Spin doctors prescribe semiconductor solutions
Cramming more and more transistors onto computer chips is becoming a headache. For years, the electronics industry has succeeded in doubling the number of transistors on a given surface area every 18 months, in line with Moore's Law. But transistor miniaturisation is becoming tougher, mainly due to power dissipation problems as the electrons travel through integrated circuits. Even when they are off, today's tiny transistors still leak current. Dissipationless switching,The solution to such energy loss, believe some researchers, is to move away from traditional charge-based microelectronics. "We want to move information without moving charges," says Dr Willem Van Roy, from the microelectronics research institute IMEC in Belgium. "The goal is dissipationless switching by using the spin of electrons." Flipping a spin's property - from up to down or down to up - avoids the energy loss experienced when moving electrons around. Electron spin is closely related to magnetism, hence the alternative name for spintronics: magnetoelectronics. The first applications of spin-based transport in integrated circuits appeared in the late 1980s. Metallic multilayers gave rise to increased storage densities in computer hard drives, among the best-known devices harnessing the power of magnetism, as well as non-volatile memory (Random Access Memory). Hard drives are metal-based devices. They use ferromagnetic metallic alloys in read heads to apply a magnetic field to reduce resistance and get a spin. But researchers would also like to make semiconductors - the core of modern information technology - magnetic by using spin functionality. "Research institutes like my own are assessing various ways of combining magnetism with semiconductor features," says Van Roy. "Using ferromagnetic materials, we can polarise the electrons' spin, then inject them into the semiconductor." Further steps in the control of spin-polarised carriers include transport, storage, manipulation and detection. ,Injection time,The injection stage can be done in several different ways. One is to polarise light in order to excite the electrons. Others include electrical injection, injection through tunnel barriers - an increasingly popular solution, and introducing magnetism in the semiconductor itself. "Ultimately, we want to move spin information around rather than the spin itself," explains Van Roy. Experiments are continuing with ferromagnetic metals on semiconductors as well as half-metallic ferromagnets. "The latter look more promising," says Van Roy. "We are still studying candidate materials for them and trying to solve conductivity mismatch problems." The race is on to create devices to inject spin-polarised electrons, with a view to manipulating the spin in the semiconductor. Recent successes in the control of spin in semiconductors include long spin lifetimes - one European research project, Feniks, achieved an acceptable lifetime of 300 picoseconds in InSb [indium antimonide], while longer lifetimes have already been demonstrated in GaAs [gallium arsenide] - and coherent transport through interfaces. "By spin dragging across an interface, we can now move spin where we want, even into different materials, for example by exciting spins in GaAs and detecting them in ZnSe [zinc selenide]." Ambient temperatures,Most importantly, spin injection and dragging have been done at room temperature - a significant improvement over the very cold temperatures previously required. Work is also focused on magnetic doping using a variety of different magnetic semiconductor families. This process is more complex than traditional doping, the addition of an impurity to a semiconductor to produce electrical conductivity. "The new form of doping can be electrical or magnetic, or both," he adds. "One of the challenges of (ferro)magnetic semiconductors is to increase the transition temperature," admits Van Roy. "If we can do this, we may be able to build more complex heterostructures that operate at room temperature." The researcher foresees a number of interesting magneto-optical applications for these new semiconductors. They include integrated magneto-optical devices, as well as spintronic lasers and switches. Optical isolators, for example, are used in combination with lasers. Currently very expensive, these isolators have recently been integrated into the laser under the IST project Isolaser (Optical isolator monolithically integrated with DFB-laser). By adding a semiconductor optical amplifier with ferromagnetic contact 200 nm from the active region, it becomes possible to control the direction in which the light is allowed to move in order to create an optical isolator. "Project Isolaser, in which my lab is involved, offers the first worldwide qualitative proof of the principle," he adds with pride. ,Niche markets,Asked about the economic impact of semiconductor spintronics, Van Roy says that this emerging science will likely be limited to niche markets, such as integrated non-volatile memories: "We do not want to replace mainstream silicon technology, we want to add to it." He adds that researchers are now looking for more intelligent switching mechanisms, citing ballistic and precessional switching. "These would allow us to control how much of a spin an electron makes, switching on and off within a fraction of a second." Magnetic quantum dots may also be used to increase magnetism, enabling control over the spin of single electrons. The Belgian scientist says the future of magnetic semiconductors remains challenging. But he is confident that they will soon exist in room-temperature versions. He also feels that tunnel-based spin injectors have been very successful: "They are robust and efficient." As for half-metals, such as manganites, they can induce up to 100 per cent spin polarisation. Yet they are hard to make and equally hard to add to semiconductors. Major challenges in the field remain. They include control and optimisation of materials for magnetic semiconductors; Van Roy coordinates the Feniks project, which is looking at this area. Better spin injection is another goal, covered by another IST-FET project, Spinosa. Future work will aim to combine spin injection, manipulation, and detection. Cryptography on horizon,Looking to the future, he believes spintronics will enable advanced cryptography before this decade is out, followed later by quantum computing. But he does not believe that spintronics will replace mainstream CMOS (complementary transistors) technology, which is "very mature and still improving." Spintronics clearly opens new avenues, such as in optical isolation and chip-to-chip communication. Van Roy concludes: "We are now at the threshold of moving from the proof of principle to the first set of tools, enabling us to make the devices we want." Source: Based on an interview with Dr Van Roy and information from projects Isolaser, Feniks and Spinosa,The IST Results service gives you online news and analysis on the emerging results from Information Society Technologies research. The service reports on prototype products and services ready for commercialisation as well as work in progress and interim results with significant potential for exploitation.,
Countries
Belgium
