3.1 Fluorescence Microscopy
The morphology of M-E blends was studied by fluorescence microscopy analysis to characterize the nature of the continuous phase and the grade of dispersion of the discontinuous phase. Fluorescence microscopy showed that even those M-E systems (see Fig. 1a-c) with small concentrations of elastomer (< 1.5 wt. %) are heterogeneous materials composed of two phases: an elastomer-rich one (ERP) and a maltene-rich one (MRP). It was also observed that the concentration of elastomer determines the distribution of these two phases. The elastomer rich phase is composed of maltenes swelled in SBS elastomer particles as can be observed in Fig. 1b. The SBS particles are easily seen as spherical particles within the maltenes continuous phase (black regions). The fact that the dispersed phase is in spherical form and that no linkages are visible between particles indicate that the UM (unmodified maltene) and the elastomer are strongly immiscible and very low interfacial adhesion was presented between the two phases [25;26;27;28;29], [30] suggested that the presence of spherical particles are due shearing and interfacial forces, and a long mixing time. However, the blend with the highest elastomer concentration (2 wt. %) of this group exhibited (see Fig. 1d) an elastomer-rich phase (ERP) with non-spherical particles, revealing that miscibility between the two phases has been improved but with SBS (E) particles acting as the dispersed phase.
From these results, it was concluded that the increase of the elastomer concentration has a stabilizing effect in the morphology of M-E blends (i.e., co-continuous). Nevertheless, it is also important to emphasize that the morphology observed on the surface of the M-E blends does not provide information concerning the bulk structure of the blends, which is likely to affect the mechanical behavior of these blends. Regarding M-E blends with high elastomer concentration Fluorescence Microscopy technique is limited, this is due to phase inversion phenomena, causing swelling of the elastomer and maltenes absorption, making it almost impossible to distinguish the maltenic phase.
3.2 Rheological characterization
Linear viscoelastic behavior of UM and M-E blends was analyzed by means of the G´, G´´, η´, η´´ and δ; the master curves give information in regard of the deformation resistance of these materials when are subjected to shear loading and the structure response to deformation. Cole–Cole, Han and van Gurp-Palmen master curves (see Figs. 2, 3 and 4) of the blends were constructed by means of time-temperature superposition principle [17; 23; 31; 32;33;34;35]. The Cole-Cole curves (see Fig. 2) η´´(ω) vs η´(ω) were used for analyzing the sensitivity of the UM and M-E blends in SAOS flow experiments, the results showed that the UM curve shows a well-defined plateau at low η´ values up to a critical point were the curves show a decreasing monotonical zone, all the blends at low concentration maintained this characteristic behavior (0.25 to 2% wt.), but the critical point is shifted to the right as the concentration increases. The curve for the elastomer (E) in Fig. 2 shows the highest values of η´´ with two slopes and a transition point, at low frequency (left part of the Fig. 2) a high slope zone with an increasing behavior is observed whereas at high frequency (right part of the Fig. 2) the slope changes to a lower value but with a decreasing behavior. This tendency is maintained by all the curves of the M-E blends with high elastomer content (50 and 80% wt.). The transition point between the behavior at low and high frequency for all M-E blends is shifted to the right as the concentration of elastomer is increased. The values of η´ at low frequency are consistently increased with the content of Elastomer.
As expected, all samples followed the TTS principle, that is, modified asphalts are thermo-rheologically simple materials [25; 36] and G’ represents the capacity of the sample to store energy and G’’ the capacity of energy dissipation, hence Han master curves (see Fig. 3) of G’ (ω) vs G’’ (ω) [32; 37] were used for analyzing the non-destructive flow dependent behavior of the UM, E and M-E blends on the frequency. In the Fig. 3, it can be seen a continuous line representing a 45-degree slope which represents the equimodulus (frequency at G’=G’’). All data below the equimodulus line represent dominant viscous behavior and all data above the equimodulus line represents dominant solid behavior. Two series of curves are clearly observed (at low and high content of elastomer in the blends). This is important to highlight since all M-E blends at low concentration and UM fall in the dominant liquid behavior zone and only part of the M-E blends curves at high content and E reach the dominant solid behavior zone at high frequency, this characteristic elastic behavior is imparted to the blends by the high content of elastomer (E).
On other side, the van Gurp-Palmen (vGP) diagram (see Fig. 4) shows two separated groups of curves, the first group corresponds to the UM and M-E blends at low elastomer content (0.25-2% wt.), these curves displays an initial plateau zone where the phase angle is independent of the complex modulus and its value of 90 degrees indicates that the material has a predominant viscous behavior at low frequencies (left part of the Fig. 4), this behavior is caused by maltenes (predominant liquid behavior) and the poor immiscibility between maltenes and the elastomer observed in the microscopy section showing a dispersed phase as observed in the Fig. 1a-c. Whereas, the other group correspond to the Elastomer (E) and the M-E blends at high elastomer content (50 and 80% wt.) where the values of the phase angle are lower indicating that these blends are more elastic materials and tend to the value of a perfect solid behavior (~ 0 Degrees) at high frequencies (right part of the Fig. 4). For this case, the E-M blend structure interacts forming a co-continuous network throughout the binder.
The rheological results suggested a possible formation, to a small extent, of crosslinking which may be formed due to thermomechanical stress during mixing. The presence of such crosslinking needs to be confirmed with other experimental evidence; however, it is consistent and could explain the macroscopic behavior and mechanical properties observed.
3.3 Multimodal Maxwell model
The multimodal model is a self-consistent approximation model that derives its expression based on analysis of deformation and stress within a composite. The M-E blends microstructure is modeled as a continuous medium where E particles are dispersed as the inclusion. Maltenes as the matrix, surrounds E particles at low concentration and generates a co-continuo phase at high concentration. The rheological behavior observed with the models in both low and high elastomer content of the M-E blends can be explained by considering that the response of the blend is mainly determined by the polymer-rich phase. The effect of increasing the elastomer content is to increase the viscoelasticity and complexity of the elastic character of the M-E blends. These results are expected since the rheological behavior of the blends is completely determined by the discontinuous polymer-rich phase at low E content, and then a phase inversion is observed as the behavior changes to a co-continuous phase at high elastomer content.
In the Tables 3 and 4 show the parameters from the multimodal Maxwell model for the mixtures with low and high polymer concentrations. The results show that six modes were used for low concentration M-E blends and unmodified Maltene (UM) (see Table 3), as comparison, UM has the longest relaxation time of low concentration samples; this is because UM is a predominant liquid material and by the addition of Elastomer (i.e., a predominant elastic material), the blend increases its elastic behavior.
Furthermore, nine modes were necessary for 50 (wt. %) in the M-E blend and eight modes for the rest of high elastomer content M-E blends and SBS elastomer (E) (see Table 4). After phase inversion (> 2 wt. %) the complexity of modified maltenes increases abruptly, and more relaxation modes needed to represent experimental data.
Table 3
Relaxation times and shear modulus for M–E systems at low concentration levels and UM.
|
2.0 wt. %
|
1.5 wt. %
|
1.0 wt. %
|
0.25 wt. %
|
UM
|
λ0 [s]
|
6.27x10− 5
|
4.05x10− 4
|
1.82x10− 5
|
1.26x10− 5
|
1.12x10− 4
|
λ01 [s]
|
1.25x10− 5
|
1.52x10− 5
|
1.70x10− 5
|
1.22x10− 5
|
1.34x10− 5
|
λ02 [s]
|
3.45x10− 4
|
3.08x10− 4
|
2.80x10− 4
|
2.30x10− 4
|
3.25x10− 4
|
λ03 [s]
|
2.51x10− 3
|
2.38x10− 3
|
4.40x10− 3
|
2.22x10− 3
|
2.35x10− 3
|
λ04 [s]
|
2.88x10− 2
|
2.04x10− 2
|
3.45x10− 2
|
3.60x10− 2
|
1.70x10− 2
|
λ05 [s]
|
1.06
|
0.672
|
1.58
|
3.14
|
0.127
|
λ06 [s]
|
8.20
|
8.20
|
7.70
|
25
|
10.3
|
G01 [Pa]
|
5x106
|
2.50x106
|
2.48x106
|
2.58x106
|
1.34x106
|
G02 [Pa]
|
3.04x104
|
3.60x104
|
8.12x104
|
4x104
|
10399
|
G03 [Pa]
|
6.54x103
|
4.75x103
|
3.48x103
|
5.21x103
|
859
|
G04 [Pa]
|
610
|
324
|
123
|
138
|
89
|
G05 [Pa]
|
3.93
|
1.56
|
0.89
|
0.22
|
5.772
|
G06 [Pa]
|
0.352
|
7.60x10− 2
|
7.10x10− 2
|
9.00x10− 3
|
0.019
|
Table 4
Relaxation times and shear modulus for M-E systems at high concentration levels and pure E.
|
E
|
80 wt. %
|
50 wt. %
|
λ0 [s]
|
3.52x10− 2
|
2.10 x10− 2
|
9.89x10− 3
|
λ01 [s]
|
1.28x10− 3
|
7.75x10− 4
|
2.36x10− 3
|
λ02 [s]
|
8.00x10− 3
|
5.68x10− 3
|
5.73x10− 3
|
λ03 [s]
|
3.2x10− 2
|
1.92x10− 2
|
1.86x10− 2
|
λ04 [s]
|
4.28x10− 2
|
5.53x10− 2
|
6.9x10− 2
|
λ05 [s]
|
0.1389
|
0.2142
|
0.208
|
λ06 [s]
|
0.4491
|
0.7701
|
0.256
|
λ07 [s]
|
2.04
|
2.88
|
0.956
|
λ08 [s]
|
10.76
|
14.68
|
6.16
|
λ09 [s]
|
--
|
--
|
10.7
|
G01 [Pa]
|
230110
|
76000
|
25361
|
G02 [Pa]
|
95290
|
45600
|
10300
|
G03 [Pa]
|
39100
|
17600
|
8840
|
G04 [Pa]
|
16750
|
11800
|
2620
|
G05 [Pa]
|
15810
|
4166
|
362
|
G06 [Pa]
|
9560
|
1357
|
328
|
G07 [Pa]
|
3520
|
360.35
|
132
|
G08 [Pa]
|
428
|
38.32
|
17
|
G09 [Pa]
|
--
|
--
|
5.67
|
Figure 5 shows the linear viscoelastic response along with the model predictions (continuous line) of UM and 2 wt% E-M blend. As expected, the effect of adding the elastomer (E) on the blend is to increase the moduli even at low concentration before the phase inversion as it was previously observed in the Cole-Cole diagram (see Fig. 2). At 2% wt. Several slope changes can be observed in the G’ curve in comparison to that of UM, this is attributed to a complex polymer-maltene interaction due probably to the improved dispersion of elastomer in the immiscible blend (see Fig. 1d). The minimum number of modes employed in the modeling represent the experimental rheological data with an averaged absolute error less than 3%.
It is important to mention that the terminal region corresponds to a Maxwell’s principal relaxation time (the inverse of the frequency when G’ (ω) = G’’ (ω)) calculated through Equations 1 and 2 and included in the discrete relaxation spectrum for these blends. It is evident from Fig. 5, that the terminal behavior does not occur in the UM data.
The slowest relaxation time corresponds to the terminal behavior for the M-E blend at 2% wt. and accounts for the maltene matrix (continuous phase)-Elastomer (dispersed phase) interaction, while the other six times are associated with the maltenes matrix only. At high polymer content, there is a phase inversion evidenced by the rheological behavior of a co-continuous phase. It is worth to mention that the maxima observed in Fig. 5 corresponds to a theoretical prediction of the model and it is related to the crossover point of the maxwell model where G’’ reaches its maximum value.
Meanwhile, Fig. 6 shows the rheological data and model predictions (continuous line); for M and 50 wt. % M-E blend, as the E content increases, the moduli values approach those of pure E. On the other hand, at 50 wt. % of E, an additional mode was implemented. In terms of number of modes, as the phase inversion approaches, M-E blend complexity increases. The E and the M-E blends rheological data show an experimental cross-over point as observed on Han curve.
In summary, based on these results, when E is well dispersed in the maltene matrix but with poor adhesion (< 2.0% wt.) the polymer-matrix interactions are evidenced as slow relaxation times (i.e., low frequency). Meanwhile, for high polymer content, the fast relaxation processes dominate the rheological behavior of the blends (high frequency region). The number of relaxation times required to model the mixtures is reduced as the polymer concentration is lowered. That is, the complexity of the materials can be interpreted based on their structure imparted by the elastomer content, the viscoelastic response after the phase inversion is mainly due to the physicochemical properties of the maltenes (see Table 1), where the polymer rich phase is a complex network of polybutadiene regions joined together by polystyrene bonds, with the polybutadiene regions providing mechanical strength and the polystyrene imparting strain hardening, which is related to the Tg of the polymer (Table 2) [12].
Finally, the mixtures with higher polymer content form more intermolecular associations than those with lower polymer content. For the above, the generalized Maxwell model allows the introduction of values of relaxation times corresponding to the crossover frequency (see Fig. 7), which gives more validity to the modeling in physical terms. The main relaxation time (Maxwell´s relaxation time) corresponds to the crossover frequency which may not be observed in the low E concentration experimental range but can be theoretically predicted.
According to [38], in the dilute and semi-dilute regime the solute-concentration dependence of the relaxation time in flexible and semi–flexible polymer solutions follows an empirical exponential model and a power law model, respectively; the exponent of the power law description is consistent with those expected by molecular theory [38]. In the case of this work, low E concentration follows an exponential model and high E concentration is described by power law model with an exponent value of 1.7427. It should be noted that, an increase of E content follows a small positive variation of relaxation time in low E mixtures. It is also clear that both low- and high-polymer blends display higher moduli values than unmodified maltenes, UM. Moreover, as E content increases, the values approach those of unmodified E, as expected. At a relatively low concentration, the model appears to be adequate to describe the mechanical behavior of M-E blends. As the polymer concentration increases, E is gradually swollen by the oily fraction of the maltenes and becomes the continuous phase and the number of modes necessary to model this behavior increases due to the complexity of the system with a maximum number of 9 modes for the 50 wt. % E system probably due phase inversion effects. These results indicated that this model could help establish target polymer contents for modified asphalt and could help set rational tolerance range for use during mixture production. Additionally, Multimodal Maxwell model provides a stress relaxation spectrum for further investigations and with possible applications in the chemical industry.