DETAILED ACTION
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 06/29/2026 has been entered.
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
This action is in response to amendments and remarks filed on 06/29/2026, which have been entered. In the current amendments, claims 1, 3, 7, 9-10, 13, 15, and 19 are amended. Claims 1-20 are pending and have been examined.
In response to amendments and remarks filed on 06/29/2026, the 35 U.S.C. 103 prior art rejections to claims 3-4, 7-8, 10, 15, and 19-20 made in the previous office action are withdrawn.
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 06/29/2026 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claims 1-2, 5-6, 9, 11-14, and 16-18 are rejected under 35 U.S.C. 103 as being unpatentable over Pathangi Sriraman (US 2019/0214223 A1) in view of Brauer (US 2021/0272273 A1) and further in view of Feng et al. (US 2021/0035833 A1).
Regarding Claim 1,
Pathangi Sriraman teaches a method (Fig. 1; Fig. 2; Fig. 3; [0012]: "a method of compensating for scanning electron microscope beam distortion-induced metrology error is provided. A metrology area image of a wafer and a design clip from a storage medium are received at a processor. The design clip is applied to the metrology area image using the processor thereby obtaining a synthesized image. One or more process change variations are suppressed and one or more tool distortions are enhanced. Metrology operations can be performed on the synthesized image" teaches a method), comprising:
receiving first data indicating a plurality of dimensions of a manufactured device based on one or more measurements of the manufactured device (Fig. 1; Fig. 2; Fig. 3; [0012]-[0013]: "a method of compensating for scanning electron microscope beam distortion-induced metrology error is provided. A metrology area image of a wafer and a design clip from a storage medium are received at a processor. The design clip is applied to the metrology area image using the processor thereby obtaining a synthesized image. One or more process change variations are suppressed and one or more tool distortions are enhanced. Metrology operations can be performed on the synthesized image … The design clip can be a design for the wafer or a synthesized design clip. The synthesized design clip can be generated using the process change variations with a machine algorithm and communicating the synthesized design clip to the storage medium. Generating the synthesized design clip can further include receiving images and data from one or more process modulated wafers at a machine learning module and learning the process change variations at the machine learning module using the machine algorithm" teaches obtaining a synthesized design clip (first data) based on a metrology images and data (dimensions) of from a process-modulated semiconductor wafer (manufactured device). [0005]: "Metrology processes can be used to measure one or more characteristics of wafers such that the performance of a process can be determined from the one or more characteristics. For example, metrology processes can measure a dimension (e.g., line width, thickness, etc.) of features formed on the wafers during the process" teaches that the metrology data includes dimensions of the wafer based on measurements of the manufactured semiconductor wafer (manufactured device)).
Pathangi Sriraman does not appear to explicitly teach providing the first data to a trained machine learning model; receiving a synthetic microscopy image generated from the trained machine learning model, the synthetic microscopy image comprising a synthetic image resembling a realistic microscopy image of a cross section depicting an internal structure of the manufactured device, wherein the synthetic microscopy image is generated in view of the first data; and performing at least one of (i) outputting the synthetic microscopy image to a display or (ii) performing one or more operations on the synthetic microscopy image.
However, Brauer teaches providing the first data to a trained machine learning model (Fig. 2; [0096]: "As shown in step 210, the one or more computer subsystems may then process the design clips generated by step 208 through the previously trained GAN. The trained GAN may then output artificially generated images, as shown in step 212. Among other things, the images output by the GAN in step 212 may be used as input data for training other learning algorithms, as shown in step 214, which may be performed as described further herein" teaches that the design clips (first data) are input (provided) to a trained GAN (machine learning model));
receiving a synthetic microscopy image generated from the trained machine learning model (Fig. 2; [0096]: "As shown in step 210, the one or more computer subsystems may then process the design clips generated by step 208 through the previously trained GAN. The trained GAN may then output artificially generated images, as shown in step 212. Among other things, the images output by the GAN in step 212 may be used as input data for training other learning algorithms, as shown in step 214, which may be performed as described further herein" teaches that the trained GAN (machine learning model) uses the design clips (first data) to output artificially generated images. [0110]: "In another embodiment, the simulated image is a reference image … This artificial image may then be used as a reference image for defect inspection. In this manner, the embodiments described herein can be used to generate an artificial optical (or other) reference image to perform die-to-database inspections. One novel feature of the embodiments described herein is therefore that they provide a method for improved optical defect inspection using a GAN or cGAN for the artificial creation of optical reference patch images which can be used for optical die-to-database inspections. Another novel feature of the embodiments described herein is that they provide a method for improved optical defect inspection using a GAN or cGAN for an unsupervised DL method for optical reference image generation" teaches that the generated artificial images are artificial optical reference patch images (synthetic microscopy images)); and
performing at least one of (i) outputting the synthetic microscopy image to a display or (ii) performing one or more operations on the synthetic microscopy image (Fig. 2; [0113]-[0014]: "The images, model or network, and information therefor may be stored with any of the other results described herein and may be stored in any manner known in the art … After the information has been stored, the information can be accessed in the storage medium and used by any of the method or system embodiments described herein, formatted for display to a user, used by another software module, method, or system, etc. For example, the embodiments described herein may generate an inspection recipe as described above. That inspection recipe may then be stored and used by the system or method (or another system or method) to inspect the specimen or other specimens to thereby generate information (e.g., defect information) for the specimen or other specimens … Results and information generated by performing the inspection on the specimen or other specimens of the same type may be used in a variety of manners by the embodiments described herein and/or other systems and methods. Such functions include, but are not limited to, altering a process such as a fabrication process or step that was or will be performed on the inspected specimen or another specimen in a feedback or feedforward manner. For example, the computer subsystem(s) described herein may be configured to determine one or more changes to a process that was performed on a specimen inspected as described herein and/or a process that will be performed on the specimen based on the detected defect(s). The changes to the process may include any suitable changes to one or more parameters of the process" teaches that the output image (synthetic microscopy image) may be stored and then formatted for display to a user or for performing operations on the output image (synthetic microscopy image)).
Pathangi Sriraman and Brauer are analogous to the claimed invention because they are directed to using machine learning models for assessing manufactured devices.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate providing the first data to a trained machine learning model; receiving a synthetic microscopy image generated from the trained machine learning model, and performing at least one of (i) outputting the synthetic microscopy image to a display or (ii) performing one or more operations on the synthetic microscopy image as taught by Brauer to the disclosed invention of Pathangi Sriraman.
One of ordinary skill in the art would have been motivated to make this modification to "provide a method for improved optical defect inspection using a GAN or cGAN for the artificial creation of optical patch images" (Brauer [0081]).
Pathangi Sriraman in view of Brauer does not appear to explicitly teach the synthetic microscopy image comprising a synthetic image resembling a realistic microscopy image of a cross section depicting an internal structure of the manufactured device, wherein the synthetic microscopy image is generated in view of the first data.
However, Feng et al. teaches the synthetic microscopy image comprising a synthetic image resembling a realistic microscopy image of a cross section depicting an internal structure of the manufactured device, wherein the synthetic microscopy image is generated in view of the first data (Fig. 4; [0086]: "FIG. 4 presents a block diagram of an example metrology system 401. As illustrated, the system includes an optical metrology tool 407 configured to probe substrates such as a substrate 403 having features 405. Optical metrology tool 407 collects optical information from features 405 and generates an optical metrology output (e.g., reflectance intensity versus wavelength or optical critical dimension information). A metrology machine learning model 409 receives the optical metrology output and generates geometric information about the features; e.g., feature profiles, CDs, contours, etc. See output 411. In certain embodiments, the machine learning model 409 is implemented as a neural network" teaches that the neural network generates a feature profile (synthetic image) of the substrate (manufactured device) based on the metrology data (first data) measured from the substrate. [0064]-[0067]: "measuring features of the processed substrate produces profile coordinates. In such embodiments, measuring features of the processed substrate may include performing microscopy (e.g., SEM, x-SEM, TEM, STEM, CD-SEM, CD-SAXS, REM, AFM). Such techniques may be employed to generate profile coordinates directly or indirectly such as by way of a set of geometric profile parameters characterizing a geometry of the feature in the processed substrate (e.g., critical dimension, side wall angles, depth, etc.) ... the geometric information of a feature may be a profile (viewed as a cross-section of the substrate in an x-z plane)" teaches that the generated profile (synthetic image) is a cross-section image of the substrate based on a microscopy image (synthetic image resembling a realistic microscopy image of a cross section). [0020]: "Current methods of evaluating semiconductor wafers and other samples used in semiconductor process development employ techniques such as x-SEM (x-ray Scanning Electron Microscopy), STEM (Scanning Transmission Electron Microscopy), and CD-SAXS (Critical Dimension Small Angle X-ray Scattering) for cross-sectional image capture" teaches that the cross-sectional image resembles a microscopy image generated by scanning electron microscope techniques (i.e. SEM and x-SEM cross section analysis exposes the internal structure, layer thickness, and subsurface defects of the semiconductor by cutting it to view the interior side profile using a scanning electron microscope)).
Pathangi Sriraman, Brauer, and Feng et al. are analogous to the claimed invention because they are directed to using machine learning models for assessing manufactured devices.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate the synthetic microscopy image comprising a synthetic image resembling a realistic microscopy image of a cross section depicting an internal structure of the manufactured device, wherein the synthetic microscopy image is generated in view of the first data as taught by Feng et al. to the disclosed invention of Pathangi Sriraman in view of Brauer.
One of ordinary skill in the art would have been motivated to make this modification to "drastically reduce the costs associated with performing destructive testing such as STEM and x-SEM" and "bring in better coupon-to-coupon consistency as random errors & LER/LWR (line edge roughness and line width roughness) on CD/contours/profiles are spatially averaged out by the optical signals" (Feng et al. [0055]-[0056]).
Regarding Claim 2,
Pathangi Sriraman in view of Brauer and further in view of Feng et al. teaches the method of claim 1.
In addition, Pathangi Sriraman further teaches further comprising: receiving second data indicating a subset of the plurality of dimensions indicated by the first data (Fig. 1; Fig. 2; Fig. 3; [0047]: "FIG. 3 shows an embodiment of a method 300 of generating and storing a synthesized design clip. Images or images and data from one or more process modulated wafers 301 are received at a machine learning module. The data can include, for example, metadata related to the alignment process and/or rendering. The images or images and data can be inputted into a machine or deep learning algorithm at the machine learning module at step 302 to learn process change variations. These process change variations are used in step 303 to generate a synthesized design clip, which in step 304 is communicated with and/or stored in an electronic data storage medium 305" teaches that the synthesized design clip (first data) is generated based on images (second data that is a subset of the first data) for process-modulated wafers (manufacturing device). Fig. 1; Fig. 2; Fig. 3; [0005]-[0006]: "Metrology processes can be used to measure one or more characteristics of wafers such that the performance of a process can be determined from the one or more characteristics. For example, metrology processes can measure a dimension (e.g., line width, thickness, etc.) of features formed on the wafers during the process. In addition, if the one or more characteristics of the wafers are unacceptable (e.g., out of a predetermined range for the characteristic(s)), the measurements of the one or more characteristics of the wafers may be used to alter one or more parameters of the process such that additional wafers manufactured by the process have acceptable characteristic(s). … With increasing pattern density and complexity in advanced technology nodes, automated in-line metrology operations that use a scanning electron microscope (SEM) are claiming an increased share in advanced process control" teaches that the images are metrology images including dimensions of the wafer);
receiving third data indicating one or more rules of design of the manufactured device (Fig. 1; Fig. 2; Fig. 3; [0047]: "FIG. 3 shows an embodiment of a method 300 of generating and storing a synthesized design clip. Images or images and data from one or more process modulated wafers 301 are received at a machine learning module. The data can include, for example, metadata related to the alignment process and/or rendering. The images or images and data can be inputted into a machine or deep learning algorithm at the machine learning module at step 302 to learn process change variations. These process change variations are used in step 303 to generate a synthesized design clip, which in step 304 is communicated with and/or stored in an electronic data storage medium 305" teaches that the synthesized design clip (first data) is generated based on data (third data) for process-modulated wafers (manufacturing device) including metadata related to the alignment process and/or rendering (rules of design)); and
providing the second data and the third data to a model configured to generate the first data in view of the second data and the third data (Fig. 1; Fig. 2; Fig. 3; [0047]: "FIG. 3 shows an embodiment of a method 300 of generating and storing a synthesized design clip. Images or images and data from one or more process modulated wafers 301 are received at a machine learning module. The data can include, for example, metadata related to the alignment process and/or rendering. The images or images and data can be inputted into a machine or deep learning algorithm at the machine learning module at step 302 to learn process change variations. These process change variations are used in step 303 to generate a synthesized design clip, which in step 304 is communicated with and/or stored in an electronic data storage medium 305" teaches that the image data (second data) and the metadata (third data) of a wafer are provided to a machine learning model to generate a synthesized design clip (first data)).
Regarding Claim 5,
Pathangi Sriraman in view of Brauer and further in view of Feng et al. teaches the method of claim 1.
In addition, Brauer further teaches wherein the trained machine learning model comprises a generator of a generative adversarial network (Fig. 2; [0096]: "As shown in step 210, the one or more computer subsystems may then process the design clips generated by step 208 through the previously trained GAN. The trained GAN may then output artificially generated images, as shown in step 212. Among other things, the images output by the GAN in step 212 may be used as input data for training other learning algorithms, as shown in step 214, which may be performed as described further herein" teaches that the trained model comprises a GAN (generative adversarial network). Fig .5; [0075]: "An example of a generator that may be included in embodiments of a GAN configured as described herein is shown in FIG. 5" teaches that the GAN comprises a generator model).
Pathangi Sriraman, Brauer, and Feng et al. are analogous to the claimed invention because they are directed to using machine learning models for assessing manufactured devices.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate wherein the trained machine learning model comprises a generator of a generative adversarial network as taught by Brauer to the disclosed invention of Pathangi Sriraman in view of Feng et al.
One of ordinary skill in the art would have been motivated to make this modification to "provide a method for improved optical defect inspection using a GAN or cGAN for the artificial creation of optical patch images" (Brauer [0081]).
Regarding Claim 6,
Pathangi Sriraman in view of Brauer and further in view of Feng et al. teaches the method of claim 1.
In addition, Feng et al. further teaches wherein the synthetic microscopy image comprises a synthetic image of a cross section of the manufactured device, wherein the synthetic image resembles a microscopy image generated by an electron-based microscopy technique (Fig. 4; [0086]: "FIG. 4 presents a block diagram of an example metrology system 401. As illustrated, the system includes an optical metrology tool 407 configured to probe substrates such as a substrate 403 having features 405. Optical metrology tool 407 collects optical information from features 405 and generates an optical metrology output (e.g., reflectance intensity versus wavelength or optical critical dimension information). A metrology machine learning model 409 receives the optical metrology output and generates geometric information about the features; e.g., feature profiles, CDs, contours, etc. See output 411. In certain embodiments, the machine learning model 409 is implemented as a neural network" teaches that the neural network generates a feature profile (synthetic image) of the substrate *manufactured device) based on the metrology data (first data) measured from the substrate. [0067]: "the geometric information of a feature may be a profile (viewed as a cross-section of the substrate in an x-z plane)" teaches that the generated profile (synthetic image) is a cross-section image of the substrate. [0020]: "Current methods of evaluating semiconductor wafers and other samples used in semiconductor process development employ techniques such as x-SEM (x-ray Scanning Electron Microscopy), STEM (Scanning Transmission Electron Microscopy), and CD-SAXS (Critical Dimension Small Angle X-ray Scattering) for cross-sectional image capture" teaches that the cross-sectional image resembles a microscopy image generated by scanning electron microscope techniques (electron-based microscopy technique)).
Pathangi Sriraman, Brauer, and Feng et al. are analogous to the claimed invention because they are directed to using machine learning models for assessing manufactured devices.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate wherein the synthetic microscopy image comprises a synthetic image of a cross section of the manufactured device, wherein the synthetic image resembles a microscopy image generated by an electron-based microscopy technique as taught by Feng et al. to the disclosed invention of Pathangi Sriraman in view of Brauer.
One of ordinary skill in the art would have been motivated to make this modification to "drastically reduce the costs associated with performing destructive testing such as STEM and x-SEM" and "bring in better coupon-to-coupon consistency as random errors & LER/LWR (line edge roughness and line width roughness) on CD/contours/profiles are spatially averaged out by the optical signals" (Feng et al. [0055]-[0056]).
Regarding Claim 9,
Pathangi Sriraman teaches a method (Fig. 1; Fig. 2; Fig. 3; [0012]: "a method of compensating for scanning electron microscope beam distortion-induced metrology error is provided. A metrology area image of a wafer and a design clip from a storage medium are received at a processor. The design clip is applied to the metrology area image using the processor thereby obtaining a synthesized image. One or more process change variations are suppressed and one or more tool distortions are enhanced. Metrology operations can be performed on the synthesized image" teaches a method), comprising:
receiving a plurality of microscopy images, wherein each microscopy image of the plurality of microscopy images is of one of a plurality of manufactured devices (Fig. 1; Fig. 2; Fig. 3; [0012]-[0014]: " a method of compensating for scanning electron microscope beam distortion-induced metrology error is provided. A metrology area image of a wafer and a design clip from a storage medium are received at a processor. The design clip is applied to the metrology area image using the processor thereby obtaining a synthesized image. One or more process change variations are suppressed and one or more tool distortions are enhanced. Metrology operations can be performed on the synthesized image … The design clip can be a design for the wafer or a synthesized design clip. The synthesized design clip can be generated using the process change variations with a machine algorithm and communicating the synthesized design clip to the storage medium. Generating the synthesized design clip can further include receiving images and data from one or more process modulated wafers at a machine learning module and learning the process change variations at the machine learning module using the machine algorithm. The machine algorithm may be a deep learning algorithm … The metrology area image can be a scanning electron microscope image or an average of a plurality of scanning electron microscope images" teaches obtaining a plurality of scanning electron microscope images as metrology area images (microscopy) for semiconductor wafers (manufactured devices). [0005]: "Metrology processes can be used to measure one or more characteristics of wafers such that the performance of a process can be determined from the one or more characteristics. For example, metrology processes can measure a dimension (e.g., line width, thickness, etc.) of features formed on the wafers during the process" teaches that the metrology data includes dimensions of the wafer);
receiving first data indicating a plurality of dimensions of the plurality of manufactured devices based on one or more measurements of each of the plurality of manufactured devices (Fig. 1; Fig. 2; Fig. 3; [0012]-[0013]: "a method of compensating for scanning electron microscope beam distortion-induced metrology error is provided. A metrology area image of a wafer and a design clip from a storage medium are received at a processor. The design clip is applied to the metrology area image using the processor thereby obtaining a synthesized image. One or more process change variations are suppressed and one or more tool distortions are enhanced. Metrology operations can be performed on the synthesized image … The design clip can be a design for the wafer or a synthesized design clip. The synthesized design clip can be generated using the process change variations with a machine algorithm and communicating the synthesized design clip to the storage medium. Generating the synthesized design clip can further include receiving images and data from one or more process modulated wafers at a machine learning module and learning the process change variations at the machine learning module using the machine algorithm" teaches obtaining a synthesized design clip (first data) based on a metrology images and data (dimensions) of from a process-modulated semiconductor wafer (manufactured device). [0005]: "Metrology processes can be used to measure one or more characteristics of wafers such that the performance of a process can be determined from the one or more characteristics. For example, metrology processes can measure a dimension (e.g., line width, thickness, etc.) of features formed on the wafers during the process" teaches that the metrology data includes dimensions of the wafer based on measurements of the manufactured semiconductor wafers (manufactured devices)).
Pathangi Sriraman does not appear to explicitly teach training a machine learning model to generate synthetic microscopy images resembling realistic microscopy images depicting cross sections of an internal structure of a manufactured device using the first data and the plurality of microscopy images, wherein training the machine learning model comprises providing the first data to the machine learning model as training input, and providing the plurality of microscopy images to the machine learning model as target output.
However, Brauer teaches training a machine learning model to generate synthetic microscopy images … using the first data and the plurality of microscopy images, wherein training the machine learning model comprises … providing the plurality of microscopy images to the machine learning model as target output (Fig. 2; [0064]: "The image and design clip pairs generated by step 202 may then be used as a training set, with the design clips designated as the training inputs and corresponding images designed as the training outputs, in step 204 in which the GAN (i.e., the generator-discriminator network (NW)) is trained using design and optical images. The training of the GAN may be performed as described further herein. In some embodiments, the computer subsystem(s) may be configured to perform steps 200, 202, and 204" teaches that the optical images (microscopy images) and design clips (first data) are used to train the GAN (machine learning model) for generating artificial (synthetic images), with the design clips (first data) being provided as training input, and the corresponding images (microscopy images) being provided as training outputs).
Pathangi Sriraman and Brauer are analogous to the claimed invention because they are directed to using machine learning models for assessing manufactured devices.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate training a machine learning model to generate synthetic microscopy images … using the first data and the plurality of microscopy images, wherein training the machine learning model comprises … providing the plurality of microscopy images to the machine learning model as target output as taught by Brauer to the disclosed invention of Pathangi Sriraman.
One of ordinary skill in the art would have been motivated to make this modification to "provide a method for improved optical defect inspection using a GAN or cGAN for the artificial creation of optical patch images" (Brauer [0081]).
Pathangi Sriraman in view of Brauer does not appear to explicitly teach generate synthetic microscopy images resembling realistic microscopy images depicting cross sections of an internal structure of a manufactured device … wherein training the machine learning model comprises providing the first data to the machine learning model as training input.
However, Feng et al. teaches generate synthetic microscopy images resembling realistic microscopy images depicting cross sections of an internal structure of a manufactured device … wherein training the machine learning model comprises providing the first data to the machine learning model as training input (Fig. 1; [0046]: "At this point, information for a full training set is available, and, as illustrated at operation 111, the process trains a metrology model using optical metrology signals produced in 105 and, optionally, 107 together with the associated feature profiles, contours, CDs, etc. produced in 109. The resulting model is tested and/or validated as indicated at operation 113. The resulting model may be installed in a metrology tool or system" teaches providing the measured metrology data (first data) to the metrology model (machine learning model) as training input. Fig. 4; [0086]: "FIG. 4 presents a block diagram of an example metrology system 401. As illustrated, the system includes an optical metrology tool 407 configured to probe substrates such as a substrate 403 having features 405. Optical metrology tool 407 collects optical information from features 405 and generates an optical metrology output (e.g., reflectance intensity versus wavelength or optical critical dimension information). A metrology machine learning model 409 receives the optical metrology output and generates geometric information about the features; e.g., feature profiles, CDs, contours, etc. See output 411. In certain embodiments, the machine learning model 409 is implemented as a neural network" teaches that the neural network generates a feature profile (synthetic image) of the substrate (manufactured device) based on the metrology data (first data) measured from the substrate. [0064]-[0067]: "measuring features of the processed substrate produces profile coordinates. In such embodiments, measuring features of the processed substrate may include performing microscopy (e.g., SEM, x-SEM, TEM, STEM, CD-SEM, CD-SAXS, REM, AFM). Such techniques may be employed to generate profile coordinates directly or indirectly such as by way of a set of geometric profile parameters characterizing a geometry of the feature in the processed substrate (e.g., critical dimension, side wall angles, depth, etc.) ... the geometric information of a feature may be a profile (viewed as a cross-section of the substrate in an x-z plane)" teaches that the generated profile (synthetic image) is a cross-section image of the substrate based on a microscopy image (synthetic image resembling a realistic microscopy image of a cross section). [0020]: "Current methods of evaluating semiconductor wafers and other samples used in semiconductor process development employ techniques such as x-SEM (x-ray Scanning Electron Microscopy), STEM (Scanning Transmission Electron Microscopy), and CD-SAXS (Critical Dimension Small Angle X-ray Scattering) for cross-sectional image capture" teaches that the cross-sectional image resembles a microscopy image generated by scanning electron microscope techniques (i.e. SEM and x-SEM cross section analysis exposes the internal structure, layer thickness, and subsurface defects of the semiconductor by cutting it to view the interior side profile using a scanning electron microscope)).
Pathangi Sriraman, Brauer, and Feng et al. are analogous to the claimed invention because they are directed to using machine learning models for assessing manufactured devices.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate generate synthetic microscopy images resembling realistic microscopy images depicting cross sections of an internal structure of a manufactured device … wherein training the machine learning model comprises providing the first data to the machine learning model as training input as taught by Feng et al. to the disclosed invention of Pathangi Sriraman in view of Brauer.
One of ordinary skill in the art would have been motivated to make this modification to "drastically reduce the costs associated with performing destructive testing such as STEM and x-SEM" and "bring in better coupon-to-coupon consistency as random errors & LER/LWR (line edge roughness and line width roughness) on CD/contours/profiles are spatially averaged out by the optical signals" (Feng et al. [0055]-[0056]).
Regarding Claim 11,
Pathangi Sriraman in view of Brauer and further in view of Feng et al. teaches the method of claim 9.
In addition, Brauer further teaches wherein training the machine learning model comprises training a generative adversarial network (Fig. 2; [0064]: "The image and design clip pairs generated by step 202 may then be used as a training set, with the design clips designated as the training inputs and corresponding images designed as the training outputs, in step 204 in which the GAN (i.e., the generator-discriminator network (NW)) is trained using design and optical images. The training of the GAN may be performed as described further herein. In some embodiments, the computer subsystem(s) may be configured to perform steps 200, 202, and 204" teaches that training the machine learning model comprises training a GAN (generative adversarial network)).
Pathangi Sriraman, Brauer, and Feng et al. are analogous to the claimed invention because they are directed to using machine learning models for assessing manufactured devices.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate wherein training the machine learning model comprises training a generative adversarial network as taught by Brauer to the disclosed invention of Pathangi Sriraman in view of Feng et al.
One of ordinary skill in the art would have been motivated to make this modification to "provide a method for improved optical defect inspection using a GAN or cGAN for the artificial creation of optical patch images" (Brauer [0081]).
Regarding Claim 12,
Pathangi Sriraman in view of Brauer and further in view of Feng et al. teaches the method of claim 9.
In addition, Feng et al. further teaches wherein the plurality of microscopy images comprise cross-sectional images of the manufactured devices, and wherein the microscopy images are generated by electron-based imaging methods (Fig. 1; [0045]: "Next, in an operation 109, the method performs a form of metrology that directly generates profiles or other geometric characteristics of the features on the test substrates. As indicated, such metrologies are often slow and/or destructive. Examples include various forms of electron microscopy including various forms of TEM (e.g., STEM) and SEM (e.g., x-SEM and CD-SEM), as well as CD-SAXS. The resulting directly measured geometric characteristics of the features are associated with the corresponding optical signals produced by the features in operation 105, and optionally operation 107" teaches that the microscopy images are generated by scanning electron microscope techniques (electron-based imaging methods). [0020]: "Current methods of evaluating semiconductor wafers and other samples used in semiconductor process development employ techniques such as x-SEM (x-ray Scanning Electron Microscopy), STEM (Scanning Transmission Electron Microscopy), and CD-SAXS (Critical Dimension Small Angle X-ray Scattering) for cross-sectional image capture" teaches that the scanning electron microscope techniques generate microscopy images comprising cross-sectional images of the wafers (manufactured devices)).
Pathangi Sriraman, Brauer, and Feng et al. are analogous to the claimed invention because they are directed to using machine learning models for assessing manufactured devices.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate wherein the plurality of microscopy images comprise cross-sectional images of the manufactured devices, and wherein the microscopy images are generated by electron-based imaging methods as taught by Feng et al. to the disclosed invention of Pathangi Sriraman in view of Brauer.
One of ordinary skill in the art would have been motivated to make this modification to "drastically reduce the costs associated with performing destructive testing such as STEM and x-SEM" and "bring in better coupon-to-coupon consistency as random errors & LER/LWR (line edge roughness and line width roughness) on CD/contours/profiles are spatially averaged out by the optical signals" (Feng et al. [0055]-[0056]).
Regarding Claim 13,
Pathangi Sriraman teaches a non-transitory machine-readable storage medium storing instructions which, when executed, cause a processing device to perform operations (Fig. 1; Fig. 2; Fig. 3; [0022]: "a non-transitory computer-readable storage medium comprising one or more programs for executing steps on one or more computing devices is provided. The steps can include applying a design clip to a metrology area image thereby obtaining a synthesized image such that one or more process change variations are suppressed and one or more tool distortions are enhanced and sending instructions to perform metrology operations on the synthesized image" teaches a non-transitory computer-readable storage medium comprising one or more programs (instructions) for executing steps (operations) on one or more computing devices (processing device)) comprising:
receiving first data indicating a plurality of dimensions of a manufactured device based on one or more measurements of the manufactured device (Fig. 1; Fig. 2; Fig. 3; [0012]-[0013]: "a method of compensating for scanning electron microscope beam distortion-induced metrology error is provided. A metrology area image of a wafer and a design clip from a storage medium are received at a processor. The design clip is applied to the metrology area image using the processor thereby obtaining a synthesized image. One or more process change variations are suppressed and one or more tool distortions are enhanced. Metrology operations can be performed on the synthesized image … The design clip can be a design for the wafer or a synthesized design clip. The synthesized design clip can be generated using the process change variations with a machine algorithm and communicating the synthesized design clip to the storage medium. Generating the synthesized design clip can further include receiving images and data from one or more process modulated wafers at a machine learning module and learning the process change variations at the machine learning module using the machine algorithm" teaches obtaining a synthesized design clip (first data) based on a metrology images and data (dimensions) of from a process-modulated semiconductor wafer (manufactured device). [0005]: "Metrology processes can be used to measure one or more characteristics of wafers such that the performance of a process can be determined from the one or more characteristics. For example, metrology processes can measure a dimension (e.g., line width, thickness, etc.) of features formed on the wafers during the process" teaches that the metrology data includes dimensions of the wafer based on measurements of the manufactured semiconductor wafer (manufactured device)).
Pathangi Sriraman does not appear to explicitly teach providing the first data to a trained machine learning model; receiving a synthetic microscopy image generated from the trained machine learning model, the synthetic microscopy image comprising a synthetic image resembling a realistic microscopy image of a cross section depicting an internal structure of the manufactured device, wherein the synthetic microscopy image is generated in view of the first data; and performing at least one of (i) outputting the synthetic microscopy image to a display or (ii) performing one or more operations on the synthetic microscopy image.
However, Brauer teaches providing the first data to a trained machine learning model (Fig. 2; [0096]: "As shown in step 210, the one or more computer subsystems may then process the design clips generated by step 208 through the previously trained GAN. The trained GAN may then output artificially generated images, as shown in step 212. Among other things, the images output by the GAN in step 212 may be used as input data for training other learning algorithms, as shown in step 214, which may be performed as described further herein" teaches that the design clips (first data) are input (provided) to a trained GAN (machine learning model));
receiving a synthetic microscopy image generated from the trained machine learning model (Fig. 2; [0096]: "As shown in step 210, the one or more computer subsystems may then process the design clips generated by step 208 through the previously trained GAN. The trained GAN may then output artificially generated images, as shown in step 212. Among other things, the images output by the GAN in step 212 may be used as input data for training other learning algorithms, as shown in step 214, which may be performed as described further herein" teaches that the trained GAN (machine learning model) uses the design clips (first data) to output artificially generated images. [0110]: "In another embodiment, the simulated image is a reference image … This artificial image may then be used as a reference image for defect inspection. In this manner, the embodiments described herein can be used to generate an artificial optical (or other) reference image to perform die-to-database inspections. One novel feature of the embodiments described herein is therefore that they provide a method for improved optical defect inspection using a GAN or cGAN for the artificial creation of optical reference patch images which can be used for optical die-to-database inspections. Another novel feature of the embodiments described herein is that they provide a method for improved optical defect inspection using a GAN or cGAN for an unsupervised DL method for optical reference image generation" teaches that the generated artificial images are artificial optical reference patch images (synthetic microscopy images)); and
performing at least one of (i) outputting the synthetic microscopy image to a display or (ii) performing one or more operations on the synthetic microscopy image (Fig. 2; [0113]-[0014]: "The images, model or network, and information therefor may be stored with any of the other results described herein and may be stored in any manner known in the art … After the information has been stored, the information can be accessed in the storage medium and used by any of the method or system embodiments described herein, formatted for display to a user, used by another software module, method, or system, etc. For example, the embodiments described herein may generate an inspection recipe as described above. That inspection recipe may then be stored and used by the system or method (or another system or method) to inspect the specimen or other specimens to thereby generate information (e.g., defect information) for the specimen or other specimens … Results and information generated by performing the inspection on the specimen or other specimens of the same type may be used in a variety of manners by the embodiments described herein and/or other systems and methods. Such functions include, but are not limited to, altering a process such as a fabrication process or step that was or will be performed on the inspected specimen or another specimen in a feedback or feedforward manner. For example, the computer subsystem(s) described herein may be configured to determine one or more changes to a process that was performed on a specimen inspected as described herein and/or a process that will be performed on the specimen based on the detected defect(s). The changes to the process may include any suitable changes to one or more parameters of the process" teaches that the output image (synthetic microscopy image) may be stored and then formatted for display to a user or for performing operations on the output image (synthetic microscopy image)).
Pathangi Sriraman and Brauer are analogous to the claimed invention because they are directed to using machine learning models for assessing manufactured devices.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate providing the first data to a trained machine learning model; receiving a synthetic microscopy image generated from the trained machine learning model; and performing at least one of (i) outputting the synthetic microscopy image to a display or (ii) performing one or more operations on the synthetic microscopy image as taught by Brauer to the disclosed invention of Pathangi Sriraman.
One of ordinary skill in the art would have been motivated to make this modification to "provide a method for improved optical defect inspection using a GAN or cGAN for the artificial creation of optical patch images" (Brauer [0081]).
Pathangi Sriraman in view of Brauer does not appear to explicitly teach the synthetic microscopy image comprising a synthetic image resembling a realistic microscopy image of a cross section depicting an internal structure of the manufactured device, wherein the synthetic microscopy image is generated in view of the first data.
However, Feng et al. teaches the synthetic microscopy image comprising a synthetic image resembling a realistic microscopy image of a cross section depicting an internal structure of the manufactured device, wherein the synthetic microscopy image is generated in view of the first data (Fig. 4; [0086]: "FIG. 4 presents a block diagram of an example metrology system 401. As illustrated, the system includes an optical metrology tool 407 configured to probe substrates such as a substrate 403 having features 405. Optical metrology tool 407 collects optical information from features 405 and generates an optical metrology output (e.g., reflectance intensity versus wavelength or optical critical dimension information). A metrology machine learning model 409 receives the optical metrology output and generates geometric information about the features; e.g., feature profiles, CDs, contours, etc. See output 411. In certain embodiments, the machine learning model 409 is implemented as a neural network" teaches that the neural network generates a feature profile (synthetic image) of the substrate (manufactured device) based on the metrology data (first data) measured from the substrate. [0064]-[0067]: "measuring features of the processed substrate produces profile coordinates. In such embodiments, measuring features of the processed substrate may include performing microscopy (e.g., SEM, x-SEM, TEM, STEM, CD-SEM, CD-SAXS, REM, AFM). Such techniques may be employed to generate profile coordinates directly or indirectly such as by way of a set of geometric profile parameters characterizing a geometry of the feature in the processed substrate (e.g., critical dimension, side wall angles, depth, etc.) ... the geometric information of a feature may be a profile (viewed as a cross-section of the substrate in an x-z plane)" teaches that the generated profile (synthetic image) is a cross-section image of the substrate based on a microscopy image (synthetic image resembling a realistic microscopy image of a cross section). [0020]: "Current methods of evaluating semiconductor wafers and other samples used in semiconductor process development employ techniques such as x-SEM (x-ray Scanning Electron Microscopy), STEM (Scanning Transmission Electron Microscopy), and CD-SAXS (Critical Dimension Small Angle X-ray Scattering) for cross-sectional image capture" teaches that the cross-sectional image resembles a microscopy image generated by scanning electron microscope techniques (i.e. SEM and x-SEM cross section analysis exposes the internal structure, layer thickness, and subsurface defects of the semiconductor by cutting it to view the interior side profile using a scanning electron microscope)).
Pathangi Sriraman, Brauer, and Feng et al. are analogous to the claimed invention because they are directed to using machine learning models for assessing manufactured devices.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate the synthetic microscopy image comprising a synthetic image resembling a realistic microscopy image of a cross section depicting an internal structure of the manufactured device, wherein the synthetic microscopy image is generated in view of the first data as taught by Feng et al. to the disclosed invention of Pathangi Sriraman in view of Brauer.
One of ordinary skill in the art would have been motivated to make this modification to "drastically reduce the costs associated with performing destructive testing such as STEM and x-SEM" and "bring in better coupon-to-coupon consistency as random errors & LER/LWR (line edge roughness and line width roughness) on CD/contours/profiles are spatially averaged out by the optical signals" (Feng et al. [0055]-[0056]).
Regarding Claim 14,
Pathangi Sriraman in view of Brauer and further in view of Feng et al. teaches the non-transitory machine-readable storage medium of claim 13.
In addition, Pathangi Sriraman further teaches wherein the operations further comprise: receiving second data indicating a subset of the plurality of dimensions indicated by the first data (Fig. 1; Fig. 2; Fig. 3; [0047]: "FIG. 3 shows an embodiment of a method 300 of generating and storing a synthesized design clip. Images or images and data from one or more process modulated wafers 301 are received at a machine learning module. The data can include, for example, metadata related to the alignment process and/or rendering. The images or images and data can be inputted into a machine or deep learning algorithm at the machine learning module at step 302 to learn process change variations. These process change variations are used in step 303 to generate a synthesized design clip, which in step 304 is communicated with and/or stored in an electronic data storage medium 305" teaches that the synthesized design clip (first data) is generated based on images (second data that is a subset of the first data) for process-modulated wafers (manufacturing device). Fig. 1; Fig. 2; Fig. 3; [0005]-[0006]: "Metrology processes can be used to measure one or more characteristics of wafers such that the performance of a process can be determined from the one or more characteristics. For example, metrology processes can measure a dimension (e.g., line width, thickness, etc.) of features formed on the wafers during the process. In addition, if the one or more characteristics of the wafers are unacceptable (e.g., out of a predetermined range for the characteristic(s)), the measurements of the one or more characteristics of the wafers may be used to alter one or more parameters of the process such that additional wafers manufactured by the process have acceptable characteristic(s). … With increasing pattern density and complexity in advanced technology nodes, automated in-line metrology operations that use a scanning electron microscope (SEM) are claiming an increased share in advanced process control" teaches that the images are metrology images including dimensions of the wafer);
receiving third data indicating one or more rules of design of the manufactured device (Fig. 1; Fig. 2; Fig. 3; [0047]: "FIG. 3 shows an embodiment of a method 300 of generating and storing a synthesized design clip. Images or images and data from one or more process modulated wafers 301 are received at a machine learning module. The data can include, for example, metadata related to the alignment process and/or rendering. The images or images and data can be inputted into a machine or deep learning algorithm at the machine learning module at step 302 to learn process change variations. These process change variations are used in step 303 to generate a synthesized design clip, which in step 304 is communicated with and/or stored in an electronic data storage medium 305" teaches that the synthesized design clip (first data) is generated based on data (third data) for process-modulated wafers (manufacturing device) including metadata related to the alignment process and/or rendering (rules of design)); and
providing the second data and the third data to a model configured to generate the first data in view of the second data and the third data (Fig. 1; Fig. 2; Fig. 3; [0047]: "FIG. 3 shows an embodiment of a method 300 of generating and storing a synthesized design clip. Images or images and data from one or more process modulated wafers 301 are received at a machine learning module. The data can include, for example, metadata related to the alignment process and/or rendering. The images or images and data can be inputted into a machine or deep learning algorithm at the machine learning module at step 302 to learn process change variations. These process change variations are used in step 303 to generate a synthesized design clip, which in step 304 is communicated with and/or stored in an electronic data storage medium 305" teaches that the image data (second data) and the metadata (third data) of a wafer are provided to a machine learning model to generate a synthesized design clip (first data)).
Regarding Claim 16,
Pathangi Sriraman in view of Brauer and further in view of Feng et al. teaches the non-transitory machine-readable storage medium of claim 14.
In addition, Pathangi Sriraman further teaches wherein the second data comprises metrology data generated from a non-destructive measurement process (Fig. 1; Fig. 2; Fig. 3; [0047]: "FIG. 3 shows an embodiment of a method 300 of generating and storing a synthesized design clip. Images or images and data from one or more process modulated wafers 301 are received at a machine learning module. The data can include, for example, metadata related to the alignment process and/or rendering. The images or images and data can be inputted into a machine or deep learning algorithm at the machine learning module at step 302 to learn process change variations. These process change variations are used in step 303 to generate a synthesized design clip, which in step 304 is communicated with and/or stored in an electronic data storage medium 305" teaches that the synthesized design clip (first data) is generated based on images (second data that is a subset of the first data) for process-modulated wafers (manufacturing device). Fig. 1; Fig. 2; Fig. 3; [0005]-[0006]: "Metrology processes can be used to measure one or more characteristics of wafers such that the performance of a process can be determined from the one or more characteristics. For example, metrology processes can measure a dimension (e.g., line width, thickness, etc.) of features formed on the wafers during the process. In addition, if the one or more characteristics of the wafers are unacceptable (e.g., out of a predetermined range for the characteristic(s)), the measurements of the one or more characteristics of the wafers may be used to alter one or more parameters of the process such that additional wafers manufactured by the process have acceptable characteristic(s). … With increasing pattern density and complexity in advanced technology nodes, automated in-line metrology operations that use a scanning electron microscope (SEM) are claiming an increased share in advanced process control" teaches that the images (second data) are metrology images that are generated using a scanning electron microscope (SEM) (i.e. a non-destructive measurement process)).
Regarding Claim 17,
Pathangi Sriraman in view of Brauer and further in view of Feng et al. teaches the non-transitory machine-readable storage medium of claim 13.
In addition, Brauer further teaches wherein the trained machine learning model comprises a generator of a generative adversarial network (Fig. 2; [0096]: "As shown in step 210, the one or more computer subsystems may then process the design clips generated by step 208 through the previously trained GAN. The trained GAN may then output artificially generated images, as shown in step 212. Among other things, the images output by the GAN in step 212 may be used as input data for training other learning algorithms, as shown in step 214, which may be performed as described further herein" teaches that the trained model comprises a GAN (generative adversarial network). Fig .5; [0075]: "An example of a generator that may be included in embodiments of a GAN configured as described herein is shown in FIG. 5" teaches that the GAN comprises a generator model).
Pathangi Sriraman, Brauer, and Feng et al. are analogous to the claimed invention because they are directed to using machine learning models for assessing manufactured devices.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate wherein the trained machine learning model comprises a generator of a generative adversarial network as taught by Brauer to the disclosed invention of Pathangi Sriraman in view of Feng et al.
One of ordinary skill in the art would have been motivated to make this modification to "provide a method for improved optical defect inspection using a GAN or cGAN for the artificial creation of optical patch images" (Brauer [0081]).
Regarding Claim 18,
Pathangi Sriraman in view of Brauer and further in view of Feng et al. teaches the non-transitory machine-readable storage medium of claim 13.
In addition, Feng et al. further teaches wherein the synthetic microscopy image comprises a synthetic cross sectional scanning electron microscope image (Fig. 4; [0086]: "FIG. 4 presents a block diagram of an example metrology system 401. As illustrated, the system includes an optical metrology tool 407 configured to probe substrates such as a substrate 403 having features 405. Optical metrology tool 407 collects optical information from features 405 and generates an optical metrology output (e.g., reflectance intensity versus wavelength or optical critical dimension information). A metrology machine learning model 409 receives the optical metrology output and generates geometric information about the features; e.g., feature profiles, CDs, contours, etc. See output 411. In certain embodiments, the machine learning model 409 is implemented as a neural network" teaches that the neural network generates a feature profile (synthetic image) of the substrate *manufactured device) based on the metrology data (first data) measured from the substrate. [0067]: "the geometric information of a feature may be a profile (viewed as a cross-section of the substrate in an x-z plane)" teaches that the generated profile (synthetic image) is a cross-section image of the substrate. [0020]: "Current methods of evaluating semiconductor wafers and other samples used in semiconductor process development employ techniques such as x-SEM (x-ray Scanning Electron Microscopy), STEM (Scanning Transmission Electron Microscopy), and CD-SAXS (Critical Dimension Small Angle X-ray Scattering) for cross-sectional image capture" teaches that the cross-sectional image resembles a microscopy image generated by scanning electron microscope techniques (electron-based microscopy technique)).
Pathangi Sriraman, Brauer, and Feng et al. are analogous to the claimed invention because they are directed to using machine learning models for assessing manufactured devices.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate wherein the synthetic microscopy image comprises a synthetic cross sectional scanning electron microscope image as taught by Feng et al. to the disclosed invention of Pathangi Sriraman in view of Brauer.
One of ordinary skill in the art would have been motivated to make this modification to "drastically reduce the costs associated with performing destructive testing such as STEM and x-SEM" and "bring in better coupon-to-coupon consistency as random errors & LER/LWR (line edge roughness and line width roughness) on CD/contours/profiles are spatially averaged out by the optical signals" (Feng et al. [0055]-[0056]).
Allowable Subject Matter
Claims 3-4, 7-8, 10, 15, and 19-20 are objected to as being dependent upon a rejected base claim, but would be allowable if rewritten in independent form including all of the limitations of the base claim and any intervening claims.
Response to Arguments
Applicant’s arguments, filed 06/29/2026, with respect to the claim rejections under 35 U.S.C. 103 to claims 3-4, 7-8, 10, 15, and 19-20 have been fully considered and are persuasive. Therefore, the 35 U.S.C. 103 prior art rejections have been withdrawn.
Applicant's arguments, filed 06/29/2026, with respect to the 35 U.S.C. 103 prior art rejections to claims 1-2, 5-6, 9, 11-14, and 16-18 have been fully considered but they are not persuasive. Applicant asserts “The Office Action rejects claims 1-2, 5-9, 11-14, and 16-20 under 35 U.S.C. § 103 as allegedly being unpatentable over Sriraman in view of Brauer and Feng. Applicant respectfully submits that claims 1, 9, and 13 are patentable over the combination of cited references because the combination does not teach or suggest all the features of the claims. …
… Sriraman is directed to compensating for scanning electron microscope beam distortion-induced metrology error. (Sriraman, Abstract.) The Office Action further cites Brauer, which is directed towards generating simulated images of specimens using generative adversarial networks and design data. (Brauer, Abstract).
The Office Action recognizes that Sriraman and Brauer do not disclose "a synthetic image of a cross section of the manufactured device, wherein the synthetic microscopy image is generated in view of the first data" and relies on the Feng reference to supply this limitation. Obviousness requires that all claim features are taught or suggested by the combination of cited references. Applicant respectfully submits that the combination of cited references fails to teach or suggest at least the bolded features.
Amended claim 1, for example, recites the limitations of:
"receiving a synthetic microscopy image generated from the trained machine learning model, the synthetic microscopy image comprising a synthetic image resembling a realistic microscopy image of a cross section depicting an internal structure of the manufactured device, wherein the synthetic microscopy image is generated in view of the first data" (Emphasis added).
Applicant submits that none of the cited references disclose the above limitations. Specifically, Feng and the other references do not generate a realistic microscopy image of a cross section of a manufactured device, and no cited reference discloses a synthetic microscopy image that depicts an internal structure of the manufactured device.
First, Feng instead discusses geometrical information about features, not realistic images of cross-sections. The Office Action cites paragraphs [0086] and [0067] of Feng. Paragraph [0086] describes a metrology machine learning model that "receives the optical metrology output and generates geometric information about the features; e.g., feature profiles, CDs, contours, etc." Feng, paragraph [0086]. The output of Feng's model is thus described as numerical geometric data such as profile coordinates, critical dimensions, and contours, not a visual image. Paragraph [0067] further explains that
"geometric information about a feature may take various forms" including "profile coordinates in two or three dimensions and/or a more global parameter such as trench depth, side wall angle, and the like." Feng, paragraph [0067].
The Office Action cites the statement in Feng paragraph [0067] that "the geometric information of a feature may be a profile (viewed as a cross-section of the substrate in an x-z plane)" to map the cross-section limitation. However, this "profile viewed as a cross-section" is geometric coordinate data describing a feature's shape, not a realistic microscopy image. Profile coordinates in two or three dimensions cannot reasonably be interpreted as "a synthetic image resembling a realistic microscopy image of a cross section" as recited by amended claim 1.
Second, the information in Feng is related to surface-level features and does not depict an internal structure of the manufactured device, as recited in claim 1. The cited passages of Feng in paragraphs [0086] and [0067] discuss generating numerical geometric parameters, e.g., profile coordinates, trench depth, side wall angle, pitch, and critical dimensions, that characterize the surface geometry of features formed on the substrate. None of the cited references generate a synthetic image that visually depicts internal structures of a manufactured device, as recited by claim l's "depicting an internal structure of the manufactured device."
Similar language is also included in independent claims 9 and 13. Thus, the combination of cited references does not teach or suggest all the features of independent claims 1, 9, and 13, and corresponding dependent claims 2-8, 10-12, and 14-20. Accordingly, Applicant respectfully requests the rejection of claims 1-2, 5-9, 11-14, and 16-20 under 35 U.S.C. § 103 be withdrawn.” (Remarks Pages 7-9).
Examiner’s Response:
The examiner respectfully disagrees. Regarding claim 1, the examiner respectfully disagrees with applicant’s assertion that “the synthetic microscopy image comprising a synthetic image resembling a realistic microscopy image of a cross section depicting an internal structure of the manufactured device, wherein the synthetic microscopy image is generated in view of the first data” is at least not taught by the cited prior arts. In particular, examiner points to paragraphs [0086], [0064]-[0067], and [0020] of Feng et al. (US 2021/0035833 A1), which specifically disclose, with respect to Fig. 4, "FIG. 4 presents a block diagram of an example metrology system 401. As illustrated, the system includes an optical metrology tool 407 configured to probe substrates such as a substrate 403 having features 405. Optical metrology tool 407 collects optical information from features 405 and generates an optical metrology output (e.g., reflectance intensity versus wavelength or optical critical dimension information). A metrology machine learning model 409 receives the optical metrology output and generates geometric information about the features; e.g., feature profiles, CDs, contours, etc. See output 411. In certain embodiments, the machine learning model 409 is implemented as a neural network" (i.e. the neural network generates a feature profile (synthetic image) of the substrate (manufactured device) based on the metrology data (first data) measured from the substrate), "measuring features of the processed substrate produces profile coordinates. In such embodiments, measuring features of the processed substrate may include performing microscopy (e.g., SEM, x-SEM, TEM, STEM, CD-SEM, CD-SAXS, REM, AFM). Such techniques may be employed to generate profile coordinates directly or indirectly such as by way of a set of geometric profile parameters characterizing a geometry of the feature in the processed substrate (e.g., critical dimension, side wall angles, depth, etc.) ... the geometric information of a feature may be a profile (viewed as a cross-section of the substrate in an x-z plane)" (i.e. the generated profile (synthetic image) is a cross-section image of the substrate based on a microscopy image (synthetic image resembling a realistic microscopy image of a cross section)), and "Current methods of evaluating semiconductor wafers and other samples used in semiconductor process development employ techniques such as x-SEM (x-ray Scanning Electron Microscopy), STEM (Scanning Transmission Electron Microscopy), and CD-SAXS (Critical Dimension Small Angle X-ray Scattering) for cross-sectional image capture" (i.e. the cross-sectional image resembles a microscopy image generated by scanning electron microscope techniques (i.e. SEM and x-SEM cross section analysis exposes the internal structure, layer thickness, and subsurface defects of the semiconductor by cutting it to view the interior side profile using a scanning electron microscope)). The machine learning model outputs generated geometric information as a feature profile (synthetic microscopy image) comprising a cross-sectional view of the substrate (manufactured device) in an x-y plane based on measurements made from performing scanning electron microscopy (SEM) for cross-sectional image capture. SEM cross-sectional image analysis uses a scanning electron microscope to image the internal structure of a substrate, meaning that microscopy image of the cross section depicts an internal structure of the substrate. In other words, the machine learning model receives an input of metrology data as a cross-sectional microscopy image characterizing geometry of features of internal structures of a substrate and outputs geometric information as a feature profile of a cross-sectional view of the substrate (manufactured device) in an x-y plane (e.g. a cross-sectional microscopy image of the internal structure of the substrate in an x-y plane).
Applicant relies on the arguments above regarding independent claims 9 and 13 and dependent claims 2, 5-6, 11-12, 14, and 16-18, therefore the response above is applicable to those claims.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to BRIAN J HALES whose telephone number is (571)272-0878. The examiner can normally be reached M-F 9:00am - 5:00pm.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Kamran Afshar can be reached at (571) 272-7796. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/BRIAN J HALES/Examiner, Art Unit 2125
/KAMRAN AFSHAR/Supervisory Patent Examiner, Art Unit 2125
Prosecution Insights