DETAILED ACTION
Claims 1, 2, 5, 7-10, 13-16, 18, and 20-22 are presented for examination.
Claims 1, 9 and 16 have been amended.
Claims 3, 4, 6, 11, 12, 17 and 19 have been cancelled.
Claims 21 and 22 are new.
This office action is in response to the amendment submitted on 18-June-2026.
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Response to Arguments – 35 USC 103
The amendments have been considered. The examiner had initially indicated that the proposed amendments appear to overcome the prior art of record, however, upon further consideration, it does appear the current prior art of record does teach the amended limitations. The rejection is maintained. Additionally, Applicant's arguments fail to comply with 37 CFR 1.111(b) because they amount to a general allegation that the claims define a patentable invention without specifically pointing out how the language of the claims patentably distinguishes them from the references.
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.
Claims 1-2, 5, 7-10, and 13-16, 18 and 20-22 are rejected under 35 U.S.C. 103 as being unpatentable over Guha et al. (US20180082826A1) in view of Malkani et al. (WO2019055576A1) and further in view of Hsu et al. (WO2021113332A1).
Regarding Claim 1, Guha teaches receiving, from a plurality of sensors, sensor data
associated with processing a substrate via a processing chamber of substrate processing equipment
([0015] “A plurality of sensors of the plasma reactor is included, and each of the plurality of sensors is configured to produce a data stream of information during operation of the plasma reactor for carrying out the plasma process”).
the sensor data comprises a first subset received from one or more first sensors and a second subset received from one or more second sensors ([0050] “As shown, the sensors 132 may include one or more of an optical emission spectrometry (OES) sensor, a pressure sensor, a voltage sensor, a current sensor, a temperature sensor, a flow rate sensor, frequency sensor, a power sensor, a metrology sensor, and combinations of two or more thereof. As an example, the following table A illustrates example information that can be obtained from various sensors of a plasma reactor.” EN: The sensors 132 include multiple types of data, each establishing a subset of distinct physical sensors and corresponding data).
the first subset being mapped to the second subset ([0074] “signals, will be classified and correlated in order of relevance to represent Processing State of the reactor. For example, capacitor tuning position in a match system (i.e., coupled to an RF power source) could be correlated with Optical Emission Spectrum (OES) from the plasma and can be correlated together. In a similar way we can define different orders of correlation, with higher order defining strong correlation and lower order defining weak correlation of sensor signals.” [0084] describes using the ML model to create the correlation between the first subset and the second subset. The first subset includes the OES sensor, and the second includes the capacitor position.)
wherein the first subset comprises one or more first types of data and the second subset comprises one or more second types of data that are different from the one or more first types of data (Please see [0050] and table A. The examiner picks any identifiable stream, like OES sensor, to be in the first subset and any remaining identifiable stream, like the pressure and capacitor position, to be in the second subset.)
wherein the second subset comprises chamber component temperature data corresponding to the first subset ([0066] “the data produced by the phenomenological model 184 may be characterized in the form of any one of the outputs produced by the sensors 132, as described with reference to FIG. 1,” and [0050] “the sensors 132 may include one or more of an optical emission spectrometry (OES) sensor, a pressure sensor, a voltage sensor, a current sensor, a temperature sensor”).
inputting, into a180 will be receiving periodic information from etch rate analysis, which may be performed after one or more substrates are tested using a metrology tool. Similar processes can be performed with a monitor wafer, which is configured to approximate the type of processing desired to be executed by the plasma reactor 100…machine learning engine 180 may be provided with information regarding reactor wall surface dynamics 182. This information may include data regarding the inferred characteristics of the chamber wall surfaces, as they change during processing” This describes the data representing the operating space of the chamber).
wherein the model input data comprises the one or more first types of data ([0065] “As an optional refinement input to the machine learning engine 180 input from a phenomenological model 184 is used, that approximates the behavior of the plasma within the chamber, given the reactor wall surface dynamics 182. The phenomenological model 184, in one embodiment, is used to approximate the nature of the chemical reactions that are occurring within the processing volume, and associated interactions with the reactor wall surfaces. Broadly speaking, a phenomenological model is sometimes referred to as a statistical model, as it is a mathematical expression that relates several different empirical observations of phenomena to each other. This relation is consistent with fundamental theory, but is not directly derived from theory. Thus, a phenomenological model does not attempt to explain why the variables in the plasma (i.e., when chemical bonds break to define different chemical species or when the recombine to define a different chemical form upon coming into contact with a surface in the reactor, e.g., a chamber wall). Generally, the phenomenological model 184 is configured to characterize the anticipated chemical kinetics of gases in the plasma of the plasma reactor, and their behavior relative to the reactor wall surface dynamics 182. These kinetics may include, for example, electron collision reactions, wall recombination reactions, wall loss reactions, etc., for different chemistries. Thus, this model simply attempts to describe the relationship, with the assumption that the relationship extends past the measured values. The phenomenological model 184 is configured to produce input to the machine learning engine 180 that is in terms of sensor output. That is, the characterization of the plasma behavior by the phenomenological model 184 is configured to produce input data to the machine learning engine 180 in the form of information similar to that which can be captured by a sensor coupled to the plasma reactor 100.”)
wherein the model output data comprises the one or more second types of data ([0066] and [0067] “Thus, it can be said that the phenomenological model 184 provides input to the machine learning engine 180 in the form of or in terms of sensor output data. The example provided above with respect to OES sensor data is just one example, in the same type of modeling that can be provided for other types of sensor data, such as sensors associated with capacitance, voltage, current, or other measurement characteristics produced by actual sensors that are coupled to the plasma reactor 100.” The phenomenological model’s output (used for the ML input) comprises capacitance, which is from the second type of data as mapped earlier.)
Guha however, does not appear to explicitly teach
receiving, as output from the physics-based model, model output data based on the model input data input into the physics-based model
the one or more calibration parameters are to be used by the physics-based model to perform one or more corrective actions associated with the processing chamber
wherein the one or more calibration parameters comprise one or more of predicted thermal contact conductance or predicted thermal contact resistance between corresponding components of the processing chamber
providing the one or more calibration parameters that have been tuned and the model input data to the physics-based model to generate updated model output data, wherein the updated model output data is used to perform one or more corrective actions associated with the processing chamber
Malkani teaches receiving, as output from the physics-based model, model output data based on the model input data input into the physics-based model (Fig. 7D shows the input and output to the model, and [0123] “i) Melt pool width: a high-fidelity physics simulation model that predicts the rapid solidification rate of the metal while the laser scans a particular region. The predicted widths are compared to experimental measurements for the width of melt pools at various operating conditions”).
providing the one or more calibration parameters that have been tuned and the model input data to the physics-based model to generate updated model output data, wherein the updated model output data is used to perform one or more corrective actions associated with the processing chamber ([0123] “(D) simulating, by the at least one processor, based on at least one AM simulation model, at least one testing adjustment to at least one of i) the one or more initial operational parameters of the AM machine or ii) the one or more initial AM build process parameters, when the first AM testing layer fails the compliance to the one or more predefined criteria;” and “(E) causing, by the at least one processor, the AM machine to implement the at least one testing adjustment in the AM build process” and [0142] Describes the physics based model simulations running and driving the calibration of the manufacturing process. Step (F) is iterative, “(F) repeating (A) through (E),” until the calibration process is completed including the feedback loop for the input data and the calibration parameters).
Guha and Malkani are analogous art because they are from the same field of endeavor in modelling and simulation using machine learning, more specifically covering manufacturing, substrates and semiconductors. Guha covers calibration, modeling and simulation for semi-conductors, while Malkani covers digital twins, calibration, modeling and simulation for manufacturing processes. Before the effective filing date of the invention, it would have been obvious to a person of ordinary skill in the art, to combine Guha and Malkani to arrive at using physics based models to calibrate the substrate chamber while avoiding their possible deficiencies. “Typically, physics-based simulation models are inaccurate relative to reality….the exemplary inventive simulation models of the present disclosure are calibrated or "tuned" to existing experimental data to at least reduce the impact of the above mentioned sources of inaccuracy.” (Malkani [0094-0095])
Hsu teaches wherein the one or more calibration parameters comprise one or more of predicted thermal contact conductance or predicted thermal contact resistance between corresponding components of the processing chamber (Fig 11 illustrates the testing system of thermal contact conductance and Fig 12 shows the calibration of the system of Fig 11. [0114] “Fig. 11 shows the equipment to quantitatively characterize the thermal contact conductance,” and [0117] “Calibration of the testing apparatus was performed every time before measurement. As shown in Fig. 12 A, a heater was put on the top surface of copper plate which is covered by a layer of polyurethane foam of 3 cm thick to avoid heat loss. The other part is the same with testing system. Fig. 12B shows the voltage reading of the heat flux sensor when different heating power is supplied” ).
Guha, Malkani, and Hsu are analogous art because they are from the same field of endeavor in modelling and simulation using machine learning, more specifically covering manufacturing, substrates and semiconductors. Guha covers calibration, modeling and simulation for semi-conductors, while Malkani covers digital twins, calibration, modeling and simulation for manufacturing processes. Before the effective filing date of the invention, it would have been obvious to a person of ordinary skill in the art, to combine Guha, Malkani to arrive at measuring the thermal conductance in wafter production. “suppression of thermal contact resistance and the engineering of surface morphology and optical property. Building energy simulation shows that the dual-mode device, if widely deployed in the United States, can save 19.2% heating and cooling energy, which is 1.7 times higher than cooling-only and 2.2 times higher than heating-only approaches.” (Hsu [0068])
Regarding Claim 2, Guha in view of Malkani and further in view of Hsu teaches the method of claim 1. Guha further teaches the first subset of the sensor data comprises one or more of spacing data, chamber pressure data, heater temperature data, or chamber flow rate data ([0050] “the sensors 132 may include one or more of an optical emission spectrometry (OES) sensor, a pressure sensor, a voltage sensor, a current sensor, a temperature sensor, a flow rate sensor, frequency sensor, a power sensor, a metrology sensor, and combinations of two or more thereof”).
PNG
media_image1.png
621
765
media_image1.png
Greyscale
Regarding Claim 5, Guha in view of Malkani and further in view of Hsu teaches the method of claim 1. Malkani further teaches the physics-based model comprises a digital twin model that is used to update processing parameters of the processing chamber ([0004] “ the initial digital twin comprises at least one of… (D) simulating, by the at least one processor, based on at least one AM simulation model... wherein the at least one testing adjustment is configured to compensate at least one difference in the actual AM part from the desired design due to the first AM testing layer; (E) causing, by the at least one processor, the AM machine to implement the at least one testing adjustment in the AM build process.” The simulation model is specified as physics based in [0123] “a high-fidelity physics simulation model” ).
Regarding Claim 7, Guha in view of Malkani and further in view of Hsu teaches the method of claim 1. Guha further teaches the one or more corrective actions comprise one or more of: providing an alert; interrupting operation of the processing chamber; or updating manufacturing parameters of the processing chamber ([0039] “The machine learning engine operates a mathematical model that is refined over time and is able to learn and correct not only the desired processing state values but also the compensation variables and its magnitude which upon translation into physical variables can be used as tuning knobs to physical controls, values, settings of a plasma reactor”).
Regarding Claim 8, Guha in view of Malkani and further in view of Hsu teaches the method of claim 1. Malkani further teaches comprising assigning a range of values to the one or more calibration parameters prior to training the machine learning model, wherein tuning of the one or more calibration parameters comprises adjusting corresponding values of the one or more calibration parameters within the range of values ([0113] “1) Digital twin uncertain parameters (i) which may be calibrated via Kennedy and O'Haggan methodology;
2) AM machine response status in time (j); and
3) AM machine hardware knobs that are calibrated via optimization to match a desired response range (k).” The range (k) is the set of acceptable values the calibration is constrained by. This constrains the data before the model is trained).
Claim 9-10, and 13-15 recite limitations similar to claims 1-2, and 5-7 respectively and are rejected under the same rationale.
Regarding Claim 16, Guha teaches a non-transitory machine-readable storage medium storing instructions which, when executed cause a processing device to perform operations comprising ([0131] “The computer system includes a central processing unit (CPU) 804, which is coupled through bus 810 to random access memory (RAM) 806, read-only memory (ROM) 812, and mass storage device 814. System controller program 808 resides in random access memory (RAM) 806, but can also reside in mass storage 814”).
The remaining limitations are similar to claim 1 and are rejected under the same rationale.
Claim 18 and 20 recite limitations similar to claims 5 and 8 respectively and are rejected under the same rationale.
Regarding Claim 21, Guha in view of Malkani and further in view of Hsu teaches the method of claim 1, further comprising: retraining the machine learning model with updated target output data comprising the updated model output data to further tune the one or more calibration parameters (Fig. 7 and [0063-0064] “This allows the machine learning engine 180 to determine when the desired processing state values 170 should be adjusted, as the true performance of the plasma reactor 100 is no longer matching the original desired processing state 170. As such, the machine learning engine 180 may be dynamically adjusting the desired processing state 170 based on its periodic validation operations, e.g., utilizing off-line metrology test data that is fed back to the machine learning engine 180. Additionally, machine learning engine 180 may be provided with information regarding reactor wall surface dynamics 182. This information may include data regarding the inferred characteristics of the chamber wall surfaces, as they change during processing. By way of example, this data can be inferred from historical measurements of wall characteristics, e.g., material buildup, flaking, roughness, consumable part usage, and other physical characteristics. This data can be inferred, as it may be provided by a model that predicts the type of physical changes that will occur on the reactor wall surfaces during operations over time. In some embodiments, this data can be dynamically updated from time to time and refined based on inspection of the reactor wall surfaces, e.g. when a chamber enters a wet clean cycle.” and [0099-0102] “For simplicity in describing the processing flow, the multivariate processing using machine learning will identify differences between the current processing state 172 and the desired processing state 170, by way of an operation of comparing of processing state 400. Verification operation 402 can also be used by the multivariate processing 150′, in order to determine whether the actual processing by the plasma reactor 100 is in sync with the desired processing state 170. The verification 402 will therefore allow for adjustments to the comparison of processing state 400, so that any adjustments take into consideration the verification 402. The output of the multivariate processing 150′will therefore generate compensation vector 184 in the virtual space. The resulting compensation vector 184 will be used as an update 406 to the model 154, which acts as a learning process, which updates the desired processing state based on actual dynamics occurring within the processing volume and the data streams being sensed and produced by the sensors. In this example, the compensation vectors and 84, which are in the virtual space, are transformed in operation 404, where the compensation vectors are converted to the real space to identify which knobs to adjust and the amount of adjustment.”)
Claim 22 recite limitations similar to claims 21 and are rejected under the same rationale.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
US11586789B2: Discloses optimizing substrate production in production chambers employing machine learning.
US20230013919A1: Discloses machine learning and process model calibration in the production of substrates.
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 AMIR DARWISH whose telephone number is (571)272-4779. The examiner can normally be reached 7:30-5:30 M-Thurs.
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, Lewis Bullock can be reached on 571-272-3759. 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.
/A.E.D./Examiner, Art Unit 2199
/DUY KHUONG T NGUYEN/Primary Examiner, Art Unit 2199
Prosecution Insights