Interfacial Polarization Dynamics of Nitrile Dipoles in Organic Synaptic Transistors for Acoustic Reservoir Computing.

Polar polymer interlayers have emerged as core functional layers in organic synaptic transistors (OSTs), enabling diverse neuromorphic behaviors through interfacial dipolar polarization and relaxation. However, the interfacial dipolar mechanisms governing memory window (ΔV) formation and modulation remain insufficiently understood. Here, polyacrylonitrile (PAN) is selected as a representative polar polymer interlayer to exploit nitrile (CN) dipolar relaxation for temporal acoustic reservoir computing (RC). It is found that the ΔV varies nonmonotonically with PAN concentration, suggesting a dipolar competition mechanism between interfacial polarization strength and dipole relaxation capability. Accordingly, dipolar relaxation underpins multiscale synaptic responses and fading memory for temporal RC. The device further exhibits stable and symmetric weight updates, with a low asymmetry ratio (AR) of 5.06% and cycle-to-cycle variations (CCVs) below 1%. Notably, the PAN-OST operates with an ultralow event energy of 50 aJ while maintaining a high signal-to-noise ratio (SNR) of 50.47 dB. The PAN reservoir achieves 97.17% accuracy in spoken digit recognition and a maximum accuracy of 94.8% in the more challenging speech emotion recognition. Overall, this work establishes an interfacial polarization-driven pathway for energy-efficient neuromorphic computing and next-generation intelligent hardware.