98 Pages
English

Quantitative time resolved neutron imaging methods at the high flux neutron source FRM-II [Elektronische Ressource] / Johannes Brunner

Gain access to the library to view online
Learn more

Description

Technischen Universit¨at Munc¨ henFakultat¨ fur¨ PhysikInstitut fur¨ Experimentalphysik E21Quantitative time resolved neutronimaging methods at the high flux neutronsource FRM-IIJohannes BrunnerVollst¨andiger Abdruck der von der Fakult¨at fur¨ Physikder Technischen Universit¨at Munc¨ hen zur Erlangung des akademischen Grades einesDoktors der Naturwissenschaftengenehmigten Dissertation.Vorsitzender: Univ. Prof. Dr. M. KleberPruf¨ er der Dissertation: 1. Univ. Prof. Dr. P. B¨oni2. Univ. Prof. Dr. W. PetryDie Dissertation wurde am 19.09.2005 bei der Technischen Universit¨at Mu¨nchen ein-gereicht und durch die Fakult¨at fu¨r Physik am 01.03.2006 angenommen.”One picture is worth a thousand words”chinese sayingAbstractEnglishIn the current work various new experimental methods and computation proceduresin the field of neutron imaging are presented. These methods have a significant technicalimportance in non-destructive material investigations.Withstroboscopicneutronradiographyperiodicprocessescanbeinvestigatedonasub-millisecond time scale. This opens great opportunities for the study and the developmentof combustion engines.Energy selective time of flight neutron radiography at neutron spallation sources usesthe energy dependence of the neutron cross section to distinguish between materials. Theenergy resolution of this technique is very good and allows the identification of specificmaterials.

Subjects

Informations

Published by
Published 01 January 2006
Reads 25
Language English
Document size 42 MB

TechnischenUniversit¨atM¨unchen
Fakult¨atf¨urPhysik
Institutf¨urExperimentalphysikE21

Quantitativetimeresolvedneutron
imagingmethodsatthehighfluxneutron
IFRM-Isource

BrunnersneJohan

Vollst¨andigerAbdruckdervonderFakult¨atf¨urPhysik
derTechnischenUniversit¨atM¨unchenzurErlangungdesakademischenGradeseines
DoktorsderNaturwissenschaften
.ertationDissgenehmigten

Vorsitzender:Univ.Prof.Dr.M.Kleber
Pr¨uferderDissertation:1.Univ.Prof.Dr.P.B¨oni
2.Univ.Prof.Dr.W.Petry

gereicDiehtunDissdduertatrchiondiewuFrdeakuamlt¨atfu¨r19.09.2005Physikbameid01er.Tec03.2006hnischenangenommUniveren.sit¨atMu¨nchenein-

”One

icturep

si

orthw

hinesec

a

ndthousa

ingysa

wrds”o

tractAbs

English

Inthecurrentworkvariousnewexperimentalmethodsandcomputationprocedures
inthefieldofneutronimagingarepresented.Thesemethodshaveasignificanttechnical
importanceinnon-destructivematerialinvestigations.
Withstroboscopicneutronradiographyperiodicprocessescanbeinvestigatedonasub-
millisecondtimescale.Thisopensgreatopportunitiesforthestudyandthedevelopment
ofcombustionengines.
Energyselectivetimeofflightneutronradiographyatneutronspallationsourcesuses
theenergydependenceoftheneutroncrosssectiontodistinguishbetweenmaterials.The
energyresolutionofthistechniqueisverygoodandallowstheidentificationofspecific
ials.rmateSoftwaretoolsforneutronradiographydataevaluationanddatavisualizationwere
programmed:calibrationalgorithms,imagedeconvolutionprocedures,toolsforimage
assessment,agraphicaluserinterfaceforfastdatainspectionandmanybatchprocessing
routinesforlargeneutronradiographyimageseries.
Thedevelopedmethodsandsoftwaretoolsareinuseattheneutronradiographyand
tomographyfacilitiesatnewresearchreactorFRM-IIinMunich,Germany.

enfassungusammZ

hDeutsc

IndervorliegendenArbeitwerdenverschiedeneneueexperimentelleMethodenund
RechenverfahrenausdemGebietderRadiographieundTomographiemitNeutronen
vorgestellt.DieseDurchstrahlungsverfahrengewinnenzunehmendanBedeutungfu¨rdie
zerst¨orungsfreiePru¨fungvonneuenMaterialien.
MitderstroboskopischenNeutronenradiographie-Methodek¨onnenperiodische
ProzesseaufeinerSubmillisekunden-Zeitskalauntersuchtwerden.Daser¨offnetviele
verheißungsvolleM¨oglichkeitenf¨urdasStudiumunddieWeiterentwicklungvonVerbren-
nungsmotoren.
Dieenergieaufgel¨osteFlugzeit-NeutronenradiographieanNeutronen-
Spallationsquellenn¨utztdieEnergieabh¨angigkeitderSchw¨achungskoeffizientenaus,um
zwischenverschiedenenMaterialienzuunterschieden.DurchdiehoheEnergieau߬osung
dieserMethodewirddasgezielteSuchennachbestimmtenMaterialienineinem
unbekanntenObjektm¨oglich.
Softwarewerkzeugef¨urdieAuswertungundVisualisierungvonNeutronenradiographie-
Datenwurdenprogrammiert:Evaluierungsalgorithmen,ProgrammezurEntfaltungder
Bildunsch¨arfeundzurKorrekturvonArtefakten,einegraphischeBenutzeroberfl¨achef¨ur
dieschnelleDateninspektion,AlgorithmenzurBestimmungderBildqualit¨atsowieRouti-
nenf¨urdieStapelverarbeitungvongroßenBildserien.
DieneuenexperimentellenMethodenundSoftwarewerkzeugewerdenanden
Neutronenradiographie-und-tomographieanlagenamneuenForschungsreaktorM¨unchen
FRM-IIeingesetzt.

Contents

1Physicsofneutronimaging2
1.1Exponentialattenuationmodel.........................3
1.2Effectsofneutronscattering...........................4
1.3Effectsoftotalreflection.............................6
1.4Spectraleffectsinneutronimaging.......................7
1.5MonteCarloSimulations.............................10

ductiontroIn

2

2Experimentalsetupforneutronimaging11
2.1Neutronsource..................................11
2.1.1TheneutronradiographyfacilityANTARESatFRM-II.......12
2.2Neutronradiographydetectors.........................14
2.2.1Scintillators................................15
2.2.2Detectoroptics..............................16
2.2.3CCDcameras...............................17
2.2.3.1LCDshutters..........................18
2.2.3.2InterlinetransferCCDs....................19
2.2.3.3CCDswithimageintensifier.................20
2.2.3.4ShutteredCCDs........................21
2.2.4TheNRdetectoratANTARES.....................22
2.3Theobjectforneutronradiographyandtomography.............23
2.4Imagequality...................................26
2.4.1Apracticalexaminationofimagequality...............27

3Dataevaluationandvisualization33
3.1HardandsoftwarefordataanalysisatANTARES..............33
3.2Artefactcorrection................................34
3.2.1Beamfluctuations............................34
3.2.2Gammaspots...............................34
3.2.3Scintillatordegradation.........................35
3.2.4Lensdistortions..............................36
3.2.5NRdeconvolution............................36
3.3Normalization...................................37
3.4ReproducibilityanduncertaintyofaNRimage................38
3.5QuantitativeNR.................................40
3.6NRdatavisualization..............................42
3.6.1Softwaretoolsfordataevaluationandvisualization..........42
3.6.2Visualizingdifferencesbetweenimages.................44
3.6.3Fusingimages...............................46

CONTENTS

vi

3.7Limitationsbythehumaneye..........................47

4Measurements51
4.1StroboscopicNR:Combustionengines.....................51
4.1.1FirstNRofarunningcombustionengine...............52
4.1.2Carcombustionengine..........................54
4.1.3Injectionnozzle..............................57
4.1.4FirstneutronradioscopyofancombustionengineatFRM-II....60
4.2EnergyselectivetimeofflightNR.......................62
4.3QuantitativeNR:acompressortyperefrigerator...............69

ANewapplication:fossilstone

BVisualizationmethod:DifferenceNRofanoilpump

esfigurofList

yiographBibl

ionsAbbreviat

soncatiPubli

diCompactcs

onsiluscCon

okOutlo

72

76

80

86

87

88

89

90

91

Introduction

whichuNeutronsethepimagingenetrati(NI)onisofaneucollectivtronseteforrmtheforinvnones-tidegationstructivofethetestiningternal(NDTstr)meucturethodsof
anobject.TheuniqueinformationdeliveredbyNeutronRadiography(NR)andNeutron
inTsighomograptintohyhistori(NT)calconandtriarbutcehetoologithcealdevobeljectopmes.ntofhightechproductsandconcedes
appTheliedfirstatNRtheresimearcageshwreaceretortakenFRMinin1940,Garchatingthewithbeginthermalningofandthesoon80’swiththefasmtnethoeudtronwsas.
Withtheuseoflarge2DdetectorsandincreasingcomputingpowertheNImethodsreally
gotestablished.InthelastyearsnewmethodslikeenergyselectiveNR,phasecontrast
NRandmanynewareasofapplicationsopenedup.Highfluxneutronsourcesanda
growinginterestfromindustryandsciencepushedtheactivitiesinthefieldandgavenew
opportunitiesformethodicalimprovements.InthelastfiveyearseightdiplomaandPhD
thesesaboutNIappeared.
Thisworkisdedicatedtonewexperimentalmethodsfortimeresolvedneutronimaging
andmethodsforquantitativedataanalysis.Itconsistsoffivechapters:
attenChapuationter1modstartsel.FwithorathewiderprincipundleersandtanthdinegphofysicsneutofronNRimandagingpresdenevtsiationtheesxpfonromentithale
simplemodellikeneutronscattering,totalreflection,effectsduetothebeamgeometry
andspectraleffectsarediscussed.
faciliThetiesfexporenrieumentrontalsimagingetupofareaaNIvailfacilitable:yisThtreeatedANTinARESchaptefacilitr2.ywAitththaecoldFRM-IneutItwrono
spectrumandtheNECTARfacilitywithafastfissionspectrum.TheANTARESfacility
ispresentedanddescribedfocussingonthebeam,thedetectorandpossibleobjectsof
.estigationvinAmajorpartofthisworkconsistsofnewandpowerfultechniquesforquantitative
evaluationandvisualizationofNRdata(chapter3).Allimportantstepsoftheevaluation
proceWitdurehinarethisdewscrorkibtedwoandnewtheexpesoftriwmareentalm”Neutrethooneydsew”erefordevviseulopedalizationandisapplexpiedlainined.prac-
timeticalofmeasflighutremNR.enInts:chtheapterstrob4boscothopticechneutrniquesonarimeagiintngrodtecucehdniqanduethandeexptheerenimeerngyts,secarlectriedive
outatneutronimagingfacilitiesinEurope,aredescribedandtheirresultsarediscussed.

1erChapt

Physicsofneutronimaging

TheobjectofinvestigationisplacedinawelldefinedneutronbeamI0(x,y)anda2D
pgraposithyionsimageensitiv(1.1).edeteThisctorshadorecordswimtheagetrconansmtaiinstteindforrmadiatationionabinoutensitttheyIin(tex,rynal),sthteructurradio-e
oftheobject.

Figure1.1:Principleofneutronradiography
Forneutrontomography,aseriesofNRimagesfromdifferentdirectionsaremeasured
byrotatingtheobject.Fromtheimagesstackthecomplete3Dvolumeoftheobjectis
reconstructed.Adetailedmathematicaldescriptionofthetomographicreconstructioncan
befoundin[Kak88,0].Thegreylevelsofeachvolumeelement(voxel)correspondtothe
attenuationcoefficientofthatelement’smaterial.Inthiswaymaterialsdifferinginthe
neutronattenuationcoefficientscanbedistinguishedandtheindividualcomponentsofa
multicompositeobjectcanbeseparated.
NeutronimagingiscomplementarytootherNDTmethodsandoftenpermitsanonde-
structiveanalysiswhenothermethodsfail.CharacteristicforNIistheuniqueability
topenetratelargemetallicobjectsandatthesametimethesensitivityforhydrogenous
materials.Inthefollowingsomeexamplesoftypicalapplicationsarelisted:
•penetrationbehaviourofwaterinsoil,wood,stone,concrete,textilesanddevelop-
mentofhydrophobicagents
•dynamicsoflubricationliquidsincombustionengines
•behaviourofsamplesundermechanical,electricalorchemicalload
•defectdetectioninmetalcasts,3Dmeasuringofthickmetallicsamplesliketurbine
adesbl

1.1Exponentialattenuationmodel3
•measurementofdistributionsoftheelectrolyteinbatteriesandrechargeablebatteries
underdifferentworkingconditions
•inspectionofhistoricalobjects
•autoradiographiesofpaintings
•localizationandpreparationoffossilsinrocks
•controlexaminationsofseals,pyro-elementsandswitchesofhighsecurityrelevance
•studiesofliquidbalancesinplants,animalsandfood
•timedependentliquiddistributionsincoolingaggregatesandheatexchangers
•measurementofborondistributionsinsteel
•inspectionofcarbonfibermaterials
•investigationofhelicopterrotorblades
•investigationofarcheologicalobjects
•measurementofgluedistributions
1.1Exponentialattenuationmodel
Howtheneutronbeamisattenuatedbyobjectsdependsontheobjectgeometry,its
thicknessanditsmaterialcomposition.Inafirstapproximationforneutronsaswellas
forx-raystheexponentialattenuationlaw,alsoknownasLambertBeerlaw,isvalid:
I(z)=I0e−ΣzΣ=Σρ∙ρ=A∙σtotamu∙ρ=σtot∙n(1.1)
I(z)...neutronintensityafteradistancezinmaterialn2
I0...neutronintensitybeforetheobjectcmn2scms
z...distanceinthematerial[1cm]
Σ...attenuationcoefficientcm
Σρ...massattenuationcoefficientcmg2
ydensitmaterial...ρσtot...totalcrosssectiongcm2
A...nuclearmassnumbcmer3[]
n...atomicdensityofthematerialcm3
amu=1,660∙10−24...atomicmassunit[g]g
Howstrongneutronsandphotonsareattenuatedpercmmaterialisdeterminedbythe
materialspecificattenuationcoefficientΣρ.Themassattenuationcoefficient(Fig.1.2)
σisindependentofthematerialdensityandallowsthecomparisonofthemicroscopic
attenForuatx-raionysprΣopertiesincreasbeestwe(benlacktheline)atomswith,thethecrossatomicsectionnums.berZbecausethephotons
isinttheeractattenwithuatitheonelecandtthronseiresnutheltinatgomconicstrastshell.betTheweenhightheerthelemephentsoton(greeneenrgiesandbarerothwneloline).wer
Theattenuationforneutronsdiffersbetweenfastandthermalneutrons.Fastneutrons

1.2Effectsofneutronscattering

Figure1.2:AttenuationpropertiesoftheelementsforneutronsandX-rays[Iae05,3]

4

interactwithmattermainlyviaelasticscatteringwhilethermalandcoldneutronsinteract
morevianuclearreactions.Thus,forfastneutrons(reddots)Σdecreaseswithincreasing
massofthetargetnucleiandforthethermalneutrons(bluedots)Σdependsontheinner
structureoftheatomiccoreshowingnoregularityalongZ.Forthermalneutronsmany
metalsaretransparentandhydrogenincontrastisstronglyattenuating.Neighboring
elementsintheperiodictablehavecompletelydifferentattenuationcoefficients.
ForsmallattenuationstheexponentialattenuationlawcanbeapproximatedbyaTaylor
s:eriseI(z)/I0=e−Σz≈1−Σz/1!Error:Σ2z2/2!(1.2)
ForΣz=0.1(attenuationI/I0≈90.4%)theapproximationerrorisonly0.5%.
TheaimofneutronimagingisthemeasurementofI/I0viathegreylevelsintheNR
imageasexactaspossible.Butthephysicalmodelofneutronattenuationanddetection
hassomelimits,whichcannotbeneglectedandaretreatedindividuallyinthischapter.
Sincetheattenuationpropertiesforfastneutronsarequitedifferent,especiallythelimits
forcoldandthermalneutronsarestudied.Iftheunderlyingprocessescanbeunderstood
andmodelled,I/I0canbemeasuredwithahigherprecisionandevenmoreinformation
canbeextractedfromtheobject.

1.2Effectsofneutronscattering

Accordingtoequation1.2onlyneutronstransmittedthroughtheobjectcontributeto
theradiographyimage.Inrealitythisisnotcorrect.Scatteredneutronscanleadto
considerableartefactsintheNRimage(Fig.1.3).Theredlineandtheblacklineare
theattenuationprofileswithandwithoutthecontributionofscatteredneutronsfromthe
.tjecob

1.2Effectsofneutronscattering

Figure1.3:Neutronscatteringintheobject

5

DeBrogliepostulatedin1923thatthewave-particledualismwasvalidnotonlyfor
photonsbutalsoformassiveparticles.Therelationsbetweenwavelengthλ,energyE
andvelocityvforthenonrelativisticcasearelistedintable1.1.

λ=mnvλA˚≈v[m]λ=√2mnEλA˚≈√E[meV]
sh9.9h9.0
E=h22E[meV]≈81.82E≈1mnv2E[meV]≈5.2∙10−6vm2
2mnλλ[˚A]2s
v=mnvs≈473∙E[meV]v=mnλvs≈λ[A˚]
2Emhm3964
Table1.1:Relationsbetweenneutronenergy,wavelengthandvelocity

Consideringtheneutronbeamasanincomingwavetheneutronsleavingtheobjectare
outgoingwaves.Inthefirstapproximationtheobjectisasourceofsphericalwaves(s-wave
scattering)andtheneutronscatteringcanbeassumedtobeisotropicinthelaboratory
system.Thus,theintensityofscatteredneutronsIshoulddiminishwithdistancedfrom
theobjectwithI∝1/d2.Thisisnotperfectlytrue,especiallyforlighttargetnucleian
isotropicscatteringinthecenterofmasssystemisabetterassumption.Thecorrection
factoraccordingto[Eme82,1]and[Due76,2]iscos(φ)=2/3AwhereisAmassofthe
targetandφtheangleofscatteredneutronsinthelaboratorysystem.Forlighttarget
nuclei(smallA)forwardscatteringissignificantlypredominant.Hydrogenforinstance
hasanaverageangleφof48◦towardstheflightdirectionoftheneutronincomparisonto
90◦inthecaseofisotropicscattering.
Forlowneutronenergiesthechemicalbondofatargetnucleusaswellasthetemperature
ofthetargetcaninfluencethescatteringprocess.Th2eprobabilit2yofscatteringofanucleus
bondinamoleculecanbeuptoafactor(1+1/A)/(1+1/M)highercomparedtofree
nuclei,whereAistheweightofthenucleusandMistheweightofthebondingsystem.The
thermalmotionofthetargetnucleichangestherelativespeedbetweenincidentneutrons
andthesenuclei.Forthispurposeeffectivecrosssectionsdependingonneutronenergy
andtargettemperaturecanbedefined.
Thestrongerthescatteringsourceisandthecloseritistothedetectorthestrongerisits

1.3Effectsoftotalreflection

6

effectontheNRimage.Theobject,thebeamstopandthedetectoritselfarethemain
sourcesofscatteredneutrons.InFig.1.4hydrogenousbiologicalobjectswereinspected
withaminimaldistancetothedetector.Thescatteredneutronsgiverisetobrighthalos
aroundtheobjectsintheNRimage.InordertovisualizethescatteringeffecttheNR
imageisdisplayedinfalsecolors.

Figure1.4:NRofbiologicalobjects,artefactsduetoscatteredneutrons

Bychoosingalargedistancebetweenobjectanddetectortheeffectofscatteredneu-
tronscaneasilybereduced.Evenifthemajorfractionofscatteredneutronsnormally
comefromtheobject,alsothebeamstoporthemirrorisasourceofscatteredneutrons.
Itmakessensetolimittheneutronbeamtothesmallestsizenecessary.Thisminimizes
thebackgroundofscatteredneutrons.

1.3Effectsoftotalreflection
Forneutronsthatimpingeunderasmallangleonaplanesurface,totalreflectionmay
occur(seeFig.1.5).Thiseffectiswellknownanditisusedforneutrontransportin
neutronguides.Inneutronimaginghowever,totalreflectionwasnotobservedyet.
Theneutronrefractionindexnisgiveninequation1.3by:

2λn2=1−NA∙bcoh∙π
NA...atomicdensitycm3
n...refractionindex[1]

(1.3)

1.4Spectraleffectsinneutronimaging7
Figure1.5:Totalreflection
bλcoh......neutcohereronntwavscatterelengthing1length[cm]
cmTheneutronrefractionindexisverysmallandformostmaterialssmallerthanone
(sesurfeacTesab.at1.2)sm.allThanisglesmeansanddthatependtotalontreflechestioncatterinforgneutrlengthonsoofccurthessconlyatterionngverymaterialplane
andthewavelengthoftheneutrons.
materialNA∙bcoh10−6A˚−2materialNA∙bcoh10−6A˚−2
Ni5813.31Aluminium2.08
CNic(keldiamond)9.4011.71VSianliconadium0.272.08
QuartzGermanium3.643.62ManTitanganesiume-2.95-1.95
3.50relvSiTable1.2:Neutronrefractionindex([Fur99,4])
Sincetheeffectisstrongeratlowerenergies,thesedisturbingartefactsappearpresum-
ablyatcoldbeamlines.
Iftheanglebetweentheincidentneutronandthesurfaceisbelowthecriticalangletotal
reflectioncanoccur.Thecriticalangleγforthetotalreflectionofneutronsis:
γ=λ∙NA∙bcoh∙360(1.4)
π2πForNi58theangleoftotalreflectionis0.1◦/˚A.
InFig.1.6thetotalreflectionofaneutronmirrorwasobservedattheANTARESfacility
atandaFRM30%-II.highTheerflNRuxi4mmagemabshoovwseththeeobsurfjeaccte.(blTheack),adistance30cbmetwlonegensuobperjectmirandrordete(m=2)ctor
isabout50cm.Therhombicformofthereflexcanbeattributedtothetiltingbetween
mirrorandthebeamdirection.
Likewisetotalreflectionsfromtherotationtablefortomographyweredetectedinsome
images.Thoughthiseffectdisturbsneutronimaging,itcouldbeusedfortheinspection
ofneutronopticaldevicesinfuture.
1.4Spectraleffectsinneutronimaging
Otherdeviationsfromtheexponentialattenuationmodelarespectraleffects.Theat-
tenuationcoefficientofamaterialΣisnotconstantbutvarieswiththeneutronenergy

1.4Spectraleffectsinneutronimaging

8

Figure1.6:NRofaneutronsupermirror(black)showsa30%higherflux(arrow)5mmabove
thereflectingsurface

Σ=Σ(E).AmeasuredattenuationcoefficientΣ¯foragivenneutronspectrumφspectrum
isaveragedaccordingto:

1Σ¯=φspectrumΣ(E)φspectrum(E)dE(1.5)
umspectr√InthecoldandthethermalenergyregimeΣ(E)isproportionalto1/v∝1/E(Fig.
1.7)formostmaterials.Thatmeanslowenergy-sloworcold-neutronsareattenuated
stronger.

Figure1.7:The1/√Eenergydependenceofthetotalneutroncrosssectionwithnuclearreso-
nancesathigherneutronenergies
Thestrongattenuationofcoldneutronssignifiesthattheyareattenuatedfirstin
thickersamples.Withincreasingpenetrationdepththeneutronenergyspectrumshifts

1.4Spectraleffectsinneutronimaging

9

towardshigherenergiesresultinginasmallerattenuation.Thiseffectiscalledbeam
hardening.Theattenuationcoefficientsforcoldneutronsweremeasuredinanexperiment
atFRM-IIforiron,aluminium,leadandgraphitefromaNRimageofstepwedgeswitha
knownthickness(seeFig.1.8,left).Thedecreasingattenuationcoefficientofiron(black
line)clearlyprovesthespectraleffectofbeamhardening(seeFig.1.8,right).

Figure1.8:NR(left)ofstepwedges,calculatedmeanattenuationcoefficients(right),ironshows
beamharding
Otherspectraleffectsatlowerenergiesappearincrystallinematerials(seeFig.1.9).
Theneutroncrosssectionshowssharpedges,socalledBraggcutoffs,atdeterminedwave-
lengths.Theexplanationforthisisthecoherentelasticscatteringmechanismonthe
atomiclatticesofcrystallitesinmicrocrystallinematerials.

Figure1.9:Coherentneutroncrosssectionofdifferentmaterialsatlargewavelengths,from
5]an01,[SInpoly-crystallinematerialsmanyorientationsexistandneutronswithwavelengths
belowthelatticespacingarescatteredoutofthebeam.Ifλ/2exceedsthelatticespac-

1.5MonteCarloSimulations

10

ingthetransmittedintensityincreasessignificantly(seeFig.1.10).Inthiswayevery
crystallinematerialshowsacharacteristicfingerprintofitslatticetypeanditslattice
parameters.Theprincipleofneutrondiffractioncanbecombinedwiththemacroscopic
spatialresolutionofneutronimaging.Usingneutronsofadefinedenergythischarac-
teristicstructureintheattenuationgivesusachancetodistinguishbetweenmaterials.

Figure1.10:SchemeoftheBragglaw(left),energydependentneutrontransmissioncurveof
iron(right),[San01,5]

1.5MonteCarloSimulations
diffNeutronerentialtranspequatortionccananbbeesdesolvcriedbbedymMonoreteprecCarloiselySimuusinglationthse.tranAftersptheortedefiniquationtion.ofThthise
calcexacutgelateduometrysing,thethesourenergycedeanpdendtheentdetecneutrtoron,cranrosssdomlyectionsshufffleordallpneutossibronletrajectargets.torieInstharee
sumitispossibletodrawaconclusionaboutthetotalneutronflux.MonteCarloSimula-
oftionnsuclareearaninistrndisumepennstsab.letoolforcomplexproblemslikethecalculationandoptimization

2erChapt

Experimentalsetupforneutron
iingmag

TheexperimentalsetupforNRimagingconsistsofthreemaincomponents:Aneutron
sourceincludingthebeamdefiningcomponentsprovidingtheneutronbeam,aposition
sensitiveNRdetectorandtheobjectofinvestigation.Thischaptergivesanoverview
ofneutronsourcesandNRdetectors,withafocusontheANTARESsetupatFRM-II.
Finallythemostimportantimageparametersarediscussed.

2.1Neutronsource

Neutronsforimagingexperimentsareusuallyextractedfromamoderatorofaspallation
sourceorareactorbymeansofbeamtubesandguides.Thevelocitiesoftheneutrons
followcloselyaMaxwell-Boltzmanndistribution:
23P(v)dv=4πm2v2e−2kmvBTdv(2.1)
Tkπ2BThemaximumofP(v)isatE=21mv2=kBT.Thisrelationexplainstheclassification
hot,thermalandcoldneutrons,referringtothemoderatortemperatureTcorresponding
toa2000Khotgraphitemoderator,waterwith320KorliquiddeuteriumwithT≈30K.
Furtherwecandistinguishbetweencontinuousandpulsedneutronsources.Accelerator
basedspallationneutronssourcesguideanintensepulsedprotonbeamonaspallation
target.Inthetargetneutronpulsesareproducedbyanuclearspallationreactionandare
moderatedafterwards.Theneutronspectrumatthesamplepositioninacertaindistance
fromthemoderatoristimedependentandcalledatimeofflightspectrum.Nuclear
reactorsincontrastarecontinuousneutronsourcesproducingneutronsbyanuclearfission
reactioninthereactorfuelelements.
Thelowrefractionindexofneutronsfavorsaquasiparallelbeamgeometryforimaging.
Suchbeamsareoptimizedfortwocriteria,whichcompetewitheachother:ahighfluxand
alowdivergency.Attoday’sneutronradiographyfacilitiesthethermalorcoldneutron
fluxisupto108to109n/cm2satthesampleposition.Thisisfiveordersofmagnitudeless
thanfortypicalX-rayssources.ThesynchrotronbeamattheID22beamlineatEuropean
SynchrotronRadiationFacilityESRF,Grenoble,Franceforinstanceisfocusedonanarea
ofabout3.5x1.5µm2withafluxupto1012ph/sbetween6.5and18keV.Thus,even
athighfluxneutronsourcestheprecisionoftheI/I0measurementisdeterminedbythe
limitedneutronflux.TypicalexposuretimesforsingleNRimagesareintherangebetween
.ms1000and100

2.1Neutronsource

12

ThedivergencyofaneutronbeamisnormallydefinedwiththeL/Dratio,whereDis
thmoederndiamefacteilritiesofittheisatcollimleastatoraboanvedL100.isAthelowbedistanceamdivbetweergencyenob(andjectaandhighL/Dcollimator.ratio)Aist
fluxrealizeacdcordiwithngatolongthedis1/Lt2ancelaLw.Iffromthethedissoutancrceetboettheweenobjeobctjecatttandhecostdetecoftorallowiserlarge,neutthrone
bofeamsizedivd=lergency/(L/D)affeinthctsecthoreiresmpageondshingarpnimageess.,Awherepoinltisinthetheobdistanjeccteisbetblweeurrnedobinjeactspanotd
ig.2.1).(Ftordetec

Figure2.1:Beamdivergencyandspatialresolution

2.1.1TheneutronradiographyfacilityANTARESatFRM-II
InExtpheerimenfollotalwingtheSetup)ANTattAREheShifghacilflituxy(AdresvearcanchedreNeactorutronFTorscomhuographngsreaktoryandMu¨ncRadiographenIhI,y
FRM-II,ispresented(Fig.2.2).

Figure2.2:ShieldingofthetomographyfacilityANTARES
AdetailedstudyincludingMonteCarloSimulationsofeveryindividualcomponent
isgivenin[Gru05,17].TheneutronsourceFRM-IIisafissionreactorwithacompact
corefuelelementofhighlyenricheduranium,whichemitsneutronswithatypicalfission
spectrum.IntheliquidD2Omoderatortankaquasi-isotropicthermalfluxisgeneratedby
moderation.Inthethermalfluxmaximum20cmoutsidethefuelelementthecoldsource,
aliquidD2bottleat25K,ispositionedandmoderatesthethermalneutronsfurtherdown
toacoldspectrum.TheneutronsareextractedbytheANTARESbeamtubeSR4and
reachthesamplepositionthroughaflighttube.Shuttersallowanopeningandclosing
oftheneutronbeamwithinsomeseconds.TheneutroncollimatorwiththediameterD
anditsdistanceLtothesamplepositiondefinesthedivergencyofthequasiparallelbeam
geometrywiththeratioL/D.

2.1Neutronsource

13

InFig.2.3thebeamgeometryissketchedinmoredetailwithsomeadditionalbeam
resparpamectiveters.ecolliThematordistandciameetersL=17.31ofmD=4.3andcmL=an16.7dD=mb2.1etwcmeenareavcollimatorailable.andTherdeteespctorectivfore
L/Dratiosare400and800.

Figure2.3:SchemeoftheneutronbeamatANTARES
Atthesampleposition(18.4mfromthebiologicalshieldingofthereactor)thebeam
diameteris40x40cm2andtheneutronfluxis1.02∙108n/cm2sand2.6∙107n/cm2s.The
neutronfluxwasmeasuredwithaprecisionofafewpercent.Onepointoftheobjectis
mappedontoaspotofthesized=l/(L/D)onthedetector,wherelisthedistancebetween
objectanddetector.Thelargerthedistancefromtheopticalaxisandthelargerthe
distancebetweenobjectanddetectorthebiggeristhedisplacement,dR=l∙R/L.Further
parametersarelistedinTable2.1.
largecollimatorsmallcollimator
800400L/DL’[m]distancetube-nosecollimator3.694.3
2.104.3D[cm]16.717.31L[m]α[◦]0.140.07
β(R)[◦]0.14∙R/(b/2)0.07∙R/(b/2)
Neutronfluxatsample1.02∙1082.5∙107
position[n/cm2s]

Table2.1:ANTARESbeamparameters
Theneutronspectrumisinthecoldregime,butnotfullymoderatedandthereforenot
perfectlycorrespondingtoaMaxwelldistribution.Itcanbeapproximatedbestwitha
MaxwellspectrumatT=42K(seeFig.2.4).Thebluepointsconnectedwiththeblack
linearetheneutronspectrumonthesamplepositionsimulatedby[Gru05,17],thered
lineisthebestfit.Thefluxmaximumofthatneutronspectrumisatawavelengthof4.8
˚A,correspondingtoaneutronenergyof3.5meVandaneutronvelocityof884m/s,to
paleincomparisonwith25meV,1.8A˚and2300m/satthethermalmaximum.
Thestabilityoftheneutronfluxwasstudiedwithaseriesofopenbeamimagesat
atimescaleofabout10seconds.Acomparisonbetweentheimagebrightnessinfour

2.2Neutronradiographydetectors

Figure2.4:NeutronspectrumoftheANTARESfacility

Figure2.5:NeutronintensityfluctuationsatfourdifferentareasintheANTARESbeam

14

differentareasofbeamrevealedthatthebeamfluctuationsareintherangebetween0.4
and0.5%(seeFig.2.5)duetothemovementsofthecontrolrod.Asaconsequenceinthe
caseofprecisemeasurementsthesebeamfluctuationsmustbecorrected.Atspallation
neutronsourcesthebeamfluctuatesmuchstrongerduetotheprotonbeamfluctuation
andsometimesevenacompletebeambreakdowncanhappen.

2.2Neutronradiographydetectors

ANRdetectormeasuresatwodimensionalimageoftheneutronfluxtransmittedbyan
object.Requirementsforsuchasystemare:

2.2Neutronradiographydetectors

15

-highdetectionsensitivity
-highspatialresolution
-hightimeresolution(ashortgatingtime)
-shortcycletimesbetweentheimages(ashortreadouttime)
-goodlinearity
-largedynamicrange
-lownoiselevel
-largedetectorarea
-theavailabilityoftheimagesindigitalform
Amongthemultitudeofavailabledetector3typeslikeconventionalfilm,imageplates,Si
basedflatpaneldetectors,GEMfoilsorHecountingtubesonedetectorismostspread
intheNRcommunity:TheNRdetectorbasedonthecombinationofaneutrontolight
converterplateandaCCDcamera(Fig.2.6).

Figure2.6:SchemeofaNRdetectorandneutronconversionprinciple
6tionmateNeutronrial,sinduwhiccehare(n,comαb)-reacinestibyoneminithttinegLi.visiblTheelighreacttiinon4πpro.AdusctsensitivionizeethCCDesccinameratilla-
detectsthelowneutroninducedphotonflux(Fig.2.6).Withthehelpofamirrorthe
cameradoesnothavetostayinthedirectbeamandcannotgetdamaged.Alighttight
boxpreventslightincidencefromoutside.
thTheecdetameraectionopticspropeanrtdiemsofainlythisbykinthdeofuseddetecCCDtorcareamera.determinedbytheusedscintillator,

2.2.1Scintillators
MostNRgroupsuseNDg(formerlyknownasNE426)scintillatorplatesof6LiF/ZnS
producedbythecompanyAppliedScintillationTechnologies(AST).Theneutrondetection
reactionintheseplatesis:6Li+n→3H+4He+4.79MeV
TheTritiumnucleus3Handtheαparticle(4He)haveahighkineticenergyof4.79MeV.

2.2Neutronradiographydetectors

16

AccordingtotheBethe-Bloch-lawthechargedparticlesarestoppedinthescintillator
materialexcitingtheZnSgrains,whichthenemitgreenorbluelightdependingonthe
dopants.Themeanrangeofthereactionproductsinthescintillatordefinethelower
limitforthespatialresolutionofaneutronradiographydetector.Theefficiencyofthe
thscein300tillatorµmisthicdeteksrmtaninedarddbyscinthetillatorhigh.cSrosimsislarectisconinoftillat6Liorsanemitdisininthetheav20%erager1.ange77∙10for5
thphethotonsickpnesersdeoftethectedsncinetutronillator..ThReescenptlyatialtheressolputionatialisredsoluetetionrmineanddbyrelativtheeegrainfficiesiznecyanofd
scsucinhatillatorconvetherterspatplateialwresereolumetionasurweasdf240or±cold10µnmeuwittronhsaby36%[Blossac02,ine13].fficForiencyain100reµspmethctictok
ausual300µmscintillatorwiththehighestefficiencyandaspatialresolutionof540µm.
Scintillatorarenevercompletelyhomogenous(Fig.2.7),butthatisnotsodisturbingif
thedataisnormalized.Thetimeresolutionofthescintillatorisgivenbythedecaytimeof
tothetehexciteddatasstheateetsofofththeematecompanrial.yTheASTli[ghAstint04,tensit11].ydInecafuystureto10%glassswithicintn85illatorµss[accCizor99,din7]g
prseopemertitoebseanprdomaisinhighgerdcandetecidattionesfeorfficiedyncynamicofn90%euattrontheimagincostg.ofaThloewyerhavlighetbouettertput.decay

Figure2.7:Structureofthescintillator(left):anareaof1.1x1.1cm2withfluctuationsuptoa
few%inbrightness;decaycurveofneutroninducedscintillationlight(right)[Ast04,11]
AttheNeutrographfacility,InstituteLaueLangevin(ILL),Grenoble,France,atNEU-
TRAfacility,PaulScherrerInstitut(PSI),Villigen,SwitzerlandandatKFKIresearch
institute,Budapest,Hungarythescintillatordegradationwithtimewasobserved.Itwas
foundthattheorganicbinderofthescintillatorloosesitstransparencyforlightdepend-
ingonscintillatorthickness.Accordingto[Hil05,6]thescintillatorefficiencydecreases
witheveryhourofirradiationwithneutronsatthethermalneutronbeamwiththehigh
fluxof3∙109n/cm2sby4%.Thatwouldcorrespondtoadegradationof0.23%/hourand
0.03%/houratANTARESforthelargeandthesmallcollimator.Becauseofthecold
spectrumatANTARESthedegradationratesareexpectedtobehigher.Measuringopen
beamimagesduringlargeNRseriestheseartefactscanbequantifiedandcorrectedmore
.yileas2.2.2Detectoroptics
Theopticsbetweenscintillatorscreenandthecamerachiportheentrancewindowofthe
imageintensifierhastoperformthefollowingtask:Dependingontheobjectsizeanarea
upto30x30cm2onthescintillatorhastobemappedontothechipareaA.Sincethe

2.2Neutronradiographydetectors

17

scintillationlightisemittedinto4π,onlyasmallfractionofthelightreachesthechip:
A/(4πd2).Thatiswhyasmalldistanceandabigapertureisthebestforahighefficiency
mapping.Objectivesarechosenbytheirfocallengthfandtheiraperture.Standard
objectiveshaveafocallengthof50mmandaviewingangleof12◦,macroobjectswith
f=28mmandf=16mmaviewingangleofa25◦and45◦.Goingtomacroobjectives
oneshouldkeepinmindsomepossibleaberrationsanddistortions.Forshortdistances
totheobjectplaneandlargeanglesfromtheopticalaxistheimagegetsdarker,blurred
anddistorted.Thisresultsfromtheincreasingdistancetothefocalsphereandincreasing
areasseenatlargesolidangle.Mostlybarrellikeopticaldistortions(somemminthe
corners)andlowerbrightnessinthecornersoftheimageduetoanglesofmorethan40◦
arevisible.Sphericalaberrationscanbeavoidedatthecostoflowerlightintensityby
reducingtheaperture.Ifthecameratoscintillatordistanceisvariablebymeansofa
translationtable,thefieldofviewcanbeadjustedbesttotheobjectsize.
2.2.3CCDcameras
TheCCDcameraisthecorepartofaNRdetectorandresponsibleformostdetector
parameters.TheprincipleofaChargeCoupledDevice(CCD)isshortlydiscussed.
EverypixelofaCCDchipconsistsofanarrayofMetalOxideSemiconductor(MOS)
capacitors.IfonthemetalcontactofaMOScapacitorapositivevoltageisapplied,in
thedepletionzoneontheoxidsemiconductorinterfacechargecanbestoraged.Chargeis
generatedbyincidentlightduetothephotoelectriceffectandbythermalexcitation.By
coolingthedevicethefractionofthermalinducedchargeisnegligibleforshortintegration
times.Inaperiodicarrayofsuchelectrodesthechargecanbeshiftedfromonepixelto
thenextbyvaryingtheappliedvoltages(seeFig.2.8).

Figure2.8:ChargetransportinaCCDpixel
thecAfthieprbevorerydercisharmogevedshiftintototaheserialnextpixelregistethr.ecFromhargethdeistseribrialutionregiofsterthethepixcelrohargewatis
transferredtothereadoutamplifier,whichconvertsitintoavoltageandamplifiesitfor
dataprocessing.Afterdigitalizationoftheimageitisstoredonharddisc.Finallythe
greylevelofeachpixelintheNRimagecorrespondstotheamountofchargecollected
inscinthetillatorpotentareaialcworell.respInondithengcastoeaofaCCDNRpixedetelisctorproptheortionamalountotofthenlighumtbemeritoftendeubytronthse
detectedonthisarea.Thus,thegreylevelintheNRimageisproportionaltothedetected
rons.neutACCDcanbecharacterizedbythefollowingphysicalparameters:
-thequantumefficiency
--ththeechgatinipgsizetimes

2.2Neutronradiographydetectors

18

--ththeefnoisrameerate
-thelinearity
TheCCDcamerapropertiesstronglyaffecttheNRimageanddeterminethepossible
appexactlicationthesh.utFteorrtheoptinewonsofmethomoddernsdevCeCDlopcedamewithrasinaretheisssewnortikal.theWhtimeilethreesolutimetion,reorsolumtionore
ofisdetimagesermineonedbcanytheacquineuretronperfltiuxme(itisisgivtypenbicallyythbeetwframeeenr100ateanofdthe1000detecms),tor.theSninceumbthere
tfrypamicealvrataleuesdepforendsa1konxthe1knpiuxemlbCerCDofcphipixelswithandaonretadheoutreadoutfrequencyfrequofen1cyMHzofthisearounCCD,d
1Hzandacompletereadouttakes1second.Processeswithatimescaleofminutescan
beinvestigatedwellinrealtime.However,periodicprocessesdownto100µscanbe
examinedbystroboscopicneutronimaging.Atriggereddetectorissynchronizedwiththe
Byrepephtitaivseepshiroftincesgstheandtrneutriggeronsingofsignalidenofticaltheptime-rocewinssddoiffewsrenoftthetimecyclwinedoarewsaccinutmheulated.cycle
cantranbsfeerseleCCDs,ctedCandCDsamowithvieacangatedbecimageomposined.tensWifieerhavanedeshxaminutterededLCCCDDssh.utters,interline

2.2.3.1LCDshutters
LCDshuttersworkbyrotatingthepolarizationoflightbetweentwoperpendicularpo-
larizers(Fig.2.9).NewFerroelectricLiquidCrystal(FLC)shuttersofferexposuretimes
downto0.2ms,buttheirstatetransitiontimeisintheorderof70-100µs[Dis05,8].The
cellrequiressomerecoverytime,soitcannotberunwitha50%dutycycle.

Figure2.9:WorkingprincipleandtransmissionofaFLCshutter[Dis05,8]
Moreimportantarethetransmissionandopacityvaluesintheopenandclosedstate,
givenintheorderof30%and0.05%(Fig.2.9).Whilethesevalueslookgoodatfirst,
theymaybesufficientforapulsedspallationsource,butinsufficientforthecontinuous
illuminationofareactorsource.Anenginerunningatidle1000rpmwillrotatewith16.7
Hz.Givenatimewindowof1ms,theshutterwillroughlybeopen16ms,butclosedfor
984msundercontinuousillumination.Theratiobetweenimagesignalandbackground
is(16x30%)/(984x0.05%)=9.8,sotheusefulimagesignalisonlytentimesthe
backgroundsignalandwillbenearlydrownedoutbythetransmissionduringclosedtime.
Inthecaseofaspallationsource,thepulselengthisspreadouttoafewmilliseconds,with
intensityrapidlydroppingtowardslongerwavelengthsandflighttimes,sotheapplication
ofLCDshuttersmaybefeasible,butwedroppedthemforcontinuoussources.

2.2Neutronradiographydetectors

19

2.2.3.2InterlinetransferCCDs
ForconventionalfullareaCCDs,thelightfluxeitherhastobeshutoffforreadoutafter
integrationtime,or-inthecaseofframetransferCCDs-thetransfertimetotheimage
storageareahastobeshortcomparedtotheexposuretime.FullareaCCDsarenot
feasibleforbelow-millisecondexposuretimeswithoutanexternalshutter.ForInterline
transferCCDstheCCDchipareaconsistsofalternatelightsensitivepixelcolumnsand
maskedverticalshiftregisters.Imageinformationfromthelightsensitivepixelscanbe
transferredintothemasked,light-insensitiveshiftregisterwithasingleclockcycle,then
bereadoutconsecutivelythroughthereadoutregisters(seeFig.2.10).

Figure2.10:CutviewofaninterlineCCD:a)chargecollectingmode,b)shutteredmode,c)
chargetransfermode[Oly05,9]
ThisPharotoeaiselesidctronedsbyareanexpcollecteosurdeinctonhetrpolotengatetialandwellabtreloansferwthegatelight(seseFiensitivg.e2.10).pixelIfarea.the
exposurecontrolgateisclockedhigh,thepotentialwellbelowthelightsensitivepixel
isextendedtowardstheevendeeperwellofascavengerdiodeline.Allcollectedphoto
electronsareimmediatelydrainedaway,whicheffectivelyservesasashutter(seeFig.2.10
a)).TheCCDcollectsphotoelectronsonlyiftheexposurecontrolgateisclockedlow,
fortimeming(seeabarFig.rier2.10betb)).weeIfnththeepoppixelosiandtetranthesfscearvgatengereisdioclodcekedandhcigh,onthtrollecingollethectedexpcharosurgee
istransferredintotheverticaltransferregister(seeFig.2.10c)).Afterclockingthe
transfergatelowagain,thecollectedchargeinthetransferregisteristransferredintothe
horizontalreadoutregister(perpendiculartothecutviewhere).Thebigadvantageofthe
interlineCCDisthatnoadditionalnoiseisintroducedbyanimageintensifier,thatitcan

2.2Neutronradiographydetectors

20

beshutteredonthechipitself,andthattheprocessofshort-timeexposureandcharge
transferintothetransferregistercanberepeatedseveraltimesbeforethechargehastobe
movedtothereadoutregister.Thedisadvantageisthat-evenwithverylowlightlevels-
thelowpixelfullwellcapacitylimitsthenumberofon-chipaccumulationstoafewdozen
times,whichisnotsufficientforstroboscopicimaging.Furtherintegrationcanonlybe
performedoff-chipinthecomputer,buttheimagestobesummedupsufferfromreadout
noiseandfromdigitizationerroriftheirintensityisverylow.Suchacamerasystemis
successfullyemployedbytheradiographygroupattheILL.
2.2.3.3CCDswithimageintensifier

Figure2.11:a)IntensifiedCCDcamera(AndorTechnology),b)imageintensifierscheme
Imageintensifiersarelightamplifyingdevicescomingoriginallyfrommilitarynight
visionequipment.ByputtingtheminfrontofaCCD,lowlevellightmeasurementsare
possible.Aphotonfluxamplificationupto108andevensinglephotondetectionbecames
possiblewiththelatestgenerationimageintensifiers.Animageintensifiersconsistsofan
evacuatedtubecomprisingaphotocathode,amicrochannelplate(MCP)andaphosphor
screen(Fig.2.11).Thephotocathodeiscoatedontheinsidesurfaceoftheinputwindow.
Whenaphotonstrikesthephotocathode,aphotoelectronisemittedanddrawntowards
theMCPbyanelectricfield.

Figure2.12:a)SchemeofaMCP,b)MicroscopyimageofaMCP
TheMCP(Fig.2.12)isa1mmthinglassdiscwithahoneycomboftypically10

2.2Neutronradiographydetectors

21

µmMCPfi,neencablihannngetlshe,peachotohwithelectraonsresitosstivloewedcoatindowg.nonAeofhighthepcotenhantialnelsisofapthplieeddisc.acrossWhenthea
phaccotoelerelecationtronhwhicashsuresffiucienltstinaenergyclou,ditofdisloelecdtrgesonsseexiticondngarytehelectMCProns,.Thwhicehidegreenturnofeulectndergoron
multiplication2.13dependsonthegainvoltageappliedacrosstheMCP.Bycascading
MCPMCPscupantobe4gatetimedsvetheryfastampldownificationtonscanbtimeefuscrthale,erinwhiccrehasedmak.eTsheICCvDoltagecameacrasrossvtherye
interestingformanyspectroscopyapplications.

Figure2.13:ElectronmultiplicationinasingleMCPchannel
tonTheoisernoiatiosebiseforequanandtifiedafterintthehesino-calledtensifier.noisTheefacSNRtorisNfbe=SstNRwithin/SNRmaximalout,thgaines,ignbutal
thedynamicrangedecreasesforoneimage.Thedynamicrangeofadetectorcanbe
expandedwithmultiplereadoutsandsummingimagesafteracquisition.Butevenifthe
readoutnoiseisnegligible,therequiredreadouttimeisnot.Thus,foralimitedbeam
consetime,quiteisnceadalovisablewertnoumrebderuceofthereadoutgainsinandordberettetornimpeurovtronedystatisticsnamic.rangeandtogetin

2.2.3.4ShutteredCCDs
Acompletelydifferenttypeofelectronicon-chipshutterwasdevelopedfortaskslike
controllingadaptiveopticsintelescopesortrackingrocketlaunchers[Rei93,10].This
techniqueworksonback-illuminatedCCDs,whichmeansthattheCCDchipisetched
verythindowntoafewtenmicrometersandisilluminatedfromthebacksideopposite
thegateelectrodes(seeFig.2.14).
Thelightsensitiveareaisflankedbytwodrainelectrodeswhichserveaschannelstops
betweenpixels.Usinglayersofdifferentdoping,apermanentpotentialwellisformed
belowthegateelectrodeofthelightsensitivepixelwithinthesiliconchip.Byapplyinga
positivepotentialtothegateelectrode,thispotentialwellcanbeextendedtotheopposite
surfaceoftheCCDtocollectphotoelectronsintheopenmode.Ifanegativevoltageis
appliedtothedrainelectrodes,arepulsivepotentialisformedbelowthedrainelectrodes.
Duetothevaryingdopingbelowthepixelarea,thepotentialwelliscompressedbutnot
eliminateduntiltheremainingpotentialwellispinchedofffromthelowerchipsurface,
effectivelyshuttingitofffromgeneratedphotoelectrons.Thismethodworksbestforshort
wavelengthlight,becausetherangeofincidentshort-wavelengthphotonsinthesiliconis
muchsmallerthanforlongerwavelengths.Themeasuredextinctionratioisgreaterthan
75000,theswitchingtimebelow55ns.Byvaryingthenegativevoltageonthedrain
electrodes,thepinch-offcanalsoberegulatedfromzerotofull,controllingtheamountof
photo-electronsreachingthepotentialwellandthusthesensitivityofthedevice.

2.2Neutronradiographydetectors

22

Figure2.14:Singleback-illuminatedpixelinaCCDwithelectronicshuttershowingthepotential
wellofthepixelandtherepulsivepotentialoftheshutterdrainswitha)shutteropenandb)shutter
10][Rei93,edclos

2.2.4TheNRdetectoratANTARES
Fig.2.15showstheNRdetectorattheANTARESfacility.It2isadetectorboxwith
a200µmthickNDgneutronscintillatorscreenof26x26cmandaCCDcameraon
atranslationtable.Dependingontheobjectsizetheobjectiveandthedistancetothe
scintillatorarechosen.Twocamerastypesareatdisposal:
•ThepeltiercooledAndorCCDcameraDW436withalargechipof2kx2kpixels
andachipsizeof26.7x26.7mm2ispreferredwhenhighspatialresolutionisneeded.
Duetothelargechipthereadouttimeforafullframeis4sathighestreadoutspeed.
Theminimalaccumulationtimeof30msislimitedbythefastshutter.Thequantum
efficiencyof95%andthenoisepropertiesoftheCCDareverygood.Thecamera
manufacturerguaranteesalinearityof1%overthedynamicrangeof16bit(65535
ls).elevgrey•TheintensifiedCCD(ICCD)cameraDH734isespeciallyusefulforstroboscopic
NRmeasurements.Thepeltiercooledchipwithachipsizeof13.3x13.3mm2and
1024x1024pixelsiscoupledtoamultichannelplatebyfiberopticalcoupling.
WiththeMCPgatingtimesdowntonscanbereachedandanexternaltrigger
inputallowssynchronizationwithanexternalprocess.Thequantumefficiency
oftheCCDchipis15%butsinglephotonandsingleneutrondetectionbecome
possiblebecauseofthehighlightamplificationfactoroftheMCP.Theamplification
factorisadjustedbytheMCPgain.Ifitistoohighthedynamicrangeof16bitis
drasticallyreduced.

Innearfutureahighspeedcamerawillexpandtheimagingpossibilitiesandallowreal
timeimagingwithmsandsub-mstimeresolution.

2.3Theobjectforneutronradiographyandtomography

.32

Figure2.15:NRdetectoratANTARES

Theobjectforneutronradiographyandtomography

23

FortheNRthesizeoftheobjectisrelevantinonedirectiononly,thethicknessinbeam
direction.Thethicknessmustbesmallenough,otherwisethecompleteneutronfluxis
absorbedandtheNRimageisblack.Iftheobjectthicknessisonthelimitofzerotrans-
mission,theneutronstatisticsgetspoorandtheinterestingcontrastsverynoisy.ForNT
thethicknessconditionmustbefulfilledinthetwodirectionsperpendiculartotherota-
tionaxis,buttomographywithmissinganglesisahotresearchtopic.Theobjectdetails,
onewantstoseeintheinspection,mustbebiggerthanthespatialresolutionandmust
produceacontrast,whichtheinspectorisabletorecognizeatthegivennoiselevel.
Thematerialpropertiesbecomerelevantinformoftheneutronmassattenuationcoeffi-
cientµρandthedensityoftheobject.Differencesbetweenconstituentindividualelements
andevenisotopescanbedetectedaccordingtoFig.1.2.Theobjectmustnotbetoostrong
attenuating,ontheotherhandthedetailmustnotbetootransparent.Assumingarela-
tivenoiseσµ/µ=1%intheopenbeamintensityandabackgroundnoiseσbgduetothe
detectorof1%inrespecttotheopenbeamsignal,theminimaldetectablethicknessis
about1/250∙L1/10andthemaximalthicknessonecandistinguishfromthebackground
isabout2∙L1/10.L1/10isthelengthoftenth,thethicknessofmaterial,whichabsorbs
90%andstilltransmits10%oftheradiation.Itcanbecalculatedfromtheattenuation
coefficientµbytherelation:
ln(1/10)2.30.69
L1/10[cm]=µ[1/cm]≈µ[1/cm]L1/2[cm]≈µ[1/cm]≈0.3∙L1/10[cm](2.2)
Thepossiblethicknessesfortheinspectionwiththermalneutronsaretheredbarsin

2.3Theobjectforneutronradiographyandtomography24
Fig.2.16forseveralmaterials,theyellowlinesonthebarsareL1/10.FortheANTARES

Figure2.16:Possibleobjectthicknessesforseveralmaterialsforthermalneutronradiography
thfacilietrmalysptheectattera.nuationEstimatincoegtfficheienctshanµgeareinthhigheerattenuthanationthosecoeinfficientliteratuwithrethebase1/d√onElathew
andthemeanenergiesof25meVatathermaland4meVattheANTARESfacilityyields
toalargerattenuationcoefficientbyafactorof2.5.Foraprecisecomparisontheindividual
spectrumandtheenergydependentcrosssectionmustbeused.Asaconsequencethe
smconalletrastsrpareenetratistronongerdepanth.dtheThedescteinctiontillatoforthdetecinmtionaterialeffisisciencyimpinrocvereasedatdliktheewicossetofwitha
thecoldspectrumduetothecrosssectionof6Li.Theaverageattenuationcoefficient
µandthecorrespondinglengthoftenthLofneutronsfortheANTARESspectrum
weremeasuredforthesomematerialsby1/usin10gstepwedges,seeTab.2.2.Measuring
Thematedrialifferethinctkntheicssknanessdesneutrinaonstepattenweudgeation,allotwhetoattsetnuudyationthecoeeffiffectcienoftbweasamhdeterardenimined.ng.
TypicallyatathicknessofL1/10beamhardeningstarts.
materialµ[1/cm]L1/10[cm]
Fe1.12±0.022.04±0.04
CAl0.090.29±±0.010.015.625.40±±0.353.5
Pb0.27±0.018.2±0.4
Cu1.05±0.022.18±0.05
TPEeflon5.110.26±±0.110.028.70.54±±0.60.01
H2O3.6±2.00.63±0.3
Oil4.4±2.50.52±0.3
Table2.2:MeasuredattenuationcoefficientsµandlengthsofatenthL1/10atthecoldspectrum
SAREANTof

2.3Theobjectforneutronradiographyandtomography

25

ForthemeasuredvaluesofTable2.2thescatteredneutronswerenotconsideredand
eseffpecetciallwasyforminH2imizOedandbyoilthethuatsemofayableadeamtodevilimiter.atingInrethesultsc.aseForofthaedothyneamricprmaterialsocesstheall
thementionedpropertiesmustbefulfilledforalimitedirradiationtime.Theconsequence
isashorterbarfortherespectivesampleinFig.2.16.
Iftheirradiationtimeislong,thesampleactivationbecomesthelimitingfactorforneu-
tronimaging.Certainnucleiareespeciallycriticalforneutronactivationandforlarge
irradiationtimesitisnecessarytoassurethatthemassofcriticalnucleiintheobjectis
smallenough(seeTab.2.2).TheNRfacilityisoptimizedforhighradiationexposure
withstronglyneutronabsorbingshieldingmaterials,whichhardlygetsactivated.Thus,
nthuecleimaarejorpartcobalt,ofthemanganactivese,ationgoldcomeandsfcoppromer.theOifrradcoursiatedetheacsample.tivationTheismosprtopproroblemtionalaticto
ththeeNRirradstationiationctianmenotandbetheentereamoundtandofnothemoreexpmeosedasureisotopmeen.tsIncanthbeecapseerfoformeandacuntivtilationthe
activationhassunkunderthealloweddosevalue.
13stableisotopeisotopicabundance[%]14isotopeafterneutroncapturehalf-life
15NC1.10.3716NC7.13s5730a
191823ONa0.210024ONa27s14.96h
272627AlMg1110028AlMg2.2min9.46min
30Si3.131Si2.62h
55Mn10056Mn2.58h
5958Fe0.35960Fe45.1d
5.2aCo100Co63Cu6364Cu12.7h
93109Nb10011940Nb20000a
197AgAu50100198AgAu2.6h127d

Table2.3:Halflifetimesofisotopes
Fortunatelymostisotopeslike14C,16N,19O,28Alhaveshortdecaytimesandafter
afewminutesthetomographystationcanbeenteredagain.Takingouttheirradiated
objectrequiresthedosefreemeasurementoftheradiationprotectionstaff.Inthecaseofan
activation,theobjectmustremaininthetomographystationuntiltheradioactivenuclei
havedecayedandtheobjectsfulfillthedoselimitof10µSv/h.Themirrorofthedetectoris
permanentlyexposedtoneutronsandshowedafterlongermeasurementsahighactivation
level,whichwasidentifiedtocomefrom24Nanucleiwithahalflifeof14.96hoursandwill
soonbereplacedbyaSiwaferasamirror.OntheANTARESsamplemanipulatorobjects
withaweightupto500kgcanbehandled,thepositioningworkswithaprecisionofamm
andtherotationispreciseto0.1◦.Formanyinvestigationsacomplexandlaborioussample
environmentisnecessarybecausediversestatesaschemicalenvironments,temperatures
orpressuresmustbecomparedinordertoanswerNDTquestions.Itshouldnotbe
forgottenthatduringthetimeconsumingpreparationofacomplexsetup,preciousbeam
timeislost.Lastbutnotleast,thesecurityrequirementsmustbefulfilled,thatmeans
aminimizationoftheamountofcombustibles,explosivesandmaterials,whichgeteasily
activated,inthetomographystation.Theirradiationtimehastobeminimized.

2.4Imagequality

26

2.4Imagequality
Itisdifficulttodefineparametersfortheimagequalityandthereexistmanydifferent
parameters,whichconfusethephysicistratherthantohelphim.Thischapter,influenced
by[Has91,19],triestogiveanorientationinthisimportantfieldandtointroducethe
parameters,describingthemeasureddatainthelastchapter.Thecombinationoffour
parametersisusefulinconsideringthequalityofanimage:thespatialresolutionΔx,the
timeresolutionΔt,therelativenoiseσandthecontrastC.
•ThespatialresolutionΔx:Therearemanydifferentdefinitionsofthespatialres-
olution.Intuitively,thespatialresolutionΔxofanimagingsystemcanbedefined
intermsofthesmallestdistancethatseparatestwoobjectsandmakesthemstill
appeardistinct.Ifthisdefinitionisnotpreciseenough,thepointspreadfunction
PSForthemodulationtransferfunctionMFTareused.ThePSFistheimageofan
idealpointobject-thetwodimensionalresolutionfunctionoftheimagingsystem.
Itisnotasingleblackpixelbutaspotduetotheblurringoftheimagingsystem,the
amountofblurringandtheshapeofthespotisameasureofthespatialresolution.
NormallythePSFcanbedescribedwellwitha2DGaussiancurveandthefullwidth
athalfmaximumFWHMofthePSFisusedasΔxPSF.ForaNRtheresponsespot
dependsonthedetectorresolutionoronthebeamdivergency.Thespatialresolution
ispositionandsignaldependentΔx=Δx(x,y,µ),whereµisthemeangreylevel-
thesignal.FortunatelyforlinearshiftinvariantsystemsΔxisconstantandtheNR
imagecanbedescribedasaconvolutionoftheobjectattenuationOBJandthePSF
(equation2.3).ThePSFistheconvolutionkernelintheequationbelow.
xyNR(x,y)=OBJ(ξ,ν)PSF(x,y;ξ,η)dξdη(2.3)
00InthiscaseanexperimentaldeterminationofthePSFcharacterizesthespatial
resolutionofthecompleteimage.SincedirectmeasurementsofthePSFaredifficult
theedgespreadfunctionESF,theresponseofasharpknifeedgeabsorber,canbe
measuredeasily.TheESFistheintegraloverthePSF.IfthePSFcanbefittedwell
byaGaussian,theESFisclosetoanerrorfunction(seeFig.2.17).
Thedistancefrom10%to90%oftheESFsignalΔx10/90isalsoacceptedasa
measureforΔx.InliteraturemostlyΔxPSForΔx10%MTFisdealtasthespatial
resolution.Therelationsbetweenthethreedefinitionsare:

ΔxPSF=0.918∙Δx10/90Δx10%MTF=0.358∙Δx10/90=2∙xforPSF(x)=0.9
(2.4)

•ThetimeresolutionΔtistheexposuretimeoftheimageandrespectivelythetime
ofneutroncollection.Δtiscontrolledbythecamerasoftware.Alongexposuretime
offersbetterimagestatisticsbutconsumesmorebeamtime.Itshouldbechosenas
theshortasrequesposstedibledetailons.theThecondtimeitionresthatolutiontheisimagebequlimitedalityibsygotheodeopnenoughingtotimereofcognizethe
opticalshutterortheimageintensifierfortheshorttimes.Theupperlimitfor
timeresolutionisthedynamicrangeofthecamera.Itmustbereadoutbefore
pixeloverflowoccurs.Thedynamicrangecanbeexpandedbymultiplereadoutand
summationofimages.

2.4Imagequality

Figure2.17:Themathematicalfunctionsusedfortheresolutiondetermination

27

•pTheositnionoiseandσissignthealdepuncertainendentt:yorσ=imσp(x,recyi,sµion).ofButhetiinmthage.ecaseNormallyofathehomogenoisneeouiss
detectorandahomogeneousbeamwecansimplifythenoisemodelandconsider
twoformsofnoise.Thebackgroundnoiseσ,limitsthedetectionofsmallsignals
andthesignaldependentnoiseσlimitsthebgprecisionofsignalmeasurement.The
componentsofσbgarethereadoutnoiseandthethermalnoiseofthecamera,the
thedeterminedrmalphbotoyacathodardkecurrnoiseentofmanieasuremntensient.fier,Thoreorneutiginronofbacthekgrousignalnd.noiItsecσanbaree
fluthectphotuatinongbeanamdinthentensiteuytronorscquinantilltumatornoise,degradatitheon.amplifiTheernstanoisedardindteheviationCCD,frthome
4thethemeanrelativσµedinvidoiseedbinyththeemopeeannbµeisamthearseiagnalwasnoiseσcalculatedorrelativforemosntoise.imaInges.cThapterhe
inverseofσisthesignaltonoiseratioSNR.
•TheneutroncontrattenastuCationisancoobeffijeciencttpropoftherteyobanjecdt.canInbetheestiNRmimatedageviaitfinthicallykniessstheandsignalthe
differencebetweenanobjectanditsbackground(seeFig.2.18).

Onlythecombinationofallparametersgivesagooddescriptionoftheimagingsystem
anddeterminestheimagequality.IfaNRimageissmoothedforinstance,therelativenoise
σcanbereducedbysacrificingspatialresolutionatthesametime.Whatimagequality
is,testidepngefnordsinonstantheceithenspecabitedlityobtojecrtecogniandzisedaefidetailnedbisythemoreincsrpeucialctor.thanInnalonowdenstruoise.ctivIne
additiontothediscussedparameters,thesizeofandetail,thewayofvisualizationand
theexperienceoftheinspectorhasanimportantinfluence(seechapter3).

2.4.1Apracticalexaminationofimagequality
ForbothNRdetectorsatANTAREStheESFofastrongknifeedgeabsorbercloseto
thedetectorwasmeasured(seeFig.2.19).AsknifeedgeabsorberaCdplateof2mm

2.4Imagequality

Figure2.18:Contrastandnoiseinaneutronradiographyprofileofthestepwedgeabove

28

thnoticknonelyssinanthdeanuwidmbtherofof1pcmixelswbasutusalsoed.inThemmknowitwnhwtheidthIDLallopwsrogramthecalculatistufenfit.proonofΔx.

Figure2.19:NRoftheCdabsorber(left)forresolutiondetermination,photographoftheCd
t)(righerabsorb

steepnTheessedgeparampreterofileofextrtheacerrtedorfufromnctionNRΔximageiswtheascrfittucialedreswitholuantionerproraramefuncter.tion.The
Preconditionsforthismethodare:
of-AthestronCdgnseuampltroneupabstoorb5◦erinwiththeabsehamarpdirkneifectionedgedoeparsnallelotstopthreadedethetectoredgepandlane:affeAtilctΔtingx
.ablyremark-Anegligiblefractionofscatteredneutrons:Scatteredneutronfromtheobjectorfrom

2.4Imagequality

29

Figure2.20:ThegreylevelprofileoftheCdedgeabove,alsocallededgespreadfunctionESF

theshieldingbroadenstheedgeprofileanddeterioratesΔx.
-AshortdistancefromtheCdknifeedgetothedetectorplane:Otherwisethebeam
divergencyincreasesΔx.
IftheseconditionsarenotfulfilledstillalowerestimationofΔxispossible.Thederivative
oftheESFinFig.2.20isthepointspreadfunctionPSF.
ThePSFinFig.2.21wasextractedfromthederivativeoftheprofile.Inmostcasesthe
PSFwillhaveaGaussshapebecauseoftheseveralinfluencesofthebeamdivergency,
thescintillatorspotsizeandtheobjectiveofthecamera.InFig.2.21thePSFwas
fittedwithaGausscurvewithaFWHMof0.6mm.Inthecasetheimageblurring
comespredominantlyfromtheconverter,thecurvewouldbeLorentian[Bac02,13].The
normalizedfouriertransformofthedatainFig.2.21isthesocalledmodulationtransfer
functionMTF(seeFig.2.22).
TheMTFisamoreprecisedescriptionofthespatialresolutionthanthedistanceof
tworesolvablestructures.TheMTFistheratiobetweenaspatialmodulationamplitude
intheobjectattenuationandthemodulationamplitudeintheimageversusthespatial
modulationfrequency.AtlowestfrequenciestheMTFisnormalizedtoone,athighest
frequenciesitmustturntozero.ThehigherthespatialfrequencyatwhichtheMTF
dropsandthesteeperthedrop,thebetteristheimagequality.Sometimesspecialopti-
calpatternsorappropriatecalibrationobjectsareusedforadirectmeasurementofthe
MTF.InthesemilogplotinFig.2.23theMTFwasfittedwithaGaussianandwithan
exponentialfunction.TheGausscurvefitsthedataslightlybetter.
ThetimeresolutionΔtforindividualimagesistheexposuretimeoftheCCDΔtexp
setbytheuser.TheusedscintillatorputsalowerlimittoΔtwithitsscintillationdecay
time.Forrealtimeimagingtheframeratebecomesrelevant.Theframeratefcanbe
approximatedforbothcameras:f=1/(Δtexp+Npixel∙Δtreadout+Ncolumns∙tvss).The

2.4Imagequality

Figure2.21:ThePointSpreadFunction

Figure2.22:MTFoftheimagingsystemwithaGaussandaLorentianfit

30

readouttimetreadoutcanbesetto1,2,8ore16µs/pixel,thetimeforaverticalshifttvss
is16µsfortheICCDand112µsfortheCCD.
Thenoisecharacteristicsbecomecrucialinthecaseofsmallcontrasts.Thebackground
noiseσbgismeasuredwithadarkimage.Thesignalnoiseσ(x,y,µ)ismainlyphotonand
neutronshotnoise,whichcandependonpositionandthesignalµ.Inafirstapproximation
itcanbecalculatedfromthemeanandthestandarddeviationinanopenbeamarea:σ
=σµ/µ.Normalizingandcorrectingtheimageforknownartefactslikedescribedinthe
nextchapterallowsamoreexactestimationoftherelativenoiseσ.Ifthebeamisnot
homogenousσmaydifferindifferentpositionsalsoinnormalizedimagesandmoreNR
imagesmustbeacquiredinordertocharacterizeσ(x,y,µ).Themostimportantnoise
componentforshortexposuretimesistheneutronquantumnoise,whichisproportional
tothesquarerootoftheneutronflux.Thatwasconfirmedimpressivelyattheneutron
spallationsourceatPSIafterabeambreakdown.Theintrinsicneutronquantumnoise

2.4Imagequality

Figure2.23:SemilogplotoftheMTFwithaGaussandaLorentianfit

31

behaviorfollowsperfectlythesquarerootdependenceasvisibleinFig.2.24.
ForbothcamerasatANTARESthemeangreylevelµandthenoiseσweremodelledin
dependenceofthecameraparametersforareactorpowerof20MWandL/D=400,see
equations2.5and2.6.DarkimagesofthecamerasshowedagreyleveloffsetCof1350±19
forthereadoutwith1µs/pixeland355±9forthereadoutwith2µs/pixelduetotheDC
offsetintheADCcircuit.Thevariationoftheoffsetisthedarkcurrent.Themeangrey
leveldependsontheobjectivebutisindependentfromthedistancedbetweencamera
andscintillator,sincethelightintensitydecreaseswith1/d2buttheareaseenonthe
scintillatorincreaseswithd2.Thevariablebinstandsforthebinning,bin=2meansthat
4pixelsarecombinedintoonepixel.InthecaseoftheICCDcameratheintensifiergain
isanadditionalparameterwhichcanbesetfrom0to255.Nisthenumberofidentical
imagesacquired.TheoffsetinthenoiseAandBvariesbetweenthemeasurementsand
mustbedeterminedexperimentally.
FortheICCDatT=-20◦Cwitha50mmobjectivewitha2mmadapterring:
√µ≈C+30∙Δt[ms]∙e−ga56in∙bin2σ∝√µ+A(2.5)
NFortheCCDatT=-50◦Canda50mmobjective:
√µµ≈C+33000∙Δt[ms]∙bin2σ∝√N+B(2.6)
sitionIfassomweellasdetailsgeabomeouttrytheandobmjectaterialofinvcompestigationositionlikofeththeegedetailsometrarey,thknoewn,materialtheprospcompec-o-
tiveattenuationandcontrastscanbeestimated.Finallythecameraparameterscanbe
calculatedsuchthatthenoiseislowenoughtoseethenecessarycontrasts.

2.4

ageIm

reFigu

wndo

yitqual

2.24:

Noiseσersusvtheentronuxfluixel(pme),anmeasreudatSIPterafabeamb32

reak

3erChapt

Dataevaluationandvisualization

Theaimoftheevaluationistheextractionoftherelevantmeasuredphysicalquantityof
measureanditsuncertainty.ForNRthisistheneutronattenuationI/I0(x,y)foreach
andeverypixel.Theevaluationconsistsoffivesteps:

1.Correctionforknownartefacts:ThefollowingNRartefactsasgammaspots,beam
fluctuations,scintillatordegradation,lensdistortionsandimagedeconvolutionare
treatedinthischapter.
2.Normalization:AllpositiondependentoffsetsandsensitivitiesoftheNRdetector
canbecorrectedbythiscalibrationmethod.
3.Calculationoftheuncertainty:InordertopredictfutureNRresultsandtostart
withquantitativeanalysisthispartisabsolutelynecessary.
4.Quantitativeanalysis:Quantitieslikevolume,areaandattenuationpropertiescan
beextracted.
5.Visualization:Sometimesquantitativeresultsarenotnecessaryandtherightvisu-
alizationanswersthequestion.

3.1HardandsoftwarefordataanalysisatANTARES

AttheANTARESinstrumentatypical16bitNRimageof2048x2048pixelhassize
ofSim8ilarMB.dataForaamounNTttsypicareallypro400ducedNRs,atorNR3.2serGBiesofofdobatajectsarelikerecordedforinstwithiancen3dynhouramics.
prNRepmearationasuremetimentse.xceAedsthmaximalemedatasuraingratetimofe1byGBfar,perbuthoustirllisallexptheecteprod.ducOfedcdouatarse,muthset
beevaluated.Thatsetsconsiderablerequirementstohardwareandsoftware.
toOuar1hGardbitwarenetworkconsistsinofordaerPCtocluredusterceinatran100sferMbittimes.netwTheork,datawhicishwillacqusoiredonbwitheupangradedIntel
Penstoredtiumon4afilePCservwither3withGHzadailandy1bacGBkupsyRAM,stemw,hwhichicchonalsotrolsservtheseascameaplatra.formThefordatadatais
isexcdhoneangebybettheweenmemthebersuseofrs.ourThgreoupdataonsevaltandaruationdPanCsd.Fithetnallyom,ogrforaphithecreccalculationstronuctionand
memoryintensivevisualizationa64bitAMDDualOpteronLinuxmachinewith2.4GHz
and16GBRAMisatourdisposal.
used.DifferenThtesoftAnwdoraretoMCDolsfanorddataIStaracqusoftwisitionare,dwriteatathperorcesawdsing,atafandromdatathedevisualitecztorationtotharee

3.2Artefactcorrection

34

ordharedrtdiospk.revAllentthdeataNRlossedatas.isThisstoredistructurn16einbitcltifudesfordmirecattinoriesaspeforcialopdienrebectoryamimstruages,cturedarkin
picimagetures,ofthertheawobdjeata,ct.aOndetailedlywithprotoccarefuollfileanofdthdileigementasurestrucmentutsrinangdofifathevdailablata,eathdeigicamdata
evaluationcanbeautomated.Innearfutureaprofessionaldatabasewillbeused.The
adatricahisleibrvaluaryateofdprwoivthenspmecatialh,pstatrogrisamticss,diemvageelopepdroincesIDL.singIDLandissignapalowproerfucelsssiofngtwarroutewithines
forLAPdAataCKannumalysis,ericalincludlibringary.mAanycoreofthestrengthbestofalgorIDLithmsisifmromageproceNumericalssing.RecMiapteshemandaticthale
andlogicaloperationscanbeappliedtomoredimensionalarrayswithsimplecommands
prandocearessinpgroofcesscomedpvleexrydatafast.Avstructuailreables.dForirecthtoreymeandmoryfileinaccetensivsseroutvisuinesalalizationlowallofk3DindNTof
datasetsthesoftwareVGStudioofthecompanyVolumeGraphicsisthetoolofchoice.A
pbigossiblimperotovehmeavnetwasremotetheaccuseessoftotheothervirtuWalindneotwsworkmacclienhinets.(VForNC)thesoftwcommare,onwhicdevhelopmadmeenitt
ofIDLprogramsaversioncontrolsoftwarelikeCVSisstronglyrecommended.

3.2Artefactcorrection
3.2.1Beamfluctuations
Especiallyatspallationneutronsourcesthebeamintensitymaystronglyfluctuatewith
time.Theconsequenceisanunknownintegralfluxandaresultingdifficultytocompare
theimagesofvaryingbrightness(showninFig.2.5fortheANTARESbeam).Forafirst
checktheaveragegreylevelofaregioninaseriesofimagesanditsstandarddeviationcan
beplottedinadiagram(programbatchinfo.prodoesthatjob)likein3.10.Theprogram
batchinfo.prochecksachosenareaineveryimageforitsmeananditsstandarddevia-
tion.Iftheeffectplaysanimportantroleitcanbecorrectedbyamultiplicationwitha
constantfactorinordertogetaequalintensitylevelineveryimage(roi_normalize.pro).

3.2.2Gammaspots
Adifficultproblemtoovercomearethewhitespotsintheimage,comingfromgammas
hittingtheCCD.Filtershavetobeusedtogetridoftheseartefacts.Fortheidentification
ofagammaspotintheimagetwostrategiesarethinkable:Tocomparethegreylevelwith
thesurroundingpixelsortoacquiremoreidenticalimagesandtocomparethesamepixels
ofeveryimageintheseries.BothwasimplementedinIDL:
displayratio=disp_ratio,,u_filtered=serienfilter(dirnamebw_ufilt)n_sig_ufilt=0.1,displaybright=disp_bright,Thesurroundingfilter,ufilter,checkstheneighboringpixelsofaboxwithaboxwidth
bwufiltandcalculatesthemeanvalueofthatbox.Ifthegreylevelofthepixelismorethan
nsigufilttimeshigherthenitisreplacedbythemeanvalue.Thisfilterisveryeffective
andmostlyused,butbwufiltandnsigufiltmustbecarefullychosen.Goodvaluesare
bwufilt=6andnsigufilt=2.5-3.5.Ifthevaluesaretoolowobjectinformationespecially
attheedgesgetslost.Ifontheotherhandthevaluesaretoohighthegammaspotsare
.dlterefinotdisplayratio=disp_ratio,,s_filtered=serienfilter(dirnamebw_sfilt)_sfilt=0.1,n_sigdisplaybright=disp_bright,

3.2Artefactcorrection

35

seriesTheandsericesalculfilteater,ssthefiltemer,dcianhec(ksthethemiddlsamee)pvaluixelseofofathesneumbgreyerblevwels.sfiltIfoftheimagesgreyofleveal
thofestheeries.pixelThisismfilorteerthwanorksnwsigellsfiltonlytimewithsmhighoreerthanthenthitisreereplimageaceds.bythemeanvalueof
seTheriesbefilstterresu(seltsehaFig.vebee3.1).nInobtainorederdtowithsaeectomhebinsucceationssofofbaothfiltfierltersas,psectarialtingmewthiothdthfore
thevisualizationwasapplied:falsecolorimagesofthedifferencebetweenfilteredand
unfilteredimage.

Figure3.1:CutoutofaNRofatelephonewithmanywhitespots(left),afterthefiltering(right)
InFig.3.2adifferenceimageshowingonlythefilteredgammasanditsgreylevel
histogramisdisplayed.Inthehistogramthegreyleveldistributionofthegammaevents
becomesvisible.Almostallgammaeventsincreasedthegreylevelonlyslightlybyabout
It100.couldInbprepincipossileblethetohiprsedtogrictamsomconetainmaterialssomesinsidinfetheormationobjeabctoubtythstuedyingammagtheenergammgiesa.
um.trecsp

Figure3.2:Differenceimageshowingtheisolatedwhitespots(left),greylevelhistogram(right)

3.2.3Scintillatordegradation
Theeffectofscintillatordegradationwasobservedforaccumulationtimesofafewhours
athighneutronfluxes[Hil05,6].ForNTmeasurementsorlongNRseriestheycanbe
identifiedquiteeasily.Takingandcomparingopenbeamimagesbeforeandafterthe
longmeasurementsshow,thatareascoveredbyastronglyattenuatingobjectbeforeare
brighterduetolessscintillatordegradation.Partiallytheseeffectscanbecorrectedby

3.2Artefactcorrection36
areslultsinearwinereterpobtainedolationbofythetakingopenbealternatelyamimopagesenbbeforeameandimagesafterandtheimagesmeasurofemethent.objeBesct.t
3.2.4Lensdistortions
Figure3.3:a)Imagewithlensdistortionsb)thesameimageafterthecorrection[Sch04,24]
InthecornersoftheNRimagedistancesbecomesmallerthantheyareinimagecenter
can(2.2.b2)edidueentotifiedopticbalestdisbytorsutionsbstitu.Suctinghthedistortiscinonstillatcanordiswithturbanquoptanicaltitatitesvetanalypatternsis..ThTheey
deviationscanbecorrectedwithaninterpolationalgorithmlikelens_distortion.pro.
3.2.5NRdeconvolution
NRimageunsharpnesscanhaveseveralreasons:theopticalmisalignment,thebeamdi-
vergencyincombinationwithalargeobjecttodetectordistanceortheintrinsicscintillator
unsharpness.IfthePSFispositionindependent,theNRimageisaconvolutionofthe
attenuationfunctionOBJandthePSFaccordingtoequation2.3.Theconvolutionin
chapter2correspondsinFourierspacetoamultiplication:
FFT(NR)=FFT(OBJ)∙FFT(PSF)(3.1)
ThecomplexfouriercoefficientsofthePSFaremultipliedwiththespectrumofthe
OBJ.KnowingthePSFitispossibletocalculatepointwisetheobjectattenuationOBJ
byinvertingtheconvolution-bydevonvolvingOBJ:
OBJ=FFT−1FFFFTT((PNSRF))(3.2)
Unfortunatelythisapproachleadstotwoproblems:
1.ThePSFmaybesingularandcanbenotbedeconvolvedbecausethefouriercoeffi-
zero.aretsncie2.alrTheeadNRydamimagepedcdonuetainstontheoise,conwvholuichtionaddsprotocessthe.signWithalthatehinoiseghampfrequencieslificati,onwhicdurhingis
theinversionprocesstheimagecontentisdestroyed.
TheWienerfilterWFisamoreadvancedapproachandtakesthenoiseRintoaccount,
imageB=OBJ+Randminimizesthevariance.
FFT(PSF)
OBJ=FFT−1(FFT(NR)∙WF)WF=|FFT(PSF)|2+|FFT(R)|22(3.3)
|FFT(NR)|

37Normalization3.3Adifficultyistheestimationofthepowerspectrumoftherelativenoise
|FFT(R)|2/|FFT(NR)|2.Itcanbedeterminedexperimentally.Ifitiszero,theWiener
FilterturnsintotheinverseFilter(equation3.2).Theresultoftheapplicationofthe
WienerfiltertoaNRimageisvisibleinFig.3.4.InthedeconvolvedNRimagemore
detailslikethethreadofthescrewortheoilrestinthetapholecanbeidentified,evenif
thenoiseincreases.
Figure3.4:NRcutoutofascrew(left)thesameNRcutoutafteradeconvolutionwiththe
Wienerfilter(right)
Forbetterdeconvolutionmoredetailednoisemodelshavetobedeveloped.There
arestatisticalandrecursivedeconvolutionalgorithmslikethemaximumlikelihoodorthe
maximumentropyalgorithm.Formoredetailssee[Jai89,21]and[Pra91,22].
3.3Normalization
Sincetheneutronbeam,thescintillator,theopticsandtheCCDarenotperfectlyhomo-
geneous,artefactsappearineveryNRimage.Theyaremisleadingintheinterpretationof
theimagedata.s)andWhendarkevercurrpeosntsibleimagethesDIdata(orisbcaliblackratedimagewiths).opOBenarebeamNRimimageagesswOBith(orexactlywhite
thsameespamearameparteamrswetersithouoftthneebueamtroniwithrradoutiationthe.obThjeectnorandmaliDIzedareimageimagesNIwisi:thexactlythe
IDRN−NI=OB−DI(3.4)
Inthiswayrelativemeasurementsareperformedandthecorrespondingprocedureis
callednormalization(see3.5,3.6).
theEvbeenam,ifscainnortillmatoralizedandNRopticsimageandseeCmsCDvderyisapflpat,ear,bectheausenoiseallinthethatinhomogeimagen(oritiesthofe
riunghct.ertaiThentyimuncertainage)sttiyllofshaowsnsucormalizeheffedictsm.ageThatσNIisvicansibbleecinalcthuelatedcornersbyofthethepropFig.agation3.6,
ofuncertainties:
12NR−DI2−1NR−DI2
σNI=OB−DIσN2R+(OB−DI)2σO2B+OB−DI+(OB−DI)σD2I
(3.5)

3.4ReproducibilityanduncertaintyofaNRimage

Figure3.5:OpenbeamNR(left),darkimage(right)

Figure3.6:RawNRofarefrigerator(left),thesameNRafternormalization(right)

38

3.4ReproducibilityanduncertaintyofaNRimage
Thedefinitequalityproofofanexperimentalmethodisitsreproducibility.Therefore
notonlyaNRimagebutalsoanimageofitsuncertaintyisnecessary.Inthefieldof
NRnotmanyworksaboutanalysisofuncertaintieswerefoundandanownevaluation
unprocceertaidurentywasallodewsveloptheecdarinefulorsderearctohguforaransysteteethematicreprerroorsducianbilidty.theThefurctheralcuimprolationvofemthenet
ofthesetup.TheIDLprogrammess_unsich.procalculatesthemeanµNRaccordingto
equation3.6andtheuncertaintyσNRofµNRforoneNRandtheuncertaintyσNRofµNR
foraseriesfromnNRimagesacquiredunderidenticalmeasuringconditionsandwith
identicalacquisitionparameters(equation3.7):
µNR=NRn(3.6)
nn

3.4ReproducibilityanduncertaintyofaNRimage

39

σ1σNR=(NRn−µNR)2σNR=√NR(3.7)
n1n−nFig.3.7showstheresultoftheuncertaintycalculationforaNRseriesofstepwedges.
Therelativeuncertaintyimage(Fig.3.7c)σ=σNR/µNR

Figure3.7:NRofstepwedges:a)meanof100NR,b)uncertainty,c)relativeuncertainty
TheyellowarrowinFig.3.7b)showsacameranoise,whichwasaccidentallyrevealed
checkingtheuncertaintyimage.Theverticallinestructurediffersfromthetypicalneutron
quantumnoiseandmustcomefromthereadoutandtheamplificationofthecamera.
Extractingalineprofile(column788fromFig.3.7)theonedimensionaldatacanbe
displayedtogetherwithitsuncertaintybars(Fig.3.8)andeverypixelfromtheprofilecan
beextractedagain,likepixel(788,405)withagreylevelof18746±3448.Inthatcase
thelargeuncertaintymustcomefromagammaevent.Knowingtheuncertaintiesofthe
NRdataitcanbecomparedandfittedmuchbettertoknowngeometriesortheoretical
dels.mo

Figure3.8:NRprofileofthestepwedgesinFig.3.7
theThesignalinvtoersenoiseoftheratioSrelatiNR.veuAccnceorrdintaingtytoσequisaationpar3.7amthetereSNRwidelymustuseindcinreaseelecwittrhonicsthe:

3.5QuantitativeNR

40

squarerootofthenumberofimagesn.Thatwasconfirmedinaexperiment,where100
identicalframeswhereacquired.Indeed,SNRincreaseswithanincreasingnumberof
frames(seeFig.3.9).InthecaseofFig.3.9theSNRimproveduptotheexcellentvalue
2500.of

Figure3.9:TheSNRversusthenumberofframes
However,itisnotalwayspossibletoimageanobjectseveraltimesinordertocalculate
theuncertainty.Inthatcasetheuncertaintycanbeestimated,ifthetheresponsefunction
ofthedetectorisknown.ThiswasdonefortypicalmeasurementswiththeNRdetectors
atANTARES.Fig.3.10showstherelationbetweenthemeanvalueµofmanypixelsand
theiruncertaintyfordifferentneutronattenuationsanddifferentgreylevelsrespectively.
Anidealsampleforthedeterminationofthedetectorresponseisawedgewithcomplete
absorptionononeside.InFig.3.10anironstepwedgewasinspectedinaseriesof100NR
imagesandtheuncertaintywasplottedversusthepixelmeanformanypixels.Again,a
squarerootlawthatconfirmsthequantumcharacterofthenoisecanberecognized.Now
theuncertaintyofasingleNRimage,acquiredwiththesamemeasurementconditions,
canbepredicted.Knowingthedetectorresponseandthenumberofimagesninthe
seriestheuncertaintyofaNRfortheseparameterscanbeestimatedwiththeformula:
√µ+120.Thesmallestµis6500duetothebackground.Dependingonmanycamera
parameterslikeaccumulationtime,numberofimages,readoutfrequency,intensifiergain
theuncertaintycanbeestimatedinadvance.Inthiswaythelowestvisiblecontrastfora
certainobjectcanbeestimated.

3.5QuantitativeNR
Knowingtheinspectedmaterialanditsattenuationcoefficient,geometryinformationfrom
theNRimagecanbeextracted:
•Distancesandareasprojectedontheconverterplate:EverypixelontheCCD
correspondstoanareaonthescintillator.Distancesbetweentwopixelsinthe
imagecorrespondtotwopointsinthescintillatorandbecauseofaparallelbeam
geometrytoaprojecteddistanceintheobject.Forsuchadistancemeasurement
theexactpositioninginthebeamiscrucial.Thelineorareatomeasureshouldbe

3.5QuantitativeNR

41

Figure3.10:Dete√ctorresponse,theredlinefitsthemostprobableuncertainty,forasingleNR
itisapproximately:µ+120.

perpendiculartothebeam.Thecalibrationofthepixelsizeandspatialresolution
diissdtoronetionswithofathestronobglyjectivabsorbemaingyobalsojectinfluwithencesharthpeedmegesasurofeamedenfitnedandsize.calibrOptiationcal
measurementsarestronglyproposed.

•Totalobjectthicknessdalongbeamdirection:Ifthematerialanditsattenuation
coefficientareknownthethicknessis:

dΣ(z)dz=−ln(II)d=−Σ1ln(II)ifΣisconstant(3.8)
000

Iisthedecreasedintensityaftertheobject,I0istheintensitybeforetheobject.
Bothareproportionaltothegreylevelintheimage.

•FpiorxelaarheomaAonogeneoustheobscinjectilltthatorevandolumethecancorresbeponcalcdinuglatedthicbkynesths.emultiplicationofa

V=−µAln(II0)ifµisconstant
pixelsofareaA
nV...volumeoftheliquidcm3
II(0z)......neneutrutrononinintentensitsityybaftereforeathedistanceobjectxinn2materialcm2s
scm

(3.9)

3.6NRdatavisualization

42

Σ...attenuationcoefficientcm1
A...pixelareacm2
BecauseoftheenergydependenceofΣandtheresultingbeamhardening,Σ=Σ(d)
candependalsoonthethicknessdofthesample.Intheareasofstrongattenuation
theuncertaintyincreasesexponentially.Ifthesignalofthetransmittedneutrons
decreasestobackgroundlevel,thethicknesscouldtheoreticallybeinfinite.
Itismuchmoreconvenienttoperformmeasurementsontomographydatasetsbecause
thegreylevelineveryvoxelcorrespondstothelogarithmoftheneutronattenuationcoef-
ficientandthedistancesarerealdistances.Morecomplexmeasurementsina3Dvolume
aretheidentificationofobjectsbygreylevelsegmentationandvolumemeasurementsby
vandoxelthecoundtiifficng.ultyThtoedisadvcalculateanthetagesuofncerNTtainarteyaroftefthacetsvofxeromlgrtheeyrecleveonlss.tructionalgorithms

3.6NRdatavisualization
Theaimofthevisualizationistodisplaytheevaluateddatainaformthatrelevantdetails
ofanobjectorartefactsoftheimagingsystembecomevisibleimmediately.Inmedicine
thismusthappeninawaythatadiagnosisispossibleafterwards,inNDTitmusthappen
inawaythattheNDTquestionisanswered.TheNDTquestionscanbeveryvarying
andmuchmorequantitative.

3.6.1Softwaretoolsfordataevaluationandvisualization
Thedataevaluationcanoftennotbeseparatedfromthevisualization.Onepartofthe
dataevaluationcanbehandledautomaticallyandthescientistonlycheckstheresults.
Theotherpartmustbetreatedindividuallyandthereforeaappropriate,flexiblesoftware
platformisabsolutelynecessary.Suchasoftwarewasrealizedwithinthescopeofthis
work.Itwasbaptized”Neutroneye”.
ThesoftwareNeutroneyeisamultipurposesoftware,whichwasdevelopedunderthe
IDL(InteractiveDataLanguage)assemblingcode.Suchatoolcanonlybedeveloped
withanobjectorientatedlanguagelikeIDLandrequiressomeprogrammingexperience.
Aconsistentstructureandaconsistentnomenclatureareabsolutelynecessary.In3.11the
graphicaluserinterfaceofNeutoneyeisvisible.Theimage,aregionofinterest(ROI)and
agreylevelhistogramoftheimagearedisplayedinthethreewindows.16bitand32bit
tiffimagescanbeloaded,processedandinspected.Onafirstglancethemostimportant
parametersoftheimagelikename,size,minimalandmaximalvaluescanbefoundon
theinterface.BychoosingaROIandadjustingthetwoscrollbarsforthelowerandthe
uppergreylevelthehuntfordetailscanstart.
Importantimageparameterslikethespatialresolution,thesizeofonepixelandthe
relativenoisecanbecalculatedwiththeESFmethodandaredisplayedinthestatus
barinthelowerregionofthegraphicaluserinterface.Theuserhasthechoicebetween
severalvisualizationmodes:classicalblackandwhiteimages,inverseimagesandfalse
colourpalettes.EveryprocessedimageorROIcanbesavedasabmpfileforuserreports.
Thegreylevelhistogramhasbecomeastandardtoolinimageprocessing.Itdisplays
thenumberofpixelsintheimageasafunctionoftheirgreylevels.Ithelpstoadjust
theslidersfortheminimalandthemaximaldisplayedgreylevelsandtofindareaswith
slightlydifferentgreylevelslikeitisnecessaryfordefects.Importantparameterslikethe

3.6NRdatavisualization

43

Figure3.11:GraphicalUserInterfaceofNeutronEyeshowingaNRofstepwedgesinfalsecolors

Figure3.12:ExtractingdetailsofaNRimageusingNeutroneye

averagegreylevelanditsstandarddeviationoftheROIarecalculatedanddisplayedin
thestatusbar.BymovingthemouseovertheROIthegreylevelofeachpixelisdisplayed
below.InthemenualmostallIDLprogramsdescribedinthisworkcanbefound.Vertical
andhorizontalprofilescanbetakenandsavedina.datfile,imagescanbenormalizedand
meananduncertaintyofaseriesofimagescanbecalculated.Byfittingtheknifeedgeof
astrongabsorberthespatialresolutioncanbecalculatedwithonemouseclick.

3.6NRdatavisualization44
3.6.2Visualizingdifferencesbetweenimages
Oneofthemostcommontasksindataevaluationisthecomparisonoftwoimages.That
caneitherbeacomparisonbetweentwosimilarobjects(likeonewithdefectsandanother
without)oracomparisonbetweentwoimagesafterdifferentevaluationsteps.Whileitis
verydifficulttofindsmalldifferencesbywatchingtwoimagesAandBonebesidetheother,
withadvancedinspectiontoolsdifferencesbetweenimagescanbefoundimmediately.
Considertwoobjects(withµ1,x1andwithµ2,x2),onebehindtheotherlikeinFig.3.13.
Figure3.13:Twosuccessiveabsorbers
Theneutronintensitiesareaccordingtotheexponentialattenuationlaw:
I1=I0eΣ1x1I2=I1eΣ2z2=I0eΣ1z1eΣ2z2(3.10)
Threemethodscouldbeusedtodisplaytheadditionalattenuationofthesecondobject:
•Calculatingoftheratiobetweenimages:
I2=eΣ2z2(3.11)
I1inThisrespmeectthotod(imageA.normalise.proTheratio)makofeBs/AsensshoeifwsthimageeaddBhitionasaladditiattenonaluatiatontenofuB.ationIf
isA=Busedtheintherationis1ormalizationandgreyopleevrelation.intheInrethesultcingaseofimagemaximalDisatt65535.enuTationhisinmeBththode
differenceD=0.Theuncertaintyfollowsfromequation3.5.
bFiguothresho3.14:wingtha)eNRattenuatimageionofofantheemptoilyonoilly)pump,b)NRimageofafilledoilpump,c)ratioof
•Calculatingthedifferencebetweenimages:
I2−I1=I0eΣ1z1eΣ2z2−I0eΣ1z1=I1(eΣ2z2−1)≈I1∙Σ2z2(3.12)

3.6NRdatavisualization

45

Thismethod(diff.pro)makessenseforimageswithslightlydifferentattenuation
intheobject,whereitisnotclearwhetherimageAorimageBshowsahigher
attenuation.Fortwo16bitimagesAandBtheprogramdiff.procalculatesthe
differenceimageD=(A-B)/2+32768.AllpixelsinAandBwiththesamegreylevels
endupatthegreylevel32768inD,tothedarkerpixelsinAahigherandtothe
brighterpixelsinAarelowergreylevelinimageDisassigned.Withanoptimized
falsecolortabletheattenuationdifferencescanbedisplayedinmuchmoredetail.
Buttheinterpretationisnottrivial,becauseinthenoiseAandBarestillvisible.
InFig.3.15twoNRofatrackedpistonweresubtracted.Theoilmovementsare
ed.izvisual

Figure3.15:a)NRofthepistonatt=0ms,b)trackedNRofthepistonatt=20ms,c)difference
imageofboth,d)differenceimageinfalsecolors

•Composingamovie:Observingaseriesofconsecutiveimages-amovie-oureyes
immediatelycandetectachangingcontrast(batchview_tif.pro).Fig.3.25shows
thatoureyeismostsensitivetofrequenciesaround20Hz.Thecompositionofa
filmissuitedbestforalargeramountofsimilarimages.Someexamplesarevisible
CD.on

3.6NRdatavisualization

46

3.6.3Fusingimages
EspeciallyinNDTitisoftennecessarytocombinedifferentkindsofdata.Thecombination
ofopticaldigicampictures,NRimages,X-rayimagesordifferenceimagesshowsmuchmore
detailsthantheindividualimagesalone.Theideaistosetoneimagesemitransparentand
tooverlaporfuseitwithanotherimage(img_align.pro).Iftheimagesarecodedwith
differentcolorstheobservercanreadthemmoreeasily.Arealchallengeistheregistration
problem:theadaptationofresolution,positionandtiltingoftheimages,sothattheyfit
tooneanother.TheimageFig.3.17isafusionoftheimagesinFig.3.16andmakesthe
oilinsidethecombustionenginevisible.

Figure3.16:a)NRimageofthepiston,b)differenceimageoftwoNRimagesshowingtheoilin
falsecolors

Figure3.17:Fusionofbothimages

3.7Limitationsbythehumaneye

47

3.7Limitationsbythehumaneye
Itmayseemstrangetomentionthelimitationsofthehumaneyeinthisplace.Butit
makessensebecauseinnondestructivetestingthemostfrequentlyaskedquestionis:Can
Iseethatdetailornot?Notonlytheimagequality(spatialresolution,timeresolution,
noise,contrast)butalsothewayitisvisualizedandtheviewermustbeconsideredin
ordertofindadefectorananomaly.Thus,itbecomesveryrelevanttoknowthelimits
ofthehumanvisionforwhomwhostudiesthemeasureddatainformofanimage.Inthe
followingthehumaneyeanditphysicalpropertiesaredescribedbasedon[Kol05,20].
Thesensitivedetectorareaofthehumaneyeistheretina(Fig.3.18),madeupoftwo
typesofphotoreceptors:therodsandthecones.Therodshaveahighsensitivityunder
lowambientlightandnightvision,alsocalledscotopicvision,whiletheconeshavea
higherspectralsensitivity.

Figure3.18:Humaneye(left),retina(right)
Theconcentrationofthetworeceptortypesvarieswitheccentricity(anglefromthe
fovea)andshowsaremarkablyhighconcentrationofconesintheFovea,thecenterofthe
retina.Soeventhoughthetotalfieldofviewforahumaneyeisabout150◦,thenormal
fieldofviewis30◦andfieldofviewwiththebestspatialresolutionisabout1◦(Fig.3.19.
ForthequantificationofthevisualacuitytheContrastTransferFunctionCTFwas
measuredwithasinewavepatternliketheonebelow.TheCTFistheresolutionfunc-
tionofthehumaneyeanaloguetotheMTFforanimagingsystem.Fig.3.20shows
themeasuredCTFofthehumaneyeindependenceofthespatialfrequencyforseveral
luminances(unit:trolands).Atthehighestluminanceahumaneyeisabletoresolve
frequenciesof10cycles/degreeatbest.Ifthecontrastbecomestoosmallthesinegrating
cannotberesolvedanymore.Dependingonthedistancebetweenimageandviewerthe
spatialfrequency,whichisresolvedbest,changes.Atlowlightlevelsthemaximumshifts
towardslowerspatialfrequencies.
Anadditionalpropertyofthehumaneyemustbementioned:Thehumaneyeisable
tosumoveracertainarea.Thiskindofpreprocessinghappensinthenetworkoftherods
ontheretinaandvarieswitheccentricity.Thestimulusofadetailisproportionaltothe
productofluminanceandthestimulatedareaontheretina.Lookingforadefect,thesize

3.7Limitationsbythehumaneye

Figure3.19:Rodandconedensitiesalongthehorizontalmeridian

48

Figure3.20:ImageforthetestofCTF(left),CTFofthehumaneyedifferentluminances(right)
[Ko20].l05,

vofaryithengcdefeconttrasmtatteanrdssasize.weIllfasthetheprocdonucttrast.ofcFonig.trast3.21andshowareasaisteslartgepatternenoughwithwecirarecleabsleof
torecognizeit.
limitedForbthyethimeageMFTinsofpeourctionimagingthatmsysteeansmthonatthethedoneetaihanlswd,eanaredonablethetoothreercognihandzebarey
tthweopCTFossiblofetheMTFshuofmananeyeimageof.theTihenspecareatorM.TInFAFig.inb3.et22weewnebseeothtwcourvpesossidesblecribCTFsewhianchd
detailTheschanumbeandeeytecetisedabanledtorecdeteognctized10inorandersinspofectmagnion.itudeofluminanceofvisiblelight
br(Δλigh=tne400ss-lev800elscnm)ancbeondsiderinistingugtished,hewhadaptatichhionasudindreerctlocwonslighequtenlevceesl.forButtheonlviysuaboutalization200
oftodolsata.orfalseDisplacoloryingdatapaletteswith.Tahedpynamicerceptionrangeofbrlargerightnthessanb8ybittherehquuirmeansceyonetrgoasestlvariogarithation-
mical(Fig.3.23)analoguetotheperceptionofsoundbythehumanear.Thehumanear
secannsitimeasvituyrefor13orhighedersrluofminansoundciesnteshnsiouldtyofbecsounonsdiwdereavedsd(Δfur=ing20anHzinsp-20ection,kHz).forTheibenstancetter

3.7Limitationsbythehumaneye

Figure3.21:Testpattern(left),Spatialsummationofthehumaneye(right)[Kol05,20]

Figure3.22:Detectabledetailsinanimage

49

byinvertingimagesorbyusingnonlinearcontrastvariationmethods(likegammacor-
rection).InthemedicalfieldthehumanbodyisinspectedintheformofinvertedX-ray
radiographies(negativeimages)becausethebonesgivehighercontrast.
Thehumaneyeandthehumanbrainarecapableofmuchmorecomplextasks.They
areforinstanceoptimizedforedgedetectionandobjectrecognition.Thisleadsto”arte-
facts”inhumanvisionlikecontrastenhancementnearbrightnessvariationsandeven
completionofpartiallyhiddenobjectandstructures.InFig.3.24werecognizestronger
contrastsneartheedgesaandanillusionarywhitesquare.
Sincethereisafiniteamountoftimerequiredtocollectandprocessinformation,there
arelimitationstotheresponsivenessofourvisualsystemtoratesofchange,Fig.3.25.
Forthedetectionofaflashoflighttheintegrationtimeisto100msforrodsand10to15
msforconesandtheeyeappearstobemostsensitivetoafrequencyof15to20Hz.For
lowlightlevelstheintegrationtimeisminimalat100msandbestatfrequenciesof5Hz.

3.7Limitationsbythehumaneye

Figure3.23:Logaritmicalperceptionofthehumaneye[Kol05,20]

Figure3.24:Machbandsand(left),illusorycontours(right)[Kol05,20]

Figure3.25:Temporalcontrastsensitivityofthehumaneye[Kol05,20]

50

4erChapt

tsneemMeasur

Withinthisworktwoexperimentalmethodsweredeveloped:stroboscopicNRandenergy
selectivetimeofflightNR.Bothtechniquesaredescribedinthischapterandexperimental
resultsarepresented.InthecaseofstroboscopicNRcombustionenginesandaninjection
nozzlewereinvestigatedatthePaulScherrerInstitut,attheInstituteLaueLangevinand
atthenewneutronsourceFRM-IIwithdifferentdetectorssystems.Thepresentedenergy
selectivetimeofflightNRexperimentwasaproofofprincipleexperimentandwascarried
outatthepulsedspallationsourceISIS.
Thefinalsectionofthischapterisdedicatedtothequantitativedataevaluation.AtKFKI
thecoolingliquidofarefrigeratorwasinspectedandintheNRseriesthevolumeofthe
coolingliquidinthebufferofthefreezerboxwascalculatedforeveryframeoftheseries.

4.1StroboscopicNR:Combustionengines

StroboscopicneutronimagingisanewtechniquedevelopedattheTechnischeUniversit¨at
Mu¨nchenwithinthiswork.Athighfluxneutronsourcestypicalintegrationtimesfora
single16bitNRimagearearound100to1000ms.Thetimeresolutionfordynamicneu-
tronradiography-orneutronradioscopy-isgivenbytheneutronintegrationtimeplusthe
readouttimeoftheCCD.ThatleadstypicallytoatimeresolutionΔtofafewseconds.
Forcyclicprocessestheperiodicityallowsthesynchronizationofdetectorandprocess.
IdenticalNRimagesfromthesametimewindowoftheprocesscanbeacquiredstrobo-
scopically.Bysummationofnimageseitheronchiporafterreadoutbysoftwareatime
resolutiondowntothemsandsubmsrangeisreached.Thequalityoftheimageisthe
sameasforaNRimagewithantimeslongerintegrationtime.Byshiftingthephaseof
thetriggersignalthetimewindowcanbeshiftedandfinallyaNRmoviecanbecomposed.
Anidealobjectofinvestigationforstroboscopicneutronimagingarecombustionengines.
Combustionenginesareanactivefieldofresearch.Theaimsaretheiroptimizationand
improvementthroughmorepower,longerlifetimes,lowerweight,lowerproductioncosts,
lessfuelconsumption,lessexhaustgases.Therearestillunansweredquestionsatmstime
scalesliketheoildistributions,thefuelinjectionorthemovementofthepistonrings,
whichcanhardlybesimulated.Therobust,metallicmotorblockmakesitverydiffi-
culttoobservethephysicalprocessesinsidewithclassicalNDTmethods.Spotsample
measurementsaretheonlywaytogetaccesstophysicalquantitieslikeoilpressureandoil
density.Opticalwindowsopenthedoortomanyopticalmethodslike”Schlieren”-images
andshadowgraphtechniques,whichdeliverinformationaboutfuelconcentration.Light
scatteringmethodsasLaserDopplerAnemometryandPhaseDopplerAnemometrycan
determinepositionandvelocityofdroplets.Spectroscopicmethodsgiveaccesstoposition

4.1StroboscopicNR:Combustionengines

52

sensitivechemicalcompositionandtemperatureinformation.Formoredetailedinforma-
tionsee[Sfb05,25].However,withthesemethodsthesystemisdisturbedandthereal
conditionsliketemperature,pressureandchemicalenvironmentcanneverbeachieved.
BeforestroboscopicNRisestablishedasastandardinspectionmethodfeasibilitystudies
mustbedoneandpioneerexperimentsmustprovetheopportunitiesofthemethod.NDT
questions,whichcanbefollowedupwithneutronsare:
-Whatisthedistributionofthelubricationliquidinsidearunningcombustionengine
(pistoncooling,pistonlubrication,backflowoftheoilintheoilpan,oiltransporttothe
ings)?areb-Howdoesthefuelinjectionoccurinsidearealsystemunderrealconditions?
Neutronimaginggivesusthepossibilitytoanswerthesequestionsbecauseoftworeasons:
1.thelargepenetrationdepthofneutronsformanymetals,especiallyaluminumoften
usedforthemotorblock
2.thehightotalneutroncrosssectionofhydrogen,andhydrogenoussubstanceslike
fuelandoilinthecaseofacombustionengine.

4.1.1FirstNRofarunningcombustionengine
Afterthemethodofstroboscopicimagingwasproventowork[Bru01,31],intheyear
2002forthefirsttimeaselfrunningcombustionenginewasinvestigatedbystroboscopic
neutronimaging.Forthefourstroke480WmodelaircraftengineofthecompanyOS
(Fig.4.1)ateststandwasprepared.Theenginerunswithnitrofuelat4800rpm.

Figure4.1:Photographofthemodelaircraftcombustionengine
Thepreparationofthetestsetupforstroboscopicimagingistimeconsuming.A
reliableandstabletriggeringsignalforthesynchronizationofmotorandintensified
cameraisneeded.NormallyaninductivesensorgeneratesaTTLsignaleverytimea
pieceofmetalpassesatamaximaldistanceof2mm.Asmallscrew,mountedonthe
rotor,gaveastableelectricalsignalbetween0Vto5Vforthecameratriggering.Inthe
firstexperimentsanexternaldelaygeneratorwasused,laterthedelaywasgeneratedby
thecamerasoftware.Itisnottrivialtogetthiskindofenginerunningandanelectric
starterisproposedforfutureexperiments.Althoughthepoweroftheengineisnothuge,

4.1StroboscopicNR:Combustionengines

53

4.1,FiguΔtre=4.2:250µSisnglewithfram4000eofaccaumuneutrlationons,radΔx=iograp0.75hymmom,vieσof=1.the7%modelaircraftengineinFig.

theairmovementsare2m3/s.Thefastmovingrotoroftheengineisdangerousand
mustbehandledwithcare.Theexhaustgasesshouldbefilteredformtheunburnedfuel
inordertopreventpollutionoftheexperiment.

2002.TheAtarunthninergmealnginefluxwofas7.5i∙m106agedn/cmat2psositandion2L/Dof=the350atNEUTRAypicalacfacilitcumy[ulationPSI05,time27]atin
NEUTRAis100ms.Withthestroboscopictechniqueamuchbettertimeresolutionis
possibleandanexposuretimeof250µswaschosen.4000imagesofthesamephasewere
summedupintotal,400wereintegratedonchipand10individualimageswereaddedby
byexpsofosurtwearetimeaftereacreh.adThout.edeteOnectorcyclewasofa12.5cooledmsawnasdsuMCPbdiviindeditensifiednto50CfrCDamescamerwitha250PI-Maxµs
6isinshocomwnbinatiinFonig.with4.2:a300oneµsinmglethicfkrameLiFZnoutSofconthveerteneutrrfonromradtheciograpomphanyymoAST.vieT[pphet1,result42].
toThetheenrotorvironmpenlane,tofthethesglowmallplugenginonetopconsistsofthofeenthegine,triggertheseexhnsoraustmpouipne,tedapesmallrpendpicotuforlar
theexhaustgasesandtheblackstronglyattenuatingfuelsupplytube.Inadditiontothe
pistonandthevalvemovementssmallchangesinthelubricationdistributionsarevisible
3).4.e(serectTheangleiminagethewasimagecharacteandrtheizedsbpatyialmeresasuroluingtionthewreaslesativtiemnoisatedeσviainantheESFopenmbeaseuamreminenthet.
Afternormalizationtherelativenoiseσwasmeasuredtobe1.7%andthespatialresolution

4.1StroboscopicNR:Combustionengines

Figure4.3:Dynamicneutronradiographyframesofthemovie,linktoCD

54

Δxwas0.75mmduetoblurringcausedbythebeamdivergency.Thefieldofviewwas
15x14.7cm2(1058x1038pixels).
ThisexperimentdemonstratedthatstroboscopicNIwithatimeresolutionof100µs
withacceptablequalityisfeasible.Oilmovementsinsidethemodelaircraftenginewere
observedatthistimescale.Duringtheexperimentsitwasrealizedthatthetestsetups
forthiskindofmeasurementsareverycomplexandrequirealongpreparationtime.

4.1.2Carcombustionengine
InacollaborationwiththeILL,theUniversityofHeidelbergandPSIafour-pistonBMW
thcomebuhighstflionuxengintestebewamasinsplineecteH9datwiILLth[thILeL05,strob28]oscinopicorndereutotronradicompareographdiffeyretecnthnfaciquileitiesat
anddetectors.TheNeutrographfacilityisdescribedin[Fer05,29].Theobjectisanorig-
inalBMWfourcylindercarcombustionengine(Fig.4.4)ofthetypeNG4withamotor
blockofaluminium.Itisdrivenbya2kWelectromotormountedonaverticaltranslation
stagebelow.Thesparkplugswereremovedtoreducethedrag.Inthiswayawatercool-
inandgclecircuitanedis.notThenecvibesrationsarysandofsucnohexhenginaustesgasemasybareeprocritical,ducedso,thatwhichtheymustshoubeldconbterolledfixed
onappropriatebearingsinordertopreventmovingorevenbreakingoftheengine.The
identicalsystemasinthemeasurementbeforewasused:aMCPintensifiedCCDcamera
PiMax[Rop04,12]withapeltiercooledchipand16bitdigitalization.Theusabledynamic
rangeislimitedbytheinherentnoiseoftheintensifier.Thefullcycle(duration120ms)
ofrotatthisions:fours150trokinediveniduginaleirunmniagesngwatith1000aexrppmosurweastspimliteofinto200120µsindeachividwualerefracamcumesovulatere2d
onchip.Inthiswaythemeasuringtimeforthefullrunwasintheorderof18minonly.
InpisFitong.ro4.5ds,abpoveistonthepfiinsrstandframepistonofthreings.recordedOfspemocialvieinisterevisiblsteis,theshowingvvisualizationalves,ofpistonthse,
pistoncoolingviaanoiljetdirectedthepistonbottom.Sincethepistonsareconnected
totheenginebodyonlyviathepistonringswithverylowheatdissipation,acontinuous
oiljetisdirectedfrombelowtothepistonbottoms,loweringthepistontemperatureby
morethan200◦C.Inthemovie[ppt1,43]thedynamicoildistributionisclearlyvisible.
Aroundtheupperturningpointofthepiston,theoilmovementatthepistonbottom
duetoinertiawasobserved.Thefieldofviewwas21.5x21.0cm2.Theobservationarea
couldbyvariedbydisplacementofthefullset-up.Theimagecharacterizationyielded
aspatialresolutionof1.8mmandarelativenoiseof2%.Thespatialresolutionisso

4.1StroboscopicNR:Combustionengines

55

Figure4.4:PhotographoftheBMWcombustionengine,theredrectangleistheFOVforthe
NR

obmojectderateandduethetodetethector,beambutdivtheergennoiscyeininctheombinmostationhomwithogeneousthelargearea,thedistancebeambetcweneenter,thise
vwaseryloused:wconasiNRderinfgramethewloaswoivnetergrlappationedwithtime.aIdniffeFig.renc4.5etheimagenew(invisualfalseizaticolorons)tecbehntwiqueene
thatframeandasecondframeofdifferentpositionintime.Thegreenandthebluecolor
meanlessattenuation,theyellowandredcolormeanmoreattenuationinthesecond
frame.Beforethesubtractionoperationthepistonwastrackedalongitsmovementinthe
diffcylinerender.tattenHoweuativer,ondiresffulterenceiniamdivagesersealwanoiseyslshoevwelandartefactsdiffe.renRtegionoffssetsofinintheterestgreywithleveal
resultsinafalsecoloroffsetofthedifferenceimage.Thepistonisintheupperturning
pointandtheoil(yellowspot)ismovingfromrighttoleftbecauseoftheinertiaandthe
specialformofthepistonbottom.Thegreenstripesshowhowtheoildisappearsduring
theup-movementofthepiston.
ThisexperimentconfirmedthatthestroboscopicNRmethodhasthepotentialforinspec-
tionofthelubricationcircuitinlargecarcombustionenginesmadeofaluminium.

4.1StroboscopicNR:Combustionengines

56

withFigure1504.5:accumNRufrlationams,eΔofxa=1.8dynmm,amicσNR=2%,moFvOieV:of21.5thexrunn21.0ingcm2engi;RneOI(abofovthe):epΔistont=(b200eloµw)s
overlappedwithadifferenceimageoftwoNRsatdifferentpositionintimeinfalsecolors:theoil
insidetheengineisvisible

4.1StroboscopicNR:Combustionengines

57

4.1.3Injectionnozzle
Afterthefirstmeasurementswithcombustionengines,itwasrealizedthatitwouldbe
verydifficulttoseethefuelinarunningengine.So,thedecisionwastakentotrywitha
commercialfuelinjectorasanextstep.Apartfromfuelinjectionsystemshighpressure
spraysareanessentialtechnologywithmanyotherapplicationsasthermalandplasma
spraycoatingandliquid-jetmachining.Liquidspraysaredifficulttoinvestigatebecause
ofthefasttimescaleandtheslightdensitychangesinrespecttoair.Usingflashlights
shadow-andreflectionimagescanbeobtainedbutnorealdensityprofileacrossthe
liquidcloudcanbemeasured.Recentlyradiographicstudiesofspayswereperformed
withX-raysandsynchrotronradiation[Mac02,30].Intheautomobileindustrycommon
raildieselinjectorsarestateoftheartandwerepurchased.Anappropriateteststand
(Fig.4.6)wasprovidedbythecompanyBosch.

Figure4.6:Dieselinjectionsetupforneutronradioscopy
Theinjectorismountedononeofthefourpositionsofthepressurerig.Theoil
pressurecanbeadjustedbetween0and1200barbythecurrentofthehighpressure
pump.Injectionnozzlesinthecarhaveseveralholeswhichinjectthefuelcloudsvery
homogenously.Duringtheexperimentsacommonraildieselinjectionnozzlewithone
singlecentralholeof0.2mmdiameterandapressureupto1000barwasinspected.Safety
precautionsweretakentoprohibitcontactwiththeoiljet.Itcanpenetratetheskinor
evencutafingerandpoisonthehumanbloodcircuit.Theoil,aspecialnonflammableoil
withaviscositysimilartoDieselfuelisusedasfuelsubstituteandarrivesfromthesmall
tankviaaprepumpandahighpressurepumptothehighpressurerig.Afterlonger
operationtimesofthenozzletheproducedoilfogcandisturbthemeasurement.The
nozzleneedsavoltageofabout100Vinordertoopentheelectronicvalveandacurrent
of2Atokeepsitopen.Aspecialpowerelectronicswithahighvoltagecapandapower
MOSFETwasusedforthisspecialtask.TheNRdetectorandthenozzlearetriggered
withthesameelectronicTTLpulse.Thepulsefrequencymustbelowerthan20Hzto
keepthepressureintherigconstant.
AtthebeamlineH9atILLtheveryhighfluxof3.2∙109n/cm2sgivesoptimalconditionsfor
shortexposuretimes.ThedetectorsystematILLusesaSensicam,anInterlineTransfer
CCDcameraofthecompanyPCOwithafastshutteroption.Theinjectorwastriggered

4.1StroboscopicNR:Combustionengines

58

withabout20Hzandkeptopenfor1ms.TheCCDcameraaccumulated5000images
synchronoustotheinjectionprocess.
In2003thefastinjectionprocessofahighpressurenozzlewasobservedforthefirsttime
withneutrons(seeFig.4.7).
Theinjectioncloudwasobservedin5stepsof100µseach.Thelackofasharpedge

µs,Figu900reµs,4.7:ΔtFi=rst100neµustronwithrad5000iograpacchiuesmuofthlationes,injecΔx=tion1.0promm,cess,σ=t=5000.28%µs,600µs,700µs,800
intheimageallowsonlyaroughestimationofthespatialresolution.Fromtheprofileof
thecylindricalnozzlethespatialresolutionisestimatedtobe1mm,atatimeresolution
Δtof100µs.ThisagreeswiththePSFestimationbasedontheL/Dratioof150of
theneutronbeamandadistanceobjecttodetectorofabout150mm.Someproblems
wereencountered:Theinjectionchamberfilledupwithanoilfogandonthealuminum
scwindinowtillatorsoildropdeterioratisformed,on,thethesnoisecintlevillatorelweasfficievenrycyhighwasandnothashomogealinneesoutrsbuctureecauseandofththee
rightsideoftheNRimagesisbrighterforanunknownreason.Themeasurementwas
continuedwiththefollowingimprovements:
-Dataimagesandopenbeamimagesweretakenalternatinglybysimplytriggeringthe
camerawithtwicethefrequencyofthenozzle.
-Thedarkcurrentcorrectionwasoptimized.
-Moreimageswereaccumulated.
-Thedataevaluationandvisualizationwereoptimizedandautomatized.
ThelastexperimentwascontinuedatILLbyMartinEngelhardt[Eng04,33]andAndreas
Hillenbach.ThebestresultsarepresentedinFig.4.8.Highpressureinjectionsoffuelwere
investigatedwithneutronsforthefirsttime.Thebestresultshaveaspatialresolutionof
1mmandarelativenoiseof0.2%.Uptonowtheinjectioninarunningenginewasnot
visualized.Itmaybepossibletoseethefuelinnearfuturebutthespatialresolutionwill
notbehighenoughtooptimizetheinjectionprocesswithNR.
theThenozzlevisuopeninalizationgisdofelathyeedinbyjec5ti00onµcsiloundreswpasectatorealthechrisingallenge.edgeThofeptheostriitionggeriinngtimesignalof
andtheattenuationofthefuelcloudisbelow0.25%.Thus,itisanidealsampleforthe
checkofNRimagequalityandinfactalotofmethodicalimprovementsresultedfrom
theseexperiments.

4.1StroboscopicNR:Combustionengines

59

Figure4.8:NRimageoftheinjectionnozzleinaction(left):Δt=100µswith12000accumu-
lations,Δx=1.0mm,σ=0.44%,FOV:62x83mm2;
NRseriesoftheinjectionprocessat350bar(right):t=500µs,600µs,700µs,800µs,Δt=100
µswith12000accumulations,Δx=1.0mm,σ=0.2%,FOVofeachinjection:10x50mm2(640x
127pixel),bluenoattenuation,yellow0.4%attenuation,red0.8%attenuation

4.1StroboscopicNR:Combustionengines

60

4.1.4FirstneutronradioscopyofancombustionengineatFRM-II
Finally,inDecember2004inthefirstreactorcycleoftheFRM-IIitwaspossibleto
ptheerformANTtheARESfirsftacistroblity.oscopicNRexperimentofaDieselcurrentgenerator(Fig.4.9)at

Figure4.9:Directinjectiondieselgeneratortestsetupfordynamicneutronradiography(above),
NRimagecomposedfrom16singleNRimages(below)

Beforethestartoftheexperimentthegeneratorwasmodifiedinthefollowingway:On
theexhaustpipeafilter(yellowcylinder)wasmounted,whichisnormallyusedfortrucks
drivingintoclosedroomsorbuildings.Thefilteredexhaustgaseswereconductedinthe

4.1StroboscopicNR:Combustionengines

61

exhaustairpipesoftheexperimentalhalloftheFRM-II.Furthermoremanypartsofthe
remogeneratorvedsucashththeatexhnoaustunestubsene,tialthepartstank,aretheexsptarosedtingtomtheotorhighwereneudistronplacfluedx.orFcorsomplecuerittelyy
threaseonexpseronimlyenatwlimitedascarefamounullytwofatcfuelhed.wasBeforeleftinthethdetanynamickandradwithiograpahysurrvuneilthelancecamgeneratorera
wresaspecsctianvenedfiebldyoftrviewanslatinofg144itxon144themsmamp2leandmanip1024ulxator1024andpixe16lswsinergleecomNRposeimagesdtoawithbiga
onewithasizeof658x658mm2and4682x4682pixels(Fig.4.9).
Fparorenthecyofstrobtheoscopimotorcmblock.easuremeThentfouthressttarrokteerdirecmotortinjweasctionremodviesedeltogengetearatorbetterunrstrwithans-
3000rpm,50Hz.Thatis20msforeachrevolutionand40msforeachcycle.Asadetector
wtheere1ICCDmsaccamcumeraulatiwithontim1024exand10241500ponixelschwipasaccusued.mulTheationsspettinerfgsrameofcforhoiceeachforoftheimaging40
framcollimatores.Becoucauseldbofeuradisedationwithsaafesmtyalrelerasonsfluxofat2.the5∙107timen/cmof2thissfionlrsty.eTxpheerimentotaltmtheasesurinmallg
timewasabout3hours.TheresultisvisibleinFig.4.10.

Figure4.10:OneframeoftheNRmovie[ppt2,44]ofthegeneratormeasuredatFRM-IIforthe
firsttime:Δt=1mswith1500accumulations,Δx=1.0mm,σ=2.5%,FOV:144x144mm2
pixels)1024x(1024

4.2EnergyselectivetimeofflightNR

62

ANRmoviewascomposed[ppt2,44].Thespatialresolutionis1mmandtherelative
noiseis2.5%.Aimsforthefuturearealongeraccumulationtimewiththebigcollimator
andtheoptimizationofthespatialresolution.Attachinganelectricalloadlikeaheaterto
thegeneratoritshouldbepossibletovisualizethefuelinjectionbecauseoftheincreased
amountofinjectedfuel.
TheANTARESfacilityisperfectlysuitedfordynamicneutronimagingbecauseofthehigh
neutronfluxandthehighspatialresolution.Duetothecoldspectrumthefacilityallows
theinspectionofthinnerandlessattenuatingobjectsincomparisontothermalfacilities
do.TheexperimentalsetupforstroboscopicNRaswellastheevaluationsoftwarewas
successfullytestedforthefirsttimewithsatisfactoryresult.

4.2EnergyselectivetimeofflightNR
Byimaginganobjectatdifferentneutronenergiesadditionalinformationcanbeobtained.
Thecontrastchangesreflecttheenergydependentcrosssectionofamaterialandallowsin
thebestcasematerialdiscrimination.Practicallytherearefivepossibilitiesforaneutron
spectrumvariation:
•Differentmoderatorsorvariationofthemoderatortemperature:Aneutronspectrum
ismostlyaMaxwelldistributionwiththetemperatureofthemoderatormediumas
themainparameter.
•Monochromatorsarequasiperfectcrystals,whichreflectasmallpart(about1%)of
theneutronfollowingtheBraggequation.
•Filtersattenuatepartsoftheneutronspectrumaccordingtothecrosssectionsof
theconstituentnuclei.Aclearchangeinthespectrumisonlyreachedwithfilters
workinginaresonanceregime.Thethickerafilteristhebetteristheenergy
discrimination,butthemoreneutronsareabsorbedandtheloweristhetransmitted
flux.Filtersarefavoredespeciallyforfastspectra,whereothermethodsaredifficult
.realizeto•Avelocityselectorisshigh-speedturbine,whichistransparentforthoseneutrons
whichmanagetopassbetweenthetwistedlamellaeinsertedintherotorinatime
intervaldefinedbytherotationspeedoftheselector.Thus,neutronsofaspecial
velocityorenergyareselected.Modernvelocityselectorsspinningwithupto30000
rpm,reachanenergyresolutionofΔE/E≈30%.
•Atpulsedneutronsourcesatacertaindistancefromthesourceatimeofflight(TOF)
spectrumcanbemeasuredduetothefinitevelocityoftheneutrons.Immediately
afterthegenerationoftheneutronpulsethegammaparticlesandthefastneutrons
arriveatthesampleposition,laterthethermalneutronsandfinallyattheendof
thecyclebeforetheconsecutivepulsethecoldneutronsarrive.Measuringthetime
betweengenerationandarrivaloftheneutrons-thetimeofflight-andtheflight
paththeneutronvelocitycanbecalculated.Fastdetectorsallowsanenergyselective
neutronmeasurements.Atmosttimeofflightinstruments3Hetubesareusedas
.stordetecThelastmethodcanbeusedforneutronimaging.3Hetubesarenotidealforimaging
becauseoftheirbadspatialresolution,butthedetectorforstroboscopicNRwiththe
ICCDcamerais.IthastheexcellentspatialresolutionofalargeareaNRdetectorand
anexcellentenergyresolutionduetothegatingpropertiesoftheintensifier.

4.2EnergyselectivetimeofflightNR

63

Bysynchronizingthisdetectorwiththeneutronpulsesandshiftingthephaseofthe
triggeringsignalenergyselectionbecomespossible.Manyframescanbeaccumulatedon
thechipbeforethereadout.Thetimebetweenneutronpulsegenerationandtheshutter
openingofthecameradefinesthestartingenergy,theopeningtimeoftheintensified
cameradefinestheenergywindows(andtheenergyresolution).Theminimalenergy
resolutionΔEofthedetectorislimitedbyscintillatordecaytimeΔtsci:

ΔE=2sE2mEnΔtsciΔE≈3.0∙104∙Δstsci∙E[eV]23(4.1)
ForΔtsci=80µsands=15mtheenergyresolutionisΔE≈0.16∙E[eV]23.The
effiregionciencyfolloofwsscithnetill1/ator√Eisbeehnaevrgyiordep(seeFendig.ent1.7).too,Athenormcrosalsizseationctionforofal6lLiinenergiestheacthecounrmalts
his.tforthTheeireabinelitrgyytodepdistinendenguishtcbrossetwseceentiontwsoandmaterialstheavisailabmainlleyneutrondeterminedflux.byRestheondanceiffesrenceinthine
highenergyregime(MeV),butalsoBraggcutoffsatcoldneutronenergies(inthemeV
regime)aremostlyusedforenergyselectiveNR.
Theproofofprincipleexperimentforthismethodwascarriedoutatthespallationsource
ISIS,attheRutherfordAppletonLaboratory,nearOxfordinUK(seeFig.4.11).The
repetitionrateis50Hzandthedurationofthepulseis20ms.Themeasuringposition
thatethliqeuidHIPmReethanxpeerimenmodteatratortheatETNGIN=100beK.amTlihneewspasatectrumadishastaancemofaxiabmuoumtn15earm2.fr0omA˚
(seeFig.4.11)andthetotalfluxisabout106n/cm2s.

fliFigughtrein4.tensit11:yspecInstrutrumment(righationt)ovplottereviewdvatersusISIStime(left)of,fliRughtht.erfordAppletonLaboratory,timeof
AdetectorwithanMCPintensifiedCCDcameraPIMax[Rop04,12]wasapplied.The
triggerinputofthecamerawasconnectedtothesynchronizationsignalofthespallation
source.Theselectionoftheenergybandandthenumberofaccumulationswasdonevia
camerasoftware.Thebeamdiameterwasabout8cmandvariedwiththeenergy.The
divergency,determinedbyneutronguide,wasnotnegligibleandthedistancebetween
objectanddetectorwasreducedtotheminimuminordertoimprovethespatialresolution.
Fromequation1.1andv=s/t(visthespeed,sistheflightpathandtisthetimeofflight
oftheneutron)followsinourcase:λA˚=t[ms]∙0.24.Themaximalresolutiondueto
thescintillatorwas0.02˚A.AstandardNDgneutronscintillatorofthecompanyASTwas

4.2EnergyselectivetimeofflightNR

64

used.2TheTheifirsdeatswampasletowdeasaterminstroneglythespatiabsorbialngreGsolutiadolionniuofmthefoilseoft5upx5ancmdtowithverifaystarthepatternenergy.
dependentcontrast.NRatthreedifferentenergiesareshowninFig.4.12.

Figure4.12:NRimagesofaGd-Foilwithastarpatternatthreedifferentenergies:at0.48˚A,at
2.64A˚and2at4.56˚A:Δt=0.1mswith20000accumulations,Δx=0.5mm,σ=10-30%,FOV:161
x166mm(1300x1340pixels)
SincetheneutronfluxisnotthesameforthedifferentenergiestheNRimagesmust
benormalized.ThereasonforthedifferentcontrastofthefoilinthethreeNRinFig.
4.12istheenergydependentneutroncrosssectionofthematerialofGadolinium.Athigh
energiesthefoilisnotvisibleatall,whilewithlowerenergiestheattenuationincreasesand
thefoilappears.Theneutronattenuationofthefoilforallmeasuredenergiesisdisplayed
in4.13.ig.F

Figure4.13:NeutronattenuationofaGdfoilatvaryingenergy
AccordingtothecrosssectionofGadolinium(Fig.1.7inchapter1)anexponential
increaseupto1.2A˚andaslowerincreaseintheattenuationatlongerwavelengthsis
expected.Theincreaseinthethermalregionisclearlyvisible,butforcoldneutronsthe
attenuationdecreasesagain.Thatcanbeexplainedbythelowcoldneutronfluxandthe
el.levnoiseghhi

4.2EnergyselectivetimeofflightNR

65

Figure4.14:EnergyselectiveNRimageoftwoelectricmotorsandamaterialstack:t=216ms,
Δt=100µswith40000accumulations,σ=10-30%,Δx=0.5mm,FOV:100x100mm(833x
l)epix833

ThesecondsampleisvisibleinFig.4.14.Astackofmaterialsshouldprooftheabilityto
discriminatematerialsandtwoelectricmotorsshouldserveatestobjects.Thematerial
fromtoptobottomare:PVC,teflon,titanium,aluminium,iron,copper,brassandsteel.
Fig.4.14showsaNRoftheobjects.
ANRseriesof39images,startingwithwavelengthsbetween3.840and3.864A˚and
endingwithwavelengthsbetween4.742and4.766˚A,wasrecorded,[ppt3,45].Forsome
materialsofthestacktherapidchangesintheattenuationcoefficientsΣneartheBragg
energy(Braggcutoff)wereobserved.SinceBraggcutoffsaredeterminedbythelattice
typeandthelatticeparameterstheyarecharacteristicforeachmaterial.Thechangesin
Σfromoneenergytothenextwerevisualizedintheformofdifferenceimages(seeFig.
4.15).Inthefalsecolordifferenceimagesthecontrastchangesarevisibleinthegreen,
blueandwhitecolor.Becauseinthedifferenceimagealonethepositionofacharacteristic
materialrelativetotheobjectsometimesisnotclear,itwasfusedwithaNRimage(see
Fig.4.16).Ironcanbedistinguishedbestfromtheotherelementsbecauseofitslarge
Braggcutoff.Theironpartsoftheelectricmotorappearbrighterandprovethatitis
possibletoresolveamaterialspositionsensitively.Inordertoobtainthebestcontrast,
oneenergyaboveandbelowtheBraggcutoffshouldbeused.Themethodisnotlimited
toironbutcanalsobeappliedforthedetectionofotherelementslikecopperorbrass,
Fig.4.17.Anecessaryconditionforthematerialdiscriminationisthepolycrystallinesolid
statewhichmostmetalshave.Becauseofthelowneutronfluxandtheshortmeasurement
periodthecameraparameterscouldnotbeoptimizedandthestatisticsaremoderate.In
atimewindowof1msabout1000countswereregisteredbythedetector.

4.2EnergyselectivetimeofflightNR

66

Figure4.15:DifferenceimagesoftwoNRimages:inblackandwhite(left)andinfalsecolors
t)gh(ri

Theuniquenessofthistechniqueisthecombinationofanexcellentenergyresolution
oftimeofflightwiththegoodspatialresolutionofalargeareaNRdetector.Afterthese
promisingresultsaconsecutiveexperimentwasplannedandprepared.Thebeaml2ines
IRISorOSIRISofferhigherfluxbyafactorof10andacoldspectrumfromaliquidH
moderatorat22K.Unfortunatelythecompletebeamtimewascancelledduetoproblems
withtheacceleratorcoils.

4.2

gyreEn

ectiveles

teim

of

tfligh

NR

Figure4.16:DifferenceoftwoNR(above)at4.368A˚and4.392˚A,differenceNR
iron(below)withthegreencolorintheimagestackaswellasinthehousingsof
motors

67

htheighlitwghotinelecgtrtheic

atNRowtofdifference),evo(ab4.17:reFiguaBraggcutoff

rofbrass(below)

e68

coppNR

oftfligh

offof

cuteimt

Bragg4.2

agyreEn

atelesectiv

NRowtofDifference

4.3QuantitativeNR:acompressortyperefrigerator

69

4.3QuantitativeNR:acompressortyperefrigerator
AfrequentproblemintheNRdataevaluationisaquantitativevolumemeasurement.As
describedinchapter3,section3.5,thevolumeindependenceofthegreylevelg,thegrey
leveloftheemptybufferg0,theattenuationcoefficientoftheliquidΣ,andtheareaApixel
correspondingtoonepixelonthescintillator,is:

(4.2)

V=−AΣpixelln(gg0)ifµisconstant(4.2)
pixelsofROI
V...volumeoftheliquidcm3
g(z)...neutronintensityafteradistancexinmaterialneutr2ons
g0...neutronintensitybeforetheobjectneutcm2rsonscms
Σ...attenuationco2efficientcm1
Apixel...pixelareacm
ThisevaluationwasappliedtoNRdataofacompressortyperefrigerator.Theevapo-
thratieonmacandhine.theInthecondensationliquidprobuffercessofofththeeevcoapolingoratorliquid(Fig.4.wass18turdiedight)inthordeecortoolingoptimizeliquid
evapcondensationoratesloiwnertheingwrtheongteplmpaceeratuofretheincircuithetinlesiaddestoofcthompleedeetepmfalfreeunzectiunonit.ofthSomeetirefrimg-es
erator.

liqFiguuidrecirc4.18:uitinThethistrefriypegerofratorefinfrigeratorrontsof(rtheight)neutronradiographydetector(left),schemeofthe
TheexperimentwasperformedatKFKIBudapest8(Fig.4.182left).Atthedynamic
neutronradiographystationatKFKI,aneutronfluxof10n/cmsataL/Dratioof170
isavailable.Theemptyrefrigeratorwaspositionedonatranslationtablesuchthatthe
liquidbuffer(see4.19)wasinthebeamandcouldbeinspected.Anexposuretimeof1ms
foroneimagewaschosen.Thespatialresolutionwas1.5mmduetothebeamdivergency.
ThedetectorwasastandardNRdetectorwithaMCPintensifiedPeltiercooledICCD
camerafromthecompanyPrincetonInstruments.
AmeasuredNRimageisvisibleinFig.4.20.WiththeIDLprogramvolume_calc.pro
theRegionOfInterst(ROI)forthevolumecalculationandtheoffsetgreylevelarechosen.
AllpixelintheROIwithalowergreylevelthantheoffsetcontributetotheliquidvolume

4.3QuantitativeNR:acompressortyperefrigerator

70

Figure4.19:Photographoftheliquidbufferofarefrigerator,boilingtemperatureoftheliquid
◦C-30andaredisplayedinfalsecolors.Inthecaseoftherefrigeratortheevaporatorwasselected
asROI,theliquidinthebufferwasvisualizedandvolumewascalculatedviathethickness
oftheliquidineverypixel.

(riFigughret),grey4.20:levelsNormalibetwzeedenNR10000oftheandliquid25000buffareerofdisplatheyedrefrinfigeratoralse(colors,left);ΔRtOI=of1.5themsliqwithuidbu12000ffer
accumulations,Δx=1.5mm,σ=4%,FOV:120x120mm2and680x680pixels
theFattorentheuationcalculaticoefficonofientthwevereoluemestimateuncedr.tainWityththteheunIDLcertaiprntiesogramofbatthechpixeprolcareseasingandis
possible:anunlimitednumberofimagescanbeprocessedwiththesameparameters
[ppt4,46].Thepaletteontherighthandsideofthebmpcolorimagehelpstointerpret
4.21).(Fig.imageethwillbTheeappliquliedidtvomolumeoreinthcomplexerefgeomrigeratoretrieswinasfusucturcee.ssfInullythecalcascueoflatedtheandrefrtheigerpatorrogramthe
precisionforthevolumecalculationwasafewpercent.Theexactpixelsizeandthe
attenuationcoefficientofthecoolingliquidwereonlyestimatedbutcouldbedetermined
byappropriatecalibrationmeasurements.Beamhardeningandscatteringmaybethe
limitingfactorsthen.

4.3

Quevtitatian

NR:

a

sorrespcom

teyp

Figu12000reac4.21:cumulSomeations,Δxframes=of1.5thmem,NRσ=4%,stack

atorrgerefri

teafrhetiqulidoluvmeccallationu:Δt=1.FOV:120x120mm2and680x680pixels

5ms71

hwit

pApAendix

Newapplication:fossilstone

Fossilsaregeologicalandhistoricalobjectsfromatimeofsomehundredmillionyears
ago.Theycanhelpustoliftsecretsliketheevolutionofplantsandanimalsonearthand
themountainformationprocesses.Afterthediscoveryofafossilaverylaborious,time
consumingandcostlypreparationfollowsuntilthefossilcanbeadmiredinamuseum
1).A.g.(Fi

FigureA.1:Pterodactyluselegans,workingplaceforfossilpreparation.
Fromthemanypromisingpiecesofrocksonlyafewreallycontainabeautifulfossil.
Acarefulandaccuratepreparationtakesnothoursbutweeksormonthsandrequires
professionaltoolslikearobuststereomicroscopeandappropriategrindingmachines.It
isworthinspectingasaccurateaspossibleafindingbeforethepreparation,butitisvery
difficulttolookdeepintothepieceofrockandtodiscoverthetracesofplantsandanimals
non-destructively.
Thefossilstone(Fig.A.2left)fromthenorthernDolomitescontainingfossilshellswas
inspectedattheANTARESfacilitywithX-rayandwithneutrons(Fig.A.2,centerand
right).Itsageisestimatedtobe240millionyearswhichcorrespondstotheTriasage.
Neutronimagingisthefavoritetechniqueduetothehighpenetrationdepthofneutrons
anditsgoodcontrastsforhydrogenousandcarbonaceousmaterials.Inthenextstepa
neutrontomographystudywasstarted.
InFig.A.3the3Dneutrontomographydatasetisdisplayedfromthreedifferentpoints
w.eviofThehistogamA.4showsthegreylevelsofthe3Ddatasetandtheassignedcolors.
Thehighpeakonthelefthandsidecorrespondstohighnumberofvoxelswithlowgrey
levels(lowneutronattenuation)likeair.Thesmallerpeakcanbeassignedtothefossil.
Thebasiccolorochershouldbesimilartothecolorofthestoneonthephotographand

73

FigureA.2:Limestoneincludingfossilshells:Photograph(left),X-rayimage(center)and
NRimage(right);thevoidisbettervisibleintheX-rayimage,theNRimagegivesmuchmore
contrastsfortheshells

FigureA.3:NTdatasetofthelimestonewithfossilshells:frontview(left),under45◦(center),
t)righ(viewside

thegreencolorhighlightstheshellsweremarked.Thebrighterthecolorintheslicedthe
strongeristheneutronattenuation.
Fossilshellswerealreadyvisibleintheopticalinspectionofthepieceoflimerock.
Thequestionwas:IsitpossibletogetenoughcontrastintheNTseethe◦shellsinsidethe
stone?AtaL/Dratioof400aseriesof400imageswithinarotationof180wereacquired,
eachwithanaccumulationtimeof1.25s.ThespatialresolutionofoneimageΔxwas0.7
mmandtherelativenoiseσ=0.7%.WiththeIDLprogramall_in_one[Sch04,24]the
tomographicreconstructionwasperformed.AfterfourhoursonanormalPCallslicesof
the3Ddatasetwerecalculated.
Fig.A.5showstwocutsofthefossilrichindetail.Thepositionandtheorientationofthe
shellscanbeclearlyidentified.Inadditionvoidsandcreepsinsidethestoneappear.In
themovie[ppt5,47]thefossilispresentedwithmoredetails.Neutronimagingmethods
areanexcellenttoolfortheinvestigationoffossils.Thehighsensitivityformanyelements

reFigu

A.4:

Histogramandolokupletaboftheosfsilofandthtehefossicapablandilitycantospuppenetrateortthethickpreparobjeationctsandelivderstheuiniqnspueectinionforinmfatiutuonre.about

the

74

eidsin

reFigu

eth

A.5:

Cutsorieniontat

of

threth

oughlslehs

theas

D3ellw

nas

eutronsoidv

tomographedu

to

ydatasetzcrystalliation

oftheand

fosskscrac

:ilTare

ehposition.evisibl

75

dan

ApBendixp

Visualizationmethod:Difference
NRofanoilpump

ThisimagingmethodhelpstovisualizedifferencesbetweentwoNR.Asatestobjectan
oilpumpwaschosen.Whiletheeyecandistinguishonlywithdifficultieswheretheoil
is,inthedifferenceimagewithfalsecolorsitisveryeasytosee.Thecolorencodesthe
neutronattenuationoftheoil.Fig.B.1showstheNRoftheoilpumpinthefilledandin
theemptystate.

FigureB.1:NRofanemptyandafilledoilpump
InthenextevaluationsteptheNRofthefilledpumpwasnormalized(dividedby)
withtheNRoftheemptypump.Theresultingimageistheneutronattenuationofthe
oilandtheuncertaintyisverylargeintheareasoflargethickness(Fig.B.2).
Thevisualizationoftheexactoildistributioninthepumpwastheaim.Thatwas
achievedwithanimagefusionoftheNRimageofthepumpandtheNRoftheoil(Fig.
B.3)showsinanimpressivewaythelocationsoftheoil.
Suchnewvisualizationtechniquesshortenthetimefornondestructivetestingand
makeresultsmucheasiertointerpret.Pre-knowledgelikegeometryorknowledgeofthe
materialsisusedintheevaluationprocedure.

reFigu

reFigu

B.2:

B.3:

NRFofusiontheofoilethinNRethpuimagempof(left),ethoilimapugempofitsanduncertainethoilniyt(rfalset)ighcolors77

ListofFigures

1.1PrincipleofNR..................................2
1.2Attenuationpropertiesoftheelementsforneutronsandx-rays.......4
1.3Scatteredneutrons................................5
1.4NRofscatteredneutrons............................6
1.5Totalreflection..................................7
1.6NRofatotalneutronreflection.........................8
1.7Energydependenceofneutroncrosssectionsforneutrondetectorcandidates8
1.8Effectsofbeamhardening...........................9
1.9Braggcutoffsofsomematerials.........................9
1.10SchemeoftheBragglaw,irontransmissioncurveversusneutronwavelength10

2.1Beamdivergencyandspatialresolution....................12
2.2ShieldingofthetomographyfacilityAnatares.................12
2.3SchemeoftheneutronbeamatANTARES..................13
2.4NeutronspectrumatANTARES........................14
2.5ANTARESneutronintensityfluctuations...................14
2.6SchemeofaNRdetectorandneutronconversionprinciple..........15
2.7Structureanddynamicpropertiesofthescintillator..............16
2.8ChargetransportinaCCDpixel........................17
2.9WorkingprincipleandtransmissionofaFLCshutter.............18
2.10WorkingprincipleofaninterlinetransferCCD................19
2.11SchemeofanintensifiedCCDcamera.....................20
2.12SchemeandelectronmicroscopyimageofaMCP..............20
2.13ElectronmultiplicationinasingleMCPchannel...............21
2.14WorkingprincipleofashutteredCCD.....................22
2.15NRdetectoratANTARES...........................23
2.16Possibleobjectthicknessesforthermalneutronradiography.........24
2.17Themathematicalfunctionsusedfortheresolutiondetermination.....27
2.18Contrastandnoiseinaneutronradiographydetermination.........28
2.19NRoftheCdforresolutiondetermination...................28
2.20EdgeSpreadFunction..............................29
2.21PointSpreadFunction..............................30
2.22ModulationTransferFunctionMTF......................30
2.23ModulationTransferFunctionMTF......................31
2.24Neutronquantumnoise.............................32

3.1Filteringofgammaspots............................35
3.2Filteringofgammaspots2...........................35
3.3Lensdistortionsbeforeandaftercorrection..................36
3.4NRbeforeandafteradeconvolutionwiththeWienerfilter.........37

SFIGUREOFLIST

3.5OpenbeamNRanddarkimage........................
3.6RawandnormalizedNRofarefrigerator...................
3.7NRofstepwedgeswithuncertaintyandrelativeuncertainty.........
3.8NRprofileofstepwedges............................
3.9TheSNRversusthenumberofframes.....................
3.10Detectorresponse.................................
3.11DataevaluationsoftwareNeutroneye......................
3.12ExtractingdetailsofaNRimageusingNeutroneye..............
3.13Schemeofdifferenceimages...........................
3.14RatioofNRimages...............................
3.15DifferenceofNRimages.............................
3.16Imagefusion...................................
3.17Imagefusion...................................
3.18Humaneyeandretina..............................
3.19Densityofreceptorsonthehumanretina...................
3.20Contrasttransferfunctionofthehumaneye..................
3.21Spatialsummationofthehumaneye......................
3.22Detectabledetailsinimage...........................
3.23Logaritmicalperceptionofthehumaneye...................
3.24Machbandsandillusorycontours........................
3.25Temporalcontrastsensitivityofthehumaneye................

4.1Photographofthemodelaircraftcombustionengine.............
4.2NRofthemodelaircraftcombustionengine..................
4.3NRseriesoftherunningmodelaircraftcombustionengine..........
4.4PhotographoftheBMWcombustionengine..................
4.5NRoftheBMWcombustionengine......................
4.6Dieselinjectionsetupforneutronradioscopy.................
4.7NRofthedieselinjection............................
4.8NRofthedieselinjection............................
4.9Dieseldirectinjectiongeneratortestsetupfordynamicneutronradiography
4.10NRframeoftheNRmovieofthedieselgenerator..............
4.11InstrumentationoverviewattheISISfacility.................
4.12EnergyselectiveNRofaGadoliniumfoil...................
4.13NeutronattenuationofaGdfoilatvaryingenergy..............
4.14EnergyselectiveNRimage...........................
4.15DifferenceimagesoftwoNRatdifferentenergiesshowing..........
4.16DifferenceoftwoNRatdifferentenergiesshowingFe.............
4.17DifferenceoftwoENRimagesoftheseriesshowingCuandbrass......
4.18RefrigeratorinfrontoftheNRdetector....................
4.19Photographoftheliquidbufferofarefrigerator................
4.20QuantitativeNRoftherefrigerator.......................
4.21LiquidvolumecalculationwiththeNRdataoftherefrigerator.......

A.1Fossilstoneandworkingplaceforfossilpreparation.............
A.2Pterodactyluselegansandaworkingplaceforfossilpreparation.......
A.3The3Ddatasetofthefossil..........................
A.4Histogramandlookuptableofthefossil...................
A.5NTcutsofthefossil...............................

B.1NRofanemptyandafilledoilpump.....................

79

383839394041434344444546464748484949505050

525354555657585960616364646566676869707071

7273737475

76

SFIGUREOFLIST

B.2B.3

FNRusiofonoftheoiNRl

iinmagethe

pumptheof

anoildptheumpimanagedoilof

initsfualsnceecrolortainsty

.

..

..

..

..

..

..

..

..

..

80

7777

Bibliography

[Kak88,0]”PrinciplesofComputerizedTomographicImaging”,A.C.Kak,M.Slaney,
IEEEPress,1988

[Eme82,1]”TheoriederKernreaktoren”,D.Emend¨orfer,K.-H¨ocker(eds),Bibliographis-
chesInstitutAGSu¨rich,1982

[Due76,2]”NuclearReactorAnalysis”,J.J.Duedestadt,L.J.Hamilton(eds),JohnWiley
&sons,1976

[Iae05,3]InternationalAtomicEnergyAgency:www-nds.iaea.org,2005

[Fur99,4]”Frontiersofneutronscattering”,AlbertFurrer,WorldScientific,1999

[San01,5]”Developmentandapplicationsofapixellatedtransmissiondetector”,J.
Santisteban,L.Edwards,A.Steuwer,PJ.Withers,M.R.Daymond,NJRhodes,E
MSchooneveld,Isisannualreport2001

[Hil05,6]”Highfluxneutronimagingforhigh-speedradiographyanddynamictomo-
graphyandstronglyabsorbingmaterials”,A.Hillenbach,M.Engelhard,H.Abele,
R.G¨ahler,NuclearInstrumentsandMethodsinPhysicsResearchA,2005

[Ciz99,Instru7]m”PenerftsaormacndeanMethodcdsiharactenPhrysisticicssofResaearcnewhsAcin424til(15-lator”,19),J.B.(1999)Cizzetal.,Nuclear

[Dis05,8]DatasheetforLCDShutteronwww.displaytech.com,2005

[Oly05,9]www.olympus-biosystems.com/templateseng/bioimaging/glossary
CCDchips.htmlDigitalImaginginOpticalMicroscopy,1005

[Rei93,10]”IntegratedElectronicShutterforBack-IlluminatedCharge-CoupledDe-
vices”,R.Reich,R.Mountain,W.McGonagle,J.Chin-MingHuang,J.Twichell,B.
Kosicky,E.Savoye,IEEETrans.onElectronDevices,Vol.40/7,1993

HYAPRLIOGBIB

[Ast04,11]Datasheetonwww.appscintech.com,2004

82

[Rop04,12]Datasheetonwww.roperscientific.com/pdfs/datasheets/pimax/
2004,1024sb.pdf

[Bac02,13]”Newfeaturesincoldneutronradiographyandtomography:PartI:thinner
scintillatorsandaneutronvelocityselectortoimprovethespatialresolution”,S.
Baechler,N.Kardjilov,M.Dierick,J.Jolie,G.K¨uhne,E.LehmannandT.Materna,
NuclearInstrumentsandMethodsinPhysicsResearchSectionA:Accelerators,
Spectrometers,DetectorsandAssociatedEquipment,Volume491,Issue3,Pages
481-491,1October2002

[Sch99,14]”NeueEntwicklungenzuRadiographieundTomographiemitthermischen
Neutronenundzuderenroutinem¨as¨igemEinsatz”B.Schillinger,Dissertationander
TechnischenUniversit¨atM¨unchen,1999

[Tre05,15]”Newwaysinneutrontomography”,W.Treimer,http://www.physik.
uni-kiel.de/kfn/Highlights/Highlight-2-engl.php,2005

[Sch04,16]”NeutronensehenwasR¨ontgenstrahlenberborgenbleibt”,B.Schillinger,T.
B¨ucherl,ZfP-Zeitung89,2004

[Gru05,17],”DesignoftheneutronimagingfacilityANTARESathighfluxneu-
tronsourceFRM-II”,F.Gr¨unauer,”DissertationanderTechnischenUniversit¨at
Mu¨nchen”,2005

[Har86,18]”MathematicsandPhysicsofNeutronRadiography”,A.A.HarmsandD.
R.Wyman,D.ReidelPublishingCompany,1986

[Has91,19]”ThePhysicsofmedicalx-rayimaging”,B.H.Hasegawa,MedicalPhysics
Publishing,MadisonWisconsin,1991

[Kol05,20]”Webvision,theorganizationoftheretinaandhumanvision”,H.Kolb,E.
Fernandez,R.Nelson,http://webvision.med.utah.edu/,2005

[Jai89,21]”Fundamentalsofdigitalimageprocessing”,AnilK.Jain,PrenticeHall
Informationandsystemssciencesseries,1989

[Pra91,22]”Digitalimageprocessing”,W.K.Pratt,AWiley-IntersciencePublication
JohnWiley&Sons,Inc,1991
[Idl05,23]”SolutionsforDataVisualizationandImageAnalysis”,http://www.rsinc.
com/

RLIOGBIBHYAP

83

[Sch04,24]”AufbaueinerR¨ontgenanlagef¨urRadiographieundTomographieamExper-
imentANTARESamFRM-II”,MichealSchulz,DiplomarbeitanderTechnischen
Universit¨atMu¨enchen,2004

[Sfb05,25]”Abschlu¨sberichtSonderforschungsbereich224MotorischeVerbrennung”,
DeutscheForschungsgemeinschaft,http://www.vka.rwth-aachen.de/sfb224/,
2005

[Bal99,26]”Characterisationofsegregationbydynamicneutronradiography”,M.Bal-
ask`oE.Svab,Actaphysicapolonica,Vol96(1999)

[PSI05,27]”TheNeutronTransmissionRadiography(NEUTRA)StationatPSI”,
2005,http://neutra.web.psi.ch/facility/index.html

[ILL05,28]”Neutrograph,radiographyandtomographywithneutronsforindustryand
science”,http://www.neutrograph.de/,2005

[Fer05,29]”ThenewstationforfastradiographyandtomographyattheILLinGreno-
ble”,T.Ferger,H.Abele,J.Brunner,R.Gaehler,B.Schillinger,J.R.Villard,
September2002

[Mac02,30]”X-rayimagingofshockwavesgeneratedbyhigh-pressurefuelsprays”,
FonMacPhteseEe,AWGal,therTateJ,MScW,hallerPoJw,ellGrunerCF,SYuM,eWY,angRenziJ.,SciMJ,ence.Ercan15;295A,(5558)Naray:1261-anan3.S,,
2002beF

[Bru01,Brun31]ner,”AufDipbaulomarbeineseitanTederststandTecehsniscfu¨hrendynUnivamersisciht¨eatMNe¨unuchtronene,nradi2001ographie”,J.

[Mue04,32]”BaueinesEntwicklungssystemsf¨urRadiographieundTomographiemit
Neutronen”,MartinMu¨hlbauer,DiplomarbeitanderTechnischenUniversit¨at
Mu¨nchen,2004

[Eng04,33]”Imagingofdynamicprocesseswithatimescaleofmicrosecondsatathermal
neutronbeam”,MartinEngelhardt,DiplomarbeitanderTechnischenUniversit¨at
Mu¨nchen,2004

[Sch05,34]”Detectionsystemsforshort-timestroboscopicneutronimagingandmeas-
urementsonarotatingengine”,NuclearInstrumentsandMethodsinPhysics
ResearchSectionA:Accelerators,Spectrometers,DetectorsandAssociatedEquip-
ment,B.Schillinger,H.Abele,J.Brunner,G.Frei,R.G¨ahler,A.Gildemeister,A.
Hillenbach,E.LehmannandP.Vontobel,2March2005

HYAPRLIOGBIB

84

[Bru05,35]”Characterizationoftheimagequalityinneutronradioscopy”NuclearIn-
strumentsandMethodsinPhysicsResearchSectionA:Accelerators,Spectrometers,
DetectorsandAssociatedEquipment,InPress,CorrectedProof,availableonline1
April2005,J.Brunner,M.Engelhardt,G.Frei,A.Gildemeister,E.Lehmann,A.
HillenbachandB.Schillinger

[Von05,high-in36]tensit”Dynyamineutrconimaginbegam”,withNuactrleariggereInstrudandmenintsandtensifiedMethoCCDdsicanPhmeraysicssysRestemearcinah
VonSectiontobel,A:G.AcFcerei,leJ.ratBorsr,unSnepre,A.E.ctromeGters,ildemeDeteister,ctorsM.anEndgelAssohardt,ciatedMarcEquh2005ipment,P.

[Kar04,37]”Energy-selectiveneutronradiographyandtomographyatFRM”,Applied
RadiationandIsotopes,Volume61,Issue4,Pages455-460N.Kardjilov,B.
Schillinger,E.Steichele,October2004

[Kar03,38]”NewfeaturesincoldneutronradiographyandtomographyPartII:applied
energy-selectiveneutronradiographyandtomography”,N.Kardjilov,S.Baechler,M.
Bastu¨rk,M.Dierick,J.Jolie,E.Lehmann,T.Materna,B.Schillinger,P.Vontobel,
NuclearInstrumentsandMethodsinPhysicsResearchSectionA:Accelerators,
Spectrometers,DetectorsandAssociatedEquipment,Volume501,Issues2-3,1,
Pages536-546,April2003

[San01,39]”Time-of-flightneutrontransmissiondiffraction”,J.R.Santistaban,L.Ed-
wards,M.E.Fitzpatrick,A.Stewer,P.J.Withers,M.-R.Daymond,M.W.Johnson,
N.Rhodes,E.M.Schooneveld,JournalofAppliedCristallography.34,289-297,2001

[Ste01,40]”StrainimagingbyBraggedgeneutrontransmission”,A.Steuwer,P.J.
M.WitW.hersJoh,J.R.nson,SanD.Wtisteanbg,an,L.Nucl.EdInwsartrds,umeG.ntsBranundo,MethM.Eo.dsFiintzpPhatricysicsk,reM.seRar.cDahSecymondtion,
20011-4,0A

[Ste01,41]”BraggEdgeDeterminationforAccurateLatticeParameterandElastic
StBrunraino,MM.Ee.Fiasuremetzpntatric”,k,A.M.R.SteuwDaer,P.ymond,J.WithM.W.ers,JohnJ.R.sSon,anD.tisteWbang,an,L.PhEys.dwstatard.s,SolG..
(a)No.2,221-230,2001

[ppt1,42]”Modelaircraftengine”(addedoncompactdisc:links/ppt/hub/hub.ppt)

[ppt1,43]”BMWmovie”(addedoncompactdisc:links/ppt/bmw/bmw.ppt)

[ppt2,44]”DynamicNRofageneratoratFRM-II”(addedoncompactdisc:
)frm2.pptfrm2/gen-links/ppt/gen-

APRLIOGBIBHY

[ppt3,45]”EnergySelectiveNR”(addedoncompactdisc:

[ppt4,46]”QuantitativeNR”(addedoncompact
)refrigerator.ppt

47]pt5,[p)ppt

”NT

of

a

ilfoss

s”eton

added(

on

tacpcom

c:sdi

disc:

85

)links/ppt/esnr/esnr.ppt

links/ppt/refrigerator/

links/ppt/fossil/fossil.

onstiAbbrevia

CCDCVSFESRFESFFTVOFIFRM-IFWHMDICCIDLILLISISKFKILSPMCPMCNPMTFNRNTNINDTFPSIPSSNROIR

...ChargeCoupledDevice
...Softwareforversionmanagement
...EdgeSpreadFunction
...EuropeanSynchrotronRadationFacility,Grenoble,France
...FastFourierTransform
...FieldOfView
...ForschungsreaktorMu¨nchenII
...FullWidthHalfMaximum
...IntensifiedCCD
...InteractiveDataLanguage,softwarefordata
...visulizationandimageprocessing
...InstitutLaueLangevin,Grenoble,France
...NeutronSpallationSource,Oxford,UK
...ResearchcenteroftheHungarianAccademyofSciences,Budapest,Hungary
...LineSpreadFunction
...MultiChannelPlate,componentofanimageintensifier
...MonteCarloNuclearParticleCode,softwareforMonteCarloSimulations
...ModulationTransferFunction
...NeutronRadiography
...NeutronTomography
...NeutronImaging
...NonDestructiveTesting
...PointSpreadFunction
...PaulScherrerInstitut,Villigen,Switzerland
...SignaltoNoiseRatio
...RegionofInterest(cutout)ofanimage

onstiicaPubl

Dip”AuflomarbbaueitaneinesderTTeechststandesnischenfu¨Unrivdersyniamt¨atiscMhe¨uncNehen,utronenr2001adiographie”,J.Brunner,

”DynamicNeutronRadiographyofacombustionengine”,J.Brunner,E.Lehmann,
B.Schillinger,NeutronRadiography(7),ProceedingsoftheSeventhWorldConference,
Rome,Italy,September15-21,2002

”ThenewstationforfastradiographyandtomographyattheILLinGrenoble”,
T.RadiFerger,ographyH.(7),AbPreleo,ceJ.edingsBrunofner,theSR.evGena¨thhler,WorldB.ScConhilferelinngecr,e,J.Rome,R.ItalyVillar,d,SepteNeutmbroner
200215-21,

Hillen”Dynbacham,icE.neutrLehmannonrad,B.iograpSchhillyofinger,acPomrobucestiedinongsenofginthee”,16thJ.WBrunorldner,ConG.Fferencerei,A.of
NonDestructiveTesting,Montreal,Canada,http://www.ndt.net/article/wcndt2004/
2004,index.htm

”Characterizationoftheimagequalityinneutronradioscopy”,NuclearInstruments
andAssoMethociateddsEinquipPmhenysicst,J.ResBrunearchner,SecM.tionEngeA:lhardAccet,G.leratorFresi,,SA.pecGiltdemromeeisteters,r,DeE.tecLethorsmanann,d
A.HillenbachandB.Schillinger,1April2005

”Detectionsystemsforshort-timestroboscopicneutronimagingandmeasurements
onarotatingengine”,NuclearInstrumentsandMethodsinPhysicsResearchSection
A:Accelerators,Spectrometers,DetectorsandAssociatedEquipment,B.Schillinger,H.
Abele,J.Brunner,G.Frei,R.G¨ahler,A.Gildemeister,A.Hillenbach,E.Lehmannand
P.Vontobel,2March2005

”DynamicimagingwithatriggeredandintensifiedCCDcamerasysteminahigh-
inA:tenAcsitceyleneratuorstron,Sbpeam”,ectromeNucters,learDeInstrtectorsumentsandandAssoMceithateoddsEinquipPhmeysicsnt,ResP.Vearconhtobel,SectionG.
Frei,J.Brunner,A.E.GildemeisterandM.Engelhardt,21March2005

talDigi

rsionev

on

compact

iscd

nsusioonclC

Inthecurrentstudyvariousexperimentalmethodsandcomputationbasedprocedures
inthefieldofneutronimagingwerepresented.Someoftheresultsarereportedworld-wide
forthefirsttime.Themostsignificantcontributionsofthecurrentthesistothefurther
developmentoftheneutronradiographycanbenamedasthefollowing:
1.Thedevelopmentofthestroboscopicneutronimagingtechnique,allowingneutron
imagingatsub-millisecondtimescaleofperiodicprocessesandthefirstvisualization
ofarunningcycleofacombustionenginerunningonitsown.
2.Thefirsttimeofflightbasedenergyselectiveneutronradiographyexperimentsusing
afastgateabledetectoratthepulsedneutronspallationsourceISIS.
3.Thedevelopmentofaquantitativeevaluationandvisualizationsoftwareforthe
calculationofuncertaintiesandfastdatainspection.
Withstroboscopicneutronimagingperiodicprocessesatsub-millisecondtimescales
canbeinspected.Bysynchronizingapositionsensitivedetectorwithshortgatingtimesto
theprocess,takingstroboscopicallyimagesofthesametimewindowandsummingthem
eitherbysoftwareoronthechip,thetimeresolutioncanbeimprovedbyafactorof1000.
Byphaseshiftingofthetriggersignalaneutronradiographymoviecanbecomposed.In
chapter4thismethodwasappliedintheinvestigationofcombustionenginesandinjection
s.nozzleThepresentedenergyselectiveneutronimagingmethodisbasedonthetimeofflight
principle.Atpulsedneutronspallationsourcestheneutronenergiesatsampleposition
aretimedependentandcanbeselected,againbythesynchronizingdetectorwiththe
neutronsourceandchangingthephaseofthetriggersignal.Incontrasttoenergyselective
neutronradiographyusingavelocityselectororfilterstheenergyresolutionismuchhigher.
Theenergiescanbeexactlytailoredtothespecificspectralattenuationpropertiesofthe
materialonewantstodetect.Atnextgenerationspallationsourcesthistechniquewill
becomeveryimportant.
Theevaluationandvisualizationofthelargedataamountsisonlypossiblewitha
highlyoptimizedandsophisticatedsoftware.Becausenocommercialsoftwareisflexible
enoughtobeadoptedandexpandedtoourspecificneeds,manyroutineswereprogrammed
inIDLandmostofthemwereembeddedinagraphicaluserinterface.Thisevaluation
software(Neutroneye)isdescribedinchapter3.
DuringthisworktheneutronimagingfacilityANTARESatFRM-IIstartedopera-
tion.Withmanytestexperimentsthefacility,thedetectorandthedatahandlingwas
characterized,improvedandoptimized.
Thedevelopedmethodsandtechniquesareavailablefortheinstrumentationstaff,for
futureusersandstudentsoftheneutronradiographyandtomographyfacilityANTARES
atthenewresearchreactorFRM-IIandcanbefurtherimprovedandexpanded.

koloOut

Stroboscopicimagingwillbeappliedtoallkindofcyclicprocesses.Thedynamics
oflubricationliquidsandcoolingagentsinmetallichousingsareideallysuitedforan
inspectionwiththismethod.Forproductdevelopmenttheobjectscouldbeadaptedto
neutronimagingbysubstitutionofimpenetrablematerialswithwindowmaterialsorby
downsizingoftheobject.
AtthenextgenerationspallationneutronsourcesenergyselectivetimeofflightNRwill
allowveryprecisematerialselectiveinspections,wherethespectrumistailoredforthe
specificproblem.Atsteadystateneutronsourcesvelocityselectorscanbeusedforenergy
on.ariativInfuturethesoftwaretoolsusedatneutronimagingfacilitieswillincreasesignificantly.
Bytheuseof”pre-knowledge”abouttheobjectlikegeometryofmaterialcompositionit
willbepossibletorecognizecomplexpatternsandtoperform3Dmeasurementswithsub
pixelresolution.
Fittingalgorithmsandevolutionarysoftwarewillallowautomatedinspectionoflarge
dataseries.ForindividualNDTproblemsasoftwarelike”Neutroneye”hastooffer
complextoolsforevaluationandvisualization,especiallyforcombinationofdifferent
datatypes.
ThecombinationofvariousNDTmethodslikeultrasonicinspection,infraredimaging,
X-rayinspection,magneticresonanceimaging,eddycurrentinspection,neutronimaging
andotherswillextendourpossibilitiesofnondestructivetestingimpressively.

Acknowledgements

AttheendofthisworkIwouldliketocordiallythankallwhohavecontributedtoits
successfulcompletion.

FirstofallIwouldliketothankProf.Dr.P.B¨onifortheuniqueopportunitytobea
scmeienmbtifiercinmishissionsscienantifidcthegroupparticanidforpationhisatcgreatonferesupnpcores,tofwhinechwchadollabaoratstrongions,inshflueortnceteronm
mypersonalscientificdevelopmentandmylife.

EspeciallyIwanttothankthescientificsupervisorduringmyPhDtimeDr.B.
Schillinger.Withhisexcellentscientificunderstandingandtechnicalknow-howhe
supportedmewhenevernecessary.Hisencouragementandhighmotivationwithoutsup-
pressingthenecessarycreativitycreatedaextremelygoodatmosphereintheANTARES
p.grou

IwanttothankallmembersoftheANTARESgroupfortheirhelp,thefruitful
diMsu¨hlbcussiaueronsandandM.theScfunhulzw.eSphadecialtogether:thanksEgo.toCalzada,Dr.N.F.KGrardju¨nauilover,forK.allLorenz,thehelM.p,
especiallyatthebeginningofmyPhDthesis.

ForthesuccessfulcooperationwiththeInstituteLaueLangevin,Grenoble,France,I
wanttothankDr.R.G¨ahler,H.Ballhausen,M.Engelhardt,T.Ferger,A.Gildermeister
andAndreasHillenbach.

wantFortothethanksuccDres.sfuE.lcLeohopmanerationn,G.witFhrei,theR.PHasaulScsaneinherrerandPInstitu.Vt,ontobVilel.ligen,SwitzerlandI

IwanttothankDr.MartonBalask`o,whoreallyimpressedmebyhisbrilliantwayof
doingNDTandmademyshorttermscientificmissioninBudapestpossible.Thankyou
forthemanynewinspirationsandideas.

FortheinjectionnozzlesetupIwanttothankDr.W.BauerfromBosch.Forthe
seRiedltupfrofomthethecardepcombartmeustionntofmeengincehaniIcamalengratefuginelerintogPanrof.dforDr.theK.moZeilindelgeairrcanrdaftDr.enH.gineJ.I
wanttothankM.Axtner.

andFfromurthertheIFwanRM-IttoIthsoftankwareallthgrouepniceforptheireoplecfromommitmethenwtoranksdhoptheirofpattheience.Reaktorstation

HYAPRLIOGBIB

92

DuringmyPhDthephysicsstudentsAndreasNeubauer,CarolaOberh¨uttinger,So¨ren
Schlimme,KonradSennandVeronikaVitztumcontributedtotheIDLprogramming,the
measurementsandthedataevaluation.

TwostudentsofthecomputersciencedepartmentoftheTU-Mu¨nchen:AlexKruppa,
FranzStrasser,AlexKruppa,FranzStrasserworkedformewithinthescopeofan
interdisciplinaryproject(IDP),whichwasveryfruitful.

ForthefinalcorrectionsofthisworkIwanttothank,againDr.B.Schillinger,Dr.
W.WaschkowskiandN.Wieschalla.

LastbutnotleastIwanttothankmyparents,mybrother,allmyfriendsandthe
mountainsofSouthTyrol,whosupportedmeallthetime.