Acoustoelectrically Driven Ferroelectric Transistors Enabling Interactive and Adaptive Neuromorphic Computation.

Neuromorphic electronics capable of interactive and adaptive signal processing are essential for next-generation intelligent systems. While most neuromorphic platforms rely on electrical, optical, or chemical stimuli, the incorporation of acoustoelectric interactions as an active computational degree of freedom remains largely unexplored. Here, we present acoustoelectrically-driven ferroelectric transistors that enable interactive and adaptive neuromorphic computation by integrating surface acoustic wave (SAW) excitation with a MoS2/Pb(Zr,Ti)O3 ferroelectric field-effect transistor. Propagating SAWs generate a direction-dependent acoustoelectric current, which cooperatively interacts with electrically programmed ferroelectric polarization, allowing mechanical-wave and electrical signals to be processed within a single device. This wave-memory coupling enables neuron-like functionalities beyond conventional synaptic switching, including spatial-temporal dendritic integration, threshold-controlled firing, and adaptive response behaviors. Furthermore, the device demonstrates reconfigurable Boolean logic operations through excitatory-inhibitory modulation governed by SAW propagation and firing thresholds, as well as associative learning inspired by Pavlovian conditioning. Finally, wireless human-activity recognition simulations demonstrate that device-level integration of acoustoelectric modulation and ferroelectric memory improves multimodal classification accuracy. By introducing acoustoelectric coupling into ferroelectric neuromorphic transistors, this work establishes a versatile materials-based framework for adaptive, multimodal neuromorphic systems that bridge mechanical wave physics, ferroelectric memory, and brain-inspired computation.