DETAILED ACTION
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Responsive to the communication dated 03/13/2026
Claims 1-20 are presented for examination
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and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114.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 03/13/2026 has been entered.
Response to Arguments – Claim Objections
Applicant’s arguments, see page 10, filed 03/13/2026, with respect to the objections to claims 5-6, 11, and 16-17 have been fully considered and are persuasive. The objections to claims 5-6, 11, and 16-17 have been withdrawn.
Response to Arguments - 101
Applicant's arguments filed 03/13/2026 have been fully considered but they are not persuasive.
Applicant argues that the newly made amendments overcome the rejections under 101.
Examiner responds by addressing the amendments and the arguments directed to the annealing/processing step.
Firstly:
the lamp array having a honeycomb shaped pattern with one or more vacancies therein;
Specifying that the array has one or more vacancies merely clarifies the form of the modelled lamp array, and is therefore merely an extension of the mental process of developing the model.
As to the annealing/processing step:
processing a substrate in the existing RTP tool using the temperature profile to anneal the substrate.
This final step of annealing the substrate using the temperature profile merely acts on the results of the abstract idea, to which the claims are directed; this is effectively equivalent to a final step of cutting hair after a mental process of designing a hairstyle to which the claims are directed.
A claim element that merely acts on a series of previous abstract steps is not indicative of integration into a practical solution nor evidence that the claim provides an inventive concept, as exemplified by ((MPEP 2106.05)(g)(Insignificant application) i. Cutting hair after first determining the hair style, In re Brown, 645 Fed. App'x 1014, 1016-1017 (Fed. Cir. 2016) and ii. Printing or downloading generated menus, Ameranth, 842 F.3d at 1241-42, 120 USPQ2d at 1854-55.)
As to the applicant’s arguments about annealing the substrate being a particular transformation, it is important to note that the particularity of the transformation is an essential component to determining whether a step is sufficient to integrate the claims into a practical application/ provide significantly more. (MPEP 2106.05(c) “An "article" includes a physical object or substance. The physical object or substance must be particular, meaning it can be specifically identified. "Transformation" of an article means that the "article" has changed to a different state or thing. Changing to a different state or thing usually means more than simply using an article or changing the location of an article. A new or different function or use can be evidence that an article has been transformed. Purely mental processes in which thoughts or human based actions are "changed" are not considered an eligible transformation. For data, mere "manipulation of basic mathematical constructs [i.e.,] the paradigmatic ‘abstract idea,’" has not been deemed a transformation. CyberSource v. Retail Decisions, 654 F.3d 1366, 1372 n.2, 99 USPQ2d 1690, 1695 n.2 (Fed. Cir. 2011) (quoting In re Warmerdam, 33 F.3d 1354, 1355, 1360, 31 USPQ2d 1754, 1755, 1759 (Fed. Cir. 1994)).
Tilghman v. Proctor, 102 U.S. 707 (1881), provides an example of effecting a transformation of a particular article to a different state or thing. In that case, the claim was directed to a process of subjecting a mixture of fat and water to a high degree of heat and included additional parameters relating to the level of heat, the quantities of fat and water, and the strength of the mixing vessel. The claimed process, which used the natural principle that the elements of neutral fat require that they be severally united with an atomic equivalent of water in order to separate and become free, resulted in the transformation of the fatty bodies into fat acids and glycerine. Id. at 729.”)
With this in mind there is not sufficient particularity recited as to how this annealing is performed, let alone how the generated temperature profile factors into this annealing process. A clear description of how this annealing is performed and specifically how the generated temperature profile is “used” in the annealing process, which is not present, would be key to showing that this is a particular transformation. Further, upon turning to the disclosure to find more details about the annealing and how the generated temperature profile affects this annealing, significant issues of support arise. See below.
It should be noted that upon further consideration and examination of the specification, there does not seem to be support for using the temperature profile to anneal the substrate, nor support for doing anything with the generated profile beyond simply outputting it. The closest disclosure in the specification to describing the use of the profile for annealing is [Par 27], which merely discloses generating a profile that will be “implemented” on a substrate, without any specifics as to what this implementation could be, for the purpose of comparing different designs. (“As will be described in greater detail below, the lamp array160 may be a theoretical or hypothetical lamp array160. That is, the lamp array160 may not be physically built. However, as a result of analysis methods, such as those described in greater detail below, the lamp array160 can be analyzed in order to determine the temperature profile that will be implemented on a substrate. As such, the output of the RTP tool can be characterized and compared to existing solutions in order to determine if the design should be built out into an actual product.”)
Further, the specification describes inputting the irradiance graph to the ML algorithm and using the algorithm to generate a “temperature profile” [Par 22], “temperature uniformity plot” [Par 38], a “temperature snapshot” [Par 36], and a “temperature across a hypothetical substrate” [Par 44]. It is unclear if these are meant to refer to the same element, but using any of these to perform an annealing operation is not disclosed, with the usage of these outputs being solely for characterizing the performance of the simulated model ([Par 44] “In an embodiment, the process690 may continue with operation696, which comprises using the irradiation graph (or graphs) as an input to the ML algorithm. The irradiation graph (or graphs) may be inputted in the ML algorithm that was trained in operation691. In an embodiment, the process690 may continue with operation697, which comprises outputting the temperature across a hypothetical substrate from the machine learning algorithm. As such, the performance of the RTP tool can be determined without the need to build the model of the RTP tool. Accordingly, many different models may be easily investigated using a similar process in order to select the best candidates for further consideration with minimal cost and development time.”) Put simply, there appears to only be support for outputting the results of the abstract analysis, not any further application of this analysis to perform a process.
Response to Arguments - 103
Applicant's arguments filed 03/13/2026 have been fully considered but they are not persuasive.
Applicant argues that no prior art teaches a lamp array having a honeycomb shaped pattern with one or more vacancies therein;
Examiner responds by explaining that this feature is taught by new reference New_Ranish (US 20060066193 A1)
In particular, New_Ranish teaches the same hexagonal/honeycomb pattern (note that “hexagonal” and “honeycomb” patterns are the same thing) as the previously cited Ranish reference ([Par 6] “A thermal processing chamber includes a substrate support rotating about a center axis and a lamphead of plural lamps in an array having a predetermined difference in radiance pattern between them. The radiance pattern includes a variation in diffuseness or collimation. In one embodiment, the center lines of all of the lamps are disposed away from the center axis. The array can be a hexagonal array, in which the center axis is located at a predetermined position between neighboring lamps.” [Fig. 5] shows the honeycomb array shape)
PNG
media_image1.png
631
575
media_image1.png
Greyscale
As for the vacancies within the array, firstly the spaces pointed out in the annotated version of Fig. 5 below read on the broadest reasonable interpretation of the lamp array “having vacancies” i.e. having positions that do not have lamps.
PNG
media_image2.png
631
575
media_image2.png
Greyscale
Further, New_Ranish also explicitly teaches removing one of the lamps to create such a vacancy, and how such vacancies help better determine the ideal configuration of the array. ([Par 52-53] “… both temperature measurements and process results indicate some radial ripple persists in the radiation pattern. Especially, the center of the wafer seems to be excessively heated and a sharp irradiance peak exists there. The ripple can be somewhat reduced by controlling the zone heating to even out the ripples. However, the ripple is still considered excessive and more fundamental solutions to the ripple problem are sought. … The schematic cross-sectional view of FIG. 7 taken along a radius of the RTP chamber illustrates the effective radial positions (±2 mm) of all lamps 36 in the hexagonal array and their radiation patterns 148. There are several radii having closely spaced lamps 36 and other radii having a gap. The maldistribution is particularly severe around the center lamp 36C. Although control of the zone heating can reduce the severity of the problem, even 15 zones appear insufficient to completely solve the ripple problem. … The simulation was repeated with the center lamp 36C removed, that is, not contributing any radiation to the total radiation pattern. The difference between a flat profile out to 50 mm and the profile without the center lamp 36C was then calculated to produce an ideal profile 156 for the center lamp 36C, as illustrated in FIGS. 8 and 10.”)
New_Ranish is analogous art because it is within the field of rapid thermal processing. It would have been obvious to one of ordinary skill in the art to combine it with Huang, Aderhold, and Lu before the effective filing date. One of ordinary skill in the art would have been motivated to make this combination in order to more uniformly process the substrate. New_Ranish notes how previous RTP systems suffer from unstable heating towards the center of the target semiconductor component ([Par 51-52] “ The radiation pattern 140 output by such an array is illustrated in FIG. 6. It includes a similar hexagonal array of bright spots 142 surrounded by a more diffuse and lower intensity background 144. However, the rotation of the wafer is intended to average out the bright spots 142 to produce a more uniform time-averaged radiation pattern. … However, both temperature measurements and process results indicate some radial ripple persists in the radiation pattern. Especially, the center of the wafer seems to be excessively heated and a sharp irradiance peak exists there. The ripple can be somewhat reduced by controlling the zone heating to even out the ripples. However, the ripple is still considered excessive and more fundamental solutions to the ripple problem are sought. We believe that a substantial cause of the ripple phenomenon in RTP is caused by two effects. First, the zone immediately surrounding the center lamp 36C does not benefit from wafer rotation since there radiation results there is no other lamp to average over. Secondly, an hexagonal array centered on the center lamp 36C and rotation axis 17 inherently produces radial oscillations, particularly near the center. The schematic cross-sectional view of FIG. 7 taken along a radius of the RTP chamber illustrates the effective radial positions (.+-.2 mm) of all lamps 36 in the hexagonal array and their radiation patterns 148. There are several radii having closely spaced lamps 36 and other radii having a gap. The maldistribution is particularly severe around the center lamp 36C. Although control of the zone heating can reduce the severity of the problem, even 15 zones appear insufficient to completely solve the ripple problem.”) To this end, New_Ranish presents a method for more uniformly processing wafers/substrates in part by removing certain lamps during the optimization process to produce an improved temperature profile ([Par 53-54] “Many simulations have been performed to quantify the geometrical effects of the finite hexagonal arrays. The radiation pattern of a standard lamp has been measured in a plane at distance from the source representative of an RTP chamber and as a function of the transverse direction (radius) from the axis of the lamp within the plane. The helical filament of a standard lamp has about eight turns extending over about 15 mm with the back of the nearest turn disposed adjacent the face of the water cooled housing or in front of it. A standard profile 152 is illustrated in the graph of FIG. 8. A computer program then calculates the total radiation intensity, as illustrated in the graph of FIG. 8 for all 409 lamps at respective positions in the array and averages this distribution for the rotation of the wafer about the center of the lamp array. The resulting total radial profile 154 of FIG. 9 exhibits a distinct peak at the center. Aside from ripple inside 50 mm, it was generally flat outwardly of the center to about 150 mm, as is desired for a 300 mm wafer. The simulation was repeated with the center lamp 36C removed, that is, not contributing any radiation to the total radiation pattern. The difference between a flat profile out to 50 mm and the profile without the center lamp 36C was then calculated to produce an ideal profile 156 for the center lamp 36C, as illustrated in FIGS. 8 and 10. … Accordingly, in one embodiment of the invention, it is desired to replace the center lamp 36C with a non-standard lamp while leaving standard lamps in the remaining sockets. In general, the center lamp 36C should produce a more diffuse pattern than the remaining lamps. This embodiment has the advantage of not requiring modification of the lamphead but only requiring modification of a replaceable lamp. The embodiment further allows all but one of the lamps to be optimized for intensity or other parameter while restricting the ripple improvement to only the center lamp 36C. It is possible to carry the lamp optimization further by separately optimizing the six lamps 36 of the innermost hexagon resulting in three different sets of lamps.”) While the removed center lamp is later replaced in some embodiments, it is clear that its removal is an essential part of the optimization process and an essential component of developing a configuration to uniformly process wafers/substrates. Overall, one of ordinary skill in the art would have recognized that combining New_Ranish with Huang, Aderhold, and Lu would result in a system capable of more uniform wafer/substrate processing, ultimately resulting in more successfully processed wafers/substrates.
Claim Rejections - 35 USC § 112
The following is a quotation of the first paragraph of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
The following is a quotation of the first paragraph of pre-AIA 35 U.S.C. 112:
The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor of carrying out his invention.
Claims 1-20 rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the written description requirement. The claim(s) contains subject matter which was not described in the specification in such a way as to reasonably convey to one skilled in the relevant art that the inventor or a joint inventor, or for applications subject to pre-AIA 35 U.S.C. 112, the inventor(s), at the time the application was filed, had possession of the claimed invention.
Claims 1,12, and 19 recite “processing a substrate in the existing RTP tool using the temperature profile to anneal the substrate.”
There is not sufficient support for using the temperature profile to anneal the substrate, nor processing the substrate using the temperature profile in general. The closest disclosure in the specification to describing the use of the profile for any kind of further is [Par 27], which merely discloses generating a profile that will be “implemented” on a substrate, without any specifics as to what this implementation could be, for the purpose of comparing different designs. (“As will be described in greater detail below, the lamp array160 may be a theoretical or hypothetical lamp array160. That is, the lamp array160 may not be physically built. However, as a result of analysis methods, such as those described in greater detail below, the lamp array160 can be analyzed in order to determine the temperature profile that will be implemented on a substrate. As such, the output of the RTP tool can be characterized and compared to existing solutions in order to determine if the design should be built out into an actual product.”) Further, the only mention of annealing in the specification is within the description of related art, generally explaining that annealing is something that RTP tools could be used for as part of a non-exhaustive list including other examples. ([Par 2] “2) DESCRIPTION OF RELATED ART In semiconductor processing environments, rapid thermal processing (RTP) tools are used, for example, in order to execute thermal treatments (e.g., anneals) and grow material layers (e.g., oxidation growth), to name a couple applications”) This brief description of prior art is not the same as a description of the disclosed invention itself. Further, even if this was taken to be a positive recitation of annealing, the use of such a temperature profile to inform the annealing process is still undisclosed.
While the thermal soak and thermal ramp down/ cooldown, which are part of the larger annealing process, are mentioned in the disclosure, the use of the profile to adjust either of these is also not disclosed.
Put simply, the disclosure describes generating such thermal profiles to analyze the performance of simulated RTP tools, but the claimed final step of using this analysis to inform the actual annealing of a substrate on an existing machine lacks sufficient support in the disclosure.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-20 are rejected under 35 U.S.C. 101 because they are directed to an abstract idea without significantly more.
Claim 1 (Statutory Category – Process)
Step 2A – Prong 1: Judicial Exception Recited?
Yes, the claim recites a mental process, specifically:
MPEP 2106.04(a)(2)(Ill): “Accordingly, the "mental processes" abstract idea grouping is defined as concepts performed in the human mind, and examples of mental processes include observations, evaluations, Judgments, and opinions.”
Further, the MPEP recites “The courts do not distinguish between mental processes that are performed entirely in the human mind and mental processes that require a human to use a physical aid (e.g., pen and paper or a slide rule) to perform the claim limitation.”
A method comprising: developing a lamp model of a rapid thermal processing (RTP) tool, wherein the lamp model comprises a plurality of lamp zones of a lamp array, the lamp array having a honeycomb shaped pattern with one or more vacancies therein;
Developing such a model is a mental process equivalent to observing and then drawing an image related to such an RTP tool with a pencil and paper.
using the irradiation graph as an input to a machine learning algorithm; outputting a temperature profile across a hypothetical substrate from the machine learning algorithm;
Determining the temperature of something based on known heat output of a heating mechanism is a mental process equivalent to observing the heating information and estimating the temperature of the heated object. For example, an experienced chef might estimate that after cooking a brisket for a few hours while an oven is set to 400 degrees, the internal temperature is getting close to the set 400 degrees.
Doing this through the use of a generic machine learning algorithm amounts to more than mere instructions to apply the exception using a general-purpose computer.
Should it be found that this is not a mental process, it is also an example of mere data gathering.
The claim also recites a mathematic concept, specifically:
calculating an irradiance graph for the plurality of lamp zones; multiplying irradiance values of the plurality of lamp zones in the irradiance graph by a power of an existing RTP tool at a given time during a process recipe; summing the multiplied irradiance values for the plurality of lamp zones to form an irradiation graph of the lamp model;
This is clearly a mathematic calculation recited in textual form, and is therefore merely a mathematic process.
Additionally, should it be found that these steps are not purely mathematic, the step of creating the graph itself after calculating its constituent data mathematically is a mental process equivalent to plotting the calculated data with a pen and paper.
Step 2A – Prong 2: Integrated into a Practical Solution?
Insignificant Extra-Solution Activity (MPEP 2106.05(g)) has found mere data gathering and
post solution activity to be insignificant extra-solution activity.
Data gathering:
using the irradiation graph as an input to a machine learning algorithm;
Generically “using” data as input is merely the act of gathering that input data into the system or algorithm and therefore amounts to no more than mere data gathering.
outputting a temperature profile across a hypothetical substrate from the machine learning algorithm.
Generically “outputting” this data without any particularities as to how this output is actually generated amounts to no more than gathering data representative of that output, and therefore amounts to no more than mere data gathering.
Post-solution activity:
…and processing a substrate in the existing RTP tool using the temperature profile to anneal the substrate.
This final step of processing the substrate using the temperature profile merely acts on the results of the abstract idea, to which the claims are directed; this is effectively equivalent to a final step of cutting hair after a mental process of designing a hairstyle to which the claims are directed. A claim element that merely acts on a series of previous abstract steps is not indicative of integration into a practical solution nor evidence that the claim provides an inventive concept, as exemplified by ((MPEP 2106.05)(g)(Insignificant application) i. Cutting hair after first determining the hair style, In re Brown, 645 Fed. App'x 1014, 1016-1017 (Fed. Cir. 2016) and ii. Printing or downloading generated menus, Ameranth, 842 F.3d at 1241-42, 120 USPQ2d at 1854-55.)
Mere Instructions to Apply (MPEP 2106.05(f)) has found that merely applying a judicial exception such as an abstract idea, as by performing it on a computer, does not integrate the claim into a practical solution.
Mere Instructions to Apply:
using the irradiation graph as an input to a machine learning algorithm; outputting a temperature profile across a hypothetical substrate from the machine learning algorithm.
Applying a machine learning algorithm to perform an abstract process at a high level of generality is simply the act of instructing a computer to perform generic functions to operate as that machine learning algorithm and carry out that process, which is merely an instruction to apply a computer to the judicial exception. The claim only recites the idea of a solution or outcome, i.e. that the temperature is output using the machine learning algorithm without reciting how this prediction is actually accomplished nor any specifics about the algorithm. Further, the computer elements claimed are cited as merely generic tools to perform the operations; for additional clarity see ([Par 38] “The structure of the ML algorithm may be any type of ML algorithm.”)
The courts have found that such mere instructions to apply are not indicative of integration into a practical application nor recitation of significantly more than the judicial exception (MPEP 2106.05(f) “Another consideration when determining whether a claim integrates a judicial exception into a practical application in Step 2A Prong Two or recites significantly more than a judicial exception in Step 2B is whether the additional elements amount to more than a recitation of the words "apply it" (or an equivalent) or are more than mere instructions to implement an abstract idea or other exception on a computer. As explained by the Supreme Court, in order to make a claim directed to a judicial exception patent-eligible, the additional element or combination of elements must do "‘more than simply stat[e] the [judicial exception] while adding the words ‘apply it’". Alice Corp. v. CLS Bank, 573 U.S. 208, 221, 110 USPQ2d 1976, 1982-83 (2014) (quoting Mayo Collaborative Servs. V. Prometheus Labs., Inc., 566 U.S. 66, 72, 101 USPQ2d 1961, 1965). Thus, for example, claims that amount to nothing more than an instruction to apply the abstract idea using a generic computer do not render an abstract idea eligible. Alice Corp., 573 U.S. at 223, 110 USPQ2d at 1983”)
Step 2B: Claim provides an Inventive Concept?
No, as discussed with respect to Step 2A, the additional limitations are Insignificant Extra-Solution Activity or mere instructions to apply and do not impose any meaningful limits on practicing the abstract idea and therefore the claim does not provide an inventive concept in Step 2B.
Insignificant Extra-Solution Activity (MPEP 2106.05(g)) has found mere data gathering and
post solution activity to be insignificant extra-solution activity.
Data gathering:
using the irradiation graph as an input to a machine learning algorithm;
Generically “using” data as input is merely the act of gathering that input data into the system or algorithm and therefore amounts to no more than mere data gathering.
A claim element that amounts to merely gathering data is not indicative of integration into a
practical solution nor evidence that the claim provides an inventive concept or significantly more, as exemplified by ((MPEP 2106.05)(g)(Mere Data Gathering) i. Performing clinical tests on individuals to obtain input for an equation, In re Grams, 888 F.2d 835, 839-40; 12 USPQ2d 1824, 1827-28 (Fed. Cir. 1989); iv. Obtaining information about transactions using the Internet to verify credit card transactions, CyberSource v. Retail Decisions, Inc., 654 F.3d 1366, 1375, 99 USPQ2d 1690, 1694 (Fed. Cir. 2011);
outputting a temperature profile across a hypothetical substrate from the machine learning algorithm.
Generically “outputting” this data without any particularities as to how this output is actually generated amounts to no more than gathering data representative of that output, and therefore amounts to no more than mere data gathering.
A claim element that amounts to merely gathering data is not indicative of integration into a
practical solution nor evidence that the claim provides an inventive concept or significantly more, as exemplified by ((MPEP 2106.05)(g)(Mere Data Gathering) i. Performing clinical tests on individuals to obtain input for an equation, In re Grams, 888 F.2d 835, 839-40; 12 USPQ2d 1824, 1827-28 (Fed. Cir. 1989); iv. Obtaining information about transactions using the Internet to verify credit card transactions, CyberSource v. Retail Decisions, Inc., 654 F.3d 1366, 1375, 99 USPQ2d 1690, 1694 (Fed. Cir. 2011);
Post-solution activity:
…and processing a substrate in the existing RTP tool using the temperature profile to anneal the substrate.
This final step of processing the substrate using the temperature profile merely acts on the results of the abstract idea, to which the claims are directed; this is effectively equivalent to a final step of cutting hair after a mental process of designing a hairstyle to which the claims are directed. A claim element that merely acts on a series of previous abstract steps is not indicative of integration into a practical solution nor evidence that the claim provides an inventive concept, as exemplified by ((MPEP 2106.05)(g)(Insignificant application) i. Cutting hair after first determining the hair style, In re Brown, 645 Fed. App'x 1014, 1016-1017 (Fed. Cir. 2016) and ii. Printing or downloading generated menus, Ameranth, 842 F.3d at 1241-42, 120 USPQ2d at 1854-55.)
Further, the specification clearly suggests that processing substrates with an RTP tool using a particular temperature configuration is a common practice in the art. ([Par 2] “DESCRIPTION OF RELATED ART In semiconductor processing environments, rapid thermal processing (RTP) tools are used, for example, in order to execute thermal treatments (e.g., anneals) and grow material layers (e.g., oxidation growth), to name a couple applications. In an RTP tool, an array of lamps are used in order to heat a substrate that is positioned below the lamps. A reflector may also be provided below the substrate in some instances. Temperature control across the surface of the substrate is a critical parameter of RTP tools. Often the temperature is desired to be substantially uniform across the diameter of the substrate.”)
Mere Instructions to Apply (MPEP 2106.05(f)) has found that merely applying a judicial exception such as an abstract idea, as by performing it on a computer, does not integrate the claim into a practical solution.
Mere Instructions to Apply:
using the irradiation graph as an input to a machine learning algorithm; outputting a temperature profile across a hypothetical substrate from the machine learning algorithm.
Applying a machine learning algorithm to perform an abstract idea process at a high level of generality is simply the act of instructing a computer to perform generic functions to operate as that machine learning algorithm and carry out that process, which is merely an instruction to apply a computer to the judicial exception. The claim only recites the idea of a solution or outcome, i.e. that the temperature is output using the machine learning algorithm without reciting how this prediction is actually accomplished nor any specifics about the algorithm. Further, the computer elements claimed are cited as merely generic tools to perform the operations; for additional clarity see ([Par 38] “The structure of the ML algorithm may be any type of ML algorithm.)
The courts have found that such mere instructions to apply are not indicative of integration into a practical application nor recitation of significantly more than the judicial exception (MPEP 2106.05(f) “Another consideration when determining whether a claim integrates a judicial exception into a practical application in Step 2A Prong Two or recites significantly more than a judicial exception in Step 2B is whether the additional elements amount to more than a recitation of the words "apply it" (or an equivalent) or are more than mere instructions to implement an abstract idea or other exception on a computer. As explained by the Supreme Court, in order to make a claim directed to a judicial exception patent-eligible, the additional element or combination of elements must do "‘more than simply stat[e] the [judicial exception] while adding the words ‘apply it’". Alice Corp. v. CLS Bank, 573 U.S. 208, 221, 110 USPQ2d 1976, 1982-83 (2014) (quoting Mayo Collaborative Servs. V. Prometheus Labs., Inc., 566 U.S. 66, 72, 101 USPQ2d 1961, 1965). Thus, for example, claims that amount to nothing more than an instruction to apply the abstract idea using a generic computer do not render an abstract idea eligible. Alice Corp., 573 U.S. at 223, 110 USPQ2d at 1983”)
The additional elements have been considered both individually and as an ordered combination in the consideration of whether they constitute significantly more, and have been determined not to constitute such.
The claim is ineligible.
Claim 2 recites “training the machine learning algorithm with training data that includes real temperature data from the existing RTP tool.”
Training a machine learning model, recited at such a high level of generality, amounts to no more than mere instructions to apply the judicial exception.
Applying a computer to perform generic training operations at a high level of generality is simply the act of instructing a computer to perform generic functions to perform that training, which is merely an instruction to apply a computer to the judicial exception. The claim only recites the idea of a solution or outcome, i.e. that the machine learning model is trained, without reciting how this training is actually accomplished. Further, the computer elements claimed are cited as merely generic tools to perform the operations; for additional clarity see ([Par 38] “For example, the ML algorithm may be a supervised ML algorithm, a semi-supervised ML algorithm, an unsupervised ML algorithm, a reinforcement ML algorithm, or the like.”)
The courts have found that such mere instructions to apply are not indicative of integration into a practical application nor recitation of significantly more than the judicial exception (MPEP 2106.05(f) “Another consideration when determining whether a claim integrates a judicial exception into a practical application in Step 2A Prong Two or recites significantly more than a judicial exception in Step 2B is whether the additional elements amount to more than a recitation of the words "apply it" (or an equivalent) or are more than mere instructions to implement an abstract idea or other exception on a computer. As explained by the Supreme Court, in order to make a claim directed to a judicial exception patent-eligible, the additional element or combination of elements must do "‘more than simply stat[e] the [judicial exception] while adding the words ‘apply it’". Alice Corp. v. CLS Bank, 573 U.S. 208, 221, 110 USPQ2d 1976, 1982-83 (2014) (quoting Mayo Collaborative Servs. V. Prometheus Labs., Inc., 566 U.S. 66, 72, 101 USPQ2d 1961, 1965). Thus, for example, claims that amount to nothing more than an instruction to apply the abstract idea using a generic computer do not render an abstract idea eligible. Alice Corp., 573 U.S. at 223, 110 USPQ2d at 1983”)
Additionally, generically training a machine learning model at a high level of generality is an example of well-understood, routine, conventional activity. See the below:
Machine learning, explained ([Page 3 Par 3])
What Is Machine Learning (ML)? ([Page 2 Par 8 – Page 3 Par 5, Page 4 Par 5- Page 5 Par 1])
Developing machine learning algorithms for meteorological temperature and humidity forecasting at Terengganu state in Malaysia ([Page 2 Par 6])
Predicting Weather Temperature Change Using Machine Learning Models ([Page 3 Par -Page 4 Par 2])
Claim 3 recites “wherein the training includes at least 25 sets of different training data.”
This merely clarifies the quantity of data used to perform the training, and is therefore no more than an extension of the mere instructions to apply.
Claim 4 recites “wherein the plurality of lamp zones includes up to 15 lamp zones.”
This merely clarifies the form of the lamp model, and is therefore merely an extension of the mental process.
Claim 5 recites “wherein the lamp array of the lamp model is different than a lamp array of the existing RTP tool.”
This merely clarifies the relationship between the existing tool and the related model made of it, and is therefore merely an extension of the mental process.
Claim 6 recites “wherein a number of lamps in the lamp array of the lamp model is different than a number of lamps in the lamp array of the existing RTP tool.”
This merely clarifies the relationship between the existing tool and the related model made of it, and is therefore merely an extension of the mental process.
Claim 7 recites “wherein the machine learning algorithm comprises two or more hidden layers.”
This merely clarifies the structure of the machine learning algorithm, and is therefore merely an extension of the mere instructions to apply.
Claim 8 recites “wherein the irradiation graph includes data points for at least 15 different positions on the hypothetical substrate.”
This merely clarifies the data included in the irradiation graph, and therefore is merely an extension of the mathematic process of calculating its form and the mental process of drawing it.
Claim 9 recites “wherein the given time during a process recipe is during a thermal soak.”
This merely clarifies what time the RTP data is from, and is therefore merely an extension of the mathematic process of calculating the form of the irradiation graph.
Claim 10 recites “wherein the given time during the process recipe is during a thermal ramp.”
This merely clarifies what time the RTP data is from, and is therefore merely an extension of the mathematic process of calculating the form of the irradiation graph.
Claim 11 recites “wherein the temperature profile across the hypothetical substrate matches a set of training data.”
This merely clarifies the relationship between the training data and output data, and is therefore merely an extension of the mental process and mere instructions to apply that mental process.
Claim 18 recites “wherein the given time during a process recipe is during a thermal soak and/or during a thermal ramp.”
This merely clarifies what time the RTP data is from, and is therefore merely an extension of the mathematic process of calculating the form of the irradiation graph.
Claims 12-17 The elements of claims 12-17 are substantially the same as those of claims 1-6. Therefore, the elements of claims 12-17 are rejected due to the same reasons as outlined above for claims 1-6.
Moreover, Mere Instructions To Apply An Exception (MPEP 2106.05(f)) has found that simply adding a general purpose computer or computer components after the fact to an abstract idea (e.g., a fundamental economic practice or mathematical equation) does not integrate a judicial exception into a practical application or provide significantly more. In light of this, the additional generic computer component elements of claim 12, namely “A non-transitory computer readable medium containing program instructions for causing a computer to perform the method comprising:” are not sufficient to integrate a judicial exception into a practical application nor provide evidence of an inventive concept.
Claim 19 (Statutory Category – Process)
Step 2A – Prong 1: Judicial Exception Recited?
Yes, the claim recites a mental process, specifically:
MPEP 2106.04(a)(2)(Ill): “Accordingly, the "mental processes" abstract idea grouping is defined as concepts performed in the human mind, and examples of mental processes include observations, evaluations, Judgments, and opinions.”
Further, the MPEP recites “The courts do not distinguish between mental processes that are performed entirely in the human mind and mental processes that require a human to use a physical aid (e.g., pen and paper or a slide rule) to perform the claim limitation.”
A method comprising: … developing a lamp model of an RTP tool, wherein the lamp model comprises a plurality of lamp zones of a lamp array, the lamp array having a honeycomb shaped pattern with one or more vacancies therein, and wherein a number of lamps in the lamp model is different than a number of lamps in the existing RTP tool;
Developing such a model is a mental process equivalent to observing and then drawing an image related to such an RTP tool with a pencil and paper, the model being drawn with a different number of lamps
using the irradiation graph as an input to a machine learning algorithm; and outputting a temperature profile across a hypothetical substrate from the machine learning algorithm; and
Determining the temperature of something based on known heat output of a heating mechanism is a mental process equivalent to observing the heating information and estimating the temperature of the heated object. For example, an experienced chef might estimate that after cooking a brisket for a few hours while an oven is set to 400 degrees, the internal temperature is getting close to the set 400 degrees.
Doing this through the use of a generic machine learning algorithm amounts to more than mere instructions to apply the exception using a general-purpose computer.
Should it be found that this is not a mental process, it is also an example of mere data gathering.
The claim also recites a mathematic concept, specifically:
calculating an irradiance graph for the plurality of lamp zones; multiplying irradiance values of the plurality of lamp zones in the irradiance graph by a power of an existing RTP tool at a given time during a process recipe; summing the multiplied irradiance values for the plurality of lamp zones to form an irradiation graph of the lamp model;
This is clearly a mathematic calculation recited in textual form, and is therefore merely a mathematic process.
Additionally, should it be found that these steps are not purely mathematic, the step of creating the graph itself after calculating its constituent data mathematically is a mental process equivalent to plotting the calculated data with a pen and paper.
Step 2A – Prong 2: Integrated into a Practical Solution?
Insignificant Extra-Solution Activity (MPEP 2106.05(g)) has found mere data gathering and
post solution activity to be insignificant extra-solution activity.
Data gathering:
using the irradiation graph as an input to the machine learning algorithm;
Generically “using” data as input is merely the act of gathering that input data into the system or algorithm and therefore amounts to no more than mere data gathering.
outputting a temperature profile across a hypothetical substrate from the machine learning algorithm.
Generically “outputting” this data without any particularities as to how this output is actually generated amounts to no more than gathering data representative of that output, and therefore amounts to no more than mere data gathering.
Post-solution activity:
…and processing a substrate in the existing RTP tool using the temperature profile to anneal the substrate.
This final step of processing the substrate using the temperature profile merely acts on the results of the abstract idea, to which the claims are directed; this is effectively equivalent to a final step of cutting hair after a mental process of designing a hairstyle to which the claims are directed. A claim element that merely acts on a series of previous abstract steps is not indicative of integration into a practical solution nor evidence that the claim provides an inventive concept, as exemplified by ((MPEP 2106.05)(g)(Insignificant application) i. Cutting hair after first determining the hair style, In re Brown, 645 Fed. App'x 1014, 1016-1017 (Fed. Cir. 2016) and ii. Printing or downloading generated menus, Ameranth, 842 F.3d at 1241-42, 120 USPQ2d at 1854-55.)
Mere Instructions to Apply (MPEP 2106.05(f)) has found that merely applying a judicial exception such as an abstract idea, as by performing it on a computer, does not integrate the claim into a practical solution.
Mere Instructions to Apply:
“training the machine learning algorithm with training data that includes real temperature data from the existing rapid thermal processing (RTP) tool.”
Training a machine learning model, recited at such a high level of generality, amounts to no more than mere instructions to apply the judicial exception.
Applying a computer to perform generic training operations at a high level of generality is simply the act of instructing a computer to perform generic functions to perform that training, which is merely an instruction to apply a computer to the judicial exception. The claim only recites the idea of a solution or outcome, i.e. that the machine learning model is trained, without reciting how this training is actually accomplished. Further, the computer elements claimed are cited as merely generic tools to perform the operations; for additional clarity see ([Par 38] “For example, the ML algorithm may be a supervised ML algorithm, a semi-supervised ML algorithm, an unsupervised ML algorithm, a reinforcement ML algorithm, or the like.”)
The courts have found that such mere instructions to apply are not indicative of integration into a practical application nor recitation of significantly more than the judicial exception (MPEP 2106.05(f) “Another consideration when determining whether a claim integrates a judicial exception into a practical application in Step 2A Prong Two or recites significantly more than a judicial exception in Step 2B is whether the additional elements amount to more than a recitation of the words "apply it" (or an equivalent) or are more than mere instructions to implement an abstract idea or other exception on a computer. As explained by the Supreme Court, in order to make a claim directed to a judicial exception patent-eligible, the additional element or combination of elements must do "‘more than simply stat[e] the [judicial exception] while adding the words ‘apply it’". Alice Corp. v. CLS Bank, 573 U.S. 208, 221, 110 USPQ2d 1976, 1982-83 (2014) (quoting Mayo Collaborative Servs. V. Prometheus Labs., Inc., 566 U.S. 66, 72, 101 USPQ2d 1961, 1965). Thus, for example, claims that amount to nothing more than an instruction to apply the abstract idea using a generic computer do not render an abstract idea eligible. Alice Corp., 573 U.S. at 223, 110 USPQ2d at 1983”)
using the irradiation graph as an input to the machine learning algorithm; and outputting a temperature profile across a hypothetical substrate from the machine learning algorithm.
Applying a machine learning algorithm to perform an abstract idea process at a high level of generality is simply the act of instructing a computer to perform generic functions to operate as that machine learning algorithm and carry out that process, which is merely an instruction to apply a computer to the judicial exception. The claim only recites the idea of a solution or outcome, i.e. that the temperature is output using the machine learning algorithm without reciting how this prediction is actually accomplished nor any specifics about the algorithm. Further, the computer elements claimed are cited as merely generic tools to perform the operations; for additional clarity see ([Par 38] “The structure of the ML algorithm may be any type of ML algorithm.”)
The courts have found that such mere instructions to apply are not indicative of integration into a practical application nor recitation of significantly more than the judicial exception (MPEP 2106.05(f) “Another consideration when determining whether a claim integrates a judicial exception into a practical application in Step 2A Prong Two or recites significantly more than a judicial exception in Step 2B is whether the additional elements amount to more than a recitation of the words "apply it" (or an equivalent) or are more than mere instructions to implement an abstract idea or other exception on a computer. As explained by the Supreme Court, in order to make a claim directed to a judicial exception patent-eligible, the additional element or combination of elements must do "‘more than simply stat[e] the [judicial exception] while adding the words ‘apply it’". Alice Corp. v. CLS Bank, 573 U.S. 208, 221, 110 USPQ2d 1976, 1982-83 (2014) (quoting Mayo Collaborative Servs. V. Prometheus Labs., Inc., 566 U.S. 66, 72, 101 USPQ2d 1961, 1965). Thus, for example, claims that amount to nothing more than an instruction to apply the abstract idea using a generic computer do not render an abstract idea eligible. Alice Corp., 573 U.S. at 223, 110 USPQ2d at 1983”)
Step 2B: Claim provides an Inventive Concept?
No, as discussed with respect to Step 2A, the additional limitations are Insignificant Extra-Solution Activity or mere instructions to apply and do not impose any meaningful limits on practicing the abstract idea and therefore the claim does not provide an inventive concept in Step 2B.
Insignificant Extra-Solution Activity (MPEP 2106.05(g)) has found mere data gathering and
post solution activity to be insignificant extra-solution activity.
Data gathering:
using the irradiation graph as an input to the machine learning algorithm;
Generically “using” data as input is merely the act of gathering that input data into the system or algorithm and therefore amounts to no more than mere data gathering.
A claim element that amounts to merely gathering data is not indicative of integration into a
practical solution nor evidence that the claim provides an inventive concept or significantly more, as exemplified by ((MPEP 2106.05)(g)(Mere Data Gathering) i. Performing clinical tests on individuals to obtain input for an equation, In re Grams, 888 F.2d 835, 839-40; 12 USPQ2d 1824, 1827-28 (Fed. Cir. 1989); iv. Obtaining information about transactions using the Internet to verify credit card transactions, CyberSource v. Retail Decisions, Inc., 654 F.3d 1366, 1375, 99 USPQ2d 1690, 1694 (Fed. Cir. 2011);
outputting a temperature profile across a hypothetical substrate from the machine learning algorithm;
Generically “outputting” this data without any particularities as to how this output is actually generated amounts to no more than gathering data representative of that output, and therefore amounts to no more than mere data gathering.
A claim element that amounts to merely gathering data is not indicative of integration into a
practical solution nor evidence that the claim provides an inventive concept or significantly more, as exemplified by ((MPEP 2106.05)(g)(Mere Data Gathering) i. Performing clinical tests on individuals to obtain input for an equation, In re Grams, 888 F.2d 835, 839-40; 12 USPQ2d 1824, 1827-28 (Fed. Cir. 1989); iv. Obtaining information about transactions using the Internet to verify credit card transactions, CyberSource v. Retail Decisions, Inc., 654 F.3d 1366, 1375, 99 USPQ2d 1690, 1694 (Fed. Cir. 2011);
Post-solution activity:
…and processing a substrate in the existing RTP tool using the temperature profile to anneal the substrate.
This final step of processing the substrate using the temperature profile merely acts on the results of the abstract idea, to which the claims are directed; this is effectively equivalent to a final step of cutting hair after a mental process of designing a hairstyle to which the claims are directed. A claim element that merely acts on a series of previous abstract steps is not indicative of integration into a practical solution nor evidence that the claim provides an inventive concept, as exemplified by ((MPEP 2106.05)(g)(Insignificant application) i. Cutting hair after first determining the hair style, In re Brown, 645 Fed. App'x 1014, 1016-1017 (Fed. Cir. 2016) and ii. Printing or downloading generated menus, Ameranth, 842 F.3d at 1241-42, 120 USPQ2d at 1854-55.)
Further, the specification clearly suggests that processing substrates with an RTP tool using a particular temperature configuration is a common practice in the art. ([Par 2] “DESCRIPTION OF RELATED ART In semiconductor processing environments, rapid thermal processing (RTP) tools are used, for example, in order to execute thermal treatments (e.g., anneals) and grow material layers (e.g., oxidation growth), to name a couple applications. In an RTP tool, an array of lamps are used in order to heat a substrate that is positioned below the lamps. A reflector may also be provided below the substrate in some instances. Temperature control across the surface of the substrate is a critical parameter of RTP tools. Often the temperature is desired to be substantially uniform across the diameter of the substrate.”)
Mere Instructions to Apply (MPEP 2106.05(f)) has found that merely applying a judicial exception such as an abstract idea, as by performing it on a computer, does not integrate the claim into a practical solution.
Mere Instructions to Apply:
“training the machine learning algorithm with training data that includes real temperature data from the existing rapid thermal processing (RTP) tool.”
Training a machine learning model, recited at such a high level of generality, amounts to no more than mere instructions to apply the judicial exception.
Applying a computer to perform generic training operations at a high level of generality is simply the act of instructing a computer to perform generic functions to perform that training, which is merely an instruction to apply a computer to the judicial exception. The claim only recites the idea of a solution or outcome, i.e. that the machine learning model is trained, without reciting how this training is actually accomplished. Further, the computer elements claimed are cited as merely generic tools to perform the operations; for additional clarity see ([Par 38] “For example, the ML algorithm may be a supervised ML algorithm, a semi-supervised ML algorithm, an unsupervised ML algorithm, a reinforcement ML algorithm, or the like.”)
The courts have found that such mere instructions to apply are not indicative of integration into a practical application nor recitation of significantly more than the judicial exception (MPEP 2106.05(f) “Another consideration when determining whether a claim integrates a judicial exception into a practical application in Step 2A Prong Two or recites significantly more than a judicial exception in Step 2B is whether the additional elements amount to more than a recitation of the words "apply it" (or an equivalent) or are more than mere instructions to implement an abstract idea or other exception on a computer. As explained by the Supreme Court, in order to make a claim directed to a judicial exception patent-eligible, the additional element or combination of elements must do "‘more than simply stat[e] the [judicial exception] while adding the words ‘apply it’". Alice Corp. v. CLS Bank, 573 U.S. 208, 221, 110 USPQ2d 1976, 1982-83 (2014) (quoting Mayo Collaborative Servs. V. Prometheus Labs., Inc., 566 U.S. 66, 72, 101 USPQ2d 1961, 1965). Thus, for example, claims that amount to nothing more than an instruction to apply the abstract idea using a generic computer do not render an abstract idea eligible. Alice Corp., 573 U.S. at 223, 110 USPQ2d at 1983”)
using the irradiation graph as an input to the machine learning algorithm; and outputting a temperature profile across a hypothetical substrate from the machine learning algorithm.
Applying a machine learning algorithm to perform an abstract idea process at a high level of generality is simply the act of instructing a computer to perform generic functions to operate as that machine learning algorithm and carry out that process, which is merely an instruction to apply a computer to the judicial exception. The claim only recites the idea of a solution or outcome, i.e. that the temperature is output using the machine learning algorithm without reciting how this prediction is actually accomplished nor any specifics about the algorithm. Further, the computer elements claimed are cited as merely generic tools to perform the operations; for additional clarity see ([Par 38] “The structure of the ML algorithm may be any type of ML algorithm.”)
The courts have found that such mere instructions to apply are not indicative of integration into a practical application nor recitation of significantly more than the judicial exception (MPEP 2106.05(f) “Another consideration when determining whether a claim integrates a judicial exception into a practical application in Step 2A Prong Two or recites significantly more than a judicial exception in Step 2B is whether the additional elements amount to more than a recitation of the words "apply it" (or an equivalent) or are more than mere instructions to implement an abstract idea or other exception on a computer. As explained by the Supreme Court, in order to make a claim directed to a judicial exception patent-eligible, the additional element or combination of elements must do "‘more than simply stat[e] the [judicial exception] while adding the words ‘apply it’". Alice Corp. v. CLS Bank, 573 U.S. 208, 221, 110 USPQ2d 1976, 1982-83 (2014) (quoting Mayo Collaborative Servs. V. Prometheus Labs., Inc., 566 U.S. 66, 72, 101 USPQ2d 1961, 1965). Thus, for example, claims that amount to nothing more than an instruction to apply the abstract idea using a generic computer do not render an abstract idea eligible. Alice Corp., 573 U.S. at 223, 110 USPQ2d at 1983”)
Additionally, generically training a machine learning model at a high level of generality is an example of well-understood, routine, conventional activity. See the below:
Machine learning, explained ([Page 3 Par 3])
What Is Machine Learning (ML)? ([Page 2 Par 8 – Page 3 Par 5, Page 4 Par 5- Page 5 Par 1])
Developing machine learning algorithms for meteorological temperature and humidity forecasting at Terengganu state in Malaysia ([Page 2 Par 6])
Predicting Weather Temperature Change Using Machine Learning Models ([Page 3 Par -Page 4 Par 2])
The additional elements have been considered both individually and as an ordered combination in the consideration of whether they constitute significantly more, and have been determined not to constitute such.
The claim is ineligible.
Claim 20 The elements of claim 20 are substantially the same as those of claim 18. Therefore, the elements of claims 20 are rejected due to the same reasons as outlined above for claim 18.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
(1) Claims 1, 4, 8-10, 12, 15, and 18 are rejected under 35 U.S.C. 103 as being unpatentable over A Design Method to Improve Temperature Uniformity on Wafer for Rapid Thermal Processing (Hereinafter Huang) in view of Aderhold (US 20100124249 A1) in further view of A Novel Spatiotemporal LS-SVM Method for Complex Distributed Parameter Systems With Applications to Curing Thermal Process (Hereinafter Lu) as well as New_Ranish (US 20060066193 A1) and Miller (US 20140170805 A1)
Claim 1. Huang teaches A method comprising: developing a lamp model of a rapid thermal processing (RTP) tool, wherein the lamp model comprises a plurality of lamp zones of a lamp array, ([Abstract] “Single-wafer rapid thermal processing (RTP) is widely used in semiconductor manufacturing. Achieving temperature uniformity on silicon wafer is a major challenge in RTP control. In this work, a lamp configuration including five concentric lamp zones is designed to obtain uniform temperature distribution on the wafer. An optics-based model is developed to determine the optimal lamp design parameters, and a uniformity criterion is proposed to evaluate the effective irradiance distribution of the tungsten–halogen lamps on the wafer. This method can be used to determine geometric parameters of the lamp array in order to achieve uniform temperature distribution on the wafer. A realistic simulation of a cold wall RTP system with five lamp rings and a 200-mm wafer is performed”) calculating an irradiance ([Page 3 Par 5] “In the following section, the lamp number N on each zone, the radius R of each zone and the height h of the lamp array are three variables chosen to calculate the effective irradiance density absorbed by a point on the wafer.”) multiplying irradiance values of the plurality of lamp zones in the irradiance ([Page 4 Equations 3 and 4] Equation 4 sums the irradiance values calculated in equation 3 at each lamp in each zone, which is calculated by an equation involving multiplication by the power of an existing RTP tool (i.e. a lamp). Note that in calculating the total irradiance in Equation 4 each initial irradiance value is multiplied by term nij before being summed.
PNG
media_image3.png
368
1419
media_image3.png
Greyscale
([Page 2 Par 1] “Simulation results show that the design yields good temperature uniformity distribution on the wafer.”)
Huang does not explicitly teach the lamp array having a honeycomb shaped pattern with one or more vacancies therein; an irradiance graph; multiplying values on a graph by a value at a given time during a process recipe; using the irradiation graph as an input to a machine learning algorithm; outputting a temperature profile across a hypothetical substrate from the machine learning algorithm; and processing a substrate in the existing RTP tool using the temperature profile to anneal the substrate.
Aderhold makes obvious ([Fig. 5] Shows an irradiance graph) multiplying values on a graph ([Par 56] “To obtain the irradiance change for a given voltage differential one needs to calculate dI(r)=dI(r)/dV*deltaV.” [Examiner’s note: dIr is the irradiance and deltaV is the change in voltage])by a value at a given time during a process recipe; ([Par 23] “In accordance with one or more embodiments, certain measurable process parameters, such as temperature and certain process signals or process control signals such as heating lamp voltages can be measured and processed to provide a good indication of process uniformity, almost or close to real-time.” [Par 39] “Measured and estimated temperature of a substrate over time windows obtained during performance of a recipe may be provided.”) using the irradiation graph ([Fig. 5] Shows an irradiance graph) ([Par 24] “Embodiments of the present invention allow monitoring of uniformity of a substrate processed in an RTP chamber based on data provided by the RTP chamber (e.g., without requiring the substrate to be analyzed in a stand alone or other metrology tool such as an ellipsometer or sheet resistance measurement tool). For example, temperature data generated by an RTP chamber while a substrate is processed may be employed to generate a temperature map, such as a contour map, of the substrate during processing. Such temperature data also may be used to identify process uniformity issues, tool problems and/or faults, and the like.” [Par 33] “Based on information from the temperature controller 116 and the maglev controller 118, the temperature measurement controller 120 may determine a temperature profile such as a temperature contour map for the substrate 110 as described further below.”)
Aderhold is analogous art because it is within the field of rapid thermal processing. It would have been obvious to one of ordinary skill in the art to combine it with Huang before the effective filing date. One of ordinary skill in the art would have been motivated to make this combination in order to make the system more accurate and cheaper to operate. As noted by Aderhold, traditional methods of monitoring and controlling RTP processes can very quickly become expensive and inaccurate ([Par 2-3] “Typically, process quality assessment on a rapid thermal processing (RTP) tool consists to a large part in processing expensive, high quality monitor wafers that need to be analyzed using metrology tools like 4-point probes for sheet resistance, or ellipsometers for film thickness … To ensure production yield, sample monitoring may be performed each shift or each day…. However, the use of monitor wafers has the tendency of being either expensive or inaccurate and it assumes and relies on a perfectly stable processing situation. With more aggressive thermal processing requirements, i.e. faster ramp rates, minimal time at temperature, and higher demands for uniformity and repeatability new approaches are needed. In RTP, the variability of the thermal properties of wafers challenges the controlled performance of the system. There is an increased need to address the smaller scale variability within the wafer during thermal processing.”) To this end, Aderhold presents a method for RTP system management that is more accurately, easily, and inexpensively ([Par 24-26] “Embodiments of the present invention allow monitoring of uniformity of a substrate processed in an RTP chamber based on data provided by the RTP chamber (e.g., without requiring the substrate to be analyzed in a stand alone or other metrology tool such as an ellipsometer or sheet resistance measurement tool). For example, temperature data generated by an RTP chamber while a substrate is processed may be employed to generate a temperature map, such as a contour map, of the substrate during processing. Such temperature data also may be used to identify process uniformity issues, tool problems and/or faults, and the like. … Because existing temperature data from an RTP chamber is employed, implementation of the present invention is inexpensive and the use of monitor substrates can be largely eliminated. For example, temperature data from "dummy" substrates or production substrates may be used to determine process uniformity.”) Overall, one of ordinary skill in the art would have recognized that combining Aderhold with Huang would result in a system that easier, less expensive to use.
The combination of Huang and Aderhold does not explicitly teach the lamp array having a honeycomb shaped pattern with one or more vacancies therein; using data as an input to a machine learning algorithm; a temperature profile across a hypothetical object from the machine learning algorithm; using a temperature profile from the machine learning algorithm to anneal the substrate.
Lu makes obvious data as an input to a machine learning algorithm; ([Page 1158 Col 1 Par 7] “. To address this challenge, the input space is mapped to a feature space in the LS-SVM. For example, Fig. 3 uses a nonlinear projection function ϕ(·) to map the nonlinear input space to a higher dimensional feature space, which transforms the complex nonlinear dynamics of the input space into a simple linear behavior in the feature space that is easier to model.” [Page 1162 Col 1 Par 3] “The procedure of the proposed spatiotemporal LS-SVM modeling method, as shown in Fig. 7, is described as follows. Step 1: The experiment data of the DPS are collected. Step 2: A distributed LS-SVM model [see (22)] at each sampling time is constructed using the collected data. Step 3: All of the Lagrange multipliers from all of the distributed LS-SVM models at the whole sampling time are collected. Step 4: A model [see (25) and (26)] of the global time dynamics is built. Step 5: The spatiotemporal dynamics of the process may be reconstructed using (27).” [Examiner’s note: an SVM is a type of machine learning algorithm]) a temperature profile across a hypothetical object from the machine learning algorithm; using a temperature profile from the machine learning algorithm ([Abstract] “Here, we propose a spatiotemporal LS-SVM modeling method for complex nonlinear DPS”) [Page 1162 Col 2 Par 4] “Here, we need to design a soft sensor, also called a spatiotemporal LS-SVM model, to predict the temperature of the whole temperature field, especially at unsampled locations” [Fig. 12] Shows predicted temperatures)
PNG
media_image4.png
328
564
media_image4.png
Greyscale
Lu is analogous art because it is within the field of semiconductor processing and modeling. It would have been obvious to one of ordinary skill in the art to combine it with Huang and Aderhold before the effective filing date. One of ordinary skill in the art would have been motivated to make this combination in order to make the modelling more efficient. As noted by Lu, previous similar temperature modelling systems have been extremely complex, making real-time processing difficult, and known methods for reducing this complexity are only applicable in very specific circumstances. ([Page 1156 Col 1 Par 1 – Col 2 par 3] “Many industrial production processes are inherently distributed in space and time [1], for example, distillation, continuous stirring reaction and processes that are involved in heat exchange in the chemical industry, and the integrated circuit (IC) cure/reflow oven that is used in the electronics packaging industry. These processes are distributed parameter systems (DPSs), in which a variety of parameters, including both the input and output, can vary both temporally and spatially and are commonly described by nonlinear partial differential equations (PDEs) [2]. The complex nonlinear structure, spatiotemporal-varying nature, and limited availability of sensors during production make modeling and controlling this DPS difficult. … Typically, the firstprinciple PDE is transformed into the ordinary differential equation (ODE), considering that the controller is designed using the ODE [3]. Commonly used methods for this transformation include the finite-difference method [4] and the finite-element method [5]. However, these common methods generally result in a high-order ODE model that may not be suitable for realtime applications [6]. Other commonly used methods, such as the spectral method [7], can result in a lower order model. Despite the successful application of the aforementioned models, these models are only applicable for situations in which the PDE structure and the parameters of the DPS are fully known [1]. Oftentimes, many practical industrial distributed parameter processes are unknown”) To this end, Lu presents a system that improves on these shortcomings, resulting in a machine-learning based temperature prediction system that is less computationally complex and can be used in a wider array of situations. ([Page 1157 Col 1 Par 1] “Here, we developed a novel spatiotemporal LS-SVM method that can consider both the time dynamics and the space properties of DPS. The space kernel function describes the nonlinear correlation between space locations. The time Lagrange multiplier represents the time dynamics. The integration of the space kernel function and the time Lagrange multiplier can reconstruct the nonlinear spatiotemporal dynamics of the DPS. Different from the traditional LS-SVM modeling method, this novel spatiotemporal LS-SVM method inherently considers the space information and may represent the linear or nonlinear relationship between space locations using the space kernel function. A practical curing thermal process and its comparison to several common DPS modeling methods demonstrate the superiority of this method in the modeling of the unknown nonlinear distributed parameter process.” [Page 1164 Col 2 Par 2] “The integration of the time Lagrange multiplier and the space kernel function in the developed spatiotemporal LS-SVM method results in the accurate modeling of complex nonlinear DPS. This proposed method is also computationally less resource-intensive.”) Overall, one of ordinary skill in the art would have recognized that combining Lu with Huang and Aderhold would result in a modelling system that is significantly more accurate and computationally efficient.
The combination of Huang, Aderhold, and Lu does not explicitly teach the lamp array having a honeycomb shaped pattern with one or more vacancies therein; performing a process to anneal the substrate.
New_Ranish makes obvious the lamp array having a honeycomb shaped pattern with one or more vacancies therein; ([Par 6] “A thermal processing chamber includes a substrate support rotating about a center axis and a lamphead of plural lamps in an array having a predetermined difference in radiance pattern between them. The radiance pattern includes a variation in diffuseness or collimation. In one embodiment, the center lines of all of the lamps are disposed away from the center axis. The array can be a hexagonal array, in which the center axis is located at a predetermined position between neighboring lamps.” [Fig. 5] shows the honeycomb array shape. As can be seen there are several vacancies (i.e. locations on the lamphead without lamps) towards the corners of the pattern. [Par 52-53] “… both temperature measurements and process results indicate some radial ripple persists in the radiation pattern. Especially, the center of the wafer seems to be excessively heated and a sharp irradiance peak exists there. The ripple can be somewhat reduced by controlling the zone heating to even out the ripples. However, the ripple is still considered excessive and more fundamental solutions to the ripple problem are sought. … The schematic cross-sectional view of FIG. 7 taken along a radius of the RTP chamber illustrates the effective radial positions (±2 mm) of all lamps 36 in the hexagonal array and their radiation patterns 148. There are several radii having closely spaced lamps 36 and other radii having a gap. The maldistribution is particularly severe around the center lamp 36C. Although control of the zone heating can reduce the severity of the problem, even 15 zones appear insufficient to completely solve the ripple problem. … The simulation was repeated with the center lamp 36C removed, that is, not contributing any radiation to the total radiation pattern. The difference between a flat profile out to 50 mm and the profile without the center lamp 36C was then calculated to produce an ideal profile 156 for the center lamp 36C, as illustrated in FIGS. 8 and 10.”)
PNG
media_image1.png
631
575
media_image1.png
Greyscale
New_Ranish is analogous art because it is within the field of rapid thermal processing. It would have been obvious to one of ordinary skill in the art to combine it with Huang, Aderhold, and Lu before the effective filing date. One of ordinary skill in the art would have been motivated to make this combination in order to more uniformly process the substrate. New_Ranish notes how previous RTP systems suffer from unstable heating towards the center of the target semiconductor component ([Par 51-52] “ The radiation pattern 140 output by such an array is illustrated in FIG. 6. It includes a similar hexagonal array of bright spots 142 surrounded by a more diffuse and lower intensity background 144. However, the rotation of the wafer is intended to average out the bright spots 142 to produce a more uniform time-averaged radiation pattern. … However, both temperature measurements and process results indicate some radial ripple persists in the radiation pattern. Especially, the center of the wafer seems to be excessively heated and a sharp irradiance peak exists there. The ripple can be somewhat reduced by controlling the zone heating to even out the ripples. However, the ripple is still considered excessive and more fundamental solutions to the ripple problem are sought. We believe that a substantial cause of the ripple phenomenon in RTP is caused by two effects. First, the zone immediately surrounding the center lamp 36C does not benefit from wafer rotation since there radiation results there is no other lamp to average over. Secondly, an hexagonal array centered on the center lamp 36C and rotation axis 17 inherently produces radial oscillations, particularly near the center. The schematic cross-sectional view of FIG. 7 taken along a radius of the RTP chamber illustrates the effective radial positions (.+-.2 mm) of all lamps 36 in the hexagonal array and their radiation patterns 148. There are several radii having closely spaced lamps 36 and other radii having a gap. The maldistribution is particularly severe around the center lamp 36C. Although control of the zone heating can reduce the severity of the problem, even 15 zones appear insufficient to completely solve the ripple problem.”) To this end, New_Ranish presents a method for more uniformly processing wafers/substrates in part by removing certain lamps during the optimization process to produce an improved temperature profile ([Par 53-54] “Many simulations have been performed to quantify the geometrical effects of the finite hexagonal arrays. The radiation pattern of a standard lamp has been measured in a plane at distance from the source representative of an RTP chamber and as a function of the transverse direction (radius) from the axis of the lamp within the plane. The helical filament of a standard lamp has about eight turns extending over about 15 mm with the back of the nearest turn disposed adjacent the face of the water cooled housing or in front of it. A standard profile 152 is illustrated in the graph of FIG. 8. A computer program then calculates the total radiation intensity, as illustrated in the graph of FIG. 8 for all 409 lamps at respective positions in the array and averages this distribution for the rotation of the wafer about the center of the lamp array. The resulting total radial profile 154 of FIG. 9 exhibits a distinct peak at the center. Aside from ripple inside 50 mm, it was generally flat outwardly of the center to about 150 mm, as is desired for a 300 mm wafer. The simulation was repeated with the center lamp 36C removed, that is, not contributing any radiation to the total radiation pattern. The difference between a flat profile out to 50 mm and the profile without the center lamp 36C was then calculated to produce an ideal profile 156 for the center lamp 36C, as illustrated in FIGS. 8 and 10. … Accordingly, in one embodiment of the invention, it is desired to replace the center lamp 36C with a non-standard lamp while leaving standard lamps in the remaining sockets. In general, the center lamp 36C should produce a more diffuse pattern than the remaining lamps. This embodiment has the advantage of not requiring modification of the lamphead but only requiring modification of a replaceable lamp. The embodiment further allows all but one of the lamps to be optimized for intensity or other parameter while restricting the ripple improvement to only the center lamp 36C. It is possible to carry the lamp optimization further by separately optimizing the six lamps 36 of the innermost hexagon resulting in three different sets of lamps.”) While the removed center lamp is later replaced in some embodiments, it is clear that its removal is an essential part of the optimization process and an essential component of developing a configuration to uniformly process wafers/substrates. Overall, one of ordinary skill in the art would have recognized that combining New_Ranish with Huang, Aderhold, and Lu would result in a system capable of more uniform wafer/substrate processing, ultimately resulting in more successfully processed wafers/substrates.
The combination of Huang, Aderhold, Lu, and New_Ranish does not explicitly teach performing a process to anneal the substrate.
Miller makes obvious performing a process to anneal the substrate. ([Par 32] “Second, the substrate, will be transferred to the loadlock where undesirable reactants/gases are removed from the substrate environment, such as oxygen (O.sub.2). The loadlock pressure (P.sub.0) is matched to the enclosure pressure (P.sub.4) . Third, the substrate will be transferred to the thermal ramp chamber (140) and be ramped up to an elevated temperature in a controlled selenium vapor environment and partial pressure (P.sub.1). Fourth, the substrate will be transferred into the thermal soak chamber (150), with an independently controlled selenium vapor environment and partial pressure (P.sub.2) to optimize film growth. Fifth, the substrate will be transferred to a third thermal chamber where in during a cooling process, with an additional independently controlled selenium vapor pressure environment and partial pressure (P.sub.3), will complete the CIGS synthesis process and provide both optimized bulk and surfaces of the absorber layer. This tertiary annealing step during cool down can be termed indirect cooling. Sixth, the substrate will be transferred to the direct cool exit loadlock (170) where excess selenium vapor is removed from the substrate environment and the substrate is cooled to even lower temperatures. Lastly, the direct cool exit loadlock pressure (P.sub.5) will be controlled to match an open atmospheric environment of which the unload section (180) resides. The substrate will then be transferred to the unload section (180).”)
Miller is analogous art because it is within the field of heat treating system controls. It would have been obvious to one of ordinary skill in the art to combine it with Huang, Aderhold, Lu, and New_Ranish before the effective filing date. One of ordinary skill in the art would have been motivated to make this combination in order to make the heat treating more efficient. As noted by Miller, heat treating components can be extremely inefficient ([Par 2] “A problem with this technology has been the large thermal budget required to heat the precursors.” [Par 3] “In typical commercial batch processing selenization equipment, it takes about an hour to process substrates with CIG or CIGS precursors due slow heating and cooling rates, resulting in long processing time, which can result in high capital equipment expenditure. Furthermore, batch processing can result in poor cooling uniformity and cause substrate bowing and/or warping, reducing the product yields. In typical inline commercial selenization equipment, the footprint is very large and the operating cost is very high because of the slow heating and cooling rates required.”) To this end, Miller presents a heat treating system that allows such treating to be more cost effective and faster ([Par 33] “Furthermore, the use of elemental selenium will allow for a decrease in operational cost and capital expenditures when compared to the alternative approach of Hydrogen Selenide (H.sub.2Se) thermal processing with its associated legal permitting, hazardous material issues, and costly delivery and safety systems. The addition of an indirect cooling section will allow for a post-reaction surface treatment in situ without having to break the pressure environment or an additional post reaction surface anneal manufacturing step. The use of high speed single substrate processing allows for a faster throughput or time per substrate than conventional batch ovens.”) Overall, one of ordinary skill in the art would have recognized that combining Miller with Huang, Aderhold, Lu, and New_Ranish would result in a heat treating system that is significantly faster and more cost effective.
Claim 4. Huang teaches wherein the plurality of lamp zones includes up to 15 lamp zones.([Abstract] “In this work, a lamp configuration including five concentric lamp zones is designed to obtain uniform temperature distribution on the wafer. ”)
Claim 8. Huang teaches wherein the irradiation ([Page 4 Par 2] “The total irradiance density received by the point x (the k-th point on the given radius) is: {equation to find irradiance density at each point)”)
Aderhold makes obvious an irradiance graph; ([Fig. 5] Shows an irradiance graph)
Lu makes obvious data from at least 15 different positions ([Fig. 12] Shows the model output at a number of continuous positions along the axes v1 and v2. Also note [Fig. which shows the placement of the 16 sensors in this plane)
PNG
media_image5.png
523
1597
media_image5.png
Greyscale
Claim 9. Aderhold teaches wherein the given time during a process recipe is ([Par 23] “In accordance with one or more embodiments, certain measurable process parameters, such as temperature and certain process signals or process control signals such as heating lamp voltages can be measured and processed to provide a good indication of process uniformity, almost or close to real-time.” [Par 39] “Measured and estimated temperature of a substrate over time windows obtained during performance of a recipe may be provided.”)
Miller makes obvious processing an element during a thermal soak. ([Par 44] “An embodiment of the present disclosure can utilize data processing methods that transform signals from temperature sensors to control individual temperatures associated with sections of the apparatus. For example, an embodiment of the present disclosure can be combined with temperature sensing instrumentation to obtain state variable temperature information” [Par 32] “ Fourth, the substrate will be transferred into the thermal soak chamber (150), with an independently controlled selenium vapor environment and partial pressure (P.sub.2) to optimize film growth”)
Claim 10. Aderhold teaches wherein the given time during a process recipe is ([Par 23] “In accordance with one or more embodiments, certain measurable process parameters, such as temperature and certain process signals or process control signals such as heating lamp voltages can be measured and processed to provide a good indication of process uniformity, almost or close to real-time.” [Par 39] “Measured and estimated temperature of a substrate over time windows obtained during performance of a recipe may be provided.”)
Miller makes obvious processing an element during a thermal ramp. ([Par 44] “An embodiment of the present disclosure can utilize data processing methods that transform signals from temperature sensors to control individual temperatures associated with sections of the apparatus. For example, an embodiment of the present disclosure can be combined with temperature sensing instrumentation to obtain state variable temperature information” [Par 32] “ Third, the substrate will be transferred to the thermal ramp chamber (140) and be ramped up to an elevated temperature in a controlled selenium vapor environment and partial pressure (P.sub.1). ”)
Claims 12 and 15. The elements of claims 12 and 15 are substantially the same as those of claims 1 and 4. Therefore, the elements of claims 12 and 15 are rejected due to the same reasons as outlined above for claims 1 and 4. Further, Aderhold makes obvious the additional elements of “A non-transitory computer readable medium containing program instructions for causing a computer to perform the method comprising:” ([Par 10] “In one embodiment, the system comprises a computer memory; and a processor enabled to read instructions and data from the computer memory, to write data to the computer memory and enabled to execute the instructions”).
Claim 18. Aderhold teaches wherein the given time during a process recipe is ([Par 23] “In accordance with one or more embodiments, certain measurable process parameters, such as temperature and certain process signals or process control signals such as heating lamp voltages can be measured and processed to provide a good indication of process uniformity, almost or close to real-time.” [Par 39] “Measured and estimated temperature of a substrate over time windows obtained during performance of a recipe may be provided.”)
Miller makes obvious processing an element during a thermal soak and/or during a thermal ramp. ([Par 44] “An embodiment of the present disclosure can utilize data processing methods that transform signals from temperature sensors to control individual temperatures associated with sections of the apparatus. For example, an embodiment of the present disclosure can be combined with temperature sensing instrumentation to obtain state variable temperature information” [Par 32] “ Third, the substrate will be transferred to the thermal ramp chamber (140) and be ramped up to an elevated temperature in a controlled selenium vapor environment and partial pressure (P.sub.1). Fourth, the substrate will be transferred into the thermal soak chamber (150),”)
(2) Claims 2-3, 7, 11, and 13-14 are rejected under 35 U.S.C. 103 as being unpatentable over A Design Method to Improve Temperature Uniformity on Wafer for Rapid Thermal Processing (Hereinafter Huang) in view of Aderhold (US 20100124249 A1) in further view of A Novel Spatiotemporal LS-SVM Method for Complex Distributed Parameter Systems With Applications to Curing Thermal Process (Hereinafter Lu) as well as New_Ranish (US20060066193A1) and Miller (US 20140170805 A1) in further view of Zhang (CN 106096116 B)
Claim 2. Aderhold teaches ([Par 23] “In accordance with one or more embodiments, certain measurable process parameters, such as temperature and certain process signals or process control signals such as heating lamp voltages can be measured and processed to provide a good indication of process uniformity, almost or close to real-time.”)
Lu makes obvious ([Page 1162 Col 1 Par 3] “The procedure of the proposed spatiotemporal LS-SVM modeling method, as shown in Fig. 7, is described as follows. Step 1: The experiment data of the DPS are collected. Step 2: A distributed LS-SVM model [see (22)] at each sampling time is constructed using the collected data. Step 3: All of the Lagrange multipliers from all of the distributed LS-SVM models at the whole sampling time are collected. Step 4: A model [see (25) and (26)] of the global time dynamics is built. Step 5: The spatiotemporal dynamics of the process may be reconstructed using (27).” [Examiner’s note: an SVM is a type of machine learning algorithm])
The combination of Huang, Aderhold, Lu, New_Ranish, and Miller does not explicitly teach training a machine learning algorithm with training data
Zhang makes obvious training a machine learning algorithm with training data ([Page 7 Par 6 – Page 8 Par 1] “Preferably, as shown in FIG. 2, obtaining the measuring data of resistance, temperature rise, establishing sample data. The sample data is divided into a training set and a test set. the training set sample data comprises each measuring resistance value R and the obtaining current I, the terminal plate material M, the terminal plate size L, the terminal plate electrifying time t, the terminal plate measuring temperature rise T. the sample data of the training set is the input data of the neural network training model; the training summary and the sample rule in the training set is used for establishing the neural network model. the test set comprises each resistance value R and the obtaining current I, the terminal plate material M, the terminal plate size L, the terminal plate electrifying time t, the terminal plate measuring temperature rise T, the terminal plate test temperature rise T. the test set is used for checking the training result of the neural network training model; if the test result reaches the set training target, then manually stopping the network training.”)
Zhang is analogous art because it is in the field of heating prediction, specifically the heating of electrical elements. It would have been obvious to one of ordinary skill in the art to combine it with Huang, Aderhold, Lu, New_Ranish, and Miller before the effective filing date. One of ordinary skill in the art would have been motivated to make this combination in order to make the processing system more convenient and accurate. As noted by, Lu, the system requires a large number of measurements ([Page 7 Col 2 Par 5] “The measured data are acquired in the thermal process with a sampling interval of Δt = 10 s, and 2800 samples are collected in the experiment. The 2800 samples from heater h2 are presented in Fig. 10. The 2800 experimental data samples from the 14 sensors (s1 –s6 ,s8 –s9 , and s11 –s16 ) are used to establish the model, and the data from the remaining two sensors (s7 and s10 ) are used to evaluate the modeling performance”) One of ordinary skill in the art would have recognized that a way to make these measurements more easily would greatly improve the ability to develop such models. To this end, Zhang presents a method that makes measuring data for physical aspects of such a system easier and more accurate ([Page 6 Par 1] “the resistance value of the terminal plate is measured by testing the positive direction, the reverse voltage and the current of the sample respectively. The circuit of the resistance test of the terminal plate is shown in FIG. 4. At present, the current resistance testing method uses the thermoelectric coupling method, the circuit design is complex, and the economic cost is high. The resistance test method of the invention is featured with simple circuit diagram design, convenient operation and accurate result”) Overall, one of ordinary skill in the art would have recognized that combining Zhang with Huang, Aderhold, Lu, New_Ranish, and Miller would make physical measurements more convenient and accurate, allowing for easier and more accurate model tuning, ultimately leading to more accurate models.
Claim 3. Zhang teaches wherein the training includes at least 25 sets of different training data. ([Page 6 Par 3] “Preferably, obtaining multiple groups of sample data, can set part of the sample data is a constant, if the set current is constant, or the terminal plate size is constant, also can set the sample data as a variable. Preferably, the need of neural network model learning and training, the sample data group can be set to 10 groups to 20000 groups.”)
Claims 13-14. The elements of claims 13-14 are substantially the same as those of claims 2-3. Therefore, the elements of claims 13-14 are rejected due to the same reasons as outlined above for claims 2-3. Further, Aderhold makes obvious the additional elements of claim 12 as inherited by claims 13-14, namely: “A non-transitory computer readable medium containing program instructions for causing a computer to perform the method comprising:” ([Par 10] “In one embodiment, the system comprises a computer memory; and a processor enabled to read instructions and data from the computer memory, to write data to the computer memory and enabled to execute the instructions”).
Claim 7. Zhang teaches wherein the machine learning algorithm comprises two or more hidden layers. ([Page 7 Par 1] “under the condition of given step length, learning rate and circulating step number, respectively taking the combination of the first hidden layer node number and the second hidden layer node number to compare, finding the error of the system when the first hidden layer is (8 to 12) nodes.”)
Zhang is analogous art because it is in the field of heating prediction, specifically the heating of electrical elements. It would have been obvious to one of ordinary skill in the art to combine it with Huang, Aderhold, Lu, New_Ranish, and Miller before the effective filing date. One of ordinary skill in the art would have been motivated to make this combination in order to make the processing system more convenient and accurate. As noted by, Lu, the system requires a large number of measurements ([Page 7 Col 2 Par 5] “The measured data are acquired in the thermal process with a sampling interval of Δt = 10 s, and 2800 samples are collected in the experiment. The 2800 samples from heater h2 are presented in Fig. 10. The 2800 experimental data samples from the 14 sensors (s1 –s6 ,s8 –s9 , and s11 –s16 ) are used to establish the model, and the data from the remaining two sensors (s7 and s10 ) are used to evaluate the modeling performance”) One of ordinary skill in the art would have recognized that a way to make these measurements more easily would greatly improve the ability to develop such models. To this end, Zhang presents a method that makes measuring data for physical aspects of such a system easier and more accurate ([Page 6 Par 1] “the resistance value of the terminal plate is measured by testing the positive direction, the reverse voltage and the current of the sample respectively. The circuit of the resistance test of the terminal plate is shown in FIG. 4. At present, the current resistance testing method uses the thermoelectric coupling method, the circuit design is complex, and the economic cost is high. The resistance test method of the invention is featured with simple circuit diagram design, convenient operation and accurate result”) Overall, one of ordinary skill in the art would have recognized that combining Zhang with Huang, Aderhold, Lu, New_Ranish, and Miller would make physical measurements more convenient and accurate, allowing for easier and more accurate model tuning, ultimately leading to more accurate models.
Claim 11. Lu teaches wherein the temperature profile across the hypothetical substrate substantially matches ([Fig. 14] Shows a comparison of measured temperature data and predicted temperature data. As can be seen, the graphs match substantially)
PNG
media_image6.png
562
768
media_image6.png
Greyscale
The combination of Huang, Aderhold, Lu, New_Ranish, and Miller does not explicitly teach a set of training data
Zhang teaches a set of training data ([Page 7 Par 6 – Page 8 Par 1] “Preferably, as shown in FIG. 2, obtaining the measuring data of resistance, temperature rise, establishing sample data. The sample data is divided into a training set and a test set. the training set sample data comprises each measuring resistance value R and the obtaining current I, the terminal plate material M, the terminal plate size L, the terminal plate electrifying time t, the terminal plate measuring temperature rise T. the sample data of the training set is the input data of the neural network training model;”)
Zhang is analogous art because it is in the field of heating prediction, specifically the heating of electrical elements. It would have been obvious to one of ordinary skill in the art to combine it with Huang, Aderhold, Lu, New_Ranish, and Miller before the effective filing date. One of ordinary skill in the art would have been motivated to make this combination in order to make the processing system more convenient and accurate. As noted by, Lu, the system requires a large number of measurements ([Page 7 Col 2 Par 5] “The measured data are acquired in the thermal process with a sampling interval of Δt = 10 s, and 2800 samples are collected in the experiment. The 2800 samples from heater h2 are presented in Fig. 10. The 2800 experimental data samples from the 14 sensors (s1 –s6 ,s8 –s9 , and s11 –s16 ) are used to establish the model, and the data from the remaining two sensors (s7 and s10 ) are used to evaluate the modeling performance”) One of ordinary skill in the art would have recognized that a way to make these measurements more easily would greatly improve the ability to develop such models. To this end, Zhang presents a method that makes measuring data for physical aspects of such a system easier and more accurate ([Page 6 Par 1] “the resistance value of the terminal plate is measured by testing the positive direction, the reverse voltage and the current of the sample respectively. The circuit of the resistance test of the terminal plate is shown in FIG. 4. At present, the current resistance testing method uses the thermoelectric coupling method, the circuit design is complex, and the economic cost is high. The resistance test method of the invention is featured with simple circuit diagram design, convenient operation and accurate result”) Overall, one of ordinary skill in the art would have recognized that combining Zhang with Huang, Aderhold, Lu, New_Ranish, and Miller would make physical measurements more convenient and accurate, allowing for easier and more accurate model tuning, ultimately leading to more accurate models.
(3) Claims 5-6 and 16-17 are rejected under 35 U.S.C. 103 as being unpatentable over A Design Method to Improve Temperature Uniformity on Wafer for Rapid Thermal Processing (Hereinafter Huang) in view of Aderhold (US 20100124249 A1) in further view of A Novel Spatiotemporal LS-SVM Method for Complex Distributed Parameter Systems With Applications to Curing Thermal Process (Hereinafter Lu) as well as New_Ranish (US20060066193A1) and Miller (US 20140170805 A1) in further view of Li (CN 105140153 A)
Claim 5. Huang teaches wherein the lamp array of the lamp model ([Abstract] “ A realistic simulation of a cold wall RTP system with five lamp rings and a 200-mm wafer is performed”)
([Page 2 Par 2] “The radiation from the lamps with its central wavelength at about 900 nm heats the wafer through a selective absorption process. The temperature of wafer can be measured by thermocouples or pyrometers, and this data is used to control the output power of lamps through a feedback circuit.” [Figure 1] Shows a diagram of an existing RTP tool)
The combination of Huang, Aderhold, Lu, New_Ranish, and Miller does not explicitly teach wherein a lamp array of a first configuration is different than a lamp arrangement of a second configuration
Li makes obvious wherein a lamp array of a first configuration is different than a lamp array of a second configuration ([Page 6 Par 5-7] “is a different length and with the short side of the long side of the glass base plate 10 performs uniform heat transfer, the plurality of bulb 110 should be evenly arranged on the lamp mounting surface 101 of the heater block 100, and should satisfy the condition of minimizing the number of the plurality of bulb 110 to reduce the manufacturing cost. That is, the glass substrate 10 of the same radiation irradiation amount per a unit area, and should minimize the number of bulb. To this end, the plurality of light bulbs installed on the lamp mounting surface 101 of heater block 100 of 110 are arranged in a linear shape or a triangle. the following, firstly, the arrangement of linear shape, and then arranged in a triangle. First, the plurality of light bulbs installed in lamp mounting surface are arranged into linear shape by 101 on example. Because the glass substrate 10 has a rectangular shape, the short side and the long side have different lengths. Therefore, when the bulb 110 disposed linearly arranged along the lamp mounting surface of the short side and long side have the same interval, along the short side of the bulb 110 of the number different from the number of bulb 110” [Figs 5, 6, and 7] Show different lamp configurations with different numbers of lamps)
Li is analogous art because it is within the field of rapid thermal processing optimization. It would have been obvious to one of ordinary skill in the art to combine it with Huang, Aderhold, Lu, New_Ranish, and Miller before the effective filing date. One of ordinary skill in the art would have been motivated to make this combination in order to further ensure uniformity. As noted by Li, there is a known need for more accurate and uniform heating in RTP systems ([Page 2 Par 4] “to solve the problem of substrate temperature is not uniform, there is a need for a method for accurately measuring and controlling technology of the substrate temperature and a method for uniform heat transfer technology of the whole area to the substrate.”) To this end, Li presents a system for optimizing the heating arrangement ([Page 9 Par 1] “for the sake of reference, by processing two-dimensional heat equation the glass base plate to the simulation result, the triangle ratio has the best thermal efficiency, and can obtain a minimum number of bulb arrangement ratio. as is known, once known measuring temperature T of the glass substrate 1 in the following equation to calculate the distributed heat distribution energy S on the heat treated glass substrate.”) Overall, one of ordinary skill in the art would have recognized that combining Li with Huang, Aderhold, Lu, New_Ranish, and Miller would result in a system that in significantly more capable of performing uniform heating in the most efficient way possible.
Claim 6. Huang teaches ([Abstract] “ A realistic simulation of a cold wall RTP system with five lamp rings and a 200-mm wafer is performed”) ([Page 2 Par 2] “The radiation from the lamps with its central wavelength at about 900 nm heats the wafer through a selective absorption process. The temperature of wafer can be measured by thermocouples or pyrometers, and this data is used to control the output power of lamps through a feedback circuit.” [Figure 1] Shows a diagram of an existing RTP tool)
Li makes obvious wherein a number of lamps in the first configuration is different than a number of lamps in the lamp arrangement of the second configuration ([Page 6 Par 5-7] “is a different length and with the short side of the long side of the glass base plate 10 performs uniform heat transfer, the plurality of bulb 110 should be evenly arranged on the lamp mounting surface 101 of the heater block 100, and should satisfy the condition of minimizing the number of the plurality of bulb 110 to reduce the manufacturing cost. That is, the glass substrate 10 of the same radiation irradiation amount per a unit area, and should minimize the number of bulb. To this end, the plurality of light bulbs installed on the lamp mounting surface 101 of heater block 100 of 110 are arranged in a linear shape or a triangle. the following, firstly, the arrangement of linear shape, and then arranged in a triangle. First, the plurality of light bulbs installed in lamp mounting surface are arranged into linear shape by 101 on example. Because the glass substrate 10 has a rectangular shape, the short side and the long side have different lengths. Therefore, when the bulb 110 disposed linearly arranged along the lamp mounting surface of the short side and long side have the same interval, along the short side of the bulb 110 of the number different from the number of bulb 110” [Figs 5, 6, and 7] Show different lamp configurations with different numbers of lamps)
Claims 16-17. The elements of claims 16-17 are substantially the same as those of claims 5-6. Therefore, the elements of claims 16-17 are rejected due to the same reasons as outlined above for claims 5-6. Further, Aderhold makes obvious the additional elements of claim 12 as inherited by claims 16-17, namely: “A non-transitory computer readable medium containing program instructions for causing a computer to perform the method comprising:” ([Par 10] “In one embodiment, the system comprises a computer memory; and a processor enabled to read instructions and data from the computer memory, to write data to the computer memory and enabled to execute the instructions”).
(4) Claims 19-20 are rejected under 35 U.S.C. 103 as being unpatentable over A Design Method to Improve Temperature Uniformity on Wafer for Rapid Thermal Processing (Hereinafter Huang) in view of Aderhold (US 20100124249 A1) in further view of A Novel Spatiotemporal LS-SVM Method for Complex Distributed Parameter Systems With Applications to Curing Thermal Process (Hereinafter Lu) as well as New_Ranish (US20060066193A1) and Miller (US 20140170805 A1) in addition to Zhang (US 20140170805 A1) and Li (CN 105140153 A)
Claim 19. Huang teaches A method comprising: ([Page 2 Par 2] “The radiation from the lamps with its central wavelength at about 900 nm heats the wafer through a selective absorption process. The temperature of wafer can be measured by thermocouples or pyrometers, and this data is used to control the output power of lamps through a feedback circuit.” [Figure 1] Shows a diagram of an existing RTP tool) developing a lamp model of an RTP tool, wherein the lamp model comprises a plurality of lamp zones of a lamp array, ([Abstract] “Single-wafer rapid thermal processing (RTP) is widely used in semiconductor manufacturing. Achieving temperature uniformity on silicon wafer is a major challenge in RTP control. In this work, a lamp configuration including five concentric lamp zones is designed to obtain uniform temperature distribution on the wafer. An optics-based model is developed to determine the optimal lamp design parameters, and a uniformity criterion is proposed to evaluate the effective irradiance distribution of the tungsten–halogen lamps on the wafer. This method can be used to determine geometric parameters of the lamp array in order to achieve uniform temperature distribution on the wafer. A realistic simulation of a cold wall RTP system with five lamp rings and a 200-mm wafer is performed”) and ([Abstract] “ A realistic simulation of a cold wall RTP system with five lamp rings and a 200-mm wafer is performed”) ([Page 2 Par 2] “The radiation from the lamps with its central wavelength at about 900 nm heats the wafer through a selective absorption process. The temperature of wafer can be measured by thermocouples or pyrometers, and this data is used to control the output power of lamps through a feedback circuit.” [Figure 1] Shows a diagram of an existing RTP tool) calculating an irradiance ([Page 3 Par 5] “In the following section, the lamp number N on each zone, the radius R of each zone and the height h of the lamp array are three variables chosen to calculate the effective irradiance density absorbed by a point on the wafer.”) multiplying irradiance values of the plurality of lamp zones in the irradiance form an irradiation ([Page 4 Equations 3 and 4] Equation 4 sums the irradiance values calculated in equation 3 at each lamp in each zone, which is calculated by an equation involving multiplication by the power of an existing RTP tool (i.e. a lamp). Note that in calculating the total irradiance in Equation 4 each initial irradiance value is multiplied by term nij before being summed.) ([Page 2 Par 1] “Simulation results show that the design yields good temperature uniformity distribution on the wafer.”)
Huang does not explicitly teach training a machine learning algorithm with training data that includes real temperature data; the lamp array having a honeycomb shaped pattern with one or more vacancies therein, and wherein a number of lamps in the first configuration is different than a number of lamps in a second configuration; an irradiance graph; multiplying values on a graph by a value at a given time during a process recipe; using the irradiation graph as an input to a machine learning algorithm; and outputting a temperature profile across a hypothetical object from the machine learning algorithm; and processing a substrate in the existing RTP tool using the temperature profile to anneal the substrate.
Aderhold makes obvious ([Par 23] “In accordance with one or more embodiments, certain measurable process parameters, such as temperature and certain process signals or process control signals such as heating lamp voltages can be measured and processed to provide a good indication of process uniformity, almost or close to real-time.”) ([Fig. 5] Shows an irradiance graph) multiplying values on a graph ([Par 56] “To obtain the irradiance change for a given voltage differential one needs to calculate dI(r)=dI(r)/dV*deltaV.” [Examiner’s note: dIr is the irradiance and deltaV is the change in voltage] [Fig. 6a] Shows another irradiance graph) by a value at a given time during a process recipe; ([Par 23] “In accordance with one or more embodiments, certain measurable process parameters, such as temperature and certain process signals or process control signals such as heating lamp voltages can be measured and processed to provide a good indication of process uniformity, almost or close to real-time.” [Par 39] “Measured and estimated temperature of a substrate over time windows obtained during performance of a recipe may be provided.”) using the irradiation graph ([Fig. 5] Shows an irradiance graph) ([Par 24] “Embodiments of the present invention allow monitoring of uniformity of a substrate processed in an RTP chamber based on data provided by the RTP chamber (e.g., without requiring the substrate to be analyzed in a stand alone or other metrology tool such as an ellipsometer or sheet resistance measurement tool). For example, temperature data generated by an RTP chamber while a substrate is processed may be employed to generate a temperature map, such as a contour map, of the substrate during processing. Such temperature data also may be used to identify process uniformity issues, tool problems and/or faults, and the like.” [Par 33] “Based on information from the temperature controller 116 and the maglev controller 118, the temperature measurement controller 120 may determine a temperature profile such as a temperature contour map for the substrate 110 as described further below.”)
Aderhold is analogous art because it is within the field of rapid thermal processing. It would have been obvious to one of ordinary skill in the art to combine it with Huang before the effective filing date. One of ordinary skill in the art would have been motivated to make this combination in order to make the system more accurate and cheaper to operate. As noted by Aderhold, traditional methods of monitoring and controlling RTP processes can very quickly become expensive and inaccurate ([Par 2-3] “Typically, process quality assessment on a rapid thermal processing (RTP) tool consists to a large part in processing expensive, high quality monitor wafers that need to be analyzed using metrology tools like 4-point probes for sheet resistance, or ellipsometers for film thickness … To ensure production yield, sample monitoring may be performed each shift or each day…. However, the use of monitor wafers has the tendency of being either expensive or inaccurate and it assumes and relies on a perfectly stable processing situation. With more aggressive thermal processing requirements, i.e. faster ramp rates, minimal time at temperature, and higher demands for uniformity and repeatability new approaches are needed. In RTP, the variability of the thermal properties of wafers challenges the controlled performance of the system. There is an increased need to address the smaller scale variability within the wafer during thermal processing.”) To this end, Aderhold presents a method for RTP system management that is more accurately, easily, and inexpensively ([Par 24-26] “Embodiments of the present invention allow monitoring of uniformity of a substrate processed in an RTP chamber based on data provided by the RTP chamber (e.g., without requiring the substrate to be analyzed in a stand alone or other metrology tool such as an ellipsometer or sheet resistance measurement tool). For example, temperature data generated by an RTP chamber while a substrate is processed may be employed to generate a temperature map, such as a contour map, of the substrate during processing. Such temperature data also may be used to identify process uniformity issues, tool problems and/or faults, and the like. … Because existing temperature data from an RTP chamber is employed, implementation of the present invention is inexpensive and the use of monitor substrates can be largely eliminated. For example, temperature data from "dummy" substrates or production substrates may be used to determine process uniformity.”) Overall, one of ordinary skill in the art would have recognized that combining Aderhold with Huang would result in a system that easier, less expensive to use.
The combination of Huang and Aderhold does not explicitly teach training a machine learning algorithm with training data; the lamp array having a honeycomb shaped pattern with one or more vacancies therein and wherein a number of lamps in the first configuration is different than a number of lamps in a second configuration; using data as an input to a machine learning algorithm; a temperature profile across a hypothetical object from the machine learning algorithm; using a temperature profile from the machine learning algorithm to anneal the substrate.
Lu makes obvious ([Page 1162 Col 1 Par 3] “The procedure of the proposed spatiotemporal LS-SVM modeling method, as shown in Fig. 7, is described as follows. Step 1: The experiment data of the DPS are collected. Step 2: A distributed LS-SVM model [see (22)] at each sampling time is constructed using the collected data. Step 3: All of the Lagrange multipliers from all of the distributed LS-SVM models at the whole sampling time are collected. Step 4: A model [see (25) and (26)] of the global time dynamics is built. Step 5: The spatiotemporal dynamics of the process may be reconstructed using (27).” [Examiner’s note: an SVM is a type of machine learning algorithm]) ([Page 1158 Col 1 Par 7] “. To address this challenge, the input space is mapped to a feature space in the LS-SVM. For example, Fig. 3 uses a nonlinear projection function ϕ(·) to map the nonlinear input space to a higher dimensional feature space, which transforms the complex nonlinear dynamics of the input space into a simple linear behavior in the feature space that is easier to model.” [Page 1162 Col 1 Par 3] “The procedure of the proposed spatiotemporal LS-SVM modeling method, as shown in Fig. 7, is described as follows. Step 1: The experiment data of the DPS are collected. Step 2: A distributed LS-SVM model [see (22)] at each sampling time is constructed using the collected data. Step 3: All of the Lagrange multipliers from all of the distributed LS-SVM models at the whole sampling time are collected. Step 4: A model [see (25) and (26)] of the global time dynamics is built. Step 5: The spatiotemporal dynamics of the process may be reconstructed using (27).” [Examiner’s note: an SVM is a type of machine learning algorithm]) a temperature profile across a hypothetical object from the machine learning algorithm; using a temperature profile from the machine learning algorithm ([Abstract] “Here, we propose a spatiotemporal LS-SVM modeling method for complex nonlinear DPS”) [Page 1162 Col 2 Par 4] “Here, we need to design a soft sensor, also called a spatiotemporal LS-SVM model, to predict the temperature of the whole temperature field, especially at unsampled locations” [Fig. 12] Shows predicted temperatures)
PNG
media_image4.png
328
564
media_image4.png
Greyscale
Lu is analogous art because it is within the field of semiconductor processing and modeling. It would have been obvious to one of ordinary skill in the art to combine it with Huang and Aderhold before the effective filing date. One of ordinary skill in the art would have been motivated to make this combination in order to make the modelling more efficient. As noted by Lu, previous similar temperature modelling systems have been extremely complex, making real-time processing difficult, and known methods for reducing this complexity are only applicable in very specific circumstances. ([Page 1156 Col 1 Par 1 – Col 2 par 3] “Many industrial production processes are inherently distributed in space and time [1], for example, distillation, continuous stirring reaction and processes that are involved in heat exchange in the chemical industry, and the integrated circuit (IC) cure/reflow oven that is used in the electronics packaging industry. These processes are distributed parameter systems (DPSs), in which a variety of parameters, including both the input and output, can vary both temporally and spatially and are commonly described by nonlinear partial differential equations (PDEs) [2]. The complex nonlinear structure, spatiotemporal-varying nature, and limited availability of sensors during production make modeling and controlling this DPS difficult. … Typically, the firstprinciple PDE is transformed into the ordinary differential equation (ODE), considering that the controller is designed using the ODE [3]. Commonly used methods for this transformation include the finite-difference method [4] and the finite-element method [5]. However, these common methods generally result in a high-order ODE model that may not be suitable for realtime applications [6]. Other commonly used methods, such as the spectral method [7], can result in a lower order model. Despite the successful application of the aforementioned models, these models are only applicable for situations in which the PDE structure and the parameters of the DPS are fully known [1]. Oftentimes, many practical industrial distributed parameter processes are unknown”) To this end, Lu presents a system that improves on these shortcomings, resulting in a machine-learning based temperature prediction system that is less computationally complex and can be used in a wider array of situations. ([Page 1157 Col 1 Par 1] “Here, we developed a novel spatiotemporal LS-SVM method that can consider both the time dynamics and the space properties of DPS. The space kernel function describes the nonlinear correlation between space locations. The time Lagrange multiplier represents the time dynamics. The integration of the space kernel function and the time Lagrange multiplier can reconstruct the nonlinear spatiotemporal dynamics of the DPS. Different from the traditional LS-SVM modeling method, this novel spatiotemporal LS-SVM method inherently considers the space information and may represent the linear or nonlinear relationship between space locations using the space kernel function. A practical curing thermal process and its comparison to several common DPS modeling methods demonstrate the superiority of this method in the modeling of the unknown nonlinear distributed parameter process.” [Page 1164 Col 2 Par 2] “The integration of the time Lagrange multiplier and the space kernel function in the developed spatiotemporal LS-SVM method results in the accurate modeling of complex nonlinear DPS. This proposed method is also computationally less resource-intensive.”) Overall, one of ordinary skill in the art would have recognized that combining Lu with Huang and Aderhold would result in a modelling system that is significantly more accurate and computationally efficient.
The combination of Huang, Aderhold, and Lu does not explicitly teach training a machine learning algorithm with training data; the lamp array having a honeycomb shaped pattern with one or more vacancies therein and wherein a number of lamps in the first configuration is different than a number of lamps in a second configuration; performing a process to anneal the substrate.
New_Ranish makes obvious the lamp array having a honeycomb shaped pattern with one or more vacancies therein; ([Par 6] “A thermal processing chamber includes a substrate support rotating about a center axis and a lamphead of plural lamps in an array having a predetermined difference in radiance pattern between them. The radiance pattern includes a variation in diffuseness or collimation. In one embodiment, the center lines of all of the lamps are disposed away from the center axis. The array can be a hexagonal array, in which the center axis is located at a predetermined position between neighboring lamps.” [Fig. 5] shows the honeycomb array shape. As can be seen there are several vacancies (i.e. places on the lamphead without lamps) towards the corners of the pattern [Par 52-53] “… both temperature measurements and process results indicate some radial ripple persists in the radiation pattern. Especially, the center of the wafer seems to be excessively heated and a sharp irradiance peak exists there. The ripple can be somewhat reduced by controlling the zone heating to even out the ripples. However, the ripple is still considered excessive and more fundamental solutions to the ripple problem are sought. … The schematic cross-sectional view of FIG. 7 taken along a radius of the RTP chamber illustrates the effective radial positions (±2 mm) of all lamps 36 in the hexagonal array and their radiation patterns 148. There are several radii having closely spaced lamps 36 and other radii having a gap. The maldistribution is particularly severe around the center lamp 36C. Although control of the zone heating can reduce the severity of the problem, even 15 zones appear insufficient to completely solve the ripple problem. … The simulation was repeated with the center lamp 36C removed, that is, not contributing any radiation to the total radiation pattern. The difference between a flat profile out to 50 mm and the profile without the center lamp 36C was then calculated to produce an ideal profile 156 for the center lamp 36C, as illustrated in FIGS. 8 and 10.”)
PNG
media_image1.png
631
575
media_image1.png
Greyscale
New_Ranish is analogous art because it is within the field of rapid thermal processing. It would have been obvious to one of ordinary skill in the art to combine it with Huang, Aderhold, and Lu before the effective filing date. One of ordinary skill in the art would have been motivated to make this combination in order to more uniformly process the substrate. New_Ranish notes how previous RTP systems suffer from unstable heating towards the center of the target semiconductor component ([Par 51-52] “ The radiation pattern 140 output by such an array is illustrated in FIG. 6. It includes a similar hexagonal array of bright spots 142 surrounded by a more diffuse and lower intensity background 144. However, the rotation of the wafer is intended to average out the bright spots 142 to produce a more uniform time-averaged radiation pattern. … However, both temperature measurements and process results indicate some radial ripple persists in the radiation pattern. Especially, the center of the wafer seems to be excessively heated and a sharp irradiance peak exists there. The ripple can be somewhat reduced by controlling the zone heating to even out the ripples. However, the ripple is still considered excessive and more fundamental solutions to the ripple problem are sought. We believe that a substantial cause of the ripple phenomenon in RTP is caused by two effects. First, the zone immediately surrounding the center lamp 36C does not benefit from wafer rotation since there radiation results there is no other lamp to average over. Secondly, an hexagonal array centered on the center lamp 36C and rotation axis 17 inherently produces radial oscillations, particularly near the center. The schematic cross-sectional view of FIG. 7 taken along a radius of the RTP chamber illustrates the effective radial positions (.+-.2 mm) of all lamps 36 in the hexagonal array and their radiation patterns 148. There are several radii having closely spaced lamps 36 and other radii having a gap. The maldistribution is particularly severe around the center lamp 36C. Although control of the zone heating can reduce the severity of the problem, even 15 zones appear insufficient to completely solve the ripple problem.”) To this end, New_Ranish presents a method for more uniformly processing wafers/substrates in part by removing certain lamps during the optimization process to produce an improved temperature profile ([Par 53-54] “Many simulations have been performed to quantify the geometrical effects of the finite hexagonal arrays. The radiation pattern of a standard lamp has been measured in a plane at distance from the source representative of an RTP chamber and as a function of the transverse direction (radius) from the axis of the lamp within the plane. The helical filament of a standard lamp has about eight turns extending over about 15 mm with the back of the nearest turn disposed adjacent the face of the water cooled housing or in front of it. A standard profile 152 is illustrated in the graph of FIG. 8. A computer program then calculates the total radiation intensity, as illustrated in the graph of FIG. 8 for all 409 lamps at respective positions in the array and averages this distribution for the rotation of the wafer about the center of the lamp array. The resulting total radial profile 154 of FIG. 9 exhibits a distinct peak at the center. Aside from ripple inside 50 mm, it was generally flat outwardly of the center to about 150 mm, as is desired for a 300 mm wafer. The simulation was repeated with the center lamp 36C removed, that is, not contributing any radiation to the total radiation pattern. The difference between a flat profile out to 50 mm and the profile without the center lamp 36C was then calculated to produce an ideal profile 156 for the center lamp 36C, as illustrated in FIGS. 8 and 10. … Accordingly, in one embodiment of the invention, it is desired to replace the center lamp 36C with a non-standard lamp while leaving standard lamps in the remaining sockets. In general, the center lamp 36C should produce a more diffuse pattern than the remaining lamps. This embodiment has the advantage of not requiring modification of the lamphead but only requiring modification of a replaceable lamp. The embodiment further allows all but one of the lamps to be optimized for intensity or other parameter while restricting the ripple improvement to only the center lamp 36C. It is possible to carry the lamp optimization further by separately optimizing the six lamps 36 of the innermost hexagon resulting in three different sets of lamps.”) While the removed center lamp is later replaced in some embodiments, it is clear that its removal is an essential part of the optimization process and an essential component of developing a configuration to uniformly process wafers/substrates. Overall, one of ordinary skill in the art would have recognized that combining New_Ranish with Huang, Aderhold, and Lu would result in a system capable of more uniform wafer/substrate processing, ultimately resulting in more successfully processed wafers/substrates.
The combination of Huang, Aderhold, Lu, and New_Ranish does not explicitly teach training a machine learning algorithm with training data; wherein a number of lamps in the first configuration is different than a number of lamps in a second configuration; performing a process to anneal the substrate.
Miller makes obvious performing a process to anneal the substrate. ([Par 32] “Second, the substrate, will be transferred to the loadlock where undesirable reactants/gases are removed from the substrate environment, such as oxygen (O.sub.2). The loadlock pressure (P.sub.0) is matched to the enclosure pressure (P.sub.4) . Third, the substrate will be transferred to the thermal ramp chamber (140) and be ramped up to an elevated temperature in a controlled selenium vapor environment and partial pressure (P.sub.1). Fourth, the substrate will be transferred into the thermal soak chamber (150), with an independently controlled selenium vapor environment and partial pressure (P.sub.2) to optimize film growth. Fifth, the substrate will be transferred to a third thermal chamber where in during a cooling process, with an additional independently controlled selenium vapor pressure environment and partial pressure (P.sub.3), will complete the CIGS synthesis process and provide both optimized bulk and surfaces of the absorber layer. This tertiary annealing step during cool down can be termed indirect cooling. Sixth, the substrate will be transferred to the direct cool exit loadlock (170) where excess selenium vapor is removed from the substrate environment and the substrate is cooled to even lower temperatures. Lastly, the direct cool exit loadlock pressure (P.sub.5) will be controlled to match an open atmospheric environment of which the unload section (180) resides. The substrate will then be transferred to the unload section (180).”)
Miller is analogous art because it is within the field of heat treating system controls. It would have been obvious to one of ordinary skill in the art to combine it with Huang, Aderhold, Lu, and New_Ranish before the effective filing date. One of ordinary skill in the art would have been motivated to make this combination in order to make the heat treating more efficient. As noted by Miller, heat treating components can be extremely inefficient ([Par 2] “A problem with this technology has been the large thermal budget required to heat the precursors.” [Par 3] “In typical commercial batch processing selenization equipment, it takes about an hour to process substrates with CIG or CIGS precursors due slow heating and cooling rates, resulting in long processing time, which can result in high capital equipment expenditure. Furthermore, batch processing can result in poor cooling uniformity and cause substrate bowing and/or warping, reducing the product yields. In typical inline commercial selenization equipment, the footprint is very large and the operating cost is very high because of the slow heating and cooling rates required.”) To this end, Miller presents a heat treating system that allows such treating to be more cost effective and faster ([Par 33] “Furthermore, the use of elemental selenium will allow for a decrease in operational cost and capital expenditures when compared to the alternative approach of Hydrogen Selenide (H.sub.2Se) thermal processing with its associated legal permitting, hazardous material issues, and costly delivery and safety systems. The addition of an indirect cooling section will allow for a post-reaction surface treatment in situ without having to break the pressure environment or an additional post reaction surface anneal manufacturing step. The use of high speed single substrate processing allows for a faster throughput or time per substrate than conventional batch ovens.”) Overall, one of ordinary skill in the art would have recognized that combining Miller with Huang, Aderhold, Lu, and New_Ranish would result in a heat treating system that is significantly faster and more cost effective.
The combination of Huang, Aderhold, Lu, New_Ranish, and Miller does not explicitly teach training a machine learning algorithm with training data; wherein a number of lamps in the first configuration is different than a number of lamps in a second configuration;
Zhang makes obvious training a machine learning algorithm with training data ([Page 7 Par 6 – Page 8 Par 1] “Preferably, as shown in FIG. 2, obtaining the measuring data of resistance, temperature rise, establishing sample data. The sample data is divided into a training set and a test set. the training set sample data comprises each measuring resistance value R and the obtaining current I, the terminal plate material M, the terminal plate size L, the terminal plate electrifying time t, the terminal plate measuring temperature rise T. the sample data of the training set is the input data of the neural network training model; the training summary and the sample rule in the training set is used for establishing the neural network model. the test set comprises each resistance value R and the obtaining current I, the terminal plate material M, the terminal plate size L, the terminal plate electrifying time t, the terminal plate measuring temperature rise T, the terminal plate test temperature rise T. the test set is used for checking the training result of the neural network training model; if the test result reaches the set training target, then manually stopping the network training.”)
Zhang is analogous art because it is in the field of heating prediction, specifically the heating of electrical elements. It would have been obvious to one of ordinary skill in the art to combine it with Huang, Aderhold, Lu, New_Ranish, and Miller before the effective filing date. One of ordinary skill in the art would have been motivated to make this combination in order to make the processing system more convenient and accurate. As noted by, Lu, the system requires a large number of measurements ([Page 7 Col 2 Par 5] “The measured data are acquired in the thermal process with a sampling interval of Δt = 10 s, and 2800 samples are collected in the experiment. The 2800 samples from heater h2 are presented in Fig. 10. The 2800 experimental data samples from the 14 sensors (s1 –s6 ,s8 –s9 , and s11 –s16 ) are used to establish the model, and the data from the remaining two sensors (s7 and s10 ) are used to evaluate the modeling performance”) One of ordinary skill in the art would have recognized that a way to make these measurements more easily would greatly improve the ability to develop such models. To this end, Zhang presents a method that makes measuring data for physical aspects of such a system easier and more accurate ([Page 6 Par 1] “the resistance value of the terminal plate is measured by testing the positive direction, the reverse voltage and the current of the sample respectively. The circuit of the resistance test of the terminal plate is shown in FIG. 4. At present, the current resistance testing method uses the thermoelectric coupling method, the circuit design is complex, and the economic cost is high. The resistance test method of the invention is featured with simple circuit diagram design, convenient operation and accurate result”) Overall, one of ordinary skill in the art would have recognized that combining Zhang with Huang, Aderhold, Lu, New_Ranish, and Miller would make physical measurements more convenient and accurate, allowing for easier and more accurate model tuning, ultimately leading to more accurate models.
The combination of Huang, Aderhold, Lu, New_Ranish, Miller and Zhang does not explicitly teach wherein a number of lamps in the first configuration is different than a number of lamps in a second configuration;
Li makes obvious wherein a number of lamps in the first configuration is different than a number of lamps in a second configuration; ([Page 6 Par 5-7] “is a different length and with the short side of the long side of the glass base plate 10 performs uniform heat transfer, the plurality of bulb 110 should be evenly arranged on the lamp mounting surface 101 of the heater block 100, and should satisfy the condition of minimizing the number of the plurality of bulb 110 to reduce the manufacturing cost. That is, the glass substrate 10 of the same radiation irradiation amount per a unit area, and should minimize the number of bulb. To this end, the plurality of light bulbs installed on the lamp mounting surface 101 of heater block 100 of 110 are arranged in a linear shape or a triangle. the following, firstly, the arrangement of linear shape, and then arranged in a triangle. First, the plurality of light bulbs installed in lamp mounting surface are arranged into linear shape by 101 on example. Because the glass substrate 10 has a rectangular shape, the short side and the long side have different lengths. Therefore, when the bulb 110 disposed linearly arranged along the lamp mounting surface of the short side and long side have the same interval, along the short side of the bulb 110 of the number different from the number of bulb 110” [Figs 5, 6, and 7] Show different lamp configurations with different numbers of lamps)
Li is analogous art because it is within the field of rapid thermal processing optimization. It would have been obvious to one of ordinary skill in the art to combine it with Huang, Aderhold, Lu, New_Ranish, Miller and Zhang before the effective filing date. One of ordinary skill in the art would have been motivated to make this combination in order to further ensure uniformity. As noted by Li, there is a known need for more accurate and uniform heating in RTP systems ([Page 2 Par 4] “to solve the problem of substrate temperature is not uniform, there is a need for a method for accurately measuring and controlling technology of the substrate temperature and a method for uniform heat transfer technology of the whole area to the substrate.”) To this end, Li presents a system for optimizing the heating arrangement ([Page 9 Par 1] “for the sake of reference, by processing two-dimensional heat equation the glass base plate to the simulation result, the triangle ratio has the best thermal efficiency, and can obtain a minimum number of bulb arrangement ratio. as is known, once known measuring temperature T of the glass substrate 1 in the following equation to calculate the distributed heat distribution energy S on the heat treated glass substrate.”) Overall, one of ordinary skill in the art would have recognized that combining Li with Huang, Aderhold, Lu, New_Ranish, Miller and Zhang would result in a system that is significantly more capable of performing uniform heating in the most efficient way possible.
Claim 20. Aderhold teaches wherein the given time during a process recipe is ([Par 23] “In accordance with one or more embodiments, certain measurable process parameters, such as temperature and certain process signals or process control signals such as heating lamp voltages can be measured and processed to provide a good indication of process uniformity, almost or close to real-time.” [Par 39] “Measured and estimated temperature of a substrate over time windows obtained during performance of a recipe may be provided.”)
Miller makes obvious processing an element during a thermal soak and/or during a thermal ramp. ([Par 44] “An embodiment of the present disclosure can utilize data processing methods that transform signals from temperature sensors to control individual temperatures associated with sections of the apparatus. For example, an embodiment of the present disclosure can be combined with temperature sensing instrumentation to obtain state variable temperature information” [Par 32] “ Third, the substrate will be transferred to the thermal ramp chamber (140) and be ramped up to an elevated temperature in a controlled selenium vapor environment and partial pressure (P.sub.1). Fourth, the substrate will be transferred into the thermal soak chamber (150),”)
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Michael P Mirabito whose telephone number is (703)756-1494. The examiner can normally be reached M-F 10:30 am - 6:30 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, Emerson Puente can be reached at (571) 272-3652. 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.
/M.P.M./Examiner, Art Unit 2187
/EMERSON C PUENTE/Supervisory Patent Examiner, Art Unit 2187
Prosecution Insights