3.1. Study of the Individual Crosslinked Networks
In the first stage, we analyzed individual crosslink networks in an XNBR matrix. Three compounds were made with increasing amounts of ZnO (2.5, 5 and 10 phr) and one pure covalent with 1 phr of S.
Figure S1 shows the values of crosslink density. The results of 1S and 10 ZnO stand out because of their similarity; thus, they can be considered as equivalent crosslinked networks.
Curing curves (
Figure 1a) show that each network has a different behavior during vulcanization; the covalent network draws a less pronounced curve showing a slow vulcanization rate. It is important to mention that the 1S recipe does not contain activators, so very long times are expected to accomplish full vulcanization. In contrast, the ionic network presents a much faster vulcanization rate. In addition, instead of reaching a
plateau, the ionic network curves have a slightly upward trend (marching modulus), this being a characteristic behavior of these types of crosslinks [
8,
11].
Table S1 summarizes the data from the curing curves. As expected, the
MH value increases with the amount of ZnO due to the increase in the ionic crosslinks points in the matrix that cause a greater resistance to shear deformation [
14].
In the ionic networks, the crosslinking of the XNBR matrix is carried out according to Equation (13):
The incorporation of metal ions yields elemental bonds in the matrix, defined as ionic pairs, and originated by the association of the anionic groups COO
− and the cations Zn
2+. The selected XNBR has 7 wt. % of carboxyl groups (-COOH); hence, according to Equation (13), the necessary amount of ZnO to obtain the saturation of these groups in 100 g of XNBR is approximately 6 g. According to the Eisenberg model [
6], the ionic pairs in the matrix are capable of forming associations called multiplets. If the proportion of multiplets is high, the ionic domains create clusters that reduce the mobility of the polymeric chains, adding rigidity to the matrix and increasing the tensile strength at room temperature.
Regarding their stress–strain curves, both networks show a very characteristic and different behavior. Despite their equivalent crosslink densities, it can be seen how the ionic network reaches very high values of tensile strength 28 ± 5 MPa with elongation at break of 472 ± 33%, while the covalent network shows a low tensile strength 4.4 ± 0.7 MPa, but high elongation at break 829 ± 54% (
Figure 1b). Such a behavior can be related to the bond energy. In covalent bonds, such as monosulfide (-S-), disulfide (-S-S-), polysulfide (-S
x-), carbon–carbon (-C-C-), among others, the weaker bond tends to give higher tensile strength due to a mechanism of stress dissipation; however, in ionic bonds, the strength of the electrostatic interactions and the formation of clusters that trap polymer chains (stiffening it) have a greater effect on the performance of the material, generating a considerable increase in mechanical resistance with a detriment of the elongation at break [
11,
31].
Tensile strength and crosslink density increase with ZnO content, showing a significant rise between 2.5 and 5 phr (
Table S2). As previously stated, the saturation of the carboxylic groups in the XNBR matrix is reached with 6 phr ZnO. Hence, the ionic clusters in 5 ZnO (close to the saturation point) would explain the increase in mechanical resistance and in crosslink density with respect to the 2.5 ZnO compounds. Meanwhile, the excess of ZnO in the 10 ZnO compound remains in the matrix without forming crosslinks, and it has no relevant effect on tensile properties. Xu et al. [
32] reported similar results when the amount of ZnO in the matrices exceeded the saturation point of the COOH groups.
The structure of each matrix has been verified by infrared spectroscopy (
Figure 2). -COOH groups are present in the XNBR matrix, as confirmed by the appearance of a band at 1700 cm
−1 in the spectrum of a raw sample (associated with carbonyl C=O vibration). When it is vulcanized, the Zn
2+ ions bind to the COO
− groups, forming ionic bonds that are evidenced by the absence of the band at 1700 cm
−1 and the appearance of two bands: one at 1595 cm
−1 and another at 1419 cm
−1, related to the
tetra- and
hexa- coordinated zinc carboxylated structure [
17,
21,
32,
33].
Figure 2a represents the infrared spectra of the 5 ZnO formulation where the aforementioned signals have been identified.
Figure 2b shows the infrared spectra of the other compounds in the vulcanized state. In the covalent formulation (1S), the -COOH signal appears unchanged at 1700 cm
−1, indicating that these groups are free in the matrix. In the case of the ionic networks, they all show the characteristic bands at 1595 cm
−1 and at 1419 cm
−1. However, for the 2.5 ZnO compound, a small signal also appears at 1700 cm
−1, associated with free –COOH groups that have not reacted since the amount of ZnO is below its saturation level.
3.2. Dual Ionic/Covalent Crosslinked Networks
Covalent networks crosslinked with sulfur are characterized by their irreversibility. On the opposite side, ionic networks are reversible and have good mechanical properties, but are unstable at temperatures above the ionic transition temperature (
Ti); in common rubbers,
Tg < RT <
Ti. In accordance with these conditions, efforts have been made to combine the best of both practices, following different strategies [
11]. In this section, the effect of combining both networks on the XNBR is studied. An illustrative scheme of the dual networks is shown in
Figure 3.
The curing curves of the dual crosslinked networks are shown in
Figure 4. The individual ionic and sulfur-cured compounds are also included for comparison purposes. It is possible to identify a trend in the dual networks in which M
H increases with the amount of ZnO. Two effects seem to be responsible for this behavior. First, a higher content of metal oxide increases the likelihood that metal ions and carboxyl groups would be in close proximity, favoring the formation of the ionic network. Second, part of the ZnO (in presence of SA) acts as an activator of the sulfur vulcanization, generating a covalent network, with a more efficient use of sulfur.
Certain similarities can be seen between the dual networks and the purely ionic one (10 ZnO). The marching modulus for the elastic component (
), as well as a maximum in the viscous component (
) curve, are characteristics of dual networks [
9,
10]. An interesting effect is also noticed on the curing curve of the dual compound with the highest ZnO content (1S-10 ZnO). Two slopes are clearly detected in the early stages of the vulcanization, which could be confirming the formation of the two networks. This effect is easily observed in
as the presence of a peak. After the typical increase in
, because of the initiation of vulcanization, these nets do not reach a
plateau but start to descend after reaching a maximum value. The descending zone is associated with the covalent network that increases the elastic component, with a corresponding decrease in the viscous component. This behavior is not detected in individual formulations, such as 1S and 10 ZnO, where
reaches a
plateau [
9,
17].
Table S3 shows a summary of the values obtained from the
curves.
The same trend shown in the rheometric curves can be observed in the study of the stress–strain curves and crosslink density (
Figure 4c and
Table S4). As the amount of ZnO increases, the crosslink density and tensile strength increase, which could be ascribed to a higher number of ionic and sulfur bonds that would eventually act as reinforcing points, increasing the tensile strength.
The internal structure of the compounds was studied by ATR-IR. In
Figure 5a, the spectra of the dual networks are observed. The bands associated with ionic bonds appear at 1595 cm
−1 and 1419 cm
−1 with an increasing amount of ZnO. For the case of 1S-5 ZnO and 1S-10 ZnO, there is enough ZnO to promote ionic vulcanization and form clusters. It is important to mention a band located at 1536 cm
−1, which is present in all compounds, but stands out to a greater extent for 1S-2.5 ZnO and 1S-10 ZnO. This band coincides with another one in the spectra of pure ZnO and ZnSt powder (
Figure 5b). Two plausible explanations can be considered. The prominent band can be related to the excess of ZnO beyond the saturation limit or to the formation of ZnSt, which is a characteristic secondary product of accelerated sulfur-based systems. This could confirm that, even at low metal oxide contents, the formation of the sulfur vulcanization activator complex begins and competes with the formation of ionic crosslinks.
3.3. Chemical and Abrasion Resistance of Individual and Dual Networks
XNBR has high chemical resistance to non-polar solvents, such as gasoline and motor oil, due to its intrinsic chemical structure influenced by nitrile (-CN) and carboxyl (-COOH) functional groups. These characteristics enable its extensive use in the automotive industry, especially for the manufacture of gaskets and hoses [
8]. For the analysis of the chemical resistance, five specimens of each formulation were immersed in three types of solvents: toluene (non-polar aromatic solvent), gasoline 95 (non-polar aliphatic solvent), and motor oil (high viscosity non-polar aliphatic solvent) for 72 h. The results of this test are shown in
Figure 6.
Both types of individual crosslinked matrices show similar chemical resistance regardless of the solvent. In general, it is observed that all the formulations have less mass change in the case of motor oil, followed by gasoline and, finally, toluene. Thus, their resistance will be higher for oil and, in contrast, lower for toluene. An increase in chemical resistance is also observed with increasing ZnO content. This behavior may be correlated with the decrease in free volume, which restricts the diffusion of the solvent in the matrix [
34,
35]. Regarding the dual networks, it is important to highlight that the combination of crosslinks yielded the highest chemical resistance, with reductions of up to
26% in compound swelling. This can be ascribed to a synergistic effect by the presence of the two types of networks that further hinder the diffusion of the solvent into the matrix. This effect can be observed in other systems with certain similarities such as Interpenetrated Polymeric Networks (IPN) in which two crosslinked nets are mixed, and physically interlaced in a molecular scale [
36]. The compound with the best performance was 1S-10 ZnO with a mass change of 176 ± 2% in toluene and 36 ± 1% in gasoline.
Abrasion resistance is the ability of a material to maintain its structure and shape after suffering damage from erosion and/or wear of their surface. This characteristic is important in the case of materials that require a use in which the shape is essential; therefore, the surface resistance would facilitate the shape retention and, thus, its functionality. In XNBR, the functional groups of the matrix originate secondary interactions that form strong bonds, causing high abrasion resistance compared to other rubbers [
8]. Even though in chemical resistance the individual crosslinked matrices show a similar behavior, it is in the abrasion resistance where their differences are indisputable.
Figure 6d shows that the individual ionic matrix has a higher abrasion resistance index 198 ± 8% than the covalent peer 106 ± 7%. This increase of
86% can be correlated with previous explanation about the strength of the bonds. In the case of the ionic matrix, the bonds have greater intensity and are stronger than in the case of the sulfur-cured matrix [
14,
31]. The dual networks also yielded the highest values of abrasion resistance in this research. A maximum is reached for the 1S-5 ZnO compound with an abrasion resistance index of 229 ± 4%. Comparing the values for the ionic and dual formulations, one can observe increments of up to
15% for the dual matrix and of up to
115% with respect to the covalent matrix (only with sulfur). This confirms a synergistic effect of both networks on this property.
3.4. Understanding the Molecular Dynamics of Dual Networks by Means of BDS
A preliminary assessment of the dynamics of the 1S, 10 ZnO, and 1S-10 ZnO compounds was done by DMA.
Figure 7a shows two relaxation zones. The
α relaxation or segmental relaxation is related to the movement of chain segments and associated with the
Tg of rubber (around −1.09 °C for 1S, 1.71 °C for 10 ZnO, and 4.43 °C for 1S-10 ZnO). This relaxation exhibits considerable changes in the shape of the peak, being wider for the ionic (10 ZnO) and dual compound (1S-10 ZnO), versus the covalent one (1S), as well as being shifted to higher temperatures (see
Figure 7a inset). This could be associated with a greater restriction of the ionic domains, as a consequence of the chains trapped in the clusters and the performance of these clusters as crosslinking points [
37]. A complementary analysis of the glass transition was also performed by DSC (
Figure 7b). The values obtained for the thermal
Tg corroborate the trend of the DMA, in which the purely ionic (10 ZnO) and dual (1S-10 ZnO) networks are presented as more restricted systems than the purely covalent (1S) one. The thermal
Tg by DSC was found at −18.8 °C for 1S, −17.4 °C for 10 ZnO, and −16.8 °C for 1S-10 ZnO (all the values reported refer to the midpoints, however, the trend is maintained both at the onset and endpoint).
In addition to
α relaxation, a second process is detected by DMA (
Figure 7a) at higher temperatures: an ionic relaxation around 50 °C that is not present in the covalent formulation (1S), which is an irrefutable proof of the presence of ionic clusters [
8,
11,
21,
23,
38] in the purely ionic compound, and in the dual one. The differences between 1S-10 ZnO and 10 ZnO could be associated with the compromise between the role of ZnO as vulcanizing agent (for the ionic network) and as a sulfur activator (for the covalent network). In this way, one can corroborate that not all the ZnO in the 1S-10 ZnO would be contributing to ionic bonds, as it does in 10 ZnO.
A systematic analysis was further performed by BDS to gain a deeper knowledge on the dynamics of the three compounds under study. In theory, dielectric and dynamic mechanical analysis should reflect the same motions of the chains and chain segments if the referred motion implies dipole motions; however, some differences in both the intensities and frequencies of the relaxations can be distinguished. In this study, a third weaker relaxation at low temperatures can be detected,
β relaxation, associated with short-range cooperative movements in the polymeric chain.
Figure 8 shows the dielectric loss spectra of the
β and
α relaxations at different temperatures and in a wide frequency range. These two relaxations can be confirmed as thermally activated processes, as evidenced by their shift towards higher frequencies with increasing temperature.
At high temperatures and low frequencies, both the electrode polarization (EP), that partially blocks the charge exchange between the sample and the electrodes, and the accelerated movement of ions generate long capacitances that translate into high dielectric constants and into a considerable increase in conductivity. Such behavior partially masks the third relaxation (ionic relaxation) in the permittivity spectrum (
). Instead, we used the modulus (
) formalism [
39] (
Figure 9a,b). Both variables describe the same electrical relaxation phenomena; however, under different conditions, a specific form offers more information with respect to the occurring physical processes. For instance, the
formalism “converts” the conductivity into a peak associated with the AC conduction and supresses the effects of DC conduction and EP. The frequency range below the peak in
corresponds to the zone dominated by two effects: EP and the movement of charge carriers over long distances. With the suppression of the contribution of EP, the second effect can be detected from changes in the size and shape of the peak. This physical process, called long-range ionic hopping, is related to large-scale molecular movements between ionic domains: it is the relaxation of the clusters. Previous studies have demonstrated that EP and AC conduction itself are consequences of this process [
40]. Other authors [
41] have conducted exhaustive and systematically research that report the superiority of the dielectric spectrum plotted in permittivity (
) rather than modulus (
) in most of studies for slow process due to motion of molecules/ions; however, in our particular case, the modulus gives us complementary information of the relaxations when the dielectric analysis is conducted, which does not contradict the literature but adds new insights.
The individual ionic compound 10 ZnO exhibits an irregular and asymmetric peak, with an observable shoulder in the low frequency range (1 Hz to 10 Hz) (
Figure 9a). This shoulder could prove the presence of a third
α′ relaxation. Some authors have attributed this new relaxation above the glass transition temperature of the matrix to an individual and secondary relaxation in rubber ionomers identified as
α′ and associated with the hard ionic domains [
21,
23,
42,
43]. In the dual network 1S-10 ZnO compound, this shoulder is less notorious (with respect to 10 ZnO), due to the dual role of ZnO (vulcanizing agent and activator) in the compound (
Figure 9b). Once the presence of the third relaxation is detected, it is possible to perform a complex permittivity fitting by using HN functions (for ionic relaxation) and power law (for conductivity), as seen in
Figure 9c,d. Other authors have not been able to detect ionic relaxations through dielectric studies, but a third relaxation (at lower temperatures than those observed by DMA, 50 °C–70 °C lower) has been associated with the Maxwell–Wagner–Sillars (MWS) relaxation [
44]. This relaxation is characteristic of multiphase systems, where each phase has different dielectric constants and conductivities. If one considers the ionic clusters as a polarizable entity, it will appear as a heterogeneity with respect to the matrix, exhibiting this physical phenomenon. In our case, the data obtained coincide in temperature with those observed by DMA. Hence, either way, both analyses are conclusive on the presence of the ionic phase.
For comparative purposes, and with the intention of discerning the effect of the molecular dynamics on the physical properties of the individual and dual crosslinked compounds, we selected a convenient temperature in which each relaxation was well resolved in the frequency domain.
Figure 10 summarizes all the relaxations at the designated temperature (−30 °C for
β relaxation, 0 °C for
α relaxation, and 35 °C for
α′ relaxation). Starting with the
β relaxation, it is slightly affected by the type of network (individual or dual) since the covalent (1S) and dual compound (1S-10 ZnO) seem more restricted at this low temperature. This relaxation can be fitted using a HN function and a power law for the high-frequency tail (front) of
α relaxation.
Table S5 summarizes fitting parameters of all relaxations at the chosen temperature. The nature of the crosslinks also does influence both the
α and
α′ relaxations, as also seen by DMA. A more restricted network is achieved in the dual system (1S-10 ZnO), evidenced by the shift of the dielectric loss to lower frequencies in the segmental relaxation zone (
Figure 10b). This slowing down of the segmental dynamics is a consequence of growing cooperativity. It is evident that, as free volume reduces (due to the crosslinks), more cooperativity is needed to accomplish segmental motions. These molecular dynamics findings are of fundamental interest for understanding, and thus tailoring, properties improvement since the structure of the material can be extracted from the relaxation of polymer chains. Hence, the restrictions on the segmental motions of the rubber chains and the lower free volume convert the vulcanizate into a stiffer and less solvent-penetrable material, improving abrasion resistance and chemical resistance, respectively.
α relaxation can be fitted using a HN function and a power law for the contributions of conductivity and high-frequency tail (front) of
α′ relaxation. Moreover, the
α′ relaxation (
Figure 10c) ratifies the formation of ionic domains in the dual network. These ionic domains also provide higher restrictions to the trapped chains within the ionic clusters. At this point, the differences in the conductivity peak in the compounds that have an ionic phase, compared to the purely covalent phase, are also interesting (
Figure 10d). This could also be an irrefutable proof of the presence of multiple ions, which increase ionic conductivity, reflected in a peak shift towards higher frequencies [
40]. In conclusion, the excellent chemical resistance and abrasion resistance, as well as the retention of tensile strength, confirm the role of these ionic domains in dual networks. To the best of our knowledge, no study has been found in the literature linking the dynamics of these relaxations with those two essential properties in most XNBR applications.
For all temperatures tested, the behavior of the relaxations was evaluated and the frequency at which the maxima appear was recorded. From these values, it was possible to obtain the graph of the logarithm of the relaxation time (
) as a function of the inverse of the absolute temperature (
) (activation diagram) (
Figure 10e). A value of
was set to reduce the effect of data fitting over a limited frequency range, according to previous studies [
45,
46]. With this consideration, a fitting can be made to each relaxation.
Table 3 shows all the fitting parameters. The
β relaxation fits perfectly to the Arrhenius function, which allows the calculation of the relaxation’s activation energy (
). From these values, we can confirm that this relaxation is slightly dependent of the crosslink type. It is observed that the formulations with covalent phase (1S and 1S-10 ZnO) present higher activation energies than the purely ionic phase (10 ZnO).
The
α relaxation departures from the Arrhenius behavior; it exhibits a curvature at high temperatures adjusting perfectly to the VFT function. This relaxation is shifted to the left, which means a higher temperature requirement for its activation compared to the
β relaxation. In addition, the ionic networks (10 ZnO and 1S-10 ZnO) are shifted to higher temperatures (increasing values of
T0), showing also different time scales between compounds. This could be explained by the higher restriction in ionic domains, which hinders the motions of rubber chain segments and slows down the
α relaxation. To deepen the analysis, two quantities can be obtained from the parameters derived from the VFT adjustment. The first of them is known as fragility strength (
), which can be calculated from Equation (14), and is related to the nature of the
Tg (strong or fragile behavior) [
47,
48]:
Second, the fragility index (or steepness index,
) [
47,
48], which is defined as an apparent activation energy and can be estimated according to Equation (15):
In polymeric materials, the strong or fragile behavior is associated with the cooperativity of the segmental chain movements. Thus, highly cooperative materials (fragile, typically with more rigid backbones) will strongly deviate from the Arrhenius behavior and will have low
values (high
), i.e., will require a higher apparent activation energy than those with low cooperativeness [
49]. According to the data shown in
Table 3, a dependence of the
values with the type of crosslink is identified. This can also be correlated with the
Tg values seen as the maximum of the
curves in DMA (
Figure 7). 1S has a lower
Tg, followed by 10 ZnO and finally 1S-10 ZnO. Although 1S and 10 ZnO compounds have equivalent crosslink densities with the swelling method, slight differences are detected in
Tg that may be attributed to the nature of crosslink, and not to an effective increase in the number of crosslinking points. According to the above, it is consistent that the values of
are higher for 10 ZnO and 1S-10 ZnO, and, consequently, the values of
are lower. From the values of
it can be concluded that, in the dual system, the fragility (low
, high
) is closest to the fragility of 10 ZnO because the dual networks impose similar restrictions on the segmental movements of the chains [
45].
The
α′ relaxation adjusts to the Arrhenius function. An attempt has been made to adjust this relaxation with a VFT function. Although adjustment is possible, with a correlation coefficient
0.98, the
T0 values obtained lack physical meaning. Considering the activated character, 1S-10 ZnO deviates from 10 ZnO due to its more restricted network, as reflected in its shifted to higher temperatures. This ionic relaxation could be thought of as a “second glass transition” associated with the hard ionic phase in the rubber matrix that gains mobility with increasing temperature [
8]. Some authors have suggested that, in the presence of ZnO, these ionic domains are an individual and differentiated phase of the rubber matrix, which contains a coordinated complex based on the ionic pairs (Zn
2+ and COO
−), as well as trapped chains. These domains form a three-dimensional zone. This zone would act as a separate polymer phase from the rest of the matrix, with its own “glass transition temperature” [
21] that can be redefined as “ionic transition temperature”. Upon exceeding this temperature, the trapped chains in clusters are released, and the ion pairs acquire mobility due to the ion-hopping mechanism, destroying coordination complexes and ionic clusters. In the dual network, these ionic domains can be surrounded (or even partially constituted) by covalent crosslinks points (S-based) that provide greater rigidity to the system. All these conclusions open a new path to study different effects on combined networks. Varying the size of the ionic domains, the cations involved and/or the proportions of the ingredients of the sulfur crosslinking system will allow for obtaining networks
on demand for specific applications.