Biology is inspiring a new type of water-based circuit that could transform computing

The future of neural network computing might be a little more sodden than we anticipated.

A team of physicists has successfully developed an ion circuit – a processor based on the motion of charged atoms and molecules in an aqueous solution rather than electrons in a solid semiconductor.

Since this is closer to how the brain carries information, their device could be the next step towards brain-like computers.

“Ionic circuits in aqueous solutions try to use ions as charge carriers for signal processing,” writes the team led by physicist Woo-Bin Jung from the Harvard John A. Paulson School of Engineering and Applied Sciences (SEAS) in a new paper.

“Here we report an aqueous ion circuit… This demonstration of the functional ion circuit capable of performing analogous calculations is a step towards a more sophisticated aqueous ion technique.”

Much of the signaling in the brain is the movement of charged molecules called ions through a liquid medium. Although it’s extremely difficult to reproduce the brain’s incredible processing power, scientists have thought that a similar system could be used for computers: pushing ions through an aqueous solution.

This would be slower than traditional silicon-based computing, but could have some interesting benefits.

For example, ions can be made from a variety of molecules, each with different properties that could be used in different ways.

But first scientists have to show that it can work.

This is what Jung and his colleagues worked on. The first step was to develop a working ionic transistor, a device that would switch or amplify a signal. Their most recent advance was combining hundreds of these transistors to work together as an ionic circuit.

The transistor consists of a “bullseye” array of electrodes with a small disc-shaped electrode in the center and two concentric ring electrodes around it. This borders on an aqueous solution of quinone molecules.

A voltage applied to the central disk creates a flow of hydrogen ions in the quinone solution. Meanwhile, the two ring electrodes modulate the pH of the solution to increase or decrease the ion current.

This transistor performs a physical multiplication of a “weight” parameter set by the gating of the ring pair by the plate voltage and produces an ionic current response.

However, neural networks rely heavily on a mathematical operation called matrix multiplication, which involves multiple multiplications.

So the team designed 16-by-16 arrays of their transistors, each capable of arithmetic multiplication, to make an ionic circuit capable of performing matrix multiplication.

“Matrix multiplication is the most widely used computation in neural networks for artificial intelligence,” says Jung. “Our ionic circuit performs matrix multiplication in water in an analogous manner, based entirely on electrochemical machines.”

Of course, there are significant limitations of the technology. The 16 streams cannot be resolved separately, meaning the operation had to be performed sequentially rather than simultaneously, significantly slowing down an already relatively slow technology.

However, its success is a step towards more sophisticated ionic computing: only by seeing the problem can we find solutions.

The next step will be to introduce a wider range of molecules into the system to see if this allows the circuit to process more complex information.

“Until now, we have only used 3 to 4 ion species, such as hydrogen and quinone ions, to enable gating and ion transport in the aqueous ion transistor,” says Jung.

“It will be very interesting to use more diverse ionic species and see how we can use them to make the content of the information to be processed richer.”

The end goal, the team says, is not to compete with ions or to replace electronics with ions, but to complement the capabilities of both, perhaps in the form of hybrid technology.

The research was published in Advanced Materials.