Modern prosthetic hands integrated with flexible sensors can detect pressure, shear force, and strain. But they are frequently unable to identify the objects they touch due to a lack of at least two sensors—one that detects static pressure and the other that detects dynamic stimuli, such as vibrations. A human hand, by contrast, can recognize braille characters on a page or the fine texture of a fabric with the slide of a finger.
Materials scientists Ningning Bai, Chuan Fei Guo (both at Southern University of Science and Technology in China), and colleagues have now developed a single flexible sensor that can capture both static pressures and high-frequency vibrations when it touches an object and slides across its surface. In that sense, the sensor mimics biological perception, which relies on the detection of both slow and fast adaptive receptors in the skin.
The sensor, shown here attached to the end of a prosthetic index finger, consists of concentric 3D-printed elliptical whorls that resemble those of a human fingerprint. It’s made of a layer of polydimethylsiloxane (bathtub caulk), a layer of ionic gel, and two flexible gold electrodes. The electric double layers, which form at the interface of the ionic layer and the gold electrode, give rise to nanoscale charge separation—and thus ultrahigh capacitance-to-pressure sensitivity.
Bai, Guo, and their colleagues attribute the high sensitivity to a subtle change in the electric-double-layer interface when the sensor is pressed into a surface. Before then, the presence of an air gap prevents contact between the electrode and the ionic gel, which produces a low initial capacitance. When pressure is applied, the smaller protrusions of the gel begin to touch the electrode, and the capacitance signal rises sharply with the buildup of charge.
The sensor can respond to static and dynamic stimuli with a resolution as fine as 15 µm in spacing and 6 µm in height and at vibrational frequencies up to 400 Hz. That’s a spatial resolution superior to that of human fingertips and a temporal resolution almost two orders of magnitude shorter than that of existing capacitive sensors.
In a demonstration of its utility, the researchers used machine-learning techniques to train the sensor to recognize 20 different textiles, including wool, corduroy, linen, and polyester. The researchers then measured the capacitance signal as the sensor was pressed on and then slid slowly across the surface of each fabric. The trained sensor recognized each textile with an accuracy of 100%. Using the same machine-learning algorithm but without the synthetic fingerprint, the recognition accuracy dropped to 54%.
The researchers envision the new sensor as a tool for improving the sensory recovery of people who wear prostheses. (N. Bai et al., Nat. Commun. 14, 7121, 2023; thumbnail image credit: Chuan Fei Guo.)