DETAILED ACTION
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 .
Response to Arguments
Applicant’s arguments received 18 May, 2026, have been fully considered. Claims 1-22 are pending. Claims 9-13 and 19-20 are withdrawn. Claims 1, 5, 7-8, 14, 17, and 21 are amended.
Applicant’s efforts to address objections to the drawings are satisfactory, therefore all objections to the drawings are withdrawn.
Applicant’s efforts to address objections to the specification are satisfactory, therefore all objections to the specification are withdrawn.
Applicant’s efforts to address rejections under 35 U.S.C. 101 have been considered.
Applicant argues that the claims do not recite judicial exceptions because they do not recite concepts that explicitly fall into the abstract idea exception categories. The examiner disagrees with this argument. Under broadest reasonable interpretation, processing measurement data to predict a critical dimension profile encompasses a mathematical process, for example. While the processing is performed by a machine learning model, at the level of detail given this is essentially similar to applying a general-purpose computer to perform a mathematical operation.
Applicant further argues that any judicial exception would be integrated into the practical application of improving semiconductor manufacturing metrology and process control. Again, the examiner disagrees. At no point does the claim language restrict the field of use to semiconductor manufacturing. The term “semiconductor” or an equivalent is not present in the claim language. Little detail is provided about what the “substrate” might be, save that was processed as part of a manufacturing process. While microscopy is explicitly recited, microscopes are used for a wide variety of purposes. Thus the field of use seems to be limited to a manufacturing process that may involve microscopy. This could apply just as easily to the biomedical industry as to the semiconductor industry. While it is true that breadth is not indefiniteness (MPEP § 2173.03), it is also the case that, when broad language is used, claims can run the risk of representing little more than an attempt to claim a judicial exception when considered under the broadest reasonable interpretation.
Again, although Applicant argues that the amended claims recite integrating the data processing result into a physical change of a manufacturing process, the examiner does not consider this to remedy the issue outlined above.
See 101 rejections below.
Applicant’s efforts to address rejections under 35 U.S.C. 103 have been considered.
Applicant argues that the prior art does not recite the amended features. Applicant argues that Pandev does not recite a generative model, and that Pandev does not process a CD profile image using a second machine learning model to produce a synthetic microscopy image. Applicant argues that the previous Office action asserts that Sakai teaches these features, but that Sakai does not, disclosing neither machine learning nor generating synthetic microscopy images.
The examiner disagrees that Pandev does not teach a generative model, referring to the arguments in the previous Office action that a model which generates output is a generative model under broadest reasonable interpretation. The examiner also disagrees with Applicant’s conclusions that the prior art of record does not teach or suggest processing a CD profile image using a second machine learning model to produce a synthetic microscopy image. While it is true that Sakai does not discuss machine learning nor the generation of synthetic images, Pandev does discuss both. Sakai is referenced because it teaches that those of ordinary skill in the art make and rely on SEM images of cross sections of substrates in the field of semiconductor manufacturing. The examiner considers that it would have been obvious to one of ordinary skill in the art to apply Sakai’s teaching to Pandev by applying machine learning, a tool of general utility, to generate a synthetic microscopy image from a CD profile image because it is common for those in the art to view semiconductor substrate cross sections as SEM images.
New grounds of rejection are written in light of the claim amendments. See 103 rejections below.
Claim Objections
Claim 1 is objected to because of the following informalities: “comprising generative model” should be replaced with “comprising a generative model”. Appropriate correction is required.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-8, 14-18, and 21-22 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
At Step 1 of the 101 analysis, all claims are directed to one of the statutory categories of invention.
Claim 1 is rejected in response to the following analysis:
At Step 2A, Prong One, the judicial exceptions are bolded in the copy of claim 1 below:
A method comprising:
processing measurement data of a substrate that was processed according to a manufacturing process using a first trained machine learning model to predict a critical dimension (CD) profile for the substrate;
generating a CD profile prediction image based on the predicted CD profile for the substrate;
processing the CD profile prediction image using a second trained machine learning model comprising generative model to generate a synthetic microscopy image of the substrate;
determining, based on the synthetic microscopy image, a dimension of the substrate; and
causing, based on the determined dimension, an update to one or more parameters of the manufacturing process to control manufacturing equipment used to process the substrate.
Processing measurement data and a CD profile prediction image both represent mathematical operations, as does predicting a CD profile.
At Step 2A, Prong Two, the additional elements do not integrate the judicial exceptions into a practical application. The first and second machine learning models represent generic computer functions. Generating a CD profile prediction image and a synthetic microscopy image represent results, however they are not particular (what is the “substrate”?) nor are they applied to a practical application. Finally, as recited, updating a parameter of a manufacturing process describes an action, however this does not describe a particular practical real-world transformation. When considered as a whole, claim 1 describes a manufacturing process on a substrate and a set of general computer methods to accept substrate data and generate a couple of images, then estimating a dimension from one of the images and using the estimate to make some adjustment to the manufacturing process. This could describe a number of procedures in a number of disparate fields.
At Step 2B, the claim as a whole does not amount to significantly more than the judicial exceptions for the reasons given above.
Claims 2 and 3 further define the measurement data but do not significantly narrow the field of use, therefore the arguments for rejecting claim 1 still apply, and claims 2 and 3 are also rejected.
Claims 4 and 5 describe the machine learning models but do not change the arguments given in claim 1, therefore claims 4 and 5 are also rejected.
Claim 6 further defines the microscopy image but does not significantly limit the field of application, therefore the arguments of claim 1 remain and claim 6 is also rejected.
Claims 7 and 8 describe measuring a dimension of a device and causing a corrective action to be performed. However the general language used to describe the manufactured device and corrective action does not significantly narrow the field of use, therefore the arguments of claim 1 remain and claims 7 and 8 are also rejected.
Claim 14 recites a non-transitory storage medium with instructions that cause the method of claim 1, and is therefore rejected for the same reasons.
Claims 15-18 recite the limitations of claims 2-3 and 5-6, respectively, and are rejected for the same reasons.
Claim 21 recites a system with a memory coupled to a processor, where the processor implements the method of claim 1. Claim 21 is rejected for the same reasons as claim 1.
Claim 22 recites that the processor implements a method of training the machine learning models (here grouped into a single machine learning model) of claim 1 and is rejected for the same reasons.
Claim Rejections - 35 USC § 103
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.
Claims 1-4, 6-8, 14-16, 18, and 21-22 are rejected under 35 U.S.C. 103 as being unpatentable over Pandev (US 20220352041 A1) in view of Sakai (US 20100314661 A1).
Regarding claim 1, Pandev discloses a method comprising: processing measurement data of a substrate (¶15: semiconductor structures are measured, and "a trained parameter conditioned measurement model predicts a set of values of each non-conditioning parameter based on measurement data and a corresponding set of predetermined values for each conditioning parameter. In this manner, the trained parameter conditioned measurement model predicts the shape of a measured structure." ¶73 clarifies that the structures may be on a wafer and ¶125 shows that the wafer generally refers to a substrate) that was processed according to a manufacturing process (¶125: the wafer was formed of a semiconductor material i.e. it was processed, and may be patterned from a manufacturing process) using a first trained machine learning model (¶105: "In some embodiments, a measurement model trained as described herein is implemented as a neural network model.") to predict a critical dimension (CD) profile for the substrate (¶19: "parameters of interest determined based on a trained parameter conditioned measurement model as described herein, include, but are not limited to: geometric parameters characterizing a measured structure…Exemplary geometric parameters include critical dimensions (CD)"; see also Fig. 6 which depicts an image comprising a critical dimension profile, and ¶15 which states that “the trained parameter conditioned measurement model predicts the shape of a measured structure”); generating a CD profile prediction image based on the predicted CD profile for the substrate (Figs. 5 and 6 depict CD profile prediction images of a hole structure in a substrate; also ¶19: “A parameter conditioned measurement model as described herein enables the trained model to reconstruct a two dimensional image or a three-dimensional image of a structure under measurement”); and
causing, based on the determined dimension, an update to one or more parameters of the manufacturing process to control manufacturing equipment used to process the substrate (¶106: “the measurement results described herein can be used to provide active feedback to a process tool (e.g., lithography tool, etch tool, deposition tool, etc.)…corrections to process parameters determined based on measured device parameter values determined using a trained parameter conditioned measurement model may be communicated to the process tool.").
Pandev does not explicitly disclose processing the CD profile prediction image using a second trained machine learning model comprising a generative model to generate a synthetic microscopy image associated with the substrate; and determining, based on the synthetic microscopy image, a dimension of the substrate.
Sakai discloses a SEM cross-section image of a semiconductor substrate (see Fig. 2 and ¶17).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to incorporate the teachings of Sakai with the invention of Pandev by processing the CD profile prediction image to generate a synthetic microscopy image associated with the substrate. Doing so would be useful because SEM images are familiar to those of ordinary skill in the art. Furthermore, it would have been obvious to perform the processing using a second trained machine learning model comprising a generative model (note that machine learning models generate output given input and are therefore generative models). Machine learning models can be trained on large amounts of correlated data to produce a desired output given an input. Since it is reasonable to assume that the details of a SEM image of a substrate feature would be correlated with the details of a CD profile prediction image, Pandev in view of Sakai would be motivated to use CD profile prediction image and SEM image data to train a machine learning model to generate a synthetic SEM image. The resulting model would be able to consistently and autonomously translate CD prediction profile images into SEM images which are familiar to those of ordinary skill in the art.
Given the above, it would have been obvious to one of ordinary skill in the art practicing the invention of Pandev in view of Sakai to determine, based on the synthetic microscopy image, a dimension of the substrate. Doing so would enable one to determine a dimension of interest displayed on the image.
Regarding claim 2, Pandev in view of Sakai teaches the limitations of claim 1, and Pandev further discloses that the measurement data comprises a profile map of at least one of a film or a feature on the substrate (¶107: "the methods and systems for metrology of semiconductor devices as described herein are applied to the measurement of memory structures. These embodiments enable optical critical dimension (CD), film, and composition metrology for periodic and planar structures.").
Regarding claim 3, Pandev in view of Sakai teaches the limitations of claim 1, and Pandev further discloses that the measurement data comprises at least one of spectral data or reflectometry data (¶40: the measurement data can include "soft X-ray reflectometry" data; ¶70: "The trained parameter conditioned measurement model is employed to estimate values of one or more parameters of interest from actual measurement data (e.g., measured spectra)").
Regarding claim 4, Pandev in view of Sakai teaches the limitations of claim 1, and Pandev further discloses that the first trained machine learning model comprises a deep neural network (¶105: the measurement model may be a "deep network model").
Regarding claim 6, Pandev in view of Sakai teaches the limitations of claim 1. Furthermore, Pandev’s image is of a cross section of a substrate (Pandev, Fig. 5) and Sakai’s image is a scanning electron microscope (SEM) image of a cross section of a substrate (Sakai, Fig. 2). Following the arguments of claim 1 for generating a virtual SEM image, then, it follows that the synthetic microscopy image would comprise a virtual scanning electron microscopy (VSEM) image of a cross section of the substrate (see rejection of claim 1).
Regarding claim 7, Pandev in view of Sakai teaches the limitations of claim 1. Furthermore, it would have been obvious to one or ordinary skill in the art practicing the invention of Pandev in view of Sakai to cause determining the dimension of the substrate to further comprise measuring a feature of the synthetic microscopy image and calculating the dimension of the substrate based on the measurement of the feature of the synthetic microscopy image. This describes a basic process of extracting dimensional data from an image, and one would have been motivated to do so in order to actually estimate a dimension of interest displayed on the image.
Regarding claim 8, Pandev in view of Sakai teaches the limitations of claim 7 and further teaches that the update to the one or more parameters of the manufacturing process may comprise updating a process recipe of the manufacturing process (Pandev, ¶106: the active feedback can include adjusting etch time in the manufacturing process).
Pandev in view of Sakai does not explicitly disclose that the update to the one or more parameters of the manufacturing process comprises one or more of scheduling maintenance of the manufacturing equipment or providing an alert to a user, however doing so would have been obvious to one of ordinary skill in the art practicing the invention of Pandev in view of Sakai. Scheduling maintenance would enable the method to autonomously address a problem determined to be caused by machine failure, and providing an alert to a user would be useful to notify a person of an issue.
Note also that Pandev does disclose that the active feedback may involve stopping an etch process (see end of ¶06), and that it would have been obvious to also provide an alert to a user so they know that the manufacturing process has ended.
Regarding claim 14, the limitations of claim 1 are found in claim 14. Claim 14 also discloses a non-transitory computer-readable storage medium storing instructions which, when executed, cause a processing device to perform the method of claim 1. Noting that the rejection of claim 1 involves computers receiving data and performing operations, these limitations would have been obvious. Claim 14 is therefore rejected on the same grounds as claim 1.
Regarding claim 21, the limitations of claim 1 are found in claim 21. Claim 21 also discloses a memory and a processing device coupled to the memory, which implements the method of claim 1. Noting that the rejection of claim 1 involves computers receiving data and performing operations, these limitations would have been obvious. Claim 21 is therefore rejected on the same grounds as claim 1.
Regarding claims 15-16 and 18, these recite the same limitations as claims 2-3 and 6, respectively, and are rejected for the same reasons.
Regarding claim 22, Pandev in view of Sakai teaches the limitations of claim 21. Furthermore, the limitations of claim 22 are encompassed by training the first and second machine learning (ML) models recited in claim 1. That is, the “machine learning model” of claim 22 may be interpreted as comprising the “first machine learning model” and “second machine learning model” of claim 1.
In claim 1, the first ML model was provided with “measurement data of a substrate”; therefore it would have been reasonable to train the first ML model with “a plurality of CD measurements associated with a substrate”. The method of claim 1 generated a “CD profile prediction image,” which is analogous to the “plurality of CD profile images” generated in claim 22. Then, since the second ML model outputs synthetic microscopy images (in the rejection of claim 1 those images are SEM images), it would have been natural to train the second ML model by providing it with a “plurality of SEM images…as target output” and a “plurality of CD profile images” as input. If one defined a ML model comprising the first and second ML models, it would have been reasonable to conclude that such a model would be trained using an input data set comprising the plurality of SEM images and the plurality of CD profile images, and that training the ML model would comprise providing the plurality of CD measurements to the ML model as training input (i.e. as input into the first ML model), and providing the plurality of SEM images to the ML model as target output (i.e. as target output for the second ML model).
Therefore, it would have been obvious to one of ordinary skill in the art practicing the invention of Pandev in view of Sakai to perform the limitations of claim 22 in order to train the machine learning models to perform the method of claim 1 (see rejection of claims 1 and 21).
Claims 5 and 17 are rejected under 35 U.S.C. 103 as being unpatentable over Pandev (US 20220352041 A1) in view of Sakai (US 20100314661 A1), and further in view of Isola (“Image-to-Image Translation with Conditional Adversarial Networks”.
Regarding claim 5, Pandev in view of Sakai teaches the limitations of claim 1 but does not explicitly teach the limitations of claim 5.
Isola teaches that conditional adversarial networks (a type of GAN) can be used as a general-purpose solution to image-to-image translation problems (Abstract; see also Fig. 1 depicting application of “the same architecture and objective, and simply [trained] on different data,” to successfully perform a wide variety of image-to-image translation tasks.).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to incorporate the teachings of Isola with the invention of Pandev in view of Sakai by causing the generative model to be an image-to-image GAN. Doing so would enable one to apply a machine learning model which is demonstrably adept at a wide variety of image-to-image translation tasks.
Regarding claim 17, claim 17 recites the same limitations as claim 5 and is rejected for the same reasons.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ETHAN WESLEY EDWARDS whose telephone number is (571)272-0266. The examiner can normally be reached Monday - Friday, 7:30am-5pm.
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, Andrew Schechter can be reached at (571) 272-2302. 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.
ETHAN WESLEY EDWARDS
Examiner
Art Unit 2857
/E.W.E./ Examiner, Art Unit 2857
/ANDREW SCHECHTER/ Supervisory Patent Examiner, Art Unit 2857
Prosecution Insights