High-thermopower ionic thermoelectric hydrogels for low-grade heat harvesting and intelligent fire protection

Zizheng Xiong† ^a, Weiguo Wang†^ab, Zhengzhong Wu^a, Zexi Zhao^a, Qingtao Zeng^a, Hongqiang Li^a, Xingrong Zeng*^a, Bowei Wu^c and Xuejun Lai*^a
aSchool of Materials Science and Engineering, Key Lab of Guangdong Province for High Property and Functional Polymer Materials, South China University of Technology, No. 381, Wushan Road, Tianhe District, Guangzhou 510640, China. E-mail: msxjlai@scut.edu.cn; psxrzeng@163.com
^bGuangdong Blue Si&F New Material Co., Ltd, No. 43, Electronic Base Factory, Science and Technology West Road, Torch Development District, Zhongshan 528437, China
^cThe Affiliated High School of South China Normal University, No. 1, Zhongshan Ave. West, Tianhe District, Guangzhou 510630, China

First published on 22nd July 2025

---

---

1. Introduction

Low-grade waste heat dissipated from sources such as industrial byproducts, biological thermal materials, and solar radiation residuals remains pervasive yet underutilized due to inherent technical challenges.^1,2 Thermoelectric (TE) materials offer a promising solution for direct thermal-to-electric energy conversion, presenting significant potential to mitigate global energy crises.^3,4 These materials leverage the Seebeck effect to generate electric potential from thermal gradients, and the phenomenon has garnered more interest in energy harvesting applications.^5–7 Traditional electronic thermoelectric (e-TE) materials, predominantly composed of inorganic semiconductors and semimetals, have been thoroughly investigated.^8,9 However, their practical applications remain constrained by intrinsic characteristics including low thermopower (∼100–200 μV K⁻¹), poor flexibility and excessive manufacturing costs.¹⁰

The advent of ionic thermoelectric (i-TE) materials has opened new ways to address the above-mentioned issues through the Soret effect, achieving exceptional ionic thermopower values at the mV K⁻¹ level.^11,12 This breakthrough paves a new path for the development of high-thermopower energy storage devices, particularly for low-grade heat recovery technologies.¹³ Nevertheless, there are still numerous challenges that require urgent resolution: phase instability in liquid-state i-TE materials under operational conditions¹⁴ and scarcity of the negative i-TE material (n-type) to complement the prevalent positive thermopower (p-type).^15,16 In recent years, researchers have focused on the p–n type conversion of quasi-solid i-TE gel (i-TEG) to construct a new type of ionic thermopile (i-thermopile) to further enhance the thermoelectric efficiency, the literature has reported i-thermopile based on systems such as 1-ethyl-3-methylimidazolium acetate (Emim:Ac)/H₂O system,¹⁷ 1-ethyl-3-methylimidazole bis(trifluoromethylsulfonyl)imide (Emim:TFSI)¹⁸ and iodide/triiodide (I⁻/I₃⁻) thermogalvanic.^19,20 These systems combine the advantages of leakage resistance, mechanical robustness, and fabrication versatility; however, they remain underdeveloped because of the absence of universal mechanisms governing the polarity conversion.²¹ Notably, extant studies predominantly concentrate on employing different kinds of i-TEGs to construct the p–n junction, while achieving selective p–n conversion and superior thermoelectric performance within the same kind of i-TEG material remains a persistent challenge. In addition, the escalating power density and increasingly stringent operational environment of the i-TEG systems necessitate rigorous fire safety standards to ensure sustained performance under demanding conditions. Recent studies include phosphorous-containing ionic liquid (PIL)-based ionogels (PILGs) with rapid fire response (1 s)²² and stretchable PAA–PEO–SDP hydrogels, exhibiting high p-type Seebeck coefficients (24.7 mV K⁻¹).²³ However, there were still shortcomings: (i) the low Seebeck coefficient (1.31 mV K⁻¹) of PILGs affected its thermoelectric conversion efficiency and (ii) the low conductivity (0.16 S m⁻¹) of the PAA–PEO–SDP hydrogels limited its use in integrated devices for energy storage.

In this work, a high-thermopower ionic thermoelectric hydrogel (HIG) for low-grade heat harvesting and intelligent fire protection was prepared. Through precise sodium chloride (NaCl) concentration modulation, we achieved unprecedented p–n-type conversion within a homogeneous material matrix, demonstrating a tunable thermopower from −1.95 to +2.14 mV K⁻¹. The p–n junction exhibited a much higher Seebeck coefficient of 3.74 mV K⁻¹. The HIG system exhibited sensitive fire-warning capability and excellent flame retardancy, while maintaining superior electrochemical performance and remarkable thermal charging energy storage capabilities, showing good potential for application in supercapacitors. Our work established a new paradigm in ionic thermoelectric materials by p–n conversion, fire safety, and energy storage in one platform. The NaCl-mediated p–n conversion strategy offers simplicity, scalability, and performance unattainable in existing systems, opening avenues for self-powered fire-protective devices.

2. Experimental section

2.1 Materials

Acrylamide (AAm, 99.0%), carboxymethylcellulose sodium (CMC-Na, 50–200 mPa s), sodium chloride (NaCl, 99.5%), N,N,N′,N′-tetramethylethylenediamine (TEMED, 99.5%), N,N′-methylenebis(acrylamide) (MBA, 97.0%) and ammonium persulphate (APS, 98.0%) were obtained from Aladdin. 2-Acrylamido-2-methyl-1-propanesulfonic acid (AMPSA, 98%) was obtained from Macklin. The above-mentioned reagents were used directly without further purification.

2.2 Preparation of the ionic hydrogel (HIG)

HIG was fabricated via a one-pot thermally initiated free-radical polymerization. First, appropriate amounts of AAm, AMPSA, CMC-Na, NaCl, MBAA, APS, and H₂O were placed in a beaker and magnetically stirred at room temperature for 1 h to obtain a transparent and homogeneous mixed solution. Subsequently, a small quantity of APS and TEMED were added and stirred at room temperature for 5 minutes. Subsequently, the mixed solution was transferred into a polytetrafluoroethylene (PTFE) mold, and the heat-induced free radical polymerization was conducted at 65 °C in an oven for 1 h, and the n-type and p-type ionic hydrogels (HIG) with high ionic thermoelectric properties were prepared. The corresponding formulation is presented in Table S1.

2.3 Preparation of the HIG thermoelectric supercapacitor and p–n junction

The p-HIG and n-HIG were cut into rectangular specimens (10 mm × 10 mm × 1.5 mm) and then laminated with carbon cloth electrodes on the upper and lower surfaces of the hydrogels. The electrode was composed of activated carbon, conductive carbon black, and polyvinylidene fluoride, and their mass ratio was 8/1/1. Subsequently, two copper foil tapes (30 mm × 10 mm × 0.1 mm) were affixed to the two end surfaces of the electrodes to connect the circuit, and the contact area between the copper foil and the carbon cloth electrode was 10 mm × 10 mm. For the convenience of experimental testing, two layers of silicone rubber (SiR) pads (10 mm × 10 mm × 2 mm) were placed at the up and down surfaces of the supercapacitor prior to its sealing with polyimide tape, and then the HIG thermoelectric supercapacitor (HIG SC) was assembled, as shown in Fig. S1a.

The schematic diagram of the p–n junction thermoelectric pair (p–n junction) is shown in Fig. S1b. The p-HIG/n-HIG (30 mm × 10 mm × 1.5 mm) was fabricated by connecting the two ends in series with copper foil tape. The p–n junction was set up in a “Π” shape to ensure that the cold and hot ends of p-HIG/n-HIG were consistent.

2.4 Preparation of ionic Wood@HIG

The prepared HIG (with a thickness of approximately 0.5 mm) was cut into the desired size and then firmly adhered to the surface of the commercial larch wood strips by taking advantage of its excellent self-adhesive performance, thereby obtaining the wood strip sample (Wood@HIG) with a fire warning function.

3. Results and discussion

3.1 Design strategy and characterization of HIG

The design strategy and functions of HIG are illustrated in Fig. 1. HIG was fabricated via a one-pot thermally initiated free-radical polymerization method, and its n-type-to-p-type transition was achieved by introducing NaCl. The p–n junction was constructed by combining the n-HIG and p-HIG with copper foils. Fig. S3a shows the Fourier transform infrared (FTIR) spectra of AAm, AMPSA and poly(AAm/AMPSA). The characteristic absorption peaks of AAm included the CC stretching vibration at 1612 cm⁻¹, the asymmetric and symmetric N–H stretching vibrations of the primary amine group (–NH₂) at 3351 cm⁻¹ and 3187 cm⁻¹, respectively, and CO stretching vibration at 1673 cm⁻¹.²⁴ For AMPSA, the characteristic vibration peaks were observed at 1083 cm⁻¹ and 1243 cm⁻¹ (attributed to –SO₃⁻ stretching vibrations), while the absorption peaks at 1666 cm⁻¹ and 1612 cm⁻¹ corresponded to the stretching vibrations of CO and CC.²⁵ The poly(AAm/AMPSA) copolymer revealed the disappearance of the CC characteristic peaks from both AAm and AMPSA monomers, confirming the completion of copolymerization. Notably, the CO stretching vibration peaks near 1668 cm⁻¹ remained observable in the copolymer. The N–H stretching vibration peak of the copolymer exhibited a red shift to 3380 cm⁻¹, presumably caused by the spatial constraints of the polymer backbone and the reduced density of –NH₂ groups.²⁶ Similarly, the –SO₃⁻ vibration peaks also red-shifted to 1041 cm⁻¹ and 1197 cm⁻¹, which could be ascribed to the intermolecular interactions between –NH₂ and –SO₃⁻ groups.²⁷ The above-mentioned results basically confirmed the successful synthesis of the poly(AAm/AMPSA) copolymer.

Fig. S3b shows the FTIR spectra of the CMC-Na and p-HIG/n-HIG gels. The addition of CMC-Na did not obviously change the positions of the functional groups in the poly(AAm/AMPSA) system. This phenomenon may be attributed to two factors: the partial overlap between the absorption bands of polar groups (–COO⁻ and –OH) in CMC-Na and those of –SO₃⁻ and –NH₂ in the copolymer; and the absence of strong chemical interactions among the three components.^24,28 Furthermore, the homogeneous distribution of micro–nanochannels in the hydrogel interpenetrating network for HIG is demonstrated in the scanning electron microscope (SEM) images displayed in Fig. 1a. The findings confirmed that the incorporation of CMC-Na did not significantly alter the chemical structural features of poly(AAm/AMPSA) hydrogels, and also revealed that the uniformly distributed micro- and nanochannel structures within them were the key morphological basis to support efficient anion/cation synergistic transport.

3.2 Mechanical properties of HIG

The mechanical properties of the hydrogels are summarized in Fig. 2. From Fig. 2a, the tensile strength and elongation at break of p-HIG decreased with the NaCl incorporation, which was attributed to the weakening of intermolecular interactions between the polymer chains by the addition of inorganic salts.²⁹ The tensile strength and elongation at break of p-HIG were 79.8 kPa and 587%, whereas the tensile strength and elongation at break of n-HIG without NaCl were 92.6 kPa and 609%, which were 1.16 and 1.02 times higher than those of p-HIG. Fig. 2b and c display the stress–strain curves of n-HIG and p-HIG for 10 stress loading/unloading cycles at 50% strain. The good overlap of stress–strain curves across 10 consecutive loading–unloading cycles at 50% strain confirmed minimal energy dissipation and structural integrity retention in both systems. HIG also possessed excellent flexibility and tensile resilience as illustrated in Fig. 2d–f. It could be observed that the hydrogel retained structural integrity after severe deformation (180° bending, twisting, and knotting) and fully recovered its original shape even after undergoing >200% tensile strain. Additionally, we investigated the variation in the mechanical properties of HIG with a NaCl content, as presented in Fig. S4. The findings indicated that although the addition of NaCl weakened the intermolecular forces, leading to a slight decline in the tensile performance of the p-HIG component, this hydrogel system still exhibited low energy dissipation, excellent flexibility, structural integrity, and shape recovery capability.

3.3 Thermoelectric properties of HIG

The ionic thermoelectric conversion performance of p-HIG, n-HIG, and their constructed ionic thermoelectric pair (p–n junction) is shown in Fig. 3. All three configurations displayed real-time response, repeatability, and linear thermoelectric behavior. Quantitative analysis revealed that when subjected to temperature gradients increasing from 5 K to 20 K, the output voltages increased from −9.2 mV to −38.6 mV for n-HIG, 10.8 mV to 43.4 mV for p-HIG, and 14.8 mV to 73.7 mV for the p–n junction. Furthermore, n-HIG, p-HIG, and the p–n junction exhibited rapid temperature response (response time <1 s) and reproducible thermoelectric performance, highlighting their potential for applications such as temperature sensing and early fire warning systems. Linear regression analysis of voltage–temperature revealed a strong linear correlation (R² = 0.999 for n-HIG and 0.998 for p-HIG) between the thermal voltage and applied ΔT for all configurations. The slope of the fitting line corresponded to the ionic Seebeck coefficient (S_i), which directly determined the thermal voltage generated per unit temperature gradient. The measured S_i values were −1.95 mV K⁻¹, 2.14 mV K⁻¹, and 3.74 mV K⁻¹ for the n-HIG, p-HIG, and p–n junction. These results indicated that the n-HIG and p-HIG possessed comparable yet opposite thermoelectric polarities, while the p–n junction achieved a synergistic enhancement in ionic thermoelectric efficiency (91.4% of the combined thermoelectric contributions from the n-HIG and p-HIG). Notably, the p–n junction exhibited enhanced thermoelectric conversion efficiency compared to the individual components. This enhancement arose from the structural compatibility and nearly equivalent thermoelectric capacities of the individual components, enabling the p–n junction to effectively amplify the cumulative thermal voltage through their series configuration.^12,30 While the p–n junction achieved 91.4% synergistic efficiency, two key factors limited further enhancement: (i) interfacial contact resistance:³¹ the copper/HIG interfaces introduced additional electrical resistance, dissipating thermal energy as Joule heat rather than convertible potential. (ii) Thermal contact losses: the copper foils exhibited high thermal conductivity (∼390–400 W m⁻¹ K⁻¹),³² which caused a portion of the heat to be absorbed by the copper foil, thereby reducing the effective temperature gradient across the p–n junction. The thermal loss directly affected the performance of the thermoelectric device by lowering the available thermal energy for conversion.

To investigate the ionic thermoelectric conversion mechanism of n-HIG and p-HIG, SEM-EDS was employed to characterize the elemental distribution across HIG cross-sections. The specific operation is shown in Fig. 4a; n-HIG and p-HIG were maintained under 10 K for 1 h and then quickly cut into six interval segments, followed by rapid freezing with liquid nitrogen and cold-drying for 24 h. EDS mapping quantified sulfur (S), chlorine (Cl), and sodium (Na) distributions across five regions spanning the hot-to-cold axis. As shown in Fig. 4b, c and S5, the concentrations of S, Cl, and Na exhibited a gradual increase from the hot end to the cold end, reaching their highest levels at the cold end. This observation indicated that free anions and cations migrated from the hot end to the cold end under thermal driving forces. In the n-HIG, S and Na contents at the hot end were measured as 2.50% and 0.26%, while those at the cold end increased to 3.66% and 0.30%. The S/Na ratio (Fig. 4d) at the cold end (12.20) significantly exceeded that at the hot end (9.62), demonstrating that the quantity of SO₃⁻ groups in the n-HIG system not only substantially surpassed that of Na⁺ ions but also exhibited greater thermal migration capability. Consequently, the accumulation of SO₃⁻ at the cold end markedly exceeded that of Na⁺. Conversely, p-HIG exhibited Cl and Na enrichment at the cold end (11.77% and 13.56%) versus hot-end concentrations (10.13% and 9.00%), with a negligible S content (0.87%). Notably, the cold-end Na/Cl ratio (0.87) decreased relative to the hot end (1.13). This phenomenon arose from two factors:^33,34 (i) the smaller ionic radius of Na⁺ compared to Cl⁻ enabled faster cation migration, leading to Na⁺ accumulation and consequent positive charge accumulation at the cold end and (ii) the dual-crosslinked anionic network formed by dissociated CMC⁻ and poly(AAm/AMPSA) copolymer chains significantly hindered the migration of Cl⁻ and SO₃⁻ anions due to electrostatic repulsion. In contrast, Na⁺ cations exhibited enhanced thermal diffusion capability under temperature gradients. These combined effects resulted in cation-selective transport behaviour within the HIG system. The ion thermoelectric conversion mechanisms of n-HIG and p-HIG are schematically illustrated in Fig. 4e:³⁵ in the absence of a temperature gradient, anions and cations remained uniformly distributed within the hydrogel matrix, maintaining electrical neutrality. After applying a temperature gradient, thermally driven ion diffusion induced directional migration of Cl⁻, SO₃⁻, and Na⁺ toward the cold end. In the n-HIG systems, the substantially higher population of SO₃⁻ groups (compared to Na⁺) led to predominant accumulation of negative charges at the cold end, generating a net negative potential. Conversely, in p-HIG systems, the faster migration rate and greater mobility of Na⁺ cations (relative to anions) resulted in excessive positive charge accumulation at the cold end. This asymmetric charge distribution elevated the local potential, thereby demonstrating p-type ionic thermoelectric behaviour. In order to clarify the effect of AMPSA on the thermoelectric properties of HIG, formulations with different AMPSA contents were systematically investigated (Table S2). As shown in Fig. S6, all HIG samples exhibited an n-type ionic thermoelectric behavior. The ionic Seebeck coefficient displayed a non-monotonic dependence on the AMPSA concentration, reaching an optimal value of −1.95 mV K⁻¹ before diminishing with further increases in AMPSA content. This was mainly attributed to the interplay between the mobile sulfonate groups (SO₃⁻) and the overall charge distribution within the HIG structure. At low AMPSA concentrations, the increased density of SO₃⁻ groups enhanced charge accumulation at the cold end under a thermal gradient, thereby amplifying the negative thermovoltage generation and leading to a higher Seebeck coefficient. However, as the AMPSA content increased beyond the optimal level, the presence of electronegative gel molecular chains and the electrostatic repulsion between free SO₃⁻ ions begin to dominate. This resulted in a gradual reduction in negative charge accumulation at the cold end, ultimately causing the ionic Seebeck coefficient to diminish. To elucidate the role of NaCl in the p–n transition, we conducted a series of experiments with hydrogel formulations containing varying trace amounts of NaCl (0, 0.067, 0.33, 0.67, 1.00, and 1.33 wt%). We measured the Seebeck coefficients of these hydrogels and compared them with those of all hydrogels listed in Table S1. The results are presented in Fig. S7 and S8. The results indicated that below 0.67 wt% NaCl, the hydrogel (n-HIG) was dominated by sulfonate groups (–SO₃⁻) from AMPSA and SO₃⁻ accumulated at the cold end by virtue of its quantity, so that a net negative thermal voltage was generated (n-type). When the sodium chloride (NaCl) content increased, the number of Na⁺ ions became sufficiently high, their faster migration rate and greater mobility led to excessive accumulation of positive charge at the cold end. Simultaneously, the dual-crosslinked anionic network formed by dissociated CMC⁻ and poly(AAm/AMPSA) copolymer chains significantly hindered the migration of Cl⁻ and SO₃⁻ anions due to electrostatic repulsion. Consequently, with increasing NaCl content, the magnitude of the ionic Seebeck coefficient of HIG also increased, ultimately resulting in p-type thermoelectric behavior (NaCl >1.00 wt%).

3.4 Fire-warning performance

Benefiting from the exceptional ionic thermoelectric properties, HIG demonstrates significant application potential in self-powered smart fire-protection systems. As shown in the schematic of fire-warning test (Fig. 5a), thermal gradients induced directional migration of ions in HIG, generating a thermovoltage through charge accumulation at the cold end. When the thermovoltage exceeded a predefined threshold (50 mV), the alarm system was autonomously triggered. Both n-HIG and p-HIG generated voltage signals without any external power input, enabling fully self-powered fire-warning responses. The voltage response profiles (Fig. 5b and c) reveal that HIG maintained a low baseline voltage at ambient temperature. Under localized flame exposure, a rapid temperature gradient induced swift ion migration, causing an abrupt voltage surge. Specifically, the n-HIG activated the alarm within 4.3 s, while the p-HIG achieved a thermoelectric potential exceeding 50 mV in merely 3.5 s, highlighting ultrafast fire-detection capabilities. Markedly, the thermovoltage spontaneously reverted to baseline after heat removal, confirming HIG's reversible fire-warning capability. Repeatability tests further verified stable performance over three consecutive ignition cycles. As shown in Fig. 5d, the p-HIG consistently achieved shorter response times during three consecutive ignition tests. This enhancement could be attributed to the incorporation of NaCl, which introduced a high concentration of smaller-sized and faster-migrating Na⁺ ions. It could significantly improve fire-warning efficiency.

Fig. 6a and b show the video screenshot of the fire-warning test of HIG, confirming HIG's repeatable fire-warning capability. When exposed to flames, the n-HIG generated a thermovoltage of 55.8 mV within 4 s, activating the alarm system. After the flame removal, the thermal potential rapidly decayed to its baseline value (3.8 mV). In cyclic ignition tests, the n-HIG demonstrated sustained responsiveness, consistently producing thermovoltage exceeding 50 mV within ∼5 s. Similarly, p-HIG achieved superior repeatability, enabling multiple warnings with shorter activation times during repeated thermal stimuli. Following three consecutive fire-warning cycles, both the n-HIG and p-HIG maintained reproducible performance. The above-mentioned results confirmed the fast response (<5 s), high output thermal voltage (>50 mV), excellent cyclic stability, and highly reproducible fire warning capability of the n-HIG and p-HIG for fire-warning applications.

3.5 Flame retardancy

Fig. 7a, b and S9 present the results of the flame retardancy tests of Wood, Wood@n-HIG, Wood@p-HIG, n-HIG and p-HIG. As can be seen from Fig. 7a and b, both the n-HIG and the p-HIG exhibited exceptional flame retardancy: after two consecutive 10 s exposures to a Bunsen burner flame, only localized charring occurred at the bottom edges, with no flame propagation, accompanied by self-extinguishing behavior after the flame removal. Compared with HIG, the flame of Wood spread rapidly after ignition and was almost completely burnt after 130 s (Fig. S9a), with a limiting oxygen index (LOI) far below 28% (Fig. 7c). Under HIG protection, both Wood@n-HIG and Wood@p-HIG remained non-ignitable after two flames burnt, sustaining only localized charring at the base while protecting the wood substrate (Fig. S9b and c). Visibly, Wood@p-HIG retained superior flame resistance even under extreme conditions (oxygen concentration of 90%) (Fig. 7d). Microscale combustion calorimetry (MCC) further quantified the thermal release properties of the polyacrylamide gel (PAAm), n-HIG, and p-HIG (Table S3). Compared to PAAm, the n-HIG and p-HIG exhibited significant reductions in heat release capacity, peak heat release rate, and total heat release. Concurrently, their PHRR temperatures increased to 371 °C and 376 °C, indicating enhanced fire safety.

3.6 Flame-retardant mechanism of HIG

Fig. 8a and b show the TG and DTG curves of Wood, n-HIG, p-HIG, Wood@n-HIG and Wood@p-HIG under air atmosphere. Comparative analysis revealed that HIG coating significantly enhanced the thermal stability of wood. The residual mass of wood at 800 °C was 7.2 wt%, exhibiting a maximum mass loss rate of 28.2 wt% min⁻¹ at 345.3 °C. In contrast, HIG-coated wood composites (Wood@n-HIG and Wood@p-HIG) demonstrated markedly improved thermal resistance, retaining approximately 16.5 wt% residual mass at 800 °C and achieving a reduced peak mass loss rate of 8.6 wt% min⁻¹ at 297.0 °C. Critically, no detectable mass loss occurred near 345.3 °C for the HIG-coated systems, indicating a marked enhancement in the thermal stability of wood.

The microstructural evolution of char layers formed during vertical combustion of Wood and Wood@p-HIG was investigated by SEM, as shown in Fig. 8g. Wood developed a loose, porous char layer after combustion (Fig. 8g(I and II)), facilitating oxygen infiltration and heat transfer to accelerate the internal thermal degradation and flammable volatile release. In contrast, Wood@p-HIG formed a continuous dense char layer with uniformly distributed micropores (Fig. 8g(III and IV)), acting as an effective thermal/flammable gas barrier to suppress combustion propagation.^36,37

The XPS analysis for the char layers of n-HIG and p-HIG is displayed in Fig. 8c–f. The retention values of flame-retardant elements in the char layer for p-HIG are as follows: 14.49% N, 1.40% Cl, and 1.09% S (Fig. 8d). Comparative C 1s spectral analysis (Fig. 8e and f) demonstrated that the C–C/CC content of the char layer increased from 61.3% to 69.0%, while the C–S and C–N contents decreased from 19.6% and 9.2% to 18.6% and 4.1%, alongside newly formed –O–CO groups after combustion. These chemical transformations indicated that more thermally stable conjugated structures were formed in the p-HIG char layer, and flame-retardant substances such as NH₃ can exert a flame-retardant effect in the gas phase.³⁸

To further elucidate the flame-retardant mechanism of HIG, thermogravimetric-Fourier transform infrared (TG-FTIR) spectroscopy was performed to analyze the thermal degradation volatiles from Wood, n-HIG, and p-HIG (Fig. 9). The thermal degradation volatiles of Wood included more flammable gases such as alkanes, CO and carbonyl compounds, which fuel combustion reactions.³⁹ In contrast, the signal intensity of H₂O generated by the n-HIG and p-HIG during combustion was about 6–8 times that of Wood, and the signal intensity of flame-retardant gases, such as NH₃, was close to 2–3 times (Fig. 8h and i), while the signal intensity of flammable gases generated was much lower. During initial pyrolysis, HIG released substantial water vapor, effectively suppressing substrate temperature rise via latent heat of vaporization.^40,41 As the combustion continued, the acylamino group in n-HIG and p-HIG started to decompose, releasing NH₃ at about 280 °C, diluting the combustible gases and oxygen,^42,43 reducing the combustion reaction in the gas phase.

In summary, HIG demonstrated exceptional flame retardancy through synergistic effects involving both gas-phase dilution and condensed-phase barrier formation (mechanism schematic is shown in Fig. 10). During initial combustion, HIG rapidly released water vapor to dilute flammable gases and suppress temperature rise via endothermic vaporization. When the temperature increased, the decomposition of –NH₂ groups in HIG generated non-flammable NH₃, further reducing oxygen availability and combustible gas concentrations to retard flame propagation. In addition, HIG's intrinsic porous network structure and gas-release behaviour promoted the formation of abundant microcavities during pyrolysis, ultimately yielding a uniform, dense expanded char layer. This hierarchical char barrier effectively blocked heat/oxygen transfer while inhibiting the thermal degradation of the gel matrix or protected substrates. This synergistic effect realized a flame-retardant effect in the gas phase-condensed phase and significantly improved the fire safety of flammable materials as a fire protection layer.

3.7 Thermal capacitance properties

HIG demonstrated exceptional ionic thermoelectric properties, positioning it as a multifunctional material for temperature sensing, fire warning systems, and low-grade heat harvesting. As illustrated in Fig. 11a, ionic thermocapacitors were constructed by assembling HIG with copper foil (Cu) electrodes, enabling thermoelectric conversion and energy storage via a four-stage capacitive cycle: thermal gradient-driven charging was the first stage. Under a temperature gradient, directional ion migration occurred within the thermocapacitor. Faster Na⁺ migration relative to Cl⁻ in p-HIG induced positive charge accumulation at the cold end, generating a forward thermovoltage. In the second stage, connecting the external circuit triggered electron transfer via the electric double-layer (EDL) effect between HIG and Cu, achieving charge equilibration and power output, and this stage was named the energy output. The third stage was about temperature gradient removal; removing the thermal gradient while disconnecting the circuit allowed ionic redistribution, creating a reverse voltage. The final stage was reverse discharging, reconnecting the load allowed electrons to migrate again, thereby delivering additional electrical energy to the external circuit. Fig. 11b–d confirm that the n-HIG, p-HIG, and p–n junction thermocapacitors followed identical operational logic: thermovoltage generation during charging, electron migration across electrodes, voltage decay during discharging, and reverse charging for energy delivery. Under a 5 K temperature gradient, all thermocapacitors exhibited slightly lower open-circuit voltages than their standalone HIG counterparts, exhibiting excellent conversion capabilities. The p–n junction thermocapacitor achieved a thermovoltage of 18.6 mV (5 K ΔT), accounting for 92% of the summed outputs from n-HIG and p-HIG units. This aligns with the theoretical relationship |S_p–n| ≈ 0.914 × (|S_n| + |S_p|), demonstrating rational energy conversion synergy.

Furthermore, as shown in Fig. S10, the open-circuit voltage and maximum output power of the n-HIG, p-HIG, and p–n junction thermocapacitors increased significantly with rising temperature gradients, indicating that a larger temperature difference enhanced the conversion of thermal energy into electrical energy. At a temperature gradient of 20 K, the maximum output power of the p-HIG thermocapacitor reached 109.0 nW, approximately 2.7 times that of the n-HIG thermocapacitor. This enhancement originated from two synergistic advantages of p-HIG: higher thermovoltage generation (2.14 mV K⁻¹) and superior mobility of mobile Na⁺ ions compared to SO₃⁻ in n-HIG, enabling a stronger current output. Additionally, the p–n junction thermocapacitor achieved a thermovoltage of −66.3 mV and a peak power of 157.4 nW, corresponding to over 90% of the summed outputs from individual n-HIG and p-HIG units. The rationally designed p–n junction architecture synergistically combined the cation-dominant transport of p-HIG with the anion-enhanced thermopower of n-HIG, collectively elevating the thermoelectric conversion efficiency of hydrogel-based systems.

3.8 Electrochemical and thermal charging properties of the HIG SC

The electrochemical impedance spectroscopy (EIS) spectra and ionic conductivity (σ_i) of HIG are presented in Fig. 12b. The p-HIG exhibited superior ionic conductivity (4.95 S m⁻¹) compared to the n-HIG, primarily attributed to its higher population of mobile ions. Both n-HIG and p-HIG demonstrated sufficiently high σ_i valued >3 S m⁻¹, making them suitable for flexible sensing (Fig. S11) and functioning as quasi-solid-state supercapacitors (denoted as n-HIG SC and p-HIG SC), as shown in Fig. 12a. Cyclic voltammetry (CV) curves of n-HIG SC and p-HIG SC were evaluated at scan rates of 10–200 mV s⁻¹ within a 0–0.75 V window (Fig. 12c). Both devices exhibited quasi-rectangular CV curves at low scan rates (10–20 mV s⁻¹), retaining centrosymmetric shapes without distortion at elevated rates, indicating ideal electric double-layer capacitive behavior and high-rate capability. Galvanostatic charge–discharge (GCD) curves further confirmed capacitive characteristics, displaying highly symmetrical triangular profiles. At 0.3 A g⁻¹, n-HIG SC and p-HIG SC achieved specific capacitances of 49.5 F g⁻¹ and 82.4 F g⁻¹, respectively (Fig. 12d). The superior performance of the p-HIG SC was attributed to NaCl-induced elevation in free ion concentration, and abundant free ions penetrated and transferred to facilitate efficient EDL formation at the electrode–electrolyte interfaces. Thus, the number of ions involved in charge exchange during capacitance formation increased, allowing the electrode surface to hold and convert more charge.⁴⁴

Leveraging the exceptional ionic thermoelectric properties of HIG, the HIG SC operated as a multifunctional device, combining quasi-solid-state energy storage with ionic thermoelectric conversion. Given its superior electrochemical and thermoelectric metrics, the p-HIG SC was selected for detailed thermal charging analysis. Its working mechanism aligned with the HIG thermocapacitor, following four key stages (Fig. 13b):^12,45 Stage I was the thermal diffusion of Na⁺. Na⁺ migrated directionally under thermal gradients, generating thermovoltage (S_pΔT, where S_p is the ionic Seebeck coefficient for the p-HIG). Afterwards, the electric charging stage began and external circuit connection triggered electron transfer, establishing EDLs. Stage III was about thermal equilibrium; the thermal gradient removal induced reverse voltage via ionic redistribution. Finally, stored charges discharge through external loads. Equivalent circuit analysis (Fig. 13c) modelled this process as a conductive pathway powered by S_pΔT, with simulated charge–discharge curve (the green curve in Fig. 13c), closely matching experimental data. In particular, the p-HIG SC exhibited an initial equilibration period prior to stable thermovoltage generation, corresponding to dynamic balancing between ionic migration and interfacial charge transfer.⁴⁶

As evidenced by the thermal charging curves in Fig. 13a and S12, the p-HIG SC demonstrated exceptional thermovoltage stability and high ionic thermoelectric efficiency across a broad temperature range (0–40 K). The maximum thermovoltage exhibited progressive enhancement with linearly increasing temperature gradients, maintaining strict linear proportionality (R² > 0.99). This was a sign of interfacial charge-transfer reliability. The linear response not only validated HIG's operational robustness under variable ambient conditions but also confirmed enhanced ionic thermoelectric efficiency with escalating thermal stimuli. Remarkably, the p-HIG SC sustained stable high-voltage outputs even under extreme thermal loads (40 K ΔT), a breakthrough performance enabling next-generation flexible energy devices for harsh-environment applications.

4. Conclusions

In this work, a high-thermopower ionic thermoelectric hydrogel system for low-grade heat harvesting and intelligent fire protection was prepared by establishing the HIG p–n junctions. The thermoelectric polarity of HIG was reversibly modulated from n-type to p-type through controlled NaCl doping within identical hydrogel matrices. The Seebeck coefficients of n-HIG and p-HIG reached −1.95 mV K⁻¹ and 2.14 mV K⁻¹, respectively, while the p–n junction exhibited a much higher Seebeck coefficient of 3.74 mV K⁻¹. When burned, HIG rapidly generated a voltage exceeding 50 mV and triggered a fire alarm within 4 s, demonstrating excellent thermoelectric efficiency and sensitive early fire-warning capability. Moreover, HIG exhibited excellent flame retardancy. Even burned on a propane torch for 20 s, HIG could not be ignited. The limiting oxygen index of wood coated with HIG (Wood@HIG) was higher than 90%, and it rapidly self-extinguished after the flame removal. HIG also demonstrated good electrochemical performance and remarkable thermal charging energy storage capabilities, showing great potential for applications in supercapacitors. This work provides new insights into the preparation and application of high fire-safety integrated thermoelectric hydrogel systems for energy conversion and storage.

Conflicts of interest

There are no conflicts to declare.

Data availability

Data are available from the authors on request. The data that support the findings of this study are available from the corresponding author, Xuejun Lai, upon reasonable request.

The formulation table of HIG; detailed testing procedures; additional material characterization and testing data (FTIR, thermoelectric properties, tensile sensing properties, electrochemical performance, et al.). See DOI: https://doi.org/10.1039/d5ta04718h.

Acknowledgements

We greatly acknowledge the National Natural Science Foundation of China (22478140, 22178134) and Guangdong Basic and Applied Basic Research Foundation (2023A1515010845, 2024A1515010340) for their financial support.

References

1. Z. Li, Y. Xu and X. Zhang, EnergyChem, 2024, 6, 100136 CrossRef CAS .
2. W. Yang, J. Bao, H. Liu, J. Zhang and L. Guo, Renewable Sustainable Energy Rev., 2023, 188, 113842 CrossRef CAS .
3. Y. Zhou, Z. Dong, Y. He, W. Zhu, Y. Yuan, H. Zeng, C. Li, S. Chen and K. Sun, Chem.–Asian J., 2022, 17, e202200850 CrossRef CAS PubMed .
4. M. Hong and Z. Chen, Acc. Chem. Res., 2022, 55, 3178–3190 CrossRef CAS PubMed .
5. I. López Tobi, M. Comamala, L. Montoro, J. R. González, J. Czarnigowski and A. Gómez, Energy, 2024, 313, 133853 CrossRef .
6. S. Jia, H. Ma, S. Gao, L. Yang and Q. Sun, Small, 2024, 20, 2405019 CrossRef CAS PubMed .
7. X. Yang, Z. Zhang, S. A. Khan, L. Sun, Z. Wang, X. Cui, Z. Huang and H. Zhang, J. Mater. Chem. C, 2024, 12, 3298–3305 RSC .
8. M. Saleem, F. Ullah, M. T. Qureshi, R. S. Abdel Hameed, M. Abdallah, O. Farghaly, M. Othman and A. Atta, J. Alloys Compd., 2022, 921, 166175 CrossRef CAS .
9. Y. Yu, X. Xu, M. Bosman, K. Nielsch and J. He, Nat. Rev. Electr. Eng., 2024, 1, 109–123 CrossRef .
10. S. Li, Y. Xu, Z. Li, S. Zhang, H. Dou and X. Zhang, J. Mater. Chem. A, 2025, 13, 3913–3921 RSC .
11. H. Cheng and J. Ouyang, J. Phys. Chem. Lett., 2022, 13, 10830–10842 CrossRef CAS PubMed .
12. Y. Pai, J. Tang, Y. Zhao and Z. Liang, Adv. Energy Mater., 2023, 13, 2202507 CrossRef CAS .
13. X. Yang, Y. Tian, B. Wu, W. Jia, C. Hou, Q. Zhang, Y. Li and H. Wang, Energy Environ. Mater., 2022, 5, 954–961 CrossRef CAS .
14. Z. Lei, W. Gao and P. Wu, Joule, 2021, 5, 2211–2222 CrossRef .
15. Q. Hu, H. Li, X. Chen, Y. Wang, H. Wu, S. Guo and K. Xu, Adv. Funct. Mater., 2024, 34, 2406968 CrossRef CAS .
16. B. Chen, Q. Chen, S. Xiao, J. Feng, X. Zhang and T. Wang, Sci. Adv., 2021, 7, eabi7233 CrossRef CAS PubMed .
17. N. Chaabene, J. Zhang, M. Turmine, E. Kurchavova, V. Vivier, F. Cuevas, M. Mateos, M. Latroche and J. Monnier, J. Power Sources, 2023, 574, 233176 CrossRef CAS .
18. N. Pereira, L. Afonso, M. Salado, C. R. Tubio, D. M. Correia, C. M. Costa and S. Lanceros-Mendez, Macromol. Rapid Commun., 2024, 45, 2400041 CrossRef CAS PubMed .
19. T. Shimono, M. Matsuki, T. Yamada, M. Morikawa, N. Yasuda, T. Fujigaya and N. Kimizuka, Chem. Lett., 2018, 47, 261–264 CrossRef CAS .
20. J. Shen, X. Huang, Y. Dai, X. Zhang and F. Xia, Nat. Commun., 2024, 15, 9305 CrossRef CAS PubMed .
21. C. Chi, M. An, X. Qi, Y. Li, R. Zhang, G. Liu, C. Lin, H. Huang, H. Dang, B. Demir, Y. Wang, W. Ma, B. Huang and X. Zhang, Nat. Commun., 2022, 13, 221 CrossRef CAS PubMed .
22. Y. Zhao, Q. Zeng, C. Jiang, X. Lai, H. Li, Z. Wu and X. Zeng, Nano Energy, 2023, 116, 108785 CrossRef CAS .
23. M. Fu, Z. Wu, X. Liu, Y. Yuan, X. Lai and K. Yue, J. Mater. Chem. A, 2024, 12, 27588–27597 RSC .
24. W. Qian, S. Jia, P. Yu, K. Li, M. Li, J. Lan, Y. Lin and X. Yang, Mater. Today Phys., 2024, 49, 101589 CrossRef CAS .
25. A. Shokry, M. M. A. Khalil, H. Ibrahim, M. Soliman and S. Ebrahim, Sci. Rep., 2019, 9, 16984 CrossRef PubMed .
26. J. Ma, F. Zeng, X. Lin, Y. Wang, Y. Ma, X. Jia, J. Zhang, B. Liu, Y. Wang and H. Zhao, Science, 2024, 385, 68–74 CrossRef CAS PubMed .
27. Y. Isikver and N. Alkan, Polym. Bull., 2024, 81, 1019–1042 CrossRef CAS .
28. N. Li, G. Chen, W. Chen, J. Huang, J. Tian, X. Wan, M. He and H. Zhang, Carbohydr. Polym., 2017, 178, 159–165 CrossRef CAS PubMed .
29. C. Li, Y. Zhang, H. Zhao, Z. Yu, J. Zhang, P. Zhang and H. Chen, J. Mater. Chem. A, 2024, 12, 9518–9526 RSC .
30. W. Zhao, Y. Zheng, A. Huang, M. Jiang, L. Wang, Q. Zhang and W. Jiang, Adv. Mater., 2024, 36, 2402386 CrossRef CAS PubMed .
31. K. Takada, N. Ohta and Y. Tateyama, J. Inorg. Organomet. Polym. Mater., 2015, 25, 205–213 CrossRef CAS .
32. J. Li, P. Zhang, H. He and B. Shi, Mater. Des., 2020, 187, 108373 CrossRef CAS .
33. C. Han, X. Qian, Q. Li, B. Deng, Y. Zhu, Z. Han, W. Zhang, W. Wang, S. Feng, G. Chen and W. Liu, Science, 2020, 368, 1091–1098 CrossRef CAS PubMed .
34. Z. Zou, N. Ma, A. Wang, Y. Ran, T. Song, Y. Jiao, J. Liu, H. Zhou, W. Shi, B. He, D. Wang, Y. Li, M. Avdeev and S. Shi, Adv. Energy Mater., 2020, 10, 2001486 CrossRef CAS .
35. C. Jiang, X. Lai, Z. Wu, H. Li, X. Zeng, Y. Zhao, Q. Zeng, J. Gao and Y. Zhu, J. Mater. Chem. A, 2022, 10, 21368–21378 RSC .
36. W. Gan, C. Chen, Z. Wang, J. Song, Y. Kuang, S. He, R. Mi, P. B. Sunderland and L. Hu, Adv. Funct. Mater., 2019, 29, 1807444 CrossRef .
37. B. Bheemisetty and S. A. Lewis, J. Appl. Pharm. Sci., 2024, 14, 51–60 CAS .
38. Y. Xu, W. Zhang, Y. Qiu, M. Xu, B. Li and L. Liu, Composites, Part B, 2022, 238, 109913 CrossRef CAS .
39. T. Takaoka, R. Nakamura and H. Habu, Fuel, 2025, 392, 134645 CrossRef CAS .
40. W. Yuan, X. Huang, J. Fu, Y. Ma, G. Li and Q. Huang, Energies, 2022, 15, 9055 CrossRef CAS .
41. J. Liu, Z. Yu, H. He, Y. Wang and Y. Zhao, Cellulose, 2021, 28, 8751–8769 CrossRef CAS .
42. L. Liu, J. Feng, Y. Xue, V. Chevali, Y. Zhang, Y. Shi, L. Tang and P. Song, Adv. Funct. Mater., 2023, 33, 2212124 CrossRef CAS .
43. Q. Zeng, X. Lai, H. Li, Z. Chen, X. Zeng and L. Zhang, J. Mater. Chem. A, 2024, 12, 1642–1653 RSC .
44. A. R. Finney and M. Salvalaglio, Faraday Discuss., 2024, 249, 334–362 RSC .
45. D. Kim, Z. A. Akbar, Y. T. Malik, J. Jeon and S. Jang, Nat. Commun., 2023, 14, 3246 CrossRef CAS PubMed .
46. L. Du Hill, M. De Keersmaecker, A. E. Colbert, J. W. Hill, D. Placencia, J. E. Boercker, N. R. Armstrong and E. L. Ratcliff, Mater. Horiz., 2022, 9, 471–481 RSC .

---

Footnote

† These authors contributed equally to this work and should be considered co-first authors.

This journal is © The Royal Society of Chemistry 2025