Tearing behaviour of two types of leather: A comparative study carried out at the local scale using the full kinematic and thermal field measurement techniques

Leather materials undergo various strain and stress states during their elaboration process and their application in numerous different functions. Among the key properties required for such materials, tearing resistance appears as one of the most important. In this paper, the tearing behaviour of two types of leather, a grain pigskin leather and a grain calf leather, was investigated at the local scale using the full‐field techniques. During the tests, thermal fields were measured at the surface of the two leathers by means of an infrared camera. Measurements of the displacement and deformation fields were also performed at the surface of the pigskin sample using the digital image correlation technique, which was not possible for the calf sample due to surface wrinkling. The results obtained enable us to discuss and compare the tearing resistance of both leathers in terms of the thermal activity in the zone of influence of the crack. The best tearing resistance was obtained for the grain calf leather that has undergone a retanning operation and whose matrix contained a plasticiser.


INTRODUCTION
Although the experimental mechanical response of leathers under tension has been studied in the literature for decades, [1][2][3][4][5][6] there are very few studies, which have investigated the tearing behaviour and failure of leather. These were carried out at the macroscopic scale and dealt with fracture resistance. In these studies, fracture energy was preferred to tensile strength or the elongation at break in order to characterise fracture resistance properties. Various fracture resistance factors have been identified: In the case of chrome-tanned bovine hides, Liu and McClintick (1997) [7] showed that water acts as a plasticiser and enhances fracture resistance, that fracture energy starts to decrease once the moisture content increases to around 90%, and that the sampling angle has little effect on the fracture energy of leather (contrary to the case of tensile strength). Furthermore, the effect of strain rates on fracture energy is not straightforward. The same authors investigated tearing tests versus tensile tests on chrome-tanned leather to clarify the difference between tear-ing and tensile behaviours and their fracture mechanisms. [8] For the tearing tests, a unique zigzag fracture pattern was observed due to the fluctuations in the tearing force, whereas the tensile tests demonstrated a uniform fracture pattern. Statistical analysis showed that tensile strength did not correlate well with tearing strength, whereas a good correlation was observed between fracture energy and tearing strength.
Such results were obtained at the global scale, and information at the microscopic scale is necessary to explore deformation mechanisms in more detail. Indeed, characterising full strain and thermal fields has been shown to enrich fracture mechanics approaches in many materials (see for instance Samaca Martinez et al. (2015) [9] for elastomers, Mogadpalli and Parameswaran (2008) 10 for composites, Molteno and Becker (2015) [11] for polymers, and Roux and Hild (2006) [12] for ceramics) by providing information on the mechanical and thermal states in the zone of influence of the crack. Some full-field measurements have already been performed using the infrared (IR) thermography in order to analyze temperature variations at the surface of leathers. Two types of approach are reported in the literature: • application to non-destructive control of final leather products. [13] In this case, pulsed-phase thermography was performed to evidence hidden scratches and the texture in depth and, finally, to evaluate the quality of the leather; • analysis of the thermal response of leathers under mechanical loading. [14][15][16] More precisely, applications concerned self-heating under cyclic loading, with the main objective of determining a limit of acceptable damage for leather products.
Nevertheless, none of these studies put into perspective the thermal response associated with the tearing resistance of leather. The present study aims to provide results on the tearing behaviour of leathers at the microscopic scale, that is, in the crack zone of influence, by using the digital image correlation (DIC) and IR thermography techniques. The paper is organized as follows: The sample geometry and loading conditions are described in Section 2. Section 3 presents the results obtained and the comparison between the two types of leather tested. Concluding remarks close the paper.

Materials and sample geometry
Two types of leather were studied, a grain pigskin leather and a grain calf leather, denoted leathers A and B, respectively, in the following. Animal skins are usually sliced into several layers: Grain leather (also called top-grain leather or top-side leather) is the upper part of the hide, whereas split leather is the underside. Samples of 17.5 mm in width were cut into plates whose thickness was equal to 1.0 mm for leather A and 1.8 mm for leather B. It should be noted that a retanning operation was carried out on leather B, not on leather A. Figure 1a depicts the sample geometry. The initial distance L between the two grips of the testing machine was 60.5 mm for leather A and 62.2 mm for leather B. A 5-mm length crack was pre-cut in both samples using a razor blade. Note that the sample geometry induces a plane stress state under tension, which is a classical mechanical state for studying tearing resistance and crack growth. Figure 1b shows the spectra of both leathers, recorded in reflection mode using a Nicolet 380-Fourier-transform IR spectroscopy spectrophotometer equipped with a Thunderdome-attenuated total reflectance (ATR; 4 cm −1 , 32-scan, single reflection ATR accessory with a diamond crystal). Both spectra exhibit classical zones for leather materials as follows: a broad band around 3,300 cm −1 attributed to the presence of numerous NH/OH functions; strong absorption bands at 1,650 and 1,550 cm −1 ascribed to the amides I (CO stretching) and II (CN stretching and NH bending) of the leather proteins; and a fingerprint region from 1,451 to 1,000 cm −1 attributed to the CH 2 wagging, CH 3 deformation, C-N stretching, and C-OH stretching of leather proteins. However, leather B exhibits in addition a band at 1,740 cm −1 (see arrow in the graph) that can be related to the presence of an ester group COOR: This chemical function is usually related to the presence of a plasticiser inside the leather matrix, typically a fatty acid used during the retanning operation. The presence of plasticiser in leather B was confirmed by thermogravimetric analysis (TGA) performed with a Mettler Toledo TGA/differential scanning calorimetry 1 with 20 mg of material introduced in a 100-L alumina crucible under airflow (see Figure 1c). The degradation of both leathers was obtained through a temperature increase from 30 to 600 • C at a rate of 10 • C/min, leading to successive weight decreases: Loss of water at 100 • C and degradation of leather fibers/proteins at approximately 300 • C are visible for both leathers; the dramatic decrease located at 180 • C for leather B is correlated with the loss of plasticiser.   Figure 2 presents a photo of the experiment. In the present study, the pre-cut sample was stretched symmetrically in one direction (vertical) by using a home-made biaxial testing machine. The crack zone of influence thus remained at the center of the thermal images captured by the IR camera during the test. The cell load capacity was 1094 N. It should be noted that this machine is composed of four independent RCP4-RA6C-I-56P-4-300-P3-M (IAI) electrical actuators (only the two vertical ones were used in this study). They were driven by a PCON-CA-56P-I-PLP-2-0 controller and four PCON-CA (IAI) position controllers. The actuators were piloted by an in-house LabVIEW program. Tensile tests were performed until failure at a loading rate of 100 mm/min. The optical and thermal cameras were located on either side of the sample.

Full kinematic field measurement
Images in the visible domain were acquired at a frequency of 25 Hz with an Imaging Development Systems' camera equipped with a 55-mm telecentric objective. The displacement field at the sample surface was determined using the DIC technique. This consists in correlating the grey levels between two different images of a given zone, each image corresponding to a different strain level. In order to improve image contrast, black paint was sprayed onto the surface before testing the samples. This leads to a random grey field. Uniform lighting of the sample surface was ensured by an LED lamp. The software used for the correlation process was SeptD. [17] For the image correlation process, the zone containing the crack is initially removed from the region of interest (defined in the undeformed state). Furthermore, the SeptD software takes into account the finite deformation framework and aspects linked to the geometrical transformation of the zones of interest (ZOI). The charge-coupled device (CCD) of the camera has 1,920 × 1,200 joined pixels. The camera was fixed on a multidirectional adjustable support, and the distance between the sample and the CCD matrix was about 100 cm. In this configuration, an area of 16.8 × 11.2 mm 2 in the sample plane was observed by the digital camera. The spatial resolution of the displacement maps, defined as the smallest distance between two independent measurement points, was set to 10 px (the ZOI size) corresponding to 0.35 mm.

Full thermal field measurement
Temperature measurements were performed using a forward-looking IR camera with a focal plane array of 640 × 512 px and detectors operating in wavelengths between 1.5 and 5.1 m. Integration time was equal to 1,000 s, and the acquisition frequency was the same as for the kinematic images, that is, 25 fps. The thermal resolution or noise equivalent temperature difference was 20 mK for a temperature range between 5 and 40 • C. The thermal image resolution was equal to 512 × 88 px. The spatial resolution of the thermal maps was 157.7 m/px. The IR camera was turned on 3 hr before testing in order to ensure internal temperature stabilization. The calibration of the camera detectors was performed with a black body using a one-point non-uniformity correction procedure. The surface emissivity of the samples was considered close to the human skin one and was set at 0.9. Figure 3 presents the mechanical response obtained in terms of force versus displacement for leathers A (Figure 3a) and B (Figure 3b). It should be noted that leather A was slightly stretched before the displacement was set to 0, explaining why FIGURE 4 Displacement fields measured for leather A at different times during the test the force is equal to 20 N at the beginning of the test. Displacement is therefore underestimated by 0.8-mm maximum. Comparing the two mechanical curves obtained clearly shows that leather A is much stiffer than leather B, even if the tests were performed with samples of different thicknesses and lengths (1 and 1.8 mm in thickness for samples A and B, respectively, 60.5 and 62.2 mm in length for samples A and B, respectively). Contrary to leather B, the mechanical response of leather A is quasilinear. The strain at failure is approximately (11/60.5) 18%. For leather B, it is equal to (83/62.2) 133%. Leather B therefore exhibits better crack growth resistance than leather A. These differences in the mechanical response of the two leathers can be explained by the presence of plasticiser in leather B (see Section 2.1).

Mechanical response
For both tests, various visible-light images are also presented in Figure 3 with respect to time in order to highlight how the crack deforms up to its final failure. The white halo behind the sample corresponds to the lens of the IR camera. Images were all acquired with the same resolution; different magnification levels were then applied numerically to better highlight the crack zone for each sample. By comparing the final and reference images, it is observed that lateral contraction in leather B is much greater than in leather A. This ability of leather B to deform laterally is in good agreement with the mechanical curve shown in Figure 3b. Furthermore, the crack tip becomes less and less singular for leather B, which limits the stress concentration intensity and delays the final failure. In this case, the crack tip resembles those found in rubbery materials. [18] This was not observed for leather A.

Kinematic fields
Kinematic fields at the sample surface were obtained using the DIC technique. Results obtained for leather A are presented in Figure 4 for the four times chosen, t 1 = 0.4 s, t 2 = 2 s, t 3 = 4 s, and t 4 = 5.88 s. It should be recalled that each datum was obtained for a ZOI of 10 × 10 px. The size of the region of interest was 480 × 320 px, corresponding to an area equal to 16.8 × 11.2 mm 2 . The displacement scale used was the same for all the images, enabling us to compare the amplitudes for FIGURE 5 Lagrangian strain fields for leather A at different times during the test each component and both leathers. Displacements along the y-axis show that the test was quasisymmetrical before the crack growth commenced. From t 3 = 4 s on, the displacement field was no longer symmetrical. In particular, it is visible at time t 4 = 5.88 s that the crack has bifurcated, increasing the dissymmetry level in the displacement field. For time t 4 = 5.88 s, the analysis can only be carried out qualitatively in the zones located on both sides of the crack, as significant out-of-plane displacements occurred (see Figure 3a). Figure 5 provides the deformation field, in terms of the Lagrangian strains xx , xy , and yy , at times t 1 -t 4 . These maps clearly highlight the observation that the zone on both sides on the crack deformed slightly and that the strain was mainly concentrated at the crack tip. As the test progressed, the strain field became less and less symmetrical, and higher strains were concentrated in the zone corresponding to the crack propagation path. The crack path could thus be predicted from the strain field. As the lateral contraction of the surface of leather B was significant and the surface became wrinkled as soon as the loading was applied, only leather A was analyzed using the DIC technique. This effect is visible in the images presented in Figure 3b. The wrinkling induced did not enable the DIC software to make correlations between the ZOIs. Figure 6 presents thermal images at times t 1 -t 4 during the extension of pre-cut leather A. As the IR camera was placed on the opposite side of the sample, the crack now appears on the left side in the images. In order to highlight temperature gradients due to strain/stress concentration at the crack tip, the superimposed white curve provides the temperature along a horizontal profile from the crack tip (see white line in the figure). The scale of each thermal map is given by a color bar. The scale of the temperature profile is given on the left-hand side of each map. As soon as the sample begins to extend (see t 1 = 0.4 s), a strong temperature gradient is observed in the crack tip vicinity. This gradient takes place over 5 px, corresponding to 0.79 mm. The maximum temperature value reached 25.2 • C, whereas the mean temperature outside the crack influence zone was equal to 24.75 • C. When the global stretch increased, the temperature at the crack tip also increased, as did the size of the crack zone of influence as follows: 2.05 mm (maximum temperature equal to 25.7 • C) at t 2 = 2 s and 2.84 mm (maximum temperature equal to 25.9 • C) at t 3 = 4 s. The crack shape remained singular. Nevertheless, just before the final failure, the crack propagated, and a strong dissymmetry occurred in the temperature field due to the crack bifurcation. In this case, the horizontal line along which the temperature profile is plotted no longer coincides with the crack tip zone. It is to be noted that just before failure, the temperature gradient was much higher than the previous; the maximum value of the temperature reached was equal to 44 • C (see the color bar of the thermal map), which is very high compared with the rest of the sample (remaining at about 25 • C), and the size of the crack influence zone strongly increased (about 4.30 mm in length). Figure 7 gives the temperature field at time t 3 . The thermal data are the same as in  full-field technique to predict the crack path. Figure 8 presents the results obtained for leather B. They strongly differ from those obtained for leather A and can be summarized as follows:

Thermal fields
• As previously observed from images in the visible domain, the crack deformed much more than in leather A. The crack shape was no longer singular. • The crack propagated symmetrically.
• The temperature gradient at the crack tip was very low, almost equal to 0 during the first 18 s (temperature field quite homogeneous). Then, a temperature gradient appeared and was present over the whole cross section. It should be noted that a heat diffusion effect is probably involved (the test duration was 8.5 times longer than for leather A), which could increase the size of the temperature gradient zone along the test. The size of the crack zone of influence could therefore be smaller than that where the temperature gradient is observed. • Just before failure, the maximum temperature value at the crack tip reached 30 • C, which is much lower than for leather A. Figure 9 summarizes the results obtained for both leathers. Note that the samples are presented here in their undeformed state (Lagragian configuration). The profiles in Figures 6 and 8 are reported in Figure 9a,b, respectively. This figure shows that in leather A, the crack propagated at quite the same rate between t 1 and t 3 . For leather B, the crack also propagated at quite the same rate between t 1 and t 2 and then between t 3 and t 6 . Between t 2 and t 3 , the propagation speed was higher. Furthermore, crack growth was much slower (about five times) in leather B than in leather A. Regarding the physical nature of the thermoelastic coupling in both leathers, the temperature variation of a material point far from the crack tip, that is, far from the damaged zone, is plotted in Figure 10 (The locations of point A for leather A and point B for leather B are given in Figure 9). In this zone, the temperature variations can be estimated simply by image subtrac-   [20] or Samaca Martinez et al. (2015). [9] Indeed, even though the temperature subtraction is not performed exactly at the same material point, no temperature gradient occurs in this zone, which does not affect the temperature change value.
The red stars represent sample failure. The curves highlight that the temperature first decreased before increasing after a certain time (2 s for leather A; 8 s for leather B). As the material temperature was equal to room temperature before stretching and room temperature was constant during the test, the cooling observed at low strains was due to thermoelastic coupling (isentropic coupling). Then, the material temperature increased. This temperature increase can be due to intrinsic dissipation and/or to entropic coupling. It should be noted that in case of some materials, especially unfilled natural rubbers, [21,22] the temperature first decreases under strain and then increases, without involving any intrinsic dissipation. In order to go further in the case of leathers, such questions are fully addressed with a calorimetric approach in Corvec et al. (2019). [23]

CONCLUSION
In this paper, the tearing response of leathers was investigated at the local scale using the full-field techniques: DIC and IR thermography. The results obtained enable us to discuss and compare the tearing resistance of two leathers in terms of thermal activity in the crack zone of influence. In particular, distinct thermal and mechanical responses were observed between a grain pigskin leather and a grain calf leather: The latter exhibited the best tearing resistance, highlighting the interest of carefully investigating these responses in order to accurately model the tearing mechanism. It should be noted that the grain calf leather had undergone a retanning operation and contained a plasticiser, which may be correlated to both the specific thermal and mechanical behaviours of this material.