As all following characterizations were carried out at neutral pH, the data clearly show that no gel-like behavior is found for the precursor solutions.
Figure 4 shows the results of the creep and creep-recovery test for samples S1 (1.67 wt% SA) and S2 (2.67 wt% SA) [
99]. All samples underwent instantaneous deformation (τ = 50 Pa), followed by a much smaller viscoelastic deformation for the remainder of the 300 s creep step, which – as expected – does not converge to a power law slope of 1, which would indicate a Newtonian creep behavior. This shows that the sample is crosslinked, i.e., has an infinite zero shear-rate viscosity [
100,
101,
102]. The recovery step shows that almost the complete deformation is recovered, again indicating that only a very small tendency of permanent (i.e., viscous) deformation is found, which is what we would expect from a covalently crosslinked gel. The higher the CaCl
2-content of the gels, the lower is the residual deformation. This is easily explained by SA being supramolecularly crosslinked by the Ca
2+-ions. The more Ca
2+-ions, the higher is the probability that the bonding sites on the SA-chain are bonding to a Ca
2+-ion. However, one should remember that only a bond between one Ca
2+-ion and 2 SA-bonding sites will lead to physical crosslinking between 2 chains (with both intermolecular intramolecular bonds, while only the intermolecular bonds will contribute to the network) [
103,
104,
105,
106]. As the bond between a Ca
2+-ion and a SA- bonding site is easier to obtain than between Ca
2+-ion and two SA-bonding sites, too many Ca
2+-ions will weaken the gel once a significant excess of Ca
2+-ions is in the system [
107,
108]. As compliance J and shear modulus G are approximately the inverse of each other, higher strain or compliance J correspond to a lower modulus [
109].
S1-0 and S2-0 show clearly different dynamic-mechanical data with significantly higher damping and lower overall modulus (especially S2-0). This can be explained by the fact that the PAAM-network is more or less unaffected by the content of Ca2+, while the SA-network does not exist, which leads to lots of dangling chains, thus increasing δ and lowering the overall network density, i.e., visible due to a lower G’.
Nonlinear rheological behavior of (PAAM/SA) hydrogels
Figure 6 shows the strain sweeps (
γ0: 0.01 → 10 → 0.01 → 30 → 0.01 → 100 → 0.01 → 300 → 0.01 → 1000 → 0.01 (%)), whose G’ (
Figure 6a) is very similar except for the last leg (
γ0 = 1000 → 0.01 (%)), which indicates that at
γ0 > 300% the sample is irreversibly damaged.
Figure 6b shows that with each cycle, G” increases by several % (while G’ is approximately constant). A peak around
γ0 ≈ 100% appears clearly after straining the sample with
γ0 = 300%, which is much larger after
γ0 = 1000%, which is assigned to damages in the network structure.
Figure 7 (a&d) show the influence of Ca
2+-ions on the strain sweeps (
γ0 =0.1→1000%) for the S1 and S2-series, respectively. The linear viscoelastic regime (G’ and G” independent of
γ0) decreases significantly upon the addition of Ca
2+-ions, while the overall linear viscoelastic values of G’ and G” as well as low damping are not affected very much by presence of Ca
2+-ions (except S2-0).
Sim et al.[
43] classified the large-amplitude oscillatory shear (LAOS) behavior of strain sweeps in 4 categories, according to which all samples behave according to Type
III (weak strain overshoot, G' decreases while G'' increases and then decreases). With an increasing
γ0 , the sample begins to enter the nonlinear region, thus the network structure of the gel samples begins to be distorted, resulting in the appearance of weak strain overshoot. The processes occurring here are ripping open the bonds between Ca
2+ and the corresponding SA-moieties, which gradually turn the SA-gel network into a polymer solution with dangling ends. Furthermore, the PAAM-network is strained, creating similar effects, albeit without breaking ion-dipole bonds. As
γ0 increases, the gel samples show a crossover point (G’ = G”). At higher deformations, G’ and G” decrease sharply, typically with power law slopes of -2 and -1, respectively [
112].
In order to quantify these differences better, four characteristic quantities are defined. The linear viscoelastic G’, the modulus at the G’-G” crossover, the nonlinearity limit, defined by a 5% deviation from the viscoelastic G’, and the deformation at the G’-G” crossover.
Figure 7(g&h) demonstrates that the influence of ion-content on the viscoelastic G’ and modulus at the G’-G” crossover is relatively small, and the characteristic deformations show a very significant dependence. The
γ0 corresponding to the nonlinear limit decreased from 5.2% and 13.7% for S1-0 and S2-0, respectively, to 0.1% and 2.37% for S1-5 and S2-5, respectively, followed by a leveling off at concentrations above ca. 30 mg/ml at
γ0 ≈ 1% [
58,
59]. Furthermore, the G” and deformation
at the G”-peak were assessed, which, however, yielded almost identical values to the G’-G”-crossover, as those coincide closely. It should be mentioned S1 has significantly lower nonlinearity limits than S2 at low Ca
2+-concentrations, which we attribute to the lower SA-concentration, leading to a sparse network, which is not fully bis-complexed and thus easy to destruct. The crossover deformation shows a rather similar relationship, but without the deep minimum found for the nonlinearity limit.
Figure 7 (i) shows the relation between intrinsic nonlinearity
Q0 and the calcium ion concentration, which exhibits a jump between 0 and 5 mg/ml Ca
2+ by 2-3 orders of magnitude, followed by a broad minimum at intermediate concentrations. Knowing that SA has a certain number of bonding moieties and that each of those moieties needs to be complexed with another moiety via a Ca
2+, it becomes clear that one Ca
2+ should be in the sample for every 2 SA-moieties. From previous discussions and reports it is known that more dangling ends in the sample lead to a stronger nonlinearity[
87]. Therefore, the closer the concentration of Ca
2+ is to the optimum concentration of 0.5 equivalent (one Ca
2+ for every two SA-moieties), the lower should be
Q0. While the concentration of the SA-moieties is not known precisely, it is clear that their concentration for the S2-series is 60% higher. Taking the minimum position of the S1 and the S2-series, it is clear that the S2-series has a minimum at higher concentration than the S1 series and quantitatively also the factor 1.6 is justifiable.
The I
3/I
1 curves of all gel samples have similar shapes, as shown in
Figure 7 (b&e). In the small strain amplitude region, the waveform of the stress response is consistent with the input strain waveform, thus I
3/I
1→0 [
64]. The I
3/I
1 value of each sample increases with an increase in
until ca. 10-30% for the samples containing Ca
2+-ions, followed by a slight decrease and finally, levelling off to the plateau value of S1-0 and S2-0 around I
3/I
1=0.1. When looking at this peak in detail it becomes apparent that the peak plateau is virtually reached at the crossover strain of the samples with Ca
2+-ions, followed by ca. 1 order of magnitude before decreasing to the level of the Ca
2+-free sample (S1-0&S2-0). This trend is better observed for the S1-series than for the S2-series as the former is “more fragile”, as discussed before.
To the best of our knowledge such a peak in I
3/I
1 has not been reported before, as no research article on double-network hydrogels with supramolecular bonds and LAOS is known to the authors. Apparently, the supramolecular bonds between Ca
2+ and the corresponding moieties of SA lead to a higher nonlinearity (I
3/I
1) than the covalent PAAM-network. Thus, as long as the SA-network is halfway intact, the nonlinearity is higher than when it has broken down due to excessive shear. This could be explained by the supramolecular bonds of SA breaking and reforming twice within one cycle, which will lead to a stronger deviation from the sinusoidal curve shape. This fundamental difference in the response caused by Ca
2+-ions is also clearly seen in the viscous Lissajous plots of S1 (
Figure 8), other Lissajous plots can be found in SI (
Figure S2-S4), where the curve shapes of
=0.1%, 300%, and 1000% are rather similar, while for the other deformations shown, S1-0 is clearly different from the other samples.
Figure 7 (c&f) show the
Q-parameters, which just like I
3/I
1 are almost indistinguishable for
γ0 > 300%, while for smaller deformations, the samples without Ca
2+-ions show a significantly higher nonlinearity
Q than S1-0&S2-0. From these data, the intrinsic nonlinearity parameter
Q0 was determined.
The Chebyshev coefficients, strain hardening rate (S), and shear thickening rate (T), as defined in the experimental part, can be used to analyze the nonlinear rheological behavior of gel samples in LAOS experiments [
113]. In LAOS testing, the third Chebyshev coefficients (e
3 and v
3) are mainly used to characterize the nonlinear response, and their physical interpretation is related to S and T: when e
3/e
1 > 0, the sample exhibits strain hardening properties similar to S > 0; when e
3/e
1 < 0, the sample exhibits strain softening properties similar to S < 0; when e
3/e
1 = 0, the gel sample is in the LVR region. When v
3/v
1 > 0, the sample exhibits shear thickening properties similar to T > 0; when v
3/v
1 < 0, the sample exhibits shear thinning properties similar to T < 0; when v
3/v
1 = 0, the gel sample is in the LVR region.
As shown in
Figure 9(a, c, and e) and SI
Figure S5 (a and b), as
γ0 increases, samples S1 and S2 undergo a transition from the LVR region to the nonlinear viscoelastic region. During this transition, samples S1-(0, 20-80) / S2-(0, 20-80) exhibit strain softening before strain hardening, while sample S1-(10-15) / S2-(10-15) directly undergoes strain hardening without any strain softening. This indicates that the samples S1-(10-15)/S2-(10-15) have a faster feedback response to strain and stress and can immediately respond when subjected to
γ0 beyond the maximum tolerance of the LVR region, which is also related to its higher strength and G'. The onset of strain hardening indicates that the gel samples have entered the nonlinear viscoelastic region. The reason for the strain hardening of gel samples is that when the gel sample is subjected to
γ0, the network structure of the sample has the ability to resist deformation. However, when the
γ0 exceeds a certain threshold, the network structure of the gel sample is destroyed, and the ability to resist deformation decreases significantly. This can also explain the phenomenon of S peaking, follow by decreasing at large deformations.
Furthermore, it should be pointed out that in comparison to I
3/I
1, the sensitivity to changes in structure is higher, as the deformation at which S, T, e
3/e
1, or v
3/v
1 deviates significantly from the linear values varies systematically stronger. E.g., when looking at the strain
γ0, at which S = 0.5, a relation very similar to
Figure 7 (i) would be obtained.