STEM‐EELSanalysisrevealsstablehigh‐densityHeinnanoporesofamorphous siliconcoatingsdepositedbymagnetronsputtering RolandSchierholza,b,*,BertrandLacroixa,VandaGodinhoa,JaimeCaballero‐Hernándeza,Martial DuchampcandAsunciónFernándeza,* aInstitutodeCienciadeMaterialesdeSevilla,CSIC‐Univ.Sevilla,Avda.AméricoVespucio49,41092 Sevilla,Spain bInstitutfürEnergie‐undKlimaforschung,GrundlagenderElektrochemie(IEK9), ForschungszentrumJülichGmbH,52425Jülich,Germany cErnstRuska‐CentreforMicroscopyandSpectroscopywithElectrons,ForschungszentrumJülich GmbH,52425Jülich,Germany * [email protected],
[email protected] Abstract Abroadinteresthasbeenshowedrecentlyonthestudyofnanostructuringofthinfilms andsurfacesobtainedbylow‐energyHeplasmatreatmentsandHeincorporationviamagnetron sputtering.Inthispaperspatiallyresolvedelectronenergy‐lossspectroscopy(EELS)inascanning transmissionelectronmicroscope(STEM)isusedtolocateandcharacterizetheHestatein nanoporousamorphoussiliconcoatingsdepositedbymagnetronsputtering.AdedicatedMATLAB programwasdevelopedtoquantifytheheliumdensityinsideindividualporesbasedontheenergy positionshiftorpeakintensityoftheHeK‐edge.Agoodagreementwasobservedbetweenthe highdensity(~35‐60at/nm3)andpressure(0.3‐1.0GPa)valuesobtainedinnanoscaleanalysisand thevaluesderivedfrommacroscopicmeasurements(thecompositionobtainedbyproton backscatteringspectroscopycoupledtothemacroscopicporosityestimatedfromellipsometry). Thisworkprovidesnewinsightsintothesenovelporouscoatings,providingevidenceofhighdensityHelocatedinsidetheporesandvalidatingthemethodologyappliedheretocharacterize theformationofporesfilledwiththeheliumprocessgasduringdeposition.Asimilarstabilization ofcondensedHebubbleshasbeenpreviouslydemonstratedbyhigh‐energyHeionimplantationin metalsandisnewlydemonstratedhereusingawidelyemployedmethodology,magnetron sputtering,forachievingcoatingswithahighdensityofhomogeneouslydistributedporesandHe storagecapacitiesashighas21at%. Keywords:spatiallyresolvedEELS,spectrumimaging,magnetronsputtering,amorphousporous siliconcoatings,closednanopores,condensedHebubbles,reducedrefractiveindex
1. Introduction Theintroductionofporosityinnanomaterialshasbeenatopicofincreasinginterestover thelastdecade,andresearcheffortshavebeenfocusedonthesynthesisandcharacterizationof newlydesignedporousandnanostructuredmaterials.Afundamentalunderstandingofthe nanostructurepermitsthecontrolanddesignofporousmaterialswithtailoredfunctionalities. Amongmanyothers,poroussiliconcurrentlyremainsoneofthemostactivelyresearched materialsfordiverseapplicationssuchassensors,photonicdevices,microelectronicsandsolar energyconversion[1‐6].Beingfullycompatiblewiththeestablishedmicroelectronicstechnology, themostattractingfeatureofporoussiliconisthewiderangeofrefractiveindexesthatcanbe achievedbyvaryingthepercentageofporosity.Inarecentwork,weproposedanewbottom‐up approachtoprepareporoussiliconcoatingsbymagnetronsputteringasanalternativeroutetothe traditionalelectrochemical‐basedprocesses[7].Inthatwork,wedemonstratedthepossibilityof producingporoussiliconcoatingswithtailoredrefractiveindices(from3.75to4.75at500nm) usingmagnetronsputteringwithHeorArastheprocessgas[7,8].Inadditiontotheadvantagesof thewell‐knownindustrialscale‐upofthemagnetronsputteringprocess,ourprocedureprovides manypossibilitiesforthedesignandproductionofvariousdevicesalternatingforexampledense andporouslayers,intrinsicordopedsilicon,etc.[7].Thelow‐refractive‐indexcoatingsproduced exhibitamicrostructureofclosedporesthatcanbeorientedtothecoating’sgrowingdirection.In thispreviouswork[7],theincorporationofahighHeamount(upto30at%)insidethecoatings wasdetectedbyprotonbackscatteredspectroscopy,anditwasproposedthatHewasmostlikely trappedinsidetheclosedporesduringdeposition[7,8].Infact,inanotherpreviouspaper,weused spatiallyresolvedelectronenergy‐lossspectroscopy(EELS)inascanningtransmissionelectron microscope(STEM)todemonstratethatamorphousporousSiNxOycoatingscanbepreparedby magnetronsputtering,revealingclosedporescontainingmolecularN2inside[9]. Atthispointoftheresearch,thechallengeinthepresentworkistoinvestigatethelocation oftheHeintheporousSicoatingsaswellastoquantifytheheliumwithnano‐scaleresolution usingtheSTEM/EELSmethodologyavailableintheTEMfacilitiesofourlaboratory.The incorporationofHeinmetalshasbeenwellinvestigatedduetoitsimportanceinnuclear technology.Ionimplantationhasbeenusedtore‐createtheconditionsofthereactorsmainlyto investigatedamagecausedinmaterials.Intheseworks,theextremelylowsolubilityofHein metalshasbeenreportedtobethecauseoftheformationofHebubblesduringionimplantation [10‐12].Inmostoftheseworks,heliumimplantedincrystallinematerialsleadstotheformationat acertaindepthofalayeroflargelydispersedbubbleswithdiametersrangingfrom2to25nm.
EELSwasusedfortheevaluationofheliumdensityinsidethesebubbles.Thefirstapproachwasto usetheblueshiftoftheHe1s→2ptransionasafunctionofHedensity[10,13].Later,spatially resolvedEELSwasreportedtobeabletorecordspectraofsinglebubblesbyseveralauthors [12,14‐16].Theadvantagehereisthattheheliumdensitycanberetrievedatthenanoscaleusing twodifferentmethods.OnemethodusestheintegratedintensityundertheHeK‐edge,andthe otherusesthealreadymentionedenergyshiftofthe1s→2ptransion.Adescriptionofboth methodsisgiveninreference12(andthereferencestherein).Thefirstaimoftheworkpresented herewastoapplyasimilarSTEM/EELSmethodologytoconfirmthattheHeincorporatedinthe amorphoussiliconcoatingspreparedbymagnetronsputteringremainsinsidetheclosedpores and,ifso,tocharacterizethephysicalparameters(density,pressure)ofHe.Bothgoalswere accomplished,asdescribedinthispaper,andanadhocMATLABroutinewasalsodeveloped. Additionally,wecomparethevaluesfortheHedensityobtainedinsidethemicroscopetothe valuesestimatedfrommacroscopicmeasurements.Notethatourresultspresentedinthispaper revealsimilarstabilizationofcondensedHeinthenanopores(Hedensitiesof~30‐50at/nm3)as thatpreviouslydemonstratedinbubblesformedbyhigh‐energyHeionimplantationinmetals [12,14‐16].AhighamountofHecanthereforebestoredandstabilizedatroomtemperatureinside theclosedpores.Tostoreasimilardensityinamacroscopictankatroomtemperature,pressures ofapproximately1kbarwouldberequired[17]. Tore‐createtheconditionsofthecoldplasmaincontactwiththenuclearreactorwalls, someresearchershaveartificiallyintroducedHeindifferentmetalsusingalsomagnetron sputteringinHe/Armixtures[18‐20].MorerecentlytheexposuretoHeplasmasandHe incorporationviamagnetronsputteringhavealsobeenpresentedasaroutefornanostructuring metals[21‐23].Inallthesearticles,althoughthepresenceofHeandnanoporeswasreportedby TEMimages,noquantificationoftheHecondensedstatewasprovided.Atthispoint,itis importanttoemphasizetherelevanceofthepresentworkthatnewlydemonstratesthatcoatings canbepreparedbymagnetronsputteringcontainingahighdensityofhomogeneouslydistributed poresfilledwithheliuminacondensedstate. 2. Experimental 2.1 Depositiondetails Inourpreviouswork[7]weshowedthedependenceoftheporesizeanddistributionon severaldepositionparametersofthemagnetronsputteringdeposition.Basedonthisinformation, adepositionpowerof300Winradiofrequency(RF)wasselectedinthispapertoobtainrather
largeandseparatedporestofacilitatetheindividualporeanalysisbySTEM/EELS(seethe micrographshowninfigure1).Themagnetronheadwasplacedata30°angletothenormalofthe substrateatadistanceof5cm(seetheinsetinfigure1left).Forcomparisonpurposesadense coatingwasalsodepositedusing150WinRF‐modeandasubstratebiasof100V.Both,thedense andporousamorphoussiliconcoatings,weredepositedfromapureSitarget(KurtJLesker 99.999%pure).Beforedepositionthebasepressurewas10‐4Pa.Theporouscoatingwasdeposited atp(He)=4.8Pa,whilethedensecoatingwasdepositedatp(Ar)=1.3Pa.Silicon(100)andquartz wereusedassubstrates.Table1summarizesthedepositionconditionsandlabelsforthestudied samples. 2.2Characterizationofthecoatings Thethicknessofthecoatingswasexaminedusingscanningelectronmicroscopy(HITACHIS‐ 4800SEM‐FEG)workingat2kVonfracturedsampleswithoutanyconductivecoating. Themicrostructureofthecoatingswasinvestigatedbytransmissionelectronmicroscopy (TEM)usingaFEITecnaiF30microscopeoperatedat300kVequippedwithaGatanGIFQuantum 963energyfilterattheLaboratoryofNanoscopiesandSpectroscopies(LANE‐ICMSE,Sevilla, Spain).Cross‐sectionalTEMspecimensofthecoatingsdepositedonSisubstrateswereprepared usingtheconventionalapproachbymechanicalpolishinganddimplegrindingfollowedbyAr+ion millinguntilelectrontransparencywasattained.ThequantitativeSTEM/EELSanalysisofindividual nano‐poresofhighlyporouscoatingsrequiresthepreparationofthinlamellaoflessthan100nm thickness,whichshouldincludeclosedfilledpores.However,overlapofporesshouldbeavoided, andverysmallporeswillmaketheSTEM/EELSanalysisofindividualporesmoredifficult.Inthis sensetheporoussampledepositedat300WisveryadequateasshowninFig.1. Thecompositionsofthethinfilmswereevaluatedusingprotonbackscattering spectrometry(p‐BS)attheNationalCenterforAccelerators(CNA,Sevilla,Spain)usinga3‐MV tandemaccelerator.Thespectrawereobtainedusingtwodifferentenergies,1.0and2.3MeV,for theprotonbeamandasurfacebarrierdetectorsetat165°.Toobtainthethicknessand compositionofthefilms,bothspectraweresimultaneouslysimulatedforeverysampleusingthe SIMNRAcode[24](seemoredetailsinref.7) TheopticalcharacterizationwasperformedoncoatingsonquartzsubstratesinaUVISEL spectroscopicellipsometerfromHORIBAJobinYvonwithanincidenceangleof60°anda wavelengthrangeof563‐1550nm.ThedatawereanalyzedusingtheDeltapsi2software developedbyHORIBAJobinYvon.
2.3STEM‐EELSspectrumimagingdataanalysis Annulardarkfieldscanningtransmissionelectronmicrographs(ADF‐STEM)andspatially resolvedelectronenergyloss‐spectra(EELS)wererecordedsimultaneouslyintheTecnaiF30on singleporesofvariousdiameters.Thespectrumimageswereacquiredinthelow‐lossregionwith apixelsizeof1nmusingsubpixelscanning(16*16).Ateachposition,anEELSspectrumof2048 pixelsoflengthwithenergydispersionof0.05eV/channelwasrecordedusinganintegrationtime between0.05and0.07s. TheHeK‐edgepeakcorrespondingtothe1sto2ptransitionwasplacedatapproximately 22eVonthehighenergytailoftheSiplasmonatapproximately17eV[25].ToprocesstheEELS spectrumimagedatasetandextracttheHesignal,aMATLABprogramwasdevelopedbasedonthe methoddescribedinreference12.Aversionofthisprogramisprovidedinthesupporting informationsection.First,therawdatawerealignedwiththezerolosspeak(ZLP)andsubjectedto multivariatestatisticalanalysis(MSA)[26]toremovethestatisticalnoise.Thepluralscattering distributionwasthenremovedfromthespectrausingFourier‐logmultiplescattering deconvolution(ProgramFLOG[27]).Thespecimenthicknessmapwascalculatedusingthelogratiomethod[27]usingthemeanfreepathcalculatedusingtheIMFProutine[27].Usingour experimentalconditions,ameanfreepathofλ=145nmwasobtainedusingtheMalisequation [28].ThecontributionoftheSiplasmonwasfittedwiththefunctiongivenineq.1fromreference 29:=/((2−2)2+(∆)2)(equation1) Here,Eprepresentstheplasmonenergy,∆Epcorrespondstothepeakwidth,andarepresentsthe amplitude.ToimprovetheHesignalextraction,asecondpeakwasrequiredatenergieshigher than23eV.ThiscontributionmaybeduetosurfaceoxideSiO2orcarboncontamination(bothat approximately24eV).Thesetwoplasmoncontributionswerefittedsimultaneouslytothedatain tworanges(from16to20.5eVandfrom23.5to40eV)onbothsidesoftheHe‐Kedge.Duringthe fit,theEpvalueofthesecondplasmonwasfixedbetween23and25eV.Thesumofthesetwo contributions(SiandSiO2plasmon)isreferredtoasthebackgroundsignalfortheHe‐Kedge.After subtractionofthebackgroundsignal,theresidualsignalwasfittedintheHeK‐edgerangefrom 20.5to23.5eV.FortheHe1s→2ptransion,asimpleGaussianfunctionwasusedforfittingas thisfunctionwasreportedtoyieldgoodresults[12].Anexampleofthefittingprocedurewillbe shownbelowinthe“ResultsandDiscussion”section(Fig.3c). ToquantifytheatomicdensityofHeateachpositionofthespectrumimage,twomethods
wereused.InMethod1[12],theHedensitynHeisgivenby nHe=IHe IZLP 1 dσHe (equation2) whereIZLPandIHearetheintegratedintensitiesofthezerolosspeak(ZLP)andtheHeK‐edge, respectively.σHeisthecross‐sectionoftheHe1s→2ptransion,whichwascalculatedusingthe programSigmak3[27].Forourexperimentalconditions(accelerationvoltageof300kV, convergencesemi‐angleα=12.4mrad,collectionsemi‐angleβ=9.6mrad,andintegrationrange of3eV),weobtainedσHe=6.410‐24m².distheporethicknessatthepixelposition.Thislatter valuewasevaluatedasthecomplementofthespecimenthicknessateachparticularpixel comparedwiththevalueatareferencepixeloutsidethepores.Thismethodwassuggestedin references14and15andwascross‐checkedusingtheSTEM‐ADFimagecontrastassuminga uniformellipsoidalvolumeoftheporesperpendiculartothegrowingdirection.TheresultsofIHe andIZLPwerestoredin2Dmaps,andthemapofHedensitywascalculatedusingtheseparameters. Toconfirmthesemeasurements,anothermethod(Method2)wasusedforcomparison. Thismethodreliesonthelinearrelationshipbetweentheblueenergyshift(ΔEHe)oftheHeK‐edge comparedwithitspositionforthefreeatom(21.218eV,ref.30)andtheheliumdensity(nHe): ∆E=C×nHe (equation3) whereCandδareconstants.Theenergyshiftisattributedtotheshort‐rangePaulirepulsion betweenelectronsoftheneighboringatoms[13].Inthiswork,weassumethattheHedensity insidetheporeswillfollowasimilarlinearlaw.Intheliterature,somediscrepanciesinCandδ valuescanbeobserved[10,12,14,15],andweusedtwoextremevaluesofC.Method2(a)usesE= 0.015nHe+0.45[15],andMethod2(b)usesE=0.044nHe+0.07eV[14].Finally,theHepressurepHe insidetheporescanbeestimatedfromthenHevaluesprovidedbythetwomethodsusinga suitableequationofstate(EOS)[11,12]dependingontheHedensityrange. 3. Resultsanddiscussion 3.1Poroussiliconcoatingwithreducedrefractiveindex Figure1showsthemicrostructureoftheinvestigatedporouscoating(a‐pSi(300),which consistsofahighlyporousstructureofclosedporesorientedinthemagnetrondirection(Fig.1a). Thehigh‐resolutionTEMimage(Fig.1b)showstheshapeoftheporesindetailandrevealsthe amorphousstructureoftheSi‐matrix.WithaRFpowerof300Wsuppliedtothemagnetron,an increaseinporesizeisobservedcomparedwiththecoatingsreportedinthepreviouswork[7].
Poreswithdiametersrangingfrom2to60nmcanbeobservedforthea‐pSi(300)coating.The poresexhibitsimilarminordiametersastheonesinprevioussamples[7](averagesizeof9±2 nm),whilethemajordiametersarelarger.TheporesareeasiertoseeontheADF‐STEMimage whichpresentsahighercontrast(Fig.1c).ThisimagepointsoutthatforourEELSanalysis(section 3.2),weneedtochooseathinregionofthespecimentofindisolatedpores(closetotheborder) butnottoothinsothattheporesremainclosed. Theprotonbackscattering(p‐BS)spectrumofthissampleiscomparedwiththatofthe densesampleinFigure2.AstrongsignalcorrespondingtoHeinsidethecoatingisobservedinthe poroussample,correspondingtoaHeamountof21at%(seeTable2).InadditiontoSi,thedense sample,a‐Si(150)(bias),whichwasdepositedunderAratmospherewithavoltageappliedtothe substrate,showsasmallsignalcorrespondingtotheincorporationofAr.TheincorporationofArin samplesdepositedbymagnetronsputteringwithanappliedsubstratebiashasbeenreported previouslybyotherauthors[31];nevertheless,thisincorporationdoesnotresultinthe introductionofporosity(detailsconcerningthemicrostructureofthisdensecoatingareprovided inreference7). Theintroductionofporosityleadstoadecreaseofboththerefractiveindexandextinction coefficient.Table3summarizesthevaluesoftherefractiveindicesanddampingconstantsofboth coatingsat605nm.Usingtheeffectivemediumapproximations[32],onecanexpresstherelative electricalpermittivityoftheporouscoating,aheterogeneousmedia(r,eff),asafunctionofthe dielectricconstantsofthecontinuousmatrix(r,c)anddispersedphase(r,d)andthevolume fraction()occupiedbythepores(eq.4,ref.32): εr,eff=εr,c 1‐3ϕεr,c‐εr,d 2εr,c+εr,d+ϕεr,c‐εr,d(equation4) Wecalculatedthedielectricconstant,ε=n²–k²,usingthevaluesofnandkmeasuredby ellipsometryat605nm,aslistedinTable3andderivedaporosityof22%fortheporouscoatingapSi(300)assumingthatthedielectricconstantofthematrixissimilartothatofa‐Si(150)(bias)and settingr,dfortheporesto1[33]. 3.2ExtractionoftheHeEELSsignalandevaluationofphysicalparameters(density,pressure) InFigure1,itispossibletoobservesomeisolatedporesinwhichthespatiallyresolvedEELS analysiscouldbeperformed.Figure3apresentstheADF‐STEMsignalacquiredinparallelwiththe spectrumimage,whichenclosesanisolatedpore.Therefore,anEELSspectrumcorrespondsto eachoneofthe1x1nm2pixelsintheimage.IntheADFsignal,thematrixwithhigheratomic
numberandscatteringcrosssectionappearsbright,andtheporesappeardark.InFigure3b,three EELSspectracorrespondingtothreeselectedpixelsatthecenter,theborderandoutsideofthe porearepresentedwithaverticalshiftforcomparisonpurposes. Fortheselectedpixel(16,10),whichcorrespondstoazoneoutsidethepore,thelowloss spectrumcontainsapeakcorrespondingtotheSiplasmonat17eV[25].Movingtowardthepore, pixel(5,10)attheborder,asmallpeakatapproximately22eVbeginstoappearthatbecomes strongeratthecenterofthepore,pixel(6,13).ThissuperimposedpeakcorrespondstotheHeK edge[25],confirmingthepresenceofHeinsidethepores. TocharacterizethephysicalparametersofHeinthecoating,thesignaloftheHeK‐edgeat eachpositionwasextractedfromtheEELSspectrafollowingtheproceduredescribedindetailin section2.3(seealsotheMATLABprograminthesupportinginformation).Toillustratethemethod, fitsofthebackgroundsignalforthematrixpositionoutside(16,10)andinside(6,13)theporeare presentedinFigures3cand3d.ThisbackgroundsignalincludesthepeakoftheSiplasmonaswell asanotherpeakforthesurfaceSiO2component.Thefiguresalsoshowtheresidualsignalobtained bysubtractionofthebackgroundandexperimentalspectra.NotethattheHesignalat approximately22eVisclearlyvisibleintheresidualsignalinsidethepore(Fig.3d),whileaflatline isobservedabove16eVforthepositionoutsidethepore(Fig.3c).Theseresultsclearlyvalidate ourfittingprocedurefortheextractionoftheHesignal.Thetwoadditionalfeaturesintheresidual signalatenergiesofapproximately5to10eV(Fig.3c)and10to15eV(Fig.3d)canbeattributed totheSi/SiO2interfaceplasmon[16,34]andthecavityplasmons[16],respectively. Figure4illustratesthemethodsusedtoevaluatetheHedensityinsidethepores.Usingour MATLABroutine(seesupportinginformation)andfollowingtheproceduresdescribedabove,the extractedheliumsignalwasfitwithaGaussianfunction(Figure3d),andthecorresponding integralsoftheHeintensity(IHe)andenergyshift(ΔEHe)relativetotheHefreeatomvaluewere storedin2Dmaps(seeFigures4aand4e).Asdemonstratedontheelementalintensitymap (Figure4a),heliumislocatedinsidethenanoporevolume.Thelocalspecimenthicknessateach pixelpositionisindicatedinFigure4b.Theporethicknessmap(Figure4c)waspreparedby selectingthepixel(17,9)asareferenceforthespecimenthickness(tref=38.8nm).Inthiscase,a maximumvalueof6.5nmwasobservedfortheporethickness,whichiscomparabletotheminor axisoftheellipsoidalpore(7nm)measuredontheADF‐imageinFigure3a.Usingtheresultsofthe Heintensityandporethickness,itwaspossibletopreparetheheliumdensitymapasdescribedin Method1(seeeq.2)andpresentedinFigure4(d).Byconsideringonlythepixelsclosetothepore center(toavoidsurfaceeffects)[14],thedensityinsidetheporerangesfrom25to54at/nm3,and
ameandensityof36at/nm3wasmeasured.ThedecreasingofnHefromthetoptothebottomof theporecanbeexplainedbyHedesorptionunderelectronirradiation,asreportedbyDavidetal. [15,16]duringSTEMrecordingofthespectrumimagingdata. FollowingMethod2(describedinsection2.3),Figure4eshowsthelocalenergyshiftΔEof theHeK‐edgefromthevalueofthefreeatoms[30].Figures4fand4gpresentthecorresponding He‐densitymapscalculatedaccordingtoeq.3usingMethods2(a)[15]and2(b)[14].Bycomparing Figures4d,4fand4g,agoodagreementbetweentheHedensityobtainedusingMethods1and 2(a)isobserved. Table4summarizesthenHevaluesobtainedintwodifferentporesusingtheabove mentionedmethodsforthecharacterizationoftheHestateinporousa‐Sicoatingsdepositedby magnetronsputtering.Forthetwopores,agoodagreementisobservedfortheHedensityusing Methods1and2a,whichprovidesagoodindicationofthereliabilityofourprocedure.Evenif differentauthorsagreeonthelinearcharacterofΔEandnHedependence,thevaluesgivenforC varysignificantly[10,12,14,15].Forourdensityrange(approximately20‐70at/nm3),webelieve thatthesystemisbetterdescribedusingMethod2abecausethelawgivenbyDavidetal.[15]was establishedforasignificantnumberofrecordeddatainthe20‐100at/nm3range.Thelawgivenby Tavernaetal.[14](Method2b)wasdeterminedusingsmallerHedensities,10‐40at/nm3,andthus doesnotcompletelydescribeoursystem.ThisideaissupportedbyLucasandco‐workers[13],who observedthattheCvalueisalsodependentontemperatureanddensity. KnowingthedensityofHeinsidethepores,theinternalpressurebecomesarelevant parametertobedetermined.Athigheratomicdensities,theidealgaslawisinaccurate.Walsh[12] discusseddifferentsemi‐empiricalequationsofstates(EOSs)tocalculatetheHepressureinside theporesfromthemeasurednHe.Forourdensityrange,theTrinkausEOS[11]isthemostsuitable. Themaximum,minimumandmeanHedensityvaluesdescribedbeforewereusedtocalculatethe Hepressure,andtheresultsarepresentedinTable4.UsingMethod1and2(a)forthefirstpore (Pore#1,presentedinFigs.3and4),ameanpressureof0.3GPawasdetermined,andthepressure locallyreachesupto0.8GPa.Inthesecondporeinvestigated(Pore#2),highermeanpressure valuesof0.5and1GPawereobserved,andthepressurelocallyreachesupto2GPa.These differencesinthetwoporesarenotnecessarilysignificantconsideringthelargeuncertaintiesof theHedensitymeasurementsthatmainlyoriginatefromthedifficultyofaccuratelymeasuringthe localporethicknessandfromHedesorptionduringthemeasurements.Nevertheless,theselocal measurementsprovideagoodindicationthattheHepressureinsidetheporeisintheGParange, whichisseveralordersofmagnitudehigherthanthepressureinthedepositionchamber,pHe=4.8
Fig.3
Fig.4