JP6268042B2 - Inspection method - Google Patents

Description

  The present invention relates to an inspection method, and more particularly to a template inspection method.

  Along with the high integration and large capacity of large scale integration (LSI), circuit dimensions required for semiconductor elements are continually miniaturized. In recent years, extreme ultraviolet (EUV) lithography and nanoimprint lithography (NIL) have attracted attention as techniques for forming such fine patterns. Since EUV lithography uses extreme ultraviolet light as a light source, it is possible to form a finer pattern than a conventional exposure apparatus using ArF light. On the other hand, in nanoimprint lithography, a pattern is formed in a resist by applying pressure to a resist on a wafer with a mold having a nanoscale microstructure. In any technique, the pattern formed on the mask or template as the original plate becomes finer than conventional ArF lithography, and high inspection accuracy is required for the inspection.

  In a mask transfer process by EUV lithography, an exposure apparatus called a stepper or a scanner is used. The exposure apparatus uses light as a transfer light source, and projects the circuit pattern provided on the mask on a wafer after reducing the circuit pattern from one-fourth to about one-fifth. That is, in the mask, a pattern having a size 4 to 5 times the circuit size formed on the wafer is formed. On the other hand, in the template in nanoimprint lithography, a pattern having a size equal to the circuit size is formed on a printing plate with a predetermined depth of engraving. For example, in the case of advanced semiconductor devices, the line width of patterns and spaces between patterns is expected to be several tens of nm to several tens of nm, and the depth of engraving is expected to be several tens of nm to 100 nm.

  On the other hand, since the pattern of the template is the same as the circuit dimensions, if the template has a defect, the influence of the defect on the pattern transferred to the wafer is larger than that of the mask pattern. Further, since the template is used for multiple times of transfer, the defect is transferred to all the wafers together with the pattern. Therefore, when inspecting the template pattern, higher accuracy is required than in the case of inspecting the mask pattern.

  In the nanoimprint technology, in order to increase productivity, a plurality of replica templates to be duplicated are produced using a master template to be an original, and then each replica template is mounted on a different nanoimprint apparatus and used. Needless to say, high accuracy is required for the master template, and the replica template needs to be manufactured to accurately correspond to the master template. Patent Document 1 discloses an inspection apparatus for detecting such a template defect.

JP 2012-26977 A

  Conventionally, the same inspection is applied to both the master template and the replica template. The inspections for these are intended exclusively to detect pattern shape defects, and a defect determination algorithm and a defect recording method suitable for it have been devised. In recent inspection apparatuses, functions for detecting defects due to line width variations have been enhanced in response to the lack of LSI manufacturing margin caused by line width variations of patterns. On the other hand, in a state-of-the-art semiconductor device, the size of a defect to be pointed out as a pattern shape defect or a line width variation is comparable to the line width variation (line width distribution) on the entire transfer surface of the template. For this reason, the problem that the shape defect and line width variation which should be detected originally cannot be detected, or the shape and line width which do not need to be detected as a defect will be detected as a defect occurs. This problem will be described in detail below.

  As methods for detecting defects, there are a die-to-database comparison method and a die-to-die comparison method. The die-to-database comparison method is an inspection method for comparing a reference image generated from design pattern data with an optical image of an actual pattern. On the other hand, the die-to-die comparison method compares optical images of the same pattern on different chips when a plurality of chip patterns having the same configuration are arranged on a part or the whole of the same template. Inspection method.

  In the die-to-database comparison method, in the step of generating a reference image, filter processing using an optical image of a typical pattern portion of a template as a model for the design pattern data, that is, learning of a filter coefficient is performed. Thereby, the reference image becomes a pattern image having a line width tendency following the pattern line width of the learned region. Therefore, when there is a distribution in the line width dimension when viewed on the entire transfer surface, the line width bias (deviation) of the pattern is inspected when inspecting an area with a pattern line width different from the pattern line width of the learned area. In a certain state, the optical image and the reference image are compared. As a result, as described above, a shape defect or line width variation that should be detected originally cannot be detected, or a shape or line width that does not need to be detected as a defect may be detected as a defect.

  Also in the die-to-die comparison method, when chips in regions having different line width distributions are compared, patterns having a line width bias (deviation) are compared. Therefore, similarly to the case of the die-to-database comparison method, there arises a problem that a defect to be detected cannot be detected, or a shape or line width that does not need to be detected as a defect is detected as a defect. .

  The present invention has been made in view of these problems. That is, an object of the present invention is to provide an inspection method capable of detecting defects that should be detected in template inspection while reducing unnecessary defect detection.

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

One aspect of the present invention is a process of acquiring an image of each pattern for a master template in which a plurality of chip regions in which the same pattern is formed is arranged;
Comparing each pattern image of the master template with a die-to-die method;
Storing first inspection information including an image of each pattern of the master template and a result of comparing each pattern image of the master template by a die-to-die method;
For replica template replica of the master template, obtaining an image of each pattern formed on the replica template;
Comparing each pattern image of the replica template with a die-to-die method;
Comparing at least a part of images of each pattern of the replica template with a corresponding coordinate image created from the first inspection information as a reference image by a die-to-die method. It relates to the inspection method used as a special report.

One aspect of the present invention is the step of storing second inspection information including an image of each pattern of the replica template;
The replica template is pressed against the substrate, and the step of transferring the pattern of the replica template to the film provided on the substrate is performed at least once until the replica template is washed, and then the replica template is washed a predetermined number of times. Obtaining an image of each pattern formed on the replica template;
For each of the images, it is preferable that the image of the corresponding coordinates read from the second inspection information is compared as a reference image by a die-to-die method.

  In one aspect of the present invention, in the step of obtaining an image of each pattern of the master template, the polarized light is irradiated with linearly polarized light with the vibration direction inclined with respect to the repeating direction of the pattern, and elliptically polarized light out of the light reflected from the pattern It is preferable to extract the components to obtain the image.

  It is more preferable to irradiate the linearly polarized light so that the vibration direction forms an angle other than the angles in the ranges of -5 degrees to 5 degrees and 85 degrees to 95 degrees with respect to the repeating direction of the pattern.

  In one aspect of the present invention, in the image created from the first inspection information, a short defect among defects detected by comparing images of each pattern of the master template with a die-to-die method. This information is preferably converted into open defect information, and the open defect information is preferably converted into short defect information.

  1 aspect of this invention WHEREIN: At least one part of the image of each pattern of the said replica template is an image determined to be a defect by the process of comparing the images of each pattern of the said replica template with a die-to-die system. It is preferable that

  In one aspect of the present invention, it is preferable that at least a part of an image of each pattern of the replica template is an image having a low S / N ratio among images of each pattern of the replica template.

  According to one aspect of the present invention, there is provided an inspection method capable of detecting defects that should be detected in the template inspection while reducing unnecessary defect detection.

It is a flowchart of the inspection method of the template by this Embodiment. It is a figure which shows a mode that the several chip area | region is arranged along the X direction and the Y direction on a template. It is a figure which shows the mounting direction of the template on an XY (theta) table. It is a figure which shows the structure of the test | inspection apparatus in this Embodiment. It is a plane schematic diagram of the transfer surface of a template. It is a figure explaining the acquisition procedure of the optical image of the pattern formed in the template. It is the figure which expanded and showed a part of chip area | region provided in the template. It is explanatory drawing of the method of producing a replica template. It is explanatory drawing of the method of producing a replica template. It is explanatory drawing of the method of producing a replica template. It is explanatory drawing of the method of producing a replica template. It is a schematic diagram of a short defect. It is a schematic diagram of an open defect. It is a schematic diagram of the defect by edge roughness.

  FIG. 1 shows a flowchart of a template inspection method according to this embodiment. More specifically, this inspection method relates to inspection of a master template and inspection of a replica template copied from the master template.

  S1 to S3 in FIG. 1 correspond to the inspection process of the master template.

  In S1, an optical image of the pattern of the master template is acquired.

  On the surface of the master template, as shown in FIG. 2, a plurality of chip regions are arranged along the X direction and the Y direction. A repeated pattern is formed in each chip region. The repetitive pattern is, for example, a wiring pattern such as a line-and-space pattern, specifically, a pattern in which a plurality of line portions are arranged at a constant pitch along the X direction. In this case, the arrangement direction (X direction) of the line portions is referred to as “repeating direction of repeating pattern”.

  FIG. 4 is a diagram illustrating a configuration of the inspection apparatus according to the present embodiment. The inspection apparatus includes an optical image acquisition unit that acquires an optical image of an inspection target and a defect detection unit that detects a defect by performing a die-to-die comparison based on the acquired optical image. The inspection target in the present embodiment is a master template or a replica template. In the present specification, when the master template and the replica template are not particularly distinguished, they are simply referred to as templates.

  In the optical image acquisition unit of FIG. 4, the template 1 is placed on a Z table 2 that is movable in the vertical direction. The Z table 2 can also be moved in the horizontal direction by the XYθ table 3.

  In order to obtain an accurate inspection result, first, the template 1 needs to be placed at a predetermined position on the XYθ table 3. Therefore, usually, an alignment mark is formed on the template 1, and the alignment of the template 1 is performed on the XYθ table 3 using this alignment mark.

  The alignment (plate alignment) of the template 1 will be described with reference to FIG.

  FIG. 5 is a schematic plan view of the transfer surface of the template 1. In this example, cross-shaped alignment marks MA are formed at the four corners of the template 1. These alignment marks MA are in a horizontal / vertical relationship with each other in design. The template 1 is provided with the plurality of chip regions (C1, C2, C3,...) Described above, and chip alignment marks CA are also formed in each chip region.

  The XYθ table 3 on which the template 1 is placed can move in the X axis direction and the Y axis direction as well as in the rotation direction (θ direction) about the Z axis.

  Specifically, the alignment step is a step of aligning the X axis and Y axis of the pattern to be inspected with the travel axis of the XYθ table 3 in a state where the template 1 is placed on the XYθ table 3.

  First, among the alignment marks MA provided at four locations, two alignment marks MA having a small Y coordinate value are photographed, and the XYθ table 3 is rotated so that both marks are exactly the same Y coordinate. Then, fine adjustment of the rotation direction of the template 1 is performed. At this time, the distance between the alignment marks MA is also accurately measured. Next, two alignment marks MA having a large Y coordinate value are photographed. As a result, the coordinates of all four alignment marks MA are accurately measured.

  From the above measurement, for example, it is assumed that the other two alignment marks MA are located at the apex of a trapezoid having two alignment marks MA having small Y-coordinate values at both ends of the base. Since the shape of the template 1 is a rectangle, the other two alignment marks MA should be positioned at the vertices of the rectangle. However, the measurement result shows that it is located at the apex of the trapezoid, and when these are considered, it can be understood that the shape of the template 1 is distorted. Therefore, it is presumed that the pattern region to be inspected also has a trapezoidal distortion similar to the above trapezoid and the expansion / contraction of the distance between the alignment marks MA.

  The template 1 may not have the alignment mark MA. In that case, among the patterns to be inspected, alignment is performed using corner vertices and edge pattern sides that are as close as possible to the outer periphery of the template 1 and have the same XY coordinates.

  The other main elements constituting the optical image acquisition unit of FIG. 4 are the optical system 4, the image sensor 13, and the laser length measurement system 15, and their functions are as follows.

  In FIG. 4, the optical system 4 irradiates the template 1 with light and causes the light reflected by the template 1 to enter the image sensor 13. Specifically, in the optical system 4, the light emitted from the light source 17 passes through the aperture stop 5 and the polarizing plate 6, is reflected by the half mirror 7, and then irradiates the transfer surface of the template 1 by the objective lens 8. .

  The polarizing plate 6 is adjusted so that the transfer surface of the template 1 is illuminated by linearly polarized light (p-polarized light). The p-polarized light is linearly polarized light whose vibration surface is parallel to a surface (incident surface) parallel to the incident direction among surfaces including the normal line of the transfer surface of the template 1. In the present embodiment, it is preferable that the repetitive direction of the repetitive pattern formed on the template 1 forms an angle of 45 degrees with respect to the vibration direction of the p-polarized light to be irradiated. FIG. 3 is a diagram showing the placement direction of the template 1 on the XYθ table 3. The vibration direction of p-polarized light indicated by a broken line is parallel to the X direction in the figure. With respect to this direction, the repeating direction of the repeating pattern is inclined 45 degrees.

  When the polarization direction of linearly polarized light is 45 degrees with respect to the repeating direction of the repeating pattern, the electric field of the incident light is equal in the vertical component and the horizontal component. In contrast, the electric field of the reflected light due to the short defect (a1 in FIG. 12) in which the line portions are short-circuited is larger in the horizontal component than in the vertical component. As a result, the polarization direction of the light reflected by the short defect is inclined in a direction perpendicular to the repeating direction of the repeated pattern. Further, in the same example, in the case of an open defect (a2 in FIG. 13) in which the line portion is disconnected, it is inclined in the repeating direction of the repeated pattern.

  On the other hand, in the case of a defect due to edge roughness (a3 in FIG. 14), the line portions are not connected to each other or the line portions are not disconnected. Since the size of the unevenness is finer than that of the short defect or the open defect, the difference in sensitivity between the electric field component of the illumination light in the horizontal direction and the vertical direction is not so large. Therefore, when the linearly polarized light is incident on the template 1 perpendicularly and the polarization direction of the linearly polarized light is 45 degrees with respect to the repeating direction of the repeated pattern, the polarization direction of the light scattered by the edge roughness is the polarization of the incident light. The value is close to 45 degrees which is the direction. However, due to the influence of the base pattern having a periodic repetition, the polarization direction is not completely 45 degrees, but takes a value slightly deviated from 45 degrees.

  As described above, the influence of the illumination light on the polarization state is different between the short defect and the open defect and the edge roughness. Therefore, by utilizing this difference, it is possible to classify defects even if the pattern is finer than the resolution limit of the optical system. Specifically, by controlling the polarization state of the illumination light and the conditions of the polarization control element (polarizing plate 9 in FIG. 4) in the optical system that forms an image of the light reflected by the template 1, the brightness and darkness due to edge roughness are controlled. Unevenness can be removed by a polarization control element, and only amplitude changes due to short defects or open defects can be extracted. That is, it is possible to obtain an optical image having a high S / N ratio by eliminating the influence of edge roughness.

  The polarization plane of light is not necessarily 45 degrees with respect to the pattern repeating direction, but 0 degrees and 90 degrees are avoided. If the plane of polarization of light is 0 degree or 90 degrees with respect to the repeating direction of the repeating pattern, the sensitivity of the light becomes the same between the defects, so that it cannot be distinguished. Therefore, the polarization plane of light is preferably set to an angle other than the angles in the ranges of −5 to 5 degrees and 85 to 95 degrees.

  In the inspection apparatus of FIG. 4, when the linearly polarized light is irradiated on the repeated pattern of the template 1, elliptically polarized light is generated in the regular reflection direction from the repeated pattern. The reason for this is as follows.

  When linearly polarized light is repeatedly incident on the pattern, a polarization component whose vibration plane direction is parallel to the repetition direction and a polarization component whose vibration plane direction is perpendicular to the repetition direction are generated. These polarization components are independently subjected to different amplitude changes and phase changes. As a result, the reflected lights of the two polarization components have different amplitudes and phases, so that the reflected light obtained by combining these polarization components becomes elliptically polarized light.

  The regular reflection direction is included in the incident plane of the linearly polarized light irradiated on the template 1 and is inclined with respect to the normal line of the transfer surface of the template 1 by an angle equal to the incident angle of the linearly polarized light irradiated. It is.

  In FIG. 4, the specularly reflected light including elliptically polarized light passes through the objective lens 8 and the half mirror 7 and enters the polarizing plate 9. The polarizing plate 9 is set so that the direction of its transmission axis is orthogonal to the transmission axis of the polarizing plate 6, that is, in a crossed Nicols state. Thereby, the polarizing plate 9 transmits only the elliptically polarized light component in the light incident on the polarizing plate 9. The light that has passed through the polarizing plate 9 passes through the aperture stop 10, is reflected by the mirror 11, passes through the imaging lens 12, and then enters the image sensor 13. Specific examples of the image sensor 13 include a TDI (Time Delay Integration) sensor.

  The optical image of the pattern of the template 1 is continuously picked up by the image sensor 13 while the XYθ table 3 moves in the X direction and the Y direction. At this time, the movement position of the XYθ table 3 is measured by the laser length measurement system 15.

  FIG. 6 is a diagram for explaining the procedure for acquiring the optical image of the pattern formed on the template 1.

In FIG. 6, the template 1 is assumed to be placed on the XYθ table 3 of FIG. As shown in FIG. 6, the inspection area on the template 1 is virtually divided by 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 template 1 in the X direction or the Y direction.

An optical image is acquired for each stripe. Therefore, the operation of the XYθ table 3 is controlled so that the stripes 20 ₁ , 20 ₂ , 20 ₃ , 20 ₄ ,... In FIG. For example, the optical image of the template 1 is acquired while the XYθ table 3 moves in the −X direction of FIG. Then, an optical image having a scanning width W as shown in FIG. 6 is continuously input to the image sensor 13 of FIG. That is, the image sensor 13, after capturing an optical image of the first stripe 20 _1, capturing an optical image in the second stripe 20 _2. In this case, after the XYθ table 3 is moved stepwise in the -Y direction, and the first stripes 20 ₁ of the optical image direction when imaging the (-X direction) while moving in reverse direction (X-direction), scanning An optical image having a width W is continuously input to the image sensor 13. When capturing an optical image of the third stripes 20 _3, after XYθ table 3 is moved stepwise in the -Y direction, and the direction when imaging the second stripes 20 _second optical image (X-direction) backward, i.e., in the same direction (-X direction) as when capturing the first stripe 20 ₁ of the optical image, XY.theta. table 3 moves. In addition, the arrow of FIG. 6 has shown the direction and order where an optical image is imaged, and the shaded part represents the area | region where the imaging of the optical image was completed.

  The manner in which the optical image is picked up will also be described with reference to FIG. FIG. 7 is an enlarged view of a part of the chip area provided in the template 1. In this figure, 2n chip regions are arranged in a partial region of the template 1. As already described, the inspection area of the template 1 is virtually divided by a plurality of stripes. The image sensor 13 captures an image of a repeated pattern formed in each chip area along the stripe in the order of the first chip, the second chip, the third chip,.

  The optical image picked up by the image sensor 13 is photoelectrically converted and further A / D (analog / digital) converted into optical image data. Thereafter, the optical image data is sent to the comparison processing unit 14 of the defect detection unit. On the other hand, the position data of the XYθ table 3 measured by the laser length measurement system 15 is sent to the position data section 16 of the defect detection section. Thereafter, the position data is sent from the position data unit 16 to the comparison processing unit 14.

  In the comparison processing unit 14, defect detection is performed by a die-to-die comparison method. This step corresponds to S2 in FIG.

  The die-to-die comparison method compares optical images of the same pattern in different chip regions of a template when a plurality of chip regions having the same pattern configuration are arranged on a part or the whole of the same template. Inspection method. That is, in the die-to-die comparison method, chip regions repeatedly formed on the template are compared with each other.

  The die-to-die comparison method will be described in more detail with reference to FIG. The image sensor 13 images the pattern in the order of the chip C1, the chip C2, and the chip C3 along the stripe 20 in FIG. The comparison processing unit 14 divides the captured stripe data into test frame units, and compares the test frame cut out from the chip C1 with the test frame cut out from the chip C2. This comparison is performed by a comparison unit (not shown) provided in the comparison processing unit 14 and having an inspection frame as a processing unit.

  In die-to-die comparison, optical image data to be inspected is compared with optical image data to be a defect determination reference using an appropriate comparison determination algorithm. Then, when the difference between the two exceeds a predetermined threshold, it is determined as a defect. For example, in FIG. 5, when the difference between the pattern in the optical image of the chip C2 and the pattern in the optical image of the chip C1 exceeds a predetermined threshold, it is determined that the chip C2 has a defect.

  Here, if there is a short defect or open defect in a repetitive pattern finer than the resolution limit of the optical system, the regularity of the pattern will be disturbed, and the gradation value of the place where the defect exists will be the surrounding gradation value. Will be different. Thereby, a short defect and an open defect can be detected. In the present embodiment, by applying the optical system 4 shown in FIG. 4, light and dark unevenness due to edge roughness is removed by a polarization control element (polarizing plate 9 in FIG. 4), and only amplitude changes due to short defects or open defects are removed. Can be extracted. The optical image data (from which defects due to edge roughness are removed) is sent to the comparison processing unit 14. Then, in the comparison processing unit 14, pixel data in the optical image is represented by a gradation value for each pixel. Further, the inspection area of the template 1 is divided into predetermined unit areas, and an average gradation value of each unit area is obtained. The predetermined unit area can be, for example, an area of 1 mm × 1 mm.

  As an example, consider a case where a plurality of chip patterns having the same configuration are arranged on a part or the whole of a template. More specifically, it is a case where a repeated pattern of the same integrated circuit transferred to the semiconductor wafer is arranged. Here, the repeating unit has a rectangular shape of the same size, and is called a die when separated from each other. One die is usually formed with one unit of integrated circuit and the like. When this repeated pattern is inspected by a die-to-die comparison method, optical images of the same pattern on different chips are compared. For example, when an optical image of the nth chip is an inspection target, the optical image of the (n−1) th chip is a reference image to be compared. At this time, if the repetitive pattern is a pattern that cannot be resolved at the wavelength of the light source of the inspection apparatus, optical images with uniform gray gradation are compared in most of the inspection region. However, in an optical image having a defect in the pattern, the defect portion is observed as a white bright spot or a black spot depending on the type or shape.

  For example, when an optical image of a repetitive pattern is acquired by irradiating the template with light and causing the reflected light to enter the sensor, if the adjacent patterns are connected and short-circuited, the light at the defect location The defect is observed as a white bright spot because it reflects in a wider area. On the other hand, if the pattern is disconnected, the pattern is missing at that portion, so that the light reflection area is reduced and the defect is observed as a black spot.

  In the comparison processing unit 14, before and after the inspection, upper and lower threshold values centered on the average gradation value are set in advance. Then, a portion where the gradation value obtained for each pixel exceeds this threshold corresponds to a portion that is observed as a white bright spot or a black spot, so that such a portion is determined as a defect.

  All the optical images picked up for the master template and the determination result in the comparison processing unit 14 are stored in the inspection information 19 as the first inspection information (S3 in FIG. 1). The determination result in this case is not related to the presence or absence of defects.

  After inspecting the master template as described above, the replica template copied from the master template is inspected. This step corresponds to the inspection (1) of the replica template in FIG. The inspection (1) of the replica template is an inspection of the performance. By this inspection, for example, in the inspection of the master template, it is grasped whether the pattern allowed near the defect determination threshold value becomes a defect in the replica template.

  In the present embodiment, as shown in FIG. 1, first, a reference image used for inspection of the replica template is created using the inspection information of the master template stored in S3 (S4 in FIG. 1). This step is performed by the reference image generation unit 18 in FIG. That is, the first inspection information of the master template stored in S3 is sent from the inspection information 19 to the reference image generation unit 18, and the reference image generation unit 18 creates the reference image. The step S4 is not necessarily performed before the step (S5) of acquiring the optical image of the replica template, and may be performed in parallel with the step S5 or may be performed after the step S5. Is possible.

  8 to 11 show an example of a method for producing a replica template. According to this method, first, as shown in FIG. 8, the master template MTP is pressed against the surface on which the hard mask HM of the quartz glass QG and the resist film RF are provided. The arrows in FIG. 8 indicate the direction in which the master template MTP is pressed. In this state, the resist film RF is cured by irradiating the resist film RF with ultraviolet light. Thereafter, as shown in FIG. 9, the master template MTP is raised in the direction of the arrow and peeled off from the quartz glass QG. Next, the hard mask HM is etched using the resist film RF as a mask, and the quartz glass QG is etched using the hard mask HM as a mask to obtain the structure shown in FIG. Subsequently, the resist film RF and the hard mask HM are removed, and the quartz glass QG is diced to a predetermined size, whereby a replica template RTP as shown in FIG. 11 is obtained.

  As is clear from the replica template manufacturing process described above, the master template and the replica template are in a relationship in which the unevenness of the pattern is inverted. For this reason, when a short defect in which lines are short-circuited is detected in the master template, this defect is detected as an open defect in which the lines are disconnected in the replica template. Therefore, in the step of creating the reference image used for the inspection of the replica template using the inspection information of the master template, the short defect information of the master template is converted into the information of open defects. Further, the open defect information of the master template is converted into short defect information.

  In addition, the master template and the replica template have a relationship in which the x coordinate of the pattern is inverted. For example, the coordinates (-1, 0) of the replica template correspond to the coordinates (1, 0) of the master template. Therefore, in the step of creating a reference image used for inspection of the replica template using the inspection information of the master template, the x coordinate of the master template is converted into the coordinate whose sign is inverted around the Y axis.

  In addition, the duty ratio defined by the width dimension and pitch of each line in the line and space pattern is also inverted between the master template and the replica template. Since the gradation value of the optical image changes when the duty ratio changes, the gradation value of the optical image of the master template is matched with the gradation value of the replica template.

  By the way, if the repeating pattern formed on the template 1 has a defect, the regularity of the repetition is disturbed, and the optical image is an image having a luminance change according to the degree of the defect in a gray image having a uniform gradation value. It becomes. This luminance change is observed as, for example, an isolated white spot or black spot. On the other hand, even if there is a defect in the master template, if you create a reference image that resembles the replica template to which the defect has been transferred, there will be no difference between the reference image and the optical image of the replica template. There is a risk that it will be determined that there is no. Therefore, a reference image is created by simply matching the background gradation value of the defect of the master template with the background gradation value of the replica template. Thereby, for example, a defect of the master template is observed as a black spot, whereas a defect of the replica template is observed as a white spot, so that the defect of the replica template can be detected.

  Next, an optical image of the pattern of the replica template is acquired (S5 in FIG. 1). This process is the same as the process of acquiring the optical image of the pattern of the master template described with reference to FIGS.

  In the optical image acquisition process of the master template, as shown in FIG. 4, polarizing plates (6) are provided on both the optical system that irradiates light to the template 1 and the optical system that causes the light reflected by the template 1 to enter the image sensor 13. 9). According to this configuration, the light emitted to the master template is linearly polarized (p-polarized), and the vibration direction of the p-polarized light is inclined 45 degrees with respect to the repeating direction of the repeated pattern and is reflected by the master template. Of the received light, only the elliptically polarized light component enters the image sensor.

  The configuration using the two polarizing plates (6, 9) as described above can increase the sensitivity of defect inspection, but the amount of light incident on the image sensor 13 is reduced by these polarizing plates. Therefore, the process for acquiring an optical image becomes long. As will be described later, since the inspection information of the master template is referred to in the replica template inspection process, an optical system having a lower sensitivity than the master template inspection process may be used. That is, an optical image can be acquired without using two polarizing plates (6, 9), thereby shortening the time required for the optical image acquisition process.

  The optical image of the replica template picked up by the image sensor 13 is photoelectrically converted and further A / D (analog / digital) converted into optical image data. Thereafter, the optical image data is sent to the comparison processing unit 14 of the defect detection unit in FIG. On the other hand, the position data of the XYθ table 3 measured by the laser length measurement system 15 is sent to the position data section 16 of the defect detection section. Thereafter, the position data is sent from the position data unit 16 to the comparison processing unit 14.

  In the comparison processing unit 14, defect detection is performed by a die-to-die comparison method. This step corresponds to S6 in FIG. 1 and is characterized by two die-to-die comparison steps.

  First, similarly to S2 of FIG. 1, the defect detection of the replica template is performed by the die-to-die comparison method (first die-to-die comparison). Specifically, optical images of the same pattern in different chip regions of the replica template are compared.

  When the template alignment was described, it was described that the shape distortion of the template can be grasped by measuring the coordinates of the alignment marks arranged at the four corners of the template. According to the die-to-die comparison method, since there is a similar shape distortion in the pattern of the chip area to be compared, it is not necessary to consider correction due to the shape distortion in defect detection.

  On the other hand, when looking at the entire transfer surface of the template, there is a dimensional distribution in the pattern, and in the most advanced semiconductor devices that are becoming finer, the dimensions of defects that should be pointed out as pattern shape defects or line width variations are The size distribution is equivalent to that of the entire transfer surface. For this reason, when trying to compare chip regions having different line width distributions, patterns with line width bias (deviation) are compared, and a defect to be detected cannot be detected or detected as a defect. There is a risk that shapes and line widths that do not need to be detected are detected as defects.

  Therefore, in the present embodiment, it is difficult to distinguish between a defective portion detected by a die-to-die comparison method that compares different chip regions of a replica template, short defects, open defects, and edge roughness. Die-to-die comparison (second die-to-die comparison) using the image created in S4 as a reference image is performed for a portion having a low S / N ratio. However, the present embodiment is not limited to this, and the entire optical image of the replica template may be compared die-to-die with the corresponding image of the master template. Further, in the inspection of the master template, the pattern allowed near the defect determination threshold may be compared with the pattern of the corresponding replica template. Thereby, it is grasped whether or not the above pattern is allowed even in the replica template.

  Specifically, in FIG. 4, the reference image generated by the reference image generation unit 18 is sent to the comparison processing unit 14. Then, in the comparison processing unit 14, a reference image created from the optical image of the replica template at the location specified by the position information from the position data unit 16 and the optical image of the master template at the same coordinates as the optical image. Are compared. In this case, since the coordinates of the images to be compared are the same, there is no need to consider the difference in the line width distribution. Therefore, even if it is an optical image acquired without using a polarizing plate, accurate defect detection in a replica template can be performed using this optical image.

  All the optical images picked up for the replica template and the determination result in the comparison processing unit 14 are stored in the inspection information 19 as second inspection information (S7 in FIG. 1). The determination result in this case is not related to the presence or absence of defects.

  The replica template determined as good in the replica template inspection (1) described above is mounted on a known nanoimprint apparatus (not shown) and used for pattern transfer onto a silicon wafer or the like.

  For example, in the nanoimprint apparatus, the replica template is held by a holding unit that is provided on the ceiling of the chamber and is movable in the Z direction. On the other hand, the silicon wafer is placed on a stage that is disposed below the holding unit and is movable in the X direction and the Y direction. The stage is moved by a linear motor, and the position of the stage is measured by a laser interferometer.

  A resist film is provided on the surface of the silicon wafer, and the replica template is pressed against the resist film when the holding portion moves in the −Z direction (downward). In this state, the resist film is irradiated with ultraviolet light and cured, and then the holding part is moved in the Z direction to peel off the replica template from the resist film. Thereby, the repeated pattern of the replica template is transferred to the resist film.

  After the transfer process to the silicon wafer is performed a predetermined number of times using the replica template, cleaning is performed to remove the resist film residue, other dust, and dust attached to the surface of the replica template. After the cleaning process is completed, the process is again mounted on the holding unit of the nanoimprint apparatus, and the process of transferring the repeated pattern to the silicon wafer is performed.

  By repeating the above steps, if the pattern of the replica template is chipped or worn, or if foreign matter or dirt that cannot be removed by the cleaning step is attached, a defect occurs in the pattern transferred to the silicon wafer. Therefore, every time the cleaning process is completed a predetermined number of times, the replica template is inspected. This inspection step corresponds to inspection (2) of the replica template in FIG.

  First, an optical image of a replica template pattern is acquired (S8 in FIG. 1). This step is the same as the optical image acquisition step of the replica template in S5.

  The optical image of the replica template picked up by the image sensor 13 is photoelectrically converted and further A / D (analog / digital) converted into optical image data. Thereafter, the optical image data is sent to the comparison processing unit 14 of the defect detection unit in FIG. On the other hand, the position data of the XYθ table 3 measured by the laser length measurement system 15 is sent to the position data section 16 of the defect detection section. Thereafter, the position data is sent from the position data unit 16 to the comparison processing unit 14.

  In the comparison processing unit 14, defect detection is performed by a die-to-die comparison method. This step corresponds to S9 in FIG.

  In step S9, the replica template is detected for defects by a die-to-die comparison method. The reference image in this process is an optical image of the coordinates (x, y) to be inspected and the coordinates (−x, y) obtained by inverting the sign of the x coordinate acquired in the inspection (1) of the replica template. Since this optical image is an optical image of the initial state of the replica template, it can be used as a reference image to grasp changes in the replica template, that is, the state of chipping or wear of patterns, the adhesion state of foreign matter or dirt, etc. Can do.

  The optical image of the replica template acquired in S8 and the determination result in the comparison processing unit 14 in S9 can be stored in the inspection information 19 as third inspection information (S10 in FIG. 1).

  In the replica template inspection (2) in FIG. 1, the replica template is pressed against the silicon wafer, and the process of transferring the pattern of the replica template onto the resist film on the silicon wafer is performed at least once until the replica template is cleaned. This process is performed every time the above process is performed a predetermined number of times. The third examination information stored in S10 is used when generating a reference image for a similar examination to be performed later, if necessary.

  In the present embodiment, a review process is performed after each process (S3, S7, S10) for storing the inspection information in FIG. 1, thereby re-cleaning or discarding the template in which the defect is detected. It is possible to determine whether to do it. The review process is a process in which the operator determines whether or not the detected defect is a practical problem. The review device can be an external device of the inspection device, but may be a part of the inspection device. Of the stored inspection information, when the defect coordinates, the optical image or reference image that is the basis for the defect determination is sent to the review device, the operator compares the optical image containing the defect with the reference image, Determine if the defect really needs to be discarded or re-cleaned. A defect can be corrected by providing a correction process after the review process.

  As described above, in this embodiment, after comparing different chip regions of the replica template (first die-to-die comparison), the defect location, short defect, and open defect detected by this comparison are compared. The image created from the inspection information of the master template is compared again as a reference image (second die-to-die comparison) for a portion having a low S / N ratio that makes it difficult to distinguish between edge roughness and edge roughness. In the second die-to-die comparison, since the coordinates of the images to be compared are the same, it is not necessary to consider the difference in the line width distribution on the transfer surface of the replica template. That is, since the patterns are compared with each other without a line width bias (deviation), an accurate inspection can be performed.

  In the present embodiment, an optical image of a replica template that has been subjected to a pattern transfer process to a silicon wafer or the like a predetermined number of times is acquired and compared with the optical image of the initial replica template. Also in this case, since the coordinates of the images to be compared are the same, it is not necessary to consider the difference in the line width distribution on the transfer surface of the replica template. That is, since the patterns are compared with each other without a line width bias (deviation), an accurate inspection can be performed.

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

  For example, in the above-described embodiment, when an optical image of the master template is acquired, both the optical system that irradiates light to the master template and the optical system that causes the light reflected by the master template to enter the image sensor are used. And a polarizing plate were arranged. Thereby, since an inspection sensitivity can be made high, it becomes possible to perform an exact inspection. Since the inspection information of the master template is used for generating a reference image used for inspection of the replica template, it needs to be accurate, but is not limited to that obtained by the configuration of the above embodiment. For example, it may be inspection information based on an SEM microscopic image, that is, an image acquired using secondary electrons generated by an electron beam. Moreover, in the said embodiment, although the polarized light was irradiated to the repeating pattern of a template, this invention is not limited to this, You may irradiate a non-polarized light to the repeating pattern of a template, and may acquire an optical image.

  In this embodiment, the inspection method of the master template and the replica template is described. However, the present invention can be applied to the inspection of the master plate and the plate replicated from the master plate.

  In this embodiment, the inspection information is recorded in the inspection apparatus. However, the inspection information may be recorded in an external server connected to the inspection apparatus via a network so that it can be shared among a plurality of inspection apparatuses.

  Furthermore, in the above embodiment, description of parts that are not directly required for the description of the present invention, such as the device configuration and control method, has been omitted, but the required device configuration and control method can be appropriately selected and used. Needless to say, you can. In addition, all inspection methods and inspection apparatuses that include the 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.

DESCRIPTION OF SYMBOLS 1 Template 2 Z table 3 XY (theta) table 4 Optical system 5,10 Aperture stop 6,9 Polarizing plate 7 Half mirror 8 Objective lens 11 Mirror 12 Imaging lens 13 Image sensor 14 Comparison processing part 15 Laser length measurement system 16 Position data part 17 Light source 18 Reference image generator 19 Inspection information 20 Stripe

Claims (7)

For a master template in which a plurality of chip regions in which the same pattern is formed are arranged, obtaining a pattern image;
Comparing each pattern image of the master template with a die-to-die method;
Storing first inspection information including an image of each pattern of the master template and a result of comparing each pattern image of the master template by a die-to-die method;
For replica template replica of the master template, obtaining an image of each pattern formed on the replica template;
Comparing each pattern image of the replica template with a die-to-die method;
Comparing at least a part of images of each pattern of the replica template with a corresponding coordinate image created from the first inspection information as a reference image by a die-to-die method. A characteristic inspection method. Storing second inspection information including an image of each pattern of the replica template;
The replica template is pressed against the substrate, and the step of transferring the pattern of the replica template to the film provided on the substrate is performed at least once until the replica template is washed, and then the replica template is washed a predetermined number of times. Obtaining an image of each pattern formed on the replica template;
2. The method according to claim 1, further comprising: comparing each image with a corresponding coordinate image read from the second inspection information as a reference image in a die-to-die manner. Inspection method.   In the step of acquiring an image of each pattern of the master template, the vibration direction is inclined with respect to the repetition direction of the pattern to irradiate linearly polarized light, and an elliptically polarized component of light reflected from the pattern is extracted to obtain the image The inspection method according to claim 1, wherein the inspection method is acquired.   The linearly polarized light is irradiated so that a vibration direction forms an angle other than an angle in a range of -5 degrees to 5 degrees and 85 degrees to 95 degrees with respect to a repeating direction of the pattern. 3. The inspection method according to 3.   In the image created from the first inspection information, among the defects detected by comparing each pattern image of the master template with a die-to-die method, short defect information is open defect information. The inspection method according to claim 1, wherein the information on open defects is converted into information on short defects.   At least a part of the images of each pattern of the replica template is an image determined to be a defect by a step of comparing images of the patterns of the replica template with each other by a die-to-die method. The inspection method according to any one of claims 1 to 5.   The at least part of the images of each pattern of the replica template is an image having a low S / N ratio among the images of each pattern of the replica template. Inspection method described in 1.

Images