A tremendous amount of effort has been dedicated to emulate the intelligent functions of neuron networks using very-large-scale integrated neuromorphic circuits containing analog, digital and mixed-mode Si-based electronic devices. However, the energy consumption of a silicon synapse is ∼10 to 100 nJ/spike, which is significantly higher than the average rate of energy consumption for a biological synapse (∼0.1 pJ/spike). This adds another technical hurdle to the challenge of scaling up the circuits to the sizes and functions comparable with the human brain (∼1011 neurons and ∼1014 synapses), and scientists are keen to find a way around the issue.

Single-walled carbon nanotubes (SWNTs) have attracted attention for various applications including field-effect transistors (FET), nonvolatile memory, logic circuits, biosensors and biomimetic synapses thanks to their nanoscale size and electronic properties. By making use of a CNT transistor with a hydrogen-doped electrochemical cell, researchers at UCLA have successfully demonstrated a spiking-neuron circuit based on a CNT transistor, which can be used to emulate a biological synapse with low-energy consumption (∼3 pJ/spike).

Crossbar architecture

The spiking-neuron circuit has a crossbar architecture in which the transistor gates are connected to its row electrodes and the transistor sources are connected to its column electrodes. An electrochemical cell is incorporated into the gate of the transistor by sandwiching a hydrogen-doped poly(ethylene glycol)methyl ether (PEG) electrolyte between the CNT channel and top-gate electrode.

An input spike applied to the gate triggers the dynamic drift of hydrogen ions in the PEG electrolyte, resulting in a post-synaptic current (PSC) through the CNT channel. When multiple spikes are fed into the row electrodes they trigger PSCs via multiple CNT transistors and PSCs cumulate in the columns. When the integration of PSCs with respect to time reaches a threshold value in the "soma" circuit, an output spike is triggered based on an integrate-and-fire mechanism.

The team hopes that the CNT spiking-neuron circuits can emulate the functions of biological neuron networks and their associated intelligent functions, and will explore this behaviour in more detail in future work.

The researchers presented their results in the journal Nanotechnology.