Nature Milestones Spin

Contents | Zoom in | Zoom out

For navigation instructions please click here

Search Issue | Next Page

march 2008 http://www.nature.com/milestones/spin Supplement to Nature Publishing Group

Spin

MILESTONES

Produced by: Nature Physics

Contents | Zoom in | Zoom out

With support from: Organic Spintronics

For navigation instructions please click here

Search Issue | Next Page

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

Organic Spintronics Organic Spintronics is a high-tech company providing cutting edge solutions in nanotechnology • Pulsed Plasma Deposition (PPD) is an innovative thin film deposition technology based on pulsed plasma and electron ablation • Organic Knudsen sources for high and ultra high vacuum (UHV)

PPD characteristics: • High quality thin film deposition of inorganic materials (i.e. TCO, DLC, complex oxides, carbon nanotubes etc.) • Applications in flexible displays, wear resistant hard coatings, thin film photovoltaics etc. • High power efficiency • Large spatial homogeneity and throughput • Development of large area (multigun) industrially oriented deposition system • Low installation and running costs • Sustainable thin film fabrication for the academy and the industry

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

MILESTONES

Spin S4 S5 S6 S6 S7 S8 S8 S9 S10 S10 S11 S12 S12 S13 S14 S14 S15 S16 S16 S17 S18 S18 S19 S20

M I L E S TO N E S

COLLECTION

Timeline

HISTORY

Physics is set spinning (1896)

S21

Answers on a postcard (1922)

P. Zeeman

The spinning electron (1925) A relative success (1928)

HISTORY

S22

Spin’s nuclear sibling (1932)

Image formation by induced local interactions: examples employing nuclear magnetic resonance

Vital statistics (1940)

P. C. Lauterbur

New resonance (1946)

HISTORY

An attractive theory (1928)

From the compass to Apollo (1950s)

S23

Mind-boggling reality (1951) Odd one out (1964)

REVIEW

S25

Super symmetry (1971)

The way to NMR structures of proteins Kurt Wüthrich

A shift in expectations (1950–1951)

Challenges for semiconductor spintronics David D. Awschalom and Michael E. Flatté

Sticking together (1972) From spectrum to snapshot (1973) Solution for solution structures (1975–1976)

REVIEW

S32

Dilute for impact (1978) A giant leap for electronics (1988)

The emergence of spin electronics in data storage Claude Chappert, Albert Fert & Frédéric Nguyen Van Dau

Read my mind (1990) Information in a spin (1990) Feel the force (1991) The difficult middle ground (1996) The rise of semiconductor spintronics (1997)

CITING THE MILESTONES

VISIT THE SUPPLEMENT ONLINE

The Nature Milestones in Spin supplement has been published using material from Nature, Nature Materials, Nature Physics and Nature Structural Biology. However, most referencing formats and software do not allow the inclusion of more that one journal name or volume in an article reference. Therefore, should you wish to cite any of the Milestones, please reference the page number (Sxx–Sxx) as a supplement to Nature Physics. For example, Nature Phys. 4, Sxx–Sxx (2008).

The Nature Milestones in Spin supplement can be found at www.nature.com/milestones/spin All featured articles will be available free for six months.

To cite articles from the collection, please use the original citation, which can be found at the start of each article.

NATURE MILESTONES | SPIN

nature

The effect of magnetisation on the nature of light emitted by a substance

CONTRIBUTING JOURNAL Nature Physics Copyright © 2008 Nature Publishing Group

SUBSCRIPTIONS AND CUSTOMER SERVICES Americas: Nature Publishing Group, Customer Services, 75 Varick Street, 9th floor, New York, NY 10013-1917, USA. Tel: +1 212 726 9200; Fax: +1 212 696 9006 E-mail: ___________ [email protected]

UK/Europe/ROW (excluding Japan): Nature Publishing Group, Subscriptions, Brunel Road, Basingstoke, Hants RG21 6XS, UK. Tel: +44 (0)1256 302 827; Fax: +44 (0)1256 812 358 E-mail: __________ [email protected] Japan: Japan Publications Trading Co. Ltd, 2-1 Sarugaku-cho 1 chome, Chiyoda-ku, Tokyo 101, Japan. Tel: +81 (03) 292 3755 CUSTOMER SERVICES: _________ [email protected]

Getty Images are used unless otherwise credited.

MARCH 2008 | S1

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

ONLINE

www.nature.com/milestones/spin Nature Milestones in Spin will be available FREE online for six months, from 28 February 2008. The website also features a Library of original research papers and review articles from the NPG journals, covering the wealth of science that is relevant to spin: • fundamental physics • applied physics • magnetic resonance in physics • magnetic resonance in chemistry • magnetic resonance in biology and medicine • spintronics Key papers in the online Library will be available FREE for six months. More information about the sponsor of Nature Milestones in Spin, Organic Spintronics, can also be found on the website. Visit the website, and read for FREE: www.nature.com/milestones/spin Produced with support from:

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

N ▶ cover: Design by Judith Shadwell. Original photograph courtesy of Shawn Zhang | Agency: Dreamstime.com

EDITORIAL OFFICES london Nature Publishing Group The Macmillan Building, 4 Crinan Street, London N1 9XW, UK Tel: +44 (0)20 7843 3620; Fax: +44 (0)20 7843 3629

coordinating editors: Andreas Trabesinger and Alison Wright

consulting editors: Stuart Cantrill and Allison Doerr

art and production: Judith Shadwell production controller: Kelly Hopkins web production editor: Deborah Anthony web design: Glennis McGregor copy editor: Victoria Kitchener marketing: Katy Dunningham publisher: Sarah Greaves publishing assistant: Claudia Banks editor-in-chief, nature publications: Philip Campbell

sponsorship: Emma Green sponsorship pages: Suzanne Coleman Copyright © 2008 Nature Publishing Group MILESTONES ADVISORS

David Awschalom Bernard Barbara Michael Berry David Deutsch Tomasz Dietl David DiVincenzo Richard Ernst Ray Freeman Erwin Hahn Lewis Kay Anthony Leggett Christoph Lehner Hideo Ohno George Pickett Sandu Popescu Jeremy Sanders Robert Tycko Frank Wilczek Edward Witten Kurt Wüthrich Jörg Wrachtrup PRODUCED WITH SUPPORT FROM:

ature Milestones are special supplements that aim to highlight the ‘milestones’ or remarkable achievements in a given field. Each breakthrough is covered in a short Milestone article, written by an editor from the Nature Publishing Group, which discusses landmark discoveries in the context of the prevailing concepts at the time and our current knowledge of the field. Milestones in Spin — the sixth supplement in the series and the first in the physical sciences — presents key developments in the story of ‘spin’. Spin describes the intrinsic angular momentum of elementary particles, a concept developed in the mid-1920s as physicists sought to explain experimental observations made decades earlier. Since then, the idea of spin has inspired major advances in physics, and has found practical application in chemistry, biology and technology. For a topic as broad as ‘spin’, it is impossible to do justice to all of the influential developments. With the help of our advisors, we have selected 23 topics, highlighting where spin has taken us so far and providing a taste of what developments might be expected in the future. We have sought, in particular, to bring out the breadth of areas that are touched by spin: from the foundation of quantum physics, to analytical chemistry and structural biology, to cognitive neuroscience and information-storage technology. Many more topics remain unmentioned: the polarization of light, the discovery and explanation of the allotropic forms of hydrogen, solid-state NMR, colossal magnetoresistance, spin qubits, spin-torque phenomena — to name but a few. But a selection had to be made, and we hope that Milestones in Spin will prove an enriching reflection on the history of the field, and one that inspires further reading. Scientific discoveries are not, for the most part, achieved in single discrete steps. Rather they involve the work of many, and the fusion of ideas, concepts and experimental evidence to arrive at general acceptance. Therefore these articles are by no means intended to offer comprehensive coverage of a particular discovery. They aim instead to highlight a few key papers and to convey a historical perspective on how a particular idea evolved. In addition to the Milestone articles, the supplement includes a Timeline — a chronology of the earliest papers connected with each Milestone — and a reprinted Collection of relevant articles and reviews from Nature, Nature Materials, Nature Physics and Nature Structural Biology. The Milestones web site also includes an extensive Library of material from the across the Nature Publishing Group. The content of the supplement, as well as selected Library articles, will be available free from 28 February 2008 for six months. A free digital edition of the print supplement can also be downloaded from the web site; there are limited numbers available, so we encourage readers to download one as soon as possible. Finally, we would like to extend our sincere thanks to the advisors and to acknowledge support from our sponsor, Organic Spintronics. As always, Nature Publishing Group takes complete responsibility for the editorial content. Andreas Trabesinger, Senior Editor, Nature Physics Alison Wright, Chief Editor, Nature Physics

NATURE MILESTONES | SPIN

nature

MARCH 2008 | S3

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

MILESTONES TIMELINE 1896

Zeeman effect (1)

1922

Stern–Gerlach experiment (2)

1925

The spinning electron (3)

1928

Dirac equation (4) Quantum magnetism (5)

1932

Isospin (6)

1940

Spin–statistics connection (7)

1946

Nuclear magnetic resonance (8)

1950s

Development of magnetic devices (9)

1950–1951

NMR for chemical analysis (10)

1951

Einstein–Podolsky–Rosen argument in spin variables (11)

1964

Kondo effect (12)

1971

Supersymmetry (13)

1972

Superfluid helium-3 (14)

1973

Magnetic resonance imaging (15)

1975–1976

NMR for protein structure determination (16)

1978

Dilute magnetic semiconductors (17)

1988

Giant magnetoresistance (18)

1990

Functional MRI (19) Proposal for spin field-effect transistor (20)

1991

Magnetic resonance force microscopy (21)

1996

Mesoscopic tunnelling of magnetization (22)

1997

Semiconductor spintronics (23) Zoom-zoom | Dreamstime.com _______

S4 | MARCH 2008

nature

www.nature.com/milestones/spin

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

Razvanjp |________ Dreamstime.com

MILESTONES

M I L E S TO N E 1

Physics is set spinning …it would take nearly three decades for this ‘anomalous Zeeman effect’ to be explained…

“It was not simply out of a spirit of contradiction that I exposed a light source to a magnetic field,” so said Pieter Zeeman, on receiving the 1902 Nobel Prize in Physics with Hendrik Lorentz. The effect of a magnetic field on light had been studied years before, most notably by Michael Faraday: in 1845, he showed that when light passes through certain materials immersed in a magnetic field, the plane in which the light oscillates is rotated — now known as Faraday rotation, this was the first experimental evidence of a connection between light and electromagnetism. Later in his life, Faraday wondered whether a magnetic field could have a direct influence on a light source — specifically, on the light emitted by atoms or molecules when excited in a flame. This investigation was the last experiment recorded in his laboratory notebook, but the result was negative: on 12 March 1862, Faraday wrote that there was “not the slightest effect demonstrable”. Possibly referring to this experiment, James Clerk Maxwell stated in 1870, on the subject of light-emitting particles, that “no force in nature can alter even very slightly either their mass or their period of oscillation”. This statement, “coming from the mouth of the founder of the electromagnetic light theory and spoken with such intensity”, greatly troubled Zeeman. Motivated by his own studies of the magneto-optic Kerr effect (a phenomenon closely related to the Faraday effect, but measured for light reflected from a magnetized medium), Zeeman decided in 1896 to revisit Faraday’s last experiment. Two recent technical advances proved useful. The first was the invention, by Henry Augustus Rowland, of diffraction gratings consisting of mirrors ruled with a large number of parallel lines; Rowland’s gratings offered unprecedented spectral resolution. The second was the steady development of photography, which brought with

NATURE MILESTONES | SPIN

nature

it the possibility of capturing spectra and analysing them later. Zeeman was soon able to show that, for a sodium flame placed in an electromagnet, there was a broadening of the sodium D-lines when the magnet was switched on. The effect was reportedly far from conspicuous; however, firm support was provided by the observation of characteristic polarization effects. This confirmed predictions made by Lorentz — immediately after seeing the first results produced by Zeeman, he had proposed a model of electrons vibrating within the lightemitting particles. In 1897, Zeeman reported much clearer splittings in the blue line of cadmium. Although Lorentz’s theory did explain the most elementary splitting patterns, it was not long before numerous exceptions were found, as more elements were studied in greater detail. “Nature gives us all, including Professor Lorentz, surprises”, concluded Zeeman. He might have been surprised to know that it would take nearly three decades for this ‘anomalous Zeeman effect’ to be explained fully — when it was realized that the splitting is a consequence of spin (Milestone 3). Andreas Trabesinger, Senior Editor, Nature Physics ORIGINAL RESEARCH PAPERS Zeeman, P. Over den invloed eener magnetisatie op den aard van het door een stof uitgezonden licht. Versl. Kon. Akad. Wetensch. Amsterdam 5, 181–184, 242–248 (1896) | Zeeman, P. Over doubletten en tripletten in het spektrum, teweeggebracht door uitwendige magnetische krachten. Versl. Kon. Akad. Wetensch. Amsterdam 6, 13–18, 99–102, 260–262 (1897) FURTHER READING Zeeman, P. The effect of magnetisation on the nature of light emitted by a substance. Nature 55, 347 (1897) | Zeeman, P. Light radiation in a magnetic field, Nobel Lecture, 2 May, 1903. ______ Nobelprize.org [online], (1903) | Rayleigh. Pieter Zeeman. 1865–1943. Obit. Not. Fellows R. Soc. 4, 591–595 (1944)

MARCH 2008 | S5

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

MILESTONES M I L E S TO N E 2

Answers on a postcard The fact that quantities such as energy, mass and charge come in discrete and indivisible amounts, or quanta, is now so fundamental to our understanding of the Universe that it is often taken for granted. It is difficult to appreciate fully just how counterintuitive an idea it was to suggest that the angular momentum of an object should take on similarly quantized values. The roots of this idea lay in the 1913 description by Niels Bohr of the structure of the atom, later refined by Arnold Sommerfeld in 1916, which suggested that not only do electrons exist in orbitals of well-defined size and shape, but that the orientation of these orbitals is strictly defined — a characteristic known as space quantization. In 1920, most physicists, including Max Born, who was one of the architects of quantum mechanics, considered the idea to be more a mathematical abstraction than a concrete physical reality. Yet, in 1922, Otto Stern and Walter Gerlach demonstrated such a reality beyond all reasonable doubt. Their experiment involved passing a collimated beam of silver atoms through an inhomogeneous magnetic field and onto a glass slide where the deposits formed a pattern. Classical models suggested that the electron orbitals around the nucleus of these atoms should be randomly and continuously distributed, and that a single,

Postcard from Gerlach to Bohr. Image courtesy of Niels Bohr Archive, Copenhagen.

broad and continuous spot of silver should form in the centre of the slide. The Bohr–Sommerfeld model, by contrast, predicted that space quantization of these orbitals should cause the beam to be split into several discrete parts in the inhomogeneous field, forming discrete lines of silver deposits on the slide. Despite its elegant simplicity, the experiment almost never happened. For one thing, for there to be any observable splitting, the alignment of the beam and the centre of the magnetic field had to be just right. More prosaically, in the middle of a worsening economic depression, funding the construction of the experiment proved to be almost as difficult. Thankfully, perseverance on the part of Stern and Gerlach, and a cheque for several hundred dollars provided by Henry Goldman (co-founder of the investment firm Goldman Sachs), allowed them to observe the splitting predicted by quantum theory — a result

M I L E S TO N E 3

The spinning electron

Pauli and Bohr watch a spinning top. Photograph by Erik Gustafson, courtesy of AIP Emilio Segrè Visual Archives, Margrethe Bohr Collection.

S6 | MARCH 2008

nature

By 1920, physicists were still struggling to make sense of the splitting of atomic spectral lines in a magnetic field, discovered by Pieter Zeeman (Milestone 1). With the Bohr atom of 1913 had come the notion of quantized orbits for electrons around the atomic nucleus. Perhaps, it was thought, the Zeeman splitting arose from the interaction

between the angular momenta of the ‘core’ of electrons in closed shells and of the ‘radiant electron’ in the outermost unclosed shell. Yet calculations made on this basis — notably by Arnold Sommerfeld and Alfred Landé, using an Ersatzmodell — failed to match experimental data. The helium atom posed a particular problem: which of its two electrons should be considered ‘core’ and which ‘radiant’? Also, no one could explain why, if an atom were in its ground state, all its electrons were not bound into the innermost shell. In 1924, Landé gave up the struggle as “impossible once and for all”. Wolfgang Pauli, however, was undeterred. He dropped the notion of an interaction between core and radiant electron and proposed instead that line splitting arose as a consequence of an intrinsic property of the electron: “eine klassisch nicht beschreibbare Art von Zweideutigkeit” — a classically indescribable twovaluedness — as he wrote in the first of his two 1925 Zeitschrift für Physik papers.

Ralph Kronig, a young physicist in Landé’s laboratory, suggested to Pauli that this might be imagined as the rotation of an electron about its own axis with one half-unit of angular momentum — in other words, spin. Pauli disliked the idea: it was still ‘classically indescribable’ and, moreover, the calculations of level splitting by Kronig were a factor of two out from measured values. Kronig did not publish, but later that year George Uhlenbeck and Samuel Goudsmit did — the same idea, beset by the same problems. Pauli still didn’t like it. Nevertheless, it was a short step from the ‘two-valuedness’ of an intrinsic electron quantum number for Pauli to realize why all electrons in an atom are not bound in the innermost state. Guided by comments made by Edmund C. Stoner, Pauli saw that, if the four numbers used to describe line splitting (denoted n, k, m and j) were thought of as quantum numbers of the electron, then once an electron existed in some state defined by particular values of n, k, m and j, no other electron could enter that state. Using this ‘exclusion principle’, Pauli could derive the exact shell structure of the atom — two electrons to close the innermost shell, eight to close the next, and so on. The deal was sealed early in 1926, with the publication by Llewellyn H. Thomas of what became known as the

www.nature.com/milestones/spin

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

that is now most famously recorded on a postcard from Gerlach to Bohr congratulating him on the success of his theory. Although the Stern–Gerlach experiment categorically disproved classical models of the atom, it was also inconsistent with the Bohr–Sommerfeld model. In fact, the observed splitting of the silver beam had nothing to do with the orbital angular momentum, but was due to the spin angular momentum of the unpaired electron in the atomic structure of silver — something that was not appreciated until years later, following the introduction of the idea of electron spin by Wolfgang Pauli (Milestone 3). Not content with realizing what is perhaps the clearest and most direct demonstration of the quantum nature of atoms, Stern went on to demonstrate and measure the quantized spin of the proton — together with the size of its magnetic moment — for which he was awarded the 1943 Nobel Prize in Physics. Ed Gerstner, Senior Editor, Nature Physics ORIGINAL RESEARCH PAPERS Gerlach, W. & Stern, O. Der experimentelle Nachweis der Richtungsquantelung im Magnetfeld. Z. Phys. 9, 349–352 (1922) | Frisch, R. & Stern, O. Über die magnetische Ablenkung von Wasserstoff-Molekülen und das magnetische Moment des Protons. I. Z. Phys. 85, 4–16 (1933) | Estermann, I. & Stern, O. Über die magnetische Ablenkung von Wasserstoff-Molekülen und das magnetische Moment des Protons. II. Z. Phys. 85, 17–24 (1933) FURTHER READING Friedrich, B. & Herschbach, D. Stern and Gerlach: how a bad cigar helped reorient atomic physics. Phys. Today 56 (12), 53–59 (2003)

Pauli’s ‘twovaluedness’ was indeed due to the spin of the electron.

‘Thomas factor’ — the missing factor of two. Thomas’ classical analysis finally won over Pauli the perfectionist (Paul Dirac would soon supply the full quantum relativistic formalism; Milestone 4). Pauli’s ‘two-valuedness’ was indeed due to the spin of the electron. Probably the great man was cheered, having written to Kronig in May 1925, “At the moment physics is again terribly confused. In any case, it is too difficult for me, and I wish I had been a movie comedian or something of the sort and had never heard of physics.” Alison Wright, Chief Editor, Nature Physics ORIGINAL RESEARCH PAPERS Stoner, E. C. The distribution of electrons among atomic levels. Phil. Mag. 48, 719–736 (1924) | Pauli, W. Über den Einfluß der Geschwindigkeitsabhängigkeit der Elektronenmasse auf den Zeemaneffekt. Z. Phys. 31, 373–385 (1925) | Pauli, W. Über den Zusammenhang des Abschlusses der Elektronengruppen im Atom mit der Komplexstruktur der Spektren. Z. Phys. 31, 765–783 (1925) | Uhlenbeck, G. E. & Goudsmit, S. A. Ersetzung der Hypothese vom unmechanischen Zwang durch eine Forderung bezüglich des inneren Verhaltens jedes einzelnen Elektrons. Naturwiss. 13, 953–954 (1925) | Thomas, L. H. The motion of the spinning electron. Nature 117, 514 (1926) FURTHER READING Pauli, W. Exclusion principle and quantum mechanics, Nobel Lecture, 13 December, 1946.______ Nobelprize.org [online], _________________ (1946) | Tomonaga, S. The Story of Spin (Univ. Chicago Press, Chicago, 1997)

NATURE MILESTONES | SPIN

nature

M I L E S TO N E 4

A relative success

A

BEMaGS F

Image courtesy of Lida Lopes Cardozo Kindersley.

nature

The idea of the spinning electron, as proposed by Samuel Goudsmit and George Uhlenbeck in 1925 (Milestone 3), and incorporated into the formalism of quantum mechanics by Wolfgang Pauli, was a solution of expediency. Yet this contrivance threw up a more fundamental question: as a 25-year-old postdoctoral fellow at the University of Cambridge formulated the problem in 1928, why should nature have chosen this particular model for the electron, instead of being satisfied with a point charge? The young postdoc’s name was Paul Dirac, and in two papers published in the Proceedings of the Royal Society of London, he set out to explain why. Using the clear, austere prose and adroit mathematics that were his hallmarks, he showed how spin emerged as a natural consequence of the correct application of special relativity to the quantum mechanics of the electron. Dirac was able to remove the nonlinearities in space and time derivatives that had confounded other attempts to marry those two great new physical theories. The corollary of his logic was that the wavefunction of the electron must have four components, and must be operated on by four-dimensional matrices. These matrices required an additional degree of freedom beyond position and momentum in the physical description of the electron. Inspection revealed them to be extensions of the two-dimensional spin matrices introduced by Pauli in his earlier ad hoc treatment. Applied to an electron in an electromagnetic field, the new formalism delivered the exact value of the magnetic moment assumed in the spinning electron model. What had emerged was an equation that, in its author’s words, “governs most of physics and the whole of chemistry”. Dirac was a famously modest man, and was not wont to exaggerate. The Dirac equation is still today the best description not just of the electron, but of all spin-1/2 particles — including all the quarks and leptons from which matter is made. When asked what had led him to his formula, Dirac replied simply “I found it beautiful”. His equation is indeed a powerful example of the deep and mysterious connection between the language of mathematics and the expressions of the physical world. Yet, however much beauty might be indicative of rightness, a physical theory is judged on its predictive power. The Dirac equation did not disappoint. The interpretation of two of its four solutions was clear: they were the two spin states of the electron. But the other two solutions seemed to require particles exactly like electrons, but with a positive charge. Dirac did not immediately and explicitly state the now-obvious conclusion — out of “pure cowardice”, he explained later. But when, in 1932, Carl Anderson confirmed the existence of the positron, Dirac’s fame was assured. He shared the 1933 Nobel Prize in Physics — its second-youngest-ever recipient — and his equation went on to become the bedrock of quantum electrodynamics, the quantum field theory of the electromagnetic interaction. Following his death in 1984, a stone was set into the floor of Westminster Abbey in London. It was inscribed with his name and iγ · дΨ = mΨ — the shortest and sweetest rendering of his extraordinary brainchild. Richard Webb, Senior Editor, Nature News & Views ORIGINAL RESEARCH PAPERS Dirac, P. A. M. The quantum theory of the electron. Proc. R. Soc. Lond. A 117, 610–624 (1928); ibid. 118, 351–361 (1928) FURTHER READING Dirac, P. A. M. The Principles of Quantum Mechanics (Int. Ser. Monograph. Phys. 27) 4th edn (Oxford Univ. Press, Oxford, UK, 1982) | Pais, A., Jacob, M., Olive, D. I. & Atiyah, M. F. Paul Dirac: The Man and his Work (Cambridge Univ. Press, Cambridge, UK, 1998)

MARCH 2008 | S7

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

MILESTONES M I L E S TO N E 5

An attractive theory

STO CKB YTE

Since the earliest observation that a lump of lodestone attracts iron, around 600 BC, magnetism has attracted numerous philosophers and physicists eager to understand its secrets. Yet the physical explanation came only with the development of quantum mechanics in the 1920s — after all, magnetism is a purely quantum effect arising from the ‘spin’ property of the electron. By convention, we say that the spin points either up or down. This means that there are two distinct spin states instead of one;

facts that an electron has spin, as well as charge, and that two identical electrons must occupy different states, are the keys to the periodic table. Until then, the force aligning the electron spins could not be explained hence, certain theories were out in terms of known interactions, none by a factor of two with respect to of which was strong enough. In the experiments. words of Paul Dirac: “the solution of Following a series of papers this difficulty […] is provided by the published in the years 1925 to 1927 exchange (austausch) interaction of — during which quantum mechanthe electrons, which arises owing to ics was developed, interpreted and the electrons being indistinguishable applied to atoms with more than one from another. Two electrons may one electron outside a closed shell change places without our knowing — Werner Heisenberg solved the it, and the proper allowance for the mystery of ferromagnetism using the possibility of quantum jumps of this concept of spin plus the exclusion nature, which can be made in a treatprinciple formulated by Wolfgang ment of the problem by quantum Pauli, which states that two electrons mechanics, gives rise to the new kind with the same energy and momentum of interaction. The energies involved, cannot occupy the same quantum the so-called exchange energies, are state. In other words, two electrons quite large.” with the same energy but different By applying such an energy spins can lie in the same orbital. The tax on indistinguishable particles,

M I L E S TO N E 6

Spin’s nuclear sibling In 1932, Werner Heisenberg mused on an odd fact. The proton and the neutron (which had been discovered only earlier that same year by James Chadwick) had almost exactly the same mass. Despite their different charges, they also responded identically to the forces that dominate within the atomic nucleus. To Heisenberg’s nose, this had a whiff of an uncovered symmetry about it. Appropriating the mathematics that Wolfgang Pauli had used to describe spin (Milestone 3), he postulated that the proton and neutron were two states of the same particle, the nucleon. These states differed only in a quantity analogous to spin — the ‘isotopic spin’, or isospin as it came to be known. The nuclear force conserved isospin, which accounted for the similarities between protons and neutrons. Other forces, such as electromagnetism, broke isospin symmetry, which explained the nucleons’ differences. As Eugene Wigner wrote of the isospin concept in a 1937 paper, “no such states are known to be of any importance […] [but they] will turn out to be very

S8 | MARCH 2008

nature

useful”. In that paper, he used isospin to predict correctly the energies of all nuclei up to atomic number 42; more recent work has extended that success to even heavier nuclei. Much like the quantum-relativistic prediction of a spinning electron by Paul Dirac (Milestone 4), isospin was an example of what Wigner would later, in a celebrated essay, describe as the “unreasonable effectiveness” of mathematics in predicting physical phenomena. And how. In 1935, Hideki Yukawa modelled a nuclear force mediated by lighter particles exchanged between the nucleons. Isospin conservation demanded three such particles. Believers in unreasonable effectiveness could not have been surprised when, some 10 years later, three particles answering the description turned up in cosmic rays and in the first accelerator experiments: the two charged and one neutral pion. In 1954, Chen Ning Yang and Robert Mills took the ideas of Yukawa further to establish their principle of ‘gauge invariance’. This was the centrepiece of

Werner Heisenberg. Photograph by Friedrich Hund, courtesy of AIP Emilio Segrè Visual Archives.

a generalized mathematical description of forces mediated by exchange particles of integer spin and isospin — the bedrock of current quantum field theories of the fundamental forces of nature. Meanwhile, however, physics was entering the accelerator age, and the discovery of a seemingly unordered menagerie of particles similar to the pions was straining the foundation of isospin symmetry. It gradually became clear that isospin was not a fundamental symmetry, but just one corner of a larger edifice. In addition, the proton and neutron were not two states of the same particle. In fact, they were not elementary particles at all, but were made up of smaller entities — quarks.

www.nature.com/milestones/spin

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

MILESTONES Heisenberg proposed a model that counted up all the spins and included the exchange interaction between nearest neighbours only; for example, in a linear chain of spins, only two neighbouring spins would count, in a square lattice, four. When these were summed, Heisenberg found a ground state (lowest-energy configuration) in which all the spins of the electrons lined up in parallel — that is, a ferromagnetic state without the need for any external magnetic field. May Chiao, Senior Editor, Nature Physics ORIGINAL RESEARCH PAPERS Heisenberg, W. Zur Theorie des Ferromagnetismus. Z. Phys. 49, 619–636 (1928) | Dirac, P. A. M. Quantum mechanics in many-electron systems. Proc. R. Soc. Lond. A 123, 714–733 (1929) FURTHER READING Mott, N. & Peierls, R. Werner Heisenberg. 5 December 1901–1 February 1976. Biogr. Mem. Fellows R. Soc. 23, 212–251 (1977)

Once deprived of its original legitimacy, one might have expected spin’s nuclear sibling to disappear. In fact, the same mathematical description resurfaced within the new fundamental paradigm as ‘weak isospin’, which is a property of quarks that is conserved in weak interactions — a testament to how deeply embedded the language of spin seems to be in the workings of the world. Richard Webb, Senior Editor, Nature News & Views ORIGINAL RESEARCH PAPERS Heisenberg, W. Über den Bau der Atomkerne. Z. Phys. 77, 1–11 (1932) | Yukawa, H. On the interaction of elementary particles. Proc. Phys. Math. Soc. Jap. 17, 48–57 (1935) | Wigner, E. On the consequences of the symmetry of the nuclear Hamiltonian on the spectroscopy of nuclei. Phys. Rev. 51, 106–119 (1937) | Yang, C. N. & Mills, R. L. Conservation of isotopic spin and isotopic gauge invariance. Phys. Rev. 96, 191–195 (1954) | Gell-Mann, M. Symmetries of baryons and mesons. Phys. Rev. 125, 1067–1084 (1962) FURTHER READING Wigner, E. The unreasonable effectiveness of mathematics in the natural sciences. Comm. Pure Appl. Math. 13, 1–14 (1960) | Robson, D. Isospin in nuclei. Science 179, 133–139 (1973) | Warner, D. D., Bentley, M. A. & Van Isacker, P. The role of isospin symmetry in collective nuclear structure. Nature Phys. 2, 311–318 (2006) | Anderson, R. & Joshi, G. C. Interpreting mathematics in physics: charting the applications of SU(2) in 20th century physics, arXiv.org [online] (2006) _____________

NATURE MILESTONES | SPIN

nature

M I L E S TO N E 7

Vital statistics The Dirac equation is monumental in physics, encapsulating so beautifully, in relativistic terms, the behaviour of a spinning electron, or indeed of any particle that has half-integer spin (Milestone 4). Wolfgang Pauli — on whose ideas the concept of a spinning electron was based — was impressed by the mathematical ‘acrobatics’ of Paul Dirac in arriving at the succinct expression, published in 1928, but he was not, however, satisfied. Pauli questioned the reliance of Dirac’s theory on the exclusion principle and the emphasis on a half-unit of spin — why should nature permit only half units? With Victor Weisskopf, Pauli set about resurrecting the Klein–Gordon equation, which describes a particle that has zero spin, but which had been all but abandoned following the unsuccessful attempt by Erwin Schrödinger in 1926 to build it into a theory of quantum-wave mechanics. Weisskopf and Pauli, however, succeeded in quantizing the Klein–Gordon equation to obtain spin-0 particles of both negative and positive charge — just as Dirac had obtained spin-1/2 particles of negative and positive charge from his equation. These spin-0 particles, moreover, did not obey the exclusion principle. Dirac’s ‘positive electron’ — the positron — was discovered (although not immediately recognized as such) by Carl Anderson in 1932, the same year that James Chadwick discovered the neutron; in 1935, Hideki Yukawa postulated the existence of the meson, and the muon was discovered in 1936. There was suddenly a growing family of particles to describe, alongside the electron, proton and photon. It was om e.c thinking about how to reconcile the Klein–Gordon and Dirac tim s am equations, and the existence of all these particles (how many re |D z u more might be discovered?) that led Pauli to one of the most l los Mi subtle concepts of modern physics — the spin–statistics theorem. In his 1940 paper, Pauli identified a vital connection between spin and quantum statistics (in the 1920s, it had been realized that something more than the Maxwell–Boltzmann variety was needed at the quantum level). According to Pauli, particles of half-integer spin obey Fermi–Dirac statistics (and, hence, are now called ‘fermions’) and those of integer spin obey Bose–Einstein statistics (‘bosons’). Mathematically speaking, the quantization of fields with half-integer spin relies on ‘plus’ commutation relations, whereas that of fields with integer spin uses ‘minus’ commutation relations. Put another way, the wavefunction of a system of bosons is symmetric if any pair of bosons is interchanged, but is antisymmetric for interchanged particles in a system of fermions. Subtle indeed, but from Pauli’s spin–statistics connection arises the exclusion principle for fermions, with its implications for atomic structure, and a ‘non-exclusion’ principle for bosons — many bosons can adopt the According to same quantum state at once, as happens in a Bose–Einstein condensate. Pauli, particles of Further particle discoveries since 1940 and the subsequent building of the half-integer spin ‘standard model’ have also served to confirm that nature works with both obey Fermi–Dirac integer and half-integer spins. Alison Wright, Chief Editor, Nature Physics

ORIGINAL RESEARCH PAPERS Pauli, W. & Weisskopf, V. F. Über die Quantisierung der skalaren relativistischen Wellengleichung. Helv. Phys. Acta 7, 709–731 (1934) | Pauli, W. The connection between spin and statistics. Phys. Rev. 58, 716–722 (1940) FURTHER READING Duck, I. & Sudarshan, E. C. G. Toward an understanding of the spin–statistics theorem. Am. J. Phys. 66, 284–303 (1998) | Tomonaga, S. The Story of Spin (Univ. Chicago Press, Chicago, 1997)

statistics and those of integer spin obey Bose–Einstein statistics.

MARCH 2008 | S9

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

MILESTONES M I L E S TO N E 8

New resonance If, during the 1920s and 1930s, the atomic nucleus had seemed of interest to few besides the (mostly) gentleman scientists studying it, by the end of the Second World War its wider importance was abundantly clear. The coming of the nuclear age was an appropriate cue for the two papers that cleared the way for arguably the most widespread practical application of nuclear spin today: nuclear

First NMR signals from water. Image reprinted with permission from Bloch, F., Hansen, W. W. and Packard, M. Phys. Rev. 70, 474–485 (1946). Courtesy of the American Physical Society.

The coming of the nuclear age was an appropriate cue… for arguably the most widespread practical application of nuclear spin today.

magnetic resonance (NMR) spectroscopy. The 1946 work of Edward Mills Purcell at the Massachusetts Institute of Technology and Felix Bloch at Stanford University gave new relevance to one object of intense gentlemanly interest before the war: the Zeeman splitting of nuclear spin states in a magnetic field (Milestone 1). The degree of splitting at a particular magnetic field strength depends on the gyromagnetic ratio of the nucleus. In NMR, a second, transverse field at the characteristic (typically radio) spin-transition frequency produces an absorption resonance — a powerful way to identify the nuclei present in a sample. Purcell et al. brought protons (1H) in solid paraffin to resonance; Bloch et al. did the same in liquid water. The coincident timing was no accident: the development of radar technologies during the war, for which several of the researchers involved had won their spurs, had made sources of radiofrequency radiation freely available for the first time. The effect itself was not entirely new. In 1938, Isidor Rabi had used it to measure magnetic moments of both atomic species in a lithium chloride molecular beam, receiving the 1944

Nobel Prize in Physics for that advance. Even earlier, the Dutch physicist Cornelis J. Gorter had looked for the resonance of 7Li in lithium fluoride and 1 H in alum, using a calorimetric method. Hampered by experimental vagaries and limited resources, he published a negative result. (In later years, on receiving a prize for his contributions to low-temperature physics, Gorter would muse on his strange ability to miss out on groundbreaking discoveries in this and other instances.) The innovations offered by Bloch and Purcell’s approaches were the transition to real liquid and solid systems, and, in Bloch’s case, the use of an induction coil to pick up and sharpen the resonance signal. These opened the way for the use of NMR in all manner of contexts, including in living tissue — where it became the lynchpin of magnetic resonance imaging (Milestone 15). In 1944, although shielded in the relative obscurity of Kazan in the steppes of Tatarstan, the Soviet physicist Yevgeny Zavoisky published the first measurements of an analogous effect involving electron spins. Electron paramagnetic resonance depends on an atom possessing an unpaired electron,

M I L E S TO N E 9

From the compass to Apollo Long before the concept of spin had been realized, the phenomenon of magnetism was a source of fascination and curiosity. The first scientific record of magnetism was made by the Greek philosopher Thales of Miletos, who, in the sixth century bc, studied the attraction of materials such as iron to loadstone (magnetite). The first magnetic device was, of course, the compass — probably invented by several cultures independently and first documented in Chinese literature in the eleventh century ad. Nearly a millennium later, and particularly since the 1950s, devices based on magnetism are once more proving significant in shaping our way of life. The magnetic tape, which was invented in 1878 by Oberlin Smith, was commercialized in the 1930s by AEG and BASF. In later

S10 | MARCH 2008

nature

decades, it was developed into, for example, videotape in 1951 and the magnetic stripes on credit cards in the 1960s. Following the invention of modern computers, technology similar to magnetic tape was the logical choice for long-term data storage. The first hard drive with a moveable head was built into the IBM 305 computer, which shipped in 1956. Its large hard disks — 24 inches in diameter — had a storage density of 2 kilobits per square inch. IBM was also a pioneer in the development of the removable floppy disk. The first floppy disk, which had a diameter of 8 inches and a storage capacity of about 80 kilobytes, dates from 1969. The cumbersome 8-inch format was soon brought down in size: the last popular format was the Sony 3.5-inch floppy.

Magnetism has also been the key to several other, historical, storage techniques. The earliest was the ‘drum memory’ in the 1950s, which consisted of rotating circular metallic plates coated with a magnetic material. Drum memory was superseded in the 1960s by the ‘core memory’ — a hand-woven grid of wires with small ferrite rings (the cores) at the intersections. Complex current pulses

Magnetic core random access memory. Image courtesy of H. J. Sommer III, Professor of Mechanical Engineering, Penn State University.

www.nature.com/milestones/spin

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

MILESTONES thereby limiting the range of its application, but making it useful for the detection and identification of free radicals. Zavoisky might also have been the first to see an NMR signal, but he did not follow it up, at least not with publications. Had the vicissitudes of the age been less, and the dissemination of scientific information easier, his claim might have been better heard in the West. As it was, the 1952 Nobel Prize in Physics went to Bloch and Purcell. Richard Webb, Senior Editor, Nature News & Views ORIGINAL RESEARCH PAPERS Gorter, C. J. Negative results of an attempt to detect nuclear magnetic spins. Physica 9, 995–998 (1936) | Rabi, I. I., Zacharias, J. R., Millman, S. & Kusch, P. A new method of measuring nuclear magnetic moment. Phys. Rev. 53, 318 (1938) | Zavoisky, E. Relaxation of liquid solutions for perpendicular fields. J. Phys. USSR 9, 211–216 (1945) | Purcell, E. M., Torrey, H. C. & Pound, R. V. Resonance absorption by nuclear magnetic moments in a solid. Phys. Rev. 69, 37–38 (1946) | Bloch, F., Hansen, W. W. & Packard, M. Nuclear induction. Phys. Rev. 69, 127 (1946) | Zavoisky, E. Spin magnetic resonance in the decimetre-wave region. J. Phys. USSR 10, 197–198 (1946) FURTHER READING Gorter, C. J. Bad luck in attempts to make scientific discoveries. Phys. Today 20 (1), 76–81 (1967) | Kochelaev, B. I. & Yablokov, Y. V. The Beginning of Paramagnetic Resonance (World Scientific, Singapore, 1995)

through the wires were able to read, as well as set, the magnetization of the cores. Despite being an intricate device, a core memory of two cubic feet, with a capacity of 4,096 words, was used in the Apollo guidance computer, onboard the NASA missions to the Moon. Computer memory was miniaturized further in the late 1970s, for example using ‘bubble memory’ in which data storage is based on small magnetic domains on a thin film. Soon afterwards, hard drives became the dominant data-storage system for computers. The history of magnetic devices illustrates well how, with a little inventiveness, the macroscopic manifestations of magnetism can be harvested to achieve amazing technological advances. However, it would take a more fundamental understanding of spin physics to achieve the next technological revolution in computing and information storage (Milestone 18). Joerg Heber, Senior Editor, Nature Materials

NATURE MILESTONES | SPIN

nature

M I L E S TO N E 1 0

A shift in expectations Nuclear magnetic resonance (NMR) spectroscopy (Milestone 8) is one of the most powerful analytical techniques in modern chemistry — a window into the world of molecules that can provide information about their structures, dynamic behaviour and how they interact with one another. Prior to the 1950s, the study of NMR was rooted firmly in the physics community. It was assumed that the frequency at which a given nucleus resonated depended only on the strength of the magnetic field in which it was placed. Physicists therefore anticipated that they could use the technique to measure — with unprecedented precision — the magnetic moments of different nuclei. When, in 1950, Warren Proctor and Fu Chun Yu set out to do this for 14N, something unexpected happened. For their experiments, they chose the compound ammonium nitrate (NH4NO3), which is highly soluble in water and contains two nitrogen nuclei per molecule — factors that were expected to improve the NMR signal. In what they described as a “surprising observation”, however, not one but two resonance frequencies were detected — one for the nitrogen nuclei in the ammonium (NH4+) ions and the other for those in the nitrate (NO3–) ions. This was the first reported observation of the phenomenon that soon became known as ‘chemical shift’, in which the local chemical environment surrounding a nucleus influences the frequency at which it resonates. The implications of NMR for the structural analysis of organic compounds became apparent soon afterwards, when, in 1951, a group of researchers from Stanford University showed that different 1H nuclei in the same molecule resonate at different frequencies. James Arnold, Srinivas Dharmatti and Martin Packard demonstrated the huge potential of NMR spectroscopy by applying the technique to ethanol (CH3CH2OH), a compound in which each molecule comprises three sets of non-equivalent 1 H nuclei. Using tiny sample volumes and placing them in the most uniform region within a magnetic field, they obtained a spectrum displaying three separate lines, corresponding to the resonant frequencies of the 1H nuclei in the CH3, CH2 and OH groups, respectively. Moreover, the relative intensities of the three signals corresponded with the number of protons in each different chemical environment. So it was possible not only to identify different molecular fragments but also to glean

Paolotoscani | Dreamstime.com

quantitative information about the number of equivalent nuclei in each. Later in 1951, Herbert Gutowsky and David McCall showed that different spin-active nuclei in the same molecule interact with one another, giving rise to fine structure in the NMR signals that encodes a wealth of information regarding molecular connectivity and structure. It did not take the chemistry community long to embrace the technique for the spectroscopic analysis of compounds. Techniques using radiofrequency pulses — rather than a continuous source — broadened the scope of NMR spectroscopy, and Fourier-transform methods of data processing notably improved the sensitivity of the method. The combination of these advances allowed the development of sophisticated multidimensional NMR experiments that revolutionized the field (Milestone 16). Stuart Cantrill, Chief Editor, Nature Chemistry ORIGINAL RESEARCH PAPERS Proctor, W. G. & Yu, F. C. The dependence of a nuclear magnetic resonance frequency upon chemical compound. Phys. Rev. 77, 717 (1950) | Hahn, E. L. Spin echoes. Phys. Rev. 80, 580–594 (1950) | Arnold, J. T., Dharmatti, S. S. & Packard, M. E. Chemical effects on nuclear induction signals from organic compounds. J. Chem. Phys. 19, 507 (1951) | Gutowsky, H. S. & McCall, D. W. Nuclear magnetic resonance fine structure in liquids. Phys. Rev. 82, 748–749 (1951) | Carr, H. Y. & Purcell, E. M. Effects of diffusion on free precession in nuclear magnetic resonance experiments. Phys. Rev. 94, 630–638 (1954) | Ernst, R. R. & Anderson, W. A. Application of Fourier transform spectroscopy to magnetic resonance. Rev. Sci. Instrum. 37, 93–102 (1966) | Aue, W. P., Bartholdi, E. & Ernst, R. R. Two dimensional spectroscopy. Application to nuclear magnetic resonance. J. Chem. Phys. 64, 2229–2246 (1976) FURTHER READING Becker, E. D. Magnetic resonance: an account of some key discoveries and their consequences. Appl. Spectrosc. 50, 16A–28A (1996)

MARCH 2008 | S11

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

M I L E S TO N E 1 1

Mind-boggling reality

David Bohm. Image from Library of Congress, New York — Telegram and Sun Collection, courtesy of AIP Emilio Segrè Visual Archives.

In 1935, Albert Einstein, Boris Podolsky and Nathan Rosen questioned whether quantum mechanics fully describes ‘physical reality’. Their paper, which was intended to illustrate that quantum mechanics is incomplete, sparked discussions that go deep into the philosophical aspects of ‘reality’ and how physics can describe it. Later that year, Einstein confessed in a letter to Erwin Schrödinger that he felt that “the main point was, so to speak, buried by erudition”, and began publishing his own versions of the ‘incompleteness argument’. All of these accounts, and the original Einstein–Podolsky–Rosen paper, made their point using continuous variables — that is, position and momentum. However, the version that is most widely discussed in the modern literature, and also forms

the basis of many experimental investigations, presents the argument in a simpler and clearer form, in terms of discrete spin variables. It was penned by David Bohm and appeared originally in his 1951 book Quantum Theory; he developed the argument further, in the context of experimental proofs, with Yakir Aharonov in 1957. Bohm and Aharonov considered a molecule made of two atoms — each having one half-unit of spin — combined such that the total spin of the molecule is zero. When the two atoms are separated, and, for one of the spins, the spin component is measured along a given direction, the same component is immediately known for the other spin — it is exactly the opposite, as the total spin still has to be zero. At first sight, it might not seem surprising that information about the properties of the second particle of a composite system can be deduced without performing any measurement on it, and without any interaction between the two particles, if the initial condition restricts how the two particles behave with respect to each other.

M I L E S TO N E 1 2

Odd one out Impurities are not always unwanted. With the right type and dose of impurity atoms, the bulk properties of a material can be tuned in a beneficial way, which is a technique made heavy use of, for example, in standard silicon technology. At the microscopic level, interesting questions arise about how an impurity atom interacts with its host. In 1964, Jun Kondo resolved a long-standing question regarding the electrical resistance of magnetic impurity-doped metals. The mystery was this: the resistance of a metal should decrease with decreasing temperature, as atomic vibrations freeze out, so that conduction electrons can move more easily through the material; however, for magnetically doped metals, the resistance was found to increase again below a certain temperature. Kondo discovered that it is the intrinsic spin of

S12 | MARCH 2008

nature

magnetic impurity atoms that leads to this anomalous resistance. The amount of scattering that electrons experience at the impurities does not decrease but increases when the temperature goes down, and this leads to the observed minimum in total resistance. Not only did this finding explain a nagging problem but it also triggered a vast amount of theoretical follow-up work. The initial issue to tackle was that the effect seemed to yield infinite resistance as zero temperature is approached — clearly an unphysical result. It was soon found that this divergence of resistance is suppressed, below a certain temperature (the Kondo temperature), by the formation of a bound state between impurity and conduction electrons, in which electron spins line up to screen the spin of the impurity atom. A later development was

Jun Kondo. Image courtesy of AIST, Tokyo.

A

BEMaGS F

For a quantum spin, however, the situation is more subtle. Quantum mechanics allows only one component of the spin to have a definite value. If, for instance, the x component of the spin is known, then the components along the y and z axes must be indeterminate; the component that is definite is determined by its measurement. Yet, in this case of two separated spin-1/2 particles, an experimenter can decide at the last minute — long after the two constituents have been separated — along which direction the first spin is measured. And this choice has immediate consequences on which component of the second, unobserved, spin is definite. How can the second spin know what has been done to the first? Is there some kind of hidden interaction that quantum theory does not account for? Does quantum mechanics allow what Einstein famously called “spooky action at a distance” (an idea he did not like)? Einstein argued that if no action at a distance can instantaneously influence the second spin, then it must have had all its components well defined from the

the extension of the theory to ‘Kondo lattices’ in which electrons interact not only with the odd magnetic impurity but rather with an array of localized spins, and are significantly slowed down by the strong interactions. This model can explain some of the unusual properties, such as anomalous superconductivity, displayed by so-called heavy-fermion compounds. Another line of research is the Kondo effect in nanometre-sized structures, such as quantum dots, in which the interactions between a single magnetic impurity and its environment can be controlled. A quantum dot can be tuned to contain an odd number of electrons — that is, an unpaired spin. Below the Kondo temperature, this localized spin can form a bound state with the free electrons in the electrodes on either side of the quantum dot, similar to the classical Kondo effect. However, this ‘Kondo resonance’ opens an additional pathway for electrons to flow through the quantum dot and, as a result, the resistance decreases, in contrast to the original effect.

www.nature.com/milestones/spin

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

outset — hence, quantum mechanics must be incomplete. A decisive step came in 1964 when John Bell, building on the Bohm–Aharonov formulation in spin variables, showed that quantum mechanics makes predictions that contradict the local-realistic world view of Einstein and do require action at a distance of some sort. Bell’s theorem has been put to the test many times since, and although there is, as yet, no single experiment that closes all possible loopholes, the weight of evidence does still favour quantum mechanics. Andreas Trabesinger, Senior Editor, Nature Physics ORIGINAL RESEARCH PAPERS Einstein, A., Podolsky, B. & Rosen, N. Can quantummechanical description of physical reality be considered complete? Phys. Rev. 47, 777–780 (1935) | Bohm, D. Quantum Theory Ch. XXII (Prentice-Hall, Englewood Cliffs, New Jersey, 1951) | Bohm, D. & Aharonov, Y. Discussion of experimental proof for the paradox of Einstein, Rosen, and Podolsky. Phys. Rev. 108, 1070–1076 (1957) | Bell, J. S. On the Einstein Podolsky Rosen paradox. Physics 1, 195–200 (1964) FURTHER READING Bell, J. S. Bertlmann’s socks and the nature of reality. J. Phys. 42, 41–62 (1981) | Sauer, T. An Einstein manuscript on the EPR paradox for spin observables. Stud. Hist. Philos. Mod. Phys. 38, 879–887 (2007)

The ideas and methods developed by Kondo and his fellow theorists turned out to be relevant to a wide range of problems that involve strong interactions between particles. As a result, the ‘Kondo effect’ — which, in truth, comprises a range of phenomena to do with collective behaviour arising from localized magnetic impurities — is an active research topic today and one that still throws up surprises. Liesbeth Venema, Senior Editor, Nature

ORIGINAL RESEARCH PAPERS Kondo, J. Resistance minimum in dilute magnetic alloys. Prog. Theor. Phys. 32, 37–49 (1964) | Anderson, P. W. A poor man’s derivation of scaling laws for the Kondo problem. J. Phys. C 3, 2346–2441 (1970) | Goldhaber-Gordon, D. et al. Kondo effect in a single-electron transistor. Nature 391, 156–159 (1998) FURTHER READING Wilson, K. G. The renormalization group: critical phenomena and the Kondo problem. Rev. Mod. Phys. 47, 773–840 (1975) | Tsvelik, A. M. & Wiegmann, P. B. Exact results in the theory of magnetic alloys. Adv. Phys. 32, 453–713 (1983) | Kouwenhoven, L. & Glazman, L. Revival of the Kondo effect. Phys. World 14(1), 33–38 (2001)

NATURE MILESTONES | SPIN

nature

A

BEMaGS F

M I L E S TO N E 1 3

Super symmetry The way that spin is woven into the very fabric of the Universe is writ large in the standard model of particle physics. In this model, which took shape in the 1970s and can explain the results of all particle-physics experiments to date, matter (and antimatter) is made of three families of quarks and leptons, which are all fermions, whereas the electromagnetic, strong and weak forces that act on these particles are carried by other particles, such as photons and gluons, which are all bosons. Despite its success, the standard model is unsatisfactory for a number of reasons. First, although the electromagnetic and weak forces have been unified into a single force, a ‘grand unified theory’ that brings the strong interaction into the fold remains elusive. Second, the origins of mass are not fully understood. Third, gravity is not included. Moreover, there are other, less obvious problems with the standard model. The two natural mass scales in nature are zero and the Planck mass, ~ 1019 GeV c−2. Neither photons nor gluons (which carry the electromagnetic and strong forces, respectively) have mass, but the W and Z bosons that are responsible for the weak force have masses of about 90 GeV c−2. Where does this mass scale come from? This ‘hierarchy problem’ can be solved by finetuning the model so that various quantum fluctuations cancel out, although many physicists are uncomfortable with this solution because some parameters must be fine-tuned to better than 1 part in 1015. However, a form of symmetry between fermions and bosons called supersymmetry offers a much more elegant solution because the quantum fluctuations caused by bosons are naturally cancelled out by those caused by fermions and vice versa. Symmetry plays a central role in physics. The fact that the laws of physics are, for instance, symmetric in time (that is, they do not change with time) leads to the conservation of energy. These laws are also symmetric with respect to space, rotation and relative motion. Initially explored in the early 1970s, supersymmetry is a less obvious kind of symmetry, which, if it exists in nature, would mean that the laws of physics do

The ATLAS experiment under construction at the Large Hadron Collider. Image courtesy of CERN.

not change when bosons are replaced by fermions, and fermions are replaced by bosons. Although it is difficult to explain supersymmetry through analogies to classical physics, its consequences are dramatic — it predicts that every fundamental particle has a superpartner with half a unit of spin less. The electron, for instance, has a spin of a half, so its superpartner (which is known as a selectron) has zero spin. This means that the superpartner of a boson is always a fermion and vice versa. Supersymmetry also plays a central role in theories that attempt to unify the forces in the standard model with gravity by treating fundamental particles as vibrating strings or membranes in 10-dimensional or 11-dimensional spacetimes. In these theories the gravitational force is carried by a spin-two boson called the graviton. Searching for supersymmetric particles will be a priority when the Large Hadron Collider comes into operation at CERN, the European particle-physics laboratory near Geneva, in 2008. Peter Rodgers, Chief Editor, Nature Nanotechnology ORIGINAL RESEARCH PAPERS Golfand, Y. A. & Likhtman, E. P. Extension of the algebra of Poincaré group generators and violation of P invariance. JETP Lett. 13, 323–326 (1971) | Neveu, A. & Schwarz, J. H. Factorizable dual model of pions. Nucl. Phys. B 31, 86–112 (1971) | Ramond, P. Dual theory for free fermions. Phys. Rev. D 3, 2415–2418 (1971) | Wess, J. & Zumino, B. Supergauge transformations in four dimensions. Nucl. Phys. B 70, 39–50 (1974) | Wess, J. & Zumino, B. A Lagrangian model invariant under supergauge transformations. Phys. Lett. B 49, 52–54 (1974) FURTHER READING Dimopoulos, S., Raby, S. & Wilczek, F. Supersymmetry and the scale of unification. Phys. Rev. D 24, 1681–1683 (1981) | Almadi, U., de Boar, W. & Furstenau, H. Comparison of grand unified theories with electroweak and strong coupling constants measured at LEP. Phys. Lett. B 260, 447–455 (1991) | Greene, B. The Elegant Universe (Vintage, London, 2000) | Kane, G. & Shifman, M. (eds) The Supersymmetric World: The Beginnings of the Theory (World Scientific, Singapore, 2000)

MARCH 2008 | S13

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

MILESTONES M I L E S TO N E 1 4

Sticking together The fact that a dirty metal can support an electric current flowing without resistance might sound exotic to some, but it is a textbook property of ‘conventional’ superconductors. When spins are involved, however, the superconductors do become ‘exotic’. The conventional mechanism of superconductivity was explained by John Bardeen, Leon Cooper and Robert Schrieffer (in what is known as the BCS theory) back in 1957. For decades, it was a mystery how electrons, which are classified as ‘fermions’, could be forced into a single ground state that was more typical for ‘bosons’ (as when they undergo Bose–Einstein condensation). Fermions cannot all pile into the same ground state because only two — one with its spin pointing upwards and the other pointing downwards — can occupy each quantum state; bosons do not heed such conventions. As it turns

out, the only way for fermions to form a condensate is for them to pair up, with the help of the crystalline lattice: when one electron passes through the lattice, the positive ions are slightly attracted to the passing negative electron; if a second electron comes along, it will sense the deformed lattice and be attracted to the net positive charge, and hence to the original electron. This kind of lattice-assisted coupling is weak, but it is strong enough for the paired electrons to drop down to a collective ground state. Naturally, people considered whether such pairing could glue other fermions together, in particular 3 He. As 4He exhibits superfluidity — that is, the liquid can flow without viscosity below a certain temperature — it was believed (hoped) that 3He would do likewise. Yet, given that 4He is a boson and 3He is a fermion, it was not clear how 3He could condense until the BCS theory came along. However, the magnetic interactions between the 3He particles are strong, and so they cannot pair up as electrons do. Instead, the pairing glue must come from another source.

M I L E S TO N E 1 5

From spectrum to snapshot The scientific principles of magnetic resonance imaging (MRI) stem from those of nuclear magnetic resonance (NMR; Milestone 8); however, now, especially in the mind of the public, the latter lies very much in the shadow of the former. The technological jump from NMR spectrum to MR image began in the early 1970s, and subsequent developments have established MRI as a priceless technique in medical research and diagnostics. MRI uses magnetic fields and radiowaves to produce, in a non-invasive manner, ‘tomographic’ images of a three-dimensional object. Paul Lauterbur introduced the scientific basis behind this mode of visualization as “image formation by induced local interactions”. His idea was to combine two magnetic fields, so that one induces an interaction whereas the other restricts this interaction to a localized region in space. He proposed the term ‘zeugmatography’

S14 | MARCH 2008

nature

to describe the technique, from a Greek word meaning ‘that which is used for joining’; however, the name never became widely accepted. Lauterbur’s pioneering experiment involved the imaging of a cross-section through two glass tubes of ordinary water (H2O) attached to the inside wall of a larger tube of deuterated water (D2O). A two-dimensional image

The first authors to propose ferromagnetic fluctuations of the spins as such a glue were A. Layzer and D. Fay. When one particle whizzes through the liquid, its spin (pointing upwards, for instance) attracts other spin-up particles and repels spin-down particles. The effective spin-up polarization can then attract another spin-up particle, leading to a spin-up–spin-up pair. When coupled in this way, the 3He atoms are able to form a superfluid. Amazingly, all the theoretical groundwork, including that by Philip Anderson and Pierre Morel, and by Roger Balian and Richard Werthamer, was laid down before the experimental confirmation of a superfluid state in 3He — that came in 1972 and earned its authors, Douglas Osheroff, Robert Richardson and David Lee, the Nobel prize in Physics in 1996. Although superfluidity was anticipated, measurements revealed several unique superfluid phases in 3He. Moreover, because of the non-zero spin of the pairs (in conventional superconductors the net spin is zero),

showing the location of the tubes of H2O was generated by combining four projections taken from different angles around the set-up. Some scepticism surrounded this initial observation. It seemed counterintuitive that radiowaves could be used to image objects that were much smaller than their wavelength. Yet it was because the interactions were restricted to certain regions that, in fact, the technique became independent of wavelength. Lauterbur recognized the potential of the concept at the time of the first simple experiment. He believed it could be used to investigate complex systems and noted the possibility of visualizing biological tissues — in particular, distinguishing between malignant tumours and healthy tissue. Many of the early practical developments in MRI were made by Peter Mansfield, who discovered how to acquire images rapidly. In recognition of their contributions, Lauterbur and Mansfield shared the 2003 Nobel Prize in Medicine. The invention of MRI has not been without controversy. Others have claimed to have produced the first

www.nature.com/milestones/spin

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

MILESTONES 3

He also yielded some unexpected properties. In 1987, a team working in Moscow discovered a pure spin supercurrent. Unlike the supercurrent in a conventional superconductor that carries charge and mass, the spin supercurrent carries only spin and there is no mass flow. 3He is truly exotic, because of its spin. May Chiao, Senior Editor, Nature Physics ORIGINAL RESEARCH PAPERS Bardeen, J., Cooper, L. N. & Schrieffer, J. R. Microscopic theory of superconductivity. Phys. Rev. 106, 162–164 (1957) | Anderson, P. W. & Morel, P. Generalized Bardeen–Cooper–Schrieffer states and the proposed low-temperature phase of liquid He3. Phys. Rev. 123, 1911–1934 (1961) | Balian, R. & Werthamer, N. R. Superconductivity with pairs in a relative p wave. Phys. Rev. 131, 1553–1564 (1963) | Layzer, A. & Fay, D. Superconducting pairing tendency in nearly ferromagnetic systems. Int. J. Magn. 1, 135–141 (1971) | Osheroff, D. D., Richardson, R. C. & Lee, D. M. Evidence for a new phase of solid He3. Phys. Rev. Lett. 28, 885–888 (1972) | Osheroff, D. D., Gully, W. J., Richardson, R. C. & Lee, D. M. New magnetic phenomena in liquid He3 below 3 mK. Phys. Rev. Lett. 29, 920–923 (1972) | Borovik–Romanov, A. S., Bun’kov, Yu. M., Dmitriev, V. V. & Mukharskii, Yu. M. Observation of phase slippage during the flow of a superfluid spin current in 3He-B. JETP Lett. 45, 124–128 (1987) FURTHER READING Leggett, A. J. A theoretical description of the new phases of liquid 3He Rev. Mod. Phys. 47, 331–414 (1975)

‘NMR image’, most notably Raymond Damadian, who had reported in 1971 the ability to distinguish between normal tissue and tumours using magnetic resonance. On the announcement of the prize, he fervently disputed the decision of the Nobel committee. The unquestionable fact remains that, although the invention of MRI — with its marriage of magnetic fields — took scientists by surprise at its conception, in its more recent lifetime it has proved to be an invaluable tool for the medical world. Alison Stoddart, Associate Editor, Nature Materials ORIGINAL RESEARCH PAPERS Damadian, R. Tumor detection by nuclear magnetic resonance. Science 171, 1151–1153 (1971) | Lauterbur, P. C. Image formation by induced local interactions: examples employing nuclear magnetic resonance. Nature 242, 190–191 (1973) | Mansfield, P. & Grannell, P. K. NMR ‘diffraction’ in solids? J. Phys. C 6, L422–L426 (1973) | Mansfield, P., Garroway, A. N. & Grannell, P. K. Image formation in NMR by a selective irradiative process. J. Phys. C 7, L457–L462 (1974) | Mansfield, P. Multi-planar imaging formation using NMR spin echoes. J. Phys. C 10, L55–L58 (1977)

NATURE MILESTONES | SPIN

nature

M I L E S TO N E 1 6

Solution for solution structures In the mid-1970s, major advances in solution nuclear magnetic resonance (NMR) spectroscopy set the stage for a revolutionary new application: solving the three-dimensional (3D) structures of proteins in the solution state. At the time, X-ray crystallography was already a well-established tool for the determination of protein crystal structures. Unlike crystallography, NMR does not require proteins to form diffracting crystals and this broadens the range of proteins that can be investigated. Furthermore, most proteins exist naturally in a solution state, or in contact with fluids, so knowledge of their properties in their native environment has physiological relevance. A unique strength of NMR is its ability to supplement molecular structures with information on dynamic processes, which may be influenced, for example, by ligand binding in solution: even before structure determination by NMR became feasible, the technique was used to obtain information about protein dynamics. In 1971, Adam Allerhand and colleagues demonstrated the existence of sub-nanosecond segmental motions in proteins, and by the middle of the decade, the groups of Brian Sykes, Robert J. P. Williams and Kurt Wüthrich presented evidence for lower-frequency motions in globular proteins. Two fundamental advances in the late 1970s set the scene for the development of NMR into a method for determining previously unknown protein structures (rather than refining incomplete structures). First, Richard Ernst, building on a breakthrough idea by Jean Jeener, demonstrated the principle of two-dimensional (2D) NMR spectroscopy. This technique, which also applies to other spectroscopies, allowed researchers to record not only chemical shifts (Milestone 10) but also the interactions between pairs of nuclear spins — it later won Ernst the 1991 Nobel Prize in Chemistry. Second, Wüthrich discovered that the nuclear Overhauser effect could be exploited in NMR experiments with proteins, allowing the mapping of networks of near-by atom pairs that are not connected through covalent bonds. Beginning in 1976, Wüthrich and Ernst joined forces, and with Kuniaki Nagayama and Anil Kumar they developed a number of 2D NMR experiments, which became the basis for solving protein structures. In 1982, Gerhard Wagner and Wüthrich published the sequence-specific assignments for a small protein, basic pancreatic trypsin inhibitor. Meanwhile, Werner Braun and Timothy Havel in the Wüthrich group were developing algorithms

NMR has become a powerful technique for protein structure determination. and software capable of calculating protein structures from NMR data. In 1985, Michael Williamson, Havel and Wüthrich reported the first solution-state protein structure — that of proteinase inhibitor IIA from bull seminal plasma. The results were met with disbelief. It was not until several structures solved initially using NMR were solved again using crystallography that the NMR technique was accepted. In 2002, Wüthrich was rewarded with the Nobel Prize in Chemistry. NMR has become a powerful technique for protein structure determination. Numerous advances made during the past two decades — including the development of three- and fourdimensional spectroscopy, isotope labelling methods, and increases in magnetic field strength — have raised the limit on the size of proteins that can be investigated. Currently, about 10% of structures being deposited in the Protein Data Bank are solved using NMR. Perhaps the most exciting frontier is the application of NMR to investigate protein dynamics, especially for large molecular machines, which will undoubtedly lead to new insights in biology. Allison Doerr, Associate Editor, Nature Methods ORIGINAL RESEARCH PAPERS Allerhand, A. et al. Conformation and segmental motion of native and denatured ribonuclease A in solution. Application of natural-abundance carbon-13 partially relaxed Fourier transform nuclear magnetic resonance. J. Am. Chem. Soc. 93, 544–546 (1971) | Wüthrich, K. & Wagner, G. NMR investigations of the dynamics of the aromatic amino acid residues in the basic pancreatic trypsin inhibitor. FEBS Lett. 50, 265–268 (1975) | Dobson, C. M., Moore, G. R. & Williams, R. J. P. Assignment of aromatic amino acid PMR resonances of horse ferricytochrome c. FEBS Lett. 51, 60–65 (1975) | Snyder, G. H., Rowan, R. & Sykes, B. D. Complete tyrosine assignments in the high-field proton nuclear magnetic resonance spectrum of bovine pancreatic trypsin inhibitor selectively reduced and carboxamidomethylated at cystine 14-38. Biochemistry 15, 2275–2283 (1976) | Aue, W. P., Bartholdi, E. & Ernst, R. R. Two-dimensional spectroscopy. Application to nuclear magnetic resonance. J. Chem. Phys. 64, 2229–2246 (1976) | Nagayama, K., Wüthrich, K., Bachmann, P. & Ernst, R.R. Two-dimensional J-resolved 1 H NMR spectroscopy for studies of biological macromolecules. Biochem. Biophys. Res. Commun. 78, 99–105 (1977) | Kumar, A., Ernst, R. R. & Wüthrich, K. A two-dimensional nuclear Overhauser enhancement (2D NOE) experiment for the elucidation of complete proton–proton cross-relaxation networks in biological macromolecules. Biochem. Biophys. Res. Commun. 95, 1–6 (1980) | Wagner, G. & Wüthrich, K. Sequential resonance assignments in protein 1H nuclear magnetic resonance spectra: basic pancreatic trypsin inhibitor. J. Mol. Biol. 155, 347–366 (1982) | Williamson, M. P., Havel, T. F. & Wüthrich, K. Solution conformation of proteinase inhibitor IIA from bull seminal plasma by 1H nuclear magnetic resonance and distance geometry. J. Mol. Biol. 182, 295–315 (1985)

MARCH 2008 | S15

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

M I L E S TO N E 1 7

Dilute for impact

Image courtesy of P. M. Koenraad, Eindhoven University of Technology.

Towards the end of the 1960s, scientists had begun exploring the technological potential of magnetism combined with semiconductor physics. Having succeeded in introducing small amounts of magnetic impurities into otherwise non-magnetic semiconductors, Robert Gałazka and colleagues presented, in 1978, remarkable data on II–VI compounds doped with manganese. In these ‘diluted magnetic semiconductors’ (DMSs), the lowconcentration defects (the manganese ions) did not compromise the quality of the material, meaning that its magnetooptical and magneto-transport properties could be probed. At the same time, pronounced magnetic properties could be observed — such as the spin splitting of electronic or impurity bands.

M I L E S TO N E 1 8

A giant leap for electronics In retrospect, it seems surprising that, although spin and charge are two of the most fundamental properties of electrons, the advantage that could be gained from combining them in a consumer device was only realized in the 1990s, when IBM introduced a new type of hard-disk drive that would revolutionize data storage. Crucial to this technological revolution was an earlier discovery made by the groups of Albert Fert and Peter Grünberg — for which the two won the 2007 Nobel Prize in Physics. In 1988, they had observed a large change in the electrical resistance of thin metal layers as a function of an external magnetic field. Although bulk magnetoresistive effects had been known for more than a century (discovered by William Thomson, later to become Lord Kelvin), they were usually only moderate. However, in the studies led by Fert and Grünberg, the effect was much more pronounced. From the outset, these thin magnetic multilayer structures

S16 | MARCH 2008

nature

A further breakthrough came in the early 1990s, with the advent of lowtemperature molecular-beam epitaxy. By growing semiconductor films under conditions that were far from thermal equilibrium, it became possible to introduce manganese impurities into III–V materials (which is more difficult under equilibrium conditions owing to the low solubility of manganese). Hideo Ohno and colleagues then demonstrated, in 1992, ferromagnetic order in the DMS (In,Mn)As — indium arsenide containing only 1.3% manganese — by measuring magneto-transport and, in particular, an anomalous Hall effect in the material. The work was followed up, in 1996, with proof of ferromagnetism in doped gallium arsenide — (Ga,Mn)As — at temperatures up to 110 K. As GaAs can be used in electronic devices that operate at room temperature, these studies established the basis for research into ‘technologically relevant’ DMSs. The 1990s also brought theoretical work by Tomasz Dietl, in collaboration with Ohno’s group, which explained the origin of ferromagnetism in (Ga,Mn)As using a model developed, by Clarence

showed magnetoresistance of up to 50%; the phenomenon was dubbed ‘giant’ magnetoresistance (GMR). GMR is based on the spin-dependent scattering of electrons travelling across metallic thin films. In its most basic realization, a GMR device consists of two thin magnetic metal films, separated by a non-magnetic metal. If the magnetic layers have a different magnetic orientation with respect to each other, the electrons scatter strongly in the trilayer structure and the electrical resistance is high. However, once the magnetic orientation of the magnetic layers is aligned using an external magnetic field, electrons with spins antiparallel to that direction scatter much less, and move more easily between the magnetic and non-magnetic layers — hence, the electrical resistance is low. This groundbreaking discovery quickly led to the use of GMR to miniaturize the recording heads of hard-disk drives. IBM had already, in 1991, developed a hard drive based on the smaller bulk magnetoresistance effect; in 1997, thanks largely to the efforts of Stuart Parkin and colleagues in the IBM laboratories, the first hard drives based on GMR were commercialized. More recently, a new magnetoresistive

A

BEMaGS F

Zener in 1950, for ferromagnetism in transition metals. According to this model, the magnetic order originates from the delocalized holes that mediate the interaction between localized magnetic moments. The importance of the work was twofold. First, the carrier-mediated magnetic order suggested the possibility of controlling the ferromagnetism using electric fields — which was soon demonstrated — and, beyond that, the development of efficient spintronics devices. Second, the Dietl model provided an effective recipe for calculating the Curie temperature of other zinc-blende and wurzite semiconductors, to advance the search for a room-temperature DMS. In particular, Dietl showed that DMSs based on zinc oxide (ZnO) or gallium nitride (GaN) could have Curie temperatures as high as 300 K. Investigations since then have indeed revealed room-temperature ferromagnetism in oxides and semiconductors that include ZnO or GaN. However, it is yet to be proved that the carrier-mediated mechanism proposed by Dietl is really at work

device has been incorporated into hard-disk drives — the magnetic tunnel junction. Magnetic tunnel junctions (introduced in 1975 by Michel Jullière) are similar in structure to the GMR trilayers, except that the metallic non-magnetic layer is replaced by an insulating layer. However, it was only in 1995, following advances in techniques for growing materials, that Jagadeesh Moodera and colleagues, and Terunobu Miyazaki and Nobuki Tezuka, were able to realize tunnel junctions with practical magnetoresistance. Further work by Parkin et al. and by Shinji Yuasa and colleagues, in 2004, proved that satisfactory room-temperature operation could be achieved when the barrier layer was made of magnesium oxide. The advance made in reducing the size of read and write heads has raised the hope that such spin-electronic effects might also be used for solid-state data storage. Indeed, the tunnel-junction structure is a useful template for such magnetoresistive random-access memory (MRAM) devices, as the two possible relative orientations of the magnetic layers could be interpreted as ‘bits’ in a storage device. Moreover, new concepts for devices are evolving continually (see also Milestone 20) — for example,

www.nature.com/milestones/spin

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

in these systems. The origin of this ferromagnetism — and its potential use in spintronics devices — is still a matter of controversy, yet the search for a carrier-mediated room-temperature DMS continues. Fabio Pulizzi, Associate Editor, Nature Materials ORIGINAL RESEARCH PAPERS Von Molnar, S. & Methfessel, S. Giant negative magnetoresistance in ferromagnetic Eu1–xGdxSe. J. Appl. Phys. 38, 959–964 (1967) | Gałązka, R. R. in Proceedings of the 14th International Conference on the Physics of Semiconductors (ed. Wilson, B. L. H.) 133 (Institute of Physics, Bristol, 1978) | Munekata, H. et al. Diluted magnetic III–V semiconductors. Phys. Rev. Lett. 63, 1849–1852 (1989) | Ohno, H., Munekata, H., Penny, T., von Molnár, S. & Chang, L. L. Magnetotransport properties of p-type (In,Mn)As diluted magnetic III–IV semiconductors. Phys. Rev. Lett. 68, 2664–2667 (1992) | Ohno, H. et al. (Ga,Mn)As: a new ferromagnetic semiconductor based on GaAs Appl. Phys. Lett. 69, 363–365 (1996) | Dietl, T., Ohno, H., Matsukura, F., Cibert, J. & Ferrand, D. Zener model description of ferromagnetism in zincblende magnetic semiconductors. Science 287, 1019–1022 (2000) | Ohno, H. et al. Electric-field control of ferromagnetism. Nature 408, 944–946 (2000) FURTHER READING Furdyna, J. K. & Kossut, J. (eds) Diluted Magnetic Semiconductors, Semiconductors and Semimetals Vol. 25 (Academic, London, 1988)

as proposed by Parkin, using the thin domain walls that separate the regions of different magnetic orientation to store data in a single nanostructure, such as a narrow wire. Joerg Heber, Senior Editor, Nature Materials ORIGINAL RESEARCH PAPERS Thomson, W. On the electro-dynamic qualities of metals: effects of magnetization on the electric conductivity of nickel and of iron. Proc. R. Soc. Lond. 8, 546–550 (1856–1857) | Jullière, M. Tunneling between ferromagnetic films. Phys. Lett. A 54, 225–226 (1975) | Baibich, M. N. et al. Giant magnetoresistance of (001)Fe/(001)Cr magnetic superlattices. Phys. Rev. Lett. 61, 2472–2475 (1988) | Binasch, G., Grünberg, P., Saurenbach, F. & Zinn, W. Enhanced magnetoresistance in layered magnetic structures with antiferromagnetic interlayer exchange. Phys. Rev. B 39, 4828–4830 (1989) | Miyazaki, T. & Tezuka, N. Giant magnetic tunneling effect in Fe/Al2O3/Fe junction. J. Magn. Magn. Mater. 139, L231–L234 (1995) | Moodera, J. S., Kinder, L. R., Wong, T. M. & Meservey, R. Large magnetoresistance at room temperature in ferromagnetic thin film tunnel junctions. Phys. Rev. Lett. 74, 3273–3276 (1995) | Parkin, S. S. P. et al. Giant tunnelling magnetoresistance at room temperature with MgO (100) tunnel barriers. Nature Mater. 3, 862–867 (2004) | Yuasa, S., Nagahama, T., Fukushima, A., Suzuki, Y. & Ando, K. Giant roomtemperature magnetoresistance in single-crystal Fe/MgO/Fe magnetic tunnel junctions. Nature Mater. 3, 868–871 (2004) FURTHER READING Chappert, C., Fert, A. & Nguyen van Dau, F. The emergence of spin electronics in data storage. Nature Mater. 6, 813–823 (2007)

NATURE MILESTONES | SPIN

nature

A

BEMaGS F

M I L E S TO N E 1 9

Read my mind

By the late 1980s, magnetic resonance imaging (MRI; Milestone 15) had become a standard technique in hospitals and laboratories for the anatomical imaging of various tissues, from muscle to brain. However, MRI was only capable of revealing static structure and physiochemical information, but not actual function. The options for functional imaging were generally cumbersome, with low spatial resolution, and could require the injection of radioactive tracers into the bloodstream, as in positron emission tomography (PET), meaning that individual subjects could be scanned only infrequently. In 1990, however, Seiji Ogawa and colleagues published a series of breakthroughs that transformed MRI into a non-invasive and relatively inexpensive means of revealing physiological activity in the brain, sparking a revolution in the study of brain and behaviour. Ogawa et al. exploited two physiological phenomena that stemmed from observations made years earlier. First, in 1890, Charles Roy and Charles Sherrington had suggested that metabolic activity in the brain could be linked to vascular changes that would refresh blood supply. Later work established a more refined view: vasculature responds in an exquisitely localized fashion to bring oxygenated blood to areas of increased neural activity. This phenomenon was already being exploited in PET and other techniques. Second, Linus Pauling and Charles Coryell had reported, in 1936, that haemoglobin — the metalloprotein in red blood cells that acts as a major transporter of oxygen in humans and other species — has different magnetic properties in its oxygenated and deoxygenated forms. In three papers published in 1990, Ogawa and colleagues now showed how these two phenomena could be detected using MRI.

First, they demonstrated that changes in the level of deoxygenated haemoglobin in blood changed the proton signal from the water molecules surrounding the vessels — an effect called bloodoxygenation-level-dependent (BOLD) contrast. Because metabolic activity in the brain involves changes in the relative levels of oxyhaemoglobin and deoxyhaemoglobin, Ogawa reasoned that it should be possible to track changes in brain activity by measuring the BOLD contrast — and in his third paper of 1990, he demonstrated exactly this. Manipulation of the brain metabolism and, hence, the blood oxygen of an anaesthetized rat — by adjusting anaesthesia or the composition of inhaled gas, or by inducing hypoglycaemia — led to changes in the BOLD contrast throughout the brain. Two years after this ground-breaking proof of principle, three independent groups (including that of Ogawa) published, almost simultaneously, demonstrations of task-related changes in the BOLD contrast in the human brain — proving not only that this method could be translated from anaesthetized animals to awake humans, but also that it could reveal localized brain function evoked by specific stimuli, such as visual images. BOLD-contrast imaging — or functional MRI (fMRI) as the technique is now commonly known — quickly became a mainstay of cognitive neuroscience. It is an accessible option for measuring brain activity with relatively high spatial resolution — resolution that has improved with advances in MRI hardware and techniques, and analysis methods. From the detailed characterization of the function of human visual brain areas to the discovery of areas that are potentially involved in higher cognitive functions, such as face recognition, empathy and self-awareness, the possibilities revealed by fMRI seem endless. I-han Chou, Senior Editor, Nature ORIGINAL RESEARCH PAPERS Roy, C. S. & Sherrington, C. S. On the regulation of the blood supply of the brain. J. Physiol. 11, 85−108 (1890) | Pauling, L. & Coryell, C. D. The magnetic properties and structure of hemoglobin, oxyhemoglobin and carbonmonoxyhemoglobin. Proc. Natl Acad. Sci. USA 22, 210–216 (1936) | Ogawa, S., Lee, T. M., Kay, A. R. & Tank, D. W. Brain magnetic resonance imaging with contrast dependent on blood oxygenation. Proc. Natl Acad. Sci. USA 87, 9868–9872 (1990) | Ogawa, S., Lee, T. M., Nayak, A. & Glynn, P. Oxygenation-sensitive contrast in magnetic resonance image of rodent brain at high magnetic fields. Magn. Reson. Med. 14, 68–78 (1990) | Ogawa, S. & Lee, T. M. Magnetic resonance imaging of blood vessels at high fields: in vivo and in vitro measurements and image simulation. Magn. Reson. Med. 16, 9–18 (1990) | Bandettini, P. A., Wong, E. C., Hinks, R. S., Tikofsky, R. S. & Hyde, J. S. Time course EPI of human brain function during task activation. Magn. Reson. Med. 25, 390–397 (1992) | Kwong, K. K. et al. Dynamic magnetic resonance imaging of human brain activity during primary sensory stimulation. Proc. Natl Acad. Sci. USA 89, 5675–5679 (1992) | Ogawa, S. et al. Intrinsic signal changes accompanying sensory stimulation: functional brain mapping with magnetic resonance imaging. Proc. Natl Acad. Sci. USA 89, 5951–5955 (1992)

MARCH 2008 | S17

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

MILESTONES M I L E S TO N E 2 0

Information in a spin The success of the digital age, and the prosperity of the world economy, owes much to the development of the silicon chip and the idea that information can be represented, manipulated and transmitted in the form of electronic charge. The power and speed of computers has doubled approximately every two years for more than half a century, but within the next decade such exponential progress in silicon-based computing is expected to end — with potentially serious consequences. In 1990, Supriyo Datta and Biswajit Das proposed a device that uses electronic spin rather than charge to process information — an idea that has the potential to go beyond the fundamental limitations of silicon electronics. The practical use of spin in the context of electronics originated from the discovery of giant

magnetoresistance (Milestone 18). When an electric current flows from a magnetized material, such as a ferromagnetic metal contact, to a nonmagnetic material, the charges that carry this flow become polarized so that their spins point predominantly in one direction. If this spin current is injected from the non-magnetic material into a second ferromagnetic contact, it will experience a resistance that depends on the relative orientation of the magnetic field of the contact. Such behaviour forms the basis of operation of the ‘spin valve’ — a device that detects the orientation of a magnetic field, facilitating substantial improvements in the storage capacity of modern computer hard disks. Yet the spin valve is little more than a passive sensor that converts information stored magnetically into a charge-based signal that can

© Intel Corp

be read by conventional electronics. By contrast, the device described by Datta and Das acts as a switch, similar in concept to the transistors of a silicon chip. Known as a Datta–Das spin field-effect transistor or

M I L E S TO N E 2 1

Feel the force Magnetic resonance imaging (MRI; Milestone 15) and atomic force microscopy (AFM) have been two of the most widely used imaging techniques of recent decades. Building on previous work in nuclear magnetic resonance, MRI can provide three-dimensional (3D) images of systems containing unpaired nuclear spins (such as the protons in water molecules), and is now used extensively in many areas of fundamental research and medicine. AFM, by contrast, is a member of the large family of scanning probe microscopies and can provide detailed images of a wide range of different surfaces. In conventional MRI, the sample is placed in a magnetic field gradient, which causes all the unpaired spins in the sample to point in the same direction as the magnetic field. A

S18 | MARCH 2008

nature

resonant radiofrequency (RF) pulse is then used to ‘flip’ the spins so that they point in the opposite direction. Over time, the spins flip back again, emitting an RF signal that contains a wealth of information about the sample. The basic idea of AFM is to scan a flexible cantilever containing a sharp tip over a surface, and to measure the deflection of the cantilever caused by the forces acting between the

Signature of a single spin. Adapted from Nature 430, 329–332 (2004).

surface and the tip. Many different forces — including van der Waals, chemical and magnetic — can be exploited in AFM to produce images of the surface with atomic resolution. In 1991, John Sidles proposed combining these two techniques to make the magnetic resonance force microscope (MRFM). The first MRFM was demonstrated by Dan Rugar, Costantino Yannoni and Sidles a year later. The resolution that is possible in MRI is limited by the use of coils to detect the RF signal. The MRFM gets around this problem by placing the sample on an oscillating cantilever and detecting the magnetic resonance mechanically (by using a laser to measure changes in the resonant frequency of the cantilever), although a coil is still needed to produce the RF field. In their initial paper, Rugar et al. reported that they had imaged the spatial distribution of electron spins in their sample with a resolution of 19 μm in one dimension. In the years that followed, performance was improved by using lower temperatures, higher gradients and more sensitive

www.nature.com/milestones/spin

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

spin-FET, the device would exploit the fact that in certain semiconducting materials with strong spin–orbit coupling — that is, the coupling between the orbital angular momentum of its electronic states and the spin of the electrons that fill these states — an applied electric field can rotate the direction of an injected spin current. Although no one has yet succeeded in building a Datta–Das spin-FET, owing to several outstanding technical challenges, its proposal planted the idea that spin could be used in its own right as a means to carry and manipulate information — and gave birth to the new field of ‘spintronics’. Ed Gerstner, Senior Editor, Nature Physics ORIGINAL RESEARCH PAPER Datta, S. & Das, B. Electronic analog of the electro-optic modulator. Appl. Phys. Lett. 56, 665–667 (1990) FURTHER READING Žutić, I, Fabian, J. & Das Sarma, S. Spintronics: fundamentals and applications. Rev. Mod. Phys. 76, 323–410 (2004) | Awschalom, D. D. & Flatté, M. E. Challenges for semiconductor spintronics. Nature Phys. 3, 153–159 (2007)

cantilevers. By 1994, nuclear spins — which have much smaller magnetic moments than electrons — had been detected and the resolution had improved to 2.6 μm in one dimension; in 2007, it reached 90 nm. The ultimate goal in MRFM is to detect a single nuclear spin, but Rugar and colleagues passed a major milestone in 2004 when they detected the spin of a single electron. Peter Rodgers, Chief Editor, Nature Nanotechnology ORIGINAL RESEARCH PAPERS Sidles, J. A. Noninductive detection of single-proton magnetic resonance. Appl. Phys. Lett. 58, 2854–2856 (1991) | Rugar, D., Yannoni, C. S. & Sidles, J. A. Mechanical detection of magnetic resonance. Nature 360, 563–566 (1992) | Rugar, D. et al. Force detection of nuclear magnetic resonance. Science 264, 1560–1563 (1994) | Stowe, T. D. et al. Attonewton force detection using ultrathin silicon cantilevers. Appl. Phys. Lett. 71, 288–290 (1997) | Rugar, D., Budakian, R., Mamin, H. J. & Chui, B. W. Single spin detection by magnetic resonance force microscopy. Nature 430, 329–332 (2004) | Mamin, H. J., Poggio, M., Degen, C. L. & Rugar, D. Nuclear magnetic resonance imaging with 90-nm resolution. Nature Nanotech. 2, 301–306 (2007)

NATURE MILESTONES | SPIN

nature

A

BEMaGS F

M I L E S TO N E 2 2

The difficult middle ground Much of physics lends itself to a convenient division between big and small: the macroscopic classical world and the microscopic world of the quantum. Yet the transition between these spheres of influence is far from clear-cut, and increasingly we are seeing examples of phenomena that are manifestly quantum mechanical encroaching into territory once believed to be the exclusive purview of classical physics. Spin is no exception. The year 1996 saw the publication, by Jonathan Friedman et al. and Luc Thomas et al., of what many consider to be the definitive proof that mesoscopic spin systems (that is, systems with sizes or numbers of component spins that place them firmly in the conceptually difficult middle ground) can undergo quantum-mechanical tunnelling of their total magnetization. These experiments revealed transitions between bulk magnetic states that were not driven by thermal fluctuations, as one would expect classically; instead, they corresponded unambiguously to quantum-mechanical tunnelling events between different collective spin states of the whole system that have been brought into resonance by an applied magnetic field. The idea that macroscopic (or rather mesoscopic) quantum phenomena might be realized in magnetic spin systems has a long and convoluted history, with hints of such behaviour stretching back at least to work by Bernard Barbara and colleagues in 1972. However, it was not until the 1980s that such ideas were placed on a firm theoretical footing. Yet more experimental results suggestive of mesoscopic quantum behaviour were soon forthcoming; however, the response of the physics community remained largely ambivalent, even sceptical, about the central claims. A key problem with the early experiments was that any tunnelling effects in the thin films and quantum dots utilized in these studies could be readily masked by other factors — such as a distribution of energy barriers between magnetic states — rendering any ‘proof’ of mesoscopic quantum behaviour ambiguous. What was needed was an experimentally ‘clean’ system: either the ability to probe experimentally the magnetization properties of a single mesoscopic particle or the availability of an ensemble of strictly identical particles. In the meantime, chemist Roberta Sessoli and colleagues were investigating small molecular clusters containing several manganese ions that were magnetically coupled to give a high-spin

The molecular magnet Mn12. Image courtesy of Jens Kortus, Tunna Baruah and Mark Pederson.

(S = 10) ground state. The motivation for their 1993 work was to develop, using the powerful ‘bottom-up’ methods of chemical synthesis, welldefined molecular materials with potentially useful magnetic properties. Yet these chemists also had in their hands, or rather in their testtubes, the very thing that the physicists needed to realize convincingly mesoscopic quantum effects: ensembles of identical magnetic molecules. By 1995, Miguel Novak and Sessoli, and Barbara and colleagues, had the first evidence of resonances between the high-spin states of these molecular magnets; and, barely a year later, came the smoking-gun demonstration of mesoscopic quantum tunnelling. Now the pursuit of exciting new physics could begin... Karl Ziemelis, Chief Editor Physical Sciences, Nature ORIGINAL RESEARCH PAPERS Barbara, B., Fillion, G., Gignoux, D. & Lemaire, R. Magnetic aftereffect associated with narrow domain walls in some rare earth based intermetallic compounds. Solid State Commun. 10, 1149–1151 (1972) | Enz, M. & Schilling, R. Spin tunneling in the semiclassical limit. J. Phys. C 19, 1765–1770 (1986) | van Hemmen, J. L. & Suto, A. Tunneling of quantum spins. Europhys. Lett. 1, 481–490 (1986) | Chudnovsky, E. M. & Gunther, L. Quantum tunneling of magnetization in small ferromagnetic particles. Phys. Rev. Lett. 60, 661–664 (1988) | Sessoli, R., Gatteschi, D., Caneschi, A. & Novak, M. A. Magnetic bistability in a metal-ion cluster. Nature 365, 141–143 (1993) | Barbara, B. et al. Mesoscopic quantum tunneling of the magnetization. J. Magn. Magn. Mater. 140, 1825–1828 (1995) | Novak, M. A. & Sessoli, R. in Quantum Tunneling of Magnetization — QTM ‘94. NATO ASI Series E: Applied Sciences Vol. 301 (eds Gunther, L. & Barbara, B.) 171–188 (Kluwer Academic, Dordrecht, 1995) | Friedman, J. R., Sarachik, M. P., Tejada, J. & Ziolo, R. Macroscopic measurement of resonant magnetization tunneling in high-spin molecules. Phys. Rev. Lett. 76, 3830–3833 (1996) | Thomas, L. et al. Macroscopic quantum tunnelling of magnetization in a single crystal of nanomagnets. Nature 383, 145–147 (1996) FURTHER READING Stamp, P. C. E., Chudnovsky, E. M. & Barbara, B. Quantum tunneling of magnetization in solids. Int. J. Mod. Phys. B 6, 1355–1473 (1992) | Barbara, B. & Gunther, L. Magnets, molecules and quantum mechanics. Phys. World. 12 (3), 35–39 (1999)

MARCH 2008 | S19

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

MILESTONES M I L E S TO N E 2 3

The rise of semiconductor spintronics 1999 from groups led by Laurens Molenkamp and Awschalom, both of which used a light-emitting diode to demonstrate spin injection into a gallium-arsenide layer, a wide range of work has established spin injection into semiconductors. Spin polarization can also be created and manipulated without ferromagnetic contacts or external magnetic fields. In 2004, Yuichiro Kato et al. demonstrated, in a gallium-arsenide layer, that an electrical current generates polarized electron spins, along with a transverse spin current, leading to spin polarization at the edges of the conducting channel. The phenomenon is known as the spin Hall effect — predicted in 1971 by Mikhail Dyakonov and Vladimir Perel — and following the first observation by Kato, it has been seen in a wide range of other materials, including metallic structures.

Demarco | Dreamstime.com

Most of the electronic devices that we use in everyday life exploit the intrinsic electronic structure of semiconductors, by using electric fields to control charge transport and optical emission. Adding control of the spin degree of freedom of electrons is a natural route forwards, leading to the integration of current technology with spintronics devices. A series of breakthroughs has brought semiconductors on a par with metals (Milestone 18) in the race towards spintronics applications. Following early work on spin effects in semiconductors — in which the groups around Ionel Solomon in Paris and Boris Zakharchenya in Saint Petersburg were prime movers — it was David Awschalom and colleagues who put non-magnetic semiconductors on the map in spintronics research. In 1997, they discovered coherence times for spin states of up to 1 ns in a II–VI compound — proving that the spin state can live for long enough, in principle, to allow external manipulation, which is an essential requirement for any spintronics device. Two years later, the same group also measured

S20 | MARCH 2008

nature

Adding control of the spin degree of freedom of electrons is a natural route forwards, leading to the integration of current technology with spintronics devices.

coherence times exceeding 100 ns in III–V compounds, and showed that, under the influence of an electric field, charges can diffuse over distances greater than 100 μm without changing their spin state. To make full use of these developments, the spins must be prepared in a well-defined state at the outset; with an eye on practical application, this should ideally happen by electrical means. In 1976, Arkady Aronov and Gregory Pikus had shown that it should be possible theoretically to inject spins with a defined orientation from external electrodes into a semiconductor. Following on from initial experimental studies in

Fabio Pulizzi, Associate Editor, Nature Materials ORIGINAL RESEARCH PAPERS Lampel, G. Nuclear dynamic polarization by optical electronic saturation and optical pumping in semiconductors. Phys. Rev. Lett. 20, 491–493 (1968) | Dyakonov, M. I. & Perel V. I. Current-induced spin orientation of electrons in semiconductors. Phys. Lett. A 35, 459–460 (1971) | Aronov, A. G. & Pikus, G. E. Spin injection into semiconductors. Fiz. Tekh. Poluprovodn. 10, 1177–1179 (1976); Sov. Phys. Semicond. 10, 698–700 (1976) | Kikkawa, J. M., Awschalom, D. D., Smorchkova, I. P. & Samarth, N. Room-temperature spin memory in twodimensional electron gases. Science 277, 1284–1287 (1997) | Kikkawa, J. M. & Awschalom, D. D., Resonant spin amplification in n-type GaAs. Phys. Rev. Lett. 80, 4313 (1998) | Kikkawa, J. M. & Awschalom, D. D. Lateral drag of spin coherence in gallium arsenide. Nature 397, 139–141 (1999) | Fiederling, R. et al. Injection and detection of a spin-polarized current in a lightemitting diode. Nature 402, 787–790 (1999) | Ohno, Y. et al. Electrical spin injection in a ferromagnetic semiconductor heterostructure. Nature 402, 790–792 (1999) | Hanbicki, A. T., Jonker, B. T., Itskos, G., Kioseoglou, G. & Petrou, A. Efficient electrical spin injection from a magnetic metal/ tunnel barrier contact into a semiconductor. Appl. Phys. Lett. 80, 1240–1242 (2002) | Kato, Y. K., Myers, R. C., Gossard, A. C. & Awschalom, D. D. Observation of the spin hall effect in semiconductors. Science 306, 1910–1913 (2004) FURTHER READING Meier, F. & Zakharchenya, B. P. Optical Orientation, Modern Problems in Condensed Matter Science Series, Vol. 8 (NorthHolland, Amsterdam, 1984) | Awschalom, D. D. & Flatté, M. E. Nature Phys. Challenges for semiconductor spintronics. 3, 153–159 (2007)

www.nature.com/milestones/spin

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

COLLECTION • This article was first published in Nature 55, 347 (1897)

NATURE MILESTONES | SPIN

nature

MARCH 2008 | S21

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

COLLECTION

Image Formation by Induced Local Interactions: Examples Employing Nuclear Magnetic Resonance P. C. Lauterbur

In 1973, Paul Lauterbur described an imaging technique that removed the usual resolution limits due to the wavelength of the imaging field. He used two fields: one interacting with the object under investigation, the other restricting this interaction to a small region. Rotation of the fields relative to the object produces a series of one-dimensional projections of the interacting regions, from which two- or three-dimensional images of their spatial distribution can be reconstructed. Application of this technique as magnetic resonance imaging is now widespread. • First published in Nature 242, 190–191 (1973)

An image of an object may be defined as a graphical representation of the spatial distribution of one or more of its properties. Image formation usually requires that the object interact with a matter or radiation field characterized by a wavelength comparable to or smaller than the smallest features to be distinguished, so that the region of interaction may be restricted and a resolved image generated. This limitation on the wavelength of the field may be removed, and a new class of image generated, by taking advantage of induced local interactions. In the presence of a second field that restricts the interaction of the object with the first field to a limited region, the resolution becomes independent of wavelength, and is instead a function of the ratio of the normal width y

z

x

Figure 1 | Relationship between a three-dimensional object, its two-dimensional projection along the Y-axis, and four one-dimensional projections at 45° intervals in the XZ-plane. The arrows indicate the gradient directions.

S22 | MARCH 2008

nature

of the interaction to the shift produced by a gradient in the second field. Because the interaction may be regarded as a coupling of the two fields by the object, I propose that image formation by this technique be known as zeugmatography, from the Greek ζευγμα, "that which is used for joining". The nature of the technique may be clarified by describing two simple examples. Nuclear magnetic resonance (NMR) zeugmatography was performed with 60 MHz (5 m) radiation and a static magnetic field gradient corresponding, for proton resonance, to about 700 Hz cm–1. The test object consisted of two 1 mm inside diameter thin-walled glass capillaries of H2O attached to the inside wall of a 4.2 mm inside diameter glass tube of D2O. In the first experiment, both capillaries contained pure water. The proton resonance line width, in the absence of the transverse field gradient, was about 5 Hz. Assuming uniform signal strength across the region within the transmitter-receiver coil, the signal in the presence of a field gradient represents a one-dimensional projection of the H2O content of the object, integrated over planes perpendicular to the gradient direction, as a function of the gradient coordinate (FIG. 1). One method of constructing a two-dimensional projected image of the object, as represented by its H2O content, is to combine several projections, obtained by rotating the object about an axis perpendicular to the gradient direction (or, as in FIG. 1, rotating the gradient about the object), using one of the available methods for reconstruction of objects from

their projections1–5. FIGURE 2 was generated by an algorithm, similar to that of Gordon and Herman4, applied to four projections, spaced as in FIG. 1, so as to construct a 20 × 20 image matrix. The representation shown was produced by shading within contours interpolated between the matrix points, and clearly reveals the locations and dimensions of the two columns of H2O. In the second experiment, one capillary contained pure H2O, and the other contained a 0.19 mM solution of MnSO4 in H2O. At low radio-frequency power (about 0.2 mgauss) the two capillaries gave nearly identical images in the zeugmatogram (FIG. 3a). At a higher power level (about 1.6 mgauss), the pure water sample gave much more saturated signals than the sample whose spin-lattice relaxation time T1 had been shortened by the addition of the paramagnetic Mn2+ ions, and its zeugmatographic image vanished at the contour level used in FIG. 3b. The sample region with long T1 may be selectively emphasized (FIG. 3c) by constructing a difference zeugmatogram from those taken at different radio-frequency powers. Applications of this technique to the study of various inhomogeneous objects, not necessarily restricted in size to those commonly studied by magnetic resonance spectroscopy, may be anticipated. The experiments outlined above demonstrate the ability of the technique to generate pictures of the distributions of stable isotopes, such as H and D, within an object. In the second experiment, relative intensities in an image were made to depend upon relative nuclear relaxation times. The variations in water contents and proton relaxation times among biological tissues should permit the generation, with field gradients large compared to internal magnetic inhomogeneities, of useful zeugmatographic images from the rather sharp water resonances of organisms, selectively picturing the various soft structures and tissues. A possible

Figure 2 | Proton nuclear magnetic resonance zeugmatogram of the object described in the text, using four relative orientations of object and gradients as diagrammed in FIG. 1.

www.nature.com/milestones/spin

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

COLLECTION c

b

a

Figure 3 | Proton nuclear magnetic resonance zeugmatograms of an object containing regions with different relaxation times. a, Low power; b, high power; c, difference between a and b.

application of considerable interest at this time would be to the in vivo study of malignant tumours, which have been shown to give proton nuclear magnetic resonance signals with much longer water spin-lattice relaxation times than those in the corresponding normal tissues6. The basic zeugmatographic principle may be employed in many different ways, using a scanning technique, as described above, or transient methods. Variations on the experiment, to be described later, permit the generation of two- or three-dimensional images displaying chemical compositions, diffusion coefficients and other properties of objects measurable by spectroscopic techniques. Although applications employing nuclear magnetic resonance in liquid or liquid-like systems are simple and attractive because of the ease with which field gradients large enough to shift the narrow resonances

The way to NMR structures of proteins Kurt Wüthrich

In 1998 Kurt Wüthrich was awarded the Kyoto Prize in Advanced Technology for having “developed a method of determining the conformations of proteins, nucleic acids and other biomacromolecules in solutions or biomembranes, where they exhibit their function”1. Wüthrich has used nuclear magnetic resonance (NMR) techniques to study proteins and nucleic acids since 1967. In a series of four papers his group outlined a framework for NMR structure determination of proteins in 1982, and in 1984 the first de novo structure of a globular protein in solution was determined. The Wüthrich group went on to solve more than 60 protein structures in solution, including the Antennapedia homeodomain, the cyclophillin A−cyclosporin A complex, and the human and bovine prion proteins. What follows is a personal recollection by Kurt Wüthrich of how he and his associates arrived at the first view of a protein structure through the NMR eye. • First published in Nature Structural Biology 8, 923–925 (2001) doi:10.1038/nsb1101–923

In the 1950s, magnetic resonance spectroscopy evolved into a useful tool in chemistry. During the period of 1962−1967, my graduate and postdoctoral research, with Professor Silvio Fallab at the University of Basel and Professor Robert E. Connick at the University of California, Berkeley, focused on the use of electron paramagnetic resonance (EPR) and nuclear magnetic resonance (NMR) spin relaxation measurements to study metal complexes

NATURE MILESTONES | SPIN

nature

in solution. With this background, I joined the Biophysics Department of Dr. Robert G. Shulman at Bell Telephone Laboratories in Murray Hill, New Jersey, where a superconducting high resolution1 H NMR spectrometer operating at 220 MHz was available for ‘research on protein structure and function’. At that time I was aware of exactly 10 papers on NMR observations of proteins and nucleic acids, which had all been published during the period of

by many line widths may be generated, NMR zeugmatography of solids, electron spin resonance zeugmatography, and analogous experiments in other regions of the spectrum should also be possible. Zeugmatographic techniques should find many useful applications in studies of the internal structures, states, and compositions of microscopic objects. P. C Lauterbur is at the Department of Chemistry, State University of New York at Stony Brook, Stony Brook, New York 11790, USA Received October 30, 1972; revised January 8, 1973. 1. 2. 3. 4. 5. 6.

Bracewell, R. N., and Riddle, A. C., Astrophys. J., 150, 427 (1967). Vainshtein, B. K., Soviet Physics–Crystallography, 15, 781 (1971). Ramachandran, G. N., and Lakshminarayan, A. V., Proc. US Nat. Acad. Sci., 68, 2236 (1971). Gordon, R., and Herman, G. T., Comm. Assoc. Comput. Mach., 14, 759 (1971). Klug, A., and Crowther, R. A., Nature, 238, 435 (1972). Weisman, I. D., Bennett, L. H., Maxwell, Sr., L. R., Woods, M. W., and Burk, D., Science, 178, 1288 (1972).

1957−1965 (REF. 2). Prominent figures in the small community of spectroscopists that ventured into direct NMR observation of biological macromolecules were William D. Phillips3, Oleg Jardetzky4 and Robert G. Shulman5. Based on the observation of empirical correlations between protein unfolding and NMR spectra2–4, there was much enthusiasm about the future of NMR for de novo protein structure determination. Nonetheless, true to my background, I initially focused on the metal ion coordination in the active sites of hemoproteins and on the electronic structure of the heme groups6. At the time, Swiss scientists who landed a job at the famous Bell Telephone Laboratories were automatically considered prime candidates for academic positions back home. In 1969, I moved to the Eidgenössiche Technische Hochschule (ETH) in Zürich, where my startup package included an EPR and three NMR spectrometers — all the instrumentation that had been available to me at Bell Telephone Laboratories. I assembled a small research group, and, with time, I was promoted to Professor of Biophysics, which is also my present position at ETH. During the first years at Zürich, my research continued to focus primarily on the metal ions in the active centers of hemoproteins2–6, and I developed a mild infatuation with polypeptide chains only in connection with the discovery of aromatic ring flipping2. My primary research interest changed in 1975, MARCH 2008 | S23

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

COLLECTION a

b

Figure 1 | The first protein structure determined by NMR. a, All heavy-atom presentation of the NMR structure of the proteinase inhibitor IIA from bull seminal plasma (BUSI IIA)12. b, Superposition of the core region of residues 23–42 in the NMR structure of BUSI IIA (green) with the corresponding polypeptide segment in the X-ray crystal structure of the homologous porcine pancreatic secretory trypsin inhibitor (PSTI) (blue)13. The drawings were prepared from the atomic coordinates obtained in REFS 12,13.

when I took some time to write a monograph on the early years of biomacromolecular NMR2. These reflections on the state of the field turned out to have been the starting point for our subsequent work on de novo protein structure determination by NMR7. Four principal elements are combined in the NMR method for protein structure determination8, 9: (i) the nuclear Overhauser effect (NOE) as an experimentally accessible NMR parameter in proteins that can yield the information needed for de novo global fold determination of a polymer chain; (ii) sequence-specific assignment of the many hundred to several thousand NMR peaks from a protein; (iii) computational tools for the structural interpretation of the NMR data and the evaluation of the resulting molecular structures; and (iv) multidimensional NMR techniques for efficient data collection. During the period 1976−1980, my research group at the ETH Zurich had grown to more than 20 scientists, all of whom made great contributions toward the structure determination method. In particular, I worked with Regula M. Keller, Sidney L. Gordon and Gerhard Wagner on developing techniques to measure NOEs for the collection of conformational constraints in proteins, and with Martin Billeter, Werner Braun and Gerhard Wagner on the sequential resonance assignment strategy and algorithms for structure calculation from NMR data. This technology passed its initial tests when we obtained partial structure determinations of the bovine pancreatic trypsin inhibitor (BPTI), cytochrome b5 and the polypeptide hormone glucagon based on data collection with one-dimensional (1D) NMR experiments. S24 | MARCH 2008

nature

In a parallel project from 1976 to 1980, Richard R. Ernst (Nobel Prize in Chemistry, 1991), who also worked at the ETH Zürich, and I joined forces to develop two-dimensional (2D) NMR techniques for applications with biological macromolecules. Kuniaki Nagayama used 2D correlation spectroscopy for amino acid spin system identification in a protein, and Anil Kumar recorded the first 2D NOE spectra during the Christmas break 1979, when he was allotted two weeks of the precious measuring time on our highest-field spectrometer operating at a proton NMR frequency of 360 MHz (REF.10). By 1981 we routinely applied a group of four homonuclear 2D 1H NMR experiments, known under the acronyms COSY, SECSY, FOCSY and NOESY9, in the protein structure determination project. This resulted in complete resonance assignments of several small proteins in 1982 and 1983 (REF.11), and in the first de novo atomic resolution NMR structure determination of a globular protein, the bull seminal protease inhibitor (BUSI)12, by Timothy F. Havel and Michael P. Williamson in 1984. The completion of the first protein NMR structure brought new, unexpected challenges. When I presented the structure of BUSI (FIG. 1a)12 in some lectures in the spring of 1984, the reaction was one of disbelief, and because of the close coincidence (FIG 1b) with results from an independent crystallographic study of the homologous protein PSTI (porcine pancreatic secretory trypsin inhibitor)13 it was suggested that our structure must have been modeled after this crystal structure. In a discussion following a seminar in Munich on May 14, 1984, Robert Huber (Nobel Prize in Chemistry, 1988)

proposed that we settle the matter by independently solving a new protein structure by X-ray crystallography and by NMR. For this purpose, each one of us received an ample supply of the -amylase inhibitor tendamistat from scientists at the Hoechst company. Virtually identical three-dimensional structures of tendamistat were obtained in our laboratory by NMR in solution and in Robert Huber’s laboratory by X-ray diffraction in single crystals. The refined tendamistat structure was published in Journal of Molecular Biology as a 50-page report14, and the addendum to that paper clearly illustrated the impact of structure determination by NMR. I quote: “Editor’s Note: We have taken the step of publishing this paper with full supporting data since it is the first high resolution structure worked out in detail by 2D NMR. We therefore think that in this one instance everything should be published in full, but it does not set a precedent, since it is hoped that in the future, such supporting data can be deposited in a data bank, as is the practice in X-ray protein crystallography”. Considering that over 2,000 NMR structures have since been deposited in the Protein Data Bank, the Editor should be commended for his vision. At that time his kind comments were comforting in the context of our structure determinations of mammalian metallothioneins, which are a class of small, metalrich proteins that we studied in collaboration with Jeremias H.R. Kägi at the University of Zürich. In June 1985 I presented the structure of rabbit metallothionein at Yale University, where I learned about a manuscript accepted for publication in Proc. Nat. Acad. Sci. USA, which described a completely different metallothionein ‘NMR structure’, and at the University of Pittsburgh, where I was confronted with a rat metallothionein crystal structure that was again very different from our NMR structure. In both instances the structural differences were very clearcut, since they involved different polypeptide folds as well as different coordinating ligands to the metals. Metallothionein had been a tough challenge for all of us involved15, and my initial reaction was to spend two nights on the phone in my US motel room rechecking step by step the sequential resonance assignments with Gerhard Wagner in Zürich. All the assignments were, of course, correct, and I am afraid that Gerhard still bears a grudge against me for ever having doubted his spectral analysis. The crystal structure, which included erroneous chain tracing and

www.nature.com/milestones/spin

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

COLLECTION identification of 11 out of a total of 20 metalcoordinating amino acid residues, eventually appeared as a feature article in Science, whereas Nature rejected our NMR structure paper. In 1992, the crystal structure of rat metallothionein was redetermined, a correction of the first structure was published, and the correct crystal structure was found to be identical with the NMR structures of the rabbit, rat and human metallothioneins that we had solved from 1985 to 1990 (REF.16). Over the years a variety of applications of the NMR structure determination method have been pursued in my laboratory. The following three examples may convey some of the excitement that was thus generated in our professional life and further indicates the wide range of NMR applications in structural biology. Studies on the structural foundations of transcriptional regulation in higher organisms pursued in collaboration with Walter J. Gehring at the Biocenter of the University of Basel, Switzerland, yielded the NMR structure of the Antennapedia homeodomain17, and provided entirely novel insights into the role of hydration water in protein−DNA recognition18. An NMR structure determination of the human cyclophilin A−cyclosporin A complex was obtained in collaboration with two of my former graduate students, Hans Senn and Hans Widmer, who had subsequently joined the Sandoz company in Basel, Switzerland. This structure determination not only introduced me to the field of immune suppression but also had an immediate practical impact on cyclosporin research, since the structure of the bound drug molecule was found to be turned inside-out when compared with the structure of free cyclosporin A19. Barely 10 days after the bovine spongiform encephalopathy (BSE) crisis in Great Britain had broken into the open in March 1996, we completed the NMR structure determination of the murine prion protein20 in a collaboration with Rudi Glockshuber, who had joined our institute at the ETH Zürich as an Assistant Professor in 1994. The observation of a long flexible tail in prion proteins21 presents on the one hand a striking illustration of the unique power of NMR to characterize partially structured polypeptide chains in physiological milieus, and on the other hand indicates novel possible avenues for the transition of the benign cellular form of prion proteins to the disease-related scrapie form. With the introduction of TROSY (transverse relaxation-optimized spectroscopy)22, the molecular weight limit for solution NMR spectroscopy has extended to ~500 kDa, and we may soon be

NATURE MILESTONES | SPIN

nature

able to obtain information on the structure of the disease-related, aggregated form of the prion protein. Kurt Wüthrich is Professor of Biophysics at the Institute of Molecular Biology and Biophysics, ETH Zürich, CH-8093 Zürich, Switzerland, Fax: 41 1-633-1151

8. 9. 10. 11. 12.

Cecil H. and Ida M. Green Visiting Professor, Structural Biology at The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, CA 92037, USA, Fax: 1 858-784-8014.

13.

1.

16.

2. 3. 4. 5. 6. 7.

Kyoto Prizes and Inamori Grants 1998, 13 (The Inamori Foundation, Kyoto; 1999). Wüthrich, K. NMR in Biological Research: Peptides and Proteins (North Holland, Amsterdam; 1976). McDonald, C.C. & Phillips, W.D. J. Am. Chem. Soc. 89, 6332–6341 (1967) Jardetzky, O. & Roberts, G.C.K. NMR in Molecular Biology (Academic Press, New York; 1981). Shulman. R.G. et al. Science 165, 251–257 (1969). Wüthrich, K. Structure and Bonding 8, 53–121 (1970). Wüthrich, K. NMR in Structural Biology — A Collection of Papers by Kurt Wüthrich (World Scientific, Singapore; 1995).

14. 15.

17. 18. 19. 20. 21. 22.

Wüthrich, K., Wider, G., Wagner, G. & Braun, W. J. Mol. Biol. 155, 311−319 (1982). Wüthrich, K. NMR of Proteins and Nucleic Acids (Wiley, New York; 1986). Anil-Kumar, Ernst, R.R. & Wüthrich, K. Biochem. Biophys. Res. Comm. 95, 1−6 (1980). Wagner, G. & Wüthrich, K. J. Mol. Biol. 155, 347−366 (1982). Williamson, M.P., Havel, T.F. & Wüthrich, K. J. Mol. Biol. 182, 295−315 (1985). Bolognesi, M. et al. J. Mol. Biol. 162, 839−868 (1992). Kline, A.D., Braun, W. & Wüthrich, K. J. Mol. Biol. 204, 675−724 (1988). Braun, W. et al. J. Mol. Biol. 187, 125−129 (1986). Braun, W. et al. Proc. Natl. Acad. Sci. USA 89, 10124−10128 (1992). Qian, Y.Q. et al. Cell 59, 573−580 (1989). Billeter, M., Güntert, P., Luginbühl, P. & Wüthrich, K. Cell 85, 1057−1065 (1996). Wüthrich, K. et al. Science 254, 953−954 (1991). Riek, R. et al. Nature 382, 180−182 (1996). Riek, R., Hornemann, S., Wider, G., Glockshuber R. & Wüthrich, K. FEBS Lett. 413, 277−281 (1997). Pervushin, K., Riek, R., Wider, G. & Wüthrich, K. Proc. Natl. Acad. Sci. USA 94, 12366−12371 (1997).

Challenges for semiconductor spintronics David D. Awschalom and Michael E. Flatté

High-volume information-processing and communications devices are at present based on semiconductor devices, whereas information-storage devices rely on multilayers of magnetic metals and insulators. Switching within informationprocessing devices is performed by the controlled motion of small pools of charge, whereas in the magnetic storage devices information storage and retrieval is performed by reorienting magnetic domains (although charge motion is often used for the final stage of readout). Semiconductor spintronics offers a possible direction towards the development of hybrid devices that could perform all three of these operations, logic, communications and storage, within the same materials technology. By taking advantage of spin coherence it also may sidestep some limitations on information manipulation previously thought to be fundamental. This article focuses on advances towards these goals in the past decade, during which experimental progress has been extraordinary. • First published in Nature Physics 3, 153–159 (2007) doi:10.1038/nphys551

Semiconductor information-processing devices are among the most sophisticated, complex high-performance structures. As the construction cost of a new fabrication line approaches $3.5 billion, and 25% of its tools are obsolete and need replacement every three years, it is reasonable to question the appeal of alternate methods of information processing. Is it even realistic to imagine that an entirely distinct method of performing logical operations might be competitive? Metallic spintronic devices, such as hard disk read heads and magnetic

random access memory (MRAM) are one of the most successful technologies of the past decade, with scaling trends outdoing even CMOS — can semiconductor analogues provide enough new functionality to warrant interest? Electro-optic modulators are also a successful roomtemperature technology. What is the advantage of replacing these devices with magnetic analogues? Quantum computing seems to be progressing rapidly with atoms and superconductors — why use spins in semiconductors? MARCH 2008 | S25

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

1994

1996

1998

2000

2002

2004

Electrical detection of spin

Spin Hall effect

Current-induced spin polarization

Single spins in nanostructures

Giant planar Hall effect

Electrical modulation of coherence

Optical modulation of coherence

Gating of ferromagnetism

Electrical injection of spin

Long spin-coherence times

Single spins in diamond

Ferromagnetic GaMnAs

Electron, magnetic ion spin-coherence

COLLECTION

2006

Figure 1 | Timeline of key experimental discoveries since 1994 in semiconductor spintronics. The pace of progress continues to increase, and with the recent demonstration of electrical detection of spin, all essential elements for a semiconductor spintronic technology have been demonstrated under some experimental conditions.

Metallic spintronic devices1, originating from the discovery of giant magnetoresistance (GMR)2, 3 in 1988 and the subsequent development of the spin valve4, can be understood by assuming that any current of spins is carried by two ‘types’ of carriers, spin-up and spin-down. The two-channel picture of spin transport proposed by Mott5 explains the behaviour of magnetoresistive devices6–10, including GMR and tunnelling magnetoresistance11 (TMR), as well as spin injection into metals12. This theoretical treatment does not treat spin ‘coherence’, in which a persistent component of the spin can be maintained transverse to an applied magnetic field or magnetization. Even though respectable spin-coherence times had been measured via electron spin resonance13 in materials used for spin transport, and although spin coherence was verified in several of the situations above (such as spin injection) it was not central to the phenomena probed, which we refer to as ‘non-coherent spintronic devices’. A second generation of metallic spintronic phenomena requires a recognition of the importance of ‘spin coherence’, in which spin orientations transverse to the magnetization of magnetic domains, to internal effective magnetic fields, or to the orientations of other spin populations, are important in the overall dynamics or material properties of a system. Examples of coherent spintronic properties include current-induced precession of the magnetization of magnetic materials14, 15 and the spin Hall effect in aluminium16. These phenomena would be used in ‘coherent spintronic devices’. By any measure the progress in metallic spintronics has been exceptionally rapid, with revolutionary commercial devices available within ten years from the discovery of fundamental physical effects. More common timelines from discovery to the marketplace are 20–30 years. S26 | MARCH 2008

nature

Semiconductor spintronic device physics is progressing along a similar path to metallic spintronics and has achieved remarkable success in the past decade. A timeline of selected key experimental demonstrations since 1994 is shown in FIG. 1. Measurements of exceptionally long room-temperature spin-coherence times in non-magnetic semiconductors (three orders of magnitude longer than in non-magnetic metals)17,18 preceded the convincing demonstration of high-efficiency semiconductor spin-injection from a spinpolarized material19–25 and of coherent spin transport in non-magnetic semiconductors26. Initial theories of electronic spin transport, within the two-channel model, emphasized the importance of drift for the motion of spin-density packets in semiconductors relative to metals27–31, and have more recently tackled the fundamentals of diffusive spin transport in magnetic fields32. The optical accessibility of spin in semiconductors, which permitted early studies of spin dynamics33, has eased the observation of coherent phenomena in spin transport, including spin precession in an internal crystal magnetic field (without any applied magnetic field)34, electrically driven motion of domain walls35, and the spin Hall effect36–39 (before it was discovered in metals)40–44. Rapid progress towards room-temperature effects has been seen over the past couple of years, suggesting that room-temperature devices based on semiconductor spintronics may be soon possible. Here we are not treating the highly successful area of spins in semiconductors for solid-state quantum computation — for a review see REF. 45. Our focus here is on ensembles of spins, especially for nearroom-temperature operation. In addition to manipulating spin dynamics within non-magnetic materials, semiconductor spintronics offers materials possibilities very unlikely in metal systems.

Electrical control of the Curie temperature46 or coercive field47 of ferromagnetic semiconductors has been demonstrated. These phenomena require depleting the carrier concentration within the materials by a substantial fraction, a requirement that would be exceptionally challenging for ferromagnetic metals (with carrier concentrations three or four orders of magnitude higher). The highly coupled spin-orbit character of the magnetic dopants present in these systems provide additional possibilities for coherent spin manipulation48, using electric fields instead of magnetic ones to manipulate the spin degree of freedom. FIGURE 2 shows a metallic spintronic MRAM device from Freescale, and the demonstration of domain-wall motion due to current in the magnetic semiconductor GaMnAs. This effect might eventually permit the development of an MRAM technology based on magnetic semiconductors. Potential advantages of semiconductor spintronics Logic. Charge-based switching devices differentiate between a ‘0’ and a ‘1’ by the location of a small quantity of charge. For example, a field-effect transistor is ‘on’ if the channel is charged and thus metallic, and is ‘off ’ if the channel is not charged, and thus insulating. This switching device can be imagined as two wells separated by a barrier of adjustable height. This barrier must be high enough so that a charge placed in one well will stay there, but must be lowered to move the charge from one well to another. In separate analyses several authors have found a minimum switching energy to move from one configuration to another, Ebi = kTln2 ~ 23 meV (REF. 49). This limit is fundamental for charge-based switching devices that are, once the charge is placed in one of the wells, allowed to reach thermal equilibrium. However, current and projected semiconductor logic devices are still far from this limit, for the minimum switching energy is derived assuming adiabatic (slow) charge motion, and the desired switching speed of semiconductor logic devices is fast and continues to increase. The projected gateswitching energy in 2018 for low-standbypower CMOS and a 10 nm gate width is 15 eV, which is three orders of magnitude larger than the theoretical minimum50. Information encoded in the electron spin orientation, rather than the position of a pool of charge, is not subject to the above limitations on switching energies. A pool of spin-polarized electrons will maintain

www.nature.com/milestones/spin

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

COLLECTION a

b

Write line 1 I

Domain wall Isense

Magnetic tunnel junction

I Write line 2 ON for sense OFF for program

I

III

II

Domain wall Iref

–jpulse

Figure 2| Electrical reorientation of magnetic domains in metals and semiconductors. a, Schematic diagram of MRAM produced by Freescale. (Figures from http://www.freescale.com/files/ memory/doc/white_paper/MRAMWP.pdf). ________________________ Isense is the sensing current. Iref the reference current. The white arrows represent the free and fixed magnetization orientations. b, A domain wall in GaMnAs is electrically controlled at low temperatures using current pulses. The domain wall image in the initial position (top panel) and final position (bottom panel) are imaged using MOKE. Three regions (I, II, III) having different GaMnAs thicknesses were made to pattern coercivity and to confine the domain wall in region II. Jpulse is the current that moves the domain wall from one side of II to the other. This experimental demonstration (from REF. 35) sets the stage for similar device developments in semiconductor spintronics.

its spin orientation without the presence of any barrier between spin-up and spin-down states for times exceeding 10 ns at room temperature in common semiconductor materials. Changing the information from a ‘1’ to a ‘0’ would consist of applying a small magnetic field, which as described below can be a real magnetic field or an ‘effective’ field, to coherently rotate the spin by 180°. Thus none of the operations involving electron spin need to raise or lower a barrier to charge motion. If the operations are done coherently the minimum switching energy derived for charge-based information processing does not apply. Even when the motion of charge (such as in an electrical gate contact) is used to manipulate spin, the switching energy of a fast spin-based device can be much closer to the fundamental limit than a charge-based device51. Semiconductor spintronic devices thus would avoid the above thermodynamic limitation of a minimum switching energy by remaining out of equilibrium for long periods of time (of the order of the spin coherence times). Coherent semiconductor spintronic devices, by virtue of the exceptionally long room-temperature spin-coherence times

NATURE MILESTONES | SPIN

nature

discovered in ordinary semiconductor materials17, 18, could in principle perform multiple independent operations before the carriers reach thermal equilibrium. Interference of spin packets is one example, whereby two packets with spin polarizations oriented at 90° to each other generate a new packet with spins oriented at 45° to the two original packets (interference in a ring structure has been demonstrated52). Routing of spins via the spin Hall effect may be possible, as the sign of the spin Hall conductivity depends on the mobility53, and thus could be tuned either by an applied ordinary voltage or an applied spin-bias54. Speed is another essential concern for next-generation information-processing devices. In charge-based devices the speed is limited by the capacitance of the device and the drive current. As the semiconductor spintronic device is a coherent one, the speed limitations are given by typical precession frequencies of electron spins, and range from GHz to THz. For example, in order to coherently rotate a spin by 180° at a THz rate, an energy splitting of the order of 3 meV must be generated between the up and down spins. This energy splitting is an

order of magnitude lower than the thermal energy at room temperature. This is both a benefit and a danger for semiconductor spintronics. Local thermal equilibrium cannot be relied on to keep the information ‘safe’, even during a logical operation. Thus the system must be sufficiently isolated from the environment (stray magnetic fields in particular) to perform its operation robustly. Storage. Non-volatile storage and logic have traditionally been performed by independent technologies. The dominant technological implementation of non-volatile storage is through the magnetic orientation of media, such as on the platters in hard disks. MRAM, which uses magnetism to store information without moving parts, has become available commercially this summer from Freescale. However MRAM, like other magnetic nonvolatile storage, relies on magnetic metals and insulators — not semiconductors. If the fundamental physical phenomena underlying MRAM can be demonstrated at room temperature in semiconductors it may be possible to integrate non-volatile storage directly into logical processors. Spin transfer torque, the current-induced reorientation of the magnetization of a magnet55, 56 that is central to second-generation MRAM, has been demonstrated in the magnetic semiconductor GaMnAs, albeit at low temperature (FIG. 2). Even more promising is the possibility of integrated magnetic transistor devices, which blend non-volatility or reprogrammability with the central transistor property of gain. These magnetic transistor devices, such as the unipolar spin transistor57 or the magnetic bipolar transistor58, have yet to be demonstrated. Long-lived polarizations are also possible within the nuclear system. Substantial effective nuclear magnetic fields (>1 T) can be generated via the Overhauser effect59, 60 and persist for many minutes with only a small applied magnetic field (0.15 T), although at low temperature. These could be used for optoelectronic coupling, or for reprogrammable logic, and have been demonstrated both in ring interference52 and in all-optical NMR (REF.61). Communications. Isolation of one optical component of a communications system from the other optical components is achieved through one-way transmission of light. This ‘non-reciprocal’ transmission requires a magnetic component, otherwise the time-reversal invariance of the apparatus will guarantee that the transmission properties for light travelling in one direction will be the same as the MARCH 2008 | S27

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

A

BEMaGS

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

F

COLLECTION a

b 2.086

20

B

2.084

M

2.083

10

2.082

295 K

2.081

R (Ω)

Polarization (%)

T = 4.2 K

2.085

40 K

0

2.080

I

2.079 2.078 0.032

Fe

–10

n

F1 N

I

+ I V I –

F1 N F2

F2

+ V –

0.031

QW p

0.030

–20 –2

–4

0

2

4

B (T)

S28 | MARCH 2008

nature

–40

–20

0

20

40

60

B (mT)

Figure 3| Spin transport through non-magnetic semiconductors and metals. a, Electroluminescence polarization in an Fe/Al0.1Ga0.9As spin-LED. A schematic of the device is shown in the lower inset. Spin-polarized electrons injected from Fe recombine with holes in the quantum well (QW). The magnetic field (B) rotates the magnetization (M) of the Fe film out of the plane (upper inset). Reprinted with permission from REF. 23. Copyright (2005) by the American Physical Society. b, Spin-valve effect (top curve, and left cartoon) and non-local electrical detection of spin injection (bottom

properties for light travelling in the other direction. Thus magneto-optical elements (Faraday rotators) play a key role in the laser systems central to current communications. Although the magnetic orientation of such an element can be changed with an applied magnetic field, the material properties themselves are inflexible. Metals and insulators cannot be sufficiently changed by applied electric fields to provide markedly different magnetic properties. Magnetic semiconductors, however, can have their Curie temperatures significantly changed46 as well as their magnetic easy axes by an electric field47. Electrically gatable Faraday rotators would provide a seamless coupling between linear optics and electrical signals, potentially with less optical loss than typical for electro-optic modulators working via the quantum-confined Stark effect. High-speed optical switches based on spin have also been suggested and demonstrated at room temperature; here, the very fast spin lifetimes in some materials, or the tunable nature of the spin lifetime with electric fields, provides advantages over charge-based switches. Spin-based lasing may provide ways to efficiently modulate high-power semiconductor lasers, as the carrier density in an active region would not need to be changed,

0.029 –60

curve, and right cartoon) in a lateral metallic spin-valve device75. R is resistance. In the ordinary spin-valve measurement, the current I flows from ferromagnetic contact F1 through the normal metal N to ferromagnetic contact F2, and the voltage is measured between F1 and F2. In the non-local measurement, the current flows from F1 into the normal metal N, and the voltage is measured between the detection contact F2 and a reference contact on the normal metal. Reprinted with permission from REF. 75. Copyright (2003) by the American Physical Society.

the intensity of the emitted coherent light could be modulated by controlling the degree of spin polarization of the current injected into the active region. Quantum computation. Spin-based solidstate approaches for quantum computation provide the potential for fixing isolated quantum degrees of freedom in space, by embedding quantum dots or ions within a solid matrix, and then addressing those degrees of freedom with small electrical contacts. Progress in understanding the coupling between spin degrees of freedom and electrical fields48, 62–64, principally through the spin-orbit interaction but also through nanomagnetic fabrication65, avoid some of the problems with producing highly localized a.c. magnetic fields. Extremely long room-temperature spin-coherence times, such as 350 μs for nitrogen vacancy centres in diamond66, and short gate times67, provide ratios of coherence times to gate times that exceed 103. Through the coupling of the spin degree of freedom of an electron to optical fields there is a clear method of coupling photons into and out of a spin-based quantum computer68, and those photons can have wavelengths compatible with current communications.

Multifunctionality. As mobile phones and other gadgets become smaller and more powerful, one could consider whether all the elements of modern information manipulation (logic, storage and communication) could be combined and performed on a single chip. Some of the applications of spintronics described above are in the area of low-power electronics, so such a multifunctional chip might make possible complex but inexpensive submicrometre remote sensors. Challenges and advances in semiconducor spintronics Logic. Improving or optimizing a spintronic device requires attention to very different problems than for charge-based devices. Optimization of charge transport usually means efficient transfer of charge current from one region to another, or conductivities that are very sensitive to gate voltages. Thus in some regions large doping densities, and in other regions large electric fields, are useful for shuttling the charges efficiently around a chip. Designers of spin devices have to worry about the loss of spin polarization or spin coherence, wherein a carrier that is spin polarized in one direction effectively turns into another ‘charge’. Two of the accelerants for spin decoherence are

www.nature.com/milestones/spin

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

COLLECTION large doping densities for semiconductors and large electric fields33, 69, 70. Both for large doping densities and large electric fields a spin-polarized carrier will experience a large, randomly oriented, internal effective magnetic field. These accelerants are now significant barriers towards achieving nearly 100% efficient spin injection from a ferromagnetic metal into a semiconductor and similarly efficient spin detection. The initial challenge for spin injection, however, was quite different, and was traced to the substantial difference in conductivity between most ferromagnets and semiconductors71. This difference made it difficult for the ferromagnet to drive a large enough chemical potential difference between spin-up and spin-down in the semiconductor. Inserting a spin-dependent interface resistance from a Schottky barrier or tunnel barrier resolves this problem72–74, and the effect of spin-density drift in the semiconductor also helps29, 30. Spin injection from a ferromagnetic metal into a semiconductor has been demonstrated experimentally with this approach using the spin-dependent resistances of Schottky barriers21, 23, and oxide insulators22. The roomtemperature efficiency of spin injection now exceeds 70% (REF. 24), and further progress is expected as materials and interface physics improves. Engineering the device doping and field profile region will therefore be required in the design of optimally performing semiconductor spintronic chips. Shown in FIG. 3 are results for optically detected spin injection into a semiconductor through a Schottky barrier (reported in REF. 23) and for electrically detected spin injection into a metal75. A successful non-local electrically detected spin-injection experiment sensitive to precession has also been performed for a semiconductor76. Spin injection and detection can be considered the ‘input’ and ‘read-out’ stages of a logic device within which the spin is manipulated by external or internal magnetic fields or by spin-selective scattering. It has been demonstrated that the internal effective magnetic fields in semiconductors with spin-orbit interactions can be used to reorient spins and also to drive magnetic resonance34. These effective internal magnetic fields can be manipulated with applied external electric fields77, 78, which implies new gating mechanisms for spin-based transistors79. Furthermore the separation of spins can be achieved through the recently demonstrated spin Hall effect, first seen in semiconductors40–44, later in metals16, and most recently at room temperature43.

NATURE MILESTONES | SPIN

nature

Control of the spin Hall effect via control of the material mobility may be used to change spin currents in magnitude or even direction54, using a controllable spin Hall effect to route spins for logic. Finally, it might be possible to do away with magnetic materials entirely due to the achievement of spontaneous spin polarization at room temperature in a non-magnetic semiconductor80, 81. Storage. Many of the ferromagnetic semiconductor materials have extremely high carrier-doping levels, and controlling the interfaces of these materials is a great challenge. If novel storage devices based on ferromagnetic semiconductors are to be attempted, then achieving ferromagnetism in lower-doped semiconductor materials will be highly desirable. There are some indications that this might occur naturally at the edges of ferromagnetic materials, as the carriers are depleted from the region but magnetism remains82. Only very recently has there been a report of a p–n diode made with a ferromagnetic material83, 84 — previous attempts led to poor diodes because the doping level in the intrinsic, or depletion, region was too high to support a significant voltage. Another recent achievement was the demonstration of exchange biasing in magnetic semiconductors85. A central element of metallic MRAM, exchange biasing will be a key element of semiconductor spintronics storage technology. Communications. Optics and ferromagnetism has turned out to be a dirty partnership so far in GaMnAs. The optical lifetimes are so short, unlike for non-magnetic semiconductors, that it was some time before they could be measured86. As the desired magneto-optical devices typically require substantial Faraday rotation without significant optical losses, magnetic semiconductors have not been successful at dislodging magnetic insulators from this niche. Experiments on CdMnTe and CdMnHgTe optical isolators, however, suggest competitive values to yttrium iron garnet for the optical rotation relative to optical loss87, 88 in a material that can be monolithically integrated on a semiconductor substrate. A semiconductor waveguide with an integrated ferromagnetic metal clad has also shown good performance as an optical isolator89. As the materials become cleaner and more controlled the magneto-optical properties should improve further. It has been discovered that much cleaner GaMnAs could be achieved through very

long low-temperature post-growth annealing. At the same time the optical properties of very lightly doped GaMnAs seem quite good, even though the material itself is not doped sufficiently to become ferromagnetic. New discoveries of ferromagnetic semiconductors suggest there should be materials with better optical properties, such as ZnCrTe. Whether this material is a carrier-mediated ferromagnet or not is not clear yet, although it is dopable and the magnetism has a large influence on the optical properties. Quantum computing. The achievement of large-scale quantum-information processing in any physical system will be a tremendous success. Recent experimental advances in semiconductor spintronic quantum computing include the demonstration of long T1 and T2 times in semiconductor quantum dots (albeit at low temperatures45), the demonstration of coherent single-spin manipulation in diamond, and numerous examples of gate operations performed on ensembles of spins90–92, but expected to be extended to single-spin manipulation in quantum dots or embedded ions in the near future. Surprises and (very) open questions Do we need magnetic materials for a functional semiconductor spintronic technology? Current-induced spin polarization and the spin Hall effect have both been demonstrated at room temperature43. If the current-induced spin polarization can be made large enough this effect could replace spin injection from a ferromagnetic metal as the central method of injecting spins into non-magnetic semiconductors at room temperature. Similarly a spin-Hall-effect detector could replace the need for a ferromagnetic detector contact to electrically measure spin polarization at room temperature. Manipulation of the spins could be done using internal effective magnetic fields34,93 or perhaps using the spin Hall effect to directly drive resonance94.

Are there interesting semiconductor spintronic devices that do not require large T times? Although much of the initial interest in semiconductor spintronics was due to the exceptionally long spin-coherence times seen in ordinary semiconductor material, theoretical work indicating that T1 times can be tuned by orders of magnitude at room temperature70 seems to provide a pathway to a new switching mechanism for a spin-field-effect transistor95, 96. MARCH 2008 | S29

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

COLLECTION Charge-based current gating (MOSFET)

Spin-based current gating (Spin-FET)

Gate

Gate Oxide

Oxide E OFF

OFF

Long T 1

E E

E

ON

Short T 1

E Channel

Drain

Source

Figure 4| Schematic structure of a MOSFET (left) and a spin-FET (right). MOSFETs work by raising and lowering a barrier to turn the current on or off. A magnetic semiconductor spin lifetime FET uses static spin-

Comparisons of the scaling properties of this approach to CMOS indicate that spin-based field-effect transistors could compete with end-of-roadmap CMOS for low-standbypower devices51. Elegant approaches to balancing precessional transport in lateral devices might permit efficient spin manipulation of ensembles, but in the diffusive regime97 rather than the ballistic regime79. An example of such a spin transistor device is shown in FIG. 4, and compared with a MOSFET. A MOSFET works by controlling the height of a barrier to electron flow, with the barrier height and width largely determined by the desired on–off current ratio and leakage current. The spin-based FET has static spin-dependent barriers (generated by a magnetic insulator or magnetic semiconductor), and works by controlling the spin lifetime in the base95. Application of an electric field generates an effective magnetic field98(a ‘Rashba field’99), which decoheres the spins. As the spin orientation of the source–channel barrier and channel–drain barrier are opposite to each other, current only flows if the spin lifetime in the channel is short. For the spinFET the inputs and outputs are incoherent, but the transistor mechanism is spintronic. It has been argued that the spin lifetime FET device shown in FIG. 4 has superior powerdissipation properties to end-of-roadmap CMOS51. The requirements for spin lifetimes

nature

E E

E Source

S30 | MARCH 2008

E

E

ON

E

Channel Magnetic semiconductor

Drain Magnetic semiconductor

dependent barriers and changes the spin character of the electrons in the channel to turn the current on or off. Reused with permission from REF. 51. Copyright 2006, American institute of Physics.

in the absence of a gate field and for spininjection efficiency, however, are very high for this device. Multi-lead devices with coherent inputs and outputs might provide even better performance. What is the connection between spincoherence times and optical coherence? Remarkable measurements of the lasing threshold dependence of a semiconductor microcavity on the T2 time of a spin in a quantum dot inside that cavity suggest that semiconductor spin-coherence could be closely tied to optical coherence in microcavity geometries100. This might provide efficient ways of transmitting information from single photons to single spins, or be used to generate a hybrid device for quantum-information processing with both optical and spin components. What might exotic materials provide for semiconductor spintronics? Individual defect centres in diamond have been coherently manipulated to demonstrate magnetic resonance90–92 at room temperature. Rapid advances in controlling the electrical and optical properties of this system may lead to future diamond-based technologies. Other wide-bandgap materials such as oxide semiconductors also seem promising for room-temperature spintronic devices.

More physics or more engineering? Perhaps the mechanism for driving the next new technology based on semiconductor spintronics is already known, and among the wide variety of interesting device physics effects produced from the past decade of intensive research. It may now be time for circuit designers and systems engineers to take a closer look at these demonstrated mechanisms to see how a semiconductor spintronics architecture should be assembled. Even if that is not yet possible, systems and circuit engineers may greatly help the field of semiconductor spintronics by identifying those areas that most need progress in order to achieve a commercially viable technology. David D. Awschalom is at the Center for Spintronics and Quantum Computation and Department of Physics, University of California, Santa Barbara, California 93106, USA Michael E. Flatté is at the Optical Science and Technology Center, Department of Physics and Astronomy, and Department of Electrical and Computer Engineering,University of Iowa, Iowa City, Iowa 52242, USA Correspondence to: emails: _____________ [email protected]; [email protected] _____________ 1.

2.

3.

Ziese, M. & Thornton, M. J. (eds) Spin Electronics (Lecture Notes in Physics series, Vol. 569, SpringerVerlag, Heidelberg, 2001). Baibich, M. N. et al. Giant magnetoresistance of (001)Fe/(001)Cr magnetic superlattices. Phys. Rev. Lett. 61, 2472–2475 (1988). Binasch, G., Grünberg, P., Saurenbach, F. & Zinn, W. Enhanced magnetoresistance in layered magnetic

www.nature.com/milestones/spin

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

COLLECTION

4. 5. 6.

7.

8.

9.

10.

11.

12. 13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28. 29.

30.

structures with antiferromagnetic interlayer exchange. Phys. Rev. B 39, 4828–4830 (1989). Dieny, B. Giant magnetoresistive in soft ferromagnetic multilayers. Phys. Rev. B 43, 1297–1300 (1991). Mott, N. F. The electrical conductivity of transition metals. Proc. R. Soc. Lond. A 153, 699–717 (1936). Camley, R. E. & Barna, J. Theory of giant magnetoresistance effects in magnetic layered structures with antiferromagnetic coupling. Phys. Rev. Lett. 63, 664–667 (1989). Barna, J., Fuss, A., Camley, R. E., Grünberg, P. & Zinn, W. Novel magnetoresistance effect in layered magnetic structures: Theory and experiment. Phys. Rev. B 42, 8110–8120 (1990). Barthélémy, A. & Fert, A. Theory of the magnetoresistance in magnetic multilayers: Analytical expressions from a semiclassical approach. Phys. Rev. B 43, 13124–13129 (1991). Valet, T. & Fert, A. Theory of the perpendicular magnetoresistance in magnetic multilayers. Phys. Rev. B 48, 7099–7113 (1993). Butler, W. H. et al. Conductance and giant magnetoconductance of Co|Cu|Co spin valves: Experiment and theory. Phys. Rev. B 56, 14574–14582 (1997). Moodera, J. S., Kinder, L. R., Wong, T. M. & Meservey, R. Large magnetoresistance at room temperature in ferromagnetic thin film tunnel junctions. Phys. Rev. Lett. 74, 3273–3276 (1995). Johnson, M. & Silsbee, R. H. Spin-injection experiment. Phys. Rev. B 37, 5326–5335 (1988). Schultz, S. & Latham, C. Observation of electron spin resonance in copper. Phys. Rev. Lett. 15, 148–151 (1965). Tsoi, M. et al. Excitation of a magnetic multilayer by an electric current. Phys. Rev. Lett. 80, 4281–4284 (1998). Katine, J. A., Albert, F. J., Buhrman, R. A., Myers, E. B. & Ralph, D. C. Current-driven magnetization reversal and spin-wave excitations in Co /Cu /Co pillars. Phys. Rev. Lett. 84, 3149–3152 (2000). Valenzuela, S. O. & Tinkham, M. Direct electronic measurement of the spin Hall effect. Nature 442, 176–179 (2006). Kikkawa, J. M., Smorchkova, I. P., Samarth, N. & Awschalom, D. D. Room-temperature spin memory in two-dimensional electron gases. Science 277, 1284–1287 (1997). Kikkawa, J. M. & Awschalom, D. D. Resonant spin amplification in n-type GaAs. Phys. Rev. Lett. 80, 4313–4316 (1998). Ohno, Y. et al. Electrical spin injection in a ferromagnetic semiconductor heterostructure. Nature 402, 790–792 (1999). Fiederling, R. et al. Injection and detection of a spinpolarized current in a light-emitting diode. Nature 402, 787–790 (1999). Hanbicki, A. T., Jonker, B. T., Itskos, G., Kioseoglou, G. & Petrou, A. Efficient electrical spin injection from a magnetic metal/tunnel barrier contact into a semiconductor. Appl. Phys. Lett. 80, 1240–1242 (2002). Motsnyi, V. F. et al. Electrical spin injection in a ferromagnet/tunnel barrier/semiconductor heterostructure. Appl. Phys. Lett. 81, 265–267 (2002). Adelmann, C., Lou, X., Strand, J., Palmstrøm, C. J. & Crowell, P. A. Spin injection and relaxation in ferromagnet-semiconductor heterostructures. Phys. Rev. B 71, 121301(R) (2005). Jiang, X. et al. Highly spin-polarized roomtemperature tunnel injector for semiconductor spintronics using MgO(100). Phys. Rev. Lett. 94, 056601 (2005).| Crooker, S. A. et al. Imaging spin transport in lateral ferromagnet/semiconductor structures. Science 309, 2191–2195 (2005). Kikkawa, J. M. & Awschalom, D. D. Lateral drag of spin coherence in gallium arsenide. Nature 397, 139–141 (1999). Aronov, A. G. & Pikus, G. E. Spin injection into semiconductors. Fiz. Tekh. Poluprovodn. 10, 1177–1179 (1976); Sov. Phys. Semicond. 10, 698–700 (1976). Flatté, M. E. & Byers, J. M. Spin diffusion in semiconductors. Phys. Rev. Lett. 84, 4220–4223 (2000). Yu, Z. G. & Flatté, M. E. Electric-field dependent spin diffusion and spin injection into semiconductors. Phys. Rev. B 66, 201202 (2002). Yu, Z. G. & Flatté, M. E. Spin diffusion and injection in semiconductor structures: Electric field effects. Phys. Rev. B 66, 235302 (2002).

NATURE MILESTONES | SPIN

nature

31. Awschalom, D. D., Samarth, N. & Loss, D. (eds). Semiconductor Spintronics and Quantum Computation (Springer, Heidelberg, 2002). 32. Qi, Y. & Zhang, S. Spin diffusion at finite electric and magnetic fields. Phys. Rev. B 67, 052407 (2003). 33. Meier, F. & Zachachrenya, B. P. Optical Orientation (Modern Problems in Condensed Matter Science series, Vol. 8, North-Holland, Amsterdam, 1984). 34. Kato, Y. K., Myers, R. C., Gossard, A. C. & Awschalom, D. D. Coherent spin manipulation without magnetic fields in strained semiconductors. Nature 427, 50–53 (2004). 35. Yamanouchi, M., Chiba, D., Matsukura, F. and Ohno, H. Current-induced domain-wall switching in a ferromagnetic semiconductor structure. Nature 428, 539–542 (2004). 36. D’yakonov, M. I. & Perel’, V. I. Current-induced spin orientation of electrons in semiconductors. Phys. Lett. A 35, 459–460 (1971). 37. Hirsch, J. E. Spin Hall effect. Phys. Rev. Lett. 83, 1834–1837 (1999). 38. Murakami, S., Nagaosa, N. & Zhang, S.-C. Dissipationless quantum spin current at room temperature. Science 301, 1348–1351 (2003). 39. Sinova, J. et al. Universal intrinsic spin Hall effect. Phys. Rev. Lett. 92, 126603 (2004). 40. Kato, Y. K., Myers, R. C., Gossard, A. C. & Awschalom, D. D. Observation of the spin Hall effect in semiconductors. Science 306, 1910–1913 (2004). 41. Sih, V. et al. Spatial imaging of the spin Hall effect and current-induced polarization in two-dimensional electron gases. Nature Phys. 1, 31–35 (2005). 42. Wunderlich, J., Kaestner, B., Sinova, J. & Jungwirth, T. Experimental observation of the spin-hall effect in a two-dimensional spin-orbit coupled semiconductor system. Phys. Rev. Lett. 94, 047204 (2005). 43. Stern, N. P. et al. Current-induced polarization and the spin Hall effect at room temperature. Phys. Rev. Lett. 97, 126603 (2006). 44. Sih, V. et al. Generating spin currents in semiconductors with the spin Hall effect. Phys. Rev. Lett. 97, 096605 (2006). 45. Hanson, R., Kouwenhoven, L. P., Petta, J. R., Tarucha, S. & Vandersypen, L. M. K. Spins in few-electron quantum dots. Preprint at _________ http://arxiv.org/condmat/0610433 ______ (2006). 46. Ohno, H. et al. Electric-field control of ferromagnetism. Nature 408, 944–946 (2000). 47. Chiba, D., Yamanouchi, M., Matsukura, F. & Ohno, H. Electrical manipulation of magnetization reversal in a ferromagnetic semiconductor. Science 301, 943–945 (2003). 48. Tang, J.-M., Levy, J. & Flatté, M. E. All-electrical control of single ion spins in a semiconductor. Phys. Rev. Lett. 97, 106803 (2006). 49. Landauer, R. Irreversibility and heat generation in the computing process. IBM J. Res. Dev. 5, 183–191 (1961). 50. International Technology Roadmap for Semiconductors (Semiconductor Industry Association, San Jose, California, USA, 2003); http://public.itrs.net. 51 Hall, K. C. & Flatté, M. E. Performance of a spin-based insulated gate field effect transistor. Appl. Phys. Lett. 88, 162503 (2006). 52. Kato, Y., Myers, R. C., Gossard, A. C. & Awschalom, D. D. Electron spin interferometry using a semiconductor ring structure. Appl. Phys. Lett. 86, 162107 (2005). 53. Engel, H.-A., Halperin, B. I. & Rashba, E. I. Theory of spin Hall conductivity in n-doped GaAs. Phys. Rev. Lett. 95, 166605 (2005). 54. Hankiewicz, E. M., Vignale, G. & Flatté, M. E. Spin-Hall effect in a [110] GaAs quantum well. Phys. Rev. Lett. 97, 266601 (2006). 55. Slonczewski, J. Current-driven excitation of magnetic multilayers. J. Magn. Magn. Mater. 159, L1–L7 (1996). 56. Berger, L. Emission of spin waves by a magnetic multilayer traversed by a current. Phys. Rev. B 54, 9353–9358 (1996). 57. Flatté, M. E. & Vignale, G. Unipolar spin diodes and transistors. Appl. Phys. Lett. 78, 1273–1275 (2001). 58. Flatté, M. E., Yu, Z. G., Johnston-Halperin, E. & Awschalom, D. D. Theory of semiconductor magnetic bipolar transistors. Appl. Phys. Lett. 82, 4740–4742 (2003). 59. Overhauser, A. W. Polarization of nuclei in metals. Phys. Rev. 92, 411–415 (1953). 60. Paget, D., Lampel, G., Sapoval, B. & Safarov, V. I. Low field electron-nuclear spin coupling in gallium arsenide under optical pumping conditions. Phys. Rev. B 15, 5780–5796 (1977).

61. Kikkawa, J. M. & Awschalom, D. D. All-optical magnetic resonance in semiconductors. Science 287, 473–476 (2000). 62. Loss, D. & DiVincenzo, D. P. Quantum computation with quantum dots. Phys. Rev. A 57, 120–126 (1998). 63. Kato, Y., Myers, R. C., Gossard, A. C., Levy, J. & Awschalom, D. D. Gigahertz electron spin manipulation using voltage-controlled g-tensor modulation. Science 299, 1201–1204 (2003). 64. Petta, J. R. et al. Coherent manipulation of coupled electron spins in semiconductor quantum dots. Science 309, 2180–2184 (2005). 65. Tokura, Y., van der Wiel, W. G., Obata, T. & Tarucha, S. Coherent single electron spin control in a slanting Zeeman field. Phys. Rev. Lett. 96, 047202 (2006). 66. Gaebel, T. et al. Room-temperature coherent coupling of single spins in diamond. Nature Phys. 2, 408–413 (2006). 67. Jelezko, F., Gaebel, T., Popa, I., Gruber, A. & Wrachtrup, J. Observation of coherent oscillations in a single electron spin. Phys. Rev. Lett. 92, 076401 (2004). 68. Leuenberger, M. N., Flatté, M. E. & Awschalom, D. D. Teleportation of electronic many-qubit states via single photons. Phys. Rev. Lett. 94, 107401 (2005). 69. D’yakonov, M. I. & Perel’, V. I. Spin relaxation of conduction electrons in noncentrosymmetric semiconductors. Sov. Phys. Solid State 13, 3023–3026 (1972). 70. Lau, W. H. & Flatté, M. E. Tunability of electron spin coherence in III-V quantum wells. J. Appl. Phys. 91, 8682–8684 (2002). 71. Schmidt, G., Ferrand, D., Molenkamp, L. W., Filip, A. T. & van Wees, B. J. Fundamental obstacle for electrical spin injection from a ferromagnetic metal into a diffusive semiconductor. Phys. Rev. B 62, R4790–R4793 (2000). 72. Rashba, E. I. Theory of electrical spin injection: Tunnel contacts as a solution of the conductivity mismatch problem. Phys. Rev. B 62, R16267–R16270 (2000). 73. Smith, D. L. & Silver, R. N. Electrical spin injection into semiconductors. Phys. Rev. B 64, 045323 (2001). 74. Fert, A. & Jaffrès, H. Conditions for efficient spin injection from a ferromagnetic metal into a semiconductor. Phys. Rev. B 64, 184420 (2001). 75. Jedema, F. J., Nijboer, M. S., Filip, A. T. & van Wees, B. J. Spin injection and spin accumulation in all-metal mesoscopic spin valves. Phys. Rev. B 67, 085319 (2003). 76. Lou, X. et al. Electrical detection of spin transport in lateral ferromagnet–semiconductor devices. Nature Phys. 3, 197–202 (2007). 77. Koga, T., Sekine, Y. & Nitta, J. Experimental realization of a ballistic spin interferometer based on the Rashba effect using a nanolithographically defined square loop array. Phys. Rev. B 74, 041302 (2006). 78. Bergsten, T., Kobayashi, T., Sekine, Y. & Nitta, J. Experimental demonstration of the time reversal Aharonov–Casher effect. Phys. Rev. Lett. 97, 196803 (2006). 79. Datta, S. & Das, B. Electronic analog of the electrooptic modulator. Appl. Phys. Lett. 56, 665–667 (1990). 80. Kato, Y., Myers, R. C., Gossard, A. C. & Awschalom, D. D. Current-induced spin polarization in strained semiconductors. Phys. Rev. Lett. 93, 176601 (2004). 81. Silov, A. Yu. et al. Current-induced spin polarization at a single heterojunction. Appl. Phys. Lett. 85, 5929–5931 (2004). 82. Rüster, C. et al. Very large magnetoresistance in lateral ferromagnetic (Ga,Mn)As wires with nanoconstrictions. Phys. Rev. Lett. 91, 216602 (2003). 83. Chen, P. et al. All-electrical measurement of spin injection in a magnetic p-n junction diode. Preprint at http://arxiv.org/abs/cond-mat/0608453 (2006). ˘ ´ I. & Sarma, S. D. Theory of spin84. Fabian, J., Zutic, polarized bipolar transport in magnetic p-n junctions. Phys. Rev. B 66, 165301 (2002). 85. Eid, K. F. et al. Exchange biasing of the ferromagnetic semiconductor Ga1–XMnXAs. Appl. Phys. Lett. 85, 1556–1558 (2004). 86. Beschoten, B. et al. Magnetic circular dichroism studies of carrier-induced ferromagnetism in Ga1– XMnXAs. Phys. Rev. Lett. 83, 3073–3076 (1999). 87. Zayets, V., Debnath, M. C. & Ando, K. Complete magneto-optical waveguide mode conversion in Cd1– XMnXTe waveguide on GaAs substrate. Appl. Phys. Lett. 84, 565–567 (2004).

MARCH 2008 | S31

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

COLLECTION 88. Onodera, K., Masumoto, T. & Kimura, M. 980 nm compact optical isolators using Cd1–XMnXHgyTe single crystals for high power pumping laser diodes. Electron. Lett. 30, 1954–1955 (1994). 89. Shimizu, H. & Nakano, Y. Fabrication and characterization of an InGaAsP/InP active waveguide optical isolator with 14.7 dB/mm TE mode nonreciprocal attenuation. IEEE J. Lightwave Tech. 24, 38–43 (2006). 90. Gruber, A., Dräbenstedt, A., Tietz, C., Fleury, L., Wrachtrup J. & von Borczyskowski, C. Scanning confocal optical microscopy and magnetic resonance on single defect centers. Science 276, 2012–2014 (1997). 91. Hanson, R., Gywat, O. & Awschalom, D. D. Roomtemperature manipulation and decoherence of a single spin in diamond. Phys. Rev. B 74, 161203(R) (2006). 92. Hanson, R., Mendoza, F. M., Epstein, R. J. & Awschalom, D. D. Polarization and readout of coupled single spins in diamond. Phys. Rev. Lett. 97, 087601 (2006). 93. Crooker, S. A. & Smith, D. L. Imaging spin flows in semiconductors subject to electric, magnetic, and strain fields. Phys. Rev. Lett. 94, 236601 (2005). 94. Duckheim, M. & Loss, D. Electric-dipole-induced spin resonance in disordered semiconductors. Phys. Rev. Lett. 2, 195–199 (2006).

˘ 95. Hall, K. C., Lau, W. H., Gündogdu, K., Flatté, M. E. & Boggess, T. F. Nonmagnetic semiconductor spin transistor. Appl. Phys. Lett. 83, 2937–2939 (2003). ` X., Ting, D. Z.-Y. & Chang, Y.-C. A resonant 96. Cartoixa, spin lifetime transistor. Appl. Phys. Lett. 83, 1462–1464 (2003). 97. Schliemann, J., Egues, J. C. & Loss, D. Nonballistic spin-field-effect transistor. Phys. Rev. Lett. 90, 146801 (2003). 98. Nitta, J., Akazaki, T., Takayanagi, H. & Enoki, T. Gate control of spin-orbit interaction in an inverted In0.53Ga0.47As/In0.52Al0.48As heterostructure. Phys. Rev. Lett. 78, 1335–1338 (1997). 99. Bychkov, Y. A. & Rashba, E. I. Oscillatory effects and the magnetic susceptibility of carriers in inversion layers. J. Phys. C 17, 6039–6045 (1984). 100. Ghosh, S. et al. Enhancement of spin coherence using Q-factor engineering in semiconductor microdisc lasers. Nature Mater. 5, 261–264 (2006).

Acknowledgements The authors would like to acknowledge the support of ONR and DARPA.

Competing interests statement The authors declare that they have no competing financial interests.

The emergence of spin electronics in data storage Claude Chappert, Albert Fert & Frédéric Nguyen Van Dau

Electrons have a charge and a spin, but until recently these were considered separately. In classical electronics, charges are moved by electric fields to transmit information and are stored in a capacitor to save it. In magnetic recording, magnetic fields have been used to read or write the information stored on the magnetization, which ‘measures’ the local orientation of spins in ferromagnets. The picture started to change in 1988, when the discovery of giant magnetoresistance opened the way to efficient control of charge transport through magnetization. The recent expansion of hard-disk recording owes much to this development. We are starting to see a new paradigm where magnetization dynamics and charge currents act on each other in nanostructured artificial materials. Ultimately, ‘spin currents’ could even replace charge currents for the transfer and treatment of information, allowing faster, low-energy operations: spin electronics is on its way. • First published in Nature Materials 6, 813 - 823 (2007) doi:10.1038/nmat2024

The interdependence between magnetization and charge transport is not a new story. For instance, anisotropic magnetoresistance (AMR), which links the value of the resistance to the respective orientation of magnetization and current, was first observed in 1856 by William Thomson, but its amplitude is weak (up to a few per cent variation in resistance on changing the relative orientation of magnetization and current). Nevertheless, the introduction by IBM in 1991 of a magnetoresistive read head based on AMR was a major technological step forward1. In hard disk drives (HDD), the head flies at constant height above the S32 | MARCH 2008

nature

magnetic domains that define the ‘bits’ in the recording medium, and senses the spatial variations of the stray magnetic field of these domains. The old ring-shaped magnetic head, invented in 1933 by Eduard Schuller (AEG) for tape recording, measured a magnetic flux, and so was approaching its sensitivity limit with the reduction of the head dimensions. In the IBM head (FIG. 1), the AMR sensor is combined with a ‘ring’ element still used for writing, and directly senses the magnetic field through its influence on the magnetization orientation in the head. The AMR head had a relative ‘magnetoresistance’ (hereafter referred to

as ΔR/R = (Rmax – Rmin)/Rmin) of the order of only 1%, but this was enough to increase the growth rate of HDD storage areal density from 25% per year, its value since nearly the introduction of HDD in 1957, up to 60% per year. Fundamentals of spin electronics The physics behind today’s fast expansion of spin electronics has also been known for a long time. A cornerstone is the ‘two currents’ conduction concept proposed by Mott2 and used by Fert and Campbell3, 4 to explain specific behaviours in the conductivity of the ferromagnetic metals Fe, Ni, Co and their alloys. In such ‘itinerant ferromagnets’ both the 4s and 3d electron bands contribute to the density of states at the Fermi level EF. Because of the strong exchange interaction favouring parallel orientation of electron spins, the ‘spin-up’ and ‘spin-down’ 3d bands are shifted in energy. This band splitting creates the imbalance between numbers nup and ndown of 3d electrons that is at the origin of the ferromagnetic moment (μ ≈–(nup – ndown) μB/atom, where μB is the Bohr magneton), whereas the conduction is dominated by the unsplit 4s band, the 4s electrons having a much higher mobility. However, the spinconserving s-to-d transitions are the main source of s-electron scattering. This has two chief consequences for transport: the spinimbalanced density of states for 3d electrons at EF results in strongly spin-dependent scattering probabilities (Fermi ‘Golden Rule’), and between two spin-flip scattering events an electron can undergo many scattering events that keep the same spin direction. Thus at the limit where spin-flip scattering events are negligible, conduction happens in parallel through two spin channels that have very different conductivities. Important length scales can be discussed in the diffusive transport model. The spindependent scattering probability results in very different mean free paths λup and λdown, or equivalently relaxation times τup and τdown. In usual thin metallic layers they scale from a few nanometres to a few tens of nanometres, with highly variable λdown/λup ratios: some impurities have strongly spin-dependent cross-sections, so that in Ni, for example, the ratio λdown/λup can reach 20 for Co impurities or decrease to 0.3 with Cr doping4. Another important length scale is the spin-conserving drift length projected along one direction, called the spin diffusion length Lsf (sf denotes spin flip). It is generally much larger than the mean free path5.

www.nature.com/milestones/spin

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

COLLECTION Inductive write element

GMR read sensor

Magnetic shield S1

Magnetic shield S2

W P2 t

N

S

S

B

N N

P1 S S

P1′

N N

S S

N N

S

Recording medium

Figure 1 | Magnetroresistance head for hard-disk recording. Schematic structure of the magnetoresistive head introduced by IBM for its hard disk drives in 1991. A magnetic sensor based on anisotropic magnetoresistance (left) is added to the inductive ‘ring-type’ head (right) still used for writing. The distances P1–P1′ and P1–P2 between the pole pieces of the magnetic shields S1 and S2 define respectively the ‘write’ and ‘read’ gaps, on which depends the minimum length B of the magnetic domains. W is the track width and t is the thickness of the recording medium. Note that in today’s hard disk recording, W and B are of the order of 100 nm and 30 nm respectively, but with a different arrangement of head and domains in ‘perpendicular recording’1.

Giant magnetoresistance and the spin-valve head The founding step of spin electronics, which triggered the discovery of the giant magnetoresistance6, 7, was actually to build magnetic multilayers with individual thicknesses comparable to the mean free paths, so that evidence could be seen for spindependent electron transport. The principle is schematized in FIG. 2 for the simplest case of a triple-layer film of two identical ferromagnetic layers F1 and F2 sandwiching a non-magnetic metal spacer layer M, when the current circulates ‘in plane’ (cf. FIG. 2b). We assume λFup >> λFdown, with λFup > tF > λFdown, for the thickness tF of the magnetic layer and tM << λM for the thickness tM of the spacer layer. When the two magnetic layers are magnetized parallel (P), the spin-up electrons can travel through the sandwich nearly unscattered, providing a conductivity shortcut and a low resistance. On the contrary, in the antiparallel (AP) case, both spin-up and spin-down electrons undergo collisions in one F layer or the other, giving rise to a high resistance. The relative magnetoresistance ΔR/R = (RAP – RP)/RP can reach 100% or more in multilayers with a high number of F/M periods. It was already 80% in the Fe/Cr multilayer of the original discovery6, hence the name of giant magnetoresistance (GMR). Elaborate theories must of course take into account other effects such as interfacial scattering and quantum confinement of the electrons in the layers8, 9.The GMR is an outstanding example of how structuring materials at the nanoscale can bring to

NATURE MILESTONES | SPIN

nature

light fundamental effects that provide new functionalities. And indeed the amplitude of the GMR immediately triggered intense research, soon achieving the definition of the spin-valve sensor10–12. In its simplest form, the spin valve is just a trilayer film of the kind displayed in FIG. 2a in which one layer (for example, F2) has its magnetization pinned along one orientation. The rotation of the free F1 layer magnetization then ‘opens’ (in P configuration) or ‘closes’ (in AP configuration) the flow of electrons, acting as a sort of valve. The standard spin valve shows magnetoresistance values of about 5–6%. More complex spin-valve stacks, with layer thicknesses controlled at the atomic scale and ultra-low surface roughness, are optimized to favour specular reflection of electrons towards the active part so that the magnetoresistance reaches about 20%. Following the introduction in 1997 by IBM of the spinvalve sensor (FIG. 2b) to replace the AMR sensor in magnetoresistive HDD read heads, the growth rate for storage areal density immediately increased up to 100% per year. Together, the sequential introduction of the magnetoresistance and spin-valve head, by providing a sensitive and scalable read technique, contributed to increase the raw HDD areal recording density by three orders of magnitude (from ~0.1 to ~100 Gbit in–2) between 1991 and 2003. This jump forward opened the way both to smaller HDD form factors (down to 0.85-inch disk diameter) for mobile appliances such as ultra-light laptops or portable multimedia players, and

to unprecedented drive capacities (up to a remarkable 1 terabyte) for video recording or backup. HDDs are now replacing tape in at least the first tiers of data archival strategies, for which they provide faster random access and higher data rates. However, the areal density growth rate started to slow down after 2003, when other problems joined the now limiting spin-valve head. Attempts were also made to develop solid-state magnetic storage. Provided that the free layer magnetization of the spin valve is constrained to take only the two opposite orientations of an easy magnetization axis, arrays of patterned spin-valve elements can be used to store binary information with resistive read-out. Spin-valve solid-state memories were indeed developed13. But the planar geometry of the spin-valve sensor intrinsically limits the integration into high-density nanoelectronics, and the low metallic resistance and ΔR/R value around 10% are not well adapted to CMOS electronics. For read heads also, the ‘current in plane’ spin-valve geometry is a limitation to the downscaling. First, it is easy to see that the integration of the planar element of FIG. 2b between the magnetic shields of the read head of FIG. 1 strongly limits the minimum dimension of the read gap between the P1 and P2 poles, a crucial requirement for reducing the bit length: indeed, two thick insulating layers are needed on either part of the sensor. The ‘current perpendicular to plane’ (CPP) geometry of FIG. 2c is much better for this purpose, as the spin-valve sensor can be directly connected to the magnetic shields. This configuration is also more favourable for reducing the track width. Simple geometrical arguments based on the average electron propagation direction also lead us to expect higher magnetoresistance values for a spin valve in the CPP configuration. The fundamental study of CPP GMR was indeed extremely productive in terms of the new concepts of spin injection and spin accumulation14. The basic principle is displayed in FIG. 3. Again we assume that τup >> τdown in the ferromagnetic metal. So when a current flows from a ferromagnetic layer (F) to a non-magnetic layer (N), away from the interface the current densities jup and jdown must be very different on the ferromagnetic side, and equal on the nonmagnetic side. The necessary adjustment requires that, in the area near the interfaces, more electrons from the spin-up channel flip their spins. This occurs through an ‘accumulation’ of spin-up electrons, that is, a splitting of the EFup and EFdown Fermi energies, which induces spin-flips and MARCH 2008 | S33

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

COLLECTION a

I

b

V F1 ac nt o C

M

ts

F2

c

F1

I

Spin-valve sensor

V

M F2

Figure 2 |The spin valve. a, Schematic representation of the spin-valve effect in a trilayer film of two identical ferromagnetic layers F1 and F2 sandwiching a non-magnetic metal spacer layer M, the current circulating in plane. When the two magnetic layers are magnetized parallel (lower scheme), the spin-up electrons (spin antiparallel to the magnetization) can travel through the sandwich nearly unscattered, providing a conductivity shortcut and a low resistance. In contrast, in the antiparallel case (top scheme) both spin-up and spin-down electrons undergo collisions in either F1 or F2, giving rise to a higher overall resistance. b, Schematic arrangement of the ‘current in plane’ spin-valve sensor in a read head. c, Schematic arrangement of the ‘current perpendicular to plane’ spin-valve sensor in a read head. In both configurations, the recording medium travels parallel to the front face of the sensor.

adjusts the incoming and outgoing spin fluxes. The spin-accumulation decays exponentially on each side of the interface on the scale of the respective spin diffusion lengths LFsf and LNsf. In this spin accumulation zone, the spin polarization of the current decreases progressively going from the magnetic conductor to the non-magnetic one, so that a spin-polarized current is ‘injected’ into the non-magnetic metal up to a distance that can reach a few hundreds of nanometres, well beyond the ballistic range. This concept has been extended to more complex interfaces between metals and semiconductors15 (FIG. 3c). Likewise, the concept applies when an interfacial resistance such as a Shottky or insulating barrier exists16. The spin-injection effect has also been demonstrated for planar geometries17, and proposed for use in threeterminal devices such as the spin transistor18. We will come back to this point later. If the magnetoresistance ratio is larger in the CPP geometry, exceeding 20%, CPP GMR samples are much more difficult to fabricate19. In particular, quantitative measurements require either superconducting contacts20 or long multilayered wires21 to ensure current lines perpendicular to the layers. And the resistance and magnetoresistance still remain too small for optimal S34 | MARCH 2008

nature

application to read heads22. Considerable industrial research is going on (see for instance REF. 23). Magnetic tunnel junction Another big step forward came from replacing the non-magnetic metallic spacer layer M of the spin valve by a thin (~1–2 nm) non-magnetic insulating layer, thus creating a magnetic tunnel junction (MTJ). In that configuration the electrons travel from one ferromagnetic layer to the other by a tunnel effect, which conserves the spin (FIG. 4a). Again, this is not a new story, as it was proposed by Jullière24 in 1975, but its practical realization with a high magnetoresistance (up to 30% at 4.2 K) had to wait until 1995 once considerable progress had been made in deposition and nanopatterning techniques25, 26. The first MTJs used an amorphous Al2O3 insulating layer between ferromagnetic metal layers: the tunnel magnetoresistance (TMR) of such stacks reached a limit around 70% at room temperature. Much higher effects were later obtained with a single-crystal MgO barrier27, 28. Within such a barrier, the tunnelling current is carried by evanescent waves of several well-defined symmetries, which, at least for high-quality interfaces, have an interfacial

connection in the metals only to Bloch waves of the same symmetry at the Fermi level29, 30. In the typical case of MgO(001) between Co electrodes for instance, the decay is much slower and the transmission higher for the evanescent waves of symmetry Δ1, so that the TMR comes from the specific high spin polarization of the states of symmetry Δ1 in a zone of the Fermi surface of Co near [001]. The MgO barrier is thus ‘active’ in selecting a symmetry of high spin polarization, leading to latest record values of ΔR/R = 1010% at 5 K, 500% at room temperature (FIG. 4b)31. There is no sharp symmetry selection in amorphous barriers such as Al2O3, which explains the much lower TMR ratios achieved. The magnetic tunnel junction is clearly a CPP ‘vertical’ device, with a magnetic behaviour similar to a spin valve but magnetoresistance values up to two orders of magnitude higher. It is stable up to reasonable breakdown voltages (above 1 V), and the equipment suppliers rapidly developed reliable techniques to scale down its dimensions to well below 100 nm. So its development had an immediate impact on storage applications. Indeed, a TMR read head was commercialized by Seagate32 in 2005 (FIG. 4c), providing a higher sensitivity. This may prove to be a short-lived option: MTJs have an intrinsic high resistance (with resistance–area products above 1 Ω cm2), and with further downscaling it will become difficult to maintain a high signal-to-noise ratio with an increasing sensor resistance. Then CPP spin valves or degenerate MTJs could become more favourable22. Magnetic random access memory The ‘high’ MTJ resistance is actually more adapted to nanoelectronics. The 1995 publications started a race to develop the magnetic random access memory, or MRAM13. FIGURE 5a shows the principle of this magnetic solid state memory, in the basic ‘cross point’ architecture. The binary information 0 and 1 is recorded on the two opposite orientations of the magnetization of the free layer along its easy magnetization axis. The MTJs are connected to the crossing points of two perpendicular arrays of parallel conducting lines. For writing, current pulses are sent through one line of each array, and only at the crossing point of these lines is the resulting magnetic field high enough to orient the magnetization of the free layer. For reading, the resistance between the two lines connecting the addressed cell is measured. In principle, this cross point architecture promises very high densities. In practice,

www.nature.com/milestones/spin

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

COLLECTION the amplitude of magnetoresistance remains too low for fast, reliable reading because of the unwanted current paths as well as the direct one through the addressed cell. So, realistic cells add one transistor per cell, resulting in more complex 1T/1MTJ cell architectures such as the one represented in FIG. 5b. Several demonstrator circuits were rapidly presented by most leading semiconductor companies, culminating with the first MRAM product, a 4-Mbit stand-alone memory33 commercialized by Freescale in 2006 (FIG. 5c), and voted ‘Product of the Year’ by Electronics Products Magazine in January 2007. The MRAM potentially combines key advantages such as non-volatility, infinite endurance and fast random access (down to 5 ns read/write time34) that make it a likely candidate for becoming the ‘universal memory’, one of the chief aims of nanoelectronics. Such a memory is able to provide data/code (Flash, ROM) and execution (DRAM, SRAM) storage using a single memory technology on the same die. Moreover, in June 2007 Freescale introduced a new version able to work in the expanded temperature range of –40 °C to 105 °C, thus qualifying for military and space applications where the MRAM will also benefit from the intrinsic resistance to radiation of magnetic storage. Nanomagnetism Progress in spin electronics cannot be separated from the development of ‘nanomagnetism’. In particular, the engineering of magnetic properties at atom level in multilayers was developed in parallel with GMR and helped to make it possible. Improved knowledge of the role of interface effects, and the use of the layer thickness as a parameter, led to the development of artificial magnetic materials with finely tuned new properties. This had a direct impact on spin storage. The magnetic storage of binary information requires the engineering of an energy barrier between two opposite orientations of the magnetization, able to stop thermally excited reversals35. This ‘magnetic anisotropy’ has several competing origins. The strongest one is usually the shape anisotropy due to the dipole–dipole magnetic interaction, which induces the well-known in-plane easy magnetization of thin films. But the main effect used in recording is the magnetocrystalline anisotropy, an atomic effect correlated to the symmetry of the immediate atomic environment. The interface anisotropy, initially proposed by Néel36, takes advantage of the break in

NATURE MILESTONES | SPIN

nature

a

Ferromagnetic

Non-magnetic Zone of spin accumulation

N

L sfF

b

L sf

E

EF

up

Spin accumulation ΔE F = E F –EF up

down

z EF

down

c Metal/metal Current spin polarization = ( jup –jdown )/( jup +j down )

Metal/semiconductor

z Figure 3 | Spin accumulation. Schematic representation of the spin accumulation at an interface between a ferromagnetic metal and a non-magnetic layer, adapted from REF. 132. a, Spin-up and spin-down current far from an interface between the ferromagnetic and nonmagnetic conductors (outside the spin-accumulation zone). LFsf and LNsf are, respectively, the spin diffusion lengths in the ferromagnetic and non-magnetic layers. b, Splitting of the Fermi levels EFup and EFdown at the interface. The dashed green arrows symbolize the transfer of current between the two channels by the unbalanced spin flips caused by the out-of-equilibrium spin-split distribution, which governs the depolarization of the electron current between the left and the right. With the current in the opposite direction, there is an inversion of the spin accumulation and opposite spin flips, which polarizes the current across the spin-accumulation zone. c, Variation of the current spin polarization when there is an approximate balance between the spin flips on both sides (metal/metal) and when the spin flips on the left side are predominant (metal/semiconductor, for example). The current densities for spin-up and spin-down electrons are jup and jdown, respectively.

translational symmetry at an interface to generate giant magnetic anisotropies, able to overcome the shape anisotropy and induce a stable perpendicular magnetization axis

(PMA) in ultrathin films and multilayers. This PMA was first observed in 1967 on single-atomic-layer films37, and achieved in 1985 in Co/Pd multilayer38 and Au/Co/Au MARCH 2008 | S35

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

A

BEMaGS

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

F

COLLECTION a

b Parallel state

1,200 1,000

eV

EF

TMR ratio (%)

800 600 400 200 0

Barrier

-300

-200

Antiparallel state

-100

0

100

200

300

External field (Oe)

c

eV

EF

Barrier

Figure 4 | The magnetic tunnel junction. a, Schematic representation of the tunnel magnetoresistance in the case of two identical ferromagnetic metal layers separated by a non-magnetic amorphous insulating barrier such as Al2O3. The tunnelling process conserves the spin. When electron states on each side of the barrier are spin-polarized, then electrons will more easily find free states to tunnel to when the magnetizations are parallel (top picture) than when they are antiparallel (bottom picture). b, Record high magnetoresistance TMR = (R max – R min)/R min for the magnetic stack (Co25Fe75)80B20 (4 nm)/MgO (2.1 nm)/(Co25Fe75)80B20 (4.3 nm) annealed at

films39 with more practical thicknesses. New materials could even be predicted from ab initio calculations40. PMA is now used for recording media in the ‘perpendicular’ HDD introduced by Seagate, Hitachi and Toshiba in 2005–06, which helped to restore the present 40% growth rate after the slow-down experienced in 2003–04. Exchange bias is another crucial effect linked to an interface, here between a ferromagnetic and antiferromagnetic layer: the antiferromagnetic layer has no net magnetic moment that could be sensitive to an applied field, but may retain a large magnetic anisotropy, which, transferred to the ferromagnetic layer through interfacial exchange interaction, contributes to stabilizing the orientation of its magnetization. This is also an old story41, but with progress in interface control42 it gained wide application S36 | MARCH 2008

nature

475 °C after growth, measured at room temperature (filled circles) and at 5 K (open circles). Reprinted with permission from REF. 31. c, Transmission electron microscope cross-section of a TMR read head from Seagate. Reprinted with permission from REF. 32. The tunnel junction stack appears vertically at the centre of the picture, with the tunnel barrier at the level of the thin white horizontal line. The thick bent lines on both sides are the insulating layers between top and bottom contacts. The two thick light grey layers on top and bottom are the magnetic pole pieces (see FIG. 1). The track width of the TMR element is typically 90–100 nm.

in the spin valve and magnetic tunnel junctions for pinning the magnetization of the reference magnetic layer. And it can also be used in a new kind of MRAM43, or to fight thermal excitations of magnetic nanoparticles44 for future storage. The growing structural quality of interfaces has led to the spin-dependent quantum confinement of the electrons in metallic ultrathin layers. Together with interfacial band hybridization this now makes it possible to control the exchange interaction between ferromagnetic layers separated by a non-magnetic layer45–49. This, for instance, enables synthetic antiferromagnets (SAF) to be built, these being trilayers where two ferromagnetic layers are kept magnetized antiparallel by exchange through a non-magnetic spacer. Depending on their exact structure and properties, such SAFs can provide

either improved non-volatility for constant writing field50, or more reliable writing in solid state devices51, or simply a weak sensitivity to applied field and minimal stray field when the net magnetic moment is brought to zero. They are thus used now as recording media for longitudinal HDD products1 as well as in Freescale’s MRAM product. And they are ubiquitous in all pinned layers of spin-valve sensors and MRAM cells where they help the exchange bias and minimize the interlayer dipole–dipole interaction. So at the beginning of this century outstanding progress had been made both in designing the magnetic properties through atomic engineering, and in understanding and controlling the spin-dependent electron transport. And materials exist that allow thermally stable magnetic particles to be produced down to sizes of a few

www.nature.com/milestones/spin

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

COLLECTION nanometres52, seemingly opening a bright future for a high-density spin storage. Writing is the problem The use of a magnetic field to write the information still remained a considerable limitation. This can easily be understood. Let us assume that information is stored in the form of the magnetization orientation of a nanoparticle of volume V. The energy barrier fighting the thermal excitations is given by KV, where K is the anisotropy constant per unit volume. Non-volatility — usually defined by a maximum error rate, for example 10-9, over a 10-year period — is obtained when KV > 50–60 kBT, where kB is the Boltzmann constant and T the temperature. So reducing V requires a corresponding increase in K, but then the writing field increases proportionally with K, whereas the power available to create it decreases as the dimensions are downscaled. The problem is well known in HDD, where it was recently postponed by first introducing the SAF longitudinal media and then changing to perpendicular recording. The future also promises heat-assisted magnetic recording (HAMR)1, where the magnetic field is helped by local heating of the media temporarily reducing the energy barrier, as in magneto-optical recording. But the fundamental limit is still there, together with other problems linked to the rotating nature of HDD storage such as mechanical tracking, slow access time (hardly reduced in several decades and still a few milliseconds) and energy consumption. The future markets of HDD, in particular through the competition with Flash storage, will depend on how such problems are solved. For instance, HDD currently maintains an areal density growth rate of roughly 40%, in line with Moore’s law defining the minimum cell size of Flash. But meanwhile multi-level Flash cells have been introduced (2 bits per cell), and huge efforts have succeeded in reducing their cost. In MRAM the writing problem was immediately worse than in HDD, as the conducting lines have much smaller dimensions, with a strong limitation in current density around 107 A cm–2 due to electromigration. Also, it is not possible in a very large-scale integration circuit to include an optimized ‘ring-like’ ferromagnetic circuit to ‘channel’ the magnetic induction to the magnetic media. A magnetic channelling was developed for MRAM53, but the effect is limited (to a factor of about two) and requires costly fabrication steps. Finally, when approaching the downscaling limits

NATURE MILESTONES | SPIN

nature

a

‘Bit’ lines

MRAM

‘Cross point’ architecture

‘Word’ lines

‘1’ ‘0’ 1T/1MTJ cell architecture

b

c

Transistor

Figure 5 | Magnetic random access memory. a, Principle of MRAM, in the basic cross-point architecture. The binary information 0 and 1 is recorded on the two opposite orientations of the magnetization of the free layer of magnetic tunnel junctions (MTJ), which are connected to the crossing points of two perpendicular arrays of parallel conducting lines. For writing, current pulses are sent through one line of each array, and only at the crossing point of these lines is the resulting magnetic field high enough to orient the magnetization of the free layer. For reading, the resistance between the two lines connecting the addressed cell is measured. b, To remove the unwanted current paths around the direct one through the MTJ cell addressed for reading, the usual MRAM cell architecture has one transistor per cell added, resulting in more complex 1T/1MTJ cell architecture such as the one represented here. c, Photograph of the first MRAM product, a 4-Mbit stand-alone memory commercialized by Freescale in 2006. Reprinted with permission from REF. 33 © IEEE 2006.

the unavoidable distribution of writing parameters, coupled to the large stray fields in such densely packed arrays, leads to spreading program errors. Freescale researchers elegantly solved this reliability problem by replacing the standard ferromagnetic free layer with a SAF layer, written using a spin-flop process33, 51, 54, and this opened the way to the first MRAM product. But this is at the expense of higher writing currents (around 10 mA), and clearly limits the achievable densities and the downscaling. As for HDD, one solution could be heat-assisted recording (TAS-RAM)43, 55. It was also proposed that writing could be assisted by microwave excitation at the ferromagnetic resonance frequency of the free layer56–58, a technique that could also be useful for hard disks. But such promising techniques do not completely suppress the need for a magnetic field.

Spin transfer — a new route for writing magnetic information The hoped for breakthrough for spin storage was provided by the prediction59, 60 in 1996 that the magnetization orientation of a free magnetic layer could be controlled by direct transfer of spin angular momentum from a spin-polarized current. In 2000, the first experimental demonstration that a Co/Cu/Co CPP spin-valve nanopillar can be reversibly switched by this ‘spin-transfer effect’ between its low (parallel) and high (antiparallel) magnetoresistance states was presented61. The concept of spin transfer actually dates back to the 1970s, with the prediction62 and observations63 of domainwall dragging by currents. Spin-transfer effects had also been predicted64 for MTJs as early as 1989. Somehow, those predictions did not immediately trigger the intense research work that followed the 1996 and

MARCH 2008 | S37

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

COLLECTION a

b

Cobalt / Copper / Cobalt

F1 thick layer

M

F2 thin layer

e– e–

e–

Writing ‘0’ S S

F1

Transverse component

e–

F2

Writing ‘1’

Figure 6 | Spin-transfer switching. a, Principle of the STT effect, for a typical case of a Co(F1)/Cu/Co(F2) trilayer pillar. A current of s electrons flowing from left to right will acquire through F1 (assumed to be thick and acting as a spin polarizer) an average spin moment along the magnetization of F1. When the electrons reach F2, the s–d exchange interaction quickly aligns the average spin moment along the magnetization of F2. To conserve

2000 publications, possibly because the required fabrication technologies were not mature enough. The principle of ‘spin-transfer torque’ (STT) writing in nanopillars is shown schematically in FIG. 6a for the usual case of 3d ferromagnetic metals (for example Co) in a spin-valve structure with a nonmagnetic metal spacer (for example Cu). Let us consider a ‘thick’ ferromagnetic layer F1, whereas the ferromagnetic layer F2 and the spacer M are ‘thin’ (compared with the length scales of spin-polarized transport65). F1 and F2 are initially magnetized along different directions. A current of s electrons flowing from F1 to F2 will acquire through F1, acting as a spin polarizer, an average spin polarization approximately along the magnetization of F1. When the electrons reach F2, the s–d exchange interaction quickly aligns the average spin moment along the magnetization of F2. In the process, the s electrons have lost a transverse spin angular momentum, which, because of the total angular momentum conservation law, is ‘transferred’ to the magnetization of F2. This results in a torque tending to align F2 magnetization towards the spin moment of the incoming electrons, and thus towards the magnetization of F1. Because the loss of transverse spin momentum happens over a very short distance (around 1 nm), the torque is an interfacial effect, more efficient on a thin layer. But a more important result S38 | MARCH 2008

nature

the total angular momentum, the transverse spin angular momentum lost by the electrons is transferred to the magnetization of F2, which senses a resulting torque tending to align its magnetization towards F1. b, Principle of STT writing of a MRAM cell: reversing the current flowing through the cell will induce either parallel or antiparallel orientation of the two ferromagnetic layers F1 and F2.

is that the amplitude of the torque per unit area is proportional to the injected current density, so that the writing current decreases proportionally to the cross-sectional area of the structure. With today’s advances in nanotechnologies and the easy access to sizes below 100 nm, this represents an important advantage of spin transfer over field-induced writing. A realistic treatment of the effect65 includes both quantum effects at the interfaces (spin-dependent transmission of Bloch states) and diffusive transport theory (spin-accumulation effects), and the dynamical behaviour can be studied through a modified Landau–Lifshitz–Gilbert equation describing the damped precession of magnetization in the presence of STT and thermal excitations65, 66. The principle of STT writing of a MRAM cell is shown in FIG. 6b. Electrons flowing from the thick ‘polarizing’ layer to the thin free layer favour a parallel orientation of the magnetizations: if the initial state is antiparallel, then beyond a threshold current density jC+ the free layer will switch. When the electrons flow from the free to the polarizing layer, it can be shown that the effective spin moment injected in the free layer is opposed to the magnetization of the polarizing layer, writing an antiparallel configuration beyond a threshold current density jC–. Since the first observation on Co/Cu/Co trilayers, STT writing has been achieved on

many different stacks, including exchangebias pinned layers and SAF layers67. It also works with tunnel junctions68, and in particular with MgO tunnel barriers69. Moreover, the threshold current densities are becoming nearly compatible with NMOS transistor output as predicted by the International Technology Roadmap for Semiconductors. And in the near future, TAS and STT writing modes could be combined for an even smaller switching current. Companies have already presented several demonstrations of ‘spin-RAM’70, 71. As shown in FIG. 7, the cell structure has now become extremely simple, opening the way to high densities. Is it enough to compete with NAND Flash on mass data storage in standalone memories? In terms of areal density, Flash still offers a smaller multi-bit cell, and three-dimensional (3D) stacking has been announced72. But MRAM has other advantages, such as potentially infinite endurance (compared with ~105 cycles for a Flash) and potential for sub-nanosecond operation66, 73, 74, that make it competitive as universal memory. Perspectives One grave limitation to ultra-high-density spin-MRAM is the requirement of one transistor per cell (1T/1MTJ). The crosspoint memory architecture provides a way of reaching very high densities75, lower fabrication costs and a potential for 3D stacking of

www.nature.com/milestones/spin

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

A

BEMaGS

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

F

COLLECTION several recording layers. Intermediate cell structures such as 1T/4MTJ76 have also been proposed. Multi-level cell operation was also recently achieved in TAS-RAM77. But in all cases increased density is obtained at the expense of smaller signal amplitude and thus much slower read. One step to fight these limitations, at least partially, would be to gain at least one order of magnitude in the amplitude of the magnetoresistance, moving towards a true ‘current switch’ with ΔR/R values comparable, for example, to those of phase-change RAM (PCRAM). This could be achieved by replacing metallic ferromagnetic layers with 100% spin-polarized conductors such as half-metallic oxides78 or Heussler alloys79, 80, or diluted magnetic semiconductors (DMS)81, 82. DMS furthermore open the way to new effects such as tunnelling anisotropic magnetoresistance83, 84, and, in the long term, to quantum dot storage devices85. Another promising concept is that of the ‘spin filter’86 where tunnelling happens through a ferromagnetic barrier: the transmission

a

varies exponentially with the square root of the barrier height, which itself depends on electron spin direction versus barrier magnetization. Such developments would of course also benefit HDD. Another approach is to build a three-terminal device that can produce both the transistor effect and magnetoresistance in a single magnetic device87–89. All these ideas show promising results at low temperature, but depend on materials issues such as obtaining Curie temperatures well above room temperature, mastering a complex stoichiometry (oxides, Heussler alloys) at an interface, or maintaining the fabrication thermal budget compatible with the CMOS process, though these new materials usually require specific high-temperature growth. Moreover, three-terminal devices can so far provide only very small currents (microamperes at most), below CMOS compatibility levels, and independent writing of one or two magnetic layers may prove difficult to scale down even using spin transfer.

b

Bit line Bipolar write pulse/read bias generator

Future magnetic mass storage could come instead from domain wall devices, in an approach conceptually close to that of the former ‘bubble’ magnetic memories90. File architecture for mass storage does not require random access to a single bit or word as proposed in spin-RAM, but can accommodate random access only to large sectors in which binary information is written or read sequentially, such as in HDD. A chain of domain walls in a magnetic stripe can indeed represent such a sector storing binary information, but a simple application of a uniform magnetic field would immediately destroy it by annihilating the reverse domain. It was first proposed91, 92 to use an oscillating magnetic field, uniform over the whole chip, acting on a specific domain wall circuit behaving as a shift register: the global character of the applied field makes downscaling and 3D stacking achievable93. Even more promising for very-high-density solidstate integration, a current injected in the magnetic stripe applies the same pressure to all domain walls along the direction of

Free layer Pinned layer

MTJ Word line Transistor

Sense amp.

100 nm 10

Source line Reference

9

BL SL

BL SL

BL

WL WL

8

Resistance (kΩ)

SL

No. of cycles 1 10 100 1K 10K 100K 1M 10M 100M 1G ‘0’

100 ns 1 cycle operation

6

WL

‘1’

7

5

WL 4 –10

–8

–6

–4

–2

0

2

4

6

8

10

Current (a.u.)

Figure 7 | The spin-RAM. a, Schematic architecture of a spin-RAM; upper panel, scheme of the memory cell, and lower panel, tentative architecture of the cell array. Reprinted with permission from REF. 70. b, Resistance versus current hysteresis loop of a spin-RAM cell. Reprinted with permission from REF. 71 © IEEE 2005. The different colours show the

NATURE MILESTONES | SPIN

nature

evolution of the loop after an increasing number (up to 1 G = 109) of writing cycles (100 ns pulses of successively positive and negative currents, see image). This demonstrates excellent stability. TEM image: TMR device size 100 nm × 50 nm; free layer CoFe (1.0 nm) / NiFe (2.0 nm); tunnel barrier MgO (1.0 nm).

MARCH 2008 | S39

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

COLLECTION a

b

R

DW propagation

e–

Tunnel barrier

e–

e–

Magnetic strip

R

DW propagation

e– Read head

Write head

Wafer

Figure 8 | Domain wall storage devices. Examples of storage devices using current-induced domain wall (DW) propagation. a, In the concept first proposed by Parkin94, the binary information is stored by a chain of domain walls in a magnetic stripe. An electrical current in the stripe, by applying the same pressure to all the domain walls, moves them simultaneously at the same speed for a sequential reading (or writing) at fixed read and write heads. A reverse current can move the domain walls in the opposite direction for resetting, or in an alternative solution the domain walls might turn on a loop. This mimics the fast passing of bits in front of the head in HDD recording, but here there is no moving part and addressing a sector would be done by CMOS electronics at microsecond access times. The initial scheme94 proposes to store data in vertical stripes: this would open the way to very compact high capacity ‘storage track memory devices’. Other schemes now propose multilayers of in-plane domain tracks, which would be easier to fabricate. b, Scheme of a MRAM cell using domain wall propagation from one stable position to another on either side of a magnetic tunnel junction (REF. 95).

electron travel, propagating the walls simultaneously at the same speed, without losing information. This mimics the fast passing of bits in front of the head in HDD recording, but here there is no moving part (hence increased ruggedness) and addressing a sector would be done by CMOS electronics with microsecond access times. This scheme thus opens the way to very compact ‘storage track memory devices’94, and could also be used in a new kind of MRAM95 (FIG. 8). But many advances are needed before its practical implementation. The theory mixes spin transfer torque (electrons crossing a domain wall transfer spin angular momentum to the non-uniform magnetization in the wall) and mechanical momentum transfer (electrons are reflected from narrow walls), leading to a domain wall propagation controlled with a current density beyond a threshold96, 97. The effect has indeed been observed in metals magnetized in plane98, 99 and perpendicular to plane100, or in DMS101 (with orders of magnitude lower currents but much higher resistance and other problems). However, even qualitative agreement between experience and theory is not straightforward102, 103. On the experimental aspect, crucial progress has to be made in reliably controlling domain wall propagation with low current densities. First non-volatile reliable trapping of the domain walls must be provided at dedicated positions in the ‘data sector’. Most works use simple notches104, 105, but S40 | MARCH 2008

nature

more elaborate pinning profiles have also been proposed106, 107. But not much has been published on the evolution of the thermal stability of the pinning when downscaling the dimensions. Besides, even in a parallel stripe and even more at artificial notches with a distribution of patterning defects, domain walls can take different structures that are close in energy, and may even change structure under a current pulse105, 108–110 . This is a potential problem for reliable operation, because the parameters of current-induced propagation should depend on the wall structure. Last but not least, the threshold current density necessary to depin a domain wall from a trap, or even to start motion in a parallel stripe, is still too high for applications (above a few 106 A cm–2). Great progress has recently been made by using the natural radiofrequency dynamics of a domain wall pinned in a potential well111–113. Note, however, that if the current density is still too high, the cross-section of a thin-film stripe is intrinsically small so the threshold current is already much smaller than is available from the smallest CMOS transistors, a key asset for developing lowpower data-storage devices (FIG. 8b). Finally, the achievable domain wall speed is also a crucial issue for fast data rates114–117. For instance, with maximum domain wall speed around 100 m s–1 and an on-track linear density of one domain wall every 200 nm, already very demanding values, the data rate would be around 0.5 Gbit s–1. This would be

in line with today’s HDD data rates, but is far from being demonstrated. In the long term, even higher densities could be reached by replacing domain walls by smaller magnetization vortices118, in a trend similar to that from bubble to Bloch line memories90. We have tried in this review to share our amazement at the outstanding progress in spin electronics over the past two decades, under the convergence of a chain of scientific breakthroughs and technology advances. If giant magnetoresistance opened the way in 1988 to control transport through magnetization, spin transfer now allows magnetization to be controlled through transport, closing the loop for a new paradigm, from magnetic recording to spin storage. Traditional hard-disk recording has gained orders of magnitude in storage capacity, thus entering the consumer electronics markets. And the MRAM magnetic solid-state memory is now in production, although as yet only for niche markets. But the intrinsic speed and endurance of magnetic recording, together with the potential of the spin-RAM to work with CMOScompatible electrical parameters, could open the way to applications where data storage would not be the primary objective although non-volatility would still be a key asset. Along this line, it has been proposed that logic calculations through magnetic interactions could be performed, in magnetic quantum cellular automata119, 120, or in domain wall logic92, for low-power massively parallel logic operations under a uniform cyclical magnetic field. MTJs can also be used for logic calculation, either directly121 (a nice idea but one whose practical realization is uncertain) or by a dense integration of MTJs into CMOS logic circuits122, 123 where they bring instant ON/OFF, run time re-programmability, and overall improved operation safety. Magnetism would thus enter the realm of the CPU. But a great step forward would be to realize three-terminal spin electronic devices that would enable a complete programmable logic function to be packed into a single nanodevice. Let us assume a source–gate–drain device where the magnetization of magnetic source and drain could be independently controlled, injecting spin-polarized electrons into a channel of spin-dependent transmission that could also be controlled by a gate voltage. Such a device has multiple inputs to control a multi-level output, realizing a logic function that can be programmed through the non-volatile magnetic configurations. The first proposition of this kind18, despite recent

www.nature.com/milestones/spin

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

COLLECTION progress in injecting spin polarization into semiconductors124, has not yet been achieved in practice. New concepts are being proposed125–127, and more could be realized, for instance, with molecules128–132. In the longer term, the use of spin injection and spin currents21, 133 may lead to the development of ‘spin logic’ devices134. Ultimately, ‘magnetic’ writing will again become a problem in much smaller and more complex devices, and new routes will have to be found. As in nanoelectronics, zero-current, gate-voltage-controlled writing would be ideal. Preliminary results have recently been obtained by using interfacial coupling with piezoelectric or even multiferroic materials135–137, or through electric field control of ferromagnetism in DMS138–139. In an even more futuristic approach, switching by spin currents only (no charge currents) has been announced140, in a pioneering step towards nanoelectronics using spin currents only. As a whole, finding solutions to the magnetic writing problem may prove to be a key issue on the way to future spin electronics, as it has been for the past evolution of magnetic recording. Claude Chappert is at the Institut d’Electronique Fondamentale, CNRS, UMR8622, 91405 Orsay, France and at Université Paris Sud, 91405 Orsay, France Albert Fert is at Université Paris Sud, 91405 Orsay, France and at Unité Mixte de Physique CNRS-Thales, 91767 Palaiseau, France Correspondence to: Claude Chappert e-mail: claude. [email protected] ___________ 1.

Moser, A. et al. Magnetic recording: advancing into the future. J. Phys. D 35, R157–R167 (2002). 2. Mott, N. Electrons in transition metals. Adv. Phys. 13, 325–422 (1964). 3. Fert, A. & Campbell, I. A. Two-current conduction in nickel. Phys. Rev. Lett. 21, 1190–1192 (1968). 4. Fert, A. & Campbell, I. Electrical resistivity of ferromagnetic nickel and iron based alloys. J. Phys. F 6, 849–871 (1976). 5. Fert, A., Duvail, J. & Valet, T. Spin relaxation effects in the perpendicular magnetoresistance of magnetic multilayers. Phys. Rev. B 52, 6513–6521 (1995). 6. Baibich, M. N. et al. Giant magnetoresistance of (001)Fe/(001)Cr magnetic superlattices. Phys. Rev. Lett. 61, 2472–2475 (1988). 7. Binasch, G., Grünberg, P., Saurenbach, F. & Zinn, W. Enhanced magnetoresistance in layered magnetic structures with antiferromagnetic interlayer exchange. Phys. Rev. B 39, 4828–4830 (1989). 8. Levy, P. M. & Mertig, I. in Spin Dependent Transport in Magnetic Nanostructures (eds Maekawa, S. & Shinjo, T.) Ch. 2, 47–112 (CRC, Boca Raton, 2002). 9. Fert, A., Barthélémy, A. & Petroff, F. in Nanomagnetism: Ultrathin Films, Multilayers and Nanostructures (eds Mills, D. M. & Bland, J. A. C.) Ch. 6 (Elsevier, Amsterdam, 2006). 10. Grünberg, P. Magnetic field sensor with ferromagnetic thin layers having magnetically antiparallel polarized components. US patent 4,949,039 (1990). 11. Dieny, B. et al. Magnetoresistive sensor based on the spin valve effect. US patent 5,206,590 (1993). 12. Dieny, B. et al. Giant magnetoresistance in soft ferromagnetic multilayers. Phys. Rev. B 43, 1297–1300 (1991). 13. Daughton, J. M. Magnetic tunneling applied to memory. J. Appl. Phys. 81, 3758–3763 (1997).

NATURE MILESTONES | SPIN

nature

14. Valet, T. & Fert, A. Theory of the perpendicular magnetoresistance in magnetic multilayers. Phys. Rev. B 48, 7099–7113 (1993). 15. Schmidt, G., Ferrand, D., Molenkamp, L. W., Filip, A. T. & van Wees, B. J. Fundamental obstacle for electrical spin injection from a ferromagnetic metal into a diffusive semiconductor. Phys. Rev. B 62, R4790–R4793 (2000). 16. Fert, A. & Jaffrès, H. Conditions for efficient spin injection from a ferromagnetic metal into a semiconductor. Phys. Rev. B 64, 184420 (2001). 17. Jedema, F. J., Filip, A. T. & van Wees, B. J. Electrical spin injection and accumulation at room temperature in an all-metal mesoscopic spin valve. Nature 410, 345–348 (2001). 18. Datta, S. & Das, B. Electronic analog of the electro-optic modulator. Appl. Phys. Lett. 56, 665–667 (1990). 19. Gijs, M. A. M., Lenczowski, S. K. J. & Giesbers, J. B. Perpendicular giant magnetoresistance of microstructured Fe/Cr magnetic multilayers from 4.2 to 300 K. Phys. Rev. Lett. 70, 3343–3346 (1993). 20. Bass, J. & Pratt, W. P. Current-perpendicular (CPP) magnetoresistance in magnetic metallic multilayers. J. Magn. Magn. Mater. 200, 274–289 (1999). 21. Fert, A. & Piraux, L. Magnetic nanowires. J. Magn. Magn. Mater. 200, 338–358 (1999). 22. Takagishi, M. et al. The applicability of CPP-GMR heads for magnetic recording. IEEE Trans. Magn. 38, 2277–2282 (2002). 23. Childress, J. et al. Fabrication and recording study of all-metal dual-spin-valve CPP read heads. IEEE Trans. Magn. 42, 2444–2446 (2006). 24 Jullière, M. Tunneling between ferromagnetic films. Phys. Lett. A 54, 225–226 (1975). 25. Moodera, J. S., Kinder, L. R., Wong, T. M. & Meservey, R. Large magnetoresistance at room temperature in ferromagnetic thin film tunnel junctions. Phys. Rev. Lett. 74, 3273–3276 (1995). 26. Miyazaki, T. & Tezuka, N. Giant magnetic tunneling effect in Fe/Al2O3/Fe junction. J. Magn. Magn. Mater. 139, L231–L234 (1995). 27. Parkin, S. S. P. et al. Giant tunnelling magnetoresistance at room temperature with MgO (100) tunnel barriers. Nature Mater. 3, 862–867 (2004). 28. Yuasa, S., Nagahama, T., Fukushima, A., Suzuki, Y. & Ando, K. Giant room-temperature magnetoresistance in single-crystal Fe/MgO/Fe magnetic tunnel junctions. Nature Mater. 3, 868–871 (2004). 29. Butler, W. H., Zhang, X., Schulthess, T. C. & MacLaren, J. M. Spin-dependent tunneling conductance of Fe/ MgO/Fe sandwiches. Phys. Rev. B 63, 054416 (2001). 30. Mathon, J. & Umerski, A. Theory of tunneling magnetoresistance of an epitaxial Fe/MgO/Fe(001) junction. Phys. Rev. B 63, 220403 (2001). 31. Lee, Y. M., Hayakawa, J., Ikeda, S., Matsukura, F. & Ohno, H. Effect of electrode composition on the tunnel magnetoresistance of pseudo-spin-valve magnetic tunnel junction with a MgO tunnel barrier. Appl. Phys. Lett. 90, 212507 (2007). 32. Mao, S. et al. Commercial TMR heads for hard disk drives: characterization and extendibility at 300 gbit/ in. IEEE Trans. Magn. 42, 97–102 (2006). 33. Engel, B. et al. A 4-Mb toggle MRAM based on a novel bit and switching method. IEEE Trans. Magn. 41, 132–136 (2005). 34. DeBrosse, J. et al. A high-speed 128-kb MRAM core for future universal memory applications. IEEE J. SolidState Circ. 39, 678–683 (2004). 35. Brown, W. F. Thermal fluctuations of a single-domain particle. Phys. Rev. 130, 1677–1686 (1963). 36. Néel, L. Anisotropie superficielle et surstructures d’orientation magnétique. J. Phys. Rad. 15, 225–239 (1954). 37. Gradmann, U. & Müller, J. Flat ferromagnetic, epitaxial 48Ni/52Fe(111) films of few atomic layers. Phys. Status Solidi B 27, 313–324 (1968). 38. Carcia, P. F., Meinhaldt, A. D. & Suna, A. Perpendicular magnetic anisotropy in Pd/Co thin film layered structures. Appl. Phys. Lett. 47, 178–180 (1985). 39. Chappert, C., Renard, D., Beauvillain, P. & Renard, J. Ferromagnetism of very thin films of nickel and cobalt. J. Magn. Magn. Mater. 54–57, 795–796 (1986). 40. Daalderop, G. H. O., Kelly, P. J. & den Broeder, F. J. A. Prediction and confirmation of perpendicular magnetic anisotropy in Co/Ni multilayers. Phys. Rev. Lett. 68, 682–685 (1992). 41. Meiklejohn, W. H. & Bean, C. P. New magnetic anisotropy. Phys. Rev. 102, 1413–1414 (1956).

42. Nogues, J. et al. Exchange bias in nanostructures. Phys. Rep. 422, 65–117 (2005). Prejbeanu, I. et al. Thermally assisted switching in exchange-biased storage layer magnetic tunnel junctions. IEEE Trans. Magn. 40, 2625–2627 (2004). 44. Skumryev, V. et al. Beating the superparamagnetic limit with exchange bias. Nature 423, 850–853 (2003). 45. Grünberg, P., Schreiber, R., Pang, Y., Brodsky, M. B. & Sowers, H. Layered magnetic structures: evidence for antiferromagnetic coupling of Fe layers across Cr interlayers. Phys. Rev. Lett. 57, 2442–2445 (1986). 46. Majkrzak, C. F. et al. Observation of a magnetic antiphase domain structure with long-range order in a synthetic Gd-Y superlattice. Phys. Rev. Lett. 56, 2700–2703 (1986). 47. Parkin, S. S. P., More, N. & Roche, K. P. Oscillations in exchange coupling and magnetoresistance in metallic superlattice structures: Co/Ru, Co/Cr, and Fe/Cr. Phys. Rev. Lett. 64, 2304–2307 (1990). 48. Bruno, P. & Chappert, C. Oscillatory coupling between ferromagnetic layers separated by a nonmagnetic metal spacer. Phys. Rev. Lett. 67, 1602–1605 (1991). 49. Bruno, P. Theory of interlayer magnetic coupling. Phys. Rev. B 52, 411–439 (1995). 50. Margulies, D. T., Berger, A., Moser, A., Schabes, M. E. & Fullerton, E. E. The energy barriers in antiferromagnetically coupled media. Appl. Phys. Lett. 82, 3701–3703 (2003). 51. Savchenko, L., Engel, B. N., Rizzo, N. D., Deherrera, M. F. & Janesky J. A. Method of writing to scalable magnetoresistance random access memory element. US patent 6,545,906B1 (2003). 52. Weller, D. et al. High Ku materials approach to 100 Gbits/in. IEEE Trans. Magn. 36, 10–15 (2000). 53. Durlam, M. et al. Low power 1 Mbit MRAM based on 1T1MTJ bit cell integrated with copper interconnects. Symp. VLSI Techn. Dig., 158–161 (2002). 54. Worledge, D. C. Spin flop switching for magnetic random access memory. Appl. Phys. Lett. 84, 4559–4561 (2004). 55. Daughton, J. M. & Pohm, A. V. Design of Curie point written magnetoresistance random access memory cells. J. Appl. Phys. 93, 7304–7306 (2003). 56. Rizzo, N. D. & Engel, B. N. MRAM write apparatus and method. US patent 6,351,409 (2002). 57. Thirion, C., Wernsdorfer, W. & Mailly, D. Switching of magnetization by nonlinear resonance studied in single nanoparticles. Nature Mater. 2, 524–527 (2003). 58. Nembach, H. T. et al. Microwave assisted switching in a Ni81Fe19 ellipsoid. Appl. Phys. Lett. 90, 062503 (2007). 59. Slonczewski, J. Current-driven excitation of magnetic multilayers. J. Magn. Magn. Mater. 159, L1–L7 (1996). 60. Berger, L. Emission of spin waves by a magnetic multilayer traversed by a current. Phys. Rev. B 54, 9353–9358 (1996). 61. Albert, F. J., Katine, J. A., Buhrman, R. A. & Ralph, D. C. Spin-polarized current switching of a Co thin film nanomagnet. Appl. Phys. Lett. 77, 3809–3811 (2000). 62. Berger, L. Prediction of a domain-drag effect in uniaxial, non-compensated, ferromagnetic metals. J. Phys. Chem. Solids 35, 947–956 (1974). 63. Freitas, P. P. & Berger, L. Observation of s–d exchange force between domain walls and electric current in very thin Permalloy films. J. Appl. Phys. 57, 1266–1269 (1985). 64. Slonczewski, J. C. Conductance and exchange coupling of two ferromagnets separated by a tunneling barrier. Phys. Rev. B 39, 6995–7002 (1989). 65. Stiles, M. & Miltat, J. in Spin Dynamics in Confined Magnetic Structures III (eds Hillebrands, B. & Thiaville, A.) (Springer, Berlin, 2006) 66. Sun, J. Z. Spin–current interaction with a monodomain magnetic body: a model study. Phys. Rev. B 62, 570–578 (2000). 67. Ralph, D. & Buhrman, R., in Concepts in Spintronics (ed. Maekawa, S.) (Oxford Univ. Press, 2006) 68. Huai, Y., Albert, F., Nguyen, P., Pakala, M. & Valet, T. Observation of spin-transfer switching in deep submicron-sized and low-resistance magnetic tunnel junctions. Appl. Phys. Lett. 84, 3118–3120 (2004). 69. Hayakawa, J. et al. Current-induced magnetization switching in MgO barrier based magnetic tunnel junctions with CoFeB/Ru/CoFeB synthetic ferrimagnetic free layer. Jpn. J. Appl. Phys. 45, L1057–L1060 (2006). 43

MARCH 2008 | S41

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

COLLECTION 70. Hosomi, M. et al. Novel nonvolatile memory with spin torque transfer magnetization switching: spin-ram. IEDM Tech. Dig. 459–462 (2005). 71. Kawahara, T. et al. 2Mb spin-transfer torque RAM (SPRAM) with bit-by-bit bidirectional current write and parallelizing-direction current read. ISSCC Dig. Tech. Papers, 480–481 (2007). 72. Jung, S. et al. Three dimensionally stacked NAND Flash memory technology using stacking single crystal Si layers on ILD and TANOS structure for beyond 30 nm node. IEDM Tech. Dig., 1–4 (2006). 73. Ito, K., Devolder, T., Chappert, C., Carey, M. J. & Katine, J. A. Micromagnetic simulation of spin transfer torque switching combined with precessional motion from a hard axis magnetic field. Appl. Phys. Lett. 89, 252509 (2006). 74. Devolder, T., Chappert, C. & Ito, K. Sub-ns spintransfer switching: compared benefits of free layer biasing and pinned layer biasing. Phys. Rev. B 75, 224430 (2007). 75. Sakimura, N. et al. A 512 kb cross-point cell MRAM. ISSCC Dig. Tech. Papers, 278–279 (2003). 76. Tanizaki, H. et al. A high-density and high-speed 1T4MTJ MRAM with voltage offset self-reference sensing scheme. Asian Solid-State Circuits Conf. Dig. Tech. Papers, 303–306 (2006). 77. Leuschner, R. et al. Thermal select MRAM with a 2-bit cell capability for beyond 65 nm technology node. IEDM Tech. Dig., 1–4 (2006). 78. Bowen, M. et al. Nearly total spin polarization in La2/ 3Sr1/3MnO from tunnelling experiments. Appl. 3 Phys. Lett. 82, 233–235 (2003). 79. Ishikawa, T. et al. Spin-dependent tunneling characteristics of fully epitaxial magnetic tunneling junctions with a full-Heusler alloy Co MnSi thin film 2 and a MgO tunnel barrier. Appl. Phys. Lett. 89, 192505 (2006). 80. Marukame, T., Ishikawa, T., Matsuda, K., Uemura, T. & Yamamoto, M. High tunnel magnetoresistance in fully epitaxial magnetic tunnel junctions with a full-Heusler alloy Co2Cr0.6Fe0.4Al thin film. Appl. Phys. Lett. 88, 262503 (2006). 81. Chiba, D., Sato, Y., Kita, T., Matsukura, F. & Ohno, H. Current-driven magnetization reversal in a ferromagnetic semiconductor (Ga,Mn)As/GaAs/ (Ga,Mn)As tunnel junction. Phys. Rev. Lett. 93, 216602 (2004). 82. Elsen, M. Spin transfer experiments on (Ga,Mn)As/ (In,Ga)As/(Ga,Mn)As tunnel junctions. Phys. Rev. B 73, 035303 (2006). 83. Gould, C. Tunneling anisotropic magnetoresistance: a spin-valve-like tunnel magnetoresistance using a single magnetic layer. Phys. Rev. Lett. 93, 117203 (2004). 84. Gould, C., Schmidt, G. & Molenkamp, L. W. Tunneling anisotropic magnetoresistance-based devices. IEEE Trans. Electron Dev. 54, 977–983 (2007). 85. Enaya, H., Semenov, Y. G., Kim, K. W. & Zavada, J. M. Electrical manipulation of nonvolatile spin cell based on diluted magnetic semiconductor quantum dots. IEEE Trans. Electron Dev. 54, 1032–1039 (2007). 86. LeClair, P. et al. Large magnetoresistance using hybrid spin filter devices. Appl. Phys. Lett. 80, 625–627 (2002). 87. Monsma, D. J., Lodder, J. C., Popma, T. J. A. & Dieny, B. Perpendicular hot electron spin-valve effect in a new magnetic field sensor: the spin-valve transistor. Phys. Rev. Lett. 74, 5260–5263 (1995). 88. van Dijken, S., Jiang, X. & Parkin, S. S. P. Room temperature operation of a high output current magnetic tunnel transistor. Appl. Phys. Lett. 80, 3364–3366 (2002). 89. Hehn, M., Montaigne, F. & Schuhl, A. Hot-electron three-terminal devices based on magnetic tunnel junction stacks. Phys. Rev. B 66, 144411 (2002). 90. Hubert, A. & Schäfer, R. Magnetic Domains (Springer, Berlin, 1998). 91. Allwood, D. A. et al. Submicrometer ferromagnetic NOT gate and shift register. Science 296, 2003–2006 (2002). 92. Allwood, D. A. et al. Magnetic domain-wall logic. Science 309, 1688–1692 (2005). 93. Cowburn, R. P. & Allwood, D. A. Multiple layer magnetic logic memory device. UK patent GB2,430,318A (2007).

S42 | MARCH 2008

nature

94. Parkin, S. S. P. Shiftable magnetic shift register and method using the same. US patent 6,834,005B1 (2004). 95. Cros, V., Grollier, J., Munoz Sanchez, M., Fert, A. & Nguyen Van Dau, F. Spin electronics device. Patent WO 2006/064022 (2006). 96. Tatara, G. & Kohno, H. Theory of current-driven domain wall motion: spin transfer versus momentum transfer. Phys. Rev. Lett. 92, 086601 (2004). 97. Li, Z. & Zhang, S. Domain-wall dynamics and spinwave excitations with spin-transfer torques. Phys. Rev. Lett. 92, 207203 (2004). 98. Grollier, J. et al. Switching a spin valve back and forth by current-induced domain wall motion. Appl. Phys. Lett. 83, 509 (2003). 99. Yamaguchi, A. et al. Real-space observation of current-driven domain wall motion in submicron magnetic wires. Phys. Rev. Lett. 92, 077205 (2004). 100. Ravelosona, D., Lacour, D., Katine, J. A., Terris, B. D. & Chappert, C. Nanometer scale observation of high efficiency thermally assisted current-driven domain wall depinning. Phys. Rev. Lett. 95, 117203 (2005). 101. Yamanouchi, M., Chiba, D., Matsukura, F. & Ohno, H. Current-induced domain-wall switching in a ferromagnetic semiconductor structure. Nature 428, 539–542 (2004). 102. Thiaville, A., Nakatani, Y., Miltat, J. & Suzuki, Y. Micromagnetic understanding of current-driven domain wall motion in patterned nanowires. Europhys. Lett. 69, 990–996 (2005). 103. Piechon, F. & Thiaville, A. Spin transfer torque in continuous textures: Semiclassical Boltzmann approach. Phys. Rev. B 75, 174414 (2007). 104. Himeno, A. et al. Dynamics of a magnetic domain wall in magnetic wires with an artificial neck. J. Appl. Phys. 93, 8430–8432 (2003). 105. Hayashi, M. et al. Dependence of current and field driven depinning of domain walls on their structure and chirality in permalloy nanowires. Phys. Rev. Lett. 97, 207205 (2006). 106. Allwood, D. A., Xiong, G. & Cowburn, R. P. Domain wall diodes in ferromagnetic planar nanowires. Appl. Phys. Lett. 85, 2848–2853 (2004). 107. Faulkner, C. C. et al. Artificial domain wall nanotraps in Ni81Fe19 wires. J. Appl. Phys. 95, 6717–6719 (2004). 108. Klaui, M. et al. Direct observation of domain-wall configurations transformed by spin currents. Phys. Rev. Lett. 95, 026601 (2005). 109. Klaui, M. et al. Current-induced vortex nucleation and annihilation in vortex domain walls. Appl. Phys. Lett. 88, 232507 (2006). 110. He, J., Li, Z. & Zhang, S. Current-driven vortex domain wall dynamics by micromagnetic simulations. Phys. Rev. B 73, 184408 (2006). 111. Saitoh, E., Miyajima, H., Yamaoka, T. & Tatara, G. Current-induced resonance and mass determination of a single magnetic domain wall. Nature 432, 203–206 (2004). 112. Thomas, L. et al. Oscillatory dependence of currentdriven magnetic domain wall motion on current pulse length. Nature 443, 197–200 (2006). 113. Thomas, L. et al. Resonant amplification of magnetic domain-wall motion by a train of current pulses. Science 315, 1553–1556 (2007). 114. Nakatani, Y., Thiaville, A. & Miltat, J. Faster magnetic walls in rough wires. Nature Mater. 2, 521–523 (2003). 115. Lim, C. K. et al. Domain wall displacement induced by subnanosecond pulsed current. Appl. Phys. Lett. 84, 2820–2822 (2004). 116. Hayashi, M. et al. Current driven domain wall velocities exceeding the spin angular momentum transfer rate in permalloy nanowires. Phys. Rev. Lett. 98, 037204. (2007). 117. Yamanouchi, M., Chiba, D., Matsukura, F., Dietl, T. & Ohno, H. Velocity of domain-wall motion induced by electrical current in the ferromagnetic semiconductor (Ga,Mn)As. Phys. Rev. Lett. 96, 096601 (2006). 118. Kasai, S., Nakatani, Y., Kobayashi, K., Kohno, H. & Ono, T. Current-driven resonant excitation of magnetic vortices. Phys. Rev. Lett. 97, 107204 (2006).

119. Cowburn, R. P. & Welland, M. E. Room temperature magnetic quantum cellular automata. Science 287, 1466–1468 (2000). 120. Imre, A. et al. Majority logic gate for magnetic quantum-dot cellular automata. Science 311, 205–208 (2006). 121. Ney, A., Pampuch, C., Koch, R. & Ploog, K. H. Programmable computing with a single magnetoresistive element. Nature 425, 485–487 (2003). 122. Black, W. C. J. & Das, B. Programmable logic using giant-magnetoresistance and spin-dependent tunneling devices. J. Appl. Phys. 87, 6674–6679 (2000). 123. Zhao, W. et al. Integration of Spin-RAM technology in FPGA circuits. Proc. ICSICT 799–802 (2006). 124. Min, B., Motohashi, K., Lodder, C. & Jansen, R. Tunable spin-tunnel contacts to silicon using low-workfunction ferromagnets. Nature Mater. 5, 817–822 (2006). 125. Hall, K. C., Lau, W. H., Gundogdu, K., Flatte, M. E. & Boggess, T. F. Nonmagnetic semiconductor spin transistor. Appl. Phys. Lett. 83, 2937–2939 (2003). 126. Hall, K. C. & Flatte, M. E. Performance of a spin-based insulated gate field effect transistor. Appl. Phys. Lett. 88, 162503 (2006). 127. Tanaka, M. & Sugahara, S. MOS-based spin devices for reconfigurable logic. IEEE Trans. Electron Dev. 54, 961–976 (2007). 128. Pasupathy, A. N. et al. The Kondo effect in the presence of ferromagnetism. Science 306, 86–89 (2004). 129. Sahoo, S., Kontos, T., Schonenberger, C. & Surgers, C. Electrical spin injection in multiwall carbon nanotubes with transparent ferromagnetic contacts. Appl. Phys. Lett. 86, 112109 (2005). 130. Hueso, L. E. et al. Transformation of spin information into large electrical signals using carbon nanotubes. Nature 445, 410–413 (2007). 131. Romeike, C., Wegewijs, M. R., Ruben, M., Wenzel, W. & Schoeller, H. Charge-switchable molecular magnet and spin blockade of tunneling. Phys. Rev. B 75, 064404 (2007). 132. Fert, A., George, J., Jaffres, H. & Mattana, R. Semiconductors between spin-polarized sources and drains. IEEE Trans. Electron Dev. 54, 921–932 (2007). 133. Kimura, T., Hamrle, J. & Otani, Y. Estimation of spindiffusion length from the magnitude of spin-current absorption: multiterminal ferromagnetic/ nonferromagnetic hybrid structures. Phys. Rev. B 72, 014461 (2005). 134. Dery, H., Dalal, P., Cywinski, L. & Sham, L. J. Spinbased logic in semiconductors for reconfigurable large-scale circuits. Nature 447, 573–576 (2007). 135. Khomskii, D. Multiferroics: Different ways to combine magnetism and ferroelectricity. J. Magn. Magn. Mater. 306, 1–8 (2006). 136. Zavaliche, F. et al. Electric field-induced magnetization switching in epitaxial columnar nanostructures. Nano Lett. 5, 1793–1796 (2005). 137. Zhao, T. et al. Electrical control of antiferromagnetic domains in multiferroic BiFeO3 films at room temperature. Nature Mater. 5, 823–829 (2006). 138. Chiba, D., Matsukura, F. & Ohno, H. Electric-field control of ferromagnetism in (Ga,Mn)As. Appl. Phys. Lett. 89, 162505 (2006). 139. Wunderlich, J. et al. Coulomb blockade anisotropic magnetoresistance effect in a (Ga,Mn)As singleelectron transistor. Phys. Rev. Lett. 97, 077201 (2006). 140. Kimura, T., Otani, Y. & Hamrle, J. Switching magnetization of a nanoscale ferromagnetic particle using nonlocal spin injection. Phys. Rev. Lett. 96, 037201 (2006).

Acknowledgments C.C. acknowledges support from the EU specific Support Action WIND (IST 033658). The authors also benefit from EU contracts Spinswitch (MRTN-CT-2006-035327) and Nanospin (STREP FET 015728). Correspondence and requests for materials should be addressed to C.C.

www.nature.com/milestones/spin

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

A

BEMaGS

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

F

*

12.040

We’re on the map: Nature Physics has its first impact factor.

Quite a landmark. * ISI Journal Citation Reports, Thomson, 2007

www.nature.com/naturephysics

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

Science news from a different angle From daily science updates to investigative journalism, from community commentary to editorial opinions, Nature brings you the most in-depth science news coverage online. Visit our new blog, The Great Beyond, for a comprehensive round-up and expert discussion of today’s science news from around the world. Subscribe to Nature and receive full access to Nature News as well as weekly issues of the most influential scientific journal in the world, in print and online.

www.nature.com/news

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

Nature Publishing Group Covering the spectrum in the physical sciences

VOL.7 NO.1 JANUARY 2008 www.nature.com/naturematerials

VOL.4 NO.1 JANUARY 2008 www.nature.com/naturephysics

Colossal predictions for platinum nanowires Charge order takes a slide

VOL.2 NO.1 JANUARY 2008 www.nature.com/naturephotonics

VOL.3 NO.1 JANUARY 2008 www.nature.com/naturenanotechnology

BIOTECHNOLOGY Nanomagnetic switches for cells

Waveguiding in photonic crystals

METEORITE IMPACTS A boon for a change

MAGNETOELECTRICITY New effects from first principles

CLIMATE CHANGE Expanding tropics

Branestorm

NAFION A clear diffusion model

QUANTUM OPTICS Few-photon fluorescence

DIAMONDOIDS Out of the sludge

SUPERCONDUCTIVITY Charge without spin QUANTUM MOLECULAR DYNAMICS Explosive simulations

Impact factor: 19.194*

Glacial water drives the flow OWEN FRACTURE ZONE Plate boundary revealed

LITHIUM BATTERIES Silicon nanowires show their potential

GLOBALIZATION OF RESEARCH Friendly competition

VOL.1 NO.1 JANUARY 2008 www.nature.com/naturegeoscience

Impact factor: 12.409*

Impact factor: 12.040*

SILICON PHOTONICS Signal regeneration SPECTROSCOPY Real-time Raman BIOLUMINESCENCE Firefly efficiency

Launched in January 2008

Launching 2009

Nature Publishing Group offers a range of journals in the physical sciences – from the broad scope of Nature to specific titles in physics, photonics, materials, nanotechnology, geosciences, chemical biology and, new for 2009, chemistry. Each journal carries a mix of research papers, reviews, news and views, and commentaries that keeps you up-to-date with the latest research in your area. Visit www.nature.com ____________ to: - register to receive table of content e-alerts - sign up for RSS feeds - enter a personal subscription - recommend the journals to your library

*Journal Citation Reports, Thomson, 2007

www.nature.com ________________

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F

Supplement to Nature Publishing Group

nature

Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page

A

BEMaGS F