Before FRIEDMAN, DAVIS and NIES, Circuit Judges.

DECISION

NIES, Circuit Judge.

Technicon Instruments Corp. (TIC) appeals from the judgment of the United States District Court for the District of Oregon holding United States Patent No. 3,804,593 ('593 patent) unenforceable because of inequitable conduct during prosecution of the patent, awarding Alpkem Corp. costs and attorneys' fees based upon that inequitable conduct,1 and further finding that Alpkem's system did not fall within the scope of asserted claims 1-4 and 6-13 literally or under the doctrine of equivalents.

Having reviewed the arguments of the parties and the stipulation they submitted to this court, we conclude:

1. The district court's finding that Alpkem's device does not fall within the scope of the asserted claims is affirmed.

2. The district court's finding that TIC engaged in inequitable conduct during prosecution of the '593 patent is clearly erroneous. Therefore, we reverse that finding and vacate the award of costs and attorney's fees based thereon. We also vacate the judgment insofar as it declares the '593 patent unenforceable.

3. In view of our disposition of the appeal on the issue of noninfringement, we do not address the patent validity issues.

Thus, we affirm-in-part, reverse-in-part, and vacate-in-part.

OPINION

* The '593 patent discloses and claims an improved method and apparatus for automatic, quantitative analysis of a stream of separate, discrete, liquid samples (e.g., blood from different sources) with respect to known substances therein. Such a procedure is known as continuous flow analysis (CFA).

The prior art disclosed what is known as a "wetted" CFA system. In a wetted system, liquid adheres to the conduit wall which reduces friction and creates smooth flow, but causes contamination between samples. Numerous segments of one sample, created by intrasample bubbles which perform a cleaning function, as well as a wash between samples, were required to remove contamination caused by the residue liquid on the wall.2 A debubbler was then used to eliminate the intrasample bubbles, thereby creating a larger segment of the sample before it entered the flow cell for analysis. Otherwise, the intrasample bubbles caused an incorrect reading. Debubbling causes intermix not only between segments from one sample, but also between samples. As the stream moves and the debubbler removes intersample bubbles, the last segment of one sample mixes with and contaminates the first segment of the next sample. A number of segments of the same sample, therefore, had to be measured until uncontaminated segments were being read as indicated by a "steady state" of measurement, at which point the clinicians would accept the analysis. The signal recorded in this system is an "S" curve.

In the patented system, a single leading segment of sample is introduced into the stream to perform a cleaning function, followed by a bubble, followed by a single trailing segment for analysis. Less cleaning is necessary in the system because the conduits are made of non-wetting material, such as Teflon.3 The non-Teflon flow cell, however, requires cleaning. By making the volume of the analysis segment at least equal to the volume of the slight passageway of the flow cell where the measurement is taken, debubbling between the two segments is not necessary and, of course, is undesirable. Contamination between samples is avoided by maintaining the intersample bubbles in the stream as it moves through the flow cell. The recorder is deactivated when bubbles are in the flow cell. The signal is a square wave form because only one segment is measured.

Alpkem makes the accused RFA-300 CFA analyzer. That device deliberately wets the conduit by mixing surfactant agents into the sample and, like the prior art, must pass numerous segments of the same sample (usually 30) to clean the conduit (which is not made of non-wetting material) as well as the flow cell before reaching a steady state for measurement. Unlike the prior art, it does not need a debubbler to remove intrasample bubbles, however, because each segment has a volume equal to the volume of the flow cell. Nor is a debubbler required to remove intersample bubbles because the bubbles are so small that they do not affect the recorder. Each segment is measured and the signal is similar to an "S" curve.

The district court interpreted the '593 patent claims as limited to a nonwetted CFA system in which neither the liquid samples nor the segmentizing media adhere to the conduit walls and cause contamination.4 It also found that the claims contemplate measurement of a single segment of a sample. Because the RFA-300 device is a wetted system and measures multiple segments of a sample, the court found no infringement either literally or under the doctrine of equivalents.

TIC argues that its invention as claimed is either a non-wetting or wetting system and that the elimination of debubbling is the principal advance over the prior art. It urges that the Alpkem CFA system appropriates all elements of the invention even though Alpkem's system is a wetted system.

II

TIC charges that the district court had a fundamental misunderstanding of patent law leading to an avalanche of errors in its infringement analysis. It asserts, for example, that the trial judge initially found literal infringement of a dependent claim, but not the underlying independent claim. Clearly, that is an incorrect interpretation of the court's analysis. The court initially stated that the claim language "read on" the accused device but that, when properly interpreted, it did not.

TIC asserts that the court limited the scope of the claims to the disclosed, preferred, non-wetted embodiment, by relying on a few lines of the specification which set forth the inventor's preferred use of conduits having nonwetting surfaces. TIC points out that the claims themselves do not mention "nonwetting" conduits, avoidance of a "liquid film," the absence of "intrasample" bubbles, or any other differences relied upon by the court.

Contrary to TIC's argument, the trial court did not simply read "nonwetting" into the claims from the specification. The court did note--properly--that the specification (and not merely in the description of the preferred embodiment) specifically describes a nonwetted system, i.e., a system in which the conduits are not wetted by either the sample segments or the segmentizing fluid. The specification of the '593 patent specifically states:

The primary object of the present invention is to improve the precision of quantitative analysis of the samples of liquid. We have discovered that this important object can be accomplished by utilizing as much as possible liquid conduits, such as Teflon tubing, which have non-wetting surfaces instead of wettable surfaces, and by washing the flow cell between the passage of successive samples therethrough with one or more bubbles of air or other gas which is inert to the liquid transmitted through the conduits and through the flow cell, whereby contamination of one sample by another is prevented or is negligible.

The prosecution history shows that TIC originally claimed "gas," which is compatible with the nonwetting system described in its application, as the segmentizing medium between samples and segments of a sample. It then attempted to add claims which specified the broader term "fluid," encompassing both liquids and gases, as the segmentizing medium. The examiner rejected such broad claims on the ground that the disclosure was nonenabling as to liquids, only gas being disclosed as the segmentizing medium. On appeal to the board, TIC argued that the prior art (namely, the Kessler '367 patent) showed both liquid and gas were known segmenting media and that equivalency between liquid and gas is "evident." The board, altering the ground for rejection, refused allowance because "fluids" would include wetting liquids which would be inoperative. TIC did not respond to the board's new ground for rejection; rather, it appealed the rejection to the Court of Customs and Patent Appeals, arguing:

[O]ne would not employ, as a segmentizing medium, any material which would adhere to the walls of the conduits or sight passageway of the flow cell. Such adherence would tend to cause contamination between successive liquid samples.

Brief for Appellants, In re Smyth, Patent Appeal Docket No. 8855, at 15-16 (July 28, 1972). The court held that the added claims were allowable, stating:

The use here of any particular "liquids" which would be inoperative, such as the examples given by the board--"colored materials," materials "adherent to the walls of the sight tube," and "liquid wetting agents"--would be predictably inoperative in the invention and thus would never be selected by one skilled in the art.

In re Smythe, 480 F.2d 1376, 1385, 178 USPQ 279, 286 (CCPA 1973).

In light of the arguments made during prosecution, including the appeal to the Court of Customs and Patent Appeals, TIC's attempt to broaden its claims to include segmentizing fluids which wet the conduit must be rejected. A party may not advance one interpretation of the claims to uphold their validity during prosecution and advance a conflicting interpretation to determine infringement. See, e.g., Mannesmann Demag Corp. v. Engineered Metal Prods. Co., 793 F.2d 1279, 1285, 230 USPQ 45, 49 (Fed.Cir.1986). Thus, we agree that the claims require avoidance of wetting (to the extent possible). As TIC asserted, wetting fluids would be inoperable given the need for greater cleaning by numerous segments.

In that regard, we agree with the court that certain amendments filed by TIC limited all of the claims with respect to the number of segments of a sample which should be passed through the system for purposes of cleaning or analysis. The claims permit division of a segment or the taking of two segments from one sample container. TIC filed amendments which specified that segments could be taken "repeatedly" from one sample container. The examiner refused to allow those amendments to include "repeatedly" (or comparable expressions), because the specification had no support for more than two offtakings from the sample container. We agree that the specification supports no more than two offtakings. Thus, the claims must be limited for validity purposes to taking no more than two segments. See Loctite Corp. v. Ultraseal Ltd., 781 F.2d 861, 867, 228 USPQ 90, 93 (Fed.Cir.1985) (to ascertain meaning of claim language, resort should be made to the prosecution history--which includes claim amendments). Further, having amended its claims to delete the objectionable material, TIC is estopped in analyzing infringement under the doctrine of equivalents from asserting more than two offtakings. See Hughes Aircraft Co. v. United States, 717 F.2d 1351, 1363, 219 USPQ 473, 481 (Fed.Cir.1983) (effect of an amendment as prosecution history estoppel limiting doctrine of equivalents may be great).

Although TIC made no attempt to broaden claim 12 concerning division of a sample, the specification states that more than one cleaning segment is "unnecessary." Thus, claim 12 may also be interpreted as "dividing" a segment only into two (one for cleaning, one for analysis), not multiple, sub-segments. See Loctite Corp., 781 F.2d at 867, 228 USPQ at 93 (to ascertain meaning of claim language to determine literal infringement, resort should be made to specification). That it may be done twice (i.e., claims 11 and 12 combined) does not expand the claim to encompass a wetted system.

Undoubtedly, the trial court gave a restricted meaning to the words of the claims, e.g., "inert fluid" must be non-wetting and "sample" means a "single sample segment." This restricted reading is entirely appropriate, however, in view of the prosecution history, the enablement provided in the specification, and the description of the invention therein. So interpreted, the claims do not encompass Alpkem's wetted system literally. Further, the prosecution history rises to the level of estoppel so as to prevent a finding of infringement by Alpkem's wetted system under the doctrine of equivalents.

III

Alpkem asserts that TIC was guilty of inequitable conduct in obtaining a reversal of the examiner's rejection of the inert fluid amendments. The district court agreed. Having reviewed the evidence, we are left with a firm conviction that the district court erred in finding that TIC's conduct met the required threshold levels of materiality and intent. See J.P. Stevens & Co. v. Lex Tex Ltd., 747 F.2d 1553, 1559-60, 223 USPQ 1809, 1092 (Fed.Cir.1984) (inequitable conduct requires proof by clear and convincing evidence of a threshold degree of materiality of the nondisclosed information and of a threshold intent), cert. denied, 474 U.S. 822 (1985).

Even if one infers that TIC should have known that the liquids disclosed in the Kessler patent would wet Teflon conduits, TIC cited the Kessler reference for a different purpose, namely, to support the more general proposition that a person of ordinary skill in the art would have known both liquid and gas could serve as a segmentizing medium. That general proposition is true and TIC did not represent that the specific liquids disclosed in Kessler would work in the invention as claimed. TIC may have been in error in believing that there were segmentizing liquids which would wet the conduits no more than the sample itself, but we fail to discern any evidence that the error resulted from gross levels of culpable intent nor materiality were crossed. Materiality and scienter are questions of fact, id. at 1562, 223 USPQ at 1094, and we hold the trial court's findings on these questions to be clearly erroneous.

1 After filing its cross-appeal from the court's refusal to include expert witness fees in that award, No. 87-1288, Alpkem has stipulated to withdrawal of that cross-appeal

2 A "sample" is defined as the liquid material, usually blood, disposed in an individual container. Each sample may be separated into multiple segments by an "intrasample" medium, i.e., bubbles of gas. "Intersample" bubbles separate one sample from the next

3 We note that a completely nonwetted CFA system is impossible as a practical matter. Teflon is not a completely non-wetting surface. By "nonwetted," therefore, we mean a CFA system in which there is insufficient adherence of sample and segmenting medium to the conduit wall so as to cause problematic contamination

4 Before the district court, TIC sought to have "nonwetting" restricted to the sample, not the segmenting fluid. The court rejected that argument, finding the restriction in "direct contradiction to [the] assertions made by Technicon in its 1972 brief to the CCPA." Slip op. at 42. We agree