Improved understanding on the reinforcement of low-temperature hydrogenated nitrile butadiene rubber composites by in situ polymerization of unsaturated metal methacrylate: influences of salt cation

Abstract

In situ reaction of unsaturated metal methacrylate (UMM) has captured scientists’ attention due to its importance in reinforcing low-temperature-grade hydrogenated acrylonitrile butadiene rubber (LTG-HNBR). In this article, LTG-HNBR composites with in situ polymerized sodium (Na⁺), magnesium (Mg²⁺) and aluminum (Al³⁺) methacrylates were successfully fabricated for the purpose of investigating the roles of their salt cations on the reinforcement of the rubber. When the cation valence rose, UMM self-polymerized to produce poly(UMM) and then converted to hybrid structures including poly(UMM) and grafting components to the rubber chains; even unreacted aggregations of UMM during the vulcanization of the matrix. Low solubility of UMM with trivalent cation (Al³⁺) complicated the composite system, decreasing its conversion of poly(UMM) and growing polymethylacrylic acid. Once UMM was fully dissolved, the poly(UMM) developed into fine, dispersed nanoparticles. Monovalent cation (Na⁺) drove these dispersed particles to arrange as band-like nano-topographies. Reinforcement of the rubber matrix was greatly affected by the generation of poly(UMM) where a tiny amount of aluminum polymethacrylate (i.e. poly(AlMMA)) gave rise to poor reinforcements. So the morphology and chemical structure of poly(UMM) and the solubility of UMM induced by its cations have a remarkable impact on reinforcement of rubber matrices. We believe that choosing the cation species of UMM may be a simple method to control the reinforcement of rubber composites.

---

Introduction

Recently, hydrogenated acrylonitrile butadiene rubber (HNBR) has attracted scientists and engineers’ great interest and been extensively used in many diverse applications, such as automobile, aviation, oilfield, and other industries. As a substitute for acrylonitrile butadiene rubber (NBR), HNBR possesses better mechanical properties, abrasion resistance, adhesion to fabrics and hot air aging resistance owing to its hydrogenated processing.¹ In order to satisfy industrial requirements, HNBR have been commoditized by classifying its acrylonitrile (ACN) content.² Among different varieties, HNBR with ∼18 wt% ACN (LTG-HNBR) is a particular type that can be applied at a temperature below −30 °C. However, to pursue high elasticity at low temperature, large side groups are introduced to the LTG-HNBR molecules, and these deteriorate other physical properties of the rubber itself, especially mechanical properties.³ Therefore, the reinforcement of LTG-HNBR becomes an urgent need, especially for high-pressure oil sealing applications.

Nanofillers such as carbon black, silica, montmorillonite, carbon nanotubes, carbon fibers and graphene have been widely used in the reinforcement of rubber.^4–11 Typically, the mechanical properties of the rubber matrix can be greatly enhanced because these nanofillers strongly restrain the molecular chains. They tend, however, to form large aggregates in rubber, leading to drawbacks in processing and poor reinforcement.⁷ An effective technology called in situ reaction emerged to resolve this problem. In this technology, an unsaturated metal methacrylate (UMM) with reactive C=C groups is blended with a certain amount of rubber and peroxide.¹² The nanoparticle phase composed of self-polymerized UMM (i.e. poly(UMM)) is created during vulcanization. Simultaneously, other UMM radicals also graft with the rubber chains. Benefiting from these complex interactions, the reinforcement of the in situ poly(UMM) on the rubber is much better than that of traditional carbon black. A variety of performance studies on this kind of rubber have been conducted. For example, HNBR, natural rubber (NR), butadiene rubber (BR), styrene butadiene rubber (SBR) and ethylene propylene diene monomer (EPDM) can be greatly reinforced by UMMs, though the reinforcing effect varies with the type of both matrix rubber and salt.^12–21 There are, however, still some basic questions remaining unclear, such as the detailed morphologies of poly(UMM) and the in situ polymerization process of UMM. Thus, to study deeply in situ self-polymerization of UMM may be very useful to serve the reinforcement of rubber.

Researchers have made attempts to understand the reinforcing mechanism of UMM by various tests, including Fourier transform infrared spectroscopy (FTIR), thermal gravity analysis (TGA), scanning electron microscopy (SEM) and mechanical testing.^{12,13,16,22,23} However, the role of ion pairs has rarely been considered. Ion bonds lack saturability and directivity which is naturally different from the van der Waals’ or hydrogen bonds of traditional fillers.²⁴ So when these organic fillers with ions are introduced to rubber, this will arouse unusual effects on the rubber’s properties. In concept, anions can be freely attracted around as many cations as possible owing to the electrostatic force. The amount of anions attracted is determined by the radius of the cations. They can be also paired according to their charge numbers, which can induce the arrangement of ions. In the case of poly(UMM), the size of its metal cation is very small, but that of the anion is larger owing to their nature of the macromolecular chain. In other words, the arrangements of poly(UMM) nanoparticles can possibly be controlled by their ion pairs in the matrix. On the other hand, grafting structures chemically fastens some polymerized UMM onto the rubber chains. Their anions can bridge rubber chains with poly(UMM) nanoparticles by the attraction of cations. These interactions are so strong that the mobility of rubber chains can be effectively restricted under stresses or strains (that is, rubber reinforcement). As a result, they are anticipated to modulate the reinforcement of UMM/rubber composites by varying the ions of UMM.

In this paper, we aimed to investigate the effects of metal cations on the self-polymerized process of their UMMs and further impacts on reinforcing LTG-HNBR. Three kinds of alkalis with sodium (Na⁺), magnesium (Mg²⁺) and aluminum (Al³⁺) were chosen to synthesize in situ poly(UMMs) in the rubber matrix by neutralizing MMA, then co-curing by peroxide. The morphologies and chemical structures of the poly(UMMs) were examined. Their interactions with the matrix were also analyzed by FTIR, visco-elasticity and stress-relaxation. After that, a possible mechanism associated with the cations of poly(UMMs) was proposed. Finally, the reinforcement of the rubber composite was checked and reasonably illustrated. We believe that these studies open a new road to tune the mechanical properties of LTG-HNBR and serve in practical applications, such as oil-sealing products.

Experimental

Materials

LTG-HNBR (Zetpol 4320, with a Mooney viscosity (ML (1 + 4) 100 °C) of 70 ± 3, a number-average molecular weight of approximately 135 000 with a polydispersity of ∼3.4, and a density of 0.98 g cm⁻³) with an acrylonitrile mass fraction of about 18.6% and 5% residual double bonds was provided by Zeon Co. (Tokyo, Japan). Methacrylic acid (MMA) was purchased from Beijing Jinlong Chemical Agent Co., Ltd. Three kinds of alkali including sodium (NaOH, analytical purity, spherical particles with sizes of 600–800 μm), magnesium (Mg(OH)₂, analytical purity, crystal powder with sizes of not more than 2 μm) and aluminum hydroxide (Al(OH)₃, analytical purity, crystal powder with sizes of 1–10 μm), and other chemicals and ingredients, such as the curing agent dicumyl peroxide (DCP) and triallyl isocyanurate (TAIC), were purchased from Beijing Chemical Reagent Co., Ltd. Sodium (NaMMA), magnesium (MgMMA) and aluminium methacrylate (AlMMA) were synthesized in situ by neutralizing the MMA with alkali during the mixing of the rubber compounds.

Preparation

LTG-HNBR was plasticized for 5 min at the temperature of 50–60 °C by a mixing roller. Then, alkali and MMA were kneaded into the rubber. The molar ratios of alkali including sodium hydroxide (NaOH), magnesium hydroxide (Mg(OH)₂) and aluminium hydroxide (Al(OH)₃) to MMA were 1 : 1, 1 : 2 and 1 : 3, respectively. The mass loadings of generated UMMs (NaMMA, MgMMA and AlMMA) were stoichiometrically held at 30 parts per hundred rubbers (phr). Other compounding and crosslinking additives including 3.0 phr antioxidant, 4.0 phr DCP and 1.5 phr TAIC were added. For comparison, pure rubber blends with the same crosslinking agents and antioxidant, 3.0 phr ZnO and 3.0 phr MgO were made. Kneading continued at room temperature for 10 min. After that, all blends were hot-pressed under a pressure of 15 MPa and vulcanized at 160 °C for 20 min, then naturally cooled down to room temperature. Finally, the vulcanized specimens were further put into an oven at the temperature of 150 °C for 4 h in order to reduce the unreacted UMMs and avoid the aging effect on their short-term properties.²⁵

Characterization

SEM (S440, Leica Cambridge Ltd., England) images were taken from the representative fractured surfaces of the rubber composites after they were gold coated. The SEM specimens were prepared by fracturing the rubbers in liquid nitrogen. A Hitachi H-800-1 TEM, which was purchased from Hitachi Ltd. (Tokyo, Japan), was used to examine the morphology of the poly(UMMs) in the composites, and the TEM specimens were prepared at −100 °C with a Reichert-Jung Ultra-cut microtome manufactured by Leica Camera AG (Leitz, Germany) and mounted onto 200-mesh copper grids. DMA measurements were made with a dynamic mechanical analyzer (VA4000, Metravib Co., French) by using cylinder specimens (their diameter and height were 10 mm and 10 mm, respectively). The temperature dependence of compressing moduli was measured in a range of −60 to 150 °C at the constant frequency of 125 Hz and the heating rate of 3 °C min⁻¹. FTIR spectra (the wavenumber range of 500–4000 cm⁻¹) were obtained from an accumulation of 100 scans at a resolution of 2 cm⁻¹ by a FTS3000 spectrometer (Bio-Rad Co., USA). The IR spectra (KBr pellets) were taken at ambient temperature. XRD patterns were recorded with a Rigaku D/max-2500 diffractometer, the X-ray beam was Ni-filtered Cu Kα (λ = 0.1542 nm) radiation operated at 40 kV and 100 mA, corresponding data were collected from 10° to 40° at a scanning rate of 2° min⁻¹. Stress relaxations of the composites were performed by stretching themselves at a strain of 100% by using a Testometric Universal Tester M350-20 kN at high across head speeds of 1000 mm min⁻¹. The plots of stress variations versus time were collected in a total time of 10 min. Tensile and tear tests were carried out at room temperature by using a Testometric Universal Tester M350-20 kN at an across head speed of 500 mm min⁻¹. In tear tests, trouser-shaped samples were selected. At least five specimens were tested for each rubber composite and their mean value was measured.

To check the conversion of UMM and the quantities of self-polymerized and grafted UMMs in the rubber composites, the method of twice-solution-extraction was used, as described in ref. 26. As-prepared composites with UMMs were firstly extracted by a 125 ml methanol/25 ml hydrochloric acid mixture at the temperature of 50–60 °C by a soxhlet extractor. Assuming that the weight variations come from residual UMMs, the conversion of UMMs (α) can be calculated by eqn (1):

where m_UMM is the total weight of UMM in the rubber compounding and m_var is the weight variation of the composite after the extraction. A secondary extraction was then performed with a methylbenzene/methanol mixture (with a weight ratio of 31 : 69) after being immersed in butanone/chloroacetic acid (the volume ratio of 70 : 30) for 3 days. During this process, the weight variations of the samples were equal to the weights of self-polymerization of UMM (particles) because the part of UMMs grafting to the rubber chains cannot be extracted by these mixtures. So, the content (ω) of self-polymerized UMM is computed by the following equation,where m_var,sec is the weight variation of the composite after secondary extraction. By extraction and then weighting, the conversion of UMM, and the content of self-polymerization and grafted UMMs was estimated.

Results and discussions

Fig. 1 shows the FTIR spectra of pure rubber, MMA and various composites in the wavenumber range of 550–4000 cm⁻¹, respectively. It has been known that the force constant of the C=O and C–O group at the same carbon atom of the COO group of UMM or poly(UMM) is averaged owing to the attraction of salt cations.²² So, the UMM or poly(UMM) can be judged by the peaks of carboxyl groups. The anti-symmetric and symmetric stretching vibrations of carboxyl groups for the composite with NaMMA are observed at 1556 cm⁻¹ and 1458 cm⁻¹, while those with MgMMA are at 1546 cm⁻¹ and 1458 cm⁻¹, respectively. These peaks, however, cannot be observed for the composite with AlMMA. We even see the absorption at 1701 cm⁻¹ from the C=O groups for this composite. So, some parts of MMA may not react with Al(OH)₃. On the other hand, the absorption at 1698 cm⁻¹ and 1636 cm⁻¹ is assigned to the C=O and C=C stretching vibration of liquid MMA. Also, the absorption at the range of 1200–1300 cm⁻¹ belongs to the stretching vibration of its C–O groups. As expected, the characteristic peak of C=C groups at 1636 cm⁻¹ for all composites disappears, suggesting that MMA becomes polymerized after vulcanisation of the matrix. Besides, the absorption of CN groups at 2235 cm⁻¹ and the bending vibration of C–H at 725 cm⁻¹ belonging to the pure rubber are also studied. Any shifts of these, however, are hardly found. This situation means that poly(UMMs) do not have strong interactions with rubber polar groups.

Fig. 1 FTIR spectra of liquid MMA, pure rubber and its composites at the wavenumber range of 550–4000 cm⁻¹.

In order to quantify further the chemical composition, the conversion of UMM and the content of poly(UMM) was tested (see Table 1). Note that the conversion α of MgMMA is highest, but AlMMA is barely self-polymerized (i.e. poly(AlMMA)). The calculated content of self-polymerized UMM, ω is shown in Table 1. The value of ω for the composite with NaMMA is approximately 1, suggesting that self-polymerized NaMAA (poly(NaMMA)) rarely graft to rubber chains. In contrast, there are the hybrid structures including self-polymerized (i.e. poly(MgMMA)) and grafting components for the composite with MgMMA. Also, the amount of grafting-poly(MgMMA) is greater than that of self-polymerized species. It is difficult to analyze the detailed components for the composite with AlMMA due to its low conversion which agrees with the observations of FTIR (most MMA cannot produce AlMMA). It is stressed that the cation of UMM has a great impact on its chemical composition in the matrix.

Table 1 Chemical structures of in situ polymerized UMMs in the matrix

Chemical structures	Rubber composites with UMMs
	NaMMA	MgMMA	AlMMA
Conversion ratio of UMMs (α)/%	86.2	98.2	4.4
Content of self-polymerized UMM particles (ω)/%	98.1	31.8	—
Content of grafting structures (1 − ω)/%	1.9	68.2	—

---

To examine the morphologies of self-polymerized UMMs in the rubber matrix, their SEM images were taken. Meanwhile, local spectrograms of Energy Dispersive Spectrometry (EDS) attached to the SEM were used to identify the poly(UMMs). Fig. 2 shows the SEM images of the pure rubber and its composites with NaMMA, MgMMA and AlMMA. Note that some white microparticles exist in the pure rubber (see Fig. 2a). The EDS spectrogram displays that they belong to rubber compounds of ZnO or MgO (see Fig. S1†). There are, however, barely any particles on the fractured surface of the composite with NaMMA as shown in Fig. 2b, implying that poly(NaMMA) has good compatibility with rubber. A similar morphology can be found in the composite with MgMMA. Careful observations reveal that large aggregations with sizes of 20–30 μm are distributed sporadically in the matrix (see the inset of Fig. 2c). These aggregations belong to poly(MgMMA) or MgMMA due to the scanned elements of C, O and Mg (see Fig. S2†). Interestingly, we find more complex surface features including multi-sized aggregations on the composite with AlMMA. We roughly identify these aggregations at two scales: one at ∼1 mm and another with at a few microns (see Fig. 2d). Moreover, the two scaled aggregations are very different. The bigger ones display regular structures more like some crystals whose scanned elements include Al, C and O (see Fig. 2e). Their compositions are not consistent with the smaller ones, which only have the elements of C and O (see Fig. 1f). To track their origin, the surface of the rubber compound was also examined by SEM before being vulcanized (see Fig. S3 and S4†). A lot of aggregations with a size of ∼1 mm also exist. So, it is deduced that the bigger aggregations in the composite are possibly generated before vulcanization, i.e. AlMMA. Owing to the low conversion of AlMMA, those smaller aggregations are possibly related to unreacted MMA.

Fig. 2 SEM images of LTG-HNBR composites with various UMMs on their fractured surfaces: (a) pure rubber; (b) NaMMA; (c) MgMMA; (d) AlMMA. The aggregations at two scales are identified from the composite with AlMMA by using EDS spectra: (e) the sizes of the bigger aggregations reach ∼1 mm (black label in the (d)); (f) the others are micron-sized particles (white label in the (d)).

Fig. 3 shows the TEM images of composites with NaMMA, MgMMA and AlMMA. Clearly, there are two phase domains, that is, a matrix and dispersed poly(UMM) phase. According to the TEM imaging principle, it can be identified that the darker phases represent the poly(UMM) regions. Some band-like nano-topographies with a thickness of 50–100 nm can, interestingly, be seen in the composite with NaMMA. In contrast, 10–30 nm granular nano-dispersions grow in the composite with MgMMA. Furthermore, there appear larger particles (with sizes of about 100 nm) in the composites with AlMMA, but their quantities are far less than those with MgMMA.

Fig. 3 Representative TEM images of the composites from in situ polymerization of UMMs: (a) NaMMA; (b) MgMMA; (c) AlMMA.

The fractural surfaces of the composites were also checked after they were twice extracted by solution, which are very important steps to identify the original dispersion and morphologies of poly(UMMs) in the rubber. SEM images of the composites are recorded in Fig. 4. There are dense micro-pores on the composites with NaMMA; on the other hand, larger pores (with sizes from the nanoscale to the microscale) can be seen for that with MgMMA. These micro-pores correspond to poly(UMM) particles that have been extracted. Yet the pores on the composite with AlMMA reach millimetres in size. Moreover, their sizes scatter and some nanoscale or micro-scaled pores coexist. These composites were also observed by TEM (see Fig. 5). The darker phases in the TEM photographs represent the residual pores after UMM and poly(UMM) was extracted. When comparing with the original TEM image in Fig. 3, it is found that the band-like morphologies of poly(NaMMA) are actually composed by dense nano-particles (see Fig. 5a). Moreover, the pore sizes in the composites with MgMMA and AlMMA are smaller than their original particles, and their quantities are less than original ones (see Fig. 5b and c). So, it is confirmed that the cations of UMMs play positive roles to control the generation, morphology and dispersion of poly(UMM) in the rubber.

Fig. 4 SEM images of LTG-HNBR composites on their fractured surface after extraction: (a) NaMMA; (b) MgMMA; (c) AlMMA.

Fig. 5 Representative TEM images of LTG-HNBR composites after extraction: (a) NaMMA; (b) MgMMA; (c) AlMMA.

Fig. 6 shows the XRD traces of the pure rubber and its composites with various UMMs. Note that the pure rubber has some sharp XRD multi-peaks as a result of compounding ZnO and MgO, moreover it possesses a broad peak at a 2θ angle of 10–30° corresponding to the amorphous state of the matrix. On the other side, there are the similar broad shapes of XRD curves for the composites with NaMMA and MgMMA, implying that the two poly(UMMs) are amorphous in the rubber. A comparison is further made that the peak of the composites becomes sharper and their intensities are higher than that of pure rubber. This situation indicates that the structures of rubber composites become more ordered due to the addition of UMMs. In addition, we notice a broad XRD peak for the composite with AlMMA, but some sharper peaks overlap it, displaying further regular structures in the matrix. Associated with these morphology and chemical composition observations, the sharp overlapping peaks are possibly derived from their multi-sized aggregations in the matrix.

Fig. 6 XRD curves of pure rubber and its composites.

Poly(UMM) possesses a strong electrostatic force in the matrix owing to the nature of its ion bonds. FTIR observations are not enough to display the special interactions because they only characterize the interactions between the organic groups with ions, instead of the ionic bond itself. Yet the ions in the rubber can be identified by applying stress or strain.²⁵ Fig. 7a shows the temperature dependence of the loss tangent (tan δ) values for the pure rubber and its composites with various UMMs. Clearly, pure rubber and its composites with NaMMA and MgMMA have only one single tan δ peak (tan δ_max). For the composites with AlMMA, however, two phase transitions are seen, as evidenced by two corresponding tan δ peaks. Such double phase peaks have been revealed by other UMM/rubber composites.²⁵ The ion pairs of UMMs develop multiplets surrounded by rubber chains with restricted mobility, and then the regions of restricted mobility overlap and form clusters (ion-rich regions).²⁷ So, the tan δ peak (tan δ_max,1) near −12 °C (T_max,1) belongs to the matrix rubber, and the peaks (tan δ_max,2) near 130 °C (T_max,2) are identified as a cluster peak arising from poly(UMM). The fewer the valence numbers of UMM cations, the smaller the is value of tan δ_max,1 for its composites. The sequence of tan δ_max,1 is: NaMMA (0.66) < MgMMA (0.88) < AlMMA (1.36). In contrast, the second peak of tan δ_max,2 is not found in the composites with NaMMA and MgMMA, even when the testing temperature reaches 290 °C, which implies that their ion pairs are well dissolved in the matrix, and rarely aggregated to form clusters. Yet, the composite with AlMMA maintains a tan δ_max,2 of 0.27. The area under the tan δ peak (TA) of the matrix can be used to examine the ion content. TA, tan δ_max and T_max of all composites are listed in Table S1.† The relation between TA, tan δ_max,1 and the valence numbers of the UMMs is fitted well by the linear equation: TA₁ = 8.0v + 21.2 and tan δ_max,1 = 0.4v + 0.23, where is v is the valence numbers of salt cations (see Fig. 7b). Because the ion content in the matrix is inversely proportional to the TA value,²⁸ there are more ion pairs in the composites with NaMMA and MgMMA than that with AlMMA. This means that salt cation valence changes the content of the ion pairs dissolved in the matrix and thus induces different interactions of poly(UMMs) with the rubber chains.

Fig. 7 (a) Temperature dependence of tan δ for pure rubber and its composites; (b) the relations among the TA, tan δ_max,1 of the composites and cation valences of the UMMs.

Similarly, stress relaxation was applied to study the interaction response between poly(UMMs) and rubber. For the rubber composites, their relaxations only descend to a certain stress, e.g. the equilibrium stress, σ_e. The relaxation time is suitable to characterize the relaxation process of rubber composites with ion pairs.²⁹ This is the time when stress reduces to e⁻¹ times the stress difference of σ₀ − σ_e (σ₀ is the original stress). The relaxation time of pure rubber and the composites with NaMMA, MgMMA and AlMMA is 27.4 s, 15.8 s, 12.2 s and 26.7 s, respectively (see Fig. 8a). The longer relaxation time is, the better the elasticity of the rubber composite is. During relaxation, the rubber chains rearrange to accommodate the constant strain and thus decrease the stress. Such stress variations can rapidly damage the ion bonds from poly(UMMs).¹⁶ The irreversible recovery of ion bonds would greatly reduce the interaction among rubber chains and therefore the stresses of the composites reduced at a quicker relaxation rate. However, the composite with AlMMA possesses an almost equal relaxation time to pure rubber. This is easily understood because only a small quantity of poly(AlMMA) is generated in the rubber composite. The Maxwell model can describe the spectrum of stress relaxation for the rubber composites where multiply exponential expressions are more suitable to demonstrate the constitutive relation of rubber composites (see the schematic in Fig. 8b).²⁹ Here, the three exponential Maxwell model is used to describe the stress relaxation process:

where τ_i is the ith relaxation time. The fitting calculations are shown in Table S2† (also plotted in Fig. 8a). As is seen in Fig. 8a, the relaxation of pure rubber and its composites is well fitted by eqn (3).

Fig. 8 (a) Stress relaxation plots of the pure rubber and its composites. The fitting plots are drawn by the calculation of eqn (3) (the coloured lines); (b) the schematic for multiply exponential Maxwell models to describe the stress relaxation of rubber composites.

On the basis of the above analysis and the previous works of Zhang’s group,^4,12,13,16 the effects of cations on the microscopic interactions between rubber and poly(UMMs) are further illustrated in our cases of LTG-HNBR composites (see Fig. 9). In spite of special cations, the ion links and crosslinking networks are developed simultaneously: UMMs are in situ polymerized in the rubber through free radical initiation of DCP at high temperature; the polymerization of UMMs is possibly terminated by macromolecular radicals of matrix rubber, resulting in grafting of poly(UMMs) onto the rubber chain and ion crosslinking; moreover, a covalent crosslinking network of rubber is also formed. So, competition between the radical cross-linking reaction of the matrix and in situ polymerization of UMMs is created. Furthermore, the conversion of UMMs becomes key to produce poly(UMMs). The solubility of UMM in the rubber affects directly this conversion. For a trivalent cation (Al³⁺), much of AlMMA cannot convert to poly(AlMMA) due to its low solubility in the matrix. In contrast, once UMM is well solved, its conversion to poly(UMM) is sufficiently high. Taking the example of NaMMA, gel permeation chromatography (GPC) shows that the number-average molecular weight of poly(NaMMA) (extracted by mixed solution) is beyond 25 000 in the matrix. The metal cations have made a tremendous impact on the generation and dispersion of poly(UMMs). When the valence numbers increase, salt cations have a smaller ion radius and allow fewer anions of MMA⁻ to be attracted. For example, the ions of Na⁺ are surrounded by more attracted MMA anions than Mg²⁺. A high concentration of MMA⁻ around Na⁺ generates more self-polymerized NaMMA. Long chains of poly(NaMMA) are wound around each other to grow nanoparticles. During this process, a lot of anions (MMA chain elements) are possibly hindered in these nanostructures. Reducing the amount of anions cannot balance (initially equimolar) cations. Thus these nanoparticles are aligned as band-like arrangements with a certain aspect ratio to increase their surface areas in order to balance charges. Also, two layers of ions protect nanoparticles and their arrangements from further aggregation. In contrast, anions cannot be effectively attracted by Mg²⁺. This situation provides enough chances for MgMMA to contact the rubber chains and then develop grafting structures. So, there are more complex chemical components in its composite. Moreover, these grafting structures help the nanoparticles bridging them with rubber chains to enhance their interactions by the attraction of Mg²⁺ owing to higher pair numbers. However, local mismatched amounts between cations and anions possibly allow the formation of ion clusters. Especially when Al³⁺ is introduced into the rubber, its high paired-ratio to anions causes more clusters to generate. Besides, the low solubility of AlMMA results in the polymerization of MMA, instead of poly(AlMMA). Such a small quantity of soluble ion pairs in the composite with AlMMA causes fewer ion crosslinks than with NaMMA and MgMMA (see Fig. S5†). Moreover, this leads to the least total crosslink density including covalent and ionic links in all composites (see ESI†). The variations of the crosslink networks are important for the performance of the rubber composites, such as elasticity and tensile properties. Therefore, cations play a really essential role in the polymeric structures of UMM and dispersed morphologies of poly(UMM), which is destined to affect the reinforcement of rubber composites.

Fig. 9 Schematic illustration of the microscopic interactions between in situ polymerized UMMs and the rubber matrix.

The dependence of the storage modulus (E′) on temperature for the rubber composites was measured by DMA, as shown in Fig. 10. The E′ values show that these composites are glassy below −30 °C, through the glass transition of the matrix phase between −30 °C and 0 °C, and exhibit a rubbery plateau region beyond 0 °C. Note that the E′ of the composites at the glassy state is less than that of pure rubber. The contrary trend, however, holds in their rubbery state, implying that the confinement of poly(UMMs) on the rubber molecule chains can work above the glassy temperature (T_g) of the matrix. Careful observations show that the E′ values of the composites at the glassy state follow the sequence: NaMMA < MgMMA < AlMMA < pure rubber, while the opposite situation occurs in their rubbery states. Modulus variations imparted by poly(UMMs) may be regarded as the product of two terms: one involves a hydrodynamic effect arising from the inclusion of polymeric fillers. In reference to the rubber composites from polymeric fillers, there is the relation among various components:³⁰

where Φ_i is the volume fraction of the ith component and E′_i is its modulus. It is seen from eqn (4) that the moduli of the composites are affected by poly(UMMs). At a lower temperature, the rubber chains cannot move and thus poly(UMM) components are more like certain plasticizers for the matrix. At the rubbery state, however, rubber chains can move freely and rubber–poly(UMM) interactions work. The rubber chains are strongly restricted by nanostructures and ion attractions and the modulus of the composite then improves. Dense Na⁺ ions arouse stronger collective interactions with the matrix than do Mg²⁺ and Al³⁺ ions due to their greater number of ion pairs. So, the modulus of the composite with NaMMA becomes the maximum. In addition, the modulus of the composite with AlMMA decreases distinctively in the temperature range of 50–150 °C. We ascribe this to the dissociation of ion clusters, reducing the ion interactions at a high temperature of above 130 °C.³¹

Fig. 10 Temperature dependence of the storage modulus E′ for pure rubber and its composites.

Tensile properties of the composites with UMMs are listed in Table 2. The dramatic increase of tensile strengths can be found in all composites, although they are lower than the reported composites (ZSC, Zeon, Co.), by adopting HNBR with higher ACN content as the matrix (beyond 40 MPa). For example, the tensile strength of the composite with NaMMA (29.5 MPa) is 9 times greater than the pure rubber (3.1 MPa). In order to display better the reinforcement, a comparison was also made when 30 phr high abrasion furnace black (N330) was added into the LTG-HNBR matrix. As expected, its tensile strength (18.3 MPa) improves greatly, but it is far smaller than the composite with NaMMA and MgMMA. When in situ poly(UMMs) are distributed in rubber matrices as nano-sized particles, the interfacial bonding from electrostatic attractions of poly(UMM) ionomers is reasonably strong. So, an effective reinforcement could be achieved, and the composites significantly improve their tensile properties. Unfortunately, the reinforcement of the composite with AlMMA is lower than that of the carbon black. This weak reinforcement is attributed to multi-sized aggregations in the rubber and tiny amounts of poly(AlMMA). It is therefore stressed that a lot of poly(UMMs) ensure the reinforcement of rubber composites. Their elongation variations at break are interesting to note. In the case of the composite with AlMMA, there are weak filler interactions with rubber due to their reduced in situ polymerization. So, AlMMA cannot effectively limit the mobility of rubber chains during the stretching and causes large elongation. In contrast, the composites with NaMMA and MgMMA have enough reinforcement to improve their ability to block the fracture of the matrix, achieving a relatively larger elongation than that of pure rubber. Besides, the improved hardness and tearing strength of the rubber composites are in concordance with the tensile strength results. Their permanent set percentage is also raised by the enhancement of the viscosity by the strong interactions of poly(UMMs).

Table 2 Tensile and tearing properties of pure rubber, the composites and the in comparison with carbon black (30 phr N330)

	Pure rubber	Rubber composites with different fillers
		NaMMA	MgMMA	AlMMA	30 phr N330
Hardness/Shore A	43	81	71	47	60
Tensile strength/MPa	3.1	26.6	29.5	12.8	18.3
Elongation at break (%)	325.9	370.0	385.9	428.4	309.8
Permanent set (%)	3.5	33.4	19.2	12.4	5.6
Tearing strength/KN m^−1	7.7	81	71	20.3	21.8

---

Conclusions

LTG-HNBR composites were successfully prepared by the in situ polymerization of UMMs with various cations. Upon the increase of the valences of the cations, their UMM tends to self-polymerize and then extend to the hybrid structures including self-polymerized and grafting structures. Self-polymerized UMMs produce well dispersed nanoparticles where a monovalent cation (Na⁺) has driven them to band-like arrangements. High paired-ratios of multivalent cations to anions make their nanoparticles larger, but the surface charges of these particles block their aggregation, to develop good dispersion. On the other hand, the low solubility of AlMMA in the rubber brings more phases, including AlMMA, poly(MMA) and small amounts of poly(AlMMA). Tensile properties and the modulus at a temperature above T_g of LTG-HNBR were greatly affected by the in situ polymerization of UMMs with different cations. However, their reinforcements are really determined by the chemical structures, morphologies and solubility of poly(UMMs), which can be controlled by the cation species of UMMs. We believe that this study may be very useful to provide a new road to the design of reinforced LTG-HNBR composites to serve in low temperature applications.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (51103033).

Notes and references

1. P. Cao, C. Huang, L. Zhang and D. Yue, RSC Adv., 2015, 5, 41098.
2. S. Hayashi, H. Sakakida, M. Oyama and T. Nakagawa, Rubber Chem. Technol., 1991, 64, 534.
3. T. Kobatake, K. Kodama, S. Hayashi and A. Yoshioka, Rubber Chem. Technol., 1997, 70, 839.
4. Y. Guan, L. X. Zhang, L. Q. Zhang and Y. L. Lu, Polym. Degrad. Stab., 2011, 96(5), 808.
5. J. Potts, O. Shankar, L. Du and R. Ruoff, Macromolecules, 2012, 45(15), 6045.
6. V. Singh, A. Shukla, M. Patra, L. Saini, R. Jani, S. Vadera and N. Kumar, Carbon, 2012, 50(6), 2202.
7. L. Wahba, M. D’Arienzo, R. Donetti, T. Hanel, R. Scotti, L. Tadiello and F. Morazzoni, RSC Adv., 2013, 3, 5832.
8. Y. Li, Q. Wang, T. Wang and G. Pan, J. Mater. Sci., 2012, 47, 730.
9. M. Bhattacharya and A. K. Bhowmick, J. Mater. Sci., 2010, 45, 6126.
10. X. Wang, A. Huang, D. Jia and Y. Li, Eur. Polym. J., 2008, 44, 2784.
11. Z. Wang, X. Fan, K. Wang, H. Deng, F. Chen, Q. Fu and B. Na, Polym. Adv. Technol., 2011, 22, 1375.
12. Y. Lu, L. Liu, C. Yang, M. Tian and L. Zhang, Eur. Polym. J., 2005, 41, 577.
13. Z. Wei, Y. Lu, Y. Meng and L. Zhang, Polymer, 2012, 53, 1409.
14. D. Yin, Y. Zhang, Y. Zhang, Z. Peng, Y. Fan and K. Sun, J. Appl. Polym. Sci., 2002, 85, 2667.
15. B. Guo, Y. Lei, F. Chen, X. Liu, M. Du and D. Jia, Appl. Surf. Sci., 2008, 255, 2715.
16. Y. Lu, L. Liu, M. Tian, H. Geng and L. Zhang, Eur. Polym. J., 2005, 41, 589.
17. A. Samui, V. Dalvi, L. Chandrasekhar, M. Patri and B. Chakraborty, J. Appl. Polym. Sci., 2006, 99, 2542.
18. Y. Nie, G. Huang, L. Qu, P. Zhang, G. Weng and J. Wu, J. Appl. Polym. Sci., 2010, 115, 99.
19. H. Yu, Y. Zhang, S. Chen, W. Ren, M. Hoch and S. Guo, J. Appl. Polym. Sci., 2011, 119, 1813.
20. S. Rajesh Kumar, P. M. Asseref, J. Dhanasekaran and S. Krishna Mohan, RSC Adv., 2014, 4, 12526.
21. L. Wang, F. Lang, F. Du and Z. Wang, J. Macromol. Sci., Part B: Phys., 2013, 52, 178.
22. S. Wen, Y. Zhou, L. Yao, L. Zhang, T. Chan, Y. Liang and L. Liu, Thermochim. Acta, 2013, 571, 15.
23. Y. Lu, L. Liu, D. Shen, C. Yang and L. Zhang, Polym. Int., 2004, 53, 802.
24. X. Hu, H. Gao, X. Zhou, Y. Cui and H. Ge, RSC Adv., 2014, 4, 13662.
25. T. Ikeda, B. Yamada, M. Tsuji and S. Sakurai, Polym. Int., 1999, 48, 446.
26. Y. Lu, Master’s Thesis, Beijing University of Chemical Technology, 2001.
27. X. Ma, J. Sauer and M. Hara, Polymer, 1996, 37(21), 4739.
28. M. Nishida and A. Eisenberg, Macromolecules, 1996, 29, 1507.
29. A. A. Apostolov, S. Fakirov and J. E. Mark, J. Appl. Polym. Sci., 1998, 69, 495.
30. K. D. Ziegel and A. Romanov, J. Appl. Polym. Sci., 1973, 17, 1133.
31. K. Sato, Rubber Chem. Technol., 1983, 56, 942.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra21688a

---