Nucleation sites and crack deflection. High-speed synchrotron x-ray 2D imaging with a super-short interval of 5 µs, exposure time of 6 µs and spatial resolution of 4 µm was employed to visualize the initiation and propagation of cracks in the stomatopod dactyl cuticle during the in situ impacting process (Fig. 2a-f). As shown in the representative images, the crack nucleation and propagation processes within the dactyl were captured while the sample was impacted by a spherical projectile with a diameter of 500 µm at speed of 160 m/s. The white strips emerged in the x-ray images featured the cracks induced by the impacting energy. As the cracks transmitted along the impacting direction, clear in-plane (indicated by the horizontal white strips) deflections toward impact surface were initiated.
The results show significant local fracture was mainly constrained within the impact region beneath the projectile contacting point with a length-scale about 160 µm. As the impacting process evolves, various damage mechanisms were observed, including sequential cleavage fracture (Fig. 2b), wing (or secondary) crack nucleation (Fig. 2c), crack deflection and coalescence (Fig. 2e)27. Even though the contrast of the acquired 2D images is not ideal, they provide valuable in situ information on the occurrence and evolution of different crack systems, of great importance for tracing and analyzing the 3D structure information from 3D ex-situ tomography experiments. To uncover the entire crack paths induced by the projectile impacting power, an ex-situ x-ray tomography imaging was performed on the same sample. As shown in Fig. 2l and 2m the full 3D crack systems induced by the impacting were successfully captured. The projection image with pixel size of 3.25 µm × 3.25 µm collected from the same perspective (Fig. 2g) as in the in situ 2D imaging offer clearer view of the crack paths within the impact and periodic region. The cracks which were propagating in the different lamellae shown low coalescence (Fig. 2h, 2k and 2m), along with branching (Fig. 2j) and deflecting behavior and were guided towards the dactyl surface in the end (Fig. 2h and 2m). Four major crack deflection sites were identified and indicated by arrows in different colors. Figure 2i-k show the sliced images of the 3D reconstruction along the dashed lines in the Fig. 2g. The multiple nucleation sites and growth of crack with low coalescence may lead to stress alleviation, which would be beneficial for the integrity of the dactyl club28. The four major deflection sites can be easily spotted in the CT slice (Fig. 2i, along the orange dash line) image, the first major deflection (green arrow) occurred before the main crack propagated to the impact-periodic interface. This phenomenon demonstrated the remaining impacting energy was not sufficient to penetrating the fronting lamella, therefore inclined to spread between lamellae looking for weak point to conquer. The crack path between the yellow and orange arrow shows that the crack travelled a long distance along the impact-periodic interface, indicating the interface between two major regions is more favorable for crack spreading compare to the interface between two lamellae within the same region (detail information in Supplementary Fig. 1 and Supplementary Note 1). The result further verified the larger modulus mismatch between impact and periodic regions (with Dundurs’ parameter of 0.33 at this interface) is important to hinder crack propagation through-thickness 15,18. The crack path between the orange and red arrow exhibits a zigzag shape within the periodic region suggesting the deflection occurred more frequently than in impact region, this may be due to the loosely packed structure and larger pore canals in this region. The frequent crack deflection significantly delayed catastrophic fracture and increased in crack surface area, leading to improved fracture resistance 29.
As shown in Fig. 2h and 2k, the crack was quickly deflected back into the impact region and finally reach the dactyl surface where the remaining energy was fully dissipated. The CT slice image (Fig. 2j, along the yellow dash line) shown that crack branching was induced in the first major deflection site. In the CT slice further away from the impacting point, the main crack turned into multiple microcrack networks, and the crack path shown in Fig. 2k and 2m appear to be discontinuous. The zigzag shaped crack propagation path can be seen more clearly from the high magnification SEM image (Fig. 3c and Supplementary Fig. 2a) collected on the cross section of the impacted dactyl. Figure 3c-i shows how the crack penetrated through a lamella when it has enough energy, with the high stress likely tearing the fiber sheets to propagate across different lamellar layers. Separation of adjacent lamellae was also found as shown in Fig. 3c and Fig. 3c-ii. Figure 3c-ii provides a rare view of how the crack is propagated within the weak interface of two neighboring lamellae, where the crack path was nearly at its end, with the remaining energy consumed by the fracture of fiber bundles (Supplementary Fig. 2b). As known previously15, in the periodic region, the fiber bundles are arranged following the shape of dactyl clubs, while the arrangement undergoes an abrupt change in the periphery of striated region, leading to abrupt crack propagation direction changes correspondingly. The complex 3D route of crack propagation results in a significant increase in fracture toughness, and demonstrated the overall energy absorption can be affected by the macroscale morphological characteristics 4.
Although equipped with a superconducting wiggler, the 3W1 beamline is based on a first-generation synchrotron source, the lack of coherence and low flux of x-ray prevent us from obtaining clearer and shorter time-interval images. For hydrated samples, the impact energy of incident projectile is not high enough to initiate visible cracks that can be observed by our in situ 2D x-ray imaging equipment. However, the hydration states of the dactyl sample are expected to play an important role in conducting its dynamic mechanical behaviors15,17. Ex-situ x-ray tomography and SEM measurements were employed to investigate the toughening and energy dissipation mechanisms of the dactyl club under the same loading conditions. According to the 3D crack networks, the volume of the main crack systems generated within the sample is about 20 times smaller than dry sample under similar impact energy. The results indicated that for the dry samples more energy was dissipated by brittle fracture such as crack initiation and extension, while for wet sample more energy tend to be consumed by other toughness mechanisms like plastic deformation and fiber bridging etc. As shown in Fig. 3d, several crack systems were nucleated at multiple sites away from the impact point, while surprisingly no main crack was found in the site directly beneath impact point, which means that there might have a zone of plastic deformation beneath the impact point.
Unlike the cracks deflected around the interface of impact-periodic region in dry sample, the cracks within hydrated samples deflected at the interface of impact surface/impact region (Denoted in Fig. 3d with black arrow). Slices of reconstructed volume for hydrated sample (Fig. 3e-f) show the main crack deflected in impact region and interface of impact surface-impact region. Distinct from the main cracks of the dry sample, the main cracks of the hydrated sample are discontinuous, as denoted in Fig. 3d with blue arrow. The discontinuity of crack system in the wet sample may is mainly caused by the herringbone architecture in the impact region which was identified as a crucial structure feature for enhancing energy dissipation in literature15. The complex fiber orientation distribution within the herringbone structure may induce abrupt disruption against crack propagation and therefore reduce the integrity of the main crack system. The impact energy were dissipated more efficiently in the impact surface for wet sample, hence the remaining energy was difficult to penetrate the herringbone structure compare to the dry sample. Another toughening mechanism observed in dactyl (which is hydration sensitive) is plasticity at the nanoscale. From SEM images of cross section of damaged hydrated dactyl club (Fig. 3g), there is a distinct plastic zone (Fig. 3g-ii) beneath the impact point, where highly misaligned mineral crystallites can be observed. The reorientation behaviors of the mineral crystallites were previously found in the impact surface of dactyl club when under high strain rate nanoindentation loadings 17, resulting from shear band localization of the stress waves. Figure 3g-iii shows an elastic zone featuring with textured arrangement of mineral crystallites. At the impact point, visualized at high-magnification in Fig. 3g-i, the existing of large pores might result from the loss of broken mineral particles. Under high stress, the crystallites fixed on the fibers were smashed into small irregular particles20 and exfoliated during polishing, leading to pores. We also observed fibers bridging the tips of opened microcracks (Fig. 3g-iv) within the elastic zone and fracture of fiber bundles (Fig. 3g-v). Under high-speed the impact, beneath the impact point the primary toughening mechanism is plasticity, which involves material phenomena like particle breakage, rotation, translation as well as microcracks, rather than crack deflection.
Deformation behavior of mineral nanoparticles. The impact surface (about 70 µm in thickness18 at the outermost region of the dactyl) which mainly consists of FAP nanoparticles16 (about 60 nm in diameter20) was considered to be mechanically crucial for providing the first shield against impact in recent studies 20. Various toughening mechanisms have been found in this region from previous studies, such as: reorientation of crystallites, particle breakage/rotation/translation 17,20. However, several new forms of deformation behavior have been revealed by investigating the damaged surface through SEM.
Figure 4a and Fig. 4b show the post-impact images of the damaged surface on the dry and hydrated dactyl club, where catastrophic fracture and spalling occurred on the impact surface when subjected to high-speed loading (160 m/s). As mentioned before, the impact induced a plastic zone surrounding the loading point on the hydrated sample. The main fracture damage was restrained in a limited region with a diameter of 700 µm (Red circle in Fig. 4c with radius labeled by r2). For convenience, we labeled the surface area with R1 (within r1) and R2 (between r1 and r2) region according to different radius from the central impacting point. Close to the loading point where largest impact forces were applied (R1; an area with a diameter of about 70 µm), misalignment of mineral nanoparticles can be observed within this region (Fig. 4f within region R1 comparison with Fig. 4g within region R2). Away from the central zone, where the impact forces were not sufficient to break the nanoparticles, particle rotation and translation seems to be the main means to consume the remaining energy within R2 (Fig. 4j within region R2 comparison with Fig. 4k within the outer region). Also, the compression of the fiber layers and the nanoparticles also contribute to the energy absorption (Fig. 4h shows the compressed fiber layers while Fig. 4i shows the uncompressed). As the damage area extended further, the spalling path was redirected towards the surface at the edge of the circle (r2). The CT images (Fig. 3e and 3f) show that the deflection of the spalling occurred at the depth about 100 µm when it reached the interface of the impact surface and impact region. The observed phenomenon further verified the hypothesis that the unique interlocking interfaces of the two regions can enhance stress redistribution and out-of-plane stiffness, acting as an important shield to prevent the catastrophic fracture going deeper. Under a higher speed (200m/s) impacting, the diameter of damage region has no significant change, but severe fracture took place at impact point, as shown in Fig. 4b.
Multiscale fiber bridging behavior. Fiber bridging can effectively reduce the local stresses and strain fields at the crack tip and acts as a prevailing toughening mechanism of the hierarchical biocomposite materials.3 The association of fiber bridging with crack evolution and energy absorption inside biocomposites has been systematically studied in many literatures 30–33. Here, many new forms of fiber bridging behavior were captured within the impacted dactyl club.
A wider range of bridge length (tens of nm to µm, shown in Fig. 4e and Fig. 5 respectively) was examined inside the samples that suffered fast impact compared to those (on the order of µm) subjected to quasi-static loading 17,18, indicating that the energy dissipation through crack bridging cannot be ignored under high-speed impact. As is known, the chitin fibrils (~ 3–4 nm in diameter) self-assemble with protein to form nanofibers (~ 20–100 nm in diameter). At the next hierarchical level, nanofibers are partially mineralized with calcium carbonate to form planar arrays of fibers (fiber sheets) stacked in a herringbone style structured manner to form lamellae at microscale within impact region34. The multiscale fiber pulling-out and bridging scheme are shown in Fig. 5d (Fiber bridging away from the impact point) and Fig. 5e (Fiber bridging near the impact point) which offers valuable information to deduce the related energy dissipation behavior: fiber sheets pulled from fiber lamellae (Fig. 5b, Fig. 5d-i and 5d-ii), fiber bundles pulled out from fiber sheets (Fig. 5d-iv and Fig. 5e-iii), fibers pulled out from fiber bundles (Fig. 5e-iii and 5e-iv) and nanofibers pull out from fibers (Fig. 5d-iv and Fig. 5e-v). The frictional energy for pulling out fibers at different scale could all contribute to the toughening. Pulling out of fiber bundles from the unique herringbone style structured interface will be much harder as different fiber layers will be involved and easily result in large scale twisted fiber bridging (Fig. 5b). Also, the out-of-plane fiber bundles running through the pore canals will exert extra resistance. Further, the pulling out process of the in-plane fibers will cause the contraction of the pore canals (Fig. 5d-ii) and forming isolated fiber sheets, along with fracture behavior of the out-of-plane fibers (Fig. 5d-iii). At the sites where fiber bridging is initiated, the nanofibers still remain connected while the surrounding matrix is fractured and full of visible nanocracks (Fig. 5d-iv), demonstrating a significant nanoscale toughening mechanism such as strain energy stored in the elastic fibers and the debonding energy for the fracture of mineral aggregation35. Remarkably, high magnification SEM images revealed how fiber bridging form in the lower hierarchical level and its association with the energy absorption. Figure 5e(i-v) shows a remarkable fractal fiber bridging behavior near the impact point across multiple structural level. The bridges connecting lamellae are large fiber bundles (tens to hundreds of micrometers, Fig. 5e-ii and Fig. 5d-i); at the end of the bridges, the fiber bundles split into single fibers or smaller fiber bundles (several micrometers, Fig. 5e-iii); further nanofibers inside those fibers are released from the encasing matrix to form the nanoscale links (Fig. 5e-iv).
3D structural characterization of impact surface and impact region. As mentioned before, the interface and junction region between the different regions of dactyl clubs can act as crack deflectors, especially when the dactyl is under hydration state. The complex geometry of the interfaces and enormous elastic modulus mismatch between the impact surface and the impact region might be key structural feature for enhance energy dissipation and crack deflection15. However, the detailed 3D structures of this region are still relatively unknown.
To gain further understanding of this region, nano-CT was used to examine the structure properties in fine details (with 3D spatial resolution of 64 nm). The reconstructive 3D volume shows that the outermost layer of the dactyl club was coated with mineral nanoparticles and no pore canals were observed (Fig. 6a-d). Beneath, densely packed mineralized chitin fibers (Fig. 6a) were distributed across the impact region forming an interpenetrating pore canal network (Fig. 6b). Surprisingly, besides the out-of-plane pore canals orient normal to the surface of the dactyl club which were found in most regions of the cuticle by SEM18,19, interconnecting in-plane pore canals framework were exposed in impact region which bundle the vertical canals together. In addition, clear evidence can be seen in the 2D slices (Fig. 6c-d) of the CT images that the pore canals were covered by mineral nanoparticle coating across all regions. As the 3D pore canal network is speculated to play an important mechanical role other than serving as channels for material transport during molting17,36, the nano-CT results furnish a new perspective how the pore canals are mechanically strengthened. The nanoparticle coatings could provide further protection of the pore canal networks from catastrophic collapse by dissipating impact energy through possible toughening mechanisms, such as particle rotation, translation and microcracks15,17,20. And also, the distribution of mineral particles in different layers surrounding the pore canal networks can also act as crack deflectors resulting from the elastic modulus mismatch of hard-soft interface37, thus increasing the fracture toughness of dactyl club. Due to the elastic modulus mismatch, cracks could be redirected and propagate perpendicular to the original direction, which has been summarized as a Cook-Gordan crack-stopping mechanism38. The strong correlation between the morphology of the mineral coating distribution and the pore canal networks as shown in Fig. 2e-i, ii and iii also suggests that the mineralization process might be highly dependent on mineral ion supply from the pore canal systems. Moreover, a thin section of segmented 3D volume of mineral particles and fiber bundles (Fig. 6e-iv) indicates irregular suture interface design of impact surface and impact region. These images show that at the interface, the fiber bundles within impact region are inserted into the mineral coating within impact surface. Geometrically interlocking interfaces enhance interfacial stiffness and strength, and interface waviness increases resistance to crack propagation39. SEM images also illustrates the insertion of fiber bundles and mineral coating (Supplementary Fig. 3e).
Dynamic finite element analysis (DFEA) was performed to gain a further insight on the mechanical role of newly uncovered structures in the impact surface and impact region. The geometrical shapes and dimensions used in the analytical model was based on the 3D architectural structure (Fig. 7a) acquired from nano-CT test. Figure 7b shows a geometric model of dactyl tissue which was converted to 3D finite element with high fidelity. To decouple the mechanical effects between the complex nanoscale geometrical feature and elastic modulus mismatch in the interface area, two specific models were defined with different material property assignments. For model I, different elastic modulus properties were assigned to the impact surface and impact region using values reported in literature15, therefore a high modulus mismatch (E1 = 70 GPa for mineral coating and E2 = 35 GPa for fiber scaffold) exists at the interface. In contrast, for model II the elastic modulus properties of mineral and fiber in the region were set to be the same on purpose, to focus on analyzing the geometrical effects on the mechanical performance.
The impact target was set to be a rigid ball with radius of 1 × 10− 6 m and mass of 1 × 10− 16 kg for both models. When subjected to impacts at a velocity of 200 m/s, the average Von-mises stress distribution cloud diagrams of model I and model II are shown in Fig. 7c and 7d, respectively. At the initial stage (first 0.5 ns), the Von-mises stress in both models were still distributed in the surrounding areas of the contact point. A peak stress (~ 1150 MPa) was obtained near the impact contact zone at 0.25 ns and the value deceased significantly (~ 300 MPa) in the next 0.25 ns. Notably, the initial semi-spherical shape stress pattern indicating a continuous propagation was disrupted in model I right after the compressive wave reached the interface of the impact surface and impact region at 0.75 ns, while no clear disruption was shown in model II. The results demonstrate the modulus mismatch in the interface acts as the first shield to hinder the impact stress propagation. As the compressive wave travelled further into the interface region, and the complex interlocking structure of the mineral coating impact surface and bulk of the impact region came into effect, the spherical shape of the stress pattern starts to exhibit obvious deformation in both models, which suggests the interlocking geometry in the interface region plays a dominant role in shielding impact waves. As the compressive wave travels deeper into the dactyl club, clear stress concentration behaviors are found around pore canal regions in model I, while no visible stress concentration signs are shown in model II (1.25–1.5 ns). The results indicated the modulus mismatch at the pore canal region is the main reason for stress concentration. The hypothesis can be further proved by the 3D mineralized pore canal networks in Fig. 6e. The modulus mismatch hinders the stress and crack propagating from the mineral coating layers into the fiber scaffolds of the pore canals, leading to the deformation and reorientation of the mineral nanoparticles and nucleation of microcracks in the mineral layer. Considering the wide distribution and large area of pore canals (specific surface area about 7.73 × 105 m2/m3) within the impact region of dactyl club, the mineral particles in the pore canal networks would dissipate considerable energy during impact event. The mineral particles surrounding pore canal networks helps spread the damage to macroscopic tissue regions from a microscale initiation point that improving the toughness of dactyl club. This kind of mineral strengthening strategy might inspire the designs of biomedical implant materials which requires high mechanical strength and microscale porosity for nutrient flow and cell proliferation 40. The FEMA results demonstrated that the elastic modulus mismatches at different length scales and the irregular suture interface of junction region can work together to prevent the stress from propagating into the bulk of impact region.