Promising directions for the development of spintronics

Spin electronic positioning and motion sensors. GMR sensors, used to detect the magnitude and direction of a magnetic field, have found wide application in the following areas: information storage and readout, programmable gate arrays, avionics, electronic machine control and automotive active safety systems. For example, the global market for automotive sensors, with annual growth of more than 10%, is one of the fastest growing, currently reaching €8.5 billion. In recent years, the main trends in the development of the global automotive industry have been the improvement of control of the internal combustion engine (in order to reduce vehicle emissions), braking and anti-skid systems, safety devices, etc. As these technologies advance rapidly, improvements in spin magnetic sensors are at the forefront: increasing sensitivity, stability, reliability and noise suppression.

Spin diode. The idea of ​​a spin two-terminal diode was first proposed by Mathews. The diode consists of a five-layer magnetic system, in which three ferromagnetic layers are separated by paramagnetic layers. One of the most successful attempts at the practical implementation of a spin diode was carried out in 2004. In the future, it is planned to use spin diodes as elementary cells of MRAM memory.

Coherent quantum spintronics. In the longer term is quantum coherent spintronics. This refers to devices whose dimensions are so small that the quantum coherence of the electron's wave function is maintained across the device, coupling incoming and outgoing electrical signals. Nanotechnology has reached such a level that today it is possible to create devices at a scale of 1 nm. A typical example is a tunnel diode (Patent Application No FR9904227, France).

Quantum computing. Researchers predict widespread use of spintronics developments in the field of quantum computing. It is believed that the next serious stage in the development of spintronics will be devices in which information will be transmitted not through electron spins, but using complex qubit pairs. For example, multi-pin spin devices, which can be based on streams of entangled qubits. In practice, such a device can be implemented on the basis of spin-electronic transistors.

The potential of spintronics is not limited to the already developed and mastered technologies described above. Despite the fact that work in this direction has been going on for more than two decades, there are many unsolved scientific and technical problems. For example, a magnetic field is now used to change the magnetization of a section of a ferromagnet. Since we can create a magnetic field only with the help of electric current (permanent magnets do not count), the problem of localizing this magnetic field in a limited area of ​​space arises. The smaller this area, the higher the density of storing information on a magnetic medium can be obtained (of course, there are still questions about selecting appropriate magnetic materials). In the laboratories of solid state physics (Zurich) and Stanford University, an experiment was carried out that shows the possibility of changing the magnetization of a material using a flow of electrons with a certain spin (such electrons are said to be spin polarized). Using photoemission from a semiconductor cathode caused by polarized light, a beam of spin-polarized electrons was obtained. This beam was passed through a magnetic film several nanometers thick. When electrons fly through the film, the spin of the electrons changes (this phenomenon is called precession ). Since nothing in nature passes without a trace, the spins of the electrons in the magnetic film also change, which means a change in the magnetization of the substance. If the number of electrons passed is comparable to the number of atoms of the substance, then the change in the magnetization of the film will be very noticeable. The effect can be used both for recording information and for reading it (at a lower intensity of the electron beam). Potentially, this technology can provide magnetization reversal speeds (that is, actually reading and writing information) of up to tens of gigahertz, but before that, researchers will have to go through

still a very long way to go.

Another interesting effect is the production of a pure spin electron flow without charge transfer. In the experiment, two counter flows of electrons with oppositely directed spin were formed. This amazing effect was achieved using two pulsed polarized lasers, the frequency of one of which is half that of the other. Thus, spin charge transfer has been achieved without the presence of a potential difference. So far this phenomenon has been observed at distances of the order of several tens of nanometers, but further research in this direction continues.

One of the problems with spintronics relates to the materials used. The fact is that spintronics requires ferromagnets, the magnetic properties of which give rise to various effects involving electron spins. But ferromagnets are metals, and modern electronics are based on semiconductors. It is the properties of semiconductors that make it possible to increase the electric current in transistors - this effect is impossible in metals. Therefore, in order to create an efficient device that uses both the spin and charge of the electron, it is necessary ferromagnet, which is a semiconductor . A new semiconductor has been created at the Pacific Northwest National Laboratory (USA) that does not lose its magnetic properties even at room temperature. This substance is titanium oxide with cobalt admixture and is grown in the form of nanometer films by molecular epitaxy. In a deep vacuum, beams of atoms in the required ratio are directed to the crystalline surface, where they form the necessary crystal structure.

Another similar material is epitaxial film made of alternating layers of gallium compounds: GaSb, GaMn. The magnetic properties of this semiconductor are maintained up to 130°C, which is sufficient for the needs of modern technology.

Another promising direction is use of organic compounds . At the University of California (Riverside), they synthesized a compound that changes its optical, electrical and magnetic properties simultaneously, depending on temperature. At a temperature of about 62°C, the substance transforms from a transparent (in the infrared spectrum) paramagnetic insulator into an opaque diamagnetic conductor. Such unique properties make it attractive not only for spintronics, but also for other promising areas, for example, photonics. True, the operating temperature of the transition is somewhat high for use, but scientists hope to reduce it by varying the composition of the substance.

Researched at Ohio State University plastic - vanadium tetracyanoethanide . Despite its organic nature, it also has magnetic properties that persist up to 130°C. In addition, plastic is much more technologically advanced than other materials, which will allow the creation of cheap plastic memory in the future.

Nanocomposites in the form of metal wires in polycarbonate or aluminum oxide. Polycarbonate film is exposed to heavy, high-energy charged particles in a nuclear reactor. Passing through polycarbonate, charged particles leave tracks with a disrupted (i.e., different, different from the rest of the array) structure. Then these tracks are etched in a concentrated alkali solution and thus uniform through cylindrical pores are formed. The pore density is determined by the length of time the membrane remains in the reactor. Membranes are produced with standard pore densities of 10 6, 10 8, 6. 10 8, 10 9 and 10 10 pores/cm2. The diameter of the pores can be obtained in a wide range from 10 to 300 nm, depending on the time of irradiation of the film, the temperature and concentration of the solution, as well as the etching time. The thickness of the membranes can range from units to hundreds of microns, the diameter is usually 13 mm. Anodized aluminum membranes can also be used for electrodeposition of nanowires. The production of a structure in the form of multilayer and granular nanowires is carried out exclusively by the method of electrolytic deposition into membrane pores (Fig.) from a single electrolyte in both potentiostatic and galvanostatic pulse modes. Before electrodeposition, a layer of gold ~0.01 μm is deposited into the pores to ensure electrical contact on one side of the membrane (Fig.).

Rice. Schematic illustrations of the polycarbonate membrane and individual multilayer nanowire (left). Geometry of applying a gold sublayer to a membrane (right).

Unlike planar electrodeposition, where the entire cathode surface is exposed to the electrolyte, in nanowire electrodeposition only a portion of the membrane surface, called the active or true deposition area, is exposed to the electrolyte. It can be calculated by knowing the number of pores over the entire area of ​​the membrane and the area of ​​one pore:

Rice. Typical current curve of nanowire growth

An increase in current after point B indicates the beginning of the film emerging over the pores. This corresponds to a nominal thickness of about 3.8 µm. After this, the wires begin to merge on the surface, and hemispherical caps appear above the wires. The thickness of the copper layer is approximately 30 nm, and the thickness of the Co-Ni alloy is 40 nm.

Rice. Installation diagram for magnetoresistive measurements (giant magnetoresistance)

To carry out magnetoresistive measurements, the presence of an upper conductive contact is required. This is achieved by depositing some material on top of the membrane. After filling the pores, the deposited substance begins to grow in the form of hemispherical bowls, which then merge with each other (Fig.)

In the case when the field is parallel to the nanowire axis, their hysteresis loops are characteristic of magnetization reversal along the easy magnetization axis. If the external field is perpendicular to the axis of the wire, then in this case much larger fields are required to rotate all the moments in this direction, and the result is a loop characteristic of magnetization along a difficult axis.

Structure, magnetic and magnetoresistive properties of spin-valve nanowires. Among magnetic nanomaterials, multilayer (or multilayer) structures occupy a special place. This is largely due to the effect of giant isotropic magnetoresistance discovered in them. The study of this phenomenon, as well as attempts by developers of various magnetic microelectronics devices to increase the magnitude of the change in electrical resistance per unit magnetic field, led to the emergence of a new, more complex family of multilayer film structures so-called “spin valve” type. They already represent periodic alternation of not two, but three or more layers with different magnetic parameters. In this case, the process of their magnetization reversal is anisotropic. When the applied external magnetic field changes in the direction of the magnetization vector, previously magnetized to saturation of the hard magnetic layer in a range less than its coercive force, the soft magnetic layer will be remagnetized in this direction in a field below its coercive force. And in the opposite direction - in a field greater than its coercive force. This difference in the magnetization reversal fields of low- and high-coercivity layers in a multilayer structure in opposite directions is the essence of the “valve” effect. The state of a multilayer structure, when the magnetic moments of the soft and hard magnetic layers are antiparallel, is unstable. And a small field of opposite direction leads to abrupt magnetization reversal of low-coercivity layers. That is why high sensitivity of the magnetoresistive element can be achieved.

Another option for increasing the magnitude of the magnetoresistive effect is to manufacture a multilayer structure in the form of nanowires. This is achieved exclusively by pulsed electrolytic deposition into the pores of nanomembranes. For nanowires, the geometry of the magnetoresistive effect is easily realized when the electric current is perpendicular to the interfaces between layers in a multilayer structure, which is impossible for conventional multilayer films with flat geometry. In this case, all conduction electrons are forced to cross magnetic layers with a periodically antiparallel direction of their magnetic moments and, therefore, the effect of their scattering will be greater compared to conventional multilayer structures. The main difficulty is that when varying the deposition potential (or cathode current density D K), it is necessary to select such deposition conditions (primarily the composition of the electrolyte and deposition modes) when only by changing them (or D K) a sufficiently large difference in composition will be achieved and the crystal structure of the magnetic layers and, therefore, their coercive force will also differ significantly. These conditions can be met by CoFeP and CoW films, in which the content of phosphorus and tungsten is a function of the current density. And, for example, at low current density (D K ~ 10-20 mA/cm 2) the phosphorus content reaches ~20-25 at.%. In this case, CoFeP 25 films are amorphous magnetically soft, and with a phosphorus content of ~5-10 at% (D K 70 mA/cm 2) they are polycrystalline and, accordingly, magnetically hard. The same is true for the cobalt-tungsten system.

For those who write and read with their heads, they usually use giant magnetoresistance effect (GMR), a quantum mechanical effect that provides the enormous capacity of today's hard drives. Using GMR
provides memory densities well above 100 gigabits per square inch. While modern hard drives have magnetic domains oriented in a plane, the next generation of hard drives will have them oriented perpendicularly. The perpendicular recording technique (Fig., below) will provide more dense packaging of information. But it will require more sensitive write and read heads, which can be made using the even more complex tunnel magnetoresistance (TMR) effect. In this sense
Computer hard drives can be considered a product of nanotechnology.

Rice. The giant magnetoresistance (GMR) effect is widely used in hard drives. Seagate Technology LLC (top). Data storage density can be increased by changing the orientation of the magnetic regions. Perpendicular recording
provides higher storage density. VDI Technologiezentrum GmbH (below)

During the month of February 2013 alone, the information technology media produced a whole bunch of very remarkable news reports about achievements in the field of spintronics. That is, about a new, fundamentally different type of electronic devices that rely not on the electric charge of carrier particles, but on their spin - a quantum property inherent in particles, the development of which promises a genuine revolution in computer technology.

This is what, offhand, looks like just some of the latest news from spintronics.

Two German universities, Mainz and Kaiserlautern, which successfully completed research on the creation of a spintronic memory chip based on the so-called Heusler compounds, received a substantial grant from the state in the amount of 3.8 million euros - to quickly bring the developed technologies to the stage of mass industrial production.

Scientists at the British University of Cambridge managed to combine in their development two of the most advanced areas of research in the field of electronics - 3D chips and spintronics. Thanks to this, they were able to create and demonstrate a prototype of the “world’s first spintronic 3D processor” (quotes are necessary here, since in fact this is far from a full-fledged processor, but the creative success of the researchers cannot be doubted).

Specialists from the University of Göttingen - again in Germany - managed to come up with and synthesize a molecule of artificial organic matter that can play the role of a stable spintronic memory cell. At this level of miniaturization, a spintronic memory device based on inexpensive organic materials could store approximately a petabyte of data (a thousand terabytes, or a million gigabytes) on a chip about an inch in size.

If we add to this same package of news a few more very recent and very impressive ones - about the successes of other research centers in the USA, Japan and other countries, which have already brought spintronic technologies very close to the industrial production phase, then it becomes obvious: big changes are indeed coming.

Well, in order to more clearly understand what kind of information technology is replacing conventional semiconductor electronics, it makes sense to take a closer look at the features of spintronics. Why is this technology so attractive, what are the most difficult problems in its development and how, finally, we manage to bypass and overcome these problems...

⇡ Natural alternative

Among experts, one can often hear the opinion that the obvious delays in the arrival of the long-awaited spintronics in our lives are caused primarily by the extremely stable and successful progress in the field of traditional semiconductor technologies. That is, the time for new technology has not come only because the old one is still around.

Moore's empirical law, as you know, establishes a rule that is in no way provable, but has been working regularly for over half a century. Thanks to the efforts of scientists and engineers, the number of elements of a typical microcircuit—in other words, the performance of chips—regularly continues to double approximately every year and a half.

Why this happens is unknown. But everyone understands that this cannot continue indefinitely. For current chip design is rapidly moving towards its physical limits. Or otherwise, all the known problems of technology - with lithography, materials, cooling - are rapidly approaching a state where overcoming them is not absolutely impossible, but turns out to be too expensive and ineffective.

In short, on the one hand, something different is clearly required. On the other hand, there has long been an understanding of exactly what this other thing will almost certainly look like.

The presence of a special property called spin in particles of matter - usually illustrated by analogies with the axis of rotation of a top or the two poles of a magnetic needle - was established in the early days of quantum mechanics. And since the quantum spin of an electron takes only two possible values, conventionally called “spin-up” and “spin-down,” a very promising information technology potential was noticed in this design quite a long time ago. In fact, in nature there is a ready-made carrier of binary information that encodes either 1 or 0 in the spin direction.

And the most remarkable thing is that we are talking about the same electron that originally figured in the foundations and core of the microelectronic revolution. Almost all semiconductor microcircuits are built on transistors, the main role in the operation of which is played by the movement of electrons. More precisely, the movement of electrical charges inherent in electrons. While electron spin - discovered almost 90 years ago - is ignored in the semiconductor industry, in fact, completely...

However, since everyone agrees that Moore’s law should continue to operate, then the technology under the general name spintronics now acts as the most natural and at the same time more progressive alternative to conventional microelectronics. This name is most often deciphered as SPIN TRansport electrONICS, that is, “electronics based on spin transfer.”

The mass of advantages and benefits of new technology is increasing day by day. Among the most important are speed and efficiency. After all, the spin of an electron can be switched from one state to another in much less time than it takes to move a charge along the circuit, and this is done with much less energy. Plus, during spin transfers, the kinetic energy of the carrier does not change, which means that almost no heat is released.

Taken together, all these features of the technology make it possible to create, based on spin and spin currents (flows of electron spins of the same polarity), significantly new transistors, logic and memory cells that will replace conventional transistors in integrated circuits. And this, in turn, will allow us to continue to adhere to the trend towards miniaturization of electronics.

Along with the development of this technology, it turns out that spintronics also opens the way to the creation of completely new types of devices. Such as light-emitting diodes (LEDs), which produce left- or right-handed circularly polarized light, which is very useful for applications in the field of protection, encoding and optoelectronic communications densification. If you look a little further into the future, it turns out that the emergence of such spintronic devices has already begun that can be used as qubits, that is, the basic design elements in quantum computers.

But in order for the spintronic revolution to occur in the semiconductor industry, it is necessary to find the optimal components of the technology, which researchers have been searching for for the past decade. Typically, there are three main tasks:

1. methods for injecting (i.e., “squirting”) spin states into a circuit;
2. spin manipulation inside the circuit;
3. detection of spin states of electrons after processing.

It is highly desirable to solve all these problems in a semiconductor environment, since these materials will most likely remain the main physical basis for electronics in the foreseeable future.

Manipulating the spin of electrons is considered a relatively simple and unsophisticated matter (since the spin, like a compass needle, reacts very sensitively to switching the magnetic field). But creating reliable injectors and detectors for fragile spins in practical applications for mass production is still a whole set of gigantic problems.

⇡ Test site and take-off area

In order to make the general state of affairs in spintronics clearer, it is necessary to emphasize that manipulation of electron spin is a large and developed business today. But only outside the semiconductor industry. In fact, metal-based spintronic devices are now ubiquitous - in the hard drives of almost every computer on the planet.

At the end of 1988, it was discovered that the flow of spin-polarized electrons in a layered coating structure (two thin layers of a ferromagnet separated by a layer of non-magnetic metal) can be significantly changed by switching to the opposite polarity of an external magnetic field. This effect, called GMR, or giant magnetoresistance, made it possible to create much more sensitive magnetic heads and, accordingly, reduce the size of the magnetic domains encoding binary data on the plates. In other words, the information capacity of hard magnetic disk drives has increased significantly.

Spin manipulation—the transfer of electron spins between two metals—is also at the heart of MRAM, magnetoresistive random access memory. That is, a new type of computer storage devices that store information without power supply.

The physics of MRAM operation is based on an effect somewhat reminiscent of GMR and known as tunneling magnetoresistance (TMR). Here, two layers of ferromagnetic metal are separated by a thin layer of insulating material such as aluminum oxide or magnesium oxide.

If in GMR there is a slow movement of spin-polarized electrons from one ferromagnetic layer to another due to classical diffusion, then in the TMR design there is a purely quantum tunnel transition through the separating layer (a classically forbidden process in which a particle passes through a potential barrier exceeding its kinetic energy).

These types of devices are called magnetic tunnel junctions, or MTJs (magnetic tunnel junctions). The main feature of the effect is that tunneling - and therefore spin transfer through the barrier - can only occur if the particle spin is “correctly” oriented.

Although spin-dependent tunneling was first demonstrated back in 1975, like most quantum phenomena, it only worked at very low temperatures. It was only shown in 1995 that this is possible at room temperature.

At first, however, it was possible to switch the aligned spins of particles in ferromagnetic layers from a parallel to an antiparallel state for only 12-18% of electrons, which is still far from sufficient for practical devices. However, by the end of the 1990s, intensive brainstorming among developers and proper financial investments led to a solution to the problem: the desired ratio was increased to 70%.

Moreover, by the mid-2000s, the latest technologies providing atomic-thick planar interfaces between metal and oxide layers made it possible to achieve TMR values ​​of the order of 400% - thanks to the special effect of coherent tunneling.

The result was that MRAM memory arrays based on tunnel magnetoresistance were put into production and sale before the end of the decade. So in the near future, as the technology becomes cheaper, MRAM will make it possible to make household computers that can turn on and off instantly. Fortunately, the state of the system will be stored in fast and non-volatile memory.

⇡ Injectors and detectors

The details in the previous story about spintronic memory were needed for this reason. The key points of this story - from the specifics of the technology to the general trajectory of its transformation from a demo sample to a mass-produced product - are very similar to the path of spintronics to semiconductor chips.

Perhaps the most important difference is that the TMR effect is based on a large number of electrons that have the desired spin state and maintain it during transitions through interfaces between ferromagnetic metals and insulating metal oxides.

Well, in order for semiconductor spintronic devices to become possible, it is necessary to achieve essentially the same behavior of electrons - but only through interfaces formed between the semiconductor and the material acting as a spin injector or spin detector.

Since silicon and gallium arsenide are the two most widely used semiconductors in the industry, the main challenge for developers is to find spin-polarized materials (substances in which the majority of electron spins are aligned in a given direction) that can be effectively combined with them.

The history of the long and difficult search for materials of this kind is still far from being written. Of course, here we could talk about several different approaches, used with varying degrees of success in many laboratories around the world to solve this most difficult problem. But it’s probably better to skip this topic for now.

Because by the end of the first decade of the 21st century, the result of all research on the introduction of spintronics into the microcircuit industry looked something like this. Despite many local successes, in general no one has been able to find suitable (ferromagnetic semiconductor) materials that work at room temperature and are suitable for use in practical semiconductor spintronics devices...

But, despite such a dismal result, this does not mean at all that progress has stalled and stopped.

⇡ Geisler compounds

An extremely important event for the history of spintronics occurred in the summer of 2010, when the discovery of physicists from the German University of Mainz was published through the journal Nature. This university has long had a reputation as one of the world's main centers for research into so-called Heusler compounds (more on the specific properties of these materials later).

Thanks to the new discovery of scientists who discovered a very special quantum state of matter in Heusler compounds - called a “topological insulator” - new wonderful prospects for the development of spintronic technologies also opened up. And not only in the field of memory devices, but also for semiconductor chips, and for new power batteries, and for many other attractive applications.

What are these Geisler materials?

First of all, it is appropriate to note that in general the German surname Heusler should be read as Heusler. However, according to the centuries-old Russian tradition, foreign names and names are pronounced in our own manner. The poet known in the world as Heine is called Heine in our country. We call Hudson Bay Hudson. For the same reason, the engineer-scientist Friedrich Heusler, who in the early 1900s discovered unusual properties of alloys of ordinary metals, is still commonly called in Russia in the old fashioned way - Heusler.

For many years now, Geisler's materials have been the focus of scientific research for the following reason. Being relatively simple chemical compounds of three basic elements, Heusler compounds can have a wide variety of different physical characteristics.

Thus, the most well-known specific feature of these compounds is that they exhibit characteristics other than what is naturally expected from the elements that make them up. Geisler's first compound, for example, was made from non-magnetic elements - copper, manganese and aluminum. However, their Cu 2 MnAl alloy behaves like a ferromagnet even at room temperature. Likewise, when three metals are combined in some other combination, the result can be a semiconductor.

In a little more detail, Heusler compounds are materials with a very general structure of composition, expressed by the formula X2YZ (where X, Y are transition metals, and Z are elements from groups III-V of the periodic table). Since each of the elements X, Y, Z can be chosen from about 10 different candidates, the total number of possible Heusler materials is roughly estimated at about 1000 (plus, there are so-called “half-Heuslers”, described by the XYZ formula and also having a range of interesting properties) .

Due to the uncomplicated and flexible structure at the base, the desired properties of Heusler compounds can be tuned by adjusting their composition. In other words, researchers have a very wide class of substances that are easy to manufacture and often consist of relatively inexpensive publicly available components, but at the same time make it possible to obtain materials with very exotic ferromagnetic or semiconductor properties.

Thanks to this, in particular, Heusler compounds are now considered a very promising material for the manufacture of solar cells and other thermoelectric generators capable of directly converting heat into electricity. For example, without moving structural parts, generate electricity from the processes of collateral heat generation of machines and devices.

When, in the mid-2000s, first theorists, and soon experimenters, discovered a completely new state of matter in nature called topological insulator, after some time it became clear that here, too, Heusler compounds turn out to be an extremely useful material.

Over the past six to seven years, topological insulators, or TIs for short, have been a very hot research topic in the field of solid-state physics and materials science. The main characteristic property of TI is the fact that, although these materials are actually insulators or semiconductors, their surfaces behave like a conductive metal - but the metal is far from ordinary. As in superconductors, in TI electrons move along surfaces without interacting with their surroundings - because they are in a previously unknown quantum state of “topological protection”.

At the same time, in sharp contrast with the physics of superconductors is another property of TI. In topological insulators, there are not one, but two currents on the surface that do not interact with each other - one for each of the spin directions, which flow in opposite directions.

And it is probably clear that these two stable spin currents, which are not affected by structural defects or contamination in the material, seem to have been created to be used in spintronics (as well as in other applications of quantum information science - such as quantum computers).

So, from these considerations alone, one can imagine how powerful interest and even, one might say, violent excitement manifested themselves in the scientific community when it turned out that Heisler’s materials, which had long been studied and mastered by scientists, possess precisely these remarkable TE properties.

There are several reasons for such excitement.

Firstly, interest in Geisler compounds is caused by their ability to exhibit what experts call a “semi-metallic” character. The term "semi-metallic character" refers to the fact that a given material is capable of simultaneously providing metallic behavior for electrons with one spin component (such as spin-up electrons) and insulating behavior for another spin orientation (such as spin-down). At the same time, materials can demonstrate a level of spin polarization of 100%, which makes them ideal candidates for spin polarizers (injectors) or, conversely, for spin detectors.

Secondly, Geisler compounds are not just a very large class of materials, numbering over 1000 representatives. It contains - according to calculations - over 50 compounds that have distinct features of topological insulators.

This also follows “thirdly”: thanks to such diversity, it now becomes possible not only to select the desired ones, but also to develop completely new physical effects. It is already quite clear that since these materials are composed of three elements, they can certainly offer a wide range of other interesting properties in addition to the quantum state of topological surface protection.

In particular, it now becomes possible to combine several unusual quantum states in one material at once, when, for example, superconductivity and a topological surface interact with each other. And this opens the way to completely new, experimentally not yet discovered characteristics, some of which have already been theoretically predicted...

Fourthly, and finally, the development of new Heusler compounds is by no means the only approach in this area to generating the desired properties of a material. Another promising alternative is the modification of already well-known materials, since they too can be structured to match the desired characteristics. Moreover, such “remodeling” can also ultimately generate materials that can be considered new.

One of the typical procedures for modifying well-developed materials is ion implantation. In this operation, a sample of a standard material is treated with a beam of ions, which produce changes in the crystal lattice and remain embedded in the structure of the material as additives. After that, the new properties of the material are the result of two factors at once: changes in the structure caused by the “bombardment” and the presence of new atoms in the structure.

Summing up all these important discoveries in relation specifically to spintronics, we can already say quite confidently that Heisler compounds are destined to play a key role here. Because it is clear that these materials allow a completely new way to overcome the known obstacles that prevent the combination of conventional ferromagnets with standard industrial technologies in the semiconductor industry.

⇡ Spintronics in 3D

Heusler materials, no doubt, are an extremely promising direction for further progress. But in order not to create the false impression that this is almost the only route for the development of spintronics today, it would be useful to review other interesting developments. Like, say, spintronics based on organic materials. Or spintronic track memory (magnetic racetrack memory, MRM). Or, finally, spintronic power sources based on magnetic tunnel junctions.

However, the length of the article is not flexible, so here, as a conclusion to the review, we will limit ourselves to only a brief story about another remarkable and completely new development. It was made by scientists at the University of Cambridge and combines two of the most promising areas in modern electronics - spintronics and 3D chips.

The idea of ​​multilayer, or stacked, as they say, designs of 3D chips has been underway for quite some time, at least since the 1990s. The essence of the idea is quite simple. If, on the same silicon base as now, we learn to make not flat, but truly three-dimensional—with many connections between layers—integrated circuits of about 100 layers, then Moore’s law will most likely continue to work properly. At least another 15 years.

But one of the biggest challenges still facing 3D chip designers is that relying on traditional electronics can never come up with a really good way to transfer information between layers. If you rely on conventional circuit transistors in this matter, then because of this, power consumption increases noticeably, and heat removal in a stacked design, on the contrary, becomes much more complicated - since most of the elements are now hidden in the internal layers of the chip.

In other words, the traditional approach to 3D chip design is not only clumsy and expensive, but also fails to keep heat dissipation within a reasonable range. And this all means that in the three-dimensional design of microcircuits it is highly desirable to rely on something else to transfer information between layers.

Scientists at the Cavendish Laboratory in Cambridge decided to use spintronics for this. That is, in a stacked multilayer design typical of 3D chips, they came up with and implemented an ingenious mechanism of vertical interlayer connections that works on the basis of the quantum spin of particles.

They called their development a “spintronic shift register,” and this design works like a kind of quantum ratchet mechanism - where bits of data and commands encoded in spins are unidirectionally pushed from one layer to another with minimal energy consumption and, accordingly, virtually no heat generation.

This “vertical register” is implemented in the form of a rather clever multilayer sandwich structure, where two different types of metal layers just a few atoms thick are alternately stacked on top of each other. The properties of the sandwich layers are selected so that the location of the information bit is shifted upward by “one register cell” for every two flips in the polarity of the magnetic field.

In other words, a certain “spin-up” domain in the magnetic layer (or cell) 12, say, after switching the magnetic field twice, appears in the cell (magnetic layer) 13. This mechanism of domain jumps across the layers-floors of the chip is, in fact, There is a basic mode of operation of the shift register in this design.

It is clear that the path from laboratory demonstration of the device to mass production of spintronic 3D processors based on it is most likely very long. But there is no doubt that the demonstrated technology is truly innovative, relies on completely standard production procedures and has no fundamental obstacles to its further development (at the moment).

For a literally newborn technology, this, one can admit, is quite a lot.

Hard drives and spintronics

• Computer hardware

Introduction

What is spintronics?

Everywhere I say here, spin and spin, but what is this?

Spin (from the English spin - rotation, spinning) is the intrinsic angular momentum of an electron not related to its movement in space. Simplifying a little, spin can be thought of as the rotation of an electron around its axis.

Let's remember a little mathematics and physics.
In classical physics, a particle’s mechanical angular momentum (or, as they also say, at the moment of momentum) is equal to:

r– radius vector of the particle;
p is the momentum vector of the particle.

At p = 0, angular momentum of a classical particle M = 0. For an electron, at p = 0, M ≠ 0.
The spin of an electron can take two values:


Rice. 1. Electron spins

In general, spin is measured in units of h (Planck's constant), and the spin is said to be equal to . The electron's own magnetic moment is associated with spin.

I think that a bunch of mathematical symbols above are enough to torment a little readers. And if so, then we will no longer use formulas.

Unlike classical charges, which create a magnetic moment only in the presence of their current (as, for example, in a solenoid), an electron has a magnetic moment at zero momentum. Not only electrons have magnetic spin, but also some other elementary particles, as well as the nuclei of some atoms.

Spintronic effects use the properties of ferrimagnetic materials. These are materials that contain atoms that have a magnetic moment (for example, Fe - iron, Co - cobalt, Ni - nickel), and at a temperature below a certain critical temperature (Curie temperature), the magnetic moments of the atoms are ordered relative to each other. When the spins are parallel, the materials are called ferromagnets, and when the spins are antiparallel, they are called antiferromagnets.

In 1989, structures consisting of ferromagnetic and non-magnetic layers were investigated. Their conductivity was studied. Let's look at the picture:

Fig.2. Three-layer ferromagnetic structure

As can be seen from the figure, both structures consist of three layers: ferromagnetic at the edges of the structure and a non-magnetic layer in the middle. A real example of such structures would be Fe-Cr-Fe (iron-chromium-iron) or Co-Cu-Co (cobalt-copper-cobalt). Moreover, the width of the non-magnetic layer is about 1 nm, or more precisely, the width of the layer should be less than the mean free path of the electron, so that there is no scattering and loss of spin during its movement. Conductivity in such a structure occurs only if the magnetizations of the outer layers are unidirectional, as can be seen in the right figure. Otherwise, we get a “metal insulator”.

And how does this apply to HDDs?

Fig.3. HDD

Only the recording/reading head is of interest within the scope of this article. I specially “gilded” it with yellow paint (as in that funny thing with Petka and Vasily Ivanovich). In general, this is not one device in the head, but two: a recording part and a reading part. Let's take a closer look at the reading part:

Fig.4. Reading head

As you can see, the head consists of four layers: iron, copper, cobalt, and an AFM antiferromagnet. AFM words, or as it is also called, the exchange layer, is designed to fix the magnetic field of the second layer. The second layer is called the fixing layer and in our case it is made of cobalt. In it, the magnetic field is always directed in one direction. The third layer is conductive, usually made of copper, and serves to separate the ferromagnetic layers. The last layer, the sensitive one, is also made of a ferromagnetic material. Unlike the fixing one, the direction of its magnetic field depends on the external field - the field of the cell. A hard disk cell contains one bit of information. Depending on the cell field orientation, the field orientation in the sensitive layer changes. If the orientations of the fields in the sensitive and recording layers coincide, then the cell, according to the principles discussed above, increases its conductivity, i.e. begins to conduct current. If the orientations of the fields are opposite, then we get a “metal insulator”. This effect of changing conductivity (or resistance, because these are just reciprocal quantities) is called GMR - Giant Magnetoresistive - the effect of giant magnetoresistance. The GMR effect was first studied in IBM laboratories in the late 80s, but it took almost 10 years for its industrial implementation.

The fact that such complex technologies surround us everywhere is very dizzying. To be continued.

Let us now consider what happens at the contact of a ferromagnet with a semiconductor (Fig. 1.17). Since the concentration of charge carriers in a semiconductor is much lower than in a ferromagnetic metal, many more electrons diffuse from the latter into the semiconductor. Dynamic equilibrium is established only when a significant potential barrier is formed at the contact - the “Schottky barrier” (Fig. 1.17,a). Because of this, in the region of the semiconductor adjacent to the contact, there is a significant bending of the bands (valence, band gap, and conduction bands).

Rice. 1.17.

When a small voltage is applied to the contact U(“+” to the semiconductor), little changes. Electric current does not flow through the Schottky barrier until the voltage reaches a value close to the height of the barrier. Then it becomes possible for electrons to tunnel through a narrow barrier (Fig. 1.17b).

Polarized electrons from the ferromagnet enter the semiconductor with energy much higher than thermal energy. Such “hot” electrons are scattered very intensely and quickly lose the orientation of their spins. Therefore injection spin-polarized electric current from a ferromagnetic metal into a semiconductor turns out to be very inefficient.

The “ferromagnetic metal – tunnel junction – semiconductor” structure turned out to be more effective in this regard (Fig. 1.17c). The bending of bands in a semiconductor separated from the metal by a dielectric is insignificant. If the thickness of the dielectric is very small (~1 nm), then tunneling begins even at low voltages. The injected spin-oriented electrons enter the semiconductor not as “hot” as in the case of the Schottky barrier. And therefore their spin relaxation time is much longer. This is why, for example, in a spin transistor with a semiconductor base (Fig. 1.6), ultrafine tunnel junctions (in Fig. 1.6, made of silicon nitride) are used between the semiconductor and ferromagnets.

Using an ultrathin tunnel junction, in 2007, using the example of a spin transistor, the structure of which is shown in Fig. 1.18, it was found that Spin-polarized electrons injected into high-purity silicon can have a fairly long spin-relaxation time and diffuse over significant (on the scale of the nano- and even microworld) distances - up to 350 μm

On a wafer of high-purity silicon ( Si(pl.)) with a thickness of 350 microns, a metallization layer was applied on top ( Al/Cu) 10 nm thick, ultra-thin tunnel layer Al 2 O 3, ferromagnetic layer (CoFe) 10 nm thick and aluminum metallization (Al). This structure served as an emitter of spin-polarized electrons. From below onto the silicon wafer ( Si(pl.)) layers of ferromagnetic were deposited (NiFe) and copper (Cu) both are 4 nm thick. A layer of silicon was grown on the latter n-type (n-Si) and ohmic contact from indium (In).

When voltage was applied to the emitter U Uh, from a ferromagnet (CoFe) into silicon through an ultrathin tunnel barrier ( Al 2 O 3 and a thin layer of metallization (Al/Cu) conduction electrons with spins oriented in the direction of the magnetization of the ferromagnet were injected. Under voltage U K1 applied to the collector layer of the ferromagnet (NiFe), these electrons drift through the silicon wafer. Their spin relaxation time and diffusion length turned out to be sufficient for a noticeable part of them to pass to the collector. The direction of spin orientation could be determined by changing the direction of magnetization of the “free” ferromagnet. In this case, the collector current decreased sharply. Silicon layer n-type (n-Si) used for additional amplification and more accurate measurement of signals.

Ferromagnetic semiconductors

The tunnel junction, while improving the conditions for injection of spin-polarized current into the semiconductor, still creates increased electrical resistance and requires increased operating voltages. Therefore, scientists paid special attention to a possible alternative - the use of semiconductor ferromagnets rather than metallic ones as a source of spin-polarized current - the so-called. ferromagnetic semiconductors(FP). Back in the 70s of the twentieth century. such PTs as europium chalcogenides and spinels such as CdCr 2 Se 4 [Nagaev E.L. Physics of magnetic semiconductors. – M.: Science. – 1979. – 431 p.]. However, they revealed ferromagnetic properties only at low temperatures.

In the last two decades of the twentieth century. were intensively studied by the so-called "dilute magnetic semiconductors"(RMP, English diluted magnetic semiconductors, DMS). These are classical semiconductors of the type A 2 B 6 and A 3 B 5, heavily, to the maximum possible solubility, doped with atoms of transition ("magnetic") metals, most often manganese ( Mn– since it has the highest solubility). Exchange interaction of electrons from partially filled d- And f- shells of magnetic ions with band charge carriers of the main semiconductor significantly changes the properties of the latter and leads to the appearance of not only ferromagnetism, but also many new phenomena that may be promising for practical applications. However, for most of these RMPs the Curie temperature turned out to be below room temperature (for example, Ga 0,95 Mn 0,05 Sb TK = 110-250 K – depending on manufacturing technology; at Ga 0,95 Mn 0,05 Sb TK = 80 K). And only for wide-gap semiconductors the Curie temperature turned out to be higher than room temperature (for Ga 1-x Mn x N, eg TK = 400 K). U GaN, doped with gadolinium (the magnetic moment of its atom is equal to 8 Bohr magnetons), thin films become ferromagnetic even when there is one gadolinium atom for almost a million gallium and nitrogen ions. Later it turned out that using additional alloying elements ( Zn, C d, etc.), it is possible to significantly increase the Curie temperature of narrow-gap semiconductors (for example, based on InSb-Mn: Zn, Cd it is possible to obtain a continuous series of RMPs with TK = 320-400 K).

In the last decade, a much wider range of magnetic semiconductors has been synthesized and studied. Ferromagnetic properties at temperatures above room temperature have been detected even in such classical semiconductors as silicon and germanium doped with manganese or other “magnetic” atoms. Here, much depends on the doping technology and on the use of additional alloying elements.

There are no significant barriers at the contact of a ferromagnetic semiconductor with a conventional semiconductor of the same conductivity type(Fig. 1.19, a, b). If the PT and a conventional semiconductor have different types of conductivity, then r-p- a transition, the passage of electric current through which is possible only in one direction (Fig. 1.19, c, d). In Fig. 1.19, in addition to the valence bands (E B1 and E B2) and conduction bands (E P1 and E P2), the bands are also conventionally shown d- And f- electrons (E fd), which are usually also present in ferromagnetic semiconductors. Depending on their position relative to the Fermi level (E Ф), they can be partially or completely filled. Even if they are partially filled, electrical conductivity through such zones is limited, since f- and d-electrons have low mobility (large effective mass).

Injection of spin-polarized current into a semiconductor from ferromagnetic semiconductors turned out to be much more effective than from ferromagnetic metals, and the degree of its spin-polarization can be much higher - up to 100%..

In the last decade, ferromagnetic semiconductor nanocomposite materials have also been actively synthesized and studied, which include magnetic structures with reduced dimensions - nanoparticles, ferromagnetic nanowires, ultrathin ferromagnetic films, which are quantum planes. The Curie temperatures for such nanocomposite semiconductors can differ significantly from the Curie temperature of the corresponding “pure” semiconductor. In addition, it becomes possible to significantly change the properties of the system using an external magnetic field

Spintronic LEDs

Using these achievements, it was possible to create, for example, prototypes spintronic LEDs and spin batteries.

Spintronic LEDs based on the -transition in differ in that their radiation is circularly polarized. This is due to the fact that, unlike conventional LEDs, spin-polarized conduction electrons or spin-polarized “holes” are injected into the heterojunction region where recombination occurs. IN AlGaAs(V GaAs and in other semiconductors of this group) optical transitions are allowed during the recombination of electrons having spin +1/2 only with holes with spin –1/2, or vice versa – electrons having spin –1/2 only with holes with spin +1 /2. Therefore, the photons that are emitted in this case have a spin of ±1, i.e. are right- or left-polarized. This is a purely quantum effect. The dynamics of rotation of the electric vector in such a circularly polarized light wave is shown in Fig. 1.20.

When absorbing circularly polarized light, the same selection rules apply. As a result of this, atoms that absorb a circularly polarized photon go into states with a magnetic quantum number that differs by ±1 from the initial state. In a number of new technologies, which we do not talk about here, this property of circularly polarized light is used for “optical magnetization” of ensembles of atoms or for their “optical pumping” - creating an inverse population of excited states of atoms.. On a gallium arsenide substrate p + (p + -GaAs) successively applied layers GaAs:Be(20 nm), ferromagnetic semiconductor nanoparticles MnAs about 3 nm in diameter, distributed in a 10 nm thick gallium arsenide matrix, aluminum arsenide tunnel barrier ( AlAs), a thin film of gallium arsenide ( GaAs, 1 nm) and ferromagnetic layer MnAs 20 nm thick. Gold contacts are formed on top to the substrate and to the layer MnAs.

If nanoparticles MnAs using an external magnetic field, remagnetize in the direction opposite to the direction of magnetization of the magnetically hard layer MnAs(a ferromagnet with a fixed magnetization), then due to the injection of spin-polarized electrons from it through a tunnel junction, an electric voltage arises at the external terminals. If you close the external electrical circuit, then to ferromagnetic nanoparticles MnAs electrons will “flow”, the spins of which are oriented in the direction of the magnetization of the “fixed” ferromagnet. These electrons, accumulating, lead to a gradual reorientation of ferromagnetic nanoparticles. If the external circuit is opened, the current stops, and along with it, the magnetization reversal of ferromagnetic nanoparticles stops.

Can be charged contactlessly. Such batteries can become an effective source of power supply for spintronic circuits and for microdevices implanted into the human or animal body.

Scientists from IBM Research and the leading European education and research center ETH Zurich have obtained images of the formation of a stable spin helix in a semiconductor for the first time in history.

­
­
­
­

“Usually, such electron spins quickly change and lose their orientation. But for the first time we managed to find a way to equalize their properties in a regular cycle of changing spins."
­
­
­
­
­
­

A little about spintronics

What is the mission of spintronics?
In the next ten to fifteen years, silicon processors will reach their limits. Therefore, scientists are already looking for new physical principles on which high-speed devices with low power consumption and heat dissipation will be built.
In spintronic devices, spin reversal requires virtually no energy expenditure, and between operations the device is disconnected from the power source. If you change the direction of the spin, the kinetic energy of the electron will not change. This means that almost no heat is generated.
Experts identify three main directions in the development of spintronics: quantum computer, spin field-effect transistor and spin memory.
According to scientists from IBM, electrons change spins very quickly - it takes about 100 picoseconds to switch (1 picosecond is one trillionth of a second). And that's the point main problem – 100 picoseconds is not enough for the microcircuits to register a change in state in the system.

No matter what

Researchers at IBM have developed a method for synchronizing electrons, increasing the spin time by 30 times - up to 1 nanosecond (equal to a microprocessor cycle with a frequency of 1 Gigahertz).
­

­
­
­
­
­

Scientists' attention was drawn to a fact not previously described by physicists - when electrons rotate in semiconductors, their spins move tens of micrometers, while rotating synchronously, like waltzing pairs.
­
­
­

“If at the beginning of the circle in a waltz the faces of all the women are turned in one direction, then after some time the rotating couples will find themselves looking in different directions.
Now we have the opportunity to record the speed of rotation of the dancers and tie it to the direction of their movement. The result is an ideal choreography - the faces of all the dancing women in a certain area of ​​the site are directed in one direction.”

At IBM Research labs, scientists used ultrashort laser pulses to observe the movements of thousands of electron spins that were launched into rotation simultaneously within an ultra-small region.
IBM researchers used time-resolved scanning microscope techniques to capture images of the synchronous “waltz” of electron spins. Synchronizing the rotation of electron spins made it possible to observe their movement over distances of more than 10 microns (one hundredth of a millimeter), which increased the possibility of using the spin to process logical operations - quickly and economically in terms of energy consumption.
­

The reason for the synchronous motion of spins is the so-called spin-orbit interaction, a physical mechanism that connects spin with the movement of an electron. The experimental semiconductor sample was made based on gallium arsenide (GaAs) by scientists from ETH Zurich. Gallium arsenide, a Group III/V semiconductor, is widely used in the production of devices such as integrated circuits, infrared LEDs, and high-efficiency solar cells.

Bringing spin electronics from the laboratory to market remains an extremely challenging task. Today's research is carried out at very low temperatures, at which electron spins minimally interact with the environment. Specifically, the research work described here was conducted by IBM scientists at a temperature of 40 degrees Kelvin (-233 Celsius or -387 Fahrenheit).
But, in any case, the new discovery gives control over the movement of magnetic “charges” in semiconductor devices and opens up new opportunities and prospects for creating small-sized and energy-saving electronics.

Tags: Add tags