The MnO₂ nanoparticles and MXene nanosheets demonstrate superior solution processability. Flexible MnO₂ film is readily fabricated by blade-coating, and paper-like MXene film is prepared by dip-coating onto an interconnected cellulose scaffold (Fig. 2c and Supplementary Fig. 4). Proton-acid treatment³⁰ can enhance the ambient stability of MXene films by suppressing water and oxygen intercalation (Supplementary Fig. 5). The as-prepared MnO₂ and MXene films both exhibit porous nanostructures, which facilitate interfacial interactions and ion transport (Supplementary Fig. 6 and 7).

Furthermore, we analyze the interfacial properties of IAIs. Zeta potential testing indicates that MnO₂ and MXene films are positively and negatively charged, respectively, as a result of surface functional group dissociation (Fig. 2d). Kelvin probe force microscopy (KPFM) testing shows surface potential difference between MnO₂ (~120 mV) and MXene film (~−285 mV, Fig. 2e and f). The work functions of MnO₂ (4.5 eV) and MXene film (4.1 eV) are  determined by  ultraviolet photoelectron spectroscopy (UPS, Fig. 2g). These results consistently confirm the reverse surface charge characteristic of the two materials, resulting in an intrinsic potential difference. Benefiting from distinct surface charge and porous nanostructures, the interfaces of MnO₂ and MXene are capable for spontaneously storing sufficient interfacial ion based on electric double layer (EDL) effect, thereby facilitating natural formation of IAIs. To further quantify the intrinsic ion storage at interface, we introduce a parameter of intrinsic interfacial capacitance (C_i), which can be measured by electrochemical impedance spectroscopy (EIS)³¹ (see the Methods section for details):1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$-{Z}^{{\prime} {\prime} }=\frac{1}{{C}_{i}}\times \frac{1}{2\pi f}$$\end{document}−Z″=1Ci×12πfwhere f and Z^” represents frequency and imaginary parts impedance, respectively. The low-frequency response in EIS can reflect the capacitive characteristics of interface, which are related to ion adsorption and accumulation at interface. As shown in Fig. 2h, the C_i of MnO₂ and MXene (2 × 2 cm²) is 62 and 83 mF, respectively, demonstrating substantial ion accumulation at interface. Therefore, the interfaces of MnO₂ and MXene demonstrate clear iontronic asymmetry and significant intrinsic storage, validating the available construction of IAIs (Fig. 2i).

The AEPG delivers cyclic electric output without pre-charging from an external electric power. After the completed release of pre-stored energy through prolonged short-circuiting discharge, the AEPG recovers to its initial voltage (~0.45 V) at re-off state, validating its inherent regenerative property (Fig. 3b and Supplementary Fig. 13). Moreover, the AEPG can be discharged in a tunable constant-current manner. Upon discharging to 0 V, the full-discharged AEPG repeatedly recovers its voltage at the re-off state, further affirming re-generating behavior (Fig. 3c and Supplementary Fig. 14).

Relying on an intermittent cyclic power generation mode, the AEPG undergoes cycles of regeneration, energy storage, and periodic energy output during discharge. This mode inspires the design of an environment-triggered energy-release gate, facilitating the device to achieve an adaptive and environmentally compatible intermittent operation. The energy-release gate can be implemented through a stimuli-responsive on/off switch, as illustrated in Supplementary Fig. 15. To verify its effectiveness, we systematically compare the current and electric energy output characteristics of the device with and without the energy-release gate. In the absence of an energy- release gate, the device generates continuous current output upon triggering on-state, but the current rapidly drops from 30 mA to 0.015 mA within ~1 min (Fig. 3d). Correspondingly, the output energy curve exhibits a plateau, with the output electricity energy density approaching a limited value of 16.5 J m⁻² ( ~ 115 h, Fig. 3e), exhibiting disposable power supply. In contrast, the AEPG with energy-release gate affords intermittent cyclic current signals (Fig. 3f). The output energy density exhibits a linear accumulation with the increasing cycles, reaching up to 543 J m⁻² ( ~ 115 h, Fig. 3g), which is ~33 times higher than that of device without energy-release gate. The result reveals that energy-release gate is available for optimizing energy conversion pathway and essential to giant electric output. In the lack of an energy-release gate, the generated energy of device is unable for accumulation, leading to continuous but weak current output that is practically unusable. In essence, the energy-release gate enables physicochemical interface modulation, which allows ions to be stored at the interface and then collectively released, thereby optimizing energy conversion pathway. The paradigm of slow harvesting and collective release, provides an alternative route for achieving giant conversion from low-grade energy.

The electrode nanostructure facilitates interfacial ion adsorption owing to its high specific surface area, thus ensuring high current output (Supplementary Fig. 16). The AEPG operates reliably over a wide temperature range (0 to 60 ^oC, Supplementary Fig. 17) and shows good compatibility with hydrogel electrolyte (Supplementary Fig. 18). The AEPG operates in various electrolytes, with performance differences arising from variations in ion mobility (Supplementary Fig. 19a). The generated voltage varied slightly with the ion concentration, while the current signal exhibited increasing variation with ascending concentration owing to enhanced C_i and ion migration (Supplementary Fig. 19b). On the account of the influence of separator thickness on device resistance, the current output of AEPG was visibly decreased with increasing spacing (Supplementary Fig. 19c and d). Different recovery time has an effect on electric output, because AEPG requires an adequate recovery period to return to its initial state (Supplementary Fig. 20).

Above-mentioned mechanism is further supported by experimental and theorical results related to interfacial ion migration and electron transfer. Electrochemical quartz crystal microbalance (EQCM) monitors trace-level mass change (Δm) through frequency change (Δf), enabling real-time tracking of ionic fluxes at interface (Fig. 4b). Upon discharging, a significant decrease in electrode mass is observed, signifying the expulsion of interfacial ion from interface (Fig. 4c). In the re-generating state, the electrode mass increases and returns to its initial value due to the ion re-migration to interface. The periodic variation in the electrode mass provides strong evidence for the reversible ion migration at interface during cycling power generation. Kinetic Monte Carlo simulation further uncovers reversible ion migration, which is modeled by electrode pairs with opposite charges (Supplementary Fig. 22). Simulated ion distribution at interfaces renders reversible variation during cyclic power generation, coinciding with the experimental results (Supplementary Fig. 23).

Along with interfacial ion-oscillation, the electrode potential exhibits dynamically reversible fluctuations during electricity generation (Fig. 4d and Supplementary Fig. 24). Cyclic voltammogram curves of electrode display a notably rectangular shape within the scanning window corresponding to fluctuated range, devoid of discernible redox peaks, elucidating their capacitive property and reliable electrochemical stability (Supplementary Fig. 25). In-situ Raman spectra illustrate that the characteristic peaks of MnO₂ and MXene electrodes remain no discernible variation during cycling electricity generation, demonstrating their stable phase structure characteristics (Fig. 4e and Supplementary Fig. 26). The morphological microstructure and chemical composition of the electrodes display negligible change after energy generation cycles, as verified by SEM and X-ray photoelectron spectroscopy results (Supplementary Fig. 27 and 28).

In the re-generating process, the accumulated interfacial charge could initiate interfacial redox reaction, which plays a critical role in ion re-adsorption and energy regeneration. Interfacial charge transfer consumes dissolved oxygen or yields reactive oxygen intermediates (e.g., hydroxyl radical)^34,38,40. To validate this mechanism, dissolved oxygen concentration in the electrolyte is monitored during cycling electricity generation using dissolved oxygen meter (Supplementary Fig. 29a). The concentration of dissolved oxygen decreases obviously after repeated electricity generation, indicating oxygen consumption during re-generating process (Fig. 4f). The consumption amount of oxygen correlates positively with the transferred charge (Supplementary Fig. 29b). When oxygen into the electrolyte is inhibited, the AEPG device fails to recover its voltage, whereas vigorous stirring greatly enhances oxygen mass transport and leads to rapid regeneration (Supplementary Fig. 30a and b). The spontaneity of the interfacial redox reaction is confirmed using a galvanic-like cell configuration, where the observed cell voltage (~0.2 V) indicates that dissolved oxygen-mediated interfacial electron transfer could be thermodynamically favorable (Supplementary Fig. 30c and d). Above results coherently corroborates the significant role of dissolved oxygen in mediating interfacial charge transfer.

EPR spectroscopy is employed to detect hydroxyl radical (·OH) during re-generating process, using 5,5-dimethyl-1-pyrroline N-oxide (DMPO) as a spin-trapping agent^40–42. As shown in Fig. 4g, no visible signal is detected on the MXene electrode prior to power generation. In contrast, the quadruplet DMPO-·OH (1:2:2:1) characteristic peaks appears during power generation, evidencing hydroxyl radical formation. Meanwhile, no DMPO-trapped radical signals are detected at the MnO₂ electrode either before or after power generation. Therefore, the observed dissolved O₂ consumption and •OH formation can verify the presence of the interfacial reaction, which mediates interfacial charge transfer and supplies the driving force for reversible ion migration and energy regeneration.

In the energy conversion process of AEPG, low-graded chemical energy derived from interfacial redox reaction could serve as the energy input. This input is harvested and stored at EDL interfaces as an intermediate state, and ultimately converted into electrical energy. The energy conversion efficiency of APEG is estimated to be about 23% (Supplementary Fig. 31, see the Supplementary Note 1 for details)

Benefiting from foldable traits of materials, the scalable AEPG device can be further transformed into miniaturized and portable reconfiguration by using stacking strategy, which fully exploit vertical space. A tiny bulk of monolithic device (2 cm × 2 cm × 5 mm) supplies an I_sc of ~143 mA, a giant area power density of 18.4 W m⁻², and a volumetric power density of up to 3680 W m⁻³ with a load resistance of 5 Ω, prior to that of reported IPGs (Fig. 5e and Supplementary Fig. 36). Additionally, the AEPG is highly flexible and durable under repeated mechanical deformations (Supplementary Fig. 37). Even after bending for 1000 cycles, AEPG still maintains 95% current level (Fig. 5f and Supplementary Video 2), which make them highly suitable for powering wearable and flexible electronics that require high resiliency to various deformations.

A flexible fiber-base electroluminescent indicator light is seamlessly integrated into clothing and efficiently powered by the integrated AEPGs. The intelligent luminescence display enhances self-visibility, thereby ensuring safer navigation in low-light conditions (Fig. 6h, Supplementary Fig. 40 and Supplementary Video 3). In addition, an ambient sensor, activated by self-powered supply system, enables real-time detection of atmospheric temperature and relative humidity. The collected data is transmitted to a microcontroller chip, which further relays the information to the textile display (Supplementary Fig. 41a). This real-time monitoring allows users to effortlessly track outdoor climate conditions, enhancing environmental awareness. This on-body interactive system could be particularly beneficial for individuals with conditions such as cerebrovascular disease or peripheral neuritis, who may have impaired sensitivity to environmental temperature and humidity (Supplementary Fig. 41b). By improving environmental perception, the system provides timely feedback and appropriate wear guidance, enhancing user comfort and safety.