Case: 21-2037 Document: 37 Page: 1 Filed: 07/11/2022

United States Court of Appeals for the Federal Circuit ______________________

LG ELECTRONICS INC., Appellant

v.

IMMERVISION, INC., Appellee ______________________

2021-2037, 2021-2038 ______________________

Appeals from the United States Patent and Trademark Office, Patent Trial and Appeal Board in Nos. IPR2020- 00179, IPR2020-00195. ______________________

Decided: July 11, 2022 ______________________

JULIE S. GOLDEMBERG, Morgan, Lewis & Bockius LLP, Philadelphia, PA, argued for appellant. Also represented by DION MICHAEL BREGMAN, ALEXANDER STEIN, Palo Alto, CA; ANDREW V. DEVKAR, Los Angeles, CA; WILLIAM R. PETERSON, Houston, TX.

JOHN DAVID SIMMONS, Panitch Schwarze Belisario & Nadel, LLP, Wilmington, DE, argued for appellee. Also represented by DENNIS JAMES BUTLER; KEITH AARON JONES, STEPHEN EMERSON MURRAY, Philadelphia, PA. ______________________ Case: 21-2037 Document: 37 Page: 2 Filed: 07/11/2022

2 LG ELECTRONICS INC. v. IMMERVISION, INC.

Before NEWMAN, STOLL, and CUNNINGHAM, Circuit Judges. Opinion for the court filed by Circuit Judge STOLL. Opinion dissenting in part filed by Circuit Judge NEWMAN. STOLL, Circuit Judge. This appeal requires us to consider how to treat a prior art reference in which the alleged teaching of a claim ele- ment would be understood by a skilled artisan not to be an actual teaching, but rather to be an obvious error of a typo- graphical or similar nature. LG Electronics Inc. appeals from the United States Patent Trial and Appeal Board’s fi- nal written decisions in a pair of inter partes review pro- ceedings challenging claims 5 and 21 of U.S. Patent No. 6,844,990. In both proceedings, the Board found that LG had not shown the challenged claims were unpatenta- ble. Because substantial evidence supports the Board’s finding that prior art disclosure critical to both of LG’s pe- titions for inter partes review was an apparent error that would have been disregarded or corrected by a person of ordinary skill in the art, we affirm. BACKGROUND I The ’990 patent relates to capturing and displaying dig- ital panoramic images. Panoramic (e.g., super-wide angle) objective lenses typically have linear image point distribu- tion functions. This means there is a linear relationship between the distance of an image point from the image’s center and the corresponding relative angle of the object point to the image’s center. While this linearity allows dig- ital panoramic images to be easily rotated, shifted, and en- larged or shrunk, it also limits image quality to “the resolution of the image sensor used when taking the initial image.” ’990 patent col. 3 ll. 1–9. This limitation on image Case: 21-2037 Document: 37 Page: 3 Filed: 07/11/2022

LG ELECTRONICS INC. v. IMMERVISION, INC. 3

quality is most noticeable when enlarging sectors of the im- age. The ’990 patent purports to improve the resolution of particular sectors of a digital panoramic image “without the need to increase the number of pixels per unit of area of an image sensor or to provide an overlooking optical en- largement system.” Id. at col. 3 ll. 35–42. Specifically, the ’990 patent specification describes cap- turing an initial digital panoramic image using an objective lens having a non-linear image point distribution function that “expands certain zones of the image and compresses other zones of the image.” Id. at col. 3 l. 62–col. 4 l. 38. The “non-linearity of the initial image” can then be corrected to produce a final panoramic image for display. Id. at col. 4 ll. 47–53. “[T]he expanded zones of the image cover” a higher “number of pixels of the image sensor” than they would with a lens having linear image point distribution. Id. at col. 3 l. 62–col. 4 l. 10. The challenged claims specify that the lens “com- presses the center of the image and the edges of the image and expands an intermediate zone of the image located be- tween the center and the edges of the image.” Id. at col. 19 ll. 48–51. Dependent claim 5, which depends from can- celled claim 1, is representative: 1. (Cancelled) A method for capturing a digital panoramic image, by projecting a panorama onto an image sensor by means of a panoramic objective lens, the panoramic objective lens having an image point distribution function that is not linear rela- tive to the field angle of object points of the pano- rama, the distribution function having a maximum divergence of at least ±10% compared to a linear distribution function, such that the panoramic im- age obtained has at least one substantially ex- panded zone and at least one substantially compressed zone. ... Case: 21-2037 Document: 37 Page: 4 Filed: 07/11/2022

4 LG ELECTRONICS INC. v. IMMERVISION, INC.

5. The method according to claim 1, wherein the objective lens compresses the center of the image and the edges of the image and expands an interme- diate zone of the image located between the center and the edges of the image. Id. at col. 19 ll. 26–51 (claim 5) (emphasis added); see also id. at col. 20 l. 51–col. 21 l. 11 (claim 21). 1 II On November 27, 2019, LG filed two petitions for inter partes review, each challenging a dependent claim of the ’990 patent. J.A. 322–66 (IPR2020-00179 challenging claim 5); J.A. 3338–87 (IPR2020-00195 challenging claim 21). Fundamental to LG’s obviousness arguments is U.S. Patent No. 5,861,999 (“Tada”), directed to a “Super Wide Angle Lens System Using an Aspherical Lens.” 2 Tada de- scribes four embodiments that share a general system structure and differ in aspects such as lens element thick- ness, separation distance, and lens shape. Each embodi- ment satisfies a set of eight conditions relating to the aspheric characteristics of various lens elements. Tada col. 2 ll. 7–67. The embodiment relevant to this appeal, Embodiment 3, is depicted in Figure 11 and described by a prescription—or set of optical parameters—set forth in Ta- ble 5. Id. Fig. 11, Tbl. 5. Tada claims priority from Japanese Patent Application No. 09-201903, which was published as JP H10-115778 (“Japanese Priority Application”). Tada “expressly

1 Independent claims 1 and 17 were cancelled in ex parte reexamination. The claims at issue here were not subject to reexamination. 2 Tada was published with the title “Super Wide An- gel Lens System Using an Aspherical Lens”; a Certificate of Correction dated December 28, 1999, updated the title to its present form. Case: 21-2037 Document: 37 Page: 5 Filed: 07/11/2022

LG ELECTRONICS INC. v. IMMERVISION, INC. 5

incorporated” these priority applications “by reference in their entireties.” Id. at col. 3 ll. 9–13. LG argued that Tada discloses, as recited in the chal- lenged claims, a panoramic objective lens having a non-lin- ear image point distribution that compresses the center and edges of an image and expands an intermediate zone of the image between the center and the edges of the image. Tada, however, does not explicitly discuss the image point distribution functions of its lenses. Instead, LG relied on its expert Dr. Russell Chipman’s declaration for the propo- sition that Tada’s third embodiment has a distribution function producing “a compressed center and edges of the image and an expanded intermediate zone of the image be- tween the center and the edges of the image” as recited in challenged claims 5 and 21. Dr. Chipman “reconstruct[ed] the lens of Figure 11 [of Tada] using the information in Table 5 of Tada” by input- ting certain “information from Table 5 [as published] . . . into an optical design program.” J.A. 1486–87 (Chipman Decl. ¶ 46). Dr. Chipman then plotted the image point dis- tribution function for the lens system at six wavelengths and testified that the “function is not linear” in any of them. J.A. 1490–93 (Chipman Decl. ¶¶ 52–53). More specifically, Dr. Chipman explained that this embodiment of Tada’s lens system “compresses the center of the image and the edges of the image and expands an intermediate zone of the image located between the center and the edges of the im- age.” J.A. 1503 (Chipman Decl. ¶ 68). LG relied exclu- sively on Dr. Chipman’s calculations and plots using the prescription in Table 5 to show that Tada’s third embodi- ment meets the compression and expansion zone limitation of the challenged claims. LG did not rely on any other prior art reference or any other portion of Tada’s disclosure for this limitation. The Board instituted inter partes review in both pro- ceedings. The parties engaged in expert discovery, with Case: 21-2037 Document: 37 Page: 6 Filed: 07/11/2022

6 LG ELECTRONICS INC. v. IMMERVISION, INC.

ImmerVision deposing Dr. Chipman and LG deposing Im- merVision’s expert, Mr. David Aikens. In its patent owner response, ImmerVision, relying on Mr. Aikens’ declaration, argued that Tada’s Table 5 includes a readily apparent er- ror that cannot form the basis of any obviousness ground. Mr. Aikens, who was specifically tasked with verifying Dr. Chipman’s work, began by following Dr. Chipman’s process, creating a lens model from the prescription, in- cluding the aspheric coefficients—values defining the sur- face shape of an aspherical lens—in Tada’s Table 5 using an optical design program. J.A. 3031–32 (Aikens Decl. ¶ 58). From the outset, Mr. Aikens noticed that something was wrong: the physical surface of his lens model based on Tada’s Table 5 and the example lens depicted in Tada’s Fig- ure 11 did not match. J.A. 3031–32 (Aikens Decl. ¶¶ 57–59). Because of this discrepancy, Mr. Aikens com- pared the sag table—a table of heights of a lens surface with respect to the optical axis—generated for his lens model with the sag table provided in Tada’s Table 6 corre- sponding to Embodiment 3. J.A. 3032–33 (Aikens Decl. ¶ 60) (“[T]he sag table can be used as a check to make sure the equation and its coefficients are correctly understood . . . this is so commonly required that a sag table is a stand- ard output of optical design codes.”). They also did not match. J.A. 3034–35 (Aikens Decl. ¶¶ 61–62). Next, Mr. Aikens reviewed the image plane for his lens model to evaluate the magnitude of the error and discovered that the output image was distorted with “precisely the kind of uncorrected field curvature that Tada was explicitly trying to prevent.” J.A. 3035–36 (Aikens Decl. ¶¶ 63–64); see also J.A. 2423 (Aikens Dep. 132:1–10) (explaining that the model “couldn’t make a usable image . . . it was so clearly wrong, there was no point in spending more time on it”). Having established that the image was severely dis- torted, Mr. Aikens began comparing other aspects of his lens model with the “diagrams of the aberrations, astigma- tism, and distortion” provided in Tada for its third Case: 21-2037 Document: 37 Page: 7 Filed: 07/11/2022

LG ELECTRONICS INC. v. IMMERVISION, INC. 7

embodiment using “standard output features” of optical de- sign code. J.A. 3036–38 (Aikens Decl. ¶¶ 65–66). For ex- ample, Mr. Aikens compared the comatic aberration plot generated for his lens model to Tada’s Figures 15A–D (co- matic aberration plots for the model lens system using Ta- ble 5 data). J.A. 3036–38 (Aikens Decl. ¶¶ 65–67). These, too, did not match. Mr. Aikens explained that “at this point, [a person of ordinary skill in the art] would be con- vinced that there was an error in [the] model and that the error was significant.” J.A. 3039 (Aikens Decl. ¶ 68); see also J.A. 2423 (Aikens Dep. 132:18–21) (“I recognized that there had to be something wrong with the aspheric coeffi- cients. This is almost always where problems occur.”). Mr. Aikens then noticed that, as depicted in the repro- duced tables below, the aspheric coefficients from Table 3, which corresponds to Tada’s Embodiment 2, “were exactly the same as in Table 5,” which corresponds to Embodiment 3. J.A. 2425 (Aikens Dep. 134:16–21). Case: 21-2037 Document: 37 Page: 8 Filed: 07/11/2022

8 LG ELECTRONICS INC. v. IMMERVISION, INC.

Tada Tbl. 3 (annotated). Case: 21-2037 Document: 37 Page: 9 Filed: 07/11/2022

LG ELECTRONICS INC. v. IMMERVISION, INC. 9

Id. Tbl. 5 (annotated). Mr. Aikens turned next to Tada’s Table 9, which pro- vides ratios of the radius of curvature and aspherical fac- tors of Tada’s aspherical lens element to the focal length of the entire lens system. J.A. 3039–40 (Aikens Decl. ¶ 69); Tada Tbl. 9. Because the focal length of the entire lens sys- tem was defined as 1 for each embodiment, the values for conditions (2), (3), and (4) in Table 9 should have matched the aspheric coefficients A4, A6, and A8 in Table 5. But, as depicted below, they did not match: Case: 21-2037 Document: 37 Page: 10 Filed: 07/11/2022

10 LG ELECTRONICS INC. v. IMMERVISION, INC.

Tada Tbl. 5 (annotated).

Id. Tbl. 9 (annotated). Finally, Mr. Aikens reviewed Tada’s Japanese Priority Application and saw that the aspheric coefficients in its Ta- ble 5—which corresponded to the same lens embodiment as Tada’s Table 5—differed from those in Tada’s Table 5. J.A. 3041–42 (Aikens Decl. ¶¶ 72–75). The relevant por- tions of these tables are reproduced below. Case: 21-2037 Document: 37 Page: 11 Filed: 07/11/2022

LG ELECTRONICS INC. v. IMMERVISION, INC. 11

Tada Tbl. 5 (annotated).

Japanese Priority Application ¶ [0032] (Tbl. 5) (annotated). It became clear to Mr. Aikens that, after “chang[ing] the aspheric coefficients [of his model] to match” those of the Japanese Priority Application, the aspheric coefficients in the Japanese Priority Application were the correct ones and that they yielded a lens surface that was “a perfect match to the surface described in Table 6.” J.A. 3042 (Aikens Decl. ¶¶ 74–75). In other words, there was a tran- scription, or copy-and-paste, error in Tada. The disclosures in Tada’s Table 5, which were intended to correspond to its Embodiment 3, were actually identical to those in Table 3, which corresponded to Embodiment 2. Case: 21-2037 Document: 37 Page: 12 Filed: 07/11/2022

12 LG ELECTRONICS INC. v. IMMERVISION, INC.

In its final written decisions, the Board found that the “disclosure of aspheric[] coefficients in Table 5 of Tada is an obvious error” that a person of ordinary skill in the art would have recognized and corrected. LG Elecs. Inc. v. Im- merVision, Inc., No. IPR2020-00179, 2021 WL 1904645, at *11 (P.T.A.B. 2021) (Final Written Decision). 3 Continuing, the Board found that because the correct aspheric coeffi- cients in Table 5 of the Japanese Priority Application do not satisfy the language of the challenged claims, LG had not met its burden to prove the challenged claims un- patentable as obvious. Id. Although LG was free to rely on the rest of the reference, it had not done so. The Board concluded that LG did not meet its burden to prove the challenged claims would have been obvious by a preponder- ance of evidence. Id. LG appeals. We have jurisdiction under 28 U.S.C. § 1295(a)(4)(A). DISCUSSION Obviousness is a legal question based on underlying findings of fact. Univ. of Strathclyde v. Clear-Vu Lighting LLC, 17 F.4th 155, 160 (Fed. Cir. 2021). We review the Board’s ultimate obviousness determination de novo and underlying factual findings, including whether a prior art reference includes an obvious typographical or similar er- ror that would be apparent to persons of ordinary skill, for substantial evidence. “The substantial evidence standard asks ‘whether a reasonable fact finder could have arrived at the agency’s decision.’” OSI Pharms., LLC v. Apotex

3 The Board issued a nearly identical decision in the proceeding concerning claim 21. LG Elecs. v. ImmerVision, Inc., No. IPR2020-00195, 2021 WL 2486694, (P.T.A.B. 2021). For brevity, we cite only the decision in IPR2020- 00179, the proceeding concerning claim 5. Case: 21-2037 Document: 37 Page: 13 Filed: 07/11/2022

LG ELECTRONICS INC. v. IMMERVISION, INC. 13

Inc., 939 F.3d 1375, 1381–82 (Fed. Cir. 2019) (quoting In re Gartside, 203 F.3d 1305, 1312 (Fed. Cir. 2000)). It is undisputed that the aspheric coefficients in Tada’s Table 5 were erroneous. Appellant’s Br. 15; see also J.A. 2903–04 (Chipman Dep. 49:2–50:24); J.A. 3039–40 (Aiken Decl. ¶¶ 68–69). And “[t]here is no dispute that if a lens were constructed using the (correct) aspherical data from Tada’s Japanese priority application, the lens would not satisfy the [compression and expansion zone] limitation of claims 5 and 21.” Appellant’s Br. 15. Therefore, the pri- mary question before us is whether substantial evidence supports the Board’s fact finding that the error would have been apparent to a person of ordinary skill in the art such that the person would have disregarded the disclosure or corrected the error. I We begin with the legal standard. Over fifty years ago, our predecessor court reversed the decision of the Board of Patent Appeals and Interferences 4 affirming the rejection of certain claims directed to a specific compound of inhala- tion anesthetic—CF3CF2CHClBr—as obvious. In re Yale, 434 F.2d 666 (C.C.P.A. 1970). The obviousness rejection relied on the errant disclosure of this compound in an arti- cle published a few years prior. Id. at 667. That article included CF3CF2CHClBr as one of nine compounds plotted on a graph with other inhalant anesthetic compounds. Id. This was the only instance of CF3CF2CHClBr within the reference; the compound CF3CHClBr appeared throughout the rest of the article. Id. At the time, CF3CF2CHClBr was not a known compound. Id. Our predecessor court set forth the standard for evaluating these types of apparent or “ob- vious typographical error[s].” Id. at 669.

4 The Board of Patent Appeals and Interferences is the predecessor of the Patent Trial and Appeal Board. Case: 21-2037 Document: 37 Page: 14 Filed: 07/11/2022

14 LG ELECTRONICS INC. v. IMMERVISION, INC.

The Yale court explained that “any number” of several pieces of evidence “individually or cumulatively would . . . alert one of ordinary skill in the art to the existence” of the error. Yale, 434 F.2d at 669. First, the court noted the inconsistency between the reference’s figures: “CF3CF2CHClBr in Fig. 3 is the only compound listed in any figure which is not also listed in Fig. 1.” Id. at 667. Second, “[a]ll eight [compounds listed in Clements] have the identical [chemical property value] in Fig. 3 that was listed for them in Fig. 1,” with the exception of the CF3CF2CHClBr compound, which “has the [chemical prop- erty value] which was assigned in Fig. 1 to CF3CHClBr.” Id. at 669. Because CF3CF2CHClBr and CF3CHClBr are two different compounds, the court explained that it would not be “likely to have the same [chemical property value].” Id. at 667. Finally, in response to a letter from a reader, one of the authors of the article stated that the reference to CF3CF2CHClBr was “of course, an error as [the reader] suppose[d,] and CF3CF2CHClBr should read CF3CHClBr.” Id. Although the court gave less probative weight to this last piece of evidence because it “had not been sworn to,” the court found it supported the conclusion that the disclo- sure of CF3CF2CHClBr was in error. Id. at 669. The court in Yale held that where a prior art reference includes an obvious error of a typographical or similar na- ture that would be apparent to one of ordinary skill in the art who would mentally disregard the errant information as a misprint or mentally substitute it for the correct infor- mation, the errant information cannot be said to disclose subject matter. Id. at 669. The remainder of the reference would remain pertinent prior art disclosure. This standard for reviewing errors in disclosures has been undisturbed for half a century and we are bound to apply it. Deckers Corp. v. United States, 752 F.3d 949, 955–56 (Fed. Cir. 2014) (discussing stare decisis). Moreover, we view Yale’s standard as sound law, ensuring that an obviously errant disclosure of a typographical or similar nature would not Case: 21-2037 Document: 37 Page: 15 Filed: 07/11/2022

LG ELECTRONICS INC. v. IMMERVISION, INC. 15

prevent a true inventor of the claimed subject matter from later obtaining patent protection. II We now address the Board’s fact finding in this case. Based on the record before it, the Board found that the as- pheric coefficients in Tada’s Table 5 were an obvious error of a typographical or similar nature that would have been apparent to a skilled artisan. Final Written Decision, 2021 WL 1904645, at *11. As explained below, we conclude that the Board’s finding is supported by substantial evi- dence. The Board correctly identified several aspects of the disclosure in Table 5 that would alert the ordinarily skilled artisan that the disclosure was an obvious error of a typo- graphical or similar nature. First, Table 5 in Tada’s Japa- nese Priority Application has different values for the aspheric coefficients than Table 5 in Tada. J.A. 3041–42 (Aikens Decl. ¶¶ 72–75). Citing Mr. Aiken’s declaration, the Board found that the discrepancy between the coeffi- cients in Tada’s Table 5 and the Japanese Priority Applica- tion’s Table 5 was “grounded [in] a transcription error in the values for A4, A6, and A8 in Tada’s Table 5, namely, inadvertent duplication of the values for the aspherical data in Table 3.” Final Written Decision, 2021 WL 1904645, at *9. Indeed, Mr. Aikens identified the “obvious typographical error in Table 5” as an error in which the “aspheric coefficients listed in Table 5 were inadvertently copied over from Table 3, which describes Embodiment 2 of Tada.” J.A. 3030 (Aikens Decl. ¶ 56). The Board explained that the “correspondence of the Tables 1, 3, 7, and 9 be- tween the [Japanese Priority Application] and Tada itself is apparent, even prior to translation, as is the incon- sistency as to the aspherical data for Table 5.” Final Writ- ten Decision, 2021 WL 1904645, at *9. Second, the Board found that an inconsistency between Tada’s Tables 5 and 9 “ma[de] it apparent that there is an Case: 21-2037 Document: 37 Page: 16 Filed: 07/11/2022

16 LG ELECTRONICS INC. v. IMMERVISION, INC.

error in Table 5’s recitation of the aspheric[] coefficients.” Id. at *8. It was undisputed that Tada’s Tables 5 and 9 are inconsistent: the aspheric coefficients A4, A6, and A8 in Tada’s Table 5 should match the values for conditions (2)–(4) in Table 9 but do not. Id. at *7–8; see Yale, 434 F.2d at 667 (describing the internal inconsistency within a ref- erence as a signal that a person of ordinary skill “would readily recognize” as portending error). As Mr. Aikens ex- plained, and Dr. Chipman agreed, because the focal length for the entire lens system is set to 1 in each of the embodi- ments, “Table 9 rather conveniently gives you the aspheric coefficients for each of the four embodiments, and it matches correctly for [Embodiments] 1, 2[,] and 4 and is totally wrong for [Embodiment] 3.” J.A. 2427 (Aikens Dep. 136:11–15); J.A. 3039–40 (Aikens Decl. ¶ 69) (The “values [in Table 9] do not match the values in Table 5 because Ta- ble 5 is in error.”); see also J.A. 2902–04 (Chipman Dep. 48:9–50:24) (conceding that the aspheric coefficients in Ta- ble 5 match the values in Table 9 for each of the embodi- ments except for Embodiment 3). Third, the Board found that having identical aspheric coefficients in Tada’s Tables 3 and 5 “is incongruous with the differences in the values of other data for the lens sys- tems.” Final Written Decision, 2021 WL 1904645, at *8. In other words, given the other significant differences be- tween the embodiments, it was unusual for Tables 3 and 5 to list the same aspheric coefficients. Id.; J.A. 2425 (Aikens Dep. 134:4–21) (“I noticed that when I was typing in Em- bodiment 2 from Table 3, the aspheric coefficients were ex- actly the same as in Table 5, and that’s never true. That could not be right.”); see also Yale, 434 F.2d at 667 (noting the improbability of two different compounds having the same chemical property value). Considering all the evidence before it, the Board rea- sonably found that Tada’s Table 5 includes an obvious error of a typographical or similar nature that would have been apparent to one of ordinary skill in the art, who would have Case: 21-2037 Document: 37 Page: 17 Filed: 07/11/2022

LG ELECTRONICS INC. v. IMMERVISION, INC. 17

substituted it with the correct information and, thus, that Table 5 cannot be said to disclose a lens that compresses the center of the image and the edges of the image and ex- pands an intermediate zone of the image located between the center and the edges of the image. Final Written Deci- sion, 2021 WL 1904645, at *11. III LG presents two additional arguments. First, LG con- tends that Yale sets forth an “Immediately Disregard or Correct” standard that imposes a temporal urgency on the discovery of the error before the error can be considered “obvious” to a skilled artisan. See, e.g., Appellant’s Br. 4–5, 15, 23, 25, 28. Applying this reading of Yale, LG argues that Mr. Aikens’ “convoluted process” that took “ten to twelve hours” to complete clearly weighed against the ob- viousness of the error. Id. at 27–28. LG reasons that be- cause Tada has remained uncorrected in the public domain for over 20 years, LG should have been able to rely on the aspheric coefficients in Tada’s Table 5 as published. LG’s suggestion that Yale requires a person of ordinary skill in the art to immediately recognize the apparent error is incorrect. As the Board correctly explained, the length of time and the “particular manner” in which the error was actually discovered “does not diminish that there is an ob- vious error in Tada within the meaning of Yale.” Final Written Decision, 2021 WL 1904645, at *10. Contrary to LG’s assertions, Yale does not impose a temporal require- ment. Nor does LG cite any other authority requiring that the error be discovered within a specified amount of time. Certainly, the amount of time it takes a skilled artisan to detect an error may be relevant to whether an error is, in fact, an apparent error under Yale. But this is just one fac- tor for the fact finder to consider as part of the overall anal- ysis. Here, the Board considered the totality of circumstances and found that Tada’s disclosure of aspheric coefficients in Table 5 is an obvious error of a typographical Case: 21-2037 Document: 37 Page: 18 Filed: 07/11/2022

18 LG ELECTRONICS INC. v. IMMERVISION, INC.

or similar nature, notwithstanding the amount of time that preceded detection of the obvious error. For the reasons explained above, this finding is supported by substantial evidence. Second, LG suggests that Yale is limited to instances in which the error is a typographical error. Appellant’s Br. 22–23. For example, LG argues that Yale should be narrowly limited to errors such as the spelling mistake in Tada’s title upon original publication (“Super Wide Angel Lens”), which was corrected soon after (“Super Wide Angle Lens”), or in Tada’s cancelled claim 1 (“arranged in this or- der form an object side”), which was also corrected (“ar- ranged in this order from an object side”). Appellant’s Br. 30. According to LG, any other interpretation of the Yale standard would “grant[] a monopoly over a resource that was previously freely available to all, destabilizing the patent system.” Id. at 24. We disagree. While our predecessor court described the error in Yale as typographical, the error at issue here is not so far afield as to warrant a different outcome. As the Board found and Mr. Aikens, testified, the error in Tada’s Table 5 was “a transcription error . . . namely, inadvertent duplication of the values for the aspherical data in Table 3.” Final Writ- ten Decision, 2021 WL 1904645, at *9; see also J.A. 3030 (Aikens Decl. ¶ 56). The distinction between the typo- graphical error in Yale and the copy-and-paste error here is a distinction without a difference. CONCLUSION We have considered the parties’ remaining arguments and find them unpersuasive. For the foregoing reasons, we affirm the Board’s final written decisions. AFFIRMED Case: 21-2037 Document: 37 Page: 19 Filed: 07/11/2022

United States Court of Appeals for the Federal Circuit ______________________

LG ELECTRONICS INC., Appellant

v.

IMMERVISION, INC., Appellee ______________________

2021-2037, 2021-2038 ______________________

Appeals from the United States Patent and Trademark Office, Patent Trial and Appeal Board in Nos. IPR2020- 00179, IPR2020-00195. ______________________

NEWMAN, Circuit Judge, dissenting in part. The court today finds an “error of a typographical or similar nature” in the specification of the Tada reference and rules that because the error is “obvious” the erroneous portion of the Tada reference 1 is eliminated as prior art. Maj. Op. at 16–17. I cannot agree that this error is typo- graphical or similar in nature, for its existence was not dis- covered until an expert witness conducted a dozen hours of experimentation and calculation. Appx2428 (LG Elecs. Inc.

1 U.S. Patent No. 5,861,999 (“Tada” or “the ’999 Pa- tent”). Case: 21-2037 Document: 37 Page: 20 Filed: 07/11/2022

2 LG ELECTRONICS INC. v. IMMERVISION, INC.

v. ImmerVision, Inc., No. IPR2020-00179, (P.T.A.B. Oct. 1, 2020), Aikens Dep. 137:3–138:3, Ex. 1018). The appearance of a few of the same numbers in two different tables in the Tada reference provides no infor- mation as to which numbers and tables are correct and which may be in error. In contrast, a typographical or sim- ilar error is apparent to the reader and may conveniently be ignored without impeaching the content of the infor- mation. The error in the Tada reference cannot properly be deemed typographical or similar. The events that preceded the expert’s discovery of the error in the Tada reference cannot be ignored. The possibly erroneous numbers in the Tada tables were not noticed by any of the patent attorneys throughout the prosecution of Tada’s U. S. application. The now “obvious” error was not noticed by the patent examiner during a complex prosecu- tion in which claims were amended and prior art distin- guished. The purportedly “typographical or similar” error was not included in the Certificate of Correction that was ob- tained for typographical errors in the issued Tada patent. This error was not noticed by two distinct Patent Trial and Appeal Boards in instituting these two inter partes review (“IPR”) petitions, despite the technological expertise of the Board. The error in Tada Table 5 was not corrected anywhere, even after 20 years of publication. Not until an expert wit- ness conducted experiments and compared the U. S. appli- cation with the Japanese Priority document 2 did anyone discover the possibly erroneous numbers in Table 5. Appx2422–2430; Appx3030–3042.

2 JP H10-115778 (July 28, 1998). Case: 21-2037 Document: 37 Page: 21 Filed: 07/11/2022

LG ELECTRONICS INC. v. IMMERVISION, INC. 3

The specifics of what led the expert, Mr. Aikens, to dis- cover the erroneous values in Table 5 also cast doubt on whether the error may be deemed “obvious and apparent.” Mr. Aikens testified that he had fully modeled Tada’s Em- bodiment 3—relying on data from Table 5—without notic- ing the error. Appx2421–22 (Aikens Dep. 130:8–22). It was only after his model was completed that he noticed the lens created a distorted image, leading him to presume there was perhaps some error in Tada. Appx2422 (Aikens Dep. 131:3–7). At this point in his experimentation, he did not know what the error was, and certainly did not know how to correct the error; he only suspected that an error existed somewhere. Appx2423 (Aikens Dep. 132:2–10). Upon realizing there was likely an error, Mr. Aikens undertook to discover it. Id. at 132:7–10 (“I wanted to un- derstand how this lens could be so wrong and be in the pa- tent. It just didn’t make sense to me.”). Mr. Aikens testified that he required several additional hours to figure out if there actually was an error in the reference and what that error was. Id. at 132:11–13. First, Mr. Aikens observed that the physical surface shape of his Embodiment 3 lens model did not match the example lens depicted in Tada’s Figure 11. Appx2424 (Aikens Dep. 133:11–14); Appx3042 (LG Elecs. Inc. v. Im- merVision, Inc., No. IPR2020-00179, (P.T.A.B. Aug. 4, 2020), Aikens Decl. ¶ 74, Ex. 2009). This suggested that an error existed, but not where the error was or how to cor- rect it. Mr. Aikens then performed various tasks such as comparing the diagrams of the aberrations, astigmatism, and distortion for Embodiment 3 to his model, and fully modeling two other embodiments—Embodiment 1 and Em- bodiment 2. Appx2425 (Aikens Dep. 134:12–15). None of these steps showed where the purported “obvious error” was located. Only after modeling the other lens embodi- ments did Mr. Aikens finally observe that the aspherical values from Table 3, which correspond to Tada’s Case: 21-2037 Document: 37 Page: 22 Filed: 07/11/2022

4 LG ELECTRONICS INC. v. IMMERVISION, INC.

Embodiment 2, “were exactly the same as in Table 5.” Id. at 134:18–19. Mr. Aikens testified that at this point of his experimen- tation he suspected there was an error in the aspherical values in Table 5, but he had yet to determine what was in error. Id. at 134:19–21. To investigate further, Mr. Aikens compared the sag table generated from his lens model with sag Table 6 from the Tada reference. He found they did not match, indicating that an error existed; however, he still did not know what the error was, nor how to correct it. Appx2425–26 (Aikens Dep. 134:22–135:4); Appx3032–35, Appx3042 (Aikens Decl. ¶¶ 60–62, 74). Mr. Aikens then compared the values in Table 9 to Table 5 and noticed that upon performing the required calculations, the aspherical values did not match between these two tables. Appx2426– 27 (Aikens Dep. 135:9–136:15). It was here, for the first time, that Mr. Aikens testified that he could confirm there actually was an error in the Tada reference. Id. at 136:9– 10. At this juncture Mr. Aikens felt confident that Table 5 contained erroneous information, but he still did not have the information to correct it. Appx2426–27 (Aikens Dep. 135:21–136:1) (“Unfortunately, Tada didn’t include a con- straint on his A10 term so that I had to optimize to find.”). Mr. Aikens testified that he was finally able to correct and confirm the error when he obtained the Japanese Pri- ority Application. Appx2420 (Aikens Dep. 129:7–11); Appx3042 (Aikens Decl. ¶ 74). The Japanese application had the correct aspherical values in Table 5, as confirmed by a skilled expert in this technology, after many hours of corrective effort that included fully modeling three sepa- rate embodiments of the lens. In sum, the error was not of “typographical or similar nature.” The facts of this case readily distinguish it from In re Yale, 434 F.2d 666 (CCPA 1970), where our predecessor Court of Customs and Patent Appeals found that the Case: 21-2037 Document: 37 Page: 23 Filed: 07/11/2022

LG ELECTRONICS INC. v. IMMERVISION, INC. 5

inclusion of the molecule CF3CF2CHClBr in a list of anes- thetics was an obvious error. In Yale the CCPA explained that CF3CF2CHClBr was not a known compound and that the obviously intended compound was CF3CHClBr, a well- known anesthetic. This error was acknowledged by the au- thors of the article. As the panel majority recounts, “any number” of the pieces of evidence mentioned by the CCPA in Yale would “‘individually or cumulatively . . . alert one of ordinary skill in the art to the existence’ of the error.” Maj. Op. at 14 (quoting Yale, 434 F.2d at 669). However, the evidence in Yale did not require calculations or experimen- tation. Yale, 434 F.2d at 667. The same cannot be said about the error in Tada, for without the Japanese Priority Application, there is no source of the correct information. In Tada, the error in Table 5 is not discoverable unless measurements are conducted, the embodiments are recre- ated, equations are recalculated, and computations are performed. Without performing these operations, the iden- tity of a few values in both the tables does not establish error. Moreover, the tables do not suggest which table might be incorrect. As Mr. Aikens demonstrated, without modeling Embodiment 3, Table 5 cannot be compared to sag Table 6 or Figs. 11–15. In contrast, in Yale it was obviously an error to replace the known chemical anesthetic compound CF3CHClBr in Fig. 1 with the unknown chemical compound CF3CF2CHClBr in Fig. 3 and list both compounds as hav- ing the same property. The CCPA reasoned that a chemist of ordinary skill would deem it extremely unlikely that these two chemicals would have the same LogPf (partial pressure) value. Yale, 434 F.2d at 667. Although the panel majority finds analogy in the view that it is highly unlikely that the Tada embodiments would have the same aspherical values, Maj. Op. at 16, such that the listing of the same values is an obvious error, there is no intrinsic reason why two different lenses could not have Case: 21-2037 Document: 37 Page: 24 Filed: 07/11/2022

6 LG ELECTRONICS INC. v. IMMERVISION, INC.

the same aspherical values. As Mr. Aikens remarked, “[a]nd so I thought, okay, well, maybe there’s a typo on the – on the aspherics, or maybe Tada is not very good.”). Appx2425 (Aikens Dep. 134:10–12). Even the Tada patent states that “[t]he basic structure of a lens system of the third embodiment is substantially the same as that of the second embodiment. Numerical data regarding the third embodiment is shown in table 5 below.” ’999 Patent col. 8 ll. 59–64. That Table 3 and Table 5 have some of the same aspherical values does not readily alert a person of ordi- nary skill that Table 5 contains an obvious error of “typo- graphical or similar nature.” The facts in Yale are not readily analogous. An im- portant consideration in Yale was that the molecule CF3CF2CHClBr was not a known chemical compound at the time. The CCPA explained that the inclusion on a table of known anesthetics of a compound that did not exist would be recognized as an error, as was shown in corre- spondence. Yale, 434 F.2d at 668–69. I agree with the panel majority that Yale establishes the correct standard to determine if an error would be ob- vious to a person of ordinary skill in the field. However, I do not agree with the majority’s application of this stand- ard to the facts herein. An “obvious error” should be ap- parent on its face and should not require the conduct of experiments or a search for possibly conflicting information to determine whether error exists. When a reference con- tains an erroneous teaching, its value as prior art must be determined. The error in the Tada reference is plainly not a “typo- graphical or similar error,” for the error is not apparent on its face, and the correct information is not readily evident. It should not be necessary to search for a foreign document in a foreign language to determine whether there is an in- consistency in a United States patent. The foundation of Case: 21-2037 Document: 37 Page: 25 Filed: 07/11/2022

LG ELECTRONICS INC. v. IMMERVISION, INC. 7

the “typographical or similar” standard is that the error is readily recognized as an error. I am concerned that we are unsettling long-standing law, and thus I respectfully dis- sent in part.