Projects
The development of emitting nanomaterial for light-emitting-diode structures and optoelectronic integrated circuits
In 2012-2013, the emitting nanomaterial development project for light-emitting-diode structures and optoelectronic integrated circuits was completed. It was supported by an international grant (Russia - Belarus) of the Russian Foundation for Basic Research (RFBR).
********************************************
Innovative technology
«Biomorphic neuroprocessor based on a combined memristor-diode crossbar for simulation of the brain cortical column operation»
In 2018 the SEC presented an innovative technology «Biomorphic neuroprocessor based on a combined memristor-diode crossbar for simulation of the brain cortical column operation» within the framework of the «Information technology (IT). Artificial Intelligence» category.
The innovative technology was created within the framework of the National Technological Initiative NeuroNet, approved on 09.06. 2015 at a meeting of the Presidium of the Presidential Council for Economic Modernization and Innovative Development of Russia. According to the NeuroNet roadmap presented at this meeting, by 2020, the first prototypes of the neuromorphic chips based on memristor technologies, as well as of the first neuromorphic computers should be developed.
We have designed an original software-based biomorphic model of a neuron. The model consists of separate functional parts and allows implementing connections between the functional parts of different neurons, which makes the modeled neurosystem very flexible.
We have formulated the principles of an ultra-large biomorphic neural network based on the aforementioned software-based neuron model. We have also adapted the neural network to the neuroprocessor.
We have developed the key elements of the neuroprocessor hardware - the memory and the logic matrices, based on the elaborate neuron electrical model.
We have presented the electrical circuits, the planar and 3D topology, and the nanotechnological basis for the manufacture of these matrices. They were based on a new nanoelectronic component – a combined memristor-diode crossbar.
Fig.1. Crossbar (a) and memristor microcircuit (b).
Fig.2. Laboratory prototype of a neuroprocessor: 1 - PIC18 controller, 2 – memristor microcircuit, 3 - I2C interface, 4 - USB interface.
We have performed SPICE modeling of how the hardware biomorphic neural network of the neuroprocessor simultaneously processes signals and engages in associative self-learning. As a result, new associations (new knowledge) were successfully formed.
Apart from the generation of new knowledge related to the creation of artificial intelligence systems, this specialized hardware is helping achieve specific technical objectives. For example, enhancing the operating speed and the energy efficiency of calculations in comparison to the currently existing computational means (servers and super-computers). This is achieved by relying on mixed analog-digital computations, including those that use new electronic elements – memristors, integrated into the composite memristor-diode crossbars.
In 2019, with the help of the nanotechnological complex «NanoFab-100», the SEC conducted experimental manufacturing and testing of a neuroprocessor memory matrix, consisting of memristor synapses.
The above research was supported by the RFBR grant No. 19-07-00272.
Research on generating new knowledge and on the improvement of the operating speed and the energy efficiency of information processing on a specialized device (in comparison with currently existing computational means of high performance) was financially supported by the RFBR grant «Aspirant-2019», No. 19-37-90030.
The manufacture of the biomorphic neuroprocessor prototype was undertaken jointly, by the «Angstrem-T» manufacturing (Moscow, Zelenograd), and the SEC Nanotechnologies, using the nanotechnological complex NanoFab-100 installed at Tyumen State University by ZAO «NT-MDT» (Zelenograd, Russia), within the framework of the Federal Target Program.
There are no analogues for the developed biomorphic neuroprocessor in Russia or abroad.
The uniqueness of the biomorphic neuroprocessor lies in the fact that it is based on a modernized biomorphic electric model of a neuron. Also, the neuroprocessor is biomorphic because it performs neural network functions – a neural network founded on an original biomorphic software-based model of a neuron.
At the moment, the information technologies (IT) sector is developing artificial neural networks based on simple neurons, as well as hardware for processing audio and video signals. In such devices, decision making occurs as a result of the analysis of information stored in memory units and the choice of the most plausible decision.
The presented biomorphic neuroprocessor is competitive in the nanoelectronics market and is more advanced than the neuroprocessor technologies produced by the industry leaders, such as IBM and Hewlett-Packard. The innovative technology will be widely implemented when current personal computers are replaced with computers of a new generation capable of new knowledge creation. This technology has a wide range of applications. First of all, it will facilitate decision making in the context of a dynamically changing informational context. It will also ease the process of managing complex dynamic objects.
In 2019, for the development of a biomorphic neuroprocessor, Sergey Udovichenko, the head of the SEC Nanotechnologies, won the Tyumen regional competition titled «Leader in Scientific and Innovative Activity» in the «Scientist of the year in innovation» nomination.
********************************************
Manufacturing high-performance glassfiber catalysts
Our research team has developed the technology for manufacturing high-performance glassfiber catalysts, based on the magnetron sputtering of thin, high adhesion, uniform in composition transition metal oxide films to the carrier.
The catalysts are designed for environmentally friendly disposal of liquid hydrocarbon waste and sludge, resulting in decreased gas emissions (in case of their possible further processing).
********************************************
Morphological study of insects and mites
Morphological properties of insects and mites are studied with the help of a scanning electron microscope, after the samples are coated in gold in an automated metal sputter coating system Quorum Q 150R.
|
|
|
