Introduction

Meeting the world’s growing demand for renewable and sustainable energy calls for innovative approaches capable of harvesting power from widely available resources1,2,3. Among these, water stands out due to the enormous energy potential stored in natural processes, such as evaporation, precipitation, and flow—estimated to exceed the global annual energy consumption by several orders of magnitude4. Hydrovoltaic technologies harness this energy by converting the interaction energy of water molecules (e.g., from vapor, droplets, or flowing streams) with nano material interfaces into electrical output (Fig. 1a)5,6,7,8. Compared with conventional solar or wind energy systems, hydrovoltaics offer broader accessibility and lower dependence on external conditions, making them highly promising for applications in unmanned platforms, low-altitude economic products, and other lightweight, distributed devices. However, despite this potential, hydrovoltaic performance under extreme temperatures remains a critical bottleneck, severely limiting real-world implementation.

Fig. 1: Physical processes and electrical performance in hydrovoltaic conversion under varying temperatures.
Fig. 1: Physical processes and electrical performance in hydrovoltaic conversion under varying temperatures.
Full size image

a Physical processes of hydrovoltaic conversion on both macroscopic and microscopic scales, including a microscopic schematic illustrating water molecules interaction with the nano material and ion transport processes at room temperature. b Relationship between the Gibbs free energy of water in different phases and temperature, where the intersection points correspond to the freezing and rapid vaporization temperatures of water. Microscopic illustrations depict hydrovoltaic conversion at corresponding high and low temperatures, demonstrating changes in material structure, the hydrogen-bond network of water molecules, and the ion transport environment. c Schematic representation of ion transport over an expanded temperature range facilitated by the molecular clustering strategy. d Model of composite clusters formed through molecular clustering. Comparison of the open-circuit voltage (Voc) and short-circuit current (Isc) between systems incorporating molecular clustering and those using sodium sulfamate (SS) alone. Data are from different experimental subjects and presented as mean ± standard deviation, n = 3. e Comparison of the maximum power density at different temperatures between the hydrovoltaic devices from this study and existing devices (Table S1). Source data are provided as a Source Data file.

The core challenge lies in the temperature-sensitive hydrogen-bond network of water (Fig. 1b). At elevated temperatures, hydrogen bonds readily break, while in subzero conditions they become rigid, in both cases disrupting ion dissociation and transport crucial for hydrovoltaic conversion. This temperature sensitivity stems from the temperature dependence of water’s Gibbs free energy. The phase with the lowest Gibbs free energy is thermodynamically most stable, and for water the gas phase is lowest at higher temperatures while the solid phase is lowest at lower temperatures, which explains water’s tendency to vaporize when heated and to freeze when cooled. To tackle these issues, researchers have tried using high concentrations of inorganic salts or organic components rich in hydrophilic groups to lower the liquid water’s Gibbs free energy across an expanded temperature range9,10. Inorganic salts enhance stability by the strong ion–water interactions, which disrupt the water’s hydrogen-bonding network and suppress ice crystallization at subzero temperatures. Moreover, at elevated temperatures, the strong hydration shells around ions increase the energy required for water molecules to escape into the vapor phase, thereby raising the fast-vaporizing point. However, high ionic concentrations can introduce strong electrostatic shielding that inhibits charge transport. Organic molecules are typically low-freezing-point solvent species that act as co-solvents by partially replacing water. These molecules depress the freezing point by increasing the fraction of non-freezable bound water. Their vapor pressures are generally lower than that of water, thereby reducing the overall vapor pressure of the hydrogel and raising the temperature at which rapid evaporation commences. Nonetheless, organic solvents often exhibit limited dissociation and high viscosity, leading to fewer mobile ions and reduced conductivity. Previous reports in hydrovoltaic research also investigated the temperature-dependent viability of energy harvesting, typically employing organic compounds, such as ethylene glycol or glycerol, and inorganic salts like lithium chloride and calcium chloride11,12,13,14,15,16. The rationale is that such additives’ intrinsic antifreeze properties can extend device operability at sub-zero temperatures. However, these reports also note that at elevated temperatures reduced internal water availability within the generator can restrict sustained ion transport, thereby limiting steady electrical output. Consequently, existing hydrovoltaic devices still exhibit modest power densities (on the order of 100 μW cm–2) across a wide temperature window, making them unsuited for deployment in harsh or variable real-world environments.

Here, we present a molecular clustering strategy to overcome these limitations, employing an integrated system of organic salts and organic molecules that induce the formation of stable composite clusters (Fig. 1c). These clusters not only increase the effective surface charge, drawing in more water molecules, but also mitigate electrostatic shielding between ions. As a result, our molecularly clustered hydrogel maintains robust thermodynamic stability and efficient ion transport across an unprecedented temperature range from –35 °C to 80 °C, achieving power densities up to an order of magnitude greater than previous reports (Fig. 1e). Furthermore, the hydrogel retains remarkable mechanical integrity under these extreme temperatures, demonstrating over 1000% strain and stable electrochemical performance. Together, these attributes offer a reliable path toward self-powered devices in both frigid and high-temperature settings. Overall, this molecular clustering concept addresses the longstanding issue of hydrogen-bond fragility in hydrovoltaic systems, enabling high efficiency under conditions where conventional approaches fail. Beyond hydrovoltaic power generation, this design principle provides a broadly applicable framework for creating functional materials capable of withstanding extreme thermal environments. We anticipate that our findings will stimulate further innovations in flexible electronics, environmental monitoring, aerospace applications, and biomedical devices, where durable and adaptive energy-harvesting materials are becoming increasingly vital.

Results

Design and characterization of molecularly clustered hydrogels

To achieve efficient hydrovoltaic performance over a broad temperature range, we first developed an all-organic hydrogel system based on a molecular clustering concept. Acrylamide (AAm) and sodium vinyl sulfonate (SV) were chosen as the polymerizable backbone, providing both a robust gel matrix and sulfonate groups that contribute mobile charges for hydrovoltaic conversion17,18,19,20,21. To further increase ion availability, we incorporated sodium sulfamate (SS), whose strongly polar –SO3Na moiety dissociates to release additional Na+ 22,23. However, without extra intermolecular interactions, the sulfamate anion (NH2SO3-) freely migrates along with Na+ in the hydrogel, leading to net charge dissipation and diminished power output (Fig. 1d).

To address this issue while balancing ion dissociation and matrix stability, we introduced glycolic acid (GA) and sodium glycolate (SG), both bearing similar hydrophilic functionalities, to induce “molecular clustering” of anions around GA (Fig. 1c). According to the principle of like-dissolves-like and Lewis acid–base interactions, GA’s hydroxyl and carboxyl groups can coordinate with the lone-pair electrons on SS/SG anions, forming robust organic clusters. Notably, SG, having a higher dissociation tendency, generates glycolate anions and Na⁺ preferentially and thus suppresses over-dissociation of GA, retaining its molecular integrity. This clustering confines anions around GA, mitigating random migration for hydrovoltaic conversion and enhancing surface charge for stabilizing water in hydrogels.

We fabricated these molecularly clustered hydrogels via UV-initiated polymerization in an open environment (Fig. S1). During UV exposure, free-radical polymerization produces a three-dimensional crosslinked matrix (Fig. S2), while controlled evaporation of excess water further promotes close contact among GA, SS, and SG, yielding a tightly crosslinked hydrogel architecture. The evaporation step facilitates the controlled transport of the internal solution, thereby promoting interactions among the components24,25. As shown in Fig. S3, Cryo-SEM imaging reveals the hydrogel microstructure after immersion in pure water for 8 h. The evaporation step densifies the porous network and reduces the pore diameter. XRD results show enhanced diffraction peaks for hydrogels prepared via the water evaporation step, as presented in Fig. S4. This finding indicates that water evaporation promotes closer packing of the polymer network26,27, thereby facilitating close contact among GA, SS, and SG.

Fourier-transform infrared (FTIR) spectroscopy revealed pronounced shifts in GA’s –OH and –COOH peaks as SS and SG were added (Fig. S5a), consistent with intermolecular interactions that facilitate the clustering process. The thickness-normalized small angle X-ray scattering (SAXS) patterns shown in Fig. 2a, b indicate that sequential incorporation of SS, SG, and GA preserves the hydrogels’ homogeneity and amorphous character28,29. A comparison of peak intensities shows that co-introduction of SS and SG produces a marked increase in scattering intensity. This behavior is attributable to the fact that both additives are dissociable small organic molecules that generate anions bearing organic functional groups upon dissociation. Electrostatic repulsion among these charged species loosens the internal network and increases structural disorder, thereby enhancing the scattering intensity. By contrast, incorporation of GA leads to a pronounced reduction in scattering intensity, to values below those observed for hydrogels containing SS or the SS-SG combination. This decrease indicates that GA reduces disorder within the hydrogel network. The scattering intensity of the SS-SG-GA hydrogel lies between that of the anion-only and the GA-only hydrogels, indicating interactions among SS, SG, and GA.

Fig. 2: Material Characterization of molecular clustering and simulation analysis of molecular clusters-enhanced water binding.
Fig. 2: Material Characterization of molecular clustering and simulation analysis of molecular clusters-enhanced water binding.
Full size image

a SAXS spectra of pure polyacrylamide/sodium vinyl sulfonate (PAAM/SV) hydrogel with sequential addition of sodium sulfamate (SS), sodium glycolate (SG), glycolic acid (GA), and GA-only molecules. b SAXS 2D mapping of SS-SG and SS-SG-GA added hydrogels. c Electrostatic potential maps compare SS and molecular clustering systems, specifically focusing on charge distributions on oxygen (O) and hydrogen (H) atoms. d Comparison of hydration enthalpies (ΔHHydration) between SS and molecular clustering systems, highlighting the enhanced water binding capability of the latter. e Radial distribution functions (g(r)) highlight the increased water binding around amino groups in molecular clustering systems compared to SS-only systems. f MD simulation snapshots showing the distribution of water molecules around amino groups in SS-only systems and molecular clustering systems, emphasizing the water stabilization effects of molecular clustering. Source data are provided as a Source Data file.

To evaluate the stability of these clusters, we performed density functional theory (DFT) calculations on three proposed binding sequences (SS-SG-GA, SG-SS-GA, and SG-GA-SS), finding similar binding energies in the range of –5.09 eV to –5.65 eV (Fig. S6). Among them, SG-GA-SS exhibited the most negative binding energy (–5.65 eV), suggesting slightly enhanced stability. Simulated infrared spectra based on this representative arrangement (Fig. S7) corroborated the experimental peak shifts, indicating that GA effectively coordinates with SS and SG anions to form stable organic clusters.

Broadening the operating temperature range

To elucidate how molecular clustering improves hydrogel stability under extreme temperatures, we first performed DFT simulations to evaluate the surface potential and hydration enthalpy of the SG-GA-SS cluster (Fig. 2c, d). Compared with a lone SS anion, the presence of molecular cluster significantly increases the negative charge on the sulfonate (from −0.641 eV to −0.658 eV) while concurrently raising the positive partial charge on amino hydrogen atoms (up to +0.331 eV). Additionally, the hydration enthalpy of the cluster drops by a factor of ~1.6 relative to SS alone, suggesting enhanced water-binding capacity that helps maintain liquid water over a broader temperature window.

Next, we conducted molecular dynamics (MD) simulations to visualize the water distribution around SG-GA-SS versus an only-SS anion. Radial distribution functions (g(r)) reveal abundant water molecules within 1–8 Å of each hydrophilic site (Fig. S8), with the amino group of clusters exhibiting particularly high local water density due to its strong polarity (Fig. 2e, f). At equilibrium, the cluster’s first and second hydration shells collectively display nearly three times more hydrogen bonds than in the SS-only scenario (Fig. 3a, b). In line with this observation, FTIR spectra show a blue shift in the O–H peaks (Fig. 3c)30, and the Raman spectrum exhibits enhanced O–H intensities at 3630 cm−1 (Fig. 3d)31. These results confirm that GA-driven clustering strengthens water–cluster hydrogen bonding while partially reducing water–water interactions.

Fig. 3: Simulation and experimental analysis of hydrogen bonding among water molecules and the thermodynamic stability of water within hydrogels facilitated by molecular clustering.
Fig. 3: Simulation and experimental analysis of hydrogen bonding among water molecules and the thermodynamic stability of water within hydrogels facilitated by molecular clustering.
Full size image

a MD simulation snapshots comparing the spatial distribution of water molecules between systems containing only SS and molecular clustering systems. b Hydrogen bond count for water molecules directly bonded to functional groups (1st hydration shell) and indirectly bonded (remaining hydration shell) in SS and molecular clustering systems. The 1st hydration shell represents water molecules directly interacting with functional groups, while the remaining shell surrounds the 1st shell, forming indirect interactions. c FTIR spectra comparison highlighting changes in water molecule chemical bonding caused by molecular clustering. d Raman spectra of water in SS and molecular clustering systems, marking peaks associated with strongly bounded water molecules directly interacting with functional groups. The corresponding ratios are presented as bar charts for clarity. e, f Comparison of molecular clustering systems, SS, and commonly used ethylene glycol (EG) in terms of vaporization enthalpy, freezing enthalpy, retention time at the onset of rapid vaporization, and retention time at the freezing point. Dashed arrows indicate that the box plot and dot plot with matching fill colors represent the same sample. g Sequential images showing multi-axial planar stretching of molecularly clustered hydrogels (1.5 cm edge length). The hydrogels were pre-exposed to environments of 80 °C and −35 °C before being subjected to immediate stretching tests. h Comparison of temperature range of the molecularly clustered hydrogel with reported applications in soft robots, flexible electronics, bioelectronics, and biomedicine. Source data are provided as a Source Data file.

To quantitatively assess water retention in high and low temperature regimes, we measured phase-change enthalpies, including isothermal tests that better reflect real-world applications (Fig. 3e, f). Compared with the SS-only system, the SG-GA-SS cluster increases the average vaporization enthalpy by ~23% and elevates the onset temperature for rapid vaporization to 84 °C, prolonging water retention by ~30%. Under freezing conditions, the cluster reduces the average freezing enthalpy by ~20% and shifts the onset of fast freezing down to −41 °C, enabling a ~ 54% longer duration of liquid water. This observation is consistent with previous DFT calculations and MD simulations, which indicate that the presence of GA molecules reduces the hydration enthalpy of the clusters. As a result, a higher proportion of water molecules become associated with the clusters. Compared to free water molecules, breaking the hydrogen bonds between these bound water molecules requires greater energy input, thereby shifting the onset temperature of rapid vaporization to a higher range. Simultaneously, the energy barrier for the bound water molecules to form ice crystals with surrounding water is increased, leading to an extension of the fast-freezing onset temperature toward lower values. These enhancements surpass those of typical glycol-based hydrogels by roughly 10 °C on both the high- and low-temperature ends (Fig. S10), underscoring how molecular clustering imparts stronger thermodynamic stability to retained water.

Finally, we evaluated the molecularly clustered hydrogel’s behavior under temperatures from −35 °C to 80 °C, focusing on optical transparency and mechanical strength (Figs. S11 S12). Throughout this range, the molecularly clustered hydrogel retains stable surface morphology and clear optical properties, with tensile strains of around 1000% and only a 3–6% deviation from room-temperature performance—significantly outperforming pure hydrogels without clusters (45–52%). In addition, the deviation in Young’s modulus for the molecularly clustered hydrogel was generally less than 15%, whereas that of the cluster-free hydrogel consistently exceeded 15%. Under multi-axial extension, it achieves large-area deformation (surface strain ~1800%) without fracturing (Fig. 3g). Rheological behavior is shown in Fig. S13, the results revealed that the hydrogel consistently exhibited a storage modulus (G’) greater than the loss modulus (G”) at –35 °C, 25 °C, and 80 °C, signifying characteristic of soft and elastic solids32. We attribute this robust resilience to the GA-induced clusters that firmly bind water, mitigating the structural damage typically caused by high-temperature vaporization or low-temperature freezing. Overall, the molecularly clustered hydrogel maintains liquid water and mechanical flexibility from −35 °C to 80 °C. Beyond the hydrovoltaic application, hydrogels employed in diverse fields, such as flexible electronics and biomedicine also demand thermal and mechanical stability of the confined liquid water under varying temperature conditions33,34, as the crystallization of water at subzero temperatures or evaporation at elevated temperatures can severely impair their structural integrity and functional performance. Compared with representative hydrogels reported for soft robotics35,36,37, flexible electronics38,39,40, bioelectronics41,42,43, and biomedical44,45,46 applications, the molecularly clustered hydrogel achieves superior thermal stability of liquid water over an extended temperature window (Fig. 3h). Furthermore, relative to state-of-the-art hydrogels used in many soft multifunctional electronic platforms, this hydrogel exhibits larger tensile strain and lower Young’s modulus over a wider temperature range, indicative of its stronger ductility (Fig. S14, Table S2 and Table S3). This broad temperature tolerance highlights the potential of our molecular clustering approach for self-powered devices and other wide-temperature applications.

Enhancing hydrovoltaic performance

Effective ion transport over a wide temperature range is crucial for hydrovoltaic conversion. Unlike most ionic conductors, hydrovoltaic effects rely on the fixation of anions and the directed migration of cations, accumulating net charges at the terminal interface47. Within the molecularly clustered hydrogel, clustering of SS and SG anions induced by GA molecules is expected to alleviate the steric hindrance experienced by Na+ during transport. We quantify the degree of steric hindrance acting on Na+ using thermodynamic entropy. The mathematical definition of entropy used in this analysis is given in the Methods section. Larger entropy values correspond to a higher degree of disorder and to greater freedom of Na+ movement. Analysis of the MD simulations indicates that Na+ ions in the molecularly clustered system exhibit higher entropy than those in the SS-only system (Fig. S15a). This finding implies that formation of molecular clusters reduces the steric hindrance acting on Na+. The resulting free Na+ transport contributes to enhanced hydrovoltaic performance. DFT calculations provide complementary evidence for this conclusion by showing that the attractive interaction between SS anions and Na+ in the molecularly cluster is markedly smaller than the corresponding interaction energy in the SS-only system (Fig. S15b). Such reduction reduces steric hindrance acting on Na+, thereby facilitating Na+ mobility.

To quantify the spontaneous ion separation within the molecularly clustered hydrogel at −35 °C, 25 °C, and 80 °C, we measured its ion conductivity (Fig. 4a) under an external electric field. Incorporation of the SG-GA-SS cluster led to conductivities of 61.17, 67.31, and 79.47 mS cm−1, respectively. Notably, the conductivity at −35 °C improved by over 50% than many previous ionogels (Fig. S16 and Table S4), indicating that molecular clustering by GA effectively promotes ion dissociation and transport. Further source-meter tests (Fig. 4b) showed that the SG-GA-SS cluster reduces the hydrogel’s electrical resistance by about 33% at low temperatures, highlighting the formation of more favorable ion channels through molecular clustering. The reduction in resistance is more significant at low temperature than at higher temperature, which aligns with the well-established temperature dependence of ionic mobility. Ionic conduction inherently diminishes at lower temperatures due to reduced thermal energy and polymer chain dynamics48. For hydrogels without clusters, the simultaneous effects of decreased ionic mobility and reduced water phase fluidity lead to a pronounced increase in electrical resistance. By contrast, SG-GA-SS cluster mitigates these adverse temperature effects.

Fig. 4: Electrical performance of molecularly clustered hydrogels under different temperatures.
Fig. 4: Electrical performance of molecularly clustered hydrogels under different temperatures.
Full size image

a Temperature-dependent ionic conductivity of molecularly clustered hydrogels. b Electrical resistance (RT) comparison of hydrogels with and without molecular clusters across varying temperatures. c Temperature gradient (ΔT)-response potential difference (ΔE) of hydrogels with and without molecular clusters. d Integrated sample photo of molecularly clustered hydrogels. e Hydrovoltaic open-circuit voltage (Voc) of molecularly clustered hydrogels maintained over an extended period at 80 °C and −35 °C. The thick solid line represents the mean, the thin solid line represents the measured data, and the shaded area shows the difference between them. f, g Schematic, photographs, and current response signals of molecularly clustered hydrogels under applied pressure, showcasing potential applications in piezoelectric devices for a wide temperature range. The shaded area separates different applied pressures. Source data are provided as a Source Data file.

In real-world conditions, temperature gradients often drive water evaporation and ion diffusion. To examine the molecularly clustered hydrogel’s hydrovoltaic behavior under these conditions, we conducted non-isothermal experiments (Fig. 4c), maintaining one end at ambient temperature (sealed) and heating the other with an evaporation outlet, applying a 10–25 K temperature difference. As the gradient increased, the SG-GA-SS cluster substantially boosted the open-circuit voltage. Under a 25 K temperature gradient, the response potential difference was enhanced by 183% relative to the control, revealing that this cluster effectively facilitates spontaneous anion–cation separation and charge accumulation. This phenomenon arises because increasing temperature enhances polymer chain dynamics and reduces viscosity, thereby facilitating ion transport49,50. Combined with the thermal stability of liquid water confined within the molecularly clustered hydrogel, these factors lead to a more pronounced increase in potential difference compared with the hydrogel without clusters.

Active metal electrodes, such as copper, aluminum, and zinc have been widely adopted to enhance electrical output and broaden application scenarios in recent hydrovoltaic studies12,15,19,51,52,53,54. Accordingly, we evaluated the open-circuit voltage (Voc) of molecularly clustered hydrogels using these three metals as electrodes. By comparing the average Voc over the same measurement period, with error bars indicating the standard deviation of fluctuations during this interval (Fig. S17), it was observed that zinc electrodes yielded a higher average Voc with smaller fluctuations, indicating superior stability. Furthermore, comparisons were made between hydrogels containing only SS and molecularly clustered hydrogels. Results demonstrated that, regardless of electrode type, the molecularly clustered hydrogel consistently exhibited high Voc, confirming that electrode selection does not alter the conclusion that molecular clustering enhances charge accumulation and hydrovoltaic performance.

Building on these improved properties, GA-based molecularly clustered hydrogels achieved consistently high hydrovoltaic performance from −35 °C to 80 °C (Fig. S18), with peak output power exceeding most reported devices by at least one order of magnitude (Table S1). We next assembled 28 individual hydrogel cells in parallel (Fig. 4d) on a flexible PET substrate, using zinc and carbon pastes as electrodes. Remarkably, the half-decay time of the hydrovoltaic open-circuit voltage maintained up to 120 h at both −35 °C and 80 °C (Fig. 4e)—outperforming many literature reports operating around room temperature (Fig. S19, Table S5). Such endurance underscores the promise of hydrovoltaic technology for sustained energy harvesting in extreme environments.

To prove that the voltage and current are generated from the hydrovoltaic effect, additional tests were conducted using two same graphite felt electrodes and asymmetric graphite felt–platinum inert electrodes. As presented in Fig. S20, when two identical graphite felt electrodes were employed, localized evaporation within the hydrogel induced an internal ion concentration gradient. This promoted directional ion diffusion, leading to charge separation and the generation of both open-circuit voltage and short-circuit current. When platinum–graphite felt inert electrodes were used, the inherent electrode potential difference synergized with evaporation-driven ion diffusion, thereby enhancing electrical output. In contrast, the slow surface oxidation of the zinc-active electrode generates a sustained flux of metal cations, which facilitates ionic diffusion and thereby enhances electrochemical performance. These findings are consistent with previously reported mechanisms in related literature15,51.

Furthermore, recognizing that water evaporation from the hydrogel during the hydrovoltaic conversion process is intrinsically unidirectional and therefore limits practical application, we evaluated a cyclic operating protocol consisting of one hour of evaporation followed by ten minutes of rehydration. As shown in Fig. S21, after six such cycles the device retained 87% of its initial open-circuit voltage and 86% of its initial short-circuit current. These results indicate that, owing to strong hydrogen-bonding interactions between molecular clusters and water molecules, dehydrated hydrogels readily reabsorb water and can undergo repeated evaporation cycles. This behavior is analogous to the charging process of conventional batteries, in which water uptake restores device functionality. Additionally, the combination of moisture absorption and evaporation has been employed to extend the duration11,15,55, while a recently reported innovative internal water circulation mechanism has demonstrated to enable sustained device operation56.

Universality and applications

To explore the universality of the molecular clustering strategy, we replaced SS in the SG-GA-SS cluster of the SV-based hydrogel with sodium methanesulfonate (SM) and examined the phase transition behavior. It was observed that, compared to the system with SM alone, the presence of the SG-GA-SM cluster significantly reduced the initial freezing transition temperature and increased the initial vaporization transition temperature of the hydrogel (Fig. S22a, b), indicating an effective broadening of the operational temperature range. In addition, when SV was substituted with 2-acrylamido-2-methylpropanesulfonic acid sodium salt (AMPS-Na) in the SG-GA-SS cluster system, the clustering strategy remained effective in widening the temperature window relative to the system with SS alone (Fig. S22c, d), further supporting the general applicability of the molecular clustering approach in enhancing the thermal stability of hydrogels. Additionally, the hydrovoltaic generators based on molecular clusters demonstrated higher open-circuit voltage and short-circuit current compared to those containing only SS or SM (Fig. S23), confirming the broad universality of the clustering strategy in improving hydrovoltaic generation performance.

To explore broader applications, we incorporated the molecularly clustered hydrogel to fabricate a pressure sensor by sandwiching it between 3D-printed stainless-steel electrodes (Fig. 4f). Under external pressing, the contact area between the hydrogel and electrodes increases, thereby enhancing the short-circuit current and enabling pressure detection. As shown in Fig. 4g, even at −35 °C and 80 °C, the sensor retained an effective response, indicating that molecular clustering preserves effective ion transport and mitigates both high-temperature vaporization and low-temperature freezing effects. These results illustrate the potential for self-powered sensing and other multifunctional devices in harsh conditions.

Discussion

This study introduces a molecular clustering strategy that overcomes the limitations of conventional hydrovoltaic systems, achieving performance across an expanded operational temperature range (−35 °C to 80 °C). By leveraging synergistic interactions between organic salts and molecules, the SG-GA-SS molecular clusters stabilize water’s thermodynamic properties and enhance ion transport. These advancements enable hydrovoltaic devices constructed from molecularly clustered hydrogels to deliver power densities exceeding those of existing technologies by an order of magnitude and to maintain stable performance under extreme environmental conditions for up to 120 h. The molecular clustering strategy not only improves hydrovoltaic conversion performance but also imparts exceptional mechanical properties to hydrogels, including over 1000% stretchability and superior electrochemical stability. The hydrogels demonstrated a 1.6-time reduction in water-binding energy barrier, a fourfold increase in strongly bound water, and expanded thermal stability, extending their practical applicability in extreme environments. Furthermore, we demonstrated the versatility of the molecularly clustered hydrogels by incorporating them to fabricate a pressure sensor capable of reliably detecting signals under harsh temperature conditions. The broadened thermal range and enhanced ion transport capabilities of these hydrogels offer promising opportunities for hydrovoltaic systems in flexible electronics, environmental monitoring, and biomedical applications. The ability to seamlessly integrate into extreme-environment technologies underscores the transformative potential of molecular clustering strategies in advancing sustainable energy solutions.

Methods

Materials

Chemicals including sodium vinylsulfonate (SV, 25 wt% in water), GA (GA, 98%), sodium glycollate (SG, 98%), acrylamide (AAM), and 2-hydroxy-4’- (2-hydroxyethoxy) -2- methylpropiophenone as UV initiator (UV-I, ≥98.0%, HPLC) were purchased from Aladdin. SS (SS, 98%) and N, N’-methylenebis (acrylamide) (MBAA, AR) were purchased from Macklin. Graphite felt with a thickness of one millimeter was customized by Xinye Electronic Material Manufacturers (Taobao, China). Zinc electrodes with a thickness of one millimeter were purchased from Zhoukou Guangjule Trading Co., Ltd., China. DI water was obtained from local sources (Dalian, China).

Fabrication of molecularly clustered hydrogels

SG-GA-SS cluster hydrogels were prepared by introducing 5.5 mmol SV solute, GA, SS, SG, 3 g AAM, 0.0037 g MBAA, and 0.0054 g UV-I in 7 ml of DI water for pre-solution, which was then stirred on a magnetic stirrer under 50 °C heating and 500 rpm for 15 min until the solution became clear. The hot solution was poured into the mold and irradiated with 365 nm ultraviolet light under 45 W of irradiation power for 40 min. The ambient temperature and relative humidity should be controlled at 25 °C and 60% to ensure a good sample yield during the polymerization process. The amount of SS in SS-based hydrogels is the sum of SG, SS and GA in SG-GA-SS cluster hydrogels. For the hydrogel without clusters in electrical measurements and hydrovoltaic experiments, the SV solute is 5.5 × 3 mmol for ensuring the same amount of Na+ ions as the SG-GA-SS cluster hydrogel.

GA, SS, and SG are key components for validating the molecular clustering strategy. Since the primary focus of this study is to investigate the effect of their clusters on hydrogel properties, we employed a fixed 1:1:1 molar ratio to ensure compositional consistency and simplify the analysis of clustering behavior. Furthermore, we evaluated the hydrovoltaic performance of the hydrogel under this ratio at varying concentrations. As shown in Fig. S24, the highest performance was achieved at 5.5 mmol. Therefore, this concentration was adopted in all experiments to ensure optimal hydrovoltaic output, thereby achieving high-efficiency hydrovoltaic performance across a broad temperature range.

Material Characterization

Characteristic groups and the O-H bond intensity of water in hydrogels were detected by Fourier transform infrared spectroscopy (FTIR, Thermo Scientific Nicolet iS20, USA). The newly synthesized hydrogel was characterized to determine the strength of the O-H bonds, thereby assessing the stability of water in the samples. When characterizing the interactions between functional groups in the hydrogel, dried samples were used to eliminate the influence of water hydroxyl groups. The hydrogen-bond intensity among water molecules in hydrogels was analyzed via Raman spectroscopy with a 532 nm excitation laser (HR Evolution). All Raman spectra were fitted by overlapping peak resolution through the Gaussian function. Surface states of molecularly clustered hydrogels under different temperatures were captured by microscopy (Jiangnan N-300M, China). The above two characterizations were performed on newly synthesized hydrogels to assess the stability of water. The porous networks in molecularly clustered hydrogels were observed by a cryo-SEM (Zeiss Sigma 300) with a cryo-transfer system (Quorum PP3010T), and the samples were pre-soaked in DI water for 24 h. The hydrogel samples used were square specimens with dimensions of 1 × 1 × 0.2 cm, which were fully immersed in water. During immersion, the pores within the hydrogel expanded, and subsequent freezing preserved the swollen structure for observation. It is worth noting that the high-water content of hydrogels limits the applicability of most characterization techniques for observing their internal micro-porous structures. Consequently, the majority of hydrogel studies employ cryo-SEM to visualize pore distribution57,58,59. Although the observed pores may differ from those in the native state and thus cannot provide a fully accurate quantification, using the same pretreatment procedure and characterization conditions allows for reliable qualitative comparison.

Thermodynamic measurements

The thermodynamic behaviors of hydrogels during vaporization and freezing processes were assessed using Differential Scanning Calorimetry (DSC; METTLER TOLEDO DSC3). All hydrogel samples used for characterization were cut from newly synthesized hydrogels. For the freezing analysis, hydrogel samples were hermetically sealed within aluminum crucibles to prevent water loss, and subjected to a cooling protocol that reduced the temperature from 30 °C to −100 °C at a controlled rate of −5 °C min⁻¹, under a nitrogen atmosphere with a flow rate of 50 mL min⁻¹. The heat flow was continuously recorded to detect exothermic phase transitions indicative of freezing. In the vaporization assessment, hydrogel samples were placed in aluminum crucibles with perforations to allow vapor release. These samples were then heated from 30 °C to 200 °C at a rate of 5 °C min⁻¹, under identical nitrogen flow conditions. The resulting heat flow data were analyzed to identify endothermic transitions corresponding to water vaporization. The onset, peak, and completion temperatures of these vigorous phase transitions were extracted from the DSC thermograms, and the associated enthalpy changes (ΔH) were determined through integration of the area under the transition peaks. The start and end temperatures of the peak were used to define the boundaries of the liquid-phase water stability temperature range.

The water retention endurance within the hydrogels was evaluated using a quasi-isothermal heat flux measurement scheme, which enabled the characterization of isothermal phase-change behavior under both high and low-temperature conditions. This approach involves maintaining the hydrogel at its initially vigorous phase transition temperature for 60 min under constant temperature and pressure conditions. The vigorous phase transition temperature refers to the onset temperature of the phase transition, i.e., the temperature corresponding to the initial peak observed during the respective heating or cooling protocol. This temperature is reached starting from room temperature at a rate of 5 °C min−1 for heating or −5 °C min−1 for cooling. Throughout the entire measurement cycle, the nitrogen gas flow rate was maintained at 50 mL min−1. The mass of each sample tested was controlled to approximately 15 mg. Prior to all measurements, the instrument was calibrated according to the manufacturer’s reference guidelines. The changes in the heat flux curve are observed, and the time corresponding to the step change indicates the maximum duration for which the water content within the hydrogel sample is maintained.

Mechanical properties characterization

Hydrogels were cut into dumbbell-shaped specimens according to reported methods to enable tensile testing60,61. Samples were equilibrated at controlled temperatures of −35 °C, 25 °C, and 80 °C prior to testing to assess thermal stability and mechanical performance across a broad temperature spectrum. Tensile tests were conducted using a material testing machine (XK-206s) operating at a constant strain rate of 50 mm min−1. The stress-strain response was quantified by normalizing the applied force to the initial cross-sectional area and the measured elongation to the initial gauge length. Young’s modulus was calculated from the linear region of the stress-strain curve, corresponding to the 0–5% strain interval62. The sample area before and after stretching was estimated using the width of a fingertip as a reference, which was then used to calculate the areal strain.

Electrochemical measurements

All experiments were conducted under meticulously controlled conditions within a programmable environmental chamber (DHS-GDJS-100B), where temperature and humidity were precisely regulated. For the evaluation of hydrovoltaic performance, high-purity zinc and graphite felt electrodes were utilized to characterize the electrochemical properties during hydrovoltaic conversion. Hydrogels, fabricated to dimensions of 1 × 5 × 2 mm, was positioned with electrodes affixed at each end. To minimize external influences, the system was hermetically sealed with cling film, leaving a 1 cm² window exposed to ensure water evaporation. The water gradient established across the hydrogel facilitated ion migration, which in turn generated an electrical output. The electrical performance of the system was continuously monitored in real-time using a multimeter (Keithley 6510).

The electrical resistance of the hydrogels was determined using the four-wire method, facilitated by a Keithley 2450 sourcemeter. During testing, the sample assembly was carried out in strict accordance with the procedures used in the aforementioned hydrovoltaic performance evaluation. This approach ensures that the measured resistance accurately reflects the resistance under hydrovoltaic operational conditions. This method minimizes the influence of lead and contact resistances, ensuring precise measurements. Samples were prepared by attaching four equidistant probes along the surface of the material. The outer two probes supplied a constant current, while the inner two probes measured the voltage drop across a defined length of the sample. This configuration allowed for direct calculation of the sample resistance according to Ohm’s law. The measurements were conducted within a programmable environmental chamber (DHS-GDJS-100B), with the sourcemeter configured to perform multiple sweeps to ensure data reliability. The collected data were processed to extract the resistance values, which were averaged over repeated measurements to account for any variability.

Temperature gradient-response ion transport was measured by positioning identical zinc electrodes symmetrically on opposite sides of the hydrogel samples with an insulating tape applied to the surface. During the testing process, the sample assembly was consistent with that used during the assessment of the hydrovoltaic performance. This consistency ensures that the characterized ion transport pathways and distances align with those during hydrovoltaic operation, thereby accurately reflecting the ion transport performance under actual operating conditions. Uniform heating was applied using a silicone rubber heater (Taobao, China), which was applied under one side of the hydrogel. The open-circuit voltage between the two electrodes was measured in real-time using a Keithley 6510 multimeter (Tektronix, USA).

The ionic conductivity of the material was assessed using a broadband dielectric impedance spectrometer (CONCEPT-80, NOVOCONTROL). We prepared thin hydrogel samples with a uniform thickness of 2 mm for measurements under different temperature conditions to ensure consistency in the experimental setup. The contact area between the electrode and the hydrogel was controlled at 2 mm2. Impedance spectra were recorded across a frequency range from 0.8 MHz to 0.1 Hz, with a voltage amplitude of 1 V to mitigate nonlinear effects17. Measurements were conducted immediately after removing the samples from the controlled environment to minimize any environmental influence. Each impedance spectrum was collected at incremental frequencies and analyzed using an appropriate equivalent circuit model to extract the real and imaginary components of impedance. The ionic conductivity (σ) was calculated from the impedance data using the following equation:

$$\sigma=\frac{\delta }{R\times A}$$
(1)

where δ represents the measured thickness of the sample, A is the cross-sectional area of the sample in contact with the electrodes, and R is the resistance obtained from impedance fitting. The resistance (R) was determined from the impedance data by analyzing the real part of the impedance (Zreal) divided by the magnitude of the impedance (∣Z∣).

Quantum chemistry calculations

All quantum chemical calculations, including the evaluation of thermodynamic properties, were conducted using Gaussian 16, ensuring a comprehensive analysis of the intermolecular interactions and hydration effects in the studied system. Geometric optimization of molecular structures was conducted using the B3LYP DFT method combined with the 6–31 G basis set. Frequency analyses were performed to verify that all optimized geometries correspond to true minima on the potential energy surface. To evaluate binding energy and hydration enthalpy values, the M05-2X functional was employed with the 6-311 + + G basis set, chosen for its efficacy in accurately describing long-range van der Waals interactions63,64. Binding energy of the SG-GA-SS cluster was computed as the difference in total energy between the fully optimized SG-GA-SS cluster and the sum of the energies of the individual SG, GA, and SS molecules. Hydration enthalpy was determined by the enthalpy difference between calculations in the gas phase and those utilizing the polarizable continuum model (PCM) to simulate the solvent environment. Enthalpy values were extracted from thermodynamic calculations using the Shermo procedure65. Electrostatic potential maps were generated and analyzed by GaussView.

Molecular dynamic simulations

MD simulations were conducted using the LAMMPS software package. The simulations employed the optimized potentials for liquid simulations all-atom (OPLS-AA) force field to model systems comprising SG, GA, SS, and water molecules66. Specifically, the SG-GA-SS system contained 40 SG, GA, and SS molecules, alongside 2830 water molecules, while the SS system included 120 SS molecules with an identical number of water molecules. Water molecules were represented using the TIP3P model, as implemented in LAMMPS. Force-field parameters and partial atomic charges for the molecules were derived using the LigParGen web-based service67. The molecular systems were placed in a cubic simulation box with dimensions of 45 × 45 × 45 nm³, and the initial configurations were generated randomly using the Packmol package. The systems underwent simulated annealing, starting with an energy minimization step to relax the system to a local energy minimum, with convergence criteria set to 1.0 × 10⁻⁷ for both energy and force and a maximum of 10,000 iterations. Following minimization, the systems were equilibrated under a number-pressure-temperature (NPT) e ensemble conditions for 1.5 million timesteps, with a timestep of 1.0 fs. Thermodynamic properties, including temperature, pressure, potential energy, kinetic energy, and volume, were recorded every 1000 timesteps during the equilibration phase to monitor system stability. Temperature and pressure were maintained at 298 K and 1 atm, respectively, using a Nosé-Hoover thermostat and barostat with damping parameters of 100 and 1000 timesteps.

The equilibrated state at timestep 50000 was selected for subsequent analyses. Radial distribution functions (RDF) were computed for specific atom pairs of interest, with cutoff distances set to 8.0 Å. The resulting RDFs were averaged over the entire simulation duration and recorded in output files. The final atomic configurations and velocities were saved in both data and custom dump formats for further analysis. Snapshots of the most probable hydration sheaths were sampled from the simulation trajectory using Material studio 2020.

Hydrogen bond (H-bond) analysis was performed using a custom Python script68. The script identified H-bonds based on a geometric criterion: the distance between two oxygen atoms was <3.0 Å, and the O–H···O angle exceeded 150°69. The script tracked bond lengths, determining the minimum and maximum lengths observed. Upon completion of the analysis, the average bond length was calculated and reported along with the minimum and maximum bond lengths. H-bond visualizations were generated using Material Studio 2020, where H-bonds inside the first and remaining hydration shell of hydrophilic groups were separately visualized.

Entropy calculations were performed based on MD simulation data70. A Python script, employing the NumPy and Matplotlib libraries, was used for data processing and visualization71. Entropy, S, was calculated using the following expression:

$$S=\frac{{k}_{B}}{2}{In}\; {{{\rm{det}}}}\left(\frac{{k}_{B}{{Te}}^{2}}{{{\hslash }}}{{{\bf{M}}}}\delta+{{{\bf{I}}}}\right)$$
(2)

where kB refers to the Boltzmann constant, ℏ is the reduced Planck constant, e is the Euler’s number and T corresponds to the temperature. M and I are the mass matrix and the unity matrix, respectively. The covariance matrix, δ, captures the coordinate fluctuations, defined as δij = 〈(xi − 〈xi〉)(xj − 〈xj〉)〉. Input data consisted of atomic coordinates obtained from an MD simulation dump file with a timestep of 1 fs and an atom count of 80. Atomic positions in the x, y, and z directions were extracted, converted from angstroms to meters, and used to construct covariance matrices reflecting atomic positional fluctuations.

Reporting summary

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