highlights - European Synchrotron Radiation Facility (ESRF)

2008www.esrf.eu

ESRFHIGHLIGHTSESRF - H

IGH

LIGH

TS 2008

BP220 - F38043 Grenoble Cedex, FranceEuropean Synchrotron Radiation Facility A light

for science

CO N T E N TS

Introduction 2

Upgrade Programme 4

Scientific Highlights 7

High Resolution and Resonance Scattering 7Materials Science 21Soft Condensed Matter 39Structural Biology 53Surface and Interface Science 71X-ray Absorption and Magnetic Scattering 83X-ray Imaging and Optics 102Methods and Instrumentation 120

The Accelerator and X-ray Source 132

Facts and Figures 138

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

E u r o p e a n S y n c h r o t r o n R a d i a t i o n F a c i l i t y

H I G H L I G H T S

2 0 0 8

HH II GG HH LL II GG HH TT SS 22 00 00 88

2 EE SS RR FF

It was obvious at the outset that2008 would be a decisive year for theESRF, with a major impact on thefuture. In retrospect, the outcomehas been positive in all aspects,blending continued scientific successwith the beginning of a successfultransition towards a bright future.The scientific highlights on thefollowing pages testify to thebreadth and quality of the scientificprogramme at the ESRF. Each resultwould merit a mention in thisIntroduction; we trust that everyreader will find an example ofoutstanding research in a field thathe or she is familiar with.

The numbers of proposals,experimental sessions, user visitsand publications in peer-reviewedjournals have hit new record highs.The figure of 2000 proposals wasexceeded for the first time andthe oversubscription on somebeamlines emphasises the need tocontinue with the efforts to maintainand extend rapid throughput andefficient operation. Further detailson user operations are included inthe chapter on Facts and Figures.The beamlines for macromolecularcrystallography continued to showthe way, introducing remote accessin 2008, unique in Europe; thisearned the Bessy Innovation Awardfor the team who developed thenecessary software. The ESRF’sleadership in the field of beamlineautomation is a result of goodcooperation, notably with the EMBL.

The world-class stability andreliability of the accelerators andX-ray source at the ESRF are animportant factor for our capability toincrease the scientific output year byyear. In 2008, the 7/8 + 1 beam fillingmode has become the new standardmode, meeting simultaneously theneeds of two very different user

communities. However, sixteen yearsafter the first beams circulated inthe storage ring, care has to betaken that old and/or obsoletetechnologies are replaced in duetime. Even “simple” componentscan cause continued concern astheir ageing leads to unpredictablephenomena or suppliers discontinuethe delivery of spares. A major stepforward in the process of continuedreplacement was the commissioningof the HQPS-2 in 2008, a next-generation system to ensure deliveryof electric power even in thepresence of spikes on the publicpower grid. HQPS-2 was graduallyput into operation and its qualitieshave contributed to the result thatstatistically, 2008 has been the bestyear on record at the ESRF in terms ofbeam availability, mean timebetween failures, and the averageduration of a failure.

The ESRF family has grown in 2008,with Slovakia being welcomed asthe 19th member, establishing theCentralsync consortium with theCzech Republic and Hungary, therebyadding 1.05% of the Members’contributions to the ESRF budget.Earlier in the year, the KurchatovInstitute in Russia and theESRF signed a Memorandum ofUnderstanding to promote differentareas of scientific collaboration.

In the area of office automation, theyear 2008 was characterised by thetransition to Windows Vista whichrequired much preparation to ensureseamless integration with the variousmanagement information systemsand, equally importantly, withscientific computing. Documentarchiving and data curation is alsohigh on the agenda in scientificcomputing as data rates and datavolumes continue to be the fastest-growing figures at the ESRF.

However, 2008 will truly beremembered as the year of the ESRFUpgrade Programme. The approvalin principle, at the June meeting ofthe Council, and that of the budgetfor the first year of the Upgrade, atthe November meeting, ensure thetransition towards a bright futurecovering at least a decade. Muchwork was necessary to detail andhone the ESRF’s proposals beforean endorsement could be given bythe Council. The work on theUpgrade Programme is documentedin a separate section immediatelyfollowing this introduction. Here,we would like to highlight theprogress made with the definitionof the building extensions, helpedby a grant from the EuropeanCommission, within the Framework 7Programme. This work has removedmany uncertainties – both technicaland financial – that would haveimpeded rapid progress withdevelopment of the first eightUpgrade beamlines.

More good news was the grant of 15million Euros by French national andregional authorities (Contrat deProjet Etat-Région) to improve theinfrastructure of the joint ILL/ESRFsite over the next three years; thisaward will allow the constructionof a new relocated site entrance, avisitor centre, and will make possiblebuildings for new partnerships withthe ILL and other leading institutes.In addition, the ESRF will benefitfrom the ambitious plans for thesurrounding area, since Grenoble hasbeen selected to host a projectwithin France’s national Campus planto aggressively modernise universityinfrastructure and governance.

The management of the ESRF hasseen two new faces arriving in 2008,with Angelika E. Röhr taking overfrom Helmut Krech as Director of

Introduction

II NN TT RR OO DD UU CC TT II OO NN

3AA LL II GG HH TT FF OO RR SS CC II EE NN CC EE

Administration, and Harald Reichertas one of two Directors of Research.Pascal Elleaume was confirmed foranother term as Director of theAccelerator and Source Division.To deal with the many newchallenges of the UpgradeProgramme, the InstrumentationServices and Development Divisionwas created, regrouping activitiesformerly located in three otherdivisions. Together with theintroduction of an ESRF-wide projectmanagement mechanism, thisreorganisation shall ensure that thetechnical, financial and managerialchallenges ahead are addressed in anadequate manner whilst at the sametime continuing to deliver world-class science. Finally, the end of theyear 2008 marked the end of BillStirling’s eight-year term as ESRFDirector General, and the handingover of the baton to Francesco Sette,who had been a Director of Researchat the ESRF for the past seven years.

The ESRF is a highly visible Europeansuccess story. Thirty publications peryear in Nature and Science for both2006 and 2007 do not remainunnoticed, nor do numerous invitedtalks at prestigious meetings likethe 2008 IUCr Congress in Osaka(Japan). The celebration, on 26November 2008, of the 20thanniversary of the signature ofthe intergovernmental conventionestablishing the ESRF was a goodopportunity to gather the ESRFfamily from all over Europe andaround the world to celebrate whathas been achieved as a result of twodecades of hard work.

We would like to use thisopportunity to extend special thanksto our Member and Associatecountries which, despite a difficultfunding context, have given thegreen light for the Upgrade

Programme. All of the colleagueswhose support was so important forthe acceptation of the UpgradeProgramme, notably the members ofthe governing bodies of the ESRF -the Council, the AFC and the SAC -also deserve our thanks. We are alsograteful to the thousands of usersfrom the scientific fields covered bythe 11 review committees, for theircontinued support and interest, andto the European Commission for theESRFUP grant that funded part of thepreparatory phase of the UpgradeProgramme. Last but certainly notleast, we wish to thank the staff ofthe ESRF who from behind the scenesmake possible the remarkablescientific investigations carried out atthe ESRF which the reader willdiscover on the following pages.

P. Elleaume, R. Dimper,C. Habfast, H. Krech, S. Larsen,H. Reichert, M. RodriguezCastellano, A.E. Röhr, F. Sette,P. Thiry and W.G. Stirling

Incoming and outgoing Directorate of the ESRF meeting at the eve of 2009. From left toright: Bill Stirling, Francesco Sette, Harald Reichert, Angelika Röhr, Pascal Elleaume, SergePerez, who will take over from Sine Larsen as Director of Research in June 2009, RobertFeidenhans'l, Chairman of the ESRF Council, and Sine Larsen. Not included in the photo isHelmut Krech, Director of Administration until October (Credit: C. Argoud).


4 EE SS RR FF

The path to approval deeplyinvolved the Council, ScienceAdvisory Committee (SAC) and UserCommunities without whom thepositive decision would not havebeen possible. The staff and themanagement of the ESRF aregrateful to the stakeholders andthe funding agencies for havinggiven a green light for theUpgrade Programme in a difficultinternational funding context.

The stage for the UpgradeProgramme was set five years agowith the Long Term Strategy, adocument looking to the future ofthe ESRF for the coming decade,which at the turn of 2003-04 receivedpositive reactions from the Counciland the subsequent User Meeting.This strategy document was followedup with the two-volume Purple Book(Figure 1), a detailed science andtechnological programme mappingout concrete ideas and proposalsfor new beamlines, infrastructureand techniques. At this time, theUpgrade was also included onthe European Strategy Forum forResearch Infrastructures (ESFRI) roadmap for large-scale European

research infrastructures, showing theimportance of a renewed ESRF forthe European Research Area.

Since its publication in October 2007,the ideas of the Purple Book havebeen honed and elaborated in muchdetail. The effort and devotion ofmany ESRF staff resulted in aconcrete plan for a Phase I of theUpgrade Programme in line with thefunding capabilities of the ESRFMembers and consolidated intonational road maps for light sourcedevelopment.

This first of two phases will last sevenyears starting in 2009, and theoverlapping phase will be submittedfor consideration in due time tocover further developments until2018. The budget envelope for PhaseI includes 177 million Euros spreadover seven years, of which 103million Euros are additional fundswith the remainder originating fromthe ESRF’s regular investmentbudget. The lion’s share of this willfund development of eight world-class beamlines unique in Europe, ifnot globally, housed in extensionsto the current experimental hall

(Figure 2). These new beamlines,targeted mostly at the exploitation ofnano-focus beams, will be supportedby new enabling technologies inareas like high precision optics andmechanics, detector and computingdevelopments, and by improvementsto the accelerator and X-ray sourcecomplex.

The detailed definition of theseelements of the Upgrade Programmehas seen major progress in 2008,preparing the ground for the hardwork to come over the next fewyears. The inclusion of the UpgradeProgramme into the ESFRI road mapfor future large-scale infrastructuresopened the door to a 5 million Eurosgrant from the Framework 7Programme (FP7) of the EuropeanCommission to catalyse preparatorywork – without which the designstudies could not have pushed aheadas quickly whilst full approval wasbeing sought. Tasks as diverse asfeasibility studies for high-fieldmagnets, data archiving, beamlineconceptual and technical designreports, improved software for usermanagement, use of the GRID forlight sources and architect designs

Upgrade Programme2008 will be remembered as the year the Upgrade Programme turned from a vision into a project thanks to itsapproval in two steps: in June, an endorsement in principle followed in November by a green light in the form of a10% increase to the budget for 2009 explicitly for the Upgrade.

Fig. 1: The two volumes of the PurpleBook outline the science cases andideas for the future science andtechnology programme of the ESRF.The Purple Book can be downloaded inPDF form from the ESRF web site.

Fig. 2: Elements of the Upgrade Programme with the greatest part of fundingbeing spent directly on new and improved beamlines and instrumentation. The

area of the inner segments reflect the staff and recurrent costs, whilst the outersegment areas represent the capital investment costs.

UU PP GG RR AA DD EE PP RR OO GG RR AA MM MM EE


for the experimental hall extensionswere all funded through this FP7grant.

Buildings on their own rarelyproduce new science; the extensionsto the existing experimental hall,necessary to house the longbeamlines for nano-focussed beams,must be designed to minimum cost,whilst their civil engineering mustinterfere as little as possible with on-going user operations. It is importantto maintain maximum access to thebeamlines for the communities whorely on regular beam time. Thedetailed definition for the newbuildings with the capacity to houseup to fifteen long beamlines alongwith some support laboratories andoffice space, is therefore complexand time consuming, and theseactivities will continue well into2009.

In spring 2008, an internationalarchitect competition was held,with the favourite design (Figure 3)now being updated with thelatest scientific and budgetaryrequirements. Particular care andeffort is being put into the design ofthe concrete slab to host the newbeamlines as its vibrational stabilityis key to the beamlines meetingtheir design performance! Workincluded experiments like that of a

1.2 ton steel ball repeatedly beingdropped on the ground in thecentre of the ring to study howvibrations propagate across the site(Figure 4). Also, soil parameterswere probed by test drillings atlocations of future hall extensions.The coming year will be decisive asthe designs for the buildingsincluding the slabs will be finalisedand the tendering of the contractsfor construction will begin. Thebuildings are expected to be readyto move into starting from 2012.

Much care will also be devoted to theselection of the eight new Upgradebeamlines. In May 2008, the SACendorsed the selection of elevencandidate beamlines from which theeight phase I beamlines will bechosen. The eleven candidates(Table 1) have in turn been distilledfrom the total of over forty proposalsoutlined in the Purple Book.

Brainstorming sessions with externalexperts and selected users were held,thanks to the FP7 grant, in the last

Fig. 4: Vibrating the floor of the ESRF(February, 2008).

UPBL1: Local probe coherent diffraction imaging and nanobeam diffraction for characterisation of individual nanostructures

UPBL2: High-energy beamline for buried interface structure and materials processing

UPBL3: Nuclear resonance beamline for the study of nanoscale materials: the interplay of growth, structure, electric and magnetic properties as well as dynamics

UPBL4: Beamline for imaging, fluorescence and spectroscopy at the nanoscale

UPBL5: Beamline for parallel and coherent beam imaging

UPBL6: High-energy resolution inelastic scattering in the hard X-ray range with micro- and nanofocus capabilities

UPBL7: Soft X-rays for nanomagnetic and electronic spectroscopy

UPBL8: Nano- and microbeam crystallography for structural and functional biology and soft matter

UPBL9: (a) Sub-microradian angular resolution small-angle scattering for probing the structure and nonequilibrium dynamics of self-assembling soft matter and biological systems(b) Structural dynamics of molecular assemblies

UPBL10: Large-scale automated screening, selection and data collection for macromolecular crystallography

UPBL11: Pushing the limits of energy-dispersive X-ray absorption spectroscopy towards the nano in spatial and temporal resolution

Fig. 3: Artist’s impression of the building extension design.

Table 1: The eleven candidates for the Upgrade beamlines, honedfrom ideas in the Purple Book.


6

Upgrade Programme

EE SS RR FF

months of 2008 to hone the sciencecases and explore the technologicallimits of the 11 candidate beamlinesleading towards detailed technicaldesign reports. Further guidance onthe new beamlines will be given bythe SAC at the meeting in May 2009.

In parallel, a large-scale exercise tomap out a floor plan for the entireESRF beamline portfolio wasundertaken in 2008, identifying thelocations of the new beamlines, withclustering of beamlines linked byscience or support facilities wheneverbeneficial and feasible. The resultwas endorsed by the SAC inNovember 2008 and is an importantstep to creating the overall plan forall ESRF beamlines at the end ofPhase I of the Upgrade. Not only willeight new beamlines be developed,but the entire portfolio of beamlinesmust be maintained and refurbishedwhere necessary, benefitting fromthe new technologies developed aspart of the Upgrade.

Also in November 2008, the SAC gavethe green light for three candidatebeamlines to already start detaileddesign work, moving them fromconceptual design into technicaldesign report phase. At subsequentmeetings, further beamlines willmove to the design phase. Thisstaggered approach adds flexibilityto the Upgrade Programme andallows the ESRF to react rapidlyshould important new needs forsynchrotron light arrive on thehorizon.

As the accelerator and X-ray sourceare another key element of theESRF’s success, a particular effort willfocus on maintaining both at theforefront. For example, severalstraight sections will be lengthenedto increase their capacity andflexibility to host undulators, andthe Klystron-based radio-frequencysystems upgraded to semiconductor-based technology to improvereliability and prepare for a futureincrease in machine current. To this

end, the development of a higher-mode-dampened prototype cavitybegan in 2008. Whereas 300 mAcurrent in the storage ring remainsan objective, implementation willultimately also depend upon the costof energy.

At the beamlines, the narrowernano-focus beams probing smallersamples will pose many technologicalchallenges to the teams responsiblefor instrumentation, beam optics,sample preparation and samplemanipulation. Also, tomorrow’sdetectors will feature larger arrayswith higher sensitivity and betterresolution, requiring in turn progressto be made in data storage and pre-processing as well as in easy-to-usedata processing. Many of these newtechnologies will be developed incollaboration with other lightsources in Europe. This is why someof the brainstorming sessions in 2008addressed these areas and involvedexperts from other light sources,securing the route to cooperation onthese developments which are bothlengthy and very costly.

Partnerships are another elementof the Upgrade Programme. Theusers of light sources are not ashomogenous a community asother users of large researchinfrastructures. Although this is oneof the reasons for the exciting anddynamic environment at synchrotronlight sources, it also means that usersincreasingly come from institutesor research laboratories withoutmuch technological or scientificbackground in photon science.

The Partnership for StructuralBiology (PSB) has shown us how tointegrate these users: a partnershipin close vicinity to the ESRF and theILL beamlines. For the PSB, thepartners include the EMBL GrenobleOutstation and the French Institut deBiologie Structurale. A laboratorywas set up offering a mix oftechniques, science, and technologyopen to partners both local and

distant. As this concept has nowproven useful in lowering the entrythreshold for users to specialisedphoton and neutron techniques,soft condensed matter is thenext candidate field for such apartnership, again to be shared withILL, and a first proposal has receiveda very positive reaction from theESRF governing bodies.

With our neighbouring sisterinstitute, the ILL, also embarking onits own upgrade programme, ILL20/20, it is important to recognise theneed for general improvements ofthe site on which both areestablished. Together, the ESRF andILL have a considerable impact uponthe local economy and on theinternational visibility of Grenoble asa scientific metropole. Frenchregional and local authoritiesconfirmed in 2008 an investment of15 million Euros spread over fouryears into partnership buildings onthe site, a new site entrance building,improved roads and a restaurantextension, new stores and goodsreception and a visitors’ centre. Inaddition, Grenoble has beenincluded into the national Campusprogramme which will, in the longerterm, profoundly change the urbanlandscape in the neighbourhood ofthe ESRF through the addition ofnew laboratory and universitybuildings, housing and leisurefacilities.

With the ESRF Upgrade, the ILL 20/20programme, the investment in thesite and the surrounding areas, thecoming years will be an exciting timefor the staff and the users of the ESRFas civil construction work andtechnology development will turninto new beamlines and facilitiescoming on-line. The new sciencethese promise to deliver should keepthe ESRF at the global forefront oflight source development for morethan a decade from now.

E. Mitchell and C. Habfast

SS CC II EE NN TT II FF II CC HH II GG HH LL II GG HH TT SS


Introduction

Traditionally the activities of the HighResolution and Resonance ScatteringGroup cover a wide scientific andmethodological range from a betterunderstanding of basic phenomena toapplications and new developments.The main fields are investigations ofelectric and magnetic properties as wellas structural dynamics, with an emphasison extreme conditions such as highpressure, high and low temperature,external magnetic fields, and confinedgeometry. This section presents a broadset of examples and is structuredaccording to basic phenomena,applications, and new developments.

A single photon can ionise only oneelectron of an atom, that is what welearn from the textbook. Doublephotoionisation with a single photoncannot occur directly but only byinteraction of the electrons with eachother within an atom. Such discoveriesalready date back a couple of years.Today, Huotari et al. present asystematic study of 3d transitionmetals in comparison with theory.Their study suggests that nearthreshold the excitation is a quasi-classical knock-out rather than aquantum-mechanical shake-off effect,and they found a universal scaling lawfor the double photoionisation energyevolution.

Hydrogen bonds are of paramountimportance in many disciplines ofnatural sciences. Even in a chemicallysimple system such as water, and despitedecades of intense research efforts, newinsights are still being gained today.Tse et al. carried out X-ray Ramanmeasurements on the solid and liquidphases of water in order to characterisethe nature of the distinct pre-edge of

the oxygen K-edge. Winkler et al.studied the dispersion of OH-stretchingmodes in diaspore (AlOOH), exploitingthe unique possibility of inelastic X-rayscattering to study these high-energyexcitations with a resolution of 6 meV.

The structures of some simple elementsmay change at high pressure becomingrather complex. Rb-IV (others includeNa, Ka, Ba) belongs to the class ofincommensurate host-guest systems,whose crystal structures have beenstudied in great detail. Loa et al.investigated the pressure dependence ofthe longitudinal acoustic phonondispersion, and found that the guestlattice vibrations closely resemble amono-atomic linear chain, indicatingvery weak coupling between the hostand the guest structure.

Applications are going towardsnanocrystalline material and catalysts.Stankov et al. were surprised that atomicvibrations in nanocrystalline grains areactually identical to those in the bulkeven for extremely small grain sizes andthat the anomalous dynamics ofnanocrystalline material arises in factfrom the atomic vibrations in thedisordered interfaces.

Spectroscopic studies of catalyticreactions have received considerablepublic attention with the Noble Prizefor Gerhard Ertl in 2007. The hard X-rayprobe at the ESRF is ideally suited tothe investigation of catalytic reactionson real catalysts under workingconditions. This is crucial to advance theunderstanding of the chemistry. Twoexamples from ID26 are presented. Thefirst concerns the oxidation of carbonmonoxide that is relevant for thepurification of feed gas for fuel cellsand automotive exhaust gas. Thesecond contribution presents thevisualisation of a catalytic reaction in

High Resolution and ResonanceScattering


8


EE SS RR FF

Playing billiards with electrons: doublephotoionisation with a single photon

Principal publication andauthorsS. Huotari (a), K. Hämäläinen (b),

R. Diamant (c), R. Sharon (c),

C.C. Kao (d), and M. Deutsch (c),

Phys. Rev. Lett. 101, 043001

(2008).

(a) ESRF

(b) Division of Materials Physics,

Department of Physics, University

of Helsinki (Finland)

(c) Physics Department, Bar-Ilan

University (Israel)

(d) National Synchrotron Light

Source, Brookhaven National

Laboratory (USA)

A single photon can directly ioniseonly one electron of an atom. Asimultaneous ejection of a secondelectron, i.e. double photoionisation,may occur only if the electrons withinthe atom interact with each other.If both electrons are ejected from theK-shell, a so-called hollow atom isformed, wherein the K shell is emptywhile higher shells are populated(Figure 5). Hollow atoms are of a greatinterest in many fields, ranging fromatomic physics to surface science.When decaying by a 1s2 → 1s12p1

transition (underlining denotes holestates) they emit a Khα1,2 hypersatellitephoton [1]. Thus, doublephotoionisation and electron-electroninteractions can be studied by high-

resolution X-ray hypersatellite spectrameasurements.

Using beamlines ID16 (ESRF) and X25(NSLS, USA) we have studied doublephotoionisation near its energythreshold, Eth, in 3d transition metals,by high-energy-resolutionhypersatellite spectroscopy. Figure 6compares the conventional Cu Kα1,2diagram spectrum of a 1s1 → 2p1

transition with its hypersatellitecounterpart. The I(Kα1)/I(Kα2) = 2intensity ratio reflects the numberratio of the LIII and LII electrons. Forhypersatellites, the measured ratio isvastly different, being determined byselection rules rather than by the LIII/LIIelectron number ratio. The Khα1 linerequires flipping the shell-changing 2pelectron’s spin, which is forbidden inthe LS coupling dominating low-Zatoms. In the range of Z studied heresuch flipping is partly allowed, and theKhα1 line has a non-zero intensity.

Figure 7 (inset) shows measuredR=I(Khα1,2)/I(Kα1,2) intensity ratios vs.ionising photon energy. This is the firstsystematic measurement of the near-threshold behaviour of R for a rangeof elements. The very long saturationrange, ~70% of Eth, stands outimmediately. Eth and the saturationranges and values in the inset differ

Fig. 5: Schematicrepresentation of theprocesses for a) single

photoionisation andb) double

photoionisation, wherewavy arrows denote

photons, and straightarrows, electron

motions. The final stateof the hypersatellite

emission process is leftwith a spectator hole,which is later filled by

another electroncreating conventional

diagram X-ray emissionlines.

Reference[1] I. Sergueev et al., PRB 78,

214436 (2008).

space and in time by means of the ESRFFReLoN camera.

New developments continued to play animportant role. The demand to accesshigh-energy Mössbauer transitions iscontinuously growing. After last year’spresentation of the nickel case, wecontinued to exploit this elementutilising synchrotron radiation basedperturbed angular correlationspectroscopy (SRPAC) [1]. In this issueantimony (37.1 keV) and the firsttellurium (35.5 keV) results arepresented. In the spotlight areskutterudites, a thermoelectric andferromagnetic material, as well as phase

change materials used in rewritableoptical data storage.

Finally, Bosak et al. utilised thermaldiffuse scattering (TDS) to visualiseportions of the Fermi surface in threedimensions via intensity modulationswhich arise from Kohn anomalies in thephonon dispersion. The experiment onZn was performed on the Swiss-Norwegian beamlines BM01, and gives aflavour of the great potential if anundulator based beamline and a state-of-the-art detector system could beutilised.

R. Rüffer

among the elements. Nevertheless,when scaled by the binding energiesof the indirectly-ionised K-electronsand the saturation values (which,incidentally, are found to decrease as~Z-1.6) an intriguing universalbehaviour is revealed: all curvescollapse onto a single curve (Figure 7).

How does the double photoionisationoccur? For photons of energiesE >> Eth, the ionised electron is ejectedwith a very high kinetic energy,changing abruptly the atomicpotential acting on the otherelectrons. The secondary electron findsitself in an excited state, and has afinite probability of being ejected. Thisis the shake-off (SO) model, which,however, is invalid close to Eth, wherethe ionising photon’s energy is justbarely enough to overcome thebinding energy of the electron. Of theother theoretical models only theknock-out (KO) model [2] agrees well(blue line in Figure 7 and lines in theinset) with the measured universalevolution of R near threshold, wherethe ejected photoelectron is slow andstays longer within the bounds of theatom. In the KO model the doublephotoionisation proceeds by thedirectly-ionised electron knocking outthe secondary one, as in billiards, onits way out. Moreover, the KO processturns out to be a quasi-classical, ratherthan a quantum, effect, in spite of itsoccurrence within the sanctumsanctorum of the quantum regime:the inner bounds of the atom.Thus, double photoionisation seemsto evolve from a quasi-classical KObilliards process near threshold to afully quantum SO effect at highionising photon energies. In conclusion, we have determineddouble photoionisation probabilitiesby measuring the evolution of the

hypersatellite spectrum from thethreshold upwards for elements23 ≤ Z ≤ 30. We suggest that near thethreshold the excitation is a quasi-classical KO effect rather than aquantum mechanical SO one, anddemonstrate a universal scaling law forthe double photoionisation energyevolution near threshold.



Fig. 6: a) The conventional Cu Kα1,2 diagram spectrum. b) The hypersatellite Cu Khα1,2

spectrum. Note the difference in the relative intensities of the two emission lines inthe two spectra.

Fig. 7: Relative double photoionisation probability R for various elements as a functionof the excess photon energy above threshold, E-Eth, scaled by the binding energy ofthe secondary ionised electron, Eth-EK, where EK is the binding energy of the directly-ionised primary electron. The inset shows the raw cross-sections for each element. Thesolid curve is the (scaled) theoretical KO prediction [2].

References[1] R. Diamant, S. Huotari,

K. Hämäläinen, R. Sharon,

C.C. Kao, and M. Deutsch, Phys.

Rev. Lett. 91, 193001 (2003).

[2] T. Schneider and J.M. Rost,

Phys. Rev. A 67, 062704 (2003).

X-ray absorption of ordered anddisordered ice

Despite extensive experimental andtheoretical investigations over manydecades, an unambiguousunderstanding of the structure andnature of the hydrogen bonding ofwater in disordered condensed phases

remains an elusive goal [1]. Theabsorption spectrum of the core 1soxygen (oxygen K-edge X-rayabsorption XAS) is a sensitive atom-specific probe for the H-bond localenvironment around the absorbing


10


EE SS RR FF

atom. From the comparison of oxygen1s electronic excitations near theabsorption threshold (pre-edge) withtheoretical calculated spectrum ofliquid water, it was postulated thatthere exists a large fraction of watermolecules with broken hydrogenbonds [2]. This hypothesis challengesthe conventional model of liquidwater where each water molecule isapproximately tetrahedrallycoordinated to four neighbouringwater molecules through hydrogenbonds (H-bond). The basic premise forthis suggestion is the observation of astrong pre-edge absorption featureanalogous to water adsorbed onsurfaces with a broken H-bondingnetwork.

To characterise the nature of thedistinct pre-edge absorption,oxygen K-edge XAS of several icepolymorphs were studied using themethod of X-ray Raman spectroscopy(XRS). Using X-ray radiation withenergy much higher than the coreionisation, XRS measures energy lossesdue to electronic excitations in amethod that is analogous to XAS.Thanks to the higher penetrationdepth of the X-ray, it is essentially abulk technique and not affected bysurface and saturation effects usuallyencountered in soft X-ray absorption.To ensure direct and unbiasedcomparison of different spectra, XRSmeasurements were made on severalice polymorphs under almost identical

instrument resolution, energycalibration and sample environment.This was achieved by taking advantageof the well-characterised thermalinduced in situ successive structuralphase transformation of a high-densityamorphous ice (HDA) to low densityamorphous (LDA), to crystalline ice Icand ice Ih and eventually to liquidwater [3]. Oxygen K-edge XRS spectrawere recorded at beamline ID16. Itwas imperative that the sample was ata well-defined temperature; therefore,experiments were carried out at 40 Kwith the sample bathed in coldflowing He gas. Radiation damage tothe sample was minimised byshortening the collection time (lessthan 1 minute) and accumulating thespectra at multiple sites on the sample.

Figure 8 compares the experimentalXAS for the various forms of ice andwater. In general, the oxygen K-XAScan be divided into three regions.A pre-edge region (I) starts from theabsorption threshold at 533 to 536 eV.A near-edge region (II) from 537 to539 eV, and a post-edge region (III)from 539 eV and beyond. All the icesamples show a pre-edge absorption at535 eV. Moreover, the intensity of thispeak decreases in the order from liquidwater to HDA, and is almost the samefor cubic ice Ic, hexagonal ice Ih, andLDA. In addition, an unusual trend isobserved in regions II and III where theXAS can be divided into two distinctgroups with similar absorption profiles.The first group consists of water andHDA ice (group A) and the secondgroup consists of ice Ic, ice Ih, and LDA(group B). In the near-edge region (II)the intensities of group A are strongerthan group B but the reverse isobserved in the post-edge region (III).Comparison of the XAS profile forgroups A and B suggests that theoscillator strengths of the excitations inthe near-edge region for group B havebeen shifted into the post-edge region.

The experimental results show thatcore-hole excitations are localised andnot strongly affected by the localenvironment. The observed XASprofiles can be correlated to thedegree of the distortion of H-bondnetwork of the ice polymorphs. Inhighly disordered ices, the intensity ofthe XAS is stronger near the oxygen-K

Fig. 8: X-ray Raman spectra for LDA, HDA, ice Ic, ice Ih, H2O, and D2O water.The spectra were normalised to equal area up to 550 eV. The XAS spectra can be

divided into three regions: (I) pre-edge, (II) near-edge, and (III) post-edge. The order ofthe structures starting from the top peak in the pre-edge region is indicated in the

lower central part of the figure. The structures that are in the highest peaks in regionsII and III are indicated in each of these regions.

Principal publication andauthorsJ.S. Tse (a), D.M. Shaw (a),

D.D. Klug (b), S. Patchkovskii (b),

G. Vankó (c), G. Monaco (c),

M. Krisch (c), Phys. Rev. Lett. 100,

095502 (2008).

(a) Physics and Engineering

Physics, University of

Saskatchewan, Saskatoon

(Canada)

(b) Steacie Institute for Molecular

Sciences, National Research

Council of Canada, Ottawa

(Canada)

(c) ESRF



References[1] Thematic issue on Water,

edited by L.R. Pratt, Chem. Rev.

102, 2625 (2002).

[2] Ph. Wernet et al., Science 304,

995 (2004).

[3] C.A. Tulk et al., Science 297,

1320 (2002).

Dispersion relation of an OH-stretchingvibration from inelastic X-ray scattering

Hydrogen bonds influence structure-property relations in a very largevariety of compounds. Consequently,numerous experimental techniques areemployed to study the detailed atomicarrangement and the dynamics ofhydrogen bonds between R-O-H andO-R’ groups, where R and R’ representparts of a structure, O-H represents asignificantly covalent bond with atypical distance of d(OH) ~1 Å. Thehydrogen bond exists between thehydrogen and the acceptor atom, witha typical distance of 1.5–1.72 Å. Thedynamics of hydrogen bonds in crystals,melts, and glasses are routinelycharacterised using infrared or Ramanspectroscopy. However, both infraredand Raman spectroscopy probe thehydrogen dynamics in the longwavelength limit and cannot provideinformation on the wave vectordependence of the stretching vibration.This is a significant limitation, as itprevents the elucidation of thecoupling of the hydrogen dynamics tothe remainder of the lattice dynamics.Due to a number of problems, such aslimited flux at high energies, intrinsiclimitation of the energy resolution andthe large incoherent scattering crosssection of hydrogen, coherent inelasticneutron scattering is not suitable forthe study of the dispersion of highfrequency OH stretching vibrations.

These limitations do not exist forinelastic X-ray scattering (IXS), and herewe demonstrate that the dispersionrelation of OH stretching frequenciescan be determined by IXS. Weperformed experiments on diaspore,α-AlOOH, a model system to studyhydrogen bonds of intermediatestrength, as it has a comparatively small

unit cell, simple chemistry, and is ofintermediate (orthorhombic) symmetry.It has recently been studied by DFTcalculations and by high-pressure singlecrystal X-ray diffraction up to 0.5 Mbar[1,2]. The hydrogen bond is slightlykinked, and the OH-stretchingfrequencies of the four symmetricallyinequivalent modes at ambient pressureare centred at ~367 meV [2]. Densityfunctional perturbation theory (DFPT)-based calculations using the CASTEPcode revealed that for some wavevectors the OH-stretching vibration hasan appreciable dispersion. Specifically,at reciprocal lattice points (0.5 0.5 0)

Principal publication andauthorsB. Winkler (a), A. Friedrich (a),

D.J. Wilson (a), E. Haussühl (a),

M. Krisch (b), A. Bosak (b),

K. Refson (c), V. Milman (d),

Physical Review Letters 101,

065501 (2008).

(a) Geowissenschaften, Goethe-

Universität, Frankfurt a.M.

(Germany)

(b) ESRF

(c) Rutherford-Appleton

Laboratory, Didcot (UK)

(d) Accelrys, Cambridge (UK)

Fig. 9: Predicted (line,offset for clarity) andexperimental (points)inelastic X-ray spectra ofdiaspore at (2.5 0 1) and(2.5 0 0), bothcorresponding to (0.5 0 0)in reduced lattice units.For these twoconfigurations the IXSspectra are dominated byeither one or the otherdoubly degeneratestretching mode. Theinsets show anenlargement of the high-energy part of thespectrum, and the verticalbars indicate thecalculated mode energy.

absorption edge. In contrast, forstructurally more ordered ices, the XASabsorption intensity is distributedbeyond the absorption edge. Thedifference in the absorption profilecan be interpreted as an enhancementof Wannier over Frenkel excitations inan ordered crystal. The results further

suggest important contributions fromcontinuum electronic states whichhave been neglected in mosttheoretical calculations. Finally, theexistence of the pre-edge feature isnot a concise indicator of themagnitude of local disorder within thehydrogen bonded network.


12


EE SS RR FF

and (0.5 0 0), we predicted that thereare two doubly degenerate vibrationswith energies of approximately 360and 378 meV, respectively, while at(0.5 0 0.5), all four stretching vibrationshave energies of ~ 361 meV.

The experiment was carried out on theIXS beamline II (ID28). The instrumentwas operated using the silicon (8 8 8)configuration, which provides a totalinstrumental energy resolution of6 meV full-width-half maximum. Froma gem quality natural sample we havecut a 750 µm thick (001)-slice. Themeasurements were performed atambient conditions. Counting timesranged from 60 to 2040 seconds perpoint. Optimal set-up parameters weredetermined prior to the experiment bycomputing theoretical spectra for a

variety of wave vector transfers fromthe DFPT-based lattice dynamicalcalculations. Figure 9 shows theexcellent agreement between thepredicted and experimentallydetermined spectra. Most notably, wefind mode energies of 357 (1) meV and374 (1) meV, in excellent agreementwith calculation. Furthermore thedispersion of 17 meV as a function ofq was confirmed by measurements foradditional reciprocal lattice points.

An analysis of the eigenvectors (shownin Figure 10) revealed that for the highenergy mode, the H-H distancesremain constant during the vibrationat about 2.42 Å, while for theantisymmetric low energy phonon, thehydrogens are further apart for mostof the time. Hence, the H-H repulsionenergy is lowered as the modeamplitude increases, lowering thepotential and consequently thefrequency of this mode.

The present experiment shows thatdispersion relations at very highfrequencies can be measured by IXS,thus offering new opportunities, forexample, in the study of phasetransformations which are governedby a change in the hydrogen bonding.

References[1] A. Friedrich, D.J. Wilson,

E. Haussühl, B. Winkler,

W. Morgenroth, K. Refson, and

V. Milman, Phys. Chem. Miner. 34,

145 (2007).

[2] A. Friedrich, E. Haussühl,

R. Boehler, W. Morgenroth,

E.A. Juarez-Arellano, B. Winkler,

Am. Mineral. 92, 1640 (2007).

Fig. 10: Eigenvectors of thephonons at the Brillouin zone

boundary at (0.5 0 0), showing thatduring the high frequency

vibration (left) the hydrogenremain at a constant distance,

while during the low frequencymotion (right) the hydrogens movefurther apart. Purple red and white

spheres represent Al, O and Hatoms, respectively.

Lattice dynamics of rubidium-IV, anincommensurate host-guest system

In recent years, a number ofsurprisingly complex crystal structureshave been discovered in the elementsat high pressures, in particularincommensurately modulatedstructures and incommensurate host-guest composite structures (see [1] fora review). The crystal structure of thehigh-pressure phase rubidium-IVshown in Figure 11 belongs to thegroup of incommensurate host-gueststructures that have also beenobserved in the elements Na, K, Ba, Sr,Sc, As, Sb, and Bi. The structurecomprises a framework of rubidiumhost atoms with open channels thatare occupied by linear chains ofrubidium guest atoms, and theperiodicities of the host and guestsubsystems are incommensurate with

each other (i.e., they have a non-rational ratio). Although considerableprogress has been made indetermining the detailed crystalstructures of the complex metallicphases at high pressure, little is knownabout their other physical properties,and the mechanisms that lead to theirformation and stability are not yetfully understood.

We investigated the lattice dynamics inincommensurate composite Rb-IV byinelastic X-ray scattering (IXS) onbeamline ID28. The focus was on thelongitudinal-acoustic (LA) phononsalong the direction of theincommensurate wavevector (parallelto the guest-atom chains). Calculationson simpler model systems predict these

Principal publication andauthorsI. Loa (a), L.F. Lundegaard (a),

M.I. McMahon (a), S.R. Evans (a),

A. Bossak (b), and M. Krisch (b),

Phys. Rev. Lett. 99, 035501 (2007).

(a) The University of Edinburgh

(UK)

(b) ESRF



Reference[1] M.I. McMahon and

R.J. Nelmes, Chem. Soc. Rev. 35,

943 (2006).

phonons to reflect theincommensurability most clearly.Phase IV of Rb is stable at pressures of16 to 20 GPa at room temperature,and a high-quality single crystal ofRb-IV was grown in a diamond anvilhigh pressure cell. In the IXSexperiment, the incident radiation wasmonochromatised at a photon energyof 17.8 keV, and two grazing-incidence mirrors focussed the X-raysonto the sample with a focal size of25 x 60 µm. The spectrum of thescattered radiation was analysed by ahigh-resolution silicon crystal analyserto yield an overall energy resolution of3 meV.

Figure 11 shows a typical IXS spectrumof Rb-IV along with its decompositioninto the elastic line, the phononexcitation peaks and a constantbackground, which were obtained byleast-squares fitting using the FIT28software. From a series of IXS spectracollected for different momentumtransfers Q, phonon dispersion curveswere obtained as shown in Figure 12a.A central result of this study is theobservation of two well-definedlongitudinal-acoustic (LA)-type phononbranches along the chain direction.They are attributed to separate LAexcitations in the host and the guestsublattices, which is a unique featureof an incommensurate compositecrystal.

A series of dispersion curves wasmeasured at different pressures, andfrom this the sound velocities of thehost and guest excitations and theirpressure dependences weredetermined (Figure 12b). While theabsolute values of the sound velocitiesin the host and the guest are rathersimilar, their pressure dependencesdiffer notably. A simple ball-and-springmodel of Rb-IV with only one springconstant reproduces theseobservations semi-quantitatively. Thissuggests that the difference in thepressure dependences is determinedlargely by geometrical factors, i.e., bythe spatial arrangement of the atomsrather than differences in the chemicalbonding in the two subsystems.

There is only very weak couplingbetween the incommensurate hostand the guest in Rb-IV, which raises a

rather interesting question. Can the 1Dchains of guest atoms in Rb-IV beconsidered a manifestation of the“monatomic linear chain” treated insolid-state physics textbooks tointroduce the concepts of crystallattice dynamics? The pressuredependence of the interatomic spacingin the guest-atom chains wasmeasured in earlier structural studiesand enables the spring constant in thelinear chain model to be determined,and also its pressure dependence. Onthis basis, the sound velocity in thelinear chains and its pressuredependence were modelled as shownin Figure 12b. The results are inexcellent agreement with the IXS datafor the guest-atom chains in thecomposite Rb-IV structure, which canthus be regarded as a manifestation ofthe monatomic linear chain modelwith regard to the LA phonons.

Fig. 11: Inelastic X-rayscattering spectrum ofRb-IV at 17.0 GPa, withthe scattering vectorq = (0 0 3.2)h referringto the host lattice. Theinset shows thecomposite crystalstructure of Rb-IV withthe rubidium host andguest atoms in blue andred, respectively.

Fig. 12: a) Dispersion relations oflongitudinal-acoustic phonons ofthe host and guest subsystem inRb-IV. b) Pressure dependences ofthe host and guest LA soundvelocities along the chain directionand results of the monatomiclinear chain model. The dashedline indicates the pressuredependence obtained with thelinear chain model.


14


EE SS RR FF

Vibrational properties of nanograinsand interfaces in nanocrystalline materials

The dynamics of nanocrystallinematerials has attracted a lot ofscientific interest during the lastdecade due to the striking differencesobserved for their atomic vibrationsrelative to the bulk counterparts.These anomalies are the enhancementof their density of phonon states (DOS)at low and high energies andbroadening of the phonon peaks [1].In addition, the energy dependence ofthe low-energy part of their phononDOS has been a source of long-standing debates. The experimentalresults are contradictory, reporting alinear dependence [2], a power lawwith n = 1.33 [3], and a quadratic(Debye-like) behaviour ([4, 5], andreferences therein). On the otherhand, the theoretical calculations havealso indicated that non-Debyedynamics could originate from theatoms located at the surfaces [6], atthe grain boundaries [7], or in theporous areas [8] of the nanocrystallinematerials. Therefore the thoroughunderstanding of the atomic dynamicsin these materials is of significantimportance not only for fundamentalphysics, but also for tailoring of theirbeneficial properties like enhanced

strength and hardness, and improvedplasticity compared to the coarse-grained materials.

In order to investigate the vibrationaldynamics of the nanograins andinterfaces systematically, we studied ananocrystalline Fe90Zr7B3 alloyprepared by crystallisation of anamorphous precursor. A ribbon withcomposition Fe90Zr7B3 enriched to 63%in 57Fe was produced by the melt-spinning technique. Severalnanocrystalline samples composed ofα-Fe nanograins and homogeneousand porosity-free interfaces wereprepared by annealing of the as-quenched ribbon in a vacuum of1.6 x 10-6 mbar. The samples werecharacterised by X-ray diffraction(XRD), which did not show thepresence of oxides or other crystallinephases. The average grain sizes weredetermined by Rietveld refinement ofthe XRD patterns and confirmed bytransmission electron microscopy.The high sensitivity of the Mössbauerspectroscopy to the local environmentof the resonant nuclei (57Fe) wasemployed to quantitatively determinethe amount of iron located within thenanograins and the interfaces atvarious crystallisation stages. Thepartial, Fe-projected DOS was obtainedfrom the nuclear inelastic scatteringspectra measured with an energyresolution of 1.0 meV at beamlineID18. A simple model allowedseparation of the DOS of thenanograins from that of the interfacesfor a wide range of grain sizes andinterface thicknesses. Figure 13 showsthat the phonon DOS of thenanograins does not depend on theirsize and remains close to that of thebulk even for 2 nm particles(Figure 13a). Furthermore, Figure 14aand Figure 14b show that the phononDOS of the interfaces is entirelyresponsible for the observed anomaliesexhibiting a shape typical for matterwith a high degree of structuraldisorder. Reducing the interfacethickness to the sub-nanometre rangeresults in a dramatic transformation ofthe phonon DOS (Figure 14c). Theexcess of phonon states at low and

Principal publication andauthorsS. Stankov (a), Y.Z. Yue (b,c),

M. Miglierini (d), B. Sepiol (e),

I. Sergueev (a), A.I. Chumakov (a),

L. Hu (b,c), P. Svec (f), R. Rüffer (a),

Phys. Rev. Lett. 100, 235503

(2008).

(a) ESRF

(b) Section of Chemistry, Aalborg

University (Denmark)

(c) Shandong University, Jinan

(China)

(d) Slovak University of

Technology, Bratislava (Slovakia)

(e) University of Vienna (Austria)

(f) Slovak Academy of Sciences,

Bratislava (Slovakia)

Fig. 13: The phonon DOS of α-Fe nanograins as a

function of their size d.

References[1] B. Fultz et al., Phys. Rev. Lett.

79, 937, (1997).

[2] U. Stuhr et al., Phys. Rev. Lett.

81, 1449 (1998).

[3] B. Roldan Cuenya et al., Phys.

Rev. B 76, 195422 (2007).

[4] T. Slezak et al., Phys. Rev. Lett.

99, 066103 (2007).

[5] S. Stankov et al., Phys. Rev.

Lett. 99, 185501 (2007).

[6] A. Kara and T.S. Rahman, Phys.

Rev. Lett. 81, 1453 (1998).

[7] P.M. Derlet et al., Phys. Rev.

Lett. 87, 205501 (2001).

[8] C. Hudon et al., Phys. Rev. B

76, 045409 (2007).



high energies is suppressedsignificantly and characteristic peaksfor the bulk Fe are observed. Thissuggests that in such thin interfaces(about 2 atomic layers thick) the atomsfollow the vibrations of theneighbouring nanograins.

In summary, the anomalous dynamicsof the bulk nanocrystalline materialsoriginate from the disorderedinterfaces, while the phonon DOS ofthe nanograins is close to that of thebulk counterparts and sizeindependent. The low-energyenhancement of the phonon statesscales linearly to the atomic fraction ofthe interfaces and perfectly obeys theDebye law, thus excluding thepresence of low dimensional effects.The deviation of the thermo-elasticproperties (vibrational entropy, mean-

square displacement,average forceconstants, specific heat)from the correspondingbulk values also followsa similar lineardependence.

Fig. 14: The phonon DOSof α-Fe nanograin interfaces

as a function of theirthickness δ.

Exploring the dynamic platinumstructure during CO oxidation

The oxidation of carbon monoxide isone of the most intensively studiedreactions in the field of catalysisbecause of its role in the purificationof feed gas for fuel cells andautomotive exhaust gas. Knowledge ofthe structure of the catalytically-activesites in these solid catalysts is essentialto understand the functioning of solidcatalysts and chemical processes.Despite the large effort in research inthis field, fundamental questionsabout the active species and reactionmechanism remain disputed. Ertl andco-workers showed that on singlecrystals under low-pressure conditions,varying reconstruction of the platinumsurface, which occurs after adsorptionof carbon monoxide, leads to carbonmonoxide-rich and oxygen-richdomains that have different reactionrates [1]. High-pressure experiments onsingle crystals however, suggested thatoxidised platinum is the active phase[2]. In the work presented here, wehave bridged both the pressure andmaterials gaps by determining thestructure of a supported platinumcatalyst during oxidation of carbonmonoxide under real conditions.Bridging the pressure and materialsgaps is essential to translate results

from single crystal studies andunderstand catalytic processes underrealistic conditions. This was achievedby combining in situ, time-resolved,and high-energy resolutionfluorescence X-ray spectroscopy(HERFD XAS) [3,4] with kineticmeasurements. HERFD yields sharperspectral features because theinstrumental energy broadening isbelow the width due to the core-holelifetime, which enables detection ofconsiderably more spectral detail.HERFD XAS were collected at beamlineID26 of ESRF and the time-resolvedXAFS data was measured at the superXAS beamline of the Swiss LightSource.

Similar to results on single crystals[2,5], we observed a low- and a high-activity state of the catalyst (Figure 15).These two regimes were only observedat ratios of oxygen to carbonmonoxide greater than stoichiometric.A sudden increase in activity to thehigh-activity regime was observedduring heating. This so-called ignitionoccurred at lower temperature withincreasing oxygen concentration. TheHERFD spectra, which were obtainedevery two minutes, showed large

Principal publication andauthorsJ. Singh (a), E.M.C. Alayon (a),

M. Tromp (b), O.V. Safonova (c),

P. Glatzel (d), M. Nachtegaal (e),

R. Frahm (f), J.A. van Bokhoven (a),

Angew. Chem. Int. Ed. 47, 9260

(2008).

(a) Institute of Chemical and

Bioengineering, ETH Zürich

(Switzerland)

(b) School of Chemistry, University

of Southampton (UK)

(c) Swiss Norwegian Beamlines

(SNBL), ESRF, Grenoble (France)

(d) ESRF

(e) Paul Scherrer Institute, PSI,

Villigen (Switzerland)

(f) Fachbereich C / Physik,

University of Wuppertal

(Germany)


16


EE SS RR FF

changes between spectra taken belowand above the ignition temperature.Below ignition, a whiteline of lowintensity with a double feature wasobserved, which is characteristic ofplatinum particles with adsorbedcarbon monoxide [4]. At temperaturesabove the ignition temperature, thespectra showed a strong increase inthe intensity of the whiteline, whilethe edge energy shifted to lowerenergy. These features arecharacteristic of oxidised platinum [4].The amount of oxidised platinumvaried with conversion; at higherconversions, higher amounts of oxidewere observed. This suggests that thehigh-activity regime is characterised bythe presence of oxidic platinum, whichis most likely present on the surface.

During the ignition, the spectrarecorded with a time resolution of0.5 seconds, using quick XANESmeasurements, showed increasingamounts of platinum oxide, whichautocatalytically increased theconversion. This suggests that it is theformation of platinum oxide thatenhances the catalytic rate. In contrastto single crystals under low pressures,which are covered by chemisorbedoxygen, nano-sized particles undergooxidation.

This work showed that the catalyst isin a different structure in the low- andthe high-activity regime: adsorbedcarbon monoxide on platinum in thelow-activity regime, that poisons thesurface, and partially oxidic platinumin the high-activity regime as observedwith in situ HERFD XAS and time-resolved XAFS. The rate-limiting stepin the low-activity regime is desorptionof carbon monoxide from the surfaceand subsequent dissociative adsorptionof oxygen. In the high-activity regime,the catalyst surface is oxidised, becauseof low concentration of carbonmonoxide at the catalyst surface. Thissurface shows a high rate of reaction.High temperature and a high oxygenconcentration benefit the formation ofthe more active catalyst. Highresolution XAS enabled thedetermination of the structure ofsupported nanoparticles in great detailand of loadings as low as 2 wt%.Because hard X-rays are involved inthis experiment, structural informationwas obtained of the catalyst underreal working conditions.

Fig. 15: Rate of oxidation of carbon monoxide over 2 wt% Pt/Al2O3 during heating(5 K/min) at oxygen to carbon monoxide ratios of 1 (green), 2 (blue), and 5 (pink).

Pt L3 edge HERFD XANES are shown in the low- and the high-activity regime.

References[1] G. Ertl, P.R. Norton, J. Ruestig,

Phys. Rev. Lett. 49, 177 (1982);

G. Ertl, Surf. Sci., 287-288, 1

(1993).

[2] M.D. Ackermann et al., Phys.

Rev. Lett. 95, 255505 (2005).

[3] P. Glatzel, U. Bergmann,

Coord. Chem. Rev. 249, 65 (2005).

[4] O.V. Safonova, M. Tromp,

J.A. van Bokhoven, F.M.F. de

Groot, J. Evans, P.Glatzel, J. Phys.

Chem. B 110, 16162 (2006).

[5] X. Su, P.S. Cremer, Y.R. Shen,

G.A. Somorjai, J. Am. Chem. Soc.

119, 3994 (1997).

Visualising the ignition of catalyticpartial oxidation of methane

Catalytic reactions are present in manychemical processes. More than 90% ofall chemical products have undergoneat least one catalytic step. Thedevelopment of new catalysts hasoften been triggered by thedevelopment of new techniques.Especially important is energy-relatedcatalysis, to make catalysts moreefficient and more stable and tosubstitute fossil fuels e.g. by biomass-

related processes. One importantreaction is the catalytic partialoxidation (CPO) of methane for theproduction of synthesis gas, which iscatalysed by noble metals such as Pt,Rh and Ir. It can be regarded a keystep in transformation of natural gasor biomass into liquid fuels such asmethanol and higher alkanes [1,2].Changes in chemical reactions like theCPO with time and/or in space are

Principal publication andauthorsB. Kimmerle (a), J.-D. Grunwaldt (a,b),

A. Baiker (a), P. Glatzel (c),

P. Boye (d), S. Stephan (d) and

C.G. Schroer (d), J. Phys. Chem. C,

DOI:10.1021/jp810319v (2009).

(a) Institute for Chemical and

Bioengineering, Department of

Chemistry and Applied Biosciences,

ETH Zürich (Switzerland)

(b) Department of Chemical and

Biochemical Engineering, Technical

University of Denmark (Denmark)

(c) ESRF

(d) Institute of Structural Physics,

TU Dresden (Germany)



often correlated to a change in stateand structure of the catalyst [3]. Togain deeper insight, it is desirable tofollow such structural changes in aspatiotemporal manner.

We studied the ignition of thecatalytic partial oxidation of methaneon a Pt-Rh/Al2O3 catalyst by spatiallyand time-resolved imaging of acatalytic reactor using a FReLoN (fastreadout low noise) camera behind themicroreactor [4]. The setup is shown inFigure 16.

The oxidation state of the noblemetals is strongly dependent on thereaction conditions. Below the ignitiontemperature of the reaction towardsCO and H2, the catalyst is in anoxidised state, whereas above theignition temperature, a reduction ofthe noble metals occurs. This can befollowed by X-ray absorptionspectroscopy (see e.g. ref. [3]). Inparticular, a strong change in X-rayabsorption occurs at the whiteline ofplatinum (11568 eV). By tuning theenergy of the incident X-rays to themaximum of the whiteline, a variationin X-ray absorption can uncover wherethe platinum constituent is reduced. Infact, by collecting X-ray transmissionimages of the microreactor with aframe rate of 4s-1, a strong structuralchange in X-ray absorption, and thusthe structure of the catalyst, could befound during the ignition of the CPOof methane. Selected images at timest1 (Pt is still in an oxidised state overthe whole catalyst bed) and t > t1 aredepicted in Figure 17. Reddish andviolet colours indicate a lower X-rayabsorption and thus reduced Pt-species.

Images 2a) to 2e) were obtained bysubtraction of the X-ray absorptionimage at time t from Figure 17f) belowthe ignition, which emphasises thedifference in absorption and correctsfor inhomogeneities in the sample.The full series taken of the ignitionprocess was compiled into a movie andwe observed that the front of reducedplatinum species moves from theoutlet towards the inlet of the reactor.The observed reduction of the noblemetal occurred at the same time as thechemical reaction ignited (detected byon-line mass spectrometry). Note thatthe ignition starts at some specific

single particles in the catalytic reactor(compare Figures 17b and 17c).

The movement of the front from theoutlet towards the inlet can beexplained with a total oxidation -reforming mechanism. Below theignition temperature methane is fullycombusted to carbon dioxide andwater over the whole length of thecatalyst bed. Upon heating thereaction self-accelerates on some ofthe catalyst grains until all oxygen isconsumed and then some new metallicspecies are formed. These new speciespromote the activation of methaneand thus lead to the formation ofhydrogen and carbon monoxide.Furthermore, the methane oxidationaccelerates and the reductionpropagates towards the inlet of thereactor, where the total oxidation ofmethane still occurs over oxidisednoble metal particles.

Fig. 16: Schematic figure of thecatalytic microreactor for partialoxidation of methane to CO andH2 on a 2.5wt%Pt-2.5wt%Rh/Al2O3

catalyst; a 1 mm x 1 mm snapshottaken with the FReLoN camera isdepicted below.

Fig. 17: a) to e): Images recorded at the whiteline energy of Pt (Pt L3-edge; 11586 eV)as a function of time with flow conditions as depicted in Figure 16: reddish colourindicates the formation of reduced Pt-species; f) X-ray absorption image recordedbelow the CPO ignition temperature, showing that the images were taken at the endof the catalyst bed composed of Pt-Rh/Al2O3 on the left and only inert aluminamaterial on the right.


18


EE SS RR FF

In conclusion, by using high resolutiontransmission X-ray imaging we could

determine the structure of a catalystnot only in situ but also in aspatiotemporally resolved manner.Thereby, important insight into thefirst stages of the ignition reactioncould be gained. This allows anunderstanding of the ignition/extinction of catalytic reactions,reaction mechanisms and also chemicaloscillations in more detail.

References[1] D.A. Hickman, and L.D. Schmidt, Science 259, 343 (1993).

[2] J.-D. Grunwaldt, L. Basini, B.S. Clausen, J. Catal. 200, 321 (2001).

[3] J.-D. Grunwaldt, S. Hannemann, C.G.Schroer, A. Baiker, J. Phys. Chem. B 110, 8674 (2006).

[4] J.-D. Grunwaldt, B. Kimmerle, A. Baiker, P. Boye, C.G. Schroer, P. Glatzel, C. Borca,

F. Beckmann, Catal. Today, on-line, doi: 10.1016/j.cattod.2008.11.002.

Nuclear resonant spectroscopy on 121Sband 125Te

Nuclear forward (NFS) and nuclearinelastic scattering (NIS) havedeveloped rapidly during the last tenyears and have been expanded tomany isotopes beyond 57Fe. NIS is apowerful tool for the study of latticedynamics. It complements othertechniques like inelastic neutronscattering: the vibrational density ofstates (VDOS) can be derived directlyfrom the measured inelastic spectrumand the technique is element specific.However, only materials comprising aMössbauer isotope can be studied.Fortunately, many interesting materialswith Mössbauer isotopes are en vogue,with even some materials consistingonly of ‘Mössbauer elements’.

High-resolution monochromators areessential for the NIS technique.To resolve phonon spectra, themonochromator has to filter out a meV

spectral bandwidth from the broadX-ray spectrum of synchrotronradiation. For X-ray energies lower than30 keV, multi-bounce high-resolutionmonochromators based on siliconcrystals have been developed verysuccessfully, but above 30 keV theirimplementation can only be achievedwith a major loss in reflectivity [1].Our studies on 121Sb and 125Te becamepossible by utilising a sapphire Braggbackscattering high resolutionmonochromator, an alternative tosilicon, and more efficient in the rangefrom 30 keV to 60 keV, becausesapphire offers sufficient angularacceptance, (sub-)meV resolution andhigh reflectivity [2]. Sapphire alsoprovides a large choice of Bragg back-reflections, matching the fixed nucleartransition energies at certaintemperatures. The 121Sb resonance at37.1298 keV could be studied with the(15 13 -28 14) reflection at a sapphiretemperature T0 = 146.54 K and the125Te resonance at 35.4931 keV couldbe studied with the (20 6 -26 2)reflection at T0 = 206.22 K. Because theenergy varies by ~100 meV per K, atemperature stabilisation in the mKregime is used. The Bragg angle is keptconstant at 89.9 degrees and theenergy of the reflected photons ischanged by a controlled change of thecrystal temperature.

An interesting compound built from‘Mössbauer elements’ is the EuFe4Sb12skutterudite, a thermoelectric andferromagnetic material, where ourrecent NIS studies on Sb havecompleted the full element specificVDOS study of all three constituents.Figure 18 shows the VDOS of Sb in

Principal publication andauthorsH.-C. Wille (a,b), R. P. Hermann (c,d),

I. Sergueev (b), O. Leupold (a),

P. van der Linden (b), B.C. Sales (e),

F. Grandjean (d), G.J. Long (f),

R. Rüffer (b), and Yu.V. Shvyd’ko (g),

Phys. Rev. B 76, 140301(R) (2007);

H.-C. Wille et al., Phys. Rev. B

(submitted).

(a) Hamburger

Synchrotronstrahlungslabor

(Germany)

(b) ESRF

(c) Institut für Festkörperforschung,

Forschungszentrum Jülich

(Germany)

(d) Department of Physics B5,

Université de Liège (Belgium)

(e) Solid State Division, Oak Ridge

National Laboratory (USA)

(f) Department of Chemistry,

University of Missouri-Rolla (USA)

(g) Advanced Photon Source,

Argonne National Laboratory,

Illinois (USA)

Fig. 18: The element specificvibrational density of states (VDOS)

of Sb, Eu and Fe (weighted byelemental content per formulaunit) in EuFe4Sb12 and of Sb in

CoSb3. The difference of the SbVDOS in the filled and unfilled

skutterudites has a maximum atthe local mode of the rattler (Eu),

which indicates rattler-cagehybridisation.



References[1] S. Tsutsui, Y. Yoda and

H. Kobayashi, J. Phys. Soc. Jpn. 76,

065003 (2007).

[2] Yu.V. Shvyd’ko and E. Gerdau,

Hyp. Int. 123/124, 741 (1999).

EuFe4Sb12 and CoSb3 and the VDOS ofEu and Fe in EuFe4Sb12. Forcomparison the difference in the SbVDOS is shown. The peak in thedifference at the same energy as thepeak in the Eu VDOS gives evidencefor a coupling of the cage (Sb) and therattler (Eu) in skutterudites. Thishybridisation is needed to obtain alow thermal conductivity, which incombination with a high electricconductivity and a large Seebeckcoefficient results in a largethermoelectric figure of merit.

Other examples of pure ‘Mössbauer’compounds are phase changematerials used in rewritable opticaldata storage that have the formulaGexSbyTez. In a first experiment, shownhere, we demonstrated theapplicability of NFS and NIS on the125Te isotope, using the sapphiremonochromator. Figure 19a shows theNFS time spectrum of 125Te metal. The125Te transition has only 1.48 ns half-life, which makes this transition thenuclear transition with the shortestlifetime ever studied by NFS and NIS.Fast detector electronics allowedcounting to start only 2 ns after theexciting synchrotron radiation bunch,considerably earlier than for standardNFS timing experiments. The extracted

quadrupole splitting in Te metal is7.85 mm/s in good agreement with7.77 mm/s obtained by classicalMössbauer spectroscopy, correspondingto 2.97 times the natural linewidthΓ0 = 0.313 µeV. The possibility forcarrying out NIS measurement wasdemonstrated with a sample ofSb2

125Te3. The count rate of forward(NFS) and inelastic (NIS) scattering withrespect to the incoming photon energyis shown in Figure 19b. The NFS signalrepresents the instrumental resolutionof the HRM, 2.6 meV. The NIS signalyields the phonon spectrum of thematerial. The experiments were carriedout at the beamline ID22N.

Principal publication andauthorsA. Bosak (a), M. Hoesch (a),

M. Krisch (a), D. Chernyshov (b),

P. Pattison (b), B. Winkler (c),

V. Milman (d), K. Refson (e),

D. Farber (f,g), D. Antonangeli (h),

submitted to Phys. Rev. B.

(a) ESRF

(b) Swiss-Norvegian Beam Lines,

Grenoble (France)

(c) Geowissenschaften, Goethe-

Universität, Frankfurt a.M.

(Germany)

(d) Accelrys, Cambridge (UK)

(e) Rutherford-Appleton Laboratory,

Chilton, Oxfordshire (UK)

(f) Lawrence Livermore National

Laboratory, Livermore (USA)

(g) University of California, Santa

Cruz (USA)

(h) Institut de Minéralogie et de

Physique des Milieux Condensés,

UMR CNRS 7590, Institut de

Physique du Globe de Paris,

Université Paris 6 et 7 (France)

Fig. 19: a) the NFS time spectrumof Te metal (black) and thetheoretical curve (brown); b) theNFS (black) and NIS (brown) energyspectra of Sb2Te3. The instrumentalfunction (NFS) shows a resolutionof 2.6 meV.

3D-imaging of the Fermi surface bythermal diffuse scattering

The Fermi surface is one of the mostimportant theoretical concepts in solidstate physics. It is the surface inelectron momentum space of a metalthat separates occupied and emptyelectronic states. Indeed, the shape ofthe Fermi surface controls the electric,magnetic and thermal properties ofmaterials. The most commonly usedexperimental techniques for thedetermination of the Fermi surfaceare the oscillation of transportproperties in a magnetic field (i.e. thede Haas-van Alphen effect and theShubnikov-de Haas effect) and angle-resolved photoemission spectroscopy(ARPES). Two-photon positronannihilation and Compton scatteringare more indirect methods of probingthe nature of the Fermi surface.

Interestingly, one of the earliesttechniques proposed to study theFermi surface was to look for theabrupt variations in the dispersions ofphonons [1]. These so-called Kohnanomalies arise from changes in thecharge screening by the conductionelectrons when the phonon wavevector q exceeds the spanning acrossthe Fermi surface k = 2kF. Here, weexploit the fact that the phonondispersions of a material give riseto thermal diffuse scattering (TDS)of X-rays that reflect the dispersionanomalies in characteristic intensityvariations of the TDS. Using only astandard X-ray diffraction set-up andan area detector, we show thatportions of the zinc Fermi surface canbe visualised in three dimensions.


20


EE SS RR FF

The experiment was performed on theSwiss-Norwegian Beam Lines (SNBL)BM01 in transmission geometry atλ = 0.722 Å using a MAR345 imageplate. A high-quality, 30-µm-thick, zincsingle crystal (MaTecK GmbH) wasmounted on a rotation stage with its<110> direction a few degrees off therotation axis. Diffuse scatteringpatterns were recorded with anincrement of 1° over an angular rangeof +/- 60°. A 15-µm-thick nickel foilwas used to reduce the fluorescencesignal from zinc. Three-dimensionalmaps were reconstructed, using theCrysAlis software (Oxford Diffraction).Figure 20a shows the reconstructeddiffuse scattering in the HHL plane.Two types of lens-like objects ofdifferent size are clearly visible. Largerlenses, i.e. centred around (0 0 2n)Bragg spots have the strongestintensity in the centre, whereas thesmaller lens-like objects, i.e. centredaround (1/2 1/2 3) appear hollow.We assign features in the intensityvariation maps of the TDS to Kohnanomalies and thus to portions of theFermi surface. To further support ourinterpretation, we performed ab initiocalculations of the Fermi surface and

the TDS patterns, using the Wien2k [2]and CASTEP [3] algorithms.As Figure 20c reveals, the large lens-like objects correspond to the innerpart of the Fermi surface around theΓ point. The shape of the inner part ofthe Fermi surface makes thecorresponding surface of dispersionanomaly locations (Kohn surface) justits homothetic transformation with ascaling factor 2 (q = 2kF). The hollowlens-like objects of smaller size are, infact, just sections of the large “lenses”(see Figure 20d). Additional support isgiven by the modelling of the TDSintensity (see Figure 20b). While theexact shape and sharpness of Kohnanomalies are not reproduced with theCASTEP calculation, the overallresemblance in terms of contrast andintensity distribution is veryconvincing.

Our proof-of-principle demonstrationunderlines the great potential of TDSin the study of lattice dynamics andelectron-phonon coupling. We haveshown that the 3D-shape of the Kohnsurface can be reconstructed withinthe timescale of hours using standardX-ray diffraction equipment. It isimportant to emphasise that, by usingan insertion device synchrotronbeamline and energy-selective areadetectors, the data quality and datacollection efficiency could be furtherincreased by some orders ofmagnitude.

Fig. 20: a) (HHL) plane reconstruction of the experimental X-ray scattering data;b) model TDS pattern derived from ab initio phonon calculations (CASTEP);

c) zinc Brillouin zone and (HHL) Fermi surface section as obtained from ab initioelectron band structure calculations (Wien2k); d) scheme of the assembly of

lens-like Kohn surfaces centred around strong Bragg spots and their dissectionby HHL planes. Experimental features not obeying the symmetry operations are

experimental artifacts due to the non-linear image plate response anddiffraction from the nickel filter.

References[1] W. Kohn, Phys. Rev. Lett. 2,

393 (1959).

[2] P. Blaha, K. Schwarz,

G.K.H. Madsen, D. Kvasnicka,

J. Luitz, Wien2k, An Augmented

Plane Wave + Local Orbitals

Program for Calculating Crystal

Properties (Karlheinz Schwarz,

Techn. Universität Wien, Austria),

2001.

[3] S.J. Clark, M.D. Segall,

C.J. Pickard, et al., Zeitschrift für

Kristallographie 220, 567 (2005).



Introduction

The properties of materials depend onthe way the constituent atoms combinewith each other, for example to formmolecules, the aggregation of moleculesand atoms into crystalline or disorderedsolids, liquids or glasses, and themicrostructural arrangement of thegrains of a material to make up the bulk.Synchrotron X-ray radiation providessome of the most powerful tools tostudy these structures and thereby gainan understanding of why materialsbehave as they do, and how propertiescan be tailored and improved for specificapplications. Many experiments at theESRF and CRG beamlines fall into thebroad category of materials science,exploiting the techniques of diffraction,scattering or imaging. Choosing just afew examples from such a broadexpanse of work is difficult. Theaccounts presented in this chapter areonly the very tip of the iceberg as far asmaterials research at the ESRF isconcerned.

Synchrotron radiation can be used tostudy the evolution of structure insystems. The first example involves thedissociation of an organometallicmolecule induced by a flash of laserlight, where the various dissociationproducts are identified, and thepathways by which these revert to thestarting structure are followed on thetimescale of nanoseconds, thus enablingfundamental chemical processes ofbond breaking and making to bestudied in detail. On a more humantimescale, experiments have revealedthe recrystallisation and growth ofnanoparticles of palladium on exposureto hydrogen gas, with implicationsperhaps for future solid hydrogen-storage systems, and of grain growth inan aluminium-based alloy on annealing.The mechanical properties of polymersdepend intimately on the conditions of

temperature and flow under which theyare formed and processed. In situ studiesof the crystallisation of high-densitypolyethylene give new insights into howmorphology and structure can be linkedto the thermo-mechanical history of thesample.

The interactions between liquids andsurfaces can have profound effects onthe properties of a system: everyoneknows that wet sand makes a bettersandcastle, and understanding how isof great importance for more seriousissues concerning landslides, or for theprocessing of powdered substances inthe food and pharmaceutical industries.Micro-tomographic imaging was used tovisualise the intricate networks of liquidthat form between small glass spheres(a model system) with different liquidcontents. In another study, high energyX-ray reflectivity probed the interactionbetween a sapphire surface and liquidscomposed of bulky charged ions (socalled ionic liquids, with potentialapplications including solvents foradvanced chemical syntheses), revealingthe extent to which the ions orderor disorder with varying chemicalcomposition and temperature.

The chemical elements, which might beexpected to be the simplest of systems,continue to surprise at temperaturesor pressures away from those ofeveryday life. Thus seven distinct phasesfor sodium were discovered within anarrow pressure and temperature rangeclose to 118 GPa, some of extraordinarystructural complexity. Sulphur at lowtemperatures and high pressures isfound to undergo a number ofstructural changes, including a changefrom a high density to low-densityamorphous phase, or at higher pressuresto an incommensurately modulatedstructure via a charge density waveinstability. For iron at high temperaturesand pressures, X-ray emissionspectroscopy, a local probe of the 3d

Materials Science


22

Materials Science

EE SS RR FF

Ultrafast X-ray solution scatteringreveals a new reaction intermediate in thephotolysis of Ru3(CO)12

Principal publication andauthorsQ. Kong (b), J.H. Lee (a),

A. Plech (c), M. Wulff (b),

H. Ihee (a), M.H.J. Koch (d),

Angew. Chem. Int. Ed. 47, 5550

(2008).

(a) Center for Time-Resolved

Diffraction, Department of

Chemistry, KAIST, Daejeon

(Republic of Korea)

(b) ESRF

(c) Fachbereich Physik der

Universität Konstanz (Germany)

(d) EMBL, Hamburg Outstation

(Germany)

spin magnetic moment, demonstratesthe collapse of the short-range magneticstate under these conditions.

The final examples in the chapterinvolve crystal structure analysis viahigh-resolution-powder or single-crystal techniques, giving insightinto the electronic behaviour andsuperconductivity in hole-doped Nd1-xSrxFeAsO; the solution of the crystalstructure of novel organometalliccomplexes of copper with peptides,which reveals a pleasing network ofhydrogen bonds; and metal-organicframework materials containing rare-earth ions that combine high thermalstability and other multifunctionalproperties (magnetism, luminescence,microporosity, hydrophobicity). Thecombination of a luminescent Ln3+

centre and the chosen ligands results in

materials that can sense ethanol in aireven in the presence of water.

Preparations are underway for theexperiments that will yield the articlesof future editions of the ESRFhighlights. Ideas and plans for improvedexperimental approaches are alwaysbeing sought, and, with the start ofthe ESRF Upgrade Programme, twobeamlines relevant to this chapter areunder consideration. The first is for high-energy diffraction using a microfocussedbeam for penetrating through or intosamples with high spatial resolution,and the second is a fully-dedicatedbeamline for pump-probe time-resolveddiffraction. These would each greatlyenhance the ESRF's capabilities in theseexciting areas.

A. Fitch

The triangular metal carbonyl clusterRu3(CO)12 is one of the simplestthermally-stable metal carbonyls.The complex is used in controlledphotoactivated synthesis wherespecific bonds in the complex arebroken upon irradiation at differentwavelengths. As the mechanismleading to the cleavage of metal-metalbonds is of great theoretical andpractical interest, the photolysis ofRu3(CO)12 has been extensively studiedby spectroscopy in solid matrices andin solution. Structural characterisationof the intermediates has, however,been very difficult. Recent ultrafastinfrared spectroscopy measurementshave shown that when solutions ofRu3(CO)12 in non-coordinating solventslike cyclohexane are excited witheither an ultraviolet (266 nm) or avisible (400 nm) optical pulse,competing reactions yield twotransient intermediates containingbridging carbonyls, Ru3(CO)11(µ-CO)(Intermediate 1) for the metal-metal

cleavage reaction channel andRu3(CO)10(µ-CO) (Intermediate 2)for the CO loss reaction channel,respectively [1]. Infrared spectroscopyspecifically monitors the time course ofthe concentration of these twointermediates via the characteristicabsorption bands of their bridgingcarbonyls. This leaves the possibilitythat other intermediates could gounnoticed, especially those containingonly terminal carbonyls withabsorption bands overlapping thoseof the parent molecule.

In contrast to infrared spectroscopy,the signal from time-resolved X-ray orelectron scattering containscontributions from all interatomicdistances in the volume probed bythe incident beam. In principle, thismakes it possible to detect all transientintermediates [2]. Except for thesimplest cases, however, there is nounique solution to recover the three-dimensional structural information



from scattering data and this inverseproblem can only be solved bymodelling based on theoreticalcalculations. This approach was usedto study the photodissociation ofRu3(CO)12 dissolved in cyclohexane in apump-probe experiment performed onbeamline ID09B. Visible (390 nm) laserpulses (2 ps) were used for excitationand 100 picosecond X-ray pulses forprobing the transient intermediates.In these experiments the sample flowsthrough a nozzle which produces athin layer of liquid. The pump-probesequence is repeated with differenttime delays between pump and probeat a frequency of 986.3 Hz and thescattered signal is accumulated on aMarCCD detector. The difference X-rayscattering intensities (q∆S(q,t))illustrating the structural changes dueto the laser excitation are shown inFigure 21a as a function of differenttime delays.

Initial attempts at fitting the curveswith only the known intermediates1 and 2 failed to give satisfactoryresults and suggested the presence ofa third intermediate. By modelling themeasured difference scatteringintensities ∆S(q,t) using the structuresof 16 putative intermediates calculatedby density functional theory (DFT),modelled in solution with moleculardynamics (MD) simulations, theconcentration of each of theintermediates could be determined(Figure 21b). The results show that

three intermediates, Ru3(CO)11(µ-CO),Ru3(CO)10(µ-CO) with bridged CO andRu3(CO)10 with terminal CO only, areformed at the onset of the reactionfrom the initial molecule Ru3(CO)12,indicating the rupture of Ru-C andRu-Ru bonds in Ru3(CO)12 by theoptical excitation. The newintermediate Ru3(CO)10 with onlyterminal CO dominates at all timedelays. It recombines non-geminatelywith one CO ligand to Ru3(CO)10(µ-CO)which eventually decays into thestarting molecule Ru3(CO)12 by non-geminate recombination with anotherCO. Ru3(CO)11(µ-CO) relaxes rapidly tothe parent molecule Ru3(CO)12through geminate recombination.

These results strikingly illustratethe complementary nature of ultrafastX-ray scattering and ultrafastspectroscopy. Indeed, a good fit to theexperimental X-ray scattering datacould only be obtained if intermediate3, which does not contain any bridging

Fig. 21: a) Time-resolved difference scattering intensities q∆S(q,t) as a function of timedelays after photolysis of Ru3(CO)12 in cyclohexane. The black dots correspond to theexperimental data and the red curves to the theoretical least-squares fits.b) Concentration changes of the relevant chemical species during the photoreaction asa function of time.

Fig. 22: Photolysis of Ru3(CO)12 incyclohexane probed in solution bytime-resolved X-ray scattering.Beside the two knownintermediates, Ru3(CO)11(µ-CO)and Ru3(CO)10(µ-CO), the hithertoundetected major intermediateRu3(CO)10 was found.


24

Materials Science

EE SS RR FF

Hydrogen-induced Ostwald ripening

The main aim in hydrogen storageresearch is to obtain high hydrogenconcentrations in a material thatpossesses suitable transport propertiesunder ambient conditions. A promisingway to favourably alter the material’sproperties is by reducing its size to theextent that surface and quantumeffects begin to play a major role.The investigation of nanocluster metalhydrides may therefore reveal novelproperties. Palladium is one of themost widely studied metals withrespect to hydrogen absorption.At elevated temperatures palladiumnanoclusters change size due toOstwald ripening: the larger clusterscapture mobile atoms at the expenseof smaller clusters [1]. As hydrogen ina metal can decrease the strength ofthe host metal bonding, this raises thequestion of how an ensemble ofpalladium nanoclusters will interactduring and after hydrogenation.A morphological or structural changeof the nanocluster ensemble mayaffect the hydrogenation propertiesas compared to a single cluster.

We investigated the effect of hydrogenexposure on a palladium clusterassembled film produced with a dual-target dual-laser vaporisation sourcewith three different techniques:extended X-ray absorption finestructure (EXAFS), X-ray diffraction(XRD) and scanning-tunnellingmicroscopy (STM). These threecomplementary methods were used todetermine the size changes of thepalladium nanoclusters upon exposureto hydrogen and oxygen. The averagegrain size was derived both from thecoordination number around palladiumatoms obtained from EXAFS (Figure 23)and the width of the XRD Bragg peaks,whereas the (lateral) diameter of thenanoclusters at the sample surface wasmeasured directly by STM. X-rayabsorption data (Figure 23) collected atthe DUBBLE beamline (BM26A) yieldedadditional parameters such as inter-atomic distances, degree of disorderand electronic information through theshift of the absorption edge; thesesupport the observation of sizechanges. Upon hydrogenation, theincrease in the nanocluster sizemeasured with XRD, EXAFS, and STM is22%, 38%, and 37%, respectively.This increase is much larger than theincrease of about 8.1% in the Pd phaseunit cell volume corresponding to thehydride formation. The cluster growthis due to an atomic reorganisation bythree-dimensional Ostwald ripening,in which the larger clusters take upmobile atoms at the expense of smallerclusters [1]. For most materials,spontaneous Ostwald ripening is anextremely slow process at roomtemperature. However, the presenceof a hydrogen atom in the metalreduces the binding energy, thusincreasing the probability of

Principal publication andauthors M. Di Vece (a), D. Grandjean (a),

M.J. Van Bael (a), C.P. Romero (a),

X. Wang (a), S. Decoster (b),

A. Vantomme (b) and

P. Lievens (a), Phys. Rev. Lett. 100,

236105 (2008).

(a) Laboratorium voor Vaste-

Stoffysica en Magnetisme &

INPAC - Institute for Nanoscale

Physics and Chemistry, K.U.

Leuven (Belgium)

(b) Instituut voor Kern- en

Stralingsfysica & INPAC - Institute

for Nanoscale Physics and

Chemistry, K.U. Leuven (Belgium)

carbonyl and is thus invisible ininfrared spectroscopy, is included inthe refinement. It would have beenvery difficult using X-ray scatteringalone to establish the existence of theminor intermediate Ru3(CO)11(µ-CO)because its contribution to the overallscattering curve is much smaller than

that of Ru3(CO)10. As its existence hasbeen unequivocally established byspectroscopy it is entirely justified toinclude it in the fit to the X-ray data.Together these techniques clearlyindicate the existence of at least threeintermediates with very differentmolecular structures (Figure 22).

References [1] E.A. Glascoe, M.F. Kling,

J.E. Shanoski and C.B. Harris,

Organomet. 25, 775 (2006).

[2] Q.Y. Kong, M. Wulff, J.H. Lee,

S. Bratos and H. Ihee, J. Am.

Chem. Soc. 129, 13584 (2007).

Fig. 23: Phase-corrected Fouriertransforms of the k3-weighted

Pd nanocluster EXAFS as a functionof the atomic distance (R).



Direct observation of 3D grain growthin Al-0.1%Mn

Principal publication andauthorsS. Schmidt (a), U.L. Olsen (a),

H.F. Poulsen (a), H.O. Sørensen (a),

E.M. Lauridsen (a),

L. Margulies (a,b), C. Maurice (c)

and D. Juul Jensen (a), Scripta

Mater. 59, 491 (2008).

(a) Center for Fundamental

Research: Metal Structures in Four

Dimensions, Risø National

Laboratory at the Technical

University of Denmark, Roskilde

(Denmark)

(b) ESRF

(c) Materials centre, Ecole des

Mines, Saint Etienne (France)

detachment of palladium atoms asshown schematically in Figure 24. Thisprocess can be qualitatively understoodin terms of the difference between thesublimation energy of palladium metaland palladium hydride. The sublimationenergy of palladium decreases withincreasing hydrogen concentration andis about 50% lower at a H/Pd ratio of0.3 [2]. The effect of hydrogen on thepalladium cluster sublimation has thesame result as if the temperature of theclusters was doubled. Since ourexperiment was performed at roomtemperature, a doubled temperaturewould correspond to approximately600 K (327ºC), a temperature at whichOstwald ripening would occur. Thusthe observed cluster size increase isattributed to Ostwald ripening inducedby exposure to hydrogen: the absorbedhydrogen decreases the sublimationenergy of palladium, which stimulatesthe detachment of atoms from theclusters at room temperature.

To conclude, the morphological andstructural changes occurring in anensemble of palladium nanoclustershave been studied after severalhydrogenation cycles with EXAFS, XRD,and STM. Initial hydrogenation

increased the cluster size, a result thatis attributed to hydrogen-inducedOstwald ripening. This phenomenonoriginates from the higher mobility ofpalladium atoms resulting from thelow sublimation energy of palladiumhydride as compared to that of thepalladium metal. The universality ofthis phenomenon makes it importantfor the application of futurenanostructured hydrogen storagematerials.

Fig. 24: A schematic impression of the palladium cluster building blocks. The palladiumatoms become free to move due to the interstitial hydrogen.

References[1] W. Ostwald, Z. Phys. Chem.

(Leipzig) 34, 495 (1900).

[2] N.V. Piskunov et al., J. Eng.

Phys. Thermophys. 74, 1217

(2001).

Metals and alloys are typicallypolycrystalline and are processed byplastic deformation and annealing.During annealing the so-called graingrowth process may occur, wherebysome of the grains in thepolycrystalline ensemble grow at theexpense of others, resulting in acoarsening of the microstructure.This has important implications forthe properties of the material; widelyknown applications are transformersteels and devices for power currentbased on Tc superconductorcompounds. Grain growth has beenstudied extensively for decades bothexperimentally - mostly by opticalmicroscopy - and theoretically, leadingto the proposition of a range of graingrowth models.

Here we present a 3D mapping studyof an Al-0.1%Mn alloy using the three-dimensional X-ray diffraction (3DXRD)microscope situated at the materialscience beamline ID11. As a functionof the treatments, the average grainvolume increases from 2.9 × 105 µm3 to4.2 × 106 µm3 and the number ofgrains within the illuminated part ofthe sample reduces from 483 to 32, seeFigure 25. The volume distribution ofthe initial grains is shown in Figure 26a.Marked in red are the 27 grains thatremained after annealing. As expectedthese are predominantly the largergrains. Notably three grains smallerthan the median size were alsoobserved to grow. As illustrated inFigure 26b the texture of both theinitial and the final distributions are


26

Materials Science

EE SS RR FF

Crystallisation and dissolution of flow-induced precursors

Easy processability of polymericmaterials is a parameter that is ofcrucial importance to promote theiruse in an increasing number of bothhigh-tech as well as commodityapplications. Production of polymericmaterials often involves shaping theproduct from the melt by applyingflow at high temperatures (thermo-mechanical history). The flow rate andtemperature can strongly influence the

final properties of the material. Forsemi-crystalline polymers, the linkbetween this thermo-mechanicalhistory and the properties is reflectedin the crystalline structure andmorphology.

The crystallisation of these entangled,i.e. spaghetti like, macromoleculesposes intriguing problems from afundamental scientific viewpoint but

AuthorsL. Balzano (a), G.W.M. Peters (a),

S. Rastogi (b), G. Portale (c),

L. Fernandez-Ballester (c) and

W. Bras (c).

(a) Eindhoven University of

Technology (The Netherlands)

(b) Loughborough University (UK)

(c) BM26, ESRF

close to random. Hence, within thestatistics of the study, no apparentcorrelation is found betweenorientation and the probability of

growth. To our knowledge, for thefirst time a bulk sample has beencompletely characterised before andafter a set of annealing treatments.

Fig. 25: a) 3D grain map after firstannealing for an Al-0.1%Mn alloy.

b) the same volume is shown after thefinal annealing. c) the four layers in the

middle of the volume are shown afterthe first annealing, top row, and final

annealing, bottom row, respectively. Theinterlayer spacing is 10 µm. Colours

symbolise crystallographic orientation.Black denote vacancies; internal vacancies

being an artefact of the program. Thepixel size in the maps is 5 µm x 5 µm.

Fig. 26: Characteristics of grains. a) Grain volume distribution before

annealing. Red colour: the sub-set ofgrains remaining after the annealing;

blue colour: all other grains. b) (111)-polefigure with respect to

sample reference system (x,y,z).Same colour scheme as for a).

obviously has implications for appliedmaterial technology as well.

It is well established that when a strongflow is applied to a polymer melt, themolecules orient and stretch with theflow. When the temperature issufficiently low, orientation and stretchof the molecules can be frozen-in withthe onset of crystallisation (Figure 27).Anisotropic (fibrillar) crystals areformed and, in the next stage,overgrown by disk-like lamellae. Thiscrystalline morphology is called shish-kebab because of the analogy with theMediterranean meat dish. The origin ofthis morphology is currently a topic ofdiscussion. It is well accepted that theinitial stages of the process are veryimportant and synchrotron scatteringtechniques, based on SAXS and WAXD,have proven indispensible for thesestudies. Experiments from our group atthe Technische Universiteit Eindhoven,performed at the Dutch Belgianbeamline BM26, as well as ID11 andID02, permitted an understanding ofthe very early stages of shish-kebabformation [1-4]. Our findings point tothe existence, prior to nucleation, offlow-induced precursors (FIPs) consistingof non-crystalline nano bundles ofstretched molecules. The key discoveryis that these FIPs can be formed attemperatures higher than the material’smelting point and represent the‘memory’ that the material has of itsthermo-mechanical history. In situobservation of FIPs in melts of mixturesof high density polyethylene (HDPE),with different molecular weights, at142°C, highlighted their fibrillarmorphology and their dynamics as afunction of the size. Large FIPs cancrystallise (become nuclei) and growinto shish-kebabs whereas small FIPsdissolve in the melt. The presence of acritical size for nucleation is inagreement with the classical theory ofnucleation. However, the same theorydoes not consider the formation ofnon-crystalline (and long lasting)structures prior to nucleation. Ourexperiments are, therefore, a bridgebetween the classical theory and newviews on (flow induced) nucleation ofmacromolecules.

The dissolution of small FIPs, for whichthe development stops in the veryearly stages, occurs on the same

timescale as the diffusion (reptation)of the longest molecules of the melt.This suggests that dissolution of thesesmall FIPs is strongly correlated withthe reptation of the longest moleculesalong the precursor backbone(Figure 28), i.e. the early FIPs are richwith the longest chains.

To summarise, in crystallisation ofentangled macromolecules likepolymers, morphology and structuredepend on the whole thermo-mechanical history [3,4]. The memoryof gradients of flow applied even athigh temperatures is stored into FIPs[2]. In the very early stages ofcrystallisation, FIPs can direct theprocess towards the formation ofhighly anisotropic crystals with shish-kebab type morphology. Theclarification of the basic mechanismsfor structure formation duringprocessing of macromolecules helpsthe exploitation of the properties ofexisting materials and could lead,eventually, to the benefit of materialswith properties tailored to theapplication.



Fig. 28: Hypotheticalstructure of a flow-induced precursors (FIP).This is a bundle made ofthe longest molecules ofthe melt that arestretched with theapplication of flow.

References[1] L. Balzano, N. Kukalyekar,

S. Rastogi, G.W.M. Peters,

J.C. Chadwick, Physical Review

Letters 100, 048302 (2008).

[2] L. Balzano, Flow induced

crystallization of polyolefins. PhD

thesis, Eindhoven University of

Technology (2008).

[3] L. Balzano, S. Rastogi,

G.W.M. Peters, Macromolecules

41, 399 (2008).

[4] L. Balzano, G.Portale,

S. Rastogi, G.W.M. Peters,

Macromolecules 41, 5350 (2008).

Fig. 27: Schematic representation of coiledmacromolecules that can be stretched with flowand crystallise into highly-anisotropic shish-kebabs.


28

Materials Science

EE SS RR FF

Morphological clues to wet granularpile stability

When dry sand is mixed with water, itacquires a visco-plastic texture, stiffenough for sculpting sand castles. Thewell-known experience at the beachthat the mechanical properties of thismaterial do not depend critically onthe liquid content is of great potentialimportance for understanding wetgranular systems in general, such asencountered in land slides, or in thepharmaceutical or food industry. Theinsignificance of the water content isparticularly unexpected if oneconsiders the enormous geometriccomplexity of the liquid structureswhich form inside the granular pile,and change dramatically with theliquid content.

The geometrical structure of the liquidwas analysed by means of X-raymicrotomography. Owing to the hightemporal resolution available atbeamline ID15A, inherent ageing

problems in the samples could becircumvented and high quality data ofhighly absorbing static and dynamicsamples were obtained. Figure 29ashows an image of a three-dimensional liquid cluster formed in apile of glass spheres. It turned out thatmost of the space between the spheresis not filled with liquid, but with air.This can be seen more clearly inFigure 29b, where the smallest liquidmorphologies are presented. For liquidcontents W < 0.02 only capillarybridges can be found in a wet pile.These morphologies are the buildingblocks for all larger morphologies, i.e.liquid clusters emerging for W > 0.02are built from capillary bridges (cb),trimers (tr), or filled tetrahedron gaps“glued together” at their edges.The surface to volume ratio of thoseextended liquid morphologies wasfound to be about constant for allobserved liquid contents.

A closer analysis shows that theinternal geometry of these intertwinedstructures of liquid, air, and the grainscan be well characterised by a fewparameters. In particular, it can bepredicted from the packing geometrythat the Laplace pressure withinequilibrated liquid clusters acquires auniversal value for liquid contents,above W ~ 0.02. It should only dependupon the surface tension of the liquid,its wetting angle, and the typical sizeand shape of the grains. That this is infact the case is shown in Figure 30,where the Laplace pressure, whichcould be inferred from thetomographic data, is shown as afunction of the liquid content. Overalmost the full range of data, it staysat the grey bar, which represents theuniversal value for the Laplacepressure as derived from the sampleparameters. The liquid morphologiesequilibrate after a typical time delayof 5 minutes for the beads used in ourexperiment, as displayed in the insetof Figure 30. Liquid morphologies witha smaller Laplace pressure will grow atthe cost of liquid morphologies withlarger Laplace pressure. In equilibriumany liquid morphology in the sample

Principal publications andauthorsM. Scheel (a), R. Seemann (a,b),

M. Brinkmann (a),

M. DiMichiel (c), A. Sheppard (d),

B. Breidenbach (e),

S. Herminghaus (a), Nature

Materials, 7, 189 (2008); J. Phys.:

Cond. Mat. 20, 494236 (2008).

(a) Max Planck Institute for

Dynamics and Self-Organization,

Göttingen (Germany)

(b) Experimental Physics, Saarland

University, Saarbrücken (Germany)

(c) ESRF

(d) Department of Applied

Mathematics, Australian National

University Canberra (Australia)

(e) Theoretical Physics Institute,

University Erlangen-Nürnberg,

Erlangen (Germany)

Fig. 29: a) X-ray microtomographyof a liquid cluster in a dense pile

of glass beads of 0.8 mmdiameter at a liquid content of W

≈ 0.11. W is defined with respectto the entire sample volume.

b) Smallest liquid morphologiesemerging in wet piles of glass

beads. Top row: as found by X-ray tomography, bottom row:

as calculated numerically. ForW < 0.02 only capillary bridges

(cb) can be found. For largerliquid contents the capillary

bridges might merge to formtrimers (tr), pentamers (pt), etc.

At low liquid content, filledtetrahedron gaps (th) can also be

found.

Fig. 30: The Laplace pressurewithin the liquid clusters, as a

function of the liquid content ofthe sample. The grey bar

indicates the universal pressurevalue calculated from the sampleparameters (surface tension and

size of spheres), its widthcorresponds to the uncertainty in

the wetting angle. The insetshows the equilibration of

individual liquid clusters afterpreparation.



will have the same Laplace pressure(see the grey bar in Figure 30).

This result entails a constant materialstiffness over the full range of data,even for small liquid content wherethe Laplace pressure deviates from thegrey bar. In this regime, the wettedarea varies as well, but in such a waythat the resulting cohesion forces

remain constant down to the lowestcontents investigated. As a result, thecohesion strength of a wet granularpile is largely independent of liquidcontent over the full rangeinvestigated. We could furthermorepresent evidence that this generalbehaviour is also valid for non-spherical granules, like real sandgrains.

Molecular layering of fluorinated ionicliquids at a charged sapphire (0001)surface

Room temperature ionic liquids (RTILs)are comprised of molecular ions.They are interesting materials from aphysical point of view as they allow usto modify the interactions betweenthe molecular ions as well theinteraction with other materials bysimply modifying the individual ionicmolecules chemically [1]. To dateseveral thousand ionic liquids havebeen synthesised and many of themare available from commercial sources.

Room temperature ionic liquids arealso very promising candidates for avariety of new technologicalapplications, such as solvents in greenchemistry, catalysis, batteries, and fueland solar cells. Although crucial forunderstanding their properties inapplications, very little is knownabout the structural arrangementof the anions and cations at solidinterfaces. In order to access theseinterfaces, high energy X-rayreflectivity (E = 72.5 keV) at theinstrument HEMD, for high-energymicrodiffraction, at beamlineID15A has been employed, whichis an ideal tool to reach deeplyburied interfaces within bulksamples. The measured X-rayreflectivity could then be usedto extract information on thelaterally-averaged electron-densityprofile at the RTIL/Al2O3 interface.

Systematic studies of the interfacialstructure as a function ofcomposition and temperature wereperformed with several RTILs at an

Al2O3 (0001) surface. By choosingdifferent combinations of anions andcations, it was possible to tune theion-ion and ion-substrate interaction.By changing the temperature, theratio between entropy and interfacialenergy could be varied, driving thesystem closer to a disordered liquid orinterfacial ordering, respectively.Figure 31 shows the normalised highenergy reflectivity for three differentRTILs containing the same anion [fap]–.All three measurements show distinctfeatures in the normalised reflectivitywhich could be attributed topronounced molecular layering at theRTIL/Al2O3 interface. Themeasurements are perfectly

Fig. 31: a) Sketch of themolecular ions: three differentcations have been combinedwith the [fap]- anion.b) Normalised high energy X-ray reflectivities of the threedifferent RTILs in contact witha Al2O3 (0001) surface.The lines are best fits using amodified distorted crystalmodel ([fap]–[bmpy]+: blue line,[fap]–[hmim]+: red line, [fap]–[tba]+: green line).

Principal publication andauthorsM. Mezger (a), H. Schröder (a),

H. Reichert (a), S. Schramm (a),

J.S. Okasinski (a), S. Schöder (a,b),

V. Honkimäki (b), M. Deutsch (c),

B.M. Ocko (d), J. Ralston (e),

M. Rohwerder (f),

M. Stratmann (f), H. Dosch (a,g),

Science 322, 424 (2008).

(a) Max-Planck-Institut für

Metallforschung, Stuttgart

(Germany)

(b) ESRF

(c) Physics Department, Bar-Ilan

University, Ramat-Gan (Israel)

(d) Condensed Matter Physics and

Materials Science Department,

Brookhaven National Laboratory,

Upton (USA)

(e) Ian Wark Research Institute,

University of South Australia,

Mawson Lakes (Australia)

(f) Max-Planck-Institut für

Eisenforschung, Düsseldorf

(Germany)

(g) Institut für Theoretische und

Angewandte Physik, Universität

Stuttgart (Germany)


30

Materials Science

EE SS RR FF

Super-soft element 11

The melting temperature of sodium isunique, being lower at very highpressures than at ambient pressure.In a recent experiment at beamlineID27, this unique feature of sodiumwas explorated in order to conduct thehighest-pressure single-crystalstructure refinements ever made. Thishas revealed that the low meltingtemperature of elemental sodium isassociated with a number of extremelycomplex new crystal structures.The complex phase diagram of this“simple” metal above 100 GPasuggests extraordinary states in boththe liquid and solid phases at extremeconditions and has implications forother “simple” metals.

For decades, a combination offundamental and applied physics hasbeen utilised to synthesise new

materials having intriguing andpotentially useful properties such assuper-hardness or superconductivity.When materials are compressed, theirbulk and shear moduli and theirstrength (all measures of howcompressible and therefore how hardmaterials are) as well as their meltingtemperatures all typically increase.For example, at a pressure of 55 GPa,the strength of elemental argonexceeds that of hardened steel underambient conditions [1], while itsmelting temperature above 130 GPais expected to exceed that of iron.The two lightest elements – hydrogenand helium – have extremely highcompressibility, and helium, owing toits chemical inertness, is commonlyused as a pressure transmittingmedium to create hydrostatic pressuresin high-pressure experiments.

Principal publication andauthors E. Gregoryanz (a), L. Lundegaard (a),

M. McMahon (a), C. Guillaume (a),

R. Nelmes (a), M. Mezouar (b),

Structural Diversity of Sodium,

Science 320, 1054 (2008).

(a) Centre for Science at Extreme

Conditions and School of Physics,

University of Edinburgh (UK)

(b) ESRF

reproduced by a modified distortedcrystal model. Figure 32 shows anexample, the reconstructed density ofthe interface [fap]–[bmpy]+/Al2O3(0001).The first layer at the interface iscomprised of cations, which may seemcounterintuitive at first sight, followedby a second layer of anions, thusensuring charge neutrality. Additionalmeasurements employing a Kelvinprobe confirmed that the substrate isnegatively charged, thereby explainingthe formation of a purely cationiclayer directly at the interface.The molecular layering decaysexponentially with a decay length ofapproximately two molecular doublelayers. Similar results have been foundfor the other two interfaces

[fap]–[hmim]+/Al2O3 and[fap]–[tba]+/Al2O3.

Dilute aqueous electrolytes containingwater-screened weakly-interacting ionshave been successfully described bymean field theories of charge doublelayers. This description neglects,however, correlations between ions. Incontrast, the interaction between theoppositely-charged RTIL ions is verystrong because no other molecule isavailable for screening. The ions are,therefore, strongly correlated. Suchcorrelations also lead to otherapparently counterintuitivephenomena like charge inversion andattraction between like-chargedobjects. Our findings are corroboratedby recent molecular dynamicssimulations of (simpler) RTILs in slitpores with Lennard-Jones potentialsfor the dispersion interactions of themolecules as well as for the wall-molecule interaction, where a clearseparation of the anions and cationsinto distinct layers was also observed[2]. Information on the structure ofsuch interfaces should help us todevelop a deeper understanding oftheir properties, such as their transportproperties, for future applications.

Fig. 32: Electron densityprofile of the interface[fap]–[bmpy]+/Al2O3 atT = –15°C: cation (red),anion (blue) and total

(black) electron densities.Red and blue lines indicatecation and anion Gaussiandistributions contributing

to the respective partialelectron density profiles

(solid lines); black line,total electron density

profile; grey bar, electrondensity of the sapphire

substrate withoutroughness.

References[1] P. Wasserscheid, T. Welton,

Ionic liquids in Synthesis (Wiley-

VCH, Weinheim Germany, 2007).

[2] C. Pinilla, et al., J. Phys. Chem.

B 111, 4877 (2007).



At pressures up to 25 GPa, the meltingtemperatures of hydrogen and heliumare the lowest of all elements and liebelow 500 K. However, their meltingtemperatures continue to increasemonotonically with pressure, reaching~1000 K at 100 GPa, while theirstrength at these pressures is such thatthey do not create even quasi-hydrostatic conditions.

Recently, it has been shown thatalthough the melting temperature ofsodium metal initially behaves“normally” and increases from 370 Kat ambient pressure to ~1000 K at30 GPa, it then decreases rapidly anddrops to just above room temperatureat pressures of 118 GPa, beforeincreasing again [2]. The meltingtemperature at 118 GPa is thus lowerthan at ambient pressure – behaviourthat is unique amongst all knownmaterials. The low meltingtemperature of sodium at 118 GPameans it is the only known “super-soft” system which can be used as atruly hydrostatic medium at pressuresabove 100 GPa. High-pressure single-crystal diffraction experiments werecarried out using the high intensityand micro-focused X-ray beamavailable at beamline ID27. Theminimum in the melting curve ofsodium at 118 GPa was shown to beassociated with a large number ofextremely complex crystal structures(Figure 33). Indeed, one of thesestructures contains more than 500atoms in the unit cell.

The low melting temperature ofsodium at 118 GPa means that it isrelatively easy to melt the sample insitu and the proximity of the meltingcurve ensures hydrostatic conditions.It was therefore possible to grow highquality single crystals at such highpressures by slow cooling from theliquid phase. A typical single-crystaldiffraction image collected underthese extreme conditions is presentedin Figure 34. The number of differentphases found in the vicinity of themelting minimum was surprising,seven in total. Very small changes inpressure and/or temperature resultedin transitions between the differentstructural forms. Several of thestructures were extremely complex –more so than anything else previously

observed for any element, even atambient conditions.

Using single-crystal methods, whichhave been developed as a three-yearlong term project at the ESRF, it hasbeen possible to determine the latticeparameters of all seven phases and,so far, to refine the atomic positionsin three of them – reaching the highestever pressure for which the full single-crystal structure refinement has beencarried out. The composite diffractionimage in Figure 34 showsrepresentative data from a primitivetrigonal structure with 90 atoms in theunit cell.

Early theoretical calculations onhydrogen have suggested that atextreme compressions it might have aliquid ground state with very unusualproperties and a family of the relatedanisotropic structures associated withits liquid state [3]. The results onsodium show something very similarbeing realised in nature, giving hopethat the theoretically-predicted bizarrestates of hydrogen do indeed existeven though they are currentlyunreachable in experimental studies.

Fig. 33: Observed phasesof sodium in the vicinityof the minimum of themelting curve.

Fig. 34: Compositediffraction image showingrepresentative data from theprimitive trigonal phasewith 90 atoms in the unitcell.

References [1] H. Mao et al., J. of Physics-

Condensed Matt. 18, S963 (2006).

[2] E. Gregoryanz et al., Phys. Rev.

Lett. 94, 185502 (2005).

[3] E. Brovman, Yu. Kagan,

A. Kholas, Soviet Physics JETP 34,

1300 (1972); Soviet Physics JETP

35, 783 (1972).


32

Materials Science

EE SS RR FF

Polyamorphic transition, charge-densitywaves and superconductivity in sulphur

Elemental sulphur, a yellowinsulating mineral, is composed ofcrown-like 8-member ring moleculesthat form a complex crystal structure.Compression has a dramatic effect onthese molecules: they break apartforming various chain and ringstructures on pressure increase [1]changing colour to red at around8 GPa. By compressing sulphur at lowtemperatures, where the phasetransitions are kinetically inhibited, wefind that the ambient pressure phaseof sulphur with S8 molecules persists ina metastable form to much higherpressures becoming amorphous above40 GPa. Sulphur becomes metallic andsuperconducting above 90 GPa withTc = 10 K [2] forming anincommensurately modulated crystalstructure at 300 K [3]. By studying thecrystal structure of metallic sulphur atlow temperatures close to thesuperconducting transition, we find acharge-density wave (CDW) instability,responsible for the structuralmodulation, to be in competitionwith the superconducting state.

Using high-pressure X-ray powderdiffraction techniques on beamlineID09A with a liquid He cryostat, wecompressed the S-I phase of sulphurat low temperatures observing atransition to an amorphous state (a-S)at pressures above 40 GPa. A change inthe diffraction pattern of a-S is seen at65 GPa and 80 K (Figure 35a),

signalling a transformation from oneamorphous form to another.Diffraction-based direct densitymeasurements give a 7% densitydiscontinuity (Figure 35b)corresponding to a transition from alow-density amorphous (LDA) to ahigh-density amorphous (HDA) form.The similarities between the densitiesand local structures of LDA and HDAsulphur with those of phases S-III(square chains) and S-IV(incommensurately modulated) arestriking, suggesting a nanocrystallinenature instead of truly amorphousform of a-S. Further compression ofthe high-density a-S resulted in itsrecrystallisation into phase S-IV at99 GPa and 40 K.

We observed sharp modulationreflections in the S-IV phase,characteristic of a long-range orderedCDW, and traced their positions andintensities as a function of pressureand temperature. From Rietveldrefinement of diffraction patterns(Figure 36a) at each P-T point weobtained the amplitude of the atomicdisplacements due to modulation inthe P range from 100 to 165 GPa andT range from 300 down to 15 K, closeto the superconducting transitiontemperature (Figure 36b). Theamplitude of the structuralmodulation (Figure 36a, inset)decreases with pressure increase,which correlates with the gradual

increase in Tc from 10 to14 K [2]. The modulationdisappears at 153 GPa,signalling the suppression ofthe CDW, leading to theenhancement of Tc from 14to 17 K at this pressure.

In conclusion, we haveshown that sulphur exhibitsa pressure-inducedamorphisation at lowtemperatures and pressuresabove 40 GPa, with a furthertransition from a low-densityto a high-density amorphousform at 65 GPa. Presentwork has extended the

Principal publications andauthorsC. Sanloup (a), E. Gregoryanz (b),

O. Degtyareva (b), and

M. Hanfland (c), Phys. Rev. Lett.

100, 075701 (2008);

O. Degtyareva (b),

M.V. Magnitskaya (d),

J. Kohanoff (e) G. Profeta (f)

S. Scandolo (g) M. Hanfland (c),

M. I. McMahon (b) and

E. Gregoryanz (b), Phys. Rev. Lett.

99, 155505 (2007).

(a) School of Geosciences and

Centre for Science at Extreme

Conditions, University of

Edinburgh (UK)

(b) SUPA, School of Physics and

Centre for Science at Extreme

Conditions, University of

Edinburgh (UK)

(c) ESRF

(d) Institute for High Pressure

Physics, Russian Academy of

Sciences, Troitsk (Russia)

(e) Atomistic Simulation Centre,

Queen’s University Belfast (UK)

(f) CNISM-Dipartimento di Fisica,

Universita` degli Studi di L’Aquila

(Italy)

(g) The Abdus Salam International

Centre for Theoretical Physics

(ICTP) and INFM/CNR

‘‘Democritos’’ National Simulation

Centre, Trieste (Italy)

Fig. 35: a) Structure factors of low-density amorphous form of sulphur

(49 and 54 GPa and 175 K, 65 GPaand 80 K) and high-density

amorphous form of sulphur (70, 91and 99 GPa and 40 K). b) Pressure

dependence of density of a-Scompared to crystalline phases.

pressure range of the direct densitymeasurements of amorphous materialsin the diamond anvil cells by a factorof 100. At a higher pressure of100 GPa, sulphur crystallises intoan incommensurately modulatedstructure, which is formed because ofa CDW instability. Sulphur now

represents a unique example of anelemental metal undergoing a CDWtransition within a superconductingstate, and joins a large group ofcomplex materials showing anenhancement of Tc upon pressure-induced suppression of a CDW.



Fig. 36: a) X-ray diffraction profileof S-IV and a Rietveld refinementon the basis of theincommensurately modulatedstructure. The inset shows thecrystal structure of S-IV.b) Observed maximum atomicdisplacement in S-IV and S-V asa function of pressure andtemperature, shown as opendiamond symbols. Triangles showTc from Ref. [2].

References[1] O. Degtyareva et al., J. Chem.

Phys. 126, 084503 (2007).

[2] E. Gregoryanz et al., Phys. Rev.

B 65, 064504 (2002).

[3] C. Hejny et al., Phys. Rev. B 71,

020101(R) (2005).

Short range magnetic collapse of ironunder high pressure at high temperatures

Knowledge of iron’s phase diagramunder high-pressure (HP) and high-temperature (HT) conditions is notonly of importance for geophysics butalso as a model system for thetheoretical understanding of 3delectronic properties. High-pressuremagnetism in particular is believed toplay a crucial role in the stability ofiron’s structural phases and elasticproperties. Theoretical improvementsbeyond the local-densityapproximation now make it possible toderive the correct magnetic groundstate at ambient conditions. However,puzzling results recently observedunder extreme conditions haveperturbed the situation anew.Superconductivity was found in Fealong with ferromagnetism at lowtemperature and high pressure [1] andhints of antiferromagnetic (AF)coupling in ε-Fe, the supposedlynonmagnetic hcp HP phase. Even morepoorly understood is the γ-Fe fcc phasestabilised at high temperature andhigh pressure. First-principlescalculations favour a noncollinearmagnetism in the form of spin spiraldensity waves [2] but results in the

γ phase are extremely scarce becauseof the HP/HT experimental constraints.Here, we demonstrate the collapse ofthe short-range iron magnetic state inthe HP/HT phases by X-ray emissionspectroscopy (XES), a local probe ofthe 3d spin magnetic moment.

The spin sensitivity of XES arises fromthe core-hole/3d exchange interactionin the final state which splits theemission spectrum into a main peakand satellite. We focus on theKβ emission line which shows thehighest sensitivity to the spin state in atransition metal. The spectra weremeasured at beamline ID27 at 33 keV,simultaneously to X-ray diffraction,using the Si(531) analyser from ID16.The sample was loaded in a diamond-anvil cell together with Al2O3 powderboth for thermal isolation andpressure-transmitting medium. HP/HTconditions were achieved in situ by adouble-sided laser heating techniquewhere an intense laser beam heats thesample inside the pressure cell.Figure 37 shows the evolution of theKβ emission spectra under highpressure at 300 K and 1400 K.

Principal publication andauthorsJ.-P. Rueff (a,b), M. Mezouar (c)

and M. Acet (d), Phys. Rev. B. 78,

100405(R) (2008).

(a) Synchrotron SOLEIL, Gif sur

Yvette (France)

(b) Laboratoire Chimie Physique -

Matière et Rayonnement, CNRS -

UPMC, Paris (France)

(c) ESRF

(d) Experimentalphysik,

Universität Duisburg-Essen,

Duisburg (Germany)


34

Materials Science

EE SS RR FF

In iron metal, the satellite is notablyweaker than that in other ironcompounds. Nevertheless, we clearlynotice a decrease in the satelliteintensity at high pressure whichsuggests a change in the iron spinstate.

The magnetic information containedin the XES spectra can be extracted bymeans of integrated absolutedifference (IAD), a quantity that islinearly related to the 3d spinmoment. Figure 38 shows the pressuredependence of the spin state from theIAD analysis. At 300 K, the results bearout the magnetic collapse when iron

enters the HP ε phase, as alreadyknown from XES and Mössbauerspectroscopy. Additionally, we observenon vanishing magnetism above thetransition that decays at higher P. Thisagrees with recent theoretical modelsuggesting that the magnetic momentis not completely suppressed in ε-Febut instead remains up to a fairly largepressure in presence of AF coupling.This is in contrast to the high-spin (HS)to low-spin (LS) transition usuallyevoked at the α-ε structural change.The behaviour in γ-Fe at 1400 K isremarkably similar to that at 300 K,suggesting that the mechanism of themagnetic transition in the γ phaseprimarily involves a HS to LS transition,besides the occurrence of a spin spiraldensity wave. Comparing our results tothe theoretical magnetic moment ofiron computed in the spin spiraldensity wave model, we find a fairagreement at low pressure, while athigh pressure a better match is foundwhen considering the low q limit ofthe spin spiral density wave structure.The latter corresponds to a magneticstate with saturated ferromagnetismand a clearly defined HS-to-LStransition.

XES therefore appears as a powerfulprobe of the 3d magnetic propertiesunder extreme conditions. In iron, theresults show hints of non-vanishingmagnetism in the ε-phase underpressure in agreement withpredictions of non-collinearmagnetism while in the γ phase, weobserve a LS collinear structure atHP/HT. This substantially minimisesthe role of spin spiral density waves inthe magnetic change at extremeconditions in the γ phase.

References[1] K. Shimizu et al., Nature

London 412, 316 (2001).

[2] S. Shallcross et al., Phys. Rev. B

73, 104443 (2006).

Fig. 37: Kβ emission spectra of iron under high pressure and temperature in theα, ε and γ phases: (a,c) full scale; (b,d) expanded intensity scale.

Fig. 38: Iron spin magnetic moment derived from the XES spectra (left scale,circles) and calculated (right scale, lines).



Hole-doping-induced superconductivityin NdFeAsO

Superconductivity at the surprisinglyhigh temperature of 55 K has beenrecently reported in fluorine-dopedrare-earth iron oxyarsenides,REFeAsO1-xFx (RE = rare earth) [1]. Themagnitude of Tc and the apparentsimilarities with the high-Tc cuprates –layered conducting FeAs slabs andproximity to antiferromagnetic andstructural instabilities – have madethese systems an intensely studiedresearch field. REFeAsO possesses atetragonal crystal structure comprisinglayers of edge-sharing FeAs4tetrahedra interleaved with REOlayers. On cooling, a structural phasetransition to orthorhombic crystalsymmetry, accompanied by thedevelopment of long rangeantiferromagnetic order occurs. Themagnetic instability is suppressed uponfluorine-doping before the onset ofsuperconductivity, while orthorhombicsymmetry survives well within thesuperconducting dome [2]. Themechanism of superconductivity inthese materials is still under debate.Inelastic X-ray measurements onNdFeAsO1-xFx on ID28 have shown thatelectron-phonon coupling could playa role in determining thesuperconducting state properties [3].The family of Fe-basedsuperconductors has now furtherexpanded to include, among others,the binary FeSe1-x phases [4].

While electron-doping in REFeAsO1-xFxhas been intensively studied, very littleis known about the structural andelectronic properties of hole-dopedphases (e.g. partial replacement of theRE ion by alkaline earths). Confirmingthe existence of superconductingphases upon hole doping, probinghow Tc varies with composition, andrevealing the influence of doping onthe structural and magnetic transitionsare of fundamental importance for theunderstanding of the pairingmechanism.

Recently, we have investigated thenew family of hole-doped systems,Nd1-xSrxFeAsO (x= 0.05, 0.1, 0.2).Inspection of the room temperature

diffraction profiles collected atbeamline ID31 reveals the tetragonalunit cell (P4/nmm) established beforefor REFeAsO (Figure 39). Rietveldanalysis shows that the a-axis increasesmonotonically with x as expectedconsidering the larger ionic radius ofSr2+, while the c-axis initially contractsat small x before undergoing aconsiderable expansion at x > 0.1.The c-axis dimensions are influenced bythe geometry of the AsFe4 pyramidalunits with the Fe-As-Fe angle, θ, firstincreasing for 0 < x < 0.1 and thendecreasing for x > 0.1 (Figure 39). Theelectronic structure of the REFeAsOsystems depends on small changes inthe geometry of the AsFe4 pyramidalunits which controls both the Fe near-and next-near-neighbour exchangeinteractions and the width of theelectronic conduction band. Thesudden increase in the compression ofthe AsFe4 pyramidal units coupled

Principal publication andauthorsK. Kasperkiewicz (a), J.W.G. Bos (a),

A.N. Fitch (b), K. Prassides (c),

S. Margadonna (a), Chem.

Commun. DOI:10.1039/B815830D

(2009).

(a) Department of Chemistry,

University of Edinburgh (UK)

(b) ESRF

(c) Department of Chemistry,

University of Durham (UK)

Fig. 39: Synchrotron X-ray (λ = 0.40301 Å)powder diffractionprofile ofNd0.8Sr0.2FeAsO at roomtemperature. Inset:evolution of the roomtemperature tetragonallattice constants as afunction of Sr2+ dopingand schematic diagramof the AsFe4 pyramidalunit.

Fig. 40: Electronic phasediagram of electron/holedoped NdFeAsO. Blue (red)squares mark thesuperconducting transitiontemperatures for electron(hole) doping.Inset: schematic diagram ofthe tetragonal structure ofSr2+ and F- doped NdFeAsOand temperaturedependence of theresistivity for Nd1-xSrxFeAsO [x = 0 (black),x = 0.05 (green), x = 0.2 (red)].


36

Materials Science

EE SS RR FF

Metal-peptide frameworksLike zeolites, the class of materialsknown as metal organic frameworks(MOFs) are both porous andcrystalline. However, while a zeoliteframework is purely inorganic and isrestricted to tetrahedral building units,the MOF coordination polymers areinorganic/organic hybrids and thepossibilities for varying the nature ofthe metal centres and of the organiclinkers to induce specific characteristicsin the resulting material areinnumerable. Interestingly, thepossibility of using peptides asconnectors has not been reported,although they offer a rich structuraldiversity and are intrinsically chiral.To explore their potential as linkers inthe fabrication of chiral metal peptideframeworks (MPFs), a series ofsyntheses using Cu2+ and Ca2+ as metal

centres and oligovalines aslinkers was performed.The combination of Cu2+ andthe Z-(L-Val)2-L-Glu(OH)-OHoligopeptide (Figure 41)produced the polycrystallinematerial “MPF-9”.

Powder diffraction methods had to beused for structure solution becausethere were no single crystals of theproduct. To this end, high-resolutionpowder diffraction data were collectedusing the diffractometer on BM01B(Swiss-Norwegian Beamline). Thepattern could be indexed with amonoclinic unit cell (a = 19.2610 Å,b = 4.8885 Å, c = 30.5096 Å,β = 92.524°) and the systematicabsences were consistent with thespace groups C2, C2/m or Cm. As anon-centrosymmetric structure wasexpected, C2 was assumed forstructure analysis.

The data proved to be of onlymodest resolution in reciprocal space(dmin of ≈ 1.4 Å), but considerablechemical information about thepeptide structure and the Cucoordination geometry was known, soa direct-space approach to structuresolution was attempted. For thispurpose, the direct-space global-optimisation algorithms implementedin the computer program FOX [1] were

Principal publication andauthorsA. Mantion (a), L. Massüger (b),

P. Rabu (c), C. Palivan (a),

L.B. McCusker (b),

A. Taubert (a,d), J. Am. Chem.

Soc., 130, 2517 (2008).

(a) Department of Chemistry,

University of Basel (Switzerland)

(b) Laboratory of Crystallography,

ETH Zurich (Switzerland)

(c) IPCMS, UMR 7504 CNRS -

Université Louis Pasteur

(Strasbourg)

(d) Institute of Chemistry,

University of Potsdam (Germany)

Fig. 41: The Z-(L-Val)2-L-Glu(OH)-OHoligopeptide.

with the larger lattice dimensions forthe x = 0.2 composition may lead to amuch smaller bandwidth and changesthe electronic behaviour as comparedto that at small x. Low temperaturediffraction profiles of all compositionsshowed the presence of a structuralphase transition from tetragonal toorthorhombic (space group Cmma) inanalogy with the undoped andelectron-doped REFeAsO systems.However, increasing the Sr2+ contentdoes not suppress the structuraltransition which occurs at almost thesame temperature for all compositions.

In agreement with the diffractionexperiments, resistivity (ρ)measurements show considerabledifferences with increasing x. Whilethe parent compound displays thetypical behaviour of REFeAsO systems,the resistivity curve for x = 0.05 revealsa drastic change with ρ increasing oncooling. It thus appears that at low x,the electron charge carriers inNdFeAsO are initially depleted pushing

the system through a transition frombad metal to semiconductor. Furtherhole doping increases the amount ofhole-type carriers, and metallicconductivity is restored. Indeed, thex = 0.2 system shows metallic-typebehaviour with ρ decreasing withtemperature and a superconductingtransition occurs at Tc =13.5 K(Figure 40).

Our experimental observations showthat hole-doping of NdFeAsO viapartial replacement of Nd3+ by Sr2+ isa successful route to inducesuperconductivity. However, thestructural and electronic response withdoping is different from and non-symmetric to that in the electron-doped side of the phase diagram(Figure 40). For NdFeAsO, higher levelsof hole doping are necessary toobserve a similar phenomenology tothe electron-doped systems as anincreased number of carriers isnecessary to overcome the initialsemiconducting behaviour.

References[1] Y. Kamihara, T. Watanabe,

M Hirano and H. Hosono, JACS

130, 3296 (2008).

[2] S. Margadonna, Y. Takabayashi,

M.T. McDonald, M. Brunelli,

G. Wu, R.H. Liu, X.H. Chen and

K. Prassides, Phys. Rev. B 79,

014503 (2009).

[3] M. Le Tacon, M. Krisch,

A. Bosak, J.-W.G. Bos, and

S. Margadonna, Phys. Rev. B 78,

140505 (2008).

[4] S. Margadonna,

Y. Takabayashi, M.T. McDonald,

K. Kasperkiewicz, Y. Mizuguchi,

Y. Takano, A.N. Fitch, E. Suard and

K. Prassides, Chem. Commun.,

5607 (2008).



used. A simplified structural modelwith limited conformational degreesof freedom was used as input, and themodel with the best fit to the datawas then taken as a starting point forstructure completion and Rietveldrefinement. For the latter, thecomplete pattern was used, and bonddistance and angle restraints wereincluded to keep the model chemicallysensible during the course of therefinement. The profile fit for the finalmodel (RF = 0.103, Rwp = 0.266,Rexp = 0.217) is shown in Figure 42.

The Cu2+ ions in the structure are eachcoordinated to two oxygen atoms ofthe peptide carboxylate groups and totwo ammonia molecules in a distortedsquare planar geometry (Figure 43).One of the coordinated NH3 moleculesforms a hydrogen bond with the L-glutamic acid residue of theneighbouring complex along the [001]direction, while the other formshydrogen bonds with two watermolecules, which form a helicalhydrogen bonding network along the[010] direction. The peptide ligandsthemselves also form hydrogen bondsto one another along the [010]direction to create a β-sheet-likestructure. This extended hydrogen-bonding network accounts for theinsolubility of MPF-9- in both water andorganic solvents, and the concomitantdifficulty in growing single crystals.

The structure is, within the limitsof the method, very well defined.It is also consistent with the resultsof the elemental analysis, TGA, EPRand magnetic measurements.

It can be described in terms of two 2-dimensional substructures. The firstof these consists of a double layer ofcopper ions in the ab plane, in whichthe two layers are diagonally displacedwith respect to one another, and thesecond is that of the peptide with thelong axis of the molecule orientedroughly along [001]. In addition to thehydrogen bonding network along the[010] direction, the phenyl groups ofneighbouring peptides along [001]form columns along [010] and areoriented to allow π-π interactions.

Although MPF-9 does not have anextended 3-dimensional Cu-peptidenetwork, magnetic interactions werefound to occur between magneticcentres; the structure does have someporosity and the peptide has beenshown to be a viable ligand meritingfurther investigation.

Fig. 42: Observed (top), calculated (middle) anddifference (bottom) profiles for the Rietveldrefinement of MPF-9. The first peak has been cutat half height to show more detail. λ = 0.50007 Å.

Fig. 43: Projections ofthe structure of MPF-9along [010] (left) andalong [110] (right)showing the H-bondingnetwork.

Reference[1] V.R. Favre-Nicolin, R. Cerny,

J. Appl. Crystallogr. 35, 734-743

(2002).

Metal-organic nanoporous structures asphotoluminescent chemical sensors

The field of multifunctional materialshas seen rapid progress with thediscovery of structures which combinea set of well-defined properties forspecific applications within the samecrystal. Properties such asnanoporosity, luminescence, andmagnetism are required for sensors,lasers, nonlinear optics, displays, andelectroluminescent devices.

Lanthanide-based porous metal-organic frameworks (Ln-MOFs) canpossess some of these properties andare therefore excellent candidates forsuch multifunctional materials forsensors. However, the pore sizes ofmost known Ln-MOF materials are toosmall to allow the diffusion ofmolecules of interest, precluding theiruse in photoluminescence sensors.

Principal publication andauthorsB.V. Harbuzaru (a), A. Corma (a),

F. Rey (a), P. Atienzar (a),

J.L. Jordá (a), H. García (a),

D. Ananias (b), L.D. Carlos (b),

J. Rocha (b), Angewandte Chemie

International Edition, 47, 1080

(2008).

(a) Instituto de Tecnología

Química, (UPV-CSIC), Valencia

(Spain)

(b) Departments of Chemistry and

Physics, CICECO, University of

Aveiro (Portugal)


38

Materials Science

EE SS RR FF

Reference[1] F. Gándara, A. de Andrés,

B. Gómez-Lor, E. Gutiérrez-

Puebla, M. Iglesias, M.A. Monge,

D.M. Proserpio, N. Snejko, Crystal

Growth & Design, 8, 378 (2008).

Fig. 44: Crystal structure ofITQMOF-2 (C grey,

F yellow: H white, O red,Eu orange). Upper-left inset:

Eu3+-containing ITQMOF-2 crystalsunder UV light.

Recently, we designed andsynthesised two novel

crystalline MOFscontaining rare-earth

cations that combinemicroporousstructure,hydrophobicity, highadsorption capacity,high thermalresistance,anisotropic

photoluminescenceand magnetic

properties.

The key design element ofthe MOFs was the inclusion of the

organic spacer, 4,4′-(hexafluoroisopropylidene)-bis(benzoicacid) (HFIPBB). This molecule has highhydrophobicity because of thepresence of aromatic rings and -CF3groups, an antenna effect from itsaromatic groups, strong interactionwith the metal centre through thecarboxylic groups, and twocoordination centres that are requiredfor the microporous 3D structure. Thechosen metal atoms were any singlelanthanide (except radioactivepromethium), or mixtures of them.Two new materials, named “ITQMOF-1” and “ITQMOF-2”, wereobtained with similar composition butdifferent crystal structures.

The structure of ITQMOF-1 could notbe solved due to severe twinning,although it has been recently describedby another Spanish group [1]; thestructure of ITQMOF-2 was solvedusing single crystal data collected atthe beamline BM16 (Figure 44).

Both materials are insoluble in waterand in common organic solvents, andhave excellent thermal stability;powder XRD of samples heated in situreveals that ITQMOF-1 is stable in air attemperatures above 400ºC, anddecomposes at ca. 450ºC.

To demonstrate the potentialapplications of these materials insensor devices, Eu3+-containingITQMOF-1 was used for detecting thepresence of ethanol in air. This wasdone by monitoring the emission at619 nm under alternating streams ofair saturated with ethanol and

ethanol-free air. In the presence ofethanol, we observed a rapid decreaseof the emission intensity, whichrecovers rapidly when the material isexposed to clean air. This behaviourmay be explained by considering thatethanol coordinates the Ln3+ ions,quenching the photoemission throughcoupling with the vibrational states ofthe O-H oscillators. The addition ofwater to both streams did not modifythese results, confirming the interest ofthe hydrophobic properties of thematerials for measuring ethanol in awater-containing air stream.

Importantly, it is possible to obtainmaterials with different emissioncolours and magnetic properties bycarefully selecting the Ln3+ ions. Forexample, Tb3+-ITQMOF-1 presents aferromagnetic interaction between themetal ions and a strong green emissionin UV light, while Eu3+-ITQMOF-2exhibits a red colour in UV light(Figure 44, inset) with paramagneticinteractions. Moreover, the (5% Eu3+

- 95% Gd3+)-ITQMOF-1 sample showsan antiferromagnetic interactionbetween the Gd3+ ions, and a redcolour in UV light due to the Eu3+ ions.Remarkably, this material presents botha 5D0 quantum efficiency q = 0.83 (toour knowledge, the largest reportedfor solid-state Eu3+ compounds withorganic ligands) and the largest valueof absolute emission quantum yield(η = 0.48, excitation wavelength of290 nm) reported for Eu3+-based MOFs.Furthermore, this emission is highlyanisotropic, depending on theorientation of the crystals relative tothe incident beam and the detector.

In conclusion, we show that ajudicious choice of the ligand (HFIPBB)allowed the design of novellanthanide-based metal-organicframeworks with high thermal stabilityand interesting multifunctionalproperties (magnetism, luminescence,microporosity and hydrophobicity).The combination of photoluminescentLn3+ centres and HFIPBB resulted inmaterials which efficiently senseethanol, even in the presence ofwater and under ambient conditions.Current studies are now focused onexploring the combination of opticaland magnetic properties for sensorapplications and data storage.



Introduction

Soft Matter studies seek to unravel thelink between a material’s microstructureand dynamics, and its macroscopicproperties. This chapter presents abroad set of examples from themorphology of synthetic polymer fibresto structural dynamics of biomoleculesin bulk and at interfaces. These studiesinvolve a diverse class of systems fromsupercooled liquids to highly self-assembled biological complexes. Thisdiversity is one of the particularstrengths of the activities at the SoftCondensed Matter Group beamlines,articulating significant overlap withother disciplines such as materialsscience, biology, and surface science.

The past year saw many importantscientific and technical developments. Ithas been a very hectic year withpreparation for the ESRF UpgradeProgramme and joint review of all fourbeamlines in the Soft Condensed MatterGroup (ID02, ID10A, ID10B, and ID13) inspring 2008. The outcome of this reviewwas very positive, without exception.The review panels recognised theoutstanding technical status andscientific accomplishments of thesebeamlines. The review panels alsoapproved the upgrade proposals for thebeamlines.

The highlights selected for this chapterrepresent only a sub-set of the manyinteresting scientific results andtechnical developments over the year.A large fraction of the articlesillustrate the unique feature ofscattering techniques in terms of thecombined access to the relevant lengthand time scales not readily feasible byother techniques. Studies of dynamicprocesses especially at interfaces arereceiving increased attention.

The first two articles deal withadsorption of biomolecules at curvedand planar interfaces. In the first article,the adsorption of proteins ontotethered polyelectrolyte brushes wasprobed in real time by SAXS, whichrevealed the sub-diffusive motions ofproteins in the grafted polyelectrolytelayer. The second article presents thestructural details of the adsorbedmonolayer of DNA on a solid substratewith a soft functionalised organic layer,elucidating their two-dimensional orderby X-ray reflectivity. This is followed by astudy of low frequency viscoelasticbehaviour of a supercooled liquid nearthe glass transition probed by XPCS atgrazing incidence which demonstratedstriking similarities in the temperaturedependence of viscosity and elasticity.The fourth article reports diffusescattering and reflectivity investigationsof the formation and self-organisationof gold nanoparticles at liquid-liquidand air-water interfaces illustrating anunexpected clustering of particles.

Soft Condensed Matter Groupbeamlines are involved in a variety oftechnical developments. The firstarticle in this category demonstratesthat hard X-ray holographic diffractionimaging is an excellent technique fordetermining electron density profiles ofnanostructures. With the availability ofhigher coherent flux, this techniqueoffers the possibility of imaging fastdynamic processes. In conventionalfibre diffraction experiments, the fibreaxis is perpendicular to the X-ray beam.With special specimen preparation,the possibility of passing themicrobeam along the fibre axis hasbeen demonstrated, which then revealsfine details of skin-core morphologywithout requiring elaborate modelling.Manipulating micrometre-size objectsin solution within a microbeam is achallenging task. In the subsequentarticle, the effectiveness of laser traps

Soft Condensed Matter


40


EE SS RR FF

for manipulating large multilamellarliposomes in a microdiffractionexperiment has been illustrated.

The final two articles present structuraldynamics of proteins when they performtheir biological functions. In the firstarticle, the quaternary transition ofhuman haemoglobin from ligand boundrelaxed state to unbound tense stateand back to the relaxed state wasstudied by nanosecond pump-probetechniques at beamline ID09B. In the lastarticle, the molecular mechanism ofmuscle braking action under rapidstretch was revealed by X-rayinterference combined with high-resolution muscle mechanics.

The joint review process of thebeamlines also set the momentum forthe Partnership for Soft CondensedMatter (PSCM) together with the ILL.The ESRF Science Advisory Committeesupported the establishment of thePSCM on a step-by-step basis, initially asa support laboratory for ancillaryequipments. As a first step, the supportlaboratories within the SCM group arebecoming better streamlined. Thescientific scope and content for thepartnership is being refined. Furtherdiscussion with the User Community isforeseen in the near future.

T. Narayanan

Directed motion of proteins alongtethered polyelectrolytes

The adsorption and immobilisation ofproteins from aqueous solutions ontosolid surfaces is a topic of considerableinterest in both soft matter andbiochemical research [1]. Manybiologically important processes areinterfacial in nature and very often,charged and uncharged polymersattached to surfaces are used to tunethe interaction with the proteins.Various biotechnological processesrequire immobilisation of enzymeswith full retention of their biologicalactivity. Reversible adsorption ofproteins onto charged surfaces is usedin protein purification by ion exchangechromatography and in many naturalprocesses such as cell adhesion.Nevertheless, unspecific adsorption ofproteins must be suppressed in manypractical applications in order toprevent foreign body reactions tomaterials from a host. Little is knownabout the kinetics of proteinadsorption and the self-organisationof biomolecules with tethered polymerchains at a molecular level.

Here we report on the first study ofthe motion of proteins (bovine serumalbumin, BSA) in a layer of tetheredpolyacrylic acid chains on colloidalpolystyrene particles (Figure 45) usingtime-resolved small-angle X-rayscattering (TR-SAXS) combined withstopped-flow rapid mixing at beamlineID02. The pH value of the aqueoussolution was kept well above theisoelectric point of the chargedproteins and the ionic strength in thesolution was adjusted to 7 mM toensure full uptake of proteins by thespherical polyelectolyte brushes (SPB).The driving force is the counterionrelease mechanism, i.e., the exchangeof confined counterions with chargedproteins. Figure 46a displays thetypical evolution of the scatteringintensity, I(q), as a function of time, t.The gradual uptake of protein can befollowed directly from the shift of theside maxima to smaller scatteringvectors q with increasing time (see thedotted vertical line). The continuouslines indicate the calculated intensitiesusing the radial electron densityprofile of SPB loaded with BSA and thescreened Coulomb interactionbetween the SPB obtained from areference interaction site integralequation theory. The TR-SAXS datarevealed that the uptake of the

Principal publication andauthorsK. Henzler (a), S. Rosenfeldt (a),

A. Wittemann (a), L. Harnau (b),

S. Finet (c), T. Narayanan (c), and

M. Ballauff (a), Phys. Rev. Lett.

100, 158301 (2008).

(a) Physikalische Chemie I,

University of Bayreuth (Germany)

(b) MPI Stuttgart and ITAP,

University of Stuttgart (Germany)

(c) ESRF

Fig. 45: Sketch of the uptake ofproteins (red spheres) by spherical

brushes consisting of a solid core(grey sphere) and grafted

polyelectrolyte chains (wavy lines).



proteins into the brush layer isrelatively fast and the final state isreached within 3 seconds. However,the time needed by a single proteinmolecule to diffuse freely through thetethered layer is less than amillisecond. This means that themotion of the proteins in the brushlayer has been slowed significantly.From the radial electron densityprofiles, the total amount of adsorbedprotein per SPB (τads) at a given timecan be derived. Figure 46b depicts thatτads increases with time as t1/4. Forunhindered diffusive motion ofproteins, the mean-square-displacement is expected to increasewith time as t1/2. The observed sub-

diffusive behaviour is due to a balanceof frictional and external forces actingon the proteins as shown schematicallyin the inset of Figure 46b.

In summary, we have demonstratedthat protein adsorption can bemonitored directly with high temporaland spatial resolution on surface-modified colloidal spheres using TR-SAXS. The uptake of chargedproteins by spherical polyelectrolytebrushes follows a sub-diffusivebehaviour. A similar directed motionof proteins and other biomoleculesthrough biological membranes couldplay a pivotal role in their self-assembly.

Fig. 46: a) TR-SAXS intensities, I(q), as afunction of time, t. For clarity, successiveupper curves have been shifted by a factor100 (as indicated in the legend). Thecontinuous lines represent the model. b) Thetime dependence of the amount of adsorbedprotein obtained from the radial electrondensity profiles derived from the fits in (a).The inset illustrates the directed motion ofproteins through the brush layer.

Reference[1] Proteins at Solid-Liquid

Interfaces, P. Dejardin (Ed),

Springer (2006).

DNA adsorption at liquid/solidinterfaces studied by X-ray reflectivity

DNA interactions with surfaces can berepulsive, leading to confinement ordepletion, or attractive. A repulsiveinteraction exists between negativelycharged phosphates of DNA andphospholipids of membranes andcontributes to enclose the geneticmaterial within the cell. Depletionforces also confine DNA within viralcapsids. Repulsive interactions are acommon requirement in micro- andnanofluidics. Attractive interactionslead to adsorption. Chemisorption is atthe core of solid-phase synthesis ofpolynucleotides. Physisorption is alsoof wide technological interest andbiological relevance. In hybridisationtechniques such as Southern blots,northern blots, or biochips, nucleicacids are immobilised on surfaces,either through chemisorption or

physisorption. Chromatographictechniques rely on a reversibleimmobilisation of DNA. It is thus ofgreat importance to understand thedetails of DNA interactions withinterfaces, liquid/solid, liquid/liquid [1]or vapor/liquid [2].

Few experimental techniques can beemployed to study the conformationof biological macromolecules on amolecular length-scale. Atomic forcemicroscopy, for example, can be carriedout within a liquid solution and hasyielded impressive atomic-resolutionimages of strongly adsorbed double-stranded DNA molecules on a solidsubstrate. Here we performed X-rayreflectivity studies at the aqueousbuffer/solid substrate interface todetect the adsorption of a

Principal publications andauthorsC. Douarche (a,b,c), R. Cortès (a),

S.J. Roser (d), J.-L. Sikorav (e) and

A. Braslau (f), J. Phys. Chem. B

112, 13676–13679 (2008);

C. Douarche (a,b), R. Cortès (a),

C. Henry de Villeneuve (a),

S.J. Roser (c), and A. Braslau (e),

J. Chem. Phys. 128, 225108 (2008).

(a) Physique de la Matière

Condensée, École Polytechnique,

CNRS, Palaiseau (France)

(b) Institut de Recherche

Interdisciplinaire, Villeneuve

d’Ascq (France)

(c) Service de Biologie Intégrative

et de Génétique Moléculaire,

CEA/Saclay, Gif-sur-Yvette (France)

(d) Department of Chemistry,

University of Bath (UK)

(e) Institut de Physique

Théorique, CNRS URA 2306,


(f) Service de Physique de l’État

Condensé, CNRS URA 2464,



42


EE SS RR FF

monomolecular layer of DNAmolecules onto (and into) a soft,organic functionalisation layer,positively charged, covalently graftedon an oxide-free, atomically flat,single-crystal silicon wafer.Experiments were carried-out onbeamline ID10B using 22.2 keV X-raysto traverse a 30 mm water path-length. The originality of the methodwas to characterise the graftedsubstrates at two different densitycontrasts (Figure 47): measured firstunder air, then under buffer solution,and then to analyse the dataconjointly, as is often the practice forneutron scattering. This severelyconstrained the model parametersyielding a simple, robust measurementof the soft structure. Thus, the verysubtle change observed in situ upon

adsorption of the somewhatdisordered monomolecular layer ofDNA could be reliably interpreted interms of adsorbed mass and thereforequantified. The high density of thislayer then implies that the thin layerof monodisperse DNA double-strandedchains must exhibit some 2D nematicordering (Figure 48), although this isnot observable directly in theexperiment.

This interpretation of the results goesagainst common beliefs about strongadsorption of macromolecules. Thisadsorption can be irreversible, yet themolecules still retain some mobility(rolling or sliding) to allowrearrangements and compactionthrough annealing. Even stronglyadsorbed on a solid surface, DNAmolecules remain accessible tomolecular interactions with enzymes,for example, as can be shown throughsome elegant biochemical experiments[3]. The tethering of single moleculesextending into the free solution istherefore not strictly necessary forbiotechnological applications and thisopens possibilities for the conceptionof new hybridisation biochips, inparticular, with aims of takingadvantage of electrical coupling withthe oxide-free semiconductorsubstrates. Enhanced efficiency of thehybridisation reaction can also begained through a reduction ofdimensionality [4] on adsorption of thereactants to a surface.

Fig. 47: Synchrotron X-rayreflectivity a) measurements

and b) density profile model.

Fig. 48: Molecular models of the organisation of the adsorbed DNA layer a) full embedding, b) intermediate and c) full coverage.

References[1] A. Goldar and J.-L. Sikorav,

Eur. Phys. J. E 14, 211 (2004).

[2] C. Douarche, J.-L. Sikorav, and

A. Goldar, Biophys. J. 94, 134

(2008).

[3] D. Rhodes and A. Klug, Nature

(London) 286, 573 (1980).

[4] G. Adam and M. Delbrück,

Reduction of Dimensionality in

Biological Diffusion Processes. In

Structural Chemistry and

Molecular Biology; A. Rich and

N. Davidson, Eds.; W. H. Freeman

and Company: San Francisco, 198

(1968).



Elastic behaviour of a supercooled liquid

It is well known that liquids flowreadily but at the same time theyexhibit elastic response under aninstantaneous deformation. Suchbehaviour can be described by theMaxwell-Debye (MD) viscoelasticmodel: After a sudden shear, thestress decays exponentially to zerowith a relaxation time given byτMD = η/G(∞), where η is the dynamicviscosity and G(∞) is the instantaneousshear modulus [1]. Thus, on timescalesmuch shorter than τMD the systemresponds elastically - like a solid -while the usual fluid behaviour isretrieved on longer timescales. TheMD picture is also useful in describingthe transition from a supercooledviscous liquid to a glass. For a glass-forming liquid in the supercooledstate the viscosity increases stronglyas the temperature decreases.Hence, τMD increases several ordersof magnitude and finally exceeds100–1000 s at the glass transitiontemperature (Tg) resulting inpredominantly solid-like response,except on very long timescales.

Contrary to the MD picture, adifferent and intriguing solid-likebehaviour was reported in the low-frequency regime for a variety ofsimple liquids and polymer melts.For instance, an elastic plateau in thestorage modulus, the real part of theshear modulus G(ω), was detected byrheometry [2]. Soft, solid-likebehaviour can be modelled by theVoigt-Kelvin (VK) approach where thematerials display elastic response toapplied stresses at low frequencies,but start flowing at higherfrequencies and/or stresses.Such behaviour is often observed incomplex fluids, e.g. gels, emulsions,and foams. Typically, these fluidsconsist of two or more separatephases and hence their structuralorganisation differs from that ofsimple liquids and polymer melts.

To study the response of asupercooled liquid close to Tg, weused grazing incidence X-ray photoncorrelation spectroscopy (XPCS) [3]from the surface. Liquid surfacesundergo continuous fluctuations due

to the presence of thermally-excitedcapillary waves and their dynamics isgoverned by the surface tension andby bulk properties like the viscosityand the shear modulus. Grazingincidence XPCS was performed toprobe the surface dynamics of thefragile glassformer polypropyleneglycol with a molecular weight of4 kDa (PPG 4000, Tg ~ 200 K).The measured dispersion relations(Figure 49) of the surface fluctuationwere extracted by fitting the intensitycorrelation functions (inset, Figure 49)at various temperatures andmomentum transfers (q). Obviously,the dispersion curves are not crossingthe origin so they deviate frompredictions of both the classicalcapillary wave theory and the MDmodel. The data can only bemodelled if elements from the VKmodel are employed accounting forthe presence of low-frequencyelasticity. A novel model combining

Principal publication andauthorsY. Chushkin, C. Caronna, and

A. Madsen, Europhys. Lett. 83,

36001 (2008).

ESRF

Fig. 50: The circles showthe temperaturedependence of theviscosity (η) and a fit withthe VTF model isindicated (solid blue line).The red squares show theelasticity (E) extractedfrom XPCS and the blackcrosses indicate Ededuced from staticscattering data. Thedashed red line is a guidefor the eye following theVTF behaviour.

Fig. 49: Low-temperaturedispersion curves. Thelines are fits with themodel. The inset displaysnormalised correlationfunctions (symbols) andfits (lines) atq|| = 7.55 x 10-5 Å-1.


44


EE SS RR FF

MD and VK viscoelasticity wasdeveloped and hence the viscosityand elasticity could be extracted as a

function of temperature (Figure 50).As expected, the viscosity follows theVogel-Tamman-Fulcher (VTF)expression and surprisingly theelasticity exhibits similar behaviour(Figure 50). This suggests a connectionbetween low-frequency elasticity anddynamical heterogeneity with densityfluctuations in the supercooled statethat appear frozen at the time scaleof the experiment.

Formation and ordering of goldnanoparticles at toluene-water and air-water interfaces

Gold nanoparticles with sizes around2 nm and below are being studiedextensively because they exhibit novelelectronic, magnetic, optical andcatalytic properties. We areinvestigating the formation of Au-55types of gold nanoparticles through areduction reaction at a liquid-liquidinterface. Diffuse X-ray scattering andreflectivity techniques [1] allow us tomonitor the formation of goldnanoparticles at liquid-liquid interface.These techniques also provideinformation regarding the out-of-plane and in-plane structure of thegold nanoparticles at liquid surfaceand liquid-liquid interface.

In situ X-ray scattering measurementswere carried out at beamline ID10Bduring the formation of gold

nanoparticles at a toluene-waterinterface. The reduction reaction [2]uses an organic precursor of gold keptin the toluene layer and a reducingagent kept in the slightly alkalinewater layer. As the reaction occursthrough the creation of fingers fromone liquid to another, the size (1 nmdiameter) of the nanoparticles remainmonodisperse. The reflectivity datacollected as the reaction progresses areshown in Figure 51a and Figure 51bwith the extracted electron densityprofiles across the interface. The peaksin the diffuse data shown inFigure 51d and the electron densityprofiles obtained from reflectivityclearly show the formation of a 1.2 nmnanoparticles core with a 1 nm organicshell which reorganize amongstthemselves to give 13 member clusterswith nanoparticle-nanoparticleseparation, d2 ~ 3 nm (Figure 51c), andcluster-cluster separation, d1 ~ 18 nm(Figure 51e). Formation of organiclayers at toluene-water interface couldbe monitored by measuring interfacialtension that reduces by an order ofmagnitude through diffuse scatteringarising from capillary waves. This insitu high-energy (required to gothrough the upper liquid) X-ray

Principal publications andauthorsM.K. Sanyal (a,b),

V.V. Agrawal (b), M.K. Bera (a,c),

K.P. Kalyanikutty (b),

J. Daillant (d), C. Blot (d),

S. Kubowicz (d), O. Konovalov (e),

C.N.R. Rao (b), J. Phys. Chem. C

112, 1739 (2008); M.K. Bera (a,c),

M.K. Sanyal (a), S. Pal (a),

J. Daillant (d), A. Datta (a),

G.U. Kulkarni (b), D. Luzet (d),

O. Konovalov (e), Europhys. Lett.

78, 56003 (2007).

(a) Saha Institute of Nuclear

Physics (India)

(b) Jawaharlal Nehru Centre for

Advanced Scientific Research

(India)

(c) S.N. Bose National Centre for

Basic Sciences (India)

(d) LIONS, CEA (France)

(e) ESRF

References[1] J.D. Ferry, Viscoelastic Properties of Polymers (Wiley & Sons. Inc. 1980).

[2] H. Mendil, P. Baroni and L. Noirez, Eur. Phys. J. E 19, 77 (2006).

[3] G. Grübel, A. Madsen, A. Robert, X-ray Photon Correlation Spectroscopy in

Soft Matter Characterization, Eds. R. Borsali and R. Pecora, pp 953-996. Springer (2008).

Fig. 51: a) X-ray reflectivity data and b) theelectron density profile obtained by fitting. c) Cross-sectional view of the interfaceshowing the three layers of nanoparticles.d) GID data and e) 3D view of the 13 membercluster-model.



scattering technique developed herewill provide us with uniqueinformation regarding asymmetricinterfacial reactions important notonly for formation of tinynanoparticles but also for severalbiological reaction mechanism. Thepublished result is one of the mostviewed papers in American ChemicalSociety journals.

We are also investigating methods toimprove the packing density ofnanoparticles on substrates bytransferring a Langmuir monolayerfrom a water surface. Hydrophobicthiol-capped gold nanoparticles werespread on a water surface and theordering of these particles was studiedby X-ray scattering techniques as afunction of in-plane pressure andtemperature. The primary purpose ofthis study is to understand thestructure and morphology of amonolayer of thiol-capped 2 nm goldnanoparticles on a water surface overthe large reversible region of isotherm(1–15 mN/m) and to develop a methodto improve coverage of thesenanoparticles in deposited monolayerfilms. Figure 52 shows (q||-qz) contourplots of the grazing incidencediffraction (GID) data collected fromthe monolayer at different surfacepressures and temperatures.Figure 52 (a,b,c) show the bending ofBragg-rod at qz = 0.19 Å-1 with highersurface pressure (13 mN/m). The rodstraightens after decompression(1 mN/m) at 10°C. The Bragg-rod is

also straightened (Figure 52 (d,e)) byannealing the monolayer at 44°C. Thedetailed analysis of the data showsthat a monolayer of gold nanoparticlesexhibits reversible buckling under acompression-decompression cycle andcan be annealed thermally. Weannealed a monolayer film at 13 mN/mfrom 16°C to 33°C and found that theannealed monolayer provides muchmore compact structures in thetransferred films on a substrate.Atomic force microscopymeasurements of these deposited filmsare shown in Figure 52 (f,g).

References[1] M.K. Sanyal, S.K. Sinha,

K.G. Huang, B.M. Ocko, Phys. Rev.

Lett. 66, 628 (1991).

[2] C.N.R. Rao, G.U. Kulkarni,

P.J. Thomas, V.V. Agrawal,

P. Saravanan, J. Phys. Chem. B

107, 7391 (2003).

Fig. 52: The (q||-qz) contour plots ofintensity at 10°C with surfacepressure a) 1 mN/m, b) 13 mN/mand c) 1 mN/m afterdecompression. Scattered intensityat d) 10°C and e) 44°C w th 13mN/m. AFM images of filmsdeposited with 13 mN/m at f) 16°Cand g) 33°C.

Holographic diffraction imaging withhard X-rays

Nanoscience needs the ability tocharacterise the structure of objects onthe nanoscale. For several reasons,hard X-rays are very attractive for thispurpose: a wavelength in the order of≈ 1 Å allows excellent spatialresolution; the high penetration intomatter permits the study of thicksamples or even buried structures in aneasy-to-realise experimental setup; andwith the upcoming hard X-ray free-electron lasers, the achievable timeresolution will reach the femtosecond

(fs) regime. Here, we demonstrate howcoherent hard X-rays can be used for adetermination of the absolute electrondensity of a lithographically tailoredgold nanostructure (the letter P) froma single diffraction experiment,yielding both the shape and theheight of the sample. We combineFourier transform holography [1],which – by a single Fourier transform –gives an unambiguous image of thesample structure, with iterative phaseretrieval procedures [2] that enable us

Principal publication andauthorsL.-M. Stadler (a), C. Gutt (a),

T. Autenrieth (a), O. Leupold (a),

S. Rehbein (b), Y. Chushkin (c),

G. Grübel (a), Phys. Rev. Lett. 100,

245503 (2008).

(a) HASYLAB at DESY, Hamburg

(Germany)

(b) BESSY, Berlin (Germany)

(c) ESRF


46


EE SS RR FF

to push the spatial resolution towardthe diffraction limit, which isdetermined by the maximum photonmomentum transfer.

In Fourier transform holography, phaseretrieval becomes particularly simplesince, in a typical setup, the object anda spatially nearby reference arecoherently illuminated. A single Fouriertransform of the recorded hologramyields the convolution of the objectand reference amplitudes, where thespatial resolution achievable iscomparable to the reference sourcesize [1]. In the present case 5 gold dotswith about 175 nm diameter wereplaced on a circle of 2.5 µm radiusaround the gold nanostructure, eachacting as reference source in thescattering process.

The corresponding hologram, which isshown in Figure 53, was recorded atbeamline ID10C using coherent 8 keVphotons. A single Fourier transform ofthe measured intensities gives the

spatial autocorrelation of the overallobject, including the cross-correlationbetween the object and the dots,which yields the object shape directlyand its complex conjugate (a 180°-rotated copy) for each dot, as can beseen in the inset of Figure 53. Theindividual images are averaged toimprove the statistics and the result –already comprising a qualitative imageof the object – is fed into furtheriterative phase retrieval algorithmsthat are commonly used in coherentdiffraction imaging [2]. Carrying outsuch phase retrieval runs yielded animage of the sample’s electron densityprojected along the beam directionwith ≈ 25 nm resolution. Knowing thatthe sample under investigation wasmade of gold and exploiting the factthat the scattering process can bedescribed as Thomson scattering, it isstraight forward to derive a meansample height, which would be thecorrect height if the sample were flat.Actually, in the present case we can tellhow well the assumption of a flatsample topography is, since thisinformation is encoded in the phasesof the hologram’s Fourier transform.For our sample, a homogeneous heightis confirmed with the actual value of≈ 235 nm, as illustrated in Figure 54.

In conclusion, we have shown thathard X-ray holographic diffractionimaging is an excellent technique fordetermining electron density profileson the nanoscale. A combination ofthe Fourier transform holographyresult and iterative phase retrievalmethods made it possible to push thespatial resolution toward thediffraction limit. By virtue of thisconcept, we determined the absoluteelectron density and derived bothshape and height of a lithographicgold nanostructure. This approach withcoherent hard X-rays is ideally suitedfor applications in materials science,where samples can be thick, mayconsist of buried structures, or have tobe measured under extreme conditionssuch as high pressure. Finally, weenvision experiments of similar type tobe carried out at future hard X-rayfree-electron laser sources, wherediffraction patterns can be collectedwithin a single shot at the time scale of≈ 10 fs, opening fascinating possibilitiesfor imaging fast dynamic processes.

Fig. 53: Recordedhologram of a goldnanostructure – the

capital letter P –surrounded by 5

reference scatterers.The inset shows the

central part of themodulus of the Fourier

transform of therecorded diffraction

intensities, imaging theobject and its rotated

copy 5 times each.Both images are shown

on a logarithmicpseudocolour scale.

Fig. 54: Visualisation of the objectgeometry (letter plus reference

dots) as determined from theelectron density profile.

References[1] S. Eisebitt, J. Lüning,

W.F. Schlotter, M. Lörgen,

O. Hellwig, W. Eberhardt, J. Stöhr,

Nature (London) 432, 885 (2004).

[2] H.N. Chapman, A. Barty,

S. Marchesini, A. Noy,

S.P. Hau-Riege, C. Cui,

M.R. Howells, R. Rosen, H. He,

J.C.H. Spence et al., J. Opt. Soc.

Am. A 23, 1179 (2006).



On-axis microstructural studies of high-performance polymeric fibres

Microfocus X-ray diffraction (µXRD) isone of the few techniques that canprobe the microstructures withinindividual polymeric fibres. Untilrecently, most µXRD studies of singlefibres were performed in the standardfibre diffraction geometry. In thisgeometry, with the fibre perpendicularto the incoming beam, skin-corevariations can be investigated basedupon the changes observed betweendiffraction patterns collected atdifferent positions across the fibrediameter. This approach is somewhatlimited by the lack of informationconcerning the origin of scatteringelements along the X-ray beam path.This makes it necessary to employmodelling in order to reconstruct skin-core morphologies in detail. Not onlydoes this require a simplification offibre structure, it risks the introductionof model bias.

A novel approach to skin-coremicrostructural studies has beendemonstrated recently at the ID13microfocus beamline. By adopting anon-axis µXRD geometry, skin-coremicrostructural variations can beimaged directly from within singlefibres exhibiting preferred orientation,and without the need for complexmodelling. The technique involvescarefully aligning a short section offibre to the beam axis and thencollecting diffraction patterns over atwo-dimensional area (Figure 55).From the resulting diffraction data it ispossible to reconstruct an image of thefibre cross-section based upondifferent microstructural parameters.For example, point-to-point variationsin rotational disorder can be visualiseddirectly between the fibre skin andcore whilst the nature of radialcrystallographic texture can be studiedin detail.

On-axis µXRD experiments on singleas-spun poly(p-phenylenebenzobisoxazole) (PBO) fibres reveal amicrostructure consisting of arelatively disordered skin and core,separated by an intermediate regionwhich exhibits a lower degree of

rotational disorder (Figure 56a). Thewidth of the fibre skin appears to varywith position around the fibrecircumference whilst the core region isslightly offset from the geometriccentre of the fibre. This general skin-core morphology can be explained interms of differences in coagulationrate and local variations in chainmobility during fibre formation.Structural asymmetry, meanwhile,suggests a heterogeneous coagulationenvironment.

High performance fibres, like manyother materials, have hierarchicalmicrostructures. It is therefore alsonecessary to consider structural featureswhich may exist over larger lengthscales when investigating skin-coremorphologies, such as those accessibleto small-angle X-ray scattering (SAXS).On-axis SAXS experiments on single poly(p-phenylene terephthalamide)(PpTA) fibres reveal variations in SAXS

Principal publications andauthorsR.J. Davies, M. Burghammer,

C. Riekel, Macromol., 40, 5038

(2007); R.J. Davies,

M. Burghammer, C. Riekel, J.

Appl. Crysta., 41, 563 (2008);

R.J. Davies, C. Koenig,

M. Burghammer, C. Riekel, Appl.

Phys. Let., 92, 101903 (2008).

ESRF

Fig. 56: a) Skin-core variations in rotational disorder for PBO fibre and b) variationof SAXS intensity over PpTA fibre cross-section. Individual diffraction patterns are

shown inset.

Fig. 55: a) Schematic of on-axis µXRD geometry and b) Scanning Electron Micrographof a PpTA fibre section mounted on a glass capillary and viewed on-axis (prepared

using a pulsed UV laser cutter).


48


EE SS RR FF

between the fibre skin and core(Figure 56b), and differences betweenfibres having different processinghistories. Following fibre spinning, theskin region contains elongatedstructures aligned parallel with thefibre’s edge whilst the remainder of thefibre exhibits an isotropic SAXS patternindicative of randomly oriented orcylindrical scattering elements.Following heat treatment, SAXS isreduced in all but the very core of thefibre. This suggests that, likecoagulation, structural modificationdue to the heat treatment process mayalso occur heterogeneously.

Aside from skin-core microstructuralvariations, the on-axis µXRD geometryalso allows radial crystallographictexture to be investigated in detail.Both PBO and PpTA fibres are found toexhibit this type of preferredorientation over their entire cross-sections. A model has been developedwhich allows azimuthal scatteringprofiles to be calculated for specifichk0 reflections under an on-axisgeometry. By minimising the modelagainst diffraction data, the nature ofradial crystallographic texture can be

probed. The results reveal that, for as-spun PBO, the a* unit cell axis is offsetfrom the fibre radial direction by 23°.This angle corresponds to an almostradial arrangement of the highlyplanar PBO molecules. This suggeststhat solvent outflow duringcoagulation is at the origin of theradial crystallographic texture. In anedge-on arrangement, the PBO chainshave the lowest drag coefficient.

Adopting an on-axis µXRD geometryhas revealed many new microstructuralfeatures for high performance fibreswhilst allowing previously knownfeatures to be examined in greaterdetail or confirmed from modelpredictions. The interest incharacterising such microstructures isconsiderable. Not only are theythought to influence fibre mechanicalproperties, they can also dictate theinterfacial performance of compositematerials. If these complex structure-property relationships can beelucidated, it could provide a cost-effective way to tailor fibre propertiesduring manufacture whilst openingthe door to producing the super-fibresof tomorrow.

X-ray microdiffraction of microsamplesmanipulated in fluidic environmentwithout mechanical contact

Sample manipulation is one of thecritical issues in synchrotron radiationmicrobeam experiments and inparticular for experiments in specialenvironments such as microfluidiccells. A common requirement is toposition and move the sample infront of the beam. This is usuallyaccomplished by fixing the sample on

a support (e.g. capillary tip, grid),which is then moved precisely intothe beam (Figure 57a). Consequently,this limits the degrees of freedom forsample orientation and manipulationand does not allow interactionsbetween two independent objects tobe studied in terms of mutualorientation, mutual growth

AuthorsD. Cojoc (a), H. Amenitsch (b),

C. Riekel (c), E. Ferrari (a),

M. Rappolt (b), M. Burghammer (c),

S. Santucci (a), G. Grenci (a),

B. Sartori (b), B. Marmiroli (b).

(a) TASC National Laboratory,

Trieste (Italy)

(b) Institute of Biophysics and

Nanosystems Research, Austrian

Academy of Sciences, Graz

(Austria)

(c) ESRF

Fig. 57: Sample manipulation in X-ray microdiffractionexperiments: a) Mechanical contact/trap and b) Laser trap

without mechanical contact. c) Microscope image of a clusterof liposomes trapped optically (dash line) inside a capillary (the

walls of the capillary are seen in black on the lateral sides),scale bar 10 µm. d) Plot of the azimuthally integrated intensity

(red) and the Lorentz function (black) fitted to experimentaldata for the diffraction pattern (inset) at one position of the

X-ray beam on an optically trapped POPE cluster.



(single cell study, real time crystalgrowth) [1].

One solution, laser tweezers haveemerged as a powerful tool tomanipulate a large variety ofmicroparticles immersed in fluids [2].We have now built a laser tweezersetup based on a custom invertedmicroscope and an infrared laser totrap, manipulate and aggregatemicrometre-scale liposome particlesinside a 100 micrometre glass capillary[3]. The micro-focused synchrotronradiation and laser beams are alignedto intersect each other, allowing theX-ray diffraction signal to beassociated with the micrometre-sizedregion of interest inside the capillary(Figure 57b). This setup has beendemonstrated at the ID13 microfocusbeamline by trapping about 50multilamelar palmitoyl-oleoyl-phosphatidylethanolamine (POPE)liposome clusters of about 10 µmdiameter (Figure 57c) and performinga scanning diffraction experiment witha 1 µm beam at about 13 keV. Theazimuthally integrated intensity of thediffraction pattern obtained from theliposome cluster at one position of theX-ray beam is shown in Figure 57d.Multiple laser traps can be createdusing diffractive optical elementsimplemented on a spatial lightmodulator [4]. This allowsmanipulation of small separatedclusters of liposomes and their fixationwithin the optical path of the X-raybeam. Using two traps, two differentclusters could be brought into contactand, consequently, a reaction betweenthem could be induced. The proof of

concept is illustrated by Figure 58.Multiple traps can be also used to trapand manipulate larger particles such asstarch granules, as shown inFigure 58c. We have improved theinitial laser tweezers setup, allowing2D optical trapping in a cylindricalcapillary (Figure 58a), to a setup withfull 3D optical trapping in a 80 µmsquared capillary (Figure 58c). Thisenables positioning and orienting thesample in any point of the capillary [5].The scanning diffraction image takenfrom a 20 x 24 µm starch granule isillustrated in Figure 58d together withrefraction from the wall of thecapillary thus demonstrating thetrapping capability far away from thecapillary wall.

The results presented here have givena first insight into this fascinating newfield - to look to single supra-molecular assemblies with micrometre-and submicrometre-sized X-ray beams,which will find multiple applications inthird- and in particular future fourth-generation synchrotron radiationsources.

Fig. 58: Two clusters of liposomes trapped at separate positions, indicated by red and blue lines: a) microscopeimage of the traps (infrared laser) and b) X-ray scanning diffraction image (azimuthaly integrated intensities,mesh 3 µm x 5 µm) from the two clusters, with the corresponding intensity profiles. c) Microscope image of alarge starch granule trapped with three laser traps (red spots) and d) scanning diffraction image (diffractionpatterns) of the starch ({001}-9 nm).

References[1] C. Riekel et al., Curr. Opin.

Struct. Biol. 63 556 (2005).

[2] D.G. Grier, Nature 424 810

(2003).

[3] H. Amenitsch et al., AIP Conf.

Proc. 879 1287 (2006).

[4] D. Cojoc et al., Appl.Phys. Lett.

91 234107 (2007).

[5] D. Cojoc et al. Nat. Meth.

submitted (2008).


50


EE SS RR FF

Probing protein structural changes insolution

Proteins are sequences of aminoacidsthat fold in 3D structures toaccomplish their biological functions.Several studies have shown thatprotein conformational changes oftenplay a crucial role in functionregulation. So far, direct structuralinformation on molecular motions hasbeen obtained only in the crystal stateby time-resolved proteincrystallography [1]. However, proteincrystals, while allowing model-freeatomic resolution reconstruction, mayinduce stress/strain in the protein andhinder biologically-importantstructural changes. For proteins insolution, only indirect information hasbeen obtained so far through time-resolved optical spectroscopy [2], thushighlighting the need of more direct,time-resolved, structural probes. In thelast decade, protein solution scatteringhas undergone tremendousimprovements in instrumentation andunderstanding of the experimentaldata, thus opening the possibility ofstudying protein structures in solutionalthough at somewhat limitedresolution of 5-10 angstroms [3].

We have therefore attempted tocombine the exceptional timeresolution achievable at ID09B (downto 100 ps) with the structuralsensitivity already demonstrated forprotein solution scattering.

We demonstrate this approach by usingit to monitor the structural changes ofdifferent proteins afterphotodissociation of the ligand.The basic idea of the experiment is asfollows: X-ray pulses are selected fromthe storage ring using a synchronisedultra-fast mechanical shutter(Figure 59). Synchronised laser pulsesare produced by the laser systemavailable at ID09B, where differentpulse lengths and laser wavelength canbe used. The laser/X-ray pulse pairs(suitably shifted by a time delay thatcan be varied electronically with anaccuracy of ~ 10 ps) are sent to thesample to excite and probe thestructure respectively. Typically, for agiven time delay, the sample has to beexposed to about 1000 single laser/X-ray pulses to make efficient use of theCCD dynamic range.

For reciprocal vector(q = 4π / λsin(θ /2)) in the range 0.05 to2 Ang-1 protein structural featuresgives rise to modulations of thescattering pattern that are fingerprintsof the protein structure thus giving adirect structural probe. Since the light-induced differences are relatively small(< 5%), the data are usuallypresentedas laseron-laseroff differencescattering patterns.

To test this new experimentalapproach we studied humanhaemoglobin because it offers a well-known example of structurally tunedbiological function. Haemoglobin ismade of four subunits each of themable to reversibly bind small moleculessuch as oxygen or carbon monoxide; itperforms its biological task efficiently(oxygen delivery) thanks to theequilibrium between twoconformations (called “R” and “T”)having a dramatically different ligandaffinity. Because of its fundamentalbiological importance, the transitionbetween the R and T conformations(called “quaternary transition”) hasattracted the interest of manyresearchers over the last decades.

Principal publication andauthorsM. Cammarata (a,b),

M. Levantino (c), F. Schotte (d),

P.A. Anfinrud (d), F. Ewald (a),

J. Choi (e), A. Cupane (c),

M. Wulff (a), H. Ihee (e), Nature

Methods 5, 881 (2008).

(a) ESRF

(b) Centre for Molecular Movies,

Niels Bohr Institute, University of

Copenhagen (Denmark)

(c) Department of Physical and

Astronomical Sciences, University

of Palermo (Italy)

(d) Laboratory of Chemical

Physics, National Institute of

Diabetes and Digestive and

Kidney Diseases, National

Institutes of Health, Bethesda

(USA)

(e) Center for Time-Resolved

Diffraction, Department of

Chemistry (BK21), Korea

Advanced Institute of Science and

Technology, Daejeon (Republic of

Korea)

Fig. 59: Schematic of the TR-WAXSexperimental technique (Adapted

by permission from MacmillanPublishers Ltd: Nature Methods

M. Cammarata et al., NatureMethods 5, 881, copyright (2008)).



In our experiment, a short laser pulsewas able to break the protein-ligandbonds thus triggering proteinstructural relaxation from the initialligand bound (R) structure to theunbound (T) structure; ligandrebinding from the solution eventuallyrestores the initial condition.

Thanks to the high photon fluxavailable at ID09B, we were able toprovide a direct observation of theR –> T transition and to measureunambiguously the “waiting time”before the quaternary transition takesplace. We found a time scale of about2 µs challenging the previousassignment obtained with absorptionspectroscopy [4] but in agreement withmore recent time-resolved Raman data[2]. The main results are shown inFigure 60. The difference scatteringpatterns (laser at a given time delay –laser off) show a clear evolution of thesignal that can be interpreted as thetransition from an intermediatestructure to the unbound equilibriumstructure. The fact that the 3 µs andthe 10 µs curves are very similarimmediately suggests a time scale forthe transition of a few microseconds.

This study shows that time resolvedscattering can be a valuableexperimental technique to gain insightinto protein structures and theirchanges. Different excitationmechanisms like pH jump or T-jumpcould also be used and preliminarytests are being carried out.

Fig. 60: Experiment onhaemoglobin: left panelsshow the schematic of theflash photolysis, right panelsshow the data at differenttime delays (green) along withthe reconstruction using twotime delays (black).

References[1] F. Schotte, M.H. Lim,

T.A. Jackson, A.V. Smirnov,

J. Soman, J.S. Olson, G.N. Phillips,

M. Wulff, P.A. Anfinrud, Science

300, 1944 (2003).

[2] G. Balakrishnan, M.A. Case,

A. Pevsner, X.Z.C. Tengroth,

G.L. McLendon, T.G. Spiro, J. Mol.

Biol. 340, 843 (2004).

[3] D.I. Svergun and M.H. Koch,

Rep. Prog. Phys. 66, 1735 (2003).

[4] J. Hofrichter, J.H. Sommer,

E.R. Henry, W.A. Eaton, Proc. Natl.

Acad. Sci. USA 80, 2235 (1983).

The molecular basis of the braking actionof muscle studied by X-ray interference

We normally think of muscles as themotors that drive the movements ofthe body, but they can also act asbrakes to resist an external force. Aneveryday example of this is the actionof the muscles in the front of the legsto resist the force of gravity whilewalking down stairs. A more dramaticexample would be deceleration of thebody on landing after a jump(Figure 61), when the brakes must beapplied in a small fraction of a second.This rapid braking response to anexternal force is an intrinsic property ofisolated skeletal muscle cells. Our recentresults at beamline ID02 have revealedits molecular basis for the first time.

Skeletal muscle cells are packed withparallel arrays of two kinds offilaments, composed of the proteins

myosin and actin respectively. Myosinis the motor protein; biochemicallydriven changes in the shape of part ofthe myosin molecule, called its motordomain, drive muscle contraction. Theactin filaments form the moleculartracks for these myosin motors; whena muscle contracts the myosin motorspull the actin filaments along the

Principal publication andauthorsE. Brunello (a), M. Reconditi (a),

R. Elangovan (a), M. Linari (a),

Y.-B. Sun (b), T. Narayanan (c),

P. Panine (c), G. Piazzesi (a),

M. Irving (b) and V. Lombardi (a),

Proc. Natl. Acad. Sci. USA 104,

20114-119 (2007).

(a) Università degli Studi di

Firenze (Italy)

(b) King’s College London (UK)

(c) ESRF

Fig. 61: The leg extensorsdecelerate the body aftera jump (adapted fromCerretelli, Fisiologiadell’Esercizio, SEU, 2001).


52


EE SS RR FF

myosin filaments. The filaments havean almost crystalline structure; motordomains are spaced at regular intervalsalong the myosin filaments, and thisregular structure produces ameasurable diffraction pattern whenan isolated muscle fibre is illuminatedby a narrow beam of X-rays. One ofthe X-ray reflections, called the M3,has two closely spaced peaks,Figure 62 (left); this fine structure isan interference effect related to thebipolar structure of the myosinfilament: each myosin filamentactually contains two arrays of myosinmolecules. This X-ray interferenceeffect has proved to be a powerfultechnique to reveal molecularmechanisms in muscle contractionbecause it can measure the movementsof the myosin motors along thefilaments with angstrom resolution inan intact single muscle cell.

Measurements of these X-rayinterference signals with sub-millisecond time resolution werepreviously used to investigate themolecular basis of force generationand active shortening in the myosinmotor [1-3]. We have now used thetechnique to determine themechanism of the braking response ofmuscle to a rapid stretchingmovement. The results showed thatthe response of a muscle to beingstretched is fundamentally different tothat during force generation or activeshortening. An important clue to themechanism came from high resolutionmechanical measurements, whichshowed that the stiffness of themuscle fibre increases in less than a

millisecond after being stretchedquickly. The integratedmechanical/structural mechanism wasrevealed by combining the mechanicaland X-ray interference results, andtaking into account the fact that eachmyosin molecule has two motordomains. Before being stretched(Figure 62a), one of the motordomains (red) of each myosin moleculeis attached to an actin monomer (darkgrey), but its partner motor (yellow) isdetached. When the muscle isstretched (Figure 62b), some of thepartner motors (pink) attach to theneighbouring actin monomer (lightgrey) and resist being stretched, withthe first motor acting as the strainsensor. This model can explainquantitatively the increase in stiffnessand the changes in both interferencepeaks of the M3 X-ray reflection. Italso provides an answer to thelongstanding question: ‘why does eachmyosin molecule in muscle have twomotor domains?’ When muscle isworking as a motor to generate forceand active movement, each myosinuses only one of its two motordomains. The partner domain is heldin reserve, to act almostinstantaneously as a brake when itsattachment is promoted by strain inthe first motor domain. In this waymuscle resists an external stretchingmovement while minimising the stresson an individual motor.

Fig. 62: Molecular basisof muscle braking;

the M3 X-ray reflectionand the conformationof the myosin motors

before a), and afterbeing stretched b).

References[1] G. Piazzesi et al. Nature 415,

659-662 (2002)

[2] M. Reconditi et al. Nature 428,

578-581 (2004)

[3] G. Piazzesi et al. Cell 131, 784-

795 (2007)



Introduction

Macromolecular crystallography hascontinued to evolve at the ESRF. Theinvestment made in automation andbeamline reliability is bearing fruit in anumber of interesting ways. It is nowclear that, in many cases, the complexityof the biological systems under studycan only be fully revealed by combiningextensive initial sample evaluation withdetailed examination of the mostpromising crystals on the beamline bestsuited to the experiment. Obtaining themaximum information from a crystalalso requires optimisation of the datacollection procedure. Only by employingmore complex experimental strategiescan we hope to obtain a successfuloutcome to large numbers ofexperiments.

The improved operational reliability ofthe macromolecular crystallography(MX) beamlines has enabled a recentchange to the way they can be accessed.Remote access was introduced during2008 and allows complete instrumentcontrol from the user’s home laboratory.This mode of access has already provenpopular, currently accounting for morethan 20% of all macromolecularcrystallography experiments, a figure weexpect to see rising. However, theadvent of remote access has revealeda number of areas where furtherinstrument development would beuseful. It has also emphasised the needfor the ESRF to adapt staffing levels andstaff profiles in order to fully supportsuch access to the beamlines.

The results presented in this chapterreveal the extent to which access to abroad portfolio of beamlines andsynchrotron techniques is important forachieving success. We observe thatboth microfocus X-ray beams andtunability of the X-ray source arecritical. In this latter context, the CRG

beamlines BM14, BM16 and BM30Aalso play an important role, providingexcellent services for their respectivecommunities. The increased study ofmacromolecular complexes has led toa resurgence of interest in small-angleX-ray scattering from protein solutions.To provide much needed extra capacityID14-3 has been converted to a SAXSbeamline in a collaboration with theEMBL; it will provide rapid access forexperiments in this area. The first suchexperiments were carried out on thebeamline in 2008 and we look forwardto a productive year in 2009.

Some of the fruits of the researchundertaken in 2008 on the ESRF MXbeamlines are reported in this chapter.We concentrate only on biologicalstructures obtained and the insights thatthese give. Methodology contributing tothis success is reported in the chapter on

Structural Biology

Fig. 63: Cryogenic cooling is used to prolong the life-time of crystals in an X-ray beam.However, in some cases flash-cooling degrades crystal quality if conditions are notideal. Annealing or tempering of samples adversely affected by cryocooling can oftenimprove diffraction quality and is now considered a standard technique inmacromolecular crystallography. A small, inexpensive automatically-operatedannealing device is installed on all ESRF macromolecular crystallography beamlinesthat allows annealing of samples either from the beamline or by users accessing thebeamline remotely. For details see: T. Giraud et al., J. Appl. Cryst. 42, 125-128 (2009).


54

Structural Biology

EE SS RR FF

An open and shut case

Bacteria are survivors and cope withdramatic changes in environmentduring their life cycle. One such stressis osmotic shock, in moving from asalty environment (such as biologicalfluids) to a salt-free environment(water), whereby bacteria experiencepressures up to 14 atm. The pressure isfelt as a swelling of the cytoplasm(turgor pressure), and, unless relieved,the consequence is rupture of themembrane. However, bacteria are ableto release this pressure build up usingthe mechanosensitive channels, MscSand MscL.

The breakthrough in the structuralunderstanding of mechanosensitivechannels came from the work of theRees lab [1]. This 3.95 Å structureshowed that MscS is a heptamer witheach monomer consisting of three

transmembrane helices at the N-terminus and a large cytoplasmicdomain. The heptamer is arrangedsymmetrically around a 7-foldsymmetry axis normal to themembrane. One of the transmembranehelices (TM3) is split into two TM3aand TM3b. Two rings of leucine sidechains on TM3a point into the centralchannel. This provides a “vapour lock”and, in essence, these side chainsprevent any ions or proteins fromtravelling from the cytoplasm into theperiplasm. Under pressure this lockmust be opened, however themechanism by which this isaccomplished remained obscure.

The route to structural insight camefrom identifying mutants which weremore likely to be open fromelectrophysiology studies.

Principal publication andauthorsW. Wang (a), S.S. Black (b),

M.D. Edwards (b), S. Miller (b),

E.L. Morrison (b), W. Bartlett (b),

C. Dong (a), J.H. Naismith (a) and

I.R. Booth (b), Science 321, 1179

(2008).

(a) Centre for Biomolecular

Sciences, University of St.

Andrews (UK)

(b) School of Medical Sciences,

University of Aberdeen (UK)

Fig. 64: The open structure of MscSis shown in orange (A subunit) and

yellow (subunits B-G).The closed structure is shown in

red (A subunit) and pink (subunitsB-G). TM1 and TM2 have rotated

around the seven-fold axis.

Methods and Instrumentation (p. 120),and in Figure 63, which presents a newannealing device available on thebeamlines.

The current degree of automationavailable on the MX beamlines is due tothe consistent hard work of the ESRFBLISS, SciSoft, and MX Groups inpartnership with the EMBL Grenobleoutstation. Their efforts were recognised

by the annual BESSY innovation award.Work on renewing the MX suite ofbeamlines will continue through 2009with the development of detailed plansfor new beamlines and modes of access.We look forward to the help andinput of our User community in thecontinuation of these projects.

G. Leonard and S. McSweeney



Thus we set out to study the crystalstructures of different mutants of MscSto see whether any showed an alteredstate. The project would have beenimpossible without the ESRF facilities,membrane protein crystals do notdiffract well and we screened manyhundreds of crystals from severalmutants before finally arriving at a3.45 Å structure for the A106V mutant.This mutation appears to destabilisethe previously observed closed formbut does not appear to stabilise theopen form. Compared to the nativestructure, the mutant MscS showsdramatic changes in the position andorientation of transmembrane helices,most notably TM1 and TM2. These twohelices have moved around 1/7 of arevolution and increased their tilt withrespect to the membrane (Figure 64).As a result of this dramatic movement,TM3a pivots at G113 in such a waythat it become parallel to themembrane normal. It is also “pulled”out of the central pore.

The movement of TM3a results in airis like motion of the leucine sidechains which create an open pore.Thus a picture emerges of TM1 andTM2 acting as the pressure sensorwhich in turn operates the valve onTM3a. The structural data of courseare only a model for the situationinside the cell. The structural modelfor the opening and closing of MscSpredicts a number of key side chaininteractions. These hypotheses weretested by site directed mutagenesisand gave strong support to thestructural model.

The significance of our work lies inunderstanding the physiology ofpathogens; the opening and closing ofchannels is an important topic inbiology. The second significant findingis that carefully chosen site directedmutants may provide an importantroute to the observation ofconformational states of membraneproteins.

Reference[1] R.B. Bass, P. Strop, M. Barclay,

D.C. Rees, Science 298, 1582

(2002).

Structural basis of Dcp2 recognition andactivation by Dcp1RNA degradation plays an importantrole in the control of gene expressionand in the elimination of aberrantmRNAs. Decapping is a critical stepsince it promotes the bulk of mRNAturnover and functions in specialisedpathways such as nonsense-mediateddecay (NMD), adenine/uridine-richelement (ARE)-mediated decay, andthe turnover of some mRNAspromoted by miRNAs [1].

Removal of the 5’ cap structure iscatalysed by the decapping complexconsisting of at least two subunits:Dcp1p and Dcp2p. Dcp1p is a smallprotein containing an EVH1 domain,which is generally a protein-proteininteraction module. The catalyticsubunit, Dcp2p, recognises cappedmRNA substrates and cleaves thepyrophosphate bond of the m7GpppNcap, releasing m7GDP as product.Dcp2p belongs to the nudix hydrolyasefamily, which contain the nudixsignature nudix motif. The crystalstructure of Dcp2p from S. pombe

reveals that a conserved N-terminalregion forms a two-lobed structurewith an α-helical domain precedingthe nudix domain [2]. That Dcp2pprefers longer RNA substrates suggeststhe presence of spatially separatedRNA-binding and catalytic sites on thenudix domain, but the nature and siteof the RNA binding domain of Dcp2phas not yet been determined.Additionally, despite the conservationof Dcp1p and its requirement fordecapping, how Dcp1p interacts withand activates Dcp2p remains elusive.

We determined the crystal structureof Dcp1p in complex with a truncatedS. pombe Dcp2p (Dcp2n) using datacollected on beamline ID29. Thestructure reveals both open and closedforms of the Dcp1p/Dcp2n complex(Figure 65). In the closed complex, acompact structure is formed with bothDcp1p and the N-terminal α-helicaldomain of Dcp2n being in closeproximity to the nudix domain of thelatter. In the open form, the α-helical

Principal publication andauthorsM. She (a), C.J. Decker (b),

D.I. Svergun (c), A. Round (c),

N. Chen (a), D. Muhlrad (b),

R. Parker (b), H. Song (a), Mol.

Cell 29, 337 (2008).

(a) Laboratory of Macromolecular

Structure, Institute of Molecular

and Cell Biology (Singapore)

(b) Department of Molecular and

Cellular Biology and Howard

Hughes Medical Institute,

University of Arizona (USA)

(c) Hamburg Outstation, EMBL

(Germany)


56

Structural Biology

EE SS RR FF

and nudix domains of Dcp2n do notinteract and are separated by aflexible linker, making the wholemolecule resemble a dumb-bell. Small-angle X-ray scattering (SAXS) analysisconfirmed that the two forms of theDcp1p/Dcp2n complex also exist insolution and that formation of theclosed complex is influenced by ligandbinding. Interestingly, we observed

that ATP is bound to the catalyticcentre (Figure 65), and that ATP, di-nucleotide (pAA) and the capanalogue m7GpppA enhance theformation of the closed complex andinhibit decapping. Mutational analysisin the flexible linker region indicatedthat the closed complex we observe is,or closely resembles, the catalyticallymore active form, suggesting that aconformational change between openand closed complexes might controldecapping. We have also identified aregion involved in RNA binding whichis located some distance from theactive site (Figure 66).

Our structural analyses coupled withbiochemical data suggest a possiblemechanism by which Dcp2p recognisesits substrate. The substrate is bound ina channel on the surface of Dcp2pwith the cap structure in the active site(Figure 66). The main body of the RNAthen wraps across the nudix domainand follows the channel along the BoxB helix. This model provides a possiblestructural explanation for thepreference of decapping enzymes forlonger RNA substrates: RNA wouldneed to be at least 12 residues long tobind in both the active site and theBox-B region on Dcp2p.

Our analyses suggest that Dcp1pstimulates the activity of Dcp2p bystabilising the closed complex, which ispromoted by the presence of substratein the active site. Finally our structurealso reveals that the Dcp1/Dcp2interface is not fully conserved,explaining why the Dcp1-Dcp2interaction in higher eukaryotesrequires an additional factor.

Fig. 65: Ribbondiagrams of the

Dcp1p-Dcp2ncomplex in the

closed and openconformations.

Dcp1p is shown ingreen, the

N-terminal domain(NTD) of Dcp2n in

pink, the nudixdomain of Dcp2n in

blue and its nudixmotif in red.

In the closed conformation, the ATP

molecule we observein the active site is

shown in a stickrepresentation.

Fig. 66: A model illustrating theputative RNA binding channel

in the closed form of the Dcp1p-Dcp2n complex.

A 12-mer poly(A) mRNA is shownin a tube representation and the

Box B region of Dcp2n is colouredin dark purple.

References[1] J. Coller, and R. Parker, Annu

Rev Biochem 73, 861 (2004).

[2] M. She, C.J. Decker, N. Chen,

S. Tumati, R. Parker, and H. Song,

(2006). Nat Struct Mol Biol 13, 63

(2006).

Structural biochemistry of a bacterialcheckpoint protein

Maintaining genome integrity is oneof the most important processes forthe sustainment of life and theprevention of cytotoxic orcancerogenic DNA aberrations. Cellspossess various DNA repair pathwaysas well as DNA damage checkpointcontrol mechanisms, to ensure that cellcycle progressions occur only with

intact chromosomes. In humans,inactivated DNA repair pathways anddamage checkpoint controls lead togenetic instability, cancer and ageingsyndromes. DisA (DNA integrityscanning A) was recently identified asa checkpoint protein in Bacillus subtilis[1]. It appears to co-localise with sitesof DNA double-strand breaks, delaying

Principal publication andauthorsG. Witte (a,b), S. Hartung (a,b),

K. Büttner (b,c) and

K.-P. Hopfner (a,b), Mol. Cell 30,

167-178 (2008).

(a) Munich-Centre for Advanced

Photonics, Munich (Germany)

(b) Center for Integrated Protein

Science at the Gene Center,

Munich (Germany)

(c) Present address: Max-Planck-

Institute for Biochemistry,

Martinsried (Germany)

the cell from entering the sporulationphase. Intriguingly, DisA appears toexist in a single, large complex insideB. subtilis that scans the chromosomefor DNA damage. To reveal themolecular and structural mechanism ofDisA we used a combination of high-resolution X-ray crystallography, small-angle X-ray scattering (SAXS) andbiochemical and biophysical analysis.

We determined the crystal structure ofthe Thermotoga maritima DisAhomolog (TmaDisA) to 2.1 Åresolution, using multiple-wavelengthanomalous dispersion data collected atbeamline ID29. According to analyticultracentrifugation, DisA forms a largeoctamer with a molecular weight of~360 kDa. We verified the likelyoctamer observed in the crystal latticeusing SAXS. A protomer of TmaDisAcan be divided into 3 domains(Figure 67): an N-terminal globulardomain (domain of unknown function,DUF147), a central helical domain anda C-terminal HhH domain. In theoctamer eight DUF147 domains formthe core of the complex, while theHhH domains form two peripheraltetrameric putative DNA bindingplatforms.

What is the function of DUF147?Surprisingly, at each of the fourinterfaces between opposing DUF147domains we found additional electrondensity for new types of cyclicdinucleotides. We identified theseligands as cyclic diadenosinemonophosphates and have shownthat DUF147 of DisA is in fact adiadenylate cyclase. It uses pairs ofATP molecules bound to opposingDUF147 domains for the cyclisationreaction. Thus we suggested renamingthe N-terminal domain DUF147 to“DAC-domain” (for diadenylatecyclase). The finding of noveldiadenylate cyclase activity and c-di-AMP as product of an enzymaticreaction is quite intriguing,considering the well known andimportant role of its cousin c-di-GMPas a second messenger in variousbacterial processes, including biofilmformation [2]. Possibly, c-di-AMP alsoacts as messenger, e.g. to coordinateDNA repair and sporulation, or moregenerally cell division since DisA is alsofound in non sporulating bacteria.

How is di-adenylate cyclase linked toDNA damage scanning? The presenceof four HhH-motifs at both ends of theDisA octamer suggests that DisA maynot recognise DNA double-strandbreaks, but rather its repairintermediates, e.g. Holliday junctions.Gel-electrophoretic mobility shiftassays showed that DisA preferentiallybinds to DNA strands resembling suchrecombination intermediates but notto plain ssDNA or dsDNA. Consistently,branched DNA substrates, but notplain ssDNA or dsDNA, stronglyinhibited the DAC-activity. Thestructural arrangement of the octameris ideally suited to allostericallyregulate di-adenylate cyclase activityby binding of branched DNA to theHhH-domains array via conformationalchanges transmitted by the helicalmiddle domain (Figure 68).

Our results suggest that DisA maydetect sites of unfinished DNA double



Fig. 67: Overall structure ofThermotoga maritima DisA. Oneprotomer of the octamericassembly is shown in cartoon style:the N-terminal DAC (DUF147)-domain is shown in green, thehelical central spine domain inblue and the C-terminal HhHdomain in orange. The otherchains in the octamer are shownas ribbons. The binding of cyclic-di-adenosine monophosphate at theinterface of two DAC-domains ishighlighted.

Fig. 68: Model for the mechanismof action of DisA in checkpointcontrol. In the absence ofchromosomal damage (scanningmode), DisA synthesises c-di-AMP.Recognition of branched nucleicacids such as stalled replication forksor recombination intermediatesmight inhibit c-di-AMP synthesisthus signalling the presence ofunsegregatable chromosomes.Several aspects of the model arespeculative, such as the precise

mode of DNA binding andthe conformationalchanges shown.


58

Structural Biology

EE SS RR FF

strand break repair, recombinationintermediates or stalled replicationforks - all of which should not occurduring sporulation or cell division. DisAmay signal the presence ofsuch structures via c-di-AMP (or lack of

c-di-AMP) although the precise role ofc-di-AMP needs to be addressed infuture studies. Intriguingly, DUF147 ispresent in many other bacteria as wellas in archaea, arguing for the existenceof other c-di-AMP associated processes. References

[1] M. Bejerano-Sagie et al., Cell

125, 679 (2006).

[2] U. Römling and D. Amikam,

Curr Opin Microbiol. 9, 218

(2006).

AcknowledgementsThis work was supported by grants from the German Research Council (SFB 684) and the European

Union (IP DNA Repair) to K.-P.H.

Pilus assembly across the bacterial outermembrane

The chaperone-usher (CU) pathway is atranslocation system used to assembleadhesive multi-subunit fibres on theouter surface of gram-negativebacteria. CU pili are formed by the non-covalent polymerisation of severalhundreds or thousands of pilus subunitswhich consist of an incompleteimmunoglobulin (Ig)-like fold lackingthe C-terminal β-strand. In theperiplasm, a cognate chaperone assistsin pilus subunit folding by donating aβ-strand to complement the truncatedIg-like fold of the pilus subunit, aprocess termed donor-strandcomplementation (Figure 69, A*) [1].Chaperone:subunit complexes are then

recruited to a pilus assembly platformin the outer membrane (OM) called the“usher”. The usher catalyses orderedsubunit polymerisation and mediatestranslocation of the nascent pilus to thecell surface. Polymerisation of pilussubunits occurs through anintermolecular fold complementationmechanism involving the first 10-20residues (termed “N-terminalextension” or Nte) of the pilus subunitnext in assembly. The Nte of a newlyrecruited subunit inserts in the groove(left by the missing strand) of thesubunit previously assembled,competing off the chaperone stillassociated with that subunit (Figure 69,B*) [2]. This so-called “donor-strandexchange” process, in which thechaperone donor-strand is exchangedfor an incoming subunit’s Nte, occurs inthe absence of an electrochemicalgradient or an hydrolysable energysource. How subunit polymerisationand fibre translocation across themembrane are orchestrated by theusher remains poorly understood,largely due to the lack of highresolution structural information onthe assembly platform.

Using data collected at the beamlinesID29 and ID23-1, we determined thecrystal structure of the detergent-solubilised translocation domain of theP pilus usher PapC to 3.2 Å resolution.Ushers are ~800 residue proteinscomprising a central β-barrel domainflanked by periplasmic soluble domainsof ~125 and ~170 residues at the N- and C-termini respectively. TheN-terminal domain forms the

Principal publication andauthorsH. Remaut (a), C. Tang (c),

N.S. Henderson (d), J.S. Pinkner (e),

T. Wang (c), S.J. Hultgren (e),

D.G. Thanassi (d), G. Waksman (a),

and H. Li (c), Cell 133, 640 (2008).

(a) Institute of Structural and

Molecular Biology, Birkbeck

College/UCL London (UK)

(c) Biology Department, BNL, NY (USA)

(d) CID, Stony Brook University,

NY (USA)

(e) DMM, Washington University

School of Medicine, MO (USA)

Fig. 69: Schematic representationof the CU biosynthetic pathway,

labelled according the type 1 pilussystem (Fim cluster).



recruitment site for chaperone:subunitcomplexes. The function of the C-terminal domain is unknown, thoughsome indications point to a possiblerole in an early activation event.The crystal structure reveals theusher translocation pore comprises a24-stranded β-barrel that gives rise to achannel with inner diameter of ~ 45 by25 Å, compatible with the translocationof folded pilus subunits (Figure 70).In its non-activated form, however, thispore is occluded by a folded plugdomain that is inserted into the loopbetween strands 6 and 7. In addition, aβ-hairpin formed by adjacent strandscurves back out of the barrel wall, intothe pore lumen and contacts the baseof the plug domain. Most likely usheractivation necessitates a conformationalchange in these two elements, resultingin opening of the pore. The crystals ofthe PapC translocation domain alsoexhibit a dimerisation interface alongthe flat surface of the kidney-shapedβ-barrel, very similar to the twinnedpore observed by electron microscopyof 2D crystals of lipid-reconstituted full-length PapC.

In combination with the single particlecryo-electron microscopy imaging ofthe type 1 pilus usher (FimD) bound toa translocating three-subunit assemblyintermediate (Figure 70), this structureprovides, for the first time, a molecularsnapshot of a twinned poretranslocation machinery responsible forCU pilus assembly at the bacterial outermembrane. Unexpectedly, only a singlepore is used for organelle secretion.The combined structures suggest a

model in which both usher protomersare required for subunit polymerisationthrough a mechanism of alternatingchaperone:subunit complexrecruitment. At each step of theincorporation cycle, the N-terminaldomain of one of the two pores isengaged in binding thechaperone:subunit complex at thebase of the growing fibre, whilst theN-terminal domain of the opposingpore forms the docking site for a newlyincoming chaperone:subunit complex.During donor-strand exchange, thechaperone at the now penultimate sitein the fibre is released, therebyliberating that usher’s N-terminaldomain for a new recruitment step.In this way, alternations inchaperone:subunit recruitment ateither usher N-terminus allow the step-wise incorporation of new pilussubunits, with the growing fibretranslocating through the activatedchannel in the twinned pore complex.

Fig. 70: (Left) Side and top views ofthe PapC translocation channel.(Right) Model of the type 1 pilusassembly intermediate(FimD2:FimC:FimF:FimG:FimH)docked into the reconstructedelectron density obtained withsingle particle cryo-EM. A newlyincoming chaperone:subunitcomplex (FimC:FimA) is modelledin, bound to the N-terminal domain of the secondusher (FimD).

References[1] F.G. Sauer, K. Fütterer,

J.S. Pinkner, K.W. Dodson,

S.J. Hultgren and G.Waksman

Science 285, 1058 (1999).

[2] H. Remaut, R.J. Rose,

T.J. Hannan, S.J. Hultgren,

S.E. Radford, A.E. Ashcroft, and

G. Waksman, Molecular Cell 22,

831 (2006).

Principal publication andauthorsT. Krojer (a), J. Sawa (a), E. Schäfer (b),

H.R. Saibil (b), M. Ehrmann (c) and

T. Clausen (a), Nature 453, 885 (2008).

(a) Research Institute for Molecular

Pathology - IMP, Vienna (Austria)

(b) Crystallography Department

and Institute of Structural

Molecular Biology, Birkbeck

College, London (UK)

(c) Centre for Medical

Biotechnology, FB Biology and

Geography, University Duisburg-

Essen, Essen (Germany)

Structural insights into the DegPprotease-chaperone

Most proteins acquire a precisely-defined 3D-structure to fulfill theircellular functions. However, thismolecular architecture is highlysusceptible to various stress conditionssuch as heat-shock. The resultantmisfolding not only impairs proteinfunctionality but also generatesmolecular aggregates that areextremely toxic for the cell. To survivestress conditions, all organisms employ

sophisticated protein-quality-control(PQC) systems in which dedicatedchaperones and proteases collaborateclosely with each other to reduce theamount of misfolded, aggregation-prone protein species produced.

The heat-shock protein DegP, amember of the HtrA family ofproteases, is an essential PQC factor inthe cellular envelope of Gram-positive


60

Structural Biology

EE SS RR FF

References[1] B. Lipinska, O. Fayet, L. Baird

and C. Georgopoulos, Journal of

Bacteriology 171, 1574 (1989).

[2] C. Spiess, A. Beil and

M. Ehrmann, Cell 97, 339 (1999).

[3] T. Krojer, M. Garrido-Franco,

R. Huber, M. Ehrmann and

T. Clausen, Nature 416, 455

(2002).

bacteria [1] and exhibits strikinglydifferent properties to those of thefunctionally-related proteasecomplexes found in the cytosol:(i) Since the periplasm is devoid ofnucleotides, DegP has to functionwithout ATP. (ii) DegP can reversiblyswitch between protease andchaperone functions in a temperature-and substrate-dependent manner [2].This switch in activity allows animmediate response to environmentalchanges that are sensed more directlyin the periplasm than in the cytosoldue to the porous character of theouter membrane. (iii) DegP actsexclusively on misfolded proteins anddoes not depend on co-factor adaptoror regulatory proteins. The autonomyof DegP is based on an intricate multi-domain organisation comprising achymotrypsin-like protease domainlinked to two C-terminal PDZ domainsthat contain different binding sites forsubstrate proteins, allosteric effectorsand even cellular membranes.

In the absence of stress conditions,DegP exists as a hexamer (DegP6), astructure that we determined in 2002using data collected at the ESRF [3].DegP6 was present in a proteolyticallyinactive state, in which the loop LAplayed a central role in stabilising thehexamer as well as in inhibiting theproteolytic activity of an adjacentprotomer. Recent data revealed thatDegP6 acts as a substrate trap thatassembles, upon substrate binding,into the functionally active particlesDegP12 and DegP24. Notably, the sizeof the transformed complexes, 12 or

24mer, depends on the mass andconcentration of substrate. Moreover,we could also show that complexformation is transient with DegPrecycling into the hexameric state aftersubstrates are chopped up. In contrastto complexes with unfolded modelsubstrates, complexes with co-purifiedouter membrane proteins (OMPs) wereremarkably stable. Interestingly, DegPselectively stabilised folded OMPprotomers, whereas unfoldedmembrane proteins were immediatelydegraded. The role of DegP insafeguarding OMP precursors throughthe periplasm was confirmed by thephenotype of a degP deletion strain,where OMP targeting to the outermembrane was significantly impaired.

We were able to crystallise theDegP24/OMP(SeMet) complex andperformed single anomalousdispersion (SAD) experiments atbeamlines ID14-4 and ID23-1. In theresulting structure (Figure 71), eightDegP trimers are located at thevertices of a cube yielding a particlewith 432 symmetry. Most importantly,oligomer assembly is correlated withthe removal of the inhibitory loop LAfrom the active site yielding a fullyfunctional DegP protease.Unfortunately, we could not observethe bound OMPs in our structure, mostlikely due to positional and chemicalheterogeneity. However, cryo electronmicroscopy (carried out by E. Schäferand H. Saibil, Birkbeck College,London) of the DegP12/OMP complexrevealed that OMPs are bound -presumably in a folded state - in theinner cavity of DegP thus confirmingour biochemical data.

Taken together, our work has revealeda novel regulatory mechanism forproteases that is based on substrate-induced oligomer assembly and hasprovided molecular insight into theantagonistic protease and chaperoneactivities of DegP. While encapsulationof folded OMP protomers is protectiveand might allow safe transit throughthe periplasm, misfolded proteins areeliminated in the proteolytic foldingchamber.

Fig. 71: Comparison of DegP6 and DegP24.The colour code is the same for the twomolecules: protease domains (green),PDZ1 (orange), PDZ2 (red).



Structure-function analysis of a bacterialtyrosine kinase involved in capsule assembly

In bacteria, some polysaccharides co-polymerases involved in capsulesynthesis have been identified asprotein-kinases belonging to a newfamily of tyrosine-kinases and calledBY-kinases. They show no sequencesimilarity with eukaryotic protein-kinases. BY-kinases contain Walker Aand B nucleotide-binding motifscharacteristic of the so-called P-loopproteins. They undergoautophosphorylation on a C-terminaltyrosine cluster and furtherphosphorylate endogenous proteinsubstrates. In proteobacteria, BY-kinases are transmembrane proteinscontaining a periplasmic domain anda cytoplasmic catalytic domain. Infirmicutes, BY-kinases are encoded bytwo adjacent genes and split in twopolypeptides, which correspond eachto one of these two domains.

In this work, we focused on the BY-kinase CapA/CapB ofStaphylococcus aureus, a pathogenresponsible for a diverse spectrum ofanimal and human diseases thatpredominates among clinical isolates.The interaction between thecytoplasmic protein-kinase CapB andthe C-terminal juxtamembranefragment of the membrane proteinCapA (CapA-Ct) is required to promoteCapB activity. Here we present thecrystal structures of the chimericalprotein CapA-Ct/CapB (CapAB) and ofits inactive P-loop mutantCapAB(K55M).

Non-denaturing electrophoresis andmass spectrometry analysisdemonstrated that the purified CapABprotein is autophosphorylated.It crystallised in 20% PEG 1000 andhigh-resolution diffraction data (1.8 Å)were collected at beamline ID23-2.The CapAB structure was solved bymolecular replacement using the P-loop ATPase MinD as starting model.It is monomeric, with CapB harbouringthe typical α/β-fold of P-loop proteins.The phosphorylated C-terminal tyrosinecluster of CapB is disordered. CapA-Ctforms α-helix followed by a β-strandthat completes the central β-sheet of

CapB. CapA-Ct is involved in nucleotidebinding via its penultimate residuePhe221a, which participates in astacking interaction with the purinering of the ADP molecule found closeto the P-loop of CapB (Figure 72).

This explains the stimulatory effect ofCapA-Ct on the kinase activity of CapB.

The structure of the inactive P-loopmutant CapAB(K55M) was solved atbeamline ID29. It presents a subunit foldhighly similar to that of the CapAB withan rmsd of 0.67 Å over 241 alignedresidues. The major difference concernsits quaternary structure. Surprisingly,the unphosphorylated protein forms aring-shaped octamer (Figure 73) with

Principal publication andauthorsV. Olivares-Illana (a), P. Meyer (a),

E. Bechet (b), V. Gueguen-

Chaignon (a), D. Soulat (b),

S. Lazereg-Riquier (c),

I. Mijakovic (d), J. Deutscher (e),

A. Cozzone (b), O. Laprévote (c),

S. Morera (a), C. Grangeasse (b)

and S. Nessler (a), PLoS Biol. 6,

e143 (2008).

(a) LEBS, CNRS UPR 3082, Gif sur

Yvette (France)

(b) IBCP, UMR 5086 CNRS,

Université de Lyon (France)

(c) ICSN, CNRS UPR 2301, Gif sur

Yvette (France)

(d) Center for Microbial

Biotechnology, DTU, Lyngby

(Denmark)

(e) LMGM, AgroParisTech, CNRS,

INRA, Thiverval-Grignon (France)

Fig. 72: A view of the active siteof the S. aureus BY-kinase mutantCapAB(K55M). The P-loop of theblue CapB-1 is shown in orangewith bound ADP-Mg in yellow. TheCapA-Ct activator domain is shownin magenta. F221a involved innucleotide binding is highlightedas are the phosphorylatable Y225from the C-terminal tyrosinecluster (red) of a neighbouringCapB-2 subunit (green).

Fig. 73: Ring-shaped octamerof the S. aureusBY-kinase mutantCapAB(K55M).CapB subunitsare alternativelycolored green

and beige withthe C-terminal

tyrosine clustershighlighted in pink.

The CapA-Ct activatordomains are shown in

blue. Bound nucleotides areshown as black sticks.


62

Structural Biology

EE SS RR FF

the C-terminal tyrosine cluster fromeach molecule inserted into the activesite of a neighbouring one (Figure 72).The last tyrosine residue in the cluster,Tyr225, points towards the phosphatetail of the bound ADP and interactswith the catalytic Asp79 that is in aposition to deprotonate the Tyr225hydroxyl group, a prerequisite for itsphosphorylation.

In the octamer, the buried surfaceareas between neighbouring CapABsubunits (more than 2000 A2) aretypical of those required for biologicalinteractions. Moreover, when acomparison is made with the other BY-kinase sequences, the absoluteconservation of the residues involvedin monomer-monomer interactionsclearly demonstrates that the octameris physiologically important.Our structural analysis clearly showsthat BY-kinase autophosphorylationproceeds via an intermolecularmechanism. Our data further suggest

that the cytoplasmic kinase domainsdissociate from each other uponphosphorylation, inducingconformational changes transmitted tothe other components of the capsulesynthesis machinery via the associatedtransmembrane activation domain.In this model, cyclicphosphorylation/dephosphorylation ofthe BY-kinase seems to be intimatelycorrelated to its polysaccharide co-polymerase activity.

This work represents the basis forstructure-based drug design projectsaiming at blocking bacterial capsuleproduction. In S. aureus, but also inother bacterial pathogens, capsularpolysaccharides are considered asmajor virulence factors. Consideringthe idiosyncratic nature of BY-kinases,which are not present in eukaryoticorganisms, these enzymes representpromising targets for the design ofdrugs with limited side effects onhost cells.

SIRP structures explain paired-receptorspecificity

The complex immune system ofvertebrates must be tightly controlledin order to respond to pathogenicchallenge without inducingunnecessary side effects. While solublemarkers such as cytokines enableimmune cells to localise to the site ofinfection, specific protein:proteininteractions formed between cells fine-tune the immunological response.

The signal-regulatory proteins (SIRPs)are a family of cell-surface proteinsexpressed primarily on myeloid cellscomprising three members: SIRPα,SIRPβ and SIRPγ [1]. While theextracellular regions of all three arevery similar, the intracellular regions ofthese proteins are not alike and theyconvey very different signals to cells.SIRPα recognises CD47, a proteinexpressed on the surface of most cellsthat acts as a molecular ‘marker ofself’, and formation of this CD47/SIRPαinteraction inhibits engulfment of cellsby macrophages. SIRPβ, on the otherhand, does not recognise CD47 and

interaction of SIRPβ with its (unknown)extracellular ligand stimulatesmacrophage engulfment. While SIRPγrecognises CD47 with lower affinitythan SIRPα, it is expressed on thesurface of T-cells rather thanmacrophages and is thought to haveno direct signalling function. The SIRPsthus typify the class of membraneprotein families called “pairedreceptors”, which comprise severalproteins with similar extracellularregions but radically differenttransmembrane/cytoplasmic regionswith distinct (activating, inhibitory orneutral) signalling potentials.

The extracellular regions of SIRPscomprise three immunoglobulin (Ig)domains in a linear array. We hadpreviously determined the structure ofthe first (amino-terminal) Ig domain ofSIRPα [2]. Using site directedmutagenesis we showed that theinteraction with CD47 was mediated bythe N-terminal (apical) loops of SIRPα,more closely resembling how antibodies

Principal publication andauthorsD. Hatherley, S.C. Graham,

J. Turner, K. Harlos, D.I. Stuart,

A.N. Barclay, Mol. Cell 31, 266

(2008).

University of Oxford (UK)

recognise antigens rather than generalIg-domain mediated cell:cell contacts.

To determine the precise mechanismby which SIRPα recognises its ligand,we solved the structure of the N-terminal SIRPα domain in complexwith the sole extracellular Ig domainof CD47 in two crystal forms bymolecular replacement using datacollected at BM14 and ID23-2.Figure 74 shows the CD47/SIRPαinterface, which is unlike those seenfor other moderate-affinity Ig domainmediated cell-surface interactions as itis highly convoluted, well fitting, andextensive, with a mutual involvementof loops from the amino-terminal endsof both SIRPα and CD47.

In contrast to SIRPα, SIRPβ and SIRPγbind CD47 weakly or not at all, despitesharing > 90% sequence identitywith SIRPα. We solved structures of theN-terminal domains of SIRPγ and twoisoforms of SIRPβ by molecularreplacement using data collected atBM14. These structures demonstratethe exquisite selectivity of the SIRPinteraction interface. Overall, thestructures of SIRPs α, β and γ are verysimilar (Figure 75). However, for bothisoforms of SIRPβ specific amino aciddifferences with respect to SIRPα couldbe identified which result in theinability of SIRPβ to bind CD47.Further, by introducing a single aminoacid mutation in the sequence of oneisoform of SIRPβ we were able toconfer the ability to bind CD47 with anaffinity approaching that of SIRPα.

The N-terminal ligand-binding domainof SIRPα has much higher sequencevariability than most other cell-surfacereceptors. Mapping this sequencevariability onto the structure of theCD47/SIRPα complex shows that mostSIRPα polymorphisms reside away fromthe interaction interface (Figure 75),implying that the sequence variabilitydoes not modulate the strength of theCD47 interaction. We propose that thedriving force for this sequencevariability might be to avoidrecognition of SIRPα by pathogens,which would exploit the ability of SIRPαto down-regulate macrophage activity.An extension of the idea that SIRPαmight be a target for pathogens is thatSIRPβ may have evolved to mimic the

extracellular regions of SIRPα, its similarstructure but opposite (activating)signalling activity serving to frustrateattempts by pathogens to exploit theinhibitory potential of SIRPα.

The structure of SIRPα in complex withCD47, together with the structures ofSIRPβ and SIRPγ in isolation, has shownus in molecular detail how this pairedreceptor binds its ligand and how



Fig. 74: The structure of human SIRPα in complexwith CD47. A schematic representation of the fiveCD47 transmembrane helices is shown.

Fig. 75: a) Superposition ofSIRPβ and SIRPγ on theCD47/SIRPα complex showsthat the SIRPs have verysimilar structures. However,only SIRPα binds CD47 withhigh affinity. b) Sites ofSIRPα amino acidpolymorphism (red spheres)partition away from theCD47/SIRPα interactioninterface.


64

Structural Biology

EE SS RR FF

Atomic structure of a molecular switchfor blood coagulation and fibrinolysis

Principal publication andauthorsL. Sanglas (a), Z. Valnickova (b),

J.L. Arolas (c), I. Pallarés (a),

T. Guevara (c), M. Solà (a),

T. Kristensen (b), J.J. Enghild (b),

F.X. Aviles (a) and F.X. Gomis-

Rüth (c). Mol. Cell 31, 598 (2008).

(a) Institut de Biotecnologia i de

Biomedicina and Departament de

Bioquímica i Biologia Molecular,

Universitat Autònoma de

Barcelona, Bellaterra (Spain)

(b) Center for Insoluble Protein

Structures at the Department of

Molecular Biology, University of

Århus (Denmark)

(c) Department of Structural

Biology at the Molecular Biology

Institute of Barcelona, CSIC (Spain).

subtle amino acid variations in theligand-binding regions account for theexquisite ligand selectivity of the SIRPfamily. The partitioning of SIRPαpolymorphisms away from the CD47interaction interface suggests thatpathogenic challenge may drive SIRPαpolymorphism. Pathogen pressure on

the inhibitory receptor provides aplausible raison d’être for theexistence of paired receptors, withtheir similar extracellular structuresbut opposing signalling potentials.

References[1] A.N. Barclay, M.H. Brown, Nat.

Rev. Immunol. 6, 457 (2006).

[2] D. Hatherley, K. Harlos,

D.C. Dunlop, D.I. Stuart,

A.N. Barclay, J. Biol. Chem. 282,

14567 (2007).

Fig. 76: Structure of the TAFIa/TCI complex. a) Richardson plot ofthe complex showing TAFIa in standard orientation. The region of

the proposed fibrinolysis switch is shown over gray background.TCI is coloured as follows: N-terminal domain (light blue), linker(red), C-terminal domain (purple). b) Close-up view of (a) after a

vertical rotation of ~90º showing the proposed fibrinolysis switchregion of TAFIa (yellow) superimposed with the equivalent region

of human pancreatic CPB1, which is taken as a highly-stablereference (green).

After blood-vessel injury, hemostasisinduces fibrin clot formation to preventblood loss and trigger vessel repair. Thisclot must be removed after tissue repairto restore blood flow. These twoprocesses are tightly regulated by thecoagulation and fibrinolytic cascadesbecause imbalance may lead tothrombosis, heart attack and stroke, orto bleeding as in hemophilia [1].Hemostasis is modulated by thrombin-activatable fibrinolysis inhibitor (TAFI).It attenuates fibrinolysis byremoving surface-exposed C-terminallysine residues from the fibrin clot.TAFI is the zymogen of a B-type zinc-dependentmetallocarboxypeptidaseof the funnelin tribe ofproteases [2,3] and it is theonly funnelin found inhuman plasma. Duringcoagulation, TAFI isprocessed by thrombin/thrombomodulin to TAFIathrough removal of itspro-domain. In contrastto pancreatic funnelins,both TAFI and TAFIa

uniquely display carboxypeptidaseactivity against larger substrates.However, while TAFIa has a half-life ofless than ten minutes, TAFI is stable incirculation [4,5].

We obtained a stable form of TAFIa invitro employing purified bovine TAFI,which was activated in the presence ofa proteinaceous inhibitor, TCI, fromthe tick. This produced a stableenzyme-inhibitor complex.



Assembly and channel-opening in abacterial drug efflux machine

Resistance to drugs remains a majorobstacle in the treatment of infectionscaused by pathogenic bacteria. Onedrug resistance mechanism originatesfrom the activity of transmembranetransporters that displace a broadrange of antibiotics and other toxiccompounds from the bacterium. In thecase of Gram-negative bacteria, thetoxic compound may need to bedisplaced across both the inner- andouter-membranes and the interstitialperiplasm in order to be properlyexpelled from the cell. A special class ofthree-component efflux pumpsaccomplishes this task. Typically, thesepumps are composed of an innermembrane protein, an outer membraneprotein, and a protein that traversesthe periplasm to link the twomembrane proteins.

A well studied representative of thefamily of multi-drug efflux pumps is theEscherichia coli AcrB/AcrA/TolCassembly. Crystal structures areavailable for each of the threecomponents of the tripartite assembly[1, 2 and references cited therein], butthe structure of the ternary assembly ispresently unknown. Accumulatingstructural and functional data providesa rough idea of the architecture of thepump, and Figure 77 summarises thesein a speculative model for the entireassembly based on docking the crystalstructures of the individualcomponents. The trimeric AcrB is theinner membrane component, and it

powers the efflux pump using theenergy from proton electrochemicalgradients. The trimeric TolC is the outermembrane protein, which is shaped likea cylinder with a hollow interior. Oneend of the cylinder is open to theextracellular medium while the othertapers towards its periplasmic end,where it is maintained in a closed stateby salt bridges and van der Waalscontacts between the adjacentsubunits. It seems clear that transport

Principal publication andauthorsV.N. Bavro (a), Z. Pietras (a),

N. Furnham (a), L. Pérez-Cano(b),

J. Fernández-Recio (b),

X. Yuan Pei (a), R. Misra (c) and

B. Luisi (a), Mol. Cell. 30, 114

(2008).

(a) Department of Biochemistry,

University of Cambridge (UK)

(b) Barcelona Supercomputing

Center (Spain)

(c) Arizona State University,

Tempe (USA)

We were able to solve the crystalstructure to high resolution usingsynchrotron radiation at beamlineID23-2. These studies revealed thatTAFIa conforms to the α/β-hydrolasefold of funnelins and that it displaystwo unique flexible loops on themolecular surface, which account forstructural instability and susceptibilityto proteolysis (Figure 76). In addition,point mutations reported to enhanceprotein stability in vivo are mainlylocated in the first loop and in anothersurface region, which is a potential

heparin-binding site. The proteininhibitor contacts both the TAFIaactive site and an exosite, thuscontributing to high inhibitoryefficiency.

Fig. 77: A cartoon schematicshowing a putative model of theAcrB/AcrA/TolC drug-effluxassembly. Crystal structures of theisolated components were dockedto model the assembly. The TolCtrimer is shown in blue, AcrBtrimer in green, three AcrAmolecules are coloured yellow.White sphere layers represent thetwo surfaces of each membrane;the two upper and two lower arefrom the outer membrane andinner membrane respectively.While the model presented heredepicts a 3:3:3 assembly, othermodels, with different subunitcompositions cannot be ruled outby the available data.

References[1] M.B. Boffa, M.E. Nesheim and M.L. Koschinsky, Curr. Drug Targets Cardiovasc.

Haematol. Disord. 1, 59 (2001).

[2] J.L. Arolas, J. Vendrell, F.X. Avilés and L.D. Fricker, Curr. Pharm. Des. 13, 349 (2007).

[3] F.X. Gomis-Rüth, Crit. Rev. Biochem. Mol. Biol. 43, 319 (2008).

[4] M.B. Boffa, W. Wang, L. Bajzar and M.E. Nesheim, J. Biol. Chem. 273, 2127 (1998).

[5] Z. Valnickova, I.B. Thogersen, J. Potempa and J.J. Enghild, J. Biol. Chem. 282, 3066 (2007).


66

Structural Biology

EE SS RR FF

requires TolC to switch to an openconformation to allow transportationof the substrate.

In our study, we mutated TolC in orderto disrupt a network of hydrogenbonding interactions at the periplasmicend that caused it to become leaky tobulky molecules in vivo. However, themutations did not affect TolC function,since it retained the capacity to form afunctional efflux assembly. We solvedthe crystal structure of the mutant intwo crystalline forms at 3.2 and 3.3 Åresolution using data collected atbeamline ID23-2. The crystal structuresreveal that three of the six pairs ofcoiled coils at the closed end of TolCmove radially outward from the centralmolecular axis to partially open thechannel (Figure 78). Conformationalchanges associated with partial channel

opening expose grooves which, wespeculated, can accommodateperiplasmic component AcrA. Wedeveloped a model of the interactionsbetween AcrA and TolC, whichsuccessfully predicted a mutation pairin the proteins that compensate eachother’s disruptive effects. Using dockingcalculations, we generated a model inwhich the open form of TolC engagesthe periplasmic crown of AcrB, theinner membrane protein.

Despite the dramatic changes in theAcrB during its work cycle [1,2], theTolC-interacting interface of AcrBchanges little, suggesting that thisinterface might only partially transmitthe conformational changes thattrigger the opening of the TolCchannel, and is likely only implicated inunlocking the periplasmic end of thechannel, while additional energy isneeded to drive the channel to fullyopen state. This suggests that AcrA ismore than just a passive bridgebetween the energised AcrB and TolC,but likely also serves as an activetransducer of energy from one to theother. The induced fit of theperiplasmic partner into the partiallyopened grooves of TolC may completethe process of channel opening.

Fig. 78: Crystal structures of theopen and closed states of the TolCchannel protein. The view is along

the three-fold axis of the TolCtrimer at the periplasmic end that

engages AcrB.

References[1] S. Murakami, R. Nakashima,

E. Yamashita, T. Matsumoto and

A. Yamaguchi, Nature 443, 173

(2006).

[2] M.A. Seeger, A. Schiefner,

T. Eicher, F. Verrey, K. Diederichs

and K.M. Pos, Science 313, 1295

(2006).

Structure of a “bonsai” Ndc80 complexsheds light on the mechanism ofchromosome segregation

Accurate chromosome segregation is aprerequisite for long-term cell andorganism survival. To ensure this, thereplicated chromosomes (known assister chromatids) establish sturdyconnections with a structure known asthe mitotic spindle. Microtubules,which are polymeric structures basedon the assembly of αβ-tubulin dimers,are the main component of thespindle. During the process of celldivision, chromosomes assemble andexpose a microtubule receptor knownas the kinetochore. The crucialmicrotubule receptor at kinetochoresis a 4-protein complex known as theNdc80 complex. Our studies haveprovided a molecular understanding of

the organisation of the Ndc80 complexand of its ability to bind microtubules.

The Ndc80 tetramer is assembled fromtwo dimers, Ndc80:Nuf2 andSpc24:Spc25 (Figure 79a). The topologyof interactions of these two dimers inthe Ndc80 complex was known fromprevious studies. Ndc80:Nuf2 haveglobular N-terminal domains and arepredicted to form a long parallelcoiled-coil with their C-terminalregions. The Ndc80:Nuf2 coiled-coilends in a tetramerisation domainwhere the Ndc80:Nuf2 dimer meetsthe Spc24:Spc25 dimer. TheSpc24:Spc25 dimer contains anotherdimeric parallel coiled-coil that

Principal publication andauthorsC. Ciferri (a), S. Pasqualato (a),

E. Screpanti (a), G. Varetti (a),

S. Santaguida (a), G. Dos Reis (a),

A. Maiolica (a), J. Polka (b), J.G. De

Luca (c), P. De Wulf (a), M. Salek

(d), J. Rappsilber (e), C.A. Moores

(f), E.D. Salmon (b) and A.

Musacchio (a), Cell 133, 427 (2008).

(a) Department of Experimental

Oncology, European Institute of

Oncology, Milan, (Italy)

(b) Department of Biology,

University of North Carolina (USA)

(c) Dept of Biochemistry and

Molecular Biology, Colorado State

University (USA)

(d) Sir William Dunn Pathology

School, Oxford (UK)

(e) Wellcome Trust Centre for Cell

Biology, University of Edinburgh

(UK)

(f) School of Crystallography,

Birkbeck College, University of

London (UK)

engages the N-terminal regions of theSpc24:Spc25 subunits, and ends with aglobular region. This organisation ofthe Ndc80 complex results in a veryelongated and rather flexiblestructure, as shown by low-resolutionelectron micrographs on negativelystained samples. The globular regionsat either end of the rod have beenshown to mediate microtubule bindingand kinetochore localisation.

Highly elongated and flexiblestructures are almost invariablydifficult to crystallise. Thus, to gainhigh-resolution structural insight intothe Ndc80 complex, we made anattempt to simplify its organisation.Because the topology of the Ndc80complex suggested that the C-terminalmoieties of the Ndc80:Nuf2 subunitsare close to the N-terminal moieties ofthe Spc24:Spc25 subunits, we tried tofuse the chains two on two to create adimeric strucure (Figure 79b). We thenremoved most of the interveningcoiled-coil region to create a BonsaiNdc80 complex containing only the“leaves” (the globular domains ofNdc80 and Nuf2), a small segment ofthe “trunk” (the coiled-coil), and the“roots” (the globular region of theSpc24:Spc25 dimer. The topology ofthis construct is shown in Figure 79b.The Ndc80bonsai construct retains theability to bind microtubules andkinetochores.

Bonsai-Ndc80 proved to be very stableand crystallised readily. Using X-raydiffraction data collected on beamlinesBM16 and ID23-1 we were able todetermine the 2.8 Å crystal structureof the Bonsai-Ndc80 complex(Figure 80a). The structure revealedthat the globular domains of Ndc80and Nuf2 each contain a calponin-homology (CH) domain (Figure 80b).CH domains have been previouslyimplicated in binding actin ormicrotubules. In the Ndc80:Nuf2moiety, the CH domains are tightlypacked and form a single largeexposed surface. We have carried outan extensive mutagenesis analysis todemonstrate that this surface providesthe main microtubule-binding area ofthe Ndc80 complex. Based on thestructure of Bonsai-Ndc80 and of cross-linking and mass spectrometry data,we have generated a model for the

full-length Ndc80 complex andrevealed the molecular basis ofchromosome segregation, a processthat is essential for the propagation ofgenetic information along successivegenerations. Our current analyses aimat understanding the broaderorganisation of multiple Ndc80complexes around microtubules.



Fig. 79: a) The Ndc80 complex is a tetrameric arrangement of Ndc80, Nuf2, Spc25 andSpc24 subunits. Ndc80:Nuf2 and Spc25.Spc24 form dimeric arrangements containingmicrotubule-binding and kinetochore-binding domains, respectively. Overall, thecomplex is dumb-bell shaped consisting of two globular domains separated by acentral coiled-coil rod. b) Bonsai version of the Ndc80 complex. The chains have beenfused two-on-two and most of the coiled coil domain has been removed.

Fig. 80: a) Ribbon model of Bonsai-Ndc80. b) Superposition of the calponin-homology(CH) domains of Ndc80 (cyan) and Nuf2 (yellow). c) Map of evolutionary residueconservation on the surface of the Ndc80 complex identifies a microtubule-bindinginterface (left).


68

Structural Biology

EE SS RR FF

Sorting SNAREs into clathrin-coatedvesicles

The protein and phospholipidcomposition of cellular membranesdefines the identity of that membrane;what binds to it and what reactionsoccur both on its surface and insidethe cavity it surrounds. Components ofthe cell’s limiting membrane areinvolved in cell signalling, homeostasis,cell contacts and recognition by theimmune system. Glycosylation,proteolytic processing and degradationof transmembrane and luminalproteins occur within cells’ organelles.

Transmembrane proteins with anenormous variety of functions aremoved between cell membranes intransport vesicles. Once cargo has beensorted into a forming vesicle, the vesiclebuds from the donor membrane and isthen transported to and fuses with thetarget membrane. The directionalityand specificity of this transport isregulated to a large extent by membersof a family of small type IItransmembrane proteins called SNAREsthat are found in both vesicle andtarget organelle membranes.After successful formation of a trans-four-helical bundle betweenhomologous helices from a SNARE inone membrane and three SNAREs inthe other, fusion of the vesicle with thetarget compartment can proceed. Thespecificity of SNARE complex formation,which leads to the specificity ofvesicle/target membrane fusion, arisesfrom the fact that only very limitedcombinations of SNAREs can formproductive SNARE complexes.The correct distribution and localisation

of SNARE proteins are therefore critical:vesicles must contain sufficient SNAREsthat can engage in complex formationat the target membrane as well asSNAREs that are being recycled to theirdonor compartment, most likely in aninactive form.

Post-Golgi transport is mainlymediated by clathrin-coated vesicles(CCVs), whose coats are composed ofan outer clathrin scaffold linked to themembrane by clathrin adaptors.Clathrin adaptors have folded domainsthat bind to short-lived membranecomponents such as phospholipidheadgroups and the GTP-bound formsof small GTPase proteins of the Arfsuperfamily and extended, flexibleregions that contain multiple shortmotifs, which bind to clathrin andother CCV components. At least one ofa very small number of transplantable‘tags’, a short linear sequence motif, oralternatively a covalently attachedubiquitin molecule, are found onnearly all transmembrane proteins andbind to the folded domains of themost common clathrin adaptors, somarking the transmembrane proteinfor incorporation as general cargo intoCCVs [1]. The mechanisms by whichSNAREs are selected for incorporationinto CCVs were unclear, although a lotwas known about how SNAREs wereincorporated into ER to Golgi COPII-coated vesicles.

Recent structural work carried out atbeamline ID14-4 has shed light on thisissue (Figure 81). The SNARE VAMP7

Principal publication andauthorsP.R. Pryor (a), L. Jackson (b),

S.R. Gray (a), M.A. Edeling (a),

A. Thompson (c),

C.M. Sanderson (c,d), P.R. Evans (b),

D.J. Owen (a), J.P. Luzio (a), Cell

134, 817-827 (2008).

(a) Cambridge Institute for

Medical Research and

Department of Clinical

Biochemistry, University of

Cambridge, Addenbrooke’s

Hospital, Cambridge (UK)

(b) Medical Research Council

Laboratory of Molecular Biology,

Cambridge (UK)

(c) Medical Research Council

Rosalind Franklin Centre for

Genomics Research, Cambridge

(UK)

(d) The Physiological Laboratory,

School of Biomedical Sciences,

University of Liverpool (UK)

Fig. 81: a) Structure of the VAMP7 (purple):Hrb (gold) complex with key residues, whosemutation abolishes the interaction betweenthe two proteins in vitro, highlighted in red.b) tagged VAMP7 (red) can be internalisedfrom the cell surface to lysosomes and lateendosomes (green) resulting in completeoverlap of the two as indicated by theresulting yellow punctae. c) A version ofVAMP7 harbouring mutations in the keyVAMP7 residues Leucine43 and Tyrosine45(red) cannot bind to Hrb and so cannot beinternalised and builds up on the cell surface.



needs to be retrieved from the plasmamembrane following its use in thefusion of late endosomes andlysosomes directly with the cell’slimiting membrane, a process thoughtto be important in delivering newmembrane to the cell surface duringmembrane repair or mitosis. Thisretrieval occurs through theinteraction of VAMP7 with the lowabundance clathrin adaptor Hrb. Thestructure of a complex between partsof VAMP7 and Hrb showed that a 20residue stretch of largely hydrophobicamino acids in Hrb wraps around agroove in the folded longin domain ofVAMP7 that preceeds its SNARE helix.Mutation of key residues in theVAMP7/Hrb interface that wereidentified from the structure of theVAMP7/Hrb complex, abolish theinteraction between the two proteins

both in vitro and in vivo where theyresult in altered trafficking of VAMP7.

Taken together with other recentstudies [2,3], these data on theVAMP7/Hrb interaction suggest thattwo parallel pathways operate inpost-Golgi CCVs for cargo selection:one for selecting general cargo that ismediated by the recognition of a fewwidely used tags, and the other forselecting the necessary SNAREs throughthe recognition of determinants thatare each found on the surface of only asingle SNARE. By having two non-competing cargo selection systemsoperating simultaneously, the cellensures that CCVs incorporate sufficientamounts of the right SNAREs to allowfusion of the vesicle with its correcttarget organelle as well as standard‘bulk’ cargo.

References[1] J.S. Bonifacino and L.M. Traub,

Annu Rev Biochem 72, 395 (2003).

[2] S. Martinez-Arca, R. Rudge,

M. Vacca, G. Raposo, J. Camonis,

V. Proux-Gillardeaux, L. Daviet,

E. Formstecher, A. Hamburger,

F. Filippini, et al., PNAS 100, 9011

(2003).

[3] S.E. Miller, B.M. Collins,

A.J. McCoy, M.S. Robinson, and

D.J. Owen, Nature 450, 570

(2007).

Principal publication andauthorsV. Adam (a), M. Lelimousin (b),

S. Boehme (c), G. Desfonds (a),

K. Nienhaus (c), M.J. Field (b),

J. Wiedenmann (d),

S. McSweeney (a),

G.U. Nienhaus (c) and

D. Bourgeois (a,b), PNAS 105,

18343 (2008).

(a) ESRF

(b) Institut de Biologie Structurale

Jean-Pierre Ebel, Commissariat à

l’Energie Atomique, Centre

National de la Recherche

Scientifique, Université Joseph

Fourier, Grenoble (France)

(c) Institute of Biophysics,

University of Ulm (Germany)

(d) National Oceanography

Centre, University of

Southampton (UK)

Structural characterisation of IrisFP, abiological photoswitch from the greenfluorescent protein family

Since the successful cloning of thegreen fluorescent protein (GFP) fromthe jellyfish Aequorea victoria in 1992,fluorescent proteins (FPs) haverevolutionised Life Science. The 2008Nobel Prize in chemistry was awardedto O. Shimomura, M. Chalfie andR.Y. Tsien for having, respectively,discovered GFP, introduced it in cells asa luminous genetic tag, and refining itto a palette of ever brighterfluorescent proteins with variouscolours. Using the fluorescent proteintoolkit, cell biologists can image a verybroad range of biological processes,including gene expression, proteintranslocation within cells, and cellmovement during development [1].

In recent years, new fluorescentproteins have been identified, notablyin Anthozoa species (e.g. corals andsea anemones found in tropical reefs),which display photo-activationproperties [2]. Proteins from the“Kaede-family” undergo a photo-conversion from green to red uponillumination with UV light. Proteinsfrom the “Dronpa-family” undergo

on/off photo-switching uponillumination with UV/blue light. Theuse of photo-activatable fluorescentproteins provided new means todynamically track proteins within livecells [3], and to develop “nanoscopy”,a revolutionary set of techniques thatdefeat the diffraction barrier in lightmicroscopy, achieving spatialresolution of ~10 nm with visible light[4]. Moreover, photo-activatablefluorescent proteins may findapplications in nano-biotechnology,for example as novel data storagemedia [5].

All fluorescent proteins discovered sofar display a so-called β-can structure,forming an almost perfect barrelcomposed of eleven β-strands.The chromophore is embedded at theheart of the β-can, and is made of3 amino-acids (X-Tyr-Gly) that undergopost-translational modifications toform an extended π-electron system.Using X-ray crystallography, UV-visspectroscopy and quantum chemistry,mechanistic details of the fluorescencemechanisms employed by photo-


70

Structural Biology

EE SS RR FF

Fig. 82: The X-raystructure of IrisFP is

shown at thecentre. Around the

structure arepictures of crystals

of IrisFP in itsdifferent forms

(switched on/off,green/red),

recorded whilefluorescing.

activatable fluorescent proteins can beobtained, which can be ofconsiderable value to rationallyengineer new variants with improvedproperties. In a collaboration betweenthe ESRF, University of Ulm and theIBS, we have developed such a

complementary approach. In thecourse of this project, we discovered“IrisFP” (the Greek goddess Irispersonifies the rainbow), a remarkablefluorescent protein that derives fromthe fluorescent protein EosFP [6] andcombines the properties of Kaede-likeand Dronpa-like proteins, i.e. photo-conversion and photo-switching(Figure 82). Following proper laserillumination schemes, IrisFP can beswitched reversibly betweenfluorescent and non-fluorescent formsin its green state, then photo-converted irreversibly from green tored, and finally switched reversiblybetween fluorescent and non-fluorescent forms in its red state.

We undertook a structuralcharacterisation of IrisFP, in particularby using the Cryobench laboratory atthe ESRF, which allows the illuminationof crystals with a variety of lasers andthe following of the induced colourchanges by in crystallo spectroscopy.We discovered that photoswitching ofgreen IrisFP involves a cis-transisomerisation of the chromophore(Figure 83), accompanied by a majorconformational change of the protein.In contrast, photo-conversion to thered-emitting state does not involve alarge structural change, but resultsfrom a cleavage of the polypeptidebackbone which induces an extensionof the conjugated π-electron system ofthe chromophore. Photoswitching ofthe red form of IrisFP appeared similarto that of the green form, involvingcis-trans isomerisation of thechromophore.

IrisFP holds great promises forapplications in cell biology. Based onthe mechanistic knowledge acquired inthis work, rational engineering will beapplied to further improve thebehaviour of the protein. Applicationstaking advantage of the multiplephototransformations displayed byIrisFP will undoubtedly emerge,including two-colour nanoscopy forhigh-resolution spatio-temporalimaging, or sequential photo-activation schemes for unravellingcomplex protein-protein interactions.IrisFP also hints at the possibility ofcombining read-only and rewritablecapabilities in future mass storagemedia.

Fig. 83: Structural changes in IrisFP upon phototransformation. Structural motionsinduced by light are represented by curved arrows of the same colour as those used

to represent light illumination at specific wavelengths.

References[1] R.Y. Tsien, Annu. Rev. Biochem. 67, 509 (1998).

[2] R. Ando et al., PNAS 99, 12651 (2002); Science 306, 1371 (2004).

[3] K.A. Lukyanov et al., Molecular Cell Biology 6, 885 (2006).

[4] E. Betzig et al., Science 313, 1642 (2006).

[5] M. Andresen et al., PNAS 102, 13070 (2005).

[6] J. Wiedenmann et al., PNAS 101, 15905 (2004).



Introduction

With the ESRF Upgrade Programmelaunched in November, the startingsignal was given for developing concreteplans for the future upgrade andrefurbishment of the surface andinterface science (SIS) beamlines.Conceptual design reports are requiredfor the meeting of the Science andAdvisory Committee in spring 2009.ID01 may be transformed into a 130 mlong beamline for nanofocusing andcoherent diffraction imaging (CDI) andID03 may be relocated to anotherstraight section. ID32 will most likely notbe influenced immediately by theUpgrade Programme.

To explore the opinions and needs ofthe user community for the upgradedID01 beamline, a highly successfulbrainstorming meeting with morethan 30 participants was held at the ESRFin November. The diffraction analysisof MEMS/NEMS (micro/nano electromechanical systems) was recognised as abrand-new and exciting avenue ofresearch. With the novel ESRF in situAFM at ID01 (see Figure 84), the X-rayanalysis of mechanical interaction onthe sub-µm scale has already beendemonstrated, which may representanother prominent mission for the newID01 beamline. The other two SIS

beamlines had a busy time as well andimprovements at ID03 have continued.In the ID03 UHV chamber in EH2,cooling was improved (< 30 K) and a highprecision and rigid HEXAPOD for samplealignment with six degrees of freedomhas now been installed. In EH1, anew small-volume flow-through reactorfor in situ X-ray diffraction wascommissioned, which permits rapid gasexchange and thus allows surfacereactions to be followed at muchincreased speed (< sec). At ID32, theX-ray standing wave and photoelectronspectroscopy capabilities have beenfurther upgraded. A new UHV systemwith a (presently unique) photoelectronanalyser was installed in EH2. Theanalyser for HAXPES (hard X-rayphotoelectron spectroscopy) studies isequipped with a low-noise 2D delay-linedetector and, once fully commissioned,will permit the analysis of electronswith up to 15000 eV kinetic energywith very high resolution (nominallydown to ∆E/E = 10-6). Fresnel zone plates,installed in front of the existing XPSsystem, have been tested and can delivera beam as small as 3 µm for spatially-resolved XPS measurements.

Surface science is dead, yet long livesurface science since with the advent ofthe fashionable term nanoscience, manyfields of (classical) surface scienceunderwent some metamorphosis, but

Surface and Interface Science

Fig. 84: A monochromatic X-raybeam of ~ 9.9 keV is focused byberyllium lenses (CRL) onto a singleSiGe island while an atomic forcemicroscopy (AFM) tip is used toapply pressure. A two dimensionalimage of the scattering close tothe SiGe(004) Bragg peak isrecorded. The shift in intensity(marked by the green arrows)reflects the compression of theisland’s lattice parameter (seehttp://www.esrf.fr/news/spotlight/spotlight63).


72


EE SS RR FF

mostly in their labelling. However, a realshift in paradigm accompanied thechange in the name of our group fromsurface science to surface and interfacescience (SIS) some years ago. The changereflected two aspects: firstly, the termsurface for the truncated region of asolid suggest a purely two dimensionalnature while the region affected by thetruncation always extends into the thirddimension. Secondly, it is apparent thatthe trend of science becoming moreand more specialised is being reversedat present owing to the increasinginterest in more complex systems.Thus, even the classical surfacescience has become more and moremultidisciplinary. Surface scientists arelooking at what’s happening above andbelow where two agents meets, whetherthese are solid and vacuum, solid andsolid, solid and liquid, etc. Contact pointsor phase boundaries are areas ofinteraction, interchange, and reactionand affect the properties of the materialsinvolved. This is of interest to almost allfields in natural science and mostindustries. Noteworthy, the increasedinteraction extends to the scientistsinvolved as well.

This year’s SIS highlights are clearlytestifying toward multidisciplinaryinteraction. Stadler et al. revealed thatthe cohesion among organic moleculescan be tuned by the adsorption on ametal (silver) surface by using X-raystanding wave (XSW) studies withphotoelectron spectroscopy at ID32.Dealing with a similar subject, also onthe borderline between physics andchemistry, and likewise using the XSWmethod at ID32, Koch et al. find that thespecific chemical/electronic interaction ofmolecules with a metal surface can leadto pronounced molecular distortions.The contribution of Le Bolloc’h et al.correlates structure and electronicproperties for a subject of fundamentalphysics. An electronic excitation, acharge density wave, was studied bydiffraction at ID01, and the authorsdiscovered that it shows an intriguinglylong coherence length of the order ofµm. Making important progress towardthe future mission of ID01, a joint team

from DESY, IKZ Berlin, JKU Linz, and theESRF (Zozulya et al.) managed to recordCDI patterns from sub-µm small SiGeislands and to reconstruct their shapeand density by a model-free phaseretrieval approach.

While the last four contributionsanalysed in some detail some of theamazing features that Mother Nature isable to create on surfaces, in thefollowing reports the authors watchedover Mother Nature’s shoulder when inaction. Thus, Coati et al. observed thedetailed nano-faceting of vicinal Nisurfaces induced by Ag deposit. Nolte etal. observe an oxygen-induced shapechange of the Rh nanoparticles on aMgO(100) during oxidation and COinduced reduction that turns out to befully reversible. Finally, in a collaborativeeffort, a team from MPI Halle, APS/ANL,and the ESRF studied the mesoscopicrelaxations in cobalt nano-islands onCu(001) and found experimentalevidence for large contractions of theinteratomic cobalt distances. With somany excellent results coming from theSIS beamlines, a selection of a fewhighlights is always difficult. I hope youenjoy our choice and I wish you a happyreading.

J. Zegenhagen



Structure and electronic properties

Tuning intermolecular interaction in long-rangeordered sub-monolayer organic films

The future development of organicelectronic devices depends on theability of tailoring the properties ofthin-films and their interfaces with(metallic and insulating) substrates.The best results can be expected fromwell-ordered films with largecrystalline grains and “perfect”interfaces. However, their propertiesare dominantly influenced by theformation of the first molecular layeron the surface representing a templatefor further growth. The developmentof the first layer – in turn – dependson the fine balance of molecule-substrate and molecule-moleculeinteraction. The latter usually isattractive due to van-der-Waals forcesand causes formation of islands andsmall crystalline grains. Here we reporton an organic adsorbate exhibiting adifferent behaviour: Sn-Phthalocyaninemolecules show a repulsiveintermolecular interaction when theyare deposited on a Ag(111) surface invery thin films (coverage below onelayer). This ensures a homogeneousfilling of the surface terraces andhence the size of structural domains isonly limited by the terrace width.

The phase diagram for SnPc/Ag(111)shows three different regions(Figure 85). A disordered overlayer isformed at low coverages and roomtemperature (RT). Like in a gas, theaverage intermolecular distancereduces with increasing density. Thisbehaviour is proven by high-resolutionlow energy electron diffractionexperiments (SPA-LEED images areshown for all regions as insets inFigure 85). In the second region athigher coverage, a series of well-ordered phases occurs, the structuralparameters of which changecontinuously with coverage.No discrete phase transitions arefound in this regime, but a continuousrotation and deformation of thesuperstructure unit cell. This effect iscaused by a substrate-mediatedrepulsive inter-molecular interactionbetween the molecules exceeding

van-der-Waals and other attractiveforces. The local molecular density isminimised at all coverages. Below 0.92monolayers (ML), the repulsion can bereversibly switched to attraction bycooling. In this third region anadditional site-specific interactionbetween molecules and substrate over-compensates the repulsion and leadsto a rearrangement of the moleculesin a commensurate pattern.

The key to understanding repulsiveinteraction is the charge transferbetween molecule and substrate.From X-ray standing waves (XSW)measurements performed at beamlineID32, we found convincing indicationsfor a significant overlap of molecularorbitals with the silver electronicstates. The adsorption geometrydiffers for high and low coverages [1].While in the monolayer-structure,all molecules adsorb with their Sn-atom below the molecular plane(“tin-down” position, Figure 86a),whereas a mixed “tin-up”/“tin-down”structure occurs at lower coverages.The molecule-surface distance is onlyabout 3 Å, significantly below van-der-Waals bonding-distances. This indicateschemisorption and necessitates spatialoverlap of molecular orbitals andsurface electronic states.An exchange of electronic charge(donation/backdonation) is the

Principal publication andauthorsC. Stadler (a), S. Hansen (a),

I. Kröger (a), C. Kumpf (a,b) and

E. Umbach (a,c), Nature Physics,

DOI: 10.1038/nphys1176.

(a) Universität Würzburg,

Experimentelle Physik II, Am

Hubland, Würzburg (Germany)

(b) present address:

Forschungszentrum Jülich GmbH,

Jülich (Germany)

(c) present address:

Forschungszentrum Karlsruhe

GmbH, Karlsruhe (Germany)

Fig. 85: Phasediagram ofSnPc/Ag(111). Gas-likephases, a region withincommensuratestructures, and acommensurate phaseare identified.Typical SPA-LEEDpatterns and amodel of the Sn-Phthalocyaninemolecule are shownas insets. XSWexperiments wereperformed at thepoints “A”, “B”, “C”and “D”.


74


EE SS RR FF

consequence. Density functionaltheory (DFT) calculations andultraviolet photoelectron spectroscopy(UPS) confirmed this finding.

The donation/backdonation effect isalso illustrated in Figure 86. Electrondonation to the substrate just belowthe molecules (yellow area) displaces

Ag electrons to the vicinity (blue).Backdonation to unoccupied orbitalsof the molecule cannot completelycompensate for this effect. Thereforeneighbouring molecules – if they areidentically oriented in “tin-down”position and located close to eachother as is the case at high coverages(Figure 86a) – compete for thedonation/backdonation effect whichcauses repulsion between themolecules. At lower coverages, therepulsion is reduced by the mixed“tin-up”/“tin-down”-orientation, butat room temperature (disordered, gas-like phase) it still remains dominant.Only upon cooling does the influenceof the substrate overcome therepulsion, and an ordered structure isformed (Figure 86b).

Intermolecular repulsion in orderedorganic overlayers is reported for thefirst time. It leads to a novel growthmode based on a homogeneous fillingof surface terraces and results in largedomain sizes. Just by tuning thetemperature, the system can beswitched from repulsive to attractiveintermolecular interaction and back.

Fig. 86: Structural modeland illustration of the

donation/backdonationeffect. a) Side view of

the incommensuratemonolayer structure(“C” and “D” in the

phase diagram,coverage 1 ML). b) Sideand perspective view of

the commensurate LTstructure (“A”).

Molecular orbitals(HOMO-1 and LUMO)are drawn in red and

green, respectively.

Reference[1] C. Stadler, S. Hansen,

F. Pollinger, C. Kumpf, E. Umbach,

T.-L. Lee, and J. Zegenhagen,

Phys. Rev. B 74, 035404 (2006).

Observation of correlations in sliding charge-densitywaves

To observe an electronic motion hasalways been a difficult task. Forexample, the sliding motion of acharge density wave (CDW) under anelectric current has been observed onlyby a few experiments. Resistivitymeasurements were the firsttechniques able to observe thisphenomenon [1]. In the sliding regime,

the Ohm law is no longer fulfilled anda characteristic electronic noiseappears. However, CDW systems, suchas blue bronze, are rarelyhomogeneous and such macroscopicmeasurements fail to give informationat the atomic scale.

X-rays could in principle bring newinformation since the periodicdisplacement associated with the CDWcan be easily measured by diffraction,leading to 2kF satellite reflections,where kf represents the Fermi wavevector. Unfortunately, the loss of thephase prevents the observation oftranslation of a whole system bydiffraction. Nevertheless, a fewconsequences of the slidingphenomenon have been measured.When the wave slides, the domain sizedecreases along the direction

Principal publication andauthorsD. Le Bolloc’h (a),

V.L. Jacques (a,b), N. Kirova (a),

J. Dumas (b), S. Ravy (c),

J. Marcus (b), and F. Livet (d),

Phys. Rev. Lett. 100, 096403

(2008).

(a) Laboratoire de Physique des

Solides bât 510, CNRS, Orsay

(France)

(b) Institut Néel, CNRS Grenoble

(France)

(c) Synchrotron SOLEIL, L’Orme

des merisiers, Saint-Aubin (France)

(d) INP Grenoble CNRS St Martin

d’Hères (France)

Fig. 87: 3D diffraction pattern ofthe 2kF satellite reflection

associated with the CDW. Thecentral peak corresponds to the2kF satellite reflection. Periodic

sequence of satellites flanking the2kF reflection appear in the sliding

regime (indicated by arrows).



Imaging of nanoislands by coherent GISAXSexperiments

High-resolution imaging ofnanoislands with hard X-rays is anexperimental challenge due to thedramatic reduction of the scatteredX-ray intensity from a sample when itssize is in the submicrometre range.The lack of focusing optics with ananometre focal spot at high X-rayenergies is another bottleneck for afull-field imaging of nanoislands. Inour work we use coherent X-rays in agrazing-incidence small-angle X-rayscattering (GISAXS) geometry to imageSiGe nanoislands grown by liquidphase epitaxy (LPE) on Si(001).A significantly enhanced diffractionpattern is obtained from an averageisland as a result of the sum of thecoherent scattering from individualislands. This is due to the narrow sizedistribution, identical shape, andorientation of the islands. We useiterative phase retrieval techniques forthe reconstruction of the projectedelectron density of the islands withtypical dimensions in the 100 nmrange. We obtained a spatialresolution of 10 to 15 nm for suchislands which is not achievable byother X-ray imaging methods.

In a conventional approach based onX-ray scattering methods, thestructural information (shape, strain,and composition) is obtained indirectlyby fitting the measured diffractionpatterns with simulated ones derivedfrom a realistic modelling of thenanostructures. Our model

independent approach yields highresolution imaging of nanostructures.As an example, we determined theelectron density of epitaxial SiGenanoislands using coherent scatteringin GISAXS geometry (Figure 88). In thisscattering geometry, the recordedintensity distribution depends on theelectron density distribution and theisland shape and size, but not onstrain. The GISAXS geometry providesan additional advantage to theconventional scattering techniques.Thelimited penetration of the X-rays intothe substrate results in a considerableenhancement of the sensitivity to thenanoislands on the sample surface.

The experiments were performed atbeamline ID01. The incidence angle αiwas taken equal to the critical angle αcfor the Si substrate αi = αc = 0.22°, foran X-ray energy of 8 keV. We usedSiGe islands of 140 nm and 200 nmsize as a model system. All islandsexhibit a truncated pyramidal shapewith a square base (see insets inFigure 89). In addition these islandstypically have a narrow sizedistribution and the same

transverse to the direction of slidingand the 2kF wave vector changes closeto the electric contacts. By usingcoherent X-ray diffraction at beamlineID01, we have observed anotherconsequence. In the sliding regime,secondary satellites appear in the closevicinity of the 2kF satellite reflection(Figure 87). The period of thecorresponding modulation isparticularly impressive since it is largerthan a micrometre, i.e. 1000 timeslarger than the period of the chargedensity wave itself (2π/2kF = 10 Å).Such long-range electronic correlations

are an unprecedented feature inelectronic systems. The existence ofthis super long-range order could be adirect consequence of theincommensurability of the CDW.In blue bronze, the 2kF wave vectoris incommensurate along the chainaxis b*, but it is close to thecommensurate value 0.750 b*. Thus,the proximity to the commensurabilitypoint could result in a lattice ofdiscommensurations, or soliton latticewith a periodic phase increment∆φ = 3π/4, leading to the secondarysatellites.

Principal publication andauthorsA.V. Zozulya (a), O.M. Yefanov (a),

I.A. Vartanyants (a),

K. Mundboth (b,c), C. Mocuta (b),

T.H. Metzger (b), J. Stangl (c),

G. Bauer (c), T. Boeck (d),

M. Schmidbauer (d), Phys. Rev. B

Rapid Commun. 78, 121304(R)

(2008).

(a) HASYLAB at DESY, Hamburg

(Germany)

(b) ESRF

(c) Institut für Halbleiterphysik,

Johannes Kepler Universität Linz,

Linz (Austria)

(d) IKZ, Berlin (Germany)

Reference[1] P. Monceau, N.P. Ong,

A.M. Portis, A. Meerschaut and

J. Rouxel, Phys. Rev. Lett 37, 602

(1976).

Fig. 88: Schematicdiagram of theexperiment in GISAXSgeometry on a singleisland in the form of atruncated pyramid witha square base.


76


EE SS RR FF

crystallographic orientation on a Sisurface. In GISAXS geometry with anincoming beam of a few hundreds ofmicrometres in size, a large number ofSiGe islands, typically 106, areilluminated simultaneously. Diffractionpatterns measured under theseexperimental conditions are shown inFigure 89.

Results of the reconstruction of theislands electron density projection arepresented in Figure 89. For comparison,a reconstruction of the island shapeobtained from a simulated GISAXSpattern is also shown on the sameFigure. We see that for our scatteringconditions the projection of theelectron density, the island shape, andtheir size are well reproduced.

We have demonstrated how thecoherent GISAXS technique can beused to image nanometre-sizedobjects. The high resolutionobtained in our experiment of about10 to 15 nm is strongly related to theenhancement of the scattered signalfrom the simultaneous illumination ofa large number of uniform islands.This approach does not depend on thecrystalline structure of such an objectand can, in principle, be applied to anymaterial system. An importantextension of this technique is itsapplication to the collection of 3Ddata measured by rotating the samplearound its axis perpendicular to thesubstrate surface, while maintaininggrazing incidence conditions [1].Such data can then be used for thereconstruction of the full 3D electrondensity of nanoislands [2].

Fig. 89: GISAXS diffraction patterns measured for different island sizes, a) 140 nm andb) 200 nm. SEM images of the samples with different magnification are presented in

the insets. c) and e) Results of the reconstruction of the islands electron densityprojections obtained from experimental GISAXS data. d) and f) The reconstruction of a

single island electron density obtained from a simulated GISAXS pattern. The size ofthe rectangular support region used in each reconstruction is indicated by a black line.

References[1] I.A. Vartanyants, et al., Phys

Rev. B 77, 115317 (2008).

[2] O.M. Yefanov and

I.A Vartanyants, Eur. Phys. J.

Special Topics 168, 81 (2009).

X-ray standing waves reveal the different bondingmechanisms of organic adsorbates

Organic electronics have becomeincreasingly popular for the varioussemiconductor applications that utiliseconjugated organic materials. Tailoringthe electronic and optical properties ofsuch organic heterostructures,however, requires a detailedunderstanding of the interfacesbetween organic molecules and metalsubstrates [1]. A detailedcharacterisation of the first molecularlayer on the substrate is crucialbecause this layer serves as a nucleusfor the growth of the organic thinfilm, and thus largely determines thecharge transfer into the organic layer.We have studied the adsorptionbehaviour of pentacene (PEN, C22H14)and perfluorinated pentacene (PFP,C22F14), two prototypical and widelystudied organic semiconductor

materials that are used in organicfield-effect transistors with high hole(PEN) and electron (PFP) mobility.

The X-ray standing wave (XSW) datarecorded at beamline ID32 reveal asurprisingly different bondingbehaviour of PEN and PFP moleculeson Cu(111). Like previous studies,which demonstrated the uniquepotential of this technique for organicsystems [2,3], the new results arebased on the high spatial resolution,the element specific information andthe sensitivity to the chemicalenvironment of the atoms. Thebonding distances d0 (Figure 90)derived from the photoelectron signalsof PEN and PFP (Figure 91a) give directevidence of their different interactionmechanism with the substrate.

Principal publication andauthorsN. Koch (a), A. Gerlach (b),

S. Duhm (a), H. Glowatzki (a),

G. Heimel (a), A. Vollmer (c),

Y. Sakamoto (d), T. Suzuki (d),

J. Zegenhagen (e), J.P. Rabe (a),

F. Schreiber (b), J. Am. Chem. Soc.

130, 7300 (2008).

(a) Humboldt-Universität zu Berlin

(Germany)

(b) Universität Tübingen

(Germany)

(c) BESSY (Germany)

(d) Institute for Molecular Science

(Japan)

(e) ESRF (France)



References[1] N. Koch, A. Vollmer, S. Duhm, Y. Sakamoto, T. Suzuki. Adv. Mater. 19, 112 (2007).

[2] L. Romaner, G. Heimel, J.L. Bredas, A. Gerlach, F. Schreiber, R.L. Johnson,

J. Zegenhagen, S. Duhm, N. Koch, E. Zojer. Phys. Rev. Lett. 99, 256801 (2007).

[3] A. Gerlach, F. Schreiber, S. Sellner, H. Dosch, I.A. Vartanyants, B.C.C. Cowie, T.-L. Lee,

J. Zegenhagen. Phys. Rev. B 71, 205425 (2005).

The analysis of the XSW data(Figure 91b) shows that PEN interactsstrongly with the Cu(111) substrateand adsorbs with an average bondingdistance of only 2.34 ± 0.02 Å, whereasthe carbon core of PFP is located atd0 = 2.98 ± 0.07 Å. Despite therelatively large distance from thesurface, the PFP molecules exhibit anadsorption geometry that does notcoincide with their planar gas phasestructure (Figure 91c).

The considerable difference of 0.64 Åfound for the positions of the carboncores of PFP and PEN demonstratesthat the later one interacts muchstronger with the Cu(111) substrate.Unlike its perfluorinated counterpartPEN molecules obviously undergo astrong chemisorptive binding.This observation is fully supported bycomplementary photoemission studiesof the valence band structure.The particular hole injection barriersfor PEN and PFP on Cu(111) can beunderstood by taking into accountthat the non-planar conformation ofPFP results in an adsorption-inducedintramolecular dipole of ~0.5 Debyeperpendicular to the surface. Overallthe example illustrates how XSWmeasurements at ID32 can beemployed to study the subtle interplaybetween structural and electronicproperties at organic/metal interfaces.

Fig. 90: An X-ray standing wave, with the periodicity of the substrate lattice dhkl, isgenerated by Bragg reflection from a single crystal. By scanning the photon energy of theincident wave through the Bragg condition the phase of the interference field changes byπ and as a consequence the nodal planes of the standing wave field are shifting by dhkl/2.Photoelectrons are used as element-specific signal. Depending on the site of the emittingatoms relative to the Bragg planes the photoelectron yield Yp(E) exhibits a characteristicshape which reveals the adsorption site d0 of a given element.

Fig. 91: a) Core-level spectra with the corresponding analysis for submonolayers ofPEN and PFP on Cu(111) which were used for the XSW measurements. b) X-ray

standing wave scans obtained on submonolayers of PEN and PFP on Cu(111). Thesymbols represent the photoelectron yield (circles) and substrate reflectivity (triangles)

data measured for both adsorbate systems. The solid lines show least-squaresfits based on dynamical diffraction theory which provide the coherent position

PH (d0 / d111 = PH , modulo n) and coherent fraction fH for each element [2,3].The coherent position thus reveals the adsorption distance d0 of PEN and PFP, whereasthe coherent fraction (,fH < 1) with fH = 0.55 for PEN/C1s and fH = 0.41 for PFP/C1s and

PFP/F1s, is related to disorder in the adsorbate system. c) Schematic conformation ofPEN (top) and PFP (bottom) on Cu(111) derived from XSW measurements, indicating

the different average positions of the F and C atoms.


78


EE SS RR FF

In situ studies

Nanofaceting of vicinal Ni surfaces induced byAg deposit

The deposit of small amounts of atomson surfaces may induce the formationof well-organised nanostructures withlong-range order. In the case of avicinal Cu surface [1], the adsorptionof Ag induces well-defined, periodicfacets of the surface. Moreover, theAg coverage can be used to tune theperiodicity and the orientation ofthese facets. The Ag/Ni couple wasthought to be very similar to Ag/Cu:Ag and Ni are not miscible in the bulk,and their cohesive energies and atomicsizes are very different, leading to Agsegregation at the Ni surface and toan abrupt chemical interface. However,the faceting behaviour of vicinalsurfaces under Ag adsorption wasfound to be different in the two cases,as observed by scanning tunnelling

microscopy (STM) and confirmedquantitatively by Grazing incidenceX-ray Diffraction (GIXD), performed onthe ID03 beamline.

Ni (322) is a surface with a miscut of11.42° with respect to the (111) planesand can be viewed as a regularsuccession of (111) terraces of 1.03 nmwidth separated by {100}-typemonatomic steps. When Ag is depositedon this surface in the monolayer range,and after thermal annealing, periodicfacets are indeed observed covering thewhole surface, but only within a verynarrow coverage range of around0.6 monolayer. Figure 92 (a and b)shows this “ideal” faceted surfaceobserved by STM and the correspondingX-ray reciprocal space map recorded atk = 2 around two nickel Bragg peaks onwhich the scattering rods originatingfrom the facets are well distinguished.Two orientations are observed for thefacets: (111) and (211). The comparisonwith a corresponding map recordedaround the Bragg peaks of the relaxedsilver deposit at k = 1.77, Figure 92c,which shows only (211) facet induceddiffuse rods arising from the Braggnodes, enables us to conclude that thesurface is made of (111) bare Ni facetsand (211) Ag covered facets. Thenarrow (111) oriented diffuse featuresin Figure 92c do not stem from silverBragg peaks and are presumably linkedto planar defects in the nickel nearsurface region.

For other coverages, in particular lowerones such as 0.3 monolayers, a surfacephase separation between facetedregions similar to the case previouslydescribed (same facets with similar localperiodicity) and bare vicinal nickel asshown in the STM image and the twocorresponding maps in Figure 92 (d, eand f) are observed. The energetics ofthe system would need to be furtherinvestigated to understand this surfacefaceting decomposition.

In addition, as already visible in theSTM image, Figure 92a, one of the

AuthorsC. Chambon (a,b), A. Coati (a),

M. Sauvage (a) and

Y. Garreau (a,b).

(a) Synchrotron SOLEIL (France)

(b) MPQ Paris (France)

Fig. 92: Ni(322) surfaceafter Ag deposit

followed by annealingat 600 K. The Ag

covered and bare Nifacets are clearly

recognised. a), b) and c) correspond

to a 0.6 monolayerdeposit, and d), e) and f)

to 0.3 monolayers. a) 3D STM image

(50x50) nm2; b) (h, l) reciprocal

space map at k = 2;c) (h, l) reciprocal space

map at k = 1.77;d) 3D STM image

(250x250) nm2; e) (h, l) reciprocal space

map at k = 2; f) (h, l) reciprocal space

map at k = 1.77.



facets shows a reconstruction. indeed,accurate GIXD data could be collectedon the reconstructed Ag covered (211)facets (Figure 93a) corresponding to a(2 x n) surface cell (8 < n < 9). A startingmodel sketched in Figure 93b involving6 silver rows on top of two successive(111) terraces of the (211) Ni substratewith (n-1) Ag atoms for n Ni atomsalong the [01] direction has been used.When optimised by quenchedmolecular dynamics simulationsinvolving 60 bulk layers, this modelrenders the diffraction data well butneeds further refinement toquantitatively account for all measuredin-plane and out-of plane structurefactors. Mixing various coverages of theterraces is being investigated.

The knowledge of the systemmorphology and its surface structureare key parameters for its possible useas a nano-organised template forfurther growth of selected nano objectswith magnetic or catalytic properties.This study shows that the couplingbetween GIXD, STM and atomisticsimulations is necessary and fruitful tounderstand such phenomena.

Fig. 93: Surface reconstructioninduced by silver on the (211)facets. a) GIXD in-plane data whichshows the (2xn) surface periodicity;b) model of the surface structureused for QMD simulations(Ag atoms in orange, Ni atoms inblue).

References[1] A. Coati, J. Creuze, and

Y. Garreau, Phys. Rev. B 72,

115424 (2005); Y. Garreau,

A. Coati, A. Zobelli, J. Creuze,

Phys. Rev. Lett. 91, 116101 (2003).

The shape of supported rhodium nanoparticles duringoxidation and reduction

The world market for the productionof chemicals is estimated today to bearound 1,300 billion Euros per year.More than 90% of these chemicals areproduced via a catalytic process.

Catalysts also play an increasinglyimportant role for the developmentof green chemistry and advancedrenewable energy technologies.

A study on catalytic nanoparticlescarried out on beamline BM32 givesan example of how in situ X-rayscattering in dedicated sampleenvironments can bridge the gapbetween structural and functionalproperties of nanostructures. Inparticular the interplay between theshape and size change of thenanoparticles and theoxidation/reduction processcan be monitored.

The current challenge for fundamentalresearch is to provide a detailedmicroscopic understanding of thedifferent physical and chemicalprocesses which take place atnanoparticles during catalytic reactions[1]. It is expected that nanoparticlesshould exhibit enhanced catalyticactivity: (i) as they possess an increasednumber of under-coordinated atoms,and (ii) as different low index facetsco-exist, facilitating mass transport andthereby lifting kinetic barriers knownfrom single crystal surfaces. Duringcatalytic cycling experiments,nanoparticles have been observed toundergo reversible size changes, whichare associated with cyclic shapechanges, material re-dispersion andsintering.

Here, we report an in situ high-resolution X-ray diffraction study of

Principal publication andauthorsP. Nolte (a), A. Stierle (a),

N.Y. Jin-Phillipp (a), N. Kasper (a),

T.U. Schülli (b), H. Dosch (a),

Science 321, 1654 (2008).

(a) Max-Planck-Institut für

Metallforschung, Stuttgart

(Germany)

(b) INAC/ SP2M Commissariat

àl’Energie Atomique, Grenoble

(France)


80


EE SS RR FF

epitaxial rhodium nanoparticles onMgO(100), during oxidation and COinduced reduction. This study broughtto light a reversible facetrearrangement of the nanoparticles,which is in direct relation to theformation of oxygen-inducedsuperstructures. We were able toaccess the average particle shape andsize with atomic resolution from aquantitative fitting of the extendedreciprocal space maps that wereobtained at elevated temperaturesand under varying gas atmospheres.

Rhodium nanoparticles were depositedin situ at a substrate temperature of670 K in the BM32 ultra-high vacuum(UHV) surface X-ray diffractionchamber, after cleaning the MgOsubstrates by sputtering and annealingunder an oxygen atmosphere. Thesample was annealed at 970 K toachieve the equilibrium shape of the

nanoparticles (see Figure 94a). Grazingincidence small angle scattering wasperformed simultaneously to wide-angle diffraction reciprocal-spacemapping.

High-resolution reciprocal space mapswere recorded from (H,K) = (-0.5,0.5)to (0.5,-0.5) and from L = 0.6 to 0.84.Figure 95a (top) shows a high-resolution map of the clean particles,observed at 600 K under UHVconditions. The data can beunderstood in a straight forward waywithin kinematical diffraction theorywhich suggests nanoparticles with atruncated octahedral shape.

In the next step, the Rh nanoparticlesare exposed to 3 • 10–5 mbar O2 at600 K and simultaneously we recordedthe X-ray diffraction pattern(Figure 95 (b and c)). We actuallyobserve a very clear change in the X-ray diffraction signal (see differencemap, Figure 95b) which consists of anintensity enhancement along the (001)rod and an intensity loss along the(111)-rods.

As a result of this analysis we find thatthe area both of the (100) side facetsand (001) top facet increases uponoxidation, which is driven by theformation of ultrathin oxide layers.This implies that only intra particlemass transport takes place during theoxidation thereby removing Rh atomsfrom the (100) side and top facets. Theaverage amount of atoms removedcorresponds to the number of atoms

Fig. 94: a) Schematic representationof the particle shape for fcc basednanoparticles. The lateral particlesize is characterised by afcc/2 *NP,its height by afcc /2 *(NT+NB) and

the missing edge atoms by NE.For a Wulff shaped particle

(without substrate) is NT = NB

and NE = NP-NT. b) (110) planereciprocal space map of clean Rhnanoparticles on MgO(100) with

an average lateral size of 8 nm (infcc bulk coordinates). The whitebox indicates the area for high

resolution scans.

Fig. 95: a) (top) (110) Diffractionmap of clean Rh particles at 600 K.

(middle) Fitted diffraction mapcorresponding to the average

particle shape given below.b) (top) Oxygen induced signal

change in the (110) plane. (middle)Simulated signal change for

particles with increased (100) sidefacet area. c) (top) Experimental

(110) diffraction map at 600 K and3 • 10–5 mbar O2 pressure.

(middle) Fitted diffraction map forparticles under oxygen exposure.

(bottom) Best fit core particleshape after oxidation.



incorporated into the surface oxidelayers on all facets. A further keyobservation is that the oxygen-inducedshape change of the Rh nanoparticle isfully reversible when the surface oxideis removed by CO exposure

(at 1 • 10–5 mbar). We havedemonstrated that in situ reciprocalspace mapping can give an atomisticinsight into the structure andmorphology of catalytically-activenanoparticles on oxide supports.

Reference[1] G. Ertl, H. Knözinger, F. Schüth,

J. Weitkamp, Handbook of

Heterogeneous Catalysis, Wiley-

VCH, Weinheim (2008).

X-ray study of mesoscopic relaxations in cobaltnanoislands on Cu(001)

Many atoms within nanostructuresexperience reduced coordination whencompared to the bulk. This leads tostrongly-modified chemical andphysical properties, one being therearrangement of the atoms in thestructure. Despite its fundamentalimportance, knowledge of the atomicgeometry in nanostructures is quitescarce. The interatomic distancerepresents the most importantparameter and a theoretical study [1]on cobalt nanoislands on Cu(001)consisting of several tens of atoms haspredicted that the interatomicdistance is dramatically reduced fromd = 2.51 Å for the bulk to values inthe 2.4 Å range, i.e. considerablyexceeding “usual” relaxations knownfor surfaces.

To provide direct evidence for therelaxations, we studied the registry ofthe cobalt-atoms with the Cu(001)substrate at beamline 33-ID-D(Advanced Light Source, ANL, USA)and at beamline ID03. In contrast tothe common view that Co-atoms residein hollow sites at a distance ofd = 2.56 Å characteristic for theCu(001) surface, a shortenedinteratomic distance (d = 2.40 Å)implies that cobalt atoms are notonly located in hollow sites but alsooccupy low symmetry positions off thehollow site.

Figure 96 shows the positions of431 cobalt atoms within the surfaceunit cell derived from moleculardynamics calculations. The distributionof the cobalt atoms around the hollowsite at (x,y) = (0,0) can be approximatedby a Gaussian function of half width athalf maximum of <u> = 0.18 Å,corresponding to a Debye parameterof B = 8π <u2> = 2.56 Å2.The determination of the static disorder

of the adsorbed Co-atoms is the basisof the X-ray analysis.

The registry of the cobalt atoms isprobed by the analysis of the intensitydistribution along the substratereflections, which are referred to ascrystal truncation rods (CTR’s).The CTR’s arise from the terminationof the crystal leading to a continuousintensity distribution, [I(qz)] betweenthe bulk Bragg-reflections. Figure 97shows the structure factor amplitudes

Principal publication andauthorsO. Mironets (a),

H.L. Meyerheim (a), C. Tusche (a),

V.S. Stepanyuk (a), E. Soyka (a),

P. Zschack (b), H. Hong (b),

N. Jeutter (c), R. Felici (c), and

J. Kirschner (a), Phys. Rev. Lett.

100, 096103 (2008).

(a) Max-Planck-Institut f.

Mikrostrukturphysik, D-06120

Halle (Germany)

(b) APS, Argonne National

Laboratory, Argonne, IL 60439

(USA)

(c) ESRF

Fig. 96: Molecular dynamicscalculation showing the positionsof cobalt atoms in the Cu(001) unitcell. The centre at (x,y) = (0,0)corresponds to the hollow site.

Fig. 97: CrystalTruncation Rods for 0.3monolayer cobalt onCu(001). Symbols andlines correspond toexperimental andcalculated structurefactor amplitudes,respectively. Curves areshifted vertically forclarity.


82


EE SS RR FF

(|F| ∝ √I) along several CTR’s. Dottedlines represent the calculated |F|-valuesfor clean Cu(001), while solid anddashed lines are fits to the data(symbols) assuming hollow siteadsorption with and without disorderof the cobalt atoms, respectively.

The solid line fits the data bestespecially close to the minima alongthe CTR’s, which is the most surfacesensitive position. Figure 98 summarisesthe cobalt Debye-parameters B versuscoverage. We find values in theB = 2.5 Å2 range, in excellentagreement with calculated valuesshown as triangles. In summary, theanalysis of the static disorder ofsurface adsorbed cobalt atoms onCu(001) in combination withmolecular- dynamics calculations basedon STM images has providedexperimental evidence for largecontractions of the interatomic cobaltdistances in monoatomic nanoislands.

Fig. 98: Experimentally-derivedDebye parameters BCo of cobaltatoms (circles) versus coverage.

Triangles represent correspondingvalues determined by molecular

dynamics calculation.

Reference[1] V.S. Stepanyuk, D.I. Bazhanov,

A.N. Baranov, W. Hergert,

P.H. Dederichs, and J. Kirschner,

Phys. Rev B 62, 15398 (2000).

Introduction

Highlights are “snap shots” in time, theypresent the latest achievements, whichare often the culmination of many yearsof work. This is not only true of thescientific programmes that lead to anexperiment at the ESRF but also for thebeamline facilities which have resultedfrom investment (human and financial)over the ~15 years that the ESRF hasbeen operating. In that time, newinstruments and support facilities havebeen built that allow today’s state-of-theart experiments to be performed. Soeven if an experiment was conceivedand the results published only recently,it is building on the work done andexperience gained over the years. We arenow at a point in time where the ESRFwill start an Upgrade Programme. Theaim is to allow experiments to beperformed that cannot be done today.Although the gradual improvement ofexisting facilities is important and willallow new science in the future, theUpgrade Programme will take us a stepfurther and enable the developmentof facilities that cannot be realisedin this incremental manner. Thesedevelopments will set the foundationsfor the scientific results of the nextdecade or more.

In the articles that follow there is astrong emphasis on magnetic materials,whether they are studies of dilutemagnetic semiconductors, multiferroicsor magnetostrictive materials. Inaddition, since absorption spectroscopy isa very powerful technique that can beapplied to many problems, there are alsohighlights resulting from a variety ofother studies. Examples includeexperiments on neptunium containingmaterials, which are interesting forenvironmental reasons; studies of thegeologically important iron oxide,

siderite, and research on alkali-dopedfullerenes with the aim of producingnew high-dimensional polymers. Also,the importance of industrial activities onthe beamlines should not be forgotten.This is particularly evident at ID24 whereToyota has been an industrial user forseveral years and is well illustrated by thehighlight article on vehicle exhaustcatalysts (see Figure 99).

In the future we can expect studies ofmagnetism and of the electronicproperties of materials to remain anexciting area of research. The UpgradeProgramme will take these activities tonew horizons. There will be moreemphasis on “dynamical properties”,“extreme conditions” and “science at thenanoscale” as these are some of thethemes of the Upgrade Programme.There will undoubtedly be otherimportant developments outside theUpgrade Programme. For instance, thepossibility of having experimentalfacilities with very high static magneticfields (> 30T) is being actively pursued.New and improved facilities for users willgive us an even stronger basis for thehighlights of the future.

N.B. Brookes



X-ray Absorption and MagneticScattering

Fig. 99: An on-lineapproach toincreasing catalystlifetimes. X-raystudies by meansof a special in situcell were used toinvestigate auto-exhaust catalystsunder realisticconditions atbeamline ID24(Image courtesy ofTOYOTA MotorCorporation).


84


EE SS RR FF

Au and Fe magnetic moments indisordered AuFe alloys

Recent developments in thirdgeneration synchrotron radiationfacilities has made possible theelement-specific determination ofmagnetic moments in multi-elementmaterials via the X-ray magneticcircular dichroism (XMCD) technique.The limit of detection is now wellbelow 0.01 µB/atom even for thin films.Moreover, the ability to separate spinand orbital magnetic momentcontributions has rendered XMCDas the most powerful tool for thecomplete magnetic characterisationof materials and for clarifying subjectsthat had remained controversial inthe past.

In this work, we have determined the5d induced magnetic moment of Auvia XMCD, for disordered fcc and bccAuFe alloys in the concentration range25 - 97 at. % Fe. The exotic fcc phase isnot predicted in the binary alloy phasediagram at room temperature.

However, one may produce singlephase fcc disordered AuFe alloys viarapid quenching of the molten alloyfor concentrations up to 53 at. % Fe.Iron is in the regular low-spin state inthe bcc Fe-rich alloys, but in the high-spin state in the ferromagnetic fccalloys. This fact may further influencethe Au induced magnetic moment andit makes the AuFe alloys extremelyinteresting from the point of view ofmagnetism; however, they could notbe thoroughly studied in the past andcontroversial results were published,due to the lack of experimentaltechniques with element specificityand shell selectivity.

Figure 100 presents the normalisedXANES and XMCD spectra recorded atthe L3,2 edges of Au in AuxFe1-xdisordered alloys. The data wererecorded at the beamline ID12 usingthe fluorescence detection scheme.The intense photon flux of the Apple IIundulator with a very high degree ofcircular polarisation was necessary inorder for high quality data to beaccumulated for the Au signal. Thesizeable XMCD signal reveals that Auhas acquired an induced magneticmoment. Knowing the direction of themagnetic field and the helicity of thebeam we were able to conclude thatAu is polarised parallel to the Femagnetic moment. Our analysisfollowed the magneto-optical sumrules. The outcome was the precisedetermination of the 5d spin andorbital magnetic moment of Au in thedisordered AuFe alloys. The resultsshow that the total 5d inducedmagnetic moment of Au in the fccalloys scales with the number N ofnearest Fe neighbours, fromMAu = 0.1 µB/atom, for N = 3, toMAu = 0.2 µB/atom, for N = 6. Iron inthese alloys is in a high spin statecarrying a total magnetic moment ofabout 3 µB/atom as our superconductingquantum interference device (SQUID)magnetometry measurements haverevealed. The maximum value of 5dinduced magnetic moment of Au is0.33 µB/atom and is exhibited when Auis placed as an impurity in a bcc Fe

AuthorsF. Wilhelm (a), P. Poulopoulos (b),

V. Kapaklis (b), J.-P. Kappler (c),

G. Schmerber (c), A. Derory (c),

N. Jaouen (a), A. Rogalev (a),

A.N. Yaresko (d), C. Politis (b,e).

(a) ESRF

(b) University of Patras (Greece)

(c) IPCMS Strasbourg (France)

(d) MPI Dresden (Germany)

(e) INT Karlsruhe (Germany)

Fig. 101: Au and Femagnetic moments by

experiment and theoryas a function of the Feat. % concentration in

disordered AuFe alloys.The lines are guides to

the eye.

Fig. 100: NormalisedXANES and XMCDspectra at the L3,2

edges of Au in AuxFe1-x

alloys, x as indicated.The spectra reveal a

straightforwardrelation between the5d induced magneticmoments of Au and

the number of Fenearest neighbours.



environment, i.e. when all its firstneighbours are Fe atoms. In the bccenvironment the total magneticmoment of Fe is 2.2 µB/atom. Thus,one may conclude that the total 5dinduced Au moment seem to follow asimple scheme of the typeMAu ∝ (N/Ntotal)*MFe regardless of thecrystallographic structure. Ourexperimental results for the Au and Femagnetic moments compare well withfirst principle calculations, as Figure 101reveals.

In conclusion, the relationshipbetween the induced magnetism ofAu and the Fe magnetic moments fordisordered bcc and fcc AuFe alloyshave been found and a simpleformula for the Au induced magneticmoments was deduced. Theexperimental results, recorded via acombination of XMCD and SQUIDresolve previous controversialexperimental data and are in goodagreement with our first principlecalculations [1].

Reference[1] F. Wilhelm, P. Poulopoulos,

V. Kapaklis, J.-P. Kappler,

N. Jaouen, A. Rogalev,

A.N. Yaresko, and C. Politis, Phys.

Rev. B. 77, 224414 (2008).

Absence of intrinsic ferromagneticinteractions in Zn1-xCoxO with highstructural perfection

Dilute magnetic semiconductors areenvisioned as sources of spin-polarisedcarriers for future semiconductordevices which simultaneously utilisespin and charge degrees of freedom.Zn1-xCoxO (Co:ZnO) was thought apromising n-type dilute magneticsemiconductor for spintronics.Although Co:ZnO materials have beeninvestigated intensively, theirrespective magnetic behaviours remaincontroversial from experimental andtheoretical points of view. Claims ofroom-temperature ferromagnetism, orits absence, have been made byexperimentalists and theoreticians [1].Thus, the possibility that metallic Conanoclusters account for the observedmagnetic order has to be ruled outwith great care.

Co:ZnO with 10.8% Co was grown bypulsed laser deposition onto c-planesapphire substrates [1], and studiedwith SQUID magnetometry andsynchrotron radiation at the Zn and CoK-edges. X-ray absorption near-edgespectra (XANES) were recorded atbeamline ID12 in total fluorescenceyield both with linear and circularpolarised light under 10° grazingincidence. To record the X-ray lineardichroism (XLD) the polarisation wasflipped using a quarter wave plate.The X-ray magnetic circular dichroism(XMCD) was recorded by reversing thecircular polarisation as well as themagnetic field direction.

Figure 102 shows the XANES and theirrespective XLD spectra recorded at theZn and Co K-edges using linearpolarised light. The XLD spectra showfeatures typical for the wurtzite latticeof the ZnO host. In addition,simulations of the XANES and XLDspectra were made using FDMNES [2],and the lattice parameters of ZnO fora Co:ZnO supercell with Co on Znsubstitutional sites. The goodagreement between experiment andsimulation suggests that more than95% of the Co dopant atoms arelocated on Zn substitutional sites. Thisresult underlines the excellentstructural quality of the Co:ZnO,devoid of metallic Co, small Co clustersor antisite disorder.

Figure 103 summarises the magneticproperties of the Co:ZnO sample. M(H)-curves with H perpendicular to

Principal publication andauthorsA. Ney (a), K. Ollefs (a), S. Ye (a),

T. Kammermeier (a), V. Ney (a),

T.C. Kaspar (b), S.A. Chambers (b),

F. Wilhelm (c), and A. Rogalev (c),

Phys. Rev. Lett. 100, 157201

(2008).

(a) Experimentalphysik,

Universität Duisburg-Essen,

Duisburg (Germany)

(b) Pacific Northwest National

Laboratory, Richland,

Washington, (USA)

(c) ESRF

Fig. 102: XANES spectraat the Zn and Co K-edges recorded withtwo orthogonal linearpolarisations and therespective XLD spectra.Simulations using theFDMNES code are alsoshown.


86


EE SS RR FF

the c-axis were recorded using SQUID(full symbols) and XMCD at the pre-edge feature of the Co K-edge asindicated in the inset. This pre-edgefeature is characteristic for Co2+ intetrahedral coordination [1] confirmingthe XLD results. The XMCD data werescaled to the SQUID data and bothtechniques consistently showparamagnetic behaviour with acomparable curvature of M(H). Threedifferent fits using Brillouin functionsfor S = 3/2 and various L are scaled tothe SQUID data. L = 3 reflects themaximum possible value of themagnetic moment for Co2+ (3d7)resulting in too strong a curvature,whereas the L = 0 fit has too weak acurvature to reproduce the data. UsingL = 1.07, (µz = 4.1µB/Co atom), whichresults in an L/S-value of 0.7 that wasalready reported by means of XMCD

data recorded at the Co L3/2-edges, thecurvature of the M(H) curves can bereproduced well. However, if onecalculates the effective magneticmoment per Co atom from themeasured magnetisation by SQUID,one finds only 1.44 µB/Co atom whichresults in a linear M(H)-curve using therespective Brillouin function indisagreement with the data.In other words, only 28(3)% of themagnetisation that can be expectedfrom the curvature of M(H), aremeasured by SQUID. If one calculatesthe abundance of isolated Co atoms,Co-O-Co pairs, triples etc on Zn cationsites, i.e. an hcp sublattice with 12 nextneighbours [3], this results in 28% ofisolated Co atoms, 18% of pairs, 11%of triples etc. for 10.8% of Co in ZnO.Therefore, the magnetisationmeasured by SQUID can be readilyexplained by a paramagnetic behaviourof the isolated Co atoms, anantiparallel alignment of the Co-O-Copairs and negligible contributions fromlarger cation clusters.

In conclusion we were able todemonstrate by means of XLD that theCo dopant atoms occupy exclusivelysubstitutional Zn sites in high-qualityCo:ZnO samples. In such samples, theisolated Co atoms behave asparamagnetic whereas the Co-O-Copairs couple antiferromagnetically.Thus we find no signs of intrinsicferromagnetic interactions.

Fig. 103: M(H)-curvesrecorded at 5 K by

SQUID and XMCD at theCo K-edge, respectively.

The inset shows therespective XMCD

spectrum. Lines areBrillouin fits (see text).

References [1] T.C. Kaspar, T. Droubay,

S.M. Heald, P. Nachimuthu,

C.M. Wang, V. Shutthanandan,

C.A. Johnson, D.R. Gamelin, and

S.A. Chambers, New J. Phys. 10,

055010 (2008).

[2] Y. Joly, Phys. Rev. B 63, 125120

(2001).

[3] R.E. Behringer, J. Chem. Phys.

29, 537 (1958).

Probing ion-specific magnetism inmultiferroics

Magneto-electric multiferroics arematerials in which mutually-coupledmagnetic and ferroelectric order-parameters coexist in a single phase.The recent discovery of such systemshas lead to a boom in research in thisfield, not only due to their potentialfor technological application, but alsobecause they offer an opportunity fordetailed studies of coupled, and oftencompeting order parameters [1]. Incertain rare-earth manganites complexmagnetic structures, such as cycloidalspin density waves, break inversionsymmetry: the essential ingredient to

obtain a spontaneous ferroelectricpolarisation. Strong, intrinsic magneto-electric coupling can give rise to thecondition whereby it is possible tomanipulate the spontaneous electricpolarisation by applying an externalmagnetic field, and thus produce apotentially important sensor material.For example, at low temperatureTbMn2O5 displays a complete reversalof electric polarisation uponapplication of a modest externalmagnetic field of 2 T [2]. To completelyunderstand this mechanism and topredict the macroscopic properties

Principal publication andauthorsR.D. Johnson (a), S.R. Bland (a),

C. Mazzoli (b), T.A.W. Beale (a),

C-H. Du (c), C. Detlefs (b),

S.B. Wilkins (d) and P.D. Hatton (a),

Phys. Rev. B 78, 104407 (2008).

(a) Department of Physics,

Durham University (UK)

(b) ESRF

(c) Department of Physics, Tamkang

University, Tamsui (Taiwan)

(d) Department of Condensed

Matter Physics and Materials

Science, Brookhaven National

Laboratory, Upton (USA)



observed, it is vital to unravelthe complex magnetic structure.Ultimately this will allow thefabrication and optimisation ofnew devices. On substitution of therare-earth ion, the RMn2O5 series(R = rare-earth, Bi or Y) displays arich phase diagram with a diverserange of macroscopic properties.Resonant X-ray scattering (RXS)selectively probes the long-rangemagnetic order present on a particularatomic species by tuning to an atomictransition corresponding to theexcitation of a core electron. Thesubsequent recombination of theexcited electron with the core holeemits a photon whose polarisationencodes precious information on themagnetic state of the resonating ions.

We have employed RXS to investigatethe magnetic behaviour of the terbiumions in TbMn2O5 at 25 K, where theelectric polarisation is most evident [3].An energy scan through the terbium LIIIabsorption edge at a constant wave-vector clearly showed two resonances,one just above the absorption edgeand the other 8 eV lower. Theseoriginate from E1-E1 and E2-E2transitions, respectively. In contrast tothe E1-E1 transitions probing the band-like 5d levels that may hybridise withthe neighbouring ions, the E2-E2transitions directly probe themagnetism of the terbium 4f shell. Thepolarisation of the emitted photons isdependent upon the incident X-raypolarisation, magnetic momentdirection and, in the case of the E2-E2transition, wave-vector. Thisdependence is probed by full linearpolarisation analysis as shown inFigure 104, a technique recentlydeveloped at beamline ID20.Simulations based upon the theory ofHannon et al. [4], of the scatteredpolarisation measured as a function ofincident polarisation, allows one torefine the magnetic moment directionsassociated with both the terbium 5dand 4f states.

In TbMn2O5, manganese ions arefound in two sublattices, Mn3+ ionscoordinated in MnO5 square basedpyramids and Mn4+ ions in MnO6octahedra. These sublattices orderbelow a Néel temperature of 43 K.It had been hypothesised that the

magnetic order on the terbiumsublattice is induced through nearestneighbour interactions solely withMn4+ ions [5]. We have confirmed thisscenario through the measurement ofa superlattice reflection sensitive tothe long range ordering of themagnetic moments which is shown toresonate at the binding energy forterbium, confirming an ordering of theterbium sublattice. Further analysis ofthe incident and exit photonpolarisation confirms that the inducedpolarisation arises solely throughinteraction with Mn4+ ions, shown bythe red line in Figure 104. We wereunable to reproduce the data whenassuming a polarisation of the terbiumions due to the Mn3+ ions, confirmingthe hypothesis of Blake et al. [5]. Byperforming a least squares fit to thepolarisation analysis, taken whentuned to the E2-E2 resonance, we haverefined the magnetic structure of theterbium ion sublattice as illustrated inFigure 105.

Fig. 104: Top:Experimental set up forfull linear polarisationanalysis. A diamondphase plate used intransmission geometryrotates the incident X-ray polarisation.The polarisation of thescattered beam may bemeasured in terms ofPoincaré-Stokesparameters (P1 and P2)using an analysercrystal. Bottom: P1 andP2 of the (4.5, 4, -0.25)magnetic reflection atthe E1-E1 transition at25 K as a function ofincident polarisation.

Fig. 105: The refined terbiummagnetic structure of TbMn2O5 at25 K. The magnetic moments lie inthe ab-plane with a slight offset inthe c-axis direction. Terbium ionsare shown in purple, oxygen in redand Mn4+ and Mn3+ in dark andlight green octahedral and square-based pyramid oxygen co-ordinations, respectively.

References[1] M. Fiebig, J. Phys. D: Appl.

Phys. 38, R123-R152 (2005).

[2] N. Hur et al., Nature 429, 392

(2004).

[3] L.C. Chapon et al., Phys. Rev.

Lett. 93, 177402 (2004).

[4] J.P. Hannon et al., Phys. Rev.

Lett. 61, 1245 (1988).

[5] G.R. Blake et al., Phys. Rev. B

71, 214402 (2005).


88


EE SS RR FF

Revealing the magnetic ground state ofCa3Co2O6

The study of frustration offerschallenges to both theoreticians andexperimentalists alike. New states,often characterised by peculiar staticand dynamic properties, are created asa result of competing interactions. Inparticular, when acting in magneticsystems, frustration reveals many of itsmost intriguing aspects and carefulinvestigations of specific systems canproduce fascinating new science.

In this respect, Ca3Co2O6 is a veryinteresting system, characterised bylow dimensional magnetism andtopological frustration. Figure 106a,shows the crystallographic structure ofthe compound, made up of chains offace-sharing polyhedra running alongthe c-axis, arranged in the ab-plane ona triangular lattice. Due to localsymmetry, one half of the Co ions havea magnetic moment (Co(II), µ ~5.1 µB),subject to (antiferro-)ferromagnetic(inter-)intrachain interactions. In thesketch, magnetic sites are located atthe centre of dark blue polyhedra,representing oxygen cages withtrigonal prismatic coordination.

Ca3Co2O6 has been the focus ofconsiderable experimental andtheoretical attention due to itspeculiar magnetic properties. Belowthe Neel temperature (TN ~25 K),

3D long range magnetic order isestablished. At low temperature, anunexpected evolution of the magneticorder parameter versus bothtemperature and magnetic field isobserved. In particular, diffractedintensities corresponding toantiferromagnetic (AFM) reflections(Figure 107 upper panel) decrease oncooling below 17 K, while the bulkmagnetisation increases in a series ofequally spaced steps versus themagnetic field, instead of smoothlyapproaching its saturation value.Depending on external conditions, therelaxation of the magnetisationtowards its equilibrium state can alsobe dramatically modified.

It is clear that a simple AFM model isnot able to account for the complexityof the magnetic ground state ofCa3Co2O6. For this reason, we decidedto investigate the magnetic propertiesof Ca3Co2O6 using resonant magneticX-ray scattering (RMXS) [1].By studying AFM reflections in zeromagnetic field and at lowtemperature, we found evidence thatthe system adopts an incommensuratemagnetic state immediately below TN.The characteristic temperatureevolution of a typical Bragg forbiddenAFM reflection in resonant conditionsis shown in Figure 107. Our results ruleout a number of previously proposedmagnetic structures, and pose seriousnew constraints on the description ofthe system.

To complete the overview of theproperties of the magnetic state andto access as much information aspossible in reciprocal space, weperformed a complementary neutronpowder diffraction (NPD) experiment[2]. Using symmetry analysis and byconsidering the possible exchangepathways within the material(Figure 106b: exchange pathsconnecting magnetic trigonal sites arerepresented) we showed thatCa3Co2O6 adopts a longitudinalincommensurate magnetic structure inthe form of a sinusoidal wavepropagating along the c-axis chains

AuthorsS. Agrestini (a), A. Bombardi (b),

C. Mazzoli (c), L. Chapon (d) and

M.R. Lees (e).

(a) Laboratoire CRISMAT, UMR

6508 CNRS-ENSICAEN, Caen

(France)

(b) Diamond Light Source Ltd.,

Rutherford Appleton Laboratory,

Chilton-Didcot (UK)

(c) ESRF

(d) ISIS Facility, Rutherford

Appleton Laboratory, Chilton-

Didcot (UK)

(e) Department of Physics,

University of Warwick, Coventry

(UK)

Fig. 106: a) Ca3Co2O6 hexagonal cell, octahedral Co(I) and trigonal Co(II) sites areshown in light and dark blue, respectively; b) proposed magnetic exchange pattern

between the Co(II) sites; c) sketch showing one possible arrangement of thenormalised Co(II) magnetic moments in the incommensurate magnetic ground state

(µ \\ c-axis, shown vertical). Some modifications to the cell and the propagation vectorlength have been applied for the sake of clarity.



(at T = 18 K the propagation vector isk~(0, 0, 0.01)). The modulation isphase shifted by 1/3 on moving fromone chain to the next, resulting in themagnetic moment arrangement asshown in Figure 106c.

As a result of our investigations,exploiting RMXS and NPD techniques,we have demonstrated that the threedimensional nature of the magneticexchange coupling in Ca3Co2O6 mustbe taken into account in order toproperly describe the magnetic groundstate realised in this low dimensionalsystem. Figure 106b shows a proposedexchange pattern (JFM >> JAFM, J’AFM)compatible with the magnetic groundstate. Our results challenge theexisting theories describing themagnetic properties of Ca3Co2O6 andprovide a basis for futureinvestigations of the unusual and asyet unexplained static (magnetisation

steps) and dynamic (relaxation of themagnetisation) properties of thissystem.

Fig. 107: Temperatureevolution of a Braggforbidden AFM reflection(320) and a chargereflection (330), in blackand grey symbols,respectively.

References[1] S. Agrestini et al., Phys. Rev. B

77, 140403R (2008); A. Bombardi

et al., Phys. Rev. B 78, 100406R

(2008).

[2] S. Agrestini et al.,. Phys. Rev.

Lett. 101, 097207 (2008).

High domain wall velocity induced bylow current densities in spin valvenanostripes

Moving magnetic domain walls innanostripes using electric currents viaspin-torque effects rather thanmagnetic field pulses is one of therecent exciting developments inspintronics [1]. Achieving reproducibledomain wall movements with lowcurrent densities and high domain wallvelocities is the main challenge in thisfield of science. The critical currentdensities necessary to move a domainwall in single layer FeNi nanostripesare of the order of 1012 A/m2 withzero magnetic field, while velocitiesspread from some m/s up to about100 m/s. These current densities aretoo high for applications. Lowercurrent density values have beenfound for spin-valve-basednanostripes. Our work shows the firstdirect evidence that in such materialsvery high domain wall velocities(exceeding 170 m/s) at zero field canbe obtained for current densities muchlower than for FeNi. We have usedphotoemission electron microscopy

(PEEM) combined with X-ray magneticcircular dichroism (XMCD) to measuredomain wall movements in thepermalloy layer of 400 nm wideFeNi(5nm)/Cu(8nm)/Co(7nm)/CoO(3nm)nanostripes (Figure 108a). Themeasurements were performed atbeamline ID08. The X-ray energy wastuned to the Ni L3 absorption edge toprobe the FeNi magnetic structure.

The remanent magnetic state obtainedfor a 90° angle zigzag nanostripe afterapplication of an in-plane externalfield perpendicular to the zigzag longaxis, is shown in Figure 108b. Magneticdomains with magnetisation pointingdownwards/upwards along thenanostripe (white/black contrast) areseparated by domain walls. A black-white-black-white contrast is visible atthe bends. This contrast is due to thefact that the stray field generated byhead-to-head or tail-to-tail domainwalls in the bends of the Co layerinduces a local antiparallel alignment

Principal publication andauthorsS. Pizzini (a), J. Vogel (a),

V. Uhlir (a), E. Bonet (a),

N. Rougemaille (a), S. Laribi (b),

F. Petroff (b), V. Cros (b),

E. Jiménez (c), J. Camarero (c),

C. Ulysse (d), G. Faini (d),

C. Tieg (e), Applied Physics

Express, accepted.

(a) Institut Néel, CNRS and UJF,

Grenoble (France)

(b) Unité Mixte de Physique

CNRS/Thales and Université Paris

Sud 11, Palaiseau (France)

(c) Dpto. Fisica de la Materia

Condensada, Universidad

Autonoma de Madrid and

IMDEA-NANO, Madrid (Spain)

(d) CNRS, PhyNano Team,

Laboratoire de Photonique et de

Nanostructures, Marcoussis

(France)

(e) ESRF


90


EE SS RR FF

between Co and FeNi layers in thevicinity of the Co domain wall [2].We found that these magnetostaticinteractions prevent domain wallmotion across the bends, but do nothinder domain wall motion away fromthe bends.

Figure 109 shows current-induceddomain wall motion in a stripe withzigzag angles of 120°. The initialmagnetic state shown in Figure 109awas obtained reproducibly in situ byapplying an ultrafast magnetic fieldpulse. Figure 109b shows the domainstructure obtained after applying one100 ns long current pulse withamplitude +2 mA. This current pulsecauses a displacement of the domainwall initially in position A, towardsposition B in the images.

Note that the other domain wallspresent in the nanostripe are notaffected by the current. This showsthat the pinning strength can be verydifferent at different sample positions.

A second current pulse with the sameamplitude and length moves the samedomain wall from B to C (Figure 109c).The two current pulses induce adomain wall movement of(1.7 ± 0.2) µm and (1.9 ± 0.2) µmrespectively, resulting in domain wallvelocities of the order of 20 m/s. Whenthe current is increased to 5 mA themeasured DW velocities are of theorder of 170 m/s. The current densitycorresponding to this current value is5 x 1011 A/m2, if we consider a uniformcurrent distribution through thetrilayer stack. These domain wallvelocities are well above literaturevalues for single FeNi layers, forcurrent densities that are at least afactor two smaller.

Currently used spin-torquephenomenological models do notallow such large domain wall velocitiesto be explained, and other spintransfer mechanisms should thereforebe considered. In our spin-valvestructures part of the incident spin fluxin the FeNi layer is transformed, in theregion around the domain wall, intospin accumulation in the Cu and Colayers. Spin accumulation in the Cuspacer below the domain wall inducesa vertical spin current that flows intothe domain wall and exerts anadditional torque on the domain wall.Our results suggest that the yield ofthis additional vertical spin transferchannel is much higher than the yieldof the in-plane spin transfer channel,as predicted recently by micromagneticsimulations on similar nanostructurescontaining a single domain wall.

Our direct observation of very fastcurrent-induced domain wall motionconfirms the potential application ofspin-valve systems to domain wall-based magnetic memories and logicdevices.

References[1] S. Parkin, M. Hayashi,

L. Thomas, Science 320, 190

(2008).

[2] J. Vogel, S. Cherifi, S. Pizzini,

F. Romanens, J. Camarero,

F. Petroff, S. Heun, and

A. Locatelli, J. Phys.: Condens.

Matt. 19, 476204 (2007).

Fig. 108: Topographic a) and magnetic b)PEEM images of a spin-valve-like nanostripewith a zigzag angle of 90°, taken at the Ni

L3-edge. The sample is magnetised with astatic in-plane magnetic field, as indicated

by the white arrow. In (b), white (black)contrast corresponds to FeNi domains with a

magnetisation component parallel(antiparallel) to the incoming X-ray

direction.

Fig. 109: XMCD-PEEM images of the FeNi layer in a zigzag nanostripe withangle of 120°; a) initial state with domain wall in position A;

b) after application of one 100 ns current pulse of +2 mA, which shifts thedomain wall to B; c) after application of a second 100 ns current pulse of

+2 mA, which shifts the domain wall to C.



Magnon dispersion in antiferromagneticcuprates measured by X-ray scattering

In the early days of high temperaturesuperconductivity it was alreadyrecognised that magnetic properties ofthese materials are intimately relatedto the superconducting ones. Whendoped, the long-range orderedantiferromagnetic background ofpristine copper oxide insulators getsfrustrated leading to short rangeantiferromagnetic fluctuations andsuperconductivity. That is why themagnetic properties of parentcompounds have attracted so muchattention since the discovery ofsuperconductivity in cuprates. The spindynamics of cuprates has been studiedup to now mostly with neutroninelastic scattering.

Recently Hill et al. [1] have found thatCu K edge resonant inelastic X-rayscattering (RIXS) spectra of undopedantiferromagnetic cuprates at 20 Kshow a sizable peak around 500 meVwhen the transferred momentum qcorresponds to the (π,0) point of thereciprocal space. This feature,suppressed by doping, has beenassigned to the simultaneousexcitation of two magnons. Magnonsare collective excitations of a latticepresenting a long range magneticorder and they correspond tochanging by one unit the magneticmoment of the system. Their non-localnature manifests itself in a dispersivebehaviour, i.e. a non-constant relationbetween momentum and energy.Following that first indication, wehave employed the same technique inthe soft X-ray range, working at the L3edge (2p → 3d transition) of Cu. Wehave seen that Cu L3 RIXS is an idealtechnique to determine magnondispersion in cuprates. In La2CuO4 andCaCuO2 we have found dispersingspectral features both at roomtemperature and at 30 K.

Figure 110 shows how Cu L3 RIXS worksin undoped cuprates, where all Cusites are divalent (3d9 configuration)and how two magnons can be therebyexcited simultaneously. In undopedcuprates, a two-dimensionalantiferromagnetic order of the spin

1/2 Cu2+ sites always characterises theground state. In the intermediatestate, the 3d10 configuration (spinzero) makes the scattering site act as amagnetic impurity that quenches thesuper-exchange locally in theantiferromagnetic lattice. Thescreening by neighbouring sites canresult in the generation of a couple ofmagnons. Similarly to optical Ramanscattering, the total spin moment ofthe system is conserved, but finiteenergy and momentum have beentransferred from the scattered photonto the system [2]. As opposed to Kedge RIXS, a single magnon can alsobe excited in L3 RIXS due to the strongspin-orbit interaction in the 2p corehole present in the intermediate state.

The measurements made at beamlineID08 with the AXES spectrometer havea combined energy resolution of400 meV, sufficient to resolvethe magnetic excitation in the 100-400 meV energy range. We areworking at present on the problem ofseparating single from mutliplemagnons. The results, summarised inFigure 111, show a clear dispersivebehaviour already in the raw data (leftpanel). The spectra are labelled asfunction of qn = q|| × a; q|| is thetransferred momentum component inthe ab plane and a is the in planelattice parameter, so qn isadimensional and equals π at the zoneboundary. In the right hand panel, theexperimental data-points arecompared to the single magnon resultsobtained on the same compound byinelastic neutron scattering and to thetheoretical prediction for bi-magnonsaccording to ref. [2].

Principal publication andauthorsL. Braicovich (a), L.J.P. Ament (b),

V. Bisogni (c), F. Forte (b,d),

C. Aruta (e), G. Balestrino (f),

N.B. Brookes (c), G.M. De Luca (e),

P.G. Medaglia (f),

F. Miletto Granozio (e),

M. Radovic (e), M. Salluzzo (e),

J. van den Brink (b,g), and

G. Ghiringhelli (a),

arXiv:0807:1140v1, (2008).

(a) CNR/INFM and Politecnico di

Milano (Italy)

(b) Universiteit Leiden (The

Netherlands)

(c) ESRF

(d) CNR/INFM and Università di

Salerno (Italy)

(e) CNR/INFM and Università

Federico II, Napoli (Italy)

(f) CNR/INFM and Università di

Roma Tor Vergata (Italy)

(g) Radboud Universiteit

Nijmegen (The Netherlands)

Fig. 110: Schematics ofRIXS at the Cu L3 edgeand RIXS excitation ofbi-magnons.


92


EE SS RR FF

These results provide the first directmeasurement of the magnon dispersionmade with inelastic X-ray scattering. Asthe peak position exceeds the energy ofa single magnon excitation measuredwith neutrons, an importantcontribution has to come from twomagnon excitations. A theoreticalcalculation for bi-magnons, afterbroadening for inclusion into theexperimental resolution, agrees verywell with the experimental data points.Nevertheless a contribution from singlemagnons cannot be excluded. In thedoped La1.85Sr0.15CuO4 sample we havemeasured a different dispersion curve,presenting a finite energy at qn = 0. Inthe doped compounds the measuredexcitations might have mixed spin-charge character.

Fig. 111: Cu L3

RIXS results ofLa2CuO4. The

peak dispersioncurve was

extracted fromthe raw data

(left panel)corresponding

to the givendots in the

Brillouin zone.

References[1] J.P. Hill, G. Blumberg, Y.-J. Kim, D.S. Ellis, S. Wakimoto, R.J. Birgeneau, S. Komiya,

Y. Ando, B. Liang, R.L. Greene, D. Casa, and T. Gog, Phys. Rev. Lett. 100, 097001 (2008).

[2] F. Forte, L.J.P. Ament and J. van den Brink, Phys. Rev. B 77, 134428 (2008).

Atomic-scale magnetostriction

Magnetostriction refers to the changein physical dimensions of a magneticmaterial, observed when there is achange in its magnetisation. It was firstdescribed by James Joule in 1842 andoriginates from spin-orbit coupling,where magnetic energy is converted tomechanical energy and vice versa.However, despite the obviousapplications of such an effect intransducer devices, the strain producedin common magnetic materials is only afew tens of parts-per-million (ppm) atmost. It was, therefore, not until thediscovery of room-temperature ‘giant’magnetostriction in rare-earthtransition metal alloys in the 1970s thatinterest developed in magnetostrictivetransducer technologies (Figure 112).

More recently, a different class ofmaterial based on binary metal alloyssuch as Fe-Al or Fe-Ga, have attractedconsiderable interest. It is well knownthat pure Fe exhibits only an extremelysmall magnetostriction (just 20 ppm),but experiments have revealed thatwhen it is alloyed with certainnonmagnetic metallic elements, themagnetostriction can be enhanced byover an order of magnitude.Compositions of Fe-Ga with around19at% Ga have reported strains of upto 400 ppm. Although this doesn’tconstitute a truly ‘giant’

magnetostriction, it is of interest fordevice applications since Fe-Ga isdevoid of expensive rare-earthcomponents, and possesses moredesirable mechanical properties than,say, the much studied Terfenol-D alloy.

From a theoretical perspective, initialproposals for the source of thisenhancement suggested it arose fromclustering of the Ga atoms within adisordered Fe lattice, to provide bothelastic and magnetoelastic defects.More recently, magnetostrictioncoefficients have been derived fromfirst-principles calculations, and themagnetocrystalline anisotropymodelled, which together, suggestedthat such defects could be betterformed by pairs of Ga atoms, arrangedrandomly throughout the material.To verify these proposals, it is clearthat atomic-scale structure and strainmeasurements are needed. Yet this iswhere problems arise.

Experimentally, magnetostrictive strainis elusive at the atomic scale. Strains ofthe order of tens of ppm onlygenerate displacements betweenatoms of the order of femtometres,which are extremely difficult to detect.Even commonly employed atomicprobes, such as X-ray AbsorptionSpectroscopy (XAS), lack the sensitivity

Principal publications andauthorsM.P. Ruffoni (a), S. Pascarelli (a),

R. Grössinger (b),

R. Sato-Turtelli (b),

C. Bormio-Nunes (c),

R.F. Pettifer (d), Phys. Rev. Lett.

101, 147202 (2008);

S. Pascarelli (a), M.P. Ruffoni (a),

R. Sato-Turtelli (b), F. Kubel (b),

R. Grössinger (b), Phys. Rev. B 77,

184406 (2008).

(a) ESRF

(b) Vienna University of

Technology (Austria)

(c) Universidade de Sao Paulo,

Lorena-SP (Brasil)

(d) University of Warwick,

Coventry (UK)



to observe such motion by up to twoorders of magnitude. As a result,experiments largely deal withmagnetostriction in a scaled-up form.Typical measurements employ straingauges on large, macroscopic samples,where the strain is easier to detect,but where atomic information is lost.

However, with the recent developmentof differential XAS (DiffXAS) onbeamline ID24 – a new techniquedesigned specifically for themeasurement of strains down to thescale of femtometres – such direct,atomic-scale measurements havebecome possible. When these arecoupled to conventional XAS data(from beamline BM29), the staticstructure may also be obtained in theabsence of an applied magnetic field,providing a complete picture of thesample behaviour.

Recently, we have carried out DiffXASand XAS experiments on samples ofFe81Ga19, where the magnetostrictionenhancement with respect to pure Fe ismaximised. From the conventional XASmeasurements, we were able toexamine the static atomic structureimmediately surrounding the Ga atoms,and found that there were no Gaclusters – in the first atomiccoordination shell around Ga, therewere no Ga-Ga bonds. Furthermore, inthe second coordination shell, wefound, on average, exactly one Ga-Ga

bond – testifying to the presence of theGa-Ga pair defects predicted by theory.

With an applied magnetic field, theDiffXAS measurements showed the waythis structure deformed. By examiningthe surroundings of the Fe and Gaatoms individually, we established thatthe strain emanated from theenvironment around the Ga atoms.Here, the magnitude of the strain wasabout 400 ppm, compared to only40 ppm around Fe. More importantly,negligible strain was found in theGa-Ga pair defect itself. We couldtherefore conclude that the pair defectsimply mediates the magnetostriction inthe material rather than providing itdirectly. Instead, and in agreement withtheory, the enhanced magnetostrictivestrain in Fe81Ga19 arises from the Ga-Febonds in the immediate vicinity of theGa-Ga pair defects.

Controlled aggregation of magneticcations in a semiconductor nitride matrix

The origin of the ferromagneticbehaviour persisting above roomtemperature (RT) and discovered insemiconductors doped with transitionmetals and rare earths may certainly beconsidered as one of the mostcontroversial issues in today’s materialsscience. Many misleading assumptionsand conclusions on the actual magneticinteractions in magnetically dopedsemiconductors and oxides have been,until recently, caused by lack of aproper correlation between fabricationparameters and structuralcharacterisation at the nanoscale.

Now, there is an increasing amount ofevidence that, owing to specificfeatures of magnetic impurities insemiconductors and oxides, theepitaxial growth of these systems canresult in the self-organised aggregationof magnetically robust nanocrystalsembedded in the host paramagneticmatrix. This finding holds enormouspotential for the fabrication of a rangeof multifunctional nanosystemsrelevant to spintronics, nanoelectronics,photonics, and plasmonics. However, ithas also been realised that difficulties inthe experimental resolution and

Principal publication andauthorsA. Bonanni (a),

A. Navarro-Quezada (a), T. Li (a),

M. Wegscheider (a), Z. Matey (b),

V. Hol’y (b), R.T. Lechner (a),

G. Bauer (a), M. Rovezzi (c),

F. D’Acapito (c), M. Kiecana (d),

M. Sawicki (d), and T. Dietl (d),

Phys. Rev. Lett. 101, 135502 (2008).

(a) University of Linz (Austria)

(b) Charles University Prague

(Czech Republic)

(c) CNR-INFM-OGG c/o ESRF GILDA

CRG Grenoble (France)

(d) Polish Academy of Sciences

Warsaw (Poland)

Fig. 112: In a typicalmagnetostrictive actuator, amagnetostrictive material (centralrod) is wrapped in a metal coil,which is in turn housed in an outercasing. By passing a currentthrough the coil, a magnetic fieldis formed within in, causing therod to extend. Alternatively, thedevice may operate in reverse as asensor, where external pressureschange the length of the rod,which in turn generates amagnetic field that induces ameasurable current in the coil.Such sensors are frequentlyemployed for non-destructivetesting applications, such asexamination of suspender cableson bridges.


94


EE SS RR FF

identification of the embeddednanostructures endure and hamper theprogress in the control of themechanisms accounting for associatedand hitherto unexplored nanoassemblyprocesses. Synchrotron radiationmicroprobe [1] and othernanoanalytical synchrotron tools, incombination with high-resolutiontransmission electron microscopy(HRTEM) [2], represent a suitable blendof techniques that, whencomplemented by magneticmeasurements, can lead to a thoroughcharacterisation of magnetically dopedsemiconductors.

The model system (Ga,Fe)N has beenfound from SQUID magnetometry datato be ferromagnetic at RT for aconcentration of magnetic ions above0.4% under the employed growth

conditions. While laboratory high-resolution X-ray diffraction (XRD) doesnot evidence any phase separation inthese samples, powder diffraction XRDmeasurements carried out at thebeamline ID31 by using a photonenergy of 15.5 keV reveal the presenceof new diffraction peaks identified asthe (002) and (111) of the phase ε-Fe3Nwith a Curie temperature TC ~ 575 K.Moreover, it has been possible toconfirm a recent theory [3], accordingto which it is possible to modify thechemical valence of transition metalcations by doping with shallowacceptors (Mg in the case of GaN) ordonors (Si for GaN), and, consequently,affect their aggregation in thesemiconducting matrix. Figure 113indicates that doping with Si doesindeed hinder the formation of Fe-richregions: panel (a) reproduces the fullω/2θ scan and panel (b) a reduced scanwith focus on the range with thediffraction peaks originating from theembedded nanocrystals. This isconfirmed by the HRTEM images in theinset of Figure 113.

The results on the effect of Si on theaggregation of the Fe ions in the GaNmatrix are also corroborated by X-rayabsorption near-edge structure(XANES) measurements at the Fe K-edge carried out at the GILDAbeamline (BM08). As evidenced inFigure 114, XANES data point to theexpected presence of Fe in the Fe3+

charge state in the nominally undoped(Ga,Fe)N samples, whereas theyconfirm the coexistence of Fe3+ andFe2+ ions in the case of doping with Si,supporting the onset of a shift of theFermi level and consequentmodification of Fe charge state in thesystem. An analogous effect has beenverified in the case of doping with Mg.

By combining HRTEM with synchrotronXRD and XANES it has been possible toidentify the distinct ways by which Feincorporates into the GaN lattice givingan important contribution to theelucidation of the origin of theferromagnetic features in magneticallydoped semiconductors. Importantly, thedoping with either acceptors or donorshampers the nanocrystal assemblingand offers a way to functionally controlthe aggregation of magnetic elementsin a non-magnetic host.

Fig. 113: Synchrotronradiation powder XRD

spectra: effect of Sidoping on the

aggregation of Fe ionsinto the GaN matrix.

a) full scan; b) reducedscan focussing on the

region with thediffraction peaks from

Fe-rich embeddednanocrystals. Inset:

HRTEM imagesconfirming the Fe

aggregation hamperedby Si doping.

Fig. 114: XANES spectrawith the region of

interest expanded in theinset: in addition to thepeak at 7114.3 ± 0.1 eV

assigned to the Fe3+

charge state, a shoulderat 7112.7 ± 0.1 eV is

visible in the Si-dopedsamples pointing to the

reduction of a part ofthe Fe ions to the Fe2+

charge state by dopingwith donors.

References[1] G. Martinez-Criado,

A. Somogyi, S. Ramos, J. Campo,

R. Tucoulou, M. Salome, J. Susini,

M. Hermann, M. Eickhoff, and

M. Stutzmann, Appl. Phys. Lett.

86, 131927 (2005).

[2] M. Jamet, A. Barski,

T. Devillers, V. Poydenot,

R. Dujardin, P. Bayle-Guillmaud,

J. Rotheman, E. Bellet-Amalric,

A. Marty, J. Cibert, R. Mattana,

and S. Tatarenko, Nature Mater.

5, 653 (2006).

[3] T. Dietl, Nature Mater. 5, 673

(2006).



Alkali metal-intercalated fullerenesunder high pressure

Alkali metal-intercalated fullerenes areattractive as precursors for newmaterials (high-dimensional polymers)exhibiting both superconducting andhigh hardness properties [1], that canbe formed via high-pressure synthesis.Nevertheless, the strong ionicinteractions between the alkali atomsand the C60 molecule were predictedto give rise to a deformation of theC60 the isolated icosahedral molecule[2]. This deformation could limit thepressure stability of the molecule.However, no experimental evidence ofsuch deformation has been providedbefore our experiments. Therefore,exploration of the P-T phase diagramof alkali metal-intercalated fullerenesis of particular interest because ofthese predictions.

We studied the body-centered cubic(bcc) Rb6C60 (a = 11.54 Å) and Cs6C60(a = 11.79 Å) systems by employingdifferent complementary techniques:X-ray diffraction (XRD), X-rayabsorption fine structure (XAFS) andRaman spectroscopies. XRD (beamlineID27) and XAFS measurements(beamline BM29) were carried out upto 15 GPa at room temperature onboth systems. By comparing thecompressibility of the total volumeobtained by XRD data with thecompressibility of the interstitialvolume between the C60 and the alkaliatoms obtained by XAFS, we observedthat its compression is accompanied bya shape-changing deformation underpressure, this is reported in Figure 115.

The distortion of the molecule isprobably due to an enhancement ofthe Coulomb interaction with thesurrounding ions, effectively increasingthe difference between the distancesfrom each molecule’s centre to those 36carbons closest to the alkali metals(grey atoms in Figure 115), on onehand, and to those 24 carbon atomslocated closest to other C60 molecules(white atoms in Figure 115), on theother hand. This deformation isanalogous to pulling the moleculethrough the three orthogonal axespointing towards the bcc facescontaining the alkali metals.

The deformation of the fullerenemolecule is greater (54% at 15 GPa)for Rb than for Cs intercalation, thistoo confirmed by our ab initiocalculations for both systems.

In spite of such slight deformation dueto the pressure enhanced Coulombinteraction, Raman measurementsperformed under pressure on Cs6C60indicated that the C60 cage is preservedup to a pressure two times higher(45 GPa) than for pristine C60, whichwas observed to amorphise at 22 GPa.

Conversely, XRD, XAS (beamline ID24)and Raman spectroscopymeasurements for Rb6C60 under HP-HTconditions showed the existence of afirst-order phase transition at around35 GPa from the bcc towards ahexagonal phase accompanied by adiscontinuous decrease of the unit cellvolume. Both XRD and Raman datacollected under pressure are shown inFigure 116. The lattice parameters ofthe hexagonal unit cell (a = 8.360 Åand c = 14.830 Å) are compatible witha distance between moleculespreviously observed in the case ofpolymerised C60. In addition, incorrespondence with the transition,the Raman data show a discontinuouschange in the frequency of theintramolecular vibrations. This isprobably associated with the breakingof some intramolecular bondsconcomitant with the formation ofintermolecular covalent bonds, as

Principal publications andauthorsR. Poloni (a), G. Aquilanti (b),

S. Le Floch (c),

M. V. Fernandez-Serra (d),

S. De Panfilis (e),

P. Toulemonde (f), D. Machon (c),

G. Morard (b),

D. Martinez-Blanco (g),

W. Crichton (b), S. Pascarelli (b),

A. San-Miguel (c), Phys. Rev. B 77,

035429 (2008); Phys. Rev. B 77,

205433 (2008).

(a) ICMAB-CSIC, Barcelona (Spain)

(b) ESRF

(c) LPMCN, University of Lyon

(France)

(d) Stony Brook, New York (US)

(e) INFM-CNR, Roma (Italy)

(f) Institut Néel, CNRS and UJF,

Grenoble (France)

(g) Universidad de Oviedo (Spain)

Fig. 115: a) Pressure evolution ofthe XAFS signal; b) Pressure-induced distortion (amplified by afactor of 30) of the C60 molecule at15 GPa observed in Rb6C60 andCs6C60.


96


EE SS RR FF

occurs in cycloaddition processescharacteristic of the C60polymerisation. Hence, we suggestthat the hexagonal HP phase could beassociated to the formation of 2Dpolymers in the (001) plane of thehexagonal structure.

The lower pressure stability of thecubic phase (and of the molecularcharacter) observed in Rb6C60compared to Cs6C60 is concomitantwith a more pronounced distortion ofthe molecule in the first compoundcompared to the latter. Nevertheless,both fullerides show a much higherpressure stability of the molecularcharacter of the system, compared tothe solid C60. We therefore infer thatthe presence of such ionic host-guestinteractions could be responsible forthe high pressure stabilisation of theC60 molecules.

Fig. 116: a) Pressure evolutionof XRD, and b) Raman

measurements for Rb6C60. Theoccurrence of a phase

transition at around 35 GPa isindicated by the appearance of

new Bragg reflections andRaman lines, respectively.

References[1] Connétable et al., Phys.

Rev. Lett. 91, 247001 (2003).

[2] Andreoni et al., Phys. Rev.

Lett. 68, 823 (1992).

Isotopic effect in extended X-rayabsorption fine structure of germanium

The structural, electronic and dynamicalproperties of crystals are mainlydependent on the atomic number ofthe constituent atoms. However, theisotopic composition has subtle butnon-negligible influence on some basicproperties, like density, phonon widths,and electronic energy gaps [1]. Isotopiceffects are relevant not only for theirbasic scientific interest, but also forseveral possible technologicalapplications [2].

The dependence of the dynamicalproperties of crystals on the isotopiccomposition is of primary importance.The force constants depend on atomicspecies and crystal structure. However,the zero-point amplitude of atomicvibrations is also influenced by the

nuclear masses, the lighter isotopesundergoing larger oscillations than theheavier ones. As a consequence ofanharmonicity, the difference in zero-point amplitude of motion reflects on adifference of interatomic equilibriumdistances and lattice parameters [1]. Inthe case of germanium, the expectedrelative change in the lattice parameterbetween 70Ge and 76Ge is as small as∆a/a ≈ 5 × 10–5. Note that these effects,of genuine quantum origin, disappearprogressively when temperatureincreases.

An investigation of the isotopic effecton the amplitudes of nearest-neighbours relative vibrations (parallelmean square relative displacement(MSRD)) and on the nearest-neighboursaverage distance in powdered samplesof 70Ge and 76Ge has been performedby extended X-ray absorption finestructure (EXAFS) spectroscopy.

Two highly isotopically enriched Gesamples with the degrees ofenrichment 98.2% for 70Ge and 99.9%for 76Ge were produced at the

Principal publication andauthorsJ. Purans (a,b), N.D. Afify (a),

G. Dalba (a), R. Grisenti (a),

S. De Panfilis (c,d), A. Kuzmin (b),

V.I. Ozhogin (e), F. Rocca (f),

A. Sanson (f), S.I. Tiutiunnikov (g)

and P. Fornasini (a), Physical

Review Letters 100, 055901:1-4

(2008).

(a) Department of Physics,

University of Trento (Italy)

(b) Institute of Solid State Physics,

Riga (Latvia)

(c) Res. Centre Soft INFM-CNR,

Roma (Italy)

(d) ESRF

(e) Res. Centre “Kurchatov

Institute”, Moscow (Russia)

(f) CNR-IFN and FBK-CeFSA,

Trento (Italy)

(g) JINR, Dubna (Russia)

Fig. 117: Difference of the EXAFSMSRDs as a function of

temperature for the two isotopes70Ge and 76Ge. The solid line shows

the difference between twoEinstein models.



References[1] M. Cardona and

M.L.W. Thewalt, Rev. Mod. Phys.

77, 1173 (2005).

[2] V.G. Plekhanov, Prog. Mater.

Sci. 51, 287 (2006).

In situ redispersion of platinumnanoparticles on ceria-based oxide forvehicle exhaust catalystsSupported precious metals are used tofacilitate many industrial catalyticprocesses. Platinum (Pt) in particular isused for the cleaning up of vehicleexhaust emissions. When the vehicleexhaust catalyst is exposed to hightemperatures (~800°C and above), thehighly dispersed metal nanoparticlesagglomerate and sinter, decreasing theactive surface area, which causes a lossof catalytic activity (i.e. deactivation).Exhaust gases exiting from gasolineengines change quickly duringoperation. Temperatures can risetransiently to around 1000°C and theexhaust gas composition itselffluctuates quickly between oxidativeand reductive compositions. Hence,in situ dynamic observation on thesintering and redispersion phenomenaof the precious metal in theautomotive catalysts is very importantindeed.

Real-time observation of thesintering/redispersion behaviour of Ptwas made possible by the fluorescenceyield variant (Turbo-XAS) of energydispersive XAFS developed byPascarelli et al. at ID24 [1].

An in situ cell optimised forfluorescence detection was designedfor this experiment [2]. Fluorescenceyield Turbo-XAS data collected at thePt LIII edge on 2 wt% Pt/Ce-Zr-Y mixedoxide (referred to as CZY) catalystsunder in situ conditions are shown inFigure 119. The signal to noise ratio onthis data is improved with respect toour previous transmission XAS studyon Pt/CZY, which showed that thecombination of low levels of Pt in thecatalysts, with high levels of heavy,absorbing, elements such as Ce and Zrseverely compromises theconventional, transmission basedexperiments, making quantitativeanalysis very difficult [3].

Kurchatov Institute (Russia), and theirGe K-edge EXAFS spectra recorded withhigh accuracy from 20 to 300 K, atbeamline BM29.

The difference of the MSRD values fortwo isotopes has been clearly evidenced(Figure 117) and is in good agreementwith a behaviour expected from theEinstein model based on the singleforce constant k0 = 8.496(40) eV/Å2 andtwo characteristic frequencies: 7.70(2)THz for 70Ge and 7.39(2) THz for 76Ge.

The effect of isotopic mass has alsobeen revealed in thermal expansion(Figure 118). The zero-point values ofthe nearest-neighbours averagedistance measured by EXAFS areconsistent with the values of distancebetween average positions measured

by Bragg diffraction, once the effects ofvibrations perpendicular to the bondare taken into account. The possibilityof detecting relative distance variationssmaller than 10 femtometres by meansof a conventional transmission EXAFSapparatus and a standard procedure ofdata analysis has been demonstrated.

Principal publication andauthors Y. Nagai (a), N. Hara (b),

K. Dohmae (a), Y. Ikeda (b, g),

N. Takagi (c, g), T. Tanabe (a),

G. Guilera (d, h), S. Pascarelli (d),

M.A. Newton (d), O. Kuno (e, i),

H. Jiang (e), H. Shinjoh (a),

S. Matsumoto (f), Angew. Chem

Int. Ed. 47, 9303 (2008).

(a) TOYOTA Central R&D Labs.,

Inc., Aichi (Japan)

(b) TOYOTA Motor Europe

Technical Centre, Zaventem

(Belgium)

(c) TOYOTA Motor Corporation

Higashi-Fuji Technical Center,

Shizuoka (Japan)

(d) ESRF

(e) TOYOTA Motor Engineering &

Manufacturing North America,

Inc., Michigan (USA)

(f) TOYOTA Motor Corporation,

Toyota (Japan)

(g) Current Address: (f)

(h) Current Address: ALBA-CELLS,

Barcelona (Spain)

(i) Current Address: (c)

Fig. 118: Difference of the nearest-neighbour average interatomicdistance in 76Ge and 70Ge,determined from EXAFS analysis(diamonds), compared with thedifference of distances betweenaverage positions determined fromX-ray backscattering (circles).

Fig. 119: Serial time-resolved Pt LIII edgeXANES spectra (6 seconds each) of2 wt% Pt/CZY catalyst under cyclicaloxidising/reducing. Variation in the firsthundred spectra under oxidising/reducingatmosphere at around 400ºC.


98


EE SS RR FF

Figure 120 shows the variation of thewhite line peak height of thenormalised XANES for the fresh Pt/CZYcatalysts under cyclicaloxidising/reducing condition at400 ~ 800°C. ∆I in Figure 120 denotesthe difference between the white linepeak height of the oxidised andreduced samples. From a previousexperiment [3], we found that the ∆Iincreased with decreasing Pt particlesize. Using the correlation betweenthe Pt particle size and ∆I, we observedthat the Pt particle size changed withtemperature. The Pt particle size ofthe fresh catalyst increases from 1.8 to4.4 nm while temperature is increasedfrom 400 to 800°C, but then decreasesto 2.7 nm as temperature is lowered to550°C. It is particularly interesting tonote that the sintering/redispersionphenomena occur reversibly and canbe controlled using the temperature.During the redispersion process,the Pt particle size was recovered from4.4 to 2.7 nm at first, and then from5.8 to 3.4 nm. In contrast, only facilesintering of Pt particles occurs inalumina supported Pt catalysts(Pt/Al2O3), and Pt redispersion is notachieved at any temperature.Therefore, the Pt redispersion on CZYcan be reasonably attributed to thestrong Pt-ceria support interaction [4].Through further kinetic measurements,using XAS and in situ transmissionelectron microscopy, an atomicmigration mechanism is shown toaccount for the observed redispersionthrough the trapping of atomic Ptspecies at sites on the Ce support that

exhibit a strong Pt-oxide-supportinteraction.

In conclusion, we have made the firstin situ time-resolved fluorescence XASmeasurements from a dilute sampleusing the Turbo-XAS acquisition mode.This technique allowed us to shed lighton catalytic processes on the secondtimescale for Pt supported on a ceria-based oxide. Analysis of the data ledto the discovery of an unexpectedphenomenon, i.e. an efficientoxidative redispersion of Ptnanoparticles during quick(~ 60 seconds) redox cycling. Thisredispersion process does lead to atangible potential for incorporationinto “on board” methodology forextending vehicle catalyst lifetimethrough curtailing or reversing theeffects of metal sintering duringoperation.

Fig. 120: Temporal dependence of the white line peak height of thePt LIII edge XANES for the fresh 2 wt% Pt/CZY catalysts under

oxidising/reducing atmosphere at 400 ~ 800°C and the schematicrepresentation of the sintering/redispersion behaviour.

4 or 20% O2/He gas and 3% H2/He gas were alternately introducedinto the cell every 60 seconds throughout the measurement.

References[1] S. Pascarelli, T. Neisius,

S. De Panfilis, J. Synchrotron Rad.

6, 1044 (1999).

[2] G. Guilera, B. Gorges,

S. Pascarelli, H. Vitoux,

C. Prestipino, Y. Nagai, N. Hara,

submitted to J. Synchrotron Rad.

[3] Y. Nagai, N. Takagi, Y. Ikeda,

K. Dohmae, T. Tanabe, G. Guilera,

S. Pascarelli, M.A. Newton,

H. Shinjoh, S. Matsumoto, AIP

conference proceedings. 882, 594

(2007).

[4] Y. Nagai, T. Hirabayashi,

K Dohmae, N. Takagi, T. Minami,

H. Shinjoh, S. Matsumoto, J. Catal.

242, 103 (2006).



Structural determination of neptuniumredox species in aqueous solutions

Neptunium (93Np) is one of the mostcritical elements for the geologicaldisposal of high-level radioactivewastes because of its considerablecontent in such wastes, and the highradioactivity, half-lives andradiotoxicity of its nuclides. From achemist’s point of view, it is also a veryinteresting element because of itsdiversity of oxidation states from III toVII [1]. Whether Np may be retained inthe waste repository for millions ofyears, or whether it will migrate to thebiosphere, depends heavily on itschemical forms (speciation). Weassume that oxidation state will have astrong influence on the speciation.Therefore detailed knowledge aboutthe interrelation between oxidationstate and the structure of chemicalspecies is critical for the developmentof safe nuclear waste repositories.This motivated us to perform X-rayabsorption fine structure (XAFS)studies to determine the complexstructure of Np species in aqueoussolutions at different oxidation states.The experiments were performed atthe Rossendorf Beamline (BM20), theonly beamline at the ESRF where suchstudies with aqueous Np samples canbe performed.

The solution samples investigated inthis study were prepared by dissolvingNpVIO2(ClO4)2·nH2O into aqueoussolutions, to give a Np concentrationof 0.04 M. The oxidation state of Np inthe samples was electrochemicallyadjusted. Np LIII-edge (17.625 keV)XAFS measurements were carried outin transmission mode using a Si(111)double-crystal monochromator, andtwo Pt-coated mirrors for rejection ofhigher harmonics.

Figure 121 shows the k3-weighted NpLIII-edge extended X-ray absorptionfine structure (EXAFS) spectra forNp(IV), -(V), and -(VI) in 1.0 M HClO4(left), and their corresponding Fouriertransforms (right). Perchlorate isknown to be a noncomplexing ligandto actinide ions in aqueous solution,and hardly coordinates to the primarycoordination sphere of actinide ions [2].

Accordingly, the chemical speciesformed in aqueous HClO4 solution areconsidered to be pure hydrate species.The EXAFS spectrum of Np(IV) iscomposed of a single oscillationpattern, giving a single peak at 1.9 Åin the Fourier transform. The structuralparameters determined by shell fittingare a Np-O distance (RNp-O) of 2.40 Åwith a coordination number (CN) of10.4, corresponding to watermolecules of the primary coordinationsphere. The EXAFS spectra for the twohigher oxidation states, V and VI,exhibit more intricate oscillationpatterns and the correspondingFourier transforms show severalsignificant backscattering peaks. Themost prominent peak at around 1.4 Åarises from the single scattering of thedouble-bond axial oxygen atoms (Oax),indicative of the neptunyl unit(NpO2

n+). The following two peaks ataround 1.9 Å arise from oxygen atomsof the water molecules in the neptunylequatorial plane (Oeq(H2O), while thesmall but sufficiently distinguishablepeak at 2.9 Å is due to the multiplescattering of Oax, respectively. Theshell fitting results reveal that theneptunyl(V) ion is coordinated in itsequatorial plane by 5.2 watermolecules with O atoms at a distanceof 2.49 Å, whereas the neptunyl(VI)ion is coordinated by 5.3 water

Principal publication andauthorsA. Ikeda-Ohno (a,b),

C. Hennig (a), A. Rossberg (a),

H. Funke (a), A.C. Scheinost (a),

G. Bernhard (a), T. Yaita (b),

Inorg. Chem. 47, 8294 (2008).

(a) Institute for Radiochemistry,

Forschungszentrum Dresden-

Rossendorf (Germany)

(b) Synchrotron Radiation

Research Center, Japan Atomic

Energy Agency (Japan)

Fig. 121: a) k3-weighted Np-LIII edge EXAFS spectra, and b) their corresponding Fouriertransforms for Np(IV), -(V), and -(VI) in 1.0 M HClO4. Solid lines represent experimentaldata and dotted lines represent theoretical fit. The line colours reflect the actualcolour of sample solutions.


100


EE SS RR FF

molecules with a shorter interatomicdistance of 2.42 Å.

Our EXAFS data show therefore astructural rearrangement of Nphydrate species with oxidation state asillustrated in Figure 122: both Np(V)and (VI) ions exist predominately aspentaaquo neptunyl complexes,

[NpO2(H2O)5]n+ (n = 1 for Np(V) and2 for Np(VI)), whereas the Np(IV) ionforms a spherically coordinateddecaaquo complex, [Np(H2O)10]4+.It is this drastic change in complexstructure between Np(IV) and Np(V),which makes the transition betweenthese two redox states almostirreversible (∆E ~0.9 V), while thetransition between Np(V) and Np(VI)requires no structural change, hence isquasi-reversible (∆E = ~0.2 V). Theseresults support our initial hypothesisthat there is a strong relationshipbetween the electrochemicalbehaviour of Np and its complexstructure.

Fig. 122: Structural rearrangementof Np hydrate species through the

redox reaction.

References[1] Z. Yoshida, S.G. Johnson,

T. Kimura, J.R. Krsul, in The

Chemistry of the Actinide and

Transactinide Elements (3rd Ed.),

L.R. Morss, N.M. Edelstein,

J. Fuger (Eds.), Springer,

Dordrecht, The Netherlands, 699

(2006).

[2] L. Sémon, C. Boehme, I. Billard,

C. Hennig, K. Lützenkirchen,

T. Reich, A. Rossberg, I. Rossini,

G. Wipff, ChemPhysChem 2, 591

(2001).

AcknowledgementThis work was supported by the Deutsche Forschungsgemeinschaft (DFG) under Contract HE2297/2-1.

Kinetics and thermodynamics ofsiderite dissolution at 300 bar between50°C and 100°C

Iron-bearing minerals are reactivephases of the subsurface environment.Siderite (FeCO3) is one of the mostabundant of these and it plays a majorrole in the deep geological storage ofCO2, both by its presence in thereservoir rock where the injectionmight be done, and by its ownconstitution, being a form ofprecipitated CO2 during its injection indeep aquifers. It is thus of utmostimportance to precisely model thestability and dissolution kinetics ofsiderite under hydrothermalsubsurface conditions.

The dissolution of siderite FeCO3in acidic hydrothermal conditions (50-100°C, 300 bar, 0.1 M HCl) hasbeen studied by X-ray AbsorptionSpectroscopy. The absorption spectraare the data on which our

methodology is based. Taken at the FeK-edge, and measured on the aqueoussolution in contact with the dissolvingmineral they allow:

• determination of the dissolution rateof siderite, as a function of time, fromthe absorption value of the solution.• determination of the speciation ofdissolved iron by comparing theabsorption spectra with referencespectra of iron complexes inhydrothermal conditions.• in situ measurement in a highpressure-high temperature autoclavewhich permits working withmacroscopic samples (siderite samplesare millimetric single crystals) +50 mm3 of solution, [1]). Results aredirectly related to subsurface CO2storage conditions.

Principal publication andauthorsD. Testemale (a), F. Dufaud (b),

I. Martinez (b), P. Bénezeth (c),

J.-L. Hazemann (a), J. Schott (c),

F. Guyot (b,d), Chemical Geology,

doi:10.1016/j.chemgeo.2008.08.01

9 (2008).

(a) Institut Néel and FAME/ESRF,

CNRS, Grenoble (France)

(b) CO2 storage research center.

Institut de Physique du Globe,

Paris (France)

(c) Laboratoire de Mécanismes

des Transferts en Géologie, CNRS-

IRD-OMP, Toulouse (France)

(d) Institut de Minéralogie et de

Physique des Milieux Condensés,

CNRS-Paris6-Paris7, Paris (France)



References[1] D. Testemale, R. Argoud,

O. Geaymond, J.-L. Hazemann,

Rev. Sci. Instrum., 76, 043905

(2005).

[2] J. van der Lee, Technical

Report LHM/RD/98/39, CIG, Ecole

des Mines de Paris,

http://chess.ensmp.fr/

[3] S. V. Golubev, P. Bénézeth,

J. Schott, J.-L. Dandurand,

A. Castillo, Chem. Geol., Special

issue on CO2 geological

sequestration, in press.

Figure 123 and Figure 124 summarisethe data collected in our experiment.Using the molalities of Fe(II) in theaqueous phase (Figure 123), ageochemical model was built usingChess [2] from which kinetics rateconstants are derived at eachtemperature, as well as the activationenergy of the system. Thanks to thismodel, the distribution of ironaqueous species is also calculated, andcompared with a very good agreementwith the speciation derived from theXANES measurements (Figure 124).This XANES signal is characteristic ofthe hydration structure of dissolvediron, and it is interpreted thanks to aparallel study where the speciation ofiron is determined as a function oftemperature, pressure and chlorideconcentration by EXAFS analysis andXANES ab initio calculations.

The kinetics rate constants obtained inthese closed-reactor pH-varyingexperiments compare favourably withvalues from recent experiments carriedout at constant chemical affinity [3]:this indicates that we can retrievekinetics parameters from batch micro-reactor experiments conducted atsynchrotron radiation facilities thatcan be coupled with spectroscopicspeciation results. These data are nowincorporated in geochemical softwaremodelling fluid-rock interactions underCO2 storage conditions.

Fig. 123: Iron molality as a functionof time for siderite dissolution at300 bar and 0.1 m HCl at a) 50°C,b) 75°C, c) 100°C. The experimentaldata points are shown as filledcircles (empty circles in c) are for a1 mol/Kg NaCl solution). Thecontinuous lines are calculated withthe Chess software using thedissolution rate constants measuredin this study [2].

Fig. 124: Comparison of XANES spectra obtained for hydrated Fe2+ (solid line) and atthe end of the siderite dissolution experiment at 100°C (dashed lines: without NaCl;

filled circles: with 1 mol kg–1 NaCl). The three spectra are very close and mainly differin the 7130–7150 eV region (indicated by the arrow), with slight differences in the

7170–7200 eV region. All the spectra show the dominance of hydrated Fe2+ species.The differences are interpreted by minor occurrence of FeCl+ and/or FeCl2 species,

more significant, as expected, in 1 mol kg–1 NaCl.


102

X-ray Imaging and Optics

EE SS RR FF

Introduction

The importance of X-ray imagingtechniques for innovative research isnow widely recognised among thesynchrotron radiation user communities.A number of X-ray imaging beamlinesare being built in new synchrotronfacilities that are emerging all aroundthe world. Two new biomedicalbeamlines have entered into operationat the end of 2008, in Canada andAustralia. The upgrade programmes ofmajor facilities such as the ESRF and APSstrongly emphasise X-ray imaging as oneof the five “highlight areas” and oneof the two “compelling avenues forresearch” respectively.

This gain in visibility results fromthe increased number of scientificcommunities that use X-ray imagingtechniques for their research, asreflected by the present chapter. Thehighlight contributions originate fromseveral scientific areas, which include: • Biomedical applications, with theinvestigation of novel contrast agents,an improved knowledge of the effect ofmicrobeams, potentially very importantfor clinical radiotherapy applications,and the characterisation of stem cellsusing infrared spectromicroscopy.• Materials Science studies, with theobservation of intergranular stresscorrosion cracking in a grain-mappedpolycrystal, and the local structure ofGaGdN, investigated with a microprobebeam and several complementarytechniques. • Cultural Heritage, where synchrotrontechniques revealed the unexpectedpresence of oil in very ancient paintingsof the Bamiyan Buddhist murals, thepresence of relics hidden in MiddleAge artistic objects, and where phasecontrast imaging was used forpaleontological purposes, to study fossilevidence hidden in Early Cretaceousopaque amber.

• Environmental Sciences with theinvestigation, using a 80 x 90 nm spot, ofcometary grains collected during theNASA Stardust mission, or helping tocharacterise the impact of carbonnanotubes on human health.

Another point we wish to emphasise isthe importance of the effort, made bythe ESRF staff and users, to implementnew (or new combinations of) imagingtechniques, opening new ways forscientific research. This is exemplified,by the first results obtained withan improved diffraction tomographymethod, as well as the biomedical phasecontrast images provided by a gratinginteferometry-based technique. Thelatter can be applied to a wide variety ofmaterials: Figure 125 shows a slice ofkiwi fruit in different contrast modes,obtained from a single phase-steppinginterferogram series.

The trends, challenges and newopportunities associated withsynchrotron radiation-based imagingare clearly defined for the next fewyears; if we wish to be able tocontribute substantially to emergingtopics in areas like nanosciences,health and environment preservation,or cultural heritage, we must improveboth the spatial (towards the nanometrescale) and temporal resolutions (for theinvestigation of evolving systems), andbe able to achieve higher chemicalsensitivity for micro- or nano-analysis.In addition, we have to pursue theimplementation of new imagingtechniques and the in situ combinationof techniques. All this will be performedin a highly competitive internationalcontext.

The ESRF upgrade programme, whereX-ray imaging plays a substantial role,intends to give an answer to these




trends and challenges, through severalprojects, which include very long (up to200 m) beamlines for Nano Imagingand Nano Analysis (NINA), and paralleland coherent beam Micro-ImagingApplications (MIA). In addition, the nexttwo years will be crucial for theimplementation of clinical radiotherapyprogrammes, at the recently refurbishedbiomedical beamline. All these upgrades,coupled with the scientific know-howalready acquired using X-ray imaging atthe ESRF, will constitute a greatincentive for both the ESRF staff andusers to push the present limits of thetechniques to reach new, currentlyunobtainable, scientific heights.

J. Baruchel

New X-ray imaging methods

3-D imaging of multi-phase polycrystalline materials bydiffraction microtomography

Powder X-ray diffraction is a techniqueused to solve new crystallinestructures, identify phases, analysephase transformation, and examinemicro-structural features. With therecent improvements in X-ray opticsand detection, powdermicrodiffraction experiments can becarried out that provide researcherswith images of high lateral resolutionin a reasonable scanning time.However, such two-dimensionalmappings only provide a globalintegral of all the diffracted intensityalong the X-ray path, thus preventingus from gaining depth resolutioninformation. To study bulk materials,we also require depth resolutioninformation. A local structural probeproviding depth resolution images ismandatory in cases where thematerials present structuralheterogeneities. Three-dimensionaldiffraction tomography has alreadybeen demonstrated for severalapplications: WAXS was used to studysoft tissue achieving millimetreresolution for biological applications,

and SAXS was used to study polymers.Furthermore, Laue diffraction andtopotomography was used to build 3Dmappings of individual grains [1,2].

Here we introduce a new scanningmethod for synchrotron X-raydiffraction tomography in order toreconstruct cross-section images ofunidentified phases in nanomaterialsand polycrystalline materials.The technique is a further extension ofthe X-ray fluorescence micro-tomography technique that hasrecently been developed at ID22.It involves simultaneous measurementof the absorption, fluorescence anddiffraction data while translating(along y) and rotating (around ωω) asample illuminated by a focused -orpencil- beam (Figure 126). Thanks tothe high flux combined with the use ofa fast readout and low noise camera(FRELON) equipped with a taper, thescanning time for a 100 µm3 sample isabout 6 hours. This process produces astack of diffraction images,fluorescence spectra and absorption

Principal publication andauthorsP. Bleuet (a), E. Welcomme (b),

E. Dooryhée (c), J. Susini (a),

J-L. Hodeau (c), P. Walter (b),

Nature Materials 7, 468 (2008).

(a) ESRF

(b) Centre de Recherche et de

Restauration des Musées de

France, CNRS-UMR 171, Palais du

Louvre, Paris (France)

(c) Institut Néel, CNRS-UPR 2940,

Grenoble (France)

Fig. 125: Gratinginterferometric radiographs ofa slice of kiwi fruit (ID19,27 keV): a) absorption image;b) differential phase contrastimage (the scattering is sostrong in the centre that thephase is not well defined);c) dark-field (i.e., scattering-contrast) image, showingfeatures that are not visible inthe absorption and the phaseimages. The field of view ofeach frame is 36.5 mm wideand 11.5 mm high;the detector pixel size was30 µm (Image courtesy:T. Weitkamp, E. Reznikova,C. David et al.).


104


EE SS RR FF

sinograms. These sinograms are arepresentation (y,ω) of a particularcrystalline phase, elementaldistribution, and attenuationcoefficient, respectively. Starting fromthis set of sinograms, a mathematicalinversion formula gives rise to cross-section images for each one of thethree modalities.

This multi-modal tomographic schemewas first tested on a textbook powderwith various grain sizes: a 300 µmcapillary was filled with a mixture ofchalcedony and iron pigmentscontaining hematite (α-Fe2O3).Chalcedony is composed of longquartz micro-fibres, generally less than100 nm in diameter. A single cross-

section image was measured andreconstructed (Figure 127). Even ifchalcedony is by far the dominantphase, at least two types of iron grainswere observed: hematite and sideriteon the one hand, and hematite andphylosilicate phases includinggreenalite on the other hand. For eachgrain, a diffraction diagram can beextracted from extremely tiny volumesof powder (as small as the voxel size),allowing subsequent data simulationor structural refinement.

Another carbon-based sample ofsimilar composition and density wasalso analysed. Five phases wereidentified and located within thesample (Figure 128). Well-crystallisedcubic diamond was located in thecentral part of the sample, embeddedwithin an amorphous carbon sp3 phasematrix. Furthermore, crystallised ferrite(α-Fe) grains located at the samplesurface were identified as well as onesingle impurity grain of calcite(CaCO3), outlining the sensitivity of themethod. Ferrite comes from somecontamination from the razor bladeduring sample extraction while calciteis probably a dust particle in contactwith the pressure cell. For this weakly-absorbing material, absorption andfluorescence tomography do notprovide any contrast, while diffractiontomography reveals the distribution ofphases (crystalline and amorphous)inside the sample.

Pencil-beam diffraction tomographybridges the gap between thequantitative structural global probe,such as X-ray and neutron diffractionmethods, and the local compositionalprobe, such as X-ray fluorescence andabsorption computed-tomographytechniques or electron diffraction.Though limited by the number ofcrystallites in the gauge volume andtheir size, this method presents anumber of advantages. It can obtaincontrast where other modalitiescannot, provide multi-modal imagesfor a complete sample characterisationand allow the reconstruction ofunknown crystalline phases withoutany a priori information. Furthermore,it is extremely sensitive due to thequantity of diffraction imagesrecorded at different positions andangles of the sample.

Fig. 126: The ID22sample environment.

Fig. 127: Reconstructed cross-sections corresponding to:

a) the entire diffractedintensity; b) chalcedony;c) hematite and siderite;

d) hematite and greenalite;e) fluorescence imaging;

f) absorption imaging.

Fig. 128: Cross-section reconstruction ofthe diamond sample. Glass capillary

(gray), sp3 amorphous (red),cubic diamond (red), ferrite (blue)

and calcite (white).

References[1] W. Ludwig, E.M. Lauridsen,

S. Schmidt, H.F. Poulsen,

J. Baruchel. J. Appl. Cryst., 40, 905

(2007).

[2] B.C. Larson, W. Yang, G.E. Ice,

J.D. Budai and J.Z. Tischler. Nature

415, 887 (2002).



Biomedical phase-contrast tomography

X-ray computed tomography (CT) is aninvaluable three-dimensional non-destructive imaging method withnumerous applications in life andmaterial science. The technique yieldsexcellent results where highlyabsorbing structures are embedded ina matrix of relatively weakly absorbingmaterial (like bones and tissue in thehuman body). However, for thedifferentiation of weakly absorbingsoft matter, e.g., pathologies in thesoft-tissue structure, the contrastachieved by the currently existing,absorption-based X-ray methods islimited, unless contrast agents areused.

Phase-sensitive X-ray imaging canovercome these limitations and yieldgood contrast for soft-tissue structures.Phase-contrast imaging relies onrefraction, or changes in the angulartrajectory of X-rays. Just as light raysbend when they enter water from air,X-rays deflect as they travel throughobjects of varying densities. We usedgrating interferometry at ID19 tomeasure this effect and thus obtaindetailed radiographic phase-contrastX-ray images. This approach can beextended to three-dimensionalphase-contrast computed tomography(PC-CT) by rotating the specimenaround an axis perpendicular to theX-ray beam and recording severalhundred differential phase-contrastprojections [1,2]. Using speciallydeveloped filtered back-projectionalgorithms, a quantitative three-dimensional volume data set of thespecimen can then be reconstructed bycomputer.

To evaluate the potential impact ofthis novel grating-based PC-CT methodparticularly for biomedical imagingapplications, we carried out several in-vitro synchrotron experiments onbiological specimens. As an example,Figure 129 shows results for a rat brainspecimen, fixed in formalin solution.The data reveals two interestingfindings. Firstly, there is good contrastbetween the brain’s white and greymatter. This can be seen in the imageof the region containing the rat’scerebellum (centre). Although such

differentiation is rarely possible withabsorption-based X-ray CT scans, ourmethod clearly resolves these smalldensity differences between the twobrain tissues. Secondly, pathologiessuch as tumours can be clearlydistinguished. The right panel ofFigure 129 shows a section through aregion in the brain containing atumour, where we can clearly see thepresence of a 9L gliosarcoma (rat braintumour) in the right caudate nucleusof the brain, even though no contrastagent was used in the experiments.This again confirms the enhancedtissue sensitivity of the method.Figure 130 shows a comparisonbetween the phase-contrast CT results(left panel), and the correspondingtomographic slice obtained byconventional absorption-based CTreconstruction (right panel). Bothresults were obtained from the samedata set, i.e., with the same exposuretime.

With this work we have demonstratedhow grating-based phase-contrast CTcan yield images of biomedicalspecimens with unprecedented softtissue sensitivity thanks to the highspatial coherence of the synchrotronbeam at ID19. In particular, we haveshown that this X-ray method can beused to discern between subtle details

Principal publication andauthorsF. Pfeiffer (a,b), C. David (a),

O. Bunk (a), T. Donath (a),

M. Bech (c), G. Le Duc (d),

A. Bravin (d), and P. Cloetens (d),

Phys. Rev. Lett. 101, 16810 (2008).

(a) Paul Scherrer Institut, Villigen

(Switzerland)

(b) EPFL, Lausanne (Switzerland)

(c) Niels Bohr Institute,

Copenhagen (Denmark)

(d) ESRF, Grenoble (France)

Fig. 129: Post mortemthree-dimensional X-ray phase-contrast CTimages of a rat brainbearing a 9Lgliosarcoma (rat braintumour), obtained atID19. The images showthree-dimensionalrenderings of thereconstructed data set.

Fig. 130: Reconstructedtomographic slicesthrough the rat’scerebellum. a) Phase-contrast CTresults, b) Conventionalabsorption CT results.


106


EE SS RR FF

in the tissue structure of animal brains,an application field which until nowhas almost exclusively been reservedfor other techniques such as histologyand non-functional imagingtechniques. Furthermore, the approachcan be combined with localtomography to obtain a detailed viewof a region of interest within a largesample. We believe that this methodcan be applied to a range of

biomedical investigations immediately,including studies of the developmentof certain pathological featuresassociated with neurodegenerativediseases (Alzheimer’s disease), or thecomplex formation of blood vessels intumour models. These results are alsopotentially interesting from a clinicalpoint of view, since a similar approachcan be implemented with standardX-ray tube sources [3].

References[1] T. Weitkamp, A. Diaz, C. David,

F. Pfeiffer, M. Stampanoni,

P. Cloetens, and E. Ziegler, Opt.

Express 13, 6296 (2005).

[2] F. Pfeiffer, O. Bunk, C. David,

M. Bech, G. Le Duc, A. Bravin, and

P. Cloetens, Phys. Med. Biol. 52,

6923 (2007).

[3] F. Pfeiffer, C. Kottler, O. Bunk,

and C. David, Phys. Rev. Lett. 98,

108105 (2007).

Biomedical applications

Gold nanoparticles functionalised by gadoliniumchelates as contrast agent for both X-ray computedtomography and magnetic resonance imaging

There is a considerable need for bettercontrast agents both for cancerdiagnosis and for monitoringtreatments. Commercial contrast agentsused in X-ray imaging and magneticresonance imaging (MRI) – twoanatomic techniques widely used todetect tumours – are not sensitiveenough to be used with small animals.Indeed, the small molecular size of thecontrast agents results in their rapidelimination by the kidneys. Moreover,most of them are specific for oneimaging technique alone, whereas touse a combination of imagingtechniques would be desirable.Numerous nanoparticulate contrastagents have been recently proposed toovercome these limitations. Theirlonger vascular half-life renders themsuitable for long-term studies ofbiodistribution, and to evaluate theanimal models response to a specifictreatment. Another great advantage of

the nanoparticle contrast agents lies intheir ability to display severalcomplementary properties in the sameobject. Consequently, somemultifunctional nanoparticles that canbe detected by several in vivo imagingtechniques have been described.

We have recently developed a synthesisof gold nanoparticles functionalised bygadolinum chelates (Au@DTDTPA-Gd).These contrast agents are composed ofa gold core (mean diameter of~ 2.5 nm) encapsulated within amultilayered organic shell of ~ 150gadolinium chelates (DTDTPA-Gd).These paramagnetic gold nanoparticlesare able to induce a contrastenhancement in magnetic resonanceimages in vitro [1] thanks to thepresence of the gadolinium (III) ionsinside the organic shell. The aim ofthe present study was to investigatethe in vivo contrast enhancement ofAu@DTDTPA-Gd50 nanoparticles(i.e. Au@DTDTPA nanoparticles with~ 50 Gd3+ ions trapped in the organicshell) for both MRI (gadolinium ions)and X-ray imaging (thanks to the strongX-ray absorption of the gold core).

The biodistribution of thesenanoparticles was monitored by MRI(at 7T) and X-ray imaging, followingtheir intravenous injection into miceand rats. X-ray imaging was performedat the biomedical beamline (ID17),

Principal publication andauthorsC. Alric (a), J. Taleb (b),

G. Le Duc (c), C. Mandon (b),

C. Billotey (b),

A. Le Meur-Herland (b),

T. Brochard (c), F. Vocanson (d),

M. Janier (b), P. Perriat (e),

S. Roux (a), O. Tillement (a), J.

Am. Chem. Soc. 130, 5908 (2008).

(a) LPCML, Villeurbanne (France)

(b) Créatis – LRMN, Villeurbanne

(France)

(c) ESRF, Grenoble (France)

(d) Laboratoire Hubert Curien,

Saint-Étienne (France)

(e) MATÉIS, Villeurbanne (France)

Fig. 131: T1-weightedimages of a mouse

before a) and after b)c)the Au@DTDTPA-Gd50

nanoparticles injection.X-ray images in

tomographic mode ofa rat at right kidney

level d) and left kidneylevel e). RK and LK forright and left kidneys

respectively;U for ureter.



References[1] P.-J. Debouttière, S. Roux et

al., Adv. Funct. Mater., 16, 2330

(2006).

[2] C. Alric, R. Serduc et al. Gold

Bull., 41, 90 (2008).

using two monochromatic X-ray beamsbracketing the gold K-edge energy.Both techniques led to the sameobservations: some areas in the bodypresent an enhanced signal due to theaccumulation of paramagnetic goldnanoparticles, this only a few minutesafter injection. The contrastenhancement occurs first in the kidneys(Figure 131) and then in the bladder.Thus, the Au@DTDTPA-Gd50nanoparticles freely circulate within thevascular system without undesirableaccumulation in liver and in spleen andare efficiently cleared by renalelimination. This was confirmed by ICPanalysis: the gold element wasessentially present in the urine andkidneys. Moreover, for both MRI andX-ray in vivo experiments, the contrastinduced by these paramagnetic goldnanoparticles allow their monitoringover a long time period compared withmolecular contrast agents.

The high photons flux of thebiomedical beamline offers twoadvantages over conventional CTscanners. First, the synchrotron imagingtechnique requires only low goldconcentrations (gold content for X-rayimaging is similar to that used in MRI).Thus, the injection of the contrastagent is less invasive. Another keyadvantage lies in the quantification ofthe contrast agent concentration in thetissues. Prior to in vivo imaging studies,in vitro experiments were carried outwith aqueous solutions of Au@DTDTPAnanoparticles containing variousconcentrations of gold (Figure 132a).We can note the concordance betweenthe experimental gold concentration(Exp) and the measurements (Meas) bysynchrotron radiation computedtomography (SRCT). The ability todetermine quantitatively the goldcontent in the tissues was used to

monitor the biodistribution ofparamagnetic gold nanoparticles in vivoand in “real time” (Figure 132b),without sacrifice of the animal.Moreover, another great advantage ofthe ID17 beamline is its ability toperform radiotherapy [2].

In conclusion, this study confirms thatAu@DTDTPA-Gd50 nanoparticles freelycirculate in the vascular system and thatthey can be used in vivo as a bimodalcontrast agent for medical imaging. TheMRI contrast enhancement comes fromthe presence of gadolinium ions in theorganic shell of the nanoparticles,whereas the gold core provides strongX-ray absorption. These nanoparticlescould therefore be a promising tool forpreclinical studies in oncology.

Fig. 132: a) Phantomimages of samplescontaining differentsgold concentrations(‘Exp’) measured bySRCT (‘Meas’). The lastphantom contains Au@DTDTPA-Gd50

nanoparticles.b) In vivo measurementof gold concentration([Au]meas) in thebladder of a rat bySRCT (blue line). B forbladder; K for kidneysand WC for the tubecollecting the urine.

Brain tumour and normal vessel responses tosynchrotron microbeam radiation therapy

Microbeam radiation therapy (MRT) isa pre-clinical form of radiosurgeryinitially dedicated to the treatment ofbrain tumours [1]. MRT requires veryintense X-rays produced by a thirdgeneration synchrotron. The treatmentis based on spatial fractionation of the

radiation dose instead of conventionalclinically-used temporal fractionation.Tumours are irradiated with very highdoses (~150-625 Gy) by an array ofparallel microbeams (~25-50 µmwidth), 100 to 400 µm spaced on-centre.


108


EE SS RR FF

Previous preclinical studies on rodentshave shown that normal tissuepresents a particularly high toleranceto radiation delivered via microbeamsand MRT has been proven to havepalliative and even curative effects oncerebral tumours implanted in ratbrain. These studies showed that MRTyields a better therapeutic index thanconventional, non-spatiallyfractionated, radiotherapy. This hasbeen largely attributed to a strongereffect of microbeam irradiation ontumoural vessels than on normalvessels. It has been hypothesised thatMRT-damaged microsegments inpoorly differentiated neovasculatureof brain tumours may not beefficiently repaired, in marked contrastto the rapid repair of radiation-damaged normal vessels [2,3].

This tumour vessel necrosis would leadto tumour asphyxia and lesion cure.

In this work, we characterised by MRIand histology the acute response of a9L tumour implanted in the mousebrain (D0) and its vasculature tocrossfired MRT performed on day 13(Figure 133).

The mean survival time of 9L tumourbearing mice increased significantly forMRT-treated as opposed to untreatedgroups (x1.34, p<0.0001). A significantincrease of apparent diffusioncoefficient (ADC) was observed24 hours after MRT in irradiatedtumours versus non-irradiated ones.ADC reflects a radiation-inducedincrease in tumour vessel permeabilitywhich could be exploited to specificallydeliver chemo-agents to brain tumoursvia the circulatory system. Indeed, nochange in vessel permeability wasobserved in the contralateral,unidirectionally irradiated,hemisphere. In the untreated group,both tumour size and vessel size index(VSI, i.e., mean vessel diameter)increased significantly (from 7.6 ± 2.2to 19.2 ± 4.0 mm2 and +21%,respectively) between the fourteenthand the twenty-first day after tumourcell inoculation. During the sameperiod, in the MRT-treated group, nodifference in tumour size wasobserved. Vessel size index measuredin MRT-treated group increasedsignificantly (+26%) but only between21 and 28 days of tumour growth. Wedid not observe significant differencein blood volume between the MRT-treated and untreated groups.Histological analysis performed atdifferent periods after irradiationrevealed that MRT presented animportant cytotoxic effect ontumoural cells while sparingunidirectionally irradiated normalbrain tissue (Figure 134). However,tumour vessel labelling (Figure 134)and MRI data suggest that MRT doesnot reduce blood volume in thistumour model (and therefore theblood supply to the tumour remainspractically unchanged), the opposite toexpectations from Dilmanian et al. [2].

To conclude, this study shows that MRTslows the 9L tumour growth in nudemouse brain. MRI and histological

Fig. 133: a) Schematicrepresentation of the

irradiation geometry. Eachmicrobeam array contains

28 microbeams (25 µmwidth, 211 µm on-centre

distance, in-beam dose:500 Gy). The two

orthogonal irradiationsproduce a composite

irradiation volume in thebrain of 6 x 6 x 6 mm at

the tumour site. MRimages from one mouse

obtained 21 days aftertumour implantation:

b) T2-weighted image andc) related apparent

diffusion coefficient,d) blood volume, and

e) vessel size index maps.The tumour, readily visible

in the left hemisphere,presents increased ADC, BV

and VSI.

AuthorsR. Serduc (a,b,c), T. Christen (a,b),

J. Laissue (d), R. Farion (a,b),

A. Bouchet (a,b), B. van der

Sanden (a,b), C. Segebarth (a,b),

E. Bräuer-Krisch (e), G. Le Duc (e),

A. Bravin (e), C. Rémy (a,b) and

E.L. Barbier (a,b).

(a) INSERM, U836, Grenoble

(France)

(b) Université Joseph Fourier,

Grenoble Institut de Neurosciences,

Grenoble (France)

(c) Present address: Centre de

Recherche Cerveau et Cognition,

UMR5549 CNRS-Université

Toulouse, and visiting scientist,

ESRF.

(d) Institute of Pathology,

University of Bern (Switzerland)

(e) ESRF

Fig. 134: HE staining a), c),e) and type IV collagen

immunochemistry b), d), f)of transversal sections of9L tumour implanted in

nude mice.Untreated (a, b)

and MRT-treated (c, d)tumour 21 days aftertumour inoculation.MRT treated at D28

tumour (e, f).



results suggest that the increase insurvival time may be due tocytoreduction rather than to a short-term effect of ionising radiation on 9Ltumour vessels. Tumour vessels weredetected and appeared perfused up to28 days after irradiation. This impliesthat our MRT parameters should beoptimised and that even higher dosescould be delivered to increase the

damage to tumour vessels. In thisstudy, measured microvascularparameters suggest that the actionmechanism of MRT on 9L gliosarcomatumour, at least under ourexperimental conditions, do notinvolve a significant microvascularcomponent and that the cellularprocesses involved require furtherinvestigation.

Principal publication andauthorsM.J. Walsh (a), T.G. Fellous (b),

A. Hammiche (a), W.R. Lin (b),

N.J. Fullwood (a), O. Grude (a),

F. Bahrami (c), J.M. Nicholson (c),

M. Cotte (d), J. Susini (d),

H.M. Pollock (a), M. Brittan (b),

P.L. Martin-Hirsch (a),

M.R. Alison (b) and F.L. Martin (a),

Stem Cells 26, 108 (2008).

(a) Lancaster University (UK)

(b) Queen Mary’s School of

Medicine and Dentistry (UK)

(c) Daresbury Laboratory (UK)

(d) ESRF

References[1] D.N. Slatkin, F.A. Dilmanian,

P. Spanne, Method for microbeam

radiation therapy, US Patent (1994).

[2] F.A. Dilmanian, T.M. Button,

G. Le Duc, Neuro-oncol 4, 26 (2002).

[3] J.A. Laissue, G. Geiser,

P.O. Spanne, et al., Int J Cancer

78, 654 (1998).

Stem cell characterisation in the human intestinederived by mid-infrared light

The human body is comprised ofdifferent cell types, which togetherform complex tissues. Cells are evolvedstructures composed of biomolecules,such as proteins, lipids, DNA, RNA,glycoproteins and carbohydrates; theirmake-up differs with function. Stemcell populations are believed to beslow-cycling, possessing an infiniteproliferative capacity and giving riseto transit-amplifying (TA) cells; thesethen differentiate into functionalterminally-differentiated (TD) cells.

Identification of stem cells is importantin order to understand fundamentaltissue biology. However, robust markersfor adult stem cells remain elusivebecause many approaches have reliedon the expression of a single cell-surface marker. The most regenerativetissue in the human and an importantstem-cell system which should beamenable to characterisation, due to itswell-compartmentalised structure, isthe gastro-intestinal (GI) tractcontaining the crypts of the small andlarge intestine. The GI tract is a modelsystem as the putative stem cells aregenerally believed to reside towardsthe base of the crypt; these divideslowly migrating upwards to form TAcells, which form TD cells after furthercell divisions. It has always beensomething of a conundrum that weseem to know a great deal aboutintestinal stem cells and yet reliablemarkers have proved to be elusive.However, it is generally agreed that inthe small intestinal crypt, the stemcells are located, at approximately cellpositions 4-5 (counting from the cryptbase), whereas in the large intestinalcrypts, they are at the very base.

Because of their different in situ roles,one might expect that these cellpopulations would have differingbiochemical profiles giving differinginfrared spectral fingerprints(Figure 135). Cellular biomoleculesabsorb the mid-infrared (λ = 2-20 µm)via vibrational transitions that arederived from individual chemicalbonds; this may yield richly structured‘‘fingerprint’’ spectra relating tostructure and conformation [1].Synchrotron infrared sources such as atESRF generate a highly collimatedbeam of photons of high brilliancethat may be delivered through a verysmall aperture.

We set out to determine whetherspecific biomolecular fingerprintsderived using infraredmicrospectroscopy could be attributedto different regions. Markeddifferences (including peak location andintensity) were observed throughoutthe infrared spectral “fingerprint”region (900 cm–1 to 1,800 cm–1).Large numbers of variables sometimesmake it difficult to identify underlyingvariance [2]. Such data may beinterrogated using principalcomponent analysis (PCA). PCA wasused for data reduction, and the

Fig. 135: Biomolecules absorb inthe mid-infrared allowinginterrogation of cells with differingbiochemistry. Chemical entitiesmay be tracked in a pixel-by-pixelfashion.


110


EE SS RR FF

output was processed using lineardiscriminant analysis (LDA). Thisallowed new variables to be foundsuch that the ratio of the between-cluster (i.e., category of cell type basedon locations) variance to the within-cluster variance was maximised. PCA-LDA highlighted νsPO2

– (1,080 cm–1) asbeing a major contributor to

segregation, pointing to an importantrole of alterations in the structure ofchromatin.

Infrared spectral image maps allowone to track the spatial distribution ofchemical entities based on levels ofrelative absorbance intensity at awavenumber in a pixel-by-pixelfashion (Figure 136). Tissues analysedat the high resolution (8 µm × 8 µm)possible at ESRF generated such imagemaps that were then superimposedexactly on the interrogated regions(Figure 136). Our findings highlightedνsPO2

– with greatest intensity locatedin a ring in the small intestinal crypt,at approximately cell positions 4-5near the base, whereas in the largebowel crypts, this occurred primarily atcell positions 1-4, at the base. Thissuggests that this approach is apowerful adjunct to ourunderstanding of stem cell biology.

Fig. 136: Stem cellhierarchy of small and

large intestine stem cells.Infrared spectral image

maps highlight νsPO2– in

the different putativeregions.

References[1] M.J. Walsh, M.J. German,

M. Singh, H.M. Pollock,

A. Hammiche, M. Kyrgiou,

H.F. Stringfellow, E. Paraskevaidis,

P.L. Martin-Hirsch and F.L. Martin,

Cancer Lett. 246, 1 (2007).

[2] M.J. German, A. Hammiche,

N. Ragavan, M.J. Tobin,

N.J. Fullwood, S.S. Matanhelia,

A.C. Hindley, C.M. Nicholson,

H.M. Pollock and F.L. Martin,

Biophys. J. 90, 3783 (2006).

Materials science

Grain mapping of stress corrosion cracks by diffractioncontrast tomography

Stainless steels can suffer fromintergranular stress corrosion crackingin certain conditions. Damage occursvia the nucleation and growth ofcracks by localised corrosion ofsusceptible, sensitised grain boundariesunder the action of stress. It is apotential failure mechanism in somepower station components, wheresensitisation can result from heattreatments, or fast neutron irradiation[1]. The susceptibility of grainboundaries to sensitisation depends ontheir structure at the atomic level,which depends on the relative crystalorientations of the grains at theboundary, as well as the orientation ofthe boundary relative to the grains.Certain special boundaries resistsensitisation, and previous experimentsat the ESRF have shown that theseresistant boundaries can form bridgesacross the advancing crack [2],increasing the resistance to stresscorrosion cracking [3]. Hence, grainboundary engineering aims to

improve material performance bymaximising the fractions of resistantboundaries [4].

Diffraction contrast tomography (DCT)is a technique recently developed atthe ESRF for mapping 3D grain shapeand orientation in polycrystallinematerials [5]. A technique related tothe ID11 3D X-ray microscope project,DCT combines microtomography withan analysis of the diffraction spotsarising from grains in the sample.As well as work reported here, othercurrent research projects are using DCTto investigate the growth of grainsduring annealing, the interaction offatigue cracks with microstructures,and on the distribution of strains andstresses in elastically anisotropicmaterials.

In the stress corrosion cracking project,DCT was used to map the grainstructure of a sensitised stainless steelsample (Figure 137). This revealed the

Principal publication andauthorsA. King (a,b), G. Johnson (a,b),

D. Engelberg (a), W. Ludwig (b,c)

and J. Marrow (a), Science 321,

382 (2008).

(a) Manchester University (UK)

(b) ESRF

(c) MATEIS, INSA de Lyon (France)



crystallographic orientation of thegrains on either side of each boundary,as well as the orientation of the grainboundary planes, which provides therequired full macroscopic descriptionof grain boundary structure. BecauseDCT is a non-destructive technique, thesample was then available for asubsequent in situ stress corrosioncracking investigation using standardmicrotomography. Both the grainmapping and in situ crackingobservations were made usingbeamline ID19. We were able to alignthe crack path with the grainboundaries as shown in Figure 137,and thus associate bridges across thecrack with particular grain boundaries(Figure 138). We could then examinethe structure of these special grainboundaries, and compare them to thepopulation of grain boundaries in thesample (~1600 boundaries in total) andalso to the subset of grain boundariesthat failed by stress corrosion cracking.

In conclusion, diffraction contrasttomography was used to map thegrain shapes and orientations in apolycrystalline stainless steel sample,revealing the crystallographiccharacter of the grain boundaries.

We could then grow an intergranularstress corrosion crack in the sample,and make in situ tomographicobservations of how the crackinteracted with the microstructure,revealing which boundary types tendto resist crack propagation.

Principal publication andauthorsG. Martínez-Criado (a),

O. Sancho-Juan (b), N. Garro (b),

J.A. Sans (a), A. Cantarero (b),

J. Susini (a), M. Roever (c),

D.-D. Mai (c), A. Bedoya-Pinto (c),

J. Malindretos (c) and A. Rizzi (c),

Appl. Phys Lett. 93, 021916-1-3

(2008).

(a) ESRF

(b) ICMUV, 46071-València (Spain)

(c) VISel, D-37077 Göttingen

(Germany)

Fig. 138: a) A bridge across thecrack, b) a rendering of the

grains around this feature, andc) the grains with the crack

opening displacements applied.

References[1] P.M. Scott, Corrosion 56, 711 (2000).

[2] L. Babout et al., Mater. Sci. Technol. 22, 1068 (2006).

[3] A.P. Jivkov, T.J. Marrow, Theoret. App. Fracture Mechanics 48, 187 (2007).

[4] D.L. Engelberg, R.C. Newman, T.J. Marrow, Scripta Mat. 59, 554 (2008).

[5] G. Johnson et al., J. Appl. Crystallogr. 41, 310 (2008).

Fig. 137: Part ofthe DCT grainmap of astainless steelsample,combined withthe crack path(shown inwhite).

Local environment of Gd in GaN

Recently, Teraguchi et al. [1] observeda Curie temperature larger than 400 Kin Ga0.94Gd0.06N. Additionally, Dhar etal. [2] reported an extraordinarily largemagnetic moment of Gd in weakly Gddoped GaN layers as well asferromagnetism above roomtemperature. Despite the greatpotential, there are some controversialresults on the ferromagneticinteractions in Ga1–xGdxN. Somefindings suggest that the crucialmechanism is either the electricpolarisation of the hexagonal GaN, ora strong long-range interactionbetween Gd atoms and certain defects.On the other hand, Gd might beinducing a magnetic moment in Gaand/or N interstitials were beingdetected; others have attributed thestabilisation of ferromagnetism tonitrogen or gallium vacancies.Moreover, according to a recent

theoretical study [3], Ga vacancies arethe most effective source of localisedholes necessary for a strongferromagnetic p-d exchange couplingin GaGdN. In addition, several reportshave not detected the formation ofany secondary phases in GaN, whereasin Gd implanted GaN the presence ofprecipitates of Gd3Ga2, GdN, and Gdhas been observed with highsaturation magnetisation. [4]Therefore, the role of Gd on thedefect formation, local atomic siteconfiguration, and/or phase separationin GaN is of key importance.

We have studied the incorporation ofGd on 0.5 µm thick GaN films grownby molecular beam epitaxy(concentrations ranging from1.14 x 1017 to 1.18 x 1019 cm–3). AllGaGdN layers are ferromagnetic atroom temperature, as measured by


112


EE SS RR FF

superconducting quantum interferencedevice magnetometry (SQUID). X-rayabsorption near edge structure(XANES) and scanning X-rayfluorescence (XRF) measurements werecarried out at the microprobe at ID22.It has been argued that high-temperature ferromagnetism inmagnetically doped nitrides resultsfrom nanoscale regions containing ahigh concentration of the magneticconstituents. However, at the lengthscale of the beam size (1.5 x 3.5 µm2),uniform patterns with no intensitychanges (< 0.02%) were observed,showing a homogeneous distributionof Ga and Gd. Whereas the expectedimpurity aggregation effects areobserved in heavily rare earth dopedGaN, our results suggest no tendencyto agglomerate at those low Gd levels.

The comparison of the X-ray lineardichroism (XLD) of the XANES at the Gd

L3 and Ga K edges supports thetetrahedral 6d3+ site distributions in thehighest Gd doped GaN (0.027%).Figure 139 shows the XANES recordedwith the electrical field vector paralleland perpendicular to the c axis of theGaGdN as well as the respective XLDsignal, reflecting directly the anisotropyof the unoccupied density of states ofthe 5d and 4p shells of the Gd and Gaatoms. The simulations of the XLDperformed with FEFF8 code wererescaled with respect to theexperimental ones. The Gd L3 edge XLDsignal is similar to the one observed atthe Ga K edge but with smalleramplitudes. The results show a stronganisotropy exhibiting the signature ofthe hexagonal GaN. There is noremarkable damping effect revealing astrong influence of the Gd impurities inany preferential crystallographicdirection. For other possible Gd siteoccupation rather than substitutional,the spectral shape and amplitude ofXLD signal is drastically different [5].

In summary, structural analysis basedon XANES and XLD revealed Gd ions ina predominantly trivalent state withtetrahedral coordination, occupyingGa sites in the wurtzite structure. Inaddition, EXAFS collections aroundGa atoms have shown no local atomicdistortion as a function of Gd fractionproviding direct evidence for thehighly short range structural order.Although the presence of very smallamounts cannot be completelyexcluded, within the sensitivity of ourexperimental techniques, neithersecondary phases nor Ga vacancieswere observed.

Fig. 139: XANES and XLD signals recorded at the Gd L3 and Ga K edges fromthe GaN (0.027% Gd).

References[1] N. Teraguchi, A. Suzuki,

Y. Nanishi, Y.K. Zhou,

M. Hashimoto, and H. Asahi, Solid

State Commun. 122, 651 (2002).

[2] S. Dhar, O. Brandt,

M. Ramsteiner, V.F. Sapega, and

K.H. Ploog, Phys. Rev. Lett. 94,

037205 (2005).

[3] L. Liu, P.Y. Yu, Z. Ma, and

S.S. Mao, Phys. Rev. Lett. 100,

127203 (2008).

[4] S.Y. Han, J. Hite, G.T. Thaler,

R.M. Frazier, C.R. Abernathy,

S.J. Pearton, H.K. Choi, W.O. Lee,

Y. D. Park, J.M. Zavada, and

R. Gwilliam, Appl. Phys. Lett. 88,

042102 (2006).

[5] A. Ney, T. Kammermeier,

E. Manuel, V. Ney, V.S. Dhar,

K.H. Ploog, F. Wilhelm, and

A. Rogalev, Appl. Phys. Lett. 90,

252515 (2007).

Cultural Heritage

Secrets of Bamiyan Buddhist murals

Forceful blasts aimed at destroying thetwo Giant Buddha statues at theBamiyan site and vandalism followingthese 2001 explosions damaged morethan just the statues. They were alsodirectly responsible for shatteringimportant mural paintings that hadonce existed on the niches above theBuddhas as well as other valuable

paintings inside the hundreds of caves(Figure 140).

The Bamiyan site is important bothhistorically and artistically for its placealong ancient cultural and materialroutes on the Silk Road linking the Eastand the West. It was included in theWorld Heritage List in Danger in 2003



and designated as a subject to beprotected through internationalcooperation. This project has beenfunded by the UNESCO/Japanese Funds-in-Trust so that the conservation project‘Safeguarding of the Bamiyan Site’could be overseen by UNESCO. In orderto perform satisfactory conservationintervention and provide a deepenedknowledge of painting technologiesstemming from Bamiyan along the vastSilk Road area, a wide range ofscientific diagnosis has been used toreveal the manufacturing techniquesand deterioration mechanisms ofvarious mural paintings from over 50caves dating back to a period betweenthe 5th and 9th centuries AD. Inparticular, synchrotron-based analyseswere performed at the ESRF with theaim of identifying the inorganicpigments and alteration products, andalso the organic binders. Thus, a multi-modal analysis was carried out on bothbeamline ID18F (micro X-rayfluorescence, µXRF, and micro X-raydiffraction, µXRD) and beamline ID21(micro Fourier Transform Infrared,µFTIR).

Elemental analyses were a first step toidentifying the pigments and grounds.We found compounds containingmetals such as lead, copper, calcium,mercury, iron and arsenic. Thesepreliminary data were then refinedthanks to micro X-ray diffraction. Thistechnique is highly valuable for theprecise identification of phases. It wasparticularly useful to distinguishpigments sharing common elementalcomposition. For example, various leadpigments were identified such asoxides, sulphates, carbonates,chlorides. It was also possible todistinguish various compositions of thesame pigment, for example leadwhites, containing differingproportions of lead carbonates(cerussite, PbCO3) and leadhydroxycarbonates (hydrocerussite,Pb3(CO3)2(OH)2). Such identification isvery important because it can giveindication of a possible choice ofpigment quality and/or a specific useof these pigments by the painter. Inparallel to these identifications, phaseswere mapped across transversal cross-sections to reveal their distribution.Indeed, these Buddhist paintingsusually present a very complex

stratigraphy with the superposition ofoften 5 or 6 layers (Figure 141). Tolook at this stratigraphy, µFTIR 2D-mappings were acquired. Thispermitted an identification of theorganic binders as well as theirlocalisation within the painting’s multi-layered structure.

Interestingly, the organic componentsalso revealed complex compositionsand distributions. A wide variety oforganic materials were used, with anapparent intentional choice correlatingto their role and their location(e.g. within sizing layers, paint layers,mordant, varnish and glaze). Oil wasidentified in many fragments, inaddition to proteins, resins andpolysaccharides. This discovery of mid-seventh century oil paintings is one ofthe oldest known to date.

Even more surprisingly, µFTIR enabledan identification of hydride materials,obtained by reaction of metallic (suchas lead or copper) compounds with oil,leading to metallic carboxylates(soaps). The use of a similarformulation was already established inancient cosmetics, but once again,their presence in such ancientpaintings was a real discovery.

Principal publication andauthorsM. Cotte (a,b), J. Susini (b),

V. Armando Solé (b),

Y. Taniguchi (c), J. Chillida (d),

E. Checroun (e), P. Walter (a),

J. Anal. At. Spectrom. 23, 820

(2008).

(a) Centre of Research and

Restoration of the French

Museums - UMR171 CNRS, Paris

(France)

(b) ESRF

(c) Japan Center for International

Cooperation in Conservation –

National Research Institute for

Cultural Properties, Tokyo (Japan)

(d) Painting restorer, Barcelona

(Spain)

(e) Painting restorer, Paris (France)

Fig. 140: A painting fragment from theFoladi cave. This painting of a Buddhafigure was identified as an oil painting(© NRICPT).

Fig. 141: Micro-FTIRanalysis of a pressedpainting fragment fromFoladi. Visible lightpicture (left). Some ofthe ingredients (organicbinders and pigments)identified and mappedby µFTIR (right).


114


EE SS RR FF

The chemical and structural study of acomplex multi-layered system such asthese paintings exemplifies thepotential of a correlative analysis usingmicro- synchrotron-based techniques.The high spatial resolution associatedwith a high detection limit and shortdwell time allows precise 2D chemicaland/or structural mappings. On a moregeneral level, this type of 2D

information together with advanceddata processing should lead to a betterunderstanding of painting techniques,both of the material compositions andthe manner in which the paint wasapplied. These results have given us acompletely new view on paintinghistory and will certainly encouragesimilar analyses of ancient paintingsworldwide.

X-ray imaging in the study of paintings

During the Middle Ages, Christianityattached great value to relics, i.e.physical remnants from the life ofChrist or remains of a saint, such as apiece of bone, textile or paper. TheScripture refers to the healing powerof objects that were touched by Christor his Apostles. Similar to modernhomeopathy, it was thought that suchremains, even when present in smalldoses, could have a beneficial effect.

In some cases relics have been hiddeninside an art object without anyexternal physical evidence of the relic’spresence. This seems to have been thepractice notably with sculpture, and insome rare instances also withaltarpieces. A good example is theNorfolk triptych from the collection ofthe Museum Boymans-van Beuningenin Rotterdam, which is considered tobe one of the oldest paintings in TheNetherlands.

The main scene in the central panelrepresents the Deposition of Christ(Figure 142). The X-ray transmission

radiography of this area shows that aninset was made into the panel, intowhich a thin plank of softwood wasinlaid with its grain directionperpendicular to the grain direction ofthe altarpiece. The primer and paintlayer have been painted on top andtherefore cover the wooden inlay. Thisraises the question of whether somesort of relic could be hidden behindthis inset (Figure 142). The objectcould be a piece of textile related tothe Shroud of Christ, depicted in thepainting itself, or a small piece ofpaper with an inscription, or maybeeven a bone or piece of wood fromthe cross of Christ.

Regular X-ray transmissionradiography has failed to properlyimage what is hidden behind thepanel inlay. Conventional radiographyprovides no three-dimensional insightinto the construction of the panel.However, a more significant limitationis the low sensitivity of X-rayabsorption to low-Z materials. Thepaint layer on the front contains heavymetals like Pb and Hg in pigments suchas lead white and cinnabar. Thewooden panel has a thickness of about4 cm. Any organic materials such aspaper or textile sandwiched inbetween the paint and the wood aretherefore virtually transparent at therelatively high energies needed topenetrate the entire object (> 25 keV).

The present study describes the first-time application of two synchrotron-based imaging techniques in the studyof paintings: X-ray computedtomography (ID17) and X-raycomputed laminography (ID19). Bothsetups have phase-contrast imagingcapability, being analyser-based and

Principal publication andauthorsK. Krug (a), L. Porra (b, c),

P. Coan (b), A. Wallert (d),

J. Dik (a), A. Coerdt (d),

A. Bravin (b), P. Reischig (a, e),

M. Elyyan (e), L. Helfen (b, e) and

T. Baumbach (e), J. Synchrotron

Rad. 15, 55 (2008)

(a) Department of Materials

Science, Delft University of

Technology (the Netherlands)

(b) ESRF

(c) Department of Physical

Sciences, University of Helsinki

(Finland)

(d) Rijksmuseum, Amsterdam (The

Netherlands)

(e) ANKA/ISS, Forschungszentrum

Karlsruhe (Germany)

Fig. 142: Christ as the man ofSorrows, also known as the

‘Norfolk tryptych’ ca. 1415–1420,Museum Boymans-van Beuningen,

Rotterdam, The Netherlands.The white frame corresponds to

the X-ray radiography of thecentral panel showing the wooden

inlay and its perpendicular graindirection.



Early structure of feathers revealed by phase contrastX-ray synchrotron imaging

A 4-cm piece of amber discoveredamong thousands of others in a quarryfrom southwestern France is helping tobridge a gap in our knowledge of theearly development of feathers. Theamber dates from the Early Cretaceousperiod, around 100 million years ago,and encases at least seven beautifullypreserved feathers of a primitive typenever seen before. They represent acritical stage in the evolution offeathers.

Conventional imaging techniquesusing transmitted light microscopeswere used first (Figure 144) but thisdid not reveal sufficient informationfor a precise assessement of themorphology of the feathers. Phasecontrast X-ray synchrotron imagingtechniques have already demonstratedtheir power on fossil inclusions inamber [1], so we decided to apply suchmethods to image the unique fossilfeathers. To reveal the finest structures

propagation-based, respectively. Thetomography setup had a 0.35 mmresolution with a 15 cm field of view,while laminography had a 7.5micrometre resolution with a 15 mmfield of view, which allowed a moredetailed inspection. The aim of thisexperiment was to see whether three-dimensional phase-contrast imagingwould be adequate for thevisualisation of an organic, fibrousobject imbedded inside a painting.

We prepared a test panel into which asmall inset was cut, mimicking thesuspected construction of the Norfolkpanel. Behind the inset we placed asmall folded piece of paper with ashort inscription in organic ink. Thepanel was given an authentic chalk-based priming layer and several layersof various paints containing heavymetals like Hg and Pb.

A computer tomography scan with0.35 mm resolution provided thethree-dimensional structure of theentire panel, visualising features suchas wood grain, and accumulation ofglue and voids inside the panel.Computed laminography with spatialresolution on the micrometre scaleallowed us to image local volumes, i.e.three-dimensional regions of interest,within the panel. In combination withpropagation-based phase-contrastimaging, we succeeded in three-dimensionally imaging the piece ofpaper sandwiched in the woodeninset, the heavy metal paint layers andthe bulk panel (Figure 143). The phase-contrast capability proved very

effective in enhancing image contrast(at all material interfaces andespecially between paper and air)which otherwise would have beendominated by the attenuation of thehighly absorbing paint layer.

In addition to the propagation-basedsetup we also imaged our panel withthe analyser-based phase-contrastimaging. This phase-contrast method isparticularly well suited to thedetection of steady variations of theprojected refractive index distribution.It permitted the hidden paper to beimaged, and even the contours of theink writing on the paper were visible.The contrast in these images stronglydepends on the refraction angle, theorientation of the analyser crystal withrespect to the wood grain and, to alesser extent, the density of the grainbeing imaged. The future availabilityof a laminography setup using ananalyser-based phase contrast couldsignificantly improve the detectabilityof such written texts. Plans areunderway to examine the actualartwork, the Norfolk triptych, withtechniques described above.

Fig. 143: Three-dimensional renderingof a mockup panelunder different viewingangles, obtained bysynchrotron-radiationcomputedlaminography. The paintlayer at the top, theperpendicular graindirections in the inlay(below) and the panel(bottom), and thefolded paper (centre)are clearly visible.

Principal publication andauthorsV. Perrichot (a), L. Marion (b),

D. Néraudeau (c), R. Vullo (c),

P. Tafforeau (d), Proc. Roy. Soc. B,

275, 1197 (2008).

(a) University of Kansas (USA)

(b) Ecobio, UMR 6553, Rennes

(France)

(c) Géosciences Rennes, UMR

6118, Rennes (France)

(d) ESRF


116


EE SS RR FF

of the feathers in three dimensions, amosaic of 6 scans was performed usinga single-distance phase retrievalapproach (holotomography) with avoxel size of 0.7 micrometres, atbeamline ID19. Complex multistepdata processing allowed anenhancement of the visibility of thefeathers and other fine structure andthen the virtual extraction of the mostwell-preserved feather from the resin.

Detailed holotomographic images(Figure 145) revealed a uniquestructure in the feathers morphologythat was previously unknown. Thefossils display a flattened primary shaft– or rachis – composed of the not-yet-fused bases of the secondary branches– or barbs – that initiate the planarform of later feathers. The barbs fuseprogressively and lack the completefusion that is observed in the rachis inall other fossil and modern feathers.The planar form of the fossil featherswas observed here for the first time.It results from the insertion of barbson the two opposite sides of the shaftwhose shape is flattened with respectto modern feathers.

This structure represents anintermediate and critical stage in theincremental evolution of feathers. Ithad been predicted by developmentaltheories [2] but was hithertoundocumented by evidence from therecent or the fossil record. Primitiveand derived feathers occur in bothearly birds and some dinosaurs knownas theropods. Interestingly, teeth fromtwo species belonging to this group offeathered dinosaurs were foundfossilised in the same quarry as theamber, raising the possibility that thefeathers come from a dinosaur ratherthan a bird.

In conclusion, the feathers’ primitivefeature differs from all other knownmorphologies, both modern andfossilised. It provides evidence for akey step in the evolution of feathersleading towards feathers for thepurpose of flight, thus in the transitionbetween dinosaurs and birds.

Fig. 144: Fossil featherspreserved in a 100

million-year-old piece ofamber from France,

observed by transmittedlight microscopy.

Scale bar 0.1 mm.

Fig. 145: 3D virtualreconstruction of thebest-preserved fossil

feather usingholotomography: long

filaments insertedparallel and opposite oneach side of a flattened

primary shaft initiatethe planar form of later

feathers.

References[1] M. Lak et al., Microsc.

Microanal., 14, 251 (2008).

[2] R.O. Prum and A.H. Brush,

Q. Rev. Biol., 77, 261 (2002).

Environmental Sciences

Hard X-ray nanoscale fluorescence imaging of Earthand Planetary Science samples

Synchrotron-based microanalysis canbe used to analyse fragile and minutesamples because it offers a unique wayto study the bulk morphology, internalstructure, crystallography, and tracecomposition of sub-millimetre grains

at the micrometre scale. Somescientific applications, however,require even higher resolution thancurrently available which leads to anadvancement of the techniques. Forinstance, nondestructive observation



of Earth and Planetary Sciencematerials requires multiscale analysisbecause samples can often featurechemical and structuralheterogeneities that span scales fromthe millimetre to the nanometre [1].

Here we present the new imagingcapabilities of the nano-imagingbeamline end-station (ID22NI) for thenon-invasive study of planetary sciencesamples. The setup is based on afocusing system permitting largebandwidth and fast nanoscale mappingoptimised for fluorescence imaging.

The optics scheme of ID22 beamlinehas to be suitably tuned in order toobtain decananometre X-ray spot sizeswith comparable horizontal andvertical dimensions. First, thehorizontal source was reduced toabout 25 micrometres by means of slits,thus creating a secondary horizontalsource 28 metres away from thesynchrotron source. The loss ofintensity is compensated for byswitching to the “pink beam” mode,whereby there is no crystalmonochromator in the beam. Thus theonly optical element between thesource and the focusing device is a flatSi mirror with Pd or Pt stripes thatfilters out high incident energies. Thisdramatically increases the photon fluxand improves the performance andstability of the focus at the expense ofthe energy resolution ∆E/E whichincreases by two orders of magnitudeup to 10–2. At 63 metres away from thesynchrotron source, the beam isfocused using a Kirkpatrick-Baez (KB)system that consists of twoindependent mirrors for focusing thebeam vertically and horizontally(Figure 146). The primary mirror,coated with a graded multilayer, actsboth as a vertical focusing device and amonochromator, yielding very high fluxof about 1012 photons/s for X-rays inthe 15-20 keV range. The minimal spotsize achieved was 80 nm in twodimensions and 40 nm in onedimension [2]. The nanoprobe featuresa flux density of up to 5.1013 ph/s/µm2,largely superior to the microprobe’s1011 ph/s/µm2, thus exhibiting muchhigher sensitivity for (ultra-) traceelement detection and allowingmapping with a greatly reduced dwelltime.

Samples are placed in the focal spot ofthe KB system on a 4-axis stage andcan be raster scanned on three axesplus one rotation. At each position afluorescence spectrum which givesinformation about the sampleelemental composition is collected.The spectra are measured using a SIINanotechnology Vortex 50 mm2 silicondrift diode (SDD) collimated detectorplaced in the horizontal plane at 75degrees from the incident beam.SDD detectors provide an excellenttrade-off between output count rate(approximately 600 kcps at 0.25 µspeaking time) and energy resolution(140 eV), although displaying lowerpeak-to-valley ratios than standardSi(Li) detectors. A scanning modebased on the acquisition of data whilecontinuously translating the samplewith a piezo-driven stage speeds upthe measurement by up to 50%.

A cometary grain from the NASAStardust cometary mission collectionwas studied [1]. The sample wasmeasured in situ, in the original lowdensity (approx. 0.01 g/cm3) silicaaerogel, in which it was trapped afterits few millimetres penetration track.

Figure 147 shows the superposition ofthe Fe 2D map with the Ca and Ni

Principal publication andauthorsP. Bleuet (a,d), A. Simionovici (b),

L. Lemelle (c), T. Ferroir (c),

P. Cloetens (a), R. Tucoulou (a),

J. Susini (a), Applied Physics

Letters 92, 213111 (2008).

(a) ESRF

(b) LGIT, OSUG, Université J.

Fourier, CNRS UMR 5559,

Grenoble (France)

(c) LST, Université de Lyon, Ecole

Normale Supérieure de Lyon,

UMR5570-USR3010 (France)

(d) Present address: CEA, LETI,

MINATEC (Grenoble)

Fig. 147: Colour-coded map of thedistribution of Fe, Ni and Ca in thecometary grain.

Fig. 146: Schematic view of theexperimental setup of the ID22NI

end-station.


118


EE SS RR FF

maps, respectively, the two other majorelements in the grain. A chemicalheterogeneity at the sub-micrometrescale is evident in this view. No post-processing of the image has been doneso that strong heterogeneities fromone 50 nm pixel to the other areshown. Micrometre Ca-rich zones andhundreds of nanometres Ni-rich phasesare embedded in an Fe-rich matrix.This chemical view starts to beconsistent with the classical petrologyof the cometary grains composed ofmicrometre Fe-rich crystals of olivinesand pyroxenes, Ca-rich pyroxenes and

plagioclases, nanometre Fe-Niprecipitates, all embedded in a silicamatrix. These excellent results suggestthat XRF studies of Stardust grainscould be used to obtain petrologyinformation in place of the alternativemore destructive analyses.

Further analyses must be carried out tobetter establish the shape and 3Dquantitative chemical composition ofthe grain. They include 2D/3Dmorphology, crystallinity and oxidationstate by means of nano-tomography,diffraction or spectroscopy.

References[1] G.J. Flynn et al., Science 314,

1731 (2006).

[2] R. Ortega, P. Cloetens,

G. Devès, A. Carmona, and

S. Bohic, PLoS ONE 9, e925 (2007).

Carbon nanotubes in macrophages: imaging andchemical analysis by synchrotron X-ray fluorescencemicroscopy

The human health impact associatedwith carbon nanotubes (CNT) is acritical issue considering theirnumerous potential applications. Basicresearch on interactions betweencarbon nanotubes and cells is needed,especially for their localisation withincells. A second requirement is to haveinformation about cell modificationsafter contact with carbon nanotubes.Both points can be addressed usingsynchrotron-based X-ray fluorescencemicroscopy (µXRF), used here for thefirst time to study carbon nanotube-cell interactions.

In this study, we consider murinmacrophages exposed to unpurified orpurified single-walled (SW) CNT or tomulti-walled (MW) CNT, the choice ofmacrophage cells being dictated by

their key role in host responses toexogenous agents (Figure 148). Two-dimensional fluorescence scans havebeen performed on cryofixed samples,on beamline ID21, with a beam sizeof 1.5 µm x 0.5 µm and with incidentX-ray energy of 7.2 keV. This energywas chosen to excite the K-lines ofelements of atomic number up toZ = 26, corresponding to iron, which isthe catalyst of the growth of thestudied nanotubes. Iron particles arefound inside MWCNT or bound toSWCNT bundles, and can therefore beused as a nanotube tracer inside cells.The purification process considered inSWCNT samples leads to a decrease ofthe amount of iron from 20 wt% to2wt% in unpurified and purifiedsamples, respectively.

Figure 149a shows fluorescence spectraand elemental maps of an unexposedcell obtained using µXRF. Phosphorusand potassium are constitutiveelements of cells and their spatialrepartition delimitates the cell shape.The P and K-rich zones in the centralpart of the cell can be attributed tothe nucleus. Iron signal is found topresent a rather homogeneousdistribution and is attributed toexperimental background.

Considering carbon nanotube-exposedcells, iron maps reveal one or severaliron-rich zone(s) inside or close to the

Principal publication andauthorsC. Bussy (a,b), J. Cambedouzou (a),

S. Lanone (b), E. Leccia (a),

V. Heresanu (a), M. Pinault (c),

M. Mayne-l’Hermite (c),

N. Brun (a), C. Mory (a),

M. Cotte (d), J. Doucet (a),

J. Boczkowski (b), and

P. Launois (a), Nano Letters 8,

2659 (2008).

(a) Laboratoire de Physique des

Solides, CNRS-Université Paris-Sud

11, Orsay (France)

(b) Institut Mondor de Recherche

Biomédicale, INSERM-Université

Paris Val de Marne/Paris 12,

Créteil (France)

(c) Laboratoire Francis Perrin,

CEA-CNRS, Saclay (France)

(d) ESRF

Fig. 148: Artistic view of amacrophage cell engulfing a

bundle of single-walled carbonnanotubes. Iron catalytic particles,

surrounded by carbon shells andbound to nanotubes, are shown

in orange (not to scale).



References[1] D.M. Brown, et al., Am. J.

Physiol. Lung Cell. Mol. Physiol.

286, L344 (2004).

[2] K. Donaldson, et al., Free

Radical Biol. & Med., 34, 1369

(2003).

AcknowledgementThe authors acknowledge financial support from the Agence Nationale de la Recherche and C'Nano

Ile de France.

cell contours (see Figure 149b). Thesezones correspond to CNT localisation.Iron-based catalyst particles could evenbe detected in the case of cellexposure to purified SWCNT, for whichiron amount is minimised, thushighlighting the high sensitivity of theµXRF method. In the case ofmacrophages exposed to unpurifiedSWCNT, an additional calcium signalfollowing the cell-shape can beevidenced (Figure 149b) although nocalcium signal is detectable inunexposed cells. This finding of extra-cellular calcium uptake by exposedcells could be of strong importance inthe perspective of understanding thetoxicological mechanisms of CNT [1,2].Moreover, pharmacological assays withcalcium chelators and inhibitors,performed following this µXRFexperiment, point toward the key roleof calcium in carbon nanotubecytotoxicity.

Calcium uptake was shown in part ofthe cells exposed to unpurified SWCNTor MWCNT but in none of thoseexposed to purified SWCNT. Chemicalor physical parameters of CNT, such astheir catalyst content (higher inunpurified SWCNT and MWCNT thanin purified ones) or their surfaceproperties (which are modified duringpurification process) could be crucial

determinants in their toxicity. Effect ofthe nanotube aspect ratio and lengthmay also be relevant. ESRF capabilitiescould allow one to go further in theunderstanding of carbon nanotube-cellinteractions: micro-XANES could beused to identify putative changes inthe chemical state of the ironnanoparticles during cell contact andnano-XRF tomography could give newinformation about CNT toxicity at thesub-cellular level.

Fig. 149: X-ray fluorescence spectra integrated over the whole scanned areas of atypical control cryo-fixed macrophage a) and of a macrophage exposed for 24h to anunpurified SWCNT suspension at 100 µg/mL b). Enlarged areas around the positions ofthe Kα and Kβ fluorescence peaks of potassium and of the Kα peak of calcium areshown, together with fits of the potassium and calcium contributions, in blue and inpurple, respectively. Elemental maps of phosphorus (P), potassium (K), iron (Fe) andcalcium (Ca) are drawn. Pixel size is 1 µm x 1 µm.


120

Methods and Instrumentation

EE SS RR FF

Introduction

Instrumentation development and thedeployment of new methodology arefundamental to synchrotron radiationresearch. At the ESRF, close collaborationbetween the instrumentation supportteams and the beamline scientistshas been the traditional source ofmost developments. In the future, theexploration of new techniques, the needfor instrument integration, the quest forstability, automation, and user-friendlyoperation will call for an even closercollaboration between beamlines anddevelopers.

A new division, the InstrumentationServices and Development Division,ISDD, will gather together supportgroups presently distributed amongthe Experiments, Computer Servicesand Technical Services Divisions in viewof the Upgrade Programme, whichwill dedicate the greatest part ofits funding to new beamlines andinstrumentation development. The

new division, together with a newlyintroduced project managementsystem, will permit an enhancedeffort, capable of responding to theinstrumentation challenges set by theUpgrade Programme.

The main themes of instrumentationdevelopment for the next few years willbe based on the integration of verydifferent technical domains and theuse of the most recent discoveriesin nanotechnology. Integration ofdifferent characterisation techniquesinto the crowded environmentsurrounding the sample is one of thechallenges of the entire programme. In2008, an important breakthrough wasmade in this direction: a beamlinespecific atomic force microscope (AFM)was developed and has been tested onselected ESRF beamlines. The microscopeis designed to visualise and positionnanostructures, track the beam profile,and manipulate and exert force on thesample while doing X-ray experiments.In one of the first experiments, ithas made the measurement of theYoung’s modulus possible for a singlenanodot [1].

The small dimensions of the atomicforce microscope are advantageousfor its installation in the beamlines,facilitating its integration (seeFigure 150). Smaller devices naturallyoffer much higher cut-off frequencies,and hence faster responses. For thesereasons they will be important playersin the instrumentation theatre of thenext few years. The first contributionin this chapter dedicated tomethodologies and instrumentation,shows how a simple silicon cantilevercan be used as a fast X-ray chopper. Thissimple experiment opens the way to theuse of fast responding MEMS and NEMSin synchrotron radiation technology.


Fig. 150: The atomic forcemicroscope installed in the end

station of beamline ID01. TheAFM is used to visualise both

the beam and the sample andto manipulate the sample

(Courtesy O. Dhez).



MEMS based high-frequency chopper

Many experiments in physics, biology,chemistry and material science requireshort bursts of X-ray radiation to studythe detailed time response of a sampleunder a particular excitation [1].

X-ray techniques are extremelyimportant for the investigation of thechemistry and structure of materials. Ina typical experiment, the system to beanalysed is excited by a short X-rayburst and information about theresponse is obtained by studying thefluorescence emitted during thesystem’s de-excitation. In a pump-probe experiment, the sample ispumped in a metastable state by laserexcitation and a delayed X-ray pulse isused to monitor the time evolution ofthe system.

The duration of the X-ray pulse iscritical to a pump-probe experiment.Since the timescale involved can besmaller than a microsecond, X-raypulses with duration less than amicrosecond are required. Short-duration X-ray bursts are producedwhen a continuous X-ray beam passesthrough a mechanical shutter that isopen for a short time. The majorlimitations of rotating mechanicalshutters are their relatively large sizeand the constraints imposed by the

operating environment. For example,they may have to work under vacuumconditions, especially if a very shortX-ray pulse is required.

A combination of X-ray techniques andmicro-electro-mechanical systems(MEMS) technology has recently beendemonstrated in a setup where amodulated X-ray beam was used toexcite a mechanical system around itsfirst order resonance [2]. We showhere that the inverse mechanism canbe used to modulate an X-ray beam.

Figure 151 shows the experimentalsetup. The mechanical system is astandard Si (100) atomic forcemicroscope (AFM) cantilever withdimensions 300 x 35 x 2 µm. Thecantilever displacement is measured bythe interference between the lightreflected from the end of a cleaved

AuthorsA. Siria (a,b), O. Dhez (c),

W. Schwartz (a,b), G. Torricelli (d),

F. Comin (c) and J. Chevrier (a).

(a) Institut Neel, CNRS-Université

Joseph Fourier, Grenoble (France)

(b) CEA/LETI-MINATEC, Grenoble

(France)

(c) ESRF

(d) Department of Physics and

Astronomy, University of Leicester

(UK)

Fig. 151: Schematic representationof the experimental setup. TheAFM cantilever (red) is used tomodulate the X-ray beam (blue)impacting around the Bragg angle.The reflection of the X-ray beam isdetected with a photodiode. Thelever position is detected via theoptical fibre (white).

Miniaturisation is also the driving forceof the second contribution where alaser beam is used to drill small holes inthe gaskets of diamond anvil cells.The next two contributions are relevantto X-ray optics, one suggesting amethod for easier retrieval ofinformation on layered systems and thesecond showing again that silicontechnology applied to synchrotronradiation can provide easy to use X-rayinterferometry up to 100 keV. The lasttwo contributions illustrate the rangeof instrumentation advances made

possible by the close collaborationbetween the MacromolecularCrystallography Group and the varioussupport teams. In the past years thisclose collaboration has led to thedevelopment of beamline automationthat earned the Bessy Innovation Awardfor the teams involved. Their work haspermitted the launch of remote accessto the macromolecular crystallographybeamlines, this is described in the finalarticle.

F. Comin

Reference[1] T. Scheler, M. Rodrigues,

T.W. Cornelius, C. Mocuta,

A. Malachias, R. Magalhães-

Paniago, F. Comin, J. Chevrier, and

T. H. Metzger, Appl. Phys. Lett. 94,

023109 (2009).


122


EE SS RR FF

optic fibre and the beam reflected bythe back of the lever. When at rest,the cantilever is in Bragg conditionsand the photodiode will detect aconstant photon flux. Whereas, whenthe AFM lever is excited by a piezo-electric ceramic, the X-ray incidenceangle is modified by the lever motion.If the lever oscillation amplitude islarger than the Bragg peak width, thecantilever will periodically sweepthrough the Bragg conditions.Therefore, a periodically-modulatedcurrent will be measured at the outputof the photodiode. The amplitude of

the photon flux oscillation is thusdirectly controlled by the oscillationamplitude of the cantilever.

The experiment was carried out at thesurface X-ray diffraction (SXRD)beamline, ID03. The incoming X-raybeam set to 18.98 keV impinged onthe cantilever at rest at Braggcondition and the X-ray photodetector was positioned at thecorresponding 2θ angle. The spot sizeof the incoming beam was 50 x 50 µmwith 1010 ph/s. Figure 152a shows thecantilever oscillation amplitude andphase lag as a function of frequencywhen the mechanical system ismechanically excited at the resonance.Figure 152b shows the current(amplitude and phase) at the outputof the X-ray photodiode as a functionof the frequency.

We have shown that an X-ray beamcan be modulated in intensity by anoscillating AFM lever. In thisexperiment the X-ray beam waschopped at 13 kHz. This operatingfrequency was related to theresonance frequency of the AFM leverused, however it could easily beincreased to values much higher than100 kHz by using others types ofcantilever. Specific setups based onMEMS could be designed so thatefficient X-ray choppers could operateat frequencies in the MHz regime,potentially opening a wealth of newexperiments based on X-rayexamination of time dependentprocesses requiring high repetitionspeed.

Fig. 152: a) Opticallymeasured response of theAFM cantilever when it is

excited around its firstresonant frequency.

b) Diode photo-currentmeasured by the

photodiode at 2θ Braggposition. In both a) and b)the black curve represents

the oscillation amplitudeand the red curve the

phase lag.

References[1] C. Rischel, et al., Nature 390,

490-492 (1997).

[2] A. Siria, et al., Nanotechnology

19, 445501 (2008).

Laser drilling of DAC gaskets

Diamond anvil cells have been usedincreasingly on synchrotron X-raybeamlines over the past decade due tothe pressure and temperature rangethey permit, and the match betweenavailable sample volume and beamsize. As the availability of micrometresized beam grows, smaller samplesmay be used with smaller diamondculets in order to reach higherpressures, up to and above onemegabar. This means of course that

the hole for the sample in the gasketwhich is clamped between the twodiamonds also has to follow the samereduction in size. The classical methodof drilling the hole in the gasket byelectro-erosion becomes increasinglydifficult when the hole size goesbelow 100 micrometres. The laser drillwe have developed allows the drillingof holes with sizes ranging from 25 µmup to 1 mm with a precision of 2 µm.Additionally, in contrast to electro-

Authors P. van der Linden (a),

M. Mezouar (a), N. Guignot (b),

Y. Dabin (a), J. Jacobs (a) and

D. Gibson (a).

(a) ESRF

(b) Synchrotron SOLEIL (France)



erosion drilling that can only be usedfor metals, the laser drill can be usedfor materials such as ceramics, saltsand even diamond.

The setup (Figure 153) uses a pulsedNd:YVO4 laser at a wavelength of1064 nm which was selected for itsshort pulse length of 10 ns. This allowsmore efficient ablation of the materialand reduces the heat affected zone.A standard beam expander andmicroscope objective focus the laserbeam on the gasket. The gasket isfixed on a high precision spindle whichwas developed in house. The firstalignment stage of the spindle is forpre-alignment by means of simplescrews; the second alignment stage isa rotating XY piezo mounted on theaxis; the mechanical support consists ofthe bearing cage, motor and slip ringcontacts. The combination of axis andbearing cage was designed tocompensate for temperature changes;the axis is belt driven by an offsetmotor to minimise the mechanicalnoise. Special care was taken tooptimise the visualisation; the constantfocus zoom allows on screen imagesfrom 900 to 120 micrometres indiameter.

The laser drill was first developed togain time drilling micro-holes formultimegabar X-ray experiments. Oneof the very first experiments requiringthe use of the laser drill was run inparallel on ID27 and ID24 for the studyof phase transformations at the deepearth pressure and temperatureconditions (Figure 154). On the highpressure beamline ID27 the sampleswere synthesised in situ by laserheating, using X-ray diffraction toverify the crystalline structure. In thesecond part of the experiment theoxidation state of iron was measuredby EXAFS on the dispersive beamlineID24.

New applications have also becomepossible: the use of non metallicmaterials in built up gaskets(Figure 155; S. Petitgirard / ID22;V. Giordano / ID16; G. Garbarino /ID27), cutting small samples (A. Bosak /ID28). Some tests have been done onthe drilling of pinholes which are to beused as beam cleaning elements.Future possibilities include drilling of

multiple holes in one gasket to usemultiple samples or to separate thepressure measurement ruby from thesample.

The authors would like to thank theircolleagues who have made theexamples available for this highlight:S. Pascarelli, V. Giordano. They alsowould like to thank R. Boehler of theMax Planck Institute who has built thefirst laser drill; P. Duboc for help withthe drawings of the spindle andI. Snigireva for electron microscopework.

Fig. 153: Setup of thelaserdrill.

Fig. 154: Fe K-edge XANES analysis of Al-bearing(Mg,Fe)SiO3 Perovskite andpost Perovskite phases up to 1.5 Mbar(image courtesy D. Andrault).

Fig. 155: Gasket drilled / filled withNaCl / drilled to avoid contact betweenthe molten Tellurium sample andRhenium gasket (courtesy V. Giordano /ID16). The hole in the centre is dark,the salt is the grey ring. The biggerwhite ring is the imprint of thediamond, 600 micrometre in size.


124


EE SS RR FF

In situ X-ray reflectometry can solve thephase problem

X-ray reflectometry is widely used todetermine the depth distribution ofthe dielectric constant in layeredsystems. This involves an inverseproblem, where one attempts to inferthe causes from observation of theeffects. Solving it may be hampered bythe ambiguity in the solution,essentially due to the absence ofinformation about the phase of theamplitude reflectivity [1].

We have developed a method of exactphase retrieval that is applicable to alayered film for which the reflectivityhas been measured in situ duringgrowth. In the simplest case, the anglebetween the X-ray beam and the filmsurface is fixed, while, at a point t intime, both the reflectivity R(t) = |r(t)|2

and the derivative dR/dt are measured.In this case, both the real R[r(t)] andimaginary I[r(t)] parts of the amplitudereflectivity r(t), hence the phase, canbe pinpointed exactly.

Using the first order differentialequation describing the variation of theamplitude reflectivity with the filmthickness we deduced the followingaccurate equation to describe thevariation of the reflectivity with timedR/dt = A(q, f, R) R[r(t)] + B(q, f, R) I[r(t)](Eq. 1), where A and B are realfunctions that do not explicitly containthe film’s parameters. They onlydepend on the reflectivity R(t) and onthe incident flux of particles q(t) andtheir chemical composition via thecomplex atomic scattering factor f(t).q and f can vary with time.

When performing in situ reflectivitymeasurements R and dR/dt are knownso that Eq. (1) is no longer a differentialequation for the function R(t). As alinear algebraic equation for the realand imaginary parts of the amplitudereflectivity, Eq. (1) can now be solvedtogether with the obvious relationR(t) = {R[r(t)]}2 + {I[r(t)]}2 (Eq. 2) todetermine both R[r(t)] and I[r(t)]directly from the experimental data,without using any model for describingthe reflection from the media.

Thus, knowing R and dR/dt at a point tin time we determine the phase of thereflected wave at the same point, i.e.,the phase can be found without anyknowledge of the pre-history of thestructure growth, and independent ofthe chemical composition inside thestructure, the presence of interlayers,and of the substrate composition. Theequation was deduced while makingtwo important assumptions: (a) thematerial polarisability is proportionalto the density, and (b) the dielectricpermeability inside the film remainsunchanged.

The reflectivity (17.5 keV X-ray energy,0.5° grazing angle) measured in situ atthe BM05 beamline [2] during thegrowth of a tungsten film on a Sisubstrate is shown in Figure 156. Theincident flux q = 7.26⋅1013 atom/cm2/sresults in a typical growth rate of 12pm/s. The measurements wereperformed every 7 s which correspondsto an increase of the film thickness ofabout 80 pm, equivalent to one thirdof a monolayer of tungsten. The phaseof the amplitude reflectivity extractedfrom the experimental curve R(t) ispresented in Figure 157. As Eq.(2) hasa quadratic form, there are twopossible solutions for the phaseretrieval problem (red curves 1 and 2).Curve 3 in Figure 157 and curve 2 inFigure 156 were calculated using thesimplest model possible, i.e., assuminga constant film density with depth.Since curves 1 and 3 in Figure 157 areclose to each other, we selected curve1 as the solution to the phase retrievalproblem. Nonetheless, both the

Principal publications andauthors I. Kozhevnikov (a), L. Peverini (b)

and E. Ziegler (b), Optics Express,

16, 144 (2008);

J. Appl. Phys, 104, 054914 (2008).

(a) Institute of Crystallography,

Moscow (Russia)

(b) ESRF

Fig. 156: Reflectivity versusdeposition time from a growing

tungsten film measured at anX-ray energy of 17.5 keV andwith a grazing angle θ = 0.5°

(dots). The solid curve wascalculated assuming constant

density in the depth of the film.



reflectivity and the phase inferredfrom the experimental data differnoticeably from the model calculation,suggesting that this simplest model isnot fully correct.

We also reconstructed the depth-distribution of the tungsten densityusing the approach described in [1],which is based on the analysis of theangular dependence of the reflectivitymeasured after deposition of a24.6 nm thick film. Two particularfeatures can be clearly observed onthe density profile: (a) the reduceddensity as compared to bulk tungstenin a region about 2 nm thick near thesubstrate and (b) the exponentialdecrease of tungsten concentrationover a depth of about 1 nm within theSi substrate, probably caused bydiffusion and implantation processes.These two features permit a completeexplanation of both the reflectivityand the phase curves.

In the future, we plan to investigatethe possibility of performing acomplete angular reflectivity curveR(θ, t) instead of a single angularvalue at any point in time. That wouldallow us to find the phase versusangle, to solve the inverse problem ofX-ray reflectometry more correctly,and to analyse the variations of thedielectric constant profile with time.

Fig. 157: Solutions of the phaseretrieval problem (curves 1 and 2)found directly from theexperimental curve shown inFigure 156. Curve 3 shows thephase evolution calculatedassuming a uniform tungsten filmwith constant density. Forillustrative purpose, we assumedthe phase to vary within the [–π / 2, 3π / 2] interval.

References[1] I.V. Kozhevnikov, Nucl.

Instrum. Methods A 508, 519

(2003).

[2] L. Peverini, E. Ziegler,

T. Bigault, I. Kozhevnikov, Phys.

Rev. B 72, 045445 (2005).

Hard X-ray interferometer based on asilicon refractive bi-lens system

Optical interferometers are widely usedfor diagnostics and metrology becauseof the availability of lasers with a highdegree of spatial coherence. X-rayinterferometry, however, has long beenhampered by the lack of coherentsources. To circumvent this problem, thefirst X-ray interferometers used the so-called perfect crystal optics in theregime of dynamical diffraction. Due tothe coherent interaction of the X-rayswith a three-dimensional crystalstructure, the crystal becomes anintrinsically coherent device andtherefore does not require spatialcoherence of the incident radiation.The classical example is the Bonse-Hartinterferometer, which utilises bothtransmission (Laue) and surfacereflection (Bragg) diffractioncomponents.

Nowadays, the widespread availabilityof bright X-ray sources with sufficientlylarge spatial coherence, such as third-generation synchrotrons, allowsresearchers to observe interference bydivision of the wave front, similar to

the classical Young’s experiment.Recently a Fresnel double slit, a Fresneldouble prism and a Fresnel doublemirror were applied to measure thespatial coherence of synchrotronbeams.

Following the successful developmentand application of refractive optics forhigh energy X-rays [1,2], we haveconstructed an X-ray bi-lensinterferometer similar to the Billet split

Authors A. Snigirev (a), V. Kohn (b),

I. Snigireva (a), M. Grigoriev (c),

S. Kuznetsov (c), V. Yunkin (c),

T. Roth (a), C. Detlefs (a) and

G. Vaughan (a).

(a) ESRF

(b) Kurchatov Instititute, Moscow

(Russia)

(c) Institute of Microelectronics

Technology, Chernogolovka

(Russia)

Fig. 158: a) Schematicdrawing of theexperimental set-up usedfor the bi-lens test.b) Scanning electronmicroscope micrographshowing a fragment ofsilicon bi-lenses chip.c) Image of the bi-lensfoci recorded at adistance F = 35 mmshows the lenses splitdistance d = 60 µm.


126


EE SS RR FF

lens in classical optics. The schematicdiagram of our X-ray bi-lensinterferometer is shown in Figure 158.It consists of two identical parallelplanar compound refractive lenses(CRLs) separated by a certain distance,d. Each lens produces an image of thesource. When the lens split distance dis smaller than the spatial coherencelength of the incoming beam, theseimages are diffraction limited and canbe treated as coherent secondarysources. At a distance of more thantwice the lens focal distance, F, theradiation cones diverging from thesecoherent secondary sources overlap,and interference occurs in the regionof superposition (see Figure 158).Identically to the Fresnel double slit,the visibility (or contrast) of therecorded interference pattern can beused as a measure of the degree of aspatial coherence of the incidentbeam, or of an equivalent effectivesource size.

A silicon chip with a set of differentlinear (1D) bi-lenses was manufacturedusing semiconductor microfabricationtechnology (Figure 158b). The lengthof each single, individual lens was 100µm and the aperture was 50micrometres. The radius of theparabola apex was 6.25 µm. Thecompound refractive lenses wereseparated by a distance d = 60 µm.

The bi-lens interferometer was testedat beamline ID06. The liquid -nitrogencooled Si-111 double crystal fixed exitmonochromator (manufactured byCINEL, Italy) was used at an X-rayenergy of 12 keV. The Si bi-lens chipwas mounted on the microoptics testbench optics stage at a distanceL0 = 55 m from the source.

All measurements of bi-lensinterference patterns were made in avertical plane. The interference patternswere recorded by means of the highresolution X-ray CCD Sensicam camerawith a spatial resolution about 1.3 µm(0.645 µm pixel size). The typicalexposure time was 5-10 sec. during 16-bunch filling mode.

The observed interference pattern isshown in Figure 159. The fringespacing or peak-to-peak width of thefringes is given by Λ = λL1/d, where λ isthe wavelength, and L1 is the distancefrom the bi-lens focus plane todetector plane. The measured spacing,Λ = 6.25 µm, is in very goodagreement with calculations. Thequality of the fringes produced by abi-lens system can be describedquantitatively using the visibilityV = (Imax – Imin)/ (Imax + Imin), whereImax and Imin are the irradiancescorresponding to the maximum andadjacent minimum in the fringesystem. It is easy to show that thevisibility V is directly related to thesource size, S, as follows:

ΛL0 lnVS = –––– ( – –––– )1/2

L1 3.56

The measured visibility was close to40%, corresponding to an effectivesource size in the vertical direction ofabout 45 µm (FWHM).

The bi-lens interferometer presentedhere has major advantages overexisting methods. Manufacturing ofmicro-slits, mirrors and prisms for hardX-rays is a challenging technologicaltask considering the requirements tothe surface and shape (edges) quality.Whereas for silicon planar lenses, thewell-developed microelectronicstechnology is used providing superiorlens quality [3]. Compared to slits,mirrors and prisms, the compoundrefractive bi-lens system can be used athigh photon energies up to 100 keV.

The applications of the bi-lenses arenot limited to coherencecharacterisation. They can easily beextended to coherent imagingtechniques. As with a classicalinterferometer, the bi-lensinterferometer generates twocoherent beams separated in spaceand then coherently recombines them

Fig. 159: a) Interferencepattern obtained using

the bi-lens interferometerof Figure 158, recordedat a distance 4 m with

monochromatic beam atenergy E = 12 keV.

b) Intensity variation fora line through the centre

of the fringe patternshows a visibility

(contrast) ofapproximately 40%corresponding to a

source size S = 45 µmvertically (FWHM).

References[1] A. Snigirev, V. Kohn,

I. Snigireva, B. Lengeler, Nature

384, 49 (1996).

[2] V. Aristov, M. Grigoriev,

S. Kuznetsov, L. Shabelnikov,

V. Yunkin, T. Weitkamp, C. Rau,

I. Snigireva, A. Snigirev,

M. Hoffmann, E. Voges, Appl.

Phys. Lett. 77, 4058 (2000).

[3] A. Snigirev, I. Snigireva,

M. Grigoriev, V. Yunkin,

M. Di Michiel, S. Kuznetsov,

G. Vaughan, SPIE 6705, 670506-1

(2007).



producing the interference pattern.One can easily insert a sample into oneof the beams while they are separated.Any interaction with the beam wouldthen induce significant changes in theinterference pattern. From the new

phase pattern, one should then beable to extract information concerningthe nature and degree of interactionof the sample with the beam, leading,eventually, to a holographicreconstruction of the sample.

Capturing the shape of macromolecularcrystals in 3DAbsorption correction algorithms havebeen used for decades, in particular bycrystallographers. The first approachesto reduce the effects of absorptionwere made by physically reshaping thesample to reach simple geometries,such as spherical or cylindrical, so thattabulated corrections could be appliedeasily. When the crystal could not bereshaped, Albrecht’s [1] method waswidely used, it relies on a precisedescription of the crystal geometry inorder to determine the correctiongraphically.

However, macromolecularcrystallography (MX) demanded adifferent approach. During MXexperiments an undefined volume ofcryoprotectant buffer is used to holdthe macromolecular crystal within anylon or Kapton loop prior tovitrification and data collection atcryotemperatures (100 K). Such acomplex system, often with highlyirregularly shaped crystals, is almostimpossible to characterise manuallywith a description of all of theelements contributing to theabsorption process, particularly theliquor surrounding the crystal. Mostscaling software used in MX todayemploys an entirely empirical functionwhich minimises the differencesbetween symmetry-equivalentreflections to correct for absorptioneffects (as well as for the correlatedcrystal decay and beam effects). Withthe use of increasingly fragile sampleshaving high radiation sensitivity,complete and redundant data sets maynot be obtainable from a singlecrystal, in turn defeating the fullyempirical method. Other methods aretherefore needed to optimise suchdata sets – of particular importancewhere small signals are being

recorded, such as anomalous datafrom sulphur or phosphorus atomsthat commonly occur naturally inproteins or nucleic acids.

Studies on two methods to supplythree-dimensional shape informationrecorded directly from the cryocooledcrystals in vitrified buffer andsupporting loop have been carried out.One uses the standard visible-lightmicroscope systems installed on all ofthe ESRF MX beamlines to build athree-dimensional model based onreconstruction from shape-from-silhouettes of the crystal, bufferand loop components; the second usesX-ray microtomography to determinethe three-dimensional shape of thecrystal within its surrounding liquidand sample holder.

Visible light silhouettes

The key aim behind this method is toprovide a routine system to buildthree-dimensional models of thecrystal-buffer-loop system byexploiting a standard on-axis samplevisualisation system that incorporates ahigh-quality on-axis camera system forsample alignment. As part of this

Principal publications andauthorsR.M.F. Leal (a,b,c),

S.C.M. Teixeira (b,c), V. Rey (a),

V.T. Forsyth (b,c) and

E.P. Mitchell (a,c); J. Appl. Cryst.

41, 729-737 (2008);

S. Brockhauser (d),

M. Di Michiel (a),

J.E. McGeehan (d),

A.A. McCarthy (d) and

R.B.G. Ravelli (d,e); J. Appl. Cryst.

41, 1057-1066 (2008).

(a) ESRF

(b) ILL (France)

(c) EPSAM and ISTM, Keele

University (UK)

(d) EMBL, Grenoble Outstation

(France)

(e) Leiden University Medical

Center (LUMC) (The Netherlands)

Fig. 160: (top) Examples of the composite imagesbuilt from visible light images taken at differentfocal lengths to overcome depth-of-field issues;(lower) The three-dimensional model of the insulincrystal, buffer and loop built from the images. Blueshows the crystal and loop components, and red thecryocooled buffer. The circle on the lower imageshows where minor errors are apparent in the model.


128


EE SS RR FF

development, a software suite (called“3DAC”) was developed specifically toreconstruct the 3D model and calculateabsorption correction factors using the2D images from the camera.A feasibility study was carried out onan insulin crystal from whichdiffraction data were collected at lowenergy (6 keV). A high quality three-dimensional model was constructedusing just 18 visible-light images atdifferent rotation angles of the sample(Figure 160) with the loop, buffer andcrystal silhouettes determined usinga mixture of automatic and manualprocedures. Absorption correctionswere determined based on simulatedbeam paths through the 3D modelthat consisted of the three separateloop, buffer and crystal components.The corrections were applied to eachreflection (whether partially or fullyrecorded) before merging.

A comparison of the reduced dataresulting from the standard empiricalapproach to scaling MX data andapplication of the absorptioncorrections calculated as aboveshowed a distinct and significantimprovement in the data when ourcorrections were applied. Metrics suchas the Rmerge, signal:noise ratio andpeak heights in phased anomalousdifference maps all improved whendata having a redundancy of less thanapproximately three was used. Forexample, in the best cases, averagepeak heights corresponding to sulphurpositions in the phased anomalousmap improved by 10% to 60%depending upon the redundancyof the data (from higher to lowerredundancy). The exact effectobserved depended on the wedgeof diffraction data used, i.e. theorientation of the crystal-loop system.However, the best data wereconsistently obtained with applicationof the quantitative absorptioncorrection.

X-ray microtomography

In X-ray tomography, a 3Dreconstruction of the sample isobtained after collecting parallel-beamX-ray images at a series of rotationangles. At higher energies (> 5 keV),macromolecular samples typically showvery little absorption contrast between

the vitrified crystal and its surroundingliquid. Fortunately, one can exploitphase contrast which results in asample to detector distance-dependentenhancement of the edges betweencrystal, liquor and sample support.

We have successfully applied thetechnique of X-ray microtomographyto obtain 3D reconstructions of avariety of protein and DNA crystals(Figure 161). The great advantage ofthe technique is that a true “X-ray”view of the crystal is obtained,showing details such as cracks, crystalsplitting and satellites, bent crystalsand other imperfections. Thereconstruction algorithms benefit fromexisting methods such as those used inelectron tomography.

X-ray microtomography leads tosuperior views of the macromolecularsample and its support compared tovisible-light methods, however, it doesrequire a special imaging setup.The X-ray beam has to fully bathe thesample, and a carefully aligned X-rayimaging detector is needed. We havedemonstrated that such a setup, whichis default on imaging beamlines, canalso be successfully installed and usedon a macromolecular crystallographybeamline. The doses needed to obtainX-ray tomographic data could staywithin a few percent of the Hendersonlimit, especially if one does not requiresub-micrometre resolution. This wouldmake the technique amenable forabsorption correction as well as forcrystal alignment.

In conclusion, both studies have shownthe feasibility of producing and usingthree-dimensional models ofmacromolecular crystals and theirsupporting buffer and loop to improvediffraction data quality. Each methodhas its own pros and cons, but givecomplementary information. Thetomography method gives stunningmodels, accurate down to the tiniestdetails but requires a specific, purpose-built installation. The visible-lightmethod uses standard beamlineequipment but produces a modelshowing only the overall bulk of thecrystal, buffer and loop. The methodsshould provide a clear advantage insituations where data acquisition islimited by crystal lifetime and should

Fig. 161: Full three-dimensionalreconstruction of a vitrified

lysozyme crystal and surroundingcryocooled buffer from X-ray

tomography.



form an essential component ofsample characterisation on futurebeamlines, as part of the overallcomplete characterisation of themacromolecular crystal. Three-dimensional shape information canalso have a practical impact on thedesign of data collection strategies forX-ray diffraction experiments and forneutron studies. Further work is nowconcentrating on expanding the

number of test cases and comparingthe absorption corrections obtainedfrom the two modelling approaches.Given the increasing interest inradiation damage and the study ofever more sensitive macromolecularsystems, the time is now right to revivethe implementation of fully non-empirical absorption correctionsthrough easy-to-use (even automatic)methodologies.

Reference[1] G. Albrecht, Rev. Sci. Instrum.,

10, 221–222 (1939).

New technical developments at the ESRFMacromolecular Crystallography Groupbeamlines

ID14-3: A New BioSAXSbeamline

In January 2008 work began on theconversion of ID14-3 from a beamlinededicated to single-wavelengthmacromolecular crystallographyexperiments to one dedicated to small-angle X-ray scattering from solutionsof biological macromolecules(BioSAXS).

The new facility is intended tocomplement already existing SAXSfacilities at the ESRF that are highlyoversubscribed for a wide variety ofexperiments in the fields of biology,chemistry, materials science, etc. Withrapidly-growing interest in BioSAXSfrom the ESRF’s MX User Community,beam-time on the existing resourceswas set to become even more soughtafter. Creating a dedicated BioSAXSfacility will alleviate the high rate ofover-subscription, improve thethroughput of such experiments byremoving the need for the station tobe optimised for every new usergroup, and result in more beam-timeon the ESRF’s existing SAXS end-stations being available forexperiments in other fields. Anadditional benefit of building this newfacility at the ESRF is the creation of acollaboration with the alreadyestablished small-angle neutronscattering (SANS) facility at the ILL.Our intention is to allow users, via asingle proposal, to access both facilitiesfor appropriate experiments. This will

allow the complementary informationprovided by neutron scattering (i.e.contrast variation) and X-ray scatteringto be obtained in a single visit to theGrenoble site.

The conversion of ID14-3 has beenundertaken in stages. The first of theseallowed testing of the proposed designand proof-of-principle experimentsbefore the major renovation of theexperimental hutch was undertaken.New beam conditioning hardware andredesigned attenuators have beeninstalled upstream of the originalID14-3 focusing optics in order tominimise parasitic scattering.To this end a “mini” hutch (the first ofits type at the ESRF) has been built inthe ID14 mezzanine (Figure 162) andhouses new beam defining slits, theexperimental shutter and beammonitoring devices. The experimentalhutch has also been refitted and now

AuthorsA. Round (a), P. Pernot (b) and

G. Leonard (b), on behalf of the

ESRF/EMBL Joint Structural

Biology Group (JSBG)

(a) EMBL Grenoble Outstation

(b) ESRF

Fig. 162: a) The ID14-3 “mini”hutch installed under themezzanine of ID14; b) the newlyrefurbished ID14-3 BioSAXSexperimental hutch.


130


EE SS RR FF

contains a 4 m long marble table whichsupports the new experimentalequipment. Two pairs of guard slits aremounted prior to the sample exposureunit (quartz capillary tube mounted invacuum). An evacuated flight tubewith variable length (0.5 to 2.5 m)and a fully-motorised beam-stop sitsbetween the sample environment anda (Vantec-2000) 2D gas filled-detector(Figure 162). The commissioning phaseof the new BioSAXS facility is nowcomplete and, following positivefeedback from a small number ofcollaborating users during October2008, the first official experiments tookplace in November 2008.

The ultimate aim of the conversion ofID14-3 to a dedicated BioSAXSresource is to offer a facility withautomatic sample loading/unloading,data collection, processing(conversion of a 2-D image to anormalised 1D X-ray scattering profile,Figure 163) and analysis. This will(rapidly) provide users with standarddata concerning the size (radius ofgyration, maximum dimension andvolume) and molecular weight ofsamples and allow on-the-fly ab initioshape reconstruction in order toprovide feedback enabling datacollection strategies to be optimised.Automation of sample loading will beincorporated on the beamline in thecoming months using a device beingconstructed in a collaborationbetween the EMBL (Grenoble andHamburg outstations) and the ESRF.Automated data analysis will beimplemented following the model ofthe SAXS facility at X-33, EMBLHamburg). Future plans extend toallowing remote access to the newfacility. This will be based on thesystem currently in use on the ESRF’s

MX end-stations, which is describedbelow.

Remote access to the ESRF’sMX facilities

Modern Structural Biology callsfor short but frequent access tosynchrotron radiation basedmacromolecular crystallographyfacilities (maximum length ofexperimental session ~6 shifts). Formany of the ESRF’s external users (bothacademic and industrial) travel to/fromGrenoble, as well as being stressful andtiring, often takes up more time thanthe experiment itself. Remote access tothe ESRF’s MX beamlines would meanmuch less travelling for users. For this,and a number of other reasons, wehave developed remote data collectionprotocols, broadly reflecting the needsof our external users. The modes ofremote access currently offered are asfollows:

1. Data collection service: Here,instructions concerning the data thatusers would like to be collected fromtheir samples (‘diffraction plans’) aresubmitted beforehand via thelaboratory information managementsystem (LIMS) ISPyB [1](http: //ispyb.esrf.fr/). ESRF staff thenoperate the beamlines and take themajority of the decisions concerningdata collection. External users canfollow progress in real-time using ISPyBand are able to intervene withsupplementary instructions should theywish to. This type of remote access ishighly labour intensive for in-housestaff and so offered only to industrialcustomers via the well-establishedMXpress service (seehttp://www.esrf.fr/Industry).

Fig. 163: Lysozyme SAXS datacollected at ID14-3:

a) 2-dimensional detector imageand b) the resulting 1-dimensional

X-ray scattering profile.



References[1] A. Beteva, S.M. McSweeney,

et al., Acta Cryst. D62, 1162

(2006).

[2] G. Leonard, J. McCarthy,

D. Nurizzo, X. Thibault,

Synchrotron Radiation News 20,

18 (2007).

2. Full remote control: In this modeexternal users can also avoid travellingto Grenoble by sending samples to theESRF via a courier company. On the dayof the experimental session, thesamples are loaded into the roboticsample changers, available on all ourMX beamlines, by ESRF staff. Theexternal users connect to the beamlinecontrol computer and, after havingbeen given control by the experimentLocal Contact, control the beamlinefrom the home laboratory in the samemanner as if they were using thebeamline locally. ESRF beamline staffremain on call to help in case ofproblems.

3. ‘Hybrid’ control: In this mode, only afew of the users from an external grouptravel to the ESRF for a scheduledexperimental session. Those remainingin the home laboratory can collect dataon the samples that they havecontributed to the session via remotecontrol of the beamline. In this mode ofremote access, sample changer loadingand other ancillary tasks are performedby those users who have travelled toGrenoble.

For the hybrid and full remote controlmodes, reliable external access to thebeamline control computers is essential.For this we use NoMachine/NX software(http://www.nomachine.com). Thisallows access to an ESRF firewallcomputer from which remote users canthen access and login to the relevantbeamline control computer. TheMxCuBE Graphical User Interface (GUI)[2] and beamline control applicationscan then be started. These applicationsare displayed and manipulated in real-time on the user’s local workstation.

To allow secure remote access to theMX beamlines we developed amaster/slave version of MxCuBE(Figure 164). Master instances of theGUI are accessed by beamline staff toset up the beamline and to hand overcontrol to a slave instance of the GUIrunning from a remote location. Themaster instance of MxCuBE allowsbeamline staff to follow the experimentand take back control or intervenewhen necessary. Slave instances ofMxCuBE have a limited scope and inparticular, cannot take control from anyother instance of the beamline GUI.

However, it is possible for slaveinstances to view all other instances ofMxCuBE running on a particularbeamline, to transfer control or tocommunicate with them via a ‘chat’ tab.

Following initial tests with the help ofgroups from the University of York(Figure 165) and the University ofMarseilles, hybrid and full remotecontrol access modes were madegenerally available to external users ofthe MX beamlines in March 2008.Currently, ~25% of academic and 30%of industrial experimental sessionsinvolve one of these two modes ofremote access. We expect these figuresto increase rapidly and that remoteaccess will soon become the principalgateway to the ESRF’s MX beamlines forour larger user groups.

Fig. 164: A remote instance of theMxCuBE beamline control GUI. Thebrick allowing transfer of controland an overview of all othermxCuBE instances running at thebeamline is circled.

Fig. 165: J. Turkenburg (right)and S. Hart collecting data onID14-2 from their office atthe University of York.


132 EE SS RR FF

Introduction

Throughout 2008, the Accelerator andSource Division has continued its effortsto ensure reliable operation as well aspreparing for the forthcoming upgrade.A number of developments have beencarried out, the most important of theseare described hereafter.

Acceleratorand X-ray Source

Beam parameters of the storage ring

Table 2 presents a summary of thecharacteristics of the storage ring’selectron beam.

Table 3 gives the main optic functions,electron beam sizes and divergences atvarious source points. For insertiondevice source points, the beta

functions, dispersion, sizes anddivergences are calculated in themiddle of the straight section.

Two representative source points foreach type of bending magnet (even orodd number) have been selected,corresponding to observation anglesof 3 and 9 mrad from the exit. Thebending magnets are such that themagnetic field is 0.4 T and 0.85 T atthe tangent point where the radiationis extracted at 3 and 9 mrad angles,respectively. Electron beam profiles areGaussian and the size and divergenceare presented in terms of rmsquantities. The associated full widthhalf maximum sizes and divergencesare 2.35 times higher. Horizontalelectron beam sizes and divergencesare given for the uniform filling modes

Energy [GeV] 6.03

Maximum current [mA] 200

Horizontal emittance [nm] 4

Vertical emittance [nm] 0.025(minimum achieved)

Revolution frequency [kHz] 355

Number of bunches 1 to 992

Time between bunches [ns] 2.82 to 2816

Even ID Odd ID Even BM Even BM Odd BM Odd BM(ID02, ID06…) (ID01, ID03…) (BM02, 4,…) (BM02, 4,…) (BM01, 3,…) (BM01, 3,…)

3 mrad 9 mrad 3 mrad 9 mrad

Magnetid Field [T] Variable Variable 0.4 0.85 0.4 0.85

Horiz. beta functions [m] 37.5 0.3 1.3 0.9 2.1 1.6Horiz. dispersion [m] 0.144 0.033 0.059 0.042 0.088 0.073Horiz. rms e- beam size [µm] 415 51 95 75 131 112Horiz. rms e- divergence [µrad] 10.3 108 115 111 102 97.4

Vert. beta functions [m] 3.0 3.0 41.2 41.7 31.6 31.7Vert. rms e- beam size [µm] 8.6 8.6 32 32 28 28Vert. rms e- divergence [µrad] 2.9 2.9 1.3 1.3 0.9 0.9

Table 2: Principalcharacteristics of the

electron beam.

Table 3: Beta functions, dispersion, rms beam size and divergence for the various source points.

RUN NUMBER TOTAL 2008-01 2008-02 2008-03 2008-04 2008-05 TOTAL2007 2008

Start 15/01/08 04/04/08 06/06/08 22/08/08 24/10/08

End 26/03/08 28/05/08 30/07/08 08/10/08 15/12/08

Total number of shifts 845 213 162 162 141 156 834

Number of USM shifts 681.96 165 130 132 113.9 144.1 685

Beam available for users (h) 5291.4 1292.5 1003 1021.2 895 1123.1 5334.7

Availability 97.82% 98.81% 97.59% 97.5% 99.46% 98.24% 98.32%

Dead time for failures 2.2% 1.2% 2.4% 2.5% 0.54% 1.8% 1.7%

Dead time for refills 0.83% 0.9% 1.2% 0.8% 1.2% 0.8% 0.97%

Average intensity (mA) 138.4 131 144 183 117 137 142.3

Number of failures 96 20 17 21 6 21 85

Mean time between failures (h) 56.8 66 61.2 50.3 151.8 54.9 64.5

Mean duration of a failure (h) 1.24 0.8 1.5 1.3 0.8 1 1.09

AA CC CC EE LL EE RR AA TT OO RR AA NN DD XX -- RR AA YY SS OO UU RR CC EE


and apply to almost all filling patternsexcept for the single bunch, for whicha slightly larger size and divergence isattained because of the increasedenergy spread of the electron beam.Vertical electron beam sizes anddivergences apply to the uniform,2 x 1/3 and hybrid filling modes only.To increase the lifetime of the storedbeam, the vertical beam sizes anddivergences are increased by about50% in the 16 and 4 bunch fillingpatterns.

The lifetime, bunch length and energyspread mainly depend on the fillingpattern. These are given in Table 4 fora few representative patterns. Notethat in both the 16-bunch and 4-bunchfilling patterns, the energy spread andbunch length decay with the current(the value indicated in the tablecorresponds to the maximum current).The bunch lengths are given for theusual radio frequency acceleratingvoltage of 9 MV (8 MV for 16-bunchand 4-bunch).

Filling pattern Uniform 7/8+1 Hybrid 16-bunch 4-bunchNumber of bunches 992 870 24x8+1 16 4Maximum current [mA] 200 200 200 90 40Lifetime [h] 75 65 30 16 9Rms energy spread [%] 0.11 0.11 0.11 0.12 0.16Rms bunch length [ps] 20 20 25 48 55

Table 4: Current, lifetime,bunch length and energyspread for a selection offilling modes.

Summary of accelerator operation

In 2008, 685 shifts (5480 hours) ofbeam were initially scheduled.Of these 5480 hours, 5335 wereeffectively delivered (including 52hours of refill). This represents a beamavailability of 98.3%, which is in linewith recent years. Dead time due tofailures accounts for the remaining1.7%. The number of failures is lowercompared to the previous year, thusleading to an all time record meantime between failures of 64.5 hours.This performance was achieved despitethe difficulties encountered with theradio frequency cavities and alsodespite the fact that electrical mainsdrops occurring between January and

June 2008 were not interceptedbecause the HQPS system was non

Fig. 166: Twenty-threedays without a singlefailure. This is thelongest deliveryperiod without beamloss of all time.

Table 5: Storage ringoperation in 2008.


134

Accelerator and X-ray Source

EE SS RR FF

functional (see below). Nineteen longdelivery periods (i.e. more than 100hours) without a single interruptiontook place in 2008. In September, thebeam was delivered for 23 dayswithout a single failure andinterrupted only by the scheduled

machine dedicated days. Figure 166highlights these four weeks of deliverywhich is the longest delivery periodever recorded without a single failure.

Table 5 presents an overview ofstorage ring operation in 2008.

Filling patterns

The filling mode labelled “7/8 + 1” isbecoming the standard multibunchmode. Developed in 2006-2007, thismode is a worthwhile compromise

offering the benefit of a long lifetimemultibunch mode while allowing timestructure studies with the single bunch.In 2008, delivery in this mode increasedto 44%, compared to 24% in 2007. Useof the uniform filling mode hasdropped to 13% of the delivery time in2008, compared to 31% in 2007. Thetendency was to replace uniform fillingwith the “7/8 + 1” mode. Delivery inthe other modes remained stablecompared to 2007 (Figure 167). Duringrun 2 of 2008, a vacuum problem onthe storage ring radio frequency cavity5 resulted in one week of delivery in7/8+1 mode instead of the initiallyscheduled 16 bunch mode.

A new high-quality power supply

The Grenoble area electricity supplyhas a dense aerial network providingcustomers with electricity. Lengthypower interruptions are rare thanks tothe 10 interconnected high voltageaerial transport lines (225 kV and 400kV). Unfortunately the aerial transportlines located in the mountainous areasare exposed to lightning duringstorms, which results in voltage andfrequency drops of the mains. Thestatistics over the last 10 years showthat 300 lightning strikes per yearoccur on the high voltage linessurrounding the ESRF. This is fourtimes more frequent than the averagefor France. Eighty of these eventsresult in a mains drop large enough totrigger an electron beam dump in thestorage ring. The biggest drops mayinduce transients that can damage theelectronics boards or leave them in anabnormal state requiring a manualreset.

The sensitivity of the various subsystemsdriving the accelerators has beenobserved since commissioning in 1992.At that time, a major storm could resultin beam interruptions of several hours,one even resulted in 5 days withoutbeam. To protect against mains drops,a high-quality power supply (HQPS)system was put into operation in 1995.It consisted of 10 rotating electro-mechanical storage systems eachcoupled to a 800 kW diesel engine via aclutch. The system was operatedwithout problems for seven years. Thediesel engines were also run to produceelectrical power for the acceleratorsystems during the 22 so-called EJP daysin winter when the cost of electricity isvery high. In 2002, a major failure ofseveral of the diesel engines occurred,caused by premature ageing of thebearings supporting the crankshaft.The origin of the problem could not beclearly identified and, although the

Fig. 167: Distribution offilling modes scheduled

for 2008.



system continued operation until 2007,it had to be stopped frequently to carryout interventions. The numerousfailures provoked premature ageing ofa number of components. Therefore, itwas decided to replace the HQPSsystem with a new system withoutdiesel engines, only involves rotatingenergy accumulators. The new systemstarted operation in July 2008. Whileseveral glitches were observed initially,it has protected the accelerator veryefficiently, allowing 4 weeks ofuninterrupted USM operation inAugust and September. Due to theabsence of diesel engines, the systemcannot maintain the full power of7.8 MW (storage ring with 200 mA andinjector in operation) for theaccelerator systems for more than3 seconds. Fortunately, such longinterruptions of the mains power arequite rare. During a long power cut,the system is nevertheless designed toswitch off the power to the radiofrequency system and to the magnetsbut to maintain power to the criticalsystems such as computer, vacuum andcontrol, using a single 1 MW dieselengine. This allows a fast restart of theaccelerator systems when the powercomes back on.

The system is manufactured by theBelgian company KST. It consists of 14rotating units each providing 800 kVA.A pair of such machines can be seen inthe photograph shown in Figure 168.In the centre are alternators rotatingat 1500 rpm while the accumulatorsare located at each extremity, theyrotate at a nominal speed of 3000 rpmand are magnetically coupled to thealternator rotating axis via windings.During charging time a specialwinding couples to the accumulatorand converts electrical power to storedenergy in the accumulator.

Figure 169 presents a schematic of theconnection of a pair of rotatingmachines to the mains and theaccelerator system. When a dropoccurs on the mains, the system reactsin one of two different ways. If themissing energy involved in the drop islow enough (very large majority ofcases), it operates as a networkconditioner and maintains a constantsupply of voltage and frequency to theaccelerator systems. In the case oflarge amplitude and/or large durationdrops (rare events), the system isolatesitself from the mains by opening the20 kV circuit breaker labeled D1 inFigure 169. When isolated, therotating power stored in theaccumulator is transformed intoelectrical power by the alternator. Theaccelerator system can run for a fewseconds. The system monitors themains on a permanent basis and triesto reconnect (by closing the D1 circuitbreaker) if the mains are back tonormal.

Fig. 168: A pair ofrotating units.

Fig. 169: Electrical connection of apair of rotating machines to the20 kV electrical grid. The rotatingunits are operated at 400 V andconnected to the 20 kVdistribution via a transformer.Seven such pairs of rotatingmachines operate in parallel withtheir own transformer andinductor.

Operation of a cryogenic in-vacuumundulatorA cryogenic permanent magnetundulator (CPMU) offers increasedpeak field for a given magnetic gapand period allowing the extension ofthe high brilliance undulator radiation

to higher photon energies beyond thatavailable from an in-vacuumundulator. To make the most of theadvantages of a CPMU, the magneticmaterial must be selected with a


136

Accelerator and X-ray Source

EE SS RR FF

maximum remanence and thus a lowcoercivity. All in-vacuum undulators inoperation at the ESRF are baked above100°C after installation on the ring inorder to minimise the Bremstrahlungsent to the beamline. Unfortunately,optimisation of a CPMU results in theselection of low coercivity magneticmaterial which prevents such baking.To minimise the risk, an initial CMPUprototype was developed which madeuse of magnetic material that can bebaked, at the expense of a reducedfield performance. In January 2008,the prototype CPMU was installed inthe ID6 straight section (See Figure 170).This undulator has a period of 18 mmand a magnetic length of 2 m. During2007, the undulator was carefully

measured with a specially designedmagnetic measuring bench operatedin-vacuum with the magnetic assemblyat cryogenic temperature [1]. Themeasurements confirmed thepredicted temperature of 150 K whichcorresponds to the maximum field.Cryogenic cooling is based on a liquidnitrogen closed loop, identical to thesystem used in many ESRF beamlinesfor the cooling of cryogenicmonochromators. The installed deviceis connected to a cryo-pump systemlocated in the technical gallery with acryogenic line of 50 m.

Temperature measurements on theundulator rely on 20 thermocouplesdistributed along the magneticassembly. The power extracted by thecryogenic loop is derived from themeasurements of the differentialpressure and temperature of the liquidnitrogen between the input andoutput of the cryogenic pump.Figure 171 shows the averagetemperature of the undulator (blue)and power extracted by the cryogenicloop (black) in 16 bunch mode. Theupper curve (green) corresponds to thegap of the undulator. It reveals thestrong dependence of the magnettemperature as a function of theundulator gap. Both temperature andextracted power are at their maximumfor a gap which is fully opened (30mm). This feature has been observedwith other types of filling patterns.It is rather unexpected and it issuspected that this originates fromhigher-order modes (HOM) of theradio frequency field resonating in theundulator tank. Table 6 presents theaverage temperature of the undulatorand power extracted by the cryogenicloop for various filling modes and gapsettings. High beam induced power of105 W is observed with a fillingpattern corresponding to the fillingpattern of the hybrid mode.

Fig. 170: The CPMUInstalled in the ID6

straight section.

Fig. 171: Averagetemperature of the

CPMU (blue) andpower extracted bythe cryogenic loop

(black) versus time inthe 16 bunch mode.

The undulator gap isshown in the upper

curve (green).

Gap Undulator temperature Total cryogenic power Power from beamFilling mode [mm] [K] [W] [W]16 bunch 30 165 285 7516 bunch 7 154 247 377/8 & uniform 6 151 240 30Hybrid 15 177 315 105No beam N/A 140 210 0

Table 6: Measured average temperature of the magnetic assembly and power extracted by the cryogenic loop in differentfilling modes and with various undulator gaps. The last column presents the net power deposited by the beam.



The measured cryogenic power of210 W without beam originates fromconduction and radiation losses insidethe undulator vessel (150 W) plus thelosses in the cryogenic line (60 W).

The average observed temperature ofthe undulator with beam varies from151 K to 177 K depending on thefilling pattern. This corresponds to achange of peak magnetic field smallerthan 0.4% [2]. A feature of majorimportance is the variation oftemperature along the 2 metre-longmagnet assemblies. This effect wascarefully studied during the magneticmeasurements. A temperaturegradient has an impact on the phaseerror in a similar manner to a gapvariation along the structure(tapering). It has been observed thatto reach a phase error of 2.5 degreesrms, a longitudinal temperaturegradient smaller than 2 K/m must beobtained. Note that the phase errororiginating from a longitudinaltemperature gradient can becompensated for by a mechanical gaptapering. Initial observations showed agradient of 7 K/m in the hybrid fillingmode. Such a large gradient was theconsequence of poor thermal contactbetween the cooling pipes and thecopper blocks located at the entranceof the undulator (see Figure 172). Thetemperature gradient was laterreduced to 2 K/m following theinstallation of a flexible thermalconnection during the summershutdown.

Cooling from 300 to 150 deg Kwithout beam has had an effect onthe vacuum. It reduced the pressure byabout one order of magnitude to afew 10-10 mbar. The pressurenevertheless increases with beamcurrent to around 10-8 mbar at 200 mAof ring current, which is notsignificantly different to a roomtemperature in-vacuum undulator.The combination of a large mass ofmagnets and support pieces at 150 Kwith a low cryogenic power loss resultsin a long thermal time constant of12.5 hours. As a result the cryo-pumpdriving the liquid nitrogen in theundulator can be switched off for afew hours and then switched on again.The temperatures fluctuates slowly but

the system restores its equilibriumwithout any further interventions.

The conclusion of a full year ofoperation with beam is that theroutine use of a CPMU in an ESRFbeamline is compatible with theconstraints imposed by storage ringoperation. While further investigationswill be carried out on the existingprototype, a new CPMU will be built in2009 using higher remanence NdFeBmaterial. The design includes anumber of improvements.

Fig. 172: Entrance of the CPMU. The installation of a flexible thermal connectionbetween the cooling tubes and the end copper blocks at the entrance of theundulator has reduced the temperature gradient by a factor of 3.5.

References[1] J.Chavanne et al., EPAC’08,

June 08, p. 2243,

http://accelconf.web.cern.ch/

AccelConf/e08/papers/wepc105.pdf

[2] C. Kitegi, PhD Thesis,

University Joseph Fourier, 2008.


138 EE SS RR FF

Facts and Figures

The beamlines

Members and associate countries(as of January 2009)

Details of the public ESRF beamlines aswell as those operated by CollaboratingResearch Groups (CRG) are given inTables 7 and 8. Figure 173 shows thelocation of the beamlines in theexperimental hall.

The transformation of ID06 into amulti-purpose instrumentationdevelopment beamline continues. TheMX crystallography station at ID14-3has been transformed into a bio-SAXSend-station dedicated to proteinsolution scattering experiments. ID14-3started user operation in November.The GRAAL CRG beamline BM07stopped operation in November andwill be dismantled.

Members’ share in contribution to theannual budget:27.5% France25.5% Germany15% Italy14% United Kingdom4% Spain4% Switzerland6% Benesync (Belgium, The Netherlands)4% Nordsync (Denmark, Finland,

Norway, Sweden)

Additional contributions(percentages refer to Members’ total contribution):1% Portugal1% Israel1% Austria1% Poland0.55% Czech Republic0.25% Hungary0.25% Slovakia

Public beamlines

Instrumentation test and

development beamlines

CRG beamlines

Free bending magnet ports

Fig. 173: Experimental hall showing location of the beamlines(public and CRG beamlines).

FF AA CC TT SS AA NN DD FF II GG UU RR EE SS


SOURCE NUMBER OF BEAMLINE STATUSPOSITION INDEPENDENT NAME

END-STATIONS

ID01 1 Anomalous scattering Operational since 07/97

ID02 1 High brilliance Operational since 09/94

ID03 1 Surface diffraction Operational since 09/94

ID06 1 ESRF instrumentation and techniques test beamline Under commissioning

ID08 1 Dragon Operational since 02/00

ID09 1 White beam Operational since 09/94

ID10A 1 Troika I + III Operational since 09/94

ID10B 1 Troika II Operational since 04/98

ID11 1 Materials science Operational since 09/94

ID12 1 Circular polarisation Operational since 01/95

ID13 1 Microfocus Operational since 09/94

ID14A 2 Protein crystallography EH 1 Operational since 07/99

Protein crystallography EH 2 Operational since 12/97

ID14B 2 Protein solution small-angle scattering EH 3 Operational since 12/98

Protein crystallography EH 4 Operational since 07/99

ID15A 1 High energy diffraction Operational since 09/94

ID15B 1 High energy inelastic scattering Operational since 09/94

ID16 1 Inelastic scattering I Operational since 09/95

ID17 1 Medical Operational since 05/97

ID18 1 Nuclear scattering Operational since 01/96

ID19 1 Topography and tomography Operational since 06/96

ID20 1 Magnetic scattering Operational since 05/96

ID21 1 X-ray microscopy Operational since 12/97

ID22 1 Microfluorescence Operational since 12/97

ID23 2 Macromolecular crystallography MAD Operational since 06/04

Macromolecular crystallography microfocus Operational since 09/05

ID24 1 Dispersive EXAFS Operational since 02/96

ID26 1 X-ray absorption and emission Operational since 11/97

ID27 1 High pressure Operational since 02/05

ID28 1 Inelastic scattering II Operational since 12/98

ID29 1 Multiwavelength anomalous diffraction Operational since 01/00

ID30 1 Machine division beamline Operational since 09/07

ID31 1 Powder diffraction Operational since 05/96

ID32 1 X-ray standing wave and surface diffraction Operational since 11/95

BM05 1 Optics - Open Bending Magnet Operational since 09/95

BM29 1 X-ray absorption spectroscopy Operational since 12/95

SOURCE NUMBER OF BEAMLINE FIELD STATUSPOSITION INDEPENDENT NAME OF RESEARCH

END-STATIONS

BM01 2 Swiss-Norwegian BL X-ray absorption & diffraction Operational since 01/95

BM02 1 D2AM (French) Materials science Operational since 09/94

BM08 1 Gilda (Italian) X-ray absorption & diffraction Operational since 09/94

BM14 1 UK CRG Macromolecular crystallography (MAD) Operational since 01/01

BM16 1 SPANISH CRG Structural biology (MAD, SAX) Operational since 01/03

BM20 1 ROBL (German) Radiochemistry & ion beam physics Operational since 09/98

BM25 2 SPLINE (Spanish) X-ray absorption & diffraction Operational since 04/05

BM26 2 DUBBLE (Dutch/Belgian) Small-angle scattering & interface diffraction Operational since 12/98 Protein crystallography + EXAFS Operational since 06/01

BM28 1 XMAS (British) Magnetic scattering Operational since 04/98

BM30 2 FIP (French) Protein crystallography Operational since 02/99FAME (French) EXAFS Operational since 08/02

BM32 1 IF (French) Interfaces Operational since 09/94

Table 8: List of the Collaborating Research Group beamlines.

Table 7: List of the ESRF public beamlines.


140

Facts and Figures

EE SS RR FF

User operation

As we enter into the 15th year ofsuccessful operation of the facility forscientific Users, we look back on theyear 2008, which once again saw thefull complement of 31 publicbeamlines, together with 11 additionalbeamlines operated by CollaboratingResearch Groups (CRGs), available for

experiments by visiting research teams.Figure 174 shows the continuingincrease in number of applications forbeamtime since 2002 confirming that,although the main beamlineconstruction effort was complete by1999 and despite the fact that moresynchrotrons are now available toEuropean scientists, there is ever-increasing demand for use of the ESRFbeamlines and the number ofapplications for beamtime continuesto rise steadily, breaking the 2000barrier for the first time during 2008.

Proposals for experiments are selectedand beamtime allocations are madethrough peer review. ReviewCommittees of specialists, for the mostpart from European countries andIsrael, have been set up in thefollowing scientific areas:• chemistry• hard condensed matter:

electronic and magnetic properties• hard condensed matter:

crystals and ordered systems• hard condensed matter:

disordered systems and liquids • applied materials and engineering• environmental and cultural heritage

matters• macromolecular crystallography• medicine• methods and instrumentation• soft condensed matter• surfaces and interfaces.

The Review Committees met twiceduring the year, around six weeks afterthe deadlines for submission ofproposals (1 March and 1 September).They reviewed a record number of2013 applications for beamtime, andselected 903 (44.9%), which were thenscheduled for experiments.

Features of this period: • increasing numbers of proposalssubmitted for research into diseasesfor an ageing society such asneurogenerative diseases (Alzheimer’s,Parkinson’s) and osteoporosis, and forradiobiology for therapy programs(Microbeam Radiation Therapy,Stereotactic Synchroton RadiationTherapy).• in the area of Soft Condensed

Scientific field Total shifts Total shiftsrequested allocated

Chemistry 3 104 1 257Hard condensed matter:• Electronic and magnetic prop. 5 355 2 031• Crystals and ordered structures 4 230 1 800• Disordered systems 1 693 699Applied materials and engineering 3 164 1 546

Environmental and cultural 2 330 699heritage mattersMacromolecular crystallography 3 507 2 381Medicine 1 109 507Methods and instrumentation 678 364Soft condensed matter 2 909 1 287Surfaces and interfaces 3 026 1 335

Totals 31 105 13 906

Fig. 175: Shiftsscheduled for

experiments, Marchto July 2008, by

scientific area.

Fig. 174: Numbers ofapplications for

beamtime, experimentalsessions and user visits,

2002 to 2008.N.B. Final numbers ofexperiments and user

visits for 2008 were notavailable at the time of

going to press.

Table 9: Numberof shifts ofbeamtime

requested andallocated for userexperiments, year

2008.



Administration and finance

Matter, around half of all proposalssubmitted were concerned withbiomaterials.

Requests for beamtime, which isscheduled in shifts of 8 hours, totalled31 105 shifts or 248 840 hours in 2008,of which 13 906 shifts or 111 248 hours(44.7%) were allocated. Thedistribution of shifts requested andallocated, by scientific area, is shownin Table 9.

The breakdown of shifts scheduled forexperiments by scientific area in thefirst half of 2008 is shown in Figure 175.This period saw yet another recordbroken with the number of user visitsgoing beyond the 3000 barrier for asingle scheduling period: 3088 visits byscientists to the ESRF under the userprogramme, to carry out 741experiments. Figure 174 shows therapid rise in the number of user visitssince 2002, the higher numbers inrecent years reflecting the increasedlevel of automation on manybeamlines, resulting in shorter andtherefore more frequent experimentsacross almost all scientific areas butparticularly for experiments carriedout by the macromolecularcrystallography BAG teams.

Overall, the number of users in eachexperimental team averaged 4 personsas in 2007, but the average staydecreased to 3 days compared with 4days in 2007, confirming thatexperiments have tended to becomeshorter thanks to many factorsincluding higher automation andincreased flux from state-of-the-artoptics. The year 2008 also saw theimplementation of remote accessexperiments for macromolecularcrystallography, whereby users may runtheir experiments on the ESRF MXbeamlines using a control station intheir home laboratories. From July toDecember 2008, of the 1366 users whoperformed public experiments on theMX beamlines, nearly 200 did soremotely from their home laboratories.User responses to questionnaires showthat the ESRF continues to maintain itsexcellent reputation concerning theassistance given by scientists andsupport staff on beamlines, and traveland administrative arrangements, inaddition to the quality both of thebeam and of the experimental stations.Facilities on site, such as preparationlaboratories, the Guesthouse and acanteen open 7 days a week, also makean important contribution to thequality of user support.

kEuro

Expenditure

Accelerator and SourcePersonnel 4 909.2Recurrent 1 849.8

Operating costs 1 590.8Other recurrent costs 259.0

Capital 4 324.1Accelerator and Source Developments 4 324.1

Beamlines, experiments and in-house research

Personnel 24 303.0Recurrent 6 743.8

Operating costs 3 692.2Other Recurrent costs 3 051.6

Capital 7 546.3Beamline developments 4 029.9Beamline refurbishment 3 516.4

Technical and administrative supportsPersonnel 16 651.7Recurrent 10 811.2Capital 3 064.3

Unexpended committed fundsFunds carried forward to 2008 0

Total 80 203.4

kEuro

Income

2007 Members’ contributions 69 742.7Funds carried forward from 2006 13.0

Other incomeScientific Associates 4 275.5Sale of beam time 2 724.5Compensatory funds 210.0Scientific collaboration and Special projects 3 237.7

Total 80 203.4

Expenditure and income 2007


142

Facts and Figures

EE SS RR FF

kEuro

Expenditure

Accelerator and SourcePersonnel 5 435Recurrent 2 040

Operating costs 1 633Other recurrent costs 407

Capital 4 932Accelerator and Source developments 4 932

Beamlines, experiments and in-house researchPersonnel 25 380Recurrent 7 240

Operating costs 4 190Other Recurrent costs 3 050

Capital 7 480Beamline developments 4 360Beamline refurbishment 3 120

Technical and administrative supportsPersonnel 17 845Recurrent 10 895Capital 4 897

Industrial and commercial activityPersonnel 500Recurrent 220

Personnel costs provision

Total 86 864

kEuro

Income

2008 Members’ contributions 70 712Funds carried forward from 2007 0

Other incomeScientific Associates 4 265Sale of beam time 2 640Compensatory funds 92Scientific collaboration and Special projects 9 155

Total 86 864

Revised expenditure and income budget for 2008

kEuro

PERSONNELESRF staff 43 876.6External temporary staff 96.4Other personnel costs 1 890.9

RECURRENTConsumables 7 664.2Services 9 295.8Other recurrent costs 2 444.7

CAPITALBuildings, infrastructure 1 312.8Lab. and Workshops 346.9Accelerator and Source incl. ID’s and Fes 4 324.1Beamlines, Experiments 7 546.3Computing Infrastructure 1 136.8Other Capital costs 267.9

Unexpended committed fundsFunds carried forward to 2007 0

Total 80 203.4

Expenditure 2007by nature of expenditure

kEuro

PERSONNELESRF staff 47 320External temporary staff 85Other personnel costs 1 965

RECURRENTConsumables 7 665Services 10 270Other recurrent costs 2 460

CAPITALBuildings, infrastructure 2 422Lab. and Workshops 400Accelerator and Source incl. ID’s and Fes 4 822Beamlines, Experiments 7 270Computing Infrastructure 1 970Other Capital costs 215

Total 86 864

Revised budget for 2008by nature of expenditure

2007 200820062005

Financial resources in 2006, 2007, 2008 and2009, by major programme

(current prices in MEuro for the respective years)

1009080706050403020100

1009080706050403020100

(in

MEu

ro)

Financial resources in 2006, 2007, 2008 and2009, by nature of expenditure

(current prices in MEuro for the respective years)

(in

MEu

ro)

2008 200920072006

Accelerator and SourceExperimentsTechnical, AdministrativeSupportPersonnel cost provision

PersonnelRecurrent

CapitalPersonnel cost prov.



Scientists, Engineers, Technicians and PhD students TotalSenior Administrators Administrative Staff

Staff on regular positions

Accelerator and Source 28 38.4 66.4

Beamlines, instruments and 212 65.2 18.5 295.7experiments*

General technical services 49.6 69.9 119.4

Directorate, administration 32.5 51.9 84.4and central services

Sub-total 322 225.4 18.5 565.9

Other positions

Short term contracts 4 7 11

Staff under “contrats dequalification” (apprentices) 23 23

European Union grants 12 2 1 15

Temporary workers

Total 338 257.4 19.5 614.9

Absences of staff(equivalent full time posts) 22.7

Total with absences 592.2

Scientific collaborators and consultants 11 11

External funded research fellows 7 18 25

2008 manpower (posts filled on 31/12/2008)

* Includingscientific staff ontime limitedcontract.

Organisation chart of the ESRF(as of January 2009)


144 EE SS RR FF

EditorG. AdmansLayoutPixel ProjectPrintingImprimerie du Pont de Claix

© ESRF • February 2009

Communication UnitESRFBP220 • 38043 Grenoble • FranceTel. +33 (0)4 76 88 20 56 • Fax. +33 (0)4 76 88 25 42http://www.esrf.eu

Cover imageHow do palladium nanoclusters behave in hydrogen? Di Vece and co-authorsobserved hydrogen-induced Ostwald ripening, the growth of larger palladiumnanoclusters at the expense of the smaller ones. This is a schematic impressionof the palladium nanoclusters with interstitial hydrogen between the palladiumatoms, which become free to move due to the interstitial hydrogen (see p24).Image courtesy M. Di Vece (K.U. Leuven) and J. Husson.

We gratefully acknowledge the help of:

C. Argoud, J. Baruchel, S. Blanchon, B. Boulanger, J-F. Bouteille, N.B. Brookes, J. Chavanne,M. Collignon, F. Comin, E. Dancer, P. Elleaume, A. Fitch, C. Habfast, L. Hardy, E. Jean-Baptiste,

A. Kaprolat, M. Krisch, S. Larsen, G. Leonard, J. McCarthy, E. Mitchell, T. Narayanan,H. Reichert, M. Rodriguez Castellano, R. Rüffer, F. Sette, W.G. Stirling, C. Stuck, S. McSweeney,J. Zegenhagen, and all the users and staff who have contributed to this edition of the Highlights.

highlights - European Synchrotron Radiation Facility (ESRF)

Documents

Transcript of highlights - European Synchrotron Radiation Facility (ESRF)