JP4599893B2 - Misalignment detection method - Google Patents

Description

  The present invention relates to a misregistration detection mark and a misregistration detection method used for pattern misregistration detection in a manufacturing process of a semiconductor element or a liquid crystal display element, and more particularly to overlay inspection of patterns formed on different layers of a substrate. The present invention relates to a preferred misalignment detection mark and misalignment detection method.

  In manufacturing processes such as semiconductor elements and liquid crystal display elements, a circuit pattern is transferred to a resist layer through a known exposure process and development process, and a predetermined material is obtained by performing processing such as etching through the resist pattern. A circuit pattern is transferred onto the film (pattern formation step). Then, by repeating the above pattern formation process many times, circuit patterns of various material films are stacked on a substrate (semiconductor wafer or liquid crystal substrate), and a circuit of a semiconductor element or a liquid crystal display element is formed. The

  Furthermore, in the manufacturing process described above, in order to accurately overlay circuit patterns of various material films (in order to improve product yield), each pattern forming process is performed after the development process and before the processing process. In addition, the positional deviation of the resist pattern on the substrate is detected. This is an overlay inspection of a resist pattern with respect to a circuit pattern (hereinafter referred to as “underground pattern”) of the underlayer formed in the previous pattern forming process.

  In overlay inspection (displacement detection), a base mark that indicates the reference position of the base pattern and a resist mark that indicates the reference position of the resist pattern are usually used. The ground mark and the resist mark are formed at the same time as the ground pattern and the resist pattern in the above pattern forming process, respectively, to form a square-shaped double mark 80 as shown in FIG. 9 (see, for example, Patent Document 1). FIG. 9 is a plan view of the double mark 80. In general, the double mark 80 has a base mark 81 on the outside and a resist mark 82 on the inside. The size D1 of the base mark 81 is, for example, about 30 μm, and the size D2 of the resist mark 82 is, for example, about 15 μm. The base marks 81 and the resist marks 82 are configured such that their centers coincide with each other when there is no positional deviation between the base pattern and the resist pattern.

At the time of overlay inspection using the base mark 81 and the registration mark 82, a measurement point including two marks (81, 82) is positioned in the field of view of the apparatus, and an image of the measurement point is displayed on a CCD camera or the like. The image sensor is used. Further, an edge image is cut out for each side of the ground mark 81 and the registration mark 82 from the captured image, and predetermined image processing is performed on the obtained partial image, whereby the center of the ground mark 81 and the registration mark are registered. The amount of positional deviation from the center of 82 is calculated. The positional deviation amount of the calculation result represents the positional deviation state of the resist pattern with respect to the base pattern.
JP-A-7-151514

  By the way, in order to cope with the miniaturization of circuit patterns accompanying the high integration of semiconductor elements and the like, it has been desired to improve the accuracy of the above-described positional deviation detection (overlay inspection). However, when the positional deviation is detected using the square double mark 80 described above, it is affected by distortion of the imaging optical system that forms an image of the double mark 80 on the imaging surface of the image sensor. As a result, the amount of misalignment between the base mark 81 and the registration mark 82 may not be detected accurately. Such a problem may occur not only in the two marks (81, 82) formed in different layers of the substrate but also in the two marks formed in the same layer of the substrate.

  An object of the present invention is to provide a misregistration detection mark and a misregistration detection method capable of reducing the influence of distortion of an imaging optical system when detecting misregistration.

Displacement detection method of the present invention, by using the positional deviation detecting mark used in the positional deviation detection of two patterns, a positional deviation detecting method for performing a positional deviation detection of the two patterns, the positional deviation detection The mark for use includes a first mark indicating one reference position of the two patterns, and a second mark indicating the other reference position of the two patterns, and the first mark is a first line. And a second linear pattern perpendicular to the linear pattern is arranged in a cross shape, and the second mark is a fourth linear pattern perpendicular to the third linear pattern and the third linear pattern. When the linear pattern is arranged in a cross shape and the first mark and the second mark are not misaligned between the two patterns, the first linear pattern and the third linear pattern The straight line directions match, And the linear direction of the said 2nd linear pattern and the said 4th linear pattern corresponds, and the length between the both ends of a longitudinal direction has the said 1st linear pattern and the said 3rd linear pattern. The first linear pattern and the third linear pattern, which are different from each other, have a straight line so that the longer linear pattern does not overlap the shorter linear pattern. It consists of two partial patterns divided into one end side and the other end side in the direction, the second linear pattern and the fourth linear pattern are different from each other in length between both ends in the longitudinal direction, Of the second linear pattern and the fourth linear pattern, the linear pattern with the longer length is one end side in the linear direction so as not to overlap the linear pattern with the shorter length. consists of two parts pattern divided into the other end side, not a said location The detection method relates to a first step of capturing images of the first mark and the second mark, and the first linear pattern and the second linear pattern of the first mark from the captured image. A second step of cutting out a partial image and cutting out a partial image related to the third linear pattern and the fourth linear pattern of the second mark, and the first step based on the cut out partial image. A linear displacement between the linear pattern and the third linear pattern is calculated, and a linear displacement between the second linear pattern and the fourth linear pattern is calculated. 3 steps, and in the second step, when cutting out a partial image related to the longer linear pattern of the first linear pattern and the third linear pattern, and Second linear pattern Ting when cutting the partial image concerning the linear pattern of the longer of the length of said fourth linear pattern, characterized in that cutting out the partial image separately for each of the partial patterns.

Of the first linear pattern and the third linear pattern in the misregistration detection mark, the shorter linear pattern, the second linear pattern, and the fourth linear pattern. linear pattern of the shorter of the length of the linear pattern, so as not to intersect with each other, each may consist of two parts pattern divided into the one end and the other end of the linear direction.

Further , at least one of the first linear pattern, the second linear pattern, the third linear pattern, and the fourth linear pattern in the misregistration detection mark is Alternatively, it may be composed of a plurality of fine linear patterns divided in the width direction.
Further , it circumscribes the longer linear pattern of the first linear pattern and the third linear pattern in the misalignment detection mark , and the second linear pattern and the At least one region among the four regions partitioned by the first mark and the second mark inside the rectangular region circumscribing the longer linear pattern of the fourth linear pattern the may include a third mark indicating at least one other reference position of the two patterns.

Further , it circumscribes the longer linear pattern of the first linear pattern and the third linear pattern in the misalignment detection mark , and the second linear pattern and the At least one region among the four regions partitioned by the first mark and the second mark inside the rectangular region circumscribing the longer linear pattern of the fourth linear pattern the may contain process information relating to at least one of the formation of the two patterns.

Further , it circumscribes the longer linear pattern of the first linear pattern and the third linear pattern in the misalignment detection mark , and the second linear pattern and the At least one region among the four regions partitioned by the first mark and the second mark inside the rectangular region circumscribing the longer linear pattern of the fourth linear pattern the may include inspection mark of exposure conditions related to at least one of the formation of the two patterns.

  According to the present invention, it is possible to reduce the influence of the distortion aberration of the imaging optical system when detecting the displacement.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
(First embodiment)
Here, an overlay inspection in a manufacturing process of a semiconductor element or a liquid crystal display element will be described as an example. Substrates (semiconductor wafers and liquid crystal substrates) that are subject to overlay inspection are in the process of forming another circuit pattern on the base pattern formed in the previous pattern formation process (that is, exposure to the resist film). It is in a state after development and before processing the material film directly under the resist film. The overlay inspection of a plurality of patterns formed on different layers of the substrate (that is, the overlay inspection of the resist pattern with respect to the base pattern) is performed by detecting the misalignment between the base pattern and the resist pattern.

The misalignment detection mark 10 of the first embodiment is a mark used for the overlay inspection described above, and is composed of a base mark 10A and a registration mark 10B as shown in FIGS. 1 (a) to 1 (c). ing. Base mark 10A is a mark of the base layer formed simultaneously with the base pattern, the size D _A, for example, about 30 [mu] m. Registration mark 10B is a mark of the resist pattern at the same time formed resist layer, a size D _B for example, about 15 [mu] m.

The ground mark 10A is a pattern in which a linear pattern 11 and a linear pattern 12 perpendicular to the linear pattern 11 are arranged in a cross shape. Center C _A of base mark 10A corresponds to the intersection of the straight line direction S ₁₂ linear direction S ₁₁ and the linear pattern 12 of the linear patterns 11. In the registration mark 10B, a linear pattern 13 parallel to the linear pattern 11 and a linear pattern 14 parallel to the linear pattern 12 are arranged in a cross shape. Center C _B of the registration mark 10B corresponds to the intersection between the linear pattern 13 and the linear pattern 14, corresponding to the intersection of the straight line direction S ₁₄ linear direction S ₁₃ and the linear pattern 14 of the linear patterns 13 .

In the figure, the linear patterns 11 and 12 are drawn thicker than the linear patterns 13 and 14, but the relationship between the thicknesses may be different from the figure.
The base marks 10A and the resist marks 10B are such that the linear directions S ₁₁ and S ₁₃ of the linear patterns 11 and ₁₃ coincide with each other when the base pattern and the resist pattern are not misaligned. 14 linear directions S ₁₂ and S ₁₄ coincide with each other. At this time, also coincides center C _B of the center C _A and registration marks 10B of the underlying mark 10A. In the following description, the direction parallel to the linear directions S ₁₁ and S ₁₃ of the linear patterns 11 and ₁₃ is defined as the X direction, and the direction parallel to the linear directions S ₁₂ and S ₁₄ of the linear patterns 12 and ₁₄ is defined as the Y direction. To do.

Further, with respect to the X direction of the misregistration detection mark 10, the linear pattern 11 of the base mark 10A and the linear pattern 13 of the resist mark 10B have a length between both ends (sizes D _A and D _{B in} FIG. 1A). Correspond to each other). The linear patterns 11 of the longer of the length between both ends, so as not to overlap the line pattern 13 of the shorter lengths across, divided between one end and the other end of the linear direction S ₁₁ The two partial patterns 11 (1) and 11 (2). The distance D _E inside the two partial patterns 11 (1) and 11 (2) is wider than the distance (size D _B ) between both ends of the linear pattern 13.

Similarly, the linear pattern 12 of the base mark 10A and the linear pattern 14 of the registration mark 10B are different from each other in the Y direction of the misregistration detection mark 10. Then, the linear patterns 12 of the longer of the length between both ends, so as not to overlap the line pattern 14 of the shorter lengths across, divided between one end and the other end of the linear direction S ₁₂ The two partial patterns 12 (1) and 12 (2) are formed. The interval between the two partial patterns 12 (1) and 12 (2) is wider than the interval between both ends of the linear pattern 14.

In the above-described misregistration detection mark 10, the distance D _E inside the linear pattern 11 (two partial patterns 11 (1) and 11 (2)) of the base mark 10 A, the linear pattern 12 (two partial patterns) The distance between the inner sides of the patterns 12 (1) and 12 (2)), the distance between both ends of the linear pattern 13 of the resist mark 10B (size D _B ), and the distance between both ends of the linear pattern 14 are as follows: It is optimized so as to have an appropriate range for the detection of misalignment.

Further, in the above-described misregistration detection mark 10, the partial patterns 11 (1) and 11 (2) of the linear pattern 11, the partial patterns 12 (1) and 12 (2) of the linear pattern 12, and the linear pattern 13 , And the linear pattern 14 are each constituted by a single pattern (a pattern having two edges in the width direction).
Next, a method for detecting misalignment (overlay inspection) between the ground pattern and the resist pattern using the misalignment detection mark 10 (the ground mark 10A and the resist mark 10B) according to the first embodiment will be described. Before describing the method, the overlay measuring apparatus 20 shown in FIG. 2 will be described.

  The overlay measurement apparatus 20 forms an optical image of the substrate 21, a stage 22 that supports the substrate 21, an illumination system (23 to 26) that irradiates the substrate 21 with illumination light L 1, as shown in FIG. An imaging system (25 to 28), an autofocus light receiving system (28 to 30), an image sensor 31, an image processing unit 32, a system control unit 33, and a stage control unit 34. Yes. Note that the misregistration detection mark 10 of the first embodiment is formed at a number of locations designated in advance on the surface (surface to be inspected) of the substrate 21.

  The stage 22 includes a holder that supports the substrate 21 in a horizontal state, and a drive unit that moves the holder in the horizontal direction (XY direction) and the vertical direction (Z direction) in accordance with an instruction from the stage control unit 34. Composed. By moving the holder of the stage 22 in the XY directions, any one of the measurement points (the misalignment detection mark 10 shown in FIG. 1) on the surface to be inspected of the substrate 21 is moved to the imaging system (25 to 28). Positioning can be performed immediately below the objective lens 26 (in the visual field region). Further, the focus adjustment of the measurement point of the substrate 21 is performed by moving the holder of the stage 22 in the Z direction. The Z direction corresponds to a direction parallel to the optical axis 6A of the objective lens 26 of the imaging system (25 to 28).

  The illumination system (23 to 26) includes a light source 23, an illumination lens 24, a half prism 25, and an objective lens 26. The light from the light source 23 passes through the illumination lens 24, the half prism 25, and the objective lens 26 (illumination light L1), and then enters the inspection surface of the substrate 21 on the stage 22 (epi-illumination). At this time, the measurement point (the misalignment detection mark 10 in FIG. 1) positioned in the visual field area is illuminated substantially vertically by the illumination light L1. Then, reflected light L2 is generated from the misalignment detection mark 10 illuminated by the illumination light L1. The reflected light L2 is guided to the imaging system (25 to 28) and the autofocus light receiving system (28 to 30).

  The imaging system (25 to 28) includes an objective lens 26, an imaging lens 27, and half prisms 25 and 28 (optical microscope section). The imaging lens 27 functions as a second objective lens. The reflected light L <b> 2 from the substrate 21 enters the imaging surface of the imaging element 31 after passing through the objective lens 26, the half prisms 25 and 28, and the imaging lens 27. At this time, an enlarged image based on the reflected light L2 (that is, an enlarged optical image of the misregistration detection mark 10) is formed on the imaging surface of the imaging element 31.

  The image pickup device 31 is a black and white area sensor (for example, a CCD camera) in which a plurality of pixels are two-dimensionally arranged. The image pickup device 31 picks up an optical image of the misregistration detection mark 10 on the image pickup surface and converts the image signal into a subsequent image. The data is output to the processing unit 32. The image signal output from the image sensor 31 is composed of a plurality of sample points, and represents a distribution relating to the luminance value for each pixel on the imaging surface. The luminance value is proportional to the intensity of the reflected light L2. The image of the misregistration detection mark 10 has a low luminance value at the edge portion of the misregistration detection mark 10.

  The autofocus light-receiving system (28 to 30) includes a half prism 28, an imaging lens 29, and a detector 30. The reflected light L 2 from the substrate 21 enters the detector 30 after passing through the objective lens 26, the half prisms 25 and 28, and the imaging lens 29. A focus signal corresponding to the in-focus state of the surface to be inspected of the substrate 21 (particularly, the measurement point in the visual field region) is output from the detector 30 to the stage controller 34.

  The stage control unit 34 controls the stage 22 in the XY directions to position one measurement point (the misalignment detection mark 10 shown in FIG. 1) on the surface to be inspected in the substrate 21 within the visual field region, and then receives the light receiving system ( 28 to 30), the stage 22 is controlled in the Z direction based on the focus signal from the detector 30 to adjust the focus of the measurement point on the substrate 21. Then, after the focus adjustment, the system control unit 33 that performs overall control of the entire apparatus controls the image processing unit 32 so that the position shift using the measurement point of the substrate 21 (the position shift detection mark 10 shown in FIG. 1). Run detection.

Here, when the stage control unit 34 controls the stage 22 to position the misregistration detection mark 10 on the substrate 21 in the field of view, the center of the misregistration detection mark 10 (that is, the center of the ground mark 10A) is roughly set. Positioning control is performed so that C _A and the center C _B of the registration mark 10B stop near the center of the visual field region. The center of the visual field region corresponds to a point on the optical axis 6A of the objective lens 26 of the imaging system (25 to 28).

The center of the positional shift detection mark 10 that is substantially coincident with the center of the viewing area (optical axis 6A of the objective lens 26), the linear direction S _11, S ₁₂ of the linear patterns 11, 12 of the base mark 10A, the resist The linear directions S ₁₃ and S ₁₄ of the linear patterns 13 and 14 of the mark 10B intersect each other near the center of the visual field region.
Further, a linear direction S ₁₁ of the linear patterns 11 of the base mark 10A, the linear direction S ₁₃ of the linear patterns 13 of the resist marks 10B are both biaxial orthogonal coordinate system having an optical axis 6A the origin (FIG. 2 Of the X-axis and Y-axis shown in (b), it is parallel to one of the axes (hereinafter referred to as “X-axis”) and is located near the X-axis. Similarly, the linear direction S ₁₂ of the linear patterns 12 of the base mark 10A, the linear direction S ₁₄ of the linear patterns 14 of the resist marks 10B are both parallel to the other axis (the "Y-axis"), Y It will be located near the axis. 2B is defined by the size of the imaging surface of the image sensor 31 and the magnification of the imaging system (25 to 28).

  1 is formed on the imaging surface of the imaging device 31 via the imaging system (25 to 28). In general, the influence of the distortion of the imaging system (25 to 28) is smaller as it is closer to the center of the field of view (the optical axis 6A of the objective lens 26), is larger as it is farther from the periphery, and is rotationally symmetrical in the radial direction. . Furthermore, as a result of many years of research, the inventor has noticed the fact that distortion is difficult to occur in the circumferential direction. The circumferential direction corresponds to a direction parallel to the Y axis on the X axis and a direction parallel to the X axis on the Y axis in an orthogonal coordinate system having the optical axis 6A as the origin (see FIG. 2B). It corresponds to.

  Furthermore, the fact that distortion is unlikely to occur in the circumferential direction means that there is no distortion in the Y-axis direction on the X-axis of the orthogonal coordinate system with the optical axis 6A as the origin, and there is no distortion in the vicinity of the X-axis even if it is off the X-axis. This means that the distortion in the Y-axis direction is small. Therefore, the enlarged optical image of the linear pattern 11 of the base mark 10A and the linear pattern 13 of the registration mark 10B (both parallel to the X axis) positioned in the vicinity of the X axis has an influence of distortion aberration in the Y axis direction. Can hardly be seen. These magnified optical images of the linear patterns 11 and 13 are used for detecting a positional deviation between the base mark 10A and the registration mark 10B in the Y-axis direction.

  Similarly, there is no distortion in the X-axis direction on the Y-axis of the Cartesian coordinate system with the optical axis 6A as the origin, and the distortion in the X-axis direction is small near the Y-axis even if it is off the Y-axis. Therefore, the magnified optical image of the linear pattern 12 of the base mark 10A and the linear pattern 14 of the registration mark 10B positioned in the vicinity of the Y axis (both parallel to the Y axis) is affected by distortion in the X axis direction. Can hardly be said. These enlarged optical images of the linear patterns 12 and 14 are used for detecting the positional deviation in the X-axis direction between the base mark 10A and the registration mark 10B.

  The image processing unit 32 captures the magnified optical image in a state where the magnified optical image of the misregistration detection mark 10 (that is, the magnified optical image of the linear patterns 11 to 14) is formed on the imaging surface of the image sensor 31. By taking in the image from the element 31 and performing predetermined image processing on this image, detection of the positional deviation in the X-axis direction and detection of the positional deviation in the Y-axis direction of the base mark 10A and the registration mark 10B are performed. In the image of the misregistration detection mark 10, luminance information corresponding to the edge portions of the linear patterns 11 to 14 appears independently.

  Among the edge portions of the linear patterns 11 to 14, the edge portions of the linear patterns 12 and 14 that are parallel to the Y axis and located in the vicinity of the Y axis are used for detecting the displacement in the X axis direction. Since these linear patterns 12 and 14 are located in the vicinity of the Y-axis and are hardly affected by distortion in the X-axis direction, by using the edge portions of the linear patterns 12 and 14, Misalignment detection can be performed accurately.

  Of the edge portions of the linear patterns 11 to 14, the edge portions of the linear patterns 11 and 13 that are parallel to the X axis and located in the vicinity of the X axis are used for detecting the displacement in the Y axis direction. Since these linear patterns 11 and 13 are located in the vicinity of the X axis and are hardly affected by distortion in the Y axis direction, by using the edge portions of the linear patterns 11 and 13, Misalignment detection can be performed accurately.

Next, the misregistration detection process in the image processing unit 32 will be described.
When detecting the positional deviation of the base mark 10A and the registration mark 10B in the X-axis direction, the image processing unit 32 moves from the image 35 (FIGS. 3A and 3B) of the positional deviation detection mark 10 to the Y-axis. Two partial images 36 (1) and 36 (2) related to the parallel linear pattern 12 (a part of the background mark 10A shown in FIG. 1B) are cut out, and the linear pattern 14 (parallel to the Y axis) ( Two partial images 37 (1) and 37 (2) relating to a part of the registration mark 10B shown in FIG.

The partial images 36 (1) and 36 (2) in FIG. 3A are cut out individually for each of the partial patterns 12 (1) and 12 (2) of the linear pattern 12. The partial images 37 (1) and 37 (2) in FIG. 3B are cut out so as not to include the intersections with the linear pattern 13 in the linear pattern 14.
When the segmentation of the partial images 36 (1), 36 (2), 37 (1), and 37 (2) is completed in this way, the image processing unit 32 performs the partial image 36 (1), In 36 (2), the luminance value of each pixel is integrated in the Y-axis direction (E direction) to generate a waveform signal as shown in FIG. 3C (projection processing). In FIG. 3C, the horizontal axis represents the pixel position, and the vertical axis represents the signal level (brightness). Note that the partial images 36 (1) and 36 (2) in FIG. 3A are separated from each other, but the projection processing is performed as a continuous partial image. The integration direction of projection processing (E direction in FIG. 3A) corresponds to a direction perpendicular to the direction of misalignment detection (here, the X-axis direction). Similar projection processing is also performed on the partial images 37 (1) and 37 (2) in FIG.

Thereafter, using a waveform signal (see FIG. 3C) generated from the partial images 36 (1) and 36 (2) in FIG. 3A, for example, by a well-known correlation method (folded correlation method, etc.) The autocorrelation calculation of the waveform signal is performed to calculate the center position C _{12 of the} linear pattern 12 in the X-axis direction (corresponding to the position in the X-axis direction of the linear direction S ₁₂ shown in FIG. 1B). Similarly, autocorrelation calculation is performed on the waveform signals generated from the partial images 37 (1) and 37 (2) in FIG. 3B, and the center position C _{14 in} the X-axis direction of the linear pattern ₁₄ (FIG. 1C). ) corresponds to the X-axis direction position of the straight line direction S ₁₄ shown in) is calculated.

Then, the difference between the center position C _12, C _14, positional displacement amount in the X-axis direction of the straight line direction S ₁₄ linear direction S ₁₂ and the linear pattern 14 of the linear patterns 12, that is, base mark 10A and the resist mark It is calculated as a displacement amount of 10B in the X-axis direction. This completes the process of detecting the displacement in the X-axis direction.
The detection of the positional deviation in the Y-axis direction is also performed by the same procedure as the detection of the positional deviation in the X-axis direction. In this case, the image processing unit 32 determines, for each partial pattern 11 (1), 11 (2), of the linear pattern 11 (a part of the base mark 10A) parallel to the X axis from the image of the misregistration detection mark 10. Individual partial images are cut out (see partial images 36 (1) and 36 (2) in FIG. 3A), and are linear in the linear pattern 13 (part of the registration mark 10B) parallel to the X axis. A partial image is cut out so as not to include the intersection with the pattern 14 (see the partial images 37 (1) and 37 (2) in FIG. 3B).

Then, in the two partial images related to the linear pattern 11, projection processing for integrating the luminance values of the respective pixels in the X-axis direction is performed, and the auto-correlation calculation of the obtained waveform signal results in the Y-axis direction of the linear pattern 11. Center position C ₁₁ (corresponding to the position in the Y-axis direction of the linear direction S ₁₁ shown in FIG. 1B) is calculated. In this case as well, the two partial images are separated from each other, but the projection processing is performed as a continuous partial image. Further, similar projection processing is performed on two partial images related to the linear pattern 13, and the center position C _{13 of the} linear pattern 13 in the Y-axis direction (FIG. 1 (c) is obtained by autocorrelation calculation of the obtained waveform signal. ) corresponds to the Y-axis direction position of the straight line direction S ₁₃ shown in) is calculated.

Then, the difference between the center position C _11, C _13, positional displacement amount in the Y-axis direction of the straight line direction S ₁₃ linear direction S ₁₁ and the linear pattern 13 of the linear patterns 11, that is, base mark 10A and the resist mark Calculated as the amount of positional deviation in the Y-axis direction of 10B. This completes the process of detecting the displacement in the Y-axis direction.
As described above, in the misalignment detection mark 10 of the first embodiment, the cross-shaped ground mark 10A composed of the linear patterns 11 and 12 of FIG. 1B and the linear pattern 13 of FIG. , 14 and the cross-shaped registration mark 10B, the center (that is, the center C _A of the base mark 10A and the center C _B of the registration mark 10B) is used as the center of the field of view (light By making it substantially coincide with the axis 6A), the following effects are produced.

  That is, the linear patterns 12 and 14 (both parallel to the Y axis) for detecting the positional deviation in the X axis direction are located in the vicinity of the Y axis of the orthogonal coordinate system with the optical axis 6A as the origin. The influence of distortion aberration in the X-axis direction (25 to 28) can be reduced. Further, since the linear patterns 11 and 13 (both parallel to the X axis) for detecting the displacement in the Y axis direction are located in the vicinity of the X axis of the orthogonal coordinate system having the optical axis 6A as the origin, the imaging system The influence of the distortion aberration in the Y-axis direction (25 to 28) can be reduced.

Accordingly, by using the edge portions of the linear patterns 12 and 14 located in the vicinity of the Y axis in the image of the positional deviation detection mark 10 captured at the time of detecting the positional deviation, the straight lines of the linear patterns 12 and 14 are used. The positional deviation amount in the X-axis direction in the directions S ₁₁ and S ₁₃ (that is, the positional deviation amount in the X-axis direction between the base mark 10A and the registration mark 10B) can be accurately calculated. Further, by using the edge portions of the linear patterns 11 and 13 located in the vicinity of the X axis, the amount of positional deviation in the Y axis direction of the linear patterns S ₁₁ and S ₁₃ of the linear patterns 11 and ₁₃ (that is, the base mark 10A). And the amount of misregistration of the registration mark 10B in the Y-axis direction) can be accurately calculated.

  Further, in the image of the misregistration detection mark 10, the edge portions of the linear patterns 12 and 14 extend along the Y axis in the vicinity of the Y axis. Significant image information can be ensured, and the S / N ratio of the waveform signal for detecting displacement is improved. For this reason, the position shift detection in the X-axis direction can be performed with good reproducibility. Similarly, since the edge portions of the linear patterns 11 and 13 extend along the X axis in the vicinity of the X axis, a lot of significant image information can be ensured when detecting the displacement in the Y axis direction. The signal-to-noise ratio of the waveform signal for detecting displacement is improved. For this reason, the position shift detection in the Y-axis direction can be performed with good reproducibility.

  In addition, according to the first embodiment, since the misalignment detection of the base mark 10A and the resist mark 10B in the X-axis direction and the Y-axis direction can be performed accurately and with high reproducibility, the resist pattern with respect to the base pattern on the substrate 21 can be highly accurate. Overlay inspection is possible. Specifically, overlay inspection can be performed with an accuracy of about 3 nm. Therefore, it is possible to cope with future process rules (minimum line width of the circuit pattern: 100 nm or less, overlay accuracy: about 30 nm or less) in the manufacturing process of semiconductor elements and the like.

Furthermore, since a lot of significant image information can be extracted from one image of the misalignment detection mark 10 for misalignment detection in the X-axis direction and the Y-axis direction, it is necessary to increase the number of times the image is captured from the image sensor 31. In this way, the throughput of misregistration detection is improved.
Further, when a partial image for detecting misregistration related to the linear patterns 11 and 12 (see the partial images 36 (1) and 36 (2) in FIG. 3A) is cut out from the image of the misregistration detection mark 10; In the linear pattern 11, partial images are cut out individually for each of the partial patterns 11 (1) and 11 (2). In the linear pattern 12, the partial patterns 12 (1) and 12 (2) are individually separated. Therefore, it is possible to detect misalignment satisfactorily without being affected by the edge portions of other linear patterns.

  Further, when a partial image for positional deviation detection related to the linear patterns 13 and 14 (see partial images 37 (1) and 37 (2) in FIG. 3B) is cut out from the image of the positional deviation detection mark 10; In the linear pattern 13, a partial image is cut out so as not to include an intersection portion with the linear pattern 14, and in the linear pattern 14, a partial image is cut out so as not to include an intersection portion with the linear pattern 13. It is possible to satisfactorily detect misalignment without being affected by the edge portion of the linear pattern.

Further, according to the first embodiment, the center of the viewing area (optical axis 6A), (center C of the center C _A and registration marks 10B of words underlying mark 10A _B) center of the positional shift detection mark 10 strictly Even if they are not matched, the linear patterns 11 to 14 can be positioned in the vicinity of the X axis and Y axis of the orthogonal coordinate system with the optical axis 6A as the origin, and the distortion aberration of the imaging system (25 to 28) can be determined. The impact can be reduced. Therefore, it is not necessary to use a very expensive stage 22 (for example, a positioning accuracy of about 1 μm or less), and the stage 22 can be configured at a relatively low cost.

Furthermore, in the first embodiment, since the autocorrelation calculation is performed using the entire waveform signal after the projection processing using a well-known correlation method (folded correlation method or the like), the ground mark 10A is hardly affected by signal noise. And the registration mark 10B can be calculated with high reproducibility. However, the positional deviation amount may be calculated based on the bottom position of the waveform signal instead of the correlation method.
(Second Embodiment)
As shown in FIG. 4, the misregistration detection mark of the second embodiment is obtained by dividing the linear patterns 11 and 13 into “an assembly of a plurality of fine linear patterns 38 and 39 divided in the width direction (submark group). ) ”, And the linear patterns 12 and 14 are formed of similar sub-mark groups. Note that the configuration of the submark group is not limited to all of the linear patterns 11 to 14 and may be applied to at least one linear pattern.

As in the misregistration detection mark of the second embodiment, at least one of the linear patterns 11 to 14 is configured by the sub mark group, so that polishing with good uniformity can be performed during CMP polishing. There is also an advantage that marks are not easily broken by CMP polishing.
Further, when the width of the fine linear patterns 38 and 39 of the sub-mark group is set within the resolving power of the objective lens 26, the width direction of the linear patterns 13 and 11 that are aggregates of the linear patterns 38 and 39 (positional deviation detection). There are a large number of edges with respect to (direction), and the reproducibility of misregistration detection is improved by the averaging effect. Furthermore, by making the widths of the fine linear patterns 38 and 39 approximately the same as the line width of the circuit pattern, the accuracy of the positional deviation detection is improved.
(Third embodiment)
As shown in FIG. 5, the misregistration detection mark of the third embodiment has a base mark 10 </ b> A and a registration mark 10 </ b> B (that is, the linear pattern 11) inside the rectangular area 40 that circumscribes the linear pattern 11 and the linear pattern 12. To 14), in the four regions a to d, a ground mark (41 to 44) indicating another reference position of the ground pattern, and a resist mark (44 to 48) indicating another reference position of the resist pattern, Are arranged so as not to overlap. The new base marks (41 to 44) and the resist marks (44 to 48) are composed of a line and space pattern (grating structure).

In the misregistration detection mark of the third embodiment configured as described above, other base marks (41 to 44) and registration marks (44) are efficiently divided into four areas a to d inside the rectangular area 40. ˜48) can be arranged, two kinds of marks can be included without increasing the occupation area.
Further, according to the misregistration detection mark of the third embodiment, the center mark C _A (FIG. 1B) of the base mark 10A composed of the linear patterns 11 and 12 and the resist mark 10B composed of the linear patterns 13 and 14 are used. In the state where the center C _B (FIG. 1 (c)) is substantially coincident with the center of the field of view (optical axis 6A), the positional deviation between the base mark 10A and the registration mark 10B is detected, and the base marks (41-44) are detected. ) And registration marks (44 to 48).

  The actual misalignment detection may be performed using both the cross-shaped marks (linear patterns 11 to 14) and the line-and-space marks (41 to 48). It may be selected at the time of inspection. The advantage of the cross-shaped mark is that since the vicinity of the optical axis is used, the position shift can be detected with high accuracy without being substantially affected by the distortion of the imaging system (25 to 28). The advantage of the line-and-space mark is that an average detection result can be obtained. By selecting the optimum one of the two types, the accuracy of detecting the displacement is improved.

In the third embodiment described above, the other base marks (41 to 44) and the resist marks (44 to 48) are arranged in four regions a to d inside the rectangular region 40. However, the present invention is not limited to this. It is not limited. A mark indicating at least one other reference position (at least one of a base mark and a resist mark) may be arranged in at least one of the four regions a to d.
(Fourth embodiment)
As shown in FIG. 6, the misregistration detection mark of the fourth embodiment has a base mark 10 </ b> A and a registration mark 10 </ b> B (that is, the linear pattern 11) inside the rectangular area 50 that circumscribes the linear pattern 11 and the linear pattern 12. 14), the process information 51 related to the formation of the base pattern, the dummy patterns 52 and 53 formed simultaneously with the base pattern, and the process information 54 related to the formation of the resist pattern The dummy patterns 55 and 56 formed simultaneously with the resist pattern are arranged. The process information 51, 54 is a reticle number or the like.

The positional shift detection mark of the fourth embodiment configured as described above, consists of the center C _A (to FIG. 1 (b)) and a linear pattern 13 and 14 of the base mark 10A consisting of linear patterns 11 and 12 In addition to being able to detect misalignment between the base mark 10A and the registration mark 10B in a state where the center C _B (FIG. 1C) of the registration mark 10B is substantially coincident with the center of the visual field area (optical axis 6A). By reading and collating the process information 51, 54, it is possible to recognize an exposure reticle error or the like. Further, by providing the dummy patterns 52, 53, 55, and 56, CMP polishing can be performed under the same uniform conditions as other portions.

In the fourth embodiment described above, the process information 51 and 54 related to the formation of the base pattern and the resist pattern are arranged in the areas a and b among the four areas a to d inside the rectangular area 50. The invention is not limited to this. Process information relating to formation of one of the base pattern and the resist pattern may be arranged in at least one of the four regions a to d.
(Fifth embodiment)
As shown in FIG. 7, the misregistration detection mark of the fifth embodiment has a base mark 10A and a registration mark 10B (that is, the linear pattern 11) inside a rectangular area 60 circumscribing the linear pattern 11 and the linear pattern 12. To 14), inspection marks 61 to 64 for exposure conditions relating to the formation of at least one of the base pattern and the resist pattern are arranged in the four regions a to d. The marks 61 to 64 are wedge-shaped SMP (Self Measurement Program) marks, and their lengths change according to exposure conditions (dose amount, focus shift amount, etc.).

The positional shift detection mark of the fifth embodiment configured as described above, consists of the center C _A (to FIG. 1 (b)) and a linear pattern 13 and 14 of the base mark 10A consisting of linear patterns 11 and 12 In addition to being able to detect misalignment between the base mark 10A and the registration mark 10B in a state where the center C _B (FIG. 1C) of the registration mark 10B is substantially coincident with the center of the visual field area (optical axis 6A). The exposure conditions (dose amount, focus shift amount, etc.) can be inspected according to the amount of change in the length of the marks 61-64.

In the fifth embodiment described above, the marks 61 to 64 are arranged in the four regions a to d inside the rectangular region 60, but the present invention is not limited to this. Similar SMP marks may be arranged in at least one of the four regions a to d.
(Sixth embodiment)
As shown in FIG. 8, the misregistration detection mark 70 of the sixth embodiment is provided with a registration mark 70B instead of the registration mark 10B of the misregistration detection mark 10 of FIG. In the registration mark 70B, a linear pattern 71 parallel to the linear pattern 11 and a linear pattern 72 parallel to the linear pattern 12 are arranged in a cross shape. Also, the linear patterns 71 and 72 is shorter in length between both ends than the linear patterns 11 and 12, so as not to intersect with each other, respectively divided into the one end and the other end of the linear direction S _71, S ₇₂ The two partial patterns 71 (1), 71 (2), 72 (1) and 72 (2) are formed.

  According to the above-described misregistration detection mark 70, when misregistration is detected, partial images of the linear patterns 71 and 72 of the registration mark 70B are obtained from the misregistration detection mark 70 image shown in FIG. Can be easily cut out. For example, in the case of the linear pattern 71, one continuous partial image 73 may be cut out from the partial pattern 72 (1) to the partial pattern 72 (2). The same applies to the case of the linear pattern 72, and one partial image continuous from the partial pattern 71 (1) to the partial pattern 71 (2) may be cut out.

Since the linear patterns 71 and 72 do not intersect with each other, even if one continuous partial image is cut out, it is possible to detect misalignment satisfactorily without being affected by the edge portions of other linear patterns.
(Modification)
In the above-described embodiment, the resist pattern overlay inspection (position shift detection) with respect to the base pattern formed on a different layer of the substrate 21 has been described as an example. However, the present invention is not limited to this. The present invention can also be applied to the case where positional deviation detection of two patterns formed on the same layer of the substrate 21 is performed.

  Furthermore, although the example which performs the process of position shift detection by the image processing unit 32 of the overlay measurement apparatus 20 has been described, the present invention is not limited to this. Even when an external computer connected to the overlay measuring apparatus is used, the same effect can be obtained.

It is a figure explaining the composition of mark 10 for position shift detection of a 1st embodiment. 2 is a configuration diagram of an overlay measurement apparatus 20. FIG. It is a figure explaining the partial image 36 (1), 36 (2), 37 (1), 37 (2) for position shift detection, and the waveform signal after a projection process. It is a figure explaining the structure of the misalignment detection mark of 2nd Embodiment. It is a figure explaining the structure of the misalignment detection mark of 3rd Embodiment. It is a figure explaining the structure of the mark for position shift detection of 4th Embodiment. It is a figure explaining the structure of the misalignment detection mark of 5th Embodiment. It is a figure explaining the structure of the position shift detection mark of 6th Embodiment. It is a block diagram of the conventional double mark 80. FIG.

Explanation of symbols

10, 70 Position shift detection mark 10A, 41-44 Base mark 10B, 45-48, 70B Registration mark 11, 12, 13, 14, 71, 72 Linear pattern 11 (1), 11 (2), 12 ( 1), 12 (2), 71 (1), 71 (2), 72 (1), 72 (2) Partial pattern 20 Overlay measuring device 21 Substrate 22 Stage 23 Light source 24 Illumination lens 26 Objective lens 6A Optical axis 27 , 29 Imaging lens 30 Detector 31 Image sensor 32 Image processing unit 33 System control unit 34 Stage control unit 35 Image 36 (1), 36 (2), 37 (1), 37 (2), 73 Partial image 38, 39 Fine linear pattern 40, 50, 60 Rectangular area 51, 54 Process information 52, 53, 55, 56 Dummy pattern 61-64 Inspection mark for exposure conditions

Claims (6)

A misregistration detection method for detecting misregistration of the two patterns using misregistration detection marks used for misregistration detection of two patterns,
The misregistration detection mark is
A first mark indicating one reference position of the two patterns, and a second mark indicating the other reference position of the two patterns,
The first mark has a first linear pattern and a second linear pattern perpendicular to the linear pattern arranged in a cross shape,
In the second mark, a third linear pattern and a fourth linear pattern perpendicular to the third linear pattern are arranged in a cross shape,
When the first mark and the second mark have no positional deviation between the two patterns, the linear directions of the first linear pattern and the third linear pattern coincide with each other, and the second mark And the linear direction of the fourth linear pattern coincides,
The first linear pattern and the third linear pattern have different lengths between both ends in the longitudinal direction,
Of the first linear pattern and the third linear pattern, the linear pattern having the longer length does not overlap the linear pattern having the shorter length so as to overlap with the linear pattern having the shorter length. It consists of two partial patterns divided into the side and the other end side,
The second linear pattern and the fourth linear pattern have different lengths between both ends in the longitudinal direction,
Of the second linear pattern and the fourth linear pattern, the longer linear pattern has one end in a linear direction so as not to overlap the shorter linear pattern. It consists of two partial patterns divided into the side and the other end side,
The positional deviation detection method is:
A first step of capturing images of the first mark and the second mark;
A partial image related to the first linear pattern and the second linear pattern of the first mark is cut out from the captured image, and the third linear pattern and the fourth mark of the second mark are cut out. A second step of cutting out a partial image related to the linear pattern;
Based on the cut out partial image, the amount of positional deviation in the linear direction of the first linear pattern and the third linear pattern is calculated, and the second linear pattern and the fourth line are calculated. A third step of calculating the amount of positional deviation of the linear pattern in the linear direction,
In the second step, when cutting out a partial image related to the longer linear pattern of the first linear pattern and the third linear pattern, and the second linear pattern And the fourth linear pattern, when the partial image related to the longer linear pattern is cut out, the partial image is cut out separately for each of the partial patterns. Method. The positional deviation detection method according to claim 1,
Of the first linear pattern and the third linear pattern in the displacement detection mark, the shorter linear pattern, the second linear pattern, and the fourth linear pattern. linear pattern of the shorter of the length of the pattern, so as not to intersect with each other, each positional displacement, characterized in that it consists of two partial patterns that are divided into a one end and the other end of the straight line direction detection Way . In the positional deviation detection method according to claim 1 or 2,
Of the first linear pattern, the second linear pattern, the third linear pattern, and the fourth linear pattern in the misregistration detection mark , at least one linear pattern has a width. A misregistration detection method comprising a plurality of fine linear patterns divided in a direction. In the position shift detection method according to any one of claims 1 to 3,
Out of the first linear pattern and the third linear pattern in the misregistration detection mark, it circumscribes the longer linear pattern, and the second linear pattern and the fourth linear pattern. Among the four regions divided by the first mark and the second mark, at least one region is inside the rectangular region circumscribing the longer one of the linear patterns. A misregistration detection method comprising a third mark indicating at least one other reference position of the two patterns. In the position shift detection method according to any one of claims 1 to 3,
Out of the first linear pattern and the third linear pattern in the misregistration detection mark, it circumscribes the longer linear pattern, and the second linear pattern and the fourth linear pattern. Among the four regions divided by the first mark and the second mark, at least one region is inside the rectangular region circumscribing the longer one of the linear patterns. A process for detecting misalignment , comprising process information relating to formation of at least one of the two patterns. In the position shift detection method according to any one of claims 1 to 3,
Out of the first linear pattern and the third linear pattern in the misregistration detection mark, it circumscribes the longer linear pattern, and the second linear pattern and the fourth linear pattern. Among the four regions divided by the first mark and the second mark, at least one region is inside the rectangular region circumscribing the longer one of the linear patterns. A misregistration detection method comprising: an inspection mark for an exposure condition related to formation of at least one of the two patterns.

Images