To address the aforementioned issues, this work presents a design for a fully foam asymmetric HEGs that is independent of the direction of water evaporation. The cathode uses foamed iron (FI) as the main metal electrode, preventing water molecules from evaporating only from one side of the HEG. The FI electrode is reinforced with the Co2+ metal-organic framework (ZIF-67), a material with a stable preparation method43,44,45. On one hand, it can enhance the surface potential to increase the voltage; on the other hand, it provides Co2+ as charge carriers, improving the current density. First-principles calculations and finite element simulations show that the addition of ZIF-67 effectively enhances the interaction between H2O and the electrode, providing a foundation for the improvement of current-voltage performance. The anode employs a CNTs-modified carbon cloth (CC) electrode (CC@CNTs), and the two electrodes are separated by melamine foam (MF). The structure is stacked to form a FI/ZIF67@CMF-MF-CC@CNTs type HEG unit. HEGs composed of FI/ZIF67@CMF exhibit an open-circuit voltage of ~0.782 V and a current density of 862 μA/cm2 at room temperature, increasing by ~50% and ~300%, respectively, compared to iron sheet electrodes. The device achieves optimal output power of 101 μW/cm2 at a 999.9 Ω load. Additionally, the device is flexible, easily bent to form wearable devices such as rings or bracelets, providing power for devices like LEDs and watches, showcasing its potential in wearable device applications. It also demonstrates high integration stability; both voltage and current increase linearly, and the generated power can be easily stored using capacitors or batteries, potentially leading to its development into self-powered supercapacitors or batteries.

The schematic diagram shown in Fig. 1a illustrates the HEG device, while the actual device and energy spectrum analysis are presented in Fig. S1 (SI). The carbon electrode (CC@CNTs) and the composite foam iron electrode (FI/ZIF67@CMF) are separated by a 0.5 mm MF, resulting in the asymmetric structure designated as FI/ZIF67@CMF-MF-CC@CNTs. The surface area of the carbon electrode is significantly increased after acidification and coating with CNTs, where the acidified CNTs wrap around the CC surface (Fig. S2, SI), effectively increasing the number of oxygen-containing groups on the CC surface, transforming the treated CC into a hydrophilic material (Fig. S3, SI). MF is employed as an isolation layer between the two electrodes (Fig. S4, SI). The porous structure of MF can store several times its own mass of water (Fig. S5, SI), thus extending the stable operating time of HEGs.

The preparation method of FI/ZIF67@CMF is illustrated in Fig. S6 (SI), where FI is initially acidified and passivated to enhance its specific surface area and form a dense oxide film to prevent oxidation. Through X-ray photoelectron spectroscopy (XPS) tests (Fig. S7, SI), the FI surface after strong acid passivation treatment simultaneously formed Fe(II) and Fe(III). Combined with X-ray diffraction (XRD) analysis (Fig. S8, SI), it can be confirmed that the passivated surface of FI forms a dense FeO layer. Subsequently, ZIF-67-deposited CMF is pressed onto the FI surface at 10 MPa for 1 min to form the FI/ZIF67@CMF composite electrode. The N adsorption-desorption isotherms are shown in Fig. S9 (SI). The BET surface area and total pore volume of the composite FI are 1.7378 m/g and 0.010295 cm³/g respectively. Fig. 1b depicts the schematic diagram of the FI composite electrode, where ZIF-67@CMF does not completely fill all the pores of FI but forms a unilateral filling. Under pressure, ZIF-67@CMF fractures and embeds into the gaps of FI (Fig. 1d-f), further increasing the specific surface area of the FI composite electrode, enhancing the effective area for interaction with water molecules. Firstly, we conducted Raman and Fourier-transform infrared analysis (FTIR) on the surface (Fig. 1g, h), confirming that the MF underwent sufficient carbonisation and that the attachment of ZIF-67 provided additional nitrogen and oxygen functional groups to the electrode surface. XPS characterisation of the FI composite electrode surface (Fig. S10, SI) revealed a significant presence of N-O bonds, C-O bonds, and oxygen functional groups (-OH), effectively promoting interaction between the electrode and HO, thereby enhancing the device's current density. Additionally, XRD characterisation of the FI composite electrode surface (Fig. 1i) indicated that a passivation layer of FeO formed on the treated FI surface, achieving stable connection with ZIF-67@CMF.

After thoroughly wetting the electrode by dripping water, the different components of the FI/ZIF67@CMF composite electrode are assembled with the CC@CNTs to form the HEG device, and its output electrical signal is tested (Fig. 1c). The current density of the HEG using only carbon materials is only 141 nA/cm and 328 nA/cm (Fig. S11, SI), much lower than that of the HEG with FI as the cathode. However, by applying pressure to form a composite electrode, the water channels on the surface of FI are further enhanced (Fig. S12, SI), increasing the interaction area between HO and the electrode, which improves the HEG current density. Further addition of ZIF-67, on the one hand, reduced the surface Zeta potential of FI (Fig. S13a, SI) and compensated for the decrease in the FI electrode potential caused by passivation (Fig. S13b, SI), thereby increasing the potential difference between the asymmetric electrodes and further improving the open-circuit voltage. On the other hand, the Co in ZIF-67 can act as charge carriers between the electrodes, boosting the HEG current density. Compared to the FI-MF-CC@CNTs HEG, the current density of the FI/ZIF67@CMF-MF-CC@CNTs HEG increases from 206 μA/cm to 832 μA/cm, and the open-circuit voltage rises from 0.5 V to 0.78 V. The current density increased by 300%, and the open-circuit voltage increased by 50%.

Fig. 2a-c) illustrates the surface charge densities of FI, ZIF-67, and FI/ZIF-67. With reference to the scale in the figure, it can be observed that in the presence of FI alone, the electron distribution on the Fe surface is uniform due to its metallic nature, resulting in a relatively low surface charge density and an overall neutral charge. In contrast, ZIF-67 exhibits characteristics of a metal-organic framework, where the surface charge is related to its elemental composition. Oxygen atoms display a slight negative charge, while carbon atoms show a slight positive charge; however, the overall structure remains electrically neutral. When ZIF-67 comes into contact with FI, a significant charge transfer occurs, with electrons from the FI surface transferring towards the oxygen, leading to a distinctly positive charge on the Fe surface and a notable negative charge on the oxygen surface. Given that the other electrode is CC@CNTs, the higher negative potential will facilitate the generation of higher voltages in the HEGs. Additionally, the increased reactivity of the oxygen atoms enhances the likelihood of forming hydrogen bonds with HO, thereby promoting interactions between the electrodes and HO. A similar scenario occurs on the CMF surface (Fig. S14, SI), where deposition of ZIF-67 leads to further electron transfer and hybridization of surface charges, resulting in an elevated potential on the ZIF-67@CMF surface. However, the electron activity on the C surface is much lower than on the FI surface, resulting in a less significant increase, consequently leading to less pronounced voltage enhancement in the resulting HEGs. From (Fig. 2d-f), it can be observed that the combination of FI and ZIF-67 leads to hybridization of internal d and f orbitals, resulting in changes in the FI electrode surface potential. Before the combination, Fe atoms have lower charge density, and electrons in each orbital are in a relatively low state, while ZIF-67, despite having higher charge density, cannot effectively transfer electrons without a collector. After the combination, FI acts as a superior collector, and the composite of FI and ZIF-67 improves the overall surface potential, forming a composite electrode with both high surface potential and excellent electron conduction ability. As a transition metal, Fe has highly active d and s orbital electrons. After hybridization with the s and d orbital electrons of elements such as C and O in ZIF-67, it is further strengthened, effectively increasing the potential difference between the two electrodes, thus resulting in a higher open-circuit voltage.

Fig. 2g and Fig. 2h illustrate the electrostatic field distribution between the FI composite electrode and the carbon electrode under wet and dry conditions, respectively. The polar water molecules transform the initially uniform electric field into a wavy pattern. As the polar water molecules move between the electrodes, the electrostatic field undergoes continuous fluctuations, generating an electrical signal. At the same time, under the influence of the electrostatic field, ions such as Co, HO, and OH in the water undergo directional movement, which can also produce an electrical signal in the external circuit. Through the combined action of both factors, the HEG is able to generate a continuous electrical signal.

The process of interaction between water and the electrodes can be broken down into three steps (Fig. 3a). Step 1, When the electrodes are not in contact with HO, there is a stable potential difference between the electrodes, but without the help of charge carriers, no current can be detected in the external circuit. Step 2, when water comes into contact with both electrodes, it interacts with the electrode surface, causing electron transfer on the electrode surface and forming a double layer (Fig. S15, SI). At the same time, the relevant ions in the electrodes undergo hydrolysis, generating charge carriers. Step 3, water moves between the electrodes through processes like evaporation. The hydrogen bonds of the water molecules drive the charge carriers to move between the electrodes. In conjunction with the schematic diagram of the two-dimensional charge carriers (electrons and ions) movement (Fig. S16, SI), after deionized water is dropped onto the HEG device, the hydrolysis-generated Co, HO, and OH ions will form ionic charge carriers that move within the HEG. Under the potential difference between the electrodes, positively charged ions move towards the CC@CNTs electrode, while negatively charged ions move towards the FI/ZIF67@CMF electrode. As the ions migrate within the internal circuit, FI/ZIF67@CMF releases electrons in the external circuit and moves towards the CC@CNTs electrode, thereby completing the charge circuit formed by the ion migration within the internal circuit.

To further increase the surface area of the FI, acidification-passivation treatment was performed on the FI before preparing the composite electrode. The treated FI surface no longer exhibits a smooth curvature but rather shows numerous nanochannels for water (Fig. S17a, SI), combined with further analysis by energy dispersive spectroscopy (EDS) revealing a uniform oxide film (Fig. S17b-c, SI). An increased number of nano- to sub-nano-sized water channels can effectively enhance the interaction area between HO and the electrodes. Additionally, under capillary action, water molecules can move more quickly across the electrode surface, facilitating evaporation and thereby increasing the current density of HEGs. To quantify the movement of HO within the water channels on the electrode surface, we employed a relationship between the grayscale in SEM images and the surface morphology of the materials. Using binary colour coding, we identified the water channels (Fig. S18a, SI) and conducted finite element analysis of the microscale water flow based on the Stokes flow equation. Fig. 3b presents the simulated results of water flow on the FI surface based on Stokes flow approximation. According to the computational fitting results, the water flow velocity in narrow regions is several times higher than that in wide regions. If it were a metal sheet, it could be considered as having an infinitely wide surface on the microscale, where water molecules would remain stationary after being affected by gravity, rather than undergoing further movement under the influence of surface tension, as on the FI surface. This implies that the rate of interaction between FI and water molecules is much higher than that with metal sheets. Furthermore, based on the Brinkman assumption satisfying the Stokes equation, an analysis of the model was conducted (Fig. S18b, SI), where the inlet and outlet positions of the water flow were exchanged under the assumption of incompressible water flow. Comparing the results with Stokes flow revealed similar pressure or velocity curves, further confirming that the microstructure effectively promotes the movement of water molecules. This also explains why the current density of the foam iron is much higher than that of metal sheets, not only because the surface area of the foam iron is higher than that of iron sheets but also because the rate of water molecule movement on the electrode surface is faster.

ZIF-67 not only enhances the electrode surface potential but also increases the interaction frequency between water molecules and the electrode. To simulate the device's performance after sufficient wetting, water molecules are placed in clusters on the material's surface, directly mimicking the evaporation process. Fig. 3e, f illustrate the movement of water molecules on the ZIF-67 (111) and Fe (111) surfaces over different time intervals. Comparatively, the interaction between ZIF-67 and water molecules is significantly more intense than that with Fe. The oxygen and nitrogen-containing groups on the ZIF-67 surface can form hydrogen bonds with water molecules, while Fe struggles to do so. Additionally, as a MOF material, ZIF-67 has pores that allow water molecules to penetrate deeper and interact with internal oxygen-containing groups, thereby promoting increased current density. In contrast, Fe, being a dense metallic crystal, restricts water molecules from entering and interacting with the material. Further comparisons of energy changes in both models over 10 ps (Fig. S19, SI) reveal that, starting from similar energy levels at 0 ps, the energy fluctuations in the ZIF-67 model are notably higher than in Fe, indicating a stronger interaction with HO. Moreover, we recorded the changes in the number of hydrogen bonds and the number of water molecules dissipating over 10 ps (Fig. 3c,d). The ZIF-67 model exhibits ten times the number of hydrogen bonds compared to the Fe model. In the ZIF-67 model, both HO-HO and HO-ZIF67 hydrogen bonds can form, whereas in the Fe model, only HO-HO bonds are present. Although the number of dissipated HO molecules is similar in both models over the same period, the greater variability in hydrogen bond numbers in the ZIF-67 model suggests that it can frequently form and break hydrogen bonds with HO. This direct interaction implies that ZIF-67 can enhance the energy collection during the water evaporation process, effectively increasing the device's current density. In summary, by combining ZIF-67 with Fe to form a composite electrode, both the surface potential of the electrode and the acquisition of evaporation energy are improved, resulting in a higher open-circuit voltage and increased short-circuit current for the device.

Since HEG generates electricity through the interaction between HO and the electrodes, environmental factors directly affect the strength of the electrical signal. A 2 × 2 cm HEG was fabricated as shown in Fig. S20 (SI), and environmental changes were controlled using a constant temperature and humidity system (Fig. S21, SI). At different temperatures, the current density of the HEG changes significantly. As the ambient temperature rises from 30  °C to 60  °C, the current density increases from 0.78 mA/cm to 1.5 mA/cm, while the voltage remains relatively stable (Fig. 4a). This could be due to the higher evaporation rate of water with increasing temperature, leading to more frequent charge transfer between the water and the electrodes. On the other hand, the increased temperature may also intensify water hydrolysis, increasing the carrier concentration between the electrodes, resulting in a significant increase in current density. Meanwhile, as the temperature increases, although the Zeta potential and electrode potential of different electrode materials change to varying degrees, both the Zeta potential difference and electrode potential difference remain relatively stable compared to the changes in current (Fig. S22, SI). Therefore, the device voltage does not exhibit significant changes with temperature variation. When the ambient humidity is lower (Fig. 4b) and wind speed is higher (Fig. 4c), water evaporation intensifies, and the current density increases significantly, while the voltage remains relatively stable (Fig. S23, SI). This further confirms that the water evaporation rate is positively correlated with current density, but has no significant effect on voltage.

Fig. 4d shows the variation in current density of the HEG device under different O concentrations. As shown in Fig. S24a (SI), to examine the power generation of the HEG device in an anaerobic environment, Ar was continuously pumped into the glove box. In the absence of O, the current density of the HEG device remained at ~10 μA/cm due to the proton flow generated by the slow movement of HO in narrow channels and the changes in electron delocalization caused by evaporation (Fig. S24b, SI). As the O concentration increased, the current density gradually increased. When the O concentration exceeded 95%, the current density even surpassed 2 mA/cm. Molecular dynamics simulations (Fig. S25a, SI) revealed that the presence of O enhanced the interaction between HO and the electrode surface, significantly increasing the number of hydrogen bonds under the influence of O, thus promoting electron transfer between the electrode and HO. Additionally, differential charge density simulations (Fig. S25b, SI) showed that the electron activity at the Fe@HO@O three-phase interface was significantly higher than that at the Fe@HO two-phase interface, further confirming that O effectively enhanced the electrical signal.

Importantly, the use of the FI composite electrode addresses the issue of unidirectional evaporation in the device, effectively promoting the evaporation rate of water molecules. When FI/ZIF67@CMF and CC@CNTs are used, one acts as the bottom electrode and the other as the top electrode, without the concern that a dense electrode would block water evaporation and cause the HEG to stop working. As shown in Fig. S26a (SI), when HEG is rotated at different angles, it is considered to have the FI/ZIF67@CMF electrode as the top electrode when the angle exceeds 90°. During rotation, HEG consistently produces stable electrical signals (Fig. 4e). As the rotation angle increases, the binding force between the two electrodes decreases under the influence of gravity. Once the angle exceeds 90°, the binding force is improved, and the current density increases again. Furthermore, the effective evaporation area is a key factor affecting HEG's output current (Fig. 4f). When the HEG device is placed in a beaker and submerged to different depths (Fig. S26b, SI), if the HEG is fully exposed to air, it relies solely on the capillary action of MF to absorb water from the beaker. With too few water molecules between the two electrodes, the current density remains low. As the submersion depth increases, the current density gradually increases. However, after exceeding 50%, the effective evaporation area decreases, and the current density declines as well.

In the natural environment, both carbon and FI electrodes convert solar energy into heat energy, which promotes water evaporation. One of the FI/ZIF67@CMF and CC@CNTs was used as the top electrode and the other as the bottom electrode, and the photo-thermal and power generation were examined under dry-humid conditions, respectively. Fig. 4g shows the current-voltage variations in different conditions. The inset in the figure shows the schematic diagram of the placement of the HEGs device, which is placed at a position of 10 cm away from the xenon lamp light source, so that the device can receive 1000 W/m of light intensity steadily. Under dry conditions, after 60 s of illumination, the device with CC@CNTs as the top electrode reached a temperature of 84.5 °C (Fig. S27a, SI), whereas with FI/ZIF67@CMF as the top electrode, the temperature was only 62.2 °C due to metallic material reflection (Fig. S27b, SI). However, no current was observed either before or after illumination, and heating the device to 60 °C also failed to produce any current (Fig. S28, SI). Conversely, when the device was in a humid state, significant currents were generated regardless of whether FI/ZIF67@CMF or CC@CNTs were used as the top electrode, the device with FI/ZIF67@CMF as the top electrode outputs a current of 96.5 μA, while the device with CC@CNTs as the top electrode outputs a current of 56.3 μA. After 60 s of illumination, due to sunlight absorption and reflection by water, the temperature of the HEGs with CC@CNTs as the top electrode was 39.4 °C (Fig. 4h), while with FI/ZIF67@CMF as the top electrode, it was 36.8  °C (Fig. 4i), both lower than in dry conditions. However, short-circuit currents were significantly enhanced, reaching 135 μA in both cases. These experiments demonstrate that the current is not a photocurrent caused by light illumination, nor a thermoelectric current induced by heating. Instead, it is a current generated by water evaporation. Furthermore, water molecules can evaporate from any electrode (whether it be FI/ZIF67@CMF or CC@CNTs), and the resulting current is independent of the evaporation location, consistently producing stable current.

To verify the device's stability, HEGs were placed under ambient conditions (21 °C <T <25 °C, 20% <RH <25%) for long-term testing (Fig. 5a). Within 10 h, the output voltage reached up to 782 mV, with a current density of up to 862 μA/cm, demonstrating high stability and showing potential for practical applications. XRD testing was performed on the FI/ZIF67@CMF electrode after working for 10 h (Fig. S29, SI). The results were consistent with those before operation, and the decrease in the ZIF-67 peak intensity was caused by the hydrolysis of Co. To confirm whether electrochemical corrosion occurs in the electrode material during use, the C-V curves of the HEG devices were tested at different scan rates. No oxidation-reduction peaks were observed in the discharge range (0-0.8 V) for all devices (Fig. 5b), confirming that no significant corrosion occurred in the electrodes during use. To further identify the potential negative voltage range where oxidation reactions might occur and to assess the stability of the device under acidic conditions, different HEG devices were tested in the range of -1.0 V to 1.0 V. The results show that the HEG device made with the original FI electrode, similar to the passivated FI/ZIF67@CMF composite electrode in pure water, did not exhibit any oxidation-reduction peaks (Fig. S30a,b, SI). However, in an acidic environment (pH=1), the original FI electrode generated distinct oxidation peaks, and as the test progressed, its surface morphology continuously changed, leading to further variations in the C-V curve (Fig. S30c, SI). In contrast, the HEG device made with the passivated FI/ZIF67@CMF composite electrode maintained a smooth C-V curve (Fig. S30d, SI), demonstrating that the composite electrode remains stable under different conditions. HEG integration can be easily achieved through series and parallel configurations. A small 1 × 1 cm unit of HEG was fabricated (Fig. S31, SI) and fixed to the front of a glass slide. By altering the circuit on the back of the glass slide, the conversion between series and parallel configurations can be achieved (Fig. S32, SI). As the number of parallel units increases (Fig. 5c), the HEG module's current rises from 0.65 mA to 3.6 mA, while the voltage of the HEG module increases from 0.71 V to 3.39 V as the number of series units increases (Fig. 5d). Fig. 5e shows the changes in current density, voltage, and power of HEG under different loads, with the power density reaching a peak value of 101 μW/cm when the load is 999.9 Ω. Compared to other recent studies (Fig. 5f), the FI/ZIF67@CMF-MF-CC@CNTs-based HEG achieves a significant increase in power generation. A detailed comparison of materials, testing environments, and other factors can be found in Table S1,2 (SI).

To verify the general applicability of the optimized approach, other metal sheet electrodes (such as Cu, Ni, and Zn) were further optimized. By switching to metal foam and forming composites with ZIF-67, both the current density and voltage were improved to varying extents (Fig. 6a). Although these metals did not show as significant an improvement as FI, the current density increased by an average of 58.17%, and the voltage increased by an average of 25.00%, demonstrating the universality of the ZIF-67-enhanced metal electrode strategy (Fig. S33, SI). No obvious redox peaks were observed for the HEGs made of different metals at different scan rates (Fig. S34, SI). The distance between the electrodes and the tightness of the connection will directly affect the output current of the HEGs. In the previously mentioned water immersion and deflection experiments, the connectivity between the two electrodes of the HEGs was also reduced due to buoyancy or gravity, resulting in lower current density. To further investigate the effect of connection sealing on the power generation of HEGs, we utilised a 50 g weight as a pressure source for further exploration (Fig. S35, SI). The current density of the device varied with the placement and removal of the weight as shown in Fig. 6b. To prevent water molecules from blocking the diffusion for a long time due to the weights, they were placed on the surface of the device and then removed. When the weights were not placed, the current density stabilised at 0.75 mA. However, on placement, the current density increased significantly, rising to a staggering 2.4 mA, further confirming the effect of the connection on the output of the device. In the absence of pressure, some of the pores in the porous MF are occupied by air, hindering current conduction. However, applying pressure to the surface displaces the air with evaporated water molecules, ensuring continuous evaporation and replenishment of water molecules on the surface of the material, thereby increasing the current density. Furthermore, pressing the device does not have a significant impact on the output voltage (Fig S36, SI), with the voltage decreasing by only 3.9% after pressing. This further confirms that external factors are unlikely to affect the device's voltage. Fig. 6c shows the V-t curves of individual devices charging commercial capacitors of different capacities. As the capacitor capacity increases, the charging time also increases, but all achieve rapid charging within 3 s, demonstrating that the electrical energy generated by HEGs can be easily harvested. To further simulate the potential solution environments for HEGs in practical applications, we used various concentrations of NaCl solution to mimic sweat (0.03 M), physiological saline (0.9 M), and seawater (3.5 M), alongside samples of natural seawater. In a still indoor environment, we tested the current output of the HEGs. Fig. S37 (SI) shows the average voltage and current density over 30 min in different solutions. As NaCl concentration increased, there was no significant change in voltage, confirming that external environments do not significantly impact voltage. However, current density increased with concentration, mainly due to the effective charge transfer between electrodes facilitated by hydrolysed Na and Cl. Compared to natural seawater, the equivalent seawater exhibits a higher current density because the larger ions like Ca, Mg, and SO in natural seawater hinder migration, affecting overall current density.

In order to achieve a high-power HEGs device, we integrated 15 HEGs units (1 × 1 cm) in series. As shown in Fig. 6d, for the purpose of circuit integration, the CC@CNTs and FI/ZIF67@CMF electrodes alternately serve as the top electrodes, with direct connections made using carbon tape. The voltage variation after the series connection is depicted in Fig. S38 (SI). Experimental results demonstrate that a series-integrated assembly of 4 units can light up a LED (Fig. 6e-Ⅰ), while 5 series-integrated units can illuminate a large LED (Figure e-Ⅱ). Furthermore, 15 series-connected units are capable of directly powering a timer (Figure e-Ⅲ) or an electronic wristwatch (Figure e-Ⅳ). Supplementary Movie 1 presents a video of the 15-series integrated assembly powering an electronic wristwatch, showing good stability through repeated power-on and power-off tests. HEGs can convert low-grade thermal energy into electrical energy, absorbing the body's emitted heat and sweat. As shown in Fig. 6f, four HEGs units were prepared in a wearable format through soft packaging. To facilitate connection, the HEGs were alternately arranged, with FI composite electrodes and carbon electrodes serving as top electrodes, connected directly by conductive carbon tape, and finally thermally encapsulated with PET. To ensure stable water adsorption-evaporation, the PET was laser-engraved for perforation. The prepared soft-packaged HEGs can be shaped through bending, and Fig. 6g shows images of the bent soft-packaged HEGs from different angles, demonstrating stable power generation after bending. Fig. S39 (SI) shows the changes in current and voltage output of the device before and after bending. There is no significant variation in either current or voltage, demonstrating that the soft-pack HEGs maintain stable current output even after deformation. When further bent and worn on a finger, the soft-packaged HEGs can light up a yellow LED (Supplementary Movie 2). These experiments also prove the practical value of HEGs in flexible wearable energy.