ARTICLE IN PRESS

(1)

Please cite this article in press as: Deng H, et al. Progress on the morphological control of conductive network ContentslistsavailableatScienceDirect

Progress in Polymer Science

j o u r n al ho me p ag e :w w w . e l s e v i e r . c o m / l o c a t e / p p o l y s c i

Review

Progress on the morphological control of conductive network in conductive polymer composites and the use as

electroactive multifunctional materials

Hua Deng

^∗

, Lin Lin, Mizhi Ji, Shuangmei Zhang, Mingbo Yang, Qiang Fu

^∗∗

CollegeofPolymerScienceandEngineering,StateKeyLaboratoryofPolymerMaterialsEngineering,SichuanUniversity,Chengdu 610065,People’sRepublicofChina

a rt i c l e i n f o

Articlehistory:

Received8November2012 Receivedinrevisedform4July2013 Accepted11July2013

Available online xxx

Keywords:

Conductivepolymercomposites Conductivenetwork

Electricalpercolationthreshold Morphologicalcontrol Sensing

Electroactivemultifunctional

a b s t ra c t

Sincetheemergenceoflargeaspectratioandmultifunctionalconductivefillers,suchas carbonnanotubes,graphenenanoplates,etc.,conductivepolymercomposites(CPCs)have attractedincreasingattention.Althoughthemorphologicalcontrolofconductivenetworks inCPCshasbeenextensivelyinvestigatedasanimportantissueforthepreparationof highperformanceCPCs,recentextensiveprogresshasnotbeensystematicallyaddressed inanyreview.Ithasbeenobservedthatthemorphologicalcontrolofconductivenetworks duringthepreparationofCPCshascrucialinfluenceontheelectricalpropertiesofthese composites.Severalmethodshavebeenshowntobeabletocontrolthenetworkstruc- ture,andthus,tunetheelectricalpropertiesofCPCs,includingtheuseofshear,polymer blends,thermalannealing,mixedfiller,latexparticleetc.Moreover,manynovelandexcit- ingapplicationshavebeenextensivelyinvestigatedforCPCs,suchasstretchableconductor, electroactivesensors,shapememorymaterialsandthermoelectricmaterials,etc.There- fore,themorphologicalcontrolofconductivenetworkinCPCsisreviewedhere.Issues regardingmorphologycharacterizationmethods,morphologicalcontrolmethods,resulted networkmorphologyandelectricalpropertiesarediscussed.Furthermore,theuseofCPCs aselectroactivemultifunctionalmaterialsisalsoreviewed.

Abbreviations: 2D,two-dimensional;3D,three-dimensional;ABS,acrylonitrile-butadiene-styrene;ABS-g-MA,maleicanhydridegraftedABS;AFM, atomicforcemicroscopy;C-AFM,conductiveatomicforcemicroscopy;CNFs,carbonnanofibers;CPCs,conductivepolymercomposites;CNTs,carbon nanotubes;CTEM,conventionalTEM;CVD,chemicalvapordeposition;DC,directcurrent;EA,ethyleneemethylacrylate;EMI,electromagneticinter- ference;FIT,fluctuationinducedtunneling;HADF-STEM,high-angleannulardarkfieldscanningtransmissionelectronmicroscopy;HCP,hexagonally close-packed;ICPs,intrinsicallyconductivepolymers;ILs,ionicliquids;MWNTs,multi-wallcarbonnanotubes;NTC,negativetemperaturecoefficient;OM, opticalmicroscopy;PDMS,polydimethylsiloxane;PP,polypropylene;PA6,polyamide6;Pc,percolationthreshold;PS,polystyrene;PA,polyamide;PC,poly- carbonate;PVAc,poly(vinylacetate);P3HT,poly(3-hexylthiophene-2,5-diyl);PU,polyurethane;PTC,positivetemperaturecoefficient;PET,poly(ethylene terephthalate);PE,polyethylene;PMMA,poly(methylmethacrylate);PCL,poly(caprolactone);PBT,poly(butyleneterephthalate);PEDOT:PSS,poly(3,4- ethylenedioxythiophene):poly(styrenesulfonate);PVDF,poly(vinylidenefluoride);SMPs,shape-memorypolymers;SEM,scanningelectronmicroscopy;

SPM,scanningprobemicroscopy;SWNTs,singlewallcarbonnanotubes;TE,thermoelectric;TEM,transmissionelectronmicroscopy;THF,tetrahydrofuran;

TPU,thermoplasticpolyurethane;WAXD,wide-angleX-raydiffraction;ZT,ﬁgureofmerit.

∗ Correspondingauthor.Tel.:+862885460953.

∗∗ Correspondingauthor.Tel.:+862885461795;fax:+862885461795.

E-mailaddresses:huadeng@scu.edu.cn(H.Deng),qiangfu@scu.edu.cn(Q.Fu).

http://dx.doi.org/10.1016/j.progpolymsci.2013.07.007

(2)

Please cite this article in press as: Deng H, et al. Progress on the morphological control of conductive network Contents

1. Introduction... 00

2. Background... 00

2.1. Conductor–insulatortransition... 00

2.2. Theoreticalanalysis... 00

3. PreparationofCPCs... 00

3.1. Meltcompounding... 00

3.2. Insitupolymerisation... 00

3.3. Solutionmixing... 00

4. MorphologicalcontrolofconductivenetworksinCPCs... 00

4.1. CharacterisationofconductivenetworkformationinCPCs ... 00

4.2. Morphologicalcontrolthroughpolymerblends... 00

4.3. Morphologicalcontrolthroughthermalannealing... 00

4.4. Morphologicalcontrolthroughashearforce ... 00

4.5. Morphologicalcontrolthroughamixedﬁller... 00

4.6. Morphologicalcontrolthroughothermethods... 00

5. ThemultifunctionalitiesofCPCs... 00

5.1. Strainanddamagesensing... 00

5.2. Stretchableconductor ... 00

5.3. Temperaturesensing... 00

5.4. Vapourandliquidsensing... 00

5.5. Shapememory... 00

5.6. Thermoelectricmaterials... 00

6. Conclusionsandoutlook ... 00

Acknowledgements... 00

References... 00

1. Introduction

Polymer composites, which are one of the most interestingandresearchedareasin nanotechnologyand composites,have demonstrated their potentialas high- performanceandmultifunctionalmaterials.Throughthe incorporationofafillerintothepolymermatrix,thechar- acteristicsofthepolymermatrix,suchaslowdensityand flexibility,canbecombinedwiththemechanicalorother physicalpropertiesofthefiller[1–4].Tremendousefforts havebeendevotedtotheresearchofpolymercomposites toimprovetheirmechanical,electrical,thermal,gasbarrier andotherproperties[1–14].

Due to their ease of processing, tunable proper- tiesandwide rangeofapplications,conductivepolymer composites (CPCs) are one of the most important and interestingareas in polymer composite research,which has remained active for several decades [4,6,15]. Most polymersarenotconductive;conductivefillersareincor- porated into various polymers to fabricate CPCs. With increasingconductive filler content, a jump in conductivity can be observed when a critical filler content is reachedinthepolymermatrix.Thisphenomenonisoften termedtheelectricalpercolationthreshold(Pc).Asshown byboth theoreticaland experimentalstudies, Pcgener- ally decreaseswith increasing filler aspect ratio [6,16].

Therefore, efforts have been made to build conductive networks in a polymer matrix using large aspectratio conductiveﬁllers.Recently,sincetheemergenceoflarge aspectratioand multifunctional conductiveﬁllers, such as carbon nanotubes (CNTs) and graphene nanoplates, this areahasattractedincreasingamounts of attention.

Anumber of recent review papers canbe foundin the literature[1–13,15].

TheelectricalpropertiesofCPCsarisefromtheircon- ductivenetworks.Moreover,itisbelievedthatconductive fillersareseparatedbyathinlayerofnon-conductingpoly- mer(intherangeofafewnm)inCPCs[6,15,17];thus,most of theoverall resistancearisesfromtunnellingbetween nearbyconductivefillers.Theadditionofmoreconductive filler to a particular type of polymer matrix often fur- therenhancestheelectricalconductivityoftheresulting composite,butthisprocedureoftenleadstopoorprocess- abilityanddeterioratedmechanicalpropertiesforrelative highfillercontents.Manystudieshavebeenconductedto reducePcviathemorphologicalcontroloftheconductive networksinthepolymermatrix.Itisbelievedthatahigher fillercontentandabetterdispersionoftheconductivefiller donotnecessarilyresultinahigherconductivity.Thecon- ductivitycanbeenhanced byaddingasmallamountof fillerandbycontrollingitsdistributionandnetworkfor- mationinthematrix.Consideringtheaboveissues, itis reasonable toanticipate thattheelectrical propertiesof CPCscanvarysignificantlyduetoprocessingrelatedissues.

AsshowninFig.1,Kovacsetal.[18]demonstratedthat different processing parameters can signiﬁcantly inﬂu- encetheconductivenetworkmorphologyandPcofCPCs thatarebasedonepoxyandCNTs.Severalothermethods havealsobeenshowntobeabletocontrolthenetwork structure, and, thus, totune theelectrical properties of CPCs.

A number of review papers have beenpublishedon different aspects of CPCs, such as graphite-filled CPCs [3], CNT-filledCPCs[1,4,6,11,15],vapour-grown carbon- nanofibre-filled CPCs[19,20] and the formation of CNT conductive networks in polymer melts [15]. Neverthe- less,themorphologicalcontrolofconductivenetworksin CPCshasyettobesystematicallyaddressedinanyreview

(3)

Please cite this article in press as: Deng H, et al. Progress on the morphological control of conductive network Fig.1.DifferentconductivenetworkmorphologyunderOMandelectricalpercolationbehaviorofCPCsbasedonMWNTandepoxy.Differentprocessing parameterscouldleadtosigniﬁcantdifferenceinelectricalproperties.[18],Reproducedwithpermissionfrom(2007)ElsevierScienceLtd.

despite its importance and recent extensive progress.

Therefore,thistopicisreviewedherefordifferentprepara- tionprocesses.Issuesregardingthemorphologicalcontrol methods,theresultingnetworkmorphologyandtheelec- trical properties are discussed. Moreover, because the morphologyoftheconductivenetworkplaysavitalrole intheﬁnalelectricalproperties,thecharacterisationofthe morphologyusingdifferenttechniquesiscrucial.Hence, recentextensiveprogressonmorphologycharacterisation methodswillalsobeaddressed.

Besidesthefunctionalitiesthatoriginatefromtheelec- tricalconductivity(includingantistatic,electrostaticpaint- ing and electromagnetic interference (EMI) shielding), CPCs have been investigated for many potential elec- troactivefunctionalities,includingstrain/damage[21–26], vapour/liquidandtemperaturesensors[27–32],stretch- ableconductors[33–36],shapememorymaterials[37]and thermoelectricmaterials[38].Duetotheirtunableproper- tiesandeaseoffabrication,anextensiveamountofwork hasbeenconducted,andrecentprogressontheseapplica- tionswillbereviewedanddiscussed.

Thisreview isdividedintoseveralsections. Theﬁrst part,theintroduction,introducesthegeneralbackground of this ﬁeld and the motivation for this review. The second part presentsthe background of this topic. The conductor–insulator transition and a theoretical analy- sisusingtheexcludedvolumetheorywillbeintroduced.

Inthethirdpart,thepreparationofCPCswillbebriefly introduced. Methods, including melting compounding, in situ polymerisationand solutionmixing, willbedis- cussed. In thefourth part,themorphologicalcontrol of conductive networks will be discussed. First, different methodsforthecharacterisationoftheconductivenetwork formationinCPCswillbereviewed.Then,differentmor- phologicalcontrolmethods,includingtheuseofpolymer blends,thermalannealing,shearforce,amixedfillerand othermethods,willbediscussedindetail.Inthefifthpart, themultifunctionalitiesofCPCswillbereviewed,suchas strainanddamagesensing,astretchableconductor,temperaturesensing,vapourandliquidsensing,shapememory propertiesandthermoelectricmaterials.Inthefinalpart

ofthisreview,theabovesectionswillbeconcluded,and theoutlookonfutureresearchtrendsinthisﬁeldwillbe presented.

2. Background

2.1. Conductor–insulatortransition

Electrically conductive polymers are mainly divided into two categories: intrinsically conductive polymers (ICPs), in which the electronic structure of thesepoly- mersis responsible for their conductivity,and CPCs, in which the addition of conductive ﬁllers to the poly- merprovidesthecompositeswithconductivity.CPCsare moreeasilyprocessedandeconomiccomparedwithICPs.

An insulator-conductor transition is observed when an increasingamountofaconductivefilleris addedtothe insulatingmatrix[39].Thistransition,which isawidely researchedtopicinthefieldofcomposites,hasbeeninves- tigatedextensivelysinceoneofthefirstreportedstudiesby Gurland[40].Alargevarietyofconductivefillershasbeen usedtofabricateCPCs.Thesefillersincludemostmetals, carbonaceousfillers(carbonblack(CB),graphite,carbon fibres,carbonnanofibres,CNTs,graphitenanosheetsand graphene),metalfibres,metal-coatedfibresandICPs.For industrialapplications,carbonaceousfillershaveattracted themostattentionduetotheirbalancedproperties;therefore,these fillers willbe themain focus of thecurrent review.

Theconductivity(1/resistivity)ofthecompositeswith increasingamountofconductiveﬁllercanbedescribedby ascalinglawaccordingtoclassicalpercolationtheory:

=0(p−pc)^t (1)

wherepcisthepercolationthresholdoftheCPC,pis theﬁllercontentintheCPC,0isascalingfactorand istheconductivityoftheCPC[26].tisanexponentthat isrelatedtothedimensionalityoftheconductivenetwork withinCPCs.Inmodels,t≈1.3andt≈2.0areusedfortwo- andthree-dimensionalnetworks,respectively.Neverthe- less,manystudieshavereportedlargevariationsinthis

(4)

Please cite this article in press as: Deng H, et al. Progress on the morphological control of conductive network exponentfrom1to12[6,41].Thereareclaimsthatalower

tvalueiscausedbyanarrowtunnellingdistancedistribu- tioninthesystemandthatahighertvalueiscausedbya broadtunnellingdistancedistribution[26].Nevertheless, thisdifferencebetweenthetheoreticalpredictionandthe experimentalobservationsisstillanunresolvedissue.

Theinsulator-conductortransitioninCPCswasinitially thoughttobeassociatedwiththeformationofnetworks with electrically connected filler particles extending throughout the whole sample. Currently, it is widely thoughtthatelectronscanflowthroughaninsulatingbar- rierviaquantummechanicaltunnellingbetweenadjacent conductiveregions.Thus,conductivenetworksconsistof tunnellingbetweenlocalconductivefillers.Thisbehaviour canbedescribedbyafluctuation-inducedtunnellingmodel [26,42].Suchamodeltakesintoaccounttunnellingthrough potentialbarriersofvaryingheightduetolocaltempera- turefluctuations.Fortheinter-nanotubegapwidthw,the conductivityofcompositesataconstanttemperaturecan beexpressedas:

lnDC∝−w (2)

w∝−p^−1/3duetospatialconsiderations,wherepisthe contentofconductiveﬁllersintheCPC[43].Therefore,Eq.

(2)canbeexpressedas:

lnDC∝−p^−1/3 (3)

BecausemostconductiveﬁllersinCPCsarenotindirect contactwitheachother,theconnectednessisthoughtto becrucialforthetransferofanelectriccurrent[44].Addi- tionally,thetunnellingdistancebetweentheelectronsof nearby conductive regions is believed to dominate the chargetransfer[15].Therefore,mostoftheresistancein thesystemisfromthetunnellingresistancebetweenthese localconductivenetworks.

2.2. Theoreticalanalysis

Theaspectratioandorientationofaconductivefiller areconsideredvitalfortheelectricalproperties ofCPCs [16].Toinvestigatetheeffectsofthesetwoissues,analyt- icalmodels,suchastheexcludedvolumetheory[45,46], canbeused.Theexcludedvolumeofanobjectisdefinedas theregionofspacethatthecentreofanothersimilarobject cannotentertoensurethatthesetwoobjectsdonotover- lap.Suchatheorycanpredictthepercolationthresholdof CPCsthatconsistofhigh-aspect-ratiocylinder-shapedpar- ticlesandanon-interactingmatrix.Thistheoryhasbeen widelyusedtodescribebothmicro-andnanocomposite systems[16,46].Fig.2showsacomparisonbetweensome resultsfrom theliterature and a theoretical calculation usingtheexcludedvolumetheory.Thepercolationthresh- oldispredictedtodecreasewithincreasingfilleraspect ratio.Thisexpectationhasbeenconfirmedbyanumber ofreportedstudiesintheliterature[6,43].Regardingthe effectofthefillerorientationonthepercolationthresh- old, a perfectly aligned conductive network has a very highpercolationthreshold,andthisvalueisconstantwith increasingaspectratio.Itshouldalsobenotedthatsome ofthereportedresultsintheliteraturearewellbelowthe theoreticallypredictedpercolationthreshold(seeFig.2).

Fig.2.Theoreticalelectricalpercolationthresholdversusconductiveﬁller aspectratioforalignedandisotropiccompositesystems(theshadowed area)accordingtocalculationusingexcludedvolumetheorytogetherwith experimentaldatareportedbyvariousreferencescitedintheworkfrom Dengetal.[16],Reproducedwithpermissionfrom(2009)Wiley-VCH.

Inthesesystems,particlescanmoveandre-aggregateto inducetheformationofconductive networks.Thisphe- nomenonisoftenobservedinsystemswitharelativelylow viscosity.Suchapercolationbehaviourcanbedefinedbya kineticpercolationthreshold[6].Otherdatafitthetheory quitewell,andthisbehaviourcanbedefinedbyastatisti- calpercolationthreshold,atwhichtheconductivephaseis randomlydispersed.

Ithasbeensuggestedthattheinteractionbetweencon- ductivefillers(especiallycarbon-basedfillers)needstobe takenintoaccountduringprocessing[6,15].Therefore,to achieveanultra-lowpercolationthreshold,carefulatten- tionshouldbepaidtothepreparationofCPCstofullyutilise thelargeaspectratiooftheseconductivefillers(suchas CNTs,grapheneandcarbonnanofibres).Alargenumberof studieshavebeenconductedonthemorphologicalcontrol ofconductivenetworksinCPCs.Asystematicreviewwill beperformedinthefollowingsections.

3. PreparationofCPCs

The preparation of CPCs involves the selection of a suitable mixing methodtoincorporate conductiveﬁller intothepolymermatrix.Asatisfactoryﬁllerdispersionis requiredtoobtainareasonableprocessabilityofthecom- posites.Moreimportantly,conductivenetworksneedtobe constructedintheCPCstoobtainthedesiredelectricalcon- ductivity.Generally,therearethreetypesofmethodsfor thepreparationofCPCs:meltcompounding,insitupoly- merisationandsolutionmixing.

3.1. Meltcompounding

Meltcompounding is aneffective method for incor- poratingconductive ﬁllers intoa viscous polymer melt.

The advantages ofthis technique are that thefillercan bedirectlydispersedintothematrix,nochemicalmodi- ficationsarerequiredandthefillersareprohibitedfrom re-aggregationbytheviscouspolymermatrix.Moreover, this methodfitsquitewellwithcurrentindustrialprac- tices. Many studies have demonstrated the successful

(5)

Please cite this article in press as: Deng H, et al. Progress on the morphological control of conductive network applicationofmeltcompoundingindispersingconductive

ﬁllers intovariouspolymer matrices;thedetailscanbe foundina numberofreviewpapers[3,5–7,13,19].Here, severalrecentstudiesontheeffectoftheprocessingcon- ditionsontheﬁllerdispersionandotherpropertieswillbe emphasised.

Huang etal.[47] reportedastudyonthemeltcom- poundingofpolydimethylsiloxane(PDMS)withMWNTs.

Therealpartofthecompositeviscositywasrecordeddur- ingmixing.Therefore,aquantitativeunderstandingofthe CNTdispersioninthepolymermatrixcouldbeobtained bymeasuringtheviscositychangesasa functionofthe nanotube-polymermixingtime.Everybatchwiththesame concentrationtendedtoexhibitasimilardispersionlevel whenmixedforalongenoughtime,t>t^*(thecriticaltime needed).Agooddispersionisindicatedbyaplateauinthe viscosityresponse.Thehighertheconcentration,thelonger thet^*neededtoachievearelativelygooddispersion.How- ever,inmostofthestudies,thesamemixingtimewasused forcompositeswithdifferentﬁllercontents,andthesemix- ingproceduresmightbetooshorttoachieveagoodﬁller dispersion.

Inadditiontothemixingtime,otherprocessingparam- eters also play important roles in the ﬁller dispersion quality and theresulting conductive network structure.

Villmowetal.[48]conductedasystematicstudyonthe meltprocessingofCNT/polymercomposites.Theauthors investigated the effect of different processing parameters onthe final properties, especially onthe electrical properties ofthecomposites. Theprocessing conditions significantly influencedtheresidencetime of themate- rials and the filler dispersion. Increases in the rotation speedandthethroughputdecreasedtheresidencetime.

The use of back-conveying elements and an extension of the processing length had opposite effects. Besides thesemachineparameters,thedesignofthescrewpro- filescanfurtherimprovethefillerdispersion,e.g.,theuse of distributivescrewconfigurationsthat containmixing elements.Theremainingmacroscopicagglomeratescould be completely dispersed using a following masterbatch dilution process,whichresulted ina verylow electrical percolationthresholdof0.24vol.%MWNTs.

Moreover,theinteraction betweenthefillerand the polymer matrixplays a significantrolein thefillerdis- persionduringmeltcompounding.Therefore,thechemical polarities of thepolymer matrix and the fillercrucially influence the final quality of the filler dispersion. For instance,conductivefillers(suchasCNTs)exhibitedlarge aggregatesinpolypropylene(PP)whenmeltcompounding wasusedasthedispersionmethod[16,49,50].Thisresult iscausedbythenon-polarnatureofthePPpolymerchain.

Thesamefillercanbeeasilywell-dispersedinapolyamide 6(PA6)matrixduetothestronginteractionbetweenthe PA6polymerchainsandtheCNTs[51].Therefore,compat- ibilisersorsurfactanthaveoftenbeenusedtoimprovethe interactionbetweenthefillerandthematrix,and,thus,the fillerdispersioninanon-polarpolymermatrix[50,52–54].

Overall,meltcompoundingisaneffectiveandefﬁcient way to disperseconductive ﬁllers in polymer matrices.

However,moreattentionneedstobepaidtothecritical mixing time and theshear stressinside the mixer.Too

highshearstressesarenotrecommendedduetoamas- sivereductionintheﬁlleraspectratioduringprocessing.

Forhighshearmixingprocesses,anoptimummixingtime betweentheﬁllerbreakageanddispersionstatesneedsto beachievedtooptimisethepropertiesofthenanocompos- ites[50].

3.2. Insitupolymerisation

In-situpolymerisationisanothertechniquefordispers- ingconductivefillersinapolymermatrix.Theadvantage ofthisprocessisthatthepolymerchainandthefillerscan bedispersedandgraftedonthemolecularscale.Thispro- cessgivesexcellentanfillerdispersionandapotentially goodinterfacialstrengthbetweenthefillerandthepoly- mermatrix.Successfulinvestigationshavebeenreported intheliteraturefromdifferentgroups[55–59].Auniform dispersionofthefillerwasobtainedandimprovedboththe mechanicalandelectricalproperties.Recently,theinsitu polymerisationmethodwasusedtofabricateCPCscon- taining graphene. A relatively hightemperature during polymerisationwasutilisedtoreducegrapheneoxideinto grapheneinthepolymermatrix[60],thusobtainingCPCs attheendofthepolymerisationprocesswithoutfurther processing.Nevertheless, itis moredifficult toadaptin situpolymerisationthanmeltcompoundingasageneral methodforthepreparationofCPCstoindustry.

However, the in-situ polymerisation process is an essential method for the preparation of thermoset and rubber-basedCPCs. Asagood example,epoxyhasbeen extensivelyinvestigatedasapolymermatrixfortheprepa- rationofarangeofCPCs[61–64].FortheseCPCs,abetter control of the conductive network structure [18] and, hence,theelectricalproperties [65]canberealiseddue to the special preparation method that is used, e.g., a vacuumbag,ﬁbrelay-upandothermethodswithaprede- ﬁnedconductivenetworkstructure,beforepolymerisation.

Similarly, silicone rubber-basedCPCs can alsobe fabri- catedthathavespecialelectricalorelectroactiveproperties [22,66–68].

3.3. Solutionmixing

Physicaltechniques, suchasmelt compounding, can leadtoscalabledispersions,butlocallyhomogeneousdis- persion states are difﬁcult toachieve withoutbreaking down the entangled ﬁllers (such as CNTs). Thus, other methodsneedtobeconsidered,suchassolutionmixing.

Nevertheless,somecharacteristicsof thefillers, suchas thesp2hybridisedstructureofthenanotubes,makethese fillersinsolubleinacommonorganicsolvent.Theintro- ductionofafunctionalisationonthefillersurfacesmight makethemsolubleandstrengthentheinteractionbetween thefillerandthematrixinthecomposites.Therefore,the choiceof therightlevelandtypeofsurface functionali- sationcanhaveaveryimportanteffectonthecomposite properties.

Duetothereasonsmentionedabove,therehasbeen alotofinterestinusingthechemicalfunctionalisationof conductiveﬁllersurfaces(suchasthefunctionalisationof CNTsortheoxidationofgraphite)tomaketheﬁllermore

(6)

Please cite this article in press as: Deng H, et al. Progress on the morphological control of conductive network solubleinagivensolventorpolymermatrix.Inthecase

oftheorganicsolventmixingmethod,ahomogenousdis- persioncanbeachievedthroughoutthesolventand,thus, thehostmatrix.Nevertheless,chemicalfunctionalisation isoftenreportedtoreducetheelectricalconductivityof CPCsduetothedestructionoftheconductiveﬁllerduring thesesurfacemodiﬁcationprocesses.Anextensivereview ontherelatedliteraturecanbefoundelsewhere[11,13].

Itisworthpointingoutthatﬁlmcastingisoftenusedasa methodtoprepareCPCsaftersolutionmixing;thisprocess oftentakesmorethan24h.Comparedwiththesolidiﬁca- tionprocessaftermeltprocessing(whichistypicallyless thanafewminutes),thisprocessoftengivesanadequate amountoftimefortheconductivenetworktore-aggregate inCPCs.Therefore,theelectricalpercolationthresholdthat isobservedinCPCsfromsolutionmixingislessthanthat frommeltcompounding[6].

4. Morphologicalcontrolofconductivenetworksin CPCs

4.1. Characterisationofconductivenetworkformationin CPCs

Becausethemorphologyof theconductivenetworks hasanimportantinﬂuenceontheelectricalpropertiesof CPCs,itiscrucialtocharacterisethemorphologicaldetails ofthese networks.Researchershave demonstrated that variousdirectandindirectmethodscanbeusedtocharac- terisethemorphologiesofnanoﬁllersandtheirconductive networksinCPCs.

A range of microscopic methods, including optical microscopy (OM) [48], scanning electron microscopy (SEM)[43,69–74],transmissionelectronmicroscopy(TEM) [75–77], scanning probe microscopy (SPM) [78,79] and atomicforcemicroscopy(AFM)[78,79],canbeusedforthe directobservationofconductivenetworksinnanocompos- ites.AsshowninFig.3,OM,TEM,SEMandHAADF-STEM havebeen usedby differentresearchgroups tocharac- terisethemorphologyofCPCsfromdifferentperspectives.

Thesemethodshavebeenwidelyusedasgeneralmicro- scopicmethodstocharacterisethemorphologyofpolymer compositesfromdifferentaspectsorondifferentscales.

OMisoftenusedtostudythemorphologyovertherange ofafewmicronsorabove.Foranythingbelowthisrange, SEM,AFMorTEMisneeded.Itis wellknownthatonly informationnearthesurfaceofspecimensiscapturedby conventionalSEMbecausesecondaryelectronshavearel- ativelyshallowescapedepth(5–50nm)duetotheirrather lowenergylevels[71].However,itwasrecentlyreported thattheSEMobservationofdeeplyembeddedCNTsand anassessmentoftheoverallCNTdispersionstatuswere possiblebyusingvoltagecontrastimaginginCPCsthatare basedonCNT/polymer[43,70–74,80].Thiscontrastmech- anismwasﬁrstreportedbyChungetal.[81]andhasbeen adaptedbydifferentresearchgroups[43,70–74,80].

Loosetal.usedconventionalSEMinthechargecon- trast imaging mode to investigate the morphology of networksthatwereconstructedwithsingle-wallcarbon nanotubes(SWNTs)[80]andgraphenesheets[82]embed- dedinpolystyrene(PS)matrices.Theauthorsreportedthat

thechargecontrastimagingoftheconductivenetworks under high acceleration voltages could provide three- dimensionalinformationonthestructureoftheconductive networks.Thismethodhasbeengainingmoreattention recently.Bauhoferetal.[70]conductedaseriesofworks toobtaindetailedinformationonthevisualisationofCNTs ininsulatingmatrices.AsshowninFig.4,factorsthathave importanteffects onthevisibilityofCNTsinCNT/epoxy composites, including theimaging mechanism, imaging depth, parameters, requirement of sample conductivity and SEM detectortype, were investigated [71].Further studyrevealedthatCNTsarestillvisiblebelow1kVand exhibit abright contrasttothedarkepoxymatrix.This resultisinterestingbecauseimagingatalowaccelerating voltageisregardedasabetteralternativetohighvoltage imaging forinvestigating CNTdispersionsintheseCPCs becausethechargingofthespecimenscanbeavoidedat alowvoltage.Finally,thekeytosuchanimagingmethod appearstobe(1)thedetectionofsecondaryelectronsthat areexcitedintheelectronbeamimpactareaand(2)theuse ofanappropriatedetectorthatissensitivetoslightcharges onthesamplesurfaces.

SEM can certainly offer valuable information on the morphologies of nanofillers and their conductive networks. However, the actual size of the nanofiller and the detailed information onthelocal network that isobtainedfromSEMarenotveryaccuratedue tolocal charging of the polymer matrix around the nanofillers.

TEMandAFMcanbeutilisedtocomprehensivelyinves- tigatetheseissues.High-angleannulardarkfieldscanning transmissionelectronmicroscopy(HAADF-STEM)hasbeen investigatedasausefultoolforobtainingareliablequan- tificationofimagestoenhancethecharacterisationofthe conductivenetworkmorphology[75].Intermsofpolymer materials,STEM hasa number of advantages over con- ventionalTEM(CTEM):theimagesareeasytointerpret duetoalackofphasecontrast,thesignalintensityislin- ear withthickness variationsanda highsignal-to-noise ratioisobtained.Theseadvantagesaremorepronounced when a high-angleannular darkfield (HAADF)detector that is capable of single-electron counting is used. For carbon-nanoparticle-filledsystems,HAADF-STEMcancre- ateexcellentcontrastbetweendifferentcomponents,as showninFig.3.Generally,itisthoughtthatHAADF-STEM canbeusedasapowerfultoolforobtaininghigh-resolution imagesofunstainedpolymersystems.

AFMisapowerfulmethodforthecharacterisationofthe topographyandpropertiesofsolidmaterials[83].AFMis equippedwithasharpprobeforscanningacrossthesam- plesurface.Theprobe-sampleinteractionsarerecordedto generatemapsofthematerialtopographyandthematerial properties,includingthemechanical,electricalandmag- neticproperties.Tkalyaetal.[82]usedconductiveatomic force microscopy (C-AFM)to investigatethe localmor- phologyofgrapheneinCPCs.Moreover,theconductivity distributionandthedetectionofpercolatingpathscanalso beobtainedwithnanometreresolution[84].

Allconventionalmicroscopytechniqueshavetheirspe- cific disadvantagesintermof imagingthedispersionof nanofillerswithina polymermatrix:opticalmicroscopy onlycharacterisesverylargeagglomeratesofnanofillers,

(7)

Please cite this article in press as: Deng H, et al. Progress on the morphological control of conductive network Fig.3.Morphologycharacterizedwithdifferentmethods:OMandTEMforPCcontaining0.688vol.%CNTs;SEMinchargecontrastmodeshowsthe distributionofMWNTsinPPmatrix;andHAADF-STEMpicturesshowsindividualCBparticlesandtheirclustersinpolymercomposites:OMandTEM, [214],Copyright2011.ReproducedwithpermissionfromElsevierScienceLtd.;SEM,[123],Copyright2009.ReproducedwithpermissionElsevierScience Ltd;andHAADF-STEM,[75],Copyright2009.ReproducedwithpermissionfromAmericanChemicalSociety.

whilesurface-basedmethods,suchasSEMorAFM,only characterise the surface or cross-section of the three- dimensionalarrangementofnanofillers,anditisdifficult todrawconclusionsonthebulkorganisationofthecom- posite fromtheTEMimages ofthin sections(thickness of∼100nm).Therefore,auxiliarymethodsareneededto providemoreinformationonconductivenetworks,suchas theorientationoftheconductivefillerandtheformation oftheconductivenetworks.Manymeasurements,including electrical conductivity, wide-angle X-ray diffraction (WAXD) and Raman spectroscopy, can be used to fur- thercharacterisethemorphologyofconductivenetworks [73,85]. From WAXD and Raman spectroscopy,the ori- entationsofthefillerandthepolymermatrixaswellas thepolymer-nanofiller and nanofiller-nanofillerinterac- tionscanbeestimated.Thedispersionofthenanofillers andthemorphologyoftheconductivenetworkscanthen bepartiallydeterminedusingthisinformation.Addition- ally,aconductivitymeasurementisoftenusedasabasic methodforevaluatingtheformationofelectricallyconduc- tivenetworksandgivesdirectevidenceoftheconductive networkmorphologyinCPCs.Alotofstudieshavebeen conductedtocharacterisetheconductivityofCPCs.Asa generalmethodformonitoringaconductivenetworkdur- ingprocessing,insituelectricalmeasurementshavebeen widelyusedtorecordtheresistivityofCPCs[73,76,86–88].

Thismethod canprovide information ontheformation anddestructionofaconductivenetworkduringdifferent processes,includingextrusion[89,90],thermalannealing [73,88]andshearing[15].Furthermore,rheologyhasalso beenusedtoinvestigatetheformationofanetworkwithin apolymermatrix[91–94].It wasdemonstratedthatthe detectionofapolymer-fillernetworkwithrheologycanbe stronglycorrelatedtothedetectionofafiller-fillerconduc- tivenetworkwithanelectricalconductivitymeasurement [93,95].

4.2. Morphologicalcontrolthroughpolymerblends

Polymerblendsarearesearchtopicthathasbeenwidely investigated.Byconstructingpolymerblendwithtwoor morepolymers,theadvantagesofthesepolymerscanbe integrated;thus,balancedand optimisedproperties can beobtainedforthefinalmaterial.Furthermore,thephase morphologyofthepolymerblendsplaysacrucialrolein thefinalproperties;therefore,polymerblendswitharange ofpropertiescanbedesignedand fabricatedbycontrol- lingtheirmorphology.Interms ofpolymer-blend-based CPCs,conductivefillerscanbeselectivelylocatedinoneof thepolymerphasesorattheinterfaces[96–100];furthermore,thevariousphasemorphologiesinpolymerblends canbeusedtoregulatetheconductivenetworkswithin

(8)

Please cite this article in press as: Deng H, et al. Progress on the morphological control of conductive network Fig.4.IllustrationofimagingmechanismunderSEMinchargecontrastmodeforCPCsbasedonepoxyandCNTsfromdifferentpreparationmethods:

theseCPCs[101–103].Therefore,polymerblendscanbe usedtocontroltheconductivenetworkmorphologyand electricalpropertiesofCPCs[103].

Double percolation has been widely investigated to reducethepercolationthresholdofCPCs[104].Inthistype ofpolymerblend,atleastoneofthepolymerphasesiscon- tinuous,andtheﬁllerislocatedinthiscontinuousphase.

ComparedwithCPCsthatarebasedonasinglepolymer, thepercolationthresholdoftheseCPCscanbesignificantly reducedduetotheselectivelocalisationoftheconductive networks.Therefore,theelectricalpropertyislargelyinflu- encedbythelocationoftheconductivefilleraswellasthe

morphologyofthepolymerblends.Thelocalisationofthe conductiveﬁllerdependsonthebalanceoftheinterfacial energiesandcanbepredictedbycalculatingthewetting parameter,ωaa,whichisdeﬁnedinEq.(4)[96].

ωa=ﬁller-polymer1−ﬁller-polymer2

polymer1,2 (4)

Inthisequation,xdenotethedifferentinterfacialten- sions betweenthefillerand polymers1and 2, andthe interfacialtensionbetweenthetwoblendphasesisinthe denominator.Thefillertendstobelocatedinpolymer1 ifthewettingcoefficientislessthan−1andtendstobe

(9)

Please cite this article in press as: Deng H, et al. Progress on the morphological control of conductive network locatedinpolymer2ifthewettingcoefﬁcientisgreater

than1.Furthermore,ifthewettingcoefﬁcientisbetween

−1and1,thefillerismorelikelylocatedattheinterface betweenthetwopolymers.Thiscoefficientcanbeusedas anindicatorofthethermodynamictendencyofthefillerto localiseinaparticularareainanimmiscibleblend.Sucha tendencycorrespondstotheminimisationoftherelevant interfacialtensions.

ElectricaldoublepercolationinCPCsthatarebasedon polymerblendscanberealisedviaaselectivelylocalised conductivefillerinoneofthepolymerphases.Thethermo- dynamicfactorsthatwerediscussedaboveoftendetermine thedistributionoftheconductivefillers. Severalgroups [104–106] have shown that the choice of the polymer matrixcansignificantlyinfluencethedistributionofthe conductive fillerinpolymer blendsdue tothedifferent interactionsbetweenthesepolymersandthefiller.Con- ductivefillersareoftenlocatedinthepolymerphasethat theyhavemoreinteractionwith.Thereareothermeth- odsthatcanbeusedtomodifythelocationofthesefillers.

Lietal.[107]showedthatthefunctionalisationofaﬁller canchangethedistributionoftheﬁllerinpolymerblends.

Moreover,Potschke et al.[100]used a reactivecompo- nenttotunetheinteractionbetweenSANandMWNTsin apolymerblendthatconsistedofPCandSAN.Theauthors observedthatMWNTsaretransferredfromthePCphaseto theSANphaseaftertheadditionofthereactivecomponent duetoanenhancedinteraction.

Otherthanthermodynamicissues,kineticfactors,such asthemixingproceduresorsequence,blendingtimeand shearstrength,alsoplayimportantrolesinthelocalisa- tionoftheconductivefiller[96,98,106].Ithasbeenshown thatthelocalisationoftheconductivefillercanbechanged bycontrollingkineticissues,suchasthemixingsequence [96,98,108,109]. Moreover,the aspectratio and surface roughnessoftheconductivefillerandtheviscosityofthe polymermatrixarealsoinfluentialfactors[99,110].

Toreduce thepercolation thresholdin CPCsthat are basedonpolymerblends,thelocalisationoftheconduc- tivefilleratthecontinuousinterfaceispreferredbecause oftheirlowvolumefraction.Manystudieshavefocused onachieving this difficulttask. For CPCsthat utilise CB asaconductivefiller,severalstudieshavebeenreported [109,111,112]. Bailly and co-workers [96,97] conducted a series of studies to localise MWNTs at the interface ofpolyamide(PA)/ethylene-methylacrylate(EA)random copolymerblends.Theauthorsobservedirreversiblepoly- merabsorptionontheMWNTsurfacesduringthetransfer fromonephasetotheother.TheseMWNTsaretrapped at the interface due to this absorption [96]. Recently, Potschkeetal.[99]illustratedthatthetransferofafiller from one phase to theother wasdriven by thermody- namicfactorsandwasinfluencedbytheshapeofthefiller.

Theauthorscompared CBwithMWNTsina blendcon- tainingpoly(styreneacrylonitrile)andpolycarbonate(PC).

Thehigh-aspect-ratioﬁller,MWNTs,transferredfromone phasetotheothereasierthanCBbecausethelow-aspect- ratioﬁllerappearedtohavemoreresistanceduringtransfer andwasthereforetrappedattheinterface.Itwassuggested thatthesphericalshapeandlowaspectratioofCBmade itmorestableattheinterfacethanMWNTs(seeFig.5).It

isratherdifﬁculttoobtainameasurableconductivityfor CPCsthatcontainalargeaspectratioﬁller(suchasCNTs) locatedattheinterface.OnlyrecentlydidWangandco- workers[113]reportaselectivedistributionofCNTsatthe interfaceofimmisciblePC/acrylonitrile-butadiene-styrene (ABS)byusingacombinationofmaleicanhydridegrafted ABS(ABS-g-MA)andanappropriateprocessingprocedure.

Byadjustingthekineticissuesaswellasthethermody- namicissues,thepreparedCPCsdemonstrateapercolation thresholdof0.05wt%.

4.3. Morphologicalcontrolthroughthermalannealing

ThermalannealinginvolvestheheatingofCPCstoabove their glass transition temperature or melting temperature.Throughsuchaprocess,thepolymermatrixbecomes mobile.Quite often, this mobility can betransferred to the imbedded conductive ﬁller and trigger the forma- tionofconductivenetworks.Thermalannealinghasbeen usedasaneffectivemethodforinducingtheformationof conductivenetworksin CPCsthatcontainvarious types ofconductiveﬁllers[15,32,43,73,88,114–116].Asshown in Fig. 6, thermal annealing can repair CNT conductive networksthatweredestroyedbyshearinginarheometer.

Amorphologicalstudyconﬁrmedthatthere-aggregation ofCNTscanbetriggeredinapolymermelt,leadingtothe formationofaconductivenetwork.

Thereal-timetracingoftheelectricalresistivityduring isothermalannealingforisotropicCPCshasbeenwidely investigated[73,90,117–120].Apercolationtimeisneeded forthecompositesthatareannealedabovetheirglasstran- sitiontemperaturetotriggertheformationofconductive networks.Itwasconcludedthatpercolationisdelayedby thebulkmobilityofthepolymerlayeraroundtheconduc- tiveparticles.Therefore,theformationoftheconductive networkissignificantlyinfluencedbythethermodynamic interactionbetweenthefillerandthematrix[119,120].

Manystudieshavebeenconductedtounderstandthe dynamicprocessofconductivenetworkformationinCPCs bycombiningexperimentalobservationsandtheoretical modelling [73,118,120–122]. Variousmodels have been proposedtoexplainthedatathatwasobtainedfrominline electrical measurement using modiﬁed percolation the- ory[73][122,123]ortheuniversalinterfacialfreeenergy model [120,124]. A relaxation time or percolation time wasobtainedfromtime-dependentconductivitymeasure- ments,anditsrelationwithtemperaturewasdescribed by an Arrheniusequation withactivation energies that werevery close tothe values that wereobtained from rheologicalexperiments.Thepercolationtimecanbechar- acterisedastheannealingtimeatwhichtheconductivity startstoincreasedrasticallyduringthedynamicpercola- tionprocess.Thetheoreticalanalysisandtheexperimental resultsrevealedthatthepercolationtimeisdirectlycorre- latedtothezero-shear-rateviscosityofthepolymermatrix, regardlessoftheﬁllerconcentrations[119,124].However, becausetheseanalyseswereperformedforisotropicCPCs, thetruemechanismfortherelaxationordynamicpercola- tionofhighlyorientedCPCsisstillnotclear.

Recently, we described a newconcept in which the skin layers in a highly oriented bicomponent structure

(10)

Please cite this article in press as: Deng H, et al. Progress on the morphological control of conductive network Fig.5.(a)Fillerwithlowaspectratiolocateattheinterfaceofpolymerblendsduringmeltcompounding.Thefigureshowsthemechanismofdecreasing drivingforce(Fcurvature)duringthetransferbetweentwopolymers.(b)Fillerwithlargeaspectratioatinterfaceofpolymerblends.Theillustrationshows thedrivingforceisnotdecreasingduringsuchatransferprocess.Typicaldistributionstatusforlow(c)andhigh(d)aspectratiofillersthatwerefirstly dispersedinablendphasewithrelativepoorwetting(indicatedasyellow)andsubsequentlymeltcompoundingwascarriedoutwithanotherpolymer whichhasmorepolarnature(blue).[99],Copyright2011.ReproducedwithpermissionfromtheAmericanChemicalSociety.

can be thermally annealed above their melting temperatureduetothesupportofaneatpolymerphase witha highermeltingtemperature[43,73,74].Insucha system,theformation,destructionandreconstructionof theconductivenetworkscanbemonitoredduringsolid statedrawingandannealing[16,73].Thisstudyprovided interestinginformationonchangesinthehighlyoriented conductivenetworkduringrelaxation.AsshowninFig.7, themorphologyoftheconductivenetworkchangedfrom anisotropic toahighlyorientedstatedue tosolid state drawing.Asubsequentannealingprocessledtotherelax- ationofthehighlyorientednetwork;thus,aconductive network consisting of relaxed oriented bundles with

“hairy”localcontactswasobtained.Duringannealing,the resistivityoftheseCPCswassignificantlyreduceddueto theformationoftheseconductivenetworks.Theresistivity ofpolymerfibres/tapesthatwereproducedfromsucha meltprocessing-basedmethodwascomparabletothatof thefibres/tapesthatwereproducedfromsolution-based orother,morecomplicated,methods[73].

4.4. Morphologicalcontrolthroughashearforce

Shearforcesareinvolvedinvariousprocessingmeth- ods,includingextrusion,injectionmoulding,spinningor stretching.Theseforcesoftenplayimportantrolesinthe ﬁllermorphologyviathestresstransferbetweenthematrix andtheﬁller.Thus,themorphologicalcontrolofthecon- ductivenetworksinCPCscanberealisedbyusingashear force[16,18].

To understand the effect of shear on the network morphologyandtheresultingelectricalproperties,many studieshave beenconducted.Shear-inducedorientation isoftenobservedduringinjectionmoulding[125],spin- ning[126]orstretching[16].Anelectricalanisotropycan alsobeobservedasaresultoftheorientationinthecon- ductivenetwork[16].AsshowninFig.8,ourrecentwork demonstratedthatsolidstatedrawingcanleadtoorienta- tionina MWNTconductivenetwork,and ananisotropy can be detected in its electrical resistivity. Eken et al.

[127] conducted a simulation study to investigate the

(11)

Please cite this article in press as: Deng H, et al. Progress on the morphological control of conductive network Fig.6. (Topleft)Sketchofexperimentsetupforelectricalmeasurementundershearandannealinginarheometer.[91],Copyright2008.Reproduced withpermissionfromElsevierScienceLtd;(bottomleft)electricalconductivityofCPCsbasedon0.5wt.%MWNTandPCmatrixundershearandthermal annealing,(right)themorphologyofMWNTnetworkinCPCshotpressedat190^◦C(up),and260^◦C(down).[76],Copyright2007.Reproducedwith permissionfromWiley-VC.

effectofshearonthemicrostructureandelectricalprop- ertiesofCNT/polymercomposites.Theauthorsobserved that the rateof the shearﬂow inﬂuenced thecompos- iteconductivitybytriggeringtheformationordestruction

ofconductivenetworks.Itshouldbenotedthatconduc- tiveclustersformedandbrokesimultaneouslyduringthe simulationwitharelativelylowﬁllerconcentration.This phenomenondirectlyledtolargeconductivityﬂuctuations.

Fig.7.SEMofMWNT/co-PPcompositesurfacefor(a)isotropicﬁlm,whererandomlydispersedMWNTsareobserved;(b)solid-statedrawntape,where highlyorientedMWNTsareobserved;and(c)solid-statedrawnandsubsequentthermalannealedtape(abovetheirmeltingtemperature),whererelaxed anisotropicnanotubebundlestructureareobserved,[74],Copyright2009.ReproducedwithpermissionfromWiley-VCH;(d)Insituelectricalmeasurement ofdrawntapesatdifferentannealingtemperatures.[73],Copyright2010.ReproducedwithpermissionfromWiley-VCH.

(12)

Please cite this article in press as: Deng H, et al. Progress on the morphological control of conductive network Fig.8. (Top)SEMimageshowingorientationofMWNTinPPmatrix

after solid state drawing, (Middle) the electrical resistivity of dif- ferentCPCsat differentdraw ratios. (Bottom)the anisotropy index (Resistivitytransverse/ResistivityLongitudinal)ofdifferentCPCsatarangeof drawratios.[16],Copyright2009.Reproducedwithpermissionfrom Wiley-VCH.

Forrelativelylargefillercontents,theconductivityfluctua- tionsdecreasedduetothepersistenceofthesepercolating clustersundershear.Furthermore,nanotubeagglomera- tionisanimportantfactorintheelectricalconductivityof CNT/polymercompositesbecauseanagglomeratedstruc- tureis beneficial for network formation. Shear-induced conductivenetworkchangesinCPCsweresystematically investigatedbyAligandco-workers[15,89].Theauthors demonstratedthat shearcouldconstructor deconstruct

conductive networks depending on the initial network morphology.WhenshearwasappliedtoCPCsthatwere basedonpolymerblends,apronouncedorientationwas observedforthenetworkofthecontinuouspolymerphase aswellasforthenetwork oftheconductive ﬁller.Both networks played important roles in the ﬁnal electrical properties,andtheshear-inducedchangesintheelectrical propertiescouldbetunedbymodifyingtheblendcompo- sition(seeFig.9)[110].

4.5. Morphologicalcontrolthroughamixedﬁller

Toachievealowerpercolationthresholdoralowercost, morethanonetypeofﬁller,particularlyﬁllerswithdiffer- entaspectratios,canbeusedtoprepareCPCs[128–134].

Thesehybridfillers,whichcontaindifferentcarbonfillers, suchasCNTs,graphite-basedfillersandCB,areoftenused duetotheirdistinctlydifferentaspectratiosandsimilar carbon-basedstructures.Atheoreticalstudyshowedthat itisnotnecessarytobuilda conductivenetworkwitha high-aspect-ratiofilleralone[131]andthatthepercolation thresholdissensitivetotheportionofhigh-aspect-ratio fillerinasystemthatcontainsfillerswithdifferentaspect ratios[135]. Thereare a number of reportedinvestiga- tionsintheliteratureonmixedcarbon-filler-filledCPCs [128,131,132,136–138].AsshowninFig.10,thepercola- tionthresholdsofthesesystemsareoftennearorbelow theaverageofthesystemsthatarefilledwitha partic- ulartype of carbon filler,indicating thata considerable amountofahigh-aspect-ratioandhigh-pricefillercanbe replaced witha low-aspect-ratio and low-price filler. It hasbeenreportedthatalow-aspect-ratiofillercanactas a bridge betweenthenetworks that areformed bythe high-aspect-ratio filler [137]or form special configura- tionconductivenetworkswiththehigh-aspect-ratiofiller [138].

Hybridﬁllersthatcontainbothanon-conductiveand aconductiveﬁllerhavealsobeenusedtofabricateCPCs.

Bilottietal.usedacombinationofneedle-likeclaywith CNTs;itwasshownthattheadditionofthenon-conductive fillercouldacceleratethedynamicpercolationbehaviour oftheCNTconductivenetwork[139].Grunlanandhisco- workers[128]observedthattheadditionof0.5wt%clay couldincrease theelectrical conductivityof CPCsbyan orderofmagnitude.Theauthorsindicatedthattheinter- actionbetweenCBandtheclaymightbethecauseofthe improvedelectricalproperties.Inaseparatestudy[130], LiuandGrunlanobservedthatSWNTsappearedtohavean affinityforclay,andSWNTsbecomemoreexfoliatedand betternetworkedinthesecomposites.Thus,thepercola- tionthresholdofthesenanocompositeswasreducedfrom 0.05wt%to0.01wt%.Additionally,Yuetal.[131]demon- stratedasignificantreductionintheelectricalresistivity upontheadditionofalargeamountofCaCO3inCPCsthat containedMWNTsastheconductivefiller.Theconceptof theeffectiveconcentrationoftheMWNTswasproposed to quantitatively evaluate theeffect of CaCO3 on CPCs.

Theextranon-conductiveﬁllerwasshowntoincreasethe effectiveconcentrateoftheconductiveﬁllerinthepolymer

(13)

Please cite this article in press as: Deng H, et al. Progress on the morphological control of conductive network Fig.9.InCPCsconsistingofPP/PEblendsandMWNTs,top:schematicshowstheselectivedistributionofCNTinPEphaseinthepolymerblends,andits networkevolutionduringsolidstatedrawingandsubsequentannealing.Blow:SEMandelectricalmeasurementshowsthemorphologicalandelectrical propertyduringtheseprocesses.Thismaterialalsoshowsthepotentialtobeusedaselectroactiveself-healingmaterials.[110],Copyright2011.Reproduced withpermissionfromtheRoyalSocietyofChemistry.

matrix. Sucha phenomenon hasalsobeenobserved for systemsthatarebasedonarangeofpolymersandﬁllers.

4.6. Morphologicalcontrolthroughothermethods

Besidesthemethodsthatwerementionedabove,there areothermethodsthatcanbeusedtocontrolthemor- phologyofconductivenetworks,suchastheuseoflatex particles,anelectricormagneticﬁeldoracombinationof differentmethods.

Polymer latex particles can be used to enhance the excludedvolumeof aconductiveﬁllerbecausea segre- gatedconductivenetworkcanbeformedattheinterfaces between the polymer latex particles [82,129,140–144].

AsshowninFig.11,Miriyalaet al.[143]preparedCPCs withpoly(vinylacetate)(PVAc)latexparticlesandCB.It wasdemonstratedthattheuseoflatexparticlesreduced the electrical percolation threshold from 8.17vol.% to 1.21vol.% via the formation of segregated conductive networks.AsimilartechniquewasusedtoprepareCPCs

Fig.10. (Left)ThemechanismofsynergisticeffectbetweenMWNTandCBfortheformationofconductivenetworkinCPCsbasedonepoxymatrix,[137], Copyright2009.ReproducedwithpermissionfromSpringerScience+BusinessMedia;(Topright),thegrape-cluster-likeconductivenetworkconstructedby CBandorientedMWNTsinPPmatrix,[138],Copyright2012.ReproducedwithpermissionfromElsevierScienceLtd;(Bottomright)theelectricalpercolation behaviorofCPCsbasedonPPcontainingdifferentﬁllerorhybridﬁllers,showinglowerpercolationthresholdcanbeobtainedforCPCscontainingMWNT andCBintheratioof1:1,[132],Copyright2012.ReproducedwithpermissionfromBME-PT.

(14)

Please cite this article in press as: Deng H, et al. Progress on the morphological control of conductive network Fig.11.(Left)Theformationofsegregatednetworkandrandomnetworkduringemulsionbasedprocessandsolutionbasedprocess,(right)andtheresulted electricalproperties,showingthepercolationthresholdcanbereducedsigniﬁcantlythroughtheformationofsegregatedconductivenetworkinpoly(vinyl acetate)latexmatrix.[143],Copyright2008.Reproduced,withpermissionfromWiley-VCH.

with various types of ﬁllers, such as CNTs [140] and graphene[82].Daltonandco-workers[141]demonstrated that a signiﬁcantreduction in theelectrical percolation threshold couldbe obtained by lockingCNTs in a pre- dominantlyhexagonalclose-packed(HCP)colloidalcrystal latticeoflatexparticles.Incontrasttotraditionallatextech- nology,in which CNTs arerandomly distributed within thematrix,theauthorsdemonstratedthattheexclusion ofCNTsfromtheinteriorvolumeofthelatexparticlespro- motedtheformationofanon-randomsegregatednetwork.

Anelectricfield canbeusedtocontrolthemorphol- ogyofaconductivenetwork.Anumberofresearchgroups havedemonstratedthatconductivefillerscanbeoriented underanelectricfield[117,145,146]andthatthedynamic percolation process can be accelerated withan electric field[120,126,145].Theformationofaconductivenetwork underan electricfield can directlyresult ina very low electricpercolationthreshold.Similartoanelectricfield, amagneticfieldcanalsobeusedtoorientaconductive fillerandtotriggertheformationofconductivenetworks [147–150].

Inadditiontothemethodsthatwerementionedabove, it is sometimes necessary to usea combination of different methods to trigger the formation of conductive networks.Guoand co-workers[138]demonstratedthat conductive networks with a special conﬁguration of a grape-cluster-like structure, in which oriented MWNTs servedasbranchesandgrape-likeCBlinkedtheseconduc- tivetubes,couldbeachievedviathecombinedmethodof amixedﬁllerandashearforce.Thisshearwasrealised throughmultistagestretchingextrusionusinganassem- blyoflaminating-multiplyingelementsinapolymermelt.

Inthisstudy,alowerpercolationthresholdorahighercon- ductivitycouldbeachievedduetotheformationofthese special conductive networks.In a seriesof studies that wereperformedbyLiandco-workers,atechniquetermed insitumicroﬁbrillisationwasusedtofabricatemultifunc- tionalCPCs[87,151–153].Thistechniqueinvolvedtheuse

ofpolymer blendsand subsequentshear;therefore,one ofthepolymerphasescouldbedeformedintomicrofib- rils.The conductivefillercouldbeselectively locatedin themicrofibrilsorattheinterface.Lowelectricalpercola- tionandmultifunctionalCPCscouldbefabricatedbysuch a technique.Inourrecentwork[110],acombination of polymerblends,shearandsubsequentthermalannealing wasusedforCPCsthatcontainedMWNTsastheconductive filler.Itwasdemonstratedthatboththepolymerphasenet- workandtheconductivenetworkplayedimportantroles in thefinal electricalproperties of theCPCs.In another recentworkofours,acombinationofalayeredstructure,a mixedfiller,solidstatedrawingandthermalannealingwas usedtocontrolthemorphologyoftheconductivenetwork [154].Insuchasystem,thedynamicpercolationinhighly orientedconductivenetworksthatformedfromdifferent carbonnanofillerswasstudiedduringdisorientationupon annealing.Itwasconcludedthatthedynamicpercolation processinhighlyorientedconductivepolymercomposites filledwithMWNTscouldbeacceleratedbytheaddition ofCBbecauselessentanglednetworksareformedinthe hybridfillersystemthaninthesystemwithMWNTsalone.

Insummary,themorphologicalcontrolofconductive networkshasbeenshowntobeextremelyimportantfor the preparation ofCPCs. First, different characterisation methods canbeutilisedtoobtaininformation onthese conductivenetworksfromdifferentperspectives.Exten- siveliteraturestudieshaveshownthatdifferentmethods, includingtheuseofpolymer blends,thermal annealing, a shear force,a mixed ﬁller, latex particles, an electric ﬁeld,oracombinationofdifferentmethods,canbeused toeffectivelycontroltheconductivenetworksand,thus, theelectricalpropertiesofCPCs.

5. ThemultifunctionalitiesofCPCs

CPCscanbeusedforapplications,suchasEMIshiel- ding, antistatic and electrostatic painting, due to their

(15)

Please cite this article in press as: Deng H, et al. Progress on the morphological control of conductive network conductivity. Moreover, these composites can also be

used for smart applications, including strain [21,22, 24,26,155–158] and damage sensing [23,63,159–165], vapour/liquidsensing [22,27–30,166–168]and temperaturesensing[32,169,170].TheoverallconductivityofCPCs dependsheavilyonthetunnellingbetweenlocalconduc- tivenetworks;therefore,achangeinthelocaltunnelling distance could lead toa significant changein theconductivity.Outer-stimuli-inducedchangesinthetunnelling distancecouldresultinasignificantchangeintheconduc- tivity.Therefore,thesechangescanbemonitoredandused forsensingpurposes.Furthermore,theelectricalconduc- tivityofCPCscanalsoresultsinotherfunctionalities,such asaflexibleconductor,electroactiveshapememoryand thermoelectricproperties.

5.1. Strainanddamagesensing

Awiderange ofapplicationshavebeenproposedfor strain sensing,includingsmart textiles,health monitor- ing,wearableelectronicsandmovementsensors[124,171].

For human motion, wearable, stretchable and sensitive devicesarerequiredtodetectresistivitychangesduring movement. Severalconductiveﬁllers, suchas CB,CNTs, graphene and metal particles, can beincorporated into aninsulating polymermatrix tofabricate strainsensors [21,158,172–175].

The sensitivity of these strain sensors is thought to be one of the most important issues because different applicationsoftenrequiredifferentstrainsensitivities.The processingconditionsandmaterialproperties,suchasthe weightfraction,diameterandconductivityoftheconduc- tiveﬁllers,theprocessingtemperatureandthetunnelling barrier height of thepolymer matrix, all play a role in determining the sensitivity [176]. To modify the strain sensitivity, severalmethods have been reported.Muru- garajetal.observedthatanarrowdistributionoftunnel gaps between overlapping nanochannels was beneﬁcial for a highsensitivity in theirCPCs [177]. Hwang et al.

reported that the sensitivity strongly depended on the concentrationofpoly(3-hexylthiophene-2,5-diyl)(P3HT), which were wrapped around CNT bundles [24]. Dang etal.observedthataCNTwithahigheraspectratiowas preferredfor a highersensitivity inCNT/siliconerubber nanocomposites[22].Recently, ourgroupdemonstrated thatmixedfillerscontainingCNTsandmetallicparticles in different volume ratioscouldbe usedto modifythe strainsensitivityofCPCsfibres[158].Itwasalsodemon- stratedthatthestrainsensitivitywasindependentofthe fillerconcentrationforthesamefillercomposition.This resultagreeswellwiththereportedstudybyZhangetal., inwhichauniversallogarithmicresistivity-strainrelation independent of theCNT concentrationin thermoplastic polyurethane(TPU)wasreported[26],asshowninFig.12.

Theauthorsdemonstratedthatastrain-inducedresistivity changecouldbeintroducedintotheﬂuctuation-induced tunnelling(FIT)modelviaanincreaseinthetunnellinggap widthduetothemacroscopicseparation ofCNTsunder strain.

ReproducedwithpermissionfromAmericanPhysicalSociety.

TheFITmodel[178]isexpressedatacertaintempera- tureasfollows[26],assuminganimageforcecorrectionto arectangulartunnelbarrier:

d(lnp)

ds =c+f(T0,T1) (5)

cisanadjustableconstant,sisthestrain,T₁isamea- sureoftheenergythat isrequiredtomoveanelectron acrosstheinsulatinggapandT0isthetemperaturebelow whichtheresistivityreducestotemperature-independent tunnelling.Thevaluesf(T₀,T₁)exhibitinsigniﬁcantvari- ationscomparedwithc,thus,auniversalresistivity-strain behaviourisobservedatarelativehighstrainforeachcom- positesystemthatcontainsthesametypeofﬁllers.

Iftheconductionissimpletunnelling,itiseasytoverify that∼exp(Bω)andd(ln)/ds=Bω,whereBisanumer- icalconstant.Theproductωonlydependsontheheight, shapeandwidthofthetunnellingbarrier[42].Asreported intheliterature[179–181],metalsandalloys(suchasthe nickel,tinandbismutheutecticalloythatwasusedinthis study)generallyhaveahighertunnellingbarrierthanCNTs becausethechargingenergyplaysamoreimportantrolein thetunnellingthatinvolvesametalparticle[182].Hence, asshowninFig.13,ahighervolumefractionofmetalinthe conductiveﬁllerscontributestoahigherresistance-strain sensitivity.

Amodel thatwasderivedfromtunnelling theoryby Simmons[183][184]canbeusedtofurtherunderstand themechanismofstrainsensing.TheoverallresistanceR ofthecompositescanbecalculatedusingEq.(6)andEq.

(7).

R=

_L

N

₈_nhs

2a²e²

exp (s) (6)

= 4

2mϕ

h (7)

Listhenumberofparticlesthatformasingleconduc- tivepath,Nisthenumberofconductingpaths,hisPlank’s