JP4652391B2 - Pattern inspection apparatus and pattern inspection method - Google Patents

Description

  The present invention relates to a pattern inspection apparatus for a sample to be inspected and a pattern inspection method, and more particularly to a pattern inspection apparatus for a mask such as a reticle used for manufacturing a semiconductor element or a liquid crystal display panel, and a pattern inspection method. It is.

  Conventionally, a pattern inspection apparatus compares optical images obtained by imaging a pattern formed on a sample to be inspected such as a reticle at a predetermined magnification, or a reference image obtained from the optical image and design data. The pattern is inspected. That is, there is a die-die inspection (DD inspection) in which optical images obtained from the same pattern at different locations on the same reticle are compared to detect defects in the reticle pattern. Alternatively, a reference image similar to the optical image of the pattern drawn on the reticle is created from the design data of the reticle, and the optical image and the reference image are compared to detect a defect in the reticle pattern. Inspection).

  Patent Document 1 discloses a method for performing a DD inspection by performing alignment processing using information on a plurality of dies on a reticle. However, this inspection does not consider the distortion of the pattern itself or the quantization error. This is because it is basically sufficient to obtain the position correction amount for each die by interpolating the positional accuracy of repeated transfer during wafer transfer. On the other hand, a high-precision inspection in each die comparison with a reticle requires a higher alignment accuracy and a reference image in which fluctuation of the reference die itself is suppressed. A die is a pattern area on a collective mask corresponding to one chip of an LSI formed on a wafer. At present, a mask pattern as an original is often formed with a size four times as large as an actual LSI pattern, but the pattern on this mask is not limited to one LSI chip. For example, 2, 3, 4,... Chips are formed on one mask, and the pattern is transferred onto the wafer at a time. In this case, the mask pattern area corresponding to each chip is called a die.

  A method for relatively aligning an optical image and a reference image in a reticle overview inspection apparatus is disclosed in Patent Document 2. However, this method only uses the alignment result in the adjacent frame as the offset at the start of alignment, and does not increase the accuracy of the alignment itself. Neither of these two alignment processes is to obtain a “highly accurate reference image for pattern inspection”.

  In recent years, semiconductors have been miniaturized, and accordingly, higher precision is required for reticle inspection. In a conventional DD inspection, an optical image of a die that is an inspection sample is aligned with a reference image of a die that is a reference, and both images are compared. However, since both images to be compared are optical images obtained by scanning, distortion and noise are generated in each of the inspection die and the reference die. There is a limit to increase the accuracy of DD inspection by increasing the alignment error and the difference amount between images due to the characteristics having the error.

  In addition, in the DD inspection of the miniaturized pattern, the resolution of the image is insufficient, and the optical image is blurred, so that the quantization error for the fine pattern becomes a signal level fluctuation, which occurs between the comparison target dies, S / N cannot be obtained sufficiently. Furthermore, since a minute portion is enlarged and imaged, various noises are stacked on the image.

  That is, distortion (in the worst reverse direction) between the inspection die to be inspected and the reference die, quantization error (in the case where the characteristics are contradictory due to the positional relationship, for example, the edge is on the quantization point or the half pixel is shifted) In addition, when noise in the reverse direction is applied to the corresponding pixel, the alignment amount for aligning the inspection die with the reference die is greatly deviated, resulting in alignment processing noise and extra level differences between pixels. This is a cause of a decrease in sensitivity of DD inspection.

In addition, due to the miniaturization of the target pattern, subtle alignment processing at the time of optical image acquisition at the critical resolution and the influence of quantization mesh displacement, especially for patterns below the resolution vicinity, the reference die The image of the image fluctuates and the reproducibility is poor. Therefore, a stable and high-quality reference image cannot be obtained. As a result, the variation of the inspection result that appears as a difference between the reference image and the inspection image becomes large, which is a constraint condition of the inspection result level that can be compensated.
JP 2004-273850 A JP 2000-147749 A

(1) The present invention is to improve the accuracy of pattern inspection of a sample to be inspected.
(2) Further, the present invention is to generate a highly accurate pattern inspection reference image and perform an alignment process on the reference image, thereby obtaining a highly accurate alignment result.

(1) The present invention is an optical image acquisition unit that acquires an optical image of a pattern of a sample to be inspected, and the optical image acquired by the optical image acquisition unit is generated from design data of the sample pattern to be inspected. An alignment processing unit that performs an alignment process including a correction process using a reference image, and a plurality of optical images of different locations based on the same design data on the sample to be inspected, processed by the alignment processing unit And a comparison processing unit that compares the optical image corrected by the alignment processing unit and the optical image combined by the synthesis processing unit .
(2) Moreover, this invention acquires the optical image of the pattern of a to-be-inspected sample, and performs the alignment process including the correction process using the reference image produced | generated from the design data of the pattern of to-be-inspected sample with respect to the said optical image. A plurality of optical images of different locations based on the same design data on the inspected sample that have undergone alignment processing are combined, and the alignment processed optical image is compared with the combined optical image Thus, there is a pattern inspection method for detecting a pattern difference of a sample to be inspected.

(1) The present invention can increase the accuracy of pattern inspection of a sample to be inspected.
(2) In addition, the present invention can generate a high-accuracy pattern inspection reference image, and by performing an alignment process on the reference image, a high-precision alignment result can be obtained as a result.

  Hereinafter, pattern inspection of a sample to be inspected such as a reticle according to an embodiment of the present invention will be described.

  The pattern inspection according to the embodiment of the present invention is created from the design data of the pattern of the sample to be inspected with respect to the optical image of the pattern of the sample to be inspected (that is, the sample on the inspection target side) acquired by the optical image acquisition unit. Then, a fitting alignment process, that is, an alignment process including the fitting process is performed using the reference image (that is, the reference side image of the inspection), and the DD inspection or the DB inspection is performed. This fitting alignment process is an alignment process including a correction process using a reference image. The pattern inspection performs a DD inspection by comparing optical images subjected to the alignment process including the correction process. In the pattern inspection, the optical image subjected to the alignment process including the correction process is compared with the reference image, and the DB inspection is performed. Alternatively, the pattern inspection is performed by combining the optical images subjected to the alignment process including the correction process, and comparing the optical image subjected to the alignment process including the correction process with the optical image subjected to the combination process. DB inspection is performed. The correction process is a correction process using a reference image generated from the design data of the sample pattern to be inspected with respect to the optical image acquired by the optical image acquisition unit.

  In the embodiment of the present invention, the alignment process including the correction process is a process of correcting and aligning the image of the sample to be inspected. For example, the reference side image of the inspection to be compared is subjected to the inspection target side. This is alignment processing for aligning images so as to overlap each other. The alignment process including the correction process includes, for example, the influence of the pixel on the reference image corresponding to the target pixel of the optical image and its peripheral pixels, with the optical image as the target side and the reference image created from the design data as the reference side. Considering this, alignment processing is performed to correct distortion of the optical image in addition to misalignment errors between the images by performing alignment processing on the optical image. By this correction processing, alignment with sub-pixel accuracy can be performed, and distortion between the optical image and the reference image, quantization error, noise in alignment processing, an extra pixel level difference, and the like can be reduced.

  The synthesis process is a process of creating a composite image by using a plurality of alignment-processed optical images including a correction process. The synthesis process is, for example, a method of synthesizing an average of a plurality of alignment-processed optical images including a correction process, that is, an image integration process.

(Pattern inspection device)
FIG. 1 shows a block diagram of the pattern inspection apparatus 10. The pattern inspection apparatus 10 includes an optical image acquisition unit 20 that acquires an optical image 12 of a pattern of a sample to be inspected such as a reticle, and a reference image that creates a reference image 14 that is similar to the optical image 12 from a database of patterns of the sample to be inspected. The creation unit 58, the alignment processing unit 60 that performs the alignment process including the correction process on the optical image 12 using the reference image 14, and the optical image subjected to the alignment process including the correction process are compared (DD inspection 620). A comparison processing unit 62 or a comparison processing unit 62 that performs comparison processing (DB inspection 622) between the alignment-processed optical image including the correction processing and the reference image is provided. The optical image 12 and the reference image 14 are compared in image processing units. The size of the processing unit is an arbitrary processing unit according to the processing function of the pattern inspection apparatus, such as a frame unit or a die unit. Can be taken. Hereinafter, processing of the optical image 12 and the reference image 14 will be described in units of dies.

  FIG. 2 shows an example of a pattern inspection procedure by the DD inspection 620. In the optical image of the reticle 300 of the sample to be inspected, an optical die 302 (optical die 1, optical die 2,...) Is shown. The reference image of the design data database 580 shows a reference die 582 (reference die 1, reference die 2,...). The optical image of the reticle 300 and the reference image of the database are assigned the same number (1, 2,...) For the same corresponding die.

  The optical die 1 of the optical image is subjected to alignment processing including correction processing using the reference die 1 of the reference image, and becomes a reference die 304 (reference die 1 ') for the DD inspection 620. The optical die 2 of the optical image, the optical die 3,... Are subjected to an alignment process including a correction process using the reference die 1 of the reference image to become an inspection die 306 (inspection die 2 ′,...). The inspection dies 2 ',... That have been subjected to the alignment process including the correction process are compared with the reference die 1', a DD inspection 620 is performed, and a defect of each die is detected. In this way, since the comparison processing targets are all subjected to alignment processing including correction processing using the reference image die 1 obtained from the design data, deformation such as distortion of the optical image is corrected, and defects in the optical image are sub- It can be found at the pixel level. Here, which is the reference die 304 and which is the inspection die 306 can be arbitrarily determined. The optical die 1 and the reference die 1 are the same corresponding die. The reference die 1 is a different die from the optical die 2, the optical die 3,..., But is designed to have the same pattern.

  FIG. 3 shows an example of a pattern inspection procedure by the DB inspection 622. The optical image of reticle 300 and the reference image of database 580 show the same die pattern as in FIG. In FIG. 3, the optical die 302 (optical die 1, optical die 2,...) Of the optical image is the die to be inspected. The optical die 1 is subjected to an alignment process including a correction process using the reference die 1, and becomes an inspection die 1 '. The inspection die 1 ′ is compared with the reference die 1 of the standard die, the DB inspection 622 is performed, and the defect of the optical die 1 is detected. Similarly, the optical die 2 is subjected to an alignment process including a correction process using the reference die 2, and becomes an inspection die 2 '. The inspection die 2 ′ is compared with the reference die 2 of the standard die, the DB inspection 622 is performed, and the defect of the optical die 2 is detected. As described above, since the comparison processing target is subjected to the alignment processing including the correction processing by the reference die 1, the reference die 2,... Of the reference image obtained from the design data, deformation such as distortion of the optical image is performed. Corrected, defects in the optical image can be found on the order of sub-pixels. Here, the reference dies for the inspection die 1 ′, the inspection die 2 ′,... Are the reference die 1, the reference die 2,.

  FIG. 4 shows an example of another procedure of pattern inspection by the DB inspection 622. In FIG. 3, the reference dies of inspection die 1 ′, inspection die 2 ′,... Are reference die 1, reference die 2,. The standard die of the die 1 ′, the inspection die 2 ′,... Is different from the pattern inspection of FIG. 3 in that it is a specific reference die 1 of the reference image. In this way, since the comparison processing targets are all subjected to alignment processing including correction processing using the reference image die 1 obtained from the design data, deformation such as distortion of the optical image is corrected, and defects in the optical image are sub- It can be found by the degree of pixels, and the inspection processing procedure can be simplified.

  In FIG. 5, a method for obtaining a reference die (the reference die in this case is referred to as a combined die) by performing a combining process using the inspection die 1 ′, the inspection die 2 ′,. Is used. An averaging process or the like is applied to the synthesis process. FIG. 5 shows an example of a pattern inspection procedure for performing the DD inspection 620 with the inspection die 1 ′, the inspection die 2 ′,. In this way, the comparison processing targets are all subjected to alignment processing including correction processing by the die 1 of the reference image obtained from the design data to correct deformation such as distortion of the optical image, and defects in the optical image are corrected to the sub-pixels. In addition, noise can be removed by performing synthesis processing due to effects such as averaging processing.

(Pattern inspection method)
FIG. 6 shows a flowchart of a pattern inspection method for inspecting an acquired optical image. The pattern inspection method includes an optical image acquisition step (S1) for acquiring a pattern formed on an inspection sample such as a reticle as an optical image 12, and a reference image 14 similar to the optical image from the design data of the pattern of the inspection sample. A reference image creation step (S2) to be created, an alignment processing step (S3) for performing an alignment process including a correction process on the optical image 12 using the reference image 14, and a comparison process between the alignment-processed optical images The comparison processing step (S4) for performing the DD inspection 620 or the comparison processing step (S4) for performing the DB inspection 622 by comparing the optical image 12 subjected to the alignment processing with the reference image 14 is provided. is doing.

  FIG. 7 shows a flowchart of the pattern inspection method in which the synthesis processing step (S5) is added to FIG. In the synthesis processing step (S5), a plurality of optical images subjected to the alignment processing are synthesized to create a synthesized image. Using this composite image as a reference image, the DD inspection 620 is performed by comparing with the optical image 12 subjected to the alignment processing.

(Configuration example of pattern inspection device)
FIG. 8 shows a configuration example of the pattern inspection apparatus 10. The pattern inspection apparatus 10 includes an optical image acquisition unit 20 that detects reflected light or transmitted light from the sample 30 to be inspected to acquire an optical image, a data processing unit 50 that stores data such as image data, and performs arithmetic processing. I have.

(Optical image acquisition unit)
The optical image acquisition unit 20 acquires the optical image 12 from the pattern of the sample 30 to be inspected such as a reticle. The sample 30 to be inspected is placed on the XYθ table 32. The XYθ table 32 is a three-axis (XY-θ) positioning stage that moves in the X and Y directions and rotates in the θ direction. The XYθ motor 34 is driven and controlled by the table control unit 56 that has received a command from the central processing unit 52, and moves the XYθ table 32 in the X, Y, and θ directions. As the XYθ motor 34, a known servo motor, step motor, or the like can be used. The position coordinates of the XYθ table 32 are measured by the laser length measurement system 28, and the output is sent to the position measurement unit 66. The position coordinates of the XYθ table 32 output from the position measurement unit 66 are fed back to the table control unit 56.

  The sample 30 to be inspected is automatically supplied onto the XYθ table 32 by the autoloader 22 under the control of the autoloader control unit 54 and is automatically discharged after the inspection is completed. A light source 24 and an irradiation unit 26 are disposed above the XYθ table 32. Light from the light source 24 irradiates the sample 30 to be inspected via the condenser lens of the irradiating unit 26. Below the sample 30 to be inspected, a signal detection unit including an expansion optical system 36 and a photodiode array 38 is arranged. The transmitted light that has passed through the specimen 30 is imaged on the light receiving surface of the photodiode array 38 via the magnifying optical system 36. The magnifying optical system 36 is automatically focus-adjusted by a focus adjusting device (not shown) such as a piezo element. This focus adjustment device is controlled by an autofocus control circuit (not shown) connected to the central processing unit 52. The focus adjustment may be monitored with an observation scope provided separately. The photodiode array 38 as a photoelectric conversion unit is a line sensor or an area sensor provided with a plurality of photosensors. By continuously moving the XYθ table 32 in the X-axis direction, the photodiode array 38 detects a measurement signal corresponding to the pattern of the sample 30 to be inspected.

  The measurement signal is converted into digital data by the sensor circuit 40 and input to the buffer memory 64 as optical image data. A plurality of buffer memories 64 can be provided. The output of the buffer memory 64 is sent to the comparison processing unit 62. The optical image data is, for example, 8-bit unsigned data, and represents the brightness of each pixel. This type of pattern inspection apparatus 10 normally reads out these pattern data from the photodiode array 38 in synchronization with a clock frequency of about 10 MHz to 30 MHz, and rearranges the appropriate data, and then performs two-dimensional raster scanning. Treated as image data.

(Optical image acquisition procedure)
An optical image of a pattern drawn on the inspection sample 30 such as a reticle is acquired by scanning the inspection sample 30 with the optical image acquisition unit 20. The pattern of the reticle 300 is acquired as, for example, a strip-shaped optical image that is cut long in the direction of one side of the reticle 300 (for example, the X direction). The strip is a stream. The stream is an optical image of a further elongated strip that is divided into four in one side direction (X direction). The stream divided into four is a substream. The substream is cut into a plurality in the direction of the other side (Y direction) orthogonal to the side (X direction). The cut optical image is a frame. The frame is, for example, an optical image having 512 pixels on one side (X direction) and 512 pixels on the other side (Y direction). One pixel has a gray scale of 256 gradations.

(Data processing part)
The data processing unit 50 is mainly similar to the optical image 12 from the design data, the central processing unit 52 that performs data processing, the autoloader control unit 54 that controls the autoloader 22, the table control unit 56 that controls the XYθ table 32 A reference image creating unit 58 that creates a reference image 14 to be performed, an alignment processing unit 60 that performs an alignment process including a correction process on the optical image 12 using the reference image 14, and an optical image processed by the alignment processing unit 60. Measurement is performed by a synthesis processing unit 68 for processing, a comparison processing unit 62 for comparing an optical image with a reference image and inspecting a defect of the inspection image, a buffer memory 64 for temporarily storing optical image data, and a laser side length system 28 The position measurement unit 66 for obtaining the position of the sample 30 to be inspected from the position data of the XYθ table 32, design data data An external storage device 70 that stores a large amount of data such as a base and programs, a main storage device 72 that stores data and programs necessary for arithmetic processing, a printer 74, a CRT 76, a bus 78, and the like are provided. For example, the design data of the reticle 300 is stored by dividing the entire inspection area into strip-shaped areas.

(Reference image creation part)
The reference image creating unit 58 expands the design data in the database 580 to form image data, and then performs a process of rounding the corners of the figure of the image data, a process of blurring the outline of the figure, and the like. The reference image 14 is created by performing a process similar to that of FIG. Since the reference image 14 is created from the design data, there is no variation such as distortion, deformation, level, gradation, etc., which occurs in the optical image acquisition unit 20.

(Alignment processing unit including correction processing)
The alignment processing unit 60 performs alignment processing including correction processing on the optical image 12 obtained by the optical image acquisition unit 20 using the reference image 14 generated from the design data by the reference image creation unit 58. By the alignment process including the correction process, the inspection-side image is aligned with the reference-side image with sub-pixel accuracy, and at the same time, the alignment is performed with sub-pixel accuracy, thereby obtaining an image free from distortion and deformation. be able to. The correction processing here refers to performing image correction on the optical image of the sample to be inspected based on the reference image created from the design data. That is, since there is distortion in the optical image, this distortion is corrected by performing an alignment process including a correction process based on the “reference image created from the design data”, and an alignment process is performed in units of subpixels. .

(Example of alignment processing including correction processing using a reference image)
For details of the example of the alignment process, first, a two-dimensional linear prediction model is set by regarding the reference image as two-dimensional input data and the optical image as two-dimensional output data. Here, a 5 × 5 two-dimensional linear prediction model using a 5 × 5 pixel region is taken as an example. Table 1 shows the suffixes (corresponding to the positions of 5 × 5 pixels) used in this model. In FIG. 9, the left figure is a reference image, and the right figure is an optical image (see Japanese Patent Application Laid-Open Nos. 2006-30518 and 2006-266860).

Let two-dimensional input data and two-dimensional output data be u (i, j) and y (i, j), respectively. The suffix of the pixel of interest is i, j, and the suffixes of a total of 25 pixels around 2 rows and 2 columns surrounding this pixel are set as shown in Table 1. A relational expression such as Expression (1) is set for a certain set of 5 × 5 region pixel data. The coefficients b _{00 to} b ₄₄ of each input data u (i, j) in the equation (1) are model parameters to be identified.

The expression (1) means that one pixel data y _k = y (i, j) of the optical image is expressed by a linear combination of 5 × 5 pixel data surrounding one pixel of the corresponding reference image. (See FIG. 9A).

(Simultaneous equation solving step (model parameter identification))
When Expression (1) is expressed as a vector, Expression (2) is obtained. Here, the unknown parameter vector α is expressed as α = [b ₀₀ ,
b ₀₁ , ..., b ₄₄ ] ^T , and the data vector x _k is x _k = [u (i-2, j-2), u (i-2, j-1),
..., u (i + 2, j + 2)] ^T.

If the coordinates i and j of the reference image and the optical image are scanned and 25 sets of data are combined, the model parameter can be identified. Actually, from a statistical point of view, n (> 25) sets of data are prepared as in Equation (3), and a 25-dimensional simultaneous equation is solved based on the following least-squares method to identify α. . Here, A = [x ₁ , x ₂ ,... X _n ] ^T , y = [y ₁ , y ₂ ,... Y _n ] ^T , x _k ^T α = y _k , and k = 1 , 2, ... n.

  For example, if the reference image and the optical image are each 512 × 512 pixels, the periphery of the image is reduced by two pixels by scanning the 5 × 5 model, so the number of equations is Equation (4), and 258064 sets Data will be obtained. Thereby, it is possible to ensure a sufficient number from a statistical viewpoint.

(Generation of optical image after alignment processing)
Substituting the identified model parameter α and the input / output image data used for identification into Equation (1), and performing a correction operation using the reference image by performing a simulation calculation that scans the coordinates i and j of the pixel. An alignment process is performed. The optical image that has undergone this alignment processing is reduced based on the least square method to reduce pixel position deviation, expansion / contraction noise, resizing processing, and sensing noise of less than one pixel.

(Composition processing part)
The combination processing unit 68 performs a plurality of alignment-processed optical image combining processes including a correction process to generate one new image. The synthesis process may be, for example, averaging a plurality of alignment-processed optical images including a correction process, or performing a median value process, thereby removing noise and improving S / N. It is done. By using this combined image as a reference image for DD inspection, the accuracy of the reference image is improved. As a result, for example, this is effective for an inspection using the volume variation of the contact hole pattern as a feature amount.

  The effect of the averaging processing method for a plurality of optical images is shown in FIG. The averaging process itself is in accordance with general image processing means. Three thin solid lines in FIG. 10 indicate cross-sectional images of the contact holes of the three optical images subjected to the alignment process including the correction process. The broken line with fine dots in FIG. 10 shows a cross-sectional image of the contact hole of the reference image obtained from the design data. A broken line of large dots in FIG. 10 shows a cross-sectional image of a contact hole obtained by averaging three optical images subjected to alignment processing including correction processing, and the averaging processing is performed as a reference image for DD inspection. It has been shown that the accuracy is improved by using the obtained image.

  Alternatively, as another example of the synthesis process, the synthesis processing unit 68 performs the following three processes using a plurality of optical images subjected to the alignment process including the correction process. (1) Position estimation: A sampling position shift at the time of photographing each image data is estimated from a plurality of image data. (2) High-band interpolation: Each image data is densified using a wide-band low-pass filter that transmits all high-frequency components of the original signal including the aliasing components. (3) Weighted sum: By taking a weighted sum using a weight corresponding to the sampling position of each densified data, the aliasing distortion is canceled and at the same time the high frequency component of the original signal is restored. By performing these three processes, it is possible to generate an image that reproduces the pattern edge position with high accuracy (less than a fraction of one pixel). Effective for inspection of minute changes (CD (critical dimension) inspection).

(Comparison processor)
The comparison processing unit 62 mainly performs comparison processing such as DD inspection comparison 620, DB inspection comparison 622, and defect analysis. The comparison processing unit 62 may have a parallel processing function. In that case, the comparison processing unit 62 includes a plurality of the same functions, and can process the comparison processes in parallel. The pattern inspection apparatus 10 can be configured by an electronic circuit, a program, a computer, or a combination thereof.

  As described above, an image obtained by performing an alignment process including a correction process using the reference image 14 generated from the design data on the optical image 12 is used for the comparison inspection. The image subjected to the alignment process including can be applied to the DB inspection as it is. Further, since the alignment process including the correction process is performed using the reference image 14, an error due to the alignment can be estimated.

  In the above embodiment, the alignment process is not performed using the optical image 12 obtained by scanning the sample 30 to be inspected, but the correction process is performed using the reference image 14 created from the design data in the database 580. A highly sensitive DD inspection can be performed by performing an alignment process including Specifically, in the above embodiment, the DD inspection and the DB inspection can be efficiently performed by one alignment process, and the relative alignment amount with the reference image of each inspection die 302 can be known. When the images of the die 302 are integrated, a super-resolution image of the inspection die 302 can be obtained using this alignment amount. Alternatively, by applying the alignment information for the reference image to each DB for DB inspection, the reference image is regenerated, so that the DB inspection with high accuracy can be performed.

  Further, according to the embodiment of the present invention, since the relative alignment amount between the inspection die of each optical image and the reference die of the reference image can be known with high accuracy more than several times the pixel accuracy, the images of the inspection die are integrated. When this is done, this alignment amount is used to reproduce the super-resolution image of the inspection die (the result of shifting the image in units smaller than the pixel, and regenerating an image finer than the imaged pixel. A super-resolution image is generated from a plurality of captured images of an object, but in this case, a super-resolution image is generated from captured images of different dies of the same pattern. Therefore, there is a great advantage that defects and defects inherent to individual dies are removed and an ideal die image is obtained.

  Further, according to the embodiment of the present invention, by applying the alignment information for the reference image to the DB inspection for each inspection die, for example, the regeneration of the reference image from the design data is first performed according to the alignment. By creating a reference image at exactly the same position, it is possible to eliminate errors caused by aligning the reference image and overlaying it on the image of the sample to be inspected, and to perform DB inspection with higher accuracy Can do.

  Since various combinations other than the above embodiment can be considered, it goes without saying that the present invention is not limited to the embodiment described here.

Block diagram of pattern inspection equipment Explanatory drawing of DD inspection between optical dies subjected to alignment processing including correction processing Explanatory drawing of DB inspection of optical die and reference die subjected to alignment processing including correction processing Explanatory drawing of other DB inspections of the optical die and the reference die subjected to alignment processing including correction processing Explanatory drawing of DD inspection of an optical die that has been subjected to a synthesis process and an optical die that has been subjected to the alignment process and an optical die that has been subjected to the alignment process including the correction process Flow chart of pattern inspection method for DD inspection and DB inspection Flow diagram of pattern inspection method for DD inspection of combined optical die and optical die subjected to alignment processing Configuration diagram of pattern inspection equipment Explanatory drawing of alignment processing including correction processing Illustration of composition processing

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 ... Pattern inspection apparatus 12 ... Optical image 14 ... Reference image 20 ... Optical image acquisition part 22 ... Autoloader 24 ... Light source 26 ... Irradiation part 28 ... Laser length measurement System 30 ... Inspected sample 300..Reticle 302..Optical die 304..Reference die 306..Inspection die 32 ... XY.theta. Table 34 ... XY.theta. Photodiode 40 ... sensor circuit 50 ... data processing unit 52 ... central processing unit (CPU)
54 ... Autoloader control unit 56 ... Table control unit 58 ... Reference image creation unit 580 ... Database 582 ... Reference die 60 ... Alignment processing unit 62 ... Comparison processing unit 620 ... DD inspection 622 ··· DB check 64 ··· buffer memory 66 ··· position measuring unit 68 ··· composition processing unit 70 ··· external storage device 72 ··· main storage device 74 ··· printer 76 ··· CRT
78 ... Bus

Claims (5)

An optical image acquisition unit for acquiring an optical image of a sample pattern to be inspected;
An alignment processing unit that performs an alignment process including a correction process using a reference image generated from design data of a sample pattern to be inspected with respect to the optical image acquired by the optical image acquisition unit;
A synthesizing unit that synthesizes a plurality of optical images at different locations based on the same design data on the sample to be inspected, which is processed by the alignment processing unit;
A pattern inspection apparatus comprising: an optical image processed by an alignment processing unit; and a comparison processing unit that performs a comparison process on the optical image synthesized by the synthesis processing unit. The pattern inspection apparatus according to claim 1,
The alignment processing unit uses a reference image generated from the design data of the sample pattern to be inspected with respect to the optical image acquired by the optical image acquisition unit, and includes alignment processing for correcting a misalignment error between images. A pattern inspection device that performs processing. The pattern inspection apparatus according to claim 1 or 2,
A pattern inspection apparatus that performs an averaging process as a synthesis process. The pattern inspection apparatus according to claim 1 or 2,
A pattern inspection apparatus that performs intermediate value processing as synthesis processing. Obtain an optical image of the pattern of the sample to be inspected,
Performing an alignment process including a correction process using a reference image generated from the design data of the pattern of the sample to be inspected for the optical image;
A plurality of optical images of different locations based on the same design data on the inspected sample subjected to the alignment process are synthesized,
A pattern inspection method for detecting a difference in a pattern of a sample to be inspected by comparing the alignment-processed optical image and the composite-processed optical image .