, 255/255, 0/255). .

FIG. 9 is a diagram for explaining Qz / HT / Cr region extraction in the pixel X4.
The left figure of Fig.9 (a) is sectional drawing which cut out the area | region of the pixel X4 shown in FIG.5 (c). And the right figure of Fig.9 (a) is a top view of the area | region of the pixel X4. The region of the pixel X4 is a region where the quartz portion occupies the entire surface among the three layers of the quartz portion, the halftone portion, and the chrome portion.

  In such a region, the area ratio occupied by the quartz portion is the ratio of the pattern region of the drawing data A of the first layer indicated by “A” in FIG. 9B, and the area ratio occupied by the halftone portion is FIG. ), The difference obtained by subtracting the pattern area ratio of the drawing data A of the first layer from the pattern area ratio of the drawing data B of the second layer indicated as “B−A, 0”, and the area ratio occupied by the chromium portion is The area ratio occupied by the quartz part indicated as “remaining” in FIG. 9B and the remainder of the area ratio occupied by the halftone part can be shown.

  Here, as shown in FIG. 9C, for example, when the resolution is obtained by dividing one pixel by an area of 0 to 255, the first layer that becomes the Cr layer formation pattern in the region where the pixel X4 is located. It is assumed that the pattern area ratio of the eye drawing data A is 255/255, and the pattern area ratio of the drawing data B of the second layer to be the HT layer formation pattern is 0/255. In this case, as shown in FIG. 9D, the area ratio occupied by the quartz portion, the halftone portion, and the chrome portion in the pixel X4 can be extracted as (255/255, 0/255, 0/255). .

  With the algorithm as described above, the Qz / HT / Cr region in the mask drawing process described with reference to FIG. 5 can be extracted.

  In S118, as a correlation calculation step, the correlation calculation circuit 350, which is an example of a correlation calculation unit, uses the area ratio of two pixels that are point-symmetric to the respective pixels from which the area ratio is extracted. Perform correlation calculation. In other words, based on the partial coherent imaging model, the image intensity corresponding to the transmission image or the reflection image is calculated by performing spatial filtering. Here, as the partially coherent imaging model, the first pixel in the vicinity of the target pixel from the multi-valued image corresponding to the extracted three layers, and the position symmetrical to the first pixel with respect to the target pixel For the second pixel, the step of obtaining a value by the product sum of all combinations of the three layers and the step of adding the product of the correlation coefficient are performed.

FIG. 10 is a diagram showing a point of interest, a region where correlation calculation is performed, and two pixels corresponding to point-symmetrical positions.
In the correlation calculation step, two pixels having a predetermined distance or less are selected from pixels located symmetrically with respect to the point of interest, and the image intensity determined by any two layers is determined by the two pixels. Weighted addition is performed according to the area when pixels are included in the layer. FIG. 10 shows a pixel P and a pixel Q that are in a point-symmetrical position with respect to the pixel of interest X1. And the neighborhood area | region which is an area | region which performs a correlation calculation with respect to the pixel X1 is shown. It is desirable to change the neighborhood area according to the inspection wavelength λ, the numerical aperture NA of the illumination lens, the illumination aperture δ, and the pixel size PS (pixel size). For example, the neighborhood region is a region corresponding to 5 to 10 pixels in diameter with the pixel of interest at the center. By calculating the correlation of the pixel of interest X1 with respect to all the pixels located in the vicinity region, it is possible to improve the accuracy when the gradation value of the pixel of interest X1 is calculated.

FIG. 11 is a diagram illustrating the area ratio between the pixel P and the pixel Q.
In FIG. 11A, the area ratio occupied by the quartz portion, the halftone portion, and the chrome portion in the extracted pixel P is indicated by (a ₁ , b ₁ , c ₁ ). Similarly, in FIG. 11B, the area ratio occupied by the quartz portion, the halftone portion, and the chrome portion in the extracted pixel Q is indicated by (a ₂ , b ₂ , c ₂ ).

  First, as an area ratio calculation step, an area ratio calculation circuit 352, which is an example of a calculation unit, for the pixel P and the pixel Q, the product of the area ratios occupied by the quartz part, the area ratio occupied by the quartz part and the halftone The sum of the product values of the area ratio occupied by the part and the product value of the area ratios occupied by the halftone part are calculated.

FIG. 12 is a diagram illustrating an equation for calculating the area ratio for the pixel P and the pixel Q.
As shown in FIG. 12, the product value SW1 of the area ratios occupied by the quartz part can be expressed as SW1 = a ₁ × a ₂ . The total value SW2 of the product of the area ratio occupied by the quartz portion and the area ratio occupied by the halftone portion can be expressed as SW2 = a ₁ × b ₂ + b ₁ × a ₂ . The product value SW3 of the area ratios occupied by the halftone portion can be represented by SW3 = b ₁ × b ₂ . Here, the product value of the area ratios occupied by the chromium part, the total value of the product values of the area ratio occupied by the quartz part and the area ratio occupied by the chromium part, the area ratio occupied by the halftone part and the chromium part are The total value of the product values with the occupied area ratio is not calculated, but the area occupied by the chrome portion is a light-shielding area, so the gradation value is “0”. Therefore, the relationship with the area ratio occupied by the chromium portion is not required to be calculated.

FIG. 13 is a diagram illustrating a weighting matrix for the pixel P and the pixel Q.
As shown in FIG. 13A, there are nine combinations of regions in the pixel P and the pixel Q. Then, weighting factors W _{11 to} W ₃₃ for each combination are determined. As shown in FIG. 13A, the weight coefficients are W ₁₁ = 1.0, W ₁₂ = W ₂₁ = t · cos φ, W ₂₂ = t ² , W ₁₃ = W ₂₃ = W ₃₁ = W ₃₂ = W ₃₃ = 0. Here, t is the amplitude transmittance, and φ is the phase. Since the area occupied by the chrome part is a light shielding area, all the weighting factors of the area related to the area ratio occupied by the chrome part are set to “0”.

Next, the weighting calculation step, the weighting calculating circuit 354 which is an example of a calculation unit, and the product value of the product value SW1 and the weight coefficient W ₁₁ between the area ratio occupied by the quartz portion calculated area ratio, quartz portion product of the product value of the sum SW2 and the weight coefficient W ₁₂ of product value between the area ratio and the halftone portion occupied area ratio, the product value SW3 and the weight coefficient W ₂₂ between the area ratio of the halftone portion occupied occupied Calculate the value.

FIG. 14 is a diagram illustrating equations for performing weighting calculation and total sum calculation for the pixel P and the pixel Q.
As shown in FIG. 14 (a), product value = SW1 and the product value SW1 and the weight coefficient W ₁₁ between the area ratio occupied by the quartz portion calculated area ratio, occupied area ratio and the halftone area quartz portion occupied product value between the total value SW2 and the weight coefficient _{W 12} and the product value = SW2 · t · cosφ, product value between the area ratio of the halftone portion occupied SW3 and the weight coefficient _{W 22} of product value of the area ratio = SW3 · a t ^2.

Area Next, a total sum calculation step, the total sum calculation circuit 356 which is an example of a calculation unit, and the product value of the product value SW1 and the weight coefficient W ₁₁ between the area ratio of quartz portion occupied, that the quartz portion is occupied and product value of the sum SW2 and the weight coefficient W ₁₂ of product value of the area ratio occupied by the specific halftone portion, and a product value SW3 between area ratio halftone portion occupied and product value of the weight coefficient W ₂₂ Calculate the sum of
As shown in FIG. 14 (b), the total sum = SW1 + SW2 · t · cosφ + SW3 · t 2.

  Then, the area ratio calculation process, the weighting calculation process, and the total sum calculation process are performed on the pixels X1 for all the pixels located in the vicinity region, and finally the total sum calculation circuit 356 calculates the sum. Here, since the calculation is performed once for two pixels located in a point symmetry with respect to the pixels located in the circle serving as the neighboring region, the calculation is completed in half the number of pixels located in the circle serving as the neighboring region. be able to.

  In S120, as a convolution process, the convolution circuit 360 convolves the point response function h (x) with the result of the correlation calculation performed on the pixel X1, thereby obtaining the first layer. A gradation value of the pixel X1 when the two-layer data of the drawing data A and the second-layer drawing data B is synthesized is calculated.

FIG. 15 is a diagram illustrating an expression for performing convolution.
FIG. 15 shows an equation for calculating the combination of two pixels that are point-symmetric with respect to the center pixel by Δ from the entire neighboring region. The variable x of Γ (x) is normalized by λ / (NA · SP) as the inspection wavelength λ, the numerical aperture NA of the illumination lens, and the pixel size SP. For example, the Γ function is represented by a primary Bessel function J1 (x) / x. g1 is a function related to the area ratio occupied by the quartz part, g2 is a function related to the area ratio occupied by the halftone part, E1 is a weighting factor W ₁₁ , E2 is a weighting factor W ₁₂ , and E3 is a weighting factor W ₂₂ ing. Then, the tone value G of the pixel X1 is calculated by convolving the point response function h (x) with the total value calculated for the entire neighboring region. In FIG. 15, the values obtained by performing the area ratio calculation process for all two pixels in the vicinity region are first summed, and then the weighting calculation process is performed. Since the result is the same even if the values subjected to the weighting calculation are summed, it does not matter.

  By performing the calculation using two pixels that are point-symmetric with the target pixel, it is possible to improve the accuracy of the value of the target pixel at the center position. Then, by calculating in the same manner for all the pixels located in the neighboring area that affects the target pixel, it is possible to obtain a highly accurate value in consideration of the influence of all the pixels.

  Then, with respect to all the pixels when the first layer drawing data A and the second layer drawing data B are combined, the steps S116 to S120 described above are performed for each pixel, whereby the first step is performed. A reference image is generated by synthesizing two-layer data of the drawing data A for the layer and the drawing data B for the second layer.

FIG. 16 is another process cross-sectional view of the full tritone mask.
Another drawing method different from FIG. 5 of the full tritone mask which is an example of the photomask 101 will be described below. First, as shown in FIG. 16A, the mask blank has a three-layer structure of a chromium film 206 as a chromium layer, a halftone film 204 as a halftone layer, and a glass substrate 202 as a quartz layer from the air side. Suppose you are.

  As shown in FIG. 16B, the chromium layer is etched using the resist pattern formed by drawing using the drawing data A 'for the first layer as a mask.

  Next, as shown in FIG. 16C, only the halftone layer is etched using the resist pattern formed by drawing using the drawing data B 'of the second layer as a mask. Through the above steps, a full tritone mask composed of three layers of a chromium layer, a halftone layer, and a quartz layer can be manufactured.

FIG. 17 is a diagram for explaining Qz / HT / Cr region extraction in the pixel X1 ′.
The left figure of Fig.17 (a) is sectional drawing which cut out the area | region of pixel X1 'shown in FIG.17 (c). And the right figure of Fig.17 (a) is a top view of the area | region of pixel X1 '. The region of the pixel X1 ′ is a region in which three layers of a quartz portion, a halftone portion, and a chrome portion are mixed.

  In such a region, the area ratio occupied by the quartz portion can be represented by a region ratio etched by the pattern of the drawing data B ′ of the second layer. In FIG. 17B, this is indicated as B '. That is, in the pattern of the drawing data B ′ of the second layer, the Cr film 206 is etched by the pattern of the drawing data A ′ of the first layer, and the exposed halftone film 204 is etched. A region etched with the pattern of data B ′ is always a quartz portion.

  The area ratio occupied by the halftone portion is a difference obtained by subtracting the area ratio etched with the pattern of the drawing data B ′ of the second layer from the area ratio etched with the pattern of the drawing data A ′ of the first layer. Can be shown. In FIG. 17B, “A′-B ′” is shown. That is, the Cr film 206 is etched with the pattern of the drawing data A ′ of the first layer, and the remaining halftone film 204 in the pattern of the drawing data B ′ of the second layer is left unetched. Tone part.

  The area ratio occupied by the chromium portion is the remainder of the area ratio occupied by the quartz portion and the area ratio occupied by the halftone portion.

  Here, as shown in FIG. 17C, for example, when the resolution is obtained by dividing one pixel by an area of 0 to 255, the first Cr layer formation pattern is formed in the region where the pixel X1 ′ is located. Assume that the pattern area ratio of the drawing data A ′ of the layer is 200/255, and the pattern area ratio of the drawing data B ′ of the second layer that becomes the HT layer formation pattern is 100/255. In this case, as shown in FIG. 17D, the area ratio occupied by the quartz portion, the halftone portion, and the chrome portion in the pixel X1 ′ can be extracted as (100/255, 100/255, 56/255). it can.

  As described above, it is desirable to change the method of extracting a multi-valued image area in accordance with the drawing data of the phase shift mask and the mask process even for the same full tritone mask.

  As described above, according to the first embodiment, the gradation value of each pixel can be calculated. As a result, it is possible to generate a highly accurate reference image using two-layer data that has not been possible in the past. Therefore, the generation of pseudo defects in the die-to-database comparison inspection can be reduced without reducing the comparison sensitivity at the boundary position between the first layer and the second layer.

Embodiment 2. FIG.
In the second embodiment, as an example of a multilayer mask, a Levenson mask in which a digging shifter serving as a phase shift layer and a chromium (Cr) film serving as a light shielding film layer are formed on a glass substrate surface (Qz: quartz) serving as a transmission layer. An algorithm of a method for generating a reference image for die-to-database comparison inspection will be described. Since the apparatus configuration may be the same as that of the first embodiment, description thereof is omitted.

FIG. 18 is a process sectional view of a Levenson mask.
A general drawing method of a Levenson mask that is an example of the photomask 101 will be described below. First, as shown in FIG. 18A, it is assumed that the mask blank has a two-layer structure of a Cr film 206 as a chromium layer and a glass substrate 202 as a quartz layer from the air side.
As shown in FIG. 18B, first, the chromium layer is first etched using a resist pattern formed by drawing with the drawing data A of the first layer. Next, as shown in FIG. 18C, the digging layer is etched using the resist pattern formed by drawing with the drawing data B of the second layer to form a shifter.

  Here, in the Levenson mask, the pattern edge of the second layer is applied to the inside of the chrome layer, so it is not necessary to consider the overlay error of the second layer. Therefore, the shift process step in S114 can be usually unnecessary.

FIG. 19 is a diagram for explaining Qz / shifter / Cr region extraction in the pixel Y1.
The left figure of Fig.19 (a) is sectional drawing which cut out the area | region of the pixel Y1 shown in FIG.18 (c). And the right figure of Fig.19 (a) is a top view of the area | region of the pixel Y1. The region of the pixel Y1 is a region where the shifter portion and the chrome portion are mixed among the three layers of the quartz portion, the shifter portion, and the chrome portion.

  In such a region, the area ratio occupied by the quartz portion is a difference obtained by subtracting the region ratio etched with the pattern of the drawing data B of the second layer from the region ratio etched with the pattern of the drawing data A of the first layer. Can show. Here, when the difference value becomes negative, the area ratio becomes the ratio value “0”. In FIG. 19B, (AB, 0) is shown. That is, in the pattern of the drawing data A of the first layer, only the Cr film 206 is etched. Then, in the pattern of the drawing data B in the second layer, a digging shifter is formed for a part of the exposed quartz portion. Therefore, the area ratio occupied by the quartz portion can be represented by a difference obtained by subtracting the pattern area ratio of the drawing data B of the second layer from the pattern area ratio of the drawing data A of the first layer. Here, in order to prevent unetched portions from being left behind, it is desirable that the pattern of the drawing data B in the second layer also includes the Cr film 206 region. Therefore, when viewed in pixel units, AB may be negative. In this case, “0” is obtained.

  The area ratio occupied by the Cr portion can be represented by a difference obtained by subtracting the pattern area ratio of the drawing data A of the first layer from the value “1”. In FIG. 19B, “1-A” is shown. That is, the Cr film 206 is etched with the pattern of the drawing data A of the first layer, and the shifter part is formed by a part of the exposed quartz part. The region that has not been etched becomes the Cr portion.

  The area ratio occupied by the shifter portion is the remainder of the area ratio occupied by the quartz portion and the area ratio occupied by the Cr portion.

  Here, as shown in FIG. 19C, for example, when the resolution is obtained by dividing one pixel by an area of 0 to 255, the first layer that becomes the Cr layer formation pattern in the region where the pixel Y1 is located. Assume that the pattern area ratio of the eye drawing data A is 100/255, and the pattern area ratio of the drawing data B of the second layer to be the shifter layer formation pattern is 255/255. In this case, as shown in FIG. 19D, the area ratio occupied by the quartz portion, the shifter portion, and the chrome portion in the pixel Y1 can be extracted as (0/255, 100/255, 156/255).

FIG. 20 is a diagram for explaining Qz / shifter / Cr region extraction in the pixel Y2.
The left figure of Fig.20 (a) is sectional drawing which cut out the area | region of the pixel Y2 shown in FIG.18 (c). And the right figure of Fig.20 (a) is a top view of the area | region of the pixel Y2. The region of the pixel Y2 is a region where the chrome portion occupies the entire surface among the three layers of the quartz portion, the shifter portion, and the chrome portion.

  In such a region, the area ratio occupied by the quartz portion is that of the drawing data B of the second layer from the pattern region ratio of the drawing data A of the first layer indicated as “AB, 0” in FIG. The difference obtained by subtracting the pattern area ratio and the area ratio occupied by the Cr portion are obtained by subtracting the pattern area ratio of the drawing data A of the first layer from the value “1” indicated as “1-A” in FIG. The difference and the area ratio occupied by the shifter part can be represented by the area ratio occupied by the quartz part indicated as “remaining” in FIG. 20B and the remainder of the area ratio occupied by the Cr part.

  Here, as shown in FIG. 20C, for example, when the resolution is obtained by dividing one pixel by an area of 0 to 255, the first layer that becomes the Cr layer formation pattern in the region where the pixel Y2 is located. Assume that the pattern area ratio of the eye drawing data A is 0/255, and the pattern area ratio of the second layer drawing data B to be the shifter layer formation pattern is 10/255. In this case, as shown in FIG. 20D, the area ratio occupied by the quartz portion, the shifter portion, and the chrome portion in the pixel Y2 can be extracted as (0/255, 0/255, 255/255).

FIG. 21 is a diagram for explaining Qz / shifter / Cr region extraction in the pixel Y3.
The left figure of Fig.21 (a) is sectional drawing which cut out the area | region of the pixel Y3 shown in FIG.18 (c). And the right figure of Fig.21 (a) is a top view of the area | region of the pixel Y3. The region of the pixel Y3 is a region where the quartz portion occupies the entire surface among the three layers of the quartz portion, the shifter portion, and the chrome portion.

  In such a region, the area ratio occupied by the quartz portion is that of the drawing data B of the second layer from the pattern region ratio of the drawing data A of the first layer indicated as “AB, 0” in FIG. The difference obtained by subtracting the pattern area ratio and the area ratio occupied by the Cr portion are obtained by subtracting the pattern area ratio of the drawing data A of the first layer from the value “1” indicated by “1-A” in FIG. The difference and the area ratio occupied by the shifter part can be shown by the area ratio occupied by the quartz part shown as “remaining” in FIG. 21B and the remainder of the area ratio occupied by the Cr part.

  Here, as shown in FIG. 21C, for example, when the resolution is obtained by dividing one pixel by an area of 0 to 255, the first layer that becomes the Cr layer formation pattern in the region where the pixel X3 is located. Assume that the pattern area ratio of the eye drawing data A is 255/255, and the pattern area ratio of the drawing data B of the second layer, which is the shifter layer formation pattern, is 0/255. In such a case, as shown in FIG. 21 (d), the area ratio occupied by the quartz portion, shifter portion, and chrome portion in the pixel Y3 can be extracted as (255/255, 0/255, 0/255).

FIG. 22 is a diagram for explaining Qz / shifter / Cr region extraction in the pixel Y4.
The left figure of Fig.22 (a) is sectional drawing which cut out the area | region of pixel Y4 shown in FIG.18 (c). And the right figure of Fig.22 (a) is a top view of the area | region of the pixel Y4. The region of the pixel Y4 is a region where the shifter portion occupies the entire surface among the three layers of the quartz portion, the shifter portion, and the chrome portion.

  In such a region, the ratio of the area occupied by the quartz portion is that of the drawing data B of the second layer from the pattern region ratio of the drawing data A of the first layer indicated as “AB, 0” in FIG. The difference obtained by subtracting the pattern area ratio and the area ratio occupied by the Cr portion are obtained by subtracting the pattern area ratio of the drawing data A of the first layer from the value “1” indicated by “1-A” in FIG. The difference and the area ratio occupied by the shifter part can be shown by the area ratio occupied by the quartz part shown as “remaining” in FIG. 22B and the remainder of the area ratio occupied by the Cr part.

  Here, as shown in FIG. 22C, for example, when the resolution is obtained by dividing one pixel by an area of 0 to 255, the first layer that becomes the Cr layer formation pattern in the region where the pixel X4 is located. It is assumed that the pattern area ratio of the eye drawing data A is 255/255, and the pattern area ratio of the drawing data B of the second layer to be the shifter layer formation pattern is 225/255. In this case, as shown in FIG. 9D, the area ratio occupied by the quartz portion, the shifter portion, and the chrome portion in the pixel Y4 can be extracted as (0/255, 255/255, 0/255).

  With the algorithm as described above, the Qz / shifter / Cr region in the mask drawing process described in FIG. 18 can be extracted.

  Then, as a correlation calculation step in S118, the correlation calculation circuit 350, which is an example of a correlation calculation unit, uses the area ratio of two pixels that are in point symmetry with the respective pixels from which the area ratio is extracted. Perform a correlation calculation for. Here, correlation calculation is performed using the area ratio of the pixel P ′ and the pixel Q ′ that are in a point-symmetrical position with respect to the target pixel Y <b> 1.

Hereinafter, the weighting coefficient is different from the first embodiment.
FIG. 23 is a diagram illustrating a weighting matrix for the pixel P ′ and the pixel Q ′.
As shown in FIG. 23A, there are nine combinations of the regions in the pixel P ′ and the pixel Q ′. Then, weighting factors W _{11 to} W ₃₃ for each combination are determined. As shown in FIG. 23 (a), the weighting coefficients are W ₁₁ = 1.0, W ₁₂ = W ₂₁ = t · cos φ, W ₂₂ = t ² , W ₁₃ = W ₂₃ = W ₃₁ = W ₃₂ = W ₃₃ = 0. Here, t is the amplitude transmittance, and φ is the phase. Since the area occupied by the chrome part is a light shielding area, all the weighting factors of the area related to the area ratio occupied by the chrome part are set to “0”. The amplitude transmittance t here may be a value close to 1 although it is smaller than 1.

  In addition, each process similar to Embodiment 1 is implemented.

  As described above, according to the second embodiment, the gradation value of each pixel in the Levenson mask can be calculated. As a result, it is possible to generate a highly accurate reference image using two-layer data that has not been possible in the past. Therefore, the generation of pseudo defects in the die-to-database comparison inspection can be reduced without reducing the comparison sensitivity at the boundary position between the first layer and the second layer.

  According to the above-described embodiment, a reference image that matches a transmission image or a reflection image of a multilayer phase shift mask such as a tritone, chromeless, and Levenson mask can be generated with a simple model. Resizing, corner rounding, and overlay correction due to the drawing process can be performed for each drawing data.

  In the above description, what is described as “˜circuit” or “˜process” can be configured by a computer-operable program. Or you may make it implement by not only the program used as software but the combination of hardware and software. Alternatively, a combination with firmware may be used. When configured by a program, the program is recorded on a recording medium such as a magnetic disk device, a magnetic tape device, an FD, or a ROM (Read Only Memory).

  The embodiments have been described above with reference to specific examples. However, the present invention is not limited to these specific examples.

  In addition, although descriptions are omitted for parts and the like that are not directly required for the description of the present invention, such as a device configuration and a control method, a required device configuration and a control method can be appropriately selected and used.

  In addition, all image generation apparatuses, image generation methods, and sample inspection apparatuses that include elements of the present invention and that can be appropriately modified by those skilled in the art are included in the scope of the present invention.

FIG. 6 is a diagram illustrating a main part of each step of the reference image generation method according to the first embodiment. 1 is a conceptual diagram showing a configuration of a sample inspection apparatus in Embodiment 1. FIG. It is a figure for demonstrating the acquisition procedure of an optical image. It is a block diagram which shows the internal structure of a reference circuit. It is process sectional drawing of a full tritone mask. It is a figure for demonstrating Qz / HT / Cr area | region extraction in the pixel X1. It is a figure for demonstrating Qz / HT / Cr area | region extraction in the pixel X2. It is a figure for demonstrating Qz / HT / Cr area | region extraction in the pixel X3. It is a figure for demonstrating Qz / HT / Cr area | region extraction in the pixel X4. It is a figure which shows an attention point, the area | region which performs correlation calculation, and two pixels which are a point symmetrical position. It is a figure which shows the area ratio of the pixel P and the pixel Q. It is a figure which shows the type | formula which performs area ratio calculation about the pixel P and the pixel Q. FIG. It is a figure which shows the weighting matrix about the pixel P and the pixel Q. It is a figure which shows the type | formula which performs the weighting calculation about pixel P and pixel Q, and total sum calculation. It is a figure which shows the type | formula which performs convolution. It is another process sectional drawing of a full tritone mask. It is a figure for demonstrating Qz / HT / Cr area | region extraction in pixel X1 '. It is process sectional drawing of a Levenson mask. It is a figure for demonstrating Qz / shifter / Cr area | region extraction in the pixel Y1. It is a figure for demonstrating Qz / shifter / Cr area | region extraction in the pixel Y2. It is a figure for demonstrating Qz / shifter / Cr area | region extraction in the pixel Y3. It is a figure for demonstrating Qz / shifter / Cr area | region extraction in the pixel Y4. It is a figure which shows the weighting matrix about pixel P 'and pixel Q'.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Sample inspection apparatus 101 Photomask 102 XY (theta) table 103 Light source 104 Magnification optical system 105 Photodiode array 106 Sensor circuit 107 Position circuit 108 Comparison circuit 109 Magnetic disk apparatus 110 Control computer 112 Reference circuit 115 Magnetic tape apparatus 202 Glass substrate 204 Halftone film 206 Cr film 208 Shifter 312, 314 Resizing circuit 322, 324 Corner rounding circuit 330 Shift processing circuit 340 Qz / HT / Cr region extraction circuit 350 Correlation calculation circuit 360 Convolution circuit 352 Area ratio calculation circuit 354 Weight calculation circuit 356 Total sum calculation circuit

Claims (3)

When the first and second multi-value images forming the first and second pattern layers are input to the substrate and the input first and second multi-value images are combined, An extraction unit for extracting an area ratio occupied by the base layer, the first pattern layer, and the second pattern layer;
A correlation calculation unit that performs a correlation calculation for each pixel using the area ratio of two pixels that are respectively point-symmetric with each pixel from which the area ratio is extracted;
A point response function is convolved and integrated with the result of the correlation calculation performed on each pixel to calculate the gradation value of each pixel, and an image obtained by combining the first and second multi-valued images is generated. A convolution integrator to
Equipped with a,
The correlation calculation unit is located in a predetermined region centered on each pixel, and all the two pixels that are point-symmetric with each pixel are
The product of the area ratios occupied by the base layer, the total value of the product of the area ratio occupied by the base layer and the area ratio occupied by the first pattern layer, and the first pattern layer A first calculation unit for calculating a product value of the area ratios;
The calculated product value of the area ratio occupied by the base layer and the first weighting factor, and the product value of the area ratio occupied by the base layer and the area ratio occupied by the first pattern layer. A second calculation unit that calculates a product value of the total value of the first and second weighting factors, a product value of the area ratios occupied by the first pattern layer, and a product value of the third weighting factor;
The calculated product value of the area ratio occupied by the base layer and the first weighting factor, and the product value of the area ratio occupied by the base layer and the area ratio occupied by the first pattern layer. A third calculation unit for calculating a sum of a product value of the total value of the first and second weighting factors, a product value of the area ratios occupied by the first pattern layer, and a product value of the third weighting factor When,
An image generation apparatus comprising: An input step of inputting the first and second multi-value images for forming the first and second pattern layers on the substrate;
An extraction step of extracting an area ratio occupied by the base layer, the first pattern layer, and the second pattern layer for each pixel when the input first and second multi-value images are combined. When,
A correlation calculation step of performing a correlation calculation for each of the pixels using the area ratio of the two pixels that are in point symmetry with the respective pixels from which the area ratio has been extracted;
A point response function is convolved and integrated with the result of the correlation calculation performed on each pixel to calculate the gradation value of each pixel, and an image obtained by combining the first and second multi-valued images is generated. A convolution integration step,
Equipped with a,
In the correlation calculation step, for all two pixels located in a predetermined area centered on each pixel, each corresponding to a point-symmetrical position with each pixel,
The product of the area ratios occupied by the base layer, the total value of the product of the area ratio occupied by the base layer and the area ratio occupied by the first pattern layer, and the first pattern layer Calculate the product of the area ratios,
The calculated product value of the area ratio occupied by the base layer and the first weighting factor, and the product value of the area ratio occupied by the base layer and the area ratio occupied by the first pattern layer. A product value of the total value of the first and the second weighting factor, a product value of the area ratios occupied by the first pattern layer and a product value of the third weighting factor,
The calculated product value of the area ratio occupied by the base layer and the first weighting factor, and the product value of the area ratio occupied by the base layer and the area ratio occupied by the first pattern layer. The sum of the product value of the total value of the first and the second weighting factor and the product value of the area ratio occupied by the first pattern layer and the product of the third weighting factor is calculated. Image generation method. When the first and second multi-value images as design data for forming the first and second pattern layers on the substrate are input and the input first and second multi-value images are combined, each pixel And an extraction unit for extracting an area ratio occupied by the base layer, the first pattern layer, and the second pattern layer,
A correlation calculation unit that performs a correlation calculation for each pixel using the area ratio of two pixels that are respectively point-symmetric with each pixel from which the area ratio is extracted;
A reference image obtained by combining the first and second multi-valued images by calculating the gradation value of each pixel by convolving and integrating a point response function with the result of the correlation calculation performed on each pixel. A convolution integrator to generate,
An optical image acquisition unit for acquiring an optical image of a sample to be inspected produced based on the design data;
A comparison unit that compares the optical image of the sample to be inspected with the reference image;
Equipped with a,
The correlation calculation unit is located in a predetermined region centered on each pixel, and all the two pixels that are point-symmetric with each pixel are
The product of the area ratios occupied by the base layer, the total value of the product of the area ratio occupied by the base layer and the area ratio occupied by the first pattern layer, and the first pattern layer A first calculation unit for calculating a product value of the area ratios;
The calculated product value of the area ratio occupied by the base layer and the first weighting factor, and the product value of the area ratio occupied by the base layer and the area ratio occupied by the first pattern layer. A second calculation unit that calculates a product value of the total value of the first and second weighting factors, a product value of the area ratios occupied by the first pattern layer, and a product value of the third weighting factor;
The calculated product value of the area ratio occupied by the base layer and the first weighting factor, and the product value of the area ratio occupied by the base layer and the area ratio occupied by the first pattern layer. A third calculation unit for calculating a sum of a product value of the total value of the first and second weighting factors, a product value of the area ratios occupied by the first pattern layer, and a product value of the third weighting factor When,
Specimen inspection apparatus characterized by having a.