failure properties of thermoplastic elastomers from polyethylene/nitrile rubber blends: effect of...

Failure Properties of Thermoplastic Elastomers fromPolyethylene/Nitrile Rubber Blends: Effect of Blend Ratio,Dynamic Vulcanization, and Filler Incorporation

Josephine George,1 N. R. Neelakantan,2 K. T. Varughese,3 Sabu Thomas1

1School of Chemical Sciences, Mahatma Gandhi University, Priyadarshini Hills P.O., Kottayam, Kerala 686 560, India2High Polymer Engineering Division, Department of Chemical Engineering, Indian Institute of Technology, Chennai600 036, India3Polymer Laboratory, Central Power Research Institute, Bangalore 560 080, India

Received 24 February 2003; accepted 3 September 2004DOI 10.1002/app.21381Published online in Wiley InterScience (www.interscience.wiley.com).

ABSTRACT: Thermoplastic elastomers from blends ofhigh-density polyethylene and acrylonitrile butadiene rub-ber were prepared by a melt-blending technique. The blendswere dynamically vulcanized using sulfur, peroxide, andmixed curing systems. The peroxide concentration was var-ied to obtain samples of varying degrees of crosslinking. Theperoxide system showed better mechanical properties. Thecrosslink density determination by the equilibrium swellingmethod revealed that the enhancement in properties can becorrelated to the extent of crosslinking. It is observed thatthe effect of dynamic vulcanization on the property im-provement is much more pronounced in rubber-rich blends.To study the effect of filler incorporation on mechanical

properties, fillers such as carbon black, silica, silane-treatedsilica, and cork-filled samples were prepared. All filled sys-tems, except cork filled, exhibited superior mechanical prop-erties. Scanning electron micrographs of selected fracturedsurfaces were analyzed to study the failure mechanism ofthe different compositions. Various theoretical models wereapplied to correlate the observed mechanical behavior withthat of theoretically predicted values. © 2006 Wiley Periodicals,Inc. J Appl Polym Sci 100: 2912–2929, 2006

Key words: thermoplastic elastomer; vulcanization; me-chanical properties; crosslinking; theoretical models

INTRODUCTION

Polymer blending is one of the easiest and mostwidely used techniques to prepare polymeric materi-als with desired properties. Blending a rubber with aplastic results in thermoplastic elastomers (TPEs).TPEs can be processed like a thermoplastic yet theypossess the properties of vulcanized elastomers. ThusTPEs bridge the gap between plastic and rubber in-dustries. One major drawback that limits the wide-spread acceptance of such systems is the incompati-bility between the phases, resulting in premature fail-ure.

It is well known that the properties of polymerblends depend on the morphology. One favorablemorphology of thermoplastic elastomers is the forma-tion of a large number of finely dispersed un-crosslinked or lightly crosslinked rubber particles in asmall amount of the plastic phase, which is just suffi-cient to retain its flowability. Such a morphology canbe achieved either by the use of a compatibilizer or bythe process known as dynamic vulcanization. In the

first case a block or graft copolymer, having segmentsidentical to the component polymers, is used. Thecompatibilizer can be formed in situ or can be pre-pared separately and then added to the system to becompatibilized. Several excellent reviews on the com-patibilization of polymer blends exist.1–6 A block orgraft copolymer locates at the interface and therebyreduces the interfacial tension, identical to the way bywhich an emulsifier acts.

Datta and coworkers7,8 described the compatibiliza-tion of poly(styrene-co-maleic anhydride)/poly(ethyl-ene-co-propylene) (SMA)/(EPM) blends by the addi-tion of primary amine-modified EPM (EPM–NH2)containing 0–3 mol % amine. A substantial improve-ment in notched Izod impact strength was reported.The improvement in impact strength can be correlatedto the formation of imide bonds that have been provedby IR spectroscopy. These authors also reported that asignificant amount of functionalized EPM can be re-placed by nonfunctionalized high-density polyethyl-ene (HDPE) without a detrimental loss of impactstrength.

The success of compatibilizers has been demon-strated in the studies of several immiscible blend sys-tems.9–14 Okada et al.15 reported on the morphologi-cal, thermal, and mechanical properties of blends of

Correspondence to: S. Thomas ([email protected]).

Journal of Applied Polymer Science, Vol. 100, 2912–2929 (2006)© 2006 Wiley Periodicals, Inc.

nylon 6 with maleated ethylene–propylene rubber. Inthis case the reaction of the polyamide amine endgroup with the grafted maleic anhydride resulted inTPEs with controlled morphology and chemical bond-ing between the phases.

Inferior mechanical properties of compatibilizedblends, compared to noncompatibilized blends, hasbeen ascribed to the macrophase separation of thecompatibilizer.16,17 If the compatibilizer has poor me-chanical properties in the pure state, phase separationmay lead to inferior mechanical properties. When acompatibilizer improves the interfacial adhesion andphase dispersion in a blend, and at the same time hasadverse effects on either or both phases, the resultingmechanical properties of the blend are determined bya combination of the two factors.16

The more physical approach toward compatibiliza-tion for TPEs is by the use of dynamic vulcanization.The principle of dynamic vulcanization is to crosslinkthe rubber phase during its melt mixing with theplastic. For blends with similar polarities, a fine mor-phology is frozen in during dynamic vulcanization.However, a coarse morphology is developed in thecase of incompatible blends.6

Several articles are quoted in the literature18–22 onthe dynamic vulcanization of rubber/plastic blends.Dynamic vulcanization of polypropylene (PP)/EPDMby sulfur18 resulted in a dramatic improvement intensile strength and elongation at break and a reduc-tion in tension set with increase in sulfur content up to2 phr. However, a sulfur system is not used commer-cially for PP/EPDM blends because PP has a highmelting point and the crosslinks lack thermal stability.The dynamic vulcanization of PP/EPDM blend byresoles19 resulted in improvement in tensile strengthand elongation at break. The main reasons for usingresoles are their activities at temperatures above 200°Cand the formation of thermally stable carbon–carbonbonds.

The blend system under discussion consists of high-density polyethylene (HDPE) and acrylonitrile buta-diene rubber (NBR). The very purpose of blendingHDPE and NBR is to develop oil-resistant thermoplas-tic elastomers. The crystallinity of HDPE and the oilresistance of NBR make the system resistant to hotoils. Technological compatibilization is used to reducethe incompatibility problem.

The present article discusses the failure propertiesof HDPE and NBR blends. The effect of blend ratioand dynamic vulcanization on the fracture behavior isanalyzed. The effect of filler incorporation on mechan-ical properties is also brought under investigation.Attempt has been made to correlate the fracture be-havior with the morphology of a few selected samples.Various theoretical models were analyzed to correlatethe experimental result with that of the theory.

EXPERIMENTAL

High-density polyethylene (HDPE; density, 0.96g/cm3; melt flow index, 7.5 g/10 min) was suppliedby M/s Indian Petrochemicals Corp. Ltd. (Baroda,India). Acrylonitrile butadiene rubber (NBR; density,0.98 g/cm3; acrylonitrile content, 32%) was procuredfrom M/s Synthetics and Chemicals (Bareilley, UP,India). Fillers such as carbon black (N �330), cork,silica, and treated silica were used. The particle sizesof carbon black and silica were 29 and 20 nm, respec-tively. Cork filler was of much higher particle size ofthe order of millimeters. Treatment of silica did notmake any difference in particle size. The filler loadingwas based on the rubber content. Only in high rubberblends was the filler loading varied to study the rein-forcement by the filler. First a rubber and filler mas-terbatch was made on a two-roll mill with all theingredients except the curatives. The requisite quan-tity of the masterbatch was then added to the plasticwhile melt mixing to get the exact composition.

Our experiments revealed that dicumyl peroxide(DCP) is the best curative for NBR and in a previousarticle we reported23 on the variation of crosslink den-sity of HDPE/NBR blends with the dosage of DCP.The effect of crosslink density on mechanical proper-ties is studied by varying the DCP dosage from 1 to 4phr based on the rubber content. The DCP used was40% active.

To get an idea about the type of crosslinks on themechanical properties, different crosslinking systemswere studied. The various crosslinking systems usedinclude sulfur, DCP, and a combination of sulfur andDCP. The sulfur curing system consisted of sulfur 0.1phr, tetramethyl thiuram disulfide (TMTD) 2.0 phr,and cyclohexyl benzthiazyl sulfenamide (CBS) 2.5 phr.The mixed curing system contained sulfur 0.05 phr,TMTD 1.0 phr, CBS 1.25 phr, and DCP 2 phr. Zincoxide (ZnO) and stearic acid were used in the propor-tion 5 : 1 in all the mixes.

The melt mixing was carried out at 160°C at a rotorspeed of 60 rpm. The mixing time was 6 min. FirstHDPE was charged into the mixer and allowed tomelt. After 2 min, the rubber or the masterbatch wasadded and allowed to mix for 2 min. The curatives (ifany) were added and mixed for 3 min in the case ofvulcanized blends. The unvulcanized blends weremixed for 4 min.

The unvulcanized blends were represented by theletter H and the subscripts following it represent theamount of HDPE in the blend. Dynamic vulcanizationusing sulfur, dicumyl peroxide, and mixed curing sys-tems were represented by the notations S, D, and SD,respectively. In filled systems C, K, Si, and TSi standfor carbon black, cork, silica, and treated silica, respec-tively, and the subscripts following each notation in-dicate the concentration of filler in the blend. Thus the

THERMOPLASTIC ELASTOMERS FROM PE/NBR BLENDS 2913

uncrosslinked and unfilled blends were representedby H100, H70, H50, H30, and H0, depending on the blendratio. In these, the subscripts indicate the weight per-centage of HDPE in the blend. The notation for filledand dynamically vulcanized blend is explained usingH50 blend as an example. Thus H50C30D4 indicates a50/50 blend of HDPE and NBR, containing 30 phrcarbon black, crosslinked by 4 phr DCP.

Mechanical properties

Tensile testing was done according to the ASTMD412-80 test method using dumbbell-shape samples ata crosshead speed of 500 mm/min using a universaltesting machine (Zwick 1465, Bamberg, Germany).

Tear strength was determined using unnicked 90°-angle test samples according to the ASTM D 624-81test method. The testing speed was identical to that oftensile testing.

Scanning electron micrographs

SEM photomicrographs of tensile and tear fracturedsamples were taken after sputter coating the fracturedsurface with gold–palladium alloy. The morphologiesof dynamically vulcanized samples were studied us-ing liquid nitrogen fractured samples. Because thesamples were crosslinked, the preferential extractionof the rubber phase was not possible. The sampleswere directly used without extraction and SEM micro-graphs were taken on a scanning electron microscope(model 500C; Philips, Eindhoven, The Netherlands).The domain dimensions were measured using an im-age analysis technique by considering a large numberof particles.

Crosslink density determination

The molar mass between crosslinks (Mc) of dynami-cally vulcanized samples were determined by equilib-rium swelling in toluene using the equation24

Mc �� pVsVr

1/3

ln(1 � Vr) � Vr � �Vr2 (1)

where �p is the density of the polymer; Vs is the molarvolume of the solvent; and Vr is the volume fraction ofswollen rubber, which is given by

Vr ��d � f1w��p

�1

�d � f1w��p�1 � As�s

�1 (2)

where d is the deswollen weight of the sample, f1 is thevolume fraction of insoluble component, w is theweight of sample, �s is the density of the solvent, andAs is the amount of solvent absorbed.

The interaction parameter is given by25

� � � �Vs

RT ��s � �p�2 (3)

where � is the lattice constant (0.34); R is the gasconstant; T is the absolute temperature, and �s and �p

are the solubility parameters of the solvent and poly-mer, respectively. The crosslink density � was calcu-lated from Mc as

� �1

2Mc(4)

RESULTS AND DISCUSSION

Brabender curves

Typical Brabender plastographs for unvulcanized anddynamically vulcanized (DCP cured) H70 are shown inFigure 1. In the unvulcanized sample, the torque de-creases with mixing time and finally registers a con-stant value, indicating complete mixing of ingredients.In dynamically vulcanized samples, the mixing torqueincreases with mixing time, which is explained by thefact that the crosslinked rubber particles exert in-creased resistance to rotation, which in turn results inincreased torque values.

The effect of DCP concentration on the final torquevalues of H70, H50, and H30 are shown in Figure 2.

Figure 1 Typical Brabender plastographs showing the ef-fect of dynamic vulcanization on the torque values of a70/30 blend of HDPE and NBR.

2914 GEORGE ET AL.

From the figure it is very clear that the final torquevalues increase with DCP concentration. The effect ofDCP on polyolefins is determined by the structure ofthe base polymer. Peroxide-initiated reactive extru-sion of polyolefin may lead to oxidative degradation,crosslinking, or chain scission, depending on the typeof polymer, compounding conditions, and partialpressure of oxygen.26 With respect to the polymer, thestructure of the pendant group (R) determines thepreferred reaction pathway. In the case of polypro-pylene (PP), (R � CH3), the positive induction effect ofthe methyl group facilitates a homolytic scission of theCOH bonds and the tertiary macroradical formed willreadily undergo chain scission even under a low par-tial pressure of oxygen.

At high oxygen pressure, oxidative degradationpredominates. In addition, the pendant methyl groupsprovide PP with steric hindrance for the coupling oftwo macroradicals. For polyethylene (PE) such an ef-fect or hindrance does not exist, and thus crosslinkingreactions by the combination of secondary macroradi-cals dominate.27,28 Thus the increase in torque withDCP concentration in the present case is a result of thecrosslinking of both the rubber and plastic phases.Also, it is evident that at a particular DCP concentra-tion, the torque increases with rubber content, whichresults from the predominant crosslinking of the rub-ber phase. Additionally when the rubber phase isdominant it becomes continuous phase and thecrosslinked continuous phase shows high viscosity.

Thus when the rubber phase forms the continuousphase the crosslinking is more efficient.

Mechanical properties

Effect of DCP dosage

DCP is capable of crosslinking both phases and thereis the chance of graft formation between HDPE andNBR. It was observed that the tensile strength ofHDPE was adversely affected upon vulcanization byDCP. This is attributed to the reduction in crystallin-ity. In the case of NBR, however, a substantial im-provement in tensile strength can be observed uponvulcanization, so a significant improvement in me-chanical properties can be expected only in rubber-rich blends.

The stress–strain behavior of unvulcanized and dy-namically vulcanized H70 blend, containing differentlevels of DCP, is shown in Figure 3. From Figure 3 it isclear that all the H70 blends show the stress–strainbehavior characteristic of a plastic (i.e., they show veryhigh initial modulus with yielding and necking). It isalso clear that with an increase in DCP content corre-sponding increases in the tensile strength and elonga-tion at break are observed. In unvulcanized H70, therubber is dispersed as large inclusions in the HDPEmatrix because of the high interfacial tension and thepoor adhesion between the phases. It is widely ac-cepted that in an immiscible blend the particles of the

Figure 2 Bar graph showing the effect of DCP concentra-tion on the final torque values of 70/30, 50/50, and 30/70blends of HDPE and NBR.

Figure 3 Stress–strain plots of unvulcanized and dynami-cally vulcanized 70/30 blend of HDPE and NBR containingdifferent levels of DCP.


dispersed phase act as stress- concentrating flaws in-troducing weak points in the matrix material. As aresult an intrinsically tough matrix breaks at a lowerstress and at a lower elongation. Thus the compara-tively large rubber particles observed in H70 act ascrack-initiating flaws, resulting in reduced tensilestrength and elongation at break. With the addition ofDCP, the rubber particles become crosslinked duringmelt mixing and the morphology is substantially mod-ified. As the percentage of DCP increases, the extent ofcrosslinking also increases, leading to the formation offine crosslinked rubber particles in the matrix. Also,morphological stability is imparted by the suppres-sion of coalescence resulting from the decreased dif-fusional mobility of the crosslinked rubber chains. Thecrosslinked rubber particles, if attached to the matrix,can withstand greater stress by undergoing increasedextent of deformation before failure, thus resulting inincreased tensile strength and elongation at break.

The stress–strain curves of unvulcanized and dy-namically vulcanized H50, containing different levelsof DCP, are shown in Figure 4. The tensile strength ofH50 is much lower than the theoretically calculatedvalue (18.4 MPa) based on the additivity rule, which isa clear indication of the incompatibility between thephases. Also, the large rubber particles dispersed inthe HDPE matrix are ineffective in stress transfer andact as stress-initiating flaws, resulting in reduced ten-sile strength. A significant change in the morphology

is observed upon dynamic vulcanization. The finerand stable morphology of the vulcanized compositioncontributes toward enhanced tensile strength.

The stress–strain plots of unvulcanized and dynam-ically vulcanized H30 are presented in Figure 5. Theunvulcanized H30 acts as a weak and soft material.The drastic reduction in the tensile strength of H30may be attributed to the higher proportion of rubberand also to the transition of the rubber phase from thedispersed state to a cocontinuous morphology. How-ever, a substantial increase in tensile strength can beobserved in vulcanized blends. On comparing thethree blends, H70, H50, and H30, it is seen that theincrease in tensile strength with DCP dosage is highestin H30. A 400% increase in tensile strength can beobserved in H30D4.

The variations of the ultimate strength with DCPdosage of the blends H70, H50, and H30 are shown inFigure 6. In all the blends, the ultimate strength in-creases with DCP dosage and the increase is substan-tial in high-rubber blends.

The morphologies of the unvulcanized and dynam-ically vulcanized H70, H50, and H30 are shown in Fig-ure 7(a)–(f). In unvulcanized blends, the rubber existsas large spherical domains in the HDPE matrix untilthe rubber content is 50% [Fig. 7(a) and (b)]. When therubber becomes the major phase, a cocontinuous mor-phology is observed [Fig. 7(c)]. On dynamic vulcani-zation, the morphology is substantially modified. Inall three blends the crosslinked rubber particles exists

Figure 4 Stress–strain plots of unvulcanized and dynami-cally vulcanized 50/50 blend of HDPE and NBR containingdifferent levels of DCP.

Figure 5 Stress–strain plots of a 30/70 blend of HDPE andNBR indicating the effect of DCP dosage on stress–strainbehavior.

2916 GEORGE ET AL.

as dispersed domains with fine morphology. Also,crosslinking imparts morphological stability by sup-pressing the tendency toward coalescence.

The crosslink densities of the blends at differentlevels of DCP were determined by an equilibriumswelling method using toluene as the solvent at 30°C.The variation of Vr (volume fraction of rubber in thesolvent swollen sample) values with DCP dosage forthe three different blends are presented in Figure 8. Inall the blends, Vr values increase with DCP dosage asthe resistance to swelling increases, which is a clearindication of the increased extent of crosslinking withDCP content. At any particular DCP level, the Vr valueis the highest for H70, followed by H50, and H30 showsthe lowest value. This is explained by the fact that theplastic phase merely acts as a filler and the solventuptake is only a result of the rubber phase. Thus H70and H30 can be considered as rubber samples withhigh and low levels of filler (plastic), respectively. Asa result, in a high plastic blend the resistance to swell-ing is high or the solvent uptake is less, thereby re-sulting in high Vr values.

Variation of yield strength and Young’s moduluswith crosslink density of H70 is presented in Figure 9.From the figure it is clear that the yield strength in-creases with increasing crosslink density. The Young’smodulus, calculated from the initial straight-line por-tion, increases up to 3 phr DCP. However, at 4 phrDCP the Young’s modulus shows a significant reduc-tion. This may be explained by the fact that majorstructural rearrangement occurs at this level of DCP,

which in turn modifies the stress–strain curve andthereby affects the modulus values.

Variation of tensile strength and Young’s moduluswith crosslink density of dynamically vulcanized H30is presented in Figure 10. The tensile strength in-creases with increasing crosslink density, increasing2.5 times from 1 to 4 phr DCP. The significant increasein tensile strength is attributed to the increased extentof crosslinking by DCP in high-rubber blends. How-ever, the Young’s modulus values initially show anincrease with increasing crosslink density and there-after a decrease.

The effect of dynamic vulcanization (4 phr DCP) onthe tear behavior of H70, H50, and H30 are shown inFigure 11. From the figure, it is clear that the dynamicvulcanization significantly improves the tear strengthvalues. The reason given for tensile strength improve-ment also holds good for tear strength. In an unvul-canized HDPE/NBR blend the dispersed phase acts asa crack-initiating flaw. In a dynamically vulcanizedblend, however, the dispersed rubber particles en-hance the energy-absorption capacity of the matrix bypromoting and controlling the deformation of the ma-trix. Crazing and shear yielding are the deformationmechanisms. The factors that control the tear strengthinclude the degree of adhesion, degree of crosslinking,and rubber particle size, for example. The particle sizeand the size distribution have a strong influence onthe tear strength. At the optimum size, a large numberof small crazes are formed and thereby enhance itsultimate properties. When the rubber particles areuncrosslinked, the molecular entanglements are insuf-ficient to prevent rapid flow and premature failureoccurs in response to an applied stress. In crosslinkedsamples, however, grafts are possible across the inter-face because the DCP can crosslink both phases. As aresult, the rubber particles can elongate to very largetensile strains, giving a crazelike structure.

The variation of tear strength with DCP dosage ofthe different blends is shown in Figure 12. From thefigure it is clear that the tear strength of the blends H70and H50 increases with DCP dosage. However, in H30,at 4 phr DCP dosage, the tear strength decreases. Thismeans that excessive crosslinking has occurred in H30.Based on the deformation processes occurring justahead of the crack tip in rubber-toughened polymers,Yee29 proposed that certain conditions must be satis-fied for maximum toughening. First, the tougheningparticles should be much smaller than the size of thecrack tip and plastic zone. Second, they shoulddebond or cavitate at stresses just below that for fail-ure of the matrix material, thereby relieving triaxialtension and initiating the formation of shear bands. Inthe present case, the highly crosslinked rubber parti-cles in 4 phr DCP-cured H30 will not effectively trans-fer the stresses to the matrix, thereby resulting inslightly reduced tear strength.

Figure 6 Variation of ultimate strength with DCP dosageof 70/30, 50/50, and 30/70 blends of HDPE and NBR.


Effect of types of crosslinksThe types of crosslinks on the mechanical properties ofthe blends are explained using H30 as an example. Thestress–strain properties of H30 vulcanized using sul-

fur, peroxide, and mixed cure systems are presentedin Figure 13. From the figure it is clear that the DCP-cured system acts like vulcanized rubbers; however,the sulfur system seems to be very weak and brittle.

Figure 7 SEM micrographs showing the effect of dynamic vulcanization on the morphologies of blends of HDPE and NBR:(a) H70, (b) H50, (c) H30 (�1000); and (d) H70D4, (e) H50D4, (f) H30D4 (�500).

2918 GEORGE ET AL.

This may be attributed to the presence of poorlycrosslinked large rubber particles present in the sul-fur-cured system. The coarse rubber particles act asstress-concentrating flaws, resulting in premature fail-ure. In the peroxide-cured system, the crosslinked

rubber particles exert increased torque during mixingand effective particle breakdown occurs. Also, there ischance for graft formation between HDPE and NBRphases. Both of these account for the very good im-

Figure 8 Effect of DCP dosage on the Vr values of differentblends of HDPE and NBR.

Figure 9 Variation of yield strength and Young’s moduluswith crosslink density of dynamically vulcanized H30.

Figure 10 Effect of crosslink density on the tensile strengthand Young’s modulus of dynamically vulcanized H30.

Figure 11 Tear behavior of unvulcanized and dynamicallyvulcanized 70/30, 50/50, and 30/70 blends of HDPE andNBR.


provement in ultimate properties. The mixed curingsystem exhibits stress–strain properties intermediateto those of sulfur- and peroxide-cured systems. Usu-ally C™C linkages present in the peroxide-cured sys-tem are thermally more stable compared to S™S link-

ages seen in the sulfur-cured system; thus some of thecrosslinks (S™S) will be lost at the time of processing.In the mixed cured system both linkages (C™C andS™S) are possible and thereby have intermediate sta-bility. Thus the morphology, crosslink density, andthe type of linkages all account for the differences inthe stress–strain properties of blends vulcanized usingdifferent curing systems.

Crosslink densities of H30 vulcanized using differ-ent curing systems are presented in Table I. It is seenthat the crosslink density of the DCP-cured blend isroughly four times that of the sulfur-cured blend. TheH30 blend, vulcanized using the mixed curing system,exhibits an intermediate crosslink density value. Be-cause crosslink density is one of the major factorscontrolling the mechanical properties of a vulcanizedblend, an increase in mechanical properties can becorrelated with the extent of crosslinking.

The effects of type of crosslinking system on theultimate properties of the blends are presented inTable II. The general trend is the same in all theblends. One can notice that with respect to propertiesthe sulfur system is least favored, followed by mixedand peroxide-cured systems. It can be observed that inthe sulfur system, the tear values are lower than thoseof unvulcanized system, although mixed and perox-ide-cured systems show substantial improvement intear strength. The efficiency of the DCP-cured systemcan best be explained by the morphological changes.In the sulfur system, the rubber is insufficientlycrosslinked and, in fact, at the processing temperaturethe low viscosity melt of the crosslinking agents justrolls over the high viscosity polymer melt and noproper mixing occurs. As a result insufficientcrosslinking occurs. Also because of the lubricatingaction of the ingredients, the already crosslinked rub-ber particles cannot exert resistance to rotation, whichin turn hinders the breaking up of the rubber intosmaller domains. In fact, the large, lightly crosslinkedrubber particles aggravate the phase separation andthe interface becomes very weak, thus resulting ininferior properties. In the DCP-cured system, the cura-tives become properly mixed with the polymer andthe crosslinked rubber particles exert increased resis-tance to rotation during mixing. This increased thrustresults in breaking up of the crosslinked particles into

Figure 12 Effect of DCP dosage on tear strength values ofdifferent blends of HDPE and NBR.

Figure 13 Types of crosslinks on the stress–strain proper-ties of a 30/70 blend of HDPE and NBR.

TABLE IEffect of Crosslinking System on the Crosslink

Density of H30

Crosslinking systemCrosslink density � 104

(gmol/g)

Sulfur 1.2583Mixed (sulfur � DCP) 2.6752DCP 4.3528

2920 GEORGE ET AL.

fine domains. The small size of the particles and theuniformity in particle size distribution both accountfor the enhancement in tear strength values of DCP-cured systems.

In the mixed cured system, the properties are inter-mediate. Here also crosslinking and breaking up of thecrosslinked particles take place but to a lesser extentcompared to that in the DCP-cured system.

The final torque values of unvulcanized and dy-namically vulcanized blends crosslinked by differentcuring agents are presented in Figure 14. It is seen thatsulfur curing results only in a slight increase in thetorque values in all the blends. In fact the extent ofcrosslinking is reflected in the torque values. The lowtorque values indicate that the sulfur systems are in-efficiently crosslinked. The high torque values of DCP-cured blends reveal that it is the most efficient vulca-nization system and the mixed curing system has in-termediate efficiency.

The morphology changes, associated with differentcuring systems, are illustrated by taking H50 as anexample. The morphologies of H50, H50S, H50SD, andH50D4 are presented in Figure 15(a)–(d). In H50 therubber is dispersed as large domains in the HDPEmatrix. The high interfacial tension and poor adhesionbetween HDPE and NBR result in aggressive phaseseparation. During melt mixing fine morphology isobtained because of high shearing action, but on sheet-ing or under service conditions particle coalescencecan occur and the coalesced particles are rejected fromthe matrix, resulting in poorly bonded large rubberparticles. However, in vulcanized samples [Fig. 15(b)–(d)] effective particle breakdown occurs and stabiliza-tion of morphology occurs because of the absence ofthe coalescence process. Also, it is clear that the size ofthe dispersed domains decreases in the order H50S� H50SD � H50D4. The number-average particle sizesfor H50, H50S, H50SD, and H50D are 14.28, 3.49, 2.36,and 1.91 m, respectively. The particle size was mea-sured by an image analysis technique that considers alarge number of particles.

The tensile fractographs of H70, H70D1, H70D2, andH70D3 are shown in Figure 16(a)–(d). In unvulcanizedH70 [Fig. 16(a)] premature failure occurs because of thepoor bonding between the rubber and plastic. How-ever, dynamic vulcanization improves the rubber–plastic adhesion by graft formation. The significantdeformation of the matrix, observed in dynamicallyvulcanized blends [Fig. 16(b)–(d)], is an indication ofimproved adhesion. Also, it is evident that the matrixdeformation increases with increasing DCP concentra-tion.

Effect of filler incorporation

Types of filler and filler loading on stress–strain properties.The stress–strain properties of dynamically vulca-nized H30 and H30, containing 30 phr of differentfillers such as carbon black, cork, silica, and treatedsilica, are shown in Figure 17. It is clear from the figurethat the tensile strength is enhanced by the incorpora-tion of carbon black, silica, and treated silica. How-ever, cork filler results in deterioration in mechanicalproperties. The reinforcement by a filler is dependent

TABLE IIEffect of Crosslinking System on the Ultimate Properties of the Blends

Samplecode

Propertiesa

Tensile strength (N/mm2) Tear strength (N/mm) Elongation at break (%)

U S SD D U S SD D U S SD D

H70 13.6 13.3 17.7 15.2 47.1 31.7 74.8 126.1 36 5 11 132H50 6.6 7.2 11.8 18.8 35.9 31.2 50 93.4 24 15 38 382H30 2.7 3.4 6.6 11.6 16.1 13.6 30.4 53.1 41 13 56 225

a U, unvulcanized; S, sulfur cured; SD, mixed cured; D, DCP cured.

Figure 14 Bar graph indicating the effect of variouscrosslinking systems on the final torque values of 70/30,50/50, and 30/70 blends of HDPE and NBR.


on so many factors such as particle size, size distribu-tion, structure, surface characteristics, and filler–ma-trix adhesion, for example. Fillers of sufficiently smallparticle size will all give about the same order ofmagnitude in reinforcement if there are no great dif-ferences in particle shape. Thus carbon black (29 nm)and silica (20 nm) give almost the same extent ofreinforcement and cork filler results in inferior prop-erties because of its comparatively large size.

The effect of carbon black loading on the stress–strainproperties of H30 is presented in Figure 18. From thefigure, it is clear that the tensile strength shows a mar-ginal increase with increasing filler loading, which is anindication of the reinforcement by the filler. The elonga-tion at break is scarcely affected by filler incorporation.

The effect of filler loading on the ultimate strengthof H30D4 containing different fillers is presented inFigure 19. From the figure it is clear that, except for

cork filler, there is an increase in tensile strength withfiller loading. Also, treatment of silica filler with silanecoupling agent results in improvement in ultimatestrength in all cases except for 30 phr filler. The en-hancement in properties may be explained by theimproved interfacial adhesion between the filler andthe rubber. Usually the coupling agents have an or-ganic and an inorganic part. The coupling agent mod-ifies the interface between the rubber and the filler andenhances the bonding between the filler and the rub-ber. Thus the interfacial adhesion is enhanced, whichis reflected in increased tensile strength.

In the case of cork filler, a reverse trend is observed(i.e., the tensile strength decreases with filler loading).This may be attributable to the insufficient wetting of thefiller by the rubber and may also be attributable to thelarge size of the filler compared to that of C-black andsilica.

Figure 15 Types of crosslinks on the morphologies of a 50/50 blend of HDPE and NBR: (a) H50, (b) H50S, (c) H50SD, and(d)H50D4.

2922 GEORGE ET AL.

Variation of Young’s modulus with filler loading ofdynamically vulcanized H30 is presented in Figure 20.A significant increase in Young’s modulus over theunfilled one is observed in all the fillers, even at 10 phrloading. Even though cork-filled samples have inferiortensile strength, the Young’s modulus is high. Thismay be attributable to the comparatively high modu-lus of the filler. This high modulus contributes towardgood initial load-bearing capacity, but in tensilestrength measurement the interfacial bonding also en-ters the picture. Thus in cork-filled samples a highYoung’s modulus is observed, even though its tensilestrength is affected by filler incorporation. In the othertwo fillers (carbon black and treated silica) theYoung’s modulus values increase but not in propor-tion to the weight fraction of the filler.

Effect of filler on tear properties

The effect of C-black, cork, silica, and treated silica onthe tear force–extension behavior of H30D4 is pre-sented in Figure 21. From the figure it is clear that anenhancement in property is obtained only with silicafiller. As observed in other properties, such as tensilestrength and elongation at break, cork filler results ina considerable decrease in tear strength. Another im-portant difference is that, unlike tensile strength andelongation at break, the tear strength of 30 phr C-black–loaded H30 is less than that of unfilled one. Thiscan be explained on the basis of the deformation pat-tern during tearing. The excessive crosslinking and thehigh loading of the filler make the matrix somewhatrigid. In tear testing the predominant ways of energyabsorption are crazing and shear yielding. For proper

Figure 16 Tensile fractographs showing the effect of DCP dosage on the deformation pattern of H70 blends (�200): (a) H70,(b) H70D1, (c) H70D2, and (d) H70D3.


energy absorption a balance between filler loadingand crosslink density should be maintained. The useof fine-particle silica at lower loadings has been re-

ported to improve the physical properties of thermo-plastic elastomer blends.30 At higher filler loadings thematerial may undergo a brittle type of failure, result-ing in reduced tear strength. This is made clear whenwe look at the effect of carbon black loading on thetear behavior of the H30 blend.

Figure 17 Stress–strain properties of dynamically vulcanizedunfilled and 30 phr filled 30/70 blend of HDPE and NBR.

Figure 18 Effect of carbon black loading on the stress–strain properties of dynamically vulcanized 30/70 blend ofHDPE and NBR.

Figure 19 Effect of filler loading on the ultimate strength ofdynamically vulcanized 30/70 blend of HDPE and NBR.

Figure 20 Effect of filler loading on the Young’s modulusvalues of dynamically vulcanized H30.

2924 GEORGE ET AL.

The effect of C-black loading on the tearing patternof H30D4 is presented in Figure 22. A substantial in-crease in tear strength is observed with only 10 phrfiller, although the strength decreases on adding fur-ther quantity of filler. This may be a result of theincreased filler–filler contact rather than the filler–rubber contact, in highly filled blends.

The tear fractographs of H50, H50D4, and H50K30D4are presented in Figure 23(a)–(c). Unvulcanized H50[Fig. 23(a)] acts as a weak and brittle material becauseof the incompatibility. When a tear force is applied,failure occurs without any matrix deformation, but inH50D4 [Fig. 23(b)] multiple fracture paths can be seen.This is explained by the fact that in dynamically vul-canized H50 a large number of small crosslinked rub-ber particles are present that enhance the rubber plas-tic adhesion. Also, these particles promote and controlthe tearing behavior by transferring the stresses to thematrix. In H50K30D4 [Fig. 23(c)], large debonded do-mains can be seen that aggravate the incompatibilityproblem. These large debonded domains cannothinder a propagating crack and premature failure oc-curs.

Theoretical modeling

In a two-phase system, theoretical models could cor-rectly predict the way by which the phases interact(respond) toward a particular deformation. In an in-compatible blend, some anomaly may be seen in theproperty–composition curve. We could account for

this by carrying out model calculations. The best-fitmodel curve to that of the experimental curve tells usabout the way in which the blend components re-spond toward an applied stress. The various modelsanalyzed include parallel, series, Halpin–Tsai,31 Co-ran’s,32 Takayanagi,33,34 Kerner,35 and Kunori36 mod-els.

According to parallel combination, the property ofthe blend M is given by

M � M1�1 � M2�2 (5)

where �1 and �2 are the volume fractions of the con-tinuous and dispersed phase, respectively. M1 and M2are the properties of phases 1 and 2. The equation forthe series combination of the components is repre-sented as

1M �

�1

M1�

�2

M2(6)

Subscripts 1 and 2 have the same significance as in eq.(1).

Halpin and Tsai31 developed equations to cover thecomplete range of moduli from the lowest lowerbound to the highest upper bound.

The Halpin–Tsai equations are

M1

M �1 � AiBi�2

1 � Bi�2(7)

Figure 22 Effect of carbon black loading on the tearingbehavior of dynamically vulcanized 30/70 blend of HDPEand NBR.

Figure 21 Force–extension behavior of unfilled and filleddynamically vulcanized 30/70 blend of HDPE and NBR.


and

Bi �M1/M2 � 1M1/M2 � Ai

(8)

Subscripts 1 and 2 represent continuous and dispersedphase. Ai is defined by the morphology of the system.For elastomer domains dispersed in a hard continuousmatrix Ai � 0.66.

The equations for Coran’s model32 are

M � f�MU � ML� � ML (9)

where f is a function of phase morphology given by

f � VHn�nVS � 1� (10)

and n contains aspects of phase morphology.VH and VS are the volume fractions of the hard and

soft phases, respectively.According to the Takayanagi model33,34

M � �1 � ��M1 � ��1 � �2�/M1 � �2/M2�1 (11)

where M1 is the property of the matrix phase, M2 is theproperty of the dispersed phase, and �2� is the vol-ume fraction of the dispersed phase and is related tothe degree of series–parallel coupling. The degree ofparallel coupling of the model can be expressed by

% Parallel � [�2(1 � �)/(1 � �2�)] � 100 (12)

In Kerner35 model a new factor, Poisson’s ratio, entersthe picture. The equation for this model is

Figure 23 Tear fractographs of 50/50 blend of HDPE and NBR (�200): (a) H50, (b) H50D4, and (c) H50K30D4.

2926 GEORGE ET AL.

Eb � Em��dEd/�7 � 5�m�Em � �8 � 10�m�Ed

� �m/15�1 � �m�

�dEm/�7 � 5�m�Em � �8 � 10�m�Ed� �m/15�1 � �m�

� (13)

where Eb is the blend property, �m is Poisson’s ratio,and � is the volume fraction. The subscripts m, d, andb stand for the matrix, dispersed phase, and blend,respectively.

Kunori and Geil36 developed a model to account forthe tensile failure of a blend arising from a lack ofadhesion between the blend components. When thereis no adhesive force between the blend components,the tensile strength of the blends

b � m�1 � Ad� (14)

where b and m are the tensile strength of the blendand the matrix, respectively, and Ad represents thearea occupied by the dispersed phase in transversecross section. Kunori and Geil36 assumed that when astrong adhesive force exists between the blend com-ponents, the dispersed phase will contribute to thestrength of the blend and therefore the parallel modelmay be modified as

b � m�1 � Ad� � dAd (15)

If the force propagates mainly through the interfacethe above equation may be written as

b � m�1 � �d2/3� � d�d

2/3 (16)

and if the force propagates through the matrix, thenthe equation becomes

b � m�1 � �d� � d�d (17)

which is same as parallel model.For filled systems, the simplest equation for the

reinforcement or the increase in rigidity attributed to afiller is given by the Einstein equation. This equation isvalid only at low concentrations of filler, when there isperfect adhesion between the phases and can be rep-resented as

G � G1�1 � 2.5�� (18)

where G is the modulus of the filled system, G1 is themodulus of the unfilled system, and � is the volumefraction of the filler. Einstein’s equation implies thatthe stiffening or reinforcing action of a filler is inde-pendent of the size of the filler particles. The equationshows that the volume occupied by the filler is inde-pendent of the size of the filler particles and the vol-ume occupied by the filler, rather than its weight, is

the important variable. The equation also assumes thatthe filler is considerably more rigid than that of thematrix.

An extension of Einstein’s theory, originally devel-oped to explain rubber reinforcement, is ascribed toGuth37 and Smallwood.38 Their equation for the in-crease in modulus arising from a rigid spherical filleris

G � G1�1 � 2.5� � 1.41�2� (19)

where each parameter has the same significance asthat of the previous equation.

The curves resulting from different models and thatof the experimental data for the variation of Young’smodulus with weight percentage of NBR of unvulca-nized blends are shown in Figure 24. From the figureit is clear that the Takayanagi model, with 15% parallelcoupling, gives the best-fit curve. The parallel andseries models show extreme deviations because in anincompatible blend, like HDPE and NBR, uniformstress distribution is virtually impossible. Also, insuch blends premature failure may occur because ofthe stress concentration at the interface.

The different model curves and experimental curvefor the variation of Young’s modulus with weightpercentage of NBR of dynamically vulcanized (4 phrDCP) blends are presented in Figure 25. The seriesmodel is the lowest bound and the parallel model isthe upper bound over the entire composition range.

Figure 24 Various theoretical models for the variation ofYoung’s modulus with weight percentage of NBR for un-filled and unvulcanized blends of HDPE and NBR.


Coran’s model shows positive and negative deviationsfrom the experimental values. Out of the various mod-els analyzed, the Takayanagi model gives the best-fitcurve over the entire composition range.

The experimental and theoretical variations ofYoung’s modulus with filler loading of dynamicallyvulcanized H30 are presented in Figure 26. The exper-imental values are higher than those of the theoreticalmodels studied.

CONCLUSIONS

Dynamic vulcanization of HDPE/NBR blends results inan increase in the mixing torque values as a consequenceof the crosslinking of the dispersed domains. Out of thethree vulcanization systems analyzed, the increase intorque was highest in DCP-cured blends, which is ex-plained by the fact that DCP is capable of crosslinkingboth phases. The increase in torque with DCP dosagewas substantial in high-rubber blends. Mechanical prop-erties indicated a significant enhancement with DCPdosage. Again, the increase was more in rubber-richblends, attributed to the increase in the extent ofcrosslinking. The nature and types of crosslinks alsohave a significant effect on the mechanical properties.The peroxide-cured system is excellent, followed by themixed cured system, and the sulfur system showed theleast improvement. Significant morphology transforma-tion occurred as a result of dynamic vulcanization. In the

unvulcanized blend, rubber exists as large particles, thusresulting in inferior properties. However, the rubber par-ticle size decreases with the efficiency of the crosslinkingsystem and its weight percentage. Thus in the presentstudy the 4 phr DCP-cured system resulted in a substan-tial number of fine rubber particles; however, in thesulfur-cured system the rubber particles are compara-tively larger. The particle size of the mixed cured systemwas between that of sulfur and the DCP-cured system.The fractured surface analysis indicated that prematurefailure occurs because the interface is weak in unvulca-nized blends. On the other hand, in dynamically vulca-nized (especially DCP-cured) blends, no such debondingoccurs. In fact in such blends yielding of the matrix canbe observed. Thus effective stress transfer could occur indynamically vulcanized blends containing small anduniformly distributed crosslinked rubber particles. Fillerincorporation revealed that the reinforcement resultsonly in high-rubber blends. Thus carbon black, silica,and treated silica could marginally improve the mechan-ical properties of the 30/70 blend of HDPE and NBR.Variation of mechanical properties with filler loadingsuggested that within the limits of study, the tensilestrength increased with filler loading, although the tearstrength was adversely affected. However, an increase intear strength can be observed at 10 phr of treated silica.Treatment of silica with silane coupling agent resulted ina marginal improvement in properties, which may beexplained by the improved interfacial bonding betweenthe polymer and the filler. Theoretical modeling sug-

Figure 25 Various theoretical models for the variation ofYoung’s modulus with weight percentage of NBR for dy-namically vulcanized blends of HDPE and NBR.

Figure 26 Theoretical models for the variation of Young’smodulus with filler loading of dynamically vulcanized H30.

2928 GEORGE ET AL.

gested that the Takayanagi model, with 15% parallelcoupling, relates well to the experimental curves in bothunvulcanized and dynamically vulcanized blends. In thecase of a filled system the experimental curve lay abovethe theoretical curves.

References

1. Xanthos, M.; Dagli, S. S. Polym Eng Sci 1991,31, 929.2. Brown, S. B. in Reactive Extrusion, Principles and Practice;

Xanthos, M., Ed.; Hanser: New York, 1992; p 75–199.3. Liu, N. C.; Baker, W. E. Adv Polym Technol 1992, 11, 249.4. Bonner, J. G.; Hope, P. S. in Polymer Blends and Alloys; Folkes,

M. J.; Hope, P. S., Eds.; Blackie: Glasgow, UK, 1993; p 46–74.5. Majumdar, B.; Paul, D. R. in Polymer Blends; Paul, D. R.; Buck-

nall, C. B., Eds.; Wiley: New York, 2000; Vol. 1, Chapter 17.6. Koning, C.; Duin, M. V.; Pagnoullie, C.; Jerome, R. Prog Polym

Sci 1998, 23, 707.7. Datta, S.; Dharmarajan, N.; Ver Strate, G.; Ban, L. in New

Advances in Polyolefins; Chung, T. C., Ed.; Plenum: New York,1993; p 197–207.

8. Datta, S.; Dharmarajan, N.; Ver Strate, G.; Ban, L. Polym Eng Sci1993, 33, 721.

9. England, W. P.; Stoddard, G. J.; Scobbo, J. J. U.S. Pat. 5,310,795(1994).

10. Kitayama, N.; Keskkula, H.; Paul, D. R. Polymer 2001, 42, 3751.11. Kim, K. J.; Jeong, W. Y. Polymer 2001, 42, 4423.12. Zhang, J.; He, J. S. Polymer 2002, 43, 1437.13. Castellano, M.; Nebbia, D.; Turturro, A.; Valenti, B.; Costa, G.;

Falqui, L. Macromol Chem Phys 2002, 203, 1614.14. Tedesco, A.; Krey, P. F.; Barbosa, R. V.; Mauler, R. S. Polym Int

2002, 51, 105.

15. Okada, O.; Keskkula, H.; Paul, D. R. Polymer 2001, 42, 8715.16. Chang, F. C.; Hwu, Y. C. Polym Eng Sci 1991, 31, 1509.17. Chen, C. C.; Fontan, E.; Min, K.; White, J. L. Polym Eng Sci 1988,

28, 69.18. Coran, A. Y.; Patel, R. Rubber Chem Technol 1980, 53, 141.19. Radusch, H. J.; Lammer, E.; Lupke, T.; Haussler, L.; Sandring,

M. Kautsch Gummi Kunstst 1991, 44, 1125.20. Olivier, E. J. U.S. Pat. 5,003,003, 1991.21. Moffett, A. J.; Dekkers, M. E. J. Polym Eng Sci 1992, 32, 1.22. Coran, A. Y.; Patel, R. Rubber Chem Technol 1983, 56, 1045.23. George, J.; Ramamurthy, R.; Varughese, K. T.; Thomas, S. J

Polym Sci Part B: Polym Phys 2000, 38, 1104.24. Khinnawar, R. S.; Aminabhavi, T. M. J Appl Polym Sci 1991, 42,

2321.25. Fedors, R. F. Polymer 1979, 20, 1087.26. Al Malika, S.; Amir, E. J. J Nat Rubber Res 1986, 1, 104.27. Chodaok, I.; Lazar, M. Angew Makromol Chem 1982, 106, 153.28. Borsig, F.; Fieldlerova, A.; Lazar, M. J Macromol Sci Chem A

1981, 16, 513.29. Yee, A. F. in Proceedings of the 3rd European Symposium on

Polymer Blends, Cambridge, UK, July 1990.30. Ramos, L. F.; Valle, D.; Ramirez, R. R. Rubber Chem Technol

1982, 55, 1328.31. Nielson, L. E. Rheol Acta 1974, 13, 86.32. Coran, A. Y. in Handbook of Elastomers, New Developments

and Technology; Bhowmick, A. K.; Stephens, H. L., Eds.; MarcelDekker: New York, 1988; p 249.

33. Dickie, R. A. J Appl Polym Sci 1973, 17, 45.34. Holsto-Miettiner, R. M.; Seppala, J. Y.; Ikkala, O. T.; Reima, I. T.

Polym Eng Sci 1994, 34, 395.35. Kerner, E. H. Proc Phys Soc London 1956, 69B, 808.36. Kunori, T.; Geil, P. H. J Macromol 1960, 218, 36.37. Guth, E. J Appl Phys 1945, 16, 20.38. Smallwood, H. M. J Appl Phys 1944, 15, 758.


failure properties of thermoplastic elastomers from polyethylene/nitrile rubber blends: effect of...

Documents