Enabling giant thermopower by heterostructure engineering of hydrated vanadium pentoxide for zinc ion thermal charging cells

Abstract

Flexible power supply devices provide possibilities for wearable electronics in the Internet of Things. However, unsatisfying capacity or lifetime of typical batteries or capacitors seriously limit their practical applications. Different from conventional heat-to-electricity generators, zinc ion thermal charging cells has been a competitive candidate for the self-power supply solution, but the lack of promising cathode materials has restricted the achievement of promising performances. Herein, we propose an attractive cathode material by rational heterostructure engineering of hydrated vanadium pentoxide. Owing to the integration of thermodiffusion and thermoextraction effects, the thermopower is significantly improved from 7.8 ± 2.6 mV K⁻¹ to 23.4 ± 1.5 mV K⁻¹. Moreover, an impressive normalized power density of 1.9 mW m⁻² K⁻² is achieved in the quasi-solid-state cells. In addition, a wearable power supply constructed by three units can drive the commercial health monitoring system by harvesting body heat. This work demonstrates the effectiveness of electrodes design for wearable thermoelectric applications.

The application of zinc ion thermal charging cells has been limited by their low kinetics and rapid energy decay. Here, the authors develop a promising material by heterostructure engineering of hydrated vanadium pentoxide for high-performance harvesting and conversion of low-grade heat.

Introduction

Portable devices are widely used in wearable applications for communication, health monitoring, and other areas. However, the unsatisfying capacity and lifetime of power systems like batteries and supercapacitors seriously hinder the development of wearable electronics^1,2. Under this consideration, power supply systems with characteristics of low-cost, high performance, sustainability, and good durability are extremely necessary for the construction of Internet of Things (IoTs)^3–6. Thermoelectric devices (TEs), as one of the key techniques, can directly convert low-grade heat into electricity under a low temperature difference of several Kelvins based on the Seebeck effect, which provides in principle the practicability of TEs in self-power supply for electronics using human body heat^7–10. Very recently, Qiu and his co-workers discovered a series of p-type ductile TE materials, AgCu(Se, S, Te) pseudoternary solid solutions for the fabrication of flexible TEs¹¹. When coupling with the n-type ductile material (Ag₂₀S₇Te₃), they developed a π-type flexible TE devices with a thickness of 0.3 mm and high normalized power density of 30 μW cm⁻² K⁻². It should point out that this device-integrated 31-pair flexible TE units adhered to human’s wrist can output a voltage of 0.2 mV and power of 70 nW. Although some applied heat-dissipation routes like copper pipes and graphite sheets would further improve the output performances to power some microelectronics, the relatively complicated integration of working units still is the key issue that needs to be addressed.

Alternatively, ionic thermoelectric devices (i-TEs) are proposed for low-grade heat harvesting by combining the diffusion of ions and redox reactions of species under the existence of temperature gradient^12,13. Owing to the rational integration of thermodiffusion and thermogalvanic effects, i-TEs have been widely concerned by researchers in practical applications¹⁴. As a proof-of-concept, Liu et al. developed a new-type ionic thermoelectric material for the harvesting of human body heat¹⁵. Typically, potassium chloride (KCl) and ferro/ferricyanide [Fe(CN)₆^4–/Fe(CN)₆^3–] were optimized in a gelatin matrix to simultaneously achieve the synergistic changes of thermodiffusion entropy and thermogalvanic entropy. As a result, a giant thermopower of 17.0 mV K⁻¹ can be obtained using copper electrode under a temperature difference of ~10 K. Most importantly, a high voltage of 2 V and a peak power of 5 mW can be generated from the harvesting of body heat by integrating 25 unipolar units. Huang et al. designed a nonaqueous ionogel with strong ion-ion interactions introduced by selective ion doping to adjust the ionic thermopower¹⁶. Detailly, ionogels based on an ionic liquid (EMIMTFSI/PVDF-HFP) shows high negative thermopower of −15 mV K⁻¹ after adding 0.5 M LiBF₄. Moreover, a peak value of positive thermopower (17 mV K⁻¹) can be achieved with the introduction of EMIMCl. According to these inspired results, the prototype device with 12 pairs of n-/p-type units, which is adhered on human arm, can produce a voltage of 0.33 V. Apart from the development of advanced ionogels, the introduction of chaotropic/host agent is an efficient strategy to enhance the thermoelectrochemical performances by increasing the concentration difference¹⁷. For example, Zhou and co-workers reported a thermosensitive guanidinium (Gdm⁺) induced Fe(CN)₆^4– crystallization to construct high-performance thermocells¹⁸. With the precipitation of crystals, both the voltage and thermopower of the system can be enhanced. However, it is difficult to adopt the precipitation and redissolution of thermosensitive crystallization into wearable applications that would meet the demands of multidimensional motion. Till now, the development of high-performance thermoelectric devices has been the top priority for self-powered electronics through human body heat harvesting.

Besides the optimization of functional electrolytes, electrode engineering is another efficient approach to improve thermoelectrochemical performances¹⁹. Based on the electrical double-layer capacitive effect or near-surface redox behavior, some electrode materials including porous carbon, carbon nanotube, and conductive polymer are widely studied in previous reports^20–26. To further boost the energy conversion efficiency and energy storage density during the thermoelectrical process, we have proposed a zinc ion thermal charging cell (ZTCC) using Zn metal anode and vanadium dioxide-porous carbon (VO₂-PC) cathode²⁷. As-fabricated ZTCC can convert low-grade heat into electricity with a thermopower of ~12.5 mV K^–1 by the combination of thermodiffusion and thermoextraction. Notably, the whole thermoelectrochemical process is dominated by the thermodiffusion effect, which may be caused by the embedding feature of VO₂ in PC matrix. It should be mentioned that the average ion diffusion coefficient of VO₂-PC is in a relatively low level (3.16 × 10⁻¹¹ cm² S⁻¹). To realize a fast thermoelectrochemical response of ZTCCs, the kinetics matching of themodiffusion and thermoextraction, as well as the ion mobility and conductivity of electrode materials, should be well considered. Importantly, vanadium oxides (especially for vanadium pentoxides) are regarded as attractive candidates for energy storage and conversion owing to their relatively large inner spacing, tuned nanostructures, and multielectron reactions^28,29. Nonetheless, the dissolution of vanadium in neutral or acidic electrolytes during electrochemical processes would result in the failure of electrodes and contamination of electrolytes. Meanwhile, the low electronic conductivity and electrostatic repulsion of layered vanadium oxides always lead to unsatisfying rate/power performances.

Herein, we propose and demonstrate a ZTCC with boosted performances using graphite modified zinc anode (Zn-G) and vanadium pentaoxide@reduced graphene oxide (V₂O₅@rGO) cathode. In detail, the properties of V₂O₅ are reasonably adjusted by the addition of rGO. Benefitting from the thermodiffusion contribution of rGO and thermoextraction behavior of V₂O₅, high-performance ZTCC in term of thermopower and power density can be achieved. Worthily, the rGO nanosheets can not only enhance the kinetics of V₂O₅ host, but also suppress the dissolution of vanadium species. Such attractive integration between V₂O₅ and rGO endows as-prepared V₂O₅@rGO composites with high thermal and electrochemical stability. In addition, the V₂O₅@rGO-1.5 based ZTCC shows great promise in the application of wearable health monitoring systems. Compared with other strategies for designing flexible thermoelectric systems, this work provides a facile way to construct a boosted ionic thermoelectrochemical device by the efficient structure regulation of electrode materials.

Results

Boosted ZTCCs enabled by V₂O₅@rGO-x

ZTCC constructed by insertion-type cathode and zinc metal anode attracts great attention due to the promising thermoelectrochemical performances during the conversion of low-grade heat and energy storage. As schematically illustrated in Fig. 1a, the thermopower of ZTCC is mainly determined by the thermodiffusion of electrolyte ions and thermogalvanic processes of electrodes. In detail, the pre-inserted cations in the cathode can be gradually extracted with heat input and transferred to the anode side under the effect of the thermal field. With the plating of diffused Zn²⁺ in Zn anode, the heat energy can be converted into electricity and stored simultaneously. Fig. 1b displays the possible voltage curve. To achieve a boosted thermoelectrochemical response, the development of electrode materials with fast kinetics and good durability becomes significant. The combination of vanadium oxides with nanocarbons shows great promise to overcome such challenges. As a demo, we proposed the V₂O₅·1.6H₂O with large interlayer spacing as a host material for charge storage and rGO as a functional matrix to enhance the kinetics and stability of composites. The density functional theory (DFT) calculations were performed on the model structure to understand the relationship between the electronic structure of V₂O₅@rGO and the superior kinetics behavior. As shown in Fig. 1c, the differential charge density of V₂O₅@rGO displays obvious charge accumulation around the V and O atoms and dispersion of surrounding electron states around the Zn atom. This charge redistribution reveals the optimized electronegativity of V₂O₅·1.6H₂O, further boosting the adsorption ability to Zn²⁺. Consequently, the V₂O₅@rGO exhibits a low negative adsorption energy of Zn²⁺ (−1.47 eV). This suggests that the ultra-large interlayer spacing of V₂O₅·1.6H₂O is beneficial to the adsorption and diffusion of Zn²⁺. Simultaneously, the crystal H₂O molecules in V₂O₅ layers can act as a “lubricant” to improve host electrochemical kinetics. Subsequently, the rGO layer on V₂O₅·1.6H₂O surface retains structural integrity and accelerates the charge transfer. In addition, the diffusion pathway of Zn²⁺ in the V₂O₅@rGO sample is shown in Fig. 1d. It is worth mentioning that the energy barrier for ions diffusion in V₂O₅@rGO is at a low level of 0.27 eV (Supplementary Fig. 1), confirming the easy diffusion of Zn²⁺ along the V₂O₅@rGO heterointerface. Such high Zn²⁺ diffusion coefficient and low diffusion barrier endow V₂O₅@rGO composites with high-power/rate performances in energy conversion and storage applications.

Fig. 1. Construction of boosted ZTCC and characterization of V₂O₅@rGO sample.

a Schematic illustration of V₂O₅@rGO based ZTCC, and b corresponding voltage distribution profile. c Differential charge density, and d the optimized diffusion pathway of Zn²⁺ in V₂O₅@rGO composite. e TEM image of V₂O₅@rGO-1.5. f XRD pattern of V₂O₅, V₂O₅@rGO-1.5, and rGO.

According to the guidelines, we propose an electrostatic interaction-induced self-assembly strategy to realize the structure regulation on the V₂O₅@rGO-x nanocomposites (x: concentration of GO solution, mg mL⁻¹). The field-emission scanning electron microscopy (FESEM) and transmission electron microscopy (TEM) images in Supplementary Fig. 2 and Fig. 1e clearly indicate that the V₂O₅@rGO-1.5 shows reasonably intertwined microstructure engineered by multilayer rGO and V₂O₅ nanobelts with several micrometers in length. It is worth mentioning that the 1D V₂O₅ nanobelts with large interlayer spacing can provide abundant electrochemical sites for ions intercalation/deintercalation and the 2D rGO nanosheets wrapped around the V₂O₅ nanobelts can act as matrix for ions adsorption/desorption as well as a pathway for charge transfer. However, the pure V₂O₅ obtained without adding GO solution exhibits random nanoparticle morphology (Supplementary Fig. 3a), which is significantly different from the nanobelt-type V₂O₅ in V₂O₅@rGO nanocomposites. Moreover, the pristine GO after hydrothermal process displays a well-developed 3D continuous framework formed by corrugated and stacked rGO nanosheets (Supplementary Fig. 3b). When considering the fully different microstructures of V₂O₅@rGO, V₂O₅, and rGO samples, it can be concluded that the introduction of GO into the preparation of V₂O₅ is a rational approach to regulate the structure of nanohybrids. In fact, the relatively acidic atmosphere (pH: ~3.0) caused by GO solution could promote the generation of positively vanadium oxide, which would adsorb on the negatively charged surface of GO and gradually evolve into nanobelts between rGO layers³⁰. The morphologies of V₂O₅@rGO composites highly depend on the content of rGO. As displayed in Supplementary Fig. 3c, the V₂O₅@rGO-1.0 holds the characteristics of V₂O₅ nanobelts, in which rGO substrate is fully covered. While for the V₂O₅@rGO-2.0, only a few V₂O₅ nanobelts are exposed from the 3D graphene network (Supplementary Fig. 3d). The corresponding high-resolution transmission electron microscopy (HRTEM) image in Supplementary Fig. 4 confirms the successful synthesis of layered V₂O₅ in V₂O₅@rGO-1.5 sample. Consistently, the ultra large interlayer spacing of around 1.1 nm can be indexed to the (001) plane of orthorhombic V₂O₅·1.6H₂O³¹. Selected area electron diffraction (SAED) pattern reveals the polycrystalline feature of formed V₂O₅·1.6H₂O nanobelts and amorphous rGO layers (Supplementary Fig. 5), further confirming the formation of heterostructure by V₂O₅·1.6H₂O and rGO. Meanwhile, the high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) and corresponding elemental mapping images detect the distribution of V, C, and O elements in V₂O₅@rGO-1.5, demonstrating the rational integration of rGO and V₂O₅ (Supplementary Fig. 6).

The phase and structure of samples are confirmed by the X-ray diffraction (XRD) patterns. As plotted in Fig. 1f, all the obvious peaks of V₂O₅@rGO-1.5 correspond to the characteristic peaks of orthorhombic V₂O₅·1.6H₂O (PDF#40–1296). The additional peak around 25.3° in the spectrum of V₂O₅@rGO-1.5 can be attributed to the (002) plane of amorphous rGO. Interestingly, the pure V₂O₅ obtained without addition of GO is in terms of orthorhombic V₂O₅ (PDF#89-0612)³². Notably, the pattern of V₂O₅@rGO-1.0 is similar with that of V₂O₅@rGO-1.5 (Supplementary Fig. 7). The interlayer spacing of pure V₂O₅ is significantly enlarged from ~0.44 nm to ~1.1 nm of V₂O₅·1.6H₂O in both V₂O₅@rGO-1.5 and V₂O₅@rGO-1.0, implying that acidic GO exhibits important role in regulating the microstructure of vanadium oxides. Moreover, as detected by the XRD pattern of V₂O₅@rGO-2.0, the V₂O₅·1.6H₂O crystals can further evolve into low-valence vanadium species like V₆O₁₃ and V₃O₇·1.6H₂O with the increasing of GO concentration due to partial reduction of high-valence vanadium species by formed rGO during hydrothermal treatment. The comparison of Raman spectra in Supplementary Fig. 8 shows typical characteristics of rGO (1330.7 and 1598.2 cm⁻¹) and VO_x (142.5, 194.4, 662.2, 404.8, 698.2 and 992.7 cm⁻¹)^33,34. The co-existence of above Raman peaks in V₂O₅@rGO-x hybrid demonstrates the rational combination of rGO and VO_x, further confirming the composition of the heterostructure. Moreover, the thermogravimetric analysis (TGA) indicates that the V₂O₅ in V₂O₅@rGO-1.5 contains ~1.62 crystal water molecules per unit, and the content of rGO is about 31.7% (Supplementary Fig. 9). In addition, the specific surface area (SSA) of sample is measured by N₂ adsorption-desorption isotherms (Supplementary Fig. 10a). Benefitting to the abundant pores and 3D-interconnected structure, the rGO shows the largest adsorbed quantity among as-prepared materials. Consequently, a high SSA value of 284.1 m² g⁻¹ can be achieved (Supplementary Fig. 10b). However, the pristine V₂O₅ only shows 5.2 m² g⁻¹. After introducing rGO, the SSA value can be reasonably improved to 13.3 m² g⁻¹ for V₂O₅@rGO-1.5. Such high SSA together with mesoporous feature of V₂O₅@rGO-1.5 can effectively boost the ions mobility. The X-ray photoelectron spectroscopy (XPS) of V₂O₅, rGO and V₂O₅@rGO-x composites was carried out to reveal the integration of carbon matrix and vanadium oxides (Supplementary Fig. 11a). Notably, the broad peak of O 1 s for V₂O₅ at 530.3 eV can be attributed to the VO_x (Supplementary Fig. 11b)³⁵. Deconvolution of O 1 s spectra for V₂O₅@rGO-x samples shows two additional peaks at 531.4 and 533.4 eV, which corresponds to the C = O groups in rGO and the water in the sample, respectively^35,36. As shown in Supplementary Fig. 11c, the peaks at 517.3 (V 2p_3/2) and 524.6 eV (V 2p_1/2) for V₂O₅ belong to V⁵⁺³⁷. After combining with rGO, a part of V⁴⁺ can be found in V 2p of V₂O₅@rGO-x composites. Detailly, the ratio of V⁴⁺ to V⁵⁺ gradually increases from 0.28 (V₂O₅@rGO-1.0) to 0.47 (V₂O₅@rGO-2.0), which could be caused by the reduction effect of rGO to V⁵⁺ during the hydrothermal procedure. Moreover, the contact angle (CA) of pristine V₂O₅ is about 26.1° (Supplementary Fig. 12a), suggesting its superhydrophilic feature. Even after standing 20 s, the CA value still can be maintained at 23.2° (Supplementary Fig. 12b). Worthily, the CA value of V₂O₅@rGO-x composite gradually increases with the increase of rGO amount, which may be caused by the stack of graphene layers. According to the change of CA values during electrolyte immersion, the V₂O₅@rGO-1.0 delivers higher electrolyte permeability than that of V₂O₅@rGO-1.5 and V₂O₅@rGO-2.0. These differences in CA values highly affect to the dissolution of vanadium species from the electrode into the electrolyte. As shown in Supplementary Fig. 13a, the electrolyte solution with pristine V₂O₅ shows a noticeable color change from transparent to yellow, suggesting the serious dissolution of V species in the electrolyte. As expected, the dissolution of V species of various V₂O₅@rGO-x electrodes can be significantly suppressed by the rGO coating layers. Furthermore, the detailed concentration of V species in electrolytes after soaking for 5 days and 10 days is summarized in Supplementary Fig. 13b. After soaking for 5 days, the concentration of V in electrolyte for V₂O₅ is 47.5 mg L⁻¹, which is several times higher than those for V₂O₅@rGO-x electrodes (8.9 ~ 17.3 mg L⁻¹). Even after soaking for 10 days, the average V element dissolution rate of V₂O₅ (7.1 mg L⁻¹ per day) is much higher than those of V₂O₅@rGO-x electrodes (1.1 ~ 2.0 mg L⁻¹ per day), confirming the important role of rGO to maintain the structure stability in V₂O₅@rGO-x composites. Besides, the ultralow electron conductivity of 1.4×10⁻⁵ S cm⁻¹ for pristine V₂O₅ can be significantly enhanced to 1.2 S cm⁻¹ for the V₂O₅@rGO-1.5 with the introduction of rGO (Supplementary Fig. 14). Thus, the coated rGO on V₂O₅ nanobelts can be acted as conductive layers in heterostructure to enhance the kinetics as well as satisfying power capability in energy conversion and storage.

Evaluation of thermoelectrochemical performances

To correlate the structure regulation of V₂O₅@rGO-x with the enhancement of thermoelectrochemical performances in the harvesting of low-grade heat, the open-circuit voltage values of systems were recorded using non-isothermal H-type cell. Detailly, the modified Zn-G foil (thickness: ~20 μm) and V₂O₅@rGO-x are used as anode in cold side and cathode on hot side, respectively. It should be pointed out that the temperature difference (ΔT) between such two chambers can be formed by water bath and determined by inserted thermocouples. To evaluate the Seebeck coefficient of rGO, V₂O₅, V₂O₅@rGO-1.0, V₂O₅@rGO-1.5, and V₂O₅@rGO-2.0, we have profiled the related results with error bands or error bars (error band/bar represents the standard deviations). Benefitting to the adsorption/desorption mechanism of ions, the kinetics of V₂O₅ can be rationally enhanced with the addition of rGO. As summarized in Fig. 2a, the V₂O₅@rGO-1.5 based ZTCC generated the highest output voltage of ~1.1 V among all other ZTCCs. In fact, such high voltage can be divided as thermal-induced voltage and self-charging-induced voltage. The self-charging behavior of Zn-related devices was carefully studied in previous works³⁸. To exclude the contribution of the self-charge process to the total voltage, we further recorded the voltage curves during self-charging processes in Supplementary Fig. 15. Owing to the continuous and dense framework of rGO, the gradual desorption of ion from interconnected channels endows rGO-contained ZTCC with relatively low self-charging rate. Thus, the V₂O₅@rGO-1.5 based ZTCC delivers the highest thermal-induced voltage of 0.72 V. Based on these results, the thermopower or Seebeck coefficient (S_i) can be determined by the relationship between thermal-induced voltage and temperature difference^15,18:

$$S_i=\frac{\mathrm{\Delta }\text{V}}{\mathrm{\Delta }\text{T}}$$ 1

Fig. 2. Thermoelectrochemical performances of H-type ZTCCs using 0.5 mol L⁻¹ Zn(CF₃SO₃)₂.

a Output voltage under various temperature differences from 5 K to 30 K, and b corresponding Seebeck coefficient. c Comparison of Seebeck coefficient and thermal-induced voltage with other reported values. d Temperature coefficients and corresponding fractional contributions of various thermal processes.

As profiled in Fig. 2b, the rGO-based ZTCC shows large S_i value of 10.8 ± 2 mV K⁻¹ based on the ultrafast adsorption/desorption. While the V₂O₅ based ZTCC only delivers 7.8 ± 2.6 mV K⁻¹ due to the relatively sluggish insertion/extraction of Zn²⁺ in its crystalline layers. When combining the merits of rGO and V₂O₅, a giant S_i value of 23.4 ± 1.5 mV K⁻¹ can be rationally achieved by adjusting the content of rGO. In detail, the slight lower S_i value of V₂O₅@rGO-1.0 based ZTCC than that of V₂O₅ based ZTCC may result from the its relatively dense structure and relatively low specific surface area. While for V₂O₅@rGO-2.0 based ZTCC, the S_i value is close to that of rGO-based ZTCC, which may be caused by the stacking of as-introduced excessive rGO into V₂O₅. Accordingly, the Seebeck coefficient highly depends on the synergy behavior of thermal adsorption/desorption and thermal insertion/extraction. As is compared in Fig. 2c, this breakthrough is also superior to most of previously reported results^{15–18,20,23,24,27,35,39,40}, indicating the great promise of structure regulation by integrating graphene and vanadium oxides for high-efficient conversion from heat to electricity. As profiled in Supplementary Fig. 16, the rGO, V₂O₅, V₂O₅@rGO-1.0, V₂O₅@rGO-1.5, and V₂O₅@rGO-2.0 based ZTCC exhibits thermo-voltage of approximately 1.2, 1.2, 0.75, 1.1, and 0.7 V with the temperature difference of 30 K. It is worth mentioning that such output voltage can be further enhanced to 1.6 V by adding power charge. After discharging with a current density of 0.1 A g⁻¹, we can calculate that the ratio of thermal charge part in fully charged rGO, V₂O₅, V₂O₅@rGO-1.0, V₂O₅@rGO-1.5, and V₂O₅@rGO-2.0 based cell is 55.2%, 68.3%, 41.2%, 69.3%, and 30.0%, respectively. This result indicates that the heat-to-current conversion of ZTCC is highly determined by the kinetics of electrodes. Supplementary Fig. 17 displays the short-circuit current plots and fitted power density curves under thermal-induced voltage for various H-type ZTCCs. As found, the V₂O₅@rGO-1.5 based ZTCC delivers high voltage of 0.67 V and current density of 5.3 A m⁻² among rGO (0.39 V, 0.42 A m⁻²), V₂O₅ (0.41 V, 0.59 A m⁻²), V₂O₅@rGO-1.0 (0.59 V, 1.7 A m⁻²), and V₂O₅@rGO-2.0 (0.59 V, 3.5 A m⁻²). Further, a relatively low resistance value of 1118.7 Ω can be calculated from the V₂O₅@rGO-1.5 based ZTCC, while the value is 3071.3 and 1491.8 Ω for V₂O₅@rGO-1.0 and V₂O₅@rGO-2.0, respectively. Besides, the rGO based ZTCC delivers relatively high inner resistance of 8217 Ω due to its poor wettability for electrolyte. While the resistance of 6150 Ω for V₂O₅ based ZTCC is mainly caused by its ultralow electron conductivity. Meanwhile, the V₂O₅@rGO-1.5 based ZTCC exhibits an excellent thermal-induced power density of 0.94 W m⁻², meaning a high P_max/(A×ΔT²) value of 1.04 mW m⁻² K⁻². Such values are greatly larger than that of 0.06 W m⁻² (0.07 mW m⁻² K⁻²), 0.06 W m⁻² (0.07 mW m⁻² K⁻²), 0.27 W m⁻² (0.3 mW m⁻² K⁻²), and 0.65 W m⁻² (0.72 mW m⁻² K⁻²) for rGO, V₂O₅, V₂O₅@rGO-1.0 and V₂O₅@rGO-2.0 based ZTCCs, respectively. As shown in Supplementary Fig. 18, the V₂O₅@rGO-1.5 based ZTCC maintains relatively stable voltage change over 6000 s (60 cycles), demonstrating its good thermal charging stability. Due to the mismatched energy/kinetics during thermoelectrochemical process, the dramatic change in the voltage during the tests can be found in the rGO, V₂O₅, V₂O₅@rGO-1.0, and V₂O₅@rGO-2.0 based ZTCCs. Moreover, a negligible voltage drop (7 ~ 29 mV) occurs in self-discharge curves of various H-type ZTCCs when eliminating the temperature gradient (Supplementary Fig. 19). Especially, the V₂O₅@rGO-x based ZTCCs show relatively fast thermoelectrochemical response, and V₂O₅@rGO-1.5 based ZTCC delivers the lowest voltage change of ~7 mV, suggesting the dense energy storage behavior of as-proposed ZTCC. The gradual growth of voltage in following process can be attributed to the chemically self-charging behaviors³⁸. To detailly study the relative contribution of thermodiffusion and thermogalvanic process to such giant value, we conducted a series of measurements for various temperature coefficients from 25 to 50 °C using the three-electrode configuration proposed in previous works^17,35. As briefly illustrated in Supplementary Fig. 20a, two same electrodes are employed as working electrode and counter electrode, respectively, and the Ag/AgCl is used as reference electrode. During tests, the temperature of setup is determined by hot water bath and recorded by thermocouple. As plotted in Supplementary Fig. 20b, the thermodiffusion of electrolyte ions can deliver a temperature coefficient of 0.54 mV K⁻¹. Notably, the temperature coefficient for Zn/Zn²⁺ and Zn_xVO/VO is 0.77 and 6.9 mV K⁻¹ (Supplementary Fig. 20c–f), respectively. Consequently, the total temperature coefficient (9.8 mV K⁻¹) can be divided as 21.8% of thermodiffusion contribution and 78.2% of electronic effect contribution (Fig. 2d). All these findings reveal that the rational integration of capacitor-type rGO and battery-type V₂O₅ can realize the ultrafast response to heat/electricity signal.

Kinetics analyses

The kinetics of as-obtained V₂O₅@rGO-1.5 cathode was reasonably studied by the cyclic voltammetry (CV) measurements and the galvanostatic intermittent titration techniques (GITT) with an operating voltage window from 0.2 to 1.6 V (Supplementary Note 1, Supplementary Figs. 21 and 22). As displayed in Fig. 3a, two pair of visible redox peaks at 0.4/0.7 V and 0.9/1.1 V in all the CV curves from 0.1 to 1.0 mV s⁻¹ can be attributed to the multiple insertion/extraction procedures of Zn²⁺ in V₂O₅ (V₂O₅ + xZn²⁺+ yH₂O + 2xe⁻ ↔ Zn_xV₂O₅·yH₂O). The well-maintained CV curves even at relatively high scan rate further suggest the good reversibility and rapid response of V₂O₅@rGO-1.5 electrode. As one of convenient indicators, the near surface-dominated procedure and bulk diffusion-controlled process can be distinguished by using i = av^b, where i and v represent the peak current and scan rate, a and b are adjustable parameters⁴¹. Typically, b = 0.5 means that the process is totally governed by diffusion, while b = 1.0 suggests the capacitive behavior. As plotted and fitted in Fig. 3b, the as-calculated b value for peaks 1–4 are 0.78, 0.86, 0.79, and 0.81, respectively, revealing that the electrochemical process of the V₂O₅@rGO-1.5 is monopolize-dominated by the capacitive process along with partial diffusion contribution. Moreover, the detail contribution of such two processes can be clarified by the following relationship:⁴²

$$i=i_{\mathrm{cap}}+i_{\mathrm{diff}}=k_{1}v+k_2{v}^{1/2}$$ 2

Fig. 3. Electrochemical kinetics analysis of V₂O₅@rGO-1.5.

a CV curves at different scan rates. b log (Current) vs. log (Scan rate) of various peaks, c the capacity contribution proportion at various scan rates. d Zn²⁺ diffusion coefficient obtained by GITT curves at charging/discharging.

As specifically shown in Supplementary Fig. 12, a high capacitive contribution of 80.8% can be achieved at the scan rate of 1.0 mV s⁻¹. Moreover, the contribution ratio of the capacitive process shows a gradual increase trend from 0.1 to 1.0 mV s⁻¹ (Fig. 3c). The capacitance dominated process in each scan rate is in line with the result discussed above. Meanwhile, the GITT was used to investigate the Zn²⁺ diffusion coefficient in electrodes to highlight the structural merits of V₂O₅@rGO-1.5 nanocomposite in Zn-based thermoelectrochemical devices. The corresponding GITT curves during charging and discharging processes are displayed in Supplementary Fig. 23. Accordingly, the Zn²⁺ diffusion coefficient (D_Zn) of V₂O₅ is in a range of 10⁻¹² to 10⁻¹¹ cm² s⁻¹ during the whole electrochemical tests (Supplementary Fig. 23a, e), demonstrating relatively sluggish insertion/extraction of Zn²⁺ in the pristine V₂O₅. With the increase of rGO content, the D_Zn value can be enhanced to a range of 10⁻¹¹ to 10⁻¹⁰ cm² s⁻¹ for V₂O₅@rGO-1.0 and V₂O₅@rGO-1.5 (Supplementary Fig. 23b, f and c, g), which is almost an order of magnitude higher than the value of V₂O₅. However, the D_Zn value shows a slight loss in the V₂O₅@rGO-2.0 (Supplementary Fig. 23d and h), which mainly caused by the dense wrapping of VO_x species by stacked rGO layers. As summarized in Fig. 3d, V₂O₅@rGO-1.5 exhibits the highest average Zn²⁺ diffusion coefficient of 4.19×10⁻¹¹ cm² s⁻¹ during the whole electrochemical processes. Such relatively fast insertion/extraction of Zn²⁺ in the V₂O₅@rGO-1.5 electrode may be owing to the relatively large interlayer spacing of V₂O₅·1.6H₂O and the high conductivity of rGO substrate. Most importantly, the crystal water may not only effectively reduce energy barrier but also mask the electrostatic interaction of Zn²⁺ via solvation effect^43–45. The slightly low D_Zn value of charging part implies the sluggish extraction kinetics of Zn²⁺ from cathode, which requires high energy input to realize such charging procedure. Based on this feature, high efficiency and density can be achieved for the conversion from low-grade heat to electricity by using V₂O₅@rGO-1.5 based ZTCCs.

Energy conversion and storage behaviors

As attractive system integrated energy conversion and storage, the zinc ion batteries (ZIBs) are assembled to evaluate the electrochemical performances. Notably, the V₂O₅@rGO-1.5 based ZIBs shows superior rate capability (Supplementary Fig. 24a), in which 375.5, 383.5, 376.9, 353.4, 320.4, 266.3, 208.0, and 145.0 mAh g⁻¹ can be recorded at 0.1, 0.2, 0.5, 1.0, 2.0, 5.0, 10.0, and 20.0 A g⁻¹, respectively. Impressively, the reversible discharging capacity can be nearly stabilized at 402.8 mAh g⁻¹ as the current density recovers to 0.1 A g⁻¹. The slight increase of specific capacity after test could be caused by the electrolyte penetration induced activation process³⁵. The rate performance of V₂O₅@rGO-1.5 cathode is also better than that of rGO, V₂O₅, V₂O₅@rGO-1.0, V₂O₅@rGO-2.0 electrodes and some of previously reposted materials (Supplementary Fig. 25a), such as VO₂-PC (82.6 mAh g⁻¹ at 20 A g⁻¹)²⁷, VPMX73 (282.2 mAh g⁻¹ at 3 A g⁻¹)⁴⁶, ZVO (167 mAh g⁻¹ at 15 A g⁻¹)⁴⁷, Zn_0.25V₂O₅·nH₂O (183 mAh g⁻¹ at 6 A g⁻¹)⁴³, MnVO (214 mAh g⁻¹ at 8 A g⁻¹)³¹, Ni_0.25V₂O₅·nH₂O (164 mAh g⁻¹ at 5 A g⁻¹)⁴⁸, NVO/CNTs (203 mAh g⁻¹ at 4 A g⁻¹)⁴⁹, NaCaVO (154 mAh g⁻¹ at 5 A g⁻¹)⁴⁵, and V₂O₅·1.6H₂O/MXene (81.2 mAh g⁻¹ at 2 A g⁻¹)⁵⁰. As plotted in Supplementary Fig. 25b, the V₂O₅@rGO-1.5 based devices exhibits the highest energy density of 265 Wh kg⁻¹ at a power density of 151 W kg⁻¹ based on the active mass loading of cathode. Even under the highest power density of 12964.3 W kg⁻¹, a high energy density of 84.7 Wh kg⁻¹ still can be retained, suggesting the superior energy storage behavior and rate capability of V₂O₅@rGO-1.5 among other electrode materials. Supplementary Fig. 24b displays the specific capacity retention of various electrodes at the current density of 10 A g⁻¹. Initially, the capacity of V₂O₅@rGO-1.5 increases to 195 mAh g⁻¹ due to the gradual activation. Such value can maintain about 122 mAh g⁻¹ even over 5000 cycles together with the corresponding Coulombic efficiency of ~100%, suggesting satisfying long cyclic stability. Notably, an obvious increase of capacity before 2000 cycles can be found in V₂O₅ cathode, this may be caused by the electrochemical activation process. However, large capacity decay of V₂O₅ cathode in following cycles may be caused by the dissolution of vanadium. Similar phenomenon can be found in V₂O₅@rGO-1.0 cathode. In addition, the cycled V₂O₅ electrode shows a distinct morphology change due to the obvious vanadium dissolution (Supplementary Fig. 26a). The cycled rGO electrode still holds the inter-connected morphology with abundant pores (Supplementary Fig. 26b), in line with its initial morphology. In sharp contrast, the cycled V₂O₅@rGO-x electrodes display similar structure and morphology to their pristine state (Supplementary Fig. 26c–e), and rGO layers can be visually observed in each electrode. From the EDX mapping images of each electrode after cycling, we can find that the zinc species exhibit in all electrodes due to the irreversible electrochemical reactions. Notably, both V₂O₅@rGO-2.0 and V₂O₅@rGO-1.5 electrodes present the relatively uniform distribution of C, O, Zn, and V elements (Supplementary Fig. 26d, e), demonstrating the effect of rGO to suppress the vanadium dissolution. However, the distribution of C and V in cycled V₂O₅@rGO-1.0 electrode is relatively uneven, which may be caused by the dissolution of vanadium. Therefore, both the V₂O₅@rGO-2.0 and V₂O₅@rGO-1.5 electrodes can achieve good stability when comparing with the V₂O₅@rGO-1.0 electrode. As plotted in Supplementary Fig. 27, the Seebeck coefficient value of each system by re-using cycled electrodes is relatively lower than that using fresh working electrodes. Especially, the ultralow Seebeck coefficient of 0.75 ± 0.1 mV K⁻¹ of pristine V₂O₅ is mainly caused by the serious dissolution of vanadium species during long-term cycling tests. However, the ZTCC with cycled V₂O₅@rGO-1.5 still can deliver a relatively high Seebeck coefficient of 17.5 ± 2.6 mV K⁻¹, further demonstrating that the rGO coating on V₂O₅ nanobelt shows key role to achieve satisfying durability and stability of vanadium-based devices by suppressing the dissolution of vanadium species to electrolyte. It should point out that the slight decrease of Seebeck coefficient value for V₂O₅@rGO-1.0, V₂O₅@rGO-1.5, V₂O₅@rGO-2.0, and rGO based ZTCCs could be the influence of byproducts.

Besides, various temperatures from 30 °C to 50 °C were adopted to further investigate the merits of heterostructure. With the increase of testing temperature (30 °C and 40 °C), the specific capacity of all cathodes can be enhanced and the electrochemical activation process can be accelerated due to the thoroughly infiltration of electrolytes (Supplementary Fig. 24c, d). It is worth mentioning that the V₂O₅@rGO-2.0 cathode exhibits better rate capability and cycling stability than other electrodes at 40 °C (Supplementary Fig. 24e, f), which is mainly attributed to its lowest dissolution rate of vanadium into electrolyte among other materials. Apart from the dissolution of vanadium, the dendrite growth and byproduct formation on Zn-G anode under relatively high temperatures (50 °C) would become the culprit of cell failure. Worthily, all as-assembled cells will be short-circuited when the total testing time is around 80 h (Supplementary Fig. 24g, h). Even vanadium dissolution at high temperatures will intensify, the cells should be operated with relatively low capacity rather than short-circuited. Thus, the biggest issue of Zn-based cells under high temperatures could be the serious growth of dendrite.

Furthermore, only a low absolute temperature coefficient of 2.5 mV K⁻¹ can be obtained from the CV curves of as-developed V₂O₅@rGO-1.5 based ZTCC, which is much lower than the Seebeck coefficient (23.4 ± 1.5 mV K⁻¹). Such relatively small polarization potential (~50 mV) indicates the fast electron transfer kinetics and the better reversibility of ZTCCs. Noted from equation S7 in Supplementary Note 2, in addition to the tested temperature coefficient, some possible factors also could affect thermopower or Seebeck coefficient of ZTCCs. The first one is related to the entropy change of redox species. The second one is the concentration difference of redox species in ZTCCs, which greatly impacts the whole Seebeck coefficient due to the possible chemical reactions between the two electrodes. The last one is the thermodiffusion of electrolyte ions in ZTCCs. Meanwhile, the CV curves of both H-type ZTCCs under different temperature differences and coin-type cells under various temperatures indicate the polarization of redox reactions between V₂O₅ and Zn can be significantly optimized by adding moderate rGO (Supplementary Note 2, Supplementary Figs. 28 and 29).

To well understand the energy conversion and storage mechanism of ZTCC, we observed the morphology change of both V₂O₅@rGO-1.5 cathode and Zn-G anode during the thermoelectrochemical measurements by SEM. As shown in Supplementary Fig. 30a1, the V₂O₅@rGO-1.5 is evenly wrapped by carbon black and PVDF in the initial electrode. After discharging to 0.2 V, some thin films were covered on the surface of the electrode along with the Zn²⁺ insertion to the V₂O₅@rGO-1.5 cathode (Supplementary Fig. 30a2), which may be caused by the generation of by-product like Zn₄(CF₃SO₃)₄(OH)₄·3H₂O. Notably, as-formed film can be gradually decomposed with the increase of temperature difference from 5 to 30 K (Supplementary Fig. 30a3–a8). Moreover, the graphite layers are covered on the surface Zn-G (Supplementary Fig. 30b1), implying the successful modification of pure Zn anode³⁷. Even discharged to 0.2 V (Zn²⁺ ions are stripped from the Zn-G anode), the graphite layers still can be maintained on the surface of Zn-G (Supplementary Fig. 30b2). Worthily, the morphology of Zn-G anode during the whole thermoelectrochemical process changes obviously. When the temperature difference is 5 K (Supplementary Fig. 30b3), some byproducts nucleate on the surface of Zn-G anode with the plating of Zn²⁺. Furthermore, the Zn-G anode displays a well-defined array structure formed by numerous thin nanosheets (Supplementary Fig. 30b4–b8). We also conducted structural characterizations of V₂O₅@rGO-1.5 cathode during various charge/discharge states with ex-situ XRD and XPS analyses to investigate such processes. Specifically, the V₂O₅@rGO-1.5 electrode at different voltages is marked as adopted temperature differences (x K) and state y, as depicted in Fig. 4a. It should point out that the sharp peaks at ~26.4° and 54.5° in all XRD patterns correspond to the graphite substrate (Fig. 4b and c). Compared to the initial electrode, the pattern of the electrode at 0.2 V after the first discharge shows the obvious right shift of characteristic peak, suggesting the transformation of the electrode from V₂O₅·1.6H₂O to Zn_xV₂O₅·nH₂O due to the co-insertion of Zn²⁺ and H₂O⁴⁶. During the thermal charge process, the (001) plane of Zn_xV₂O₅·nH₂O gradually shifts to a low-angle region and shows a tendency to change back to V₂O₅·1.6H₂O (Fig. 4b), indicating the thermoextraction of Zn²⁺/H₂O from crystals. Moreover, the formed Zn_xV₂O₅·nH₂O almost fully recovers as V₂O₅·1.6H₂O (state I) after the following electrochemically charging process to 1.6 V (Fig. 4c). Notably, the reversible shift of (001) plane from state I to state V well confirms the highly reversible conversion between Zn_xV₂O₅·nH₂O and V₂O₅·1.6H₂O, demonstrating high reversibility for energy storage. This phenomenon also suggests the co-insertion of Zn²⁺ and H₂O into V₂O₅·1.6H₂O during the discharging process, which then adjusts back to V₂O₅·1.6H₂O owing to the extraction of ions during the following charging process. Notably, the weak peak around 17.1° during the thermoelectrochemical process can be attributed to the presence of Zn₄(CF₃SO₃)₄(OH)₄·3H₂O, confirming the formation of byproduct on electrode surface^42,51,52.

Fig. 4. The mechanism analyses of V₂O₅@rGO-1.5 based ZTCCs.

a The voltage change curves at different processes. b, c The corresponding ex-situ XRD patterns of V₂O₅@rGO-1.5 cathode during b thermal charging states at various temperature differences and c electrochemically charging/discharging states at 30 K. d–f The corresponding ex-situ XPS spectra of d Zn 2p, e V 2p, and f O 1 s.

The chemical states and electronic interaction during fully charge and discharge stages are further investigated by using XPS measurements (Fig. 4d–f). As shown in Fig. 4d, two strong peaks located at 1022 and 1045 eV from state III can be assigned to the Zn 2p_3/2 and Zn 2p_1/2 signals, which is reasonably caused by the insertion of Zn²⁺ into the vanadium oxide layers. After charging to state V, the Zn 2p signals with lower intensity indicate that a lot of Zn²⁺ can be reversibly extracted, while partial residual zinc species still present in V₂O₅@rGO-1.5 cathode³⁵. When considering the valence change of V element during electrochemical processes, the electrode at state III shows a strong V⁴⁺ peak together with a high V⁴⁺/V⁵⁺ ratio of 1.06, meaning a significant reduction of V₂O₅ with the insertion of Zn²⁺ (Fig. 4e). Upon recharging to state V, the V⁵⁺ peak becomes strong and the V⁴⁺/V⁵⁺ ratio (0.31) is almost recovered to the initial state (0.23), in consistent with the extraction of Zn²⁺. In addition, the content of H₂O in O 1 s spectra exhibits significant change after discharge and charge (Fig. 4f), confirming the solvation effect endowed insertion of H₂O associated with the major insertion of Zn²⁺⁴³. According to the above results, the possible electrochemical reactions can be proposed as follows:

Cathode:

$${\mathrm{V}}_2{\mathrm{O}}_{5}1.6{\mathrm{H}}_2\mathrm{O}+\mathrm{xZ}{\mathrm{n}}^{2+}+4{\mathrm{H}}^{+}+\left(\mathrm{n}-1.6\right){\mathrm{H}}_2\mathrm{O}+2\mathrm{x}{\mathrm{e}}^{-}\leftrightarrow {\mathrm{H}}_4\mathrm{Z}{\mathrm{n}}_{\mathrm{x}}{\mathrm{V}}_2{\mathrm{O}}_5\cdot \mathrm{n}{\mathrm{H}}_2\mathrm{O}$$ 3

$$4{\mathrm{H}}_2\mathrm{O}\leftrightarrow 4{\mathrm{H}}^{+}+4\mathrm{O}{\mathrm{H}}^{-}$$ 4

$$2\mathrm{Z}{\mathrm{n}}^{2+}+{2\mathrm{Zn}\left(\mathrm{C}{\mathrm{F}}_3\mathrm{S}{\mathrm{O}}_3\right)}_2+4\mathrm{O}{\mathrm{H}}^{-}+3{\mathrm{H}}_2\mathrm{O}\leftrightarrow \mathrm{Z}{\mathrm{n}}_4{\left(\mathrm{C}{\mathrm{F}}_3\mathrm{S}{\mathrm{O}}_3\right)}_4{\left(\mathrm{OH}\right)}_4\cdot 3{\mathrm{H}}_2\mathrm{O}$$ 5

Anode:

$$\mathrm{xZn}\leftrightarrow \mathrm{xZ}{\mathrm{n}}^{2+}+2\mathrm{x}{\mathrm{e}}^{-}$$ 6

Demonstration of wearable devices

Quasi-solid-state devices shows great promise in multi-functional applications due to their flexibility, safety, and wearability⁵³. Here, the wearable ZTCC was using polyacrylamide (PAM) based gel electrolyte, which is sandwiched by Zn-G anode and V₂O₅@rGO-1.5 cathode, as illustrated in Supplementary Fig. 31. It is worth mentioning that the temperature difference between two electrodes is generated by a resistive heater (12 V, 7 W) attached on the cathode side. The as-assembled device delivers the output voltage of 0.38 V at an ultralow temperature difference of 8 K among other solid-state devices (Fig. 5a). After deducting the self-charging contribution (Supplementary Fig. 32), the thermal-induced voltage is 55, 74, 89, 101, 111, 117, 126, and 122 mV at the temperature difference from 1 to 8 K, respectively. As summarized in Fig. 5b, the corresponding Seebeck coefficient of V₂O₅@rGO-1.5 based solid-state device is calculated to be around 11.9 ± 1 mV K⁻¹, reflecting good thermal-electrical response in near room temperature low-grade heat harvesting. Such value also is much higher than that of rGO (10.4 ± 1 mV K⁻¹), V₂O₅ (8.9 ± 1 mV K⁻¹), V₂O₅@rGO-1.0 (5.2 ± 0.7 mV K⁻¹), and V₂O₅@rGO-2.0 (9.4 ± 0.7 mV K⁻¹) based devices. Remarkably, the highest thermal-induced power density of 0.12 W m⁻² can be obtained by V₂O₅@rGO-1.5 among all as-fabricated solid-state devices (Fig. 5c), signifying an ultrahigh normalized power density of 1.9 mW m⁻² K⁻² in state-of-the-art reported systems (Supplementary Table 1). Inspired by such impressive performances of V₂O₅@rGO-1.5 based solid-state ZTCC, an external load with a resistance of 10 kΩ was employed to examine its stability and durability in energy conversion. As displayed in Fig. 5d, the output voltage was stably maintained at ~0.4 V with the adoption of temperature gradient, implying the satisfying thermal stability and promise potential of solid-state ZTCC. However, the relatively high voltage decay of rGO and V₂O₅ based solid-state ZTCC are possibly caused by the poor capacity and sluggish ion diffusion kinetics, respectively. In contrast, the slight voltage changes of V₂O₅@rGO-1.0 and V₂O₅@rGO-2.0 based solid-state ZTCC could be attributed to their relatively low ion diffusion coefficient. Even after 7 days, the V₂O₅@rGO-1.5 based solid-state ZTCC still can realize the thermal charge and discharge processes (Supplementary Fig. 33). This result demonstrates that as-developed ZTCCs can be used repetitively for relatively long-term cycling rather than being a one-time energy source. As a demo, three ZTCCs connected in series can easily light up a white light-emitting diode (LED) with a temperature difference of ~10 K (Fig. 5e). When considering the temperature difference between body heat and ambient, V₂O₅@rGO-1.5 based solid-state ZTCC is very attractive as one of power input source to replace traditional batteries for integrated health monitoring systems (Fig. 5f). Figure 5g shows the digital photo of a brief health monitoring system. Typically, the selected electro-cadiography sensor module (BMD101) can be powered by three ZTCCs connected in series. By Bluetooth wireless, the recorded data can be presented and processed in the mobile phone or laptop (Fig. 5h). All above-mentioned demonstrations further confirm the application prospect of as-constructed ZTCCs in energy conversion and storage fields, which makes it possible in terms of electronics integration.

Fig. 5. Construction of quasi-solid-state ZTCCs using PAM based gel electrolyte and applications.

a Thermal charging curves under various temperature differences. b Fitted Seebeck coefficient. c Thermal-induced current density (hollow symbol) and power density (solid symbol). d Thermal stability with a loading resistance of 10 kΩ. e Digital photo of a white LED lighted by three ZTCCs in series. f Illustration for health monitoring system. g Body heat-charged health monitoring system and h the recorded data.

Discussion

In summary, this work has developed high-performance ZTCCs by cathode engineering. By sequentially introducing 2D rGO into 1D V₂O₅·1.6H₂O nanobelts, both the thermal response and kinetics of as-constructed ZTCCs are modified. Notably, the V₂O₅·1.6H₂O nanobelts provide enough electroactive sites for dense Zn²⁺ storage, suggesting high capacity and energy density. The rGO layers wrapped around V₂O₅·1.6H₂O nanobelts can not only keep structural stability and inhibit V dissolution, but also accelerate the charge transfer, implying good durability and excellent power density. As a result, a giant thermopower of 23.4 ± 1.5 mV K⁻¹ can be achieved by V₂O₅@rGO-1.5 based ZTCC, which synergistically integrates the fast thermodiffusion and thermoextraction processes. Moreover, a high thermal-induced voltage of 0.72 V and normalized power density of 1.04 mW m⁻² K⁻² can be delivered at the temperature difference of 30 K. The ex-situ experiments and DFT calculations reveal that the rGO can boost the charge transfer at the heterointerfaces formed by rGO and V₂O₅·1.6H₂O. Besides, the crystal water layer in V₂O₅·1.6H₂O can promote the diffusion of Zn²⁺ with low energy barrier due to its reduced electrostatic repulsion effect. It is worth mentioning that as-proposed V₂O₅@rGO-1.5 based ZTCC shows great promise in self-power supply for health monitoring systems. By coupling with hydrogel electrolyte, three solid-state ZTCCs attached to the arm can drive the Bluetooth wireless module by an ultralow temperature difference formed between body heat and atmosphere, demonstrating potential application of ZTCCs in energy conversion and storage as well as wearable electronics areas.

Methods

Preparation of V₂O₅@rGO-x

To prepare V₂O₅@rGO-x, the concentration of commercial graphene oxide (GO, CARMERY, Institute of Coal Chemistry, Chinese Academy of Sciences) solution was firstly adjusted to 1.0, 1.5, and 2.0 mg mL⁻¹. Then, 120 mg of vanadium pentoxide (V₂O₅, Xiya Reagent) was dissolved in 60 mL of GO solution and stirred magnetically for about 30 min at room temperature. Subsequently, above solution was transferred into a 100 mL Teflon-lined stainless steel autoclave and kept at 180 °C for 12 h. Finally, the products were collected by centrifugation at 9166 RCF xg for 3 min and washed several times using deionized water and ethanol, followed by drying overnight. The samples are labeled as V₂O₅@rGO-x, where x represents the GO concentration. For comparison, we carried out control experiments with pure GO solution (1.5 mg mL⁻¹) and V₂O₅ in water by the same conditions to obtain rGO and V₂O₅, respectively.

Material characterizations

The morphology was observed by using a field emission scanning electron microscope (FESEM, FEI NANO SEM430) and a high-resolution transmission electron microscope (HRTEM, JEOL JEM-2100). X-ray diffraction (XRD) patterns were recorded from PANalytical Empyrean diffractometer with Cu Kα radiation (λ = 1.5406 Å). The surface chemical states of samples were detected by X-ray photoelectron spectrometer (XPS, Thermo Scientific K-Alpha). The Raman spectra were collected by the Horiba Scientific LabRAM HR with an excitation wavelength of 532 nm. Thermogravimetric analysis (TGA) curve was recorded on a Mettler-Toledo STARe SW 15.00 analyzer with a heating rate of 10 °C min⁻¹ under air flow. The wettability of materials was measured by a CA tester (XG-CAMC33) using electrolyte drop (10 μL). The dissolution rate of vanadium into electrolyte was detected by Inductively Coupled Plasma Optical Emission spectroscopy (ICP-OES, Thermo Fisher iCAP PRO). The SSA of samples were measured by the ASAP 2460 analyzer (micromeritics), and were analyzed by Brunauer-Emmett-Teller (BET) method.

Electrochemical measurements

Thermoelectrochemical performance tests of ZTCCs were conducted in a non-isothermal H-type cell with 0.5 mol L⁻¹ Zn(CF₃SO₃)₂ electrolyte on a standard electrochemical workstation (CHI 760E). Zn-G foil (thickness: ~20 μm) and as-prepared electrodes were used as counter/reference electrodes and working electrodes, respectively. The Zn-G anode was modified by graphite using the pencil drawing method. To prepare the working electrode, V₂O₅@rGO-x, acetylene black, and polyvinylidene fluoride were firstly mixed together in a mass ratio of 7:2:1. Then, the slurry was painted on graphite paper with a diameter of 1.2 cm (mass loading: 1.2 mg cm⁻²) as working electrode. CV curves of ZIBs were recorded from 0.1 to 1.0 mV s⁻¹ with a voltage window of 0.2–1.6 V by the Biologic VMP-300 workstation. GITT, rate capability, and cyclic stability were measured using the CT3001A Land Battery Test System.

The quasi-solid-state ZTCC was assembled by Zn-G foil anode and V₂O₅@rGO-x cathode together with hydrogel electrolyte. It should mention that both the Zn-G and V₂O₅@rGO-x electrodes hold the area of 2×3 cm², and the active mass loading in the V₂O₅@rGO-x electrode is about 8 mg. In addition, the PAM gel electrolyte was prepared by following steps. 2 g of acrylamide (AM, Macklin) and 0.4 g of acrylic acid (AA, Macklin) were dispersed into 10 mL mixed solvent of deionized water and glycerol in a volume ratio of 1:2. Subsequently, 1.5 mg of N,N′-methylenebisacrylamide (Macklin), 10 μL of N,N,N′,N′-tetramethylethylenediamine (Aladdin), and 0.25 g of potassium persulfate (Aladdin) were added into above solution part by part with vigorous stirring for 4 h under ice-water bath. After that, the transparent solution was injected in a mold and polymerized by using a UV lamp (365 nm, 60 W) for 20 min to prepare the hydrogel matrix. Moreover, such hydrogel soaked with 0.5 mol L⁻¹ Zn(CF₃SO₃)₂ electrolytes can be used for the construction of quasi-solid-state ZTCCs.

Author contributions

Z.L. carried out the experiment. Z.L., Y.X. and J.C. evaluated the electrochemical performances. L.W. analyzed the structural characterization and electrochemical results. Z.L. and Y.X. organized the figures. H.D. and X.Z. supervised the whole work. All authors co-discussed the results and wrote the manuscript.

Peer review

Peer review information

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

Data availability

All relevant data that support the findings of this study are presented in the manuscript and supplementary information file. Source data are available from the corresponding author upon reasonable request.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-023-42492-z.