Together with their colleagues from Germany and the Netherlands, scientists at the Moscow Institute of Physics and Technology (MIPT) have found a way to significantly improve computer performance. In their paper published in Nature Photonics, they propose the use of the so-called T-waves, or terahertz radiation as a means of resetting computer memory cells. This process is several thousand times faster than magnetic-field-induced switching.
“We have demonstrated an entirely new way of controlling magnetization, which relies on short electromagnetic pulses at terahertz frequencies. This is an important step towards terahertz electronics. As far as we know, our study is the first to make use of this mechanism to trigger the oscillations of magnetic subsystems,” says Anatoly Zvezdin of Prokhorov General Physics Institute and MIPT, a coauthor of the paper and a USSR State Prize-winning scientist heading MIPT’s Laboratory of Physics of Magnetic Heterostructures and Spintronics for Energy-Saving Information Technologies.
The rapidly increasing amounts of digital data that have to be manipulated, along with the growing complexity of the computation tasks at hand, compel hardware designers to achieve ever higher computational speeds. Many experts believe that classical computation is currently approaching a limit, beyond which no further increase in data processing speed will be practicable. This is motivating scientists all over the world to investigate possibilities of entirely different computer technologies. One of the weak spots in modern computers holding back their evolution is memory: it takes time to complete every set/reset operation for a magnetic memory cell, and reducing the duration of this cycle is a very challenging task.
A group of scientists including Sebastian Baierl of the University of Regensburg, Anatoly Zvezdin, and Alexey Kimel of Radboud University Nijmegen (the Netherlands) and Moscow Technological University (MIREA) proposed that electromagnetic pulses at terahertz frequencies (with wavelengths of about 0.1 millimeters, i.e., between those of microwaves and infrared light) could be used in memory switching instead of external magnetic fields. A more familiar device that makes use of terahertz radiation is the airport body scanner. T-rays can expose weapons or explosives concealed under a person’s clothing, without causing any harm to live tissues.
To find out whether T-rays could be used for convenient memory states switching (storing “magnetic bits” of information), the researchers performed an experiment with thulium orthoferrite (TmFeO?). As a weak ferromagnet, it generates a magnetic field by virtue of the ordered alignment of the magnetic moments, or spins of atoms in the microcrystals (magnetic domains). In order to induce a reorientation of spins, an external magnetic field is necessary.
However, the experiment has shown that it is also possible to control magnetization directly by using terahertz radiation, which excites electronic transitions in thulium ions and alters the magnetic properties of both iron and thulium ions. Furthermore, the effect of T-rays proved to be almost ten times greater than that of the external magnetic field. In other words, the researchers have devised a fast and highly efficient remagnetization technique—a solid foundation for developing ultrafast memory.
The scientists expect their “T-ray switching” to work with other materials as well. Thulium orthoferrite, which was used in the experiment, happens to be convenient for the purposes of demonstration, but the proposed magnetization control scheme itself is applicable to many other magnetic materials.
“There was a Soviet research group that used orthoferrites in their studies, so this was always kind of a priority field for us. This research can be seen as a follow-up on their studies,” says Anatoly Zvezdin.
It prepares specialists in theoretical and applied physics, applied mathematics, and related disciplines. It is sometimes referred to as “the Russian MIT.”
MIPT is famous in the countries of the former Soviet Union, but is less known abroad. This is largely due to the specifics of the MIPT educational process (see “Phystech System” below). University rankings such as The Times Higher Education Supplement are based primarily on publications and citations. With its emphasis on embedding research in the educational process, MIPT “outsources” education and research beyond the first two or three years of study to institutions of the Russian Academy of Sciences. MIPT’s own faculty is relatively small, and many of its distinguished lecturers are visiting professors from those institutions. Student research is typically performed outside of MIPT, and research papers do not identify the authors as MIPT students. This effectively hides MIPT from the academic radar, an effect not unwelcome during the Cold War era when leading scientists and engineers of the Soviet arms and space programs studied there.
The word “phystech,” without the capital P, is also used in Russian to refer to Phystech students and graduates.
The main MIPT campus is located in Dolgoprudny, a northern suburb of Moscow. However the Aeromechanics Department is based in Zhukovsky, a suburb south-east of Moscow.
Moscow Institute of Physics and Technology (MIPT) research articles from Innovation Toronto
- Personalized medicine will employ computer algorithms – June 16, 2016
- New 3D hydrogel biochips prove to be superior in detecting bowel cancer at early stages – May 29, 2016
- New type of graphene-based transistor will increase the clock speed of processors by orders of magnitude – May 21, 2016
- Physicists build “electronic synapses” for neural networks – April 22, 2016
- Scientists develop a 100 times faster type of memory cell – March 21, 2016
- Physicists get a perfect material for air filters with many possible uses – March 2, 2016
- Physicists promise a copper revolution in nanophotonics – February 27, 2016
- Scientists build a neural network using plastic memristors – January 28, 2016
- Faster computers via new metamaterial – January 3, 2016
- Scientists Find Way to Maintain Quantum Entanglement in Amplified Signals – July 30, 2014
- Nature Inspires New Submarine Design
Russian scientists at the Moscow Institute of Physics and Technology (MIPT), the Joint Institute for High Temperatures of the Russian Academy of Sciences (JIHT RAS), and Gamaleya Research Centre of Epidemiology and Microbiology found that treating cells with cold plasma leads to their regeneration and rejuvenation. This result can be used to develop a plasma therapy program for patients with non-healing wounds.
The paper has been published in the Journal of Physics D: Applied Physics.
Non-healing wounds make it more difficult to provide effective treatment to patients and are therefore a serious problem faced by doctors. These wounds can be caused by damage to blood vessels in the case of diabetes, failure of the immune system resulting from an HIV infection or cancers, or slow cell division in elderly people. Treatment of non-healing wounds by conventional methods is very difficult and in some cases impossible.
Cold atmospheric-pressure plasma refers to a partially ionized gas (the proportion of charged particles in the gas being close to 1?%) with a temperature below 100,000 K. Its application in biology and medicine has been made possible by the advent of plasma sources generating jets at 30–40?°C.
An earlier study established the bactericidal properties of low-temperature plasma, as well as the relatively high resistance of cells and tissues to its influence. The results of plasma treatment of patients with non-healing wounds varied from positive to neutral. The authors’ previous work prompted them to investigate the possibility that the effect of plasma treatment on wound healing could depend on application pattern (the interval between applications and the total number of applications).
Two types of cells were used in this study, viz. fibroblasts (connective tissue cells) and keratinocytes (epithelial cells). Both play a central role in wound healing.
The effect of plasma treatment on cells was measured. In fibroblast samples, the number of cells increased by 42.6?% after one application (A) and by 32.0?% after two applications (B), as compared to the untreated controls. While no signs of DNA breaks were detected following plasma application, an accumulation of cells in the active phases of the cell cycle was observed, alongside a prolonged growth phase (30 hours). This means that the effect of plasma could be characterized as regenerative, as opposed to harmful.
The proliferation of cells that had been treated daily over a period of three days (group C) was reduced by 29.1?% relative to the controls. Keratinocytes did not show noticeable changes in proliferation.
The researchers also performed an assay of the senescence-associated ?-galactosidase, which is measured at pH 6.0. The concentration of this enzyme in a cell increases with age. Plasma treatment significantly reduced the content of this substance in the samples. This, together with a prolonged exponential growth phase of the culture, suggests a functional activation of cells—their rejuvenation.
‘The positive response to plasma treatment that we observed could be linked to the activation of a natural destructive mechanism called autophagy, which removes damaged organelles from the cell and reactivates cellular metabolic processes,’ says Elena Petersen, a co-author of the paper and the head of the Laboratory of Cellular and Molecular Technologies at MIPT.
The scientists are planning additional research into the molecular mechanisms underlying the effects of plasma on cells. They also aim to determine the influence of a patient’s age on the effectiveness of plasma therapy.
Learn more: Cold plasma will heal non-healing wounds
Physicists from the Moscow Institute of Physics and Technology (MIPT) have found that the two-dimensional form of carbon, known as graphene, might be the ideal material for manufacturing plasmonic devices capable of detecting explosive materials, toxic chemicals, and other organic compounds based on a single molecule, says an article published in Physical Review B.
Scientists have long been fascinated by the potential applications of a quasiparticle called the plasmon, a quantum of plasma oscillations. In the case of a solid body, plasmons are the oscillations of free electrons. Of special interest are the effects arising from the surface interactions of electromagnetic waves with plasmons—usually in the context of metals or semimetals, as they have a higher free electron density. Harnessing these effects could bring about a breakthrough in high-accuracy electronics and optics. One possibility opened up by plasmonic effects is the subwavelength light focusing, which increases the sensitivity of plasmonic devices to a point where they can distinguish a single molecule. Such measurements are beyond what any conventional (classical) optical devices can achieve. Unfortunately, plasmons in metals tend to lose energy quickly due to resistance, and for this reason they are not self-sustained, i.e. they need continuous excitation. Scientists are trying to tackle this issue by using composite materials with predefined microstructure, including graphene.
Graphene is an allotrope of carbon in the form of a two-dimensional crystal. It can be visualised as a one-atom-thick honeycomb lattice made of carbon atoms. Two MIPT graduates, Andre Geim and Konstantin Novoselov, were the first to isolate graphene, which won them a Nobel Prize in Physics. Graphene is a semiconductor with extremely high charge carrier mobility. Its electrical conductivity is also exceptionally high, which makes graphene-based transistors possible.
Although, plasmonic devices have seemed an exciting prospect to pursue from the start, to take advantage of them, it was first necessary to find out whether the technology behind them was feasible. To do this, scientists had to find a numerical solution to the relevant quantum-mechanical equations. This was accomplished by a team of researchers at the Laboratory of Nanostructure Spectroscopy headed by Prof. Yurii Lozovik: they formulated and solved the necessary equation. Their research has led them to develop a quantum model that predicts plasmonic behaviour in graphene. As a result, the scientists described the operation of a surface-plasmon-emitting diode (SPED) and the nanoplasmonic counterpart of the laser—known as the spaser—whose construction involves a graphene layer.
A spaser could be described as a device similar to a laser and operating on the same basic principle. However, to produce radiation, it relies on optical transitions in the gain medium, and the particles emitted are surface plasmons, as opposed to photons produced by a laser. An SPED is different from a spaser in that it is an incoherent source of surface plasmons. It also requires considerably lower pump power. Both devices would operate within the infrared region of the spectrum, which is useful for studying biological molecules.
‘The graphene spaser could be used to design compact spectral measurement devices capable of detecting even a single molecule of a substance, which is essential for many potential applications. Such sensors could detect organic molecules based on their characteristic vibrational transitions (‘fingerprints’), as the light emitted/absorbed falls into the medium infrared region, which is exactly where the graphene-based spaser operates,’ says Alexander Dorofeenko, one of the authors of the study.
Learn more: ‘Sniffer plasmons’ could detect explosives
Research findings prove the capabilities of silicon nanoparticles for flexible data processing in optical communication systems
A team of physicists from ITMO University (Saint Petersburg) and Moscow Institute of Physics and Technology (MIPT) has demonstrated the potential of silicon nanoparticles for effective non-linear light manipulation. Their work lays the foundation for the development of novel optical devices with a wide range of functionalities. These silicon nanoparticles based devices would allow to transmit, reflect, or scatter incident light in a specified direction, depending on its intensity. They could be integrated into microchips that would enable ultrafast all-optical signal processing in optical communication lines and the next generation optical computers.
Non-linear antennas Electromagnetic waves of a wide spectral range are used to transmit information: from radio waves that carry radio signals over the air to infrared radiation and visible light used in telecommunications to transfer data through fibre optics. An essential component of any equipment that relies on electromagnetic waves for information transmission and processing is a device called the antenna, which is designed to either receive or transmit signals in a particular direction. It is often the case that incoming signals need to be flexibly processed. This requires the use of a reconfigurable antenna, i.e. one whose characteristics (e.g. its radiation pattern) can be changed in a specific manner during signal processing. One possible solution relies on the use of a non-linear antenna, which can be switched by the incident light itself.
Denis Baranov, a PhD student at MIPT and one of the authors of the study, comments on the research findings: ‘It is a top priority — and at the same time a major challenge — to develop such tuneable antennas operating at infrared and optical frequencies. Nowadays, we can already transmit information through fibre optics at incredible speeds of up to hundreds of Gbit/s. However, silicon-based electronics are unable to process the incoming data at that rate. Non-linear nanoantennas that work at optical wavelengths could help us to resolve this problem and make ultrafast all-optical signal processing possible.’
Silicon nanoparticles To demonstrate non-linear switching, the authors of the paper, which was published in ACS Photonics, have studied a dielectric nanoantenna — an optically resonant spherical nanoparticle made of silicon. While spherical particles of all sizes show resonances, it is the size of the particle that determines its resonant wavelength. The first of these resonances, which can be observed at the longest wavelength, is the magnetic dipole resonance. Incident light of a certain wavelength induces a circular electric current in the particle, similar to the current that flows in a closed circuit. Because silicon has a high refractive index, particles with diameters approaching 100 nm will already show the magnetic dipole resonance at optical frequencies, making them useful for enhancing various optical effects at the nanoscale. The team has used silicon nanosphere resonances to enhance Raman scattering in an earlier study, which is detailed in another article.
The optical properties of a non-linear silicon nanoantenna are manipulated by means of electron plasma generation (Fig. 2). As silicon is a semiconductor, there are almost no electrons in its conduction band under normal conditions. However, exposing it to a laser pulse of high intensity and very short duration (≈100 femtoseconds, i.e. about 10⁻¹³ or one ten-trillionth of a second) excites the electrons into the conduction band. This significantly alters the properties of the material as well as the behaviour of the silicon nanoantenna itself, causing it to scatter light in the direction of the incident pulse. Thus, by exposing a particle to a short and intense pulse, its behaviour as an antenna can be dynamically controlled.
In order to demonstrate ultrafast nanoantenna switching, the authors of the study carried out a series of experiments, which involved the irradiation of an array of silicon nanoparticles with a short and intense laser pulse and a continuous measurement of their transmittance. The researchers observed that the transmission coefficient of a structure changed by several per cent within 100 femtoseconds and then gradually returned to its initial value.
On the basis of the experimental results, the researchers went on to develop an analytical model that describes the ultrafast non-linear dynamics of the nanoantenna examined in the study, as well as the generation and relaxation of electron plasma in silicon. According to the model, a radical change in the scattering diagram of the antenna (Fig. 3) occurs within a very short period of time — on the order of 100 femtoseconds. Before the pulse arrival, the amount of energy scattered by the particle in the forward direction is nearly the same as in the backward direction. However, driven by a short pulse, the antenna switches to almost perfectly unidirectional forward-scattering. Theoretical predictions backed by the experimental data suggest that an antenna of this kind would have a bandwidth of about 250 Gbit/s, whereas conventional silicon-based electronics rely on components with bandwidths limited to only tens of Gbit/s.
Concluding remarks: there’s more to come The experiments performed by the authors of the study have demonstrated ultrafast nanoantenna switching between different light scattering modes, which is caused by the interaction of an intense laser pulse with the silicon of the nanostructure. The researchers have developed an analytical theory describing the behaviour of such non-linear nanoantennas. ‘The research shows that silicon nanoparticles might well become the basis for developing ultrafast optical nanodevices. Our model can be used to design nanostructures containing silicon particles that are more complex, which would enable us to manipulate light in a most unusual way. For example, we hope to eventually control not just the amplitude of an optical signal but also its direction. We expect to be able to “turn” it by a specified angle on an ultrafast timescale,’ says Sergey Makarov, a senior researcher at the Department (Chair) of Nanophotonics and Metamaterials of the ITMO University.
Dmitry Fedyanin from the Moscow Institute of Physics and Technology and Mario Agio from the University of Siegen and LENS have predicted that artificial defects in the crystal lattice of diamond can be turned into ultrabright and extremely efficient electrically-driven quantum emitters. Their work published in New Journal of Physics demonstrates the potential for a number of technological breakthroughs, including the development of quantum computers and secure communication lines, which, in contrast to previously proposed schemes, would be able to operate at room temperature.
The research conducted by Dmitry Fedyanin and Mario Agio is focused on the development of efficient electrically-driven single-photon sources – devices that emit single photons when an electrical current is applied. In other words, using such devices, one can generate a photon “on demand” by simply applying a small voltage across the devices, the probability of an output of zero photons is vanishingly low and generation of two or more photons simultaneously is fundamentally impossible.
Until recently, it was thought that quantum dots (nanoscale semiconductor particles) are the most promising candidates for true single-photon sources. However, they operate only at very low temperatures, which is their main drawback – mass application would not be possible if a device has to be cooled with liquid nitrogen or even colder liquid helium, or using refrigeration units, which are even more expensive and power-hungry. At the same time, it was known that certain point defects in the crystal lattice of diamond, which occur when foreign atoms (such as silicon or nitrogen) enter the diamond accidentally or through targeted implantation, can efficiently emit single photons at room temperature. However, this has only been achieved by optical excitation of these defects using external high-power lasers. This method is ideal for research in scientific laboratories, but it is very inefficient in practical devices. Experiments with electrical excitation, on the other hand, did not yield the best results – in terms of brightness, diamond sources lost out significantly (by several orders of magnitude) to quantum dots. As there were no theories describing the photon emission from colour centres in diamonds under electrical excitation, it was not possible to assess the potential of these single-photon sources to see if they could be used as a basis for the quantum devices of the future.
The new publication gives an affirmative answer – defects in the structure of diamond at the atomic level can be used to design highly efficient single-photon sources that are even more promising than their counterparts based on quantum dots.
Operation at the single‐photon level will not only increase the energy efficiency of the existing data processing and data transmission devices by more than one thousand times, but will also lay the foundations for the development of novel quantum devices. Building quantum computers is still a prospect of the future, but secure communication lines based on quantum cryptography are already starting to be used. However, today they do not use true single-photon sources; instead, they rely on what are known as attenuated lasers. This means that not only is there a high probability of sending zero photons into a channel, which greatly reduces the speed of data transfer, but there is also a high probability of sending two, three, four, or more light quanta simultaneously. One could intercept these “extra” photons and neither the sender nor the recipient would know about it. This makes the communication channel vulnerable to eavesdropping and quantum cryptography loses its main advantage – fundamental security against all types of attacks.
For quantum computing it is also essential to have the ability to manipulate individual photons. The quantum of light can be used to represent a qubit – the fundamental unit of quantum information processing, – which is a superposition of two or more quantum states. For example, a qubit can be encoded in the polarization of a single photon. The advantage of the optical quantum computing paradigm is that one can natively combine quantum computations with quantum communication and design high-performance, large and scalable quantum supercomputers, which is not possible to do using other physical systems, such as superconducting circuits or trapped ions.
Dmitry Fedyanin and Mario Agio are the first to successfully reveal the mechanism of electroluminescence of colour centres in diamond and develop a theoretical framework to quantify it. They found that not all states of colour centres can be excited electrically, despite the fact that they may be “accessible” under optical excitation. This is because under optical pumping defects behave like isolated atoms or molecules (such as hydrogen or helium), with virtually no interaction with the diamond crystal. Electrical excitation, on the other hand, is based on the exchange of electrons between the defect and the diamond crystal. This not only brings limitations, but also opens up new possibilities. For example, according to the researchers, certain defects can emit serially two photons at two different wavelengths from two different charge states in a single act of the electroluminescence process. This feature could lead to the development of a fundamentally new class of quantum devices that had simply been disregarded before because these processes are not possible with optical excitation of colour centers. But the most important result of the study is that the researchers found out why high-intensity single-photon emission from colour centers was not observed under electrical pumping. The reason for this was the technologically complex process of doping of diamond by phosphorus, which cannot provide sufficiently high density of conduction electrons in diamond.
The calculations show that using modern doping technologies it is possible to create a bright single-photon source with an emission rate of more than 100,000 photons per second at room temperature. It is truly remarkable that the emission rate only increases as the device temperature increases achieving more than 100 million photons per second at 200 degrees Celsius. “Our single-photon source is one of few, if not the only optoelectronic device that should be heated in order to improve its performance, and the effect of improvement is as high as three orders of magnitude. Normally, both electronic and optical devices need to be cooled by attaching heat sinks with fans, or by placing them in liquid nitrogen,” says Dmitry Fedyanin from the Laboratory of Nanooptics and Plasmonics at MIPT. According to him, the technological improvement of diamond doping will further increase the brightness 10-100 times.
One hundred million photons is very low compared to household light sources (e.g. a normal light bulb emits more than 10^18 photons per second), but it should be emphasized that the entire flow of photons is created by a tiny (~10^-10 metres in size) defect in the crystal lattice of diamond and, unlike a light bulb, photons follow strictly one after the other. For the quantum computers mentioned above, around ten thousand photons per second would be enough – the possibility of developing a quantum computer is currently limited by entirely different factors. In quantum communication lines, however, the use of electrically-driven diamond single-photon sources will not only guarantee complete security, but will also greatly increase the speed of information transfer compared to the pseudo single-photon sources based on attenuated lasers used today.
Scientists from MIPT and their colleagues have developed a novel compact and powerful ceramic-based laser – it will be used as a minimally traumatic and inexpensive laser scalpel for surgical operations, and also for cutting and engraving composite materials. The results of the study have been published in Optics Letters.
Today, lasers are used everywhere: in consumer electronics devices, in medicine, metallurgy, metrology, meteorology, and many other areas. Laser beams occur due to the effect of stimulated emission in an active medium, which could be a gas, liquid, crystal, or glass. The wavelength of a laser and the efficiency of converting energy into radiation are both dependent upon the parameters of the active medium.
Ivan Obronov, a researcher at MIPT, and his colleagues from the Institute of Applied Physics of the RAS and the company IRE-Polus used a ceramic obtained from compounds of rare-earth elements – lutetium oxide with added thulium ions (Tm3+:Lu2O3). It was the thulium ions that enabled the ceramic to generate laser radiation.
“Ceramics are a promising type of medium for lasers because they are produced by sintering powders into a polycrystalline mass. They are cheaper and easier to manufacture than single crystals, which is extremely important for mass adoption. In addition, it is easy to alter the chemical composition of ceramics, which in turn alters the laser properties,” explains Obronov.
The laser they have developed converts energy into radiation with an efficiency of more than 50%, while other types of solid state lasers have an average efficiency of approximately 20%, and it generates infrared radiation with a wavelength of about 2 microns (1966 and 2064 nanometres). The wavelength is what makes this laser so useful for medical purposes.
Researchers from the Moscow Institute of Physics and Technology (MIPT), the Engelhardt Institute of Molecular Biology (EIMB RAS), the Institute of Bioorganic Chemistry (IBCh) and a number of other Russian research centers have developed a new method of diagnosing colorectal cancer. The results of the study have been published in Cancer Medicine.
The scientists have created a hydrogel-based biochip to help detect bowel cancer i.e. colorectal cancer (CRC). CRC is the third most common type of cancer and it develops with minimal clinical symptoms in the early stages. Despite doctors’ efforts, the 5-year survival rate does not exceed 36%. Treatment is only effective, and patients only have a good chance of recovery, if the cancer is detected early.
Diagnostic methods that are currently in use are not sufficient. Analyses carried out in vitro have low specificity and invasive studies such as colonoscopy are not only traumatic, but they are also not always suitable for an early diagnosis, as they do not give a complete picture of the development and distribution of colorectal cancer.
The method proposed by scientists from EIMB RAS, MIPT, the Russian Scientific Center of Surgery, Sechenov First Moscow State Medical University, the Institute of Bioorganic Chemistry, and Buyanov City Clinical Hospital is based on the simultaneous detection of various substances in patients’ blood. These substances are autoantibodies against tumor-associated glycans, which can be found in serum at the early stages of cancer, immunoglobulins of different classes, and oncomarkers (molecules produced by tumor cells).
Scientists have developed a new type of graphene-based transistor and using modeling they have demonstrated that it has ultralow power consumption compared with other similar transistor devices
Scientists have developed a new type of graphene-based transistor and using modelling they have demonstrated that it has ultralow power consumption compared with other similar transistor devices. The findings have been published in a paper in the journal Scientific Reports. The most important effect of reducing power consumption is that it enables the clock speed of processors to be increased. According to calculations, the increase could be as high as two orders of magnitude.
“The point is not so much about saving electricity – we have plenty of electrical energy. At a lower power, electronic components heat up less, and that means that they are able to operate at a higher clock speed – not one gigahertz, but ten for example, or even one hundred,” says the corresponding author of the study, the head of MIPT’s Laboratory of Optoelectronics and Two-Dimensional Materials, Dmitry Svintsov.
Scientists from MIPT have succeeded in growing ultra-thin (2.5-nanometre) ferroelectric films based on hafnium oxide that could potentially be used to develop non-volatile memory elements called ferroelectric tunnel junctions. The results of the study have been published in the journal ACS Appl. Mater. Interfaces.
Humans are constantly expanding the volume of stored and processed information, which according to statistics is doubling every 1.5 years. To store this information, we need increasing amounts of computer memory, especially non-volatile memory, which stores information even in the event of a power outage. Scientists all over the world are trying to develop faster and more compact storage devices. The ideal would be a “universal” memory device with the speed of RAM, the capacity of a hard drive, and the non-volatility of a flash drive.
There are many known principles that can be used to build memory, but each one has its drawbacks. This is why modern computers and mobile devices have multiple types of memory.
Non-volatile memory based on ferroelectric tunnel junctions is a promising development that has not yet been fully implemented. A ferroelectric is a material that is able to “remember” the direction of an externally applied electric field by the residual polarization charge.
Thin-film ferroelectrics have for a long time been used to make non-volatile memory devices, however it is extremely difficult to miniaturize them in order to achieve high density / storage capacity and, in addition to this, they are made of materials that are “incompatible” with the production processes used in modern microelectronics.
A team of scientists from the Moscow Institute of Physics and Technology (MIPT) have created prototypes of “electronic synapses” based on ultra-thin films of hafnium oxide (HfO2). These prototypes could potentially be used in fundamentally new computing systems.
The paper has been published in the journal Nanoscale Research Letters.
The group of researchers from MIPT have made HfO2-based memristors measuring just 40×40 nm2. The nanostructures they built exhibit properties similar to biological synapses. Using newly developed technology, the memristors were integrated in matrices: in the future this technology may be used to design computers that function similar to biological neural networks.
Memristors (resistors with memory) are devices that are able to change their state (conductivity) depending on the charge passing through them, and they therefore have a memory of their “history”. In this study, the scientists used devices based on thin-film hafnium oxide, a material that is already used in the production of modern processors. This means that this new lab technology could, if required, easily be used in industrial processes.
“In a simpler version, memristors are promising binary non-volatile memory cells, in which information is written by switching the electric resistance – from high to low and back again. What we are trying to demonstrate are much more complex functions of memristors – that they behave similar to biological synapses,” said Yury Matveyev, the corresponding author of the paper, and senior researcher of MIPT’s Laboratory of Functional Materials and Devices for Nanoelectronics, commenting on the study.
Synapses – the key to learning and memory
A synapse is point of connection between neurons, the main function of which is to transmit a signal (a spike – a particular type of signal, see fig. 2) from one neuron to another. Each neuron may have thousands of synapses, i.e. connect with a large number of other neurons. This means that information can be processed in parallel, rather than sequentially (as in modern computers). This is the reason why “living” neural networks are so immensely effective both in terms of speed and energy consumption in solving large range of tasks, such as image / voice recognition, etc.
Over time, synapses may change their “weight”, i.e. their ability to transmit a signal. This property is believed to be the key to understanding the learning and memory functions of the brain.
From the physical point of view, synaptic “memory” and “learning” in the brain can be interpreted as follows: the neural connection possesses a certain “conductivity”, which is determined by the previous “history” of signals that have passed through the connection. If a synapse transmits a signal from one neuron to another, we can say that it has high “conductivity”, and if it does not, we say it has low “conductivity”. However, synapses do not simply function in on/off mode; they can have any intermediate “weight” (intermediate conductivity value). Accordingly, if we want to simulate them using certain devices, these devices will also have to have analogous characteristics.
The memristor as an analogue of the synapse
As in a biological synapse, the value of the electrical conductivity of a memristor is the result of its previous “life” – from the moment it was made.
There is a number of physical effects that can be exploited to design memristors. In this study, the authors used devices based on ultrathin-film hafnium oxide, which exhibit the effect of soft (reversible) electrical breakdown under an applied external electric field. Most often, these devices use only two different states encoding logic zero and one. However, in order to simulate biological synapses, a continuous spectrum of conductivities had to be used in the devices.
“The detailed physical mechanism behind the function of the memristors in question is still debated. However, the qualitative model is as follows: in the metal–ultrathin oxide–metal structure, charged point defects, such as vacancies of oxygen atoms, are formed and move around in the oxide layer when exposed to an electric field. It is these defects that are responsible for the reversible change in the conductivity of the oxide layer,” says the co-author of the paper and researcher of MIPT’s Laboratory of Functional Materials and Devices for Nanoelectronics, Sergey Zakharchenko.
The authors used the newly developed “analogue” memristors to model various learning mechanisms (“plasticity”) of biological synapses. In particular, this involved functions such as long-term potentiation (LTP) or long-term depression (LTD) of a connection between two neurons. It is generally accepted that these functions are the underlying mechanisms of memory in the brain.
The authors also succeeded in demonstrating a more complex mechanism –spike-timing-dependent plasticity, i.e. the dependence of the value of the connection between neurons on the relative time taken for them to be “triggered”. It had previously been shown that this mechanism is responsible for associative learning – the ability of the brain to find connections between different events.
To demonstrate this function in their memristor devices, the authors purposefully used an electric signal which reproduced, as far as possible, the signals in living neurons, and they obtained a dependency very similar to those observed in living synapses (see fig. 3).
Fig.3. The change in conductivity of memristors depending on the temporal separation between “spikes”(rigth) and thr change in potential of the neuron connections in biological neural networks.
Source: MIPT press office
These results allowed the authors to confirm that the elements that they had developed could be considered a prototype of the “electronic synapse”, which could be used as a basis for the hardware implementation of artificial neural networks.
“We have created a baseline matrix of nanoscale memristors demonstrating the properties of biological synapses. Thanks to this research, we are now one step closer to building an artificial neural network. It may only be the very simplest of networks, but it is nevertheless a hardware prototype,” said the head of MIPT’s Laboratory of Functional Materials and Devices for Nanoelectronics, Andrey Zenkevich.