The limitation that matters in US20260215351A1 is the last clause of claim 1. The application, published 23 July 2026, assigned to Samsung Electronics Co., Ltd. and naming Dongsik Shim as sole inventor, recites twenty claims across three independents. Every one of those independents ends the same way: the pre-bonding comprises evaporating the liquid in a vacuum.

A method comprising: pre-bonding, by a liquid, a wafer to a chip, while the chip is attached to a jig; and detaching the jig after the pre-bonding, wherein the pre-bonding comprises evaporating the liquid in a vacuum.— Method of Bonding Wafer and Chip, US20260215351A1

Claim 1, element by element

Four limitations, taken in order. First, pre-bonding by a liquid — the joining agent is a liquid, not an adhesive, a solder or a direct dry contact. Second, a wafer to a chip while the chip is attached to a jig — the pre-bond occurs with the tool still holding the chip, which fixes the sequence rather than leaving it open. Third, detaching the jig after the pre-bonding — the release is a positively recited step, not an implied consequence. Fourth, the pre-bonding comprises evaporating the liquid in a vacuum.

That fourth element is where the scope narrows hardest. A process that pre-bonds with a liquid and lets it evaporate at ambient pressure is outside claim 1 as written. So is a process that removes the liquid by any means other than evaporation, and one that evaporates it after rather than as part of the pre-bonding step, since the claim recites the evaporation as a component of the pre-bonding rather than as a subsequent operation. The vacuum requirement is the cleanest design-around target in the claim, and it is present in all three independents, so there is no independent claim that reaches a non-vacuum process.

Note also what claim 1 does not require. It does not recite surface tension, any material for the chip or the wafer, any temperature, any pressure, or any permanent bond at all. Claim 1 covers the temporary attachment and the tool release. On its face it stops before the join is made durable.

What the second and third independents add

Claim 9 recites the same pre-bond and jig detachment, then adds post-bonding, by heat and pressure, the wafer to the chip after the pre-bonding. That is the claim that reaches a completed bond. Its dependents supply the process window: claim 10 recites a pressure in a range of 0.5 atm to 4 atm, and claim 11 recites "a temperature in a range of 200° C. to 400° C." — reproduced here with the record's own formatting. Claim 12 recites that the post-bonding changes the bond caused by the pre-bonding into a covalent bonding, which is the clearest statement in the document of what the two stages are respectively for.

Claim 19 is the narrowest independent and the most interesting. It recites pre-bonding by a surface tension of a liquid, folds in the post-bonding by heat and pressure, and adds a limitation the other independents leave to dependents: wherein the chip includes an optical waveguide or an antenna. Claim 20 then combines both process-window ranges into a single dependent.

That last limitation is where the record disagrees with itself. Claim 5 recites that the wafer comprises a device, and claim 6 recites that the device comprises an optical waveguide or an antenna — putting the optical structure on the wafer side. Claim 19 recites the optical waveguide or antenna as part of the chip. Claims 15 and 16, depending from claim 9, follow the wafer-side reading. So the published set supports both configurations, but each only inside its own chain, and the two are structural inverses of one another. Any construction has to pick a chain. Blending them produces a description of an invention that no single claim recites.

It is worth being explicit about which reading carries more weight as filed. The wafer-side configuration appears twice, in the chain hanging off claim 1 and again in the chain hanging off claim 9, while the chip-side configuration appears once, inside claim 19. But claim 19 is an independent claim and the other two are dependents, so the single appearance is the one written into standalone scope. Frequency does not settle it; position does. An examiner working this file would have to address both, and the antecedent language in claim 19 gives no basis for reading its recitation as anything other than what it says.

The material limitations sit entirely in dependents. Claims 2 and 13 recite that the chip comprises a group III-V semiconductor. Claims 3 and 14 recite that the liquid comprises oxygen; claim 4 recites that it comprises water. Claims 7 and 17 recite manufacturing a laser diode by processing the chip. None of these constrains an independent claim, which means the independents read on the recited process regardless of the materials involved — the III-V and laser-diode context is prosecution fallback, not present scope.

For anyone mapping around this, the practical boundaries as published are: perform the liquid removal outside a vacuum, or hold the chip by something other than a jig during pre-bonding, or complete the bond without a distinct pre-bond stage at all. Each avoids limitations common to all three independents. The surface-tension recitation in claim 8 and claim 19 is narrower still, since a liquid pre-bond relying on a mechanism other than surface tension would fall outside those specific claims while potentially remaining within claim 1.

The standing caveat applies with full force here. This is a published application, not an issued patent. The claims described above have no enforceable effect in this form, have not been through public examination, and the independents in particular are broad enough that narrowing during prosecution should be expected rather than treated as a surprise. Published is not granted, and the scope discussed here is the scope as filed.