Low Field H RMN Investigation of Plasticizer and Filler Effects in EPDM

SAVE OFFLINE

LOW FIELD 1H NMR INVESTIGATION OF PLASTICIZER AND FILLER EFFECTS IN EPDM RICHARD J. PAZUR,1,* D. LEE,1 F. J. WALKER,1 MAXIM KASAI2 1

FREUDENBERG-NOK GENERAL PARTNERSHIP (FNGP), PLYMOUTH, MI 48170-2455 2 FREUDENBERG FORSCHUNGSDIENSTE KG, WEINHEIM, GERMANY

RUBBER CHEMISTRY AND TECHNOLOGY, Vol. 85, No. 2, pp. 295–312 (2012)

ABSTRACT A series of compounds based on peroxide-cured EPDM were prepared with varying amounts of paraffinic plasticizer and carbon black. Modeling of the NMR relaxation signal was successfully carried out by either a biexponential or triexponential fitting procedure. It was found the degree of plasticization correlated directly with the average molar mass between chain entanglements (Me) calculated from the short decay constant T21. Values of Me correlated to the dynamic properties (storage modulus and tan d) in the unvulcanized state, thus providing a measure of processability. An increase in carbon black concentration brought about a decrease in Me because of increased interactions between the filler and the polymer chain. A new parameter Mchain–filler is introduced to estimate the average molar mass between polymer chains and fillers. Compared with the chain entanglement density, the overall magnitude of this interaction appears to be weak in the mobile zone of the compound matrix. As in the case of plasticization, a relatively good correlation is obtained between Me and the dynamical properties in the unvulcanized state. Compression-set resistance is shown to directly follow the average molar mass between cross-links (Mc) before and after aging. The carbon black study results can be understood within the context of a morphological model containing different zones of chain mobility—a thin shell of immobilized chains, an intermediate zone of limited mobility, and a dominant mobile phase consisting mainly of entangled and cross-linked polymer chains. [doi:10.5254/rct.12.88944]

INTRODUCTION It is a well-known fact that cross-link density is a fundamental property of thermoset rubbers because of its direct influence on modulus, tensile, and tear strengths; on ultimate elongation, flex fatigue, and hysteresis; and on heat buildup, tension, compression set, as well as its resistance to abrasion.1 Probably the most powerful analytical technique to quickly characterize the state of cure in rubber networks is low field 1H solid-state Nuclear Magnetic Resonance (NMR), which involves the measurement of the transverse T2 spin–spin relaxation by using a Hahn-Echo (HE) or Carr– Purcell–Meiboom–Gill (CPMG) pulse sequence.2 In the simplest of terms, it is possible to relate the NMR relaxation data to the polymer chain mobility, which, in turn, is influenced by the presence of physical (chain entanglements) and chemical cross-links. Besides polymer chain constraints caused by chemical cross-links or chain entanglements, the total NMR relaxation signal contains additional information that is important in characterizing the state of the elastomer chain within a typical rubber compound. The polymer chain mobility and hence, the relaxation signal, is directly affected by reinforcing fillers, such as carbon black (size and structure), polymer crystallinity, and the presence of the so-called ‘‘sol’’ or liquid fraction.3,4 Multiple quantum (MQ) NMR has been shown to be independent of the reinforcing filler effect when compared with analysis by HE relaxation.5 The NMR relaxation studies involving the determination of chemical cross-link and physical entanglement densities for EPDM are well understood.6–8 Plasticizers are often added to rubber compound formulations to depress the glass transition point to improve the functional properties of the material when in use at lower temperatures. NMR relaxation curves of FKM swollen with acetone have been reported, showing that the relaxation time constant T2 increases as a function of

*Corresponding author. Current address: Department of National Defence, Quality Engineering Test Establishment, Polymeric Materials & Advanced Textiles, Ottawa, Ontario K1A 0K2 Canada; Ph. 819-997-9087; email: [email protected]

295

296

RUBBER CHEMISTRY AND TECHNOLOGY, Vol. 85, No. 2, pp. 295–312 (2012)

swelling time.3 To the best of our knowledge, a systematic study examining the effect of the addition of a plasticizing agent to rubber, such as EPDM, on the T2 relaxation signal has not been thoroughly undertaken. The nature of filler–elastomer interactions is a subject of current interest in rubber academia and industry.9 The HE pulse sequence has been employed for investigating filler–rubber interactions in natural rubber,10–13 cis-polybutadiene,14,15 styrene-butadiene,16–19 silicone,20 polyacrylate,21 and EPDM14,22 rubbers. Reinforcing fillers, such as carbon black or silica, were examined. In the case of carbon-black–elastomer systems, a morphological model has been proposed to describe the different zones of polymer mobility between the bulk polymer and the carbon black surface.9 A thin layer of tightly bound rubber, a loosely bound intermediate rubber zone, and finally, an unbound rubber (pure gumlike or bulk polymer) region make up the three identifiable regions of different molecular mobilities according to NMR investigations. Connective filaments or bridges from the intermediate zone are believed to attach the filler aggregates together. It is estimated that the unbound or extractable rubber portion varies from 70 to 90%, leaving the loosely and tightly bound polymer chains at 10 to 30%, depending on the compound. The immobilized regime (so-called bound rubber) in EPDM has been estimated to be approximately 1 nm thick, with the remaining polymer chains comprising loosely bound and mobile sections and a few adsorption network junctions with the carbon black surface.22 Adequate curve fitting of the magnetization decay in EDPM compounds can be accomplished by using a combination of exponential functions as displayed in Eq. 1,       MðtÞ t t t ¼ A exp þ B exp þ C exp ð1Þ Mo T21 T22 T23 where T21, T22, and T23 represent the decay times of the rigid (chemical cross-links and physical entanglements), dangling chain-end contribution and highly mobile fractions due to the sol fraction (T21 < T22 < T23). The values A, B, and C reflect the relative amount of proton magnetization of the rigid, dangling chain end and the sol fractions in the compound. Eq. 1 may be shortened to a biexponential equation in cases (T23  ‘) where a long-term exponential tail is not visible. A calculation of the number average molar mass between the cross-linked chains (Mc) is now possible, as shown in Eq. 2,6 Mc ¼

C‘ T21 MU aT2rl N

ð2Þ

where C‘ represents the number of backbone units per statistical Kuhn-segment, MU is the average molar mass of one monomer unit, a is described as a theory coefficient dependant on the angle between the segment axis and the internuclear vector for the closest nuclear spins on the main chain, T2rl is a measure of the strength of the intrachain proton–proton interactions in the rigid lattice. and N is the number of backbone bonds per monomer. For EPM/EPDM polymers, C‘ ¼6.62 at 363 K, the coefficient a is 6.2 – 0.7, and the value of T2rl is 10.4 – 0.2 ls at 140 K.6 Assuming tetrafunctional cross-linking, the conversion of Eq. 2 to a cross-link density (mc) can be accomplished by using mc ¼ q/(2Mc), where q is the density of the elastomer. For gum elastomers or unvulcanized compounds, Mc can be replaced by Me, which provides an estimate of the number average molar mass between physical entanglements. Finally, it is assumed for the sake of this investigation that the chosen filler and plasticizer levels do not invalidate the use of Eq. 2 because of possible non-Gaussian chain behavior. The purpose of this study is twofold. The effect of adding either a plasticizer or carbon black on the HE NMR relaxation curves for a peroxide-curable EPDM will be investigated. In both cases, this effect will be quantified in terms of Me and Mc for the unvulcanized and vulcanized compounds,

LOW FIELD 1H NMR

297

TABLE I PEROXIDE-BASED EPDM RECIPES OF VARYING PLASTICIZER AND CARBON BLACK LEVELS

Ingredient

Amount, phr

EPDM N550 Plasticizer TMQ Proc. aid DCP (40%) Total

100 0–75 0–30 1 1 5 107–212

respectively. An effort will be made to determine the carbon black contribution (because of its interaction with polymer chains) to the total, effective cross-link density, as measured by HE NMR. Correlations of NMR-derived data to dynamic mechanical results in the unvulcanized state will also be carried out. EXPERIMENTAL MATERIALS

The EPDM (Vistalon 2504) base elastomer (57.5 wt% ethylene, 4.7% wt ethylidene norbornene, ML 1þ4@125 8C¼25 Mooney units) was supplied by Exxon-Mobil (Irving, TX). The N550 grade of carbon black (Sterling 6630) provided by Cabot Corporation (Boston) possesses a surface area of 40 m2/g and a dibutyl phthalate absorption of 120 cm3/100 g. 2,2,4-trimethyl-1,2dihydroquinoline (TMQ) was used as antioxidant (Crompton Corp., Middlebury, CT). A highlyparaffinic plasticizer (Mw ¼690 g/mol) from Holly Corp. (Dallas), known as Sunpar 2280, was used as plasticizing agent. The processing aid (Aflux 42M) which comprises a blend of fatty acids, fatty acid esters, and fatty alcohols came from Lanxess Corp (Pittsburgh). Finally, a 40% active dicumyl peroxide (DCP, Di-Cup 40KE) on a clay carrier from Arkema Corp. (King of Prussia, PA) was used to initiate vulcanization. These six materials comprise the base formulation combinations presented in Table I. For the plasticizer study, the N550 carbon black loading was fixed at 60 phr as the six plasticizer concentrations were mixed at 0, 5, 10, 15, 20, and 30 phr. No plasticizer was used in the carbon black investigation, for which six formulations of 0-, 15-, 30-, 45-, 60-, and 75-phr carbon black were prepared. Mixing of the formulations presented in Table I took place in a laboratory-size Brabender (Duisburg, Germany) internal mixer, possessing a mixing chamber capacity of 375 cm3 and Banbury-type rotors. The rotors turned at 25 rpm, and the starting temperature was 25 8C. The EPDM polymer was introduced at time 0 and allowed to masticate for 1 min; thereafter, all ingredients were introduced to the mixing chamber and mixed an additional 3 min before dumping the batch. A sweep was performed at 2.5 min. Afterward, the batches were passed at the same frequency on a 15 cm 3 33 cm two-roll open mill, set at 25 8C. The total mixing procedure was scrupulously adhered to, to prevent any overmixing or undermixing between successive batches. A ‘‘mixed-EPDM’’ batch subjected to the same mixing procedure was also included to investigate the effect of mixing alone on EPDM without addition of filler, processing aid, antioxidant, and peroxide. Processing properties of the mixed batches were measured by using a rheoTech MDpt moving die processability tester, operating at variable frequency and strain levels at 125 8C, according to the

298

RUBBER CHEMISTRY AND TECHNOLOGY, Vol. 85, No. 2, pp. 295–312 (2012)

procedure outlined in ASTM Standard D 6204, Method A. Tensile slabs were prepared on a Lawton (De Pere, WI) 45-metric-ton compression mold set at 180 8C using a cure time of 6 min. Standard ASTM procedures were followed for hardness (type A), stress–strain, and compression-set testing (plied specimens). The molecular weights of the virgin and mixed EPDM, as well as selected compounds mixed for the carbon black study, were determined with gel permeation chromatography (GPC). All samples were first subjected to a Soxhlet extraction for 24 hours at room temperature with methanol to remove, in particular, the organic peroxide. Dissolution of the extracted samples took place in trichlorobenzene at 145 8C at a concentration of 3.5 mg/mL; 1.0 mg/mL antioxidant was added to that solution, and afterward, the samples were placed in an oven at 145 8C for 5 hours. A series of glass microfiber filters (1.0 lm and 1.6 lm pore sizes) provided by Whatman (Maidstone, Kent, U.K.) were used for sample high temperature filtration. A flow rate of 1.2 mL/min on a Jordi DVB mixed-bed column (500 3 10 mm i.d.; Jordi Labs, Bellingham, MA) was employed. The column temperature was maintained at 145 8C, and the injection size was 200 lL. Polystyrene standards with a concentration of 0.5 mg/mL were used for calibration purposes (molecular weights: 8910K, 4410K, 1090K, 899K, 184K, 100K, 17.6K, 9.73K, 3.47K, and 590) with an injection volume of 100 lL. The samples were monitored at a sensitivity of 16 and a scale factor of 64 using a Waters (Wilmslow, Cheshire, U.K.) 150-C instrument. Jordi GPC software was used for data acquisition and analysis. Molecular weight fractions were distinguished by using a refractive index detector. Each sample was run in duplicate through the column (average value is reported), and the maximum error in each molecular weight average was approximately 10%. T2 relaxation data were collected on a Bruker (The Woodlands, TX) Minispec mq 20 NMR spectrometer. set at a resonance frequency of 20 MHz. The standard HE pulse sequence [908/t/1808/ t(acquisition)] was employed using a time delay between pulses of 1 s. The EPDM-sample relaxation data were measured at 363 K and subsequently analyzed using curve-fitting software (Origin, OriginLab, Northampton, MA). The experimental error (–3r) of the values of T21 is estimated to be no higher than 1.5%.23 Considering that the constants used for calculations in Eq. 2 only act as multiplying factors to transform the values of T21 into a Mc or Me value, only the error due to the calculation of T21 is reported. Finally, a cautious approach to the interpretation of curve fitting of NMR relaxation is advised because of the total error in curve-fitting parameters (Eq. 1), which rises when the total number of parameters increases.24 RESULTS AND DISCUSSION PLASTICIZER EFFECTS IN EPDM

Baseline data of the uncured, dynamical properties (storage modulus and tan d) of the six EPDM formulations are illustrated in Figure 1a,b. Upon low (7%) or high strain (100%) imposition, the elastic modulus increases as a function of the frequency for all compounds. A systematic lower modulus is measured on addition of the plasticizer, with the effect correlating well with the increasing plasticizer concentration. Plasticizer addition increases the viscous component of the overall viscoelastic behavior of the compound, as illustrated by the higher tan d values. These observations are in line with expectations because highly plasticized compounds are predicted to have better flow properties, which include improved flowability at rubber-processing temperatures and, conversely, cold flow issues at ambient temperatures. The unvulcanized and vulcanized EPDM compounds were subsequently tested in the NMR spectrometer and selected spin–spin relaxation curves are presented in Figure 2. In the graph notation, U and V correspond to unvulcanized and vulcanized, respectively, with the ensuing number representing the phr level of plasticizer. A wide range of exponential decay type curves was

LOW FIELD 1H NMR

299

FIG. 1. — Frequency sweep at 125 8C for the plasticized (phr) EPDM at 7 and 100% strains for (a) storage modulus, and (b) tan d.

recorded with the main trend of an increasing baseline (long decay portion) with increasing plasticizer content (vulcanized or not). To accurately curve-fit the series of relaxation curves for the six plasticized EPDM formulations, the triexponential fitting function (Eq. 1) was necessary. All output results of this computational operation are tabulated in Table II. The concentration of plasticizer is presented in both phr and vol%, and the amplitude values of A, B, and C were normalized to 100%. The total error from curve fitting for the latter values was about 1%. The increasing baseline effect, as shown in Figure 2, is mirrored in the values of C, which reflect the relative amount of magnetized protons due to the liquid or highly mobile portion of the decay curve. The increase in the amplitude of C can be safely attributed to the increase in plasticizer concentration, which appears to correlate directly to the vol% of plasticizer added to the recipe (unvulcanized and vulcanized). The amount of the rigid

300

RUBBER CHEMISTRY AND TECHNOLOGY, Vol. 85, No. 2, pp. 295–312 (2012)

FIG. 2. — Selected NMR relaxation curves for plasticized unvulcanized and vulcanized EPDM.

portion, represented by A, decreases simultaneously as C increases in value. The vulcanization process at high temperatures causes the liberation of small molecules (organic peroxide byproducts, essentially) giving rise to a shift downward in the C value of about 2 to 3%. This shift downward of the baseline of the relaxation curve is also observed in postcured samples because of the loss of volatiles and small organic molecules.3,24 TABLE II TRIEXPONENTIAL CURVE FITTING OF THE NMR RELAXATION RESULTS SHOWN IN FIGURE 2

Plasticizer phr

vol%

Unvulcanized 0 0 5 3.5 10 6.7 15 9.8 20 12.6 30 17.8 Vulcanized 0 0 5 3.5 10 6.7 15 9.8 20 12.6 30 17.8

A, %

B, %

C, %

T21, ms

T22, ms

T23, ms

Me or Mc, g/mol

79 75 71 67 64 60

18 19 19 20 20 20

3 6 10 13 16 20

1.15 1.22 1.29 1.35 1.40 1.52

5.7 6.1 6.3 6.5 6.7 7.2

32 31 30 31 31 32

1930 2050 2160 2260 2350 2550

89 86 81 77 73 67

10 11 12 13 14 15

1 3 7 10 13 18

0.91 0.98 1.08 1.13 1.19 1.31

7.4 8.1 8.0 8.0 7.9 7.9

67 36 31 30 29 30

1530 1640 1810 1900 2000 2200

LOW FIELD 1H NMR

301

FIG. 3. — Average molar mass between entanglements/cross-links for plasticized unvulcanized and vulcanized EPDM compounds.

The Me and Mc values calculated in Table II were plotted on Figure 3 versus the plasticizer concentration. Both of these values increase linearly with plasticizer addition. The similar slope observed between the unvulcanized and vulcanized compounds is good evidence that the amount of chemical cross-linking is constant as a function of plasticization. Assuming that the total entanglement density measured by NMR is due to a combination of entangled polymer chains and casual or temporary contacts between adjacent chains, it is likely that plasticizer addition, in this case, the paraffinic hydrocarbon molecules, insert themselves between the long polymer chains increasing the free volume and hindering some of the temporary contacts between adjacent chains. The additional lubricity provided by these highly mobile molecules could also conceivably help to disentangle some of the polymer chains. It is likely that both of these behaviors are operating at the molecular level upon addition of the plasticizer. The long decay time representing the sol fraction is remarkably constant around 30 ms for the unvulcanized samples. In the case of the vulcanized samples, that value is achieved for higher plasticizer loadings. The lower plasticizer levels, in particular, 0 to 5 phr, are better modeled using a biexponential fitting function, given the error in the calculated C value (.1%). Similar values of Me are obtained using both multiexponential fitting functions. In addition, it appears that an acceptable deconvolution of the A, B, and C contributions to the overall magnetization decay is possible; however, some overlap of each contribution during the curve fitting procedure is still possible, causing some skewing of fitted results. Because polymer chain entanglement density has a major effect on rubber processability, the values of Me derived from NMR relaxation measurements were plotted against the storage modulus and tan d data from Figure 1a,b. The resulting graphs are presented in Figure 4a,b. Excellent linear correlations are observed between Me and these dynamic quantities. Plasticization increases Me values, which, consequently, causes the storage modulus to decrease and tan d to increase. Obviously, a higher Me value is indicative of a system possessing less elastic behavior, which provides, in the unvulcanized state, a material possessing enhanced (more viscous) processing characteristics. The excellent correlation between dynamic properties and Me, calculated through

302

RUBBER CHEMISTRY AND TECHNOLOGY, Vol. 85, No. 2, pp. 295–312 (2012)

FIG. 4. — Correlation of Me to (a) storage modulus, and (b) tan d for plasticized unvulcanized EPDM.

the HE NMR relaxation, provides confidence in the latter method for discerning plasticizing effects in rubber compounds. CARBON BLACK EFFECTS IN EPDM

For this investigation, the virgin EPDM polymer and a mixed-EPDM sample were both included for direct comparison with the six compounded EPDMs of varying carbon black loadings. As in the plasticizer study, frequency sweeps at low (7%) and high (100%) strains were carried out and are plotted in terms of modulus and tan d in Figure 5a,b. The storage modulus is observed to

LOW FIELD 1H NMR

303

FIG. 5. — Frequency sweeps at 125 8C and 7 and 100% strains for (a) storage modulus, and (b) tan d at various carbon black loadings in EPDM.

increase as a function of frequency, and a regular progression of higher modulus can be seen in going from unfilled to the highest carbon black loading. It is well known that favorable interactions between the polymer chains and the carbon black surface are responsible for the increase in moduli values and the overall reinforcement effect. Very little difference, however, was observed between the EPDM, mixed EPDM, and 0 phr N550 EPDM compound. The viscoelastic behavior depicted by tan d appears to show more differentiation between the samples, in particular, at low frequencies and low strains. In Figure 5b, the mixing of EPDM alone causes tan d to increase slightly. Addition

304

RUBBER CHEMISTRY AND TECHNOLOGY, Vol. 85, No. 2, pp. 295–312 (2012)

FIG. 6. — Selected NMR relaxation curves for unvulcanized and vulcanized EPDM at various carbon black loadings.

of the peroxide, processing aid, and antioxidant causes tan d to slowly decrease, but on addition of carbon black, tan d slowly increases, then decreases. At a higher frequency, all tan d values are similar, with a slight trend of higher tan d with increasing carbon black concentration. Generally higher tan d values are seen at higher strains when the carbon black content is increased, but there appears to be more variation in the results. The tan d data for these series of samples appear to suggest that the mixing process may affect the molecular characteristics of the EPDM polymer in the form of chain breakage due to the shearing action during the mixing process. In Figure 6, selected spin–spin relaxation curves are plotted for both the unvulcanized and vulcanized EPDMs compounds. It is immediately apparent that the curves can be separated into two distinct populations, based on whether vulcanization took place or not. The virgin and mixedEPDM samples both belong to the unvulcanized series of curves. Vulcanization causes the initial relaxation to fall at a quicker rate (because of the addition of chemical cross-links) and the baseline relaxation zone to decrease. This was also readily apparent in the plasticizer study in Figure 2. Tables III and IV summarize the biexponential and triexponential curve fitting results for unvulcanized and vulcanized EPDM, respectively. According to Table III, the unvulcanized EPDM relaxation curves are best modeled using a triexponential fitting function because of the nonnegligible C contributions of up to 5%. This fitting function will more accurately describe the longtail chain behavior due to the presence of small sol-like molecules in the unvulcanized matrix. Long decay-time relaxation values given by T23 for higher carbon black loadings are similar to those values obtained in the plasticizer investigation. The Me values in the range of 1900 to 2100 g/mol are in good agreement with reported literature values for NMR relaxation for EPDM.6,8 On the other hand, the vulcanized EPDM relaxation curves are best fitted using the biexponential function because little difference in values is observed in adding an extended exponential tail to the fitting process. The T23 values also show higher variation (and error) because of the low relative amount of highly mobile concentration (C < 1%). As seen in the case of the plasticizer study, the baseline shift observed in going from unvulcanized to vulcanized samples (C

LOW FIELD 1H NMR

305

TABLE III CURVE-FITTING RESULTS FOR THE UNVULCANIZED EPDM COMPOUNDS

Carbon black A, %

B, %

C, %

EPDM Mixed EPDM 0 0 15 6.4 30 12.0 45 17.1 60 21.6 75 25.6

84 84 83 82 82 82 81 81

15 15 16 17 17 17 18 18

1 1 1 1 1 1 1 1

EPDM Mixed EPDM 0 0 15 6.4 30 12.0 45 17.1 60 21.6 75 25.6

82 82 79 79 79 78 78 78

15 15 16 17 17 18 18 19

3 3 5 4 4 4 4 3

phr

vol%

T21, ms

T22, ms

Biexponential 1.20 8.5 1.20 8.5 1.29 11.3 1.27 10.5 1.24 9.6 1.22 8.9 1.22 8.5 1.20 8.0 Triexponential 1.17 6.2 1.17 6.2 1.23 6.5 1.23 6.5 1.20 6.2 1.18 6.0 1.17 5.7 1.16 5.6

T23, ms

Me, g/mol

— — — — — — — —

2010 2010 2160 2140 2090 2050 2040 2010

44 44 50 43 39 36 33 33

1970 1960 2070 2060 2020 1980 1970 1950

values) is likely due to the removal of highly mobile (and volatile) molecules during and after completion of the molding process. In both unvulcanized and vulcanized EPDM compounds, a slight downward shift is observed in both the magnitudes of A and C, whereas B increases upon addition of carbon black. These results suggest that the dangling chain contribution to the total magnetization decay increases at the expense of both the rigid and sol phases. Selected, unvulcanized EPDM samples were analyzed to characterize both molecular weight and molecular weight distribution by GPC (see Table V) to verify any differences from effects caused by the internal mixing process. The most striking result is the decrease in Mz (the proportion of very long EPDM chains) upon mixing with or without carbon black. This agrees with the increased tan d observed in the low-frequency zone of Figure 5b. Afterward, the molecular weight characteristics appear to be quite constant through the mixing process, independent of the carbon black content. Chain degradation in EPDM can take place during the mixing process, lowering the molecular weight and narrowing the molecular weight distribution.25 Our results support that observation. The longer EPDM chains are most sensitive to the shearing effects, and a certain fraction of them break to form shorter ones, while causing the overall polydispersity to narrow. Based on these results, the increase in the amount of dangling chain ends, as seen by increasing the carbon black concentration, cannot be wholly explained by molecular weight data. Figure 7 displays, in graphical form, the values of Me and Mc that were calculated for the unvulcanized (including the mixed and virgin EPDM) and vulcanized EPDMs using the triexponential and biexponential functions, respectively. No significant change is seen for Me between EPDM and mixed EPDM, which suggests that the mixing process does not significantly alter the density of chain entanglements in the absence of any other additives or fillers. Even though

306

RUBBER CHEMISTRY AND TECHNOLOGY, Vol. 85, No. 2, pp. 295–312 (2012) TABLE IV CURVE-FITTING RESULTS FOR THE VULCANIZED EPDM COMPOUNDS

Carbon black phr

vol%

Biexponential 0 0 15 6.4 30 12.0 45 17.1 60 21.6 75 25.6 Triexponential 0 0 15 6.4 30 12.0 45 17.1 60 21.6 75 25.6

A, %

B, %

C, %

T21, ms

T22, ms

T23, ms

Mc, g/mol

92 92 92 91 90 90

7 8 8 9 10 10

1 0 0 0 0 0

0.85 0.87 0.88 0.91 0.91 0.91

9.4 9.3 8.9 8.4 8.0 7.6

— — — — — —

1430 1470 1490 1520 1520 1530

92 92 91 90 90 90

7 7 8 9 9 10

1 1 1 1 1 0

0.85 0.87 0.88 0.90 0.90 0.91

7.2 7.6 7.7 7.6 7.2 7.2

59 56 57 64 58 86

1420 1460 1480 1510 1520 1530

a certain proportion of long chains are broken during mixing, the chains are still long enough to entangle, ensuring a constant physical-entanglement density. The jump in value from 1960 g/mol (mixed EPDM) to 2070 g/mol (EPDM, 0 phr carbon black) is most likely due to a small plasticizing effect by adding 4 phr of small organic molecules (processing aid, antioxidant, and peroxide) to the recipe. These molecules will insert themselves between the polymer chains and provide extra lubrication and mobility for the chains. As carbon black is progressively added into EPDM, the Me value now decreases because of the additional physical interactions between polymer chains and the carbon black surface. As shown by the graph, this interactive effect is additive and linear as carbon black loading is increased. In addition to the physical chain entanglements, vulcanizing the compounds by way of thermally initiated peroxide decomposition adds chemical cross-links in the form of carbon–carbon linkages, thus lowering the overall average molar mass between constraints. The Mc value contains contributions from Me (which now contains a fraction of trapped entanglements due to chemical cross-linking) and from any remaining interactions between the polymer chains and the active carbon black. Upon vulcanization, Mc values increase slightly as a function of carbon black loading. TABLE V GPC RESULTS OF SELECTED UNVULCANIZED EPDM SAMPLES

Carbon black phr

vol%

EPDM Mixed EPDM 15 6.4 45 17.1 75 25.6

Mn, kg/mol

Mw/10, kg/mol

Mz/10, kg/mol

Mw/Mn

63 66 61 62 61

28 25 25 25 24

101 67 66 66 66

4.4 3.8 4.1 4.0 3.9

LOW FIELD 1H NMR

307

FIG. 7. — Effect of carbon black loading on average molar mass between entanglements (Me) and total cross-links (Mc).

This is opposite to the effect that was observed with the plasticization study. The increase of the Mc values correlates directly with carbon black loading, so it is likely that its surface chemistry inhibits the peroxide curing mechanism to a small degree causing less chemical cross-links to be formed. The presence of acidic groups can consume part of the active peroxide concentration through a nonproductive acid catalyzed reaction.26 Also, the occurrence of carbon black flocculation during the vulcanization process cannot be ruled out.27 The presence of this phenomenon would increase the values of Mc, but because the NMR relaxation cannot discriminate between chemical and physical cross-links, the extent of filler flocculation is unknown. Owing to the addition of reinforcing agents, such as carbon black, an additional series of interactions taking place between the filler’s surface and polymer chains is expected. Using, as a baseline, the value of Me for the compound (Me,nf ), which contains no carbon black but has the effect of adding small molecules, and assuming that the physical entanglements and polymer chain– filler interactions are decoupled and additive, it is possible to estimate a perceived average molar mass (Mchain–filler) due to the latter contribution by way of Eq. 3: 1 1 1 »  ð3Þ Mchainfiller Me Me;nf Conversion of Mchain–filler to a density value was envisioned but is problematic because of the numerous types of interactions that are possible between the backbone chain and the free chain ends with the black surface. Figure 8 displays the results of applying Eq. 3 to the unvulcanized Me data in Figure 7. The value of the Mchain–filler decreases as the volume fraction of carbon black increases, indicating that more interactions are taking place between the polymer chain and carbon black. The nature of the curve makes sense because the Mchain–filler is expected to tend toward very high values (infinity) as the carbon black concentration reaches 0%. The total entanglement density is also plotted to compare with the magnitude of the Mchain–filler values. That comparison strongly suggests that the contribution of the polymer chain entanglement is still the most important source of the physical cross-links, compared with the interactions between the polymer and carbon black. An earlier study on EPDM also echoes the same observation.22

308

RUBBER CHEMISTRY AND TECHNOLOGY, Vol. 85, No. 2, pp. 295–312 (2012)

FIG. 8. — Estimation of the polymer chain/carbon black interaction in terms of Mchain–filler versus carbon black loading.

In Figure 9, the compression-set resistance, measured after heat aging 168 hours at 150 8C, is plotted against carbon loading and is compared directly with Mc before and after aging. Calculated values of Mc after heat aging are also tabulated in Table VI. Because of the low baseline for heataged samples, a biexponential fit is satisfactory for describing the molecular behavior (i.e., C¼0%). Both compression set and Mc values increase in a fairly linear fashion as carbon black is added into the system, illustrating that there is a good correlation between them. All values of Mc decrease by about 50 g/mol on average after the heat-aging process for each carbon black loading. Oxidative heat aging of EPDM brings about additional cross-linking in EPDM, causing the hardness to rise while elongation is lost.28 Comparing amplitude values (B%) between Tables VI and IV

FIG. 9. — Compression-set resistance (168 hours/150 8C) and average molar mass before and after heat aging.

LOW FIELD 1H NMR

309

TABLE VI BIEXPONENTIAL CURVE-FITTING RESULTS FOR HEAT-AGED (168 HOURS/150 8C) EPDM SAMPLES

Carbon black phr 0 15 30 45 60 75

vol%

A, %

B, %

T21, ms

T22, ms

Mc, g/mol

0 6.4 12.0 17.1 21.6 25.6

92 92 91 90 90 89

8 8 9 10 10 11

0.82 0.84 0.86 0.87 0.88 0.88

7.8 7.8 7.6 7.2 6.8 6.4

1370 1410 1450 1460 1470 1480

(biexponential fit), the free chain-end contribution increases by about 1%, suggesting that besides cross-linking, some chain scission may be taking place as well. Figure 10a,b attempts to correlate the calculated Me values (which include polymer chain to carbon black interactions) to the storage modulus and tan d, respectively, both measured as a function of frequency and strain (7 or 100%) for the unvulcanized EPDM samples. The general trend noticed is the decrease in modulus as Me increases in value. Assuming physical entanglement density is relatively constant in all samples, the most important factor causing this trend is the additional interactions occurring between the polymer chains and the filler particles (i.e., lower carbon black loading leads to lower storage modulus). It is also manifest that this trend is linear with the best curve fitting occurring at higher frequencies of measurement. Parallel to the prior plastification investigation, these observations strongly suggest that the HE NMR technique can be used to measure quantitatively the effect of carbon black loading in rubber, given the good correlation to dynamic properties in the unvulcanized state. A morphological model of the levels of chain mobility that have been characterized in rubber using NMR has been simplified from ref 9 in Figure 11. Three regions of differing molecular mobility as measured by NMR are generally accepted: 1. 2. 3.

Tightly bound rubber – very limited motion of chains (chains absorbed on surface) Loosely bound rubber including connective filaments (chains possess an intermediate mobility) Unbound rubber (bulk elastomer mobility)

Given the timescale used for the NMR relaxation collection in this investigation (.0.08 ms), the information gathered in Tables III and IV pertain predominantly to the unbound rubber phase and its changes when carbon black is added to the system. It appears that the bulk mobile phase is being affected or perturbed by the presence of carbon black, resulting in additional constraints of the polymer chains being detected through NMR relaxation. It is generally assumed that the longer polymer chains are attracted to the disordered carbon black surface.9 Free chain ends, consequently, do not appear to be directly absorbed on the surface and are expelled back toward the loosely bound and unbound rubber phases. This phenomenon will automatically decrease values of the rigid portion while increasing the total number of free chain ends (see Table III). In the presence of carbon black at different loadings, the molecular weight of the EPDM was shown to be constant by GPC. Unless additional constraints (physical or chemical cross-links) are being developed in the interface region between the mobile bulk chains and the intermediate zone, the Mchain–filler parameter appears to relate directly to chains that are absorbed directly on the carbon black surface and their effect on the bulk mobile phase. Further investigation to examine these interactions and the nature of the

310

RUBBER CHEMISTRY AND TECHNOLOGY, Vol. 85, No. 2, pp. 295–312 (2012)

FIG. 10. — Correlation of Me to (a) storage modulus, and (b) tan d for unvulcanized EPDM of varying carbon black content.

interface by low field NMR in EPDM and other elastomer systems is solicited. Promising work in examining the nature of filler/elastomer interactions has recently been published by using a combination of MQ NMR and equilibrium volume swell techniques.29,30 CONCLUSIONS The effect of plasticizer and carbon black addition in EPDM compounds on the NMR spin– spin relaxation signal has been elucidated. Either biexponential or triexponential curve fitting was necessary to successfully model the relaxation response. It was found that the degree of

LOW FIELD 1H NMR

311

FIG. 11. — Simple two-dimensional model depicting the morphological zones in presence of carbon black aggregates (not to scale) in the unvulcanized state. The zones are described as tightly bound (immobilized) rubber (dark gray), loosely bound rubber (light gray), and remaining area (white) extractable rubber. Filamentlike bridges composed of extended EPDM chains (light gray) are believed to connect carbon black aggregates together. A magnified area showing the arrangement of the polymer chains within the three zones is shown to the right.

plasticization is directly correlated with the Me calculated from the short decay constant T21. Me correlates very well with the dynamic properties in the unvulcanized state, signifying the potential usefulness of HE NMR relaxation to measure processability characteristics. An increase in carbon black concentration without the presence of a plasticizer brought about a decrease in Me. The lowering of the Me value is likely due to increased interactions between the filler and the polymer chain. A new parameter Mchain–filler is introduced to estimate the average molar mass between polymer chains and fillers. Compared with the chain entanglement density, the overall magnitude of this interaction appears to be weak in the mobile zone of the compound matrix. As in the case of plasticization, a relatively good correlation is obtained between Me and dynamical properties in the unvulcanized state. Compression-set resistance is shown to directly follow Mc before and after heat aging. The carbon black investigative results can be readily understood within the context of a morphological model containing different zones of chain mobility—a thin shell of immobilized chains, an intermediate zone of limited mobility, and a dominant mobile phase consisting mainly of the polymer chains. ACKNOWLEDGEMENTS The authors would like to thank FNGP for permission to publish this article. Jordi Labs is also acknowledged for providing the GPC data and analysis.

REFERENCES 1

M. A. Fath and C. P. Radar, ‘‘Vulcanization,’’ in Elastomer Technology: Special Topics, 1st ed., K. Baranwal and H. Stephens, Eds., Rubber Division, ACS, Akron, 2003, 204.

2

Modern NMR Techniques for Chemistry Research, A. E. Derome, Ed., Pergamon Press, Oxford, U.K., 1987, Vol. 6.

312 3

RUBBER CHEMISTRY AND TECHNOLOGY, Vol. 85, No. 2, pp. 295–312 (2012)

M. Viol, K. Beck, and A. Voda, ‘‘Possibilities of the Use of Nuclear Magnetic Resonance Crosslink Density Imaging and Spectroscopy,’’ Paper 2004-010870, presented at the SAE World Congress, Detroit, March 8–11, 2004.

4

K. Beck, A. Voda, M. Bescher, V. Peterseim, and M. Viol, Kautsch. Gummi Kunstst, 60, 100 (2007).

5

K. Saalw¨achter, Prog. Nucl. Magn. Reson. Spec. 51, 1 (2007).

6

V. M. Litvinov, W. Bararendswaard, and M. van Duin, RUBBER CHEM. TECHNOL. 71, 105 (1997).

7

R. A. Orza, P. C. M. M. Magusin, V. M. Litvinov, M. van Duin, and M. A. J. Michels, Macromol. Symp. 230, 144 (2005).

8

R. J. Pazur and F. J. Walker, ‘‘State of Cure Measurements in Peroxide and Sulfur-Cured EPDM,’’ Paper 49, presented at the 174th Fall Meeting of the Rubber Division, ACS, Louisville, KY, October 14–16, 2008.

9

J. L. Leblanc, Prog. Polym. Sci. 27, 627 (2002).

10

J. C. Kenny, V. J. McBrierty, Z. Rigbi, and D. C. Douglass, Macromolecules 24, 436 (1991).

11

U. Heuert, M. Kn¨orgen, H. Menge, G. Scheler, and H. Schneider, Polym. Bull. 37, 489 (1996).

12

J. Leisen, J. Breidt, and J. Kelm, RUBBER CHEM. TECHNOL. 72, 1 (1999).

13

U. Hoffmann, W. Gronski, G. Simon, and A. Wutzler, Angew. Makromol. Chem. 202/203, 283 (1992).

14

S. Kaufman, W. P. Slichter, and D. D. Davies, J. Polym. Sci. A-2 9, 829 (1971).

15

J. O’Brien, E. Cashell, G. E. Wardell, and V. J. McBrierty, Macromolecules 9, 653 (1976).

16

G. Simon, B. G¨otschmann, D. Matzen, and H. Schneider, Polym. Bull. 21, 475 (1989).

17

H. L¨uchow, E. Breier, and W. Gronski, RUBBER CHEM. TECHNOL. 70, 747 (1997).

18

F. Yatsuyanagi, H. Kaidou, and M. Ito, RUBBER CHEM. TECHNOL. 72, 657 (1999).

19

H. Luo, M. Kl¨uppel, and H. Schneider, Macromolecules 37, 8000 (2004).

20

M. Gussoni, F. Greco, M. Mapelli, A. Vezzoli, E. Ranucci, P. Ferruti, and L. Zetta, Macromolecules 35, 1722 (2002).

21

J. Berriot, F. Martin, H. Montes, L. Monnerie, and P. Sotta, Polymer 44, 1437 (2003).

22

V. M. Litvinov and P. A. M. Steeman, Macromolecules 32, 8476 (1999).

23

Internal FNGP control data.

24

R. J. Pazur and F. J. Walker, ‘‘Practical State of Cure Measurements by NMR,’’ Paper 45, presented at the 176th Fall Meeting of the Rubber Division, ACS, Pittsburgh, October 13–15, 2009.

25

K. Baranwal and H. L. Jacobs, J. Appl. Poly. Sci. 13, 797 (1969).

26

P. R. Dluzneski, Rubber World 224, 34 (2001).

27

G. G. A. B¨ohm and M. N. Nguyen, J. Appl. Poly. Sci. 55, 1041 (1995).

28

R. C. Keller, ‘‘General Purpose Elastomers—D Ethylene-Propylene Rubber,’’ in Basic Elastomer Technology, 1st ed., K. C. Baranwal and H. L. Stephens, Eds., Rubber Division, ACS, Akron, 2001, 287.

29

I. Mora-Barrantes, J. L. Valentin, P. Posadas, A. Marcos, A. Rodriguez, L. Ibarra, and L. Gonzalez, ‘‘Reinforcement Effect of Novel Modified Fumed Silica in Styrene–Butadiene Rubbers,’’ Paper 96, presented at the 176th Rubber Division Meeting, ACS, Pittsburgh, October 13–15, 2009.

30

J. L. Valentin, I. Mora-Barrantes, J. Carretero-Gonzalez, M. A. Lopez-Manchado, P. Sotta, D. R. Long, and K. Saalw¨achter, Macromolecules 43, 334 (2010). [Paper 90, presented at the Fall 178th Technical Meeting of the Rubber Division, ACS (Milwaukee) 12–14 October 2010] [Received August 2011, Revised November 2011]