Introduction

Solar cells are a prime choice for powering portable electronics because they offer mobility and independence from traditional power sources1,2,3,4. However, their reliance on sunlight, which is inherently intermittent, limits their effectiveness as a consistent and sustainable energy supply. Photovoltaic powered batteries (PVBs) present a promising solution by integrating solar energy harvesting, conversion, and storage within a single device, thereby enhancing the feasibility of solar power for portable electronics and other self-sufficient smart technologies5,6,7.

Traditional PVBs typically comprise externally connected independent solar cells and batteries. While this configuration is straightforward and relatively efficient, it lacks the potential for high integration, miniaturization, flexibility, and cost-effectiveness. An advanced, fully integrated design could overcome these limitations by harnessing the combined advantages of solar cell and battery technologies. However, developing such integrated devices faces significant challenges, including the need to achieve high overall energy conversion efficiency (ηoverall), substantial energy capacity, and long-term cycling stability8,9,10,11. The pursuit of highly efficient light-active materials with well-engineered surface properties, is critical for optimizing the power conversion efficiency (PCE) from solar to electrical energy12,13,14. Concurrently, the development of stable electrode materials capable of high-capacity multi-electron redox reactions is crucial for enhancing battery capacity and ensuring long-term cycling stability, while minimizing polarization to maximize energy storage efficiency15,16,17. Perovskites, known for their effective application in solar cells as absorbers and in rechargeable batteries as cathodes, show great potential in addressing these challenges and advancing the development of efficient PVB devices18,19,20,21,22,23. To fully leverage their advantages, a holistic strategy is needed to optimally integrate their photovoltaic and electrochemical functionalities, enabling an integrated system for highly efficient energy conversion and storage.

In this work, we explore a dual-functional modulation approach by sharing-using of ethyl viologen diiodide (EVI2) both in perovskite solar cells (PSCs) and rechargeable batteries. The strong aromatic conjunction inherent of EV fosters pronounce interactions with perovskite octahedral units, accelerating electron transfer in the interface of perovskite and charge transport layer in PSCs, while concurrently augmenting the intrinsic charge transfer capacity within the EVSn2I6 cathode in batteries. The strategy demonstrates highly efficient and stable PSCs with a PCE of 26.11%, retaining 96.2% of the initial PCE after 1000 h under ISOS-L-1 protocol. Meanwhile, batteries using EVSn2I6 perovskitoid cathodes deliver an impressive EV0/EV+/EV2+ and I−/I0/I+ multi-electron redox capacity of 296.1 mAh g−1 at 0.5 A g−1, and a capacity of 223.1 mAh g−1 at 5 A g−1 with 89% retention after 10,000 cycles (capacity decay of 0.0011% per cycle). Benefiting from the advanced material design and internal stacking device structure, the integrated PVB devices achieve a ηoverall of 18.54%, while the fabricated flexible PVB devices maintain a ηoverall of 17.62%. This device reliably powers a wearable glucose monitor for continuous 24-h operation, enabling real-time data tracking and transmission, thus highlighting the potential of flexible PVB devices in portable electronics.

Results

Interaction of ethyl viologen diiodide with perovskite

The impacts of EVI2 as a modification layer on the perovskite structure and morphology were characterized using grazing-incidence wide-angle X-ray scattering (GIWAXS) and scanning electron microscopy (SEM) (Supplementary Figs. 1 and 2), indicating that its introduction does not alter the crystal structure or induce the formation of a 2D phase, but rather functions as a surface modifier. Based on this, molecular simulations were employed to construct a schematic representation of the interaction between EVI2 and the perovskite surface, as illustrated in Fig. 1a. To probe the changes in the chemical states of the perovskite surface upon EVI2 modification, we performed X-ray photoelectron spectroscopy (XPS). As shown in Fig. 1b, the introduction of EVI2 results in a shift of Pb 4 f core level from 143.4 eV to 143.1 eV, suggesting an increase in electron density around Pb, likely due to the electron donation from I ions and strong conjugation in EVI224,25,26,27,28,29. Similar shifts can also be observed in the I 3 d core levels, as presented in Supplementary Fig. 3. For further concretizing the moieties acting on the perovskite surface, we investigated the XPS N 1 s spectra of both pristine EVI2 and EVI2-treated perovskite. In Fig. 1c, the N 1 s spectrum of EVI2 reveals two distinct peaks at 399.0 eV and 401.0 eV, corresponding to nitrogen radicals (−N*) and positively charged nitrogen species (−N+) from the viologen group30. Upon deposition onto the perovskite surface, the −N+ peak shifts to 401.4 eV, suggesting an electronic interaction between the viologen group and the perovskite, accompanied by a reduced electron density around the viologen31.

Fig. 1: Interaction of EVI2 with perovskite.
figure 1

a Schematic illustration of EVI2-modified perovskite (angst. = angstrom). b XPS spectra of Pb 4 f for pristine perovskite and EVI2-modified perovskite. c XPS spectra of N 1 s for pristine EVI2 and EVI2-modified perovskite. d The projected density of states (pDOS) plot of the EVI2 extracted from EVI2/perovskite model based on DFT calculation, and e the zoom-out location of conduction band minimum (CBM) + 1 level, originally shown in (d). f The pDOS plot of the perovskite extracted from EVI2/perovskite model based on DFT calculation. g Kelvin Probe Force Microscopy (KPFM) images of perovskite films without and with EVI2 in the dark (CPD contact potential difference, w/o means without, w means with). Scale bar is 500 nm. h, i Light-dependent KFPM potential curves of indium tin oxide (ITO)/p-type organic small molecules/perovskite/C60( ~ 5 nm) films without and with EVI2. The illumination was applied from the top of the sample surface. j Light-dependent CPD statistics of ITO/p-type organic small molecules/perovskite/C60( ~ 5 nm) films without and with EVI2.

Full size image

To further explore the charge redistribution implied by XPS results, we carried out density functional theory (DFT) simulations, including charge-density difference analysis and projected density of states (pDOS). As depicted in Supplementary Fig. 4, incorporation of EVI2 induces an obvious peak at ~0.2 eV and a downward shift in the conduction band minimum (CBM) from ~0.9 eV to ~0.5 eV. These electronic modifications point to a substantial influence of EVI2 on the perovskite surface states, aligning well with the trends observed from XPS measurements. Additionally, charge-density difference analysis (Supplementary Fig. 5) indicates that the electron density accumulates around the viologen group when the EVI2 interacts with perovskite, further corroborating the charge transfer mechanism.

We further extracted the charge-density difference of the EVI2 and perovskite, respectively, based on the computed charge density corresponding to the CBM (Supplementary Fig. 5) and presented their pDOS in Figs. 1d to 1f. Notably, the CBM + 1 position of EVI2 aligns with the CBM of the perovskite, suggesting that electrons can be more easily injected from the conduction band of the perovskite into EVI2 molecules. This band/orbital alignment not only lowers the energy barrier at the interface but also reduces chance of electron-hole recombination, overcoming a limitation of passivation molecules that struggle to efficiently facilitate electron transport. This can be attributed to the strong aromatic conjugated viologen groups, which give rise to efficient carrier transport on the perovskite surface.

The improved electronic structure at the EVI2/perovskite interface is further reflected in optoelectronic measurements. After EVI2 modification, the photoluminescence (PL) intensity of the film is significantly enhanced, and the time-resolved photoluminescence (TRPL) lifetime is prolonged from 223.5 ns to 457.4 ns (Supplementary Figs. 6, 7 and Supplementary Table 1), both indicating reduced non-radiative recombination. In addition, space-charge-limited current (SCLC) measurements, presented in Supplementary Fig. 8, reveal that the trap-filled limit voltage (VTFL) decreases from 0.56 V in the unmodified perovskite film to 0.31 V after EVI2 treatment, indicating a reduction in defect state density32.

To further assess changes in surface properties and charge transport, we used Kelvin Probe Force Microscopy (KPFM) to examine the perovskite films following EVI2 incorporation. Figure 1g shows the surface potential distributions under dark conditions along with the corresponding contact potential difference (CPD). The pristine perovskite film exhibits an inhomogeneous surface CPD ranging from −0.4 to −0.3 V. Whereas the EVI2-modified perovskite film displays a reduced CPD range, indicating a more uniform surface potential distribution. Moreover, further analysis shows that the EVI2-modified perovskite film has a smaller potential difference between the grain boundaries and the grain interior (Supplementary Fig. 9)33,34. The improvement of the uniformity of surface potential distribution is beneficial for effective charge carrier extraction and inhibition of nonradiative recombination35,36. We further fabricated ITO / p-type organic small molecules / perovskite / C60( ~ 5 nm) films and conducted KPFM under different light intensities (Supplementary Figs. 10 and 11). The light-dependent CPD values are summarized in Figs. 1h to 1j. Compared to the pristine perovskite film, the EVI2-modified perovskite film exhibits a more significant shift in CPD values with increasing light intensity. Since photoexcitation at the perovskite / C60 interface causes electron-hole separation and electron extraction to C60, a larger change in CPD indicates more efficient electron extraction from the perovskite to C6037,38,39. This result further validates the effective charge transport facilitated by EVI2, as predicted by the DFT calculations above.

Photovoltaic performance of perovskite solar cells

To further evaluate device performance, we constructed p-i-n structured PSCs as illustrated in Fig. 2a, utilizing C60 and p-type organic small molecules as the electron and hole transport layers (ETL and HTL), respectively. A cross-sectional SEM image of the device is included in Supplementary Fig. 12. The current-voltage (J-V) profiles of the optimal PSCs, without and with EVI2 treatment, under simulated AM 1.5 G illumination are shown in Figs. 2b and 2c. The best-performing EVI2-modified device achieves a PCE of 26.11%, featuring a short-circuit current density (JSC) of 26.17 mA cm−2, an open-circuit voltage (VOC) of 1.186 V, and a fill factor (FF) of 84.12%. This device also exhibits minimal hysteresis, with a forward scan PCE of 25.82%, and a certified maximum power point tracking (MPPT) efficiency of 25.43%, as shown in Supplementary Fig. 13. By comparison, the unmodified PSC yields a lower PCE of 23.70%, with a VOC of 1.121 V and FF of 81.06%. The external quantum efficiency (EQE) spectrum of the champion device is provided in Fig. 2d, closely matching the J-V results, with only slight variation in integrated current. The stabilized performance of the EVI2-treated device is shown in Supplementary Fig. 14, demonstrating a steady current density of 24.67 mA cm−2 and a stable PCE of 25.66%.

Fig. 2: Photovoltaic performance.
figure 2

a Semantic illustration of perovskite solar cell (PSC) structure. ETL electron transport layer, PVK perovskite, HTL hole transport layer, ITO indium tin oxide. b Current-voltage (J-V) curves of the best-performing control and EVI2-modified PSCs (w/o means without, w means with). c J-V curves of the best-performing EVI2-modified PSCs with forward and reverse scans. VOC opencircuit voltage, FF fill factor, JSC short-circuit current density, PCE power conversion efficiency. d External quantum efficiency (EQE) curves and integrated JSC of the EVI2-modified PSC. e Operational stability of PSCs without and with EVI2 with maximum power point (MPP) tracking under continuous simulated 1 sun illumination (ISOS-L-1). f Devices efficiency evolution under light on-off cycle test (12-h/12-h) using simulated 1 sun illumination (ISOS-LC-1).

Full size image

The encapsulated solar cells were tested under ambient air conditions and continuous illumination following the ISOS-L-1 protocol at room temperature, with their output monitored via MPPT. As depicted in Fig. 2e, the EVI2-modified devices exhibit great operational durability, maintaining over 96.2% of their initial value after 1000 h of continuous operation. In comparison, control devices without EVI2 retain only 83.2% of their original PCE after 800 h. To further evaluate long-term stability, a more stringent test based on the ISOS-L-3 protocol was performed, as shown in Supplementary Fig. 15. After 1000 h, the EVI2-modified devices preserve 93.7% of their starting efficiency, whereas untreated ones drop to 76.2%. Additionally, light on-off cycling tests were conducted under ISOS-LC-1 conditions, as presented in Fig. 2f. After 30 cycles (720 h) of alternating 12-h illumination at a temperature of 22 ± 3 °C and 46 ± 8% relative humidity, the EVI2-modified devices exhibit only about 2% efficiency loss, while their unmodified counterparts degrade by 11%. The PSCs incorporating EVI2 achieve great photovoltaic performance and stability, compared with the state-of-the-art PSCs that use other passivation methods (see Supplementary Table 2).

Electrochemical performance of rechargeable batteries

For rechargeable batteries, EVI2 has recently been shown as a promising cathode material due to its high theoretical capacity based on multi-electron redox chemistry40. However, EVI2 cathodes suffer from limited reversibility and sluggish kinetics, leading to poor cycling stability and severe polarization. To address these issues, we introduced Sn cation into EVI2, synthesizing EVSn2I6 perovskitoid. EVSn2I6 exhibits an organic-inorganic hybrid 1D structure, where the cation-π conjugated EV2+ cations are electrostatically bound to [Sn2I6]2−, differing from the fully organic EVI2 crystal structure, as shown in Supplementary Figs. 16 to 18 and Supplementary Table 3. EVSn2I6 is a low-dimensional (1D) perovskitoid, which inherently exhibits superior stability compared to traditional 3D perovskites. To verify the air stability of EVSn2I6, we conducted XRD and XPS analyses on both freshly synthesized samples and samples exposed to air for one week. As shown in Supplementary Fig. 19, the XRD patterns exhibit no structural degradation after air exposure, indicating that the material maintains its crystallinity over time. Furthermore, XPS Sn 3 d spectra confirm that Sn2+ remains the dominant state, with negligible signs of Sn4+ formation even under ambient conditions, indicating that EVSn2I6 does not undergo oxidation upon air exposure. The strong interaction between organic cations and the inorganic framework can limit the diffusion of oxygen into the structure. Moreover, the π-conjugation of EV2+ can provide additional electronic shielding and steric hindrance, reducing the likelihood of Sn2+ oxidation.

Figure 3a shows the configuration of the rechargeable batteries. The coin cells were assembled using synthesized EVSn2I6 cathode, lithium metal anode and modified ether-based electrolyte (1 M lithium bis(trifluoromethane sulfonimide)+ 0.1 M LiCl in dimethoxyethane /1,3-dioxolane, 1: 1 by volume, with 1 wt% LiNO3 additive). For comparison, EVI2 cathodes were assembled under the same conditions. Compared to EVI2, EVSn2I6 offers stronger adsorption energies for iodine species (see Fig. 3b and Supplementary Figs. 20 and 21), suggesting that the EVSn2I6 cathodes entail greater potential for achieving good cyclic performance by suppressing the polyiodide shuttle behavior41. Supplementary Fig. 22 shows the computed charge-density difference of I2, I3−, and I5− on the surface of EVSn2I6, confirming the strong interaction between EVSn2I6 and the iodine species. Moreover, in Supplementary Fig. 23, the computed pDOS of a model slab of EVSn2I6 presents a narrower band gap near the Fermi level, compared to that of EVI2 slab, indicating higher charge transfer ability42,43. These DFT calculation results support the electrochemical advantages of the developed EVSn2I6.

Fig. 3: Electrochemical performance of rechargeable batteries.
figure 3

a Schematic diagram of battery configuration with EVSn2I6 as cathode and lithium metal as anode. b Adsorption energy comparison for the surface of EVSn2I6 and EVI2 toward I2, I3−, and I5−. c Comparison of CV curves at 0.5 mV s−1 and GCD profiles at 1 A g−1 for EVSn2I6 and EVI2 cathodes. d Rate performances and e corresponding GCD curves of EVSn2I6 cathodes from 0.5 to 5 A g−1. f Cycling performance of EVSn2I6 and EVI2 cathodes at 1 A g−1. g Prolonged cycling performances at 5 A g−1. The cathode material loadings are 1.5–1.8 mg cm−2.

Full size image

To evaluate the lithium plating/stripping behavior and Coulombic efficiency (CE) of the electrolyte, Li | |Cu half-cells were performed. As shown in Supplementary Fig. 24, our electrolyte demonstrates a high CE of 99.21% (sample: 4 distinct cells, average CE: 99.07%, standard deviation: 0.13%), indicating efficient and reversible lithium plating/stripping without lithium dendrite formation. For morphological analysis of the deposited lithium, the coin cell was disassembled after plating 3 mAh cm−2 of lithium onto the copper substrate at a current density of 1 mA cm−2, followed by SEM imaging. As shown in Supplementary Fig. 25, the deposited lithium exhibits a dense and smooth surface, which is advantageous for achieving high CE and suppressing dendrite formation. Furthermore, to investigate the composition of the solid electrolyte interphase (SEI), XPS analysis was conducted. The XPS results reveal the presence of inorganic species, including Li3N, LiNxOy and Li2O, suggesting that LiNO3 undergoes reduction to form a Li3N-rich inorganic SEI layer (Supplementary Fig. 26). The inorganic SEI is essential for improving lithium metal stability and supporting extended cycling performance.

Figure 3c presents the cyclic voltammetry (CV) curves recorded at 0.5 mV s−1, alongside the galvanostatic charge-discharge (GCD) profiles measured at 1 A g−1. The CV curves reveal that both cathode materials display four distinct pairs of redox peaks, corresponding to their respective redox processes of EV0/EV+, EV+/EV2+, I−/I0, and I0/I+ from low to high voltage, respectively, demonstrating the reversible electrochemical activity of both EV2+ cations and I− anions22,44. The EVSn2I6 cathode exhibits reduced polarization compared with EVI2 cathode, implying improved reaction kinetics45. The GCD curves depict four distinct charge-discharge plateaus corresponding to the CV curves. Notably, the stronger iodine redox peaks of EVSn2I6 contribute to the higher average voltage of EVSn2I6 (2.98 V) than that of EVI2 (2.76 V).

The rate performances of EVSn2I6 and EVI2 cathodes were assessed from 0.5 to 5 A g−1, as illustrated in Fig. 3d. At 0.5 A g−1, the EVSn2I6 cathode exhibits a high reversible capacity of 296.1 mAh g−1, and retains 212.4 mAh g−1 at a high specific current of 5 A g−1. When the specific current returning to 0.5 A g−1, the capacity recovers to 275.8 mAh g−1. In contrast, the EVI2 cathode exhibits dramatically decreased capacity during cycling. Despite its higher theoretical capacity (343.6 mAh g−1) compared to EVSn2I6 (309.1 mAh g−1), the EVI2 cathode suffers from poor capacity retention, resulting in considerably lower practical reversible capacity over prolonged cycling. Herein, the capacity fading of EVI2 is mainly attributed to the dissolution of its organic molecular structure in the organic electrolyte40,46. In contrast, EVSn2I6 exhibits better cycling performance due to its unique composition and electronic structure. The GCD profiles of EVSn2I6 in Fig. 3e reveal distinguishable charge/discharge plateaus at different rates, underscoring its great redox reversibility and kinetics.

Figure 3f illustrates the cycling stability of the batteries at 1 A g−1. The EVSn2I6 exhibits an initial capacity of 274.0 mAh g−1 and maintains 93% after 200 cycles, corresponding to a capacity fade rate of 0.035% per cycle. As shown in Fig. 3g, the prolonged cycling performance was evaluated over 10,000 cycles at a high specific current of 5 A g−1. The EVSn2I6 initially delivers 223.1 mAh g−1 and retains 89%, translating to a low per-cycle degradation rate of 0.0011%. Supplementary Fig. 27 shows the morphology of Li metal anode in Li | |EVSn2I6 cell after the long-term cycling. The plated Li shows a uniform and dense morphology with a large granular structure. Herein, the lithium utilization rate is ~11.6% with a cathode loading of 1.62 mg cm−2. To investigate the impact of lithium utilization rate, we tested a higher utilization design (~53.8%) by increasing the cathode loading to ~8.1 mg cm−2. This approach resulted in a capacity retention of 81% after 200 cycles (capacity decay of 0.095% per cycle, Supplementary Fig. 28). As summarized in Supplementary Table 4, the EVSn2I6 cathodes show great cycling stability. Therefore, the EVSn2I6 cathode materials can provide fast and stable multi-electron storage with high capacity and low polarization for rechargeable batteries.

Performance of photovoltaic-powered battery devices

The EVI2 and the derived EVSn2I6 demonstrate high performances in PSCs and rechargeable batteries, respectively. Building on these successes, we designed all-perovskite-based PVB devices on rigid substrates for achieving integrated solar energy conversion and storage (Supplementary Fig. 29). The structure of the PVB device is presented in Fig. 4a. In this device, a micro-module design with 4 sub-cells (total area: 0.768 cm2, Supplementary Fig. 30) was applied as the photovoltaic unit to ensure that sufficient voltage and steady current can be provided to charge the battery. Figure 4b shows the J-V curves of the micro-module under simulated 1 sun illumination. The modules achieved a PCE of 23.60%, with a VOC of 4.41 V, a JSC of 6.455 mA cm−2, and an FF of 82.92%. The silver electrode and the indium tin oxide (ITO) electrode from the PSC part are connected to the anode and the EVSn2I6 cathode of the battery component, respectively. To prevent the leakage of the battery’s electrolyte, the interlayer of the device is isolated with a polyethylene terephthalate (PET) film. During photo-charging, the photogenerated electrons in the perovskite layer transfer through the ETL to the anode. The solar-generated holes transfer through the HTL toward ITO and then to the cathode to achieve the oxidation reaction of the EVSn2I6. While discharging, the stored energy can be released according to the power supply demand as a standard rechargeable battery.

Fig. 4: Performance of fabricated all-perovskite-based photovoltaic powered battery (PVB) device.
figure 4

a Exploded-view illustration of the PVB device. ITO indium tin oxide, HTL hole transport layer, PVK perovskite, ETL electron transport layer, PET polyethylene terephthalate, red arrow of h+ hole transfer, blue arrow of e− electron transfer. b, c J-V curves of rigid (b) and flexible (c) micro-module under simulated 1 sun illumination, the inset is the cyclic discharge capacity and overall energy conversion efficiency (ηoverall) of the PVB device under 1 sun illumination. The cathode material loading of PVB is ~1.6 mg cm−2. d Signals recorded by the intelligent controlled flexible PVB device-powered continuous glucose monitor for 24 h. e Battery voltage-time profiles of the intelligent controlled flexible PVB device-powered continuous glucose monitor under different light sources for 24 h. f Schematic diagram of the continuous glucose monitor powered by the intelligent controlled flexible PVB device.

Full size image

The photo-charge and galvanostatic discharge experiments were controlled manually with a solar simulator and battery tester. To activate the battery, the PVB device was galvanostatic charged and discharged for the initial 10 cycles before photo-charging. The ηoverall for the photo-charge and galvanostatic discharge experiments can be calculated by the equation: \({\eta }_{{{\rm{overall}}}}=\frac{{E}_{{{\rm{d}}}}}{P\times S\times t}\times 100\%\), where Ed, P, S, and t are discharge energy, light power density, effective area of PSCs, and photo-charge time, respectively8. The inset of Fig. 4b shows the cyclic discharge capacity and ηoverall of the PVB device under 1 sun illumination. It demonstrates high stability over 100 cycles, with a gradual decrease in capacity from 208.1 mAh g−1 to 192.9 mAh g−1. As shown in Supplementary Figs. 31 and 32, the photo-charge/galvanostatic discharge curves are similar to the galvanostatic charge/discharge curves, revealing the great energy harvesting and storage capability of the device. The ηoverall of the PVB device reaches 18.54% and stays above 18.37% throughout the cycles, demonstrating great performance relative to previously reported PVB devices with internal energy storage (Supplementary Fig. 33 and Supplementary Table 5)5,7,8,47,48,49,50,51,52,53,54,55,56,57.

The Photo-charge/galvanostatic discharge tests of the PVB device were conducted under different light intensities and varying light sources. As shown in Supplementary Fig. 34, the ηoverall of the PVB device decreases as light intensity decreases, while the charging duration and specific capacity increase, due to the reduction in charging current at lower light intensities. Nevertheless, the device maintains stable charge-discharge performance across various light intensities, demonstrating its reliability. When the light source was switched to room light, the device sustained a stable voltage of ~3.49 V over an extended charging duration of 2000 min.

To achieve portable applications, the rigid glass substrate was replaced with flexible polyethylene naphthalate, enabling the flexible manufacturing of the PVB device. Under 1 sun illumination, the flexible micro-modules can deliver a VOC of 4.38 V, a JSC of 6.337 mA cm−2 with an FF of 80.87% and a PCE of 22.45% (Fig. 4c and Supplementary Fig. 35). Furthermore, the flexible micro-module can retain 98.89% of its initial performance after 1000 bending cycles at a curvature radius of 8 mm, demonstrating good mechanical durability (Supplementary Fig. 36).

As shown in the inset of Fig. 4c, the fabricated flexible PVB device maintains a ηoverall of up to 17.62% and above 17.42% for 100 cycles, with a slow capacity decay from 212.9 mAh g−1 to 190.7 mAh g−1. To assess the mechanical durability of the flexible device, we conducted bending stability tests under repeated bending cycles at a curvature radius of 8 mm. The device maintains about 88% of its ηoverall and 82% of its discharge capacity after 1,000 bending cycles (Supplementary Fig. 37), indicating its bending durability for practical applications. To evaluate temperature tolerance and mechanical robustness, we tested the flexible PVB device on a − 20 °C frozen saltwater block and a 40 °C heated platform, representing low-temperature and high-temperature conditions (Supplementary Fig. 38a, b). The device successfully powered an LED bulb in both scenarios, demonstrating the device’s thermal reliability and safety. Additionally, repeated hammering tests simulated external impacts in wearable applications. Despite these stresses, the PVB retained its structural integrity and functionality, continuously powering the LED bulb, confirming the effectiveness of our encapsulation strategy in preventing mechanical failure (Supplementary Fig. 38c).

Moreover, to incorporate an intelligent control feature into the device, we designed a charging voltage protection board for the PVB device to prevent overvoltage during charging. As shown in Supplementary Fig. 39, the designed charging voltage protection board can determine whether to accept current from the PSC based on the voltage of the battery. When the battery voltage exceeds 3.70 V ( ± 50 mV), it blocks the charging current and ensures that energy is not released from the battery to the PSC. To evaluate the intelligent controlled flexible PVB device for application in portable electronics, we integrated the device with a mobile continuous glucose monitor, as shown in Supplementary Fig. 40. The device can enable continuous 24-h glucose monitoring and send the glucose data through Bluetooth in real time, owing to the stable power self-supply (Fig. 4d). We further tested the battery voltage-time profiles of the intelligent controlled flexible PVB device powered continuous glucose monitor under sunlight, room light, and dark conditions to study its operational capability in different environments. As shown in Figs. 4e and 4f, the battery can be charged to 3.70 V within 0.08 h under sunlight. The battery powers the glucose monitor every five min, causing a momentary voltage to drop. With the combined effect of sufficient sunlight intensity and the charging voltage protection board, the battery voltage remains around 3.70 V. When the light intensity transitions to indoor lighting, the reduced light intensity lowers the supply current, resulting in a decreased battery voltage, stabilizing around 3.44 V. In the dark, the battery can continue to operate for over 8 h. Throughout the entire power supply process, the battery voltage remains above 2.69 V. Consequently, the PVB devices show great potential as wire-less photo-charging power sources for portable electronics.

Discussion

In this work, we demonstrate an effective material-sharing strategy by using EVI2 to bifunctionally modulate the performance of PVB devices, leading to a significant advancement in efficient and stable energy conversion and storage technologies. The EVI2-modified PSCs achieve a high PCE of 26.11% with 96.2% maintenance after 1000 h of operation tracking under continuous simulated 1 sun illumination (ISOS-L-1), while the EVI2-derived EVSn2I6 cathode materials exhibit an ultralong high-rate cycling stability with an impressive EV0/EV+/EV2+ and I−/I0/I+ multi-electron redox chemistry, delivering a specific capacity of 223.1 mAh g−1 at 5 A g−1 with 89% retention after 10,000 cycles (capacity decay of 0.0011% per cycle) in batteries. More significantly, the integrated PVB devices demonstrate a high ηoverall of 18.54%. Beyond the performance metrics, the flexible PVB devices with a ηoverall of 17.62% and good stability over 100 cycles, further showcase potential for powering portable electronics. When integrated with an intelligent control unit, this device can reliably provide uninterrupted power for a continuous blood glucose monitoring device, enabling 24-h real-time tracking and seamless data transmission. These results constitute a significant advance toward efficient and reliable power solutions of using PVB devices as sustainable, self-sufficient energy sources for personalized and portable electronics.

Methods

Materials

Methylammonium chloride (MACl, 99.99%), Cesium iodide (CsI, 99.999%) and Formamidinium iodide (FAI, 99.99%) were purchased from GreatcellSolar. Lead iodide (PbI2, 99.99%), p-type organic small molecules including [2-(3,6-Dimethoxy-9H-carbazol-9-yl)ethyl] phosphonic acid (MeO-2PACz, 98%) and [4-(3,6-Dimethyl-9H-carbazol-9-yl)butyl] phosphonic acid (Me-4PACz, 99%) were purchased from TCI. Ethyl viologen diiodide (EVI2, 99%), tin(II) oxide (SnO, 97%), hydriodic acid (HI, 55-58 wt.% in H2O), hypophoaphoeous acid (H3PO2, AR, 50 wt.% in H2O), lithium chloride (LiCl, 99%), and methanol (99.9%, anhydrous) were purchased from Sigma Aldrich. Methylammonium iodide (MAI, 99.9%), C60 (purity of 99.5%), bathocuproine (BCP, purity of 99.9%), and Octane-1,8-diamine Dihydroiodide (ODADI, 99.5%) were purchased from Xi’an Yuri Solar Co., Ltd. High purity silver (99.99%) was purchased from Zhongnuo Advanced Material Co., Ltd. Solvents, including isopropanol (IPA, 99.5%, anhydrous), dimethylformamide (DMF, 99.8%, anhydrous), dimethyl sulfoxide (DMSO, 99.9%, anhydrous), and chlorobenzene (CB, 99.8%, anhydrous) were purchased from J&K and used as received. Polyethylene naphthalate (PEN)/indium tin oxide (ITO) or Glass/ITO substrates (15 Ω sq−1) were received from Advanced Electron Tech. (China). Electrolyte of 1 M lithium bis(trifluoromethane sulfonimide) in dimethoxyethane/1,3-dioxalane = 1:1 vol.% with 1 wt% LiNO3 (1 M LiTFSI in DME/DOL = 1:1 vol.% with 1% LiNO3, commercial product) was purchased from Duoduo Chem Co., Ltd. Lithium metal (Li, >99.99%, 20 μm deposited on copper foil), Ketjen black (EC-300J), N-Methyl-2-pyrrolidone (NMP, 99.9%), and polyvinylidene difluoride (PVDF, 99.5%) were purchased from Canrd (China).

Fabrication perovskite solar cells and the micro-modules

Rigid glass/ITO or flexible PEN/ITO substrates, each with a sheet resistance of 15 Ω sq−1, were subjected to sequential ultrasonic cleaning. The process involved immersing the substrates in detergent, deionized water, acetone, and isopropanol, each for a duration of 20 min. Following the cleaning steps, the substrates were dried in an oven at 100 °C. They then underwent oxygen plasma treatment for 10 min. After that, the substrates were immediately transferred into a N2-filled glovebox. A solution was prepared by dissolving MeO-2PACz and Me-4PACz in methanol at concentrations of 0.3 mg mL−1 each, maintaining a volumetric ratio of 1:2, respectively. This precursor mixture was deposited onto the ITO surface via spin-coating at 4000 revolutions per minute for 30 sec. Finally, the coated substrates were subjected to thermal annealing at 100 °C for 5 min.

To prepare the perovskite precursor solution, a concentration of 1.5 M was achieved by dissolving 19.5 mg of CsI, 219.3 mg of FAI, 23.8 mg of MAI, and 760.7 mg of PbI2 (which included a 5% molar excess) into a 1 mL mixed solvent system consisting of DMF and DMSO in a 4:1 volume ratio. The resulting composition corresponds to the chemical formula Cs0.05FA0.85MA0.1PbI3. To improve the perovskite film quality, MACl was introduced at a molar concentration of 12%, along with 0.6 mol% of ODADI as an additive. The complete mixture was stirred continuously for 2 h to ensure homogeneity of the precursor solution. The perovskite precursor solution was deposited onto pretreated substrates containing p-type organic semiconductors through a two-step spin-coating process. Initially, the substrates were spun at 1000 revolutions per minute for 10 sec, followed by a second stage at 5000 revolutions per minute for 40 sec. During the final 5 sec of this stage, 200 μL of CB was dispensed at the center of the spinning substrate to facilitate crystallization. Immediately after spin-coating, the films were placed on a hotplate and thermally annealed at 100 °C for 30 min. For surface treatment, a solution of EVI2 (1 mg mL−1) was prepared using IPA as the solvent. This solution was dynamically spin-coated onto the previously formed perovskite layers at 5000 revolutions per minute for 30 sec, followed by thermal treatment at 100 °C for 10 min to complete the surface engineering process. The spin-coating operations were carried out inside a N2-filled glovebox with a controlled temperature maintained at 18 − 23 °C using an integrated air-conditioning system, and both moisture and oxygen concentrations were controlled below 5 parts per million (ppm).

For the fabrication of PSCs, successive thermal evaporation was employed to deposit functional layers on the perovskite film. A 25 nm thick C60 layer was first deposited at a rate of 0.5 Å s−1, followed by the deposition of a 6 nm BCP layer at the same rate. Subsequently, a 100 nm Ag electrode was thermally evaporated at a deposition rate of 1.0 Å s−1. These depositions were conducted under a high vacuum environment, with the pressure maintained below 4 × 10−6 Torr. The device area was defined and characterized as 0.0506 cm2 by metal shadow mask.

For the fabrication of micro-modules, C60, BCP, and Ag were sequentially deposited on the perovskite film using the same thermal evaporation protocol employed for fabricating PSCs. To define interconnections within the module structure, two laser scribing steps were carried out using a laser marking system (Han’s Laser PV-HL0303) operating at a wavelength of 532 nm. The scribing parameters for the P2 and P3 lines were set to 2.5 W at 100 kHz (line width: 100 μm) and 5 W at 100 kHz (line width: 80 μm), respectively. The final micro-module consists of 4 sub-cells (each with an area of 0.192 cm2) connected in series, with a total area of 0.768 cm2.

Synthesis of EVSn2I6

A total of 58.5 mg of EVI2 was dissolved in a solvent mixture comprising 3 mL of HI and 0.5 mL of H3PO2. In a separate step, 33.5 mg of SnO was dissolved in a solution containing 2 mL of HI and 0.2 mL of H3PO2. Both mixtures were heated and stirred at 120 °C until complete dissolution was achieved, resulting in clear solutions. Subsequently, the SnO solution was gradually added into the EVI2 solution, and the temperature was increased to 140 °C. The combined solution was maintained under continuous stirring for 2 h, during which it turned transparent. Cooling was then carried out at a controlled rate of 5 °C per hour until the mixture reached room temperature. The resulting crystalline product are washed several times with deionized water and IPA, and dried in a vacuum oven at 60 °C for 12 h.

Fabrication of batteries

The electrodes were formulated by combining 80 wt.% EVSn2I6, 10 wt.% Ketjen black, and 10 wt.% binder of PVDF in a solvent of NMP. The mixture was stirred for 2 h via a magnetic mixer at around 25 °C in air. The resulting slurry was uniformly coated onto flexible carbon cloth using doctor blade and dried overnight at 60 °C in a vacuum oven. Circular electrodes with a diameter of 1.2 cm were cut using a precision disc cutter, yielding a cathode mass loading of 1.5–1.8 mg cm−2. These electrodes were further vacuum-dried at 60 °C for 8 h by a vacuum oven before being transferred into an Ar-filled glovebox. The electrolyte was prepared by dissolving LiCl (0.1 M) into a commercial electrolyte consisting of 1 M LiTFSI in a 1:1 volume ratio of DME/DOL, which also contained 1 wt.% LiNO3. This solution was thoroughly mixed using a magnetic mixer for 24 h inside an Ar-filled glovebox to dissolve fully (The ether-based DME/DOL solvent system ensures high ionic conductivity and low viscosity, enabling efficient lithium-ion transport even at sub-zero temperatures. LiNO3 contributes to the formation of a stable solid electrolyte interphase (SEI) on the lithium metal surface, significantly reducing dendrite formation and enhancing cycling safety. Moreover, LiTFSI provides superior thermal and electrochemical stability compared to conventional LiPF6-based electrolytes.) Lithium metal in the form of a 20 μm-thick layer deposited on copper foil (1.2 cm in diameter) was employed as anode. A Celgard 2325 trilayer microporous membrane, 25 μm thick and 1.6 cm in diameter, served as separator. CR2032 coin cells were assembled at room temperature (~25 °C) inside an Ar-filled glovebox. For sufficient electrochemical reaction at the cathode, excessive lithium metal and electrolyte (40 μL per cell) were used. The thin-film lithium anode on the copper substrate offers advantages such as enhanced flexibility, mechanical robustness, and uniform lithium availability. EVI2 cathodes were fabricated and integrated into cells under identical conditions. All specific capacity values were calculated based on the mass of the cathode material.

Assembly of photovoltaic powered battery devices

The PVB devices were constructed by integrating battery components onto the PSC module within an Ar-filled glovebox. A PET film was adhered to the module using UV glue, to isolate the perovskite solar cell from the battery section, while also providing protection to the perovskite layer. Cathodes and anodes were cut into 1 cm × 1 cm squares. Copper foil, serving as the current collector, was placed in direct contact with the silver electrode of the PSC. Lithium metal, pre-deposited onto the copper foil, was used as the anode, followed by the placement of a separator. The electrode and separator (1.2 cm × 1.2 cm) were affixed and sealed using UV glue to prevent electrolyte leakage. After injecting 30 μL of electrolyte, the EVSn2I6 cathode, supported by an aluminum foil current collector, was positioned on top. A connection between the cathode and the ITO electrode of the PSC was established using silver-plated copper wire. The aluminum packing foil (which provides great water and oxygen barrier properties), UV glue (which forms a flexible, tight, and chemically stable seal), and PET tape (which enhances mechanical flexibility and structural integrity) were used for encapsulation, ensuring both mechanical robustness and environmental durability. For the application, a charging voltage protection board was connected between the PSC’s ITO electrode and the EVSn2I6 cathode and was built into the commercial continuous glucose monitor.

Material characterization

SEM and elemental mapping were carried out using a QUATTRO S field-emission scanning electron microscope. XPS analysis was performed with an EscaLab XI+ spectrometer. For both SEM and XPS characterizations of cycled Li metals, the cells were disassembled within an Ar-filled glovebox and the Li metals were transferred using a sealed sample holder under inert Ar atmosphere to prevent exposure to ambient air or moisture (temperature maintained at ~25 °C). XRD patterns were acquired using a Rigaku SmartLab diffractometer. PL spectra were collected on an Edinburgh FLS980 spectrometer equipped with a 450 nm excitation source. KPFM related characterizations were conducted using the SHIMADZU SPM-9700HT system. CPD was recorded under both dark and illumination conditions, with white LED illumination directed at a 15° angle from the top of the sample, providing a power density of 5 mW cm−2.

Device measurements

Photovoltaic performance measurements were conducted inside a N2-filled glovebox at ambient temperature, employing a Xenon arc lamp solar simulator (Enlitech SS-F5, Taiwan). The illumination intensity was calibrated to 100 mW cm−2 using a certified silicon reference cell equipped with a KG2 optical filter. To improve light transmittance through the device, a 120 nm thick magnesium fluoride anti-reflection layer was pre-deposited onto the rear side of the ITO substrate. Current-voltage characteristics were recorded using a Keithley 2400 source meter. Measurements were performed under both reverse bias (from 1.20 V to −0.01 V) and forward bias (from −0.01 V to 1.20 V) scanning modes, with a voltage sweep rate of 0.01 V s−1 and an integration delay of 10 ms per data point. For the micro-modules, current-voltage measurements were carried out over a voltage range of −0.1 V to 5.0 V, employing a scan rate of 0.05 V s−1 with a delay time of 10 ms per data point. The stabilized power output was evaluated by tracking the steady-state current density at the MPP, which was determined from the reverse scan of the J-V characteristics. EQE spectra were acquired using an Enlitech QE-R measurement system. Long-term device stability was assessed following the ISOS-L-1 testing protocol under continuous illumination from a 1-sun equivalent LED light source. The tests were conducted under ambient conditions, with the temperature maintained at 22 ± 3 °C and relative humidity controlled at 46 ± 8%. The PSCs were biased at MPP voltage, and the power output was tracked by using a CHI1040C multi-potentiostat. To ensure accurate MPP tracking, the current density-voltage curves were recorded every 12 h, allowing periodic adjustment of the load conditions. For the on-off light cycling stability test, the 1 sun AM1.5 G equivalent LED light source was employed to simulate a diurnal illumination pattern consisting of 12 h of light followed by 12 h of darkness. This cycling was repeated over 30 cycles, corresponding to a total duration of 720 h. Electrochemical evaluations were conducted using 20 μm lithium metal directly on a copper foil substrate as anodes and took place under ambient condition with room temperature (~25 °C). CV analyses were performed utilizing a CHI 760E electrochemical workstation. The GCD cycling at different specific currents was carried out with a CT3002A LAND battery tester. For PVB devices, the photo charge/galvanostatic discharge cycling were assessed using a combination of a solar simulator and the battery tester.

Computational methodology

All first-principles density-functional theory (DFT) calculations were conducted using the Vienna Ab Initio Simulation Package (VASP 6.4)58,59,60. To investigate the interaction between EVI2 and the perovskite (see Supplementary Figs. 4 and 5), the Perdew-Burke-Ernzerhof (PBE) functional within the generalized gradient approximation (GGA) framework was employed60. For modeling the surface structures of both pristine and EVI2-treated FAPbI3 perovskite with a (001) PbI2-terminated facet, a 2 × 2 × 1 k-point mesh for vacuum-bearing slab, using the Monkhorst-Pack scheme. The electronic constituents considered were 2 s and 2p for C and N, 6 s and 5 d for Pb, and 1 s for H. Considering the heavy element atoms’ (Pb) strong relativistic effect, spin-orbit coupling (SOC) was included in all calculations related to electronic properties, though excluded during geometric optimization. Projector augmented wave (PAW) pseudopotentials were utilized, with a cut-off energy of 600 eV. To account for van der Waals interactions between protonated amines and high-electronegativity groups, the DFT-D3 method by Grimme (with zero damping) was applied with the PBE functional during geometry optimization61,62,63. Structural relaxation was performed until the force on each atom was less than 10−4 eV Å−1, ensuring mechanical equilibrium. A vacuum spacing of 25 Å was introduced on the surface structures to eliminate interactions between fixed terminal layers.

To compare the performance of EVI2 and EVSn2I6 as cathode materials, surface models were constructed using crystallographic data (CIF files) and XRD patterns (see Supplementary Figs. 16 to 18). Specifically, the (1 3 − 1) plane was built for EVI2, while the (011) surface was modeled for EVSn2I6. In both cases, the PAW method was employed to describe the ion-electron interactions, with a kinetic energy cutoff of 520 eV. To account for dispersion Van der Waals interactions, the Grimme’s DFT-D3 scheme was applied. The Brillouin zone was sampled using a gamma k-point Monkhorst-Pack grid, with a larger 5 × 5 × 1 k-point used for structure optimization and electronic property computation. The calculations were performed until the total energy was below 10−5 eV per atom and the force was less than 0.02 eV Å−1. A vacuum layer of 15 Å along the z-direction was introduced to largely reduce the interaction between the periodically repeated unit cells. The adsorption energy (ΔEad) can be calculated by: ΔEad = ΔETotal – ΔEbare – ΔEadsorbate, where ΔETotal, ΔEbare, ΔEadsorbate represent the electronic energy of adsorption system, bare system, and adsorbate (see Supplementary Figs. 20 and 21) respectively.

Reporting summary

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