JP2013200456A - Method of manufacturing substrate for mask blank, method of manufacturing mask blank, method of manufacturing transfer mask, and method of manufacturing semiconductor device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To detect existence of an optically non-uniform region on a translucent substrate more appropriately in a method of manufacturing a substrate for mask blank.SOLUTION: Provided is a method of manufacturing a substrate for mask blank by using a translucent substrate, which includes: a substrate preparing step S102 in which a translucent substrate having two main surfaces facing to each other are prepared; an imaging step S202 in which one main surface of the translucent substrate in the state where laser beam is applied at least to one main surface; a detection step S204 in which image processing is performed to an image of the imaged main surface and laser beam is applied to an optically non-uniform region which exists in the translucent substrate, by which existence of light scattered from the region is detected; and a selection step S206 in which a translucent substrate having no scattered light in the detection step S204 is selected as a substrate for mask blank, and the image processing includes processing of obtaining a light quantity difference of the captured image by a remote differential.

Description

  The present invention relates to a mask blank substrate manufacturing method, a mask blank manufacturing method, a transfer mask manufacturing method, and a semiconductor device manufacturing method.

  In photolithography technology, a transfer mask having a thin film with a pattern formed on a light-transmitting substrate such as a glass substrate is used, and exposure light transmitted through the transfer mask is used as an object (resist film on a semiconductor wafer). Etc.), the pattern is transferred. It is necessary to use a translucent substrate used for a transfer mask and a mask blank serving as an original plate thereof that does not have an optically non-uniform region that causes a reduction in the amount of exposure light.

  However, a translucent substrate used for manufacturing a mask blank may have a concave defect such as a scratch on the main surface in a process of polishing the main surface. In addition, air bubbles or foreign matter may be mixed inside or internal striae may occur at the stage of the glass ingot before cutting into the shape of the translucent substrate. A mask blank is manufactured by forming a thin film for pattern formation on the main surface of the translucent substrate having such defects, and a transfer mask is formed by forming a transfer pattern on the thin film of the mask blank by dry etching or the like. If a portion where the transfer pattern from which the thin film is removed does not exist (white portion) is arranged in a portion where the defect exists, there is a possibility that a transfer failure due to the defect occurs. For this reason, it is preferable that the translucent board | substrate with the concave defect etc. of a magnitude | size beyond a predetermined value, or internal striae is excluded.

  With respect to such a problem, conventionally, with respect to the translucent substrate after the main surface has been polished, concave defects and the like existing on the main surface and optically non-uniform portions existing inside (bubbles, foreign matters, internal veins) Patent Document 1 discloses a method for inspecting the presence or absence of internal defects. In this inspection method, the angle at which the laser beam is incident on the inside of the transparent substrate is set so that the incident laser beam is totally reflected when it hits the main surface through the inside of the transparent substrate. The laser beam repeats total reflection at the main surface and end face inside the translucent substrate. Then, the laser beam hits the portion where the concave defect or the internal defect exists and deviates from the total reflection condition, thereby detecting the light leaking from the translucent substrate. This is to inspect the presence or absence of concave defects and internal defects. For detecting the leaked light, the main surface is photographed by a CCD, and the captured image is processed to detect a concave defect or an internal defect.

Japanese Patent Laid-Open No. 11-24001

  In recent years, there has been an increasing demand for further pattern miniaturization in patterning using a photolithography technique, and in response to this, the wavelength of exposure light used for exposure transfer has been shortened and the output has been increased. Specifically, a laser beam having a wavelength of 200 nm or less such as an ArF excimer laser having an exposure wavelength of 193 nm has been used. Along with this, higher accuracy is also required for defect detection of a light-transmitting substrate.

  On the other hand, in the method for inspecting a light-transmitting substrate disclosed in Patent Document 1, depending on the type and state of a defect or the like, it may be difficult to detect with high accuracy. For example, when light due to an internal defect leaks, the inspection light may be emitted from the main surface with the amount of light attenuated. As a result, the contrast in the image of the main surface photographed by the CCD may be lowered.

  Further, in order to improve the inspection throughput, it is desirable to widen the area of the main surface of the translucent substrate that is imaged at once with the CCD. However, the image captured by the CCD tends to become blurred as it moves away from the focus position (focal position) of the CCD lens system. For this reason, if the portion of the main surface where the laser beam leaks is at a position away from the focal point of the image taken by the CCD, the leaked light image becomes blurred and the contrast becomes low.

  Here, in the method for inspecting a light-transmitting substrate described above, the determination of the presence or absence of a defect is usually automatically performed by a computer algorithm (machine). Further, in image processing (image analysis) used in this inspection method, there is a case where the shape of what is shown in the image is recognized by performing difference processing (differential processing) between pixels in the image. In this image processing, since it is usually necessary to grasp the exact shape of the object shown in the image, priority is given to edge detection. For this reason, it is necessary to obtain a finer difference, and usually, a difference for obtaining a light amount difference from adjacent pixels is calculated.

  However, when such a difference process in conventional image analysis is performed, if the light amount difference between adjacent pixels is small, the light amount difference does not reach the detection threshold value, so that the device itself cannot be recognized. There is. That is, there is a possibility that a problem of overlooking the leaked laser beam may occur. In addition, if the threshold value itself is lowered, it may be determined that there is a defect in the automatic determination by the machine even though it is determined that the defect does not exist visually this time. .

  As a result of earnest research, the inventor of the present application more specifically, for example, the surface of the light-transmitting substrate is slightly swelled compared to the periphery, and the light-transmitting state is not clear. It has been found that in the case of a defect (moy) or the like where the surface of the conductive substrate is rough, the leaked laser light is likely to be overlooked in conventional image processing. Moreover, when the wavelength of exposure light shortened to 200 nm or less, it discovered that it was necessary to detect such a defect appropriately in the translucent board | substrate used as a mask blank board | substrate.

  Therefore, conventionally, there has been a demand for a method capable of appropriately detecting a defect or the like even when the contrast of the image is low or when a defect or the like that is overlooked by conventional image analysis exists. Then, an object of this invention is to provide the manufacturing method of the mask blank board | substrate which can solve said subject, the manufacturing method of a mask blank, the manufacturing method of the mask for transfer, and the manufacturing method of a semiconductor device.

In order to solve the above problems, the present invention has the following configuration.
(Configuration 1) A method of manufacturing a mask blank substrate using a light-transmitting substrate, the step of preparing a light-transmitting substrate having two opposing main surfaces, and laser light on at least one main surface An imaging process for photographing one main surface of the translucent substrate in the state of contact, and image processing is performed on the photographed image of the photographed main surface, and the laser beam is optically present in the translucent substrate. A detection process for detecting the presence or absence of light scattered from the area by hitting a non-uniform area, and a selection process for selecting a light-transmitting substrate that does not scatter light in the detection process as a mask blank substrate. The image processing includes processing for obtaining a light amount difference of the photographed image by remote difference.

  In this case, the gradation difference (light quantity difference) between the pixels (pixels) of the captured image is not calculated between adjacent pixels, but is calculated between two pixels separated by a predetermined number of pixels ( (Calculated by remote difference). Thereby, the gradation difference between two pixels is emphasized compared to the case of calculating between adjacent pixels (calculating with a proximity difference), and it becomes easy to detect scattered light.

  Therefore, if it does in this way, presence of an optically non-uniform | heterogenous area | region in a translucent board | substrate can be detected more appropriately, for example. More specifically, for example, even when the captured image has a low contrast, the presence of an optically non-uniform region on the translucent substrate can be appropriately detected. Further, it is possible to detect a defect such as a stain or a haze that reduces a light amount difference between adjacent pixels in a captured image. This also makes it possible to appropriately select the mask blank substrate in the selection step.

  In addition, the state which applied the laser beam to the main surface includes the case where it applies from the inside of a translucent substrate, and the case where it applies from the outside of a translucent substrate. The case where the light is applied from the inside of the translucent substrate means, for example, that the laser beam is totally reflected at the main surface and the end surface inside the translucent substrate, as in the inspection method of the translucent substrate disclosed in Patent Document 1. Is repeated. Moreover, the case where it irradiates from the outer side of a translucent board | substrate is a case where a main surface is directly irradiated with a laser beam from the outer side of a translucent board | substrate, for example.

  (Configuration 2) In the imaging process, laser light is introduced into the translucent substrate and propagated through the translucent substrate while repeating total reflection between the two main surfaces. One main surface is photographed, and the detection process performs image processing on the photographed image of the photographed main surface, so that the laser beam propagating in the translucent substrate exists in the translucent substrate. By hitting a uniform area, the presence or absence of light leaking from any of the main surfaces is detected.

  In this case, a laser beam that repeats total reflection between the two main surfaces is substantially only in the case where it deviates from the total reflection condition in an optically non-uniform region. It will leak from the main surface. Therefore, in this way, for example, by detecting light leaking from the translucent substrate, various defects and the like of the translucent substrate can be more appropriately detected with high accuracy.

  (Configuration 3) The translucent substrate has an end face at the outer edge of the opposing main surface, and has a chamfered surface at each corner where the main surface and the end face intersect. One of the chamfered surfaces. If it does in this way, a laser beam can be appropriately introduced in a translucent board | substrate, satisfy | filling the conditions which repeat total reflection between two main surfaces, for example.

  (Configuration 4) The surface for introducing the laser beam and the main surface for photographing the translucent substrate are polished to a mirror surface. If it does in this way, the conditions which repeat total reflection about the laser beam introduced in the translucent board | substrate can be satisfy | filled more appropriately, for example.

  (Configuration 5) In the imaging step, at least one main surface is irradiated with laser light from the outside of the translucent substrate.

  Even in this case, for example, by detecting the response to the laser light at each position on the main surface of the translucent substrate from the photographed image, it is possible to appropriately detect the presence of an optically nonuniform region on the translucent substrate. It can be detected. This also makes it possible to appropriately select the mask blank substrate in the selection step.

  (Configuration 6) Image processing is processing for creating a light amount distribution of a captured image by calculating a light amount difference between a processing target pixel of the captured image and a pixel that is not adjacent to the pixel. In this way, for example, image processing using remote difference can be appropriately performed.

  (Configuration 7) Image processing is performed by calculating a light amount difference between a processing target pixel and a pixel separated from the processing target pixel by the number of pixels corresponding to a distance of twice or more the size of the defect to be detected in the captured image. This is a process of creating a light amount distribution of the captured image by calculating. In this way, for example, a defect to be detected can be detected more appropriately by image processing using remote difference.

  The defect size is, for example, a range in which an optically non-uniform region existing in the translucent substrate spreads. As the size of the defect, for example, the length of the defect in the longitudinal direction can be used. Further, the size of the defect to be detected may be the maximum size in the range of the size of the defect to be detected in the detection process.

  (Configuration 8) A method of manufacturing a mask blank substrate using a light-transmitting substrate, the step of preparing a light-transmitting substrate having two opposing main surfaces, and laser light in the light-transmitting substrate An imaging process for imaging one main surface of the translucent substrate in a state where it is introduced and propagated through the translucent substrate while repeating total reflection between the two main surfaces, and imaging of the captured main surface Presence or absence of light leaking from one of the main surfaces when image processing is performed on the image and laser light propagating in the translucent substrate hits an optically non-uniform area in the translucent substrate And a selection step of selecting a translucent substrate that does not leak light as a mask blank substrate in the detection step, and image processing is processing for obtaining a light amount difference of a photographed image by remote difference including.

  In this case, the laser beam that repeats total reflection between the two main surfaces is substantially separated from the main surface only when it deviates from the total reflection condition in an optically non-uniform region. It will leak out. Therefore, if it does in this way, the various defects of a translucent board | substrate, etc. can be detected appropriately with high precision by detecting the light which leaks from a translucent board | substrate, for example.

  (Configuration 9) A mask blank manufacturing method, wherein a pattern forming thin film is formed on a main surface of a mask blank substrate manufactured by the mask blank substrate manufacturing method according to any one of Configurations 1 to 8. Process. In this way, a mask blank substrate selected with high accuracy can be used. Thereby, a highly accurate mask blank can be manufactured appropriately.

  (Configuration 10) A method for manufacturing a transfer mask, which includes a step of forming a transfer pattern on a pattern forming thin film of a mask blank manufactured by the method for manufacturing a mask blank described in Configuration 9. In this way, a mask blank manufactured with high accuracy can be used. This also makes it possible to appropriately manufacture a highly accurate transfer mask.

  (Configuration 11) Using the transfer mask described in Configuration 10, a circuit pattern is formed on a semiconductor wafer. In this way, a transfer mask manufactured with high accuracy can be used. Thereby, the circuit pattern can be formed on the semiconductor wafer with high accuracy.

  According to the present invention, for example, in the method for manufacturing a mask blank substrate, the presence of an optically non-uniform region in the translucent substrate can be detected more appropriately.

It is a flowchart which shows an example of the manufacturing method of the board | substrate for mask blanks concerning one Embodiment of this invention. It is a figure explaining in detail about inspection process S104. FIG. 2A shows an exemplary configuration of the inspection apparatus 100. FIG. 2B is a diagram illustrating an example of a state of laser light introduced into the translucent substrate 10. It is a figure which shows an example of the detailed structure of CCD102 which image | photographs the picked-up image used as the object of image processing. It is a figure explaining the image processing performed in the image processing apparatus. FIG. 4A is a diagram for explaining image processing performed by a method different from that of the image processing apparatus 106 of this example as a reference. FIG. 4B is a diagram for describing image processing performed in the image processing apparatus 106 of this example. It is a figure explaining the effect which performs image processing using remote difference. FIG. 3 is a diagram illustrating an example of a mask blank 20 and a transfer mask 30 that are manufactured using a translucent substrate 10. It is a figure explaining the 1st example of an experiment. It is a figure explaining the 2nd experiment example. It is a figure which shows the other example of the test | inspection performed by imaging | photography process S202 and detection process S204.

  Hereinafter, embodiments according to the present invention will be described with reference to the drawings. FIG. 1 is a flowchart showing an example of a method for manufacturing a mask blank substrate according to an embodiment of the present invention. In this example, the mask blank substrate manufacturing method is a method of manufacturing a mask blank substrate using a translucent substrate, and includes a substrate preparation step S102 and an inspection step S104.

The substrate preparation step S102 is a step of preparing a translucent substrate having two opposing main surfaces, and prepares a translucent substrate that has been processed to become a mask blank substrate. Substrate preparation step S102 may be a step of preparing a translucent substrate to be a mask blank substrate by the same or similar method as a known method. As the translucent substrate, for example, a translucent synthetic quartz glass substrate is preferably used. Moreover, in this example, the low thermal expansion glass used as a board | substrate of a reflective mask blank can also be used as a translucent board | substrate. Examples of the low thermal expansion glass include SiO ₂ —TiO ₂ glass.

  In this example, in the substrate preparation step S102, first, a synthetic quartz glass substrate having a predetermined shape and size is cut out from a translucent synthetic quartz glass ingot. The shape of the synthetic quartz glass substrate to be cut out is a shape having a rectangular upper and lower main surfaces and four side surfaces orthogonal to both main surfaces and connecting each side of both main surfaces.

  Then, chamfering is performed between both main surfaces of the synthetic quartz glass substrate and the side surfaces orthogonal to each other to form a chamfered surface. In the following description, in the synthetic quartz glass substrate after chamfering, a portion of the side surface that is not chamfered and is orthogonal to both main surfaces is referred to as an end surface.

  In this example, both the main surface, each end surface, and each chamfered surface of the synthetic quartz glass substrate are all polished into a mirror surface. By mirror polishing, for example, the surface roughness Ra (arithmetic average roughness) of both main surfaces of the synthetic quartz glass substrate is set to about 0.5 nm or less, and the surface roughness Ra (arithmetic average roughness) of each end face and each chamfered surface. Is about 2 nm or less. The surface roughness of the main surface is preferably 0.2 nm or less in terms of root mean square roughness (Rq). In the substrate preparation step S102, the main surface and the end surface may be further subjected to precision polishing or ultraprecision polishing. In addition, the dimension of the synthetic quartz glass substrate after this mirror polishing is performed is, for example, about 152.1 mm × about 152.1 mm × about 6.25 mm. As described above, the substrate preparation step S102 prepares a translucent substrate to be inspected in the next inspection step S104.

  Inspection process S104 is a process of inspecting the translucent substrate prepared in substrate preparation process S102. In this example, the inspection step S104 is a step of inspecting the presence or absence of an optically non-uniform region in the translucent substrate, and includes an imaging step S202, a detection step S204, and a selection step S206.

  The photographing step S202 is a step of photographing one main surface of the translucent substrate in a state where laser light is applied to at least one main surface. Further, in this example, the photographing step S202 is a step of photographing in a state where laser light is applied to the main surface from the inside of the translucent substrate, and the laser light is introduced into the translucent substrate, and two main steps are performed. One main surface of the translucent substrate is photographed in a state where it is propagated through the translucent substrate while repeating total reflection between the surfaces. More specifically, in the imaging step S202, for example, in the same manner as the inspection method of a light-transmitting substrate disclosed in Patent Document 1, the laser light is totally reflected on the main surface and the end surface inside the light-transmitting substrate. Repeat and shoot.

  The detection step S204 is a step of detecting an optically nonuniform region in the translucent substrate based on the photographed image photographed in the photographing step S202. In this example, the detection step S204 performs image processing on the photographed image of the main surface, and the laser beam propagating through the translucent substrate is within the translucent substrate (in addition to the inside of the translucent substrate, The presence or absence of light leaking from one of the main surfaces is detected by hitting an optically non-uniform region existing in the main surface).

  More specifically, the detection step S204 detects, for example, the presence or absence of light that scatters from an optically nonuniform region such as a scratch in the translucent substrate and leaks out of the translucent substrate as a result. To do. Further, in the detection step S204, the optical angle of the laser beam is refracted by striae or the like in the optically nonuniform region such as striae inside the substrate, and the incident angle that enters the boundary of the main surface is the total reflection condition. As a result, the presence or absence of light leaking out of the translucent substrate is detected.

  In this example, detection process S204 performs image processing including processing for obtaining a light amount difference of a photographed image by a remote difference for obtaining a light amount difference between non-adjacent pixels (pixels) as image processing for the photographed image. More specifically, this image processing is, for example, processing for creating a light amount distribution of a photographed image by calculating a light amount difference between a processing target pixel of the photographed image and a pixel not adjacent to the pixel. is there.

  In addition, detection process S204 may detect the presence or absence of the light of the same wavelength as the laser beam introduce | transduced in the translucent board | substrate, for example. In this way, for example, light leaking from the main surface of the translucent substrate due to the influence of an optically nonuniform region can be detected more appropriately.

  The selection step S206 is a step of selecting a translucent substrate that passes in the inspection step S104 as a mask blank substrate. In this example, in the selection step S206, based on the result of the detection step S204, a translucent substrate that does not leak light from the main surface due to scattering in an optically non-uniform region is selected as a mask blank substrate. .

  Here, in the photographing step S202, the laser light that repeats total reflection between the two main surfaces of the translucent substrate substantially deviates from the total reflection condition in an optically non-uniform region. Only if it leaks. Therefore, according to this example, by detecting light leaking from the translucent substrate in the detection step S204, various defects and the like of the translucent substrate can be appropriately detected with high accuracy. Accordingly, the mask blank substrate can be appropriately selected in the selection step S206. Therefore, according to this example, for example, a transparent substrate having a defect can be quickly and appropriately eliminated, and the productivity of the mask blank substrate can be improved.

  In the inspection step S104, various known inspections for selecting a mask blank substrate may be further performed. Moreover, you may perform reproduction | regeneration processes, such as re-mirror polishing, with respect to the translucent board | substrate which was not selected as a mask blank board | substrate in test | inspection process S104. In this case, it is preferable to perform the inspection step S104 again on the translucent substrate after the regeneration process.

  FIG. 2 is a diagram for explaining the inspection step S104 (see FIG. 1) in more detail. First, the configuration of the inspection apparatus 100 for performing the inspection step S104 will be described. FIG. 2A shows an exemplary configuration of the inspection apparatus 100. In this example, the inspection apparatus 100 is an apparatus for inspecting the non-uniformity of the translucent substrate 10 by introducing laser light into the translucent substrate 10, and includes a laser light source 110, a plurality of mirrors 112, a table 114, A driving unit 116, a lens system 108, a CCD 102, an A / D converter 104, and an image processing apparatus 106 are provided.

  The laser light source 110 is a light source that generates laser light to be introduced into the translucent substrate 10. As the laser light source 110, for example, a light source that generates laser light having a wavelength corresponding to the size of a defect to be detected by the inspection apparatus 100 is preferably used.

  For example, in a transfer mask manufactured from a mask blank substrate, when laser light having a wavelength of 200 nm or less, such as a 193 nm ArF excimer laser, is used as an exposure wavelength, a laser having a wavelength of 488 nm (blue) or a wavelength of 515 nm (green) A laser or the like can be suitably used. Further, the laser light source 110 may be a multi-line type laser light source that outputs both of these laser beams. If comprised in this way, a high output laser beam can be output appropriately. The beam diameter of the laser light is preferably about 0.7 mm (for example, 0.5 to 0.9 mm), and the beam spread angle is preferably about 1 mrad (for example, 0.5 to 1.5 mrad). Further, the wavelength of the laser light generated by the laser light source 110 may be the same as the exposure wavelength used in the transfer mask manufactured from the mask blank substrate.

  The plurality of mirrors 112 are mirrors that reflect the laser light from the laser light source 110 and introduce it into the translucent substrate 10. In this example, the plurality of mirrors 112 are provided so that the angle of the mirror 112 can be adjusted with respect to the laser light source 110 and the translucent substrate 10, and the angle of the laser light incident on the translucent substrate 10 can be changed. It adjusts so that the conditions which repeat a total reflection between 10 two main surfaces may be satisfy | filled.

  The table 114 is a table that holds the laser light source 110 and the plurality of mirrors 112. Further, the driving means 116 moves the table 114 relative to the translucent substrate 10.

  In this example, the translucent substrate 10 is fixed and held at a predetermined position by a known substrate holder, for example. The driving means 116 moves the table 114 with respect to the fixed laser light source 110 to change the position where the laser light is incident on the translucent substrate 10. As a result, the driving means 116 translates the laser light in the direction of one side of the translucent substrate 10 and sequentially changes the position where the laser light passes through the translucent substrate 10.

  The lens system 108 is a lens for imaging the main surface of the translucent substrate 10. In this example, the lens system 108 is an imaging optical system that forms an image of the leaked light from the translucent substrate 10 on the CCD 102, and in a state where the lens system 108 is focused on one main surface of the translucent substrate 10. It is provided between the optical substrate 10 and the CCD 102. The CCD 102 is an image sensor for photographing one main surface of the translucent substrate 10 in the photographing step S202 (see FIG. 1).

  The CCD (Charge Coupled Device) is an example of an image sensor that can photograph the main surface of the translucent substrate 10. As this imaging device, various sensors having a function as a two-dimensional area sensor can be used instead of the CCD. For example, a CMOS image sensor or the like may be used instead of the CCD. It is also conceivable to use a photomultiplier or the like.

  The A / D converter 104 is a converter that converts an analog signal detected by the CCD 102 into a digital signal. In this example, the A / D converter 104 transmits the converted digital signal to the image processing apparatus 106.

  The image processing apparatus 106 is an apparatus that performs image processing on a captured image captured by the CCD 102. In this example, the image processing apparatus 106 is, for example, a computer, and performs image processing in the detection step S204 (see FIG. 1) based on a signal received from the CCD 102 via the A / D converter 104. The image processing apparatus 106 further selects a mask blank substrate in the selection step S206 (see FIG. 1) based on the result of the image processing. The image processing performed in the image processing apparatus 106 will be described in more detail later.

  Subsequently, the state of the laser light introduced into the translucent substrate 10 will be described in more detail. FIG. 2B is a diagram illustrating an example of a state of laser light introduced into the translucent substrate 10.

  As described with reference to FIG. 1, in this example, translucent substrate 10 has an end surface 304 (T surface) at the outer edge of opposing main surface 302, and main surface 302 and end surface 304 intersect. Each corner is provided with a chamfered surface 306 (C surface). In this example, the chamfered surface 306 serves as a laser light introduction surface to the translucent substrate 10.

  In addition, the inspection apparatus 100 introduces laser light from any one of the chamfered surfaces 306 by adjusting the incident angle and position of the laser light with the plurality of mirrors 112. Thereby, for example, the laser beam can be more appropriately introduced into the translucent substrate 10 while satisfying the condition for repeating the total reflection between the two main surfaces 302.

  Further, the main surface 302, the end surface 304, and the chamfered surface 306 are polished to a mirror surface. Therefore, according to this example, for example, total reflection between the two main surfaces 302 can be appropriately repeated with respect to the laser light introduced into the translucent substrate 10.

  When the introduction surface (the chamfered surface 306) is mirror-polished as in this example, the introduced laser light is not diffused by the introduction surface, and the light is propagated as parallel light. Therefore, in the total reflection state described below, all of the light incident on the main surface 302 which is a total reflection surface and the end surface 304 which is a folded surface is totally reflected. As a result, the response of the laser beam in the optical non-uniform portion and the uniform portion in the translucent substrate 10 becomes more critical, and the contrast is improved.

  On the other hand, when the introduction surface is not polished, light diffuses on the introduction surface and light in multiple directions propagates. In this case, since the respective ray trajectories cannot be predicted, it becomes difficult to satisfy the condition of total reflection. However, with regard to the laser light to be introduced into the translucent substrate 10, it is also conceivable that the chamfered surface 306 is not mirror-polished as long as the laser light can be introduced so that all of the laser light is totally reflected at the main surface 302 that strikes first. It is done. For example, by applying a matching oil having the same refractive index as that of the translucent substrate 10 in order to form a pseudo mirror surface on the rough chamfered surface 306 to be the introduction surface, the condition of total reflection can be set. The laser beam can be introduced by filling.

  Here, in this example, the laser light incident on the translucent substrate 10 repeats total reflection on the main surface 302 and the end surface 304 of the translucent substrate 10 and seems to be almost confined in the translucent substrate 10. It becomes a state. The almost confined state means that the introduced laser light propagates through the translucent substrate 10 and is incident on the chamfered surface 306 that is the introduction surface, that is, incident on the main surface 302 and the end surface 304. As far as it is concerned, it means that the total reflection is continued and propagated through the translucent substrate 10.

  Therefore, for example, as shown in FIG. 2B, the thickness direction of the translucent substrate 10 is in the Z direction, and the end surface 304 sandwiching the chamfered surface 306, which is the laser light introduction surface, between the main surface 302 is orthogonal. When the direction is the Y direction, and the direction orthogonal to the Z direction and the X direction is the X direction, the laser light incident so as to travel in the Y direction by repeating reflection between the main surfaces 302 is transmitted through the translucent substrate 10. Scanning is performed indefinitely by propagation due to the total reflection of the light itself so as to reach an area (inspected area) having one cross section (YZ cross section) when cut in the Y direction. As a result, the laser light introduced into the translucent substrate 10 is in a state where it is almost confined within the translucent substrate 10.

  That is, if there are no non-uniform portions such as scratches in the region to be inspected of the translucent substrate 10, the laser light introduced into the translucent substrate 10 repeats total reflection on the surface and is transmitted into the translucent substrate 10. Becomes confined (laser light spreads). Therefore, in this case, the laser light does not substantially leak to the outside at the main surface 302 and the end surface 304.

  On the other hand, if there is a scratch or the like on the surface of the translucent substrate 10 due to, for example, contamination of foreign matter during polishing, the total reflection condition is not satisfied at that position, and light leaks from the portion corresponding to the scratch. . Further, even in the presence of a defect having the same transmittance but different refractive index, which is characteristic of the striae of glass, the original orbit (light path) deviates from the place where the refractive index is different. It will leak out. That is, if there is an optically non-uniform portion in the translucent substrate 10, if it is originally uniform, the optical path (path) through which it passes is deviated, and the total reflection condition is not satisfied, and light is transmitted from the surface of the translucent substrate 10. Leaks out.

  Therefore, by detecting the presence or absence of light leaking from the translucent substrate 10 with the above-described configuration, first, the translucent region of the one-line inspection region as viewed from the main surface 302 side of the translucent substrate 10 is firstly displayed. An optically non-uniform area in the substrate 10 can be detected appropriately.

  In this example, the table 114 is moved by the driving means 116, so that the laser light incident on the translucent substrate 10 is translated in the X direction. Thus, the one-line inspection region is sequentially shifted in the X direction, and an optically non-uniform region in the inspection region is detected. Therefore, according to this example, the optical non-uniformity inspection can be appropriately performed on the entire translucent substrate 10.

In addition, about conditions, such as an angle which introduce | transduces a laser beam into the translucent board | substrate 10, the conditions similar to a well-known method which are disclosed by patent document 1, for example can be used. More specifically, for example, a refractive index n _i of the outer medium in contact with the refractive index of the translucent substrate 10 with respect to the wavelength λ of introducing laser light n _t, the translucent substrate 10 (the atmosphere), translucent when the angle of the light incident on the major surface 302 initially strikes since the laser beam is incident on the sex substrate 10 and theta _i, theta _i is the main surface 302, represented by sinθ _c = _n i / _{n t} becomes critical angle theta _c or more and, at the end face 304 so that the (90 ° -θ _i) is the critical angle theta above, may be introduced laser beam from the introducing surface.

  With this configuration, the singular point where the laser light leaks geometrically does not substantially exist in the main surface 302 and the end surface 304, and light can be substantially confined in the translucent substrate 10. . In addition, this makes it possible to appropriately introduce the laser light so that the laser light introduced only from the chamfered surface 306 is emitted.

  Here, the reason why it is geometrically optical is that light caused by an optical change due to the inherent property of the material of the translucent substrate 10, for example, Rayleigh scattered light or the like is not considered here. Also, the singular point is substantially absent because of the influence of the spread angle of the laser beam itself, after a total reflection is repeated many times on the main surface 302 and the end surface 304, a very small amount of laser beam is totally reflected. This is because it is considered that the condition is not satisfied, and the leakage occurs out of the translucent substrate 10. However, the leaked light due to the spread angle of the laser light itself leaks in the direction along the surface of the translucent substrate 10, so that it is not detected by the CCD 102 and does not affect the detection sensitivity.

Further, in this example, if you want to see the incident angle theta _i of the laser beam with respect to the translucent substrate 10, for example, a wedge-shaped transparent member, such as an optical member 120 shown by a broken line in FIG. 2 (b) In addition, it may be installed on the main surface of the translucent substrate 10 via matching oil or the like. In this way, the incident angle θ _i can be obtained from the refraction angle γ of the light emitted from the optical member 120, the apex angle of the optical member 120, and the like. A member such as the optical member 120 can be used as an incident window for introducing laser light into the translucent substrate 10, for example. In this way, laser light can be introduced from other than the chamfered surface 306 of the translucent substrate 10.

  In addition, the inspection method of this example is an object to be inspected because light is substantially confined in the translucent substrate 10 using geometric optical total reflection, which is a physical critical phenomenon. The response to the laser light in the non-uniform portion and the uniform portion in the translucent substrate 10 is also critical, and the non-uniform portion appears with dramatic contrast as leakage light in the background of the dark uniform portion. As a result, defects such as extremely fine scratches in the translucent substrate 10 are observed as if the light inside the container leaks out from the pinholes present in the light-shielding container such as a dark box. . Therefore, according to this example, an optically non-uniform region in the translucent substrate 10 can be appropriately detected with high accuracy.

  Further, in this example, the laser light passes through an optically non-uniform region in the translucent substrate 10 having a refractive index higher than that in the air. Therefore, according to this example, light is easily scattered by the short wavelength effect. Therefore, according to this example, even when the size of the optically non-uniform region is a fine size that cannot be detected by the wavelength of the laser light in the air, it can be appropriately detected with high accuracy.

  As described above, in this example, a defect or the like is detected for the translucent substrate 10 whose entire surface is mirror-finished. However, for example, when inspecting the translucent substrate 10 at a stage where the mirror finish is not applied, the same method as in this example is applied to the translucent substrate 10 whose part or the entire surface is not mirror finished. Defects and the like may be detected with In this case, it is preferable to perform inspection by applying a liquid such as matching oil onto a surface that is not mirror-finished so that the surface is a mirror-finished surface (liquid free surface, pseudo-mirror surface).

  In this example, laser light is introduced from one side of the translucent substrate 10 as described above. However, the method of introducing laser light is not limited to this, and, for example, it is conceivable to introduce laser light in two directions from two adjacent sides. In this way, for example, a directional defect can be detected with higher accuracy. That is, when the main surface is a rectangular translucent substrate, the angle formed between the traveling directions of the two laser beams is 90 degrees.

  3 and 4 are diagrams for explaining the image processing performed in the image processing apparatus 106 in more detail. FIG. 3 shows an example of a detailed configuration of the CCD 102 that captures a captured image to be subjected to image processing. The CCD 102 is an image sensor in which a plurality of pixels 202 are arranged in a two-dimensional array. Each pixel 202 in the CCD 102 corresponds to each pixel (pixel) of the captured image photographed in the photographing step S202.

  Here, the operation of the CCD 102 in the photographing step S202 will be described. In the photographing step S <b> 202, the light leaked from the translucent substrate 10 is imaged on the CCD surface of the CCD 102 by the lens system 108. In the photographing step S202, as described above, the one-line inspection region is scanned over the entire main surface of the translucent substrate 10. In this example, the shutter of the CCD 102 is kept open during this time. Further, the CCD 102 sends the acquired data to the image processing device 106 via the A / D converter 104.

  For example, the image processing apparatus 106 associates the amount of movement of the table 114 during scanning of the inspection area with data received from the CCD 102, for example, and thereby each position of the main surface of the translucent substrate 10 and each pixel of the CCD 102. 202. Accordingly, the image processing apparatus 106 acquires a captured image obtained by capturing the entire main surface of the translucent substrate 10.

  Note that since the number of pixels 202 in the CCD 102 is limited, when shooting a captured image, a method of capturing the entire main surface of the translucent substrate 10 at a low magnification at once, or dividing the main surface into a plurality of regions. Therefore, it is necessary to use one of the methods for photographing each area at a high magnification. In this example, any method may be used depending on the size of the defect to be detected, the performance of the lens system 108, and the like.

  FIG. 4 is a diagram for describing image processing performed in the image processing apparatus 106. In this example, the image processing apparatus 106 detects a pixel with a large amount of light from a captured image acquired by the CCD 102 in order to detect light leaking from the main surface of the translucent substrate 10. For this purpose, a light amount difference from surrounding pixels is calculated for each pixel of the captured image.

  FIG. 4A is a diagram for explaining image processing performed by a method different from that of the image processing apparatus 106 of this example as a reference. Image processing in the case of using a proximity difference for calculating a light amount difference between adjacent pixels is illustrated. Show. In this method using the proximity difference, a light amount difference between the processing target pixel 204 and an adjacent pixel is calculated as a gradation difference corresponding to the processing target pixel 204 in the captured image.

  More specifically, for example, in this method, first, the processing target pixel 204 is sequentially selected from the pixels 204 constituting the captured image. In addition, in order to perform a three-direction difference process (differential process) that is a non-linear filter, three pixels 204 that are adjacent to the pixel 204 to be processed in each of predetermined vertical, horizontal, and diagonal directions are used as the other pixel for calculating the difference. Select further.

  FIG. 4A shows a case where the light quantity of the pixel 204 to be processed is X00, and the light quantities of the adjacent pixels 204 in the three vertical, horizontal, and diagonal directions are X10, X01, and X11, respectively. In this case, the light amount difference between the processing target pixel 204 and each adjacent pixel 204 is X00-X10, X00-X01, and X00-X11, respectively. Of these, the one having the largest absolute value is set as the gradation difference corresponding to the pixel 204 to be processed.

  In this way, for each pixel 204, as a representative value indicating a gradation difference, a difference value in the direction in which the change in the amount of light is the largest among the three directions can be appropriately selected. This also makes it possible to appropriately detect a change in the amount of light even when, for example, a defect whose appearance differs depending on the direction exists.

  FIG. 4B is a diagram for describing image processing performed in the image processing apparatus 106 of this example. In this example, the image processing apparatus 106 performs a process for obtaining a gradation difference corresponding to each pixel 204 of the captured image by a remote difference for obtaining a light amount difference between two pixels 204 that are separated by a predetermined number of pixels. Further, this remote difference processing is performed in three directions, ie, vertical, horizontal, and diagonal, as in the method described with reference to FIG. More specifically, for example, first, the processing target pixel 204 is sequentially selected from the pixels 204 constituting the captured image. Further, three pixels 204 that are separated from the pixel 204 to be processed by a predetermined number of pixels in each of vertical, horizontal, and diagonal predetermined directions are further selected as counterpart pixels for calculating the difference.

  FIG. 4B shows that when obtaining a remote difference with a pixel that is n pixels away from the processing target pixel 204, the light amount of the processing target pixel 204 is X00, and the light amount of the pixel that is n pixels away in the vertical direction is Xn0. The figure shows a case where the light amount of a pixel separated by n pixels in the horizontal direction is X0n, and the light amount of a pixel separated by n pixels in the oblique direction is Xnn. In this case, the light amount difference between the processing target pixel 204 and each of the counterpart pixels 204 whose differences are to be calculated is X00-Xn0, X00-X0n, and X00-Xnn, respectively. Of these, the one having the largest absolute value is set as the gradation difference corresponding to the pixel 204 to be processed.

  Here, the process of detecting the light leaking from the translucent substrate 10 based on the photographed image is performed, for example, when the gradation difference calculated by the difference process is a predetermined value regardless of whether the proximity difference or the remote difference is performed. This is done by discriminating pixels 204 that are above the slice level (threshold). Further, for example, by discriminating the pixels 204 whose gradation difference is equal to or higher than the slice level, an edge such as a defect is obtained, and the presence or absence of an optically nonuniform region such as a defect is identified.

  However, for example, when the contrast of the photographed image is low, or when there is a defect or the like that reduces the light amount difference between adjacent pixels 204 in the photographed image, such as a spot or a haze, the slice level is determined by the gradation difference obtained by the proximity difference. This is not the case and an edge such as a defect may not be recognized. As a result, it may be impossible to identify the presence or absence of an optically nonuniform region such as a defect.

  On the other hand, when the light amount difference is obtained by the remote difference as in this example, the gradation difference corresponding to the pixel 204 is emphasized compared to the case of performing the proximity difference, and is in-plane with respect to the main surface of the translucent substrate 10. In the light amount distribution, it becomes easy to detect the peak of the gradation difference. As a result, even when the contrast of the captured image is low or there is a defect or the like in which the light amount difference between adjacent pixels 204 is small, the defect or the like can be detected appropriately. Therefore, according to this example, the mask blank substrate can be selected more appropriately in the selection step S206.

  Here, in the remote difference processing, the number of pixels separating the processing target pixel 204 for obtaining the gradation difference and the partner pixel 204 for calculating the difference (hereinafter referred to as the number of separated pixels) is a defect to be detected. It is preferable to set according to the size. For example, the number of separated pixels is preferably equal to or greater than the number of pixels corresponding to a distance that is twice or more the size of a defect or the like to be detected. If comprised in this way, the defect etc. which become the object of detection can be detected appropriately.

  In the image processing performed in the image processing apparatus 106, for example, a plurality of types of remote differences in which the number of separated pixels is different may be performed for each pixel 204 to be processed. More specifically, for example, for one pixel 204 to be processed, a first remote difference process in which the number of separated pixels is the first number N1, and a second number N2 in which the number of separated pixels is different from N1 It is conceivable to perform the second remote difference processing. In this case, the result of obtaining a larger gradation difference may be the gradation difference corresponding to the pixel 204. It is also conceivable to determine the type of defect or the like by comparing the results of a plurality of remote differences. Further, in addition to image processing by remote difference, image processing by proximity difference may be further performed.

  In this example, the number of separated pixels is set to be the same in the vertical, horizontal, and diagonal directions. However, for example, when the number of separated pixels is large, the difference between the actual distance at which the two pixels are separated in the vertical and horizontal directions and the actual distance at which the two pixels are separated in the oblique direction is large. Therefore, the separation distance in the oblique direction may be set smaller than the separation distance in the vertical and horizontal directions.

  FIG. 5 is a diagram illustrating the effect of performing image processing using remote difference. In FIG. 5, the defect form, the light amount, the proximity difference result, and the remote difference result are shown with feature emphasis and simplification for the convenience of illustration.

  First, a case where image processing using remote difference is applied to various defect forms will be described. Examples of the defect that becomes an optically non-uniform region in the translucent substrate 10 include a scratch, a stain, and a haze. The scratch is, for example, a groove-like or hole-like concave defect on the surface of the translucent substrate 10. For example, the stain is a defect that slightly rises on the surface of the translucent substrate 10 as compared with the periphery. Moreover, a haze is a defect which the surface of the translucent board | substrate 10 is rough in the state where a boundary is not clear, for example. In FIG. 5, the column of the form shows an outline of the cross-sectional shape of the defect of each form.

  Further, in FIG. 5, the light quantity column indicates the light quantity that leaks from the translucent substrate 10 in response to defects of each form when laser light is introduced into the translucent substrate 10 by the inspection apparatus 100 of this example. The distribution of. In this example, the light amount distribution is a light amount distribution obtained from a captured image. As shown in the figure, when the defect is a flaw, strong light leaks out corresponding to the position of the flaw. Therefore, the peak of the light amount distribution becomes high. However, in the case of stains and haze, only weak light leaks out compared to scratches. As a result, the light quantity distribution also has a low peak and a small change in the light quantity.

  In FIG. 5, the proximity difference column indicates a distribution of gradation differences calculated by image processing using the proximity difference described with reference to FIG. The detection column below indicates whether or not each type of defect can be detected by this image processing.

  As shown in the figure, when the defect is flawed, the peak of the light amount distribution is clear, and therefore a clear and high peak exists in the gradation difference distribution due to the proximity difference. For this reason, it is easy to distinguish from noise that gradually changes the amount of light due to illumination unevenness, imaging unevenness, etc., and by detecting pixels with a gradation difference equal to or higher than a predetermined slice level, defects can be detected appropriately. Can be detected.

  However, in the case of a stain or a haze, since the peak of the light amount distribution is low and the change in the light amount is small, there is no clear peak in the gradation difference distribution. For this reason, it is difficult to distinguish from noise, and if an appropriate slice level that does not pick up noise is set, there is a high possibility that a defect cannot be detected.

  On the other hand, in FIG. 5, the remote difference column indicates a distribution of gradation differences calculated by the image processing using the remote difference described with reference to FIG. The detection column below indicates whether or not each type of defect can be detected by this image processing.

  As shown in the figure, when the defect is flawed, the peak of the light amount distribution is clear, and therefore, a clear and high peak exists in the distribution of gradation difference due to the remote difference. Therefore, it is easy to distinguish from noise, and a defect can be detected appropriately by detecting a pixel whose gradation difference is equal to or higher than a slice level.

  Further, when using the remote difference, as described above, the gradation difference at the defect position is emphasized as compared with the case of using the proximity difference. Therefore, even in the case of a stain or a haze, a certain clear peak occurs in the distribution of gradation differences. As a result, it is possible to distinguish from noise, and it is possible to detect a defect appropriately by detecting a pixel having a gradation difference equal to or higher than a predetermined slice level.

  As described above, when image processing using the proximity difference is performed, a stain or a haze may not be detected appropriately. On the other hand, even a defect such as a stain or a haze that cannot be properly detected by the proximity difference can be appropriately detected by performing image processing using the remote difference as in this example. Therefore, according to this example, a defect etc. can be detected appropriately with higher accuracy.

  In the above description, only the defects on the substrate surface have been described for convenience of description. However, the point that the gradation difference at the position of the defect or the like is emphasized by the remote difference is also the same for the optically nonuniform area such as striae inside the substrate. Therefore, according to this example, it is possible to appropriately detect defects and the like inside the substrate with higher accuracy.

  Next, a case where image processing using remote difference is applied to a captured image with low contrast will be described. When shooting a shot image, if the lens system 108 is out of focus and out of spec of the subject depth, the shot image is in a so-called out-of-focus state. As a result, the contrast is reduced in the captured image.

  Further, when the captured image is captured, the entire main surface of the translucent substrate 10 or a certain constant area is photographed with the lens system 108 focused at a certain position. For this reason, even when the focus is in the center of the captured image, the focus may be shifted and the contrast may be low at a position away from the center in the captured image. Furthermore, due to the limit of the performance of the lens system 108 itself, there may be a case where the light cannot be completely collected due to the aberration of the optical system even in a focused state. As a result, the contrast of the captured image may be lowered.

  In many cases, the focus of the lens system 108 in the depth direction of the translucent substrate 10 at the time of capturing a captured image is adjusted to the vicinity of the main surface on the side where the lens system 108 of the translucent substrate 10 is disposed. In this case, if there is a defect or the like on the main surface on the opposite side, the inspection light leaks from the main surface side, so that the focus of the lens system 108 is shifted, resulting in a defocused state. In addition, there is an optically non-uniform region inside the translucent substrate 10, and the inspection light is scattered in the region, and light leaks from the main surface opposite to the side where the lens system 108 is disposed. In this case, the focus of the lens system 108 is similarly shifted, resulting in a defocused state.

  Then, when the contrast of the photographed image becomes low due to these causes or the like, as shown in the figure, when the light amount distribution leaking from the translucent substrate 10 is obtained from the photographed image, the peak is low and the change in the light amount is small. Distribution. As a result, there is no clear peak in the gradation difference distribution calculated by the image processing using the proximity difference. Therefore, when image processing using proximity differences is performed, the possibility that a defect cannot be detected increases because the contrast of the captured image is low.

  On the other hand, in the distribution of gradation differences calculated by image processing using remote differences, a certain clear peak occurs as a result of enhancement of gradation differences at positions such as defects. Therefore, it is possible to detect a defect even when the contrast of the captured image is low.

  As described above, according to this example, even a defect or the like that cannot be properly detected by the image processing using the proximity difference can be appropriately detected by using the image processing using the remote difference. Therefore, according to this example, the translucent substrate 10 can be appropriately inspected with higher accuracy. Thereby, the mask blank substrate can be appropriately manufactured with high accuracy.

  In detecting a defect or the like, it is conceivable to not only detect the presence / absence of a defect but also to further determine the position, type, size, and the like of the defect. In this case, for example, the image processing apparatus 106 compares the light amount distribution obtained from the captured image and the gradation difference distribution obtained by the remote difference with the discrimination information stored in advance. Thus, the position, type, size and the like of the defect are further determined. If it does in this way, a substrate for mask blanks can be selected more appropriately in selection process S206, for example. Furthermore, for example, it is possible to determine whether or not the translucent substrate 10 that has not been selected as an acceptable product in the selection step S206 can be reused by re-polishing, the re-polishing method, and the like.

  FIG. 6 shows an example of a mask blank 20 and a transfer mask 30 that are manufactured using the translucent substrate 10. FIG. 6A shows an example of the configuration of the mask blank 20. The translucent substrate 10 that has been accepted in the selection step S206 and selected as the mask blank substrate is then used for manufacturing the mask blank 20. In the manufacturing process of the mask blank 20, the pattern forming thin film 12 is formed on the main surface of the translucent substrate 10 selected as the mask blank substrate by, for example, a known method. In this way, a mask blank substrate selected with high accuracy can be used. Moreover, thereby, the highly accurate mask blank 20 can be manufactured appropriately.

  The pattern forming thin film 12 may have any structure of a single layer structure, a multi-layer laminated structure, or a composition-graded structure. The mask blank here includes a configuration in which a hard mask film used as an etching mask when the pattern forming thin film 12 is patterned is formed on the pattern forming thin film 12. The mask blank here includes a configuration in which a resist film made of an organic material is formed on the pattern forming thin film 12 or the hard mask film. The mask blank 20 manufactured in this way is used for manufacturing the transfer mask 30.

  FIG. 6B shows an example of the configuration of the transfer mask 30. In the manufacturing process of the transfer mask 30, the pattern forming thin film 12 of the mask blank 20 is patterned by etching, for example, by a known method to form a transfer pattern. In this way, by using the mask blank 20 manufactured with high accuracy, the transfer mask 30 with high accuracy can be appropriately manufactured.

  The mask blank 20 referred to here is provided with a multilayer reflective film (for example, an alternate laminated film of Si film and Mo film) that reflects exposure light between the translucent substrate 10 and the pattern forming thin film 12. Configuration is also included. Examples of such a mask blank include a reflective mask blank in which EUV (Extreme UltraViolet) light (wavelength 13 to 14 nm or the like) is applied to exposure light. In the case of this reflective mask blank, when a multilayer reflective film is formed on a translucent substrate having defects such as scratches on the main surface, the size and depth are amplified starting from the defects such as scratches. As a result, a fatal phase defect may occur in the multilayer reflective film. In addition, when a reflective mask (transfer mask) manufactured from a reflective mask blank is set on a mask stage of an exposure apparatus, it is different from a side on which a multilayer reflective film or a pattern forming thin film is formed. The opposite side (back side) is chucked. For this reason, the presence of defects such as scratches on the main surface on the back surface side of the translucent substrate may reduce the accuracy of the chuck. For these reasons, it is important that defects such as scratches can be accurately detected even in a light-transmitting substrate used for a reflective mask blank.

  In the manufacturing process of the reflective mask, the transfer pattern is formed by patterning the pattern forming thin film by etching using, for example, a known method using the reflective mask blank. In this way, a reflective mask (transfer mask) with high accuracy can be appropriately manufactured by using the reflective mask blank 20 manufactured with high accuracy.

  These manufactured transfer masks 30 are used for manufacturing semiconductor devices. In the manufacturing process of a semiconductor device, a circuit pattern is formed on a semiconductor wafer using a transfer mask 30 by a known method, for example. In this way, the circuit pattern can be formed on the semiconductor wafer with high accuracy by using the transfer mask 30 manufactured with high accuracy. This also makes it possible to appropriately manufacture a high-quality semiconductor device free from malfunction defects.

  Next, the effect of performing image processing using remote difference will be described using a specific experimental example. FIG. 7 is a diagram for explaining the first experimental example. In the first experimental example, first, a mask blank substrate on which standard particles having a size of 90 nm were attached was prepared. Then, a photographed image (raw image) obtained by photographing the mask blank substrate was photographed by an inspection apparatus similar to the inspection apparatus 100 described with reference to FIG. Further, in order to make it easier to confirm the effect of performing image processing using remote difference, an image taken with the focus slightly shifted was prepared as a taken image. The raw image uses a known inspection apparatus similar to the inspection apparatus 100 of FIG. 2 and a known mask blank substrate on which standard particles as described above are attached, except for the image processing apparatus 106. By this, photographing is possible.

  Then, for the same raw image, the remote distance is set to 2, 4, 10, 12, and 20 so that the remote distances are different, and the remote difference is used by the method described with reference to FIG. 4B. Image processing was performed to obtain a distribution of gradation differences corresponding to the number of separated pixels. Further, each processed image was created using the calculated gradation difference of each pixel as a gradation value. For comparison, the same raw image was subjected to image processing using the proximity difference by the method described with reference to FIG. 4A, and the gradation difference distribution in that case was obtained. Further, similarly to the case of remote difference, a processed image was created using the calculated gradation difference of each pixel as a gradation value. In this experimental example, the level of the gradation value when the luminance (differential luminance) is the highest is 4095, and the level of the gradation value when the luminance is the lowest is 0.

  For each gradation difference distribution, the value of the gradation difference at the peak of the distribution was used as the detection level. As shown in the graph in FIG. 7, it can be seen that as the number of separated pixels increases and the remote distance increases, the detection level increases and the defect portion enhancement effect increases. Therefore, it can be confirmed from this experimental example that defects can be easily detected by image processing using remote differentiation.

  Here, in the experimental environment of this experimental example, even when the mask blank substrate has no defect, the background noise level of the photographed image is about 100. Therefore, for example, a practical slice level for distinguishing noise from a signal can be set by giving a further 20% margin and setting the level at which the gradation value becomes 120 as the slice level. In this experimental example, the detection level exceeds the practical slice level by setting the number of separated pixels to 12 or more.

  FIG. 8 is a diagram illustrating a second experimental example. In the second experimental example, first, a mask blank substrate on which standard particles having a size of 120 nm were attached was prepared. Then, as an image corresponding to a photographed image obtained by photographing a defect or the like, a photographed image obtained by photographing the mask blank substrate while varying the focus position by an inspection apparatus similar to the inspection apparatus 100 described with reference to FIG. Six (raw images) were taken.

  In the graph of FIG. 8, the horizontal axis indicates the focus relative position. The focus relative position indicates the focus position when the photographing apparatus (CCD camera) is fixed at a predetermined position, the focus position is closest to the mask blank substrate, and the focus position in other cases is relatively indicated. It is. As shown in the drawing, the six raw images are images taken at focus relative positions at intervals of 20 μm.

  Then, the image processing using the remote difference was performed on each of the six raw images by the method described with reference to FIG. 4B to obtain a gradation difference distribution corresponding to each focus position. Further, each processed image was created using the calculated gradation difference of each pixel as a gradation value. In this experiment, the number of separated pixels was 12. For comparison, image processing using proximity differences is performed on each of the six raw images using the method described with reference to FIG. 4A to obtain a gradation difference distribution corresponding to each focus. It was. Further, similarly to the case of remote difference, a processed image was created using the calculated gradation difference of each pixel as a gradation value.

  Then, for each obtained gradation value distribution, a detection level that is a peak value of the distribution was obtained. In the graph of FIG. 8, circles indicate the detection levels corresponding to the respective focus positions when using the remote difference. The cross mark indicates the detection level corresponding to each focus position when the proximity difference is used. As shown in the graph, it can be confirmed that the detection level is increased by using the remote difference at all focus positions.

  In the case of this experimental example, it can be estimated that the detection level is about 600 when the focus position is the optimum place (best focus position). In this case, for example, a case where the detection level exceeds 300, which is half of the detection level, can be said to be a stable detection level. Therefore, in the case of this experimental example, the range of the focus position where the detection level exceeds 300 can be defined as the focus depth.

  Then, when the focus depth is calculated according to this definition based on the graph shown in FIG. 8, the focus depth when the proximity difference is used is ± 25 μm with respect to the estimated best focus position. Therefore, when the proximity difference is used, it is necessary to set the focus position at least within a range of ± 25 μm from the best focus position in order to stably detect standard particles having a size of 120 nm.

  On the other hand, when using the remote difference, the focus depth is ± 35 μm with respect to the estimated best focus position. Therefore, when using a remote difference, in order to stably detect a standard particle having a size of 120 nm, the focus position may be set within a range of ± 35 μm from the best focus position. Therefore, it can be confirmed from this experimental example that the use of remote difference increases the depth of focus and improves the detection power.

  FIG. 9 is a diagram illustrating another example of the inspection performed in the imaging step S202 and the detection step S204, and illustrates an example of the configuration of the inspection apparatus 100 used in this inspection. Except as described below, in FIG. 9, the configuration denoted by the same reference numeral as in FIG. 2 has the same or similar function as the configuration in FIG.

  The inspection apparatus 100 of this example irradiates one main surface of the translucent substrate 10 with laser light from the outside of the translucent substrate 10 by the laser light source 110 in the imaging step S202. Further, the CCD 102 captures a captured image of the main surface irradiated with laser light from the outside of the translucent substrate 10.

  In the detection step S204, the image processing device 106 performs image processing using the remote difference on the captured image in the same manner as the image processing device 106 described with reference to FIG. Accordingly, the image processing apparatus 106 detects the presence / absence of light scattered from an optically non-uniform region present in the translucent substrate 10.

  Also in this example, by detecting the response to the laser light at each position on the main surface of the translucent substrate 10 by the captured image, it is possible to appropriately detect the presence of an optically non-uniform region in the translucent substrate 10. . Accordingly, the mask blank substrate can be appropriately selected in the selection step S206.

  Note that, for example, when the laser beam is irradiated from the outside to the main surface of the translucent substrate 10 as in this example, compared with the case where the laser beam repeats total reflection inside the translucent substrate 10, it is optically inferior. It is also conceivable that the difference in optical response between the uniform area and the other normal areas is reduced. As a result, even if there is an optically nonuniform region, the contrast of the captured image may be lowered. Therefore, when the main surface is irradiated with laser light from the outside of the translucent substrate 10, detection with high accuracy may be difficult even if a defect or the like is detected by image processing using proximity difference. On the other hand, according to this example, by performing image processing using remote difference, it is possible to detect defects more appropriately even when the contrast of the captured image is low. Therefore, according to this example, even when the main surface of the translucent substrate 10 is irradiated with laser light from the outside, a defect can be detected more appropriately.

  As described above, the present invention has been described using the embodiment, but the technical scope of the present invention is not limited to the scope described in the above embodiment. It will be apparent to those skilled in the art that various modifications or improvements can be added to the above embodiment. It is apparent from the description of the scope of claims that embodiments with such changes or improvements can be included in the technical scope of the present invention.

  The present invention can be suitably used for, for example, a method for manufacturing a mask blank substrate.

DESCRIPTION OF SYMBOLS 10 ... Translucent board | substrate, 12 ... Thin film for pattern formation, 20 ... Mask blank, 30 ... Mask for transfer, 100 ... Inspection apparatus, 102 ... CCD, 104 ... A / D converter, 106 ... image processing device, 108 ... lens system, 110 ... laser light source, 112 ... mirror, 114 ... table, 116 ... drive means, 120 ...・ Optical member, 202... Pixel, 204... Pixel, 302... Main surface, 304.

Claims (11)

A method of manufacturing a mask blank substrate using a translucent substrate,
Preparing a translucent substrate having two opposing main surfaces;
A photographing step of photographing one main surface of the translucent substrate in a state in which laser light is applied to at least one of the main surfaces;
Image processing is performed on the photographed image of the photographed main surface, and the presence or absence of light scattered from the region is detected when the laser light strikes an optically non-uniform region present in the translucent substrate. A detection process;
A selection step of selecting a translucent substrate free of scattered light as a mask blank substrate in the detection step;
The method of manufacturing a mask blank substrate, wherein the image processing includes processing for obtaining a light amount difference of a photographed image by remote difference. In the photographing step, laser light is introduced into the translucent substrate and propagated through the translucent substrate while repeating total reflection between two main surfaces. Shoot the main surface of
In the detecting step, image processing is performed on the photographed image of the photographed main surface, and laser light propagating in the translucent substrate hits an optically non-uniform region existing in the translucent substrate. The presence or absence of the light which leaks from either main surface is detected by this, The manufacturing method of the board | substrate for mask blanks of Claim 1 characterized by the above-mentioned. The translucent substrate has an end face at an outer edge of the opposing main surface, and has a chamfered surface at each corner where the main surface and the end face intersect,
3. The method for manufacturing a mask blank substrate according to claim 2, wherein the surface into which the laser beam is introduced is any one of chamfered surfaces. 4. The method for manufacturing a mask blank substrate according to claim 2, wherein a surface into which the laser light is introduced and a main surface on which the translucent substrate is photographed are polished to a mirror surface. 5.   2. The method for manufacturing a mask blank substrate according to claim 1, wherein in the imaging step, the at least one main surface is irradiated with laser light from the outside of the translucent substrate. 3.   The image processing is processing for creating a light amount distribution of a photographed image by calculating a light amount difference between a processing target pixel of the photographed image and a pixel not adjacent to the pixel. Item 6. A method for producing a mask blank substrate according to any one of Items 1 to 5. The image processing calculates a light amount difference between the processing target pixel and a pixel separated from the processing target pixel by the number of pixels corresponding to the distance of twice or more the size of the defect to be detected in the captured image. 7. The method of manufacturing a mask blank substrate according to claim 1, wherein the process is a process of creating a light amount distribution of a photographed image. A method of manufacturing a mask blank substrate using a translucent substrate,
Preparing a translucent substrate having two opposing main surfaces;
Photographing one main surface of the translucent substrate with laser light introduced into the translucent substrate and propagating through the translucent substrate while repeating total reflection between the two main surfaces Shooting process to
Image processing is performed on a photographed image of the photographed main surface, and laser light propagating in the translucent substrate hits an optically non-uniform region existing in the translucent substrate, A detection step of detecting the presence or absence of light leaking from the main surface of
In the detection step, and having a selection step of selecting a translucent substrate that does not leak light as a mask blank substrate,
The method of manufacturing a mask blank substrate, wherein the image processing includes processing for obtaining a light amount difference of a photographed image by remote difference. 9. A mask blank manufacturing method comprising a step of forming a pattern forming thin film on a main surface of a mask blank substrate manufactured by the method for manufacturing a mask blank substrate according to claim 1. Method. A method for producing a transfer mask, comprising a step of forming a transfer pattern on the thin film for pattern formation of a mask blank produced by the method for producing a mask blank according to claim 9. A method of manufacturing a semiconductor device, wherein a circuit pattern is formed on a semiconductor wafer using the transfer mask according to claim 10.