A battery you can bake

Drop a lithium-ion cell into a 200-degree oven and it does one of two things: it swells and vents, or it catches fire. Ordinary rechargeable batteries live in a narrow comfort zone, roughly 0 to 45°C, and they punish you for leaving it. So it reads as a small act of defiance that a Japanese company started shipping samples, in August 2024, of a rechargeable cell that charges at a 20C rate inside a 200°C furnace and keeps working down to 40 below. Brief excursions to 300°C are fine.

The battery is made of glass. Not glass as in a fragile case. Glass as in the cathode is glass, the anode is glass, and the electrolyte between them is glass, each melted and crystallized the way you’d make the substrate inside a flat-panel display.

The company shipping it is not a battery company. It is Nippon Electric Glass, or NEG, a 76-year-old maker of display glass and fiberglass that spun out of NEC in 1949 and, by its own accounting, did about $1.9 billion in revenue last year selling mostly the glass sheets inside LCD panels and the glass fiber that reinforces everything from circuit boards to wind blades.

The number nobody is reporting

Read across the US grant record and NEG is not dabbling. Since 2018 it has been granted 15 US patents on sodium-battery materials and cells. Narrow to grants issued since 2022 that mention sodium-ion batteries and NEG sits at the top of the entire field with 10, ahead of CATL’s battery-recycling arm, ahead of the university labs, ahead of the specialist chemistry startups. Narrow one more notch, to US patents with “solid-state sodium” in the title, and it stops being a race: NEG holds five, and no other assignee holds more than one. Not LG Energy Solution. Not Toyota. Not Samsung. A glass company.

To see why that should bother the companies racing to build solid-state batteries the usual way, you have to understand what actually goes wrong inside one.

Why solid-state is hard, and how glass cheats

Every solid-state battery fights the same enemy: the interface. In a normal cell, a liquid electrolyte soaks into every pore of the electrodes and wets everything it touches. Swap in a solid electrolyte and now you have two hard materials pressed face to face, making contact only where their microscopic bumps happen to line up. Ions have to hop across that ragged gap. Then, every time the cell charges and discharges, the electrodes breathe, expanding and contracting by a few percent, and the brittle join cracks, delaminates, and the cell quietly dies.

NEG has a patent, granted in February 2025, devoted entirely to this failure. It specifies roughening the electrolyte surface to a controlled texture so the electrode layers become, in the document’s words, “difficult to peel.” When a company is patenting the microscopic roughness of a seam, you know the seam is where batteries go to die.

Glass has a trick that pressed ceramic powders don’t. Heat it past its softening point and it flows. NEG’s process melts the components so the glass electrode and the glass electrolyte fuse into one continuous body, the interface welded shut rather than clamped together and hoping. The company calls it “softening fluidity.” It is exactly what a glass melt line does all day, and exactly what a battery startup with a powder press cannot do. The single hardest manufacturing problem in solid-state batteries turns out to be the thing glassmakers solved generations ago.

A door that has been open for 50 years

The ion conductors NEG uses are not new. They are β″-alumina and NASICON, two of the oldest known fast sodium-ion conductors, and their backstory is the real adjacent-possible move here.

β-alumina got its first job in 1966, when Neill Weber and Joseph Kummer at Ford Motor Company were trying to build an electric-car battery and needed a wall that would pass sodium ions but block electrons. They found it in a quirk of aluminum oxide. The result was the sodium-sulfur battery: molten sodium on one side, liquid sulfur on the other, a β-alumina membrane between them, all running above 300°C. It worked. It was also, essentially, a furnace with terminals. NASICON, a name compressed from “sodium super-ionic conductor,” arrived a decade later, in 1976, out of John Goodenough’s lab. Goodenough went on to co-invent the cobalt-oxide cathode that made the lithium-ion battery possible, and his sodium conductor faded into a footnote.

These materials have sat in the literature for half a century, wheeled out mostly for hot molten-metal grid batteries. What changed is not the chemistry. It is that someone who knows how to process oxide glass at industrial scale worked out how to build a fully solid cell around those old conductors and assemble it near room temperature, using heat to fuse the interface. The door was openable the entire time. It took a glass company to walk through it.

What’s actually inside the cell

The cathode is where NEG’s glass fluency shows most. Instead of the layered sodium oxide most sodium-ion developers use, a 2025 patent describes a crystallized-glass cathode built from a sodium-iron-or-cobalt phosphate, melted as a glass and then crystallized into the active material. Iron is the version NEG pushes commercially, and that choice is the whole cost argument: sodium and iron are among the cheapest, most abundant elements in the crust. The cell uses no lithium, no cobalt, no nickel, and, because every layer is an oxide, no sulfides, chlorides, or fluorides. There is nothing inside to off-gas anything toxic if the cell is crushed.

The specifications are modest, and NEG is honest about it. The heat-resistant cell is a 60-by-50-millimeter wafer one millimeter thick, 2.9 volts, four milliamp-hours. A standard version reaches 80 mAh. Nobody is putting these in a car.

Who cares, and why

That ceiling is the point, and also the tell. NEG is not aiming at the electric vehicle. It is aiming at every place a lithium cell cannot survive: a spacecraft, a downhole oil-and-gas sensor, semiconductor process equipment, an engine bay, a sterilization autoclave, a high-temperature wireless node. These are small, high-margin, capability-defined markets where “works at 200°C and cannot ignite” is worth far more than energy density. The rest of the solid-state sodium patent list says the same thing in a different accent: the other names filing are China Glaze, another glass-and-ceramics maker, and Ceramatec, a ceramics shop. The people who see this opportunity are the people who already melt oxides for a living.

There is a larger frame here for anyone tracking the solid-state race by watching QuantumScape’s share price and Toyota’s press releases. Those bets are all lithium, all aimed at the car, and all fighting the exact problem NEG sidesteps: how do you make a defect-free, paper-thin, solid electrolyte sheet cheaply and at volume? A glass company has been answering that question for display substrates since the 1960s. If solid-state batteries end up being won on the melt line rather than in the powder lab, the incumbents are not in Silicon Valley or Toyota City. They are in Otsu, Shiga, and they have been making the glass in your television the whole time.

NEG has not announced a mass-production date, and the cells shipping now are specialty samples, not a product line. But the patent record is the part a company can’t spin. When a firm files 15 times on one idea across seven years and out-files the entire field on the hardest version of it, that is not a side project. That is a company that has seen a door.

Method. Patent counts come from roughly 9.3M US utility grants sourced from USPTO bulk grant XML, searched for sodium-ion and solid-state sodium battery filings and matched to assignees; each company’s total combines variant spellings and subsidiary filings, so treat exact counts as close approximations, not audited figures. The grant record runs through mid-June 2026. Historical context on β-alumina (Ford Motor Company, 1966) and NASICON (1976) is drawn from the published battery literature; cell specifications, materials, and the August 2024 sample-shipment date come from Nippon Electric Glass’s own product disclosures and press releases. Patent counts measure filing activity and technical position, not shipping volume or commercial success; the cells described here are low-capacity specialty samples, not mass-market products.