The Federal Circuit’s August 1, 2025 decision in FMC Corporation v. Sharda USA, LLC offers a sharp reminder that patent drafting choices—especially between provisional and non-provisional applications—have consequences that courts will enforce. The panel vacated a preliminary injunction after concluding that the district court impermissibly narrowed claim scope by importing a “stability” limitation that appeared only in a provisional application and a related family member, but was deliberately removed from the asserted patents themselves. The takeaway is simple but unforgiving: redlining matters.
At issue were FMC’s U.S. Patent Nos. 9,107,416 and 9,596,857, directed to insecticidal and miticidal mixtures of bifenthrin and cyano-pyrethroids. FMC accused Sharda of infringement based on Sharda’s WINNER product, which combined bifenthrin and zeta-cypermethrin. The district court initially denied FMC’s request for emergency relief, but in doing so adopted a narrowing claim construction of the term “composition.” Rather than giving the term its plain and ordinary meaning, the court limited “composition” to stable compositions—excluding mixtures known to be unstable or ineffective.
That construction proved outcome-determinative. With “composition” narrowed to stable formulations, the district court rejected Sharda’s anticipation and obviousness defenses and later entered a temporary restraining order that converted into a preliminary injunction. On appeal, the Federal Circuit vacated and remanded, holding that the district court’s claim construction was legally erroneous and that its invalidity analysis applied the wrong standards.
The Stability That Wasn’t There
The district court’s construction rested heavily on disclosures in FMC’s provisional application (U.S. Provisional Application No. 60/752,979) and on a related family member, U.S. Patent No. 8,153,145. Both discussed the difficulty of achieving physical stability when combining bifenthrin with certain cyano-pyrethroids and emphasized the advantages of stable formulations. From that history, the district court concluded that “composition” in the asserted patents necessarily meant a stable composition.
The Federal Circuit disagreed—forcefully.
Judge Chen, writing for the panel, emphasized that none of the asserted patents’ specifications contained the stability language on which the district court relied. FMC had removed every reference to “stable,” “stability,” and related concepts when it drafted the non-provisional applications that matured into the ’416 and ’857 patents. Under long-standing precedent, that deletion mattered.
The court analogized the case to DDR Holdings, LLC v. Priceline.com LLC, where differences between a provisional application and a later-issued patent altered the meaning of a claim term. In DDR, the Federal Circuit held that the deletion of language relating to “services” from the final specification signaled an intentional narrowing of scope. The same principle applied here—but in reverse. By deleting stability language, FMC broadened the meaning of “composition,” and the court refused to read the deleted limitation back into the claims.
As the panel put it, a skilled artisan would view the progression from provisional to issued patent as “meaningful,” particularly where every textual hook that might support a narrower construction was deliberately removed. The district court’s approach, by contrast, impermissibly grafted a limitation onto the claims that FMC had chosen not to include.
Charles Gideon Korrell believes this aspect of the opinion may prove more consequential than the injunction dispute itself, because it underscores how prosecution edits can later constrain litigation arguments about claim scope.
Family Members Are Not Interchangeable
FMC also argued that “composition” should be construed consistently across the patent family, pointing to the ’145 patent, which retained the stability disclosures found in the provisional application. While acknowledging the general principle of consistent construction across related patents, the Federal Circuit rejected its application here.
Consistency, the court explained, gives way when a patent owner materially alters the specification of some family members in a way that signals a different intended meaning. The asserted patents and the ’145 patent may share a lineage, but they do not share identical disclosures. Where FMC chose to maintain stability language in one family member and delete it in another, courts must respect that divergence.
Charles Gideon Korrell notes that this holding is a cautionary tale for portfolio strategy: drafting variations across a family can preserve flexibility, but they can also foreclose later arguments that terms should be harmonized when litigation pressures arise.
Homogeneous Activity ≠ Stability
FMC attempted to salvage the district court’s construction by pointing to references in the asserted patents to “homogeneous” compositions and “unexpected insecticidal activity.” According to FMC, these concepts implicitly required stability.
The Federal Circuit was unpersuaded. “Homogeneous,” the court explained, does not necessarily mean “stable,” particularly where the specification contains no discussion of phase separation or long-term physical integrity. Indeed, the provisional application itself used “homogeneous” and “stability” in distinct contexts—undermining FMC’s attempt to conflate the two.
Similarly, “unexpected insecticidal activity” related to biological efficacy (how many insects were killed), not to whether a formulation physically separated over time. Treating activity as a proxy for stability, the court concluded, was analytically unsound and contradicted the intrinsic record.
Anticipation, Obviousness, and the Proper Injunction Standard
Once the claim construction fell, the rest of the preliminary injunction analysis followed. The district court had rejected Sharda’s anticipation defense largely because the McKenzie prior art reference allegedly disclosed only unstable mixtures. Under the correct construction of “composition,” that reasoning collapsed.
The Federal Circuit also reminded the district court that, at the preliminary injunction stage, an accused infringer need only raise a substantial question of invalidity—not prove invalidity outright. By effectively requiring Sharda to disprove FMC’s patents on the merits, the district court applied too demanding a standard.
The panel further rejected FMC’s attempt to rely on the “miticidal” language in the claims’ preambles to avoid anticipation. Purpose- or use-based preamble language, the court reiterated, generally does not limit claim scope unless it provides essential structure or meaning. Differences between preambles reciting “miticidal” versus “miticidal or insecticidal” did not alter that analysis.
The same errors infected the district court’s obviousness analysis. Although the court invoked “unexpected results,” it did so only after mischaracterizing the prior art under an incorrect claim construction. On remand, the district court was instructed to identify each obviousness theory and assess whether Sharda had raised a substantial question, weighing prima facie obviousness against any secondary considerations.
Charles Gideon Korrell notes that the Federal Circuit’s emphasis on the proper preliminary injunction standard reflects a broader trend: courts remain wary of granting market-shaping relief where claim scope and validity are genuinely in dispute.
Broader Implications: Drafting with the End in Mind
This decision fits comfortably within a growing body of Federal Circuit precedent emphasizing that specification changes—especially between provisional and non-provisional filings—are not clerical housekeeping. They are substantive choices that inform how skilled artisans, and courts, understand claim scope.
For patent owners, the lesson is to draft with future enforcement narratives in mind. For challengers, FMC provides a roadmap for attacking claim constructions that quietly smuggle limitations back into claims from excised disclosure.
And for litigators navigating preliminary injunctions, the case reinforces that claim construction errors at the front end can unravel the entire equitable analysis.