A group of American scientists successfully collected the ion chip, which consists of transistors operating in a liquid environment, and current flow is not generated by electrons, as in the case of solid semiconductor transistors, but by charged molecules and atoms.
The project was developed by a group of scientists led by Wu Bin Jun from the John Paulson Harvard School of Engineering and Applied Sciences. The transmission of brain signals is done through ions in liquid environments. It is still extremely difficult to reproduce the computational productivity of the human brain, and silicon components are still showing higher performances, but a simplified version of the scheme has been created, and in the long run it can offer its advantages. For example, ions can be created from different molecules, and in each case they will have different properties and their scope of application.
In the first phase, engineers built a functioning ion transistor, a component that controls the input signal, and then several hundred such transistors combined into a coherent ion mixer. The ion transistor consists of three electrodes: one disc-shaped in the centre and two rings around it. When the voltage is applied to the central disc, the electromechanic reaction is formed by the ion current from it in the direction of the liquid environment. The speed of the reaction can be controlled by changing the pH of the environment, which occurs when the ring electrodes are captured or, on the contrary, produced hydrogen ions themselves. This allows the transistor to perform a multiplication operation, and when they are combined into a array of 16x16 dimensions, the diagram makes it possible to multiply the matrix, the most common artificial intelligence operation.
The technology has significant limitations in its current performance. For example, it is not possible to produce all 16 conclusions simultaneously, i.e. operations have to be performed in sequence, which further slows down the already slow components. However, the authors have managed to achieve the basic work of the model and will now improve: for example, they plan to introduce a wider range of molecules into it, which in theory will allow them to process more complex information.
The authors of the study are not going to replace electronics with ionics; the new technology can complement existing solutions or create a hybrid with the capability of both approaches.