Water is the contaminant that sodium cells cannot tolerate, and the conventional answer has been to drive it out with heat. A published application that appeared in the 16 July 2026 US drop, US20260204669A1, assigned to CONTEMPORARY AMPEREX TECHNOLOGY CO., LIMITED, proposes something closer to the opposite instinct: leave the water in place, inject the electrolyte on top of it, and then use the cell's own first charge to react the water away electrochemically. The filing is titled “BATTERY MANUFACTURING PROCESS, BATTERY TREATMENT DEVICE, BATTERY, AND ELECTRIC DEVICE,” lists six inventors — Kai WU, Lan XIE, Ying SHA, Shuai SONG, Zhen LIN and Xiao LI — and is classified under CPC H01M 10/446, H01M 10/054 and H01M 10/058. It is a published application, not a granted patent.

The background section states the problem the filing is directed against in its own words: In the prior art, water is generally removed from batteries through high-temperature baking, but in this water removal method, the water inside the battery cannot be effectively removed. That is the baseline the record picks — baking, not any form of ambient moisture control. The description goes further and names a case where baking specifically struggles: a cell on which water removal through “high-temperature baking” is difficult, such as a cell with a Nasicon-type phosphate positive electrode material system. Heat has to pull moisture out through the electrode sheets and separator from the outside in; the record's position is that this does not reach the water that matters.

The voltage window is the whole invention

Independent claim 1 is startlingly short. It recites a battery manufacturing process comprising acquiring a battery to be treated, injecting an electrolyte into the battery, and charging the battery before sodium in the battery is reduced at a negative electrode. That is the entire independent claim. Worth being precise here: the phrase “sodium-ion” does not appear in claim 1 at all. Sodium enters the claim only through the words “sodium in the battery is reduced at a negative electrode,” and enters the record more broadly through CPC H01M 10/054, which covers non-lithium secondary cells. The claim as literally written is about a battery whose negative electrode can reduce sodium — not a claim that names a chemistry. The mechanism lives one level down, in claim 2, which supplies the parameter that makes the whole scheme work:

charging the battery at a first charge voltage, wherein the first charge voltage is higher than a reduction potential of H+, and the first charge voltage is lower than a reduction potential of sodium.— BATTERY MANUFACTURING PROCESS, BATTERY TREATMENT DEVICE, BATTERY, AND ELECTRIC DEVICE, US20260204669A1

Read that as a bracket. Above the lower bound, hydrogen ions will take electrons at the negative electrode; below the upper bound, sodium will not. Park the cell inside that bracket and you have an electrode surface that is electrochemically active toward water and inert toward sodium. The abstract describes the two routes water can take: water may receive electrons at the negative electrode and undergo a reduction reaction, producing hydrogen gas and hydroxide, or water may first decompose in the electrolyte to generate hydrogen ions which are then reduced at the negative electrode to hydrogen gas. Either way the water leaves as H2. The description writes the reaction as 2H2O+4e−→H2+2OH− — reproduced here as printed in the record, where it is not balanced as written — and gives a second path involving the electrolyte salt, 2H2O+NaPF6+4e−→NaPO2F2+2H2+4F−.

The ordering of the process steps is not incidental. Claim 1 injects electrolyte before the charge step, which at first looks like the wrong sequence if your goal is a dry cell. It is the right sequence for this mechanism: the electrolyte is the transport medium. Water trapped in the electrode sheets and separator has to diffuse into the liquid phase before it can reach the negative electrode and be consumed. Baking has no such carrier. Claims 4 and 5 add heat — the battery is heated during the first-voltage charge, at 25° C. to 70° C., preferably 45° C. to 60° C. — but here heat is doing a different job than in the prior-art baseline, speeding diffusion of water into the electrolyte rather than trying to evaporate it out of the stack.

Pulling the product out to drive the reaction

Claims 6 and 7 evacuate the cell during the first-voltage charge, at a vacuum of at least −80 kPa and below 0 kPa, preferably −30 kPa to −10 kPa. This is straightforward Le Chatelier logic applied to a manufacturing line: the reaction produces hydrogen gas, so removing the hydrogen as it forms pushes the equilibrium toward producing more of it, which means consuming more water. Claim 3 gives the first charge voltage as 1.0 V to 2.9 V, preferably 1.7 V to 2.9 V, more preferably 2.5 V to 2.7 V; claim 8 holds it there for 4 to 48 hours, preferably 6 to 18 hours. The description states that at 12 h, the water removal efficiency is close to 90% — a statement made in the description about the disclosed process, not a validated cell specification, and these ranges are preferred embodiments in a filing rather than any disclosed production setting.

Only then does claim 9 lift the ceiling. The second charge voltage is above sodium's reduction potential and/or above the reduction potential of an electrolyte additive — greater than 2.9 V and up to 4.2 V. By the time sodium metal is deposited, per the abstract, the water it would otherwise have met is gone. The record explains the stakes plainly: The reduced sodium metal has high activity and is prone to violent reactions with water in the battery, generating gases such as H2, thereby causing significant capacity loss. The same H2 either way — the filing's argument is that it is far better to generate it deliberately, at a controlled voltage, with a vacuum line attached, than to generate it accidentally against fresh sodium metal. The record also notes that leftover water hinders sodium deposition, and that uneven deposition easily causes local dendrites, leading to short circuits in a cell. Claim 11 closes the sequence with electrolyte injection, evacuation, protective gas injection and sealing. Independent claim 12 recites the same logic as a battery treatment device with an acquisition unit and a charge unit; claims 15 and 16 reach the resulting battery and an electric device containing it.

The rest of the company's 16 July cohort sits somewhere else entirely. Five other applications published under Contemporary Amperex assignee strings that day: US20260204759A1, a cell housing routing both tabs out the same end surface; US20260204612A1, a winding mandrel whose secondary members retract so the assembly narrows for jelly-roll release; US20260204611A1, a pressing plate that bends a cell tab onto the terminal post; US20260201509A1, an aluminum alloy for battery boxes formulated to suppress the Al2Cu phase; and US20260200307A1, a vehicle with a raised battery platform under the seat cushion. Cell hardware, assembly equipment, materials, pack structure. Among these six records, the hero is the only chemistry and formation-process filing and the only one that touches sodium at all — an observation about what published on one date, and nothing more.