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 .
Priority
Receipt is acknowledged of certified copies of papers required by 37 CFR 1.55.
Information Disclosure Statement
The information disclosure statement (IDS) submitted on August 02, 2023 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Specification
The lengthy specification has not been checked to the extent necessary to determine the presence of all possible minor errors. Applicant’s cooperation is requested in correcting any errors of which applicant may become aware in the specification.
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on February 16, 2026 has been entered.
Response to Amendment
The Amendment, filed on February 16, 2026, has been received and made of record. Claims 1-6 & 8-20 are pending. Claim 7 is canceled. Claims 1, 12 & 18 have been amended. Applicant’s amendments to the Claim(s) did not need to overcome objections and/or any U.S.C. 112(b) rejections in the Final Office Action mailed October 10, 2025, hereafter referred to as the Final Office Action.
Response to Arguments
Applicant’s arguments, please see applicant’s remarks pp. 8-9, filed February 16, 2026, with respect to the rejection(s) of amended independent claim(s) 1, 12 & 18 (all contain similar claim language) under U.S.C. § 103 have been fully considered and are persuasive. Therefore, the rejection has been withdrawn. However, in light of the amendments upon further consideration, a new ground(s) of rejection is made in view of Chen et. al. (US 2012/0075108 A1, hereinafter, Chen), in view of Turner et al. (US 5576629 A, hereinafter, Turner), in view of Hansen (US 2012/0123737 A1, hereinafter, Hansen), in view of Mahoney, (US 2006/0180570 A1, hereinafter, Mahoney), in view of Carter (US 2021/0351007 A1, hereinafter, Carter), and further in view of Gadre et al. (US 2023/0061513 A1, hereinafter Gadre), for amended independent claims 1 & 18, and Chen, in view of Turner, in view of Hansen, in view of Mahoney, in view of Pang et al. (US 6193802 B1, hereinafter, Pang), in view of carter, in view of Gadre, in view of Kapoor et al. (US 2022/0037135 A1, hereinafter, Kapoor), in view of Wiklund et al. (US 2006/0036404 A1, hereinafter, Wiklund), and further in view of Shaw et al. (US 2022/0139674 A1, hereinafter, Shaw), for amended independent claim 12.
In response to the applicant’s arguments, please see pp. 8-9 of applicant’s remarks, with respect to the rejection of amended independent claims 1 & 18 under U.S.C § 103, that the prior art references, Chen et al. (US 8587321 B2, hereinafter, Chen), in view of Turner et al. (US 5576629, hereinafter, Turner), in view of Hansen (US 8120376 B2, hereinafter, Hansen), and further in view of Mahoney (US 2006/0180570 A1, hereinafter, Mahoney), and with respect to the rejection of amended independent claim 12 under U.S.C § 103, that the prior art references, Mahoney, in view of Pang et. a. (US 6193802 B1, hereinafter, Pang), in view of Turner, and further in view of Hansen, as cited by the applicant, fail to teach, disclose, and/or suggest, individually or in combination, the amended features, “a recommendation module configured to: match the detected deviation with one or more deviation signatures stored in a corrective action database to identify a likely case of the excursion, wherein each of the deviation signatures is associated with a previously identified excursion cause,” and “generate one or more recommendations based on the likely cause to control the excursion…”.
In light of the amendments in independent claims 1, 12 & 18, new grounds of rejections are made over Chen et. al. (US 2012/0075108 A1, hereinafter, Chen), in view of Turner et al. (US 5576629 A, hereinafter, Turner), in view of Hansen (US 2012/0123737 A1, hereinafter, Hansen), in view of Mahoney, (US 2006/0180570 A1, hereinafter, Mahoney), in view of Carter (US 2021/0351007 A1, hereinafter, Carter), and further in view of Gadre et al. (US 2023/0061513 A1, hereinafter Gadre) for amended independent claims 1 & 18, and new grounds of rejections are made over Chen, in view of Turner, in view of Hansen, in view of Mahoney, in view of Pang et al. (US 6193802 B1, hereinafter, Pang), in view of carter, in view of Gadre, in view of Kapoor et al. (US 2022/0037135 A1, hereinafter, Kapoor), in view of Wiklund et al. (US 2006/0036404 A1, hereinafter, Wiklund), and further in view of Shaw et al. (US 2022/0139674 A1, hereinafter, Shaw) for amended independent claim 12. The examiner respectfully disagrees with the applicant’s contentions that Chen, in view of Turner, in view of Hansen, in view of Mahoney, in view of Carter, and further in view of Gadre, fail to disclose, teach, and/or suggest individually or in combination, the previously stated amended features of the claimed invention for amended independent claims 1 & 18, and Chen, in view of Turner, in view of Hansen, in view of Mahoney, in view of Pang, in view of carter, in view of Gadre, in view of Kapoor, in view of Wiklund, and further in view of Shaw, fail to disclose, teach, and/or suggest individually or in combination, the previously stated amended features of the claimed invention for amended independent claim 12. The new grounds of rejections further disclose the additional limitations that have been amended and included in independent claims 1, 12 & 18, and meet these requirements. Therefore, the Applicant’s arguments are unconvincing and the rejections of amended independent claims 1, 12 & 18, and dependent claims 2-6, 8-11, 13-17 & 19-20, which depend from and incorporate the limitations of amended independent claims 1, 12 & 18, are respectively maintained. Rejections based on the newly cited prior art references follow.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1, 5-6, 8 & 18 are rejected under 35 U.S.C. 103 as being unpatentable over Chen et. al. (US 2012/0075108 A1, Pub. Date Mar. 29, 2012, hereinafter, Chen), in view of Turner et al. (US 5576629 A, Pat. Date Nov. 19, 1996, hereinafter, Turner), in view of Hansen (US 2012/0123737 A1, Pub. Date May 17, 2012, hereinafter, Hansen), in view of Mahoney (US 2006/0180570 A1, Pub. Date Aug. 17, 2006, hereinafter, Mahoney), in view of Carter (US 2021/0351007 A1, Pub. Date Nov. 11, 2021, hereinafter, Carter), and further in view of Gadre et al. (US 2023/0061513 A1, Fil. Date Aug. 27, 2021, hereinafter Gadre).
Regarding independent claim 1: Chen, teaches:
An excursion detection and control system for detecting excursion in plasma processing ([Abstract]), and controlling an in-situ plasma processing (Disclosed in combination: Chen: [Abstract] & [0006]-[0007]: “detecting plasma excursions in a plasma chamber comprises directly sensing a radio frequency (RF)” refers to “in-situ plasma processing”; Carter: [Abstract]: also teaches controlling an in-situ plasma processing), the excursion detection and control system comprising ([Abstract] & [0002]):
Chen, is silent in regard to:
a transducer assembly configured to:
track one or more harmonics that are produced due to nonlinearity of an impedance in a plasma environment, and that comprise at least one of voltage or current;
create a fingerprint of energy distribution in frequency space based on the one or more harmonics,
However, Turner, further teaches:
a transducer assembly configured to:
track one or more harmonics ([Col.1, ll. 6-13]) that are produced due to nonlinearity (Fig. 4; [Col. 7, ll. 23-31]: 62; “a non-linear portion that has already shifted some of the radio frequency energy from the signal”) of an impedance in a plasma environment ([Col.3, ll. 23-41]), and that comprise at least one of voltage or current ([Col.1, ll. 6-13], [Col. 3, ll. 23-41], [Col. 5, ll.10-14 & 55-60] & [Col. 15, ll. 25-33]: teaches a transducer/sensor assembly that tracks voltage and current harmonics resulting from the non-linear impedance of plasma);
one or more spectrum analyzers configured to (Fig. 4; [Col. 7, ll. 14-18 & 57-67]: 60):
create a fingerprint of energy distribution in frequency space based on the one or more harmonics ([Col. 17, ll. 14-19]: teaches creating a fingerprint in frequency space based on the tracked harmonics),
PNG
media_image1.png
522
746
media_image1.png
Greyscale
It would have been obvious to one having ordinary skill in the art before the effective filing date of the claimed invention was made to modify the voltage-based excursion detection system of Chen/Carter to include the harmonic-tracking transducers and harmonic fingerprinting methodologies of Turner, according to known methods. A POSITA would have been motivated to make this combination for the following reasons: to increase detection sensitivity, and to gather rich diagnostic data. By upgrading the transducer assembly to track the specific harmonic-rich frequency signal generated by the plasma’s non-linear impedance, as taught by Turner, the system gains a sensitive, multi-dimensional view of the plasma state. A harmonic fingerprint provides a unique frequency-space signature of what is happening inside the chamber, which is essential for diagnosis the specific type of excursion. The motivation to combine Turner is to improve the sensitivity and diagnostic richness of the plasma excursion detection system. Further, a POSITA would be motivated to integrate Turner’s teaching of tracking voltage and current harmonics, which are produced by the non-linear impedance of the plasma, to create harmonic fingerprints. This predictable modification allows the detection system to monitor complex, multi-dimensional shifts in the plasma state rather than relying on analog thresholds. The use of a known technique (capturing harmonic fingerprints from a non-linear plasma load, as taught by Turner) to improve a similar known device (e.g., an in-situ plasma excursion monitor as taught by Chen) yield the predictable result (KSR) of a robust, highly sensitive plasma diagnostic tool.
Chen, in combination with Turner, are silent in regard to:
compare, in real-time, the fingerprint with a reference spectrum of an ideal plasma processing, stored in a historical database, to detect a deviation of the fingerprint with the reference spectrum, and
detect an excursion in the plasma processing based on the deviation of the fingerprint with the reference spectrum; and
However, Hansen, further teaches:
compare, in real-time ([0008]-[0009] & [0027]: teaches comparing real-time data spectrum to a prerecorded signature stored in a database to detect a fault/deviation), the fingerprint with a reference spectrum ([0008]-[0009]: signature is the “created fingerprint”) of an ideal plasma processing ([0020] & [0023]: properly functioning plasma refers to “ideal plasma processing”), stored in a historical database ([0019]: signature signal database refers to “historical database”), to detect a deviation of the fingerprint with the reference spectrum ([0019]-[0020]: signature signal refers to “fingerprint”), and
detect an excursion in the plasma processing based on the deviation of the fingerprint with the reference spectrum ([0023]: excursion detection is referred to as “arcing”, comparator compares the test signal to the signature signal to determine if signals vary); and
It would have been obvious to one having ordinary skill in the art before the effective filing date of the claimed invention was made to incorporate the database comparison architecture of Hansen into the excursion detection system of Chen/Turner, according to known methods. A POSITA would have been motivated to make this combination for the following reasons: to accommodate complex data structure and to improve excursion detection accuracy. A POSITA would recognize that a single, preset value threshold, as taught by Chen, is inadequate for evaluating the complex harmonic fingerprints of energy distribution in frequency space taught by Turner. To effectively use Turner’s spectral fingerprints for excursion detection, the system needs to be upgraded to compare real-time multi-frequency spectra against a stored baseline spectrum (e.g., Hansen’s known-state signature). Further, simple analog thresholds are prone to false positives or missing subtle drifts. By storing a reference spectrum of an ideal plasma processing in a historical database (Hansen’s prerecorded signature signal in a known state), the system can detect deviations across the entire frequency space simultaneously, providing a more accurate and robust detection of plasma instability. The motivation to combine Hansen with the primary combination of Chen and Turner is to provide a robust computational method for evaluating complex harmonic signals. Further, a POSITA would be motivated to integrate Hansen’s signature database architecture to allow the system to compare real-time harmonic fingerprints against a stored ideal reference spectrum. This predictable substitution upgrades the detection logic to handle complex spectral data, increasing the sensitivity and accuracy of the excursion detection system. Applying Hansen’s known fault-detection comparison technique (real-time vs. database reference signal) to Turner’s known harmonic fingerprint signals yield the predictable result (KSR) of an accurate, automated excursion detection system capable of monitoring complex plasma states.
Chen, in combination with Turner, and Hansen, are silent in regard to:
a recommendation module configured to:
via at least one of workpiece conditions, process chamber pressure, gas mixture, RF power, or electrode spacing.
However, Mahoney, further teaches:
a recommendation module configured to ([0035]-[0037]):
via at least one of workpiece conditions ([0003]-[0005] & [0035]-[0037]), process chamber pressure ([0003]-[0005] & [0035]-[0037]), gas mixture ([0003]-[0004], [0022] & [0037]: gas flow manifolds are inherently known to control the flow of “process gas mixture” and “chamber pressure”), RF power ([0003]-[0005], [0022] & [0035]-[0037]), or electrode spacing ([0003]-[0006], [0022], [0037] & [0042]: probes or sensors can also be referred to as an “electrode and/or induction element”, wafer-based probe “device may be disposed in fixed arrays within the processing equipment itself”).
It would have been obvious to one having ordinary skill in the art before the effective filing date of the claimed invention was made to incorporate the smart FDC recommendation and correction modules of Mahoney into the excursion detection system of Chen/Turner/Hansen, according to known methods. A POSITA would have been motivated to make this combination for the following reasons: to automate process recovery and increase yield and to translate detection into actionable control. A primary goal in semiconductor manufacturing is to decrease process variability and increase product yield. By combining Hansen’s accurate excursion detection with Mahoney’s decision trees, the system can automatically adjust parameters (like RF power or gas mixture) in real-time to save an ongoing plasma process, rather than alarming and forcing a manual tool shutdown. Turner and Hansen provide multi-dimensional data (harmonic fingerprints). A POSITA would recognize that to fully utilize this data, an expert system like Mahoney’s is required to map the complex deviations to specific hardware corrections (e.g., advising an operator or system on what input variables to adjust). The motivation to combine Mahoney is to provide a closed-loop, automated correction mechanism. A POSITA would be further motivated to integrate Mahoney’s smart FDC recommendation module to automatically translate detected spectral deviations into root-cause diagnoses and actionable hardware adjustments (such as changing RF power or pressure), reducing process downtime, decreasing variability, and preventing wafer loss. Applying a known fault classification and correction algorithm (Mahoney) to a known fault detection output (Hansen) is a simple substitution of one known element for another to obtain predictable results (KSR).
Chen, in combination with Turner, Hansen, Mahoney, and Carter, are silent in regard to:
match the detected deviation with one or more deviation signatures stored in a corrective action database to identify a likely cause of the excursion, wherein each of the deviation signatures is associated with a previously identified excursion cause, and generate one or more recommendations based on the likely cause to control the excursion
However, Gadre, further teaches:
match the detected deviation with one or more deviation signatures stored in a corrective action database to identify a likely cause of the excursion, wherein each of the deviation signatures is associated with a previously identified excursion cause, and generate one or more recommendations based on the likely cause to control the excursion ([0021]-[0025], [0029], [0032] & [0037]-[0040]: teaches matching a detected fault pattern (one or more deviation signatures) against a stored library of known fault patterns (corrective action database) to identify the root cause (previously identified cause) and recommending a corrective action via process parameters)
It would have been obvious to one having ordinary skill in the art before the effective filing date of the claimed invention was made to incorporate the fault library and pattern-matching diagnostic methodologies of Gadre into the smart FDC and excursion detection system of Chen/Mahoney, according to known methods. A POSITA would have been motivated to make this combination for the following reasons: to reduce equipment downtime and prevent misdiagnosis, as stated by Gadre, relying on isolated sensor data or manual operator troubleshooting can allow deteriorating conditions to go undetected or lead to unnecessary replacement of operational components while trying to determine the cause of a failure; and to automate root-cause diagnostics. Integrating Gadre’s historical fault library allows the control system to instantly correlate complex frequency-space fingerprints (Turner/Hansen) with historical root causes, removing the need for manual operator investigation. The use of historical fault databases to classify the current sensor deviations is a well-known technique in semiconductor manufacturing informatics. The motivation to combine Gadre is to improve the accuracy and speed of the FDC system’s diagnostic capabilities. By integrating Gadre’s historical fault library, the FDC system of Mahoney is upgraded from a basic responsive decision-tree to an adaptive, database-driven expert system that reduces tool downtime and prevents operator misdiagnosis by directly matching detected harmonic deviations against previously identified root causes to automate hardware corrections. Applying Gadre’s database-matching technique to the FDC controllers of Mahoney yields the predictable result of an automated, self-correcting plasma chamber that can intelligently adjust variables like RF power or pressure based on past learning (KSR).
Regarding dependent claim 5, Chen, teaches:
The excursion detection and control system of claim 1 (Disclosed in combination: Chen: [Abstract]; Carter: [Abstract]),
Chen, in combination with Hansen, are silent in regard to:
wherein the deviation is created due to a variation in controllable subassemblies of the plasma processing that comprises at least one of incoming workpiece conditions, process chamber pressure, process gas mixture, Radio Frequency (RF) power, or electrode spacing.
However, Turner, in combination with Mahoney, and Gadre, further teach:
wherein the deviation is created due to a variation in controllable subassemblies of the plasma processing (Disclosed in combination: Turner: [Col. 14, ll. 34-39]: confirms that the harmonic fingerprint deviations represent physical variations in the hardware and chemistry; Mahoney: [0003]-[0005], [0009] & [0012]: teaches that the system model identifies when the controllable input variables are in error (variation); Gadre: [0023]-[0024]: teaches that the detected deviations (fault patterns) are caused by failures/variations in the chamber’s subsystems (controllable assemblies))
It would have been obvious to one having ordinary skill in the art before the effective filing date of the claimed invention was made to combine the hardware-correlation teachings of Turner with the root-cause FDC diagnostic modules of Mahoney and Gadre, according to known methods. A POSITA would have been motivated to make this combination to upgrade the system from a passive monitor into an active, self-correcting controller. Further, a POSITA would recognize that plasma deviations do not happen in a vacuum, but are physical manifestations of hardware or input variations. By integrating Gadre and Mahoney’s fault libraries, the system is programmed to understand that a specific spectral deviation (as taught by Turner) is created by a variation in a controllable assembly (e.g., drift in chamber pressure or fluctuations in RF power). This predictable mapping is the required mechanism that allows the system to identify the root cause and automatically manipulate the correct hardware input to save the process, and yield predictable results (KSR).
Chen, in combination with Turner and Hansen, are silent in regard to:
that comprises at least one of incoming workpiece conditions, process chamber pressure, process gas mixture, Radio Frequency (RF) power, or electrode spacing.
However, Mahoney, in combination with Gadre, further teach:
that comprises at least one of incoming workpiece conditions (Disclosed in combination: Mahoney: [0035]-[0037]: lists RF Power, “flows/chemistry balance” (gas mixture),“pressure”, and electrode spacing, as the system input factors that cause the responses; Gadre: [0040]: lists pressure and flow rate (gas mixture)), process chamber pressure (Mahoney: [0003]-[0005], [0036]-[0037] & [0042]), process gas mixture (Mahoney: [0003]-[0004], [0022], [0037] & [0042]: gas flow manifolds are inherently known to control the flow of “process gas mixture” and “chamber pressure”), Radio Frequency (RF) power (Mahoney: [0003]-[0005], [0022], [0036] & [0042]), or electrode spacing (Mahoney: [0003]-[0006], [0022] & [0042]: probes or sensors can also be referred to as an “electrode and/or induction element”, wafer-based probe “device may be disposed in fixed arrays within the processing equipment itself”).
It would have been obvious to one having ordinary skill in the art before the effective filing date of the claimed invention was made to configure the excursion detection system such that the detected spectral variations are recognized as being created by variations in controllable assemblies (such as RF power, pressure, gas mixture, or electrode spacing), as taught by Mahoney, Gadre, and Turner, according to known methods. A POSITA would recognize that plasma does not fluctuate randomly; excursions are drive by physical variations in the tool’s subsystems. By integrating the teachings of Mahoney and Gadre, the FDC system is configured to link the detected spectral deviation directly to a “root cause” variation in a controllable assembly (e.g., a drift in the Mass Flow Controller causing a gas mixture variation, or a voltage drop causing an RF power variation). This predictable mapping of spectral deviation to hardware variation allows Mahoney and Gadre’s systems to automatically generate recommendations to correct specific hardware assemblies, and yield predictable results (KSR).
Regarding dependent claim 6, Chen, teaches:
The excursion detection and control system of claim 1 (Disclosed in combination: Chen: [Abstract]; Carter: [Abstract]),
Chen, in combination with Turner, and Hansen, are silent in regard to:
to determine a reason behind the excursion and provide the one or more recommendations.
However, Mahoney in combination with Gadre, further teach:
to determine a reason behind the excursion and provide the one or more recommendations (Disclosed in combination: Mahoney: Fig. 6; [0012]-[0013] & [0037]: teaches that the system uses “decision trees” to diagnose the error advise what variables to adjust and further states that this analysis is used for “fault detection and classification (FDC)” and “advanced process control”, where the output is a “FDC Report” that provides a classification and, by implication, a recommendation for action; Gadre: [0023]: teaches that the matching process determines the “root cause” (reason) and outputs a “corrective action” (recommendation)).
PNG
media_image2.png
816
763
media_image2.png
Greyscale
It would have been obvious to one having ordinary skill in the art before the effective filing date of the claimed invention was made to incorporate the fault library and historical pattern-matching diagnostic methodologies of Gadre into the smart FDC decision-tree framework of Mahoney, according to known methods. A POSITA would have been motivated to make this combination to improve diagnostic speed, accuracy, and learning capability of the FDC system’s diagnostic modules. While Mahoney teaches using decision trees to find root causes and adjust hardware like RF power or pressure, it lacks a mechanism to leverage historical failure data. A POSITA would be motivated to integrate Gadre’s historical fault library into Mahoney’s FDC system. This predictable modification upgrades the FDC controller from a decision-tree model to an adaptive, database-driven system that can instantly match real-time deviation signatures against previously identified root causes to automate hardware corrections, thus reducing tool downtime, preventing misdiagnosis, and yield predictable results (KSR).
Chen, in combination with Turner, Hansen, Mahoney, and Carter, are silent in regard to:
wherein the recommendation module is configured to match the deviation with one or more deviations stored in the corrective action database
However, Gadre, further teaches:
wherein the recommendation module is configured to match the deviation with one or more deviations stored in the corrective action database ([0023], [0032] & [0037]-[0040]: teaches matching an observed fault pattern to a database of known historical fault patterns)
It would have been obvious to one having ordinary skill in the art before the effective filing date of the claimed invention was made to configure the recommendation module of the primary combination (Chen/Turner/Hansen) to match detected deviations against a corrective action database to determine the reason behind the excursion and provide recommendations, as taught by Gadre, according to known methods. A POSITA would be motivated to integrate the “fault library” diagnostic architecture of Gadre into the control system. By matching the specific shape or signature of the real-time deviation against historically stored deviations, as taught by Gadre, the system predictably transforms raw fault detection data into actionable intelligence, identifying the exact root cause (reason) and generating hardware adjustments to save the process, yielding predictable results (KSR).
Regarding dependent claim 8, Chen, teaches:
The excursion detection and control system of claim 1 (Disclosed in combination: Chen: [Abstract] & [0028]; Carter: [Abstract]), further comprising a controller configured to at least one of (Disclosed in combination: Chen: [Abstract] & [0028]: teaches a control system that receives the excursion data; Carter: [Abstract]: teaches the hardware controllers for the plasma chamber)
Chen, is silent in regard to:
and stop the plasma processing based on the excursion such that the subsequent substrates in the plasma processing are saved from the excursion.
However, Chen, in combination with Turner, further teach:
and stop the plasma processing based on the excursion such that the subsequent substrates in the plasma processing are saved from the excursion (Disclosed in combination: Chen: [0028]: teaches shutting down the system upon detecting excursion, which inherently saves subsequent substrates from being processed in a faulty chamber; Turner: [Claim 29], [Claim 36] & [Claim 43]: teaches a halt signal to prevent further processing).
It would have been obvious to one having ordinary skill in the art before the effective filing date of the claimed invention was made to configure the controller of the of the primary excursion detection system to shut down the process to save subsequent substrates, as taught by Chen/Turner, according to known methods. A POSITA would be motivated to configure the system’s controller to execute the two standard industry responses to a detected fault. If the excursion is severe or uncorrectable, then the controller falls back to the safety mechanism taught by Chen and Turner, immediately halting the plasma processing. The secondary response is explained below after the secondary mapping. This prevents the tool from continuing to process and destroy expensive subsequent wafers. Combining these known hardware control methods with the excursion detection logic yields the predictable result (KSR) of a fully automated, self-preserving plasma chamber.
Chen, in combination with Turner, and Hansen, are silent in regard to:
control an associated controllable subassembly based on the one or more recommendations such that the excursion is reduced in subsequent substrates,
However, Mahoney, in combination with Gadre, further teach:
control an associated controllable subassembly based on the one or more recommendations such that the excursion is reduced in subsequent substrates (Disclosed in combination: Mahoney: Fig. 6; [0002], [0007], [0012]-[0013], [0035]-[0037] & [Claim 4]: teaches using the FDC system to adjust system inputs to return the chamber to nominally acceptable conditions, which reduces the excursion for subsequent processing and increases yield; Gadre: [0017], [0023]-[0024], [0032] & [0037]-0040]),
It would have been obvious to one having ordinary skill in the art before the effective filing date of the claimed invention was made to configure the controller of the of the primary excursion detection system to either adjust a subassembly to fix the excursion, as taught by Mahoney/Gadre, or shut down the process to save subsequent substrates, as taught by Chen/Turner, according to known methods. A POSITA would be motivated to configure the system’s controller to execute the two standard industry responses to a detected fault. If the expert system (Mahoney/Gadre) has a high-confidence recommendation, the controller automatically adjusts the subassembly (e.g., changes the gas flow or RF power) to return the chamber to nominal conditions, ensuring the next wafer processed is defect-free. If the excursion is severe or uncorrectable, then the controller falls back to the safety mechanism taught by Chen and Turner, immediately halting the plasma processing. This prevents the tool from continuing to process and destroy expensive subsequent wafers. Combining these known hardware control methods with the excursion detection logic yields the predictable result (KSR) of a fully automated, self-preserving plasma chamber.
Regarding independent claim 18, Chen, teaches:
A method for detecting excursion in plasma processing ([Abstract]) and controlling the plasma processing (Disclosed in combination: Chen: [Abstract] & [0006]-[0007]: “detecting plasma excursions in a plasma chamber comprises directly sensing a radio frequency (RF)” refers to “in-situ plasma processing”; Carter: [Abstract]: also teaches controlling an in-situ plasma processing), the method comprising:
Chen, is silent in regard to:
tracking one or more harmonics that are produced due to nonlinearity of an impedance of a plasma environment, wherein the one or more harmonics comprise at least one of voltage or current;
creating a fingerprint of energy distribution in frequency space based on the one or more harmonics;
However, Turner, further teaches:
tracking one or more harmonics ([Col.1, ll. 6-13]) that are produced due to nonlinearity (Fig. 4; [Col. 7, ll. 23-31]: 62; “a non-linear portion that has already shifted some of the radio frequency energy from the signal”) of an impedance of a plasma environment ([Col.3, ll. 23-41]), wherein the one or more harmonics comprise at least one of voltage or current ([Col.1, ll. 6-13], [Col. 3, ll. 23-41], [Col. 5, ll.10-14 & 55-60] & [Col. 15, ll. 25-33]: teaches a method of tracking voltage and current harmonics resulting from the non-linear impedance of the plasma);
creating a fingerprint of energy distribution in frequency space based on the one or more harmonics ([Col. 8, ll. 44-47], [Col. 9, ll. 66-67], [Col. 10, ll. 1-2], & [Col. 17, ll. 14-19]);
It would have been obvious to one having ordinary skill in the art before the effective filing date of the claimed invention was made to modify the voltage-based excursion detection methodology of Chen/Carter to include the harmonic fingerprinting methodologies of Turner, according to known methods. A POSITA would have been motivated to make this combination for the following reasons: to increase detection sensitivity, and to gather rich diagnostic data. By upgrading the transducer assembly to track the specific harmonic-rich frequency signal generated by the plasma’s non-linear impedance, as taught by Turner, the system gains a sensitive, multi-dimensional view of the plasma state. A harmonic fingerprint provides a unique frequency-space signature of what is happening inside the chamber, which is essential for diagnosis the specific type of excursion. The motivation to combine Turner is to improve the sensitivity and diagnostic richness of the plasma excursion detection system. Further, a POSITA would be motivated to integrate Turner’s teaching of tracking voltage and current harmonics, which are produced by the non-linear impedance of the plasma, to create harmonic fingerprints. This predictable modification allows the detection system to monitor complex, multi-dimensional shifts in the plasma state rather than relying on analog thresholds. The use of a known technique (capturing harmonic fingerprints from a non-linear plasma load, as taught by Turner) to improve a similar known device (e.g., an in-situ plasma excursion monitor as taught by Chen) yield the predictable result (KSR) of a robust, highly sensitive plasma diagnostic tool.
Chen, in combination with Turner, are silent in regard to:
comparing, in real-time, the fingerprint with a reference spectrum of an ideal plasma processing, stored in a historical database, to detect deviation of the fingerprint with the reference spectrum;
detecting an excursion in the plasma processing based on the deviation of the fingerprint with the reference spectrum;
However, Hansen, further teaches:
comparing, in real-time ([0008]-[0009], [0027] & [Claim 23]: teaches a method of comparing a real-time data spectrum to a prerecorded signature stored in a database to detect a deviation), the fingerprint with a reference spectrum ([0008]-[0009]: signature signal is the “created fingerprint”) of an ideal plasma processing ([0020], [0023] & [Claim 23]: properly functioning plasma refers to “ideal plasma processing”), stored in a historical database ([0019]: signature signal database refers to “historical database”), to detect deviation of the fingerprint with the reference spectrum ([0019]-[0020]: signature signal is the “fingerprint”);
detecting an excursion in the plasma processing based on the deviation of the fingerprint with the reference spectrum ([0023] & [Claim 23]: excursion detection is referred to as “arcing” and comparator compares the test signal to the signature signal to determine if signals vary);
It would have been obvious to one having ordinary skill in the art before the effective filing date of the claimed invention was made to incorporate the database comparison architecture of Hansen into the excursion detection Methodology of Chen/Turner, according to known methods. A POSITA would have been motivated to make this combination for the following reasons: to accommodate complex data structure and to improve excursion detection accuracy. A POSITA would recognize that a single, preset value threshold, as taught by Chen, is inadequate for evaluating the complex harmonic fingerprints of energy distribution in frequency space taught by Turner. To effectively use Turner’s spectral fingerprints for excursion detection, the system/methodology needs to be upgraded to compare real-time multi-frequency spectra against a stored baseline spectrum (e.g., Hansen’s known-state signature). Further, analog thresholds are prone to false positives or missing subtle drifts. By storing a reference spectrum of an ideal plasma processing in a historical database (Hansen’s prerecorded signature signal in a known state), the system/methodology can detect deviations across the entire frequency space simultaneously, providing a more accurate and robust detection of plasma instability. The motivation to combine Hansen with the primary combination of Chen and Turner is to provide a robust computational method for evaluating complex harmonic signals. Further, a POSITA would be motivated to integrate Hansen’s signature database architecture to allow the system to compare real-time harmonic fingerprints against a stored ideal reference spectrum. This predictable substitution upgrades the detection logic to handle complex spectral data, increasing the sensitivity and accuracy of the excursion detection methodology. Applying Hansen’s known fault-detection comparison technique (real-time vs. database reference signal) to Turner’s known harmonic fingerprint signals yield the predictable result (KSR) of an accurate, automated excursion detection methodology capable of monitoring complex plasma states.
Chen, in combination with Turner, and Hansen, are silent in regard to:
generating one or more recommendations based on the likely cause to control the excursion via at least one of workpiece conditions, process chamber pressure, gas mixture, RF power, or electrode spacing.
However, Mahoney, in combination with Gadre, further teach:
generating one or more recommendations based on the likely cause to control the excursion (Disclosed in combination: Mahoney: [0035]-[0037]: supports the method by teaching parameter adjustments via FDC recommendations including RF power, pressure, and chemistry (gas) balance; Gadre: [0037]-0040]: teaches generating a recommended corrective action to adjust specific process parameters) via at least one of workpiece conditions (Mahoney: [0003]-[0005] & [0035]-[0037]), process chamber pressure (Mahoney: [0003]-[0005] & [0035]-[0037]), gas mixture (Mahoney: [0003]-[0004], [0022] & [0037]: gas flow manifolds are inherently known to control the flow of “process gas mixture” and “chamber pressure”), RF power (Mahoney: [0003]-[0005], [0022] & [0035]-[0037]), or electrode spacing (Mahoney: [0003]-[0006], [0022], [0037] & [0042]: probes or sensors can also be referred to as an “electrode and/or induction element”, wafer-based probe “device may be disposed in fixed arrays within the processing equipment itself”).
It would have been obvious to one having ordinary skill in the art before the effective filing date of the claimed invention was made to incorporate the smart FDC recommendation Mahoney in combination with Gadre’s recommended corrective action to adjust process parameters into the excursion detection system/methodology of Chen/Turner/Hansen, according to known methods. A POSITA would have been motivated to make this combination for the following reasons: to automate process recovery and increase yield and to translate detection into actionable control. A primary goal in semiconductor manufacturing is to decrease process variability and increase product yield. By combining Hansen’s accurate excursion detection with Mahoney’s decision trees, and Gadre’s correction action(s), the system/methodology can automatically adjust parameters (like RF power or gas mixture) in real-time to save an ongoing plasma process, rather than alarming and forcing a manual tool shutdown. Turner and Hansen provide multi-dimensional data (harmonic fingerprints). A POSITA would recognize that to fully utilize this data, an expert system like Mahoney’s is required to map the complex deviations to specific hardware corrections (e.g., advising an operator or system on what input variables to adjust). The motivation to combine Mahoney is to provide a closed-loop, automated correction mechanism. A POSITA would be further motivated to integrate Mahoney’s smart FDC recommendations and Gadre’s recommended corrective action to automatically translate detected spectral deviations into root-cause diagnoses and actionable hardware adjustments (such as changing RF power or pressure), reducing process downtime, decreasing variability, and preventing wafer loss. Applying a known fault classification and correction algorithm (Mahoney/Gadre) to a known fault detection output (Hansen) is a simple substitution of one known element for another to obtain predictable results (KSR).
Chen, in combination with Turner, Hansen, Mahoney, and Carter, are silent in regard to:
matching the detected deviation with one or more deviation signatures stored in a corrective action database to identify a likely cause of the excursion, wherein each of the deviation signatures is associated with a previously identified excursion cause, and
However, Gadre, further teaches:
matching the detected deviation with one or more deviation signatures stored in a corrective action database to identify a likely cause of the excursion, wherein each of the deviation signatures is associated with a previously identified excursion cause, and generate one or more recommendations based on the likely cause to control the excursion ([0021]-[0025], [0029], [0032] & [0037]-[0040]: teaches matching a detected fault pattern (one or more deviation signatures) against a stored library of known fault patterns (corrective action database) to identify the root cause (previously identified cause) and recommending a corrective action via process parameters)
It would have been obvious to one having ordinary skill in the art before the effective filing date of the claimed invention was made to incorporate the fault library and pattern-matching diagnostic methodologies of Gadre into the smart FDC and excursion detection system of Chen/Mahoney, according to known methods. A POSITA would have been motivated to make this combination for the following reasons: to reduce equipment downtime and prevent misdiagnosis, as stated by Gadre, relying on isolated sensor data or manual operator troubleshooting can allow deteriorating conditions to go undetected or lead to unnecessary replacement of operational components while trying to determine the cause of a failure; and to automate root-cause diagnostics. Integrating Gadre’s historical fault library allows the control system to instantly correlate complex frequency-space fingerprints (Turner/Hansen) with historical root causes, removing the need for manual operator investigation. The use of historical fault databases to classify the current sensor deviations is a well-known technique in semiconductor manufacturing informatics. The motivation to combine Gadre is to improve the accuracy and speed of the FDC system’s diagnostic capabilities. By integrating Gadre’s historical fault library, the FDC system of Mahoney is upgraded from a basic responsive decision-tree to an adaptive, database-driven expert system that reduces tool downtime and prevents operator misdiagnosis by directly matching detected harmonic deviations against previously identified root causes to automate hardware corrections. Applying Gadre’s database-matching technique to the FDC controllers of Mahoney yields the predictable result of an automated, self-correcting plasma chamber that can intelligently adjust variables like RF power or pressure based on past learning (KSR).
Claims 2-4, 9-11 & 19-20 are rejected under 35 U.S.C. 103 as being unpatentable over Chen, in view of Turner, in view of Hansen, in view of Mahoney, in view of Carter, in view of Gadre, in view of Kapoor et al. (US 2022/0037135 A1, Pub. Date Feb. 3, 2022, hereinafter, Kapoor), and further in view of Wiklund et al. (US 2006/0036404 A1, Pub. Date Feb. 16, 2006, hereinafter, Wiklund).
Regarding dependent claim 2: Chen, teaches:
The excursion detection and control system of claim 1 ([Abstract]), wherein the one or more spectrum analyzers are configured to (Disclosed in combination: Chen: [Abstract]; Turner: [Col. 7, ll. 57-66]; Hansen: [Abstract] & [0008]-[0009]: foundational system established by the combination, with Turner and Hansen establishing the use of spectrum analyzers to process the harmonic signals):
Chen, is silent in regard to:
calculate, when plasma and chemical conditions of the plasma processing are stable, at least one of an average or a standard deviation of an amplitude of the one or more harmonics, and
filter the one or more harmonics by removing harmonics from the one or more harmonics that have the amplitude and the standard deviation less than a pre-defined threshold value.
However, Turner, in combination with Kapoor, further teach:
calculate (Turner: [Col. 14, lines 34-39]: quantifying is referred to as “calculating”), when plasma and chemical conditions of the plasma processing are stable (Turner: [Col. 11, ll. 4-9, 18-21, & 26-32] & [Col. 18, ll. 24-29]), at least one of an average (Turner: [Fig. 9; [Col. 10, ll. 55-64] & [Col. 11, ll. 4-9]: 112, average algorithm, teaches calculating the mean (average) of voltage and current harmonics during a stable plasma period) or a standard deviation (Disclosed in combination: Turner: Fig. 2; [Col. 2, ll. 5-7] & [Col. 14, ll. 34-39 & 47-50]: 40; standard deviation is referred to as “variability”; Kapoor: [0009]-[0010]: teaches calculating standard deviations of spectral densities (frequency amplitudes) while the plasma is maintained under stable/nominal conditions) of an amplitude (Turner: [(Col. 8, ll. 34-47]) of the one or more harmonics (Turner: [Col.1, ll. 6-13], [Col. 8, ll. 42-47] & [Col. 11, ll. 26-32]), and
PNG
media_image3.png
742
870
media_image3.png
Greyscale
PNG
media_image4.png
868
722
media_image4.png
Greyscale
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to integrate the standard deviation threshold calculations of Kapoor into the stable baseline averaging of Turner and the reference spectrum framework of Hansen, according to known methods. A POSITA would have been motivated to make this combination to improve the statistical accuracy and robustness of the excursion detection system/methodology. Relying solely on average means or static limits makes a system/methodology vulnerable to false alarms triggered by background noise. By integrating Kapoor’s teaching to calculate the standard deviation under nominal/stable conditions, the system can dynamically set thresholds based on the statistical variance of plasma. Thus, would predictably reduce false-positive alarms, ensuring that only statistically significant deviations trigger the FDC corrective actions of Mahoney and Gadre. Turner teaches a system that monitor RF harmonics in a plasma process and performs statistical process control (SPC) on the data, including averaging and standard deviation (“three sigma”) to establish control limits (thresholds) for go/no-go decisions, Chen further teaches that it is known to compare a processed plasma signal to a preset value. The combination of teachings would improve and optimize a plasma monitoring system/methodology, and yield predictable results (KSR).
Chen, in combination with Turner, Hansen, Mahoney, Carter, Grade, and Kapoor, are silent in regard to:
filter the one or more harmonics by removing harmonics from the one or more harmonics that have the amplitude and the standard deviation less than a pre-defined threshold value.
However, Wiklund, further teaches:
filter the one or more harmonics by removing harmonics from the one or more harmonics that have the amplitude and the standard deviation less than a pre-defined threshold value (Fig. 9; [0029]-[0030]: teaches selecting (equivalent of “removing harmonics…that have the amplitude and standard deviation less than a pre-define threshold value” (i.e., noise-floor zeroing)) specific frequency “bins” (harmonics) out of a Fast Fourier Transform (FFT) spectrum based on criteria including having the least magnitude (amplitude) and the lowest standard deviation, and applying those characteristics to digitally filter the data).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to integrate the digital frequency bin filtering logic of Wiklund into the harmonic spectrum analyzer system of Turner, Hansen, and Kapoor, according to known methods. A POSITA would have been motivated to make this combination to eliminate spectral noise. Real-world plasma environments generate large amounts of “dark current” and baseline electromagnetic noise across the frequency bins. Further, a POSITA would recognize that feeding all of the raw, low-valued data into the FDC comparison would consume unnecessary processing power and risk false alarms. Applying Wiklund’s teachings, the system zeroes out/removes the specific harmonics that fall below the amplitude and standard deviation thresholds (i.e., the bins that contain static noise), cleaning the signal before it reaches the fault detection algorithm. Further, the application of a known Digital Signal Processing (DSP) filtering technique (removing low-magnitude, low-variance FFT bins as taught by Wiklund) to a known harmonic spectrum diagnostic system (Turner/Kapoor) is a simple substitution for one known element for another to obtain the predictable result of a cleaner robust diagnostic signal, and yield predictable results (KSR).
Regarding dependent claim 3, Chen, teaches:
The excursion detection and control system of claim 1 (Disclosed in combination: Chen: [Abstract]; Carter: [Abstract]),
Chen, and Turner, are silent in regard to:
wherein the one or more spectrum analyzers are configured to
However, Turner, in combination with Hansen, and Kapoor, further teach:
wherein the one or more spectrum analyzers (Disclosed in combination: Turner: Fig. 9; [Col. 10, ll. 28-48] & [Claim 29]; Hansen: [0017]: the detection device can be a “photodetector” to “detect changes in light wavelengths” and convert them to an “electronic signal spectrum,” a device that measures intensity vs. wavelength/frequency is a form of spectrum analyzer (e.g., an optical spectrometer); Kapoor: [0007]-[0009]: teaches a signal processor configured to compute a spectral density using a Fast Fourier Transform (FFT) (i.e., a spectrum analyzer)) are configured to (Disclosed in combination: Turner: Fig. 9; [Col. 10, ll. 28-48] & [Claim 29]; Hansen: [0020] & [0023]: excursion detection is referred to as “arcing”; Kapoor: [0007]-[0009])
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the harmonic fingerprinting of Turner, the database-comparison architecture of Hansen, and the statistical standard deviation thresholding of Kapoor to replace the primitive voltage-spike detection of Chen, according to known methods. A POSITA would be motivated to make this combination to achieve a high-fidelity, noise-resistant detection. Further, a POSITA would recognize that to accurately monitor complex plasma physics, one must capture frequency data (Turner’s harmonic fingerprints), however, frequency spectra are complex and an automated comparison to a historical baseline is required (Hansen). The motivation to combine Turner, Hansen, and Kapoor is to create a statistically robust, high-fidelity plasma excursion detection system. To prevent false alarms from normal plasma fluctuations, the POSITA would be further motivated to mathematically define the boundary for this comparison by utilizing standard deviation thresholds derived from the reference spectrum (Kapoor). This combination predictably yields a detection module that relies on statistical process control rather than analog limits, and yield predictable results (KSR).
Chen, in combination with Hansen, Mahoney, Carter, and Grade, are silent in regard to:
detect the excursion when the deviation is greater than a control limit.
However, Turner, in combination with Kapoor, further teach:
detect the excursion when the deviation is greater than a control limit (Disclosed in combination: Turner: Fig. 9; [Col. 10, ll. 28-48] & [Claim 29]: teaches checking control limits to halt or adjust processing based on deviations; Kapoor: [0007]: teaches that the FFT processor detects the anomalous event (excursion) when the signal differs by a threshold amount (control limit)).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to configure the spectrum analyzers of Chen, Turner, and Hansen to detect an excursion when the deviation is greater than a control limit, as taught by Kapoor, according to known methods. A POSITA would be motivated to configure the spectrum analyzer’s processing logic using Kapoor’s statistical threshold amount (i.e., Turner’s control limits). By programming the spectrum analyzer to only trigger a detection signal when the deviation breaches a defined control limit (such as standard deviation threshold), the system filters out normal operational noise. This predictable application of a known Statistical Process Control (SPC) thresholds to known spectral comparison methods ensures that statistically significant excursions trigger the FDC corrective action modules of Mahoney and Gadre, reducing false alarms, improving overall yield, and yielding predictable results (KSR).
Regarding dependent claim 4, Chen, teaches:
The excursion detection and control system of claim 3 (Disclosed in combination: Chen: [Abstract]; Carter: [Abstract]),
Chen, in combination with Turner, are silent in regard to:
wherein the control limit is associated with the reference spectrum.
However, Turner, in combination with Kapoor, further teach:
wherein the control limit is associated with the reference spectrum (Disclosed in combination: Turner: Fig. 9; [Col. 10, ll. 28-48] & [Claim 29]: teaches the system’s use of control limits for bounding process parameters; Kapoor: [0009]-[0010]: associated the threshold/control limit directly with the spectrum, calculating it as a standard deviation of the nominal reference spectral density).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to configure the excursion detection system of Chen, Turner, and Hansen such that its control limit is associated with the reference spectrum, as taught by Kapoor, according to known methods. A POSITA would be motivated to integrate Kapoor’s statistical architecture into the system. By associated the control limit directly with the reference spectrum itself (e.g., configuring the threshold to be exactly one or two standard deviations relative to the baseline reference spectrum), the FDC system becomes called to the specific plasma process being run. This predictable modification ensures the system’s sensitivity is mathematically tied to the variance of the ideal state, thus filtering out normal background noise, preventing false alarms while capturing excursions (true not false-positive), and yield predictable results (KSR).
Regarding dependent claim 9, Chen, teaches:
The excursion detection and control system of claim 1 (Disclosed in combination: Chen: [Abstract] & [0028]; Carter: [Abstract]),
Chen, is silent in regard to:
wherein the plasma processing is performed for at least one of depositing or removing a film on a substrate, and wherein the reference spectrum for comparison with the fingerprint is selected based on one or more control inputs set by a user for the plasma processing.
However, Turner, in combination with Gadre, further teach:
wherein the plasma processing is performed for at least one of depositing or removing a film on a substrate (Disclosed in combination: Turner: [Col. 16, ll. 23-47]: teaches applying the system to both deposition (PECVD) and film removal (etching); Gadre: [0004]-[0006], [Claim 1], [Claim 11] & [Claim 19]: also teaches depositing film),
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the harmonic fingerprinting diagnostics of Turner with the fault library and pattern-matching recommendation module of Gadre, according to known methods. A POSITA would have been motivated to make this combination to automate the interpretation of complex diagnostic data and to enable self-correcting tool control. While Turner teaches extracting harmonic fingerprints to detect shifts in plasma hardware and chemistry, it lacks an automated expert system to diagnose those shifts. A POSITA would be motivated to integrate Gadre’s historical fault library to match Turner’s harmonic deviations against known failure patterns. This predictable combination upgrades the monitoring system into an active controller capable of instantly diagnosing root causes and automatically recommending hardware adjustments, preventing wafer scrap, reducing reliance on manual operator troubleshooting, and yielding predictable expected results (KSR).
However, Turner, in combination with Carter, and Kapoor, further teach:
and wherein the reference spectrum for comparison with the fingerprint is selected based on one or more control inputs set by a user for the plasma processing (Disclosed in combination: Turner: [Col. 3, ll. 42-62] & [Col. 4, ll. 3-10 &14-44]: teaches inputting process parameters (control inputs) to dictate the process control; Carter: [0128] & [0130]-[0131]: teaches a mechanism of selecting a predefined stored function from memory based on user control inputs via an interface; Kapoor: Fig. 6B; [0007]-[0010] & [0056]-[0062]: teaches that the system holds multiple references).
standard deviation of the nominal reference spectral density).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to configure the excursion detection system of Chen, Turner, and Hansen such that the reference spectrum is selected from the historical database based on one or more control inputs set by the user (such as a recipe or process parameter), as taught by Turner, Carter, and Kapoor, according to known methods. A POSITA would understand that a single semiconductor processing chamber is typically used to run different recipes (e.g., etching different materials, or running at 500W RF power versus 1000W RF power). Due to the physics of the plasma changing drastically depending on the recipe, a nominal or ideal harmonic fingerprint for an etch process will look different than an ideal fingerprint for a deposition process. Therefore, if the FDC system maintains a historical database of “one or more reference spectral densities” (Kapoor), it must know which baseline spectrum to use for the comparison so as not to trigger an immediate false alarm. A POSITA would be further motivated to integrate Carter’s teaching of selecting predefined stored functions via user inputs into the system. By linking the user’s input process parameters/recipe (Turner) to the database, the system automatically selects the correct baseline reference spectrum that corresponds to the specific process being run. This predictable combination of known user-interface selections and known database architectures is standard process in configuring multi-recipe process tools, that would yield predictable results (KSR).
Regarding dependent claim 10, Chen, teaches:
The excursion detection and control system of claim 9 (Disclosed in combination: Chen: [Abstract] & [0028]; Carter: [Abstract]),
Chen, is silent in regard to:
wherein the one or more control inputs are configured to identify on at least one of a purpose of the plasma processing, a type of substrate, an amount of substrate, a process chamber pressure, a type of gas, or an amount of gas.
However, Turner, in combination with Mahoney, and Gadre, further teach:
wherein the one or more control inputs are configured to identify on at least one of (Disclosed in combination: Turner: Fig. 9; [Col. 10, ll. 11-13 & 28-36], [Col. 11, ll. 4-9] & [Col. 17, ll. 55-64]: 102, teaches inputting process parameters including gas flow amounts; Mahoney: [0035]-[0038]: teaches system input factors corresponding to gas amounts, gas types, and pressure; Gadre: [0023]-[0024], [0027]-[0029], [0032]-[0034] & [0037]-[0040]: teaches that the inputs (recipe parameters) identify the process chamber pressure and the amount/flow of gas) a purpose of the plasma processing (Turner: [Abstract], [Col. 5, ll. 54-63], [Col. 8, ll. 34-42] & [Col. 10, ll. 2-10]), a type of substrate (Mahoney: Fig. 6; [0003]-[0004], [0009]-[0010], [0013], [0022] & [0035]-[0037]: critical process metrics such as profiles refer to “type of substrate” and measurements also collect “in-situ rate and uniformity (etch or deposition) metrology data”), an amount of substrate (Mahoney: Fig.6; [0003]-[0004], [0009]-[0010], [0013], [0022] & [0035]-[0037]: critical process metrics such as etch rates, depths and profiles refer to “amount of substrate”), a process chamber pressure (Mahoney: [0003]-[0006], [0022], [0031], [0035]-[0037] & [0042]: this is a primary variable managed by the system), a type of gas (Mahoney: [0003]-[0006], [0022], [0031], [0035]-[0037], & [0042]: gas flow manifolds and controls manage “identification of type of gas” and “plasma inputs” include “flow” and “chemistry balance”), or an amount of gas (Mahoney: [0003]-[0006], [0022], [0031], [0035]-[0037] & [0042]: gas flow manifolds and controls manage “identification of amount of gas” and “plasma inputs” include “flow” and “chemistry balance”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to configure the excursion detection system such that the user’s control inputs (used to selected the reference spectrum) identify processing targets such as process chamber pressure or gas flow rates, as taught by Turner, Mahoney, and Gadre, according to known methods. A POSITA would understand that a control input for a plasma chamber is the process recipe entered by the user. A FDC system must know the physical parameters of the recipe being run to select the correct baseline. Integrating the recipe definitions taught by Gadre and Mahoney (e.g., pressure settings, flow rates, and chemistry balances) into the control system is the predictable application of standard industry practice. Ensures the diagnostic system known what physical environment (e.g., what pressure or gas mixture) it is monitoring, allowing it to correctly correlate the real-time harmonic fingerprints to the appropriate baseline, and yield predictable results (KSR).
Regarding dependent claim 11, Chen, teaches in first embodiment:
The excursion detection and control system of claim 1 (Disclosed in combination: Chen: [Abstract] & [0028]; Carter: [Abstract]),
Chen, is silent in regard to:
wherein the one or more spectrum analyzers comprise any combination of an RF input attenuator, a pre-selector, a low-pass filter, a local oscillator, a mixer, an Intermediate Frequency (IF) gain module, an IF filter, an analog-to-digital converter, a digital IF, a fast Fourier transform module, a video bandwidth filter, and a display, working in tandem.
However, Chen, in combination with Turner, Kapoor, and Wiklund, further teach:
wherein the one or more spectrum analyzers (Disclosed in combination: Chen: [Abstract], [0005] & [0007]: spectrum analyzer is described, “measuring the radio frequency (RF) waveform generated at or near an RF power supply”, “The RF waveform is typically measured in analog form and digitized, followed by digital signal processing) comprise any combination of an RF input attenuator (Disclosed in combination: Chen: Fig. 7; [0023]), a pre-selector (Disclosed in combination: Chen: Fig.7; [0024]-[0025], [0027]-[0028], [0036]-[0037], [Claim 7], [Claim 8], [Claim 15] & [Claim 17]: bandpass filter refers to “pre-selector”, where specification states in paragraph [0065] that the pre-selector can also be a low-pass filter), a low-pass filter (Disclosed in combination: Chen: Fig. 7; [0024]-[0025], [0036]-[0037], [Claim 7], [Claim 8], [Claim 15] & [Claim 17]: teaches the analog-to-digital converter and low-pass filters working tandem; Kapoor: [0004] & [0008]: teaches the combination of a low-pass filter and an FFT module), a local oscillator (Disclosed in combination: Chen: Fig. 6; [0018]: local oscillator, as stated in the specification, paragraph [0065], is used for “changing the frequency of a signal”, “unfiltered signal includes both RF driving frequency and other frequencies, such as noise or plasma excursions”, thus changing the frequency of the signal, including additional frequencies such as the RF driving frequency that’s set to 13.56 MHz; Turner: [Col. 7, ll. 11-32]), a mixer (Disclosed in combination: Turner: [Col. 7, ll. 11-32]), an Intermediate Frequency (IF) gain module (Disclosed in combination: Chen: Fig. 6; [0023]-[0025], [0027], [0035]-[0036] & [0038]: IF gain module is well-known in the art to inherently and fundamentally be an amplifier), an IF filter (Disclosed in combination: Chen: Fig. 6; [0023]-[0025], [0027], [0035]-[0036] & [0038]: high pass filter 234 is referred to as the IF filter), an analog-to-digital converter (Disclosed in combination: Chen: Fig. 6; [0005], [0028] & [0038]: analog to digital converter 242, teaches the analog-to-digital converter and low-pass filters working in tandem; Wiklund: [Claim 29]: teaches an analog-to-digital converter (ADC)), a digital IF (Disclosed in combination: Chen: Fig. 6; [0028], [0030] & [0038]: digital IF is referred to as a digital alarm signal 244), a fast Fourier transform module (Disclosed in combination: Turner: [Col. 8, ll. 11-33]; Kapoor: [0004], [0008], [0049], [0062], [0064] & [0066]: teaches the combination of a low-pass filter and an FFT module; Wiklund: Fig. 7; [0022], [0024], [0028]-[0030], [0033] & [Claim 10]), a video bandwidth filter (Disclosed in combination: Kapoor: [0048]: well known in the art that FPGAs implement “video bandwidth filters”), and a display (Disclosed in combination: Chen: Fig. 8; [0016] [0030]: all embodiments have a personal computer 310 that connects to system controller (150) with “data logging and diagnostic software”), working in tandem (Disclosed in combination: Chen: Figs. 2 & 5-8; [Abstract], [0005], [0007], [0028], [Claim 15] & [Claim 17]: various components are all connected and operate together in a system to achieve the goal of plasma excursion detection, teaches the analog-to-digital converter and low-pass and high-pass filters working in tandem; Turner: [Col. 7, ll. 11-32]; Kapoor: [0004] & [0008]: taches the combination of a low-pass filter and an FFT module; Wiklund: [0033] & [Claim 29]: teaches the combination of an analog-to-digital converter (ADC) feeding into a Fast Fourier transform (FFT) module).
PNG
media_image5.png
633
782
media_image5.png
Greyscale
PNG
media_image6.png
798
1012
media_image6.png
Greyscale
It is recognized that the citations and evidence provided above are derived from potentially different embodiments of a single reference. Nevertheless, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to employ combinations and sub-combinations of these complementary embodiments, to configure the spectrum analyzers of the control system to comprise a combination of low-pass filters, analog-to-digital converters, and Fast Fourier Transform (FFT) modules working in tandem, as taught by Chen, Turner, Kapoor, and Wiklund, according to known methods, and otherwise motivating experimentation and optimization. While Kapoor and Wiklund teach analyzing harmonic frequencies using digital FFTs, a POSITA would understand that the analog signal from the plasma chamber must first pass through an analog-to-digital converter, as taught by Wiklund and Chen and/or a mixer/oscillator as taught by Turner, to be processed by the digital FFT algorithm. Furthermore, it is standard digital signal processing (DSP) practice to apply a low-pass filter (anti-aliasing filter, as taught by Kapoor and Chen) prior to the ADC to prevent high-frequency noise from corrupting the FFT spectrum. Therefore, integrating the combination of components (Mixer/Oscillator>Low-Pass Filter>ADC>FFT module) is the application of known elements, standard digital signal processing architecture to the known fault detection systems, yielding the predictable result (KSR) of generating a clean digital spectrum for the FDC module to analyze.
Regarding dependent claim 19, Chen, teaches:
The method of claim 18 ([Abstract]), further comprising:
Chen, is silent in regard to:
calculating, when plasma and chemical conditions of the plasma processing are stable, at least one of an average and standard deviation of amplitude of the one or more harmonics,
However, Turner, in combination with Kapoor, further teach:
calculating (Turner: [Col. 14, ll. 34-39]: quantifying is referred to as “calculating”), when plasma and chemical conditions of the plasma processing are stable (Turner: [Col. 11, ll. 4-9, 18-21, & 26-32] & [Col. 18, ll. 24-29]), at least one of an average (Turner: [Fig. 9; Col. 10, ll. 55-64] & [Col. 11, ll. 4-9]: 112, average algorithm: teaches calculating the mean (average) of voltage and current harmonics during a stable plasma period) and standard deviation (Disclosed in combination: Turner: Fig. 2; [Col. 2, ll. 5-7] & [Col. 14, ll. 34-39 & 47-50]: 40; standard deviation is referred to as “variability”; Kapoor: [0009]-[0010]: teaches calculating standard deviations of spectral densities (frequency amplitudes) while the plasma is maintained under stable/nominal conditions) of amplitude (Turner: [(Col. 8, ll. 34-47]) of the one or more harmonics (Turner: [Col.1, ll. 6-13], [Col. 8, ll. 42-47] & [Col. 11, ll. 26-32]);
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to integrate the standard deviation threshold calculations of Kapoor into the stable baseline averaging of Turner and the reference spectrum framework of Hansen, according to known methods. A POSITA would have been motivated to make this combination to improve the statistical accuracy and robustness of the excursion detection system/methodology. Relying solely on average means or static limits makes a system/methodology vulnerable to false alarms triggered by background noise. By integrating Kapoor’s teaching to calculate the standard deviation under nominal/stable conditions, the system can dynamically set thresholds based on the statistical variance of plasma. Thus, would predictably reduce false-positive alarms, ensuring that only statistically significant deviations trigger the FDC corrective actions of Mahoney and Gadre. Turner teaches a system that monitor RF harmonics in a plasma process and performs statistical process control (SPC) on the data, including averaging and standard deviation (“three sigma”) to establish control limits (thresholds) for go/no-go decisions, Chen further teaches that it is known to compare a processed plasma signal to a preset value. The combination of teachings would improve and optimize a plasma monitoring system/methodology, and yield predictable results (KSR).
Chen, in combination with Turner, Hansen, Mahoney, Carter, Grade, and Kapoor, are silent in regard to:
or filtering the one or more harmonics by removing harmonics from the one or more harmonics that have the amplitude and the standard deviation less than a pre-defined threshold value.
However, Wiklund, further teaches:
or filtering the one or more harmonics by removing harmonics from the one or more harmonics that have the amplitude and the standard deviation less than a pre-defined threshold value (Fig. 9; [0029]-[0030]: teaches selecting (equivalent of “removing harmonics…that have the amplitude and standard deviation less than a pre-define threshold value” (i.e., noise-floor zeroing)) specific frequency “bins” (harmonics) out of a Fast Fourier Transform (FFT) spectrum based on criteria including having the least magnitude (amplitude) and the lowest standard deviation, and applying those characteristics to digitally filter the data).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to integrate the digital frequency bin filtering logic of Wiklund into the harmonic spectrum analyzer system of Turner, Hansen, and Kapoor, according to known methods. A POSITA would have been motivated to make this combination to eliminate spectral noise. Real-world plasma environments generate large amounts of “dark current” and baseline electromagnetic noise across the frequency bins. Further, a POSITA would recognize that feeding all of the raw, low-valued data into the FDC comparison would consume unnecessary processing power and risk false alarms. Applying Wiklund’s teachings, the system zeroes out/removes the specific harmonics that fall below the amplitude and standard deviation thresholds (i.e., the bins that contain static noise), cleaning the signal before it reaches the fault detection algorithm. Further, the application of a known Digital Signal Processing (DSP) filtering technique (removing low-magnitude, low-variance FFT bins as taught by Wiklund) to a known harmonic spectrum diagnostic system (Turner/Kapoor) is a simple substitution for one known element for another to obtain the predictable result of a cleaner robust diagnostic signal, and yield predictable results (KSR).
Regarding dependent claim 20: Chen, teaches:
The method of claim 18 ([Abstract]),
Chen, in combination with Turner, are silent in regard to:
wherein the detecting occurs when the deviation is more than a control limit,
However, Turner, in combination with Kapoor, further teach:
wherein the detecting occurs when the deviation is more than a control limit (Disclosed in combination: Turner: Fig. 9; [Col. 10, ll. 28-48] & [Claim 29]: teaches checking control limits to halt or adjust processing based on deviations; Kapoor: [0007]: teaches detecting the anomalous event when the signal differs by a threshold amount (control limit)),
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to modify the excursion detection method of Chen, Turner, and Hansen by configuring the system/methodology such that detection occurs when the deviation exceeds a control limit that is associated with the reference spectrum, as taught by Kapoor, according to known methods. A POSITA would be motivated to implement Kapoor’s statistical method amount (i.e., Turner’s control limit) to determine a substantial difference, comparing a real-time signal to a reference spectrum to find a deviation. By associating the control limit with the reference spectrum (e.g., setting the limit as one or two standard deviations relative to the nominal references spectrum, as taught by Kapoor), the FDC system ensures minor plasma fluctuations would be ignored, and only statistically significant excursions trigger the control system/methodology. This predictable application of known statistical process control (SPC) techniques to known spectral comparison methods yields an accurate and scaled fault detection system/methodology (KSR).
Chen, in combination with Turner, Hansen, Mahoney, Carter, and Gadre, are silent in regard to:
and wherein the control limit is associated with the reference spectrum.
However, Kapoor, further teaches:
and wherein the control limit is associated with the reference spectrum ([0009]-[0010]: associates the threshold/control limit directly with the spectrum, calculating it as a standard deviation of the nominal reference spectral density).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to modify the excursion detection method of Chen, Turner, and Hansen by configuring the system/methodology such that detection occurs when the deviation exceeds a control limit that is associated with the reference spectrum, as taught by Kapoor, according to known methods. The motivation to combine Kapoor with the primary combination is to provide a precise, mathematical boundary for detecting excursions. While Hansen teaches detecting fault when a signal differs from a reference spectrum, a POSITA would be motivated to define this difference using Kapoor’s standard deviation thresholding. By associating the control limit with the reference spectrum (e.g., setting the limit as one or two standard deviations relative to the nominal references spectrum, as taught by Kapoor), the FDC system ensures minor plasma fluctuations would be ignored, and only statistically significant excursions trigger the control system/methodology. This predictable application of known statistical process control (SPC) techniques to known spectral comparison methods yields an accurate and scaled fault detection system/methodology (KSR), reducing false alarms and improving the statistical reliability of the plasma control system/methodology.
Claims 12-17 are rejected under 35 U.S.C. 103 as being unpatentable over Chen, in view of Turner, in view of Hansen, in view of Mahoney, in view of Pang et al. (US 6193802 B1, Pat. Date Feb. 27, 2001, hereinafter, Pang), in view of Carter, in view of Gadre, in view of Kapoor, in view of Wiklund, and further in view of Shaw et al. (US 2022/0139674 A1, Fil. Date Oct. 26, 2021, hereinafter, Shaw).
Regarding independent claim 12, Chen, teaches:
An excursion detection and control system for detecting excursion in a downstream plasma processing and controlling the downstream plasma processing, the excursion detection and control system comprising (Disclosed in combination: Chen: [Abstract]: teaches an excursion detection and control system; Pang: [Col. 2, ll. 23-33]: teaches downstream plasma processing):
Chen, in combination with Turner, Hansen, and Mahoney, are silent in regard to:
an electrode assembly inserted in an exhaust line of a plasma processing chamber to receive its effluent as a feed gas;
a power generator configured to generate an energy for ionizing the feed gas in the electrode assembly, wherein the ionization of the feed gas leads to a formation of a plasma environment to create a small capacitively coupled plasma;
However, Pang, further teaches:
an electrode assembly inserted in an exhaust line of plasma processing chamber to receive its effluent as a feed gas ([Abstract], [Col. 3, ll. 57-61 & 64-67] & [Col. 4, ll. 1-4, 13-15, 38-41, 45-48 & 61-67], [Col. 5, ll. 1-4], [Col. 10, ll. 5-11], [Col. 12, ll. 36-50], [Claim 17] & [Claim 19]: teaches an electrode assembly positioned in the exhaust line (foreline) that receives the chamber’s effluent (exhaust gas) as feed gas);
a power generator configured to generate an energy for ionizing the feed gas in the electrode assembly ([Abstract], [Col.3, ll. 57-61] & [Col. 27, ll. 29-31]), wherein the ionization ([Col. 27, ll. 29-31]) of the feed gas ([Abstract]) leads to a formation of a plasma environment ([Col. 3, ll. 64-67] & [Col. 4, ll. 1-5]) to create a small capacitively coupled plasma ([Title]; [Col. 3, ll. 57-67], [Col. 4, ll. 1-4 & 13-15], [Col. 12, ll. 36-50], [Claim 1] & [Claim 12]: teaches a power generator to ionize the feed gas between parallel plate electrodes, which inherently creates a capacitively coupled plasma environment);
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of Chen, Turner, Hansen, Mahoney, Pang, Gadre, and Shaw, to arrive at the downstream excursion detection and control system, according to known methods. The primary combination, Chen, Turner, and Hansen focus on in-situ chamber monitoring. A POSITA would recognize that unreacted gases and particles in the exhaust line (foreline) pose massive clogging and emission risks. Furthermore, it would be obvious to a POSITA, to apply the parallel-plate downstream plasma generation apparatus of Pang to create a small capacitively coupled cleaning plasma from the chamber’s effluent. The motivation to apply the excursion detection system of the primary combination to the downstream plasma is to ensure the exhaust scrubber operates correctly and efficiently, yielding expected predictable results (KSR).
Chen, in combination with Hansen, Mahoney, Pang, Carter, Gadre, Kapoor, and Wiklund are silent in regard to:
a dual directional coupler coupled between the electrode assembly and the power generator, and configured to track one or more harmonics that are produced due to non-linearity of an impedance in the plasma environment, wherein the one or more harmonics are associated with at least one of voltage or current;
However, Turner, in combination with Shaw, further teach:
a dual directional coupler coupled between the electrode assembly and the power generator, and configured to track one or more harmonics (Disclosed in combination: Turner: [Col.1, ll. 6-13]; Shaw: [0041] & [0045]) that are produced due to non-linearity (Turner: Fig. 4; [Col. 7, ll. 24-32]: 62; “a non-linear portion that has already shifted some of the radio frequency energy from the signal”) of an impedance in the plasma environment (Turner: [Col.3, ll. 23-41]), wherein the one or more harmonics are associated with at least one of voltage or current (Disclosed in combination: Turner: ([Col.1, ll. 6-13] & [Col. 7, ll. 24-32]): teaches configuring the sensors to track voltage/current harmonics generated by the non-linear impedance of the plasma; Shaw: [0041] & [0045]: teaches the hardware, a dual directional coupler coupled between the generator and the plasma load);
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of Chen, Turner, Hansen, Mahoney, Pang, Gadre, and Shaw, to arrive at the downstream excursion detection and control system, according to known methods. The primary combination, Chen, Turner, and Hansen focus on in-situ chamber monitoring. To monitor the downstream plasma, the system needs accurate RF sensors. A POSITA would be motivated to select the dual directional coupler, as taught by Shaw, because it is standard RF industry practice to simultaneously measure forward and reflected power without signal degradation. It would be further obvious to route these signals into Turner’s harmonic analyzer to extract the non-linear harmonic fingerprints, providing a sensitive diagnostic picture of the downstream plasma health, and yield expected predictable results (KSR).
Chen, in combination with Mahoney, Pang, Carter, Gadre, Kapoor, Wiklund, and Shaw are silent in regard to:
one or more spectrum analyzers configured to:
create a fingerprint of energy distribution in frequency space based on the one or more harmonics,
compare, in real-time, the fingerprint with a reference spectrum of an ideal plasma processing, stored in a historical database, to detect deviation of the fingerprint with the reference spectrum, and
detect an excursion m the downstream plasma processing based on the deviation of the fingerprint with the reference spectrum; and
However, Turner, in combination with Hansen, further teach:
one or more spectrum analyzers configured to (Turner: Fig. 4; [Col. 7, ll. 14-18 & 57-67]: 60):
create a fingerprint of energy distribution in frequency space based on the one or more harmonics (Turner: [Col. 8, ll. 44-47], [Col. 9, ll. 66-67], [Col. 10, ll. 1-2], & [Col. 17, ll. 14-19]: teaches creating a harmonic fingerprint representing the energy distribution),
compare, in real-time (Hansen: [0008]-[0009], [0027] & [Claim 23]: teaches a method of comparing a real-time data spectrum to a prerecorded signature stored in a database to detect a deviation), the fingerprint with a reference spectrum (Hansen: [0008]-[0009]: signature signal is the “created fingerprint”) of an ideal plasma processing (Hansen: [0020], [0023] & [Claim 23]: properly functioning plasma refers to “ideal plasma processing”), stored in a historical database (Hansen: [0019]: signature signal database refers to “historical database”), to detect deviation of the fingerprint with the reference spectrum (Hansen: [0008]-[0009], [0019]-[0020] & [Claim 23]: signature signal is the “fingerprint”, teaches comparing a test signal in real-time to a “prerecorded signature” (reference spectrum) stored in a historical database of the component operating in a “known state” (ideal processing) to detect a fault (deviation/excursion)), and
detect an excursion in the downstream plasma processing based on the deviation of the fingerprint with the reference spectrum (Hansen: [0008]-[0009], [0023] & [Claim 23]: excursion detection is referred to as “arcing” and comparator compares the test signal to the signature signal to determine if signals vary); and
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the harmonic fingerprinting sensors of Turner, with the automated, database-driven comparison architecture of Hansen, according to known methods. A POSITA would have been motivated to make this combination to automate the real-time interpretation of complex plasma diagnostics data and to enable recipe-specific monitoring. While Turner teaches extracting multi-dimensional harmonic fingerprints to monitor plasma, interpreting the data manually is inefficient and prone to error. A POSITA would have been further motivated to integrate Hansen’s automated comparison architecture, which compares live signals to an ideal historical reference spectrum. This predictable combination allows the system to autonomously evaluate Turner’s complex harmonic data in real-time, instantly flagging an excursion the moment the live plasma signature deviates from the known baseline, and yield predictable results (KSR).
Chen, in combination with Turner, Hansen, Pang, Carter, Kapoor, Wiklund, and Shaw are silent in regard to:
a recommendation module configured to:
match the detected deviation with one or more deviation signatures stored in a corrective action database to identify a likely cause of the excursion, wherein each of the deviation signatures is associated with a previously identified excursion cause, and
generate one or more recommendations based on the likely cause to control the excursion via at least one of workpiece conditions, process chamber pressure, gas mixture, RF power, or electrode spacing.
However, Mahoney, in combination with Gadre, further teach:
a recommendation module configured to (Mahoney: [0035]-[0037]):
match the detected deviation with one or more deviation signatures stored in a corrective action database to identify a likely cause of the excursion, wherein each of the deviation signatures is associated with a previously identified excursion cause (Gadre: [0021]-[0025], [0029], [0032] & [0037]-[0040]: teaches matching a detected fault pattern (one or more deviation signatures) against a stored library of known fault patterns (corrective action database) to identify/determine the root cause (previously identified cause) of the failure and recommending a corrective action via process parameters), and
generate one or more recommendations based on the likely cause to control the excursion (Disclosed in combination: Mahoney: [0035]-[0037]: supports the method by teaching parameter adjustments via FDC recommendations including RF power, pressure, and chemistry (gas) balance; Gadre: [0037]-0040]: teaches generating a recommended corrective action to adjust specific process parameters) via at least one of workpiece conditions (Mahoney: [0003]-[0005] & [0035]-[0037]), process chamber pressure (Mahoney: [0003]-[0005] & [0035]-[0037]), gas mixture (Mahoney: [0003]-[0004], [0022] & [0037]: gas flow manifolds are inherently known to control the flow of “process gas mixture” and “chamber pressure”), RF power (Mahoney: [0003]-[0005], [0022] & [0035]-[0037]), or electrode spacing (Mahoney: [0003]-[0006], [0022], [0035], [0037] & [0042]: probes or sensors can also be referred to as an “electrode and/or induction element”, wafer-based probe “device may be disposed in fixed arrays within the processing equipment itself”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of Chen, Turner, Hansen, Mahoney, Pang, Gadre, and Shaw, to arrive at the downstream excursion detection and control system, according to known methods. The primary combination, Chen, Turner, and Hansen focus on in-situ chamber monitoring. A POSITA would recognize that unreacted gases and particles in the exhaust line (foreline) pose massive clogging and emission risks. Furthermore, it would be obvious to a POSITA, to apply the parallel-plate downstream plasma generation apparatus of Pang to create a small capacitively coupled cleaning plasma from the chamber’s effluent. The motivation to apply the excursion detection system of the primary combination to the downstream plasma is to ensure the exhaust scrubber operates correctly and efficiently. A POSITA would be motivated to select the dual directional coupler, as taught by Shaw, because it is standard RF industry practice to simultaneously measure forward and reflected power without signal degradation. It would be further obvious to route these signals into Turner’s harmonic analyzer to extract the non-linear harmonic fingerprints, providing a sensitive diagnostic picture of the downstream plasma health. To automate the correction of the downstream plasma, it would have been obvious to use Hansen’s historical database comparison to spot spectral deviations in real-time, and subsequently feed those deviations into the expert FDC modules of Gadre and Mahoney. This predictably allows the system to instantly match the downstream excursion to a known fault library, identifying the root cause and automatically adjusting the RF power or gas mixtures feeding the exhaust line to restore the scrubber’s efficiently without manual operator intervention, and yield predictable results (KSR).
Regarding dependent claim 13, Chen, teaches:
The excursion detection and control system of claim 12 (Disclosed in combination: Chen: [Abstract]: teaches an excursion detection and control system; Pang: [Col. 2, ll. 23-33]: teaches downstream plasma processing),
Chen, is silent in regard to:
wherein the one or more spectrum analyzers are configured to:
calculate, when plasma and chemical conditions of the downstream plasma processing are stable, at least one of an average or a standard deviation of amplitude of the one or more harmonics, and
However, Turner, in combination with Kapoor, further teach:
wherein the one or more spectrum analyzers (Disclosed in combination: Turner: Fig. 9; [Col. 10, ll. 28-48] & [Claim 29]; Kapoor: [0007]-[0009]: teaches a signal processor configured to compute a spectral density using a Fast Fourier Transform (FFT) (i.e., a spectrum analyzer)) are configured to (Disclosed in combination: Turner: Fig. 9; [Col. 10, ll. 28-48] & [Claim 29]; Kapoor: [0007]-[0009]) :
calculate (Turner: [Col. 14, ll. 34-39] & [Col. 18, ll. 20-34]: quantifying is referred to as “calculating”, teaches calculating the mean (average) of the harmonic amplitudes during a stable plasma period), when plasma and chemical conditions of the downstream plasma processing are stable (Turner: [Col. 11, ll. 4-9, 18-21, & 26-32] & [Col. 18, ll. 24-29]), at least one of an average (Turner: [Fig. 9; Col. 10, ll. 55-64] & [Col. 11, ll. 4-9]: 112, average algorithm: teaches calculating the mean (average) of voltage and current harmonics during a stable plasma period) or a standard deviation (Disclosed in combination: Turner: Fig. 2; [Col. 2, ll. 5-7] & [Col. 14, ll. 34-39 & 47-50]: 40; standard deviation is referred to as “variability”; Kapoor: [0009]-[0010]: teaches calculating standard deviations of spectral densities (frequency amplitudes) while the plasma is maintained under stable/nominal conditions) of amplitude ([Col. 8, ll. 34-47]) of the one or more harmonics (Turner: [Col.1, ll. 6-13], [Col. 8, ll. 42-47] & [Col. 11, ll. 26-32]), and
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to integrate the standard deviation threshold calculations of Kapoor into the stable baseline averaging of Turner and the reference spectrum framework of Hansen, according to known methods. A POSITA would have been motivated to make this combination to improve the statistical accuracy and robustness of the excursion detection system/methodology. Relying solely on average means or static limits makes a system/methodology vulnerable to false alarms triggered by background noise. By integrating Kapoor’s teaching to calculate the standard deviation under nominal/stable conditions, the system can dynamically set thresholds based on the statistical variance of plasma. Thus, would predictably reduce false-positive alarms, ensuring that only statistically significant deviations trigger the FDC corrective actions of Mahoney and Gadre. Turner teaches a system that monitor RF harmonics in a plasma process and performs statistical process control (SPC) on the data, including averaging and standard deviation (“three sigma”) to establish control limits (thresholds) for go/no-go decisions, Chen further teaches that it is known to compare a processed plasma signal to a preset value. The combination of teachings would improve and optimize a plasma monitoring system/methodology, and yield predictable results (KSR).
Chen, in combination with Turner, Hansen, Mahoney, Pang, Carter, Grade, and Kapoor, are silent in regard to:
filter the one or more harmonics by removing harmonics from the one or more harmonics that have the amplitude and the standard deviation less than a pre-defined threshold value.
However, Wiklund, further teaches:
filter the one or more harmonics by removing harmonics from the one or more harmonics that have the amplitude and the standard deviation less than a pre-defined threshold value (Fig. 9; [0029]-[0030]: teaches selecting (equivalent of “removing harmonics…that have the amplitude and standard deviation less than a pre-define threshold value” (i.e., noise-floor zeroing)) specific frequency “bins” (harmonics) out of a Fast Fourier Transform (FFT) spectrum based on criteria including having the least magnitude (amplitude) and the lowest standard deviation, and applying those characteristics to digitally filter the data).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to integrate the digital frequency bin filtering logic of Wiklund into the harmonic spectrum analyzer system of Turner, Hansen, and Kapoor, according to known methods. A POSITA would have been motivated to make this combination to eliminate spectral noise. Real-world plasma environments generate large amounts of “dark current” and baseline electromagnetic noise across the frequency bins. Further, a POSITA would recognize that feeding all of the raw, low-valued data into the FDC comparison would consume unnecessary processing power and risk false alarms. Applying Wiklund’s teachings, the system zeroes out/removes the specific harmonics that fall below the amplitude and standard deviation thresholds (i.e., the bins that contain static noise), cleaning the signal before it reaches the fault detection algorithm. Further, the application of a known Digital Signal Processing (DSP) filtering technique (removing low-magnitude, low-variance FFT bins as taught by Wiklund) to a known harmonic spectrum diagnostic system (Turner/Kapoor) is a simple substitution for one known element for another to obtain the predictable result of a cleaner robust diagnostic signal, and yield predictable results (KSR).
Regarding dependent claim 14, Chen, teaches:
The excursion detection and control system of claim 12 (Disclosed in combination: Chen: [Abstract]: teaches an excursion detection and control system; Pang: [Col. 2, ll. 23-33]: teaches downstream plasma processing),
Chen, and Turner, are silent in regard to:
wherein the one or more spectrum analyzers are configured to
However, Turner, Hansen, and Kapoor further teach:
wherein the one or more spectrum analyzers (Disclosed in combination: Turner: Fig. 9; [Col. 10, ll. 28-48] & [Claim 29]; Hansen: [0017]: the detection device can be a “photodetector” to “detect changes in light wavelengths” and convert them to an “electronic signal spectrum,” a device that measures intensity vs. wavelength/frequency is a form of spectrum analyzer (e.g., an optical spectrometer); Kapoor: [0007]-[0009]: teaches a signal processor configured to compute a spectral density using a Fast Fourier Transform (FFT) (i.e., a spectrum analyzer)) are configured to (Disclosed in combination: Turner: Fig. 9; [Col. 10, ll. 28-48] & [Claim 29]; Hansen: [0020] & [0023]: excursion detection is referred to as “arcing”; Kapoor: [0007]-[0009])
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the harmonic fingerprinting of Turner, the database-comparison architecture of Hansen, and the statistical standard deviation thresholding of Kapoor to replace the primitive voltage-spike detection of Chen, according to known methods. A POSITA would be motivated to make this combination to achieve a high-fidelity, noise-resistant detection. Further, a POSITA would recognize that to accurately monitor complex plasma physics, one must capture frequency data (Turner’s harmonic fingerprints), however, frequency spectra are complex and an automated comparison to a historical baseline is required (Hansen). The motivation to combine Turner, Hansen, and Kapoor is to create a statistically robust, high-fidelity plasma excursion detection system. To prevent false alarms from normal plasma fluctuations, the POSITA would be further motivated to mathematically define the boundary for this comparison by utilizing standard deviation thresholds derived from the reference spectrum (Kapoor). This combination predictably yields a detection module that relies on statistical process control rather than analog limits, and yield predictable results (KSR).
Chen, in combination with Hansen, Mahoney, Pang, Carter, and Grade, are silent in regard to:
detect the excursion when the deviation is greater than a control limit.
However, Turner, and Kapoor further teach:
detect the excursion when the deviation is greater than a control limit (Disclosed in combination: Turner: Fig. 9; [Col. 10, ll. 28-48] & [Claim 29]: teaches checking control limits to halt or adjust processing based on deviations; Kapoor: [0007]: teaches that the FFT processor detects the anomalous event (excursion) when the signal differs by a threshold amount (control limit)).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to configure the spectrum analyzers of Chen, Turner, and Hansen to detect an excursion when the deviation is greater than a control limit, as taught by Kapoor, according to known methods. A POSITA would be motivated to configure the spectrum analyzer’s processing logic using Kapoor’s statistical threshold amount (i.e., Turner’s control limits). By programming the spectrum analyzer to only trigger a detection signal when the deviation breaches a defined control limit (such as standard deviation threshold), the system filters out normal operational noise. This predictable application of a known Statistical Process Control (SPC) thresholds to known spectral comparison methods ensures that statistically significant excursions trigger the FDC corrective action modules of Mahoney and Gadre, reducing false alarms, improving overall yield, and yielding predictable results (KSR).
Regarding dependent claim 15, Chen, teaches:
The excursion detection and control system of claim 14 (Disclosed in combination: Chen: [Abstract]: teaches an excursion detection and control system; Pang: [Col. 2, ll. 23-33]: teaches downstream plasma processing),
Chen, and Turner, are silent in regard to:
wherein the control limit is associated with the reference spectrum.
However, Turner, in combination with Kapoor, further teach:
wherein the control limit is associated with the reference spectrum (Disclosed in combination: Turner: Fig. 9; [Col. 10, ll. 28-48] & [Claim 29]: teaches the system’s use of control limits for bounding process parameters; Kapoor: [0009]-[0010]: associated the threshold/control limit directly with the spectrum, calculating it as a standard deviation of the nominal reference spectral density).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to configure the excursion detection system of Chen, Turner, and Hansen such that its control limit is associated with the reference spectrum, as taught by Kapoor, according to known methods. A POSITA would be motivated to integrate Kapoor’s statistical architecture into the system. By associated the control limit directly with the reference spectrum itself (e.g., configuring the threshold to be exactly one or two standard deviations relative to the baseline reference spectrum), the FDC system becomes called to the specific plasma process being run. This predictable modification ensures the system’s sensitivity is mathematically tied to the variance of the ideal state, thus filtering out normal background noise, preventing false alarms while capturing excursions (true not false-positive), and yield predictable results (KSR).
Regarding dependent claim 16, Chen, teaches:
The excursion detection and control system of claim 12 (Disclosed in combination: Chen: [Abstract]: teaches an excursion detection and control system; Pang: [Col. 2, ll. 23-33]: teaches downstream plasma processing),
Chen, in combination with Turner, and Hansen, are silent in regard to:
to determine a reason behind the excursion and provide the one or more recommendations.
However, Mahoney in combination with Gadre, further teach:
to determine a reason behind the excursion and provide the one or more recommendations (Disclosed in combination: Mahoney: Fig. 6; [0012]-[0013] & [0037]: teaches that the system uses “decision trees” to diagnose the error advise what variables to adjust and further states that this analysis is used for “fault detection and classification (FDC)” and “advanced process control”, where the output is a “FDC Report” that provides a classification and, by implication, a recommendation for action; Gadre: [0023]: teaches that the matching process determines the “root cause” (reason) and outputs a “corrective action” (recommendation)).
It would have been obvious to one having ordinary skill in the art before the effective filing date of the claimed invention was made to incorporate the fault library and historical pattern-matching diagnostic methodologies of Gadre into the smart FDC decision-tree framework of Mahoney, according to known methods. A POSITA would have been motivated to make this combination to improve diagnostic speed, accuracy, and learning capability of the FDC system’s diagnostic modules. While Mahoney teaches using decision trees to find root causes and adjust hardware like RF power or pressure, it lacks a mechanism to leverage historical failure data. A POSITA would be motivated to integrate Gadre’s historical fault library into Mahoney’s FDC system. This predictable modification upgrades the FDC controller from a decision-tree model to an adaptive, database-driven system that can instantly match real-time deviation signatures against previously identified root causes to automate hardware corrections, thus reducing tool downtime, preventing misdiagnosis, and yield predictable results (KSR).
Chen, in combination with Turner, Hansen, Mahoney, Pang, and Carter, are silent in regard to:
wherein the recommendation module is configured to match the deviation with one or more deviations stored in the corrective action database
However, Gadre, further teaches:
wherein the recommendation module is configured to match the deviation with one or more deviations stored in the corrective action database ([0023], [0032] & [0037]-[0040]: teaches matching an observed fault pattern to a database of known historical fault patterns)
It would have been obvious to one having ordinary skill in the art before the effective filing date of the claimed invention was made to configure the recommendation module of the primary combination (Chen/Turner/Hansen) to match detected deviations against a corrective action database to determine the reason behind the excursion and provide recommendations, as taught by Gadre, according to known methods. A POSITA would be motivated to integrate the “fault library” diagnostic architecture of Gadre into the control system. By matching the specific shape or signature of the real-time deviation against historically stored deviations, as taught by Gadre, the system predictably transforms raw fault detection data into actionable intelligence, identifying the exact root cause (reason) and generating hardware adjustments to save the process, yielding predictable results (KSR).
Regarding dependent claim 17, Chen, teaches:
The excursion detection and control system of claim 16 (Disclosed in combination: Chen: [Abstract]: teaches an excursion detection and control system; Pang: [Col. 2, ll. 23-33]: teaches downstream plasma processing),
Chen, is silent in regard to:
and stop the downstream plasma processing based on the excursion such that the subsequent substrates in the downstream plasma processing are saved from the excursion.
However, Chen, in combination with Turner, further teach:
and stop the plasma processing based on the excursion such that the subsequent substrates in the plasma processing are saved from the excursion (Disclosed in combination: Chen: [0028]: teaches shutting down the system upon detecting excursion, which inherently saves subsequent substrates from being processed in a faulty chamber; Turner: [Claim 29], [Claim 36] & [Claim 43]: teaches a halt signal to prevent further processing).
It would have been obvious to one having ordinary skill in the art before the effective filing date of the claimed invention was made to configure the controller of the of the primary excursion detection system to shut down the process to save subsequent substrates, as taught by Chen/Turner, according to known methods. A POSITA would be motivated to configure the system’s controller to execute the two standard industry responses to a detected fault. If the excursion is severe or uncorrectable, then the controller falls back to the safety mechanism taught by Chen and Turner, immediately halting the plasma processing. The secondary response is explained below after the secondary mapping. This prevents the tool from continuing to process and destroy expensive subsequent wafers. Combining these known hardware control methods with the excursion detection logic yields the predictable result (KSR) of a fully automated, self-preserving plasma chamber.
Chen, in combination with Turner, and Hansen, are silent in regard to:
further comprising a controller configured to at least one of control an associated controllable subassembly based on the one or more recommendations such that the excursion is reduced in subsequent substrates,
However, Mahoney, in combination with Gadre, further teach:
further comprising a controller configured to at least one of control an associated controllable subassembly based on the one or more recommendations such that the excursion is reduced in subsequent substrates (Disclosed in combination: Mahoney: Fig. 6; [0002], [0007], [0012]-[0013], [0035]-[0037] & [Claim 4]: teaches using the FDC system to adjust system inputs to return the chamber to nominally acceptable conditions, which reduces the excursion for subsequent processing and increases yield; Gadre: [0017], [0023]-[0024], [0032] & [0037]-0040]),
It would have been obvious to one having ordinary skill in the art before the effective filing date of the claimed invention was made to configure the controller of the of the primary excursion detection system to either adjust a subassembly to fix the excursion, as taught by Mahoney/Gadre, or shut down the process to save subsequent substrates, as taught by Chen/Turner, according to known methods. A POSITA would be motivated to configure the system’s controller to execute the two standard industry responses to a detected fault. If the expert system (Mahoney/Gadre) has a high-confidence recommendation, the controller automatically adjusts the subassembly (e.g., changes the gas flow or RF power) to return the chamber to nominal conditions, ensuring the next wafer processed is defect-free. If the excursion is severe or uncorrectable, then the controller falls back to the safety mechanism taught by Chen and Turner, immediately halting the plasma processing. This prevents the tool from continuing to process and destroy expensive subsequent wafers. Combining these known hardware control methods with the excursion detection logic yields the predictable result (KSR) of a fully automated, self-preserving plasma chamber
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to HUGO NAVARRO whose telephone number is (571)272-6122. The examiner can normally be reached Monday-Friday 08:30-5:00 pm EST.
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, Eman Alkafawi can be reached at 571-272-4448. 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.
/HUGO NAVARRO/ Examiner, Art Unit 2858 April 9, 2026
/EMAN A ALKAFAWI/Supervisory Patent Examiner, Art Unit 2858 4/16/2026
Prosecution Insights