Researchers at MIT have built a tiny device that uses motion, not only electric charge, to process information. The work is published in Science Advances. The platform is meant to put computing and memory in one small, low-power part, for uses such as edge computing, medical monitors, and adaptive machines.
The idea comes from living systems that compute through physical change. An octopus keeps much of its nervous system in its arms, so sensing and response can happen locally. Mechanical computing works in a related way: calculation is done by movement or compression. This is a nenoscale device. At that size a small shift in shape can change how a material conducts electricity,. A challenge is that two surfaces that come very close tend to stick and will not separate again.
How the device works
The researchers placed a super-thin film of polydimethylsiloxane, or PDMS, a soft polymer familiar from laboratory molds and some medical parts, between two metal electrodes. The polymer acts as a nano-spring and balances the adhesive forces that would otherwise weld the metals together, so the gap can open and close in a controlled way. When voltage is applied, the plates pull toward each other, squeeze the film, and change the current that flows. PDMS is viscoelastic, which means that after it is compressed it returns to its first shape only after a delay. That delay lets the device keep a short memory of the forces and voltages it has seen and turn that history into an electrical signal.
The same behavior is used as an artificial neuron. In the brain a neuron gathers input until a threshold is reached, then fires and later settles. In this device, repeated voltage slowly compresses the film. Past a threshold the device fires, then relaxes. Computing and memory sit in the material itself, so extra parts such as capacitors are not required. The researchers say the memory of the polymer can be tuned across a wide range of designs. Possible later uses include prosthetics that handle touch data on the spot and wearable patches that read health signals with little power. Future work aims to add sensing to the same mechanical platform.