In the Tyumen State University, as part of the modeling of neural networks of the brain, performed TSU and LLC "TASO" received
Memristor acts as a resistance whose value varies depending on the current passing through it. By lowering the resistance memristor connection between logic elements, which is involved in the formation of this memristor is more effective — learning takes place. With increasing resistance of the memristor is forgetting or braking. Read the information recorded in the memristor can be measured by its resistance. By its action is similar to the memristor synapse — the connection between nerve cells in the brain (neurons).
Work on memristivnoy chip in a clean room of Tyumen State University, March 2012
The existence of the memristor predicted in 1971 a researcher from the University of Berkeley, Leon Chua, but in practice, the memristor was the world’s first produced in 2008 in the Hewlett-Packard Corporation team led by R. Stanley Williams.
As the TSU rector of Innovation Development and Information Technology, General Director of "TASO" and the project manager to create neuromorphic systems and modeling of neural networks of the brain Vadim Filippov, "… We used to be able to create a software model based on powerful supercomputers, the memristor to give us an opportunity to create a self-learning memristivnye chip that will be embedded in the hardware and software systems that simulate the brain’s cortex. Creating a memristor likely eventually lead to a complete review and revision of the entire architecture of modern computers, bringing them to the associative organization of living systems. "
Memristors in TSU derived titania (TiO2) — semiconductor in pure form having a high resistance. However, when alloyed with other elements of TiO2, these dopants (for example oxygen ions), high electric field can drift in the direction of the electric current. Bias voltage supply through a thin film of titanium dioxide causes the impurities spread in the amount of TiO2 and thus lowers the resistance. The power supply in the other direction moves back impurities, increasing resistance of TiO2. Memristor can take not only two positions — either 0 or 1, but any other value, working in analog mode. And this is one element that can reduce the size of the memristor to a few nanometers, and response speed — up to nanoseconds.
Memristors were synthesized using the TSU acquired a modular technology platform for the formation of complexes with nanotechnology cluster layout "NANOFAB-100." For memristor were deposited on the substrate conductive paths cross, which was covered by a layer of titanium dioxide (TiO2) with thickness of 15 nanometers. Over it were applied longitudinal conductive tracks. At the intersections of tracks obtained memristivny pronounced effect of reducing the resistance and potentiation of communication depending on the lead current.
The principal difference between the memristor from most types of modern semiconductor memory and its main advantage over them is the fact that it does not store its properties as a charge. Memristor is not afraid of leakage of charge in the transition to nanometer-scale integrated circuits, and it is completely nonvolatile, and the data can be stored in the memristor as long as there are the materials from which it is made. Memristors today are the only non-living material, approaching in its functions to the synapses of the living brain. Synaptic connections can be modeled and CMOS transistors, transistor circuits but incomparably more slowly, more complex energy-consuming and more expensive than memristivnye.
According to the materials, LLC "TASO"