dissenting-in-part.
I join the court’s opinion in all respects but one. The District Court erred in finding infringement of claim 2 of the '183 patent under the doctrine of equivalents, and the court today errs in sustaining the judgment of the District Court on this issue.
Claim 2 calls for a multilayer pharmaceutical tablet comprising naproxen and a triptan wherein “substantially all of said triptan is in a first layer of said tablet and substantially all of said naproxen is in a second, separate layer.” This claim language begged the question of how much by volume of both naproxen and triptan less than 100% is “substantially all.” The District Court answered the question, construing “substantially all” without objection to mean:
“[A]t least 90%, and preferably greater than 95%, of the total triptan present in the tablet is included within one distinct layer and at least 90%, and preferably greater than 95%, of the naproxen present in the tablet is included within a second distinct layer.”
The accused products do not literally meet the “substantially all” limitation. That is why the patentee could not assert literal infringement of claim 2, and why he was forced to rely on the doctrine of equivalents to establish infringement. The actual composition of the claimed and accused tablets, with regard to the “substantially all” limitation, is shown on the following chart.
*1173DRL’s ANDA product Claimed Product Par’s ANDA product
—85% total amt. of sumatriptan Layer 1 >90% total amt. of sumatriptan, —100% total amt. of sumatriptan,
<10% total amt. of naproxen —15% total amt. of naproxen (granules)
Layer 2 >90% total amt. of naproxen, —85% total amt. of naproxen —100% total amt. of naproxen (granules),
¿10% total amt. of sumatriptan —15% total amt. of sumatriptan (sprayed on granules)
As shown in the table above, each of the Par and DRL products includes one more or less “pure” layer meeting the “substantially all” limitation as defined by the District Court. Par’s product has substantially all the required triptan in one layer along with 15% of the tablet’s naproxen, but the second Par layer has less than 90% of naproxen. So the question is whether 85% of naproxen is equivalent to 90%, when “substantially all” means “at least 90% and preferably greater than 95%.”
DRL’s product is also designed to avoid the “substantially all” limitation. It has substantially all the required naproxen in one layer, along with 15% of the triptan, but the second DRL layer has only 85% of the total triptan in the product. So, again, the question is whether 85% can be “substantially all” given the District Court’s construction of the limitation.
Pozen candidly admits in its brief to this court that the District Court never directly addressed the question of whether a layer containing 85% of a necessary ingredient is an equivalent of a layer containing at least 90% and preferably 95% of the necessary ingredient. And Pozen says it does not argue that 85% can be a numerical equivalent of 90%. Instead, Pozen argues that the District Court properly elided the numeric equivalence issue by asking only if the accused products had an “equivalent of a second layer” which could be viewed as containing substantially all of the required ingredient. Pozen and the District Court both see the “substantially all” limitation as requiring the tablet to have one more or less pure layer, and not an actual second layer but an “equivalent” second layer that could be said to be equivalent to a more or less pure layer even if it failed to contain substantially all of the required ingredient. In short, Pozen and the District Court used the notion of an equivalent layer simply to avoid answering the question of whether 85% is the equivalent of 90% or preferably 95%.
Even if the equivalent layer notion has merit, it still cannot be disconnected from the language of the claim. The equivalent layer would, in any event, have to be the equivalent of a more or less pure layer, i.e., one with at least 90% and preferably 95% of the required ingredient in it. How can a layer with only 85% of the necessary ingredient in it be an equivalent of a layer with at least 90% and preferably 95% of the required ingredient in it?
The District Court, with no contest from Pozen, recognized that “substantially all” inherently contains a range from something less than 100% up to 100%. Rather than leave the inherent equivalent range embedded in the claim language, the District Court put boundaries on the claim language. “Substantially all” means at least 90% and preferably 95%. The defendants cite our precedent, including Moore *1174U.S.A., Inc. v. Standard Register Co., 229 F.3d 1091 (Fed.Cir.2000), for the proposition that the spread from 85% to 90% is too great to be an equivalent. Pozen appreciates the force of those cases, but argues they are inapplicable here because the District Court did not answer the numeric equivalence question but instead turned the infringement decision on a flawed layer equivalence notion.
In my view, the District Court erred by not asking itself if under claim 2 a layer, viewed from the outside or from the inside, can be equivalent if is numerically nonequivalent. It cannot. The majority states that “a reasonable person could determine that a tablet layer with 85% of the agent is within the scope of the doctrine of equivalents.” Respectfully, I disagree.