Shape characteristic value estimation device, shape characteristic value estimation method, and storage medium

By using the CD estimation model and the substrate impact model before processing in the shape characteristic value calculation device, the problem of low accuracy of the estimation of the shape characteristic value value of the object film under the influence of multi-layer film on the substrate is solved, and high-precision CD estimation is achieved.

CN119063637APending Publication Date: 2024-12-03TOKYO ELECTRON LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410638969.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Priority Date
2023-06-01
Filing Date
2024-05-22
Publication Date
2024-12-03

Smart Images

  • Figure CN119063637A_ABST
    Figure CN119063637A_ABST
Patent Text Reader

Abstract

The invention provides a shape characteristic value estimation device, a shape characteristic value estimation method, and a storage medium capable of estimating a characteristic value of a shape of a target film formed on a substrate with high precision. A shape characteristic value estimation device (CD estimation device) is provided with a post-processing image acquisition unit (102), an estimation unit (104), a pre-processing substrate influence model creation unit (107), and an estimation result correction unit (108). A pre-processing substrate impact model creation unit (107) divides the plurality of post-processing images into a first set and a second set. Creating a first candidate, which is a candidate of a pre-processing substrate influence model, on the basis of the first set, and creating a second candidate, which is a candidate of the pre-processing substrate influence model, on the basis of the second set; a pre-processing substrate influence model is determined by comparing a result obtained by correcting the estimation result by the estimation result correction unit (108) on the basis of the first candidate with a result obtained by correcting the estimation result by the estimation result correction unit (108) on the basis of the second candidate.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to a shape characteristic value estimating device, a shape characteristic value estimating method and a storage medium. Background Art

[0002] Patent Document 1 discloses calculating the film thickness of a film formed on a substrate based on an image obtained by photographing the substrate surface. In this case, correlation data is disclosed that associates pixel values ​​obtained from a prepared photographed image with film thickness measurement values ​​at each coordinate corresponding to the pixel values.

[0003] Prior art literature

[0004] Patent Literature

[0005] Patent Document 1: Japanese Patent Application Publication No. 2015-215193 Summary of the invention

[0006] Technical problem to be solved by the invention

[0007] The present invention provides a technology for estimating the characteristic value of the shape of a target film formed on a substrate with high accuracy.

[0008] Technical solutions for solving technical problems

[0009] A shape characteristic value estimation device according to one aspect of the present invention includes a processed image acquisition unit, which acquires a processed image including image information about the surface of a substrate after a target film is processed. The shape characteristic value estimation device includes an estimation unit. The estimation unit estimates the shape characteristic value of the target film by applying an estimation model for estimating the shape characteristic value of the target film to the processed image, wherein the estimation model involves the correlation between information about the color of the substrate contained in the processed image and a shape characteristic value as a characteristic value of the shape of the target film formed on the substrate. The shape characteristic value estimation device includes a pre-processed substrate influence model creation unit. The pre-processed substrate influence model creation unit creates a pre-processed substrate influence model, which is a model of the correlation between the difference between the estimation result of the shape characteristic value of the target film obtained by the estimation unit and the shape characteristic value of the target film obtained without using the estimation model and the information about the color of the pre-processed substrate, wherein the pre-processed substrate is the substrate before the target film is processed. The shape characteristic value estimation device includes an estimation result correction unit, which corrects the estimation result of the shape characteristic value of the object film obtained by the estimation unit based on the pre-processing substrate influence model. The pre-processing substrate influence model creation unit is configured to be able to perform a process of dividing a plurality of processed images into a first set and a second set. The pre-processing substrate influence model creation unit is configured to be able to perform the following process: creating a first candidate as a candidate for the pre-processing substrate influence model based on the first set, and creating a second candidate as a candidate for the pre-processing substrate influence model based on the second set. The pre-processing substrate influence model creation unit is configured to be able to perform the following process: determining the pre-processing substrate influence model by comparing the result of the estimation result correction unit correcting the estimation result based on the first candidate with the result of the estimation result correction unit correcting the estimation result based on the second candidate.

[0010] Effects of the Invention

[0011] According to the present invention, there is provided a technique for estimating the characteristic value of the shape of a target film formed on a substrate with high accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 This is a schematic diagram showing an example of a schematic configuration of a substrate processing system.

[0013] Figure 2 It is a schematic diagram showing an example of a coating and developing device.

[0014] Figure 3 It is a schematic diagram showing an example of an inspection unit.

[0015] Figure 4 This is a block diagram showing an example of the functional structure of the control device.

[0016] Figure 5 This is a diagram schematically showing changes in information on color acquired from image data obtained by imaging a plurality of wafers.

[0017] Figure 6 This is a block diagram showing an example of the hardware configuration of the control device.

[0018] Figure 7 This is a flowchart showing an example of a CD estimation method.

[0019] Figure 8 This is a flowchart showing an example of a CD estimation method.

[0020] Fig. 9 This is a schematic diagram for explaining an example of the deviation of the estimation result of the CD estimation model.

[0021] Fig.10 It is a schematic diagram illustrating an example of a model affecting a substrate before processing.

[0022] Fig.11 is a flow chart showing the process of creating a pre-processing substrate impact model.

[0023] Fig.12 is a diagram illustrating a method of partitioning a sample involved in creating a pre-processing substrate effect model.

[0024] Description of Reference Numerals

[0025] 1…substrate processing system, 2…coating and developing device (shape characteristic value estimation device), 3…exposure device, 4…carrier block, 5…processing block, 6…interface block, 11-14…processing module, 30…housing, 31…holding unit, 32…linear driving unit, 33…shooting unit, 34…reflecting unit, 35…camera, 36…half-reflecting mirror, 37…light source, 100…control device, 101…pre-processing image acquisition unit, 102…post-processing image acquisition unit, 103…image information storage unit, 104…estimation unit, 105…estimation model storage unit, 106…shape characteristic value information storage unit, 107…pre-processing substrate influence model creation unit, 108…estimation result correction unit. DETAILED DESCRIPTION

[0026] Hereinafter, various exemplary embodiments will be described in detail with reference to the accompanying drawings. In addition, in each of the drawings, the same reference numerals are given to the same or corresponding parts.

[0027] [Substrate processing system]

[0028] The substrate processing system 1 is a system for forming a photosensitive coating on a substrate, exposing the photosensitive coating, and developing the photosensitive coating. The substrate to be processed is, for example, a semiconductor wafer W.

[0029] The substrate processing system 1 includes a coating and developing device 2 and an exposure device 3. The exposure device 3 performs an exposure process of a resist film (photosensitive coating) coated on a wafer W (substrate). Specifically, the exposure device 3 irradiates an energy line to an exposure target portion of the resist film by a method such as liquid immersion exposure. The coating and developing device 2 performs a process of coating a resist film on the surface of the wafer W (substrate) before the exposure process performed by the exposure device 3, and performs a development process of the resist film after the exposure process. By performing a series of processes, a resist film of a predetermined pattern is formed.

[0030] [Substrate processing device]

[0031] Hereinafter, as an example of a substrate processing device, the structure of the coating and developing device 2 will be described. Figure 1 and Figure 2 As shown, the coating and developing device 2 includes a carrier block 4, a processing block 5, an interface block 6 and a control device 100. The coating and developing device 2 as a substrate processing device described in this embodiment is equivalent to a shape characteristic value estimating device that estimates the shape characteristic value of the shape of the object film formed on the substrate. The "shape characteristic value" of the shape of the object film in this embodiment is equivalent to the characteristic quantity of the shape of the object film. As an example, the shape characteristic value can include a critical dimension (CD: Critical Dimension) such as a line width and an aperture of the object film or a film thickness of the object film. In the following embodiment, the case where the coating and developing device 2 is used as a shape characteristic value estimating device to estimate the CD of the object film is described. The function of estimating the CD of the coating and developing device 2 will be described later.

[0032] The carrier block 4 is used to introduce the wafer W into the coating and developing device 2 and to guide the wafer W out of the coating and developing device 2. For example, the carrier block 4 can support a plurality of carriers C (storage units) for the wafer W, and has a conveying device A1 including a transfer arm built therein. The carrier C stores a plurality of, for example, circular wafers W. The conveying device A1 takes out the wafer W from the carrier C and delivers it to the processing block 5, and receives the wafer W from the processing block 5 and returns it to the carrier C. The processing block 5 has a plurality of processing modules 11, 12, 13, and 14.

[0033] The processing module 11 is equipped with a plurality of coating units U1, a plurality of heat treatment units U2, a plurality of inspection units U3, and a conveying device A3 for conveying wafers W to these units. The processing module 11 forms a lower film on the surface of the wafer W using the coating unit U1 and the heat treatment unit U2. The coating unit U1 of the processing module 11, for example, rotates the wafer W at a predetermined rotation speed while applying a processing liquid for forming the lower film on the wafer W. The heat treatment unit U2 of the processing module 11 performs various heat treatments accompanying the formation of the lower film. The heat treatment unit U2, for example, is equipped with a hot plate and a cooling plate, and the wafer W is heated to a predetermined heating temperature using the hot plate, and the heated wafer W is cooled using the cooling plate to perform heat treatment. The inspection unit U3 performs processing for inspecting the state of the surface of the wafer W, for example, acquiring a surface image as information representing the state of the surface of the wafer W.

[0034] The processing module 12 is equipped with a plurality of coating units U1, a plurality of heat treatment units U2, a plurality of inspection units U3, and a conveying device A3 for conveying wafers W to these units. The processing module 12 forms an intermediate film on the lower film by using the coating unit U1 and the heat treatment unit U2. The coating unit U1 of the processing module 12 forms a coating film on the surface of the wafer W by applying a processing liquid for forming the intermediate film on the lower film. The heat treatment unit U2 of the processing module 12 performs various heat treatments accompanying the formation of the intermediate film. The heat treatment unit U2, for example, is equipped with a hot plate and a cooling plate, and the wafer W is heated to a predetermined heating temperature by using the hot plate, and the heated wafer W is cooled by using the cooling plate to perform heat treatment. The inspection unit U3 performs processing for inspecting the state of the surface of the wafer W, for example, obtaining a surface image as information representing the state of the surface of the wafer W.

[0035] The processing module 13 is equipped with a plurality of coating units U1, a plurality of heat treatment units U2, a plurality of inspection units U3, and a conveying device A3 for conveying wafers W to these units. The processing module 13 forms a resist film on the intermediate film through the coating unit U1 and the heat treatment unit U2. The coating unit U1 of the processing module 13, for example, applies a processing liquid for forming a resist film on the intermediate film while rotating the wafer W at a predetermined rotation speed. The heat treatment unit U2 of the processing module 13 performs various heat treatments accompanying the formation of the resist film. The heat treatment unit U2 of the processing module 13 performs a heat treatment (PAB: Post Applied Bake) on the wafer W formed with the coating film at a predetermined heating temperature, thereby forming a resist film. The inspection unit U3 performs a process for inspecting the state of the surface of the wafer W, for example, obtaining a surface image as information representing the state of the surface of the wafer W.

[0036] The processing module 14 has built-in a plurality of developing units, a plurality of heat treatment units U2, a plurality of inspection units U3, and a conveying device A3 for conveying wafers W to these units. The processing module 14 uses the developing unit and the heat treatment unit U2 to perform a development process on the resist film R after exposure. The developing unit of the processing module 14, for example, supplies a developer to the surface of the exposed wafer W while rotating the wafer W at a predetermined rotation speed, and then washes off the developer with a rinse liquid, thereby performing a development process on the resist film R. Such a development process is a process for the resist film R (target film), specifically, a process for removing a portion of the resist film R. The heat treatment unit U2 of the processing module 14 performs various heat treatments accompanying the development process. As specific examples of the heat treatment, a heat treatment before the development process (PEB: Post Exposure Bake), a heat treatment after the development process (PB: Post Bake), etc. can be cited. The inspection unit U3 performs a process for inspecting the state of the surface of the wafer W, for example, acquiring a surface image as information indicating the state of the surface of the wafer W. Such a surface image is a processed image including image information on the surface of the wafer W after the resist film R (target film) has been processed, ie, developed.

[0037] A shelf unit U10 is provided on the side of the carrier block 4 in the processing block 5. The shelf unit U10 is divided into a plurality of small chambers arranged in the vertical direction. A conveying device A7 including a lifting arm is provided near the shelf unit U10. The conveying device A7 lifts and lowers the wafer W between the small chambers of the shelf unit U10.

[0038] A shelf unit U11 is provided on the interface block 6 side in the processing block 5. The shelf unit U11 is divided into a plurality of small chambers arranged in the vertical direction.

[0039] The interface block 6 transfers wafers W between the interface block 6 and the exposure device 3. For example, the interface block 6 has a conveyor device A8 including a transfer arm built therein, which is connected to the exposure device 3. The conveyor device A8 transfers the wafer W arranged in the shelf unit U11 to the exposure device 3, receives the wafer W from the exposure device 3 and returns it to the shelf unit U11.

[0040] [Inspection unit]

[0041] The inspection unit U3 included in the processing modules 11 to 14 is described. The inspection unit U3 included in the processing modules 11 to 13 has a function of acquiring image data (pre-processing image) of the surface of the wafer W before the resist film R is processed (development process), that is, the pre-processing substrate. In addition, the inspection unit U3 included in the processing module 14 has a function of acquiring image data (post-processing image) of the surface of the wafer W after the resist film R is processed (development process).

[0042] like Figure 3 As shown, the inspection unit U3 includes a housing 30, a holding part 31, a linear drive part 32, a photographing part 33, and a light projection and reflection part 34. The holding part 31 holds the wafer W horizontally. The linear drive part 32 uses a motor as a power source, for example, to move the holding part 31 along a horizontal linear path. The photographing part 33 has a camera 35 such as a CCD camera. The camera 35 is arranged at one end side in the inspection unit U3 in the moving direction of the holding part 31, facing the other end side of the moving direction. The light projection and reflection part 34 projects light to the shooting range and guides the reflected light from the shooting range to the camera 35 side. For example, the light projection and reflection part 34 has a half mirror 36 and a light source 37. The half mirror 36 is arranged in the middle of the moving range of the linear drive part 32 at a position higher than the holding part 31, and reflects the light from below to the camera 35 side. The light source 37 is arranged on the half mirror 36, and irradiates the illumination light downward through the half mirror 36.

[0043] The inspection unit U3 operates as follows to obtain image data of the surface of the wafer W. First, the linear drive unit 32 moves the holding unit 31. As a result, the wafer W passes under the half mirror 36. During this passage, the reflected light from each part of the surface of the wafer W is sequentially sent to the camera 35. The camera 35 images the reflected light from each part of the surface of the wafer W and obtains image data of the surface of the wafer W. When the shape (e.g., line width, etc.) of the film formed on the surface of the wafer W changes, for example, the color of the surface of the wafer W changes according to the change in shape, etc., the image data of the surface of the wafer W captured by the camera 35 changes. That is, obtaining the image data of the surface of the wafer W is equivalent to obtaining information about the shape of the film formed on the surface of the wafer W.

[0044] The image data acquired by the camera 35 is sent to the control device 100. The control device 100 can estimate the shape characteristic value of the film on the surface of the wafer W based on the image data, and the estimation result is stored as an inspection result in the control device 100. In addition, the image data itself is also stored in the control device 100.

[0045] The CD of the line width is the minimum width of the pattern of the target film (e.g., resist) after development, and is about several nm to several μm. Here, it is assumed that the imaging unit 33 of the inspection unit U3 for acquiring the pre-processing image and the post-processing image does not have a resolution (resolution) to the extent that the CD of the pattern can be accurately measured from the captured image. In the image captured by the imaging unit 33, the color of the pattern portion varies depending on the CD size. Therefore, the information about the color of the surface acquired for each pixel (here, the grayscale value) can vary depending on the shape of the pattern formed on the target film.

[0046] [Control device]

[0047] An example of the control device 100 is described in detail. The control device 100 controls each element included in the coating and developing device 2. The control device 100 is configured to be able to perform a process including forming each of the above-mentioned films on the surface of the wafer W and performing a developing process. In addition, the control device 100 is configured to be able to perform a process for estimating the shape characteristic value of the formed film as a main part of the shape characteristic value estimation device. Here, a configuration example of the control device 100 in the case of estimating the CD of the target film as the shape characteristic value in the coating and developing device 2 is described.

[0048] like Figure 4 As shown, the control device 100 has, as a functional structure, a pre-processing image acquisition unit 101, a post-processing image acquisition unit 102, an image information storage unit 103, an estimation unit 104, an estimation model storage unit 105, and a shape characteristic value information storage unit 106. Furthermore, the control device 100 has a pre-processing substrate influence model creation unit 107 and an estimation result correction unit 108. Figure 4 Each functional unit shown is a type of shape characteristic value estimating means and is a functional unit for realizing a function as a CD estimating means.

[0049] Before describing each functional unit, an overview of the processing (inspection) performed by the coating and developing device 2 including the control device 100 as a device (shape characteristic value estimation device) for inspecting a substrate is described. In the coating and developing device 2, a process of estimating the CD of a film formed on the surface of the wafer W is performed based on an image obtained by photographing the surface of the wafer W. When the film on the surface of the wafer W is developed, the color of the surface changes depending on the CD of the pattern. Taking advantage of this, in the coating and developing device 2, the CD at each point on the surface of the wafer W is estimated based on the image data including information on the color of the surface of the wafer W.

[0050] As an example, the process of CD estimation is as follows. That is, first, prepare a plurality of wafers whose CDs of films to be estimated (target films) are known. Then, create a CD estimation model (shape characteristic value estimation model) based on the information about the color of each pixel in the image information obtained by photographing the surface of these wafers and the correlation between the CD of the film on the wafer surface at the position where the pixel is photographed. After that, obtain an image of the surface of a wafer that has been developed after forming the target film to be estimated, and estimate the CD of the film on the wafer surface based on the information about the color of each pixel contained in the image data and the CD estimation model. In this way, the CD of the target film on the wafer surface can be estimated.

[0051] Figure 5Schematically shows the change of information about color obtained from image data obtained by photographing a plurality of wafers. Figure 5 In FIG. 1 , information about the color of the surface (here, grayscale value) obtained from image data obtained by photographing the surface of the wafer after the target film is formed is shown. Figure 5 As shown, since each wafer exhibits a different color from another, the CD of the film on the wafer surface is estimated using the difference in color.

[0052] However, in the above method, it may not be possible to create a CD estimation model that takes into account the condition of the lower layer of the film to be estimated CD. As described above, a plurality of films are formed on a wafer. Therefore, when the film to be estimated CD is a resist film, a lower film and an intermediate film are stacked under the resist film. Therefore, Figure 5 The difference in the color of the wafer surface of each wafer as shown may not reflect the change caused by the CD of the film to be estimated by CD, but may reflect the deviation of the state of the resist film and the base portion. In the case where the film thickness of the film formed on the surface is thin as in the wafer used as the object in this embodiment, it is considered that the deviation of the state of the base portion, etc. is reflected in the deviation of the change in the color of the wafer surface.

[0053] The above-mentioned CD estimation model is a model obtained by estimating the correlation between the CD of the resist film and the color information in the image data, but it can be said that it does not take into account the situation where the film thickness of the resist film and the films below it are different. For example, if the film thickness of the intermediate film below the resist film changes, the color of the wafer surface may change depending on the thickness of the intermediate film even in the state before the resist film is applied. However, in the CD estimation model, the influence of such a lower film may not be fully reflected. Considering the above-mentioned problem, it is also a solution to create a CD estimation model using a wafer that changes the condition of the lower film (film thickness, etc.). However, it is considered that it may be difficult to prepare a corresponding number of wafers corresponding to the various conditions required to create a model with high estimation accuracy.

[0054] Therefore, in the coating and developing device 2 described in the present embodiment, a model is created to estimate the influence of the difference of the base portion of the substrate before processing on the CD estimation result, using the image information (pre-processing image) obtained by photographing the surface of the substrate before processing, wherein the substrate before processing is a wafer after the resist film is formed and before the processing of the resist film, that is, the development processing. Specifically, for a wafer after the film to be formed is processed, a model is created based on the CD estimation result using the CD estimation model, the CD value obtained by the method not using the CD estimation model, and the image information of the surface of the wafer before the film to be developed, that is, the substrate before processing. This model (pre-processing substrate influence model) is a model that estimates the influence of the resist film and the base portion on the CD estimation result of the target film obtained by the CD estimation model based on the image data of the substrate before processing. In the coating and developing device 2 described in the present embodiment, the pre-processing substrate influence model is used to remove the CD error caused by the deviation of the film thickness of the base portion, etc., and achieve a CD estimation result with higher calculation accuracy. The control device 100 performs such a high-precision CD estimation process.

[0055] Figure 4 The pre-processing image acquisition unit 101 of the control device 100 shown has a function of acquiring image information (pre-processing image) of the surface of a pre-processing substrate, wherein the pre-processing substrate is a wafer before the processing, i.e., the development processing of the target film to be formed. The pre-processing image acquisition unit 101 acquires the pre-processing image of the target wafer W by, for example, controlling the inspection unit U3 of the processing module 13.

[0056] The processed image acquisition unit 102 has a function of acquiring a processed image including image information about the surface of the wafer W after the target film has been processed (specifically, after the development process). The processed image acquisition unit 102 acquires the processed image of the target wafer W by, for example, controlling the inspection unit U3 of the processing module 14.

[0057] The image information storage unit 103 has a function of storing the pre-processing image acquired by the pre-processing image acquisition unit 101 and the post-processing image acquired by the post-processing image acquisition unit 102. The image information stored in the image information storage unit 103 is used to estimate the CD of the target film formed on the wafer W.

[0058] The estimation unit 104 has a function of estimating the CD of the target film based on the processed image stored in the image information storage unit 103. The CD estimation model is used in the estimation of the CD by the estimation unit 104. The CD estimation model is an estimation model for estimating the CD of the target film, and relates to the correlation between the information on the color of the surface of the wafer W included in the processed image and the CD of the target film formed on the wafer W. The estimation unit 104 estimates the CD of the target film by applying the estimation model to the processed image.

[0059] The estimation model holding unit 105 has a function of holding a CD estimation model used in the estimation of CD by the estimation unit 104. The CD estimation model is a model for calculating CD based on information about the color of each pixel in the image data obtained by photographing the surface of the wafer W after the target film has been processed (more specifically, after the development process), and is a model that represents the correspondence between the information about the color and the CD. By using such a model to obtain information about the color of the surface of the wafer W at each position of the image data, CD can be estimated based on the information about the color. The method for creating the CD estimation model is not particularly limited, and it can be created using a known statistical processing method or the like. Based on the information about the color of the surface of the substrate before processing contained in the pre-processing image acquired by the pre-processing image acquisition unit 101 and the information about the color of the surface of the substrate after processing contained in the post-processing image acquired by the post-processing image acquisition unit 102, information related to the change in color of the substrate surface caused by the formation of the film is acquired. On this basis, the CD estimation model can be created by calculating the correlation between the change in color and CD. However, the above method is an example and is not limited to the above example. For example, instead of using information about the color of the surface of the substrate before processing, only information about the color of the surface of the substrate after processing can be used to calculate the correlation between the color change and the CD, thereby creating a model. In addition, the CD estimation model can be created by the coating and developing device 2, or by other devices, for example.

[0060] The shape characteristic value information holding unit 106 has a function of holding information about the CD of the target film corresponding to the shape characteristic value about the target film obtained without using the CD estimation model. As the CD of the target film obtained without using the CD estimation model, for example, a measured value of the CD of the target film measured by a certain method can be used. In addition, when the operation of the coating and developing device 2 is stable, for example, a film is formed on the inspection substrate during a detailed inspection (QC inspection) related to the operation confirmation of the device, and the inspection result at this time can also be used as the CD. In addition, as long as the characteristics of each of the multiple units (coating unit, heat treatment unit, etc.) involved in film formation included in the coating and developing device 2 can be grasped to a certain extent, it can also be configured to obtain CD based on which unit is processed. In addition, if the gradual change of CD after film formation is grasped in advance based on the characteristics of the device, the value obtained by adding the measured value measured regularly to the predicted value of the change over time can also be used as the CD. Such a CD is a value with a certain degree of reliability, and it can be a value obtained (or calculated) without using the CD estimation model, and can be obtained using various methods. The CD stored in the shape characteristic value information storage unit 106 may be configured to be acquired while performing the process related to the film formation on the wafer W, as in the case of actual measurement by a certain method, or may be configured to be acquired (calculated) in advance when a predicted value or the like is used as the CD. In addition, CD may be set to one value for each wafer W, for example, multiple values ​​may be defined for one wafer W (for example, for each test structure (shot) or bare chip).

[0061] The pre-processed substrate influence model creation unit 107 has the following function: based on the estimation result of CD, the CD of the target film obtained without using the CD estimation model, and the pre-processed image which is the image of the wafer W before the target film is processed, that is, the pre-processed substrate. In detail, the pre-processed substrate influence model creation unit 107 creates the pre-processed substrate influence model, which is a model about the correlation between the difference between the estimation result and the CD of the target film obtained without using the CD estimation model and the pre-processed image (information about the color of the pre-processed substrate). The pre-processed substrate influence model is created by accumulating the estimation results of the CD estimation model to some extent. The details of the pre-processed substrate influence model creation unit 107 will be described later.

[0062] The estimation result correction unit 108 has a function of correcting the estimation result of the CD of the target film obtained by the estimation unit 104 based on the pre-processing substrate influence model created by the pre-processing substrate influence model creation unit 107. The pre-processing substrate influence model can be used to estimate whether the resist film and the base portion, etc. have an influence on the CD estimation result. Therefore, by removing the influence component of the resist film and the base portion, etc. from the CD estimated by the CD estimation model, a more accurate CD estimation result can be obtained.

[0063] The control device 100 is composed of one or more control computers. For example, the control device 100 has Figure 6 The circuit 120 shown. The circuit 120 has one or more processors 121, a memory 122, a storage 123 and an input / output port 124. The storage 123 has, for example, a storage medium that can be read by a computer, such as a hard disk. The storage medium stores a program for causing the control device 100 to execute the substrate inspection process described later. The storage medium can also be a removable medium such as a non-volatile semiconductor memory, a magnetic disk, and an optical disk. The memory 122 temporarily stores the program loaded from the storage medium of the storage 123 and the calculation results of the processor 121. The processor 121 executes the above-mentioned program by cooperating with the memory 122, thereby constituting the above-mentioned functional modules. The input / output port 124 inputs and outputs electrical signals between the components that are the control objects according to the instructions from the processor 121.

[0064] In addition, the hardware structure of the control device 100 is not necessarily limited to each functional module being configured by a program. For example, each functional module of the control device 100 may include a dedicated logic circuit or an application specific integrated circuit (ASIC) obtained by integrating the dedicated logic circuit.

[0065] In the following embodiments, the control device 100 includes the above-mentioned structure, but the control device 100 may not include all the above-mentioned functions. For example, the image information storage unit 103, the estimated model storage unit 105, the shape characteristic value information storage unit 106, etc. may be provided in an external device as a functional unit of the database.

[0066] [Processing flow]

[0067] Next, as an example of coating and developing processing, a process flow executed in the coating and developing device 2 will be described.

[0068] In the process flow, first, the control device 100 controls the conveying device A1 to convey the wafer W to be processed in the carrier C to the shelf unit U10 , and controls the conveying device A7 to place the wafer W in the chamber for the process module 11 .

[0069] Next, the control device 100 controls the conveying device A3 to convey the wafer W in the shelf unit U10 to the coating unit U1 and the heat treatment unit U2 in the processing module 11. In addition, the control device 100 controls the coating unit U1 and the heat treatment unit U2 to form a lower film on the surface of the wafer W. After that, the control device 100 controls the conveying device A3 to return the wafer W with the lower film formed thereon to the shelf unit U10, and controls the conveying device A7 to place the wafer W in the small chamber for the processing module 12.

[0070] Next, the control device 100 controls the conveying device A3 to convey the wafer W of the shelf unit U10 to the coating unit U1 and the heat treatment unit U2 in the processing module 12. In addition, the control device 100 controls the coating unit U1 and the heat treatment unit U2 to form an intermediate film on the lower film of the wafer W. For example, the control device 100 controls the coating unit U1 so that the intermediate film is formed by coating the lower film of the wafer W with a processing liquid for forming the intermediate film. Next, the control device 100 controls the heat treatment unit U2 to perform heat treatment on the intermediate film. After that, the control device 100 controls the conveying device A3 to return the wafer W to the shelf unit U10, and controls the conveying device A7 to arrange the wafer W in the small chamber for the processing module 13.

[0071] Next, the control device 100 controls the conveying device A3 to convey the wafer W of the shelf unit U10 to each unit in the processing module 13, and controls the coating unit U1 and the heat treatment unit U2 to form a resist film on the intermediate film of the wafer W. For example, the control device 100 controls the coating unit U1 so that the resist film is formed by coating the intermediate film of the wafer W with a processing liquid for forming the resist film. Next, the control device 100 controls the heat treatment unit U2 to perform heat treatment on the resist film. After the resist film is formed, the control device 100 controls the conveying device A3 to convey the wafer W to the inspection unit U3, and controls so that the inspection unit U3 is used to photograph the surface of the wafer W and obtain image information (pre-processing image). After that, the control device 100 controls the conveying device A3 to convey the wafer W to the shelf unit U11.

[0072] Next, the control device 100 controls the conveyor A8 to convey the wafer W in the shelf unit U11 to the exposure device 3. Thereafter, the control device 100 controls the conveyor A8 to receive the wafer W subjected to the exposure process from the exposure device 3 and place it in the chamber for the processing module 14 in the shelf unit U11.

[0073] Next, the control device 100 controls the conveying device A3 to convey the wafer W of the shelf unit U11 to each unit in the processing module 14, and controls the developing unit and the heat treatment unit U2 to perform a developing process on the resist film R of the wafer W. After the developing process, the control device 100 controls the conveying device A3 to convey the wafer W to the inspection unit U3, and controls so that the surface of the wafer W is photographed using the inspection unit U3 to obtain image information (processed image). After that, the control device 100 controls the conveying device A3 to return the wafer W to the shelf unit U10, and controls the conveying device A7 and the conveying device A1 to return the wafer W to the carrier C. Through the above, the process is completed.

[0074] [CD calculation method]

[0075] Next, refer to Figure 7 to Figure 10 , a CD estimation method in the processing modules 11 to 14 performed by the control device 100 is described. The CD estimation method is a method for inspecting the wafer W after film formation performed in the inspection unit U3 provided in the processing modules 11 to 14. In the inspection unit U3, by estimating the CD, it is evaluated whether the desired film formation is performed on the wafer W after film formation, and in particular, whether the film formation of the desired CD is performed.

[0076] In addition, in the following embodiments, the case of estimating the CD of the resist film is described. Therefore, the object process is a development process, and the wafer W before the development process is implemented (that is, the wafer formed to the resist film) is described as the substrate before the process. However, the object film is not limited to the resist film. For example, the object film can also be used as an intermediate film to estimate the film thickness of the intermediate film. In this case, the substrate before the process becomes the wafer before the intermediate film is formed, that is, the wafer on which the base film is formed. In this way, the setting of the substrate before the process can be changed according to the difference of the object film that becomes the object of CD estimation. In addition, the substrate before the process can also be a substrate that has been processed (such as film forming, etching, cleaning, etc.) using other devices. In this way, the substrate before the process can be any substrate before the object process, without any special limitation.

[0077] Figure 7 1 is a flowchart for explaining the steps of the CD estimation method in a series of processes until the CD estimation using the CD estimation model. First, the control device 100 executes step S01. In step S01, a base substrate is prepared and sent to the inspection unit U3. The base substrate here refers to the wafer W that has been film-formed to the intermediate film in the coating unit U1 and the heat treatment unit U2. The sent base substrate is held in the holding portion 31.

[0078] Next, the control device 100 executes step S02. In step S02, a resist film is formed as an object film on the base substrate. Next, the pre-processing image acquisition unit 101 of the control device 100 executes step S03. In step S03, the surface of the substrate before processing is photographed by the photographing unit 33. Specifically, while the holding unit 31 is moved in a prescribed direction by the driving of the linear driving unit 32, the surface of the substrate before processing is photographed by the photographing unit 33. Thus, image information (pre-processing image) of the surface of the substrate before processing is acquired in the photographing unit 33. The pre-processing image is held in the image information holding unit 103 of the control device 100.

[0079] Next, the control device 100 executes step S04. In step S04, a part of the target film is removed by development. The wafer W that has been subjected to development is sent to the inspection unit U3. The sent wafer W that has been subjected to development is held by the holding portion 31.

[0080] Next, the processed image acquisition unit 102 of the control device 100 executes step S05. In step S05, the surface of the processed wafer W is imaged by the imaging unit 33 in the same manner as step S02. Specifically, the surface of the wafer W is imaged by the imaging unit 33 while the holding unit 31 is moved in a predetermined direction by the driving of the linear driving unit 32. Thus, image information (processed image) of the surface of the wafer W is acquired in the imaging unit 33. The processed image is stored in the image information storage unit 103 of the control device 100.

[0081] Next, the estimation unit 104 of the control device 100 executes step S06. In step S065, the CD of the target film on the surface of the wafer W is estimated based on the processed image of the surface of the wafer W. In the estimation of the CD by the estimation unit 104, the CD estimation model held in the estimation model holding unit 105 is used.

[0082] The specific method for estimating the CD of the processed image is as follows. First, information about the color of the surface of the chip W is obtained for each pixel from the processed image. At this time, the process of calculating the correlation with the base image before film formation can also be performed. Whether to perform this process can be appropriately changed according to the condition of the CD estimation model. After that, it is compared with the CD estimation model maintained by the estimation model holding unit 105. Thus, the CD of the area photographed by the pixel can be estimated for each pixel. Thus, the CD of the target film at each position on the surface of the chip W can be estimated.

[0083] Next, in the CD estimation method described in this embodiment, Figure 8 Modification of the substrate impact model before treatment is shown.

[0084] like Figure 8As shown, the pre-processing substrate influence model creation unit 107 of the control device 100 sequentially executes steps S11 to S13 related to the creation of the pre-processing substrate influence model. In step S11, first, the CD estimation result y of the target film obtained by the CD estimation model (F) in a specific pixel and the CD information Y of the target film obtained without using the CD estimation model are obtained. In step S12, a pre-processing image of the wafer W from which the CD is estimated is prepared. Then, in step S13, a pre-processing substrate influence model (G) is created regarding the correlation between the difference between the CD estimation result y and the CD information Y and the color information included in the pre-processing image. Then, after the pre-processing substrate influence model (G) is created, the estimation result correction unit 108 of the control device 100 executes step S14. In step S14, the pre-processing substrate influence element y' calculated based on the pre-processing substrate influence model (G) is subtracted from the CD estimation result y of the target film, thereby correcting the CD estimation result. These steps are described below.

[0085] The CD estimation result of the CD estimation model in the processed image is set to y, and the CD information of the target film obtained without using the CD estimation model is set to Y. In addition, the CD estimation model created based on the image after the development process (processed image) is set to F. The CD estimation result y is calculated based on the CD estimation model, so it should satisfy the relationship of the following mathematical formula (1).

[0086] y=F(processed image)…(1)

[0087] Here, when the accuracy of the CD estimation model is sufficiently high, the CD estimation result y should be consistent with the CD information Y. However, in reality, the state of the following equation (2) is achieved.

[0088] Y=F(processed image)+error1…(2)

[0089] exist Fig. 9 In the formula (2), it is schematically shown that the CD estimation result may contain the error "error1". Fig. 9 In the example, for each wafer W to be the target, the relationship between the CD estimation result y and the CD information Y in a specific pixel is plotted. When the CD estimation result y and the CD information Y are consistent, the points should be arranged in Fig. 9However, in reality, the CD estimation result y does not match the CD information Y. There are groups T1 where the CD estimation result y is larger than the CD information Y and groups T2 where the CD estimation result y is smaller than the CD information Y. In "error1" in the mathematical formula (2), in addition to simple errors, errors caused by insufficient accuracy of the CD estimation model F may also be included. One of the reasons for the reduction in accuracy of the CD estimation model F is that the variation of the substrate before the target processing cannot be fully removed as described above.

[0090] Therefore, as the pre-processing substrate influence model, the correlation between the color information obtained from the pre-processing image and the portion corresponding to "error 1" is grasped. If the pre-processing substrate influence model is set to G, the pre-processing substrate influence model is a model that satisfies the following equation (3).

[0091] yY=G(image before processing)+error2…(3)

[0092] As described above, "error1" in the mathematical formula (2) is considered to include the variation amount derived from the substrate before processing, but as described above, the component derived from the substrate before processing is treated as the substrate before processing influence model G. As a result, in "error2" in the mathematical formula (3), the component derived from the substrate before processing is removed, and only the so-called pure error is included.

[0093] exist Fig.10 , the pre-processing substrate influence model G is schematically shown. The pre-processing substrate influence model G is, for example, a model that represents the correlation between the information about the color of the pre-processing substrate surface in the pre-processing image (for example, the brightness value in the pre-processing image) and yY. Fig.10 As shown in FIG. 1 , by modeling the correlation between information on the color of the substrate surface before processing (e.g., brightness value in the pre-processing image) and yY, it is possible to apply a correction corresponding to the color of the substrate surface before processing. That is, the variation amount (corresponding to yY) that is considered to be included in the estimated value of CD estimated using the CD estimation model F and is derived from the substrate before processing can be grasped based on the pre-processing substrate influence model G.

[0094] If the pre-processing substrate influence factor obtained based on the pre-processing substrate influence model in a specific pixel is set to y', then y' satisfies the relationship of the following equation (4).

[0095] y'=G(image before processing)…(4)

[0096] When the CD influencing factor y' is subtracted from the CD estimation result y estimated using the CD estimation model F, as shown in formula (5), the CD estimation result y can be made closer to the CD information Y, that is, closer to a more accurate value. As shown in formula (5), this y-y' includes error2, which is a simple error. However, as described above, error2 does not include the model error, so in y-y', the model error that may be included in the estimation result when only the CD estimation model F is used can be removed.

[0097] y-y'=y-(yY)+error2=Y-error2...(5)

[0098] In addition, Fig.10 In FIG. 1 , the state where the substrate influence model G before processing becomes a linear function of the brightness value of the image before processing is shown, but this is a simpler example for the purpose of explanation. Fig.10 As shown in FIG. 1 , the pre-processing substrate influence model G can be obtained as a linear function using only one value as information about the color of the pre-processing image, but a more complex statistical method can also be used to achieve higher-precision modeling. That is, as information about the color of the pre-processing image when creating the pre-processing substrate influence model G, not only the brightness value but also multiple parameters such as RGB values ​​can be used, and a statistical method using these feature quantities can be used to create the model. The pre-processing substrate influence model G can also be created using a known statistical method such as regression analysis in addition to multiple regression analysis, nonlinear analysis, and decision tree analysis.

[0099] The creation of the pre-processing substrate influence model G and the correction of the CD estimation result using the pre-processing substrate influence model G described above ( Figure 8 The series of processes shown in the figure can be performed at a stage where the CD estimation results of the target film using the CD estimation model F have been accumulated to a certain extent. This is because it is impossible to grasp the extent to which the CD estimation model F includes the variation amount derived from the substrate before processing (i.e., the component equivalent to the substrate influence model G before processing) at the stage of preparing the CD estimation model F. Therefore, after CD estimation is performed on a plurality of wafers W using the CD estimation model F, the creation of the substrate influence model G before processing and the correction of the CD estimation results using the substrate influence model G before processing are performed. In order to improve the accuracy of the substrate influence model G before processing to a certain extent, it is considered that the number of data used to create the substrate influence model G before processing can be about several dozen, but it is not limited to this number.

[0100] In addition, the pre-processed substrate influence model G may be corrected at predetermined intervals. For example, when the number of wafers W processed in the coating and developing apparatus 2 exceeds a predetermined number, the pre-processed substrate influence model G may be re-created using information related to the most recent dozens of wafers W, and the information may be reflected in the subsequent estimation of the CD of the wafer W. In this way, the timing of creating the pre-processed substrate influence model G and its changes may be appropriately changed in consideration of the operating conditions of the coating and developing apparatus 2, the characteristics of the wafer W, and the like.

[0101] Furthermore, the data used when creating the pre-processed substrate influence model G is not limited to one data for each wafer W. Since the CD estimation model F is used to estimate the CD for each test structure, the pre-processed substrate influence model G may be created by taking in the estimation result for each pixel.

[0102] [Details of the influence of the substrate on model creation before processing]

[0103] Next, the details of the creation of the pre-processing substrate influence model by the pre-processing substrate influence model creation unit 107 will be described. The method of creating the pre-processing substrate influence model by the pre-processing substrate influence model creation unit 107 is not limited to the method described below.

[0104] The pre-processed substrate influence model creation unit 107 performs a process for suppressing overfitting only when the number of post-processed images (i.e., the number of wafers that become samples) is greater than a specified value, and then creates a pre-processed substrate influence model. Overfitting refers to the state of a model that can fit a learning sample but cannot fit unknown data. Fig.11 and Fig.12 , the creation process of the pre-processing substrate influence model while suppressing over-matching is explained.

[0105] Fig.11 is a flow chart showing the process of creating a pre-processing substrate impact model. Fig.11 As shown, the pre-processing substrate influence model creation unit 107 first determines whether the number of samples S as the processed images is greater than a specified value (here, 75) (step S31). If the number of samples S is not greater than the specified value in step S31, the set division described later is not performed, and the pre-processing substrate influence model creation unit 107 creates a pre-processing substrate influence model based on all samples S (step S32).

[0106] On the other hand, when the number of samples S is greater than a predetermined value, Fig.12 As shown in (a) of FIG. 1 , the samples S are arranged in time series (step S13). Then, as Fig.12As shown in (b), among the total number of N samples S, for example, N / 3 new samples S in time series are taken as a set (third set) of verification samples S1 (step S34).

[0107] Then, if Fig.12 As shown in (c) of FIG. 1 , the pre-processing substrate influence model creation unit 107 randomly divides N / 3 samples S2 from the remaining samples S into the first set (step S35). Fig.12 As shown in (d), the pre-processing substrate influence model creation unit 107 divides the remaining N / 3 samples S3 into a second set (step S36).

[0108] Next, the pre-processing substrate influence model creation unit 107 implements learning of the first set (step S37), and creates a first candidate model formula CD as a candidate of the pre-processing substrate influence model. 1 (1st candidate). 1st candidate model type CD 1 As shown in the following formula (6).

[0109] CD 1 =CD 0 +FUNC1(Gray)…(6)

[0110] In the above formula (6), CD 0 is the estimated value of CD estimated by the CD estimation model, Gray is the grayscale value of the image before processing, and FUNC1 is a function representing the correction amount obtained by learning the substrate before processing.

[0111] Then, the estimation result correction unit 108 corrects the first candidate model formula CD 1 , derives verification data after correcting the estimated value of CD of verification sample S1 (step S38). Specifically, the estimation result correction unit 108 derives CD obtained using the first candidate model formula by the following formula (7): 1 Verification data of sample S1 is used for verification.

[0112] CD 1verify =CD 0verify +FUNC1(Gray verify )…(7)

[0113] In addition, based on the first candidate model CD 1 , as shown in the following formula (8), the CD data (learning data) of the sample S3 of the second set is derived. group2 This represents the second set. Such data for learning is CD data obtained by learning in the first generation (first set).

[0114] CD 1group2 =CD0group2 +FUNC1(Gray group2 )…(8)

[0115] Next, the pre-processing substrate influence model creation unit 107 implements learning of the second set (step S39), and creates a second candidate model formula CD as a candidate of the pre-processing substrate influence model. 2 (2nd candidate). 2nd candidate model type CD 2 As shown in the following formula (9).

[0116] CD 2 =CD 1 +FUNC2(Gray)…(9)

[0117] In the above formula (9), CD 1 FUNC2 is the CD data after the first substrate learning correction, and FUNC3 is a function representing the correction amount obtained by using the second substrate learning before processing.

[0118] Then, the estimation result correction unit 108 corrects the second candidate model formula CD 2 , derives the verification data after correcting the estimated value of CD of the verification sample S1 (step S40). Specifically, the estimation result correction unit 108 derives the CD using the second candidate model formula by the following formula (10): 2 The verification data of the verification sample S1 is obtained.

[0119] CD 2verify =CD 1verify +FUNC2(Gray verify )…(10)

[0120] Next, it is determined whether the above-mentioned processing of S35 to S40 has been performed a predetermined number of times (step S41), and the processing of steps S35 to S40 is repeated until the predetermined number of times. In this way, the pre-processing substrate influence model creation unit 107 repeatedly performs the processing of dividing the plurality of samples S into the first set, the second set, and the set (third set) of the verification sample S1 while changing the combination of the samples S included in the first set and the second set.

[0121] Then, when the processing of S35 to S40 is performed for a predetermined number of times, the pre-processing substrate influence model creation unit 107 compares all the verification data based on the model formula for each first set and the verification data based on the model formula for each second set. The pre-processing substrate influence model creation unit 107, for example, determines the candidate model formula with the smallest value of the change value of the verification data (for example, 3σ of the difference from the CD value obtained without using the estimated model) as the best model as the pre-processing substrate influence model (step S42). In addition, when there is an already created pre-processing substrate influence model, the pre-processing substrate influence model creation unit 107 may compare the change value of the new pre-processing substrate influence model (for example, 3σ of the difference from the CD value obtained without using the estimated model) with the change value of the already created pre-processing substrate influence model. Moreover, the new pre-processing substrate influence model may be adopted only when it is confirmed that the change value has improved. When the change factor of the sample is small, there is a risk that the learning model of the generational progress becomes over-learned. In this case, the change value of the verification sample becomes large, so the past (old) learning model is selected. Thereby, the number of factors can be safely increased.

[0122] [effect]

[0123] The shape characteristic value estimation device (CD estimation device) of the present embodiment includes a processed image acquisition unit 102, which acquires a processed image including image information about the surface of a wafer W after processing the target film. The shape characteristic value estimation device includes an estimation unit 104. The estimation unit 104 estimates the CD of the target film by applying a CD estimation model to the processed image, the CD estimation model involving the correlation between the information about the color of the surface of the wafer W included in the processed image and the CD of the target film formed on the wafer W. The shape characteristic value estimation device includes a pre-processed substrate influence model creation unit 107. The pre-processed substrate influence model creation unit 107 creates a pre-processed substrate influence model, which is a model of the correlation between the difference between the estimation result of the CD of the target film estimated by the estimation unit 104 and the CD acquired without using the CD estimation model, and the information about the color of the pre-processed substrate, wherein the pre-processed substrate is the wafer W before the target film is processed. The shape characteristic value estimation device includes an estimation result correction unit 108 that corrects the estimation result of the CD of the object film estimated by the estimation unit 104 based on the pre-processing substrate influence model. The pre-processing substrate influence model creation unit 107 is configured to be able to perform a process of dividing a plurality of processed images into a first set and a second set. The pre-processing substrate influence model creation unit 107 is configured to be able to perform the following processes: creating a first candidate as a candidate of the pre-processing substrate influence model based on the first set, and creating a second candidate as a candidate of the pre-processing substrate influence model based on the second set. The pre-processing substrate influence model creation unit 107 compares the result of the estimation result correction unit 108 correcting the estimation result based on the first candidate and the result of the estimation result correction unit 108 correcting the estimation result based on the second candidate. Thus, the pre-processing substrate influence model is determined.

[0124] In the shape characteristic value estimation device of the present embodiment, first, the CD of the target film is estimated using the CD estimation model. Next, a pre-processed substrate influence model is created, which is a model of the correlation between the difference between the estimated result of the CD of the target film and the CD of the target film obtained without using the CD estimation model, and the information about the color of the pre-processed substrate contained in the pre-processed image. Then, based on the pre-processed substrate influence model, the estimation result is corrected. By adopting such a configuration, the variable component of the estimation result caused by the color of the pre-processed substrate, etc. can be corrected based on the pre-processed substrate influence model. Therefore, the characteristic value (for example, CD) of the shape of the target film formed on the wafer W can be estimated with higher accuracy. In addition, in the shape characteristic value estimation device of the present embodiment, the processed image is divided into a first set and a second set, and candidates for the pre-processed substrate influence model are created for each set. Then, the estimation result of the shape characteristic value estimated by the estimation unit 104 is corrected for each candidate of the pre-processed substrate influence model, and the correction result of each candidate is compared, thereby selecting (determining) an appropriate pre-processed substrate influence model. When creating a pre-processing substrate influence model based on a plurality of processed images, if there are too many samples for learning, overmatching may sometimes occur (a state in which the pre-processing substrate influence model can be matched with the samples for learning but cannot be matched with unknown data). In this case, the correction accuracy of the pre-processing substrate influence model is reduced. In this regard, a plurality of processed images are grouped, and candidates for the pre-processing substrate influence model are created by learning according to each group, thereby reducing the number of factors learned at one time and suppressing overfitting. Then, by comparing the correction results of each candidate of the pre-processing substrate influence model to select (determine) an appropriate pre-processing substrate influence model, a pre-processing substrate influence model with high correction accuracy can be created. By using the pre-processing substrate influence model created in this way to correct the inference result, the characteristic value of the shape of the object film formed on the wafer W can be inferred with high accuracy.

[0125] The pre-processing substrate influence model creation unit 107 divides the plurality of processed images into a first set, a second set, and a third set. The pre-processing substrate influence model creation unit 107 compares the result of the estimation result correction unit 108 correcting the estimation result for the third set based on the first candidate with the result of the estimation result correction unit 108 correcting the estimation result for the third set based on the second candidate. Thus, the pre-processing substrate influence model is determined. In this way, by preparing learning samples and verification samples based on a plurality of processed images, the number of factors for one learning can be reduced, and verification samples can be prepared efficiently and appropriately.

[0126] The pre-processing substrate influence model creation unit 107 repeatedly performs the process of dividing into the first set, the second set, and the third set for a plurality of processed images while changing the combination of processed images included in the first set and the second set. Then, the pre-processing substrate influence model creation unit 107 determines the pre-processing substrate influence model by comparing the results after correction of the estimated results for each first set with the results after correction of the estimated results for each second set. In this way, while changing the samples included in each set, the first candidate and the second candidate of a plurality of patterns are created respectively, and the correction results based on each candidate are compared with each other, thereby being able to select a more appropriate pre-processing substrate influence model based on the plurality of correction results. That is, it is possible to select a pre-processing substrate influence model that is corrected with higher accuracy while increasing the number of input factors.

[0127] The pre-processed substrate influence model creation unit 107 may also perform the process of dividing a plurality of processed images into the first set and the above-mentioned second set only when the number of processed images is greater than a specified value. The pre-processed substrate influence model creation unit 107 may also create the pre-processed substrate influence model without performing the process of dividing into the first set and the second set when the number of processed images is less than a specified value. If the above-mentioned grouping is performed when the number of processed images is small, it is possible that under-fitting of factors that cannot be modeled is generated. In this regard, the above-mentioned grouping is performed only when the number of processed images is greater than a specified value to perform learning, and when the number of processed images is less than the specified value, learning is performed without grouping, thereby suppressing both over-fitting and under-fitting. In this way, a pre-processed substrate influence model with high correction accuracy can be created, and the characteristic values ​​of the shape of the target film formed on the wafer W can be estimated with high accuracy.

[0128] Various exemplary embodiments have been described above, but the present invention is not limited to the exemplary embodiments described above, and various omissions, substitutions, and changes may be made.

[0129] For example, in the above-mentioned embodiment, the case where the "shape characteristic value" is a limit dimension (CD) such as line width is described, but the "shape characteristic value" may also be the film thickness of the target film. In this case, the processing of the target film means coating the target film on the substrate, and the processed image means the image of the surface of the substrate coated with the target film. That is, in the case of CD estimation, the processed image means the image of the state in which a part of the target film is removed after the development processing, but in the case of film thickness estimation, the processed image means the image of the state in which the target film is not removed before the development processing.

Claims

1. A shape characteristic value estimation device, characterized in that: include: a processed image acquisition unit that acquires a processed image including image information on the surface of the substrate after the target film has been processed; an estimating unit that estimates the shape characteristic value of the target film by applying an estimating model for estimating the shape characteristic value of the target film to the processed image, wherein the estimating model relates to a correlation between information on the color of the substrate included in the processed image and a shape characteristic value that is a characteristic value of the shape of the target film formed on the substrate; a pre-processed substrate influence model creation unit that creates a pre-processed substrate influence model, the pre-processed substrate influence model being a model of correlation between a difference between an estimation result of the shape characteristic value of the target film obtained by the estimation unit and the shape characteristic value of the target film obtained without using the estimation model, and information on the color of the pre-processed substrate, wherein the pre-processed substrate is a substrate before the target film is processed; and an estimated result correction unit for correcting the estimated result of the shape characteristic value of the target film obtained by the estimation unit based on the pre-processed substrate influence model, The pre-processing substrate influence model creation unit is configured to perform the following processing: The plurality of processed images are divided into a first set and a second set; creating a first candidate as a candidate of the pre-processing substrate influence model based on the first set, and creating a second candidate as a candidate of the pre-processing substrate influence model based on the second set; and Processing of the pre-processing substrate influence model is determined by comparing a result obtained by the estimation result correction unit of correcting the estimation result based on the first candidate with a result obtained by the estimation result correction unit of correcting the estimation result based on the second candidate.

2. The shape characteristic value estimation device according to claim 1, characterized in that: The processing of the pre-substrate influences the model creation unit: dividing the plurality of processed images into the first set, the second set and the third set, The pre-processing substrate influence model is determined by comparing a result obtained by the estimation result correction unit correcting the estimation result regarding the third set based on the first candidate with a result obtained by the estimation result correction unit correcting the estimation result regarding the third set based on the second candidate.

3. The shape characteristic value estimation device according to claim 2, characterized in that: The processing of the pre-substrate influences the model creation unit: while changing the combination of the processed images included in the first set and the second set, repeatedly performing a process of dividing the plurality of processed images into the first set, the second set, and the third set a plurality of times; The pre-processing substrate influence model is determined by comparing the results of the estimation result correction unit correcting the estimation results of the third set based on the first candidate for each of the first sets with the results of the estimation result correction unit correcting the estimation results of the third set based on the second candidate for each of the second sets.

4. The shape characteristic value estimation device according to any one of claims 1 to 3, characterized in that: The processing front substrate affects the model creation unit: Only when the number of the processed images is equal to or greater than a predetermined value, performing processing of dividing the plurality of processed images into the first set and the second set, When the number of the processed images is less than the predetermined value, the pre-processed substrate influence model is created without performing the process of dividing into the first set and the second set.

5. A shape characteristic value estimating method for estimating a shape characteristic value, the shape characteristic value being a characteristic value of a shape of a target film formed on a substrate, characterized in that: include: a step of acquiring a processed image including image information about the surface of the substrate after the object film has been processed; The step of estimating the shape characteristic value of the target film by applying an estimation model for estimating the shape characteristic value of the target film to the processed image, wherein the estimation model relates to a correlation between information on the color of the substrate contained in the processed image and the shape characteristic value of the target film formed on the substrate; a step of creating a pre-processed substrate influence model, the pre-processed substrate influence model being a model of correlation between a difference between an estimated result of the shape characteristic value of the target film and the shape characteristic value of the target film obtained without using the estimated result, and information on the color of the pre-processed substrate, wherein the pre-processed substrate is a substrate before the target film is processed; and The step of correcting the estimated result of the shape characteristic value of the target film based on the pre-processing substrate influence model, In the creation of the pre-processing substrate impact model the following steps are performed: The step of dividing the plurality of processed images into a first set and a second set; The steps of creating a first candidate as a candidate of the pre-processing substrate influence model based on the first set, and creating a second candidate as a candidate of the pre-processing substrate influence model based on the second set; and The step of determining the pre-processing substrate influence model by comparing a result obtained by correcting the estimated result based on the first candidate and a result obtained by correcting the estimated result based on the second candidate.

6. The shape characteristic value estimation method according to claim 5, characterized in that: dividing the plurality of processed images into the first set, the second set and the third set, The pre-processing substrate influence model is determined by comparing a result obtained by correcting the estimated result regarding the third set based on the first candidate and a result obtained by correcting the estimated result regarding the third set based on the second candidate.

7. The shape characteristic value estimation method according to claim 6, characterized in that: while changing the combination of the processed images included in the first set and the second set, repeatedly performing a process of dividing the plurality of processed images into the first set, the second set, and the third set a plurality of times; The pre-processing substrate influence model is determined by comparing the results of correcting the estimated results of the third set based on the first candidate for each of the first sets with the results of correcting the estimated results of the third set based on the second candidate for each of the second sets.

8. The shape characteristic value estimation method according to any one of claims 5 to 7, characterized in that: Only when the number of the processed images is equal to or greater than a predetermined value, performing processing of dividing the plurality of processed images into the first set and the second set, When the number of the processed images is less than the predetermined value, the pre-processed substrate influence model is created without performing the process of dividing into the first set and the second set.

9. A computer-readable storage medium, characterized in that: A program for causing the device to execute the shape characteristic value estimating method according to any one of claims 5 to 7 is stored.