H I G H L I G H T S O F I L L R E S E A R C H

Neutrons and magnetismA review of ILL research on magneticmaterials and phenomena

This booklet is the fourth of a series

devoted to the application of neutron

techniques in different research areas.

We have already published:

Exploring matter with neutrons (May 2000)

Neutrons and life (May 2001)

Neutrons and new materials (May 2002)

Neutrons and the Universe (May 2003)

The next one, to be published in 2005, will focus on the use of

neutrons to study soft matter.

1

or centuries, people have been aware of magnetic materials – theeffects of a magnetic force are easily observable. In everyday life,

magnets are found everywhere from cars, through mobile phones, tohigh-speed trains. But magnetic systems exist on all scales: elementaryparticles such as the electron and neutron act like tiny magnets; planetssuch as the Earth, Jupiter and Saturn (and once upon a time Mars, p.9)have powerful magnetic fields generated by massive dynamos at theircores. The Earth’s magnetosphere actually protects us from high-energysolar radiation so can be considered essential for life. The Sun and other stars have complex magneticstructures which generate powerful storms, and galaxies also reveal large-scale magnetic turbulence.

Back on Earth, for the past two centuries, scientists have been trying to understand the origin ofmagnetism. In the 19th century, the first theoretical concepts and experimental methods were developed andbulk properties of magnetic materials defined. In the 20th century, as understanding grew of the nature ofmagnetic interactions at the microscopic level, new kinds of magnetic behaviour were proposed. Thesepredictions were not confirmed until the mid-1950s with the development of neutron diffraction experiments.

Because of its elementary magnetic moment, the neutron can probe the magnetic properties of materialsat the atomic level and even down to the nuclear level: it acts as a tiny compass exploring the inner structureof matter. It can detect and characterise the vibrational motions of the individual magnetic moments. Fromthe characteristics of the vibrational modes, scientists and engineers can determine the interactions betweenthe local moments, which are so important for the bulk properties of magnetic materials. In this way they havebeen able to optimise and develop more efficient magnets for technological applications.

Louis Néel, who won the Nobel Prize for his seminalcontributions to our understanding of magnetism, was one of thefounding fathers of ILL. And so, research in magnetism has playeda major role at ILL right from the beginning. Expertise in methodsand techniques was developed in-house early on, bringing ILL tothe forefront of research in this area.

The ILL is now engaged in an ambitious programme ofinstrument renewal to prepare for future experimental challenges.It is our responsibility to ensure that the laboratory offers theadequate tools needed by the scientific community to develop thematerials to be used in tomorrow’s technology. Thanks to theseefforts, neutrons are seen to contribute to a deeper understandingof properties of matter and to the growth of advanced technology.The ILL can be proud of its achievements in the study ofmagnetism with neutrons.

1

N E U T R O N S A N D M A G N E T I S M

F

Foreword

Dr Christian VettierAssociate DirectorHead of Science Division

Louis Néel (1904-2000) played a major role in the birthand the international success of research in Grenoble.A 2-day workshop, organised in November 2004commemorates the 100th anniversary of his birth

AIP

Em

ilio

Seg

rè V

isu

al A

rch

ives

33

Contents

N E U T R O N S A N D M A G N E T I S M

1

4

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

22

Foreword Christian Vettier

Introduction Gerard Lander, Christian Vettier and Roland Currat

M A G N E T I C S T R U C T U R E S

The power of polarised neutrons Jane Brown

Orbital order in manganites Tapan Chatterji

Animal magnetism Andrew Wills and Eddy Lelièvre-Berna

How Mars lost its magnetism Bachir Ouladdiaf

Magnetism chills out Clemens Ritter

M A G N E T I C I N T E R A C T I O N S

The vanadium triangle Grégory Chaboussant

Molecules seize the moment Dante Gatteschi

Two of a kind Jiri Kulda and Stephen Hayden

Spin chains with a chiral twist Jiri Kulda and Hans-Benjamin Braun

A P P L I C A T I O N S

Towards better recording media Thomas Thompson, Stephen Lee and Charles Dewhurst

Superconductivity: resistance is useless Robert Cubitt and Charles Dewhurst

Inside modern magnets Gerard Lander

Reflections on magnetic multilayers Valeria Lauter

A new way to study magnetic multilayers Vincent Leiner

Glossary

Contacts

eople have been exploiting magnetism for morethan a millennium. The ancient Greeks discovered

that certain stones – lodestones – could attract iron.This magnetic force could then be imparted to an ironobject by rubbing it with the stone. The Chineseshowed that magnetised needles pointed north-southwhen suspended. Such magnetic compasses havebeen guiding ships across the seas ever since.

Today, magnetic materials have many more uses.Apart from obvious applications as fridge magnets anddoor catches, they form the central components ofmotors used in transport and electrically-drivenmachinery, and are also found in communicationdevices from telephones to loudspeakers. Extremelypowerful magnets are used in hospital scanners, and inbasic research – in huge machines to accelerate andstudy subatomic particles.

One of the most significant technologicaldevelopments over the past decades has been theexploitation of magnetic materials in storinginformation as minute magnetised particles – first ontapes and then on computer disks. Indeed, the rapidgrowth of our information-based economy has beendriven not only by the development of ever smallerelectronic devices (the latest 10 gigahertz chips willhave transistors 30 nanometres in size) but also ofmore densely packed magnetic computer memories

(MRAM devices now include memory cells of 1.4 squaremicrometres). The next generation of computer chips,as well as memory devices, could well be based onnew, exotic magnetic materials, designed andfabricated at the ‘nano’ scale of atoms and molecules.

Understanding the magnetic forceThese advanced ideas depend on a detailedunderstanding of the magnetic force at the most basiclevel. During the past two centuries, scientists haveelucidated its subtle nature. Hans Christian Oersted inDenmark and Michael Faraday in the UK saw the linkbetween electricity and magnetism in the early 19thcentury, which led Scottish physicist James ClerkMaxwell, shortly afterwards, to derive his brilliantequations explaining mathematically the combinedelectromagnetic force.

In the early 20th century, our insights into magneticinteractions advanced rapidly with the development ofquantum theory, which built on the ideas of Maxwellamongst others. Pioneers of quantum mechanics,Werner Heisenberg, Paul Dirac and Wolfgang Pauli,were able to explain magnetism as an inherentproperty of the electron. According to their ideas,particles like the electron have a quantum propertycalled spin, which carries a half-integer value. Theeffect is that of a bar magnet, or magnetic moment,which can be flipped up or down to give a value of +1/2 or –1/2.

Everyday matter is made up of atoms containingelectrons, so what makes some materials magnetic andnot others? Electrons in atoms and molecules tend topair off so that their magnetic moments of +1/2 or –1/2cancel, leaving no residual magnetism. However, somemetals and metallic salts (and quite a few molecularmaterials) contain unpaired electrons. In crystallineform, the atoms or molecules carrying these unpairedelectrons are arranged in a regular geometric way on alattice. If the electrons can interact with each other (viathe ‘exchange’ energy, as predicted by quantum theory)their moments will align to give a bulk magnetic effect.

In the familiar permanent magnet, such as iron, themoments are aligned in the same direction generatinga north and south pole. Not surprisingly, this is calledferromagnetism. Ferromagnetic materials include iron,

P

Investigating magnetism with neutrons

Studies of magnetism with neutrons is a vibrant research area, which is

underpinning a new generation of electronic devices as well as leading

to a deeper understanding of Nature at a fundamental level

4

N E U T R O N S A N D M A G N E T I S M

>> G E R A R D L A N D E R ,C H R I S T I A N V E T T I E R A N D

R O L A N D C U R R A T

The IN14instrument

cobalt and nickel, and a range of metallic compoundsinvolving heavier elements – lanthanides and actinides.Above a certain temperature called the Curietemperature, the thermal energy causes the alignmentsto break up. In a few materials, the moments align inalternately opposite directions – an effect calledantiferromagnetism. This phenomenon was firstproposed by Louis Néel working in Grenoble, and thetemperature below which a substance becomesantiferromagnetic is named after him.

We now know of many materials where themagnetic ordering is a good deal more complicated,depending on how the moments interact, as some ofthe examples in this booklet show. The ordering mayvary according to direction – especially in materialswith layers or chains of differing atoms. The momentsmay be canted to each other, or form a helicalarrangement (p.14). In electron-rich molecules withunpaired electrons the moment may be spread outover several atoms (p.12). Single molecules consistingof clusters of metal atoms, each with one or severalunpaired electrons, can form ‘nanomagnets’ where thespins combine and interact in fascinating ways (p.11).Because electrons also orbit the atomic nucleus, theymay produce an additional magnetic contribution dueto the moving electric charge, and this can furthercomplicate the magnetic interactions.

Studies of these exotic magnetic properties involvea strong interplay between theory and experiment.Models of magnetic interactions often have relevanceto other physical systems which can be described usinga similar underlying mathematical approach. Exploringmagnetic behaviour thus not only leads to newapplications but also plays an important role indemonstrating the universality and beauty of the lawsof Nature.

Experimental studiesMany of the models were developed, by Néel andothers, before the experimental tools – or the materials– existed to test them out. It was not until beams ofneutrons became available from nuclear reactors in the1940s that it became possible to study in detail howmagnetic moments are arranged in a material.

Neutrons, like electrons, have a spin 1/2, so have a

magnetic moment. They can themselves thereforeinteract with unpaired electrons in a material, and canbe scattered by the lattice of magnetic moments. Theresulting diffraction pattern provides informationabout the magnetic structure, and even the dynamicsof the interactions.

The theory of neutron scattering from magneticsolids had already been developed by Otto Halpernand Montgomery Hunt Johnson at New York Universityin a famous paper in 1939. A decade later, Cliff Shulland Ernie Wollan successfully applied the ideas usingthe reactor at Oak Ridge National LaboratoryTennessee. Neutrons have turned out to be the perfecttool for probing magnetism at a microscopic level. Themodels proposed by Néel were verified and newperspectives started to be explored, leading to the richfield of magnetism studies we have today, in which ILLplays a major part.

New techniques combined with ever improvinginstrumentation have led to a wide range ofexperiments. Scattering at small angles providesstructural information on a larger scale – that ofdomains of different magnetisation in a material.Neutrons can also be reflected off surfaces andinterfaces, allowing researchers to look at structuresbased on ultrathin magnetic layers, to be used in‘spintronic’ devices in which information is carried bythe electron’s spin rather than by its mobility (p.18).

The availability, now, of polarised neutrons (p.6)means we can pinpoint magnetic information inexperiments much more precisely, allowing us to lookat other exotic materials. These include unusualsuperconducting materials, in which the stronginteractions between electrons are reflected in themagnetic behaviour (p.13), and molecules with unusualmagnetic structures. It may soon be possible to followthe paths of single electrons in biological systems asrevealed through their magnetism (p. 8).

Neutron magnetic scattering remains a burgeoningfield of essential technical and theoretical importance.This booklet highlights the wide range of research onmagnetism tackled at the ILL.

5

I N T R O D U C T I O N

The EVAinstrument

The power of polarised neutrons

Magnetically aligned neutron

beams are an excellent probe

of magnetic materials

N E U T R O N S A N D M A G N E T I S M

>> J A N E B R O W N

antiparallel to those in the inner ones (total spin, -6).Polarised neutron scattering confirmed this picture, andthe magnetisation values at the various manganesesites agree well with those predicted theoretically.

Polarised neutron diffraction has also been used topinpoint the unpaired electrons responsible formagnetism in an organic compound containing nometals. It consists of a fluorocarbon ring attached to aring of one carbon, two sulfur and two nitrogen atoms.The unpaired electrons are found on the latter, five-membered ring (figure 2).

Polarisation analysisAdding a second flipper and a polarisation analyserbetween the sample and detector allows us to carryout measurements that specifically distinguish magneticscattering from the normal nuclear variety. Data fromfour combinations of measurements made by switchingeach flipper can be obtained: the intensities of scatteredpolarised neutrons both parallel and antiparallel to theincident polarised neutrons, and the same with theincident polarisation reversed. This technique can beused to study antiferromagnetism in which the directionof magnetic moments in a crystal alternate.

More sophisticated measurements of the magnitudeand direction of the scattered polarisation for variousorientations of the incident polarisation will give theabsolute magnetic configuration of materials such asmagneto-electric crystals (in which an electric fieldinduces magnetisation or vice-versa). For example,such polarimetric experiments on the magneto-electriccrystal chromium oxide (Cr2O3) determine the absoluteorientation, relative to the surrounding oxygen atoms,of the oppositely directed moments on pairs ofchromium ions. Figure 3 shows the configurationstabilised by cooling in different combinations ofelectric and magnetic fields.

6

Unpolarised white beam

Polarised monochromaticbeam

Magnetisedmonochromatingcrystal Magnetised

collimators Sample magnetised by superconducting magnet

Spin flipper

Neutron detector

Cr3+

O2-

E H E H

N N

S S

C

oth neutrons and electrons have a spin of 1/2,so have a magnetic moment and can interact

magnetically with matter. The strength of theinteraction depends not only on the size of theelectronic magnetic moments, but also on their relativeorientations. Beams of neutrons with all their spinsparallel – a polarised beam – can therefore be used tostudy materials in which the electronic magneticmoments are ordered in a structurally interesting way.The intensities of the neutrons scattered by the sample(usually a single crystal), are measured in a detector.

However, the best way to extract preciseinformation about the sample’s magnetic properties isto compare scattering patterns obtained with neutronbeams polarised in opposite directions. A device calleda spin flipper reverses the polarisation of the incidentneutron beam (figure 1), so that the ratio between thescattered intensities for the two orientations can bemeasured – the flipping ratio. Precise measurements ofthis ratio can be used to map the distribution ofmagnetic moments within a crystal – as was shown byClifford Shull and his students at MIT in pioneeringexperiments in the 1960s investigating the classicferromagnetic metals iron, cobalt and nickel. Theydemonstrated that although the electrons responsiblefor the magnetism are also involved in conductingelectricity they are well-localised in space.

Since then, this method has been used to look atmagnetisation in many materials. One of the moreexotic compounds studied is a large moleculecontaining 12 manganese atoms linked to acetategroups, commonly known as Mn12-Ac, and which hasunusual low-temperature magnetic properties. Themolecule contains an inner core of four manganeseions, each with three unpaired electrons, surrounded byan outer ring of eight manganese ions with a total of32 unpaired electrons. The overall spin of the complexis obtained by summing all the magnetic moments inthe molecule. Magnetisation measurements give a netspin of 10, which suggests that the spins on the outermanganese atoms (total spin,16) are oriented

BFigure 1A simple polarised neutrondiffractometer. The neutronsfrom a reactor source arereflected by a magnetised crystalwhich selects neutrons with aparticular wavelength andpolarised parallel to themagnetisation direction. Theseneutrons are then scattered by amagnetised single crystalsample and the intensities of thereflected neutrons are recorded

Figure 2 The magnetisation distributionin the CNSSN ring of the organicmagnet, p-O2NC6F4CNSSN

A crystal for polarisingneutrons on instrument IN20

Figure 3 The crystal structure of chromium oxide showing themoment orientations stabilised by cooling in (a) paralleland (b) antiparallel magnetic (H) and electric (E) fields

D3 diffractometer used forpolarised neutron diffraction

ompounds called manganites – derivedfrom a structure consisting of a rare-earth

element such as lanthanum combined withmanganese and oxygen (LaMnO3) – have causedgreat excitement in the past decade becausethey show a huge change in resistivity when amagnetic field is applied. This colossalmagnetoresistance (CMR) effect could be the key to the next generation of magnetic memory devices,magnetic-field sensors, or transistors.

The key to this CMR effect is the manganites’complex electronic and magnetic structure, whichdepend subtly on how the outer ‘d’ electrons of themanganese atoms are arranged in their orbitals. In theparent lanthanum manganite, the triply-chargedmanganese atom (Mn3+) is bound to six neighbouringoxygen atoms in an octahedral shape (figure 1a). TheMn3+ ion has one electron involved in bonding; itoccupies one of two possible d orbitals, which aregeometrically different and are associated withdiffering bond lengths. The octahedral structure is thusactually slightly distorted. The lanthanum ions, alsotriply charged (La3+), sit between layers of manganese-oxygen octahedra.

If a percentage of the lanthanum ions is replacedwith doubly-charged metal ions such as strontium(Sr2+), the electrons redistribute to give some Mn4+ ionswith an empty d orbital (a hole). This ‘hole-doping’allows the remaining electrons to hop from manganeseto manganese atom so that the material conducts likea metal. Furthermore, when the 30 or 40 per-cent hole-doped manganite is cooled below a certaintemperature, the spins of these electrons align so thatthe material becomes ferromagnetic. It is this thatincreases the resistivity of the material because thealigned electrons scatter oppositely-aligned electronstrying to pass through.

Using a combination of X-ray and neutrondiffraction, we investigated the interplay between theordering of the spins, the charge and the orbitaloccupancy in the 50 per-cent hole-doped, layeredmanganite (LaSr2Mn2O7). Neutron studies enabled usto measure the manganese-oxygen bond lengths,which indicated which d orbitals were occupied. Wefound that this compound has equal numbers of Mn3+

and Mn4+ ions arranged alternately, and a ‘staggered’arrangement of bonding orbitals (figure 2) below 225K.At 170K this charge-orbital order ‘melts’ asantiferromagnetic order sets in.

An unusual transitionMore recently, we have investigated the orbitalordering further, in the parent lanthanum manganiteover a range of temperatures. The bonding d orbitalsare again arranged in an alternate staggered patternwithin each plane of manganese-oxygen octahedra(figure 1b), but unlike in LaSr2Mn2O7, all the manganeseions are Mn3+ and have a bonding d orbital. Theoctahedra are distorted giving three different bondlengths. The compound retains this structure up to750K – when something dramatic happens. Thedistortion disappears, the staggered arrangement oforbitals is lost, and the solid material abruptly contractsin volume. Such a transition in a solid, which dependspurely on changes in the electronic arrangement, israther unusual. We think that in the high-temperature,orbitally disordered state all the octahedra packtogether more efficiently. This is similar to whathappens when ice melts into water, although in thatcase the contraction is due to the water moleculeslosing positional order.

7

Combined neutron and X-ray

diffraction has revealed a new type of

electronic behaviour in an important

class of magnetic materials

Orbital order inmanganites

M A G N E T I C S T R U C T U R E S

Research team T. Chatterji, G. J. McIntyre, A. Stunault and B. Ouladdiaf (ILL), and F. Fauth (ESRF, Grenoble)

(a)

La,Sr(1) La,Sr(1)

Mn3+

La,Sr(1')1.92 Å 1.96 Å

1.92 Å

1.91 Å

1.96 Å

1.92 Å

O(3) O(3')

Mn4+

O(3')

La,Sr(1')

C

>> T A P A N C H A T T E R J I

Figure 1(a) The structure oflanthanum manganite;(b) the staggeredordering of d orbitals

Figure 2The structure of 50 per-cent hole-dopedlanthanum strontium manganite(LaSr2Mn2O7), at a temperature of 165K,looking down through the manganese-oxygen octahedra. Neutron diffractionallowed us to determine the position ofboth manganese and oxygen atoms, andthus the Mn-O bond distances. Thisrevealed the ordering of the Mn3+ andMn4+ ions (a) and the ordering of thebonding orbitals (b)

(b)

a

b

eactions in biological systems, such as energy-producing respiration, rely on a helping hand

from enzymes containing metal atoms. These metallo-proteins catalyse the transfer of electrons in areduction-oxidation (redox) mechanism. A typicalexample is cytochrome c oxidase (above). Suchmolecular assemblies are large and complex, and thedetails of the redox process are still uncertain;conventional studies using X-ray scattering, resonanceor spectroscopic techniques are complicated by thepresence of the many atoms in the enzyme’sframework and the ambiguity of the data.

Polarised neutrons, however, have the potential toprovide the information to unravel these mechanisms.Most enzyme-catalysed redox reactions involve thetransfer of single electrons (each with a magneticmoment). They can therefore be located and followed,via the accompanying magnetic changes in themolecular system, using polarised neutron diffraction(p.6), without the interfering effects of the multitude offramework atoms.

In fact, the main obstacle in using neutron diffractionis that proteins such as cytochrome c oxidase aredifficult to prepare in the required single-crystal form.For this polarised neutron technique to be used,it must first be modified to deal with powder andpolycrystalline samples.

Powder diffraction is a well-established method ofstudying such materials. Instead of measuringindividual reflections in three dimensions, as happensin single-crystal diffraction, the data are obtained as aone-dimensional set of complex, overlapping peaksfrom which information has to be extricated by acomputer. In proteins, the complexity and large size ofthe protein structure amplifies the overlapping peakproblem, so requires an optimised powderdiffractometer.

A polarisation filterThe main technical difficulty to overcome is providingan intense-enough source of polarised neutrons over awide wavelength range. Traditionally, ferromagneticsingle crystals (p.6) are used, which produce polarisedbeams only at specific wavelengths. A new polarisationtechnique has recently been developed at ILL, whichhas a broader application and can be used with ‘white’beams of pulsed neutrons (as provided by spallationneutron sources), as well as the continuous neutronsource from a nuclear reactor such as ILL’s.

The method is based on spin-polarised helium-3which absorbs neutrons of one spin type, effectivelyacting as a spin filter. The device is compact (only a fewcentimetres long) and convenient. It is ideal for usingwith powder diffractometers, and will be suitable forexperiments using the next generation of spallationsources that aim to follow the path of redox reactionswith timed neutron pulses.

In June 2003, we carried out the first experiment todetermine the applicability of this approach using thepowder diffractometer D1B at the ILL. Data collectedfrom a variety of organometallic and intermetallicmaterials clearly indicate that experiments can bedesigned to detect unpaired electrons in biologicalsystems, thus heralding a new way of probingfundamental questions in biology – using magnetism.Chemists should also be able to apply the sametechnique to design novel redox catalysts.

8

Developing polarised neutrons to

reveal the magnetic side of proteins

N E U T R O N S A N D M A G N E T I S M

Research team A.S. Wills (University College London and The Royal Institution of Great Britain) and E. Lelièvre-Berna (ILL)

>> A N D R E W W I L L S A N D E D D Y L E L I È V R E - B E R N A

R

Animal magnetism

The experimental teamon instrument D1B

The active site ofcytochrome c oxidase

The helium-3 cellused as a source ofpolarised neutrons

ur planetary neighbour Mars is currently excitinggreat interest as various space missions explore

its surface and atmosphere. One of the key areas ofinterest is Mars’ magnetic field and what happened toit in the distant past. Researchers believe that Marsonce had a global magnetic field, like Earth’s, but theiron-core dynamo that generated it shut down billionsof years ago leaving behind only patches of magnetismdue to magnetised minerals in the Martian crust.

NASA’s Mars Global Surveyor spacecraft mappedthese magnetic left-overs in 1999. The results weremysterious. While crustal magnetisation in thenorthern lowlands was negligible, it was much strongerin the older highlands of the southern hemisphere –with an especially strong magnetic field in a stripbetween longitudes 130°E and 240°E, but almost nomagnetisation around the giant impact craters ofHellas and Argyre (figure 1).

The north-south divide probably arose because,when the Martian dynamo shut down, the younger,thinner northern crust had not yet solidified and thusbecame demagnetised – while the solid southerncrust’s magnetisation was already ‘frozen’ in. However,there is no apparent difference between themagnetised and non-magnetised areas in the southernterrain. One explanation is that the Hellas and Argyreimpacts occurred after the dynamo shut down, andthat the accompanying pressure shock-waves (ratherthan heat) demagnetised the crust around those areas.

Possible evidence for this scenario lies with one ofthe minerals identified in Martian meteorites ascarrying the remanent Martian magnetisation –pyrrhotite, Fe7S8. This mineral is ferromagnetic below atransition temperature of 325°C. This is due toalternate stacking of partially and fully-filled iron layersin the mineral which, although coupledantiferromagnetically, results in a residual magneticmoment. At moderately high pressures, however,pyrrhotite loses this magnetism, but the exact pressureat which this happens was not known.

9

How Mars lost its magnetism

Research team B. Ouladdiaf (ILL), R. Ballou and G. Fillion (Louis Néel Laboratory, Grenoble, France), P. Rochette (CEREGE University of Aix, France) and L. Hood (University of Arizona, USA)

O

>> B A C H I R O U L A D D I A F

M A G N E T I C S T R U C T U R E S

Pyrrhotite under pressureWe decided to pin down the pressure-demagnetisationrelationship more precisely using neutron diffraction.Samples of natural pyrrhotite were placed in pressure-clamp cells of up to 2 and 3 gigapascals (GPa), usingdeuterated ethanol as a medium to transmit evenpressure, and placed on diffractometers D1B and D20respectively. Two sets of peaks were isolated as beingfrom the mineral; one set could be ascribed tomagnetic scattering since it disappeared above thetransition temperature (in an independent experiment,figure 2). The study then showed that this magneticpeak decreases with pressure and vanishes between2.6 and 3.1 GPa, figure 3. An independent experimentbased on a remanence measurement after pressure-release led to the same result.

Calculations indicate that the shock waves fromimpacts on the Martian surface would have indeedresulted in a maximum pressure contour of 3 GPa atthe distance that coincides with the boundary betweenmagnetised and non-magnetised crust. Thisstrengthens the case of pyrrhotite as the mineralresponsible for the Martian remanent magnetic field.

Besides nicely predicting the global magnetisationmap, the results also have amajor implication for thepaleomagnetic signal of theMartian meteorites. All Martianmeteorites have been shockedto pressures above 3 GPa, andsince their remanence is alsocarried by pyrrhotite, theywould have been demagnetisedwhen they hit the Earth.Therefore, their paleomagneticsignal postdates shock.

350

20

36

52

450

550

650

Scattering angle Temperature (K)

Inte

nsity 50

100

Figure 2Part of the neutron scatteringpattern collected on D1Bfrom the pyrrhotite powdersample at room pressure andversus temperature

Figure 3 The neutron scattering patternversus pressure: (a) showinghow the intensity of themagnetic peak goes down withpressure; and (b) its evolutionwith increasing pressure, asmeasured on D20

Figure 1The topography of Mars showingthe computed peak shock pressurein GPa around the Hellas and Argyreimpacts (grey), and the 20 and 40nanotesla contours of the magneticfield (black) measured by MarsGlobal Surveyor at an altitude of380 to 420 kilometres

0°

60°

30°

0°

-30°

-60°

30° 60° 90° 120°

-8 -4 0 4 8km

12 16

150° 180° 210° 240° 270° 300° 330° 360°

Pressure (GPa) Scattering angle

(001) peak

Inte

nsity

00.0

0.5

1.0

1 2 3 23 24 25 26

3.1 GPa

2.6 GPa

2.1 GPa

0.95 GPa

D1BD20

Mars has a complex and patchy

magnetic field; neutron experiments

may explain why

hen a magnetic field is applied to a magneticmaterial, the constituent magnetic moments

align, and the material warms up. When the field isswitched off, the moments become disorderly againand the material cools. In certain alloys, thismagnetocaloric effect is substantial – large enough tosuggest a novel type of refrigeration based on cyclingthe magnetic field-switching and allowing the heatgiven off to escape. Companies are already interestedin applying this environmentally-friendly technology toair-conditioning and supermarket chillers.

The giant magnetocaloric effect was first discoveredin 1997 in the gadolinium silicon germanium alloy,Gd5(Si1-xGex)4, and has since been found in similarmetal compounds. A field of 1 tesla (20,000 times theEarth’s field) can produce a change of 3 to 4 degrees inthese materials, and depending on the proportion ofgermanium contained, the effect can be tuned to appearacross a range of temperatures between 20 and 295K.

The extent of the magnetocaloric effect is believedto depend on an intimate coupling between thematerial’s crystal structure and its magnetism. X-rayand magnetisation measurements indicate that as thetemperature drops, a strong magnetic moment

appears alongside a change incrystal structure from monoclinic toorthorhombic. Figure 1 shows howbonds between silicon orgermanium atoms of every secondlayer break, allowing the layers toshift sideways to a new structure.

Changing phasesThe obvious question is whether the structural transitionstrictly coincides with the onset of magnetism, and if sowhich triggers which. We set about finding out usingneutron diffraction. We chose a variant of the alloycontaining terbium rather than gadolinium becausethe latter strongly absorbs neutron energy; the terbiumcompound still shows the magnetocaloric effect.

Starting at a temperature of 110K, well above themagnetic transition temperature, we found thatapplying a field as high as 5 tesla induced little changefrom the monoclinic to the orthorhombic structure(figure 2). Even when lowering the temperature to105K, only half the sample changed. But what wassurprising was that the magnetic scattering patternobtained indicated that the remaining monoclinicphase had become magnetic as well.

To explore this unexpected magnetic phase, wedecided to do a slow and detailed scan with an intenseneutron beam, in a zero field, lowering the temperatureat a rate of about 2 degrees an hour. The resultingdiffraction pattern in figure 3 shows the magneticpeaks for the monoclinic (M) and orthorhombic (O1)structures, revealing that magnetism first emerges inthe former and only at about 104K does the latterappear. The monoclinic peaks disappear at about 92K,marking completion of the structural transition. Wealso found that the monoclinic phase was largelyferromagnetic, similar in nature to the magneticstructure of the orthorhombic phase. Neutron studiesare the only way that could distinguish between thetwo magnetic phases.

Clearly, these results are significant in understandingthe magnetocaloric effect. The magnetic phase transitionhappens at a higher temperature (about 112K) than thestructural phase transition (about 105K). Even an appliedfield of 5 tesla could not induce the change of morethan 50 per cent in the crystal structure, showing thatthe interplay between crystal structure and magneticproperties is more complex than previously thought.

10

Magnetism chills out

N E U T R O N S A N D M A G N E T I S M

Research team C. Ritter (ILL), L. Morellon, C. Magen, P. A. Algarabel and M. R. Ibarra (University of Zaragoza, Spain)

>> C L E M E N S R I T T E R

WFigure 1 The transition frommonoclinic to orthorhombicstructure in the alloy,Tb5Si2Ge2. The red balls aresilicon or germaniumatoms, yellow ballsrepresent terbium atoms.The arrow marks where astrong shear movementleads to breaking of thesilicon or germanium bonds

Figure 3Neutron diffraction contour plot ofTb5Si2Ge2 in a selected angular andtemperature range as recorded on thehigh-intensity D1B diffractometer. Thediffraction peaks marked are of purelymagnetic origin coming from themonoclinic (M), the orthorhombic (O1),or both crystallographic structures

0 1 2 3 4 5

Perc

enta

ge o

f m

onoc

linic

stru

ctur

e

Applied field (tesla)

120K

115K

110K

105K

100K

10

0

20

30

40

50

60

M+O181.202

92.05

103.028

113.958

124.995

Tem

pera

ture

(K)

O1 M

Figure 2The percentage of monoclinic phase induced byincreasing the magnetic field at selected temperaturesas determined from high-resolution D2B data

Materials that behave as

magnetic refrigerants have

surprisingly complex structures

11

M A G N E T I C I N T E R A C T I O N S

Research team G. Chaboussant, S. T. Ochsenbein, A. Sieber, R. Bircher and H.-U. Güdel (University of Bern, Switzerland),B. Barbara (Louis Néel Laboratory, Grenoble, France), A. Müller (University of Bielefeld, Germany) and H. Mutka (ILL)

olecules that contain a cluster of magnetic metalatoms are fascinating physicists and chemists

alike because of the surprising ways in which theconstituent spins can interact. Such a molecule is aheavily hydrated complex of 15 vanadium atoms(K6[V15As6O42(H2O)]·8H2O). Each vanadium ion has aspin 1/2. However, the spins arrange themselvesantiferromagnetically to give a total molecular spin ofonly a 1/2 in the lowest energy, or ground state.

The reason behind this weak magnetic effect isshown in figure 1. The vanadium atoms are arrangedas two hexagons with a triangle sandwiched between.The spins in the separate hexagons form stronglycoupled non-magnetic pairs, which also interact weaklywith individual vanadiums of the central triangle, sothat these vanadiums are also weakly coupled witheach other. At very low temperatures (2 to 3K), thesystem is effectively a magnet made up of threeindirectly interacting spins. Since at least two of thesespins are always forced to be parallel, while they wouldprefer to be antiparallel, the system is said to be‘frustrated’. It also means that there are two possiblelowest quantum energy levels for the molecule,depending whether the spins are up or down.

Magnetisation experiments exploring this couplinghave shown that the situation is actually morecomplicated. First, increasing and then reversing anapplied magnetic field at different rates produces apeculiar kind of hysteresis effect (the lag inmagnetisation/demagnetisation of the moleculebehind the changing field) which suggests that thealignment of spins is affected by the vibrations of thecrystal lattice (phonons). Secondly, at zero field the twoground states are not equivalent but differ by a smallamount of energy.

Inelastic neutron scattering While magnetisation experiments have successfullyexplored the phonon-coupling phenomenon, high-resolution neutron scattering, combined with magneticfields, offers the best method to investigate the originof the split ground state. When neutrons scatter offmolecules they can exchange energy – inelasticscattering – so that the atoms jump to another energystate, for example, by changing spin direction.

Measuring the energies of theinelastically scattered neutronsthus gives the energy of thetransition and identifies thequantum state.

By investigating the varioustransitions at different neutronwavelengths and magnetic fields,we could determine the energiesof the vanadium cluster’s lowestmagnetic states. First, however,the hydrogen atoms of the water molecules in thevanadium complex had to be replaced by deuterium toreduce the ‘background’ from the strongly scatteringhydrogens. The sample was cooled to between 40 and50 millikelvin and the experiment was carried out withmagnetic fields going up to 1 tesla.

Figure 2 shows the main experimental results atdifferent wavelengths. They reveal five types ofinelastic transitions, which can be explained by threesets of unequal couplings betweenthe vanadiums making up thetriangle of antiferromagnetic spins.There is a tiny but significantenergy gap of 30 microlectronvoltsbetween the two lowest states.The gap is most likely induced bythe fact that the vanadium triangleis spatially distorted as a result ofone water molecule sitting insidethe cluster and disordered watermolecules in the crystal lattice.

The data also suggest that theground states are ‘entangled’ – a property of quantum systems inwhich states are intrinsically linked.Entangled states are of greatinterest in the burgeoning field ofquantum computing as a futureway of encoding and processinglarge amounts of data.

M

Figure 1 Molecular structure ofthe vanadium (V15)cluster showing thethree vanadium ions(light blue balls)

The vanadium triangle

Single molecule magnets can display unusual

magnetic interactions in which their local

environment plays a significant role

>> G R É G O R Y C H A B O U S S A N T

(I)+(II)

(I)(II)

0.65 T

0.5 T

Energy (meV)0.1

0.0

0.1

0.2

0.340 60 80 100 120 140

0.4

0.2 0.3 0.4 0.5

0 T

0.5 T

0.75 T

1.0 T

(IV)

(V)

(III)

Inte

nsity

λ =7.5Å

λ =9.0Å

λ =11Å

E (µeV)

Figure 2 Inelastic neutron scattering spectra at different wavelengthsand applied magnetic fieldsshowing the five transitions, I, II,III, IV and V. The two lowest-energypeaks are shown in the inset

The IN5 instrument

Molecules seizethe moment

Molecules with

unpaired electrons

offer a new type of

magnetic material

agnets based on molecular building blockshave the potential to produce materials with

novel characteristics of both theoretical and practicalinterest. The advantages of molecular magnets are thatthey can be altered by fine-tuning their structurechemically, they are soluble in organic solvents so canbe processed easily, and they may interact with light ina technologically useful way. One intriguingapplication is in the new field of quantum computingwhich relies on encoding information in superposedquantum states.

Contrary to classical magnets, where themagnetisation density is localised on the atoms ofmetallic lattices, in molecular magnets the density maybe spread across several of the constituent atoms. It isoften difficult to understand the magnetic interactionwithout a detailed knowledge of how thismagnetisation density is distributed. Here, polarisedneutron diffraction techniques (p. 6) provide the mostsignificant results.

Organic magnets The first materials investigated in this way were organicmolecules containing a nitrogen-oxygen single bondcarrying an unpaired electron – a nitroxide radical (tobe magnetic, molecules must have unpaired electrons).Normally, radicals are very reactive, as the unpairedelectron tries to seek a mate to form a stable chemicalbond, but the radical can be stabilised by protectingthe nitroxide bond with surrounding bulky groups ofatoms. The first genuine ferromagnet made, with nometal atoms, is shown in figure 1 – although it wasmagnetic only below 0.6K. It contains a six-memberedcarbon ring (a benzene ring) attached to a nitrogenwith attendant two oxygens (a nitro group). Polarisedneutron studies showed that the unpaired spin densityis not limited to the nitroxide bonds but actuallyspreads through the benzene ring to the nitro group.The nitro group of one molecule can then coupleferromagnetically with the nitroxide group of aneighbouring molecule. The measured spin densityconfirms this and agrees well with theoretical predictions.

Another fascinating class of magnetic moleculescontain metal ions bound to flat, ringed organicstructures called semiquinones, which again have an

unpaired electron. When bound to a nonmagnetictitanium ion (Ti4+), they generate moderately strongferromagnetic coupling. Figure 2 shows a beautifulexample, with two semiquinone-based structuresattached to the titanium so that their planes areperpendicular to each other. The unpaired electronsresponsible for the magnetism are smeared out inorbitals above and below the rings. We still do notunderstand how they couple ferromagnetically: thesmeared-out orbitals could overlap directly; or theunpaired electrons could couple through theintervening orbitals of the metal ion. Polarised neutrondata suggest that the latter is the case, as unpaired spindensity is observed on the titanium atom.

Clusters of metal atoms bound in a molecularframework to create ‘single molecule magnets’ havealso generated enormous interest. As described on p.6,the magnetic spins of the metal atoms can combine incomplex ways. One of the first single moleculemagnets discovered was a cluster of eight iron ionsbridged by oxide and hydroxide ions – basically a fleckof rust enclosed in a flower-like organic structure.Magnetic measurements showed that the total spin ofthe molecule is 10. This suggests that since each ironatom has five unpaired electrons (spin 1/2 each, sototal 5/2), six iron atoms must have their spins up (spin 15) and two must have their spins down (spin 5).Polarised neutron analysis of the magnetisation densityas in figure 3 confirms this arrangement.

12

N E U T R O N S A N D M A G N E T I S M

Research team D. Gatteschi, A. Dei, A. Caneschi and R. Sessoli (University of Florence, INSTM, Italy), B. Gillon (Laboratoire L. Brillouin, Saclay, France),D. Ressouche, J. Schweizer and Y. Pontillon (DRMFC, CEA Grenoble, France)

NO2

O

O

N

N

>> D A N T E G A T T E S C H I

M

Figure 1 A nitroxide-based organic magnet

Figure 3 Map of the spindensity of the singlemolecule magnetwith a cluster ofeight iron atoms

5µB

0µB

-4µB

Figure 2 Map of the spin density of a titaniumsemiquinone radical-ion

KEY Red: positive spin densityBlue: negative spin density

Instrument D1Bused for neutronpowder diffraction

How do spins pair up in

superconductors?

Two of a kind

etals become superconducting when anattractive interaction causes the conduction

electrons to couple into ‘Cooper pairs’. The pairedelectrons can combine their 1/2 spins to give an integertotal spin of either 0 or 1. According to quantumtheory, such particles are able to share the same energystate, unlike the individual constituent electrons, and itis this quantum property that gives a superconductorits unique properties – like the absence of electricalresistivity. In the earliest superconductors discovered –mercury and lead – the pairing interaction is mediatedby vibrations of the metal’s crystal lattice (phonons).It is strongest for electrons with opposite spins, so thatthey align antiparallel with their total spin equal to 0.Such materials are known as ‘conventionalsuperconductors’.

In recent years, many new forms of superconductivityhave been discovered such as the high-temperaturecuprates (p.16) and ‘heavy fermion’ compounds –heavy-metal alloys in which the electrons interact in astrong and complex way. Many of these materials haveproperties which are fundamentally different fromconventional superconductors. In particular, thepairing interaction in these ‘unconventional’superconductors seems to be different in nature.The crystal lattice appears to play little part in theinteraction, and spins have a parallel arrangement.

Probing Cooper pairsAn excellent way of probing the type of pairing insuperconductors is to measure how easily the spins areinfluenced by an applied magnetic field – the spinsusceptibility. A conventional magnetometer, however,cannot be used because electric currents on thesurface of a superconducting sample screen out themagnetic response inside the bulk of the material (the Meissner effect). So, to find out about the spatial distribution of the magnetisation inside asuperconductor, polarised neutron diffraction is themethod of choice. Neutrons scatter off the periodicarrangement of induced magnetic moments within thesuperconductor’s crystal lattice to produce a series ofdiffraction peaks, whose intensities are then measured.

In the conventional superconductors, the electronsare paired antiparallel; their magnetic moments cancelso they are not affected by a magnetic field. Thismeans that at zero temperature the spin susceptibilitymust be zero. However, as the temperature rises, theincreasing thermal energy starts to break up some of

the Cooper pairs into normal electrons, and the spinsusceptibility rises. Neutron scattering data fromexperiments on the conventional vanadium-siliconsuperconductor (V3Si) confirm this picture (figure 1a).

In a superconductor with parallel spins, when thefield is applied along certain directions, the spinsusceptibility actually remains finite even at zerotemperature. A recently discovered superconductorstrontium ruthenate (Sr2RuO4) was thought to showthis behaviour. Indeed, carrying out the same kinds ofmeasurements, we found that there was no drop in thespin susceptibility when strontium ruthenate becamesuperconducting at 1.5K (figure 1b), as would beexpected with conventional superconductors. Thissuggests that the electrons constituting the Cooperpairs have a non-zero spin moment, indicating thattheir spin states are the same.

This is particularly interesting because strontiumruthenate (figure 2) appears structurally similar to thehigh-temperature superconducting cuprates, which,however, exhibit a spin-1/2 pairing. Strontiumruthenate behaves much more like helium-3 in itssuperfluid state in which the two electrons are in equalspin states.

13

Figure 2Strontium ruthenate is arecently discoveredsuperconductor which has manyunusual physical properties suchas a very large electronicspecific heat at lowtemperatures. It also has verystrong magnetic fluctuations(excitations)

M A G N E T I C I N T E R A C T I O N S

Research team S. M. Hayden and J. A. Duffy (University of Bristol, UK) and J. Kulda (ILL)

>> J I R I K U L D A A N D S T E P H E N H A Y D E N

M

Tc

T (K)

Shull & Wedgewoodthis work

Spin

susc

eptib

ility

00 5 10 15 20 25 30

0.1

0.2

0.3

0.4

0.5

0.6 Tc = 1K

0.2 0.4 0.6 0.8

T (K)

1 1.2 1.4 1.60

0.6

0.8

0.2

0

0.4

Figure 1aThe spin susceptibility in the conventional superconductorV3Si measured by neutron scattering. This experiment wasfirst performed by F. A. Wedgwood and Clifford Shull at MIT.Shull received the Nobel prize in 1994 for his contributionsto neutron scattering

Figure 1bThe temperature dependence of thesusceptibility in Sr2RuO4 measured byneutron scattering

The IN20 instrument

Spin chains witha chiral twist

One-dimensional magnetic systems, called spin chains,

show unusual dynamic behaviour that may shed light on

how high-temperature superconductors work

hiral objects are those that cannot besuperimposed onto their mirror images, such

as our right and left hands. They occur at every scale inNature from spiral seashells to molecules and elementaryparticles. Some of the most intriguing systems studiedby condensed-matter physicists – like the hightemperature cuprate superconductors (p.16) or low-dimensional spin chain systems – have no magneticorder in the traditional sense. However, the spins oftheir magnetic atoms do behave in a much morecomplicated way, than would correspond to simpledisorder. This situation is thought to be linked topossible chiral arrangements of their quantum states.

This motivated us to study a simple prototype of aspin-chain system, caesium cobalt bromide (CsCoBr3),in which chirality might emerge and be observedexperimentally. In this compound, the spins of thecobalt ions are arranged in an up-down fashion alongone crystal axis, forming an antiferromagnetic chain,and point in the opposite direction to those in theneighbouring chains. A similar arrangement with allthe spins reversed has the same energy, so both statescan be present, separated by a ‘domain wall’ as in figure 1.In fact, as a result of thermal fluctuations, one spin willoccasionally flip over, triggering the neighbouring spinsin the chain to flip sequentially. This creates a dynamicdomain wall like a travelling solitary wave, or soliton,which destroys the antiferromagnetic order.

Spiral solitonsIn the classical picture, these domain walls are just anabrupt switch-over between two neighbouring atoms(figure 1). Researchers have demonstrated previously,by neutron scattering, that because of ‘quantumfluctuations’ (due to the quantum uncertaintyprinciple) the solitary wave joining the two possiblespin arrangements is actually spread over a largenumber of adjoining atoms. Using spin-polarisedneutrons, we have now demonstrated that thistransition range has a well-defined internal structure:the spins turn in one direction or the other, creating aspin chain with a moving right-handed or left-handedtwist (figure 2). As a result, the moving solitary waveshave a clear handedness.

How spin-polarised neutrons scatter from a spiralstructure depends on its handedness. If the neutronspin is parallel to the helicity, the neutron passesunaffected but is scattered if the spin is antiparallel.The spiral thus acts like a spin filter for neutrons. Whilethis is a simple way to distinguish a static magneticstructure of fixed helicity, detecting the chirality ofmoving domain walls is far more challenging.

As long as the right-handed and left-handedsolitons in the spin chain have the same range ofenergies, the neutrons cannot distinguish the twochiral states of the solitons since their effects cancel.This explains why chirality has remained undetected sofar. We overcame this problem by applying an externalmagnetic field, which induces a slight energy differencebetween the populations of the right and left-handedsolitons. In accordance with theoretical predictions, theresponse then shows a small asymmetry betweenneutrons, which had their spins parallel to the field andthose with spins in the opposite direction.

This proves that quantum fluctuations induce asurprising metamorphosis, whereby theantiferromagnetic order is not simply lost by anappearance of moving domain walls but is replaced bya chiral arrangement of the magnetic moments.Thediscovered chirality may therefore shed new light onthe behaviour of systems with exotic magnetic states atthe limit between order and disorder, including thosethat lie at the heart of the still unexplained high-temperature superconductors.

14

N E U T R O N S A N D M A G N E T I S M

Research team H. B. Braun (ETH, Zurich, Switzerland), P. Boeni (TU Munich, Germany), B. Roessli (PSI Villigen, Switzerland) and J. Kulda (ILL)

>> J I R I K U L D A A N D H A N S - B E N J A M I N B R A U N

π/2–π/2

Ener

gy

Soliton momentumC

Figure 1A domain wall, or soliton, separatesthe two types of antiferromagneticorder in a spin chain. They have equalenergies so without quantumfluctuations, the domain wall can beplaced at any point along the chainwithout changing its energy.

Figure 2Quantum fluctuations inducetransitions between spin statesshown in figure 1 and lead to asoliton band. The states of lowestenergy are left- and right-handed

1 2

Drawing handsby M.C. Escher

The IN14 instrument

diffraction pattern for the as-deposited film. Thecoherent diffraction ring obtained confirms theirregularity. The size of the particles is 4 nanometres andthe average interparticle distance is 6.5 nanometres.

After annealing this diffraction ring disappears,indicating that the process has disrupted the orderingof the particles and that they have started toagglomerate. To estimate the sizes of the particleclusters formed, we compared the diffraction resultswith those predicted from theory. The SANS dataindicate that significant agglomeration occurs for all three samples. The particle-cluster size increasesfrom 4 nanometres for the as-deposited film to 6.2 nanometres for the film annealed at 580°C for 30 minutes, and to 16 nanometres for the 650°C/5minutes film. In the case of the film annealed at 700°Cfor 5 minutes, the median size appears to increaserather dramatically to 66 nanometres.

The challenge of understanding and inhibitingparticle agglomeration is currently the goal of anumber of research groups worldwide, as issynthesising the L10-phase nanoparticles without theneed for annealing. Although many hurdles must beovercome, the high coercivities found at roomtemperature demonstrate that these nanoparticles doindeed offer significant potential as very high-densityrecording media.

anotechnology – manipulating matter at thescale of a billionth of a metre – offers the

potential for many exciting applications. In particular,creating ferromagnetic particles just a few nanometresacross would allow data to be recorded and stored atdensities of around 0.16 terabit per square centimetre(1 terabit per square inch) – 10 times that currentlyavailable on magnetic disks and tapes.

However, there is a problem: as the particles getsmaller, thermal motions are more likely to preventthem from remaining magnetically aligned. This meansthat they can’t be magnetised and demagnetised as isneeded for recording data – the coercivity goes to zero.The hunt is now on to find ways of creating magneticallyordered arrays of nanoparticles that remain thermallystable at room temperature. Clusters of roughly equalnumbers of iron and platinum atoms in a highlymagnetically ordered arrangement (L10 phase),4 nanometres across, turn out to be good candidates.

Recently, researchers at IBM developed a method forpreparing the L10 phase by depositing the metalparticles on a surface in a polymer matrix to giveevenly-sized nanoparticles arranged in well-behavedlayers. Following deposition, the films must beannealed at more than 500°C for about half an hour toform the desired phase. This burns off the organicmaterial, leaving behind particles embedded in acarbon-containing matrix. However, the annealingprocess also tends to cause the particles toagglomerate into larger, less-ordered clusters. Tooptimise the films for data storage, it is vital tounderstand just how the final structural and magneticproperties of the particles are affected by theannealing conditions.

SANS reveals allSmall angle neutron scattering (SANS) was able toreveal the extent of agglomeration at differentannealing temperatures. The nanoparticles were firstdeposited on double-sided silicon substrates to createfilms consisting of three layers of particles withcomposition Fe58Pt42 as in figure 1. The films wereannealed over a range of temperatures (580 to 800 °C)and times (2 to 120 minutes); the magnetisation of thefilms was also measured. Figure 2 shows the neutron

Magnetic nanoparticles could be

the next medium for data storage

Towards better recording media

15

Figure 2Diffraction image measured at room temperature for as-depositednanoparticles with a diameter of 4 nanometres and a separation of 6.5 nanometres

A P P L I C A T I O N S

Research team T. Thomson and S. Sun (IBM Almaden Research Center, USA), S. L. Lee and C. J. Oates (University of St Andrews, UK),F. Y. Ogrin (University of Exeter, UK) and C. D. Dewhurst (ILL)

>> T H O M A S T H O M P S O N , S T E P H E N L E EA N D C H A R L E S D E W H U R S T

N

0.15

0.1

0.05

0

-0.05

-0.1

-0.15

Annealing at 580°C

-0.15 -0.1 -0.05 0 0.05 0.1 0.15

q y(Å

-1)

qx (Å-1)

a

i

ii b

Figure 1(a) Polymer-mediated self-assembly of nanoparticles,(i) is the polymer and (ii) showsthe surfactant-coated particles;(b) electron micrograph of theevenly-sized, evenly-spacednanoparticles

The D11 small angleneutron scattering

diffractometer

The behaviour and motion of magnetic fields in superconductors

is paramount in determining their usefulness for carrying large

resistance-free electrical currents and generating strong magnetic fields

uperconductors are materials that conductelectricity without resistance when cooled below

a critical temperature, Tc . They can therefore transmitpower with minimal loss as well as generate strongmagnetic fields.

Until recently, the only superconducting materialsexploited commercially (in powerful magnets) werealloys with Tcs not much above absolute zero. They hadlimited use because they required expensive liquidhelium as a coolant. Then in 1987 a new class ofsuperconductors was discovered – the ceramiccuprates – some of which have Tcs ranging upwardsfrom 92K. These high-temperature (HTC) materials arerevolutionising superconducting applications becausethey can be cooled with much cheaper liquid nitrogen.Indeed, they are already being exploited in powertransmission, and in high-field magnets for levitatingtrains and medical imaging machines.

Their efficient application, however, requires anunderstanding of their superconducting properties atthe microscopic level, in particular the behaviour ofmagnetic fields generated within them. The magneticfield is bundled into tubes called flux-lines or vorticessustained by circulating currents. Each vortex repels itsneighbours, like bar magnets aligned in the samedirection, and they naturally arrange themselves in asolid, hexagonal vortex lattice (VL).

The ability of superconductors to carry large currentsdepends crucially on the response of the VL. When acurrent passes perpendicular to the vortices they feel aforce, in the same way as a current-carrying wire in anelectrical motor feels a force in a magnetic field. Ifthere is nothing to resist this force, the VL moves andenergy is lost in dragging vortices though the material,so that the electrical resistance returns. Fortunately,real materials are often full of defects that form siteswhere the superconductivity is locally weaker than inthe surrounding material. Vortices prefer to sit onthese weak regions or ‘pinning centres’. Only if the VL is‘tacked down’ by these pinning centres can a significantbulk current pass though the superconductor withoutresistance. For high-current applications, manufacturersgo to great lengths to introduce impurities in order toimprove the electrical characteristics.

Seeing vortices A neutron beam provides an excellent probe of theseeffects. The rows of vortices reflect neutrons to form apattern of diffraction peaks characteristic of theirarrangement. Vortex-pinning can be observed directlyas broadened peaks in the diffraction images.

Of particular interest are the newer HTC materialswith higher operating temperatures. In these cases, theincreased thermal energy can shake the vortices soviolently that the whole vortex ‘solid’ becomesunpinned and may even melt into a liquid of vortexlines. Such vortex liquids are useless for high-currentapplications. In most materials, the melting happensonly at temperatures approaching the Tc , but a notableexception is a bismuth-containing cuprate – a favouredmaterial for making superconducting tapes. Small-angle neutron diffraction experiments on this materialhave provided one of the first confirmations of thistheoretically predicted phenomenon.

The ‘Holy Grail’ of superconductivity is thus not onlyto discover a material that superconducts at roomtemperature but also one in which the vortices can berigidly pinned so as to support a large current. Neutrondiffraction will continue to play a pivotal role in thistechnologically important quest.

16

N E U T R O N S A N D M A G N E T I S M

Research team R. Cubitt, C. D. Dewhurst and S. Levett (ILL), E. M. Forgan et al (University of Birmingham, UK), D. McK. Paul et al (University of Warwick,UK), S. L. Lee et al (University of St. Andrews, UK), M. R. Eskildsen et al (Notre Dame University, USA)

Thermal energy

Vortex lattice

Pinning

pins

>> R O B E R T C U B I T T A N D C H A R L E S D E W H U R S T

S

The vortex lattice (VL) in superconductorscreates an undulating magnetic field, much

like the shape of the inside of an egg-box.The vortices are like threads of magnetic field.

They repel each other and usually form ahexagonal lattice. Material defects can act as

‘pins’ to tack-down the VL and stop it frommoving when an electrical current is passed.

‘High’ temperatures close to the Tc canviolently shake the VL from its pinning centres

and even melt the VL into a vortex liquid

Small-angle neutron diffraction images showing peaks from the vortex lattice (VL)in (a) niobium, Tc = 9.2K and (b) the recently discovered magnesium diboridesuperconductor (Tc = 38K) where the VL splits into two equivalent domains of

hexagonal lattice. If the currents around vortices are not circular due to theunderlying crystal symmetry, the VL can become rhombic as in yttrium nickel boron

carbide (c), Tc = 15.5K; or even square as in strontium ruthenate (d), Tc = 1.5K

Superconductivity:resistance is useless

The D22 small angleneutron scatteringinstrument

a b c d

U Fe

Neutron experiments combined

with X-ray studies are leading to

better permanent magnets

Inside modernmagnets

lthough iron is the most familiar magneticmaterial, many of today’s practical magnetic

devices are made from a combination of magneticelements – so-called transition metals such as iron orcobalt, and rare-earth elements such as samarium orneodymium. Powerful permanent magnets made ofsamarium and cobalt (SmCo5) or neodymium/iron/boron (Nd-Fe-B) compounds are in wide commercialuse. The magnetic moments of the different types ofatoms in these materials interact in complex ways andwe still do not fully understand how. To find out moreabout their magnetic structure, and thus improve theirperformance, we need to look at ‘model’ systems andmeasure the fundamental interactions.

One such magnetic material is a compoundcontaining iron, aluminium, and a rare earth (M) such asdysprosium (Dy) or holmium (Ho), or an actinide suchas uranium (U). The formula is MFe4Al8, where M = Dy,Ho or U. The electrons responsible for iron’s magnetismsit in the so-called 3d electronic shell in the atom, whilein dysprosium or holmium, which are heavier elementswith more electrons, they sit in the 4f shell. In uranium,the heaviest element, they sit in an outer 5f shell.

A complex structure We have been studying in considerable detail how thevarious electronic shells (3d, 4f and 5f) develop theirordered magnetic moments as the temperature islowered. This has been the subject of considerablecontroversy in the past. Using polarised neutrons (p.6), we could obtain information about theorientation of magnetic moments. From these results,coupled with magnetic X-ray measurements made atthe neighbouring European Synchrotron RadiationFacility, emerges a unified picture of the magneticordering process.

As we expected, the magnetic ordering isdominated by the 3d electrons of iron, although thetype of ordering is influenced by the M atom.Surprisingly, in the rare-earth based materials the initialmagnetic ordering occurs only in the array of ironatoms, whereas the rare-earth magnetic momentsremains disordered. In contrast, in the uraniumcompound, the iron 3d and uranium 5f electron shellsinteract enough to cause the two sets of atoms to

order at the same temperature.The X-ray studies show that ordering in the 4f

electrons is more complicated and involves theinformation being transmitted to the 4f shells via theintermediary 5d electrons of the rare-earth element,but they do order when the temperature is lowenough. Through polarised neutron measurements, weshowed that the 4f magnetic moments dance alongwith the (dominant) iron 3d moments, but with theirspins oriented at an angle to those of the 3d momentsdepending on temperature. This phase angle isdifferent for each element, and its origin is notunderstood. At the lowest temperatures, the rare earthmoments try and align themselves ferromagnetically(all in the same direction), but the iron prevents themfrom doing so, and the subsequent competition givesrise to unusual behaviour in a magnetic field. Theseexperiments are good examples of how neutron and X-ray studies together can analyse the complex magneticstructure of strategically important materials.

17

The unusual magneticconfiguration of theuranium iron aluminium(UFe4Al8) compound

A P P L I C A T I O N S

Research team J. A. Paixão (University of Coimbra, Portugal), P. J. Brown (ILL), B. Lebech (Risø National Laboratory, Denmark),G. H. Lander (Institute for Transuranium Elements, Karlsruhe, Germany)

>> G E R A R D L A N D E R

A

IN8 – one of the instrumentsinvolved in the experiment,which has recently been rebuilt

agnetic films and multilayers, just a few atomsthick, are used in a wide variety of applications

including magnetic media, recording heads, magneticrandom access memory, sensors, and actuators. Theinterfaces and surfaces of these thin structures oftenintroduce novel and useful characteristics. Indeed,most of the recent explosive growth in magnetic datastorage has been due to new discoveries and a betterunderstanding of the magnetic and electronicproperties of complex magnetic thin films.

Studying the magnetism of these systems isextremely challenging. Most experimental methodsgive information averaged over all the layers. However,the technique of polarised neutron reflectometry – inwhich polarised neutrons are reflected off the surfaceand interfaces – provides an excellent method of probingin detail their complete magnetic and nonmagneticstructure both across and along the layers.

This analysis is achieved via two types ofsimultaneous measurement: that of normal, or specular,reflection of polarised neutrons – which givesinformation about the transverse structure of layers;and that of diffuse, or off-specular, scattering – in whichdifferences in the lateral structure (magnetic andnonmagnetic) cause the reflected neutrons to deviatefrom the specular direction.

Giant magnetoresistance Using this approach we investigated a phenomenon ofconsiderable commercial significance, giantmagnetoresistance (GMR), whereby an externalmagnetic field induces a very large change in electricalresistance in a material with alternating magnetic andnon-magnetic layers. The effect results from changes inmagnetisation alignment in the layers whichcorrespondingly influence the degree of scattering ofelectrons (responsible for resistance) passing throughthe layers. GMR materials are revolutionising datastorage since they can be used in the devices that readthe magnetised areas of computer disks.

One material we studied consisted of alternatingiron (magnetic) and chromium (nonmagnetic) layers.We were able to determine how the magnetisation wasoriented across and along the layers, from analysingthe line shape in the specular and off-specular

reflectivity patterns (figure 1a). This revealed that inthe presence of a magnetic field the direction ofmagnetisation of the iron atoms is twisted up throughthe layers in a complex way – the first time this hadbeen seen. The magnetisation within the layers fallsinto lateral domains about 200 to 300 nanometresacross. Within each vertical domain column, themagnetisation orientation twists in alternate directionsin alternate iron layers, with the difference in tiltincreasing from the centre towards the top and bottomlayers. The resulting configuration, figure 1b, has no netmagnetic moment perpendicular to the external fieldand therefore is stable.

Polarised neutron reflectometry can also be used tostudy GMR multilayer devices, with potential in read-head applications, called exchange-biased spin valves.These consist of two ferromagnetic layers separated bya nonmagnetic spacer. An additional antiferromagneticlayer is coupled to one of the layers, effectively ‘pinning’down its orientation. When the ferromagnetic layersare aligned parallel the resistance is low, but when anapplied magnetic field reverses the magnetisation ofthe ‘free’ layer the resistance rises dramatically.

Figure 2 shows what can happen during themagnetisation reversal of an exchange-biased structurebased on cobalt oxide-cobalt-silicon layers. The spins inthe magnetic cobalt layers can re-orient to the newfield direction either starting in small clumps whichrapidly grow in size (the domain walls therefore move),or rotating all together. The two scenarios producedifferent off-specular reflectivity patterns (a and b).

18 Research team V. Lauter (TU Munich and ILL), H.J. Lauter (ILL), B. Toperverg (PNPI, St Petersburg, Russia) and V. Ustinov (MPI, Ekaterinburg, Russia)

±40°

H

A

B

ϕn

±75°

±49°

±70°

±56°

±63°

±63°

±56°

±70°

±49°

±75°

±40°

M

Figure 1bConfiguration of magnetisation orientation within

the iron layers in an iron-chromium multilayerfitting the experimental data. The only possible

two types of magnetic domains are depicted.Brackets A and B indicate the two transverse

antiphase parts of one lateral domain

N E U T R O N S A N D M A G N E T I S M

Nano-sized magnetic films have huge commercial potential

but have complex characteristics; polarised neutron

reflectometry offers the best way to see inside them

>> V A L E R I A L A U T E R

Reflections on magnetic multilayers

Figure 2Reversing the magnetic

field in an exchange-biasedspin valve (CoO/Co/Si) as

‘seen’ by neutrons

a

b

Figure 1aExperimental two-dimensional map of theintensity scattered froman iron-chromiummultilayer

p i+p f,Å

-1

0.1

0.05

0

-0.06 -0.04 -0.02 -0 0.02 0.04 0.06

pi - pf , Å-1

J

J

J

J

J

J

J

J

TcTN

Interleaved layers of iron and vanadium

exposed to hydrogen offer a new experimental

approach to study magnetic systems with

variable coupling between layers

A new way to studymagnetic multilayers

evices based on ultra-thin multilayers ofmagnetic and nonmagnetic materials have

important applications in the burgeoning area of‘spintronics’ whereby information is transmitted orstored by manipulating electrons’ spin. An importantaspect of current research is to determine how themagnetic interactions between the layers can be tuned– for example, by altering the inter-layer spacings – tocreate useful properties. We have prepared a multilayersystem that offers a new experimental tool for this kindof investigation.

More than 30 years ago Robert Griffiths, when atCarnegie-Mellon University in Pittsburgh, proposed, in athought-experiment, a model system of spins arrangedon a cubic lattice (figure 1). This lattice could beregarded as a series of planes of magnetic momentswith a fixed, ferromagnetic coupling within the planesbut a variable coupling between moments in adjacentplanes. The interlayer coupling strength can be variedfrom positive, through zero, to negative. A positivecoupling denotes that the magnetic interactionbetween the sheets causes the moments in adjacentplanes to align and the whole system becomes aferromagnet. If the coupling is negative, however,neighbouring ferromagnetic planes align in anantiparallel way so that the material becomes anantiferromagnet. With zero coupling there’s nointeraction between the magnetic sheets – the ferro-magnetic planes no longer ‘see’ each other – and eachlayer becomes an isolated two-dimensional magnet.

Griffiths presented the properties of this system as a‘phase diagram’ plotting the point at which the overallmagnetic order vanishes in the system (criticaltemperature) against the strength of the magneticcoupling (figure 2a). This reveals that as adjacent layersdecouple, the critical temperature also drops towards aminimum value at zero coupling – the system displaysa strongly reduced but finite ordering temperature of atwo-dimensional magnet. Interestingly, in the absenceof an external magnetic field, the phase diagram issymmetrical around the temperature axis, indicatingthat the coupling behaviour over temperature is thesame for both the ferromagnetic and antiferro-magnetic regimes.

This kind of model which presents what happens as

the interlayer coupling is varied is highly suitable forstudying real 2D magnetic systems, and also forexploring the gradual transition from both aferromagnetic and antiferromagnetic bulk magnet to aplanar system.

A real experimentWe have now succeeded in realising this thoughtexperiment of a fixed intra- and variable interlayercoupling for the first time, and have established thephase diagram. First, we prepared a multilayer systemof two planes of iron atoms (ferromagnetic) and 13planes of vanadium atoms (nonmagnetic) – repeated ina series of 200 composite layers. Neutron reflectivityand magnetometry studies show that the interlayercoupling is antiferromagnetic.

We could then alter this coupling by introducinghydrogen gas. Previous work had shown that thevanadium layers readily and exclusively take uphydrogen, expanding the metal lattice, which has astrong effect on the electronic properties of the hostmetal. In our experiment, a very small amount ofhydrogen was sufficient to alter the strength andindeed the sign of the interlayer coupling.

We combined neutron reflectivity experiments withmagnetometry studies to investigate the properties ofour iron-vanadium multilayer system at differenthydrogen concentrations and temperatures, and thenmapped out a phase diagram (figure 2b). This showedthat the experimental phase diagram is in very goodagreement with Griffiths’ theoretical one.

We think that metallic multilayer systems in whichhydrogen is incorporated in one of the sublayers offersa new experimental approach to the investigation ofmagnetic phase diagrams, changes in the dimension ofa magnetic system, and the critical conditions at whichmagnetic properties change.

19

Figure 1Spins placed on a cubiclattice with fixed intralayercoupling (J) but variableinterlayer coupling (J’)

Figure 2a Griffiths’ schematic magneticphase diagram of a model systemof spins on a cubic lattice (T istemperature, J’ is the interlayercoupling strength and H is theexternal magnetic field)

b The experimental phasediagram of an iron-vanadiumFe2/V13 multilayer as a functionof hydrogen concentration andtemperature

0.00

0.0 0.5 1.0 1.5

120

100

80

60

40

20

0

0.02 0.04 0.06

A P P L I C A T I O N S

Research team V. Leiner (ILL and University of Bochum, Germany), K. Westerholt and H. Zabel (University of Bochum, Germany),A. M. Blixt and B. Hjörvarsson (Uppsala University, Sweden)

>> V I N C E N T L E I N E R

D

H

T FerromagneticAntiferromagnetic

J

T(K

)

ba

The ADAM instrument

Glossary

20

N E U T R O N S A N D M A G N E T I S M

AntiferromagnetismA type of magnetic order in which the magneticmoments are alternatelyoppositely aligned.

ChiralityThe notion of ‘handedness’,referring to an object whosemirror image cannot besuperimposed on it.

Colossal (or giant)magnetoresistanceA phenomenon in whichcertain materials show verylarge changes in electricalresistance with magnetic field.The effect was first seen ineuropium chalcogenide thinfilms, where it is referred to asgiant magnetoresistance (GMR).Even larger resistance changes,colossal magneto-resistance(CMR), have since beenobserved in a family of dopedmanganese oxides.

CoercivityThe magnetic field intensityneeded to demagnetise a fullymagnetised material.

Crystal latticeThe regular three-dimensionalarray of atoms or molecules ina crystalline material.

DeuteriumA heavier isotope of hydrogenhaving a neutron as well as aproton in the nucleus.

Electron orbitalThe quantum state of anelectron in an atom or ion.

Electron shellAtomic electrons are arrangedin shells composed of specificorbitals – s, p, d, f.

DomainA region in a magnetic materialin which all the magneticmoments are aligned; domainswith different alignments areseparated by walls.

FerromagnetismA type of magnetic order inwhich the magnetic momentsare all aligned.

Giant magnetoresistanceSee left: colossalmagnetoresistance.

Gigapascal (GPa)A unit of pressure – 1 gigapascalis a billion pascals, or 9870atmospheres.

Ground stateThe lowest quantum energylevel a system can occupy.

High temperaturesuperconductors (HTCs)A class of ceramic oxides whichbecome superconducting attemperatures above the boilingpoint of liquid nitrogen.

HysteresisThe effect whereby theresulting change due to a forcelags behind its application –particularly the magnetisationof a material in a changingmagnetic field.

Magnetic momentAn effect arising from aspinning electric charge. Atomshave magnetic moments as aresult of the spin and orbitalmotions of any unpairedelectrons. Neutrons also have amagnetic moment.

Magnetic multilayersAdvanced materials consistingof alternating ultrathin layers of magnetic and nonmagneticmaterials. Some of thesematerials behave as spin valves showing giantmagnetoresistance.

Magnetic scatteringNeutrons have a magneticmoment which interacts withthe magnetic moment of anunpaired electron in an atom.The arrangement of electronmagnetic moments in a crystallattice therefore produces acharacteristic magneticdiffraction pattern.

Magnetocaloric effectAn effect whereby somemagnetic materials heat upwhen placed in a magneticfield and cool down whenremoved.

Magnetoelectric crystalA crystal that develops amagnetisation when immersedin an electric field and adielectric polarisation whenimmersed in a magnetic field.

Molecular magnetSingle molecules with one orseveral unpaired electronswhich show a variety ofmagnetic behaviours.

Molecular orbitalThe quantum state of anelectron associated with amolecule.

Monoclinic crystalA crystalline structure withthree axes of unequal length,two of which are perpendicularto the third axis, but not toeach other.

NanometreOne-billionth of a metre (10-9 metres).

NanotechnologyThe manipulation of materialsand their physical and chemicalproperties at the scale of ananometre.

NeutronOne of the two particles foundin the atomic nucleus. Like allquantum particles, the neutronhas both particle-like (mass,velocity, kinetic energy,gravitational effects) and wave-like (wavelength, diffraction andrefraction effects) properties.

Neutron diffractionNeutrons can be reflected,or scattered, off a material in which the interatomicdistances are similar to theneutron wavelength. Thescattered waves interfere toproduce a characteristicdiffraction pattern.

Off-specular reflectionSee ‘Specular reflection’, right.

Orthorhombic crystalA crystalline structure withthree perpendicular axes ofdifferent length.

Permanent magnetsMaterials such as iron thatremain magnetic, with theconstituent moments aligned,even in the absence of anexternal magnetic field.

Phase transitionThe abrupt change in theproperties of a material underdefined conditions such astemperature and pressure.

PhononA quantum of atomic vibrationin a crystal lattice.

Polarised neutronsA beam of neutrons whosespins are all aligned.

Polarised neutron diffractionA class of neutron experimentsusing polarised neutrons toinvestigate the magneticproperties of materials.

Polarised neutronreflectometryA technique using polarisedneutrons to investigatemagnetic properties at surfacesand interfaces.

Powder diffractionCoherent scattering from apolycrystalline material.

Quantum computingA new type of computingbased on manipulatingquantum states which areintrinsically connected in a way that allows much moreinformation to be processed.

Quantum fluctuationsThe mathematics underpinningquantum theory predicts that there is always a smalluncertainty in connectedproperties such as position andmomentum, or energy and time. This results in inherentfluctuations in quantum states,for example, spin states that are similar in energy.

Quantum stateThe defined energy state of aquantum system.

Small angle neutronscatteringThe measurement of neutronscattering at small angles usedto investigate large-scalestructures such as polymers orbiological moleculararrangements.

Specular reflectionNormal reflection from aninterface or a multilayerstructure in which the angle ofincidence of the neutron beamequals the angle of reflection.Specular reflection givesinformation about thetransverse structure of layers.In addition, the lateral layerstructure (magnetic andnonmagnetic roughness)causes the reflected neutronsto deviate from the speculardirection (diffuse or off-specular reflection).

SpinThe term describing theinternal angular momentum of a quantum particle such asthe electron or neutron. It hasthe value of 1/2 or multiples of that value.

SpintronicsA new technology in whichinformation is transferred orstored using the spin ofelectrons.

Spin valveA novel device that operates byallowing electrons with onespin state, but not the other,to pass a junction.

SuperconductorA material that has no electricalresistance below a certaintemperature.

21

Contacts

Scientific Coordination Office (SCO) Institut Laue Langevin6 rue Jules HorowitzBP 156F-38042 Grenoble Cedex 9FranceEmail: cicognani@ill.fr or kjenkins@ill.fr

Page 6

Dr Jane Brown Email: brown@ill.fr

Page 11

Dr Grégory ChaboussantLaboratoire Léon Brillouin (LLB-CNRS-CEA)CEA Saclay91191 Gif-sur-Yvette Cedex, FranceEmail: chabouss@llb.saclay.cea.fr

Page 7

Dr Tapan Chatterji Email: chatt@ill.fr

Pages 16

Dr Robert CubittEmail: cubitt@ill.fr

Page 4

Dr Roland CurratEmail: currat@ill.fr

Page 15 and 16

Dr Charles DewhurstEmail: dewhurst@ill.fr

Page 12

Professor Dante GatteschiDepartment of ChemistryUniversity of FlorencePolo Scientifico Universitariovia della Lastruccia 3 I-50019 Sesto FiorentinoItalyEmail: dante.gatteschi@unifi.it

Pages 13 and 14

Dr Jiri KuldaEmail: kulda@ill.fr

Pages 4 and 17

Dr Gerard LanderInstitute for Transuranium ElementsPO Box 2340D-76125 KarlsruheGermanyEmail: gerard.lander@cec.eu.int

Page 18

Dr Valeria LauterEmail: vlauter@ill.fr

Page 19

Dr Vincent LeinerInstitute for Experimental Physics 4Ruhr University of BochumD-44780 BochumGermanyEmail: vincent.leiner@gkss.de

Page 8

Dr Eddy Lelièvre-Berna Email: lelievre@ill.fr

Page 9

Dr Bachir OuladdiafEmail: ouladdia@ill.fr

Page 10

Dr Clemens Ritter Email: ritter@ill.fr

Page 1 and 4

Dr Christian VettierEmail: vettier@ill.fr

The authors are based at ILL unless otherwise stated

Published by the Scientific

Coordination Office

Institut Laue Langevin

Email: sco@ill.fr

Editor: Nina Hall

Tel: 44 (0)20 8248 8565

Fax: 44 (0)20 8248 8568

Email: ninah@ealing.demon.co.uk

Design: Spaced

Email: pete@spaced-out.biz

www.spaced-out.biz

Photography (unless shown otherwise):

ILL and Artechnique – artechnique@wanadoo.fr

Main cover photo: NASA

The planet Mars has a complex andpatchy magnetic field. Neutronexperiments may explain why

Institut Laue Langevin6, rue Jules HorowitzBP 15638042 Grenoble

Cedex 9France

web: www.ill.fr

September 2004