Damage to PE-HD will occur either due to craze-crack failure or due to fracture of the plastically yielded zone after shear deformation. Depending on whether the yield stress or the craze stress is lower, the prevailing micro-deformation mechanism shifts from crazing to shear deformation or vice versa. Crazing is associated with brittle failure and shearing with ductile failure. Thus, transitions in these deformation mechanisms may also be denoted as brittle-ductile fracture transitions. Furthermore, the occurrence of brittle fracture is a safe indicator that craze-crack mechanisms such as SCG and ESC participated in the damage process (section 2.3.6).
There are two possibilities to create internal surfaces in crazing mechanism:
(1) At low temperatures and high strain rates, the characteristic time for disentangling a polymer chain is long and internal surface will be created when the chains break. In this regard, both terms in equation 19 contribute to the surface energy Eis.
(2) At elevated temperatures (with respect to Tg) and low strain rates, chain relaxation becomes
faster compared to the timescale of deformation processes. Additional internal surfaces will predominantly be formed by intermolecular separation and chain disentanglement. The surface energy is then dominated by the van der Waals contribution (Eq. 19) [3].
Since the brittle-ductile transition at elevated temperatures is related to molecular disentanglement phenomena, an increased resistance to crazing and cracking is supposed to be found with increased molecular mass (higher transition temperature and more tie molecules, section 2.1). Longer molecular relaxation times shift the disentanglement transition towards higher temperatures, loading times and lower strain rates. Due to the inherent time dependency, the SCG/ESC mechanisms are dominant at long-time intervals.
Therefore, predominantly brittle failure due to SCG/ESC only occurs, if the external macroscopic stress is appropriate. Qian et. al supposed that predominantly ductile failure results as soon as the external stress becomes much higher than about one-half the yield strength [66]. At that point, a brittle-ductile transition due to a transition from the craze-crack to the shear deformation mechanism occurs. Consequently, tests to examine SCG/ESC behavior have to be performed at low external stresses. In this respect, Qian et al. recommended to apply stresses lower than one-half of the yield strength. However, the results obtained in this study indicate a direct relation of the brittle- ductile transition to the yield strength (sections 5.3.2 and 5.3.3).
It has to be noted that rapid crack propagation (RCP) also leads to brittle failure generated by shock stress (e.g. in Charpy impact tests) [53, 67] and smooth, flat fracture surfaces are obtained. Therefore, they are termed truly brittle fracture surfaces. In differentiation to this, fracture surfaces that result from slow crack growth mechanisms should be termed brittle-like or pseudo-brittle. They might expose macroscopically brittle surfaces but are not truly brittle compared to RCP. This can be concluded from the occurrence of microscopic fibrillar structures that are noticed on SCG/ESC fracture surfaces obtained by scanning electron microscopy (SEM). Such fibrillar structures indicate a microscopically ductile failure (section 5.1.3). In contrast, RCP fracture surfaces also exhibit flat and brittle features on this smaller, microscopic length scale. In this study, mainly SCG/ESC phenomena are addressed that may result in macroscopically brittle and microscopically ductile fracture. Therefore, SCG/ESC fracture behavior is referred to as ‘brittle’ for the sake of simplicity. If the microscopically occurring brittle fracture behavior resulting from RCP is especially intended to be addressed, it is henceforth referred to as ‘truly brittle’ fracture. From these considerations, an assignment of damage mechanisms and fracture surface appearances becomes evident: RCP leads to truly brittle, SCG leads to brittle and shear deformation (SD) results in (macroscopically) ductile fracture surfaces.
In SCG/ESC test methods (section 2.5), only one of the damage periods, crack initiation or crack propagation, is commonly in focus. To examine crack propagation caused by the craze-crack mechanism, a defined imperfection (notch) is usually introduced into the specimen to cause crack initiation prior to loading. Therefore, the contribution of crack initiation (e.g. to the failure time) is deliberately omitted in such tests. The force applied takes effect perpendicularly to the direction of the notch. However, cracks propagate within the notch plane. Due to a reduction of the residual fracture surface area and an associated increase of the local mechanical stress, brittle as well as ductile fracture behavior occurs during a test (first brittle, then ductile behavior) [62,68]. Hence, a distinct fracture surface exhibits characteristic features of craze-crack mechanism (macroscopically brittle) as well as shear deformation (final failure, macroscopically ductile). Therefore, the overall classification of a fracture surface is essentially ascribed to the failure mode (craze-crack or shearing) which dominated under the given conditions and predominantly led to fracture of the material [11]. In practical testing, conditions have to be selected that lead to predominantly brittle fracture surfaces of specimens on a macroscopic length scale since their occurrence is the prerequisite to consider the
test as representative for SCG/ESC. In this respect, the initial stress applied to a specimen is crucial. In FNCT, the integral transition from predominantly brittle to ductile fracture surfaces can be determined by a plot of actual initial stress applied σL over a broad range vs. time to failure tf (Fig. 6)
[13].
Figure 6: Typical stress-time to failure curve of a PE-HD container material over a broad range of actual stresses applied, according to ISO 16770 [13].
SCG/ESC behavior with typical brittle fracture surfaces is attributed to the strong decreasing linear part at lower stresses and higher failure times (Fig. 6) in the double-logarithmic depiction of time to failure tf and initial stress applied σL. Due to a significant tf difference at higher/lower σL, the
transition point (‘knee’ in the curve) can usually be identified easily.