This case is before us on remand from a decision of the United States Supreme Court vacating our judgment in Morton International, Inc. v. Cardinal Chemical Co., 959 F.2d 948, 22 USPQ2d 1231 (Fed.Cir.), reh’g denied, (Fed.Cir.) and reh’g in banc denied, 967 F.2d 1571, 23 USPQ2d 1362 (Fed.Cir.1992). Cardinal Chem. Co. v. Morton Int’l, Inc., — U.S. —, 113 S.Ct. 1967, 124 L.Ed.2d 1, 26 USPQ2d 1721 (1993). The case was first decided by this court on March 20, 1992, when we affirmed the December 14, 1990 judgment of the United States District Court for the District of South Carolina, No. 6:83-889-OK, holding Morton International, Inc.’s U.S. Patents 4,062,881 (’881) and 4,120,845 (’845) not infringed by Cardinal Chemical Company, and invalid. The majority of the panel concluded that the issue of patent invalidity need not be reached, since the court affirmed the district court’s finding of lack of infringement and under precedents of this court and the United States Supreme Court, the issue of validity was then moot. One judge of the panel concurred in the result, but dissented from the conclusion of mootness. The court denied a petition for reconsideration and did not accept a Suggestion for hearing in banc, although three judges of the court dissented from the decision not to hear the case in banc.
The Supreme Court granted certiorari on the question of mootness and, on May 17, 1993, reversed the decision of this court, holding that the issue of validity is not mooted when a finding of noninfringement is affirmed because “the Federal Circuit is not a court of last resort,” — U.S. at —, 113 S.Ct. at 1975, 26 USPQ2d at 1727, and because “[ejven if it may be good practice to decide no more than is necessary to determine an appeal, it is clear that the Federal Circuit had jurisdiction to review the declaratory judgment of invalidity.” Id., — U.S. at —, 113 S.Ct. at 1976, 26 USPQ2d at 1728. The Court concluded that it was an abuse of discretion not to consider validity in light of the fact that “[t]he two patents at issue here have been the subject of three separate lawsuits, and both parties have urged the Federal Circuit to resolve their ongoing dispute over the issue of validity.” Id., — U.S. at —, 113 S.Ct. at 1978, 26 USPQ2d at 1730. The Court thus remanded the case to our *1467court for consideration of the issue of validity-
Morton has suggested that, on remand, we hear the case in banc. Upon circulation of that suggestion to the full court, the court did not accept the suggestion. In addition, Morton moved for reargument. The panel denied that motion, the case having been thoroughly argued earlier. The case is now back to us on remand for redecision on the merits in light of the Supreme Court’s holding that the question of validity is not moot when a holding' of noninfringement is affirmed.
Morton contends that the district court erred in holding its patents not infringed and invalid. Cardinal contends that the district court erred in failing to award attorney fees for an exceptional case under 35 U.S.C. § 285 and that we should sanction Morton’s appeal as frivolous. We affirm the court’s judgment in all respects. We also reject Cardinal’s assertion that this appeal is frivolous.
BACKGROUND
The patents in suit are directed to a family of compounds which are used as additives or heat stabilizers for polyvinyl chloride (PVC). The heat stabilizers protect the PVC from decomposition, discoloration, and loss of strength. The particular heat stabilizers disclosed in the ’881 and ’845 patents are directed to organotin compounds, in particular to organotin mercaptoalkyl carboxylic acid ester sulfides. Only Claim 1 of each patent is at issue.1 The key limitations of the claims are (1) the bonding of the organotin compound, referred to by the parties as “partial connectivity,” wherein a tin atom is bonded to a sulfur atom, the sulfur atom is bonded to a second tin atom, and the second tin atom is bonded to a mercaptoalkyl ester (referred to by the parties as a “reverse ester”); and (2) “the organotin compound [has] an amount of tin within the range from 10 to 42% by weight and an amount of sulfur within the range from 8 to 42% by weight.”
The ’881 and ’845 patents, as noted, have had considerable exposure to the courts, as Morton has filed infringement actions in three-separate district courts against different defendants. In the first, Morton filed suit against Argus Chemical Corporation in the United States District Court for the Eastern District of Louisiana. After trial, the district court entered judgment, holding the patents invalid and not infringed. Morton Thiokol, Inc. v. Witco Chemical Corp., et al. No. 84-5685 (E.D.La., filed November 28, 1984).
On appeal, we affirmed the district court’s judgment of" non-infringement and vacated the invalidity judgment, ruling that the affir-mance made it unnecessary to review the issue of invalidity. Morton Thiokol, Inc. v. Argus Chemical Corp., 873 F.2d 1451, 11 USPQ2d 1152 (Fed.Cir.1989) (non-precedential).
In the second action, the appeal of which is now before us, Morton again filed suit, this time against Cardinal, alleging, that Cardinal sold mixtures containing the claimed compounds. After trial, the district court made lengthy findings and conclusions -and held that the claims were invalid under 35 U.S.C: § 112 because the specification was not enabling and the claims were indefinite. It further held that Cardinal’s heat stabilizer products did not infringe those claims, either literally or under the doctrine of equivalents. Morton and Cardinal appealed and cross-appealed.2
*1468DISCUSSION
I. Infringement
A finding of literal infringement requires that the asserted claims, as properly construed, read on the accused product. Envirotech Corp. v. Al George, Inc., 730 F.2d 753, 758, 221 USPQ 473, 477 (Fed.Cir.1984). The patentee has the burden of proving infringement by a preponderance of the evidence. SmithKline Diagnostics, Inc. v. Helena Labs. Corp., 859 F.2d 878, 889, 8 USPQ2d 1468, 1477 (Fed.Cir.1988). The first step in determining infringement is thus to construe the claims. Fonar Corp. v. Johnson & Johnson, 821 F.2d 627, 631, 3 USPQ2d 1109, 1112 (Fed.Cir.1987), cert. denied, 484 U.S. 1027, 108 S.Ct. 751, 98 L.Ed.2d 764 (1988). The second step is to decide whether each limitation in the properly construed claims is found, either literally or equivalently, in the allegedly infringing compounds. Id. Although claim interpretation is a question of law, subject to de novo review on appeal, the district court’s ultimate finding on infringement, as well as subordinate findings relating to proper claim construction, are issues of fact, reviewed under a clearly erroneous standard. See H.H. Robertson, Co. v. United Steel Deck, Inc., 820 F.2d 384, 389, 2 USPQ2d 1926, 1929 (Fed.Cir.1987) (interpretation of a claim may depend on evidentiary material about which there is a factual dispute); Moleculon Research Corp. v. CBS, Inc., 793 F.2d 1261, 1269-70, 229 USPQ 805, 811 (Fed.Cir.1986) (determination of infringement is a question of fact), cert. denied, 479 U.S. 1030, 107 S.Ct. 875, 93 L.Ed.2d 829 (1987).
Morton argues that the district court erred in interpreting Claim 1 and that the court’s misinterpretation resulted in errors in the court’s conclusions as to infringement and invalidity. More particularly, Morton claims that the district court misconstrued the claims by restricting the type of evidence Morton could use to show the existence of the claimed compound in Cardinal’s mixture to: (1) isolation and (2) nuclear magnetic resonance (NMR) data showing simultaneous tin-tin coupling and Carbon-13 splitting as the only method to prove the partial connectivity of Claim 1. We disagree.
On review of the record, it is clear that the district court did not construe Claim 1 to require isolation. The court merely required Morton to identify the claimed compounds in Cardinal’s mixtures to confirm their existence. The court specifically found that “[n]o amount of testing by equipment or methods available [] can identify, isolate or separate any compound claimed by the ’881 or ’845 patents.” Morton’s reliance on In re Breslow, 616 F.2d 516, 205 USPQ 221 (CCPA 1980), to show error in the district court’s finding is misplaced.
In Breslow, the compound in question could not be isolated, but was conceded to exist. The court expressly recognized that the existence of the compound was not in question. Id. at 518, 205 USPQ at 224. In the present case, however, the existence of the claimed partial connectivity in the Cardinal mixture is the essence of the dispute, and its existence has never been established. The court specifically acknowledged that under Breslow, “it is not imperative that the compounds be isolated in order to be claimed.” The court, in determining whether the' claims were infringed, used isolation as simply a possible means of such proof.
At trial, one of the ways Morton suggested one could analytically prove the existence of the claimed partial connectivity in Cardinal’s mixtures was high-resolution Tin-119 NMR. However, the court observed the conflicting testimony of Morton’s expert, Dr. Mesch, and ‘Cardinal’s NMR expert, Dr. Ellis, and found Dr. Ellis’s more credible. The court found Dr. Mesch’s identification of “at least eight peaks in a spectrum of the Cardinal product as possibly having the claimed [partial connectivity]” discredited by Dr. Ellis’s conclusions that some of Mesch’s identified peaks did not indicate the partial connectivity, but were caused by other elements present in the sample. The court determined that:
*1469even after Morton and Cardinal have each expended extraordinary sums of money on state of the art testing, these organotin components of the accused Cardinal products cannot be identified as containing the claimed connectivity as specified in the ’881 and ’845 patents.
It further determined that there is “no objective support for the actual existence of the claimed connectivities.” Therefore, the court had no reason to determine whether Cardinal’s product contained the specified weight percentages of tin and sulfur claimed in the patents at issue.
Morton also argues that its building block theory shows the existence of the claimed partial connectivity, and that the district court failed to consider it. This theory is essentially that the Cardinal stabilizers contain four moieties-butyltin, sulfide, mercapto-ethyltallate, and mercaptoethanol — and that these structures or “building blocks” can combine to form particular compounds, at least some of which contain the claimed connectivity. The compounds containing the claimed connectivity are assumed to be indicated by certain portions of the NMR spec'tra.
The record shows that the district court considered and rejected Morton’s building block theory, concluding that the theory did “not demonstrate the existence of claimed compounds in Cardinal’s product which is a complex mixture resulting from Cardinal’s process.” Even assuming Morton’s position that certain molecular fragments can be positively identified in the NMR spectra as being other compounds (non-claimed compounds and prior art stabilizers), our review of the-record shows that it is mere speculation to assume that what remains are claimed compounds, compounds that have the partial connectivity. Cardinal’s evidence, based on expert testimony from Drs. Story and Eisch, showed that by performing the reactions allegedly taught in the ’881 patent, one skilled in the art would not necessarily obtain the compounds containing the claimed partial connectivity. In fact, Dr. Story determined that a number of compounds within the claimed ranges of the tin and sulfur percentages are theoretically possible, but that they did not have the weight ranges of Claim 1. Morton’s theory still results in a number of unknown compounds remaining in the spectra. Thus, there is only a possibility of the existence of the claimed compounds in the spectra, and that was not convincing enough to the court to prove infringement.
In summary, to show infringement, Morton was required to prove that every limitation in the claims was met, including the limitation that the partial connectivity existed in Cardinal’s stabilizers.' The district court found that Morton failed to prove that Cardinal’s products contained the claimed partial connectivity. That finding was not clearly erroneous.
II. Validity
35 U.S.C. § 112 requires that an invention be described “in such full, clear, concise, and exact terms as to enable any person skilled in the art ... to make and use the same.” At trial, Cardinal had the burden of proving by clear and convincing evidence facts establishing lack of enablement of the ’881 and ’845 patents. See 35 U.S.C. § 282. We review the court’s determination of lack of enablement as a question of law. Moleculon Research Corp. v. CBS, Inc., 793 F.2d 1261, 1268, 229 USPQ 805, 811 (Fed.Cir.1986), cert. denied, 479 U.S. 1030, 107 S.Ct. 875, 93 L.Ed.2d 829 (1987). The specification purports to teach, with over fifty examples, the preparation of the claimed compounds with the required connectivity. However, the district court found that
[e]ven with the aid of sophisticated analytical instrumentation and the use of model systems which attempt to provide the compounds claimed in the ’881 patent, however, there is no evidence that such compounds exist. The clear and convincing evidence has shown that the examples of the ’881 patent do not produce the postulated compounds. Rather, the examples and procedures produce a complex mixture of alkyltin mercaptides and alkyltin sulfides. The evidence established that a number of these are prior art compounds known to be useful as heat stabilizers.
These findings are supported by the record. On review of the record, there is considera*1470ble evidénce showing that those skilled in the art could not make the claimed compounds using the procedures of the specification, and no evidence that such compounds even exist. The fifty-odd examples in the patent obviously teach something, but the evidence shows that they did not teach what was allegedly defined in the claims. Because Cardinal’s clear and convincing evidence was not rebutted by Morton, such as by showing that the examples do indeed produce the compounds containing the claimed connectivity, we see no error in the district court’s determination.
Morton argues that the court misapplied the burden of proof by requiring Morton to prove that the patents are valid. We disagree. The court correctly required Cardinal to prove by clear and convincing evidence facts establishing lack of enablement. The court found that Cardinal met its burden and that Morton failed to rebut that evidence, such as by showing that the examples do indeed produce the compounds containing the claimed connectivity. We see no error in the district court’s determination.
The district court also concluded that Cardinal had proven by clear and convincing evidence that the claims were invalid for indefiniteness under 35 U.S.C. § 112, second paragraph, which requires that the “specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.” Whether a claim is invalid for indefiniteness requires a determination whether those skilled in the art would understand what is claimed when the claim is read' in light of the specification. Orthokinetics, Inc. v. Safety Travel Chairs, Inc., 806 F.2d 1565, 1576, 1 USPQ2d 1081, 1088 (Fed.Cir.1986) (citations omitted).
The court found that the claimed compounds cannot be identified by testing and that one skilled in the art could not determine whether a given compound was within the scope of the claims. The record supports this conclusion. Since the evidence shows that the claims at issue here are not sufficiently precise to permit a potential competitor to determine whether or not he is infringing, we also agree with the district court’s determination that the claims are invalid for failure to satisfy the “definiteness” requirement of section 112, second paragraph. See Amgen, Inc. v. Chugai Pharmaceutical Co., 927 F.2d 1200, 1218, 18 USPQ2d 1016, 1030 (Fed.Cir.1991).
III. Cross-appeal
Cardinal asserts in its cross-appeal that the district court erred in failing to award attorney fees and in failing to find this case “exceptional” under 35 U.S.C. § 285, because, inter alia, Morton effectively retried a case that it earlier lost. Morton Thiokol, Inc. v. Witco Chemical Corp., et al., No. 84-5685 (E.D.La., filed November 28, 1984). Whether a case is exceptional or not is a factual determination subject to appellate review under the clearly erroneous standard. Reactive Metals and Alloys Corp. v. ESM, Inc., 769 F.2d 1578, 1583, 226 USPQ 821, 824 (Fed.Cir.1985). At the end of the trial, the district court acknowledged that “[t]his case is close”; then, in a lengthy and well-reasoned opinion, the court specifically .found that “[djespite having some serious reservations as to Morton in fact retrying the Argus case, this court does not find that this action constitutes an exceptional case under 35 U.S.C. § 285.” We find no clear error in this result.
Based on reasoning similar to its arguments on the exceptional case issue, Cardinal also asks this court to sanction Morton for a frivolous appeal. We decline to do so. Even though we agree with the district court’s finding against Morton on the question of infringement, we accept that Morton did have an argument that was new to this case, one that apparently it was not in a position to raise earlier. Thus, based on a review of the record, we determine that Morton had a basis in law and fact for filing suit and bringing this appeal. Cf. Connell v. Sears, Roebuck & Co., 722 F.2d 1542, 1154-55, 220 USPQ 193, 203-04 (Fed.Cir.1983). Thus, we affirm the district court’s denial of attorney fees and decline to impose sanctions for a frivolous appeal. We also see no merit in Cardinal’s contention that the' district court failed to consider evidence of Morton’s inequitable conduct before the Patent and Trademark Office. The record reveals that *1471the district court considered but rejected the evidence.
CONCLUSION
For the foregoing reasons, the district court’s judgment is affirmed.
COSTS
No costs.
AFFIRMED.
. Claim 1 of the ’881 patent is representative for our purposes:
1. A monoorganotin or diorganotin mercap-toalkyl ester of a carboxylic acid or mercapto hydroxyalkyl ester of a carboxylic acid mono or polysulfide useful as a stabilizer for improving the resistance to deterioration of vinyl chloride polymers when heated at 350°F, containing at least one tin ato [sic, atom] having one to two hydrocarbyl groups having from 1 to 20 carbon atoms and selected from the group consisting of alkyl, aryl, cycloalkyl, aral-kyl and alkenyl and linked to the tin through carbon, at least one mercaptoalkyl ester of a carboxylic acid group linked to tin through the sulfur of the mercaptoalkyl group and at least one mono or polysulfide sulfur group bonded exclusively to tin, the organotin compound having an amount of tin within the range from 10 to 42% by weight and an amount of sulfur within the range from 8 to 42% by weight.
. A third action, filed by Morton against Atochem North America, Inc. in the United States District Court for the District of Delaware for infringing the same Morton patents, is pending. Morton Int'l, Inc. v. Atochem North Am., Inc., No. 87-60-*1468CMW (D.Del., filed February 9, 1987). Atochem filed a brief as amicus curiae in the present appeal, urging the affirmance of the district court’s judgment of invalidity. The American Intellectual Property Law Association also filed a brief as amicus curiae.