The world’s greatest supercomputers still look like crude counting tools compared to the human brain, but a new chip has pinched a human technique for improving the efficiency of computation.

The human brain is made up of close to a hundred billion neurons, each with their own network of synapses and circuits which can not only process data, but are strengthened and weakened depending on the needs of the whole system. This is commonly referred to as ‘learning’.

Materials scientists at the Harvard School of Engineering and Applied Sciences have now created a new type of transistor that mimics the behaviour of a synapse. The device simultaneously modulates the flow of information in a circuit and physically adapts to changing signals.

By exploiting some strange material properties, the chips could create a new kind of artificial intelligence – not formed by coded algorithms but generated by the physical architecture of the computer.

“There’s extraordinary interest in building energy-efficient electronics these days,” says principal investigator Shriram Ramanathan, associate professor of materials science at Harvard. Our brain is easily the most efficient device around, running all of its phenomenal computing power on roughly 20 Watts of electricity.

“The transistor we’ve demonstrated is really an analogue to the synapse in our brains,” says co-lead author Jian Shi, a postdoctoral student.

“Each time a neuron initiates an action and another neuron reacts, the synapse between them increases the strength of its connection. And the faster the neurons spike each time, the stronger the synaptic connection. Essentially, it memorizes the action between the neurons.”

A system integrating millions of the tiny synaptic transistors and neuron terminals could create ultra-efficient parallel computing capabilities.

It has been made possible by the use of ‘nickelate’, a material from the class known as ‘correlated electron systems’, which can suddenly change its conductance when exposed to certain fields.

The mind-moulded device consists of the nickelate semiconductor squished between platinum electrodes, next to a small pocket of ionic liquid. An external circuit multiplexer converts the time delay into a magnitude of voltage which it applies to the ionic liquid, creating an electric field that either drives ions into the nickelate or removes them. The entire device, just a few hundred microns long, can be embedded in a silicon chip.

“This system changes its conductance in an analogue way, continuously, as the composition of the material changes,” explains Shi.

“It would be rather challenging to use CMOS, the traditional circuit technology, to imitate a synapse, because real biological synapses have a practically unlimited number of possible states—not just ‘on’ or ‘off.’”

“We exploit the extreme sensitivity of this material,” says Ramanathan, referring to the nickelate material

“A very small excitation allows you to get a large signal, so the input energy required to drive this switching is potentially very small. That could translate into a large boost for energy efficiency.”

“The beauty of this type of a device is that the 'learning' behaviour is more or less temperature insensitive, and that’s a big advantage...we can operate this anywhere from about room temperature up to at least 160 degrees Celsius.”

“In our proof-of-concept device, the time constant is really set by our experimental geometry... in other words, to really make a super-fast device, all you’d have to do is confine the liquid and position the gate electrode closer to it,” says Ramanathan.

Details of the new device are available from the journal Nature Communications.