Trustees of Boston University v. Everlight
Trustees of Boston University v. Everlight Electronics Co., Ltd., et al.
2016-2576, 2016-2577, 2016-2578, 2016-2579, 2016-2580,
2016-2581, 2016-2582, 2016-2591, 2016-2592, 2016-2593,
Decided: July 25, 2018
If a patent claim is construed to cover several different embodiments and the patent’s specification fails to teach how to make one of them, does the claim meet §112’s enablement requirement? In a case involving patented light emitting diodes, the Court of Appeals for the Federal Circuit held that it does not.
LEDs are typically made of multiple layers of semiconductor materials. These materials may be mono- or polycrystalline, in which their constituent atoms are arranged in one or more highly ordered lattice structures; or amorphous, in which their constituent atoms lack the long-range structure of a crystal.
The Trustees of Boston University sued LED maker Everlight Electronics for infringing US patent 5,686,738, which claims a semiconductor structure that includes (1) a crystalline substrate layer; (2) a "non-single crystalline buffer layer" grown on the substrate; and (3) a "growth layer" made of doped gallium nitride grown on the buffer layer.
BU asked the trial court to construe the phrase "non-single crystalline buffer layer" to cover both polycrystalline and amorphous buffer layers. The District Court agreed; the result of this (and other claim construction rulings) was that the claim language covered several different embodiments, including one—in which a monocrystalline gallium nitride "growth layer" is grown directly on an amorphous "buffer layer"—that was the focus of the parties’ enablement dispute.
At trial, defendant Everlight argued that at the time the patent was filed, no one had ever grown a monocrystalline semiconductor layer directly over a layer of amorphous semiconductor; and because the patent specification fails to teach how to do this, the claim is invalid for failing to meet §112’s enablement requirement. The District Court disagreed, concluding that the patent did not have to enable an embodiment with a monocrystalline growth layer formed directly on an amorphous buffer layer, so long as it enabled an embodiment—also covered by the claims as the District Court had construed them—with an (unclaimed) intermediate layer between the amorphous buffer layer and the monocrystalline growth layer.
At the Federal Circuit, BU presented two main arguments in favor of affirmance: (1) enablement was not an issue, because researchers had recently shown that it is possible to grow a crystalline layer directly on top of an amorphous layer; and (2) under the District Court's claim construction—which covers 6 different embodiments—the other five were undisputedly enabled, which is "sufficient" for §112 purposes.
The Court of Appeals rejected both arguments. Regarding the first, the panel said, “the inquiry is not whether it was, or is [now], possible to make the full scope of the claimed device…. The inquiry is whether the patent’s specification taught one of skill in the art how to make such a device without undue experimentation as of the patent’s effective filing date”—which the ‘738 patent plainly failed to do.
Regarding BU’s argument that a specification enabling most of the embodiments covered by a claim is "sufficient" for §112 purposes, the Federal Circuit said that its precedent requires a patent’s specification to enable the full scope of the claimed invention. And although an artisan’s knowledge of the prior art and routine experimentation may fill gaps in a patent disclosure, this gap filling cannot substitute for a complete lack of disclosure as to one or more embodiments.
The court concluded by noting that BU created its own enablement problems by seeking (and getting) a construction that was far broader than its patent’s teachings.