3.1. Absorption Measurements
The comparison of the Hansen solubility parameters of the polymer material and the solvent already allows an assessment of the interaction and aggressiveness of the solvent to the polymer. In particular, the ESC behavior depends on the size of the solvent molecules and the molar volume of the SCA which depends on the molecular structure. The smaller the size of the solvent molecule, the better the solvent penetrates into the polymer material. The more similar the Hansen solubility parameters of solvent and polymer are, the more aggressive the solvent. The Hansen and Hildebrand solubility parameters of PMMA and the SCAs used, and their molar volumes are listed in Table 1
Comparing only the molar volume of the SCA, DI is the solvent with the lowest molar volume and thus the highest diffusion rate in PMMA. Therefore, EG and NH should be absorbed more slowly.
In contrast to the molar volume, NH comes closest to PMMA with 15.3 when comparing the total Hildebrand parameter δT of the solvents with the value of PMMA (20.0). This indicates that NH is more aggressive compared to DI and EG as it has higher chemical compatibility. Therefore, NH may dissolve PMMA faster over time or at elevated temperatures.
To get a first insight into the absorption properties of DI, NH, and EG in PMMA, absorption tests were performed. The uptake of the SCA was measured by the increase in the weight of the samples. The results are shown in Figure 2
at 23 °C (a) and 48 °C (b).
None of the SCAs led to a strong dissolution of PMMA at 23 °C or 48 °C. Thus, all solvents can be used as SCA for the ESC measurements under dynamic loading. The absorption curve for DI is approximately Fickian, with a weight gain after 100 d of 1.5% at 23 °C (a). This is the expected behavior for a liquid with limited chemical compatibility but small molecular size [20
]. Swelling of the material was not observed. The investigations at 48 °C (b) showed that the diffusion takes place at a higher speed and that a plateau is reached after about 50 d. The diffusion rate in PMMA is therefore much higher compared to lower temperatures. A possible explanation regarding the large error bar is the strong diffusion rate on one side and the rapid increase in interaction between fluid and polymer. The possible movement allows for water to diffuse into the polymer but also out of the polymer. The steady increase in weight achieved in 100 days at room temperature is achieved 10 times faster.
Compared to DI, the diffusion behavior for EG and NH is completely different regardless of temperature. Despite the strong difference in molar volume—the value for EG is much smaller than the value for NH—and δT
(NH is much closer to the value of PMMA than EG), both SCAs initially showed only a small weight loss over the measurement time (less than 0.5%). This could be due to the solubility of mobile components in PMMA. This effect was previously described in the literature [19
]. From the results shown, it can be concluded that there is little or no interaction of the solvent with PMMA. Fracture-mechanical fatigue crack propagation tests, in which diffusion plays a minor role, can help to investigate the influence on the mechanical properties in more detail.
3.2. Fatigue Crack Growth Experiments
For the second part, the influence of SCA on the mechanical behavior of PMMA in the fatigue crack propagation test was investigated. The focus was on how the SCA and the wetting process affect crack propagation. This was assessed by using the characteristic values: ΔKth
, the slope of the FCP curve in area II, and Kcmax
. Figure 3
shows the FCP measurements for the pure PMMA without medium and the ESC behavior for the type of wetting, (a) previous immersion of the CT sample with crack for 30 d at 48 °C and in situ wetting, (b) the entire sample up to the crack tip.
When comparing the crack propagation curve of the pure PMMA material and the samples previously immersed in EG (blue curve) and NH (green curve) (a), no influence can be seen after 30 d at 48 °C storage condition. All three curves are almost congruent. The characteristic values ΔKth
, steepness of the curve and Kcmax
are identical within the measurement accuracy (Table 2
). The analysis of the fracture surfaces of the specimens by SEM support our findings (Table 3
). The fracture surfaces of the pure PMMA and the previously stored samples with EG and NH all look identical. These results are somewhat surprising since FCP measurements are very sensitive and can detect minimal differences in crack propagation behavior. From this it can be concluded that there is no influence of EG and NH on the crack propagation behavior under these storage conditions. Even the crack of about 1 mm in the CT sample does not accelerate the diffusion of SCA into the sample or to the crack tip. When comparing the fracture surfaces of the pure PMMA and the samples 8N/NH/st48/30d and 8N/EG/st48/30d previously stored in EG or NH, no differences are discernible. All surfaces are structureless and smooth, as is usually the case with a brittle material such as PMMA.
A possible reason for the fact that no influence was visible may be due to the way the samples were stored in the medium. Only a layer close to the surface was saturated with the medium. No medium evidently penetrated through the crack into the crack tip running transversely to the surface, since the crack apparently closed again due to relaxation to such an extent that the capillary forces were not sufficient to transport the medium into the volume in front of the crack tip by diffusion. These results are in contrast to the results of Arnold [19
]. He found a strong decrease in the elongation at break and thus in the embrittlement in tensile tests on samples that were only immersed in EG for 1 min. In the tensile test, the surface has a completely different meaning than in the CT sample in the FCP test. Here the crack runs across the volume of the sample.
The FCP measurements show a completely different behavior when the samples are wetted in situ with EG and NH (b). Continuous wetting of the sample and thus the crack tip with EG leads to the same ΔKth
as with pure PMMA. However, the crack growth accelerates more slowly, as shown by the lower gradient in region II (Table 2
). The water that has diffused into the crack tip leads to a softening effect and thus to a delayed unraveling of the molecular chains. This is associated with a more ductile crack propagation behavior. This also results in an increase in Kcmax
from 0.82 MPa√m for neat PMMA 8N to Kcmax
0.99 MPa√m for the EG treated sample. In addition, the experimental scatter of the measurement points is significantly higher at higher K values in the samples wetted with SCAs. This may be due to the fact that the solvent has less time at higher da/dN values to diffuse into the crack tip and interact with the PMMA material there. Thus, the crack is stopped by EG or proceeds normally as in the pure material (lamination). Here, too, the molar volume and the chemical nature play an important role. Because there is no diffusion into the material, the SCA can only act while being squeezed into the crack tip while the crack tip is dynamically (10 Hz) opening and closing. The influence described is clearly visible in Figure 4
. Starting from the edge of each lamella in the direction of crack propagation, ramps and threads appear due to plastic deformation (marked by arrows).
The medium NH causes a fundamentally more brittle fatigue crack propagation behavior compared to EG and compared to pure PMMA. The characteristic values show a shift of the crack propagation curve to the left (Table 2
). This embrittlement is due to a lower solubility according to Hansen and Hilde. The fracture surfaces of the NH-wetted sample have a much finer striation pattern and are therefore less prone to crack propagation. As in EG, striation formation can also be observed in NH, but the distance is much wider.
The factors responsible for the different behavior of the two wetting methods in the fatigue crack propagation test are, on the one hand, the lower diffusion rate of the medium and, on the other hand, the chemical interaction of EG and NH in the PMMA. Previous wetting of the CT sample despite the presence of a crack has no direct influence on the micromechanical behavior under dynamic loading. The in situ wetting process allows the SCA to act directly at the crack tip, where it can directly influence crack propagation behavior, which is reflected in a more brittle or more ductile material behavior. The physical and chemical nature of the SCA plays a decisive role in the mode of action of the solvent.
An interesting behavior was exhibited by DI in conjunction with PMMA. Figure 5
a shows the fatigue crack propagation behavior between the in situ wetted sample and the sample previously immersed for 30 days at 48 °C prior to testing. In (b) the comparison of the influence of different storage times on the dynamic-mechanical behavior is shown.
In situ wetting of PMMA with DI (Figure 5
a) leads to the same results as with EG. A shift of the entire curve to the right implies a more ductile behavior with the same slope of the crack propagation curve in region II. Compared to the large influence on the crack propagation behavior, there are no major differences at the microscopic level visible (SEM images shown in Figure 6
). In comparison to NH and EG, no formation of stirrup is visible in area II. This could be due to the accelerated diffusion of DI in the PMMA. It can therefore be assumed that the crack tip is always slightly wetted with DI, regardless of whether the sample is loaded or unloaded.
The crack propagation behavior of CT samples stored 30 days in DI at 48 °C differs significantly from the samples with in situ wetting. Up to a ΔK of 0.44 MPa√m, the curve is as usual. With an increase in ΔK, the crack propagation rate, da/dN, also increases. With a higher ΔK, the crack propagation rate remains almost constant at a value of 10−4
mm/cycle in the double logarithmic scale, regardless of the ΔK value. Below ΔK of 0.44 MPa√m, the crack propagation curve is similar to that of pure PMMA. This is also reflected in the comparison of the region I fracture surfaces of the two CT-specimens in the SEM (Figure 6
). When this plateau is reached with slope = 0, the crack grows at a constant propagation rate despite an increase in ΔK (slope = 0). This plateau suggests that it is caused by DI diffusing into the crack tip. In this load range, this can lead to the crack in the PMMA growing as a result of the macromolecules continuously disentangling as a result of the penetration of the medium and the associated higher molecular mobility. This slower process of disentanglement is only possible in this low crack propagation range. Above a ΔK value of about 0.6 MPa√m, regular crack growth begins again, which is then presumably dominated by chain scission. Then, the crack proceeds in the same way and identically as in situ wetted samples. However, Kcmax
decreases from 1.03 MPa√m for the in situ wetted sample to 0.86 MPa√m, which is the same as for pristine PMMA.
This decrease in Kcmax could be a consequence of a lower local temperature at the crack tip. Comparing the curves of 8N with those of 8N/48, please consider that the PMMA was measured under temperature, a shift in the curve and therefore in Kcmax can be seen at the same level as the change for the samples with the two different wetting methods.
The plateau is also clearly visible on the fracture surface of the PMMA sample in area II (Figure 6
). The fracture surfaces in area I of the pure PMMA and the stored sample (8N/DI/st48/30d) are very similar. Upon reaching region II, a visible change in the appearance of the fracture surface can be seen. Compared to the still quite smooth surface of pure PMMA, the sample stored in DI shows plastic deformation due to the formation of striations. Due to the high plastic deformation, so-called humps appear at the edges of the ramp, which give the ramp edge a bead-like thickening. In addition, the formation of crazing can be observed. As previously described, this is evidence of a significant slowdown in the propagation rate.
In a further experiment, the influence of the storage time on the plateau in the propagation rate was examined. Crack propagation curves are presented for the samples stored in DI at 48 °C for 14, 30 and 60 days (Figure 5
b) together with the pure PMMA and the PMMA with in situ wetting.
The difference is that with increasing storage time in the medium and thus longer time for the diffusion of DI into the sample, the da/dN value at the plateau decreases (values in Table 3
). This can be attributed to the diffusion depth and the formation of the plastic zone in front of the crack tip. The plastic zone increases in diameter at higher loads and, at a certain load, hits the volume in the PMMA that has already been penetrated with DI. Below this load, the plastic zone only spreads in the non-wetted part of the PMMA sample. This is also reflected in the fracture surfaces of this plateau at the different immersion times.
Another difference lies in the fact that a shift in the overall curves is observed with longer storage times. This effect is best observed below the plateau where DI has no effect on crack propagation behavior. Comparing the curve below the plateau, the shift leads to the values of PMMA measured under temperature (brown curve of sample 8N/48). Here, the temperature leads to slight embrittlement, which can be clearly seen in the SEM images of area I in Figure 6
. Compared to the pure PMMA sample without temperature and SCA treatment, a smoother fracture surface is observed in the dipped samples under temperature. This served as evidence of why a left shift is visible. This does not come from the SCA but from temperature storage, which is shown by the measurement under temperature.
The formation of a plateau with a constant crack propagation rate that is independent of ∆K only occurs in tests with storage in DI. In this case, there is a complex interplay between the diffusion depth of the medium and the geometric formation of the plastic zone around the crack tip. The longer the immersion time in DI, the deeper the medium has already penetrated through the surface into the sample. Due to the physics of Fick’s law, it is not possible to saturate the entire sample (thickness 4 mm) with DI in a realistic experimental time. Therefore, the stored samples are still dry in the core region. The amplitude of the stress intensity factor ΔK at the crack tip creates a plastic deformation zone. With increasing stress intensity ΔK, the zone increases in volume. At a certain stress intensity ΔK, the formation of the plastic zone is strongly influenced by the medium in the surface layer of the specimen—perpendicular to the crack propagation direction (see arrow in Figure 6
sample 8N/DI/st48/60d). The medium in the surface layer has an additional plasticizing effect, so that an increase in the stress intensity is compensated for by further volume work in the form of disentanglement phenomena in the surface layer penetrated by the medium. Above a certain threshold, this causes only a very small increase in the crack propagation rate (da/dN) (plateau area). If ΔK increases further, the crack propagation is accelerated again, since micromechanically, it is no longer the disentanglement processes of the macromolecules which dominate the crack propagation, but rather chain fracture. This means that in a certain ΔK area, particularly the penetrating medium delays the crack initiation. The ΔK level that has to be reached for the penetrating medium to influence crack growth depends strongly on the storage time at 48 °C.