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 Amendment
Claims 1-22 are pending.
Applicant’s amendments to the claims, filed 01/22/2026, are accepted, with independent Claims 1, 11, and 21 are amended.
Applicant's arguments filed 01/22/2026 have been reviewed and fully considered.
With regard to interpretation of Claim 21 under 35 U.S.C. § 112(f), Applicant’s arguments (REMARKS, Pg6, Paragraph 2) regarding Examiner’s notice of how the claim limitation was interpreted for examination are understood as follows. Examiner notes claim limitations containing functional language, as described in previous office action and described in MPEP 2181/subsection I, where functional language in a limitation is not modified by sufficient structure, material or acts for performing the function within the limitation, invoke interpretation under 35 U.S.C. § 112(f). Using this interpretation, Examiner consulted specification to find meaning and/or structural details to facilitate examination, as explained in previous office action. As noted in previous office action “If applicant does not intend to have the limitation interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, applicant may: (1) amend the claim limitation(s) to avoid it/them being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph (e.g., by reciting sufficient structure to perform the claimed function); or (2) present a sufficient showing that the claim limitation(s) recite(s) sufficient structure to perform the claimed function so as to avoid it/them being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph.”
Examiner has fully reviewed Applicant’s Remarks and amended claim limitations, but does not find that Applicant has (1) amended claim limitations to avoid item being interpreted under 35 U.S.C. 112(f) or (2) presented sufficient showing that the claims limitation recites sufficient structure. Applicant argues (Pg.6): “addition to the portions of the application cited in the Office Action, the corresponding structure for the claim elements is further found in, e.g., paragraphs [0112]-[0114].” Examiner understands this remark is consistent with guidance found in MPEP for interpretation where 35 U.S.C. § 112(f) is invoked, i.e., referring to the specification for guidance in interpretation of claim limitation, and understands, based on the recited statement above that Applicant agrees with this guidance for facilitating examination. Examiner will refer to specification paragraphs according to Applicant’s remarks.
With regard to rejection of Claims 1-22 under 35 USC § 101, Examiner has reviewed and fully considered Applicant’s arguments but they are unpersuasive. Examiner maintains rejection for claims as currently amended under 35 USC § 101. Examiner acknowledges Applicant’s detailed arguments (Pg.6, Para 3) regarding legal standards and evaluation of eligibility, as set forth for rejections under 35 USC § 101. Examiner points to MPEP 2106, with examples discussed in the 2024 Guidance Update on Patent Subject Matter Eligibility, Including on Artificial Intelligence, as noted by Applicant (Remarks, Pg. 9), to be particularly relevant for examination of the claimed invention. Applicant argues that claim limitations, specifically limitation language of independent claims 1, 11, and 21, do recite mathematical processes using input data. Examiner notes the term “mental process” was used in the rejection based on the high level of generality of limitation language. Examiner notes, as pointed out by Applicant, once the claim limitation has been established to be in an eligible stator category (Step 1), it must be determined whether the claim is directed to a judicial exception (Step 2A, with two prongs). As noted by Applicant, Claim limitations found in independent claims 1, 11, and 21 do recite statutory eligibility, and also recite a judicial exception of “Abstract Idea”. Examiner notes using guidance found in MPEP 2106.04(a), there are three enumerated groupings of abstract ideas: 1) Mathematical concepts – mathematical relationships, mathematical formulas or equations, mathematical calculations (see MPEP § 2106.04(a)(2), subsection I); 2) Certain methods of organizing human activity – fundamental economic principles or practices (including hedging, insurance, mitigating risk); commercial or legal interactions (including agreements in the form of contracts; legal obligations; advertising, marketing or sales activities or behaviors; business relations); managing personal behavior or relationships or interactions between people (including social activities, teaching, and following rules or instructions) (see MPEP § 2106.04(a)(2), subsection II); and 3) Mental processes – concepts performed in the human mind (including an observation, evaluation, judgment, opinion) (see MPEP § 2106.04(a)(2), subsection III). In previous office action, applying broadest reasonable interpretation, Examiner noted the limitations of Claim 1 (and similarly claims 11 and 21) recited an abstract idea in the Mathematical concept OR Mental Process groupings. Applicant points out that there is no evidence that the limitations fall into a Mental Process grouping. Using broadest reasonable interpretation, Examiner respectfully disagrees in that a judgement or observation, or a simple mathematical calculation may fall into them Mental Process grouping. Thus, claims recited at a high level of generality may fall into either of the recited groupings. Examiner asserts that the language OR does not imply that any/all of the recited ideas must be accomplished using a mental process but allows for the possibility that some processes as recited may. Both mathematical concept and mental process are enumerated as possible manifestations of the Abstract Idea judicial exception.
Applicant argues that claim limitations found in claim 1 do not “set forth or describe any mathematical relationships”. Examiner respectfully disagrees, using broadest reasonable interpretation and plain meaning, finding proper cause to proceed to Step 2, Prong 2, for further consideration to determine eligibility. Applicant argues (Remarks, Pg. 13) that proper analysis was not carried out under Step 2A Prong 2. Examiner again respectfully disagrees, finding the claim limitation does not recite elements that integrate the Abstract Idea into a practical application. Applicant argues (Remarks, Pg. 13) that proper analysis was not carried out under Step 2A Prong 2. Examiner respectfully disagrees. Under further consideration of Step 2B, evaluation of dependent claims to ascertain the presence additional elements reciting an inventive concept or integration into a practical application, where additional elements were considered individually and in combination revealed that the claim limitation is directed to the identified judicial exception.
Applicant argues (Remarks, Pg. 13-14) using specific recitations from the specification as evidence of Examiner’s failure to properly perform evaluation under Step 2A, Prong 2, namely to consider whether the claimed invention sets forth an improvement in technology. Examiner notes under a broadest reasonable interpretation, words of the claim must be given their plain meaning, unless such meaning is inconsistent with the specification. While understanding of the claims may be guided by the specification, it is improper to import claim limitations from the specification (MPEP 2111.01, I.-II.), i.e., language from the specification cannot be imported into claim limitations. Examiner maintains rejection of Claims 1-22, as currently amended, under 35 U.S.C. 101, with further detailed reasoning and rationale below.
With regard to rejection of Claims 1-22 under 35 U.S.C. § 103, over obvious
combination of prior art, Examiner finds arguments are not persuasive. Further consideration and search as necessitated by amendments, with new grounds for rejection is detailed below.
Specifically, Applicant argues (Remarks, Pg. 16) that the combination of prior art of HONDA (US 20180356807 Al), in view of QUI (US 20220122103 Al) fail to teach each and every element of Claims 1 and 21 as currently amended. Because amended necessitate further search to address limitations not previously considered, a new grounds of rejection is proper, as detailed below. Applicant further argues that QUI “operates in a fundamentally different manner than the presents claims”. Examiner notes that the reference field of endeavor may be considered as different from the claimed invention and still be considered a proper reference. Applying guidance found in MPEP 2141.01(a) I: “A reference is analogous art to the claimed invention if: (1) the reference is from the same field of endeavor as the claimed invention (even if it addresses a different problem); or (2) the reference is reasonably pertinent to the problem faced by the inventor (even if it is not in the same field of endeavor as the claimed invention).” Examiner has considered the problem faced by the inventor, based on reading and guidance of specification in view of claim limitations and finds that a person of ordinary skill would have reason to consult and apply the teachings found in the disclosures of HONDA, QUI, and SU (US 20200356011 A1), and would understand each reference to be “reasonably pertinent” to the claimed invention.
Detailed response addressing Applicant arguments, with attention to reasoning and rationale as applied to establish a prima facie case of obviousness in determination that the claimed invention, with claim limitations as currently amended, does not distinguish over prior art is presented below with new grounds of rejection as necessitated by amendment.
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 1-22 are rejected under 35 U.S.C. § 101 because the claimed invention is directed to an abstract idea without significantly more. These claims fall into eligible statutory categories as set forth in 35 U.S.C. § 101 (MPEP § 2106.03). limitations are directed to a judicial exception, as explained below.
Claim 1 is held to be patent ineligible, as explained below.
Claim 1 recites the following abstract ideas, emphasized in bold below:
“A method for characterizing a device on a sample comprising:
obtaining measured metrology data from the device;
and
determining, based on the measured metrology data, at least
one parameter of the device with a machine learning model that uses composite metrology data,
wherein each composite metrology datum comprises a merger of a
metrology datum measured for a reference device and a first synthetic metrology datum for a first model of the reference device, wherein the first model of the reference device is a physical model of the reference device.”
STEP 1: Claim 1, and similarly for independent claims 11 and 21, recites an eligible statutory category, namely a process or method, based on input of metrology data. Claims 11 and 21 recite the eligible statutory category machine (system). (MPEP § 2106.03).
STEP 2A-Prong1: Claim 1 recites a judicial exception. (MPEP2106.04) Claim 1 describes, as discussed above, using broadest reasonable interpretation, processes
that fall within definition of Abstract Idea in the Mathematical Concept grouping.(MPEP
2106.04(a)(2), subsection I).
Specifically, referring to line numbers above, lines (1), (3), and (4) recited mathematical processes carried out in performing the method (emphasized in bold) or “Mental Process” grouping: concepts performed in the human mind (including an observation, evaluation, judgment, opinion) (see MPEP § 2106.04(a)(2), subsection III). This interpretation is supported by referring to specification in at least
FIG. 3 with [0018]: “graph illustrating a parameter space for metrology using machine learning”
FIG.s 4 A-E, with [0019]-[0023], where in figures are described as “workflow for generation of composite metrology data”, based on: [0019]“FIG. 4A…based on merging measured metrology data and synthetic metrology data to be included in training data.”; [0020]“FIG. 4B…combining synthetic metrology data calculated for a modified reference device with a variation between measured metrology data from a reference device and synthetic metrology data calculated”, [0021]“FIG. 4C…combining synthetic metrology data calculated for a modified reference device with a variation between measured metrology data from a reference device and synthetic metrology data calculated for the reference device”; [0022]“FIG. 4D…combining measured metrology data from a reference device with a variation between synthetic metrology data calculated for the reference device”; and [0023]“FIG. 4E…combining measured metrology data from a reference device with a variation between synthetic metrology data calculated for the reference device and synthetic metrology data calculated”
FIG. 5, with [0024] “workflow for generation of composite reference data based on merging parameters values”
FIG.s 11 and 12, each further describing processes involved in mathematical calculations, including machine learning model training.
Examiner notes terms shown above in bold recited mathematical processes as would be understood by one of ordinary skill in the art, and points further the mathematical (graphical) representations of results represented graphically in FIGs. 6A-B, 7A-B, 8A-B, 9A-B. In addition, specification further confirms mathematical concept judicial exception in at least:
[0006] “machine learning models trained using composite metrology data.”
[0035] “Simulated metrology data, e.g., simulated optical data, may be
calculated for the physical model using rigorous coupled-wave analysis (RCW A), Finite-Difference Time-Domain (FDTD) or Finite Element Method (FEM), or other similar techniques.”
[0039] “training data may be augmented using synthetic (simulated) data
calculated from physical models.”
[0054] “composite metrology data that is generated based on a combination
of theoretically calculated metrology data”
[0063] “fit to a physical model… Calculated metrology data is generated for
the physical model using a modeling technique, such as RCW A, FDTD, FEM, etc., which is fit to the measured metrology data.”
Examiner further notes, that while specific mathematical equations may not appear in claim limitations, such explicit mathematical equations for carrying out the judicial exception are found in the specification in at least [0066] “perturbation, which may be defined for a parameter P as follows: (Eq. 1)” and [0067] “random perturbation as a function of POCD_Fit may be defined as follows: (Eq. 2)”.
Examiner asserts that terms in bold above, including “determining”, “machine learning model”, “merger”, “first model”, “simulation”, and “generation” would be understood as an explicit as an explicit reference to a mathematical process, which in the context of a machine-learning based method, may be carried out via use of various computational algorithms.
STEP 2A – Prong 2: Claim 1 does not integrate the recited judicial exception into a practical application. In this step, the claim is evaluated to determine if limitations recite additional elements that integrate the exception into a practical application of that exception. This judicial exception is not integrated into a practical application because there is no improvement to another technology or technical field; improvements to the functioning of the computer itself; a particular machine; effecting a transformation or reduction of a particular article to a different state or thing. Examiner notes that since the claimed methods and system are not tied to a particular machine or apparatus, they do not represent an improvement to another technology or technical field.
Step 2B - Claim 1 does not amount to significantly more than the recited judicial
exception. Evaluation of additional elements identified in Claim 1, as noted above in underline, do not amount to significantly more than the judicial exception. Specifically, terms such as “obtaining measured metrology data”, measured for a reference device” recite are considered necessary data gathering or acquiring numerical input values to be used in the mathematical concept, and as such, these limitations do not integrate the abstract idea into a practical application. As recited in MPEP section 2106.05(g), necessary data gathering (i.e. receiving data) is considered extra solution activity in light of Mayo, 566 U.S. at 79, 101 USPQ2d at 1968; OIP Techs., Inc. v. Amazon.com, Inc., 788 F.3d 1359, 1363, 115 USPQ2d 1090, 1092-93 (Fed. Cir. 2015). Limitation language such as “on a sample” recite indicates field of use or technological environment in which the judicial exception is carried out. (MPEP2106.05(f)) As such, these terms is interpreted as nothing more than a series of mathematical calculations using input data, with support for this interpretation found in specification as noted above.
Further, examiner notes additional elements are well known, routine and conventional as evidenced by in the relevant art based on the prior art of record, including, for example: QUI (US 20220122103 A1), or HONDA (US 20180356807 A1) or SU (US 20200356011 A1), among others, directed to same or related technical field. The limitation language does not apply or use the judicial exception in some other meaningful way beyond generally linking the use of the judicial exception to a particular technological environment, such that the claim as a whole is more than a drafting effort designed to monopolize the exception. (see MPEP 2106.05(e) and Vanda Memo).
Independent Claims 11 and 21 are likewise held to be patent ineligible using the same rationale and reasoning as applied to Claim 1 above.
Dependent Claims 2-10, 12-20, and 22 further limit the abstract ideas without
integrating the abstract concept into a practical application or including additional limitations that can be considered significantly more than the abstract idea.
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 text of those sections of Title 35, U.S. Code not included in this action can
be found in a prior Office action.
Claims 1, 3-10, and 21-22 are rejected under 35 U.S.C. § 103 as being unpatentable over HONDA (US 20180356807 A1) in view of QUI (US 20220122103 A1) and further in view of ROTHSTEIN (US 20230017097 A1).
With regard to Claims 1 and 21, HONDA teaches:
A method for characterizing a device on a sample comprising: obtaining measured metrology data from the device; (HONDA is in same technical field and directed to similar field of use, [0002]: “ relates generally to semiconductor manufacturing processes, and more particularly, to methods for generating more robust predictions for targeted process variables”; HONDA teaches using data from a device in a manufacturing process, FIG.3 with [0046]: “general method 300 for building a robust production-worthy ML model that is focused on one or more targets of the semiconductor manufacturing process…targets include independent variables… relating to specific features of the semiconductor device and which are used to characterize the to the ML model.”; HONDA teaches obtaining metrology data from device, FIG. 1 with [0025]: “main idea is to basically understand the temporal dependencies of the independent variables, which are most commonly sensor measurements and/or parametric test measurements for semiconductor manufacturing applications”; and [0028]: “typical semiconductor manufacturing process 100…input data can be collected from the process at every step and sub-step of a production run, and yield and other performance characteristics may be calculated from the input data”)
and determining, based on the measured metrology data, at least one parameter of the device with a machine learning model that uses metrology data, (HONDA teaches using measured device data to implement ML model development, [0004]: “ML model can be constructed for a specific process parameter by sampling relevant data in order to build one or more training sets of data to represent expected performance of the process with regard to that parameter.”; HONDA implicitly teaches use of composite data for model development, FIG. 1, with [0028]: “input data can be collected from the process at every step and sub-step of a production run”.)
wherein the first model of the reference device is a physical model of the reference device. (HONDA teaches use of a physical model, [0061]: “Drift and the other variations in underlayer thickness can be determined using a physics-based model, also known as a white box model…the white-box model is a physics-based numerical method for finding solutions that satisfy these equations and may consist of performing a non-linear least-square (“NLLS”) fit to the reflectometry (i.e. spectral) data in order to determine the physical parameters of interest”; Examiner interprets “physical model” to mean generally a math-based, process-driven model rooted in fundamental physical laws which may act as a trusted baseline to validate, verify or inform models generated using machine learning techniques, as taught by reference.)
HONDA does not teach:
machine learning model that uses composite metrology data
wherein each composite metrology datum comprises a merger of a metrology datum measured for a reference device and a first synthetic metrology datum for a first model of the reference device,
QUI teaches:
machine learning model that uses composite metrology data (QUI is pertinent prior art, directed to field of machine learning model development applied to a manufacturing environment, Abstract: “customized product performance prediction method based on heterogeneous data difference compensation fusion” and teaches implementation of machine learning for prediction of product characteristics using composite data, [0006]: “customized product performance prediction method based on heterogeneous data difference compensation fusion, including the following steps” with [0007]: “[step] (1) with a configuration parameter of a customized product as an input feature and performance of the customized product to be predicted as an output feature, collecting and obtaining data samples; collecting actual measurement performance data of an existing product”; Examiner interprets “composite data” as analogous to reference “heterogenous data difference compensation fusion” to mean use of multiple source/types of data for model development.; QUI suggests machine learning method via implementation of BP neural network model for prediction, as would be clear and understood by one of ordinary skill in the art, FIGs. 2 and 3, with [0010]: “(4) selecting a BP neural network model as a performance prediction model of the customized product, and taking the input feature and the output feature selected in the step (1) as the input and output of the prediction model”)
wherein each composite metrology datum comprises a merger of a metrology datum measured for a reference device and a first synthetic metrology datum for a first model of the reference device, (QUI teaches amalgamation of measured data with predictive data for training models, using comparative analysis to produce a reference model of product, see [0009]: “[step] (3) performing difference compensation correction on the calculation simulation data set on the basis of the historical actual measurement data set: encoding the historical actual measurement data set and the calculation simulation data set on the basis of a depth auto-encoder”, and FIG.1, box 5, with [0010]: “[step] (4) selecting a BP neural network model as a performance prediction model of the customized product…taking the input feature and the output feature selected in the step (1) as the input and output of the prediction model; using the calculation simulation data set after the difference compensation correction as the training sample set, and training and constructing an optimal BP neural network model combined with a tabu search algorithm; and then testing the model by using the historical actual measurement test set”)
It would have been obvious to one of ordinary skill in the art before effective filing date of the claimed invention to modify the method disclosed by HONDA to include a machine learning model that uses composite metrology data and wherein each composite metrology datum comprises a merger of a metrology datum measured for a reference device and a first synthetic metrology datum for a first model of the reference device, such as that of QUI because it would be understood as a way to increase data volume and diversity to produce a more reliable and accurate model. One of ordinary skill would understand several advantages of combining the more advance machine-learning technique taught by QUI with the measurement and modeling method of HONDA, including the advantage of using merged (composite) data to help mitigate impact of noise or unexpected anomalies among individual datum, reduce data sparsity, and improve the efficiency of model development, saving computational time and expense.
With respect to Claims 3 and 22, HONDA in view of QUI teaches limitations of Claims 1 and 21, respectively.
HONDA does not teach:
wherein each composite metrology datum comprises the merger of the metrology datum and the first synthetic metrology datum and further a second synthetic metrology datum for a second model of a modified reference device that is changed with respect to the first model.
QUI further teaches:
wherein each composite metrology datum comprises the merger of the metrology datum and the first synthetic metrology datum and further a second synthetic metrology datum for a second model of a modified reference device that is changed with respect to the first model. (QUI teaches implementation of composite data techniques for model development, as above, [0006]; QUI further teaches model refinement, using measured and simulated data, [0017]: “calculation simulation data set ESets after the difference compensation correction is taken as the training sample set, the historical actual measurement verification set ESethvalid is taken as the verification sample set, the historical actual measurement test set ESethtest is taken as the test sample set, and an optimal BP neural network model BPNNsopt is trained and constructed in combination with the tabu search algorithm to serve as the final prediction model.”)
It would have been obvious to one of ordinary skill in the art before effective filing date of the claimed invention to further modify HONDA, as modified by QUI as taught above, to include method step of wherein each composite metrology datum comprises the merger of the metrology datum and the first synthetic metrology datum and further a second synthetic metrology datum for a second model of a modified reference device that is changed with respect to the first model, such as that further disclosed by QUI because it would be advantageous in minimizing the need for unnecessary measurements that require time and expense. Combining this technique taught by QUI with the method of HONDA would address “real-world” or “real-time” data scarcity to provide sufficient opportunity for a model to learn and improve and ultimately make accurate predictions. This motivation is suggested by HONDA (see, for example, [0005]). One of ordinary skill would also understand that using composite data for model training can address issues of bias from measured data that may occur during a measurement operation
With respect to Claim 4, HONDA, in view of QUI, teaches the limitations of Claim 3.
HONDA further teaches:
wherein the second model of the modified reference device has at least one parameter that is varied with respect to the first model. (HONDA teaches model refinement using device parameter, see [0004]: “application of machine learning (“ML”) algorithms…ML model can be constructed for a specific process parameter by sampling relevant data in order to build one or more training sets of data to represent expected performance of the process with regard to that parameter.”)
With respect to Claim 5, HONDA, in view of QUI, teaches the limitations of Claim 3.
HONDA further teaches,
wherein the machine learning model further uses reference parameters for the modified reference device, a first set of key parameter values generated for the first model, and a second set of key parameters values generated for the second model. (HONDA teaches, as above, model refinement using device parameters, [0004]; and FIG. 3, depicting incorporation of temporal models; FIGs 6,7,8 teaching progressive model development using key measured parameters, with [0056]:”results are shown in FIGS. 6-8…FIG. 6 is a plot of predicted depth versus actual measured depth for a first ML model (model A); FIG. 7 is a plot of predicted depth versus actual measured depth for a second ML model (model B); and FIG. 8 is a plot of predicted depth versus actual measured depth for a third ML model (model C).”)
HONDA does not teach:
wherein the machine learning model further uses composite reference parameters for the modified reference device based on a merger of reference parameters of the reference device
QUI further teaches:
wherein the machine learning model further uses composite reference parameters for the modified reference device based on a merger of reference parameters of the reference device (QUI teaches, as above, implementation of composite parameters, [0019]: “beneficial effects…adopting the neighborhood association method and the similarity difference compensation method, the associated connection between historical actual measurement data and calculation simulation data is realized…using high-fidelity historical actual measurement data, such that the difference compensation fusion between the calculation simulation data and the historical actual measurement data can be effectively realized”; and as reiterated in Claim 1.)
It would have been obvious to one of ordinary skill in the art before effective filing date of the claimed invention to further modify HONDA, as modified by QUI as taught above, to include method steps of using composite reference parameters for the modified reference device based on a merger of reference parameters of the reference device, such as that further disclosed by QUI because it would allow for the enhancement of the method taught by HONDA by improving accuracy, robustness, and generalization of machine learning models for device characterization. One of ordinary skill would see the advantage of this obvious combination as a way to mitigate inconsistencies in device data to improve the accuracy and reliability of the training process, and ultimately improve the final model of a device and device parameter.
With respect to Claim 6, HONDA, in view of QUI teaches the limitations of Claim 3.
HONDA further teaches:
wherein the second model of the modified reference device is produced by changing at least one parameter of the first model. (HONDA teaches variation to develop refined models, as above, [0006]; and [0034]: “process parameters and other metrology from upstream processes and metrology can also be used to train a machine learning algorithm that is focused on the overlay error”; and FIGs 6-8 depicting use of parameter variation to develop models.)
With respect to Claim 7, HONDA, in view of QUI teaches the limitations of Claim 3.
HONDA does not teach:
wherein second synthetic metrology data for the second model of the modified reference device is generated based on a variation between metrology data measured from the reference device and first synthetic metrology data for the first model.
QUI further teaches:
wherein second synthetic metrology data for the second model of the modified reference device is generated based on a variation between metrology data measured from the reference device and first synthetic metrology data for the first model. (QUI teaches successive model development, [0005]: “based on heterogeneous data difference compensation fusion…uses a BP neural network model as a prediction model”; QUI teaches using model data with measured data, [0041]: “associating the encoded calculation simulation data set ESets with the historical actual measurement training set ESethtrain by using the neighborhood association method.”; and in Claim 3.)
It would have been obvious to one of ordinary skill in the art before effective filing date of the claimed invention to further modify HONDA, as modified by QUI, as taught above, to include method steps of wherein second synthetic metrology data for the second model of the modified reference device is generated based on a variation between metrology data measured from the reference device and first synthetic metrology data for the first model, such as that further disclosed by QUI because this technique provides way to leverage the method of HONDA to include a more advanced training technique that allows for better analysis of discrepancies between real measurements from a reference device and the predictions of a first model. This step provides a motivating advantage to improve device parameter modeling in a way that is compatible with logistics that require real-time measured data.
With respect to Claim 8, , HONDA, in view of QUI teaches the limitations of Claim 3.
HONDA does not teach:
wherein the composite metrology data is generated by modifying metrology data measured from the reference device with a determined difference between first synthetic metrology data for the first model and second synthetic metrology data for the second model of the modified reference device.
QUI further teaches:
wherein the composite metrology data is generated by modifying metrology data measured from the reference device with a determined difference between first synthetic metrology data for the first model and second synthetic metrology data for the second model of the modified reference device. (QUI teaches, as above, use of composite data, and further teaches model refinement using model generated data, [0052]: “2) inputting the samples into the model in sequence, and performing forward calculation to calculate the corresponding output”; and Claim 3: “model under the current combination of h1, h2 and h3 is trained by using the training sample set, and the model obtained by training is verified by using the verification sample…optimized by using the tabu search algorithm…if requirements are met, the model is selected as the final prediction model, or if requirements are not met, the number of hidden layers of the BP neural network model is reset, and a new network model is retrained.”)
It would have been obvious to one of ordinary skill in the art before effective filing date of the claimed invention to further modify HONDA, as modified by QUI, as taught above, to include method steps of wherein the composite metrology data is generated by modifying metrology data measured from the reference device with a determined difference between first synthetic metrology data for the first model and second synthetic metrology data for the second model of the modified reference device, such as that further disclosed by QUI because it would be understood as a way to minimize the need for extensive, and possibly costly physical “real-time” measurements necessary for accurate model development. One of ordinary skill would understand that combining the modeling techniques of QUI with the machine learning method of HONDA would improve the training process to enable meaningful virtual predictive models.
With respect to Claim 9, HONDA, in view of QUI teaches the limitations of Claim 3.
wherein the composite metrology data comprises a plurality of sets of composite metrology data for a corresponding plurality of modified reference devices.
HONDA further teaches:
metrology data comprises a plurality of sets of metrology data (HONDA teaches, as above, use of multiple data sets for ML model development, [0004].)
HONDA does not teach:
composite metrology data comprises a plurality of sets of composite metrology data
QUI further teaches:
composite metrology data comprises a plurality of sets of composite metrology data corresponding plurality of modified reference devices. (QUI teaches, as above, use of composite data for ML model development, [0052] “2) inputting the samples into the model in sequence, and performing forward calculation to calculate the corresponding output; 3) calculating the loss lbatch of the training samples of the batch size according to the loss function; 4) performing error back propagation, and updating the weights and the deviations by using the small batch gradient descent method; 5) repeating 1-4 times until the training samples in the entire training sample set MSets are traversed”)
It would have been obvious to one of ordinary skill in the art before effective filing date of the claimed invention to further modify HONDA, as modified by QUI, as taught above, to include method step of composite metrology data comprises a plurality of sets of composite metrology data, such as that further disclosed by QUI because it would be a way to strengthen the final model by providing a more comprehensive and robust training environment. One of ordinary skill would understand that combining the modeling techniques of QUI with the machine learning method of HONDA would improve the model development process and ultimately provide a more generalized model with improved prediction capacity for a device parameter.
With respect to Claim 10, HONDA, in view of QUI teaches the limitations of Claim 1.
HONDA further teaches:
wherein the measured metrology data comprises measured spectra and the composite metrology data comprises composite spectra. (HONDA teaches collection of spectral data for model development, FIG. 5, with [0014]: “graph plotting spectral intensity as a function of wavelength for three different layer thicknesses”, and FIGs 10-12; and see [0042]: “metrology data can be collected during etch processes. Optical emissions spectra or spectral data from photoluminescence can be utilized as input data”.)
Claims 2, and 11-20 are rejected under 35 U.S.C. 103 as being unpatentable over HONDA, in view of QUI, as applied to Claim 1 above, and further in view of SU (US 20200356011 A1).
With respect to Claim 2, HONDA, in view of QUI teaches the limitations of Claim 1.
HONDA, as modified by QUI as taught above, does not teach:
wherein the first model is produced by fitting metrology data measured from the reference device to synthetic metrology data for the first model.
SU teaches:
wherein the first model is produced by fitting metrology data measured from the reference device to synthetic metrology data for the first model. (SU is in same technical area, [0006] “method comprising: obtaining training data set including an optical proximity correction corresponding to a spatially shifted version of a training design pattern; and training, by a hardware computer system, a machine learning model configured to predict optical proximity corrections”; SU teaches fitting data to test model, see [0070]: “a model is usually given a dataset of known data on which training is run (training dataset), and a dataset of unknown data (or first seen data) against which the model is tested (testing dataset). The goal of cross validation is to define a dataset to “test” the model in the training phase (i.e., the validation dataset), in order to limit problems like overfitting, give an insight on how the model will generalize to an independent data set”)
It would have been obvious to one of ordinary skill in the art before effective filing date of the claimed invention to further modify HONDA, as modified by QUI, as taught above, to include method step of producing first model by fitting metrology data measured from the reference device to synthetic metrology data for the first model, such as that of SU because provide a way to enhance and better use the machine learning method disclosed by HONDA as modified by QUI to ensure a more reliable and accurate model of a device or product under test. This combination of SU with HONDA as modified by QUI is an obviousness combination to improve a model by using comparative fitting techniques early in the model development process as a way to produce a best model for a specific measured device or product.
With respect to Claim 11, HONDA teaches:
A metrology system configured for supporting characterizing a device on a sample (HONDA method and system, as above, discussed in Claim 1, [0025] and [0030])
obtain measured metrology data from the device; (HONDA teaches, as above, discussed in Claim 1, obtaining metrology data from device, FIG. 1 with [0025].)
determine, based on the measured metrology data, at least one parameter of the device with a machine learning model (HONDA teaches using measured device data to implement ML model development, as above, Claim 1, [0004].)
wherein the first model of the reference device is a physical model of the reference device. (HONDA teaches use of a physical model, with interpretation as above, Claim 1)
HONDA does not teach:
a source configured to generate radiation to be incident on the device on the sample;
at least one detector configured to detect radiation from the device produced in response to the radiation that is incident on the device; and
at least one processor coupled to the at least one detector, wherein the at least one processor is configured to: obtain metrology data from device,
a machine learning model that uses composite metrology data,
wherein each composite metrology datum comprises a merger of a metrology datum measured for a reference device and a first synthetic metrology datum for a first model of the reference device.
QUI teaches:
a machine learning model that uses composite metrology data, wherein each composite metrology datum comprises a merger of a metrology datum measured for a reference device and a first synthetic metrology datum for a first model of the reference device. (QUI teaches this limitation, parallel to that found in Claim 1, above [0006], [0009], and FIG.1, box 5, with [0010])
It would have been obvious to one of ordinary skill in the art before effective filing date of the claimed invention to modify the method disclosed by HONDA to include in a system, based on metrology data , at least one parameter] with a machine learning model that uses composite metrology data, wherein each composite metrology datum comprises a merger of a metrology datum measured for a reference device and a first synthetic metrology datum for a first model of the reference device, such as that of QUI because this method provides a way to increase data volume and diversity to produce a more reliable and accurate model. The motivational advantages of combining the more advance machine-learning technique taught by QUI with the measurement system and method for modeling taught by HONDA include the impact of merged (composite) data to help mitigate impact of noise or unexpected anomalies among individual datum, reduce data sparsity, and improve the efficiency of model development, saving computational time and expense.
HONDA, as modified by QUI, as taught above, does not teach:
a source configured to generate radiation to be incident on the device on the sample;
at least one detector configured to detect radiation from the device produced in response to the radiation that is incident on the device; and
at least one processor coupled to the at least one detector, wherein the at least one processor is configured to: obtain metrology data from device,
SU teaches:
a source configured to generate radiation to be incident on the device on the sample; (SU is directed specifically to optical metrology, [0006]; SU teaches optical source, see FIG.1, with [0054]: “FIG. 1 illustrates an exemplary lithographic projection apparatus …components include illumination optics…which may be a deep-ultraviolet excimer laser source or other type of source”)
at least one detector configured to detect radiation from the device produced in response to the radiation that is incident on the device; (SU teaches optical measurements, FIG. 2 and [0056]: “measuring device may comprise an optical measurement device configured to measure a physical parameter of the substrate”; Examiner notes one of ordinary skill would understand “measurement device” would imply a detector in the case of optical metrology.)
at least one processor coupled to the at least one detector (SU teaches basic computational/computer interfacing components that would be known to one of ordinary skill for computer-based data acquisition and computational processes, [0173]: “computer system 100 in response to processor 104 executing one or more sequences of one or more instructions”, or [0176]: “Computer system 100 may include a communication interface 118 coupled to bus 102. communication interface 118 sends and receives electrical, electromagnetic or optical signals”)
It would have been obvious to one of ordinary skill in the art before effective filing date of the claimed invention to further modify HONDA, as modified by QUI as taught above to include components of a at least one processor source configured to generate radiation to be incident on the device on the sample and at least one detector configured to detect radiation from the device produced in response to the radiation that is incident on the device and at least one processor coupled to the at least one detector, such as that of SU because this technique is an efficient way to practically implement the acquisition of measured sample data for use in a machine learning model, as taught by HONDA, modified by QUI. While HONDA does teach optical measurements, SU explicitly recites a source and detector for such data acquisition. It would be seen as an obvious advantage to include such components to practically functionalize the disclosure of HONDA as modified by QUI to broaden the range of applications to include optical data. Examiner notes that while HONDA and QUI do not explicitly recite presence of standard computational components (i.e., processor), one of ordinary skill would understand such components are implied based on functions and processes described therein, making the combination with SU obvious and reasonable.
With respect to Claim 12, HONDA, in view of QUI, and further in view of SU teaches the limitations of Claim 11.
HONDA does not teach
wherein the first model is produced by fitting metrology data measured from the reference device to synthetic metrology data for the first model.
SU further teaches:
wherein the first model is produced by fitting metrology data measured from the reference device to synthetic metrology data for the first model. (SU teaches this limitation, as discussed above for parallel limitation in Claim 2, [0006], [0070])
It would have been obvious to one of ordinary skill in the art before effective filing date of the claimed invention to further modify HONDA, as modified by QUI and SU, as taught above, to include method step of producing first model by fitting metrology data measured from the reference device to synthetic metrology data for the first model, such as that further disclosed by SU because it would be understood as a way to enhance and better use the machine learning method disclosed by HONDA with QUI to ensure a more reliable and accurate model of a device or product under test. One of ordinary skill would understand the advantage and obviousness of this combination to use fitting techniques early in the model development process as a way to produce a best model for a specific measured device or product.
With respect to Claim 13, HONDA, in view of QUI, and further in view of SU teaches the limitations of Claim 11.
HONDA does not teach:
wherein each composite metrology datum comprises the merger of the metrology datum and the first synthetic metrology datum and further a second synthetic metrology datum for a second model of a modified reference device that is changed with respect to the first model.
QUI further teaches:
wherein each composite metrology datum comprises the merger of the metrology datum and the first synthetic metrology datum and further a second synthetic metrology datum for a second model of a modified reference device that is changed with respect to the first model. (QUI teaches use of composite data, as above, and further teaches model refinement using model generated data, [0052])
It would have been obvious to one of ordinary skill in the art before effective filing date of the claimed invention to further modify HONDA, as modified by QUI and SU, as taught above, to include in a system, wherein each composite metrology datum comprises the merger of the metrology datum and the first synthetic metrology datum and further a second synthetic metrology datum for a second model of a modified reference device that is changed with respect to the first model, such as that further disclosed by QUI because it would be a way to optimize model development while avoiding the need for extensive, and possibly costly physical “real-time” measurements necessary for accurate model development. Combining the modeling techniques of QUI with the machine learning method of HONDA would improve the training process to enable meaningful virtual predictive models.
With respect to Claim 14, HONDA, in view of QUI, and further in view of SU teaches the limitations of Claim 13.
HONDA further teaches:
wherein the second model of the modified reference device has at least one parameter that is varied with respect to the first model. (HONDA teaches implementing variation to develop refined models, [0006]: “pick design parameters that are insensitive to known manufacturing and environmental variations”; and [0034]: “process parameters and other metrology from upstream processes and metrology can also be used to train a machine learning algorithm that is focused on the overlay error”; and as above, FIGs 6-8 depicting use of parameter variation to develop models.)
With respect to Claim 15, HONDA, in view of QUI, and further in view of SU teaches the limitations of Claim 13.
HONDA does not teach:
wherein the machine learning model further uses composite reference parameters for the modified reference device based on a merger of reference parameters of the reference device, a first set of key parameter values generated for the first model, and a second set of key parameters values generated for the second model.
QUI further teaches:
wherein the machine learning model further uses composite reference parameters for the modified reference device based on a merger of reference parameters of the reference device, a first set of key parameter values generated for the first model, and a second set of key parameters values generated for the second model. (QUI teaches composite data techniques, as discussed above, [0006] and [0017].)
It would have been obvious to one of ordinary skill in the art before effective filing date of the claimed invention to further modify HONDA, as modified by QUI and SU as taught above, to include method step of wherein each composite metrology datum comprises the merger of the metrology datum and the first synthetic metrology datum and further a second synthetic metrology datum for a second model of a modified reference device that is changed with respect to the first model, such as that further disclosed by QUI because provides an advantageous way to avoid the need for measurements that require time and expense in the goal of developing an accurate and reliable model. Combining this further technique taught by QUI with the method of HONDA to use composite data would address “real-world” or “real-time” data scarcity to provide sufficient opportunity for a model to learn and improve and ultimately make accurate predictions. This motivation is suggested by HONDA (see, for example, [0005]). One of ordinary skill would also understand that using composite data for model training can address issues of bias from measured data that may occur during a measurement operation
With respect to Claim 16, HONDA, in view of QUI, and further in view of SU teaches the limitations of Claim 13.
wherein the second model of the modified reference device is produced by changing at least one parameter of the first model.
HONDA further teaches:
wherein the second model of the modified reference device is produced by changing at least one parameter of the first model. (HONDA teaches variation to develop refined models, as above [0006]; and [0034]: “process parameters and other metrology from upstream processes and metrology can also be used to train a machine learning algorithm that is focused on the overlay error”; and as above, FIGs 6-8 depicting use of parameter variation to develop models.)
With respect to Claim 17, HONDA, in view of QUI, and further in view of SU teaches the limitations of Claim 13.
HONDA does not teach:
wherein second synthetic metrology data for the second model of the modified reference device is generated based on a variation between metrology data measured from the reference device and first synthetic metrology data for the first model.
QUI further teaches:
wherein second synthetic metrology data for the second model of the modified reference device is generated based on a variation between metrology data measured from the reference device and first synthetic metrology data for the first model. (As above, parallel limitation, Claim 3, [0005]; QUI teaches using model data, [0041]; and see Claim 3.)
It would have been obvious to one of ordinary skill in the art before effective filing date of the claimed invention to further modify HONDA, as modified by QUI and SU, as taught above, to include method steps of wherein second synthetic metrology data for the second model of the modified reference device is generated based on a variation between metrology data measured from the reference device and first synthetic metrology data for the first model, such as that further disclosed by QU because it would be understood as a way to leverage the method of HONDA to include a more advanced training technique that would allow for better analysis of discrepancies between real measurements from a reference device and the predictions of a first model. One of ordinary skill would see the advantage of this combination to improve device parameter modeling in a way that is compatible with logistics that require real-time measured data.
With respect to Claim 18, HONDA, in view of QUI, and further in view of SU teaches the limitations of Claim 13.
HONDA does not teach:
wherein the composite metrology data is generated by modifying metrology data measured from the reference device with a determined difference between first synthetic metrology data for the first model and second synthetic metrology data for the second model of the modified reference device.
QUI further teaches:
wherein the composite metrology data is generated by modifying metrology data measured from the reference device with a determined difference between first synthetic metrology data for the first model and second synthetic metrology data for the second model of the modified reference device. (QUI teaches, as above, use of composite data, and further teaches model refinement using model generated data, [0052].)
It would have been obvious to one of ordinary skill in the art before effective filing date of the claimed invention to further modify HONDA, as modified by QUI and SU, as taught above, to include in a metrology system, the steps of wherein the composite metrology data is generated by modifying metrology data measured from the reference device with a determined difference between first synthetic metrology data for the first model and second synthetic metrology data for the second model of the modified reference device, such as that further disclosed by QUI because it provides as a way to avoid additional and possibly costly physical “real-time” measurements necessary for accurate model development. One of ordinary skill would understand that combining the modeling techniques of QUI with the machine learning method of HONDA as already modified by QUI and SU, would improve the training process to enable meaningful virtual predictive models.
With respect to Claim 19, HONDA, in view of QUI, and further in view of SU teaches the limitations of Claim 13.
HONDA further teaches:
wherein metrology data comprises a plurality of sets of metrology data (HONDA teaches use of multiple data sets for ML model development, as above, [0004])
HONDA does not teach:
wherein the composite metrology data comprises a plurality of sets of composite metrology data for a corresponding plurality of modified reference devices.
QUI further teaches:
wherein the composite metrology data comprises a plurality of sets of composite metrology data for a corresponding plurality of modified reference devices. (QUI teaches, as above, use of composite data for ML model development, [0052].)
It would have been obvious to one of ordinary skill in the art before effective filing date of the claimed invention to further modify HONDA, as modified by QUI and SU as taught above, to include in a metrology system steps of composite metrology data comprises a plurality of sets of composite metrology data, such as that further disclosed by QUI because it would be a way to strengthen the final model by providing a more comprehensive and robust training environment. Combining the modeling techniques of QUI with the machine learning method of HONDA would improve the model development process and ultimately provide a more generalized model with improved prediction capacity for a device parameter.
With respect to Claim 20, HONDA, in view of QUI, and further in view of SU teaches the limitations of Claim 11.
HONDA further teaches:
wherein the measured metrology data comprises measured spectra and the composite metrology data comprises composite spectra. (See above, discussion of parallel limitation, Claim 10, HONDA teaches collection of spectral data for model development, see FIG. 5, with [0014].)
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
PANDEV (US 10215559 B2) - teaches similar machine learning based techniques for metrology in a manufacturing setting, including explicit use of a physical model.
ROTHSTEIN (US 20230017097 A1) - teaches similar machine learning based methods for metrology of devices, specific to field of semiconductor manufacturing, including use of a physical model for reference in model development.
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 TONI D SAUNCY whose telephone number is (703)756-4589. The examiner can normally be reached Monday - Friday 8:30 a.m. - 5:30 p.m. ET.
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, Catherine Rastovski can be reached at 571-270-0349. 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.
/TONI D SAUNCY/Examiner, Art Unit 2857
/Catherine T. Rastovski/Supervisory Primary Examiner, Art Unit 2857