A versatile and simple rectangular-topology strategy enables robust hydrogels

Abstract

High‑strength, tough hydrogels are urgently needed for biomedicine and soft robotics, yet single‑network architectures suffer from an inherent trade‑off between mechanical robustness and high water content. Herein, we propose a rectangular topological network (RN) composed of rectangular motifs in a single-network hydrogel via stoichiometric crosslinking of long and short polymer chains. This architecture achieves a high elastic modulus via dense intermolecular hydrogen bonds while maintaining large mesh sizes. Driven by a strain-induced hardening mechanism, the hydrogel achieves a simultaneous increase in strength and toughness. Successful fabrication using two crosslinkers demonstrates broad applicability. A representative PEG‑based RN hydrogel exhibits a Young’s modulus of 21.97 MPa, a toughness of 31.04 MJ·m⁻³, and a tensile strength of 7.72 MPa. Thus, we show that this facile and versatile RN strategy offers a route for high‑energy dissipation and can be extended to other platforms.

Tough hydrogels have potential in a range of applications, but balancing mechanical robustness and high water content is challenging. Here, the authors report the development a rectangular topological network for high mechanical robustness in a single network hydrogel.

Introduction

The pursuit of high strength and toughness is a paramount goal in materials science¹. However, these two properties are often mutually exclusive due to the inherent limitations of a single structural motif². Hydrogels are three-dimensional hydrophilic polymer networks that retain large amounts of water. Their wide applications range from environmental engineering and agroforestry to daily chemicals and emerging fields such as controlled drug delivery and, particularly, flexible bioelectronic human-machine interfaces^3–9. The fundamental requirement of high-water absorbency necessitates a long-chain network, i.e., a low crosslinking density within the hydrogel network. This specific network architecture, unfortunately, compromises its load-bearing capacity and energy dissipation efficiency, leading to an intrinsically soft and weak mechanical character^10,11. Consequently, hydrogels typically exhibit a strength <30 kPa and a toughness <1 kJ·m⁻³, which are substantially inferior to the mechanical properties of many load-bearing biological tissues, such as cartilage and tendons (strength > 5 MPa, toughness > 10 MJ·m⁻³)^12–14.

To address the challenge of simultaneously achieving high strength and high toughness within a single hydrogel structural motif, prevailing strategies involve: i. integration of structural motif with distinct roles in strength and toughness, and ii. incorporating high-energy-dissipation mechanisms between two structural motifs^15,16. Regarding the integration of strength‑conferring and toughness‑conferring structural motifs, double-network (DN) hydrogels represent a seminal advancement by ingeniously combining a rigid/brittle first network with a flexible/ductile second network^17–22. Gong et al. developed a seminal DN hydrogel consisting of a brittle, strong poly(2-acrylamido-2-methylpropanesulfonic acid) network (sacrificing at a low stress) and a ductile, tough polyacrylamide (PAAm) network (sustaining stress at large extension), as shown in Fig. 1a, yielding a hydrogel with a fracture strength of >1.2 MPa and a toughness of 10 MJ·m^{−3 23,24}. The DN hydrogel developed by Gong and colleagues fundamentally changed the conventional perception that hydrogels are inherently fragile. Nanocomposite hydrogels are formed by incorporating nanofillers (e.g., clay, silica) as a rigid structural motif into a tough hydrophilic polymer network, thereby creating a composite of rigid and dissipative motifs. The dispersion stability of the nanoparticles, their biocompatibility, and the potential compromise to the material’s optical transparency represent critical trade-offs that must be considered in their practical application^25–27.

Fig. 1. Rectangular topological network (RN) hydrogel: structural design, deformation mechanisms, and mechanical performance.

a Schematic illustrations of network deformation and fracture mechanisms during stretching for RN, hydrogen-bonded single-network (HN), and double-network (DN) hydrogels. Schematic of the motif structure of RN hydrogels in the b dried and c hydrated states, and the corresponding multiple hydrogen-bonding interchain interactions. d Optical image of the RN hydrogel. e SEM micrograph of the cryo-fractured surface of a dried RN hydrogel. Comparison of mechanical properties of f tensile strength (MPa) and toughness (MJ·m⁻³) and g tensile strength (MPa) and tear energy (kJ·m⁻²) for hydrogels with different reinforcement mechanisms from reported studies^{28–31,36,46,52–58}.

The second strategy, which introduces high-energy-dissipation mechanisms between two structural motifs, primarily encompasses a range of mechanisms such as slide-ring structures, homogeneous tetra-arm polymer networks, topological entanglements, crystal reinforcement, and multiple hydrogen bonding^28–42. Suo et al built upon a highly entangled network by introducing interchain cross-links with short-chain molecules that bridge the primary long polymer chains; by modulating the ratio of these cross-linkers, they produced a hydrogel whose strength at 500% strain increase from 30 to 390 kPa, with a toughness of 0.68 MJ·m^{−3 31}. The aforementioned strategies have significantly enhanced the strength and toughness of hydrogel materials. Resolving the long-standing conflict between achieving both high strength and high toughness within a single hydrogel structural motif remains a central goal that the field strives to achieve.

We propose a rectangular topology strategy to achieve high strength and toughness in a single hydrogel network, a design concept that can be extended to broader material systems and applications. As illustrated in Fig. 1b, c, this RN motif, which is formed by crosslinking long and short polymer chains in equimolar amounts, can be regarded as a repeating structural unit. First, RN motif achieves a high elastic modulus under low crosslinking density by leveraging strong intermolecular hydrogen bonding, while maintaining large network mesh size, high water content, and transparency (shown in Fig. 1d, e); second, RN motif significantly enhances both fracture strain and fracture strength through strain-induced topological rearrangement and hydrogen bond reorganization within the structural motif (shown in Fig. 1a). This design is different from the well-known H-bond mechanism. Detailed discussion can be found in Supplementary Fig. 1. We designed two distinct RN hydrogels using poly (ethylene glycol) (PEG), a widely used hydrogel material, demonstrating its simplicity and versatility. After applying the RN strategy, the fracture strength of hydrogel increased nearly 20 times from 0.35 to 7.72 MPa, while 50 times of improvement in toughness from 0.62 to 31.04 MJ·m⁻³ at 500% strain. The mechanical performance of our hydrogels and the corresponding literature values are presented in Fig. 1f. The tear energy of this hydrogel, together with literature data, is shown in Fig. 1g. A more comprehensive literature survey is provided in Supplementary Table 1. The RN design exhibits substantial energy dissipation via reversible hydrogen bond reorganization guided by its rectangular topology, offering a topological design framework that complements the conventional chemical composite approach. This toughening mechanism is governed by topology rather than solely by conventional hydrogen-bonding interactions. The locking effect—resulting from the precise covalent crosslinking of long and short chains—creates a confined rectangular framework in which hydrogen bonds form internally, fundamentally distinguishing this design from conventional PEG crystallization. This work not only provides guiding principles for designing high-performance hydrogel materials but also shows promising potential for extending this strategy to other polymer material systems for enhanced mechanical properties.

Results and discussion

To illustrate the potential for extending our strategy to broader systems, two RN hydrogel systems were prepared. Using Cardol as the starting material, we designed and synthesized a crosslinker cardanol-derived polyols (OH-GE-Cardol) with an average –OH functionality of 8 (see Supplementary Fig. 2). The crosslinker OH‑GE‑Cardol was subsequently crosslinked with a Pre‑PU (pre‑reacted from PEG and the crosslinker IPDI) to form a Cardol‑based hydrogel, designated as System 1. A tetrafunctional crosslinker bis(2,3-dihydroxypropyl) disulfide (BDPD)⁴³ was also prepared and employed the same methodology to fabricate a BDPD-based hydrogel series, designated as System 2 (see Supplementary Fig. 3). The key distinction between the two crosslinkers lies in their molecular architecture: the Cardol-based derivative possesses both higher functionality and the inherent rigidity of a benzene ring. The synthetic routes and molecular structures are depicted in Fig. 2a.

Fig. 2. Synthesis and structural motifs of RN hydrogels using two distinct crosslinkers.

a Molecular structure schematics for the synthesis of PEG-OH-GE-Cardol and PEG-BDPD hydrogels. Synthesis and structural motif of b PEG_n = 9/90/OH-GE-Cardol and c PEG_n = 9/OH-GE-Cardol.

The PEG_n = 9/90/OH-GE-Cardol hydrogel (System 1) was fabricated as schematically illustrated in Fig. 2b. Initially, PEG_n = 9, PEG_n = 90, and the chain extender IPDI were pre-reacted in a molar ratio of N (PEG_n = 9): N (PEG_n = 90): N(IPDI) = 1:1:4, yielding an –N = C = O– terminated prepolymer (Pre-PU, Fig. 2b_ii). Subsequently, an appropriate amount of N, N-dimethylacetamide (DMAC) was introduced to dilute the system. The solvent content was determined based on Flory-Huggins solution theory and rubber elasticity theory to maintain the polymer chains in a sufficiently extended conformation during crosslinking (see Supplementary Fig. 4 for details). The crosslinker OH-GE-Cardol was then added in an equimolar amount relative to PEG_n = 9. Owing to the equimolar ratio of long and short chains, the system forms topological motifs with a rectangular structural motif during curing (Fig. 2b_iii). Solvent exchange ultimately yields the hydrogel, which features abundant hydrogen bonds in the form of PEG-H₂O-PEG within its rectangular architecture (Fig. 2b_iv). To elucidate the critical role of the long-and-short-chain combination, a control hydrogel containing only a single chain length was prepared for System 1 (Fig. 2c), designated as PEG_n = 9/OH-GE-Cardol. The total molar amount of PEG, as well as the types and quantities of all other materials and experimental procedures, remained consistent. With only one chain length and a molar ratio of N (PEG_n = 9): N (IPDI) = 2:1, tetrahedral structural motifs form during curing (Fig. 2c_iii), ultimately resulting in a topological network with tetrahedral geometry after further curing and solvent exchange (Fig. 2c_iv). The alternating long and short chains create larger network meshes, which is expected to reduce the effective crosslinking density. This structural feature contributes to the observed enhancement in fracture strain. Concurrently, the rectangular architecture facilitates the formation of extensive intermolecular hydrogen bonds between adjacent hydrated long chains in the presence of water molecules (PEG-H₂O-PEG hydrogen bonds), significantly strengthening the material. In contrast, the uniform chain length in the PEG_n = 9/OH-GE-Cardol hydrogel leads to high crosslinking density and greater average distance between chains. This structural configuration impedes the formation of effective interchain hydrogen bonding upon hydration, limiting the overall strength enhancement.

Following the same approach as for System 1, System 2 (BDPD‑based) was also synthesized to comprise both an experimental RN hydrogel and a control ref. hydrogel. Specifically, while maintaining the same PEG molecular weights, stoichiometric ratios, and synthesis methodology, the crosslinker was replaced with BDPD, a tetra-hydroxy-functional crosslinker. The experimental group, hydrogels synthesized using both PEG_n = 9 and PEG_n = 90, is designated as PEG_n = 9/90/BDPD. The corresponding control group, prepared using only PEG_n = 9, is designated as PEG_n = 9/BDPD (More structural details in Supplementary Fig. 5).

The reactants, intermediates, and resulting hydrogel in System 1 were characterized by FT-IR. As shown in Fig. 3a, Pre-PU exhibits a distinct characteristic peak at 2263 cm⁻¹, corresponding to the asymmetric stretching vibration of the isocyanate group (–NCO). This peak completely disappears in the spectrum of PEG_n = 9/90/OH-GE-Cardol, indicating the complete consumption of the –NCO-terminated prepolymer. The peak at 1707 cm⁻¹ is attributed to the C = O stretching vibration of the urethane group, while the broad absorption band in the range of 3428–3212 cm⁻¹ corresponds to the N–H stretching vibration of secondary amines⁴⁴. Furthermore, the aromatic C = C skeletal vibration peak, observed at 1592 cm⁻¹, shifts in the spectrum of PEG_n = 9/90/OH-GE-Cardol, indicating a change in the chemical microenvironment of the benzene rings due to their successful covalent incorporation into the polymer network. The NMR spectra further corroborated the synthesis of these materials (see Supplementary Figs. 6 and 7). The sol-gel transition of the hydrogel was visually confirmed by the vial-inversion test, and its transparency was observed before and after gelation, as depicted in Fig. 3c.

Fig. 3. Structural characterization, mechanical properties, and model parameters of RN hydrogels and controls.

a FT-IR spectra of OH-GE-Cardol crosslinker, Pre-PU intermediate, PEG_n = 9/90/OH-GE-Cardol, PEG_n = 9/OH-GE-Cardol, and PEG_n = 90. b XRD patterns of PEG_n = 9/90/OH-GE-Cardol hydrogel, PEG_n = 9/OH-GE-Cardol hydrogel, and PEG_n = 90. c Sol-Gel transition of the hydrogel. d Tensile stress-strain curves of PEG_n = 9/90/OH-GE-Cardol and PEG_n = 9/OH-GE-Cardol hydrogels with corresponding fittings based on the H‑bonds Arruda-Boyce model (HAB)/the H-bonds strain‑induced ordering model (HSI) model. e Schematic of the crosslinking node structure in PEG_n = 9/90/OH-GE-Cardol hydrogel. f Schematic of the crosslinking node structure in PEG_n = 9/90/BDPD hydrogel. g Comparison of fitted parameters (H-bond density parameter, Shear modulus, H-bond contribution, Hardening modulus, Max-ordering degree, H-bond reorganization) for the PEG_n=9/90/OH-GE-Cardol and PEG_n = 9/OH-GE-Cardol hydrogels. h Tensile and i compressive stress-strain curves of hydrogels with fixed OH-GE-Cardol crosslinker and varying long-to-short chain ratios (1:1, 2:1, and 1:2). j AFM characterization of the System 1 RN hydrogel, including the phase image, the height image, and a magnified view of the rectangular structures within it.

To confirm the presence of intermolecular hydrogen bonds between adjacent hydrated long chains in the presence of water molecules (PEG-H₂O-PEG hydrogen bonds), the PEG_n = 9/90/OH-GE-Cardol and PEG_n = 9/OH-GE-Cardol samples were further analyzed via X-ray diffraction (XRD) patterns (Fig. 3b). While pure PEG is semi-crystalline and exhibits sharp crystalline diffraction peaks, both samples display broad, amorphous halos, indicating that PEG has been incorporated into the crosslinked network. Notably, compared with PEG_n = 9/OH-GE-Cardol, the PEG_n = 9/90/OH-GE-Cardol sample shows slightly sharper diffraction peaks at 19.30° and 5.30°. The signal at 19.30° corresponds to the (120) plane of the orthorhombic crystal system of PEG, representing a typical characteristic of PEG crystalline structure. The diffraction peak at 5.30° is associated with a long-period ordered structure of PEG chains. The corresponding d-spacing is approximately 15.5 Å, significantly larger than the motif cell parameters of PEG (~5 Å), a feature that is likely attributed to the presence of water molecules within the crystal lattice. The emergence of these two diffraction peaks provides evidence that the rectangular topology enables adjacent PEG chains to bridge via hydrogen bonds through a few molecular layers of confined water. This interaction gives rise to larger-scale, ice-like microstructures (Fig. 2b), thereby significantly enhancing the hydrogel’s strength⁴⁵. The FTIR and XRD data for the materials in System 2 are provided in Supplementary Fig. 8.

To demonstrate the mechanical enhancement imparted by the RN structure, the tensile mechanical properties of the as-prepared hydrogels were analyzed. As shown in Fig. 3d, the stress-strain curve of the hydrogel with a rectangular topology (System 1) exhibits an extremely high initial modulus, with stress increasing rapidly in the initial stage. Subsequently, the stress continues to rise with increasing elongation until fracture. In contrast, the tetrahedral network hydrogel with uniform chain length (the control in System 1) displays a much more gradual deformation response throughout the entire stretching process. The rectangular topology hydrogel demonstrates a 45-fold higher tensile strength, a 52-fold greater toughness, and a 100-fold increase in Young’s modulus compared to its tetrahedral counterpart. The stress-strain curve of the PEG_n = 9/90/BDPD hydrogel (System 2) exhibits a very high initial modulus with rapid stress increase, followed by a more gradual progression until fracture. In contrast, the entire tensile process of PEG_n = 9/BDPD remains relatively smooth and gradual. The fracture strength of the former is 11 times that of the latter, confirming the versatility of our RN strategy. In addition, using the same strategy, we prepared hydrogels from a non‑PEG polymer (Polyvinyl Alcohol), which also showed comparable performance differences (See Supplementary Fig. 9). Notably, the RN structures constructed in System 1 and System 2 show distinct differences in mechanical properties such as toughness, tensile strength, Young’s modulus, and tear energy. (See Supplementary Table 2) These differences can be attributed to the structural characteristics of the crosslinkers. Compared to the BDPD crosslinker, the Cardol-based OH-GE-Cardol crosslinker provides more hydrogen bonding sites, while its inherent benzene rings contribute greater molecular rigidity, as schematically illustrated in Fig. 3e, f. The Cardol-derived crosslinker OH-GE-Cardol contains an average of eight hydroxyl groups, surpassing the tetrafunctional BDPD crosslinker (four hydroxyl groups), enabling each OH-GE-Cardol node to form more intermolecular hydrogen bonds with the hydrated PEG chains (PEG-H₂O-PEG) during network formation. Additionally, the intrinsic benzene ring in OH-GE-Cardol imparts greater molecular rigidity compared to the flexible aliphatic backbone of BDPD, due to a higher rotational energy barrier, providing pronounced local stiffness at the network nodes.

To elucidate the reinforcement mechanism of the RN structure, a tensile model was developed to analyze the stress-strain behavior of the hydrogels. All strains reported in this work are engineering strains, calculated as $\epsilon =\left(L-L_0\right)/L_0$. This model accounts for the deformations of both the polymer network and the hydrogen bonds. The entire tensile process was divided into two distinct regimes: elastic deformation (segment AB in the corresponding stress-strain curve) and irreversible network rearrangement (segment BC). Within the elastic deformation region, where initial response is dominated by elastic deformation of the molecular network and H-bonds, the stress-strain curves were fitted based on the Arruda-Boyce model⁴⁶ incorporating a parameter ${\rho }_H$, which characterizes the hydrogen bond density⁴⁷. On this basis, we established a modified version, termed the H‑bonds Arruda-Boyce model (HAB), to better capture the contribution of hydrogen‑bonding interactions to the overall mechanical response. The corresponding equation is given as follows:

$${\sigma }_{\mathrm{AB}}={\mu }_0\left(1+{\rho }_{\mathrm{H}}\right)\left(\lambda -{\lambda }^{-2}\right){\sum }_{i=1}^5{c}_i{\overline{\mathrm{I}}}_1^{i-1}$$ 1

where ${\sigma }_{\mathrm{AB}}$ is the stress, ${\mu }_0$ represents the initial shear modulus of the polymer network, ${\rho }_{\mathrm{H}}$ denotes the hydrogen bond density parameter, $\lambda$ is the stretch ratio, ${\overline{\mathrm{I}}}_1={\lambda }^2+2/\lambda$, and $c_i$ are the Arruda-Boyce coefficients.

As the strain increases, the network structure undergoes irreversible network rearrangement. The rectangular network architecture facilitates an ordered rearrangement of polymer chains and the reconstruction of intermolecular hydrogen bonds. To capture this behavior, we introduced H-bonds effects into the strain‑induced ordering model⁴⁸, yielding the H-bonds strain‑induced ordering model (HSI) as described by the following equation:

$${\sigma }_{\mathrm{BC}}={\mathrm{\sigma }}_{\mathrm{y}}+E_{\mathrm{h}}\left(\epsilon -{\epsilon }_0\right)+k_{\mathrm{H}-\mathrm{reorg}}\left[1-{\mathrm{e}}^{-\beta \left(\epsilon -{\epsilon }_0\right)}\right]$$ 2

Where ${\sigma }_{\mathrm{BC}}$ is the stress, ${\sigma }_{\mathrm{y}}$ represents the equivalent yield stress at 60% strain, $E_{\mathrm{h}}$ denotes the hardening modulus, $\epsilon$ is the total strain, ${\epsilon }_0$ is the yield strain, $k_{\mathrm{H}-\mathrm{reorg}}$ corresponds to the maximum contribution from hydrogen bond reorganization, and $\beta$ describes the saturation rate for hydrogen bond reorganization.

Our piecewise constitutive model accurately captures the distinct mechanisms governing the AB and BC regions (R² ≥ 0.97), focusing on the dominant physics in each segment. While a unified model that fully couples elasticity, hardening, and rearrangement in the transition remains a future challenge, the present work clarifies the core mechanisms underlying the high strength and toughness of RN hydrogels. Further experimental and theoretical efforts are needed to comprehensively resolve the transition region. The model provides an accurate description of the experimental data (the coefficient of determination, R² = 0.97), validating the proposed reinforcement mechanism for the high strength and toughness of the RN hydrogels (See Supplementary Code 1 for the code; see Supplementary Figs. 10–13 for fitting details). A comparison of the fitted parameters for the RN hydrogels in both systems is shown in Fig. 3g. These molecular features are quantitatively captured by two key parameters in our model. The hydrogen bond density parameter (${\rho }_{\mathrm{H}}$), which quantifies the contribution of hydrogen bonding to the initial modulus, is 1.7 times higher for the OH-GE-Cardol-based RN hydrogel than for the BDPD-based counterpart, directly reflecting the higher density of potential hydrogen-bonding sites from the additional hydroxyl groups. Furthermore, the structural ordering saturation rate parameter ($\beta$), which describes how rapidly the hydrogen bond network reaches an ordered, saturated state under strain, is significantly higher for the OH-GE-Cardol system. This can be rationally attributed to the rigidity conferred by its benzene rings, which helps preserve the RN framework during deformation, thereby facilitating faster and more directional cooperative rearrangement of adjacent chain segments and hydrogen bonds, manifesting as a more pronounced strain-hardening response. Based on the above topological architecture, a clear distinction can be made between topological rearrangement and ordinary molecular chain conformational rearrangement. The molecular chain conformational rearrangement is typically confined to local-scale chain orientation or entanglement disentanglement, whereas the topological rearrangement specifically refers to cooperative network structural adjustments that span multiple chain segments, guided by the rectangular topological framework. This is manifested as the cooperative displacement of network junctions, the directional dissociation and reorganization of the hydrogen-bond network, and the resulting characteristic two-stage mechanical response—an initial high-modulus elastic regime (AB segment) followed by a subsequent strain-hardening regime (BC segment). The hydrogen bond density in the former is 1.7 times that of the latter, accompanied by a significantly higher degree of structural ordering. The fitting results align well with the molecular structural design, collectively affirming the validity of the reinforcement mechanism and underscoring the effectiveness of the Cardol-based crosslinker.

To investigate the influence of rectangular topology on material properties, we systematically varied the dimensions of the structural motif by adjusting the ratio between long and short polymer chains, and evaluated the corresponding mechanical performance of the hydrogels. As shown in Fig. 3h, i, using OH-GE-Cardol as the crosslinker with fixed short (PEG_n = 9) and long (PEG_n = 90) chains, three mixing ratios (1:1, 1:2, and 2:1) were prepared and tested. Tensile tests revealed that the 1:1 sample delivered the best overall performance, whereas the 2:1 formulation exhibited greater load-bearing capacity under compression. This difference can be structurally interpreted: a moderately higher proportion of short chains promotes the formation of a denser crosslinked network, thereby enhancing compressive resistance, while an excessively high short-chain content may restrict chain extension and compromise tensile properties. These results directly demonstrate that the formation of the rectangular topology relies on the precise coexistence of long and short chains, which serves as the most fundamental design and functional evidence of the locking effect. To further optimize performance, the effect of long-chain degree of polymerization was examined. It was found that an order-of-magnitude difference in chain length between the short and long chains provided an optimal balance between toughness and strength (see Supplementary Fig. 14). Together, these findings reveal a design principle for such RN hydrogels: their formation critically depends on both precise dimensional matching and optimized proportional blending. Specifically, an approximate tenfold difference in chain length provides the necessary geometric contrast for rectangular motif formation, while specific mass ratios (e.g., 1:1) facilitate effective and ordered assembly of these motifs. This precise molecular-level design is essential for transitioning from a disordered crosslinked network into a long-range ordered topological architecture, and clarifies why conventional, broadly defined formulations fail to achieve comparable mechanical excellence.

To further demonstrate the existence of the rectangular structural motifs in the RN hydrogel, we performed nanoscale Atomic force microscopy (AFM) topography and mechanical mapping on the dried hydrogel, with the results shown in Fig. 3j. Regular rectangular microdomains with a width of approximately 20–30 nm and a length of approximately 70–80 nm can be clearly observed, matching the theoretical dimensions of the alternating long‑short chain rectangular unit. The phase image reveals that these microdomains exhibit higher stiffness than the surrounding amorphous regions, as indicated by their brighter contrast. A control AFM image of annealed pure PEG (Supplementary Fig. 15) shows typical spherulitic (spherical) crystals, whereas the RN hydrogel exhibits only rectangular microdomains, ruling out the possibility that the rectangular features arise from PEG crystallization. Cryo‑TEM directly visualizes large‑scale nanostructures in the hydrated RN hydrogel, corroborating the AFM results and confirming that the RN structure is a genuine bulk architecture rather than a surface artifact (see Supplementary Fig. 16). Additionally, Small-angle X-ray scattering (SAXS) characterization further reveals a mesoscopic periodicity of about 70 nm for the RN network, with a 4.4‑fold higher intensity than the short‑chain control, confirming the enhanced electron density contrast from the rectangular topology (see Supplementary Fig. 17). Together, AFM, Cryo‑TEM, and SAXS provide consistent multi‑scale evidence—from nanoscale surface morphology to bulk hydrated networks and statistical periodicity—confirming the existence of the rectangular locked structure in the RN hydrogel.

To investigate the hydrogel microstructure and validate the rectangular topology reinforcement mechanism, the morphologies of hydrogel samples were characterized using scanning electron microscope (SEM). As shown in Fig. 4a, b, the fracture surface of the dried PEG_n = 9/90/OH-GE-Cardol hydrogel exhibits distinct rectangular microstructures. These structures remain intact and maintain their rectangular morphology even after freeze-drying, demonstrating structural stability. In contrast, the PEG_n = 9/OH-GE-Cardol hydrogel lacks these structures (see Supplementary Fig. 18). To investigate the effect of chain length ratio, we prepared a long-chain-only control hydrogel (PEG_n = 90/OH-GE-Cardol). However, without short chains essential for the RN, its crosslinking density was too low to form an intact gel network for mechanical testing. As shown in Fig. 4c, optical and SEM images show a collapsed, featureless structure, confirming its inability to sustain loads. See Supplementary Fig. 19 for details.

Fig. 4. Morphology, mechanical properties, swelling/water loss kinetics, fatigue resistance, and comparative benchmarking of RN hydrogels.

SEM images of hydrogels: a PEG_n = 9/90/OH-GE-Cardol; b freeze-dried PEG_n = 9/90/OH-GE-Cardol; c PEG_n = 9/OH-GE-Cardol. Comparative performance of PEG_n = 9/90/OH-GE-Cardol and PEG_n = 9/OH-GE-Cardol hydrogels: d compressive stress-strain curves; e swelling kinetics in water; f water loss kinetics. Cyclic loading tests for: g PEG_n = 9/90/OH-GE-Cardol and h PEG_n = 9/90/BDPD. i Comparison of tensile strength, toughness, and Young’s modulus between the hydrogel prepared in this work (PEG_n = 9/90/OH-GE-Cardol) and representative single-network hydrogels reported in the literature.

The significant divergence in the compression curves (Fig. 4d) can be attributed to the fundamental differences in their network architectures. The PEG_n = 9/OH-GE-Cardol hydrogel possesses only a single amorphous covalent network, where the flexible chains are prone to buckling failure under compression. In contrast, the PEG_n = 9/90/OH-GE-Cardol hydrogel forms an ordered structure in which dispersed PEG-H₂O-PEG microcrystalline domains act as rigid reinforcing points, substantially enhancing the network’s resistance to deformation and its modulus. During compression, the rectangular topology reinforces the network’s deformability, endowing the hydrogel with an efficient energy dissipation mechanism and thereby markedly improving its macroscopic compressive strength. These findings are consistent with the conclusions drawn from XRD, swelling, and tensile tests. The water retention capacity and swelling kinetics of the two hydrogels are shown in Fig. 4e, f. The higher equilibrium swelling ratio of PEG_n = 9/90/OH-GE-Cardol is associated with the formation of a larger network mesh size due to its lower effective crosslinking density. Concurrently, its water retention capability is slightly inferior to that of the more densely crosslinked PEG_n = 9/OH-GE-Cardol hydrogel. Complementary compression, swelling, and water uptake tests conducted on hydrogels in System 2 revealed consistent trends and inter-group differences with the main experimental findings (Supplementary Figs. 20 and 21), further validating the reliability of the initial conclusions and the general applicability of the construction strategy. We conducted cyclic loading‑unloading tests to evaluate the short‑term cyclic stability of the rectangular topology‑reinforced hydrogels. After 50 loading cycles, no significant failure was observed, indicating good cyclic stability under the tested conditions. The results for Systems 1 and 2 are presented in Fig. 4g, h. The observed mechanical behavior is attributed to progressive network rearrangement and hydrogen‑bond reorganization guided by the rectangular topology. To rule out the influence of water loss, sample mass was measured before and after 50 cycles; no significant change was observed (variation < 1%), confirming that dehydration had a negligible effect on the mechanical response. The prepared hydrogels are thermally stable, as demonstrated in Supplementary Fig. 22. For System 1, the PEG_n = 9/90/OH-GE-Cardol hydrogel demonstrated an initial tensile strength of 22.51 MPa in the cyclic loading test. Although the initial stiffness decreased during the initial cycles, it stabilized after the second cycle, indicating structural reorganization during the fatigue process while retaining mechanical integrity. This behavior aligns with the reported cyclic tensile response of hydrogels reinforced by dynamic bonds such as hydrogen bonds⁴¹.

As shown in Fig. 4i, we systematically compared the three key mechanical properties, including toughness, Young’s modulus, and tensile strength, of our RN hydrogel with representative results from the literature. The comparison demonstrates that the hydrogel developed in this study achieves leading performance in all three key parameters, showcasing its overall mechanical properties.

This study presents a design strategy for highly strong and tough hydrogels based on a RN motif. Our experimental results demonstrate that this specific structure enables a high elastic modulus through hydrogen bonding while maintaining large network chain lengths, and significantly enhances both the fracture strain and fracture strength via a strain-hardening mechanism, thereby achieving a simultaneous enhancement of strength and toughness. This toughening mechanism is governed by topology rather than solely by conventional hydrogen-bonding interactions. Compared with conventional homogeneous single-network hydrogels, the topologically structured hydrogel developed here simplifies the preparation process and exhibits higher tensile strength, Young’s modulus, and fracture toughness, indicating its effectiveness in mechanical reinforcement. Furthermore, the reinforcement mechanism based on the rectangular topology can be extended to a broader range of single-network hydrogel systems, providing an important material foundation for broadening the application of hydrogels in medical devices, human-machine interfaces, and implantable systems. While this study emphasizes material design and mechanical principles, the potential applicability of this topology-based reinforcement strategy to other systems merits further research aimed at elucidating the regulatory mechanism of the rectangular topology and expanding its application scope to areas such as flexible electrodes and implants.

Methods

Materials

Cardol glycidyl ether (GE-Cardol) was purchased from Nasurfar Bio-Tech Co., Ltd. (Suzhou, Jiangsu, China)^49,50. Thioglycerol (TG, >95% purity) was supplied by Yuankuo Technology Co., Ltd. (Hubei, China). The photoinitiator 2,2-dimethoxy-2-phenylacetophenone was sourced from Energy Chemical (Shanghai, China) with analytical purity. Lithium hydroxide, dibutyltin dilaurate, and isophorone diisocyanate (IPDI), all of analytical purity, were obtained from Aladdin Biochemical Technology Co., Ltd. (Shanghai, China). Deionized water, ethanol, dichloromethane, and phthalic anhydride, all of analytical purity, were provided by Lingfeng Chemical Reagent Co., Ltd. (Shanghai, China). Pyridine, phenolphthalein, DMAC, and PEG (n = 9, 18, 45, and 90) with analytical purity were acquired from Macklin Biochemical Co., Ltd. (Shanghai, China). The PEG was purified by vacuum drying at 110 °C prior to use. High-purity nitrogen gas (99.999%) was supplied by Changyuan Industrial Co., Ltd. (Nanjing, China).

Preparation of hydrogel

Under a nitrogen atmosphere, purified PEG_n = 9 (4 g), PEG_n = 90 (40 g), and IPDI (8.9 g, molar ratio of 1:1:4) were charged into a three-neck round-bottom flask. The reaction vessel was immersed in an oil bath at 70 °C with mechanical stirring at 100 rpm for 20 min. Subsequently, dibutyltin dilaurate (0.02 g, 0.05 wt% relative to the total mass of PEG monomers) was added as a catalyst, and the reaction was allowed to proceed for another 30 min. Then, DMAC (15 mL) was introduced as a solvent, and the mixture was reacted for an additional hour to afford an –NCO-terminated polyurethane prepolymer (Pre-PU) solution. Upon completion of the reaction, the stirring rate was reduced to 30 rpm, and the system was allowed to cool naturally to room temperature.

Under a nitrogen atmosphere, the Pre-PU solution was added dropwise into a DMAC solution of the crosslinker (0.1 mol/L) at room temperature. The mixture was mechanically stirred at 100 rpm for 10 min, followed by degassing under reduced pressure to ultimately yield the polyurethane hydrogel. Two types of laboratory-synthesized crosslinkers were utilized in this study, namely cardanol-derived polyols (OH-GE-Cardol) and BDPD. The synthetic procedures for these two polyol crosslinkers are described as follows:

1. Synthesis of OH-GE-Cardol

   GE-Cardol (9.38 g, 0.022 mol), TG (10.62 g, 0.98 mol), and 2,2-dimethoxy-2-phenylacetophenone (0.187 g, 0.74 mmol) were placed in a three-neck round-bottom flask. The mixture was mechanically stirred at 30 rpm and irradiated with a 365 nm LED UV lamp (20 W) at 35 °C for 36 h. Subsequently, a mixed solution of lithium hydroxide (0.34 g, 0.014 mol) in ethanol (15 g, 0.326 mol) was added to the reaction system, and the reaction was continued at 35 °C for 4 h. The crude product was washed four times with deionized water, extracted with dichloromethane, and then concentrated by rotary evaporation to remove the dichloromethane, yielding the target product OH-GE-Cardol.

2. Synthesis of BDPD

TG (5.58 g) was dissolved in DMSO (10 mL) in a round-bottom flask. The mixture was stirred magnetically at 90 °C and 30 rpm for 12 h, then cooled to room temperature. The product was precipitated in dichloromethane (500 mL), collected by suction filtration, and this purification process was repeated three times. The resulting solid was then rotary-evaporated at 50 °C to constant weight, affording BDPD as a white powder in a yield of 92.70%⁵¹.

Characterization

Fourier transform infrared (FT-IR) spectra were recorded on a spectrometer (Bruker Equinox 55, Germany) at a scan rate of 0.4 cm/s with 32 scans over the wavenumber range of 500–4000 cm. Nuclear magnetic resonance (NMR) spectra were acquired using a spectrometer (Bruker AVANCE III 400 MHz, Germany). Samples were dissolved in deuterated chloroform, with ¹H NMR spectra obtained using 16 scans and a pulse delay of 2 s, and ¹³C NMR spectra collected with 1024 scans and a pulse delay of 1.5 s. Thermogravimetric analysis was performed using an analyzer (NETZSCH STA 209 F1, Germany) under a nitrogen atmosphere, with a heating rate of 15 °C·min⁻¹ from −45 °C to 100 °C. XRD patterns were collected on a diffractometer (Smartlab 9KW, Japan) using Cu Kα radiation (40 kV, 40 mA), scanning from 0° to 80° at a rate of 10°·min⁻¹. The fracture surface morphology of the hydrogels was examined using a field-emission SEM (Hitachi S4800, Japan). Hydroxyl value determination was conducted according to the ISO 4327:2019 standard. AFM was performed on a Bruker Dimension Icon in tapping mode to obtain height and phase images; phase contrast reflects local stiffness. SAXS was conducted at SSRF BL19U2 ($\lambda$ = 0.124 nm, $q$ = 0.005–0.5${\mathrm{\AA }}^{-1}$), and periodicity. Cryo-TEM was performed on a Talos Arctica (200 kV) under low-dose mode using vitrified samples to directly visualize the network morphology of RN and Ref. hydrogels.

Mechanical properties, including tensile, cyclic tensile, and compressive behaviors, were evaluated using a KJ-1065 universal testing machine following ISO 23529:2016. For tensile tests, dumbbell-shaped specimens (25 × 4 × 2 mm³) were prepared and stretched at 60 mm·min⁻¹ to obtain stress-strain curves. Cyclic tensile tests were conducted over 50 consecutive cycles at a crosshead speed of 5 mm·min⁻¹. All cyclic tests were performed under ambient conditions (23 ± 2 °C, 50 ± 5% relative humidity). The sample mass was recorded before and after the 50-cycle test to monitor potential water loss. For compression tests, cylindrical specimens measuring 12 mm in thickness and 29 mm in radius were compressed at 5 mm·min⁻¹ up to 60% strain to evaluate compressive performance. Tear energy (Gc) was determined using single-edge notched tension tests on rectangular specimens (30 × 10 × 2 mm³) with a 15 mm pre-cut notch at a rate of 10 mm·min⁻¹, calculated as Gc = U·A⁻¹, where U is the work of fracture (integral area under the force-displacement curve) and A is the fractured cross-sectional area²².

The swelling behavior of the samples was determined gravimetrically. Pre-weighed dry specimens (10 × 10 × 3 mm³) were immersed in an excess of deionized water under a constant temperature of 25 ± 0.5 °C⁵². At selected time intervals, the samples were removed, carefully blotted with filter paper to remove excess surface water, and immediately weighed to obtain the wet weight ($W_{\mathrm{t}}$). The sampling frequency was higher during the initial rapid swelling stage and gradually decreased as the swelling process approached equilibrium. Swelling was considered to have reached equilibrium when the deviation between three consecutive measurements was less than 1%, and the corresponding weight was recorded as the equilibrium weight ($W_{\mathrm{e}}$). The swelling ratio (SR) and equilibrium swelling ratio (ESR) were calculated according to the following equations:

$$SR\left(\%\right)=\frac{W_{\mathrm{t}}-W_0}{W_0}\times 100$$ 3

$$ESR\left(\%\right)=\frac{W_{\mathrm{e}}-W_0}{W_0}\times 100$$ 4

The water retention capacity ($WR$) of the hydrogels was evaluated by exposing the equilibrium-swollen samples to a constant temperature and humidity environment. The initial mass was recorded as $W_0$, and the mass at time T was recorded as $W_{\mathrm{T}}$. The $WR$ was calculated using the following formula:

$$WR\left(\%\right)=\frac{W_{\mathrm{T}}}{W_0}\times 100$$ 5

Author contributions

L.C., L.Z., and N.B. supervised the project, conceived the idea and designed the experiments. Z.Y. and Y.Z. planned and performed the experiments, and collected and analyzed the data. Z.H. helped with characterization of the hydrogel. Z.Y. wrote the initial draft; L.C. and N.B. guided the manuscript revision and reviewed the final version., and all authors discussed the results and commented on the manuscript.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Funding

N.B. disclose support for the research of this work from the National Key Research and Development Program of China (Grant No. 2024YFB3816400), the National Natural Science Foundation of China (Grant No. 52350322, 52432009), the Zhejiang Provincial Natural Science Foundation (Grant No. LD24E010007, LD25B060003), and the Zhejiang Provincial Natural Science Foundation of China under Grant No. 2024ZY01019.

Data availability

The Source Data generated in this study have been deposited in the Figshare repository under the accession link https://doi.org/10.6084/m9.figshare.32101108. The data are also available from the corresponding author upon request.

Code availability

The custom code used in this study is available as Supplementary Code 1.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-76723-w.