Natural Rubber–Filler Interactions: What Are the Parameters

Article pubs.acs.org/Langmuir

Natural Rubber−Filler Interactions: What Are the Parameters? Alan Jenkin Chan,*,†,‡ Karine Steenkeste,† Alexis Canette,§ Marie Eloy,‡ Damien Brosson,‡ Fabien Gaboriaud,‡ and Marie-Pierre Fontaine-Aupart† †

Institut des Sciences Moléculaires d’Orsay (ISMO), CNRS, Univ. Paris-Sud, Université Paris-Saclay, F-91405, Orsay, France Manufacture Française des Pneumatiques Michelin, 23 place des Carmes Déchaux, 63040 Clermont Ferrand Cedex 9, France § INRA AgroParisTech, Micalis, UMR 1319, 91300 Massy, France ‡

S Supporting Information *

ABSTRACT: Reinforcement of a polymer matrix through the incorporation of nanoparticles (fillers) is a common industrial practice that greatly enhances the mechanical properties of the composite material. The origin of such mechanical reinforcement has been linked to the interaction between the polymer and filler as well as the homogeneous dispersion of the filler within the polymer matrix. In natural rubber (NR) technology, knowledge of the conditions necessary to achieve more efficient NR−filler interactions is improving continuously. This study explores the important physicochemical parameters required to achieve NR−filler interactions under dilute aqueous conditions by varying both the properties of the filler (size, composition, surface activity, concentration) and the aqueous solution (ionic strength, ion valency). By combining fluorescence and electron microscopy methods, we show that NR and silica interact only in the presence of ions and that heteroaggregation is favored more than homoaggregation of silica−silica or NR−NR. The interaction kinetics increases with the ion valence, whereas the morphology of the heteroaggregates depends on the size of silica and the volume percent ratio (dry silica/dry NR). We observe dendritic structures using silica with a diameter (d) of 100 nm at a ∼20−50 vol % ratio, whereas we obtain raspberry-like structures using silica with d = 30 nm particles. We observe that in liquid the interaction is controlled by the hydrophilic bioshell, in contrast to dried conditions, where hydrophobic polymer dominates the interaction of NR with the fillers. A good correlation between the nanoscopic aggregation behavior and the macroscopic aggregation dynamics of the particles was observed. These results provide insight into improving the reinforcement of a polymer matrix using NR−filler films. conventional one is to disperse the fillers in the polymer matrix by direct melt compounding.10 Another approach, also in dry technology, is to functionalize directly the surface of the filler particles with a coupling agent to improve the compatibility between the filler and the polymer matrix11,12 or to graft the fillers on the surface of the polymers.13 Surface-initiated redox and plasma polymerization have also been tested to achieve a better filler dispersion.14 An alternative approach, involving liquid mixing, first requires the dispersion of binary colloids in liquid to allow the interaction to take place, after which they are dried.15−18 This approach takes advantage of the micellar structure of NR particles in aqueous suspension to enhance the NR and (hydrophilic) filler interactions, which improves the dispersion of the filler in the resulting composite. Oberdisse has reported the structural−mechanical property relationship of hard filler (silica) and soft (polymer matrix) composite, which revealed a promising effect of the strong reinforcement capabilities using

I. INTRODUCTION Natural rubber (NR) latex is a commercially successful biobased polymer widely used because of its excellent intrinsic physical properties.1,2 However, it is seldom used in its raw state. Processes involving the transformation from a linear polymer into a composite material are necessary to achieve the improved elastomeric properties required for most applications. Conventionally, this process involves mechanical mixing of a solid block of NR with compounding ingredients or additives, such as fillers, oils, and powders (e.g., sulfur).3 Fillers are added as a reinforcing material to increase the tensile strength, improve the tear strength, and enhance the abrasion resistance of the rubber product.4 At present, the mechanisms of reinforcement are not fully understood; however, there is a general agreement on two factors that have essential contributions to the final properties of the NR composite material: the dispersion of the filler in the rubber matrix and the interaction between NR particles and the filler.5−8 The ability to homogeneously disperse the filler in the NR matrix remains one of the most challenging tasks in rubber processing.9 Several routes have been designed; the most © 2015 American Chemical Society

Received: August 29, 2015 Revised: October 18, 2015 Published: October 21, 2015 12437

DOI: 10.1021/acs.langmuir.5b03244 Langmuir 2015, 31, 12437−12446

Article

Langmuir liquid mixing.15 The study, however, relied on the fusion of the polymer and silica when the solution was dried. As a result, the heteroaggregation mechanisms remain unknown because of the difficulties involved in examining the aggregation process during drying. Several authors pointed out that an in-depth understanding of the mechanisms that affect NR−filler interactions prior to dehydration is the key to opening new perspectives toward tuning reinforcement effects. This approach can also open new fields of study, such as making composite materials using synthetic latex.15,18 This has led us to investigate the physicochemical parameters that control the aggregation process in liquid medium. Our aim is to identify the key properties related to the filler (type, size, surface functionality) and the aqueous medium (ionic strength, ion valency) that trigger NR−filler interactions and, subsequently, aggregation. We performed experiments using the combined methods of fluorescence correlation spectroscopy (FCS) and fluorescence intensity imaging to identify the interactions within and to elucidate the aggregate structure in solution on the nanoscale. Transmission electron microscopy (TEM) was also used to provide higher spatial resolution images following a less destructive sample preparation protocol. We also performed aggregation assays to correlate the nano- and macroscopic behaviors of NR and the filler. Unless otherwise mentioned, experiments were performed under dilute conditions to minimize homoaggregation and maximize heteroaggregation between NR and the filler.

Data acquisition was performed with the SymPhoTime software, which directly correlates the fluorescence fluctuation signals. Data acquisition time for all samples was 30 s, and at least 30 curves were acquired for each sample. The autocorrelated curves were fitted with a two-component 3D Brownian diffusion (eq 1) model.20−24 ⎧ ⎤1/2 ⎤⎡ 1 ⎪ ⎡⎢ 1 1 ⎥ ⎥⎢ ⎨α 8 N ⎪ ⎢⎣ 1 + (τ /τd1) ⎥⎦⎢⎣ 1 + (ω0 /z 0)2 (τ /τd1) ⎥⎦ ⎩

g (τ ) =

⎤1/2 ⎫ ⎡ ⎤⎡ ⎪ 1 1 ⎥ ⎬ ⎥⎢ + (1 − α)⎢ 2 ⎢⎣ 1 + (τ /τd2) ⎥⎦⎢⎣ 1 + (ω0 /z 0) (τ /τd2) ⎥⎦ ⎪ ⎭ (1) where α and 1 − α are the fractions of the molar concentrations of the two diffusive species, τd1 and τd2 are their respective translational diffusion times, N is the number of fluorescent species inside the excitation volume, and ω0/z0 is the lateral/axial radii ratio of the laser beam (ω0/z0 = 0.30 ± 0.03 was determined elsewhere).24 For twophoton excitation, the diffusion time obtained in eq 1 is related to the diffusion coefficient (D) (eq 2).22

τd =

ω02 8D

(2)

Assuming spherical particles, it is possible to calculate the hydrodynamic size of the nanoparticles using the Stokes−Einstein relationship (eq 3). D=

kT 6πηr

(3)

where k is the Boltzmann constant, T is the temperature, η is the viscosity of the medium, and r is the radius of the spherical particle. When eq 1 cannot be fitted to the autocorrelation curve, the mean diffusion time is estimated directly from the curve by reading the time at half amplitude of the autocorrelation function g(τ). II.C. Confocal Laser Scanning Microscopy. Fluorescence images were acquired using the Leica TCS SP5-AOBS confocal laser scanning microscope implemented at the Centre de Photonique Biomedicale (CPBM, Orsay, France). The sample (volume = 500 μL) was deposited on a coverslip for 2 h at room temperature. The nonadhered particles were then washed off three times with solvent by tilting the coverslip and allowing a stream of the solvent to gently run down. Then, the coverslip was placed in the attofluor chamber, and 1 mL of solvent was added to maintain an aqueous environment. The attofluor chamber was mounted directly onto the microscope. The samples were imaged using a 63×, 1.4 N.A. plane apochromat oil immersion objective. The resolution of the confocal images was 512 × 512 pixels2 recorded on 8-bits with varying magnification values (2×, 5×, and 10×). A 543 nm helium−neon visible laser line was used as the excitation source for all of the fluorescent fillers. The fluorescence was collected within the spectral range of 560−700 nm. Each image corresponds to an average of four frames. II.D. Transmission Electron Microscopy. Transmission electron microscopy (TEM) was performed using an HT7700 transmission electron microscope (Hitachi, Japan) equipped with an 8 million pixels format CCD camera driven by the image capture engine software AMT, version 6.02, at the MIMA2 microscopy platform (Jouy-enJosas, France). The samples were fixed with 1% osmium tetraoxide (Electron Microscopy Sciences, LFG distribution, France) and 20 mM Tris buffer (Sigma-Aldrich) for 1 h at room temperature. The salt used for the Tris buffer was either NaNO3 or MgSO4, depending on the sample. The suspension was deposited on a 300-mesh nickel grid with a Formvar carbon film (Electron Microscopy Sciences, LFG distribution, France) for 1 h at room temperature. Then, the samples were washed three times using Tris buffer for ∼5, ∼2, and ∼1 min and dried between washes using Whatman grade no. 1 cellulose filter paper. Observations were made at 100 kV in high-resolution mode with an objective aperture adjusted to obtain sufficient contrast.

II. EXPERIMENTAL SECTION II.A. Sample Preparation and Materials. High ammonia NR latex concentrate (∼60 wt %, HA latex) was obtained from Trang Latex Co. Ltd., Thailand. In contrast to the fillers classically used in reinforced polymers, which resemble a fractal structure but are not fluorescent,19 all filler particles used in this study need to be max fluorescent (λmax ex = 570 nm and λem = 585 nm) and are thus spherical: nonfunctionalized silica particles (plain silica), carboxylate-functionalized silica particles (COOH silica), and carboxylate-functionalized polystyrene particles of 30, 50, and 100 nm in diameter (d) were purchased from Kisker Biotech, Germany. The NR concentration was kept constant at a dilute concentration of 1.8 × 10−2 wt % (except for macroscopic aggregation assays), and the concentration of the fillers was appropriately adjusted to give a final volume percent ratio (dry silica/dry NR) of 20%. This ratio was used for all experiments except when the effect of concentration on the NR−silica interaction was studied. The fillers were sonicated in a water bath for ∼15 min prior to mixing with NR. The solutions used are nonionic (Milli-Q water) and ionic. Ionic solutions were composed of sodium nitrate (NaNO3) and magnesium sulfate (MgSO4) (Sigma-Aldrich). The concentrations of the salts were appropriately adjusted to obtain an ionic strength of 100 mM. II.B. Fluorescence Correlation Spectroscopy. Fluorescence correlation spectroscopy (FCS) was performed using a home-built two-photon excitation setup. In this setup, a drop of the NR−filler suspension was placed on a coverslip, which was immediately followed by FCS measurement. After 1 h, another drop taken from the same suspension was tested. The coverslip was placed on a sample holder and mounted on a 63× 1.4 N.A. oil immersion objective. The sample was biphotonically excited at 800 nm using a Ti:sapphire pulsed laser (150 fs, 76 MHz, MIRA 900, Coherent Inc.) pumped in the green region by a continuous wave solid-state diode laser (11 W, VERDI, Coherent Inc.). The fluorescence was then collected through the same objective and separated from the excitation beam by a dichroic mirror before being focused on a hybrid photomultiplier detector (PMA hybrid 40, PicoQuant, Germany). Fluorescence intensity fluctuations were recorded using a TimeHarp260 module (PicoQuant, Germany). 12438

DOI: 10.1021/acs.langmuir.5b03244 Langmuir 2015, 31, 12437−12446

Article

Langmuir II.E. Macroscopic Aggregation Assays. Aggregation tests were conducted at a constant volume percent ratio (dry silica/dry NR) of 20%, similar to the nanoscopic measurements, but the concentration of NR was increased to ∼0.70 wt %. The solutions used were nonionic (Milli-Q) and ionic (NaNO3 and MgSO4, Sigma-Aldrich). The concentrations of the salts were appropriately adjusted to an ionic strength of 100 mM. If precipitates formed at the bottom of the test tube and the solution showed no visual sign of turbidity after 24 h, then heteroaggregation was considered to have taken place.

III. RESULTS Our aim is to quantify the different NR−filler interactions in aqueous suspensions by systematically varying certain parameters while keeping others constant. III.A. Effects of Ions on Silica−Silica and NR−NR Interactions. The stability of the separate suspensions of plain silica particles (d = 30, 50, and 100 nm) and NR particles was determined using FCS and confocal fluorescence imaging, respectively. We performed these experiments under three conditions: (1) in the absence of salts, (2) in the presence of Na+ ions at 100 mM ionic strength, and (3) in the presence of Mg2+ ions also at 100 mM ionic strength. For plain silica particles in the absence of salts, the typical normalized fluorescence autocorrelation curves are shown in Figure 1a. They were fitted with the two-component diffusion model (eq 1). The diffusion time of the first component (Table 1) corresponds to the monomeric form of the particles, as ascertained by their respective hydrodynamic size calculated using eqs 2 and 3. For each autocorrelation curve, a second component was observed with a longer diffusion time of 58 ± 12 ms. This may be attributed to aggregates normally present at ∼25 ± 3% of the total population. In Na+ and Mg2+ suspensions, the autocorrelation curves and diffusion times obtained for 30 and 50 nm silica particles were similar to those in the absence of salts. However, for 100 nm silica particles, we obtained large, irregular, and irreproducible autocorrelation curves (data not shown) typical of large aggregates. Because of the irregularity of these curves, eq 1 could not be applied. However, an approximation (see Experimental Section) gave a diffusion time ∼120−150 times greater than that of a single 100 nm particle. As far as NR particles are concerned, they are considered to have a bimodal distribution of small (∼100−200 nm) and large (∼500−800 nm) rubber particles.25,26 In the absence of salts, we illustrated in a previous study that NR particles are mainly monomeric, in agreement with the literature.27 In the presence of salts (Na+ and Mg2+), using confocal imaging in transmission mode, we observed both monomers (isolated particles) and aggregates of NR particles, as shown in Figure 1b, also in agreement with the literature.28 III.B. Effects of Ions on NR−Silica Interactions: Kinetics and Morphology. The interactions between NR and plain silica particles will be described in terms of the kinetics and morphology. In cases where an interaction was observed, confocal and TEM images are provided to visualize the heteroaggregate structures. For the mixed NR−plain silica particles in the absence of salts, Figure 2 shows that the normalized autocorrelation curves and the corresponding calculated diffusion times are comparable to those of the free silica particles (Table 1). This was observed for all three sizes of silica particles, showing a lack of interaction between the two types of particles. However, in Na+ and Mg2+ suspensions, the normalized autocorrelation curve shifts to yield the same curve at a longer diffusion time for the

Figure 1. (a) Normalized autocorrelation curves of the separated suspensions of plain silica particles of d = 30 nm (○), d = 50 nm (△), and d = 100 nm (☆) in the absence of salts. In NaNO3 and MgSO4 suspensions at 100 mM ionic strength (IS), similar curves were obtained except for the 100 nm plain silica particles. (b) Transmission image of NR particles in a MgSO4 suspension (100 mM IS) obtained from a fluorescence microscope. Similar images were obtained for NR particles in a NaNO3 suspension at 100 mM IS.

three different suspensions (NR−30 nm plain silica, NR−50 nm plain silica, NR−100 nm plain silica), also shown in Figure 2. An approximation provided a mean diffusion time of ∼70− 100 ms, which corresponds to d = ∼1−1.4 μm and is consistent with the size of the NR−plain silica complex. In terms of the aggregation dynamics, it should be recalled that in the presence of ions 100 nm plain silica particles yielded irreproducible curves due to the formation of large silica−silica aggregates. However, this was not observed for the mixed NR−silica suspension. Therefore, the results suggest that under dilute conditions heteroaggregation between NR and silica particles is favored over the homoaggregation between silica−silica particles. In terms of the heteroaggregation kinetics, comparison of the FCS measurement was performed on two time scales: immediately and 1 h after mixing NR and silica. In the Mg2+ suspension, the FCS measurement showed that the particles interacted immediately. However, when the Na+ suspension was measured immediately, we observed a curve similar to that of the free plain silica particles; it took ∼1 h to observe the shift characteristic of the NR−plain silica particle interaction. 12439

DOI: 10.1021/acs.langmuir.5b03244 Langmuir 2015, 31, 12437−12446

Article

Langmuir

Table 1. Diffusion Time and the Corresponding Calculated Size of the Silica Particles and NR−Silica in Water and Salt Suspensions wth ions (Na+ or Mg2+)

without ions suspension plain silica particles

NR−plain silica particles a

M size (d, nm)a 30 50 100 30 50 100

τd1 (ms) 2.0 3.0 6.0 1.9 3.0 6.0

± ± ± ± ± ±

0.2 0.3 0.3 0.2 0.2 0.3

C size (d, nm)b

τd1 (ms)

± ± ± ± ± ±

1.9 ± 0.3 2.8 ± 0.05 ≥720

29 ± 3.0 43 ± 6 ≥1.1 × 104

∼70−100

(1.1−1.5) × 103

31 46 110 30 46 110

3.0 5.0 32 3.0 4.0 32

C size (d, nm)b

Manufacturer. bCalculated.

Figure 2. Normalized autocorrelation curves of the mixed suspensions of NR and plain silica particles of d = 30 nm (°), d = 50 nm (△), and d = 100 nm (☆) in the absence of salts. These curves are similar to the ones obtained on the separate suspensions of silica particles shown in Figure 1. In the NaNO3 and MgSO4 suspensions at 100 mM ionic strength, the autocorrelation curves (●) are similar regardless of the size of the silica particles. This shift to the same curve at a longer diffusion time indicates an interaction between NR and plain silica.

We also visualized the structure of the heteroaggregates in the ionic suspensions, where NR−plain silica particle interactions were observed. For the Na+ suspension, typical superimposed fluorescence and transmission images obtained from the fluorescence confocal microscope after 2 h of interaction are shown in Figure 3a. The transmission image (gray) corresponds to NR, whereas the green objects correspond to the fluorescent silica particles. Two general observations can be made: (i) monomers of large rubber particles (LRPs, ∼500−600 nm) either do not or rarely interact with silica particles and (ii) patches of fluorescent particles (∼1−2 μm) are seen homogeneously distributed on the substrate. Extending the duration of the interaction between NR and silica particles to 4 h seems to result in the growth of the fluorescent patches into large NR−silica heteroaggregates (including LRPs) with dendritic morphology, as shown in Figure 3b. The fluorescent patches can be either homoaggregates of silica particles or heteroaggregates of silica and small rubber particles (∼100−200 nm). The lateral resolution limit of the confocal microscope (∼450 nm) prevented us from identifying the nature of the aggregates. For this reason, we used TEM. Typical micrographs showed that silica and NR have relatively similar opacities, but they can still be differentiated through their charge distribution or morphology, as shown in Figure 4a−d. NR particles show a uniform charge distribution. Therefore, they display a smooth morphology, whereas silica

appears to be granular as a result of the nonuniform charge distribution since silicon (Si) has a greater electron density than oxygen (O). However, we do not exclude the possibility that the silica particles are nanoporous. After 2 h, mostly NR−silica interactions were observed, but we could not rule out silica− silica aggregates. The NR−100 nm plain silica particle complexes took several forms: from a simple 1:1 interaction (Figure 4a) to the formation of irregularly shaped (dendritic) heteroaggregates on the scale of ∼1 μm (Figure 4b), which is similar to the size of the fluorescent patches observed in the fluorescence intensity images (Figure 3). For the mixed NR−100 nm plain silica particles in the Mg2+ suspension, dendritic morphology was observed using fluorescence confocal microscopy. Between the behavior of the NR and the filler in the Na+ and Mg2+ solutions, the difference lies in the kinetics of their interaction, which is in agreement with the results obtained from the FCS measurements. The formation of NR−100 nm silica heteroaggregates in the Mg2+ suspension was 2 times faster than that in the Na+ suspension, as shown in Figure 3c,d. In the Mg2+ suspension after 2 h of interaction, the size of the dendritic-shaped aggregates was already comparable to that obtained after 4 h in the NaNO3 suspension. Videos of the NR−silica heteroaggregation process in the separate suspensions of NaNO3 and MgSO4 recorded in real-time with the fluorescence confocal microscope are also provided to demonstrate the heteroaggregation process as well as the difference in the interaction 12440

DOI: 10.1021/acs.langmuir.5b03244 Langmuir 2015, 31, 12437−12446

Article

Langmuir

Figure 3. Superimposed transmission and fluorescence intensity images of the mixed NR−100 nm plain silica in (a, b) NaNO3 and (c, d) MgSO4 suspensions. Images after (a, c) 2 h and (b, d) 4 h of interaction. NR (black) and silica particles (green). Ionic strength of the solutions: 100 mM.

Figure 4. TEM micrographs of NR−100 nm plain silica after 2 h of interaction in (a, b) NaNO3 and (c, d) MgSO4 suspensions at 100 mM ionic strength. (b, d) Magnified image of the red rectangular zones in (a) and (c), respectively. Inset in (a) is a magnified image of a 1:1 NR−plain silica particle interaction.

and Ca2+ as cations. Results obtained for K+ and Ca2+ were comparable to those with Na+ and Mg2+, respectively (images not shown). Interestingly, the morphology of the NR−silica heteroaggregate seems to be dependent on the size of the silica particle. For the mixed NR−30 nm plain silica in the Mg2+ suspension,

kinetics (Supporting Information). Results from TEM are also in agreement with the results from confocal microscopy, as shown in Figure 4c,d, in which aggregates larger than 2 μm were also observed. To verify that the nature of the ion does not influence the rate of aggregation, we also performed experiments using K+ 12441

DOI: 10.1021/acs.langmuir.5b03244 Langmuir 2015, 31, 12437−12446

Article

Langmuir

plain complex (∼70−100 ms). In terms of the morphological structure, NR−COOH silica particle complexes exhibited both small (∼1−2 μm) and large (>2 μm) types of dendritic aggregates (observed in fluorescence confocal microscopy and TEM), not large heteroaggregates only, as was observed with the plain silica nanoparticles (Figure 4a,b). Furthermore, NR− COOH silica particle interactions were not always observed, as opposed to NR−plain silica interactions. Because the COOH silica particles have a higher surface charge density, a higher interaction energy barrier (DLVO interaction potential energy) was expected for the NR−COOH silica interaction and therefore less heteroaggregation was expected compared to the case of NR−plain silica.29 III.C. Effect of Concentration of NR and Filler on Their Interaction. The concentration of the two binary components, NR and filler, can be important in the growth of aggregates. Thus, we varied the concentration of the silica particles while keeping the NR concentration constant. Figure 6 displays the structural morphological evolution of the binary mixture of NR and silica particles (d = 100 nm, plain) in the presence of Mg2+ ions at 100 mM ionic strength as a function of the volume percent ratio (dry silica/dry NR). We observed that at low silica concentrations (∼≤5%) little or no interaction between the NR and silica particles occurred; both NR and silica particles were seen to be independently and homogeneously dispersed in the substrate. A stronger interaction was observed at volume ratio of ∼10%, where the interaction yields two different heteoaggregate structures: a raspberry-like and dendritic morphologies. A complete structure/phase transformation was seen at volume ratios of ∼20 and ∼50%. At these concentrations, we observed only the dendritic morphology of the NR−silica particle heteroaggregates. If the silica concentration continued to increase, then the interaction between the NR and silica continued to exist, but the result suggested that the dendritic clusters were less frequently observed. III.D. Macroscopic Visualization. Finally, we assessed whether the above-reported nanoscopic behaviors correlate with the macroscopic visualization of heteroaggregate formation. The results obtained for the NR−plain silica suspensions (d = 100 and 30 nm) under three different liquid conditions (in water, NaNO3, and MgSO4 at 100 mM ionic strength) are displayed in Figure 7. In the absence of salts, we observed a turbid solution, indicating that there is no interaction between the binary particles. After several weeks in the absence of interaction, the water suspension containing the NR and silica particles separated into three fractions: (i) the bottom pink sediments correspond to the fluorescent fillers, (ii) the top floating milky white liquid corresponds to the natural rubber particles, and (iii) between the two fractions is the supernatant. In the presence of Na+ ions, the suspension remained slightly turbid or there was no distinct separation among the suspension, silica, and NR even after several days. In Mg2+ solution, we observed precipitation of the NR−silica heteroaggregates. Remarkably, the aggregation kinetics was rapid (∼3−5 min), in agreement with the results obtained from the nanoscale measurements. Similar results were obtained with the 30 nm silica particles. In addition to the interaction and aggregation rates, we also observed the ability of these NR− silica precipitates to be redispersed through external agitation (i.e., mechanical vortex or hand mixing). These precipitates would reflocculate again if left to stand undisturbed. The time

the dendritic morphology was no longer observed, as shown in Figure 5. The 30 nm silica particles were not intercalated

Figure 5. (a) TEM micrograph and (b) fluorescence intensity image of the NR−30 nm plain silica particle suspension after 2 h of interaction in a MgSO4 suspension at 100 mM ionic strength.

between two or several NR particles, but, instead, they surround the NR particles entirely, thus resembling a raspberry-like structure (Figure 5a). This full surface coverage by the 30 nm silica particles resulted in a homogeneously fluorescent image, as shown in Figure 5b. For the mixed NR− 50 nm plain silica in Mg2+ suspension, both dendritic and raspberry-like morphologies were observed (data not shown). We also described how functionalization of the surface of the silica particles with carboxylate groups could influence their properties and interaction dynamics with NR particles. First, without NR, the COOH silica particles (d = 100 nm) remained monomeric in both ionic suspensions (NaNO3 and MgSO4) at 100 mM ionic strength, as revealed by FCS measurements (results not shown), unlike its plain silica counterpart. This was expected because electrophoretic mobility measurements in water revealed that when silica particles are functionalized they become more negatively charged (μ = ∼−2.6 μm cm/(V s) for plain silica particles and μ = ∼−4.0 μm cm/(V s) for COOH silica particles). For the mixed NR−COOH silica suspension, results from the FCS measurement showed an interaction between the two particles only in the presence of ions, as was observed with the plain silica particles. The mean diffusion time extracted from the autocorrelation curve at half amplitude of the autocorrelation function was similar to the values reported for NR−silica 12442

DOI: 10.1021/acs.langmuir.5b03244 Langmuir 2015, 31, 12437−12446

Article

Langmuir

Figure 6. Confocal fluorescence intensity images of the heteroaggregation process and structural evolution of the NR−100 nm plain silica complex with increasing volume percent ratio (dry silica/dry NR) in a MgSO4 suspension at 100 mM ionic strength. The concentration of silica increases, and the NR concentration was kept constant.

Figure 7. Photographs of mixed solutions of NR−100 nm plain silica in water, NaNO3, and MgSO4 at 100 mM ionic strength. Aggregation was observed only in the presence of ions. Similar results were obtained for NR−30 nm plain silica suspensions. NR concentration: ∼0.7 wt %; volume percent ratio (dry silica/dry NR): 20%.

scopic mechanisms involved in the process of aggregate formation have remained a mystery, which, if solved, could open the possibility of regulating reinforcement effects. Our study, for the first time, tackles this intriguing question and deciphers the important physicochemical parameters that influence the aggregation dynamics prior to dehydration as a function of the chemical nature, size, and concentration of the filler as well as the composition of the solution (ions and ion valence) under dilute conditions. The results are summarized in Table 2. IV.A. NR−NR and Silica−Silica Homoaggregation vs NR−Silica Heteroaggregation. Natural rubber is a hydro-

scale for the reflocculation was similar to that presented in Figure 7.

IV. DISCUSSION As mentioned in the Introduction, different mixing procedures are constantly being developed to incorporate and improve filler dispersion in the NR matrix. In the liquid route, the NR and filler are first dispersed in an aqueous suspension and then the liquid is dried-out, forcing the particles to coalesce and form nanocomposite films. At the moment, the limited available literature is concerned with the properties of the film (final product). The underlying interactions and, hence, the micro12443

DOI: 10.1021/acs.langmuir.5b03244 Langmuir 2015, 31, 12437−12446

Article

Langmuir

Table 2. Homo- and Heteroaggregation of Natural Rubber and Plain Silica Particles as a Function of Silica Size in NaNO3 and MgSO4 Salt Suspensionsa,b,c plain silica (d, nm) NR 30 nm plain silica 50 nm plain silica 100 nm plain silica

NR

30

50

100

isolated particles and aggregates raspberry raspberry and dendritic dendritic

raspberry isolated particles n/a n/a

raspberry and dendritic n/a isoated particles n/a

dendritic n/a n/a aggregates

a Ionic strength, 100 mM. Heteroaggregation is ∼2 times faster in MgSO4 than it is in NaNO3 suspenion. bNR−COOH silica particle interactions yield similar results, but the reaction is not always guaranteed to occur. cIn water, the separate suspensions of NR and all sizes of silica particles remained monomeric (isolated particle) and there is no interaction between NR and silica particles.

particles, as opposed to dry conditions where the interaction is heavily influenced by the hydrophobic polymer (cis-polyisoprene). IV.B. NR−Silica Heteroaggregation Kinetics. The composition of the NR biomembrane is one of the factors that could influence the NR−silica heteroaggregation dynamics. Visual comparison of the confocal fluorescence intensity images captured for the NR−silica aggregates in NaNO3 suspensions after 2 and 4 h of interaction (Figure 3a,b) indicated that NR− filler aggregation most probably started with small rubber particles (SRPs). This can be related to the smaller amount of lipids in SRPs than in LRPs.27 In liquid, NR was stabilized partly by lipids due to the negative charge of the phosphate groups. Therefore, charge neutralization by the ions and hence the aggregation would be more favorable for SRPs (lower lipid concentration). Another factor that modulated the NR−silica heteroaggregation kinetics is the ion valency. Nanoscopically, the interaction of NR and silica in Mg2+ suspensions is ∼2−4 times faster than that in Na+ suspensions. This was again consistent with their macroscopic behavior. The difference in the aggregation kinetics could not be explained by comparing their respective EDL because, at identical ionic strength, the compression of the Debye length would be similar. A more plausible explanation should be based on the physicochemical effect relating to the cation valence. Aside from their influence in decreasing the interaction energy barrier, the ions (Na+ and Mg2+) can also act as electrostatic bridges between the surfaces of adjacent particles,33 with divalent ions being able to form two bonds, whereas only a single bond can be formed with monovalent ions. Therefore, the rate of aggregate formation would be faster in MgSO4 than in NaNO3. Additional evidence for this assumption was based on the similarity of heteroaggregation behavior when the ions were changed to K+ and Ca2+. These results demonstrated the capability of regulating the interaction kinetics using ions, which is advantageous, especially in systems involving several compounds that must chemically react at appropriate points in time during processing. IV.C. Modulating NR−Silica Heteroaggregates. Another highlight of this study is the dependence of the NR−silica aggregates’ morphology on the size ratio of the two particles: the 30 nm silica particles adsorbed and entirely covered the surface of the NR particles, forming a raspberry-like structure, whereas for the 100 nm silica particles, dendritic clusters were formed instead. Furusawa et al.34 and Yates et al.35 have shown similar results on their studies with silica/polymeric lattices. They explained this behavior based on the effect of the size ratio of the two binary particles on the surface coverage area. Specifically, they calculated that the number of particles

phobic core of polyisoprene surrounded by a thin (∼10−20 nm) hydrophilic layer principally composed of protein− phospholipid complexes.28,30 Under the studied pH conditions (∼5.7−6), when no ions were present, the biomembrane of the NR particles remained significantly negatively charged, thereby maintaining the colloidal stability of the NR latex.31 In the presence of Na+ and Mg2+ ions at 100 mM ionic strength, the magnitude of the electrostatic repulsion was reduced due to the screening of the electrostatic double layer (EDL). Thus, even under dilute conditions, as was done in this study, the NR particles can form aggregates. This was also observed for the 100 nm plain silica particles but not for the 30 and 50 nm particles (the particles remained monomeric). For the mixed NR and plain silica suspensions in the presence of ions, nanoscopically, the FCS autocorrelation curves were stabilized. Macroscopically, the formation of NR− plain silica precipitates showed not only an interaction between the NR and all sizes of plain silica particles but also the dominance of heteroaggregation over homoaggregation (silica− silica and NR−NR interaction). Because both NR and silica particles are negatively charged, heteroaggregation was observed only in the presence of ions. In an attempt to further reduce the filler−filler interaction, we used COOH-functionalized silica particles, which are more negatively charged than the plain silica particles (electrophoretic mobility of COOH-modified silica particles is ∼−4.0 μm cm/(V s), whereas it is ∼−2.6 μmcm/(V s) for plain silica particles). Another point of interest for using carboxyl groups is that in dry NR−filler technology the presence of chemical groups (i.e., carboxyl, phenol, etc.) at the surface of the fillers could facilitate interaction.32 Indeed, we show that the presence of carboxylate groups significantly reduced filler−filler interactions (at 100 mM ionic strength). Unfortunately, however, it also decreased the possibility of NR−silica interactions, indicating that using fillers with excess negative charge is not the best approach. Perhaps, in liquid, a more appropriate solution is to use a highly positively charged filler to promote NR−filler interactions while preventing filler−filler interactions or to attach neutral molecules (hydrophilic polymers) to the fillers that have a high affinity to NR. We also tested if heteroaggregation could be improved by replacing silica particles for polystyrene particles (PS, carboxylate-modified) in which the hydrophobic polymer is exposed or accessible to the environment. The results from FCS measurements and macroscopic aggregation assays showed that the PS particles did not interact at all with NR particles in all three suspensions (water, NaNO3, and MgSO4; Supporting Information 2). These results, therefore, revealed that NR−filler interactions in liquid are mainly controlled by the hydrophilic biomembrane (proteins and lipids) of the NR 12444

DOI: 10.1021/acs.langmuir.5b03244 Langmuir 2015, 31, 12437−12446

Langmuir

■

required to cover the surface of another particle decreases as the particle size ratio increases: for a size ratio ∼1, one silica particle can adhere only to 3−4 latex particles (and vice versa). A closer inspection of our TEM micrographs confirmed this finding: NR particles of 100−200 nm in diameter were surrounded by more than 20 silica particles of 30 nm diameter, but for the 100 nm silica particle, only 3−5 particles surrounded the NR. In the latter, large open spaces between the silica particles exist and thus favored the formation of irregularly shaped heteroaggregates (i.e., dendritic clusters). We find the 50 nm silica to be the size at which both dendritic and raspberry-like structures were observed. This result was interesting when considering the prospects for its applications, as it demonstrated the possibility of tuning the structure of heteroaggregates. In this context, another point of interest is to show that varying the concentration (volume percent ratio silica/NR) could also modulate the formation of dendritic clusters. Here, the concentration of the 100 nm silica particles was changed while keeping the NR concentration constant. Dendritic clusters were observed only at ratios of 20−50%. Decreasing these ratios and thus the silica particle concentration diminished the frequency of collisions with NR particles, resulting in little or no interaction. In contrast, an increase in the silica particle concentration resulted in filler networking (silica−silica interaction) and therefore also decreased NR− filler interactions.

AUTHOR INFORMATION

Corresponding Author

*Phone: +33 652780554. E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors thank ANRT for CIFRE funding (2011/1324). This work was supported by grants from Michelin. We also thank Catherine Le Bris, Julien Vincent, Christophe Charrière, and Marc Hilaire for their technical assistance with the upgrade of the FCS setup. We also thank Christine Longin for her expertise and access to equipments for TEM. We also express our gratitude to Andrew Mayne and Cesare Mikhail Cejas for their english revision of the manuscript.

■

REFERENCES

(1) Cornish, K. Biochemistry of Natural Rubber, a Vital Raw Material, Emphasizing Biosynthethic Rate, Molecular Weight and Compartmentalization, in Evolutionary Divergent Plan Species. Nat. Prod. Rep. 2001, 18, 182−189. (2) Rippel, M. M.; Leite, C. A. P.; Galembeck, F. Elemental Mapping in Natural Rubber Latex Films by Electron Energy Loss Spectroscopy Associated with Transmission Electron Microscopy. Anal. Chem. 2002, 74, 2541−2546. (3) Hashim, A. S.; Kawabata, N.; Kohjiya, S. Silica Reinforcement of Epoxidized Natural Rubber by the Sol-Gel Method. J. Sol-Gel Sci. Technol. 1995, 5, 211−218. (4) Sarkawi, S. S.; Dierkes, W. K.; Noordermeer, J. W. M. Elucidation of Filler-Filler and Filler-to-Rubber Interactions in Silica-Reinforced Natural Rubber by TEM Network Visualization. Eur. Polym. J. 2014, 54, 118−127. (5) Mermet-Guyennet, M. R. B.; Gianfelice de Castro, J.; Varol, H.; Habibi, M.; Hosseinkhani, B.; Martzel, N.; Sprik, R.; Denn, M. M.; Zaccone, A.; Parekh, S. H.; Bonn, D. Size-Dependent Reinforcement of Composite Rubbers. Polymer 2015, 73, 170−173. (6) Edwards, D. C. Review Polymer-Filler Interactions in Rubber Reinforcement. J. Mater. Sci. 1990, 25, 4175−4185. (7) Wolff, S. Chemical Aspects of Rubber Reinforcement by Fillers. Rubber Chem. Technol. 1996, 69, 325−346. (8) Frohlich, J.; Niedermeier, W.; Luginsland, H.-D. The Effect of Filler-Filler and Filler-Elastomer Interaction on Rubber Reinforcement. Composites, Part A 2005, 36, 449−460. (9) Heinrich, G.; Kluppel, M.; Vilgis, T. A. Reinforcement of Elastomers. Curr. Opin. Solid State Mater. Sci. 2002, 6, 195−203. (10) Manas-Zloczower, I. Mixing and Compounding of Polymers: Theory and Practice, 2nd ed.; Carl Hanser Verlag GmbH and Co. KG: Munich, 2009. (11) Nguyen, V.; Yoshida, W.; Cohen, Y. Graft Polymerization of Vinyl Acitate on Silica. J. Appl. Polym. Sci. 2003, 87, 300−310. (12) Suzuki, K.; Siddiqui, S.; Chappell, C.; Siddiqui, J. A.; Ottenbrite, R. M. Modification of Porous Silica Particles with Poly (Acrylic Acid). Polym. Adv. Technol. 2000, 11, 92−97. (13) Hubner, E.; Allgaier, J.; Meyer, M.; Stellbrink, J.; PyckhoutHintzen, W.; Richter, D. Synthesis of Polymer/Silica Hybrid Nanoparticles Using Anionic Polymerization Techniques. Macromolecules 2010, 43, 856−867. (14) Nah, C.; Huh, M. Y.; Rhee, J. M.; Yoon, T. H. Plasma Surface Modification of Silica and its Effect on Properties of Styrene-Butadiene Rubber Compound. Polym. Int. 2002, 51, 510−518. (15) Oberdisse, J. Aggregation of Colloidal Nanoparticles in Polymer Matrices. Soft Matter 2006, 2, 29−36. (16) Grossiord, N.; Loos, J.; Regev, O.; Koning, C. E. Toolbox for Dispersing Carbon Nanotubes into Polymers to Get Conductive Nanocomposites. Chem. Mater. 2006, 18, 1089−1099. (17) Grunlan, J. C.; Mehrabi, A. R.; Bannon, M. V.; Bahr, J. L. WaterBased Single-Walled-Nanotube-Filled Polymer Composite with an

V. CONCLUSIONS From the perspective of applications of the current work, a fundamental step in NR reinforcement is the improvement of the interactions between NR and the filler particles. To our knowledge, our results are the first to report on the physical processes involved in the interaction of NR and silica under aqueous conditions using systematically varying parameters (Table 2). NR and silica interactions are electrolyte-responsive, where the kinetics increases with the ion valency. Such interactions also occur only at critical volume percent ratios (∼20−50%). Concerning the morphology of the heteroaggregates, it is strictly dependent on the size ratio between NR and filler particles, with dendritic clusters being observed for particles of comparable sizes. The parameter that most distinguishes these results compared to those under dry conditions is that the interaction of NR in liquid is mainly controlled by its biosurface (hydrophilic character, lipid concentration, etc.). Overall, the good agreement between the nano- and macroscopic interaction behaviors validates and supports the integrity of all of the results presented.

■

Article

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.langmuir.5b03244. Normalized autocorrelation curves of mixed suspensions of COOH polystryrene in the absence and presence of NR (PDF) Movie showing the NR−100 nm plain silica heteroaggregation process in NaNO3 and MgSO4 suspensions (AVI) 12445

DOI: 10.1021/acs.langmuir.5b03244 Langmuir 2015, 31, 12437−12446

Article

Langmuir Exceptionally Low Percolation Threshold. Adv. Mater. 2004, 16, 150− 153. (18) Martinez-Pedrero, F.; Alousque, F.; De Gaudemaris, B.; Berriot, J.; Gaboriaud, F.; Bremond, N.; Bibette, J. Making an Elastomeric Composite Material via the Heteroaggregation of a Binary Colloidal Dispersion. Soft Matter 2012, 8, 8752−8757. (19) Klüppel, M.; Schuster, R. H.; Heinrich, G. Structure and Properties of Reinforcing Fractal Filler Networks in Elastomers. Rubber Chem. Technol. 1997, 70, 243−255. (20) Chen, Y.; Müller, J. D.; Berland, K. M.; Gratton, E. Fluorescence Fluctuation Spectroscopy. Methods 1999, 19, 234−252. (21) Krichevsky, O.; Bonnet, G. Fluorescence Correlation Spectroscopy: The Technique and Its Applications. Rep. Prog. Phys. 2002, 65, 251−297. (22) Schwille, P.; Haustein, E. Fluorescence Correlation Spectroscopy: An Introduction to Its Concepts and Applications; Biophysics Textbook Online: Gottingen, Germany, 2002. (23) Briandet, R.; Lacroix-Gueu, P.; Renault, M.; Lecart, S.; Meylheuc, T.; Bidnenko, E.; Steenkeste, K.; Bellon-Fontaine, M. N.; Fontaine-Aupart, M. P. Fluorescence Correlation Spectroscopy to Study Diffusion and Reaction of Bacteriophages Inside Biofilms. Appl. Environ. Microbiol. 2008, 74, 2135−2143. (24) Habimana, O.; Steenkeste, K.; Fontaine-Aupart, M. P.; BellonFontaine, M. N.; Kulakauskas, S.; Briandet, R. Diffusion of Nanoparticles in Biofilms is Altered by Bacterial Cell Wall Hydrophobicity. Appl. Environ. Microbiol. 2011, 77, 367−368. (25) Cornish, K.; Siler, D. J.; Grosjean, O. K.; Goodman, N. Fundamental similarities in rubber particle architecture and function in three evolutionary divergent plant species. J. Nat. Rubber Res. 1993, 8, 275−285. (26) Winsunthorn, S.; Bonfils, F.; Pochat-Bohatier, C.; Bouyer, D.; Deratani, A.; Dupuy, C. Comparative study of the elasticity and permeability of vulcanized films made with skim and cream natural rubber latex. J. Appl. Polym. Sci. 2008, 108, 960−968. (27) Chan, A. J.; Steenkeste, K.; Eloy, M.; Brosson, D.; Gaboriaud, F.; Fontaine-Aupart, M. P. Lipid Content in Small and Large Natural Rubber Particles. Rubber Chem. Technol. 2015, 88, 248−257. (28) Gaboriaud, F.; De Gaudemaris, B.; Rousseau, T.; Derclaye, S.; Dufrene, Y. F. Unravelling the Nanometre-Scale Stimuli-Responsive Properties of Natural Rubber Latex Particles using Atomic Force Microscopy. Soft Matter 2012, 8, 2724−2729. (29) Guo, D.; Xie, G.; Luo, J. Mechanical Properties of Nanoparticles: Basics and Applications. J. Phys. D: Appl. Phys. 2014, 47, 013001. (30) Cornish, K.; Wood, D. F.; Windle, J. J. Rubber particles from four different species examined by transmission electron microscopy and electron-paramagenetic-resonance spin labeling, are found to consist of a homogeneous rubber core enclosed by a contiguous, monolayer membrane. Planta 1999, 210, 85−96. (31) Rochette, C. N.; Crassous, J. J.; Drechsler, M.; Gaboriaud, F.; Eloy, M.; de Gaudemaris, B.; Duval, J. F. L. Shell structure of natural rubber particles: evidence of chemical stratification by electrokinetics and cryo-TEM. Langmuir 2013, 29, 14655−14665. (32) Rashid, A. A.; Yahya, S. R. In Natural Rubber Materials: Composites and Nanocomposites; Thomas, S., et al., Eds.; 8; Royal Society of Chemistry: Cambridge, UK, 2014; Vol. 2, Chapter 19, pp 550−573. (33) Blackley, D. C. High Polymer Lattices; McLaren and Sons: London, 1966; Vol. 2. (34) Furusawa, K.; Anzai, C. Heterocoagulation Behaviour of Polymer Latices with Spherical Silica. Colloids Surf. 1992, 63, 103−111. (35) Yates, P. D.; Franks, G. V.; Biggs, S.; Jameson, G. J. Heteroaggregation with Nanoparticles: Effect of Particle Size Ratio on Optimum Particle Dose. Colloids Surf., A 2005, 255, 85−90.

12446

DOI: 10.1021/acs.langmuir.5b03244 Langmuir 2015, 31, 12437−12446