Any modern computer or smartphone has a processor — a universal device designed to execute programs. The principles of operation of a traditional processor were laid down in the forties of the last century and have not changed much since then: the CPU reads commands and executes them one at a time per cycle. On modern processors, these cycles can be several billion per second, and due to this, it seems to a person that his computer performs many tasks simultaneously, but he still performs them in turn.

Neuromorphic processors work radically differently — they repeat the structure of the human brain, which is really multitasking. Recently, a similar processor was launched into mass production by the Russian company Kaspersky, so neuromorphic electronics is no longer science fiction. However, this technology is still in great need of new research and discoveries that can multiply its effectiveness.

The laboratory "Neuroelectronics and Memristive Nanomaterials" (Neuromena) is engaged in this frontier of scientific research at the Southern Federal University. Most recently, a junior researcher at this laboratory, Daniil Khakhulin, spoke at the "School of Young Scientists" within the framework of the All-Russian Forum "Microelectronics-2023". His report was recognized as the best among young scientists in Russia in the section "Artificial intelligence, digital twins and neuropodic systems", the work was devoted to lithium niobate as a promising neuromorphic material.

"Despite the external differences, fundamentally integrated circuits and the brain are very similar: in both systems, information is presented in the form of a charge, the passage of which is limited by energy barriers and directed by modulation of conducting channels," said Daniil Khakhulin.

The basic element in the human brain is a neuron, or nerve cell. Neurons are interconnected by means of synapses. Several tens of billions of neurons make up a complex self-learning system, which is still far from any computer. In computing, the smallest "brick" is considered to be a transistor: there are several billion such microscopic elements in a typical computer or smartphone processor. The discipline known as neuromorphic engineering has for several decades set itself the task of reproducing, at least partially, the structure of the human brain in the form of electronic circuits.

At the same time, most experts in the world agreed that the best material for creating a neuromorphic processor — a computer with "neurons" and "synapses" — is hafnium oxide. The results of the research of the laboratory "Neuromena" of the Southern Federal University show that in some cases it makes sense to consider other materials.

"Lithium niobate (LiNbO3) films obtained by pulsed laser deposition meet the same requirements for a number of parameters as other materials for synaptic devices. If we continue experimenting with this compound, getting new composites based on it, then it is very likely that one day we will get a material superior to analogues, and neuromorphic processors will be made from it," Daniil Khakhulin stressed.

SFedU scientists already have on hand not only samples of new material, but also innovative methods of managing its condition. By changing the current, voltage and pulse duration, the staff of the Neuromena laboratory at SFedU learned to control the resistance and plasticity of lithium niobate.

"A number of materials have gained more attention of researchers in solving problems of neuromorphic microelectronics primarily due to the combination of optimal electrophysical properties and manufacturability. However, this does not exclude the search for new materials to improve the performance of the end devices of neuroelectronics. The main advantage and at the same time disadvantage of our LiNbO3 films is the high resistance. However, the flexibility of pulsed laser deposition technology allows us to produce composite films by varying the impurity concentration both in film thickness and on its surface. This opens the way to the search for a LiNbO3—based composite with the best combination of properties in relation to the tasks of neuromorphic electronics," Daniil Khakhulin summed up.

The laboratory "Neuroelectronics and Memristive Nanomaterials" (Neuromena) was created within the framework of the megagrant project and meets the objectives of the strategic project of the SFedU "Intelligent control and Information Processing technologies in advanced robotic complexes and hybrid systems" of the federal program "Priority-2030" (national project "Science and Universities"). Its fundamental goal is to develop the element base of hybrid neuromorphic systems based on biocompatible memristive nanomaterials and composites based on them. According to scientists, the laboratory's developments can be used in robotic systems, neuroprocessor computer architecture of a new generation and wherever there is a possibility of implementing artificial intelligence systems.