JP2015105897A - Inspection method of mask pattern - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an inspection method of a mask pattern enabling an inspection to be performed even for a region where vignetting occurs.SOLUTION: A mask is mounted on a stage, and an optical image obtained by emitting light to an inspection region where a pattern of the mask is formed is compared with a reference image, and a place in which a difference between both images exceeds a predetermined threshold value is determined as a defect. The emission of the light to the mask is performed operating autofocusing means. When a pellicle is not mounted on the mask, a first threshold value is used, and when the pellicle is mounted on the mask, the first threshold value is used for a comparison between the optical image of a region where the region occurred with vignetting due to the pellicle is not overlapped with the inspection region and the reference image, and a second threshold value relatively looser in sensitivity than the first threshold value is used for the comparison between the optical image of the region where the region occurred with the vignetting due to the pellicle and the inspection region are overlapped and the reference image.

Description

  The present invention relates to a mask pattern inspection method.

  With the high integration and large capacity of large scale integration (LSI), the circuit dimensions required for semiconductor devices are becoming narrower. For example, recent typical logic devices are required to form patterns with a line width of several tens of nanometers.

  A photolithography technique is used in the manufacturing process of the semiconductor element. In this technology, an original image pattern (referred to as a mask or reticle; hereinafter collectively referred to as a mask) on which a circuit pattern is formed is exposed and transferred onto a wafer by a reduction projection exposure apparatus called a stepper or a scanner. For LSIs that require a large amount of manufacturing cost, it is essential to improve the yield in the manufacturing process. The defect of the mask pattern is a major factor that decreases the yield of the semiconductor element, and therefore an inspection process for detecting this is important.

  Factors that reduce the yield of semiconductor elements include other than mask pattern defects, and dust adhering to the mask surface is a typical example. This is because the dust on the mask surface is exposed and transferred onto the wafer together with the mask pattern in the semiconductor manufacturing process and becomes a defect.

  Therefore, in order to prevent dust from adhering to the mask, a pellicle is attached to the mask. The pellicle is formed by attaching a transparent pellicle film to a square frame, and is attached to the mask by fixing the frame around the pattern forming portion of the mask by adhesion or the like. By using the pellicle, dust is attached to the pellicle instead of the mask. On the other hand, the focal point at the time of exposure is set to match the pattern formed on the mask. Therefore, according to this method, it is possible to prevent the dust attached to the pellicle from being exposed and transferred to the wafer.

  In the mask inspection process, first, the presence or absence of a defect is inspected with respect to the mask before mounting the pellicle. If there is a defect, it is sent to the correction process and corrected, and then the pellicle is mounted on the mask. At this time, dust adheres to the pellicle or generates dust from the frame, so that the mask after the pellicle is mounted is also inspected. If necessary, the pellicle is peeled off, washed, etc., and then the pellicle is mounted again for inspection.

  The mask inspection is performed as follows. First, a mask is placed on the XY stage and moved to a desired position. Next, the mask is illuminated with light of a predetermined wavelength via the objective lens while the XY stage is fed in the X direction at a predetermined speed, and an optical image of the mask pattern is acquired by the sensor. At this time, the autofocus means operates so that the objective lens is focused on the pattern surface of the mask. The acquired optical image is compared with a reference image, and the presence or absence of a defect in the mask pattern is determined from the result. Here, in the case of the inspection method based on the die-to-database method, the reference image to be compared with the optical image of the mask is a reference image created based on the drawing data (design pattern data). On the other hand, in the case of the inspection method using the die-to-die method, the reference image is an optical image different from the optical image to be inspected.

  By the way, when the mask is to be inspected with the pellicle mounted, a part of the light to illuminate the mask is blocked by the frame or the inner wall of the frame is illuminated in the vicinity of the frame. As a result, a phenomenon (vignetting) in which the amount of illumination light is insufficient or scattered light from the inner wall becomes stray light occurs. Then, a clear optical image cannot be obtained by the sensor, and an accurate inspection cannot be performed.

  In the inspection process, when the XY stage moves in the X direction from the outside of the pellicle frame at one end of the mask, the autofocus spot straddles the frame of the pellicle. Here, when the autofocus spot reaches the frame, the autofocus means operates so that the objective lens is focused on the frame surface and does not focus on the pattern surface of the mask. Then, along with the movement of the XY stage, the autofocus spot moves away from the frame and returns to the mask pattern surface again. However, the autofocus means cannot follow the operation of the XY stage and does not focus on the mask pattern surface. Occurs. If the objective lens is not focused on the pattern surface of the mask, a clear optical image cannot be obtained and accurate inspection cannot be performed as described above.

  In order to solve the problem in the inspection of a mask with a pellicle, Patent Document 1 measures in advance how close the inspection point and the inner wall of the frame are to each other for each type of pellicle film or frame. An inspection apparatus that sets an inspectable area based on information is disclosed. Patent Document 2 describes that a region near the pellicle that cannot be inspected by the inspection apparatus is calculated from information on the size and position of the pellicle and the inspection capability of the inspection apparatus, and the region is visually observed.

JP-A-5-150442 JP-A-8-320295

  As seen in Patent Documents 1 and 2, in the prior art, an area where vignetting occurs is specified and excluded from the inspection target of the inspection apparatus. The pattern in the removed area is not inspected even if it is originally required to be inspected, or it is inspected visually, so that the level of quality assurance of the mask is lowered or the time required for the inspection process There is a problem that becomes longer.

  The present invention has been made in view of these points. That is, an object of the present invention is to provide a mask pattern inspection method capable of inspecting an area where vignetting occurs.

  Other objects and advantages of the present invention will become apparent from the following description.

According to one embodiment of the present invention, a mask is placed on a stage, light is irradiated onto an inspection region where a mask pattern is formed, and at least one of light transmitted through the mask and light reflected by the mask is an objective lens A mask pattern inspection method that compares an optical image obtained by forming an image on a sensor through a reference image with a reference image and determines that a difference between both images exceeds a predetermined threshold value as a defect,
The mask is irradiated with light while operating the autofocus means to focus the objective lens on the mask.
When the pellicle is not attached to the mask, the first threshold is used as the predetermined threshold,
When a pellicle is mounted on the mask, the first threshold value is used to compare the optical image of the region where the vignetting is not overlapped with the inspection region by the pellicle and the reference image, and the vignetting region and the inspection are performed by the pellicle. The second threshold value, which is relatively less sensitive than the first threshold value, is used for comparison between the optical image and the reference image in the region where the region overlaps.

Another aspect of the present invention is to place a mask on a stage, irradiate light onto an inspection region where a mask pattern is formed, and to objectively target at least one of light transmitted through the mask and light reflected by the mask. Comparing an optical image obtained by forming an image on a sensor through a lens with a reference image, a mask pattern inspection method for determining a point where a difference between both images exceeds a predetermined threshold as a defect,
The mask is irradiated with light while operating the autofocus means to focus the objective lens on the mask.
When the pellicle is not mounted on the mask, the first reference image generated from the mask pattern design data is used as the base image,
When a pellicle is mounted on the mask, the first reference image is used as a reference image to be compared with an optical image of an area where the area where vignetting occurs and the inspection area do not overlap with the pellicle, and the area where vignetting occurs due to the pellicle A second reference image that simulates blurring due to vignetting is used as the first reference image as a reference image to be compared with an optical image in a region that overlaps the inspection region.

  The second reference image is preferably generated by filtering the first reference image.

The first reference image is generated from the mask pattern design data through filtering,
The second reference image is preferably generated from the design data of the mask pattern using a filter coefficient different from the filter coefficient of the filter process when the first reference image is generated.

  It is preferable that the vignetting area is an area where the autofocus means does not normally focus on the mask pattern surface.

  The area where vignetting occurs is preferably specified by using the numerical aperture of the objective lens and the height of the frame constituting the pellicle.

  If there is a pattern that requires higher accuracy than the specified accuracy in the area where the vignetting area and inspection area overlap due to the pellicle, this pattern is inspected by issuing a warning in real time during the inspection. It is preferable to stop.

According to another aspect of the present invention, a mask is placed on a stage, light is irradiated onto an inspection region where a mask pattern is formed, and at least one of light transmitted through the mask and light reflected by the mask is used as an object. Comparing an optical image obtained by forming an image on a sensor through a lens with a reference image, a mask pattern inspection method for determining a point where a difference between both images exceeds a predetermined threshold as a defect,
The mask is irradiated with light while operating the autofocus means to focus the objective lens on the mask.
When the pellicle is not attached to the mask, the first threshold is used as the predetermined threshold,
When a pellicle is mounted on the mask, the first threshold value is used to compare the optical image of the region where the vignetting is not overlapped with the inspection region by the pellicle and the reference image, and the vignetting region and the inspection are performed by the pellicle. For an area that overlaps with the area, a warning is issued to determine whether or not to use the first threshold value for comparison between the optical image of the overlapping area and the reference image.

  If it is determined that the first threshold value is not used for the comparison between the optical image of the overlapping region and the reference image, the second threshold value, which is relatively less sensitive than the first threshold value, is used instead of the first threshold value. It is preferable to use it.

  If it is determined that the first threshold value is not used for comparison between the optical image of the overlapping area and the reference image, the optical image of the overlapping area and the reference image are changed using the first threshold value after changing the inspection area. Is preferably compared.

  According to the mask pattern inspection method of the present invention, it is possible to inspect even an area where vignetting occurs.

1 is a schematic configuration diagram of an inspection apparatus according to Embodiment 1. FIG. It is a figure explaining the acquisition procedure of an optical image. It is explanatory drawing of a mode that the frame of a pellicle blocks illumination light. It is a figure explaining the area | region where vignetting arises. 3 is a flowchart illustrating an example of an inspection method according to the first embodiment. 6 is a flowchart illustrating another example of the inspection method according to the first embodiment. 5 is a flowchart illustrating an example of an inspection method according to a second embodiment. 10 is a flowchart illustrating another example of the inspection method according to the second embodiment. It is a cross-sectional schematic diagram of the mask with the pellicle mounted. It is a typical top view of the mask with which the pellicle was mounted | worn. 6 is a flowchart illustrating another example of the inspection method according to the first embodiment. 10 is a flowchart illustrating another example of the inspection method according to the second embodiment. 10 is a flowchart illustrating an example of an inspection method according to a third embodiment.

Embodiment 1 FIG.
FIG. 1 is a schematic configuration diagram of an inspection apparatus used in the mask pattern inspection method of the present embodiment. In FIG. 1, constituent parts necessary for the present embodiment are illustrated, but other known constituent parts necessary for inspection may be included.

  The inspection method of the present invention includes a die-to-database comparison method, a die-to-die comparison method, a cell comparison method, and a pixel of interest in one image. And the surrounding pixels may be used, and the present invention can also be applied to inspection of a pattern that cannot be resolved at the light source wavelength of the inspection apparatus, such as inspection of an equal-magnification template. Hereinafter, a die-to-database comparison method will be described as an example. In this method, a reference image created from design data of a pattern to be inspected is a standard image, that is, an image that is compared with the optical image of the pattern for the purpose of defect detection.

  As illustrated in FIG. 1, the inspection apparatus 100 includes a configuration unit A that is an optical image acquisition unit, and a configuration unit B that performs processing necessary for inspection using the optical image acquired by the configuration unit A.

  The component A includes a light source 103, an XYθ stage 102 that can move in the horizontal direction (X direction, Y direction) and the rotation direction (θ direction), an illumination optical system 170 that constitutes a transmission illumination system, and autofocus means. It has a magnifying optical system 104, a photoarray sensor 105, a sensor circuit 106, a laser length measurement system 122, and an autoloader 130 that are configured to automatically adjust the focus.

  In the component A, an optical image of the mask 101 to be inspected, that is, measurement data is acquired. The measurement data is an image of a mask on which a figure based on the figure data included in the design pattern data of the mask 101 is drawn. For example, the measurement data is 8-bit unsigned data and represents the brightness gradation of each pixel.

  The mask 101 is placed on the XYθ stage 102 by the autoloader 130.

  The inspection apparatus 100 can inspect both the mask with the pellicle mounted and the mask without the pellicle mounted. In the present embodiment, the former, that is, the pellicle is mounted on the mask 101. A description will be given assuming the state. FIG. 9 is a schematic cross-sectional view of the mask 101 with the pellicle mounted. The pellicle has a square frame and a pellicle membrane attached to the frame. The pellicle is attached to the mask 101 by fixing the frame around the pattern forming portion of the mask 101 with an adhesive or the like. In the inspection process, the pattern formation portion of the mask 101 becomes an inspection region.

  The autoloader 130 is driven by the autoloader control circuit 113. The autoloader control circuit 113 is controlled by the control computer 110. When the mask 101 is placed on the XYθ stage 102, light is emitted from the light source 103 disposed above the XYθ stage 102 to the pattern formed on the mask 101. More specifically, the light beam emitted from the light source 103 is applied to the mask 101 via the illumination optical system 170.

  Below the mask 101, an enlargement optical system 104, a photoarray sensor 105, and a sensor circuit 106 are arranged. The light that has passed through the mask 101 forms an optical image on the photoarray sensor 105 via the magnifying optical system 104.

  The focus of the magnifying optical system 104 is adjusted to the pattern surface of the mask 101 by the autofocus means. Known auto-focus means can be used. Specifically, the height position of the mask 101 is detected, and servo is applied so that this position is maintained at the focal position of the objective lens constituting the magnifying optical system. For driving the mask 101 in the height direction, for example, a piezo element that can be driven minutely is used.

  For detecting the height of the mask 101, a method of irradiating the mask 101 with light for focusing and detecting the optical axis variation of the reflected light is used. For example, the mask 101 is irradiated with light from an oblique direction, and the optical axis variation of the reflected light is detected by a position sensor such as a two-divided sensor to obtain the height of the mask 101. Then, the mask 101 is finely moved up and down by the servo circuit so that the mask height signal becomes constant.

  In the inspection apparatus 100, a reflection illumination system may be used instead of the transmission illumination system. For example, it is possible to adopt a configuration in which light is irradiated from below the mask 101 and light reflected by the mask 101 is guided to the photoarray sensor via the magnifying optical system. Moreover, it is good also as a structure provided with both the transmission illumination system and the reflection illumination system, According to this structure, it is possible to acquire each optical image by the light which permeate | transmitted the mask 101, and the light reflected by the mask 101 simultaneously. is there.

  When an attempt is made to inspect the mask 101 with the pellicle mounted, a part of the light to illuminate the mask 101 is blocked by the frame or the inner wall of the frame is illuminated in the vicinity of the frame. As a result, a phenomenon (vignetting) in which the amount of illumination light is insufficient or the scattered light from the inner wall becomes stray light occurs, and a clear optical image cannot be obtained. That is, the acquired optical image has insufficient brightness.

  When the XYθ stage 102 moves in the X direction from the outside of the pellicle frame at one end of the mask, the autofocus spot straddles the frame of the pellicle. Then, when the autofocus spot reaches the frame, the autofocus means operates so that the focus of the magnifying optical system 104 is focused on the frame surface, so that the pattern surface of the mask 101 is not focused. Thereafter, as the XYθ stage 102 moves, the autofocus spot moves away from the frame and returns to the pattern surface of the mask 101 again. However, the operation of the autofocus means cannot follow the operation of the XYθ stage 102, and the pattern surface of the mask 101 is moved. There will be places that are out of focus. If the focus of the magnifying optical system 104 is not aligned with the pattern surface of the mask 101, the optical image becomes unclear and blurred, and accurate inspection cannot be performed.

  FIG. 10 is a schematic top view of the mask 101 on which the pellicle is mounted. For the sake of explanation, the pellicle film is omitted. In the inspection area of this figure, the vicinity of the frame is affected by vignetting, the influence of the autofocus means operating to focus on the surface of the frame, and the influence of the autofocus means not following the XYθ stage 102. This is a portion where a clear optical image cannot be obtained. The other part, that is, the inspection area other than the vicinity of the frame is a part where a clear optical image is obtained without being affected by vignetting.

  The pattern image of the mask 101 formed on the photoarray sensor 105 is photoelectrically converted by the photoarray sensor 105 and further A / D (analog / digital) converted by the sensor circuit 106. In the photoarray sensor 105, a plurality of light receiving elements (not shown) are arranged. An example of this sensor is a TDI (Time Delay Integration) sensor. In this case, the pattern of the mask 101 is imaged by the TDI sensor while the XYθ stage 102 continuously moves. Here, the light source 103, the magnifying optical system 104, the photoarray sensor 105, and the sensor circuit 106 constitute a high-magnification inspection optical system.

  In the configuration unit B, the control computer 110 serving as a control unit that controls the entire inspection apparatus 100 performs the position circuit 107, the comparison circuit 108, the reference circuit 112, the expansion circuit 111, the expansion circuit via the bus 120 serving as a data transmission path. The circuit 140, the autoloader control circuit 113, the stage control circuit 114, the magnetic disk device 109 as an example of a storage unit, the network interface 115 and the flexible disk device 116, the liquid crystal display 117, the pattern monitor 118, and the printer 119 are connected. The XYθ stage 102 is driven by an X-axis motor, a Y-axis motor, and a θ-axis motor controlled by a stage control circuit 114. For example, an air slider and a linear motor or a step motor can be used in combination for these drive mechanisms.

  When what is described as “˜circuit” in FIG. 1 is configured by a program, the program can be recorded in the magnetic disk device 109. For example, each circuit of the autoloader control circuit 113, the stage control circuit 114, the expansion circuit 111, the expansion circuit 140, the reference circuit 112, the comparison circuit 108, and the position circuit 107 may be configured by an electric circuit, and is controlled by the control computer 110. It may be realized as software that can be processed. Further, it may be realized by a combination of an electric circuit and software.

  The control computer 110 controls the stage control circuit 114 to drive the XYθ stage 102. The movement position of the XYθ stage 102 is measured by the laser length measurement system 122 and sent to the position circuit 107.

  Further, the control computer 110 controls the autoloader control circuit 113 to drive the autoloader 130. The autoloader 130 automatically transports the mask 101 and automatically unloads the mask 101 after the inspection is completed.

  The optical image of the mask 101 is acquired as follows.

  FIG. 2 is a diagram for explaining an optical image acquisition procedure for detecting a defect in the pattern formed on the mask 101. In this figure, it is assumed that the mask 101 is placed on the XYθ stage 102 of FIG.

As shown in FIG. 2, the inspection area on the mask 101 is virtually divided into a plurality of strip-shaped inspection areas, that is, stripes 20 ₁ , 20 ₂ , 20 ₃ , 20 ₄ ,. Each stripe can be, for example, a region having a width of several hundred μm and a length of about 100 mm corresponding to the entire length of the mask 101 in the X direction or the Y direction.

An optical image is acquired for each stripe. That is, when the optical image is acquired in FIG. 2, the operation of the XYθ stage 102 is controlled so that each of the stripes 20 ₁ , 20 ₂ , 20 ₃ , 20 ₄ ,. . For example, an optical image of the mask 101 is acquired while the XYθ stage 102 moves in the −X direction of FIG. Then, an image having a scanning width W as shown in FIG. 2 is continuously input to the photoarray sensor 105 of FIG.

To explain in detail the above example, first, to acquire the image of the first stripe 20 _1. Then, to obtain the image in the second stripe 20 _2. In this case, after the XYθ stage 102 is moved stepwise in the -Y direction, obtaining an optical image while moving in the opposite direction (X direction) and the first stripe 20 in _one direction at the time of acquisition of the image in the (-X direction) Then, an image having a scanning width W is continuously input to the photoarray sensor 105. When acquiring the image of the third stripes 20 _3, after XYθ stage 102 is moved stepwise in the -Y direction, and the direction (X direction) to obtain the image in the second stripe 20 _second reverse, i.e. , in the direction (-X direction) acquired the image in the first stripe 20 _1, XY.theta. stage 102 moves. In addition, the arrow of FIG. 2 has shown the direction and order where an optical image is acquired, and the shaded part represents the area | region which acquired the optical image.

  The pattern image formed on the photoarray sensor 105 in FIG. 1 is photoelectrically converted by the photoarray sensor 105 and further A / D (analog-digital) converted by the sensor circuit 106. Thereafter, the optical image is sent from the sensor circuit 106 to the comparison circuit 108.

  The A / D converted sensor data is input to a digital amplifier (not shown) capable of offset / gain adjustment for each pixel. The gain for each pixel of the digital amplifier is determined in the calibration process. For example, in the calibration process for transmitted light, the black level is determined during photographing of the light-shielding region of the mask 101 that is sufficiently large with respect to the area imaged by the sensor. Next, the white level is determined during imaging of the transmitted light region of the mask 101 that is sufficiently large with respect to the area imaged by the sensor. At this time, in anticipation of light quantity fluctuation during inspection, for example, the offset of each pixel is set so that the amplitude of the white level and the black level is distributed from 10% to 240% corresponding to about 4% to about 94% of the 8-bit gradation data. And adjust the gain.

  In the case of inspection by the die-to-database comparison method, the reference for defect determination is a reference image generated from design pattern data. Next, a reference image generation method will be described with reference to FIG.

  In the inspection apparatus 100, pattern data (design pattern data) used when forming the pattern of the mask 101 is stored in the magnetic disk device 109.

  The development circuit 111 reads pattern data from the magnetic disk device 109 through the control computer 110. Next, the read pattern data is converted into binary or multi-value image data (design image data). Specifically, the expansion circuit 111 expands the drawing data to data for each graphic, and interprets a graphic code, a graphic dimension, and the like indicating the graphic shape of the graphic data. Then, the pattern data is developed into binary or multi-valued image data as a pattern arranged in a grid having a grid with a predetermined quantization dimension as a unit. Further, the occupation ratio occupied by the graphic in the design pattern is calculated for each area (square) corresponding to the sensor pixel, and the graphic occupation ratio in each pixel becomes a pixel value.

  The image data converted by the expansion circuit 111 is sent to the reference circuit 112.

  The reference circuit 112 performs an appropriate filtering process on pattern data that is graphic image data. The reason is as follows.

  The pattern formed on the mask 101 has a rounded corner, a finished line width and the like in the manufacturing process, and does not exactly match the design pattern. Further, the optical image obtained from the sensor circuit 106 in FIG. 1 is blurred due to the resolution characteristics of the magnifying optical system 104 and the aperture effect generated in each light receiving element of the photoarray sensor 105, in other words, a spatial low-pass. The filter is active. Therefore, prior to the inspection, a mask to be inspected is observed, a filter coefficient that simulates a change in the manufacturing process and the optical system of the inspection apparatus is learned, and a two-dimensional digital filter is applied to the pattern data. In this way, a process for making the reference image resemble an optical image is performed.

  The learning of the filter coefficient may be performed using a mask pattern serving as a reference determined in the manufacturing process, or using a part of the pattern of the mask to be inspected (mask 101 in the present embodiment). You may go. In the latter case, a filter coefficient based on the pattern line width of the area used for learning and the degree of rounded corners is acquired and reflected in the defect determination standard for the entire mask.

  Note that when learning a filter coefficient using a mask to be inspected, there is an advantage that the filter coefficient can be learned while eliminating the influence of variations in manufacturing lots and condition variations of the inspection apparatus. However, if there is a dimensional variation in the mask plane, it will be the optimum filter coefficient for the part used for learning, but it will not necessarily be the optimum coefficient for other areas. Can be. Therefore, it is preferable to learn near the center of the mask which is not easily affected by dimensional variations in the plane. Alternatively, learning may be performed at a plurality of locations in the mask surface, and an average value of a plurality of obtained filter coefficients may be used.

  The filtered reference image is sent to the comparison circuit 108 and used as a defect inspection standard in the optical image of the mask 101. Note that data indicating the position of the mask 101 on the XYθ stage 102 output from the position circuit 107 is also sent to the comparison circuit 108.

  As described above, the figure included in the pattern data is a basic figure of a rectangle or a triangle. The magnetic disk device 109 includes information such as coordinates (x, y) at the reference position of the figure, side lengths, figure codes serving as identifiers for distinguishing figure types such as rectangles and triangles, and each pattern figure. Graphic data defining the shape, size, position, etc. Data hierarchized using clusters (or cells) is arranged in stripes, but the stripes are divided into appropriate sizes to form sub-stripes. Then, the sub stripes cut out from the optical image and the sub stripes cut out from the reference image corresponding to the optical image are input to the comparison unit in the comparison circuit 108.

  The sub-stripe input to the comparison circuit 108 is further divided into rectangular small areas called inspection frames. Then, in the comparison unit, the defect is detected by comparison in units of frames. The comparison circuit 108 is equipped with several tens of comparison units so that a plurality of inspection frames are processed simultaneously in parallel. Each comparison unit captures an unprocessed frame image as soon as one inspection frame is processed. As a result, a large number of inspection frames are sequentially processed.

  Specifically, the processing in the comparison unit is performed as follows. First, the optical image and the reference image are aligned. At this time, in addition to the parallel shift in units of sensor pixels so that the edge position of the pattern and the position of the luminance peak are aligned, the luminance values of neighboring pixels are proportionally distributed, and the alignment of less than the sensor pixels is also performed. . After completing the alignment, evaluate the level difference for each pixel between the sensor frame image and the reference frame image, or compare the differential values of the pixels in the pattern edge direction. Therefore, the defect is detected. In the configuration shown in FIG. 1, the transmission images are compared with each other. However, in the configuration using the reflection optical system, comparison between the reflection images or a comparison determination algorithm combining transmission and reflection is performed. Used. As a result of the comparison, when the difference between the two exceeds the defect determination threshold value, the point is determined as a defect.

  In the present embodiment, an area where vignetting is caused by the pellicle mounted on the mask 101 is specified, and inspection is performed by changing the defect determination threshold value between an area where the specified area and the inspection area overlap and an area where the inspection area does not overlap. This point will be described in detail below.

As described above, the focus of the magnifying optical system 104 (specifically, the objective lens) is adjusted to the pattern surface of the mask 101 by the autofocus means. For example, the mask 101 is illuminated obliquely, and the height of the mask 101 is obtained by detecting the optical axis variation of the reflected light with a position sensor such as a two-divided sensor. Then, the mask 101 is finely moved up and down by the servo circuit so that the mask height signal becomes constant. In this case, the two-divided sensor as the position sensor is composed of two diodes, and light is incident on both. If one output is α and the other output is β, the ratio of α and β changes due to movement of the optical axis.


By obtaining the value of, the position of light can be detected. Here, in order not to be affected by the inclination or deflection of the mask 101, the light needs to be illuminated within the field of view of the sensor.

  By the way, when trying to inspect the mask with the pellicle mounted, as shown in FIG. 3, in the vicinity of the frame, part of the light that should illuminate the mask is blocked by the frame, or the inner wall of the frame is illuminated. To do. As a result, a phenomenon (vignetting) in which the amount of illumination light is insufficient or scattered light from the inner wall becomes stray light occurs. Then, a clear optical image cannot be obtained by the sensor, and an accurate inspection cannot be performed.

  When the XYθ stage 102 moves in the X direction from the outside of the pellicle frame at one end of the mask in a state where the autofocus means is operated with respect to the mask 101 on which the pellicle is mounted, the autofocus spot is changed to the pellicle. It will straddle the frame. When the autofocus spot reaches the frame, the autofocus means operates so that the focus of the magnifying optical system 104 (specifically, the objective lens) is focused on the frame surface, so that the pattern surface of the mask 101 is focused. No longer. Thereafter, as the XYθ stage 102 moves, the autofocus spot moves away from the frame and returns to the pattern surface of the mask 101 again. However, the operation of the autofocus means cannot follow the operation of the XYθ stage 102, and the pattern surface of the mask 101 is moved. There will be places that are out of focus.

  In the example of FIG. 3, vignetting occurs in the region R, and a clear optical image cannot be obtained by the sensor. Also, the region R is such that the autofocus means operates to focus on the surface of the frame or the autofocus means cannot follow the XYθ stage 102. That is, in the region R, the pattern surface of the mask 101 is largely deviated from the focal position of the magnifying optical system 104. Thus, the area where vignetting occurs is where the autofocus means does not normally focus on the pattern surface of the mask 101. Specifically, the focus servo signal immediately after crossing the pellicle frame shows an extremely different value from normal. Can be considered. In the example of FIG. 3 in which oblique incident light and reflected light are used for the focus mechanism, the region R is recognized as a region in which vignetting occurs by signal processing by the autofocus means itself without needing to observe by the follow-up delay described above. it can.

  The autofocus means is not limited to the above example, and may be, for example, a pattern projection method. In this method, first, illumination light is vertically incident on the mask by the illumination optical system, and a focus position detection pattern is projected onto the mask. At this time, the pattern is projected so as to be imaged on the focal plane of the objective lens. Further, the two sensors are arranged so that the in-focus plane comes to a position shifted back and forth along the optical axis with the image plane of the focal position detection pattern interposed therebetween. In other words, these sensors are used to detect the focus position when the mask surface is shifted forward (front pin) or the mask surface is shifted backward (rear pin) with respect to the focal plane of the objective lens. The pattern is arranged at a position where the contrast value of the pattern becomes maximum. The shift amount and the shift direction from the in-focus position of the mask are detected based on the magnitude relationship between the contrast values of the patterns detected by the two sensors. Based on the detected result, the height of the mask is adjusted by the servo circuit so as to be positioned on the focal plane of the objective lens.

  Even in the case of autofocus means using the pattern projection method, illumination light is blocked near the frame of the pellicle, so that the amount of illumination light is insufficient or scattered light from the inner wall of the frame becomes stray light as described above. This causes a problem that a clear optical image cannot be obtained by the sensor. Similarly, the autofocus unit operates so as to focus on the surface of the frame, or the autofocus unit cannot follow the XYθ stage, so that a clear optical image cannot be obtained. Therefore, the region where vignetting occurs in this case can also be specified by the control signal from the focus servo system indicating an abnormal value as described above.

  In the present embodiment, the region where vignetting occurs can be obtained as follows.

The numerical aperture NA of the objective lens is represented by the sine of θ, as shown in Equation (2), where θ is the angle formed by the peripheral ray of the illumination light and the optical axis.

In FIG. 4, assuming that the height of the frame is H, when the distance between the pattern formed on the mask 101 and the frame is equal to or less than D, vignetting occurs in a part of the light incident on the objective lens. As a result, the amount of illumination light is insufficient, or the scattered light from the inner wall becomes stray light, so that a clear optical image cannot be obtained by the sensor and accurate inspection cannot be performed. Therefore, in this case, the range from the frame to the distance D is an area where vignetting occurs. The distance D is given by equation (3).

  After obtaining the area where vignetting occurs as described above, it is determined whether or not the area overlaps the inspection area. If the two do not overlap, defect determination is performed using the threshold S1 (first threshold) for the entire inspection region. That is, all the optical images acquired in the inspection region are compared with the reference image corresponding to each optical image, and each difference is obtained. When the difference exceeds the threshold value S1, the location is determined as a defect.

  On the other hand, when the area where vignetting occurs and the inspection area overlap, the defect determination threshold S2 (second threshold) in the overlapped area is set to be relatively less sensitive than the threshold S1. This is due to the following reason.

  In a region where vignetting occurs, it is difficult to obtain a bright optical image with the photoarray sensor 105, and the autofocus unit does not normally focus on the pattern surface of the mask 101, so that the obtained optical image is blurred. Become. Therefore, if a strict defect determination threshold value is applied, there is a possibility that the portions determined to be defects may continue, which is not practical.

  By the way, the mask 101 is usually formed with a mixture of various types of patterns depending on the application. All of these patterns are not necessarily formed with high accuracy. For example, the types of main patterns used in semiconductor integrated circuits need to be formed with high accuracy because they affect device quality. On the other hand, the power pattern of the peripheral circuit, the dummy pattern for adjusting the drawing accuracy, the pattern reflecting the substrate information and the information required in the semiconductor manufacturing process are not required to be as accurate as the main pattern. Such information patterns include, for example, an ID pattern for the operator to visually distinguish the mask, a bar code pattern for each device used in the semiconductor manufacturing process to identify the mask, a logo pattern for the company name, and exposure by a stepper. Sometimes, an arrow pattern for confirming the direction of the mask, an end mark pattern for confirming the end point of etching, and the like are included.

  For the main pattern that needs to be formed with high accuracy, the detection accuracy in the inspection process is important as well as the accuracy during pattern formation. That is, even a very small defect needs to be detected. On the other hand, there is no problem even if a pattern that does not need to be formed with high accuracy is not detected if it is a defect of a certain size.

  In general, the main pattern that needs to be formed with high accuracy is arranged near the center of the mask 101. On the other hand, a pattern that does not need to be formed with high accuracy is arranged in the peripheral portion of the mask 101. Here, the area where the vignetting occurs is the periphery of the pattern formation portion of the mask 101 because the frame of the pellicle is fixed around the pattern formation portion of the mask 101. Therefore, even if the defect determination threshold value of the inspection area overlapping the area where vignetting occurs is set loosely, the pattern in this area is not required to be formed with high accuracy, and thus there is often no practical problem.

  As described above, there is no problem in the inspection even if the threshold value S2 is set to be lower than the threshold value S1, and even if the optical image obtained by the photoarray sensor 105 is inferior in clarity by doing in this way, Inspection using can be performed. That is, according to the present embodiment, it is possible to inspect a vignetting region that has been removed from the inspection target of the inspection apparatus in the prior art.

  Information on the area where vignetting occurs, for example, (1) the position at which the control signal from the focus servo system shows an abnormal value, or (2) theoretically the basis for calculating the area where the peripheral ray of illumination light is vignetted The numerical aperture NA, the height H of the pellicle frame, and the like are stored in the magnetic disk device 109 of FIG. The development circuit 140 reads area data such as an area where vignetting occurs, an inspection area, and an area where both overlap from the above information through the control computer 110 from the magnetic disk device 109. And the sensitivity designation information of each area | region data is acquired. For example, if the area where vignetting occurs and the inspection area overlap, the sensitivity designation information is defined as level 1, and if both do not overlap, the sensitivity designation information is defined as level 2.

  The inspection area may be made into finer area data, and sensitivity designation information corresponding to each may be defined. For example, when the circuit pattern provided in the inspection area is examined in detail, it consists of several patterns with different importance levels. For example, a pattern used as a clock signal line when operating as an LSI is a highly important pattern that requires low line width error and edge roughness, and needs to be inspected with high sensitivity. On the other hand, the dummy and shield patterns are not required to have a slight dimensional variation or the presence of defects. Furthermore, the importance of the pattern used for the power supply is about the middle between the clock pattern and the dummy or shield pattern, and the inspection sensitivity is between these. Therefore, for example, if the area data is a pattern used as a clock signal line, the sensitivity designation information is level 2-3. If the area data is a pattern used as a power supply, the sensitivity designation information is level 2-2. If the area data of the pattern is used, the sensitivity designation information is defined as level 2-1. Then, the development circuit 140 sets a threshold corresponding to each area data. By doing in this way, generation | occurrence | production of a pseudo defect can be suppressed and shortening of inspection time can be aimed at.

  It is also conceivable that there is a pattern that is highly important and needs to be formed with high accuracy in a region where the vignetting region and the inspection region overlap. In such a case, it is preferable to define a plurality of region data corresponding to the importance level as a region where both overlap, and to define sensitivity designation information for each region data. For example, for area data of a pattern that is low in importance and has no problem in the inspection using the threshold value S2, the sensitivity designation information is set to level 1-1. On the other hand, the sensitivity designation information is set to level 1-2 for the area data of the pattern that has high importance and is difficult to be substantially inspected with the threshold value S2.

  The importance level information of the pattern is stored together with the position information, for example, in the magnetic disk device 109. In this case, it is preferable that the magnetic disk device 109 further stores a conversion table indicating the correlation between the importance information of the pattern and the sensitivity designation information. The expansion circuit 140 uses the conversion table stored in the magnetic disk device 109 to acquire sensitivity specification information of the area data.

  Next, the expansion circuit 140 expands the area data into area image data. In the area image data, multivalued values are set in squares corresponding to pixels, and the value of each pixel is an occupation ratio occupied by a figure in the design pattern.

  Next, the development circuit 140 synthesizes the pattern-developed area image data. For example, the reference image and the optical image may be combined with the size of the comparison unit, or may be combined with one mask.

  Here, since the area indicated by the area data has a dimension in which a margin is added to the corresponding pattern dimension, the areas may overlap each other. That is, a plurality of area data with overlapping areas may be input to the expansion circuit 140. In such a case, it is preferable to create the combined region image data as follows.

  When the area data of the area where the vignetting area overlaps the inspection area and the area data of the area where these areas do not overlap, the expansion circuit 140 indicates the area obtained by synthesizing the plurality of area data where the areas overlap. Create area image data. Then, S2 is applied as a defect determination threshold to the overlapping area indicated by the combined area image data. Here, if S1 is applied as the defect determination threshold value, an unclear optical image is determined with a strict threshold value, so that there is a possibility that pseudo defects will continue and substantial inspection cannot be performed. Such a situation can be prevented by setting S2 as a threshold value. Note that when inspecting the overlapping area indicated by the above-described combined area image data by applying S2 as the defect determination threshold, the inspection apparatus 100 may issue a warning.

  If the area data based on the importance of the pattern set by the designer overlaps with the area data by the external software that analyzes the pattern shape, the development circuit 140 reflects the intention of the designer and outputs the area image data. create.

  The area image data created as described above is output to the comparison circuit 108.

  The comparison circuit 108 takes in the area image data created by the development circuit 140, uses the defect determination threshold S1 or S2 determined by the area image data, and uses an optical image and a reference image according to a predetermined algorithm for each pixel of the area image data. Compare When the error exceeds the threshold value, the location is determined as a defect. If it is determined as a defect, the coordinates and the optical image and reference image that are the basis for the defect determination are stored in the magnetic disk device 109 as the inspection result.

  Assuming that there is a pattern that needs to be formed with high accuracy in a region where the vignetting region and the inspection region overlap, for example, level 1-1 or level 1-2 as described above as sensitivity designation information Is defined, it is preferable that a warning is issued from the inspection apparatus 100 as follows.

  For example, the comparison circuit 108 reads the importance level information of the pattern stored in the magnetic disk device 109 through the control computer 110. Further, the comparison circuit 108 grasps the inspection position based on data indicating the position of the mask 101 on the XYθ stage 102 output from the position circuit 107. Then, the inspection position is matched with the position information in the pattern importance information, and a warning is issued if the area is defined by level 1-2. The warning can be displayed on the liquid crystal display 117 from the comparison circuit 108 through the control computer, for example.

  For the area where the warning is issued, the inspection is temporarily stopped, and thereafter, the threshold is changed and the inspection is performed again, or the visual inspection is also performed. Thereby, the inspection accuracy with respect to the pattern which needs to be formed with high accuracy can be increased. Note that it is obvious that the visual inspection area in this case is narrower considering that the area of the visual inspection performed in the prior art is the entire area where vignetting occurs.

  The inspection result stored in the magnetic disk device 109 is reviewed by the operator. The review is an operation of determining whether or not the detected defect is a problem in practical use. For example, the operator compares the reference image that is the basis for the defect determination with the optical image including the defect, and determines whether the defect needs to be corrected. The defect information determined through the review process is also stored in the magnetic disk device 109. When at least one defect to be corrected is confirmed in the review process, the mask 101 is sent together with the defect information list to a correction device (not shown) which is an external device of the inspection apparatus 100. Since the correction method differs depending on whether the defect type is a convex defect or a concave defect, the defect type and the defect coordinates including the distinction between the irregularities are attached to the defect information list.

  As described above, the comparison circuit 108 determines the defect while changing the defect determination threshold value of the comparison target area according to the area image data. That is, even in a region where vignetting occurs, it is possible to perform defect determination by the comparison circuit 108 by using a loose defect determination threshold. Therefore, according to the present embodiment, it is possible to inspect the entire mask on which the pellicle is mounted, and the area can be minimized even if visual inspection is necessary. As well as improving the quality assurance level, it is possible to shorten the time required for the inspection process.

  In addition, the inspection apparatus 100 of the present embodiment has two development circuits (140, 111). The development circuit 111 converts the pattern data used when forming the pattern of the mask 101 into binary or multi-value image data (design image data). On the other hand, the expansion circuit 140 converts area data such as an area where vignetting occurs, an inspection area, and an area where both overlap each other into area image data, and determines a defect determination threshold value. By doing in this way, the inspection apparatus 100 can perform two conversion processes in parallel.

  However, the present embodiment is not limited to the above, and the processing in the development circuit 140 may be performed by the control computer 110. In this case, the determined defect determination threshold is stored in the magnetic disk device 109. At the time of inspection, the control computer 110 reads the defect determination threshold value from the magnetic disk device 109. Then, the comparison circuit 108 compares the optical image with the reference image with reference to this value.

  FIG. 5 is a flowchart showing an example of a mask pattern inspection method according to this embodiment.

  First, in S101, mask loading is performed. This is executed by the following procedure, for example. The load button is selected by the operator on the operation screen shown on the liquid crystal display 117 of the inspection apparatus 100. Thereby, a command signal is issued from the control computer 110, and the stage control circuit 114 moves the XYθ stage 102 to the load position (X, Y) based on this command signal.

  Next, in S102, an image of the entire mask 101 is taken. From the captured image, it is determined whether or not the pellicle is mounted on the mask 101 (S103). If the pellicle is not mounted, the inspection is started. The defect determination threshold value in this case can be set to S1 (first threshold value), for example.

  If it is determined in S103 that the pellicle is mounted, after specifying the area where vignetting occurs, the area overlapping with the inspection area is further specified, and the defect determination threshold of the area is relaxed. For example, when S1 is used as the defect determination threshold in the inspection area where no vignetting occurs, S2 is used as the threshold in the overlapping area. Specifically, this procedure is performed through S104 to S108.

  First, in S104, accurate alignment of the mask 101 on the XYθ stage 102 is performed. Subsequently, after the calibration described above is performed (S105), an area A where vignetting occurs is specified (S106). In the area A, the servo system control signal in the autofocus means may indicate an abnormal value, and the equation (3) is used from the angle θ formed by the peripheral ray of the illumination light and the height H of the frame. You may ask. As shown in the equation (2), since the sine of the angle θ is the numerical aperture NA of the objective lens, the area A can be obtained using the numerical aperture NA of the objective lens and the height H of the frame.

  Next, in S107, it is determined whether or not the area A and the inspection area B overlap. If these areas do not overlap, the inspection is started without relaxing the defect determination threshold. In the above example, the threshold value S1 (first threshold value) is used.

  If it is determined in S107 that there is an overlap between the area A and the inspection area B, the defect determination threshold of the overlapping area is relaxed (S108). In the above example, the inspection is performed with the threshold value relaxed from S1 to S2 (second threshold value).

  The inspection process in S109 is as described above. That is, an optical image of the mask 101 is acquired and compared with a reference image generated from the pattern data of the mask 101. For example, when the difference between these images exceeds the threshold value S1 or the threshold value S2, the location is determined as a defect.

  FIG. 6 is a flowchart showing another example of the mask pattern inspection method according to this embodiment.

  In FIG. 6, S201 to S207 are the same as S101 to S107 of FIG.

  If it is determined in S207 that there is an overlap between the vignetting area A and the inspection area B, the process proceeds to S208, and it is determined whether or not there is a highly important pattern in the overlapping area. For example, the importance level information of the pattern stored in the magnetic disk device 109 of FIG. 1 is referred to. If there is no pattern with high importance, the relaxed threshold value obtained in S209 is used, and the process proceeds to S210 for inspection.

  On the other hand, if there is a highly important pattern, a warning is issued and the inspection is stopped. About the area | region where the warning was issued, after finishing the test | inspection by the test | inspection apparatus 100, a test | inspection by visual observation is performed.

  The inspection process of S210 is performed according to the method described above. That is, an optical image of the mask 101 is acquired and compared with a reference image generated from the pattern data of the mask 101. For example, when the difference between these images exceeds the threshold value S1 or the threshold value S2, the location is determined as a defect.

  FIG. 11 is a flowchart showing another example of the mask pattern inspection method according to this embodiment.

  In FIG. 11, S501 to S507 are the same as S101 to S107 of FIG.

  If it is determined in S507 that there is no overlap between the area A and the inspection area B, if these areas do not overlap, the defect determination threshold is not relaxed and the inspection is started. In this case, for example, a threshold value S1 (first threshold value) is used.

  On the other hand, if it is determined in S507 that there is an overlap between the area A and the inspection area B, the defect determination threshold of the overlapping area is relaxed (S508). For example, the threshold value is relaxed from S1 to S2 (second threshold value).

  In the example of FIG. 11, when the threshold value is relaxed in S508, the inspection apparatus 100 issues a warning (S509). Next, it is determined whether or not the examination is forced using the second threshold value S2 (S510). This determination is made by, for example, processing in the control computer 110 by reading out pattern importance information stored in the magnetic disk device 109 of FIG. For example, if there is no pattern with high importance in the overlapping area, it is determined that the inspection is forced, and if there is a pattern with high importance, it is determined that the inspection is not forced.

  If the inspection is not forced, the process returns to S508 and the second threshold is reset. And the process of S508-S510 is repeated until it becomes the threshold value according to the importance of the pattern. Note that the inspection area B may be changed instead of resetting the threshold value. In this case, after determining that the inspection is not forced in S510, the process returns to S507. As a result of changing the inspection area B, if it is determined in S507 that there is no overlap between the area A and the inspection area B, the process proceeds to S511 and the inspection is performed.

  The inspection process in S511 is as described above. That is, an optical image of the mask 101 is acquired and compared with a reference image generated from the pattern data of the mask 101. For example, when the difference between these images exceeds the threshold value S1 or the threshold value S2, the location is determined as a defect.

  In the example of FIG. 11 as well, it is possible to inspect the entire mask on which the pellicle is mounted. Therefore, it is possible to suppress a reduction in the level of quality assurance of the mask due to the removal of the area where vignetting occurs in the conventional method from the inspection target.

  In the above example, the die-to-database comparison method has been described as an example, but the same applies to the die-to-die comparison method. That is, when vignetting has occurred on one of the dies, the defect determination threshold value applied to the vignetting region is replaced with the first threshold value, and the second threshold value, which is relatively less sensitive than the first threshold value, This makes it possible to perform an inspection even in an area where vignetting occurs.

Embodiment 2. FIG.
The present embodiment is directed to an inspection method using a die-to-database method. In this method, the reference image to be compared with the optical image of the mask is a reference image created based on drawing data (design pattern data).

  In the first embodiment, an area where vignetting is caused by the pellicle mounted on the mask is specified, and inspection is performed by changing the defect determination threshold value between an area where the specified area and the inspection area overlap and an area where the inspection area does not overlap. On the other hand, in the present embodiment, the inspection in the overlapping area is made possible by processing the reference image in the die-to-database inspection method.

The inspection method of the present embodiment can be implemented using the inspection apparatus 100 of FIG. 1 described in the first embodiment. The inspection process is also substantially the same as that in FIG. 5 of the first embodiment, but differs in that the reference image is processed instead of the threshold reduction in S108. That is,
(1) When no pellicle is attached to the mask 101,
(2) When the pellicle is mounted, but the area where vignetting occurs and the inspection area do not overlap,
(3) The reference image used in (1) and (2) and the reference image used in (3) are changed when the pellicle is mounted and the region where vignetting occurs and the inspection region overlap.

  As described in the first embodiment, pattern data (design pattern data) used at the time of pattern formation of the mask 101 is stored in the magnetic disk device 109 of FIG. 1, and the development circuit 111 reads out this pattern data to obtain image data ( (Designed image data). After the image data is sent to the reference circuit 112, the reference circuit 112 performs a filtering process to generate a reference image. In the above cases (1) and (2), the comparison circuit 108 detects a defect in the optical image using this reference image.

  On the other hand, in the case of (3), since vignetting occurs in the inspection region, it is difficult to obtain a clear optical image with the photoarray sensor 105. Here, the reference image used in (1) and (2) is generated by performing processing (filter processing) for making the image data resemble an optical image. The optical image at this time is bright and blurred. There is no clear image. Therefore, if it is attempted to perform defect determination by comparing a reference image generated assuming a clear image with an unclear optical image, pseudo defects may occur successively, which is not practical.

  Therefore, in the present embodiment, processing is performed to make the reference image resemble the optical image obtained in the area where vignetting occurs. That is, with respect to the reference image used in (1) and (2), the optical image becomes dark due to vignetting, the autofocus means is focused on the surface of the frame, or cannot follow the XYθ stage 102. Then, processing is performed in consideration of blurring of the optical image. For example, a filter coefficient that simulates blurring due to vignetting is further applied to image data after filter processing corresponding to the reference image used in (1) and (2) (hereinafter referred to as the first reference image). Perform the filtering process used. When vignetting occurs, the black level rises above the level determined in the calibration process, and the white level falls below the level determined in the calibration process. Based on this, a reference image suitable for the case of (3) ( Hereinafter, this is referred to as a second reference image.). Alternatively, in addition to the filter coefficient different from that used when generating the first reference image for the image data sent from the development circuit 111 to the reference circuit 112, specifically, changes due to the manufacturing process and the optical system of the inspection apparatus The second reference image may be generated by learning a filter coefficient that simulates blurring due to vignetting and then applying a two-dimensional digital filter to the pattern data. In any method, the generation of the second reference image can be performed by the reference circuit 112 in FIG.

  FIG. 7 is a flowchart showing an example of the inspection method according to the present embodiment.

  In FIG. 7, S301 to S307 are the same as S101 to S107 in FIG.

  In S307 of FIG. 7, it is determined whether or not there is an overlap between the area A and the inspection area B where vignetting occurs. If there is no overlap in these areas, it can be said that it corresponds to the above (1) or (2), so the process proceeds to S309, and an inspection is performed using the first reference image.

  On the other hand, if it is determined in S307 that there is an overlap between the area A and the inspection area B, the first reference image is processed to generate a second reference image (S308). The second reference image is obtained by adding blurring of the optical image due to vignetting to the first reference image.

  After the second reference image is generated, the process proceeds to S309 for inspection. The inspection process in S309 is as described in the first embodiment. That is, an optical image of the mask 101 is acquired and compared with a reference image generated from the pattern data of the mask 101. At this time, the first reference image is used in the inspection region B that does not overlap with the region A where vignetting occurs, and the second reference image is used in the inspection region B that does not overlap with the region A where vignetting occurs. As a result of the comparison, when the difference between the optical image and the reference image exceeds a predetermined threshold, the location is determined as a defect.

  FIG. 8 is a flowchart showing another example of the inspection method according to the present embodiment.

  In FIG. 8, S401 to S407 are the same as S101 to S107 in FIG.

  If it is determined in S407 that there is no overlap between the vignetting area A and the inspection area B, the process proceeds to S410, and an inspection is performed using the first reference image.

  On the other hand, if it is determined that there is an overlap between the area A and the inspection area B, the process proceeds to S408, and it is determined whether there is a pattern having a high importance in the overlapping area. For example, the importance level information of the pattern stored in the magnetic disk device 109 of FIG. 1 is referred to. If there is no pattern with high importance, the reference image is processed in S409, and a second reference image in which the optical image is blurred due to vignetting is generated. Thereafter, the process proceeds to S410, and an inspection is performed using the second reference image.

  On the other hand, if there is a highly important pattern, a warning is issued and the inspection is stopped. About the area | region where the warning was issued, after finishing the test | inspection by the test | inspection apparatus 100, a test | inspection by visual observation is performed.

  The inspection process in S410 is the same as S309 in FIG.

  The inspection results obtained as described above are stored, for example, in the magnetic disk device 109 of FIG. 1 and then reviewed by the operator. The review is an operation of determining whether or not the detected defect is a problem in practical use. The operator, for example, compares the first reference image or the second reference image that is the basis for the defect determination with the optical image including the defect, and determines whether the defect needs to be corrected.

  The defect information determined through the review process can also be stored in the magnetic disk device 109. When at least one defect to be corrected is confirmed in the review process, the mask 101 is sent together with the defect information list to a correction device that is an external device of the inspection apparatus 100. Since the correction method differs depending on whether the defect type is a convex defect or a concave defect, the defect type and the defect coordinates including the distinction between the irregularities are attached to the defect information list.

  The inspection result in the inspection apparatus 100 is stored in a common server (not shown). The data may be stored in a storage unit (for example, the magnetic disk device 109 in FIG. 3) in the inspection apparatus 100.

  The operator reads the inspection result from the common server by using a review device (not shown), and determines whether or not the detected defect needs to be corrected. The defect information determined through the review process is stored in the common server as a review result.

  When at least one defect to be corrected is confirmed by the review apparatus, the mask 101 is sent to a correction apparatus (not shown). The correction device reads the review result from the common server and makes necessary corrections accordingly. Since the correction method differs depending on whether the defect type is a convex defect or a concave defect, the review result includes the defect type including the unevenness and the defect coordinates.

  FIG. 12 is a flowchart showing another example of the inspection method according to this embodiment.

  In FIG. 12, S601 to S607 are the same as S101 to S107 in FIG.

  If it is determined in S607 that there is no overlap between the vignetting area A and the inspection area B, the process proceeds to S611 and the inspection is performed. The reference image (first reference image) used in the inspection is an image generated by performing processing (filtering processing) to resemble the optical image on the image data obtained by converting the mask pattern data. Here, the model optical image is a clear image that is bright and free of blur.

  On the other hand, if it is determined in S607 that there is an overlap between the area A and the inspection area B, the process proceeds to S608. In S608, the first reference image is processed to generate a second reference image. The second reference image is obtained by adding blurring of the optical image due to vignetting to the first reference image.

  In the example of FIG. 12, when the second reference image is generated in S608, the inspection apparatus 100 issues a warning (S609). Next, it is determined whether or not the examination is to be performed using the second reference image (S610). This determination is made by, for example, processing in the control computer 110 by reading out pattern importance information stored in the magnetic disk device 109 of FIG. For example, if there is no pattern with high importance in the overlapping area, it is determined that the inspection is forced, and if there is a pattern with high importance, it is determined that the inspection is not forced.

  If the inspection is not forced, the process returns to S608 and the first reference image is reprocessed. And the process of S608-S610 is repeated until it becomes the 2nd reference image according to the importance of a pattern. Note that the inspection region B may be changed instead of reprocessing the reference image. In this case, after determining that the inspection is not forced in S610, the process returns to S607. As a result of changing the inspection area B, if it is determined in S607 that there is no overlap between the area A and the inspection area B, the process proceeds to S611 and the inspection is performed.

  The inspection process in S611 is as described in the first embodiment. That is, an optical image of the mask 101 is acquired and compared with a reference image generated from the pattern data of the mask 101. At this time, the first reference image is used in the inspection region B that does not overlap with the region A where vignetting occurs, and the second reference image is used in the inspection region B that does not overlap with the region A where vignetting occurs. As a result of the comparison, when the difference between the optical image and the reference image exceeds a predetermined threshold, the location is determined as a defect.

  Also by the example of FIG. 12, it is possible to inspect the entire mask on which the pellicle is mounted. Therefore, it is possible to suppress a reduction in the level of quality assurance of the mask due to the removal of the area where vignetting occurs in the conventional method from the inspection target.

Embodiment 3 FIG.
In the first embodiment, handling of an area where an area where vignetting occurs and an inspection area overlap is described. That is, the threshold value S2 having a low sensitivity is applied to such an overlapping area, but when there is a pattern that needs to be formed with high accuracy in the overlapping area, a warning is issued from the inspection apparatus. . Here, the warning is issued in real time during the inspection. For example, in FIG. 1, the comparison circuit 108 reads the importance level information of the pattern stored in the magnetic disk device 109 through the control computer 110. Further, the comparison circuit 108 grasps the inspection position based on data indicating the position of the mask 101 on the XYθ stage 102 output from the position circuit 107. Then, the inspection position is matched with the position information in the pattern importance information, and if the pattern has a high importance pattern, a warning is issued and the inspection is stopped.

  On the other hand, in the present embodiment, whether or not to apply the same threshold value (first threshold value) as the inspection region not affected by vignetting to the region where the region where the vignetting occurs and the inspection region overlap, Make a warning and make a decision before testing. Hereinafter, the inspection method of the present embodiment will be described in detail.

  FIG. 13 is a flowchart showing an example of a mask pattern inspection method according to this embodiment. This inspection method can be implemented using the inspection apparatus 100 of FIG. 1 described in the first embodiment. Therefore, the following description will be made with reference to FIG.

  First, before the inspection, a first threshold is determined (S701). Thereafter, the inspection process starts. In the inspection process, first, in S702, mask loading is performed. This is executed by the following procedure, for example. The load button is selected by the operator on the operation screen shown on the liquid crystal display 117 of the inspection apparatus 100. Thereby, a command signal is issued from the control computer 110, and the stage control circuit 114 moves the XYθ stage 102 to the load position (X, Y) based on this command signal.

  Next, in S703, an image of the entire mask 101 is taken. From the captured image, it is determined whether or not the pellicle is mounted on the mask 101 (S704). If the pellicle is not mounted, the process proceeds to S710 and the inspection is started. In this case, the defect determination threshold is the first threshold.

  If it is determined in step S704 that the pellicle is mounted, accurate alignment of the mask 101 on the XYθ stage 102 is performed in step S705. Subsequently, after the calibration described in the first embodiment is performed (S706), an area A in which vignetting occurs is specified (S707).

In the area A, the servo system control signal in the autofocus means may indicate an abnormal value, and the equation (4) is used from the angle θ formed by the peripheral ray of the illumination light and the height H of the frame. You may ask. A range from the frame to the distance D is defined as a region A where vignetting occurs.

Furthermore, as shown in Expression (5), since the sine of the angle θ is the numerical aperture NA of the objective lens, the region A can also be obtained using the numerical aperture NA of the objective lens and the frame height H.

  Next, in S708, it is determined whether or not the first threshold value is applied to the overlapping area between the area A and the inspection area B.

  In S708, when it is determined to perform the inspection using the first threshold value for the overlapping region, the process proceeds to S710 and the inspection is performed. The inspection process is as described in the first embodiment. That is, an optical image of the mask 101 is acquired and compared with a reference image generated from the pattern data of the mask 101. For example, when the difference between these images exceeds the first threshold value, the point is determined as a defect. Although there is a possibility that pseudo defects frequently occur by applying the first threshold value, in such a case, a warning may be issued again.

  On the other hand, if it is determined in S708 that the examination is not forced, the second threshold is set by relaxing the threshold in S709. Then, it progresses to S710 and an inspection is performed.

  The determination as to whether or not to perform the inspection in S708 can be made with reference to, for example, the importance level information of the pattern stored in the magnetic disk device 109 of FIG. In this case, if there is no highly important pattern in the overlapping area, the inspection using the first threshold value is not forced, and the process proceeds to S709 to relax the threshold value.

  Also according to the present embodiment, it is possible to inspect the entire mask on which the pellicle is mounted. Therefore, it is possible to suppress a reduction in the level of quality assurance of the mask due to the removal of the area where vignetting occurs in the conventional method from the inspection target. Further, since it is determined whether or not the inspection is to be performed with the first threshold before the inspection is performed, it is possible to prevent the occurrence of pseudo defects due to the inspection with the first threshold.

  In the present embodiment, the inspection region B can be changed instead of reducing the threshold value. For example, after determining that the inspection is not forced, the inspection region B is changed so that there is no highly important pattern in the overlapping portion with the region A. Thereafter, the inspection area B is inspected using the first threshold value.

  In the first to third embodiments, the inspection of the mask on which the pellicle is mounted has been described. Here, a general mask inspection is performed before and after mounting the pellicle. For example, first, the presence or absence of defects is inspected with respect to the mask before mounting the pellicle (inspection I). If there is a defect, it is sent to the correction process and corrected, and then the pellicle is mounted on the mask. At this time, dust adheres to the pellicle or dust is generated from the frame, so the mask after the pellicle is mounted is also inspected (inspection II). Then, if necessary, the pellicle is peeled off, washed, etc., and then the pellicle is mounted again for inspection (inspection III). In such a case, the first to third embodiments can be applied to the inspection II and the inspection III.

  The present invention is not limited to the above-described embodiments, and various modifications can be made without departing from the spirit of the present invention.

  In the above embodiments, descriptions of parts that are not directly required for the description of the present invention, such as the device configuration and control method, are omitted. However, the required device configuration and control method may be appropriately selected and used. Needless to say, you can. In addition, all inspection apparatuses or inspection methods 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.

  Further, in this specification, what is described as “to circuit” or “to part” can be configured by a computer-operable program. However, not only software programs but also hardware and software It may be implemented by a combination or a combination with firmware. When configured by a program, the program can be recorded on a recording device such as a magnetic disk device.

DESCRIPTION OF SYMBOLS 100 Inspection apparatus 101 Mask 102 XY (theta) stage 103 Light source 104 Enlargement optical system 105 Photoarray sensor 106 Sensor circuit 107 Position circuit 108 Comparison circuit 109 Magnetic disk apparatus 110 Control computer 111,140 Expansion circuit 112 Reference circuit 113 Autoloader control circuit 114 Stage control circuit 115 Network Interface 116 Flexible Disk Device 117 Liquid Crystal Display 118 Pattern Monitor 119 Printer 120 Bus 122 Laser Length Measuring System 130 Autoloader 170 Illumination Optical System

Claims (8)

A mask is placed on the stage, light is irradiated onto the inspection area where the mask pattern is formed, and at least one of the light transmitted through the mask and the light reflected by the mask is connected to the sensor through the objective lens. An inspection method of a mask pattern that compares an optical image obtained by imaging with a reference image and determines that a difference between both images exceeds a predetermined threshold value as a defect,
Irradiating the mask with light while activating autofocus means for focusing the objective lens on the mask;
When no pellicle is attached to the mask, the first threshold is used as the predetermined threshold,
When a pellicle is attached to the mask, the first threshold value is used to compare the optical image of the area where the vignetting area and the inspection area do not overlap with the reference image, and the pellicle A second threshold value having a relatively lower sensitivity than the first threshold value is used to compare the optical image of the region where the vignetting region and the inspection region overlap with the reference image. Inspection method. A mask is placed on the stage, light is irradiated onto the inspection area where the mask pattern is formed, and at least one of the light transmitted through the mask and the light reflected by the mask is connected to the sensor through the objective lens. An inspection method of a mask pattern that compares an optical image obtained by imaging with a reference image and determines that a difference between both images exceeds a predetermined threshold value as a defect,
Irradiating the mask with light while activating autofocus means for focusing the objective lens on the mask;
When a pellicle is not attached to the mask, a first reference image generated from design data of the mask pattern is used as the reference image,
When a pellicle is attached to the mask, the first reference image is used as a reference image to be compared with an optical image of a region where the vignetting region and the inspection region do not overlap with the pellicle, and the pellicle A second reference image that simulates blurring due to the vignetting is used for the first reference image as a reference image to be compared with an optical image of a region where the vignetting region and the inspection region overlap. Mask pattern inspection method.   The mask pattern inspection method according to claim 2, wherein the second reference image is generated by filtering the first reference image. The first reference image is generated through a filter process from design data of the mask pattern,
The second reference image is generated from design data of the mask pattern by using a filter coefficient different from a filter coefficient of a filter process when the first reference image is generated. Item 3. A mask pattern inspection method according to Item 2.   If there is a pattern that requires higher accuracy than a predetermined accuracy in the region where the vignetting area and the inspection region overlap, a warning is issued in real time during inspection and the pattern 5. The mask pattern inspection method according to claim 1, wherein the inspection is stopped. A mask is placed on the stage, light is irradiated onto the inspection area where the mask pattern is formed, and at least one of the light transmitted through the mask and the light reflected by the mask is connected to the sensor through the objective lens. An inspection method of a mask pattern that compares an optical image obtained by imaging with a reference image and determines that a difference between both images exceeds a predetermined threshold value as a defect,
Irradiating the mask with light while activating autofocus means for focusing the objective lens on the mask;
When no pellicle is attached to the mask, the first threshold is used as the predetermined threshold,
When a pellicle is attached to the mask, the first threshold value is used to compare the optical image of the area where the vignetting area and the inspection area do not overlap with the reference image, and the pellicle A warning is issued for a region where the vignetting region and the inspection region overlap, and it is determined whether or not the first threshold value is used for comparison between the optical image of the overlapping region and the reference image. A mask pattern inspection method characterized by the above.   If it is determined that the first threshold value is not used for the comparison between the optical image of the overlapping region and the reference image, the sensitivity is relatively less than the first threshold value instead of the first threshold value. The mask pattern inspection method according to claim 6, wherein a second threshold is used. If it is determined that the first threshold value is not used for comparison between the optical image of the overlapping area and the reference image, the overlapping area is changed using the first threshold value after changing the inspection area. The mask pattern inspection method according to claim 6, wherein the optical image is compared with the reference image.