Bioinspired nano-architecture for cellulose fibers with spider silk–like toughness

Abstract

Cellulose fibers derived from renewable biomass exhibit exceptional tensile strength through molecular alignment and dense packing, yet their toughness remains limited, typically below 50 MJ m⁻³. Drawing inspiration from the helical nanoarchitecture of cherry bark, we introduce a bioinspired nano-orientation strategy to fabricate regenerated cellulose fibers with a biaxially oriented structure via a scalable microfluidic spinning technique. Combining experimental characterization and molecular simulations, we demonstrate that this biaxial nano-architecture effectively redistributes stress and suppresses crack propagation during deformation, achieving a remarkable fracture strain of 41% alongside a tensile strength of 553 MPa. This synergy yields a toughness of 184 MJ m⁻³, exhibiting highly competitive performance relative to most previously reported cellulose fibers and synthetic polymers, and achieving a mechanical performance on the same order of magnitude as natural spider silk. Moreover, this enhancement extends seamlessly from single fibers to woven fabrics, highlighting its potential for sustainable, high-performance materials in textiles, automotive components, and aerospace applications. Our findings illuminate a design for overcoming the intrinsic brittleness of cellulose fibers, advancing their applicability as eco-friendly structural materials.

Cellulose fibers derived from renewable biomass offer great tensile strength performance, but their toughness remains limited. Here the authors develop cellulose fibers with uniquely high toughness and tensile strength, that can form from single fibers to woven fabrics.

Introduction

Natural materials often derive their exceptional mechanical performance from intricate hierarchical structures, where alignment and architecture synergistically optimize strength, toughness, and energy dissipation^1,2. For example, spider silk fibers achieve remarkable toughness (16 ~ 350 MJ m⁻³) and impact absorption through a combination of highly aligned protein nanocrystals and amorphous regions arranged in a semi-crystalline fibrous structure, enabling efficient stress redistribution under dynamic loads^3–5. Similarly, wood exhibits outstanding toughness and fracture resistance owing to its multilayered secondary cell wall, where cellulose microfibrils are oriented at varying angles across distinct lamellae⁶. This spatially heterogeneous alignment balances stiffness and ductility, allowing wood to withstand complex mechanical stresses. Beyond alignment, twisted and helically wound fiber structures found in biological tissues—such as the spiral architectures in tendons⁷ and plant bark⁸—further enhance strain capacity and toughness by promoting energy dissipation and impact absorption through controlled deformation mechanisms⁹.

Building on these natural design principles, engineered cellulose fiber materials have made significant strides by harnessing molecular and nanoscale alignment through methodologies such as interface hydrogen bond modulation¹⁰, topological regulation¹¹, orientation control^12,13, crystalline engineering², structural modification¹⁴, and fiber twisting^9,13 to achieve tensile strengths approaching or exceeding 1.0 GPa. Advances in fabrication techniques, notably fluidic and microfluidic spinning, enable precise control over fiber morphology and orientation at multiple length scales, facilitating the production of high-performance fibers with tailored mechanical properties^15–17. These methods allow continuous, scalable manufacturing of aligned or hierarchically structured fibers from a variety of biopolymers and synthetic precursors. The integration of fluid dynamics with materials chemistry has yielded fibers exhibiting exceptional stiffness and strength, while recent innovations focus on introducing architectural complexity to simultaneously enhance toughness and strength¹⁸. Such engineered fibers hold immense potential for applications spanning textiles, composites, and structural components, contributing to lightweight, sustainable material solutions.

Inspired by the helical nanoarchitecture of cherry bark^19,20, we report a bioinspired nano-orientation engineering strategy to fabricate ultra-tough regenerated cellulose fibers with a biaxially oriented configuration via a scalable microfluidic spinning system. Experimental characterization and molecular simulations reveal that this biaxial alignment effectively disperses mechanical stress and delays failure, achieving a fracture strain of 41% and a tensile strength of 553 MPa. The resulting fibers exhibit an outstanding toughness of 184 MJ m⁻³—outperforming most reported cellulose fibers and synthetic polymers and achieving a mechanical performance on the same order of magnitude as natural spider silk. This distinctive, scalable approach provides a pathway toward high-strength, high-toughness biomass fibers, offering a sustainable alternative for advanced structural materials with reduced environmental impact.

Results

The bark of the White Ash tree (Fig. 1a) and Cherry (Fig. 1b, Supplementary Fig. 1) exhibits distinctly different textures and structural organizations. The White Ash bark displays a unidirectional alignment along the growth axis²¹, whereas the Cherry bark features a characteristic two-dimensional (2D) helical spring structure (Supplementary Fig. 2)^19,20, as evidenced by deformation patterns in the equatorial and meridional lines of the 2D small-angle X-ray scattering (SAXS) diffraction pattern. Notably, these structural differences correspond to markedly divergent mechanical properties. The 2D-oriented Cherry bark demonstrates a fracture strain of 73.9%, which is 403% higher than that of the unidirectional White Ash bark, and a toughness of 30.5 MJ m^-3, representing a 534% increase, while maintaining comparable mechanical strength (~46.9 MPa) (Fig. 1c). These results indicate that engineering spatially nano-oriented architectures can significantly enhance the toughness of cellulose-based materials.

Fig. 1. Fabrication of bioinspired ultra-tough cellulose fibers in a microfluidic spinning system.

a Photograph of White Ash. b Photograph, schematic diagram, and 2D SAXS diffraction pattern of the Cherry bark with helical spring structure. c Stress-strain curves for Cherry bark and White Ash bark. d Illustration of the fabrication process of bioinspired fibers in a microfluidic spinning system with the green central stream of the modified cellulose solution, the blue sheath streams of good solvents (LiOH/urea/H₂O solution), and the red sheath stream of the coagulating agent (phytic acid solution). e Surface morphology of the wet-spun traditional cellulose fibers. f The axial cross-section morphology of the wet-spun traditional cellulose fibers. g Surface morphology of the bioinspired fibers. h The axial cross-section morphology of the bioinspired fibers. i Toughness of the resulting bioinspired fibers (184 MJ m^-3) is shown to be higher than those of typical cellulose fibers (Lyocell fiber²⁴, Wood fiber¹³, Microfluidic-spun CNF fiber¹⁵, Viscose fiber²³, Bamboo macrofiber²², Wet-spun regenerated fiber¹⁴), error bar represents the standard deviation of 5 independent experiments.

Utilizing a customizable microfluidic spinning system to reconstruct the cross-linked cellulose network, we implemented a precision nano-orientation strategy that endows the fibers with a radial biaxial orientation architecture, successfully realizing the construction of scalable, ultra-tough regenerated cellulose fibers (Fig. 1d). The central channel introduces a loosely cross-linked modified cellulose solution (crosslinking density about 60 ~ 80 mol m^-3) prepared in an 8 wt.% LiOH/15 wt.% urea aqueous system, which not only exhibits excellent spinnability at room temperature (Supplementary Figs. 3, 4, Supplementary Movie 1) but also promotes axial alignment of the cellulose network. As the flow proceeds, it passes through dual sheath streams composed of the good solvent (LiOH/urea aqueous solution) and the coagulation solvent (phytic acid solution). The good solvent sheath disrupts the cellulose flow by inducing dilution, shear, and turbulence, thereby reorienting the surface cellulose network perpendicularly (circumferentially) to the main flow direction and establishing a biaxial alignment in the sol phase. This 2D orientation is subsequently fixed during the sol–gel transition upon contact with the coagulation solvent. Compared to traditional wet-spun cellulose fibers with unidirectional nanostructures (Fig. 1e, f), the bioinspired fibers display perpendicular nanostructures and densely stacked nanofibers both on the surface (Fig. 1g) and within the inner layers (Fig. 1h), closely resembling the natural architecture of cherry bark. Mechanical characterization reveals that these bioinspired fibers achieve an exceptional toughness of up to 184 MJ m^-3, significantly surpassing other typical cellulose fibers, including Wet-spun regenerated cellulose fibers¹⁴, Bamboo macrofibers²², commercial Viscose fibers²³, Microfluidic-spun CNF fibrers¹⁵, Wood fibers¹³, and commercial Lyocell fibers²⁴ (Fig. 1i).

Subsequently, we scrutinized the fluid dynamics within the microfluidic chip using a combination of computational fluid dynamics (CFD) simulations (Fig. 2a), in situ Raman imaging (Fig. 2b), and in situ laser confocal scanning (Supplementary Fig. 5). As depicted in Fig. 2a, the flow behavior of the cellulose solution within the microfluidic chip can be divided into three distinct stages: 1) prior to interaction with the good solvent in the central channel (Position 1, Fig. 2a), the modified cellulose solution exhibited relatively low flow velocities (Fig. 2a); 2) upon introduction of the good solvent (Position 2, Fig. 2a), the cellulose solution underwent dilution and shear alignment, resulting in a pronounced velocity gradient (Fig. 2a), where surface flow velocities were markedly lower than those in the interior. This phenomenon arises from the shear and disruption caused by the oblique inflow of the good solvent, which perturbs the surface velocity field of the cellulose solution, leading to deceleration of the surface network concurrent with acceleration of the inner flow; 3) finally, upon merging with the coagulating solvent (Position 3, Fig. 2a), the central solution experienced further shear acceleration (Fig. 2a) and subsequent gelation, yielding biaxially oriented gel fibers.

Fig. 2. Flow behavior of cellulose solutions and structural characterization of bioinspired fibers within microchannels.

a The computational fluid dynamics (CFD) simulations of the fluid dynamics within the microfluidic chip: (left) schematic diagram of the solution flow process in the chip, the green central stream of the modified cellulose solution, the blue sheath streams of good solvents (LiOH/urea/H₂O solution), and the red sheath stream of the coagulating agent (phytic acid solution); (middle) velocity field distribution of multi-phase solutions within the microfluidic chip; (right) the velocity distribution at three key locations of the spinning system. b In situ Raman monitoring during fiber spinning: (left) experimental setup for in situ Raman analysis; (middle) reconstructed 3D XZ plane Raman image accompanied by detecting C−H stretching intensities (2895 cm^-1); (right) a series of representative Raman spectra selected at equal intervals from the outside point (i) to the inside point (ii). c SEM images of regenerated cellulose fibers prepared under different conditions. d Derjaguin Müller Toporov (DMT) modulus distribution map of bioinspired cellulose fiber cross-section measured via atomic force microscopy, including modulus profiles from the outside point (i) to the inside point (ii). e 2D SAXS patterns, corresponding azimuthal integration profiles and calculated orientation factors for conventional wet-spun cellulose fiber and the bioinspired fiber.

To explore the formation mechanism of the biaxial orientation structure, we utilized Raman imaging to investigate the C–H stretching vibrations of cellulose chains, specifically monitoring the peak at 2895 cm⁻¹, within the region between the first and second sheath layers^25,26. The 3D Raman spatial imaging and positional Raman spectroscopy maps (Fig. 2b) revealed a pronounced contrast in packing density between the surface and the interior of the cellulose dope. This indicates that the shear-induced dilution effect exerted by the good solvent in the first sheath flow results in a relatively looser, less densely packed cellulose network on the surface of the pre-crosslinked cellulose dope, compared to the more compact interior region. Such differential packing density is critical for establishing the biaxial nano-orientation by enabling differential alignment and structural reorganization during subsequent coagulation and gelation steps.

Additionally, we conducted a detailed morphological characterization of gel fibers prepared from both low (3 wt.%) and high (5.5 wt.%) concentration cellulose solutions using two fabrication methods: traditional wet spinning and the bioinspired microfluidic spinning technique (Fig. 2c). Our findings revealed that within the commonly employed wet spinning system, variations in cellulose concentration primarily affect the packing density of the resulting gel fibers. However, regardless of concentration, the outer layers of wet-spun gel fibers consistently exhibit a well-defined uniaxial alignment along the fiber axis (Fig. 2c). In stark contrast, gel fibers produced via the microfluidic spinning system display distinct protruding nanostructures on their surfaces that are circumferentially aligned, a consequence of flow direction deviations induced by the sheath flows in the microchannels (Fig. 2c). Upon air-drying, these protruding features densely consolidate into circumferentially oriented nanostructures on the fiber surface (Fig. 2c), establishing an outer circumferential orientation that coexists with the inner axial orientation. This results in a unique biaxial-oriented architecture in the bioinspired fibers. Collectively, these findings highlight the critical importance of the complex, multi-directional flow dynamics within the microfluidic channels for inducing and stabilizing the biaxial nano-orientation structure in cellulose fibers.

Furthermore, atomic force microscopy Derjaguin–Müller–Toporov (AFM-DMT) modulus distribution mapping of the bioinspired cellulose fiber cross-section reveals a clear modulus difference between the inner and outer regions, induced by the spatial nanoarchitecture (Fig. 2d).  Specifically, the elastic modulus of the outer layer is notably lower than that of the central region. This pronounced modulus gradient can plausibly be attributed to dilution effects resulting from solvent perturbations acting on the outer layer during fiber formation. Such effects lead to a relatively loose network structure and reduced stiffness in the outer region compared to the more compact, highly aligned central core. Subsequent 2D synchrotron SAXS analyses were conducted on both traditional wet-spun and bioinspired fibers to probe their internal nanostructured orientation. The traditional wet-spun fibers exhibited diffraction patterns exclusively along the equatorial direction, consistent with a uniaxial (axial) orientation of nanostructures. In contrast, the bioinspired fibers showed distinct diffraction features along both equatorial and meridional directions, indicative of a biaxial orientation of elementary fibrils (Fig. 2e). The corresponding azimuthal integration curves further confirmed this nearly perpendicular dual-orientation pattern within the bioinspired fibers. Together, these experimental findings strongly validate that precise nano-orientation engineering within the microfluidic channels is critical for fabricating single cellulose fibers with robust biaxial orientation nanostructures. This unique hierarchical architecture holds promise for imparting exceptional mechanical properties and functional performance to the fibers, closely mimicking natural high-performance biological materials.

To elucidate the relationship between structural alignment and mechanical performance, we systematically evaluated and compared the mechanical properties of traditional wet-spun cellulose fibers with those of microfluidic-spun bioinspired fibers. The results demonstrated that the bioinspired fibers show substantial enhancements in mechanical behavior: specifically, the fracture strain increased to 41%, and the toughness improved to 184 MJ m^-3. Additionally, the fracture strength also rose from 473 MPa in the traditional fibers to 553 MPa in the bioinspired ones (Fig. 3a). Even without chemical pre-crosslinking, the mechanical performance of microfluidic-spun cellulose fibers is markedly enhanced compared to those fabricated via traditional wet spinning (Supplementary Fig. 6). These improved mechanical properties, particularly the remarkable increase in toughness, are primarily attributed to the unique biaxially oriented nanostructure engineered within the bioinspired fibers. The circumferential alignment of nanostructures on the fiber surface promotes more effective stress transfer during tensile deformation, facilitating pronounced energy dissipation through mechanisms such as crack propagation. Notably, the presence of “Z”-shaped crack deflections observed in Supplementary Fig. 7 acts to shield the internally aligned axial nanostructures by diverting crack paths, thereby enhancing the fiber’s damage tolerance. When benchmarked against a range of other one-dimensional oriented biomass fibers—including cellulose I fibers such as bamboo macrofibers¹⁹ and microfluidic-spun cellulose nanofibril (CNF) fibers¹⁵, cellulose II fibers like Lyocell²⁷, viscose fibers²³, and regenerated cellulose fibers¹⁴, as well as other biopolymers such as chitosan fibers²⁷—the bioinspired fibers distinctly outperformed in terms of both fracture strain and toughness (Fig. 3b, c). This compelling evidence confirms that nano-orientation engineering, particularly the introduction of a biaxial nanostructure, is a highly effective strategy for markedly enhancing the toughness of cellulose-based fibers, overcoming the typical brittleness associated with uniaxially oriented biomaterials.

Fig. 3. Study on the mechanical properties and deformation mechanisms of bioinspired fibers.

a Tensile stress-strain curves of bioinspired fibers and wet-spun traditional fibers, with schematics of crack propagation during fiber fracture (bottom-left: wet-spun traditional fibers; upper-right: bioinspired fibers). b Tensile stress-strain curves of bioinspired fibers and typical cellulose fibers (Bamboo macrofiber, Microfluidic-spun CNF fiber, Wet-spun regenerated fiber, Lyocell fiber, Viscose fiber)^{14,15,22–24}. c Comparison of the toughness and tensile strain of bioinspired fibers and other typical cellulose fibers, the bubbles represent the numerical ranges^{14,15,18,22–24,27}. d Molecular dynamic simulations of uniaxially and biaxially oriented cellulose molecular chain bundles, structural evolution snapshots of two types of molecular chain bundle under tension. The white and rainbow-colored chains represent the circumferential and axial cellulose chains, respectively. e Simulated stress-strain curves of two types of cellulose molecular chain bundles, with strain up to 160%. f Toughness comparison between bioinspired fibers and typical natural^{14,22,39–43} and synthetic fibers^{28–34,36–39}. BC fiber means Bacterial cellulose fiber, RC fiber means Regenerated cellulose fiber, TDCF means Twist de-acetylated cellulose fiber. Error bar represents the standard deviation of 5 independent experiments. g Radar plot comparing comprehensive performance (Toughness, Strength, Weavability, Scalability, and Cost) of four typical cellulose fibers^15,22,24.

Molecular dynamics (MD) simulations are performed to elucidate the role of nano-orientation on the mechanical properties of cellulose fibers, focusing specifically on tensile deformation behavior of simplified molecular models representing uniaxially and biaxially oriented cellulose chain bundles (Fig. 3d, Supplementary Fig. 8). For the uniaxially oriented model, failure predominantly occurred through extensive interchain sliding events along the fiber axis (Fig. 3d), which manifested as pronounced saw-tooth features on the simulated stress-strain curve. This model exhibited a fracture strain of approximately 120% and a toughness of about 588.6 MJ m^-3 (Fig. 3e, Supplementary Fig. 9). The saw-tooth pattern corresponds to sequential breaking and reforming of hydrogen bonds during interchain sliding, characteristic of brittle failure with limited extensibility. Conversely, the biaxially oriented model, which incorporates circumferentially aligned surface chains surrounding an axially oriented core, demonstrated markedly different deformation mechanisms. Upon tensile loading, the circumferential surface chains gradually reoriented toward the tensile axis (Fig. 3d). Because these surface chains are hydrogen-bonded and interact via Van der Waals forces with multiple spatially separated core chains, their reorientation not only facilitates improved load transfer to the core—thereby enhancing strength—but also induces abundant interchain sliding between the surface and core regions. This mechanism effectively delays catastrophic failure by dissipating energy through sliding-induced deformation, thereby increasing toughness.

Quantitative analysis from the simulations revealed that the biaxial architecture increased the fracture strain by 25% (reaching 150%), boosted fracture strength by 21% (reaching 1783 MPa), and enhanced toughness by 59% (up to 935.5 MJ m⁻³) compared to the uniaxial model (Fig. 3e, Supplementary Figs. 9, 10). These molecular insights strongly corroborate the experimental findings, providing mechanistic validation that biaxial nano-orientation substantially improves toughness by combining enhanced strength with increased extensibility.

Owing to this nano-orientation engineering, the bioinspired fibers exhibit highly competitive toughness relative to a broad spectrum of benchmark materials. This includes organic fibers (such as PET²⁸, PAN²⁹, Kevlar³⁰, Nylon³¹, Polyster³², and PBI³³), inorganic fibers (like Graphene³⁴, GO³⁵, Glass³⁶, Steel³⁷, Carbon³⁶, Basalt³⁸, CNTs³⁹), as well as natural and regenerated fibers (including Regenerated cellulose¹⁴, Bamboo²², Palm⁴⁰, Vakka⁴⁰, Coconut⁴⁰, Date⁴⁰, Bacterial cellulose⁴¹, Silk⁴², and Spider silk⁴³) (Fig. 3f). Impressively, advancing toward the mechanical properties of natural spider silk. Moreover, compared with conventional cellulose fibers such as Lyocell fiber²⁴, Bamboo microfiber²², and CNF fibers¹⁵ (Fig. 3g), the bioinspired fibers demonstrate significant advantages in terms of cost efficiency, processing time, manufacturability, as well as mechanical strength and toughness. This advancement substantially broadens the processing strategies and potential applications for high-performance cellulose-based fibers in sustainable materials science.

Textiles represent a major application field for fibers. Hence, we systematically investigated both the fabricability and the textile performance of the bioinspired fibers. As depicted in Fig. 4a, bioinspired fibers, characterized by their outstanding mechanical properties, were successfully woven on commercial textile machinery (Supplementary Movie 2), enabling the production of bioinspired fabric samples with dimensions around 15 cm ⨯ 90 cm. This achievement demonstrates the practical feasibility of scaling up these fibers for textile manufacturing. To evaluate the structural robustness of the bioinspired fabrics under environmental and mechanical stresses, we employed a CEAST drop-tower impact system to examine their impact resistance (Fig. 4b, Supplementary Fig. 11). Four typical commercial cellulose-based textiles—Cotton, Modal, Cuprammonium, and Viscose fabrics—were chosen as benchmarks. All samples were standardized to comparable thicknesses and subjected to identical testing conditions. During impact testing, we recorded the fracture time-force curves (Fig. 4b) and fracture energies (Fig. 4b). The results evidenced that the bioinspired fabric exhibited an extended impact duration of approximately 5.5 s, followed by a deformation phase lasting up to 12 s, without any noticeable permanent deformation or visible damage (Fig. 4b, c). In stark contrast, these commercial cellulose fabrics showed shorter impact durations (less than 3 s), rigid fracture patterns, and irreversible physical damage post-impact (Fig. 4b, c). To further probe the limits of the bioinspired fabric’s mechanical resilience, the drop height was incrementally increased to induce failure. Under these more severe conditions, the bioinspired fabric demonstrated a maximum failure force reaching up to 607 N, maintained an impact duration of about 5.5 s (Supplementary Fig. 12), and exhibited fracture energy as high as 1.75 J (Fig. 4b, Supplementary Fig. 13). These values substantially surpass those recorded for the commercial textiles. Collectively, these findings verify that bioinspired fibers engineered via nano-orientation techniques can be processed into fabrics exhibiting exceptional mechanical performance, combining ultrahigh toughness and high strength. This positions them as up-and-coming candidates for developing high-performance, durable textiles that outperform conventional cellulose-based materials.

Fig. 4. Fabrication and performance evaluation of bioinspired fabric.

a Fabrication of bioinspired fabric: photograph of the bioinspired fibric manufacturing process (left); photograph of the bioinspired fabric (15 cm × 90 cm) (right). b Impact resistance performance of bioinspired fabrics and conventional cellulose-based fabrics (Cotton fabric, Cuprammonium fabric, Viscose fabric, Modal fabric): schematic of impact testing for cellulose fabrics (left); time-force curves of fabrics during impact (middle); duration and fracture energy of fabrics during impact (right). The error bar represents the standard deviation of 3 independent experiments. c Photographs of fabrics after impact. d Schematic illustration of Molecular dynamics simulation of impact-induced mechanical response in bioinspired fabric, traditional fabric (up) and bioinspired fabric (down). e Simulated force-displacement curves of bioinspired fabric and traditional fabric. The curve showing larger force at the same displacement corresponds to bioinspired fabric. f Deformation trajectories (side view and cross-section view where different chains are in different color) of a single bundle from each model. The white and rainbow-colored chains represent the circumferential and axial cellulose chains, respectively. The structural integrity in bioinspired fabric explains the larger simulated forces.

We conducted MD simulations to further elucidate the mechanical behavior of bioinspired fabrics under indentation loading. Two fabric models were constructed for this purpose (Fig. 4d): one composed of uniaxially oriented cellulose molecular bundles representing traditional fabrics, and another composed of biaxially oriented bundles mimicking the bioinspired fabric structure. Indentation was simulated by imposing a constant velocity displacement via an indenter, as presented in the Methods section. The results (Fig. 4e) show that the bioinspired fabric model exhibits a higher resistance to indentation compared to the traditional fabric model, consistent with experimental observations where bioinspired fabrics sustain significantly greater failure stresses. Moreover, the area under the force-displacement curve—corresponding to the energy dissipation density—is substantially larger for the bioinspired model, in agreement with experimental data showing improved fracture energy of bioinspired fabrics.

Admittedly, MD simulations are subject to intrinsic length- and time-scale limitations. The simulation models contain a finite number of molecular chains and employ periodic boundary conditions, which may not fully capture the larger-scale structural heterogeneities and defects present in real fibers. Furthermore, the simulated deformation and indentation processes occur on nanometer length scales and sub-nanosecond time scales, which are orders of magnitude smaller and shorter than experimental conditions. Consequently, the absolute mechanical responses obtained from MD simulations cannot be directly equated to experimental values. Instead, these simulations are primarily intended to provide mechanistic insight and enable a comparative analysis between the traditional and bioinspired fibers within a consistent computational framework.

To interpret these findings mechanistically, cross-sectional snapshots were examined under identical indentation forces. The uniaxially oriented bundles readily underwent bending, dispersion, and localized structural damage (Fig. 4f, left), evidenced by pronounced deflections and shape distortions. In contrast, the biaxially oriented bundles maintained their shape and structural integrity (Fig. 4f, right). This behavior underscores the critical role of circumferentially oriented surface chains in reinforcing fiber architecture, effectively preserving fiber integrity, and enabling exceptional mechanical performance under indentation. Together, the combined experimental and simulation results confirm that the bioinspired dual-oriented structural design successfully preserves the enhanced mechanical properties of individual fibers when assembled into fabrics. This significantly broadens the application prospects of cellulose-based textiles, particularly in demanding environments requiring high mechanical robustness. Furthermore, the fabric exhibits excellent washing stability, even after repeated rigorous washing cycles, it successfully maintained its overall structural integrity without any signs of loosening (Supplementary Fig. 14). Beyond cellulose, this microfluidic-enabled nano-orientation strategy serves as a highly versatile platform, readily extendable to other biopolymers such as chitin and chitosan (Supplementary Figs. 15, 16). This indicates a generalizable route for fabricating high-performance, tough, and strong biofabric materials from diverse polysaccharide sources.

In summary, we have developed a bioinspired nano-orientation strategy enabling the scalable fabrication of regenerated cellulose fibers with a biaxially aligned structure through a microfluidic spinning system. Both experimental characterizations and molecular dynamics simulations demonstrate that this biaxial architecture effectively redistributes and disperses stress during deformation. Consequently, the fibers retain a high tensile strength of 553 MPa while significantly enhancing extensibility to 41%, culminating in an exceptional toughness of 184 MJ m^-3. In the context of current literature, this value compares favorably against various traditional cellulose and synthetic fibers, demonstrating a mechanical profile reminiscent of natural spider silk. Furthermore, these outstanding mechanical properties are successfully translated from individual fibers to fabrics fabricated from the bioinspired fibers, which exhibit improved mechanical robustness compared to conventional commercial cellulose-based textiles. Our work provides fundamental insights into the role of nano-orientation in fiber mechanics and paves the way for sustainable, high-performance cellulose fiber applications across various sectors, including textiles, automotive, and aerospace industries, thereby advancing the development of environmentally friendly and mechanically resilient materials. Although our current lab-scale focus on complex radial structural control entails a throughput trade-off relative to mature Lyocell processes, we envision bridging this gap for industrial-scale demands through two feasible pathways: (i) implementing multi-channel spinning arrays, and (ii) optimizing dope rheology for higher-velocity extrusion while preserving delicate flow dynamics.

Methods

Materials

Cellulose cotton linter pulp (α-cellulose content higher than 95%) was provided by Hubei Chemical Fiber Group Co., Ltd (Xiangyang, China). LiOH, urea, and Na₂SO₄ were purchased from XiLong Science Co., Ltd (Guangzhou, China). Phytic acid solution was purchased from Aladdin Chemical Co., Ltd (Shanghai, China).

Chemically pre-crosslinked cellulose solution preparation

A total of 5.5 wt.% and 3 wt.% cellulose solutions were prepared by dissolving cotton linter pulp (5.82 g and 3.09 g) in 100 g of pre-cooled LiOH (8 wt.%)/urea (15 wt.%)/H₂O (77 wt.%) separately. Epichlorohydrin (1 mL), the chemical crosslinker, was then added dropwise to the cellulose solution at −5 °C and stirred for 2 h. After centrifugation to remove air bubbles, the pre-crosslinked cellulose solutions were obtained and employed as the core flow in the microfluidic spinning system. The proposed protocol and the reagents used pose no safety risks. The process incurs negligible mass loss.

Bioinspired fiber’s fabrication

We used a specialized three-phase, five-channel microfluidic chip with a core flow of pre-crosslinked cellulose solution, a first sheath flow of good solution- LiOH (8 wt.%)/urea (15 wt.%)/H₂O (77 wt.%), and a second sheath flow of coagulation solution, including phytic acid (5 wt.%)/Na₂SO₄ (5 wt.%) solution. Flow rates were precisely controlled at 150, 300, and 400 μL/min using a syringe pump. Hydrogel fibers were formed within the chip, then wound onto a metal plate rotating at 20 rpm. After thorough washing with deionized water to remove residual solvent and coagulant, bioinspired fibers were obtained through air drying. The proposed protocol and the reagents used pose no safety risks. The process incurs negligible mass loss.

Characterization

SAXS patterns were performed using the Genix 3D X beamline with wavelength λ = 1.54 Å at Xeuss 2.0. The sample-to-detector distances were 2497.7 mm. The 1D SAXS profiles were acquired by circularly averaging the intensity using the built-in software. In situ Raman imaging characterization was carried out using a confocal Raman microscope (inVia, Renishaw, UK) with a scanning range of 135 × 30 × 150 μm³, a spatial resolution of 400 nm, and an excitation wavelength of 785 nm. The 2D imaging of bioinspired fiber was achieved through the application of the multivariate curve resolution method provided by additional software. Scanning electron microscope (SEM) images of freeze-dried and air-dried fibers, coated with a layer of gold vapor, were captured using a field emission scanning electron microscope (SIGMA 500, Zeiss, Germany) with an accelerating voltage of 5 kV. The mechanical properties of bioinspired fibers and traditional fibers were characterized utilizing a nanomechanical actuating transducer (NMAT) on a Universal Testing Machine (Nano UTM T150, KLA-Tencor, USA) with a strain rate of 0.0027 s^-1. All measurements were performed under controlled conditions (25 °C and 50 ± 5% relative humidity). Engineering stress was calculated using the fiber cross-sectional area, derived from SEM-measured diameters assuming a circular cross-section. Results are reported as mean ± standard deviation (n = 5 ~ 10 independent specimens). Detailed mechanical testing parameters for all fibers are provided in Supplementary Table 1. AFM elastic modulus mapping of the bioinspired fiber was performed by scanning adhesion curves in an area of 15 × 7.5 μm² with a silicon nitride probe (RTESP-300, BRUKER) at a resolution of 512 × 512 pixels and fitting the Johnson–Kendall–Robert model. The impact testing of bioinspired fabric and other cellulose fabrics was carried out using an Instron impact testing machine (CEAST droptower impact system, CEAST 9310). The impact energy was controlled by adjusting the drop height (40 ~ 60 cm) and the weight of the hammer (0.8 ~ 1 kg). The impact energy was determined using the formula: E = 1/2 mV², where E represents the energy, m is the mass, and V is the velocity. Both force and energy curves were automatically recorded by sensors on the hammer head during the testing process. Details regarding the calculations of the crosslinking density and the orientation factor (f) can be found in the “Supplementary Methods” section of the Supplementary Information.

Computational fluid dynamics simulation

We employed computational fluid dynamics (CFD) simulations to verify the fluid transport within microfluidic chips, including momentum transfer due to velocity differences. To investigate the spontaneous transport of physical quantities (such as velocity, pressure, mixing, and enrichment between cellulose solution and solvent) from high-intensity to low-intensity regions during the spinning process, we initially divided the spinning channel into nearly 5 million units with a diameter of 20 nm. Subsequently, we utilized COMSOL’s laminar flow field simulation software, employing a multiphase mixing model to analyze the solution flow throughout the entire channel. A fixed wall boundary condition was applied at the interface between the flow and the chip wall, neglecting free-molecular motion on the solid surface. The flow velocity and Reynolds number within the channel were generally low, indicating predominantly laminar flow.

Molecular dynamics simulation

The fiber model of the traditional fiber contains 36 cellulose molecular chains (uniformly oriented), and the simulation box is periodic along the length direction. These 36 cellulose molecular chains are arranged into a 6 × 6 array⁴⁴. Each individual cellulose molecular chain in the periodic simulation box consists of 14 monomers (i.e., C₁₂H₂₀O₁₀). Then, for each chain, we randomly choose two neighboring monomers and convert them to chain ends (delete the covalent bond between them and terminate unsaturated atoms with H atoms), so that the molecular chains in this periodic model system have finite length. The fiber model of bioinspired fiber is constructed by placing the model of the traditional fiber into the double helix structure formed by two cellulose molecular chains of finite length (each consisting of 90 monomers). Molecular dynamics simulations are performed using LAMMPS⁴⁵_, and the ReaxFF potential is employed⁴⁶. The model is equilibrated by an NPT ensemble simulation at 300 K and with a pressure of 0 atm along the length direction. After the equilibrated structure is obtained, the uniaxial tension along the fibril direction is applied to the simulation box by a constant strain rate of 10^-6fs^-1 subjected to canonical ensemble (NVT) at 300 K using a timestep of 0.25 fs. The stress component along the loading directions is computed and averaged over every 2000 timesteps. The hydrogen bond cutoff is 7.5 Å. Visualization is performed using OVITO⁴⁷. The fabric models are constructed by two identical fibrils (either model of the traditional fiber and bioinspired fiber) that are in contact and are oriented orthogonally. For computational efficiency, each traditional fiber contains 9 cellulose molecular chains arranged in a 3 × 3 array. This setting remains the same for the inner core of bioinspired fiber. During the simulation, the four ends of the two bundles are fixed. The model is equilibrated by an NVT ensemble simulation at 300 K. After the equilibrated structure is obtained, a spherical indenter with a radius of 50 Å is employed to exert a force on the center of the model at a velocity of 10^-4Å fs^-1. The forces acting on the indenter along the pushing directions are computed and averaged over every 1000 timesteps. Other settings are the same as the fiber model.

Author contributions

D.Y. conceived the concept. D.Y. and S.Z. provided guidance and support throughout the project. X.F. took lead roles in performing the experiments, analyzing data, and drafting the manuscript. T.Y. and Y.Z. contributed to mechanical experimental characterization. C.J. contributed to the characterization of fiber structure. X.Z. and S.Z. contributed to the molecular dynamics analysis. All authors contributed intellectually to the study design and approach, discussed the results, and reviewed and edited the final manuscript.

Peer review

Peer review information

Nature Communications thanks Atif Rasheed and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. [A peer review file is available].

Funding

D.Y. discloses support for the research of this work from the National Natural Science Foundation of China [grant number 52473090], the Excellent Youth Scientists Fund from the Natural Science Foundation of Anhui Provincial Project [grant number 2408085Y025]. S.Z. discloses support for publication of this work from the National Natural Science Foundation of China [grant number 12002304], the Distinguished Young Scientists Fund from the Natural Science Foundation of Zhejiang Province [grant number LR23A020001] and the “Pioneer” R&D Program of Zhejiang [grant number 2023C03007]. X.F. discloses support for the research of this work from the National Natural Science Foundation of China [grant number 52503126], the National Natural Science Foundation of Anhui Province [grant number 2508085QE172]. X.Z., T.Y., Y.Z., and C.J. declare no relevant funding.

Data availability

The data supporting the findings of this study are available within the article and its Supplementary Information files and are available from the corresponding authors upon request. All data generated in this study are provided in the Supplementary Information/Source Data file. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-74052-6.