Intel Corporation v. Tela Innovations, Inc.
Filing
175
CLAIM CONSTRUCTION ORDER. Signed by Judge William H. Orrick on 11/4/2019. (jmdS, COURT STAFF) (Filed on 11/4/2019)
1
2
3
4
UNITED STATES DISTRICT COURT
5
NORTHERN DISTRICT OF CALIFORNIA
6
7
INTEL CORPORATION,
Plaintiff,
8
TELA INNOVATIONS, INC.,
Re: Dkt. No. 163
Defendant.
11
United States District Court
Northern District of California
CLAIM CONSTRUCTION ORDER
v.
9
10
Case No. 3:18-cv-02848-WHO
12
Before me are six patents from the same patent family, all assigned to declaratory
13
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
21
BACKGROUND
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
28
assignee.
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
Claim construction is a matter of law. See Markman v. Westview Instruments, Inc., 517
8
9
U.S. 370, 372 (1996); Vitronics Corp. v. Conceptronic, Inc., 90 F.3d 1576, 1582 (Fed. Cir. 1996).
“Generally, a claim term is given its ordinary and customary meaning—the meaning that a term
11
United States District Court
Northern District of California
10
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).
“The appropriate starting point . . . is always with the language of the asserted claim itself.”
19
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
28
1
The patents at issue in the case have evolved since it was initiated.
2
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
claim language by demonstrating how the inventor understood the invention and whether the
11
United States District Court
Northern District of California
10
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
28
3
1
2
DISCUSSION
I.
THE TECHNOLOGY
The patents at issue aim to improve the design and manufacturability of integrated circuits
3
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.
Transistors are made up of a substrate, a source region, a drain region, and a gate. A
10
United States District Court
Northern District of California
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
28
2
This account given in this Order does not detail every step involved in semiconductor
fabrication.
4
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.
At the time of the ’352 Patent, transistor feature sizes had decreased and were approaching
9
45 nm (nanometers).3 ’352 Patent 1:27-30. Because those feature sizes are smaller than the
11
United States District Court
Northern District of California
10
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
21
The parties agree on the construction of the following two terms:
Claim Term
Agreed Construction
“diffusion region”
selected portions of the substrate within which
impurities have been introduced to form the
source or drain of a transistor
“a lithography process”
plain and ordinary meaning, i.e., a process by
which a pattern is imprinted on a resist or
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.
17
18
19
20
24
25
26
27
28
3
Also at the time of the ’352 Patent, improvement in chemical mechanical polishing (CMP)
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,
ridges, and troughs can cause breaks in the interconnect lines that cross them.” Id. at 17:13-22.
CMP can help flatten the surface of the semiconductor wafer to facilitate stacking. Id. at 17:2328.
5
1
2
3
4
5
6
7
8
9
A. “linear gate electrode segment, linear conductor segment(s), linear conductive
segment(s), and (interconnect) linear conductive structures”
Tela’s Proposal
Intel’s Proposal
Court’s Ruling
a 3D conductive structure
having a consistent vertical
extending in a single
having a rectangular shape of cross-section shape and
direction over the substrate
a given width defined in a
extending in a single
plane parallel to a top surface direction over the substrate
of the substrate and defined to
have a length that extends in
one direction
ʼ352: 1, 17; ʼ966: 2, 31, 33; ’012: 2, 8, 11, 13, 28
The parties first dispute the term “linear,” which is found in the ’352, ’966, and ’012
Patents. Tela argues that “linear” is to be defined and understood from the top-down view, while
Intel counters that it should be understood in terms of a cross-section view. Because the claims
11
United States District Court
Northern District of California
10
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
14
1.
The plain and ordinary meaning of the claim language
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
28
6
1
2
the top view).
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
13
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
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
the only embodiments, or all of the embodiments, contain a particular
limitation. We do not read limitations from the specification into
claims; we do not redefine words. Only the patentee can do that. To
constitute disclaimer, there must be a clear and unmistakable
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.
10
United States District Court
Northern District of California
11
12
17
The intrinsic evidence
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
28
7
1
2
3
4
5
6
One part of the specification of the ’352 Patent provides:
The dynamic array is defined such that layers (other than the diffusion
region layer 203) are restricted with regard to layout feature shapes
that can be defined therein. Specifically, in each layer other than the
diffusion region layer 203, only linear-shaped layout features are
allowed. A linear-shaped layout feature in a given layer is
characterized as having a consistent vertical cross-section shape
and extending in a single common direction over the substrate.
Thus, the linear-shaped layout features define structures that are ondimensionally 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.
United States District Court
Northern District of California
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
’352 Patent, Figure 6. As is clear in the image above, the diffusion region does not run in a
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
8
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
9
Next Intel contends that Figures 3C and 3D support its construction because those figures
show three-dimensional features with consistent cross-sectional shapes. But the specification
describes these figures as “exemplary linear-shaped feature[s]” that are “defined to be compatible
11
United States District Court
Northern District of California
10
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
14
15
16
17
18
19
20
21
22
23
24
In addition, the
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
9
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.
Figures 101A, 101B, and 101C, while showing rectangular shapes, do not support Intel’s
9
construction for a two reasons. First, they are limited to a single embodiment. See ’966 Patent
11
United States District Court
Northern District of California
10
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.
The patents’ description of some benefits of the dynamic array, on the other hand, do seem
18
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
27
28
4
Tela notes the difficulty of achieving a uniform topology: “A skilled artisan would understand
that, in such a standard CMOS process, as features traverse across the substrate, there will be
variations in the cross-sectional shape, particularly when traversing across shallow trench isolation
regions.” Reply 4.
10
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
the hearing that it was comfortable with the second half of Intel’s proposal, and because I
11
United States District Court
Northern District of California
10
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
14
15
16
17
B. “gate structure(s) and gate electrode feature(s)”
Tela’s Proposal
Intel’s Proposal
Court’s Ruling
feature that can form a gate(s) linear-shaped feature
feature comprising a gate(s)
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
ʼ334: 1, 2, 4, 10, 20, 22; ʼ335: 1, 2, 4, 10, 20, 22; ’523: 1, 2, 4, 10, 18, 22, 26
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
28
not cross an active portion of the substrate, no transistor is formed; in other words, the gate is a
11
1
dummy gate. Below is Figure 5 from the ’334 Patent, which Tela excerpts and annotates in its
2
Opening Brief.
3
4
5
6
7
8
9
10
United States District Court
Northern District of California
11
12
13
14
The parties substantially agree, but I will adopt Intel’s proposal for three reasons. First, I
15
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
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.5 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.
Finally, Tela seeks a construction that would require gate electrode features / gate
25
26
27
28
5
Intel further notes, “To the extent Tela’s construction means ‘[does or does not] form a gate of a
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.
12
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
6
7
8
9
10
United States District Court
Northern District of California
11
12
13
14
15
16
17
To the extent that there is a contact gate, it should be defined as above the gate it contacts—but
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.6 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 III.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
28
6
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—
“the gate contact”—despite the absence of gate contacts with some gate structures/ gate electrode
features, Tela’s proposal is not precise.
13
1
C. “gate electrode”
Tela’s Proposal
Intel’s Proposal
Court’s Ruling
portion of the [linear gate
a portion of the conductive
a portion of the conductive
electrode segment (’352
shape in the gate layer that
shape that extends over the
Patent) / linear conductive
extends over and parallel with diffusion region and is used
segment (’966 and ’012
a diffusion region to form a
to control the flow of
Patents) / gate structure (’334 transistor gate
electrical current between the
and ’335 Patents) / gate
source and drain regions of a
electrode feature (’523
Patent)] used to control the
transistor
flow of electrical current
between the source and drain
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
2
3
4
5
6
7
8
The parties next dispute the term “gate electrode,” which is found in all six patents at issue.
9
10
United States District Court
Northern District of California
11
12
The parties dispute (i) whether to construe gate electrode as extending over and parallel with a
diffusion region and (ii) whether to describe its structure (as forming a transistor gate) or its
function (as controlling the flow of electrical current).7
At the hearing, Tela generally expressed agreement that the gate electrode extends over the
13
14
15
16
17
18
19
20
21
22
23
24
25
diffusion region, and language in the patents supports this understanding. See ’352 Patent at
Abstract (“Each linear gate electrode track . . . extends over . . . a diffusion region . . . .”); ’966
Patent at Abstract (providing that the “diffusion region layout shapes to be formed within a portion
of a substrate . . . a gate electrode level above the portion of the substrate”); id. at 8:20-29
(describing “gate electrode portions which extend over one or more of the [p/n]-type diffusion
regions to form respective [P/N]MOS transistor devices”); id. at 12:38-40 (providing that “gate
electrode features 207 are defined above the diffusion regions 203 to form transistor gates”); Op’g
14 (noting “the gate of a transistor is formed by the portion of the gate electrode feature or gate
structure that extends over the active portions of the substrate”). Tela raised at the hearing its
concern that because there must be a channel between a transistor’s source and drain region, a
literal understanding of Intel’s proposal would prevent a transistor from being formed on a
physical chip.8 With the caveat that I do not construe this term in a way that would prevent the
26
27
28
7
Intel argues that its proposal is substantially the same as a construction for gate electrode Tela
agreed to adopt in a different case. According to Tela, different patents were at issue in that case.
8
Tela has no concerns with this aspect of the construction with respect to the layout view.
14
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 7:2123 (providing, “the linear-shaped layout features in a given layer extend in a common direction
11
United States District Court
Northern District of California
10
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
25
26
27
28
9
At the hearing, Intel clarified that it will not argue that complete overlap is necessary in order to
“extend over.” It understands “extend over” to mean in the vertical direction.
10
Intel’s reliance on Figure 2 of the ’966 Patent is not persuasive because it is “an illustration
showing a generalized stack of layers used to define a dynamic array architecture, in accordance
with one embodiment of the present invention.” ’352 Patent 6:15-17.
15
1
2
3
4
5
D. “gate electrode contact, gate contact structure(s), contact structure”
Tela’s Proposal
Intel’s Proposal
Court’s Ruling
a structure that passes
conductive structure(s) in
conductive structure(s) in a gate
through an insulator to enable a gate contact layer above contact layer above and separate
connection of the gate
and separate from the gate from the gate layer and below
electrode feature to the
layer and below and
overlying metal conduction
separate from interconnect and separate from interconnect
layers
lines
layers
’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.
United States District Court
Northern District of California
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
28
16
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
8
9
10
United States District Court
Northern District of California
11
12
13
14
15
16
gate electrode
gate contact
interconnect
17
18
All of 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
17
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
8
9
10
United States District Court
Northern District of California
11
12
13
E. “interconnect level region”
Tela’s Proposal
Intel’s Proposal
area within a layer having
an area of a layer with
metal and/or via conductive
conductive structures that
structures where the layer is
traverse over the substrate to
located above the diffusion
enable desired routing and
contact layer/gate electrode
connectivity
contact layer and enables
definition of the desired
circuit connectivity
’966: 31, 33, 35; ’012: 26, 28, 30
Court’s Ruling
an area within a layer having
conductive structures where
the layer is located above the
diffusion contact layer/gate
electrode contact layer and
enables definition of the
desired circuit connectivity
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
28
18
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
form interconnects. For these reasons, I will adopt a modified version of Tela’s proposed
11
United States District Court
Northern District of California
10
construction.
12
13
14
15
F. “[gate / metal / contact] gridline(s)”
Tela’s Proposal
Intel’s Proposal
virtual projected lines along
gridline(s) along which [gate
which [gate /metal / contact]
/ metal / contact] structures
features are defined
are formed
Court’s Ruling
virtual projected lines along
which [gate /metal / contact]
features are positioned
’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
The parties next dispute the term “grideline(s),” which is found in the ’334, ’335, and ’523
16
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
27
28
11
Intel does not actually dispute the accuracy of this description of the layers’ relative
orientations. With the clarification that I recognize some claims may omit contact layers, I aim to
assuage Intel’s concerns that the requirement of such a layer will be improperly added into some
claims.
19
Using the word “positioned” resolves the dispute between “defined” and “formed.”
1
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 . . .
positioned to extend lengthwise in a y-direction in a substantially centered manner along an
11
United States District Court
Northern District of California
10
associated gate gridline”). At the hearing, both parties agreed with the tentative after clarifying
12
their views.12
13
14
15
16
G. “physically and electrically separated by a [conductor] line end spacing of
minimum size”
Tela’s Proposal
Plain and ordinary meaning.
Not indefinite.
’352: 1
Intel’s Proposal
Indefinite.
Court’s Ruling
Indefinite.
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
28
12
Specifically, Intel confirmed that structures are positioned on virtual gridlines on the chip itself,
and Tela confirmed that the positioning of the structures along gridlines is determined during the
layout process.
20
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).
The claim language surrounding this term reads:
at least one of the linear gate electrode tracks having multiple linear
gate electrode segments adjacently defined thereover in an end-to-end
manner such that facing ends of adjacent linear gate electrode
segments are physically and electrically separated by a line end
spacing of minimum size, wherein a size of each line end spacing
within a window of lithographic influence is substantially the same,
wherein the minimum size of the line end spacing corresponds to a
substantially full occupancy of the at least one linear gate electrode
track by the multiple linear gate electrode segments . . .
8
9
10
United States District Court
Northern District of California
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
’352 Patent 21:23-35. The specification devotes little attention to defining “spacing of minimum
size.” It provides,
When a given gate electrode track is required to be interrupted, the
separation between ends of the gate electrode track segments at the
point of interruption is minimized to the extent possible taking into
consideration the manufacturing capability and electrical effects. . . .
Minimizing the separation between ends of the gate electrode track
segments at the points of interruption serves to maximize the
lithographic reinforcement, and uniformity thereof, provided from
neighboring gate electrode tracks.
Tela argues that these descriptions are sufficient to allow one of ordinary skill in the art to
understand “that the line-end spacing is placed and sized in a manner to maximize lithographic
reinforcement of neighboring gate electrode segments while taking into consideration electrical
effects.” Op’g 25.
I disagree. The intrinsic evidence reveals where “line end spacing” is located and why
“minimum size” is desirable,13 but it provides no objective way to determine what “minimum
size” means. Because the intrinsic evidence does not provide an objective boundary, it is
27
28
13
As articulated by Tela, “the minimum size allows for substantially full occupancy of the gate
electrode track.” Op’g 24.
21
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
manufacturing yield might approach 100%, but as the space decreases, the yield will similarly
11
United States District Court
Northern District of California
10
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”
28
indefinite even though the user “predefined” the value for maximum time value before beginning
22
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
United States District Court
Northern District of California
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
23
1
CONCLUSION
2
For the reasons set forth above, the disputed terms are construed as follows:
Term
Court’s Ruling
“linear gate electrode segment, linear
extending in a single direction over the
conductor segment(s), linear conductive
substrate
segment(s), and (interconnect) linear
conductive structures”
“gate structure(s) and gate electrode
feature comprising a gate(s) of a transistor(s)
feature(s)”
or a dummy gate
“gate electrode”
a portion of the conductive shape in the gate
layer that extends over the diffusion region
and is used to control the flow of electrical
current between the source and drain regions
of a transistor
“gate electrode contact, gate contact
conductive structure(s) in a gate contact layer
structure(s), contact structure”
above and separate from the gate layer and
below and separate from interconnect layers
“interconnect level region”
an area within a layer having conductive
structures where the layer is located above the
diffusion contact layer/gate electrode contact
layer and enables definition of the desired
circuit connectivity
“[gate / metal / contact] gridline(s)”
virtual projected lines along which [gate
/metal / contact] features are positioned
“physically and electrically separated by a
Indefinite
[conductor] line end spacing of minimum
size”
3
4
5
6
7
8
9
10
United States District Court
Northern District of California
11
12
13
14
15
16
17
18
19
IT IS SO ORDERED.
Dated: November 4, 2019
20
21
William H. Orrick
United States District Judge
22
23
24
25
26
27
28
24
Disclaimer: Justia Dockets & Filings provides public litigation records from the federal appellate and district courts. These filings and docket sheets should not be considered findings of fact or liability, nor do they necessarily reflect the view of Justia.
Why Is My Information Online?