thermally induced performance decay in conductive polymer composites

270 POLYMER COMPOSITES, JUNE 2004, Vol. 25, No. 3

INTRODUCTION

Conductive polymer composites consisting of in-trinsically insulating polymer matrices and dis-

persed electrical conductive fillers are finding increas-ing applications because of their light weight, goodprocessability by the techniques common for unfilledpolymers, chemical stability, design capability, cost ef-fectiveness, and easy regulation of electrical conduc-tivity, and mechanical performance within a wide range(1). One of the most attractive features of conductivethermoplastic composites is the positive temperaturecoefficient (PTC) effect, which is characterized by adrastic rise in volume resistivity as temperature ap-proaches the melting point of the matrix polymer. Atroom temperature, the conducting filler particles areclosely packed with intimate contact to the neighbors,forming conduction paths throughout the composite.During heating, the polymer expands much more thanthe fillers and the effect is strongly enhanced whenapproaching the melting temperature of the polymer

matrix. Then, the conducting paths established by thefiller particles are obviously interrupted because of theenhanced thermal expansion of the polymer. This leadsto the huge jump in the resistivity. Since the variationis reversible, it can be expected that when an electricfield is applied, the heat output can be automaticallyadjusted in response to the increasing or decreasingtemperature. As a result of the “smart” nature describedabove, polymer based PTC material is capable of servingin self-regulating heaters, microswitches, sensors, etc.

As summarized by Strümpler and Glatz-Reichen-bach (2), a strong PTC of resistance in conductivecomposites close to the melting temperature was firstdiscovered by Pearson at Bell Labs in 1939 (3), thenstudied by Frydman (4) in 1945, forgotten for a while,and rediscovered in 1961 by Kohler (5), who observeda significant resistivity change of about one order ofmagnitude. Since the work of Kohler, much attentionhas been paid to polymeric PTC materials (6�9).

Although polymer based PTC composite materialsexhibit better properties than organic PTC ones, theirservice life is still affected by structural variation result-ing from chemical and physical aging. That is, whenthe composites are exposed to repeated thermal-coldcycling, composite resistivity gradually increases withoperation time. In the meantime, the electrothermal

Thermally Induced Performance Decayin Conductive Polymer Composites

YAN HUI HOU1,2, MING QIU ZHANG2*, and MIN ZHI RONG2

1Key Laboratory for Polymeric Composite andFunctional Materials of Ministry of Education

Zhongshan UniversityGuangzhou 510275, People’s Republic of China

2Materials Science InstituteZhongshan University

Guangzhou 510275, People’s Republic of China

In the course of long-term service, electrically conductive polymer composites act-ing as positive temperature coefficient (PTC) materials are faced with performancedecay characterized by gradually increased room temperature resistivity and de-creased PTC intensity. To reveal the influencing factors and to find appropriateways for solving the problems, thermal-cold cycling experiments (which simulatethe extreme operating conditions of PTC type materials in a laboratory environ-ment) and electrification tests are carried out in the current work. The resultsdemonstrate that irreversible damage of partial conductive networks and, in partic-ular, oxidation degradation induced crystallizability deterioration of the matrixpolymer are responsible for the electrical performance decay. Additionally, an in-crease in the contact resistance formed at the metallic electrode/composite contactsexerts a negative influence on the service life of the composites. Polym. Compos.25:270–279, 2004. © 2004 Society of Plastics Engineers.

*To whom correspondence should be addressed.© 2004 Society of Plastics EngineersPublished online in Wiley InterScience (www.interscience.wiley.com).DOI: 10.1002/pc.20021

Thermally Induced Performance Decay

POLYMER COMPOSITES, JUNE 2004, Vol. 25, No. 3 271

transformation ability and PTC intensity (defined asthe ratio of the maximum resistivity to the room-tem-perature resistivity calculated from the temperature de-pendence of composite resistivity) decrease. The pres-ent work focuses on the factors related to the propertydegradation of PTC type composites, so as to provide aknowledge basis for determining proper measures toproduce qualified materials.

As a continuation of our works on carbon black(CB)�filled immiscible polyblends, CB/ethylene�vinylacetate copolymer/low density polyethylene (CB/EVA/LDPE) composites are employed in this paper as thetarget materials. In this context, the conclusions ofthe previous studies on the same materials (e.g., Ref-erences 10�14) might be helpful in the discussion ofthe experimental data produced in the current article.

EXPERIMENTAL

LDPE and EVA, with identical melt-flow indexes of2.0 g/10 min, were used as the matrix polymers. Anelectrically conductive CB was employed as the filler.

The polymers and CB with the desired proportions(LDPE/EVA � 80/20 and the weight percentage of CBin EVA/LDPE is 18%, which are fixed throughout thepresent work to reduce the complexity of the compos-ites) were melt-mixed in a laboratory-size BrabenderPlasticorder Model XB20-80 at 160°C and 20 rpm. Ineach case, LDPE and CB were mixed for 5 min andthen EVA was incorporated into the mixer for theblending time of 10 min. After being removed andgranulated, the composites with pre-embedded elec-trodes were compression-molded at 140°C and thencooled in air to room temperature. The sheet samples(65 � 45 � 3 mm3) were allowed to rest overnight be-fore subsequent measurements.

Room-temperature crosslinking was conducted byirradiation with a 60Co �-ray source in air to minimizethe undesired negative temperature coefficient (NTC)effect. The radiation dose was 25 Mrad. After irradia-tion, the materials were annealed in an oven at 75°Cfor 10 h. Then, the samples were cooled inside theoven by simply switching off the electricity supply.

The volume resistivity of the composites was meas-ured with a four-lead system. Two current electrodesmade of brass net (45 � 5 mm2) were pre-embeddedat the ends and through the breadth of the rectangu-lar specimens. Two potential electrodes (a pair ofbrass rods 1 mm in diameter and 45 mm in length)were also pre-embedded through the breadth of thespecimens. The center-to-center distance between acurrent electrode and its neighboring potential elec-trode was 8 mm and that between the two potentialelectrodes was 44 mm. In some cases, two-lead meas-urements were carried out by connecting the potentialelectrodes to an ohmmeter.

To examine the variation in the structure-propertyrelationship of the composites in response to alternat-ing temperatures, thermal-cold cycling experimentswere conducted in air. In each cycle, the compositespecimens were heated from room temperature to

130°C within 10 min, kept at 130°C for 5 min, andthen cooled to room temperature. After certain cycles,the temperature dependence of resistivity was recordedby heating the specimens at a heating rate of 2°C/min.In addition, the specimens were electrified to simulatea practical service situation. In this case, alternatingcurrent was applied to the potential electrodes pre-embedded in the specimens. The consumed powerand surface temperature of the specimens were si-multaneously recorded as a function of time.

Differential scanning calorimetry (DSC) studies ofthe composites that had experienced thermal-cold cy-cling were carried out on a TA DSC 2910 apparatus inN2 atmosphere to characterize the melting and crys-tallization behavior. The samples were heated fromroom temperature to 140°C at a rate of 10°C/min andkept for 5 min to eliminate thermal history. Then, thesample was cooled to room temperature at a rate of10°C/min. The variation in heat capacity as a func-tion of time was recorded. Thermogravimetric analysis(TGA) of the composites was made by using a Perkin-Elmer TGA-II apparatus in N2 atmosphere at a heat-ing rate of 10°C/min.

RESULTS AND DISCUSSION

As stated in the Introduction, variations in roomtemperature resistivity and PTC intensity as a functionof working time deteriorate the performance stability ofconductive polymer composites acting as PTC materi-als. To accelerate property decay in the laboratory envi-ronment and to obtain information within a limitedtime, thermal-cold cycling experiments under extremeconditions were conducted, as mentioned in the Exper-imental section. It is worth noting that the ceiling tem-perature of the experiments, i.e. 130°C, would rarely bereached in practice because of the temperature self-reg-ulating ability of the composites themselves.

Figures 1a and 2a show the temperature depend-ence of resistivity of the composites that had experi-enced different cycles of thermal-cold cycling tests.The corresponding changes in room temperature re-sistivity and PTC intensity are plotted in Figs. 1b and2b, respectively. Except for the great difference be-tween the temperature dependence of resistivity of thecomposites as-manufactured (thermal-cold cycles � 0)and the others (Fig. 1a, Fig. 2a ), it is seen that theelectrical reproducibility of the crosslinked compositesis much better than that of the uncrosslinked ver-sions, as reflected by the overlapped curves of the for-mer (cf. Figs. 1a and 2a).

Crosslinking of the matrix polymer was applied be-cause the unwanted NTC effect (characterized by resis-tivity decrease above the melting temperature of thecrystalline phase) can be effectively eliminated (15).On the one hand, the three-dimensionally crosslinkedpolymer network helps anchor the conduction pathsformed by the conductive carbon black particles; on theother hand, the free radicals generated during irradia-tion treatment on the matrix macromolecules mightreact with the active groups on CB, further enhancing

the stability of the conduction paths (14). Therefore,the crosslinked composites are able to withstand therepeated heating load.

It is known that crosslinking of polyolefins occursmainly in the amorphous phase and at the crys-talline/amorphous interface (16, 17). When the tem-perature is high enough to result in melting of mostcrystalline portions in polyethylene, some vulnerablespots on the crosslinked nets might be broken down

consequently. This damage can hardly be healed bythe volume contraction during the subsequent coolingstep. Hence the CB particles fixed by the crosslinkedmatrix networks cannot be restored to their originalplaces, which explains the significant increase inroom temperature resistivity and the resultant de-crease in PTC intensity at the beginning of thermal-cold cycling experiments (Fig. 2b ). Comparativelyspeaking, the electrical networks damaged because of

Yan Hui Hou, Ming Qiu Zhang, and Min Zhi Rong


Fig. 1. (a) Temperature dependence of resistivity, �, of uncrosslinked CB/EVA/LDPE composites experiencing different cycles of thermal-cold cycling tests; (b) room temperature resistivity, �RT, and PTC intensity of the composites versus thermal-cold cycles.

the drastic volume expansion at the melting point ofthe matrix in the uncrosslinked composites are some-what recoverable because of the absence of the cross-linking structure, which leads to a far smaller incre-ment of room temperature resistivity at the initial stage(Fig. 1b).

Examination of Figs. 1b and 2b reveals that when thenumber of thermal-cold cycles exceeds 450, the roomtemperature resistivity of the composites increases

slightly and the PTC intensity decreases with increasingthermal-cold cycles. This means that a more carefulstructural analysis is required to understand the mech-anism involved in the long-term performance degrada-tion. Figures 3a and 4a illustrate the melting behaviorof the composites. In comparison with the specimensas-manufactured, the peak endothermic temperature ofthe composites that had undergone 450 thermal-coldcycles does not change remarkably. This is particularly



Fig. 2. (a) Temperature dependence of resistivity, �, of crosslinked CB/EVA/LDPE composites experiencing different cycles ofthermal-cold cycling tests; (b) room temperature resistivity, �RT , and PTC intensity of the composites versus thermal-cold cycles.

true for the crosslinked composites, suggesting thatthere is no substantial variation in the microstructureof the matrix. Since the electrical decay is rather ob-vious at the beginning of the cyclic heat treatment,it again suggests that the partial breakage of theconducting pathways as a result of matrix melting isresponsible for the attenuation of properties within ashort time.

With a rise in time, either the melting temperatures

or the crystallization temperatures of the compositesdecrease significantly (Figs. 3 and 4). In addition, theprofiles of both the endothermic and exothermic peaksbecome more and more irregular. These strongly dem-onstrate the morphological variation of the matrix poly-mer. That is, the crystalline structure is graduallydestroyed, leading to a rise in the number of the im-perfect crystals or the amorphous portions. Since anincrease in the number of disordered regions in the



Fig. 3. (a) DSC heating traces of uncrosslinked CB/EVA/LDPE composites experiencing different cycles of thermal-cold cycling tests;(b) DSC cooling traces of the composites.

matrix would result in volume expansion of the com-posites (because the density of crystallites is alwayshigher than that of amorphous ones), the conductivenetworks established by CB particles throughout thecomposites have to be further damaged in the courseof the thermal-cold cycling experiments, leading to arise in the room temperature resistivity. On the otherhand, damage of the crystalline phase would alsoreduce the driving force for the PTC effect according

to generally accepted theory (1, 15). Therefore, the PTCintensity of the composites decreases with increasingcycling times of the thermal-cold experiments.

A comparison of Fig. 4 with Fig. 3 reveals that thethermal stability of the crosslinked composites is su-perior to that of the uncrosslinked versions. This canbe attributed to the fact that the crosslinking struc-ture hinders the oxygen diffusion to a certain extentand improves the resistance of the composites to



Fig. 4. (a) DSC heating traces of crosslinked CB/EVA/LDPE composites experiencing different cycles of thermal-cold cycling tests; b) DSC cooling traces of the composites.

physical aging. This deduction can be supported bythe pyrolytic behavior of the composites exhibited inFigs. 5 and 6. After 450 cycles of thermal-cold experi-ments, the pyrolytic temperature corresponding to 50%weight loss, T50%, of the uncrosslinked composites is significantly lowered (Fig. 5), while the decrement of T50% of the crosslinked composites is quite small.When the thermal-cold cycles approach 2700 in num-ber, T50% of the uncrosslinked composites is furtherreduced but that of the crosslinked ones is still main-tained close to a higher value. Since oxidation is mainlyresponsible for aging of polyethylene in air (18) andcan be shielded by crosslinked nets, it is known thatthe damage of the crystalline phases should result fromthe effect of oxidative degradation (19). This means thatchain scission of the matrix polymer takes place as aresult of the action of thermal-oxidation when the com-posite samples experience thermal-cold cycling tests,which reduces the ordered degree of the crystallinezones and leads to increased room temperature resis-tivity and reduced PTC intensity.

Figures 7 and 8 show the results of the specimenselectrified with alternating current. In contrast to theabove experiments conducted in an oven to study theeffect of physical aging, the present tests consider theresponses of the materials to a real working environ-ment. It is seen that both power and surface tem-perature of the composites reach the maxima rapidlyand then decrease with a rise in time. Clearly, an in-creased resistance of the systems, which reduces theelectrothermal performance, should account for thephenomenon. This roughly coincides with the results

summarized in Figs. 1 and 2, but it is still question-able whether the resistivity of the composites in-creases so quickly (e.g. within 24 h) under an alter-nating electric field. Therefore, it is worth studying thevariation in resistivity of the composites before andafter they are electrified (Fig. 9). As far as we know,the four-terminal method is able to eliminate the in-fluence of contact resistances, but the two-terminalmethod cannot. Figure 9 indicates that before they areelectrified, the resistivity of the composites measuredby the two-lead method is close to that measured bythe four-lead method, suggesting the contact resist-ance is negligible. After they are electrified, the situa-tion changes. The resistivity determined by the two-lead method is significantly higher than that by thefour-lead method, besides the fact that the resistivityvalues determined by the two methods are all higherthan those of the composites before being electrified.The former phenomenon has to be related to the re-markably increased contact resistance between themetallic electrodes and the composites, and the lattercan be explained by the aforementioned increase ofcomposite resistivity induced by structure degrada-tion.

Norman reviewed a number of mechanisms of con-tact resistance (20), but the authors of the presentwork believe that the increase in contact resistancerevealed in Fig. 9 results from one of two causes: (i)debonding at the interface between the electrodes andthe composites as a result of the different thermalexpansibilities of the materials, and (ii) the transientlocalized overheat generated by the localized electric



Fig. 5. TGA curves of uncrosslinked CB/EVA/LDPE composites experiencing different cycles of thermal-cold cycling tests (T50%denotes the temperature corresponding to 50% weight loss).

field concentration leads to melting of the matrixpolymer surrounding the electrodes (21). The moltenmatrix facilitates a quick formation of a polymer-richcoverage on the electrodes in consideration of the dif-ferences among the components’ surface free energies

(i.e., 127 mJ/m2 for brass, 42 mJ/m2 for CB, and �26mJ/m2 for LDPE and EVA). For the moment, the ex-perimental data are not sufficient to confirm themechanism. Evidently, further work is needed to findthe solution for the problem.



Fig. 6. TGA curves of crosslinked CB/EVA/LDPE composites experiencing different cycles of thermal-cold cycling tests (T50% denotesthe temperature corresponding to 50% weight loss).

Fig. 7. Time dependence of power and surface temperature of uncrosslinked CB/EVA/LDPE composites electrified with 110V A.C.



Fig. 8. Time dependence of power and surface temperature of crosslinked CB/EVA/LDPE composites electrified with 220V A.C.

Fig. 9. Room temperature resistivity, �RT , of CB/EVA/LDPE composites as-manufactured and those electrified with 220V A.C. for 24 h.



CONCLUSIONS

The work presented here analyzes the mechanismsinvolved in performance deterioration of electricallyconductive polymer composites serving as candidatesfor PTC materials. It is found that redistribution ofcarbon black particulates as a result of repeated melt-ing and crystallization of the matrix polymer accountsfor the electric performance decay of the compositesat the initial stage of thermal-cold cycling. Crosslink-ing of the matrix improves the stability of the materi-als yet makes the damage of the conduction pathwaysirreversible. With respect to the long-term performancedecay of the composites working under thermal-coldcycling conditions, the diminished crystallization abil-ity of the matrix induced by oxidative degradationmust take responsibility. On the other hand, whenthe composites are electrified, an increase in contactresistance between the composites and the metallicelectrodes further accelerates the performance degra-dation of the system in addition to the above-men-tioned causes dealing with the structural variation ofthe composites themselves.

ACKNOWLEDGMENTS

The financial support of the National Natural ScienceFoundation of China (Grant 50133020), the Team Pro-ject of the Natural Science Foundation of Guangdong(Grant 20003038), the Key Programs of the Ministry ofEducation of China (Grant 99198), and the TalentTraining Program Foundation of the Higher EducationDepartment of Guangdong Province is gratefully ac-knowledged.

REFERENCES

1. M. Q. Zhang and H. M. Zeng, “Conducting ThermoplasticsComposites,” in Handbook of Thermoplastics, p. 837,O. Olabisi, ed., Marcel Dekker Inc., New York (1997).

2. R. Strümpler and J. Glatz-Reichenbach, J. Electroceram.,3, 329 (1999).

3. G. Pearson, U.S. Patent 2,258,958 (Oct. 14, 1941).4. E. Frydman, U.K. Patent Spec. 604 695 I 718 14S (July

8, 1948)5. F. Kohler, U.S. Patent 3,243,753 (Mar. 29, 1966).6. D. M. Moffatt, J. P. Runt, A. Halliyal, and R. E. Newham,

J. Mater. Sci., 24, 609 (1989).7. K. Ohe and Y. Natio, Jpn. J. Appl. Phys., 10, 99 (1971).8. J. Meyer, Polym. Eng. Sci., 13, 462 (1973).9. A. Voet, Rubber Chem. Technol., 54, 42 (1981).

10. M. Q. Zhang, G. Yu, H. M. Zeng, H. B. Zhang, and Y. H.Hou, Macromolecules, 31, 6724 (1998).

11. G. Yu, M. Q. Zhang, H. M. Zeng, Y. H. Hou, and H. B.Zhang, Polym. Eng. Sci., 39, 1678 (1999).

12. G. Yu, M. Q. Zhang, H. M. Zeng, Y. H. Hou, and H. B.Zhang, J. Appl. Polym. Sci., 73, 489 (1999).

13. Y. H. Hou, M. Q. Zhang, K. C. Mai, M. Z. Rong, G. Yu,and H. M. Zeng, J. Appl. Polym. Sci., 80, 1267 (2001).

14. Y. H. Hou, M. Q. Zhang, M. Z. Rong, G. Yu, and H. M.Zeng, J. Appl. Polym. Sci., 84, 2768 (2002).

15. M. Narkis and A. J. Vaxman, Appl. Polym. Sci., 29, 1639(1984).

16. G. Wang and B. Jiang, Chin. J. Appl. Chem., 12(6), 17(1995) (in Chinese).

17. G. Wang and B. Jiang, Chin. J. Appl. Chem., 13(4), 68(1996) (in Chinese).

18. M. Narkis, A. Ram, and Z. Stein, Polym. Eng. Sci., 21,1049 (1981).

19. J. Meyer, Polym. Eng. Sci., 14, 10 (1974).20. R. H. Norman, Conductive Rubbers and Plastics, p. 7,

Elsevier, Amsterdam (1970).21. M. Q. Zhang and H. M. Zeng, Acta Mater. Compos.

Sinica, 8, 1 (1991) (in Chinese).

thermally induced performance decay in conductive polymer composites

Documents