Nanoscopic Electrolyte-Gated Vertical Organic Transistors with Low Operation Voltage and Five Orders of Magnitude Switching Range for Neuromorphic Systems

Abstract

Electrolyte-gated organic transistors (EGOTs) are promising candidates as a new class of neuromorphic devices in hardware-based artificial neural networks that can outperform their complementary metal oxide semiconductor (CMOS) counterparts regarding processing speed and energy consumption. Several ways in which to implement such networks exist, two prominent methods of which can be implemented by nanoscopic vertical EGOTs, as we show here. First, nanoscopic vertical electrolyte-gated transistors with a donor−acceptor diketopyrrolopyrrole− terthiophene polymer as an active material can be used to reversibly switch the channel conductivity over five orders of magnitude (3.8 nS to 392 μS) and perform switching at low operation voltages down to −1 mV. Second, nanoscopic EGOTs can also mimic fundamental synaptic functions, and we show an interconnection of up to three transistors, highlighting the possibility to emulate biological nerve cells.

Donor Engineering Tuning the Analog Switching Range and Operational Stability of Organic Synaptic Transistors for Neuromorphic Systems

“…For example, Eckel and co‐workers recently designed a nanoscopic vertical organic electrolyte‐gated transistors that achieves large conductance changes of five orders of magnitude after 1000 pulses. [ 25 ] However, the complex device fabrication process and nonlinear resistance tuning limit its application as an artificial neural network. In organic electronic devices, channel conductance can be limited by the electronic mobility and ionic mobility of organic semiconductor materials.…”

Sub-3 V, MHz-Class Electrolyte-Gated Transistors and Inverters

“…These devices employ gate potentials to modulate the conductivity of conducting polymers in direct contact with electrolytes. Currently, EGTs are again enjoying a resurgence of interest associated with their applications in physiological recording, − neuromorphic computing, − and biosensing, − as well as in fundamental physics, − where the high charge densities achieved in electrolyte-gated semiconductors are leading to exciting discoveries.…”

“…In our view, there are clear reasons why EGTs will not rival the speed of Si MOSFETs (though perhaps not everyone agrees), but for envisioned applications of EGTs, gigahertz speeds do not appear necessary. As EGTs are being developed as physiological recording amplifiers, , and integrated into circuits ,− and neural nets, ,,,, it is nevertheless important to establish the limits of performance that can be obtained by rational design. We note that there are other limitations to EGT performance like high dielectric loss tangents and quasistatic leak currents associated with electrolytes that are important for power consumption, but these are not our focus here.…”

Sub-3V, MHz-Class Electrolyte-Gated Transistors and Inverters

“…These devices employ gate potentials to modulate the conductivity of conducting polymers in direct contact with electrolyte. Currently, EGTs are again enjoying a resurgence of interest associated with their applications in physiological recording [8][9][10] , neuromorphic computing [11][12][13][14][15][16] , and biosensing [17][18][19][20][21][22] , as well as in fundamental physics [23][24][25][26] , where the high charge densities achieved in electrolyte-gated semiconductors are leading to exciting discoveries.…”

“…In our view there are clear reasons why EGTs will not rival the speed of Si MOSFETs 2 (though perhaps not everyone agrees 38 ), but for envisioned applications of EGTs, gigahertz speeds do not appear necessary. As EGTs are being developed as physiological recording amplifiers 9,10 , and integrated into circuits 36,[39][40][41] and neural nets 4,11,13,14,16 , it is nevertheless important to establish the limits of performance that can be obtained by rational design. We note that there are other limitations to EGT performance like high dielectric loss tangents and quasi-static leak currents associated with electrolytes that are important for power consumption 2 , but these are not our focus here.…”