Summary
finding circularity of proposed construction a "compelling reason" to reject it
Summary of this case from Lecat's Ventriloscope v. MT Tool & Mfg.Opinion
No. 96-1526
DECIDED: June 10, 1997 Rehearing Denied; Suggestion for Rehearing En Banc Declined August 13, 1997.
Appealed from: U.S. District Court for the Northern District of California, Senior Judge Orrick, Jr.
Henry C. Bunsow, Keker Van Nest, of San Francisco, CA, argued for plaintiff-appellee. With him on the brief was James M. Emery and Denise M. DeMory.
Roger L. Cook, Townsend and Townsend and Crew LLP, of San Francisco, CA, argued for defendant-appellant. With him on the brief was William J. Bohler.
Before ARCHER, Chief Judge, NEWMAN, and BRYSON, Circuit Judges.
IXYS Corporation appeals from a district court's ruling on summary judgment that IXYS infringed U.S. Patent No. 4,364,073 (the '073 patent), owned by Harris Corporation, and that the patent is not invalid. A summary judgment may be upheld only if there are no material issues of fact in dispute and the judgment is not premised on any error of law. Because the district court erred in construing the claims of the '073 patent and in concluding that the patent satisfies the enablement requirement of 35 U.S.C. § 112, we reverse the grant of summary judgment on infringement and enablement, vacate the summary judgment on the other issues of patent validity, and remand the case to the district court.
I. A.
Harris and IXYS manufacture semiconductor devices called IGBTs (insulated-gate bipolar transistors). An IGBT is used to control the flow of electric power. It contains four regions of semiconductor material of alternate conductivity type, and it has three external terminals. The device controls the flow of power in response to a signal applied to one of its terminals, called the gate. The presence of an appropriate gate signal turns the device on and allows electric current to flow through it; removing the gate signal turns the device off, blocking the flow of current.
A thyristor is a semiconductor device similar to an IGBT. Like an IGBT, a thyristor contains four regions of semiconductor material of alternate conductivity type and has three external terminals, one of which is the gate. As in the IGBT, applying an appropriate gate signal turns the thyristor on, allowing the flow of electric current through it. Unlike an IGBT, however, removing the gate signal from a thyristor does not shut off the flow of electric current through the device. Once a thyristor turns on in response to the application of a gate signal, it cannot be turned off simply by removing the gate signal. The thyristor thus exhibits "latching" behavior: In response to the application of an appropriate gate signal, the device turns on and remains on even if the gate control signal is removed. Turning a thyristor off typically requires reduction of the current flowing through the device below a threshold level.
The latching property of the thyristor arises from the structure of the device. The four alternating semiconductor regions in a thyristor inherently incorporate two three-layer combinations, each of which has a forward current gain, denoted as (1 and (2, respectively. It is well known that a thyristor will not latch if the sum of (1 and (2 is less than one, as is recited in a 1967 prior art reference of record in this case. See F.E. Gentry, Four Layer Semiconductor Switch with the Third Layer Defining a Continuous, Uninterrupted Internal Junction, U.S. Pat. No. 3,324,359 (issued June 6, 1967) at col. 3, ll. 62-67 (Gentry).
A MOSFET (metal-oxide-semiconductor field-effect transistor) is also a three-terminal device that is used to control the flow of electric power. Unlike IGBTs and thyristors, however, MOSFETs have only three semiconductor regions. A MOSFET controls the flow of power through the device in response to an appropriate control signal applied to its gate terminal. MOSFETs are similar to IGBTs in that they can be used to control the flow of electric power by selectively applying and removing an appropriate gate signal. MOSFETs do not exhibit the "latching" behavior of thyristors, but thyristors can typically carry larger amounts of electric power.
IGBTs combine the controllability of a MOSFET with the high-power-carrying capabilities of a thyristor. Because they incorporate a four-layer structure similar to a thyristor, however, IGBTs incorporate two three-layer combinations of regions of alternate conductivity and therefore exhibit latching if subjected to certain electrical conditions, such as high voltages.
B.
The '073 patent has 17 claims. Claim 1, the only independent claim, recites an IGBT-like structure. It provides (emphasis added):
A vertical MOSFET device, comprising:
a semiconductor substrate, including in series, adjacent source, body, drain and anode regions of alternate conductivity type;
the body region being adjacent to a surface of the substrate;
the source and drain regions being spaced so as to define a channel portion in the body region at said surface;
the source, body and drain regions having a first forward current gain (1 and the anode, drain and body regions having a second forward current gain (2, such that the sum (1 + (2 is less than unity, and no thyristor action occurs under any device operating conditions.
The dispute in this case centers on the emphasized portion of the claim.
In addition to Gentry, the prior art of record includes a U.S. patent issued in 1980 to Plummer. See James D. Plummer, Monolithic Semiconductor Switching Device, U.S. Pat. No. 4,199,774 (issued Apr. 22, 1980) (Plummer). Plummer discloses a four-layer semiconductor device with a structure identical to that of the device recited in claim 1 of the '073 patent. Plummer's device operates as a transistor at lower voltages and currents (i.e., it exhibits no latching behavior), but beyond a certain threshold of voltage or current it acts as a thyristor (i.e., it latches).
C.
In 1994, Harris brought this action against IXYS, alleging infringement of the '073 patent. IXYS moved for summary judgment of non-infringement and patent invalidity. Harris cross-moved for summary judgment, arguing that the IXYS devices literally infringe claim 1 of the '073 patent, and that the patent is not invalid.
The district court denied IXYS's motion for summary judgment and, adopting Harris's claim construction, granted Harris's motion. The court ruled as a matter of law that IXYS's IGBTs infringe the '073 patent, and that the '073 patent is not invalid because of indefiniteness, obviousness, anticipation, or non-enablement.
II.
On appeal, the parties debate the proper construction of claim 1 of the '073 patent. Specifically, they disagree as to the meaning of the last clause of the claim, which recites "such that the sum (1 + (2 is less than unity, and no thyristor action occurs under any device operating conditions." IXYS construes that clause as covering only four-layer devices that, because of their structure, never act as thyristors (i.e., devices in which the sum (1 + (2 is less than one under all circumstances). Harris, on the other hand, construes the disputed clause as covering four-layer devices that have a transistor (or non-latching) mode of operation in addition to a threshold point beyond which the devices act like thyristors, as long as the devices were intended to be operated below the thyristor threshold. We agree with IXYS.
A.
With respect to the first part of the disputed clause, the phrase "such that the sum (1 + (2 is less than unity" merely restates a basic characteristic of four-layer semiconductor devices that is well known to any electrical engineering student and is recited in the 1967 Gentry reference. A four-layer device functions as a transistor whenever the sum of (1 and (2 is less than one, and as a thyristor whenever the sum of (1 and (2 is greater than one, as the specification of the '073 patent acknowledges. That part of the clause therefore does nothing to limit the scope of the claim language. Moreover, the second part of the clause, which refers to "device operating conditions," includes within the scope of the claim "any" device operating conditions, without qualification. Nothing in the language of the claim suggests that it should be construed as limited to conditions under which the device was "intended" to operate in order to ensure that it would not latch.
A further, compelling reason for rejecting Harris's proposed construction of the "any device operating conditions" limitation is that Harris's construction would make the limitation entirely circular. To say that the invention of claim 1 does not latch when the sum of (1 and (2 is less than one is to say only that the device does not latch when the sum of the forward current gains is not great enough to cause latching. The circularity of Harris's proposed construction is revealed by a passage in its brief in which it explains that the devices described in the `073 patent "will not latch `under any device operating conditions,' i.e., so long as they are properly functioning as they were designed to operate." But Harris further argues that "the patented devices [are] `operable' so long as the devices continue to operate as a transistor and are latch-free." In plain terms, Harris's claim construction argument is that the intended "operating conditions" of the patented device do not include conditions that cause the device to latch, and therefore the device does not latch under its intended "operating conditions."
We decline to assume, without a stronger textual basis, that the "any device operating conditions" limitation was intended simply to convey that the claimed device does not function as a thyristor as long as it is not operated under conditions that trigger thyristor action. That construction would contribute nothing but meaningless verbiage to the definition of the claimed invention. Nor can we accept Harris's argument that when an IGBT latches "under extreme adverse non-operating conditions," it has failed and therefore cannot be regarded as functioning under "device operating conditions." As the record before the district court makes clear, latching may lead to device failure, but does not necessarily do so. Two technical articles that were before the district court, one of them authored by an inventor of the '073 patent, say as much. It is therefore incorrect to suggest, as Harris does, that a latched state cannot be an "operating condition" for an IGBT.
An additional problem with Harris's proposed construction of claim 1 is that it would cause claim 1 to read on the prior art described in the Plummer patent. Harris attempts to distinguish Plummer's device on the ground that it has a low thyristor threshold and is intended to operate as a thyristor. The inventors of the '073 patent, according to Harris, came up with the novel idea of using the four-layer device in the transistor range, rather than in the thyristor range. Yet despite Harris's assertion that its device has a higher thyristor threshold than Plummer's device as a result of changing the geometries and conductivities of the device, the language of claim 1 of the '073 patent does not reflect that structural distinction.
Moreover, the fact that the inventors recognized that the Plummer device could be used as a transistor rather than as a thyristor would not be sufficient to avoid the effect of the prior art. It is well established that the discovery of a new property or use of an old product does not render the old product patentable. See In re Spada, 911 F.2d 705, 708, 15 USPQ2d 1655, 1657 (Fed. Cir. 1990); Titanium Metals Corp. of Am. v. Banner, 778 F.2d 775, 780-82, 227 USPQ2d 773, 777-78 (Fed. Cir. 1985). Thus, because claims should be read in a way that avoids ensnaring prior art if it is possible to do so, see Modine Mfg. Co. v. United States Int'l Trade Comm'n, 75 F.3d 1545, 1557, 37 USPQ2d 1609, 1617 (Fed. Cir.), cert. denied, 116 S.Ct. 2523 (1996); Decca, Ltd. v. United States, 544 F.2d 1070, 1080, 191 USPQ 439, 447 (Ct.Cl. 1976), the effect of the Plummer prior art device is to render Harris's proposed construction of claim 1 much less plausible than IXYS's competing construction.
B.
The specification of the '073 patent provides further support for IXYS's proposed construction of the claim as reciting a device that does not latch under any conditions. The summary of the invention characterizes the invention as the product of manipulating "conductivities and geometries . . . so that the device functions substantially as a [non-latching] transistor." '073 patent at col. 1, ll. 58-60. Later, in the course of describing the claimed invention, the specification explains that although the invention "is a four layer device, it does not function as a thyristor having regenerative [i.e., latching] properties." Id. at col. 3, ll. 53-55.
C.
The prosecution history of the '073 patent also provides substantial support for IXYS's construction of the disputed language of claim 1. The applicants repeatedly argued to the examiner that their claimed invention was structurally distinct from the prior art and that the structure of the claimed device prevented it from latching. When the examiner expressed concern that the applicants were merely seeking a patent on the Plummer device operating at voltage and current levels below its thyristor threshold, the applicants amended claim 1 to specify that the device would not latch under "any" operating conditions. In light of the series of exchanges between the applicants and the examiner, Harris cannot persuasively argue that the "any operating conditions" language of claim 1 should be read to include only those conditions under which the designer intended the claimed device to operate.
In the original application, the last clause of claim 1 recited "the source, body and drain regions having a first forward current gain and the anode, drain and body regions having a second forward current gain, such that the sum of the current gains is less than unity." The examiner initially rejected all of the claims as anticipated by Plummer and as obvious over Plummer in view of Gentry. In response, the applicants attempted to distinguish their claimed invention from Plummer's device by arguing that their "invention is structured so as not to operate as [a thyristor]" and that their "invention operates as a transistor at all times" (emphasis added). The applicants also argued that, unlike Gentry, "the sum of the current gains ((1 + (2) in Applicants' invention is less than 1 at all times," and they represented to the examiner that, because of its structure, their invention "operates as a transistor at all times . . . and this is true at both low and high voltage operating conditions." In responding to the examiner's obviousness rejection, the applicants asserted that "[t]here is no suggestion [in the prior art] of a four layer transistor structure such as [the applicants' structure], wherein the sum (1 + (2 is maintained less than 1 for all operating voltages" (emphasis added).
In the second office action, the examiner again rejected all of the claims as anticipated by Plummer and as obvious over Plummer in view of Gentry. The examiner characterized the applicants' invention as a "device configured to act as a transistor at all times, that is, so the sum [(1 + (2] is less than unity at all intended operating voltages. This [limitation], however, is not in the claim." The applicants responded to the second office action by amending the last clause of claim 1 so that it recited "the source, body and drain regions having a first forward current gain (1 and the anode, drain and body regions having a second forward current gain (2, such that the sum (1 + (2 is less than unity under device operating conditions, and no thyristor action occurs." In remarks accompanying their amendment, the applicants stated that they had amended claim 1 "so as to definitively specify that the structure of the present invention behaves at all times as a transistor and not as [a thyristor]" (emphasis added). The applicants characterized Plummer's device as functioning as a thyristor "[b]eyond some threshold voltage," and argued that in contrast to Plummer's invention, their device was "structured such that it will never operate as [a thyristor]" (emphasis added). They contended that their device "does not have a threshold point above which it will change from a transistor to [a thyristor]" and that it "is not merely [a thyristor] operating in the [pre-latching] region." They further noted that they had amended the claim "to further emphasize this structural distinction." The examiner again rejected all of the claims. The examiner pointed out that for the Plummer device, under certain voltage and current levels, the sum of the forward gains is less than one, "so that no thyristor action occurs under those operating conditions." The language of claim 1, the examiner explained, "does not appear to preclude the existence of device operating conditions under which thyristor action occurs, it appears only to state that there are device operating conditions under which thyristor action does not occur." During a subsequent interview, the examiner suggested to the applicants that "language such as `no thyristor action occurs under any device operating conditions' would appear to overcome the rejection" of claim 1. The applicants then adopted that language, which is the current text of the claim. In remarks accompanying the amendment, the applicants stated that "[t]he present language of Claim 1 now precludes the existence of device operating conditions under which thyristor action occurs." The examiner subsequently allowed the claims, and the '073 patent issued.
Harris argues that the prosecution history of the '073 patent supports the proposition "that the '073 patent covers devices that . . . remain latch-free while they are functioning effectively as they were designed to operate." The prosecution history, however, does not support Harris's argument. As the excerpts quoted above illustrate, the applicants repeatedly sought to distinguish their claimed device from Plummer's device on the basis of their device's structure, not its intended operating conditions. The applicants stated to the examiner in their responses to the office actions that, because of its distinct structure, their claimed invention would not operate as a thyristor. In particular, the applicants repeatedly represented to the examiner that the claimed device "never" acts as a thyristor, that the device "does not have a [thyristor] threshold point," and that the device operates as a transistor "at all times." Nothing in the applicants' submissions to the examiner suggested an intention to qualify the terms "never" and "at all times" so as to limit those terms to the particular conditions under which the designer intended the device to operate. Instead, the applicants represented to the examiner that, because of structural differences between the claimed invention and the prior art, their device would "never" act as a thyristor, and we construe the claim consistently with those representations.
To be sure, as Harris points out, during the second office action the examiner characterized the applicants' claimed invention as a device in which the sum of (1 and (2 is "less than unity at all intended operating voltages." In light of the overall prosecution history of the `073 patent, however, that single remark is not sufficient to justify importing a qualification into the plain language of the claim, especially since the remark was not in a statement made by the applicants, but rather appeared in the examiner's characterization of the applicants' claimed invention. Moreover, it was long after the examiner's remark that the applicants substituted the final claim language ("no thyristor action occurs under any device operating conditions") for the earlier version ("under device operating conditions . . . no thyristor action occurs"), a change that emphasizes that thyristor action will not occur at all, rather than that it will not occur under certain conditions. We therefore construe claim 1 as reciting a four-layer IGBT device in which latching does not occur.
III.
Once claim 1 is construed, the issue of infringement is straightforward. Harris acknowledges in its brief that under certain conditions the accused IXYS devices exhibit latching behavior. To be sure, as IXYS's evidence in the district court made clear, IXYS takes steps to avoid latching in its devices, such as rating its products for levels of current, voltage, temperature, and switching speed at which latching is unlikely to occur, or attaching external circuits that prevent the IGBTs from being exposed to conditions that would cause latching. But the fact that IXYS attempts to avoid latching does not mean that "no thyristor action occurs" in the IXYS devices "under any device operating conditions," as is required by claim 1 of the '073 patent. Moreover, as we have noted, latching does not necessarily lead to device failure in IGBTs, even though latching may defeat the purposes for which the device was designed and sold. Because we construe "any device operating conditions" not to be limited to those conditions under which a particular device was intended to function, the district court was incorrect in concluding, as a matter of law, that IXYS's devices literally infringe claim 1 of the '073 patent. The district court therefore erred in granting summary judgment to Harris on that issue.
IV.
We next turn to the question of the validity of the '073 patent. The district court held that the '073 patent is not invalid. With respect to enablement, IXYS contends that the issue was not presented to the district court in the parties' motions, while Harris argues that the district court did not err by including the enablement issue in the judgment. In any event, there is no doubt that the district court ruled on the issue. The court explicitly held that "[t]he '073 patent claims and specification also comply with paragraph one of Section(s) 112 because the detailed description of how the invention should be assembled enables someone skilled in the art to build the device. Therefore IXYS's claim that the patent is invalid for lack of enablement is denied." We interpret the district court's order as granting summary judgment to Harris on the issue of enablement, and we reverse the district court's judgment on that issue.
To constitute an enabling disclosure, "the specification of a patent must teach those skilled in the art how to make and use the full scope of the claimed invention without `undue experimentation.'" In re Wright, 999 F.2d 1557, 1561, 27 USPQ2d 1510, 1513 (Fed. Cir. 1993). The specification of the '073 patent, however, contains no discussion of how to make or use the claimed invention so that it will exhibit the characteristics set forth in the last clause of claim 1.
The specification recognizes that latching behavior is present in four-layer semiconductor devices and that the claimed invention is a four-layer device. The latching problem can be solved, the specification states, by manipulating "the conductivities and geometries of the four semiconductor regions . . . so as not to form a . . . thyristor." '073 patent at col. 1, ll. 41-44; see also id. at col. 1, ll. 57-59 ("The conductivities and configurations of the body and drain regions are adjusted so that the device functions substantially as a . . . transistor."); id. at col. 3, ll. 49-55 (advancing similar propositions). Neither that passage nor anything else in the specification, however, sets forth how the "conductivities and geometries of the four semiconductor regions" can be manipulated so that the claimed device will not exhibit thyristor action.
The specification's reference to manipulating the device conductivities and geometries merely restates broad general principles of semiconductor device design well known in the prior art. Designing new semiconductor devices invariably involves manipulating device geometries and conductivities. In fact, a 1978 prior art reference that was before the district court teaches changing conductivities and "varying the geometrical layout" of the structure to create new semiconductor devices. Likewise, the inventors' reference to the relationship between the sum of forward current gains in a four-layer device and the onset of thyristor action merely recites a principle well known in the art. Thus, because the applicants' disclosure does not teach a person having ordinary skill in the art how to make an IGBT that acts as a transistor "at all times," the most we can credit them with is having predicted — rather than invented — such a device. We therefore conclude that the district court erred in granting summary judgment in Harris's favor on the issue of enablement.
The district court's rulings on the issues of anticipation, obviousness, and indefiniteness — as well as IXYS's challenges to those rulings — are all premised on the district court's claim construction, which we have rejected. In light of our discussion of enablement, however, the district court may find it unnecessary to revisit those issues. We therefore do not address the merits of IXYS's arguments on the issues of anticipation, obviousness, and indefiniteness, but instead vacate the summary judgment on those issues and remand for further proceedings consistent with this opinion.
REVERSED-IN-PART, VACATED-IN-PART, AND REMANDED.