JP2022149848A - Measurement device, lithographic apparatus, and article manufacturing method - Google Patents

Abstract

To provide a technology advantageous for highly precisely measuring positional information of a measurement object.SOLUTION: A measurement device measuring positional information of a measurement object is provided. The measurement device includes a scope which captures an image of the measurement object and generates image data, and a processing unit which obtains a characteristic amount of the image data, and determines an optical parameter which is to be set in the scope for the measurement based on the characteristic amount. The processing unit may obtain an evaluation value of the characteristic amount of the image data about each of a plurality of optical parameter candidates and may determine the optical parameters based on each evaluation value.SELECTED DRAWING: Figure 24

Description

The present invention relates to a metrology apparatus, a lithographic apparatus and a method of manufacturing an article.

A lithographic apparatus such as an imprint apparatus and an exposure apparatus may be used in a lithographic process for manufacturing articles such as semiconductor devices. A lithographic apparatus may transfer a pattern of an original onto a shot area of a substrate. In an imprinting apparatus, a mold is brought into contact with an imprinting material placed on a shot region of a substrate, and the imprinting material is cured to form a pattern of a cured imprinting material on the shot region. It is formed. In the exposure device, a latent image of the pattern of the original is formed on the photosensitive material by projecting the pattern of the original onto the shot area of the substrate coated with the photosensitive material. The latent image is transformed into a physical pattern by a development process. In such a lithography apparatus, a technique for measuring the relative positions of the marks on the substrate and the marks on the original with high accuracy is required in order to align the shot area of the substrate and the original with high accuracy.

Patent Literature 1 discloses a method of estimating an alignment correction value using machine learning.

Japanese Patent No. 4601492

It has been found that alignment accuracy depends on the optical parameters used in the alignment metrology. Therefore, it can be said that there is room for improving the alignment accuracy by optimizing the optical parameters according to the pattern formation target.

SUMMARY OF THE INVENTION An object of the present invention is to provide an advantageous technique for highly accurate measurement of position information of a measurement object.

According to one aspect of the present invention, there is provided a measuring device for measuring position information of an object to be measured, comprising: a scope for capturing an image of the object to be measured to generate image data; and a processing unit that determines optical parameters to be set in the scope for the measurement based on the feature amount.

According to the present invention, it is possible to provide an advantageous technique for highly accurate measurement of the position information of the object to be measured.

4A and 4B are diagrams illustrating the operation of the imprint apparatus; FIG. FIG. 2 is a diagram exemplifying the configuration of an imprint apparatus; The figure which illustrates the structure of an article manufacturing system. FIG. 4 is a diagram illustrating a method of generating a model; FIG. 4 is a diagram illustrating a method of calculating an alignment error amount (correction value) using a model and performing correction; 4A and 4B are diagrams illustrating image data of alignment marks and alignment waveforms; FIG. FIG. 4 is a diagram illustrating a method of calculating mark position information (temporary position information) from alignment mark image data; FIG. 5 is a diagram illustrating a method of extracting or calculating a feature amount related to a non-measurement direction from image data of alignment marks; FIG. 5 is a diagram illustrating a method of extracting or calculating a feature amount related to a non-measurement direction from image data of alignment marks; FIG. 5 is a diagram illustrating a method of extracting or calculating a feature amount related to a non-measurement direction from image data of alignment marks; FIG. 4 is a diagram exemplifying the arrangement of alignment marks and overlay inspection marks; FIG. 5 is a diagram illustrating an amount of alignment error of alignment marks; FIG. 4 is a diagram illustrating the amount of alignment error and the degree of certainty of alignment marks; The figure explaining a moiré measurement system. FIG. 4 is a diagram exemplifying waveforms of image data of an alignment mark in a measurement direction and a non-measurement direction; FIG. 4 is a diagram exemplifying waveforms of image data of an alignment mark in a measurement direction and a non-measurement direction; FIG. 2 is a diagram exemplifying the configuration of an exposure apparatus; The figure which illustrates the structure of a measuring device. FIG. 4 is a diagram illustrating the configuration of an illumination aperture stop; 4A and 4B are diagrams illustrating configurations of alignment marks; FIG. FIG. 4 is a diagram for explaining optical parameter dependence of a signal waveform (measurement direction) obtained from image data of an alignment mark; FIG. 4 is a diagram for explaining dependence of a signal waveform (non-measurement direction) obtained from image data of an alignment mark on optical parameters; 4A and 4B are diagrams illustrating the asymmetry of alignment marks; FIG. 4 is a diagram illustrating the asymmetry of a plurality of alignment marks; FIG. 4 is a diagram illustrating a method of calculating an alignment error amount (correction value) using a model and performing correction; The figure explaining the article manufacturing method.

<First embodiment>
Hereinafter, embodiments will be described in detail with reference to the accompanying drawings. In addition, the following embodiments do not limit the invention according to the scope of claims. Although multiple features are described in the embodiments, not all of these multiple features are essential to the invention, and multiple features may be combined arbitrarily. Furthermore, in the accompanying drawings, the same or similar configurations are denoted by the same reference numerals, and redundant description is omitted.

In the first embodiment, an imprint apparatus will be described as an example of a lithography apparatus, but the imprint apparatus and the exposure apparatus have much in common regarding the alignment technique between the shot region of the substrate and the original. Therefore, the alignment technique described below can also be applied to the exposure apparatus. Therefore, an exposure apparatus will be described as a second embodiment.

FIG. 2A schematically shows the configuration of the imprint apparatus IMP according to the first embodiment. The imprinting apparatus IMP cures the imprinting material IM in a state in which the imprinting material IM on the shot area of the substrate S and the pattern area MP of the mold M are in contact with each other. Imprint processing for separating the mold M is performed. A pattern made of the cured imprint material IM is formed on the substrate S by this imprinting process.

As the imprint material, a curable composition (also referred to as an uncured resin) that cures when energy for curing is applied is used. Electromagnetic waves, heat, and the like can be used as energy for curing. The electromagnetic wave can be light having a wavelength selected from the range of 10 nm or more and 1 mm or less, such as infrared rays, visible rays, and ultraviolet rays. The curable composition can be a composition that is cured by irradiation with light or by heating. Among these, the photocurable composition that is cured by irradiation with light contains at least a polymerizable compound and a photopolymerization initiator, and may further contain a non-polymerizable compound or a solvent if necessary. The non-polymerizable compound is at least one selected from the group consisting of sensitizers, hydrogen donors, internal release agents, surfactants, antioxidants, polymer components and the like. The imprint material can be placed on the substrate in the form of droplets, or in the form of islands or films formed by connecting a plurality of droplets, by an imprint material supply device (not shown). The viscosity of the imprint material (viscosity at 25° C.) can be, for example, 1 mPa·s or more and 100 mPa·s or less. Materials for the substrate include, for example, glass, ceramics, metals, semiconductors, and resins. If necessary, a member made of a material different from that of the substrate may be provided on the surface of the substrate. The substrate is, for example, a silicon wafer, a compound semiconductor wafer, quartz glass.

In this specification and drawings, directions are indicated in an XYZ coordinate system with the horizontal plane as the XY plane. In general, the substrate S is placed on the substrate holder 102 so that its surface is parallel to the horizontal plane (XY plane). Therefore, hereinafter, the directions perpendicular to each other in the plane along the surface of the substrate S are defined as the X-axis and the Y-axis, and the direction perpendicular to the X-axis and the Y-axis is defined as the Z-axis. Further, hereinafter, directions parallel to the X axis, Y axis, and Z axis in the XYZ coordinate system are referred to as the X direction, Y direction, and Z direction, respectively. Also, rotation about the X-axis, rotation about the Y-axis, and rotation about the Z-axis are indicated by θx, θy, and θz, respectively.

The imprint apparatus IMP includes a substrate holding unit 102 that holds the substrate S, a substrate driving mechanism 105 that drives the substrate S by driving the substrate holding unit 102, a base 104 that supports the substrate holding unit 102, and a substrate holding unit 102. A position measurement unit 103 that measures a position can be provided. Substrate drive mechanism 105 may include, for example, a motor such as a linear motor. The imprint apparatus IMP can include a sensor 151 that detects a substrate driving force (alignment load) required for the substrate driving mechanism 105 to drive the substrate S (substrate holding unit 102) during alignment. The substrate driving force in the alignment performed while the imprint material IM on the substrate S and the pattern region MP of the mold M are in contact with each other corresponds to the shear force acting between the substrate S and the mold M, for example. The shear force is a force acting mainly in the planar direction of the substrate S and the mold M. As shown in FIG. The substrate driving force during alignment has a correlation with, for example, the magnitude of the current supplied to the motor of the substrate driving mechanism 105 during alignment, and the sensor 151 detects the substrate driving force based on the magnitude of the current. can do. The sensor 151 is an example of a sensor that measures the influence (shear force) that the mold M receives during pattern formation. A drive request (command value) issued to the substrate drive mechanism 105 by the controller 110, which will be described later, is called a stage control value.

The imprint apparatus IMP can include a mold holding unit 121 that holds the mold M, a mold driving mechanism 122 that drives the mold M by driving the mold holding unit 121 , and a support structure 130 that supports the mold driving mechanism 122 . Mold drive mechanism 122 may include a motor such as, for example, a voice coil motor. The imprint apparatus IMP can include a sensor 152 that detects the release force (separation load) and/or the pressing force. The release force is a force required to separate the hardened material of the implant material IM on the substrate S from the mold M. As shown in FIG. The pressing force is the force with which the mold M is pressed in order to bring the mold M into contact with the imprint material IM on the substrate S. The release force and the pressing force are forces acting mainly in a direction perpendicular to the planar direction of the substrate S and the mold M. As shown in FIG. The mold release force and the pressing force are correlated with, for example, the magnitude of the current supplied to the motor of the mold driving mechanism 122, and the sensor 152 detects the separating force and the pressing force based on the magnitude of the current. be able to. The sensor 152 is an example of a sensor that measures the influence (mold release force and/or pressing force) that the mold M receives during pattern formation. A drive request (command value) issued to the mold drive mechanism 122 by the control unit 110, which will be described later, is also called a stage control value.

The substrate driving mechanism 105 and the mold driving mechanism 122 constitute a driving mechanism that adjusts the relative position and the relative attitude between the substrate S and the mold M. As shown in FIG. Adjustment of the relative positions of the substrate S and the mold M by the drive mechanism is for contact of the mold with the imprint material on the substrate S and for separation of the mold from the cured imprint material (pattern of cured material). including driving The substrate driving mechanism 105 can be a driving mechanism having a plurality of degrees of freedom (eg, three axes of X, Y, and θZ, preferably six axes of X, Y, Z, θX, θY, and θZ). The die drive mechanism 122 can also be a drive mechanism with multiple degrees of freedom (eg, three axes of Z, θX, and θY, preferably six axes of X, Y, Z, θX, θY, and θZ).

The imprint apparatus IMP can include a mold transport mechanism 140 that transports the mold M and a mold cleaner 150 . The mold conveying mechanism 140 can be configured, for example, to convey the mold M to the mold holding unit 121 and to convey the mold M from the mold holding unit 121 to an original stocker (not shown), the mold cleaner 150, or the like. The mold cleaner 150 cleans the mold M with ultraviolet rays, chemicals, or the like.

The mold holding part 121 can include a window member 125 that forms the pressure control space CS on the back surface of the mold M (the surface opposite to the pattern area MP in which the pattern to be transferred to the substrate S is formed). The imprinting apparatus IMP controls the pressure in the pressure control space CS (hereinafter referred to as cavity pressure) to place the pattern region MP of the mold M on the substrate S as schematically shown in FIG. A deformation mechanism 123 may be provided for deforming into a convex shape. Also, the imprint apparatus IMP can include an alignment scope 106 , a curing section 107 , an imaging section 112 and an optical member 111 .

The alignment scope 106 illuminates the alignment mark (first mark) of the substrate S (first member) and the alignment mark (second mark) of the mold M (second member), and an optical image formed by both alignment marks. can generate image data. The alignment scope 106 or the control unit 110 can detect relative position information between marks by processing image data obtained by imaging. The alignment scope 106 can be positioned by a driving mechanism (not shown) according to the position of the alignment mark to be observed. The image data generated by imaging with the alignment scope 106 is hereinafter also referred to as an alignment image. Further, the result of measurement using the alignment scope 106 is also called an alignment measurement value.

An example of the alignment image observed by the alignment scope 106 can be image data generated by capturing optical images formed by reflected light from the first and second marks. Another example of the alignment image can be image data generated by imaging moiré (interference fringes), which is an optical image formed by the first marks and the second marks.

The curing unit 107 irradiates the imprint material IM with energy (for example, light such as ultraviolet light) for curing the imprint material IM through the optical member 111, thereby curing the imprint material IM. The imaging unit 112 images the substrate S, the mold M, and the imprint material IM via the optical member 111 and the window member 125 . Image data obtained by imaging by the imaging unit 112 is also called a spread image.

The imprinting apparatus IMP may comprise a dispenser 108 for placing imprinting material IM onto the substrate S. FIG. The dispenser 108 discharges the imprint material IM so that the imprint material IM is arranged on the substrate S according to, for example, a drop recipe indicating the arrangement of the imprint material IM. The imprint apparatus IMP includes a control unit 110 that controls a substrate driving mechanism 105, a mold driving mechanism 122, a deformation mechanism 123, a mold transport mechanism 140, a mold cleaner 150, an alignment scope 106, a curing unit 107, an imaging unit 112, a dispenser 108, and the like. can be provided. The control unit 110 is, for example, PLD (abbreviation of Programmable Logic Device) such as FPGA (abbreviation of Field Programmable Gate Array), or ASIC (abbreviation of Application Specific Integrated Circuit), or program 113 is incorporated. It can be configured by a general-purpose computer, or a combination of all or part of these.

FIG. 3 illustrates the configuration of an article manufacturing system 1001 for manufacturing articles such as semiconductor devices. Article manufacturing system 1001 may include, for example, one or more lithographic apparatus (imprint apparatus IMP and/or exposure apparatus). In addition, the article manufacturing system 1001 can include one or more inspection devices 1005 (eg, overlay inspection device, foreign matter inspection device) and one or more processing devices 1006 (etching device, deposition device). Furthermore, the article manufacturing system 1001 may also include a model generation device 1007 that generates a model for calculating the registration error amount. These devices are connected to a control device 1003 which is one of the external systems via a network 1002 and can be controlled by the control device 1003 . Examples of the control device 1003 include MES, EEC, and the like. The model generation device 1007 can be configured by, for example, a PLD such as FPGA, ASIC, a general-purpose computer in which a program is installed, or a combination of all or part of these. An example of the model generation device may be a server called EdgeServer. The model generation device 1007 may be incorporated in the control unit of the imprint apparatus IMP or the exposure apparatus, the control device 1003, or the like. A system including a lithographic apparatus, such as an imprint apparatus IMP or an exposure apparatus, and a model generator 1007 may be understood as a lithographic system.

The alignment scope 106 and the control unit (processing unit) 110 of the imprint apparatus IMP can constitute a measurement apparatus that measures or detects position information of the measurement target. In another aspect, imprint apparatus IMP includes a measurement apparatus that measures or detects position information of a measurement target. A measuring device can operate as a measuring device that measures or detects position information of a measurement object. For example, the measurement apparatus can measure the position information of the measurement object in the diffraction direction of the diffraction grating that forms the alignment mark, that is, in the first direction, which is the measurement direction. The measurement device can also be configured to measure position information of the measurement object in a second direction (non-measurement direction) different from the first direction (for example, a direction perpendicular to the first direction). The processing unit can determine the position information by obtaining provisional position information of the measurement object based on the image data and correcting the provisional position information based on the correction value based on the feature amount of the image data. For example, the processing unit determines the position of the measurement object in the first direction based on the temporary position information of the measurement object in the first direction obtained from the image data and the correction value based on the feature amount of the image data in the second direction. Location information can be determined. The measuring device may further include a model for obtaining correction values based on feature quantities. Moreover, the measuring device may further include a machine learning unit that generates the model by machine learning.

In the following, a lithographic method using alignment corrections will be described. The registration error amount can be estimated by machine learning from the alignment images. In this embodiment, the alignment error amount and its confidence are estimated from the alignment image by machine learning. The alignment error amount corresponds to the error of the alignment mark position calculated from the alignment image. Confidence is an index (information) that indicates the degree of confidence that the estimated registration error amount is highly accurate. good.

In addition to information on the measurement direction of the alignment image, information on the non-measurement direction can be used as the feature amount of the alignment image. The information in the non-measurement direction means information in a direction different from the measurement direction. As an example, the information on the non-measurement direction can be the feature quantity in the direction orthogonal to the measurement direction. Alternatively, the information on the non-measurement direction can be a difference amount in each coordinate of the measurement direction of signal sequences of a plurality of measurement directions integrated in different ranges in the non-measurement direction, which will be described later. By using such information, it becomes possible to estimate the registration error amount and confidence with high accuracy. For example, machine learning is used to calculate the registration error amount and confidence. A model for estimating the registration error amount from the information from the alignment image is learned in advance from the information on the alignment image and the measurement value of the overlay inspection. After that, during actual alignment, using the calculated model, the alignment error amount and the degree of confidence, which are errors in the mark position, are calculated from the alignment image, and used for the alignment operation.

The lithography method can include a measurement method for measuring the position information of the measurement object, a measurement method for measuring alignment errors between the shot regions of the substrate and the mold, and an alignment method for aligning the shot regions of the substrate and the mold. . In the lithography method, an alignment error amount as a correction value or correction amount is estimated from image data of an inspection object. Here, the inspection object can be (an optical image of) a mark or an optical image (for example, moire) formed by the first mark and the second mark. The alignment error amount is the error amount (estimated error amount) of the mark position (position information) calculated based on the image data of the inspection object, or the relative position between the first mark and the second mark (relative position information) error amount (estimated error amount).

FIG. 1 shows, as an embodiment of the lithographic method, a lithographic method performed in an imprint system including an imprint apparatus IMP. The operations shown in FIG. 1 can be controlled by the controller 110 .

In step S101, a substrate transport mechanism (not shown) transports the substrate S from a transport source (for example, a relay unit between the pretreatment apparatus and the imprint apparatus IMP) onto the substrate holding unit 102 . In steps S102 to S106, imprint processing (pattern formation) is performed on a shot area selected from a plurality of shot areas on the substrate S. FIG. In step S102, the imprint material IM is placed by the dispenser 108 on the selected shot area. This process can be performed, for example, by discharging the imprint material IM from the dispenser 108 while driving the substrate S by the substrate driving mechanism 105 . In step S103, the substrate S and the mold M are relatively driven by at least one of the mold driving mechanism 122 and the substrate driving mechanism 105 so that the pattern region MP of the mold M contacts the imprint material IM on the shot region. be. In one example, the mold driving mechanism 122 drives the mold M so that the pattern area MP of the mold M contacts the imprint material IM on the shot area. In the process of bringing the pattern region MP of the mold M into contact with the imprint material IM, the pattern region MP of the mold M can be deformed into a convex shape toward the substrate S by the deformation mechanism 123 . At this time, the cavity pressure can be controlled and its value accumulated. In addition, in the process of bringing the pattern region MP of the mold M into contact with the imprint material IM, the image capturing unit 112 can perform image capturing, and the captured image (spread image) can be accumulated.

In step S104, the shot area of the substrate S and the pattern area MP of the mold M may be aligned. Alignment can be performed by using the alignment scope 106 to measure the relative positions of the first marks of the shot area and the second marks of the mold M so that the relative positions fall within the allowable range of the target relative positions. In alignment, the substrate S and the mold M can be relatively driven by at least one of the mold driving mechanism 122 and the substrate driving mechanism 105 . Here, the target value of the relative driving amount between the substrate S and the mold M can be obtained by correcting the provisional position information (provisional relative position information) based on the alignment error amount (correction value). Temporary position information (temporary relative position information) is information determined based on image data obtained using the alignment scope 106, and can indicate a temporary relative position between the shot area of the substrate S and the mold M. The registration error amount can be calculated based on the image data obtained using the alignment scope 106 . The registration error amount can be calculated using a model generated by the model generator 1007 and provided to the controller 110 of the imprint apparatus IMP. The correction of the provisional position information (provisional relative position information) by the registration error amount (correction value) may be performed during the entire alignment execution period, or when the relative position between the shot area and the mold M is equal to or less than the reference value. It may be executed after the point in time when it is settled. The control unit 110 can accumulate image data obtained using the alignment scope 106 and provide the accumulated image data to the model generation device 1007 . The model generation device 1007 can generate a model for determining the registration error amount based on the image data thus provided from the control unit 110 of the imprint apparatus IMP.

Here, a method for measuring the position of the mark will be exemplified. FIG. 6A illustrates a mark image (image data) 401 obtained by imaging a mark for measuring the position in the X direction, and FIG. 6B illustrates an alignment waveform obtained from the mark image 401. 406 is illustrated. The substrate S may have a mark corresponding to the mark image of FIG. 6(a) and a mark rotated by 90 degrees. A mark corresponding to the mark image in FIG. 6(a) is used to measure the position in the X direction, and the measurement direction is the X direction. A mark obtained by rotating the mark corresponding to the mark image in FIG. 6A by 90 degrees is used to measure the position in the Y direction, and the measurement direction is the Y direction.

Using a first mark for measuring the position in the X direction provided in the shot area of the substrate S and a first mark for measuring the position in the Y direction provided in the shot area of the substrate S, the shot The position of the region in the X direction and the position in the Y direction can be detected as temporary position information. Also, using the second mark for measuring the position in the X direction provided on the mold M and the second mark for measuring the position in the Y direction provided on the mold M, the position of the mold M in the X direction is measured. The position and the position in the Y direction can be detected as temporary position information. The controller 110 can determine the correct relative position (alignment error) between the shot area of the substrate S and the mold M based on the temporary position information and the correction value described above.

Alternatively, from the moiré fringes formed by the first marks for the X direction provided in the shot region of the substrate S and the second marks for the X direction provided on the mold M, A relative position can be detected as provisional relative position information. Similarly, from the moire fringes formed by the first marks for the Y direction provided in the shot region of the substrate S and the second marks for the Y direction provided on the mold M, the Y direction between the shot region and the mold M is detected. can be detected as temporary relative position information. The controller 110 can determine the correct relative position (alignment error) between the shot area of the substrate S and the mold M based on the temporary relative position information and the correction value described above.

FIG. 7 shows how the alignment scope 106 is used to measure the position of the mark. A method of measuring the mark position 402 will be described below using the mark image of FIG. 6A as an example. A mark position 402 to be measured is the center position of the mark image in the measurement direction (the X direction in FIG. 6A), which is also the center position of the mark corresponding to the mark image. In this example, the measuring direction 404 is the X direction and the non-measuring direction 405 is the Y direction.

In step S<b>501 , the control unit 110 obtains a mark image 401 (image data) by imaging the mark using the alignment scope 106 . In step S<b>502 , controller 110 generates alignment waveform 406 based on mark image 401 . The alignment waveform 406 can be generated, for example, by calculating an integrated value for each of the pixels at the same position in the measurement direction 404 (X direction) among the plurality of pixels forming the measurement area 403 including the mark image 401 .

In step S<b>503 , the control section 110 calculates the mark position 402 based on the alignment waveform 406 . As an example of the calculation method, there is a method in which the position of the center of gravity of the alignment waveform 406 is used as the mark position 402 . As another example, there is a method of calculating the mark position by calculating the phase of the alignment waveform by Fourier transform or the like, or a method of calculating the mark position using a pattern matching method.

Returning to FIG. 1, in step S105, the imprint material IM between the substrate S and the pattern region MP of the mold M is irradiated with energy for curing the imprint material IM by the curing unit 107 . Thereby, the imprint material IM is cured to form a cured product of the imprint material IM. In step S106, the substrate S and the mold M are relatively driven by at least one of the mold driving mechanism 122 and the substrate driving mechanism 105 so that the cured imprint material IM and the pattern region MP of the mold M are separated. be done. In one example, the mold M is driven by the mold driving mechanism 122 so that the cured imprint material IM and the pattern region MP of the mold M are separated. Also when the cured imprint material IM and the pattern region MP of the mold M are separated, the pattern region MP of the mold M can be deformed into a convex shape toward the substrate S. Further, imaging is performed by the imaging unit 112, and the state of separation between the imprint material IM and the mold M can be observed based on the captured image.

In step S107, the control unit 110 determines whether the imprint processing of steps S102 to S106 has been performed on all the shot regions of the substrate S. Then, the control unit 110 proceeds to step S108 when the imprint processing of steps S102 to S106 has been performed on all the shot areas of the substrate S, and proceeds to step S102 when there is an unprocessed shot area. return. In this case, the imprint processing of steps S102 to S106 is performed on shot areas selected from unprocessed shot areas.

In step S108, the substrate S is transported from the substrate holding unit 102 to a transport destination (for example, a relay unit with the post-processing device) by a substrate transport mechanism (not shown). The operations shown in FIG. 1 are performed for each of a plurality of substrates when a lot consisting of a plurality of substrates is processed.

Next, a model generation method in the model generation device 1007 will be described with reference to FIG. Note that, as described above, the model generation device 1007 may be incorporated in the imprint apparatus IMP (for example, the control unit 110), in which case the model is generated in the imprint apparatus IMP.

In step S201, first, the model generation device 1007 acquires the measurement values of one shot area of the substrate S measured by the overlay inspection device. The acquired measurement value can be the result of measuring the overlay accuracy of at least one point belonging to each shot region of the substrate S. FIG. The measured value can be, for example, the overlay deviation amount between (the overlay inspection mark of) the underlying layer of the substrate S and the layer (the overlay inspection mark of) formed thereon by the imprint apparatus IMP. Then, in step S201, the model generating apparatus 1007 calculates the difference between the measurement value by the overlay inspection apparatus and the final measurement value by the imprint apparatus IMP (final alignment error in step S104) as the registration error amount. .

In step S202, the model generation device 1007 first acquires the mark image (image data) of the mark in the shot area for which the measurement value was acquired in the immediately previous step S201. This mark image is acquired using the alignment scope 106 in step S104, and can be provided from the imprint apparatus IMP to the model generation apparatus 1007 at any timing after the end of step S104. In step S202, the model generation device 1007 further obtains the feature amount of the acquired mark image. This feature quantity includes at least a feature quantity relating to the non-measurement direction, and may additionally include a feature quantity relating to the measurement direction. The measurement direction is the X direction in the example of FIG. The non-measurement direction is a direction crossing the X direction in the example of FIG. 6, for example, the Y direction.

In step S203, the model generation device 1007 determines whether or not steps S201 and S202 have been performed for all of the plurality of shot areas of the substrate S, and if there is an unexecuted shot area, the process is performed on the unexecuted shot area. S201 and S202 are executed. Then, when steps S201 and S202 are completed for all of the plurality of shot areas, in step S204, the model generating device 1007 generates a model for estimating the registration error amount based on the feature amount.

Model generation can be performed, for example, by machine learning. Specific examples are as follows. First, a new layer (pattern) is formed under the same conditions in a plurality of shot regions on the substrate by the imprint apparatus IMP. Then, the overlay inspection device 1005 measures the amount of overlay deviation between (the overlay inspection mark of) the underlying layer and (the overlay inspection mark of) the newly formed layer in each shot area. Next, the model generation device 1007 acquires the measured amount of misalignment of each shot area, and determines the difference between the amount of misalignment and the final measured value when a new layer is formed in the shot area. Calculate as a quantity. Then, the model generation device 1007 performs machine learning using the feature amount of the mark image of each shot area used when forming a new layer as input data for the model, and using the calculated alignment error amount as teaching data. conduct. At this time, if an abnormal value exists in the input data and/or the teacher data, it is preferable to perform machine learning while excluding that data.

In machine learning, preprocessing may be performed on the registration error amount. Examples of preprocessing include a method of adding an offset value to the amount of alignment error, and a method of multiplying the amount of alignment error by a value to change the scale of the error amount.

As a machine learning method, for example, Bayesian inference can be cited, which performs inference considering uncertainty by treating variables as probabilities. When Bayesian estimation is used, the model is a function that inputs feature values and outputs a probability distribution of registration errors, and internal variables can be optimized by machine learning. Examples of models include generalized linear models, hierarchical Bayesian models, and Gaussian processes. The expected value of the probability distribution of the output registration error amount can be used as the inference value of the registration error amount.

At step S205, the model generator 1007 saves the model generated at step S204. Also, the model generating apparatus 1007 can provide the model generated in step S204 to the control unit 110 of the imprint apparatus IMP.

Here, the reason for estimating the registration error amount based on the feature amount of the mark image in the non-measurement direction will be described. FIG. 14D shows the shot area of the substrate S and the shot area of the substrate S based on moire fringes, which are optical images formed by the first marks provided in the shot area of the substrate S and the second marks provided on the mold M. The principle of measuring relative position information with the mold M is shown. FIG. 14(d) shows the first marks 3a provided in the shot area of the substrate S and the second marks 2a provided in the mold M. As shown in FIG. The alignment scope 106 has an illumination optical system that illuminates the mark, and the illumination optical system has a pupil plane P. FIG. IL1, IL2, IL3, and IL4 represent illumination light from poles formed on the pupil plane P. FIG.

Illumination lights IL1 and IL2 are used to measure the relative positions of the shot area of the substrate S and the mold M in the X direction. As illustrated in FIG. 14A, in the measurement of the relative position in the X direction, the illumination lights IL3 and IL4, which are not used for the measurement of the relative position in the X direction, are the edges of the first mark 3a and the second mark 2a. can generate scattered light. This scattered light may become flare and be mixed into a moire fringe signal (moire fringe image data). FIG. 14(c) illustrates the signal intensity distribution (light intensity distribution on the light receiving surface of the imaging element of the alignment scope 106) in the X direction of the moire fringe signal in FIG. 14(a). It can be seen that the scattered light from the edges of the first mark 3a and the second mark 2a has large peaks on the left end side and the right end side in the signal intensity distribution. Of the four moire fringe signals, two periods on the left end side and right end side are affected by the scattered light, which affects the measurement accuracy of the relative position. The same is true for relative position measurement in the Y direction, and the illumination lights IL1 and IL2 that are not used for relative position measurement in the Y direction may generate scattered light at the edges of the first mark 3b and the second mark 2b. Then, this scattered light becomes flare light and can be mixed into the moire fringe signal. The above explains that the light intensity distribution in the measurement direction can be affected by flare. By the same principle, the light intensity distribution in the non-measurement direction can also be affected by flare and changed. A change in the light intensity distribution in the non-measurement direction can reduce the relative position or position measurement accuracy in the measurement direction.

15(a) and 16(a) show the signal values of pixels at the same position in the measurement direction (X direction) among the plurality of pixels forming the image data obtained using the alignment scope 106. A signal waveform obtained by calculating the integrated value is illustrated. The signal waveforms in FIGS. 15(a) and 16(a) can be understood as signal waveforms in the measurement direction. 15(b) and 16(b) show signal values of pixels having the same position in the non-measurement direction (Y direction) among a plurality of pixels forming image data obtained using the alignment scope 106. A signal waveform obtained by calculating the integrated value of is exemplified. The signal waveforms in FIGS. 15(b) and 16(b) can be understood as signal waveforms in the non-measurement direction. The examples of FIGS. 16A and 16B are more affected by flare light than the examples of FIGS. 15A and 15B. The signal waveform in the measurement direction illustrated in FIG. 16A is more distorted than the signal waveform in the measurement direction illustrated in FIG. Further, the signal waveform in the non-measurement direction illustrated in FIG. 16(b) has greater distortion and signal value variation than the signal waveform in the non-measurement direction illustrated in FIG. 15(b). ing. That is, it can be seen that the signal waveform in the non-measurement direction has a correlation with the signal waveform in the measurement direction, that is, the measurement result in the measurement direction. Therefore, the feature amount of the image data related to the non-measurement method is obtained, and the position information of the object to be measured is obtained by correcting the provisional position information of the object to be measured in the measurement direction obtained from the image data based on the feature amount. It can be determined with high accuracy.

Here, the feature quantity obtained from the image data in the non-measurement direction (second direction) can include multiple values 603 corresponding to multiple positions in the non-measurement direction, as illustrated in FIG. The plurality of values 603 includes a plurality of integrated values, and each of the plurality of integrated values is an integrated value of the signal values of pixels having the same position in the non-measurement direction among the plurality of pixels forming the image data. Possible. Alternatively, the plurality of values 603 may include signal values of a plurality of pixels on a line parallel to the non-measurement direction among the plurality of pixels forming the image data. Alternatively, the plurality of values 603 may be a plurality of values obtained by processing signal values of a plurality of pixels on a line parallel to the non-measurement direction among the plurality of pixels forming the image data. Alternatively, the plurality of values 603 are obtained by performing a basis conversion on a plurality of integrated values, and each of the plurality of integrated values has a position in the non-measurement direction among the plurality of pixels forming the image data. It can be the sum of the respective signal values of equal pixels. Alternatively, the plurality of values 603 may be values obtained by performing basis conversion on signal values of a plurality of pixels on a line parallel to the non-measurement direction among the plurality of pixels forming the image data. Alternatively, the plurality of values 603 are obtained by processing the signal values of the plurality of pixels on a line parallel to the non-measurement direction among the plurality of pixels forming the image data, and performing basis conversion on the plurality of values. It can be a value obtained by

Alternatively, in FIG. 9, the difference between the result of integrating the signal values of the pixels at the same position in the measurement direction in the region 701 and the result of integrating the signal values of the pixels at the same position in the measurement direction in the region 702 is calculated. It is good also as a feature-value in a non-measurement direction.

An example of calculating or determining the feature amount of image data in the non-measurement direction will be described below with reference to FIG. 10 . In FIG. 10, x1, x2, . y1, y2, . . . represent the Y coordinates (pixel positions in the Y direction) of the image data. Below, the pixel value of a pixel whose X coordinate is x1 and whose Y coordinate is y1 is represented as x1y1. Here, the coordinates x1, x2, . . . , y1, y2, .

In one example, by accumulating the pixel values of pixels having the same y-coordinates such as (x1y1+x2y1+x3y1+...), (x1y2+x2y2+x3y2...), etc., the feature of the signal waveform in the non-measurement direction is obtained as a feature quantity. can be obtained as Such a method is effective when diffracted light and/or scattered light are generated along the non-measurement direction.

When diffracted light and scattered light are locally generated, (x1y1), (x1y2), (x1y3), (x1y4), (x1y5), (x1y6), (x2y1), (x2y2), . , the pixel value of the pixel at each coordinate may be used as it is as the feature quantity in the non-measurement direction. Here, feature amounts may be determined as (x1y1+x1y2), (x1y3+x1y4), (x1y5+x1y6), (x2y1+x2y2), (x2y3+x2y4), . Thus, by adding the pixel values of a plurality of pixels in the y direction, it is possible to reduce the number of data representing the feature amount, thereby reducing the amount of calculation in calculating the correction value based on the feature amount. can be done. The total value of pixel values in each group may be extracted as a feature quantity such that the average coordinates of groups of pixels are in ascending order, such as (x1y1+x1y3), (x1y2+x1y4), and so on. Alternatively, the total value of the pixel values in each group may be extracted as a feature value such that the coordinates of groups of pixels partially overlap, such as (x1y1+x1y2+x1y3), (x1y3+x1y4+x1y5), and so on. good. Alternatively, addition in the x and y directions such as (x1y1+x1y2+x2y1+x2y2), (x1y2+x1y3+x2y2+x2y3), (x1y3+x1y4+x2y3+x2y4), . When diffracted light and/or scattered light are generated along an oblique direction, addition is performed in an oblique direction such as (x1y1+x2y2), (x2y2+x3y3), (x1y2+x2y3), (x2y3+x3y4), . may be extracted.

Alternatively, the pixel value of each pixel may be multiplied by constants α, β, γ, . As a result, it is possible to arbitrarily lower the weights of the feature quantities for which the effect of correction is small. Also, the feature amounts may be determined as (αx1y1+βx1y2+γx1y3), (ax1y2+bx1y3+cx1y4), (px1y3+qx1y4+rx1y5), . Here, α, β, γ, a, b, c, p, q, and r are constants by which pixel values are multiplied. When α=a=p=-1, β=b=p=2, and γ=c=r=-1, the gradient in the non-measurement direction can be obtained as a feature quantity.

A new feature amount may be obtained by performing basis conversion on the obtained feature amount in the non-measurement direction. Examples of basis transformation include a method of obtaining phase and amplitude by performing Fourier transformation, and a method of obtaining basis by principal component analysis and performing basis transformation to reduce the amount of information. Also, a new feature amount can be obtained by adding or multiplying an offset value to the feature amount. In addition to the feature quantities in the non-measurement directions, the feature quantities in the measurement directions may be used to determine the correction values based thereon. An arbitrary point of the waveform (image data) in the measurement direction may be sampled and used as a feature amount, and basis conversion, offset addition, and multiplication can be performed in the same manner as the feature amount in the non-measurement direction.

The process of obtaining the feature amount from the image data may be executed by the control unit 110 or the like of the imprint apparatus IMP.

Hereinafter, the processing executed in step S104 (alignment) described above will be described with reference to FIG. In this process, the registration error amount is calculated using the model described above, and the provisional position information regarding the measurement direction obtained based on the image data is corrected based on the registration error amount (correction amount).

In step S<b>301 , the control unit 110 of the imprint apparatus IMP acquires the model generated by the model generation apparatus 1007 . Note that the acquisition of the model need not be performed immediately before the next step, S302, and may be performed at any timing, such as before the above-described step S102.

In step S302, the control unit 110 acquires the image data captured by the alignment scope 106 in step S104, and extracts or calculates at least the feature amount regarding the non-measurement direction from the image data. The feature amount extraction or calculation method in step S302 is the same as the feature amount extraction or calculation method executed by the model generation device 1007 in step S202.

In step S303, the control unit 110 calculates the registration error amount using the model acquired in step S301 and the feature amount extracted or calculated in step S302. For example, when Bayesian estimation is used for learning, the feature amount is input to the model acquired in step S301, the expected value of the probability distribution output from the model is acquired as the registration error amount, and the probability distribution Variance can be taken as confidence. The registration error amount can be used as a correction value.

In step S304, the control unit 110 obtains the position information of the object to be measured in the measurement direction as temporary position information based on the light intensity distribution in the measurement direction of the image data captured by the alignment scope 106 in step S104. This provisional position information is provisional position information that is obtained without considering the feature amount regarding the non-measurement direction of the image data.

In step S305, the control unit 110 adjusts the measurement object in the measurement direction based on the temporary position information obtained in step S304 and the correction value based on the feature amount of the image data regarding the non-measurement direction obtained in step S303. Find the location information of . Specifically, the control unit 110 subtracts the correction value based on the feature amount of the image data regarding the non-measurement direction obtained in step S303 from the provisional position information obtained in step S304, thereby reducing the position of the measurement object in the measurement direction. location information can be obtained.

In step S305, the control unit 110 detects an abnormality using the confidence factor obtained in step S303 or a value calculated based on the confidence factor. In one example, control unit 110 determines that there is an abnormality when the degree of certainty exceeds a threshold. In response to this abnormality detection, the imprint apparatus IMP can perform control different from normal alignment.

A specific example of the control different from the normal alignment will be described. Consider control for aligning the relative positions of the mold and the substrate with the target relative positions by using the measured values of one or more alignment marks arranged in each shot area of the substrate during exposure. . Specifically, the target relative positions s _x , s _y , θ _x , θ _y , β _x and β _y that minimize the following evaluation formula V are obtained.

However, n is the number of alignment marks, i is a variable indicating the position, and is an integer from 1 to n. x _i and y _i indicate the designed positions of the alignment marks in the shot area in the X and Y directions, respectively. dx _i and dy _i are the alignment measurements in the X and Y directions, respectively. s _x and s _y represent shift components of the target relative position, θ _x and θ _y represent rotation components, and β _x and β _y represent elongation components. Position control is performed.

If the alignment measurement values dx _i and dy _i contain an extremely large abnormal value, each component of the relative position is strongly affected by the abnormal value, making it impossible to calculate an accurate value. Therefore, the control unit 110 calculates the certainty of the marks in the shot area, and if the certainty of a certain mark is below the threshold, does not use the measured value of that mark and uses the measured values of the remaining marks. Calculate the relative position. This makes it possible to calculate relative positions that are not affected by abnormal values. Alternatively, by obtaining the target relative position by multiplying the measured value of each mark by the weight w according to the degree of certainty, it is possible to reduce the influence of the abnormal value without setting a threshold. The evaluation formula in this case is represented by the following formula.

In addition, the control unit 110 can extend the alignment control time when an abnormality is found, and perform normal alignment when the certainty increases. This is effective when an alignment mark image is not formed normally due to insufficient filling of the imprint material in the pattern portion and is detected as an abnormality. In such a case, the imprint material may be filled as time elapses, and the alignment mark image is normally formed, and the confidence of the mark increases.

The verification results of this embodiment are shown below. In this verification, the amount of alignment error was corrected for eight alignment marks 801, 803, 805 and 807 in the X direction and alignment marks 802, 804, 806 and 808 in the Y direction in the shot area as shown in FIG. did 809 is a mark used in the overlay inspection apparatus. The alignment error amount of each alignment mark in this verification is calculated based on the overlay measurement results obtained from the overlay inspection marks in the vicinity of the alignment mark. For training to generate the model, data of 20 substrates x 69 shot areas are used, and correction is applied to 6 substrates x 69 shot areas that are different from the data used for training. did. The position and direction of alignment marks were learned and corrected independently.

FIG. 12 shows the standard deviation of the alignment error amount of all data before and after correction, and shows how much the alignment error amount varies. The purpose of the present embodiment is to reduce this variation, and it can be confirmed from the graph that the variation in the amount of alignment error is reduced by about 16% at maximum.

FIG. 13 is a graph showing the alignment error amounts of each data before and after correction of the alignment mark 805 side by side. For example, looking at the circled part on the right side of the graph, it can be confirmed that the variation is reduced by the correction. On the other hand, looking at the encircled portion on the left side of the graph, the amount of registration error is extremely large, and there are data that have not been improved even after correction. The degree of certainty of this data is extremely small compared to other data, so it can be confirmed that an abnormality has occurred in the alignment mark and correct measurement cannot be performed.

As described above, as an embodiment, an example in which a correction value calculated from data indicating the state of the imprint apparatus IMP during imprint processing is applied to superimposition of the shot areas has been described, but the present invention is not limited to this. For example, a correction value may be provided from the imprint apparatus IMP to the control device 1003, and the correction value may be used in subsequent processing. For example, a correction value obtained from information indicating the state of the imprint apparatus IMP during imprint processing may be applied during alignment of another shot area such as the next shot area. Alternatively, the correction value may be applied during alignment of the shot area at the same position on the next substrate.

In the above-described embodiment, the method of calculating the correction value and confidence factor during imprinting has been described, but the present invention is not limited to this. Anomaly detection may be performed by calculating the certainty of alignment measurement for a substrate that has been exposed.

Also, in the above-described embodiments, the imprint apparatus has been described, but the present invention is not limited to this. As mentioned above, it is also applicable, for example, to other lithographic apparatus, such as exposure apparatus. In the exposure apparatus, it is possible to similarly correct the alignment position using the alignment mark image when forming the pattern.

Further, in the above-described embodiment, an example in which a model generation device calculates (learns) a model for calculating a correction value has been described, but the present invention is not limited to this. As described above, the model generation apparatus may be realized by using (using together) the control unit of the imprint apparatus or the exposure apparatus, the control apparatus 1003, or the like.

The lithography method using alignment correction according to the present embodiment has been described above. Improvement in alignment accuracy is expected by using such correction.

On the other hand, the alignment accuracy depends on the optical parameters (for example, parameters such as illumination wavelength, σ stop, NA, etc.) in the imaging of the alignment marks in the alignment scope 106, and the stepped structure on the substrate surface (for example, the circuit pattern to be exposed, the pre-process A stepped structure formed by an underlying circuit pattern or the like formed in .).

An example of the relationship between the optical parameters and the step structure of the alignment scope 106 and the alignment accuracy will be described with reference to FIGS. 22 and 23. FIG. FIG. 22(a) shows a plan view of the alignment mark AM in one example as seen from the Z direction. Here, the measurement direction is the X direction, and in this case the alignment mark AM is expressed as one mark extending in the Y direction. 22(b) and 22(c) show cross-sectional views of the alignment mark AM along the XZ plane. The alignment mark AM can have a structure in which a metal material such as tungsten is embedded in an insulating material such as SiO2. A resist 2000, which is a photosensitive material, is applied over the alignment marks AM. An image of the alignment mark can be captured by irradiating the alignment mark with light and obtaining reflected light.

In semiconductor processes, CMP (Chemical Mechanical Polishing) may be used as a technique for flattening a substrate surface. In that case, the shape of the alignment mark AM may be changed by being microscopically scraped by the process. FIG. 22(b) shows a state (so-called dishing) in which the surface shape of the alignment mark AM is ground to an ideal shape. Here, the alignment marks AM are cut symmetrically with respect to the center line C extending in the Y direction. However, in an actual CMP process, the polishing conditions (polishing direction, polishing material, etc.) may deviate from the ideal state. FIG. 22(c) shows a state in which the polished state deviates from the ideal. Here, it is cut asymmetrically with respect to the center line C, and the lowest position is shifted by ΔX from the center in the X direction.

FIG. 23(a) shows a plan view of the alignment mark AM in another example as seen from the Z direction. The alignment mark AM in FIG. 22(a) is a single mark extending in the Y direction, but the alignment mark AM in FIG. 23(a) is a plurality of marks extending in the Y direction. In the example of FIG. 23A, the alignment marks AM are four marks extending in the Y direction. 23(b) and 23(c) show cross-sectional views of the alignment mark AM along the XZ plane, similar to FIGS. 22(b) and 22(c). In the examples of FIGS. 23(b) and 23(c), in addition to the aforementioned dishing phenomenon, a state (so-called erosion) occurs in which portions other than marks (between marks) are also scraped away. FIG. 23(b) shows a case where such a phenomenon occurs symmetrically with respect to a center line C extending in the Y direction and passing through the center of the four marks in the X direction. On the other hand, FIG. 23(c) shows a state in which erosion asymmetrical with respect to the center line C occurs. The amount of occurrence of this erosion phenomenon can vary depending on the density (eg, line width, pitch), material, polishing conditions (eg, polishing direction, polishing material) of the plurality of alignment marks AM.

When an alignment mark having such an asymmetrical shape is irradiated with light, an optical path difference occurs between the optical path of the reflected light and the ideal optical path, resulting in an alignment error in the measured value. The alignment error due to this optical path difference depends on the wavelength of the irradiation light.

As a method of solving this problem, a method of irradiating light from a plurality of light sources that generate light with wavelengths different from each other is conceivable. The amount of error can be reduced by adjusting the intensity of the light sources so that the light from each light source cancels each other's alignment errors. Alternatively, the amount of error can be reduced by irradiating light from a light source having a wide range of wavelengths. At this time, the wavelength width of the light source for irradiation can be selected, and the wavelength width appropriate for the measurement target (that is, the wavelength width can be selected such that the amount of alignment error is reduced). Appropriate intensity ratios and wavelength widths may vary depending on the step structure on the substrate surface to be measured.

As another method, a plurality of alignment marks having different conditions such as alignment mark density (line width or pitch), material, and polishing conditions (polishing direction, polishing material) are formed in advance, and the optimum measurement value is obtained. can be obtained by selecting alignment marks. This means that the alignment mark conditions themselves can also be treated as the aforementioned optical parameters. It is also possible to select the optimum condition from the combination of alignment mark condition, detection method (bright field, dark field, moiré, self-interference method, etc.), wavelength, and other optical parameters. be.

By selecting the optimum optical parameters in this manner, an improvement in alignment accuracy can be expected. These optimal optical parameters may differ depending on whether the measured values are optimized before correction without considering the correction, or when the measured values are optimized after correction with consideration given to the correction. Conceivable. As an example, continue to consider the selection of the optimum illumination wavelength. As described above, combining multiple wavelengths is expected to be robust against alignment mark shape asymmetry, but it is also expected to increase the complexity of the illumination light and alignment mark images. Considering the correction by machine learning described above, an increase in the complexity of the estimation target is expected to adversely affect the learning of the error amount estimation model. Conversely, measurement using a single wavelength is likely to be affected by asymmetry, but since the measurement light is simple, it is expected that learning and estimation of the error amount will be easy, and high-precision correction will be possible. In other words, assuming correction, it may be possible to obtain better alignment accuracy with a single wavelength depending on the conditions.

As another example, it is assumed that there is a parameter that is normally not adopted because the measurement resolution is good but the stability is poor because abnormal values are likely to occur. At this time, assuming that anomaly detection can be performed during alignment, it is considered that by properly excluding anomalous values through correction, stable measurement can be performed with high resolution using this parameter.

From the above example, it is considered that better alignment can be performed by optimizing parameters and performing correction on the premise of correction, rather than performing correction under parameters selected without considering correction. Therefore, in the present embodiment, optical parameters are optimized in alignment measurement on the premise of correction.

The procedure for determining the optimum parameters and the alignment correction method will be described below. The basic imprinting procedure is the same as steps S101 to S108 shown in FIG.

In addition, model creation for calculating the amount of alignment error and the degree of certainty in steps S201 to S205 shown in FIG. 4 is basically performed in the same manner. In step S204 described above, data exposed under the same exposure conditions (optical parameters) are used, but here, data exposed under a plurality of optical parameters are used in the optimum optical parameter determination procedure. Multiple models may be trained separately, using separate models for each parameter. That is, the measurement device may include multiple models corresponding to multiple optical parameters. Alternatively, the metrology device may use one model common to all parameters. When an individual model is used, in step S202, parameter information is acquired in addition to alignment mark image information, and is used for learning and estimation, thereby improving estimation accuracy.

A method of calculating an alignment error amount (correction value) and performing correction in this embodiment will be described. The control unit 110 functions as a processing unit that determines optical parameters to be set in the alignment scope 106 for measurement based on feature amounts of image data. The processing unit applies an optical parameter candidate selected from the plurality of optical parameter candidates to the alignment scope 106 . Then, the processing unit acquires the image data of the measurement target imaged by the alignment scope 106, obtains the feature amount of the image data, and obtains the evaluation value for the feature amount. The processing unit does this for each of the plurality of optical parameter candidates. The processing unit then determines optical parameters based on each evaluation value.

A specific example of this processing will be described below with reference to FIG. FIG. 24 shows a flow chart of processing executed in step S104 (alignment) that can be used as an alternative to FIG. In step S2101, control unit 110 of imprint apparatus IMP selects an optical parameter to be used from among a plurality of optical parameters used for model learning, and selects an optical parameter corresponding to the selected optical parameter from among a plurality of models. Get the model. In step S2102, the control unit 110 acquires the image data captured by the alignment scope 106 in step S104, and extracts or calculates feature amounts at least in the non-measurement direction from the image data. The feature amount extraction or calculation method in step S302 is the same as the feature amount extraction or calculation method executed by the model generation device 1007 in step S202.

In step S2103, the control unit 110 uses the model acquired in step S2101 and the feature amount extracted or calculated in step S2102 to calculate the registration error amount (correction value), confidence, and evaluation value. For example, when Bayesian estimation is used for learning, the feature amount is input to the model acquired in step S2101, the expected value of the probability distribution output from the model is acquired as the registration error amount, and the variance of the probability distribution can be obtained as the certainty. The registration error amount can be used as a correction value. A method of calculating the evaluation value will be described later.

In step S2104, control unit 110 determines whether or not the calculation of the alignment error amount, certainty factor, and evaluation value for all optical parameters has been completed. If these calculations have not been completed, the process returns to step S2101 to calculate the registration error amount, confidence factor, and evaluation value for other optical parameters. If these calculations are finished, the process proceeds to step S2105.

In step S2105, control unit 110 compares the evaluation values of the respective optical parameters and selects the optical parameter with the best evaluation value. Then, based on the light intensity distribution in the measurement direction of the image data obtained by imaging with the alignment scope 106 using the selected optical parameter, the control unit 110 uses the position information of the measurement object in the measurement direction as temporary position information. Ask. This provisional position information is provisional position information that is obtained without considering the feature amount regarding the non-measurement direction of the image data.

In step S2106, the control unit 110 adjusts the measurement object in the measurement direction based on the temporary position information obtained in step S2105 and the correction value based on the feature amount of the image data regarding the non-measurement direction obtained in step S2103. Find the location information of . Specifically, the control unit 110 subtracts the correction value based on the feature amount of the image data in the non-measurement direction obtained in step S2103 from the provisional position information obtained in step S2105, thereby reducing the position of the measurement object in the measurement direction. location information can be obtained.

In the above embodiments, the example of determining the optimum optical parameter during imprinting and applying it to shot superimposition has been described, but the present invention is not limited to this. Optical parameters determined for a certain shot area may be used as optical parameters for other shot areas such as the next shot area and/or shot areas of another substrate for measurement and correction. In this case, it is not always necessary to perform the process of selecting the optimum optical parameter from the plurality of optical parameters as described above in the separate shot area or the shot area on the separate substrate.

Also, in the present embodiment, the optimum optical parameters are determined during imprinting, but the optimum optical parameters are determined after imprinting and used as the optical parameters for another shot region and/or another substrate shot region. You may In this case, there is no need to select the optimum optical parameter among multiple optical parameters during imprinting. For example, parameters may be switched for each shot area or substrate to obtain data, and optimum parameters may be determined from the data.

Finally, an example of evaluation values will be described. As a simple example, a model-determined confidence can be used as an evaluation value. As described above, the confidence factor is an index that indicates the degree of confidence that the estimated registration error amount is highly accurate. Based on this certainty, a parameter that is expected to be corrected with the highest accuracy can be determined as the optimum parameter.

As another example, the magnitude of the estimated registration error amount can be used as the evaluation value. If the estimated amount of registration error is small, it can be expected that the measurement has been performed with high accuracy. Therefore, the parameter with the smallest estimated amount of registration error can be determined as the optimum parameter.

In addition, when the optimum parameters are obtained after imprinting, the shot area after imprinting can be measured by the overlay inspection apparatus. An actual measurement value of the registration error amount of each parameter may be calculated, and the deviation between the estimated registration error amount and the actual measurement value may be calculated as the evaluation value. As a result, the parameter with the smallest deviation from the measured value can be determined as the optimum parameter.

According to the above embodiments, it is possible to reduce alignment measurement errors, control exclusion of abnormal alignment marks, and determine optimal parameters for exposure conditions, thereby improving alignment accuracy.

<Second embodiment>
FIG. 17 schematically shows the configuration of the exposure apparatus EP according to the second embodiment. The exposure apparatus EP can include an illumination device 1800 , a reticle stage RS on which a reticle 1031 (original) is placed, a projection optical system 1032 , a substrate stage WS on which a substrate 1003 is placed, a measuring device 1802 and an arithmetic processing unit 1400 . A reference plate 1039 is arranged on the substrate stage WS. The control unit 1803 is electrically connected to the illumination device 800, the reticle stage RS, the substrate stage WS, and the measuring device 1802 and controls them. In this embodiment, the control unit 1803 can also perform correction calculation and control of the measured value of the surface height position of the substrate 1003 by the measuring device 1802 . The control unit 1803 can be configured by, for example, a PLD such as an FPGA, an ASIC, a general-purpose computer in which a program is installed, or a combination of all or part of these. For example, the control unit 1803 can include a processor 1803a and a storage unit 1803b that stores programs and data.

The illumination device 1800 includes a light source unit 1810 and an illumination optical system 1801, and illuminates a reticle 1031 on which a circuit pattern for transfer is formed. The illumination optical system 1801 can have a function of uniformly illuminating the reticle 1031 and a polarized illumination function. Light source unit 1810 uses, for example, a laser. An ArF excimer laser with a wavelength of about 193 nm, a KrF excimer laser with a wavelength of about 248 nm, or the like can be used as the laser, but the type of light source is not limited to the excimer laser. For example, an F2 laser with a wavelength of about 157 nm or EUV (Extreme Ultraviolet) light with a wavelength of 20 nm or less may be used.

An illumination optical system 1801 is an optical system that illuminates a surface to be illuminated using a light beam emitted from a light source unit 1810. In this embodiment, the light beam is formed into an exposure slit having a predetermined shape that is optimal for exposure, and the reticle 1031 to illuminate the The illumination optical system 1801 can include lenses, mirrors, optical integrators, diaphragms, and the like. For example, from the light source side, the condenser lens, the fly-eye lens, the aperture stop, the condenser lens, the slit, and the imaging optical system are arranged in this order.

The reticle 1031 is made of, for example, quartz, has a circuit pattern to be transferred formed thereon, and is supported and driven by the reticle stage RS. Diffracted light emitted from the reticle 1031 passes through the projection optical system 1032 and is projected onto the substrate 1003 . The reticle 1031 and the substrate 1003 are arranged in an optically conjugate relationship. The pattern of the reticle 1031 is transferred onto the substrate 1003 by scanning the reticle 1031 and the substrate 1003 at the speed ratio of the reduction ratio. The exposure apparatus EP is provided with a reticle detection device (not shown) of oblique incidence system, and the position of the reticle 1031 can be detected by the reticle detection device and arranged at a predetermined position.

The reticle stage RS supports a reticle 1031 via a reticle chuck (not shown) and is connected to a moving mechanism (not shown). The movement mechanism is composed of a linear motor or the like, has a plurality of degrees of freedom (for example, three axes of X, Y, and θZ, preferably six axes of X, Y, Z, θX, θY, and θZ), and operates as a reticle stage. The reticle 1031 can be moved by driving RS.

The projection optical system 1032 has a function of forming an image of the light flux from the object plane onto the image plane. The projection optical system 1032 includes an optical system consisting of a plurality of lens elements, an optical system having a plurality of lens elements and at least one concave mirror (catadioptric optical system), a plurality of lens elements and at least one kinoform lens. An optical system having a diffractive optical element such as .

A substrate 1003 is an object to be processed, and a photoresist is applied on the substrate. Incidentally, in this embodiment, the substrate 1003 is an object to be detected from which the positions of the marks 1032 on the substrate 1003 are detected by the measuring device 1802 . Further, the substrate 1003 is an object to be detected whose surface position is detected by a surface position detection device (not shown). Note that the substrate 1003 may be a liquid crystal substrate or another object to be processed.

The substrate stage WS supports the substrate 1003 with a substrate chuck (not shown). Like the reticle stage RS, the substrate stage WS is composed of a linear motor or the like, and has a plurality of degrees of freedom (for example, three axes of X, Y, and θZ, preferably X, Y, Z, θX, θY, and θZ). 6 axes) to move the substrate 1003 . Also, the position of the reticle stage RS and the position of the substrate stage WS are monitored by, for example, a 6-axis laser interferometer 1081 or the like, and the two are driven at a constant speed ratio by the stage controller 1804 .

FIG. 18A shows a configuration example of the measuring device 1802. As shown in FIG. The measurement device 1802 can include an illumination system that illuminates the substrate 1003 with light emitted from the optical fiber 1061 and an imaging system that forms an image of the mark 1072 provided on the substrate 1003 . The illumination system can include illumination optics 1062 , 1063 , 1065 , illumination aperture stop 1064 , mirror M 2 , relay lens 1067 , polarization beam splitter 1068 , λ/4 plate 1070 and objective optics 1071 . The imaging system includes an objective optical system 1071 , a λ/4 plate 1070 , a detection aperture stop 1069 , a polarizing beam splitter 1068 and an imaging optical system 1074 . It is configured. The measuring device 1802 obtains the coordinate position of the mark 1072 based on the position information of the substrate stage WS measured by the laser interferometer 1081 and the signal waveform detected for the mark 1072 .

In the measurement device 1802 , light emitted from the optical fiber 1061 passes through illumination optical systems 1062 and 1063 and reaches an illumination aperture stop 1064 arranged at a position conjugate with the substrate 1003 . At this time, the beam diameter at the illumination aperture stop 1064 is sufficiently smaller than the beam diameter at the optical fiber 1061 . Light passing through the illumination aperture stop 1064 is guided to the polarization beam splitter 1068 through the illumination optical system 1065, the mirror M2, and the relay lens 1067. FIG. The polarization beam splitter 1068 transmits P-polarized light parallel to the Y direction and reflects S-polarized light parallel to the X direction. Therefore, the P-polarized light that has passed through the polarization beam splitter 1068 passes through the detection aperture stop 1069 and the λ/4 plate 1070, is converted into circularly polarized light, passes through the objective optical system 1071, and is formed on the substrate 1003. Koehler illumination is applied to the marked mark 1072 .

The light reflected, diffracted, and scattered by the mark 1072 passes through the objective optical system 1071 again, passes through the λ/4 plate 1070 , is converted from circularly polarized light into S-polarized light, and reaches the detection aperture stop 1069 . Here, the polarization state of the light reflected by the mark 1072 is circularly polarized in the opposite direction to the circularly polarized light irradiated on the mark 1072 . That is, when the polarization state of the light applied to the mark 1072 is right-handed circularly polarized light, the polarization state of the light reflected by the mark 1072 is left-handed circularly polarized light. Further, the detection aperture stop 1069 switches the numerical aperture of the reflected light from the mark 1072 by changing the aperture amount according to a command from the control unit 1803 . The light passing through the detection aperture stop 1069 is reflected by the polarizing beam splitter 1068 and then guided to the imaging sensor 75 via the imaging optical system 74 . Therefore, the polarizing beam splitter 1068 separates the optical path of the illumination light to the substrate 1003 from the optical path of the reflected light from the substrate 1003 , and an image of the mark 1072 provided on the substrate 1003 is formed on the imaging sensor 75 .

The optical fiber 1061 guides light from a light source 1050 composed of a separate halogen lamp, metal halide lamp, plasma light source, LED, or the like. Between the light source 1050 and the optical fiber 1061 is arranged a wavelength filter 51 that changes the wavelength to be transmitted. Wavelength filter 51 is configured to pass a wavelength band selected based on the quality of the image obtained when observing mark 1072 (for example, contrast and measurement deception, which will be described later).

The illumination aperture diaphragm 1064 has a switching mechanism (for example, a rotating mechanism) (not shown), and can change the shape of the transmitted light source distribution. For example, the illumination aperture stop 1064 can selectively change the size of the aperture (so-called illumination σ value) to enable modified illumination and the like. The shape of the opening can also change the quality of the image of the mark 1072, as will be described later. The illumination aperture stop 1064 has four openings 1055a, 1055b, 1055c, and 1055d as shown in FIG. 18B, for example. placed on the optical axis. Thus, by selecting an aperture to be placed on the optical axis, the lighting conditions can be changed. It should be noted that the mechanism for changing the illumination state is not limited to the one shown in this example, and those skilled in the art can easily conceive that similar effects can be obtained by various shapes and combinations. deaf.

As described above, by changing optical parameters such as wavelength, illumination aperture stop 1064, detection aperture stop 1069, etc., the quality of the image of mark 1072 can be changed. An example will be described with reference to FIGS. 19 and 20. FIG.

FIG. 19 shows an example of marks 1072 in the XY plane. The marks 1072 include two marks 1052X1 and 1052X2 aligned in the X direction and two marks 1052Y1 and 1052Y2 aligned in the Y direction. The measurement of the mark 1072 by the measuring device 1802 may cause a measurement error due to the effects of processes such as CMP, etching, uneven coating of resist, and the like. In particular, in the case of such a mark for performing measurement in two directions, there is a possibility that the mark shape asymmetry described later may occur not only in the measurement direction but also in directions other than the measurement direction.

FIG. 20A shows only one mark (for example, mark 1052X2) of the marks 1072 shown in FIG. 19 and shows a cross section (ZX cross section) of the mark. This mark has a stepped structure 1054 on which a resist 1053 is applied. The resist 1053 is generally a liquid resist applied to the substrate by spin coating. If the liquid resist 1053 is applied to the mark, it may be applied asymmetrically with respect to the stepped structure 1054 . If the resist is applied in such an asymmetrical state, the observed alignment image will also be asymmetrical, resulting in measurement fraud.

FIGS. 20B to 20E show examples of signal waveforms of mark images. Here, the horizontal axis represents the position in the X direction, and the vertical axis represents the signal intensity. FIG. 20B shows signal waveforms when the optical parameters (wavelength of illumination light, illumination σ value) are set to predetermined values. In this signal waveform, the signal intensity difference at the mark edge is small and the contrast is low. When the contrast is low like this, the measurement accuracy is lowered, so the control unit 1803 changes the optical parameters so as to enhance the contrast. FIG. 20(C) shows the signal waveform of the mark image obtained by changing the wavelength of the illumination light among the optical parameters when the signal waveform of FIG. 20(B) is obtained. Although the signal intensity difference at the mark edge is amplified in this signal waveform, the signal intensity between the left and right mark edges becomes asymmetrical because the resist 1053 is applied in an asymmetrical state, resulting in a measurement error.

FIG. 20D shows the signal waveform of the mark image acquired by changing the illumination σ value to a small value among the optical parameters when the signal waveform of FIG. 20C is obtained. By decreasing the illumination σ value, the contrast is enhanced by amplifying the signal intensity difference at the mark edge while maintaining the symmetry of the signal intensity between the left and right mark edges. This phenomenon can be adjusted by changing the focus position of the mark 1072 . Specifically, by changing the focus position of the mark 1072, it is possible to change the amount of contrast enhancement, the position in the X direction where the enhancement occurs, and the like. Therefore, the measurement device 1802 can perform measurement at an optimum focus position according to measurement conditions (contrast, measurement deception, etc.).

FIG. 20(E) shows the signal waveform of the mark image acquired by changing the wavelength of the illumination light among the optical parameters when the signal waveform of FIG. 20(B) is obtained to a larger value. Since the resist 1053 is applied on the mark, the interference conditions between the concave portion and the convex portion in the stepped structure 1054 are different. Relationships change. In contrast to the simple stepped mark of this example, in marks used in actual semiconductor processes, changes in contrast caused by changing the wavelength of illumination light are often accompanied by more pronounced changes. In any case, by changing the wavelength according to the mark, the appearance of the obtained mark image will be different. In step S2103 (see FIG. 24), the control unit 1803 uses the feature amount of this asymmetric signal to calculate the correction value, confidence factor, and optimum optical parameter for each alignment measurement.

FIG. 21A is a plan view (XY plan view) of a mark coated with resist, focusing only on mark 1052X1 in mark 1072 shown in FIG. The mark has a stepped structure 1054 on which a resist 1053 is applied. The resist 1053 is generally a liquid resist applied to the substrate by spin coating. When the liquid resist 1053 is applied to the mark, it may be applied asymmetrically to the stepped structure 1054 in the non-measurement direction (the Y direction in this figure) as well as in the measurement direction. If the resist is applied in such an asymmetrical state, the edge shape in the non-measurement direction of the observed alignment image will also be asymmetrical. The degree of this asymmetry is similar to the degree of asymmetry in the measurement direction, and corresponds to the magnitude of the measurement error in the measurement direction. In this way, the intensity of the signal obtained by integrating the non-measurement direction along the horizontal axis (X-axis in this figure) is constant when the resist is uniformly applied, but when the resist is applied non-uniformly. the change will be large. Therefore, it can be said that the intensity of the signal obtained by integrating in the non-measurement direction is a feature quantity that can more clearly determine the degree of non-uniformity in the coating state of the resist, that is, the degree of asymmetry.

FIGS. 21B to 21E show examples of waveforms of signal intensities obtained by integrating the mark image in the non-measurement direction. Here, the vertical axis represents the position in the Y direction, and the horizontal axis represents the signal intensity. FIG. 21B shows signal waveforms when the optical parameters (wavelength of illumination light, illumination σ value) are set to predetermined values. This signal waveform has a low signal intensity (contrast). When the signal intensity is low in this way, the measurement accuracy is degraded, so the control unit 1803 changes the optical parameters so as to increase the signal intensity. FIG. 21(C) shows the signal waveform of the mark image obtained by changing the wavelength of the illumination light among the optical parameters when the signal waveform of FIG. 21(B) is obtained. Although the signal intensity is amplified in this signal waveform, the change in the signal intensity due to the asymmetric coating state of the resist is emphasized, and a measurement error occurs in this state.

FIG. 21D shows the signal waveform of the mark image acquired by changing the illumination σ value to a small value among the optical parameters when the signal waveform of FIG. 20C is obtained. By decreasing the illumination σ value, the signal intensity is emphasized at the mark edge portion, and the contrast can be increased. This phenomenon can be adjusted by changing the focus position of the mark 1072 . Specifically, by changing the focus position of the mark 1072, it is possible to change the amount of contrast enhancement, the position in the X direction where the enhancement occurs, and the like. Therefore, the measurement device 1802 can perform measurement at an optimum focus position according to measurement conditions (contrast, measurement deception, etc.). When the focus position is properly adjusted, the influence of the asymmetry of the resist is reduced, the change in the obtained signal intensity is reduced, and a state suitable for alignment can be obtained.

FIG. 21(E) shows the signal waveform of the mark image acquired by changing the wavelength of the illumination light among the optical parameters when the signal waveform of FIG. 21(B) is obtained to a large value. Since the resist 1053 is applied on the mark, the interference conditions between the concave portion and the convex portion are different, and if the phase difference is large, the relationship between the signal intensity of the concave portion and the signal intensity of the convex portion changes. . The signal waveform of FIG. 21(E) has a slightly larger change than the signal waveform of FIG. 21(D). In marks used in the actual semiconductor process, not the simple step marks in this example, the change in the signal in the non-measurement direction caused by changing the wavelength of the illumination light is often accompanied by a more pronounced change. . In any case, by changing the wavelength according to the mark, the appearance of the obtained mark image will be different. In step S2103 (see FIG. 24), the control unit 1803 calculates the correction value, confidence factor, and optimum optical parameter for each alignment measurement using the feature amount of the signal intensity in the non-measurement direction.

<Embodiment of article manufacturing method>
INDUSTRIAL APPLICABILITY The article manufacturing method according to the embodiment of the present invention is suitable for manufacturing articles such as microdevices such as semiconductor devices and elements having fine structures. The method for manufacturing an article according to the present embodiment includes a step of forming a latent image pattern on a photosensitive agent applied to a substrate using the above exposure apparatus (a step of exposing the substrate), and and developing the substrate. In addition, such manufacturing methods include other well-known steps (oxidation, deposition, deposition, doping, planarization, etching, resist stripping, dicing, bonding, packaging, etc.). The article manufacturing method of the present embodiment is advantageous in at least one of article performance, quality, productivity, and production cost compared to conventional methods.

A pattern of a cured product formed using an imprint apparatus is used permanently on at least a part of various articles, or temporarily used when manufacturing various articles. Articles are electric circuit elements, optical elements, MEMS, recording elements, sensors, molds, or the like. Examples of electric circuit elements include volatile or nonvolatile semiconductor memories such as DRAM, SRAM, flash memory, and MRAM, and semiconductor elements such as LSI, CCD, image sensors, and FPGA. Examples of the mold include imprint molds and the like.

The pattern of the cured product is used as it is or temporarily used as a resist mask as at least a part of the article. After etching, ion implantation, or the like in the substrate processing step, the resist mask is removed.

Next, an article manufacturing method will be described. In step SA of FIG. 25, a substrate 1z such as a silicon substrate having a surface to be processed 2z such as an insulator is prepared. to give Here, a state is shown in which a plurality of droplet-like imprint materials 3z are applied onto the substrate.

In step SB of FIG. 25, the imprint mold 4z is opposed to the imprint material 3z on the substrate with the side on which the uneven pattern is formed. In step SC of FIG. 25, the substrate 1z provided with the imprint material 3z and the mold 4z are brought into contact with each other and pressure is applied. The imprint material 3z is filled in the gap between the mold 4z and the workpiece 2z. In this state, when light is irradiated through the mold 4z as energy for curing, the imprint material 3z is cured.

In step SD of FIG. 25, after the imprint material 3z is cured, the mold 4z and the substrate 1z are separated to form a pattern of the cured imprint material 3z on the substrate 1z. The pattern of this cured product has a shape in which the concave portions of the mold correspond to the convex portions of the cured product, and the convex portions of the mold correspond to the concave portions of the cured product. It will be done.

In step SE of FIG. 25, when etching is performed using the pattern of the cured product as an anti-etching mask, portions of the surface of the workpiece 2z where the cured product is absent or remains thinly are removed to form grooves 5z. In step SF of FIG. 25, by removing the pattern of the cured product, an article having grooves 5z formed on the surface of the workpiece 2z can be obtained. Although the pattern of the cured product is removed here, it may be used as an interlayer insulating film included in a semiconductor element or the like, that is, as a constituent member of an article, without being removed after processing.

(Other embodiments)
The present invention supplies a program that implements one or more functions of the above-described embodiments to a system or apparatus via a network or a storage medium, and one or more processors in the computer of the system or apparatus reads and executes the program. It can also be realized by processing to It can also be implemented by a circuit (for example, ASIC) that implements one or more functions.

The invention is not limited to the embodiments described above, and various modifications and variations are possible without departing from the spirit and scope of the invention. Accordingly, the claims are appended to make public the scope of the invention.

IMP: imprint apparatus, S: substrate, M: mold, 110: control unit, 102: substrate holding unit, 105: substrate driving mechanism, 121: mold holding unit, 122: mold driving mechanism, 1007: model generation device

Claims (20)

A measuring device for measuring position information of a measurement object,
a scope that captures an image of the object to be measured and generates image data;
a processing unit that obtains a feature amount of the image data and determines an optical parameter to be set in the scope for the measurement based on the feature amount;
A measuring device comprising: The processing unit is
applying an optical parameter candidate selected from among a plurality of optical parameter candidates to the scope to obtain image data of the object to be measured captured by the scope, obtaining a feature amount of the image data, and obtaining the feature amount obtaining an evaluation value for each of the plurality of optical parameter candidates;
determining the optical parameter based on each evaluation value;
The measuring device according to claim 1, characterized by: 3. The measuring apparatus according to claim 2, wherein the processing unit determines an optical parameter candidate corresponding to the best evaluation value as the optical parameter. The processing unit is
Obtaining temporary position information of the measurement object based on the image data;
determining the position information by correcting the provisional position information based on a correction value based on the feature quantity;
4. The measuring device according to claim 2 or 3, characterized in that: 5. The measuring device according to claim 4, wherein the processing unit obtains the size of the feature quantity as the evaluation value. 5. The processing unit according to claim 4, wherein the processing unit obtains, as the evaluation value, a difference between the position information of the measurement object measured by an external inspection device and the position information determined by the processing unit. measuring device. 7. The measuring apparatus according to any one of claims 4 to 6, further comprising a model for obtaining said correction value based on said feature quantity. 8. The measuring device according to claim 7, further comprising a machine learning unit that generates the model by machine learning. The machine learning unit uses the feature amount as input data for the model, and uses the difference between the position information of the measurement object measured by an external inspection device and the position information determined by the processing unit as teacher data. 9. The measuring device according to claim 8, wherein learning is performed. 5. The measuring apparatus according to claim 4, further comprising a model for obtaining said correction value and a degree of certainty representing certainty and/or reliability of said correction value based on said feature quantity. 11. The measuring device according to claim 10, wherein the processing unit obtains the evaluation value based on the degree of certainty. 12. The measuring device according to claim 10, further comprising a machine learning unit that generates the model by machine learning. The machine learning unit uses the feature amount as input data for the model, and uses the difference between the position information of the measurement object measured by an external inspection device and the position information determined by the processing unit as teacher data. 13. The measuring device according to claim 12, wherein learning is performed. The machine learning is performed using Bayesian inference,
The processing unit obtains the expected value of the probability distribution output from the model as the correction value, and obtains the variance of the probability distribution as the confidence factor.
14. The measuring device according to claim 13, characterized by: having multiple models corresponding to each of the multiple optical parameters,
The processing unit acquires a model corresponding to an optical parameter selected from the plurality of optical parameters, from among the plurality of models.
The measuring device according to any one of claims 7 to 14, characterized in that: Based on temporary position information of the measurement object in a first direction obtained from the image data and a correction value based on a feature amount of the image data in a second direction different from the first direction, the processing unit , determining the position information of the object to be measured in the first direction. 17. The measuring device according to claim 16, wherein said second direction is a direction orthogonal to said first direction. 18. The measuring apparatus according to any one of claims 1 to 17, wherein the optical parameter includes at least one of a wavelength of illumination light generated by the scope, an illumination σ value, and NA. A lithographic apparatus for transferring a pattern of an original onto a substrate,
The measuring device according to any one of claims 1 to 18 configured to measure the relative position between the substrate and the original,
aligning the substrate and the original based on the output of the measurement device;
A lithographic apparatus characterized by: a transfer step of transferring a pattern onto a substrate using a lithographic apparatus according to claim 19;
a processing step of processing the substrate that has undergone the transfer step;
A method for manufacturing an article, comprising obtaining an article from the substrate that has undergone the treatment step.

Images