properties and applications of filled conductive polymer composites

Polymer International 44 (1997) 117È124

Properties and Applications of FilledConductive Polymer Composites

Xiao-Su Yi,* Guozhang Wu and Yi Pan

Zhejiang University, Department of Polymer Science and Engineering, CN 310027 Hangzhou, PeopleÏs Republic of China

(Received 9 August 1996 ; revised version received 28 February 1997 ; accepted 24 March 1997)

Abstract : The electrical properties of polymers Ðlled with di†erent types of con-ducting particles are reviewed. Following a theoretical description of a generale†ective media (GEM) equation, the experimental conductivityÈvolume fractiondata for thermoplastic Ðlled with vanadium oxide particles as well as thermo-setting polymer composites, are Ðtted to the equation. The calculated property-related parameters in the equation are discussed. The electrical conductivity ofthe composites is combined with an extremely large positive temperature coeffi-cient (PTC) e†ect, depending on the Ðller type or carbon black), as well as(V2O3on its distribution and volume fraction. Both melting and recrystallizationbehaviour are responsible for the PTC e†ect. Due to a conductive Ðlamentarynetwork across the medium, a localized thermal e†ect comes into existence,leading to self-heating of the body. This gives the composites potential applica-tion, for example, in plastic welding. Preliminary experimental results are report-ed.

Polym. Int. 44, 117È124 (1997)No. of Figures : 11. No. of Tables : 4. No. of References : 12

Key words : general e†ective media equation, vanadium sesquioxide, carbonblack, positive temperature coefficient of resistance, thermal expansion, self-heating, plastic welding.

INTRODUCTION

The use of Ðlled conductive polymer composites isattractive for many reasons. They have been successfullyused, for example, for electrostatic dissipation, electro-magnetic interference shielding, and wave absorption.In recent years, interest has been renewed because of theneed for the design of composites which exhibit so-called product properties.1 The most common of theseis a thermistor made up of Ðlled conductive polymers.The thermal resistance e†ect is a result of the product ofthe thermal expansion of the polymer matrix and theformation of conducting pathways in the Ðller phase.The smartness of this kind of composite can be illus-trated by the extremely large positive temperature coef-Ðcient (PTC) anomaly e†ect of electrical resistivity,

* To whom all correspondence should be addressed.

which can sometimes be additionally combined with anegative temperature coefficient (NTC) e†ect.

In this paper, we Ðrstly concentrate on a generale†ective media (GEM) equation. Its applicability todescribe the insulatorÈconductor transition behaviourof conductive polymer composites is tested and dis-cussed. Secondly, PTC and NTC e†ects of di†erentcomposites are demonstrated, particularly of the com-posite made by incorporating vanadium suboxides intoconventional polymer matrices, which exhibit a “wellÏshaped temperature dependence of the electrical resis-tivity. Thirdly, as a potential application of Ðlled con-ductive polymers, it is demonstrated that plastics can bewelded by using Ðlm heaters made up of these compos-ites.

Although most of the data presented are taken fromactual work done at the Institute of Polymers and Pro-cessing (IPP) of Zhejiang University, this is merely a

1171997 SCI. Polymer International 0959-8103/97/$17.50 Printed in Great Britain(

118 X.-S. Y i, G. W u, Y . Pan

matter of accessibility and in no way implies that thesedata are the best or the only data of this nature avail-able.

GENERAL EFFECTIVE MEDIA EQUATION

In the process of studying physical properties of multi-phase mixtures, various models have been given todescribe the dependence of physical properties upon therelative concentration (i.e. volume fraction), but each ofthem may Ðt only a special case, in which strict assump-tions are made.

As far as the electrical conductivity of a binary systemis concerned, McLachlan et al.2 postulated a GEMequation after studying and summarizing various pre-vious works. The equation is written as :

f (pl1@t[ pm1@t)pl1@t] Apm1@t

]r(ph1@t[ pm1@t)ph1@t] Apm1@t

\ 0

where p stands for the conductivity, and the subscriptsm, l and h represent medium-, low- and high-conductivity components, respectively ; f and r arevolume fraction of low- and high- conductivity com-ponents, respectively. f] r\ 1 always holds. A isdeÐned as :

A\ (1 [ rc)/rc \ fc/(1 [ fc)

is the critical volume fraction of the high-rc (\1 [ fc)conductivity component at which the insulator to con-ductor transition occurs.

is related to the geometries and orientations ofrcboth components according to :

rc\ L r/(1 [ L f ] L r) for oriented ellipsoids

rc\ mf/(mf ] mr) for random ellipsoids

where and are phenomenological “demagnet-L r L fizationÏ constants of the two component particles,respectively. As the particle shape is spherical, L is 1/3.If the component is Ðbre- or layer-shaped and is orient-ed in the electric current direction across the medium,L \ 0 ; perpendicular to the current, L \ 1. andmr mfare parameters for a random case. Actually, an e†ectiveL is often used, even in random cases.

Exponent t is deÐned by the following equations :

t \ 1/(1 [ L f ] L r) for oriented ellipsoids

t \ (mf ] mr)/(mf ] mr) for random ellipsoids

Consequently, L and m may be related by the followingequations :

L r\ rc/t L f \ 1 [ [(1[ rc)/t]

mr\ t/rc mf \ t/(1 [ rc)

The GEM equation gives complete information onhow the volume fraction of components a†ects the

medium conductivity by taking into account the intrin-sic conductivities, geometries, arrangement and theorientations to the applied electric Ðelds of both com-ponents. After arbitrarily giving a value to t, andp1, ph

can be calculated by Ðtting the experimental data torcthe GEM equation (1). The whole process is called four-parameter Ðtting. In a similar way to this process, three-parameter Ðtting can be also used by giving aphreasonable value.3

It has been assumed in introducing the GEM equa-tion that : (1) the binary system is microscopicallyhomogeneous ; (2) the particle size distributions are inÐ-nitely wide and that they are in contact with each other,with no voids remaining ; (3) the electrical contactpotential between like and unlike particles is negligible.It was shown in Ref. 2 how most aspects of the Brugge-manÏs e†ective media and percolation theories could becombined into this single GEM equation.

INSULATOR–CONDUCTOR TRANSITION3

The resistivity against volume fraction relation)(omÈrfor conductive particle-Ðlled composites made by com-pression sinter moulding is shown in Fig. 1. The resis-tivity falls abruptly from º1010 to ¹101) cm at Ðllervolume fractions between 0É1 and 0É4, indicative of thematrix polymers. The computer-Ðtted curves run per-fectly through most of the experimental data points.The Ðller material will be discussed in detailV2O3below.

Figure 2 shows the resistivity against volume fractionfor two polyethylene-based matrices. The average sizesof the as-produced and ball-milled Ðller particles are5É96 and 1É66 km, respectively. The ball-milled Ðllergives smaller critical volume fractions and a lower rateof decrease of resistivity. This may suggest that ÐnerÐller favours the formation of an electrical percolation

Fig. 1. Resistivity plotted against Ðller volume fraction forcomposites ; the solid lines indicate the ÐttingV2O3Èpolymer

curves, while the experimental values are designated as points.

POLYMER INTERNATIONAL VOL. 44, NO. 2, 1997

Filled conductive polymer composites 119

Fig. 2. Resistivity plotted against Ðller volume fraction forHDPE and LLDPE composites Ðlled with as-produced andball-milled particles. The lines indicate the Ðtting curvesV2O3

while the experimental values are designated as points.

network across the medium and forms more brancheswithin the network.

Fitting data to the GEM equation gives all the vari-able parameters listed in Table 1. The resistivity values

of 0É037 and 0É092 ) cm are the measured originalohvalues for the as-produced and ball-milled Ðllers,V2O3respectively, whereas the values for the matriceso1result from the Ðtting calculation. As reported in Table1, the calculated values are very di†erent, even wheno1these should be the same original values for the samematrix material (for example linear low-density poly-ethylene (LLDPE). It is also found that most values of tand L (e†ective demagnetization constant of areV2O3)relatively close to each other, independent of the matrix.

From the correlation of size and geometry in the caseof Ðbre-like conductive Ðllers, it follows that large Ðllerparticles favour the formation of conducting paths atlow percolation concentrations,4 whereas for carbonblack (CB) Ðlled polymers, the opposite is true. It isshown that low percolation concentrations often corre-spond to reduced diameters of the primary CB par-ticles.5 Our experiments on (PE)V2O3Èpolyethylene

composites are in agreement with this behaviour ;namely, the Ðner the Ðller size, the lower the criticalvolume fraction.

The fact that all values are smaller than 1/3 inL rTable 1 indicates that the dimension of the par-V2O3ticles along the electric current (perpendicular to thecompression direction during formation) is longer thanthat in the other directions. In fact, an aspect ratio ofabout 5 : 1 is observed by scanning electron microscopy(SEM) measurements.3 The e†ect of matrix materials onthe values is also discussed in Refs 3 and 4. InL rgeneral, L provides the e†ective geometry and orienta-tion of the smallest electroconductive or insulating unitforming the percolation path or insulating network.This unit might be the individual particles or particleaggregates (segregated chains), depending on the parti-cle size and the mixing method.

PTC TRANSITION6

In this paper, materials indicating PTC behaviour arecomposites made up of semi-crystalline polymers asinsulators, characterized by a high thermal expansionand a Ðller of conducting particles, whose concentrationis close to the critical volume fraction. When the tem-perature is raised, the volume of insulator increases,while the volume of the conductor phase, initially abovethe percolation threshold, falls below the critical value,and the conductance of the composite decreases.

Figure 3 illustrates resistivity versus temperature forfour di†erent base polymers, Ðlled with the as-produced

particles.7 The volume fraction for all the com-V2O3posites was constant at r\ 50%. Depending on thepolymers, there is always a steep increase in the resis-tivity near the melting point. At this temperature, thePTC transition occurs.

InÑuence of Ðller loading on the room temperatureresistance and on the magnitude of the PTC e†ect isdemonstrated for CB-Ðlled PE composites in Fig. 4 ;these results were obtained in our laboratory and willbe published elsewhere. At room temperature, Rdecreases with the increase of the CB volume fraction.

TABLE 1. Resulting values by three-parameter fitting4

Samplea rl(1010 O cm) r

h(O cm) t r

cLr L

fd (%)

PVC-1 838 0·037 2·77 0·100 0·030 0·673 1·3

HDPE-1 3·10 0·037 2·16 0·297 0·138 0·675 2·2

HDPE-2 2·26 0·092 2·32 0·276 0·119 0·688 1·4

LLDPE-1 13·5 0·037 2·47 0·324 0·132 0·726 2·5

LLDPE-2 6·20 0·092 2·65 0·317 0·120 0·743 1·4

LDPE-1 4·26 0·037 2·78 0·380 0·137 0·777 2·2

Epoxy-1 172 0·037 1·90 0·261 0·137 0·611 4·3

a The samples are named by the matrix material and a number : 1 ¼as-produced filler ; 2 ¼ball-

milled filler.


120 X.-S. Y i, G. W u, Y . Pan

Fig. 3. Resistivity plotted against temperature for four di†er-ent matrices and polymer composites.V2O3

In the range of the melting point, the resistanceincreases drastically. The lower the volume fraction, thelarger the PTC. At a volume fraction of 20%, the PTCjump is as high as 107.

After the PTC transition, R decreases immediatelyand the NTC transition follows. The NTC e†ect isprobably due to the reorientation, reaggregation orreassembling of the conductive units. It is believed that,in the temperature range of polymer melting, the ini-tially dispersed particles may obtain a mobility whichhelps in self-repair of the broken percolation network.7The NTC magnitude seems to be proportional to theprevious PTC magnitude.

The NTC e†ect can mostly be alleviated by cross-linking (Fig. 5). By this treatment, the PTC jump isfound to be intensiÐed ; it is almost two orders higherthan that before the treatment. In addition, the thermalcycling behaviour is also improved. After three to fourthermal cycles, the room temperature resistivity and thePTC magnitude become stable and reproducible. Many

Fig. 4. Resistance plotted against temperature for carbonblackÈpolyethylene composites, depending on the Ðller

volume fraction.

Fig. 5. E†ect of cross-linking on the PTC transition forPEÈCB composites.

cycles and long-term ageing do not show any degrada-tion. A disadvantage of the cross-linking is that theroom temperature resistivity increased slightly, whichcan be seen as a price one has to pay.

THERMAL EXPANSION BEHAVIOUR

The mechanism for the PTC anomaly in semi-crystalline polymer composites is generally attributed tothe relatively large change in speciÐc volume of thepolymer at its melting temperature. For this reason, thethermal expansion behaviour, measured by the speciÐcvolume(l)Ètemperature relation for thermoplastic semi-crystalline polymers has been studied. An example isplotted in Fig. 6.

Under isobaric conditions, as the polymer melt isvery slowly cooled, the speciÐc volume decreases Ðrstlinearly with temperature. At a sharp bend isT \ Tsolid ,registered in the lÈT curve, indicating the beginning ofrecrystallization of this semi-crystalline polymer. On

Fig. 6. SpeciÐc volumeÈtemperature relation for a matrixpolymer on cooling and on heating.



Fig. 7. PTC transition of a PEÈCB composite on heating andon cooling.

further cooling, the lÈT curve runs “parabolicallyÏdownwards with temperature. At the end of the parab-ola, the curve becomes nearly linear again. In this lowtemperature range, the temperature dependence isslightly lower than that in the high temperature range.

If one heats up the sample from room temperature atexactly the same temperature rate and pressure, it isfound that the speciÐc volume Ðrst increases linearlywith temperature. This is, followed by a parabolic,change which is di†erent from the previous cooling. Theheating curve shifts to a higher temperature range, sothat the melting point measured is higher thanTm Tsolid .The temperature di†erence *T between on heatingTmand on cooling has been measured for someTsolidtypical polymers, for example : *T B 18 K for PE,

*T B 34 K for POM (polyoxymethylene) and*T B 50 K for polypropylene. The hysteresis loopenclosed by the cooling and heating curves is physicallydue to a suppressed crystallization during cooling.8

Similarly to this transition phenomenon, we Ðnd thatthe resistance change of CB-Ðlled PE is also di†erent oncooling, and on heating (Fig. 7). The PTC transitiontemperature upon cooling is obviously lower than thatupon heating, and the PTC intensity is less. This rela-tion between the PTC and lÈT transitions suggests thatthe rapid change in speciÐc volume must be responsiblefor the rapid change in resistance. All factors inÑuencingthe melting behaviour, such as temperature rate andpressure, inÑuence the PTC behaviour of the compos-ites. Further investigations are planned. It is hoped toobtain more detailed information on quantitative rela-tions between the PTC and lÈT transitions.

VANADIUM SUBOXIDE–POLYMER

COMPOSITES3,6

Vanadium suboxides have been studied since the 1950s.It was found that some structural phase transitionsoccurring in vanadium suboxides were accompanied byelectrical resistivity, as well as speciÐc volume changes.For example, in the transition from anti-V2O3 ,ferromagnetic insulator (AFI) to paraelectric metallicconductor (PMC) took place at 160 K, accompanied bya resistivity change from 105) cm to 10~2) cm. Afurther transition followed from paraelectric metallicconductor to paraelectric insulator (PI) at 400 K, corre-sponding to a resistivity increase of up to 103È104) cm.

Fig. 8. Resistivity plotted against temperature for and composites, with regard to the volume fraction and particleV2O3 VO2ÈPEdistribution.


122 X.-S. Y i, G. W u, Y . Pan

Fig. 9. VoltageÈcurrent relation for highly Ðlled PE compos-ites. The experimental values are designated as points for anelectrode distance of 10 mm, and as triangles for a distance of

60 mm.

For a similar transition occurred from semicon-VO2 ,ductor to metallic conductor at 340 K. Because of diffi-culties of sintering and fragmentation during the phasetransitions, these potential thermistor materials havenot yet been applied.

In our study, and powders were preparedV2O3 VO2by reducing in atmosphere at high tem-V2O3 H2perature and in a solution with a dissolved reducingagent. The volume fraction dependence of room tem-perature resistivity for the compositesV2O3Èpolymerhas been reported in Figs 1, 2 and 3.

The resistivityÈtemperature relation is plotted in Fig.8 during heating. Above a critical volume fraction, for

example 30% for in PE, the composites show aV2O3square-well in the oÈT proÐle by combining the NTCand PTC e†ects. The decrease in o at about 173 K(certainly slightly higher than 160 K (113¡C)) is due tothe phase transition from AFI to PMC in Simi-V2O3 .larly, a poorly deÐned change in the compos-VO2ÈPEite happens at 238 K (65¡C). Both the higher transitiontemperatures may be due to the heat transfer resistancewithin the composites.

The NTC resistivity changes suggest that the physicalproperties of the vanadium suboxides are maintained inthe composites. No fragmentation was detected.

DC CONDUCTION

In the case of DC conduction, the currentÈvoltage char-acteristic in most Ðlled conductive polymers satisÐes theequation I\ AUb, where U stands for the appliedvoltage, I the corresponding current, and A the recipro-cal of the resistance. If b \ 1, we have ohmic behaviour.Non-ohmic behaviour is often reported, perhaps(b D 1)due to contact resistance.9 OhmÏs law remains validinside the body of the composite, while at the same timethe overall sample exhibits non-linear currentÈvoltagecharacteristics.9

We measured the IÈU behaviour of two highly Ðlledconductive PE composite samples, designated as S,using a four-electrode device. Here and in the followingtext, “highly ÐlledÏ means a Ðller volume fraction overthe critical threshold. Current readings were takenwithin a few seconds of applying the voltages in orderto avoid joule heating. Figure 9 shows that the samplesbehave ohmically at room temperature, independent ofthe distance (b) between the electrodes.

Fig. 10. Body temperature plotted against time for two PE-based compositions, with respect to the voltages applied.



It is well known that electrical power dissipation in aconductive composite leads necessarily to a rise in bodytemperature. Thus, a self-heating phenomenon occursdue to localized thermal e†ects (Fig. 10). In the experi-ment, the equilibrium voltage measured was obtainedvery quickly. Body temperature was measured by insert-ing a small thermocouple.

Generally, the lower the sample resistance (e.g.sample the higher the heating-up rate, and theS4),lower the voltage applied. A further measurement ofresistance versus temperature (Fig. 11) exhibits an NTCtransition in whereas gives only a straight lineS32 , S4of slightly increased gradient through the melting pointof its matrix material high-density polyethylene (HDPE)(about 130¡C).

Phenomenologically, both the NTC transition andcurrent-controlled negative resistance (CCNR)10 reÑecta common property of highly Ðlled conductive polymercomposites. It is the CCNR e†ect that causes a fasterincrease in temperature due to the reduced resistivityafter the NTC transition of the sample. TheS32 S4sample is thermostabilized, and hence is more easilyelectrically controlled in the use of dissipation energy.

Fig. 11. Resistance plotted against temperature for the twocompositions presented in Fig. 10.

The detailed properties of and will be discussedS32 S4in the next section. The compositions and processingtechnique of samples. andS11, S12 , S21, S22 , S31, S32 S4are listed in Table 2.

Theoretically, the electrical current-induced self-heating may be calculated and predicted by heat equa-tions. The temperature behaviour of the body may bereferred to a model of Ðlamentary networks. Accordingto this, a conductive Ðlamentary network is partiallyformed throughout the composite so that the currentpreferentially Ñows in channels, which are the Ðrst partto be heated, and hence become more conductive abovethe transition temperature of the polymer.

PLASTIC WELDING USING CONDUCTIVE

POLYMERS11,12

The welding principle using highly Ðlled conductivepolymer, usually as Ðlms, is comparable with resistancewelding, for example, electrofusion welding of thermo-plastics. The matrix polymers used for the Ðlms aremostly the same as the plastics to be welded. Manykinds of conductive Ðllers are available includingcarbon blacks, Ðbres as well as metallic powders, Ñakesand Ðbres.

The components are mixed together with some addi-tives and processing aids to form pellets. They are theninjection moulded or extruded to form Ðlms which Ðtthe individual shapes of parts to be welded. At the sametime, electrodes are placed. The welding procedure isthe same as electrofusion welding.

To produce welds of high quality, balanced electricaland mechanical properties of the composite Ðlm are ofimportance, because the addition of Ðller to polymers ata loading level required to achieve high conductivityusually has the e†ect of degrading the toughness charac-teristics. A comparison is reported in Table 3.

The Ðrst three pairs of samples have either a highresistivity with good mechanical properties, or a lowresistivity with poor mechanical properties. The best

TABLE 2. Samples for plastic welding

Sample Polymer matrix Fillera Filler content (wt%) Processinga

S11

HDPE2480 graphite black 25 blending

S12

HDPE2480 graphite black 30 blending

S21

HDPE2480 acetylene black 25 blending

S22

HDPE2480 acetylene black 30 blending

S31

HDPE2480 CB HB-4B 25 blending

S32

HDPE2480 CB HB-4B 30 blending

S4

HDPE2480 CB HB-1 20 blending

a The blending condition was the same for all samples : shear milling at 180¡C for 15 min


124 X.-S. Y i, G. W u, Y . Pan

TABLE 3. Influence of composition on important

properties

Sample r (O cm) sb

(MPa) eb

(%)

S11

130 15·3 10

S12

32 9·7 5

S21

140 24·5 210

S22

30 21·3 120

S31

150 27·1 300

S32

35 27·0 260

S4

2·0 27·3 180

result is obtained with by optimization of its com-S4position. Moveover, because of the high Ðller loading,the sample exhibits a constant resistivity independentS4of the sample geometry and voltage applied. Combinedwith its thermal stability (Fig. 11), is expected to beS4able to be used in welding Ðlms.

The welding conditions (welding pressure p, weldingvoltage U and the welding time t), as well as the weldingresults (weld strength strength coefficient a) arep

b,

given in Table 4. The PE matrix plastic has an originaltensile strength of 28É6 MPa. The welds produced by the

TABLE 4. Welding conditions and weld properties

Sample r (kg cmÉ2) U (V) t (s) sb

(MPa) a

S11

5–10 150 150 9·7 0·34

S22

5–10 45 100 23·4 0·82

S32

5–10 45 70 25·1 0·87

S4

5–10 15 30 27·0 0·94

conductive Ðlm exhibit the highest weld strengthS4among the samples tested. The voltage used is absolu-tely safe, and the welding time is acceptable practically.

ACKNOWLEDGEMENTS

The authors acknowledge the National AdvancedMaterials Committee of China, the Science & Tech-nology Committee of Zhejiang Province and Qilu Pet-rochem. Co. for their Ðnancial support.

REFERENCES

1 Newnham, R. E. & Ruschau, G. R. J. Am. Ceram. Soc., 74 (1991)463.

2 McLachlan, D. S., Blaszkiewicz, M. & Newnham, R. E., J. Am.Ceram. Soc., 73 (1990) 2187.

3 Pan, Y., Wu, G. & Yi, X.-S., J. Mater. Sci., 29 (1994) 5757.4 Lux, F., J. Mater. Sci., 28 (1993) 285.5 Gilg, R. G., in Elektrisch leitende Kunststo†e (1.AuÑ.). Carl Hanser

Verlag, Munich, 1986, p. 55.6 Pan, Y., Wu, G. & Yi, X.-S., Book 2 in Proceedings of Electro-

ceramics V (International Conference on Electronic Ceramic),1996, Aveiro, Portugal, p. 195.

7 Narkis, M., Ram, A., & Flashner, F., J. Appl. Polym. Sci., 25 (1980)1515.

8 Stuart, H. A., Die Physik der Hochpolymeren, Bd. 3. SpringerVerlag, Berlin, 1955.

9 Reboul, J.-P., in Carbon BlackÈPolymer Composites, ed. E. K.Sichel. Marcel Dekker, New York, 1982, p. 82.

10 Reboul, J.-P. in Carbon BlackÈPolymer Composites, ed. E. K.Sichel. Marcel Dekker, New York, 1982, p. 98.

11 Yi, X.-S. & Wu, G., Chinese Patent CN 1083767A, 1994, Beijing.12 McMills, C. J., Altos, L., et al., US Patent No. 5,286,925, 1994.


properties and applications of filled conductive polymer composites

Documents