Highly Stable Quasi‐Solid Thermocells for Continuous Power Generation Across a Broad Humidity Range

ABSTRACT

Ionic hydrogel‐based quasi‐solid (IH‐QS) thermocells are promising for low‐grade heat harvesting, but suffer from humidity‐induced instability (i.e., severe dehydration at low humidity and excessive swelling at high humidity), degrading their mechanical and thermoelectric performances. Herein, we develop humidity‐stable IH‐QS thermocells via the incorporation of amphiphilic networks and hygroscopic electrolytes. This design ensures operational stability across a wide relative humidity (RH) by synergistically combining hydrophilic ─SO₃ ⁻ groups and hygroscopic electrolytes for high water retention at low humidity and hydrophobic phenyl moieties that form physical crosslinks to suppress swelling at high humidity. Furthermore, the ─SO₃ ⁻ and phenyl moieties of amphiphilic networks enhance the entropy difference of the [Fe(CN)₆ ⁴⁻/Fe(CN)₆ ³⁻] redox couple via electrostatic and ion‐π interactions, boosting thermopower. The resulting IH‐QS thermocell achieves high thermopower and normalized power density across 30%–99% RH (e.g., 2.17 mV K⁻¹ and 0.19 mW m⁻² K⁻² at 30% RH; 1.63 mV K⁻¹ and 0.55 mW m⁻² K⁻² at 99% RH), overcoming a critical barrier to their practical deployment.

Benefiting from the built‐in pressure‐sensitive transmembrane potential across the gradient polyelectrolyte conductors, the construction of gradient polycation/polyanion heterojunction can vastly boost the output voltage, power density, and sensing performance of ionic diode‐based generators. The resulting gradient ionic diode show great potential toward various applications such as wearable power supplies, all‐ion‐based DC power system, and pressure sensors.

1. Introduction

The ever‐increasing emphasis on energy conservation and carbon neutrality has pushed researchers to develop efficient low‐grade heat utilization technologies, given that a substantial fraction of global wasted energy exists in the form of low‐grade thermal energy [1, 2, 3]. Over the past decades, diverse thermoelectric conversion systems, ranging from conventional electronic thermoelectric generators [4, 5, 6, 7, 8] to emerging ionic thermocells [9, 10], have been extensively explored for low‐grade thermal energy utilization. Notably, ionic hydrogel‐based quasi‐solid (IH‐QS) thermocells have emerged as particularly promising candidates, combining the exceptional thermopower (Seebeck coefficient) of liquid‐phase ionic thermocells with improved structural stability that mitigates electrolyte leakage issues [9, 10, 11, 12, 13]. These advantages position IH‐QS thermocells as potential sustainable power sources for the continuous operation of wearable electronics and Internet of Things (IoT) devices in the future.

The thermoelectric and mechanical properties of IH‐QS thermocells are intrinsically governed by their ionic hydrogel‐based thermoelectric (IH‐TE) materials, which consist of crosslinked polymer networks swollen with thermogalvanic ionic solutions [3, 9, 13]. Recent advances in the IH‐TE material design [9, 10, 11, 12, 13, 14], including the development of double‐network architectures [9], oriented structures [14], and synergistic molecular‐structural engineering approaches [11, 12, 14], have remarkably enhanced mechanical robustness, thermopower, and Carnot‐relative efficiency (η_r ). Nevertheless, conventional IH‐QS thermocells exhibit significant limitations in their water regulation capacity under varying humidity conditions [11, 15, 16]. Specifically, they exhibit severe dehydration‐induced performance degradation (reduced flexibility, ionic conductivity, and power output) at low humidity, while suffering from substantial deterioration of mechanical and thermoelectric properties due to excessive swelling at high humidity [15, 16, 17, 18, 19]. Although material encapsulation can effectively stabilize water content in quasi‐solid thermocells, the encapsulation strategy is subject to the decreased heat transfer efficiency, complex encapsulation process, and poor electrode/encapsulating material interface [11, 20, 21].

Current strategies to mitigate low‐humidity dehydration in ionic hydrogels primarily focus on enhancing water absorption capacity through the incorporation of hygroscopic salts [11, 22] and non‐volatile hydrophilic additives (e.g., glycerol, ionic liquids, deep eutectic solvents) [15]. However, these modified hydrogels still exhibit considerable water content fluctuations in response to ambient humidity changes [15]. This instability arises from the pronounced reduction in physical crosslinking density within conventional hydrogel networks as water content increases, resulting from weakened inter‐chain entanglements and diminished supramolecular interactions (e.g., hydrogen bonding and electrostatic interactions, Figure 1a) [23, 24, 25]. Consequently, water absorption at high humidity reduces inter‐chain crosslinking, which lowers water chemical potential and promotes excessive swelling of hydrogels [15, 17]. Conversely, water loss at low humidity increases crosslinking density and raises water potential, thereby accelerating dehydration of hydrogels [15]. This behavior creates a performance paradox: although enhanced water absorption facilitates stable operation under dry conditions, it exacerbates swelling under high humidity, leading to significant deterioration in both mechanical and thermoelectric properties in humid environments [15, 23, 24, 25]. The challenge of maintaining appropriate and stable water content in the IH‐TE materials represents a major obstacle to achieving high‐efficiency thermoelectric conversion of IH‐QS thermocells in real‐world environments with fluctuating humidity [11, 26, 27].

FIGURE 1.

Design principle of the P(SPAK‐co‐PEA)_DES IH‐TE materials. (a) Schematic illustration comparing the variations in swelling ratio and physical crosslinking density between conventional IH‐TE materials and the developed IH‐TE materials as a function of RH. (b) Schematic illustration of P(SPAK‐co‐PEA)_DES amphiphilic networks has stronger intermolecular interaction with Fe(CN)₆ ⁴⁻ than with Fe(CN)₆ ³⁻. (c) Water contents of different IH‐TE materials as a function of RH. (d) Stress–strain curves of P(SPAK‐co‐PEA) IH‐TE materials under different RH. (e) Comparison of mechanical properties between our P(SPAK‐co‐PEA) IH‐TE materials and conventional IH‐TE materials. (f) Thermopower (Seebeck coefficient, S_e) of different IH‐TE materials.

In this work, we develop a strong IH‐QS thermocell by integrating amphiphilic polymer networks with hygroscopic quaternary electrolyte solution, enabling efficient and stable low‐grade heat harvesting across a broad humidity range (30%–99% RH). The electrolyte consists of a Fe(CN)₆ ⁴⁻/Fe(CN)₆ ³⁻ redox couple, a deep eutectic solvent (DES), LiCl, and water, which synergize with the amphiphilic network to provide three essential functionalities (Figure 1a): (i) enhanced water retention facilitated by hydrophilic ─SO₃ ⁻ groups and hygroscopic electrolytes, ensuring sustained ionic conductivity and redox activity at low humidity; (ii) increased resistance to water adsorption through hydrophobic phenyl‐mediated physical crosslinking, which mitigates excessive swelling and prevents associated mechanical and thermoelectric degradation at high humidity; (iii) entropy‐driven thermopower amplification by selective intermolecular interactions with the Fe(CN)₆ ⁴⁻/Fe(CN)₆ ³⁻ redox pair [3, 11, 12, 13, 14]. As a result, the fabricated thermocells can maintain an optimal water content range (33.6–60.3 wt.%) and deliver outstanding mechanical and thermoelectric performance across the entire tested humidity range (30%–99% RH). The devices achieve a high thermopower of 2.17 mV K⁻¹ and output power density of 0.19 mW m⁻² K⁻² at 30% RH, while retaining competitive performance (1.63 mV K⁻¹ and 0.55 mW m⁻² K⁻²) at 99% RH. Practical viability is demonstrated by successfully powering wearable electronics (e.g., LEDs and watches) under varying ambient humidity conditions, overcoming a major obstacle to the real‐world application of IH‐QS thermocells.

2. Results and Discussion

2.1. Design Principle

The ionic hydrogel thermoelectric (IH‐TE) material comprises a P(SPAK‐co‐PEA) amphiphilic copolymer network and a quaternary electrolyte system containing Fe(CN)₆ ⁴⁻/Fe(CN)₆ ³⁻ redox couple, LiCl, DES (choline chloride/glycerol = 1/2), and water (Figure 1a). The amphiphilic Poly(3‐sulfopropyl potassium acrylate‐co‐2‐phenoxyethyl acrylate) (P(SPAK‐co‐PEA)) network exhibits three key functional characteristics: (1) The hydrophilic ─SO₃ ⁻ groups enable efficient water absorption under low‐humidity conditions; (2) The hydrophobic phenyl groups induce microphase separation via π‐π stacking when humidity rises, increasing physical crosslinking density [28, 29, 30, 31]. This in turn elevates the chemical potential of water within the amphipathic polymer networks, and hinders the uptake of excessive waters at high humidity, thereby inhibiting the degradation of mechanical and thermoelectric conversion performances of resulting IH‐TE materials; (3) Both hydrophilic ─SO₃ ⁻ groups and hydrophobic phenyl groups synergistically increase the redox entropy difference of [Fe(CN)₆ ⁴⁻/Fe(CN)₆ ³⁻] through differential electrostatic and ion‐π interactions, thereby enhancing thermopower (Figure 1b) [9, 12, 13, 14, 32]. Besides, the quaternary electrolyte system contributes additional functionality: the hydrophilic LiCl and DES components enhance water absorption capacity, while the high DES content mitigates excessive volume shrinkage under low‐humidity conditions, beneficial to maintain adequate interchain spacing for suppressing the formation of physical crosslinks.

Collectively, the resulting IH‐TE materials demonstrate three key synergistic advantages: (1) exceptional hygroscopicity at low humidity, enabled by abundant hydrophilic moieties and a low crosslinking density; (2) self‐limiting water uptake at high humidity, mediated by dynamically enhanced crosslinking [28, 29, 30, 31]; (3) high thermopower generation, driven by the amphiphilic P(SPAK‐co‐PEA) network‐induced entropy difference of Fe(CN)₆ ⁴⁻ and Fe(CN)₆ ³⁻ (Figure 1a,b) [9, 12, 13, 14, 32]. Consequently, the resulting IH‐TE materials are expected to maintain stable hydration levels across a wide humidity range (30%–99%), ensuring robust mechanical integrity, consistent conductivity, and efficient thermoelectric conversion under varying environmental conditions.

The fabrication process of humidity‐stable P(SPAK‐co‐PEA)_DES IH‐TE material is depicted in Figure S1. The P(SPAK‐co‐PEA) amphipathic networks were prepared by copolymerizing the potassium 3‐sulphonatopropyl acrylate (SPAK) with ethylene glycol phenyl ether acrylate (PEA) within the dimethyl sulfoxide (DMSO) solvent, in which the total mass fraction of SPAK and PEA was fixed at 30 wt.%. After the solvent replacement in the distilled water for 24 h and the [Fe(CN)₆ ⁴⁻/Fe(CN)₆ ³⁻]‐LiCl‐DES‐water quaternary electrolyte system for 48 h, the humidity‐stable P(SPAK‐co‐PEA)_DES IH‐TE materials were obtained. As a proof of concept, we fabricated a humidity‐stable P(SPAK‐co‐PEA)_DES IH‐TE material with an SPAK/PEA molar ratio of 1/2 by using a quaternary electrolyte solution. The quaternary electrolyte solution was prepared by mixing an ionic water solution containing 0.4 m K₄Fe(CN)₆/K₃Fe(CN)₆ and 25 wt.% LiCl with DES at mass ratios of 10 and 90 wt.%, respectively. Then, we compared the water content under different RH between our P(SPAK‐co‐PEA)_DES IH‐TE material and conventional IH‐TE materials (Figure 1c). Here, five kinds of conventional IH‐TE materials were employed, including the poly(3‐sulfopropyl potassium acrylate‐co‐ acrylamide)_DES (P(SPAK‐co‐AM)_DES) and poly(2‐acrylamido‐2‐methylpropanesulfonic acid‐co‐acrylamide)_DES (P(AMPS‐co‐AM)_DES) IH‐TE materials prepared with the Fe(CN)₆ ⁴⁻/Fe(CN)₆ ³⁻‐LiCl‐DES‐water quaternary electrolyte solution, P(SPAK‐co‐PEA)_H2O, P(SPAK‐co‐AM)_H2O, and P(AMPS‐co‐AM)_H2O IH‐TE materials prepared with the Fe(CN)₆ ⁴⁻/Fe(CN)₆ ³⁻‐LiCl‐water ternary electrolyte solution (see more detailed recipe in Experimental Section). The water content of our P(SPAK‐co‐PEA)_DES IH‐TE material is as high as 33.6 wt.% at a low RH of 30%, which just increases to 60.3 wt.% as the RH increases to 99%. In stark contrast, the conventional P(SPAK‐co‐AM)_DES, P(AMPS‐co‐AM)_DES, P(SPAK‐co‐PEA)_H2O, P(SPAK‐co‐AM)_H2O and P(AMPS‐co‐AM)_H2O IH‐TE materials present much lower water contents of 15.11, 13.86, 4.28, 0.476, and 0.373 wt.% at 30% RH, respectively, which, however, pronouncedly increases to 70.32, 75.3, 87.62, 95.3, and 93.77 wt.%, at 99% RH (Figure 1c). Based on the Flory‐Rehner theory shown in Note S1, we further calculated the theoretical water content of an ideal elastic gel maintaining constant crosslinking density under varying relative humidity (RH) conditions (Figure 1c). Comparative analysis revealed that when the RH was elevated from 30% to 99%, the humidity‐stable P(SPAK‐co‐PEA)_DES IH‐TE material exhibited a significantly lower water content increment than both the ideal elastic gel model and conventional IH‐TE materials. This observation strongly implies an enhancement in physical crosslinking density within the P(SPAK‐co‐PEA)_DES IH‐TE material system under humid conditions.

The water absorption generally has side effects on the practical load‐bearing applications of IH‐TE materials, resulting from its plasticizing effect [33, 34]. The excessive water absorption of conventional P(SPAK‐co‐AM)_DES, P(AMPS‐co‐AM)_DES, P (SPAK‐co‐PEA)_H2O, P(SPAK‐co‐AM)_H2O, and P(AMPS‐co‐AM)_H2O IH‐TE materials under high RH conditions greatly deteriorates their mechanical properties, as reflected by vastly decreased tensile strength and breaking elongation at high RH (Figure S2). Besides, excessive water loss at low RH renders these conventional IH‐TE materials extremely brittle. In this regard, benefiting from the relatively stable water content and increased physical crosslinking density, our P(SPAK‐co‐PEA) IH‐TE material just presents much smaller degradation in mechanical properties at both low and high RH (Figure 1d). It can withstand high tensile stresses of 189 and 46.6 kPa and large strain of 700% and 430% at the RH of 30% and 99%, respectively, far superior to the conventional IH‐TE materials (Figure 1d,e). The outstanding mechanical properties offers the basis for the application of our IH‐TE materials in reality.

Besides the more stable water content and mechanical properties, the P(SPAK‐co‐PEA)‐based thermocells also present higher thermopower compared to the traditional ones (Figure 1f). This demonstrates the amphiphilic P(SPAK‐co‐PEA) network can enlarge the entropy difference of Fe(CN)₆ ⁴⁻ and Fe(CN)₆ ³⁻ [9, 10, 11, 12, 13, 14]. To enhance the output performance of P(SPAK‐co‐PEA)‐based IH‐QS thermocells under varying RH, we optimized the molar ratio of SPAK/PEA and the concentrations of various additives (redox couple, DES, and LiCl) in the quaternary mixture solution (Figures S3–S6). The detailed operational steps are shown in Note S2. The molar ratio of SPAK/PEA was set as 1/2, and in the quaternary electrolyte solution, the mass fractions of the DES and the ionic water solution (i.e., the mixture solution of Fe(CN)₆ ⁴⁻/Fe(CN)₆ ³⁻, LiCl, and water) were fixed at 90 and 10 wt.%, respectively. The concentrations of K₄Fe(CN)₆/K₃Fe(CN)₆ and LiCl in the ionic water solution were fixed at 0.4 m and 25 wt.%, respectively.

2.2. Mechanism Discussion for High Thermopower

The intermolecular interactions between monomer units and the Fe(CN)₆ ⁴⁻/Fe(CN)₆ ³⁻ redox couple were systematically investigated based on UV–vis spectra analysis [9, 10, 11, 12, 13, 14]. Upon introduction of AMPS/AM, SPAK/AM, and SPAK/PEA, the UV absorption bands of Fe(CN)₆ ⁴⁻ exhibited distinct blue shifts, with peak positions shifting from 218 to 215 nm, 214 nm, and 207 nm (Figure 2a). However, the absorption profile of Fe(CN)₆ ³⁻ remained largely unaltered after the addition of these polymeric components (Figure 2b). These observations indicate that the ─SO₃ ⁻ and phenyl functional groups in the amphiphilic P(SPAK‐co‐PEA) networks exhibit significantly stronger intermolecular interactions with the Fe(CN)₆ ⁴⁻‐based clusters than with those formed by Fe(CN)₆ ³⁻ [9, 10, 11, 12, 13, 14]. This conclusion is further corroborated by complementary molecular dynamics (MD) simulations and density functional theory (DFT) calculations (Figure 2c–e). Detailed parameter settings are shown in Notes S3 and S4. In the IH‐TE materials, the Fe(CN)₆ ⁴⁻ are surrounded with higher density of cationic species relative to Fe(CN)₆ ³⁻, stemming from stronger electrostatic interaction between cations and Fe(CN)₆ ⁴⁻ (Figure 2c,d). Further quantitative analysis of interaction energies demonstrates that Fe(CN)₆ ⁴⁻‐based clusters presents stronger electrostatic and cation‐π interactions with SPAK and PEA moieties, exhibiting binding energies of −28.28 and −17.79 kcal/mol, respectively (Figure 2e and Figure S7). These values are markedly lower than those computed for Fe(CN)₆ ³⁻‐based clusters with SPAK (−25.86 kcal mol⁻¹) and PEA (−4.71 kcal mol⁻¹). This selective affinity amplifies the differential mobility (and consequently, the entropy differential) between the two redox species, thereby providing a mechanistic basis for the enhanced thermopower observed in the P(SPAK‐co‐PEA) IH‐TE materials [3, 11].

FIGURE 2.

Intermolecular interaction between P(SPAK‐co‐PEA) amphiphilic networks and Fe(CN)₆ ⁴⁻/Fe(CN)₆ ³⁻ redox couple. (a,b) UV–vis spectra of the (a) K₄Fe(CN)₆ aqueous solution and the (b) K₃Fe(CN)₆ aqueous solution with the addition of AMPS/AM, SPAK/AM, and SPAK/PEA. (c) The particle distribution (or spatial distribution of particles) within the simulated system at the final states of the molecular dynamics (MD) simulation. (d) The interaction between Fe(CN)₆ ³⁻/Fe(CN)₆ ⁴⁻ and different cations contained within the (SPAK‐co‐PEA) IH‐TE materials. (e) The binding energies between Fe(CN)₆ ⁴⁻‐based clusters/Fe(CN) ₆ ³⁻‐based clusters and SPAK/PEA, as determined by DFT calculations.

2.3. Mechanism Discussion for High and Stable Water Content

The water content of IH‐IE materials is highly associated with both their hydrophilicity (hydration power) and crosslinking density (hydration resistance). DFT calculations were conducted for revealing the molecular mechanisms underlying the outstanding water retention capacity at low humidity. The interaction energies for the SPAK/PEA‐water, LiCl‐water, and DES‐water were calculated as −12.7, −21.52, and −15.63 kcal/mol, respectively, all of which are lower than −4.78 kcal/mol for water–water interactions (Figure 3a). This indicates that the P(SPAK‐co‐PEA) IH‐TE material possesses the strong hydrogen bonding interaction with water [13, 35]. Furthermore, the physical crosslinking densities of the IH‐TE materials were quantitatively assessed based on shear modulus measurements obtained via dynamic rheometer (Figure S8). The detailed calculation method is shown in Note S5. The P(SPAK‐co‐PEA)_DES IH‐TE material features the ultralow physical crosslinking density at a low RH of 30% (Figure 3b), which contributes to a reduced chemical potential of water within the material and lowers its resistance to water absorption under low RH. Collectively, the synergistic interplay between diminished crosslinking density and intensified hydrogen‐bonding interactions of the IH‐TE matrix with water molecules endows the material with superior water uptake and retention capabilities at low RH.

FIGURE 3.

Mechanism for high and stable water content. (a) Interaction energies of various components of (SPAK‐co‐PEA)_DES IH‐TE material with water (red, white, grey, blue, pink, yellow, purple, and green spheres represent O, H, C, N, Li, S, K, and Cl atoms, respectively). (b) Physical crosslinking density (α_p ) of different IH‐TE materials as a function of RH. (c,d) The (c) I–q curves and (d) SAXS patterns of P(SPAK‐co‐PEA) IH‐TE material under different RH at 25°C. (e) RH‐dependent scattering intensity at q = 0.05 and 0.25 Å⁻¹. (f) LF ¹H NMR spectra of IH‐TE material under varying RH. (g,h) 2D LF ¹H NMR spectra of IH‐TE material at (g) 30% RH and (h) 90% RH.

It is also worth noting that the physical crosslinking density of our P(SPAK‐co‐PEA)_DES IH‐TE material exhibits a pronounced increase in response to elevated RH (Figure 3b), primarily attributed to enhanced π‐π interactions [30, 31]. This structural reinforcement effectively suppresses excessive water uptake at high RH, thereby ensuring a relatively stable water content across varying humidity conditions. In contrast, conventional IH‐TE materials demonstrate high physical crosslinking densities at low RH, which undergo significant reduction upon humidity increase (Figure 3b and Figure S8). This substantial variation in crosslinking density results in pronounced fluctuations in water content in response to RH changes (Figure 1a).

To investigate the mechanism governing the increased physical crosslinking density at elevated RH, we systematically examined the evolution of intermolecular interactions in P(SPAK‐co‐PEA)_DES IH‐TE materials under progressively increasing environmental humidity using Fourier‐transform infrared (FTIR) spectroscopy (Figure S9) [36]. Spectral analysis revealed a pronounced blueshift in the C═C stretching vibration peak of the benzene ring with increasing RH. This observation provides direct evidence for the enhancement of π‐π stacking interactions within the IH‐TE material matrix at high humidity conditions, which accounts for the corresponding increase in physical crosslinking density [36]. The small‐angle x‐ray scattering (SAXS) was employed to analyze the microstructural evolution of P(SPAK‐co‐PEA)_DES IH‐TE material under the rising environmental humidity at 25°C (Figure 3c–e). As the RH increases, the peak value q_max of scattering intensity I shifts to a higher q region (Figure 3c), and the diffraction circle gradually moves away from the beamstop (Figure 3d). These variations imply the generation of large amounts of phase‐separated microdomains within IH‐TE material at high RH due to the π‐π interactions of phenyl groups, leading to a smaller long period L (i.e., average distance between neighboring microdomains) [36, 37]. According to the Bragg's law, the L can be calculated by L = 2π/q_max [38]. The average spacing of phase‐separated microdomains at 30% RH (L_30% ) is estimated to be 176 nm, and reduces to 107 nm (L_90% ) after the RH rises to 90% (Figure S10). It is worth mentioning that the diffraction circles are buried within a beamstop at the low RH, and thus the L at this case are roughly depicted based on the variation trend [39, 40]. The scattering contributions can be roughly divided into two states: the low‐q state (q < 0.1 Å⁻¹), where scattering is dominated by the large separated phases; and the high‐q region (q > 0.1 Å⁻¹), where scattering is primarily contributed by the local structure of the polymer matrix [36]. As illustrated in Figure 3e, the scattering intensities at q = 0.05 Å⁻¹ and q = 0.25 Å⁻¹ were examined under varying RH conditions. As the RH rises, the scattering intensity in the high‐q region experiences a gradual decline, whereas that in the low‐q region undergoes an incremental increase. This observation also signifies that the rising RH can promote microphase separation of P(SPAK‐co‐PEA) IH‐TE material, thereby enlarging its physical crosslinking density [36, 37].

To gain molecular insights into the microstructure evaluation of IH‐TE under the rising RH, we collected the humidity‐dependent 1D/2D low‐field ¹H NMR spectra (Figure 3f–h). Low‐field ¹H NMR can be employed to monitor the activity changes of hydrogen atoms via spin‐spin relaxation time (T₂) [36]. There exist two T₂ peaks on the 1D ¹H NMR spectra at low RH (Figure 3f). Here, the lower T₂ peak represents the confined protons on the P(SPAK‐co‐PEA) polymer networks and water bonded onto the networks, while the higher one represents the less restricted protons in the electrolyte solution. With the RH increasing, the uptake of water from the surrounding atmospheric environment can enhance the activity of hydrogen atoms, leading to the shift of T₂ peaks to high T₂. Different from the traditional P(AMPS‐co‐AM)‐based IH‐TE materials, the lower T₂ peak of P(SPAK‐co‐PEA)_DES IH‐TE material broadens and split into two peaks at high RH, indicating the phase transition of the P(SPAK‐co‐PEA)_DES IH‐TE material after water absorption (Figure 3f and Figure S11) [41, 42]. Besides, the higher T₂ peak also split into two peaks at high RH. The newly emerged lower T₂ peak stems from the confined electrolyte solution in the phase‐separated domains. 2D LF‐¹H NMR T₁‐T₂ spectra, where T₁ represents the spin‐lattice relaxation time, are used to provide more precise microstructure evaluation information of IH‐TE material (Figure 3g,h). Here, a lower T₁/T₂ ratio means a higher mobility (the diagonal line with equal T₁ and T₂ denotes a completely mobile liquid state [41, 42]. There exist two species in the 2D T₁‐T₂ plot of the P(SPAK‐co‐PEA)_DES IH‐TE material at 30% RH (Figure 3g). The species 1 and 2 are related to the protons on the polymer networks and the electrolyte solution, respectively. However, four species can be observed in the 2D T₁‐T₂ plot of IH‐TE material after the water absorption at 90% RH (Figure 3h). Compared with the 2D T₁‐T₂ plot at 30% RH, the newly emerged species 3 with much higher T₁/T₂ ratio is contributed by the severely confined protons of polymer networks at phase‐separated domains, and species 4 with a lower T₁/T₂ is attributed to the protons from the confined electrolyte solution in the phase‐separated domains. These observations further substantiate the occurrence of microphase separation of P(SPAK‐co‐PEA) amphipathic polymer networks at high RH, accounting for the considerably increased physical crosslinking density in this case.

2.4. Thermoelectric Performance Under the Varying RH

The IH‐QS thermocells were readily constructed by placing the P(SPAK‐co‐PEA) IH‐TE material between two platinum strip electrodes (Figure 4a). When the two platinum electrodes are connected to the hot source and cold source, respectively, a reversible redox reaction Fe(CN)₆ ³⁻ + e⁻ ↔ Fe(CN)₆ ⁴⁻ occurs (Figure 4b) [9, 10, 11, 12, 13, 14]. At the hot side, ions in the reduced state (Fe(CN)₆ ⁴⁻) prefer to oxidize, accompanied by electron transfer to the hot electrode. Whereas at the cold side, ions in the oxidized state (Fe(CN)₆ ³⁻) prefer to reduce, along with the electron capture from the cold electrode. In the meantime, the Fe(CN)₆ ⁴⁻ and Fe(CN)₆ ³⁻ ions could continuously move toward the hot side and cold side, respectively, through diffusion, convection, and migration [9, 10, 11, 12, 13, 14]. Such a sustainable redox reaction enables the continuous transfer of electrons from the hot to the cold side and unremitting power output between the two electrodes.

FIGURE 4.

Output performance of (SPAK‐co‐PEA)_DES IH‐TE material in different environments. (a) Digital image of the platform for thermopower measurements. The cross‐sectional area of the IH‐TE material is 0.2 cm², and the distance between the two electrodes d is 1.1 cm. (b) Schematic illustration of the IH‐TE material added with the redox couple [Fe(CN)₆ ⁴⁻/Fe(CN)₆ ³⁻] located in a temperature gradient field. (c) Schematic of the platform for thermopower measurements. (d) Thermopower of P(SPAK‐co‐PEA)_DES IH‐TE materials and conventional IH‐TE materials in different RH. (e) Effective ionic conductivity of P(AMPS‐co‐AM)_DES IH‐TE materials, P(SPAK‐co‐AM)_DES IH‐TE materials, and P(SPAK‐co‐PEA) IH‐TE materials under varying humidity levels. (f) Output power densities of the P(SPAK‐co‐PEA) IH‐TE materials as a function of load resistance when RH changes from 30% to 99%. (g) Output power of the P(SPAK‐co‐PEA) IH‐TE materials as a function of load resistance when ΔT changes from 10°C to 40°C at 90% RH. (h) The P_max of different IH‐TE materials under a 10°C temperature difference at different RH. (i) The P_max /(ΔT)² of different IH‐TE materials under a 10°C temperature difference at different RH.

Figure 4d shows the thermopowers of different IH‐TE materials under the varying RH. The P(SPAK‐co‐PEA)_DES IH‐QS thermocell presents an ultrahigh thermopower of 2.17 mV K⁻¹ at 30% RH, which undergoes a slight decrease with increasing the RH. This may stem from the increased water content and decreased interaction between the polymer network and redox couple [Fe(CN)₆ ⁴⁻/Fe(CN)₆ ³⁻]. In contrast, the traditional P(SPAK‐co‐PEA)_H2O, P(SPAK‐co‐AM)_H2O, and P(AMPS‐co‐AM)_H2O IH‐QS thermocells can even not operate stably and present extremely small thermopower at low RH of 30%–50%, arising from the extremely low water content within these IH‐TE materials. Although the P(AMPS‐co‐AM)_DES and P(SPAK‐co‐AM)_DES IH‐QS thermocells can provide stable power output at low RH, they suffer from the low thermopower under all humidity conditions (30%–99%), as well as low ionic conductivity at low RH (Figure 4d,e).

The material maintains a stable thermoelectric voltage output under a sustained temperature gradient (Figure S12a). At 40% RH, with a series external resistance of 500 Ω and a temperature difference of 10 K, a peak output power density of 24.7 mW m⁻² is achieved. As the discharge progresses, the power density gradually decays and eventually stabilizes at a steady‐state value (Figure S12b). This gradual decline in power density is primarily attributed to electrode polarization occurring during continuous power generation and consumption of the thermal diffusion potential [3]. The optimal thermoelectric output can be fully restored by re‐equilibrating the material in the quaternary electrolyte solution [3]. Then we tested the output power density (P _out) connected to an external electronic circuit with different load resistance (R _load) in different RH. As the R _load increases, the current decreases, while the P _out increases first and then decreases (Figure 4f). The P _out reaches its maximum value (P _max) when the R _load is equal to the internal resistance (R_in ), aligning with the prediction of P _max = V ²/4R_in [9]. The P _max of P(SPAK‐co‐PEA) IH‐QS thermocell increases from 18.54 to 54.81 mW m⁻² as the RH increases from 30% to 99%, stemming from the increase in the ionic conductivity (Figure 4f). When the temperature difference increases to 40 K at 90% RH, the maximum power density rises to 856.74 mW m⁻² (Figure 4g). In this respect, limited by extremely low ionic conductivity and small thermopower (Figure 4d,e), the conventional P(SPAK‐co‐AM)_DES, P(AMPS‐co‐AM)_DES, P(SPAK‐co‐PEA)_H2O, P(SPAK‐co‐AM)_H2O, and P(AMPS‐co‐AM)_H2O IH‐QS thermocells exhibit extremely low P _max at low RH (Figure 4h). Additionally, the P _max of P(SPAK‐co‐PEA) IH‐QS thermocell is nearly proportional to ΔT ², consistent with the theoretical analysis (Figure S13) [9, 10, 13, 14]. Its normalized power density (P_max /ΔT ²) is as large as 0.54 mW m⁻² K⁻² at 99% RH (Figure 4i), superior to most of the previously reported advanced IH‐QS thermocells (Table S1) [9, 13, 43, 44, 45, 46]. Even at a low RH of 30%, the normalized power density of P(SPAK‐co‐PEA)_DES IH‐QS thermocell can reach up to 0.19 mW m⁻² K⁻². Besides, the normalized power densities of P(SPAK‐co‐AM)_DES, P(AMPS‐co‐AM)_DES, P(SPAK‐co‐PEA)_H2O, P(SPAK‐co‐AM)_H2O, and P(AMPS‐co‐AM)_H2O IH‐QS thermocells are also much lower than our P(SPAK‐co‐PEA) IH‐QS thermocells at high RH, since the excessively water absorption vastly reduces the interaction between the polymer network and redox couple [Fe(CN)₆ ⁴⁻/Fe(CN)₆ ³⁻], and hence degrades the thermopowers of traditional IH‐TE materials (Figure 4h,i).

The P(SPAK‐co‐PEA)_DES IH‐TE material is thermally insulated. Its thermal conductivity increases slightly from 0.209 to 0.362 W m⁻¹ K⁻¹, as the RH increase from 30% to 99% (Figure S14). Benefiting from the intrinsically low thermal conductivity and the high Seebeck coefficient of the P(SPAK‐co‐PEA)_DES IH‐TE material, the thermoelectric figure of merit (ZT) values are higher than 0.002 across all RH levels, which reaches a maximum of 0.0032 at 90% RH (Figure S15 and Note S6). We also assess the output thermoelectric performance of the P(SPAK‐co‐PEA) IH‐QS thermocell by figuring out the energy conversion efficiency (η) and the Carnot‐relative efficiency (η_r ) under different RH following the method described in Note S7. Even at a low RH of 30%, the η and η_r of P(SPAK‐co‐PEA)_DES IH‐QS thermocell can reach up to 0.00177% and 0.054%, respectively (Figure S16). The η and η_r further increase to 0.00465% and 0.141% at 90% RH, comparable to previously reported ones [13].

To assess the stability of the P(SPAK‐co‐PEA) IH‐QS thermocell under real‐world conditions (∼50% RH and 25°C), the thermocell was subjected to compression, stretching, and 90° bending under a temperature difference of ∼10°C. The voltage and current changes in response to external stimuli were recorded using a digital source meter (Figure S17). The P(SPAK‐co‐PEA) IH‐QS thermocell maintained stable voltage and current output in response to the compression and bending. To evaluate mechanical durability and operational reliability, the output stability of the P(SPAK‐co‐PEA)_DES IH‐QS thermocell was examined during repeated bending and stretching cycles. Prior to the tests, the thermocell was connected to an external load (500 Ω) under a fixed temperature gradient (10 K). The thermocell maintained a stable current output throughout approximately 100 bending and stretching cycles (Figure S18). This consistent performance under mechanical perturbation indicates favorable mechanical durability and robust operational reliability of the IH‐QS thermocell. When subjected to stretching strain, its output current slightly decreases because of the resistance increase, and the voltage remains stable. The current immediately recovers after removing stretching strain. Therefore, the quasi‐solid thermocell can provide stable and satisfying power output under various environmental stimuli, demonstrating its adaptability to complex environments. The P(SPAK‐co‐PEA)‐based IH‐QS thermocells maintain relatively high and stable Seebeck coefficients and tensile strength after storage for 48–120 h under conditions ranging from 30% to 99% RH (Figure S19). Furthermore, in real‐world applications, ambient humidity is constantly changing. To evaluate performance under such conditions, we conducted a cyclic humidity fluctuation test alternating between low and high humidity. The low‐humidity condition was set to 30% RH and the high‐humidity condition to 90% RH. After equilibrating for 12 h at each humidity level, the Seebeck coefficient and output power density were measured under a temperature gradient of 10 K. The results demonstrate that the P(SPAK‐co‐PEA)_DES IH‐QS thermocells maintain stable Seebeck coefficient and output power density under simulated humidity‐fluctuating environments, meeting the requirements for practical applications (Figure S20).

The outstanding and stable mechanical and thermoelectric properties enable versatile, sustainable applications of P(SPAK‐co‐PEA) IH‐QS thermocells toward wearable electronics and IoT devices in various environments ranging from dry desert areas to high‐humidity coastal areas. To validate the practical value of thermocells, a thermoelectric device capable of illuminating a light‐emitting diode (LED) was constructed using a voltage‐boosting device (Figure 5a). After the implementation of a 10°C temperature gradient between the two sides of the thermocell, a gradual increase in the output voltage is observed during the thermal charging process (Figure 5b). Once the temperature difference between the two sides of the thermocell has been stabilized, the output reaches a steady state. When an external voltage booster was utilized to augment the voltage, the thermocell could continuously light up red, yellow, and green LEDs at different RH (Figure 5c). Reducing RH results in a dimming light of LED (Figure S21).

FIGURE 5.

Potential applications of humidity‐stable IH‐QS thermocells. (a) Schematic illustration of the IH‐QS thermocells lighting up LEDs by harvesting low‐grade heat. (b) Continuous voltage outputs curve of IH‐QS thermocells in response to a ΔT of 25°C at 30% RH. (c) Photographs of the IH‐QS thermocells that continually lights up LEDs under a temperature gradient of 20°C at 30% RH. (d) Output voltage–time curve of thermoelectric array consisting of 36 thermocell units connected in series under an applied temperature difference of 30°C. (e) Photograph of the thermoelectric array fixed on a human hand for powering an electronic wristwatch under a temperature difference of ∼30°C. (f) Voltage‐time curve showing the ability of the thermocell array to charge a supercapacitor (0.1 F) under an applied temperature difference of 30°C. (g–i) The thermoelectric array consisting of 72 thermocell units for powering an electronic wristwatch in simulated environments maintained at 25°C with different humidity levels: (g) desert (30% RH), (h) urban (50% RH), and (i) seaside (90% RH).

In addition to the use of a voltage booster, high output voltage can also be achieved by connecting individual thermocell elements (dimensions: 10 × 10 × 2 mm) in series and integrating them into large‐scale arrays (Figure 5d and Figure S22). The total voltages produced by these arrays are, however, lower than the arithmetic sum of the voltages generated by each unit, which is likely attributable to resistive losses at the electrode interconnections [9, 10, 11, 12, 13, 14]. In our experiments, a thermoelectric generator array consisting of 36 series‐connected thermocell units was fabricated. When worn on the dorsum of the hand under ambient conditions of 60% relative humidity and −5°C, the array produced an open‐circuit voltage of 1.53 V. This output was sustained by maintaining a temperature differential greater than 30°C across the entire device. (Figure 5d). Leveraging the temperature gradient between the hand surface and the ambient environment, the array was able to power an electronic wristwatch at its rated voltage of 1.5 V (Figure 5e). Beyond direct device powering, effective management of surplus electrical energy is essential for expanding practical applications. This was demonstrated by employing the thermoelectric array to charge a 0.1 F supercapacitor to 1.5 V within approximately 200 s (Figure 5f). Furthermore, a separate array consisting of 72 thermoelectric units was fabricated. When attached to a human arm at 36°C, this array effectively powered an electronic wristwatch under simulated ambient conditions of 20°C across varying humidity levels, namely, desert (30% RH), urban (50% RH), and seaside (90% RH) environments (Figure 5g–i). Prior to each test, the thermocell array was equilibrated at the respective humidity level for 2 days to ensure steady‐state moisture absorption, after which it was applied to the arm to supply power to the electronic wristwatch. These functional tests confirm the practical viability of the P(SPAK‐co‐PEA) IH‐TE materials for wearable electronics. The results validate our material design strategy and demonstrate effective mitigation of the issue of moisture‐induced performance variation commonly encountered in unencapsulated thermoelectric devices.

3. Conclusion

In summary, this work develops strong IH‐QS thermocells by integrating amphiphilic polymer networks with hygroscopic quaternary electrolyte solution, enabling efficient and stable low‐grade heat harvesting a broad RH range. The coordinated water retention by hydrophilic moieties and hygroscopic electrolytes allows thermocell to maintain high water content at low RH, ensuring high ionic conductivity and activity of redox couple. Concurrently, hydrophobic phenyl groups of polymer network generate extensive physical crosslinking networks that effectively suppress excessive water uptake at high RH, preserving mechanical and thermoelectric stability across wide RH ranges. Moreover, the ─SO₃ ⁻ and phenyl moieties of amphiphilic network amplifies the entropy difference of redox couple [Fe(CN)₆ ⁴⁻/Fe(CN)₆ ³⁻] via selective intermolecular interactions, boosting thermopower of IH‐QS thermocell. These synergies enable the thermocell to maintain stable water content (33.6–60.3 wt.%) and deliver exceptional mechanical and thermoelectric performance in response to large environmental humidity fluctuations (30%–99% RH). The resulting IH‐QS thermocell can provide high thermopower and normalized power density across 30%–99% RH (e.g., 2.17 mV K⁻¹ and 0.19 mW m⁻² K⁻² at 30% RH; 1.63 mV K⁻¹ and 0.55 mW m⁻² K⁻² at 99% RH). Practical applicability is demonstrated by powering wearable electronics (e.g., LEDs, watches) under different ambient humidity conditions, removing the blocks to the application of IH‐QS thermocells in real‐world environments.

4. Experimental Section

4.1. Materials

Acrylamide (AM, > 98%), potassium ferricyanide (K₃Fe(CN)₆, ≥99%), potassium ferrocyanide trihydrate (K₄Fe(CN)₆·3H₂O, ≥99.99%), Choline chloride (CC, ≥98%) and N,N'‐Methylenebis (2‐propenamide) (MBAA, ≥98%) were purchased from Shanghai Macklin Biochemical Technology Co., Ltd. 2‐phenoxyethyl acrylate (PEA), 3‐sulfopropyl potassium acrylate (SPAK), 2‐acrylamide‐2‐methylpropane sulfonic acid sodium salt solution (AMPS, 50 wt.%) and α‐ketoglutaric acid (≥98%) were purchased from Shanghai Aladdin Biochemical Technology Co., Ltd. Dimethyl sulfoxide (DMSO, ≥99.5%), glycerol (≥99%) and lithium chloride (LiCl, ≥97%) were supplied by Sinopharm Chemical Reagent Co., Ltd. Platinum electrode, peltier chips and sodium calcium glass sheets were purchased from commercial sources in China (Taobao Mall).

4.2. Fabrication of P(SPAK‐co‐PEA) and P(SPAK‐co‐PEA)_H2O IH‐TE Materials

Fabrication of P(SPAK‐co‐PEA) hydrogel: The P(SPAK‐co‐PEA) hydrogel was synthesized through UV‐initiated copolymerization of SPAK and PEA monomers. The total monomer concentration in the precursor solution was fixed at 30 wt.%. Taking the monomer ratio of SPAK/PEA = 1/2 as an example: 0.775 g of SPAK, 1.28 g of PEA, 0.015 g of N,N'‐methylenebis(acrylamide) (MBAA, crosslinker), and 0.005 g of α‐ketoglutaric acid (photoinitiator) were sequentially dissolved in 4.4 g of dimethyl sulfoxide (DMSO). The mixture was first dispersed by magnetic stirring for 5 min and then sonicated for 10 min to achieve complete dissolution. Subsequently, the precursor solution was injected into a mold composed of two soda‐lime glass slides separated by a 2 mm thick silicone spacer. The filled mold was placed in a UV chamber and subjected to bilateral irradiation with 365 nm UV light for 4 h to initiate polymerization. The as‐prepared P(SPAK‐co‐PEA) organogel was sectioned into blocks measuring 1 cm × 2 cm × 2 mm. These specimens were subsequently immersed in ultrapure water for 24 h, with the water replaced every 8 h to facilitate complete extraction of DMSO. After this purification process, the P(SPAK‐co‐PEA) hydrogel was dehydrated in a drying oven at 30°C for approximately 12 h until its volume returned to the pre‐immersion state.

Preparation of the quaternary electrolyte solution: The deep eutectic solvent (DES) was prepared by mixing choline chloride and glycerol at a molar ratio of 1:2. An appropriate amount of water was added to achieve a DES‐H₂O mixture with a water content of 10 wt.%. The mixture was sealed and stirred in the 60°C water bath for 1 h to obtain a homogeneous solution. After cooling to room temperature, 100 mL of the solution was taken, and 1.69 g of potassium ferrocyanide trihydrate (K₃Fe(CN)₆·3H₂O) and 1.32 g of potassium ferricyanide (K₄Fe(CN)₆) were added under continuous stirring to form the DES‐H₂O‐[Fe(CN)₆ ⁴⁻/Fe(CN)₆ ³⁻] mixture. Subsequently, 2.5 g of LiCl was added per 100 g of this mixture, and the sealed system was stirred at room temperature for 12 h to obtain the quaternary electrolyte mixture solution.

Preparation of P(SPAK‐co‐PEA) and P(SPAK‐co‐PEA)_H2O IH‐TE materials: The P(SPAK‐co‐PEA) and P(SPAK‐co‐PEA)_H2O IH‐TE materials IH‐TE material were prepared by submerging the P(SPAK‐co‐PEA) hydrogel into 50 mL of the prepared quaternary electrolyte solution and the aqueous solution containing 0.4 m K₄Fe(CN)₆/K₃Fe(CN)₆ for 48 h, respectively.

4.3. Fabrication of Conventional IH‐TE Materials

Preparation of P(AMPS‐co‐AM) and P(SPAK‐co‐AM) hydrogels: The precursor solution was prepared by mixing AM with either AMPS or SPAK at a molar ratio of 3:1 in ultrapure water, with the total monomer concentration fixed at 30 wt.%. To 10 g of the mixture, MBAA (crosslinker) was added at 0.75 wt.% relative to the total monomer concentration, and α‐ketoglutaric acid (photoinitiator) was added at 0.25 wt.% relative to the total monomer concentration. The resulting solution was injected into a mold composed of two soda‐lime glass slides separated by a 2 mm thick silicone spacer. Polymerization was carried out at 50°C for 4 h in an oven, yielding P(AMPS‐co‐AM) and P(SPAK‐co‐AM) hydrogels.

Preparation of P(AMPS‐co‐AM)_DES/P(SPAK‐co‐AM)_DES IH‐TE materials: The P(AMPS‐co‐AM) and P(SPAK‐co‐AM) hydrogels were cut into blocks measuring 1 cm × 2 cm × 2 mm. The samples were first immersed in ultrapure water for 24 h to remove residual impurities, followed by dehydration in a drying oven until their volumes returned to the pre‐immersion state. Subsequently, the hydrogels were equilibrated by immersion in 50 mL of the prepared quaternary electrolyte mixture solution for 48 h, yielding the final P(AMPS‐co‐AM)_DES and P(SPAK‐co‐AM)_DES IH‐TE materials.

Preparation of P(AMPS‐co‐AM)_H2O/P(SPAK‐co‐AM)_H2O IH‐TE materials: The P(AMPS‐co‐AM)_H2O and P(SPAK‐co‐AM)_H2O IH‐TE materials were prepared by submerging the P(AMPS‐co‐AM) and P(SPAK‐co‐AM) hydrogel with dimensions of 1 cm × 2 cm × 2 mm into 50 mL of an aqueous solution containing 0.4 m K₄Fe(CN)₆/K₃Fe(CN)₆ for 48 h.

4.4. Characterization

The UV–vis spectra were recorded on a UV‐3600 plus spectrophotometer. In the UV–vis spectra measurements, the concentration of the [Fe(CN)₆]⁴⁻/[Fe(CN)₆]³⁻ was maintained at 0.2 mm for all groups. For the AMPS/AM system, the concentrations of AMPS and AM were 0.19 and 0.56 mm, respectively. For the SPAK/AM system, the concentrations of SPAK and AM were 0.19 and 0.56 mm, respectively. For the SPAK/PEA system, the concentrations of SPAK and PEA were 0.25 and 0.5 mm, respectively. The FTIR spectrum of the samples was collected on a Nicolet iS50 spectrometer. The storage modulus (G) of IH‐TE materials under different RH (30%–99%) were characterized at a fixed strain of 1% and an oscillation frequency of 1 Hz by a rheometer (TA DHR3). Low‐field nuclear magnetic resonance (LF‐NMR) images were acquired using a Niumag MesoMR23‐060H‐I medium‐sized nuclear magnetic resonance imaging and analysis system. Small‐angle x‐ray scattering (SAXS) data were collected using a Xenocs Xeuss 2.0 x‐ray small‐angle scattering instrument. The thermoelectric performance was measured with a homemade device using Pt strips (0.1 cm × 0.1 cm × 1 cm) as electrodes, which were placed into a constant temperature and humidity chamber for humidity control. During the measurements of the voltage and current, the temperatures were controlled by two commercial Peltier chips (HT‐12706), and two thermocouples were attached to both ends of the thermoelectric gel to monitor the temperature difference (ΔT). All measurements were performed on samples in the unencapsulated state. The effective ionic conductivity σ_eff for the IH‐TE materials was calculated from the slope of the current–voltage curve. The thermal conductivity (κ) of the thermocell was calculated by the equation: κ = a × c × r, in which the detailed heat in different RH (c, J g⁻¹ K⁻¹) were measured by differential scanning calorimetry (TA Instruments, Waters, DSC 250) at a scanning rate of 10 K min⁻¹, the density of IH‐TE materials (ρ, g cm⁻³) were calculated by the mass and volume, and the thermal diffusivity (α, m² s⁻¹) were measured by the laser‐lash diffusivity instrument using the rectangle‐shaped thermocell with dimensions of 20 mm × 20 mm × 2 mm (Hot Disk‐TPS 2500 S). The mechanical properties of IH‐TE materials were tested with a universal mechanical tester (WDW5T) under a gauge length of 20 mm at a stretching velocity of 20 mm min⁻¹. For the proof‐of‐concept demonstration of lighting LEDs, we used one unit of the thermocell with a temperature gradient of about 10°C. A voltage amplifier (AD620) with a tunable magnification of 1.5–1000 times parallel connected to the IH‐QS) thermocells. The working voltage of an LED was about 1.5–1.8 V and three LEDs were connected in parallel.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: smll73890‐sup‐0001‐SuppMat.docx.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.