Controlling\nLong-Term Plasticity in Neuromorphic Computing\nThrough Modulation of Ferroelectric Polarization

Electrolyte-gated\ntransistors (EGTs) have significant potential\nfor neuromorphic computing because they can control the number of\nions by mimicking neurotransmitters. However, fast depolarization\nof the electric double layer (EDL) makes it difficult to achieve long-term\nplasticity (LTP). Additionally, most research utilizing organic ferroelectric\nmaterials has been focused on basic biological functions, and the\nimpact on nonvolatile memory properties is still lacking. Herein,\nwe present a polyvinylidene fluoride (PVDF)-based ion-gel synaptic\ndevice using PVDF and poly(vinylidene fluoride-<i>co</i>-hexafluoropropylene) (P(VDF-HFP)) to implement LTP through the introduction\nof ferroelectric materials. The PVDF-based polymer slows the escape\nrate of TFSI anions from the electrolyte/channel layer through residual\npolarization. The fabricated synaptic devices successfully demonstrate\nLTP by controlling ion adsorption under the influence of PVDF-based\npolymers. Furthermore, it implements synaptic functions including\npaired pulse facilitation (PPF), high-pass filtering, and neurotransmitter\ncontrol. To validate the potential of neuromorphic computing, we successfully\nachieved high recognition rates for artificial/convolutional neural\nnetwork (A/CNN) simulations via sequential adsorption and desorption\nunder ferroelectric polarization with long-term potentiation/depression\n(LTP/D). This study provides a rational ion adsorption strategy utilizing\nthe ferroelectric polarization caused by the introduction of a PVDF-based\npolymer in the dielectric layer.