University of Osaka researchers built a solid-state nanopore that opens and closes on its own as mineral deposits form and dissolve, producing signals that machine learning can use to tell molecules apart.
Most sensors are passive. They register whatever passes through or touches them, and they respond the same way each time. A thermometer does not care what the temperature was an hour ago.
Researchers at the University of Osaka have built a molecular sensor that behaves differently. In a study published in ACS Nano, a team from the university’s Institute of Scientific and Industrial Research (SANKEN) and collaborating institutions describes a tiny pore that senses molecules, generates electrical signals and retains a short-term memory of recent events, all without outside control.
“Our nanopore works differently,” said lead author Makusu Tsutsui.
Holes that read molecules
Nanopores are openings only a few billionths of a meter wide. When they sit in a thin membrane separating two salty solutions and a voltage is applied, ions flow through the pore and create a measurable current. A molecule passing through briefly blocks part of that flow, and the resulting dip in current reveals information about the molecule.
The approach has already transformed DNA sequencing. Devices that thread DNA strands through protein nanopores, reading the bases from changes in current, have been commercially available for about a decade and are now used in labs and field settings around the world. Solid-state nanopores, drilled into synthetic materials rather than built from proteins, are a promising alternative because they can be more robust and easier to manufacture.
The basic principle is older still. In the 1950s, the engineer Wallace Coulter patented a method for counting blood cells by drawing them through a small opening and measuring the momentary changes in electrical resistance as each cell passed. Nanopores apply the same idea at the scale of single molecules.
Most nanopores, however, are passive openings whose behavior is controlled by external electronics. The Osaka team set out to make one that would reshape itself.
A pore that opens and closes itself
The key is chemistry inside the pore. “Under a constant voltage supply, chemical reactions inside the pore repeatedly build up and dissolve tiny mineral deposits,” Tsutsui said.
As these deposits grow and dissolve, the pore opens and closes on its own, producing bursts of electrical signals. The device does not need to be switched or tuned during operation. It runs through its cycles simply because a steady voltage is applied.
Molecules passing through the pore interfere with those ongoing reactions, changing how the pore evolves over time. Because the pore’s state at any moment reflects what has happened before, the device carries a record of recent events.
Senior author Tomoji Kawai explained that different molecules change the size, duration and timing of the electrical spikes in distinctive ways, leaving what he described as “unique electrical signatures.” “As each signal also depends on the nanopore’s recent memory, the device behaves as a remembering, chemically active sensor,” he said.
Telling molecules apart
To decode those signatures, the team used machine learning. The approach could distinguish all four DNA nucleotides, the building blocks of genetic code, and could measure mixtures of them. The researchers then extended it to seven amino acids, the building blocks of proteins.
All of this was done without actively controlling the pore during sensing. The release does not give accuracy figures for the classifications.
Identifying amino acids is of particular interest. Sequencing proteins directly, one amino acid at a time, remains far harder than sequencing DNA, partly because there are 20 standard amino acids rather than four bases, and many differ only subtly. Nanopores are among the leading candidates for eventually making single-molecule protein sequencing practical.
Toward iontronics
“This represents a shift in how nanopores are designed,” Tsutsui said. Instead of serving as passive channels that simply let molecules through, he said, nanopores can actively respond to their chemical environment.
Devices whose electrical response depends on their own history are already of interest in electronics. Memristors, for example, are components whose resistance changes with the current that has passed through them, and researchers are exploring them as building blocks for computers that mimic the brain.
The researchers suggest that such multifunctional nanopores could be useful in biomedical research and diagnostics. They also point to iontronics, an emerging field that processes information using ions rather than electrons. Living nervous systems already work this way, with ions moving across cell membranes to carry signals, and a device whose response depends on its own history has some of the character of a memory element.
The work is at an early stage. The release does not identify the minerals involved, report how reliably molecules were classified, or say how the device performs on complex biological samples. The research was supported by the Japan Society for the Promotion of Science.
Notes
- Tsutsui, M., et al. “Autonomous molecular sensing with a chemically stateful solid-state nanopore.” ACS Nano, 30 July 2026. DOI: 10.1021/acsnano.6c08258.
- The University of Osaka. “New smart sensor identifies present molecules by remembering the past.” EurekAlert!, 30 July 2026. https://www.eurekalert.org/news-releases/1138160.




Leave a Reply