Inspection method, inspection device, and substrate processing device

Abstract

The present invention determines an arrangement state of a substrate through a spectrum indicating characteristics of light absorption or reflection, the spectrum being obtained by irradiating the substrate with light. This inspection method comprises: a step in which a measurement area set on a substrate is irradiated with light to measure a characteristic spectrum such as a reflectance spectrum; and a step in which, when the characteristic spectrum changes according to the structure of the substrate in the measurement area, a training model generated through training on the characteristic spectrum for a plurality of substrates in association with the thicknesses of the substrates or film thicknesses when a film is formed on the surface of each of the substrates, is used to determine the arrangement state of the substrate in the measurement area from the structure of the substrate, the structure thereof being identified by comparing a spectrum to be inspected, which is the characteristic spectrum measured in the step for measuring the characteristic spectrum, with a verification spectrum, which is the characteristic spectrum obtained from the training model.

Description

Inspection method, inspection device, and substrate processing device

[0001] The present disclosure relates to an inspection method, an inspection apparatus, and a substrate processing apparatus.

[0002] In a substrate processing apparatus, after a film is formed on a substrate such as a wafer, the film may be inspected to determine whether the film has a predetermined film thickness and whether there are any abnormalities such as voids inside the film. Patent Document 1 describes a film thickness measurement unit for measuring the film thickness of a thin film or the like, which irradiates a laser beam toward the wafer and evaluates the thin film by using spectroscopic ellipsometry to measure the light reflected from the wafer.

[0003] Japanese Patent Application Laid-Open No. 2007-56336

[0004] The present disclosure provides a technique that can determine the placement state of a substrate based on a spectrum that indicates the absorption or reflection characteristics obtained by irradiating the substrate with light.

[0005] The method disclosed herein includes the steps of: irradiating a measurement area set on a substrate with light, and measuring a characteristic spectrum that indicates the absorption or reflection characteristics for the wavelength of the light; and, when the characteristic spectrum changes depending on the structure of the substrate in the measurement area, using a learning model generated by learning on a plurality of substrates to associate the characteristic spectrum with the film thickness, if any, of the substrate, or the thickness of the substrate, and determining the placement state of the substrate in the measurement area from the structure of the substrate identified by comparing the inspection target spectrum, which is the characteristic spectrum measured in the step of measuring the characteristic spectrum, with a reference spectrum, which is the characteristic spectrum obtained from the learning model based on the film thickness or substrate thickness identified from the substrate from which the inspection target spectrum was obtained.

[0006] According to the present disclosure, the arrangement state of a substrate can be determined from a spectrum indicating the absorption or reflection characteristics obtained by irradiating the substrate with light.

[0007] 12. A structural diagram showing an inspection apparatus according to the first embodiment. A configuration diagram showing a control device in the first embodiment. A partial plan view showing positioning of a measurement area in a comparative embodiment. A longitudinal sectional side view of a substrate W showing a process of moving a measurement area to an inspection area. A graph illustrating the reflectance spectra in (a) to (c) of FIG. 4. A graph showing change in wavelength of a singular point with respect to the film thickness of a SiN film. A partial plan view showing positioning of a measurement area in the first embodiment. A flowchart showing a generation operation. A flowchart showing a determination operation and a movement operation. A longitudinal sectional side view illustrating a substrate having another film. A graph showing reflectance spectra at each measurement position shown in FIG. 10. A longitudinal sectional side view illustrating a substrate on which a taper is formed. A graph showing reflectance spectra at each measurement position shown in FIG. 12. A partial plan view showing positioning in a modified example of the first embodiment. A flowchart showing a generation operation in the modified example. A flowchart showing a determination operation and a movement operation in the modified example. A plan view showing a film formation processing system. A longitudinal sectional side view showing a film formation module. A plan view showing a state in which a measurement area is positioned to an alignment mark in a second embodiment. 10 is a plan view showing a state in which a measurement region is misaligned from an alignment mark in a second embodiment. FIG. 11 is a longitudinal side view illustrating positioning to an alignment mark. FIG. 12 is a longitudinal side view of substrates showing an upper substrate and a lower substrate in a state before they are bonded together. FIG. 13 is a longitudinal side view showing an upper substrate and a lower substrate bonded together in an aligned state. FIG. 14 is a longitudinal side view showing an bonding apparatus. FIG. 15 is a structural diagram showing an inspection apparatus according to a fourth embodiment. FIG. 16 is an explanatory diagram showing the principle of spectroscopic measurement using HSC. FIG. 17 is a partial plan view showing a measurement region where spectroscopic measurement of the fourth embodiment is performed. FIG. 18 is a graph showing characteristic spectra by spectroscopic measurement of the fourth embodiment. FIG. 19 is a graph showing reflectance spectra when spectroscopic measurement is performed at different illuminances. FIG. 19 is a longitudinal side view showing an example of a film deposition module provided with an inspection apparatus using HSC. FIG. 19 is a cross-sectional plan view showing the film deposition module. FIG. 19 is a reflectance spectrum showing the results of Experiment 1. FIG. 19 is a table summarizing the conditions for spectroscopic measurement of Experiment 2. FIG. 19 is a first graph showing the results of Experiment 2. FIG. 19 is a second graph showing the results of Experiment 2.10 is a top perspective view showing a turntable on which a substrate to be subjected to spectroscopic measurement in Experiment 3 is placed. FIG. 11 is a first graph showing the experimental results of Experiment 3. FIG. 12 is a second graph showing the experimental results of Experiment 3.

[0008] First Embodiment An inspection device 1 according to the first embodiment will be described in detail below with reference to the drawings. The inspection device 1 according to the first embodiment has a function of guiding measurement light for performing spectroscopic measurement of a substrate W to a correct irradiation position. FIG. 1 is a structural diagram showing the inspection device 1 according to this embodiment. FIG. 2 is a configuration diagram showing a control device 100 according to this embodiment. The inspection device 1 according to this embodiment is disposed, for example, in a film formation processing system 10 that performs a film formation process described below, and inspects the substrate W after film formation together with the control device 100 that controls the film formation process of the film formation processing system 10.

[0009] As shown in FIG. 3 (plan view) and FIG. 4( a) (cross section), the substrate W to be inspected has a base layer 31 constituting the surface of a semiconductor wafer, with a concave-convex pattern 32 formed by etching or the like. A surface film 33 is then formed on the concave-convex pattern 32 by a film formation process in the film formation processing system 10. The diagonal lines in the inspection area T1 in FIG. 3 are a simplified representation of the concave-convex pattern 32. The base layer 31 is made of, for example, silicon (Si), which also constitutes the substrate W. The surface film 33 is, for example, a silicon nitride (SiN) film or a silicon oxide (SiO) film. The inspection device 1 performs a film thickness inspection, for example, by measuring the film thickness of the surface film 33 and determining whether the film thickness is within a predetermined range. This film thickness inspection is performed on the inspection area T1, where the surface film 33 is formed on the concave-convex pattern 32, to inspect whether the surface film 33 is formed to an appropriate thickness, for example, without including voids (air gaps).

[0010] As shown in FIG. 1 , the inspection apparatus 1 includes a measurement unit (inspection unit) 11 for performing spectroscopic measurement, a stage 21 on which a substrate W is placed, and a control device 100. The inspection apparatus 1 irradiates an inspection area T1 of the substrate W on the stage 21 with measurement light and performs a film thickness inspection of the surface film 33 by an optical method using reflected light, which will be described later. Furthermore, as will be described later, when the irradiation of the measurement light and the reception of the reflected light are performed from a common position, the measurement area R for film thickness inspection is the area on the substrate W irradiated with the measurement light and is a range in which the reflected light from the substrate W can be detected. The measurement area R is the area on the substrate W where spectroscopic measurement for film thickness inspection is effectively performed. Accurate film thickness measurement cannot be performed unless the entire area is located within the inspection area T1. In other words, the measurement light must be irradiated so that the non-inspection area T2 adjacent to the inspection area T1 is not included within the measurement area R. Otherwise, if data indicating a film thickness outside the allowable range is obtained in the film thickness inspection, it will be impossible to determine whether this is due to a defect in the surface film or a misalignment of the measurement area R, and accurate inspection results will not be obtained. Therefore, the inspection device 1 performs positioning in advance to place the measurement region R in the inspection region T1 before the film thickness inspection.

[0011] Hereinafter, "the entire range of the measurement region R is placed in the inspection region T1" may also be expressed as, for example, "the measurement region R is properly placed in the inspection region T1." "At least a part of the measurement region R is not placed in the inspection region T1," which prevents accurate inspection results from being output, may also be expressed as, for example, "the measurement region R is not properly placed in the inspection region T1."

[0012] The measurement unit 11 of the inspection device 1 includes a collimator lens 12, a coaxial probe 13 that irradiates measurement light via the collimator lens 12 and receives the reflected light, an illuminator 14, and a measurement instrument 15 for spectroscopic measurement. The measurement unit 11 also includes a movement mechanism (not shown) that displaces the positions of the collimator lens 12 and the coaxial probe 13. The height direction distance of the coaxial probe 13 and the collimator lens 12 relative to the substrate W is adjusted so that the measurement region R can be positioned without protruding from the inspection region T1.

[0013] The illuminator 14 incorporates a light source for emitting measurement light in the wavelength range from visible light to ultraviolet light, which is used to position the measurement region R, and optical elements such as mirrors and lenses. The coaxial probe 13 is composed of a center portion for emitting light and an outer portion for receiving light, and the center portion is connected to the illuminator 14 and the outer portion is connected to the measuring instrument 15 by optical fibers 16. The measurement light emitted from the coaxial probe 13 irradiates the measurement region R, which is at least a portion of the inspection region T1 on the substrate W that is the target of film thickness inspection, and the light reflected from the measurement region R is detected by the measuring instrument 15 via the outer portion of the coaxial probe 13 and the optical fiber 16.

[0014] 2 is a diagram showing an example of a schematic configuration of a control device 100 according to this embodiment. The control device 100 is, for example, an information processing device such as a computer. The control device 100 has an external I / F (interface) unit 111, a display unit 112, an input unit 113, a storage unit 114, and a control unit 115. In addition to these functional units shown in FIG. 2, the control device 100 may also have various functional units that are included in known computers.

[0015] The external I / F unit 111 is an interface for inputting and outputting information to and from other devices. For example, the external I / F unit 111 is an interface for controlling communications with other devices. One embodiment of the external I / F unit 111 is a network interface card such as a LAN card. For example, the external I / F unit 111 transmits and receives various data to and from the film forming processing system 10 or other devices via a network. The external I / F unit 111 may be an interface such as a USB (Universal Serial Bus) port. The external I / F unit 111 may be an interface such as a USB (Universal Serial Bus) port.

[0016] The display unit 112 is a display device that displays various types of information. Examples of the display unit 112 include a liquid crystal display (LCD) and a cathode ray tube (CRT). The display unit 112 displays various types of information. The input unit 113 is an input device that inputs various types of information, such as a mouse or a keyboard. The input unit 113 accepts operation input from an operator or the like, and inputs operation information indicating the content of the accepted operation to the control unit 115.

[0017] The storage unit 114 is a storage device that stores various types of data. For example, the storage unit 114 is a storage device such as a hard disk, a solid state drive (SSD), or an optical disk. Note that the storage unit 114 may also be a data-rewritable semiconductor memory such as a random access memory (RAM), a flash memory, or a non-volatile static random access memory (NVSRAM).

[0018] The storage unit 114 stores an operating system (OS) and various programs executed by the control unit 115. For example, the storage unit 114 stores a program containing instructions (steps) for executing film thickness inspection, positioning operations, and film formation processes. Furthermore, the storage unit 114 stores various data used by the programs executed by the control unit 115. For example, the storage unit 114 stores process condition data 120, learning data 121, and model data 122. The storage unit 114 can also store other data in addition to the data exemplified above. The various programs and data may be stored in a computer-readable computer recording medium (e.g., a hard disk, an optical disk such as a DVD, a flexible disk, a semiconductor memory, etc.). The various programs and data may also be transmitted from other devices and used online.

[0019] The control unit 115 is a device that controls the control device 100. Specifically, the control unit 115 is an electronic circuit such as a central processing unit (CPU) or a microprocessing unit (MPU), or an integrated circuit such as an application-specific integrated circuit (ASIC) or a field-programmable gate array (FPGA). The control unit 115 has an internal memory for storing programs and data, reads various programs stored in the storage unit 114, and executes the processes of the read programs. The control unit 115 functions as various processing units when the programs run. Specifically, the control unit 115 includes a system control unit 130 that controls each unit of the film formation processing system 10 and controls the film formation process, a first measurement control unit 131, a second measurement control unit 132, a generation unit 133, and a determination unit 134. Note that, in this embodiment, the control unit 115 includes the system control unit 130 to the determination unit 134, but the functions of the system control unit 130 to the determination unit 134 may be distributed among multiple control units.

[0020] Before describing the configuration of the control device 100 related to film thickness inspection and positioning, a brief description of the positioning of this embodiment will be given. In recent years, as semiconductor devices have become increasingly miniaturized, miniaturization has already progressed to the nm (nanometer) range in, for example, the semiconductor manufacturing process of VLSI (Very Large Scale Integration). For this reason, higher accuracy is also required for the positioning of the measurement region R to the inspection region T1. To clearly explain the positioning of this embodiment, which provides high accuracy, a comparative example of positioning that differs from this embodiment will first be described.

[0021] Positioning in the comparative embodiment is performed using image processing. Specifically, the inspection apparatus in the comparative embodiment includes an imaging unit, such as a CCD (Charge Coupled Device), that acquires a surface image of the substrate W, and a signal control unit that processes image signals from the imaging unit. The signal processing unit is connected to a control device 100, which has an image processing program. As shown schematically in FIG. 3 , which illustrates the positioning of the measurement region R in the comparative embodiment, the image processing program first identifies a reference point RP from the surface image of the substrate W and positions the measurement region R at the reference point. Then, based on a predetermined movement distance L from the reference point RP to the inspection region T1, the coaxial probe 13 is moved to properly position the measurement region R in the inspection region T1.

[0022] The positioning accuracy of this comparison depends on the accuracy of identifying the reference point RP identified from the surface image. For example, when the reference point RP is set at the intersection of the contour line F (thick line in FIG. 3 ) of the scribe line formed by the flat pattern 34, the reference point RP is identified by identifying the contour line F from the surface image. The contour line F is identified by obtaining a differential signal, such as the brightness of adjacent pixels among the many pixels that make up the surface image, and identifying the boundary of adjacent pixels where the difference is greatest. For this reason, the accuracy of identifying the reference point depends on the adjacent pixels that are part of the many pixels that make up the surface image.

[0023] Furthermore, in line with the miniaturization of semiconductor devices as described above, even higher resolution surface images are required. Furthermore, because highly confidential patterns of the substrate W are reflected in the surface image, handling of the acquired surface image can be problematic. In contrast to the above-described comparative inspection apparatuses, the inspection apparatus 1 of this embodiment performs highly accurate positioning without using such image processing. Instead, it uses, for example, either an absorbance or reflectance spectrum acquired by spectroscopic measurement of reflected light across the entire measurement region R. In this embodiment, in addition to the absorbance spectrum or reflectance spectrum, the intensity spectrum or absorbance intensity spectrum of reflected light from the measurement region R can also be used. In this disclosure, a spectrum that indicates the absorption or reflection characteristics for the wavelength of measurement light irradiated onto the measurement region R is referred to as a "characteristic spectrum." The following description will be given of an example in which a reflectance spectrum is used.

[0024] Next, the characteristics appearing in the reflectance spectrum of reflected light used for positioning in this embodiment will be described. Figures 4A to 4C are longitudinal side views of a substrate W illustrating the process of moving a measurement region R from a non-inspection region T2 adjacent to the inspection region T1 toward the inspection region T1. The non-inspection region T2 shown in the figures can be, for example, a region on a flat pattern 34 such as a scribe line, and the surface film 33 is formed with a generally uniform thickness relative to the surface of the underlayer 31 of the substrate W. In Figure 4A, the measurement region R is located at measurement position I on the non-inspection region T2. ​​In Figure 4B, the measurement region R is located at measurement position II, which is the boundary between the inspection region T1 and the non-inspection region T2. ​​In Figure 4C, the measurement region R is located at measurement position III on the inspection region T1 (measurement region where spectroscopic measurement is performed). Furthermore, Figure 4D illustrates a case in which the surface film 33 formed on the uneven pattern 32 contains voids 35 when the measurement region R is located at measurement position III. FIG. 5 is a graph illustrating the reflectance spectra when the measurement region R is arranged as shown in FIGS. 4( a ) to 4 ( c ).

[0025] In each reflectance spectrum (characteristic spectrum) shown in Figure 5, it can be confirmed that there is an intersection point where the same reflectance (value) is obtained at the same wavelength regardless of the arrangement (measurement position) of the measurement region R. As shown in Figures 4(a) to 4(c), the wavelength at which multiple reflectance spectra obtained by changing the surface pattern (substrate structure) of the substrate W corresponding to the measurement region R intersect is hereinafter referred to as a "singular point." Meanwhile, the reflectance spectra of the inspection region T1, the non-inspection region T2, and the boundary region between them are generally different from each other, except for the singular points P1 and P2. Therefore, by measuring the reflectance spectrum, it can be determined whether the target is located in the inspection region T1, the non-inspection region T2, or the boundary region between the inspection region T1 and the non-inspection region T2.

[0026] At least one of the singular points P1 and P2, including singular points other than the intersection points described below, appeared in the reflectance spectra measured on substrates W having different shapes of the concave-convex pattern 32 and different film properties and thicknesses of the surface film 33. The inventors therefore made the following assumption and came up with the idea of ​​using the singular points to position the measurement region R.

[0027] The measurement light includes, for example, light of multiple wavelengths. When the measurement light is generally reflected by the patterns 32 and 34 below the surface film 33, the optical characteristics of the reflected light, such as reflectance, change for each wavelength. In other words, the reflectance spectrum can be said to change depending on the shape of the patterns 32 and 34. Meanwhile, in the film formation process performed by the film formation system 10, the surface film 33 is formed to a generally uniform thickness on the upper end surfaces 32a and 34a of the patterns 32 and 34, which effectively reflect the measurement light. The reflected light from the reflection of the measurement light is also an interference wave that undergoes reflection and interference depending on the wavelength, and the interference wave changes depending on the interference factor and the light absorption factor. Both the interference factor and the light absorption factor share the refractive index and film thickness due to the film composition. Therefore, even if the waveform, etc., of the reflectance spectrum varies depending on the shape of the patterns 32 and 34, it can be assumed that a singular point will appear at which the reflectance is the same at at least one wavelength, as long as the composition and film thickness of the surface film 33 are generally the same.

[0028] 5, the reflectance spectra of the surface films 33 formed on different pattern shapes and having approximately the same film thickness have mutually different waveforms in the wavelength ranges before and after the singular points P1 and P2. Hereinafter, the adjacent wavelength ranges before and after each wavelength of the singular points P1 and P2 may be referred to as adjacent wavelength ranges.

[0029] Among the adjacent wavelength ranges, if the wavelength ranges outside the singular points that do not include the wavelength of the singular points as shown in Figure 5 are designated Δλ1, Δλ2, and Δλ3, the reflectances at measurement positions I to III are different in the wavelength ranges Δλ1 and Δλ3, i.e., the waveforms of the reflectance spectra are different. In the wavelength range Δλ2, the reflectances at measurement positions I and II are generally the same, i.e., the waveforms of the reflectance spectra are generally the same. On the other hand, even within the same wavelength range Δλ2, the reflectances at measurement positions I and II are different from those at measurement position III, and the waveforms of the reflectance spectra are different. Thus, the difference in the waveforms of the reflectance spectra caused by the difference in the pattern shape provided in the measurement region R can be clearly seen from the reflectance of a single wavelength and the waveform of the reflectance spectrum in the wavelength range outside the singular points.

[0030] Therefore, when spectroscopic measurement is performed by displacing the measurement region R on a surface film 33 of the same thickness, even if the waveform of the reflectance spectrum changes with the displacement, it can be said that a singular point appears at a predetermined wavelength where the reflectance is the same. Furthermore, the wavelength of such a singular point changes depending on the film thickness of the surface film 33 when the surface film 33 has the same composition. For example, FIG. 6 is a graph showing the change in the wavelength of the singular point with respect to the film thickness of a SiN film. As illustrated in FIG. 6, the film thickness of a surface film 33 of the same composition and the wavelength of the singular point are, for example, proportional to each other, and the wavelength of the singular point also increases as the film thickness increases. In this embodiment, the measurement region R is positioned using the characteristics of such a singular point and the measurement results of the reflectance spectrum in the inspection region T1. Hereinafter, data showing the correspondence between film thickness and wavelength of the singular point, such as that shown in FIG. 6, may also be referred to as film thickness singularity correlation data.

[0031] In this embodiment, the positioning of the measurement region R is performed by a generation operation that generates a prediction model (learning model) based on learning data such as film thickness and reflectance spectrum of the inspection region T1, and a determination operation that determines whether the measurement region R is located in the inspection region T1 based on the prediction model. If it is determined in the determination operation that the measurement region R is not located in the inspection region T1, the measurement region R is moved and the determination operation is performed again, and these operations are repeated to position the measurement region R. The reflectance spectrum for generating the prediction model is measured using a learning substrate W1 that is formed in the same manner as the substrate W to be inspected.

[0032] Returning to the description of the control device 100, the learning data 121 in the storage unit 114 is learning data used to generate a prediction model. The learning data 121 is multiple sets of correspondence data stored in association with one another, including the reflectance spectrum measured for the inspection region T1 on the learning substrate W1 (described later) calculated by the second measurement control unit 132, singularity information on the reflectance spectrum, and the film thickness of the surface film 33 on the learning substrate W1. As a specific example, the reflectance spectrum is a reflectance spectrum in an adjacent wavelength range, as described above. The adjacent wavelength range of the reflectance spectrum used as learning data cannot be generalized because reflectance spectra exhibit various waveforms. However, for example, a wavelength range including the wavelength before (short wavelength) and after (long wavelength) the singularity wavelength, or a wavelength range before or after the singularity, is appropriately set. The width of the adjacent wavelength range can be, for example, approximately 50 nm to 200 nm. For efficiency of the generation and determination operations, it is preferable that this width be as narrow as possible. For example, the lower limit of the adjacent wavelength range Δλm shown in FIG. 5 is a wavelength at which the difference in reflectance between waveforms is large enough to allow the waveforms to be clearly distinguished, such as the wavelength λm at which the reflectance at measurement positions I and II reaches its maximum value. The upper limit λn of the adjacent wavelength range Δλm is set to "λn = λt × 2 - λm" so that the difference in wavelength from the wavelength λt at singular point P1 to the upper limit λn is the same as the difference from the wavelength λt to the lower limit wavelength λm. When multiple singular points P1 and P2 exist as shown in FIG. 5, the width of the adjacent wavelength range Δλm can be set to approximately 50 nm. In this case, the difference in reflectance spectra becomes clear, allowing the generation and determination operations to be performed effectively and efficiently.

[0033] Here, the correspondence data constituting the learning data 121 used to generate the prediction model are stored as a plurality of sets of data with different film thicknesses of the surface film 33. Furthermore, taking into consideration fluctuations in the reflectance spectrum and singular points due to film formation, a plurality of points of correspondence data for the same film thickness may be acquired. The correspondence data is acquired by acquiring the reflectance spectrum measured by the measuring instrument 15 via the external I / F unit 111, and manually inputting the film thickness and singular points via the input unit 113 by the operator, and is stored in the storage unit 114 as the learning data 121.

[0034] The generation unit 133 performs machine learning on the training data 121 and stores a prediction model of the characteristic spectrum in association with the film thickness of the surface film 33 as model data 122 in the storage unit 114. For example, the generation unit 133 performs machine learning on the training data 121 using a linear regression algorithm to generate a prediction model that learns the relationship between multiple sets of corresponding data in the training data 121. For example, the generation unit 133 generates a prediction model that learns the film thickness of each corresponding data and the reflectance spectrum of an adjacent wavelength range using linear regression. Incidentally, the film thickness included in each corresponding data in the training data 121 has a one-to-one correspondence with the wavelength of the singularity as shown in FIG. 6 , and therefore can be replaced with singularity information. Therefore, the technical scope of the present disclosure also includes a case where a prediction model is generated from singularity information including at least the wavelength and the reflectance spectrum. Furthermore, the singularity information in such correspondence data may be the wavelength and reflectance of the singularity, or may be only the wavelength without including the reflectance.

[0035] In this embodiment, a linear regression algorithm is used as machine learning to generate a prediction model so as to prevent overfitting of the training data 121. However, advanced machine learning algorithms such as neural networks and more advanced regression analysis than linear regression can also be used. Note that the term "learning model" in this disclosure is not limited to a learning model created based on the results of machine learning of the training data 121. For example, the term "learning" in this disclosure also includes a case in which singular points and reflectance spectra are acquired for multiple training substrates W1 for each film thickness, and a prediction model showing the correspondence between film thickness and singular points or reflectance spectra (reflectance at each wavelength) is generated by arithmetic averaging.

[0036] The second measurement control unit 132 controls spectroscopic measurement of the substrate W and the learning substrate W1 to calculate the reflectance spectra of the substrate W and the learning substrate W1, respectively. Specifically, the second measurement control unit 132 controls the irradiator 14 and the measuring instrument 15 to irradiate measurement light from the irradiator 14 and detect reflected light from a measurement region R, which is the irradiated region, with the measuring instrument 15. The measuring instrument 15 outputs data on the signal intensity of the detected light to the external I / F unit 111. The second measurement control unit 132 performs spectroscopic analysis, such as Fourier transform, on the signal intensity data output by the measuring instrument 15 to calculate the reflectance spectrum.

[0037] In the positioning determination operation, the determination unit 134 determines whether the measurement region R is located in the inspection region T1 using a prediction model from the reflectance spectrum measured for the substrate W that is the target of film thickness inspection. Specifically, the determination unit 134 first derives a reflectance spectrum for an adjacent wavelength range in the inspection region T1 that corresponds to the film thickness from the model data 122, using the film thickness as an explanatory variable and the reflectance spectrum for the adjacent wavelength range as a target variable. Then, in the reflectance spectrum measured for the substrate W to be inspected, the determiner 134 acquires the reflectance at the wavelength of a singular point that corresponds to the film thickness of the surface film 33, and sets this as singular point information. By determining whether the singular point information acquired from the substrate W to be inspected matches the singular point information derived from the reflectance spectrum of the prediction model, it is determined whether the placement state of the measurement region R can be determined by the determination operation of this embodiment.

[0038] If the determination unit 134 determines that the derived reflectance spectrum matches the measured reflectance spectrum, the determination unit 134 compares the derived reflectance spectrum with the measured reflectance spectrum to determine whether they match and determine the placement state of the measurement region R relative to the inspection region T1. Specifically, if the derived reflectance spectrum and the measured reflectance spectrum are determined to generally match, the determination unit 134 determines that the measurement region R is located in the inspection region T1. If the determination unit 134 determines that the derived reflectance spectrum and the measured reflectance spectrum do not match, the determination unit 134 determines that the measurement region R is not located in the inspection region T1. An example of the comparison of the reflectance spectra being generally consistent is when the reflectances match at wavelengths by, for example, 80% or more, preferably 90% or more. Furthermore, the determination of whether the reflectance spectra match is not limited to a strict match at each wavelength. For example, when the reflectance is expressed as a percentage, if the measured reflectance spectrum at each wavelength is within ±5% of the reflectance in the derived reflectance spectrum, the reflectance spectra may be determined to match.

[0039] The first measurement control unit 131, which controls the measurement of the film thickness inspection, detects the interference intensity spectrum of light reflected from the upper and lower surfaces of the surface film 33, for example, using the measuring instrument 15, and calculates the film thickness using the refractive index of the surface film 33 and parameters related to the substrate W stored in the memory unit 114 (spectral interference film thickness measurement). The determination unit 134 determines whether the calculated film thickness of the surface film 33 is within a preset allowable film thickness range.

[0040] Various positioning operations in this embodiment will be described below. Fig. 7 is a plan view showing the substrate W, the inspection region T1 and the non-inspection region T2 of the learning substrate W1 in positioning the measurement region R in this embodiment. Fig. 8 is a flowchart showing the generation operation, and Fig. 9 is a flowchart showing the determination operation and the movement operation.

[0041] In the generation operation, first, a learning substrate W1 is prepared, on which the same pattern and surface film 33 as those of the substrate W for which film thickness inspection is to be performed are formed (step S101). The learning substrate W1 is, for example, a substrate having the same pattern as the substrate W, and a film formation process is performed in a film formation processing system 10 described below to form the surface film 33. The surface film 33 of the learning substrate W1 does not include voids in the range where the measurement region R is located, and is not defective in film thickness, i.e., not defective in film formation.

[0042] The second measurement control unit 132 sets the thickness of the surface film 33 on the learning substrate W1 (step S102). This thickness is the average thickness of the surface film 33 on the learning substrate W1 in the inspection region T1 and the non-inspection region T2 in its surrounding region, and is not limited to an actual measurement value obtained by film thickness measurement, but may be, for example, an estimated film thickness value. The estimated film thickness value is, for example, an estimated film thickness determined from the film formation conditions of the film formation process in the film formation processing system 10, or a target film thickness set in the film formation process, and can be considered to be the same as the average film thickness obtained by actual measurement.

[0043] Next, the learning substrate W1 is placed on the stage 21, and the collimator lens 12 is moved by the moving mechanism to place the measurement region R in the inspection region T1 (FIG. 7). In the learning stage, the measurement region R can also be placed in the inspection region T1 using a hyperspectral camera or comparative image processing.

[0044] The measurement unit 11 and the second measurement control unit 132 measure the reflectance spectra of the inspection region T1 and the non-inspection region T2 on the learning substrate W1 (step S103). For example, after measuring the reflectance spectrum of the inspection region T1, the collimator lens 12 (measurement region R) is moved to the non-inspection region T2 to measure a reflectance spectrum different from the reflectance spectrum of the inspection region T1.

[0045] Then, a singular point in the reflectance spectrum of the inspection region T1 is identified (step S104). The singular point can be identified, for example, from the intersection of the reflectance spectrum of the inspection region T1 and the reflectance spectrum of the non-inspection region T2. ​​Alternatively, the singular point may be identified by identifying a wavelength from the film thickness of the learning substrate W1 using the film thickness singularity correlation data described above, and then identifying the reflectance from the reflectance spectrum of the inspection region T1 measured based on that wavelength. In this case, it is not necessary to measure the reflectance spectrum of the non-inspection region T2 in step S103, and it is not necessary to identify the singular point (step S104).

[0046] The generation unit 133 stores the set film thickness of the learning substrate W1, the measured reflectance spectrum of the inspection region T1, and the identified singularity information as mutually associated data in the learning data 121 (step S105). The wavelength of the singularity is also stored as the singularity information. The learning data 121 is then stored with the film thickness as an explanatory variable and the reflectance spectrum as a target variable. The reflectance spectra stored in the learning data 121 may be stored for the entire measured wavelength range, or may be stored only for adjacent wavelength ranges to reduce the amount of data.

[0047] Next, the generation unit 133 determines whether all of the planned learning data has been acquired (step S106). The planned learning data is a plurality of sets of correspondence data for substrates W provided with surface films 33 of various thicknesses required for generating the model data, and a prediction model. The number of substrates W and the film thicknesses are determined in advance according to the number of substrates W and the film thicknesses of the surface films 33 required for generating the prediction model.

[0048] If it is determined that all of the planned learning data have not been acquired (No in step S106), the film thickness is changed (step S107), a substrate W is prepared (step S101), and the above steps S102 to S106 are similarly performed. By repeating this process until it is determined that all of the planned learning data have been acquired (Yes in step S106), learning data 121 is stored as multiple sets of data in which film thickness, reflectance spectrum, and singularity information are associated with each other for each substrate W including a surface film 33 with a different film thickness. Thereafter, the generation unit 133 performs machine learning on the learning data 121 to generate a prediction model (step S108), stores the prediction model as model data 122 in the storage unit 114, and terminates the generation operation.

[0049] 9 is a flowchart showing an example of the determination operation and movement operation in the method according to the embodiment. In the determination operation, first, a substrate W to be inspected, on which a surface film 33 is formed, is prepared (step S201), and the substrate W is placed on the stage 21 by a transport mechanism such as a transport arm. The second measurement control unit 132 sets the film thickness of the surface film 33 on the substrate W (step S202). The film thickness is set to be the same as the film thickness of the learning substrate W1 in the generation operation.

[0050] The determination unit 134 derives a reflectance spectrum corresponding to the set film thickness using the prediction model stored as model data 122 (step S203). Specifically, the determination unit 134 acquires the reflectance spectrum of the adjacent wavelength range from the prediction model using the acquired film thickness as an explanatory variable and the reflectance spectrum of the adjacent wavelength range as a target variable. Furthermore, the determination unit 134 reads out singularity information (singularity wavelength) stored in association with the film thickness as learning data 121, and derives the reflectance at the singularity (model reflectance) using the reflectance spectrum acquired from the prediction model.

[0051] Next, the movement mechanism of the inspection apparatus 1 moves the collimator lens 12 and the coaxial probe 13 to position the measurement region R at the initial measurement position (step S204). The initial measurement position is any position between the inspection region T1 and the non-inspection region T2. ​​The measurement region R is positioned at the initial measurement position, for example, based on previously known pattern layout information of the substrate W, by visually estimating the inspection region T1 using a magnified image captured by a camera to an extent that the pattern layout cannot be recognized, or by using a magnifying glass. Alternatively, the measurement region R may be positioned at the initial measurement position by presetting an approximate relative movement amount from a reference position, such as an alignment mark on the substrate W, to the inspection region T1, and positioning the inspection region T1 from the reference position based on that relative movement amount. This positioning of the measurement region R does not require high precision compared to the positioning performed by the determination and movement operations of the present disclosure. The measurement region R positioned in this manner is positioned in an area including the inspection region T1 or the non-inspection region T2, or outside these areas.

[0052] The measurement unit 11 and the second measurement control unit 132 measure the reflectance spectrum by spectroscopic measurement at the first measurement position (step S205, a process of measuring a characteristic spectrum). The determination unit 134 identifies a singular point in the measured reflectance spectrum. The singular point is identified, for example, by identifying the wavelength of the derived singular point information and the reflectance at that wavelength from the reflectance spectrum (measured reflectance). The determination unit 134 then determines whether the model reflectance at the wavelength at the singular point matches the measured reflectance (step S206).

[0053] If the determination unit 134 determines that the difference between the model reflectance and the measured reflectance is equal to or greater than a preset tolerance and that the singular points do not match (No in step S206), the determination operation is terminated. Such a mismatch is likely due to external factors, such as an insufficient film thickness of the surface film 33 at the initial position, an incorrect setting of the measurement region R at the initial position, or an error in the inspection device 1 itself, and the subsequent determination operation is likely to be inappropriately performed, so the determination operation is terminated. If the determination unit 134 determines that the identified model reflectance and the measured reflectance match (the singular points match) (Yes in step S206), the above-mentioned external factors are not anticipated, and the determination operation is likely to be inappropriately performed, so the determination operation is continued.

[0054] Next, the determination unit 134 compares the measured reflectance spectrum in the adjacent wavelength range (inspection target spectrum) with the derived reflectance spectrum in the adjacent wavelength range of the inspection region T1 (verification spectrum) and determines whether they match (step S207). This step can be said to determine the arrangement state of the inspection target substrate W relative to the measurement region R (a process of determining the arrangement state). The comparison may be performed by comparing the waveforms of the reflectance spectra in the adjacent wavelength ranges themselves, or by comparing the reflectance or the amount of change in reflectance at wavelengths other than the wavelength of the singular point in the adjacent wavelength range. Comparing the reflectance spectra in adjacent wavelength ranges in this way makes it possible to clearly compare the reflectance spectra and also reduces the data processing load on the determination unit 134, thereby speeding up the determination process.

[0055] If it is determined that the reflectance spectra in the adjacent wavelength ranges match each other (Yes in step S207), it is determined that the measurement region R is located in the inspection region T1, and the positioning of the measurement region R is complete, so the determination operation is terminated. If it is determined that the reflectance spectra in the adjacent wavelength ranges do not match each other (No in step S207), it is determined that the measurement region R is not located in the inspection region T1, and it is assumed that it is located in the non-inspection region T2 or at the boundary between the inspection region T1 and the non-inspection region T2. ​​In this case, the positioning of the measurement region R is not complete, so a movement operation is performed. For example, the movement operation is performed by the movement mechanism of the inspection device 1 moving the collimator lens 12 and the coaxial probe 13 to move the measurement region R again (step S208, the process of moving the measurement region). The movement of the measurement region R is performed in the same manner as the positioning at the initial measurement position (step S204) described above.

[0056] Then, the aforementioned steps S205 to S207 of the determination operation are performed. These determination and movement operations are repeated until it is determined that the reflectance spectra in adjacent wavelength ranges match (Yes in step S207) and it is determined that the measurement region R has been placed in the inspection region T1, thereby positioning the measurement region R in the inspection region T1. As described above, according to the positioning method of this embodiment, it is determined whether the measurement region R has been positioned in the inspection region T1 based on the reflectance spectrum of light reflected from the entire measurement region R. Therefore, the determination method of this embodiment uses an optical technique that is less susceptible to measurement errors and uses the reflectance spectrum of light reflected from the entire measurement region R, so it can determine the position of the measurement region R with higher accuracy than the image processing of the comparative embodiment.

[0057] Once the positioning of the measurement region R in the inspection area T1 is completed, the inspection device 1 performs a film thickness inspection. In the film thickness inspection, the first measurement control unit 131 calculates the film thickness using the reflectance spectrum measured in the positioned measurement region R and the refractive index of the surface film 33 and parameters related to the substrate W stored in the memory unit 114. The determination unit 134 determines whether the calculated film thickness of the surface film 33 is within a preset film thickness tolerance range, and inspects whether the surface film 33 has been properly formed.

[0058] Furthermore, such film thickness inspection may be repeated at multiple positions within the inspection area T1. In this case, the measurement area R is moved by a preset displacement amount to repeatedly inspect the film thickness. The preset displacement amount is, for example, a length equal to or greater than the diameter of the measurement area R in one direction. Then, each time the measurement area R is moved, the aforementioned judgment operation steps S205 to S207 are performed, and the film thickness inspection is performed after confirming that the measurement area R is positioned within the inspection area T1. This allows inspection within the inspection area T1 without going beyond the inspection area T1.

[0059] (Other Examples of Reflectance Spectra) Examples of reflectance spectra different from the example illustrated in Fig. 5 will be described. Fig. 10 is a longitudinal sectional side view illustrating a substrate W having a film 36 that easily reflects light in an underlying portion of the surface film 33, and Fig. 11 is a graph showing the reflectance spectra at each measurement position (I to III) shown in Fig. 10. Fig. 12 is a longitudinal sectional side view illustrating a substrate W in which a taper is formed in the concave-convex pattern 32, and Fig. 13 is a graph showing the reflectance spectra at each measurement position (I to III) shown in Fig. 12.

[0060] 10, a film 36 that has a tendency to reflect light may be formed, for example, in the non-inspection region T2 between the base layer 31 and the surface film 33. In this case, as shown in Fig. 11, the reflectance of the spectroscopic measurement at measurement position I increases overall, and the waveform of the reflectance spectrum deviates upward from the waveforms of the reflectance spectra at measurement positions II and III.

[0061] Therefore, unlike the example shown in FIG. 5 , in the example shown in FIG. 12 , the reflectance spectra at measurement positions I to III do not intersect, and there is no intersection. In such a case, singularity point P3 is set at the position where the difference in reflectance values ​​of the reflectance spectra is minimum. In this case, the reflectance of singularity point P3 is the reflectance of the reflectance spectra at measurement positions I to III at the wavelength of singularity point P3. In other words, singularity point P3 in this example has the same wavelength, but the reflectance is set for each reflectance spectrum at measurement positions I to III. In this case, for example, reflectance is not included in the singularity information described above in this embodiment, and step S206, which compares singularities, is not performed. Furthermore, as the adjacent wavelength range when there is no intersection in the reflectance spectra as described above, a wavelength range spanning the wavelength of singularity point P3 is set, for example, and the width of this adjacent wavelength range is approximately 50 nm to 200 nm.

[0062] 12 includes convex portions that become wider toward the bottom, and the side surfaces of the concave-convex pattern 32 include inclined side surfaces 37. In this case, the measurement light tends to be diffused by the inclined side surfaces 37, and the reflected light tends to be incident from outside the range of measurement positions I to III, and as shown in Fig. 13, the reflectance spectra at measurement positions I to III are generally similar. However, even in this case, two singular points P4, which are the intersection points of the reflectance spectra, appear.

[0063] It has been confirmed that the wavelength of the singular point P3 described using Figure 11 generally coincides with the wavelength of the singular point determined from the film thickness of the surface film 33 using the correlation data as shown in Figure 6. It has also been confirmed that, for the two singular points P4 that appeared as described using Figure 13, the wavelength of the two singular points P4, for example the longer wavelength, generally coincides with the wavelength of the correlation data as shown in Figure 6. From the above examples of the appearance of singular points P3 and P4, it can be seen that even if a different film such as film 36 or a pattern with a special shape is present at measurement positions I to III, a singular point corresponding to the film quality and film thickness of the surface film 33 will appear, and the generation operation and determination operation of this embodiment can be efficiently performed by utilizing this singular point.

[0064] (Effects) According to the inspection device 1 of this embodiment, the positional state of the substrate W to be inspected relative to the measurement region R can be determined from the reflectance spectrum (characteristic spectrum) obtained by irradiating the substrate W with measurement light.

[0065] (Variations) In the above-described embodiment, in step S206 of FIG. 9 , a comparison of reflectances at singular points is performed to determine whether to continue the evaluation operation. In this case, an example has been described in which the reflectance at the singular point of the reflectance spectrum derived from the prediction model (model reflectance) is used as the comparison target with the reflectance at the singular point of the measured reflectance spectrum (measured reflectance). The method for identifying the model reflectance is not limited to this example. For example, in step S105 of FIG. 8 , machine learning may be performed to create a prediction model using film thickness as an explanatory variable and singular point information (the wavelength of the singular point and the reflectance at that wavelength) as a target variable. In this case, in step S206 of FIG. 9 , the singular point information (the reflectance at the singular point) obtained based on the result of inputting the film thickness into the prediction model may be used as the model reflectance to compare the reflectances.

[0066] Furthermore, it is preferable that the density of the concave-convex pattern 32 be substantially uniform throughout the entire inspection region T1. This ensures that the density of the concave-convex pattern 32 is uniform regardless of the position of the measurement region R within the inspection region T1, and therefore, substantially the same reflectance spectrum can be obtained regardless of the position of the measurement region R within the inspection region T1. By comparing the reflectance spectrum of the prediction model with the measured reflectance spectrum, it is possible to determine with high accuracy whether or not the measurement region R is within the inspection region T1. Furthermore, in this embodiment, the measurement unit 11 can also be used for spectroscopic measurement for positioning, so that film thickness inspection is performed using spectral interference film thickness measurement. However, this is not limiting, and film thickness inspection may also be performed using, for example, reflectance spectroscopy or spectroscopic ellipsometry. Furthermore, the positioning of the present disclosure is not limited to being performed using an inspection device for film thickness inspection, but may also be performed using other inspection devices for in-line inspection of substrates W.

[0067] Although the inspection device 1 of this embodiment performs positioning using either the absorbance spectrum or the reflectance spectrum as the characteristic spectrum, the characteristic spectrum that can be used for positioning is not limited to this. For example, even if the intensity spectrum of reflected light or the absorbance intensity spectrum (FIG. 28) which is the difference between the intensity spectrum of irradiated light and the intensity spectrum of reflected light is used, a singular point similarly appears, and therefore, it can be treated in the same way as the reflectance spectrum.

[0068] (Modification of First Embodiment) Positioning in this modification will be described with reference to Figs. 14 to 16. In the description of this modification, differences from the first embodiment will be mainly described, and a description of the same configuration as in the first embodiment will be omitted. Fig. 14 is a plan view showing the substrate W, inspection region T1 and non-inspection region T2 of the learning substrate W1 in positioning of the measurement region R in this modification. Fig. 15 is a flowchart showing the generation operation of this modification, and Fig. 16 is a flowchart showing the determination operation and movement operation of this modification.

[0069] 14 , in positioning the measurement region R in this modified example, there are adjacent regions Ta, Tb, Tc, and Td, each including a different pattern formed on the base layer 31 for the non-inspection region T2. ​​The different reflectance spectra of the adjacent regions Ta to Td are measured, and data that correlates each reflectance spectrum with region information indicating each of the adjacent regions Ta to Td is stored in training data 121 to generate model data 122. In the determination operation, the model data 122 generated in this manner is used to determine in which region T1, Ta to Td the measurement region R is located, and the measurement region R is quickly positioned toward the inspection region T1 based on the relative positional relationship of each of the adjacent regions Ta to Td with respect to the inspection region T1.

[0070] 14 and 15 , the operation of generating a prediction model in this modified example will be described. In the non-inspection region T2 of the substrate W and the learning substrate W1, various uneven patterns or flat patterns different from the uneven pattern 32 of the inspection region T1 are formed around the inspection region T1. Because these adjacent patterns each contain different patterns, reflectance spectra of different waveforms are measured by irradiating the measurement light from the measurement unit 11. The regions in which each adjacent pattern is provided are set as adjacent regions Ta to Td, respectively. The ranges of the inspection region T1 and the adjacent regions Ta to Td, and the relative positions of the adjacent regions Ta to Td, may be determined in advance from design information for the substrate W, etc., and can be acquired in the generation operation described below.

[0071] As in the first embodiment, first, a first learning substrate W1 is prepared (step S301), the thickness of the surface film 33 is set for the learning substrate W1 (step S302), and the reflectance spectrum in the inspection region T1 is measured (step S303). Next, for example, the measurement region R is moved toward the adjacent region Ta, and the reflectance spectrum of the adjacent region Ta is measured (step S304). If the relative position of the adjacent region Ta is not known from design information or the like, for example, the measurement region R is moved from the inspection region T1 in the positive direction of the Y axis shown in FIG. 14 while measuring the reflectance spectrum. Then, when the reflectance spectrum clearly changes from the reflectance spectrum of the inspection region T1, the region where the measurement region R is located is set as the adjacent region Ta, and the movement to the adjacent region Ta is completed. In this case, a movement vector A of the measurement region R may be identified and set as the relative position of the adjacent region Ta.

[0072] Next, singular points are identified from the reflectance spectrum of the inspection region T1 and the reflectance spectrum of the adjacent region Ta (step S305). Next, the singular point information, each measured reflectance spectrum, film thickness, and each region information are stored in the training data 121 as mutually corresponding data (hereinafter also referred to as relative data) (step S306). Specifically, the region information indicating the inspection region T1, the reflectance spectrum of the inspection region T1, the singular point information, and the film thickness are mutually corresponding to each other, and the region information indicating the adjacent region Ta, the reflectance spectrum of the adjacent region Ta, the singular point information, and the film thickness are mutually corresponding to each other and stored in the training data 121, respectively. These film thicknesses and singular point information generally match, but the reflectance of the singular point information may differ as described above.

[0073] The above operation will be described in detail below. The film thickness is used as an explanatory variable, and the reflectance spectrum and area information are used as objective variables to learn the data and stored in the learning data 121. The relative position may be stored in the learning data 121 together with the area information or instead of the area information as data corresponding to the area information, or may be stored in the storage unit 114.

[0074] Next, the generation unit 133 determines whether all of the adjacent regions Ta to Td have been measured (step S307). If the relative positions of the adjacent regions Ta to Td are not known from design information or the like, the determination is made based on, for example, whether it can be determined that the relative data for each adjacent region required for positioning has been acquired based on the accumulated relative data in the training data 121. If it is determined that all of the planned training data has not been acquired (No in step S307), the measurement region R is moved to another adjacent region, for example, adjacent region Tb, and the reflectance spectrum of adjacent region Tb is measured (step S308). The movement of the measurement region R to adjacent region Tb is performed in the same manner as the movement from the inspection region T1 to adjacent region Ta in step S304, and the relative position of adjacent region Tb can also be determined from the movement amount vector B.

[0075] As was done for adjacent region Ta in step S306, data in which the reflectance spectrum, region information, and film thickness measured for adjacent region Tb are correlated with each other are stored in training data 121 (step S306). The above steps S308, S306, and S307 are also performed for the other adjacent regions Tc and Td. When it is determined that measurements have been made for all adjacent regions Ta to Td (step S307), multiple sets of data for the inspection region T1 and the adjacent regions Ta to Td for the first training substrate W1 are stored in the training data 121.

[0076] Next, the generation unit 133 determines whether the training substrate W1 has been measured with all of the planned film thicknesses (step S309). If it determines that the training substrate W1 has not been measured with all of the planned film thicknesses (No in step S309), a second substrate W with a different film thickness is prepared (steps S310 and S301). Then, as with the first training substrate W1, the generation unit 133 measures the reflectance spectrum for each of the regions T1 and Ta to Td, and stores the data correlating the regions T1 and Ta to Td in the training data 121 (steps S302 to S308). By repeating the above process until it is determined that the training substrate W1 with all film thicknesses has been measured (Yes in step S309), the training data 121 stores multiple sets of data correlating the regions T1 and Ta to Td for multiple training substrates W1 with different film thicknesses. The generation unit 133 performs machine learning on the learning data 121 to generate a prediction model (step S311), stores the prediction model as model data 122 in the storage unit 114, and ends the generation operation.

[0077] The determination and movement operations of this modified example will be described with reference to FIG. 16 , a flowchart illustrating an example of the determination and movement operations in the method according to this embodiment. First, a substrate W to be inspected, on which a surface film 33 is formed, is prepared on the stage 21 (step S401), and the film thickness of the surface film 33 is set (step S402). Next, using a prediction model stored as model data 122, reflectance spectra for each of the regions T1 and Ta to Td corresponding to the set film thickness are derived (step S403). Specifically, the determination unit 134 acquires the reflectance spectra for each of the regions T1 and Ta to Td from the prediction model using the set film thickness as an explanatory variable, the corresponding region information, and the reflectance spectrum as a response variable. Furthermore, the determination unit 134 reads out singularity information (singularity wavelengths) stored in association with the film thickness as learning data 121, and derives the reflectance at the singularity (model reflectance) using the reflectance spectrum acquired from the prediction model.

[0078] Next, the measurement region R is placed at the initial measurement position on the substrate W (step S404), and the reflectance spectrum is measured at the initial measurement position (step S405, a step of measuring a characteristic spectrum). It is then determined whether the reflectance of a specific point identified from the measured reflectance spectrum (measured reflectance) matches the reflectance of a specific point in the reflectance spectrum derived from the prediction model (model reflectance) (step S406). If it is determined that these reflectances match (Yes in step S406), it is then determined whether the derived reflectance spectrum of the inspection region T1 matches the measured reflectance spectrum (step S407).

[0079] If it is determined that these reflectance spectra do not match (No in step S407), a movement operation is performed. First, the determination unit 134 identifies the region information and relative position of the measurement region R (step S408). The region information and relative position are identified, for example, by identifying the reflectance spectrum of each of the derived adjacent regions Ta to Td that matches or is most similar to the measured reflectance spectrum, and then using the region information and relative position associated with the identified reflectance spectrum. The region information and relative position may be identified by identifying only the relative position without identifying the region information.

[0080] Based on the identified relative position, the movement mechanism of the measurement unit 11 moves the measurement region R toward the inspection region T1 (step S409). To give a specific example using Figure 14, for example, if the measured reflectance spectrum matches the region information indicating the adjacent region Tc and the reflectance spectrum linked to the relative position (movement vector C), the measurement region R is moved by a vector C' in the opposite direction to the movement vector C indicating the relative position. According to the method of this modified example, which is configured to repeatedly perform such movement operations and the determination operations of steps S405 to S407, the measurement region R can be positioned in the inspection region T1 efficiently and with high accuracy.

[0081] (Film Forming Processing System 10) The following briefly describes the film forming processing system 10 in which the inspection apparatus 1 of this embodiment and its modified examples is installed. Fig. 17 is a plan view showing the film forming processing system 10. The film forming processing system 10 includes a loader module 61, a load lock module 65, a vacuum transfer module 66, and first to third processing modules 7A to 7C.

[0082] The loader module 61, the load lock module 65, and the vacuum transfer module 66 are arranged in this order in the front-to-rear direction.

[0083] The loader module 61 includes a housing whose interior is at atmospheric pressure, a transport mechanism 62 for substrates W provided in the housing, and four load ports 63, and the inspection apparatus 1 according to this embodiment is disposed in the housing. A transport container 64 called a FOUP (Front Opening Unified Pod) for storing substrates W is placed on each load port 63. The transport mechanism 62 includes, for example, an articulated arm that can move left and right, and is capable of transporting substrates W between the transport containers 64 on each load port 63, each load lock module 65, and the inspection apparatus 1.

[0084] In this example, there are provided three load lock modules 65. Each load lock module 65 has a housing, and the housing is connected to the loader module 61 and the vacuum transfer module 66 via gate valves G provided at the front and rear of the housing, respectively.

[0085] Each load lock module 65 is configured so that the pressure inside the housing can be freely changed between atmospheric pressure and vacuum pressure when the gate valves G at the front and rear of the housing are closed. A stage (not shown) is provided inside the housing, and the stage can carry a substrate W and transfer the substrate W to and from the transport mechanism 62 and a vacuum transport mechanism 67 (described later) that access the load lock module 65.

[0086] The vacuum transfer module 66 includes a housing 66a and a vacuum transfer mechanism 67 provided inside the housing 66a. The housing 66a is connected to an exhaust mechanism (not shown), such as a turbomolecular pump, via an exhaust pipe, and the inside of the housing 66a is maintained in a vacuum atmosphere by exhausting air from the exhaust mechanism.

[0087] The first to third processing modules 7A to 7C are arranged side by side on both the left and right sides of a housing 66a of the vacuum transfer module 66, as seen from the front side. Substrates W are transferred between the first to third processing modules 7A to 7C and the load lock module 65 by a vacuum transfer mechanism 67 including, for example, an articulated arm.

[0088] The diagram shows the correspondence between the processes performed in each of the first to third process modules 7A to 7C and the series of processes described above. The first process module 7A performs a COR process, which alters the oxide film formed on the surface of the base layer 31 of the substrate W to generate reaction products, and the second process module 7B performs a PHT process, which sublimes and removes the reaction products. The third process module 7C performs a film formation process, which forms a surface film 33. Each of the process modules 7A to 7C includes a process chamber 71 that is evacuated to create a vacuum atmosphere inside, and a mounting table 72 that is provided within the process chamber 71 and on which the substrate W is placed. Each process step is performed within the process chamber 71.

[0089] The system control unit 130 of the control device 100 outputs control signals to each part of the film forming processing system 10 according to each program stored in the storage unit 114, and controls the operation of each part. Specifically, the system control unit 130 controls the operation of the first to third processing modules 7A to 7C, the opening and closing of the gate valves G and G1, the transfer mechanism 62, the vacuum transfer mechanism 67, etc.

[0090] The transport path of the substrate W in the film forming processing system 10 is first the transport container 64 → loader module 61 → load lock module 65 → vacuum transport module 66 → first processing module 7A → second processing module 7B → third processing module 7C in this order. During this transport process, the oxide film on the surface of the underlayer 31 of the substrate W is removed, and then a surface film 33 is formed on the substrate W. The substrate W is then transported in the order of the third processing module 7C → vacuum transport module 66 → load lock module 65 → loader module 61 → inspection device 1. The substrate W on which the surface film 33 has been formed during this transport process undergoes a film thickness inspection by the inspection device 1. The substrate W that passes the film thickness inspection is transported to the transport container 64 and then transported together with the transport container 64 to the next processing device by a transport device (not shown). The substrate W that fails the film thickness inspection is stored in a storage area for rejected substrates W (not shown) provided in, for example, the loader module 61.

[0091] (Film Formation Module) Next, the third process module 7C, which is a film formation module, will be described as a representative of the process modules 7A to 7C, using the vertical side view of Figure 18. The description will be focused on a case where film formation is performed as substrate processing. The third process module 7C is configured to be airtight, and the process vessel 71 is cylindrical. A mounting table 72 is provided within the process vessel 71.

[0092] The mounting table 72 is electrically grounded and made of a metal such as nickel. The mounting table 72 supports the substrate W placed on its upper surface horizontally. The lower surface of the mounting table 72 is electrically connected to a support member 73 made of a conductive material. The mounting table 72 is supported by the support member 73. The support member 73 is supported by the bottom surface of the processing vessel 71, and its lower end is electrically connected to the bottom surface of the processing vessel 71. Therefore, the processing vessel 71 is also grounded.

[0093] The mounting table 72 has a built-in heater 74, which can heat the substrate W placed on the mounting table 72 to a predetermined temperature. The mounting table 72 is provided with lifter pins 75 for raising and lowering the substrate W. When loading a substrate W into the third processing module 7C, the lifter pins 75 are raised by an elevation mechanism (not shown) to receive the substrate W from the vacuum transfer mechanism 67 ( FIG. 17 ), and after the vacuum transfer mechanism 67 leaves, the lifter pins 75 are lowered to place the substrate W on the mounting table 72.

[0094] A substantially disk-shaped shower head 76 is provided on the inner surface of the processing vessel 71 above the mounting table 72. The shower head 76 is supported on the upper part of the mounting table 72 via an insulating member 71a such as ceramic, thereby electrically insulating the processing vessel 71 from the shower head 76. The shower head 76 is made of a conductive metal such as nickel.

[0095] The shower head 76 contains a gas diffusion space 76a, has a shower plate 76b at its bottom facing the mounting table 72, and is provided to close the interior of the processing vessel 71 from above. The shower plate 76b has a number of gas ejection holes 76c dispersedly formed therein, which open toward the gas diffusion space 76a and the mounting table 72. A gas supply path 77a is connected to the top of the shower head 76 to introduce various gases into the gas diffusion space 76a. A gas supply unit 77 is connected to the gas supply path 77a.

[0096] The gas supply unit 77 has gas supply lines connected to gas supply sources of various gases used in forming the surface film 33, which is a SiN film. Each gas supply line branches appropriately according to the film formation process and is provided with control devices for controlling the flow rate of the gas, such as valves such as an open / close valve and flow rate controllers such as a mass flow controller. The gas supply unit 77 supplies a purge gas, a source gas, and a reaction gas to the gas supply path 77a. For example, the purge gas is an inert gas, and the source gas is SiH ４ (silane) gas, and the reactive gas is NH ３These gases are diffused in the gas diffusion space 76a and discharged from the gas discharge holes 76c.

[0097] The space surrounded by the lower surface of the shower head 76 and the upper surface of the mounting table 72 forms a processing space where a film formation process is performed. The shower head 76 is paired with the mounting table 72 and functions, for example, as an electrode plate for forming capacitively coupled plasma (CCP) in the processing space. A high-frequency power supply 78 is connected to the shower head 76 via a matching box 78a. When generating plasma in the processing space, the high-frequency power supply 78 supplies high-frequency power (RF power) in accordance with the gas supplied from the gas supply unit 77. This applies the high-frequency power to the gas supplied from the shower head 76 to the processing space, generating plasma in the processing space.

[0098] An exhaust port connected to an exhaust pipe 79a is formed at the bottom of the processing vessel 71, and the exhaust pipe 79a is connected to an exhaust device 79 having a vacuum pump and a pressure adjustment valve. By the operation of the exhaust device 79, the inside of the processing vessel 71 is depressurized to a predetermined vacuum level.

[0099] Next, a brief description will be given of the process by which the third processing module 7C performs film formation on the substrate W under the control of the system control unit 130, which controls film formation in the control device 100. First, the substrate W is placed on the mounting table 72 by the vacuum transfer mechanism 67 of the film formation processing system 10. The substrate W thus loaded has the uneven pattern 32 exposed on its surface, and no surface film 33 is formed thereon. When performing film formation on such a substrate W, the system control unit 130 controls the exhaust device 79 to reduce the pressure inside the processing vessel 71. The system control unit 130 controls the gas supply unit 77 to supply various gases used in film formation from the gas supply unit 77 and introduce the processing gas into the processing vessel 71 through the shower head 76. The system control unit 130 then controls the high-frequency power supply 78 to supply high-frequency power from the high-frequency power supply 78 to generate plasma in the processing space, thereby performing film formation on the substrate W. In this manner, the surface film 33 is formed on the substrate W by the third processing module 7C. The lifter pins 75 are then protruded from the stage 72 , and the rear surface of the substrate W is supported by the lifter pins 75 to lift the substrate W from the stage 72 and transport it out by the vacuum transport mechanism 67 .

[0100] Second Embodiment An inspection apparatus 1 according to a second embodiment will be described below with reference to FIGS. 19A, 19B, and 20. The inspection apparatus 1 according to the second embodiment has a function of identifying the position of an alignment mark AM formed on a substrate W by utilizing a reflectance spectrum (characteristic spectrum) obtained by irradiating the substrate with measurement light. FIG. 19A is a plan view showing a state in which a measurement region R is positioned relative to an alignment mark in this embodiment, and FIG. 19B is a plan view showing a state in which the measurement region R is displaced from the alignment mark in this embodiment. In FIGS. 19A and 19B, the measurement region R is shaded, and the periphery of the measurement region R is indicated by a dashed line. FIG. 20 is a vertical side view illustrating the positioning of the measurement region R relative to an alignment mark on a substrate having a resist layer and a hard mask layer provided on its surface.

[0101] The inspection apparatus 1 of this embodiment uses spectroscopic measurement by the measurement unit 11 to accurately position the measurement region R relative to the alignment mark AM, as mentioned in the comparative example shown in FIG. 3 . As shown in FIGS. 19A and 20 , the alignment mark AM is, for example, a convex portion formed on the surface of the base layer 31. The upper surface of the alignment mark AM forms a reference region AMa for indicating a reference position, and a characteristic uneven pattern AMb different from the various uneven patterns on the substrate W is formed in the reference region AMa. Therefore, by utilizing the positioning technique described in the first embodiment, instead of the measurement positions II and III shown in FIGS. 4( b) and 4( c), the reflectance spectra of the reference region AMa and its surrounding peripheral region 31b on the upper surface of the base layer 31 are compared, as shown in FIGS. 19A and 19B. This allows the measurement region R to be accurately positioned relative to the alignment mark AM.

[0102] Here, in this embodiment, the measurement region R is set to have, for example, the same shape and area as the reference region AMa. Therefore, when the measurement region R and the reference region AMa are positioned so that their center points overlap in a planar view, that is, when the measurement region R is aligned with the reference region AMa, the measurement region R is positioned within the reference region AMa without protruding from the reference region AMa. Then, in the positioning generation operation, the reflectance spectrum when the measurement region R is aligned is measured and learned, and stored in the model data 122. In the determination operation, the agreement between the measured reflectance spectrum and the reflectance spectrum when aligned according to the prediction model is determined, and if they do not match, a movement operation is performed, so that the center position of the measurement region R can be aligned with the center of the alignment mark AM.

[0103] Furthermore, the determination operation of the present disclosure makes it possible to distinguish between the reflectance spectrum in the aligned arrangement shown in Fig. 19A and the reflectance spectrum when the measurement region R is positioned outside the reference region AMa shown in Fig. 19B. Therefore, according to this embodiment, it is possible to determine whether the measurement region R is aligned with the alignment mark AM, and to perform positioning with high accuracy.

[0104] 20, a hard mask layer 38 containing metal to enable high selectivity etching and lacking transparency, and a resist layer 39 may be provided on the surface film 33n to be etched. In this case, positioning by image processing as shown in the comparative example cannot recognize the alignment mark AM during imaging, making positioning extremely difficult. However, in this embodiment, if the irradiator 14 is provided with a light source that emits measurement light in the infrared wavelength range, positioning can be performed using spectroscopic measurement using infrared light that transmits through the hard mask layer 38 and the resist layer 39, making it possible to use this method for substrates W having a variety of layers.

[0105] The measurement area R may be set to a shape similar to the reference area AMa but slightly smaller than the reference area AMa, for example, as long as the positional deviation from the alignment mark AM is within an allowable range.

[0106] Third Embodiment An inspection apparatus 1A of a third embodiment performs an alignment inspection to check whether a laminated (bonded) substrate Wc, which is formed by bonding the surfaces of an upper substrate Wa and a lower substrate Wb, is aligned without misalignment. The substrates for which the alignment inspection is performed will be described below with reference to Figures 21 to 23. Figure 21 is a vertical cross-sectional side view of the upper and lower substrates before they are bonded together, Figure 22 is a vertical cross-sectional side view of the upper and lower substrates bonded together in an aligned state, and Figure 23 is a vertical cross-sectional side view of the upper and lower substrates bonded together in a misaligned state.

[0107] First, the structure and bonding of the upper substrate Wa and the lower substrate Wb will be described using Figure 21. The upper substrate Wa and the lower substrate Wb have substantially the same configuration and include a base substrate 31A, an operation unit 20a, and a wiring layer 20b. Hereinafter, the surface of the upper substrate Wa or the lower substrate Wb that is bonded to the other substrate Wa or Wb will be referred to as the bonding surface SF. The operation unit 20a is formed by including a portion of the base substrate 31A. The operation unit 20a includes, for example, a semiconductor device such as a transistor. The base substrate 31A is, for example, a semiconductor wafer. The wiring layer 20b is, for example, a multi-layer wiring. The wiring layer 20b includes wiring 22, a metal pad 23, a first insulating film 24, and a second insulating film 25.

[0108] The wiring 22 is provided in multiple layers. The wiring 22 is electrically connected to the calculation unit 20a. The metal pad 23 is provided on the bonding surface SF side of the wiring 22 that is located farthest from the base substrate 31A. The metal pad 23 is electrically connected to the wiring 22. The metal pad 23 is electrically connected to the calculation unit 20a via the wiring 22. The surface of the metal pad 23 is exposed. The first insulating film 24 is, for example, an interlayer insulating film that fills the spaces between the wirings 22.

[0109] The second insulating film 25 is provided on the bonding surface SF side of the first insulating film 24. The surface of the second insulating film 25 is exposed. The exposed surface of the second insulating film 25 is flush with, for example, the exposed surface of the metal pad 23. The second insulating film 25 is, for example, a SiC film. In this way, the upper substrate Wa and the lower substrate Wb each have, at the bonding surface SF, a region where the second insulating film 25 is exposed and a region where the metal pad 23 is exposed. Note that the second insulating film 25 is an example of an insulating film, and the metal pad 23 is an example of a conductive film.

[0110] The bonding of the upper substrate Wa and the lower substrate Wb will now be described. First, the surfaces of the upper substrate Wa and the lower substrate Wb are planarized by chemical mechanical polishing (CMP) to form a bonding surface SF, which is then coated with, for example, an adhesive to form an adhesive film 26. As shown in Fig. 21, the upper substrate Wa is placed on top and the lower substrate Wb is placed on the bottom of the upper substrate Wa. The exposed pad surfaces 23a of the respective metal pads 23 are positioned so as not to misalign with each other, and pressure is applied to bond the upper substrate Wa and the lower substrate Wb to form an overlapped substrate Wec. As shown in Fig. 22, it is preferable that the upper substrate Wa and the lower substrate Wb are not misaligned in a direction parallel to the bonding surface SF, and that the pad surfaces 23a of the respective metal pads 23 are aligned without misalignment in a direction parallel to the bonding surface SF. In detail, it is preferable that the pad surface 23a of one metal pad 23 on the upper substrate Wa or the lower substrate Wb has substantially no portion that is shifted from the pad surface 23a of the other metal pad 23 and does not overlap with the other pad surface 23a.

[0111] However, as shown in Figure 23, there is a small chance that the bonding surfaces SF of the upper substrate Wa and the lower substrate Wb may be misaligned during manufacturing, resulting in the formation of an overlapped substrate Wc in which the substrates are bonded together with the pad surfaces 23a of their metal pads 23 misaligned. Therefore, the alignment inspection by the inspection device 1A according to this embodiment makes it possible to detect such an overlapped substrate Wc in an unaligned state with high accuracy and ease. The alignment inspection of this embodiment will be described below, with a focus on the differences from the first embodiment.

[0112] The measurement unit 11 ( FIG. 1 ) in the alignment inspection of this embodiment irradiates infrared light, particularly infrared light in a wavelength range of 1000 nm or more, as measurement light to measure the internal structure of the overlapped substrate Wc. The thickness of the overlapped substrate Wc and the learning overlapped substrate Wc1 in the generation operation corresponds to the film thickness of the surface film 33 on the substrate W and learning substrate W1 of the first embodiment. Therefore, the singular point can be identified from the thickness of the overlapped substrate Wc and learning overlapped substrate Wc1.

[0113] In this embodiment, it is preferable to provide an infrared absorbing sheet between the stage 21 and the substrate W to prevent noise from being mixed in from the stage 21. Alternatively, for example, the coaxial probe 13 may not be a coaxial structure, but may be provided as two separate probes, one connected to the irradiator 14 and the other connected to the measuring instrument 15, with a through-hole extending vertically through the stage 21. The probe on the irradiator 14 side may be disposed below the through-hole in the stage, and the probe on the measuring instrument 15 may be disposed above the through-hole, and the infrared light irradiated onto the substrate W may be received. This also prevents noise from being mixed in from the stage 21. In this case, the measurement region R is positioned in an inspection region T3 (described later) by shifting the substrate W placed on the stage 21. The characteristic spectrum of the measurement light transmitted through the substrate W may be, for example, the intensity spectrum of the transmitted light received by the probe on the measuring instrument 15 side, or the absorbance intensity spectrum, absorbance spectrum, or transmittance spectrum, which are the differences between the intensity spectrum of the irradiated light, the intensity spectrum of the transmitted light, and the intensity spectrum of the reflected light. Since the transmitted light is the difference between the light irradiated onto the measurement region R and the light reflected and absorbed by the substrate W, it can be said that the intensity spectrum and transmittance spectrum of the transmitted light measure the characteristics corresponding to the reflection and absorption of the measurement light for each wavelength. Therefore, the intensity spectrum and transmittance spectrum of the transmitted light also correspond to the characteristic spectrum of the present disclosure.

[0114] In the alignment inspection, for example, the pad surfaces 23 a of the upper substrate Wa and the lower substrate Wb of the overlapping substrate Wc and the overlapping substrate Wc1 correspond to patterns such as the concave-convex pattern 32 described in the first embodiment and its modified examples. According to this correspondence relationship, in this embodiment, the aligned state or misaligned state of the metal pads 23 bonded to each other on the overlapping substrates Wc and Wc1 appears as a change in the reflectance spectrum excluding singular points, and therefore the aligned state or misaligned state can be determined from the reflectance spectrum.

[0115] In the alignment inspection, spectroscopic measurements are performed on the same inspection area T3 on the overlapped substrate Wc and the learning overlapped substrate Wc1, for example, in the generation operation and the determination operation, to measure the reflectance spectrum. The inspection area T3 is, for example, a portion including the front and back surfaces of the overlapped substrate Wc that is arbitrarily set, and includes areas where the same pattern of pad surfaces 23a is arranged on the upper substrate Wa and the lower substrate Wb. In the generation operation, the reflectance spectra and singular points measured for the inspection area T3 on the learning overlapped substrates Wc1 of different thicknesses are stored in the memory unit 114 as learning data 121. Then, machine learning is performed on the learning data 121 to generate a predictive model of the reflectance spectrum corresponding to the thickness of the learning overlapped substrate Wc1, and the model data 122 is stored in the memory unit 114.

[0116] In the determination operation, the reflectance spectrum of the inspection region T3 of the superposed substrate We is measured, and the reflectance spectrum derived by inputting the thickness of the superposed substrate We into the prediction model is compared with the measured reflectance spectrum to determine whether they match. From this result, it is determined whether the superposed substrate We is in a matched state or otherwise, i.e., in an unmatched state. This makes it possible to sort the superposed substrate We determined to be matched as a passing product and the superposed substrate We determined to be unmatched as a failing product with high accuracy.

[0117] Furthermore, in the alignment test, the reflectance spectrum in the misaligned state changes depending on the multiple misalignment directions along the bonding surface SF and the amount of misalignment, which is the amount of displacement from the aligned state. Therefore, for example, the amount of misalignment can be changed within a predetermined range in each of two mutually orthogonal directions along the bonding surface SF, and the reflectance spectrum can be learned in advance by the generation operation. In this case, the amount of misalignment in each misalignment direction can be understood in association with the region information and relative position in the modified example of the first embodiment, thereby making it possible to estimate the direction and amount of misalignment of the superposed substrate Wc to be inspected.

[0118] (Bonding Apparatus 7D) The following briefly describes the bonding apparatus 7D, which is installed together with the inspection apparatus 1A of this embodiment and bonds the upper substrate Wa and the lower substrate Wb. FIG. 24 is a vertical cross-sectional side view showing the bonding apparatus 7D. The upper substrate Wa is loaded into the bonding apparatus 7D through a substrate loading / unloading port (not shown) formed in the processing vessel 71A on the transfer region A1 side. The upper substrate Wa is then transported to a position adjustment mechanism 81 by a substrate transport body 80 that is movable in the Y and Z directions. The position adjustment mechanism 81 is configured to rotate each of the substrates Wa-Wc it holds around its center to adjust their horizontal positions. The upper substrate Wa, which was loaded earlier, has its position adjusted by the position adjustment mechanism 81, and is then transferred to a holding arm 82a of an inversion mechanism 82 with the upper substrate Wa facing upward with its bonding surface SF facing upward.

[0119] The inversion mechanism 82 inverts the holding arm 82a in the transfer region A1 to invert the front and back surfaces of the upper substrate Wa it holds, so that the bonding surface SF of the upper substrate Wa faces downward. The inversion mechanism 82 then passes through a transfer port 83a provided in a partition wall 83 separating the transfer region A1 from the processing region A2, and moves toward the upper chuck 84 in the processing region A2. The upper chuck 84 then sucks the back surface of the upper substrate Wa and holds it. The upper chuck 84 is then moved by a chuck driver 84a, which is movable in the Y direction, to a position above and facing the lower chuck 85. The upper substrate Wa waits in this state until the lower substrate Wb is transferred to the bonding device 7D.

[0120] Next, the lower substrate Wb is carried into the bonding apparatus 7D, and is carried to the position adjustment mechanism 81 by the substrate transport body 80. Next, a nozzle arm 91, which is movable in the Y and Z directions and to which a nozzle 93 is attached, moves the nozzle 93 to above the center of the lower substrate Wb. Next, the position adjustment mechanism 81 rotates the lower substrate Wb, and an adhesive supply mechanism connected to the nozzle 93 applies adhesive to the surface of the lower substrate Wb by spin coating, and the horizontal orientation of the lower substrate Wb is further adjusted by the position adjustment mechanism 81.

[0121] Next, the lower substrate Wb is transported by the substrate transport body 80 to the lower chuck 85 in the processing region A2, and the back surface of the lower substrate Wb is held by suction on the lower chuck 85 with the bonding surface SF facing upward. Next, the horizontal positions of the lower substrate Wb held on the lower chuck 85 and the upper substrate Wa held on the upper chuck 84 are adjusted while capturing an image of their bonding surface SF with, for example, a CCD camera. Specifically, the position adjustment is performed by adjusting the horizontal position of the upper substrate Wa by the upper chuck 84 based on the captured image so that a predetermined reference point (not shown) on the surface of the lower substrate Wb coincides with a reference point (not shown) on the surface of the upper substrate Wa ( FIG. 21 ).

[0122] Next, the lower chuck 85 is raised by a chuck driving unit 85a configured to raise and lower the lower chuck 85, and the lower substrate Wb held by the lower chuck 85 and the upper substrate Wa held by the upper chuck 84 are brought into contact with each other at their bonding surfaces SF and pressure-bonded together ( FIGS. 22 and 23 ). As a result, the upper substrate Wa and the lower substrate Wb are bonded together via the adhesive, and a superimposed substrate Wec is formed. The formed superimposed substrate Wec is carried out of the bonding device 7D and carried into the inspection device 1A in the reverse order to the order in which one of the upper substrates Wa, Wb was carried into the processing region A2.

[0123] The inspection device 1A performs an alignment inspection on the loaded overlapped substrate Wc as the substrate to be inspected, and determines whether the overlapped substrate Wc is in an aligned state or an unaligned state. As described above, according to the determination operation of the present disclosure, it is possible to accurately and simply detect a defective overlapped substrate Wc that is determined to be in an unaligned state.

[0124] (Fourth Embodiment) An inspection device 1B according to a fourth embodiment will be described with reference to Figures 25 to 31. In these figures, structures similar to those of the first embodiment are assigned common reference numerals, and redundant explanations will be omitted, with the focus being on differences from the first embodiment. Figure 25 is a structural diagram showing the inspection device 1B according to the fourth embodiment, and Figure 26 is an explanatory diagram showing the principle of spectroscopic measurement by a hyperspectral camera (hereinafter referred to as HSC) 11B constituting the inspection device 1B.

[0125] As shown in Figures 25 and 26, the HSC 11B includes a lens 12, a spectroscope 17, and a camera 18. In the fourth embodiment, the HSC 11B is configured to perform spectroscopic measurement by arranging, for example, 900 optical elements side by side. That is, the HSC 11B has a large number of measurement regions R in the inspection apparatus 1 described with reference to Figure 1 arranged side by side to form an elongated rectangular measurement region RB. By scanning the substrate W while moving this measurement region RB, efficient spectroscopic measurement of the entire surface of the substrate W can be performed. Specifically, the HSC 11B introduces incident light from the elongated rectangular measurement region RB through the lens 12 into the spectroscope 17 via a horizontal slit (not shown) and disperses the incident light, and then receives the light at an area sensor 19 provided in the camera 18.

[0126] As shown in FIG. 26 , the area sensor 19 has, like the measurement area RB described above, 900 optical elements arranged in a row in the x′ direction (horizontal direction). These optical element rows are arranged in a vertical direction intersecting the horizontal direction within the area irradiated with the light dispersed by the spectroscope 17. With this configuration, the area sensor 19 receives dispersed light of each wavelength in a matrix fashion using optical elements arranged within a vertical width (Y) in each of 900 regions (measurement areas R) obtained by dividing the horizontal width (X) of the measurement area RB. The HSC 11B configured in this way can generate the characteristic spectrum described above based on the results of measuring the luminous intensity of light received by each light-receiving element for each wavelength at multiple positions (900 locations in this example) in the horizontal direction of the measurement area RB in a single scan. The HSC11B performs spectroscopic measurement at short intervals while scanning the measurement area RB in a direction (e.g., the Y direction) that intersects the longitudinal direction of the measurement area RB (corresponding to the aforementioned "horizontal direction"), thereby decomposing the surface of the substrate W into a high-density matrix and performing spectroscopic measurement.

[0127] A data set including a large number of reflectance spectra acquired by one scan has a data size of, for example, terabytes and is stored in a memory (not shown) within the control device 100. These data sets are used by each part of the control unit 115 for each process of positioning and film thickness inspection. The above-described operation of acquiring data sets is repeated for subsequent scans.

[0128] Such an HSC 11B simultaneously performs spectroscopic measurement at each lateral position of the measurement region RB during one scan, eliminating variations in illuminance due to different measurement times. Furthermore, since the illuminance is generally the same at multiple adjacent positions in the X direction in the measurement region RB during one scan, the illuminance of light incident on the optical element at these corresponding positions can be considered to be generally the same. Furthermore, since the upper surface of the substrate W can be scanned in a short time and spectroscopic measurement can be performed in a high-density matrix, it is considered that there is almost no difference in illuminance between multiple adjacent positions along the Y direction, as well as in the X direction. Therefore, for each region on the surface of the substrate W that is spectroscopically measured in a high-density matrix, consisting of multiple adjacent positions in the X and Y directions where there is almost no difference in illuminance, it is not necessary to consider the effect of differences in illuminance of light irradiated onto the substrate W on the reflectance spectrum.

[0129] Fig. 27 is a partial plan view showing the measurement region RB where spectroscopic measurement is performed in the fourth embodiment. Fig. 28 is a graph illustrating exemplary characteristic spectra in the regions T1, T2a, and T2b shown in Fig. 27 , with the vertical axis representing the absorbance spectrum, which is the difference between the intensity of irradiated light and the intensity of reflected light, among the various characteristic spectra described above. It is assumed that a surface film 33 of the same thickness is formed in these regions T1, T2a, and T2b. In this example, scanning is performed while moving the measurement region RB on the substrate W. The absorbance spectrum measured in the measurement region RB at the position shown in Fig. 27 results in a number of slightly shifted absorbance spectra measured at each measurement position in the width direction of the measurement region RB. Therefore, when the absorbance spectra obtained by scanning the regions T1, T2a, and T2b with the measurement region RB are superimposed, the result is a group of characteristic spectra (absorbance spectrum group) including a range of intensity (light quantity) values, rather than the single linear characteristic spectrum with no range in reflectance values ​​as in Fig. 5 (Fig. 28). However, even in a group of absorbance intensity spectra with different intensity values, the intensity values ​​converge to approximately one point at the wavelength where the spectra in regions T1, T2a, and T2b intersect, and the aforementioned singular points P7 and P8 appear. Note that even when other characteristic spectra such as those described in the first and third embodiments are measured, characteristic spectrum groups and singular points with a certain range are obtained.

[0130] Therefore, according to the spectroscopic measurement of the fourth embodiment, the positioning of the present disclosure can be performed using a method similar to that of the first embodiment. For example, based on the flowchart described with reference to FIG. 9 , it is assumed that a determination is made as to whether the measurement region RB is located in the inspection region T1, the non-inspection region T2b, or the boundary region between the regions T1 and T2b. In this case, even if the measurement region RB shown in FIG. 27 is fixed and scanned without moving, a large number of reflectance spectrum groups (e.g., 900 light receiving elements, i.e., 900 pixels) can be obtained by the measurement in step S205 of FIG. 9 , as described above. Therefore, in subsequent step S206 of FIG. 9 , it is determined whether the singular points of these reflectance spectrum groups match those of the prediction model. If a reflectance spectrum matching the singular points is included (step S206: Yes), it is then confirmed whether the reflectance spectrum matches the reflectance spectrum of the inspection region according to the prediction model (step S207). If the reflectance spectra do not match (step S207: No), the measurement region RB is moved (step S208), and the operations of steps S205 to S207 are repeated.

[0131] Furthermore, as described above, when using the HSC 11B capable of spectroscopic measurement of a large number of points in a short period of time, it is not necessary to determine the positional state of the substrate W based on the flowchart of Figure 9. For example, a group of reflectance spectra obtained by moving the measurement region RB and scanning the entire surface of the substrate W may be collectively determined to determine whether or not any of the spectra includes a singular point that matches the reflectance spectrum of the inspection region T1 obtained by the prediction model. If any of the reflectance spectra matches the singular point and the reflectance spectrum of the prediction model, the position within the surface of the substrate W at which the reflectance spectrum was acquired is identified. The position at which the reflectance spectrum was acquired can be identified, for example, from the position of the measurement region RB at which the reflectance spectrum was acquired during the scanning period and the position of the pixel (optical element) within the measurement region RB.

[0132] Furthermore, as will be described in detail in the examples below, according to the positioning and film thickness inspection according to the present disclosure, particularly the fourth embodiment, blurring due to focus shift or changes in illuminance are unlikely to affect the identification of singular points, reflectance spectrum, etc. Therefore, even when applied to in-line inspection incorporated in a processing module where focus shift and illuminance changes are expected to occur frequently due to disturbances such as vibration generation and illuminance fluctuations, positioning and the like can be performed almost unaffected by these disturbances.

[0133] An example of this disturbance, illuminance fluctuation, will be described. Figure 29 is a graph illustrating the reflectance spectra obtained when a measurement region RB is positioned across two adjacent regions (e.g., regions T1 and T2b in Figure 27) having the same film thickness but different patterns, and the measurement region RB is spectroscopically measured under different illuminances L1 to L3. The solid line in the figure indicates a representative reflectance spectrum for one of regions T1 and T2b, while the dashed line indicates a representative reflectance spectrum for the other region. As described above, in the spectroscopic measurement of the fourth embodiment, in a single scan of spectroscopic analysis, the illuminance can be considered to be approximately the same at multiple horizontally consecutive positions within the measurement region RB. Therefore, the reflectance spectra of the adjacent regions measured in a single scan indicate that the measurement was performed at approximately the same illuminance.

[0134] On the other hand, as scanning is repeated for a large number of substrates W, the illuminance of the irradiated light may change due to, for example, contamination of a window through which the irradiated light and reflected light pass. In this regard, FIG. 29 shows that even when reflectance spectra are acquired under different illuminances L1 to L3, the wavelength of the singular point where the reflectance spectra acquired in regions T1 and T2b, which have different pattern shapes, intersect remains unchanged. Therefore, according to the spectroscopic measurement of the fourth embodiment, even if illuminance fluctuations expected in inline inspection occur, the positional state of the substrate W can be stably identified based on the wavelength of the singular point, which is less susceptible to such illuminance fluctuations. In the following modified example, an example of inline inspection using a film forming module 7E equipped with an inspection device 1B will be described.

[0135] (Application Example of the Fourth Embodiment) In this application example of the fourth embodiment, the inspection apparatus 1B is attached to a semi-batch type film formation module 7E (substrate processing apparatus) installed in place of the third processing module 7C where the film formation processing is performed, and an in-line inspection is performed immediately after the film formation processing in the film formation module 7E. FIGS. 30 and 31 are a longitudinal side view and a transverse plan view, respectively, of the semi-batch type film formation module 7E. As shown in these figures, the processing vessel 71B is, for example, a flat, cylindrical vacuum vessel, and a horizontal rotating table (table) 72B serving as a mounting table is provided therein. The rotating table 72B rotates around a central axis (rotation axis) M by a rotation mechanism 73B provided at its center. A plurality of substrates W are placed on the rotating table 72B, and the substrates W are revolved around the central axis M as the rotating table 72B rotates.

[0136] 31 , a source gas nozzle 51 for supplying a source gas and a reaction gas nozzle 52 for supplying a reaction gas are provided in a region through which the substrate W passes during revolution. Two regions between the source gas nozzle 51 and the reaction gas nozzle 52 are provided with purge gas nozzles 53 for supplying purge gas. These source gas, reaction gas, and purge gas nozzles 51, 52, and 53 are arranged circumferentially spaced apart from one another and extend horizontally from the sidewall of the processing vessel 71B toward the center of the processing vessel 71B. The source gas, reaction gas, and purge gas nozzles 51, 52, and 53 are provided with a number of gas discharge holes 50 ( FIG. 30 ) along the longitudinal direction for discharging the various gases supplied into the processing vessel 71B.

[0137] 31 , by disposing partition members 56 in each of the two regions surrounded by the two purge gas nozzles 53, the space within the processing vessel 71B is partitioned into a region 54 to which a source gas is supplied and a region 55 to which a reactant gas is supplied. This prevents the source gas and the reactant gas from mixing within the processing vessel 71B. The region 54 to which the source gas is supplied and the region 55 to which the reactant gas is supplied are connected to an exhaust device 79 ( FIG. 30 ) via exhaust paths 57 and 58 equipped with valves V1 and V2, respectively.

[0138] The gas supply unit 77 (process gas supply unit) is composed of supply paths independent of each other and includes source gas, reactive gas, and purge gas supply systems 77b, 77c, and 77d, each having its own gas supply source. The source gas, reactive gas, and purge gas supply systems 77b, 77c, and 77d are each provided with a flow rate controller such as a valve or a mass flow controller, and are connected to the source gas, reactive gas, and purge gas nozzles 51, 52, and 53. As a result, the source gas, reactive gas, and purge gas nozzles 51 to 53 are continuously supplied with the source gas, reactive gas, and purge gas from their respective supply sources.

[0139] In this film formation module 7E, as the turntable 72B rotates, the substrate W passes through the source gas supply region 54, causing the source gas to be adsorbed onto the wafer surface. Next, as the substrate W passes through the reactive gas supply region 55, the source gas and the reactive gas on the wafer surface react to form a film. In this way, as the substrate W passes alternately through the source gas supply region 54 and the reactive gas supply region 55, a surface film 33 is formed on the substrate W by the ALD method, as in the first embodiment.

[0140] The processing vessel 71B has an opening at a portion above the region 55 to which the reaction gas is supplied, and the opening is closed by a window 76B that transmits light for the spectroscopic measurement described above. To maintain the airtightness of the processing vessel 71B, the periphery of the window 76B is sandwiched via an O-ring between the processing vessel 71B and a top lid 76C attached to the top of the processing vessel 71B. The top lid 76C is open above the central region of the window 76B that transmits light so as not to block the central region.

[0141] HSC 11B is disposed above window 76B and is configured to receive light reflected from measurement region RB via the central region of window 76B. HSC 11B is installed with lens 12 facing so that measurement region RB is positioned within a range that encompasses the diameter of the revolving substrate W ( FIG. 36 ). Furthermore, the orientation of lens 12 of HSC 11B is adjusted so that measurement region RB extends radially from the sidewall of processing vessel 71B toward the center of processing vessel 71B, i.e., toward central axis M, in a plan view, similar to each of nozzles 51-53.

[0142] After the surface film 33 is formed, each substrate W is scanned by the HSC 11B while revolving within the film formation module 7E. That is, after each revolving substrate W enters the area below the measurement region RB, its entire surface passes below the measurement region RB. The HSC 11B performs spectroscopic measurement of the moving measurement region RB at short intervals, thereby measuring the reflectance spectrum of the entire upper surface of the substrate W. Such in-line inspection in the film formation module 7E allows the measurement region RB to be positioned in the inspection region T1 efficiently and accurately, and film thickness inspection to be performed.

[0143] Here, in the past, it was considered difficult to use high-precision spectroscopic measurement such as the HSC 11B in in-line inspection of the film deposition module 7E due to the influence of various disturbances. Examples of disturbances include changes in illuminance over time due to clouding of the window 76B caused by various gases supplied during the film deposition process, and poor diagnosis due to blurred images. Other disturbances include tilting of the turntable 72B from the center toward the outside and defocusing caused by rotational vibration of the turntable 72B. It was considered difficult to obtain accurate measurement results using spectroscopic measurement using the conventional HRC 11B due to such disturbances. In contrast, by using the reflectance spectrum and its singular points as disclosed herein, positioning and film thickness inspection can be performed accurately and efficiently without being affected by disturbances, as will be described in the examples below.

[0144] (Other Modifications) The HSC 11B of the fourth embodiment is not limited to one that scans the linear measurement region RB at short intervals while moving. For example, the HSC 11B may be another type of HSC, such as one that performs spectroscopic measurement by changing multiple spectral filters corresponding to different bands while keeping the area-shaped measurement region RB fixed. Alternatively, instead of the HSC 11B, multiple measurement units 11 like those in the first embodiment may be prepared, and the coaxial probes 13 of each may be moved integrally, thereby performing spectroscopic measurement at the same time while each measurement region R moves integrally like the measurement region RB.

[0145] Furthermore, the inspection device 1B is not limited to being attached to the film formation module 7E, but may be attached to another processing module, or may be disposed in the film formation processing system 10 as a stand-alone inspection device as shown in the first embodiment without being attached to a processing module. In this case, the inspection device 1B is not limited to being attached to a rotary stage, but may be placed on a rail-like stage that moves in one direction and spectroscopic measurement may be performed while moving the substrate, or the measurement region RB may be moved relatively by moving the HSC 11B.

[0146] (Experiment 1) In order to confirm the effectiveness of the alignment inspection of the third embodiment, unbonded wafers were superimposed with an extremely small amount of misalignment, and the reflectance spectrum of a predetermined inspection area was measured.

[0147] A. Experimental Conditions Two TEG (Test Element Group) wafers were prepared, each with the same L / S pattern formed on a surface layer with a thickness of 350 nm. The thickness of the portion of these wafers that did not include the L / S pattern was 775 μm, and the portion with this thickness was covered with silicon oxide (SiO ２ The wafers include a structure in which metal such as metal pads 23 is embedded in a concave-convex pattern of silicon nitride (SiN) or the like. These wafers were stacked unbonded with their surface layers facing upward, with no misalignment of the L / S patterns, and placed on a flat table. A small seismic intensity was then applied to the table to create an extremely small amount of misalignment between the wafers that makes quantitative measurement difficult, and five experiments were conducted to measure the reflectance spectrum (Experiments 1 to 5). Each experiment was measured five times, and it was confirmed that there was no change in the reflectance spectrum for all five runs.

[0148] B. Experimental Results Figure 32 shows the reflectance spectra showing the results of Experiment 1. In Figure 32, the reflectance spectra of Experiments 1 to 5 are displayed using different line types, but since some reflectance spectra overlap with other reflectance spectra, they are shown using solid lines, dashed lines, one-dot chain lines, and two-dot chain lines.

[0149] As shown in Figure 32, no change was observed in the reflectance spectra of Experiments 1 to 5 over the five measurements. This confirmed that the reflectance spectra corresponded to the misalignment direction and amount, and that the prediction model could be used to estimate the misalignment direction and amount from the reflectance spectrum measured for the superimposed substrate We to be inspected. Furthermore, in Figure 32, singular points P4 and P5 appeared where the five reflectance spectra intersected, and it was confirmed that the wavelength at singular point P4 corresponded to the wavelength corresponding to the wafer thickness. Therefore, it was confirmed that the singular points also had the characteristics described above. Therefore, it was confirmed that the alignment inspection of the third embodiment could easily detect even extremely small misalignments of the superimposed substrate We with high accuracy.

[0150] (Experiment 2) Under conditions where defocusing and illuminance differences occur as expected in in-line inspection as described above, the trends of the singular points and reflectance spectra identified by the spectroscopic measurement of the fourth embodiment were confirmed.

[0151] A. Experimental Conditions A bare wafer having multiple sets of test patterns formed on its surface was prepared as the substrate W to be used in the experiment. Each set of test patterns had two different pattern areas formed thereon, with a SiN film of the same thickness formed thereon, in order to identify a singular point for each set. The film thicknesses of the test patterns in each set were different in order to identify a singular point corresponding to the film thickness. The entire top surface of the substrate W was then scanned, and the two pattern areas of each set of test patterns were spectroscopically measured. Reflectance spectra of the two different pattern areas for each set of test patterns were obtained, and the singular points corresponding to the film thickness of each set of test patterns were identified from the intersections of these spectra.

[0152] In order to reproduce an environment with focus deviation and illuminance differences, spectroscopic measurement was performed at low resolution, where the width of the measurement region RB on the substrate W was made relatively large by adjusting the height of the lens 12, and at high resolution, where the width was made relatively small and the resolution was about 110 times higher than the low resolution. The high resolution reproduced a state in which the illuminance was relatively high and in focus, and the low resolution reproduced a state in which the illuminance was relatively low and out of focus.

[0153] In addition, for the reflectance spectra of each data set acquired by the high- and low-resolution scans, data sets were also prepared by converting them into absolute reflectance in order to cancel out the illuminance difference. Figure 33 is a table summarizing the spectroscopic measurement conditions <1> to <5>. Data sets were prepared for each of the conditions <1> to <5>, and the trends of the singular points identified under each of the conditions <1> to <5> were confirmed from these data sets. Figure 33 also lists the wavelengths of the singular points measured under conditions <1> to <4> for a test pattern among multiple sets of test patterns. Condition <5> was obtained by spectroscopic measurement in the first embodiment using a single measurement area R.

[0154] Then, it was confirmed whether the singular point wavelengths identified by the measurement of each set of test patterns under each of the conditions <1> to <5> corresponded to the film thickness of each set of test patterns. As an additional test, singular points in test patterns with relatively large voids were also confirmed by low-resolution spectroscopic measurement.

[0155] B. Experimental Results Figure 34 is a graph showing the wavelengths of the singular points, illustrating the results of Experiment 2. The figure plots the singular point wavelengths of each set of test patterns for each of the above conditions, with the horizontal axis representing the singular point wavelength [nm] obtained from the reflectance spectrum of condition <1> (high resolution, no absolute value conversion), and the vertical axis representing the singular point wavelength [nm] obtained from the reflectance spectrum of each of conditions <2> to <5>. In other words, Figure 34 plots the singular point wavelength (vertical axis) of each set of test patterns obtained under condition <1> against the singular point wavelength (horizontal axis) measured for the same set of test patterns under each of conditions <2> to <5>.

[0156] As shown in FIG. 34 , the plots of the singularity wavelengths under condition <1> and the singularity wavelengths under conditions <2> to <4> overlap on a straight line with a slope of 1. As shown by these plots, the singularity wavelengths measured under conditions <1> to <4> are identical and do not differ depending on the film thickness of each test pattern, regardless of whether the measurements are low or high resolution or whether conversion to absolute reflectance is performed. Therefore, it was confirmed that the spectroscopic measurement of the fourth embodiment enables positioning and other tasks based on the reflectance spectrum utilizing the singularity of the present disclosure, without being affected by fluctuations in the focal length or illuminance of the lens 12. It was also confirmed that the singularity can be accurately identified from the reflectance spectra of the measured relative values ​​without converting to absolute reflectance to cancel differences in illuminance.

[0157] It was confirmed that the singularity wavelengths under conditions <1> to <4> shown in Fig. 34 coincide with the wavelengths of the singularities corresponding to the respective film thicknesses. Fig. 35 shows the correspondence between some of the singularity wavelengths identified under conditions <3> to <5> and film thicknesses. As shown in the figure, the singularity wavelengths under conditions <3> and <4> are proportional to film thickness, as described in the first embodiment, and it was confirmed that there is also a correspondence between film thickness and the singularity in the spectroscopic measurement according to the fourth embodiment.

[0158] Although singularities appeared in the pattern region where the film thickness was 0 nm, i.e., no SiN film was provided, their wavelengths were values ​​that deviated from the proportional relationship with film thickness observed for other singularity wavelengths for film thicknesses of 0 nm or more. Furthermore, as shown in Figures 34 and 35, the singularity wavelengths under condition <5> in the spectroscopic measurement of the first embodiment, including the 0 nm film thickness, were each approximately 5 nm shorter than the singularity wavelengths under conditions <1> to <4> in the spectroscopic measurement of the fourth embodiment. This is thought to be due to a deviation specific to the calibration of the spectrometer used under condition <5>. Therefore, it was inferred that if this deviation was identified and corrected in advance, the spectroscopic measurement of the first embodiment would have the same level of accuracy as the fourth embodiment.

[0159] Furthermore, Figure 35 shows the results of additional testing, showing singular points identified by the spectroscopic measurement of the fourth embodiment for another test pattern in which a SiN film containing voids was formed. Singular points appeared at wavelengths outside the singular point wavelength corresponding to the film thickness. Thus, it was confirmed that accurate singular points appeared and SiN films containing voids could be identified even with the low-resolution spectroscopic measurement of the fourth embodiment. Furthermore, it was inferred that low-resolution spectroscopic measurement, which can measure a wide range in one scan, makes it easier to find the reflectance spectra of different pattern areas within one scan, thereby enabling rapid identification of singular points. It was also inferred that spectroscopic measurement can be performed quickly in a high-density matrix on the substrate W, enabling efficient spectroscopic measurement.

[0160] (Experiment 3) When performing positioning and film thickness inspection using spectroscopic measurement by the HSC 11B in the film forming module 7E shown in the modified example of the fourth embodiment, it is confirmed whether the position on the substrate W can be detected for the multiple reflectance spectra contained in each data set.

[0161] A. Experimental Conditions Figure 36 is a top perspective view of multiple substrates W that were the subject of spectroscopic measurement in Experiment 3. In Experiment 3, spectroscopic measurement was reproduced for multiple substrates W that were placed on a turntable 72B and revolving within a film formation module 7E, as shown in the modified example of the fourth embodiment. The position of the lens 12 was set under low-resolution conditions so that the width (X) of the measurement region RB was longer than the diameter of the substrate W. Then, the measurement region RB, which was positioned in an elongated state in the direction toward the central axis M, was revolved relative to the multiple substrates W, and spectroscopic measurement was performed every short period of time to obtain a large number of data sets.

[0162] B. Experimental Results Figures 37 and 38 are graphs showing the experimental results of Experiment 3. Figure 37 shows the intensity profile of one wavelength acquired at each position in the measurement region RB at time t1. The intensity range inside the two dashed lines in Figure 37 is the intensity range used for positioning and film thickness inspection in the present disclosure. Reflected light with an intensity falling within this intensity range can be determined to have been reflected on the substrate W and then incident on the HSC 11B, while reflected light with an intensity weaker than this intensity range is considered to have been reflected outside the substrate W, for example, on the turntable 72B. It can be seen that the boundary positions of these reflected lights with different intensities correspond to the black dots A and white dots B on the periphery of the substrate W shown in Figure 36.

[0163] Figure 38 is a graph showing the time-dependent changes in the positions of points A and B in the measurement region RB, which were identified using the correspondence between position and intensity in the measurement region RB as shown in Figure 37. Times t0 to t5 in Figure 38 correspond to times t0 to t5 in Figure 36. As shown in Figure 38, the intersecting positions of the solid and dashed lines indicating the positions of points A and B represent the initial or final entry positions of the substrate W below the measurement region RB. Points A and B gradually move away from this entry position, and at time t2, when points A and B are furthest apart, it is believed that the diameter of the substrate W is located below the measurement region RB. After time t2, points A and B gradually move closer to each other and become the same position. The period during which points A and B are not detected corresponds to the period until the next substrate W arrives at the measurement region RB. The period during which point A moves from the inside to the outside near time t5 is the period during which the measurement region RB moves along the notch.

[0164] As shown in Fig. 38, by determining the positions of points A and B in measurement region RB, it is possible to identify the position on substrate W of measurement region RB, which relatively revolves, from the change in the measurement range in measurement region RB from minimum width (initial entry position) → maximum width (substrate diameter position) → minimum width (final entry position) and time. The position in measurement region RB at which a certain acquired reflectance spectrum was measured can be determined by comparing the data with data indicating the peripheral position of substrate W, as shown in Fig. 37. As described above, in-line inspection such as that shown in the modified example of the fourth embodiment makes it possible to identify the measurement position of a certain reflectance spectrum acquired during film thickness inspection or positioning of inspection region T1.

[0165] As shown in FIG. 36 , the relative movement speed of the measurement region RB relative to the substrate W is smaller inside the measurement region RB and larger outside it, and the displacement directions at these positions are not strictly parallel. Therefore, differences in the amount and direction of movement inside and outside the measurement region RB may affect the reflectance spectrum at each position, potentially resulting in errors in the positioning using the singular points of the present disclosure. To address this issue, a separate test pattern was moved irregularly on the revolving or rectilinearly moving substrate W, and spectroscopic measurements in the fourth embodiment were performed under various conditions, such as low resolution, high resolution, and with or without absolute value conversion, as shown in Experiment 3, to confirm the trends of singular points, etc. In this experiment, as in Experiment 3, singular points corresponding to film thicknesses were identified from the reflectance spectrum measured for the test pattern. This also confirms that accurate positioning, etc., can be achieved without errors, even in in-line inspection such as the modified fourth embodiment.

[0166] It should be noted that the embodiments disclosed herein should be considered to be illustrative and not restrictive in all respects, and various omissions, substitutions, modifications, and combinations may be made to the above-described embodiments without departing from the scope and spirit of the appended claims.

[0167] W: Substrate Wc: Laminated substrate R: Measurement area 33: Surface film 121: Learning data

Claims

1. A method for inspecting a substrate by irradiating it with light, comprising: a step of irradiating a measurement area set on the substrate with light to measure a characteristic spectrum indicating the absorption or reflection characteristics for the wavelength of the light; and a step of, when the characteristic spectrum changes according to the structure of the substrate in the measurement area, correlating the characteristic spectrum with the film thickness if a film is formed on the surface of the substrate or the thickness of the substrate using a learning model generated by learning about a plurality of substrates, and determining the arrangement state of the substrate in the measurement area from the structure of the substrate identified by comparing the inspection target spectrum, which is the characteristic spectrum measured in the step of measuring the characteristic spectrum, with a reference spectrum, which is the characteristic spectrum obtained from the learning model based on the film thickness or substrate thickness identified from the substrate from which the inspection target spectrum was obtained.

2. The method according to claim 1, wherein the characteristic spectrum is any one of a reflected light intensity spectrum, an absorbance intensity spectrum, an absorbance spectrum, and a reflectance spectrum from the measurement area.

3. The method according to claim 1, wherein the light irradiated to the measurement area is measurement light for performing spectroscopic measurement of a film on a surface of the substrate on which a concave-convex pattern, which is a structure of the substrate, is formed, and in the step of determining the arrangement state of the substrate, the learning model generated by learning the characteristic spectrum formed in accordance with the concave-convex pattern in correspondence with the film thickness of the film is used to determine whether or not the measurement area is located within the area on which the concave-convex pattern is formed from a comparison result between the inspection target spectrum obtained by identifying the film thickness and the comparison spectrum, and when it is determined in the step of determining the arrangement state that the measurement area is not located within the area on which the concave-convex pattern is formed, the step of moving the measurement area, the step of measuring the characteristic spectrum, and the step of determining the arrangement state are repeated.

4. The method according to claim 3, wherein the uneven pattern is formed in a measurement area where the spectroscopic measurement is performed.

5. The method according to claim 4, wherein an adjacent region of the measurement region of the substrate is provided adjacent to the measurement region, and the adjacent region is formed with a different unevenness pattern from the unevenness pattern of the measurement region, and the step of determining the arrangement state of the substrate uses the learning model generated by learning the characteristic spectrum formed according to the unevenness patterns of the measurement region and the adjacent region, and determines whether the measurement region is located in the measurement region or the adjacent region from a comparison result between the inspection target spectrum obtained by identifying the film thickness and the comparison spectrum, and if the step of determining the arrangement state determines that the measurement region is located in the adjacent region, the step of moving the measurement region moves the measurement region based on a direction and movement distance preset based on the arrangement relationship between the measurement region and the adjacent region.

6. The method according to claim 3, wherein the uneven pattern is an alignment mark formed in a preset area on the surface of the substrate, the alignment mark being provided for adjusting the alignment of the substrate.

7. The method according to claim 3, further comprising a step of specifying values ​​of the inspection target spectrum and the comparison spectrum at the singular point after the step of measuring the characteristic spectra, when the wavelength of the light at a position where the multiple types of characteristic spectra obtained by changing the structure of the substrate intersect, or the wavelength of the light at a position where the difference between the values ​​of the multiple types of characteristic spectra is minimum when the multiple types of characteristic spectra do not intersect, is called a singular point, and when the difference between the value specified in the inspection target spectrum and the value specified in the comparison spectrum is equal to or greater than a preset tolerance, the step of determining the arrangement state of the substrate is not executed.

8. The method according to claim 7, wherein the singular point varies depending on the thickness of a film on the surface of the substrate, and the step of identifying a value at the singular point includes identifying the value at the singular point that corresponds to the film thickness.

9. The method according to claim 1, wherein the substrate is a bonded substrate formed by bonding a first substrate and a second substrate, the light irradiated to the measurement area is infrared light for measuring an internal structure of the bonded substrate, and in the step of determining the arrangement state of the substrate, the learning model generated by learning the characteristic spectrum formed according to the bonding state between the first substrate and the second substrate in correspondence with the thickness of the bonded substrate is used, and whether or not the first substrate and the second substrate are bonded in an accurate position is determined based on a comparison result between the inspection target spectrum obtained by identifying the thickness of the substrate and the comparison spectrum.

10. The method according to claim 9, wherein the bonding state is a misalignment of metal pads provided on the first substrate and the second substrate, respectively, which connect the first substrate and the second substrate to each other.

11. The method of claim 1, wherein the step of measuring the characteristic spectrum includes measuring the characteristic spectrum with a hyperspectral camera.

12. The method according to claim 11, wherein the measurement area of ​​the characteristic spectrum measured by the hyperspectral camera is configured as an elongated rectangle whose longitudinal dimension is greater than the diameter of the substrate, and in the measurement by the hyperspectral camera, the measurement area is moved relative to a plurality of the substrates placed on a stage in a fixed direction, thereby causing the measurement area to scan the entire surface of the substrate.

13. An apparatus for inspecting a substrate, comprising: a stage; an inspection unit which irradiates light onto a measurement area set on the substrate placed on the stage and receives reflected light; and a control unit, wherein the control unit outputs a control signal to execute the steps of: irradiating light onto the measurement area and measuring a characteristic spectrum which indicates the absorption or reflection characteristics for the wavelength of the light; and, when the characteristic spectrum changes according to the structure of the substrate in the measurement area, correlating the characteristic spectrum with the film thickness if a film is formed on the surface of the substrate or the thickness of the substrate, using a learning model generated by learning about a plurality of substrates, and determining the arrangement state of the substrate in the measurement area from the structure of the substrate identified by comparing the inspection target spectrum, which is the characteristic spectrum measured in the step of measuring the characteristic spectra, with a reference spectrum, which is the characteristic spectrum obtained from the learning model based on the film thickness or substrate thickness identified from the substrate from which the inspection target spectrum was obtained.

14. The apparatus of claim 13, wherein the characteristic spectrum is one of a reflected light intensity spectrum, an absorbance intensity spectrum, an absorbance spectrum, and a reflectance spectrum from the measurement area.

15. The device described in claim 13, wherein the inspection unit is configured to be able to move the measurement area, and the light irradiated onto the measurement area is measurement light for performing spectroscopic measurement of a film on a surface of the substrate on which a concave-convex pattern that is a structure of the substrate is formed, and the control unit, in the step of determining the arrangement state of the substrate, uses the learning model generated by learning the characteristic spectrum formed in accordance with the concave-convex pattern in correspondence with the film thickness of the film, and determines whether or not the measurement area is located within the area on which the concave-convex pattern is formed from a comparison result between the inspection target spectrum obtained by identifying the film thickness and the comparison spectrum, and if it is determined in the step of determining the arrangement state that the measurement area is not located within the area on which the concave-convex pattern is formed, outputs a control signal for repeatedly executing a step of moving the measurement area, a step of measuring the characteristic spectrum, and a step of determining the arrangement state.

16. The apparatus according to claim 15, wherein the uneven pattern is formed in a measurement area where the spectroscopic measurement is performed.

17. The device described in claim 16, wherein an adjacent region of the measurement region of the substrate is provided adjacent to the measurement region, and the control unit, in the step of determining the arrangement state of the substrate, uses the learning model generated by learning the characteristic spectrum formed in accordance with the unevenness patterns of the measurement region and the adjacent region, and determines whether the measurement region is located in the measurement region or the adjacent region from a comparison result between the inspection target spectrum obtained by specifying the film thickness and the comparison spectrum, and if it is determined in the step of determining the arrangement state that the measurement region is located in the adjacent region, then in the step of moving the measurement region, outputs a control signal for moving the measurement region based on a direction and movement distance preset based on the arrangement relationship between the measurement region and the adjacent region.

18. The apparatus according to claim 15, wherein the uneven pattern is an alignment mark formed in a preset area on the surface of the substrate, the alignment mark being provided for adjusting the alignment of the substrate.

19. The apparatus described in claim 15, wherein the wavelength of light at a position where multiple types of characteristic spectra obtained by changing the structure of the substrate intersect, or the wavelength of light at a position where the difference between the values ​​of the multiple types of characteristic spectra is minimum when the multiple types of characteristic spectra do not intersect, is called a singular point, the control unit includes a step of specifying values ​​of the inspection target spectrum and the comparison spectrum at the singular point after the step of measuring the characteristic spectra, and outputs a control signal so as not to execute the step of determining the placement state of the substrate if the difference between the value specified in the inspection target spectrum and the value specified in the comparison spectrum is equal to or greater than a preset tolerance.

20. The apparatus described in claim 19, wherein the singular point changes depending on the thickness of a film on the surface of the substrate, and in the step of identifying a value at the singular point, the control unit outputs a control signal for identifying the value at the singular point corresponding to the film thickness.

21. The device described in claim 13, wherein the substrate is a bonded substrate formed by bonding a first substrate and a second substrate, and the light irradiated to the measurement area is infrared light for measuring an internal structure of the bonded substrate, and the control unit, in the step of determining the arrangement state of the substrate, uses the learning model generated by learning the characteristic spectrum formed in accordance with the bonding state of the first substrate and the second substrate in correspondence with the thickness of the bonded substrate, and outputs a control signal for determining whether the first substrate and the second substrate are bonded in an accurate position based on a comparison result between the inspection target spectrum obtained by identifying the thickness of the substrate and the comparison spectrum.

22. The device according to claim 21, wherein the bonding state is a misalignment of metal pads provided on the first substrate and the second substrate, respectively, which connect the first substrate and the second substrate to each other.

23. The apparatus of claim 13, wherein the inspection unit comprises a hyperspectral camera that measures the inspection target spectrum.

24. The apparatus of claim 23, wherein the measurement area of ​​the hyperspectral camera is configured as an elongated rectangle whose longitudinal dimension is greater than the diameter of the substrate, and the measurement area scans the entire surface of the substrate by moving the measurement area relative to the multiple substrates placed on the stage in a fixed direction.

25. A substrate processing apparatus comprising: the apparatus according to claim 24; a rotation axis for rotating the stage; a processing vessel housing the stage on which the plurality of substrates are placed and which revolves around the rotation axis; and a processing gas supply unit for supplying processing gas to the substrates, wherein the processing gas is supplied to the substrates placed on the stage to process them, and then the substrates are inspected.

Images