The spin doctors Researchers around the world are looking to develop advanced computers based on electrons’ spin, rather than just their charge. Matthew Chalmers examines how close the promised devices are to becoming reality DAVID AWSCHALOM DAVID

46 | Chemistry World | May 2009 www.chemistryworld.org It’s a promise perpetuated in Processing problems — for example due to weaker ‘spin- In short popular magazines, academic Spintronic data storage devices orbit’ coupling (the interaction papers and grant proposals: hyper- based on magnetic metals are already  Computing devices between an electron’s spin and its functional magnetoelectronic big business. They lie in the hard- based on electron spin motion through the electromagnetic computing devices that harness disk units of almost all computers, promise to be far faster field of the nucleus, which shifts the spin of electrons. By allowing and originate from the Nobel- and more powerful that atomic energy levels). Moreover, bits of information to be stored prize winning discovery of giant current devices, which in such spin-orbit and processed natively in the same magnetoresistance (GMR) made process information effects can be tuned via band-gap material, rather than in the separate independently in 1988 by physicists using electron charge engineering, altering conduction memory and processor components Albert Fert and Peter Grünberg alone and valence bands and the chemical of today’s computers, ‘spintronics’ (see box). But the full promise of  Today’s computer hard potential, which allows the spins could revolutionise information spintronics – storing and processing drives already exploit to be manipulated. In 1999 David technology. information together – cannot spintronic effects Awschalom’s group at the University Since the invention of the be realised with metals. Instead,  New of California in Santa Barbara, US, transistor in 1947, microelectronics researchers need to perfect the art of materials are needed maintained record spin coherence has transformed our way of life spin manipulation in more functional before spintronic data over a distance of 100 micrometres in by shunting electronic charge materials such as semiconductors. processing becomes a a chunk of the semiconductor gallium around transistors etched in ever A major goal of spintronics is to practical reality arsenide, for example, and last year smaller wafers of semiconductor, develop reliable transistors – the Ian Appelbaum, now at the University namely silicon. But were such workhorses of logic – that switch of Maryland, US, and co-workers components also able to process the or amplify a signal depending on its managed to transport spin-polarised intrinsic angular momentum – spin spin. Designs for spin transistors electrons in silicon over a distance of – of electrons, they would permit have been around since 1990, but a several millimetres. devices with far greater speed and potential commercial device seems Getting a useful spin current into functionality and which emit much far away. Despite huge progress in a semiconductor in the first place is less heat. Spintronics may be the only semiconductor spintronics in the a different matter (see box p48). One way to continue the miniaturisation past 15 years, researchers are only well trodden approach is to inject spin trend of semiconductor electronics, now beginning to find efficient ways by shining circularly polarised light which in the past four decades to inject, transport and detect spin- on a semiconductor and transferring has seen the density of transistors polarised electrons in silicon. the angular momentum of the on silicon chips double every two Semiconductors, especially silicon, photons to electrons (and detecting it years (the latest processors pack are better at preserving the spin of via the reverse optical process). This around one billion in a few square electrons than metals because they works well in gallium arsenide, but centimetres). And to take that step have fewer scattering mechanisms not in silicon due to the material’s into semiconductor spintronics, new materials are key. Predicted in the late 1920s by Paul Storage solutions Dirac, spin is a quantum mechanical The Nobel prize winning neighbouring ferromagnetic property of particles that, in discovery of giant layers are parallel (controlled conjunction with the Pauli exclusion magnetoresistance (GMR) by passing an principle, is partly responsible for in 1988 kick-started the through an input line) electrons the structure of the periodic table, spintronics revolution, becoming are able to tunnel between them, dictating how many electrons occupy a commercial success for data generating a low resistance each shell. An electron has two storage within an astonishing 10 which represents a ‘1’; the possible spin states – plus or minus years. If you allow spin-polarised antiparallel state, in which the a half in units of Planck’s constant electrons (those in the same spin exclusion principle reduces the divided by 2pi, but often simply called state) to diffuse through a thin chances of electron tunnelling, ‘up’ and ‘down’ – which causes it to film of alternating ferromagnetic is taken to represent a ‘0’. Room line up like tiny bar magnet with or and nonmagnetic metal layers, temperature TMR was first against the direction of an external their motion will be restricted demonstrated in 1995, and in . because the magnetisation of 2006 Freescale Semiconductor Because quantum mechanics adjacent ferromagnetic layers announced a commercial 1MB permits an electron to be in a naturally lines up in opposite MRAM chip, while Motorola has superposition of two spin states, directions. Switch those layers demonstrated a 4MB MRAM

of which there are an infinite to being parallel using an chip. IBM combination, spintronics offers a external magnetic field, however, Stuart Parkin also studies MRAM A third type of metallic route to a quantum computer with the and they line up to reduce the spintronic device was ability to solve problems intractable resistance – by at least 70 per tunnel magnetoresistance demonstrated in 2008 by Stuart with conventional microelectronics. cent in today’s devices. This (TMR) promises non-volatile Parkin and colleagues at IBM Even a simple device that reads and ‘spin-valve’ behaviour has memory that stores information Almaden. Called ‘racetrack’ writes information in terms of the allowed the magnetic read permanently without requiring memory, it uses spin currents to angle of a single particle’s spin could heads of hard disks to be over a electrical power, allowing directly manipulate the magnetic lead to an enormous increase in thousand times more sensitive to computers to switch on instantly. state of nanoscale domain walls storage density and computing magnetic changes, allowing vast Magnetic random access within magnetic nanowires via a power. ‘I’d say this is the major quantities of information to be memory (MRAM) exploits phenomenon called spin-torque justification for working in stored in the magnetic regions on quantum tunnelling across switching, promising ultra dense semiconductor spintronics,’ says the disk surface. ferromagnetic-insulating layers. solid-state memory with similar Kees de Groot of the University of A similar effect called If the magnetic moments of costs to magnetic disk storage. Southampton, UK. www.chemistryworld.org Chemistry World | May 2009 | 47 Spintronics

Scanning tunnelling efficiency of 90 per cent, although at micrograph showing low temperatures and with the aid of ferromagnetic an external magnetic field. interactions (red and Dilute magnetic semiconductors yellow) in a gallium (DMS), made ferromagnetic by arsenide semiconductor doping the semiconductor with doped with manganese transition metal atoms, produce an army of spin-aligned electrons ready for injection across an interface into a pure semiconductor. The challenge is to find materials that can achieve this at room temperature without too much doping (which makes band DRS A. YAZDANI & D.J. HORNBAKER / SCIENCE PHOTO LIBRARY PHOTO SCIENCE / HORNBAKER D.J. & YAZDANI A. DRS engineering difficult) or the help of an external magnetic field. Efficient injection also requires atomically flat interfaces between the DMS and semiconductor in order to reduce chemical mixing between the materials, which disrupts the spin- states of passing electrons.

Effective injection Despite promising theoretical predictions for high ‘Curie’ temperatures (above which materials lose their ferromagnetism) in DMS materials and a decade spent trying to realise them, the current ‘Semiconductor record – nearly 200K in GaMnAs by researchers at the University of spintronics Nottingham, UK, and the Czech requires a Academy of Sciences – still isn’t sufficient for device applications. breakthrough in Many researchers are working on basic materials’ alternative approaches. ‘Currently, metallic tunnel junctions are the weaker spin-orbit coupling and its at Tohoku University in Japan, only promising route towards room indirect band structure. In any case, injected spin from the ferromagnetic temperature spin injection,’ says nobody wants cumbersome optical semiconductor GaMnAs into GaAs Georg Schmidt at the University of elements around when etching chips without using magnetic fields, while Würzburg in Germany. at the nanoscale. Laurens Molenkamp and colleagues Metallic tunnel junctions (MTJs), One alternative is to inject spin at Würzburg University, Germany, which underpin TMR in metallic polarised electrons from magnets. employed the dilute magnetic spintronics (see box p47), allow In 1999, Awschalom’s group, in semiconductor BeMnZnSe to inject electrons to tunnel with their spins collaboration with researchers spin into GaAs with a reported intact from a magnetic metal into a nonmagnetic semiconductor. In 2001 Klaus Ploog and co-workers at Avoiding injections the Paul Drude Institute in Berlin, In 2004, David Awschalom’s of a conducting channel – Germany, demonstrated spin group unearthed a effectively separating the spin of injection from a layer of iron into phenomenon called the spin electrons from their charge and gallium arsenide at room temperature which allows spin making devices more compatible via an MTJ. Unfortunately, DAVID AWSCHALOM DAVID polarisation to be achieved with existing semiconductor engineering sufficiently thin directly in semiconductors manufacturing techniques. and abrupt interfaces between by electrical means, without Awschalom’s team observed ferromagnetic metals and silicon is the use of magnetic fields and the effect in GaAs at 20K, but much more difficult. circumventing the problems it has since been detected at In May 2007, Appelbaum’s group with spin injection. ‘The spin Hall room temperature in thin surface made a breakthrough in silicon effect could further revolutionise Spins separated layers of ZnSe. A hot area now, spintronics. They injected spin- spintronics by allowing say some researchers, is the polarised electrons through a thin efficient detection of spins,’ flowing through a material and quantum spin Hall effect recently film of Co84Fe16 into silicon and says Ramamoorthy Ramesh of a transverse external magnetic observed in the semiconductor transported them for a distance the University of California at field, the spin Hall effect HgTe by scientists at Würzberg of about 10µm at a temperature of Berkeley, US. harnesses the spin-orbit coupling University, Germany, which is 85K. To overcome the resistance Like the classical Hall effect, of electrons to cause ‘spin up’ an intrinsic property of some mismatch between the two materials, in which a voltage develops at and ‘spin down’ electrons to materials and may be more the team used a voltage drop to right angles to both a current accumulate at opposite edges robust to material defects. make the electrons energetic or ‘hot’ – effectively firing them across the 48 | Chemistry World | May 2009 www.chemistryworld.org Could MRAM be the microscopy to investigate spin IBM memory of the future? transport and GMR effects in simple alkane chains such as pentane. ‘In terms of components, people don’t yet have a totally clear idea of what spintronics will do, so there are a lot of different approaches.’ Two decades since the discovery of GMR and virtually all the world’s data is stored and accessed depending on which direction the intrinsic angular momentum of electrons points. Yet the hurdles facing researchers building the spintronics revolution are numerous. ‘In semiconductor spintronics a breakthrough in basic materials is required,’ says Shinji Yuasa at AIST in Japan. Harnessing the potential of organic spintronics in particular requires intensive numerical simulations to work out how different molecules bond to the metal contacts and how their shape and vibrational modes affect spin transport. Despite being tantalisingly barrier like bullets. close to demonstrating transistor- Although the efficiency like behaviour in the lab, most of Appelbaum’s silicon spin researchers see a commercial spin injection was only 2 per cent (his Researchers at the transistor as a long way off. ‘This team subsequently has achieved University of California in Riverside, is research,’ says Appelbaum, who a polarisation of 37 per cent), the US, have used multiwalled nanotube declines to put a date on when researchers used a magnetic field to channels to demonstrate GMR such a device may exist. Kees de make the electron spins precess as effects of 6 per cent at temperatures Groot reckons a quantum computer they travel. They also detected the up to 14K, and many groups are probably is still 20 years away. spin-polarisation electrically using trying to fabricate sheets of graphene It’s time, thinks Awschalom, for a a Ni80Fe20 film attached to the other to further understand spin transport new breed of ‘quantum engineer’ to side of the silicon which, like the and manipulation in organic take a closer look at the spintronic CoFe injection layer, acted as a spin materials. In 2006 a team at the phenomena already unearthed in filter. National Institute of Standards and university labs. ‘You can’t simply Such control is essential for a Technology in the US sandwiched a replace “charge” for “spin” in today’s spintronic transistor and thus for the layer of octanethiol between cobalt technology and think things will low-power, reprogrammable logic and nickel contacts which acted as a be better,’ he points out. ‘The real that may eventually overthrow the molecular spin valve, allowing only impact of this field is to make a fixed-function logical units possible electrons in one spin state to pass discontinuous change in technology by pushing just electric charge through the insulating carbon layer. – a totally different way at looking around. Researchers have had some ‘Molecular spintronics has only at systems architecture.’ Other than success in programming a metallic taken off in the last few years,’ dropping the power consumption MRAM element (see box p47) to says Walther Schwarzacher of the of existing electronic devices, carry out basic logical operations. University of Bristol in the UK, Awschalom says it’s not even obvious who is using scanning tunnelling that a spin transistor alone would The organic movement launch a spintronic revolution. ‘We Carbon-based materials such as can now approach industry with nanotubes, diamond and graphene robust, controllable quantum effects (sheets of carbon just a single at room temperature,’ he says. ‘It’s a atom thick), which can behave as pretty interesting time for us.’ DAVID AWSCHALOM DAVID semiconductors and insulators, display even weaker spin-orbit Matthew Chalmers is a freelance interactions than silicon, due to science writer based in Bristol, UK carbon’s lower atomic number and wider band gaps. That should allow electron spins to be preserved for Further reading  I Appelbaum et al, Nature, 2007, 447, 295 longer and at higher temperatures, (DOI: 10.1038/nature05803) although graphene so far has  D Awschalom and M Flatté, Nature Phys., performed worse than expected David Awschalom says 2007, 3, 153 (DOI: 10.1038/nphys551) and some researchers question ‘quantum engineers’  D Awschalom et al, Sci. Am., 2007, 297, 58  T Bland et al, Phys. World, January 2005, p24 whether spin transport is really being are needed to exploit  I Žutic et al, Rev. Mod. Phys., 2004, 76, 323 observed in organic semiconductors. spintronics know-how (DOI: 10.1103/RevModPhys.76.323)

www.chemistryworld.org Chemistry World | May 2009 | 49