1 2 3 4 UNITED STATES DISTRICT COURT 5 NORTHERN DISTRICT OF CALIFORNIA 6 7 INTEL CORPORATION, Case No. 3:18-cv-02848-WHO
8 Plaintiff, CLAIM CONSTRUCTION ORDER v. 9 Re: Dkt. No. 163 10 TELA INNOVATIONS, INC., Defendant. 11
12 13 Before me are six patents from the same patent family, all assigned to declaratory 14 judgment defendant Tela Innovations, Inc., and all asserted against plaintiff Intel Corporation. 15 The patented technology aims to improve the design and manufacturability of integrated circuits 16 by ameliorating difficulties associated with the lithographic gap, or the size difference between 17 ever-shrinking semiconductor features and the wavelength of light used to fabricate them. The 18 parties have asked me to construe seven terms from the asserted claims. My constructions are 19 below. 20 BACKGROUND 21 Between November 4, 2008 and January 22, 2019, the United States Patent and Trademark 22 Office (“PTO”) issued United States Patent Nos. 7,446,352 (“the ’352 Patent”), 7,943,966 (“the 23 ’966 Patent”), 7,948,012 (“the ’012 Patent”), 10,141,334 (“the ’334 Patent”), 10,141,335 (“the 24 ’335 Patent”), and 10,186,523 (“the ’523 Patent”) (collectively, the “patents in suit”). See 25 Declaration of Frank Liu (“Liu Decl.”), Exs. 1-6 [Dkt. Nos. 166-2, 166-3, 166-4, 166-5, 166-6, 26 166-7]. All of the patents in suit are part of the same patent family, all claim priority to 27 Provisional Application No. 60/781,288, filed on March 9, 2006, and all list Tela as the sole 1 Intel filed this declaratory judgment action on May 15, 2018.1 Dkt. No. 1. Since that time, 2 I have resolved a motion to transfer, several motions to dismiss and strike, a disputed motion for a 3 protective order, and several discovery disputes. See Dkt. Nos. 64, 70, 86, 162. The parties 4 briefed claim construction starting on June 13, 2019, and each submitted an electronic technology 5 tutorial in advance of the claim construction hearing. See Dkt. Nos. 163, 173. After providing the 6 parties with my tentative opinions, I heard argument on September 27, 2019. Dkt. Nos. 172, 173. 7 LEGAL STANDARD 8 Claim construction is a matter of law. See Markman v. Westview Instruments, Inc., 517 9 U.S. 370, 372 (1996); Vitronics Corp. v. Conceptronic, Inc., 90 F.3d 1576, 1582 (Fed. Cir. 1996). 10 “Generally, a claim term is given its ordinary and customary meaning—the meaning that a term 11 would have to a person of ordinary skill in the art in question at the time of the invention.” 12 Howmedica Osteonics Corp. v. Zimmer, Inc., 822 F.3d 1312, 1320 (Fed. Cir. 2016) (internal 13 quotation marks and citation omitted). In determining the proper construction of a claim, a court 14 begins with the intrinsic evidence of record, consisting of the claim language, the patent 15 specification, and, if in evidence, the prosecution history. Phillips v. AWH Corp., 415 F.3d 1303, 16 1313 (Fed. Cir. 2005); see also Vitronics, 90 F.3d at 1582. “A claim term used in multiple claims 17 should be construed consistently . . . .” Inverness Med. Switzerland GmbH v. Princeton 18 Biomeditech Corp., 309 F.3d 1365, 1371 (Fed. Cir. 2002). 19 “The appropriate starting point . . . is always with the language of the asserted claim itself.” 20 Comark Commc’ns, Inc. v. Harris Corp., 156 F.3d 1182, 1186 (Fed. Cir. 1998). “[T]he ordinary 21 and customary meaning of a claim term is the meaning that the term would have to a person of 22 ordinary skill in the art in question at the time of the invention, i.e., as of the effective filing date 23 of the patent application.” Phillips, 415 F.3d at 1312. “There are only two exceptions to this 24 general rule: 1) when a patentee sets out a definition and acts as his own lexicographer, or 2) when 25 the patentee disavows the full scope of a claim term either in the specification or during 26 prosecution.” Thorner v. Sony Computer Entm’t Am. LLC, 669 F.3d 1362, 1365 (Fed. Cir. 2012). 27 1 Such redefinition or disavowal need not be express to be clear. Trustees of Columbia Univ. in City 2 of New York v. Symantec Corp., 811 F.3d 1359, 1364 (Fed. Cir. 2016). 3 Courts read terms in the context of the claim and of the entire patent, including the 4 specification. Phillips, 415 F.3d at 1313. The specification is “the single best guide to the 5 meaning of a disputed term.” Vitronics, 90 F.3d at 1582. “The construction that stays true to the 6 claim language and most naturally aligns with the patent’s description of the invention will be, in 7 the end, the correct construction.” Renishaw PLC v. Marposs Societa’ per Azioni, 158 F.3d 1243, 8 1250 (Fed. Cir. 1998). The court may also consider the prosecution history of the patent, if in 9 evidence. Markman, 52 F.3d at 980. The prosecution history may “inform the meaning of the 10 claim language by demonstrating how the inventor understood the invention and whether the 11 inventor limited the invention in the course of prosecution, making the claim scope narrower than 12 it would otherwise be.” Phillips, 415 F.3d at 1317 (citing Vitronics, 90 F.3d at 1582-83); see also 13 Chimie v. PPG Indus., Inc., 402 F.3d 1371, 1384 (Fed. Cir. 2005) (“The purpose of consulting the 14 prosecution history in construing a claim is to exclude any interpretation that was disclaimed 15 during prosecution.”) (internal quotations omitted). 16 In most situations, analysis of the intrinsic evidence alone will resolve claim construction 17 disputes, Vitronics, 90 F.3d at 1583; however, a court can further consult “trustworthy extrinsic 18 evidence” to compare its construction to “widely held understandings in the pertinent technical 19 field,” Pitney Bowes, Inc. v. Hewlett-Packard Co., 182 F.3d 1298, 1309 (Fed. Cir. 1999). 20 Extrinsic evidence “consists of all evidence external to the patent and prosecution history, 21 including expert and inventor testimony, dictionaries, and learned treatises.” Markman, 52 F.3d at 22 980. All extrinsic evidence should be evaluated in light of the intrinsic evidence, Phillips, 415 23 F.3d at 1319, and courts should not rely on extrinsic evidence in claim construction to contradict 24 the meaning of claims discernible from examination of the claims, the written description, and the 25 prosecution history, Pitney Bowes, 182 F.3d at 1308 (citing Vitronics, 90 F.3d at 1583).
26 27 1 DISCUSSION 2 I. THE TECHNOLOGY 3 The patents at issue aim to improve the design and manufacturability of integrated circuits 4 by creating solutions to manage the lithographic gap. ’352 Patent 1:49-51. Integrated circuit 5 chips are the building blocks of devices like computers, smart phones, and tablets, and transistors 6 are the building blocks of integrated circuit chips. Today, a single integrated circuit chip includes 7 billions of transistors, which form the bottom layer of the chip, connected to the layers above by 8 metal interconnects. Transistors are effectively switches that control the flow of electrical current 9 through a circuit. 10 Transistors are made up of a substrate, a source region, a drain region, and a gate. A 11 semiconductor material forms the substrate. The source and drain regions have the same charge, 12 either positive or negative, which is created by introducing impurities during the fabrication 13 process. The transistor gate can be made of metal or polysilicon. Voltage applied to the transistor 14 gate determines whether a channel forms underneath the gate, allowing charge to flow between the 15 source and drain regions. When the opposite charge is applied to the gate, a current begins to flow 16 through the substrate between the source and drain regions (i.e., the transistor is “on”). 17 Fabrication of integrated circuits occurs one layer at a time, beginning with the bottom 18 transistor layer, known as the front end. To fabricate transistors, different materials are added, 19 altered, and removed until the desired features are present. The Asserted Patents are primarily 20 directed to one tool used during fabrication, called photolithography, or lithography. Lithography 21 is used to create a specific pattern of gates on the substrate. Once the gate material has been 22 deposited onto the substrate, a material called photoresist, which is sensitive to light, is placed on 23 top.2 A light is shone through a patterned mask, altering the chemical nature of the photoresist 24 that it reaches and creating the desired pattern. When the photoresist is developed, depending on 25 what type of photoresist was used, either the parts that were exposed to light or the parts that were 26 not exposed to light will remain. The exposed gate material, i.e. without photoresist on top, is 27 1 chemically etched away, leaving the desired gate pattern. Finally, ashing removes the remaining 2 photoresist. 3 When transistors are too close, they can electrically interfere with one another. With up to 4 billions of transistors on a single chip, they might be separated by only the space of only one one- 5 hundredth of a human hair. Despite this proximity, there are a few ways to prevent transistors 6 from interfering with one another. Dummy gates, which lack source and drain regions, can 7 separate transistors. In addition, field oxide can be used as an insulator to cover the portions of the 8 substrate that do not have active transistors, and gates can be formed on top of the field oxide. 9 At the time of the ’352 Patent, transistor feature sizes had decreased and were approaching 10 45 nm (nanometers).3 ’352 Patent 1:27-30. Because those feature sizes are smaller than the 11 wavelength of light, unintended interactions can occur between neighboring features during 12 lithography. See id. at 1:24-27. Specifically, unwanted shapes may be created (constructive 13 interference) or desired shapes may be removed (destructive interference). Id. 1:35-41. The 14 patented technology aims to create a solution “for managing lithographic gap issues as technology 15 continues to progress toward smaller semiconductor device features sizes.” Id. at 1:49-51. 16 II. CLAIM CONSTRUCTION 17 The parties agree on the construction of the following two terms: Claim Term Agreed Construction 18 “diffusion region” selected portions of the substrate within which 19 impurities have been introduced to form the source or drain of a transistor 20 “a lithography process” plain and ordinary meaning, i.e., a process by which a pattern is imprinted on a resist or 21 semiconductor wafer using light using a mask 22 Joint Claim Construction and Prehearing Statement [Dkt. No. 163] 2. The parties dispute seven 23 terms, and I construe them as follows. 24 25 3 Also at the time of the ’352 Patent, improvement in chemical mechanical polishing (CMP) 26 allowed more interconnect layers to be stacked together. Id. at 1:21-23. The topology of the different interconnect layers can limit how many layers can be stacked together because “islands, 27 ridges, and troughs can cause breaks in the interconnect lines that cross them.” Id. at 17:13-22. 1 A. “linear gate electrode segment, linear conductor segment(s), linear conductive 2 segment(s), and (interconnect) linear conductive structures” Tela’s Proposal Intel’s Proposal Court’s Ruling 3 a 3D conductive structure having a consistent vertical extending in a single 4 having a rectangular shape of cross-section shape and direction over the substrate a given width defined in a extending in a single 5 plane parallel to a top surface direction over the substrate of the substrate and defined to 6 have a length that extends in one direction 7 ʼ352: 1, 17; ʼ966: 2, 31, 33; ’012: 2, 8, 11, 13, 28 8 The parties first dispute the term “linear,” which is found in the ’352, ’966, and ’012 9 Patents. Tela argues that “linear” is to be defined and understood from the top-down view, while 10 Intel counters that it should be understood in terms of a cross-section view. Because the claims 11 themselves do not support the limitation Intel seeks to place on the term, nor does the specification 12 clearly do so, I agree with Tela’s position on the term “linear.” 13 1. The plain and ordinary meaning of the claim language 14 I begin by analyzing the language of the claims themselves. Claim 2 of the ’966 Patent 15 reads in part, “wherein the gate electrode level region includes a plurality of linear conductive 16 segments each formed to have a respective length and a respective width as measured parallel to 17 the substrate region . . . .” ’966 Patent 27:45-48. Claim 2 of the ’012 Patent reads, “wherein the 18 gate electrode level region includes a plurality of linear conductive segments each formed to have 19 a respective length and a respective width as measured parallel to the substrate region . . . .” ’012 20 Patent 33:3–6. 21 This language shows—and the parties agree—that “linear” at the very least means free of 22 bends on the x-y axis. Op’g 10; Resp. 11. Indeed, that is the plain and ordinary meaning of the 23 term: a straight line. According to Tela, this understanding is enough to construe the term 24 because “linear” is properly understood according to the x-y axis, from the top-down view. 25 Because the patents are directed to layout files, and features in a layout file are defined from a top 26 view, this term too should be understood from the top view. See Liu Decl. Ex. 7, Declaration of 27 Daniel Foty (“Foty Decl.”) [Dkt. No. 166-8] ¶ 83 (asserting that the patentee used “linear” from 1 the top view). 2 The difficulty with construing this term arises from the fact that—despite the patents’ 3 focus on the x- and y-axes rather than the z-axis—“linear features” are three-dimensional. The 4 specifications of the Asserted Patents do not describe or represent linear features only from the top 5 view, although Tela rightly points out that most figures show that perspective. See Reply 3. But 6 the patents do include and describe some three-dimensional figures. Because the claim language 7 does not expressly address the z-axis of linear-shaped features, it is necessary to review the 8 intrinsic evidence to determine whether it clearly communications anything about the z-axis. My 9 review of the intrinsic evidence is guided by this admonition from the Federal Circuit: [E]ven where a particular structure makes it ‘particularly difficult’ to 10 obtain certain benefits of the claimed invention, this does not rise to the level of disavowal of the structure. It is likewise not enough that 11 the only embodiments, or all of the embodiments, contain a particular limitation. We do not read limitations from the specification into 12 claims; we do not redefine words. Only the patentee can do that. To constitute disclaimer, there must be a clear and unmistakable 13 disclaimer. 14 Thorner v. Sony Computer Entm’t Am. LLC, 669 F.3d 1362, 1366–67 (Fed. Cir. 2012) (internal 15 citation omitted). 16 2. The intrinsic evidence 17 The specification serves as “the single best guide to the meaning of a disputed term.” See 18 Phillips, 415 F.3d at 1315. The specification of the patents at issue primarily discusses linear 19 features by reference to the x and y directions. For example, the ’352 Patent reads, “It should be 20 appreciated that the linear-shaped feature may be oriented to have its length 305 extend in either 21 the first reference direction (x), the second reference direction (y), or in diagonal direction defined 22 relative to the first and second reference directions (x) and (y). . . . Also, it should be understood 23 that the linear-shaped feature is free of bends, i.e., change in direction, in the plane defined by the 24 first and second reference directions.” ’352 Patent 9:4–9, 14–17.
25 26 27 1 One part of the specification of the ’352 Patent provides: 2 The dynamic array is defined such that layers (other than the diffusion region layer 203) are restricted with regard to layout feature shapes 3 that can be defined therein. Specifically, in each layer other than the diffusion region layer 203, only linear-shaped layout features are 4 allowed. A linear-shaped layout feature in a given layer is characterized as having a consistent vertical cross-section shape 5 and extending in a single common direction over the substrate. Thus, the linear-shaped layout features define structures that are on- 6 dimensionally varying. 7 °352 Patent 7:1—10 (emphasis added). Although Intel relies on this language to support its 8 || proposed construction, the language does not carry the dispositive weight Intel assigns to it. First, 9 || it is telling that the description specifically omits the diffusion region from the requirement that 10 || only linear-shaped features are allowed. The arrows below point to the diffusion region. 1] r-503 12 , meme mmmo FE ina 601 Fe] ee Pe Ee
bp fF TIE! □□ Ete FL Fl
2 EI le Fa Ee □□
50 BB □□□ [EBs [=| (=| fo 21 503 22 eee 23 Fig. 6 24 25 °352 Patent, Figure 6. As is clear in the image above, the diffusion region does not run ina 26 || straight line on the x-y axis; instead, it has bends and direction changes. By contrast, the darker 27 || gate electrode features are linear—they run in straight, parallel lines. 28 It is not completely clear that the consistent-cross-section description above is limited to
1 || just one embodiment, and Intel argues that following language shows that it applies broadly to all 2 || linear-shaped features: “The specific configurations and associated requirements of the linear- 3 shaped features in the various layers 207-223 are discussed further with regard to FIGS. 3-15C.” 4 *352 Patent 7:23-26. Intel argues that with this language, the specification expands the cross- 5 sectional definition to apply to all of the figures. I disagree. That language instead serves to 6 || clarify that other embodiments of linear-shaped features will have different “associated 7 || requirements”; they will not necessarily be required to have a consistent cross-sectional shape. 8 Next Intel contends that Figures 3C and 3D support its construction because those figures 9 || show three-dimensional features with consistent cross-sectional shapes. But the specification 10 || describes these figures as “exemplary linear-shaped feature[s]” that are “defined to be compatible 11 with the dynamic array, in accordance with one embodiment of the present invention.” °352 (12 || Patent 8:58-60, 9:18-20. Accordingly, the figures are limited to just one embodiment. ~ 13 301 3 14 3055 15
16 E Ca Fi Cc 617 37 é O 303 Z 18 19 aly 20 C a1 _ 21 22 23 30 \_ Fig. 3D 7, 24 || In addition, the we specification clarifies that 25 || while Figures 3C and 3D have rectangular and trapezoidal cross-sections, “it should be understood 26 || that the linear shaped features having other types of cross-sections can be defined within the 27 || dynamic array.” Id. at 9:45—49 (emphasis added). Indeed, “essentially any suitable cross- 28 sectional shape of the linear-shaped feature can be utilized so long as the linear-shaped feature
1 is defined to have a length that extends in one direction . . . .” Id. at 9: 49–52 (emphasis 2 added). I agree with Tela that this language shows that “irrespective of the cross-sectional shape, 3 the defining characteristic of a linear feature is whether its length extends in one direction in an x- 4 y plane and has consistent width (i.e., free of bends).” See Op’g 13–14. While all linear features 5 in that embodiment must have a consistent cross-sectional shape, that shape is not limited to a 6 rectangle or a trapezoid as long as it is consistent across the feature. Figures 3C and 3D are 7 accompanied by arrows that point in the x and y directions rather than the z direction, confirming 8 that the x-y plane is the focus. 9 Figures 101A, 101B, and 101C, while showing rectangular shapes, do not support Intel’s 10 construction for a two reasons. First, they are limited to a single embodiment. See ’966 Patent 11 11:28-31 (“FIG. 1 is an illustration showing a number of neighboring layout features and a 12 representation of light intensity used to render each of the layout features, in accordance with one 13 embodiment of the present invention.”). Second, they are layout features in a mask used during 14 the lithography process rather than three-dimensional conductive structures. See id. at 11:31-34 15 (“Specifically, three neighboring linear-shaped layout features (101A-101C) are depicted as being 16 disposed in a substantially parallel relationship within a given mask layer.”). These figures do not 17 support Intel’s argument that all linear features necessarily have consistent cross-sections. 18 The patents’ description of some benefits of the dynamic array, on the other hand, do seem 19 to favor Intel’s proposed construction. The dynamic array architecture aims in part to achieve 20 substantially uniform topologies in order to facilitate the stacking of more interconnect layers, to 21 improve the effectiveness of the CMP procedure, and to reduce the unpredictability of light 22 interaction during lithography. ’352 Patent 17:13-18:49; see Foty Decl. ¶¶ 75-77. As the 23 specification lays out the difficulties of these processes, widely varying topologies would prevent 24 the realization of these benefits.4 But this is not enough to support the narrowing of the claim 25 language that Intel proposes. See Thorner, 669 F.3d at 1366–67 (“[E]ven where a particular 26 4 Tela notes the difficulty of achieving a uniform topology: “A skilled artisan would understand 27 that, in such a standard CMOS process, as features traverse across the substrate, there will be 1 structure makes it ‘particularly difficult’ to obtain certain benefits of the claimed invention, this 2 does not rise to the level of disavowal of the structure.”). As Tela argued at the hearing, the 3 primary benefit of the technology is related to lithography, and those benefits are realized by 4 placing layout features in a mask in a way that makes light interactions more predictable. See ’352 5 Patent 18:32-35 (“The regular architecture implemented within the dynamic array allows the light 6 interaction unpredictability in the via lithography to be removed . . . .”). 7 The plain meaning of “linear” is a straight line. As Tela argued at the hearing, depth is 8 simply not a defining characteristic of the technology claimed in the patents. Intel’s proposed 9 construction would improperly limit otherwise broad claim language. Because Tela indicated at 10 the hearing that it was comfortable with the second half of Intel’s proposal, and because I 11 conclude that its wording would be more helpful to the jury, I adopt the following construction of 12 linear: “extending in a single direction over the substrate.” 13 B. “gate structure(s) and gate electrode feature(s)” Tela’s Proposal Intel’s Proposal Court’s Ruling 14 feature that can form a gate(s) linear-shaped feature feature comprising a gate(s) 15 of a transistor(s) defined comprising a gate(s) of a of a transistor(s) or a dummy below the gate contact transistor(s) or a dummy gate gate 16 ʼ334: 1, 2, 4, 10, 20, 22; ʼ335: 1, 2, 4, 10, 20, 22; ’523: 1, 2, 4, 10, 18, 22, 26 17 The parties next dispute the terms “gate structure” and “gate electrode feature,” which are 18 found in the ’334, ’335, and ’523 Patents. There are three main disputes here. First is the question 19 of whether to define these terms as being “linear-shaped features.” To support its proposal that the 20 structures should be construed as linear, Intel relies on specification language that describes a 21 “rectangular shape,” arguing that the specification was referring to rectangular cross-sections. 22 Resp. 14-15. I cannot agree. Not only does Intel’s argument depend on my acceptance of its 23 construction of linear—which I rejected—but the claims themselves provide no indication that 24 gate structures/ gate electrode features must be linear. It would be improper to impose this new 25 requirement during claim construction. 26 The parties next dispute how to communicate the fact that gate structures/electrode features 27 can, but may not necessarily, form the gate of a transistor. If gate structures/electrode features do 1 dummy gate. Below is Figure 5 from the ’334 Patent, which Tela excerpts and annotates in its 2 || Opening Brief. 3 S03 503 503 0 ana 503 503 503 4 2 . . © — =< ti i i i i oo 5 | | Transistor Gate “pred □□ Feature/Gate 7 | i soe! Structure 8 BOTA me tks OTA sos 409 10 wae oe 1] 405 WL =| (=) io) io) i) : 2 13 503 L593 93 L-gg3 90 “593
3 15 The parties substantially agree, but I will adopt Intel’s proposal for three reasons. First, I
16 || agree with Intel that the language “comprising” is more accurate. A single gate structure can form
17 || one or more transistor gates without its entirety being a transistor gate, as Tela’s annotations above
2 18 || acknowledge. Second, I agree with Intel’s critique that Tela’s language improperly suggests that 19 || dummy gates “can form a gate(s) of a transistor(s).” See Resp. 16. Instead, dummy gates are 20 || incapable of forming the gate of a transistor because they lack a source or drain.> In addition, even 21 within the same gate electrode feature/ gate structure, some parts are not capable of forming a 22 || transistor because they are not above diffusion regions. Third, importantly, Tela does not dispute 23 || that the term “dummy gate” is an inaccurate way to describe the portions of a gate structure/ gate 24 || electrode contact that do not form transistor gates. 25 Finally, Tela seeks a construction that would require gate electrode features / gate 26 2 . 7 > Intel further notes, “To the extent Tela’s construction means ‘[does or does not] form a gate of a 2g || transistor,’ this language is meaningless and imposes no limitation at all—everything on earth does or does not form a gate of a transistor.” Resp. 16.
1 structures to be defined below the gate contact, while Intel argues that dummy gates are generally 2 || not defined below a gate contact. I agree with Intel. Figure 6 of the ’352 Patent shows two 3 dummy gates (the dark rectangles on the right-hand side), neither of which is below a gate contact 4 || (labeled as 601). 5 5-503 6 □ (=) fo) fm) 601
5 13 Beso: | □ J aon i i) i im 14 —503 15 oe
N 16 Fig. 6
17 || To the extent that there is a contact gate, it should be defined as above the gate it contacts—but
Z 18 || because some gate structures/ gate electrode features do not have associated gate contacts, it does 19 || not make sense to include that requirement as part of their construction.® See Resp. 16 (“Thus, 20 || while it makes sense to define a gate contact as above the gate that it contacts—as Intel does in 21 Section II].D—it does not make sense to define a gate structure as below a contact that may or 22 || may not exist.”). 23 24 25 26 27 ® If Tela’s proposal defined gate structures/ gate electrode features as below the layer of gate contacts, perhaps its construction would be appropriate. But given the singular construction— 28 “the gate contact”—despite the absence of gate contacts with some gate structures/ gate electrode features, Tela’s proposal is not precise.
1 C. “gate electrode” Tela’s Proposal Intel’s Proposal Court’s Ruling 2 portion of the [linear gate a portion of the conductive a portion of the conductive 3 e Pl ae tc et nro t)d /e l is ne eg am r e cn ot n ( d’ u3 c5 t2 iv e s eh xa tep ne d i sn ot vh ee r g aa nte d l pa ay re ar l lt eh la wt ith s dh ifa fp ue s it oh na t r ee gx it oe nn d as n o dv ie s r u t sh ee d segment (’966 and ’012 a diffusion region to form a 4 Patents) / gate structure (’334 transistor gate to control the flow of and ’335 Patents) / gate electrical current between the 5 electrode feature (’523 source and drain regions of a Patent)] used to control the transistor 6 flow of electrical current between the source and drain 7 regions of a transistor ’352: 1, 16, 17; ’966: 2; ’012: 2 , 8, 11, 13; ’334: 1, 4, 22; ’335: 1, 4, 22; ’523: 1, 4, 22, 26 8 The parties next dispute the term “gate electrode,” which is found in all six patents at issue. 9 The parties dispute (i) whether to construe gate electrode as extending over and parallel with a 10 diffusion region and (ii) whether to describe its structure (as forming a transistor gate) or its 11 function (as controlling the flow of electrical current).7 12 At the hearing, Tela generally expressed agreement that the gate electrode extends over the 13 diffusion region, and language in the patents supports this understanding. See ’352 Patent at 14 Abstract (“Each linear gate electrode track . . . extends over . . . a diffusion region . . . .”); ’966 15 Patent at Abstract (providing that the “diffusion region layout shapes to be formed within a portion 16 of a substrate . . . a gate electrode level above the portion of the substrate”); id. at 8:20-29 17 (describing “gate electrode portions which extend over one or more of the [p/n]-type diffusion 18 regions to form respective [P/N]MOS transistor devices”); id. at 12:38-40 (providing that “gate 19 electrode features 207 are defined above the diffusion regions 203 to form transistor gates”); Op’g 20 14 (noting “the gate of a transistor is formed by the portion of the gate electrode feature or gate 21 structure that extends over the active portions of the substrate”). Tela raised at the hearing its 22 concern that because there must be a channel between a transistor’s source and drain region, a 23 literal understanding of Intel’s proposal would prevent a transistor from being formed on a 24 physical chip.8 With the caveat that I do not construe this term in a way that would prevent the 25 26 7 Intel argues that its proposal is substantially the same as a construction for gate electrode Tela 27 agreed to adopt in a different case. According to Tela, different patents were at issue in that case. 1 formation of a transistor, I will adopt the construction that the gate electrode extends over a 2 diffusion region.9 3 The parties next dispute whether the gate electrode should be defined as being parallel to 4 the diffusion region. Neither the claim language nor the intrinsic record supports the addition of 5 this requirement. Although Intel is right that the patent claims and specifications use the word 6 parallel, the intrinsic evidence Intel points to does not specifically address the relationship between 7 the gate electrode and the diffusion region, instead talking more generally about the substrate.10 8 See ’012 Patent 32:63-33:2 (providing, “a gate electrode level region . . . formed above and over 9 the Substrate region . . . oriented substantially parallel to the Substrate region”); ’352 Patent 7:21- 10 23 (providing, “the linear-shaped layout features in a given layer extend in a common direction 11 over the substrate and parallel with the substrate”). 12 Finally, the parties dispute the appropriateness of a construction that describes the function 13 the gate electrode performs. Intel argues that Tela is improperly inserting a functional requirement 14 into a feature that the patents define only by its structure. Resp. 19. Tela counters that a person of 15 ordinary skill in the art (“POSITA”) would understand that the gate electrode controls the flow of 16 electrical current. Reply 9-10. I agree, and the addition would aid the fact finder. For these 17 reasons, and drawing from each party’s proposal in a way that would be most helpful to the jury, I 18 will construe gate electrode as “a portion of the conductive shape in the gate layer that extends 19 over the diffusion region and is used to control the flow of electrical current between the source 20 and drain regions of a transistor.”
21 22 23 24
9 At the hearing, Intel clarified that it will not argue that complete overlap is necessary in order to 26 “extend over.” It understands “extend over” to mean in the vertical direction.
27 10 Intel’s reliance on Figure 2 of the ’966 Patent is not persuasive because it is “an illustration 1 D. “gate electrode contact, gate contact structure(s), contact structure” Tela’s Proposal Intel’s Proposal Court’s Ruling 2 a structure that passes conductive structure(s) in conductive structure(s) in a gate 3 t ch or no nu eg ch ti a on n i on fs tu hl ea t go ar tt eo enable a a ng da t se e pc ao rn at ta ec t f rl oa mye r th a eb go av te e c fro on mta c tht ela gy ae tr e a lb ayo ev re aa nn dd bs ee lp oa wra te electrode feature to the layer and below and 4 overlying metal conduction separate from interconnect and separate from interconnect lines layers layers 5 ’352: 17, 19; ’334: 1, 5, 22; ’335: 1, 5, 22; ’523: 1, 22, 26 6 The parties next dispute the term “gate contact” and “gate contact structure,” which is 7 found in the ’352, ’334,’335, and ’523 Patents. Intel argues that gate contacts “are their own 8 distinct structure in a separate layer,” while Tela argues that “some level of physical overlap” is 9 necessary between the contact and the gate and interconnects in order for an electrical connection. 10 See Op’g 19-21; Resp. 19-21. 11 The intrinsic evidence shows that the gate contact structures are a separate structure in a 12 layer that is distinct from the gate layer and the interconnect layer. First, the patents separately 13 name and describe gate structures, contact structures, and interconnects, which “strongly implies 14 that the named entities are not one and the same structure.” See HTC Corp. v. Cellular Commc’ns 15 Equip., LLC, 701 F. App’x 978, 982 (Fed. Cir. 2017). The Patents further describe the 16 relationship between the different structures, which would not be necessary if they were part of the 17 same structure. See ’334 Patent 30:63-65 (describing “contact structures positioned and sized to 18 overlap both edges of the top surface of the gate structure to which it is in physical and electrical 19 contact”), 30:31-32 (describing “a first-metal layer formed above top surfaces of the gate 20 structures”). The specification language confirms this understanding. See ’966 Patent 9:15-18 21 (“FIG. 6 is an illustration showing a gate electrode contact layer defined above and adjacent to the 22 gate electrode layer of FIG. 5, in accordance with one embodiment of the present invention.”), 23 12:40-42 (“Gate electrode contacts 209 are defined to enable connection between the gate 24 electrode features 207 and the conductor lines.”), 12:45-46 (“Interconnect layers are defined above 25 the diffusion contact 205 layer and the gate electrode contact layer 209.”), 18:34-37 (“In the gate 26 electrode contat layer, gate electrode contacts 601 are drawn to enable connection of the gate 27 electrode features 501 to the overlying metal conduction lines.”), 18:44-48 (“Also, it should be 1 appreciated that in the present invention, the gate electrode contact 601 is oversized in the 2 || direction perpendicular to the gate electrode contact features 501 to ensure overlap between the 3 gate electrode contact 601 and the gate electrode feature 501.”). 4 Figures 5, 6, and 8A from the ’966 Patent, shown here with colors added by Intel in its 5 || responsive brief, show three distinct stacked layers. Resp. 20 (showing the gate electrode in red, 6 || the contact in purple, and the metal interconnect in orange). 7 oi) (pagege 1 Toes ate 9 fe fe | ~ “ "Te | OF ee shen ip EG 10 i ET be . ee les a | i nh bh | Pet el io 11 | ¢ con 8 | “ws he a= □ 12 ~ 3 oe ul cm hears _ ris (el (ol fel Ge ler | |Soseee | □ 13 ee ee a — □ 504 504 a saa = = —— □
Q 16 || gate electrode gate contact interconnect
2 18 || Allof this evidence is sufficient to establish that the gate contact is positioned in a separate layer 19 || above the gate and below the interconnect. 20 Tela critiques Intel’s proposal by arguing that “[o|ne of ordinary skill in the art would 21 || understand that to form the contacts described in the patent would require some level of physical 22 || overlap between the contact and the gate and interconnects to enable the electrical connection 23 || between the gate and interconnect features through the contact structure.” Op’g 20-21. But Intel’s 24 || proposal does not eliminate the possibility of physical contact; indeed, claim 1 of the ’334 Patent 25 expressly provides that the gate structures “have a respective top surface in physical and electrical 26 || contact with a corresponding one of the at least six contact structures.” ’334 Patent 30:55-58; see 27 || also id. at 11:34-38 (“Gate electrode contacts 209 are defined to enable connection between the 28 || gate electrode features 207 and conductor lines. For example, the gate electrode contacts 209 are
1 defined to enable connection between transistor gates and their respective conductor nets.”), 2 17:39-43 (describing Figure 6) (“In the gate electrode contact layer, gate electrode contacts 601 3 are drawn to enable connection of the gate electrode features 501 to the underlying metal 4 conduction lines.”). The construction adopted here does not eliminate physical contact but rather 5 clarifies that the gate contact is a separate structure in a separate layer from the gate and the 6 interconnects. 7 E. “interconnect level region” Tela’s Proposal Intel’s Proposal Court’s Ruling 8 area within a layer having an area of a layer with an area within a layer having 9 m stre uta cl t ua rn ed s / wor h v ei ra e c tho en d lau yc eti rv ie s c tro an vd eu rsc et i ov ve es rt r tu hc et u sure bs s tt rh aa tt e to c tho en d lau yc et riv ie s ls otr cu ac tetu dr e as b ow vh ee tr he e located above the diffusion enable desired routing and 10 contact layer/gate electrode connectivity diffusion contact layer/gate contact layer and enables electrode contact layer and 11 definition of the desired enables definition of the circuit connectivity desired circuit connectivity 12 ’966: 31, 33, 35; ’012: 26, 28, 30 13 The parties next dispute the term “interconnect level region,” which is found in the ’966 14 and ’012 Patents. Dependent claim 31 of the ’966 Patent reads in part, “An integrated circuit 15 device as recited in claim 2, further comprising: a first interconnect level region that forms part of 16 an overall first interconnect level of the integrated circuit device.” ’966 Patent 30:45-53. The 17 parties disagree over whether the interconnect level region should be construed as located above 18 the diffusion contact layer and gate contact layer. Intel argues that Tela’s proposal improperly 19 adds a “contact layer” requirement even though the claims at issue here do not include language 20 about a contact layer—while other, unasserted claims do. Resp. 22. Tela counters that its 21 proposal is consistent with the specification, that Intel agreed that contact structures are located 22 below the interconnect level in the context of other claims, and that claim terms in the same patent 23 family should be read consistently with one another. Reply 12. 24 I agree with Tela that the construction of this term should specify the location of the 25 interconnect level region relative to other levels in order to be consistent with the specification and 26 aid the fact finder. See ’966 Patent 12:45-46 (“Interconnect layers are defined above the diffusion 27 contact 205 layer and the gate electrode contact layer 209.”). Accordingly, to the extent that the 1 various layers are described in a claim,11 the interconnect level region is defined above the 2 diffusion contact layer and the gate electrode contact layer. I will construe “interconnect level 3 region” as above the gate contacts. I also note that the parties agree that this term not cover the 4 gate electrode layer, which is a distinct claim term and thus a distinct level. See Resp. 22 (laying 5 out reasons why Tela is incorrect that Intel’s construction would improperly encompass the gate 6 electrode level). 7 Finally, Intel criticizes Tela’s narrow construction of interconnects as metal and via 8 structures, arguing that those materials are listed only in specific embodiments of the invention. I 9 agree. The claim language does not support this narrower understanding of the materials that can 10 form interconnects. For these reasons, I will adopt a modified version of Tela’s proposed 11 construction. 12 F. “[gate / metal / contact] gridline(s)” Tela’s Proposal Intel’s Proposal Court’s Ruling 13 virtual projected lines along gridline(s) along which [gate virtual projected lines along 14 w feh ati uch re [ sg aa rt ee d/m efe inta el d / contact] / a rm e e ft oa rl m / ec do ntact] structures w feh ati uch re [ sg aa rt ee p/m ose it ta iol n/ ec do ntact] 15 ’334: 1, 2, 5, 10, 18, 20, 22; ’335: 1, 2, 5, 10, 18, 20; ’523: 1, 2, 6, 8, 10, 16-19, 22, 23 16 The parties next dispute the term “grideline(s),” which is found in the ’334, ’335, and ’523 17 Patents. The parties are in substantive agreement on this term and only dispute the accuracy of 18 one another’s choice of words. It is clear from the intrinsic evidence, and the parties agree, that 19 the gate/ metal/ contact structures are positioned on a chip along virtual gridlines. See Resp. 23 20 (“Tela’s only criticism of Intel’s construction is based on the misapprehension that it requires 21 gridlines to be a ‘physical construct.’ It does not.”). Tela criticizes Intel’s use of the word 22 “formed,” arguing that the features are defined and positioned on the gridlines during the layout 23 process, not the fabrication process. Reply 13. Instead, “There is no need for the gridlines to be 24 used to form the features, because the features were positioned prior to the lithography process.” 25 Id. at 14. 26 11 Intel does not actually dispute the accuracy of this description of the layers’ relative 27 orientations. With the clarification that I recognize some claims may omit contact layers, I aim to 1 Using the word “positioned” resolves the dispute between “defined” and “formed.” 2 Indeed, as the claim language shows, the gate/ metal/ contact structures are defined along virtual 3 gridlines during the layout process in preparation for lithography. See ’335 Patent 30:5-8 4 (providing, “gate structure layout shapes used as an input to a lithography process, the gate 5 structure layout shapes positioned in accordance with a gate horizontal grid”). They are then 6 formed on a chip in positions that are also along a virtual grid as predetermined during the layout 7 process. See ’334 Patent 30:3-6 (providing, “. . . gate structures formed within a region of a 8 semiconductor chip, the gate structures positioned in accordance with a gate horizontal grid that 9 includes at least seven gate gridlines”), 30:8-12 (providing, “each gate structure in the region . . . 10 positioned to extend lengthwise in a y-direction in a substantially centered manner along an 11 associated gate gridline”). At the hearing, both parties agreed with the tentative after clarifying 12 their views.12 13 G. “physically and electrically separated by a [conductor] line end spacing of 14 minimum size” Tela’s Proposal Intel’s Proposal Court’s Ruling 15 Plain and ordinary meaning. Indefinite. Indefinite. Not indefinite. 16 ’352: 1 17 Finally, the parties dispute the term “physically and electrically separated by a [conductor] 18 line end spacing of minimum size,” which is found in claim 1 of the ’352 Patent. Intel argues that 19 the term is indefinite because the specification fails to provide an objective boundary for 20 determining “how much spacing between gate segments is permitted before it is no longer 21 considered ‘minimum.’” Resp. 23–24. Tela counters that the intrinsic evidence gives a POSITA 22 enough information to understand this term with reasonable certainty. Op’g 23-24. 23 “[A] patent is invalid for indefiniteness if its claims, read in light of the specification 24 delineating the patent, and the prosecution history, fail to inform, with reasonable certainty, those 25 skilled in the art about the scope of the invention.” Nautilus, Inc. v. Biosig Instruments, Inc., 572 26
27 12 Specifically, Intel confirmed that structures are positioned on virtual gridlines on the chip itself, 1 U.S. 898, 901 (2014). While “absolute precision is unattainable,” the claim language must be 2 “precise enough to afford clear notice of what is claimed, thereby apprising the public of what is 3 still open to them.” Id. at 901, 910. Terms of degree must have objective boundaries. Berkheimer 4 v. HP Inc., 881 F.3d 1360, 1363–64 (Fed. Cir. 2018). “Because the claims of a patent are afforded 5 a statutory presumption of validity, overcoming the presumption of validity requires that any facts 6 supporting a holding of invalidity must be proved by clear and convincing evidence.” Budde v. 7 Harley-Davidson, Inc., 250 F.3d 1369, 1376 (Fed. Cir. 2001). 8 The claim language surrounding this term reads: at least one of the linear gate electrode tracks having multiple linear 9 gate electrode segments adjacently defined thereover in an end-to-end manner such that facing ends of adjacent linear gate electrode 10 segments are physically and electrically separated by a line end spacing of minimum size, wherein a size of each line end spacing 11 within a window of lithographic influence is substantially the same, wherein the minimum size of the line end spacing corresponds to a 12 substantially full occupancy of the at least one linear gate electrode track by the multiple linear gate electrode segments . . . 13 ’352 Patent 21:23-35. The specification devotes little attention to defining “spacing of minimum 14 size.” It provides, 15 When a given gate electrode track is required to be interrupted, the separation between ends of the gate electrode track segments at the 16 point of interruption is minimized to the extent possible taking into consideration the manufacturing capability and electrical effects. . . . 17 Minimizing the separation between ends of the gate electrode track segments at the points of interruption serves to maximize the 18 lithographic reinforcement, and uniformity thereof, provided from neighboring gate electrode tracks. 19 Tela argues that these descriptions are sufficient to allow one of ordinary skill in the art to 20 understand “that the line-end spacing is placed and sized in a manner to maximize lithographic 21 reinforcement of neighboring gate electrode segments while taking into consideration electrical 22 effects.” Op’g 25. 23 I disagree. The intrinsic evidence reveals where “line end spacing” is located and why 24 “minimum size” is desirable,13 but it provides no objective way to determine what “minimum 25 size” means. Because the intrinsic evidence does not provide an objective boundary, it is 26 27 1 appropriate to consider extrinsic evidence. See Berkheimer, 881 F.3d at 1364 (noting that without 2 “objective boundary or specific examples of what constitute[d] ‘minimal’ in the claims, 3 specification, and prosecution history, the district court [had] properly considered and relied on 4 extrinsic evidence”). 5 According to Intel’s expert, by requiring consideration of both “manufacturing capability” 6 and “electrical effects,” the specification introduces two measures that are inherently subjective. 7 See Subramanian Decl. ¶¶ 92-96. In the semiconductor industry, “manufacturability involves a 8 yield tradeoff.” Id. ¶ 93; see ’352 Patent 4:50-67 (acknowledging that products are 9 “manufacturable with a specified probability”). Where there is more space between features, the 10 manufacturing yield might approach 100%, but as the space decreases, the yield will similarly 11 decrease. See Subramanian Decl. ¶ 94. As Intel points out, whether any given manufacturing 12 yield is acceptable will depend on a subjective determination based on the manufacturer’s goals— 13 “e.g., research (low yield acceptable); small-scale production (medium yield acceptable); mass 14 production (only very high yield acceptable).” Resp. 24. The same is true for electrical effects: 15 smaller spacing will increase the potential for undesirable electrical effects like interference, 16 parasitic capacitance, or cross-talk. Subramanian Decl. ¶¶ 95-96. Whether or not these effects are 17 acceptable depends on a subjective determination of the tradeoffs. The ’352 Patent provides no 18 line by which to measure “minimum space,” instead explicitly inviting these subjective 19 considerations to guide the determination of what that spacing should be. 20 Tela asserts that a POSITA would understand minimum size “to refer to the minimum 21 spacings set forth in the design rules,” which account for the lithographic concerns referred to in 22 claim 1 of the ’352 Patent. Reply 15. Because “[t]he design rules are set before the claimed 23 semiconductor device is manufactured,” according to Tela using the design rules to define the 24 spacing is not subjective. Id. This argument is unpersuasive. The design rules for any given 25 semiconductor device cannot serve as the objective bounds for determining minimum size; the 26 patent must do that. See Power Integrations, Inc. v. ON Semiconductor Corp., No. 16-CV-06371- 27 BLF, 2018 WL 5603631, at *12-13 (N.D. Cal. Oct. 26, 2018) (finding “maximum period of time” 1 the power discharge); compare id. at 12 (“Under the claim, ‘you have to choose a value, and once 2 you’ve chosen a value, you have to meet it. And if you do that, you meet the claim.’”) with Reply 3 15 (“If that semiconductor device has line-end spacing that is no larger than the minimum allowed 4 by the design rules, then it meets that claim element.”). 5 Tela argues that numerical precision is not required; the patentee intentionally chose to 6 claim “minimum” feature sizes rather than a specific value in order to account for advancements 7 that would allow for smaller feature sizes. Reply 14. While Tela may be right on both counts, the 8 law requires an objective boundary. The ’352 Patent provides none; this term is indefinite.
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 1 CONCLUSION 2 For the reasons set forth above, the disputed terms are construed as follows: 3 Court’s Ruling “linear gate electrode segment, linear extending in a single direction over the 4 conductor segment(s), linear conductive substrate segment(s), and (interconnect) linear 5 conductive structures” “gate structure(s) and gate electrode feature comprising a gate(s) of a transistor(s) 6 feature(s)” or a dummy gate 7 “gate electrode” a portion of the conductive shape in the gate layer that extends over the diffusion region 8 and is used to control the flow of electrical current between the source and drain regions 9 of a transistor 1 “gate electrode contact, gate contact conductive structure(s) in a gate contact layer 0 structure(s), contact structure” above and separate from the gate layer and 11 below and separate from interconnect layers “interconnect level region” an area within a layer having conductive a 12 structures where the layer is located above the diffusion contact layer/gate electrode contact 13 layer and enables definition of the desired 14 circuit connectivity “Tgate / metal / contact] gridline(s)” virtual projected lines along which [gate 15 /metal / contact] features are positioned “physically and electrically separated by a Indefinite 16 [conductor] line end spacing of minimum size” 17
18 IT IS SO ORDERED. 19 Dated: November 4, 2019 20 . 21 liam H. Orrick United States District Judge 23 24 25 26 27 28