Neuromorphic organic electrochemical transistor with agarose hydrogel for high-endurance plasticity

With an increasing interest in information processing and the development of biocompatible technologies, the opportunities regarding neuromorphic structures have risen. Neural networks have the ability of storing information through irreversible chemical modifications, obtaining long-term plasticity. On the other hand, short-term plasticity, which is defined by the ability of temporarily store information, relates to an induced strengthening/weakening of the synaptic weight that is dissipated after a characteristic time constant. Both the processes can be emulated by different structures as memristors, transistors or capacitors. Considering organic electrochemical transistors (OECT) as suitable components for these applications, our research group has recently developed a method that induces long-term plasticity involving direct electropolymerization of poly(3,4-ethylenedioxythiophene):polystyrene sulfonate (PEDOT:PSS) on the gate electrode, thanks to a series of voltage pulses. [1] Following the same direction, to enhance the compatibility with hardware electronics and durability of neuromorphic OECTs, the application of an agarose-based hydrogel as a solid electrolyte was investigated. The channel, connecting the source and drain electrodes, and the gate electrode are encapsulated in a hydrogel composed by agarose, EDOT and NaPSS. The hydrogel composition can be tuned to optimize the device performances, in terms of long-term plasticity emulation and transconductance. The device is prepared by drop casting the liquid gel precursor on the OECT, covering the channel and the gate electrode. Once the crosslinked network is physically formed, with the application of voltage pulses at the gate electrode between -0.5 and 1.3 V and of a fixed drain potential (Vd) of -0.3 V, the electropolymerization starts. Short-term plasticity was also investigated, observing an increase in characteristic time constant because of the presence of a porous 3D network, opposed to the previous aqueous solution. Future perspectives involve the optimization of the composition, to increase the device transconductance, and the study of neuromorphic OECTs as control devices.