Following Elon Musk's announcement of the Neuralink project, the attention of many has focused on the development of new nanotechnology in the fields of medicine and bioengineering. In recent months, a team of researchers at the Micro Nano Research Facility (MNRF) in Melbourne has created a nanocell capable of simulating the human brain. No device had so far come so close to the real behavior of our brain, as it is able to base its evolution on previous experiences.
THE TURNING POINT: THE MEMRISTOR
The ultrafast cell developed at MNRF works as a memristore, a passive nonlinear component capable of changing its electronic characteristics according to the voltage applied to it. Unlike traditional CMOS cells (which can represent only two logical states, namely '1' and '0'), it allows the creation of multilevel memory cells, which are well suited to the simulation of synaptic junctions. Just think of an electric dimmer that unlike a normal on / off switch, allows you to adjust at will the brightness of the bulb.
When a potential difference is applied to the electrodes, a current flow is obtained and a consequent change in the conduction capacity of the material, thanks to which it is possible to emulate an important biological function of synapses, namely synaptic plasticity (ability of the nervous system to change the intensity of the relationships between neurons, to establish new ones and to eliminate some of them).
The change in resistive state can be maintained even after the electrical signal is removed. In this way, the memristor is able to preserve the history of previously applied electrical stimuli.
Similarly, the human brain, based on past experience, modifies the tendency of the electrical impulse to travel through each interneural connection: the so-called "synaptic weight" is represented by the variation of the resistance of the memristor.
MATERIALS AND STRUCTURE
The electronic cell was created on a nanofilm composed of an alloy of oxidized metals, including Titanium and Strontium, which following previous studies had proved to be the most suitable for applications of this type. Oxidizing the semiconductor (Sn, Strontium) with molecules of TiO3, we obtain a material with excellent memristive properties, which can be amplified with a further doping by acceptor atoms.
CONCLUSION
The development of this new technology opens the way not only to structures able to host more and more complex neural networks and artificial intelligences or to the design of smaller but at the same time larger memories, but also to a progress in the biomedical field: getting closer and closer to the real electrical functioning of the human brain, it would be possible to observe more closely the evolution of neurodegenerative diseases and, at the same time, to conduct experiments without worrying about ethical barriers. So we can expect a continuous and growing development of memristor technologies, which will become more and more used in every technological field for its immense potential.
Written by Sebastiano Vittoria e Marco Vitali del VGen Engineering Hub
