UCLA

Fabrication and operation of carbon nanotube (CNT) based electronic devices called "synapstors," with the goal of emulating the functions of biological synapses, are reported. These synapstors have a structure akin to field-effect transistors, utilizing a random network of single-wall semiconducting CNTs as its conducting channel. Analog spike signal processing with low power consumption was demonstrated. These synaptic devices are capable of carrying out logic, learning, and memory functions, all in a single element. Analog spike neuromorphic circuits, composed of CNT synapstors and integrate-and-fire (I&F) circuits, are also reported. A positive voltage spike, applied on the gate electrode of a synapstor, could generate a postsynaptic current (PSC) via the CNT channel of the device. Multiple input spikes could be applied on individual CNT synapstors, which could be either excitatory or inhibitory and be connected in a neuromorphic circuit, inducing excitatory or inhibitory PSCs, respectively. These PSCs flow collectively to an I&F circuit, triggering output spikes. Such neuromorphic circuits are capable of emulating biological neural networks, enabling parallel signal processing, dynamic learning, and memory functions with low power consumption.