Introduction

The rapidly expanding markets for electric vehicles, portable electronics, and grid storage have fueled the remarkable commercial success of lithium-ion batteries1,2. However, the high costs, safety concerns, and uneven geographical distribution of raw materials, such as lithium and cobalt, underscore the increasing need for the development of complementary energy storage solutions3,4. In this context, rechargeable zinc metal batteries (RZMBs) present a more economically viable option due to their lower cost5, enhanced safety6, and greater abundance of zinc (Zn) in the Earth’s crust7. Besides, the use of low-toxicity aqueous electrolytes in RZMBs also mitigates environmental risks associated with battery production and disposal, enhancing their sustainability. However, the commercialization of RZMBs is still hindered by Zn metal anodes (ZMAs)-related challenges, including uncontrolled dendritic growth and severe interfacial side reactions, leading to the poor stability and reversibility of ZMAs in practical scenarios8,9.

To meet commercial targets, ZMAs should exhibit an ideal Coulombic efficiency (CE) of 100% under demanding conditions: an anode loading of 5 mAh cm−2 with 80% Zn utilization per cycle and a fast-charging rate of 2 C (10 mA cm−2)10. However, these harsh electrochemical conditions—characterized by high areal capacities (≥5 mAh cm−2), high current densities (≥10 mA cm−2), and high depth of discharge (DOD; ≥80%)—often lead to premature failure of the ZMAs11,12,13. During Zn plating, both Zn2+ and the electric field tend to concentrate at the protuberances with high surface energy14,15, a phenomenon known as the tip effect. In particular, when plating Zn under high current densities, high fluxes of Zn2+ concentrate on such tips and rapidly deposit, leading to the rapid soaring of big protrusions, especially upon a high areal capacity16. Additionally, high DOD incurs a great volume change of ZMAs, leading to pulverization of Zn foil and the formation of dead Zn12,14. The resulting Zn dendrite and pulverization compromise the integrity of the solid electrolyte interphase (SEI), exposing fresh Zn continuously to the aqueous electrolyte and intensifying side reactions, e.g., Zn corrosion and hydrogen evolution. Collectively, these factors reduce Zn reversibility and limit the calendar life in practical RZMB applications.

Significant efforts have been devoted to addressing these challenges, including the development of artificial protective layers on Zn17,18,19,20,21, three-dimensional (3D) Zn hosts22,23,24,25,26, and surface modification of plating substrates13,27,28. However, the large volume changes and potential pulverization during Zn plating/stripping can compromise the long-term stability of protective layers over time20,29, while 3D frameworks often incorporate a high proportion of inactive materials, reducing overall energy density. Furthermore, modification of plating substrates, while effective, often entails complex procedures and additional energy consumption. Electrolyte engineering—such as concentrated5,30,31, water-organic hybrid32,33,34,35, or additive-containing electrolytes36,37,38—has shown promise in improving ZMAs’ reversibility and stability. However, these approaches often introduce new challenges, such as higher cost and elevated flammability31,39. Moreover, most of these approaches are often evaluated under moderate electrochemical conditions (e.g., small current densities), and very small DODs (e.g., below 1%)10. These limitations underscore the need for a balanced strategy that effectively mitigates dendrite growth and side reactions at a practical level while maintaining cost-effectiveness, safety, and energy density to advance RZMB commercialization.

Notably, the electrodeposition of Zn—encompassing both nucleation and growth processes—occurs directly at the electrolyte/Zn interface. This makes the electrolyte surface tension (σ) a particularly relevant parameter, as it significantly influences interfacial properties and should impact the deposition behavior of Zn. However, a fundamental and quantitative understanding of how σ governs Zn nucleation and growth remains lacking. Traditional electrolyte design has primarily focused on suppressing water activity by using high organic solvent content (>55 vol%)34,40 or modifying the adsorption layer on Zn through functional additives in electrolytes41,42. While these approaches have shown some improvement, they have not established σ as a targeted design parameter, nor have they provided quantitative relationships between σ and deposition outcomes. Furthermore, the impact of σ on battery performance under practical conditions (high current, high capacity, high DOD, Ah-level pouch cells) and its overall effects on dendrite suppression, SEI stabilization, and side reaction mitigation remain insufficiently explored.

In this work, we report a low-σ electrolyte strategy that rationally tunes the Zn nucleation and growth behaviors for the high-performance practical RZMBs. Unlike previous studies that overlooked the significance of σ, we conduct a fundamental analysis of how electrolyte σ influences Zn electrodeposition behavior. Specifically, we establish a quantitative theoretical framework based on classical nucleation theory, identifying σ as the key parameter controlling both nucleation energy barrier (nucleation process) and growth resistance (growth process). This low-σ design fundamentally alters sparse nucleation/coarse-grained growth into dense nucleation/fine-grained growth, even under harsh electrochemical conditions. Experimental and computational studies reveal that low-σ electrolytes promote uniform Zn deposition, stabilize the SEI, and suppress side reactions. These synergistic effects yield highly stable and reversible ZMAs: a high 99.72% average CE over a lifespan of 8400 hours (350 days) at 1 mA cm−2/1 mAh cm−2, and high stability exceeding 2200 (17% DOD) and 300 hours (85% DOD) at harsh electrochemical conditions (10 mA cm−2/10 mAh cm−2). The practical viability of our approach is further demonstrated in a 1.27 Ah pouch cell that operates under a lean electrolyte condition and a low N/P ratio. Our work establishes quantitative relationships between σ and Zn electrodeposition behavior, and demonstrates highly stable ZMAs in low-σ electrolytes even under harsh electrochemical conditions. These findings highlight σ as a crucial yet previously overlooked electrolyte property, introducing it as a design criterion of electrolytes for high-performance practical RZMBs.

Results

Theory of σ-governed Zn nucleation and growth

By combining classical nucleation theory with electroplating dynamics, we establish electrolyte’s σ as the governing parameter for Zn electrodeposition, controlling both nucleation (Step I) and growth (Step II). Typically, the nucleation process in liquid systems can be categorized into homogeneous nucleation and heterogeneous nucleation43,44, depending on whether a substrate is involved (Supplementary Fig. 1). During Zn electroplating on a copper (Cu) substrate (Fig. 1a), Zn undergoes heterogeneous nucleation, where the change of Gibbs energy ∆Ghet can be expressed as:

$$\Delta {G}_{{\mathrm{het}}}=f\left(\theta \right)\cdot \Delta {G}_{\hom }={{\rm{f}}}\left(\theta \right)(\Delta {G}_{{{\rm{S}}}}+\Delta {G}_{{{\rm{V}}}})$$
(1)

where ∆Ghom is the change of Gibbs energy during homogeneous nucleation, and f(θ) is a constant (0 ≤ f(θ) ≤ 1) determined by the contact angle (θ) between the Zn nucleus and substrate. \(\Delta {G}_{{\mbox{S}}}\) and \(\Delta {G}_{{\mbox{V}}}\) refers to surface energy and volume energy, respectively. Note that \(\Delta {G}_{{\mbox{S}}}\) (>0) represents the energy input required to form a solid surface during nucleation. This energy also needs to be overcome in the subsequent growth processes. In contrast, \(\Delta {G}_{{\mbox{V}}}\) (<0) reflects the energy release during the transition of Zn2+ (aq) to Zn (s) under a negative bias voltage \(\Delta \varphi\) (vs. Zn2+/Zn) (Fig. 1a, Supplementary Fig. 2).

Fig. 1: σ-governed Zn nucleation and growth.
Fig. 1: σ-governed Zn nucleation and growth.
Full size image

a Schematic illustration of the thermodynamics of Zn heterogeneous nucleation and growth. b Correlations among σ of electrolytes, Zn heterogeneous nucleation energy barrier (\(\triangle {G}_{{\mbox{het}}}^{*}\)), and Zn critical nuclei radius (r). c Correlations between σ and Zn particle radius (R) under low and high plated capacities. d Schematics of Zn nucleation and growth in low- and high-σ electrolytes. The gray hemispheres in a and d represent Zn deposited on the substrate. e Correlations among ԑ, solvent polarity, and σ. f Evolution of σ of electrolytes with varying concentrations of low-polarity solvents. g Evolution of Zn nucleation energy barrier and the ionic conductivity with varying concentrations of TEP.

In the nucleation stage (Step I), the nucleation energy barrier \((\Delta {G}_{{{\rm{h}}}{{\rm{et}}}}^{*})\), which is the maximum \(\Delta {G}_{{\mbox{het}}}\) required to form a stable nucleus, can be expressed as (Supplementary Fig. 3):

$$\Delta {G}_{{\mathrm{het}}}^{*}={{\rm{f}}}\left(\theta \right)\cdot \Delta {G}_{\hom }^{*}=\frac{{{\rm{f}}}(\theta )16\pi {M}^{2}}{3{{n}^{2}\rho }^{2}{F}^{2}{\Delta \varphi }^{2}}{\sigma }^{3}$$
(2)

and the corresponding critical nuclei radius (r) (Supplementary Fig. 4), can be expressed as:

$$r=\frac{-M}{F\rho \Delta \varphi }\sigma$$
(3)

where M is the molar mass of Zn; n is the mole of electrons transferred per mole of Zn; ρ is the density of Zn; F is the Faraday constant; \(\Delta \varphi\) (vs. Zn2+/Zn) is the negative bias voltage during the electroplating process. Detailed derivations of these equations can be found in Supplementary Note 1.

As such, the effect of σ on Zn nucleation can be quantified in Eqs. 2 and 3, and is summarized in Fig. 1b. Specifically, the energy barrier of Zn nucleation is proportional to σ3 (Eq. 2), indicating that reducing σ lowers the energy barrier and accelerates the nucleation process. Additionally, a lower σ also diminishes the critical nuclei radius (Eq. 3), resulting in smaller nuclei. Thus, a low-σ electrolyte is expected to promote denser and finer Zn nuclei compared to those in higher-σ counterparts, fostering a nuclei-rich growth mode.

In the growth stage (Step II), the ratio of two opposing terms (−ΔGS/ΔGV) should be minimized to enhance the thermodynamic favorability of Zn growth. Therefore, the radius of Zn particles (R) during growth can be expressed as:

$$R=\frac{-3M}{{nF}\rho \Delta \varphi }\times \frac{1}{({-\Delta G}_{{{\rm{S}}}}/\Delta {G}_{{{\rm{V}}}})}\times \sigma$$
(4)

Therefore, in high-σ electrolytes, Zn grows from fewer nuclei, forming large particles to reduce −ΔGS/ΔGV. These large particles readily evolve into big protrusions under a high plated capacity, raising short-circuit risks. Conversely, for the same −ΔGS/ΔGV, low-σ electrolytes reduce growth resistance—that is, the surface energy per unit area, equivalent to the electrolyte surface tension σ—yielding smaller, denser particles (Supplementary Note 1).

Importantly, it should be noted that the difference of particle size in high-σ electrolyte (RHigh σ) and low-σ electrolyte (RLow σ) is very little when the plated capacity is low (Fig. 1c). However, it amplifies sharply at the high plated capacity (Fig. 1c). This disparity highlights the critical role of σ on the growth of Zn, especially under high areal capacities that are required in practical applications of RZABs.

Collectively, σ governs both Zn nucleation (Step I) and growth (Step II), as summarized in Fig. 1d. Low-σ electrolytes enable (i) facile and dense nucleation (Step I), (ii) smaller nuclei size (Step I), and (iii) fine-grained growth even under high areal capacities (Step II).

Guided by the theoretical prediction that identifies σ as the governing parameter, we engineered low-σ electrolytes. This was achieved by incorporating solvents with low polarity—and consequently low dielectric constants (ε)—into a reference aqueous electrolyte (Fig. 1e, f). The high σ of aqueous electrolytes typically arises from strong dipole-dipole interactions between water molecules, specifically the H2O-H2O hydrogen bonds. By adding low-polarity solvents with weaker intermolecular attractions, we aim to disrupt these strong dipole-dipole networks and consequently lower σ. It enables us to probe the effect of σ as the primary variable. The σ of the reference electrolyte (2 mol kg−1 zinc trifluoromethanesulfonate in water with 1.5 wt% zinc acetate dihydrate) decreases upon the addition of solvents with lower ε, including ethylene glycol (EG), triethyl phosphate (TEP), dimethoxyethane (DME), bis(2-methoxyethyl) ether (G2), and tetraethylene glycol dimethyl ether (G4) (Fig. 1e, f). Electrolytes containing G4, G2, DME, and TEP exhibit lower σ compared to electrolytes containing EG, because of their lower ε values. Particularly, TEP enables the lowest σ due to its minimal ε and non-linear molecular structure (Fig. 1f). The addition of only 5 vol% TEP (denoted as 5% TEP) significantly reduces the σ to 55% of the reference electrolyte, while further increasing the TEP content to 10, 30, and 50 vol% results in only marginal additional reductions (Fig. 1f). The variation trend of σ aligns well with that of the contact angles as shown in Supplementary Fig. 5.

The nucleation energy barriers in electrolytes with different contents of TEP were also calculated and compared under a fixed bias voltage of 0.1 V based on Eq. 2. As shown in Fig. 1g, 5% TEP dramatically lowers the nucleation energy barrier to 2 × 10−22 J, which is only 17% of that in the reference electrolyte. However, further increasing TEP content leads to a substantial decline in electrolyte ionic conductivity, yet a negligible reduction in the nucleation energy barrier. Thus, the addition of merely 5 vol% TEP can effectively lower both the σ (to 55% of the reference electrolyte) and the nucleation energy barrier (to 17% of that in the reference electrolyte) without significantly compromising the ionic conductivity (still up to 90% of that of the reference electrolyte). Therefore, 5% TEP was selected for subsequent studies to achieve an optimal balance between low σ and high ionic conductivity.

Dense nucleation and fine-grained growth of Zn in low-σ electrolytes

To evaluate the impact of σ on the nucleation and growth of Zn, we examined the morphologies of Zn deposits on Cu foil. σ of the reference electrolyte (denoted as 0% Addi) and the 5% TEP are 68.2 and 37.8 mN m−1, respectively. Compared to 0% Addi, 5% TEP exhibits a slightly wider electrochemical stability window (2.55 V) but lower ionic conductivity (Supplementary Fig. 6). Zn||Cu half cells in these two electrolytes were assembled in a custom-made setup (Supplementary Fig. 7), with a 1 cm distance between Zn and Cu electrodes to allow the observation of morphological changes at high plating capacities.

As shown in Fig. 2a, Zn deposits in the high-σ electrolyte of 0% Addi exhibit sparse nuclei, resulting in large Zn protrusions during growth. At the beginning (3 minutes) of electroplating with a high current density of 10 mA cm−2, the Zn deposits appear as widely spaced large islands. Extending the plating duration to 30 minutes produces prominent Zn protrusions. After 2 hours, these protrusions further develop into larger, porous dendrites (Fig. 2b). This morphological evolution decreases the interelectrode spacing and elevates the discharge plateau (Supplementary Figs. 8 and 9). In striking contrast, Zn deposits in the low-σ electrolyte of 5% TEP exhibit dense nucleation and compact growth. The initial 3 minutes of plating yields uniformly distributed small islands (Fig. 2c), indicating the high density of small nuclei. Even after 5 hours of plating, the Zn deposits still maintain a dense, fine-grained morphology. Cross-sectional analysis also reveals a smooth, dendrite-free layer in 5% TEP (Fig. 2d). TEP’s ability to enable dense, uniform Zn deposition is further validated across various current densities in coin cells (Supplementary Fig. 10).

Fig. 2: Characterization of Zn deposits in high- and low-σ electrolytes.
Fig. 2: Characterization of Zn deposits in high- and low-σ electrolytes.
Full size image

a–d Top (a, c) and cross-sectional (b, d) views of Zn deposits at 10 mA cm−2 in high-σ electrolyte of 0% Addi (a, b) or low-σ electrolyte of 5% TEP (c, d). The plated capacities are 0.5, 5, 10, and 50 mAh cm−2. e–h RMS roughness of Zn deposits with plated capacities of 5, 10, 30, and 50 mAh cm−2 at a current density of 10 mA cm−2 (e, g) and the LSCM images of Zn deposits at 10 mA cm−2/50 mAh cm−2 (f, h) in 0% Addi (e, f) and 5% TEP (g, h). i, j The finite element simulation of the post-nucleation growth of Zn in different electrolytes. k, l XRD patterns of Zn deposits after 30 minutes, 1 hour, and 5 hours of electroplating at 10 mA cm−2 in 0% Addi (i) and 5% TEP (j). m Projected area of Zn deposits after 3 minutes of plating at 10 mA cm−2 in electrolytes with different σ. n RMS roughness of Zn surfaces after 5 hours of plating at 10 mA cm−2 in electrolytes with different σ.

For quantitative comparison, we measured the root-mean-square (RMS) roughness of Zn deposits. The 0% Addi sample shows obvious capacity-dependent roughness escalation, rapidly growing from 6.7 to 434 μm when the plating capacity increases from 5 to 50 mAh cm−2 (Figs. 2e, f, and Supplementary Fig. 11a). In stark contrast, 5% TEP maintains low RMS roughness at ~5 μm (Fig. 2g, h and Supplementary Fig. 11b), reaching only 5.2 μm at 50 mAh cm−2—a 98.8% reduction compared to 434 μm in 0% Addi. These dramatic differences confirm that high-σ electrolytes promote sparse nucleation and uneven growth, leading to rapidly escalating surface roughness with capacity. Conversely, low-σ electrolytes enable dense nucleation and uniform plating, maintaining a consistently smooth surface even at high capacities.

Finite element simulations of the deposition behavior correlate well with the experimental results, demonstrating stark contrasts in deposition behavior between high- and low-σ electrolytes. As shown in Fig. 2i, j, the initial modeling exhibits sparse and large nuclei on Cu foil in 0% Addi, whereas dense and small Zn nuclei in 5% TEP. Under the external electric field, Zn2+ fluxes flow to Zn nuclei along the trajectories of current vectors (Supplementary Fig. 12). As the simulation progresses, Zn2+ keeps accumulating at these sparse nuclei, resulting in large protrusions in 0% Addi (Fig. 2i and Supplementary Fig. 13). In contrast, the dense Zn nuclei in 5% TEP promote more uniform Zn deposition (Fig. 2j and Supplementary Fig. 13), diminishing the inhomogeneity of Zn growth.

Notably, we found that σ also regulates the orientation of crystalline facets of Zn deposits. In comparison to the (101) facet, the (002) facet of Zn not only contributes to the lateral growth of Zn but also exhibits high resistance to Zn corrosion and hydrogen evolution45,46,47. In 0% Addi, Zn deposits are dominated by the (101) facet, and the (002)/(101) intensity ratio is below 0.27 (Fig. 2k). In contrast, Zn deposits in 5% TEP show a significant enhancement in the (002) crystal plane, and the (002)/(101) intensity ratio rises from 0.22 to 1.18 as plating time extends from 30 minutes to 5 hours (Fig. 2l). The preferential growth of the (002) plane in 5% TEP also facilitates the high stability and reversibility of ZMAs as demonstrated in the later part.

Importantly, the favorable regulation of Zn nucleation and growth is versatile across a range of low-σ electrolytes. To compare the density of Zn nuclei in different electrolytes, we quantify the projected areas of Zn deposits on Cu after 3 minutes of electroplating at 10 mA cm−2 (Fig. 2m and Supplementary Fig. 14). The projected area—defined as the two-dimensional footprint of Zn deposits—serves as an indicator for nuclei density, with larger areas indicating higher nuclei density. Low-σ electrolytes, including 5% TEP, 5% G2, 5% G4, 5% DME, and 30% DME, exhibit much larger projected areas (35.2%, 36.5%, 36.5%, 25.1%, and 27.2%, respectively) than high-σ electrolytes (18.3% in 0% Addi, 12.7% in 5% EG, and 13.3% in 30% EG), indicating dense nucleation process. For the subsequent growth process, we analyze the RMS roughness of Zn deposits in different electrolytes, with smaller roughness values indicating smaller particles of Zn. After 5 hours of electroplating (Fig. 2n and Supplementary Figs. 15 and 16), Zn deposits in low-σ electrolytes show significantly lower RMS roughness, being 5.2 μm for 5% TEP, 16.7 μm for 5% G4, 25 μm for 5% G2, 30 μm for 30% DME, to 34.5 μm for 5% DME—less than 10% of that in 0% Addi (434 μm), indicating finer grains in low-σ electrolytes Additionally, Zn deposits in these low-σ electrolytes also exhibit a higher proportion of the (002) crystal plane. The (002)/(101) intensity ratio reaches 2.61 in 5% DME, 3.83 in 5% G2, 4.97 in 30% DME, and 9.49 in 5% G4 (Supplementary Fig. 17). These results highlight the universality of this strategy with good potential for extensive applications.

Stable SEI and suppressed side reactions in low-σ electrolytes

The dense nucleation and fine-grained growth enabled by the low-σ electrolyte enhance the SEI stability, effectively suppressing side reactions even under harsh electrochemical conditions. The SEI chemistry of deeply cycled Zn (10 mA cm−2/10 mAh cm−2 for 10 cycles) in 0% Addi and 5% TEP was studied using X-ray photoelectron spectroscopy (XPS). The O 1s spectra of deeply cycled Zn show characteristic peaks of Zn(OH)2-related product at ~531.0 eV and ZnCO3-related product at ~532.1 eV48,49 (Fig. 3a and Supplementary Fig. 18). The former is derived from the byproducts of Zn corrosion in aqueous electrolyte48,49, while the latter is attributed to the effective SEI component derived from Zn(Ac)2 in electrolytes31,49. In 0% Addi, ZnCO3-related product remains below 40% at all depths, while Zn(OH)2-related product reaches as high as ~40% after 150 seconds of sputtering (Fig. 3b). In sharp contrast, 5% TEP enables high ZnCO3 content of over 60% and low Zn(OH)2 content below 20% at all depths. This beneficial SEI composition is also consistent across other low-σ electrolytes (5% G2, 5% G4, and 30% DME) (Supplementary Fig. 19).

Fig. 3: σ-controlled SEI and side reactions of Zn.
Fig. 3: σ-controlled SEI and side reactions of Zn.
Full size image

a XPS spectra of O 1s of Zn in 5% TEP after 10 plating/stripping cycles at 10 mA cm−2/10 mAh cm−2. b Composition of ZnCO3- and Zn(OH)2-related components on Zn after various durations of Ar+ sputtering in 0% Addi and 5% TEP. c Relative intensity of ZnO−, F−, and S− fragment ions derived from deeply cycled Zn as a function of sputtering time. Insets: depth profiles of fragment ions. d Net amount of H2 generated during Zn plating/stripping at 5 mA cm−2/5 mAh cm−2. Zn electrodes were pre-cycled at 10 mA cm−2/10 mAh cm−2 for 10 cycles to achieve passivation before the measurement. e Arrhenius analysis of Zn/electrolyte interface in 0% Addi or 5% TEP. ZTR: Zn2+ transport resistance. f–i Schematics (f, g) and operando observation (h, i) of interfacial behaviors upon Zn plating/stripping under harsh electrochemical conditions in high-σ (f, h) and low-σ (g, i) electrolytes. Images in h and i were captured at 10 mA cm−2 with a plating duration of 450 seconds. Scale bar is 100 μm.

Time-of-flight mass spectrometry (TOF-MS) further confirms the suppressed Zn corrosion and robust SEI in low-σ electrolytes. We characterized the deeply cycled ZMAs by tracking ZnO− (byproducts of Zn corrosion) along with F− and S− fragments (byproducts of Zn(OTf)2 decomposition) as functions of sputtering depth (Fig. 3c). The signals of ZnO−, F− and S− are strong and intensify with detection depth, indicating a thick byproducts layer caused by the corrosion of continuously exposed fresh Zn in 0% Addi. In contrast, these signals are significantly weaker in 5% TEP and diminish with detection depth, manifesting the suppressed Zn corrosion and electrolyte decomposition (Supplementary Fig. 20). The TOF-MS results align well with the XPS depth profiling at extended sputtering durations (Supplementary Fig. 21).

Differential electrochemical mass spectrometry (DEMS) demonstrates the suppressed hydrogen evolution reaction (HER) in low-σ electrolytes. In 0% Addi, the net H2 evolution in the pre-cycled symmetric Zn||Zn cell peaks at 8.3 nmol min−1 and then continues to fluctuate around 7.5 nmol min−1 when the cell is charged/discharged at 5 mA cm−2/5 mAh cm−2 (Fig. 3d). In contrast, the H2 evolution in 5% TEP is significantly suppressed, showing only an initial peak at only 1.8 nmol min−1, and then gradually declining to close to zero. Similar suppression can also be observed at harsher conditions, i.e., 10 mA cm−2/10 mAh cm−2 (Supplementary Fig. 22). Severe HER generates OH− ions that corrode Zn, forming various zinc hydroxide hydrates. Consistent with this, XRD patterns of deeply cycled Zn show prominent zinc hydroxide hydrate peaks50,51 in 0% Addi, while no such impurities are detected in low-σ electrolytes (5% TEP, 5% G2, 5% G4, and 30% DME) (Supplementary Fig. 23).

The suppressed side reactions on Zn are also supported by the electrochemical analysis. Arrhenius analysis of the Zn/electrolyte interface (Fig. 3e, Supplementary Fig. 24, and Supplementary Table 1) reveals a significantly lowered Zn2+ transport resistance (ZTR) through the SEI in 5% TEP (62.4 kJ mol−1) than that in 0% Addi (74.2 kJ mol−1). The high ZTR indicates a thick by-product-covered surface on Zn in 0% Addi, impeding Zn2+ transport. Additionally, Tafel plots of pre-cycled ZMA show a more positive potential and a lower current, also confirming the reduced corrosion and HER in 5% TEP (Supplementary Fig. 25).

Figure 3f and g illustrates the impact of σ on SEI during Zn stripping/plating under harsh electrochemical conditions. Due to the sparse nucleation and coarse-grained growth, high-σ electrolytes lead to the formation of dendrites and coarse Zn particles, which mechanically fracture the SEI, persistently exposing fresh Zn into electrolytes and accelerating Zn corrosion and HER (Fig. 3f). In contrast, low-σ electrolytes facilitate dendrite-free Zn deposition and a stable SEI, attributed to the dense nucleation and fine-grained growth, which shields fresh Zn from electrolyte-induced side reactions (Fig. 3g). Operando imaging of planar Zn||Cu cells (Supplementary Figs. 26 and 27) support this distinction: high-σ electrolytes produce porous, dendritic Zn deposits (Fig. 3h), while low-σ electrolytes yield compact, homogeneous Zn layers (Fig. 3i). Notably, such dendrite-free and dense interface persists in 5% TEP even under harsher electrochemical conditions (45 mA cm−2, 22.5 mAh cm−2; Supplementary Movies 1 and 2).

Mechanism of TEP-induced reduction of electrolyte’s σ

Instead of entering the primary solvation shell of Zn2+, TEP disrupts the hydrogen-bonding framework, particularly at the Zn/electrolyte interface, thereby reducing the σ of electrolytes. The Zn K-edge X-ray absorption near-edge structure (XANES) spectra of 0% Addi and 5% TEP overlap completely (Fig. 4a), suggesting the almost identical electronic structure of Zn2+ in both electrolytes. The Fourier transformation of extended X-ray absorption fine structure (EXAFS, Fig. 4b) reveals a peak at ~1.60 Å, corresponding to the Zn-O bond length in [Zn(H2O)x]2+33,52. This result remains consistent even when the TEP content is increased to 30 vol% (Supplementary Fig. 28), suggesting that TEP does not alter the primary solvation shell of Zn2+. These findings align with the radial distribution function (RDF) of 5% TEP (Fig. 4c). The absence of TEP in the primary solvation shell indicates TEP is electrochemically stable against reductive decomposition during Zn plating. This is further supported by the lack of phosphorus species in the XPS spectra of deeply cycled Zn and the absence of parasitic currents in the LSV measurements (Supplementary Fig. 29). Therefore, the robust SEI does not originate from the reductive decomposition of TEP but instead results from the favorable regulation of Zn nucleation and growth in low-σ electrolytes. Furthermore, it also rationalizes the significantly improved stability and reversibility of ZMAs despite the trace TEP added, as TEP is not consumed during cycling.

Fig. 4: The role of TEP in low-σ electrolytes.
Fig. 4: The role of TEP in low-σ electrolytes.
Full size image

a, b Zn K-edge XANES (a) and EXAFS spectra (b) of 0% Addi and 5% TEP. The XANES spectra in a were normalized by subtracting a linear pre-edge background and setting the post-edge absorption to unity. c RDF of 0% Addi and 5% TEP. d, e Snapshots of MD simulation of 5% TEP and 0% Addi on Zn surface. The up panel is the enlarged view of the Zn/electrolyte interface. Atoms are color-coded as follows: oxygen, red; hydrogen, white; phosphorus, olive green; carbon, dark turquoise. f Average numbers of H2O-H2O hydrogen bonds (H-bonds) as a function of the distance to the Zn surface. Data points show the mean from n = 100 replicates (sampled every 0.2 ns over 20 ns). Error bars represent standard deviation. g Comparison of the reduction of H2O-H2O H-bonds at the interface (0.2 nm), near the interface (0–2 nm), and throughout the overall electrolyte.

Since σ in aqueous media generally originates from the dipole-dipole interactions (H2O-H2O hydrogen bonds) at interfaces53, we propose that TEP reduces σ by regulating Zn/electrolyte interfacial H2O-H2O interactions. As shown in Fig. 4d, e, and Supplementary Data 1, the average number of H2O-H2O hydrogen bonds as a function of distance from the Zn electrode was calculated from the trajectories of all molecules in 20 ns (sampled every 0.2 ns). As shown in Fig. 4f, the number of H2O-H2O hydrogen bonds exhibits a maximum at 0.2 nm from the Zn surface and decreases to a minimum at 0.4 nm for both electrolytes. The decrease is due to the lower density of water molecules between the electric double layer and the bulk electrolyte. Beyond 0.4 nm, the number of H2O-H2O hydrogen bonds remains stable up to 2 nm. Notably, the addition of 5 vol% TEP significantly reduces the number of H2O-H2O hydrogen bonds at the interface (0.2 nm) compared to 0% Addi. As shown in Fig. 4g, the H2O-H2O hydrogen bonds at the interface (0.2 nm), close interface (0–2 nm), and throughout the overall electrolyte are reduced by 30%, 10%, and 6%, respectively, when comparing 5% TEP to 0% Addi. Raman spectroscopy further confirms this interfacial disruption (Supplementary Fig. 30) by showing a shift of the ~3238 cm−1 peak to higher wavenumbers and a decrease in relative intensity. The primary mechanism of disruption is the adsorption of TEP molecules at the Zn/electrolyte interface (Supplementary Fig. 31), which disrupts the water network, reduces dipole-dipole interactions, and lowers σ.

Electrochemical stability and reversibility of ZMAs

We evaluated the electrochemical stability of ZMAs using symmetric Zn||Zn cells under harsh electrochemical conditions. Zn||Zn symmetric cells in 5% TEP operate stably across 2700 hours at 5 mA cm−2/5 mAh cm−2 (Fig. 5a). In contrast, cells in 0% Addi undergo rapid short-circuiting after only 150 hours. Under more demanding conditions (10 mA cm−2/10 mAh cm−2, 17% DOD), Zn||Zn cells in 5% TEP sustain stable operation for 2216 hours, while cells in 0% Addi fail within 110 hours (Fig. 5b). 5% TEP, therefore, extends the cycle life of Zn||Zn cells by 18-fold at 5 mA cm−2/5 mAh cm−2 and 20-fold at 10 mA cm−2/10 mAh cm−2 compared to 0% Addi. This enhanced stability is maintained under moderate cycling conditions (2 mA cm−2/2 mAh cm−2) (Supplementary Fig. 32).

Fig. 5: Zn plating/stripping under harsh electrochemical conditions.
Fig. 5: Zn plating/stripping under harsh electrochemical conditions.
Full size image

a, b Long-term stability of symmetric Zn||Zn cells in 0% Addi or 5% TEP at 5 mA cm−2/5 mAh cm−2 (a) or 10 mA cm−2/10 mAh cm−2 (b). Insets of b: enlarged profiles at different time slots. c Long term stability of symmetric Zn@Cu||Zn@Cu cells with 85% DOD at 10 mA cm−2/10 mAh cm−2. Insets: enlarged profiles at 1st, 37th, 75th, and 112th cycle. d CE of Zn||Cu cells at 1 mA cm−2/1 mAh cm−2. e Comparison of average CE and lifespans of half cells between this work to other approaches. f Comparison of current densities, areal capacities, and cumulative capacities plated of Zn between this work to other reported ZMAs. The numbers in (e, f) correspond to the numbering of the data points. Refer to Supplementary Table 2 (e) and Supplementary Table 3 (f) for the reference, description, and values corresponding to each data point.

Even at harsher electrochemical conditions of 85% DOD and 10 mA cm−2/10 mAh cm−2, Zn@Cu||Zn@Cu symmetric cells in 5% TEP demonstrate stable operation for 300 hours—150 times longer than in 0% Addi (Fig. 5c). The enhanced stability of ZMAs is not limited to 5% TEP: Zn plating/stripping in other low-σ electrolytes (e.g., 5% G2, 5% G4, and 30% DME) all exhibit improved cycle life compared to the reference electrolyte (Supplementary Fig. 33).

Notably, ZMAs show high reversibility in a long cycle life. The Zn||Cu asymmetric cell in 5% TEP sustains stably for 4200 cycles at 1 mA cm−2/1 mAh cm−2 (Supplementary Fig. 34). It achieves a high average CE of 99.72% over a lifespan of 8400 hours (350 days, Fig. 5d). This result shows improvement compared to other approaches, which typically exhibit average CE of 97.5–99.6% over short lifespans of 200–4000 hours19,27,31,54,55,56,57,58,59,60,61 (Fig. 5e and Supplementary Table 2). In sharp contrast, Zn||Cu cells in 0% Addi fail after 260 cycles, accompanied with abnormal long charging plateaus (Supplementary Fig. 34). Note that a minor yet observable CE fluctuation occurs between cycles 400–1400 in 5% TEP. This behavior, stemming from slight side reactions on the Cu current collector under low current density (Supplementary Figs. 35 and 36), does not compromise the overall stability of Zn plating/stripping in 5% TEP. Even at tenfold higher current density and capacity (10 mA cm−2/10 mAh cm−2), Zn||Cu asymmetric cells in 5% TEP still maintain stable cycling for 230 cycles, exhibiting highly repeated profiles and a high average CE of 99.2% (Supplementary Fig. 37).

We further compare the current densities, areal capacities, cumulative plated capacities, and CE of ZMAs in 5% TEP to other reported approaches5,27,54,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76 in Fig. 5f and Supplementary Table 3. Our strategy enables high stability and reversibility of ZMAs, with cumulative plated capacities of 6.75 Ah cm−2 (5 mA cm−2/5 mAh cm−2), 11.07 Ah cm−2 (10 mA cm−2/10 mAh cm−2) and CE of 99.72%, showing improvement compared to most reported values. Critically, this performance is achieved without compromising safety or cost-effectiveness, establishing the low-σ electrolyte strategy as a promising paradigm for highly stable and reversible ZMAs.

Electrochemical performance of full cells

Full cells incorporating Zn negative electrodes and high-loading NVO positive electrodes were evaluated. In 5% TEP, Zn||NVO batteries (NVO: ~17 mg cm−2) achieve high capacities of over 350 mAh g−1 at 5 mA cm−2 and 230 mAh g−1 at 34 mA cm−2, corresponding to high areal capacities of 6.1 and 4 mAh cm−2, respectively (Fig. 6a and Supplementary Table 4). When further increasing NVO mass loading to an industry-level of ~34 mg cm−2, Zn||NVO batteries still exhibit a high areal capacity of 12.6 mAh cm−2 at 5 mA cm−2 and 6.2 mAh cm−2 at 68 mA cm−2 (Supplementary Fig. 38 and Supplementary Table 4). The Zn||NVO batteries (NVO: ~17 mg cm−2) in 5% TEP exhibit stable cycling for 3000 cycles at 34 mA cm−2, with an average capacity of 194 mAh g−1. It presents a significant improvement over those in 0% Addi, which suffers from substantial degradation since the 200th cycle (Fig. 6b).

Fig. 6: Electrochemical performances of Zn||NVO full cells.
Fig. 6: Electrochemical performances of Zn||NVO full cells.
Full size image

a Voltage profiles of Zn||NVO full cells (NVO: ~ 17 mg cm−2) in 5% TEP evaluated at 5, 10, 17, and 34 mA cm−2. b Long-term stability of Zn||NVO full cells in 0% Addi or 5% TEP at 34 mA cm−2. Zn thickness: 100 μm, N/P ratio: 9.6. c Long-term stability of Zn||NVO full cells at a low N/P ratio of 1.9 and a high current density of 34 mA cm−2. Zn thickness: 20 μm. All tests in a–c were performed in coin cells. d Long-term stability of the 77 mAh single-layer Zn||NVO pouch cell at a low N/P ratio of 2.8 and high current densities of 5, 10, and 18 mA cm−2. e Optical images and the schematic of the 1.27 Ah Zn||NVO pouch cell. f, g Voltage profiles (f) and long-term stability (g) of the 1.27 Ah Zn||NVO pouch cell at 60 and 139 mA g−1 with an N/P ratio of 3.3.

To assess the feasibility of low-σ electrolyte, the N/P ratio was further decreased to 1.9 by using a NVO positive electrode (~17 mg cm−2) with a capacity of 6.1 mAh cm−2 and a thin ZMA (20 μm) with a capacity of 11.7 mAh cm−2. At a high operating current density of 34 mA cm−2, the low N/P ratio cell in 5% TEP still delivers an average capacity of 166 mAh g−1 (2.8 mAh cm−2) for 2000 cycles with an average CE of 99.97% and a slow capacity decay of 0.03% per cycle (Fig. 6c). In comparison, the low N/P ratio cell in 0% Addi cell decays rapidly in 500 cycles, ascribed to the depletion of Zn metal. These pronounced differences demonstrate the superior reversibility of ZMAs in 5% TEP electrolyte, resulting from effective suppression of both dendritic growth and side reactions.

To assess the practical application and scalability of our strategy, we fabricated a single-layer pouch cell (3 cm × 4 cm) with a total capacity of 77 mAh at a low N/P ratio of 2.8. The cell demonstrates excellent rate capability, achieving areal capacities of 6.4, 5.5, and 4.2 mAh cm−2 at high current densities of 5, 10, and 18 mA cm−2 respectively (Fig. 6d). The Zn||NVO pouch cell in 5% TEP maintains 85% capacity retention over 500 cycles at 5 mA cm−2, validating the scalability of our low-σ electrolyte strategy.

In addition, a 1.27 Ah pouch cell, measuring 10.2 cm in length and 9.2 cm in width, was fabricated based on a bipolar configuration (Fig. 6e, f). It employs a limited amount of ZMA (40 μm, 23.4 mAh cm−2, N/P = 3.3:1), an NVO positive electrode (7 mAh cm−2), and a lean electrolyte condition (5%TEP: 11.3 g Ah−1) (Supplementary Table 5, Supplementary Fig. 39). The pouch cell was first galvanostatically cycled at a low specific current of 60 mA g−1 (delivering an energy density of ~68 Wh L−1, Supplementary Fig. 40, Supplementary Note 2), and then cycled at a high specific current of 139 mA g−1. The cell maintains stable cycling for 120 cycles, after which it shows a high capacity retention of 84% (relative to the 6th cycle at 139 mA g−1) and an average CE exceeding 99.9% (Fig. 6g).

Discussion

Mitigating dendrite growth and side reactions of Zn at a practical level is a persistent challenge. In this work, we have proposed a universal low-σ electrolyte strategy that fundamentally transforms Zn deposition behavior. The key attribute of the low-σ electrolyte is the ability to enable dense Zn nucleation and fine-grained growth, which is ascribed to the reduced nucleation energy barrier and critical nuclei size. This regulated nucleation and growth mode enables dendrite-free Zn deposition, stabilizes the SEI, and suppresses side reactions even under harsh electrochemical conditions. Therefore, ZMAs in low-σ electrolytes achieve a high average CE of 99.72% at 1 mA cm−2/1 mAh cm−2 over a lifespan of 8400 hours (350 days)—a 2- to 40-fold improvement compared to other common approaches (typically 97.5–99.6% over 200–4000 hours). Even under harsh electrochemical conditions, e.g., 10 mA cm−2/10 mAh cm−2, ZMAs still exhibit significantly extended lifespans of 2,200 hours (at 17% DOD) and 300 hours (at 85% DOD), representing 20-fold and 150-fold improvements over conventional electrolytes, respectively. Ah-level pouch cells operating under lean electrolyte (11.3 g Ah−1) and low N/P ratios (3.3) also demonstrate the practical viability of this strategy. This work provides a potentially cost-effective, scalable solution that addresses the core challenges of RZMBs without sacrificing safety or energy density. The principles developed here could potentially be extended to other metal anode systems such as Li and Na metal anodes.

Methods

Chemicals

Zinc trifluoromethanesulfonate (Zn(OTf)2, 98%), zinc acetate dihydrate (Zn(Ac)2·2H2O, 99.995%), and tetraethylene glycol dimethyl ether (G4, ≥99%) were purchased from Macklin. Ethylene glycol (EG, anhydrous, ≥99.8%) and triethyl phosphate (TEP, ≥99.8%) were purchased from Sigma-Aldrich. Dimethoxyethane (DME) and bis(2-methoxyethyl) ether (G2, ≥99.5%) were purchased from Aladdin.

Preparation of electrolytes

To prepare the reference electrolyte of 0% Addi, 2 mol kg−1 Zn(OTf)2 was dissolved in deionized water, followed by the addition of Zn(Ac)2·2H2O salt as a SEI-forming additive with a weight ratio of 1.5 wt%. To prepare 5% TEP electrolyte, 5 vol% DI water in the reference electrolyte was replaced with 5 vol% of TEP. Similarly, 5 vol% or 30 vol% of DI water in the reference electrolyte was replaced with equivalent volumes of EG, DME, G2, and G4 to prepare 5% EG, 30% EG, 5% DME, 5% DME, 5% G2, and 5% G4. The surface tension of the electrolytes was measured using the tensiometer of SDC 350KS.

Characterizations

Zn||Cu cells containing 3.5 mL of 0% Addi or 5% TEP electrolyte were assembled in a custom-made setup (Supplementary Fig. 7) with a fixed distance of 1 cm between the Zn and Cu electrodes to prevent short circuits under high areal capacities. Zn electroplating on Cu was performed in this setup at a current density of 10 mA cm−2 for durations of 3 minutes, 30 minutes, 1 hour, and 5 hours. After electroplating, the cells were disassembled to retrieve the Zn-deposited Cu electrodes. These electrodes were rinsed with deionized water to remove residual electrolytes before the following characterization. Scanning electron microscopy (SEM) images were acquired using a Tescan VEGA3 instrument with a beam voltage of 20 kV. X-ray diffraction (XRD) patterns were recorded using a Rigaku SmartLab diffractometer with Cu Kα radiation. Data were collected in a continuous scan mode over a 2θ range of 30° to 50° with a scan speed of 5° per minute and a step size of 0.02°. The surface roughness of Zn deposits was measured using a laser scanning confocal microscope (KEYENCE VK-X200).

After cycling at 10 mA cm−2/10 mAh cm−2 for 10 cycles, the Zn||Zn coin cells were disassembled to retrieve the cycled electrodes. These electrodes were rinsed with deionized water to remove residual electrolytes before X-ray photoelectron spectroscopy (XPS) characterization. XPS was performed using a Thermo Fisher Scientific Nexsa spectrometer with an Al Kα X-ray source, operating at a beam energy of 12 keV. Time-of-flight secondary ion mass spectrometry (TOF-SIMS, TOF-SIMS5, Germany) was employed to analyze the SEI composition as a function of depth. Sample preparation followed the same protocol as for XPS analysis. Depth profiling was conducted using a 1 keV Cs+ sputter ion beam for material removal, with analysis performed in negative ion mode over an area of 50 × 50 μm−2. The pre-cycled Zn electrodes (10 cycles in Zn||Zn coin cells at 10 mA cm−2/10 mAh cm−2) were reassembled into Zn||Zn cells, and in situ hydrogen gas evolution during charging and discharging at 5 or 10 mA cm−2 was monitored using a differential electrochemical mass spectrometry (DEMS) system (Ling Lu Instruments) with a helium carrier gas flow rate of 0.7 mL min−1. After stabilizing the baseline with helium gas-blowing, the pre-cycled Zn||Zn cells were subjected to galvanostatic charging/discharging. The net H2 release was determined by subtracting the baseline.

The X-ray absorption spectra of 0% Addi, 5% TEP, and 30% TEP were collected from the Hard X-ray micro-analysis (HXMA) at the Canadian light source (CLS). The samples were measured in a liquid cell equipped with an open window sealed by Kapton tape, giving a total thickness of approximately 5 mm. The incident photon energy was calibrated using a Zn foil, with E0 set to 9659 eV, and the spectra were collected over an energy range of 9459–10207 eV (12 k). Data were acquired in fluorescence yield mode using a 32-element Ge detector. All XAS data were analyzed with the Athena software77, and the EXAFS data were processed using the Artemis program. A linear function was subtracted from the pre-edge region, then the edge jump was normalized using Athena software. The k2-weighted χ(k) data were Fourier transformed after applying a Hanning window function.

Computational study

Finite element simulations

To elucidate the dynamics of Zn electrodeposition, three-dimensional finite element simulations were performed using the COMSOL Multiphysics 5.6 software28. We employed the tertiary current distribution to model the electrochemical processes. The simulation accounted for electrode morphology evolution during plating/stripping via the Deformed Geometry module. Ion transport under the applied electric field was governed by the Nernst-Planck equation, while electrode kinetics were described by the Butler-Volmer equation at the relevant boundaries. All simulation parameters are provided in Supplementary Table 6.

Molecular dynamics (MD) simulations

The MD were conducted with GROMACS (version 2021.3)78,79. Initial molecular geometries were optimized using ORCA, and the simulation boxes were constructed with Packmol80. The optimized potentials for liquid simulations all-atom (OPLS-AA) force field81 was applied, with atomic charges derived via the RESP2 method in Multiwfn82. Following energy minimization, systems were equilibrated for 1 ns under NPT ensemble conditions using the Berendsen barostat. Production runs (20 ns) were then performed in the NPT ensemble at 300 K with a 1 fs timestep, employing the V-rescale thermostat (τT = 1 ps) and Parrinello-Rahman barostat (τP = 2 ps) for temperature and pressure control, respectively. Hydrogen bond analysis was performed using built-in GROMACS tools. To simulate the variation of H2O-H2O hydrogen bonds at the Zn (001)/electrolyte interface in 5% TEP, the Zn/electrolyte interface is modeled by adding 1675 H2O molecules, 9 TEP molecules, 64 Zn2+, 120 OTf− and 8 Ac− into a box above the Zn slab. The Zn slab model contains 4 layers of Zn (001) plane with an orthogonal supercell of 8√3*16. The area of the Zn/electrolyte interface is 3.62 nm × 4.18 nm. The electrolyte layer with a thickness of 4 nm was placed above the Zn slab, and the total thickness of the simulated box was 4.93 nm. For the 0% Addi electrolyte, the 9 TEP molecules were removed while keeping all other components unchanged. Hydrogen bonds in the simulation were defined by an O H distance of less than 3.5 Å and the angle of H-O-O less than 30°.

Electrodes fabrications

We prepared the NaV3O8·1.5H2O (NVO) powder through the following steps. 3.0 g commercial V2O5 powder (Sigma-Aldrich, ≥99.9%) was mixed with 100 mL NaCl aqueous solution (2 mol kg−1) under vigorous stirring at 25 ± 2 °C. After 96 hours, the product was centrifuged, washed with deionized water and ethanol, and freeze-dried for 24 h. The as-prepared NVO powder was mixed with multi-walled carbon nanotubes (conductive agent, Canrd, ≥99%) and polytetrafluoroethylene (binder, Canrd, 60 wt% dispersion in water) in an agate mortar at a mass ratio of 6:2:2 with deionized water (solvent-to-solid ratio of 1.1:1) to form a homogeneous, dough-like mixture. The mixture was then rolled on a Ti mesh (100 mesh, ≥99.9%, 270 μm, 28.9 mg cm−2) with a roller to get single-coated NVO electrodes. The electrode was dried at 60 °C for 72 hours. After drying, the electrode sheet was cut into rectangles using a precision cutter. The dimensions were 3 cm × 4 cm for the single-layer pouch cell and 9 cm × 10 cm for the 1.27 Ah pouch cell. The mass loading of NVO is ~17 mg cm−2, and the thickness of NVO is ~420 μm. The same methods were followed to prepare the NVO electrodes with a mass loading of ~ 34 mg cm−2.

Pouch cell assembling

The 1.27 Ah pouch cell was constructed with a bipolar structure (Fig. 6e) with two Zn foils (9 cm × 10 cm × 40 μm, >99.999%, MTI Corporation) clamped in the middle sharing one Ti foil current collector (9 cm × 10 cm × 10 μm, >99.99%, MTI Corporation) and two pieces of NVO electrodes (9 cm × 10 cm × 420 μm) on both sides separated by the glass fiber (9.2 cm × 10.2 cm × 240 μm, Whatman). The optical images and detailed parameters of these components are seen in Supplementary Figs. 39 and 40, and Supplementary Table 5. 5% TEP was used as the electrolyte. An Al packaging bag was used to accommodate the cell. The two positive electrode tabs and one negative electrode tab were bonded to the Al packaging films by the adhesive polymer tape (MTI Corporation) via hot pressing. The pouch cell bag was filled with 5% TEP electrolyte and left standing overnight to ensure complete wetting of the NVO electrodes. Subsequently, the bag was sealed under vacuum using the automated vacuum sealing machine (ShineYe Corporation), with a final 10 mL (14.4 g) of 5% TEP loaded in the pouch cell, equivalent to an electrolyte-to-capacity ratio of 11.3 g Ah−1. All the assembling procedures were completed in the atmosphere at 25 ± 2 °C. The pouch cell was tested under an external pressure of ~0.1 MPa. The calculation of the energy density of this Ah-level pouch cell can be found in Supplementary Note 2. The single-layer pouch cell was assembled with one piece of NVO electrode (3 cm × 4 cm × 420 μm) and one piece of ZMA (3 cm × 4 cm × 30 μm), with glass fiber separated in the middle (3.2 cm × 4.2 cm × 240 μm).

Electrochemical measurement

Galvanostatic charge and discharge tests

The galvanostatic cycling of Zn||Zn and Zn||Cu cells (Fig. 5a–d) and Zn||NVO cells (Fig. 6a–c) were performed in CR 2032-coin cells (SS 304) using glass fiber separator and 120 μL electrolyte. For Zn||Zn symmetric and Zn||Cu (≥99.99%, 20 μm, MTI Corporation) asymmetric cells, a 100 μm thick Zn foil was used. Before use, Zn foil was polished by softback sanding sponges (3 M, USA) and then rinsed with ethanol to remove surface oxides or impurities. Electrodes were punched into disks with a diameter of 12 mm for the coin cell assembly. Zn||Zn symmetric cells were evaluated at 5 mA cm−2/5 mAh cm−2 and 10 mA cm−2/10 mAh cm−2. Zn||Cu asymmetric cells were evaluated at 1 mA cm−2/1 mAh cm−2 with the cut-off voltage at 0.5 V. Symmetric Zn@Cu||Zn@Cu cells for 85% DOD testing were prepared as follows. First, 11.76 mAh cm−2 of Zn (equivalent to 10 mAh cm−2 at 85% DOD) was pre-deposited onto Cu disks (12 mm diameter) at 10 mA cm−2 in Zn||Cu coin cells. These Zn@Cu electrodes were then retrieved and assembled into symmetric Zn@Cu||Zn@Cu cells, which were evaluated at 10 mA cm−2 and 10 mAh cm−2. For Zn||NVO cells, the Zn foil thickness was 100 μm for cells with a N/P ratio of 9.6 and 20 μm for cells with a N/P ratio of 1.9. Electrodes were punched into disks with a diameter of 8 mm for the coin cell assembly. The Zn||NVO cell was first discharged to 0.2 V and then charged to 1.6 V at different current densities.

Arrhenius analysis of the Zn/electrolyte interface was performed using Zn||Zn symmetric cells. First, a stable SEI was pre-formed by galvanostatic cycling at 5 mA−2 and 5 mAh cm−2 for 10 cycles. Subsequently, electrochemical impedance spectroscopy (EIS) was conducted under potentiostatic mode at the open-circuit potential (0 V vs. OCV) after resting at OCV for 30 min. The impedance spectra were acquired with a perturbation amplitude of 5 mV over a frequency range from 100 kHz to 10 mHz with a total of 41 measurement points. This EIS protocol was repeated at 5 different temperatures (20, 30, 40, 50, and 60 °C) to obtain the temperature-dependent resistance for the Arrhenius plot. Tafel plots (Supplementary Fig. 25) were performed in a typical three-electrode system with the EC-Lab electrochemical workstation (Biologic VMP3, France), where pre-cycled Zn foil (10 cycles in Zn||Zn cells at 10 mA cm−2/10 mAh cm−2) acted as the working electrode, graphite rod as the counter electrode, and Ag/AgCl as the reference electrode. The Tafel plots were tested at a scan rate of 2 mV s−1 and a potential of ± 0.3 V vs. open-circuit potential. The exchange current densities were calculated by fitting Tafel plots in the voltage range from −30 to 30 mV vs. Zn2+/Zn.