The work proposes a self-powered transistor-like iontronics (STI) device based on an MXene/Bi 2D heterojunction as a compact tactile interface for HMII. Conventional HMII platforms often rely on complex multi-sensor arrays with heavy wiring, signal crosstalk and latency. Here, the authors instead design a single flexible, all-solid-state device that combines sensing, encoding and low-power readout in one structure. The free-standing STI (FSTI) device employs interdigitated MXene@Zn (source) and MXene@Bi (drain) electrodes, a PVDF-HFP-GO solid polymer electrolyte with high ionic conductivity, and a cellulose nanofiber isolation layer. The MXene/Bi heterostructure enlarges MXene interlayer spacing and suppresses restacking, providing abundant active sites and fast Zn 2+ diffusion.
Mechanistically, the device mimics a p-channel MOSFET: without pressure, the CNF layer blocks ion channels, and the device is “off”. Under pressure, electrode-electrolyte contact increases, ion channels form, Zn is oxidized at the source and intercalated at the drain, and the resulting ion migration drives an external electronic current. A fixed intrinsic potential difference of ~1.1 V determined by electrolyte concentration, while external pressure modulates the internal resistance via the number of ion channels, tuning the open-circuit voltage. Electrochemical impedance spectroscopy confirms that pressure lowers charge-transfer resistance and enhances Warburg diffusion. Voltage-mode readout yields extremely low power consumption compared with current-mode measurement.
Optimized devices achieve an output up to 1.1 V and 2.3 μA, linear sensitivity (R 2 = 0.995), fast response/recovery (66.59/44.18 ms), broad frequency tolerance (up to 2.22 Hz), and outstanding durability over 50,000 loading cycles. Demonstrations include self-powered monitoring of radial pulse, object weight and surface tension, as well as robotic joint motion, with wireless Bluetooth transmission and direct LED driving. Finally, the authors integrate a single FSTI on the wrist with a neural-network model on Arduino to decode five hand gestures with 95.83% accuracy and to provide tactile feedback for a robotic hand, enabling delicate manipulation (e.g., grasping tofu without damage). The work establishes a promising iontronics paradigm for compact, intelligent HMII interfaces.
Science Bulletin
Experimental study