Alignment mark detection method, alignment mark detection device, and member alignment device

The use of a hyperspectral camera to correct alignment mark contours based on film thickness estimation enhances alignment accuracy and uniformity in semiconductor wafer processing.

WO2026240035A1PCT designated stage Publication Date: 2026-11-19TOKYO ELECTRON LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
PCT/JP2026/017665
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2025-05-16
Filing Date
2026-05-08
Publication Date
2026-11-19

AI Technical Summary

Technical Problem

Existing alignment mark detection methods are inaccurate due to film thickness variations, leading to misalignment and non-uniform film deposition on semiconductor wafers.

Method used

Utilizing a hyperspectral camera to capture reflectance spectra and estimate film thickness, correcting the provisional contour of alignment marks to accurately detect their centers, ensuring precise alignment and uniform film deposition.

Benefits of technology

Enables highly accurate detection of alignment marks, allowing for precise positioning of semiconductor wafers and uniform film thickness across the wafer surface, improving manufacturing consistency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure JP2026017665_19112026_PF_FP_ABST
    Figure JP2026017665_19112026_PF_FP_ABST
Patent Text Reader

Abstract

The detection method according to the present disclosure for detecting an alignment mark provided on a first member in order to detect the position of a second member relative to the first member includes: an assumption step for capturing an image of the alignment mark, which is covered with a film, and detecting a provisional contour of the alignment mark; a spectra acquisition step for acquiring characteristic spectra indicating reflection or absorption characteristics at multiple detection points that are set on the basis of the provisional contour; a film thickness estimation step for estimating the film thickness of the film at each detection point on the basis of wavelengths determined for each detection point by comparison between each characteristic spectrum and a reference characteristic spectrum serving as a point of comparison common to each characteristic spectrum; and a contour detection step for correcting the provisional contour on the basis of the film thickness estimated at each detection point and detecting a contour of the alignment mark.
Need to check novelty before this filing date? Find Prior Art

Description

Alignment mark detection method, alignment mark detection device, and member alignment device

[0001] The present disclosure relates to an alignment mark detection method, an alignment mark detection device, and a member alignment device.

[0002] In manufacturing a semiconductor device from a substrate such as a semiconductor wafer, a predetermined operation is performed after adjusting the relative position between one member and the other member. Specific examples of this predetermined operation include placing the substrate on a stage when one member is the stage and the other member is the substrate, connecting a chip to the substrate when one member is the substrate and the other member is the chip, and performing processing by a processing unit when one member is the substrate and the other member is the processing unit that performs processing on a specific portion of the substrate.

[0003] The above adjustment of the relative position is performed by capturing an image of a mark called an alignment mark whose relative position is fixed with respect to one member or the other member. Patent Document 1 shows a configuration example of an apparatus in which the relative position between a substrate and a nozzle constituting a processing unit is adjusted based on an image obtained by capturing a circular alignment mark formed on the substrate, and then a processing liquid is supplied from the nozzle to the substrate.

[0004] Japanese Unexamined Patent Application Publication No. 2015-186771

[0005] The present disclosure provides a technique capable of highly accurately detecting an alignment mark provided for detecting the relative position between a first member and a second member.

[0006] The alignment mark detection method of the present disclosure is a detection method for detecting alignment marks provided on a first member for detecting the relative position of a second member with respect to a first member, and includes: a hypothesis step of imaging the alignment marks covered with a film and detecting a provisional contour of the alignment marks; a spectrum acquisition step of acquiring characteristic spectra showing the reflection or absorption characteristics at a plurality of detection points set based on the provisional contour; a film thickness estimation step of estimating the film thickness at each detection point based on a wavelength determined for each detection point by comparing each characteristic spectrum with a reference characteristic spectrum which is a common comparison target for each characteristic spectrum; and a contour detection step of correcting the provisional contour based on the film thickness estimated at each detection point and detecting the contour of the alignment marks.

[0007] This disclosure enables highly accurate detection of alignment marks provided for detecting the relative position between a first member and a second member.

[0008] This is a longitudinal cross-sectional side view of a film deposition apparatus according to the first embodiment of this disclosure. This is a plan view of the film deposition apparatus and a transport mechanism for the film deposition apparatus. This is a schematic diagram of an image of alignment marks (AMs) obtained in a comparative example. This is a schematic diagram of an image of AMs obtained in a comparative example. This is a longitudinal cross-sectional side view of an AM provided on a wafer W processed by the film deposition apparatus. This is a schematic diagram of an image of an AM. This is a longitudinal cross-sectional side view of the AM. This is a schematic side view of a hyperspectral camera (HSC) provided in the film deposition apparatus. This is a schematic diagram of the components of the HSC. This is an explanatory diagram showing the relationship between the image target and the reflectance spectrum. This is an explanatory diagram showing the relationship between the image target and the reflectance spectrum. This is a schematic diagram of an image of the AM. This is a schematic diagram of an image of the AM. This is a schematic diagram of an image of the AM. This is a characteristic diagram showing the reflectance spectrum of each point. This is a characteristic diagram showing the reflectance spectrum of each point. This is a correlation diagram showing the correspondence between singular wavelengths and film thickness on the AM. This is a plan view of the AM and the transport mechanism. This is a characteristic diagram showing the reflectance spectrum of each point. This is a block diagram of a control device provided in the film deposition apparatus. This is a flow diagram of the operations performed by the film deposition apparatus and the transport mechanism. This is a characteristic diagram showing the relationship between brightness and position in an image. This is a characteristic diagram showing the relationship between brightness and position in an image. This is a characteristic diagram showing the relationship between brightness and position in an image. This is a characteristic diagram showing the relationship between brightness and position in an image. This is a schematic diagram of the AM image. This is an explanatory diagram showing the relationship between the imaging target and the reflectance spectrum. This is a longitudinal cross-sectional side view showing an AM of another configuration. This is a schematic diagram of the AM image. This is a schematic diagram of an AM image of yet another configuration. This is a schematic diagram of the AM image. This is an explanatory diagram showing the relationship between the imaging target and the reflectance spectrum. This is an explanatory diagram showing the relationship between the imaging target and the reflectance spectrum. This is a longitudinal cross-sectional side view of the wafer W to show another example of the AM. This is a cross-sectional plan view of a bonding apparatus according to the second embodiment. This is a cross-sectional front view of the bonding apparatus. This is a plan view of the wafer W bonded by the bonding apparatus. This is a plan view of the wafer W bonded by the bonding apparatus. This is a side view of a bonding apparatus according to the third embodiment. This is a side view of another bonding apparatus according to the third embodiment. This is a side view of a bonding apparatus according to the fourth embodiment. This is a side view showing a modified example of the joining device according to the third embodiment. This is a graph showing the results of the evaluation test.

[0009] [Outline Configuration of the Film Deposition Apparatus of the First Embodiment] The film deposition apparatus 1, which is the first embodiment of the alignment mark detection device of this disclosure, will be described with reference to Figure 1, a longitudinal cross-sectional side view, and Figure 2, a plan view of some of the components of the apparatus. In the figures, the horizontal directions that are orthogonal to each other are shown as the X direction and Y direction, and the vertical direction is shown as the Z direction. The film deposition apparatus 1 comprises a processing container 11, a stage 21 provided inside the processing container 11, a shower head 12 provided on the ceiling of the processing container 11 above the stage 21, a gas supply mechanism 13 equipped with a valve and a gas supply source, and an exhaust mechanism 14 including a vacuum pump and a valve for adjusting the amount of exhaust by the vacuum pump.

[0010] The processing container 11 is evacuated by the exhaust mechanism 14 and maintained at a vacuum pressure. With a vacuum atmosphere formed inside the processing container 11, the film-forming gas supplied from the gas supply mechanism 13 to the shower head 12 is discharged onto the semiconductor wafer (hereinafter referred to as wafer), which is a circular substrate placed on the stage 21, and a film-forming process is carried out on the wafer W.

[0011] A transport port provided on the side wall of the processing container 11 is opened and closed by a gate valve 15. The wafer W is supported from the back side by a transport mechanism 19 equipped with a support that is movable in both the X and Y directions, and the wafer W is transported between the inside and outside of the processing container 11 through this transport port. The transport mechanism 19, which forms the wafer W positioning mechanism, is, for example, a SCARA type robot, and the support is an end effector; only the support among the elements constituting the transport mechanism 19 is shown in the illustration. The film deposition apparatus 1 is equipped with three pins 22 extending in the Z direction and a lifting mechanism 23 that raises and lowers the pins 22 in the Z direction to make them protrude and retract from the upper surface of the stage 21, and the wafer W is transferred between the transport mechanism 19 and the stage 21 via the pins 22.

[0012] The first component, the stage 21, incorporates an electrostatic chuck capable of adsorbing a wafer W, and the central part of the horizontal upper surface of the stage 21 is set as a mounting area 24 formed by the electrostatic chuck. The second component, the wafer W, is placed on this mounting area 24, and a film-forming gas is supplied uniformly to all parts of the surface of the wafer W from the showerhead 12, and a film 33 is formed with high uniformity by the deposition of components in this film-forming gas. A heater (not shown) is embedded in the stage 21, and the wafer W on the stage 21 is heated so that the components of the film-forming gas can be deposited.

[0013] A portion of the upper surface (front) of the wafer W protrudes upward, forming multiple, for example, four, protrusions. Each of these protrusions is configured as an alignment mark (AM) 2. Therefore, the relative position of each AM 2 with respect to the wafer W is fixed. Each AM 2 is circular in plan view.

[0014] A portion of the ceiling of the processing container 11 is configured as a window 16 that can transmit visible light. Outside the processing container 11, above the window 16, is an illumination 17 and an imaging mechanism, a hyperspectral camera (HSC) 20. Figure 2 shows the image acquisition area 25 where imaging by the HSC 20 takes place, enclosed by a dashed line. The illumination 17 shines light into the image acquisition area 25 through the window 16 to enable this imaging. Imaging by the HSC 20 is performed with the transport mechanism 19 supporting the wafer W and positioned above the stage 21. In this state, a portion of the wafer W on the mounting area 24, including each AM2 and the area around each AM2, fits within the image acquisition area 25.

[0015] The film deposition apparatus 1 is equipped with a control device 40, which is a computer. Data acquired by imaging with the HSC 20 is transmitted to the control device 40. The control device 40 detects the center of each AM2 in a plan view from the image acquired from the data and controls the operation of the transport mechanism 19 so that the center of each AM2 in a plan view is positioned at a predetermined relative position with respect to the stage 21. That is, the wafer W is aligned with respect to the stage 21 based on the center of the AM2 in a plan view. This alignment is performed so that when the pin 22 rises to receive the wafer W from the transport mechanism 19 and then descends, the wafer W is placed on the mounting area 24. When the wafer W is transported into the processing container 11, it is conceivable that the position of the wafer W may be misaligned with respect to the support of the transport mechanism 19. Therefore, in order to prevent the wafer W from being placed misaligned from the mounting area 24, the transport mechanism 19 is aligned based on the image of the AM2 in this way. The transport mechanism 19, together with the film deposition apparatus 1, which is an alignment mark detection device, constitutes a component positioning device. In the following explanation, "the center of AM2 in a plan view" may sometimes be simply referred to as "the center of AM2."

[0016] [AM Center Detection in Comparative Example] Before further explaining the configuration of the film deposition apparatus 1 and the method for detecting the center of AM2, we will first describe the comparative film deposition apparatus 1' and the method for detecting the center of AM2 using this film deposition apparatus 1'. In the case of the film deposition apparatus 1', instead of the HSC20, an RGB camera such as a CCD camera or CMOS camera is provided as the imaging mechanism. It differs from the film deposition apparatus 1 according to the first embodiment in that the center of AM2 is calculated from the image acquired by this RGB camera by the control device 40' provided in the film deposition apparatus 1' and the operation of the transport mechanism 19 is controlled accordingly. Figure 3 schematically shows the AM2 in the image acquired by the RGB camera in this way for illustrative purposes.

[0017] The procedure for calculating the center position of AM2, performed by the control device 40', will now be explained. Since a step is formed between AM2 and the upper surface of the wafer W where AM2 is not formed, this step appears as a difference in brightness in the image acquired by the RGB camera. Binarization processing is performed on this image, and the difference in brightness increases. Then, the contour 31 of AM2 is identified based on the difference in brightness in the image. It should be noted that the identification of this contour 31 can also be said to be the identification of the outer edge of AM2.

[0018] Next, multiple points on the contour 31 are selected using a predetermined algorithm. In Figure 3, these selected points are shown as 32. After the selection of these points 32, a virtual circle is calculated from each point 32 using a calculation method such as the least squares method. That is, the virtual circle is set up so that each point 32 is located on or close to this virtual circle. Then, the position of the center P1 in this virtual circle is determined.

[0019] Figure 3 shows an image of AM2 when it is exposed without being covered by the film 33 described later. When such an image is obtained, the actual contour of AM2, the contour 31 of AM2 obtained by image processing, and the virtual circle obtained from point 32 on the contour 31 all coincide or nearly coincide. Therefore, the center P1 of AM2 calculated from the virtual circle coincides or nearly coincides with the actual center P0 of AM2. As a result, the wafer W can be placed on the mounting area 24 by aligning the wafer W by the transport mechanism 19 so that the wafer W is positioned at a predetermined relative position according to the center P1 of each AM2. The dotted lines in Figure 3 are virtual auxiliary lines drawn to facilitate understanding the positional relationship between each point 32 and the center P1.

[0020] However, the film 33 is formed on the wafer W before it is transported to the film deposition apparatus 1, so that it covers the surface of the wafer W. Therefore, it is possible that the film 33 is covering AM2 when imaging is performed by the RGB camera.

[0021] Figure 4 is a schematic diagram illustrating an image of AM2 covered with film 33, and Figure 5 is a longitudinal cross-sectional side view of AM2 from which this image was acquired. When AM2 is covered with film 33 in this way, the film 33 is formed not only on the top surface of AM2 but also on the sides, so the position of the step shifts outward from AM2 compared to when AM2 is not covered with film 33. Therefore, in the image obtained by the RGB camera, the position with a relatively large difference in brightness does not coincide with the outer edge of AM2, so the contour 31 of AM2 identified after the binarization process of the image described above will deviate from the actual contour of AM2 shown by the dotted line in Figure 4.

[0022] Even if a misalignment of the contour occurs in this way, as shown in Figures 4 and 5, if the film thickness of the film 33 formed on the side surface of AM2 (hereinafter referred to as the lateral film thickness J) is uniform in all parts of the periphery of AM2, then the center P1 of AM2 identified from the contour 31 will coincide with the actual center P0 of AM2. Therefore, the position of the transport mechanism 19 can be controlled based on the position of this center P1, and the wafer W can be placed on the mounting area 24.

[0023] However, the lateral film thickness J is not necessarily uniform across the entire periphery of the AM2. Figures 6 and 7 show a schematic diagram of the AM2 image and a longitudinal cross-sectional side view of the AM2, respectively, similar to Figures 4 and 5. In this example, the lateral film thickness J is non-uniform across the entire periphery of the AM2. When the lateral film thickness J is non-uniform in this way, the center P1 of the AM2 identified from the contour 31 does not coincide with the actual center P0 of the AM2. Therefore, if the position control of the transport mechanism 19 is performed based on the position of this center P1, the wafer W will be placed offset from the mounting area 24. Consequently, there is a risk that the film thickness formed on the wafer W by the film deposition process in the film deposition apparatus 1 will vary significantly across the surface of the wafer W.

[0024] [Explanation of HSC] In order to resolve these problems, the film deposition apparatus 1 is equipped with an HSC 20 as an imaging unit, as described above. The control device 40 analyzes the imaging data acquired by the HSC 20 to detect the center P1 of AM2. Any known HSC 20 can be used as appropriate without any particular restrictions, but an example of its configuration will be briefly explained with reference to Figures 8 and 9. Figure 8 is a side view showing the schematic configuration of the HSC 20, and Figure 9 is a schematic diagram of the components included in the HSC 20. In Figures 8 and 9, the optical path is schematically shown by a dashed line.

[0025] For the HSC 20, for example, a so-called line scan type configuration can be used, which includes an elongated imaging area V extending in the X direction, and moves in the Y direction together with the HSC 20 by the movement mechanism 29 shown in Figure 1, thereby imaging the entire image acquisition area 25 shown in Figure 2. The HSC 20 includes a light guide member (not shown), a spectrometer 26, and a detection unit 27. The light guide member includes, for example, a lens or a plate that forms a slit in the optical path, and guides the light incident on the HSC 20 from the imaging area V to the spectrometer 26. In this example, the light incident on the HSC 20 is the reflected light from the illumination 17, which is reflected from objects such as AM2 in the image acquisition area 25 or the upper surface of the wafer W outside of AM2. The spectrometer 26 includes, for example, a diffraction grating or a prism, and spectrally separates the light guided from the light guide member. The spectrally separated light is then irradiated onto the detection unit 27. The detection unit 27 is a so-called image sensor and is composed of a number of optical elements 28 arranged in a matrix, for example, along the X direction and the Y' direction which is perpendicular to the X direction. An electrical signal corresponding to the light reception intensity is output from each optical element 28. Note that the Y' direction may coincide with the Y direction.

[0026] Assume that the imaging area V of the HSC 20 is divided into numerous areas V1 in its length direction (X direction). Light from different areas V1 is spectrally separated in the Y' direction by the spectrometer 26 and enters optical elements 28 that are located at different positions in the X direction. In other words, as shown in Figure 8, light from area V1 is incident on a row of optical elements 28 that are aligned in the Y' direction corresponding to that area V1. Then, as shown in Figure 9, although light emitted from the same area V1 is incident on each optical element 28 that is located at the same position in the X direction and is aligned in the Y' direction, the wavelength bands of the incident light are different for each optical element 28 due to spectral separation. The spectrometer 26 and the detection unit 27 are configured so that the light entering from the imaging area V can be spectrally separated into a large number of bands, for example, 100 or more, and received by the optical elements 28.

[0027] As described above, the HSC 20 is used to image the image acquisition area 25, and the control device 40 obtains information on the light intensity in 100 or more bands for each position in the image acquisition area 25 as imaging data. Using this imaging data, the control device 40 can acquire the reflectance spectrum at each position in the image, identify the specific wavelength from the reflectance spectrum as will be described in detail later, estimate the lateral film thickness J, correct the contour 31 of AM2, and detect the center P1 of AM2.

[0028] [Regarding Singular Wavelengths] Here, we will explain singular wavelengths with reference to Figure 10, an explanatory diagram showing the relationship between the object being imaged by the HSC20 and the reflectance spectrum obtained from that object. For the purpose of this explanation, Figure 10 shows side views of structures 37A and 37B, respectively, consisting of a base film 35 and an upper film 36 covering the base film 35 from above. In the illustrated example, the thickness of the base film 35 differs between structures 37A and 37B, with the base film 35 of structure 37A being thicker. The relationship between structures 37A and 37B and the AM2 described above will be explained later.

[0029] If structures 37A and 37B are imaged from above using the aforementioned HSC20, reflectance spectra can be obtained from position 38A on structure 37A and position 38B on structure 37B, respectively, as shown in the lower part of Figure 10. Under conditions where the illuminance between positions 38A and 38B is the same or approximately the same, i.e., illumination light of the same or approximately the same intensity is incident, the reflectance spectra 39A and 39B obtained from positions 38A and 38B will be different due to the influence of differences in the composition (in this case, thickness) of the undercoat 35. The wavelength at which the difference in reflectance between the reflectance spectra 39A and 39B obtained in this way is called the singular wavelength. In the example shown in Figure 10, the wavelength shown as E0 nm is the singular wavelength. In the example shown in Figure 10, the waveforms of reflectance spectra 39A and 39B do not intersect, but they may intersect. When waveforms intersect in this way (i.e., when an intersection point exists in the waveforms), the difference in reflectivity at that intersection point is zero and therefore the minimum, and thus the wavelength at that intersection point is a singular wavelength.

[0030] If the film thickness of one of the upper layers 36 of structures 37A and 37B changes from the state shown in Figure 10, the reflectance spectrum obtained from that one will change, and the singular wavelength will shift accordingly. The applicant's verification has confirmed that the amount of this singular wavelength shift corresponds to the amount of change in the film thickness of the upper layer 36, that is, there is a correlation between the singular wavelength and the film thickness of the upper layer 36.

[0031] This peculiar wavelength shift will be explained in detail with reference to Figure 11, which, like Figure 10, shows the relationship between the imaging target and the reflectance spectrum. The structures 37A and 37B shown in Figure 11 differ from the structures 37A and 37B shown in Figure 10 only in the thickness of the upper layer film 36 of structure 37A. If structures 37A and 37B in Figure 11 are imaged from above with the HSC 20, similar to structures 37A and 37B in Figure 10, the reflectance spectrum 39A obtained from position 38A will be different from the reflectance spectrum 39A shown in Figure 10. Therefore, the peculiar wavelength E1 identified from the reflectance spectra 39A and 39B obtained from structures 37A and 37B in Figure 11 is different from the peculiar wavelength E0 in Figure 10. Then, if we denote the thickness of the upper film 36 of structure 37A in Figure 10 as F0 and F1 as the thickness of the upper film 36 of structure 37A in Figure 11, respectively, the amount of singular wavelength shift [E1-E0] corresponds to the difference in thickness of the upper film 36 [F1-F0].

[0032] In this embodiment, by utilizing the correlation between film thickness and specific wavelength, the film thickness of the film 33 at various points on the periphery of AM2 is determined from the imaging data obtained by imaging the image acquisition region 25 shown in Figure 2 with HSC 20. These film thicknesses are the film thickness in the thickness direction of the wafer W, as shown as H in Figure 7. Subsequently, the contour of AM2 in the image is corrected separately for each part of the contour using each film thickness H to calculate the contour with high accuracy, and then the center P1 of AM2 is obtained.

[0033] In Figures 10 and 11, the underlayer film 35 corresponds to the surface layer (upper layer) of the wafer W, and the upper layer film 36 corresponds to the film 33 remaining on this surface layer. Structure 37A, shown as having a large underlayer film thickness, corresponds to the area on the wafer W where AM2 is formed, and structure 37B, shown as having a small underlayer film thickness, corresponds to the area on the wafer W where AM2 is not formed. The positions 38A and 38B from which the reflectance spectra 39A and 39B are acquired correspond to the film thickness detection point α and the reference point R, respectively, which will be explained later.

[0034] Regarding reflectance, for example, a specific light intensity can be set in advance as 100%, and the reflectance can be calculated as the ratio of the intensity of reflected light obtained from each optical element 28 of the HSC 20 to that specific light intensity. In other words, the reflectance from each optical element 28 can be obtained by dividing the intensity of reflected light obtained from each optical element 28 by the same light intensity. This specific light intensity may be the light intensity of illumination light, or it may be the light intensity obtained by the HSC 20 when imaging a specific object.

[0035] [Detection of the AM Center in This Embodiment] The outline of the detection of the AM2 center P1 in this embodiment will be explained in more detail. First, with the transport mechanism 19 supporting the wafer W and positioned on the stage 21, the HSC 20 captures the image acquisition area 25, and the captured data is transmitted to the control device 40, after which the control device 40 performs data processing. For the purposes of this explanation, it will be assumed that this captured data is acquired with the lateral film thickness J of the AM2 being non-uniform, as shown in Figure 7. Therefore, Figure 7 will be referred to as appropriate in the following explanation.

[0036] As part of the data processing described above, a binarization process is first performed to enhance and detect the contour 31 of AM2. This binarization process will be described in detail below. As the HSC20 is configured as described above, the control device 40 can obtain images for each of more than 100 bands. This binarization process is performed, for example, by calculating the sum of the brightness values ​​of the images for each band at each position in the acquired image acquisition area 25, comparing this sum with a preset threshold, and determining whether or not it exceeds the threshold.

[0037] Let's explain this specifically assuming a wavelength range of 5 nm for one band. If, for a pixel at a certain position in the acquired image, the brightness in the 400 nm image is B1, the brightness in the 405 nm image is B2, the brightness in the 410 nm image is B3, and so on, then the calculation B1 + B2 + B3... = Bx is performed. The sum of these brightness values, Bx, is compared to a threshold to determine whether the pixel will be output as white or black after binarization. The contour 31 of AM2 is identified from the black and white contrast of the image obtained after binarization in this way. This contour 31 is corrected later, so it is sometimes referred to as a provisional contour 31.

[0038] The processing of imaging data after detection of the provisional contour 31 will be explained with reference to Figures 12 to 15, which are schematic diagrams of the AM2 image in an arbitrary band. Once the provisional contour 31 is identified as described above, film thickness detection points α are set inside this provisional contour 31 according to a predetermined algorithm. As shown in Figure 12, multiple film thickness detection points α are set and may be distinguished from each other by numbering them as α1, α2, α3...αn. In the illustrated example, 16 film thickness detection points α are set, so n=16. Each film thickness detection point α is set inside the contour 31 and in the vicinity of the contour 31 so that it is expected to be located on the periphery of the AM2. Figure 7 shows film thickness detection points α1 and α9 as examples.

[0039] Furthermore, a point outside the contour 31 but in the vicinity of the contour 31 is designated as the reference point R. Therefore, the reference point R is set to be a point at a different location from AM2, that is, a point outside AM2 and on the wafer W. Since the reference point R is set to be a point in the vicinity of the contour 31, it is set for each AM2. In other words, four reference points are set, the same as the number of AM2s. The reason for setting the reference point R to a point in the vicinity of the contour 31 in this way is that if the distance between the film thickness detection point α and the reference point R is too large, the illuminance difference between the film thickness detection point α and the reference point R may become large, as will be described later. If the illuminance difference is large, the reflectance spectrum obtained from each point will be affected by that illuminance difference, and the aforementioned peculiar wavelength and film thickness will no longer correspond. To prevent this, the reference point R is set in the vicinity of the contour 31 (which is also the vicinity of each film thickness detection point α).

[0040] Specifically, it is preferable that each film thickness detection point α and the reference point R be set such that the distance between each film thickness detection point α and the reference point R is less than 3 cm. Note that "less than 3 cm" means that the distance in the image is less than 3 cm when converted to an actual distance. Since the image is acquired as a plan view image, more specifically, it is preferable that each point be set such that the actual distance between each film thickness detection point α and the reference point R in a plan view is less than 3 cm. The reference point R may be set based on the temporary contour 31 so that it is in a predetermined positional relationship with the temporary contour 31, or the reference point R may be a predetermined point in the image regardless of the position of the contour 31.

[0041] After setting the film thickness detection points α1 to αn and the reference point R in this manner, the reflectance spectrum C at each of the film thickness detection points α1 to αn and the reflectance spectrum at the reference point R (which may hereafter be referred to as the reference reflectance spectrum C0) are calculated from the imaging data. The reflectance spectra C of the film thickness detection points α1 to αn are sometimes represented by the same numbers 1 to n as the source film thickness detection points α1 to αn. In other words, the reflectance spectra obtained from film thickness detection points α1, α2, α3, etc. are sometimes represented as C1, C2, C3, etc. As will be described in detail later, the film thickness H at each film thickness detection point α is calculated, and corresponding points β and correction points γ are set for each film thickness detection point α. These film thickness H, corresponding points β and correction points γ are also sometimes represented by the same numbers as the source film thickness detection points α, similar to the reflectance spectrum C.

[0042] After acquiring the reflectance spectra C1 to Cn and the reference reflectance spectrum C0, each of the reflectance spectra C1 to Cn is compared with the reference reflectance spectrum C0. In other words, the comparison is performed for each set number of film thickness detection points α, such as comparing the reflectance spectrum C1 at film thickness detection point α1 with the reference reflectance spectrum C0, comparing the reflectance spectrum C2 at film thickness detection point α2 with the reference reflectance spectrum C0, comparing the reflectance spectrum C3 at film thickness detection point α3 with the reference reflectance spectrum C0, and so on. This comparison yields the singular wavelengths explained in Figures 10 and 11. As shown in Figures 10 and 11, the singular wavelengths shift according to the film thickness. Therefore, each singular wavelength obtained by comparing the reflectance spectra C1 to Cn at film thickness detection points α1 to αn with the reference reflectance spectrum C0, which is set as a common comparison target for those reflectance spectra C1 to Cn, corresponds to the film thickness H at each of the film thickness detection points α1 to αn.

[0043] This will be specifically described with reference to FIG. 7, and FIGS. 16 and 17 which are characteristic diagrams showing reflectance spectra. FIGS. 16 and 17 respectively show the reflectance spectrum C1 at the film thickness detection point α1, the reflectance spectrum C9 at the film thickness detection point α9, and the reference reflectance spectrum C0. In addition, the singular wavelength obtained by comparing the reflectance spectrum C1 and the reference reflectance spectrum C0, and the singular wavelength obtained by comparing the reflectance spectrum C9 and the reference reflectance spectrum C0 are denoted as D1 and D9, respectively. As shown in FIG. 7, since the film thickness H1 at the film thickness detection point α1 and the film thickness H9 at the film thickness detection point α9 are different from each other, the singular wavelength D1 obtained from the film thickness detection point α1 shown in FIG. 16 and the singular wavelength D9 obtained from the film thickness detection point α9 shown in FIG. 17 are different, and these singular wavelengths D1 and D9 correspond to the film thicknesses H1 and H9, respectively.

[0044] As the correspondence relationship between the singular wavelength and the film thickness H, for example, the proportional relationship shown in FIG. 18 can be considered to exist, and the singular wavelength increases as the film thickness H increases. A test is performed in advance, and the correspondence relationship between the singular wavelength and the film thickness H (hereinafter referred to as the wavelength-film thickness correspondence relationship) as exemplified in the correlation diagram of FIG. 18 is acquired in advance. Then, after acquiring the singular wavelength from each film thickness detection point α as described above, the film thicknesses H1 to Hn of the film 33 at the film thickness detection points α1 to αn are calculated based on this wavelength-film thickness correspondence relationship.

[0045] Here, when film formation is performed on the object to be processed, if the film formation is substantially isotropic, the larger the film thickness H at the film thickness detection point α, the larger the side film thickness J in the vicinity of this film thickness detection point α (see FIG. 7). That is, it can be considered that the film thickness H and the side film thickness J are correlated. Therefore, assuming that the side film thicknesses J in the vicinity of the respective film thickness detection points α1 to αn are J1 to Jn, respectively, it is possible to estimate the side film thicknesses J1 to Jn from the film thicknesses H1 to Hn calculated as described above.

[0046] Points on the provisional contour 31 that are in the vicinity of the film thickness detection points α1 to αn are set to correspond to the film thickness detection points α1 to αn, and these points are designated as corresponding points β1 to βn (see Figure 12). After obtaining the lateral film thickness J1 to Jn, correction points γ1 to γn are obtained by moving these corresponding points β1 to βn towards the center of the region enclosed by the provisional contour 31 by an amount corresponding to the lateral film thickness J1 to Jn, as shown in Figure 13. In other words, correction point γ is obtained inside the provisional contour 31, and the larger the lateral film thickness J, the greater the position that correction point γ is moved from the corresponding point β towards the center of the region enclosed by the provisional contour 31. As described above, the correction point γ is the point obtained by moving the corresponding point β on the provisional contour 31 corresponding to the film thickness detection point α by a distance corresponding to the film thickness H and, consequently, the lateral film thickness J.

[0047] Then, a virtual circle is calculated from the correction points γ1 to γn using a calculation method such as the least squares method, as shown in Figure 14. That is, the virtual circle is set so that the correction points γ1 to γn are located on or close to this virtual circle. The virtual circle set in this way is the contour 34 of the AM2 corrected to the provisional contour 31 so that the influence of the lateral film thickness J1 to Jn on each part of the side surface of the AM2 is canceled out. In other words, the contour of the AM2 is detected with high accuracy so as to approximate the contour of the actual AM2. After the contour 34 is obtained in this way, for example, a predetermined algorithm selects multiple points on the contour 34, and from each of these selected points, a predetermined calculation is performed to calculate the center P1 of the contour 34 (which is also the centroid of the region enclosed by the contour 34), as shown in Figure 15, and this center P1 is determined as the center of the AM2.

[0048] [Specific Example of Setting Method for Each Point] An example of a specific setting method for the film thickness detection points α1 to αn, corresponding points β1 to βn, and correction points γ1 to γn shown in FIGS. 12 to 14 will be described. First, when a temporary contour 31 is acquired, a circle approximating the temporary contour 31 is calculated, and the center of gravity G (which is also the center of the circle) of this circle is calculated (see FIG. 12). The calculation of the circle may also be performed by selecting a plurality of points forming the contour 31 with a predetermined algorithm, and using a method such as the least squares method from the selected points. Then, a plurality of virtual straight lines L1 passing through the above-mentioned center of gravity G are set. Each virtual straight line L1 is inclined relative to each other at a predetermined angle. The intersection points between these virtual straight lines L1 and the temporary contour 31 are detected, and positions separated from the intersection points by a predetermined distance along the virtual straight lines L1 are set as the film thickness detection points α1 to αn. In addition, by setting the intersection points as the corresponding points β1 to βn, the film thickness detection point α and the corresponding point β with the same number are positioned on the same virtual straight line L1, and the film thickness detection point and the corresponding point β with the same number are set to be adjacent to each other.

[0049] After that, when the film thicknesses H1 to Hn at the film thickness detection points α1 to αn are acquired as described above, these film thicknesses H1 to Hn are treated as being the same as the side film thicknesses J1 to Jn, respectively. Therefore, in the present setting method, the acquisition of the film thickness H is also the acquisition of the side film thickness J. Then, the respective points corrected to be shifted toward the center of gravity G along the virtual straight line L1 passing through each of the corresponding points β1 to βn by the magnitude of the side film thicknesses J1 to Jn from the corresponding points β1 to βn are determined as the correction points γ1 to γn (see FIG. 13). That is, regarding the correction point γ1, it is determined as a point shifted from the corresponding point β1 toward the center of gravity G along the virtual straight line L1 passing through this corresponding point β1 by the same magnitude as the side film thickness J1 (= film thickness H1). Note that the relationship between the magnitude of the side film thickness J and the amount of shifting the point from the corresponding point β in the image (correction amount), which is necessary for determining the correction point γ from the side film thickness J, is acquired in advance.

[0050] The above setting method for each point is an example shown for ease of understanding, and is not limited to this setting method. For example, a predetermined calculation formula for calculating the side film thickness J from the film thickness H is prepared, and the side film thickness J is determined by performing an operation according to this calculation formula, so that the correction points γ1 to γn may be specified.

[0051] [Placement of wafer W after detection of center of AM2] As described above, for example, four AM2s are provided, and the center P1 is similarly determined for each AM2. After calculating the center P1 in this way, as shown in the plan view of Figure 19, the position of the transport mechanism 19 is controlled based on the positions of the four center P1s, and the position of the wafer W is corrected. Then, the pins 22 move up and down, and the wafer W is placed on the placement area 24. In other words, the wafer W is placed on the stage 21 in such a way that the displacement relative to the placement area 24 on the stage 21 is suppressed. After placement, when the film-forming gas is supplied to the wafer W and a film is formed, the displacement is suppressed, so a film with a highly uniform thickness is formed on each part of the surface of the wafer W. Although the position of the wafer W relative to the stage 21 is performed based on the center P1 of the AM2 as described above, as explained above, the center P1 is calculated from the corrected contour 34 of the AM2, so it can be said that the positioning is performed based on the contour 34.

[0052] [Supplementary information on the detection points of singular wavelengths and film thickness] In Figures 10 and 11, it was stated that intersections may occur in the reflectance spectra 39A and 39B obtained from structures 37A and 37B, respectively, and in such cases, the wavelength of the intersection is the singular wavelength. As shown in the characteristic diagram of Figure 20, there may be two or more intersections. In other words, there may be multiple singular wavelengths. Even in such cases, the center P1 of AM2 can be detected by processing the imaging data as described above. Specifically, one of the multiple intersections may be treated as the singular wavelength, and a wavelength-film thickness correspondence relationship may be obtained as illustrated in Figure 18. Alternatively, each of the multiple intersections may be treated as a singular wavelength, and a wavelength-film thickness correspondence relationship may be obtained from each singular wavelength, and the film thickness H may be determined from multiple singular wavelengths by taking the average of each singular wavelength and the film thickness obtained from each wavelength-film thickness correspondence relationship.

[0053] Furthermore, although the above explanation uses 16 for n, which is the number of film thickness detection points α, n can be 2 or more for the purpose of correcting multiple positions of the provisional contour 31 of AM2 as described later, and is not limited to n=16. Note that the larger the number of n, the more correction points there will be for the provisional contour 31, and the acquired contour 34 can be approximated more closely to the actual contour of AM2. From this perspective, it is preferable to set n to 4 or more.

[0054] [Configuration of the control device] Next, the control device 40 will be described with reference to the block diagram in Figure 21. As described above, the control device 40 is a computer and includes a microcomputer having a CPU (Central Processing Unit), ROM (Read Only Memory), RAM (Random Access Memory), input / output ports, and various circuits. As shown in the figure, the HSC 20 and the transport mechanism 19 are connected to the control device 40, and operation control can be performed by receiving data from the HSC 20 and transmitting control signals to the transport mechanism 19. Although not shown in the figure, the control device 40 is also connected to various parts other than the HSC, such as the gas supply mechanism 13, and operation control can be performed by transmitting control signals in order to perform the film deposition process.

[0055] The control device 40 includes a storage unit 42, which stores the OS (Operating System) and various programs executed by the control device 40, as described later. This program is configured to contain instructions (step groups) for executing a series of operations (as shown in the flowchart described later) performed by the film deposition apparatus 1. Furthermore, the storage unit 42 stores various data necessary for executing this series of operations, such as the wavelength-film thickness correspondence shown in Figure 18. These programs and data may be stored on a computer-readable storage medium (e.g., a hard disk, optical disc such as a DVD, or semiconductor memory). Additionally, these programs and data can be transmitted online from other devices as needed.

[0056] Next, the control unit 41, which constitutes the control device 40, will be described. The control unit 41 is a device that controls the control device 40. Specifically, the control device 40 is composed of electronic circuits such as a CPU (Central Processing Unit) and an MPU (Micro Processing Unit), and integrated circuits such as an ASIC (Application Specific Integrated Circuit) and an FPGA (Field Programmable Gate Array). The control unit 41 has an internal memory for storing programs and data, and reads and executes various programs stored in the storage unit 42. As each program is executed, the control unit 41 functions as a data acquisition unit 43 that receives imaging data from imaging by the HSC 20, an image processing unit 44 that performs processing from the aforementioned binarization process to the detection of the center P1 of the AM2 using the imaging data, a position adjustment unit 45 that controls the operation of the transport mechanism 19 based on the detected center P1 of each AM2, and a processing operation control unit 46 that controls operations related to the film deposition process, such as pressure control in the processing container 11 by the exhaust mechanism 14 and the supply of each gas by the gas supply mechanism 13.

[0057] With the control device 40 configured as described above, the acquisition and processing of imaging data described above can be performed, and the operation of each part of the film deposition apparatus 1 and the transport mechanism 19 can be controlled. Therefore, this control device 40 corresponds to a spectrum acquisition unit that acquires each reflectance spectrum, an estimation unit that estimates the film thickness H at each film thickness detection point α, and a detection unit that detects the AM contour 34. The operation control of each part of the film deposition apparatus 1 and the transport mechanism 19 is performed by the control device 40 outputting control signals to each part of the film deposition apparatus 1 and the transport mechanism 19.

[0058] [Wafer Placement and Processing Flow] Figure 22 is a flow chart showing, in order, the operation of the film deposition apparatus 1, the operation of the transport mechanism 19, and the data processing by the control device 40, from the transport of the wafer W into the processing container 11 as described above to the processing of the wafer W. The flow chart will be explained below. First, with the wafer W supported by the transport mechanism 19 positioned on the stage 21, the image acquisition area 25 is imaged by the HSC 20, and the image data is transmitted to the control device 40 (step S1).

[0059] Binarization is performed on the brightness of the images obtained from each band (step S2), and a provisional outline 31 of AM2 is detected from the image obtained from the binarization process, as explained in Figure 12 (step S3). Furthermore, each film thickness detection point α is set so that it is expected to be located on the periphery of AM2 based on the provisional outline 31, while a reference point R is set outside the provisional outline 31 (step S4). Subsequently, the reflectance spectrum at each of the film thickness detection points α and reference point R is obtained, and a unique wavelength is detected for each film thickness detection point α by comparing the reflectance spectrum C of the film thickness detection point α with the reference reflectance spectrum C0, as explained in Figures 16 and 17 (step S5). After that, the film thickness H of each film thickness detection point α is calculated from the wavelength-film thickness correspondence relationship exemplified in Figure 18 and the detected unique wavelength, and the lateral film thickness J near each film thickness detection point α is estimated (step S6). As previously described, the film thickness H at the film thickness detection point α may be used as the lateral film thickness J, or the value obtained by substituting this film thickness H into the calculation formula may be used as the lateral film thickness J.

[0060] Next, as explained in Figure 13, corresponding points β1 to βn, which are set to correspond to the film thickness detection points α1 to αn on the provisional contour 31, are corrected based on the lateral film thickness J1 to Jn, and corrected points γ1 to γn are obtained. Then, as explained in Figure 14, the contour 34 of AM2 is calculated from these corrected points γ1 to γn (step S7), and further, as explained in Figure 15, the center P1 of AM2 is calculated from the contour 34 (step S8). After that, as shown in Figure 19, the transport mechanism 19 is operated based on the center P1 of each AM2 to align the wafer W, and the wafer W is placed on the stage 21 via the pins 22. Film deposition gas is supplied to this wafer W and the film deposition process is performed (step S9).

[0061] [Regarding the illuminance of the image acquisition area] The image acquisition area 25 shown in Figure 2 can be set as a relatively large area. It is technically difficult to make the brightness uniform in such a relatively large area using illumination 17. Therefore, the illuminance of each of the four AM2s distributed in the image acquisition area 25 is different. As explained above, using HSC20 to image the AM2s is advantageous even considering such variations in illuminance. To demonstrate the advantages of using HSC20, we will describe the case in which the AM2s are imaged using the comparative example film deposition apparatus 1' equipped with the aforementioned RGB camera and the center P1 is detected. For the sake of explanation, we will assume that each AM2 is not covered with film 33 at the time of imaging. For the explanation, refer to the characteristic diagrams shown in Figures 23 to 26, which show the relationship between brightness and position in the image acquired by the RGB camera. In these characteristic diagrams, the dotted lines indicate the position of the boundary between the AM2s and the areas on the upper surface of the wafer W where AM2s are not formed (hereinafter sometimes referred to as AM non-formed areas).

[0062] When the illumination in AM2 is appropriate, the difference in brightness between the boundary between AM2 and the AM-non-formed region and the position offset from that boundary is large in the image acquired by the RGB camera, as shown in Figure 23. In other words, the boundary forms a clear contrast in the image. Consequently, this boundary becomes even clearer in the image obtained by the binarization process, and the contour 31 of AM2 acquired from the image has high accuracy in shape and position.

[0063] However, as mentioned above, the illuminance varies in each part of the image acquisition area 25, which may result in the illuminance of some of the AMs among the multiple AM2s not being appropriate. If the illuminance is too high or too low, the difference in brightness between the boundary and the position shifted from that boundary will be low, as shown in Figures 24 and 25, respectively. Consequently, the shape and position of the contour 31 obtained after binarization processing may be inaccurate. As a result, there is a concern that the center P1 of the AM2 calculated from this contour 31 may be significantly different from the actual center P0 of the AM2.

[0064] It is conceivable to perform brightness correction as an image processing step before binarization, but this brightness correction does not create a steep brightness difference between the boundary between AM2 and the AM non-formed region and a position shifted from that boundary, as shown in Figure 23. Figure 26 shows the relationship between brightness and position in an image when brightness correction is performed on an image in which the brightness difference shown in Figure 24 has been formed. As shown in the figure, when brightness correction is performed, the image becomes one in which the brightness changes smoothly with respect to changes in position. That is, in an image in which brightness correction has been performed, the position of the boundary described above becomes ambiguous, so performing such brightness correction does not solve the problem of reduced accuracy of the shape and position of contour 31 due to variations in illumination.

[0065] The brightness of each pixel in an image obtained with an RGB camera is determined by parameters of only three bands (i.e., R value, G value, and B value). However, as mentioned earlier, the HSC20 can measure more than 100 bands, and since images can be obtained for each of these bands, the brightness of each pixel can be determined using more than 100 parameters. With so many bands as parameters, even if a clear brightness contrast cannot be formed between the boundary and a position offset from the boundary in one band, a clear contrast can be formed in other bands.

[0066] Therefore, in the first embodiment, using the HSC20 to perform imaging and using the brightness of each band when performing binarization processing, as described above, allows for more accurate detection of the boundary (i.e., contour 31) compared to the comparative example. Consequently, in this embodiment using the HSC20, the influence of variations in illumination in the image acquisition area 25 can be suppressed, and the shape and position of the contour 31 can be acquired with high accuracy. Since image processing is performed on the acquired contour 31 as described in the flowchart of Figure 22, it is advantageous because the center P1 of AM2 can be detected with high accuracy.

[0067] [Regarding the setting of the reference point R] In Figure 11, it was explained that the reference point R is set in the vicinity of the temporary contour 31 of each AM2. The reason for setting the reference point R in this way is that, as described above, variations in illumination may occur in the image acquisition area 25. Setting it in the vicinity means, for example, that the distance between the film thickness detection point α and the reference point R is less than 3 cm, as described above. Also, for example, if the acquired image contains multiple AM2s as in this embodiment, the reference point R is set so that the reflectance spectrum to be compared is closer to the temporary contour 31 from among the multiple temporary contours 31 calculated for each AM2. In other words, for the reference point R and temporary contour 31 that are in a closer positional relationship among the multiple reference point Rs and multiple temporary contours 31, the reflectance spectrum C0 of the reference point R is compared with the reflectance spectrum C of the film thickness detection point α set from the temporary contour 31 so that a unique wavelength is acquired.

[0068] A detailed explanation will be given with reference to the schematic diagram of the image acquired by the HSC20 in Figure 27. In Figure 27, the two AM2s are distinguished and shown as 2A and 2B, and the provisional contours 31 obtained from these AM2A and 2B are shown as 31A and 31B, and the reference points R set for the provisional contours 31A and 31B are shown as RA and RB. The reflectance spectrum of reference point RA will be compared with the reflectance spectrum of the film thickness detection point α set on contour 31A, but RA will be set closer to contour 31A where the reflectance spectrum to be compared is acquired than contour 31B. Similarly, the reflectance spectrum of reference point RB will be compared with the reflectance spectrum of the film thickness detection point α set on contour 31B, but RB will be set closer to contour 31B where the reflectance spectrum to be compared is acquired than contour 31A. By setting a reference point R for each of the multiple AM2s in this way, the effect of the illuminance difference described above is suppressed.

[0069] Incidentally, in the explanation using structures 37A and 37B in Figures 10 and 11, the reference point R (shown as position 38B in Figures 10 and 11) was described as a point in the region where the film 33 (upper layer film 36 in Figures 10 and 11) is formed. However, as shown in Figure 28, the reference point R may also be a point in the region where the film 33 is not provided. In that case as well, a peculiar wavelength appears, similar to the case where the reference point R is a point in the region where the film 33 is formed, and a shift in the peculiar wavelength occurs due to the change in film thickness H at the film thickness detection point α.

[0070] Therefore, if there is a region on the upper surface of the wafer W in which the film 33 is not formed, the reference point R may be a point in that region. In the following description, we will describe a case in which AM2 is configured as a region of a different material or a region with an uneven pattern than the reference point R, but even in that case, the reference point R may be a point in the region not covered by the film 33.

[0071] Incidentally, the reference point R is not limited to being set to one per temporary contour 31; multiple reference points R may be set as long as the same reflectance spectrum can be obtained from each of them. Then, the unique wavelength may be detected by comparing the reflectance spectra C at multiple film thickness detection points α obtained from the temporary contour 31 with the reference reflectance spectrum C0 obtained from any of the reference points R. In other words, it is sufficient that the reflectance spectra C of each of the multiple film thickness detection points α set in the temporary contour 31 are compared with a common reference reflectance spectrum C0, and the sources from which this reference reflectance spectrum C0 is obtained do not have to be different.

[0072] Furthermore, while it has been described that a reference point R is set from the imaging data in which the film thickness detection point α is acquired and a reference reflectance spectrum C0 is obtained, the acquisition of the reference reflectance spectrum C0 is not limited to the said imaging data. By conducting a test in advance, the reference reflectance spectrum C0 can be acquired from a position near a location on the wafer W that is likely to become the film thickness detection point α and has the same or approximately the same illuminance as the film thickness detection point α, and stored in the control device 40. When aligning the wafer W, the specific wavelength may be detected using the reference reflectance spectrum C0 acquired in advance in this manner. However, as described above, acquiring the reference reflectance spectrum C0 from the imaging data is preferable from the viewpoint of eliminating the need for the above-mentioned prior test and simplifying the preparation.

[0073] [Other Examples of AM] The AM to be detected is not limited to the examples described above. For example, while the AM has been shown as a circular protrusion on the wafer W, it may also be a circular recess on the wafer W, as shown in the longitudinal cross-sectional side view of Figure 29, where the AM 51 is formed. In this case, contrary to the case where the AM is a protrusion, in the schematic diagrams shown in Figures 10 and 11, the region where the AM 51 is formed corresponds to the structure 37B with a small thickness of the underlayer film 35, and the area outside the region where the AM 51 is formed corresponds to the structure 37A with a large thickness of the underlayer film 35. However, just as in the case where the AM is a protrusion, it is possible to acquire a specific wavelength. Therefore, as in the examples shown above, it is possible to detect the contour 34 and center P1 of the AM 51 from the specific wavelength.

[0074] When formed as a recess, as in AM51, it is covered with film 33, so the provisional contour 31 is detected closer to the center of AM51 than the actual contour of AM51, unlike when formed as a convex portion, as in AM2 (see Figure 30). Figure 30 is a schematic diagram showing the image acquired by HSC20, similar to Figure 12. When detecting the center P1 of AM51, the reference point R, film thickness detection point α, corresponding point β, and correction point γ can be set in the same way as in the detection process of the center P1 of AM2, except that the positional relationship of the correction point γ with respect to the contour 31 is different from the positional relationship in the detection process of the center P1 of AM2. As previously described, in the detection process of the center P1 of AM2, the correction point γ is acquired as a point inside the provisional contour 31, but in the detection process of the center P1 of AM51, since the provisional contour 31 is located closer to the center of the contour than the actual contour of AM51, the correction point γ is acquired as a point outside the provisional contour 31. To avoid complexity, only γ1 is shown in the diagram among the correction points γ1 to γn.

[0075] In Figure 30, the film thickness detection point α is set inside the temporary contour 31 but away from it. However, it may also be set inside the temporary contour 31 but in contact with it. The reason for setting the film thickness detection point α inside the temporary contour 31 is that if the film thickness measurement point α is set on or outside the temporary contour 31, the point in the region where the film is piled up along the side wall of the recess (shown as ε in Figure 29) becomes the film thickness measurement point α, making it impossible to accurately obtain the film thickness H.

[0076] Incidentally, AM is not limited to being circular in plan view, but can be any shape. For example, AM may be rectangular, or it may be a cross in plan view, as shown in AM52 in Figure 31. In the case of AM being rectangular or a cross, as described above, it may be formed as a convex part relative to the surroundings, or as a concave part. Figure 31 is a schematic diagram of an image obtained when AM52 is formed as a convex part.

[0077] Even if the AM is cross-shaped, as in AM52, when detecting the contour 34 and center P1 while it is covered with the film 33, similar to the case of a circle, after detecting a provisional contour 31, for example, the centroid G of the region enclosed by the contour 31 can be determined, and the film thickness detection points α1 to αn and corresponding points β1 to βn can be set based on this centroid G. The centroid G can be calculated by a predetermined algorithm, such as approximating the provisional contour 31 as two intersecting wide straight lines and using the center of the intersection region of those lines as the centroid G. In the example shown in Figure 31, the film thickness detection points α1 to αn (n=12 in the illustrated example) are set to have a predetermined positional relationship with the acquired centroid G, and the points on the provisional contour 31 that are closest to the film thickness detection points α1 to αn are set as corresponding points β1 to βn. Therefore, if the provisional contour 31 is formed by straight lines, the corresponding point β is the intersection point obtained by drawing a perpendicular line from the film thickness detection point α to this straight line. The perpendicular line drawn from the film thickness detection point α1 is denoted as L2 and shown in Figure 31.

[0078] Figure 32 shows a schematic diagram of the AM52 image, similar to Figure 31. After setting the film thickness detection point α and corresponding point β as described above, and then determining the lateral film thickness J1 to Jn, the correction points γ1 to γn are calculated as shown in Figure 32. The correction points γ1 to γn can be determined by setting the corresponding points β1 to βn in advance for each corresponding point β, and then moving the corresponding points β1 to βn in different directions. As explained in the description of the detection process of the center P1 of AM2, this amount of movement corresponds to the lateral film thickness J estimated from the film thickness detection point α, which was the source of the setting of the corresponding point β.

[0079] After determining the correction points γ1 to γn, multiple straight lines extending in each direction are found to fit these correction points γ1 to γn. The contour 34 of the AM 52 formed by these lines is determined, and the centroid of this contour 34 is determined as the center P1 of the AM 52. This centroid can be calculated in the same way as the centroid G from a provisional contour 31. As described above, the contour 34 and center P1 can be obtained even if the AM is not circular. In the case of a rectangular AM, the lateral film thickness J is determined in the same way as in the case of a circle or cross, and the contour 34 and center P1 can be calculated from the correction point γ obtained by moving the corresponding point β corresponding to the film thickness detection point α from which the lateral film thickness was set.

[0080] Incidentally, in Figures 10 and 11, it was stated that when the thickness of the undercoat 35 differs between structures 37A and 37B, it is possible to acquire a singular wavelength, and that a shift in the singular wavelength occurs in accordance with the variation of the upper layer 36 of one of these structures. However, the case in which it is possible to acquire a singular wavelength and a shift in the singular wavelength occurs due to the variation of the upper layer 36 of one of the structures is not limited to when the thickness of the undercoat 35 differs between structures 37A and 37B. Hereinafter, the explanation will be given with reference to Figures 33 and 34, which are explanatory diagrams showing the relationship between the imaging target and the reflectance spectrum, similar to Figures 10 and 11.

[0081] In the example shown in Figure 33, the thickness of the undercoat 35 is the same between structures 37A and 37B. However, the undercoat 35 of structure 37A and the undercoat 35 of structure 37B are made of different materials. Even when the materials of the undercoat 35 differ between structures 37A and 37B, the specific wavelength shifts in accordance with the variation in the upper layer 36 of one of these structures 37A or 37B. Therefore, although the examples described so far have shown that the heights of the position set as the film thickness detection point α and the position set as the reference point R are different, it is not limited to the formation of a height difference at each setting position. The film thickness detection point α may be set to be located on the AM, while the reference point R may be set outside the AM and at the same height as the AM, but in a region formed of a different material.

[0082] Furthermore, as shown in Figure 34, if at least one of the underlayer films 35 has an uneven pattern, a unique wavelength appears, similar to the case where the materials of the underlayer films 35 are different, and this unique wavelength changes depending on the thickness of the upper layer film 36. In the example in Figure 34, an uneven pattern is formed on the underlayer film 35 of structure 37A, and an uneven pattern is not formed on the underlayer film 35 of structure 37B. However, the underlayer film 35 of structure 37B may have an uneven pattern of a different shape than that of the underlayer film 35 of structure 37A.

[0083] Therefore, as shown in the longitudinal cross-sectional side view of the wafer W in Figure 35, the AM (shown as 2A forming a convex portion) is not limited to having a flat top surface as previously described, but may also be a region on which an uneven pattern is formed on its top surface. The reference point may be set in a flat region outside the AM 2A that is the same height as the AM 2A, or the reference point R may be set in a region that is the same height as the AM 2A but has an uneven pattern different from the uneven pattern of the AM 2A. Figure 35 shows an example in which the reference point R is set in a region where an uneven pattern is formed in this way.

[0084] [Apparatus Configuration Without HSC] The film deposition apparatus 1 may be configured in a way that does not use an HSC 20. For example, in the film deposition apparatus 1, an RGB camera as described in the comparative example is provided instead of the HSC 20, so that a provisional contour 31 can be obtained from the image obtained by the RGB camera. In addition to this RGB camera, a spectrometer equipped with a probe and measuring instrument, and a moving mechanism are provided in the film deposition apparatus 1. The probe is configured to be able to move horizontally to any position on the window 16 by the moving mechanism. This probe receives reflected light from a measurement point below the probe, and this reflected light is guided to the measuring instrument body. This light is spectrally separated in the measuring instrument body, and the spectrally separated light of each wavelength is incident on an optical element 28 provided in the measuring instrument body. The output of the optical element 28 is transmitted to a control device 40, which can obtain the reflectance spectrum at the measurement point. In this example of apparatus configuration, the RGB camera corresponds to the imaging unit, and the spectrometer and control device 40 correspond to the spectrum acquisition unit.

[0085] Assuming that AM2 is imaged using an RGB camera, once the film thickness detection points α are determined based on the provisional contour 31 described above, the probe is moved to receive light from each film thickness detection point α and the reference point R, and the reflectance spectra C1 to Cn of the film thickness detection points α1 to αn and the reflectance spectrum C0 of the reference point R are obtained. After obtaining the reflectance spectra in this manner, the contour 34 of AM2 and the center P1 can be detected in the same way as when using the HSC20. However, when using an apparatus configuration that uses an RGB camera and a spectrometer without providing the HSC20, it is necessary to perform the imaging operation for obtaining the provisional contour 31 with the RGB camera and the measurement operation for obtaining the reflectance spectrum with the spectrometer in sequence. On the other hand, when using the HSC20, the imaging operation and the measurement operation can be performed together, so the time required from imaging the image acquisition area 25 to placing the wafer W on the stage 21 can be made relatively short. Therefore, the throughput of the film deposition apparatus 1 can be improved.

[0086] [Other supplementary information regarding the first embodiment] It has been explained that the reflectance spectrum is obtained as the characteristic spectrum of reflection by the HSC20, and that the singular wavelength is identified from this reflectance spectrum. However, instead of the reflectance spectrum, the intensity spectrum of reflected light may be obtained as the characteristic spectrum of reflection, and the singular wavelength may be identified from this spectrum. In other words, the parts that have been referred to as the reflectance spectrum in the above explanation can be read as the intensity spectrum of reflected light, and the vertical axis in the spectral data exemplified in Figure 10, etc., can be the intensity of reflected light instead of reflectance. The singular wavelength can be the wavelength at which the difference in intensity of reflected light between the intensity spectra of reflected light is minimized.

[0087] Furthermore, although it has been described that the wafer W is aligned after determining the center P1 of the AM, this alignment does not have to be performed. For example, the distance between the center P1 of each AM2 and the wafer W when the wafer W is placed on the stage 21 is obtained by the pins 22, and if each distance is within the acceptable range, the wafer W is considered normal, while if any distance is outside the acceptable range, the wafer W is considered abnormal, and the control device 40 marks the wafer W. In other words, instead of alignment, the detected center P1 of the AM2 may be used to determine an abnormality regarding the placement position of the wafer W. Note that this determination of an abnormality in the placement position is not limited to determining an abnormality in the distance to the center P1 of the AM2, but may also be performed to determine an abnormality in the distance to any point on the contour 34. Therefore, although it has been explained that the center P1 of the AM is detected from the contour 34, the detection of the center P1 does not have to be performed after the detection of the contour 34. Also, although four AMs are provided in the example shown in Figure 2, it is sufficient to provide the number necessary for the alignment of the wafer W and the determination of an abnormality in the placement position, and it is not limited to four.

[0088] Although it was stated that a provisional contour 31 is obtained by binarization, this binarization process is not essential for obtaining the provisional contour 31. For example, if it is confirmed by prior testing that a step difference between AM2 and the AM-non-formed region is obtained with relatively clear contrast from the surrounding area in the image obtained from a specific band, then the step difference in that image can be determined as the provisional contour 31. However, it is preferable to perform binarization because it makes the settings of each point, such as the film thickness detection point α, more appropriate.

[0089] While it was stated that the HSC 20 can move using the movement mechanism 29 to perform imaging and acquire imaging data for the image acquisition area 25, the HSC 20 is not limited to moving in this manner when acquiring imaging data. In other words, if the HSC 20 has a relatively wide field of view, imaging data may be acquired by imaging the entire image acquisition area 25 at once without moving.

[0090] [Second Embodiment] In the second and subsequent embodiments described below, the technologies described in the first embodiment can be applied as appropriate, even without special mention. When applying these technologies, the configuration of each part of the device can be changed as appropriate. As a second embodiment, a joining device 6, which is an example of an alignment device incorporating an AM detection device, will be described with reference to the cross-sectional plan view and longitudinal front view in Figures 36 and 37, respectively. In the description of this joining device 6, the X direction is the left-right direction, and the Y direction is the front-back direction.

[0091] The bonding apparatus 6 holds wafer W1 on the upper side and wafer W2 on the lower side, both in a horizontal position, and positions the semiconductor device formation surfaces 50 of wafers W1 and W2 facing each other. Then, it detects the AM53 and AM54 provided on the device formation surfaces 50 of wafer W1 and wafer W2, respectively, and performs alignment to change the relative position of wafers W1 and W2 in a plan view. After that, wafers W1 and W2 are brought close together and bonded. This alignment and proximity between wafers W1 and W2 is performed by the relative movement of stages 74 and 75 that adsorb and hold the back surfaces (main surfaces opposite to the device formation surfaces 50) of wafers W1 and W2. Wafers W1 and W2 are transported to the bonding apparatus 6 in a pre-treated state so that this bonding can be performed. Furthermore, wafers W1 and W2 correspond to the first and second components, respectively, and AM53 and AM54 correspond to the first and second alignment marks, respectively.

[0092] AM53 and AM54 are coated with a film (not shown) by processing performed on wafers W1 and W2 before they are transported to the bonding apparatus 6. Therefore, the alignment between wafers W1 and W2 is performed by acquiring imaging data of the device formation surface 50 of wafers W1 and W2 using the HSC20, and detecting the contours 34 and center P1 of AM53 and AM54 from this imaging data, as in the first embodiment. In the following description, the wafer W into which wafers W1 and W2 have been bonded may be referred to as bonded wafer W3. Also, wafers W1 and W2 before bonding and bonded wafer W3 may be referred to simply as wafer W without distinguishing between them.

[0093] The bonding apparatus 6 comprises a housing 61. A transport opening 63, which is opened and closed by a shutter 62, is formed in the rear side wall of the housing 61. Inside the housing 61, an inner wall 65 with a wafer W passage 64 divides the area into a transport region 66 on the right and a bonding region 67 on the left. The transport opening 63 opens into the transport region 66. In the transport region 66, a mounting section 68, a transport mechanism 71, a reversing mechanism 72, and a holding position adjustment mechanism 73 are provided in order from rear to front. A transport mechanism (not shown) provided outside the housing 61 transfers the wafer W to the mounting section 68.

[0094] The transport mechanism 71 includes a base 71A that rotates around the Z-axis (vertical axis) and a wafer support 71B that moves back and forth on the base 71A, enabling the transfer of wafers W between the mounting section 68, the inversion mechanism 72, the holding position adjustment mechanism 73, and the stage 74 of the bonding region 67. The holding position adjustment mechanism 73 includes a rotating section 73A that supports and rotates the back surfaces of wafers W1 and W2, and an optical sensor 73B that irradiates light onto the peripheral edges of wafers W1 and W2 during this rotation. The light irradiation by the optical sensor 73B is performed to detect notches, which are cutouts provided on the peripheral edges of wafers W. After detecting these notches, the transport mechanism 71 receives wafers W1 and W2 so that the notches are oriented in a predetermined direction, and transports them to the stage 74 and the inversion mechanism 72, respectively.

[0095] The inversion mechanism 72 includes a holding arm 72A that can rotate around a horizontal axis while holding the wafer W1 received from the transport mechanism 71, and a drive mechanism 72B. The drive mechanism 72B can rotate the holding arm 72A to transport the wafer W between the transport area 66 and the bonding area 67, and can also raise and lower the holding arm 72A. By using this inversion mechanism 72, when the wafer W1 is transferred from the transport mechanism 71 to the stage 75, the device forming surface 50 can be inverted from an upward state to a downward state.

[0096] In the bonding region 67, a stage 74 for holding wafer W2 and a stage 75 for holding wafer W1 are provided on the lower and upper sides, respectively, and the wafers W2 and W1 held on the stages 74 and 75 face each other. The stage 74 is connected to a drive mechanism 76 which forms a position adjustment mechanism for the stage 74 and the wafer W2 held on the stage 74. The drive mechanism 76 is composed of a horizontal moving part 76A and a rotating part 76B, with the rotating part 76B connected to the horizontal moving part 76A and the stage 74 connected to the rotating part 76B. The horizontal moving part 76A is a so-called XY stage and moves the stage 74 in the X and Y directions, respectively, together with the rotating part 76B. The rotating part 76B rotates the stage 74 in the direction of the vertical axis (θ direction). Therefore, the stage 74 is movable in the X, Y, and θ directions.

[0097] On the other hand, stage 75 is movable in the Z direction by being connected to a lifting mechanism 77 that raises and lowers stage 75. HSC20 is provided on stages 74 and 75, respectively. If the HSC20 provided on stages 74 and 75 are designated as HSC20A and 20B, respectively, HSC20A images the device formation surface 50 of wafer W1 on stage 75, and HSC20B images the device formation surface 50 of wafer W2 on stage 74.

[0098] As the bonding apparatus 6 is configured as described above, wafer W1 is transported in the order of mounting section 68 → holding position adjustment mechanism 73 → inversion mechanism 72 → stage 75, and wafer W2 is transported in the order of mounting section 68 → holding position adjustment mechanism 73 → stage 74. Then imaging is performed by HSC 20A and HSC 20B, and the positions of the stage 74 in the X, Y, and θ directions are adjusted based on the acquired images AM 53 and AM 54. That is, the wafer W2 is aligned with wafer W1 by horizontal movement. After that, the stage 75 descends, and the device formation surfaces 50 of wafers W1 and W2 are joined together to form a bonded wafer W3. The bonded wafer W3 is transported from the stage 74 to the mounting section 68 following the reverse path of the path through which wafer W2 was transported, and is discharged from the bonding apparatus 6.

[0099] Figures 38 and 39 are plan views showing the positional relationship of the device formation surfaces 50 of wafers W1 and W2 to illustrate the above alignment example, with Figures 38 and 39 showing the state before and after alignment, respectively. Each chip formation region on the device formation surface 50 of wafer W1 is shown as 50A, and each chip formation region on the device formation surface 50 of wafer W2 is shown as 50B. AM53 and AM54 are shown as similar square recesses.

[0100] As described above, imaging is performed using HSC20A and 20B to obtain provisional contours 31 in AM53 and AM54. After that, corrected contours 34 and centers P1 are detected in AM53 and AM54 in the same manner as in the first embodiment. For example, in a plan view, the alignment may be performed so that each contour 34 of one AM53 and AM54 is contained within each contour 34 of the other AM. In the examples of Figures 38 and 39, the contour 34 of AM54 on wafer W2 is larger than the contour 34 of AM53 on wafer W1, and the alignment is performed so that the contour 34 of AM53 is contained within the contour 34 of AM54 as wafer W2 moves. In addition, the method of aligning wafers W1 and W2 using the contours 34 or centers P1 of AM53 and AM54 can be determined as appropriate, such as aligning so that the distance to the centers P1 of each AM53 and each AM54 is within an acceptable range in a plan view. Furthermore, even when aligning using the respective center P1 of AM53 and AM54, the center P1 is obtained from the contours 34 of AM53 and AM54, so it can be said that the alignment is performed based on the contours 34 of AM53 and AM54.

[0101] The contours 34 or center P1 of AM53 and AM54 are obtained with high accuracy, similar to the first embodiment, by canceling out the influence of the film covering these AMs. Therefore, misalignment between wafers W1 and W2, which is performed based on these contours 34 or center P1, is suppressed. As a result, a decrease in the yield of products manufactured from the bonded wafer W3 is suppressed. As shown in the second embodiment above, the scope of application of this technology is not limited to the apparatus for aligning wafers W with respect to the stage 21 shown in the first embodiment.

[0102] [Third Embodiment] Next, bonding devices 8A and 8B that perform bonding using AM according to the third embodiment will be described with reference to Figures 40 and 41, which are side views of the respective devices. Note that the description of the AM configuration will be omitted from this third embodiment onward, but for example, it can be configured as described above. In the third embodiment, an electronic component, a chip 59 that has an electronic circuit formed by including circuit elements such as resistors, capacitors and / or coils, is bonded to a substrate 58. Bonding device 8A connects a plurality of chips 59 onto a chip-holding substrate 56 on which a resin layer 55 for fixing the chips is formed, and then forms a bonded body 57 by curing the resin layer 55. Bonding device 8B performs bonding to bond the chips 59 held in the bonded body 57 to the substrate 58. Bonding devices 8A and 8B have different objects to hold and bond, but are configured in general the same way as the bonding device 6 of the second embodiment.

[0103] Further explanation will be given regarding the bonding device 8A. The bonding device 8A includes a stage 81 on which the non-formed surface of the resin layer 55 on the chip holding substrate 56 is placed and which holds the non-formed surface by suction, and a chip holding unit 82 above the stage 81 which holds the chip 59 by suction with the non-bonding surface 59B of the chip 59 facing downward. The stage 81 and the chip holding unit 82 can move relative to each other by being connected to a drive mechanism 76 and a lifting mechanism 77, respectively, similar to the stages 74 and 75 in the second embodiment. The drive mechanism that moves the stage 81 in the X, Y, and θ directions corresponds to the position change mechanism. AM56A and AM56B are formed on the formation surface of the resin layer 55 on the chip holding substrate 56. AM59A is formed on the non-bonding surface 59B of the chip 59.

[0104] An HSC20 is provided below the stage 81 to image AM59A and AM56A. To enable imaging of AM59A and AM56A, the stage 81 is provided with holes (not shown). The chip holding substrate 56 is made of a substrate that transmits visible light, such as a glass substrate, and the resin layer 55 is also made of a material that transmits visible light, just like the chip holding substrate 56.

[0105] As described in the second embodiment, the stage 81 is moved based on AM59A and AM56A in the imaging data to align the chip holding substrate 56 with the chip 59, and then the chip 59 is lowered to connect to the uncured resin layer 55. Note that AM59A of the chip 59 is covered with a film, similar to the AM in the previous embodiments, and this film is not made of a transparent material like the resin layer 55. Even though AM59A is covered with a film, the contour 34 and center P1 of AM59A can be detected with high accuracy, so the misalignment of the chip 59 with respect to the chip holding substrate 56 is suppressed. AM56A is imaged through the chip holding substrate 56, but since this chip holding substrate 56 is transparent, contour correction like that for AM59A is not required. After the chip 59 is connected to the resin layer 55 in this way, the resin layer 55 is cured by irradiating it with ultraviolet light from an ultraviolet irradiation unit (not shown), and a bonded body 57 is formed in which the chip 59 is fixed to the chip holding substrate 56. The chip 59 corresponds to the first component, and the chip holding substrate 56 corresponds to the second component.

[0106] Next, the bonding device 8B will be further described. The bonding device 8B includes a stage 83 that holds the substrate 58 by suction with the bonding surface of the chip 59 on the substrate 58 facing upward, and a stage 84 that holds the chip-holding substrate 56 of the bonding body 57 by suction above the stage 83 with the chip 59 facing downward. The stages 83 and 84 can move relative to each other by being connected to a drive mechanism 76 and a lifting mechanism 77, respectively, similar to the stages 74 and 75 in the second embodiment. The drive mechanism that moves the stage 83 in the X, Y, and θ directions corresponds to the position change mechanism.

[0107] AM58B is formed on the bonding surface of the chip 59 on the substrate 58. The HSC20, which is provided above the stage 84, images the AM58B of the substrate 58 and the AM56B of the bonding body 57 through a hole (not shown) provided in the stage 84. As mentioned above, the chip holding substrate 56 and the resin layer 92A can transmit visible light, so the AM56B and AM58B can be imaged in this way.

[0108] As described in the second embodiment, the stage 83 is moved based on AM56B and AM58B in the imaging data to align the substrate 58 and the bonding body 57, and then the bonding body 57 is lowered to bond the chip 59 to the substrate 58. The AM58B of the substrate 58 is covered with a film, similar to the AM in the previous embodiments, and this film is not made of a transparent material like the resin layer 55. Even though the AM58B is covered with a film, the contour 34 and center P1 of the AM58B can be detected with high accuracy, so the misalignment of the chip 59 on the substrate 58 is suppressed. The AM56B is imaged through the chip holding substrate 56, but since this chip holding substrate 56 is transparent, contour correction like that for AM58B is not required. After the connection of the chip 59 to the substrate 58, the resin layer 92A is heated by a heating unit (not shown) to peel the chip holding substrate 56 from the chip 59. The substrate 58 corresponds to the first component, and the chip holding substrate 56 corresponds to the second component.

[0109] [Fourth Embodiment] The substrate processing apparatus 9 according to the fourth embodiment will be described with reference to the side view of Figure 42. This substrate processing apparatus 9 is a device for supplying a coating solution to a substrate 91, which is a substrate for manufacturing organic EL (electroluminescent) or color filters, and forming a coating film, similar to the apparatus described in Patent Document 1 above.

[0110] The substrate processing apparatus 9 includes a stage 92 on which a substrate 91, which is a first component, is placed; a drive mechanism 93; a plurality of nozzles 94A to 94C positioned above the stage 92; coating liquid supply mechanisms 95A to 95C, which include valves and a source for supplying coating liquid to each of the nozzles 94A to 94C; and an HSC 20. The drive mechanism 93 can move the stage 92 in the X, Y, and θ directions, similar to the drive mechanism 76 of the second embodiment.

[0111] A partition wall 96 called a bank is formed on the surface of the substrate 91, and this partition wall 96 forms a plurality of recesses. Red, blue, and green coating liquids 97A, 97B, and 97C are discharged from the second members, nozzles 94A, 94B, and 94C, respectively, into each recess to form a coating film. Before the discharge of these coating liquids 97A to 97C onto the substrate 91, the AM 91A on the surface of the substrate 91 is imaged by the HSC 20 provided above the stage 92. The AM 91A is assumed to be covered with a film 98. From the imaging data, the contour 34 and center P1 of the AM 91A are detected, similar to the first embodiment. Then, by moving the stage 92 based on the contour 34 or center P1 of AM91A, the positional relationship of the substrate 91 with respect to the nozzles 94A to 94C is adjusted, and then the coating liquid 97A to 97C is discharged from the nozzles 94A to 94C onto the substrate 91, preventing the coating liquid 97A to 97C from being supplied to the outside of the recess.

[0112] As described above, this technology can also be applied when aligning a substrate with a processing unit that performs processing on the substrate, as exemplified by nozzles 94A to 94C. In this embodiment, the processing units, nozzles 94A to 94C, correspond to the second member, and the substrate 91 corresponds to the first member.

[0113] [Supplementary explanation of the device configuration] Although the HSC20 has been described as receiving visible light (wavelength 380 nm to 780 nm) to obtain imaging data, the HSC20 may also be configured to receive infrared light (wavelength 780 nm to 2500 nm) to obtain imaging data. Therefore, although the bonding apparatus 8A and 8B of the third embodiment shown in Figures 40 and 41 has been described as being constructed of a material that can transmit visible light in order to image the chip 59 and substrate 58 which are further away from the HSC20 than the chip holding substrate 56, it is not limited to being constructed of such a material. For example, a substrate such as a Si (silicon) substrate can be used for the chip holding substrate 56.

[0114] Up until now, the HSC20 and the probe of the spectrometer have been described as receiving reflected light reflected from an object and obtaining a reflectance spectrum or reflectance spectrum. However, in implementing this technology, obtaining such a spectrum related to reflected light is not limited to that. Figure 43 is a side view showing a modified example of the bonding apparatus 8B of the third embodiment. A light source 89 is provided below the stage 83, and infrared light is irradiated from the light source 89 at a predetermined intensity toward the region where the AM58B of the substrate 58 is formed and the region where the AM56B of the chip holding substrate 56 is formed, which overlap each other in a plan view. The HSC20, positioned above the stage 84, receives the transmitted light that has passed through each of these regions. In this case, each optical element 28 of the HSC20 receives light at a light intensity corresponding to the absorbance in each part of each region, so the control device 40 can obtain the spectrum related to absorbance shown in the figure.

[0115] Specifically, the absorbance spectrum is an absorbance (absorption coefficient) spectrum or an absorbance intensity spectrum. When obtaining an absorbance spectrum, a specific light intensity at which the absorbance reaches 100% is determined in advance, and the spectrum can be obtained from the absorbance obtained by calculating the ratio of the light intensity detected by each optical element 28 to that specific light intensity. When obtaining an absorbance spectrum or absorbance intensity spectrum in this way, a singular wavelength can be obtained in the same way as when obtaining a reflectance spectrum or reflectance intensity spectrum. Therefore, the singular wavelength should be the wavelength at which the difference between the absorbance spectrum at the film thickness detection point α and the absorbance spectrum at the reference point R is minimized. In the figure, the singular wavelength is shown as E2. After obtaining the singular wavelength of the absorbance spectrum in this way, the lateral film thickness of the AM can be detected in the same way as when obtaining the singular wavelength from the reflectance spectrum. As described above, the characteristic spectra that can be used in this technology include reflectance spectra and reflectance intensity spectra, which are spectra related to reflectance, and absorbance spectra and absorbance intensity spectra, which are spectra related to absorbance.

[0116] Furthermore, although the substrate processing apparatus to which the wafer W is transported in the first embodiment was described as a film deposition apparatus, this substrate processing apparatus may be a substrate processing apparatus other than a film deposition apparatus, such as an etching apparatus. The etching apparatus can have the same configuration as the aforementioned film deposition apparatus 1, except that, for example, etching gas is supplied as the processing fluid instead of film deposition gas to etch the surface of the wafer W. Even if a misalignment of the wafer W occurs relative to the transport mechanism 19 when the wafer W is transported to a substrate processing apparatus other than a film deposition apparatus, the wafer W can be placed on the placement area 24 by aligning the wafer W with respect to the stage 21 as described in the first embodiment. When the wafer W is placed in the correct position in this way, the positions to which the processing fluid is supplied to each part of the wafer W become correct, and the temperature distribution of the wafer W formed by the heater of the stage 21 becomes correct, so that appropriate processing can be performed on each part of the surface of the wafer W. It should be noted that the embodiments disclosed herein should be considered as illustrative in all respects and not restrictive. The above embodiments may be omitted, replaced, modified and combined in various forms without departing from the scope and spirit of the attached claims.

[0117] [Evaluation Tests] The evaluation tests conducted in relation to this technology are described below. ・Evaluation Test 1 As part of Evaluation Test 1, multiple cut pieces of other Si substrates (hereinafter referred to as Si pieces) were placed on a Si substrate on which a copper film was formed on its surface. A ZrO film of different thicknesses was deposited on the surface of each Si piece, with thicknesses of 4 nm, 5 nm, or 6 nm. The Si substrate was imaged using HSC20. A reflectance spectrum C was obtained at points on each Si piece, designated as film thickness detection points α, while a reference reflectance spectrum was obtained at points on the Si substrate where no Si pieces were formed, designated as reference points R. As described in the embodiment, the specific wavelength at each film thickness detection point α was detected by comparing the reflectance spectrum of each film thickness detection point α with the reference reflectance spectrum.

[0118] The graph in Figure 44 shows the results of Evaluation Test 1, with the horizontal axis representing the thickness of the ZrO film on the Si piece set as the film thickness detection point α (unit: nm), and the vertical axis representing the singular wavelength (unit: nm). As shown in this graph, there was an approximately linear relationship between the singular wavelength obtained from each film thickness detection point α and the thickness of the ZrO film at each film thickness detection point α. Therefore, Evaluation Test 1 demonstrated that it is possible to detect film thickness H based on the singular wavelength, as described in the embodiment.

[0119] Furthermore, in this evaluation test 1, although the difference in film thickness between the three Si pieces is very small, the unique wavelengths of each film thickness detection point α are detected as different wavelengths from each other. Therefore, as explained in the first embodiment, the detection accuracy of film thickness H is high when obtaining film thickness H from the unique wavelengths. Consequently, the lateral film thickness J can also be calculated with high accuracy, and this evaluation test 1 confirmed that the contour and center point of AM can be calculated with high accuracy using this technology.

[0120] • Evaluation Test 2: Evaluation Test 2 was conducted in much the same manner as Evaluation Test 1, but the distance between the Si substrate on which the copper film was formed and the HSC20 was changed from the distance in Evaluation Test 1. As a result, imaging was performed with a focus shift (out of focus) between the HSC20 and the Si substrate.

[0121] In this evaluation test 2, the unique wavelengths obtained from each film thickness detection point α on the Si piece were approximately the same as the unique wavelengths obtained from the film thickness detection point α in evaluation test 1. Therefore, it was confirmed that this technology is preferable because it can suppress the occurrence of detection errors of the AM2 contour 34 and the center P1 of the AM2 due to the focus shift described above.

[0122] • Evaluation Test 3 A wafer W made of Si was placed in a predetermined location, and light was irradiated onto its surface from an illumination source. The surface was then imaged using HSC20. The reflectance intensity at each position in the region along the diameter of the wafer W was obtained from the image data, and it was confirmed that the reflectance intensity differed between these positions. In other words, it was confirmed that the illuminance differed at each position in the region along the diameter of the wafer W in this imaging environment. The difference between the maximum and minimum reflectance values ​​was greatest at a wavelength of 530 nm, with the maximum value being 2750 and the minimum value being 1800.

[0123] A wafer W was prepared, composed of Si with a 60 Å thick SiN (silicon nitride) film formed on its surface to create a textured pattern. Imaging was performed in this imaging environment. In evaluation test 3-1, points along the diameter of the wafer W surface with relatively high illuminance and a textured pattern were designated as film thickness detection points α, while points without the textured pattern were designated as reference points R. Reflectance intensity spectra were acquired from each point and unique wavelengths were detected. In evaluation test 3-2, points along the diameter of the wafer W surface with relatively low illuminance and a textured pattern were designated as film thickness detection points α, while points without the textured pattern were designated as reference points R. Reflectance intensity spectra were acquired from each point and unique wavelengths were detected.

[0124] As a result, the singular wavelength showed the same value at 645.3 nm in evaluation tests 3-1 and 3-2. The reflectance corresponding to the singular wavelength of 645.3 nm in evaluation test 3-1 was 360, and the reflectance corresponding to the singular wavelength of 645.3 nm in evaluation test 3-2 was 280. Therefore, these test results confirm that even with relatively large differences in illuminance at each film thickness detection point α, the influence on the singular wavelength is suppressed. This means that the singular wavelength can be detected even when the resolution of the acquired image is relatively low due to relatively low illuminance, and from this perspective, this technology is desirable.

[0125] C Reflectance spectrum C0 Reference reflectance spectrum H Film thickness W Wafer α Film thickness detection point R Reference point 2 Alignment mark (AM) 21 Stage 31 Provisional contour 33 Film 34 Contour

Claims

1. A detection method for detecting alignment marks provided on a first member for detecting the relative position of a second member with respect to a first member, comprising: a hypothetical step of imaging the alignment marks covered with a film to detect a provisional contour of the alignment marks; a spectrum acquisition step of acquiring characteristic spectra showing the reflection or absorption characteristics at a plurality of detection points set based on the provisional contour; a film thickness estimation step of estimating the film thickness at each detection point based on wavelengths determined for each detection point by comparing each characteristic spectrum with a reference characteristic spectrum which is a common comparison target for each characteristic spectrum; and a contour detection step of correcting the provisional contour based on the film thickness estimated at each detection point to detect the contour of the alignment marks.

2. The alignment mark detection method according to claim 1, wherein the characteristic spectrum is a spectrum of reflectance, reflected light intensity, absorbance, or absorbance intensity.

3. The alignment mark detection method according to claim 2, wherein the wavelength determined for each detection point is the wavelength at which the difference in any of the reflectance, reflected light intensity, absorbance, or absorbance intensity is minimized between the characteristic spectrum obtained from the detection point and the reference characteristic spectrum.

4. The alignment mark detection method according to claim 3, wherein the contour detection step is a step of detecting the contour of the alignment mark based on correction points obtained by moving each corresponding point in the provisional contour associated with each detection point by a distance corresponding to the film thickness at each detection point.

5. The alignment mark detection method according to claim 4, wherein the alignment mark is a convex or concave portion, and in the case of a convex portion, the correction point is obtained inside the provisional contour, and in the case of a concave portion, the correction point is obtained outside the provisional contour.

6. The alignment mark detection method according to claim 1, wherein the imaging of the alignment marks is performed by a hyperspectral camera, and the spectrum acquisition step is a step of acquiring the characteristic spectrum from the light spectrally separated by the hyperspectral camera.

7. The alignment mark detection method according to claim 6, wherein the assumption step includes the step of obtaining the provisional contour from an image obtained by performing a binarization process on an image obtained by capturing the alignment marks with the hyperspectral camera.

8. The alignment mark detection method according to claim 1, wherein the detection point is set as a point inside the provisional contour.

9. The alignment mark detection method according to claim 1, wherein the reference characteristic spectrum is a characteristic spectrum obtained from an outer reference point with respect to the provisional contour.

10. The alignment mark detection method according to claim 9, wherein the distance between each of the reference point and the plurality of detection points is less than 3 cm.

11. The alignment mark detection method according to claim 9, wherein there are multiple alignment marks to be imaged, the reference point is set for each of the provisional contours, and in the film thickness estimation step, a comparison is made between the optical spectrum of the reference point and the optical spectrum of the detection point set based on the provisional contour for the reference point and the provisional contour that are in a closer positional relationship among the multiple reference points and the multiple provisional contours.

12. The alignment mark detection method according to claim 1, further comprising a position change step of changing the relative position between the first member and the second member based on the contour of the alignment mark detected in the contour detection step.

13. The alignment mark detection method according to claim 12, wherein the first member is a substrate, and the second member is a stage on which the substrate is placed for processing.

14. Alignment mark detection method according to claim 12, wherein one of the first member and the second member is a substrate, and the other of the first member and the second member is a processing unit that performs processing on a substrate, a chip, or a substrate which is one of the first member and the second member.

15. The alignment mark detection method according to claim 12, wherein the alignment marks include a first alignment mark provided on the first member and a second alignment mark provided on the second member, the assumption step, the spectrum acquisition step, the film thickness estimation step and the contour detection step are performed on the first alignment mark and the second alignment mark to detect their respective contours, and the position change step is a step of changing the relative position of the first member and the second member based on the contours of the first alignment mark and the second alignment mark.

16. An alignment mark detection device for detecting alignment marks provided on a first member for detecting the relative position of a second member with respect to a first member, comprising: an imaging unit for imaging the alignment marks covered on a film; a spectrum acquisition unit for detecting a provisional outline of the alignment marks from imaging data acquired by imaging and acquiring characteristic spectra showing the reflection or absorption characteristics at a plurality of detection points set based on the provisional outline; an estimation unit for estimating the film thickness at each detection point based on a wavelength determined for each detection point by comparing each characteristic spectrum with a reference characteristic spectrum which is a common comparison target for each characteristic spectrum; and a detection unit for correcting the provisional outline based on the film thickness estimated at each detection point and detecting the outline of the alignment marks.

17. A member alignment device comprising: an alignment mark detection device according to claim 16; and a position changing mechanism that changes the relative position between the first member and the second member based on the contour of the alignment mark.