Additive Manufacturing of 3D Structures Composed of Wood Materials

(1)

www.advmattechnol.de

Additive Manufacturing of 3D Structures Composed of Wood Materials

Doron Kam, Michael Layani, Sheer Barkai Minerbi, Donna Orbaum, Shir Abrahami Ben Harush, Oded Shoseyov,* and Shlomo Magdassi*

DOI: 10.1002/admt.201900158

up to fabrication of a final object, with minimal waste material. The cost of pro- duction is independent of the number of printed objects, and therefore, manufacturing even a small number of copies can be economically justified. Currently available materials for additive manufacturing include thermoplastics, photopolymers, ceramics, and metals.^[2–5] However, to date, there are no reports or industrial processes that enable 3D printing of objects solely from wood components. Herein, we report on the 3D printing of objects composed of 100% wood-based materials, using jet binder and extrusion printing technologies.

In a novel additive manufacturing approach presented here, we utilize wood flour (WF), a finely pulverized wood or reclaimed wood, considered a low value byproduct of the wood industry, in combination with a binder composed of cellulose nanocrystals (CNCs) and xyloglucan (XG), materials that can potentially be extracted from industrial side streams. CNCs are highly crystalline, nanosized, rod-shaped particles that are extracted from natural cellulose sources by hydrolysis,^[6] while XG is the most abundant hemicellulose found in the primary cell wall of spermatophytes (except grasses),^[7] where it inter- twines with cellulose and other cell wall components, and is hypothesized to contribute to the plant cell wall extensibility.^[8]

Cellulose and hemicellulose have previously been printed sepa- rately, primarily to produce 3D hydrogel structures.^[9,10] The nanocomposite structure of wood consists of complementary materials and cellular structures, chemically bound to provide trees with the superior material properties, such as low density and the thermal resistance required to withstand extreme environmental conditions.^[11] At the plant cell wall dimension, cellulose crystallinity is the main strength-providing component in cellulose microfibrils and hemicelluloses, such as xyloglucan, glue the microfibrils together into a composite structure that is both strong and tough.^[12]

Recent reports related to 3D-printed wood include several printing approaches. For fused deposition modeling (FDM) 3D printers, “wood-like” filaments embedded in a synthetic polymer (poly(vinyl alcohol) (PVA) or acrylonitrile butadiene styrene (ABS))^[13,14] were printed by extrusion above the melting point of the polymer. The filaments, mainly composed of polylactic acid, are already commercially available. For extrusion-based 3D printing, a mixture of sawdust with a soluble cellulosic derivative (methylcellulose)^[15] or with a commercial 3D objects composed of 100% wood components are 3D printed utilizing

wood flour microparticles dispersed in a matrix composed of cellulose nanocrystals and xyloglucan. In the printed object, a wood waste product is “glued” with extracted wood products, to be a substitute for pristine wood. 3D printing is used to maximize conversion of low value materials into final products that exhibit visual, textural, and physical properties of natural timber. Several 3D printing technologies are applied to achieve a wide range of densities, mechanical properties, colors, and morphologies as well as high thermal insulation. Furthermore, the 3D printing process enables predesigning of fiber layout in the printed wood, which enables control of shrinkage orientation.

D. Kam, Dr. M. Layani, S. Barkai Minerbi, D. Orbaum, S. Abrahami Ben Harush, Prof. S. Magdassi

Casali Center of Applied Chemistry Institute of Chemistry

The Hebrew University of Jerusalem Jerusalem 91904, Israel

E-mail: shoseyov@agri.huji.ac.il D. Kam, Prof. O. Shoseyov

Department of Plant Sciences and Genetics in Agriculture Robert H. Smith Faculty of Agriculture

Food and Environment

The Hebrew University of Jerusalem Rehovot 76100, Israel

E-mail: magdassi@mail.huji.ac.il

The ORCID identification number(s) for the author(s) of this article can be found under https://doi.org/10.1002/admt.201900158.

Printed Wood

The word “material” originates from the Latin word “ma¯teria”, which means wood, which has always been an essential raw material. Wood processing methodologies are continually being developed, with the general aim of maximizing forest- based economies by utilizing side streams to create sustainable wood-based products as replacements for petroleum-derived materials.^[1] At present, wood resources are very rich, and the conventional wood processing technology is very mature, and is based either on top-down subtractive processes, where a tree is first chopped down and then either carved or cut into smaller pieces or on molding processes based on synthetic resin materials. Conversely, 3D printing is a bottom-up, additive manufacturing process that builds up the object of interest through a layer by layer deposition process. This process enables formation of complex structures, from computer design

(2)

wood glue,^[16] was printed. For binder jet 3D printing, wood powder was used, while the binder, composed of lignin sul- fonate, sodium silicate, or PVA solutions, was inkjet printed.^[17]

Another additive manufacturing approach used a blend of wooden chips with powders composed of cement, gypsum, methylcellulose, and sodium silicate, onto which water was sprayed through a series of masks.^[18] However, the binders typ- ically used in the wood are toxic and are generally categorized into two main synthetic families: formaldehyde-based resins and methylene diphenyl diisocyanate (MDI). Today, tight legis- lation of formaldehyde content in consumer products (less than 0.05 ppm formaldehyde and moving toward “no added formaldehyde” products) and the high toxicity of MDI^[19] restrict and limit their use.

Inspired by the natural structure of wood and its basic building blocks, we attempted to 3D print wood objects that mimic the chemistry, structural, mechanical, and thermal properties of natural timber. The process exploited a low-cost wood product in combination with key wood-based additives via 3D printing. To this end, wood objects were printed using water-based inks prepared from varying ratios of XG, CNCs as a binder, and WF particles, and their physical properties were analyzed after drying. Our study was performed in two stages:

exploring and optimizing ink compositions for the fabrication of wood-based objects by molding, and utilizing the optimal compositions in 3D printing (Table S1, Supporting Information).

The 3D objects were fabricated via extrusion and inkjet-based printing technologies. In extrusion-based technologies, ink is extruded through a nozzle to build up a 3D object, whereas, in inkjet technology, ink droplets are jetted out and can serve as a binder at desired locations (binder-jet printers).

In the extrusion approach, we utilized two techniques, direct ink writing (DIW) under ambient conditions and direct cryo writing (DCW) (Figure 1), a technique that integrates both 3D printing and cryotemplating as the ink is extruded onto a cold plate and frozen in a controlled fashion. In both techniques, the ability to extrude ink relies on the rheological properties of the ink. CNCs exhibit shear thinning rheology,^[20] in which the viscosity decreases as a result of applied stress, while removal of the stress enables reversion of the viscosity toward its initial value.^[21] While addition of XG and WF did not interfere with this general behavior, only some of the ink compositions could be extruded (highlighted in yellow in Figure S1 in the Supporting Information). As the piston of the extruder applies stress, the ink flows out from the nozzle but recovers its high viscosity once deposited, thus preventing its flow once it is deposited on the building platform. The yield stress (Table S3, Supporting Information), storage and loss modulus, which are rheological properties critical to maintenance of the shape of the extruded filaments in DIW processing, are shown in Figures S2–S3 in the Supporting Information.

Both techniques require postprocessing to dry the printed wooden sample. In the DIW method, samples were left to dry at room temperature (RT), which caused a volumetric decrease of

≈40% and a concurrent density increase from 350 to 850 kg m⁻³. As water is removed, CNCs and XG bind to the WF particles to form a dense rigid structure. The resulting structures were extremely dense and could be processed with conventional tools utilized for natural wood (Video S1, Supporting Informa- tion). Figure S4 in the Supporting Information compares the cross-sections of polished printed objects with that of commercial engineered wood. In the DCW approach, sublimation

Figure 1. A) Schematic illustration of the extrusion-based 3D printing technique. B) Direct ink writing (DIW) of mashrabiya, a wooden “harem window”.

C) Multimaterial printing of two wood types into a chess board model. D) Direct cryo writing (DCW) of a nut and screw and E) cross-section of a DCW-printed sample.

(3)

of the ice prevents shrinkage, and resulted in very low-density materials (≈100 kg m⁻³) (Figure 1D). Utilization of these printing approaches yielded wood-based objects that covered a wide range of densities, which matched those of natural wood.

In natural wood, density depends on the type of the tree, and ranges from 160 kg m⁻³ for light woods such as balsa, up to 1300 kg m⁻³ for a dense wood such as ebony.

Interestingly, since DCW extrudes ink onto a cold stage, directional freezing occurs within the printed material, resulting in internal structuring of the deposited material. Cross-sectional light microscope images (Figure 1E) show alignment perpen- dicular to the printing platform, which demonstrates a cellular structure resembling the structure of natural wood.^[22]

When simultaneously extruding with multiple extruders, a multimaterial 3D-printed object can also be fabricated. As an example, a model chessboard was printed using a combination of maple and eucalyptus-based inks (Figure 1C). We noted that the object appeared homogenous, with no delamination between different parts, since the same binder composition was used for both inks (Figure S5, Supporting Information).

Complex structures, which require support and high-resolution objects are difficult to obtain using extruder-based printing.

DIW resolution is strongly dependent on nozzle diameter, typi- cally resulting in moderate resolution of 200–1000 µm,^[23] and in extreme cases by using microcapillary extruder, the resolution can go as low as 10–100 µm.^[24,25] In contrast, typical binder-jet printers enable fabrication of objects with higher resolution due to the inherent performance of commercial printheads (600–300 dpi corresponding to 42–84 µm).^[26] More- over, the ability to print objects without the need for additional

support materials, and recycling the residual powder material is one of the great advantages of binder jet.^[27] In binder-jet printers, a binder is jetted from an inkjet printhead onto pre- defined locations on a thin bed of powder (Figure 2A). For this purpose, we prepared inkjet ink comprised of CNC and XG. Of note, 2D inkjet printing has been reported previously,^[28] however, CNCs were not used as a binder.

The first step prior to 3D binder-jet printing was to print on a solid substrate. The morphology of dried CNC inkjet droplets printed on a silicon wafer presented as uniform and repeti- tive “coffee stain” rings (Figure 2C; Figures S4–S6, Supporting Information), which formed due to flow of the CNCs toward the rim of the droplets.^[29] To create a continuous pattern, several layers were printed, as shown in Figure 2B.

Next, we used a multicolor binder-jet printer, in which the powder material was WF, and the binder was composed of XG and CNC. Each of the binder components was placed in a separate section of the printhead cartridge, thus enabling separate control of XG and CNC. This setup enabled control of the ratio between the two binder components as well as the ratio of binder to WF powder by control of the number of printed droplets. Figure 2D,E presents typical 3D wood-based objects prepared by this approach.

Extrusion-based 3D printing enables control of direction- ality via shear-induced alignment caused by forcing material out from a nozzle.^[30] Thus, printed objects can exhibit anisotropic mechanical distortion through careful design of printing pathways. Trees are also anisotropic, with cylindrical sym- metry, which results in uneven shrinkage during drying, mani- fested by different warping aspects such as, cup, bow, twist, or

Figure 2. A) Schematic illustration of the inkjet-based 3D printing technique. B) 2D inkjetted CNC to form the HUJI symbol on a silicon wafer. C) AFM image of one inkjet droplet. D) Binder jet window model and E) binder jet cylinder model.

(4)

crook.^[31] Figure 3 illustrates successful 3D-printed wood object designed to exhibit the wood warping behavior of real wood upon drying.

After printing, we quantified the physical properties of objects printed with ink containing varying rations of XG and CNC. Mold casting was achieved by either casting in a mold followed by drying at RT or freeze casting followed by lyophilization. Different ratios of XG:CNC in aqueous suspension were mixed with WF from Eucalyptus at a constant total solid mass (Table S2, Supporting Information). It was found that the compressive modulus and strength increased with increasing CNC concentrations, for samples obtained by drying at RT (Figure S9, Supporting Information).

Next, while maintaining a fixed ratio of XG:CNC, we evaluated the effect of binder/

WF ratio, in the range of 0.001–0.03. Sam- ples that were freeze cast exhibited low density and were highly porous (Figure S10, Supporting Information). In the freeze casting process, samples were directionally frozen on a cold stage, resulting in vertical ice crystal growth, such that an aligned and porous structure is obtained after ice sublimation.^[32] As shown in Figure S7B–D in the Supporting Information, increases in the binder concentration led to a significant increase in the compressive modulus and strength of the wooden object structure, indi- cating strong bonding between the XG:CNC binder and the WF particles.

Unconfined compression tests performed on cylindrical objects prepared by the different processes indicated that the modulus and strength increased as the proportion of binder increased (Figures S8–S12, Sup- porting Information). Ashby’s plots were used to compare the mechanical properties of the 3D-printed wood-based objects to those of natural woods (Figure 4). As shown, the modulus and strength of the printed woods were within the range of natural woods. The exact mechanism leading to similar mechanical properties to those of wood is not known at this stage, and we hypothesize that it could be assigned to the binder material, which was similar to that of natural wood cell. Preliminary three-point bending tests conducted on samples 14.5% maple resulted in Young’s modulus of 1200 MPa.

Ongoing quantitative analyses of other samples are being performed and will be presented in a future report.

Aside from exploitation of its mechanical properties, wood is also used in construction to improve thermal insulation. The thermal conductivity of the wood-based objects 3D printed by DIW was 0.085 W mK⁻¹, which is at the low range of thermal Figure 3. Images of direct ink writing (DIW) printed wood and 3D scans of different

predesigned lumber cut warping conditions. Arrows indicate printing pathway directions, corresponding to plant cell arrangement (scale bar: 10 mm).

Figure 4. Mechanical properties, determined by an unconfined compression test versus density (Ashby plots) of 3D-printed wood (triangles: red—

CNC-CF, pink—CNC-LAB), mold samples (circles: black—Eucalyptus, blue—freeze cast), and natural balsa, oak, and MDF (raw data presented in Figures S9–S12, Supporting Information).

(5)

conductance of natural woods, while that of the wood by 3D-printed DCW was 0.05 W mK⁻¹, which is remarkably low for woods objects. For comparison, medium-density fiber- board (MDF) and beech plywood have thermal conductivities of 0.1069 and 0.1581 W mK⁻¹, respectively.^[33] Moreover, when immersing a multimaterial 3D-printed object in water, under constant stirring for one hour, some disintegration of the printed objects was observed, mainly starting at the interface between the two materials, in addition to slight absorption of water (Video S2, Supporting Information). After drying, the object returned to its initial properties.

In this work, we present, for the time, additive manufacturing using ink composed of 100% wood and extracted wood products without any synthetic binder. The presented approaches for 3D printing of wood-based objects, enabled hierarchical structuring, and control over the macroproperties of the resulting objects. The ink components both bear a low environmental footprint, and avoided usage of fossil oil-based resins that are commonly used in industrial engineered wood. We expect that the presented printing approaches and material compositions will open new directions in the field of additive manufacturing, overcome traditional wood industry barriers, and exploit wood waste.

Experimental Section

Materials: Four types of WF were tested: Eucalyptus wood self- grinded using an angel grinder—referred to as Eucalyptus WF, WF from hardwood with particle size distribution of 92% < 75 µm and 2% > 150 µm (MOD-EASY FIBER-75, LA.SO.LE EST)—referred to as WF 75, 80 mesh of pine (8020, American Wood Fibers, Inc.)—referred to as Pine WF, and 80 mesh of maple (8010, American Wood Fibers, Inc.), referred to as Maple WF. XG from tamarind seed was supplied by Megazyme Inc., Bray, Ireland (Lot #150901). Two types of CNC were used: commercially available CNC obtained from Celluforce Inc., Montreal, Canada (freeze dry, LOT #2015-009), referred to as CNC- CF, and homemade CNC, prepared as described below, referred to as CNC-LAB.

Preparation of CNC-LAB: 80 g of dried bleached, TEMBEC softwood kraft pulp (60 °C in oven for over 48 h) were vigorously mixed, with a Teflon stirrer, with 1.4 L of 61% sulfuric acid (AR-p), in a 2L double- walled reactor, at 50 °C, 90 rpm, 2.5 h. The reaction was terminated by mixing the slurry with tenfold of cold distilled water (DW), following with three centrifugations (20 °C, 6000 rpm, 10 min), to remove the remaining acid. Finally, CNC was dialyzed against distilled water for at least two weeks (12-1400 Daltons, Medicell Membranes LTD) and then sonicated and filtered (Watmann 41 filter paper).

Ink Preparation: All XG suspensions started as 2 wt% XG, which was prepared by vigorously mixing XG with DW in glass vial for 1 h, at 80 °C, until a clear suspension was achieved. CNC-CFs were suspended in DW (10 wt%) and sonicated. Different ratios of CNC:XG:WF (Table S2, Supporting Information) were mixed for 5 min, using a planetary mixer (AR-100, THINKY Co. Ltd, Japan).

DIW Extrusion Printing: Wood ink (Table S2, Supporting Information) were DIW-printed using a Hyrel3D 30M (Hyrel International, Inc.

Norcross, USA) printer equipped with a disposable syringe extruder (SDS-10) with a 10 mL luer-lock syringe mounted with 1.291 mm (16G) smoothflow tapered tip (Nordson EFD, USA). G-code files were prepared via Slic3r software, and object were printed with speed rate of 5 mm s⁻¹.

DCW Extrusion Printing: A DIW apparatus with a temperature- controlled copper platform was used. A copper plate was glued on a thermoelectric cooler (TEC) using conductive grease, while the TEC was mounted on a water block heat exchange (customthermoelectric.com).

The TEC temperature was controlled by a power supply (EA-PS 2042-20B, Elektro-Automatik). Different wood inks were used for DCW (Table S2, Supporting Information). The printed samples were post-treated by lyophilization (Labconco Freezone 2.5, Missouri, USA).

Inkjet Printing: Inkjet printing were performed using a Dimatix DMP- 2831 piezoelectric ink jet printer (Dimatix-Fujifilm Inc., Santa Clara, USA), equipped with a 10 pL cartridge (DMC-11610). The print head had 16 nozzles, each with a diameter of 21 µm. CNC-CF (2 wt%, Table S2, Supporting Information) was printed on a 3:1 piranha-cleaned P(100) silicon wafer (Item# 452, University Wafers, Inc. South Boston, USA).

The jetting frequency was set to 1000 Hz, using 200 or 600 dpi. Platform temperature was set to 60 °C.

Binder Jet Printing: Binder jet experiments were performed using a Cometrue T10 binder jet printer (MicroJet Technology Co., Ltd. Taiwan) equipped with two print heads (TJ-865CM and TJ-865YK). Each print head was empty thoroughly with at least 1 L of DW after original ink was empty. Then, the clear chambers were filled with binder-CNC (CNC-CF) ink, while color chambers (cyan, magenta yellow, and black) were filled with binder-XG (Table S2, Supporting Information). The surface tension of the inkjet ink was adjusted by adding BYK 348 (BYK, Germany) as a wetting agent. The dependence of surface tension on surfactant concentration is presented in Figure S13 in the Supporting Information.

High-resolution printing setup was selected, with level 3 counter width and layer height of 0.08 mm. WF 75 was used as the powder in all experiments.

Molded Samples: Two grams of the different inks (Table S2, Supporting Information) were inserted into a cylinder mold (D = 10 mm, H = 20 mm) with an adjustable cap and dried at 40 °C. After 24 h, samples were removed from the mold and dried at 70 °C for at least another 24 h.

Freeze-Casting Molding: CNC:XG (1:10) was mixed with WF at different concentrations (0.1–3 wt%, Table S2, Supporting Information). Ink (2 g) was placed in a mold (D = 21 mm, H = 10 mm) on a copper plate placed on top of a TEC. Then, directional ice crystal froze the sample from substrate upper followed by lyophilization, until the entire water content sublimated.

Mechanical Testing: Unconfined compression tests were performed using an Instron universal testing machine (Model 3345, Instron Corp.) equipped with a 100 N or 5 kN load cell, set at a displacement rate of 2 mm min⁻¹. Three repeat specimens were tested and analyzed (apart from Figure S8 in the Supporting Information where two specimens per concentration were tested) using MATLAB. An average of the measurements and error was plotted. Modulus was calculated from the slope of the linear region, while the stress was measured from the end of the linear region.

Three-point bending tests were performed on five rectangle specimens (6 mm thickness, 18 mm width) with 60 mm support spam at a rate of 1.5 mm min⁻¹.

Rheology: A Haake Rheostress 6000 Rheometer, coupled with a temperature controller RS6000 and controlled by Rheowin 4.60 software (Thermo Fisher Scientific Inc.), was used for rheology measurements.

All measurements were performed at RT and were operated by a MP 35 lower plate and P35 TiL upper plate. Three independent 1 mL sample tests were obtained, but only one representative curve is presented in Figures S1–S3 in the Supporting Information.

Viscosity evaluations were performed in controlled rate (CR) mode (0.1–25 1 s⁻¹ shear rate). Oscillation amplitude sweep tests were performed in controlled stress (CS) mode, at a frequency of 1 Hz between 0.1 and 1000 Pa stresses. Oscillation frequency sweep tests were performed in CS mode, at a 10 Pa stress between 0.1592 and 15.92 Hz frequencies.

Atomic Force Microscopy (AFM): CNC dilute suspensions (0.005 wt%) were spin coated (WS-650-23, Laurell Technologies Corporation, North Wales, USA) at 4000 rpm for 30 s onto 3:1 piranha-cleaned P(100) silicon wafers (Item# 452, University Wafers, Inc. South Boston, USA).

A NanoWizard 4 (JPK instrument, Berlin, Germany) atomic force microscope was used to characterize CNC inkjet droplet morphology.

Images were acquired in noncontact mode, under ambient conditions,

(6)

using PPP-NCHAuD cantilevers (spring constant 42 N m⁻¹, resonant frequency ca. 330 kHz, Nanosensors, Switzerland).

Thermal Conductivity: Thermal conductivity of cylindrical samples (D = 10 mm, H = 20 mm) was measured using a TCi Thermal Conductivity Analyzer (C-Therm Technologies Ltd, Canada). Modified Transient Plane Source (MTPS) Sensors were used.

Optical Microscopy: Morphology and surface structure of printed objects were characterized under an optical microscope (CX31, Olympus, Tokyo, Japan), equipped with a camera (E330 with 1.2X attached, Olympus).

Optical Profile: Dried inkjet droplets were coated with 50 nm gold and were imaged using a Contour GT-K1 (Bruker, Tucson, USA) optical profile, equipped with x20 and x50 objectives. Collected data were analyzed using vision64 software.

Surface Tension: Surface tension of the different inks was measured using an OCA15EC device and SCA software (Dataphysics GmbH).

Seven measurements were taken and average interfacial tension and standard error were calculated.

3D Scans: 3D scans of wooden objects were obtained by a DAVID SLS-3 scanner, equipped with a 3D automatic turntable pro (Hewlett- Packard Development Company, USA). Curvature mapping of the scans was performed with 3ds Max software (Autodesk Inc., USA).

Supporting Information

Supporting Information is available from the Wiley Online Library or from the author.

Acknowledgements

The authors thank Dr. S. Eliav and Dr. I. Sheky from the HUJI Nanocenter for technical help with optical profile imaging, M. Chasnitsky and Prof.

I. Braslavsky for providing the peltier plate setup and helping with AFM imaging. Useful discussions with Dr. T. Abitbol of RISE Bioeconomy are gratefully acknowledged. D.K., M.L., O.S., and S.M. conceived and designed the project and experiments. D.K., S.B.M, and D.O. printed objects via DIW. D.K. and S.A.B.H printed objects with a binder jet.

D.K. prepared all other 3D-printed samples, performed and analyzed all experiments. D.K., M.L., O.S., and S.M. wrote the manuscript.

All authors discussed the results and contributed to revisions of the manuscript.

Conflict of Interest

Patent applications have been filed under serial numbers 62/614437 and 62/614438 with the U.S. Patent and Trademark office.

Keywords

3D printing, nanocellulose, natural materials, sustainable, wood

Received: February 20, 2019 Revised: April 17, 2019 Published online: June 11, 2019

[1] J. Song, C. Chen, S. Zhu, M. Zhu, J. Dai, U. Ray, Y. Li, Y. Kuang, Y. Li, N. Quispe, Y. Yao, A. Gong, U. H. Leiste, H. A. Bruck, J. Y. Zhu, A. Vellore, H. Li, M. L. Minus, Z. Jia, A. Martini, T. Li, L. Hu, Nature 2018, 554, 224.

[2] J. R. Tumbleston, D. Shirvanyants, N. Ermoshkin, R. Janusziewicz, A. R. Johnson, D. Kelly, K. Chen, R. Pinschmidt, J. P. Rolland, A. Ermoshkin, E. T. Samulski, J. M. DeSimone, Science 2015, 347, 1349.

[3] J. H. Martin, B. D. Yahata, J. M. Hundley, J. A. Mayer, T. A. Schaedler, T. M. Pollock, Nature 2017, 549, 365.

[4] Z. C. Eckel, C. Zhou, J. H. Martin, A. J. Jacobsen, W. B. Carter, T. A. Schaedler, Science 2016, 351, 58.

[5] F. Kotz, K. Arnold, W. Bauer, D. Schild, N. Keller, K. Sachsenheimer, T. M. Nargang, C. Richter, D. Helmer, B. E. Rapp, Nature 2017, 544, 337.

[6] N. Lin, J. Huang, A. Dufresne, Nanoscale 2012, 4, 3274.

[7] H. V. Scheller, P. Ulvskov, Annu. Rev. Plant Biol. 2010, 61, 263.

[8] Y. B. Park, D. J. Cosgrove, Plant Cell Physiol. 2015, 56, 180.

[9] G. Siqueira, D. Kokkinis, R. Libanori, M. K. Hausmann, A. S. Gladman, A. Neels, P. Tingaut, T. Zimmermann, J. A. Lewis, A. R. Studart, Adv. Funct. Mater. 2017, 27, 1604619.

[10] K. Markstedt, A. Escalante, G. Toriz, P. Gatenholm, ACS Appl.

Mater. Interfaces 2017, 9, 40878.

[11] P. Fratzl, R. Weinkamer, Prog. Mater. Sci. 2007, 52, 1263.

[12] D. Fengel, G. Wegener, Wood: Chemistry, Ultrastructure, Reactions, Walter De Gruyter, Berlin 2011.

[13] D. Rigotti, V. Duc, H. Nguyen, A. Cataldi, A. Pegoretti, Mater. Today Commun. 2018, 15, 1.

[14] X. Feng, Z. Yang, S. S. H. Rostom, M. Dadmun, Y. Xie, S. Wang, J. Appl. Polym. Sci. 2017, 134, 45082.

[15] M. Rosenthal, C. Henneberger, A. Gutkes, C.-T. Bues, Eur. J. Wood Wood Prod. 2017, 76, 1.

[16] M. Kariz, M. Sernek, M. K. Kuzman, Eur. J. Wood Wood Prod. 2016, 74, 123.

[17] H. Zeidler, D. Klemm, F. Böttger-Hiller, S. Fritsch, M. J. Le Guen, S. Singamneni, Procedia Manuf. 2018, 21, 117.

[18] K. Henke, S. Treml, Eur. J. Wood Wood Prod. 2013, 71, 139.

[19] G. A. Ormondroyd, in Wood Composite (Ed: M. Ansell), Woodhead Publishing Lt, Cambridge, UK 2015, pp. 47–66.

[20] T. Abitbol, D. Kam, Y. Levi-Kalisman, D. G. Gray, O. Shoseyov, Langmuir 2018, 34, 3925.

[21] R. Koppolu, T. Abitbol, V. Kumar, A. K. Jaiswal, A. Swerin, M. Toivakka, Cellulose 2018, 25, 6055.

[22] R. A. Savidge, Bull. Geosci. 2007, 82, 301.

[23] S. Kyle, Z. M. Jessop, A. Al-Sabah, I. S. Whitaker, Adv. Healthcare Mater. 2017, 6, 1.

[24] A. Camposeo, L. Persano, M. Farsari, D. Pisignano, Adv. Opt. Mater.

2019, 7, 1800419.

[25] J. A. Lewis, Adv. Funct. Mater. 2006, 16, 2193.

[26] J. Ingaglio, J. Fox, C. J. Naito, P. Bocchini, Constr. Build. Mater. 2019, 206, 494.

[27] B. Utela, D. Storti, R. Anderson, M. Ganter, J. Manuf. Processes 2008, 10, 96.

[28] M. Roman, F. Navarro, in Model Cellulosic Surfaces (Ed: M. Roman), American Chemical Society, Washington DC 2010, pp. 157–171.

[29] X. Mu, D. G. Gray, Cellulose 2015, 22, 1103.

[30] A. Sydney Gladman, E. A. Matsumoto, R. G. Nuzzo, L. Mahadevan, J. A. Lewis, Nat. Mater. 2016, 15, 413.

[31] S. Pang, Wood Sci. Technol. 2002, 36, 75.

[32] S. Deville, E. Saiz, R. K. Nalla, A. P. Tomsia, Science 2006, 311, 515.

[33] W. Sonderegger, P. Niemz, Eur. J. Wood Wood Prod. 2009, 67, 313.