Slow-water strategy for fracture-resistant hydrogels

Abstract

Water in hydrogels has often been overlooked or deemed detrimental to the mechanical properties of hydrogels due to the plasticizing effect of water, which weakens interchain interactions and accelerates crack propagation. Here, we propose a slow-water strategy that imparts enhanced fracture resistance to hydrogels. By converting free water into slowly migrating intermediate water, we suppress excess chain mobility by reducing the free volume around polymer chains, thereby enhancing interchain interactions and substantially extending the load-transfer length. Notably, even with only 13% polymer content and without additional network design, the hydrogels achieve orders-of-magnitude improvements in fracture toughness and fatigue threshold solely through medium design, and these enhancements are demonstrated to be dominated by the intermediate water. The strategy demonstrates versatility, providing insights into the role of intermediate water in biological systems and enabling innovative approaches to reinforcing biomimetic materials under complex stress conditions.

The high water content of hydrogels can lead to limited mechanical robustness, but is also key to favourable properties. Here, the authors report the development of a water strategy with free water converted to intermediate water, suppressing chain mobility and improving properties.

Introduction

In living systems and their analogs, like hydrogels, water can make up more than 80% of the total mass¹. Yet, the mechanical properties of hydrogels are often thought to depend solely on the polymer network². The conventional view of water in hydrogels is often overly simplistic. A prevalent simplifying assumption in the field has been to model all water in hydrogels as free water, which fails to capture the impact of water-polymer interactions on mechanical properties^3,4. This perception overlooks the impact that different states of water can bring on hydrogel performance. Consequently, theoretical predictions of hydrogel properties often deviate from experimental outcomes, particularly at low water content. For instance, while the predictions of the Lake-Thomas model align well with hydrogels at high water content, they become significantly inaccurate as water content decreases⁵. Similarly, the entropic elasticity assumption does not hold for dehydrated hydrogels⁶. These discrepancies stem from changes in the state of water, impacting polymer chain interactions.

Water in hydrogels exists in three states: free water (FW), intermediate water (IW, or weakly bound water), and strongly bound water (SBW)¹, classified according to binding strength, freezability, correlated spectroscopic features, and metrics of molecular mobility (Table 1). This operational classification differs from the broader concepts of confined water, which emphasizes spatial restriction within the network, and hydration water, which emphasizes local association with polymer chains or solutes. Free water, which constitutes over 70% of the total water in hydrogels, can migrate freely within the gel network and serves as an effective solvent⁷. Strongly bound water, which does not freeze below 0 °C, is adsorbed onto macromolecules and provides structural support⁸. Intermediate water, which freezes below 0 °C, neither acts as a solvent nor provides rigidity to polymer chains⁹. Water with different states possesses distinct migration rates, with free water moving much faster than intermediate and strongly bound water, resulting in large free volume¹⁰. The proportion and distribution of water in different states are crucial to the stability and functionality of biological structures. For instance, the concentration of bound water in human bones decreases with age, correlating with bone strength and toughness¹¹. Similarly, the hydration of elastin and collagen and the redistribution of intrafibrillar water in living tissues are linked to stress relaxation and stiffness rates¹¹. The decrease in bound water, especially intermediate water, is associated with an increased risk of diseases. Neurodegenerative diseases involve amyloid plaques, where soluble oligomers develop into fibrils¹². Bound water inhibits amyloid fibril growth, while free water favors it. Besides, malignant tissues have a higher proportion of free water compared to normal tissues. These observations underscore the role of water states in governing the structural and functional robustness of biological tissues and their constituent macromolecules.

Table 1.

Classification of three types of hydrated water

Name	Mode of binding	Freezability	Correlation time τc (s)	in-situ ATR-IR O-H stretching region (cm−1)
Free water	Scarcely bound	Freezing at 0 °C	10−12-10-11	3400 − 3200
Intermediate water	Weakly bound	Freezing below 0 °C	10−10-10-9	3400
Strongly bound water	Tightly bound	Non- Freezing below 0 °C	10-8−10-6	3600

In typical hydrogels, such as polyacrylamide with 87% water content, the water exists in a ratio of ~7:2:1 for free, intermediate, and strongly bound water. Most of the water is in a free state, which contrasts significantly with tough biological tissues like skin and muscle, where intermediate water often exceeds 50%¹³. According to the Fox-Flory free volume theory¹⁴, a high content of water, acting as a plasticizer, provides substantial free volume for polymer chains. This weakens interactions among chains, allowing them to move rapidly and thereby failing to reach effective stress dispersion. Free water leads to overlooked impacts such as reduced fracture energy and fatigue threshold or delayed fracture^5,15.

Hydrogel reinforcement has traditionally been guided by the Lake-Thomas model, which links fracture resistance to network parameters such as chain length and polymer volume fraction¹⁶. The approaches of network design have successfully enhanced stress dispersion and fracture resistance in hydrogels¹⁷. The attention here, however, is rather focused on designing the medium around the polymer network to create tougher, fatigue-resistant hydrogels. The use of medium modifiers like sugars and polyols in the stabilization of biomolecules is an old trick of nature¹⁸. These modified media counteract the various stress conditions that an organism encounters, though the exact mechanisms are not fully understood^19,20.

To highlight the role of the medium, we constructed a unified hydrogel network with consistent polymer content and cross-linking density, integrating it with media featuring distinct water states. We report a medium design strategy that uses high intermediate water content to achieve stress dispersion and enhance fracture resistance. By engineering a medium with reduced free water content, we suppress excess free-volume-driven chain mobility, thereby improving interchain interactions and enlarging the stress-distribution zone at crack tips. This strategy can be achieved by using disaccharides such as trehalose, sucrose, or maltose. These disaccharides, known for stabilizing macromolecules, exhibit strong hydration and hydrogen-bond interactions with polymer chains^20,21. These disaccharides exhibit a critical concentration threshold at which they significantly alter the dynamics of the aqueous medium, rather than the limit of their solubility or their maximum possible integration into the polymer network. Beyond this point, they extensively disrupt the tetrahedral hydrogen-bond network of bulk free water, stabilizing it around the disaccharide molecules and inducing a drastic slowing of water dynamics, constituting the slow-water effect^9,21,22. The properties of this stabilized aqueous medium display a strong dependence on disaccharide concentration, with water dynamics undergoing an abrupt transition near this threshold²³. Immersing the polymer network in this medium allows the disaccharides to anchor onto polymer chains via hydrogen bonds and capture surrounding free water molecules through potent hydration, restricting their mobility and converting them into intermediate water (Fig. 1a). Here, intermediate water is used as an operational descriptor of a trehalose-induced slow-water medium, in which disaccharide hydration and polymer–solute hydrogen bonding jointly reduce free-water-driven plasticization and restrict chain mobility.

Fig. 1. Slow-water strategy and characterization.

a The slow-water strategy is achieved through strong hydrogen bonding and hydration. b ATR-IR spectra showing the changes of –OH stretching peak. c The low-field nuclear magnetism curves of PAAm-L, PAAm-M, and PAAm-H hydrogels (here, PAAm hydrogels with low, medium, and high concentrations of intermediate water, respectively). d Relaxation time T₂ and correlation time τ_c of hydrogels. e Raman spectra showing the fitting peaks representing IW and FW in the PAAm-H hydrogels. f The content of SBW, IW, and FW in PAAm-L, PAAm-M, and PAAm-H hydrogels. g The temperature sweep of three hydrogels was tested by a dynamic thermomechanical analyzer from −30 °C to 20 °C. The curves are the loss factor with temperature, where the peak position of the loss factor corresponds to the temperature range of the glass transition of the materials. PAAm denotes polyacrylamide; PAAm-L, PAAm-M and PAAm-H denote PAAm hydrogels with low, medium and high intermediate-water contents, respectively. ATR-IR, attenuated total reflectance infrared spectroscopy; FW, free water; IW, intermediate water; SBW, strongly bound water; T₂, proton spin–spin relaxation time; τc, correlation time; tan δ, loss factor.

Results

Synthesis and characterization of slow-water hydrogels

Here, trehalose is utilized to exemplify the implementation of the slow-water strategy. Three polymer chain media designated as PAAm-L, PAAm-M, and PAAm-H were formulated under conditions necessary to induce the slow-water effect, specifically with trehalose concentrations of 0 M, 0.4 M, and 1.1 M, the latter exceeding the critical 1 M threshold^22,24 (Supplementary Fig. 1). Notably, solvent variation shows negligible impact on the network topology (Supplementary Fig. 2). To avoid conflating the slow-water mechanism with additional high-concentration effects (e.g., sharply increased viscosity driving forces or potential phase-related complications), we did not further extend trehalose concentration in the current study.

ATR-IR spectroscopy was employed to differentiate the -OH stretching vibration peaks in the 3100–3500 cm⁻¹ range (Fig. 1b). Peaks around 3200 cm⁻¹ correspond to free water, while those near 3345 cm⁻¹ are indicative of intermediate water²⁵. As the trehalose concentration increased, the intermediate water peak in PAAm-L was lower, whereas in PAAm-H, the intermediate water peak was surpassed. This transition indicates a redistribution of water populations toward more IW-like environments. In particular, the decrease of the FW-associated band together with the increase of the IW-associated band is more consistent with a shift in population weights across operationally defined water-state windows than with a merely uniform slowing-down of a single bulk-like free-water population.

LF-NMR was used to characterize the proton spin-spin relaxation times ($T_2$) in three types of hydrogels and trehalose solutions (Fig. 1c and Supplementary Fig. 3). $T_2$ values in the range from 0.1 to 200 ms were associated with strongly bound water fully captured by trehalose and chains, which shows minimal mobility. By comparison, from 200 to 10,000 ms range shows the coupled peaks of intermediate water and free water with greater mobility²⁶. We determined the results for $T_2$ and correlation times ${\tau }_c$ of three types of hydrogels (Fig. 1d). As trehalose concentration increased, the $T_2$ values of water molecules in the hydrogels decreased, and ${\tau }_c$ values increased, indicating a reduction in water molecule mobility. LF-NMR peak area analysis allowed for the quantification of different water states, with strongly bound water representing 19.5%, 13%, and 10% of the total water content in PAAm-H, PAAm-M, and PAAm-L, respectively. The strongly bound water on the chains consistently accounted for 10% of the total hydrogel content across all types.

To distinguish IW and FW, we analyzed the Raman spectral band from 3200 to 3800 cm⁻¹, which captures the stretching modes of -OH groups in water molecules²⁷. By comparing the areas under these fitting peaks (Fig. 1e), the ratio of intermediate water to free water is 2.33:1. Similar analyses of PAAm-M and PAAm-L hydrogels revealed ratios of 1:1.55 and 1:2.7, respectively (Supplementary Fig. 4). These ratios are intended primarily as relative, semi-quantitative indicators for comparing samples, rather than exact absolute fractions. Combining these results with the LF-NMR data, we calculated the content of different water states in three hydrogel formulations (Fig. 1f). Taken together, these results support a redistribution of operationally defined water populations toward higher IW fractions and lower FW fractions in the trehalose-rich hydrogels. To assess the effect of intermediate water on chain mobility, we performed temperature slope scanning using a dynamic thermomechanical analyzer (DMA) (Fig. 1g, Supplementary Fig. 5). The obtained loss factor curves, which reflect the glass transition temperature (T_g)²⁸, reveal that a higher T_g corresponds to reduced segmental mobility. Our findings show that increased intermediate water content correlates with elevated T_g values, confirming the mobility-constraining effect of intermediate water. While the reduction in glass transition temperature from the dry to the hydrated state is mainly governed by strongly bound and intermediate water, in the fully hydrated regime relevant to this study, a higher fraction of free water mainly increases free volume and facilitates chain mobility under deformation, whereas intermediate water suppresses this effect and enhances interchain cooperativity.

Mechanical properties and mechanism

The mechanical properties of hydrogels with varying intermediate water contents were evaluated. Tensile tests demonstrated a significant improvement in both fracture strength and elongation with increasing intermediate water content, while the modulus remained constant. (Fig. 2a, Supplementary Fig. 6, Table 2). Fracture strength rose from 25±2.1 kPa (PAAm-L) to 33±2.6 kPa (PAAm-M) and 90±6.07 kPa (PAAm-H), while stretch increased from 15±2.4 to 27±3.2 and 55±5.1. The underlying mechanism involves intermediate water-induced enhancement of interchain cooperativity and local solvent-polymer interactions, which homogenize the stress distribution and mitigate stress localization at network defects. The softening of PAAm-H and PAAm-M results from hydrogen bonds determined by Raman spectroscopy and ^¹³C NMR spectroscopy (Supplementary Figs. 7–8). We assessed the fracture toughness using the Rivlin-Thomas method²⁹ (Supplementary Fig. 9). Conventional PAAm-L hydrogels displayed a fracture toughness of only 650±105 J/m². PAAm-M and PAAm-H hydrogels, nevertheless, demonstrated significantly higher fracture toughness values of 2800±344 J/m² and 9320±1230 J/m². (Fig. 2b). Notably, the PAAm-H hydrogel exhibited increased resistance to crack propagation. The striking differences in the mechanical behavior between PAAm-L and PAAm-H hydrogels underscore the critical role of medium in resisting fracture. In conventional PAAm-L hydrogels, over 70% of the water molecules exist as free water, interacting weakly with the polymer chains and exchanging positions with other water molecules on the time scale of ~1–100 picoseconds³⁰ (Fig. 2c). Abundant free water weakens interchain interactions, which diminish stress transfer pathways, making chain segments near the crack tip more susceptible to slippage or fracture. In PAAm-H hydrogels, the disaccharide can form strong hydrogen bonds²⁰ with polymer chains (Supplementary Figs. 7–8). Meanwhile, the disaccharide captured surrounding free water to intermediate water by disrupting the tetrahedral hydrogen-bond network of water²⁰. Compared to free water, intermediate water migrates 1–2 orders of magnitude slower, reducing the free volume for polymer chain movement, enhancing interchain interactions, and promoting chain cooperativity. Notably, the slow-water strategy is distinct from the high-viscosity solvent strategy, which slows the redistribution of stress in stretchable polymer networks, leading to stress concentration at crack tips and hindering effective dissipation, ultimately reducing toughness³¹. Meanwhile, fluorescence recovery was observed in both PAAm-L and PAAm-H hydrogels under the tested conditions (Supplementary Fig. 10). The increased interchain interactions compact the polymer network, endowing the hydrogels with excellent self-healing properties (Supplementary Fig. 11). Moreover, the intermediate water enhances toughness without introducing significant hysteresis, even under large stretch (Supplementary Fig. 12).

Fig. 2. Impact and versatility of the slow-water strategy on hydrogel mechanics.

a Stress-stretch curves of PAAm-L, PAAm-M, and PAAm-H hydrogels. b Fracture toughness of three types of hydrogels. Γ denotes fracture toughness, W(λ) is the strain-energy density at stretch λ, and H is the initial specimen height. c The conversion of free water into intermediate water reduces the free volume surrounding polymer chains and suppresses excess chain mobility. d Nonlinear elastic length and dissipative length of PAAm-L, PAAm-M, and PAAm-H hydrogels. l and ξ denote the nonlinear elastic length and dissipative length, respectively. e Comparison of slow-water strategy with network design strategy, filler reinforcement strategy, and dehydration strategy. f Crack extension per cycle dc/dN versus applied energy release rate G for PAAm-L and PAAm-H hydrogels. G denotes the applied energy-release rate, and dc/dN denotes the crack-extension rate per loading cycle. g The single-notch test of PAAm-H hydrogels under cyclic loading with the maximum stretch = 6. N denotes the number of loading cycles. h–k Validation of the versatility and efficacy of the slow-water strategy in enhancing hydrogel mechanics. h Stress-stretch curves of PAAm networks modified by sucrose. i Toughness of PAAm and PAAm-H-S hydrogels. j Fatigue threshold of PAAm-H-S hydrogels. k Stress-stretch curves of trehalose-modified PEG network. For a, b, d, h–k, data are presented as mean ± s.d.; n = 3 independently prepared hydrogel specimens per group. Shaded bands in a, h and k and error bars in b, d and i denote s.d. All individual data points are shown in b and i. Source data are provided as a Source Data file.

Table 2.

Hydrogel content, chain length, and shear modulus

	${\varphi }_{AAm}$	${\varphi }_{Water}$	${\varphi }_{Trehalose}$	N	G(kPa)
PAAm-L	12.5%	87.5%	0	2,000	3.17±0.11
PAAm-M	12.5%	80.75%	6.75%	2,000	3.14±0.09
PAAm-H	12.5%	65.75%	21.75%	2,000	3.10±0.12

Where$N=\frac{N_{AA\mathrm{m}}}{2N_{MBAA}}$. N denotes the nominal number of acrylamide units per network strand; and G denotes the shear modulus. G values are presented as mean ± s.d. n=3 independently prepared hydrogel specimens per formulation.

We examined the nonlinear elastic length l~Γ/E and dissipative length ξ ~Γ/W*, which are both critical in understanding fracture behavior for highly deformable soft materials³² (Fig. 2d). The results reveal that altering the medium properties can alone increase the dissipative length ξ from 2 mm to 24 mm, expanding the dissipation zone by an order of magnitude. The nonlinear elastic length l increased from 20 mm to 160 mm, and means l ≫ ξ, confirming materials are classified as soft tough materials, capable of sustaining large deformations while maintaining enhanced fracture resistance.

We compared the slow-water strategy with other toughening mechanisms employed in PAAm single-network hydrogels with crosslink densities ranging from 2.5 × 10⁻⁴ to 1.5 × 10⁻³ such as network design³³, filler reinforcement^34,35, and dehydration⁵ (Fig. 2e). Notably, the slow-water strategy overcomes the limitations of low crosslinking density and polymer content, significantly enhances the toughness and enables the design of materials with properties exceeding those of the compared strategies.

We quantified the fatigue resistance of these hydrogels by measuring their fatigue thresholds (Fig. 2f, Supplementary Figs. 13–14). PAAm-L shows a threshold of 9.2±2.2 J/m², while PAAm-H exhibits a significantly higher threshold of 212±10.6 J/m². PAAm-H withstands 20,000 cycles at a 6-fold stretch ratio without crack propagation (Fig. 2g, Supplementary Movie 1). Notably, these monotonic improvements in fracture toughness and fatigue resistance parallel the progressive reduction in water mobility (decreased T₂ and increased τ_c), the increase in T_g, and the reduction in polymer-chain free-motion space R_p across the hydrogel series, supporting a direct link between water-state regulation and mechanical reinforcement (Supplementary Fig. 15). This enhancement results from effective stress dissipation, which is achieved through strengthened chain interactions caused by reduced chain free volume and local solvent-polymer interactions in the intermediate water-dominant hydrogel. This process extends the load-transfer length, thereby activating more polymer chains during fatigue loading. Meanwhile, disaccharides can further improve energy dissipation within the network by forming strong hydrogen bonds between adjacent polymer chains²⁰. Importantly, the slow-water strategy improves mechanical properties without introducing viscoelasticity, as shown by relaxation tests (Supplementary Fig. 16), variable-rate loading (Supplementary Fig. 17 and Supplementary Fig. 18), and rheological frequency sweep analyses (Supplementary Fig. 19). These observations suggest that the equilibrium water-state distribution remains a reasonable approximation during mechanical loading, as any local perturbations of hydrogen-bonded water environments are expected to relax rapidly relative to the loading timescale and do not give rise to pronounced rate dependence, hysteresis, or mechanical drift. In contrast to glassy hydrogels, the slow-water hydrogels exhibit no significant temperature or rate dependence, while showing no characteristic viscoelastic signatures.

Versatility of the slow-water strategy

The slow-water strategy is rooted in two fundamental properties: (i) the ability of an additive to form hydrogen bonds with polymer chains, and (ii) its capacity for strong hydration. To assess the universality of the strategy beyond trehalose, sucrose, a compound with similar properties, was employed to formulate an intermediate water-dominant medium. This modification enhanced the fracture elongation and tensile strength of the PAAm hydrogel from 12±1.1 to 21±3.5 and from 24±1.2 kPa to 56±2.8 kPa, respectively (Fig. 2h). Additionally, the fracture toughness and fatigue threshold of the PAAm-H-S hydrogel (PAAm with sucrose-induced intermediate water-rich medium) increased significantly, from 650±105 J/m² and 9.2±2.2 J/m² to 3,720±450 J/m² and 50±3.8 J/m², respectively (Fig. 2i, j).

The slow-water strategy effectively reinforced a variety of hydrogels. To decouple its effect from specific solvent-polymer hydrogen bonding, we applied it to model systems, including tetra-arm polyethylene glycol (PEG) hydrogels with a near-ideal, defect-free network. In such perfect networks, trehalose’s hydrogen bonds, known to repair network defects, yield negligible mechanical improvement due to the lack of enough binding sites²⁰. Crucially, for the PEG-H hydrogel (the ideal PEG networks with an intermediate water-rich medium), fracture elongation and strength were significantly enhanced compared to the PEG-L hydrogel (the ideal PEG networks with a water medium) (Fig. 2k). This result provides definitive evidence that the reinforcement is primarily driven by the slow-water enhancement of interchain cooperativity. Consistent with this universal mechanism, significant improvements in fatigue thresholds were also observed in PAAm-alginate and PAAm-PAMPS hydrogels (Supplementary Fig. 20).

Slow-water induced chain constraints

The rearrangement of polymer chains and their medium at the molecular level necessitates the availability of free space, occupied by the atoms involved in these movements. The free volume theory³⁶ suggests that as the quantity of free water diminishes, the free volume accessible to the polymer chains reduces, hindering the overall structural rearrangement. We employed molecular dynamics (MD) simulations to investigate the chain dynamics induced by changes in the medium. Packmol³⁷ was used to position a specified number of PAAm polymer chains, trehalose molecules, and water molecules in a cubic box, forming the initial structure model for MD simulation (Fig. 3a). Detailed molecular modeling methodologies can be found in the Table 3 and Methods. We replaced water with trehalose to modify the medium while keeping polymer content constant. Diffusion coefficients were calculated to assess water migration at varying trehalose concentrations (Fig. 3b), revealing a significant decrease in diffusion as trehalose concentration increased, due to trehalose trapping water molecules. The diffusion coefficients of trehalose molecules within the network were consistently two orders of magnitude lower than those of water molecules (Fig. 3c), indicating that trehalose remains anchored to polymer chains, forming the intermediate water-rich medium. We also examined the effect of intermediate water-rich medium on polymer chain mobility by analyzing the radial distribution function (RDF) of the polymer chains. We defined the radius of the free motion space (R_p) as the value corresponding to the first peak in the RDF results (Fig. 3d). Increasing trehalose concentration reduced the radius of free motion R_p (Fig. 3e). Accordingly, a decrease in the free volume of the polymer chains would restrict segmental motion and enhance network interactions. Water molecules become entrapped around the structures formed by trehalose and polymer chains, with the dynamics of these molecular chains influenced by the medium. A reduction in free water decreases polymer chain mobility, thereby suppressing local free-volume-assisted deformation and swelling at the crack tip. Concurrently, enhanced interchain interactions help maintain network stability and promote uniform stress distribution. This synergistic effect effectively suppresses crack propagation and significantly improves fracture toughness. The intermediate water-dominant medium reduces the mobility of the molecular chains while enhancing their stability. This phenomenon underscores the critical role of intermediate water in maintaining the structural integrity of macromolecules within biological systems. While these simulations capture the essential dynamic signatures at the present scale, future extensions to larger systems and explicit interaction energetics would further refine the molecular picture.

Fig. 3. Medium variation modulates polymer chain dynamics.

a The molecular dynamics model mixing PAAm chains, trehalose, and water molecules. b The diffusion coefficient of water molecules changes with trehalose content, which is picked as the steady-state value. The small figure illustrates the method of taking the steady-state value using 0.45 M as an example. c The diffusion coefficient of trehalose molecules changes with trehalose content, which is picked as the steady-state value. The inset shows the steady-state migration rates at corresponding trehalose concentrations during the simulation process using 0.45 M as an example. d RDF of a polymer molecule containing 20 trehalose molecules. e The radius of the free motion space of the chain R_p varies with the increasing trehalose molecules. The two circles show the schematic diagrams of the polymer chain with different R_p values. RDF denotes the radial distribution function, and R_p denotes the radius of the free-motion space of the polymer chain.

Table 3.

Formulation of the prepared trehalose-PAAm systems

Specimen	1#	2#	3#	4#	5#	6#	7#	8#	9#	10#
Acrylamide	200	200	200	200	200	200	200	200	200	200
Trehalose	0	10	20	30	40	50	60	70	80	100
Water	7000	6665	6330	5995	5660	5325	4990	4655	4320	3650

Slow-water hydrogel for complex stress conditions

Slow-water hydrogels not only address the demands for toughness and fatigue resistance in traditional applications such as sensors, actuators, optics, electronics, drug delivery, and tissue engineering, but also show improved deformation stability observed in conventional hydrogels under complex stress conditions³⁸. To evaluate this feature at the device-demonstration level, we fabricated PAAm-H hydrogel balloons and examined their stability under biaxial inflation and cyclic deformation.

The slow-water hydrogel balloon exhibits short-term static and cyclic deformation stability under biaxial inflation (Fig. 4a). Initially, a set volume of gas was pumped into the chamber and sealed for two hours (Fig. 4b(i), Supplementary Movie 2). The results revealed no significant changes in shape. We cycled the inflation and deflation of the container (Fig. 4b(ii), Supplementary Movie 3). Over the test course, the PAAm-H hydrogel exhibited minimal shape change.

Fig. 4. Functionality and applications.

a Diagram of the hydrogel stress testing device. b Stability test of PAAm-H hydrogels under biaxial stress. (i) Static stability test of PAAm-H hydrogel under biaxial stress with constant stretch of 4 in the horizontal direction. (ii) Dynamic stability test of PAAm-H hydrogel under biaxial stress with maximal stretch of 4 in the horizontal direction. c Schematic illustration of an ex vivo tissue-like soft-support model using a hydrogel balloon. d Device-level deformation-stability demonstration of hydrogel balloons in a porcine stomach model. (i) Support-capacity comparison between PAAm-L and PAAm-H hydrogel balloons. (ii) Short-term static and cyclic deformation stability of the PAAm-H hydrogel balloon in the porcine stomach model.

We employed a molding method to fabricate PAAm-H hydrogel balloons coated by EcoFlex, avoiding dehydration or swelling (Supplementary Fig. 21), which simulate both dynamic and static support scenarios (Supplementary Figs. 22–23), such as those encountered in an ex vivo tissue-like soft-support model (Fig. 4c). The PAAm-H balloon was able to support nearly the entire volume of the porcine stomach, under the tested conditions, while the PAAm-L hydrogel balloon could only support 30% of the volume (Fig. 4d(i)). We further confirmed that the PAAm-H hydrogel balloon showed minimal shape change during the 2 h static test and cyclic inflation–deflation test (Fig. 4d(ii) and Supplementary Movies 4–5). Notably, this hydrogel balloon is engineered for gentle, conformable tissue support at ultra-low pressures, leveraging its tissue-like softness and fatigue resistance, rather than for the high-pressure mechanical dilation typical of commercial latex/polyurethane devices. Additionally, PAAm-H hydrogels can fabricate artificial fish bladders for rapid depth adjustment in bionic fish (Supplementary Fig. 24, Supplementary Movie 6). The fatigue and fracture tests provide material-level evidence for the intrinsic resistance of the slow-water hydrogel to crack initiation and fatigue crack growth. The hydrogel-balloon experiments further demonstrate deformation stability in a representative device geometry under defined cyclic inflation conditions, where biaxial deformation and geometric constraints are involved. Together, these complementary tests support the potential of slow-water hydrogels for mechanically stable soft-device designs, while future application-specific studies will further clarify device-level durability under more complex loading conditions

Discussion

Since Balcar et al. first divided water into free and bound states in 1919³⁹, insights into its pivotal role in governing material behavior have undergone continuous refinement. Recent observations of dehydration or solvent exchange in hydrogels have revealed performance alterations¹⁷, but the influence of water states, particularly intermediate water, on polymer systems has not been fully appreciated. It is now apparent that polymers require a stable medium to ensure both structural and functional reliability. While the plasticizing effect of water has been shown to weaken the mechanical performance of hydrogels, most strengthening strategies have focused on polymer network design. Our results identify suppression of excess free-water-induced chain mobility as a promising route to fracture-resistant hydrogels. The critical element of this design is intermediate water, which occupies a smaller free volume. Intermediate water constrains the mobility of polymer chains, facilitating stress distribution and improving the fracture resistance without sacrificing flexibility. High contents of intermediate water can be achieved by integrating disaccharides, which serve to anchor onto polymer chains via hydrogen bonds and capture free water through potent hydration. By contrast, inorganic salts or small organic molecules, despite also engaging in hydrogen bonding, may alter water structure differently or even plasticize the network. This strategy has shown applicability across various crosslinked hydrogels, significantly enhancing their mechanical properties. The enhanced fracture resistance is attributed to this coupled medium effect rather than to macroscopic dehydration, increased polymer concentration, or network densification. Moreover, experimental results provide clear evidence that the mechanical enhancement stems from the intermediate-water strategy, not from effects related to solvent viscosity or local solvent-polymer interactions. Importantly, the presence of an intermediate water-dominant medium has minimal impact on the modulus and viscoelastic properties of the polymer network, which fundamentally distinguishes it from glassy hydrogels. We surmise that strongly bound water may affect the modulus of the polymer network, while the viscoelasticity is primarily derived from the inherent nature of the network. Hydrogels modified with a slow-water strategy exhibit static and dynamic fatigue resistance under complex stress conditions, providing possibilities for designing highly stable materials for demanding applications. Insights into water states within polymer systems have broader implications, potentially shedding light on biological processes such as cataract formation driven by disrupted medium stability and the influence of water-stabilized metastable intermediates on fiber formation in neurodegenerative diseases. Delving deeper into these water states could pave the way for biomedical interventions.

Methods

Materials

All the chemicals were purchased online without further modification. Acrylamide (AAm, Lot # C2204091), Sucrose (Lot # S112226), Trehalose (Lot # C2203186), Gluconic acid δ-lactone (Lot # D1929063), Ethylenediaminetetraacetic acid calcium Disodium salt hydrate (Lot # C2025126), 2-Acrylamido-2-methylpropane-1-sulfonic acid (AMPS, Lot # B2104072), 2-Hydroxy-4^′-(2-hydroxyethoxy)−2-methylpropiophenone (Irgacure 2959, Lot # H137984), α-Ketoglutaric acid (α-keto, Lot # K2013084) were purchased from Aladdin, Shanghai. Sodium alginate (Lot # C10751109), N, N’-methylenebisacrylamide 98% (MBAA, Lot # C10700404) were purchased from Macklin, Shanghai. Tetra-amine–terminated PEG (PEG-NH₂) (Lot # PS2-N-5K) and tetra-NHS–terminated PEG (PEG-NHS) (Lot # PS4-H-20K) were purchased from Ponsure, Shanghai.

Synthesis of polyacrylamide hydrogels with an intermediate water-dominant medium

PAAm-L, PAAm-M, and PAAm-H hydrogels were each prepared from 20 mL precursor solutions containing 2.84 g of AAm, with trehalose concentrations of 0, 0.4, and 1.1 M, respectively. Add 0.1 mL of 0.1 mol/L crosslinker MBAA solution, 0.4 mL of 0.1 mol/L Irgacure 2959 solution to every 20 mL precursor solution. The obtained solutions were fully mixed, defoamed, then poured into the designed molds, and put into the UV(365 nm) curing box, polymerized for 3 hours.

Synthesis of alginate-polyacrylamide hydrogels with intermediate water-dominant medium

Sodium alginate (SA) and water were mixed at a ratio of 1:49 to form 2% SA solutions. Three precursor solutions (Alginate-PAAm-L, M, H) were obtained by varying the trehalose concentration (0, 0.35, or 1.1 M) in the 17.5 mL 2% SA solvent, each containing 3.2 g of dissolved AAm. 0.05 g Ethylenediaminetetraacetic acid calcium disodium salt hydrate, 0.1 g Gluconic acid δ-lactone, 0.4 mL of 0.1 mol/L Irgacure 2959 solution, 0.1 mL of 0.1 mol/L MBAA solution were added to the above solutions. Then the liquid precursors were completely mixed and defoamed, then poured into molds, and placed in a UV curing box for 3 hours.

Synthesis of double-network hydrogels with intermediate water-dominant medium

20.7 g AMPS, 10 mL 0.1 mol/L MBAA solution, and 0.25 mL 0.1 mol/L α-Ketoglutaric acid solution were dissolved in 79.3 g water. The prepared solution was mixed, defoamed, and placed under the UV light, cured for 3 hours. The obtained hydrogel was cut into small particles, then put in a lyophilizer to completely freeze and dry for 3 days. The micro dried particles were ground into powder as microgels. Double-network hydrogels were each prepared from 20 mL precursor solutions containing 0.5 g of microgels and 5.68 g of AAm, with trehalose concentrations of 0 and 1.1 M, respectively. Then 0.8 mL of 0.1 mol/L Irgacure 2959 solution, 0.2 mL of 0.1 mol/L MBAA solution were added into the above solutions. The prepared solution was fully mixed, degassed, and transferred into acrylic molds. Then the molds were exposed to UV light for 3 hours.

Synthesis of polyacrylamide hydrogels with sucrose-induced intermediate water-rich medium

PAAm-H-S hydrogels were prepared from 20 mL precursor solutions containing 2.84 g of AAm, with sucrose concentrations of 1.1 M. Add 0.1 mL of 0.1 mol/L crosslinker MBAA solution, 0.4 mL of 0.1 mol/L Irgacure 2959 solution to precursor solutions. The obtained solutions were fully mixed, defoamed, then poured into the designed molds, and put into the UV curing box, polymerized for 3 hours.

Synthesis of tetra-arm PEG hydrogels with an intermediate water-dominant medium

For PEG-H, 0.08 g of tetra-amine–terminated PEG macromer (PEG-NH₂), 0.08 g of tetra-NHS–terminated PEG macromer (PEG-NHS), and 1 g of trehalose were dissolved in 2.2 mL of PBS solution. For PEG-L, 0.08 g of PEG-NH₂ and 0.08 g of PEG-NHS were dissolved in 3.2 mL of PBS solution without trehalose. The precursor solutions were stirred until fully dissolved, injected into the molds, and allowed to react undisturbed for 24 h.

Characterization

Swelling characterization

To verify that the preparation methods did not alter the network topology of the hydrogels, we examined the swelling behavior and mechanical properties at equilibrium for three types of hydrogels. The prepared hydrogels were immersed in excess pure water, with samples weighed at regular intervals until the mass change fell below 1% between consecutive measurements (indicating swelling equilibrium). Supplementary Fig. 2a presents the recorded mass ratio during the swelling process until equilibrium was reached. All three hydrogels achieved swelling equilibrium within 48 hours, exhibiting an initial rapid swelling phase followed by a gradual slowdown. Given the differing preparation parameters that resulted in varying densities among the hydrogels, the mass ratios were converted to media ratios, as shown in Supplementary Fig. 2b. The inset reveals slight differences in the initial content of medium ratio due to the distinct masses of trehalose and water per unit volume. At swelling equilibrium, the medium content became identical across all samples (Supplementary Fig. 2b), confirming complete leaching of trehalose and leaving only pure water as the medium. The consistent media ratio (water content) at equilibrium suggests an identical initial network topology. What is more, as demonstrated in Supplementary Fig. 2c, the three hydrogels exhibited nearly identical fracture elongation and tensile strength at swelling equilibrium. Furthermore, the swollen PAAm-H hydrogel was dehydrated until its polymer volume fraction matched that of the PAAm-L hydrogel. The resulting hydrogel was stored in a sealed bag for two days to ensure uniform solvent distribution within the gel. Finally, the soaked and dehydrated PAAm-H hydrogel was subjected to tensile testing alongside the untreated control PAAm-L hydrogel. The results (Supplementary Fig. 2d) showed that the two hydrogels exhibited highly similar initial modulus, ultimate fracture strength, and fracture elongation.

Rheological test

The rheological tests were performed in the rotary rheometer (HAAKE MARS). Tested hydrogels were cut into cylinders (R = 25 mm, H = 2 mm). Then the samples were placed on the rotary rheometer at room temperature, 25 °C, whose upper and lower surfaces were glued to the rheometer, and tested over a range of frequencies f (0.01 to 320 rad/s) at a shear strain of 0.1%.

Rate-dependence test

The mechanical characterization of rate-dependence was performed in the dynamic thermomechanical analyzer (RSA-G2). The tested hydrogels were designed to a specific size (10 mm × 2 mm × 2 mm). The samples were tested over a range of loading rates (0.4–40 mm/s) at room temperature, 25 °C.

Near-quasi-static loading

We presented a custom-built humidity-control platform to prevent hydrogel dehydration during prolonged mechanical testing (Supplementary Fig. 18a). The system comprises a sealed chamber for sample isolation, a precision humidity sensor for real-time environmental monitoring, and a humidifier connected via a humidification tube to dynamically regulate relative humidity at 70–80%—a critical range for maintaining hydrogel stability under sustained loading conditions. Supplementary Fig. 18b presents the loading curves of PAAm-L and PAAm-H hydrogels under a near-quasi-static loading rate of 0.002 s⁻¹. The results demonstrate that PAAm-H hydrogel exhibits significantly enhanced stretchability and tensile strength compared to PAAm-L. Notably, due to the height limitation of the humidity-controlled chamber, the PAAm-H hydrogel reached the maximum displacement limit of the testing apparatus without fracture. Further comparative analysis in Supplementary Fig. 18c reveals nearly identical mechanical responses of PAAm-H hydrogel at both 0.002 s⁻¹ and 1 s⁻¹ loading rates, confirming its rate-independent mechanical stability even under extremely slow deformation conditions.

Dynamic mechanical analysis (DMA)

Dynamic mechanical analysis tests were performed using DMA Q800 equipment. Hydrogels with dimensions of 12 mm × 9 mm × 2 mm were fixed on a cantilever beam at a frequency of 1 Hz with an amplitude of 50 μm and a heating rate of 5 °C/min from −30 to 20 °C.

Tensile test

All the tensile tests were performed by a tensile machine (Instron 5965) via the pure shear samples (50 mm × 10 mm × 2 mm) at an ambient temperature of 25 °C. The reason for not adopting the dog-bone shape specimen is that the influence of its transition zone becomes non-negligible under large stretch (more than 10). The hydrogels were tested at a strain rate of 6 min⁻¹. The nominal stress was calculated by the tested force divided by the cross-sectional area of the samples. The amount of stretch was evaluated by the elongated length divided by the initial length. All tests were performed on at least three specimens.

Fracture toughness measurement

To obtain the fracture toughness of the hydrogels, we prepared pure shear-type hydrogels of the same size according to the Rivlin-Thomas method and divided them into two batches, one without any treatment, while the other with a 10 mm thin cut at one end of the hydrogel. Both batches were stretched until fracture. The stress-stretch curves were recorded. The stress-stretch curves of the hydrogels without a pre-set crack were integrated to obtain the work of fracture W(λ). The maximum stretch λ_max is recorded for the hydrogels containing a pre-set crack, and the energy release rate of the hydrogel is calculated by Γ= W(λ_max) H, which is the fracture toughness of the hydrogel.

Fatigue test

The pure shear-shaped hydrogel specimens with pre-set 10 mm cracks were tested on the fatigue machine for cyclic tensile tests. Each sample was subjected to cyclic load-unload of a specified maximum stretch at the strain rate of 5 s⁻¹. The energy release rate G was calculated by the formulation G = W(λ)H. The W(λ) was evaluated from the stress-stretch curves of uncracked pure shear hydrogel samples subjected to 100 loading-unloading cycles, with the area enclosed by the 100^th unloading curve used for calculation. We conducted 20,000 loading-unloading cycles for each pre-notched hydrogel specimen while monitoring crack propagation length. We considered the crack to be non-propagating if no extension exceeding 1 mm was observed, which corresponds to an effective resolution of 50 nm/cycle, a scale widely accepted by researchers in this field.

Fluorescence recovery after photobleaching (FRAP)

The hydrogel precursor solution was doped with a fluorescent tracer at 1 wt% of the total monomer concentration. Post-polymerization, selected regions of interest in the hydrogels were subjected to laser bleaching. The subsequent fluorescence recovery was monitored by time-lapse confocal microscopy imaging (OLYMPUS FV 3000).

¹³C-NMR spectroscopy

Four types of samples, PAAm-L, PAAm-M, PAAm-H, and trehalose, respectively, were used for Nuclear Magnetic Resonance (NMR) tests. All the tested samples were fully frozen and dried in a lyophilizer to obtain dry polymer. These dry polymers were ground into powders. The solid powders were tested by Bruker 400 M with 10 kHz MAS spin speed, 4 s recovery time, and 6.5 μs prescan delay.

Raman spectroscopy

Four types of samples, made by PAAm-L, PAAm-M, PAAm-H, and trehalose, were conducted by Thermo Scientific DXR with a 785 nm laser as a light source from 500–4000 cm⁻¹.

Gaussian peak fitting of Raman spectra

The stretching modes of -OH groups in water molecules were captured by Raman spectra from 3200 to 3800 cm⁻¹. Peaks at 3233 and 3401 cm⁻¹ correspond to free water with four complete hydrogen bonds, whereas peaks at 3514 and 3630 cm⁻¹ are indicative of intermediate water associated with weaker hydrogen bonds. Through peak deconvolution and Gaussian fitting of the Raman spectra, four distinct peaks were identified, each representing a specific water state. These assignments should be regarded as operationally resolved windows within a continuum of water environments, because both spectral deconvolution and relaxation-time analysis are to some extent model-dependent and may influence the absolute fractions obtained, whereas the relative trends across samples remain robust.

Fourier transform infrared spectroscopy (FTIR)

ATR-IR tests were performed on Thermo Fisher Nicolet iS20 (USA) from 400-4,000 cm⁻¹. The spectra were obtained at room temperature with 32 scans and a resolution of 4 cm⁻¹.

Low-field nuclear magnetic resonance (LF-NMR)

T₂ of hydrogels were measured using the MesoMR23-060H-I under 0.25 T and 12 MHz at 25 °C resorting to the Carr–Purcell–Meiboom–Gill (CPMG) sequence. The maximum point of every second echo was accumulated using a 90° pulse of 12 μs and a 180° pulse of 24 μs. The delay time between the 180° pulses was 200 μs. The amplitude of every second echo was measured, a total of 20,000 points were collected, and scans were coded using a repetition time of 3 s. The proton spin-spin relaxation times T₂ and the correlation times τ_c can be linked through the Bloembergen–Purcell–Pound equation²⁸:

$$1/T_2=C/2\left(3{\tau }_c+5{\tau }_c/\left(1+{\omega }_0^2{\tau }_c^2\right)+2{\tau }_c/\left(1+4{\omega }_0^2{\tau }_c^2\right)\right)$$ 1

where C is the kinematic constant for water with a value of C = 5.33 × 109 s⁻² and ω₀ is the Larmor frequency.

Biaxial stress measurement

We designed an experimental apparatus to evaluate the performance of PAAm-H hydrogels under complex biaxial stresses. A closed chamber was created by affixing the PAAm-H hydrogel to the top of an open container, which was inflated to apply biaxial stress. For cyclic inflation and deflation of the container, each cycle was controlled to produce a horizontal elongation of stretch = 4, and the deformation of hydrogel across 1000 cycles was observed. Measurements were taken at N = 1, 100, 200, 500, and 1000 cycles, with images captured both at the onset of inflation (N = 1, 200, 1000) and at the point of maximum strain (N = 100, 500).

Stability measurement of PAAm-H hydrogel balloon in porcine stomach

A thin layer of Ecoflex was applied as a waterproof coating to the outer surface of the PAAm-H hydrogel balloon to minimize the impact of swelling in humid conditions, creating an ideal synergistic combination where the Ecoflex coating provides essential hydrophobic protection while the PAAm-H hydrogel provides mechanical robustness under the tested conditions (Supplementary Fig. 21). Besides, Ecoflex is a commercially available silicone elastomer with well-documented biocompatibility and stability in biomedical contexts, which can act as a physiologically inert barrier, preventing direct contact between the hydrogel and the tissue. The hydrogel balloons were placed inside a collapsed porcine stomach, and gas was continuously pumped into the balloons until rupture. To evaluate the static stability of the PAAm-H balloon, the hydrogel balloon was inflated to half the volume of the porcine stomach and sealed to maintain a constant gas volume. Over two hours, no obvious change in shape or size was observed under the tested conditions. Cyclic inflation and deflation were conducted to examine the dynamic stability (Supplementary Movie 5). Over 200 cycles, the supported shape showed minimal change under the tested cyclic inflation–deflation protocol. Fresh porcine stomachs were obtained from a local commercial supplier in Hangzhou, China. The tissues were food-industry by-products, and no animals were procured or euthanized specifically for this study.

MD simulation

In this study, the PAAm chain was constructed using the Build Polymers module in Materials Studio software, which polymerized 20 acrylamide monomers to form the PAAm molecular chains. Packmol³⁷ was used to place a certain number of PAAm polymer chains, trehalose molecules, and water molecules in a cubic box to construct the initial structure model for MD simulation. The further kinetic simulations were carried out using large-scale atomic/molecular massively parallel simulator LAMMPS⁴⁰ (Supplementary Data 1).

The SPC model⁴¹ was adopted to describe the water molecules. The OPLS-AA all-atom force field, developed by Jorgensen et al.^42,43, was chosen to describe the intermolecular interactions between trehalose molecules and polymer systems. The atomic interactions in the system were calculated using the force field, with the long-range Coulombic forces calculated using the PPPM method with a nonbonded cutoff radius of 12 Å. The bond length between hydrogen atoms and their connected atoms was constrained using the Shake algorithm⁴⁴. The velocity Verlet algorithm was employed to calculate the motion equations, with the position coordinates of each atom updated at each time step of 1 fs. The relaxation simulation was carried out under the NPT ensemble at a temperature of 300 K and a pressure of 1.01 × 10⁵Pa. The system temperature and pressure were controlled using the Nose-Hoover thermostat and barostat^45,46, with the temperature damping coefficient set to 100 and the pressure damping coefficient set to 1000. The periodic boundary conditions were applied, and the system was relaxed for a total of 100 ns until equilibrium was achieved. This ensured that the initial model of the system was stable and well-organized for subsequent MD simulations. The number of acrylamides, trehalose, and water monomers in 10 systems with different trehalose concentrations is listed in Table 3.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Z.H. and S.Q. conceived the concept. Z.H. and Y.L. designed and performed the experiments and processed the data. H.Y. performed the molecular dynamics simulation. Z.H., Y.L., H.Y., Y.M., S.L., and J.W. analyzed and interpreted the results. Z.H. and Y.L. drafted the manuscript. All authors contributed to the writing or editing of the manuscript. Z.H., S.Q. and W.Y. supervised the research.

Peer review

Peer review information

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

Funding

This work is supported by the National Natural Science Foundation of China (Nos. 12321002, 12302224, U22A20255, 12132014), and the 111 Project (No. B21034).

Data availability

The source data underlying the figures in the main text and Supplementary Information are provided with this paper as a Source Data file. The initial and final configurations of the molecular dynamics simulations are provided as Supplementary Data 1. Data supporting the findings of this study are also available from the corresponding authors upon request. Source data are provided with this paper.

Competing interests

Authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-76816-6.