Science Arts & Métiers (SAM)
is an open access repository that collects the work of Arts et Métiers Institute of
Technology researchers and makes it freely available over the web where possible.
This is an author-deposited version published in:
https://sam.ensam.eu
Handle ID: .
http://hdl.handle.net/10985/18176
To cite this version :
Anne-Laure LEBAUDY, Raphaël PESCI, Boris PIOTROWSKI - Multilayer CdHgTe-based infrared
detector: 2D/3D microtomography, synchrotron emission and finite element modelling with stress
distribution at room temperature and 100 K - Materialia - Vol. 9, p.100511 - 2020
Any correspondence concerning this service should be sent to the repository
Multilayer
CdHgTe-based
infrared
detector:
2D/3D
microtomography,
synchrotron
emission
and
finite
element
modelling
with
stress
distribution
at
room
temperature
and
100
K
A.L.
Lebaudy
∗,
R.
Pesci
,
B.
Piotrowski
ENSAM-Arts et Métiers ParisTech, LEM3 UMR CNRS 7239, 4 rue Augustin Fresnel, 57078 Metz Cedex 3, France
a
b
s
t
r
a
c
t
ThemechanicalbehaviourofaCdHgTe-basedinfrareddetectorwasevaluatedafterprocessingatseveral temper-aturestodeterminetheimpactofthermomechanicalloadingonresidualstressandreliability.Thearchitecture ofthedetectorwasfirstentirelycharacterized,relyingonSEM,X-raymicrotomographyanddiffractionanalysis, inordertogetthenature,themorphologyandthecrystallographicorientationofalltheconstitutivelayers, andinparticulartheindiumsolderbumps.Theresultsnotablyshowedtheunexpectedsinglecrystalaspectof theindiumbumpswitharepeatabletruncatedconegeometry.Toobtainthethermomechanicalresponseofthe structureafterprocessingandintherangeofoperatingtemperatures(from430Kto100K),a3DFiniteElement modelwasthendeveloped.Asexpected,thenumericalresultsshowedastressgradientevolutioninthe struc-turefromhightolowtemperatures,withhighlocanjvvlstressaround30MPaintheCdHgTeat100K,mainly duetothethermalexpansioncoefficientmismatchbetweenthedifferentlayers.Theyhighlightedthesignificant influenceofthegeometryandthesinglecrystalnatureofthebumpsaswellasthebehaviourlawofthedifferent materials.
1. Introduction
Thedemandtoincreasefunctionalityandperformanceofelectronic deviceshasledtothegradualminiaturizationofelectronicpackaging. Theflipchipprocesswherethechipisassembledfacedownontothe circuitboardisidealforsizeconsiderations[1]:thisarchitecturedoes notneedextracontactareaonthesidesofthecomponentandthe con-nectionlengthisminimised,whichisrelevantforhighfrequency ap-plications.Intheinfrareddetection field,theflipchiptechnologyis exploitedforseveralphotodiodecircuitsin InSb,GaN, CdHgTe dedi-catedtodifferentcutoff wavelengths.Thepackagingiscomplexand con-tainsseveralmaterialssuchasmetal,semi-conductororunderfillresin, whichhaveverydifferentmechanicalbehaviours(thermalexpansion, elasticconstants,amongothers).Theinterconnectsizedecreasingand theglobalarchitecturecomplexificationleadtosevere thermomechan-icalloadingsduringthemanufacturingstepsandrunningconditions: theycaninduceresidualstress,mainlycomingfromthemismatchin thethermalexpansioncoefficient(CTE)ofthevariousmaterialsused
[2],warpageormechanicalfailure[3]inthemicroelectronic assem-bliesandaffecttheirreliability.Suchresidualstresseshaveoftenbeen observedandquantifiedinthinorthickplates,forexampleafter weld-ing[4,5],withagreatimpactoftensilevaluesonthestructureintegrity
orfatiguelimit,sothattheyshouldberationalised.Byapproximating thestructuretoathinfilmandtheStoneyformulation[6]usingthe curvatureofthesubstrate,itispossibletocalculatetheresidualstresses inthechipabovebyneglectingtheinterconnectionlayer.Thismethod isverylimitedanddoesnotallowforanestimationoftheevolution ofstressthroughallthelayers.Someexperimentaltechniquessuchas Ramanspectroscopy[7], X-raydiffraction[8]orKossel Microdiffrac-tion[9]can alsobe used, buttheyarelimitedin scopedue tolocal heating,lowelectrical conductionorthethicknessofthelayers.The mostcommonapproachtoestimatethestressvaluesistouseFinite El-ementMethod(FEM)andsimulatethestackofchipandconnections underthermalloading.Ifliteraturepresentssomepapersabout2D rep-resentations[10,11],theincreasingnumberofsolderbumpsinthe as-sembliesledtothedevelopmentof3Dmodels[12].Theresultsenable toobserveastressgradientthroughoutthearchitectureandinall di-rections.However,manyhypothesesareoftenformulatedconcerning thematerialproperties,withthedifferentlayersincludingsingle crys-tals usuallyassumedtobe isotropic,andeitherplanestressorplane strainconditionsareconsidered[13].Inthecaseoftheinterconnection zone,theshapeofthesolderbumpscanberoughlyassimilatedto cylin-dersor truncatedspheres whilethematerialanisotropyisneglected, andtolimitthenumberofelementsinmodels,manyauthorspropose toreplacetheinterconnectionlayerwithahomogeneouslyequivalent material[14].Theseassumptionshelptodecreasethecomputingtime andtomodelthewholeassembly,butitmakesitimpossibletorender
Fig.1.ArchitectureoftheIRdetectorconsidered.
astressgradientineachlayer[15]andtheycanleadtoa misestima-tionof thestressvalues.Today, themainchallengeforstress analy-sisinflipchippackagesisthesignificantgradientatthemicrometric scale.Theinterconnectionlayercanbeverythincomparedtothetwo interconnectedcircuitswhich canintroducehighstrain inthesolder bumps.Moreover,themechanicalbehaviouroftheconstitutive materi-alsisoftennotwellknown.Themodellingscalehastobeadaptedto therequiredobservationsandtheimpactoftheinputdataontheresults mustbeevaluatedtodeterminewhatthemostinfluentialparameters are.
Thepresentstudyaimsatestimatingthelevelofstressthatappears inaninfraredCdHgTe-baseddetectorafterprocessingatroom temper-atureandat100Kduringservicelife.A3DFEMmodelisdeveloped toprovide abetterunderstandingofthestrain/stressgradientinthe structureunderthermalloadingandtoimprovethereliability.The rel-evanceofsuchamodelatalocalscalestronglydependsontheaccurate knowledgeoftheactualarchitectureofthecomponent.First,a charac-terisationsteprelyingonSEMcoupledtoEDSandEBSD,X-ray microto-mographyandsynchrotronemissionhasbeencarriedouttoapprehend thegeometry,thechemicalcompositionandthecrystallographyofall thelayersofthestructure,especiallytheindiumsolderbumpsinthe in-terconnectionlayer,whichseemstobethemostcriticalinthestructure, consideringtheobservedfailuremodes.Fromtheseconsiderations,a3D FEMmodelwasdevelopedandthermalloadingsrepresentativeofthe componentlifecyclewereapplied.Thesesimulationsmadeitpossibleto showthestressgradientsarisingintheseverallayersatroom tempera-turebutalsoatoperatingtemperature(77K).Aparametricstudyonthe inputparameterswasthencarriedouttodeterminethemostinfluential variablesonthestressvalues.Finally,thestressdistributionwas com-paredtoexperimentalvaluesobtainedat77Kduringapreviouswork usinganX-raydiffractionmethodforsinglecrystal.
2. Experimental procedure
2.1. Presentationofthedetectorarchitecture
Thisworkconsidersaflipchipinfrareddetector(midwave-infrared MWIR)witha30μmpitch320×256I/OpixelsIRfocalplanearray.The correspondingarchitectureispresentedinFig.1:aCdHgTethinlayer (detectorcircuit)isinterconnectedtoasiliconsubstrate(readout cir-cuit)throughindiumbumpsandanepoxyunderfill.
Thisarchitectureleadstosomesymmetrywithorthotropic proper-ties:theindiumsolderbumpsarealignedparallel/perpendiculartothe edgesofthedetectioncircuit.Fourmainprocessstepsarenecessaryto manufacturethedetector(allstepsaredescribedin[16]):theindium bumpdepositingonthesiliconcircuit,thehybridisationoftheCdHgTe detection circuit,theunderfilling stepandthefinaldetectioncircuit thinning.Duringthehybridisationstep,heatinguntil430Kisapplied (aboveindiummeltingtemperature);thenbothmechanicaland chemi-calpolishingsofthedetectorcircuitareperformedtokeeptheactive de-tectionCdHgTelayerverythincomparedtothethicknessofthesilicon circuit.Duringmanufacturingandservicelife,thecomponentis there-foresubjectedtointensetemperaturevariationsfrom430Kto77K.The CTEmismatchbetweentheCdHgTedetectorchip,theinterconnection layerandthesiliconcircuitismainlyresponsibleforresidualstress initi-ationandwarpageintheassemblythatcanleadtoseveralfailuremodes suchassolderbumpdamageorcracksintheCdHgTechip.Inorderto better understandtheappearanceof residualstress,especiallyin the detectioncircuitandthesolderbumps,thegeometryandthematerials constitutiveofthedifferentlayershadtobepreciselyknowntobuild thefiniteelementmodel.Acompletecharacterisationoftheassembly wasthereforecarriedoutandparticularemphasiswasputonthe inter-connectionlayerandthesolderbumps.
Fig.2. (a)Solderbumpsvisualisationbymicrotomography(b)Exampleofapost-processingofthestackwithImageJ® (c)Solderbumprepresentedasatruncated cone.
Table1
Indiumsolderbumpaveragethicknessinboththecentreandtheedgesofthecomponent.
Mean value Standard deviation Solder bump thickness (μm) Component centre 6.52 6.54
Component edge 0.11 0.13 Diameter in contact with CdHgTe (μm) Component centre 17.4 0.16 Component edge 16.8 0.19 Diameter in contact with silicon (μm) Component centre 12.8 0.12 Component edge 12.3 0.14 2.2. Geometricalcharacterisation
WhilethegeometryofthesiliconandCdHgTecircuitsiswellknown, themorphologyoftheinterconnectionlayercomposedofsolderbumps inanepoxymatrixstillremainstobe confirmedafterthe hybridisa-tionstepbecausevisuallyinaccessible.Inordertoaccuratelymodelthe structure,effortshavebeenmadeinX-raymicrotomographytoobtain theprecise3Ddescriptionofthesolderbumpsinthecentreandonthe edgesoftheassembly.Ithastheadvantageofbeingnon-destructive, whileachievingaveryacceptablespatialresolution.Thismethod ex-ploitstheselectiveabsorption propertiesof theX-raysbythematter toobservethe sampleinside[17]: ina tomograph,thesampleis il-luminatedwithanX-raysourceandtheintensitymodulationsofthe transmittedbeamarerecoveredas animagebya 2Ddetector. Dur-ingan acquisition cyclewhose duration varies from a fraction of a secondtoseveral hours,some projectionsarecollected accordingto differentangularpositions ofthesampleinrotation.Thedistanceto thedetector,thechoiceoftheacquiredimagenumberaswellasthe anglesteparerealisedaccordingtothedesiredresolution.The recon-structionof3D objectsismadefrom theangularprojections:a grey levelis associatedwitheach voxelof thevolume,whichinforms on theattenuationcoefficientof thematter. Theanalyses wererealised
using aPhoenixv|tome|xm 300/180tomographfromGeneral Elec-tricMeasurementandControlequippedwitha16 megapixel Dyn41-100 detector.Thetubeof thesetup canreach anacceleration volt-age of 180kV andgenerate anX-ray beamallowinga veryfine de-tail detection (downto0.5μmvoxel size).Without any cutting,the analyses wereperformed witha1μm3 resolution andled toagood
estimation of the solder morphology andrepartition. To be statisti-cally representative, a large volume of thesample, approximatively 2.0×2.0×1.0mm3 was analysed,corresponding to a field of about
4500soldersbumps,bothinthecentreandattheedges ofthe com-ponent.Afterthresholdapplication,the3Dvisualisationofthesolder bumpswiththesoftwareMyVgl® wasveryprecise,asshowninFig.2
(a).
TheopensourceimageJ® softwarepresentedin[18]wasusedfor partofthepost-processingintheinterconnectionzone:anexampleis giveninFig.2(b).Byaveragingsolderdiametersinallimages,itwas possibletovisualizethetypicalgeometryofanindiumsolderbump:the detaileddimensionsaregiveninTable1,withsimilarresultsinboththe centreandtheedgesoftheIRdetector.Thefinal3Drepresentationcan beeasilyassimilatedtoatruncatedcone(Fig.2(c)),withmaximumand minimumdiametersincontactwiththeCdHgTecircuitandthesilicon substrate,respectively.
Fig.3.CrystallographicorientationofthesiliconandCdHgTesinglecrystals.
2.3. Crystallographicanalyses
Afterthegeometriccharacterisationoftheindiumsolderbumps,it wasnecessarytodeterminethecrystallographicnature ofeach crys-tallinelayerof thestructurein order toidentifytherelevant consti-tutiverelation of each material tosupply thefinite element model. ElectronBackscattereddiffraction(EBSD)analyseswereperformed us-ingaJEOL7001F FEGSEMequippedwitha CCDcamera from Ox-fordInstruments.The CdHgTeandthesilicon wereeasily accessible withtheelectronbeambecausetheyhademergentsurfaces.The anal-ysesestablished that they weretwo single crystals(111) and (100) oriented,respectively,asshowedinFig.3,withamisorientationless than1°.
Somecomponentswerecutandpreparedwithspecifictechniquesin ordertoaccesstheinterconnectionzone,especiallytheindiumsolder bumps.Thesectionswerefirstcarefullyprepared(sawingand polish-ing)usingaLeicaEMTXPfollowedbyionicpolishingwithaLeicaEM TIC3X.Thislastpolishingtechniqueprovidedidealsurfaceconditions forperformingEBSDmappingswithoptimalindexingrate.The equip-mentusedallowedioncross-sectioncuttingatroomtemperatureorat lowtemperaturewhenthesetupwassuppliedwithliquidnitrogen.Due tothelowmeltingtemperatureofindiumanditsdiffusionproperties, theionicpolishingwascarriedoutat153Kwithanacceleration volt-ageof10kVandperformedinthecentreoftheassembly.Then,phase andorientationmappingswereconductedonseveralsolderbumpsin theionicpolishedarea.EBSDmappingswereperformedwithabeam energyof20kV,withastepsizeofroughly0.5μm;theywerethen post-processedusing theChannel5suite. Intotal,adozen indiumsolder bumpsweremapped,showingthattheywereallsinglecrystalswith randomorientation,asshowninFig.4.TheEBSDorientationimages arecolour-codedtoindicatethegrainorientationgivenbythestandard triangle(tetragonallattice).
Inordertoconfirmtheseobservationsconsideringagreater num-berofsolderbumps,ringdiffractionexperimentswerecarriedouton thebeamlineID11attheESRF(EuropeanSynchrotronRadiation Facil-ity)inGrenoble,France.Asexploitedby[19,20],thismethodallowsto identifytheseveralphasespresentinasample.Themeasurementswere performedwitha78.5keV(𝜆=0.158Å)monochromaticX-raybeamin transmissionmode.Aschematicrepresentationoftheexperimental de-viceisshowninFig.5(a).The0.2×0.2mm2incidentbeam
correspond-ingtoafieldof about40 indiumsolderbumpsentered normallyto thespecimens,formingcompleteDebye–Scherrerringsfromtheseveral crystallographicphasesofthestructure.The2Ddiffractionringswere recordedbyaFrelon2DCCDcamerawitharesolutionof2048×2048 pixels (47.2×47.2 μm2 pixel size). A distance sample/camera of
288.2mmwaschosentorecoveramaximumofdiffractionringswhile keepinganacceptableresolutionfortheirproperindexing.Thesamples
Fig.4. EBSDmappingsshowingthemeancrystallographicorientationofthree indiumsolderbumps.
wereplacedinsuchawaythatthebeamwasfocusedinthecentreofthe assembly,"full-chip"zone,toacquirethediffractionringsofthe detec-tioncircuit,theinterconnectionzoneandthereadingcircuit simultane-ously.TheideawastopenetratefirstthesiliconsubstratewiththeX-ray beamandexitthroughthethinneddetectioncircuitsidetominimizethe amountofmaterialcrossedbythediffractedbeamcomingfromindium soldersandlimitthediffractionringattenuation.Anexampleofa gen-erateddiffractionpatternacquiredbythecameraisshowninFig.5(b). The diffraction patternsobtained haveboth diffraction spotsand completerings.Eachringcorrespondstodiffractingplanes{hkl}ofa definitephase:itiscomposedofseveralindividualspotsdiffractedat thesameBraggangle,eachspotbeinggeneratedbyagrainina diffrac-tionposition.Itisthereforepossibletoidentifythephasespresentinthe assembly,theircrystallographicstructureaswellastheirpolycrystalline ormonocrystallinenature.Giventhesmallbeamsizeused,acomplete ring willbe characteristicofapolycrystallinephasewithsmallgrain size.InFig.5(b),theveryintensespotscorrespondtosiliconand Cd-HgTesinglecrystals;othersareduetoremainingphases.Inorderto in-dexthediffractionringsandassociatethemtotheirrespectivephases,it isnecessarytoproceedtoapreliminarysteptodeterminetheir associ-ateddiffractionangles.TheFit2Dsoftwarewasusedforpost-processing tointegratetheDebye–Scherrerringsandgeneratetheequivalentofa 2𝜃 scan[21].Theinputdataincludethepixelsizeofthecamera,the wavelengthoftheradiationconsideredaswellasthecamera-to-sample distance.Theseparametersarethenrefined;theuserhasthepossibility tomanuallycorrectthecentreofthebeamandthenon-orthogonality ofthedetectorwithrespecttotheincomingbeam.Thisstepcanalso be performedmoreaccuratelyusing acalibrationmaterialsuchasa nanocrystallinepowder[22].Inourcase,thecalibrationisbasedon diffraction peaksof thesilicon,thelessstrainedlayer(becausevery thickcomparedtoothers).AsshowninFig.6,severaldiffuse continu-ousdiffractionlines/diffractionringscanbeattributedtoverythin poly-crystallinelayersintheassemblysuchastheUnderBumpMetallization (UBM).Thediffractionringsofindiumhavealsobeenidentified accord-ingtotheirdiffractionangle𝜃 associatedtothe{hkl}diffractingplanes (surroundedlines/diffractionringsinthe2𝜃 scanpresentedFig.6):they arediscontinuous,showingthepresenceofonlyafewgrainswith dif-ferentorientations.
This resultconfirmstheconclusions from EBSDmappingson the cross sectionsanalysed:each indiumsolderbumpis asinglecrystal. Thiscanbeofcrucialimportancesincethetetragonalcrystalofindium leadstoahighlyanisotropicthermomechanicalbehaviouralongthe dif-ferentcrystallographicaxes.Thefactthatitispresentinasinglecrystal
Fig.5. (a)Diffractionringsetup(b)Diffractionpatternofthecomponentacquiredbythe2Ddetector.
Fig.6. Ringdiffraction:identificationofthediscontinuousdiffractionringsof indium.
formwithrandomorientationsintheIRdetectorcangeneratea me-chanicalresponsethatcandiffergreatlyfromoneindiumsolderbump toanother(onepixelofthedetectortoanother).Consideringthese var-iousgeometricandcrystallographicobservations,a3Dfiniteelement modellingwasbuilttosimulatethecompletestressdistributioninthe IRdetector.
3. Finite element modelling 3.1. Architectureandmeshing
The3DFEMmodeldevelopedinthisworkaimstoshow,atthescale ofthelayerorofasolder(thatisapixeloftheIRdetector),thepresence ofstrain/stressgradientsunderthermalloadingsandthereforeto eval-uatetheimpactoftemperaturevariationsonthestresslevelreached
Fig.7. Boundariesconditionsconsideredforthe3Dmodel.
inCdHgTe,indiumandsilicon.Itmustbeasrepresentativeofthe ac-tualgeometryoftheassemblyaspossiblewithoutnecessarilymodelling theentirestructure,inordertohighlighttheeffectofstressgradients onaverylocalscalebyconsideringmaterialswithveryhigh hetero-geneousproperties.Theunderbumpsmetallizationbetweenthe detec-tion/readoutcircuitsandindiumsolderwasignoredinordertosimplify themodelandanidealbondingalongthematerialinterfaceswas as-sumed.Onlythethreemainlayersoftheassemblywereconsideredin thesimulations:
- ThedetectioncircuitinCdHgTe.
- The interconnection zone consisting of a field of indium solder bumpsembeddedinanepoxymatrix.
- Thesiliconsubstrate.
Thesolderbumpsweremodelledastruncatedconesasobservedin theX-raymicrotomographyreconstructions.
ThemeshoftheCdHgTelayerandtheinterconnectionzone(solders andepoxymatrix)wasrealisedusinglinearhexahedralelements.Atthe interfacewiththeinterconnectionzone,thefirst50μmofsiliconwere alsofinelymeshedwithlinearhexahedralelements;then,adecreasing meshwasappliedusingquadratictetrahedralelementsinorderto de-creasethecalculationtime.Thismeshallowedtoobtainreliablestress
50 100 150 200 250 300 350 400 450
Epoxy element
deactivation
Epoxy element
activation
« strain free »
Underill step
Underill
temperature
er
ut
ar
e
p
m
e
T)
K(
Room temperature
Melting temperature
of indium
Hybridisation
100 K
Operating
temperature
Fig.8. Thermalloadingsconsideredinthemodel.
distributionatthesiliconsurface:indeed,this layerismore than40 timesthickerthanthecombineddetectioncircuitandinterconnection layer,sothatitisrelevanttoconsiderthatitisonlyunderloadingat itsextremesurface.Ameshconvergenceanalysisshowedthatasizeof 1.3μmisnecessarynottoinfluencethestressvalues.
3.2. Geometrichypotheses,boundaryconditionsandthermalloadings
Duringhybridisation,beforethinning,thedetectioncircuitconsists oftwomainlayers,theCdHgTeepitaxiallayerandaCdZnTesubstrate
[16].Inordertominimisethenumberofnodesinthemodel,the pres-enceoftheCdZnTelayerwasnottakenintoaccountintheassembly. Thethermalcycleisthereforeappliedtothegeometryofthefinal detec-tioncircuit,sowithonlytheactivelayerofCdHgTe.Thisisarequired assumptiontodecreasethenumberofelementstomeshthesilicon sub-strate,whichissignificantlythickerthantheCdHgTe.Theboundary conditionsofthemodelareshowninFig.7.Thestructurehastwoplanes ofsymmetry:conditionsUx=0andUy=0respectivelyappliedontwo facesofthemodel(allthelayersconsidered).Then,theunderfillstep inducestheformationofanepoxy"beading"surroundingandrisingon theflanksofthedetectioncircuit:inordertotakeitintoaccount (con-sideringitsthicknessandthehighdifferenceofthermalexpansionwith theCdHgTe layer),thedisplacements of thelatterwerealsolimited (Ux-CdHgTe=0andUy-CdHgTe=0).
Althoughthethermalconductivityisdifferentforalltheconstitutive materials,theassemblyisassumedtohaveahomogeneoustemperature ateachcomputationtime.TheadoptedthermalcycleisshowninFig.8. Inordertosimulatetheunderfillstep,itwasdecidedtodeactivatethe epoxyelements("strainfree")atthebeginningofthecalculationandto reactivatethemat353K(Abaqusoption),theglasstransition temper-atureofepoxy,abovewhichitwasconsideredthatepoxydidn’thave anyimpactonthemechanicalbehaviourofthestructure.
Table2
ElasticconstantsofCdTeandHgTe.
References C 11 C 12 C 44 Uncertainty Temperature (K) CdTe [25] 53.51 36.81 19.94 + / − 0.2% 298 [26] 61.5 43.0 19.6 + / − 18% 77 [24] 53.8 37.4 20.2 + / − 3% 298 56.2 39.3 20.6 + / − 3% 77 HgTe [27] 54.8 38.1 20.4 + / − 0.5% 290 58.7 41.0 21.7 77 [28] 53.61 36.60 21.23 + / − 0.8% 298 58.63 40.59 22.41 78 [29] 50.8 35.8 20.5 200 3.3. Materialproperties
Thedetectioncircuitconsistsofa(111)-orientedCdHgTesingle crys-tal.WhileCdHgTehashighperformanceinthefieldofinfrared detec-tionduetoitsopticalandelectricalproperties,ithasmuchlower me-chanicalpropertiesthanothermaterialsusedforthisapplication.Itis generallyconsideredasa"soft"material[23],withaverylow mechan-icalstrength.Duetothedifficultyinmeasuringtheelasticproperties of athin-film material,fewdataaboutthemechanicalbehaviourare availableinliterature:onlyvaluesfortheCdTeandHgTemanufactured withmethodsallowingtoobtainlargesinglecrystalscanbefound.In thisstudy,thestiffnesstensorsatroomtemperatureand77Kproposed by[24]fortheCdTehavebeenused(Table2).
TheplasticbehaviourofCdHgTeisalsonotwellknown.Theyield stresshasbeenestimatedat12MPabyBalletetal.[30].Thisvaluecan beimpactedbyapossiblediffusionofthezincpresentintheCdZnTe epitaxialsubstratetotheCdHgTeactivelayerbeforethinning.With4% zincintheCdHgTe[31],theyieldstresscanbemultipliedby4for ex-ample,leadingtoavalueofabout60MPa;ithasbeenfixedat30MPa inourmodel.TheCTEproposedbyWilliamsetal.[32]andBagotetal.
Fig.9. EvolutionoftheCTEforindiumsinglecrystalasafunctionoftemperature[36].
Table3
Elasticconstantsofindiumsinglecrystalaccordingto[37,38,39].
[37] [38] [39] Elastic constants 300 K 77 K 298 K 300 K 351 K 371 K 391 K 412 K 422 K 429.7 K C 11 45.35 52.6 445 45.1 43.3 425 42.1 41.4 41.1 40.9 C 33 45.15 50.8 444 45.3 43.5 42.9 42.3 41.7 41.3 41.1 C 44 6.51 7.6 6.55 6.53 6.3 6.17 6.03 5.89 5.83 5.78 C 66 12.07 16 12.2 11.9 10.7 10.3 9.88 9.46 9.21 9.05 C 12 40.01 40.6 39.5 39.7 39.8 39.6 39.5 39.1 39.2 39.1 C 13 41.51 44.6 40.5 41.1 40.9 40.4 40.3 40.1 39.9 39.5
[33]correspondingtoatemperaturerangefrom293Kto693Kandfrom 0Kto80Krespectively,havealsobeenused.Indiumischaracterised byalowmeltingpoint,asmallyieldstressandaveryhighductility at293Kand77K.ThesepropertiescombinedwithitshighCTE com-paredtoCdHgTeandsiliconareveryinterestingfortheinterconnection layerwhenoperatingatlowtemperature.Themechanicalpropertiesof polycrystallineindiumarewellknown:thebehaviourlawsfrom4Kto 295KhavebeendeterminedbyReedetal.[34],confirmingthehigh levelofplasticityandtheCTEgivenbyTouloukian[35].Onthe con-trary,fewdataaboutindiumunderitssinglecrystalformareavailable, buttheypointtoaconsiderableanisotropyduetothetetragonallattice withanotable ratiobetween thetwolatticeparameterscanda.For example,Collinsetal.[36]showedthattheCTEisverydifferentalong theaxisofthecrystalconsidered.Fig.9presentstheresultsobtained fordifferentauthors,withIn(ǁ),theCTEmeasuredalongthequadratic axisofindium,In(⊥),thatmeasuredaccordingtoaplanenormaltothe quadraticaxis,andIn(av),theaverageCTE.
The6elasticconstantsofindiumforseveraltemperatureswerealso determinedbyseveralauthors(Table3).Severalsimulationswerethen
Table4
MechanicalpropertiesofEpotek301–2asafunctionoftemperature.
Temperature (K) Young modulus (MPa) Poisson’s ratio
293 3664 0.358
250 4109 0.365
200 4474 0.349
150 5745 0.334
100 6993 0.350
launchedtocomparetheresultsconsideringthepropertiesofindiumas bothpolycrystalandsinglecrystal.
Epoxypolymers(commonlyknownasepoxies)arewidelyusedin microelectronicsforthemanufactureofflipchipassembliestofillthe interconnectionzonebetweenthechipanditssubstrate,thatistosayto fillthegapbetweenthesolderbumps.ThegradeusedintheIRdetector isEpotek301–2andwascharacterisedbyCeaseetal.[40]for temper-aturesrangingfrom300Kto100K.ThevaluesofYoung’smodulusE andPoisson’sratioobtainedarepresentedinTable4.
Fig.10. Sectionalviewofthestressdistributioninthedifferentlayersoftheassemblyat293K.Therightpartisazoomoftheframedzonewithafinerscale.
Atroomtemperature,Epotek301–2hasapurelyelasticbehaviour withayieldstress/strength around40MPa.Itwillthenvary signifi-cantlywithtemperature,asmostofthermoplastics,withaglass transi-tiontemperatureTg=213K;themanufacturerindicatesthattheCTEis 61.10−6K−1belowthistemperatureand180.10−6K−1beyondit.
Thesubstrateofthecomponentisanorientedsiliconsinglecrystal (100),thebehaviourof whichis elastic,witha yieldstress/strength closeto130MPa.TheCijklelasticityconstantsconsideredforthemodel
camefrom[41]andtheCTEasafunctionoftemperaturefrom[42]. Forallthesinglecrystalmaterials,theelasticconstantsare imple-mentedintheFiniteElementmodelinatwo-step-process.First,basing ontheexperimentalresults,thecrystallographicorientationofCdHgTe, siliconandtheindiumbumpsareidentifiedandtheelasticconstantsare giveninthecrystalbasisforeachmaterial.Then,theorientationofeach crystalisintroducedintheFiniteElementmodel(whereaglobal coor-dinatesystemimposestheglobalorientation)throughaspecific calcu-lationthatactsasatransfermatrixbetweeneachcrystalandtheglobal coordinatesystem.
4. Results and discussion
4.1. Simulationswithpolycrystallineindiumsolderbumps
The3Dmodellingstrategyallowstovisualizethestressgradients alongthethickness,inallthelayersoftheassembly,througha well-defined cuttingplane.Themaindrawbackof theFEMsimulationsis thecomputationaltime.Toseethestrain/stressgradientatthesolder scale,itisnotpossibletomodelthetotalityofthestructure(morethan 80,000solderbumps).Firstsimulationswith36,144,324and1296 solder bumps werelaunched andshowed that144 were representa-tiveofthewholeassembly,withconvergingresults.Inthispaper,all therepresentations ofthestress variationappearingintheassembly werevisualizedusingdiagonalsections.Onlyhalfofthearchitecture is presented in thefollowing figuresshowing theedgeandthe cen-treofthestructureduetothesymmetryconditions.At293K,we no-ticethatintheCdHgTelayer,thecomponentsofthestresstensor tak-ingthehighestvaluesare𝜎xxand𝜎yy,closeto30MPatheedgesand
zz yy yz xz xy
100 K
xx 10 MPa 30 MPa 20 MPa -20 MPa -10 MPa -30 MPa 0 MPa 10 MPa 30 MPa 20 MPa -20 MPa -10 MPa -30 MPa 0 MPa 10 MPa 30 MPa 20 MPa -20 MPa -10 MPa -30 MPa 0 MPa 10 MPa 30 MPa 20 MPa -20 MPa -10 MPa -30 MPa 0 MPa 10 MPa 30 MPa 20 MPa -20 MPa -10 MPa -30 MPa 0 MPa 10 MPa 30 MPa 20 MPa -20 MPa -10 MPa -30 MPa 0 MPa 4 MPa 12 MPa 8 MPa -8 MPa -4 MPa -12 MPa 0 MPa 4 MPa 12 MPa 8 MPa -8 MPa -4 MPa -12 MPa 0 MPa 4 MPa 12 MPa 8 MPa -8 MPa -4 MPa -12 MPa 0 MPa 4 MPa 12 MPa 8 MPa -8 MPa -4 MPa -12 MPa 0 MPa 4 MPa 12 MPa 8 MPa -8 MPa -4 MPa -12 MPa 0 MPa 4 MPa 12 MPa 8 MPa -8 MPa -4 MPa -12 MPa 0 MPaComponent edge
Component centre
Fig.11. Sectionalviewofthestressdistributioninthedifferentlayersoftheassemblyat100K.Therightpartisazoomoftheframedzonewithafinestscale.
Component edge
Purely elastic polycristalline indium σyy
σxx
Component edge Component centre
σyy
Reference model
Component edge Component centre
20 MPa 60 MPa 40 MPa -40 MPa -20 MPa -60 MPa 0 MPa
CdHgTe with a yield stress of 60 MPa
10 MPa 30 MPa 20 MPa -20 MPa -10 MPa -30 MPa 0 MPa 10 MPa 30 MPa 20 MPa -20 MPa -10 MPa -30 MPa 0 MPa Out of scale
Out of scale Out of scale
a) c) b) Component centre σxx σxx σyy
Fig.12. Comparisonofthenumericalresults(a)Standardmodel(reference)(b)CdHgTewithayieldstressof60MPa(c)Polycrystallineindiumwithapurelyelastic behaviour.
around10MPainthecentre(Fig.10);azoomoncertainzonesmakes itpossible tohighlightgreater tensile values inthe areasin contact withthesolderbumps.Theepoxyisinuniformbiaxialtensionaround 15MPaandtheindiumsolderbumpsarealsoinslighttensionfor𝜎xx
and𝜎yy(2MPa);thesiliconisinbiaxialtensionwithvaluesranging from4MPato12MPaundertheindiumsolderbumps.Finally,itcanbe notedthattheaffectedsilicondepthunderthesolderisapproximately 8μm.
Whencoolingat100K,interestingresultscanbenoted(Fig.11):
- Biaxialtensionisintensifiedalong𝜎xxand𝜎yyintheCdHgTe,with valuesfrom30MPaabovethesolderbumpsto18MPainthe inter-solderareas;theepoxy shows somevalues outsidethescale pre-sented(greyarea:35MPaonaveragefor𝜎xxand𝜎yy).Thisis con-sistentwiththeexperimentalstressvaluesobtainedin[43],inwhich theauthorsusedX-RayDiffraction(XRD)witha0.8mm2beamto
makesome mappingsfromthecentretotheedgeoftheCdHgTe chip:inparticular,themeasurementsat77Kshowedtensionstress valuesuptoabout30MPa.
3 MPa 9 MPa 6 MPa -6 MPa -3 MPa -9 MPa 0 MPa
Single crystal indium local stress scale
Component edge Component centre
One single crystal indium solder (orientation 1)
10 MPa 30 MPa 20 MPa -20 MPa -10 MPa -30 MPa 0 MPa Out of scale 3 MPa 9 MPa 6 MPa -6 MPa -3 MPa -9 MPa 0 MPa
Single crystal indium local stress scale σxx
σxx
Component edge Component centre
Reference model: elasto-plastic polycristalline indium
10 MPa 30 MPa 20 MPa -20 MPa -10 MPa -30 MPa 0 MPa Out of scale
Component edge Component centre
One single crystal indium solder (orientation 3)
10 MPa 30 MPa 20 MPa -20 MPa -10 MPa -30 MPa 0 MPa Out of scale 3 MPa 9 MPa 6 MPa -6 MPa -3 MPa -9 MPa 0 MPa
Single crystal indium local stress scale σyy X Y Z Global Abaqus coordinate system
Crystallographic orientation of one single crystal indium solder bump surrounded by polycrystalline solder bumps
Orientation 1 Orientation 2 Orientation 3
c a b c a b
Orientation 2 with a 45 rotation around a axis σyy
Component edge Component centre
One crystal indium solder (orientation 2)
10 MPa 30 MPa 20 MPa -20 MPa -10 MPa -30 MPa 0 MPa Out of scale 3 MPa 9 MPa 6 MPa -6 MPa -3 MPa -9 MPa 0 MPa
Single crystal indium local stress scale σxx
σyy
Fig.13. Influenceofoneindiumsolderbumpmodelledassinglecrystal:differentcrystallographicorientationsandthermalexpansioncoefficientsconsidered (100K).
- Thesolderbumpsattheedgeoftheassemblyareintension,while thoseinthecentreareincompressionwithtensionintheirperiphery (forboth𝜎xxand𝜎yy).Thisparticularstressstateisrelatedtothe differenceintheCTEbetweentheepoxymatrixandindium(61.10−6
K−1and30.10−6K−1,respectively).
- Thestressvariationsinthesiliconareenhancedwiththeapparition ofcompressionvaluesofabout−7MPaaccordingto𝜎xxand𝜎yyin thezonesundertheepoxy.
- Thelayerwiththemostimportantvaluesin𝜎zzisthe interconnec-tionzone,withsolderbumpsincompressionandtheepoxyin ten-sion.
- Thevaluesof𝜎yzand𝜎xzareamplified,especiallyintheCdHgTe andthesiliconbetweenthesolderbumps.Theepoxyisalsoslightly affected.
- Finally,thevaluesof𝜎xyaregenerallylow:onlytheepoxylocally reachescompressionlevelsaround−12MPaintheareasincontact withtheindiumsolderbumps.
Theresultsshowthegreatinfluenceof themechanicalproperties andtheCTEmismatchbetweeneachlayerofthestructureonthestress values.Thestressgradientscanbeappreciatedattwoscales:between theedgeandthecentreoftheassembly,andatthelocalscale consid-eringonesolderbump/pixeloftheIRdetector.Thedifferencebetween theedgeandthecentreistypicallyduetotheCTEmismatchbetween theCdHgTeandthesilicon.Indeed,theCdHgTedetectioncircuitcanbe assimilatedtoathinfilmcomparedtothesiliconcircuitthickness.The CdHgTehasaCTEtwicesmallerthanthesilicon,whichexplainsthe in-ducedtensilestressstateappearingduringcoolingfrom430Kto100K. Thestressgradientishigherat293Kthanat100KbecausetheCdHgTe hasstarted toyield,sothatthebiaxialtensile stateof CdHgTetends tobeuniformatlowtemperature(around30MPaaccordingto𝜎xxand
𝜎yy).Tokeepthestaticequilibriumofthestructureundertheincreasing tensilestateoftheCdHgTelayer,alowcompressivemeanstateappears
inthesiliconsubstrate.Moreover,stressheterogeneitiesinthesolder zoneappearat293Kandareenhancedat100K:thisphenomenon cor-respondstothemismatchbetweentheCTEofCdHgTe,indium,epoxy andsilicon.Thesiliconisincompressionundertheepoxy(−10MPato −20MPa)andintensionunderthesolders(8–18MPa)withinadepth of afewmicrons.Thus,usingapolycrystallinebehaviourlawfor in-dium,heterogeneitiescanalreadybeobservedunderthermalloading, whetheritispixelbypixelorbetweenthecentreandtheedgeofthe detector.
4.2. Influenceoftheinputparameters
Inthissection,thefocusismadeontheinfluenceoftheinput pa-rameters ofthemodelonthenumericalresults.Asexplainedin2.4, oneofthesourcesofuncertaintiestopredictthestressdistributionin theIRdetectoristhelackofknowledgeabouttheCdHgTe thermome-chanicalproperties.Severalsimulationsconsideringthedifferentelastic constantsproposedbyalltheauthors(Table3)haveshownaverylow impactonthestressvaluesinallthelayersat100K(<1%).Thequestion oftheplasticityoftheCdHgTeisalsoveryimportant,especiallydueto thepossiblediffusionofZncomingfromtheCdZnTesubstratebefore thinning.IntheresultspresentedinSection3.1(Fig.12(a)),theyield stressoftheCdHgTewasfixedat30MPa,avalueestimatedfrom lit-eratureandtheresidualstressmeasurementspresentedin[43].Other simulationswithayieldstressof 60MPa werelaunched (Fig.12(b)) andshowedthegreatimpactonthenumericalresultssincethe max-imal stressvalues observedwereof thesameorder ofmagnitude.A strongerstressgradientwasalsoobservedintheCdHgTelayerbetween theedgeandthecentreoftheassembly.Onthecontrary,neither sili-connorindiumwasreallyaffected,duetotheirthicknessandtheirhigh plasticity,respectively.Itisthereforeveryimportanttowelldefinethe mechanicalbehaviouroftheCdHgTeinordertogetstressvaluesas rep-resentativeaspossiblewiththemodel.IntheflipchipFEMfield,most
ofthe authorsmadethehypothesisthat circuitandsoldermaterials couldbeconsideredaspurelyelastic[11,14].Thematerialproperties ofindiumannouncedbyPlötneretal.[44]showedavery lowyield stresswithhighplasticity.Apureelasticlawforindiumwasalso imple-mentedinthemodelforcomparison(Fig.12(c)).Thevaluesof𝜎xxand
𝜎yyinthiscasewerethusverydifferent:thesiliconbelowtheindium solderswasmorestressedwithnegativevaluesoutoftherepresented scale(−30MPa),whiletheCdHgTejustabovewasincompression con-trarytothepreviousresults.Asforindium,itwasinhightensionwith stressvaluesoutofscale.Theseresultswerethedirectconsequenceof theabsenceofplasticityinindium:thestressvaluesincreasedwithout anylimitation.
4.3. Influenceofcrystallographicorientationforindiumsolderbumpsas singlecrystals
SincetheEBSDandXRDresults showedthateach indiumsolder bumpwasasinglecrystal,complementarysimulationswerelaunched consideringthisparticularaspect.Morespecifically,theaimwastosee ifthegreatanisotropyofthecrystallographicorientationandtheCTE ofindiumpresentedbyCollinsetal.[36]couldincreasethestress gra-dientsintheassembly,especiallyatlowtemperature,sincethesolder expansion/contractionunder thermalloadingwassupposedtoaffect thelevelofstressintheotherlayers.Severalsimulationswerelaunched consideringonlyoneindiumsolderbumpasasinglecrystalandthe othersaspolycrystalline.Fig.13presentstheresultsobtainedforthree differentorientationsofonesolderbumpat100K,inparticularthosefor whichthedifferenceismaximised(anisotropyenhancedformechanical propertiesandCTE),whenconsideringthedifferentaxesofthe tetrago-nallatticeofindium.Theyshowaweakimpactonthestressdistribution intheCdHgTejustabove,andiftheindiumsolderbumpsaremostlyin tensionwithlowcompressioninthecentreat293Kand100K,when polycrystallineorsinglecrystalswithorientation1or3,someofthem canbefullyincompressionwithhigherstressvaluesduetotheir crystal-lographicorientationandtheparticularevolutionoftheCTE(especially fororientation2).Itshouldnotedthatthiseffectissimilarwhen consid-eringallthesolderbumpsassinglecrystals,withorientationscloseto orientation2alwaysinducinghigherstress/deformationinthebump. Theselocalvariationsinthestressvalues canleadtopossiblesolder cracksattheinterfacewiththeCdHgTeand/orsiliconlayersand ex-plainthe“randomlysupposed” apparitionofdefectssuchasdeadpixels inthedetector.
5. Conclusions
Inthiswork,thestrain/stressevolutioninanIRdetectorsubjected tothermalloadingsduringitscyclelifehasbeennumericallystudied. Conclusionsofthisstudycanbesummedupasfollows:
- Theexperimentalresultsgaveaverydetaileddescriptionofthe de-tectorstructure.Thetomographyanalysesledtoanexcellent defini-tionofthe3Dsolderbumpgeometryafterthelastprocessstep.EBSD andringdiffractionprovidedafirsttwo-dimensionaldescriptionof thenatureandcrystallographyofthedifferentlayers.Themost im-portantpointsarethepresenceofCdHgTeandsiliconsinglecrystals aswellasthesinglecrystalnatureofindiumsoldersandtherandom repartitionoftheircrystallographicorientation,afeaturethatuntil nowwascompletelyunknown.Theycouldsubsequentlybe imple-mentedinthefiniteelementmodelwiththeiranisotropic mechan-icalproperties,whichwasessentialforthemodeldevelopmentto simulatestrain/stressheterogeneitiesatthescaleofonepixelofthe detector.
- Thefiniteelementmodelwasveryusefultoobservelocal hetero-geneitiesinsidetheCdHgTeandsiliconlayersandtoestimatethe stresslevelineachindiumsolderbump.Heterogeneitiesappeared atthescaleofthesolderbumpat293Kandwereamplifiedat100K.
WhereasthebiaxialtensilestateofCdHgTetendedtobeuniformat lowtemperature(around30MPaaccordingto𝜎xxand𝜎yyconsistent withthevaluesobtainedexperimentallyat77K[43]),siliconwasin compressionunderepoxy(−10a−20MPa)andintensionunderthe solderbumps(8–18MPa)withinadepthofafewmicrons.Thus, us-ingapolycrystallinebehaviourlawforindium,heterogeneitiescan alreadybeseenpixelbypixelwhenapplyingthermalfieldsbetween thecentreandtheedgeofthedetector.Inordertohaveamore pre-ciseestimationofthestressstateineachpixelofthedetector,other simulationswerelaunchedconsideringthesinglecrystalnatureof eachsolderbumpaswellastheinfluenceofthecrystallographic ori-entationandtheanisotropicCTEofindium.Thenumericalresults showedthattheCdHgTelawwasaweakimpacted,butaccording totheircrystallographicorientation,theindiumsolderbumpscould bemoreincompressionat293Kand100K:thisdifferenceinthe stress/deformationcouldleadtopossiblecracksattheinterfacewith theCdHgTeand/orsiliconlayersandexplainsomedefectssuchas thedeadpixelsappearinginthedetectorduringelaborationand ser-vicelife.
Declaration of Competing Interest
Theauthorsdeclarethattheyhavenoknowncompetingfinancial interestsorpersonalrelationshipsthatcouldhaveappearedtoinfluence theworkreportedinthispaper.
Acknowledgements
ThisworkwassupportedbyRegionLorraineinFrancein collabora-tionwithCEA(Commissariatà l’EnergieAtomique).
References
[1] J.H. Lau , Flip Chip Technologies, McGraw-Hill, New York, 1996 .
[2] M. Song , Z. Wei , B. Wang , L. Chen , J.A. Szpunar , Study on copper protrusion of through-silicon via in a 3-D integrated circuit mat, Sci. Eng. A 755 (2019) 66–74 . [3] H.M. Tong , Y.S. Lai , C. Wong , Advanced Flip Chip Packaging, Springer, 2012 . [4] J.Y. Wan , J.W. Jiang , J. Li , G. Sun , D.K. Kim , W. Woo , Weld residual stresses in
a thick plate considering back chipping: neutron diffraction, contour method and finite element simulation study, Mater. Sci. Eng. A 699 (2017) 62–70 .
[5] W. Jiang , W. Chen , W. Woo , S.T. Tu , X.C. Zhang , V. Em , Effects of –low-temperature transformation and transformation-induced plasticity on weld residual stresses: nu- merical study and neutron diffraction measurement, Mater. Des. 147 (2018) 65–79 . [6] G.G. Stoney , The tension of metallic films deposited by electrolysis, Proc. R. Soc. A
82 (1909) 172–175 .
[7] J. Chen , I.D. Wolf , Theoretical and experimental raman spectroscopy study of me- chanical stress induced by electronic packaging, IEEE Trans. Comp. Pack. Techol. 28 (2005) 484–492 .
[8] A. Declémy , P.O. Renault , XRD measurement of thermal strain in insb-based devices, Phys. Status Solidi 204 (2007) 1041–1046 .
[9] R. Pesci , K. Inal , S. Berveiller , E. Patoor , Stress analysis by kossel microdiffraction on a nickel-based single crystal superalloy during an in situ tensile test - Comparison with classical X-Ray diffraction, Mat. Sci. Forum 681 (2011) 1–6 .
[10] S.C. Chen , Y.C. Lin , C.H. Cheng , The numerical analysis of strain behavior at the solder joint and interface in a flip chip package, J. Mater. Process. Technol. 171 (2006) 125–131 .
[11] M.L. Sham , J.K. Kim , J.H. Park , Thermal performance of flip chip packages: nu- merical study of thermo-mechanical interactions, Comput. Mater. Sci. 43 (2008) 469–480 .
[12] X. Cheng , C. Liu , V.V. Silberschmidt , Numerical analysis of thermo-mechanical be- havior of indium micro-joint at cryogenic temperatures, Comput. Mater. Sci. 52 (2012) 274–281 .
[13] Q. Yao , J. Qu , Three-dimensional versus two-dimensional finite element modeling of flip-chip packages, J. Elect. Packag. 121 (1999) 196–201 .
[14] W. Kpobie , N. Bonfoh , C. Dreistadt , M. Fendler , P. Lipinski , 3D modeling of flip chip assemblies with large array and small pitch: validation of the proposed model, in 14th international conference on thermal, mechanical and multi-physics simulation and experiments in microelectronics and microsystems, in: EuroSimE, 2013, pp. 1–6 . [15] X. Zhang , Q. Meng , Y. Lü, J. Si , Assessment of InSb infrared detector arrays assembly
procedure employing ANSYS, Opt. Quant. Elec. 51 (2019) 92 .
[16] P. Castelein , F. Marion , J.L.J.P. Baylet , N. Moussy , O. Gravrand , A. Durand , J.P. Cha- monal , J.L. Destefanis , Megapixel hgcdte mwir focal plane array with a 15-μm pitch, Proc. SPIE 5074 (2008) 5074/1-5074/8 .
[17] E. Maire , J.Y. Buffiere , L. Salvo , J.J. Blandin , W. Ludwig , J.M. Letang , On the appli- cation of X-ray microtomography in the field of materials science, Adv. Eng. Mater. 3 (2001) 539–546 .
[18] M.D. Abramoff, P.J. Magalhães , S.J. Ram , Image processing with imagej, Biophys. Intern. 11 (2004) 36–42 .
[19] J.D. DeFouw , J.P. Quintana , D.C. Dunand , In situ X-ray synchrotron diffraction study of mgb2 synthesis from elemental powders, Acta Mater. 56 (2008) 1680–1688 . [20] I. Weißensteiner , M. Petersmann , P. Erdely , A. Stark , T. Antretter , H. Clemens ,
V. Maier-Kiener , Deformation-induced phase transformation in a Co–Cr–w–Mo al- loy studied by high-energy X-ray diffraction during in-situ compression tests, Acta Mater. 164 (2019) 272–282 .
[21] M.L. Young , J.D. Almer , M.R. Daymond , D.R. Haeffner , D.C. Dunand , Load parti- tioning between ferrite and cementite during elasto-plastic deformation of an ultra- high-carbon steel, Acta Mater. 55 (2007) 1999–2011 .
[22] V. Taupin , R. Pesci , S. Berbenni , S. Berveiller , R. Ouahab , O. Bouaziz , Lattice strain measurements using synchrotron diffraction to calibrate a micromechanical model- ing in a ferrite-cementite steel, Mat. Sci. Eng. A 561 (2013) 67–77 .
[23] P. Capper, Properties of narrow-gap cadmium-based compounds, london: INSPEC, the institution of electrical engineers, 1994.
[24] R.D. Greenough , S.B. Palmer , The elastic constants and thermal expansion of sin- gle-crystal cdte, J. Phys. D 6 (1973) 587–592 .
[25] H.J. McSkimin , D.G. Thomas , Elastic moduli of cadmium telluride, J. Appl. Phys. 33 (1962) 56–59 .
[26] D. Berlincourt , H. Jaffe , L.R. Shiozawa , Electroelastic properties of the sulfides, se- lenides, and tellurides of zinc and cadmium, Phys. Rev. 129 (1963) 1009–1017 . [27] T. Alper , G.A. Saunders , The elastic constants of mercury telluride, J. Phys. Chem.
Solids 28 (1967) 1637–1642 .
[28] R.I. Cottam , G.A. Saunders , Ultrasonic attenuation in hgte from 2 to 300 K, Philos. Mag. 32 (1975) 1231–1243 .
[29] J.G. Mavroides , D.F. Kolesar , Elastic constants of hgte, Solid State Commun. 2 (1964) 363–364 .
[30] P. Ballet , X. Baudry , B. Polge , D. Brellier , J. Merlin , P. Gergaud , Strain determina- tion in quasi-lattice-matched lwir hgcdte/cdznte layers, J. Elec. Mater. 42 (2013) 3133–3137 .
[31] K. Guergouri , R. Triboulet , A. Tromson-Carli , Y. Marfaing , Solution hardening and dislocation density reduction in cdte crystals by Zn addition, J. Cryst. Growth 86 (1988) 61–65 .
[32] M.G. Williams , R.D. Tomlinson , M.J. Hampshire , X-ray determination of the lattice parameters and thermal expansion of cadmium telluride in the temperature range 20–420 °C, Solid State Commun. 7 (1969) 1831–1832 .
[33] D. Bagot , R. Granger , S. Rolland , Elastische eigenschaften von einkristallinem sili- cium in abhangigkeit von der temperatur und der versetzungsdichte, Phys. Status Solidi 177 (1993) 295 .
[34] R.P. Reed , C.N. McCowan , R.P. Walsh , L.A. Delgado , J.D. McColskey , Tensile strength and ductility of indium, Mater. Sci. Eng. A 102 (1988) 227–236 . [35] Y.S. Touloukian, Thermophysical properties of matter, volume 13: thermal expan-
sion, 1977.
[36] J.G. Collins , J.A. Cowan , G.K. White , Thermal expansion at low temperatures of anisotropic metals: indium, Cryogenics (Guildf) 7 (1967) 219–224 .
[37] B.S. Chandrasekhar , J.A. Rayne , Elastic constants of indium from 1.4° to 300°K, Phys. Rev. 124 (1961) 1011–1014 .
[38] D.R. Winder , C.S. Smith , Single-crystal elastic constants of indium, J. Chem. Solids 4 (1958) 128–134 .
[39] C.L. Vold , M.E. Glicksman , E.W. Kammer , L.C. Cardinal , The elastic constants for single-crystal lead and indium from room temperature to the melting point, J. Chem. Solids 38 (1977) 157–160 .
[40] H. Cease , P.F. Derwent , H.T. Diehl , J. Fast , D. Finley , Measurement of Mechanical Properties of Three Epoxy Adhesives at Cryogenic Temperatures for CCD Construc- tion, 2006, pp. 1–19. Fermi Lab Report Fermilab-TM-2366-A .
[41] H.J. McSkimin , Measurement of elastic constants at low temperatures by means of ultrasonic waves–data for silicon and germanium single crystals, and for fused silica, J. Appl. Phys. 24 (1953) 988–997 .
[42] T. Middelmann , A. Walkov , G. Bartl , R. Schödel , Thermal expansion coefficient of single-crystal silicon from 7K to 293K, Phys. Rev. B 92 (2015) 174113/1–174113/7 . [43] A.L. Lebaudy , R. Pesci , M. Fendler , X-ray diffraction residual stress measurement at room temperature and 77K in a microelectronic multi-layered single-crystal struc- ture used for infrared detection, J. Electron. Mater. 47 (2018) 6641 .
[44] M. Plötner , B. Donat , A. Benke , Deformation properties of indium-based solders at 294 and 77K, Cryogenics (Guildf) 31 (1991) 159–162 .