JP4957027B2 - Overlay measuring device - Google Patents

Description

  The present invention relates to an overlay measurement apparatus that measures the overlay state of each layer of a substrate.

In the manufacturing process of a semiconductor element, a liquid crystal display element, etc., the overlay state of each layer of the substrate is measured. The overlapping state of each layer is an important parameter that affects the yield of elements and the performance of a completed device, and an apparatus for measuring the overlapping state of each layer is recognized as an important manufacturing apparatus.
In the overlay measurement apparatus, for example, an image of each mark formed on the base layer and the resist layer of the substrate is captured by an imaging means such as a camera, and this image is processed to determine the overlay displacement amount of each mark ( For example, see Patent Document 1).

When processing an image, after setting the measurement direction of the overlay displacement amount according to the coordinate axis on the image, the straight line portion extending in the non-measurement direction perpendicular to the measurement direction from the straight line portion of each mark appearing in the image In general, the change in brightness in the measurement direction of this straight line portion is integrated in the non-measurement direction (projection processing) to reduce signal noise. Then, the overlay deviation amount is obtained using the signal waveform of each mark obtained by the projection processing.
JP 2004-108957 A

  However, the above apparatus has the following problems due to the projection processing. In other words, when the straight line portion selected when processing the image is not perpendicular to the measurement direction (slightly tilted), the projection processing is performed if the contrast of the change in brightness in the measurement direction is not constant in the selected straight line portion. The signal waveform obtained by the above becomes asymmetrical, and an error may occur in the overlay deviation amount. Note that the inclination of the straight line portion described above corresponds to an image rotation error, and is caused by an attachment error of the imaging unit. Further, the non-uniformity of the contrast of the straight line portion is caused by the non-uniformity of the reflectance of the mark on the substrate.

  The object of the present invention is to perform overlay measurement that can accurately determine the amount of misalignment without being affected by the inclination or contrast non-uniformity of the straight line selected from the image even when signal noise is reduced by projection processing. To provide an apparatus.

An overlay measuring apparatus according to a first aspect of the present invention is an apparatus for taking in images of first marks and second marks formed on different layers of a substrate, and among straight portions of the first marks appearing in the images. First processing means for generating, as a first signal waveform, a signal waveform related to a change in brightness when the first straight line section is crossed in the measurement direction based on a first straight line section that intersects a predetermined measurement direction. Based on the second straight line portion that intersects the measurement direction among the straight line portions of the second mark that appear in the image, a signal waveform relating to a change in brightness when the second straight line portion is crossed in the measurement direction. Using the second processing means for generating the second signal waveform, and the first signal waveform and the second signal waveform, the overlay deviation amount in the measurement direction between the first mark and the second mark is determined. Third processing means to be obtained, The first processing means extracts a light / dark change when the first straight portion is crossed in the measurement direction at a plurality of different locations on the first straight portion, and the light / dark change is perpendicular to the measurement direction. The second signal processing unit crosses the second straight line portion in the measurement direction at a plurality of different locations of the second straight line portion, respectively. The brightness change at the time is extracted, the contrast of the brightness change is aligned with each other, and then the brightness change is integrated in the vertical direction to generate the second signal waveform.
In a second aspect based on the first aspect, the second processing means determines a contrast magnitude of the light / dark change of the second straight line portion, and aligns the contrast of the light / dark change based on the determination. Determine whether or not.
In a third aspect based on the second aspect, the second processing means aligns the contrast of the plurality of portions of the second straight line portion to an average value.

According to a fourth aspect of the present invention, there is provided an overlay measurement apparatus including: a capturing unit that captures images of a first mark and a second mark formed on different layers of a substrate; and a linear portion of the first mark that appears in the image. First processing means for generating, as a first signal waveform, a signal waveform related to a change in brightness when the first straight line section is crossed in the measurement direction based on a first straight line section that intersects a predetermined measurement direction. Based on the second straight line portion that intersects the measurement direction among the straight line portions of the second mark that appear in the image, a signal waveform relating to a change in brightness when the second straight line portion is crossed in the measurement direction. Using the second processing means for generating the second signal waveform, and the first signal waveform and the second signal waveform, the overlay deviation amount in the measurement direction between the first mark and the second mark is determined. Third processing means to be obtained, The first processing means extracts a change in brightness when the first straight line portion is crossed in the measurement direction at a plurality of different locations of the first straight line portion, and aligns the contrast of the light-dark change with each other. Thereafter, the change in brightness is integrated in a direction perpendicular to the measurement direction to generate the first signal waveform, and the second processing means is configured to respectively apply the plurality of different portions of the second straight line portion to the second linear portion. After extracting the change in brightness when the second straight line section is crossed in the measurement direction, and aligning the contrast of the brightness change with each other, the brightness change is integrated in the vertical direction to obtain the second signal waveform. Is to be generated.
In a fifth aspect based on the fourth aspect, the first processing means determines a contrast magnitude of the brightness change of the first straight line portion, and the second processing means is the second straight line portion. The magnitude of the contrast of the light / dark change is determined, and based on the determination of the first and second processing means, it is determined whether or not the contrast of the light / dark change of the first and second linear portions is to be made uniform.
In a sixth aspect based on the fifth aspect, the first processing means aligns the contrasts of the plurality of portions of the first straight line portion to an average value, and the second processing means includes the second straight line portion. The contrast of the plurality of locations is made equal to the average value.
According to a seventh invention, in any one of the first to sixth inventions, at least one of the first mark and the second mark is in the direction of integration in the first and second marks. The contrast is different.
In an eighth invention according to any one of the first to seventh inventions, at least one of the first mark and the second mark intersects the integrating direction.
According to a ninth aspect, in any one of the first to eighth aspects, the plurality of different portions of the first and second linear portions are set along the direction of integration.

An overlay measuring apparatus according to a tenth aspect of the present invention is an apparatus for taking in images of a first mark and a second mark formed on different layers of a substrate, and a linear portion of the first mark that appears in the image. First processing means for generating, as a first signal waveform, a signal waveform related to a change in brightness when the first straight line section is crossed in the measurement direction based on a first straight line section that intersects a predetermined measurement direction. Based on the second straight line portion that intersects the measurement direction among the straight line portions of the second mark that appear in the image, a signal waveform relating to a change in brightness when the second straight line portion is crossed in the measurement direction. Using the second processing means for generating the second signal waveform, and the first signal waveform and the second signal waveform, the overlay deviation amount in the measurement direction between the first mark and the second mark is determined. Third processing means to be obtained The first processing means extracts a light / dark change when the first straight part is crossed in the measurement direction at a plurality of different locations on the first straight part, and the light / dark change is perpendicular to the measurement direction. The second signal processing unit crosses the second straight line portion in the measurement direction at a plurality of different locations of the second straight line portion, respectively. And the third signal processing unit generates the second signal waveform by integrating the light and dark changes in the vertical direction, and the third processing unit includes the first signal waveform and the second signal waveform. After determining the amount of overlay deviation using, based on the contrast of each light and dark change extracted from the first straight line portion and the contrast of each light and dark change extracted from the second straight line portion, It corrects the amount of misalignment. .

  According to the overlay measurement apparatus of the present invention, even when signal noise is reduced by projection processing, the overlay displacement amount can be accurately obtained without being affected by the inclination of the straight line portion selected from the image or the unevenness of contrast. Can do.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
(First embodiment)
The overlay measurement apparatus 10 (FIG. 1) of this embodiment is an apparatus that measures the overlay state of each layer of the substrate 11 in the manufacturing process of a semiconductor element, a liquid crystal display element, and the like. The substrate 11 is a semiconductor wafer, a liquid crystal substrate, or the like, and is in a state after exposure / development on the resist layer and before processing (etching process) on a predetermined material film. A large number of measurement points are prepared on the substrate 11 for overlay inspection, and an overlay mark 11A shown in FIG. 2 is formed at each measurement point. 2A is a plan view, and FIG. 2B is a cross-sectional view.

  The overlay mark 11A is a double mark, and includes a square-shaped resist mark 31 indicating the reference position of the resist layer, and a square-shaped base mark 32 indicating the reference position of the base layer. The resist mark 31 and the base mark 32 are formed on different layers of the substrate 11 and have different sizes. In the measurement of the overlay state (overlay inspection), the position of each mark is detected, and the overlay deviation amount between the base mark 32 and the registration mark 31 is calculated.

  As shown in FIG. 1, the overlay measurement apparatus 10 of the present embodiment includes a stage 12 that supports a substrate 11, illumination optical systems (13 to 19), an imaging optical system (18 to 25), and a camera 26. A control unit 27 and a pupil division type focus detection unit 28 are provided. Details of the pupil division type focus detection unit 28 are described in, for example, Japanese Patent Application Laid-Open No. 2002-40322, and thus the description thereof is omitted here.

  The stage 12 moves the substrate 11 in the horizontal direction based on the overlay inspection recipe, and sequentially positions the overlay marks 11A (FIG. 2) at the respective measurement points on the substrate 11 within the field of view (for example, the center of the field of view). To do. In addition, the stage 12 moves the substrate 11 in the vertical direction based on the same recipe and the focus signal from the focus detection unit 28, so that the overlay mark 11 </ b> A positioned in the visual field region is placed on the imaging surface of the camera 26. To focus.

  The illumination optical system (13 to 19) includes a light source unit 13, an aperture stop 14, a condenser lens 15, a field stop 16, an illumination relay lens 17, a beam splitter 18, and a first objective lens 19. The The light source unit 13 includes a light source 3A that emits light having a wide wavelength band (for example, white light), a wavelength switching mechanism 3B, a lens 3C, and a light guide fiber 3D.

By switching the optical filter of the wavelength switching mechanism 3B and inserting it into the illumination optical path, it is possible to select light in a predetermined wavelength range corresponding to the transmission characteristics of the optical filter from light having a broad wavelength emitted from the light source 3A. it can. The light after passing through the optical filter of the wavelength switching mechanism 3B is guided to the aperture stop 14 via the lens 3C and the light guide fiber 3D.
The aperture stop 14 limits the diameter of the light emitted from the light source unit 13 to a specific diameter. The condenser lens 15 condenses the light from the aperture stop 14. The field stop 16 is an optical element that limits the field area of the overlay measurement apparatus 10. The illumination relay lens 17 collimates the light from the field stop 16. The beam splitter 18 reflects light from the illumination relay lens 17 downward (toward the first objective lens 19).

  In the above configuration, the light from the light source unit 13 uniformly illuminates the field stop 16 via the aperture stop 14 and the condenser lens 15. The light passing through the field stop 16 is guided to the beam splitter 18 via the illumination relay lens 17, reflected by the reflection / transmission surface thereof, and then guided to the first objective lens 19. The first objective lens 19 receives light from the beam splitter 18 and focuses it on the substrate 11 on the stage 12.

  As a result, the overlay mark 11A (FIG. 2) positioned in the field of view within the substrate 11 is vertically illuminated by the light in the predetermined wavelength range that has passed through the first objective lens 19 (epi-illumination). Then, diffracted light is generated from the illumination region of the substrate 11. Diffracted light includes zero-order diffracted light (that is, reflected light), ± first-order diffracted light, and the like. Diffracted light from the substrate 11 is guided to an imaging optical system (18 to 25) described later.

In addition to the beam splitter 18 and the first objective lens 19, the imaging optical system (18 to 25) includes a second objective lens 20, a beam splitter 21, a field stop 22, a first imaging relay lens 23, and the like. The aperture stop 24 and the second imaging relay lens 25 are configured. The beam splitters 18 and 21 are half prisms that perform light amplitude separation.
The diffracted light generated from the substrate 11 is collimated by the first objective lens 19, passes through the beam splitter 18, and enters the second objective lens 20. The second objective lens 20 condenses the diffracted light from the beam splitter 18. The beam splitter 21 transmits a part of the diffracted light from the second objective lens 20 and reflects the remaining part. The light transmitted through the beam splitter 21 is guided to the focus detection unit 28.

  The light reflected by the beam splitter 21 is condensed on the field stop 22 of the imaging optical system (18-25) and guided to the first imaging relay lens 23. The first imaging relay lens 23 collimates the light from the field stop 22. The aperture stop 24 is disposed on a plane conjugate with the virtual pupil plane of the first objective lens 19 and limits the diameter of light from the first imaging relay lens 23 to a specific diameter. The second imaging relay lens 25 re-images the light from the aperture stop 24 on the imaging surface of the camera 26.

In the imaging optical system (18 to 25), when the overlay mark 11A (FIG. 2) of the substrate 11 is positioned in the visual field region, an optical image of the mark is formed on the imaging surface of the camera 26.
The camera 26 is arranged so that its imaging surface coincides with the image plane of the imaging optical system (18-25). The camera 26 includes an area sensor (for example, a CCD sensor) in which a plurality of pixels are two-dimensionally arranged, captures an optical image of the overlay mark 11A on the substrate 11, and outputs an image signal to the control unit 27. The image signal represents a distribution (luminance distribution) related to the luminance value for each pixel of the optical image on the imaging surface of the camera 26.

  The control unit 27 captures an image (two-dimensional image) of the overlay mark 11A on the substrate 11 based on an image signal from the camera 26 during overlay inspection, and performs overlay inspection on the image. Apply image processing. That is, the position of each of the registration mark 31 and the background mark 32 constituting the overlay mark 11A is detected, and the overlay deviation amount between the registration mark 31 and the background mark 32 is calculated.

  This overlay displacement amount is a relative displacement amount of the center of gravity position of each mark, and represents the overlay displacement amount of the resist layer with respect to the base layer on the substrate 11. This overlay deviation amount is also referred to as a “superposition measurement value”. After calculation of the overlay deviation amount, it is preferable to correct a measurement error caused by the substrate called WIS (Wafer Induced Shift) to obtain a value close to the actual overlay deviation amount after the etching process.

Next, a specific example of a procedure for processing an image of the overlay mark 11A (hereinafter, “inspection image”) will be described.
For example, as shown in FIG. 3A, straight lines 41 to 48 having lower luminance than the background appear in the inspection image. Each of these linear portions 41 to 48 corresponds to each edge portion of the overlay mark 11A (FIG. 2A). In the inspection image, the straight portions 41 to 44 of the registration mark 31 and the straight portions 45 to 48 of the base mark 32 appear at different positions.

  Then, the processing for the inspection image (FIG. 3A) is performed for each coordinate axis (x axis, y axis) on the image. That is, when an image is processed, the measurement direction of the overlay deviation amount is set according to the coordinate axes (x axis, y axis) on the image, and the process is performed according to this measurement direction. Of the coordinate axes (x axis, y axis), the direction of the x axis corresponds to, for example, the direction of the scanning line. The direction of the y axis corresponds to, for example, a direction perpendicular to the scanning line.

  When the x-axis direction is set as the measurement direction, the signal processing ranges of the straight portions 42, 44, 46, and 48 are respectively set by the x-axis measurement frames 51 to 54 shown in FIG. For example, a known correlation operation is performed using the luminance value of each pixel included in the signal processing range. Furthermore, when the direction of the y-axis is the measurement direction, the signal processing ranges of the linear portions 41, 43, 45, and 47 are respectively set by the y-axis measurement frames 54 to 58 shown in FIG. Designate and perform the same correlation calculation in the signal processing range. Hereinafter, the case where the x-axis direction is the measurement direction will be described.

  In the measurement in the x-axis direction, among the linear portions 41 to 44 of the registration mark 31 appearing in the inspection image (FIG. 3A), a non-measurement direction (vertical direction) perpendicular to a predetermined measurement direction (x-axis direction) ( The straight portions 42 and 44 extending in the y-axis direction) are selected, and the measurement frames 51 and 52 (FIG. 3B) are set in the straight portions 42 and 44. Next, in the measurement frames 51 and 52, the change in brightness when the straight portions 42 and 44 are crossed in the measurement direction is integrated in the non-measurement direction (projection process), and a registration mark as shown in FIG. 31 signal waveforms 31L and 31R are generated.

  Similarly, the straight portions 46 and 48 extending in the non-measurement direction (y-axis direction) are selected from the straight portions 45 to 48 of the base mark 32 appearing in the inspection image (FIG. 3A). Measurement frames 53 and 54 (FIG. 3B) are set on the straight portions 46 and 48, respectively. Next, in the measurement frames 53 and 54, the light and dark changes in the measurement direction of the straight portions 46 and 48 are integrated in the non-measurement direction (projection processing), and the signal waveform of the ground mark 32 as shown in FIG. 32L and 32R are generated.

Thereafter, a superposition correlation between the signal waveforms 31L and 31R (FIG. 4A) of the registration mark 31 and its self-reversal waveform is obtained, and the position where the correlation value is maximized is the barycentric position C of the registration mark 31 in the x-axis direction. _Set to ₁ . Underlying mark 32 of the signal waveform 32L, operation to 32R (FIG. 4 (b)) is also performed in the same manner, as a result, the center-of-gravity position C ₂ in the x-axis direction of the underlying mark 32 is detected. Then, using the result of position detection, an overlay deviation amount L in the measurement direction between the registration mark 31 and the base mark 32 is calculated.

  Such processing is performed when the linear portions 41 to 48 appearing there are horizontal or vertical with respect to the coordinate axes (x axis, y axis) on the image, as in the inspection image shown in FIG. Therefore, it is possible to accurately obtain the overlay deviation amount L. However, the image as shown in FIG. 3A is obtained in an ideal case where there is no mounting error of the camera 26 and the camera 26 is not inclined with respect to the substrate 11.

However, in an actual apparatus, since the mounting error of the camera 26 is not zero, the camera 26 is inclined with respect to the substrate 11. The tilt of the camera 26 is a tilt in the rotation direction about the optical axis of the camera 26 (that is, the optical axis of the imaging optical system (18 to 25)).
Then, due to the influence of the inclination, a rotation error δθ as shown in FIG. 5A occurs in the inspection image, and the linear portions 41 to 48 are present with respect to the coordinate axes (x axis, y axis). Will tilt. The rotation error δθ is equal to the tilt angle of the camera 26 with respect to the substrate 11 and is, for example, about several minutes. In FIG. 5 (a), the rotation error δθ is shown larger than the actual value for easy understanding.

The rotation error δθ (FIG. 5A) of the inspection image is one of the factors of the machine difference of the overlay measurement apparatus 10 (that is, the measurement error caused by the apparatus adjustment error). The overlay measuring apparatus 10 is under development to eliminate machine difference factors as much as possible, but there is a limit to reducing the rotation error δθ.
Therefore, in the overlay measurement apparatus 10 of the present embodiment, a rotation error δθ as shown in FIG. 5A is generated in the inspection image, and the linear portions 41 to 48 are relative to the coordinate axes (x axis, y axis). Even in the case of tilting, measurement is performed without being affected by this, and the measurement error due to the rotation error δθ of the image (error of the overlay deviation amount L) can be reduced.

  However, the magnitude of the measurement error due to the image rotation error δθ depends on the nonuniformity of the reflectance of the overlay mark 11A on the substrate 11, and the like. As a tendency, the non-uniformity of the reflectance of the overlay mark 11A is small, and in the measurement frames 51 to 54 shown in FIG. 5B (similar to FIG. 3B), the straight portions 42, 44, 46, If the contrast difference between the light and dark changes in the 48 measurement directions is small, the measurement error due to the rotation error δθ can also be suppressed.

When this state is illustrated and described, the change in brightness in the measurement direction of the straight portions 46 and 48 in the measurement frames 53 and 54 is determined for each position y _{1 to} y _{3 in the} non-measurement direction shown in FIG. For example, as shown in FIG. Here, the position y ₂ is the center of the measurement frames 53 and 54. In FIG. 6, the change in brightness at each of the positions y _{1 to} y ₃ is shifted in the measurement direction (positions x _{1 to} x ₃ ) reflecting the inclination (rotational error δθ) of the straight portions 46 and 48. The contrasts C _O (y ₁ ) to C _O (y ₃ ) have almost no difference.

In such a case, when the signal waveform (see the signal waveforms 32L and 32R in FIG. 4B) is generated by integrating the light-dark changes in the measurement direction (FIG. 6) in the non-measurement direction by the projection processing described above, it was obtained. The center-of-gravity position C ₂ obtained from the signal waveform by correlation calculation or the like coincides with the center-of-gravity position C ₂ (y ₂ ) obtained in the same manner as a provisional signal waveform with the change in brightness at the center (position y ₂ ) of the measurement frames 53 and 54. Will do.

The same applies to the straight portions 42 and 44 in the measurement frames 51 and 52. As shown in FIG. 7, the change in brightness in the measurement direction at each position y ₄ , y ₂ and y ₅ is the straight portions 42 and 44. Is reflected in the measurement direction (positions x _{4 to} x ₆ ) reflecting the inclination of the image (rotation error δθ), and the contrasts C _i (y ₄ ), C _i (y ₂ ), C _i (y ₅ ) are almost different. If not, the center-of-gravity position C ₁ obtained from the signal waveform after integration by projection processing (see the signal waveforms 31L and 31R in FIG. 4A) and the change in brightness at the center (position y ₂ ) of the measurement frames 51 and 52 Centroid position C ₁ (y ₂ ) determined as a temporary signal waveform.

  Therefore, if the non-uniformity of the reflectance of the overlay mark 11A is small and the contrast difference of the change in brightness in the measurement direction of the linear portions 42, 44, 46, 48 is small in the measurement frames 51 to 54, the integration by the projection process is performed. By performing correlation calculation using the subsequent signal waveforms (FIGS. 4A and 4B), the overlay deviation amount L can be accurately obtained. That is, the measurement error due to the rotation error δθ of the inspection image can be kept small.

On the other hand, the nonuniformity of the reflectance of the overlay mark 11A on the substrate 11 is large, and the straight portions 42 and 44 and / or the straight portions 46 and 48 in the measurement frames 51 to 54 (FIG. 5B). If the contrast difference between light and dark changes in the measurement direction is large, the measurement error due to the rotation error δθ also increases reflecting this contrast difference.
For example, the contrast change in the measurement direction of the straight portions 42 and 44 in the measurement frames 51 and 52 has a very small difference in contrast as in FIG. 7, and the measurement direction of the straight portions 46 and 48 in the measurement frames 53 and 54. Let us consider a case where the contrast difference is large in the light-dark change.

In this case, the change in brightness in the measurement direction of the straight portions 46 and 48 in the measurement frames 53 and 54 is, for example, as shown in FIG. 8 for each position y _{1 to} y _{3 in the} non-measurement direction shown in FIG. become. In FIG. 8, the change in brightness at each of the positions y _{1 to} y ₃ is shifted in the measurement direction (positions x _{1 to} x ₃ ) reflecting the inclination (rotation error δθ) of the straight portions 46 and 48, and the contrast C _O (y ₁ ) to C _O (y ₃ ) are greatly different.

When such light-dark changes (FIG. 8) are integrated in the non-measurement direction as they are (projection processing) to generate signal waveforms (similar to the signal waveforms 32L and 32R in FIG. 4B), the obtained signal waveforms are left and right. becomes asymmetric, the center-of-gravity position C ₂ as determined by such a correlation calculation from the signal waveform, the center of gravity was determined in the same manner the light-dark change in the central (position y ₂₎ of the measurement area 53 as the temporary signal waveform position C ₂ ( y ₂ ) no longer matches.

That is, the barycentric position C ₂ obtained from the signal waveform after the integration by the projection processing uses a change in light and dark at a position (eg, position y ₃ ) deviated from the center (position y ₂ ) of the measurement frames 53 and 54 as a temporary signal waveform. Similarly, it coincides with the obtained center-of-gravity position C ₂ (y ₃ ).
Therefore, as described above, if the contrast difference of the change in brightness in the measurement direction of the linear portions 42 and 44 in the measurement frames 51 and 52 is small (FIG. 7), the center of the measurement frames 53 and 54 in FIG. The center-of-gravity position C ₂ (y ₂ ) obtained from the change in brightness at the position y ₂ ) as a tentative signal waveform, and the change in brightness at a position (for example, position y ₃ ) deviated from the center (position y ₂ ) of the measurement frames 53 and 54. Is a difference δx (= C ₂ (y ₃ ) −C ₂ (y ₂ )) from the center of gravity position C ₂ (y ₃ ) obtained as a temporary signal waveform is an error of overlay deviation L (hereinafter referred to as “measurement error”). δx ”).

  If the linear portions 41 to 48 have a high contrast in the inspection image (FIG. 5A), the contrast difference in the measurement frames 51 to 58 is relatively small, and the measurement error δx is small. . On the contrary, if the linear portions 41 to 48 are low contrast, the contrast difference is relatively large, and the measurement error δx is also large. As described above, the difference in contrast in the straight line portion of each mark can cause an uncertain error factor.

  Here, the ITRS 2004 edition (International Technology Roadmap for Semiconductors 2004 Edition) requires 2.3 nm as total measurement accuracy (TMU) for a half pitch of 65 nm, which is expected to start mass production in 2007. In this case, the total measurement accuracy (TMU) is defined as the following equation (1), and the machine difference required for the overlay measurement apparatus 10 is 1 nm or less.

本実施形態の重ね合わせ測定装置１０では、上記のような測定誤差δｘを低減して正確に（つまり機差１nm以下で）重ね合わせずれ量Ｌを求めるために、次のような手順で検査用の画像（図５(ａ)）を処理する。ここでも、ｘ軸方向を測定方向とする例で説明する。

In the overlay measurement apparatus 10 of the present embodiment, in order to accurately determine the overlay error L by reducing the measurement error δx as described above (that is, with a machine difference of 1 nm or less), the overlay measurement apparatus 10 is used for inspection in the following procedure. The image (FIG. 5A) is processed. Here again, an example in which the x-axis direction is the measurement direction will be described.

Of the straight portions 41 to 44 of the registration mark 31 that appear in the inspection image (FIG. 5A), the straight portions 42 and 44 that intersect the measurement direction (x-axis direction) are the measurement frames 51 and 52. Assume that the contrast difference in FIG. 5B is very small.
In this case, similar to the procedure described using the image of FIG. 3A, the change in brightness in the measurement direction is extracted at each of a plurality of different portions of the linear portions 42 and 44 (see FIG. 7). The change in brightness is integrated in the non-measurement direction (projection process), and the signal waveform of the registration mark 31 (FIG. 4A) is generated.

  Of the straight portions 45 to 48 of the base mark 32 that appear in the inspection image (FIG. 5A), the straight portions 46 and 48 that intersect the measurement direction (x-axis direction) are the measurement frames 53 and 54. It is assumed that the contrast difference in FIG. 5B is large. In this case, if the signal waveform (FIG. 4B) of the ground mark 32 is generated by the procedure described using the image of FIG. 3A, the measurement error δx becomes large.

For this reason, in the overlay measurement apparatus 10 of the present embodiment, the change in brightness in the measurement direction is extracted at each of a plurality of different locations of the straight portions 46 and 48 (for example, for each scanning line) (see FIG. 8). The contrast (C _O (y ₁ ) to C _O (y ₃ ) etc.) is calculated. The contrast is equal to the difference between the maximum gradation value and the minimum gradation value of the change in brightness.
Next, an average value Cave is obtained from each calculated contrast (C _O (y ₁ ) to C _O (y ₃ ), etc.), and using this average value Cave, each of the extracted light-dark changes (FIG. 8) is obtained. The gradation value is multiplied by Cave / C _O (y). As a result, each change in brightness (FIG. 8) is converted into a change in brightness with a constant contrast (= average value Cave) as shown in FIG.

Then, after the contrasts of the respective light / dark changes are aligned with each other (FIG. 8 → FIG. 9), the obtained light / dark changes of the constant contrast (= average value Cave) are integrated in the non-measurement direction (projection processing). ), A signal waveform of the base mark 32 (FIG. 4B) is generated.
Therefore, the center-of-gravity positions C ₁ and C ₂ obtained from the respective signal waveforms (FIGS. 4A and 4B) of the registration mark 31 and the base mark 32 by the correlation calculation or the like are both the centers of the measurement frames 51 to 54 ( In other words, the change in brightness at the position y ₂ ) in FIGS. 5C and 7A coincides with the center-of-gravity positions C ₁ (y ₂ ) and C ₂ (y ₂ ) obtained similarly as temporary signal waveforms. Become.

For this reason, even when the signal noise is reduced by the projection processing, it is not affected by the inclination (rotation error δθ) of the linear portions 41 to 48 selected from the inspection image (FIG. 5A) and non-uniform contrast. Thus, the overlay deviation amount L can be obtained accurately.
That is, in the overlay measurement apparatus 10 of the present embodiment, the inspection image (FIG. 5A) is processed by the above-described procedure, so that the measurement error δx due to the rotation error δθ is reduced and accurately (that is, machine difference). The overlay deviation amount L can be determined (below 1 nm). As a result, the reliability of the overlay deviation amount L, which is a measured value, is increased, which contributes to an improvement in yield in semiconductor manufacturing.

  In the first embodiment described above, the case where the contrast difference between the linear portions 42 and 44 of the registration mark 31 appearing in the inspection image (FIG. 5A) is very small (FIG. 7) has been described as an example. However, the present invention is not limited to this. When the contrast difference between the linear portions 42 and 44 is large, processing for aligning the contrast with each other for each change in brightness in the measurement direction extracted from the linear portions 42 and 44 (FIG. 8 → FIG. 9) is performed, and then the projection processing is performed. The signal waveform (FIG. 4A) may be generated by performing integration according to the above.

Further, when the contrast difference between the linear portions 42 and 44 of the registration mark 31 is large and conversely the contrast difference between the linear portions 46 and 48 of the base mark 32 is small, the same effect can be obtained by applying the present invention. Can do.
Further, the process of aligning the contrast of each light / dark change (FIG. 8 → FIG. 9) may be performed individually for the light / dark change itself extracted from the inspection image (FIG. 5 (a)). It is not limited to. For example, after dividing each measurement frame into multiple areas with different positions in the non-measurement direction, integrating multiple adjacent light and dark changes in the divided areas, and aligning the contrast with each other for each light and dark change in each divided area (FIG. 8 → FIG. 9) may be performed.

(Second Embodiment)
Here, the inspection image (FIG. 5A) is processed in the following procedure without performing the uniform contrast (FIG. 8 → FIG. 9) described in the first embodiment. Here again, an example in which the x-axis direction is the measurement direction will be described.
Among the straight line portions 41 to 44 of the registration mark 31 appearing in the inspection image (FIG. 5A), a plurality of brightness changes in the measurement direction are obtained from the straight line portions 42 and 44 intersecting the measurement direction (x-axis direction). Extraction is performed (see, for example, FIG. 7), and this light / dark change is integrated in the non-measurement direction (projection processing) to generate a signal waveform of the registration mark 31 (FIG. 4A).

Similarly, from the straight line portions 45 to 48 of the ground mark 32 appearing in the inspection image (FIG. 5A), from the straight line portions 46 and 48 intersecting the measurement direction (x-axis direction), the brightness in the measurement direction is dark and dark. A plurality of changes are extracted (see, for example, FIG. 8), and this light / dark change is integrated in the non-measurement direction (projection processing) to generate a signal waveform (FIG. 4B) of the background mark 32.
Then, the measurement error δx (FIG. 8) is obtained by using the gravity center positions C ₁ and C ₂ obtained from the respective signal waveforms (FIGS. 4A and 4B) of the registration mark 31 and the base mark 32 by correlation calculation or the like. The overlay deviation amount L is calculated.

Next, in order to perform addition / subtraction of the measurement error δx (FIG. 8) included in the overlay deviation amount L, the contrast of each brightness change extracted from the linear portions 42 and 44 of the registration mark 31 (for example, C _i (y in FIG. 7). ₄ ), C _i (y ₂ ), C _i (y ₅ ), etc.) are calculated. Similarly, contrasts (for example, C _O (y ₁ ) to C _O (y ₃ ) in FIG. 8) extracted from the linear portions 46 and 48 of the background mark 32 are also calculated. The calculated contrast is stored in a memory (not shown) together with position information (y _{1 to} y _5, etc.) of the brightness change in the non-measurement direction. The position information (y _{1 to} y ₅ etc.) is, for example, the address of each scanning line.

  Further, in order to perform addition / subtraction of the measurement error δx (FIG. 8), the following calculation formula (2) is prepared.

計算式(２)は、加減算用の補正係数としての測定誤差δｘを算出するための一次近似式であり、検査用の画像（図５(ａ)）の回転誤差δθをパラメータとして含んでいる。回転誤差δθは、上記したように、カメラ２６の傾き角と等しい。このため、予め測定を行い、装置パラメータとして登録することができる。回転誤差δθの測定には、市販の（例えばコグネックス社製など）の画像解析ソフトを用いればよい。

The calculation formula (2) is a linear approximation formula for calculating the measurement error δx as a correction coefficient for addition and subtraction, and includes the rotation error δθ of the inspection image (FIG. 5A) as a parameter. The rotation error δθ is equal to the tilt angle of the camera 26 as described above. For this reason, it can measure beforehand and can register as an apparatus parameter. For measurement of the rotation error δθ, commercially available image analysis software (for example, manufactured by Cognex Corporation) may be used.

In the calculation formula (2), the first term on the right side represents an error component caused by the inner registration mark 31, and the second term represents an error component caused by the outer base mark 32. Further, each of these terms indicates, for each scanning line (each position in the non-measurement direction), an amount (y · δθ) by which the linear portion of each mark shifts in the measurement direction with the rotation error δθ, and contrast C _It means an average value obtained by weighting with _i (y) and C _O (y).

In the present embodiment, according to the calculation formula (2), the contrast C _i (y) related to the registration mark 31 obtained from the inspection image (FIG. 5A) and the position information in the non-measurement direction ( y) and the rotation error δθ (C _i (y) · y · δθ) are calculated, integrated in the measurement frames 51 and 52, and the obtained integrated value is used as the contrast C _i (y The error component due to the registration mark 31 of the first term is calculated.

Similarly, the contrast C _O (y) related to the background mark 32 obtained from the image (FIG. 5A) is calculated by multiplying the position information (y) in the non-measurement direction by the rotation error δθ (C _O (y ) · Y · δθ) is calculated and integrated in the measurement frames 53 and 54, and then the obtained integrated value is divided by the integrated value of the contrast C _O (y) to obtain the background of the second term. An error component caused by the mark 32 is calculated.
Further, a difference between the error component (first term) caused by the registration mark 31 and the error component (second term) caused by the background mark 32 is obtained, and a measurement error δx as a correction coefficient for addition / subtraction is calculated.

Finally, using this correction coefficient, the measurement error δx (FIG. 8) included in the overlay deviation amount L calculated without performing the above-described contrast uniformity (FIG. 8 → FIG. 9) is added and subtracted. The overlay deviation amount L is corrected.
Therefore, even when the signal noise is reduced by the projection processing, it is not affected by the inclination (rotation error δθ) of the linear portions 41 to 48 selected from the inspection image (FIG. 5A) and the non-uniformity of the contrast. The overlay deviation amount L can be accurately obtained.

That is, also in this embodiment, it is possible to reduce the measurement error δx due to the rotation error δθ and accurately obtain the overlay deviation amount L (that is, with a machine difference of 1 nm or less). As a result, the reliability of the overlay deviation amount L, which is a measured value, is increased, which contributes to an improvement in yield in semiconductor manufacturing.
In the second embodiment described above, the rotation error δθ, which is the parameter of the calculation formula (2), has been described as an example of measuring in advance using commercially available image analysis software. However, the present invention is not limited to this. For example, as shown in FIG. 10, a pair of measurement frames 53A and 54A and measurement frames 53B and 54B are set at different positions in the non-measurement direction in the straight line portions 46 and 48 of the image, and known according to the interval δy between them. , The contrast C _i (y) of the light and dark changes extracted from the measurement frames 53A and 54A, the contrast C _O (y) of the light and dark changes extracted from the measurement frames 53B and 54B, and the like in the calculation formula (2). Therefore, the rotation error δθ may be obtained by calculation. The rotation error δθ can be registered in advance as an apparatus parameter, but may be calculated every time an inspection image is captured.

  In the second embodiment described above, the rotation error δθ of the image is used as the parameter of the calculation formula (2), but the present invention is not limited to this. When the inclination of each straight line portion differs between the registration mark 31 and the ground mark 32, the inclination angle of the straight line portion of the registration mark 31 is used as δθ of the first term on the right side, and the ground mark 32 of the ground mark 32 is denoted as δθ of the second term. The inclination angle of the straight line portion may be used. Since both the registration mark 31 and the base mark 32 are formed using the same exposure apparatus, a common δθ may be normally used. However, it may be appropriate to use a different δθ because the formation timing is shifted.

1 is a schematic diagram illustrating an overall configuration of an overlay measurement apparatus 10. It is a schematic diagram explaining the configuration (a), (b) of the overlay mark 11A. It is a figure explaining the linear parts 41-48 (a) which appeared in the image for a test | inspection, and the measurement frames 51-58 (b), (c) set to this. It is a figure explaining the signal waveform after integration by projection processing. It is a figure explaining the rotation error (delta) theta of the image for a test | inspection, and the inclination of the linear parts 41-48 with respect to a coordinate axis (x axis, y axis). It is a figure explaining the light-dark change of the measurement direction of the linear parts 46 and 48 in the measurement frames 53 and 54 (when contrast difference is small). It is a figure explaining the light-dark change of the measurement direction of the linear parts 42 and 44 in the measurement frames 51 and 52 (when contrast difference is small). It is a figure explaining the light-dark change of the measurement direction of the linear parts 46 and 48 in the measurement frames 53 and 54 (when a contrast difference is large). It is a figure explaining the state after aligning the contrast of each brightness change mutually. It is a figure explaining the modification of how to obtain | require rotation error (delta) (theta).

Explanation of symbols

10 overlay measurement apparatus; 11 substrate; 13-19 illumination optical system;
18-25 imaging optical system; 26 CCD camera; 27 control unit; 31 registration mark;
32 ground mark; 41 to 48 linear portion; 51 to 58 measurement frame;
31L, 31R, 32L, 32R Signal waveforms after projection processing

Claims (10)

Capture means for capturing images of the first mark and the second mark formed on different layers of the substrate;
Based on the first straight line portion that intersects the predetermined measurement direction among the straight line portions of the first mark that appear in the image, a signal related to the change in brightness when the first straight line portion is crossed in the measurement direction. First processing means for generating a waveform as a first signal waveform;
Based on a second straight line portion that intersects the measurement direction among the straight line portions of the second mark appearing in the image, a signal waveform related to a change in brightness when the second straight line portion is crossed in the measurement direction is obtained. Second processing means for generating a second signal waveform;
Using the first signal waveform and the second signal waveform, a third processing means for obtaining an overlay deviation amount in the measurement direction between the first mark and the second mark,
The first processing means extracts a light / dark change when the first straight part is crossed in the measurement direction at a plurality of different locations on the first straight part, and the light / dark change is perpendicular to the measurement direction. The first signal waveform is generated by integrating in a different direction,
The second processing unit extracts a change in brightness when the second straight line portion is crossed in the measurement direction at a plurality of different locations of the second straight line portion, and matches the contrast of the light-dark change to each other. Then, the second signal waveform is generated by integrating the change in brightness in the vertical direction. In the overlay measurement apparatus according to claim 1,
The second processing means determines a contrast magnitude of the light / dark change of the second straight line portion, and determines whether or not to match the contrast of the light / dark change based on the determination. Combined measuring device. In the overlay measurement apparatus according to claim 2,
The overlay measuring apparatus, wherein the second processing means aligns the contrast of the plurality of portions of the second straight line portion with an average value. Capture means for capturing images of the first mark and the second mark formed on different layers of the substrate;
Based on the first straight line portion that intersects the predetermined measurement direction among the straight line portions of the first mark that appear in the image, a signal related to the change in brightness when the first straight line portion is crossed in the measurement direction. First processing means for generating a waveform as a first signal waveform;
Based on a second straight line portion that intersects the measurement direction among the straight line portions of the second mark appearing in the image, a signal waveform related to a change in brightness when the second straight line portion is crossed in the measurement direction is obtained. Second processing means for generating a second signal waveform;
Using the first signal waveform and the second signal waveform, a third processing means for obtaining an overlay deviation amount in the measurement direction between the first mark and the second mark,
The first processing means extracts a light / dark change when the first straight portion is crossed in the measurement direction at a plurality of different locations of the first straight portion, and matches the contrast of the light / dark change to each other. Then, the first signal waveform is generated by integrating the change in brightness in a direction perpendicular to the measurement direction,
The second processing unit extracts a change in brightness when the second straight line portion is crossed in the measurement direction at a plurality of different locations of the second straight line portion, and matches the contrast of the light-dark change to each other. Thereafter, the second signal waveform is generated by integrating the change in brightness in the vertical direction.
An overlay measuring apparatus characterized by that. In the overlay measurement apparatus according to claim 4,
The first processing means determines the contrast magnitude of the brightness change of the first straight line portion, the second processing means determines the contrast strength of the brightness change of the second straight line portion, An overlay measuring apparatus that determines whether or not the contrasts of light and dark changes in the first and second straight line portions are made uniform based on the determination of the first and second processing means. In the overlay measurement apparatus according to claim 5,
The first processing unit aligns the contrast of the plurality of portions of the first straight line portion with an average value, and the second processing unit aligns the contrast of the plurality of portions of the second straight line portion with an average value. An overlay measuring apparatus characterized by the above. In the overlay measurement apparatus according to any one of claims 1 to 6,
At least one of the first mark and the second mark has a different contrast in the direction of integration in the first and second marks. In the overlay measurement apparatus according to any one of claims 1 to 7,
At least one of the first mark and the second mark intersects the integrating direction, the overlay measuring apparatus. In the overlay measurement apparatus according to any one of claims 1 to 8,
A plurality of different portions of the first and second straight line portions are set along the integrating direction, the overlay measurement apparatus. Capture means for capturing images of the first mark and the second mark formed on different layers of the substrate;
Based on the first straight line portion that intersects the predetermined measurement direction among the straight line portions of the first mark that appear in the image, a signal related to the change in brightness when the first straight line portion is crossed in the measurement direction. First processing means for generating a waveform as a first signal waveform;
Based on a second straight line portion that intersects the measurement direction among the straight line portions of the second mark appearing in the image, a signal waveform related to a change in brightness when the second straight line portion is crossed in the measurement direction is obtained. Second processing means for generating a second signal waveform;
Using the first signal waveform and the second signal waveform, a third processing means for obtaining an overlay deviation amount in the measurement direction between the first mark and the second mark,
The first processing means extracts a light / dark change when the first straight part is crossed in the measurement direction at a plurality of different locations on the first straight part, and the light / dark change is perpendicular to the measurement direction. The first signal waveform is generated by integrating in a different direction,
The second processing means includes a plurality of second linear portions at different locations of the second linear portion, respectively.
Is extracted in the measurement direction, and the second signal waveform is generated by integrating the change in brightness in the vertical direction,
The third processing means obtains the overlay deviation amount using the first signal waveform and the second signal waveform, and then compares each contrast change extracted from the first straight line portion with the first and second contrast waveforms. The overlay deviation amount is corrected based on the contrast of each light and dark change extracted from the two straight line portions.
An overlay measuring apparatus characterized by that.

Images