VULCANIZATION FOR REINFORCEMENT OF RUBBER

The two critical issues governing properties of rubber vulcanizates are sulfur cross-linking and filler reinforcement. Although many studies have been conducted to gain a better understanding of the two issues,^1–5  there are still unknown aspects about vulcanization (sulfur cross-linking) and the filler–fillings for rubbers. Therefore, to achieve a paradigm shift of rubber science and technology, it is necessary to conduct more in-depth research on the essential nature of sulfur cross-linking and reinforcement. Except for a few special cases, such as rubber adhesives, cross-linking is a requirement for raw rubber to be serviceable as an elastic solid material by forming a three-dimensional rubber network structure. Even when highly reinforcing nanoparticles are mixed with rubber, nearly all rubber products cannot demonstrate a stable elasticity to make them useful without any cross-linking.⁵ 

According to Kuhn's discussion of the term paradigms,⁶  we may assume that the vulcanization paradigm was established in the late 1970s.^4,7  Since the technology has matured, rubber scientists and engineers have usually focused on modifying and/or improving existing systems rather than developing new curing methods. Based on the established vulcanization paradigm, one can now select proper reagents for the vulcanization process. However, there are still unknown aspects of rubber vulcanization. Therefore, it is now necessary to conduct more in-depth research on the essential nature of cross-linking, especially sulfur cross-linking, to enhance the role of the reinforcing filler for rubber products.

To date, our group has revealed the effects of the combination and composition of sulfur cross-linking reagents on the network formation in isoprene rubber (IR). The combination of zinc oxide (ZnO) with other reagents was interestingly found to be crucial not only for chemical cross-linking of the rubber molecules but also for controlling structural network inhomogeneity in the sulfur cross-linked IR using small-angle neutron scattering (SANS).⁸  A two-phase inhomogeneous network structure was proposed for the sulfur cross-linked IR, as shown in Figure 1, where there are network domains of high network-chain density embedded in the rubber network matrix (the mesh network). The network domains were also reported to be generated around ZnO particles. In addition, the amounts of ZnO and stearic acid (StH) were found to control the mesh size in the two-phase inhomogeneous structure. In situ time-resolved zinc K-edge X-ray absorption fine structure (XAFS) spectroscopy also indicated the formation of the two-phase network.⁹  Notably, two distinct sulfur cross-linking reactions for mesh and network-domain formation occur nearly simultaneously. Furthermore, the formation of network domains was confirmed using atomic force microscopy (AFM) nanomechanical mapping with the two-dimensional mapping of the reliability index.¹⁰  Several studies on the inhomogeneous structures of vulcanizates by AFM have been also reported.^11–14 

We discovered a new intermediate generated from ZnO and StH at high temperatures around the vulcanization temperature using a combination of in situ time-resolved zinc K-edge XAFS spectroscopy and in situ time-resolved Fourier-transform infrared spectroscopy.¹⁵  It was a bridging bidentate zinc/stearate complex. The complex has a surprising zinc/stearate ratio of 1. The density functional theory calculations for identifying the intermediate suggested that its fundamental skeleton is a dinuclear-type bridging bidentate zinc/stearate complex composed of [Zn₂(μ-O₂CC₁₇H₃₅)₂]²⁺·4X, where X is the hydroxyl group, water and/or rubber segment (intermediate I). The intermediate I has been unknown despite the long history of rubber science and technology. According to the ACS News Service Weekly PressPac, our findings were expected to be useful for producing important rubber materials, especially ecofriendly tires.¹⁶  This aspect may relate to the concepts of numbers 7, 9, 11, and 12 of the sustainable development goals,¹⁷  because the findings, if used to improve tire performance, for example, could mean higher gas mileage for consumers and less air pollution.

The combined experimental and computational investigation also proposed roles for the intermediate I and its derivatives in the sulfur cross-linking reaction.¹⁸  The novel zinc/stearate complexes accelerate the sulfur cross-linking of IR like an enzyme. Furthermore, when ZnSt₂¹⁹ or a combination of ZnO and StH²⁰  were used as activators, the sulfidic linkage of a disulfide type was dominant in the N-(1, 3-benzothiazol-2-ylsulfanyl)cyclohexanamine (CBS)–accelerated system. Time-domain nuclear magnetic resonance spectroscopy and AFM were used to attribute not only the sulfur cross-linking but also the formation of the highly homogeneous network structure to the function of the intermediate I.²¹  These findings are significant in rubber science and technology because specific network structures contribute to preparing high-performance rubber materials.

In this study, we show how network inhomogeneity affects the reinforcement of CBS-accelerated IR vulcanizates prepared in the presence of ZnO as an activator. Using AFM, we first evaluate network–domain formations and their continuous structures, termed the network-domain cluster and network-domain network. The formations of these structures are then discussed by combining the mechanical characteristics of the vulcanizates from the viewpoint of the reinforcement effect. Recently, we partially and briefly presented this content at the 200th Technical Meeting of the Rubber Division.²²  Thus, in this new article, the study on vulcanization for reinforcement of rubber is described in detail. The findings provide a practical hint for producing high-performance rubber materials. The novel insights will be useful in achieving breakthroughs in the traditional, yet indispensable, technique of vulcanization.

According to the recipes shown in Table I, the rubber compounds were prepared by conventional mixing at room temperature on a two-roll mill with a water-cooling system (6 ×15 test roll; Kansai Roll Co., Ltd., Osaka, Japan). Isoprene rubber (IR2200) was supplied from JSR Co. (Tokyo, Japan). Elemental sulfur (S₈, powder, 150 mesh), ZnO (average diameter, 0.29 μm), and CBS (Sanceler CM-G) were commercial grades for rubber processing and used as received. They were supplied from Hosoi Chemical Industry Co., Ltd. (Tokyo, Japan); Sakai Chemical Industry Co., Ltd. (Osaka, Japan); and Sanshin Chemical Industry Co., Ltd. (Yamaguchi, Japan), respectively. The rubber compounds obtained were individually cured under heat pressing at 144 °C in a mold to make cross-linked rubber sheets of approximately 1 mm thickness. The press-heating times to prepare vulcanizates using the activator of ZnO were determined from the maximum torques shown in the cure curves (Figure 2) obtained at 144 °C using the JSR CURELASTOMETER III. To control the network-chain density of S-IR-4, the heat-pressing time of this sample was determined from 80% value of its maximum torque.

Three specimens of the vulcanizate, approximately 3 mm × 4 mm × 1 mm in size, were swollen in a large amount of toluene at 25 °C for 48 h to their equilibrium swelling, and each volume change was measured using a CCD camera (VC1000 Digital Fine Scope, OMRON Co., Kyoto, Japan). A network-chain density (ν) of the vulcanizate was determined by the modified Flory−Rehner equation.²³  The Flory−Huggins interaction parameter (χ) used in the calculation was 0.427 +0.112ϕ_R², where ϕ_R is the volume fraction of the vulcanizate in the swollen sample.^24,25  The degree of equilibrium swelling by volume, Q, of the vulcanizate was calculated using the equation [Q = (V_s/V₀)], where V₀ and V_s are the volumes before and after swelling, respectively. An average of network-chain densities of three specimens for each vulcanizate are shown in Table I.

The AFM measurement for phase imaging was performed using a commercial AFM system (SPM–9700, Shimadzu, Kyoto, Japan) with a tapping mode under ambient conditions, referring our previous report.¹⁰  A cantilever used in this mode was made of monolithic silicon (NCHR, NANOWORLD, Neuchâtel, Switzerland) with a nominal spring constant of 42 N/m and a tip radius less than 8 nm. The scan sizes of the phase images were 5 μm × 5 μm and 2 μm × 2 μm, and their resolutions were 256 × 256 pixels. The vulcanized rubber sheets were cut by an ultra–microtome (MT–XL CR–X, RMC Boeckeler Instruments, Inc., Tucson, AZ, USA) at approximately –80 °C to prepare the flat surface.

For the AFM nanomechanical mapping, the force volume measurement was conducted by operating with an applied force of 1.4 nN. The loading speed was set at 6720 nm/s. Cantilevers (OMCLRC800PSA43) were supplied by Olympus Co. (Tokyo, Japan) and were made of SiN. The nominal spring constant of 0.76 N m⁻¹ was calibrated by Sader's method²⁶  before the measurement. The nominal tip radius was 15 nm, which was used to calculate the Young's modulus. Force–distance curves were randomly collected over the selected surfaces area of 2 μm × 2 μm with a resolution of 64 × 64 pixels. Each force–distance curve was converted to the force–deformation (F-δ) curve according to the method described in our previous work.¹⁰  All experimental withdrawing force–deformation curves were analyzed by using the JKR contact mechanics model.²⁷  For the analysis, we adopted the “linearized fitting” method proposed by Nakajima et al.²⁸  The AFM measurements for phase imaging and nanomechanical mapping were conducted at least three times for each sample to confirm the representativeness of their results. Note that one typical phase image and one typical Young's modulus mapping image for each are displayed.

Dynamic mechanical analysis was carried out, in which the temperature dependence of the storage modulus (E′), loss modulus (E″), and loss tangent (tan δ) of the vulcanizates were measured on a dynamic mechanical analyzer (Rheospectolar DVE-4, UBM Co., Kyoto, Japan) at a frequency of 10 Hz. The strain mode was used, and the heating rate was 2 °C/min in a range from –130 °C to 150 °C. The sample size was approximately 20 mm × 5 mm × 1 mm.

Differential scanning calorimetry (DSC) measurements were obtained using the Rigaku Thermoflex (DSC-8230) under nitrogen at a heating rate of 10 °C/min in a range from –130 °C to 150 °C. The amount of the sample mass was about 10 mg encapsulated in an aluminium pan.

Tensile measurements were carried out on ring-shaped samples using a tailor-made tensile tester (ISUT-2201, Aiesu Giken, Co., Kyoto, Japan) at room temperature (approximately 25 °C). The outside and inside diameters of the ring-shaped samples were 13.7 and 11.7 mm, respectively. The stretching speed was 100 mm/min (i.e., the strain speed was about 4.98 min^–1). Three specimens of each vulcanizate were subjected to the measurement.

The AFM method is a useful technique that enables us to obtain a nanoscaled visualization of the material surface properties.^29–31  In polymer materials science, AFM has been used extensively to evaluate the morphological characteristics of plastics, rubbers, and fibers, because several polymeric materials are observed to possess inhomogeneous structures on a nanometer scale.^30,32,33  Thus, the three samples prepared with ZnO activator were first measured. The AFM phase images of the vulcanizates using different amounts of CBS and sulfur are shown in Figure 3, with a color change from blue to red indicating the changes from a large phase angle to a small phase angle. Note that Figure 3 shows typical images of each sample obtained by the measurement three times. Good reproducibility of the results was confirmed. The areas of smaller phase angles than the rubber matrix are clearly observed in each vulcanizate, as shown in the phase images of this figure. Generally, a large phase angle shows a soft part of a large energy absorption between a sample and a cantilever. Therefore, we detected inhomogeneous network structures on the surfaces of these samples. As predicted, the two-phase network structures consist of a hard phase (red, orange, and yellow colors) in a soft matrix (blue) in each vulcanizate. Most of the hard parts in S-IR-1 were distorted circle and/or elliptic forms and well dispersed throughout the soft matrix. However, the hard parts in the matrix of S-IR-2 appear in long irregular shapes and in small spherical shapes, which are located between the long irregular phases. Notably, such irregular and small hard parts tend to connect to form a specific structure like a network in S-IR-3. An imperfect network structure is also recognized in S-IR-3. In addition, as the amount of CBS and sulfur increases, the area of the hard phases in the soft matrix also increases. The separation of hard phases in the soft rubber matrix is accelerated in S-IR-3 compared with S-IR-2. This phenomenon is similar to the micro-phase separation of thermoplastic elastomers. Because the same amount of ZnO was added to each sample, a factor promoting the formation of continuous hard phases in this study is mainly attributed to the amounts of sulfur cross-linking reagents in the IR matrix.

These findings are consistent with our previous SANS results.⁸  As described in the introduction, we found that increasing the amounts of sulfur and accelerator (CBS) increases the size of the network domain in the two-phase network structure of the vulcanizates. This finding was explained by the absorption of sulfur and CBS onto ZnO particles,³⁴  which was then followed by a sulfur cross-linking reaction around each ZnO particle. Therefore, the degree of sulfur cross-linking reaction around the ZnO particles became larger by increasing the amounts of sulfur and CBS, which increased the network-domain size and network-chain density. Thus, the specific hard phases in the matrix of S-IR-2 and S-IR-3 are assumed to be formed by overlapping each network domain as the amount of CBS and sulfur increases. In addition, the soft phases shown in blue are more visible in S-IR-3 than in S-IR-2. This result implies that increasing the amounts of CBS and sulfur mildly promotes the aggregation of ZnO particles in the rubber matrix. Consequently, the formation of clusters and networks composed of the network domains is accelerated, and a significant phase separation between the soft and hard areas occurs. Generally, it is difficult to quantitatively compare the difference in phase images among samples³¹ ; thus, AFM nanomechanical mapping was performed next to support this consideration.

Young's modulus measurement has been used to perform quantitative visualization of the network structures using AFM nanomechanical mapping.^11,32  Generally, this is a powerful tool for quantitatively comparing the surface properties among samples. This technique has especially been used to investigate filler dispersion and the interface between hard fillers and soft rubber phases.^14,35  The AFM nanomechanical mapping gave us reasonable Young's modulus mapping images with nearly good reproducibility for each sample. Typical Young's modulus images of the vulcanizates are shown in Figure 4d–f. The specific morphology in the Young's modulus mapping image of each sample corresponded well with that obtained from phase imaging (Figure 4a–c), where the phase images are also shown at a scan size of 2.0 μm. The R-factor_(AFM) images of the vulcanizates are also shown in Figure 4g–i. Note that the use of different colors from yellow to black in the figures indicates the accuracy of the measurement from good to bad, respectively. We found that most R-factors_(AFM) are significantly low in all images, which means that the Young's moduli calculated by the linearized fitting method²⁸  based on the JKR model²⁷  are sufficiently reliable and can be used for comparison of samples. The histograms of the R-factor_(AFM) images supported this consideration. These results of the AFM nanomechanical mapping clearly show inhomogeneity in S-IR-1, S-IR-2, and S-IR-3. All samples can be roughly divided into three areas: low, middle, and high Young's modulus areas. The low Young's modulus area of approximately 2 MPa (blue) in Figure 4d–f is attributed to a mesh network matrix phase. The high Young's modulus (≥15 MPa) areas shown in red are ascribable to the presence of ZnO particles and/or the thin rubber-covered ZnO particles in the vulcanizates. The intermediate area is recognized as mainly located between the soft rubber matrix and the hard phases, including the ZnO particles. Therefore, this intermediate area includes thick rubber-covered ZnO particles and/or the rubber network phases with higher network-chain densities than that of the matrix.

This discussion is supported by our previous study,¹⁰  in which the effects of the rubber layers existing above and below the ZnO particles on the measured Young's moduli were named the blanket effect and mattress effect, respectively. The soft rubber phase's blanket and mattress effects result in a small Young's modulus of ZnO. It is also necessary to recognize that the ZnO particles existing under the soft phase increase the real Young's modulus of the soft phase. This phenomenon is the inverse of the mattress effect and is named the bone effect for convenience in this study. The blanket effect caused by the rubber-covered layer and the bone effect caused by the remaining ZnO particles were observed more frequently in S-IR-2 than in S-IR-3. This might be due to two factors. The first is a better dispersion of sulfur cross-linking reagents in the former than in the latter, which may increase the fraction of rubber-layered phases in S-IR-2. The second factor is the smaller progress of the sulfur cross-linking reaction due to the smaller amounts of reagents of CBS and sulfur in the former than in the latter, which may remain in ZnO in S-IR-2. The red phases in S-IR-3 were the lowest among the samples, and lower Young's moduli were detected in S-IR-3 than in S-IR-2. Histograms of the Young's modulus support the presence of soft matrix phases (Figure 5), where the results of curve fitting using a Gaussian function are represented as dotted lines. The peak top of the Young's modulus of 1.9 MPa is apparently observed as one component in S-IR-1 and S-IR-3. Unfortunately, the component was not detected as one peak in S-IR-2 because of its small fraction. These results imply that the inhomogeneity of Young's moduli in the vulcanizates is ascribed to the generation of the network domain and its continuous phases (clusters and networks) in the soft network matrix. Note that the amount of the remaining ZnO in the network domain and its contentious phases depends on the degree of cross-linking reaction (i.e., the degree of consumption of ZnO in the sulfur cross-linking reaction in this study). The continuous phases are termed the network-domain cluster and network-domain network, respectively. Their schematic figures are illustrated in Figure 6.

Dynamic mechanical and thermal properties of the vulcanizates are satisfactorily explained in terms of the reinforcement effects of network domains and their continuous phases. The dynamic mechanical analysis are shown in Figure 7 and summarized in Table II. We found that increasing the amounts of CBS and sulfur increases the storage moduli (E′) at 25 °C and 100 °C and the temperature of the tan δ peak top, which corresponds to the glass transition temperature (T_g). The higher T_g indicates the lower mobility of the rubber chains caused by the constraint of the larger areas of the network domains and the continuous phases in the inhomogeneous network structures of the vulcanizates. This is consistent with the variation of T_g measured under the static condition by DSC (Table II). Note that, however, the variation of T_g by DSC is not so large, which may be attributed to the presence of similar soft matrices of approximately 2 MPa in all samples. The heights of the tan δ peaks of the three samples are comparable, as shown in Figure 7c, which may support the consideration for T_g by DSC. Therefore, the increase in E′ over T_g shows that the reinforcement effects of the network domains and their continuous phases having higher network-chain densities than the mesh network. Furthermore, we discuss the relationship between the network morphology and the dynamic mechanical properties of the samples.

As shown in Figure 3, the Young's moduli of the soft matrices between S-IR-1 and S-IR-3 are comparable, but the fractions and shapes of the network domains differ. It is also observed in this figure that the formation of continuous network domains occurs more often in S-IR-3 than in S-IR-1 because of the higher amounts of CBS and sulfur in the former than in the latter. Therefore, the T_g and the E′ over T_g of S-IR-3 are higher than those of S-IR-1 as expected, even though the remaining fraction of ZnO is smaller in the former than in the latter. In general, it is well-known that the reinforcement effect of ZnO is significantly smaller than that of carbon black (CB). Specifically, ZnO may not have a reinforcement effect by itself, but ZnO can produce the network domains and its continuous phases depending on the amounts of cross-linking reagents used. Therefore, the ZnO particles at least indirectly act as a reinforcing filler. Interestingly, the reinforcement effects by the network domains and their continuous phases are very similar to the well-known effect by CB. Figure 8 shows the morphology of nanofiller CB⁵ : a fundamental particle as a model in Figure 8a: a primary aggregate of CB covered by bound rubber in rubber matrix, and the ultimately higher aggregate (i.e., agglomerate) are shown in Figure 8b,c, respectively. The CB particles are aggregated, and only the primary aggregates and/or the higher aggregates are present in the rubber matrix with bound rubber. The semiflexible nanofiller CB network involving the bound rubber in the rubber matrix was semiquantitatively found to be effective for reinforcing rubber materials.³⁶  The morphological difference of the CB particles in vulcanizates mixed with different amounts of CB is one of the main factors governing the mechanical properties.^2–5  Thus, similar to the morphology effect of the CB particles, our observations in this study indicate that not only the network-chain density but also the network inhomogeneity due to the morphology of the network domains should be designed to produce excellent rubber materials.

This consideration is supported by the tensile properties of the vulcanizates, which are shown in Figure 9. Nearly good reproducibility for the stress–strain curves of each vulcanizate was detected. It is apparent that the stresses in each strain were high, in the following order: S-IR-3 > S-IR-2 > S-IR-1. Generally, the increases in stresses are discussed in terms of cross-linking density and/or filler content.^2–5  Here, we can add the effects of the network domain and its continuous phases (network-domain clusters and network-domain networks) in the inhomogeneous networks to characterize the tensile properties of the vulcanizates. The network-domain network shows the highest reinforcement effect among the vulcanizates. The samples' upturn stresses will be reported elsewhere, considering the strain-induced crystallization behavior, which is predicted to be significantly influenced by the network morphology.

To better understand the reinforcement effect of network-domain networks, a comparison between S-IR-3 and S-IR-4 is conducted, with the network-chain densities of the two samples set to be comparable (Table I). First, the Young's modulus mapping image of S-IR-4 is shown in Figure 10b with that of S-IR-3 in Figure 10a. A comparison of the histograms of the two samples is also shown in Figure 10e. The network of S-IR-4 is significantly more homogeneous than that of S-IR-3. However, note that the peak top of the Young's modulus of the former is approximately 3 MPa, which is higher than that (1.9 MPa) of the soft matrix of the latter. Furthermore, the dispersion of the Young's modulus of the former is significantly narrow. However, stresses in each strain of the tensile stress–strain curves of two samples are higher in S-IR-3 than in S-IR-4 (Figure 10f). Because the nearly good reproducibility for the stress–strain curves of each vulcanizate was confirmed, this result is explained on the basis of the difference in network morphology discussed in this study. We already detected a similar phenomenon in the cure curves between S-IR-3 and S-IR-4, where the torque of S-IR-3 was found to be significantly higher than that of S-IR-4 (Figure 2). These results imply that we have to pay attention to using only the network-chain density when studying the physical properties of vulcanizates. It is emphasized that the properties of cross-linked rubbers should be discussed using the network-chain density and the network morphology. For filler-filled cross-linked rubbers, therefore, the reinforcement effects not only by the filler filling but also by cross-linking finally have to be taken into account including viewpoints of filler morphology and rubber network morphology for studying the characteristics of the composite materials.

Zinc compounds are important reagents for the vulcanization and reinforcement of rubber. ZnO is revealed as one of the critical keys in controlling network inhomogeneity and the hierarchical structure in vulcanizates. ZnO particles can produce network domains and their continuous phases of the domains by absorbing sulfur and CBS on the surfaces of particles. Thus, by changing the amounts of ZnO and other cross-linking reagents, inhomogeneity and the hierarchical structure of networks can be controlled. The morphology of ZnO and its clusters in the rubber matrix, as predicted in our SANS study in 2009,⁸  can serve as a template for the controls. This aspect is critical for the rubber industry to produce high-performance rubber materials, because of their network domains and continuous phases to reinforce rubber products. In addition, a good reinforcement effect of filler cannot be obtained without an appropriate cross-linked structure. Therefore, the most essential issue for reinforcing rubber materials remains the cross-linking reaction, especially vulcanization. The following first four points below (i)–(iv) and their overall designs are required to produce high-performance rubber materials. The continuous phase of the domains is referred to as the network-domain cluster and the network-domain network in this study.

• Sulfur cross-linking reaction of rubber molecules

• Formation of network domains

• Formation of network-domain clusters

• Formation of network-domain networks

• Formation of filler networks

Note that the points of (ii), (iii), and (iv) are significantly similar to those of reinforcing fillers such as CB. The morphology of the reinforcing filler in a rubber matrix has been identified as key for reinforcing rubber by filler filling, where a fundamental particle of filler, its primary aggregate, and agglomerate (filler network) are present in a rubber matrix.^2–5  It is known that the filler network is the origin of the reinforcement effect of rubber. In terms of filler network formation, it is noticed that the controls of the hierarchical structures of rubber and filler networks are essential for next-generation rubber science and technology. Finally, it is emphasized that the design of five network structures, including (v), is important for reinforcing ubiquitous 21st-century rubber materials to achieve sustainable development goals.¹⁷ 

This work was partially supported by the Japan Science and Technology Agency (JST) ALCA program (grant number JPMJAL1501) and Fuji Seal Foundation for Y.I. and by the Descente and Ishimoto Foundation for the Promotion of Sports Science for K.M. The authors thank Messrs. T. Sato and T. Sugiyama, Ms. Y. Kitada, and Drs. H. Kobayashi and Yuta Sakaki for their support. This study was partially conducted at the Center for Rubber Science and Technology, Kyoto Institute of Technology.