Microprocessors in smartphones, computers, and data centers process information using electrons through solid semiconductors, but our brains (Computer neural networks and water) have a different mechanism. They rely on the organization of ions and liquids to process information. Inspired by the brain, researchers wanted to create “ionics” in water solutions.
Although ions in water move more slowly than electrons in semiconductors, scientists believe that different types of ions with different physical and chemical properties can be manipulated for rich and complex information processing. good. However, ion computing is still in its infancy.
So far, laboratories have developed various ion devices such as ion diodes and transistors, but no one has assembled many such devices into a complex computer circuit – until now. A group of researchers from the Harvard John A. Paulson School of Engineering and Applied Sciences (SEAS), in collaboration with DNA Script, a biotech startup, developed an ion circuit (Computer neural networks and water) with hundreds of ion transistors and performed the basic calculations of a neural network.
This research is published in high quality materials.
The researchers started by building a new type of ion transistor using their recently developed method. A transistor consists of an aqueous solution of quinone molecules, which has two concentric annular electrodes with a central electrode in the form of a disk, like a bull’s eye. The two ring electrodes descend and adjust the local pH around the central disc by producing and trapping hydrogen ions. An electric current mounted on the central disk causes an electrochemical reaction to produce an ion current through the disk into the water.
The reaction rate can be sped up or slowed down — by increasing or decreasing the ion current — by adjusting the local pH. In other words, the pH controls, or triggers, the ionic current of the disc in the aqueous solution, creating the ionic counterpart of the electronic transistor.
They then designed a pH-triggered ion transistor so that the disk current is a quantile of the disk voltage and the “range” represents the pH of the region triggering the transistor. They organized these transistors into a 16 × 16 network to spread the analog numbers of individual transistors into an analog matrix, with the network of local pH values serving as the weight matrix in the network. nerves.
“Matrix growth is the most popular collection of neural networks for artificial intelligence,” said Woo-Bin Jung, a postdoctoral fellow at SEAS and first author of the paper. “Our ion circuit does the multiplication of the matrix and the water in an analog way based entirely on the electrochemical mechanism.”
“Microprocessors manipulate digital electronics to operate multiple matrices,” said Donhee Ham, Gordon McKay Professor of Electrical Engineering and Applied Physics at SEAS and co-author of the paper. “Although our ion circuits cannot be as fast or precise as digital microprocessors, the enrichment of the electrochemical matrix in water is interesting in itself and can be very efficient.”
Now the group wants to improve the chemical complexity of the system.
“So far, we have only used 3-4 types of ions, such as hydrogen and quinone ions, to enable ion gating and migration in liquid ion transistors,” Jung said. “It will be very interesting to use different ionic species and to see how we can use them to make the content of the information to be resolved.”
