Design and additive manufacturing of bionic hybrid structure inspired by cuttlebone to achieve superior mechanical properties and shape memory function

Lightweight porous materials with high load-bearing, damage tolerance and energy absorption (EA) as well as intelligence of shape recovery after material deformation are beneficial and critical for many applications, e.g. aerospace, automobiles, electronics, etc. Cuttlebone produced in the cuttlefish has evolved vertical walls with the optimal corrugation gradient, enabling stress homogenization, significant load bearing, and damage tolerance to protect the organism from high external pressures in the deep sea. This work illustrated that the complex hybrid wave shape in cuttlebone walls, becoming more tortuous from bottom to top, creates a lightweight, load-bearing structure with progressive failure. By mimicking the cuttlebone, a novel bionic hybrid structure (BHS) was proposed, and as a comparison, a regular corrugated structure and a straight wall structure were designed. Three types of designed structures have been successfully manufactured by laser powder bed fusion (LPBF) with NiTi powder. The LPBF-processed BHS exhibited a total porosity of 0.042% and a good dimensional accuracy with a peak deviation of 17.4 μm. Microstructural analysis indicated that the LPBF-processed BHS had a strong (001) crystallographic orientation and an average size of 9.85 μm. Mechanical analysis revealed the LPBF-processed BHS could withstand over 25 000 times its weight without significant deformation and had the highest specific EA value (5.32 J·g−1) due to the absence of stress concentration and progressive wall failure during compression. Cyclic compression testing showed that LPBF-processed BHS possessed superior viscoelastic and elasticity energy dissipation capacity. Importantly, the uniform reversible phase transition from martensite to austenite in the walls enables the structure to largely recover its pre-deformation shape when heated (over 99% recovery rate). These design strategies can serve as valuable references for the development of intelligent components that possess high mechanical efficiency and shape memory capabilities.

Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence.Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.

Introduction
Throughout billions of years of evolution, organisms have developed natural structures that exhibit unparalleled superior performance, enabling them to survive in the fierce competition of nature [1][2][3].Indeed, organisms exhibit effective structural design strategies that enable the mechanical performances of biological materials exceed the expected limitations of their components [4].By emulating these natural biological structures, engineers can design novel structures with exceptional mechanical properties and multifunctionality, thereby advancing the development of engineering components [5].Achieving satisfactory mechanical properties or functionality with bionic porous structures constructed with minimal material is one of the most prominent examples [6].For example, several examples have been reported of bionic structures exhibiting unusual phenomena of compression-induced twisting performance [7], non-global structural buckling [1,8] and twisting stress conduction path [9,10].A primary goal in all these bionic designs is to develop porous structures with high load-bearing and energy-absorbing capacity, making them valuable in numerous applications such as aerospace, automobiles, electronics, and others.Therefore, the challenge of constructing porous structures that meet target requirements resides in the requirement for ingenious structural design to ensure sufficiently strong to withstand the mechanical loading without damage and to ensure that the failure of the structure is non-catastrophic in case the strength is exceeded.
A large number of studies have proved that the load-bearing and energy absorption (EA) capability of components can be significantly improved by imitating natural biological structures [11][12][13].In recent years, the rapid advancements in additive manufacturing technology have facilitated the production of complex bionic structures.A fascinating example is the cuttlebone, a highly porous cellular structure primarily composed of a brittle mineral [14,15].Surprisingly, partially gas-filled cuttlebone exhibits remarkable resistance to external pressure in deep-sea environments [16].This is attributed to the unique microstructure of 'wall-septa', where lamellar septa are separated and supported by vertical walls with an optimal waviness gradient [17][18][19].Notably, studies have shown that the destruction of cuttlebone is noncatastrophic, as progressive fracture and subsequent densification of one chamber do not affect the integrity of adjacent chambers [14,17].Inspired by these extraordinary characteristics, various mechanical structures have been developed over the past decade.For instance, cuttlebone-like lattices designed through topology optimization achieved high specific compressive modulus and energy absorption efficiency (EAE) [20][21][22].Moreover, advancements in analytical technologies, such as microcomputed tomography (Micro-CT), have provided researchers with powerful tools to unravel the three-dimensional structures of cuttlebone and design bionic structures based on its characteristics.Multilayer biomimetic structures, based on the three-dimensional structural features of cuttlebone, exhibit layer-by-layer failure under uniaxial compression, resulting in stress-strain curves with wide elastic regions and prolonged plateaus, thereby achieving exceptional strength and EA properties [17].Although the multilayer structural characteristics of bionic cuttlebone structure have been clearly recognized in the mechanical field, which can control the maximum stress on the wall in a clear position at the chamber level to achieve conspicuous mechanical performance [14], the structural design strategies considering the characteristics of the cuttlebone structure has not yet been established.
Metallic components play a crucial role in modern industries [23].Further advancements in bionic metal structures offer opportunities to enhance the performance and functionality of metal components [9].However, most bionic cuttlebone structures are constructed using nonmetallic materials.Currently, Nickel-Titanium (NiTi) shape memory alloy has garnered attention as a raw material for designing intelligent components due to its unique properties, including superelasticity (SE) and shape memory effect (SME) [24].Nevertheless, existing NiTi porous structures, such as triply periodic minimal surface structures [25] and cellular lattice structures [26,27], are prone to severe stress concentration accompanied by irreversible plastic deformation under loading, or catastrophic failure if their strength is exceeded.These limitations hinder the full utilization of the NiTi alloy's potential.Therefore, understanding the optimal waviness gradient of cuttlebone's vertical walls, which enables high strength and progressive failure at the chamber level, can provide valuable insights for developing lightweight intelligent protection systems that fully exploit the material characteristics of the NiTi alloy.
Inspired by the construction strategy of cuttlebone, a novel bionic hybrid structure (BHS) has been proposed.The vertical support wall of BHS becomes more tortuous from bottom to top.Structures were fabricated by laser powder bed fusion (LPBF) using NiTi powder as raw material.First, the forming quality, including porosity and dimensional accuracy of LPBF-processed structure, was quantitatively evaluated, and the phase transformation behavior and microstructure characteristics were analyzed.Subsequently, the mechanical behavior of bionic structure was investigated by performing compression experiments.Finally, the shape recovery capabilities and EA behavior of structures were evaluated by strain controlled compression experiments and finite element analysis.This systematic study could provide guidance for designing and manufacturing lightweight intelligent protection systems with high load-bearing and high EA while having shape memory functions.

Microstructure of cuttlebone and bionic structure design
A specimen of Sepiella japonica with a cuttlebone was acquired from a fish market.The cuttlebone was dissected from Sepiella japonica and sectioned into a 1 cm 3 cube.The prepared samples were then cleaned with alcohol using ultrasonic treatment to ensure the removal of any residue.The internal shell of the cuttlefish (figure 1(a)), located on the dorsal side, serves as a rigid buoyancy tank despite being composed of heavy calcium carbonate [28].The density of the cuttlefish can be adjusted by changing the amount of liquid filled in the porous structure (porosity >90%) of its bone (figure 1(b)), allowing it to maintain a stable position in the deep sea with minimal effort [16].Furthermore, the cuttlebone acts as a rigid skeleton that enables the cuttlefish to withstand significant external pressure, which increases with depth.The multilayer structure inside the cuttlebone, reconstructed from micro-CT images (figure 1(c)), reveals the porous chambered structure of the cuttlebone, with walls supporting adjacent septa (movies S1 and S2, supporting information).To investigate the failure mode of the cuttlebone for EA, a compression test on a cube-shaped cuttlebone with a side length of 1 cm was conducted (movie S3, supporting information).The compressive stress-strain curve of the cuttlebone exhibits three stages (figure 1(d)): elastic deformation (E), plateau (P) and densification (D).The stress-strain curve in the plateau stage fluctuates periodically (figure 1(e)), and the period of fluctuation corresponds to the number of chambers.Within the fracture process of an individual layer, the stress-strain curve of the cuttlebone displays minor troughs (S1 in figure 1(f)), followed by a continuous decrease in stress due to wall failure (S2 in figure 1(f)).In the densification stage (S3 in figure 1(f)), the stress increases rapidly as the fractured walls gradually compact.
Distinctive labyrinthine patterns were observed in specific wall-septa structures of the reconstructed cuttlebone, near both the top (figure 2(a)) and the bottom (figure 2(b)).It was evident that the walls exhibited an increasing degree of waviness from the bottom to the top.The cross-sectional profile of a single wall at the bottom resembled a sine wave.The reconstructed individual wall, depicted on the left side of figure 2(c), demonstrated this sinusoidal contour feature.Furthermore, a plot illustrating the normalized contour length versus normalized height for a single wall, displayed on the right side of figure 2(c), clearly indicated the increasing tortuosity of the wall from the bottom to the top.On the other hand, the crosssectional profile of a single wall at the top represented a hybrid wave, which is composed of a superposition of the bottom sine wave and a sine wave with a smaller period (figures 2(d)-(f)).
Taking inspiration from the wall-septa characteristics of cuttlebone, a BHS characterized by two horizontal septa supported by walls with hybrid waves was proposed (figure 2(g)).
The top profile of the wall can be controlled using a hybrid sine curve, expressed by equation (1): (1) Similarly, the bottom profile of the wall can be controlled by a sine curve, expressed by equation ( 2): where A, B, C, and D have fixed values of 0.5, 1, 0.5, and 3, respectively.The wall of the BHS was formed by extruding the top profile to the bottom profile in the height direction.By combining the top and bottom septa, the unit cell of the BHS was obtained.The single-layer BHS structure was obtained by arranging the unit cells in an array (figure 2(g)).
The thickness of walls and septa was designed to be 0.2 mm.For comparison, this work also designed a regular corrugated structure (RCS) with both the top and bottom profiles of regular sinusoids (figure 2(h)) and a straight wall structure (SWS) with straight top and bottom profiles (figure 2(i)) respectively.Among them, both the top and bottom profiles of RCS were controlled by equation (2).The other structural parameters of RCS and SWS were consistent with those of BHS.

LPBF process and microstructural characterization
All designed structures were fabricated using a self-developed LPBF system.The pre-alloyed Ni 53.34 Ti 46.65 (at%) alloy powder which can deliver precise chemical composition was used as raw material (figures 3(a)-(c)).The powder particles had a nearly spherical shape with an average of 26.96 µm, which ensured the required fluidity during the powder laying process in LPBF.The processing parameters were set as follows: laser power (P) of 125 W, scanning speed (v) of 1 400 mm•s −1 , powder layer thickness (h) of 30 µm, and hatch spacing (t) of 50 µm.After fabricated, the samples were cut off from the substrate using electric discharge wire-cutting technology.They were washed in an ultrasonic cleaner with alcohol for 10 min to ensure the removal of residual powder and cooling medium.
The cuttlebone sample and LPBF-processed sample were characterized by the Micro-CT with a Micro-CT scanner (d2, Diondo, Germany) at 90 kV voltage, with the voxel size of 7 µm.The 3D models were rendered using VGStudio MAX software.The phase constitution was determined through xray diffraction (XRD) analysis using a Thermo Scientific TM ARL TM X'TRA x-ray diffractometer with Cu-Kα radiation (λ = 0.154 18 nm).The scanning angle 2θ ranged from 20 • to 90 • with a scanning speed of 2 • •min −1 .The thermal behavior was analyzed using Differential Scanning Calorimetry (DSC) on a TA Instruments DSC Q20 instrument, with a cooling/heating rate of 10 • C•min −1 .Polished cross-sections of the powder and fracture morphology of the structure were characterized using a field emission scanning electron microscope (SEM) (TESCAN LYRA3, Czech Republic) equipped with an energy-dispersive x-ray spectroscopy (EDS) system (Bruker XFlash Detector 6160, USA).

Mechanical testing and numerical simulation
Compression tests were performed using a CMT5205 testing machine (MTS Industrial System, China) with a strain rate of 1 mm•min −1 .To evaluate the functional performance, cyclic compression tests were conducted with strain increments of 0.5% and a fixed strain rate of 10 −3 mm•s −1 .After the last cycle of loading and unloading, the shape memory process was carried out by immersing the samples in a water bath at 50 • C for 10 min.Subsequently, took out the sample from the water bath and cooled it to room temperature and its height was measured as L 1 .Therefore, the recovery rate η of the sample can be calculated as [29][30][31]: where L 0 is the original height of the sample.The EA indicators were defined and calculated according to the ISO 13314:2011 standard [32], which recommends a compressive strain upper limit of 50% for calculating EA and EAE.The absorbed energy (EA) could be defined by the area under the compressive stress-strain curve, and determined by the following relation [32]: where σ is the compressive stress (MPa), ε 0 is the upper limit of the compressive strain (50%).The EAE was used to evaluate the energy absorbing performance and could be expressed as the following [32]: where σ 0 is the maximum compressive stress within the strain range.The specific energy absorption (SEA) considered the density of the structure was adopted to further evaluate the EA, which can be expressed as the following [33]: where ρ nomi is the nominal relative density, which was determined by the ratio between the volume of the designed model and the spatial region of the structure; ρ s is the density of the base material.
The mesoscopic model with a three-dimensional size of 1 000 µm × 600 µm × 200 µm was employed based on the finite volume method to evaluate the thermal behavior of the molten pool during LPBF processing (figure 3(d)).The establishment details and boundary conditions for this model can be obtained elsewhere [34,35].The model took into account thermophysical behaviors such as the transition of solid/liquid phases, liquid surface tension, thermocapillary force, buoyancy force, and the recoil pressure associated with the LPBF process.The thermal physical parameters of solid NiTi alloy utilized in the model can be referenced from our previous studies [36,37].For simulating the mechanical behaviors of LPBF-processed structures under compression (figure 3 2019 was utilized.The structure was meshed using tetrahedral elements with a mean element size of 0.2 mm and a total of 186 717 elements, ensuring convergence and accuracy in the computed results.The parameters employed in the constitutive model can be obtained from our previous work [38].

Forming quality
The as-processed sample exhibited an extremely low porosity, with the presence of very small-sized pores, as depicted in the defect distribution (figure 4(a)).Statistical analysis of the defect volume indicated that the majority of defects within the structure consisted of a large number of pores with a volume below 0.5 × 10 −4 mm 3 , resulting in a total porosity of 0.042% (figure 4(b)).The insets provided further insights into the three-dimensional reconstructed morphologies of typical defects, including spherical and irregular pores.Spherical pores were attributed to undestroyed gases that remained in the solidified material due to the unstable collapse of vapor cavities [39].On the other hand, irregular pores were likely related to insufficient fusion [40].Furthermore, the defects observed in the wall and septum (figures 4(c) and (d)) primarily comprised small-sized pores.Notably, the porosity in the wall (0.045%) exceeded that in the septum (0.033%), which was clearly visible in the magnified inserts.This phenomenon could be attributed to the presence of more overhanging surfaces in the wall, which trapped some gas when the molten liquid on the suspended surface flowed into the underlying powder layer [9].
The three-dimensional morphology of the LPBF-processed BHS (figure 5(a)) exhibited the absence of obvious macrodefects, and the internal connections within the structure appeared continuous.The cross-sectional patterns of the walls near the top and bottom (figures 5(b) and (c)) displayed regular waveforms without noticeable distortion.The surface deviation analysis of the entire structure revealed a Gaussian distribution with a peak deviation of 17.4 µm (figure 5(d)).Moreover, the surface deviations of the wall and septum (figures 5(e) and (f)) followed a Gaussian distribution, with peak deviations close to zero.These results indicate that the LPBF-processed BHS exhibited good dimensional accuracy.In overhanging surfaces, a higher proportion of areas with elevated surface deviations was observed (figure 5(g)).This can be clearly seen from a comparison of the surface deviation diagrams along the facing and building directions.The crosssectional morphologies of the walls also revealed substantial dimensional deviations at the overhanging surfaces (highlighted in yellow in figure 5(h)).The wall surface exhibited the presence of adhered powders and a fish-scale-like staircase effect (indicated by a yellow dashed line in figure 5(i)), which may contribute to the dimensional deviations observed in the LPBF-processed BHS.

Phase transformation and microstructure
DSC was utilized to investigate the temperature-dependent phase transformation behaviors, which are closely associated with the shape memory function of the structure.The DSC analysis revealed single-stage transformations characterized by a single peak, indicating a direct solid-solid transformation between B19 ′ martensite and B2 austenite during the heating/cooling process (figure 6(a)) [41].The presence of a single peak also suggests good uniformity in the chemical composition and microstructure of the LPBF-processed BHS [42].The martensitic transformation temperatures, determined using the tangent method [43], indicated that the structure primarily consists of B2 austenite at room temperature [44][45][46].This can be attributed to the consumption of Ni in the matrix caused by the evaporation of Ni and the formation of Ni-rich phases [44,47].XRD patterns revealed the existence of B2 (figure 6(b)) and a small amount of B19 ′ with a large probability of the absence of other phases.The detected B19 ′ may be the accumulated residual martensite generated by reversible transformation caused by cyclic thermal stress during the LPBF process.It tends to form at B2 grain boundaries and aggregate at the boundaries of the melt pool [48].The weak compositional segregation leads to the precipitation of fine Ni-rich precipitates in the side regions of the tracks [49], and the stress-induced effects between tracks promote their growth and precipitation [50].These precipitates induce incoherency stresses and modify the local Ni concentration in the surrounding matrix, thereby facilitating the formation of fine martensitic plates [51].The partial cross-section of the LPBF-processed BHS exhibited no noticeable macroor micro-defects (figure 6(c)).Some adhered powders were observed on the surface, particularly on the overhang surface.BSE (figure 6(d)) and EDS images (figures 6(e) and (f)) further confirmed a uniform distribution of components within the sample, without macroscopic segregation.
Numerical simulation was employed to gain insights into the thermophysical behavior of the LPBF-processed tracks and the overlap region of two tracks during LPBF processing.The as-manufactured tracks exhibited a regular shape and were devoid of noticeable defects (figure 7(a)), which can be attributed to the uniform distribution of the powder bed with a high packing density [52].During the LPBF process, the surface temperature of the area irradiated by the laser beam can easily reach boiling point values, leading to the formation of vapor recoil pressure that exerts additional forces on the liquid surface [53].This phenomenon creates depressions in the molten pool.Furthermore, the Marangoni effect, induced by the strong temperature gradient at the front of the molten pool, promotes melt flow inside the pool and increases the melt depth (figure 7  The phase distribution observed in the electron backscatter diffraction (EBSD) results confirmed the findings from the XRD analysis, indicating that the sample predominantly consists of B2 phase with only a small amount of B19 ′ phase (figure 8(a)).The texture change in the longitudinal section of the wall was evaluated using EBSD analysis.The texture index I tex was used to evaluate the local texture intensity, which could be calculated by the following equation [55]: where f (g) is the orientation distribution function in the Euler space, and g is the Euler space coordinates.It was generally found that the texture index of anisotropic materials was larger than 1 [6].The maximum intensity of the texture index

Mechanical properties and shape memory function
The deformation processes and the progressive failureJ mechanisms of LPBF-processed BHS under uniaxial compressive were revealed (movie S4, supporting information).Local buckling of the walls was observed near the bottom part, as indicated by a yellow arrow at a strain of 10%.Subsequent to the failure at the local buckling position, the stress significantly decreased (figure 9(a)).Interestingly, the walls of the BHS only failed below the buckling position (area between the two yellow dashed lines), while the residual walls remained intact.Because the partial wall above the destruction location was more tortuous, it had more materials to possess greater stiffness to resisted the fracture.During the subsequent compression process, local buckling and failure occurred successively at the bottom of the remaining walls until densification was achieved.This progressive failure mechanism effectively improved material utilization efficiency and EA capacity, particularly for materials with limited ductility.In contrast, the RCS and the SWS exhibited The work-hardening rates as a function of true strain for the three LPBF-processed structures before the failure of the walls were also presented (figure 9(g)).As the structures were gradually compacted during compression, the work-hardening rates initially increased with increasing true strain.This indicated an enhancement in the deformation uniformity of the structures, leading to an increase in material strength.However, following the onset of buckling deformation and the subsequent reduction in strength (figure 9(d)), the work-hardening rates of the structures decreased significantly (figure 9(g)).To evaluate the EA capacity of the structures, EA curves were plotted against compressive strain (figure 9(h)).The EA curve of the BHS exhibited a characteristic 'staircase' shape, with a plateau stage (Stage II) sandwiched between two rapid-increase stages (Stages I and III).In contrast, the EA curves of the RCS and the The energy absorption versus the strain.Comparison of energy absorption indexes between the three structures at the upper limit strain of 50%: (i) energy absorption efficiency and (j) specific energy absorption.
SWS showed an initial rapid-increase stage (Stage II) followed by a quasi-plateau stage (Stage III).The EAE of the BHS was significantly improved compared to the RCS and the SWS (figure 9(i)).Furthermore, considering the differences in mass among the three LPBF-processed structures, the EA capacities were quantified using SEA.The BHS exhibited the highest SEA value (5.32 J•g −1 ), representing a 333.11% increase compared to the SWS (figure 9(j)).These results indicate that the BHS possesses excellent load-bearing performance and superior EA capacity.
The fracture surface of the wall exhibited a relatively smooth and flush appearance (figure 10 .These observations confirm that the fracture occurred through a quasi-cleavage mechanism.By examining the morphologies of the compressed structural fragments (figures 10(e)-(g)), it was observed that the walls of the BHS disintegrated completely during the compression test, leaving behind scattered fragments.In contrast, the walls of the RCS and the SWS fractured into large chunks, forming intact fragments.This indicates that the BHS exhibited a higher material utilization rate compared to the other two structures.
The mechanical properties of LPBF-processed samples were evaluated through strain-controlled cyclic compression, with a strain increment of 0.5% until a final applied strain of 7.5%.During each unloading-reloading cycle (figures 11(a)-(c)), the loading-unloading curve exhibited a hysteretic loop, which was attributed to the reversible stressinduced martensitic transformation [57].This indicated that all LPBF-processed structures demonstrated excellent superelastic behavior.The peak stress of the BHS gradually increases with the increase in the number of loading cycles and slightly decreases after reaching the maximum stress of 18.58 MPa (figure 12(a)).This behavior can be attributed to the local buckling deformation of the BHS, which promoted the formation of a new deformation band at the pre-existing strained band front, thereby increasing the resistance to subsequent deformation.In contrast, for the RCS and the SWS, the peak stress initially increased and then decreased, with the values consistently lower than those of the BHS (figures 12(b) and (c)) [58].Notably, the BHS also exhibited advantages in terms of elastic modulus (figures 12(d)-(f)).
The variations in mechanical energies (ME) were calculated for different loading-unloading cycles.The ME included the mechanical energy consumed by irreversible deformation E consumed , the viscoelastically dissipated energy E dissipated and the elastically restored energy E elastic .ME total represented the sum of these three energies (as shown in the inset of figure 11(c)).The ME total of all three samples increased as the loading cycle progressed (figure 11(d)), with BHS demonstrating superior EA capabilities, as reflected in the larger area of each hysteresis loop compared to the RCS and the SWS.It is worth noting that the ME total of the RCS exhibited an initial increasing trend with loading but became nearly constant after approximately six cycles.This behavior can be attributed to the rapid decrease in strength following buckling deformation.During each loading-unloading cycle, E elastic played a dominant role in the total energy dissipation ME total , followed by the E dissipated generated by the reversible transformation and the E consumed contributed by irreversible deformation (figures 11(e)-(g)).The E consumed was directly related to the irreversible strain, which increased gradually for all samples with an increase in the number of cycles (figures 12(g)-(i)).The continuous accumulation of residual strain can be attributed to the progressive increase in the amount of residual martensite generated upon unloading [59], which is due to the obstruction of the reverse transformation of martensite to austenite caused by high-density dislocations [59,60].The reorientation or detwinning of the self-adapted martensite in the samples resulted in the maximum ∆ε occurring in the first cycle.However, reoriented or detwinned martensite cannot undergo reverse transformation without external energy excitation, leading to a sharp drop in ∆ε during the second cycle.Subsequent cycles exhibited an increase in ∆ε due to the accumulation of residual martensite.
The shape memory function of the deformed samples was further assessed through water-bath heating.The morphologies of the three structures after strain-controlled cyclic compression revealed buckling deformation (figure 11(h)), consistent with the findings in figure 9(a).Figures 11(i) and (j) demonstrate the process of water-bath heating and structural morphologies after shape recovery, respectively.Remarkably, all structures exhibited a significant SME after water-bath heating, with recovery rates of 99.10%, 98.94%, and 99.74% for the BHS, the RCS, and the SWS, respectively (figure 11(k)).These results highlight the ability of the BHS to achieve efficient EA while maintaining a high recovery efficiency.
Finite element analyses were conducted to investigate the stress distribution (SD) and phase transformation behavior of the three structures under a 7.5% strain.It was observed that the positions of stress concentration (indicated by yellow arrows in the SD plot of figure 13) in the three structures corresponded well with the locations of buckling deformation observed in the experiments (marked by yellow arrows at a strain of 10% in figures 9(a)-(c)).These stress-concentrated positions were more prone to local failure during subsequent loading.This finding also explained why the walls of the BHS failed near the bottom (figure 9  to present the nodal stresses of the finite element mesh extracted from a single vertical wall in each of the three structures (Histograms in figure 13).In contrast to the wider SD in the RCS and the SWS, the narrower nodal SD of the BHS demonstrated its more uniform SD.This indicates that the vertical walls of the BHS, characterized by hybrid waveforms, were beneficial in dispersing stress and enhancing the stability of the structure.Notably, the average values of the nodal stresses (σ Avg ) in the BHS, the RCS and the SWS were 773.87 MPa, 364.26 MPa and 376.69 MPa, respectively, indicating that BHS had good load-bearing capacity.Corresponding to the SD, the distribution of martensitic volume fraction (MVF) was uniform in the vertical walls of BHS, while the martensitic transformation mainly occurred in the stress-concentrated positions in the RCS and the SWS (MVF plot in figure 13).This demonstrates that the BHS could dissipate more imposed mechanical energy through martensitic transformation compared to the RCS and the SWS, which explains that the BHS has higher viscoelastically dissipated energy (E dissipated ) (figures 11(e)-(g)).Taken together, these results summarized the BHS enhanced ability of load-bearing and EA, achieved through the homogenization of SD in the vertical walls with hybrid waveform characteristics.

Conclusions
In this work, the microstructural characteristics of cuttlebone were revealed, and a novel BHS was proposed and manufactured by LPBF using NiTi powder.The forming quality, including porosity and dimensional accuracy of LPBFprocessed BHS was quantitatively evaluated, and the mechanical behavior and the shape recovery capability of LPBFprocessed structures were investigated.The main conclusions are as follows: 1.The cuttlebone was found to have a porous chambered structure with walls supporting adjacent septa.The walls exhibited increasing waviness from the bottom to the top.Labyrinthine patterns were observed in specific wall-septa structures of the reconstructed cuttlebone.The stress-strain curve of cuttlebone exhibits three stages: elastic deformation, plateau, and densification.The plateau stage fluctuates periodically, and the period of fluctuation corresponds to the number of chambers.2. A dense and designed shape-consistency LPBF-processed BHS was achieved, the total porosity was 0.042%, and the porosity in the wall (0.045%) was greater than the septum (0.033%).The LPBF-processed BHS exhibited good dimensional accuracy, with surface deviations following a Gaussian distribution and a peak deviation of 17.4 µm.

Figure 1 .
Figure 1.Structural characteristics of cuttlebone.(a) Morphology of the cuttlefish.(b) Cuttlebone dissected from cuttlefish.(c) A multilayer structure inside the cuttlebone reconstructed from micro-CT image.(d) Stress-strain curve of the cube cuttlebone specimen.(e) Magnified view of the stress-strain curve in (d).(f) The progressive failure of the walls of the cuttlebone.

Figure 2 .
Figure 2. The wall features of cuttlebone and the design process of bionic structure.The labyrinthine patterns of cross-sections of walls near (a) the top and (b) the bottom.(c) A representative reconstructed individual wall and the normalized length plotted against the normalized height of individual wall cross-section.(d) The superimposed labyrinthine pattern of the cross-sections of walls near the top and the bottom.(e) The shape of the cross section of typical corrugated walls separated from (d).(f) Pattern feature extraction and analysis in (e).The design processes and geometric parameters of (g) the bionic hybrid structure (BHS), (h) regular corrugated structure (RCS) and (i) the straight wall structure (SWS).

Figure 3 .
Figure 3. NiTi alloy powder and finite element simulation models.(a) Morphology and (b) particle-size distribution of pre-alloyed NiTi alloy powder.(c) Morphology of the cross-section of NiTi alloy powder.(d) Overview of the established mesoscopic model and the scanning strategy during simulation.(e) Schematic of the finite element (FE) model for simulating the compression behavior of LPBF-processed structure.

Figure 4 .
Figure 4.The micro-CT reconstructed images from the LPBF-processed bionic hybrid structure.(a) The three-dimensional distribution of defects inside the as-processed sample and corresponding (b) statistical defect volume, the inset in (b) shows the three-dimensional reconstructed morphologies of pores.The three-dimensional distribution of defects and corresponding statistical defect volumes inside (c) the wall and (d) the septum.
(b)) [54].To understand the temperature history of the LPBF-processed tracks and the overlap region of two tracks, three temperature detection points (A, B, and C) were established in the cross-sectional view of the temperature field (figure 7(c)).Point A was located in the middle of the first track, point C was in the middle of the second track, and point B was positioned between points A and C at the overlap of the two tracks.The temperature at both detection points A and C reached the boiling point of Ni (3005 K) once, while the temperature at detection point B exceeded the boiling point of Ni twice (figures 7(d)-(f)).This suggests that the overlap region of the two tracks depletes Ni in the matrix and exacerbates the chemical composition differences between the NiTi powder and the as-manufactured alloys.

Figure 5 .
Figure 5.The micro-CT reconstructed images from the LPBF-processed bionic hybrid structure.(a) The three-dimensional morphology, cross-sectional morphologies of walls near (b) the top and (c) the bottom.The statistical surface deviation distribution and corresponding surface deviation map, (d) the whole structure, (e) the wall and (f) the septum.(g) The surface deviation maps facing and along the building direction.(h) Cross-sectional morphologies of walls.(i) Side surface morphology of the wall shows some adhered powders and the fish-scale-like staircase effect.

recorded as 7 .
49 (figure8(e)), the grains exhibited a strong (001) crystallographic orientation (figure8(b)).This strong texture is attributed to the preferential growth of grains in the (001) direction along the building direction[56].Notably, the grain size ranged from 3.39 µm to 73.60 µm, with an average size of 9.85 µm (figures 8(f) and (g)).The presence of fine grains can be attributed to the extremely rapid cooling rate during the LPBF process.Furthermore, the Kernel Average Misorientation revealed characteristic distortions that represent local deformations and high dislocation densities primarily concentrated in the enrichment area of low-angle grain boundaries (figure8(c)).A significant number of low-angle grain boundaries were enriched on the upper side of the wall (figure 8(d)), exhibiting concentrated misorientation angles ranging from 1.5 • to 4.5 • (figure 8(h)), with a particularly high frequency around 2 • .This indicates the presence of numerous deformed grains in the material.

Figure 6 .
Figure 6.Phase transformation behavior and chemical elements of LPBF-processed bionic hybrid structure.(a) DSC curve and (b) XRD pattern.(c) OM image of a partial cross-section.(d) The BSE image the box selection area in (c).Distribution of (e) Ni and (f) Ti elements obtained by EDS measurements in the region corresponding to the BSE image.

Figure 7 .
Figure 7. Simulation results of LPBF-processed tracks.(a) The 3D temperature fields.(b) The longitudinal sectional view of velocity fields.(c) The cross-sectional view of temperature fields.Temperature-time curves of (d) detection point A, (e) detection point B and (f) detection point C.

Figure 8 .
Figure 8. EBSD results of the longitudinal section of LPBF-processed bionic hybrid structure.(a)-(d) Showing the phase distribution, inverse pole figures (IPF), kernel average misorientation (KAM) and grain boundaries maps in the longitudinal section of the wall of LPBF-processed bionic hybrid structure.(e) Pole figures (PF) showing a slight concentration of (001) crystallographic orientation along building direction.(f)-(h) Grain size distribution, misorientation angle distribution and local misorientation angle distribution in the longitudinal section.

Figure 9 .
Figure 9. Mechanical properties and deformation mechanism of three LPBF-processed structures.(a)-(c) Deformation processes and failure mechanisms.(d) Stress-strain curves.(e) and (f) The load-bearing test of LPBF-processed bionic hybrid structures, the cartoon showing the bearing capacity of the structure in the right of (e), (f) the details in the left of (e), the inserts in (f) show the morphologies of the tested samples without no obvious deformation.(g) Variations in the work-hardening rate with the true strain for the three structures.(h)The energy absorption versus the strain.Comparison of energy absorption indexes between the three structures at the upper limit strain of 50%: (i) energy absorption efficiency and (j) specific energy absorption.

Figure 10 .
Figure 10.The failure mechanism of LPBF-processed structures.(a)-(d) Fracture morphologies of the failure wall under various magnification.The morphologies of the compressed structural fragments from (e) bionic hybrid structure (BHS), (f) regular corrugated structure (RCS) and (g) the straight wall structure (SWS).
(a)), characterized by river patterns and cleavage steps (figure 10(b)).Additionally, crystalline patterns, dimples, and micropores were observed (figures 10(c) and (d)).These observations confirm that the fracture occurred through a quasi-cleavage mechanism.The fracture surface of the wall exhibited a relatively smooth and flush appearance (figure 10(a)), characterized by river patterns and cleavage steps (figure 10(b)).Additionally, crystalline patterns, dimples, and micropores were observed (figures 10(c) and (d))

Figure 11 .
Figure 11.The shape memory function of LPBF-processed structures.Cyclic loading-unloading stress-strain curves for LPBF-processed structures: (a) the bionic hybrid structure (BHS), (b) the regular corrugated structure (RCS), (c) the straight wall structure (SWS).(d) Variations of total mechanical energies at different loading cycles of three structures.Variations of mechanical energies at different loading cycles: (e) BHS, (f) RCS, (g) SWS.(h) Structural morphologies of LPBF-processed structures after the last cycle.(i) Schematic of water-bath heating.(j) Structural morphologies and (k) recovery rates of LPBF-processed structures after heating after the last cycle.
(a)), while the middle positions of the vertical walls of the RCS and the SWS were observed to fracture first (figures 9(b) and (c)).Comparatively, the BHS appeared to have a weaker stress concentration in the vertical walls, as indicated by the SD plot of figure 13.To further support this observation, histograms were plotted

Figure 12 .
Figure 12.Mechanical behavior of three LPBF-processed structures under different loading cycles.Variation of peak stress with an increasing cycle of (a) the bionic hybrid structure (BHS), (b) the regular corrugated structure (RCS), and (c) the straight wall structure (SWS).The change of elastic modulus of (d) BHS, (e) RCS, and (f) SWS with the increase of loading cycle.Irreversible strain and irreversible strain increments ∆ε at different loading cycles: (g) BHS, (h) RCS, and (i) SWS.

Figure 13 .
Figure 13.The FE simulation results of designed structures under 7.5% strain.Showing the Von Mises stress distribution (SD), histograms of nodal stress in a single vertical wall, and martensite volume fraction (MVF) for the bionic hybrid structure (BHS), the regular corrugated structure (RCS) and the straight wall structure (SWS).The insets in the histograms reflect the stress distribution in a single vertical wall.

3 .
LPBF-processed BHS underwent single-stage transformations during the heating/cooling process and mainly consisted of the B2 phase with a small amount of the B19 ′ phase.The overlap region of the two tracks depletes Ni in the matrix and exacerbates the chemical composition differences between the NiTi powder and the as-manufactured alloys.A strong (001) crystallographic orientation was observed in the LPBF-processed BHS, with grain sizes ranging from 3.39 µm to 73.60 µm and an average size of 9.85 µm.4. The LPBF-processed BHS exhibited excellent load-bearing capabilities, as evidenced by distinct first peak stresses of 18.2 MPa.It demonstrated the ability to withstand pressures equivalent to 25 000 times its weight without significant deformation.During compression, the LPBFprocessed BHS exhibited local buckling and progressive failure, which played a crucial role in efficient EA.The unique wall shape of BHS reduced stress concentration, and facilitated viscoelastic and elastic energy dissipation, allowing for largely recovering the pre-deformation by heating (over 99% recovery rate).