The glass transition, segmental relaxations and viscoelastic behaviour of particulate-reinforced natural rubber

European Polymer Journal 67 (2015) 232–241

Contents lists available at ScienceDirect

European Polymer Journal journal homepage: www.elsevier.com/locate/europolj

The glass transition, segmental relaxations and viscoelastic behaviour of particulate-reinforced natural rubber Menglong Huang 1, Lewis B. Tunnicliffe 1, Alan G. Thomas, James J.C. Busfield ⇑ Materials Research Institute, Queen Mary University of London, Mile End Road, London E1 4NS, United Kingdom

a r t i c l e

i n f o

Article history: Received 1 September 2014 Received in revised form 14 March 2015 Accepted 17 March 2015 Available online 24 March 2015 Keywords: Glass transition Elastomer Rubber Reinforcement Carbon black Precipitated silica

a b s t r a c t The role of filler particles in defining local changes to the dynamics of elastomer polymers – specifically in relation to the reinforcement of commercial elastomer systems – is as yet incompletely resolved. This work examines the glass transition of filled, crosslinked natural rubber in relation to filler reinforcement mechanisms using a variety of complimentary experimental techniques. Carbon black and precipitated silica filler particles with a wide range of surface areas, structures and surface activities are used as the reinforcing phase. No effect of the filler particles on the calorimetric glass transition is observed in terms of shifting, broadening or transition strength. Dielectric spectroscopy measurements show that the polymer relaxation times of the filled rubbers in the glass-to-rubber transition zone remain essentially equivalent to that of the corresponding unfilled material. Dynamic mechanical analysis demonstrates that the storage moduli of the filled elastomers are significantly amplified on the rubbery side of the glass transition. Elastic stiffening mechanisms are discussed in the context of contributions from filler networking. The reason for a distinct lack of evidence for filler-induced polymer modification may be due to the nature of polymer confinement associated with the imperfect dispersion state of the fillers in the sample and the fact that the fillers themselves are formed from aggregated primary particles rather than being true nanoparticulates. Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction Crosslinked rubbers (elastomers) reinforced with carbon blacks and precipitated silicas are a widely used and well established type of high performance composite material. However the precise mechanisms of filler reinforcement of elastomers remain incompletely understood despite the large scale utilisation of filled elastomers for a wide range of safety–critical engineering applications. A particularly industrially important aspect of reinforcement

⇑ Corresponding author. Tel.: +44 (0)20 7882 8866. E-mail addresses: [email protected] (M. Huang), l.tunnicliffe @qmul.ac.uk (L.B. Tunnicliffe), [email protected] (A.G. Thomas), j.busfi[email protected] (J.J.C. Busfield). 1 These authors contributed equally to this work. http://dx.doi.org/10.1016/j.eurpolymj.2015.03.024 0014-3057/Ó 2015 Elsevier Ltd. All rights reserved.

is the modification of the strain-dependent viscoelastic spectrum of the elastomers imparted by the presence of the filler phase. For example, in passenger car tyres the performance factors such as wet grip and rolling resistance (contributing to total fuel efficiency) are intimately linked to the viscoelastic properties of the tyre tread material. Some of these modifications can be understood in terms of the structural dynamics of dissipative yielding and stress softening of fractal filler structures under increasing dynamic loadings; but even at small, linear viscoelastic strains where such yielding does not take place, the modification of the viscoelastic spectrum cannot be fully explained in an analytical sense by hydrodynamic, rigid volume reinforcement concepts alone [1–3]. To address this problem, a number of authors have proposed and/or demonstrated that the local segmental

M. Huang et al. / European Polymer Journal 67 (2015) 232–241

dynamics of certain rubber polymers in the vicinity of a range of filler particles can be substantially altered versus the corresponding bulk polymer material [4–6]. Furthermore gradients of chain mobility proceeding from the polymer in direct contact with the filler surface to bulk behaviour have been reported as well as ‘bridges’ of glassy polymer between filler aggregates where confinement of the polymeric macromolecules is particularly extreme. Some authors have proposed a theoretical model for these effects based around the percolation of small, localised density fluctuations within the polymer phase [7]. It has also been proposed that the existence of a broadened glass transition can account for a range of phenomena associated with particulate reinforcement such as the Payne and Mullins effects [8]. Conversely there are a large number of studies that report no significant effect of the presence of filler on the local segmental dynamics of the polymer and the bulk glass transition. This disagreement within the literature (up to 2008) has been highlighted by the topical review of Robertson and Roland [9]. While a considerable amount of work has been published on the behaviour of thin polymer films [10,11] with some reports of equivalence between confinement effects in thin films and bulk nanocomposite properties [12], no universal conclusions have yet been reached. Table 1 highlights the findings of a brief, and in no way exhaustive, selection of historical and more contemporary investigations which examined such glass transition effects specifically for the case of filled elastomers. Some of the studies described in Table 1 were performed on conventional, commercial filled elastomers, others were performed on elastomers filled with esoteric, non-conventional fillers and a number were conducted on model nanocomposite systems comprising nanoscale particulates of near perfect dispersion within an elastomer matrix. Such a wide variety of sample types may have led to a range of different conclusions and interpretations as to the effect of filler particles on the polymer glass transition. This work evaluates a series of silica and carbon blackfilled natural rubbers using Differential Scanning Calorimetry (DSC), Broadband Dielectric Spectroscopy (BDS) and Dynamic Mechanical Analysis (DMA) techniques in an attempt to determine if the segmental dynamics and glass transition of the polymer in proximity to the filler surface are significantly modified and to what extent this can be correlated with the viscoelastic properties of the filled elastomers.

2. Experimental section 2.1. Specimen preparation 2.1.1. Carbon black-filled specimens The preparation and detailed characterisation of the carbon black-filled samples has been reported in a separate publication [41]. A series of commercial carbon blacks were used, along with ‘graphitised’, thermally deactivated counterparts which display a significantly reduced interaction with the polymer phase. Key characteristics of the fillers are reproduced in Table 2 from Tunnicliffe et al. [41].

233

The Natural Rubber (NR) used was SMR CV60 grade obtained from the Tun Abdul Razak Research Centre (TARRC), Hertfordshire, UK. All carbon blacks were compounded with NR in a Banbury internal mixer. Mass loadings of fillers were adjusted to maintain equal volume fractions of 0.20; equivalent to 50 phr (parts per hundred rubber) of unmodified carbon black. Subsequently 2 phr of dicumyl peroxide (Fisher Scientific) was added on a laboratory 2-roll-mill. Samples were cured as 2 mm thick flat sheets using a compression mould and hot-press. The curing temperature was 150 °C and the curing time was 100 min – which was determined via an Alpha Technologies MD2000 rheometer to be the time required for near total decomposition and reaction of the peroxide. Subsequent analyses were performed with samples cut from the cured sheets. 2.1.2. Precipitated silica-filled specimens Commercial precipitated silicas were investigated, together with tri-ethoxyvinyl silane (TEVS) modified counterparts. Conventional (CS) and highly dispersible (HDS) grades of silica were used. For the conventional silica the surface modification was pre-reacted with the silica by the supplier. For the highly dispersible silica the surface modification was performed in situ during compounding. All silicas were compounded with NR on a laboratory 2roll-mill with the exception of the HDS with TEVS surface modification. This was prepared in a Banbury internal mixer in order for the mix to reach a sufficiently high temperature to promote reaction of TEVS with the silica surface. TEVS (Fisher Scientific) was added to the HDS on a nitrogen surface area basis to obtain a theoretically equivalent fractional surface coverage with the TEVS-pretreated CS. Sample mass loadings were adjusted to maintain equivalent volume fractions with the carbon black-loaded specimens (0.20). Further specimen processing was identical to that detailed for carbon black-filled specimens. Key characteristics of the fillers are reproduced in Table 2. 2.1.3. Specimens for BDS For BDS, samples having a thickness of roughly 0.7 mm were made by the same compression moulding method. Samples were cut into 25 mm  25 mm  0.7 mm square sheets then gold electrodes were coated on both sides of each specimen by evaporation deposition. The diameter of the concentric gold circle electrodes was 22 mm. Due to the carbon black’s inherent conductivity which, when the filler is loaded at significant volumes both below and above the electrical percolation threshold, can mask the polymer chain segmental relaxation process, the carbon black series of samples were not extensively investigated using BDS. However as an example, one carbon black sample (N330) was prepared at a volume fraction (0.05) well below the filler electrical percolation threshold [41] and subjected to BDS testing. 2.2. Experimental techniques 2.2.1. Differential scanning calorimetry Calorimetry measurements were made using a Perkin Elmer Pyris Diamond DSC. Small samples (10 mg) were

234

M. Huang et al. / European Polymer Journal 67 (2015) 232–241

Table 1 Overview of a number of relevant publications. Author

Year

Technique(s)

Polymer/filler

Conclusion

Mason [13] Smit [14] Waldrop and Kraus [15] Kraus and Gruver [16] Kaufman, Slichter and Davis [17] Kraus, Rollmann and Gruver [18] O’Brien et al. [19] Struik [4] Stacer and Husband [20] Tsagaroupoulos and Eisenberg [21] Wang [5] Berriot et al. [6]

1960 1966 1969

Dilatometry Mechanical 1 H NMR

NR/CB SBR/CB SBR/CB

1970

Thermal expansion

SBR/CBs

No change in Tg Absorbed, modified polymer layer. Thickness >2 nm No effect of filler on polymer segmental motion even for ‘bound’ rubber fraction Little effect of carbon black on the Tg

1971

1

BR, EPDM/CB

1971

Mechanical

SBR/PS

Immobilised and constrained regions of polymer detected in ‘bound’ rubber fraction Modifications of dynamic properties are not associated with Tg effects

1976 1987 1990

1

H NMR Mechanical Mechanical

BR/CB SBR/CB Various/alumina

Evidence of a glassy layer of polymer surrounding filler Evidence of extended Tg in filled elastomers Interphase relaxation behaviour reported

1995

DMA

PVAC, PMMA, PS, P4VP/silica

Interfacial layer of polymer with restricted mobility inferred from temperature dependence of loss tangent

1998 2002

1

Mechanical H NMR

SBR/CB PEA/silica

Fragiadakis et al. [22] Fragiadakis et al. [23] Amanuel et al. [24] Bogoslovov et al. [25] Meier and Klüppel [26] Robertson et al. [27] Ding et al. [28] Robertson et al. [29]

2005

DSC, TSDC and BDS

PDMS/silica

2006

BDS

PDMS/silica

2008

DSC

PS/silica

Gradient of Tg around filler particles Identified gradient of Tg around filler particles in model, filled elastomers Interfacial layer of polymer of 2.1–2.4 nm thickness with restricted mobility Restricted interfacial layer of polymer of 3 nm thickness. Layer decreases in thickness with increasing temperature No alteration or broadening of Tg

2008

PVAC/silica

No effect of filler on Tg or polymer segmental dynamics

2008

DSC, rheometry, BDS, dilatometry DMA and DBS

Various elastomers/CB

Immobilisation of polymer in gaps between filler aggregates

2008

Rheometry, AFM

SBR/Silica

No effect of filler on dynamic loss modulus around Tg

2009 2009

Polyisoprene/C60 PVAC/silica

C60 slows both segmental and chain relaxation processes No change in temperature dependence of segmental relaxation

No evidence of glassy or non-bulk behaviour in thin polyisoprene droplets Restricted interfacial layer detected for homogenously dispersed nanoparticles. No modification of dynamics detected for aggregated fillers Immobilisation of polymer between filler aggregates subject to temperature dependence Concludes that ‘second glass transition’ of Tsagaroupoulos and Eisenberg [21] is related to polymer terminal flow rather than Tg No effect of commercial fillers on the polymer segmental dynamics

H NMR

Göritz et al. [30]

2010

BDS DMA, DSC, BDS, Pressure dependence AFM

Fragiadakis et al. [31]

2011

DSC, TSDC and BDS

Polyisoprene droplets on graphitic substrate NR/silica

Fritzche and Klüppel [32] Robertson et al. [33] Robertson et al. [34] Vo et al. [35]

2011

DMA and DBS

SBR/CB

2011

Rheometry

BR/CB

2011

Rheometry, DSC and AFM BDS

SBR/silica

Papon et al. [36] Hernández et al. [37] Kummali et al. [38] Wu et al. [39] Mujtaba et al. [40]

2011 2012 2010– 2012 2013 2013 2014

1

H NMR and DSC BDS

Electric Force Microscopy BDS 1

H NMR, DMA

SBR/CB, silica, nanoclay PEA/Silica NR/graphene, silica, vulcanizing additives SBR/silica VPR/graphene oxide (GO) SBR/silica

cut from cured sheets and placed in aluminium crucibles. Samples were cooled to 193 K and then heated at a rate of 10 K min1 to 238 K under a flushing nitrogen atmosphere. Heat flow traces were weight-normalised to the weight of elastomer in the samples (as opposed to total sample weight).

Higher Tg for interfacial layer. Up to 80 K higher than bulk depending on the filler Gradient of Tg around filler particles in model, filled elastomers No evidence of a shift of Tg by introducing nanofiller Equivalent segmental relaxation dynamics between interfacial layer and bulk polymer Segmental dynamics depends on interfacial bonding and GO volume fraction Small volumes of immobilised polymer detected varying as a function of temperature

2.2.2. Broadband dielectric spectroscopy BDS was performed on a Novocontrol Concept 40 dielectric spectrometer. All coated samples were dried in a vacuum oven at 100 °C for 24 h prior to testing. Experiments were performed over a frequency range from 0.1 Hz to 20 MHz and a temperature range from 100 °C to

M. Huang et al. / European Polymer Journal 67 (2015) 232–241

235

Table 2 Filler suppliers, surface areas and densities.

  à

Filler code (Trade Name)

Supplier

Nitrogen-specific surface area/m2 g1

Density/ g cm3

N134 N134g N330 N330g N990 N990g HDS (Zeosil 1165) HDS-TEVS (Zeosil 1165 + TEVS) CS (VN2) CS-TEVS (Coupsil 6508)

Cabot Modified Cabot Modified Cancarb Modified Solvay Modified

134 136 77 78 8 8 165 165

1.80  1.96 1.80  1.90 1.80  1.95 2.00à 2.00à

Evonik Evonik

130 126

2.00à 2.00à

Literature value from Donnet et al. [42]. Average of literature values from Hewitt [43]. Fig. 1. Schematic of a typical DSC data trace schematically highlighting four analysis parameters – onset and midpoint temperatures, transition gradient and heat capacity step magnitude.

50 °C. The accuracy of the temperature controlling system was ±0.1 °C. Frequency domain dielectric spectra were collected at temperature steps of 1 K with an isothermal dwell time of approximately 10 min. 2.2.3. Dynamic mechanical analysis Temperature-dependent dynamic mechanical properties of the filled elastomers were measured using a Perkin Elmer 8000 DMA in tensile deformation mode. Rectangular samples were cut from sheets of elastomer. Samples were placed into the tensile clamps at room temperature and cooled to 183 K (about 40 K below the Tg). The clamps were fully tightened under cryogenic conditions. Data were collected at temperature steps of 2 K from 197 to 237 K with an isothermal step dwell time of 10 min. The dynamic strain was 0.1% (within the linear viscoelastic domain) with a static tensile preload of 0.1 N at a frequency of 1 Hz. 3. Results 3.1. Glass transition evaluation by DSC A typical DSC data trace showing the step change in heat flow at the glass transition is presented in Fig. 1. A number of different parameters can be extracted from such data. As indicated in Fig. 1, the transition onset and midpoint temperatures were determined for all samples and the gradient of the transition was also extracted. Papon et al. [36] have recently demonstrated a relationship between the broadening of the step change in heat capacity observed in scanning calorimetry measurements and the gradient of polymer glass transition around silica filler particles in PEA nanocomposites. A similar trend is observed in the DSC data presented by Fragiadakis et al. [31] for samples exhibiting a significant volume of interfacial polymer exhibiting retarded dynamics. Accordingly a linear fit to the step in heat capacity may be indicative of the broadening of Tg due to the presence of a gradient of immobilised polymer near the filler surface. Additionally the magnitude of the step change in heat capacity when

passing through the Tg has previously been used as a measure of the amount of polymer participating in the glass transition process [29,31]. This value is extracted for all samples in the study and is normalised to the weight of polymer in the sample. In Figs. 2–5 the parameters determined from the DSC traces are plotted versus the nitrogen specific surface area of the filler for all samples considered in this study. The filler type and surface activities are indicated on the plots. As can be seen from Figs. 2–5 neither the glass transition onset or midpoint temperatures, the transition gradient or the step change in heat capacity at the Tg depend significantly or systematically on the presence of filler, the filler surface area or the filler surface activity. A number of repeats were performed on the unfilled sample to determine the error associated with these values. As can be seen from examination of the error bar for the unfilled sample, most of the variation in data lies within the repeatability of the measurement. The DSC data is not sufficiently accurate enough to indicate any dependence of the glass transition temperature or glass transition gradient associated with the incorporation of any of the filler particles used in this study at these volume fractions. 3.2. Segmental dynamics Fig. 6A presents the frequency and temperature dependence of the dielectric losses for unfilled NR. The strong dielectric relaxations observed correspond to chain segmental (a) relaxations. Below the Tg the a relaxation is not observable within the test frequency range. Upon increasing the temperature, the a relaxation is first fully observed at around 213 K – just above the glass transition and subsequently shifts to higher frequencies with increasing temperature. Fig. 6B is an example of one of the filled NR dielectric data sets and presents a temperature and frequency domain plot of the dielectric loss for the HDS-TEVS filled NR compound. Several differences from the corresponding unfilled spectra are apparent. Firstly, tending towards lower frequencies, a gradual rise in the dielectric loss is

236

M. Huang et al. / European Polymer Journal 67 (2015) 232–241

A VogelFulcherTammann (VFT) fitting procedure (Eq. (3)) was performed on the extracted relaxation frequency data in order to quantitatively determine the glass transition temperatures from Arrhenius-type plots. The Tg is defined here as the temperature at which the relaxation time is 100 s [28]. The VFT plot and extracted Tg values are presented in Fig. 8 which also includes the data obtained for the low volume fraction N330-filled compound. Glass transition temperatures for these compounds are found to vary only slightly (maximum ±1 K) and nonsystematically from that of the unfilled elastomer.

e ðf Þ ¼ e1 þ

Fig. 2. Tg onset temperatures as a function of nitrogen-specific surface areas of fillers.

De a b

ð1 þ ðif =f 0 Þ Þ

ð1Þ

HN fitting function where e is the dielectric strength, f0 is the characteristic peak relaxation frequency, a and b are shape parameters and f is the frequency

e00 ¼ að2pf ÞðsÞ

ð2Þ

observed. This may be attributable to Maxwell–Wagner– Sillars (MWS) polarisation effects [44] and/or low frequency conductivity effects [25] and can be deconvoluted from the underlying polymer chain segmental relaxation using a single Havriliak and Negami (HN) fitting function (Eq. (1)) in combination with an additional low frequency term (Eq. (2)) [39]. An example of the deconvolution process is show in Fig. 7. In contrast to other investigations [31,35,39] we find no evidence of a secondary, lower frequency a relaxation which may be ascribed to a polymer interphase exhibiting retarded dynamics. Secondly, at higher frequencies and lower temperatures another, nonpolymer, dissipation becomes apparent. This dissipation process appears to be a partially resolved relaxation and could potentially be attributed to the relaxation of confined, hydrogen bonded water molecules at the silica surface which has previously been detailed for silica-filled SBR compounds [45]. In this paper we focus on the temperature dependence of the a relaxation. The characteristic frequency of the relaxation process was determined for each temperature step using an HN fit to each frequency domain spectrum.

MWS/conductivity dispersion term where a and s are fitting parameters and f is the frequency

Fig. 3. Tg midpoint temperatures as a function of nitrogen-specific surface areas of fillers.

Fig. 4. Transition gradient as a function of nitrogen-specific surface areas of fillers.

logf p;a ¼ logf 0 

A T  T0

ð3Þ

VFT equation where f0 and A are fitting parameters. T0 is the Vogel temperature (the ideal glass transition temperature) which is generally 30–70 K below the glass transition temperature, Tg. fp,a is the frequency of the a relaxation peak at temperature T. 3.3. Dynamic mechanical properties DMA data are presented in Figs. 9A and 10. The data were collected at a suitably small strain to ensure that the measurements were within the filled samples’ linear viscoelastic regions. This means that the storage moduli reflect the material properties without any contribution from the softening or yielding of the percolated filler network – which would significantly complicate interpretation. Testing was performed under isothermal

M. Huang et al. / European Polymer Journal 67 (2015) 232–241

237

type of data in terms of a broadening of the glass transition very difficult as it is not possible to discriminate contributions to total material stiffness arising from the load-bearing filler network and any immobilised polymer phase. However, contrasting the temperature dependence of the polymer relaxation time from BDS with the storage moduli (Fig. 9A versus Fig. 9B) indicates that this substantial modulus amplification may not be attributable to any filler-induced modifications of the polymer phase dynamics. In Fig. 9B the polymer segmental relaxation times, s, were calculated using Eq. (4) from the characteristic relaxation frequency, f0, and found to be in line with values computed from previously published data [37].

s¼ Fig. 5. Step change in heat capacity at the Tg as a function of nitrogenspecific surface areas of fillers.

stepwise conditions comparable to those employed for the BDS testing procedure. Fig. 9A details the temperature dependence of the storage moduli of the silica-filled samples along with the unfilled control sample. The corresponding viscous moduli data are presented in Fig. 10. Onset temperatures for the glass-rubber transition were extracted by linear intersection extrapolation of the E0 data and are presented as an inset in Fig. 9A. There is no systematic change in the onset temperature of the glass transition from E0 data; in line with the DSC data. We conclude from these data that a bulk change in the glass transition temperature does not occur upon the introduction of filler particles. Filler-induced storage modulus amplification (reinforcement) is observed on the rubbery side of the glass transition. Such stiffening effects have been attributed to the presence of glassy polymer around the filler particles [6] and to filler networking effects [34]. The presence of a percolated filler network makes interpretations of this

1 2pf 0

ð4Þ

The polymer segmental relaxation time, s, calculated from the relaxation frequency, f0. 4. Discussion The data presented here suggests that the dynamics of the polymer phase in these samples are not significantly modified by the presence of the various filler particles. The DSC data show no filler-induced changes in the Tg. Previously, DSC analyses of filled elastomers which display a pronounced polymer interphase have been shown to be sensitive to the broadening and magnitude of the step change in heat capacity at the Tg [31,36]. In light of this, our DSC data indicates that if such an interphase exists in the samples studied here, then the total volume fraction of modified polymer is small and that the associated thermal transitions are within the measurement error of this particular instrument. The temperature dependence of the dielectric a relaxation in the vicinity of the glass transition for filled samples shows little variation from that of the unfilled material for

Fig. 6. The frequency and temperature dependence of the dielectric loss for (A) Unfilled NR and (B) HDS-TEVS filled NR.

238

M. Huang et al. / European Polymer Journal 67 (2015) 232–241

Fig. 7. Example de-convolution of the low frequency relaxation process from the polymer a relaxation using an HN function and a conductivity contribution.

all silica-filled samples and for a carbon black-filled sample well below the apparent electrical percolation threshold loading. This is significant as the transition region and lower temperature rubbery region are where one may have expected to observe the maximum effect of any filler-induced perturbation of the elastomer dynamics. Filled sample small strain storage moduli measured using DMA are significantly amplified on the rubbery side of the Tg by the presence of the filler phase. The lack of any systematic correlation of these reinforcing effects with temperature-dependent changes in the underlying polymer dynamics measured by BDS and from the dynamic loss moduli suggests that this amplification may be associated with the mechanical nature of the filler network within the sample. In the context of previous investigations this may not be so surprising. For example Fragiadakis et al. [31] demonstrated the sensitivity of the polymer interphase to the degree of filler particle aggregation. The filler particles

Fig. 8. Arrhenius plots for the characteristic dielectric relaxation frequency (f0). The VFT fit to the unfilled material data is shown as a solid line. Inset: extracted Tg values from VFT fittings.

studied in this paper are all industrial grade fillers (or derived thereof) and as such display a significant level of primary particulate aggregation. These particles cannot therefore be classed as true nanoparticles, but rather as nanostructured aggregates – with correspondingly reduced filler-polymer contact areas. In addition, upon shear processing and thermal curing of the samples, filler aggregate agglomeration (flocculation) may occur, which can act to further deplete the surface area of filler in intimate contact with the polymer and detrimentally affect the degree of filler dispersion. Therefore the nature of confinement of polymer in these systems may be very different to previously published reports of filled elastomers and thermoplastics where near-perfect dispersions of isolated spherical nanoparticles have been shown to significantly alter the dynamics of the polymer in proximity to the filler [6,12,31,36]. Example TEM and SEM micrographs of individual silica aggregates and fracture surface filler morphology are presented in Fig. 11 and serve to illustrate the aggregated and agglomerated nature of the filler particles and filler particle networks studied in this work respectively. The precise role of particle dispersion or aggregation in dictating a local modification of the Tg is beyond the scope of this paper but may in part explain why no significant interphase can be detected for the commercially relevant compounds studied here. It is also worthwhile considering the very recent publication of Mujtaba et al. [40]. They report a small volume of immobilised polymer in commercial silica

Fig. 9. (A) Small strain storage moduli for silica-filled and unfilled samples as a function of temperature. (B) Segmental relaxation time calculated from BDS testing as a function of temperature.

M. Huang et al. / European Polymer Journal 67 (2015) 232–241

239

detection associated with the bulk characterisations performed in this study. Interestingly, previous calorimetric studies of the Tg of so-called ‘bound rubber’ in SBR/Silica commercial systems showed no significant deviations from the bulk rubber Tg [34]. Note that the term ‘bound rubber’ used in this sense refers to the fraction of polymer in an un-crosslinked, filled rubber, which is insoluble in solvent. Solvent extraction removes all soluble polymer chains from the compound and what remains is polymer in intimate contact with the filler. One may have expected this type of study to increase the sensitivity of the DSC technique to any small volume of immobilised polymer and thereby more clearly expose the effects of the filler on the polymer Tg. 4.1. Origins of modulus amplification Fig. 10. Small strain loss moduli for silica-filled and unfilled samples as a function of temperature.

aggregate-filled SBR determined using solid state 1H NMR relaxometry experiments. Their reported volumes of glassy polymer are small – of the order of 1–3% total volume of the composite – which could be within the limits of

Fig. 11. (A) TEM image of individual CS filler aggregates and (B) SEM image of the CS filler dispersion state at a filled sample fracture surface prepared under cryogenic conditions.

It is pertinent at this point to discuss mechanisms other than a filler-induced broadening of the Tg which could possibly account for the observed trends in small strain modulus amplification in the rubbery region. It has previously been proposed that occlusion or shielding of a part of the total volume of polymer from the globally applied strain by the fractal filler network increases the effective volume fraction of the solid reinforcing phase [46]. Additionally, local strain amplification due to the presence of filler inclusions raises the average stored energy density in the polymer phase upon deformation [47,48]. Accordingly the presence of a geometrically complex filler network can act to amplify the storage modulus significantly above that determined analytically via the Einstein–Smallwood and Guth–Gold hydrodynamic equations [49,50]. Unfortunately it is very difficult to experimentally verify whether or not these effects alone could account for the modulus amplification observed for the systems studied here. Bergström and Boyce [51] have demonstrated using stochastic 3D finite element simulations that random dispersions of fillers in a hyperelastic matrix can accurately match, and in some cases exceed, the stiffening at small strains observed in experimental data. Both occlusion of elastomer phase and strain amplification mechanisms are apparent in their simulation results. Gusev and Rozman [52] performed a finite element study of the effects of filler particle volume fraction and morphology on the small strain stiffness of hyperelastic composites. They found that fillers exhibiting what they termed ‘honeycomb’ morphologies could reinforce compounds up to levels 100 times that of the unfilled rubber modulus at realistic volume fractions. This is reasonably consistent with experimental observations [6]. Furthermore Akutagawa et al.’s finite element analyses of three dimensional representations of real filler dispersions (determined experimentally using a microtomography technique) embedded in an hyperelastic continuum have demonstrated that such mechanisms can actually overestimate the magnitude of experimentally observed stiffening [48]. Importantly all the aforementioned simulations did not incorporate any alteration of the dynamics of the elastomer phase by the filler and yet were still able to approximate the appropriate stiffening magnitude – suggesting that stiffening can be explained by the nature of filler networking.

240

M. Huang et al. / European Polymer Journal 67 (2015) 232–241

In fact, in some of Payne’s original work [53] on filler reinforcement he demonstrated that at small strains the shear moduli of carbon black-filled paraffin oil approaches that of equivalently filled rubber – indicating that the observed filler-induced stiffening of rubber may be relatively unrelated to the rubber phase and instead may be more dependent on the mechanical properties of the filler network. Robertson and Wang [54] have presented similar findings for silica-filled diisodecyl adipate oil. PérezAparicio and co-workers have studied silica-filled NR compounds which are reasonably similar to those described in this paper. They have demonstrated that the linear viscoelastic storage modulus amplification observed for their compounds exceeds the experimentally-discriminated levels of stiffening arising from strain amplification effects alone [55] and therefore contains a contribution to stiffening which is not related to the general entropic elasticity of the polymer network. A key issue is therefore; what mediates the load transfer between fractal aggregates composing the filler network? This paper suggests that load transfer cannot be attributed to large volumes of immobilised polymer between filler aggregates. The role of very small quantities of immobilised polymer (which may be within the detection limits of the bulk characterisation techniques used in this paper) is open to further investigation. 5. Conclusion A set of complimentary experimental techniques were employed to study the glass transition, polymer segmental dynamics and particulate reinforcement of natural rubber. A series of carbon black and silica particles were used as the reinforcing phase, varying in particulate morphology, surface area and surface activity. Little evidence was found for a substantial effect of the fillers on the polymer phase dynamics, despite using techniques which have previously been demonstrated to be sensitive to such phenomena. No effect of differing surface activities of the particles on the Tg was detectable. The reason for a distinct lack of filler-induced polymer modification may be due to the nature of polymer confinement associated with the imperfect dispersion state of the fillers in the sample and the fact that the fillers themselves are formed from aggregated primary particles rather than being true nanoparticulates. These findings were discussed in the context of possible small strain reinforcement mechanisms for these materials. Acknowledgements L.B.T. would like to thank EPSRC (UK) and Sibelco for doctoral funding. M.H. would like to thank the Chinese Scholarship Council for doctoral funding. The authors are grateful to Prof. Ren Wei of Xi’an Jiaotong University for allowing access to the dielectric spectrometer. The authors thank Cancarb, Cabot Corporation and Solvay Silicas for the kind supply of their filler materials and TARRC for their assistance with some of the sample preparation.

Appendix A

PVAC P4VP PEA VPR BR CB PDMS PMMA PS SBR

Polyvinyl acetate Poly(4-vinylpyridine) Polyethyl acrylate Butadiene-styrene-vinyl pyridine rubber Polybutadiene rubber Carbon black Polydimethyl siloxane Poly(methyl methacrylate) Polystyrene Styrene butadiene rubber

References [1] Suphadon N, Thomas AG, Busfield JJC. Polym Test 2010;29:440–4. [2] Suphadon N, Thomas AG, Busfield JJC. Polym Test 2011;30:779–83. [3] Busfield JJC, Jha V, Hon AA, Thomas AG. Constitutive models for rubbers IV – Proceedings of the 4th European conference for constitutive models for rubber, vol. 4, 2005. p. 459. [4] Struik LCE. Polymer 1987;28:1535–41. [5] Wang MJ. Rubber Chem Technol 1998;71:521–89. [6] Berriot J, Montes H, Lequeux F, Long D, Sotta P. Macromolecules 2002;35:9756–62. [7] Long D, Lequeux F. Eur Phys J E: Soft Matter Biol Phys 2001;4:371–87. [8] Merabia S, Sotta P, Long DR. Macromolecules 2008;41:8252–66. [9] Robertson CG, Roland CM. Rubber Chem Technol 2008;81:506–22. [10] Keddie JL, Jones RAL, Cory RA. Europhys Lett 1994;27:59–64. [11] Fryer DS, Peters RD, Kin EJ, Tomaszewski JE, de Pablo JJ, Nealey PF. Macromolecules 2001;34:5627–34. [12] Bansal A, Yang H, Li C, Cho K, Benicewicz BC, Kumar SK, et al. Nat Mater 2005;4:693–8. [13] Mason P. J Appl Polym Sci 1960;4:212–8. [14] Smit PPA. Rheol Acta 1966;5:277–83. [15] Waldrop MA, Kraus G. Rubber Chem Technol 1969;42:1155–66. [16] Kraus G, Gruver JT. J Polym Sci Pol Phys 1970;8:571–81. [17] Kaufman S, Slichter WP, Davis DD. J Polym Sci Pol Phys 1971;9:829–39. [18] Kraus G, Rollmann KW, Gruver JT. Rubber Chem Technol 1971;44:92–6. [19] O’Brien J, Cashell E, Wardell GE, McBrierty VJ. Macromolecules 1976;9:653–60. [20] Stacer RG, Husband DM. Rheol Acta 1990;29:152–62. [21] Tsagaropoulos G, Eisenberg A. Macromolecules 1995;28:396–8. [22] Fragiadakis D, Pissis P, Bokobza L. Polymer 2005;46:6001–8. [23] Fragiadakis D, Pissis P, Bokobza L. J Non-Cryst Solids 2006;352:4969–72. [24] Amanuel S, Gaudette AN, Sternstein SS. J Polym Sci Pol Phys 2008;46:2733–40. [25] Bogoslovov RB, Roland CM, Ellis AR, Randall AM, Robertson CG. Macromolecules 2008;41:1289–96. [26] Meier JG, Klüppel M. Macromol Mater Eng 2008;293:12–38. [27] Robertson CG, Lin CJ, Rackaitis M, Roland CM. Macromolecules 2008;41:2727–31. [28] Ding Y, Pawlus S, Sokolov AP, Douglas JF, Karim A, Soles CL. Macromolecules 2009;42:3201–6. [29] Robertson CG, Bogoslovov R, Roland CM. Rubber Chem Technol 2009;82:202–13. [30] Göritz D, Hofmann RWP, Bissem HH. Rubber Chem Technol 2010;83:323–30. [31] Fragiadakis D, Bokobza L, Pissis P. Polymer 2011;52:3175–82. [32] Fritzsche J, Klüppel M. J Phys-Condens Matter 2011:035104. [33] Robertson CG, Mindaugas R. Macromolecules 2011;44:1177–8. [34] Robertson CG, Lin CJ, Bogoslovov RB, Rackaitis M, Sadhukhan P, Quinn JD, et al. Rubber Chem Technol 2011;84:507–19. [35] Vo LT, Anastasiadis SH, Giannelis EP. Macromolecules 2011;44:6162–71. [36] Papon A, Montes H, Hanafi M, Lequeux F, Guy L, Saalwächter K. Phys Rev Lett 2012;108:065702. [37] Hernández M, Mar Bernal MD, Verdejo R, Ezquerra TA, LópezManchado MA. Compos Sci Technol 2012;73:40–6.

M. Huang et al. / European Polymer Journal 67 (2015) 232–241 [38] Kumalli MM, Miccio LA, Schwartz GA, Alegría A, Colmenero J, Otegui J, et al. Polymer 2013;54:4980–6. [39] Wu S, Tang Z, Guo B, Zhang L, Jia D. RSC Adv 2013;34:14549–59. [40] Mujtaba A, Keller M, Ilisch S, Radusch H-J, Beiner M, Thurn-Albrecht T, et al. ACS Macro Lett 2014;3:481–5. [41] Tunnicliffe LB, Kadlcak J, Shi Y, Morris MD, Thomas AG, Busfield JJC. Macromol Mater Eng 2014;299(12):1474–83. [42] Hess WM, Herd CR. In: Donnet JB, Bansal RC, Wang MJ, editors. Carbon black. New York: Marcel Dekker Inc.; 1983. p. 147 [Chapter 3]. [43] Hewitt N. In: Hewitt N, editor. Compounding precipitated silica in elastomers. Norwich: William Andrew Publishing; 2007. p. 5 [Chapter 1]. [44] Otegui J, Schwartz GA, Cerveny S, Colmenero J, Loichen J, Westermann S. Macromolecules 2013;46:2407–16.

[45] [46] [47] [48] [49] [50] [51] [52] [53] [54] [55]

241

Klüppel M. J Phys-Condens Mater 2009;21:035104. Medalia AI. J Colloid Interf Sci 1970;32:115–31. Mullins L, Tobin NR. J Appl Polym Sci 1965;9(9):2993–3009. Akutagawa K, Yamaguchi K, Yamamoto A, Heguri H, Jinnai H, Shinbori Y. Rubber Chem Technol 2008;81:182–9. Smallwood HM. J Appl Phys 1944;15:758–66. Guth E. J Appl Phys 1945;16:20–5. Bergström JS, Boyce MC. Rubber Chem Technol 1999;72: 633–56. Gusev AA, Rozman MG. Comput Theor Polym Sci 1999;9:335–7. Payne AR. Rubber Chem Technol 1965;38:387–99. Robertson CG, Wang X. Phys Rev Lett 2005;95:075703. Pérez-Aparicio R, Vieyres A, Albouy PA, Sanséau O, Vanel L, Long DR, et al. Macromolecules 2013;46:8964–72.