DETAILED ACTION
This office action addresses Applicant’s response filed on 27 January 2026. Claims 1-4, 7-10, 13-18, and 22-27 are pending.
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
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.
Claim(s) 1-3, 7, 22, 24, and 27 is/are rejected under 35 U.S.C. 103 as being unpatentable over Tripodi (US 2019/0378012) in view of Ravid (US 2008/0243433), Liman (US 2020/0335406), and Chung (US 2016/0041548).
Regarding claim 1,Tripodi discloses a method for training a machine learning model to predict metrology measurements of a current substrate being processed at a manufacturing system (¶74), the method comprising:
identifying historical spectral data collected during performance of a prior substrate process for a prior substrate using manufacturing equipment of the manufacturing system and extracting one or more spectral data items from the historical spectral data wherein the one or more spectral data items and the correspond to a particular portion of the prior substrate (¶¶76-77, 86);
generating training data for the machine learning model, wherein generating the training data comprises:
a mapping between the one or more spectral data items and coordinates for the portion of the prior substrate (¶86): generating a first training input comprising historical spectral data associated with a portion of a prior substrate previously processed at manufacturing equipment of the manufacturing system (¶¶76-77); and
generating a first target output for the first training input, wherein the first target output comprises historical metrology measurements associated with the particular portion of the prior substrate at the coordinates, the historical metrology measurements collected at least one of during or after the performance of the prior substrate process (¶¶79, 83); and
providing the training data to train the machine learning model on (i) a set of training inputs comprising the first training input and (ii) a set of target outputs comprising the first target output (¶79).
If Tripodi is found to be unclear regarding historical spectral data associated with a portion of a prior substrate previously processed at the manufacturing system, Ravid discloses the same (¶8). Additionally, persons having ordinary skill in the art, reading Tripodi, would understand that Tripodi’s locations would comprise coordinates, since the substrate locations corresponding to training samples would typically or necessarily be defined by coordinates. Nevertheless, if Tripodi is found to be unclear regarding coordinates, Ravid discloses the same (Figs. 5 and 7). Finally, Tripodi does not appear to explicitly disclose identifying historical non-spectral data associated with the prior substrate, extracting the one or more non-spectral data items corresponding to a particular portion of the prior substrate, and generating a first training input comprising the one or more spectral data items, one or more non-spectral data items, and coordinates for the portion of the prior substrate. Ravid discloses
identifying historical spectral data collected for a prior substrate previously processed at manufacturing equipment of the manufacturing system and historical non-spectral data associated with the prior substrate (Figs. 5 and 7; ¶¶8, 59), extracting one or more spectral data items from the historical spectral data and one or more non-spectral data items from the historical non-spectral data, wherein the one or more spectral data items and the one or more non-spectral data items correspond to a particular portion of the prior substrate (Figs. 5 and 7; ¶8), and generating a first training input comprising the one or more spectral data items, the one or more non-spectral data items, and coordinates for the portion of the prior substrate (Figs. 5, 7, and 8).
Liman also discloses that the training inputs comprises historical spectral data and historical non-spectral data (¶124).
It would have been obvious to persons having ordinary skill in the art before the effective filing date of the application to combine the teachings of Tripodi, Ravid, and Liman, because doing so would have involved merely the routine use of a known technique to improve similar devices in the same way to achieve the predictable results of training a neural network (NN) using additional information such as non-spectral data and associated coordinates for the spectral and non-spectral data, in order to improve NN prediction. KSR Int’l Co. v. Teleflex Inc., 82 U.S.P.Q.2d 1385, 1396. Tripodi discloses metrology NNs for predicting characteristics of a structure on a wafer, the NN being trained with training samples comprising spectral data and associated locations. Liman teaches using both historical spectral data and historical non-spectral data to improve NN predictions. Ravid teaches that the spectral, non-spectral, and location/coordinate data should be mapped to each other to provide data for metrology processes. The teachings of Ravid and Liman are directly applicable to Tripodi in the same way, so that Tripodi’s NN would use historical spectral data, historical non-spectral data, and associated coordinates to improve predictions of target characteristics.
Tripodi does not appear to explicitly disclose that the historical non-spectral data is sensor data collected by one or more sensors of the manufacturing equipment and representing at least one of a state of one or more components of the manufacturing equipment or an interior environment of the manufacturing equipment during the performance of the prior substrate process for the prior substrate. Chung discloses these limitations (¶¶27, 34, 41); Chung also discloses that the historical metrology measurements collected at least one of during or after the performance of the prior substrate process (¶42). It would have been obvious to persons having ordinary skill in the art before the effective filing date of the application to combine the teachings of Tripodi, Ravid, Liman, and Chung, because doing so would have involved merely the routine use of a known technique to improve similar devices in the same way to achieve the predictable results of considering processing parameters in predicting metrology measurements. KSR Int’l Co. v. Teleflex Inc., 82 U.S.P.Q.2d 1385, 1396. Tripodi discloses training models for predicting metrology measurements. Chung teaches that models for predicting metrology measurements incorporate sensor data from processing equipment. The teachings of Chung are directly applicable to Tripodi in the same way, so that Tripodi’s metrology measurement prediction would similarly consider processing equipment parameters.
Regarding claim 2, Tripodi discloses receiving, from a substrate measurement subsystem of the manufacturing system, a first set of measurements associated with the prior substrate, wherein the first set of measurements comprises the historical spectral data (¶¶76). Tripodi does not appear to explicitly disclose additional historical non-spectral data associated with the prior substrate, and wherein the first training input further comprises the additional historical non-spectral data. Ravid (¶59) and Liman (¶124) disclose these limitations. Motivation to combine remains consistent with claim 1.
Regarding claim 3, Tripodi discloses that generating the first target output comprises: receiving, from a metrology system communicatively coupled to the manufacturing system, the historical metrology measurements associated with the prior substrate previously processed at the manufacturing system, wherein the first target output is generated based on the received historical metrology measurements (¶¶79, 83).
Regarding claim 7, Tripodi discloses that each training input of the set of training inputs is mapped to a target output of the set of target outputs (¶¶74, 79). Motivation to combine remains consistent with claim 1.
Regarding claim 22, Tripodi does not appear to explicitly disclose that the first training input further comprises at least one of eddy current data or capacitance data. Liman discloses the same (¶124). Motivation to combine remains consistent with claim 1.
Regarding claim 24, Tripodi does not appear to explicitly disclose that the manufacturing equipment comprises at least one of a processing chamber, a load lock, or a transfer chamber. Chung discloses these limitations (¶¶3, 27). Motivation to combine remains consistent with claim 1.
Regarding claim 27, Tripodi does not appear to explicitly disclose that the sensor data comprises at least one of a temperature of a heater of a substrate support assembly of the manufacturing equipment, a voltage of an electrostatic chuck of the substrate support assembly, an electrical current of the one or more components of the manufacturing equipment, a power of the one or more components, a voltage of the one or more components, a temperature of the interior environment of the manufacturing equipment, or a pressure of the interior environment; Chung discloses these limitations (¶¶27, 41). Motivation to combine remains consistent with claim 1.
Claim(s) 4 is/are rejected under 35 U.S.C. 103 as being unpatentable over Tripodi in view of Ravid, Liman, Chung, and Hu (US 2017/0024509).
Regarding claim 4, Tripodi discloses that generating the first target output comprises: receiving, from a client device of the manufacturing system, the historical metrology measurements associated with the prior wafer previously processed at the manufacturing system, wherein the first target output is generated based on the received historical metrology measurements (¶¶79, 83). If Tripodi is found to be unclear regarding receiving the metrology measurements from a client device of the manufacturing system, Hu discloses the same (Fig. 1; ¶¶24, 26). It would have been obvious to persons having ordinary skill in the art before the effective filing date of the application to combine the teachings of Tripodi, Ravid, Liman, Chung, and Hu, because doing so would have involved merely the routine combination of known elements according to known techniques to produce merely the predictable results of allowing computerized control of optical metrology and generation of metrology measurements. KSR Int’l Co. v. Teleflex Inc., 82 U.S.P.Q.2d 1385, 1395. Tripodi discloses receiving metrology results from a metrology tool. Persons having ordinary skill in the art, reading Tripodi, would understand that the metrology tool includes a client device that produces the measured results. Hu provides explicit disclosure of the metrology tool including a client device that controls the metrology tool and produces the measurement results. The teachings of Hu are directly applicable to Tripodi in the same way, so that Tripodi’s optical metrology tool would similarly include a client device to allow control of the metrology tool and generation of metrology results.
Claim(s) 8 is/are rejected under 35 U.S.C. 103 as being unpatentable over Tripodi in view of Li, Feng, Liman, Chung, and Ophir (US 2022/0318987).
Regarding claim 8, Tripodi does not appear to explicitly disclose that the machine learning model is configured to generate one or more outputs indicating a level of confidence of a metrology measurement for the current substrate being processed at the manufacturing system. Ophir discloses these limitations (¶21). It would have been obvious to persons having ordinary skill in the art before the effective filing date of the application to combine the teachings of Tripodi, Ravid, Liman, Chung, and Ophir, because doing so would have involved merely the routine use of a known technique to improve similar processes in the same way to achieve the predictable results of indicating confidence of results estimated by the machine learning model and determining need for, e.g., re-training the model. KSR Int’l Co. v. Teleflex Inc., 82 U.S.P.Q.2d 1385, 1396. Tripodi is directed to training a machine learning model to predict metrology measurements. Ophir teaches determining a confidence score for results predicted by the machine learning model. The teachings of Ophir are directly applicable to Tripodi in the same way, so that Tripodi would similarly determine a confidence score for results predicted by the model, to indicate confidence of the results and need for, e.g., retraining the model.
Claim(s) 9, 10, 13, and 16-18 is/are rejected under 35 U.S.C. 103 as being unpatentable over Tripodi in view of Ravid, Liman, Feng (US 2021/0035833), and Chung.
Regarding claim 9, Tripodi discloses an apparatus comprising: a memory to store a trained machine learning model (¶74); and a processing device coupled to the memory (¶74), the processing device to:
identify spectral data collected during a performance of a current substrate process for a current substrate using manufacturing equipment of a manufacturing system and extract one or more spectral data items from the spectral data, wherein the spectral data items correspond to a target portion of the current substrate (¶82);
provide the one or more spectral data items and coordinates for the portion of the current substrate as input to the trained machine learning model (¶¶82, 86); obtain one or more outputs from the trained machine learning model and extract, from the one or more outputs, a metrology measurement for the target portion of the current substrate being processed at the manufacturing system (¶¶74, 82).
Persons having ordinary skill in the art, reading Tripodi, would understand that Tripodi’s locations would comprise coordinates, since the substrate locations corresponding to training samples would typically or necessarily be defined by coordinates. Nevertheless, if Tripodi is found to be unclear regarding coordinates, Ravid discloses the same (Figs. 5 and 7). Tripodi also does not appear to explicitly disclose identifying non-spectral data associated with the substrate and extracting the one or more non-spectral data items corresponding to a target portion of the substrate. Ravid discloses these limitations (Fig. 5). Liman also discloses that the training inputs comprises historical spectral data and historical non-spectral data (¶124).
It would have been obvious to persons having ordinary skill in the art before the effective filing date of the application to combine the teachings of Tripodi, Ravid, and Liman, because doing so would have involved merely the routine use of a known technique to improve similar devices in the same way to achieve the predictable results of training a neural network (NN) using additional information such as non-spectral data and associated coordinates for the spectral and non-spectral data, in order to improve NN prediction. KSR Int’l Co. v. Teleflex Inc., 82 U.S.P.Q.2d 1385, 1396. Tripodi discloses metrology NNs for predicting characteristics of a structure on a wafer, the NN being trained with training samples comprising spectral data and associated locations. Liman teaches using both historical spectral data and historical non-spectral data to improve NN predictions. Ravid teaches using associated spectral, non-spectral, and location/coordinate data for metrology processes. The teachings of Ravid and Liman are directly applicable to Tripodi in the same way, so that Tripodi’s NN would use historical spectral data, historical non-spectral data, and associated coordinates to improve predictions of target characteristics.
If Tripodi, Ravid, and/or Liman is found to be unclear regarding providing spectral data, non-spectral data, or coordinates from current substrates as inputs to the trained machine learning model, it is well-known that for inference using machine learning models, inputs to the trained model should correspond to the inputs used to train the model, as taught by Feng (¶¶29, 53); i.e. models are trained using values of the variables/features that will be input to the trained model during inference to ensure accuracy. It would have been obvious to persons having ordinary skill in the art before the effective filing date of the application to combine the teachings of Tripodi, Ravid, Liman, and Feng, because doing so would have involved merely the routine use of a known technique to improve similar devices in the same way to achieve the predictable results of ensuring NN prediction accuracy by using input variables corresponding to variables in the training data. KSR Int’l Co. v. Teleflex Inc., 82 U.S.P.Q.2d 1385, 1396. Tripodi discloses a metrology NN that uses data about a substrate structure, including coordinates, to predict characteristics of the structure. It is well-known that input variables for NN inference/prediction should correspond to the variables used to train the NN, as taught by Feng, to ensure NN prediction accuracy. The well-known principle, taught by Feng, is directly applicable to Tripodi in the same way, so that Tripodi would similarly use input variables to the trained NN that correspond to the variables used to train the NN, such as coordinates of substrate structures, to ensure NN prediction accuracy.
Tripodi does not appear to explicitly disclose that the historical non-spectral data is sensor data collected by one or more sensors of the manufacturing equipment and representing at least one of a state of one or more components of the manufacturing equipment or an interior environment of the manufacturing equipment during the performance of the prior substrate process for the prior substrate. Chung discloses these limitations (¶¶27, 34, 41). It would have been obvious to persons having ordinary skill in the art before the effective filing date of the application to combine the teachings of Tripodi, Ravid, Liman, and Chung, because doing so would have involved merely the routine use of a known technique to improve similar devices in the same way to achieve the predictable results of considering processing parameters in predicting metrology measurements. KSR Int’l Co. v. Teleflex Inc., 82 U.S.P.Q.2d 1385, 1396. Tripodi discloses training models for predicting metrology measurements. Chung teaches that models for predicting metrology measurements incorporate sensor data from processing equipment. The teachings of Chung are directly applicable to Tripodi in the same way, so that Tripodi’s metrology measurement prediction would similarly consider processing equipment parameters.
Regarding claim 10, Tripodi discloses that the processing device is further to: receive, from a substrate measurement subsystem of the manufacturing system, a set of measurements associated with the current substrate, the set of measurements comprising the spectral data (¶¶74, 84). Tripodi does not appear to explicitly disclose the set of measurements comprising additional non-spectral data associated with the current substrate, wherein the additional non-spectral data is provided as an additional input to the trained machine learning model. Liman discloses the same (¶124); as noted above with regard to claim 9, persons having ordinary skill in the art would understand that inputs to the trained model for inference/prediction would correspond to the inputs used to train the model, as taught by Feng (¶29, 53). Motivation to combine remains consistent with claim 9.
Regarding claim 13, Tripodi does not appear to explicitly disclose that the processing device is further to: cause, via a client device of the manufacturing system, the metrology measurement for the current substrate being processed at the manufacturing system to be provided to a user of the manufacturing system via a graphical user interface (GUI); Feng discloses the same (¶¶60, 82). It would have been obvious to persons having ordinary skill in the art before the effective filing date of the application to combine the teachings of Tripodi, Ravid, Liman, Chung, and Feng, because doing so would have involved merely the routine use of a known technique to improve similar devices in the same way to achieve the predictable results of allowing a user to see metrology results. KSR Int’l Co. v. Teleflex Inc., 82 U.S.P.Q.2d 1385, 1396. Tripodi discloses a metrology NN that predicts substrate characteristics. Feng teaches providing the predicted characteristics to a user through a GUI. The teachings of Feng are directly applicable to Tripodi in the same way, so that Tripodi would similarly provide the predicted characteristics to a user through a GUI to allow the user to see the metrology results.
Regarding claim 16, Tripodi discloses that the trained machine learning model is trained with an input-output mapping comprising an input and an output, the input based on historical spectral data associated with a surface of a prior substrate previously processed at the manufacturing system, and the output identifying a historical metrology measurement associated with the prior substrate previously processed at the manufacturing system (¶¶74, 79). If Tripodi is found to be unclear regarding these limitations, Feng discloses the same (¶35). Tripodi does not appear to explicitly disclose historical sensor data collected by one or more sensors of the manufacturing equipment and representing at least one of a state of one or more components of the manufacturing equipment during the performance of the prior substrate process or an interior environment of the manufacturing equipment during the performance of the prior substrate process. Chung discloses these limitations (¶¶27, 34, 41). Motivation to combine remains consistent with claim 9.
Regarding claim 17, Tripodi discloses a non-transitory computer readable storage medium comprising instructions that, when executed by a processing device, cause the processing device (¶74) to:
receive input spectral data associated with a current substrate processed at manufacturing equipment of a manufacturing system and extract one or more spectral data items from the spectral data, wherein the spectral data items correspond to a target portion of the current substrate (¶82);
process the one or more spectral data items and coordinates for the target portion of the current substrate using a trained machine learning model (¶¶82, 86); obtain, based on the processing of the one or more spectral data items and the coordinates for the target portion of the current substrate, one or more outputs indicating a metrology measurement for the portion of the current substrate being processed at the manufacturing system (¶¶74, 82).
Persons having ordinary skill in the art, reading Tripodi, would understand that Tripodi’s locations would comprise coordinates, since the substrate locations corresponding to training samples would typically or necessarily be defined by coordinates. Nevertheless, if Tripodi is found to be unclear regarding coordinates, Ravid discloses the same (Figs. 5 and 7). Tripodi also does not appear to explicitly disclose identifying non-spectral data associated with the substrate and extracting the one or more non-spectral data items corresponding to a target portion of the substrate. Ravid discloses these limitations (Fig. 5). Liman also discloses that the training inputs comprises historical spectral data and historical non-spectral data (¶124).
It would have been obvious to persons having ordinary skill in the art before the effective filing date of the application to combine the teachings of Tripodi, Ravid, and Liman, because doing so would have involved merely the routine use of a known technique to improve similar devices in the same way to achieve the predictable results of training a neural network (NN) using additional information such as non-spectral data and associated coordinates for the spectral and non-spectral data, in order to improve NN prediction. KSR Int’l Co. v. Teleflex Inc., 82 U.S.P.Q.2d 1385, 1396. Tripodi discloses metrology NNs for predicting characteristics of a structure on a wafer, the NN being trained with training samples comprising spectral data and associated locations. Liman teaches using both historical spectral data and historical non-spectral data to improve NN predictions. Ravid teaches using associated spectral, non-spectral, and location/coordinate data for metrology processes. The teachings of Ravid and Liman are directly applicable to Tripodi in the same way, so that Tripodi’s NN would use historical spectral data, historical non-spectral data, and associated coordinates to improve predictions of target characteristics.
If Tripodi, Ravid, and/or Liman is found to be unclear regarding providing spectral data, non-spectral data, or coordinates from current substrates as inputs to the trained machine learning model, it is well-known that for inference using machine learning models, inputs to the trained model should correspond to the inputs used to train the model, as taught by Feng (¶¶29, 53); i.e. models are trained using values of the variables/features that will be input to the trained model during inference to ensure accuracy. It would have been obvious to persons having ordinary skill in the art before the effective filing date of the application to combine the teachings of Tripodi, Ravid, Liman, and Feng, because doing so would have involved merely the routine use of a known technique to improve similar devices in the same way to achieve the predictable results of ensuring NN prediction accuracy by using input variables corresponding to variables in the training data. KSR Int’l Co. v. Teleflex Inc., 82 U.S.P.Q.2d 1385, 1396. Tripodi discloses a metrology NN that uses data about a substrate structure, including coordinates, to predict characteristics of the structure. It is well-known that input variables for NN inference/prediction should correspond to the variables used to train the NN, as taught by Feng, to ensure NN prediction accuracy. The well-known principle, taught by Feng, is directly applicable to Tripodi in the same way, so that Tripodi would similarly use input variables to the trained NN that correspond to the variables used to train the NN, such as coordinates of substrate structures, to ensure NN prediction accuracy.
Tripodi does not appear to explicitly disclose that the historical non-spectral data is sensor data collected by one or more sensors of the manufacturing equipment and representing at least one of a state of one or more components of the manufacturing equipment or an interior environment of the manufacturing equipment during the performance of the prior substrate process for the prior substrate. Chung discloses these limitations (¶¶27, 34, 41). It would have been obvious to persons having ordinary skill in the art before the effective filing date of the application to combine the teachings of Tripodi, Ravid, Liman, and Chung, because doing so would have involved merely the routine use of a known technique to improve similar devices in the same way to achieve the predictable results of considering processing parameters in predicting metrology measurements. KSR Int’l Co. v. Teleflex Inc., 82 U.S.P.Q.2d 1385, 1396. Tripodi discloses training models for predicting metrology measurements. Chung teaches that models for predicting metrology measurements incorporate sensor data from processing equipment. The teachings of Chung are directly applicable to Tripodi in the same way, so that Tripodi’s metrology measurement prediction would similarly consider processing equipment parameters.
Regarding claim 18, Tripodi discloses that the the input spectral data is received from a substrate measurement system of the manufacturing system (¶75).
Claim(s) 14 and 15 is/are rejected under 35 U.S.C. 103 as being unpatentable over Tripodi in view of Ravid, Liman, Feng, Chung, and Ophir.
Regarding claim 14, Feng discloses that the one or more outputs comprise (i) a metrology measurement for a prior wafer processed at the manufacturing system (¶¶32, 34, 60), but does not appear to explicitly disclose a level of confidence that the current substrate being processed at the manufacturing system is associated with the metrology measurement for the prior substrate; Ophir discloses these limitations (¶21). It would have been obvious to persons having ordinary skill in the art before the effective filing date of the application to combine the teachings of Tripodi, Ravid, Liman, Feng, Chung, and Ophir, because doing so would have involved merely the routine use of a known technique to improve similar processes in the same way to achieve the predictable results of indicating confidence of results estimated by the machine learning model and determining need for, e.g., re-training the model. KSR Int’l Co. v. Teleflex Inc., 82 U.S.P.Q.2d 1385, 1396. Tripodi is directed to training a machine learning model to predict metrology measurements. Ophir teaches determining a confidence score for results predicted by the machine learning model. The teachings of Ophir are directly applicable to Tripodi in the same way, so that Tripodi would similarly determine a confidence score for results predicted by the model, to indicate confidence of the results and need for, e.g., retraining the model.
Regarding claim 15, Feng does not appear to explicitly disclose that to extract the metrology measurement for the current wafer being processed at the manufacturing system from the one or more outputs, the processing device is to determine that the level of confidence satisfies a threshold condition. Ophir discloses these limitations (¶21). Motivation to combine remains consistent with claim 14.
Claim(s) 23 is/are rejected under 35 U.S.C. 103 as being unpatentable over Tripodi in view of Ravid, Liman, Chung, and Warnaar (US 2020/0249576).
Regarding claim 23, Tripodi does not appear to explicitly disclose that the first target output comprises at least one of an etch rate, an etch rate uniformity, a critical dimension uniformity, or an edge-to-edge placement error. Warnaar discloses these limitations (¶¶87, 88). It would have been obvious to persons having ordinary skill in the art before the effective filing date of the application to combine the teachings of Tripodi, Ravid, Liman, Chung, and Warnaar, because doing so would have involved merely the routine use of a known technique to improve similar devices in the same way to achieve the predictable results of predicting typical metrology characteristics. KSR Int’l Co. v. Teleflex Inc., 82 U.S.P.Q.2d 1385, 1396. Tripodi discloses NNs for predicting any metrology characteristics. Warnaar teaches that such metrology characteristics include at least one of an etch rate, an etch rate uniformity, a critical dimension uniformity, or an edge-to-edge placement error. The teachings of Warnaar are directly applicable to Tripodi and/or Feng in the same way, so that Tripodi and/or Feng would similarly use the NN to predict typical metrology characteristics.
Claim(s) 25 is/are rejected under 35 U.S.C. 103 as being unpatentable over Tripodi in view of Ravid, Liman, Chung, Hsu (US 8,486,589), and Hashimoto (2019/0346769).
Tripodi discloses that extracting the one or more spectral data items from the historical spectral data comprises: determining, for each spectral data item of the spectral data, whether the respective spectral data item corresponds to a particular structural feature associated with a prior substrate process performed of the prior substrate using the manufacturing equipment, wherein the one or more spectral data items are extracted from the historical spectral data responsive to a determination that the one or more spectral data items and correspond to the particular structural feature (¶¶67, 76, 86). If Tripodi is found to be unclear regarding determining whether data corresponds to a structural feature and extracting the data responsive to the determination, Hsu (claim 12) and Hashimoto (¶120) disclose the same. In particular, Hu and Hashimoto both teach selectively extracting, from existing data, the data that specifically corresponds to structures of interest, which persons having ordinary skill in the art would understand to already be present in Tripodi since Tripodi explicitly discloses training samples comprising data associated with specific targets.
It would have been obvious to persons having ordinary skill in the art before the effective filing date of the application to combine the teachings of Tripodi, Ravid, Liman, Chung, Hsu, and Hashimoto, because doing so would have involved merely the routine use of a known technique to improve similar devices in the same way to achieve the predictable results of extracting training samples for structures of interest from acquired data. KSR Int’l Co. v. Teleflex Inc., 82 U.S.P.Q.2d 1385, 1396. Tripodi discloses collecting various metrology data and generating training samples comprising data for specific targets. Hsu and Hashimoto teach extracting the data associated with structures of interest from the metrology data. The teachings of Hsu and Hashimoto are directly applicable to Tripodi in the same way, so that Tripodi would similarly produce training samples by extracting data associated with structures of interest from acquired metrology data.
Claim(s) 26 is/are rejected under 35 U.S.C. 103 as being unpatentable over Tripodi in view of Ravid, Liman, Feng, Chung, Hsu, and Hashimoto.
Tripodi discloses that to extract the one or more spectral data items from the historical spectral data from the historical non-spectral data the processing device is to: determine, for each spectral data item of the spectral data, whether the respective spectral data item corresponds to a particular structural feature associated with a prior substrate process performed of the prior substrate using the manufacturing equipment, wherein the one or more spectral data items are extracted from the historical spectral data responsive to a determination that the one or more spectral data items correspond to the particular structural feature (¶¶67, 76, 86). If Tripodi is found to be unclear regarding determining whether data corresponds to a structural feature and extracting the data responsive to the determination, Hsu (claim 12) and Hashimoto (¶120) disclose the same. In particular, Hu and Hashimoto both teach selectively extracting, from existing data, the data that specifically corresponds to structures of interest, which persons having ordinary skill in the art would understand to already be present in Tripodi since Tripodi explicitly discloses training samples comprising data associated with specific targets.
It would have been obvious to persons having ordinary skill in the art before the effective filing date of the application to combine the teachings of Tripodi, Ravid, Liman, Feng, Chung, Hsu, and Hashimoto, because doing so would have involved merely the routine use of a known technique to improve similar devices in the same way to achieve the predictable results of extracting training samples for structures of interest from acquired data. KSR Int’l Co. v. Teleflex Inc., 82 U.S.P.Q.2d 1385, 1396. Tripodi discloses collecting various metrology data and generating training samples comprising data for specific targets. Hsu and Hashimoto teach extracting the data associated with structures of interest from the metrology data. The teachings of Hsu and Hashimoto are directly applicable to Tripodi in the same way, so that Tripodi would similarly produce training samples by extracting data associated with structures of interest from acquired metrology data.
Response to Arguments
Applicant’s arguments have been considered but are moot in view of the new grounds of rejection. Applicant asserts that the prior art fails to teach newly-added limitations, which are addressed above using newly-cited prior art.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ARIC LIN whose telephone number is (571)270-3090. The examiner can normally be reached M-F 07:30-17:00 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, Jack Chiang can be reached at 571-272-7483. 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.
23 March 2026
/ARIC LIN/ Examiner, Art Unit 2851