Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Response to Arguments
Applicant's arguments filed 02/27/2026 have been fully considered and in light of the amendments, are persuasive. A new rejection appears below under 35 USC 103 in view of Cherian U.S. Publication 2018/0150052 and Cherian in view of Kitajima U.S. Patent #9,811,077. Additionally, a number of new rejections under 35 USC 112 appear below due to the amendments.
Claim Rejections - 35 USC § 112
The text of those sections of Title 35, U.S. Code not included in this action can be found in a prior Office action.
Claims 1-16 are rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the enablement requirement. The claim(s) contains subject matter which was not described in the specification in such a way as to enable one skilled in the art to which it pertains, or with which it is most nearly connected, to make and/or use the invention.
The independent claims 1, 4, 9, and 12, disclose “the autoencoder configured to reconstruct the input preset spectrum data to output the reconstructed preset spectrum data as the output data”. However, this limitation is not supported in the specification in sufficient detail that one of ordinary skill in the art can make or use the invention. It is not defined exactly what happens within the autoencoder. The specification discloses training the autoencoder on ideal data but there are no calculations, algorithms, or steps disclosed that the autoencoder performs on the input data. The autoencoder is essentially a black box. Correction is required.
Additionally, the limitation, “when the optical film-thickness measuring system works properly” is not defined in the specification and seems to conflict with other teachings that disclose some of the preset spectra data is classified as “abnormal” (P.0054). “Works properly” is relative term that has no definition in the specification. There might be some middle ground for preset data from something that “works properly” yet still outputs “abnormal” data, however it is unclear and not disclosed in the specification such that one of ordinary skill in the art would understand the inventor’s intention. Correction is required.
With respect to claims 2, 6, 10, and 14, the limitation “classifying the plurality of preset spectra data…according to algorithm of clustering” is not sufficiently supported because the breadth and depth of clustering algorithms leave too much guess and test to a person of ordinary skill in the art without teaching as to how to apply it. The number of groups and how to determine whether one is normal or which set to use for training is not disclosed. The clustering algorithm is treated as a black box covering any and all possible clustering algorithms which the specification does not have support for. Correction is required.
Claim 1-8 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
With respect to claims 1-8, the limitation “wherein the preset spectrum data is used for optical film-thickness measurement during polishing” is unclear if optical film-thickness measurement is actually performed. The limitation is not a positive limitation in the method steps or for the processor to configure to perform, but rather an aside about the intentions of how to use the preset spectrum data. Clarification is required.
The balance of claims that are dependent upon the listed claims above fail to correct the deficiencies noted above.
Claim Rejections - 35 USC § 103
The text of those sections of Title 35, U.S. Code not included in this action can be found in a prior Office action.
Claims 1, 2, 4, 5, 6, 9, 10, 12, 13, and 14 are rejected under 35 U.S.C. 103 as unpatentable over Cherian U.S. Publication 2018/0150052.
With respect to claim 1, Cherian discloses a spectrographic monitoring using a neural network comprising:
Method for measuring abnormality detection of preset spectrum data for optically measuring film thickness of a workpiece (abstract, film thickness = thickness of an outer layer)
Configuring an optical film-thickness measuring system for use in polishing with the preset spectrum data before polishing of the workpiece (P.0009, P.0032, wherein “during polishing” means that polishing occurs afterwards as well, so measurement system is intended to be used before polishing is finished)
Inputting the preset spectrum data to an autoencoder which is a trained model constructed by machine learning using training data including a plurality of normal preset spectrum data (P.0049, P.0051, P.0064)
and outputting output data from the autoencoder to evaluate whether the optical film thickness measuring system is operating under normal calibration conditions, the autoencoder being configured to reconstruct the input preset data to output the reconstructed preset data as the output (P.0068, P.0071-2, wherein operating under normal calibration conditions = has reached a desired endpoint, reconstruct input preset data = principal component values P reconstructed to spectrum Q)
Calculating a difference between the output data output from the autoencoder and the preset data representing a deviation from normal conditions ( P.0071, preset spectra data = principal component values P, output data = reconstructed spectrum Q)
Determining there is an abnormality in the preset spectrum data when the difference is large (P.0071, if the difference is large, the associated characteristic value can be ignored by the process module =considered abnormal)
Polishing the workpieces when the difference is small, the preset spectrum data is used for optical film-thickness measurements during polishing when the abnormality is not determined (P.0071, P.0072)
Cherian fails to disclose comparing the difference to a threshold.
It would have been obvious to one of ordinary skill in the art at the time of filing to use a threshold for determining abnormalities since the machine learning needs a specific cut off number to classify things. Thresholds are almost inherent to comparisons, being ubiquitous to analyzing data results. Thresholds give meaning to data. Cherian discloses “if the difference is large” but large can mean many different things, so providing an exact threshold is common sense for repeatability.
It should be noted that the limitation “wherein the preset spectrum data is used for optical film thickness measurement during polishing when the abnormality is not determined” is not limiting on the claims since it is not a positively cited method step.
With respect to claim 4, Cherian discloses a spectrographic monitoring using a neural network comprising:
Method for measuring abnormality detection of preset spectrum data for optically measuring film thickness of a workpiece (abstract, film thickness = thickness of an outer layer)
Providing a plurality of preset spectra data acquired in the past when an optical film thickness measuring system works properly (P.0049, inherent that it was working properly or the processing would not be carried out)
Determining reference spectrum data which is one of a plurality of preset spectra data acquired in the past (P.0025)
Generating a latest preset spectrum data obtained by the optical film thickness measuring system (P.0071, latest preset spectrum data = reconstructing the spectrum)
Calculating a difference between the reference spectrum data and the latest preset spectrum data representing a deviation from normal conditions ( P.0071, preset spectra data = principal component values P, output data = reconstructed spectrum Q)
Polishing the workpiece when the difference is smaller than the threshold value (P.0071, P.0072)
Determining there is an abnormality in the preset spectrum data when the difference is large (P.0071, if the difference is large, the associated characteristic value can be ignored by the process module =considered abnormal)
Polishing the workpieces when the difference is small, the preset spectrum data is used for optical film-thickness measurements during polishing when the abnormality is not determined (P.0071, P.0072)
It should be noted that the limitation “wherein the preset spectrum data is used for optical film thickness measurement during polishing when the abnormality is not determined” is not limiting on the claims since it is not a positively cited method step.
With respect to claim 9, Cherian discloses a spectrographic monitoring system using a neural network comprising:
An optical film-thickness measuring apparatus for optically measuring film thickness of a workpiece (abstract, film thickness = thickness of an outer layer)
A light source configured to emit light (P.0033)
An optical sensor head configured to irradiate the workpiece with the light emitted by the light source and receive reflected light from the workpiece (P.0033, optical sensor head = optical monitoring system, Figure 1)
A processing system configured to determine film thickness of the workpiece based on spectrum measurement data of the reflected light from the workpiece and preset spectrum data (P.0034, processing system = controller, P.0038, P.0051)
The processing system including an autoencoder which is a trained model (P.0060, P.0011) configured to:
Input the preset spectrum data to the autoencoder and outputting output data from the autoencoder, the autoencoder being configured to reconstruct the input preset data to output the reconstructed preset data as the output (P.0068, P.0071-2, wherein operating under normal calibration conditions = has reached a desired endpoint, reconstruct input preset data = principal component values P reconstructed to spectrum Q)
Calculate a difference between the output data output from the autoencoder and the preset data representing a deviation from normal conditions ( P.0071, preset spectra data = principal component values P, output data = reconstructed spectrum Q)
Determine there is an abnormality in the preset spectrum data when the difference is large (P.0071, if the difference is large, the associated characteristic value can be ignored by the process module =considered abnormal)
Determine the film thickness of the workpiece based on the preset spectrum data when the difference is smaller than the threshold value (P.0071, P.0072, P.0073, wherein an endpoint indicates a desired thickness so inherently is measuring thickness indirectly)
Cherian fails to disclose comparing the difference to a threshold.
It would have been obvious to one of ordinary skill in the art at the time of filing to use a threshold for determining abnormalities since the machine learning needs a specific cut off number to classify things. Thresholds are almost inherent to comparisons, being ubiquitous to analyzing data results. Thresholds give meaning to data. Cherian discloses “if the difference is large” but large can mean many different things, so providing an exact threshold is common sense for repeatability.
It should be noted that the limitation “wherein the preset spectrum data is used for optical film thickness measurement during polishing when the abnormality is not determined” is not limiting on the claims since it is not a positively cited method step.
With respect to claim 12, Cherian discloses a spectrographic monitoring system using a neural network comprising:
An optical film-thickness measuring apparatus for optically measuring film thickness of a workpiece (abstract, film thickness = thickness of an outer layer)
A light source configured to emit light (P.0033)
An optical sensor head configured to irradiate the workpiece with the light emitted by the light source and receive reflected light from the workpiece (P.0033, optical sensor head = optical monitoring system, Figure 1)
A processing system configured to determine film thickness of the workpiece based on spectrum measurement data of the reflected light from the workpiece and preset spectrum data (P.0034, processing system = controller, P.0038, P.0051)
The processing system including an autoencoder which is a trained model (P.0060, P.0011) configured to:
Import a plurality of preset spectra data acquired in the past when the optical film-thickness measuring apparatus works properly (P.0051, inherent that it works properly otherwise the measurement would not be performed)
Determining reference spectrum data which is one of a plurality of preset spectra data acquired in the past (P.0025)
Create a latest preset spectrum data before polishing the workpiece (P.0071, latest preset spectrum data = reconstructing the spectrum, in-situ with polishing, meaning there is more polishing after the calculations)
Calculating a difference between the reference spectrum data and the latest preset spectrum data representing a deviation from normal conditions ( P.0071, preset spectra data = principal component values P, output data = reconstructed spectrum Q)
Determining there is an abnormality in the preset spectrum data when the difference is large (P.0071, if the difference is large, the associated characteristic value can be ignored by the process module =considered abnormal)
Determine the film thickness of the workpiece based on the preset spectrum data when the difference is smaller than the threshold value (P.0071, P.0072, P.0073, wherein an endpoint indicates a desired thickness so inherently is measuring thickness indirectly)
Cherian fails to disclose comparing the difference to a threshold.
It would have been obvious to one of ordinary skill in the art at the time of filing to use a threshold for determining abnormalities since the machine learning needs a specific cut off number to classify things. Thresholds are almost inherent to comparisons, being ubiquitous to analyzing data results. Thresholds give meaning to data. Cherian discloses “if the difference is large” but large can mean many different things, so providing an exact threshold is common sense for repeatability.
It should be noted that the limitation “wherein the preset spectrum data is used for optical film thickness measurement during polishing when the abnormality is not determined” is not limiting on the claims since it is not a positively cited method step.
With respect to claim 2, 6, 10, and 14, Cherian discloses all of the limitations as applied to claim 1, 4, 9, and 12. In addition, Cherian discloses:
Providing/importing a plurality of preset spectra data acquired in the past when the optical film thickness measuring system works properly (P.0049, inherent that it was working properly or the processing would not be carried out)
Classifying the plurality of preset spectra data acquired in the past into groups (P.0049, P.0051, P.0052, groups = groups of eigenvectors)
Producing the training data including a plurality of normal preset spectra data belonging to one of the groups (P.0052, P.0059)
Performing the machine learning using the training data to construct the autoencoder which is the trained model (P.0061-62, where matrix T is generated in P.0056 from the various reference spectra A in P.0055)
However, Cherian fails to disclose algorithm clustering to classify the data into groups.
The examiner takes Official Notice that it would have been obvious to one of ordinary skill in the art at the time of filing to use algorithm clustering on the reference spectrum set of Cherian. Algorithm clustering is a generic data processing method that isolates typical data from outliers. Clustering historical measurements of Cherian and selecting the normal cluster to use going forward into training in order to reduce data and get a more accurate data set representative of the mean.
With respect to claim 5 and 13, Cherian discloses all of the limitations as applied to claim 4 and 12 above. In addition, Cherian discloses:
The difference is a sum of squared difference (P.0071)
However, Cherian fails to disclose the difference is a Euclidean difference.
It would have been obvious to one of ordinary skill in the art at the time of filing to use a Euclidean difference rather than sum of squared as in Cherian since both are well known comparison methods. The examiner takes Official Notice of that the difference between two spectra may be computed as a Euclidean distance, which is the square root of the sum of squared differences disclosed by Cherian.
Claims 3, 7, 8, 11, 15, and 16 are rejected under 35 U.S.C. 103 as unpatentable over Cherian U.S. Publication 2018/0150052 in view of Kitajima U.S. Patent #9,811,077.
With respect to claim 3, 8, 11, and 16, Cherian discloses all of the limitations as applied to claim 1, 4, 9, and 12 above. In addition, Cherian discloses:
Using in-situ intensity spectrum data (P.0053)
However, Cherian fails to disclose the preset spectrum data and latest preset spectrum data is one of base intensity data, dark level data containing background intensity, or light monitoring data containing intensity of the irradiation light.
Kitajima discloses polishing with pre deposition spectrum measurement comprising:
Correcting raw spectrum data with a stored based spectrum and one or more dark spectra (Col.9, l 11-33)
It would have been obvious to one of ordinary skill in the art at the time of filing to apply the stored correction traces of Kitajima to the raw spectrum data of Cherian since corruption of the base spectrum or dark spectrum would introduce errors into the thickness analysis that both references are trying to calculate. It is well known in the art to use based spectrum data or dark data to correct spectrum measurements to negate influence not attributed to the sample being measured.
With respect to claim 7 and 15, Cherian discloses all of the limitations as applied to claim 1 and 12 above. In addition, Cherian discloses:
A stored set of reference spectra, a newly measured spectrum, a difference between them reconstructed and treating a large difference as indicating an abnormality (P.0048-52, P.0071)
However, Cherian fails to disclose normalizing the plurality of preset spectra data acquired and normalizing the latest preset spectrum data.
Kitajima discloses:
Raw optical spectra for film-thickness monitoring are normalized and comparing the normalized spectrum to a reference (abstract, Col.8, l 48-49, Col.9, l 1-4)
It would have been obvious to one of ordinary skill in the art at the time of filing to apply the normalization of Kitajima to both sides of the comparison of Cherian (the original preset and the latest preset spectrum data) since the normalization would remove factors such as lamp drift, window transmission, or other intensity based error sources is simply applying a known technique with expectation of success.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to REBECCA CAROLE BRYANT whose telephone number is (571)272-9787. The examiner can normally be reached M-F, 12-4 pm.
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, Kara Geisel can be reached at 571-272-2416. 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.
/REBECCA C BRYANT/Primary Examiner, Art Unit 2877
Prosecution Insights