highlights in research at the ILLEXPLORING MATTER WITH NEUTRONS

Institut Laue Langevin, Grenoble, France

foreword

introduction

biologyNeutrons help fight obesityDavid Pignol and Peter Timmins

Highlighting the composition of celluloseHelmut Schober, Christoph Czihak and MartinM ller

New insights into cellulose structureHenri Chanzy, Paul Langan and YoshiharuNishiyama

A new model of a biological membraneGiovanna Fragneto-Cusani

Mapping lipid membranesChristian M nster and Tim Salditt

soft matterA new tool to observe ultra-thin polymerfilmsPeter M ller-Buschbaum

Detergents, foods and pastes understanding how they flowSimona Bare and Adrian Rennie

liquids & glassesThe glass transition made easyF. Javier Bermejo

The sound of waterCaterina Petrillo

Super cool water!John Dore

chemistry & structureThe many phases of waterMichael Koza and Helmut Schober

2

3

4

5

6

7

8

9

10

11

12

13

14

EXPLORING MATTER WITHNEUTRONS contents

materialsOil generation in fractal rocksAndrzej Radlinski and Peter Lindner

Probing the hidden strength of alloysGernot Kostorz

Helium bubbles in steelGiovanna Cicognani and Roland May

Reflections on novel glasses for fasterelectronicsRalf Siebrecht

thin filmsMagnetic spirals in very thin filmsVincent Leiner and Hartmut Zabel

Explaining giant magnetoresistanceValeria Lauter-Pasyuk, Boris Toperverg and Hans Lauter

magnetismQuantum tunnelling in molecular magnetsIsabelle Mirebeau

New class of magnetic phase transitionsdiscoveredJiri Kulda

Inside modern magnetsGerard Lander

Relaxing spins in glassy magnetsRobert Cywinski

fundamental physicsDropping neutron wavesGerbrand van der Zouw

quantum systemsSuperfluid helium in porous mediaBj rn F k

glossary

contacts

15

16

17

18

19

20

21

22

23

24

25

26

27

29

especially benefit those areas of research that arecurrently creating a great deal of excitement amongstthe scientific community, either because they havetremendous commercial potential such aselectronics, catalysis, pharmaceutical development,materials and engineering, or are at the frontiers ofknowledge such as magnetism and particle physics.

The ILL has an exciting future ahead. However, wecan also look back at the many successes over thepast 30 years. In the pages that follow you will find across-section of recent results which illustrate therange of research carried out at ILL. I hope you enjoyreading about them.

Professor Dirk Dubbers Director

Progress in science increasingly depends onobserving and measuring subtle effects in ever moresophisticated experiments. It is vital, therefore, thatcentral facilities like ILL provide instrumentation andresources that match users requirements. With this inmind, we have now set in place an ambitiousprogramme the Millennium Programme toupgrade ILLs instruments and infrastructure to newstandards of excellence, and to attract new users.The programme will be implemented in two parts overthe next 10 years, and aims at improving the overallinstrument performance by a factor of 10. This willallow our user community to carry out new types ofhighly sensitive experiments not possible before. Theprogramme is not yet finalised but we have alreadystarted the upgrade of a first group of instruments.W e hope to continue with other instruments and torenew the neutron guides. These refurbishments will

2

foreword

About the ILL The Institut Laue Langevin (ILL) in Grenoble, France,

is a major international facility for scientific research,

operating the most intense continuous source of

neutrons in the world. It possesses a wide range of

instruments for carrying out simultaneously

many different kinds of experiments to

investigate the structure of matter at a

microscopic level across the disciplines of

biology, materials science, engineering,

chemistry and physics.

About 1200 scientists visit the Grenoble

site every year to work with our on-site

scientists in carrying out cutting-edge

research. Although much of the research is

in fundamental science, a substantial

amount is of direct industrial relevance.

Neutron scattering experiments have made

significant contributions to our understanding of

the structure and behaviour of biological and soft

condensed matter, to the design of new chemicals

such as drugs and polymers, and to materials

used in electronics and structural engineering.

Neutron studies also offer unique insights into the

nature of complex systems at the most

fundamental level.

ILL was founded in 1967 as a bi-national

enterprise between France and Germany with the

UK joining later in 1973. As well as these three

Associate Members, six Scientific Members now

participate in ILL: Spain, Italy, Switzerland, Austria

and Russia, and more recently the Czech

Republic. The Institute has a German director

(Dirk Dubbers) and two associate directors, one

from France and one from the UK (Christian

Vettier and Colin Carlile).

Left to right: ILL directors Dirk Dubbers,Christian Vettier and Colin Carlile

positions in biological materials, particularly thosecontaining water (H2O). What is more, hydrogenscatters very differently from its heavier isotope,deuterium. This means that selected components in astructure can be highlighted by substituting hydrogenwith deuterium. The degree of isotopic substitution canalso be controlled such that the scattering strength inone part of a structure is the same as in thesurrounding medium, rendering it invisible so thatanother part of the structure stands out in contrast .

As well as determining structure, neutron scatteringalso reveals the motions of atoms and molecules via anexchange of energy between the neutrons and thesample (inelastic scattering). Very small changes inenergy measured at low temperatures may also beused to measure subtle quantum processes in exoticmaterials.

Neutrons also have a spin, or magnetic moment,which can interact with the electron spins in magneticmaterials. Beams of polarised neutrons (in which all thespins are aligned) offer a unique tool for characterisingexotic materials with complex magnetic structures anarea of growing interest.

Another area of increasing importance, is the studyof thin films and multilayers. Many of the most importantbiological, chemical and electronic/magneticphenomena occur at surfaces and interfaces. Underappropriate experimental conditions, neutrons can bereflected off surfaces and interfaces so that the depth ofthe layers and their structure and dynamics can bedetermined.

Finally, the neutron itself is used to investigate thelaws of Nature. ILL has facilities to prepare beams ofneutrons of very low energies (cold neutrons) forextremely precise experiments on gravity and onparticle properties significant in developing a unifiedtheory of the fundamental particles and forces of theUniverse.

N eutrons are subatomic particles found in thenuclei of atoms. They are emitted during certainnuclear processes including the fission of uranium-235 so can be obtained from a nuclear reactor. Likeall subatomic particles neutrons obey the laws ofquantum mechanics, which means they behave likewaves as well as particles. The wavelength ofneutrons (tens of nanometres) corresponds to thedistances between atoms and molecules in solids andliquids. Consequently when they interact with matter for example, a regular array of atoms or molecules ina crystal lattice the neutron waves are reflected, orscattered. Waves reflected from similarly orientedplanes of atoms in the crystal interfere and re-inforceeach other periodically to produce a characteristicdiffraction pattern (like water waves on a lake thatmeet after being reflected off two rocks). The patterncan be recorded as a series of peaks of the scatteredneutron intensity, which provides information aboutthe position of the atoms and the distance betweenthem.

There are many variations of the scatteringprocess which gives the technique its wideapplicability to many different kinds of materials. Forinstance, the structure of larger molecular or atomicassemblies such as biological membranes, polymers,engineering structures and minerals, can be studiedby measuring very small angles of scattering (thewider the distance between atoms the smaller theangle of reflection). This is called small angle neutronscattering (SANS) and requires a special, very longinstrument to resolve the small angles.

Neutrons scatter off hydrogen atoms quite strongly(unlike X-rays) so are also used to establish their

3

introduction to neutron scattering

Neutrons are excellent probes of all kinds of matter. They are more penetratingthan X-rays and provide complementary information on structure and dynamics

complex with PL, and bile salts which solubilise the fattyacids from the triglyceride breakdown to form minutespherical structures called micelles.Analysing the active complexW e wanted to understand at a molecular level howthese components come together to form an activecomplex. A technique called small angle neutronscattering, as described on page 3, is ideal forinvestigating large molecular structures such as thisone. By using a method called contrast variation, wecould visualise specific parts of the complex whilst otherparts remain invisible. The principle behind this is thesame as the familiar school laboratory experimentwhereby sugar is slowly added to water in which isimmersed a glass rod. At a certain sugar concentrationthe glass rod becomes invisible because it has thesame refractive index as the sucrose solution. In otherwords, light is scattered equally by the glass and thesolution. We say that there is no contrast between therod and the solution. In the case of neutron scattering,heavy water is added (water in which hydrogen issubstituted by its heavier isotope deuterium, D2O). So,for example, when the water contains 10 per cent heavywater to 90 per cent ordinary water, the lipase andcolipase are visible and the bile salt invisible.

Using these techniques we have been able to showthat the complex contains two molecules of colipaseattached to each pancreatic lipase molecule and theyare connected and stabilised by a micelle of the bilesalts. The model of the complex is shown in the Figure.

This knowledge could enable us to design drugs thatcan prevent the complex from forming in the first placeand thus limit the amounts of fat assimilated by thebody. Our colleagues in Marseille are currentlyconducting preliminary clinical studies using rats. l

M ammals, including humans, digestfood with secretions produced in severalorgans along the digestive tract thesalivary glands in the mouth, the stomach,the gall bladder, the pancreas and theintestine. The secretions contain enzymeswhich break down the complex foodmolecules proteins, fats andcarbohydrates into simpler molecules thatcan then be absorbed into the bloodstream.W e are particularly interested in the way fats arebroken down. The reason is that the process is quitecomplicated. Ninety-five per cent of fats in theW estern diet consist of molecules known astriglycerides which, because they do not dissolve inwater, exist in the form of oily droplets. To be digested, an enzyme produced by the pancreas called pancreatic lipase (PL) mustfirst attach itself to the oil globules. However, PL can t actalone. Its molecular shape is such that its so-calledactive site the area in the enzyme structure wherethe triglyceride would interact and be broken up ishidden away. Two other components are needed tounmask the site and help the enzyme adsorb onto thedroplet surface. These are colipase, which forms a

Studies revealing the shape and structure of a molecular complex involved in fat

digestion could lead to new slimming drugs

Neutrons help fight obesity

biology

4

DAVID PIGNOL & PETER TIMMINS

Calculated model of theactive complex built with

two lipase/colipasemolecules per micelle

corresponding to observedcrystal structure

Research team D. Pignol, J. Hermoso and J. C. Fontecilla-Camps(LCCP, CEA-CRNS Grenoble), L. Ayvazian, B. Kerfelec, I. Crenon

and C. Chapus (CNRS Marseille), P. Timmins (ILL)

method and the contrast variation method, described onthe previous page, offers an excellent probe. Inelasticscattering happens when some of the energy of theincoming neutrons is absorbed by the atoms in a material,as they oscillate during the scattering process, causing achange in the energies of the scattered neutrons. Thischange in energy reflects the characteristic vibrations ofthe atoms depending on how they are bonded to theirimmediate molecular environment. At the same time, bysubstituting certain hydrogen atoms with deuterium inchosen regions of the material we can render thedisordered regions invisible and highlight the dynamicresponse the spectrum of atomic vibrations of just thecrystalline microfibrils, and vice versa.

From these results we found that neutron inelasticspectra of the disordered parts of cellulose were distinctlydifferent from those of the crystalline parts, closelyresembling that of completely disordered artificial cellulose.These spectra therefore constitute true signatures of therespective parts of cellulose. The universal character ofthese signatures indicates that the disordered regions aresimilar in all types of this material and do not depend onits source. From the amount of hydrogen that can besubstituted by deuterium we also concluded that thedisordered regions are mainly formed by the surfaces ofthe crystallites.

These results will be useful in the industrial processingof cellulose and for understanding howcellulose is made in plants. l

The bio-material cellulose is found everywhere. Allplants and certain marine animals and bacteriacontain cellulose, as do the materials we use derivedfrom plants, such as wood, paper and textiles.Despite being an important resource, we still havemuch to learn about its structure. Cellulose is acomposite carbohydrate with a basic molecularstructure consisting of a chain of 1000 to 20 000linked glucose (sugar) molecules. In the native state,these polymer chains tend to align into compactneedle-like crystals called microfibrils which arestabilised by weak hydrogen bonds and electrostaticforces and are about 0.01 micrometres across (seeFigure 1). In between the microfibrils, are regionswhere the chains are arranged in a disorderly way.Cellulose fibres like the plant material flax consist ofbundles of many microfibrils.Microfibril structureHowever, not only is the crystal structure of themicrofibrils still controversial but also the exact waythese microfibrils are held together. Are they gluedtogether by a few disordered polymer chains lying onthe surfaces of the crystals or by more extendedregions of disordered cellulose polymers? Thisinformation is important because the mechanicalproperties of cellulose strongly depend on thearrangement of crystalline regions and disorderedregions within the material.

A combination of the neutron inelastic scattering

Highlighting thecomposition of

Cellulose is an extremely importantcomposite material. Even after more tha20 years of extensive research itsstructure is not well understood. Inelastneutron scattering may offer some of thmissing clues

HELMUT SCHOBER, CHRISTOPH CZIHAK & MARTIN MULLER

cellulose

molecule (<10 ̄ )monoclinic unit cell

(10 ̄ )microfibrils (20 200 ̄ )

Crystalline packing withdisordered interface layer

c: fibre axis

PLANT CELL WALLS

5Research team C. Czihak (Univ. Vienna and ILL), G. Vogl (Univ. Vienna and HMI Berlin), M. M ller (ESRF, Grenoble),H. Schober (ILL), Y. Nishiyama (Univ. Tokyo)

biology

Figure 1 (left) Hierarchicalorganisation of cellulose. The polymer(top left in figure) aggregatesto cellulose nano-crystallites,which are interconnected bydisordered interface layers(bottom right). Thesecrystallites are organised inmicrofibrils of diameter 2 to20 nanometres (bottom left)

Figure 2 (below) Thedynamic response ofdisordered regions ofcellulose samples of differentorigin

Importance of hydrogen bondingNeutron diffraction provides the only method that canoffer a detailed visualisation of the hydrogen bondingnetwork in cellulose fibres. The experiments werecarried out on an instrument called D19. This is theonly instrument in the world that can carry out this sortof study (in the past it has also been used to carry outwork on other biological molecules such as otherpolysaccharides, DNA, filamentous viruses).

The idea behind the experiment is simple. Neutronsare fired at the sample of cellulose and the pattern ofneutrons scattered by the sample is recorded. Thispattern can be used to determine the structure of thecellulose. Since hydrogen atoms cause significantscattering of neutrons their position in the cellulose canbe determined as well. To be absolutely sure of theprocess, crystalline fibres were also prepared in whichthe hydrogen in all the OH groups were replaced bydeuterium (deuterium scatters neutrons even morestrongly than hydrogen does) and the same diffractionexperiment repeated.

Recently we studied cellulose II fibres that resultfrom the swelling of fibres in concentrated sodiumhydroxide in this way. There were two suggested modelstructures. Studies using X-ray crystallography couldnot distinguish between the two models. Using ourneutron diffraction data we were able to decide whichof the models was correct.

Figure 1 shows the diffraction patterns recordedfrom cellulose II. The left half of the picture shows thediffraction pattern recorded from normal cellulose andthe right half shows that from deuterated cellulose(each picture has been cut in half and they are shownside by side for comparison). The differences betweenthese two patterns arise from the hydrogen/deuteriumatoms alone. These patterns were used to calculatemaps that show where the hydrogen atoms are locatedFigure 2(a) shows the map, and Figure 2(b) shows thehydrogen bonding network in cellulose that wasdetermined from it. This study is the first three-dimensional description of a hydrogen bonding systemin a fibrous polysaccharide. l

C ellulose is often said to be the most abundant polymeron Earth. It is certainly one of the most important structuralelements in plants and other living systems. In Nature itoccurs as slender rod-like crystalline microfibrils. One of thekey features of cellulose is that each of the units making upthe polymer bears three hydroxyl groups (OH). It is thesehydroxyl groups and their ability to bond via weak hydrogenbonding that not only play a major role in directing how thecrystal structure forms but also in governing importantphysical properties of cellulose materials.

Cellulose can occur in a number of different forms. Naturalcellulose is known as cellulose I, but other forms such ascellulose II, cellulose III and cellulose IV have beendescribed. We are involved in a long-term study at ILL tounravel the fine details of these different forms.

6

New insights into

cellulosestructure

Neutron fibrediffraction offers an

excellent way tostudy the intricatebonding in natural

polymers likecellulose

biologyHENRI CHANZY, PAUL LANGAN &YOSHIHARU NISHIYAMA

Research team H. Chanzy (CERMAV-CNRS,Grenoble), Y. Nishiyama (Univ. Tokyo), P. Langan (Los

Alamos National Laboratory), T. Forsyth (ILL)

Figure 2 (a) the resultingmap of cellulose II;

(b) hydrogen-bondingnetwork in cellulose

Figure 1 A comparison ofthe two diffraction patternsof cellulose from normaland deuterated cellulose

(b)

(a)

InstrumentD19 at ILL

W e then used neutron reflectivitymeasurements to follow the evolutionof the system. Neutrons are reflectedoff an interface just like light. Ourexperiment consisted in bouncing a

neutron beam off the bilayers at the solid/waterinterface and collecting the reflected neutrons on adetector while rotating the sample. The signal fromthe reflected neutrons at the different angles wasanalysed to obtain the structure at the interface.

The reflectivity curves obtained at differenttemperatures, both in the gel, at the transition pointand in the fluid phase, are shown in Figure 1. Thosecurves were interpreted by using mathematicalmodels (continuous lines in Figure 1). This allowed usto determine the structure of the bilayers veryprecisely. Results are shown in Figure 2.

Around the transition point between the twophases we observed a giant swelling (a big increaseof the thickness of the water layer between the twobilayers) and an increase in the unevenness of thefree bilayer. Well above and below the phasetransition the systems present a similar structure andin both phases they are stable. This means that thefluid phase of the free bilayer is suitable for

investigating the interaction of lipids with proteins.Moreover, the giant swelling might be the first directobservation of a phenomenon theoretically predictedmore than 20 years ago. l

All living cells have an outside wall, or membrane,formed mainly from lipids. These long-chainmolecules consist of a head which likes water and ahydrocarbon tail that hates water. They align in adouble layer, or bilayer, with the tails tucked awayinside an internal impermeable core and the headspointing outwards so as to form a protective barrierpermeable only to certain very small molecules. Thestructure of the bilayer plays an important role in theactivity of the cell and is very complicated. Forexample, the membrane also contains proteins whichact as guardians regulating what goes in and out ofthe cell, and whose behaviour is influenced bychanges in the lipid bilayers. Such systems aredifficult to study at the molecular level.

To study membrane structure and behaviour moreeasily, researchers resort to simple models ofbilayers, in particular those based on a class of lipidscalled phospholipids. We have succeeded inpreparing a new, and we think improved, modelsystem a phospholipid bilayer freely floating inwater, which will enable us to study interactionsbetween membrane bilayers and molecules likeproteins.

A double bilayer on a surfaceThe model was created by assembling two bilayers(double bilayer) on a flat solid surface in contact withwater. The first bilayer is strongly adsorbed on thesurface and the second the free bilayer floats justabove it. At a certain temperature depending on thecomposition, phospholipid bilayers change from a gelphase, where the lipid chains are rigid and well-ordered, to a fluid phase, where the chains aredisordered. In cells, membranes are fluid andfluctuate in structure, but since it is more difficult todeposit fluid bilayers on solid substrates we formedthe bilayer in the gel phase and then raised thetemperature to get in the fluid phase.

7

A new model for studying cell membranes is helping researchers to understand howthese important structures work

biology

A new model of a

biological membrane

GIOVANNA FRAGNETO-CUSANI

Research team G. Fragneto and E. Bellet-Amalric (ILL), F. Granerand L. Perino-Gallice (UJF, Grenoble), T. Charitat (ICS, Strasbourg)

l 25.4 Cl 51.5 Cl 55.4 Cl 64.1 C

25.4 C

51.5 C

55.4 C

64.1 C

Figure 2 Possibleinterpretation of modelsthat fit the data measured

GiovannaFragneto-Cusani(right) withcolleagues,Laurence Perinoand Isabelle Grillo

Figure 1 Measured reflectivityprofiles and curves from

mathematical models for the lipiddouble bilayers in D2O (heavywater) at various temperatures.

The x-axis Q is a quantityproportional to the incident angle

gives information on the vertical profile of lipid membranes the thickness of the layers, the roughness of theinterfaces between the layers, and so on. But there is alsothe possibility of observing diffuse, nonspecular scatteringwhich reflects properties along the surface or interfacesuch as lateral fluctuations, elasticity, and density variationsin the layers.Using specular and nonspecular scatteringBy preparing the sample on a transparent silicon substrate,with the neutron beam incident from either the air or substrate side, the diffuse scattering can bemeasured for a range of positive and negative angles ofincidence and exit. This method gives previouslyinaccessible information on the lateral membrane structure,in particular, concerning thermal fluctuations.

For the experiments, samples of stacks of severalthousand bilayers had to be oriented perfectly on a siliconsubstrate. The samples were either partially hydrated(there is a water layer of 1 to 1.5 nanometres betweenmembranes) and kept at constant temperature in ahumidity cell or completely immersed in water.

The samples produced a tremendous amount of diffuseintensity reflecting the softness of the system. The data obtained (Figure 1) on the specular reflectionsshowed that the average distance between the lipidbilayers is 4.98 nanometres corresponding to a water layerof about 1.4 nanometres between adjacent membranes.The nonspecular, diffuse scattering pattern reveals therange of fluctuations in the layers and how they correlateacross several adjacent layers. These results can then becompared with those predicted from theoretical models.

When antibacterial peptides such as Magainin areinserted in the lipid layers we noted that there were drasticchanges in the specular and nonspecular reflectivity as theratio of peptide to lipid increased. This was a result of thedistortion of the lipid membranes inflicted by the presenceof the peptide. The data also provided information on theconfiguration of the peptides with respect to the lipidbilayer. We demonstrated that this technique can reveallateral structures on length-scales from a nanometre up toseveral millimetres. In the future we hope to apply thissame method to many interesting lipid membrane-basedmaterials, including lipid/peptide, lipid/protein or lipid/DNAsystems. l

Lipid bilayers (see page 7) are the basic buildingblocks of biological membranes. Understanding theirphysical properties such as their elasticity, small dynamicfluctuations, and the intermolecular interactions thatgovern how they assemble is extremely important invarious research areas, including engineering syntheticbiomaterials and drug development.

In this last context, the interaction between lipidbilayers of cell walls and antimicrobial peptides (short-chain proteins) is of particular interest, since the latter

show broad antibacterial, fungicidal andantitumour activity. However, the structuralchanges accompanying the interaction arecurrently poorly understood. We are, therefore,studying fundamental physical properties in

model lipid systems withand without antimicrobialpeptides such asMagainin derived fromthe skins and intestinesof frogs.

When combined withthe contrast variationmethod described onpage 4, neutronreflectivity offers uniquepossibilities for studyingthe structure of lipidmembranes at themolecular level. Specularscattering, when theangle of reflection equalsthe angle of incidence,

Neutron reflectivity reveals muchmore about the structure of

multilayer lipid membranes thanwas previously thought possible

biology

Research team C. M nster, and T. Salditt (Univ. Munich), B. Bechinger (MPI Biochemie, Martinsried), R. Siebrecht and

V. Leiner (ILL and Univ. Bochum)

Mapping lipidmembranes

8

Figure 2 (a) A peptide P(green) adsorbed on the lipidbilayers L (blue); (b) reflectivitycurves at increasing peptideconcentration from top tobottom: pure lipid (P/L=0), P/L= 0.025, 0.05, 0.1. Thepeptides affect the multilamellarorder

(a)

(b)

Figure 1 (a) Sketchof two lipid bilayer

membranes; (b) scattering pattern

of a stack ofmultilamellarphospholipidmembranes Q x

Q z

Qz

Christian Münster & Tim Salditt

developing this method (grazing incidence smallangle neutron scattering, GISANS) to obtainscattering signals that reveal peaks at positionscharacteristic of the polymer film s structure.

Dewetting in polymer filmsOne experiment showing the potential of thistechnique was the investigation of the dewetting ofconfined polymer films. Dewetting is frequentlyobserved in daily life, for example, when rain dewetson a car windscreen a once-continuous film breaksup into spherical drops. Using GISANS we couldstudy a confined polymer film, measuring the averagediameters of the droplets and the distance betweenthem. The results gave similar values to thoseobtained with AFM (Figure 2). A simple theoreticalmodel, describing the dewetting of thicker films alsosupported these results.

In samples of polymer blends GISANS can also beused to distinguish the chemical structures andgeometrical arrangements of thedifferent polymer phases usingthe contrast method described onpage 4. Again hydrogen issubstituted with deuterium in onepolymer.

Naturally, the technique is notlimited to polymer samples. Itopens up new possibilities toprobe the topology and chemicalcomposition of any type ofsample surface. l

Polymer films are used in many applications such ascoatings and increasingly in electronics. Withminiaturisation becoming the major technical driver inmost advanced technologies, ultra-thin polymer filmswill be increasingly in demand. However, as thethickness of the film is reduced to that close to thepolymer s molecular dimensions, its properties changefrom those of the bulk material. To achieve an eventhinner film, the polymer molecules must be confined orsqueezed and thus their conformation the internalarrangement of atoms is changed. This so-calledconfinement is not only limited to single polymer filmsbut also those fabricated from polymer mixtures.

To investigate such confined polymer films requiresrather special experimental techniques. Atomic forcemicroscopy (AFM) which relies on a microscopic probeto feel the polymer s topology at atomic resolution canbe used but is limited to very small areas of thesample. Neutron scattering methods therefore have tobe applied in order to obtain a statisticallyrepresentative description of the film. Unfortunately, thethinness of the sample means that the diffractionsignals obtained by passing the neutrons through thefilm are too weak to measure. The trick is instead toreflect the neutron beam off the polymer film at grazingangles of about 1 degree (see Figure 1). We have been

9

Extremely thin polymer films are finding use in many high-tech areas. A new advanced neutron scattering techniquecan probe their structure and chemical composition

soft matter

A new tool to observe

ultra-thin polymer films

PETER MÜLLER-BUSCHBAUM

Research team P. M ller-Buschbaum and W. Petry (TU Munich), J.S. Gutmann, O. Wunnicke and M. Wolkenhauer (MPIPolymerforschung Mainz), R. Cubitt (ILL), M. Stamm (IPF Dresden)

(b)(a)Figure 1 (a) a schematicdrawing of the experimental set-up used;

Figure 2 GISANS (circles)and AFM (crosses) givesimilar data which aredescribed by a simpletheoretical model (solidline)

shampoos. At high concentrations they form bilayers, likethe lipids mentioned on the previous page, that pack intolayered sheets, or lamellae (see Figure 1).

Experiments with detergentsW e were interested to know how the lamellae wereoriented to the direction of flow at different shear ratesand at different concentrations. A full study of thisproblem clearly requires measurement of the orientationin all three spatial dimensions (see Figure 2). For thisstudy a special cell that could maintain uniform shearflow was used that could be rotated in all directions in theneutron beam. The diffracted neutron intensity from thebilayer structure can then be used to monitor thealignment of the surfactant layers.

W e carried out the experiments with a non-ionicsurfactant called C12E O4 in heavy water. When wemeasured the orientations of the lamellae we found thatthere was a large difference between samples with 40per cent of surfactant by weight and those with 60 percent. Surprisingly the 60-per-cent sample showed amaximum in the alignment at an angle of about 45degrees to the flow direction as shown in Figure 3. Thisbehaviour had not been seen in previous experiments onsurfactants but does resemble that seen in dispersions ofrigid, plate-like particles. l

M any familiar materials such as thixotropic paints,foods like tomato ketchup, the mineral and latexdispersions used to coat paper and detergents arecolloids. These are dispersions of particles oraggregates of molecules suspended in a fluid. Acommon feature of most of these materials is that theviscosity changes with the rate of flow. The pressurerequired to maintain a flow decreases as the shearrate, or gradients in speed of flow, increases.

The industrial importance of these liquids hasmeant that scientists have made considerable effort tounderstand the behaviour of these materials in termsof the structure and interactions of the particles andthe fluid. Neutron scattering provides manyadvantages in this work because it can probe theparticles at the required dimensions in the range of1 nanometre to a few micrometres. Exploiting thecontrast obtained by substituting deuterium forhydrogen in molecules, as described in previousarticles, allows particular structural regions to belocated. A class of materials that we are particularlyinterested in, which also behaves rather like theparticulate systems, are concentrated dispersions ofsurfactant (for example, soap) molecules in water.These are the sort of material used in detergents or

10

Pouring tomato sauce orwashing-up liquid out of a

bottle involves complexchanges at the molecular

level, as neutronexperiments are revealing

soft matterSIMONA BARE AND ADRIAN RENNIE

Research team S. Bare and A. R. Rennie (Kings CollegeLondon), E. Bellet-Amalric and G. Fragneto (ILL)

Figure 3 Plots ofthe normalised

diffracted intensityagainst the anglefor C12E O4 at 60

per cent weight anda shear rate of 30

s-1

Detergents, foods and pastes understanding

their flow d

C12E O4

0 20 40 60 80-80 -60 -40 -20

Intensity

Incidentbeam

FlowFlo

0.6

0.5

0.4

0.3

0.20.1

Figure 1 The lamellar phasestructure of surfactantsshowing the inter-lamellarspacing d used to determinethe orientation

Figure 2 Orientation ofparticles or lamellae ina shear flow. Thealignment with respectto the sample cell canbe measured byrotating it in the beam

real material that behaved like thehard-needle model and then studywhat happens as it approaches theglass transition. Fortunately, acommon material, pure ethanol,exists in two forms, or phases, thatbehave just like the hard-needlemodel above and below the glass

transition (Figure 2). Both are crystallinewith the molecules sitting at the nodes ofa cubic lattice. However, in one phase the rotator phase the molecules canrotate like the needles. In the other phase,this orientational disorder has become frozen in. The factthat the glass transition from the first phase to the secondinvolves molecular rotations only, while preserving theoverall crystal structure, gives us an opportunity todescribe this phenomenon just in terms of the possiblemolecular reorientations within a cubic crystal environment a much simpler situation than in a supercooled liquid inwhich the molecules are free to move around.

W e used quasielastic neutron scattering whichmeasures very small energy changes to study the changein dynamics through the rotational glass transition in theethanol phases and compared the spectra obtained withthose calculated for the hard-needle model. We found thatrotational freezing within the real material is closelymimicked by that shown by the hard-needle model. Theimplications of such a finding are far-reaching since itshows that this glass transition can be understood in termsof a purely dynamic phenomenon. l

As has been known since the days of thefamous French chemist Joseph Louis Gay-Lussac (1778-1850), a pail of pure water left outside during winter mayavoid freezing into an ice block if the temperaturedecreases fast enough through the freezing point. Thestate attained by the liquid under such conditions isreferred to as metastable . This means that it takes justa small perturbation to push the liquid into the state that sthe most stable under the ambient conditions, in otherwords, normal ice. For many substances, subsequentcooling of this metastable state usually referred to as a supercooled liquid leads to a transformation into asolid phase known as a glass, which shares somephysical properties with many everyday materials suchas window glass.

The details of how the atoms or molecules movewithin a supercooled liquid, as well as the mechanismswhich drive the solidification of the liquid into the glassphase (glass transition), have mesmerised scientists formany decades, and the topic has been identified as oneof the main challenges in the field of condensed matter.Very active research during the past few decades hasled scientists to realise that the behaviour of glasses andsupercooled liquids is also shared by a good numberof objects, which span from disordered crystallinematerials to cosmological black holes.

The hard-needle modelThe characteristics of these disparate systems inparticular, the way the motions of the particlescomposing the supercooled liquid come to a halt whenapproaching the glass-transition temperature can bedescribed by a conceptually simple model (Figure 1).This is a system of hard, infinitely thin needles whichsit on the nodes of a cubic crystalline lattice and mayrotate freely.

To profit from this analogy and understand glasstransitions better, our research group (a team ofEuropean and American physicists) decided to find a

11

An international team has been carrying outexperiments to test a conceptually simple modelof how a supercooled liquid changes to a glass

liquids & glasses

The glass transition made easy

F. JAVIER BERMEJO

Figure 2 TheGibbs free-energyof the differentphases ofcondensedethanol

Bernard Frick oninstrument IN16which was usedto obtain theseresults

Ui

L

a

i

Figure 1 A system ofhard-needles sitting onthe nodes of a body-centred cubic lattice asimple model forparticles approaching aglass transition

Research team M. JimØnez-Ruiz, A. Criado, F.J. Bermejo and C. Cabrillo (CSICMadrid), F.R. Trouw and R. FernÆndez-Perea (Argonne National Laboratory), H.L wen (Univ. D sseldorf), H.E. Fischer (Orsay), G.J. Cuello (ILL)

into complex networks. It also links water to biologicalmolecules such as proteins. Many of the complexdynamic characteristics of water are thought to be relatedto hydrogen bonding.

Strange acoustic vibrationsIt turns out that at low frequency the molecular vibrationscorrespond to the normal sound, which can be heard byhumans. These vibrations often extend up to frequenciesas high as 1 terahertz (one million million vibrations asecond) and are responsible for many properties ofliquids. In the case of water, however, researchers haveobserved that the velocity of these high-frequency soundwaves is much greater (about 3000 metres per second)than that of normal sound (about 1500 metres persecond). Scientists consider this effect quite anomalousand they had not yet been able to offer a simpleexplanation.

Inelastic neutron scattering is often used to study thevibrations of atoms, as mentioned on page 5, and it is thistechnique that we used at the ILL to investigate thevibrations in water. We employed heavy water becausedeuterium atoms are much stronger scatterers thanhydrogen, and are comparable to oxygen in this respect,thus giving us a strong neutron spectrum for analysis.Analysing the results of the neutron scatteringexperiment, we have been able to give a good accountfor the strange behaviour of the high-frequency molecularvibrations in water. We have identified two differentvibrations, that reflect the presence in water of small groups of molecules tightly bound together. Thereare the vibrations inside each group of molecules (theseare optical vibrations, so-called because they diffuseoptical light) and the high frequency sound which allowsthe vibrations of one group to propagate to another groupof molecules (acoustic vibrations).

W e have been able to compare the energies of thedifferent vibrations (Figure 1) with previous data obtainedfrom the companion technique of X-ray scattering, anexperiment performed at the nearby EuropeanSynchrotron Radiation Source in Grenoble. It has beenparticularly satisfying to see that the neutron scatteringresults give a more complete account of the X-ray data.The neutron scattering experiment also enabled themeasurement of the intensity of the two vibrations atdifferent wavelengths, as shown in Figure 2. l

12

The sound of

water

A team of German andItalian scientists has been

investigating soundpropagation in water, whichshould help shed light on its

complex physical andbiological behaviour

liquids & glassesCATERINA PETRILLO

Research team C. Petrillo and F. Sacchetti (INFM, Perugia), B. Dorner (ILL), J. B. Suck (TU Chemnitz)

w(meV)

Squared structure factor

Q (̄ -1)

Q (̄ -1)

Figure 2 Theintensity of the

different vibrations inwater. The dotsrepresent the

intensities of the low-frequency vibrations,the triangles those ofthe high-frequencyvibrations, and theopen symbols the

sum of both

Figure 1 The energy ofthe different vibrationsin water. The lines are

deduced from theneutron scattering

experiment; the dotsare the results of the

previous X-rayexperiment. The

dashed lines representthe expected behaviour

of the sound waveswhen the velocity is

assumed to be about3000 metres per

second or 1500 metresper second

W ater is certainly the most common substance in oureveryday world but, most importantly, it is vital to theexistence of life. Because of the basic role that waterplays in living systems, it is one of the most studiedliquids. And because of its complex and uniquecharacteristics it is also one of the most fascinating. Oneof the properties of water that deserves detailedinvestigation is its dynamic behaviour how theconstituent molecules vibrate which is known to affectbiological systems. For example, the functions ofbiological molecules are driven by the interplay of thevibrational and diffusional motions of the surroundingwater molecules. A weak type of bonding calledhydrogen bonding links these water molecules together

point, eventually forming a defective version ofcubic ice instead of hexagonal ice. Theformation of cubic ice was unexpected as it is metastable in the bulk state, converting tohexagonal ice above 200K.

The depression of the freezing point is greater in the smaller-sized pores, and recently

we found that we could super-cool water even furtherby confining it in a new type of ordered mesoporoussilica (see right) with pores between 2.5 to 3.5 nanometres across.

Neutron diffraction studies with heavy waterperformed on ILL instrument D4 over a range oftemperatures revealed that the water remains liquiddown to 235K and then undergoes a reversibletransformation to cubic ice. This low-temperature state ofwater offers the possibility of investigating waterproperties in a very unusual condition. The enhancednetwork connectivity due to the hydrogen bonds meansthat the properties are very different from those ofnormal water. However, the water structure remainsdelicate in the sense that it is easily modified byenvironmental influences. It is therefore conceivable thatthese features are significant in defining the behaviour ofconfined or interfacial water in biological systems. Waterdoes seem to have some perplexing characteristics butthere is no doubting the important influence that it hasover our lives! l

The properties of water and ice are familiar toeveryone and the simple chemical composition, H2O,is the first (and in some cases the only!) formula thatanyone learns at school. Yet, the behaviour of wateris still not well understood. Its structure at a molecularlevel is remarkably complex as the result ofinteractions between the water molecules calledhydrogen bonding. These weak forces are, forexample, responsible for the arrangement ofmolecules in the normal crystalline form of water,hexagonal ice. Each molecule has only four nearestneighbours leading to an unusually low density, whichis why ice floats on water.

Hydrogen bonding also influences the structuralproperties of liquid water, as neutron or X-rayscattering studies reveal. The pattern for most liquidschanges very little with temperature over the whole ofthe liquid range. Water, however, is different. It ismuch more sensitive to temperature which affects thehydrogen bonding and thus the relative positions andorientations of neighbouring molecules.

One interesting feature is that liquid water can becooled well below its normal freezing point. Variousstudies have shown that the degree of hydrogenbonding increases in this metastable regime, leadingto a more random network structure characteristic of alow-density glassy ice formed when water iscondensed onto a cold plate at temperatures lowerthan 140K. The properties of bulk super-cooled waterin this region are more like a gel than a liquid.

W ater in a solid matrixW e have found that a similar effect occurs, quitenaturally, when water is confined in the pores of asolid matrix, such as a mesoporous silica with a poresize of about 10 nanometres. Furthermore, the waterremains in the liquid state below its normal freezing

13

Water is a surprisingly complex substance forming many structural variations in different physicalenvironments. Under certain conditions, it can even remainliquid at 40 degrees below its normal freezing point!

liquids & glasses

Super cool water!

JOHN DORE

Research team J. Dore (Univ. of Kent, Canterbury), C. Haggenm ller (Univ. M nchen), P. Behrens (Univ. Hannover),H. Fischer (Orsay)

The geometry ofmesoporous MCMsilicas; a) hexagonal,b) cubic

A schematic model of low-temperature water basedon the structure of low-density amorphous ice

Instrument D4

Water exists in anextraordinary number of

forms, some of themdiscovered only recently.

Neutron scatteringexperiments are helping

researchers to understand

the subtle forces thatgovern the transitions from

one form to another

There is another intriguing aspect of the new phases.Having been observed in totally different regions of water sphase diagram, the amorphous phases and ice XII wereoriginally thought to be unconnected. However, HDA wasfound to be contaminated with an unidentified crystallinephase. Using ILL instruments we have since shown that thephase is in fact ice XII! This implies that from an stability poinof view the formation of HDA competes with that of ice XII. Byaltering the compression rate it is possible to favour theproduction of either ice XII or HDA. Under rapid compressionthe formation of ice XII dominates.

Ice XII, therefore, is more common in water s phasediagram than originally thought, existing in at least twoseparate regions. The study of the conversion of hexagonalice to ice XII contributes valuable clues to the understandingof the formation of amorphous and crystalline phases underpressure. Crystallisation implies a reorganisation of water shydrogen-bond network. This requires a high molecularmobility as can be provided by transient melting. Moreexperiments are certainly necessary to determine definitelythe lowest temperature at which such melting can occur. l

Although, water has been the object of extensiveexperimental and theoretical investigation it still rewardsus with new and unexpected properties. Because of itshydrogen-bonded molecular network (as mentioned inthe previous article) the structure of crystalline water(ice) is strongly influenced by pressure and temperature.To date, not less than 12 different ice phases (structuralforms) are known. The stability of these phases over arange of temperatures and pressures can be shown ona so-called phase diagram (see right). The hexagonalform is the stable one at ambient pressure. Several icephases are metastable in their region of existence onthe phase diagram, transforming into more stable formsof ice if we wait long enough.

Melting ice at -200 CResearchers discovered so far two distinctlydifferent non-crystalline (amorphous) forms ofice containing random networks of watermolecules, and only recently also the twelfthcrystalline ice phase. Called ice XII, it wasproduced at about half a gigapascal pressure(equivalent to a load of 5 tonnes per squarecentimetre!) and a temperature of -13 C. XII isbuilt up from seven and eight-membered rings ofwater molecules leading to an incredibly densestructure. The two amorphous phases werefound at much lower temperatures. High densityamorphous ice (HDA) is produced bycompressing hexagonal ice by applying apressure of at least 1 gigapascal at temperaturesof about -200 C. On heating to -150 C HDA transformsinto a low-density amorphous phase (LDA). Oneinteresting question here is whether the collapseof hexagonal ice to HDA is analogous to meltingat normal pressures and temperatures where aslight pressure applied to the delicate openstructure of hexagonal ice at just below 0 Ccauses it to liquefy (the molecular mobilityinvolved, though, in this melting is orders ofmagnitude slower). Can we therefore think ofHDA as a type of frozen liquid water but existing

at -200 C?

14

The many phases of

waterchemistry & structure

MICHAEL KOZA & HELMUT SCHOBER

Research team M. M. Koza, A. T lle and F. Fujara(Univ. Dortmund), H. Schober and T. Hansen (ILL)

The phase diagram of water; lrepresents the regionin which ice XII has been observed (temperature:260K; pressure 0.55: gigapascals). Note that thisregion is fully surrounded by ice V

(a)

(b)

(c)

Transition of ice XIItowards hexagonal iceupon heating. The topphotograph (a) shows

ice XII as recovered fromthe pressure cell at lowtemperature. In contrast

to high-densityamorphous ice it has amilky appearance. Thefollowing photographs

show the flocking of iceXII as it transforms to the

lower density formscubic ice and hexagonal

ice (b and c)

prepared series of rocks, and used them topredict the extent of oil generation inrecently-drilled oil wells. The figure below

illustrates how extensive the fractal properties of asedimentary rocks can be. The figure shows acombination of SANS data taken on three different ILLand ORNL instruments. It may be seen that all threedata sets overlap smoothly with no adjustableparameters. To our knowledge, these data representthe largest range to date over which fractal behaviourhas been observed in a natural system. Such anextent of fractal microstructure in a rock is remarkable,when compared with the limited size range over whichthe fractal properties are usually observed typically1.3 orders of magnitude.

This study extends the widest fractal length rangepreviously observed in sedimentary rocks, whichcovered two decades (orders of magnitude) in lengthscale and 7.5 decades in intensity, and shows thatsedimentary rocks are in fact one of the most extensivefractal systems found in Nature. As observed inPhysical Review Focus: the constancy of the fractaldimension over so many scales is astounding ,considering what a messy, heterogeneous materialsedimentary rock appears to be ... this study willenhance the idea that you can describe rock withsimple concepts ... There will be certain bona fide usesof fractal concepts, and one of them will be in thegeological sciences. l

Sedimentary rocks are dirty systems. They areformed from a mixture of organic and inorganic debrisdeposited in an aqueous environment of lakes ormarine incursions, in other words, from mud. The mudgets buried and compacted at elevated temperaturesover geological periods of time. The resultingsedimentary rock develops a pore system through along and complex redistribution of matter between itssolid and liquid components. In the end, many typesof sedimentary rocks have a fractal structure, whichmeans that on average their pore structure looks thesame at any magnification. This has been longrecognised by traditional geology, where variousobjects are used to define the scale of rocks shown inphotographs.

On the microscale, cold (low energy) neutrons aremuch better than the direct imaging techniques (likeoptical or electron microscopy) to detect fractal micro-structure in rocks. The imaging techniques reveal anoverwhelming wealth of detail on every scale, andtime-consuming statistical analysis of many imagesneeds to be done to determine the generalcharacteristics of the averaged rock micro-structure.In contrast, neutrons penetrating rocks are naturallysorted according to the internal microstructureaveraged over the volume of the entire sample(typically 1 centimetre square and 0.1 centimetrethick).

Rock micro-architectureW e have used small angle neutron scattering (SANS),see page 3, at ILL and at the Oak Ridge NationalLaboratory (ORNL) in Tennessee to determine themajor properties of rock micro-architecture theinternal surface area of the pores and their fractalscaling. When applied to petroleum geology, theSANS technique can detect the deformation of thepore space caused by the internal pressure due to thethermally-driven decomposition of large organicmolecules into hydrocarbons, as well as observe theconsequent transport of these hydrocarbons throughthe increasingly larger pores. We observed thesephenomena in both the natural and laboratory-

15

Neutrons can penetrate porous rocks, revealing their‘fractal’ nature and how trapped organic matter istransformed into oil which then seeps out of the rock.These findings help companies prospecting for oil

materials

Oil generation in fractal rocks

ANDRZEJ RADLINSKI & PETER LINDNER

Research team A. P. Radlinski (Australian Geological Survey Organisation, Canberra), E.Z. Radlinska (The Australian National University, Canberra), M. Agamalian and G. D. Wignall (Oak Ridge National Laboratory), P. Lindner and O. G. Randl (ILL)

Scanning electronmicrograph of an oil-bearing sedimentary rock

A view of the detectortube of the 80-metresmall angle scatteringinstrument D11

Scattering intensity dS/dW

(cm-1)

Scattering vector Q (̄ -1)

SANS data fromsedimentary rockshowing that the pore-rock interface is asurface fractal overthree orders ofmagnitude in length-scale and 10 orders ofmagnitude in cross-section (intensity)

neutrons while being subjected to conditions similar tothose in real processing. For example, we can followwhat happens as the precipitates form during heattreatment by monitoring the small-angle scatteringpatterns over time.Microstructure of superalloysOur recent experiments have focused on investigatingthe microstructure of alloys of nickel with aluminium andmolybdenum. The strength of these alloys is influencedby how well the dimensions of the crystal lattices of theordered precipitates and the surrounding matrix match.The molybdenum serves to modify any mismatchbetween precipitates and matrix. A large mismatchintroduces strain into the material and affects the rate atwhich the precipitates grow. We know from electronmicrographs that when the mismatch is small,precipitates are almost spherical, but they becomecuboidal (with rounded corners and edges) withincreasing misfit. We followed the evolution of the sizeand shape of the precipitates with small angle scatteringat various temperatures. Figure 1 shows the results forone particular situation.

Titanium is another element often incorporated innickel alloys. Like those with nickel and aluminium,nickel-titanium alloys also form ordered precipitates. Theprecipitation sequence is, however, more complex.While in the case of aluminium the cuboidal precipitatesare the final precipitate phase, with titanium they areonly an intermediate stage, and an ordered hexagonalphase develops after sufficient ageing. Again, small-angle scattering experiments offer insight into thedevelopment of the microstructure. The small-anglescattering patterns of single crystals reveal the differentsymmetries of the scattering from the cuboidal and thehexagonal, plate-like precipitates. Figure 2 showsexamples taken at three different temperatures. Theyindicate various stages of transition from the cuboidalphase forming initially, to the plate-like phase.

These data along with results from electronmicroscopy and large-angle X-ray diffraction serve as abasis for modelling and simulating the behaviour ofalloys under real processing and service conditions, inthe interest of improving the properties of thesematerials. l

N owadays, nickel-rich alloys are widely used instructural components which are exposed to very hightemperatures. The so called nickel-base superalloyshave a particularly high mechanical strength and arevery resistant to corrosive environments. They areused, for example, to make turbine blades orchemical-reaction vessels. The high-temperaturestrength of these alloys is primarily due to very strong,often microscopically small crystalline particles, orprecipitates, which are embedded in large numbers inthe nickel-rich matrix.

The technological development of superalloys isquite advanced, but there is still a broad need forreliable experiments to understand better the basicphysical processes underlying the changes that theprecipitates might undergo during use changes insize, shape and population. The longevity of structuralcomponents depends on the stability of an optimisedmicrostructure under various environmental changes.

Because the precipitates are a nanometre to amicrometre across, small angle neutron scattering isextremely useful to study their structure andbehaviour (see page 3). Recently, our research teamat the ETH Z rich, together with researchers at ILLand the Hahn Meitner Institut in Berlin, havedeveloped a new device a high-temperature cell in which samples of alloy can be studied with

Probing the hidden strength of

Experiments at ILL using newlydeveloped equipment are

following the microstructuralchanges in nickel-base alloys at

high temperatures. The work willhelp to optimise industrially

important materials

materialsGERNOT KOSTORZ

Research team J. Hecht, M. Kompatscher, G. Kostorz, J.-M. Schneider and B. Sch nfeld (ETH Z rich), B. DemØ (ILL)

Figure 2 Small-anglescattering of nickel-richnickel-titanium single

crystals homogenised at1440K and aged in situ for(a) 46 minutes at 1240K,(b) 10 minutes at 1220K,(c) 43 minutes at 1200K.One cuboidal main axis isthe horizontal direction.The inclined streaks are

perpendicular to the close-packed planes of the face-centred cubic structure of

the matrix

alloys

(a) (b) (c)

Figure 1 (a) Small-anglescattering patternmeasured on D11 for anickel-rich nickel-aluminium-molybdenumsingle crystal aged for 3hours at 1073K. The dottedlines represent lines ofequal intensity ascalculated for a cuboidalprecipitate; (b) outline ofthe best-fitting cuboidalshape. The size is about 20nanometres

[100]

[011] 0.05 ̄ -1

(a)

(b)

16

formation, evolution, and change in size.These effects can be simulated by bombarding

samples of steels with charged helium atoms (alphaparticles) in particle accelerators. It is difficult to observethe helium bubbles with an electron microscope becausethey are so small (see left). Small angle neutronscattering, on the other hand, is a particularlyappropriate technique for this study, for several reasons:the neutrons have a wavelength of similar size to theseparticles; they penetrate relatively thick samples easily(up to several millimetres); and finally, because neutronsin a magnetic field can sense the magnetisation of thesteel, this property can be used to enhance the visibilityof the non-magnetic helium bubbles.

W e investigated samples of a particular steel F82H,which were bombarded with helium particles at thecyclotron facility at the Karlsruhe Research Centre. Theywere then heat-treated at 525 C, 825 C and 975 C fortwo hours under high vacuum. These implantationconditions are representative of the effect of helium infusion reactors.

The neutron scattering measurements were thencarried out at ILL. From the scattering data we calculatedthe relative numbers of helium bubbles of different sizes(in terms of the total volume of bubbles of a given radius)for three different annealing temperatures. From theresults shown in the figure below we concluded that thethermal treatment and coalescence of the initiallyimplanted helium atoms is responsible for the growth oflarger bubbles at the expense of the smallest ones.Neutrons thus have turned out to be an essential tool forcharacterising the bubble-growth mechanism in thesesteels in a quantitative way. l

Scientists hope that our future energy needs will bemet by electricity produced from fusion reactors. Theidea is to reproduce the nuclear reactions that takeplace in stars, where a mixture (called plasma) ofelectrically-charged light elements (such as hydrogen)is burnt to heavier ones (such as helium) producingvast amounts of energy in the form of radiation.Thereaction conditions extremely high temperatures andpressures and powerful magnetic fields are extremelydifficult to realise in a man-made device. One of theproblems to be solved is the choice of suitablematerials for the reactor. The thermal stresses and veryhigh radiation will induce dramatic changes in both themechanical properties and the microstructure of thereactor walls, especially in the first wall the structuralelement closest to the plasma.

Radiation damageThe fusion process induces radiation damage such thathelium accumulates in the first wall. Helium is notsoluble in solids and it forms gaseous bubbles, whosegrowth at higher temperatures causes the swelling andconsequent embrittlement of the wall components.These phenomena are key factors determining the life-time and the reliability of the steels that would be used as structural materials for fusion reactors. In order tounderstand better how these helium bubbles grow, we investigated the influence of temperature on their

17

Neutrons probe the growth of heliumbubbles in a type of steel developed forfuture fusion reactors

materials

Helium bubbles in steel

GIOVANNA CICOGNANI & ROLAND MAY

Research team R. Coppola (ENEA-Casaccia), G. Cicognani, R.P. May(ILL), M. Magnani (ENEA, Bologna), A. Moeslang (FZ Karlsruhe)

A transmissionelectron micrograph(left) of helium bubblesin F82H steel afterimplantation of heliumat a temperature of250 C

The distribution of heliumbubbles according to size (inangstroms) obtained from theneutron scattering data at a)250 C, b) 825 C and c) 975 C.The shaded areas representthe 80% confidence band. Atthe lower temperatures (a andb) we observe one well-defined peak centred around1.5 nanometres correspondingto a dense population of smallbubbles. At 975 C (c), asecond distribution of bubbles, 10 times larger, appears. Thedensity of both bubblepopulations is almost oneorder of magnitude larger thanat 825 C. There is an increaseof a factor of five in the relative value ofthe bubble volume fractionfrom 250 C to 825 C and ofabout 50 per cent from 825 Cto 975 C

the nanometre range, surface effects aredominant which significantly influences the

magnetic properties. Such composite materials offer newtechnological possibilities. For example, electronicinformation is often stored as tiny magnetised areas on adisk or tape, and MNCGs offer the possibility for high-densityinformation storage. Used in new ultra-fast switching devices,MNCGs could also replace classical electronic devices basedon semiconductors.

Polarised neutron reflectivityTo tailor MNCGs with specific properties, researchers needto know how the clusters are distributed down through thesurface layer, the range of sizes of cluster and how theyinteract with each other. Therefore, a non-destructive methodis needed to analyse the sample s structure and especiallythe magnetic properties. A promising method of choice ispolarised neutron reflectivity

As explained in the Introduction, polarised neutrons (theirmagnetic moments aligned) provide an indispensable tool forstudying magnetic behaviour. Reflecting the polarised beamfrom a sample at angles just grazing the surface should helpunravel the magnetic properties of the implanted magneticparticles.

First of all, we performed reflectivity experiments withunpolarised neutrons on a nickel-implanted MNCG todetermine the range of sizes of the nickel clusters at variousdepths. The glass was bombarded with two beams of nickelions of different energies and intensities. Figure 1 shows the

reflectivity curve (open circles) we obtained. We thencompared the result with that obtained from anothermethod called Rutherford backscattering spectroscopy(RBS). Even though both techniques give the sameoverall chemical structure, RBS cannot reveal thesample s magnetic properties related to the nickel-cluster distribution.

This unpolarised reflectivity experiment is the firststep to set a more general approach to MNCGs usingneutron-based techniques, to study their magneticproperties. Recently, we went on to perform reflectivityexperiments with polarised neutrons. We obtainednew promising looking data which is currently beinganalysed. Soon, we hope to shed some light on thephysical nature of magnetic MNCGs. l

A favoured method of makingthe complex materials required, forexample, by the electronicsindustry is ion implantation,whereby a material is bombardedwith highly accelerated ions(charged atoms or molecules) sothat they become embedded justbelow the surface. Of particularinterest are glasses implanted withmetal ions, which form tiny clustersof atoms, only nanometres across,

in a thin surface layer.These metal nanocluster composite glasses, or

MNCGs, show so-called nonlinear optical propertieswhich are useful in optoelectronics using light ratherthan electricity to transfer information. In particular therefractive index of the MNCGs varies depending onthe intensity of the light (optical Kerr effect). Thisfeature could be exploited in high speed opticalswitching devices which would operate at rates ofonly a few thousand billionths of a second.

Futhermore, MNCGs obtained by implanting ionsof metals like iron, nickel or chromium are importantfor their magnetic properties. When the size ofmagnetic particles inside a non-magnetic matrix is in

18

Reflections on novelglasses for fasterelectronics

Glasses with tiny metalclusters embedded inthem could lead to a

new generation ofinformation storage and

fast-switching devices

materialsCLAUDIA MONDELLI & RALF SIEBRECHT

Research team C. Mondelli (ILL) and R. Siebrecht (ILL and Ruhr-Univ. Bochum), G. Battaglin, E. Cattaruzza and F. Gonella (INFM-Venice),

F. D Acapito (INFM and ESRF, Grenoble), P. Mazzoldi (INFM-Padua)

Figure 1Reflectivity

measurements(open circles) takenat ILL; the solid lineis put in to fit to the

data

Figure 2 Theexperimental in-depth nickel distributionmeasured with Rutherfordbackscatteringspectroscopy (RBS), left,and reflectivity, right

Ralf Siebrecht withinstrument ADAM

used to obtainthese results

Depth profile RBS Reflectivity 2dx10-5

1.0 0.8 0.6 0.4 0.2 0.0

0

500

100

0

150

0

200

0

250

0

0

500

100

0

150

0

200

0

250

0

2.0 2.5 3.0 3.5

120 ¯

90 ¯

200 ¯

100 ¯

Glass SiO2

Depth (̄)

Depth (̄)

Surface

unexpected results. Our results agreed with previousstudies on holmium films as concerns the ordering ofthe magnetic structure as a function of temperature.However, the transition temperature from disorder tomagnetic ordering was only 105K, considerably lessthan in the bulk, and the turn angle was bigger at 20K(46 ). Below 20K, the helical magnetic orderingremained incommensurate. However, we had expectedmuch more dramatic changes in a holmium film whichis only slightly thicker than the distance required toachieve one complete turn in the spin helix. Assuming aturn angle of 30 as in the bulk at 20K, 12 latticeplanes, or single atomic layers, are required for acomplete turn, which amounts to a film thickness of 3.4nanometres. Theorists had suggested that the turnangle of the spin helix should get smaller as thenumbers of layers of holmium atoms is reduced, andthat below a film thickness of 10 nanometres the helicalarrangement of spins should collapse to a moreconventional ferromagnetic state.

Instead we found that the fewer the number of layersthe larger the turn angle, thus shortening the pitch ofthe helix. This indicated that the spin-helix structure israther robust. In future we hope to investigate whathappens in even thinner films which are less than thedimension of a complete helical turn. We clearly stillhave a lot to learn about spin-helical magnetism. l

Today it is possible to grow ultra-thin magnetic filmsof very high quality. This allows us to raise thequestion of how magnetic properties of the bulkmaterial are modified in thin films either because ofthe effect of reducing interactions to just twodimensions or because of interactions with adjacentlayers of other materials. Information on such films isimportant not only in understanding complex magneticbehaviour at a fundamental level but also in thedesign of novel magnetic nanostructures integrated innew electronic devices. The most completeinformation on such ultra-thin structures comes frompolarised neutron scattering, especially when themagnetic ordering is complex. In this context, wehave been studying the magnetic behaviour in singleultra-thin films of holmium. This rare-earth metal is anexcellent model system for such studies since itsstrong magnetic moment and long-range magneticorder result in intense magnetic scattering patterns.

Holmium s helical magnetic structureHolmium is also interesting because in the bulk, themagnetic moments order helically below a transitiontemperature of about 132K. The spins orderferromagnetically in the basal plane that is, within amonolayer of atoms in the crystal. Then from onelayer to the next, their orientation turns a certainangle, setting up a magnetic spiral perpendicular tothe planes. However, the periodicity of the helix doesnot coincide with that of the crystal planes, in otherwords, the spin helix is incommensurate . At thetransition temperature, the turn angle is 50 butdecreases continuously to about 30 at 20K. Below20K the magnetic moments tilt upwards out of thebasal plane locking into a commensurate helicalmagnetic-cone structure with a turn angle of 30 (Figure 1).

W e were interested in investigating the changes inmagnetic ordering of very thin holmium films only 4.6nanometres thick. The film was grown on a sapphiresubstrate sandwiched between two layers of niobiumand yttrium. This prevented the holmium fromoxidising and ensured that both surfaces of the filmwere symmetrical structurally and magnetically.

The magnetic scattering data gave us some

19

Investigating very thin layers of complex magnetic materials onlya dozen or so atoms thick is difficult. Neutron scattering, however,rises to the challenge, offering unique insights

thin films

in very thin filmsMagnetic spirals

VINCENT LEINER & HARTMUT ZABEL

Research team V. Leiner and R. Siebrecht (ILL), D. Labergerie, Ch. Sutter, and H. Zabel (Univ. Bochum)

Commensuratemagnetic conestructure

0 20 132 T(K)

Incommensurate spin helix

Figure 2 Radial neutronscans of the magnetic peakfor the 4.6 nanometre-thicksingle epitaxial holmiumfilm for differenttemperatures. The peakshifts to higher scatteringvectors due to anincreasing turn angle of theholmium spin helix withincreasing temperature.Simultaneously theintensity decreases due toa loss of long-rangemagnetic order withincreasing temperature. Inthe inset, the magneticorder parameter is plottedas a function oftemperature

Figure 1Phase

diagram ofbulk holmium

will be high. If the orientation of one of the magnetic layersis then switched by a magnetic field fewer electrons will bescattered and the resistance will fall dramatically.

A more complex spin arrangementIn fact, the actual arrangement of spins in these multilayersturns out to be quite complicated, and we have beenstudying just how the electron scattering leading to GMRarises. The multilayer structure we studied is shown inFigure 1a. It consists of alternating layers of iron andchromium, 6.8 and 0.98 nanometres thick respectively. Ineach iron layer the spins are all aligned, but themagnetisation of each iron layer also interacts with that ofthe neighbouring iron layers through the chromium layers.For a certain thickness of the chromium layer the ironlayers are antiferromagnetically ordered, in other words,the spin orientation of neighbouring iron layers is in theopposite direction. This arrangement shows a typical GMReffect.

Polarised neutron reflectivity is a very good techniquefor exploring the spin arrangements in these materials.Reflecting polarised neutrons off the surface and interfacesof the magnetic layers reveals not only information aboutthe layer structure but also the layers magnetisation. Wemeasured two components of scattering: the specularreflections (where the angle of reflection afequals theangle of incidence ai) which gives a vertical profile of thelayers including thickness (Figure 1a); and off-specularreflections which reveals variations horizontally along thelayers (Figure 1b).

Figure 2 shows the experimental data for specularscattering. The intensity variation along the diagonal ridgereflects the periodic composition of the multilayer. But thereare also intensity wings on the specular line due to off-specular magnetic scattering and this reflects the lateralmagnetic structure of the layers. Theoretical calculationssuggest that what this shows is that there are tinydomains in each layer, about 200-300 nanometres indiameter and extending column-like across all the multi-layer. They are antiferromagnetically coupled (Figure 1b).W e are still analysing the results but we think that thesemagnetic domains play an essential role in the electronscattering process responsible for the GMR effect. Webelieve they offer a more exact description than the simpleexplanation usually given in terms of stacks ofhomogeneously-ordered antiferromagnetic layers. l

20

thin filmsVALERIA LAUTER-PASYUK, HANS LAUTER &BORIS TOPERVERG

Research team V. Lauter-Pasyuk (University of Munich and JINR Dubna), B. Toperverg, (ILL and PNPI St Petersburg), H. J. Lauter (ILL), O. Nikonov (ILL and JINR

Dubna), E. Kravtsov, M. A. Milyaev, L. Romashev and V. Ustinov (IMP Ekaterinburg)

Figure 2 (a) An intensitymap of specularscattered and off-specular scatteredneutrons from the iron-chromium multilayer as afunction of the incidentand outgoing scatteringangle. The colours reflectthe scattered intensity ona log-scale. Thehorizontal colour linescrossing the pictures arebackground effects; (b) amodel calculation tosimulate theexperimental results

A few years ago, French and German scientistsdiscovered a remarkable phenomenon called giantmagnetoresistance (GMR) which is revolutionising thestorage of computer data. Data are stored on harddisks as minute magnetised areas whose fields aredetected as changes in electrical resistance in a readhead an effect called magnetoresistance. GMR is200 times more powerful than ordinarymagnetoresistance.

GMR arises in ultra-thin multilayers of magneticmaterials. The simplest arrangement consists of twomagnetic layers with a nonmagnetic layer in between.The conventional explanation of GMR is as follows. When a current passes through the layers, theelectrons oriented in the same direction as theelectron spins in a magnetic layer go through whilethose oriented in the opposite direction are scattered(leading to electrical resistance). If the sets of spins inthe two magnetic layers are oppositely aligned thenmany electrons will be scattered and the resistance

Figure 1 (a) an iron-chromium multilayer withscattering geometry forspecular reflection. The

arrows inside the multilayerindicate the idealised

antiferromagneticmagnetisation direction in

the iron layers; (b) thesame multilayer with

magnetic domains creatingoff-specular scattering seenin the scattering geometry.The spin configuration of

the iron-layer magnetisationis idealised

antiferromagnetic order, butthe spins are turning

somewhat into the externalfield direction

(a)

(b)

Explaining giant

magnetoresistancePolarised neutron reflection experiments are revealingthat the GMR effect used in the latest magnetic storagedevices may be more complicated than at first thought

(this is a classical process), but it is also possibleto tunnel through the barrier. Tunnelling isa curious quantum process which is aresult of the statistical nature of thequantum mechanics that there is always asmall probability that the molecule can exist in analternative state. Solving this fascinating questionmay have important practical consequences,since magnetic molecules could be used in future asmemories for new generations of computers,and the study of their magnetic reversal helps todetermine the size-limit for information storage.

W e used inelastic neutron scattering at very lowtemperatures to investigate the transitions. Theneutron can exchange energy with the cluster, whenthe cluster passes from one energy state to another.It is not possible to see the tunnelling effect directlybecause it is too low in energy. However, we hopedto see it indirectly by a broadening of some energylevels. This broadening modifies the inten-sity andenergy of the thermal transitions.

W e found that we could see the transi-tionsbetween these energy sublevels very clearly (seeright). At 1.5K the spectrum revealed a well-definedtransition from the lowest energy level (M =10) to the first excited one (M = 9). At highertemperatures, the other levels start to bepopulated. At 23.8K all levels are popu-lated.The giant spin then starts to break up. At23.8K, we could see observe how thetransitions near the central peak areinfluenced by tunnelling. We were able toascribe the observed effects to a termrelated to tunnelling in the equationdescribing the energy of the system.

Thanks to the neutron probe, the energyterm responsible for the tunnelling wasdetermined very precisely. Althoughextremely small, it could explain the maintunnelling transitions. More work needs to bedone in understanding the whole tunnellingeffect. l

Small clusters of atoms that behave as tiny magnetsare currently of great practical interest as potentialinformation storage devices. They are also of theoreticalimportance because their behaviour lies on the borderbetween classical and quantum physics. Among them,molecular clusters are very attractive because of theirsimplicity: they are all identical objects, arranged overregular arrays, and weakly interacting with each other.

These large molecules contain typically a dozensimilar metal ions whose magnetic spins (arising fromthe unpaired electrons) combine into one giant spin .These molecules are fascinating because even thoughthey are quite large, they still behave as quantumobjects. A molecule containing 12 manganese ionslinked with acetate groups (Mn12-ac) is the best studiedof these spins clusters. The direction of the giant spin(S) is along an axis perpendicular to the plate-like shapeof the molecule. According to quantum rules, the totalspin S could have 2S+1 possible projections along thisaxis, yielding 2S+1 energy levels for the molecule. InM n12-ac, the arrangement of the manganese spinsinside the cluster yields a total spin S = 10, that is, thereare 21 magnetic energy levels (M ) from -10, -9, to +9,+10.

Magnetic transitionsW e are interested to know how the giant spin flips fromone energy level to another. This is a challengingquestion for theoreticians. Roughly speaking, twomechanisms could be considered. The energy barrierbetween the levels can be overcome by supplying heat

21

A European team has been studying thequantum ‘peculiarities’ in a large molecule that behaves as a mini-magnet

magnetism

in molecular magnetsQuantum tunnelling

ISABELLE MIREBEAU

Research team I. Mirebeau and M. Hennion (LLB), H. Casalta (ILL),H. Andres and H. U. G del (Univ. Bern), V. Irodova (Inst. KurchatovMoscow), A. Caneschi (Univ. Florence)

Energy spectra of theM n12-ac spin cluster attwo temperatures

View of the core of theM n12-ac cluster, in whichonly the metal atoms andthe bridging oxygens (smallspheres) are shown. Thetotal spin S=10 is built byan antiparallel arrangementof four Mn4+ ions (largespheres), and eight Mn3+

ions (medium spheres)

IN5 technicianSteve Jenkinsworking on theexperiment

exponents. These describe how typical bulk properties likemagnetisation, specific heat or magnetic susceptibilitychange with temperature as the phase transitiontemperature is approached either from above or below. Thecritical exponents reflect the structural symmetry of thematerial and its dimensionality whether the ordering is inone, two or three dimensions. According to S. Kawamura,the chiral ordering should have its own characteristic set ofchiral critical exponents and would represent a newuniversality class of magnetic phase transitions.

Measuring chiral critical exponentsNo one had, however, determined these critical exponentsexperimentally, so my colleagues Sergey Maleyev andVladimir Plakhty from St Petersburg in Russia devised astrategy to measure them making use of fundamental rulesrelating the spin rotation and energy transfer in thescattering of polarised neutrons. At first we investigatedtheir scattering by CsMnBr3 at various temperatures abovethe transition, looking for traces of chiral ordering in thecrystal which would fluctuate over time and location. Thedynamics of this process would be revealed by small changes in the energies of thescattered neutrons correlated with their initial spinorientation. By successively orienting the spins of incidentneutrons parallel and anti-parallel to an applied magneticfield and measuring the difference in the probability of thepositive and negative energy transfers we have obtainedinformation about the dynamic chirality. Ultimately, thetemperature-dependence of this probability difference isdescribed by the critical exponent associated with the chiralordering.

W e also looked at the ordered state, below the transitiontemperature of 8.22K, and found that unequal populationsof left and right-handed domains were frozen in. Again wewere able to use the neutron data to calculate another ofthe chiral critical exponents related to the spin ordering.Combining the two measured critical exponents, we could determine a third criticalexponent. We found they were all in excellent agreementwith the predicted values for this new class of magneticmaterials. l

There is currently a greatinterest in materials withexotic magnetic properties,for example, compounds containing magnetic atoms inwhich the direction of their electron spins are ordered inan unusual way. One such material is caesiummanganese tribromide (CsMnBr3). The manganese ionsare arranged on a triangular lattice, in flat layers. Thisprevents their magnetic moments from lining up in anorderly antiparallel (antiferromagnetic) way as might beexpected for a similar material, say, with a cubic crystallattice. Instead, the magnetic moments of neighbouringmanganese ions lie at angles of 120 to each other. Thiseffect is called frustration and CsMnBr3 is a well-knownexample of a frustrated two-dimensional triangular latticeantiferromagnet.

While frustration in such triangular antiferromagnetshas been established for a long time, the collectivebehaviour of the spins is controversial. Two alternativespin configurations can exist, as shown below: passingalong a line in the crystal lattice from one magnetic atomto the next you can see that the spins can either turnsystematically to the left or to the right. The two spinpatterns cannot be mapped onto each other by a simpletranslation or by rotating all the spins, and are said to bechiral , or handed . As with conventional magnetism,where there is a phase transition at a certain temperaturefrom a state where magnetic moments point in anydirection to an ordered magnetic state, this chiralordering is also expected to set in below a certaintransition temperature. Such phase transitions can beclassified according to quantities called critical

22

New class of

magnetic phase transitions discovered

A mixed metal bromide (CsMnBr3) has anunusual type of ‘handed’ magnetic order.Neutron experiments have confirmed that

it represents a new kind of magneticbehaviour

magnetismJIRI KULDA

Research team V.P. Plakhty and E.Moskvin (PNPI Gatchina St. Petersburg), J. Kulda (ILL Grenoble), D. Visser (Univ. Warwick Coventry),

J. Wosnitza (Physik. Inst. Univ. Karlsruhe), R.K. Kremer (MPI Stuttgart)

The clockwise (a)and counter-

clockwise (b) chiraldomains in the

magnetic structureof CsMnBr3

(a) (b)

Jiri Kulda (right)setting up anexperiment

information about the orientation of magnetic moments.From these results, coupled with certain magnetic X-raymeasurements made at the ESRF, emerges a unifiedpicture of the magnetic ordering process.

As we expected, the magnetic ordering is dominated bythe 3d electrons of iron, although the type of ordering isinfluenced by the M atom. Surprisingly, in the rare-earthbased materials the initial magnetic ordering occurs only inthe array of iron atoms, whereas the rare-earth magneticmoments remain disordered. In contrast, in the uraniumcompound, the iron 3d and uranium 5felectron shellsinteract so that the two sets of atoms order at the sametemperature.

The X-ray studies show that ordering in the 4felectronsis more complicated and involves the information beingtransmitted to the 4fshells via the intermediary 5delectrons of the rare-earth element, but they do orderwhen the temperature is low enough. Using the neutronpolarimeter, we showed that the 4fmagnetic momentsdance along with the (dominant) iron 3d moments, butwith their spins oriented at an angle (which depends ontemperature) to those of the 3d moments. This phaseangle is different for each element, and its origin is notunderstood. At the lowest temperatures, the rare-earthmoments try to align themselves ferromagnetically (all inthe same direction), but the iron prevents them from doingso, and the subsequent competition gives rise to unusualbehaviour in a magnetic field. These experiments are anexcellent example of how neutron and X-ray studiestogether can analyse the complex magnetic structure ofstrategically important materials. l

A material is magnetic when its constituent atoms ormolecules have a magnetic moment (due to the spin ofsingle, unpaired electrons) which are aligned, orordered in some way. The most familiar material is ironbut today many practical magnetic devices are madefrom a combination of magnetic elements so-calledtransition metals such as iron or cobalt, and rare-earthelements such as samarium or neodymium. Powerfulpermanent magnets made of samarium and cobalt(SmCo5) or neodymium/iron/boron (Nd-Fe-B)compounds are in wide commercial use. The magneticmoments of the different types of atoms in thesematerials interact in complex ways and we still do notunderstand them. To find out more about their magneticstructure, and thus improve their performance, we needto look at model systems and measure thefundamental interactions.

One such magnetic material is a compoundcontaining iron, aluminium, and a rare earth M such asdysprosium or holmium, or an actinide uranium(MFe4Al8, where M = Dy, Ho, U). The electronsresponsible for iron s magnetism sit in the so-called 3delectronic shell in the atom, while in dysprosium orholmium, which are heavier elements with moreelectrons, they sit in the 4fshell. In uranium, theheaviest element, they sit in an outer 5fshell.

Combining neutron and X-ray studiesW e have been studying in considerable detail how thevarious electronic shells (3d, 4fand 5f) develop theirordered magnetic moments as the temperature islowered. This has been the subject of considerablecontroversy in the past. Taking advantage of a newpolarimeter at ILL, which supplies polarised neutronsin any chosen orientation and then measures changesin orientation of the outgoing neutrons, we could obtain

23

Ingenious neutron scattering experiments combined with X-ray studies are for the first time revealing the complexelectronic structures of modern permanent magnets

magnetism

Inside

modern magnets

GERARD LANDER

Research team J. A. Paixªo (University of Coimbra), P. J. Brown (ILL), B. Lebech (Risł National Laboratory), G. H. Lander (Institute for Transuranium Elements, Karlsruhe)

The unusual magneticconfiguration of theUFe4 Al8 compound

U Fe

Gerard Lander(right) with Nick Bernhoeft of CEA Grenobleand theinstrument IN8

Magnetic materials called spinglasses in which the spinsare highly disordered are

used as universal models ofcomplexity. However, there

is still much to learn abouttheir dynamic behaviour

does it take longer with a stretched exponential timedependence as found in some complex and stronglyinteracting systems?Neutron spin echoSpin relaxation happens over a period of between ahundred-millionth to a million-millionth of a second, whichexactly matches the time window that can be probed by atechnique called neutron spin echo. Here, polarisedneutrons are passed down a fixed-length tube in a veryuniform magnetic field. Their spins precess , and thenumber of precessions achieved depends on their velocity,or the time spent in the tube. After being scattered by thesample the neutrons are spin-flipped and then travel downa second tube of similar length and magnetic field. Theprecession is therefore reversed and the neutron spinsshould end up back as they were originally. If, however,there is a slight change in neutron energy due tointeractions with the spin relaxation process in the samplethen there will be a change in their velocity and their spinswill not have wound back to what they were. The spin echoexperimental set-up is such that the energy changes can bemeasured more precisely than with any other neutrontechnique, and it has been possible to follow the evolutionof very slow magnetic spin relaxation not only down to thespin glass transition, but also below it.

Our measurements have provided important newinsights. For example, we find that the relaxation processescontinue to evolve through the glass temperature, althoughwith different time-scales on either side of the transition.The measurements also show quite clearly that the spinrelaxation follows the complex stretched exponential timedependence, just as predicted by several theoreticalmodels of strongly interacting disordered systems. l

Look closely at an ancient stained glass window andyou might notice that the glass at the bottom of thewindow is thicker than at the top. This is because thematerial has flowed a little, like a liquid. In molten (liquid)glass the molecules move randomly but when it cools toa certain temperature (the glass transition) and solidifies,the random positions of the molecules become frozen in.

A question interesting scientists is whether the glasstransition is a true transition. Enormous amounts of workhave been done on trying to understand the behaviour ofglasses since they are a classic example of a complexsystem. Even more intriguing, however, are magneticmaterials in which the spins interact randomly to give ahighly disordered magnetic state a so-called spinglass. Spin glasses also undergo a glass transition to astate in which the random orientations of the spins arefrozen in.

Physicists are fascinated by spin glasses. They are,in many ways, easier to theorise about than structuralglasses, and mathematical models of spin glassbehaviour have been applied in areas dealing withcomplexity as diverse as protein folding and biologicalevolution. However, there is still significant controversyregarding the spin glass transitions. Again, somephysicists ask: does a spin glass transition really exist;and how universal is the nature of glass transitions? Toinvestigate further, we have been studying a particular

model spin glass system a non-crystalline or amorphous alloy oferbium and iron (Er7Fe3). The bestway to investigate the spin glasstransition is to follow what happensto the spins at varioustemperatures around the glasstransition. When individual spins inthe sample align how long does ittake for them for them to becomeunaligned? Does this spinrelaxation decay awayexponentially as for manyanalogous physical processes, or

24

Relaxing spins in

glassymagnets

magnetismROBERT CYWINSKI

Research team P. Manuel and R. Cywinski (Univ. St. Andrews),R.I. Bewley (ISIS, CLRC), P. Schleger and B. Farago (ILL)

Figure 1 (right) Theseneutron spin echo

measurements show howthe alignment of the

magnetic spins in glassyEr7Fe3 decay with time,

and how this decaychanges as the sample iscooled the solid linesrepresent a theoretical

model describing complexspin relaxation processes

Relaxation rate probabilityTemperature (K)

Log (t)

36

32

28

24

20-10

010

Figure 2 The evolution of thedistribution of magnetic

relaxation rates close to theglass transition in amorphous

Er7Fe3 as determined by acombination of neutron spinecho, dynamic susceptibility

and magnetisationmeasurements. Notice thedramatic broadening of thedistribution as we go below

the glass temperature

to gravity. Because all COW-typeexperiments so far have beencarried out using essentially thesame neutron interferometer, it ismuch more likely that the intricatetheoretical aspects of thisinstrument are not yet fullyunderstood.

To verify this hypothesis wehave repeated the COWexperiments with a completelydifferent type of neutroninterferometer: our interferometer for very coldneutrons at ILL. This one-metre-long interferometeruses particularly slow neutrons flying at only about 40metres per second (144 kilometres per hour). Thismakes it very sensitive to gravity. Thanks to adding aneutron filter to our interferometer in 1999, wemanaged to determine all parameters in ourexperiment with high enough accuracy to be able todecide whether the earlier experiments or the theorywere right.

Our experiment consisted of measurements of theshift of the interference pattern (gravitational phase-shift) at different amounts of tilt of the interferometer(Figure 2). The measurement of the velocitydistribution of the neutrons and of some otherparameters results in a theoretical prediction withwhich we can compare our experimental results. Wefind that our experiment agrees with theory on theexperimental level of error of about 0.4 per cent. Atthe same time it excludes the earlier results that didnot agree with theory.

W e think we can now safely say that neutronsdrop as theory predicts; also when they happen to bebehaving like waves. l

By dropping masses off the tower of Pisa, GalileoGalilei for the first time demonstrated that all bodies fallthe same way, when starting from the same position withthe same velocity. This observation forms the basis ofEinstein s General Theory of Relativity describing gravity.

It is interesting to ask what happens when you dropan object, such as the neutron, that is governed by thelaws of that other great physical theory of our time:quantum mechanics. This theory predicts that very smallparticles can behave either as waves or as particles,depending on how you look at them. Wave-likebehaviour has been observed in several beautifulexperiments for increasingly heavier particles electrons, neutrons and atoms. Recently Markus Arndtand colleagues at the University of Vienna showed thateven so-called fullerenes football-shaped moleculesconsisting of 60 or 70 carbon atoms behave as waves.

The COW experimentUntil 1975 there had been no experiment thatdemonstrated what happens when you drop a particlein the quantum limit, that is, while it is behaving as awave. In that year R. Collela, A.W. Overhauser and S.A.W erner (COW) at the University of Missouri (Columbia)observed just that using a neutron interferometer. This isa device that specifically exploits the wave-like aspect ofthe neutron. Inside the instrument the neutron wave issplit and goes two separate ways. When the two parts ofthe neutron wave come together again this leads tointerference, as with water-waves behind a rock. Atsome places the neutron wave becomes stronger, whileat other places it disappears. COW observed how thegravitational force shifts this interference pattern (seeFigure 1).

Unfortunately the shifts in the interference patternobserved in the COW experiment and in later, moresophisticated versions of it did not quite agree withwhat quantum theory predicted. Does this meantheoreticians should start to worry? Probably not.Another team recently showed using an atom inter-ferometer that theory and experiment agree almostperfectly on a level of 7 parts in a billion.

This leaves us with the very unlikely hypothesis thatthe neutron might be somehow different when it comes

25

How does gravity affect a particle obeying quantum laws?Scientists at the University of Vienna and ILL have beeninvestigating how neutron waves respond to gravity

fundamental physics

Dropping neutron waves

GERBRAND VAN DER ZOUW

Research team G. van der Zouw and A. Zeilinger (Univ. Vienna),P. Hłghłj, R. G hler, P. Geltenbort and J. Butterworth (ILL)

Figure 1 The COWexperiment: when aneutron interferometeris tilted one path travelsat a greater height thanthe other. This causesa shift (phase-shift) ofthe interference patternat the exit of theinterferometer wherethe two paths cometogether

Figure 2 This graph showsthat an interference pattern(which here looks like asine) is shifted by tilting theinterferometer. The amountof this phase-shift ispredicted by theory

insights into technologically important quantum systemssuch as high temperature superconductors (whichconduct electricity without resistance). These materialshave complex disordered structures whose electronicbehaviour is still not well-understood.

The large-scale properties of superfluid helium areknown to be modified by immersion in porous media,and we wanted to relate these changes to the quantumexcitations using inelastic neutron scattering. Weexamined superfluid helium in two forms of porous silica aerogel and Vycor. In the aerogel the superfluidtransition is lowered by a few millikelvin whereas inVycor it is lowered to 1.95K.

What we found was that while the roton energies andthe roton life times in both systems were the same as inbulk superfluid helium, there were additional two-dimensional excitations associated with the layers ofhelium atoms that are adsorbed and held in place on thesilica surfaces. These layer modes had lower energiesthan the three-dimensional bulk rotons and accountedfor the modifications in physical properties.

Vycor showed a further peculiar effect. In bulk heliumthe intensity of the rotons should drop more or lessproportionately with the superfluid fraction as thetemperature is raised to the superfluid transition. At thispoint the rotons should disappear. They don t! There isstill a roton peak in the neutron spectrum above thesuperfluid transition. In fact, theoretical work hassuggested that the roton intensity should relate directlyto the number of helium atoms that have undergoneBEC rather than the superfluid fraction. We thereforethink that in Vycor the temperature at which the BECoccurs is actually above the superfluid transition. Inother words, superfluidity sets in only after the BECfraction has reached a certain level as the temperatureis dropped. This separation of the superfluid and BECtransition points is very strange indeed but must relatedto the disorder induced by the confined environmentfurnished by Vycor. We still have a lot to learn aboutthese intriguing quantum systems. l

H elium is the only material that does not freeze,even close to absolute zero temperature. Instead, itundergoes some extraordinary changes. At atemperature of 2.17K, bulk liquid helium-4 (the normalisotope with two neutrons and two protons) changesto a superfluid a bizarre form of matter which flowswithout friction or viscosity and can even climb out ofits container!

This state is the result of the helium-4 atomsundergoing a change predicted by quantum theorycalled a Bose-Einstein condensation (BEC).According to quantum mechanics, at a low enoughtemperature, certain kinds of indistinguishableparticles like helium-4 will all collect together in thelowest quantum energy state (ground state) andbehave as a single quantum entity. Like microscopicquantum particles, this macroscopic quantum statecan be excited to higher energy levels; in the case ofsuperfluid helium, these excitations are densityfluctuations called phonons and rotons. Rotons inparticular have been well studied and their energyspectra related to the large-scale physical propertiesof liquid helium such as specific heat, the superfluidtransition temperature, and the fraction of the liquidthat is in a superfluid state.

Introducing disorderRecently, researchers have become interested instudying superfluid helium adsorbed on surfaces andin porous solids. These confined systems areexpected to show some disorder, and could offer

26

Pure superfluid helium is well-known for its bizarre quantum properties.When confined in porous silica, its behaviour is even stranger

quantum systems

Research team B. F k and O. Plantevin (CEA Grenoble),H.R. Glyde and N. Mulders (University of Delaware),

G. Coddens (LLB Saclay), H. Schober (ILL)

Superfluidhelium

in porous media

Temperaturedependence of the

roton energymeasured for

different aerogels(symbols). Nodeviations are

found from bulkhelium-4 (line)

The IN12instrument used inthis experiment

BJ RN F¯K BJÖRN FÅK

Enzyme A type of protein responsible for mediating aspecific biochemical reaction in living systems.

felectrons The electrons occupying the f orbitals in anatom. Atomic electrons sit in orbitals, s, p, d, f, classifiedaccording to the laws of quantum mechanics.

Ferromagnetism A type of magnetic order in which theelectron spins are all aligned.

Fractal A type of complex structure that is the same onany scale. Fractals are classified according to the waythey scale.

Giant magnetoresistance A phenomenon shown bycertain complex magnetic materials in which an appliedmagnetic field produces a large change in electricalresistance.

Gigapascal Unit of pressure equal to a billion newtons per square metre and equivalent to 10 000 times theatmospheric pressure.

Glass A solid material in which the atoms or moleculesare randomly arranged.

Glass transition The change at a certain temperaturefrom a glass to a liquid or vice versa.

Heavy water W ater in which hydrogen has beenreplaced by its heavier isotope deuterium

Hexagonal ice The normal form of ice in which eachwater molecule has four nearest neighbours.

Hydrogen bonding A weak form of bonding in which the proton of a hydrogen atom is electrostaticallyattracted to pairs of electrons on atoms like oxygen andnitrogen. It plays a crucial part in the structure of water and also ofmany biological materials.

Incommensurate structure A term applied tomaterials in which a periodic ordering (for example,magnetic spins) does not match that of the crystal lattice.

Inelastic scattering Scattering of particles likeneutrons from a sample in which there is an exchange ofenergy between the neutrons and the molecules oratoms in the sample, thus giving information about thedynamics of the sample molecules or atoms.

Lipid A molecule consisting of a long hydrocarbon chainwith an electrically charged group of atoms at one end.Lipids arrange themselves in layers and are the basis ofbiological membranes.

Magnetic moment Magnetic effect arising from aspinning electric charge. Atoms have magnetic momentsas a result of the spin and orbital motions of anyunpaired electrons. Neutrons also have a magneticmoment.

Amorphous state A state of solid matter in which theatoms or molecules are arranged randomly rather than ina regular crystalline array.

Angstrom (¯) A unit of length equal to 10-10 metres.

Antiferromagnetism A type of magnetic order in whichthe magnetic moments of the atoms or molecules arealternately aligned in opposite (antiparallel) directions,giving a null net magnetisation.

Bilayer A term most often applied to double layers oflipid molecules.

Bose-Einstein condensation A quantum phenomenonin which particles which have a whole-number spin(bosons) drop into a single quantum state usually at verylow temperatures. Particles with half-integer spin(fermions) such as electrons cannot occupy the samestate and must spread themselves over a series of statesas in an atom. Examples of bosons are helium-4 nucleiand Cooper pairs the electron pairs responsible forsuperconductivity.

Cellulose A natural carbohydrate consisting of a chainof simple sugar molecules in a complex configuration.

Contrast variation A technique in which particularatoms in a sample are substituted by another isotopewith different scattering strength in a way thatpreferentially enhances the scattering pattern of particularcomponents of interest.

Crystal lattice The regular three-dimensional array ofatoms or molecules in a crystal.

Critical exponent A quantity used to classifysubstances according to the type of phase transitionsthey undergo. Close to a phase transition, certainproperties vary with the difference between thetemperature of the material and the phase transitiontemperature raised to some characteristic power thecritical exponent. These coefficients reflect the symmetryof the material.

d electrons The electrons occupying the d orbitals inan atom. Atomic electrons sit in orbitals, s, p, d, f,classified according to the laws of quantum mechanics.

Deuterium A heavier isotope of hydrogen having aneutron as well as a proton in the nucleus.

Electronic shell Electrons in atoms are arranged inconcentric shells subdivided into orbitals (see delectrons).

Electron spin One of the quantum properties ofelectrons (and other subatomic particles). It has a valueof one-half. Electrons prefer to form pairs with spins inopposite directions; unpaired electrons in atoms ormolecules are responsible for the magnetic properties ofmaterials.

27

glossary

Quantum mechanics The theory that describes thebehaviour of matter at the microscopic level in terms ofprobability waves.

Quantum tunnelling A phenomenon in which a quantumsystem can tunnel through an energy barrier as a result ofthe Uncertainty Principle which allows for a small probabilitythat the system can exist on the other side of the barrier.

Quasielastic neutron scattering Measurement of verysmall changes in neutron energy after scattering, whichrelate to subtle dynamic changes in the sample.

Rare-earth elements A large group of heavy metals in thePeriodic Table from cerium (atomic number 58) to lutetium(atomic number 71) with similar chemical and physicalproperties.

Roton A type of quantum energy excitation found insuperfluid helium.

Silica Silicon oxide.

Small angle neutron scattering Measurement of neutronscattering at small angles used to investigate structures withlarge interatomic distances such as polymers or biologicalstructures.

Specific heat The ratio of heat supplied to a material to itssubsequent rise in temperature.

Specular scattering Reflection of waves in which theangle of reflection equals the angle of incidence.

Spin glass A magnetic material in which the ordering ofspins is random.

Superalloy A group of metal alloys often containing nickelwith high strength at elevated temperatures.

Superconductivity The property of conducting electricitywith no resistance.

Supercooled water Liquid water cooled below its normalfreezing point.

Superfluidity A property in which liquid helium flowswithout friction and viscosity.

Synchrotron radiation The electromagnetic radiationproduced when charged particles are accelerated and bentin a magnetic field. Synchrotron radiation is coherent andcan be produced in intense narrow beams suitable for X-rayanalysis experiments.

Transition metal A large group of heavy metals in thePeriodic Table with a rich variety of chemical and physicalproperties derived from the fact that they have unfilled dorbitals.

X-ray scattering A technique used to determine thestructure of materials. X-rays are reflected, or scattered, offa material in which the interatomic distances are similar tothe X-ray wavelength such that the scattered waves interfereto produce a characteristic diffraction pattern.

Magnetic susceptibility The ease with which amaterial can be magnetised by an external magneticfield.

Nanometre One billionth of a metre (10-9 metres).

Neutron One of the two particles found in the atomicnucleus. It is electrically neutral, but has a spin of a halfand a mass of 1, equal to that of the proton.

Neutron reflectivity A technique in which neutrons arereflected off a surface or interface. It is used tocharacterise the structure of surfaces and thin layers.

Neutron diffraction or scattering Like othersubatomic particles, neutrons have a characteristicwavelength depending on energy. When reflected, orscattered, off a material in which the interatomicdistances are similar to the neutron wavelength, thescattered waves interfere to produce a characteristicdiffraction pattern.

Neutron spin echo A method of measuring very small changes in energy in a neutron beam afterinteracting with a sample, and thus subtle aspects ofdynamics in the material, from changes in precession ofneutron spin after passing through two successivemagnetic fields which oppose each other.

Nonspecular scattering Reflection of neutrons from asurface or interface in which the angle of reflection isdifferent from the angle of incidence. It gives informationabout the lateral structure of the surface.

Phase diagram A schematic representation of thevarious structures adopted by a material over a range oftemperatures and pressures.

Phonon A quantum of atomic vibration in condensedmatter.

Phospholipid A class of lipid containing a phosphategroup.

Plasma A hot gas composed of charged particles.

Polarised neutrons A beam of neutrons whose spinsare all aligned.

Polymer Large molecules consisting of chains of similarchemical units.

Polysaccharide A polymer made up from sugar units.

Quantum energy levels Systems obeying the laws ofquantum mechanics can occupy only certain energylevels.

28

glossary

Scientific Coordination

Office (SCO)

Institut Laue Langevin

6 rue Jules Horowitz

BP 156

38042 Grenoble Cedex 9

France

Email: kjenkins@ill.fr or

sco@ill.fr

page 4

Dr Peter Timmins

ILL

Email: timmins@ill.fr

page 5

Dr Helmut Schober

ILL

Email: schober@ill.fr

page 6

Dr Henri Chanzy

CERMAV-CNRS

BP 53X

F-38041 Grenoble Cedex 9

France

Email:

chanzy@cermav.cnrs.fr

page 7

Dr Giovanna Fragneto-

Cusani

ILL

Email: fragneto@ill.fr

page 8

Dr Christian M nster

Sektion Physik

LM Universit t M nchen

Geschwister Scholl Platz 1

D 80539 M nchen

Email: christian.muenster@

physik.uni-muenchen.de

page 9

Dr Peter M ller-Buschbaum

Technische Univ. M nchen

Physik-Department

Lehrstuhl E13

James-Franck-Str 1

D-85747 Garching

Germany

Email: muellerb@physik.tu-

muenchen.de

page 10

Dr Adrian R. Rennie

Dept. Chemistry,

King s College

Strand,

London WC2R 2LS

U K

Email:

rennie@colloids.ch.kcl.

ac.uk

page 11

Professor F. Javier Bermejo

Inst. Estruct. Materia

C.S.I.C.

Serrano 123

E-28006 Madrid

Spain

Email: javier@langran.iem.

csic.es

page 12

Dr Caterina Petrillo

Dip. di Fisica

Univ. di Perugia

Via A. Pascoli

I-06123 Perugia

Italy

Email: petrillo@pg.infn.it

page 13

Professor John Dore

School of Physical

Sciences

Univ of Kent,

Canterbury CT2 7NR

U K

Email: j.c.dore@ukc.ac.uk

page 14

Dr Helmut Schober

ILL

Email: schober@ill.fr

page 15

Dr Peter Lindner

ILL

Email: lindner@ill.fr

page 16

Professor Gernot Kostorz

Eidgen ssische Technische

Hochschule Z rich

Institute of Applied Physics

CH-8093 Z rich

Switzerland

Email: kostorz@iap.phys.

ethz.ch

page 17

Dr Giovanna Cicognani

ILL

Email: cico@ill.fr

page 18

Dr Ralf Siebrecht

ILL

Email: siebrecht@ill.fr

page 19

Vincent Leiner

Institut fr

Experimentalphysik IV

(Festk rperphysik)

Ruhr Universit t Bochum

D 44780 Bochum

Germany

Email: leiner@ill.fr

page 20

Dr Hans Lauter

ILL

Email: lauter@ill.fr

page 21

Dr Isabelle Mirebeau

Laboratoire LØon Brillouin

CE-Saclay

91191 Gif sur Yvette

France

Email: mirebea@llb.saclay.

cea.fr

page 22

Dr Jiri Kulda

ILL

Email: kulda@ill.fr

page 23

Dr Gerard Lander

European Commission

JRC

Institute for Transuranium

Elements

Postfach 2340

D-76125 Karlsruhe

Germany

Email: lander@ill.fr

page 24

Professor Robert Cywinski

School of Physics and

Astronomy

Univ of St Andrews

St Andrews KY16 9SS

U K

Email:

Robert.Cywinski@st-

and.ac.uk

page 25

Dr Gerbrand van der Zouw

Institut fr

Experimentalphysik

Universit t Wien

Boltzmanngasse 5

A-1090 Wien

Austria

Email:

gerbrand.van.der.zouw

@univie.ac.at

page 26

Dr Bj rn F k

CEA Grenoble

DRFMC/SPSMS/MDN

17 rue des Martyrs

F-38054 Grenoble Cedex 9

France

Email: fak@cea.fr

29

contacts