Standard method for microCT-based additive manufacturing quality control 4 : Metal powder analysis

Method

article

Standard

method

for

microCT-based

additive

manufacturing

quality

control

4:

Metal

powder

analysis

Anton

du

Plessis

a,

*

,

Philip

Sperling

b

,

Andre

Beerlink

b

,

Willie

B.

du

Preez

c

,

Stephan

G.

le

Roux

a

a

CTScannerFacility,StellenboschUniversity,Stellenbosch,SouthAfrica

b_{YXLONInternationalGmbH,Hamburg,Germany}

c_{DeptofMechanicalEngineeringDept,CentralUniversityofTechnology,FreeState,SouthAfrica}

ABSTRACT

X-raymicrocomputedtomography(microCT)canbeapplied toanalysepowderfeedstockusedinadditive

manufacturing.Inthispaper,wedemonstrateadedicatedworkflowforthisanalysismethod,specificallyfor

Ti6Al4Vpowdertypicallyusedincommercialpowderbedfusion(PBF)additivemanufacturing(AM)systems.The

methodologypresentedincludessamplesizerequirements,scanconditionsandsettings,reconstructionand

imageanalysisprocedures.Weenvisagethismethodwillsupportstandardizationinpowderanalysisinthe

additivemanufacturingcommunity.Thisisaimedatultimatelyimprovingthequalityofadditivelymanufactured

parts,throughtheidentificationofimpuritiesanddefectsinpowders.

 MicroCTanalysisofmetalpowdersforadditivemanufacturing

 Methoddescribesastandardworkflowsimplifyingusageofthetechnique

 Samplerequirementsandimageanalysisworkflowisdescribed

©2018TheAuthor(s).PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBYlicense(http://

creativecommons.org/licenses/by/4.0/).

ARTICLE INFO

Methodsname:StandardmethodformicroCT-basedadditivemanufacturingqualitycontrol4:metalpowderanalysis Keywords:Additivemanufacturing,MicroCT,X-ray,Tomography,Non-destructivetesting,Powder,Particle,Characterization Articlehistory:Received7May2018;Accepted11October2018;Availableonline23October2018

*Correspondingauthor.

E-mailaddresses:[email protected](A.duPlessis),[email protected](P.Sperling),

[email protected](A.Beerlink),[email protected](W.B. duPreez),[email protected](S.G.leRoux). https://doi.org/10.1016/j.mex.2018.10.021

2215-0161/©2018TheAuthor(s).PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBYlicense(http:// creativecommons.org/licenses/by/4.0/).

ContentslistsavailableatScienceDirect

MethodsX

SpecificationsTable

SubjectArea Engineering

Morespecificsubjectarea: Additivemanufacturing/advancedmanufacturing/mechanical&industrialengineering Methodname: MicroCTanalysisofmetalpowder-standardmethod

Nameandreferenceof originalmethod

Noneyet,onlyisolatedcasesofindividualresearcherswhohavedoneslightvariationsofthe technique,ascitedinthepaper

Resourceavailability Alldescribedinpaperalreadywithreferences:typicalmicro/nanoCTscanner,3Dimage analysissoftwareeg.VolumeGraphicsVGStudioMax3.2

Methoddetails

Powderanalysisistraditionallydoneusingalaserdiffractionmethod,suchasdescribedinASTM B822 – 17. This laser diffraction method is simple, fast and provides estimated particle size distributions;howeverthisisbasedonasphericalparticleassumption.Oftentheparticlesfoundmay besignificantlynon-spherical.Somestudieshavealsomadeuseofmicroscopyandimageanalysisto analysethemorphologyofintactmetalpowderparticles.However,themicroscopymethodcanonly providepseudo-3Dimages,nottrue3Dimages,henceonlyqualitativeanalysisispossibleandinternal porosityinsideparticlescannotbevisualized.Sectioningofparticlesembeddedinresinandimaging oftheseparticlesusingamicroscopeispossibleandhasbeenusedincombinationwithstereological imageanalysistoprovideparticlesizedistributionsandshapeinformationandinthiscaseinternal porositymaybevisualized.However,thismethodhasthedisadvantageofbeingverytimeconsuming andstatisticallychallengingtocalculateproperparticlesizes,duetothesectioningofparticlesbeing inherentlynotthroughthemiddleinmostcases.Anotherdisadvantagewithsectioningisthatthe sectioningprocessmaysmearoversmallporesandmaythereforeaffecttheimagesobtained.

MetalpowderanalysisinAMhasbeenappliedtomonitorchangesinpowderqualityuponmany cyclesofre-use[1].Inthisstudyitwasshownthatpowderparticlesbecomelesssphericalandhavean increasinglyroughersurfacewithanincreasingnumberofre-usecycles.Itmayalsobethatother typesofpowderpartiallyfusewhichcandecreasepowderbedflowabilityproperties,butthishasnot beendirectlyreportedinthescientificliteraturetoourknowledge.Inanycasethequalityofre-used powderneedscharacterizationtoensuremaintenanceofoptimalproperties.

Ithasbeendemonstratedthatporosityinsidepowdersmaybetransferredtothemeltpooland hencetothefinalpart[2],inasynchrotrontomographystudy.Itisalsoknownthattheparticlesize distributionandthesphericityoftheparticlesaffecttheflowabilityofthepowders,whichinturn affectsthepowderbedqualityintermsofspreadingandpackingdensity.

TheuseofX-rayCTforanalysisofparticleshapeswasoriginallydemonstratedasearlyas2002 [3] andmorerecentlythemethodwascomparedwithvariousothermethodsformetalpowder analysis for additive manufacturing [4]. This work demonstrated theadvantage of CTand also discussedtheeffect ofrecycling ofpowder. Theuseoflaboratory microCTfor imagingof small particlessuchasmetalpowderswasdemonstratedinafewmorestudiesrecently,usingdifferent proceduresandsamplepreparation.Inonesuchstudytheinterestwassimplytovisualizeporosity inpowders,withoutdiscussionoftheprocedureused[5].Inastudyoftheparticleshapesofsmaller very irregular particles in the range 50–150

m

m, it was shown that microCT can be applied successfullytocharacterizetheparticleshapes[6].Inanotherpaperamethodologywasdescribed for microCTscans upto3

m

mresolution,witha dedicatedimage analysisprocedure[7].Inthis study,theparticleswereembeddedinresinandtheresinmachinedtoarodgeometry.Thisallows stabilityandeaseofmountingofthesampleinthemicroCTinstrumentforhighmagni_fication.The sameauthorsmorerecentlyextendedthisworktosmallerpowdersandscanresolutiondownto 0.7

m

m[8].Thispaper describesa workflowforobtaining powder porosityby microCTbutthe descriptionforobtainingparticlesizedistributionisnotclear,andtheproceduremakesuseof user-dependantproceduresforde-noisingandthresholding.Nevertheless,itdemonstratesfeasibilityof themethodandapplicabilitytocharacterizationofpowderstypicallyusedinAM,anddoesprovide afirststeptowardsstandardization.Alltheabove-mentionedstudiesmakeuseofcarefullymounted

particlesin resin.Thisprocedureof samplepreparationis timeconsumingandlimitsthewider uptakeofthismethod.

In the method presented here, we demonstrate a simplified methodology where no sample preparationisrequired:theparticlesareloadedinasmallcuportubeandscannedat0.7–1.5

m

m resolution(dependingontheparticlesizesexpected),foratotalscantimeofapproximately2_–3hper sample.Dependingontheanalysisrequiredtheimageanalysisprocedureinvolvesroughlythesame timeinvestmentasscanningtime,whichcanallowoptimizedworkflowforlargenumbersofsamples (imageanalysisof_firstsampledoneduringscanofsecondsample,etc).Wehaveappliedasimpli_fied versionofthismethodrecentlytotheanalysisofheavymineralsands,asshownin[9].Theprocedures describedhereindetailrequireshighresolutionscanningpossiblewithanysystemcontainingan X-raysourceandassociated hardwareallowingnanoCT, ie.submicronsourcespotsizeand system stability.Themethod alsouses imageanalysisroutinesavailable in commercialsoftware,which removes potential human bias from the methodology. Such simplified unbiased methods are importanttotheproperuseofthetechnologytosupporttheadditivemanufacturingcommunity,and isoneofanumberofstandardizedmethodsdevelopedinourgroup[10–12]andmentionedinarecent reviewofthetechnologyappliedtoAM[13].AsdescribedinSeifietal[14],thereiscurrentlyanurgent needforstandardizationintheAMcommunityandthequalityinspectionofmetalpowdersispartof thisrequirement.

Themethod

X-raymicrocomputedtomography[15]wasusedinthisstudyusingoptimizationproceduresas describedin [16]. Metal powder was acquired from a recentstudy of powdersused in differentcommercial systems[17],withthetwodemonstratedhereoriginatingfromcommercialsupplierTLSTechnikGmbH withlargesizefraction(LENSpowder,40–100

m

m)andtheotherwithsmallparticlesizedistribution (<40

m

m)foraDMLSAMmachine.Thesecoverthetypicalsizerangesofpowdersinusecommerciallyin AMsystems.Themethodologyisdemonstratedforthelargerpowder,whilethesmallerpowderisshown inthelastfigureandintheassociatedimageanalysisworkflowvideo(Supplementarymaterial).

Sampleswereloadedintoaplasticcup(forthelargerpowder)orathinplastictube(forthesmaller powder),thiswasfixedonaglassrodandmountedascloseaspossibletotheX-raysource;thesample mountscontainingpowdersareshowninFig.1.Thisallows,withhighqualityparametersandreasonable scantimes,voxelsizesof1.5

m

mforthecupand0.7

m

mforthetube.Thelargerpowdercuphasatotal widthofmetalpowderofapprox.2.5mm,andthisrequiresa0.1mmcopperfiltertopreventbeam hardeningartefacts.ThescansettingsarewithanX-rayspotsizeapproximately2

m

m.Forthesmaller containerwithtotalpowderwidthof0.7mm,nobeamfilterwasnecessary.InthislattercasetheX-ray spotiskeptbelow0.9

m

musingsuitableapertures(systemspecific).Inbothcasesthebeamhardening correctionappliedwasverystrongtoensurenogreyscalevariationacrossthediameterofthecup,which canaffectthesegmentationstep.ThescanparametersusedareshowninTable1.Thebestcontrastis obtainedwhentheentiresamplewidthfitsthefieldofview,thecurrentisincreasedtothemaximum allowedfortheX-rayspotsize(usuallysystemcontrolledlimits),andthenoiseislimitedbykeepingthe detectorascloseaspossibletothesource.Thismeansthatthesampleisveryclosetothesource,which requiresaverypreciseglassrodwithnoexcessmaterialwhichcanlimittherotation(seeFig.1).Scan settingsincludedetectorshift,toremovepossibleringartefactsandaveragingof2imagesateachstep positionwhilethefirstimageateachsteppositionisdiscarded.Afullrotationiscompletedwithupto 3000steppositions.Thepowdermustsettleinthecontainersoitissuggestedtorunadummyscanprior totherealscan,thisalsoallowsthesystemtothermallystabilizeandlimitsX-rayspotdrift.Itshouldbe mentionedthat10

m

mresolutionscansofpowderhavebeensuggestedinatleastoneaerospacequality controlguideline,toensurenoimpuritiesarepresentinthepowder.Suchascandoesnotresolvepowder particlesbutcanbeveryfastandcanthereforebeadditionallydonepriortohigherresolutionscans.Such fastscanswillimmediatelyindicatethepresenceofdenseimpuritiesbutmoredetailedimagesare requiredforporosityanalysisorfurtheranalysisasdescribedinthispaper.

FollowingagoodqualitymicroCTscanattheparametersinTable1,reconstructionusingastrong beamhardeningcorrectionandde-noisingusingadefaultadaptiveGaussfilterinVGStudioMAX3.1 (VolumeGraphics,Heidelberg,Germany),theresultingmicroCTsliceimagesforthelargeparticle

powderareshowninFig.2.Moredetailscanbeseeninthestepsshowninthesupplementaryvideos. The3Dimageshowstheexteriormorphologyoftheparticleswhilethesliceimagesshowinternal porosity andmore detailsof the morphology.This imagecan beusedtoassess thepresence of impurities,withoutanyfurthercomplexanalysis.

Evaluationofporosityinthepowderscanbechallengingassomeporesmightbeopentothe surfaceoftheparticlewhileothersarenot.Thesuggestionistomanuallyevaluatetheporosityinslice images.A morequantitative(optional)assessmentisdescribed here– thisinvolvesselectingthe closedporosityonly.Thiscanthenbeusedtoassessmanuallytheextentofopenporosityvsclosed porosity.Thesegmentationmethodinvolvesapplyinganadvancedsurfacedetermination(usingthe autofunction,nohumanbias)withandwithout“removeallvoids”.IneachcaseanROIisselected fromthesurfacedetermination,andthetwoROIsaresubtractedfromoneanothertoleaveonlythe internalporesasanROI.ThisROIisusedinacustomdefectmaskporosityanalysis,toprovidecolour codedporosityinformationoftheclosedporesasshowninFig.3.

Table1

Scansettingsforeachtypeofscan.

Voxelsize(mm) Voltage(kV) Current(mA) Scantime(hrs) Fieldofview(mm)

2 100 100 2 2.5

0.7 100 280,withapertures 3 0.7

Fig. 1.Samplemounting–thepenisforscaleindication,thepowderisloadedinthecuportubeasshown–sampleonleftisfor 1.5mmscan,sampleonrightisforsmallerpowdersfor0.7mmscan.

Thenextstepisforanalysisoftheparticlesizesandshapes,forwhichtheVGStudioMAX3.1foam structureanalysismoduleisused,applyingthealgorithmtothematerial.Defaultsettingsareapplied toobtaintheanalysisasshowninFig.4,whichprovidesforeachparticleavolumeasshowninthe colourcoding.Theparticlesizedistributioncanthereforebeanalysedindetail(Fig.5a),aswellasthe sphericitydistribution(Fig.5b).Sphericityisheredefinedastheratioofthesurfaceareaofasphere withthesamevolumeastheparticle,relativetothesurfaceareaoftheparticleitself.Forstatistical analysisthedataforeachparticleisextractedinaCSVfileasaspreadsheet.Thereisnosegmentation step,suchastypicalwatershedalgorithmusedinothersoftwaretools,butthesplittingoftouching

Fig.2.CTscanresultsshowing(a)3Dsurfaceviewand(b)CTsliceimageclearlyindicatingparticleswithporespaces(black circles).VisualizationsperformedwithVGStudioMAX3.1.

Fig.3. Porosityanalysisofpowders(closedporesonly).

Fig.5.StatisticalinformationobtainedbymicroCTof(a)particlesizedistributionand(b)sphericitydistribution–atotalof 62,137particleswereanalysedinthisdataset.

Fig.6.CTsliceimageofEOSpowderwithpeakof40mmat(a)scanresolution1mmusingthelargercontainer,andanimproved scanusingasmallercontainerat(b)0.7mm.Thesmallersamplesizerequiresasmallercontainerwhichresultsinimproved scanquality.

particlesisaffectedbya“mergethreshold”valuewhichisbydefaultsetto5%andworkswellinmost cases.Whenitisobservedthattoomuchortoolittlesplittingoccurs,thisvaluecanbeadjusted,and thiswilldependonthescanqualityandresolutionrelativetoparticlesize.

Theabove-mentionedexampleclearly showswhatispossible, whenthescanresolutionis 1.5

m

mand theparticlesizedistributionpeakisaround100

m

m,thereforebasedonthisscananaverageof66voxels arerequiredacrossameanparticlefortheabovedetailedanalysis.Notonlytheresolutionbutalsothe contrast is importanthere. In cases when thepowder is smaller,this may result in poor contrast and hence thedetailedanalysesarenotpossible.Inthiscaseimpuritiescanstillbecheckedandestimatescanbe madeofthemorphologyofthepowders.OneexampleofthisisshowninFig.6a,wherethescan resolutionof1

m

mforpowderwithanexpectedpeakof<40

m

misshown.Thesmallsizeofthepowder limitedthescanqualityandhencelimitsthefurtherprocessingofthedatawhenscannedusingthe 2.5mmwidecup.Besidesresolution,thereisalsopoorpenetrationandsub-volumescanning,reducing thedataquality.Thebestcontrastisfoundwhentheentirewidthofthesamplefitsinthefieldofview.A smallerfieldofviewallows morepenetrationandhencebettercontrast.Fig.6bistheresultofan improvedscanofthesamepowderusingasmallertubeandahigherscanresolutionat0.7

m

mforthe samepowder.Thelatterscanat0.7

m

m,allowedquantitativeanalysisasshowninFig.7.Theimagesin Fig. 7demonstratethat thequantitativeanalysesdescribedcanbe applied to smallerpowders in thesame wayasdescribedabove forlarger powders.Thoughnotthetopic ofinvestigation of thismethod description,thissmallerpowderwasmeasuredashavingameanparticlediameterofonly14

m

m.

This methodwasrecentlyapplied tovirginTi6Al4Vpowderaspartofa roundrobinstudy,and interesting_“powderinsidepowder_”wasobservedforthe_firsttimetoourknowledge,thisisshownin[18]. Conclusion

AsimplemethodwasdescribedwhichallowshighresolutionmicroCTscansanddetailedanalysis ofTi6Al4Vmetalpowder,typicalforlaserpowderbedfusionsystems.Whilethemethoddescribed hereisforTi6Al4Vparticles,itmaybemodifiedslightlyandappliedinasimilarmannerforloweror higherdensityparticles,andforparticleswithsmallerorlargersizedistributions.Smallerparticles willrequireahigherresolutionscanandpotentiallyasmallersampletube.Forlargerparticles,alarger fieldofviewisrequiredandlargercup,toensurenoparticlesarecutoffattheiredges.Denserparticles mightrequirelongerscantimesandthesmallestdetectedparticlesmightbelargerduetoincreased powerrequiredwhichincreasestheX-rayspotsize.

Asastandardmethod,itissuggestedthatwhenanunknownpowderistobetested,thefirstscanis doneat1.5

m

masdescribedabove,whichallowstoroughlycheckforimpurities,morphologyand porosity.Ifthepowderislargeenough(approx.>100

m

m),quantitativeanalysisisalsopossiblewith thisdata,asdemonstrated.Ifquantitativeanalysisisrequiredbutthepowderisfoundtobetoosmall foraclearsegmentation,ahigherresolutionscanissuggestedusingasmallertubeasshownfor 0.7

m

m.Theexamplesshownherecovertherangeofsizesexpectedformostmetalpowderbedfusion systemsandcanthereforebeusedforthisapplicationdirectlywithoutfurthermodificationofthe parameters.Inthiscase,powderwithmeansizeof14

m

mwassuccessfullyanalysedindetailusingthe 0.7

m

mscansettings.

This imageanalysis methodologyprovides information on internalporosity (open or closed), particlemorphology(volume,surfacearea,sphericity)andonthepresenceofimpurities suchas denserparticles.However,themethoddoesnotprovideinformationonoxygencontent,whichisan issueinre-usedorexposedpowders,anditdoesnotprovideinformationonparticlesorporessmaller thanthevoxelsize.Italsodoesnotnecessarilyprovideinformationonmulti-particles,butthismight beaninterestingtopicforfurtherinvestigation.Itshouldthereforebeusedaspartofaholisticquality inspection,alsoincorporatingothermethods.

Acknowledgements

Fig.7.SubmicronCTscanofDMLSpowder(<40mmspecification).Thisseriesshowsthesliceimagewithoutandwithanalysis, a3Dviewoftheanalysisforsize,anda3Dviewoftheuncodedparticles–coloursavariedbetweenadjacentparticlesto highlightthelargenumberofparticles.

TheCollaborativePrograminAdditiveManufacturing(CPAM),(Contract CSIR-NLC-CPAM-15-MOA-CUT-01),fundedbytheSouthAfricanDepartmentofScienceandTechnology,isacknowledgedfor financialsupport.YXLONinternationalisalsoacknowledgedforfinancialsupportofthiswork. AppendixA.Supplementarydata

Supplementarymaterialrelatedtothisarticlecanbefound,intheonlineversion,atdoi:https://doi. org/10.1016/j.mex.2018.10.021.

References

[1]H.P.Tang,M.Qian,N.Liu,X.Z.Zhang,G.Y.Yang,J.Wang,Effectofpowderreusetimesonadditivemanufacturingof Ti-6Al-4VbyselectiveElectronbeammelting,JOM67(2015)555–563,doi:http://dx.doi.org/10.1007/s11837-015-1300-4. [2]R.Cunningham,A.Nicolas,J.Madsen,E.Fodran,E.Anagnostou,M.D.Sangid,A.D.Rollett,Analyzingtheeffectsofpowder

andpost-processingonporosityandpropertiesofelectronbeammeltedTi-6Al-4V,Mater.Res.Lett.5(2017)516–525,doi: http://dx.doi.org/10.1080/21663831.2017.1340911.

[3]E.J. Garboczi, Three-dimensional mathematical analysis of particle shape using X-raytomography and spherical harmonics:applicationtoaggregatesusedinconcrete,Cem.Concr.Res.32(2002)1621–1638,doi:http://dx.doi.org/ 10.1016/S0008-8846(02)00836-0.

[4]J.A.Slotwinski,E.J.Garboczi,P.E.Stutzman,C.F.Ferraris,S.S.Watson,M.A.Peltz,Characterizationofmetalpowdersusedfor additivemanufacturing,J.Res.Inst.Stand.Technol.119(2014)460–493,doi:http://dx.doi.org/10.6028/jres.119.018. [5]F.Léonard,S.Tammas-Williams,I.Todd,CTForAdditiveManufacturingProcessCharacterisation:AssessmentofMelt

StrategiesonDefectPopulation,(2016).(AccessedFebruary20,2018)www.3dct.at.

[6]C.Rothleitner,U.Neuschaefer-Rube,J.Illemann,Sizeandshapedeterminationofsub-millimetersizedabrasiveparticles withX-raycomputedtomography,6thConf.Ind.Comput.Tomogr.,(2016),pp.1–9.(AccessedFebruary20,2018) www.3dct.at.

[7]K.Heim,F.Bernier,R.Pelletier,L.-P.Lefebvre,HighresolutionporesizeanalysisinmetallicpowdersbyX-raytomography, CaseStud,Nondestruct.Test.Eval.6(2016)45–52,doi:http://dx.doi.org/10.1016/J.CSNDT.2016.09.002.

[8]F.Bernier,R.Tahara,M.Gendron,AdditivemanufacturingpowderfeedstockcharacterizationusingX-raytomography,Met. PowderRep.73(2018)158–162,doi:http://dx.doi.org/10.1016/j.mprp.2018.01.002.

[9]A.Rozendaal,S.G.LeRoux,A.duPlessis,C.Philander,GradeandproductqualitycontrolbymicroCTscanningoftheworld classNamakwaSandsTi-ZrplacerdepositWestCoast,SouthAfrica:anorientationstudy,Miner.Eng.(2017),doi:http://dx. doi.org/10.1016/j.mineng.2017.09.001.

[10]A.duPlessis,P.Sperling,A.Beerlink,L.Tshabalala,S.Hoosain,N.Mathe,S.G.leRoux,StandardmethodformicroCT-based additivemanufacturingqualitycontrol1:porosityanalysis,MethodsX5(2018)1102–1110.

[11]A.duPlessis,P.Sperling,A.Beerlink,L.Tshabalala,S.Hoosain,N.Mathe,S.G.leRoux,StandardmethodformicroCT-based additivemanufacturingqualitycontrol2:densitymeasurement,MethodsX(2018).

[12]A.duPlessis,P.Sperling,A.Beerlink,O.Kruger,L.Tshabalala,S.Hoosain,S.G.leRoux,StandardmethodformicroCT-based additivemanufacturingqualitycontrol3:surfaceroughness,MethodsX5(2018)1111–1116.

[13]A.duPlessis,I.Yadroitsev,I.Yadroitsava,S.G.LeRoux,X-raymicrocomputedtomographyinadditivemanufacturing:a reviewofthecurrenttechnologyandapplications,3DPrintingAddit.Manuf.5(3)(2018)227–247.

[14]M.Seifi,M.Gorelik,J.Waller,N.Hrabe,N.Shamsaei,S.Daniewicz,J.J.Lewandowski,Progresstowardsmetaladditive manufacturingstandardizationtosupportqualificationandcertification,JOM69(2017)439–455,doi:http://dx.doi.org/ 10.1007/s11837-017-2265-2.

[15]A.duPlessis,S.G.leRoux,A.Guelpa,TheCTScannerFacilityatStellenboschUniversity:anopenaccessX-raycomputed tomographylaboratory,Nucl.Instrum.MethodsPhys.Res.Sect.BBeamInteract.withMater.Atoms.384(2016)42–49,doi: http://dx.doi.org/10.1016/j.nimb.2016.08.005.

[16]A.duPlessis,C.Broeckhoven,A.Guelpa,S.G.leRoux,Laboratoryx-raymicro-computedtomography:auserguidelinefor biologicalsamples,Gigascience6(2017)42–49,doi:http://dx.doi.org/10.1093/gigascience/gix027.

[17]K.Thejane,S.Chikosha,W.duPreez,CharacterisationandmonitoringofTI6AL4V(ELI)powderusedindifferentselective lasermeltingsystems,SouthAfricanJ.Ind.Eng.28(2017)161–171,doi:http://dx.doi.org/10.7166/28-3-1853. [18]A.duPlessis,S.G.leRoux,StandardizedX-raytomographytestingofadditivelymanufacturedparts:aroundrobintest,