This section starts by presenting the formability limits of the aluminium AA1050-H111 and copper sheets, follows with the morphology of the fractured surfaces, and ends with the determination of fracture toughness.
3.1. Formability Limits
The formability limits by necking (FLC) were determined by means of tensile, Nakajima and bulge tests using the methods and procedures that were previously described in
Section 2.2. The fracture forming limits by tension (FFL) were determined by means of the tensile, Nakajima and bulge tests, plus the DNTTs using the methods and procedures that were also described in
Section 2.2. The shear fracture forming limits (SFFL) made use of the shear tests with different ligament sizes
.
Table 5 presents the fracture loci equations of the FFL and SFFL for the aluminium AA1050-H111 and copper sheets. As seen, the FFL and SFFL of aluminium AA1050-H111 have slopes of −0.68 and +1.38, whereas the FFL and SFFL of copper have slopes of −0.70 and +1.41, respectively. These slopes are different from the theoretical estimates of −1 and +1, because the experimental conditions deviate from the simplifying assumptions that Isik et al. [
5] used in their theoretical model. Despite these deviations, the perpendicularity between the FFL and the SFFL maintains.
Figure 5 presents the formability limits and the corresponding failure strains for the aluminium AA1050-H111. The strain paths for the DNTT, staggered DNTT and shear tests were obtained by means of the DIC system, and all the fracture strain pairs were determined from measurements of the final thickness of the specimens after testing (
Section 2.2).
The staggered DNTT fracture strains pairs present smaller values of the minor strain
with the increase of the specimen’s ligament angle
until a value of 90°, which corresponds to pure shear conditions. Results of the staggered DNTT specimen with ligament angles
of 60, 70 and 80° were revealed as appropriate to characterise the transition mixed-mode fracture region located in-between the FFL and SFFL (refer to the detail in
Figure 5).
A fractography analysis was performed on the fracture surface of the DNTT, staggered DNTT and shear test specimens to investigate the crack opening mode and to correlate the observations with the FFL and SFFL of aluminium AA1050-H111. The SEM images of the fracture surfaces are given in
Figure 6. They were obtained with a magnification of 1500×, and are representative of the entire length of the fracture surface of the specimens.
The analysis of the fracture surface of the DNTT specimen shown in
Figure 6a reveals a circular dimpled structure typical of a normal fracture caused by remote loading orthogonal to the fracture surface. These results are consistent with the fracture strains of DNTT being located on the FFL (
Figure 5), corresponding to the fracture forming limit by tension (mode I).
Analogously, the fracture surface of the shear test specimen shown in
Figure 6c reveals elongated, parabolic dimpled structures that are different from the circular ones due to loading conditions. The open ends of the parabolic dimples are directing the shearing direction, and the overall structure is typical of fracture by sliding caused by in-plane shear. This result is consistent with the fracture strains of the shear test specimens being located on the SFFL (
Figure 5) corresponding to the shear fracture forming limit (mode II).
The fracture surface of the staggered DNTT specimen shown in
Figure 6b reveals a parabolic dimpled structure in-between the typical circular dimpled structure of mode I and the elongated parabolic dimpled structure of mode II. This observation allows us to consider failure by a fracture mixed-mode consisting of opening by modes I and II, which is consistent with the corresponding fracture strains being located in the transition zone between the FFL and SFFL in principal strain space. Moreover, these results are in accordance with a recent work by Gerke et al. [
20], who presented an experimental SEM analysis of the fractured surfaces of a biaxial cruciform X0-specimen under proportional and non-proportional loading conditions.
A staggered DNTT specimen with a ligament angle
and a ligament size
mm was also analysed to investigate the fracture surface of a specimen with a smaller ligament angle located in the transition zone between the FFL and SFFL (refer to the detail in
Figure 5).
Figure 7a presents an SEM picture of the fracture surface side with a magnification of 60× (macro view) to identify the two different locations from which the SEM magnifications of 1500× were taken (
Figure 7b,c).
The first location, identified as region ‘I’ in
Figure 7a, and shown with a magnification of 1500× in
Figure 7b, reveals a near circular dimple-dominated structure typical of normal fracture (mode I). The second location, identified as region ‘II’ in
Figure 7a and shown with a magnification of 1500× in
Figure 7c, reveals an elongated dimpled structure that is characteristic of sheared fracture (mode II). This result reinforces the above-mentioned conclusion that staggered DNTT specimens of aluminium AA1050-H111 fail by fracture in mixed-mode and, therefore, are capable of providing fracture strains in the transition zone between the FFL and SFFL in principal strain space.
Figure 8 presents the formability limits and the corresponding failure strains for copper in principal strain space. As expected, the fracture strains obtained from the DNTT and shear tests are located on the FFL and the SFFL, respectively.
Just as in aluminium AA1050-H111, the fracture strains obtained for the staggered DNTTs present smaller values of the minor strain with the increase of the ligament angle , for the same ligament size . However, and in contrast to aluminium AA1050-H111, all staggered DNTTs of copper present fracture strains on the FFL. In fact, even the specimen with provides fracture strain pairs on the FFL.
The SEM images of the fracture surfaces of selected DNTT, staggered DNTT and shear test specimens of copper are given in
Figure 9. The main conclusion arising from the observation of these images is that surface fractography fully corroborates the results plotted in principal strain space (
Figure 8). In fact, the fracture surface of the DNTT and of the staggered DNTT specimens shown in
Figure 9a,b reveal circular dimples typical of normal fracture (mode I), whereas the fracture surface of the shear test shown in
Figure 9c reveals elongated parabolic dimples that are characteristic of sheared fracture (mode II).
The main difference to the results previously obtained for aluminium AA1050-H111 is that staggered DNTT specimens of copper fail by normal fracture and, therefore, are unable to provide fracture strains in the transition zone between the FFL and SFFL in principal strain space. The explanation for the different results provided by the staggered DNTTs of aluminium AA1050-H111 and copper is attributed to the fact that crack opening modes are not solely dependent on the geometry of the specimens, but also on the material properties, namely on strain hardening. This explanation is better understood by observing the experimental distributions of the major strain
at the onset of fracture for both materials, obtained from DIC (
Figure 10).
As seen in
Figure 10a, the plastic deformation region of DNTT specimens of copper is wider than that of aluminium AA1050-H111, and a similar conclusion may be drawn for the staggered DNTT (
Figure 10b) and shear (
Figure 10c) test specimens. The consequence of the plastic deformation region being wider is two-fold. In one hand, it justifies the deviation of copper staggered DNTT specimens from pure shear conditions, because larger strain hardening coefficients
diminish localisation effects, and therefore, reduce the absolute values of strains at the onset of fracture. This justifies the reason as to why copper staggered DNTT specimens could not provide fracture strains in the transition region between the FFL and SFFL in principal strain space.
On the other hand, it also justifies the larger slopes of the linear fittings of the total specific work per unit of area that were obtained for the copper specimens, when compared to those obtained for the aluminium AA1050-H111 specimens (
Figure 11). This will be analysed in more detail during the next section of the paper focused on the determination of fracture toughness.
3.2. Fracture Toughness
Determination of fracture toughness
by means of the essential work of fracture (refer to the method described in
Section 2.3), was successfully applied to the DNTT, staggered DNTT and shear tests that are listed at the bottom half of
Table 4.
Figure 11 presents the evolutions of the total specific work
as function of the ligament length
for aluminium AA1050-H111 (
Figure 11a) and copper (
Figure 11b).
As seen, the evolutions are linear, and are in good agreement with the results obtained by Cotterell et al. [
13], namely in what concerns the increase in the slope of the linear fitting of the total specific work
with the specimen’s ligament angle
. This means that the slope of the total specific work increases from the DNTT specimens (
= 0°, mode I) to the shear specimens (
= 90°, mode II).
Table 6 presents the experimental values of fracture toughness for the AA1050-H111 aluminium and copper sheets obtained by extrapolating the total specific work
to the limiting condition where the starting ligament length
is zero (refer to
Section 2.3).
The values obtained for aluminium AA1050-H111 can be classified into three different groups, corresponding to fracture toughness in opening mode I (, mode II ( and mixed-mode (. The highest value was found for fracture toughness in mode II , but the difference to the values of corresponding to normal fracture surfaces (DNTT and staggered DNTTs with ) and corresponding to mixed-mode fracture surfaces (staggered DNTTs with ), is small.
Moreover, the results included in
Table 6 also allow a conclusion that fracture toughness
of staggered DNTTs specimens of aluminium AA1050-H111 with
is independent from ligament angle
. This result is opposite to that observed by Mai [
21] for aluminium B1200-H14, which has a strain hardening coefficient
5, comparable to that of aluminium AA1050-H111
.
In contrast, the values obtained for copper can only be classified into two groups corresponding to fracture toughness
in opening mode I (
and mode II (
. This is because the experimental fracture strains in principal strain space (
Figure 8) and the SEM images of the staggered DNTT fractured surfaces did not reveal failure by mixed-mode (
Figure 9). In other words, it was not possible to determine fracture toughness
.
This last conclusion is important, because it points out the paramount importance of combining the formability limits in principal strain space, and the SEM images to characterise the opening mode of staggered DNTT specimens. Otherwise, one may be wrongly assuming a type of fracture, when in fact it does not exist.
Under these circumstances, the main conclusion derived from the experiments with copper is that fracture toughness determined from the entire set of DNTT and staggered DNTT specimens show a tendency of diminishing with the increase in ligament angle . The highest value is obtained for DNTTs (= 0°), and the lowest value is obtained for the staggered DNTT with = 85°.
The above-mentioned results reveal some dependency on the ligament angle , in contrast to what was found for the tests performed in aluminium AA1050-H111, and the explanation may once again be related to the differences in strain hardening (: Aluminium AA1050-H111 and : Copper). In fact, high strain hardening coefficients, leading to significant levels of strengthening during plastic deformation, seem to prevent mixed modes in staggered DNTTs with the staggered geometry, because smaller circular dimpled structures (typical of less ductile materials) have more difficulty evolving into elongated parabolic dimpled structures, as the stagger angle increases more than coarser circular dimpled structures (typical of ductile materials).