Northern Telecom Limited v. Samsung Electronics Co., Ltd. And Samsung Semiconductor, Inc., Defendants-Cross

215 F.3d 1281, 55 U.S.P.Q. 2d (BNA) 1065, 2000 U.S. App. LEXIS 13696, 2000 WL 763743

• Court: Court of Appeals for the Federal Circuit
• Decided: June 13, 2000
• Docket: 99-1208, 99-1227
• Status: Published

Opinion

CLEVENGER, Circuit Judge.

Northern Telecom Limited (“Northern Telecom”) appeals from the United States District Court for the Northern District of California,-which granted summary judgment invalidating Northern Telecom’s United States Patent No. 4,030,967 (“the ’967 patent”) for failure to disclose the best mode contemplated by the inventor of carrying out the invention. See Northern Telecom Ltd. v. Samsung Elec. Co., Ltd., No. C-95-449, slip op. at 1 (N.D.Cal. Sep. 16, 1996). Samsung Electronics Co., Ltd., and Samsung Semiconductor, Inc. (collectively “Samsung”) cross-appeal the district court’s grant of summary judgment in favor of Northern Tele-com on claim construction and its finding that Samsung literally infringed the ’967 patent. Because we disagree with the district court’s conclusion that the inventors of the ’967 patent failed to disclose the best mode of carrying out their invention, we reverse the court’s grant of summary judgment of invalidity and remand for further proceedings. We affirm, however, the district court’s rulings on claim construction and literal infringement.

I

Northern Telecom owns the ’967 patent, titled “Gaseous Plasma Etching Of Aluminum And Aluminum Oxide.” The ’967 patent contains one independent claim: “A process for gaseous etching of aluminum and aluminum oxide, including an initial step of plasma etching in the presence of a gaseous trihalide comprising at least in part, a boron trihalide.”

Aluminum etching is used in the manufacture of integrated circuit semiconductor devices to create conductive lines of aluminum or aluminum alloy between various electronic devices on a silicon substrate. A typical manufacturing process begins with a silicon wafer coated with a conductive layer of aluminum or aluminum alloy. The manufacturer first covers the conductive layer with a mask of nonetchable material. Next, an aluminum etching process is applied to remove portions of the conductive film not protected by the mask. Finally, the manufacturer removes the masking material, leaving the desired pattern on the surface of the semiconductor substrate.

Until the mid 1970’s, most integrated circuit manufacturers used wet chemicals to remove aluminum from the silicon substrate during the etching process. However, because wet etching tended to “undercut” desirable aluminum beneath the mask, this technique caused poor pattern definition and limited attempts to reduce conductive line widths. Wet etching also involved the use of toxic chemicals which were expensive and environmentally damaging. Dry etching offered an alternative to wet etching that would potentially avoid the undercutting problem and the costs associated with using and disposing of toxic chemicals.

In 1975-76, Northern Telecom undertook a research and development project for the United States Army Electronic Command to develop a new etching technique that would permit smaller semiconductor devices. The Northern Telecom researchers focused on plasma etching, a type of dry etching. Previous attempts to plasma etch aluminum had failed due to the presence of a naturally-occurring layer of aluminum oxide that forms on the surface of aluminum when exposed to air. These failures occurred because chlorine and bromine-based gases would not etch through the aluminum oxide layer. The ’967 inventors discovered that certain etch gases — specifically boron trihalides— would remove the oxide layer and permit etching of the aluminum underneath. The use of boron trihalides in plasma etching enabled etching of aluminum line widths below 2 microns, a significant advance in the miniaturization of semiconductor devices.

There are three types of dry etching techniques important to this case: (1) sputter etching; (2) plasma etching; and (3) reactive ion etching. Sputter etching is essentially a mechanical process whereby the metal being etched is bombarded with energetic ions that physically dislodge atoms from the exposed metal. There are no chemical reactions involved in sputter etching. Although sputter etching was known at the time of the ’967 invention, it was considered a very slow process and, thus, undesirable.

Plasma etching utilizes a gas plasma, which is created by applying voltage to a low pressure gas. The gas plasma comprises a mixture of particles: reactive neutrals, excited neutrals, electrons, positively charged chemical species, and negatively charged chemical species. Plasma etching is primarily a chemical process in which a reaction occurs between an active species in the plasma and the metal surface. The parties dispute whether plasma etching, in the context of the.’967 patent, necessarily excludes the presence of charged ions that mechanically dislodge atoms from the exposed metal, as occurs in the sputter etching process. The district court held that “plasma etching refers to a chemical process without excluding the non-chemical process of ion bombardment.”

Reactive ion etching combines the mechanical ion bombardment of sputter etching with the chemical process of plasma etching. In this technique, the metal surface to be etched is intentionally charged with a negative voltage, which causes positively charged ions in the gas plasma to accelerate toward the metal surface, bombarding the surface with enough energy to dislodge exposed metal atoms. Simultaneously, active radicals in the gas plasma chemically react with the metal surface to form volatile compounds that are then removed from the metal surface.

The final step in manufacturing an integrated circuit semiconductor device is typically to heat or “sinter” the etched product at temperatures above 400° C. Sintering improves the contact between the conductive lines and the silicon substrate. However, during sintering, a phenomenon known as “spearing” may occur, which results in tiny metal protrusions extending downward from the bottom of the aluminum layer into the silicon substrate. Spearing occurs because elevated temperatures cause silicon to diffuse upward into the aluminum layer, allowing aluminum to fill the spaces vacated by the silicon. Spearing is undesirable because it may cause short circuits between multiple conductive layers in an integrated circuit semiconductor device.

One solution to the spearing problem is to use an aluminum silicon alloy as the metallic layer rather than pure aluminum. If the silicon content of the aluminum alloy exceeds the solid solubility limit for silicon in aluminum at the sintering temperature, then no diffusion of silicon from the substrate will occur and, thus, spearing will be prevented. It is undisputed that the inventors of the ’967 patent knew that aluminum silicon alloy could be used in this manner and, indeed, believed that the use of aluminum silicon alloy was necessary to manufacture semiconductors below 2 microns.

Northern Telecom’s infringement action against Samsung, originally filed in the United States District Court for the Northern District of Texas, was transferred to the District Court for the Northern District of California in February 1995. On September 17, 1996, the district court issued an opinion setting forth its interpretation of two disputed claim elements, “plasma etching” and “aluminum and aluminum oxide.” The district court construed “plasma etching” to be a chemical process, but one that does not necessarily exclude the mechanical process of ion bombardment. It construed “aluminum and aluminum oxide” to refer solely to pure aluminum and its native layer of aluminum oxide, and not to alloys such as aluminum silicon.

Samsung’s accused reactive ion etching process employs a boron trichloride plasma, which chemically reacts with and removes metal surface atoms. In addition, the metal surface is negatively charged, which causes positively charged ions in the plasma to accelerate toward the surface and mechanically remove some metal atoms. Samsung described this process as “synergistieally ■ combining] energetic ion bombardment with chemically active radicals to achieve an etch rate far in excess of what would be achieved by plasma etching or sputter etching alone.”. Notwithstanding the additional element of ion bombardment, the district court found that, because Samsung’s process included boron trihal-ide plasma etching, it fell within the literal scope of the ’967 patent. In addition, the court found that, although the ’967 patent specifically claims a method of plasma etching pure aluminum and aluminum oxide, Samsung’s use of an aluminum silicon alloy did not avoid this limitation because pure aluminum is present within the aluminum silicon alloy.

After discovery, Samsung moved for summary judgment of invalidity of the ’967 patent under 35 U.S.C. § 112, which requires an inventor to set forth in the patent specification “the best mode contemplated by the inventor of carrying out his invention.” Samsung alleged two best mode violations: (1) failure to disclose the use of aluminum silicon alloy; and (2) failure to disclose the use of a “cold trap” in the vacuum pump associated with the reactor used to generate the gas plasma.

The parties agree that the primary use of the ’967 patent is in the manufacture of semiconductor devices. Also undisputed is the fact that fine line etching — that is, etching of lines of less than 2 microns — is preferred for improved semiconductor devices. With those facts in hand, the district court concluded that the best mode of carrying out the invention is to achieve fíne line etching. Since the use of aluminum silicon alloy is necessary to preclude the spearing phenomenon during the sin-tering phase, which occurs after the plasma etching phase, the district court held that disclosure of aluminum silicon alloy was required to satisfy the best mode requirement of 35 U.S.C. § 112. Because the inventors knew that aluminum silicon alloy is necessary to prevent spearing, and since the alloy is not disclosed, the patent was held invalid.

The district court entered final judgment in favor of Samsung without reaching the issue of the “cold trap.” Northern Telecom filed a timely motion to set aside the judgment of invalidity, which was denied by the district court on December 19, 1998. This appeal followed, vesting this court with jurisdiction pursuant to 28 U.S.C. § 1295(a)(1) (1994).

II

We independently review the district court’s grant of summary judgment, see Johnson Worldwide Assocs., Inc. v. Zebco Corp., 175 F.3d 985, 987, 50 USPQ2d 1607, 1609 (Fed.Cir.1999), and its interpretation of the patent, see Augustine Med., Inc. v. Gaymar Indus., Inc., 181 F.3d 1291, 1297, 50 USPQ2d 1900, 1904 (Fed.Cir.1999). Summary judgment is appropriate only if there is no genuine issue of material fact. See Fed.R.Civ.P. 56(c). To this end, the court must draw all reasonable factual inferences in favor of the non-movant. See Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 255, 106 S.Ct. 2505, 91 L.Ed.2d 202 (1986).

Compliance with the best mode requirement is a question of fact, see Chemcast Corp. v. Arco Indus. Corp., 913 F.2d 923, 928, 16 USPQ2d 1033, 1037 (Fed.Cir.1990), which is composed of two best mode inquiries. The first inquiry is whether, at the time of filing the patent application, the inventor considered a particular mode of practicing his invention superior to all other modes. This inquiry is wholly subjective; that is, it focuses on the inventor’s state of mind at the time he filed his patent application. See id., 913 F.2d 923, 16 USPQ2d at 1037. The second inquiry is whether the inventor’s disclosure is adequate to enable one of ordinary skill in the art to practice the best mode of the invention. This inquiry is objective and depends upon the scope of the claimed invention and the level of skill in the relevant art. See id.

A

As we have repeatedly held, the contours of the best mode requirement are defined by the scope of the claimed invention. See Engel Indus., Inc. v. Lockformer Co., 946 F.2d 1528, 1531, 20 USPQ2d 1300, 1302 (Fed.Cir.1991) (“The best mode inquiry is directed to what the applicant regards as the invention, which in turn is measured by the claims. Unclaimed subject matter is not subject to the disclosure requirements of § 112.”); see also Chemcast, 913 F.2d at 927, 16 USPQ2d at 1037 (an “objective limitation on the extent of the disclosure required to comply with the best mode requirement is, of course, the scope of the claimed invention”); Randomex, Inc. v. Scopus Corp., 849 F.2d 585, 588, 7 USPQ2d 1050, 1053 (Fed.Cir.1988) (“It is concealment of the best mode of practicing the c laimed invention that section 112 ¶ 1 is designed to prohibit” (emphasis in original).); Zygo Corp. v. Wyko Corp., 79 F.3d 1563, 1567, 38 USPQ2d 1281, 1284 (Fed.Cir.1996) (“The focus of a section 112 inquiry is not what a particular user decides to make and sell or even in what field the invention is most likely to find success. Rather, in keeping with the statutory mandate, our precedent is clear that the parameters of a section 112 inquiry are set by the claims” (emphasis in original).).

This consistent body of law postulates that the first task in any best mode analysis is to define the invention at hand. The definition of the invention, like the interpretation of the patent claims, is a legal exercise, wherein the ordinary principles of claim construction apply. See, e.g., Johnson Worldwide, 175 F.3d at 989, 50 USPQ2d at 1610; Vitronics Corp. v. Conceptronic, Inc., 90 F.3d 1576, 1582, 39 USPQ2d 1573, 1576-77 (Fed.Cir.1996). We begin, as always, with the language of the claim itself. See Johnson Worldwide, 175 F.3d at 989, 50 USPQ2d at 1610. Claim language is given its ordinary and accustomed meaning except where a different meaning is clearly set forth in the specification or where the accustomed meaning would deprive the claim of clarity. See id. at 989-90, 175 F.3d 985, 50 USPQ2d at 1610; Renishaw PLC v. Marposs Società per Azioni, 158 F.3d 1243, 1249, 48 USPQ2d 1117, 1121 (Fed.Cir.1998).

When the invention is defined, the best mode inquiry moves to determining whether a best mode of carrying out that invention was held by the inventor. If so, that best mode must be disclosed.

In this case, Northern Telecom admits that the best mode of manufacturing fine line semiconductor devices requires use of aluminum silicon alloy. However, Northern Telecom contends .that this admission is irrelevant to the best mode inquiry since its patent is not drawn to a process for manufacturing fine line semiconductor devices. Indeed, Northern Telecom points out that its claimed process is advantageous in the manufacture of semiconductor devices, without regard to line width, because of the many uncontested advantages of gaseous plasma etching over wet chemical processes that are made possible by plasma etching in the presence of a gaseous trihalide as taught in claim 1.

Northern Telecom also points out that its patent does not prohibit Samsung from manufacturing semiconductor devices with fine lines. The patent relates, instead, to a process for gaseous etching of aluminum and aluminum oxide that employs a gaseous trihalide. So long as Samsung does not employ the patented process, Northern Telecom cannot use its patent to restrain manufacture of fine line semiconductor devices. From this analysis as well, Northern Telecom asserts that its patent, for best mode purposes, should not be tied to a process for making fine line devices. In short, Northern Telecom asserts that the district court erred in requiring disclosure of a mode for carrying out something other than the claimed invention.

Samsung selects a single word to describe Northern Telecom’s position: “disingenuous.” Such is apt, according to Samsung, because even the inventors concede that the preferred outcome in semiconductor device manufacturing is the achievement of fine line etching. Since the inventors’ process can be used to achieve that goal, the best mode of avoiding the spearing phenomenon — which occurs at a step remote from the patented gaseous etching step — must be disclosed in the inventors’ patent. For support, Samsung refers to our decision in Dana Corp. v. IPC Ltd. Partnership, 860 F.2d 415, 418, 8 USPQ2d 1692, 1695 (Fed.Cir.1988).

B

We agree with Northern Telecom that the district court misunderstood the invention at hand. What is claimed in claim 1 is a process for plasma etching of aluminum and aluminum oxide in the presence of a gaseous trihalide. Fine line etching is simply not part of the claimed invention.

Claim 1 states in its entirety: “A process for gaseous etching of aluminum and aluminum oxide, including an initial step of plasma etching in the presence of a gaseous trihalide comprising at least in part, a boron trihalide.” The preamble of the claim clearly and unambiguously sets forth the utility of the claimed invention, namely “etching aluminum and aluminum oxide.” 1

Samsung does not dispute that the function stated in the preamble has an ordinary and accustomed meaning in the art which does not deprive the claim of clarity. Instead, Samsung asserts that the phrase should be read to include the narrower utility of fíne line etching because the method of claim 1 was conceived and intended solely for the purpose of producing semiconductor devices with “fine aluminum conductive lines.” In support of this assertion, Samsung relies on extrinsic evidence, that is, evidence outside of the patent itself. For instance, Samsung points to an interim report drafted by Northern Telecom during its research project for the Army Electronics Command, in which Northern Telecom articulated the goal of developing a “practical 2 micron CMOS process.” Samsung also points to statements of Northern Telecom’s technical experts and attorneys as proof that semiconductor miniaturization was the overall “goal” of the ’967 patent.

In this case, the scope of the claimed invention is clearly set forth in the claim itself, and the written description does not require a contrary reading. The use of extrinsic evidence to construe the scope of a claim is improper where the ordinary and accustomed meaning of a claim term does not render the scope of the claim unclear and where the patentee has not chosen to be his own lexicographer. See Johnson Worldwide, 175 F.3d at 990, 50 USPQ2d at 1610. Therefore, to the extent that the district court concluded that the invention set forth in claim 1 is limited to fine line etching, the court erred.

The cases cited by Samsung do not compel a different result. In Dana, 860 F.2d at 419, 8 USPQ2d at 1695, we held that the failure to disclose an unclaimed fluoride treatment, which was necessary for satisfactory performance of the claimed seal, was a violation of the best mode requirement. Similarly, in Northern Telecom, Inc. v. Datapoint Corp., 908 F.2d 931, 940-41, 15 USPQ2d 1321, 1328 (Fed.Cir.1990), we held that the failure to disclose special audio tapes for capturing data, which the inventors preferred to conventional audio cassettes, was a violation of the best mode requirement, where the claim included the use of magnetic tapes. See id. at 940, 908 F.2d 931, 15 USPQ2d at 1323.

Dana and Datapoint are different from the present case because each involved a situation in which the omitted best mode related directly to the claimed invention. For instance, the disputed claim in Dana included, inter alia,, “a valve stem seal for sealing between said valve stem and said valve guide.” 860 F.2d at 416, 8 USPQ2d at 1694. The claim stated that the invention was for use “[i]n an internal combustion engine.” Id. We held that the inventor’s failure to disclose an unclaimed fluoride treatment violated the best mode requirement because the treatment was necessary for satisfactory performance of the valve stem seal. See id. at 419, 860 F.2d 415, 8 USPQ2d at 1695. Thus, the asserted best mode in Dana directly related not only to the claimed utility of the invention (sealing poppet valve stems in an internal combustion engine), but also to a specific claim limitation (a valve stem seal).

Likewise, in Datapoint, the disputed claims were directed to a method of capturing source data comprising, inter alia, storing data on magnetic tape. We affirmed the district court’s finding that the inventors violated the best mode requirement by not disclosing a specially designed magnetic tape that was preferred to ordi nary audio tape cassettes. See Datapoint, 908 F.2d at 940-41, 15 USPQ2d at 1328. Again, the asserted best mode in Data-point directly related both to the claimed utility (capturing source data) and a specific claim limitation (storing data on magnetic tape). Accord Spectra-Physics, Inc. v. Coherent, Inc., 827 F.2d 1524, 1536, 3 USPQ2d 1737, 1746 (Fed.Cir.1987) (failure to disclose a six-stage braze cycle violated best mode requirement, where a claim limitation required a means for attaching two components).

Dana and Datapoint are consistent with other decisions of this court in which we have held that unclaimed matter that is unrelated to the operation of the claimed invention does not trigger the best mode requirement. For example, in Randomex, 849 F.2d at 590, 7 USPQ2d at 1054, we held that an inventor’s deliberate concealment of a preferred cleaning fluid formula did not violate the best mode requirement because his “invention neither added nor claimed to add anything to the prior art respecting cleaning fluid.” Likewise, in DeGeorge v. Bernier, 768 F.2d 1318, 1325, 226 USPQ 758, 763 (Fed.Cir.1985), we held that an inventor’s nondisclosure, of a word processor configuration did not violate the best mode requirement. We held that, “[b]eeause the properly construed count does not include a word processor, failure to meet the best mode requirement here should not arise from an absence of information on the word processor.” Id.

In the present case, the best mode alleged by Samsung is the use of aluminum silicon alloy — instead of pure aluminum— as the interconnect layer in manufacturing integrated circuits with conductive lines of less than 2 microns. However, as we have discussed above, claim 1 is not limited to the use of gas plasma etching to produce semiconductor devices with fine conductive lines. Instead, the claim is directed to a general process for plasma etching of aluminum and aluminum oxide in the presence of a gaseous trihalide. Accordingly, the inventors of the process were not required, under 35 U.S.C. § 112, to disclose the best mode of achieving fíne line etching — only the best mode of carrying out the claimed method of plasma etching aluminum and aluminum oxide with a gaseous trihalide. It is undisputed that this requirement was satisfied in the specification.

C

In its summary judgment motion, Samsung argued that the inventors of the ’967 patent also violated the best mode requirement by failing to disclose the use of a “cold trap” in conjunction with a vacuum pump disclosed in the written description. The district court declined to reach this issue, finding that the failure to disclose aluminum silicon alloy was a sufficient basis upon which to grant summary judgment. On appeal, Samsung urges this court to consider the cold trap issue as an alternative ground for affirming the district court’s grant of summary judgment. We decline to do so, since the best mode conclusion rests on questions of fact which should be decided by the district court in the first instance. We therefore express no view on” the “cold trap” issue, which remains open for adjudication upon remand.

Ill

Samsung cross appeals the district court’s construction of claim 1 and grant of summary judgment of infringement in favor of Northern Telecom. In particular, Samsung challenges: (a) the district court’s construction of “aluminum and aluminum oxide” to mean pure aluminum and its native oxide layer without further limitation; and (b) the district court’s conclusion that “plasma etching,” as claimed in the ’967 patent, encompasses processes that combine chemical reactive etching and ion bombardment. We take these arguments in turn.

A

To the district court, Samsung asserted that "aluminum and aluminum oxide," in claim 1, means pure aluminum and aluminum oxide-that is, not any other metals or alloys in combination with aluminum. See Northern Telecom, No. C-95-449, slip op. at 31. After analyzing the intrinsic evidence of record, the district court agreed with Samsung's construction and concluded that the `967 patent "refers solely to substantially pure aluminum and its native layer of aluminum oxide, and not to alloys such as aluminum silicon or aluminum silicon copper." Id., slip op. at 33. Nevertheless, the district court found that Samsung's process literally infringes the `967 patent because pure aluminum is indisputably present in the aluminum silicon alloy that is etched in that process. See id., slip op. at 34. That is, the district court concluded that there was no genuine dispute of fact that Samsung's process does indeed etch "aluminum and aluminum oxide" as claim 1 requires. See id. The court reasoned that the additional etching of the silicon and copper in the Samsung process was simply an additional step or aspect of the accused process and thus did not prevent a finding of literal infringement. See id.; see also e.g., A.B. Dick Co. v. Burroughs Corp., 713 F.2d 700, 703, 218 USPQ 965, 967 (Fed.Cir.1983) ("It is fundamental that one cannot avoid infringement merely by adding elements if each element recited in the claims is found in the accused device.").

On appeal, Samsung does not dispute that its accused process etches "aluminum" as defined by the district court: that is, aluminum-in the aluminum silicon or aluminum silicon copper alloys-is etched by the Samsung process. Instead, Samsung presents to this court yet another interpretation of the "aluminum and aluminum oxide" limitation of claim 1. Specifically, Samsung now asserts that the aluminum must be arranged in a "layer" to meet the requirements of the "aluminum and aluminum oxide" limitation, notwithstanding that the term “layer” does not appear anywhere in the text of claim 1.

We note at the outset that we look with “extreme disfavor” on appeals that allege error in claim constructions that were advocated below by the very party now challenging them. Key Pharms. v. Hercon Labs. Corp., 161 F.3d 709, 714-15, 48 USPQ2d 1911, 1915 (Fed.Cir.1998). In this case, because we find little merit to its arguments on this point, we need not address whether Samsung is judicially es-topped from challenging its own claim construction adopted by the trial court. See, e.g., Yniguez v. Arizona, 939 F.2d 727, 738 (9th Cir.1991) (judicial estoppel “preventfs] a party from changing its position over the course of judicial proceedings when such positional changes have an adverse impact on the judicial process.”).

Samsung's first argument is that the district court's construction of "aluminum"-as pure elemental aluminum-would work (in its view) the "absurd" re-suit that claim 1 could cover processes where even very small (i.e., "trace") amounts of pure aluminum were etched. In essence, Samsung appears to suggest that the additional "layer" limitation must be read into the claim to ensure that only larger (i.e., non-"trace") quantities of etched aluminum are covered by the `967 patent. We reject this contention. Not only is the term "layer" not found in claim 1, but neither is a limitation that establishes a minimum quantity of aluminum that must be etched to meet the claim. This court has repeatedly and clearly held that it will not read unstated limitations into claim language. See, e.g., Renishaw PLC v. Marposs Societa' per Azioni, 158 F.3d 1243, 1248, 48 USPQ2d 1117, 1120 (Fed.Cir.1998); Vitronics Corp. v. Conceptronic, Inc., 90 F.3d 1576, 1582, 39 USPQ2d 1573, 1576 (Fed.Cir.1996); Markman v. Westview Instruments, Inc., 52 F.3d 967, 979-90, 34 USPQ2d 1321, 1329-30 (Fed.Cir.1995) (en banc), aff'd, 517 U.S. 370, 116 S.Ct. 1384, 134 L.Ed.2d 577, 38 USPQ2d 1461 (1996). In any event, we fail to see how Samsung’s suggestion would resolve what it views as an “absurd” result: the addition of a “layer” requirement would only mandate that the aluminum (in whatever quantity) be disposed in a roughly planar fashion. Thus, very small amounts of aluminum (“trace”), if etched, would appear to meet even Samsung’s proposed definition, thus perpetuating the result that Samsung finds “nonsensical”. The limitation “aluminum” stands unmodified in claim 1; merely characterizing the claim language as “absurd”, without more, does not convince us that “aluminum” must be read to incorporate a “layer” limitation. See, e.g., Johnson Worldwide Assoc., Inc. v. Zebco, Corp., 175 F.3d 985, 989, 50 USPQ2d 1607, 1610 (Fed.Cir.1999) (canons of claim construction require that “[g]eneral descriptive terms will ordinarily be given their full meaning; modifiers will not be added to broad terms standing alone.”); Virginia Panel Corp. v. MAC Panel Co., 133 F.3d 860, 865-66, 45 USPQ2d 1225, 1229 (Fed.Cir.1997); Bell Communications Research, Inc., v. Vitalink Communications Corp., 55 F.3d 615, 621-22, 34 USPQ2d 1816, 1821 (Fed.Cir.1995).

Samsung’s second argument is that the term “layer” must be read into the “aluminum” limitation of claim 1 because the written description of the ’967 patent describes embodiments of the invention with references to “layers” of aluminum or “lines” of aluminum. See, e.g., ’967 Patent, col. 1, 11. 55-60, col. 1, 11. 64-65, col. 3, 1. 8. This we cannot do. In order to overcome the presumption that the meaning of the claim term “aluminum” is other than its plain and ordinary meaning- — what the district court described as “pure” or elemental aluminum — Samsung must demonstrate either that (1) the patentee provided an alternate definition of “aluminum” in the intrinsic evidence; or (2) that the ordinary meaning of “aluminum” deprives the claim of clarity. See Johnson Worldwide, 175 F.3d at 990, 50 USPQ2d at 1610. Samsung does not argue that the ordinary meaning of “aluminum” is unclear; instead, Samsung appears’ to suggest that the several references to aluminum “layers” or “lines” in the written description provide a special definition of “aluminum” — i.e., that, in the context of claim 1, “aluminum” means only “aluminum layers” or “aluminum lines”. Yet, a review of the written description undermines Samsung’s arguments: throughout the ’967 patent, aluminum is referenced both in terms of its elemental variety (by its symbol Al) and as a “layer”. Compare, e.g., ’967 Patent, col. 1, 11. 46, col. 1, 11. 67, col. 3, 1. 8, col. 3, 1. 9, col. 3, I. 16, col. 3 (referring to aluminum as “Al”), with ’967 Patent, col. 1, 11. 55-56, col. 1, 1. 61, col. 1, 11. 64-65, col. 3, 1. 27 (aluminum as “aluminum layer”). As we held in Johnson Worldwide, the “[v]aried use of a disputed term in the written description demonstrates the breadth of the term rather than providing a limited definition.” 175 F.3d at 991, 50 USPQ2d at 1611 (reference to “heading” as both direction of boat and as direction of motor supports view of “heading” as encompassing both definitions); see also Enercon GmbH v. International Trade Comm’n, 151 F.3d 1376, 1385, 47 USPQ2d 1725, 1731-31 (Fed.Cir.1998) (refusing to limit a word used “interchangeably” in the written description to only one of the uses of the term). Thus, Samsung’s argument based on the written description fails.

Samsung’s final argument on this point is that the district court’s construction of “aluminum” as “pure” aluminum renders the “aluminum oxide” limitation superfluous. That is, Samsung suggests that because aluminum oxide also includes “aluminum” as defined by the district court, the district court has effectively negated the “aluminum oxide” limitation. Samsung’s argument is misplaced. The “aluminum oxide” limitation is an additional limitation in claim 1. That is, to infringe claim 1, both aluminum and aluminum oxide must be etched. Or, to state it differently, both elemental aluminum and the molecular combination of aluminum and oxygen must be etched. Any accused process that fails to etch either aluminum or the molecular combination of aluminum and oxygen (ie., “aluminum oxide”) will not infringe. For example, a process that etches only aluminum silicon alloy will not, of course, etch aluminum oxide. Thus, the district court’s definition of aluminum does nothing to render the “aluminum oxide”-limitation as other than a limit on the scope of claim 1.

Accordingly, we hold that the district court’s construction of “aluminum and aluminum oxide” in claim 1 was correct.

Samsung also argues that the district court erred when it construed “plasma etching” to allow reactive ion etching. According to Samsung, “plasma etching,” as used in claim 1, specifically excludes all forms of “ion bombardment” etching. Because the accused process uses reactive ion etching — a combination of both plasma etching and ion bombardment — Samsung asserts that no infringement can lie. Samsung asserts that both the specification and prosecution history of the ’967 patent support its view of “plasma etching”. We disagree.

Some background is in order. There is no dispute between the parties that plasma etching, as that term is commonly used in the field, is a “chemical” process. In plasma etching, electrical power is applied to a gas, creating a plasma of chemically-active radicals, which react with the workpiece, causing etching. There is also no dispute between the parties that ion bombardment is a “mechanical” process, whereby an electrical field creates a plasma that emits charged particles, which physically strike the workpiece, causing etching.- Reactive ion etching, used in the Samsung process, is a combination of the chemical and mechanical processes. In reactive ion etching, the plasma created by the electrical field creates both chemically-active and charged particles, allowing etching to pro-céed both by chemical reaction and by physical bombardment.

Claim 1 of the ’967 patent requires “an initial step of plasma etching.” The claim does not mention ion bombardment or reactive ion etching. Thus, there is no dispute between the parties that the ’967 patent does not claim ion bombardment as an etching mechanism. But this in itself does not allow the Samsung process, to escape infringement: the reactive ion etching of the Samsung process, as we noted above, uses “plasma etching” (the chemical process) as part of the etching mechanism. The use of ion bombardment as another part of the etching mechanism, an additional element of the accused process, will not ordinarily prevent a finding of infringement. See, e.g., A.B. Dick, 713 F.2d at 703, 218 USPQ at 967. Samsung, recognizing this tenet of the law, argues instead that the limitation “plasma etching” in claim 1, read in light of the intrinsic and extrinsic evidence, specifically requires the exclusion of ion bombardment as any part of the etching mechanism.

Samsung does not argue that the ordinary meaning of “plasma etching” itself excludes ion bombardment. See Johnson Worldwide, 175 F.3d at 989, 50 USPQ2d at 1610 (“The general rule is, of course, that terms in the claim are to be given their ordinary and accustomed meaning.”); Vitronics, 90 F.3d at 1582, 39 USPQ2d at 1576 (“[WJords in a claim are generally given their ordinary and customary meaning.”). Instead, Samsung argues that the intrinsic and extrinsic evidence of record compels the conclusion that “plasma etching” in claim 1 must be limited to circumstances where chemical processes are the only etching mechanism. See Johnson Worldwide, 175 F.3d at 989-90, 50 USPQ2d at 1610 (describing circumstances under which a term’s ordinary meaning may be set aside). Thus, the issue is not whether the ’967 patent contemplates the presence of ion bombardment, but whether it contemplates it and specifically excludes it.

We begin with the specification. See Vitronics, 90 F.3d at 1582, 39 USPQ2d at 1577 (“[I]t is always necessary to review the specification to determine whether the inventor has used any terms in a manner inconsistent with their ordinary meaning.”). Samsung is correct in noting that the written description broadly describes “plasma etching” in terms of a chemical reaction. See ’967 patent, col. 1., 11. 37-42. Indeed, the only mention of “ion bombardment” is the statement that “the RF power should be sufficiently low as not to damage the photoresist through ion bombardment.” See id. at col. 2, 11. 54-56. Samsung seizes upon this statement as evidence that the claimed “plasma etching” must completely exclude ion bombardment. We disagree. By suggesting — in connection with a description of the preferred parameters for the claimed “plasma etching” process — that steps “should” be taken to reduce or avoid ion bombardment, the patentees fall far short of excluding ion bombardment altogether. Cf. Laitram Corp. v. Morehouse Indus., Inc., 143 F.3d 1456, 1463, 46 USPQ2d 1609, 1614-15 (Fed.Cir.1998) (adhering to clear assertions in written description that the invention “requires” a narrow construction of a claim term). See also Johnson Worldwide, 175 F.3d at 991, 50 USPQ2d at 1611 (discussing Laitram). Indeed, wholly contrary to Samsung’s view, that sentence clearly contemplates that ion bombardment may be present in the claimed process, but states that the patentees prefer that it be reduced or eliminated. In other words, the low level of ion bombardment in conjunction with the claimed plasma etching is plainly a preferred embodiment of the ’967 patent. This court consistently declines to construe claim terms according to the preferred embodiment. See, e.g., Johnson Worldwide, 175 F.3d at 990-92, 50 USPQ2d at 1610-12; Renishaw, 158 F.3d at 1248, 48 USPQ2d at 1120-22; Constant v. Advanced Micro-Devices, Inc., 848 F.2d 1560, 1571, 7 USPQ2d 1057, 1064 (Fed.Cir.1988). We see no reason to abandon that practice in this case.

Samsung’s reliance on Figure 2 of the ’967 patent is no more persuasive. Figure 2 shows the wafer (i.e., the “workpiece”) located in a plasma region, rather than in a DC electric field as would be the case if ion bombardment was to be the primary or only etching technique. While we can agree that Figure 2 makes clear that the patentees preferred gas plasma etching over ion bombardment, that fact, of course, cannot lead to the conclusion that ion bombardment was specifically excluded from the scope of the claims. As we note above, the entire specification plainly expresses a preference for gas plasma etching over ion bombardment. But preferred embodiments, without more, do not limit claim terms. See, e.g., Johnson Worldwide, 175 F.3d at 990-92, 50 USPQ2d at 1610-12; Renishaw, 158 F.3d at 1248, 48 USPQ2d at 1120-22; Advanced Micro-Devices, 848 F.2d at 1571, 7 USPQ2d at 1064.

Samsung next points to passages in the prosecution history, where, it asserts, the patentees unambiguously disclaimed the existence of any ion bombardment in the claimed process. During prosecution of the ’967 patent, the Examiner rejected several claims as obvious over Shockley, in view of Zielinski and Fraser. Shockley discloses a sputter etching process. Fraser and Zielinski disclose methods of reactive ion etching aluminum and aluminum oxide, respectively. None of the references discloses etching in the presence of boron trihalide. In response to the Examiner’s rejection, the applicants amended claim 1 and remarked:

[1] It is not agreed that the references cited in any way disclose the pres ent invention as defined by the claims originally filed or as now filed, nor do the references foreshadow the present invention.

[2] The references ... are concerned with a totally different process. In sputter etching, the plasma is en- ■ tirely incidental to the' ion bombardment, which bombardment is the etching mechanism. The effect can as readily be affected by using a non-plasma condition by bombarding the surface with ions from an ion gun. Further, a plasma is defined as neutral environment, that is one that has equal members of positive and negative ions. In ion etching (as compared to plasma etching), the target does not sit in the plasma but in a positive space charge.

[3] In the references the plasma plays no part in the etching, being incidental. In the present invention the plasma is part of the etching process in that the etching process uses neutrals, (uncharged particles) in the reaction, these being from the plasma.

Response to Office Action dated January 7, 1977, pp. 1-2 (paragraph numbers supplied). Samsung argues that the above remarks clearly portray “plasma etching” as a process that must exclude ion bombardment because the inventors remarked that the asserted references (which disclosed ion bombardment and reactive ion etching) are “concerned with a totally different process,” in paragraph 2.

We find the passage does not support Samsung’s argument. In the main, paragraphs 1 and 2 above are descriptions of the asserted references, as opposed to a description of "the “plasma etching” process of claim 1. By contrast, it is paragraph 3 which describes the invention, and the specific way that the claimed process differs from the asserted references. Paragraph 3 describes the plasma as “part” of the etching process, and states that the “etching process uses neutrals” in the reaction. These statements, while describing features of plasma etching,- do not exclude the possibility of ion bombardment. Indeed, describing the plasma as “part” of the etching mechanism rather than as “the” (or, perhaps, “the only”) etching mechanism, the language instead supports the inference that other mechanisms may also be “part” of the etching process. Again, much like the specification, this language evinces the inventors’ clear focus on plasma etching. It does not explicitly call for ion bombardment; but neither does it specifically state its exclusion as part of the invention. See, e.g., Cybor Corp. v. FAS Techs., Inc., 138 F.3d 1448, 1458, 46 USPQ2d 1169, 1175-76 (en banc) (refusing to limit scope of claim language where prosecution history did not clearly call for a narrower definition).

The description of the asserted references as “concerned with a totally different process” in paragraph 2 is far too slender a reed to support the judicial narrowing of a clear claim term. The inventors do not specify (except in paragraph 3, of course) how the references are “totally different.” Perhaps the inventors considered the references as “totally different” due to their failure to disclose plasma etching in the presence of a boron trihalide, or because of the relative importance of gas plasma etching in the ’967 invention. Or, as Samsung suggests, perhaps the inventors considered the presence of ion bombardment as creating a “totally different process.” Or it could be any number of other unstated reasons. Like the district court, we simply cannot tell.

In sum, we find that the prosecution history fails to prove Samsung’s assertion that “plasma etching” in claim 1 of the ’967 patent requires the exclusion of ion bombardment. That is, under these circumstances, we cannot conclude that Samsung has demonstrated that the patentees — with reasonable clarity and deliberateness, see In re Paulsen, 30 F.3d 1475, 1480, 31 USPQ2d 1671, 1674 (Fed.Cir.1994) — de fined “plasma etching” as excluding ion bombardment. See Johnson Worldwide, 175 F.3d at 989, 50 USPQ2d at 1610 (noting the “heavy presumption in favor of the ordinary meaning of claim language”).

Samsung next argues that ambiguity in the prosecution history places this case `within the ambit of our holding in Athletic Alternatives, Inc. v. Prince Mfg., Inc., 73 F.3d 1573, 37 USPQ2d 1365 (Fed.Cir.1996). Samsung suggests that Athletic Alternatives established a rule that when the prosecution history presents an unclear choice between a broader and narrower meaning of a claim term, then the narrower meaning controls. Accordingly, Samsung asserts that the confusing language from the prosecution history noted above requires that a narrower meaning of "plasma etching"-that which specifically excludes ion bombardment-applies. This is a misreading of Athletic Alternatives.

In Athletic Alternatives, our review of the prosecution history led to "{tjwo strong and contradictory interpretive strands.... Each strand, considered alone, leads to a coherent and distinct meaning of the disputed claim." 73 F.3d at 1580, 37 USPQ2d at 1371. That is, Athletic Alternatives did not turn on ambiguity-it turned on clarity. The inventors in Athletic Alternatives had provided, in the prosecution history, two "coherent and distinct" definitions of the relevant claim limitation. When presented with the situation where two clear yet contradictory definitions are provided by the patentee, we stated that this court would choose the narrower of the two, as such a practice would "best serve[ I" the notice function of the claim. Id. at 1581, 37 USPQ2d at 1372. Samsung appears to read Athletic Alternatives as requiring that courts choose a narrow definition of a claim limitation whenever there is a dispute over meaning and ambiguity in the intrinsic evidence. This is incorrect. The plain and ordinary meaning of claim language controls, unless that meaning ren ders the claim unclear or is overcome by a special definition that appears in the intrinsic record with reasonable clarity and precision. See, e.g., Johnson Worldwide, 175 F.3d at 990-91, 50 USPQ2d at 1610. Vagueness and inference cannot overcome an ordinary meaning of a claim term; nor can it serve to invoke the rule of Athletic Alternatives. Under Samsung’s reading, Athletic Alternatives would substitute for reasoned analysis. To the contrary, Athletic Alternatives considers the case where reasoned analysis leads to two clear and distinct definitions of claim language. It does not apply here, where confusing statements in the prosecution history simply fail to overcome the ordinary meaning of the “plasma etching” limitation.

Samsung's final argument is that extrinsic evidence supports a construction of "plasma etching" that requires the exclusion of any ion bombardment. In particular, Samsung points to an article published in August 1976, written by Gordon Poul-sen, a named inventor of the `967 patent. In that article, Mr. Poulsen described "plasma etching" as "unlike reactive ion etching," being "a strictly chemical process." Poulsen, Plasma etching in integrated circnit manufactnre-A review, J. Vac. Sci. Technol., Jan./Feb.1977, at 266-274. Samsung argues that this contemporaneous record-the Poulsen article appeared two weeks after the `967 patent's application was filed-demonstrates that the term "plasma etching" was intended by the inventors to exclude any ion bombardment.

We first note, of course, that extrinsic evidence is rarely, if ever, probative of a special and particular definition of a limitation found in a claim. This is because extrinsic evidence “may not be used to vary or contradict the claim language” as discerned from the intrinsic record. Vitronics, 90 F.3d at 1584, 39 USPQ2d at 1577. That is, claims are assumed to take on their ordinary meaning. See Johnson Worldwide, 175 F.3d at 990-91, 50 USPQ2d at 1610; Vitronics, 90 F.3d at 1582, 39 USPQ2d at 1576. That ordinary meaning may in some cases be overridden by clear definitional statements in the intrinsic record, but it may not, of course, be altered by reference to extrinsic evidence — for that would allow the public record of the patent to be changed for purposes of litigation, thus abrogating the notice function of the patent documents. See Vitronics, 90 F.3d at 1583, 39 USPQ2d at 1577 (warning against the loss of the public’s right to ascertain the patent’s scope from public documents); Southwall Techs., Inc., v. Cardinal IG Co., 54 F.3d 1570, 1576, 34 USPQ2d 1673, 1676 (Fed.Cir.1995) (same).

In any event, we find Samsung’s reliance on extrinsic evidence in this case to be unpersuasive. Poulsen’s statement that plasma etching was “unlike” reactive ion etching merely states technologic fact: plasma etching, a chemical process, is different from reactive ion etching, which uses both mechanical and chemical techniques to etch. His statement in no way, however, makes the claim that plasma etching must exclude any ion bombardment. Indeed, the Poulsen article is much like the intrinsic evidence proffered by Samsung' — it supports the argument that plasma etching is different from ion bombardment, but it does not support the view that “plasma etching”, as claimed, must preclude the presence of any ion bombardment.

Samsung also points to statements made by the inventors in the course of prosecuting a related Japanese application. There, the inventors again distinguished plasma etching from processes using ion bombardment as an etching mechanism, arguing that references disclosing sputter etching (i.e., ion bombardment) or reactive sputter etching (i.e., reactive ion etching) were not “identical” according to Japanese patent law. To the extent that statements construing terms in different claims in a different application, made to distinguish different references according to different legal standards, are relevant, we again find that they demonstrate little more than the inventors’ view that plasma etching and ion bombardment are different (i.e., not “identical”) etching mechanisms. But, as we noted before, such evidence cannot convince us that the inventors of the ’967 patent accorded “plasma etching” a meaning in claim 1 that required the exclusion of any ion bombardment.

Even if we were to find persuasive the extrinsic evidence proffered by Samsung, we would, of course, have to balance that against the strong evidence submitted by Northern Telecom. In particular, we note that Northern Telecom’s expert, Dr. Hess, explained that those of skill in the art would unquestionably recognize that ion bombardment would result from the etching process described in the ’967 specification, that the depiction of Figure 2 of the ’967 patent “makes clear that ions will bombard the surface” of the workpiece, and that the results reported in the written description indicate the presence of ion bombardment — thus supporting Northern Telecom’s argument that the specification specifically contemplates the presence of ion bombardment. Further, Dr. Hess testified that experts in the field used the term “plasma etching” to apply to both a purely chemical process, as well as the chemical portion of the combination chemical-mechanical reactive ion etching technique.

In sum, we find that Samsung’s argument in favor of a narrow meaning of “plasma etching”— one that requires the exclusion of any ion bombardment — falls short. While we agree that the record makes clear that the patentees considered plasma etching to be different from ion bombardment, we cannot agree that this mandates a finding of noninfringement. That is, if a patent requires A, and the accused device or process uses A and B, infringement will be avoided only if the patent’s definition of A excludes the possi bility of B. See, e.g., Northern Telecom, Inc. v. Datapoint Corp., 908 F.2d 931, 945, 15 USPQ2d 1321, 1332 (Fed.Cir.1990) (“The addition of features does not avoid infringement, if all the elements of the patent claims have been adopted.”); Uniroyal, Inc. v. Rudkin-Wiley Corp., 837 F.2d 1044, 1057, 5 USPQ2d 1434, 1444 (Fed.Cir.1988) (“Adding features to an accused device will not result in noninfringement if all the limitations in the claims, or equivalents thereof, are present in the accused device.”); A.B. Dick, 713 F.2d at 703, 218 USPQ at 967. Statements simply noting a distinction between A and B are thus unhelpful: what matters is not that the patent describes A and B as different, but whether, according to the patent, A and B must be mutually exclusive. While Samsung’s “plasma etching” argument has demonstrated that plasma etching and ion bombardment are indeed different techniques, it has failed to show that the ’967 patent requires, as a part of claim 1, that no ion bombardment be present.

Accordingly, we hold that the district court’s claim construction of “aluminum and aluminum oxide” as well as “plasma etching” was correct, and affirm the partial summary judgment based thereon.

IV

In order to establish a best mode violation, the party asserting invalidity must show that the asserted best mode relates directly to the claimed invention. Because the district court improperly construed the claimed invention to require fine line etching, and because the asserted best mode does not relate to a claim limitation, we reverse the court’s finding that Northern Telecom violated the best mode requirement by not disclosing the use of aluminum silicon alloy. We affirm, however, the district court’s claim construction of “aluminum and aluminum oxide” and “plasma etching,” and the partial summary judgment based thereon. We remand the case for further proceedings not inconsistent with this opinion.

COSTS

No costs.

AFFIRMED-IN-PART, REVERSED-IN-PART, AND REMANDED.

1 . Preamble language in a claim may provide an indication of how the inventor intended to "carry out" his invention. See Carl Zeiss Stiftung v. Renishaw PLC, 945 F.2d 1173, 1181-82, 20 USPQ2d 1094, 1101 (Fed.Cir.1991) (where the preamble identified the purpose of the claimed invention, the definiteness requirement of 35 U.S.C. § 112 ¶2 was met because "the claimed device [was] capable of performing its claimed purpose.” (emphasis in original)); Cf. C.R. Bard, Inc. v. M3 Systems, Inc., 157 F.3d 1340, 1350, 48 USPQ2d 1225, 1230-31 (Fed.Cir.1998) (a preamble that merely states the intended use or purpose of the invention does not necessarily limit the scope of the claim.)