IBM Patents a Two-Layer Chip Surface to Keep Medical Sensors Reading Accurately

Source: Patentlyze (October 9, 2026). AI-generated summary by biochip.com, published October 10, 2026. Not independently reviewed.

IBM has filed a patent application for a biosensor chip surface designed to keep its biological sensing layer attached more reliably in liquid samples. The approach uses two or more stacked oxide or nitride layers beneath the recognition coating, rather than the single layer used in the chip designs described in the filing. In tests on nine sample devices and a control, IBM reports that the layered structures produced a more stable signal over time. If the approach holds up in further testing, it could address a persistent materials problem in transistor-based medical sensors: readings that drift as the sensing surface degrades.

A layered foundation for biosensing

The proposed device is a biosensor, a chip that detects a target biological molecule by converting a molecular interaction into an electrical signal. It is built around a transistor, an electronic switch in which a gate electrode controls electrical current through a semiconductor channel.

On top of that gate electrode sits a biorecognition layer, the chemical coating that selectively captures a target, such as a virus or protein. When the target binds to the coating, it changes charge near the chip surface. That change alters the current flowing through the device, producing a measurable signal.

Why single-layer surfaces can drift

The challenge is similar to a sticker gradually lifting from a repeatedly washed bottle. In a liquid environment, water can break the chemical bonds that anchor a biosensor's recognition molecules to its surface, a process called hydrolysis.

As molecules detach, the sensing layer loses its ability to capture targets consistently. IBM says this can weaken the signal, reduce the lowest concentration a sensor can detect, and cause measurements to wander even when the sample itself has not changed.

IBM's stacked gate design

IBM's application, published as US 2026/0309672 A1, describes a gate electrode built from at least two stacked layers made from oxides or nitrides. The biorecognition layer is chemically grafted to the uppermost layer, creating the interface between the electronics and the sample.

One example starts with a roughly 4-nanometer base layer of hafnium dioxide or titanium nitride and adds a roughly 1.5-nanometer zirconium dioxide top layer. The process then adds a phosphonic acid compound that self-assembles into a single-molecule-thick film. A biotin-based sensing molecule may be linked to that surface.

The design focuses on the material immediately beneath the biological coating, rather than changing the basic transistor sensing principle. IBM also describes possible bipolar junction transistor and optical sensor versions, suggesting the surface chemistry could be applied beyond the primary semiconductor channel design.

Early device results

IBM reports testing nine sample devices alongside a control. The stacked-layer devices maintained a stable signal over time, while the single-layer devices showed large signal swings.

That is an encouraging engineering result, but it is not yet evidence of a finished diagnostic test. The application does not identify a commercial product, a disease target, or clinical testing in patient samples. A patent application also defines a proposed intellectual-property claim, not a regulatory clearance or proof of clinical performance.

Why This Matters

Many rapid and near-patient tests depend on small sensors operating in fluids, where the interface between the chip and its biological capture molecules can determine whether a reading remains trustworthy. A more durable attachment layer could be useful when a sensor needs to monitor a sample over time rather than provide a single quick measurement.

IBM filed the application on April 7, 2025, and it was published on October 8, 2026. The next meaningful questions are whether the layered gate improves performance with specific clinical targets, how consistently it can be manufactured, and whether it remains stable across real-world sample conditions.

Topics: Biosensors

Start here: Biosensors explained