Impact damages detection on composite materials by THz imaging

Any correspondence concerning this service should be sent to the repository

administrator:

staff-oatao@listes-diff.inp-toulouse.fr

O

pen

A

rchive

T

OULOUSE

A

rchive

O

uverte (

OATAO

)

OATAO is an open access repository that collects the work of Toulouse researchers and

makes it freely available over the web where possible.

This is an author-deposited version published in:

http://oatao.univ-toulouse.fr/

Eprints ID: 16554

To cite this version:

Destic, Fabien and Bouvet, Christophe Impact damages detection

on composite materials by THz imaging. (2016) Case Studies in Nondestructive Testing

and Evaluation, vol. 6. pp. 53-62. ISSN 2214-657

Impact

damages

detection

on

composite

materials

by

THz

imaging

Fabien Destic

a,

∗

,

Christophe Bouvet

a,b

a_{InstitutSupérieurdel’Aéronautiqueetdel’Espace(ISAE-SUPAERO),UniversitédeToulouse,31055ToulouseCedex4,France} b_{InstitutClémentAder(ICA),UniversitédeToulouse,CNRS,INSAT,ISAE-SUPAERO,MinesAlbi,UPS,31077ToulouseCedex4,France}

a

b

s

t

r

a

c

t

This paper presents aNon-Destructive Testing (NDT)method basedon the penetration propertiesofterahertz(THz)waves.ACWraster-scanningTHzimagingsetup,usinga3.8 THzQuantumCascadeLaserasasource,isusedtoperformNDTof polypropylene/polypro-pylenecompositesamples.

ResultsfromtransmissionandreflectionTHzimagingarecomparedtoultrasoundC-scan. THz imagesinreflectiongive similarresultstoC-scanwhereasTHztransmissionimages providemoreinformationaboutdelaminationsandcracksinthefiberfabrics.

1. Introduction

ThefieldofTHzwavesliesintheelectromagneticspectruminthefrequencyrangebetween300GHzand10THz,orin wavelengthsfrom30 μmto1 mm.Ontheonehand,theenergyofTHzphotonsisverylow,below40 meV.Ontheother hand,most non-metallic andnon-polar media are “transparent” to THzwaves. Thesetwo properties make THzwaves a potentialcandidateforNon-Destructive Testinginavarietyoffields,suchasartconservationscience[1],plasticweldjoints [2],qualitycontrolonchocolatebars[3]ortheabsorptionofwaterintopolyamideandwoodplasticcomposite[4].

Acompositematerialisacombinationoftwo(ormore) materialsworkingtogetherthatgive thecompositeitsunique properties. Thisdenomination is applied to many materials and manysectors: glass, aramide or carbonfibre reinforced compositesusedintheaeronauticsandautomotiveindustriesorsportsequipment,andwoodcompositeinfurniture man-ufacturing,amongothers.

NDT usingTHzwaves can be an alternative when“traditional” methods are not very efficient(composite made from softepoxyresin,forinstance)ortoimprovespatialresolution.Inasimplifiedmanner,therearetwokindsofTHzradiation technology:pulsedandContinuousWave(CW).

Inpulsedsystems,afemtosecondTi:Saormode-lockedfiberlaser(duration

∼

100 fs)produces anoptical-pulsetrainsent on aTHzemitter (photoconductiveantennaornonlinear crystal)to formelectromagneticpulses.Afterbeingtransmitted throughorreflected bythe sample,the resultingwaveformiscoherentlydetected intime-domain usingadelay lineand FourierTransformedtobeanalyzedinthefrequencydomain,providingabroadspectrumthatcoverstypicallythefrequency range[0.1–3]THz[5].

*

Correspondingauthor.

Fig. 1. Experimental setup.

The familyofCWorquasi-CWsystemscanbe dividedinto twotypesdepending onthe detectionused: coherent de-tection,generallyusedwithlowpowertunablesources,alsousesanopticaldelaylineandhasthesamecharacteristicsas pulsedsystems,anddirectdetectionwherephaseinformationisnotavailablebutoperationsarefasterandeasier.

Overthepasttenyears,some researchhasbeenconductedoncharacterizationorimaging,intheTHzrange,ofvarious composite materials,mostofwhichuse pulsedTHz-TimeDomain Spectroscopysystems(THz-TDS).Forinstance,THz-NDT havebeenappliedtotitaniumnanospheresinlowdensitypolyethylene(LDPE)[6]andglass-fibrereinforcedpolymers[7,8]. Fire damage testingon carbon orglass fibre was alsoexplored [9–11].The European FP7project, DOTNAC,is dedicated to the development of a TDSsetup to inspect aeronauticscomposite materials. Defects on glassfiber reinforcedplastics (foreign materialinserts,delaminations,ormoisturecontamination)canbevisualizedorcoatingsanalyzedoncarbonfiber samples[12].Amenabaretal.proposeareviewtoTHzNon-DestructiveTestingofcompositematter[13].

Continuous wave THz inspection is also possible as demonstrated on sprayed-on foam insulation (SOFI) with a gas laser at 1.63 THz [9] or carbon fibre at 0.6 THz with Gunn diodes [10]. Composite materials diagnostic using CW THz wavesproducedbyQuantumCascadeLaseronthinsamplesofglassfibreorKevlarcompositehavebeenstudied[14].This approachseemstobeverypromisingasQCLsareTHzsourcesemittingafewtensofmW,atfrequenciesinthe[1–4]THz range,withaconstantlyimprovingbeamquality.

2. Experimentalsetup

The raster-scanningimagingexperimental setup (Fig. 1)uses a 3.8THz(

λ

=

79

μ

m)Quantum CascadeLaser(QCL)as a THzsource.Thelaser, basedontheResonant-Phonondepopulationscheme,hasbeendesignedbyR. Colombelli’sgroup atUniversité Paris-Sud[15].Whencooled at

∼

10Kbya closedcyclecryocooler,itemits35mWpeak powerinpulsed operation [16].An ILXLightwaveLDX-36010-35pulsedcurrentsourcedrivestheQCLat1.4Aby 950 μslong pulsesata 21 Hzrepetitionrate.

L1 isaTPX lens thatworksbothasalens andacryostatwindow. Thecollimated laserbeamdiameteris

∼

13 mmin

horizontaland

∼

15 mminvertical,divergencebeing

∼

5 mrad.AsecondTPXlens,L2,isusedtofocalizethelaserbeamin

a800 μmdiameterspotwherethesampleundertestisplaced.Theaveragepoweris

∼

130 μW.

The sample to be imagedis moved inX andY by two Newport SMC100 motorizedtranslationstages. Thesetup can operate either in transmission mode with the off-axis parabolic mirrors inside the red dotted rectangle or in reflection mode by meansof a HRFZ-Si beamsplitterand mirrors inthe black dotted rectangle.At each point or pixel,the mean andstandarddeviationof10intensityvaluesfromthedetector(TydexGolaycell,NEP

=

113 pW.Hz−1/2)isacquiredviaa Signal Recovery7270DSPlock-inamplifierand, so,atransmissionorreflectionimagecanbeconstructed.Usingametallic resolutiontarget,aspatialresolutionof0.5 mminverticalandhorizontalhasbeenmeasuredforthetwoarrangementsof theimagingsetup.Theacquisitionofa30 mm

×

30 mmimagewith0.5 mmstepslastsabout1 h25 min.

InordertocharacterizetheTHzimagesasobjectivelyaspossible,wewillusetwofiguresofmerit:

•

thedynamicrange, DR

=

10

.

log



Max min



whereMax andmin arethemaximumandtheminimumamplitudevaluesintheimage

•

thesignal-to-noiseratio, SNR

=

10

.

log



[

Si,j

Ni,j

]



whereSi,jandNi,j

=

σ

(

Si,j

)

arethesignalamplitudeandthenoiseatpixel

(

i

,

j

)

. 3. Impactdamagedetection

3.1. Samples

Thesamplesusedtodemonstratedamagedetectionwere 3 mmthickpolypropylene compositeplates,commercialized by Goodfellow under referencePP403300. This kindof composite hasapplications in automotive components, industrial cladding,audioproducts,personalprotectiveequipmentandsportsgoods.

The composite deals with 3 mm thickness platemanufactured from 20woven plies oriented at 0◦ of thermoplastic self-reinforcing compositeproducedby compacting polypropylenefabric.Thistype ofcompositepossesses aunique com-binationofimpact/crash resistanceandstiffness/strength.The densityis0.92 g/cm3_{,thetensilemodulus5.0GPa andthe}

tensilestress is 180 MPa.In self-reinforcedcomposite, thepolymer matrix is reinforcedwithhigh-tenacity fibers ofthe samepolymerfamily:inthepresentcase,itdealswithPPmatrixandPPfibers.Duringthehotcompactingmanufacturing process,partofthefibersaremeltedandrecrystallizeduponcoolingtobondthestructuretogether.Theglobalmorphology oftheobtainedcompositeisnotasclearasaclassicalcompositewithwellseparatedplies,and,consequently,theresulting impact damage doesnot consistin delamination andmatrixcracking (Fig. 2(c)) butratherin a damaged zone including matrixcrackingandfiber/matrixdebonding.

3.2. Damagecreation

Fig. 2(a)showsthefallingweightimpacttestingsystemused.Thefallingweight(mass

=

2.03 kg)isinfreefallguidedin atube.Theimpactenergyisknownfromthevelocity,measuredbysensors[17].Two100 mm

×

150 mmplates(Fig. 2(b)) havebeenimpacted,intheircenter,withenergiesatlevelsof10 J(sampleS1)and20 J(sampleS2). Theimpactdamage (Fig. 2(c))consistsof:

•

dentdepthduetotheimpactorcontact

•

matrixcracking

•

delamination(failureofinterfacebetweenconsecutiveplies)

•

failureoftheplylocatedbackside

Ofcourse,thehighertheimpactenergylevel,thegreaterthedamage. 4. Results

4.1. ReflectionTHzimaging

Forboth samples,a reflection image is acquiredwith the setup andconditions described in Fig. 2.For each sample, S1 andS2,two imagesaretaken: onewiththe sample illuminatedonthe impactside, so-called frontside orrecto,one illuminatingthe sample opposite to the impact i.e.back side orverso. The THzimages are then compared toultrasonic mapsfroma KrautkramerUSD30equipment.Toperformthesemaps,calledC-scanimages,theplateisfullyimmersedin water.Asingletransducerof3MHzfrequency(

∼

500 μminwavelength)operatingasatransmitter–receiverisscannedina planeparalleltothecompositeplatesurfaceandonlythereflectionfromthebackfaceismonitored.Thistechniqueallows tomeasureboththeamplitude ofthereflectionandthetime offlight,i.e.thetimewhichisneededforanemitted pulse reflectedfromthebackfaceofthesampletothereceiver,andthentoevaluatethedepthofthefirstdamage.Asoursetup doesnotgivephaseinformation,theamplitudeC-scanultrasoundsimagesarecomparedtotheTHzreflectionimages(after convertingintensitytoamplitude).Onallimages,amplitudeisnormalizedrelativetoitsmaximumanddisplayedonacolor scale.

Foreachsample,pairedimagesareshownandcompared:THzreflectionimageandultrasonicamplitudeC-scan,sorted accordingtoilluminationsideandimpactenergy.

Inorder to easily comparethe imagesobtainedwithTHzandC-scan techniques,thedamage was highlighted witha dottedwhiteformandtheimpactlocation washighlightedwithanarrow(Figs. 3–7).Ofcoursethedamageformwasthe sameforeveryimage ofthesameimpact,andsymmetrywas doneforversoobservations(Figs. 5–7).Thesolepurposeof thedamageformwastocomparethedamageobtainedwiththedifferentobservationsbutitcannotbe consideredas the

Fig. 2. Damage creation.

exactdamagedzone.Infact,thecomplementarityofthedifferentobservationswasusedinordertoplotthedamagedzone asaccuratelyaspossible.

4.1.1. Recto

Regarding thedamagecreatedonsampleS1 bythe10Jimpact(Fig. 3),theoverallsize oftheobserveddamagezones issimilar.IntheTHzimage(Fig. 3(a)),thecentralspot,correspondingtotheimpactpoint,showsthemaximumamplitude signal, whileitisthemostdamaged zone.Itisprobablyduetothe dentdepthcreatedbytheimpactor(Fig. 2(c)),which mayactasaconcavereflectorandasalightfocalizer.

Fig. 4summarizestheresultsobtainedontheplateimpactedwithanenergyof20J.TheTHzimage(Fig. 4(a))showsno significant differencewithFig. 3(a),correspondingtothe10J impact.Indeed,inbothcases,thedamagearea isexpressed byamaximumsignallocatedwithinacircleof3–4 mmdiameterwhichisprobablyduetothedentdepthoftheimpactor inthe“soft”materialand,so,thecreationofaconcavereflectorthatmayactasalightfocalizer.Thereisnotanysignificant size differenceonthe bluecrescent-shapedareas(signalminimum).Intheultrasonicimage Fig. 4(b),thedamageis con-tainedwithinazoneof20 mmhighand15 mmwide.Withinthiszone,threemaximumspotsareseen.Theyareassumed to correspond tonon-damagedareas,which seems unlikely regardingtheir locationsandsizes. Ratherthey are probably measurementartifacts

Fig. 3. Sample S1 recto.

4.1.2. Verso

ResultsforthebacksideorversoobservationsarereportedinFig. 5andFig. 6forthe10Jand20Jimpacts,respectively. Inallcases,theversoobservationsare clearerthan rectoobservations,forbothTHzandforultrasonicinvestigations. This ismostlikelyduetoduetotheshapeofthebacksidewhichislessdeformedthanthefrontside.Indeedthedentdepth leftbytheimpactoratthefrontsideprobablydisturbsthewave penetrationintheplate.Atthebackside,inspiteofthe plyfailure(Fig. 2(c)andFig. 8),theplateshapeissmoother.

At the two impact energies, the signal levels (eq. colors), shapesand sizes produced either by the THzsetup or the ultrasoundC-scanarequitesimilar:an ellipticalblue–greenzonecorresponding toalowsignal forsampleS1 inthe 10J impactcase(Fig. 5(a)andFig. 5(b)),a“potato-like”zoneforsampleS2inthe20Jcase.But,bothfor10Jimpactandfor 20Jimpact,thedamagedzoneevaluatedusingTHzinvestigationseemsclearerthanwiththeultrasonicone.

4.2. TransmissionTHzimaging

Then,atransmissionimageisacquiredforbothsamples.ThesampleisilluminatedbytheQCLontherecto.Theaverage attenuationduetosampleabsorptionandreflectanceisabout30dB.

Fig. 7(a)showstheresultingimageforsampleS1,impactedat10J.Itpresentsatransmissionminimum(redandbrown) inan area ofabout2 mmwide and7 mmlong(dimensions aredetermined fromthe intensityprofile,notshownhere). Itcorresponds toanon-impactedside crack,causedby thetraction,whichfollowstheweavingpattern. Thiscrack,inthe thicknessoftheplate, canbe likenedtoa slotthat diffractsterahertz radiationthusthe signalisreduced. Inthe caseof

Fig. 4. Sample S2 recto. (For interpretation of the references to color in this figure, the reader is referred to the web version of this article.)

sample S2i.e.20Jimpactenergy(Fig. 7(b)),thetransmissionimage showstwoperpendicularslotswhosedimensionsare approximately11mmwideand10mmhigh,correspondingtotwocracksat90◦ anglebytheweavedirections.

4.3. ConsiderationsaboutTHzimages

Uncertaintiesare difficulttoevaluate.Infact,the C-scanis averyclassic methodwhichisused hereasthereference. Unfortunately, in this case, the C-scanof woven PP composite givesuncertain results. Thisis mostprobably due to the complex morphology ofthe impact damage duetothe hot compacting manufacturing process whichmelts thefibers of consecutivepliestogether.

Table 1summarizestheDRandSNRobtainedforthedifferentTHzimages.Ontheonehand,dynamicrangeishigherfor reflectionimages(

∼

19–20 dB)thanfortransmission,probablyduetotheshapeofsamples.Forrectoillumination,thedent depthoftheimpactorfocalizethereflectedlightandthesignalishigh.Rectoilluminationwerealsousedfortransmission images butinthesecases, thelight isabsorbed anddiffractedduring propagation throughthe material.In thedamaged zone,diffractioneffectsaremuchhigherandso,thesignalintensityislower.

Ontheotherhand,SNRisbetterfortransmissionthanforreflectionexperiments.Thisbehaviorisaresultofthelock-in amplifierresponsetosignaldynamicrange:giventhattheamplifierinputrangeisfixed regardingthehighlevel,thenthe lowlevelsignalmeasurementisnoisierandtheSNRdecreasesastheDRincreases.

Fig. 5. Sample S1 verso. (For interpretation of the references to color in this figure, the reader is referred to the web version of this article.)

Table 1

FiguresofmeritoftheTHzimages.

Fig. Test parameters DR SNR

3(a) S1 refl. recto 19.2 dB 17.8 dB

4(a) S2 refl. recto 19.2 dB 17.8 dB

5(a) S1 refl. verso 19.4 dB 13.6 dB

6(a) S2 refl. verso 19 dB 10.8 dB

7(a) S1 trans. 19.8 dB 14.7 dB

7(b) S2 trans. 19 dB 10.8 dB

4.4. Microscopeobservationsofthedamage

In order to validate our results, the two samples were cut at the impact point, perpendicularly to the crack. After polishing, the damage was observed witha Scanning ElectronMicroscope (SEM). Corresponding imagesare reported in Fig. 8.

Inboth cases,a crackisvisibleontheback sideoftheimpact (verso)and, aroundthiscrack,a delaminationzone.In picture8(b)ofdamageto20J,itisclearlyvisiblethattherearfaceoftheplateisslightlyconvex.

Fig. 6. Sample S2 verso.

ThisbacksidecurvaturecausesthedecreaseinthereflectedsignalandformsobservedbothontheTHzimagesandon C-scanimages.Weavefibersaredeformed,uptodelaminations.Asconfirmedbythe8(a)photography,plateS1,impactedat 10J,islessdamagedthanS2,impactedat20J.Thecurvatureofthebacksideisalsolighter.TheSEMobservationshowsa delaminationdamage

∼

2 mmforthesampleimpactedat10Jandabout6to7mmforthesecondsample.Theobservation ofthetwoplatecutsareingoodagreementwiththeTHztransmissionimageswheredamagewidthsofthesameorderof magnitudeareobserved.

5. Conclusion

Impactdamagedetectiononcompositefiberpolypropylene/polypropyleneplateswasperformedbyTHzimagingin trans-mission andin reflection.The resultswerecompared withultrasonicC-scanNDT, followedbymicroscopic observationof induceddamage.TheC-scanrevealsdamageparalleltotheplate,thusdelamination.Inthecaseofunidirectionallaminates, thisdamageissharpandflatandassuchitiseasily detectedbyC-scan.Inthecaseoffabrics,wherethedamageismore complex withcracks betweenpleatsbutalsointhe fold, betweenthetwo directionsofweavecalledmeta-delamination [18],theC-scangivesanindication aboutthedamagedareabutwithoutspecifyingthenatureandlocationofthedamage. The THzreflectionreverseside appears togive thesameindication astheC-scanwhile theTHztransmissionappears to highlight thefiber cracks.The two THzimagingtechniquesare complementary andcanprovide aNDT withinformation aboutthelocationandtypeofthedamage,crucialforunderstandingdamageincompositestructures.

Fig. 7. THz transmission image. (For interpretation of the references to color in this figure, the reader is referred to the web version of this article.)

References

[1]Gallerano G,Doria A,Giovenale E,Messina G,Petralia A,Spassovsky I,etal.THz-ARTE:non-invasiveterahertzdiagnosticsforartconservation.In:33rd internationalconferenceoninfrared,millimeterandterahertzwaves,IRMMW-THz2008.2008.p. 1–2.

[2]Wietzke S,Jördens C,Krumbholz N,Baudrit B,Bastian M,Koch M.Terahertzimaging:anewnon-destructivetechniqueforthequalitycontrolofplastic weldjoints.JEurOptSoc,RapidPubl2007;2:07013.

[3]Jördens C,Rutz F,Hasek T,Koch M.Towardsreal-timeterahertzqualityassuranceofchocolateproducts.In:Joint31stinternationalconferenceon infraredmillimeterwavesand14thinternationalconferenceonterahertz electronics,IRMMW-THz2006.2006.p. 332.

[4]Jördens C,Wietzke S,Scheller M,Koch M.Investigationofthewaterabsorptioninpolyamideandwoodplasticcompositebyterahertztime-domain spectroscopy.PolymTest2010;29(2):209–15.

[5]Zhang X-C,Xu J.IntroductiontoTHzwavephotonics,vol. 29.Springer;2010.

[6]Rutz F,Koch M,Khare S,Moneke M.Qualitycontrolofpolymericcompoundsusingterahertzimaging.ProcSPIE2005;5727:115–22.

[7]Rutz F,Wietzke S,Koch M,Richter H,Hickmann S,Trappe V,etal.Non-destructivetestingofglass-fibrereinforcedpolymersusingterahertz spec-troscopy.In:Europeanconferenceonnon-destructivetesting,ECNDT2006.2006.

[8]Jördens C,Scheller M,Wietzke S,Romeike D,Jansen C,Zentgraf T,etal.Terahertzspectroscopytostudytheorientationofglassfibresinreinforced plastics.ComposSciTechnol2010;70(3):472–7.

[9] Redo-Sanchez A,Karpowicz N,Xu J,Zhang X-C.Damageanddefectinspectionwithterahertzwaves.In:4thinternationalworkshoponultrasonicand adavancedmethodsfornondestructivetestingandmaterialcharacterization.2006.www.n.net.

[10]Karpowicz N,Dawes D,Perry MJ,Zhang X-C.FiredamageoncarbonfibermaterialscharacterizedbyTHzwaves.ProcSPIE2006;6212(1):62120G.

[11]Stoik CD,Bohn MJ,Blackshire JL.Nondestructiveevaluationofaircraftcompositesusingtransmissiveterahertztimedomainspectroscopy.OptExpress 2008;16(21):17039–51.

[12]Ospald F,Zouaghi W,Beigang R,Matheis C,Jonuscheit J,Recur B,etal.Aeronauticscompositematerialinspectionwithaterahertztime-domain spectroscopysystem.OptEng2013;53(3):031208.

[13]Amenabar I,Lopez F,Mendikute A. Inintroductory reviewtoTHznon-destructivetestingofcompositemater.JInfraredMillimTerahertzWaves 2013;34(2):152–69.

[14]Destic F,Petitjean Y,Massenot S,Mollier J,Barbieri S.THzQCL—basedactiveimagingappliedtocompositematerialsdiagnostic.In:35thinternational conferenceoninfraredmillimeterandterahertzwaves,IRMMW-THz.2010.p. 1–2.

[15]Chassagneux Y,Wang Q,Khanna S,Strupiechonski E,Coudevylle J,Linfield E,etal.LimitingfactorstothetemperatureperformanceofTHzquantum cascadelasersbasedontheresonant-phonondepopulationscheme.IEEETransTerahertzSciTechnol2012;2(1):83–92.

[16]Destic F,Petitjean Y,Mollier J.THzabsolutepowermeasurement:asimpleandreliablemethod.In:37thinternationalconferenceoninfrared, millime-ter,andterahertzwaves,IRMMW-THz.2012.p. 1–2.

[17]Bouvet C,Rivallant S.Dynamicdeformation,damageandfractureincompositematerialsandstructures.WoodheadPublishing;2016.

[18]Vieille B,Casado V,Bouvet C.Abouttheimpactbehaviorofwoven-plycarbonfiber-reinforcedthermoplastic- andthermosetting-composites:a com-parativestudy.ComposStruct2013;101(0):9–21.