Composites: Part A 102 (2017) 88–95

Contents lists available at ScienceDirect

Composites: Part A

journal homepage: www.elsevier .com/locate /composi tesa

Spatial Confining Forced Network Assembly for preparation ofhigh-performance conductive polymeric composites

http://dx.doi.org/10.1016/j.compositesa.2017.07.0271359-835X/� 2017 Elsevier Ltd. All rights reserved.

⇑ Corresponding authors at: State Key Laboratory of Organic-Inorganic Compos-ites, Beijing University of Chemical Technology, Beijing 100029, China (D. Wu).

E-mail addresses: wudaming@vip.163.com (D. Wu), hy06@163.com (Y. Huang),zguo10@utk.edu (Z. Guo).

Daming Wu a,b,⇑, Xiaolong Gao b, Jingyao Sun b, Dan Wub, Ying Liu a,b, S. Kormakov b, Xiuting Zheng c,Lili Wu d, Yao Huang b,⇑, Zhanhu Guo d,⇑a State Key Laboratory of Organic-Inorganic Composites, Beijing University of Chemical Technology, Beijing 100029, ChinabCollege of Mechanical and Electrical Engineering, Beijing University of Chemical Technology, Beijing 100029, ChinacPolymer Material Processing Equipment Engineering Research Center of the Ministry of Education, Beijing 100029, Chinad Integrated Composites Laboratory (ICL), Department of Chemical & Biomolecular Engineering, University of Tennessee, Knoxville, TN 37996, USA

a r t i c l e i n f o

Article history:Received 11 May 2017Received in revised form 8 July 2017Accepted 27 July 2017Available online 29 July 2017

Keywords:Self-assemblyForced-assemblyNetworkConductive compositeSpatial Confining Forced Network Assembly

a b s t r a c t

Constructing a network of conductive fillers in polymeric matrix is essential for the preparation of con-ductive polymer composites. Although the conductivity of the composites could increase remarkablyafter the percolation threshold, it is still much lower than expected due to a limited self-assembly inter-action between filler particles. In this paper, high-performance conductive polymer composites were pre-pared by the method of Spatial Confining Forced Network Assembly (SCFNA). The compound ofhomogenous polymer and conductive fillers, prepared by conical twin-screw mixer, was placed in a com-pression mold with confining space to carry out two-stage compression, free compression and spatialconfining compression. The electrical conductivity of the SCFNA prepared polypropylene/short carbonfibers was increased up to 4 orders of magnitude higher than that of by ordinary compoundingtechnology.

� 2017 Elsevier Ltd. All rights reserved.

1. Introduction

Conductive polymer composites (CPCs) are considered as one ofthe most important functional materials which have found wideapplications for decades in manufacturing antistatic, electric andthermal conductive products to meet the needs of electronicdevices, aircraft accessories, personal computers, light emittingdiode (LED) chips, electromagnetic interference (EMI) shieldingand sensing materials, medical devices, smart biomaterial for tis-sue engineering, auto parts, household appliances, mine pipes,etc. [1–4]. Because most polymers are electrically insulating, con-ductive fillers are incorporated to fabricate CPCs. More efforts focuson building conductive networks in a polymer matrix using con-ductive fillers, including carbon black (CB), carbon fibers (CFs), gra-phite flakes, carbon nanotubes (CNTs), graphene, etc. [2,5]. Meltcompounding, in situ polymerization, and solution mixing aremain methods to prepare CPCs, of which the melting compoundingis the most popular and efficient solution to get homogeneouslydispersed polymer composites.

Co-rotation twin screw extruders, planetary extruders, and Bussextruders are the most available efficient melt compoundingequipment, which could provide a powerful shear effect to bringabout the improved dispersion of fillers in the polymer matrix[5,6]. Melt compounding method is the most favorite and firstchoice in the preparation of CPCs incorporated with micro scale fil-lers of graphite, carbon black, carbon fibers, etc. [7–9], moreover ithas demonstrated feasibility in preparing CPCs incorporated abovementioned micro scale fillers with nanoscale fillers of graphene,CNTs, carbon nano fibers, carbon black, etc. [10–12]. In situpolymerization can promote the fillers to distribute evenly inpolymeric monomers with very low viscosity [5]. This processgives excellent filler dispersion and a potentially good interfacialstrength between the filler and the polymer matrix. In situpolymerization has been reported for the preparation of CPCssuch as polypyrrole/graphene oxide (GO), polyacry-lonitrile/GO,biodegradable poly (butylene succinate)/GO and N-hydroxy-methyl acrylamide/CNTs [13–16]. The in-situ polymerization pro-cess cannot be widely applied as melt compounding process dueto very little choice of available polymerizable monomers, howeverit is an essential method for the preparation of thermoset andrubber-based CPCs [5]. As a good example, epoxy has been exten-sively investigated as a polymer matrix for the preparation of a

D. Wu et al. / Composites: Part A 102 (2017) 88–95 89

large range of CPCs containing CNTs [17–20]. Solution mixingmakes it much easier to achieve locally homogeneous dispersionof entangled fillers without breaking down [5]. Film casting isthought as a sensible process to prepare CPCs after solution mixing.Compared with the several minutes solidification process of meltcompounding process, the solidification process of solution mixingoften takes more than 24 h. Such long period give adequate timefor the aggregation of conductive network and reach the electricalpercolation threshold with lower filler loading than that for meltcompounding process [5,21].

Generally, there are mainly three approaches to obtain a contin-ual conductive network by self assembly of conductive fillers inpolymeric matrix after melt compounding, solution mixing, orco-continuous immiscible polymers phase with selected distribu-tion of conductive fillers in one phase, as well as segregated con-ductive structure, as shown in Fig. 1. A widely adoptedprocessing technology is to increase the fraction of the conductivefillers to a certain level until reaching the percolation threshold[2,5,6,30–32]. Another method is to adopt a system of two immis-cible polymers and conductive fillers. With suitable fractions oftwo polymers and at certain thermodynamic conditions of temper-ature and enthalpy, a co-continuous two-phase structure withselected location of fillers in one polymer could be obtained, theone containing fillers serves as continual conductive network[33,34]. The third method is the formation of a segregated struc-ture in the composites by compressing a mixture of polymer gran-ules decorated with conductive fillers via dry or solution mixing toconstruct the segregated conductive networks [6,28,29,35,36]. Forexample, Khastgir [22] reported that the conductivities of compos-ites Ethylene Vinyl Acetate/Conductive Carbon Black (EVA-CCB),Ethylene Vinyl Acetate/Short Carbon Fiber (EVA-SCF) (by melt mix-ing), and Ethylene Vinyl Acetate/Short Carbon Fiber/Multi WallCarbon Nano Tubes (EVA-MWCNTs) (by solution mixing) at a per-colation loading of conductive fillers with mass fraction of 30, 15and 5 wt%, respectively, increased from 10�8 S/m of neat EVA to

Fig. 1. Approach to construct the conductive networks in polymeric composites.(For interpretation of the references to colour in this figure legend, the reader isreferred to the web version of this article.)

2.5 � 10�4 S/m, 5.0 � 10�3 S/m and 2.5 � 10�2 S/m, respectively.Although a conductive network had been formed, the electric con-ductivities of above composites still had 5–7 orders of magnitudelower than that of CCB, SCF and MWCNT fillers. This can beexplained that the possible aggregation force of the fillers, includ-ing van der Waals force, liquid bridge, electrostatic force, hydrogenbonding, is not large enough to repel all polymer away in between,the network by such a self-assembly mechanism could not be verycompacted.

A transition from insulator to conductor for polymer compos-ites was firstly reported by Gurlan and remained up to now asone of the most important topics in preparing high-performanceCPCs [5,21]. Many models for predicting electric conductivity andpercolation threshold were proposed, including Scaling law basedon classical percolation theory [5], Voet Model of non-Ohmicassumption [22], Scarisbrick model of Ohmic assumption [23],McCullough model based on transport phenomena [24], Buechemodel for gel formation process in polymerization [25], modifiedScarisbrick Model considering the filler aspect ratio, surface areato volume ratio, and conductivity of the polymer [24]. Theoreticalpredictions and experiments have confirmed that the formation ofa continual conductive network is the key to obtain high-performance CPCs [5–7,10,21,22]. The average gap between theconducing fillers in the network plays a crucial role in all of param-eters and conditions affecting the level of conductivity of the com-posites. The transportation mechanism of electrons betweenadjacent conductive particles in the polymeric matrix is explainedas an Ohmic conduction with an average gap between adjacentparticles less than or equal to 10 Å on the basis of Franckel’s theory[5], and non-Ohmic transmission of tunneling effect or electronemission from one particle to the next with an average gapbeyond 10 Å [23]. Adequate theoretical and experiment investiga-tions had qualitatively and quantitatively proven the conclusiveeffect of the gap between adjacent particles on the conductivityof the composites [5–7,10,22–26]. But,up to now, the morphologyof conductive network of either microfillers or nanofillers with anaverage gap of less than or equal to 10 Å between adjacent parti-cles prepared by existing compounding methods was rarelyreported. Rather than theoretical prediction, the possibility of con-trolling the average gap of particles in the self-assembly network-ing as thin as designed or as required seems even less prospectsubject to the filler loading limit and processing windows[5,33,34]. As a new approach to improve the electrical conductivityof the polymeric composites, the three dimensionally intercon-nected carbon aerogel-epoxy composites of aerographite-epoxyand aerographene-epoxy with extremely low percolation thresh-old to 0.007 vol% and orders of magnitude higher electrical conduc-tivity compared to CNT-based polymer nanocomposites atcomparable filler content were prepared with prefabricated net-work of graphite or graphene by chemical vapor deposition(CVD) process and freeze casting technology [37–39]. Porous car-bon nanotubes/chitosan (CNTs/CHI) nanocomposites prepared byice templating technology demonstrated dramatically improvedelectrical conductivity up to 8 orders of magnitude higher thanpure chitosan scaffolds due to unidirectionally oriented lamellarstructure with uniform layer thickness and channels in submicronand micron scale respectively [40].

In order to construct a more compacted network in polymercomposites and markedly enhance electric conductivity of CPCs,a Spatial Confining Forced Network Assembly (SCFNA) methodwas proposed and used to prepare CPCs of thermosetting poly-dimethylsiloxane (PDMS) and short carbon fibers (SCFs). The elec-trical conductivity of PDMS/SCFs composite by SCFNAmethod withpercolation threshold of 1.5 wt% reached to 103.2 S/m, almost 3orders of magnitude higher than by ordinary compoundingmethod [27]. In this paper, the mechanism and procedure of SCFNA

Fig. 2. Scheme of technological pathway of SCFNA and conventional compounding approach. (For interpretation of the references to colour in this figure legend, the reader isreferred to the web version of this article.)

Fig. 3. Model of SCFNA: free compression and spatial confining compression. (For interpretation of the references to colour in this figure legend, the reader is referred to theweb version of this article.)

90 D. Wu et al. / Composites: Part A 102 (2017) 88–95

was clarified in details and a verifying test on SCFNA was carriedout in preparing CPCs of thermoplastic PP and SCFs. The main ideaof SCFNA was introducing a mechanical delivered additional inter-action to carry out a forced assembly of network, in this way, amore compacted network can be attained. It should be pointedout that the object of forced assembly was the free assembly net-work, not homogeneously distributed filler particles. As shown inFig. 2, the SCFNA process involves three steps. Firstly, a homoge-neous dispersion mixture of conductive fillers and polymer wereobtained by melt compounding or solution mixing. Secondly, thehomogeneous dispersed mixture was compressed by a pressingmold to initiate and finish the networking of the conductive fillersby self-assembly mechanism. Thirdly, the sample was further com-pressed after forming a self-assembled network until to a thicknessless than average diameter of the conductive network threads, dm.In this way, the network was compulsively compressed to be morecompacted in a confining space. The thickness of the sample is crit-ical in SCFNA process. As shown in Fig. 3, when the thickness of thesample was much larger than dm, the compression could only causethe self-assembled network wiggling in polymer matrix withoutdensification of the network threads. When the sample was com-pressed to a thickness of less than dm, a noteworthy densificationof the network threads could be formed by forced compressionof the mold to the network. Due to the restriction of wiggling free-dom of the network and the prevention of aggregate force betweengranules from falling apart, the filler at the threads will tend to getcloser and extra polymer between the filler granules will besqueezed out in special confining compression process. The mech-anism of forced assembly can be explained as follows. In spacialconfining compressing process, characterized by a thickness ofsample below dm, the compression force could be transmitted tothe granules to break the balance of self-assembly network. Thegranules under compression tend to get much closer until a new

balance was established. In this way, the self-assembly networkwas converted to a forced-assembly network with enhanced elec-trical conductivity.

2. Experimental

2.1. Materials

PP of YUPLENE BX3900 with a density of 0.905 g/cm3, meltingpoint of 167 �C and an average molecular weight of 1.15x105 g/molwas bought from SK global chemical, Korea. The short carbon fibersof 7 lm in diameter, 4 mm in length and 2.6x104 S/m in electricalconductivity were provided by Toray Group, Japan.

2.2. Experimental setup

A conical twin-screw mixer of HAAKE MiniLab was used to pre-pare PP/SCF compound. The experimental hot embossing devicewas designed and the setup was shown in Fig. 4, which could pro-vide a pressing force up to 50 kN with a pressing speed of 0.005–0.5 mm/s in working stroke. The servo motor of this embossingdevice was accurately controlled by a PLC system and the positionaccuracy of ±3 lm for the up platen was guaranteed. The com-pressing mold was made of two flat plates with electrical heaters.The temperature of the molds can be controlled from ambient tem-perature to 220 �C with a control accuracy of ±1 �C. The effectiveworking area was 160 � 160 mm.

2.3. Preparation of composite samples

The PP and SCFs of 5, 7, 10, 15, 25 and 35 wt% in mass fractionwere manually stirred for 2 min and then the mixtures of each SCF

Fig. 4. Sketch of the experimental hot embossing device. (For interpretation of thereferences to colour in this figure legend, the reader is referred to the web version ofthis article.)

D. Wu et al. / Composites: Part A 102 (2017) 88–95 91

mass fraction were loaded into the HAAKE MiniLab conical twin-screw mixer with a rotational speed of 70 rpm for 15 min at175 �C. A calculated weight of molten PP/SCF compound preparedby HAAKE MiniLab conical twin-screw mixer was put on the downmold with a temperature of 190 �C and the two stage compressionstarted. In the first stage of compression, free compression, the upmold with a temperature of 190 �C went down to compress thelump of PP/SCF compound from original thickness of about10 mm to 2.0 mm within 120 s and kept 5 min to initiate and

Fig. 5. Morphology evolution of PP/10 wt% SCF composites in process compression: (a) faafter the second stage of compression with a thickness of 0.4 mm; (e) after second stathickness of 0.2 mm; (g) section of the sample with a thickness of 0.2 mm after spatialcamera). (For interpretation of the references to colour in this figure legend, the reader

complete the self assembly of network before the second stage.In the second stage of compression, forced compression, the sam-ples were compressed from a thickness of 2.0 mm to a final thick-ness of 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 1.0 and 1.5 mm,respectively, for the sample of 10 wt% SCF, and 0.1, 0.2, 0.3 and0.4 mm, respectively, for other samples within 60 s and then thecompressed samples were taken out of the compression mold afterbeing cooled to 150 �C. In order to make a comparison of the effectof self assembly network and forced assembly network on electri-cal conductivities of the composites, the samples with a thicknessof 2.0 mm after the first stage of compression were taken, as theCPCs by conventional compounding method.

2.4. Characterization

The JTVMS-1510T 3D Image Measurement System, product ofChina Gongguan Janten Instrument Co. was used for characterizingthe morphology of the PP/SCF composites. A digital camera ofCanon SX710 HS of Canon INC. of Japan was used to observe themacrostructures of the samples. The direct current (DC) electricalconductivities of the composites over 10�6 S/m were measuredby Keithley 4200-SCS Parameter Analyzer (America) with a stan-dard four probe method, while electrical conductivities below10�6 S/m were measured by a ZC-90D resistivity meter fromShanghai Taiou Electronics (China). All of the conductivity mea-surements were carried out at an ambient condition.

3. Results and discussion

Fig. 5 showed the morphology evolution process in the prepara-tion of CPCs of PP/10 wt% SCF. Fig. 5(a) showed a uniform disper-sion state after mixing in a conical twin-screw mixer of HAAKE

st cooled extrudate; (b) before compression; (c) after first stage of compression; (d)ge of compression with a thickness of 0.2 mm; (f) overview of the sample with aconfining compression (a–e and g: by 3D Image Measurement System; f: by digitalis referred to the web version of this article.)

92 D. Wu et al. / Composites: Part A 102 (2017) 88–95

MiniLab, which was essential to obtain a continuous and uniformconductive network. Since it took about 5 min to add the com-pound extrudate to the mold, the self assembly of networkinghad already started for CPCs of PP/10 wt% SCF, referring to Fig. 5(b). After the first stage of compression, i.e. free compression, seeFig. 5(c), the self assembly network was already built up, but thethreads of the network were relatively incompact. Fig. 5(d,e)showed the forced assembly network of the samples after the sec-

Fig. 6. Morphology evolution of PP/5 wt% SCF composites in process compression. (a) Fasthickness of 0.4 mm; (d) after second stage of compression with a thickness of 0.2 mm; (sample with a thickness of 0.1 mm.

Table 1Electrical conductivities Ks and Kc of the PP/SCF composite by SCFNA and compounding m

Sample

Mass fraction of the SCF (%) Thickness (mm)

5 0.100.200.300.40

7 0.100.200.300.40

10 0.100.200.300.400.50.60.70.81.01.52.0

15 0.100.200.300.40

25 0.100.200.300.40

35 0.100.200.300.40

ond stage of compression, i.e. spatial confining compression. Obvi-ously, a compact network was found in the sample of PP/10 wt%SCF and the samples with a thickness of 0.2 mm and 0.1 mmshowed much compact threads of the network than that of withthickness of 0.4 mm. Fig. 5(f) showed the overall images of theforced assembly network by SCFNA. Fig. 6 showed the morphologyevolution of the CPCs of PP/5 wt% SCF in process of two stage com-pression. It seems very similar to the case of Fig. 5. But there were

t cooled extrudate; (b) before compression; (c) after first stage of compression with ae) after second stage of compression with a thickness of 0.1 mm; (f) overview of the

ethods.

Electrical conductivities Ks/Kc

Kc (S/m) Ks (S/m)

1.12E�12 1.76E�08 1.57E+49.52E�11 8.17E+11.65E�11 1.47E+13.84E�12 3.43

2.06E�04 6.95E+01 3.37E+54.10E+01 1.99E+52.03E+01 9.85E+48.52 4.14E+4

4.54E�04 6.73E+01 1.48E+54.74E+01 1.04E+58.45 1.86E+41.98 4.36E+31.51E+00 3.33E+31.02E�03 2.255.55E�04 1.225.51E�04 1.215.02E�04 1.114.82E�04 1.064.54E�04 1.00

2.18E�03 1.74E+02 7.98E+41.05E+02 4.82E+46.68E+01 3.06E+44.65E+01 2.13E+4

1.29E�01 3.08E+02 2.39E+31.33E+02 1.03E+38.59E+01 6.66E+23.77E+01 2.92E+2

3.04E�01 5.73E+02 1.88E+34.86E+02 1.60E+32.32E+02 7.63E+21.37E+02 4.51E+2

Fig. 7. Comparison of electrical conductivities of PP/SCF composites by SCFNA andcompounding method. (For interpretation of the references to colour in this figurelegend, the reader is referred to the web version of this article.)

D. Wu et al. / Composites: Part A 102 (2017) 88–95 93

crucial difference in two aspects. Firstly, the relative integratednetwork was formed only when the sample was compressed to athickness of 0.1 mm. Secondly, by comparison of Figs. 6(e) to 5(c), we can find that the density of the threads was just comparableto that of self assembly network in the sample of PP/10 wt% SCF.The morphology pictures of the samples of PP/5 wt% SCF had a cor-respondence with the conductivities, ref. to Table 1, that the sam-ple kept very low level of conductivities until to be compressed to0.2 mm and a sharply increase 3 orders of magnitude in conductiv-ities appeared when the sample was compressed to 0.1 mm. Thisphenomenon indicated that SCFNA technology could decrease thepercolation threshold, as in the case of CPCs PP/5 wt% SCF, whichwas impossible to form a network by ordinary compoundingmethod. The driving force of for the formation of more and morecompacted network of SCF in process of forced assembly is mainlyderived from the mechanical pressing force which could take effectonly after the sample to be compressed to a thickness less thanmean diameter of free assembly network threads. As an evidenceof the argument of driving force in process of forced assembly,Fig. 9 showed a variation of conductivity with thickness of thesample. It can be seen from Fig. 9 that the conductivity kept almostconstant with the sample being compressed from a thickness of 2to 0.7 mm and sharply increased after the sample being com-pressed to about 0.6 mm. Because the free assembly network hada wiggling freedom when the thickness was greater than a dm ofabout 0.5 mm, see Fig. 5(c), the compressing force could mainlycause the network wiggling in polymer matrix. After being com-pressed to be less than dm, a forced compression of the networkcan be effectively carried out because the network already lostits wiggling freedom. In process forced compression, the aggrega-tion force between fillers played also very important role, whichguaranteed a needed effect of being more closer of the fillers andbeing squeezed out of the filled polymer in network.

A significant difference of the aggregation state of SCF beforeand after SCFNA process can be observed in Figs. 5 and 6. FromFigs. 5(c) and 6(c), it can be seen that the threads of network werenot compact enough and many SCFs were located in polymerphase, not on conductive network. From Fig. 5(c), a roughly deter-mination of the mean threads diameter of 0.51 mm for free assem-bly network could be made. This implied, according to the SCFNAmodel, that the forced assembly networking for that sample shouldstarted from the moment when it was compressed to about0.5 mm. From the section view of the forced compressed samplewith a thickness of 0.2 mm in Fig. 5(g), it can be seen that thethreads of the network were compactly compressed to a ellipse-like shape with a mean width of 0.631 mm and most SCFs werelocated on the network, from which we can expect a significantlyimproved electrical conductivity of the composites.

Table 1 summarizes the electrical conductivities of the compos-ites with 5, 7, 10, 15, 25, and 35 wt% SCF produced by SCFNA (Ks)and ordinary compounding (Kc) methods. In order to make a com-parison of the difference of SCFNA method and ordinary com-pounding method, a ratio of electrical conductivities of twomethods Ks/Kc was also listed. The thickness of the samples bycompounding was 2 mm. The electrical conductivity of the neatPP of YUPLENE BX3900 was 3.32x10�12 S/m. The comparison ofelectrical conductivity of the PP/SCF composite by SCFNA withordinary compounding method was shown in Fig. 7.

It can be seen from Fig. 7 that the percolation threshold was fal-len at the same area of SCF contents from 5 to 10 wt% for bothSCFNA method and compounding method. This is because thatthe forced assembly was carried out only after formation of freeassembly network. The conversion from free compression to forcedcompression in SCFNA process for PP/10 wt% SCF compositeoccurred when the sample was compressed to a thickness of0.6 mm, as shown in Fig. 9. Further compression after the conver-

sion thickness, the electrical conductivity of the compositeincreased so sharply, that the second percolation threshold bySCFNA appeared. From a general view of the experimental data,the electrical conductivities of the PP/SCF composites preparedby SCFNA technology could increase up to 4 orders of magnitudehigher than that of composites prepared by ordinary compoundingtechnology. The tremendous increase in the electrical conductivityof the PP/SCF composites by SCFNA technology can only beascribed to the forced assembly network, because the formulationsand all processing conditions were kept the same in carrying outthe contrast experiments, except the addition of forced compress-ing for the SCFNA method. From Fig. 7, it can be also seen that in awide range of SCF content after percolation threshold, all thecurves were basically parallel at a high level of Ks/Kc ratio, whichdemonstrated a structure stability of the forced assembly networkby SCFNA technology. As an application of SCFNA, the PDMS/SCFcomposites were tested [27] and the ratio of Ks/Kc was in the rangeof 11.6–70.1 with a SCF content changing from 2 wt% to 4 wt%.

The main difference of the PDMS/SCF processing condition ofthe work [27] from present PP/SCF was the viscosity of matrixpolymer and the effect of bolting cloth. The reason of a lower ratioof Ks/Kc for PDMS/SCF than that of PP/SCF could be explained thatthe PDMS before solidification had very low viscosity like water,so the conductive fillers could be easier to construct a relativelycompact network by free assembly mechanism. Another reasonwas the increase of filler density in the main region of the PDMS/SCF composites by volume exclusion effect of the bolting cloth inSCFNA process. Much bigger increase in the electrical conductivityof PP/SCF than PDMS/SCF in the process of SCFNA can be explainedthat the free assembly network of the PP/SCF was much looser thanthat of PDMS/SCF due to much higher viscosity of the PP matrix.Therefore, the SCFNA showed more effects on the thermoplasticpolymer composites.

Fig. 8 showed how the electrical conductivity increased withthe decrease of the sample thickness. With the decrease of samplethickness from 0.4 to 0.1 mm, the electrical conductivity increasedby 3.74 times for the PP/15 wt% SCF composites by SCFNA. It fol-lowed from above results that the compression ratio played a veryimportant role in constructing more compact forced assemblynetwork. From Fig. 5(c), the Characteristic diameter dm of thePP/10 wt% SCF was observed to be about 0.5 mm. According tothe proposed Model of SCFNA, refer to Fig. 3, the free assembly net-work of PP/10 wt% SCF composites in our experiment lost their

Fig. 8. The effect of the sample thickness on the electrical conductivity of PP/SCFcomposites by SCFNA technology. (For interpretation of the references to colour inthis figure legend, the reader is referred to the web version of this article.)

Fig. 9. Two stage compression of the PP/SCF composites (SCF content: 10 wt%). (Forinterpretation of the references to colour in this figure legend, the reader is referredto the web version of this article.)

94 D. Wu et al. / Composites: Part A 102 (2017) 88–95

freedom of wiggling in the polymer matrix after being compressedto 0.5 mm, from that moment further compression would causeforced assembly of the network. In fact, the contribution to theforced assembly of the network in process of SCFNA included alsothe consolidation of multilayer network and the compression ofthe incorporative network.

The 1D fillers, like CFs and CNT, are more easy to construct anetwork in polymeric matrix and available to prepare high electri-cal conductive polymer composites by SCFNA technology. The 2Dfillers, like flake graphite, is also available for preparation of poly-mer composites with enhanced conductivity by SCFNA technology.For 0D fillers, like nano CB, the SCFNA may difficult develop itscapability because a non-rational thickness of several decadesnanometers needed for forced assembly will bring special difficultyin fabrication. But the 0D fillers could apply together with 1D fillersto bring further enhancements in conductivities by filling the gapsbetween 1D fillers. In this work, the high electrical conductive PP/SCF composite films or thin sheets were prepared by SCFNAmethod. This does not mean that the SCFNA method is onlyavailable for flat and thin product. More thicker and intricatelyshaped products could be fabricated by SCFNA technology withspecialized molds and mechanism. The work on fabrication of high

electrical conductive PDMS/SCF composites [27] demonstratedthat SCFNA method is also available for fabricating compositeswith thermosetting polymer matrix.

4. Conclusions

In this paper, high electrical conductive polymer compositeswere prepared by a method of Spatial Confining Forced NetworkAssembly (SCFNA). The mechanism of the SCFNA method was toapply a compression stress onto a self-assembly network whenthe compound was pressed in a confined space, in this way, theself-assembly network lost its wiggling freedom and had to beforce compressed, leading to a more compacted network bysqueezing out the polymer between aggregated fillers. A morphol-ogy evolution model of SCFNA process was proposed and provedpracticable to prepare PP/SCF composites with high conductivity.A forced-assembly network may be formed when the compoundwas compressed to a thickness less than the average diameter ofthe threads of self-assembly network. A forced-assembly networkwith 5–35 wt% SCFs was obtained in the PP/SCF composites bythe SCFNA technology, which made it possible to increase the elec-trical conductivity up to 4 orders of magnitude higher than that ofcomposites prepared by the ordinary compounding technology.With the increase of the compression ratio, after second percola-tion threshold by SCFNA, the electrical conductivity of the samplescould increase several times due to the increase of forced assemblydegree. SCFNA technology could decrease the percolation thresh-old of CPCs composites, in this research the conductive networkwas built up with 5 wt% SCF in PP matrix, which was impossibleto form a network by ordinary compounding method.

Acknowledgements

This work was supported by the National Natural Science Foun-dation of China (Grant No. 51673020) and China PostdoctoralScience Foundation (Grant No. 2016M600032).

References

[1] Sharma M, Gao S, Mäder E, Sharma H, Wei LY, Bijwe J. Carbon fiber surfacesand composite interphases. Compos Sci Technol 2014;102:35–50.

[2] Ma Y, Wu D, Liu Y, Li X, Qiao H, Yu ZZ. Electrically conductive and super-toughpolypropylene/carbon nanotube nanocomposites prepared by meltcompounding. Composites: Part B 2014;56:384–91.

[3] Chung DDL. Comparison of submicron-diameter carbon filaments andconventional carbon fibers as fillers in composite materials. Carbon2001;39:1119–25.

[4] Díez-Pascual AM, Naffakh M, Marco C, Gómez-Fatou MA, Ellis GJ. Multiscalefiber-reinforced thermoplastic composites incorporating carbon nanotubes: areview. Curr Opin Solid State Mater Sci 2014;18:62–80.

[5] Deng H, Lin L, Ji MA, Zhang S, Yang M, Fu Q. Progress on the morphologicalcontrol of conductive network in conductive polymer composites and the useas electroactive multifunctional materials. Prog Polym Sci 2014;39:627–55.

[6] Tadmor Z, Gogos CG. Princirples of polymer processing. A John & Sons Inc.Publication; 2006.

[7] Sengupta R, Bhattacharya M, Bandyopadhyay S, Bhowmick AK. A review on themechanical and electrical properties of graphite and modified graphitereinforced polymer composites. Prog Polym Sci 2011;36:638–70.

[8] Yu GC, Wu LZ, Feng LJ, Yang W. Thermal and mechanical properties of carbonfiber polymer-matrix composites with a 3D thermal conductive pathway.Compos Struct 2016;149:213–9.

[9] Ma PC, Liu MY, Zhang H, Wang SQ, Wang R, Wang K, et al. Enhanced electricalconductivity of nanocomposites containing hybrid fillers of carbon nanotubesand carbon black(44). ACS Appl Mater Interfaces 2009;5:1090–6.

[10] Bauhofer W, Kovacs JZ. A review and analysis of electrical percolation incarbon nanotube polymer composites. Compos Sci Technol 2009;69:1486–98.

[11] Kuilla T, Bhadra S, Yao D, Kim NH, Bose S, Lee JH. Recent advances in graphenebased polymer composites. Prog Polym Sci 2010;35:1350–75.

[12] Al-Saleh MH, Sundararaj U. A review of vapor grown carbon nano fiber/polymer conductive composites. Carbon 2009;47:2–22.

[13] Wang JP, Xu YL, Zhu J, Ren PG. Electrochemical in situ polymeriza-tion ofreduced graphene oxide/polypyrrole composite with highpower density. JPower Sources 2012;208:138–43.

D. Wu et al. / Composites: Part A 102 (2017) 88–95 95

[14] Wu F, Lu Y, Shao G, Zeng F, Wu Q. Preparation of polyacry-lonitrile/grapheneoxide by in situ polymerization. Polym Int 2012;61:1394–9.

[15] Wang XW, Zhang CA, Wang PL, Zhao J, Zhang W, Ji JH. Enhanced performanceof biodegradable poly(butylene succinate)/graphene oxide nanocompositesviain situ polymerization. Langmuir 2012;28:7091–5.

[16] Nayak S, Bhattacharjee S, Singh BP. Preparation of transparent and conductingcarbon nanotube/N-hydroxymethyl acrylamide composite thin films by in situpolymerization. Carbon 2012;50:4269–76.

[17] El Sawi I, Olivier PA, Demont P, Bougherara H. Processing and electricalcharacterization of a unidirectional CFRP composite filled with double walledcarbon nanotubes. Compos Sci Technol 2012;73:19–26.

[18] Combessis A, Bayon L, Flandin L. Effect of filler auto-assembly on percolationtransition in carbon nanotube/polymer composites. Appl Phys Lett 2013;102.011907/1-4.

[19] Thostenson ET, Chou TW. Real-time in situ sensing of damage evolution inadvanced fiber composites using carbon nanotube networks. Nanotechnology2008;19. 215713/1-6.

[20] Thostenson ET, Chou TW. Carbon nanotube networks: sensing of distributedstrain and damage for life prediction and self healing. Adv Mater2006;18:2837–41.

[21] Gurland J. An estimate of contact and continuity of dispersions in opaquesamples. Trans Met Soc AIME 1966;212:642–6.

[22] Sohi NJS, Bhadra S, Khastgir D. The effect of different carbon fillers on theelectricalconductivity of ethylene vinyl acetate copolymer-based compositesand the applicability of different conductivity models. Carbon2011;49:1349–61.

[23] Voet A, Whitten Jr WN, Cook FR. Electron tunneling in carbon blacks. KolloideKeitschrift and Zeikhriji fur Polymerey 1965;21:39–46.

[24] Scarisbrick RM. Electrically conducting mixtures. J Phys D: Appl Phys1973;6:2098–110.

[25] McCullough RL. Generalized combining rules for predicting transportproperties of composite materials. Compos Sci Technol 1985;22:3–21.

[26] Bueche F. Electrical resistivity of conducting particles in an insulating matrix. JAppl Phys 1972;43:4837–8.

[27] Gao X, Huang Y, Liu Y, Kormakov S, Zheng X, Wu D, et al. Improved electricalconductivity of PDMS/SCF composite sheets with bolting cloth prepared by aspatial confining forced network assembly method. RSC Adv 2017;7:14761–8.

[28] Zhang C, Wang L, Wang J, Ma CA. Self-assembly and conductive net-workformation of vapor-grown carbon fiber in a poly(vinylidenefluoride) melt.Carbon 2008;46:2053–8.

[29] Zhang C, Zhu J, Ouyang M, Ma CA. Electric field controlled formation anddissociation of multiwalled carbon nanotube conductive pathways in apolymer melt. Appl Phys Lett 2009;94. 111915/1-3.

[30] Zhu J, Wei S, Ryu J, et al. In situ stabilized carbon nanofiber (CNF) reinforcedepoxy nanocomposites. J Mater Chem 2010;20(23):4937–48.

[31] Gu H, Guo J, Wei H, et al. Strengthened magnetoresistive epoxy nanocompositepapers derived from synergistic nanomagnetite-carbon nanofibernanohybrids. Adv Mater 2015;27(40):6277–82.

[32] Liu H, Gao J, Huang W, et al. Electrically conductive strain sensingpolyurethane nanocomposites with synergistic carbon nanotubes andgraphene bifillers. Nanoscale 2016;8(26):12977–89.

[33] Lan Y, Liu H, Cao X, et al. Electrically conductive thermoplastic polyurethane/polypropylene nanocomposites with selectively distributed graphene.Polymer 2016;97:11–9.

[34] Zhang X, Yan X, He Q, et al. Electrically conductive polypropylenenanocomposites with negative permittivity at low carbon nanotube loadinglevels. ACS Appl Mater Interfaces 2015;7(11):6125–38.

[35] Zhang K, Yu HO, Shi YD, et al. Morphological regulation improved electricalconductivity and electromagnetic interference shielding in poly (l-lactide)/poly (e-caprolactone)/carbon nanotube nanocomposites via constructingstereocomplex crystallites. J Mater Chem C 2017;5(11):2807–17.

[36] Gong T, Liu MQ, Liu H, et al. Selective distribution and migration of carbonnanotubes enhanced electrical and mechanical performances in polyolefinelastomers. Polymer 2017;110:1–11.

[37] Fukuda SGT, Mecklenburg M, et al. Electro-mechanical piezoresistiveproperties of three dimensionally interconnected carbon aerogel(Aerographite)-epoxy composites. Compos Sci Technol 2016;134:226–33.

[38] Wang Z, Shen X, Kim Jang-Kyo, et al. Ultralow electrical percolation ingraphene aerogel/epoxy composites. Chem Mater 2016;28:6731–41.

[39] Wang Z, Shen X, Garakani MA, et al. Graphene aerogel/epoxy composites withexceptional anisotropic structure and properties. ACS Appl Mater Interfaces2015;7:5538–49.

[40] Yan J, Wang H, Wu T, et al. Elastic and electrically conductive carbonnanotubes/chitosan composites with lamellar structure. Compos A2014;67:1–7.