Introduction

Flexible electronics have produced a paradigm shift in the wearable technology sector1,2,3. Remarkable advancements were made in developing wearable sensors that are thin, conformal, and stretchable4,5,6. In particular, ultraflexible devices have garnered immense attention due to their ability to seamlessly integrate with the human skin7,8,9,10,11, providing more accurate data acquisition and an enhanced degree of user comfort.

Advances in wearable technology are highly dependent on the development of flexible energy devices, which should offer high efficiency, durability, and constant power output and possess the capacity for effortless integration12,13. Most commercial wearables rely on bulky coin cells or rechargeable batteries as power sources. These contribute significantly to the system’s overall rigidity, posing limitations to their mechanical compliance and often necessitating frequent charging or replacement.

The next frontier is to produce ultraflexible energy sources, especially flexible energy harvesting-storage systems (FEHSSs) that efficiently generate and store power, and adapt to curved surfaces and moving objects. While there has been remarkable progress in ultraflexible energy harvesters, integrating these with flexible energy storage devices to form an efficient and robust power system brings a set of challenges. These include achieving outstanding power conversion and storage efficiency without compromising the mechanical compliance, establishing a stable interfacial connection, and realizing human-friendly user interface.

There is growing interest and progress in integrated FEHSSs. Self-powered energy modules like thermoelectric generators, bioenergy harvesters that involve triboelectric and piezoelectric generators, and photovoltaic-supercapacitor combinations, have demonstrated early success in this field13,14,15,16. Documented devices typically exhibit a thickness in the range of 1 mm to 600 µm, with a power output below 5 mWh cm2 17,18. The current level of flexibility and efficiency within these systems is very limited, necessitating extensive enhancements to ensure their effective interplay with ultraflexible sensors, and the soft, dynamic attributes of the human body.

Organic photovoltaics (OPVs) featuring lightweight, biosafety, and typically short energy payback time are especially suitable for wearable applications19. Recent investigations have substantially escalated the power conversion efficiency (PCE) of flexible OPVs, propelling it beyond the 10% threshold20,21,22,23. Simultaneously, intense efforts have been made to enhance the performance robustness to withstand harsh outdoor conditions22,24,25,26. However, ultraflexible OPVs that exhibit the best performance are mostly individual units with a limited size. Ultraflexible OPVs and modules with an active area > 5 cm2 and an areal power output > 10 mW cm–2 are still quite rare. Besides, energy derived solely from OPVs has intermittent availability under fluctuating light conditions, which undermines their efficacy in integrated wearable systems that require constant power sources.

Lithium (Li)-ion or Li-polymer pouch cells serve as prevalently used energy storage devices. However, they have to be encapsulated in a strong foil to meet safety standards. Their irregular deformation during wear can lead to cracks and delamination of the battery electrodes, resulting in diminished electrochemical properties and posing a potential risk of toxic liquid electrolyte leakage27. For electronics to be worn by people, whether on-skin or in-textiles, the emphasis on safety is even greater. As a result, safer battery systems, e.g., zinc (Zn)-ion batteries28,29 (ZIB), would facilitate a user-friendly design in a thin configuration. To achieve ultra-flexibility in an integrated FEHSS, the thickness of each individual component needs to be tailored to balance the electrochemical performance, mechanical properties, and durability30. The use of gel and solid-state electrolytes in batteries has indubitably contributed to their flexibility. These electrolytes must simultaneously serve as the separator while preserving adequate ionic conductivity, an intricate task. In other words, developing ultrathin electrolytes poses an ongoing challenge due to the dual requirement of sustaining conductivity and managing the heightened risk of potential short-circuiting31. Therefore, there persists a demand for flexible energy storage solutions, particularly focusing on the development of thin, efficient, and safe electrolytes and electrodes.

The primary challenge in system integration involves reconciling the disparate components while accounting for their diverse physical, optoelectronic, and electrochemical properties. This process also seeks to generate high voltage output across both the energy harvesting and storage modules. A 650 µm-thick FEHSS was demonstrated, consisting of OPVs and a textile-based Li-ion battery17. The use of a textile battery indeed improved the conformability of the overall system compared to a foil battery. In a different approach, homo-tandem OPVs were used to double the open circuit voltage output to 1.92 V, facilitating the charging of Li-ion and Na-ion batteries32. A monolithically integrated photo-rechargeable power source was developed using Si photovoltaics and Li-ion batteries18. A bipolar stacked solid-state battery configuration was used, resulting in an overall voltage output of 5.4 V from the battery module. Despite the scarce efforts devoted toward merging the energy harvesting and storage components, their potential to enable compliant, efficient, and stable wearable technology remains largely unexplored.

In this work, we present a 90 µm-thick, highly efficient, fully integrated energy harvesting and storage system that meets the needs discussed above. It is made possible by ultrathin, high-performance OPV modules paired with thin Zn-ion batteries. First, the ultraflexible OPV cells have a PCE of up to 16.18%, a shelf lifetime exceeding 1500 h, and retain over 92 % of their initial PCE under a continuous illumination of 50 mW cm–2 for 500 h. The freestanding 4 µm-thick OPV modules have state-of-the-art areal power outputs exceeding 10 mW cm–2. Second, a notable innovation in the battery is that we have realized a drastic reduction of the hydrogel electrolyte thickness, from millimeters to 10 µm, while maintaining its high electrochemical performance. Third, we designed ultraflexible organic diodes with a rectification ratio in the order of 105 in the ± 2 V range, to effectively prevent undesired current reflux from the battery to the OPVs in dark conditions. The integrated system has an energy density greater than 5.82 mWh cm−2, and an overall conversion and storage efficiency of 6.91%, along with excellent operational and storage stability. Finally, we demonstrate an all-in-one energy harvesting and storage system to power wearable electronics, including wearable biosensors, small gadgets like smartwatch and smartphone, etc. Our FEHSS represents a potentially universal power source that can be effortlessly integrated on both human body and textiles, paving the way for a more connected, seamless experience with wearable technology.

Results

Design of the FEHSS

Figure 1A, B show an illustration of the structure and the material composition of our FEHSS. The ultraflexible OPV module (chemical structure of the active materials in Fig. 1C), with a total thickness of 4 µm, is positioned atop the flexible Zin-ion battery (composition detailed in Fig. 1D). The two components are connected in parallel by an anisotropic conductive film (ACF) tape. Ultrathin (1.5 µm) parylene films serve as the passivation layer for the OPV and the battery, effectively creating segregation between their active materials. The resulting FEHSS has a thickness of approximately 90 µm, one of the thinnest structures in documented energy systems with exceptional mechanical compliance. It can be feasibly attached to the human skin (Fig. 1E) or textiles (Fig. 1F). Bent to a radius of less than 1 mm for 500 cycles or being compressed to a strain of 10% for 100 cycles, the FEHSS maintains over 80% of its overall power conversion and storage efficiency (Fig. 1G), showing mechanical durability for practical applications.

Fig. 1: Structural overview and material composition of the FEHSS.
figure 1

A Illustration of the FEHSS functioning as a power source for an on-skin electrocardiogram (ECG) sensor patch for healthcare monitoring. B Schematic of the layered configuration of the flexible OPV module and Zn-ion batteries. C The chemical structures of the primary donor (PM6) and acceptor (Y6) materials, and the third component (O-IDTBR), forming the active layer of the flexible OPV. D Schematic of the Zn/MnO2-graphite battery, with the components of the hydrogel electrolyte highlighted. E Photograph showing a 90 µm-thick FEHSS attached to the skin of a volunteer’s shoulder. Scale bar: 2 cm. F Photograph showing the same FEHSS attached to a textile. Scale bar: 1 cm. G Mechanical durability of the FEHSS for over 500 cycles under a 1 mm bending radius (left), and 100 cycles of compression-stretching under 10% strain (right). The y-axis shows the normalized efficiency of photo-conversion and storage. Inset schematics show the process of bending and compression test cycles. Source data are provided as a Source Data file.

High-performance ultraflexible OPVs

To construct highly efficient OPVs, we use poly[(2,6-(4,8-bis(5-(2-ethylhexyl-3-fluoro) thiophen-2-yl)-benzo[1,2-b:4,5-b’] dithiophene))-alt-(5,5-(1’,3’-di-2-thienyl-5’,7’-bis(2-ethylhexyl) benzo[1’,2’-c:4’,5’-c’] dithiophene-4,8-dione)] (PBDB-T-2F or PM6), a well-known derivative from the poly[(2,6-(4,8-bis(5-(2-ethylhexyl)thiophen-2-yl)-benzo[1,2-b:4,5-b’]dithiophene))-alt-(5,5-(1’,3’-di-2-thienyl-5’,7’-bis(2-ethylhexyl)benzo[1’,2’-c:4’,5’-c’]dithiophene-4,8-dione)] (PBDB-T) family, as the donor material33, and 2,2′-((2Z,2′Z)-((12,13-bis(2-ethylhexyl)-3,9-diundecyl-12,13-dihydro-[1,2,5]thiadiazolo[3,4-e]thieno[2,“3′’:4’,5′]thieno[2′,3′:4,5]pyrrolo[3,2-g]thieno[2′,3′:4,5]thieno[3,2-b]indole-2,10-diyl)bis(methanylylidene))bis(5,6-difluoro-3-oxo-2,3-dihydro-1H-indene-2,1-diylidene))dimalononitrile (BTP-4F or Y6) as the acceptor material (Fig. 1C)34. They provide an outstanding performance, most notably, a PCE surpassing 19% in a rigid OPV configuration35. While the system shows promise, it has a relatively limited stability36. With increasing time and under continuous illumination, there is a marked loss of PCE, which can be ascribed to aging-induced morphological changes and the formation of light-induced traps that increase the probability of recombination36.

To improve the inherent stability, we added (5Z)-3-ethyl-5-[[4-[15-[7-[(Z)-(3-ethyl-4-oxo-2-sulfanylidene-1,3-thiazolidin-5-ylidene)methyl]-2,1,3-benzothiadiazol-4-yl]-9,9,18,18-tetra(nonyl)-5,14-dithiapentacyclo[10.6.0.03,10.04,8.013,17]octadeca-1(12),2,4(8),6,10,13(17),15-heptaen-6-yl]-2,1,3-benzothiadiazol-7-yl]methylidene]-2-sulfanylidene-1,3-thiazolidin-4-one (O-IDTBR, Fig. 1C), into the above mix37. This effectively broadens the light absorption range (Figure S1A) and leads to an energy cascade path in the bulk heterojunction (Fig. 2A) that facilitates exciton dissociation38. Moreover, the inclusion of additional active components in a bulk heterojunction is known to aid in reducing the number of trap sites, thus decreasing the recombination probability and increasing charge collection efficiency39.

Fig. 2: Performance of the ultraflexible OPV single unit.
figure 2

A Energy level alignment of PM6, Y6, and the additive O-IDTBR in the active layer. B J-V characteristics of ultraflexible OPVs based on a PM6:Y6 binary blend (black) and a PM6:O-IDTBR:Y6 ternary blend (red). C EQE spectrum of ultraflexible OPVs based on a PM6:Y6 binary blend (black) and a PM6:O-IDTBR:Y6 ternary blend (red). D Light intensity dependence of VOC for OPVs based on a PM6:Y6 binary blend (black) and a PM6:O-IDTBR:Y6 ternary blend (red). E Light intensity dependence of JSC for OPVs based on a PM6:Y6 binary blend (black) and a PM6:O-IDTBR:Y6 ternary blend (red). F 2D GIWAXS data of PM6:Y6 binary blend. G 2D GIWAXS data of PM6:O-IDTBR:Y6 ternary blend. H Cake cuts from 2D GIWAXS data of the binary blend (black) and the ternary blend (red). Cake cuts in in-plane and out-of-plane directions are shown in dashed and solid lines, respectively. I Stability evaluation of PM6:O-IDTBR:Y6 ternary OPVs under different conditions, including in inert atmosphere in the dark (green), in ambient conditions in the dark (blue), and under a continuous illumination of 50 mW cm–2 in ambient conditions (red). J J-V characteristics of OPVs with pure ZnO (red) and EDT-ZnO (blue) as the electron transport layer. K Light intensity dependence of VOC for PM6:O-IDTBR:Y6 ternary OPVs with pure ZnO (red) and EDT-ZnO (blue) as the ETL. L Photo-stability evaluation of ultraflexible PM6:O-IDTBR:Y6 OPVs with ZnO-EDT as the ETL (blue), compared to the counterpart device using ZnO as the ETL (red). The devices were placed under continuous light illumination at 50 mW cm–2 for over 500 h. Source data are provided as a Source Data file.

Representative current density-voltage (J-V) characteristics for the ultraflexible OPVs based on PM6:Y6 binary and PM6:O-IDTBR:Y6 ternary blends are shown in Fig. 2B. The statistical analysis of their performance is summarized in Figure S1B and Table S1. Ternary-blend OPVs show notable performance improvement in current density (JSC) and fill factor. For the best-performing ultraflexible devices, JSC reaches 25.2 mA cm–2, and the fill factor reaches 73.8%, surpassing the values of 23.5 mA cm–2 and 73% observed from the binary-blend counterpart devices. The open-circuit voltage (VOC) has a moderate increase of 0.02 V, reaching 0.82 V in ternary-blend devices. As a result, the ternary-blend OPV has a superior PCE of 15.2%, surpassing the 13.7% of its binary-blend counterparts – a comparison made while both systems are in an ultraflexible configuration (Table S1). From the external quantum efficiency (EQE) spectra (Fig. 2C), the ternary-blend OPV exhibits a high EQE (75–83%) in the wavelength range of 550 nm to 850 nm, while the binary-blend counterpart shows a lower EQE (70–78%) in the wavelength range from 400 nm to 850 nm.

The performance of OPV devices is highly dependent on the exciton dissociation and charge recombination processes40. To probe the charge recombination in the OPVs, we evaluated the light intensity dependence of JSC and VOC. The ideality factor (n) of photovoltaics, which indicates the measure of how closely they follow the ideal diode equation41, can be derived from the dependence of VOC on incident light intensity following:

$${V}_{{{{\rm{OC}}}}}\,=\,n\frac{{k}_{{{{\rm{B}}}}}T}{q}{{{\mathrm{ln}}}}\left({I}\right)+{{{\rm{constant}}}}$$
(1)

where kB is for the Boltzmann’s constant, T is temperature, q is elementary charge, and I stands for light intensity. The ideality factor was estimated as 1.30 for PM6:Y6 binary-blend OPVs and 1.22 for PM6:O-IDTBR:Y6 ternary-blend OPVs, as shown in Fig. 2D. The ideality factor, observed to be closer to 2 in this binary case, suggests that the dominant carrier loss mechanism is trap-assisted recombination. JSC depends on the illumination intensity following the power law,

$${J}_{{{{\rm{SC}}}}}\propto {I}^{{{{\rm{\alpha }}}}}$$
(2)

where JSC is the short circuit current density and α indicates the extent of bimolecular recombination loss in the devices42. For the PM6:Y6 binary-blend and PM6:O-IDTBR:Y6 ternary-blend OPVs, we found respective α values of 0.95 and 0.99, as shown in Fig. 2E. The observed increase in α following the incorporation of O-IDTBR, confirms an increased charge collection efficiency and a reduced incidence of bimolecular recombination. The results highlight the presence of trap sites in both devices and that the binary OPVs have a larger loss of carriers through trap-assisted charge recombination43.

To better understand the microstructure of the binary and ternary blends, we perform grazing incidence wide-angle X-ray scattering (GIWAXS). Figure 2F, G show the 2D GIWAXS data of binary blend (Fig. 2F) and ternary blend (Fig. 2G) films, with their cake cut profiles presented in Fig. 2H. Note that the peak at 0.47 Å−1 is from the Kapton foil, which is used to seal the measurement chamber. Both samples show a (100) scattering peak at 0.30 Å−1, which is ascribed to the lamellar stacking of PM6. π-π stacking peaks are observed at 1.70 Å−1 in the binary blend and at 1.73 Å−1 in the ternary blend. A slight enhancement and peak shift in the π-π stacking peak is observed in PM6:O-IDTBR:Y6 ternary system from 1.70 Å−1 to 1.73 Å−1 in the out-of-plane (OOP) direction, corresponding to a shrinkage of π-π distance from 3.70 Å to 3.63 Å. Both the increased intensity and peak shift to a higher q value are beneficial for the OPV use as a closer π-π stacking distance and more crystallites can cause faster and more efficient charge carrier transfer and transport.

Figure 2I summarizes the stability performance of PM6:O-IDTBR:Y6 OPVs in an inert atmosphere in the dark (green), in ambient conditions in the dark (blue), and under continuous illumination in ambient conditions (red). When stored in the dark for 1500 h, the ultraflexible OPVs retain 98% of their initial PCE in an inert atmosphere. In ambient conditions, their PCEs maintain 87% of the initial values, with a shelf lifetime \({\tau }_{80}\) (the time for the PCE to decline to 80% of its original value)44, longer than two months. In contrast, the PCE of PM6:Y6 binary OPVs decreases to 82% of their original values when stored in ambient conditions in the dark for 1500 h (Figure S2A). Under continuous illumination at an intensity of 50 mW cm–2, both the ternary and binary OPVs show a noticeable degradation. In these circumstances, the \({\tau }_{80}\) values are approximately 380 h for the ternary OPVs, and 330 h for their binary counterparts.

To improve the photostability, we add 1,2-ethanedithiol (EDT)45 to passivate the electron transport layer (ETL), i.e., zinc oxide (ZnO), which commonly suffers from surface defects46. Fig. 2J shows the J-V characteristics of ultraflexible OPVs based on pristine ZnO and EDT-modified ZnO (denoted as EDT-ZnO below) as the ETL. The devices with EDT-ZnO has a JSC of 26.3 mA cm–2, a VOC of 0.815 V, and a fill factor of 75.5%, resulting in a PCE of 16.18% (Fig. 2J). The modified ETL has a noticeable effect on the dependence of both VOC and JSC on the light intensity. We observe that both the α and n values of the devices approach unity (Fig. 2K and Figure S2B). As depicted in Fig. 2K, n for the modified devices decreases to 1.08, suggesting that using EDT-ZnO as the ETL effectively reduces trap-assisted recombination losses. Additionally, the α value in devices using EDT-ZnO reach 1.005 (Figure S2B), a marked improvement from the initial value of 0.99 identified in devices with ZnO as the ETL. Figure S2C shows the integrated JSC from EQE spectrum of PM6:O-IDTBR:Y6 ternary blend devices, which agrees well with the J-V curve of the OPV devices. The stability of JSC as a function of time (Figure S2D) shows no degradation. The photocurrent is stable under maximum power point, with minor fluctuations of less than 0.003 mA cm−2. Figure S2E plots the photocurrent density with respect to the effective voltage (Jph-Veff) on a logarithmic scale. At low Veff values, the photocurrent density increases linearly in all devices and quickly saturates at Veff < 0.3 V, indicating an effective charge dissociation. The EDT modified ternary-blend devices has a significantly higher saturation current (Jsat = 29.2 mA cm–2) compared to the devices with pure ZnO ETL. The exciton dissociation probability Pdiss (Jph/Jsat) is calculated to be 93.6%, 94.2%, and 96.5% for ZnO/PM6:Y6, ZnO/PM6:O-IDTBR:Y6, and EDT-ZnO/PM6:O-IDTBR:Y6 devices. With Mott-Shockley analysis the trap densities (NA) of the devices are probed47. Figure S2F summarizes the Mott-Shockley plots of C–2-V (C denotes capacitance, V stands for voltage) for OPVs based on ZnO/PM6:Y6 (black), ZnO/PM6:O-IDTBR:Y6 (red), and EDT-ZnO/PM6:O-IDTBR:Y6 (blue). The dashed lines represent the linear fit of the curves. EDT-ZnO/PM6:O-IDTBR:Y6 devices exhibit the highest slope of –2.58 × 1018 as compared to –1.65 × 1018 and –0.95 × 1018 for ZnO/PM6:O-IDTBR:Y6 and ZnO/PM6:Y6 devices. Consequently, they yield the lowest trap density (1.14 × 1016 cm–3), while the devices based on ZnO/PM6:Y6 and ZnO/PM6:O-IDTBR:Y6 show higher trap densities of 3.1 × 1016 cm–3 and 1.78 × 1016 cm–3 (Table S1). These results further highlight the enhanced charge dissociation and collection efficiency of the EDT-ZnO/PM6:O-IDTBR:Y6 devices.

In a durability test, where the devices are subjected to continuous illumination at 50 mW cm–2 for 500 h, ultraflexible OPVs using EDT-ZnO as ETL retained over 92% of their initial PCE (Fig. 2L). This is a significant improvement compared to the τ80 value of 380 h of devices based on ZnO ETL. Detailed parameters including the VOC, JSC, and fill factor changes under continuous illumination are shown in Figure S3. After 500 h of continuous illumination, devices using EDT-ZnO ETL have a loss of 17% in the fill factor, whereas those based on ZnO ETL decreases to less than 70% of their initial values (Figure S3A). The VOC (Figure S3B) and JSC (Figure S3C) for devices with EDT-ZnO ETL maintained rather stable, showing negligible loss after 500 h of continuous illumination.

Designable ultraflexible OPV modules

We scale up the ultraflexible OPV single cells to configurable modules, using the PM6:O-IDTBR:Y6 ternary blend as the active material, and EDT-ZnO as the ETL. The overall voltage and current output of these devices are determined by adjusting the series and parallel connections within arrays and changing the area of the individual active cells. To meet the specific power needs of different applications, from low-powered devices such as miniaturized biosensors requiring microwatt (µW) range power, to high-powered intricate electronics like smartwatches and electronic displays needing milliwatt (mW) power, we engineer multiple arrays with different output characteristics. Figure 3A shows our ultraflexible OPV module consisting of 12-cells-in-series and 14 of these groups in a parallel connection, occupying an effective working area of 6.72 cm2. The delamination of this module from a glass support is depicted in Figure S4A. From this 6.72 cm2 freestanding module, we achieve an overall power conversion efficiency of 10.5%, a peak power (Pmax) exceeding 68.9 mW (Fig. 3B, green traces), resulting in an areal power output (Pareal) over 10.2 mW cm–2. Other modules of different sizes, i.e., 1.28 cm2, and 16 cm2, have their representative J-V curves shown in Fig. 3B, and individual performance indices summarized in Table 1. Their power output can meet the energy demand for wearable applications, particularly small sensors and gadgets, which typically require less than 100 mW of power13. Compared to documented flexible OPV modules22,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71, our modules show outstanding areal power output while being extremely thin and flexible (Fig. 3C and Table S2). Though the PCE of ultraflexible OPV single unit has surpassed 10% since 201822 and continued to rise in recent years22,48,49,50,51,52, their scale were limited and there were few cases showing Pareal exceeding 10 mW cm–2.

Fig. 3: Structure and performance of ultraflexible OPV modules.
figure 3

A Photographs showing an ultraflexible OPV module that has 14 in-parallel groups of 12 cells in-series. Each OPV unit has an active area of 0.04 cm2. The total working area is 6.72 cm2, on a 25 cm2 substrate. Scale bar, 1 cm. B J-V characteristics (solid lines) and power output (dashed lines) of ultraflexible OPV modules in the freestanding state. The modules have working areas of 1.28 cm2 (pink), 6.72 cm2 (green), and 16 cm2 (blue). C The areal power output by our ultraflexible OPV modules (active areas: 1.28 cm2, 6.72 cm2, and 16 cm2) and from documented results22,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71. Green dots denote common flexible devices (thickness range: 100–200 µm)53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71 Hollow orange spheres denote ultraflexible devices with a thickness below 10 µm22,48,49,50,51,52. Red stars represent ultraflexible devices developed in this study. D J-V characteristics (solid lines) and power output (dashed lines) of the 6.72 cm2 module under different light intensities, i.e., 1-sun (black), 0.5 sun (blue), 0.3 sun (green), and 0.1 sun (pink) illumination. E Power output of the same 6.72 cm2 module under varying light intensities of 7000 lux (maroon), 3500 lux (red), 1500 lux (peach), and 700 lux (pink). F Mechanical durability of the 6.72 cm2 module tested under a bending radius of 1 mm (green) and under 50% compression strain (blue) for 1000 stretching cycles. Source data are provided as a Source Data file.

Table 1 Performance parameters of ultraflexible OPV cells and modules with different active areas

The performance of ultraflexible OPV modules under different light intensities is examined (Fig. 3D). Under simulated 0.5-sun (50 mW cm–2) and 0.3-sun (30 mW cm–2) illumination, our 6.72 cm2 module has a power output of 35 mW and 20 mW, respectively. When subjected to 0.1-sun (10 mW cm–2) illumination, its JSC drops to ~0.2 mA cm–2 and VOC is ~ 7.6 V. This produces a power output of 6 mW, which is sufficient to charge four ZIB batteries connected in series, each of which has a charged voltage of approximately 1.8 V, thereby enabling the operation of small wearable sensors72.

Our ultraflexible OPV module can efficiently produce power in various lighting conditions, even with dim or indoor illumination. For instance, under an overcast sky that yields an average light intensity of approximately 7000 lux, the 6.72 cm2 module generates a power output of 3.5 mW (Fig. 3E). As the light intensity decreases to 700 lux in a lit room, the ultraflexible module continues to function effectively, producing an output of around 0.2 mW, underscoring its performance robustness.

The ultraflexible OPV modules display exceptional mechanical durability. As shown in Fig. 3F, after being bent to a 1 mm radius for 1000 cycles, the 6.72 cm2 module retains 90% of its original power output. In additional test, the module is attached to a pre-stretched rubber membrane and then allowed to release, which introduced a compressive strain of 50%. After 1000 cycles of compression-stretching, the module continues to generate 50 mW of power, maintaining its current and voltage output at 9.8 mA and 6.4 V, respectively. We strategically placed relatively brittle materials, specifically the indium tin oxide (ITO) electrode, between flexible parylene films and positioned them close to the neutral mechanical plane. This configuration effectively distributes strain in the direction perpendicular to the plane, and minimizes the strain experienced by the brittle components, consequently enhancing the overall mechanical durability of the module73. Scanning electron microscopic (SEM) images of the OPV module before and after subjected to the mechanical durability test are depicted in Figure S4B, C. Following the continuous bending test for 1000 cycles, there is a noticeable increment in the formation of wrinkles across the surface of the ultraflexible OPV module. On the other hand, there are no visible manifestations of cracking or interfacial delamination within the examined regions. These observations together with the performance stability are indicative of the OPV module’s resilience to mechanical strains.

Performance of an ultrathin Zn-ion battery

The layer composition of the thin zinc-ion battery is illustrated in Fig. 4A, which mainly constitutes a thin hydrogel electrolyte sandwiched between the Zn anode and MnO2-graphite cathode. Note that we have substantially reduced the overall thickness of the battery using two strategies. First, we introduce an ultrathin hydrogel electrolyte, i.e., 10 µm-thick poly(vinyl alcohol)/graphene oxide (PVA-GO) hydrogel, formed by a cold-lamination method we have previously reported74. Fig. 4B shows the fabrication of ultrathin, uniform PVA-GO hydrogel films tailored to specific sizes. SEM images of the hydrogel film display a porous structure (Fig. 4C) and a bending radius down to 60 µm (Fig. 4D), showing extreme mechanical compliance. Second, we substitute the commonly used carbon cloth, typically with a thickness of 0.5 mm, with thin graphite paper only 25 µm thick. The perimeter is sealed with 10 µm-thick polyethylene terephthalate (PET) double-sided tape, and copper foil is used as the contact tabs for both the Zn anode and MnO2-graphite cathode. Finally, 1.5 µm thick parylene films are used as the encapsulating medium at both the front and back sides to provide a flexible yet effective containment for the battery. The total thickness of the battery is minimized to 85 µm. This marked thickness reduction is a crucial advance in the ongoing development of thin, flexible, and safe energy storage systems.

Fig. 4: Fabrication and performance of the ultrathin Zn-MnO2 battery.
figure 4

A Illustration detailing the structure and multi-layered configuration of the battery. B Depiction of the synthesis of the thin hydrogel electrolyte film for the battery, by a cold-lamination procedure. C, D SEM images showing the topography of the ultrathin hydrogel (C) unstrained and (D) strained to a bend radius of 60 µm. E Voltage-capacity profile of flexible Zn-MnO2 batteries, featuring hydrogel electrolyte thicknesses of 10 µm (red), 100 µm (blue), and 1 mm (cyan), charged by an external power supply. F Rate performance of the battery at 0.2 C, 1 C, and 5 C charging rates. G Stability of the battery over 200 consecutive charge-discharge cycles. Source data are provided as a Source Data file.

To quantitatively investigate the impact of hydrogel thickness in the performance of ZIBs, we conduct electrochemical evaluations of hydrogels with different thicknesses, i.e., 10 µm, 100 µm, and 1 mm. Nyquist plots (Figure S5A) depict the corresponding battery performance under open circuit voltage conditions. The intercepts of these curves reveal an increase in ohmic resistance from 2.7 Ω to 81 Ω as the hydrogel thickness increases from 10 µm to 1 mm. The ionic conductivity (σ) of hydrogel electrolytes is then calculated following:

$$\sigma=\frac{L}{R\times S}$$
(3)

where L stands for the hydrogel thickness (i.e., the distance between the two electrodes), R stands for the bulk resistance of the hydrogel, and S stands for the area of electrode/electrolyte interface. The ionic conductivities are determined to be 7.1, 6.7, and 6.2 mS cm–2 for the hydrogel with thicknesses of 10 µm, 100 µm, and 1 mm, respectively. Reducing the thickness of hydrogels generally enables faster ion diffusion between the anode and cathode by shortening the distance and reduces the impedance resistance.

While the reduction in electrolyte thickness enhances ionic conductivity by reducing the impedance, it may concurrently have a negative impact on battery capacity (Q). The voltage-capacity curves illustrated in Fig. 4E reveal a 4.5% decrease in discharge capacity when the hydrogel thickness decreases from 1 mm to 10 µm. This reduction in capacity may be attributed to the limited electrolyte volume and ion availability in thinner hydrogels compared to their thicker counterparts75. Moreover, ultrathin hydrogel electrolytes are likely to more rapidly saturate with active ions at the electrode-electrolyte interface, constraining the ions available for electrochemical reactions, which could, in turn, decrease specific battery capacity and discharge duration76. Despite the relatively lower capacity of the ultrathin battery, it remains adequate for most wearable applications. The significant benefits in flexibility offered by ultrathin electrolytes compensates for the minor capacity losses. Consequently, the 10 µm hydrogel electrolyte is chosen for our FEHSS, prioritizing its exceptional flexibility and ionic conductivity.

Cyclic voltammograms (CVs) of the thin Zn/MnO2-graphite battery at scan rates of 0.1 mV s–1 (grey), 0.2 mV s–1 (blue), 0.3 mV s–1 (green), 0.5 mV s–1 (orange) and 1.0 mV s–1 (pink) in the voltage range of 1.0–1.8 V are recorded (Figure S5B). Two pairs of peaks, i.e., oxidation peaks at 1.50 and 1.64 V, reduction peaks at 1.24 and 1.38 V, can be clearly observed, highlighting the two-step redox reaction associated with Zn2+ ion insertion and extraction from the MnO2/graphite cathode. Over multiple scan cycles, a well maintenance of the as-mentioned redox peaks of H+ and Zn2+ co-intercalation are observed, indicating the good ionic conductivity of the hydrogel. The charge-discharge profiles of the flexible battery (Fig. 4E) show voltage plateau at around 1.45–1.50 V, associated with the interaction of Zn2+. The discharge capacity increased under stable voltage until all active sites in MnO2 lattice are effectively embedded, agreeing well with the CV characteristics. With an electrode area of 1.75 cm2, the battery reaches an overall capacity of 6.8 mA h, a specific capacity of 3.88 mA h cm–2, and a coulombic efficiency up to 98%. Figure S5C shows the storage stability of the 85 µm-thick battery. After being stored in ambient conditions for 10 days, it preserves 94% of its original voltage, maintaining a value exceeding 1.58 V. Energy density (Ed) of the battery is estimated following:

$${E}_{d}=\frac{V\times Q}{A}$$
(4)

where V is the nominal voltage, Q is the capacity, and A is the active area of the battery. Our exceptionally thin and flexible zinc-ion battery, featuring an active electrode area of 1.75 cm2, boasts an energy density exceeding 5.82 mW h cm–2. This marks one of the highest energy densities observed in flexible batteries, which hardly achieve energy densities beyond 5 mW h cm–2 13.

The rate performance of the 85 µm-thick battery is analyzed at three charging rates, 0.2 C, 1 C, and 5 C. Figure 4F illustrates a slight decrement in specific capacities, namely the capacity per gram, at both low and high cycling rates. This phenomenon can be ascribed to the natural distortion of the MnO2 cathode, triggered by the MnO2 lattice deformation during Zn2+ intercalation/de-intercalation cycles.

The freestanding ZIB’s cycling stability is assessed across 200 charge-discharge cycles at 5 C charging rate. The battery exhibits a steady coulombic efficiency ranging between 94%–98% and retains more than 86% of its initial specific capacity after the cycling process (Fig. 4G). The voltage-specific capacity profiles of the flexible battery at the 1st and 200th cycles are shown in Figure S5D, highlighting excellent cycling stability of the ZIB, despite its ultrathin nature.

The mechanical robustness of ultrathin ZIBs is pivotal for their practical application across various wearable scenarios. The 85 µm-thick ZIB exhibits excellent mechanical stability, retaining 88% of its original coulombic efficiency after being subjected to 500 continuous bending cycles at a bending radius of 1 mm (Figure S5E). To further assess the effects of bending strain on the battery’s performance, we analyze Nyquist plots obtained before and after the bending experiments. As shown in Figure S5F, a modest increase in the ohmic resistance (approximately 5%) and the charge transfer resistance (around 8.5%) is observed, suggesting subtle performance alterations. SEM images presented in Figure S6 provide visual elucidation of the battery’s structural integrity before and after the bending test. These findings collectively suggest the robust mechanical resilience of the ultrathin ZIB, capable of enduring multiple bending cycles without significant performance degradation.

The integrated FEHSS

The integrated FEHSS is assembled by connecting an OPV array with flexible batteries in a back-to-back stacked structure using ACF tapes. The anode of the battery is connected to the anode of the OPV module, and the cathodes are connected in parallel (Figure S7A). Notably, current reflux from the battery to the OPV is a major concern for hybrid solar-battery systems when there is no power management circuitry in between. To prevent the current backflow under dark conditions, we design a simple blocking diode between the battery and the OPV module. As depicted in Fig. 5A, B, a diode with an active area of 0.02 cm2 is connected to the positive terminal of each in-series group of cells to block the current reflux. The diodes are serially connected with the OPV module (Fig. 5B), then the OPV-diode unit is connected in parallel to the flexible ZIB to form an integrated FEHSS. The leads from the FEHSS are further connected to the load (circuit diagram in Figure S7B). The J-V characteristic of the diode shows a rectifying ratio in the order of 105 for ± 2 V (Figure S7C), which is sufficient for ZIBs. We test the rectifying performance of the diode by comparing voltage levels of the battery connected to the OPV in both dark and illuminated conditions (Fig. 5C). When the light is turned off, the diode effectively prevents current reflux, and the battery voltage remains above 1.4 V (solid curve). Without the diode, on the other hand, the battery voltage is observed to continuously fall below 1.3 V as the light intensity decreased (dashed curve).

Fig. 5: Performance of the FEHSS.
figure 5

A Schematic highlighting the layout of the current-management diode in the integrated system. B Schematic showing the side view of the diode structure implemented within the OPV module. C V-t curve of the system with (solid line) and without (dashed line) the diode. The orange area represents the FEHSS in the dark. D Photo-charging (solid lines) and discharging (dashed lines) characteristics of an integrated FEHSS having one ZIB (blue), and another integrated FEHSS having three ZIBs connected in series (green). E Schematic showing the layout of the FEHSS having a top OPV module (12 cells-in-series and 14 of them in-parallel) stacked on three ZIBs that are connected in series, with a PET adhesive frame and ACF tape at the positive and negative terminals to allow the charge flow. F Comparison of the charge-discharge curves of the integrated system being charged by (black) an external power supply, and (red) by photo-charging. G, H Stability characterization of the ultrathin FEHSS: (G) charge-discharge stability, and (H) storage stability in ambient conditions. I Radar plot comparing the performance metrics of reported FEHSSs based on solar energy harvesting and battery storage. PCS-ZIB stands for a perovskite solar cell integrated with a zinc-ion battery, SiPV-LIB stands for a silicon photovoltaic integrated with a lithium-ion battery, FIED stands for a fiber-shaped integrated energy device, and OPV-LIB represents an organic photovoltaic integrated with a lithium-ion battery. Source data are provided as a Source Data file.

After integration, the OPV array maintains its high-power output, and the flexible battery sustains its normal charge-discharge characteristics. As shown in the voltage-capacity profiles, a solitary ZIB can be fully charged to approximately 6.5 mA h, with an output voltage of 1.8 V, through spontaneous photo-charging by a 1.28 cm2 OPV module under low illumination conditions of 1 mW cm–2 (Fig. 5D, blue plots). One can conveniently assemble additional ZIBs in series (Fig. 5E) to intensify the output voltage. For example, three ZIB cells aligned in series can be charged by the same OPV module under higher illumination intensity of 50 mW cm–2 and achieve an output voltage exceeding 5.4 V (Fig. 5D, green plots). The charge-discharge profiles of an FEHSS by a conventional external power source (black traces, the initial five cycles) and by photo illumination (red traces, the final five cycles) are compared in Fig. 5F. The comparison illustrates a near-equal charging efficiency and displays steady charge-discharge characteristics across successive cycles.

We further evaluate the charge-discharge profiles of the FEHSS under varying light intensities to simulate outdoor and indoor lighting conditions and assess performance impacts. The charging rate generally declines with a decrease in the output current of the OPV module, which correlates directly with reduced light intensities (Figure S8). As depicted in Figure S8A, under 1 sun illumination, i.e., light intensity of 100 mW cm–2, a ZIB integrated with a 6.72 cm2 OPV module can be fully charged within 37 min (black), and within 72 min at a reduced intensity of 50 mW cm–2 (blue). Indoor light levels typically range from 500 to 700 lux, translating to approximately 75–100 μW cm–2 77. A notable increase in charging time is observed as the light intensity is decreased to indoor light conditions (Figure S8B). As light intensity diminished from 1000 µW cm–2 to 75 µW cm–2, the charging duration of the same ZIB escalated from approximately 400 min to 580 min. Although full charging to 1.8 V indoors exceeds 6.5 h, the system exhibits a noteworthy interim charging capability. Within 2 h, the ZIB battery can be partially charged to 1.5 V, becoming capable of powering low-energy devices. The integration with an OPV module facilitates continuous charging, even under low lighting conditions, enhancing the system’s versatility and utility for both outdoor and indoor applications.

To gain insights into the overall performance of the integrated FEHSS, we calculate the overall photo-electric conversion and storage efficiency (\({E}_{{{{\rm{overall}}}}}\)) following:

$${E}_{{{{\rm{overall}}}}}=\frac{Q\times V}{I\times {A}_{{{{\rm{OPV}}}}}\times {t}_{{{{\rm{charge}}}}}\,}\times 100$$
(5)

where I is the light intensity, AOPV is the effective OPV area, and tcharge is the photo-charging time (in hours). In an outdoor environment with a light intensity of approximately 65 mW cm–2, an FEHSS comprised of a 1.28 cm2 OPV module and a ZIB can be spontaneously charged to a capacity of 5.5 mA h within 1.5 h. This process results in an overall efficiency of 6.61%. With a reduced light intensity of 10 mW cm–2, the same FEHSS can be charged to a capacity of 5.9 mA h within 10 h, yielding an overall efficiency of 6.91%, which stands among the highest efficiencies reported for variable kinds of FEHSS17,18,78,79,80,81. Supplementary Table S3 provides a comprehensive evaluation of the photo-charging capabilities of FEHSS under various light intensities, specifically detailing performance metrics with small (1.28 cm2) and relatively large (6.72 cm2) OPV modules. Increased incident light intensities generally correlate with faster charging rates for both sizes of OPV modules. Nonetheless, this acceleration in charging rates inversely affects the battery’s capacity82. As the light intensity diminishes from 100 mW cm–2 to 10 mW cm–2, the overall efficiency with a 1.28 cm2 OPV module rises from 5.27% to 6.91%. Above results suggest that both the charging rate and the overall efficiency of FEHSS are intricately linked to the incident light intensity I, which directly influences the current output of the OPV module and subsequently impacts the charging behavior of the battery. Therefore, a balance must be achieved between charging rate and overall efficiency, as excessively low charging rate may limit the practical utility of the integrated energy system.

Assessments of stability, encompassing conditions like continuous operation, in storage, and subjected to mechanical strain, highlight the enduring performance of our FEHSS. In particular, the FEHSS retains more than 80% of its photoconversion and storage efficiency across over 60 uninterrupted charge-discharge cycles (Fig. 5G). When exposed to ambient conditions, our FEHSS demonstrates a shelf lifetime (\({\tau }_{80}\)) that surpasses two weeks (Fig. 5H). Under mechanical strain, such as bending to a 1 mm radius over 500 cycles or enduring repeated 10% compression strain for 100 cycles, the system preserves over 80% of its original light harvesting-storage efficiency (Fig. 1G). It sustains a voltage output of 1.48 V and a capacity exceeding 5.3 mA h. Figure S9 depicts cross-sectional SEM images of the as-prepared FEHSS (Figure S9A), FEHSS under bending strain (Figure S9B), and after cyclic bending test (Figure S9C). The ultrathin FEHSS can achieve a bending radius as low as 120 µm (Figure S9B). Following 500 cycles of repeated bending test, slight deformation and kinks were observed at the battery side within the FEHSS, which might be the major cause of the performance loss. Enhancement of the interfacial strength among layers in the battery could mitigate the observed deformation, representing a potential avenue for future optimization.

The radar chart depicted in Fig. 5I, along with the summarized performance metrics in Table S4, provides a comprehensive overview of the documented FEHSS based on photovoltaic energy harvesters and battery storage systems. Noteworthy among these systems is our FEHSS, distinguished by its unparalleled flexibility, exceptional electrical performance, and inherent safety features. Demonstrating mechanical compliance with a bending radius down to 120 µm, our system maintains robust functionality across both individual battery and OPV modules, as evidenced by excellent PCE, energy density, specific capacity, and overall energy harvesting-storage efficiency. The common trade-offs between electrical and mechanical performance within these systems have been effectively addressed. Moreover, our FEHSS stands out from other documented systems with the incorporation of a current management diode, a feature enhancing its practical applicability. These findings highlight the efficiency and versatility of our FEHSS, positioning it as a promising solution for wearables applications.

We subsequently evaluate its feasibility of self-charging and concurrently powering compact electronic devices such as LEDs. This assessment is conducted using a smaller-scale FEHSS, comprised of a 6.72 cm2 OPV module and two ZIBs in series connection that generate 3.6 V voltage. As shown in Figure S10A, in the absence of light illumination (top panel), the output voltage of the FEHSS declines to 2.4 V after continuously powering the LED for 30 min. Even in dim conditions, featuring a light intensity of 10 mW cm–2, the FEHSS manages to charge autonomously while powering the LED (bottom panel). This process effectively decelerates the battery’s discharge process, as further depicted in Figure S10B.

Lastly, we demonstrate the use of FEHSS to power a variety of electronic devices. Wearable electronics generally require a power input in the range of 1–100 mW to operate14. A small FEHSS based on a 5\(\times\)5 cm2 substrate can generate power up to 68.9 mW. With three ZIB cells connected in series (Fig. 5D), the energy density of over 5.82 mWh cm–2 and an output voltage exceeding 5.4 V can be realized, offering great feasibility to power wearable sensors and even daily gadgets. Figure 6A, B show our on-shirt FEHSS powering a biosensor. For effective sunlight harnessing, the FEHSS is positioned on the fabric at the shoulder of the wearer (Fig. 6A). The sensor is an on-skin single-lead electrocardiogram (ECG) electrode patch, which is connected to a flexible printed circuit board (PCB) for signal processing and transmission (Fig. 6B). Using a Bluetooth module, the signal waveform and associated data could be displayed on a smartphone in a real-time manner (Supplementary Movie 1). Figure 6C shows a cutout of an ECG signal waveforms within 6 s, clearly displaying characteristic P, QRS, and T waveforms that reflect atrial and ventricular activities – a testament to the successful operation and use of the FEHSS as a power source for wearables.

Fig. 6: FEHSS as the power source for wearables and gadgets.
figure 6

A, B Photographs of an all-flexible ECG sensing system powered by our FEHSS. Scale bar: 5 cm. C ECG signal obtained from the flexible ECG electrodes powered by our FEHSS, showing clear, continuous P, QRS, and T waveforms. D Photograph demonstrating an on-skin FEHSS charging a smartphone outdoors under a light intensity of 65 mW cm–2. Scale bar: 3 cm. E Smartwatch (left) and smartphone (right) charging rate using commercial USB (pink) and FEHSS (purple) charged under the light intensity of 100 mW cm–2. Source data are provided as a Source Data file.

Furthermore, our FEHSS demonstrates the capability to charge a smartphone and a smartwatch (Fig. 6D, E) – a feature that has particular significance in activities such as wilderness adventures or long-distance trips where reliable power sources may not be available. Figure 6D demonstrates a user-friendly interaction where the FEHSS was adhered to the hand dorsum of a user, enabling smartphone charging without interfering with the user’s ongoing activities (complete setup layout shown in Figure S10C). Here, this FEHSS consists of a 6.72 cm2 OPV module connected to four ZIBs underneath, all based on a 5\(\times\)5 cm2 ultrathin substrate. Figure 6E shows the charging efficiency for a smartwatch (left) and a smartphone (right), using the commercial USB (pink) as compared to our FEHSS (purple). When coupled with a compact power regulator, our FEHSS consistently produces a power output of 0.5 W – approaching the performance of a conventional USB 3.0 charging unit that typically outputs several watts. When applied to a smartwatch (setup layout shown in Figure S10D) with a total capacity of 455 mA h, our system managed to charge it to 28% within an hour. Using a conventional USB, the smartwatch is charged to 45% within the same timeframe. Our FEHSS successfully charges a smartphone from 0% of battery to 6% within an hour, whereas a conventional USB charger brought the same battery capacity to 21% within the same timeframe. Despite a comparatively reduced charging speed, FEHSS facilitates adequate power for smartphone/smartwatch charging, in an entirely eco-friendly manner. Expanding the FEHSS lateral scale, fine-tuning the power regulator, and optimizing the wiring and interfacial connections will undoubtedly boost the charging speed for larger electronic devices. The prototype demonstrations presented herein already underscore the potential applications of ultrathin FEHSS for daily power requirements, with convenience and an eco-friendly energy basis.

While individual energy harvesting and storage components can function independently, their integration addresses inherent limitations. In FEHSS, the OPV module continuously charges the battery, thereby prolonging its lifespan, while the battery ensures power provision under low-light conditions when the OPV’s output is insufficient. Visual evidence of this synergy is provided in Supplementary Movie 2 and Movie 3. When the LED is powered solely by the OPV, its operation relies on high light intensity and ceases once the incident light diminishes (Supplementary Movie 2). Utilizing FEHSS effectively circumvents this limitation, as the ZIB within the FEHSS sustains the LED’s brightness even in darkness (Supplementary Movie 3). These scenarios highlight the critical importance of integrating both the OPV and ZIB within a unified system.

Discussion

We have developed a highly flexible energy harvesting and storage system as a reliable energy source for wearable applications. Consisting of an organic photovoltaic module as the energy harvesting component and zinc-ion batteries as the energy storage component, the self-powered FEHSS can be integrated with textiles and even be worn directly on the skin, to effectively power wearable devices in a sustainable fashion.

For the energy harvesting component, we have boosted the PCE of ultraflexible OPVs up to 16.18%. The freestanding OPVs demonstrate exceptional long-term storage stability that extends beyond two months, and operational stability for over 500 h under continuous illumination. We also scaled up the devices into solar modules. Demonstrating our prototype, we develop an ultraflexible module with an effective area of 6.72 cm2, which delivers an areal power output reaching 10.2 mW cm–2, generating power over 68.9 mW, which is sufficient to operate small electronics.

For the energy storage component, we have effectively reduced the thickness of the Zn-ion battery to 85 µm without sacrificing its electrochemical performance. This is enabled by introducing an ultrathin PVA-GO hydrogel as the electrolyte and replacing the carbon cloth with thin (25 µm) graphite paper. This flexible battery had a high capacity of 6.8 mA h and good electrochemical stability, i.e., over 86% of its initial specific capacity after cycling over 200 charge-discharge cycles.

Finally, the OPV module and batteries are integrated to produce an all-in-one energy harvesting and storage system for wearable applications. It’s noteworthy that we innovatively introduced ultraflexible organic diodes, which have a rectification ratio in the order of 105 in the ± 2 V range, in between the energy harvesting and storage modules, to effectively prevent undesired current reflux from the battery to the OPVs in dark conditions. The integrated FEHSS shows an overall energy conversion and storage efficiency up to 6.91%, a \({\tau }_{80}\) surpassing two weeks in ambient conditions, excellent working stability and mechanical durability. With an overall thickness of mere 90 µm, the system easily conforms to curved and soft surfaces including the human skin and textiles. It can power various electronic components, including electrophysiological sensors for healthcare monitoring, gadgets like the smartwatch and smartphone, etc.

In the realm of energy solutions for electronics, a multitude of investigations have underscored the potential of OPVs and flexible batteries. Their fusion into a self-sustaining, lightweight, and adaptable power system heralds a substantial breakthrough for wearable applications. The integration of customizable, ultraflexible OPV modules and 85 µm-thick Zn-ion batteries, each with an excellent performance, allows this FEHSS to be used for long-term operation and should provide the basis for many medical monitoring advances.

Methods

Fabrication of flexible OPV arrays

Glass substrates were cleaned successively with detergent, acetone, and isopropanol using ultrasonication, then blow-dried with a nitrogen gun, followed by oven drying at 100 °C for 1 h. A fluorinated polymer layer (Delo D3, Shenzhen Sino-fluorine:Novec 7100, 3 M Company) was spin-coated on the glass substrates as the sacrificial layer, followed by the deposition of 1.5 µm-thick parylene substrates. The parylene film was coated using a parylene deposition system (SCS PDS2010E Labcoter) and annealed at 180 °C for 1 h in a glove box. A 100 nm-thick ITO layer was then sputtered as the transparent conductive cathode using a SKY Technology-RH40 Magnetron sputtering device. The substrates were treated with oxygen plasma (Diener) for 5 min, followed by the deposition of the electron transport layer by spin-coating of a ZnO precursor solution (5000 rpm for 30 s) with subsequent annealing at 180 °C for 30 min. A ZnO solution was prepared by dissolving zinc acetate dehydrate (0.25 g) and ethanolamine (0.08 mL) in 2.5 mL 2-methoxy ethanol. To modify the ETL, the devices were treated with an EDT solution (1,2-ethanedithiol (0.5% v/v) in acetonitrile) by spin coating on the ZnO layer and annealing at 110 °C for 5 min. The substrates were then transferred to a glovebox for active layer deposition. Active solutions were prepared by dissolving PM6:Y6 (1:1.2, w/w) and PM6:O-IDTBR:Y6 (1:0.18:1.02, w/w) in chlorobenzene and 1-chloronaphthalene (99.5:0.5 v/v) at a concentration of 18 mg mL–1, and were dynamically spin-coated at 1500 rpm for 60 s, followed by annealing at 110 °C for 10 min. A thin (7 nm) layer of MoOx was thermally deposited on top of the active layer in a high vacuum evaporator (< 10–4 Pa), immediately followed by the deposition of an Ag anode (100 nm) in the vacuum chamber. The active area of each device was defined by shadow masks. The complete devices were then encapsulated with a 1.5 µm-thick parylene film. External gold contacts were conformally added to the devices using anisotropic conductive tape (ECATT 9703, 3 M Company) before encapsulation to ensure good contact during device characterization. The ultraflexible OPV devices, only 4 µm thick, were then peeled off from the supporting glass into a freestanding state.

Characterization of the OPV devices

The OPV devices were tested under AM 1.5 G simulated solar illumination (100 mW cm–2) produced by a SS-X50 Enlitech solar simulator based on a 300 W Xe lamp. The light intensity was calibrated with an SRC-2020, EnliTech, a standard silicon PV. Current density-voltage (J-V) characteristics of the OPV devices were obtained in forward scan direction, with 0.02 V scanning step and 10 ms delay time using a Keithley 2400 source meter under ambient conditions. Shadow masks were used to define the active area. EQE spectra of the OPV devices were measured with a PD-QE equipment (Enlitech), which uses filters to produce monochromatic light for the EQE measurement. The measurement setup was calibrated with a commercial silicon photodiode before conducting the EQE test. The Pareal was calculated over the total effective area of each module. Grazing-incidence wide-angle X-ray scattering (GIWAXS) was performed at the P03 beamline, PETRA III, DESY in Hamburg (Germany), using an X-ray beam energy of 11.83 keV (1.023 Å). The sample-to- detector distance (SDD) was set to 227 mm and the incident angle to 0.1°. The 2D GIWAXS transformation to q-space and the cake cuts for the GIWAXS data were processed by the Matlab-based package GIXSGUI.

The stability of the OPV devices was evaluated under different conditions, i.e., in the dark in inert atmosphere, in the dark in ambient air, and under direct illumination in ambient air. For the inert condition, the devices were stored in dark containers inside a glove box. The air stability was tested with devices stored in a dark desiccator with a controlled temperature of 25 °C and humidity of 55%. The photostability was tested by exposing the OPV devices to direct and continuous illumination under the solar simulator.

Fabrication of flexible ZIBs

We prepared each part of the flexible Zn-MnO2 battery with intrinsically soft materials: 20 μm thin Zn metal foil (99.99% metals basis; Innochem), 50 μm conductive graphite paper, MnO2 powder (99.5%; Innochem) and synthesized PVA-GO gel polymer. PVA-GO hydrogels with excellent ionic conductivity were synthesized following a three-step procedure: (i) dissolve 8.5 g PVA powder (weight-average molecular weight of 146,000 to 186,000 and degree of hydrolysis of 98.0 to 98.8%; Acros) in 100 mL of deionized water under 85 °C, and adding a 1500 μL GO dispersion (5 mg/mL; one to five layers; XFNano) drop by drop into the stirring solution. (ii) A cold lamination technique was used to prepare ultrathin films of PVA-GO hydrogel. For this procedure, a small amount of PVA-GO precursor (~2 mL) was carefully dropped onto a hydrophobic PET film using a 5 mL dropper. This was covered with another hydrophilic PET film for protection and passed through the cold laminator, which uniformly spread the PVA-GO precursor between the two PET films to form a thin sheet of the desired thickness. For this study, we prepared 50 µm thick hydrogel films. After passing the hydrogel through the cold laminator, any bubbles that appeared in the film were removed. The PVA-GO hydrogel was cured by freeze-thawing to induce crosslinking and hydrogen bond formation between GO and PVA. For this step, the PET laminated hydrogel films were refrigerated at −6 °C for 18 h and melted at room temperature for at least 1 h. Repeating this operation increases the degree of crosslinking and increases mechanical strength. Once the hydrogel was fully cured, the supporting hydrophobic PET film was peeled off, and the hydrogel thin film was dried at 120 °C for 30 min, which reduced its thickness to 10 µm. After completely drying, the hydrogel was soaked in the salt solution. (iii) Soak the as-prepared gel polymer in the 1 M ZnSO4 (99.5%; Innochem)/0.3 M MnSO4 manganese (II) sulfate monohydrate (99.99%; Innochem), potassium thiosulfate (95%; Sigma-Aldrich) solution for 24 h. We prepared a 7:2:1 slurry with MnO2 powder, conductive carbon and PVDF for coating on conductive graphite paper to make a flexible cathode film. Using Zn-metal foil as the anode, MnO2-graphite paper as the cathode, and the prepared PVA-GO film as electrolyte, we assembled the flexible Zn-MnO2 battery in a stacked structure and encapsulated it with ultrathin parylene films, sealed by a 10 µm-thick double sided PET tape. Copper foil was used as the contact tabs for both the cathode and anode.

Characterization of flexible ZIBs

Zn-MnO2 batteries were fabricated in a sandwiched structure for the electrochemical tests. We conducted cyclic voltammetry testing at scan rates of 0.1 mV s–1, 0.2 mV s–1, 0.5 mV s–1, and 1.0 mV s–1. The rate performance was characterized at set rates of 0.2 C, 1.0 C, and 5.0 C with a battery test system (NEWARE CT-4000). Long-life operation was also tested in the NEWARE CT-4000 with 5 C rate cycling at a current density of 0.5 A g–1.

Integration of OPV arrays and ZIBs

The all-in-one FEHSS was assembled using ACF tape at the electrode terminals. The negative terminals of the OPV module and ZIB were directly connected with ACF tape. Small rectifying diodes using the same structure and materials as the OPVs were placed at the positive terminals of the OPV and connected to the positive terminal of the battery. The diodes were added to restrict current reflux to the OPV unit under dark conditions. Copper contact tabs were added for connecting the energy system with the electronics of the application.

The use of wearable integrated FEHSS

For the ECG measurement, an electrode patch was connected to a flexible printed circuit board (FPCB) that consisted of an analog front-end chip AD8232 and Bluetooth low-energy module CC2640. The flexible electrodes were made using a ink of PEDOT:PSS/EEG/LiTFSI following our previously established recipe. The signals were sampled at 1250 Hz (8 bit), filtered with a high-pass filter (0.5 Hz) and a lowpass filter (40 Hz) in AD8232, and transmitted to mobile phone by CC2640.

The smartphone charging was conducted by connecting the FEHSS to the smartphone using a type-C cable, and an additional direct current to direct current (DC-DC) converter to regulate the voltage stably at 5.0 V. For the smartwatch charging test, a wireless charger (HUAWEI WATCH charging cradle CP81-1) was used, connected to the FEHSS with an USB cable. For comparison, the smartphone and smartwatch were also charged directly through power supply using an USB cable.

Reporting summary

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