Dendritic Organic Electrochemical Transistors Grown by Electropolymerization for 3D Neuromorphic Engineering

Abstract

One of the major limitations of standard top-down technologies used in today's neuromorphic engineering is their inability to map the 3D nature of biological brains. Here, it is shown how bipolar electropolymerization can be used to engineer 3D networks of PEDOT:PSS dendritic fibers. By controlling the growth conditions of the electropolymerized material, it is investigated how dendritic fibers can reproduce structural plasticity by creating structures of controllable shape. Gradual topologies evolution is demonstrated in a multielectrode configuration. A detailed electrical characterization of the PEDOT:PSS dendrites is conducted through DC and impedance spectroscopy measurements and it is shown how organic electrochemical transistors (OECT) can be realized with these structures. These measurements reveal that quasi-static and transient response of OECTs can be adjusted by controlling dendrites' morphologies. The unique properties of organic dendrites are used to demonstrate short-term, long-term, and structural plasticity, which are essential features required for future neuromorphic hardware development.

Keywords: bipolar electropolymerization; long-term memory; organic electrochemical transistors; short-term memory; structural plasticity; synaptic plasticity.

Figure 1

Structural plasticity with dendritic PEDOT:PSS fibers. a) Schematic representation of the experimental set‐up for dendritic growth with bipolar electro‐polymerization. A periodic square signal of V _pp with frequency f _p of 40, 80, 160, and 320 Hz is applied in between the two freestanding Au wires. b) Temporal evolution of the formation of PEDOT dendrites with f _p = 160 Hz. c) Comparison of the morphologies achieved for different f _p. d) Current–voltage characteristics of the different dendrites. Inset: conductance evolution with f _p. Error bars are calculated with three repetitions of the experiments. e,f) SEM images of dendrites grown at 80 Hz. g) Normalized current density map for the dendrite obtained at f _p = 160 Hz based on image analysis and electrical simulations. h) Calculated value of resistivity from the experimental resistance value and image's predicted resistance value.

Figure 2

Electrical characteristics of dendritic OECTs. a) Schematic of the OECT setup with Ag/AgCl as a gate electrode. b) Transfer characteristic of the OECTs evidencing accumulation and depletion mode at negative and positive V _G, respectively. c) Transconductance values for the forward (square symbols) and backward (circle symbols) currents at V _G = 0 V. d) Impedance spectroscopy of standard OECT deposited by spin‐coating and dendritic OECT at f _p = 160 Hz. The n value, which is 0 for a perfect resistor, 1 for ideal capacitor and 0.5 for Warburg element.

Figure 3

Short‐term plasticity effect demonstrated for various dendritic morphologies. a) Square‐shaped pulses of 10 s were applied to the gate with continuous recording of source–drain voltage of 0.1 V. b) Typical SD current response to a square shape pulse. c,d) Source–drain current responses for dendritic OECTs grown at c) f _p = 40 Hz and d) f _p = 320 Hz with pulse amplitude from ‐0.4 to 0.4 V with step of 0.1 V. Potentiation (depression) is observed at negative (positive) gate voltages. e) Normalized responses for dendritic OECTs grown at 40, 80, 160, and 320 Hz with gate pulses of 0.4 and ‐0.4 V. f,g) Variation in time constant of charging/discharging regions.

Figure 4

In situ network evolution and long‐term memory effects. a) Formation of a third dendrite using the electropolymerization process with f _p = 80 Hz. b) Schematic of the read operation with gate terminal disconnected. c) Schematic of the programming of the dendritic OECTs with positive/negative sweep applied on the gate. d) Successive program/read sequences with e) V _sweep increased from (±) 0.1 to 0.4 V (step of 0.1 V) in between each sequences. Long term potentiation (Depression) is obtained at negative (positive) bias. Time interval in between two successive programming was around 30 s.