An isotope battery does not store charge and then give it back. It converts the energy of radioactive decay directly into electrical current, typically by letting emitted particles create electron-hole pairs in a semiconductor junction. The appeal is duration: output is set by a half-life rather than by a charge cycle, which suits sensors and implants that must run for years without service. The obstacle has never really been the physics. It is that the source material is expensive, and the conventional way of applying it — depositing a film across the whole semiconductor surface — spends a great deal of it on area that does not need it.

US20260231568A1, published on August 6, 2026 and assigned to LG Energy Solution, Ltd., is addressed to that problem. The device it describes is simple to state: a first conductivity type semiconductor layer; a radiation source positioned along one side of it; and a second conductivity type semiconductor layer on the far side of the source, so the source sits between two layers of opposite type. What distinguishes it from a generic betavoltaic stack is the shape the source is required to take. It has, in the claim's words, an at least partial droplet shape.

That geometry is a consequence of how the layer is made, and the fabrication method is where the disclosure does its real work. Instead of coating, the source is prepared as a slurry and dispensed as a plurality of discrete droplets spaced along the surface. The droplets are cured in place. Only then is the second semiconductor layer formed over the cured source and the exposed surface between the droplets. Material lands where it is wanted and essentially nowhere else.

A method of fabricating the isotope battery is disclosed in such a way that the isotope battery may be mass-produced by a simple process without wasting expensive materials.— Isotope Battery and Method of Fabricating the Same, US20260231568A1

The parameters that make it a process

The dependent claims read like a process window rather than a wish. The droplets are specified at 10 micrometres to 10 millimetres in diameter as they are discharged toward the surface — four orders of magnitude, which is less a precision spec than a statement that the method is meant to scale from microfabrication to something closer to printing. The slurry viscosity is given as 0.5 to 2000 centipoise at 25 °C, a range running from roughly water to something like a light syrup. And one claim requires that the ejected droplets form a free surface across their total surface area before landing — that is, that they detach and fly rather than being drawn out in contact with the substrate, which is the distinction between jetting and extrusion.

Surface behaviour after landing is handled by an alternative pair. One claim requires the source to meet the semiconductor surface at a contact angle greater than 90 degrees; another, depending from the same parent, requires less than 90 degrees. These are mutually exclusive, and deliberately so — a bead that stands proud of the surface and a bead that wets and spreads are different devices with different absorption geometries, and the application claims both rather than choosing. A separate branch etches a concave pit into the first layer and seats the droplet at least partly inside it, with a further claim allowing part of the source to protrude from the pit and spread onto the adjacent surface.

Two further claims are worth flagging for anyone reading the disclosure as a device rather than a process. One places a photon generating layer between the radiation source and the first semiconductor layer — an indirect conversion path, where decay products produce light and the junction harvests the light, rather than the particles striking the junction directly. Another arranges the discrete droplet regions as nodes of a grid extending along the surface, which turns a single source pad into a patterned array whose spacing is a design variable.

What the record does not say

It does not name the isotope. Nowhere in the abstract or the twenty claims is a specific radionuclide identified, and the half-life, activity and emission type that would determine power output are all absent. The claims are drafted around geometry and process, not around the source chemistry. Nor does the record disclose an output figure, an efficiency, a device lifetime or an application. The classification codes on the record sit in the H10H 20-series — semiconductor light-emitting device territory — which is consistent with the photon generating layer variant but is not itself a statement about the device's purpose.

The context is that this is one record out of forty-three that the same applicant published on August 6, and the other forty-two are lithium-ion work: an intumescent fire-protection coating that expands between cells, a cylindrical cell whose crimp bend is held under 0.175 mm, a lithium-rich manganese cathode with a carbon coating, an electrolyte injection apparatus. This desk has covered several of that pack-safety line over recent drops. The isotope battery is a different kind of object entirely, and it arrives with a manufacturing method attached rather than as a materials result — which is the detail that makes it legible as coming from a company whose core competence is making cells in volume.

It remains a published application. The claims have not been examined to allowance, no device has been shown to exist, and publication reflects a filing made roughly eighteen months earlier. What is on the record is a described method for putting an expensive material only where it is needed.