“Wood-nacre”: Development of a Bio-inspired Wood-Based Composite for Beam and 3D-Surface Elements with Improved Failure Mechanisms

Following the natural structure of the nacre, the material studied consists of a multitude of hexagonal tiles that are glued together in an offset manner with a ductile adhesive. This so-called “wood nacre” consists of macroscopic tiles of birch wood veneer with a thickness of 0.8 mm and a size of 20 or 10 mm in diameter in order to mimic the aragonite tiles and the ductile PUR-adhesive corresponds to the layers of collagen in between. E-modulus (MOE), bending strength (MOR) and impact bending strength of the samples were determined and compared with reference samples of birch laminated wood. The hierarchical layered structure of the tiles does not cause any relevant loss in stiffness. Like nacre, “wood nacre” also shows tough fracture behaviour and a high homogenization effect. However, strain hardening and high fracture toughness of the natural model could not be fully achieved. The reason for this is the insufficient ratio between the strength and stiffness of the veneer layers and the adhesive. By adjusting the size of the tiles, increasing the strength and surface roughness of the veneers, e.g. by densification, and using more ductile adhesives that can be applied in smaller layer thicknesses, it should be possible to better reproduce the natural ratios of nacre and thus achieve a significant improvement in the material properties of “wood nacre”. In addition to the mechanical properties, the high potential of the new material lies in the possibility of producing 3D shell-shaped elements for lightweight wood hybrid construction.


Introduction
Nacre is a natural composite material consisting of aragonite tiles (calcium carbonate) embedded in an organic matrix that provides the material with sufficient fracture toughness in addition to high-strength properties [1]. The excellent fracture toughness is based on the hierarchical structure of two material phases with very different mechanical properties, the very hard mineralised aragonite tiles of CaCO 3 with an E-modulus (MOE) in the range of 60-90 GPa and the very soft protein-matrix with a MOE that is multiple orders of magnitude smaller. The biomimetic principal of this modulus mismatch upon increasing toughness was described by Murali et al. [2]. The aragonite tiles arranged in layers are staggered in relation to each other. In the case of crack formation, different toughening mechanisms like crack deflection at the ductile boundary layers, crack bridging and tablet interlocking occur [1,3,4]. Despite the high volume fraction of the brittle aragonite tiles, a relatively high fracture toughness is achieved through the interaction with the ductile boundary layers. Similar mechanisms based on hard and soft phases arranged in complex hierarchical structures and ranging from the nano-to the macro-scale are also observed in bones, teeth and bamboo [4]. The resulting hybrid materials often have unique combinations of strength and toughness with low weight but are difficult to replicate synthetically.
Wood consisting of cellulose, hemicellulose, and lignin is another hierarchically structured natural material [5]. In relation to its low density, ranging between 200 and 1000 kg/ m 3 , wood has a relatively high strength and MOE varying between 8 and 16 GPa [6]. In compression, the natural 1 3 material shows high energy absorption at failure. When loaded transversely to the grain, wood shows some similarities to the failure mechanism of a honeycomb structure. This is due to the collapse of individual cells [7]. Longitudinally to the fibre, the so-called kink banding occurs under compressive stress [8], a buckling of the cellulose microfibrils into the rather ductile lignin matrix of the cell wall. Therefore, a pronounced ductile failure behaviour can also be observed in the longitudinal direction of the wood. By contrast, due to the semi-crystalline properties of cellulose wood shows brittle failure in tension with an elongation at break of around 1-2% depending on the microfibril angle (MFA) within the predominant cell wall layer S2 [9].
It has been shown that wood undergoes brittle fracture under bending on the tensile side, resulting in a primarily transversal cell wall failure [10,11]. The area of the stembranch connections, however, exhibit different fracture processes when overloaded. The crack propagates there in a zigzag pattern due to the alternating orientation of the tracheids in the vicinity of branches [12,13], resulting in considerable fracture toughness. Furthermore, the optimised hierarchical structure of the wood cell wall and the interaction of the rather ductile properties of lignin with the stiff and brittle properties of cellulose also contribute to the excellent mechanical performance of branch-stem junctions.
Due to its good and sustainable availability and low cost, wood has a high potential to be used not only for building applications but also for other high-end material purposes. The use of wood for mechanical engineering and vehicle technology purposes has been intensively investigated and discussed in recent research work [14][15][16][17][18][19]. Hybrid composite materials and structures can be used to develop components from wood and other biomaterials that meet the high crash requirements. For example, Baumann et al. [20,21] showed that crash-relevant elements can be technically realised. Furthermore, Asada et al. [22] and Bauernfeind et al. [23,24] confirmed that energy and resource input, as well as CO 2 -footprint can be minimised by means of such composites.
The developed side impact beam by Baumann et al. [20,21] is based on two highly rigid surface layers of birch plywood (Betula pendula Roth.), which was laminated unidirectionally. A fish-belly-shaped core of paper honeycomb was glued in between. The beam is usually fixed in the doorframe between the A and B pillar of the chassis. In the event of an impact, the beam exhibits a high moment of inertia due to its special shape. If the impactor further intrudes towards the car's interior, the honeycomb core collapses and the moment of inertia of the side impact beam drops to a minimum. Subsequently, the beam acts as a tension element and transmits high forces to the A and B pillars. This ensures that the high impact energy of the impactor is converted into deformation energy.
Other research, such as Moradpour et al. [25], is investigating wood-based materials reinforced by synthetic fibre materials to improve their mechanical properties. However, for reasons ecological assessment, these approaches are increasingly being critically evaluated. Therefore, solutions for homogeneous, mechanically efficient, and formable biobased materials that require minimal use of synthetic materials are being sought.
Pramreiter et al. [26] attempted to develop an energyabsorbing wood composite, inspired by the ultrastructure of mollusc shells. This was achieved by applying the hierarchical crossed lamellar structure (CLS) of the mollusc shells within a wood composite. They demonstrated, that a hierarchical structure made of wooden elements and a synthetic adhesive can be produced without the use of other materials, whereas wood corresponds to the hard component and the glue represents the soft and ductile component of the ultrastructure of the mollusc shells. The goal of the study was to maximise the fracture and deformation energy of beam elements subjected to bending stress by means of targeted crack dissipation. The structure works effectively and also offers the possibility to homogenise the highly varying wood properties.
Compared to man-made fibre-reinforced composites, sheet metal or metal components, crash-relevant wooden composites presented by Pramreiter et al. [26] and Baumann et al. [20,21] have certain limitations in terms of shaping. Especially the concept of Pramreiter et al. [26] involves relatively high costs and is very labour intensive, which limits the use of such wood composites for crash-relevant components and make it difficult to transfer the construction principles from the laboratory scale into practice and thus into industrialised production processes. In addition, it is difficult to produce 3D-surface elements from this structure. Therefore, other technical solutions of hierarchically structured materials with sufficient strength and fracture toughness made out of wood and other bio-based materials are needed. These solutions must enable a technological industrial implementation at acceptable production costs yet easily fabricated into 3D-surface elements.
The solution to the problem proposed in this article is inspired by the construction principles of nacre. Multi-scale interfaces were incorporated into a hierarchical planar macrostructure to mimic the stepwise crack growth behaviour of the hard and damage-resistant bivalve (Saxidomus purpuratus) or nacre. Therefore, hexagonal structures were cut into thin veneers using a laser cutter and were glued on top of each other in a staggered manner to form flat elements. Thus, veneer layers represent the hard-mineralised aragonite tiles of CaCO 3 while a ductile adhesive corresponds to the very soft protein-containing intermediate layers.
In Fig. 1, the possible fracture mechanisms of "wood nacre" are described. If the material fails in the matrix like the natural model nacre, shifting and pulling out of the wooden tiles as well as strain hardening in the stress-strain curve should be observed (see Fig. 1b) This requires having sufficient interfacial shear strength to ensure a proper load transfer from the weak matrix to the stiffer wood tiles. Shallow wood fracture along the adhesive joint and tensile fracture of the wooden tiles (see Fig. 1c) lead to a reduction or even prevention of the desired effect of increasing the fracture energy in case of failure of "wood nacre".
It should be emphasised that the overall aim of this research to see whether the failure behaviour of natural nacre can be replicated using wood veneers and a ductile adhesive. Furthermore, it is hypothesised that the high variability in strength and work of fracture usually associated with wood should decrease through a reproducible failure behaviour. The findings of this study should enable future optimisation of properties of veneers and ductile adhesives related to the production of bioinspired wooden elements.

Material and Methods
Mechanical properties, size and proportions of the pentagonal to hexagonal aragonite tiles and protein-containing intermediate layers were taken from Mayers et al. [27] and Bouville [28] which are as follows: tiles diameter = 8-10 µm, tiles thickness = 0.5 µm, thickness of the protein-containing layers = 20-30 nm, MOE of the aragonite tiles = 79-92 GPa, MOE of the protein layer = 1*10 -9 GPa.
Graded Finnish birch (Betula pendula) veneers (purchased from Koskisen, Järvelä, Finland) with a thickness of 0.8 mm ± 0.05 mm and a supplier's grading quality of A/B [29] were used for the production of the "woodnacre". To exclude knots, fibre deviations in and out of the plane and other defects, the veneers were visually sorted before processing. The veneers were cut into 600 × 400 mm sheets and stored in a standard climate at 20 °C ± 2 °C and 65% ± 5% relative humidity according to standard ISO 554 [30] before further processing with a laser cutter (Epilog Zing 24, Cameo Laser GmbH, Stuhr, Germany). For the laser cutting, a vector graphic was created with AutoCAD 2020 (v23.1, Autodesk, California, USA). The cell size for the hexagonal patterns was 10 and 20 mm based on natural relations derived from literature [27,28] as follows.
The cell size of the wooden tiles resulted from the following considerations. The dimension of the wooden tiles was derived from the diameter and thickness of the aragonite. For nacre, the tile diameter is about 16-20 times the thickness [27]. With a veneer thickness of 0.8 mm and a glue line thickness of 0.1 mm, this results in a tile diameter of 14.4 to 18 mm.
In addition to the size ratios of tile thickness and diameter, the desired failure mechanism is a second consideration for determining the ideal diameter of tiles for the "wood nacre". To reproduce the desired effect from the natural model nacre, it is necessary that the material fails cohesively in the ductile adhesive layers. Therefore, the tiles must be small enough to prevent them from breaking in tension, as this would lead to a brittle failure. Corresponding to this limitation, the concept of critical fibre length was used to determine the maximum tile size. Elements smaller than the critical fibre length will be pulled out and, as with nacre, material failure will occur due to shear within the soft phase of the composite. Therefore, the size of the wood-nacre tiles was estimated as follows: With the breaking force (F max ), the length of an imbedded fibre (l ef ), the fibre diameter (d f ) and the interfacial shear stress (IFSS) (τ), the critical fibre length is calculated according to Eq. (1) of Kelly and Tyson [31].
Corresponding to Zarges and Heim [32] with Eq. (2), the resulting critical fiber length (l c ) can be calculated using the determined IFSS and the fiber tensile strength (σ f ).
For an estimation of a suitable tile length of the "woodnacre", the formula was applied and for using an IFSS of about 3 MPa according to Salca et al. [33] and a tensile strength of about 140 MPa for the birch veneer according to Sell [34] and Wagenführ [35] a critical tile length of 18.67 mm was calculated. To cover the optimal range of the tile size, it was decided to produce "wood nacre" with a tile size of 10 and 20 mm.
In the corners, the material was not completely cut through, which meant that the elements were held together Failing of the matrix material and shifting and pulling out of the tiles. c Fracture of the wooden tiles by thin webs (0.5 mm and 1 mm) preventing the hexagonal tiles from falling apart. After laser cutting, the veneer sheets were sanded (grain 80). A one-component polyurethane adhesive from Collano AG (Sempach, Switzerland) (RP 2760) was used for gluing. A spreading quantity of 140 g/m 2 was applied and distributed with a glue spatula. The individual veneer layers (N = 23) were laid on top of each other with an offset of 5 mm in both directions as shown in Fig. 2. Then, they were pressed in a hot press at a pressure of 0.7 MPa and 40 °C for 180 min to form 20 mm thick panels (see Fig. 3). The resulting glue joint thickness was subsequently determined optically using an incident microscope (Olympus) and the image analysis software ImageJ (Image-J 1.52a; National Institutes of Health, Bethesda, USA). The quantity of polyurethane adhesive corresponded to a glue joint thickness of 0.1 mm.
The panels were stored in a standard climate according to ISO 554 [30] for at least 7 days to ensure complete curing of the adhesive. Subsequently, the panels were cut to 40 × 20 × 400 mm (width × height × length) using a conventional circular saw. The resulting samples were used for the evaluation of mechanical and physical properties. Prior to mechanical tests, the density of each sample was calculated according to the standard DIN 52182 [36].
3-point bending experiments according to EN 408 [37] were carried out by means of a universal testing machine (Zwick/Roell Z-100, Ulm, Germany) to determine MOE, bending strength (MOR), and the work of fracture (WoF). MOE was measured between 10 and 40% of the maximum load in the 3-point bending experiment. The deflection of the specimens was measured with a mechanical extensometer Macrotens (Zwick/Roell, Ulm, Germany) with an accuracy of ± 1.5 µm in the centre of the beam according to the standard. The force was determined with a load cell (max. load capacity = 100 kN, resolution = 1/600,000). Before the bending tests, the specimens were preloaded with a force of 20 N and subsequently tested at a constant crosshead speed of 15 mm/ min. After a force drop of 95%, the test was stopped. The WoF was calculated as the work done up to a residual load of 90% of the maximum force reached, resulting from the force-displacement curves in relation to the original cross-section of the specimen. The energy was calculated per unit volume based on the dimensions of the specimens and the corresponding loads. The bending tests were evaluated using the control software of the universal testing machine testXpert III (Zwick/Roell, Ulm, Germany). Further statistical evaluations were carried out with the statistical software SPSS (version 27, IBM, New York, United States).
The impact bending test was investigated in accordance with the DIN 52 189 [38] standard. For this purpose, test specimens with the dimensions 40 × 20 × 300 mm (W × H × L) were produced and tested at standard climate with a pendulum machine (Wolpert, Vienna, Austria). The impact bending strength was calculated according to Eq. (3), where α is the impact bending strength, W is the work required to break the sample, and w and h corresponds to the width and thickness of the specimen.
To assess the failure mechanism and to observe the fracture surfaces, the specimens were photographed and filmed during the test with a conventional digital camera and camcorder, and the fracture surfaces were examined after the test with a DSX 1000 digital microscope (Olympus, Shinjuku, Tokyo, Japan). Therefore, a brightfield observation mode was chosen at a total magnification of 40 × (objective lens DSX10-XLOB3X). To acquire the images beyond the field of view of the microscope, processing modes were selected for vertical and horizontal image compilation using DSX10 software (version 1.1.2.2, Olympus, Shinjuku, Tokyo, Japan). For the 3D acquisition mode, the entire specimen cross-section was scanned in the vertical direction. The images were merged in the horizontal direction with an overlap of 10% on each side.

Results and Discussion
The selection process of the veneer sheets led to a relatively homogeneous density distribution of the produced "wood nacre", which was 0.714 ± 0.021 g/cm 3 for samples with a tile size of 20 mm and 0.769 ± 0.016 g/cm 3 for the smaller tile size of 10 mm. Both "wood nacre" samples are characterised by a higher density than normal birch wood of 0.64 g/cm 3 [35]. The higher density can be explained by the higher density of the resin (~ 1.1 g/cm 3 ) and the resin content of approx. 15%. A rough estimate considering the resin content gives a density of 0.71 g/cm 3 (0.15 × 1.1 g/ cm 3 + 0.85 × 0.64 g/cm 3 = 0.71 g/cm 3 ). This value is lower than the recorded density of smaller tile samples probably due to their higher number of laser cuts that are filled with PUR resin during bonding.
The MOE of "wood nacre" (see Fig. 4a) with a tile size of 20 mm were found to be slightly higher than solid birch wood reported by Sell [34] and Wagenführ [35], which is in the range of 13.3 GPa up to 16.5 GPa. Regardless of this, samples with a smaller tile size of 10 mm show significantly lower MOE values than samples with a 20 mm tile size, which can be probably explained by a higher elastic deformation of the bonding phase between the wooden elements. The relatively high MOE of the samples with a tile size of 20 mm indicates that the stiffness of the "wood nacre" is no longer negatively affected above a certain tile size. Whereas, if the tile size is too small, it can be assumed to have a reducing effect on the stiffness.
Wagenführ [35] reported MOR values for birch clear wood samples in the range of 120 MPa, which is significantly higher than the values measured for "wood nacre" (see Fig. 4b). The lower strength values can be explained by a different fracture mode in the tensile zone at the bottom of the beam element. In general, solid wood is characterised by a lower compressive strength than tensile strength [39], which initially leads to compression failure when a beam element is overstressed. In the second phase, the stresses in the tensile zone of the beam element are continuously increased and finally there is an abrupt tensile failure and a complete softening of the flexural element. In the compression zone, a typical buckling of the cells (so-called kink banding) can be observed, while in the tension zone a pronounced fibrous fracture surface is the result of an abrupt and brittle tensile failure. The tensile zone of "wood nacre" was weakened by the transverse laser cuts, which presumably causes the specimens to fail in tension rather than first in compression as the stress increases. Therefore, it can be assumed, that the laser cross-cuts between the individual tiles led to a reduction of the MOR in the case of "wood nacre" for both tile sizes. However, significantly higher strength losses are obtained for samples with the smaller tile size (Fig. 4b). For the specimens with the large tile size, the specimens failed as desired within or along the adhesive layer (see Fig. 6a). In the specimens with the smaller tile size, pull-out of the wood elements from the bond was also observed in some cases, indicating that the critical fibre length was not reached in some cases (Fig. 6f). This means that the tile size of 10 mm was probably chosen too small and the optimal tile size for a veneer thickness of 0.8 mm is somewhere between 10 and 20 mm.
The hypothesis, that "wood nacre" fails at first in tension, is supported by the assessment of the force-displacement curves from the bending test. Wood shows a clearly ductile fracture behaviour in bending. After the yield point, which is reached at about 60-70% of the ultimate load, the linear elastic behaviour is merging in a ductile phase (see Fig. 5 continuous line). As already mentioned, this can be explained by the increasing compression failure in the upper compression zone of the bending element. When the maximum tensile stress of the wood material at the underside of the beam is reached, an abrupt force drop can be observed (see Fig. 5). Corresponding to the similar MOEs of solid wood and "wood nacre", their curves (dotted and dash-dotted line in Fig. 5) are close in the elastic phase of the samples. Ductile behaviour is less pronounced for "wood nacre" with tile size 10 and 20 mm and the curves are characterised by higher force drop after reaching ultimate force, which can be possibly explained by a The force-displacement curve of "wood nacre" after reaching the maximum force behaves similarly to wood or plywood, which means that the force drops abruptly to about 50%. Only after this initial brittle fracture does the desired fracture mechanism of nacre take place and the increased fracture energy due to the generation of large fracture surfaces along the tiles comes into action. Thus, from this point, the force-displacement curve falls continuously. However, nacre shows a completely different fracture behaviour than the presented "wood nacre". After the linear-elastic range, the tough collagen between the very stiff aragonite tiles requires a lot of energy to force the crack front through the material. The result is strain hardening, which means that the force partly increases even further or is at least kept at a high level and the material absorbs a significant amount of energy overall.
The assumption that "wood nacre" firstly fails in tension is also in good agreement with the images from the fracture surface, presented in Fig. 6. However, as with the natural model Nacre, this leads to crack dissipation and a back-and-forth jumping of the crack front and thus a stepped fracture pattern (Fig. 6).
For the improvement of the "wood nacre" this means that not only a better ratio for the tile size has to be found, but that the adhesive has to absorb significantly more energy to achieve the desired fracture toughness. In nacre, the MOE of the collagen phase (1*10 -9 GPa) is far below the MOE of the aragonite tiles (79-92 GPa) [27]. Kumpenza et al. [40] give an MOE of about 1 GPa for a similar one-component polyurethane adhesive from the same manufacturer (Collano AG, Sempach, Switzerland). Pramreiter et al. [41] measured birch veneers from Koskisen (Järvelä, Finland) and report a tensile modulus for the material of 13.78 GPa. In the case of "wood nacre", the difference between the E-moduli of the wood tiles and the one-component polyurethane adhesive is consequently significantly smaller than for the natural model nacre. A relatively high viscosity (over 7000 mPa s) PUR was used for the samples produced, which penetrates only slightly into the wood structure. Low-viscosity aminoplastic adhesives such as PRF or MUF show a much bigger interphase [42,43], which may have a positive effect on the mechanical properties of the material. Veneer-based materials with high fracture energy are primarily sought for new highly technical applications of wood [16,17,23,24]. For such applications, aminoplastic resins seem to be less suitable due to their formaldehyde emissions. Therefore, this group of adhesives is not a possible solution. Apart from that, this adhesive group is rather characterised by brittle fracture behaviour.
For a better imitation of the biological model nacre, two limitations have to be met. First, an adhesive that has a sufficiently low viscosity when uncured must be found as the low viscosity allows the adhesive to be applied as uniformly thin as possible. This is a challenge because ductile adhesives usually have relatively high viscosities. This means that the amount of adhesive applied cannot be reduced as well. Therefore, a tough and low-viscosity adhesive is sought that penetrates sufficiently deep into the substrate to stabilise the partly damaged wood structure. It is possible that alternative adhesives such as PVAc, acrylates and epoxy resins give better results. Due to their higher costs, acrylates and epoxy resins are not yet widely used and have been little studied in wood science. However, for new high-quality products with corresponding mechanical and technological performance, it is conceivable that these adhesives can be used successfully.
Second, the wood surface has a natural roughness due to the cellular wood structure, which requires a certain amount of adhesive application for complete wetting of the surface, which is in the order of about 70 g/m 2 [44]. The use of highly densified veneers as the base material for the tiles would bring several advantages. Due to the densification, the surface roughness is highly reduced and the surface is thus much smoother. And the densification more than doubles the tensile MOE and strength of the veneers [45]. In wood bonding, shallow wood fracture often occurs at the boundary of the interphase, causing the two joined parts to fail in a rather brittle way. Wood fracture is also clearly visible in the fracture patterns in Fig. 6. This effect can also be minimised by densification, which can generate an increased cohesive failure in the bonded joint as demanded to imitate the biological model nacre. Higher strength could be achieved by using other wood species with a higher density.
Comparing the fracture energies determined by means of impact pendulum tests and 3-point bending tests yielded only slightly higher fracture energies in the case of the dynamic impact pendulum tests (see Fig. 7). The difference in strain rate dependence also indicates that the fracture process in "wood nacre" is strongly influenced by the failure of the adhesive, which makes the choice of the adhesive a key parameter to increase fracture toughness. The change in adhesive properties offers the corresponding potential with regard to a possible optimisation of the material. Nevertheless, "wood nacre" provides lower fracture toughness compared to solid wood, which is mainly caused by lower MOR values and the lack of compressive failure of the hexagonal tiles on the top of the beam.
Based on the observed fracture surfaces and the significantly lower strength values than solid birchwood, it can be assumed that the tile size and the adhesive bonding of the individual veneer layers can be significantly optimised. Regardless of the tile size, the samples failed primarily at the interface between wood and adhesive (see Fig. 6). The Fig. 7 Work of fracture (WoF) of "wood nacre" measured in 3-point bending (static) and by means of an impact pendulum (dynamic) fracture patterns mainly show a shallow wood failure, which indicates that the potential of the bonding strength between wood and adhesive is not fully utilised. Further improvement of the material is therefore associated with varying the bonding of the wood tiles. In addition to increasing the strength of the veneer and preventing shallow wood failure, tougher adhesives that form very thin glue joints must be used. With these improvements, the proposed "wood nacre" would show better mechanical performance compared to other woodbased composites [46,47].
In addition to the proposed optimizations of the "wood nacre", the material already has the following potentials and advantages in its present form. In relation to its low density (the density varies approximately between 0.2 and 1.2 g/cm 3 depending on the wood species), wood is a highperformance material with very good strength and stiffness properties [5,6]. Typically, the modulus of elasticity of wood parallel to the grain varies between 8 and 16 GPa, while tensile strength values between 80 and 200 MPa may be achieved [6]. The basis for this is the extreme performance of the bio-polymer cellulose. The wooden cell wall exhibits an E-modulus of more than 30 GPa and a tensile strength of up to 1000 MPa [48]. Due to the ductile properties of lignin, in which the cellulose microfibrils with an E-modulus of 70-140 GPa and tensile strength of 7500 MPa are embedded [49], wood exhibits very ductile behaviour in compression, but very brittle fracture behaviour in tension. This means that without special pre-treatment with high losses in strength and stiffness, wood cannot be deformed and deep-drawn under tension such as thermoplastics and metals without damage. 3D shell structures are, therefore, limited in the case of wood hybrid structures. "Wood nacre" represents a possible strategy to overcome these limitations. The layers, produced in the form of hexagonal tiles, make it possible to form them in 3D. In addition, the arrangement of the tiles and the layered structure of the composite create a pronounced homogenization effect of the composite.

Conclusion and Outlook
The initial stiffness of the "wood-nacre" was not affected by the wood-tile structure used as an imitation of the natural building principle. However, the desired ductile fracture behaviour was not observed after the maximum force was reached. Like other wood-based materials, "wood nacre" breaks relatively brittle. Only after the first drop in force crack initiation did occur and the crack propagation typical of nacre follow the natural pattern. In this process, the zigzag crack propagation predominantly follows the glue lines, gaining a corresponding fracture toughness.
It is assumed that the optimal tile size lies within the currently chosen range of 10-20 mm. However, the fracture did not occur within the glue line as desired and the glue was probably too brittle and stiff compared to the low stiffness of the wood material. A lower stiffness of the adhesive will result in a slight decrease in the MOE of "wood nacre", but this may achieve the desired crack deviation and fracture work.
The laboratory samples produced show, that the structural design of the "wood-nacre" composite can be produced highly efficiently and also in an industrially scalable way. With the structure, a way was found to produce a potentially stiff, strong and tough engineered wood-based composite. The construction principle can be used for both beam and surface construction elements. Due to the hexagonal laser cuttings, it is also possible to press the veneer layers spherically to a certain extent into 3D shapes. With these findings, "wood nacre" can also be used as a 3D-shell element and thus offers a large field of application in diverse structural components.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.