Neural network computing could be a little more advanced than we thought.
An ionic circuit is a processor based on the movements of charged atoms and molecule in a solution rather than electrons in a solid Semiconductor has been developed by a team of physicists
Their device could be the next step in brain-like computing since it is closer to how the brain transports information.
In a new paper, a team led by physicist Woo-Bin Jung of the Harvard John A. Paulson School of Engineering and Applied Sciences explain how ion circuits can be used to process signals.
This demonstration of the functional ionic circuit capable of analog computing is a step in the right direction.
The ion movement in the brain is a major part of signal transmission. Scientists think that a similar system could be used for computing: pushing ion through a solution.
It might have some interesting advantages over conventional, Silicon-based computing.
Different properties of different molecule can be exploited in different ways.
Scientists need to demonstrate that it works.
Jung and his coworkers have been working on this. A functional ionic transistor is a device that can switch on or off. Combining hundreds of transistors was their most recent advancement.
The transistor consists of a "bullseye" arrangement of electrodes, with a small disk-shaped one in the center. The interface is with a solution of quinones.
There is a current of hydrogen ion in the solution. The ionic current can be increased or decreased by the two ring electrodes.
The answer to the ionic current is produced by the multiplication of the "weight" parameters set by the ring pair.
Neural networks rely on a multiplication operation called matrix multiplication.
To create an ionic circuit that can perform matrix multiplication the team designed 16-by-16 array of their transistors.
Jung says matrix multiplication is the most common calculation in neural networks. The matrix multiplication in water is performed by the ionic circuit in an analogue way.
Significant limitations to the technology are present. The operation had to be sequential rather than concurrent because the 16 currents can't be separated.
It's only by seeing the problem that we can find solutions that it's a step closer to more sophisticated ionic computing.
The next step is to see if introducing a wider range of molecules will allow the circuit to process more complicated information.
Jung says that they have only used 3 to 4 ionic species to enable the gating and ionic transport in the transistor.
It will be fascinating to see how we can use more diverse ionic species to make more money.
The goal is not to compete with or replace electronics with ionics, but to complement, perhaps in the form of hybrid technology with the capabilities of both.
The research has appeared in a journal.
All Rights Reserved © 2024
